f

43Mfr-.-<;

COLLEGE ALGEBRA

BY

HENTtY BUECHARD EINE

Pkofessor of Mathematics in Princeton Univeesity

GINN AND COMPANY

NEW YORK CHICAGO LONDON BOSTON ATLANTA DALLAS COLUMBUS SAN FRANCISCO

ENTEKED AT STATIONERS' HALL

COPYRIGHT,
1901, 1904,

BY

HENRY

B.

FINE

ALL RIGHTS RESERVED

PRINTED IN THE UNITED STATES OF AMERICA

532.10

^^^oy

c t

f t

CANS AND COMPANY PRIETORS BOSTON

PROU.S.A.

PKEFACE
In this book
I

have endeavored to develop the theory of the

algebraic processes in as elementary and informal a manner as possible, but connectedly and rigorously, and to present the processes themselves in the form best adapted to the

purposes of practical reckoning. The book is meant to contain everything relating to algebra that a student is likely to need during his school and college

and the effort has been made to arrange this varied material in an order which will properly exhibit the logical
course,

interdependence of It has seemed to

its

me

related parts. best to divide the book into

two

parts,

a preliminary part devoted to the number system of algebra and a principal part devoted to algebra itself. "I have based my discussion of number on the notion of
cardinal
first

number and the notion

of order as exhibited in the

1, 2, 3, .

instance in the natural scale

There are con-

siderations of a theoretical nature in favor of this procedure But experience has coninto which I need not enter here.

vinced

me
is

method

the best.

that from a pedagogical point of view also this The meaning of the ordinal definition of

an irrational number, for example, can be made clear even to a young student, whereas any other real definition of such a

number

is

too abstract to be always correctly understood

by

advanced students.

My

discussion of

number may be thought unnecessarily

fundamental

elaborate.

But

in dealing with questions of this

character a writer cannot with a good conscience omit points

which properly belong

to his discussion, or fail to give proofs

iii

IV

PREFACE

of statements which require demonstration. I hope the details of the discussion will interest the more thoughtful class of

students; but learn from it

all
is

the

that the general student need be asked to ordinal character of the real numbers

and of the relations of equality and inequality among them, and that for all numbers, real and complex, the fundamental
operations admit of definitions which conform to the
tative, associative,

commu-

and distributive laws. In the second or main part of the book I begin by observing that in algebra, where numbers are represented by letters, the laws just mentioned are essentially the definitions of the fundamental operations. These algebraic definitions are stated in detail, and from them the entire theory of the algebraic processes and the practical rules of reckoning are subsequently
derived deductively. I shall not attempt to describe this part of the book minutely. It will be found to differ in essential features from
the text-books in general use. I have carefully refrained from departing from accepted methods merely for the sake of novelty. But I have not hesitated to depart from these methods

when

this

seemed to me necessary

in order to secure logical

consistency, or when I saw an opportunity to simplify a matter of theory or practice. I have given little space to special devices either in the text or in the exercises. On the other

hand, I have constantly sought to assist the student to really

mas'-.er

the general methods of the science.

Thus, instead of relegating to the latter part of the book the method of undetermined coefficients, the principal method
of investigation in analysis, I have introduced
it

very early

and have subsequently employed it wherever this could be done to advantage. This has naturally affected the arrangement of topics. In particular I liave considered partial fractions in the chapter on fractions.

They belong

there logically,
in ele-

and when adequately treated, supply the best practice mentary reckoning that algebra affords.

PREFACE

mation and

Again, I have laid great stress upon the division transforits consequences, and in connection with it have

introduced the powerful method of synthetic division. This earlier chapters on equations will be found to contain a pretty full discussion of the reasoning on which the solution of
equations depends, a more systematic treatment than is customary of systems of equations which can be solved by aid of the quadratic, and a somewhat elaborate consideration of the graphs
of equations of the first and second degrees in two variables. The binomial theorem for positive integral exponents is

treated as a special case of continued multiplication, experience having convinced me that no other method serves so well to

convey to the student the meaning of this important theorem. I have introduced practice in the use of the general binomial theorem in the chapter on fractional exponents, but have
deferred the proof of the theorem itself, together with all that relates to the subject of infinite series, until near the end of the book.

In the chapters on the theory of equations and determinants

there will be found proofs of the fundamental theorems regarding symmetric functions of the roots of an equation and a

more important properties of resultants. These subjects do not belong in an elementary course in algebra, but the college student who continues his mathematdiscussion of the
ical studies will

need them.

The

like is to be said of the chap-

and of the chapter on properties of continuous functions with which the book ends. The ideas which underlie the first part of the book are
ters

on

infinite series

those of

and Georg Cantor.

Kowan Hamilton, Grassmann, Helmholtz, Dedekind, But I do not know that any one hitherto

has developed the doctrine of ordinal number from just the point of view I have taken, and in the same detail.
tions
to

In preparing the algebra itself I have profited by suggesfrom many books on the subject. I wish in particular

acknowledge

my

indebtedness to the treatises of Chrystal.

vi

PREFACE

The book has been several years in preparation. Every year since 1898 the publishers have done me the courtesy to issue for the use of the freshmen at Princeton a pamphlet
containing what at the time seemed to me the most satisfactory treatment of the more important parts of algebra. With the assistance of my colleagues, Mr. Eisenhart and Mr. Gillespie,

good and
of times.

new trial to select what had proved what had proved unsatisfactory. As a consequence, much of the book has been rewritten a number
I endeavored after each
to discard
;

No doubt subsequent experience will bring to light further possibilities of improvement but I have hopes that as the book stands it will serve to show that algebra is
many
not only more intelligible to the student, but also more interesting and stimulating, when due consideration is given to the
reasoning on which
its

processes depend.

Princeton University
June, 1905

HENRY

E.

FINE

K-^

CONTENTS
PAET FIRST NUMBEES
I.

The Natural Numbers

tiplication

Counting,

PAGE Addition, and Mul1

II.

III.

Subtraction and the Negative Division and Fractions

Irrational Numbers

16 27

IV.

S9
70

V.

Imaginary and Complex Numbers

PART SECOND ALGEBRA

I.

II.

III.

Preliminary Considerations The Fundamental Operations Simple Equations in one Unknown Letter
. >

79

93110-

IV.

Systems of Simultaneoi^s Simple Equations V. The Division Transforsiation

.... ....

127155-

176 Factors of Rational Integral Expressions VII. Highest Common Factor and Lowest Common Multiple 196 213 S^III. Rational Fractions

VI.

IX.

Symmetric Functions
Evolution
Irrational
Functions.

245
252

X. The Binomial Theorem

XL
XII.

260

Radicals

and

Fractional

XIII.

271 Exponents 298Quadratic Equations XIV. Discussion of the Quadratic Equation. Maxima and 304. Minima XV. Equations of Higher Degree which can be solved by

Means of Quadratics
vii

309-

viii

COXTENTS
XVL
PAGE Simultaneous Equations which can be solved by Means of Quadratics 317
340
.

XVII. Inequalities XVIII. Indeterminate Equations of the First Degree XIX. Ratio and Proportion. Variation

342
347

XX. Arithmetical Progression

XXI. Geometrical Progression XXII. Harmonical Progression
XXIII.

354 357

362

Method of Differences. Arithmetical Progressions OF Higher Orders. Interpolation

.

364
374

XXIV. Logarithms XXV. Permutations and Combinations XXVI. The Multinomial Theorem
XXVII. Probability XXVIII. Mathematical Induction XXIX. Theory of Equations

303

408 409
424

425
483
492
. .

XXX. Cubic and Biquadratic Equations

XXXI. Determinants and Elimination XXXII. Convergence of Infinite Series
. .

520

XXXIII. Operations with Infinite Series 639 XXXIV. The Binomial, Exponential, and Logarithmic Series 563 XXXV. Recurring Series 660 XXXVI. Infinite Products 564 XXXVII. Continued Fractions 666 XXXVIII. Properties of Continuous Functions 677

INDEX

691

A COLLEGE ALGEBRA
PAKT FIRST NUMBERS
I.

THE NATURAL NUMBERS COUNTING, ADDITION, AND MULTIPLICATION

GROUPS OF THINGS AND THEIR CARDINAL NUMBERS

In our daily experience things present Groups of things. themselves to our attention not only singly but associated in groups or assemblages.

The fingers of a hand, a herd of cattle, the angular points of a polygon are examples of such groups of things. think of certain things as constituting a group, when we distinguish them from other things not individually but

We

as a whole, and so our attention.

make them
let

collectively a single object of

For convenience,
Equivalent groups.
of letters

us call the things which constitute a

group the elements of the group.

One-to-one correspondence.

ABC

and

DEF

are so related that

The two groups we can combine

all their elements in pairs by matching elements of the one with elements of the other, one element with one element. Thus, we may match A with D, B with E, and C with F.

Whenever

it is

possible to

match
shall

all

the elements of two

groups in this manner,

say that the groups are equivalent ; and the process of matching elements we shall call bringing the groups into a one-to-one relation, or a relation
1

we

of one-to-one corresjyondence.

".\:'

'.,

(|;p;iXi:GE

algebra
to

Thcoiem:^
ffrottp,

J/,

iwoypyjt^^^^
to

they are equivalent

equivalent one another.

the

same third

For, by hypothesis,

we can

bring each of the two groups

into one-to-one correspondence with the third But group. the two groups will then be in one-to-one correspondence

with each other, if we regard as mates every two of their elements which we have matched with the same element of the third group.
Cardinal number. may think of all possible groups of things as distributed into classes of equivalent groups, any two given groups belonging to the same class or to different
classes, according as
it
is,

We

or

is

not, possible to bring

them

into one-to-one correspondence. Thus, the groups of letters A

same

class,

the groups

ABCD

and

BCD and EFGH belong to the EFG to different classes.

The property which is common to all groups of one class, and which distinguishes the groups cf one class from those of another class, is the number of things in a group, or its cardinal
number. In other words,
in a group, or its cardinal number, is to the group itself and every be brought into one-to-07ie correspondence
is

The number of things

that property which

common

group which with it.

may
:

The cardinal number of a group of things that property of the group which remains unchanged if we rearrange the things within the group, or replace them one by one by other things"; or again, "it is that property of the group which is independent of the character of the things themselves and of their arrangement within the group."
is

Or we may say

"

For rearranging the things or replacing them, one by one, by other things will merely transform the group into an equivalent group, 2. And a property which remains unchanged during all such changes in the group must be independent of the character of the things and of their arrangement.

THE NATURAL NUMBERS

Part.

when

say that a first group is ^ijmrt of a second group the elements of the first are some, but not all, of the elements of the second.

We

Thus, the group

ABC
it

is

From
If the
second a

this definition
first
ptf^'H't

a part of the group ABCD. immediately follows that

of three groups he a part of the second, and the of the third, then the first is also a part of

the third.
Finite and infinite, groups.

blage

is finite

when
it is

it

is

We say that a group ar -assemequivalent to no one of its parts

;

i7ifinite

when

equivalent to certain of

its

parts.*

Thus, the group ABC is finite; for it cannot be brought into one-toone correspondence with BC, or with any other of its parts.

But any never-ending sequence of marks or symbols, the never-ending for example, is an infinite assemblage. sequence of numerals 1, 2, 3, 4,

We

can, for instance, set

1, 2, 3, 4,

assemblage

up a one-to-one relation between the entire and that part of it which begins at 2, namely,
1, 2, 3, 4, 5,

between

(a)

and
by matching
1

2,3,4,5,6,...,
in (a)

(b)

with 2 in

(b),

being for every numeral that

we

2 in (a) with 3 in (b), and so on, there may choose to name in (a) a corresponding

numeral in (b). Hence the assemblage

is infinite.

(a) is

equivalent to

its

part

(b).

Therefore

(a)

Less and greater cardinal numbers.

Let

M and N denote any

two

finite

groups.
1.

It

must be the case that

or or

2.
3.

M and N are equivalent, M equivalent to a part of N,

is

N is

equivalent to a part of M.

* Of course we cannot actually take account of all the elements, one by one, an infinite group or assemblage, as it is more often called. We regard such an assemblage as defined when a law has been stated which enables us to say of every given thing whether it belongs to the assemblage or not.
of

4
In the
first
4,

A COLLEGE ALGEBRA
case

we

say that

M and N have the same cardinal

;

or equal cardinal numbers in the second case, that the cardinal number of is less than that oi N; in the third,

number,

that the cardinal

Thus,
is
if

number
group

of

is

greater than that of N.

3f

is tlie

of letters c6c,

and

N the

group

de/pr,

then

JIf

equivalent to a part of iV, to the part de/, for example. Hence the cardinal number of is less than that of iV, and the cardinal

number

of

is

greater than that of

M.

9
no

Note.

definition of finite group, 7, that there is ambiguity about the relations "equal," "greater," and "less" as

It follows

from the

here defined.

Thus, the definition does not make it possible for the cardinal number to be at the same time equal to and less than that of iV, since this would mean that 3/ is equivalent to and also to a part of iV, therefore
of

that iV

is

equivalent to one of

its

parts, 3,

and

therefore, finally, that

is infinite, 7.

10

Corollary.

If the first of three cardinal numbers

be less

than
is

the second, a nd the second less than the third, then the first also less than the third.

For

M
11

cardinals,
is

denote any groujis of things of which these are the therefore equivalent to a part of N, and iV to a part of to a part of P, 3, 0. equivalent
if

M, N,

is

The system of cardinal numbers. By starting with a group which contains but a single element and repeatedly "adding"
one new thing,
:

we

are led to the following

list

of the cardinal

numbers 1. The cardinal number

but a single element.

of a " group " like

I,

which contains

2. The cardinal number of a group like II, obtained by adding a single element to a group of the first kind. 3. The cardinal number of a group like III, obtained by adding a single element to a group of the second kind.

4.

And

so on, without end. "

We name these successive cardinals

and represent them by the signs

one,"

"

two,"

"

three,"

1, 2, 3,

THE NATURAL NUMBERS

Observations on this system.
of

'

Calling the cardinal

number

12

group a finite cardinal, we make the following observations regarding the list of cardinals which has just

any

finite

been described.
First.

Every cardinal contained

I

in this list is finite.

'

lent, 7

For the group is finite, since it has no part to wliich to be equivaand each subsequent group is finite, because a group obtained
;

by adding a single thing

because
I

to a finite

group
;

is itself finite.*

Thus,

II

is finite

is

III

is finite

because

II

is

and so

on.

Second.

Every

finite cardinal is
finite

contained in the
is

list.

number of some For, by definition, every But we can construct a group of marks finite group, as M. equivalent to any given finite group M, by making one mark for each object in And this group of marks must have a last mark, and therefore be M.
cardinal
the cardinal
III

I

included in the
therefore
itself,

list

of 11, since otherwise

it

it

would be never-ending and

and with

M, be

infinite, 7.

Third.

No two

of these cardinals are equal.

For, as just shown, all of the true of every two of them that one
8.

This follows from the definition in

groups
is

I,

II,

III,

are finite

and

it is

equivalent to a part of the other.

* "We
If

may

denote a finite group,

prove this as follows (G. Cantor, Math. Ann., Vol. 46, p. 490) and e a single thing, the group Me, obtained
:

by adding

For
If

let

G=

e to

H denote that the groups

P

M,

is

also finite.

G and
to

H are equivalent.
of its parts, 7.
'

which is matched with e in Me, and represent Let f donote the element of the rest of P by Pi. and e =/, we have 3/= Pi. since Then P\ f But this is impossible, since 3/ is finite and Pi is a part of M, 7. does contain p. (2) Suppose that is matched with e in Me, for then the rest of P, It cannot be tliat e in which is a part of M, would be equivalent to M. But suppose that c in is matched with some other element, as g in Me, and p is matched with/ in /'. that e in If P oe true on this hypotliesis, it must also he true if we reoombine the elements e. /, g so as to mat<-h r in P with r in Me, ami /' in P with g in Me. But, as 'just shown, we sliould then liave a pnvt of eiiuivaieut to M. Hence this liypothesis also is impossible

Me is not finite, it must be equivalent Let P denote this part, so that Me = P. not contain e. (1) Suppose that P does

some one

Me=

Me =

A COLLEGE ALGEBRA
THE NATURAL SCALE. EQUATIONS AND INEQUALITIES

13

call the signs 1, 2, 3, oi " " one," two," three," positive integers or natural numbers. Hence

The natural numbers.

We

their

names

"

A
14

natural number
scale.

is

a sign or symbol for a cardinal mnnber.

in

The natural

Arranging these numbers

an order

corresponding to that already given the cardinals which they represent, 11, we have the never-ending sequence of signa

1,2,3,4,5,...,
or "one," "two," "three," "four," "five," ., which the Jiatural scale, or the scale of the natural numbers.

we

call

15

Each sign in the scale indicates the number of the signs in that part of the scale which it terminates.
Thus, 4 indicates the number of the signs
of signs in turn,
1, 2, 3, is

1, 2, 3, 4.
I,

same as the number of groups the same as the number of marks in the
4
is

the

II,

For the number and this,

III,

lill,

last

group,

lill,

8.

And
16

so in general.

The

ordinal character of the scale.

The natural

scale,

by

itself

considered, is merely an assemblage of different signs in which there is a first sign, namely 1 to this a definite next follow;

ing sign, namely 2 to this, in turn, a definite next following and so on without end. sign, namely 3
;

In other words, the natural scale is merely an assemblage of different signs which follow one another in a definite and

known

order,

and having a

first

but no last sign.

Regarded from this point of view, the natural numbers themselves are with merely marks of order, namely of the order in which they occur

respect to time

when the scale

is

recited.

17

It is evident that the scale, in

common with

all

other assem-

blages whose elements as given us are arranged in a definite and known order, has the following properties :

THE NATURAL NUMBERS

1. may say of any two of its elements that the one " and the other " " follows," and these words precedes precede " and " follow " have the same meaning when applied to

We

"

any one pair of the elements as when applied to any other pair. If any two of the elements be given, we can always deter2. mine which precedes and which follows. If a, h, and c denote any three of the elements such that 3. a precedes h, and b precedes c, then a precedes c.

An assemblage may already possess these properties when given us, or we may have imposed them on it by some rule of arrangement of our own choosing. In either case we call the
assemblage an ordinal system.
Instances of the
of events in time
;

first

kind are

(1)

the natural scale itself

(2)

a sequence

a horizontal

line.

a row of points ranged from An instance of the second kind

(3)

left to right
is

along a group of men

arranged according to the alphabetic order of their names.

" " assemblage may also have coincident elements. Thus, group of events two or more may be simultaneous. We call such an assemblage ordinal when the relations 1, 2, 3 hold good among its non-coincident elements it being true of

An

18

in a

the coincident elements that

4.

If
c.

a coincides with

b,

and

with

c,

then a coincides
then a precedes

with
5.

If a coincides with

b,

and

b precedes

c,

c.

by numbers indicate

It is

their relative order in the scale that the natural

19

relations

of greater

and

less

among the

cardinal numbers.

For of any two given cardinals that one

natural

is

greater wliose

number occurs
:

later in the scale.

And the relation " if the first of three cardinals be less than the second, and the second less than the third, then the first is less than the third," is represented in the scale by the relation "if a precede b, and b precede c, then a
:

precedes

c."

8
In
fact,

A COLLEGE ALGEBRA
other method than this for comparing do not compare the cardinal numbers of groups of things the by the method of 8. On the contrary, we represent them by natural numbers, and infer which are greater and which less

we seldom employ any

cardinals.
directly,

We

appropriate from the relative order in which these natural numbers occur in the scale. The process causes us no conscious effort of thought, for the scale is so when any two of the natural numbers vividly impressed on our minds that, are mentioned, we instantly recognize which precedes and which follows. and B, that the population of A is Thus, if we are told of two cities, and that of B, 125,000, we immediately conclude that B has the

120,000,

because greater number of inhabitants, in the scale than 120,000 does.

we know

that 125,000 occurs later

20

word
21

In what follows, the Numerical equations and inequalities. " number " will mean natural number, 13 and the
;

letters a,

h, c

will denote

we wish to " coincide " in the natural scale, ber, or read " a a

When

any such numbers. indicate that a and h denote the same num-

we employ

the equation

= b,

equals i."

22

But when we wish

in the natural scale,

to indicate that

we employ one
''

a precedes and 5 follows of the inequalities

than
b

a <h, read
h

is less

"
;

> a,

read "

b is

greater than

<z."

23

Of course, strictly speaking, these words, "equal," <'less," and " greater," refer not to the signs a and b themselves, but " Thus, the phrase, a to the cardinals which they represent.
is less than ^," is merely an abbreviation for, "the cardinal which a represents is less than the cardinal which i represents." But all that the inequality a < b means /or the signs a and b themselves is that a precedes b in the scale. From 17, 18 and these Rules of equality and inequality.

24

definitions, 21, 22, it

1. 2.

immediately follows that

Tf
If

=b a <
a

and

ft

and
and

3.

li a = b

= b < b <

r,

then a then a

c,

= <

c.

c.

c,

then a <c.

THE NATURAL NUMBERS

COUNTING
Arithmetic
is

primarily concerned with the ordinal relations

25

existing among the natural numbers, and with certain operations by which these numbers may be combined.

The operations of arithmetic have their origin in counting. To discover what the cardinal number of a given group of objects is, we count the group. The process is a very familiar one. We label one of the objects "one," another "two," and so on, until there are no
Counting.

26

objects left

"two,"

being careful to use these verbal signs "one," without omissions, in the order of their occurrence

that suits our

in the scale, but selecting the objects themselves in any order whim or convenience and the sign or label with
;

which the process ends

cardinal

is

what we

seek,

the name

of the

of the group itself. For owing to the ordinal character of the scale, this last sign indicates how many signs

number

have been used all told, 15, and therefore how many objects there are in the group, 8. Thus, the process of counting may be described as bringing the group counted into one-to-one correspondence, 2, with a
part of the natural scale namely, the part which begins at " one " and ends with the last number used in the count.

counting

Observe that the natural numbers serve a double purpose in (1) We use a certain group of them as mere counters
:

in carrying out the process, and (2) we used to record the result of the count.

employ the

last

one so

We

have intimated that

it is

immaterial in what order

we
:

select the objects themselves.

This

may

be proved as follows

Theorem.

The

result of counting a finite

group of

objects is

27

the same, whatever the order in

which we

select the objects.

Suppose, for example, that the result of counting a certain group were 99 when the objects are selected in one order, P, but 97 when they are selected in another order, Q.

10

A COLLEGE ALGEBRA
consists of the first 97 objects in the ordeT

The group which

would then be equivalent to the entire group in the order Q, for, by hypothesis, both have been matched with the first 97 numbers of the natural scale, 3. But this is impossible, since it would make a part of
the group equivalent to the whole hypothesis, a finite group, 7.
;

whereas the group

is,

by
the

28

Another definition of cardinal number.

cardinal

We may make
:

theorem just demonstrated the basis of a definition of the

number

of a finite group,

namely

The cardinal number of a finite group of things is that property of the group because of which we shall arrive at the same natural number in whatever order we count the group.
This
led
if

is

the definition of cardinal

to

we choose

make

number to which we are naturally the natural scale, defined as in 16, our starting

point in the discussion of number.

ADDITION
29
Definition of addition.

To add 3
number,
8,

to 5 is to find

what number

occupies the third place after 5 in the natural scale.

We may

find this
scale,

by counting three numbers

6,

forward in the

beginning at

thus

6, 7, 8.

We
And

indicate the operation

writing 5

+ 3 = 8.

by the sign +, read "plus,"

add & to a is to find what number occupies the bth place after a in the natural scale. Since there is no last sign in the scale, this number may always be found. We call it the siim of a and b and represent it in terms of a and b by the expression a -\- b.
in general, to

30

Note. The process of finding a + 6 by counting forward in the scale corresponds step for step to that of adding to a group of a things the elements of a group of b things, one at a time. Hence (1) the result of the latter process is a group of a finite cardinals, so also does a +

+
b.

6 things, 8,

and

(2) if

a and b denote

See footnote,

p. 6.

THE NATURAL NUMBERS

Since a
on,

11
and so
all

1,

-\-

2,

and

so on, denote the 1st, 2d,

-\-

31

numbers

after a, the sequence a

1,

-\-

2,

denotes

which follows a. Hence any given number after a may be expressed in the form a + d, where d denotes a definite natural number. The process. To add large numbers by counting would be very laborious. We therefore memorize sums of the smaller numbers (addition tables) and from these derive sums of the " laws " of addition larger numbers by applying the so-called
that portion of the scale

32

explained in the following sections. The laws of addition. Addition is a " commutative " and an
" associative " operation
;

33

that

is, it

conforms to the following

a,

two laws The commutative

:

law.

-{-

b
is

-\-

34

The
a
to b.

result of

adding b
a

^o

the

same as that of adding

35

The

associative law.

(1) -{-

c)

= (a +

b)

-\- c,

The

result of first
to a,
is

adding

c ^o

OAid then

adding the
^o

sum so
a and

obtained

the

same as that of first adding b

then adding c

to the

sum

so obtaiiied.

Note. In practice, we replace the expression (a + 6) + c by a + 6 + c, our understanding being tliat the expression a + & + c + represents

36

tlie

result of

adding 6 to

a, c to

the

sum

so obtained,

and so

on.

Proofs of these laws.

First.

We may
lata
.3

prove these laws as follows.

37

The comm%itative

-\-

-\-

a.

Thus, the sums 3

2 and 2

are equal.
first

For

.3

2 represents the

number found by

counting
scale.

off

three

numbers, and after that two numbers, on the natural

Thus,
(a)

The group counted

the counters,

1, 2,
1, 2,

.3,

4, 5,
1, 2.

.3,

(b)

But as there is a one-to-one relation between the groups of signs (a) and and every one-to-one relation is reciprocal, 2, we may interchange the roles of (a) and (b) that is, if we make (b) the group counted, (a) will
(b),
;

represent the group of coimters.

12
Hence finding 3

A COLLEGE ALGEBRA
+2
is

equivalent to counting the group of signs

1, 2, 3,
1, 2.

(b)

In like manner, finding 2

is

equivalent to counting the group

1,2,

1,2,3.

(c)

But as (b) and (c) consist of the same signs and differ only in the manner in which these signs are arranged, the results of counting them
are the same,

27

that

is,

3.
b.

Similarly for any two natural numbers, a and

Second.
For

The

associative

law

a +(b

-\-

c)

= (a

-\-

b)

-}- c.

+ b, and then to the cth sign after this, namely to {a + b) + c, we count b + c signs all told, and hence arrive at the (6 + c)th sign after a, namely at a + (& + c).
in counting to the 6th sign after a,

namely

to a

The notion of cardinal number is involved in the proofs just But addition may be defined and its laws established independently of this notion, as is shown in the footnote below.*
given.
* The Italian mathematician Peano has defined the system of natural numbers without using the notiou of cardinal number, by a set of "postulates" which we may state as follows where " number " means " natural number." 1. The sign 1 is a number. 2. To each number a there is a next following mmiber call it a +. 3. This number a + is never 1. 4. If a += 6 + tben a= &. 5. Every given number a is present in the sequence 1, 1 +, (1 +)+, .

The numerals 2, 3, are defined thus 2=l + ,3=2 + ,---. The sum a+b is to mean the number determined (because of 5) by the series of formulas a+l = a+,a + 2=(a+l)+, . The series of formulas just written is equivalent to the single formula 6. a + (6 + l)= (a + b) + l. From 6, by " mathematical induction," we may derive the laws of addition: 7. a+ (b + c)= {a + b) +c. 8. a + b=b + a. If 7 is true when c= k, it is also true when c= k+1. First. For, by 6 and 7, = [a + (6 + k)] + 1 a+[b+{k + l)]-^a + [(b + k) + 1] = [{a + b) + k] + l= {a + b) + {k + l). But, by 6, 7 is true when c= 1. Hence 7 is true when c= 2, when c= ^, .. when c= any number, by 5. Second. We first prove 8 for the particular case 8'. a + 1 = 1 + a. If 8' is true for a=k, then {* + 1) + 1 = (1 + A") + 1 = 1 + (A; + 1), by 6. Hence if 8' is true for a k, it is also true for a= k+ 1. for a= 3, Hence, since 8' is true for a = 1, it is true for a = 2, Finally, if 8 be true for b = k, it is true for b = k + 1. For, by 7 and 8', a + {k + I) = (a + k) + 1= \ + (a + k) = 1 + (^ + a) = (1 + A-) + a= ik+l) + a. for 6=3, ... Hence, since 8 is true (by 8') when b= 1 it is true for b=2,

..

.-.

.-.

See Stolz and Gmeinor, Thcoretiache Arifhnipfik, pp. 13 ff., and the references to Peano there given; al.so Hnntiiiitfou in liiilli't'ni of the American Mathematical Society, Vol. IX, p. 40. H. Grassmanu (Lehrbuch der Arithmetik) was the first to derive 7 and 8 from 6.

THE NATURAL NUMBERS

General theorem regarding sums.

13
38

cation of these laws, 34, 35,

it

By making repeated applican be shown that

The sum of any finite number of numbers will be the same, whatever the order in xohich we arrange them, or whatever the

manner in which we group them, xohen adding them.

Thus, a

For

a4-fe

+ c + d=a + c + 6 + d. + c + = a + (6 + c) + = a + (c + 6) + d =a+c+&+
cZ
cZ.

(Z

35
34

35
39

Rules

of equality

definition of sum, 29,

1.

and inequality for sums. First. From the and the ndes of 24, it follows that

If a If If

2.

3.

= b, a < a > b,
b,

then a

-{-

then a
then a

=b +c<b c > b
c
-\-

-\- c.

-{- c.

-\- c.

Here
If

1 is

We may

obvious, since it a prove 3 as follows,

b,

then a and b denote the same number.

2 similarly.

and
(6

a>6, let a = 6 + d, 31. Then a + c = (& + d) + c =

Second.

c)

d, 34, 35,

.-..

>b +

c.

From
4.
5.
6.

1, 2,

3
c

it

follows conversely that

-\- c,

li a
If

-\-

=b
Z

then a

b.

If
b

+c< + +^>^+
then a

c,

then a <ib.

c,

then

a>

b.

Thus,
Fox-

it

-\-

c,

b.

otherwise
or else

we must have

either

a<b

and therefore a
c

;by

2),

a>b

and therefore a

Ob +
1, 2,

+ c<6 +

(by 3).

Third.
7.

It also follows If

from
c

3 that

8.
9.

If a
If a

= <

b, b,

and

and
and

>

b,

= d, c < d, c > d,
b

then a
then a

-^ c
-\-

then a
and
if c

+c= < c >

=
d,

b b
b

-\-

d. d.
d.

-{-

-\-

Thus,

if

a
c

=
b

b,

then a
d.

+c=

c,

then b

+c=

+ d.

Hence a

14

A COLLEGE ALGEBRA
MULTIPLICATION

40

Definition of multiplication.

To multiply a by
is a.

6 is to find the

sum

of b numbers, each of which

call this

We

terms of a and

sum the product of a by J and bhy a x b, ov a b, or simply ab.

ab

express

it

in

Hence, by definition,
41

-\-

to

b terms.

42 43

also call a the multiplicand, b the multiplier, the factors of ab.

We

and a and

The process. To find products by repeated addition would be very laborious. We therefore memorize products of the smaller numbers (multiplication tables), and from these derive
tion

products of the larger numbers by aid of the laws of addiand the laws of multiplication explained in the following
sections.

44

The laws of multiplication. Multiplication, like addition, is a commutative and an associative operation, and it is " dis-

with respect to addition the following three laws

tributive
:

"

that

is, it

conforms to

45

The commutative law.

ab

=
a,

ba.

The

result of multiplying
a.

by h

is

the

same as that of mul-

tiplying b by
Thus, 2

and 3 2

=
a

6.

46

The

associative law.

(l)c)

= (aF)

c,

of multiplying a by the product be that of m,ultiplying the product ab by c.

result

The

is

the

same as

Thus, 2

(3

4)

12

24

and

(2

3)4

24.
36.

In practice

we

write ahc instead of {ah)c.

Compare

47

The

distributive law.

(b -f c)

=
is

ab

+
c,

ac,

The

result

of first adding b and

ing a by the

sum

so

obtained,

the

and then multiplysame as that of first

THE NATURAL NUMBERS

multiplying a hy b and a by
so obtained.
c,

15
the products

and then adding

4

Thus, 3

(4

5)

27

and

12

15

27.

Proofs of these laws.

First.

We may

prove these laws as follows

-{-

48

The +
ac

distributive law: ab
{a + a a + a +
-\-

ac
(a

=
+

a(b
a

-\-

c).

(1)
41

For ah

= =

ioh terms)
a
c

+
c)

to (6

+
ah

terms

to c terms) + = a (6 + c). 35,

41

Hence
Thus, a{b

a(J)

-\-

+ )=
+
c)

-\-

ac

+
-\-

(2)

+c+
have

d)

a(b

+
be

ad

ah

+
b)

ac
c.

ad

by

(1)

and

35

We

also

ac

+
+

= (a +
-j

(3)
to c terms) {a

Eor ac

he

= =

{a
(a

+ +

+ 6) +
a

to c terms)
6')

(a

to

(h + h + terms =

b)c.

38

Second.

The commutative law: ab

a6

ba.

= = =

(1

l-h
6

1 + + l-b +6+

to a terms) b
-]

to a

terms

by

(3)

to a

terms

ba.

41

Third.

The
(ab) c

associative law: (ab)c =^ a (be),

= ab + ab + = a{b + b +

to c

terms
terms)

41

to

{be).

by

(2)

and

41

General

theorem regarding

products.
finite

These

laws

can

be

49

extended to products of any

number

of factors.

Thus,

The j)roduct of any finite number of factors is independent of the order in which the factors are multiplied together.
Rules
1.

of equality

and inequality

for products.
4. 5.

These are
be,

50
b.

If a If a If
a,

2.

3.

= b, < b, >
b-

then ac
then ac

= be.
<
be.

If ac If ac If

then a

then ac

> be.

6.

< be, ac > be,

then a <,b.

then a ">

b.

16
Here
1 is

A COLLEGE ALGEBRA
obvious, since iia

b,

We may prove 3 as follows, and 2 similarly. n a>b, let a = b + d. Then ac = (b + d)c = bc + dc,
The rules 4, 5, 6 are the converses of the reasoning used in 39.
1, 2,

then a and b denote the same number.

.: >bc. 3 and follow from them by

From
that

1, 2, 3,

by the reasoning employed

in 39, it follows
bd.

If a

li a
If

b <b

and

c c

d,

then ac

and
and

a> b

< d, c> d,

then ac
then ac

< bd. > bd.

II.

SUBTRACTION AND THE NEGATIVE

THE COMPLETE SCALE

51

Subtraction. To subtract 3 from 5 is to find what occupies the 3d place before 5 in the natural scale.

number

We find this number, 2,

by counting three numbers backward

:

in the scale, beginning at 4, thus 4, 3, 2, indicate the operation by the sign,

We

writing 5

read "minus,"

2.

And,

in general, to subtract b
^th

from a

is

to find

what number

52

place before a. call this number the remainder obtained by subtracting b from a, and represent it in terms of a and b by the expression a b. We also call a the minuend and b the subtrahend. Addition and subtraction inverse operations. Clearly the third number before 5 is also the number from which 5 can be obtained by adding 3.

occupies the

We

we may describe the remainder a b either before a, or as the number from which a can be obtained by adding b, that is, as the number which is
And,
in general,

as the

bth.

number

defined by the equation

53

(a

b)

-Jr

a.

SUBTRACTION AND THE NEGATIVE

17

Again, since saying that 7 occupies the 3d place after 4 is equivalent to saying that 4 occupies the 3d place before 7, we have 4 + 3 3 And so, in general, 4,

=
b
-\-

(a -\-b)

b=

a.

54
;

Since a
since a

-\-

b = a, b a,

^ 54,

subtraction undoes addition

and

55

53, addition

undoes subtraction.
2iiQ

We

therefore say that addition and subtraction

tions.

inverse opera-

The complete scale. The natural scale does not fully meet the requirements of subtraction for this scale has first number, 1, and we cannot count backward beyond that number.
; di,

56

Thus, on the natural scale

it is

impossible to subtract 4 from

2.

But there are important advantages backward as freely as forward. And

is itself

in being able to count

since the natural scale

there

is

merely a system of signs arranged in a definite order, no reason why we should not extend it backward by
:

placing a new ordinal system of signs before it. therefore invent successively the signs place before 1 1, which we place before

We

0,
;

which we 2, which

we place before 1 In this manner we

...,-5,

and so on.
create the complete scale

-4,-3,-2,

-1,0,1,2,3,4,5,...,
last sign or "

which has neither a first nor a which it is therefore possible

to count

number," and on backward, as well as

57

forward, to any extent whatsoever. Observe the symmetry of this scale with respect to the sign 0. As 3 is the third sign after 0, so 3 is the third sign
before
;

and so
of

in general.

Meaning

the

new numbers.

One of these new

signs,

58

be said to have a cardinal meaning. Thus, counting backward from 3 corresponds to the operation of removing the elements of any group of 3 things, one at a time.

namely

0,

may

This operation

may

be continued until all the elements have

18

A COLLEGE ALGEBRA

been removed, and we may call the sign of the cardinal number of the resulting " group " of no elements. therefore as one of the natural numbers. often regard

We

But

1,

2,

3,

have no cardinal meaning whatsoever.

the other hand, all these new signs have the same ordinal character as the natural numbers. Every one of them occupies

On

a definite position in an ordinal system which includes the And we may consider it defined by this natural numbers also.
position precisely as we may consider each natural number defined by its position in the scale. regard this as a numbers. sufficient reason for calling the signs 1, 2, 3

We

59

Positive and negative.

2,
the old

3,

distinguish the new numbers 1, as a class from the old, we call them negative,

To

])ositive.

60

The numbers of both kinds, and 0, are called integers to distinguish them from other numbers to be considered later. Let a, b, c denote any numAlgebraic equality and inequality.
bers of the complete scale. According as a precedes, coincides b. b, a b, or a with, or follows b, we write a

<

>

61

Since by definition the complete scale is an ordinal system, thus. 17, the rules of 24 apply to it also
;

If a

<

b and b
is,

<

c,

then a

<

c.

62

When

< b,

that

when

a precedes b in the complete

less

customary to say that a is algebraically or that b is algebraically greater than a.

scale, it is

than

b,

Observe that the words "less" and "greater," as thus used, mean this and nothing more. in the complete scale " 20 " 20 is less than - 18 " means 18." precedes merely Thus,

"precede" and "follow,"

63

Absolute or numerical values.

We call 3 the numerical value of

| 1

and use the symbol 3 to represent = 3. Similarly for any negative number. it, writing | 3| The numerical value of a positive number, or 0, is the number

or its absolute value,

itself.

Thus[3|

= 3.

SUBTRACTION AND THE NEGATIVE

Numerical equality and inequality.

19
64

any two numbers of the complete

numerically less than, equal
as
|a|

to,

Furthermore we say of and b, that a is or greater than b, according

scale, as a

<, =,

or

>
-

|b|.

Thus, while 3 than 2, and while 3. greater than

is

algebraically less than 2,

7 is algebraically less

it is

numerically greater
3,
it

than

is

numerically

OPERATIONS WITH NEGATIVE NUMBERS

also invent operations by which the New operations. negative numbers and may be combined with one another and with the natural numbers, as the latter are themselves

We

65

combined by addition, multiplication, and subtraction. We call these operations by the same names, and indicate them in the same way, as the operations with natm-al nimibers
to

which they correspond.

Employing

a,

as in 60, to denote

plete scale, but a and h to denote natural may define these new operations as follows
Definitions of addition and subtraction.
1.

any number of the comnumbers only, we

:

These are
after
a.

66

2.
3.

a
a

& is to
Z*

mean

the ith the

hih.

number

is

to

mean

number before
the same

a.

4.
5.

+ and a a + ( V) is to a ( h) is to

are to

mean

number
as a as a

as

a.

mean the same number

&.

mean

the same

number

i.

In other words, adding a positive number h to any number is to mean, as heretofore, counting b places forward in the scale; subtracting it, counting b places backward: while adding and subtracting a negative number are to be equivalent respectively to subtracting and adding the corresponding positive number.

20
Thus, byl,

A COLLEGE ALGEBRA

3
2

by by

2,

2=
5

1,

since

1 is

the 2d

number

after

3. 2.

= ==-

3,

since

is

the 5th

number

before

4,
5,

by

- 5 + (- 6 - (-

2)
2)

2 2

==-

7 (by 2).

4 (by 1).
:

67

Definition of multiplication.
1.

This

is

and
J)

are to

mean

0.

2.

a (

and (
h)
is

a) b are to

mean

ab.

3.

a)

to

mean

ah.

In other words, a product of two

0, is to

factors, neither of

which

is

be positive or negative according as the factors have the sam,e or opposite signs. And in every case the numerical

value of the product values of the factors.

Thus, by
2, 3,

is

to be the product of the numerical

x X

by

2 2

= - G, = 6.

and

= - 6.

68

the statements of ^%

Observe that and 67 are neither assumptions nor theorems requiring demonstration, but what we have called them definitions of new ojjerations.
(^()

The

origin and significance of these definitions.

it would be absurd to attempt to prove that 2 ( 3) = 2 3 with nothing to start from except the definition of multiplication of natural numbers, 40, for the obvious reason that 3 is not a natural number. The phrase "2 taken 3 times" is meaningless.

Thus,

But why should such operations be invented ? To make the negative numbers as serviceable as possible in our study of relations among numbers themselves and among things in the world about us.

The new operations have not been invented arbitrarily on the contrary, they are the natural extensions of the old operations to the new numbers.
;

tion as

In dealing with the natural numbers, we first defined addiand then showed that 2i counting forward process

SUBTRACTION AND THE NEGATIVE

21

the results of this process have two properties tvhich are independent of the values of the numbers added, namely
:

1.

+
:

b=^b

-\-

a.

2.

(1}

c)

= {a + b) +

c.

Similarly
properties
3.

we proved
4.

that products possess the three general

ah

= ba.

(be)

= (ai)

c.

5.

a(b

-\-

c)

ab

ac.

When we employ

letters to

denote numbers, these properties

1-5 become to all intents and purposes our icorking definitions of addition and multiplication for, of course, we cannot then actually carry out the processes of counting forward, and so on.
;

Clearly if corresponding operations with the new numbers are to be serviceable, these " definitions " 1-5 must apply
to

them

also.
:

And

66, 67 merely state the solution of

the problem

To make such an extension of the meanings of addition, multiand subtraction that sums and jjroducts of any numbers of the complete scale may have the properties 1-5, and that
plication,

subtraction
Thus,
(1)

may

continue to be the inverse of addition.

define adding a positive

when we

forward, and subtracting it as counting backward, the old definitions of addition and subtraction.
(2)

number 6 to a as counting we are merely repeating

From
if

this definition of addition

But
~-b

the commutative law a

b

+
6.

{b), and therefore 6

b

follows that 6 + 6 = 0. + aisto hold good, we must have 6 = 0, + ( 6) = or, since 6

it
;

we must have

+ (

6)

This suggests the definition a 6. 6) = a ( (3) If our new addition and subtraction are, like the old, to be inverse ( 6) = a + 6. Operations, we must also have, as in 66, 5, a (4) Again, to retain the old connection between addition and multiplication, 41,

we must
(

have, as in

67, 2,

a) 6

= =

a a

+ () + to 6 terms a to ^^ terms =

ab.

(5)

If
=^

the commutative law ab

ba

is

6)

6)

= 6a =

to hold good,

we must also have

a6, as in 67, 2,

22
(6)

A COLLEGE ALGEBRA
Similarly,

to

a terms

our wish to conform to the law ah = 0. a = and a namely,

= 0, and this fact together with ha leads to the definitions of 67, 1,

(-a)(-& +
hold good,
6)

it (7) Finally,

follows from

(6)

that
is

= -a-0 =

0.

But

if

the distributive
6)

law

to

we must

also

have

(_a)(-6 +
we
69

= {-a)(-6) + (-a)6 = (-a)(-6)-a6, by (4). We therefore have ( a) ( 6) - a6 = 0. And since also ab ab =

are thus led to define

0,

{ a)(b)

as ab, as in 67, 3.

The operations
ciative,

and distributive laws.

just defined conform to the commutative, assoIt remains to prove that the new-

with the laws which operations are in complete agreement suggested them.

To begin

with,

we have
a

+ (6 + c) = a + 5 + c, - + c) = a - J (Z.

(1) (2) (3)

c,

3i^l,-b=&-b + h = a,
as follows

from the definitions of addition and subtraction as counting forward and backward, by the reasoning in 37, 52.
I,

The commutative law, a

+ b = b + a.
d

First,

-a + b = b+(-a).
let

Forifa>6,
Then

b.

31,34

-a + b = -{d + b) + h
b

and

= _d-b + b = -d; + (-a) = b-(b + d), = b-b-d = -d.

by

(2)

and

(3)

66, 4

by

(2)

Proceed in a similar manner when b>a.

Second,
YoY
II.

+ (

b)

=b
b)

-{-

i^-

{-

b)

= -{a +

-{b

( a). + a)= - b + {-

a),hy

(2)

and 66,4.

The

associative law, a

First, a

+ [^ + (let

+ (b + c) = (a + b) + = a + + (- c). t)]
/^

c.

rorif6>c,

c.

31,34

SUBTRACTION AND THE NEGATIVE

Then
and
a

23

+
a.

[6

(-c)]

a a

+[d +

c c

+ {-c) =

+(- c)] = + {-c) =

a a

+ +

d,
d.

by

(3)

and

66,

Proceed in a similar manner when c>b.

Second, a

+ [(
I
Z<

b)

c^

= & + ( b) +
a

c.

This follows from

and the case

just considered.

Third, a

+[- +(- c)]= +((2)

^')

(-

c).
c)

This follows from

III.

and

66, 4, since

+ (

(&

+ c).

The commutative law, ab

ba.

First,

{ a)b^b{

a).

For

{-a)h = -ah = -ha = b{-a).

45; 67,2

Second,
For

(-

a)

(-

b)

= (-

b)

(-a)(-6) =
The

a6

(- a). = &a=:(-6)(-a).

45; 67,3

IV.

associative law, a (be)

= (ab)c.
a)
a)

First,

(-

a)

[(a)

b)

(6)

0)]

= [(= (=: c)

&)]

(ahc,

c).

For and

(-

[(-

(-

c)]

6c
c)

=-

46

67, 2,

[(- a)(- 6)](-

ai-

(-

=- a&c.

67,2,3

Second, the other cases

V.

may

be proved in the same way.

The

distributive law, a (b

c)

ab

+ ac.

First,

a[b

+{ c)] = ab
c)]

-{-a

c).
.

For

[6

+ (-

Hence

to a terms + (- c)] + [6 + (- c)] + = 6 + 6 + -to a terms +( c) + ( c) + -to a terms ^ha + {-c)a. II and III 41 67, 2 a[6 + ( c)] = a6 + a( c) by III
[6

Second, from this case the others readily follow.

Thus,

(-a)[6+(-c)]=-a[6 + (-c)] = - Ca& + a(- c)] = (-

'

a)&

+(-

a)

(-

c).

24
70

A COLLEGE ALGEBRA

result. As has already been observed, 68, in arithmetic or algebra, the laws a -\- b b + a, and so on, are equivalent to definitions of addition and multiplication, even when the letters a, b, c denote natural numbers. And we

The general

literal

have

now shown

that these definitions apply to all numbers

of the complete scale. By means of these laws

we may change

the form of a literal

expression without affecting its value, whatever numbers of the complete scale the letters involved in the expression may
denote.
Thus, whether
a, 6, c,

d denote positive or negative integers, we have

(a

6) (c

d)

= (a + = ac +

6) c

hc

+ (a + 6) d + ad + bd.

71

reasoning similar to that in

Rules of equality and inequality for sums. 39 that

a

We may

prove by

According as
so
is

< = <, =
,

or or

> b, >b+c

and conversely.

Hence
72
or

it is

true for positive and negative

numbers

alike that

An equation remains an equation and the sense of an inequality

remains unchanged when the same number
is

is

added

to

both sides,

subtracted from both.

73

Rules of equality and inequality for products. Observe that changing the signs of any two numbers a and b reverses the
order in which they occur in the complete scale,
Thus, we have

57.
2.

<-

2,

but 3

>2

< 2,

but 5

>it

From

this fact

and the reasoning of

a
ac

50

follows that

According as
so
is

<, =,
<, =,
>, =,

but

a(
Hence

c)

> b, or > or < b( c);

or
b^^,

and conversely.

SUBTRACTION AND THE NEGATIVE

25
74

Multiplying both sides of an equation by the same member,

positive or negative, leaves it an equation. Multiplying both sides of an inequality by the

same

positive

number leaves its sense imchanged. But multiplying both sides of an inequality by thv same negative number changes its sense, from <.to'>, or vice versa.

From
cation

the
0,

first

of these rules
a

and the definition of multipli-

by
:

namely
1.

= 0, we derive the following important

then ac
then a

theorem

If

= b,

2.

If ac

= be,

= be. = b, unless e = 0.

75

The exceptional case under 2 should be carefully observed. = 3 0, of course it does not follow Thus, from the true equation, 2
that 2

3.

Zero produets.

If a product be
0, either a

0,

one of

its

factors must be

0.

76

Thus,

if

ab

or b

0.

For, since

is

also equal to 0,

we have
and therefore
Numerical values

ab

= a =

b,

0,

unless b

= 0.

75
77

of products.
is

net of two or more factors

The numerical value of a prodthe product of the numerical

values of the factors.

Thus, |(-2)(- 3)(-4)|

= |-24|=24;

andl-2|-| -3|-| -4|

24.

The numerical value of a sum of Numerical values of sums. two numbers is the sum of their numerical values when the numbers are of like sign, but the numerical difference of these values when the numbers are of contrary sign.
Thus,

78

|-3 +

(-5)|

But

|2-|-(-5)|

= |-3I =

= |-8| = 8;and|-3|-M-5| = 3-f5 = 3; and|-5|-2 = 3.

8.

26

A COLLEGE ALGEBRA
THE USE OF INTEGRAL NUMBERS
IN

MEASUREMENT

79

Measurement.

We

use numbers not only to record the

results of counting groups of distinct things, but also to indicate the results of measuring magnitudes, such as portions of

80
81

time, straight lines, surfaces, and so on. measure a magnitude by comparing

We

lar

it with some particumagnitude of the same kind, chosen as a unit of measure.

If the

magnitude contains the unit a certain number of

times exactly,

we

call this

number

its

measure.

In particular, we

call the

measure of a line segment the

length of the segment.

Thus, we may measure a line segment by finding how many times we can lay some chosen unit segment, say a foot rule, along it.
If
it is

we

find that

it

three feet long, or that its length

contains the foot rule exactly three times, that is, its measure

we say
is 3.

that

82

The usefulness
scale,

of the natural

numbers

in

measurement

is

due to the fact that, by their relative positions

they indicate the relative measures tiiey are.
sizes

in the natural

of the magnitudes whose

83

We often Application of the negative numbers to measurement. have occasion to make measurements in opposite " directions " from some fixed " point of reference."
Thus, we measure time in years before and after the birth of Christ, longitude in degrees loest and east of Greenwich or Washington, temperature in degrees below and above zero.

then distinguish measurements made in the one in the other by the simple device of representing the one by positive numbers, the other by
direction

We may

from those made

84

negative numbers. Thus, consider the following figure:

4

-32-1
P-^
P-, P_i

P_4

Pi

Pj

Ps

P^"

DIVISION

AND FRACTIONS

27

Here the fixed point of reference, or origin, is 0, the unit are such that OPi, and the points P^, P3, , P_i, P_2, = P_iO = P_2^-i = OPi = P1P2 = A^3 = Above these points we have written in their proper order the numbers of the complete scale, so that comes over O. The distance of each point P from 0, that is, the length
is

of the segment OP, is then indicated by the numerical value of the number written above it and the direction of P from O
;

is

indicated by the sign of that number.

Thus,

3 over

P_ 3

indicates that

P_3 is distant 3 units to the left of

0.

Moreover, the order in which the points occur on the line is indicated by the order in which the corresponding numbers
occur in the scale.
Points used to picture numbers.
to-one relation,
0,
1,

Inasmuch

as there

is

a one-

85

between the system of points , P_2, P_i, and the system of numbers , 2, 1, 0, Pi, P2, 2, , either system may be used to represent the other.
2,

In what follows we shall frequently use the points to picture

the numbers.

III.

DIVISION

AND FRACTIONS

DIVISION REPEATED SUBTRACTION

The two kinds

the

of division.

There are

tivo

operations to which

86

name

one may

division is applied in arithmetic and algebra. The be described as repeated subtraction, the other as the

inverse of vmltiplicatio7i. There is a case in which the coincide. call this the case of exact division.

two
87

We

Division repeated subtraction. of these senses is to answer the

To divide 7 by 3 two questions 1. What multiple of 3 must we subtract from 7 remainder which is less than 3 ?
:

in the first

to obtain a

2.

What

is

this remainder ?

28

A COLLEGE ALGEBRA

may find the answer to both questions by repeatedly 4 and 4 3 3 Thus, since 7 subtracting 3. 1, we must subtract 3 twice, or, what comes to the same thing, we must

We

subtract

3x2.

And

the remainder
is

is 1.

This kind of division, then,

tion.

equivalent to repeated subtracIts relation to subtraction is like that of multiplication

7, 3, 2, 1

to addition.

Observe that the four numbers

the equation

are connected

by

7=3-2 +

And so in general, if a and b are any two natural numbers, to divide a by b, in the sense now under consideration, is to find two natural numbers, q and r, one of which may be 0, such that
88
89

^ bq

-^ r

and

<

b.

We
Note.

call

a the dividend,

b the divisor, q the quotient,

and

the remainder.

90

When
if

a and h are given, two numbers q and r satisfying

88

may always

be found.

a < 6, we have q = and r = a. Thus, If a>6, it follows from 31, 35 that we can continue the sum until it either equals a or will become greater than a if we add +6+

another

b.

And

if

have, 41, either a

q denote the number of terms in this sum, hq^ or a = hq + r, where r < 6.

b are given^ but one pair of

we
q

shall

Again,

when a and

numbers

and

satisfying 88 exists.

For were there a second such

bq-\-r

pair, say q', r\

we should have

bq'

r',

and therefore b{q

q')

r'

r.

But
but b{q

this is impossible, since r'

would be numerically
b.

less

than

6,

q')

not numerically less than

91
b,

Exact
as

division.

If the dividend a
6

when a

12 and

= 3,

is a multiple of the divisor the remainder r is 0. We then

b.

say that a is exactly divisible by of 88 reduces to a bq, or

In this case the equation

92

qb

a.

DIVISION

AND FRACTIONS

29
93

Hence when a is exactly divisible by b, we may also define the quotient q as the yiumber which, inidtiplied byh, will produce a. In this case, furthermore, we may indicate the division
thus, a
-i-

94

b,

and express the quotient q

in

terms of a and
as well as qb

by

one of the symbols j ov a/b, writing q

=j

= a.

THEOREMS AND FORMULAS RESPECTING EXACT DIVISION

Theorem
operations
;
1.

Exact division and multiplication are inverse

is,

95

that

^b

X b

a,

a7id a

x b

-=-

a.

These equations follow from the definitions

hi

93 and

87 respectively.

Theorem

2.

WheJi divisionis exact, miiUiplyirig dividend and

96

divisor by the

same number
a
?

leaves the quotient xmchanged.

For

if

qb,

then
then

am =
q

bm.

50, 4G

That

IS, if

= -
b

=
bm

94 97

Theorem
with

3.

Exact

division, like multiplication, is distributive

;

resjject to

addition and subtraction

that

is,

a,b =
I

+
c

b
>

a7id

,a
c
b

b
c

For

if

= =

qc,

and

q'c,

we have
Hence

+
c

qc

+
,

q'c

=
c

{q

q') c.

39, 47

b d + b a ^ = Q+g' = -+. c

94

And

similarly for subtraction.

Thus,

1^
3

+^=
3

+ 9; and 1^ ^
3

2-1

9.

Formulas

for adding

and subtracting quotients.

These are
be

98

ad

-\-

be
'

ad

bd

l^d^

bd

'b~d^

30

A COLLEGE ALGEBRA
a
c

^
And

ad

be

ad

be

similarly for subtraction.

18

Thus,

[-

10 = ^ + 2 = 8; 6
5

^ and

18

10 3

^^^

3-5

=
120
15

8.

99

Formula

for multiplying quotients.

This

is

ac

I'd^hd
For
if

qh,

and

g'd,

we have
c a d

ae
,

gg'

6d.

50, 45, 46

_
.

Hence
15 6 2

Z)

= =

g'

= _.
bd
^. and

ac

94

Thus, '

,^

15

15-6 3-2

90 = ,^
6

15.

100

Formula
is exact.

for dividing one quotient

by
,

another,
,

when

this division

This

is

a
h

a a

d
also q

he
=
Q

_ad
he

For

if

g&, c

g'd,

and
a
r

g"g',

we have
and

-^

~ d

-^

<i'

<l"i

94

Thus, '

^ = i^ = l = q bqd 10 24 = 4--2 = and 2;

be
, ,

g".

96,94

-=-

'

24-5 6-10

=2.
120

60

101

Exact division for negative numbers. Evidently the definition of quotient given in 93 has a meaning for negative numbers also, whenever the numerical value of the dividend is exactly divisible by that of the divisor. Expressing these quotients
as in

94,

we have
4.

the following theorem

is

102

Theorem

If

a.

exactly divisihle hy b, then

~b~^~b'

rb~~b'

a_a

^~b"

a_a

DIVISION
For
if

AND FRACTIONS
a

31
73, 67

qb,

we have

Hence

= ( q)b. q=

94

And

similarly in the other cases.

1.

Zero in relation to exact division.

definition of quotient given in divisor is 0.

On
is

93

meaningless

the other hand, the when the

103

For g X

number

is

0= 0, no matter what number q one which multiplied by gives

will give a,

may
;

denote.
(2)

Hence
is

(1)

every

and

there

no number

which multiplied by

0/0 would denote

2.

In other words, according to the definition of 93 and 94, the Symbol every number, and a/0 no number whatsoever.

But when the dividend

is

is.O,

ber b which
fact,

not 0, the definition of 93 has a meaning.

is 0.

and the divisor some numIn

the quotient denoted by 0/b

For, according to 94, 0/6 should denote the number which multiplied and is that number (and the only one), since 0-6 = 0. by b gives
;

FRACTIONS.

DIVISION

THE INVERSE OF MULTIPLICATION

mentioned in

The second kind

fixictions

of division

86

is

the gener-

alization of exact division defined as in

93.

It requires that

ordinal definition of these

be introduced into the number system. We seek an new numbers, like that given the
56.
c,

negative numbers in

theorem, in which

a, b,

One is suggested by the following d denote natural numbers.

104

Theorem 5. When a is exactly divisible bijh, and c by di, the quotients a/b and c/d occur in the natural scale in the same relative order as the products ad and be / that is,

a/b <, =,
1.

or

>

c/d, according as ad <, =, or

J = -, d
then

"> be.

For

if

-6d =
b

-d6.

50
93,94

But

-6

a,

and -d
a

c.

Hence

ad

= be.

32

A COLLEGE ALGEBRA
can show in a similar manner that
if

And we
If
2.

a/b<c/d, then ad<hc; and

But from
all this it follows,

a/h>c/d, then ad>bG.

conversely, that

If

ad

be,

then a/b

c/d.

For otherwise we should have either (1) a/b<c/d, and therefore ad<hc, or (2) a/b>c/d, and therefore ad>bc. And we can show in the same way that
If

ad<bc, then a/b<c/d; and

if

ad>bc, then a/b>c/d.

105

But the relative ordei Enlarging the ordinal number system. of ad and be in the scale is known, whether the values assigned
a, b,
c,

d be such

as

make a

divisible

by

b,

and

by

d, or not.
b,

Therefore, take any two natural numbers, a and

b is not 0,

of w^hich

and

w'ith

them form the expression

-, or a/b.
b'

exactly divisible by b, let a/b denote, as heretofore, the natural number which is the quotient of a \yj b; but if
is

If a

a/b for the moment merely as a new symbol, read whose relation to division is yet to be given, 122. give to all such symbols a/b, c/d, and so on, the property of order already possessed by those which denote natural numbers, by supposing them arranged in accordance
not, regard

"a over Then

6,"

with the rule

a/b

shall precede, coincide with, or follow c/d,

according as ad precedes, coincides with, or follows be. Or, employing the signs <, =, >, as heretofore, to
"

precede,"

" coincide " with," follow,"

mean

106

Let a/b <, =, or

Thus, 4/5
Again, 2/3
since
is

>

c/d, accordiiig as ad <, =,

is,

or^

be.

is

to precede 7/8, that

4/5<7/8,

since

to
1
;

0-3<2-

He between and 1, or 0<2/3<l. and 2/3<l/l, since 2 -Kl -3.

4-8<7-5. For 0/l<2/3,

107

To such

of the symbols

a/b

as denote natural

numbers

this rule assigns their proper places in the scale itself; while to the rest it assigns places between consecutive numbers of

the scale.

DIVISION
Note.

AND FRACTIONS

33
108

any particular symbol a/b with numbers of the scale, we have only to reduce a to the form = bq + r, where r<6, 88. Then if r = 0, so that a = bq, our rule a makes a/b comcide with q. But if r is not 0, our rule places a/b between q and q + I.
find the place thus given

To

respect to the

The entire assemblage of symbols a/b thus defined and like the natural scale which forms part of it arranged is an ordinal system.

109

For

it

has

all

the properties of an ordinal system which were enumerated

in 17, 18.

Thus,

if

a/b<c/d, and c/d<e/f, then a/b<e/f.

For

if

a/b<c/d, and c/d<e/f,

ad<bc,

we have

and cf<ed.
by

106
the corre-

Multiplying the sides of the first of these inequalities sponding sides of the second, we have

adcf<bced.

50

Hence
and therefore
Fractions.

af< be,
a/b<e/f.

50

106
110

When a/b
fraction ; and

we

call it

Su

does not denote a natural number, we call a its numerator, b its denomi-

nator,

and both a and

is

b its terms.

Hence

a symbol of the form a/b, defined by its position in an ordinal system which includes the natural numbers.

A fraction
From an

ordinal point of view, therefore,

we

are justified in

calling fractions numbers.*

* The rule of lOfi may also be used to define symbols of the form 1 /O, 2/0, and so on, ordinally. Thus, by the rule, 1 /O will follow every number a/b whose denominator h is not 0. For l/0>a/6, since 1 6>a-0. Asnin, 1/0, 2/0, and so on, will occupy the same place in our ordinal system. For 1/0=2/0, since 10= 2 0. But the rule will give nodefinite position to the symbol 0/0. Forwhatever the values of a and h, we should have 0/0= a/h, since 6 = a 0.

34
111

A COLLEGE ALGEBRA

Negative fractions.

We
or

also form fractions

whosB numerintegers,

ators,
-1

denoiniuators,

b a

both,

are

nerjatlve

as

-J defining

them ordinally
a

as follows:

^a

2.

Every negative fraction shall

irrecede 0.

3. Negative fractions shall be arranged with respect to one another {and 7iegative integers) In accordance with the rule :

a -<,=,
112

or

> -)

according as

ad <,

=,

or

> be.

fractions alike

To distinguish integers and numbers. from other numbers which we have yet to conAnd we call the system sider, we call them rational numbers. which consists of all these numbers the rational system. This system possesses an important property which does
The system
of rational

not belong to
113

its part,

the integral system, namely:

is

The rational system

dense;

that

is,

between every two

unequal rational numbers there are other rational numbers.

For
let

and - be any two d

.

fractions, such that

<a
,

We can prove

as follows that the fraction

be

ad

...
lies

between - and
^

a
b

c
--

2bd

Since -

OL

<-, we have ad<bc.

d
to both sides of

106

b
1.

If

we add ad

ad<bc, we have, by

39, 50, 106,

2ad<bc +
2.

ad,

.:

a{2bd)<b{be

o
ad),
'''

be

-{-

ad

b'^~2bd

If

we add

be to

both sides of ad < be, we have similarly

6c
.-.

bc

+ ad<2be,
3

(be

+
,

ad)d<c{2bd),
4- 5 2

.-.

+ ad c <- 2bd d

,5 - and Thus, between 6 4

we have

+
4

3 -6 o

38 7^ 4o

= ^" 24

19

DIVISION

AND FRACTIONS

35
114

Hence, when speaking of rationals, one must carefully avoid such expressions as the " next number greater or less than a To each integer given number for no such number exists.
''
;

such a next integer, but between any rational and a rational assigned as the next, there are always other rationals. In what follows let a, b, c, d denote Operations with fractions.
there
is

115

any given integers, positive or negative. In 98-102 we proved that, when a/b and c/d denote integers, we have
a
'

ad

-\-

be

^
b

ad
ad = -
be

be
bd
J

d
-

bd
.

d
-7-

^
3.

a c -- = ac
b

4.

a
b

be

c d

when

ad
be

-; is

an integer.

But the second member of each of the equations 1, 2, 3, 4 has a meaning even when a/b and c/d are not integers. Each of them is a definite fraction of the kind defined in 110, 111.

Hence

1, 2, 3,

4 at once suggest an extension of the mean-

ings of addition, subtraction, multiplication, and division which will make these operations applicable to fractions, namely The sum of two fractions a/b and o/t? is to mean the fraction
:

116

{ad

bc)/bd.
difference obtained by subtracting the fraction c/d is to mean the fraction (ad bc)/bd.

The

from
the

117

the fraction a/b

The product
fraction ac /bd.

of

two

fractions

a/b and c/d

is

to

mean

118

The quotient resulting from dividing the fraction fraction c/d is to mean the fraction ad /be.

a/b by the

119

Observe that these definitions are equivalent to the rules for reckoning with fractions given in elementary arithmetic.

The commutative,

associative,

and distributive laws control

120

these generalized operations.

mu

ac

Ca

poiid

orv

36
121

A COLLEGE ALGEBRA
rules of equality and inequality, 71, 73, also hold

The

good

for these operations.

Thus,
For
if

.r

a
b

if

f
f
ad

c e -;-^'

^,

then

a r
b

--

f
f

d
ce6/.

^.f = -.?,
b

then

aed/=

118,100,111
73, 106, 111

Hence

cb,

d and therefore -

122

Definition of a fraction as a quotient.

The

fraction a/I>

may

now

be described as the number which multiplied by b tvill produce a, that is, as the number which is defined by the equation

123

'^b b
Tor

a.

-.6 = -. - =
b
6
1

-- = -- = a.
1-6
1

106,111,118
119,
it

124

Division the inverse of multiplication.

From 118,
c

follows that

T'^d^d^H

""'"^

b^d^d'V

in other words, that multiplication and division, as defined in 118, 119, are inverse operations. Compare 55.
For, by 118, 119 and 106, 111,

we have
c c

a
b

_ad
be

adc
bed

_a
b

dc cd

_a

b' b

__ac bd

_ acd _a
bdc
b

rd de

_a
b

Hence we may
125
by

deseril)e the

kind of division now before us

as the inverse of multiplication

and say
find a mimher tvhich multiplied

To divide a/b by c/A

is to

c/d

tvill

produce a/b.

By
made

it

introducing fractions into our number system, we have possible always to find such a number, except when the

divisor

c/d

is 0.

DIVISION
This
is

AND FRACTIONS

37

algebra.

the usual meaning of division in arithmetic and the generalization of exact division, 93. Irreducible fractions. Reducing a fraction to its lowest terms.
It is

126

If the numerator
factor,

and denominator of a fraction have a common we can remove it from both without changing the value

of the fraction.

For

=
bin

-, since
b

am

-b

a- bm,

106.

When
tion
is

all

such

common
its

factors have been removed, the frac-

said to be in
7/"

lowest terms, or to be irreducible.

Theorem.
other fraction
tiples

a/b

be
is

an

irreducible fraction,
to
it,

and

a'/b'

any

127

which

equal

then

a' a7id b' a7'e

equimul-

of a and b
a' /b'

resjiectively.

For since

a/b, and therefore a'b

ab',

is

a factor of
b.

a'b.

But, by hypothesis, a has no factor in be a factor of a', 492, 1.

common

with

Hence a must

have a' = ma, where m is some integer. But substituting ma for a' in a'b = ab', we have mab = db', and therefore

We therefore
b'

mb,

50.

Corollary.

If two irreducible fractions are equal, their nunier-

128

ators

must

be equal,

and

also their denominator's.

THE USE OF FRACTIONS

IN

MEASUREMENT
129

The definition of length given in 81 Fractional lengths. only applies to such line segments S as contain the unit segment s a certain number of times exactly.
But even
or
its

if

S does not contain

s
;

exactly,

it

may

still

be

commensurable with

that

is, it

some other

aliquot part of s
:

contain the half, the third, In that case we define exactly.

may

length as follows

If a given line segment co7itains the hth part of the unit segment a times exactly, we say that its length is the fraction a/b.

130

38

A COLLEGE ALGEBRA
s exactly 7 times, the

(in

Thus, if S contains the 10th part of terms of s) is 7 / 10.

length of

131

Note.

Observe that

to this definition, so also

For

if

is the length of S in terms of s according every fraction of the form ma/mh. contains the 6th part of s exactly a times, it will contain the
if

a/h

is

mhth. part of s exactly

ma

times.

132

Fractions are useful in measurement for the same reason

that integers are useful namely, by their relative positions in the rational system, they indicate the relative sizes of the
:

segments whose lengths they are.

For
if

a/6 and c/d

are ad/hd and hc/hd, exactly ad times, in

are the lengths of S and T in terms of s, so also 131 that is, the bdih. part of s is contained in S
;

T
or

exactly 6c times.

Hence
that
is,

S<, =, S<, =,
It

or

>T, >T,

according as

ad<, =,

or >6c,

according as a/6<,

or >c/d.

106

133
is

hardly need be said that the definition of length here given equivalent to the definition ot fraction given in elementary arithmetic,
Note.
lesser fractions are there defined as fractions v?hich

and that greater or

correspond to greater or lesser line segments or other magnitudes.

134

Rational numbers pictured by points.

integers,

may
3
I

Fractions, as well as be pictured by points on an indefinite straight

line, 85.

_4
I

S-2
T
I

1
I
I

1
I

253
I I

L^

>
that

Thus, to construct a point, P, which will picture 7/3 in the same way A pictures 1, we have only to start at the origin and lay off the third part of the unit OA seven times to the right.
is the picture of 7/3. P', the corresponding point to the left of 0, "We proceed in a similar manner in the case of any given fraction,

positive or negative.

135

All such points are arranged along the line in an order corresponding to that of the rationals which they picture. With this in mind we often speak of one rational as lying to the left or right of another rational, or as lying between two
other rationals.

IRRATIONAL NUMBERS
IV.

39

IRRATIONAL NUMBERS

PRELIMINARY CONSIDERATIONS
Definitions.

square

"
;

aaa ...

The product aa is represented by a^, read "a. " the product the product aaa, by a^, read " a cube to w factors, by a", read " the wth power of a."
;

136

In the symbols
exponents; a itself

a^,
is

a^

a",

the numbers

2, 3,

n are

called

called the base.

called squaring a
;

Finding
finding
a",

a^

from a

is to

finding

a^,

cubing a

the wth power. The operation which consists in raising a given a given power is also called involution.

raising a

number to
137

Roots and logarithms.

If,

as

we

number, and n a positive integer, a" Call this number b then

;

are supposing, a is a rational is also a rational number.

a"

5.

This equation suggests two new problems To assign values to 7i and b, and then find First.
:

a.

Second.
Thus,

To
n

assign values to a and

b,

and then find

n.

(1) let

and

&

9.

The equation then becomes

and we

find that a
(2) let

Again,

=3 =2 =
3

or

and

-3 6 =
;

for both 32
8.

= 9 and (- 3)2 = 9. The equation then becomes =

8,

2"

and we

find that

for 2^

8,

When
1.

= 5,
b,

138

is

called the wth root of

b

and

is

expressed in terms

of "

by the symbol V^, the simpler symbol V^, read square root of b," being used when n = 2. 2. n is called the logarithm ofh to the base a, and is expressed in terms of a and b by the symbol log b.

n and

40

A COLLEGE ALGEBRA

3 are squai-e roots of

2

9,

Thus, since 3^ = 9 and ( 3)'- = 9, both 3 and and both may be written Vl) but see 139.
;

Again, 2

is

the logarithm of 9 to the base 3

that

is,

logs 9.

139

Note.

V9, we 3, by

Instead of representing both the square roots of 9 by the symbol may represent the positive one, 3, by Vg, and the negative one,

V9.

This

is

the usual

method
it.

of representing square roots in

elementary algebra, and

we

shall follow

140

Evolution and finding logarithms.

The operation by which

V6

is

found,

when n and

h are given, is called extracting the

xdh root ofh, or evolution.

The operation by which

given,
is

log^Z* is

found,

when a and

b are

Both
'i.41

called ^'fZiw^ the logarithm ofh to the base a. these operations are inverses of involution, 55,

124

involution has two inverses, while addition and multiplication each has but one, will be seen by comparing the three

Note.

The reason that

equations
\.

-\-h

c.

2.

ab

c.

3.

a^

c.

+ + a, ba, the problem : Given c and 6 in 1 or Given c and a, find b. 2, find a, is of the same kind as the problem But since a^ is not equal to b", the problem Given c and b in 3, find
Since a
b b

and ab

a, is

wholly different

in

kind from the problem

Given

and

a, find b.

143

New numbers new operations

point which

needed.

We
;

shall subsequently study these

in detail

for in algebra they are second in

importance to the four fundamental operations only. But the now concerns us is this: 27ie// necessitate further
extensions of the

number system.
at

In fact,

it is

once evident that

Va

can denote a rational

number

in exceptional cases only.

Thus, to cite the simplest of illustrations, neither denote a rational number. For
1.

V 1

nor

V2

can

Since the square of every rational

is

num ber
1

is

positive,

no rational

exists

whose square number.

2.

1.

Hence

cannot denote a rational

is 2. For clearly 2 is not as follows that it is not the

No

rational

number

exists

whose square

the square of any integer, and

we can show

square of any fraction.

IRRATIOXAL NUMBERS
Suppose p/q
to be a fraction in its lowest terms,

41
such that

{p/(/)^

2,

orpVg- =

2/l.

But
this,

since p'^/q^ is in its lowest terms, 492, 2, it would follow 128, that p- = 2, which is impossible, since pj is an integer.

from

Therefore
It

2 cannot denote a rational number. can be .shown in tlie same way that if a/b be any fraction in

its

lowest terms, ^a/b cannot denote a rational number, unless both a and b are nth powers of integers.

We are to make good this deficiency in our number system by creating two new classes of numbers the irrational numbers, of which v2 is one, and the imaginary numbers, of which
:

V 1
We

is

one.

shall treat the irrational

numbers

in the present chapter

and the imaginary numbers

in the chapter

which

follows.

THE ORDINAL DEFINITION OF IRRATIONAL NUMBERS

In the present chapter the letters a, b, c, and so on, will denote any r^ational numbers, whether positive or negative,
integral or fractional.

The rational numGeneral properties of the rational system. bers constitute a system which has the following properties 1. It is an ordinal system.
:

143

2.

It is dense

that

is,

between every two unequal numbers

lie
still

of the system, a

and

b,

there

other numbers of the

system.
3. The sum, difference, product, and quotient of every two numbers of the system are themselves numbers of the system, the quotient of any number by excepted. By the definitions which follow, we shall create a more extended system which possesses these same three properties, and which includes the rational system. 1. The number ^ separates the Separations of the first kind. remaining numbers of the rational system into two classes
:

144

42

A COLLEGE ALGEBRA

than)
than)

the one class consisting of all rationals wliicli precede (are less J, the other of all rationals which follow (are greater
A.

Let us name these two classes of numbers Ci and

c^
\

C'g

respectively.

pictures of

In the figure, the half line to the left of the point \ contains the pointall numbers in the class Ci, and the half line to the right the
all

point-pictures of

numbers

in the class C2, 134.

From
1. 2.

109, 111, and 113, it immediately follows that Each number in L\ precedes every number in Co.
is

There

no

last

number

in Cj,

and no

first in Cj.

it

Thus, were there a last number in Ci, there would be numbers between and 1/3, 113, which is impossible since, by hypothesis, all rationals
than
1

less

/3 are in C\.

145

2.

Instead of thus separating the rational system into the

three parts
Ci'

C'j, \, Co, we may join ^ to C'l, so forming a class made up of C\ and J, and then say The number 1 separates the entire rational system into two
:

parts, Cj'
1. 2.

and
is

C.^,

such that

Each number
There

in C\' precedes

a last

number

in

C'l',

every number in Cg. namely i, but there

is

no

first

number

in C\. to Co, call the residting class Cj',

146

3.

Or we may join \
:

and

then say
parts, Ci
1. 2.

The number \ separates the

and
is Cj',

entire rational system into tivo

such that

Each number
There

in Ci precedes every
Ci,

number
is

in
a,

C2'.

no last number in
J.

but there

first

number

in Co',

namely

It is

evident that each of the rational

numbers defines

similar separations of the rational system.

147

Conversely, if we are able, in- any way, to separate the entire rational system into two parts, Bi and B^, such that each

IRRATIONAL NUMBERS
number

43

either a last

in B^ precedes every number in Bo and that there is number in B^ or a J77'st in B^, the separation will

numbers and,
Thus,
let

serve to distinguish this last or first number from all other in that sense, to define it.
us assign the negative rationals to Bi and the remaining is the first There is then no last number in i>i, but

rationals to B^.

number

in Bz-

And

zero

is

distinguished from

called the first

number

in Bo, as perfectly as

all other numbers when by the symbol 0.

Note.
in B2.
11.3,

Obviously there cannot be both a last number in Bi and a first For there must then be rationals between these two numbers,

148

whereas, by hypothesis, every rational belongs either to Bi or to 2-

Separations of the second kind.

But we can

also, in various

149

ways, separate the entire rational system into a part Ai in which there is no last number, and a part A^ in which there is no first number.
every rational

Thus, since no rational exists whose square is 2, 142, is either one whose square is less than 2, or

one whose square is greater than 2. Let A2 consist of all positive rationals whose squares are greater than 2, and let A^ consist of all the other rational numbers. Then
1.

Each number
let ai

in vli precedes every

number

in A^.

be any number in Ai, and a^ any number in A2. Evidently ai<a2, if a\ is negative or 0; and if ai is positive, a-^Ka-i^,

For

and therefore
2.

ai
is

< a^.
no
last

There

number

in A^,

and no

first in A^.

For when any

less
less

than than

2, 2,

we can always
;

last in

Ax.

whose square is whose square is also hence no number can be assigned which is the 183, 2 (.3) Similarly no rational can be assigned which is the ^rs^ in A^.
positive rational, ai, has been assigned

find a greater rational

The new number a

classes

V2.

The

relation

between the two

150

numbers, A^ and A^, is therefore precisely the same as that between the classes Ci and C2 in the separation
of

corresponding to

^,

which was described

in 144.

44

A COLLEGE ALGEBRA

But no rational number exists which can be said to correto be defined by it. spond to the separation A^, Ao, or For since every rational belongs either to ^i or to A^, no rational exists which lies between A^ and Az, SiS ^ lies betweenCi and Cg. And since there
is

A 2, no

rational exists

no last number in Ai and no first which corresponds to this separation

Cj',
C'g

in

as

i corresponds

to the separation

of

145, or to the

separation C\, C^' of

146.

(Compare

147.)

Hence this separation .lu .1 2, creates a place for a 7iew ordinal number, namely, a number which shall follow all numbers
in

A I and

We
sent
it

precede all in A 2. invent such a number.

by the
a,
9.

letter a

later,

For the present we may reprewhen multiplication has been

defined for replace

151

we

shall

find that a^

2,

and we can then

symbol V2, 182. We then define this new number a as that number which lies between all positive ratlonals whose squares are less than 2
significant

by the more

and

all

whose squares are greater than

also express this definition
!

2.

We may
where
a-^

by the formula

<a<
<
is

a2

and

a>2,

denote any numbers whatsoever in A^ and A^

respectively, and, as heretofore,

means "precedes."
of the

152

Note.
tions of

Observe that
tlie

this definition

same kind

as the defini-

Like negative and fractional numbers given in 56, 110. these numbers, a is a symbol defined by its position in an ordinal .system of symbols which includes the natural numbers. It therefore has precisely

same right as they to be called a number. Our reason for inventing this and similar numbers is also the same as our reason for inventing negative numbers and fractions. They serve a useful purpose in the study of relations among the numbers which we already possess, and among things in the world about us.*
the
* We may add that there would bo no ohjection from an ordinal point of view to our inventing more tliaii one inuiilxT to corrpspond to the seiiaration the formula o, <a< b < Oj. A^, A^, say two numbers, a and b, detined ordinally by But there are objections of another kind to our inventing more than one such number. See page 67, footnote (3).

.IRRATIONAL NUMBERS

45
153

The irrational numbers in general. The real system. The particular separation of the rational system which we have been considering is but one of an infinite number of possible separations of a similar character. For every such separation we invent a new number, defining ordinally with respect to the numbers of the rational system V2 in 151, precisely as we have defined the number a
it

To
bers,

distinguish these

new numbers from the

rational

num-

irrational numbers, or simply irrationals. Again, to distinguish the rational and irrational numbers
call

we

them

alike

from the imaginary numbers, which we have yet to

consider,

we

call

them

rQ.al

numbers.

Finally, we call the system which consists of all the rational and irrational numbers the sijstem of real numbers, or the real
system.

Hence, using a to denote any irrational number, following general definition of such a number
:

we have

the

An irrational number, a, is defined whenever a law is stated which will assign every given rational to one, and but one, of two classes, Aj, Ag, such that (1) each number in A^ precedes
every number in A and (2) thei'e is no last number in A^ and no first number in Aj tlie defiiiition of a then being : it IS the one mimber which lies between all numbers in Ai and
;

154

all in Ag.
It is

A2;

also that

here implied that there are numhers in both the classes Ai and Ai and Ao together comprise the entire rational system.

An

irrational
it

number,

a,

is

said to be negative or positive

155

according as

that is, the numbers The real which constitute it are arranged in a definite and known order, For the definition of each irrational indicates how it 17. lies with respect to every rational: and from the definitions of any two given irrationals we can at once infer how they lie
:

precedes or follows 0. system is an ordinal system

156

with respect to one another.

46

A COLLEGE ALGEBRA.
let a

Thus
1.

and b denote any two given irrationals

then

If every rational which precedes a also precedes b, and every rational which follows a also follows b, the numbers a and b occupy the same position relative to the nmnbers of the

rational system.
therefore,

By

154, a

our definition of an irrational number, and b denote one and tte same number.

We

indicate this

by the formula:

2.

If

among
b,

the rationals which follow a there are some

which precede

then a itself must precede b ^or b follow

:

We

a).

indicate this by the formula

a

<

b or b

>

a.

among the rationals which precede a there are some which follow b, then a itself must follow b (or b precede a).
3.

If

We
157

indicate this by the formula

a

>

b or b

<

a.

different real numbers which precedes and which follows; also, that we may always draw the following conclusions with respect to three given real numbers, a, b, c

It thus

appears that

when any two

are given,

we can

at once infer

If a

b,

and b

= < <

c,

then a
then a

= < <

c.

If a
If a

< b, = b,

and b
and
b

c,

c.

c,

then a

c.

158

system is dense. For there are rational numbers not only between any two unequal rationals, 113, but also between any two unequal irrational numbers, and between
real

The

any

two numbers one of which

159

is

rational

and the other

irrational,

The

real

system
first

is

continuous.

Tlie real system, therefore,

possesses the

and second of the properties of the rational

IRRATIONAL NUMBERS

47

system enumerated in 143. But it possesses an additional property not belonging to the rational system, namely
:

If the entire veal system be separated into two j)(^^'ts, Ri and R2J such that each nmnher in E-i jjvecedes everij number
in Rjj there
not both.
is

either a last riumber in

Rj or a

first in Rj; I'ut

For in separating the real system into the parts i?i and E^ we separate the rational system into two parts, Ai and Ao, the one part consisting of all the rationals in Ei, the other of all the rationals in R2.

Every rational belongs either to ^1 or to An, and each rational in Ai precedes every rational in A^. Let a be the number which the separation Ai, A^ defines, 147, 154. Then either a is a rational namely the last number in Ai or the first

in A2, 147,

or, if there

be no last number in
154.

Ai and no

first in

A2,

is

an irrational lying between Ai and ^2,

If

1.

is

the last

number

in Ai,

it

is also the last in

Ei; for
it

were
a,

there any number in Ei after a, there that is, rationals in Ai after a, which
2. 3.
7?2-

would be
is

rationals

between

and

impossible.
in A2-,
it is

Similarly,
If If

if

is

the
it

first

number

also the

first in 2^2-

is

irrational,
i?i,

a belongs to
rationals in

must, by hypothesis, belong either to 7i or to for were there any it is the last number in Ei
;

number
that
if
is,

in i?i after a, there

Ai

after

would be rationals between a, which is impossible. And,

it

and

a, 158,

in like

manner,

a belongs to E^,

it is

the

first

number

in E^.

Finally, there cannot be both a last number in Rj and a first in JB2, since there would be rationals between these two numbers, 158, that is,
rationals belonging neither to

Ai nor

to

Ao

which

is

impossible.

indicate that the real system is dense and at the same time possesses the property just described, we say that it is

To

continuous.

Theorem.

real number,

a,

either rational or irxational,

is

160

defined, ivhenever

real system

may

stated by means of which the entire be separated into two parts, Rj, R2, such that

a law

is

each number in Ri precedes every number in R2

tlien

this

number,

a,

being either the last number in Ri or the first in R2.'

is

This

an immediate consequence of 147, 159.

48

A COLLEGE ALGEBRA
APPROXIMATE VALUES OF IRRATIONALS

161

Given any irrational number, a, defined as in 154. By method illustrated below we can find a pair of rationals, the one less and the other greater than a, which differ from each other as little as we please. Such rationals are called
the

approximate values of

a.

Let a be the irrational, V2, which lies between all positive rationals whose squares are less, and all whose squares are greater than 2. 1. We may find between what pair of consecutive integers a lies by

computing the squares of which is greater than 2.

1,

2,

3,

successively, until

we reach one

We see
2.

at once that 1^
lies

< 2 and

22

> 2.

Hence a
computing
than
2.

between
then find
. . .

We may

l.l^, 1.22,

and 2, or 1 <a<2. between what pair of consecutive tenths a lies by successively, until we reach one which is greater
1

We
3.

thus obtain 1.42<2 and 1.52>2; for 1.42= j.ge^ 1.52

lies

= 2.25.

Hence a

between

1.4

and

1.5, or

1.4<a<1.5.

By

a similar procedure

we

find, successively,

1.41

<a< 1.42,

1.414

<a< 1.415,
number

and so on without end.

4.

Let

rti

denote the nth number in the sequence

?ith

thus obtained, and ao the

1.4, 1.41, 1.414, in the sequence 1.5, 1.42, 1.415,

Then

ai<a<a2 and

a2

Oi

1/10",

and by choosing n great enough we can make 1/10" less than any positive number, as 5, we may choose to assign, however small. 5. We call 1.4, 1.41, 1.414 the approximate values of a = V2 to the first, second, third place of decimals; and so on.

Evidently the
to

process

thus

illustrated

may

be applied

number, a; for all that the process requires is a test for determining whether certain rationals are less or greater than a, and the definition of a, 154, will always supply such a test. We therefore have the following theorem

any given

irrational

IRRATIONAL NUMBERS
Let a denote any
number, as
8,

49
If any positive small, we can
162

rjiven irrational
it

number.

be assigned,

matters not

how

always find two

7'ationals, a^, a^y

such that

ai

<a<

a2 aiul ao

ai <

8.

Evidently this theorem is true of rationals also. Tlius, if a denote a given rational number, and
tto

ai

=
ai

1/10",

we have ai<a<a2, and we can make

a^

1/10",

2/10" as

small

a.s

we

please by choosing n sufficiently great.

ADDITION, SUBTRACTION, MULTIPLICATION, DIVISION

It

ties of the rational

remains to give the real system the third of the propersystem enumerated in 143. For this we

shall require the following

theorem

Theorem.
1.
2.

Let Ai and Ag be two classes of ixitionals such that Each number in Aj is less than every number in Ao,

163

There is no last nmvber in Aj and no first in K^, For every positive number, 8, that may be assigned, it matters not how small, ice can find i?i A^ a number ai, and in A2 a number a2, such that
3.

a2

ai <

8.

We may
That there

then conclude that between A^ and A2 there

one.

lies

one

number and but

is at least

one such

number

follows from

and

2,

by

154.

That there cannot be more than one such number follows from 3. For suppose that between every ai in ^1 and every ao in ^2 there were the two rationals d and d\ as indicated in the figure
:

Then

for every a^, a^

we should have
aoXi', and

and therefore
which
is

aa
it

ai

>d'

- ai> d,
3.

d,

73,121
39, 121

impossible, since

contradicts

50

A COLLEGE ALGEBRA
of

Nor can there be two numbers, one or both

;

which are

irrational^

for between these two numbers there lying between every ai and a^ would be two rationals also lying between every ai and ao, 158, which

we have
164
Note.
154, in

just

shown

to be impossible.
differs

This theorem
that
it is

from the

definition of an irrational

number,
lies

not here a part of the hypothesis that every rational

either in

^i

or in A2.

165

Addition.

Let a and b denote any two

rational or irrational,

and

let Ui, a^,

l>i,

h^,

(/iveii real numbers, denote any rationals

whatsoever such that

i

<a<
is

a^,

and
last

^i

< b < h^.

(1)

Observe that there

no

number

of the kind denoted

or ^1, and no first number of the kind denoted by a^ or and that if any positive number, as 8, be assigned, it matters not how small, we can always choose a^, a^, and 7*1, h^, 162, so that both ! < S and h.^ h^ < 6. rtj (2)

by

ffi

Z2

When
from

find their sum,

(1),

both a and b are rational, say a a + yS, by the rule of

116

a and b = yS, we can and it follows

;

by

121, that
i

^>l

<

i8

<

^2-

Moreover, whether a and b are rational or from (1), by 121, that

1

not, it follows

^'1

<

^2-

(3)

These considerations lead us

to define the
:

sum

of a

and

b,

when one
166

or both are irrational, as follows

a.

The sum of
which
as
lies

and

b,

written a

b.

is

to

?m// that number

between all the nnwhrrs

it is

+ \\.

In other words,
a-i

the

all the numbers number defined by the formula

aj

bi

and

bi

<
ao

a 4- b

<
bj

ao

+ ba,

where

ax, aj, bj,

bo denote
ai

any rationals whatsoever such that

<a<

and

< b < bg.

IRRATIONAL NUMBERS
To
1.

51
is

justify this definition

we must show

that there

one and but one

such number a

Each
There

(iTi

+ b. + 6i is
no
bi

This follows from

less

163

for

than every a2
5i

+ +

b^.

2.

is

last ai

and no
a^

first a-z

+
;

b^.

Thus, ai

+ +

cannot be the

last ai

6i

for since there

is

no

last

a-i

and no
3.

last 6i,

we can choose
6i

and

bi so

that

ai>ai and
choose

6i>6i',

and

therefore ai
If

> a\' +

6i'.

any

positive rational,

5,

be assigned,

we can

ai,

a^-,

&i, &25

so that

a2-ai<5/2
and therefore
{a^

and

62

-&i<5/2,
hi)

102
121

60)

(cli

<

5.

as in Definition of a, i, a^ have the same meanings 165. Considerations like those in 165 lead us, when a is

a.

Let

167

irrational, to define

a as follows

The symbol

is to

meaii the number defined by the formida

168

a2 <C
where
ai, a2

a <C

a^,

denote

any rationals whatsoever such that

a^ "^
Si

"\ a2.
is

It

a
1.

follows from 163 that there for

one and but one such

number

Each
There

2.

az,
3.

is less than every ai, since ai<a2. 73, 111 no last ao and no first ai. Thus, were there a last there would be a first a2 but no such number exists. We can always choose ai, a^, so that

az

is

Subtraction.

ao)

ao

< 5.
from
is, a,

162

b,

is

to

The resiilt of mean the number

subtracthir/ h

written

169

+ (

b)

that

a-b = a+(-b).
The meaning
It follows

of a

+ (
ai

b) itself

is

known from

166, 168.

from

166, 168 that a

b may also be defined by the formula

b

60

<a

< a2

61,

where

ai, 02, 61, 62

denote any rationals whatsoever such that

ai<a<a2 and 6i<b<62.

52
170

A COLLEGE ALGEBRA
Let a and b be any two
h.j,

Multiplication, both factors positive.

given positive numbers, and whatsoever such that

ai

a^, a,, b^,

any

piositive rationals

<a<

cr2

and

^i

<b<
a, b :=
J

b^.

(1)
it

When
(1),

and b are

rational, say a
,

y8,

follows from

by

121, that

^ n^

and

in every case

it

follows that
a^bi

<

a2^2-

(2)

are therefore led, when one or both of the are irrational, to define their product thus
:

We

numbers

a,

171

mean

The product of two positive numbers a and b, ivritten ab, is to that number which lies between all the numbers aibi and In other words, ab is the number defined all the numbers aobo.
ajbi

by the formula

< ab <
any

a^bo,

where

ai,

ao,

bi,

b2 denote
ai

positive rationals whatsoever

bi

such that

<a<

a.2

and

< b < bg.

It follows

from
is

1G3 that

there

is

one and but one such number

ab

for

1.

Each
There

aihi
is

less

2.

no

last aibi

than every and no

a^bifirst a-zbo-

(Compare proof,
ai,
a-z,

100, 2.)
?'2i

3.

Any

positive 5 being given,

nb2

we can choose
fri^'i

6i,

so that

(''-2

<

5.

For

a^bo
Ui,

Oi&i

b\)

bi (a2

fii),

and we can choose

&2

a-z,

&i, &2,

1^2, so that
(1) (2)

&i

and therefore

Oo (62

and ff2 ai < 5/2 ?^i, < 5/2 61) + bi (uo c/i) < 5.
ffl2

We may
ai, 02, so

make such

a clioice of

ai,

rto,

?);,

60,

as follows:
621

First take

any particular number

of the kind

as

b^',

and then choose

(3)

that

ao- ai<d/2b/.

IRRATIONAL NUMBERS
Next, using the a^ thus found, choose
62
61, ^o,

53

so that

&i<5/2a2,
62'

as in
it

(1).

Since &i

< bo' and

therefore 5/2
02

< 5/2 61,

follows from

(3)

that
(1).

i<5/2 6i,
0.

as in

Multiplication, one or both factors negative or

Let a and b

172

denote any two given positive numbers.

1.

Then

b)

and (

a) b are to

mean

ab.

2.
3.

(
a

a)

b) is to

mean

ab.
0.

and

a are to

mean

Definition of 1/a.
!, O25

Let a be any given positive number, and any positive rationals vi^liatsoever such that
i

173

<a<

Os-

Considerations like those of

tional, to define

165 lead

us,

when

a is irra-

1/a

as follows

The symbol 1/a

is to

mean

the num,ber defined

bij

the

formula

174

l/2< l/a<
where
ai, ao

l/oi,

denote

any j^ositive
Oi

rationals whatsoever such that

(7,.

<a<
is

It

follows from 163 that there

;

one and hut one such

number

1/a
1.

for

Each l/ao
There
is

is less

2.
3.

no

last

Any

positive 6

than every 1/ai, 106. l/ao and no first 1/ai. (Compare proof, being given, we can choose ai, ao so that

168, 2.)

1/ai

- l/a2<5.
if

For

l/fli
if

- l/2<5,
a2

a2- ai<5-ai2Oi,

106,117

But
tti,

ai

denote any particular number of the kind

we can choose

ao so that ai

> a/ and

fli

< 5 a/-, and

therefore

< 5airt2175

Definition

of

l/(

a).

Let a denote any given positive

is

number.

Then l/(

a)

to

mean 1/a.

54
176
Division.

A COLLEGE ALGEBRA
The quotient of a a 1 /b, that is, a

by b
1

(b

not 0)

is to

mean the

number

^b-

of a 1/b itself is known from the preceding definitions. a and b are positive, it follows from 171, 174 that we may also define a/b by the formula

The meaning

When

ai/62<a/b<a2/6i,
where
ai, a^, 6i, &2

denote any positive rationals whatsoever such that

ai

< a < a2 and

6i

< b < ftg-

177

The operations just defined are extensions of the corresponding operSubtraction continues to be the ations for rational numbers.
The commutative,
associative,

and distributive laws.

inverse of addition, and multiplication of division. Finally, additiofi and multiplication continue to conform to the commutative, associative,

and
and
c

distributive laivs.
are any three positive

Thus,
170,

if

a,

b,

numbers defined, as

in

by the formulas
!

< a < a2,

a (b

bi<h<b-2,

Ci

< c < C2,

we have
For by
166, 171, a (b
ai('>i
ffli^i

c)

=
+

ab

ac.

c)

and ab

ac are defined by the formulas

(1)

And
the

since aj (bi

+ ci)<a(b + c)<a2(62 + C2), + ttiCi < ab + ac < a2&2 + 2C2. + Ci) aibi + aiCi and a2 (62 + c^) = aobo + a2C2,
(1)

(2)
120,

numbers defined by

and

(2)

are the same.

178

The
for

rules of equality and inequality.

These also hold good

sums and products

as just defined,

namely

According as
so
is

a<, =,
a
-t-

or
or or

<, =,

also,

ac

<, =, >, =,

but

ac

or

> b, > b + c; > be, < be,

if

if c

> 0, <
0.

IRRATIONAL NUMBERS
Thus,
if

55

For
Ci

let

a < b, then a + c < b + c. d and d + a be any two rationals between a and

b,

and choose
a,
166.
(1)

so that Ci

< c < Ci +

a.

Then, since a<d and c<Ci + a, we have a + c<d + Ci + and since d + a < b and Ci < c, we have d + a + Ci < b + c, But from (1) and (2) it follows, 157, that a + c<b + c.

(2)

The proof that, if a < b and c > 0, then ac < be, is similar. But in this case we choose Ci so that Ci<c<Ci(l + a/d).

From
c

these rules

it

<

d,

then a

+c<b+

a, b, c,

d are positive,

39, that if a < b and and so on also, as in 50, when that if a < b and c < d, then ac < bd,

follows, as in
d,

179

and so on. On approximate

values.

1.

Having now defined subtraction

180

for irrational numbers, 169,

we can

state the

theorem of

162 as follows

is
2ko,

Whe7i any irrational a is given, and any jiositive rational 8 assigned, however small, we caii always find rationals, aj and

which will
For, by 162,

differ from a hy less

than

8.

we can
it

But from a<a2

that a

and a^ such that a\<a.<a<2, and a2 ! < 5. ai<a2 ai, and therefore follows, 178, that a
find ai

Oi

< 5.

In like manner, since

<

cti,

we prove

that a2

Thus,

161,

we have V2 -

1.41

<. 01 and 1.42

it

a < 5. - V2<.01.

We

say of such an a^ or a^ that

8.

represents a with an

error not exceeding

In practical reckoning we employ approximate values of 2. irrational numbers more frequently than the numbers themselves.

If !

and

hi

are approximate values of a

will be an

and b respec-

approximate value of the sum a + b. But to insure that the error of a-^ -\- h^ shall not exceed 8, we must ordinarily choose a^ and b^ so that their respective This follows from the proof in errors shall not exceed 8/2.
tively,

then

a-i

+ bi

166.

ab,

Similar rules for finding approximate values of a b, and a/b with errors not exceeding 8, may be derived from

the proofs in 168, 171, 174.

56

A COLLEGE ALGEBRA
INVOLUTION AND EVOLUTION

181

Powers.
rationals,

In the case of irrational numbers, as in that of

182

Roots.

we represent the products The wth rout of any given

aa, aaa,

by

a-, a^,

positive

number

b,

written
;

Vb,
that

is
is,

to

mean
is

that positive

number whose

Vb

to denote the positive

?7ith power is b number which is defined

by the formula

b. ( Vb)'" must show that one, and but justify this definition, we accomplish one, such number as it implies actually exists.

To

We

this as follows

183

Theorem.
tive real
1.

The real system contains number b.

??ith

the

mth

root of every posi-

If b is the
is

power of a rational number, the truth of

(2/3)3, then \/b

the theorem
Thus,
2.
if

obvious.

8/27

2/3.

power of a rational, its mth. root is that real number a which lies between all positive rationals, than b and all positive rationals, !, whose mth powers are less
If b is not the mth.
a^,

whose

??ith

powers are greater than

b.

Compare

151.

It follows

from

154 that there

since (1) every positive rational is than every a.i, and (3) there is no last a\ and no

one, and but one, such number a, either an rti or an a^, (2) each ai is less
is

first 03.

We may
If

prove (3) as follows there be a last ai, call it p.

:

Then

since

p < b,

there are rationals

Let one of them be p + 5. We have only to show that we can find a rational q>p such that g"' <p'" + S, or q'" p'" < S for we shall then have p"'<q"' < b, so that p is not the last rtj.

between

p"'

and

b.

But

q"*

-p'^

{q

- P) ((/"'-' + q"'-'^P p)?H-a./"'-i, <((/

<5,

-\

qjy"-^

+ p'""')

308
a2,

if 02'

he any particular
S/mao.'"'-^.

iiq^p +
is

We

ean show in a similar manner that there

it

no

first a^.

This established,

may

readily be proved that a

= "v b.

IRRATIONAL NUMBERS
For, since

57
171, 181
02"*.

But b
Hence
Rules

is

the only

ai<a<a2, we have ai < a < 02. number between every Ui"' and every
b,

a'"

that

is,

= V b.
Let a and
h

of equality

and inequality.

denote any

184

positive real numbers,

and

any positive

integer.

Then

According as
so is

a <, =, or
a"'

<, =,

or

> > >

b,

b"',

(1) (2)

and

Va<,
prove
(1)
if

=, or

Vb.

We may
Thus,

by repeated use

of 179.
is,

a<b, then a-

a<b

-b,

that

a^

< 6-

and so on.

We derive (2) from (1). Thus, if a Va < or > V 6, we should have a < or > b.
Rules
bers,
1.

6,

then

Va = v6

for

were

of exponents.
rn

and
a'"

Let a and b denote any two and n any two positive integers. Then

real

num-

185

a"

= a'" +
a?
a!^

2.
acta

('")''

'"".

3.

(aby
a-'^

= a^b"

urn

Thus,

aa

aaaaa

a^

+-

177

(a2)3

= =

2.a2.a2
a?>

=a2 + 2 + 2

=a2.3

byl
177

(a6)3

a6

aft

aaa bbb

a^

b^

And

similarly for

any other

positive integral values of

and

n.

A
real

theorem regarding

roots.

Let o and

denote any positive

186

numbers, and

any positive

integer.

Then

Va Vi =
For

Va/>.
(V6)'

(Va

V6)'

= V^)
(

a&

182, 185, 3

and
Hence
and therefore

(Vaby^

=
=

ab.

182

(Va
m

V5)'

(va6)'
in

m^ ^ Va v 6 = Va6.
,.

184, (2)

58

A COLLEGE ALGEBRA
VARIABLES AND LIMITS

187

Variables.
bers,

We

say that a never-ending sequence of

a I,
112,

num

such as
^35
'
' '

^H>

'

known, if the value of every particular term a is known, or can be computed, when the index n which shows its position in the sequence is given. We often have occasion to consider variables which are such given but never-ending supposed to be running through
is

given or

sequences of values.
Thus, 1, f, |, such a variable
is,

is

if

is such a given never-ending sequence, and ;-^, we suppose it to be running through this sequence,

that

to be taking successively the values i, f ,

J,

188

Limits.

As x runs through

tinually approaches the value

1,

the sequence i, , |, , it conand in such a manner that if

how small, assign any positive number, as 8, it matters not x will ult ately become and remain less the difference 1 than the number so assigned. Thus, after x reaches the 100th
we

term of the sequence, 1

will

remain

less

than

.01.

We

express

all

this

the sequence
general

i, , f,

by saying that, as x runs through And in it approaches 1 as limit. ,

189

never-ending sequence of values,

a as limit, if the difference a

variable x, xvhvsh is supposed to be running through a given is said to approach the number

x will ultimately become and

ptositioe

remain numerically

less

than every

number

that

we

may
also

assign.
it is
;

not enough that a - x become less than 5 it must to approach a as limit. the difference Thus, if X run through the sequence i, 0, f 0, f 0, - X will become less than every 5 that we can assign, but it will not

Observe that

remain

less, if

is

remain

less

than this

5,

and x

will not approacli

;

as limit.

In particular, &

may become

that

is,

may

reach

its

limit a.

IRRATIOXAL NUMBERS

59
190

To indicate that x is approaching the limit a, we write either x ^ Sl, read " x approaches a as limit," or lim x a., read " the limit of x is a."

Whether a variable x approaches a

limit or not depends

191

entirely on the character of the sequence of values through which it is supposed to be running.
Thus, while x approaches a limit when it runs through the sequence f plainly it does not approach a limit when it runs through the
)

i) f

>

sequence

1, 2, 3, 4,

-,

or the sequence

1, 2, 1, 2,

Hence the importance

Theorem
1.

of the following theorems

the other hand, remains always less than it appjvaches a limit. And this limit

If the varialde x continually increases, but, on some given ?iumber c,

is

192

either c

or some

number which

is

less

than

c.

For by hypothesis there are numbers which x will never Assign all such numbers to a class R^, and all other numbers, that is, all numbers which x will ultimately exceed,
exceed.
to a class
^'^
i?i.

thus obtain a separation of the entire system of real numbers into two parts, Ri, R^, so related that each number
in Ri
is less than every number in T?.,Obviously there is no last number in Ri.

We

Hence,
a.

160,

there

is

first

number

in R^.

Call this

number

As x

it will approach a as limit. For however small 8 may be, if only positive, a S belongs to the class of numbers i?i, which x will ultimately exceed. Hence x will ultimately remain between a 8 and a, and therefore differ from a by less than 8.

increases,

In the same manner

the variable

it

may

be demonstrated that

x continually decreases, but, on the other hand, If remains always greater than some given number c, it approaches a limit. And this limit is either c or some number tvhich is
greater than
c.

193

60
194

A COLLEGE ALGEBRA

It is not necessary, however, that x Regular sequences. should always increase or always decrease, if it is to approach a limit.

Thus, X is sometimes increasing and sometimes decreasing, as it runs J, |, 5, 15, but it approaches as hmit. through the sequence

shall prove that x will or will not approach a limit, -, a, through according as the sequence of values a^, a^, which it runs, has or has not the character described in the

We

following definition

195

The sequence
eve7'i/

ai, aj,

, a,

is

said

to be regular,

if for

positive test number 8 that may be assigned a corresponding term a^ can he found, which ivill differ numerically from

every subsequent term by

less

than

8.

Thus, the sequence

1.4, 1.41, 1.414,

(1), 161, is

regular.

For the difference between the first term, 1.4, and every subsequent term is less than 1/10; that between the second term, 1.41, and every subsequent term is less than 1 / 10- that between the nth term and every subsequent term is less than 1/10". Now, however small 5 may be, we can give n a value which will make 1/10" smaller still and if k denote such a value of n, the kth term of will differ from every subsequent term by less than S. 1.4, 1.41, Thus, if we assign the value 1/500000 to 5, we have 1 /lO^ < 5, so that the sixth term of 1.4, 1.41, will differ from every subsequent term by
; ;

less
2.

than this value of

5.

The following sequences are

1

also regular

'

J 5

/0\

17...
1,

{0\

-h

-h

tV,

(4)
is

2,

1,

1,

(5)

Observe that

in (2)

a lesser term, in (4) We sometimes encounter regular sequences like (5), all of whose terms after a certain one are the same. Evidently a variable which runs through such a sequence will ultimately become constant, that is, will
reach
3.
its limit.

followed by a greater term, in (3) by sometimes by a greater term, sometimes by a lesser.

each term

The following sequences are not regular

1,

2,

3,

4,

...,

(6)

I, i,

i,

i, ..-.

(7)

IRRATIONAL NUMBERS
For
in (6) the difference

61
'*

between a term and a subsequent one may always be' indefinitely great, and in (7) it may always be |, and therefore not less than every number, I for instance, that we can assign.

Formulas
relation

for regular sequences.

a^.

1.

We may

indicate

the
a^,

196

between the term

and every subsequent term,

by the formula,
\ap
2.
vi^ill

63

af.\<8
a,,

for every

p>

k.

(1)

Again, since any terms a^ there

lie

between
a^.

lie

between
a^
3.

8
ttp

may be which are > a^. and a^ + 8, and any which are < a^. will and a,., we may also write

Z<

<

Qj.

for every
if

p>

k.

(2)

It follows

from

(2) that

some of the terms a^

8.

are less,

and some are greater than

these terms

a^,

the difference between two of

may

exceed

8,

but not 2

But we can always find a term, a which corresponds to 8/2 as a^. corresponds to 8. The difference between every two terms after , will then be numerically less than 2(8/2), or 8 that is, the relation between every two of these terms
;

will be that indicated

by the formula
^
foi'

|p
Theorem
to

^.jI

<

every

>q>

I.

(3)

2.

The

variable

x vill ajyjyrnach a limit if the

197

sequence of values

aj, ao,

a^,

through tvhich

it is

supposed

run, is a regular sequence.

For there are numbers to whose right x will ultimately remain as it rnns through the sequence a^, ^2, , ,. (1)

Thus,

if S

and ak have the meanings above explained, x

will

remain

to

the right of

a<.

5 after it

reaches the value

a^.,

196

(2).

Assign
bers

all

that

such numbers to a
all

class,

Tii,

and

all

other

num-

is,

numbers

to

whose right x

will not remain

to a class, R^.

62

A COLLEGE ALGEBRA
thus obtain a separation of the entire system of real into two parts, Ri, R^, so related that each number
less
a,
if

We
in Ri

numbers
is

than every number in R^.

By

160, a definite

number,
Thus,
constitute

exists at

which

this separation occurs.

i?i,

the sequence be \, J, \, ^^, , the negative rationals and the positive rationals, Eo. and a itself is 0. but
;

As X runs through number a as limit.

Since (1) such that
small.
is

the sequence (1),

test

it

will approach this

For assign any positive

number,

S, it

matters not
a,,

how
(3),

regular,

we can

find a term,

196

[ttp

But since a

Og] < S/2 8/2 belongs

for every
to
/?i, all

2^>

q> m.
+

(2)

the values of x after a

lie to the right of a 8/2 8/2. And since a belongs to i?2) among these values there will be some after a, which lie to the left of a 8/2 for otherwise a 8/2 would

certain one will

belong to
Thus,
if

i?i,

since x

would ultimately remain to

its right. jL^ all values of

is,

the sequence be
Jg, lie

|,

i,

|,

j-\,

and

X after the /oitriA, - 25 and Jo-

between a

5/2 and
x.

5/2, that

between

Let a^ denote such a value of

Then

a-S/2<<<a + 8/2,
or

|a-<|<8/2.
From
(2)

(3)

and
1^

(3), since y'

>

m,

it

follows, 78, 178, that

<^;,|

<

for every

>

</'.

In other words, after x reaches the value a^ the difference a X remains numerically less than 8.
Therefore x approaches a as limit,
198
Conversely, if

189.
a,

x
,

is

approaching a

limit,

the sequence of

values

ai, ao,

a^,

through ivhich

it is

supposed

to

run,

must be regxdar.

IRRATIONAL NUMBERS
For since the difference
remain numerically

a.

63

will ultimately

become and
8,

less
a/,

than every assigned positive number,

so that

189,
1^

we can choose
''^I

<

S/2 and

|a

a^,|

<
a^.|

8/2

for every

p>
j^

k;
k.

whence

[a^
aj, a^,

<

S
is

for every

>

Hence the sequence

regular,

196

(1).

We may
a limit

combine 197, 198

in the single statement:

The sufficient and necessary condition that a variable approach is that the sequence of values through which it is supposed to run he a regular sequence.

199

SOME IMPORTANT THEOREMS REGARDING LIMITS

In the present section a and h will denote any given real numbers, and x and y variables which are supposed to run
through given never-ending sequences of values. The limit 0. From the definition of limit, 189,
d lately follows that
1. If the variable x will ultimately become and remain numerically less than every positive number, 8, that may be and conversely. as limit assigned, then x approaches
;

it

imme-

200

2.

If X approaches a as limit, then a

x approaches

as

limit;

and conversely.
the limit 0, as
it

Thus X approaches
and
1

X approaches the limit 0, as x runs tlu'ough the sequence

variable
1
.

runs through the sequence \t\i\-,"-',

j,

|j

whose limit
yi

is

is

called an infinitesimal.

Theorem
ically less

and A and B remain numerIf than some fixed number^ c, as x and y vary, then

and y

= 0,

201

Ax

-V

By = 0.
positive

For assign any

Since X

number,

5, it

matters not

how

small.

= 0, X will ultimately remain numerically

<S/2c.

200,1

64
Since y

A COLLEGE ALGEBRA

= Q,

will ultimately

remain numerically

< 5/2 c.

200, 1
..

Hence

Ax + By

will ultimately

remain numerically
1.

< 2c

2c

<

5,

and therefore approaches and y Thus, if X

as limit, 200,

= 0,
may
and

then (xy

3)

?/

0.

202

Note.
variables.

This theorem

readily be extended to

any finite number

of

Thus,

if

= 0, =
2/

= 0,

then

Ax + By + Cz:^ 0.

203

limit of the sum, difference, product, quotient of two variables ivhich approach limits is the sum, difference,
2.

Theorem

The

product, quotient of these limits : that the limits a and b respectively, then
1. 2.
a-

is,

if x

and y approach

+y= + :^ a
ft 1/

5.

3.

xi/==ab.

b.

4.

x/ij

a /b, unless

= 0.

For, since a - x

and

- yz^O,

200, it

follows from

201 that
(1)

A{a-x) + B(b-y) = 0.
The formulas
1. 1, 2, 3,

4
{X

may

be derived from

(1).
.:

Thus,

a
is,

y)

that
2.

a
is,

{X

y)

= (a- X) + (b - y) + y = a + 6. = (a x) {b y)
~
y

i 0,
^ 0,
0,

by(i)

.-.

that
3.

x
ab

=a
-\-

b.

xy

{a

x)b

{b

y)x

.-.

that

is,

xy ==
x

ab.

'
that

= /a

x\

/x

x\

\1

,,

b-y
is,

{b-b)'-{b-y)

^''-^^b-^'-'^by--'

x/y :^a/b.
If
3.
:s.

204
205

Corollary.

=
is

a,

then x"

= a".
is,

Theorem

The

limit of the nth root of a variable which

the nth root of that limit: that

approaches a limit

If X

= a,

then

V x =L Va.

IRRATIONAL NUMBERS
1,

65

When

a
0,

Since x ==

0. Assign any positive number, S. x will ultimately remain numerically

< 5".

200, 1

Hence

V x will
a
is

ultimately remain numerically

< 5.

184

Therefore
2.
is

Vx==0.
notO.
It follows

When

from a

later section,

200, 1 a 308, that x

always exactly divisible by Vx Va, and that the quotient Q does not approach the limit when x =L a. It therefore follows from 203 (1), by setting A = 1/ Q and B = 0, that

Vx - V^ =

(X

a) /

Q = 0,

that

is,

V2

Va.

200,

RELATION OF THE IRRATIONAL NUMBERS TO MEASUREMENT

Length
line
of a line

segment incommensurable with the

unit.

If a

206

segment S be incommensurable with the unit segment s,that is, if, as when S and s are diagonal and side of the same
square,
is

we can prove that no aliquot part of s, however small, the definition of length given in contained in S exactly 130 does not apply to S.

But there

is

then a definite irrational number,

:

a,

which
s fall

stands in the following relation to S The segments which are commensurable with the unit
into

two distinct classes, those which are less than S and those which are greater than S. The rational nimibers which are their lengths, 130, fall into two corresponding classes, which we may call Ay and JjEvery positive rational belongs either to Ai or to
A^, each

number

in A^ precedes every number in A 2, and, finally, there is neither a last number in A^ nor a first in A^*

lies

There is, then, 154, a definite irrational number, between all numbers in Ai and all in A 2. We

a,

which

call this

* For were there a last number in ^,, then among the segments commensurable with s and less than S there would be a greatest, say S'. But no such seffniont exists. For according to the Axiom of Archimedes, explained in the following footnote, we could find an aliquot part of s which is less than S S' and the sum of S' and this part of s would be commensurable
;

with

s, less

than S and greater than

S'.

66

A COLLEGE ALGEBRA
a the length of S.
:

number
207
s,

We
S,
a,

therefore have the following

definition

The length of any segment,

is

incommensurable with the unit,

ivhich lies between all rationals
all rationals

that irrational number,

which are lengths of segments less than S and which are lengths of segments greater than S.
Thus,

Vi

is

the length of the diagonal of a square in terms of the side.

208 209

If the length of S in terms of s is a, that whether a is rational or irrational.

we

write S

as,

and
134,

Real numbers pictured by points.

As

in the figure of

as origin; also take any right line and on it a fixed point some convenient unit, s, for measuring lengths. And by the of any point P of the line, understand the distance from

length of the segment

OP

in

terms of

s,

130, 207.

choose as the picture of any given number, a, that point, P, is the tiumerical value of a., of the line tchose distance from the point being taken to the right or left of 0, according as a is
positive or negative.
is a rational number, we can actually construct P, 134. the contrary, if a is irrational, we usually cannot conthen assume that P exists, in other words, that struct P.

We

If a

On

We

on the line there

is

a single point, P, lying between

less

all

points

which picture rationals

* This

than a and

all

which picture

rationals greater than a.*

is

we may mention tlie followinsj because measurement now under consideration.

1.

not the place for a discussion of the axioms of geometry; hut of their relation to the subject of
of

Axiom
ice

Archimedes.
ji ml

If s and S denote two line segments such that

intPfier, ni,

S<S,
2.

can always

an

such that

ms>S.

of continuity. //" all the points of a right line be separated into two classes, K, and K.,, siirh tliat earli point in R, //>.s" to the left of every point MiR there is either a last point in R, orajirst in R,. (i) The Axiom of Arcliimedcs is involved in the assumption tliat every For the first steji in measurinuj S in terms of line segment can be measured. is. S is to find an iiitefjer, )/;, such tliat (i 1) (2) The axioms 1 and 2 cnnlile us to prove the assumption in 209 that for

Axiom

s<S<

every given irrational a there exists a corresponding point, P. For a separates the rational system into two parts, which we may name B and C respectively. Call the points corresponding to the numbers in each the

IRRATIONAL NUMBERS

67
210

Conversely, when P is given, we can find a, at least approxi sign to mately, by measuring OP and attaching the + or the result, according as P is to the right or left of O. Thus, if P is to the right of O, and we can lay s along OP Jive times, the tenth part of s along the part left over seven times, and the hundredth
part of s along the part still left over six times, then 5.76 will be the value of a to the second place of decimals.

ence, 2,

In this manner we set up a relation of one-to-one correspondbetween all the real numbers and all points on the
;

211

line

and

if a

and b denote any two

real

numbers, and

P and Q

the corresponding points, P will lie to the left or right of Q according as a is less or greater than b.
Thus, if a and b are positive and a < b, and if c denote a rational lying between a and b, and R the corresponding point, we have, 206,

OP<OR

and

OR<OQ

and therefore

OP<OQ.

are to show that there is in the JS-points and the C'-points respectively. line a definite point, P, which separates all the iJ-points from all the C'-points. First assign the ii-points and, all intermediate points to a class i?, and all denote the point which this points to the right of these to a class R,, and let separation defines, by 2. Next assign the C'-points and all intermediate points to a class S^ and all points to their left to a class S^, and let Q denote the point which this separation defines, by 2. The points F and Q must coincide. For if not, let denote the line segment between them. By 1, we can find an integer, m, such that

We

PQ

mPQ>s,
this is impossible. c such that to b and c respectively, we

and therefore
select

PQ>s/m.

But

For we can
have

number

cb<l/m. And if L

from B a number b and from C be the points corresponding and

LM<s/m, and PQ<LM, and therefore PQ<s/m. It is this one point, P or Q, that corresponds to a according to 209.
{?>) Finally, observe that corresponding to 2 the system of real numbers has the property described in KiO, and corresponding to 1 the property //'a and b are any two positive real numbers, we can always find an integer,
:

m, such (hat mb>a. For, by 108, 176,

therefore wib > a.

178,

we can choose an integer, m, such that ?n>a/b and

at least not without a real system would not possess this property sacrifice of some of its other properties were we to invent more than one irrational for a separation of the rational system of the kind described in 154. Thus, if every rational is either an , or an a, and Oj<b<c<02 for every a^, 178 and proof of 1()3. ctj, fflj! we should have c-'b<a^ But however small a positive number, S, we might assign, we could find no b)>S. integer, m, so great that )/((c For it would then follow that c b>6/m, which is impossible since

The

b<a2 Oj and we

can choose

a^, a^

so that

a^a^K^/m,

68
212
Theorem.

A COLLEGE ALGEBRA
If the length of S in terms of T is a, and that of is b, then the length of S in terms of s is ab.

in terms of s
1.

When

and b are
and b

rational.

Let a
Since
itself

= a/b

c/cZ,

where

a, 6, c,

S contains

the 6th part of

cZ denote integers. a times, 130, 6S will contain

T
(1)

a times, that

is,

Similarly

6S = dT =

aT.
cs.

(2)

But from

(1)

and

(2) it

readily follows that

hdS

atZT,

and adT

acs,

and therefore
That
2.
is,

6dS

acs.

the length of

in

terms of

s is

(XC

(X

or
b

hd

130

Wlien

and

b,

one or both, are irrational.

thai,

Let Si and S2 denote any segments commensurable with T, such

Si

< S < S2,

in

and

let ai, 02

be the lengths of Si, S2

Si

terms of T, so that

aiT and S2

2T, where rti<a<ff2s,

208
such

Similarly, let Ti,

T2 denote any segments commensurable with

that

Ti<T<T2,
and
let 61, 62

be the lengths of Ti, T2 in terms of

s,

so that

Ti

61S, .and

T2 =

ios,

where 6i<b<62-

Then

since

Si
1,

OiT, and

T>Ti, and Ti = M,
Si > ai6is.

we

have, by case
Similarly

S2 < ajb^s.
fli^is

Hence
and therefore

< Si < S < S2 < 02628,

ai6is

< S < 02628.

all

We
and

have thus demonstrated that

the

numbers

a 161 and 0262 are

lengths, in terms of s, of segments respectively less and greater than S. Therefore ab, the one number which lies between all the numbers ai6i
0262, 171, is the length of

itself in

terms of

s,

207.

/
IRRATIONAL NUMBERS
Corollary.
b respectioely,

69
213

If the lengths of S and T in terms of s are a and then the length of S in terms ofT is a/b.

For
jf s
is

let

Then

the length of S in terms of T be x. since the length of S in terms of T is x, and that of b, the length of S in terms of s is xb, 212.

in

terms

But, by hypothesis, the length of

in
a,

terms of

s is a.

Hence
and therefore

xb

x = a/b.
variable.

The continuous
is

One

of our

most familiar intuitions

214

that of continuous motion^

Suppose the point

along the line

to be

moving continuously from A

the origin.

to

OAB\ and let a, x, and OA, OP, and OB respectively, O being

b denote the lengths of

According to the assumption of 209, the segment AB contains a point for every number between a and b, through which, of course, P must pass in its motion from ^4 to B. This leads us to say that as P moVes continuously from A to B, x increases
or that X varies continuously

from the value a to the value b through all intermediate values, from a to b.
Of course
it is

impossible actually to trace the variation of this x, since its values there is no next following value. If we attempt to reason about x mathematically, we must content ourselves with defining it thus (1) x may take every given value between a and b,
to

any given one of

and (2) if p and q denote any given pair of these values, and p < 5, then X will take the value p before it takes the value q. We may add that x is
often called a continuous variable
is

when only

the

first

of these properties

attached to
Ratio.

it.

Let

and

By the measure of M in terms of N, or the ratio of M to iV, we mean the very same numbers which we have defined as lengths in 81, 130, 207, when M and N denote
same kind.
line segments.

denote any two magnitudes of the

215

70

A COLLEGE ALGEBRA

Hence the theorems of 212, 213 regarding lengths hold good for the measures or ratios of any magnitudes of the same In particular, kind.
216

If the measures of

and

in terms
to

of the same
a./^).

tinit

are a

and

b respectively, the ratio

o/M

is

V.

THE IMAGINARY AND COMPLEX NUMBERS

PURE IMAGINARIES

217

The real system does not contain the even roots of negative numbers for the even powers of all real numbers are positive. Thus the real system does not contain the square root of 1. To meet this difficulty, we invent a new system of signs
;

called

imaginary or complex numbers.

218

The simplest of these new signs is i, called the unit of With this unit and the real numbers, a, we form imaginaries. signs like ai, which we then regard as arranged in the order We in which their " coefficients," a, occur in the real system.
thus obtain a

new continuous

ordinal system of "numbers,"

which we

call 2>M^e imaginaries.

Thus, proceeding as when developing the real system,

we may

rirst

form the complete

scale of imaginaries

i,

i,

i,

0,

i,

z,

i,

then enlarge this into a

fractional coefficients,

system by introducing imaginaries with and finally into a, continuous system by introducing
deji.se

imaginaries with irrational coefficients. Here 2 i is merely the name of one of our

new numbers.

Its

only

property

is

a definite position in the

have defined multiplication, we 2 X J or i X 2. Similarly every pure imaginary

But when we ordinal system. shall see that 2 i also represents the product

new

ai.

Hence we write for i. i as 0. In particular we shall define Observe that is the only number which is common to the real system
and the system
of pure imaginaries.

IMAGINARY AND COMPLEX NUMBERS

71
219

addition and midti^jUcation.

equations
1.

For these new numbers we invent operations which we call They are defined by the following "
:

ai

+ bi = (a

-\-

b)i.
3.

2.

a-bi

= bi-a = abi.

ai -bi = ab.

Thus, 3, the product of two pure imaginaries, ai and bi, is to mean the real number, ab, obtained by multiplying the

coefficients of ai

and

bi together

and changing the sign of

the result.

We

define

power as

in 136.

Thus, (aiy

= ai

ai.

The system of pure imaginaries contains the square all negative numbers in the real system, namely :

roots

of

220

V 1 =
For
i'^

and
i-i

V =
a'-^

ai.

= \i-\i=-l.

219,3

The refore, i is a square r oot of 1, 138. We indicate this root by V 1, and thus have i = V 1. In like manner, it may be shown that i is a square root of 1. We indicate this root by V 1.
Similarly, since (ai)2

ai- ai

a^,

we have

ai

a^.

COMPLEX NUMBERS
To secure a number system which will contain the higher even roots of negative numbers, we invent comjilex numbers. These are expressions like a + bi, formed by connecting a real
number,
a,

221

with a pm-e imaginary,

bi,

by the sign +.

They

are also often called imaginary numbers.

Until addition has been defined for complex numbers the expression 6z is to be regarded as a single symbol and the sign + as merely a

part of this symbol.

Since

<z

and

bi

bi,

real

numbers and pure

222

imaginaries are included

among the complex numbers.

72
223

A COLLEGE ALGEBRA

We regard the complex numbers as arranged in rows and columns in such a manner that all numbers a + bl which have the same b lie in the same row and are arranged in this row from left to right in the order of their a's while all numbers which have the same a lie in the same column and are arranged in this column from below upward in the order of
;

their

6's.

And we may
its

ber defined by

consider any particular complex num" two-dimensional ordinal position in this

arrangement." In 238 we shall explain a method of picturing this arrange-

ment

for all values of a

for integral values of a

and b. and b.

We may indicate it as follows

-2 + 2i -2 + -2 2 i

-I + 2i -1 + -1 1 i
i

2i
i

1 l

+ +

2i
i

+2i

2+i
2
2

-2-2i
This arrangement

-l-2i
may

-2i
i

l-2i

2-2i

also be described as

whose elements are rows (or columns), each system of signs of the form a + bi.

of

an ordinal system, 17, which is itself an ordinal

224

Definition of equality.
erjual

Two complex numbers

are said to be

the same position in the two-dimensional ordinal arrangement jiist described. Pleuce,

when they occupy

225

If a

bi

=c

-f di,

In particular, iia+bi

then a = c and b = d = 0, then = and i =

and conversely. and conversely.

Of two unequal complex numbers, like 2 + Hi and 8 + i, we cannot that is, precedes or follows say that the one is less or yreater than the other, since complex numbers do not form a simple ordinal system.

226

Definitions of addition, subtraction, multiplication.

dlffereyice,

The sum^

are to

and product of two complex numbers a + hl^ c + di, mean the complex numbers which form the second
of the following equations
:

members

IMAGINARY AND COMPLEX NUMBERS

1. 2. 3.

73

(a

+ ^-0 + (c + (U) =.(a + c) + (b + ^i. - (c + dl) = (a -c) + (b -d) {a + hi) = (ac hd) + (ac? + he) (a + ^0 (c + di)
i.

i.

According to 1 and 2, addition and subtraction are inverse operations. In particular, by 1, (a + i) + (0 + 6i) = ( + 0) + (0 + 6) z = a + f'i that of a and hi, according to the definition 1. is, a + hi is the sum These definitions are in agreement with tlie commutative, associative,
;

and

distributive laws.

In fact,

we

arrive at

them by combining these

laws with definitions previously given.

(a

Thus,
hi) c
-\-

hi) (c

di)

{a

-\-

{a

hi)

di di

+ hi c + a di + hi = (ac hd) .+. {ad + be) i,

ac

since

i^

1.

Corollary.

A product
+
&i) (0

vanishes tvhen a factor vanishes.

(

227

For

(a

+ ^) =

0)

(a

+
-\-

0)

0.

Division.

We

define the quotient oi a

hi

hj

-{-

di as the

228

complex number which multiplied hy c + di will give a + hi. When c + di is not 0, there is one and but one such number,
namely, that in the second member of the equation

a
c

+ hi ^ ac + hd he ad "^ + + di c' + d'

6-2

rf2

But when

+ di

is 0,

no determinate quotient
of 220.

exists.

For the product of the right member of a + hi, as the reader may easily verify by aid

this equation

by

di

is

We
If

discover that this

number

is

the quotient as follows

c

a number exists which multiplied by

+
a
a
h.

di will give a

hi, let it

be

Then
or
{ex

(x

+
-\-

yi) {c

+ +

di)

= = =

+ +

hi.

(1)
(2)

- dy)
=

{dx

cy)i
cij

hi.

and therefore ex

dy

a and dx

(3)

225

Solving this pair of equations for

x and

y,
.

we

obtain
.,

ac

hd
5

he
c2

c-2-i-d^

ad
J

unless c^

+
,

a^

j.>

a 0.

h\
(4)

tZ^

74
Moreover, since

A COLLEGE ALGEBRA
(4)
is

satisfy (o), the corresponding multiplied by c + di will give

It is

the only pair of values of x and y which will number x -\- yi is the only number which

l>i-

c^ -\- d- = our definition of quotient and d 0, since otherBut if c- + d^ = 0, both c = meaningless. and wise we should have a positive number equal to 0. And if c = d = 0, the divisor c + di is 0.

evident from

(4)

that

when

is

229

The commutative, associative, and distributive laws. The operations just defined evidently include the corresponding operaLike the latter, thei/ conform to the tions with real numbers.
commutative, associative, and distributive
Thus,
(a
(6

laivs.

a'i) (b
b'i)

b'i)

ah

a'b' b'a'

+ +

{aV
(h'a

a'h)
ba')

i,

(]

and

+
+

{a

+
+

a'i)

ba(2)
{b

i.

(2)

But the second members

Hence
(a

of (1)

and

are equal,
b'i)

by

177.

a'i) (b

b'i)

{a

a'i).

And
230

the remaining laws

of equality.
1.
2.

may

be established similarly,

Rules

Let
a

a, b, c

denote any complex numbers.

If

If If If a

3. 4.
1.

= b, a -f c = b + a = b, ac = be,
a

then a
c,

then then

+c=b+ a = b.
ac

c.

= be.
= b,
=c+
b'.

then
and
b'i,

a
c

unless c
c'L

= 0.

For

let

a'i,

If

a
a
a
{a

= b + b'i, + a'i = b +
and

then

= =

b
b

a'

225

Hence and therefore

that
2.
is,

+
c)

+ c and a' + c' = b' + c', + {a' + c')i = {b + c) + {b' + c')i, a + c = b + 0.

c

178

225 226

If

a a

c
c)

we have
and therefore
3 and
4.

c-|-(
-'

=b+ =b+ = b.

c,

+ (

c),

byl

226

The proofs

of these rules are similar to those of

and 2

respectively.

IMAGINARY AND COMPLEX NUMBERS

Corollary.

75
231

If a product vanish, owe of

230, 4,

its

factors must vanish.

This follows from

by the reasoning

of 76.

Absolute value of a complex number.

The

positive real

number

232

Va'-

+
is

6^ is called

the absolute or numerical value of a

+ bi

and

represented by \a

bi\.

Hence, by definition,

Thus,

|2 4-

il

= V4 +

^ Vs.

definition of numerical value reduces to that already For a geometrical interpretation of this given for real numbers, 63. definition see 239.
6
0, this

When

We

also say of

two complex numbers that the

to,

first is

numer-

233

or greater than the second, according as the absolute value of the first is less than, equal to, or
ically less than, equal

greater than that of the second.

Thus, 2

For

|2

+ 3 is numerically greater than 3 + + 3i| = Vl3 and [3 + i| = VlO, and Vl3>VlO.

i
i.

Theorem

The

absolute value of a product of two complex

234

members

is eqxial to

the product of their absolute va'lues.

Let the numbers he a

Since

= ah = =

+ a'i and ab - a'b' +

a

b + b'i. {ab' + a'b)

h

i,

226

we have

|ab|

V(a6

a'6')-

(^'

+
it

a'b)^-

232

But on carrying out the indicated operations

{ab

will be

found that

a'b'Y
^

+
+

{ab'
{(ib'

+ +

a'b^
a'by^

=
=

(a^

a'^) (62

h"^).

Hence V(a6 That

is,

a'b')

= Va^ +
|a|-|b|.

a'2

Vt^+b^.

186

|ab]

Theorem

2.

The

number's cannot exceed the

absolute value of a sum of two complex smn of their absolute values.

as in 234.
b')^

235

Employ
Then
if

the

same notation

Va2 +
a'2

a2

-j-

62

+ 6'2^ V(a + 6)2 + (a' + + 2 V(a2 + a'2) (62 + 6'2) > a2 + 62 + a'2 + 6'2 + 2 {ab + a'b')
a'^

Vfts

(1)

6'2

184

76
.-.if
if

A COLLEGE ALGEBRA
V(a2
a262

.-.

a'W^

+ a'2) (62 + 6'^) ^ab + a'b' + a26'2 + o'262 > aW + a'26'2 +

a26'2

178
2 a6a'6'
(2)
is

184 178

.-.

if

a'262

> 2 aha'h'

/.if

{a6'-a'6)2>0.
(2) is

178
positive

But
(or 0).

always true since the square of every real number which proves our theoi'em. Hence (1) is always true

Thus,

[2

+
+

i|

= V5 +
1.

and

|1

3i|

= V]L0.
Vs + VlO.
of a

But

(2

i)

(1

3
1

5,

and 5 <

236
(a

Powers and

roots.

The

?ith

lyoxver

hi,

written

hi)'",

is

is a + hi. complex number, as

the product of n factors each ot which It follows from 226, 3, that this product is a
to
c

mean

di.

185, that the laws of exponents hold good for powers of complex numbers as thus defined. If {a + hiy = c + di, we call a + hi an 7ith root of c + di, 2.

It

may

be proved, as in

and we may indicate

it

thus,

di.

We

shall
c

number

prove subsequently that every given complex di has n such wth roots in other words, that in
:

the system of complex numbers there are n different numbers

whose wth powers equal

+ di.
1 are

- 1/2 = f, the number Thus, since (l/Vi + i/v^)- = 1/2 + 2 i/2 I/V2 + i/Vi is a square root of i, and therefore a fourth root of 1.
The remaining
three fourth roots of

-\/V2 +
237 system
is

i/V2,

I/V2 -i/V2,

-I/V2 -i/V2.
number

General conclusion.

'No further enlargement of the

For, as has just been pointed out, 226, 236, the complex system meets all the requirements of the four fundamental operations and evolution. And while certain
necessary.

other operations with numbers have a place in mathematics, among them the operation of finding logarithms of numbers, these operations admit of definition by infinite series, 140,

if

like

III

-f M2

+ W3 +

wliose terms are complex

numbers

and

such a series have a sum, that

sum

is

a complex number.

IMAGINARY AND COMPLEX NUMBERS

GRAPHICAL REPRESENTATION OF COMPLEX NUMBERS

77

Complex numbers may be pictured by points in a plane, the points being called the graphs of the corresponding numbers.
Take any two
the origin
s for
1.
;

238

also

right lines X'OX, Y'OY intersecting at right angles at some fixed unit segment

measuring lengths. We represent each real number, a, by that point A on X'OX whose distance from 0, in terms of s, is |a|, 209, taking A to the right or left of O, according as a
is

positive or negative.
2.

We

represent each 'pure imaginary,

by that point B on Y'OY whose disB above or tance from is \h\, taking below O, according as 6 is positive or
hi,

negative.

A
in

represent each complex number by the point P, which is obtained by the following construction Find A and B, the graphs of a and 6i, as in 1 and 2. Then through and B draw parallels to Y'OY and X'OX respectively. The point T which these lines meet is the graph of a + bi.
3.

We

6t

We
By

call

X'OX

the azis of real numbers

and

Y'OY

the axis of

pun

imaginaries.

this

method we bring the system

of complex

numbers

into a relation of one-to-one correspondence, 2, with the assemMoreover we obtain a complete blage of all points in a plane.

representation of the two-dimensional ordinal character of the complex system, 223.

Observe that the graphs of all numbers which have the same imaginary lie on the same parallel to X'OX, and that the graphs of all numbers which have the same real part lie on the same parallel to Y'OY.

part

The
its

graph from

absolute value of any complex the origin.

number

is

the distance of

239

For since the lengths of

respectively, the length of

OA and AT in the figure of 238 OT is VoM^ or |a + 6i|, 232.

are a

and

78
240
a

A COLLEGE ALGEBRA
of the siim

The graphs

a'i,

=^+

b'i,

may

and product of two complex numbers be found as follows

:

Given

and B,

the graphs of

a and b

respectively.

Join

OA

and
is

OB
the

and complete gram OACB.

graph 0/ a

the

parallelo-

Then C

b.

For, draw the perpendiculars

BD,
a,
a',

AE, CF, AG.

h, b'

Then

are the

lengths of OE,
respectively,
gles

EA, OD,

DB

ODB
a
a'

and the trianand AGO are

congruent.

OF = OE + EF = OE + OD = EC = FG + GC = EA + DB =
Therefore

+ b. + b\

in length,

in length.

When
drawing
Since

AC OC<OA + AC,

the graph of a + 6 + (a'+ b')i, or a -|- b, 220, 1. are in a straight line (and always) C may be found by 0, A, equal in length and direction to OB.
is

The graph

of the difference a

i.e.<OA+ OB, we have |a b is that of the sum

-|-

b|<|a|

|bl.

+ (

b).

Given A and B, the graphs of a and b, and let I denote the graph of 1. Join OA, OB, lA, and on OB construct the triangle OBC similar to OIA and such that were OB turned about O
until
it

lay along

OX, OC would
is

lie

along OA.

Then C

the

graph of ab.

This rule will be proved later and

rules derived

from

it

for constructing

graphs of quotients and porvers. When b = i, OC is OA turned 90 " counter-clockwise " about 0.

241

It follows that identical rela-

tions

among complex numbers

be translated into geometrical theorems.

may

Hence

imagi-

nary numbers

may

express relations

among

real things.

- a)/ 2 shows that the diagonals Thus, the identity (a + b)/2 = a + (b of a parallelogram bisect each other; for the graphs of (a + b)/2 and in tlie first figure, 240. a + (b - a)/2 are the midpoints of OC and

AB

PART SECOND ALGEBRA

I.

PRELIMINARY CONSIDERATIONS

ON THE USE OF LETTERS TO DENOTE NUMBERS

In algebra a letter is often used to Constants and variables. denote any numher whatsoever. Thus, in the formula ab = ba the letters a and b denote every two numbers, the meaning of the formula being the product of any first number by any second
:

242

number

the same as the product of the second by the first. In many algebraic discussions we find it convenient to make
is

the following distinction between two letters which have this meaning, as between the letters b and x in the expression x -{-b. First. regard the one, b, as having had a paHicular

We

value, but
it is

we

any that we please, assigned it at the outset, which Such a letter then to retain throughout the discussion. shall call a known letter or number, or a constant.

Second.

On

regard the other,

to

the contrary, throughout the discussion we x, as free to take every possible value and

change from any one value to any other.

Such a

letter

we
243

shall call a variable.

Unknown
letter

letters.

But

letters are also

to

employed
be found.

to denote

particular numbers whose values are

Such a

we

shall call

an unknown letter or number.

are not at liberty to assign any value we please to an unknown letter, as we are to a constant or variable letter.
Thus,
in the

We

equation 2 x

0,

is

an unknown

letter

whose value

In the expression 2x readily find to be 5/2. any value we please, but in the equation 2x 5 =

we

we may assign x we can assign x no

5

other value than 5/2.

79

80
244 The

A COLLEGE ALGEBRA
choice of letters.
is

The only necessary restriction on onr that no single letter be made to represent two different numbers at the same time.
choice of letters

It is customary, however, to represent known numbers or constants by the earlier letters of the alphabet, as a, b, c unknown numbers and variables by the later letters, as x, y, z.

Besides simple letters

we sometimes use
a',

letters

affected

with accents or subscripts: thus, a second," "a third"; and a^^,

a sub-one," " a sub-two."

a", a'"

read

"a prime,"

a^,

a^

read "a. sub-null,"

245

On reckoning with
letters, as a,
h, c,

letters.

we can

When we represent numbers by only indicate the results of combining

them by the operations of arithmetic. Thus, to add b to a merely means to form the expression a + b, which we therefore call the sum of a and b. Similarly, the product of a by
Z

is

the expression ab. Inasmuch as the literal expressions thus obtained denote numbers, we may reckon with them by the operations of

arithmetic.

But

in

such reckonings we cannot make any

direct use of the values of the expressions, since these are

can merely connect the expressions by the the form of appropriate signs of operation and then simplify the result by changes which we know will not affect its value,
not given.
no matter what this

We

may

be.

Now, as we have seen, 68, all the changes that can be made in the forms of sums and products without affecting
their values are
1.

embodied

in the following

formulas

3.

a + (b + c) = (a + b)+c. 2. + b = b^a. (ab) a (b + c)= ab + ac. 5. 4. a (be) ab = ba.

c.

It may therefore be said that the formulas 1-5 are practi-^ of addition and multiplication that we cally all the definition either need or can use when combining literal expressions and
;

the like

is

true of the remaining operations of arithmetic.

PRELIMINARY CONSIDERATIONS
Thus, to add 2x
to

81

+ Sy and 4 x + 5 ?/ merely means to find the simplest which the expression 2x + Sy + {ix + by) can be reduced by We thus obtain applying formulas 1-5, and adding given numbers.
form

2x + + (4x + 5?/) = 2x + (3y + 4x) + 52/ = 2x + 4x + Sy + 5y = (2 + 4) X + (3 + 5)

.3

2/

2/

= 2x + 3?/ + 4x + 6j/ = 2x + {4x + 3y) + 52/ = (2x + 4x) + (3?/ + 5y) = 6 X + 8 y, the sum required,

by by 2 and

2
1

by 2

by 3 and 5

THE FUNDAMENTAL RULES OF RECKONING

In accordance with what has just been said, we

may

regard

246

addition, subtraction, multiplication, division, involution, evolution as defined for algebra by the following rules

and and

formulas, which

we

shall therefore call the

fundamental

rules

of reckoning. In these formulas denote any

bers of all kinds in the first part of the book

a, b, c

which have been established for num the

letters

numbers whatsoever, and the sign of equality, =, means "represents the same number as."
finite

Addition.

The

-\-

b.

We

call this

result of addintj i to a is the expression It has expression the sum of a and b.

247

a value, and but one, for any given values of a and b. In particular, a-\-0 = 0-\-a = a. Addition is a cnmm^itative and an associative operation that
;

248

is, it

conforms to the two laws, 34, 35

a

+b=b
If

-\-

a,

-\-

(b

-\-

c)

= (a

-\-

b)

c.

Thfe following rules of equality are true of sums,

39

249

a
-{-

b,
-\- c,

then a then

If a
Subtraction.

c ^= b

+c=b a = b*

-{- c.

Tliis is the inverse of addition, 55.

b,

two numbers, a and

* I,ater

there

is

Given any always a number, and but one,

good when
c is infinite.

250

we

shall find that this rule does not hold

82

A COLLEGE ALGEBRA
call this

from which a can be obtained by adding b. We bar the remainder which results on subtracting

from

we represent

it

by the expression a
(a

b)

-\-

= a.
b

a,

numand

b.

Hence, by definition,

In particular, we represent
251
Multiplication.

of

by

b.

The

result

multlphjlng a by i

is

the

a by b. It has a value, and but one, for any given values of a and b. = a = 0, whatever finite value a may In particular, a
expression
ab.

We

call ab the j^^'oduct of

have.

When
252

Zi

is

Multiplication

a positive integer, ab = a -\- a + to b terms. is a commutative and an associative operation,

;

and it is distributive with respect to addition forms to the three laws, 45-47
:

that

is, it

con-

ab

= ba,
If

a (be)

= (ab)

c,

(b

-}-

c)

ab

ac.

253

The following rules of equality

a
z=
b,

are true of products,

75,

76

then ac then
then

If ac
If ac

be, = 0,
is

= be. a = b, unless a = 0, or

c c

= 0.* = 0.
^

254

Division.

This

the inverse of 77iultiplication,

;

12A.

Given

any two numbers, a and b except when b is 0, there is always a number, and but one, from which a can be obtained by We call this number the quotient which multiplying by b. results on dividing a by b, and we represent it by the expression - or a/b.

Hence, by definition,

e
255
* Later

- \b

a,

except

when

0.

Involution. This is a case of repeated multiplication. to n factors by a" and represent the continued product a a call it the nth power of a.

We

we

shall find that this rule does uot hold

good when

c is infinite.

PRELIMINARY CONSIDERATIONS
In the symbol

83

a", n is called the exponent, and a the base. Involution, or raising to a power, conforms to the following thi'ee laws, called the laws of exponents, 185
:

256

a"'

n a"

am + n

(')

a""",

(ab)"

a^b""

The following
If If

rules of equality are true of powers,

184

257

= a" =
a

b,
b",

then a"

then

= a = b,

ft".

or a

b.

The second
is

of these rules and the general rule of which a particular case will be demonstrated later.

it

Evolution.

This

is

one of the inverses of involution, 138,

258

Given any jjositive number a, there is a positive number, and but one, whose nth power is equal to a. We call this number the p)'>'^ncipal nth root of a, and we represent it by v a,
140.
or,

when n

2,

by va.

Hence, by definition,

(^y =

a.

But this positive number, Vo, is not the only number whose n\\\ power is equal to a. On the contrary, as will be shown subsequently, there are n different ininiV)ers whose th powers are equal to a and this is true not only when a is
;

positive, but also

when a

is

any other kind of number.

is

When

is

positive

and n

odd, the principal nth root of

is

va.
reversibility of the preceding rules.
tlie

On the
certain of

We

have called

259

rules just enumerated rules of equality ; call the rest rules of combination.

we may

Observe that

all

the rules of combination and the rules of

equality for sums are reversible, but that the rules of equality for products and powers are not completely reversible.
Thus, according to the distributive law, a{h -\- c) = ah + ac, which is one of the rules of combination, we can replace a{b + c) by ab + ac, or reversely, ab + ac hy a (b + c).

84
Again,
if

A COLLEGE ALGEBRA
a
if

=
if

6,

we may always conclude

that a

c i= b

c,

and

reversely, that,

c,

then a

b.

But while,
trary,
c is
if

= 6, we may
=

ac

be,

we can conclude
6
it

always conclude that ac = bc; on the conthat a = b only when we know that

not

0.

And
260 The

while from a

only follows that either a

b or

always follows that a^ a = b.

=
b

b^,

from

a^

is

b^ it

rules of inequality.

The formula a
real numbers, a

^
and

means

" a

not

equal to b."

Of two given unequal

b,

the one

is alge 62.

braically the greater, the other algebraically the lesser, If a is the greater and b the lesser, we write

"> b or b <. a.

In particular, we have a
261
positive or negative. For any given real

>

or a

<
c,

0,

according as a

is

numbers

a,

b,

we have the

rules,

178, 184:
1.

If

If

If
2.

=b =b a < b
a
as

and

and
and

= < b <
b
c

c,

then a
then a

=
<. <.

c.

c,
c,

c.

then a

c.

According

a <, =, or

so

is

<, =, or

and
but
3.

ac <, =, or
ac

>, =,

or

> i, > -f > be, if c>0; < be, if c < 0.

^>

r,

When

a and b are positive,

according as
so
is

a <, =, or
a"

<, =,

or
or

and

Va

<, =,

> b, > i"; > V^.

As has already been pointed

which involve only the sign
bers also.

This
c,

is

out, the rules under 2 and 3 hold good of imaginary numIf a = also true of the rule and b = c,

ft

then a

which we may

call the general rule

of equality.

PRELIMINARY CONSIDERATIONS
ADDITIONAL ALGEBRAIC SYMBOLS

85

Besides the symbols whose meanings have been explained

in the preceding sections, the following are often in algebra:
1.

262

employed

Various

sir/ns

of aggregatioji, like the parentheses ()

^

employed above, and [ ], |, to indicate that the expression included by them is to be used as a single symbol. 2. The double signs , read "plus or minus," and q:, read "minus or plus."
Thus, in a

6 =f

c,

signs being read together

3.
4.
5.

which means a + h c or a h and the lower signs similarly.

for hence or therefore. for and so on.

-\-

c,

the upper

The symbol The symbol

;

.*.

than

for not greater than ; <^ for not Also, . for smce ; ^ for greater or less than.

less

ALGEBRAIC EXPRESSIONS

Any expression formed by combining letters, or letters and numbers, by the operations just described, is called an algebraic
expression.
Note.

263

The number

an expression 1 + X + x^ +

may

of times that an operation is involved in such be limited, as in 1 + x + x^, or unlimited, as in supposed to be continued without end. In the one

264

we say that the expression is finite, in the other, infinite. present we shall have to do 'vith finite expressions only.
case

For the

It is customary to classify algebraic expressions as follows, according to the manner in which the variable (or unknown) letters under consideration occur in them
:

265

expression is called integral if it does not involve an indicated division by an expression in which a variable letter
occurs
;

An

266

fractional, if it does.

86
Thus,
if

A COLLEGE ALGEBRA
X and y are the variable
axletters,

but

a, &, c

constants,

then

+
X
1

hx

and b

+ Vx

are integral,
.

but

2/

and
1

+x X

are fractional.

..

267

An expression is called rational if it does not involve an indicated root of an expression in which a variable letter
occurs; irrational,
Thus, a
if it

does.

+ Vbx
1.

is

rational, but

V^ + y/y

is

irrational.

268

Notes.

In applying these terms to an expression,

we suppose

it

reduced to its simplest form. Thus, Vx^ + 2 xy 4- y'^ is rational, since it can be reduced to the rational form x -\- y. 2. The terms integral, rational, and so on, have nothing to do with the numerical values of the expressions to which they are applied. Thus, X + 2 is a rational integral expression, but it represents an It represents a fraction for integer only when x itself represents one.
every fractional value of x, and an irrational value of X.

number

for every irrational

269

When

an algebraic expression

is

with the signs immediately preceding them are called the terms of A.
connected by
or
signs, these parts

made up

of certain parts

Thus, the terms of the expression

a+

a^c-{b +

c)

[d

+ e]- {f+g} +

+ i+j\- n +p

+m

are a, a-c, {b + c), and so on, those of the terms which themselves consist of more terras than one being enclosed by parentheses or some

equivalent sign of aggregation, 262,

1.

270

Integral expressions are called monomials, binomials, trinomials, and in general polynomials, according to the number of
their terms.

271

In any monomial, the product of the constant factors

is

called the coefficient of the product of the variable factors.

Thus, in 4
ab'^xhj*, 4 ab^ is

At

the

same

time,

it is

proper to

the coefficient of x^y*. call any factor the coefficient of the

rest of the product.

PRELIMINARY CONSIDERATIONS
In every monomial the coefficient should be written
coefficient is expressed,
it is 1.

87
"When no

first.

Thus,

1 is

the coefficient of x^y.

Like terms are such as differ in their coefficients at most.

Thus,

272

2 x'^y

and bxhj are

like terms.
is

The

degree of a monomial

the

sum

of the exponents of such

it.

273

of the variables under discussion as occur in

Thus,
if

of ax^, three; that of

the variables are x and y, the degree of 4 ab'^x^y* 6, zero (see 595).

is

seven

that

The degree of a polynomial is the degree of its term or terms of highest degree and the degree of any integral expression is that of the simplest polynomial to which it can be reduced.
;

274

Thus, the degree of ax^ + hx-y 1) (x 2) is two. degree of (x

cy^

dx-

ey

+ / is

three ;

and the

It is convenient to arrange the tenants of a polynomial in the order of their degrees, descending or ascending, and if there are several terms of the same degree, to arrange these in the

275

order of their degrees in one of the variables.

This order
is

observed in the polynomial given in

is

274.

polynomial

said to be homogeneous

when

all its

terms

276

are of the

same degree.

Thus, 5 x^

x'^y

xy"^

+ y^

is

homogeneous.

Rational integral expresPolynomials sions in a single variable, as x, are of especial importance. They play much the same role in algebra as integral numbers
in arithmetic. In fact we shall find that they possess many properties analogous to those of integral numbers. They can always be reduced to the form of a polynoviial in x, that is, one of the forms
:

in a single variable.

277

UqX
or, as

-}-

!,

CTqCc^

aiX

a.,,

a^pc^
:

a^x"^

2^

+ ^3?

>

we

shall say, to the

form

88

A COLLEGE ALGEBRA

where n denotes the degree of the expression, and the dots stand for as many terms as are needed to make the entire

number of terms n-\-l. The coefficients a^, a^,

of any kind.

denote constants, wliich

In particular, any of them except a^ the polynomial then being called incomplete.

may be may be 0,

Observe that in each term the sum of the subscript of a and the exponent of x is the degree of the polynomial.
Thus, in 5 x^
as

x'

1,

04

2, as

+ 2 X- + X 3, we = 1, tte = 3.
or

have n

= 6,

ao

5,

ai

= 0,
a-

Oo

= 0,

278
x^

Functions.

Clearly an algebraic expression like

2 or

+ y,

which involves one

more

variables, is itself a vari-

We call X -\- 2 & function of x because its value depends on that of x in such a way that to each value of x there corresponds a definite value q/" x + 2.
able.
call x^ + ?/ a function of x and y and, every algebraic expression z.f\inction of all the variables which occur in it.

For a like reason we

in general,

we

call

279

What we have just termed integral or fractional, rational or irrational expressions in a-, x and y, and so on, Ave may also term integral or fractional, rational or irrational functions of
X,

X and

?/,

and so

on.

280

often represent a given function of x by the symbol /(), read "function of a-." We then represent the values of the function which correspond to a; 0, 1, i, by
shall

We

/(O), /(I),

/(ft).

6 + 2. And, 3, f{b) Thus, if /(x) = X + 2, we have /(O) = 2, /(I) in general, if /(x) represent any given expression in x, /(6) represents the result of substituting h for x in the expression.

When dealing with two or more functions of , we may represent one of them by f{x), the others by similar symbols,
as F{x^, <t>(x), </'(^)In like manner, we
ables,

may

X and

y,

by the symbol

represent a function of two variand so on. /(a;, y),

PRELIMINARY CONSIDERATIONS
EXERCISE
1.
I

89

X,

the degree of x-yz^ + 2 x^y^z^ + 3 x'^y'^z^ with respect to y, and z separately ? with respect to y and z jointly ? with respect
is

What

to X, y, z jointly ?
2.

What

is

the degree of (x

3.

Given 3 x'^

+ 1) (2 x2 + 3) (x^ - 7) ? + x^ 4 x* + x^ 12 what are the values of

;

n, ao, ai,

in the notation of 277 ?

4.
5.

If

/(x)

2 x3
{x2

If

/(x) /(x)

=
= =
?/)

- x2 + - 3x + +

3, find /(O),

/(

1),

/(3), /(8).

2)/{2x

5),

find/{0),/(- 2),/(6).

6.

If

+ V;^ +
3,

3, find /(I), /(4), /(5).

is

7.
8.

If /(x)
If /(x,

2X

what
X

/(x

2) ? /(x^

1) ?
:

x^

?/

8, find
1),

the following
/(I,
1),

/(0,0),

/(I, 0),

/(O,

/(-2, -3).

IDENTICAL EQUATIONS OR IDENTITIES

If. A denotes the very same expression as B, or one which can be transformed into B by the rules of reckoning, 247-

281

258,

we say that A is identically equal to B. The notation A = B means ".4 is identically equal
Thus, X
(X

to

B"

2)

is

identically equal to x2

2 (x

2).

For

X (X

2)

+ 4 = (x2 +

2 x)

= x2 +

(2x

4)

= x2 +

2(x

2).

248,252

We
An
slo7i,

call

=B

an identical equation, or identity.

Hence
282

identical equation

A=B

zs

a statement that a first expres-

A, can be transformed into a second expression, B, by means

of the rules of reckoning.

Id particular, an identical equation like

283

3_8 + 2 = 4 + T-14
in

which no

letters occur, is called a numerical identity.

90

A COLLEGE ALGEBRA
is

The following very useful theorem

284
Theorem.

implied in

282.

If two polijnomials in x are identically equal, their

is,

corresponding coefficients are equal ; that

If aox"
then

+ aix-i + + ao = bo,

a,

= aj =

boX"
bi,

+
,

bix"-'
a

+ b,

b^.

For were these

coefficients different, the

first

as they stand and the the rules of reckoning.

polynomials would be different could not be transformed into the second by

Thus,
If,

if

ax2

3X

- 3 = 2 x2 +

6x

c,

then a

2, &

3, c

=-

3.

instead of being constants, the coefficients ao, fli, , bo, 6i, denote algebraic expressions ivhich do not involve x, it follows from the
^oX" + ^ix"-! aix''-i + identity aoX" in other words, that the expressions denoted

that ao

= bo,

Ui

= bi,

by corresponding

coefficients,

ao and

bo,

and so on, are

identically equal.

285

similar theorem holds good of two identically equal polynomials whose terms are products of powers of two or more variables with constant coefficients.
Thus,
if

bx

+
b

cy

dx"^

exy +fy^

+
d'x'^

= a' + b'x + c'y +

then

e'xy

+fy^ +

a',

b',

c',

d', e

e',f =/',

286

Properties of identical equations.

make constant use

Theorem
1
.

In algebraic reckoning of the following theorems

:

we

7/"

A=

B, then

B=

A.

For the process by which

since
it

A may

be transformed into

is

reversible

involves only rules of combination, 259. into A. will transform

But the reverse process

Thus, we

may
x2

reverse the transformation in the example in 281.

For

2 (x

2)

= x2 +

(2

4)

= (x2 + 2x) + 4 = x(x + 2) + 4.

Theorem
For since
2.

248,252

If

A = C and B = C, then A = B= C, we have C = B.

and

B.

by Theorem

Hence

A=C

C ^B, and

therefore

A = B.

PRELIMINARY CONSIDERATIONS
Thus, since

91
248, 252
248, 252

x (x
x2
,

and

we have

x(x

+ 4 = x2 + + 2) = x2 + + 2) + 4 = x2 +
+
2)

2x
2

2 (X

+ 4, + 4, 2 (x + 2).
X
identity

Theorem
operation
is

3.

An

identity remains
its

an

when

the

same

ijerformed on both

members.

This follows from the rules of equality, 249, 253, 257.

Thus,

A = B,

then

A + C=B
is

-\-

C,

and so on.
of

On proving identities. A and B, that A = B, it

As 286, 2, and B to the same form C. The following theorem supplies another very useful method.
into B.

two given expressions, not necessary actually to transform shows, it is sufficient, if we can reduce

To prove

287

If from a sujjposed identity, A = B, a known identity, C = D, can he derived by a reversible process, the supposed identity

288

A=B

is

true.
is
is

For since the process

Therefore, since

reversible,
true,
-\-

C = -D

A = B can be ^ = B is also true.

b b
is

derived from

C = D.

Example.
If
it

Prove that a

identically equal to a.
(1)
b.

we suppose
[(a

+
-

b=a + b=a +
and the step

will follow that

b)

b]

(2)

249

But

(2) is

known

identity, 250,

(1) to (2) is reversible.

Therefore

(1) is true.

That it is not safe to draw the conclusion A = B unless the process from ^ = i?to C = Z)is reversible may be illustrated thus
:

If
it

we suppose
but

will follow that

=x x^ = (- x)2.
x
it

(1)

(2)
(1) is true,
is false.

Here

(2) is true,
(2) is

does not follow from this that

since

the step (1) to

not reversible, 259.

And

in fact, (1)

It is important to remember that Identity and equality. identity is primarily a relation of form rather than of value. At the same time.

289

If A and

B are finite expressions, and A = B, then A and B have

letters

equal values for all values of the

which

may occur in

them.

92

A COLLEGE ALGEBRA
A

into B by a limited number of For, by hypothesis, we can transform = b + a, and so on. But a + b and b + a applications of the rules a -\- b have equal values whatever the values of a and b and so on.
;

The reason
later.

for restricting the theorem to finite expressions will appear

in

Conversely, if and B, then

and

A =B.

have equal values for all values of the letters This will be proved subsequently.

Hence in the case of finite expressions we may always replace the sign of identity of form, =, by the sign of equality B. shall usually of value, =, and when A B, write A

We

follow this practice. This use of the sign

that described in

is to

be carefully distinguished from

325.

ON CONVERSE PROPOSITIONS
290
Consider a proposition which has the form

If A, then B,
fully expressed then a certain other statement, B,
or,
:

(1)
is

more

If a certain statement, A,
is

true,

also true.
it is

Thus,

If
If

a figure

is

a square, then

a rectangle.
1

1,

then X

0.

291

Interchanging the hypothesis, A, and the conclusion, B, of (1) we obtain the converse proposition

If B, then

A*
:

(2)

Thus, the converses of the propositions just cited are

If If

a figure

is

a rectangle, then

it is

a square.

0,

then x

1.

292

As

the

first

of these examples illustrates, the converse of

true jyroposition

may

be false.

*A proposition like If A ami B, thou C, wliich has a doiihlp hypothesis, has two converses: namely, If C and B. then ^4, and If A and (\ then B. Simiare three converses; and so on. larly, if there be a triple hypothesis there

THE FUNDAMENTAL OPERATIONS

93
293

But the converse of a true proposition: If A, then B, is always true when the process of reasoning by which the conis derived from the hypothesis, A, is reversible; for by reversing the process we may derive A from B, in other

clusion, B,

words, prove If B, then A. The method of proving a proposition by proving its converse by a reversible process is constantly employed in algebra. An
illustration of this

method has already been given

:

in 288.

When

a proposition

If

yl

then B,

is true,

we

1.

call

suffi-

294

cient condition of B,

and

B
If

a necessary condition of A.
x

Tims, the

proposition
is

1,

then

(cc

(x

1)

(a;

2)

=
2)

is

true.

Hence x
(x

= 1) (x

sufficient
is

condition

that

1) (x

0,

and

2)

a necessary condition that x

.1,

When

both the proposition If

If B, then A, are true, we say that A sary condition of B and inee versa.
;

is

then B, and its converse the sufficient and necesand

x

295

Thns, both

(1)

If

1,

then x
is

= 0,

(2) If

0,

then

1,

are true.

that X

Hence x = 1 and vice versa.

the sufficient and necessary condition

XL

THE FUNDAMENTAL OPERATIONS

ADDITION AND SUBTRACTION

Sum

and remainder.

Let

A and B

denote any two algebraic

296

sum of A and B, and by the remai7ider to be found by subtracting B from A, we shall mean the simplest forms to which the expressions A + B and A B can be
expressions.

By

the

reduced by aid of the rules of reckoning, 247-258. Some useful formulas. In making these reductions the

fol-

297

lowing formulas are very serviceable, namely

1.

a a

3.

+ b c = a c b. c)= a + b -]-(b a (b 5.
-{-

2.

c.

4.

c')

ab

(b c) = a b c. (b c) = a b + ac.
-{-

c.

94

A COLLEGE ALGEBRA

These formulas may be described as extensions of the commutative, associative, and distributive laws to subtraction. We may prove 1 and 2 by aid of the rule, 249
:

Two
the

expi'essions are equal if the results obtained

to

by adding

same expression
1.

both are equal.

a-\-b

c^=a
c

c-\-b.

For the
Thus,

result of adding
[(a

to each

member

is

i.

h)

c\

h,

250

and

(a-c)-\-b
2.

+ c-{a-c) + c +
c.

a-\-b.

248, 250

(b + c) = a b
b
c)
c)
-\-

For the result of adding

Thus,
[a

c to
a,

each

member +

is a.

(6

c)]

(6
(6

+ +

250

and

a-6 c +
prove

= a 6-C + = a-6 + 6 =
:

a.

248, 250

We may
Since

3, 4,

5 as follows

h={b-c)+c,
3.

250

we

have,

4.

5.

+ 6 c = a + [(6 c) + c] c = a+{b-c) + e -c = a+{b-c). by 1 and a b + c = a \_{b c c']-\= a(b c)c-{-c = a-(b-c). ab ac = a[^(b r) + r] ac = a(b c)+ ac ac = a(b-c). by land
a
() -{-

248

250

by 2

250

252

250

Observe that

it

follows from

248 and the formulas 1-4

that a series of additions and subtractions any order whatsoever.

may

be

performed in

THE FUNDAMENTAL OPERATIONS

Thus,

95

+ +

b
c c

= =
Rules
of sign.

+ c~d + e = a + c - (b + d) + e = a + c + e b d.
" " rules of sign

e
'

+
(b

e,

byl
d),

-}-

by 2 and

by 2

The

which follow are par-

298

ticular cases of the formulas 3, 4, 5 just established.

1.

-\-

3.

a (
1, 2,

=a = ac. c)
c)
:

c.

2.

(
a)

c)

-\- c.

4.

= c)

ac.

"We obtain

3 at once

by setting

in 297, 3, 4, 5 respectively.

We may

prove 4 as follows

(-a)(-c) = (-o)(0-c) = (-a)0-(-a)c

297, 5

=
Rule
2, 3, 4,

oc)

ac.

by

and 3
297,

of parentheses.

From

the formulas

248 and
:

299

we obtain the following important

rule

Parentheses preceded by the

theses preceded by the

sign may he removed ; parensign m.ay also he I'emoved, if the sign

of every term within the parentheses he changed.

Parentheses

may

be introduced

in

accordance with the

same

rule.

Thus,

a-\-b

a-\-b

{c

e).

To simplify an expression which involves parentheses

within parentheses, apply the rule to the several parentheses
successively.
Thus,

\b

[c

{d

e)]\

a a a

h b

+ +

[c c

{d (d

e)']

= =

e)
e.

6-l-c

Of course the parentheses may be removed in any order but by beginning with the outermost one (as in the example) we avoid changing any sign more than once.

96
300

A COLLEGE ALGEBRA
From

Rules for adding and subtracting integral expressions. the formulas of 248, 252, 297 we derive the rules
:

To odd
coefficients,

(or subtract) tioo like terms,

add

(or subtract) their

all their terms in

and

affix the

common

letters to the residt.

To add two or more polynomials, write

succession ivith their signs unchanged,

and then simplify by

combining like terms.

To subtract one polynomial from another, change the sign of every term in the subtrahend aiid add.
Example
1.

Add

ab"^

and

(i

5 a6-

also subtract

aJfi

from 4

ab"^.

We have
and

4 a62

4 a62
2.

+ (- ba) = - - 5 ab"^) =
(

Example

Add

x^

+
+

ax'^y

- b) ab"- = - ab"- {- 5)] ab"- = 9 ab"^. [4 + 2 ab^ and bx'^y 5 ab^.

;

248

297, 5

We have

x^

axT-y

= x3 + = + = x3 +
a;3

ax^y ax-y
(a

+ + +
5)

2 ab^ 2 a63

+
+

{bx'^y

5 ab^)

bx"y

-bab^
5 ab^

299
248

bx-y
x^y
ab-

+ 2 a6' - 3 ab^
b^

252, 297, 5

Example

3.

Subtract 2
a^

a% a-b
a-'-b

from

a^

a-b

b^.

We

have

+ + -

+
+

b^ 63

(2 a-b

ab'^

= =

a' a^

2 a26

ab"-

+ ft?) - b^

299

a-b

ab-.

252, 297

When
terms,

it is

the polynomials to be added (or subtracted) have like convenient to arrange these terms in columns and

then to add (or subtract) by columns.

Example
subtract 5
4.

Add

a*

+ a% -

a"-b"-

and

ab'^

a"-b'^

a^b,

and

a^b"^

ab^

from the
a*

result.

We have

a^b

2 S

a%aW-

+
a6^

6*

-a^b + gi

5 a%"-

+
-I-

a6

- 4 a262

2a65

- &*

THE FUNDAMENTAL OPERATIONS

^V
r,

97

EXERCISE n
-Qax-y,
5 bx'^y,

1.
2. 3.
4.

Add4ax2y,

and

Sbx-y.

+ 2 a - 62, Sa + b^--2a^, and b^-4a-ia^. Add3x2-5x + 6, x2 + 2x - 8, and - 4x2 + 3x - 7. i a^ 5 a* - 6 a62 - a% Add 4 a3 + a"-b - d b\ + 10 6% and 6 6' - 15 a62 _ 4 a26 - 10 a^.
Add
7 a2

5.
6.

Subtract

4a 26 + 6c from

3a

c.

Subtract 2 x2

5x

from x^

6 x2

5.

7. 8.

What must
From
x^

be added to a^
y^

5 a"b to give a^

+
+

6^ p

6x + 5y + + ~
\

take the

sum
2 x2

of

-2x2-6x + 72/-8
9.

and
a

x-^

+
a

2/

9.

Simplify
Simplify

{a

b)

{2

b)\

6 (a

4b).

10.

11.
12.

6x-^4x + [2x- (3x + 5x + 7 - l) + 3] -8^. - [4 - c + 3 a - (4 6 - c) - (6 + 3 c) - 6 c]. Simplify 2 a Subtract x - (3 + 2 2) from 2 - [3 x + (y + 5 2)].

^ ^

?/

13.

To what should
To what should

xx*

14.

8x

5 be

added

to give x^

y"^

7 ?

9 x2

3 y be added to give

7 ?

MULTIPLICATION
Product.

By
shall

and

B,

we

mean

the jjroduct of two algebraic expressions, A the simplest form to which the expression

301

AB
1. 2.

Of

can be reduced by means of the rules of reckoning. especial importance in such reductions are
:

3.

The commutative, associative, and distributive The law of exponents a'"- a" = a m + n The rules of sign
:

laws.

a(

b)

= (

a)b

= ab;

a)

b)

= ab.
303

1. To find the Rules for multiplying integral expressions. product of two monomials, multip)ly the product of the numerical

98

A COLLEGE ALGEBRA

the latter by factors hy that of the literal factors, simjMfijing adding exponents of powers of the same letter.

Give the result the

or

sign, according as the

monomials
or

have
2.

like or unlike signs.

To find the product of a polynomial by a monomial

each pohjnomial, multiply each term of the multiplicand by term of the multiptlier and add the products thus obtained.

The
law

first

rule follows

the law of exponents.

;

from the commutative and associative laws and The second rule follows from the distributive

thus,
{a

c)

{m

-{ n)

b + c)n {a + b + c)m + {a + = am + bm + cm + an + bn + en.

The

first

monomials.

rule applies also to products of more than two When an odd mimber of these monomials have
;

the sign of the product is otherwise it is +. product of more than two polynomials may be found by repeated applications of the second rule.

signs,
A

Example

1.

Find the product of

We have - 4 a:^b^x^

2 6x*

- 4 a-b-x^, 2 6x^ and - 3 a^x. - 3 a^x = 24 a'^b'^x'^bx^a^x = 24 a^b^x^.

b"^ -\- a^. Example 2. Find the product of a 2 6 and ah For convenience we arrange both factors in descending powers and choose the simpler factor as multiplier. We then have

of a,

(a2

+ ab-

b^) (a

2 6)

= =

a^

+
_

a^-b

ah^

-2a"-b-2 aW- +

268

a3

aPh

ab'^

b^.

303

The degree of the product with respect to any letter (or set of letters) is the sum of the degrees of the factors with respect to that letter (or set of letters).
This follows from 302,
degree in any product the factors.
is
1, and the fact that the term of highest the product of the terms of highest degree in

and the degree

is jive.

Thus, the degrees of x^ + 1 and x^ of the product (x^ +

1) (x^

are two and three respectively, - 1), or x^ + x" - x'- - 1,,

THE. FUNDAMENTAL OPERATIONS

99
is

When

both factors are homogeneous,

276, the product

304

homogeneous.
For if all the terms of each factor are of the same degree, all the products obtained by multiplying a term of the one by a term of the other are of the same degree. Hence the sum of these products is a homogeneous polynomial.

When both factors are polyArrangement of the reckoning. nomials in x or any other single letter, or when both are
homogeneous functions of two letters, it is convenient to arrange the reckoning as in the following examples.
Example
2
n
3
a;2

305

1.

Multiply 2x3
A.
r.

- x^ +

_
I

by x - 3 + x^. ^^ arrange both factors in descend-

X
J

.3
'

^^^ ^^^ ascending) powers of x and

,

2
_ 2

place multiplier under multiplicand. We then write in separate rows the

_ 2x5+ x^-7x3 + 8S + 5xIl5

r
3 4.

"

1 f;

P^i'tial

products

"

corresponding to

''^"'^^ '"^' ^^ *^" multiplier, placing them so that like terms, that
'^^'

is,

Finally

terms of the same degree, are in the same column. we add these like terms by columns.
2.
2/2

Example
x2

Multiply

x"^

y'^

xy hy 2 y

-\-

x.

+
4-

2 x?/
2
?/

X
x^

In this case both factors are homogeneous functions of x and y.

+
+

2 ofiy 2 x"y

x^

4 x-y

xy'^ + 4 xy'^ + 3 xy'^

We
2y^ 2y^
of x,

arrange them both in descending powers and therefore in ascending powers of y

in Ex.
1.

and then proceed as

coefficients. In the reckoning illustrated in 305, the terms are so arranged that their positions suffice to indicate what powers of x occur in them. may make

Detached
1,

306

Ex.

We

use of this fact to abridge the reckoning and writing the coefficients only, and it

by suppressing x
is

always worth

while to do this
coefficients.

when the given polynomials have numerical

is

If either polynomial

incomplete, care viust he taken

coefficient.

to

indicate every missing term by a

100
Example.

A COLLEGE ALGEBRA
Multiply x^
2

3 x2

by

x^

+ 3 x2 -

2.

1
X -r

S4-O + "^
"''
.-.

We
Ex.
1,

arrange the reckoning as in 305, but write the coefficients only, indicoefficients.

o'

-r

v/

d
d

U y

+ +

eating the missing terms by

2+6

^^
^

partial product correspondInsert^^ ^^^ Q ^gj.j^ ^ j.j^g multiplier.

j^j^g

^j^j^

+ 09+4-

result
is

ing the appropriate powers of x in the final beginning with x^ since the sum
six

of the degrees of the factors - 9z* + 12 x^ - 4. namely, x^

we

obtain the product required,

The degree of the product, terms, seven, in the result

.six, is

also indicated

by the number
277.

of

1+0-9 + + 12-0-4,

This is called the method of detached coefficients. It applies not only to polynomials in a single letter, both arranged in but also to descending or ascending powers of that letter,

homogeneous polynomials in two letters. For in arranging two such polynomials in descending powers of one of the letters, we at the same time arrange them in ascending powers of the

what powers
307

other letter, so that the position of any coefficient will indicate of both letters go with it.

Formulas derived by the method

sider the following examples.

of detached coefficients.

Con-

Example

1.

Prove the truth of the identity

(a*

+ a% +

a262

afts

54)

(_?,)

?,5_

We
1
1

perform

the multiplication indicated in

coefficients,
,

the
1
1

first

member by detached
,.
.

and
,.

r
1

so obtain the coefficients of the product arranged '^

m
.

descendmg powers
b.

of

a and

m
.

ascendnig

powers of

"^ We know in advance that the degree of the in product is five, which is also indicated by the number of terms, six, Hence the product is the final result.

a5

+
2.

a^6

a''62

a26

0- ah*

b^,

or a^

b^.

Example

Prove the truth of the

(a2

identities

(aS

a^-b

-ab + + ab"- -

&2) (a
b^) {a

+ +

6)
b)

= =

flS

a*

+ -

63. b*.

(1)
(2)

THE FUNDAMENTAL OPEI^ATICNS

Proceeding precisely as in Ex.
1
1,

101

we haTS
1

(1)

(2)

1+1

1+1

1-1+1 1-1+1
1

1-1+1-1 1-1+1-1
i.e.

1,

a^

6^

+ 0-1,

i.e.

a'

- .

By

the method illustrated in these examples

we may prove

the truth of the following identities, of which the examples are special cases, namely
:

For

everi/ positive integral

value of

?i

we have

308

(a"-i 4-

a"-^

-\

h ah"-^

+ b"'^) (a-b)=a'' b\
7i,

For every positive odd value of

(a"-i

we have

309
b)

a''-H

-\

ab"-"-

+
-

b"-^) (a

a"

+
-

b\

And

for every positive even value of n,

we have

310
b".

(a"-^

a^-^b

-\

[-

ab"-^

b"'^) (a

+ b) =

a"

Powers
of a

of a binomial.

We

can compute successive powers

311

by repeated multiplications. These multiplications are readily performed by detached coefficients.

6

As the coefficients of the multiplier are always 1 + 1, it is only necessary to indicate for each multiplication the partial We thus obtain products and their sum.
(1)
1

102
partial "products"

A COLLEGE ALGEBRA
and
so'

obtain the coefficients of the next

:

power, we are merely applying the rule

312

To any coefficient in a power already found add the which precedes it ; the sum will be the corresjyonding
in the next power. All the coefficierits of this next power, excejit the first can he found by this rule ; these are 1 and 1
Thus, the coefficients of
3
(4)

coefficient

coefficient

and

last,

which correspond
obtain 4

to 3, 3, 1 in (3) are

or

4,

+
5.

3 or 6,

3 or 4.

Applying the rule

10, 1

to (4),

we

or

5,

4 or 10, 4

-t-

6 or

4 or
(a

Hence
6)5

a5

5 a*b

10 a^b^

10

a"-b^

+ 5a +

.
b

Evidently the coefficients of any given power oi a be obtained by repeated applications of this rule.
Example.
Find successively
of
(a

-\-

can

6), (a

+ Wi
the

[c-

^)^-

313

Products

two binomial

factors

of

first

degree.

The

student should accustom himself to obtaining products of this

kind by inspection.
{x

We
b)

have

+ a) {x +
aj)
(^0^:;

X'

{a^x

^i)

=^ o^o-^^

+ {a + b)x + ah. + (o^i + aibo) x

(1)
-f a^b^.

(2)

In the product (1) the coefficient of x is the stim and the final term is the product of a and b. In the product (2) the first and last coefficients are products
of the
first coefficients

and of the

last coefficients of the factors,

and the middle ao^i and a j JoExample

(X
1.

coefficient is the

sum

of the " cross-products "

Find the product

5) (x

(x

5)

(a;

8).

+
2.

8)

x2

-f

(5

(x

8)x

40
(x

x2

- 3x -

40.

Example
(X

Find the product

3y)

+
7/2

lOy).

Sy)

(X

10

2/)

x2

(3

10)X2/

30

^i

+ l3xy +

30?/2.

THE FUNDAMENTAL OPERATIONS

Example
(2x
3.

103

Find the product

(2

+
3

3)

(4x

+
3

7).

3)(4x
4.

7)

2 -4x2

+
+

(2

+
a

4)x

8x2

+ 26x +

21.

Example

By (x-10){x-

the methods just explained find the products

15),

{3a

46)

(5

-66),
x.

(7

y)(5x

Sy).

Product of any two polynomials in

(UoX^ + a^x"^ + = ttoboX^ + (ao^i +
ac,x
-\-

Consider the product

biX

314

a^) (h^x-

bo)

i^o)^"*

+ (ai^2 +
The product
is

2^1

+ a2bo)x^ + 3^o) ^^ + (dzh + 3^l) ^ + dsK

+ (o^2
+i^i

of the degrees of the factors.

a polynomial in x whose degree is the sum And the coefficient of each term

may
K)

one of the numbers

^1) ^2-

be obtained by the following rule, in which a^, denotes Oq, a-^, 02, ((3, and b^. one of the numbers

and

I^ind the diffei'ence between the degree of the product the degree of the term, and then form and add all the k equals this difference. products ajjb^ in xvhich h

Thus, to obtain the coefficient of x^,

we

find the difference 5

2,

or

3,

and then form and add ahbk in which h + k = 3.

aih^,

a-zbi,

03601

these being

all

the products

X of the form
cates

This rule applies to the product of any two polynomials in + , and ^ya-" + -\- b. It also indiOo-^"' +

any particular coefficient of the product when the factors have numerical coefficients.
Example
(aox^s
1.

how

to obtain

Find the

coefficient of x^^" in the

product
-\
\-

aja;74

aux +

ays) (&oa;o

6ix59

659X

660).

The degree of the product is 75 + 60 or 135 and 135 - 100 = 35. Hence the coefficient of x^oo is 00635 + 01634 + + 03461 + 03560.
;

Similarly the coefficient of

x^

is

040650

O41659

+
+

+
-

O74626

O75635.

Example

2.

Find the

coefficient of x^ in the

product
6X
3).

(3 X*

2 x3

x2
is

8 X
2)

7) (2 x3

5 x2

The required

coefficient

(-

(-

3)

+ (-

8) 5

2,

or

14.

104
Example and of x23_
3.

A COLLEGE ALGEBRA
In the product of Ex.
1,

find the

coefficients

of a;"*

Example

4.

In the product of Ex.

2,

find separately the coefficients

of x6, x^^ X*, x2,

and

x.

315

The following Products found by aid of known identities. formulas or identities are very important and should be carefully memorized.
(a

(a

+ hf ^a'' + 2ab + h\ -by = a''-2ab + b^

(1) (2)

(a

b)(a

-b)=a^- b\

(3)

To

this list

may
'

be added

the formulas given in 308, 309,

:

310, and the following,

311

{a-\-bf

a^

+ ^ a%
b

-f-

3 ai^

yz^

(4)

may be replaced by any algebraic expressions whatsoever, these formulas supply the The simplest means of obtaining a great variety of products.
Inasmuch
following examples will

as the letters a

and

make

this clear.

Example
(3x

1.

Find the product (3x

5?/)2.

5?/)2

=
2.

{Zxf

3x

5?/

(5
{x2

(/)2

9x2

sqx?/

252/2.

by

(2)

Example
(x2

Find the product

(x2

+
y'^)

xy

y-) (x^

xy

+ xy +

2/2)

-xy +

2/2)

= =

[{x2
(X2

2/2)2

+ xy] [{x2 + _ x^y2 = a;4 +

2/2)

+ ?/2). - xy] +
y4.

a;-2y2

by

(3)^ (1)

Example
(X

3.

Explain the steps in the following process.

2/

2) {X

-y

+ z)(x + y - z) {x - y - z) = [x + (y + z)] [X - + z)] [X + - 2)] [X = [X2 - + 2)2] [X2 - - 2)2] = [(x2 - 2/2 - 22) -2yz]- [(x2 _ _ 22) + 2 yz] = [x2-(2/2 + z2)]2_4 2/222 = X* - 2 X2 (2/2 + 22) + (2/2 + 22)2 _ 4 y222 = X* + + 2* - 2 x22/2 - 2 y2z2 _ 2 22x2.

(2/

(2/

(2/

z)]

(2/

(2/

2/2

2/*

THE FUNDAMENTAL OPERATIONS

105

Observe in particular that by this method we may derive from (1) and (4) the square and cube of any polynomial.
Thus, we have
(a

+ +

+ +

c)2

= =

[(a a(a

+ 6) + c]2 = (a + + 62 + + 2 ab +
c'-

6)^

2 (a

6) c

c2

2 ac

2 6c.

(a

c)3

= =

6)3

+
c3

3 (a

hy^c

3 (a

6) c2

c^

a3

+ 63 +

+ 3a26 + 362a + 362c + 3c26 + 3c2a + 3a2c + 6a6c.

of these results
is

Generalizing the

first

we have

the theorem

The square of any

jjolynom'ial

equal

to the

sum of

the

316

squares of all its terms together with twice the products of every two of its terms.

Example Example Example

1.

Fiud the product

Find the product

(a
(1

2c

Zd)\

2.

3.

Find the product

+ 2x4- 3x'-)2. x'-y + xy- (x^

y^)^.

Powers

of

monomial products.

By

the

7ith.

power of any '317

algebraic expression, A, we shall mean the simplest form to which the expression A" can be reduced by the rules of

reckoning. From the laws of exponents derive the following rule

:

('")"

a""'

and (ab)"

a"b''

we

To raise a monomial
numerical of each
ing as
literal factor

exjyression

to the

nth

2)oive)',

raise its

^18

coefficient to the

nth p>oiver and multiply the exponent

by n.

If the sign of A be

give the result the sign

or , accord-

is

even or odd.

Thus,

2 ax-^if)*

= (

2Ya*xy"-^

16 a^x^yss.
a"6

For by repeated applications of the law

(o6)"

we have

(-

2 ax2y7)4

= (_

2)4a4 (x2)^
{a'")"

(?/7)4,

and by repeated applications

of tlie

law

a'""

we have

(-

2)4a' (x-)* (i/^)*

= lea^xV^.

106

A COLLEGE ALGEBRA
EXERCISE

m
also the identities of 315.

In the following examples perform each multiplication by the most In particular employ detached coefficients expeditious method possible.

where
1.

this

can be done with advantage

Multiply 3x5
Multiply 5 x3

_ -

2x*

xs

7x2

_ 6x +
a^
_|-

5 by 2x2

- 3x +
2 a2.
y.

1.

2.
3.

3 ax2

2 a-x

by 3 x2

ax

Multiply x5

4.
5.

Multiply 3 x3
Multiply
7

x'-yS xy* - y^ hy x*y + x^y^ - 2 x2 + 7 by 2 x3 - 3 X + 5. 2 by 4 5 y, by inspection,

?/

.X

6.

Multiply a2
Multiply X*

ax +

6x

x2

by

i)

x.
x'-^

7.
8.

5x2

Multiply 2 x"

3 x -2

- X. by 3 + x -3 by x-2 _ ^n-s. 5
x*

9.

Multiply a2 Multiply x
Multiply

10.
11. 12.

Multiply
Multiply

- a& + 3 62 by a2 + a6 - 3 62. + 3y-2zbyx-32/ + 2z. X- + xy + y- + X y + 1 hy X y 1, a2 + + c2 + 6c + ca - a6 by a + 6 - c.

6'-

13.
14. 15. 16. 17.
18.
19.

Multiply x

Find the
Also
Also
(x2
{x

5byx-4y + 6. 3zby2x + 2/-8z. product (6 + x) (6 - x) (62 + x2).

3x-2?/ +

+
X

7?/

+ +

1) (x2

+ +

1) (X*

x2

+ +

1).

z)

{- X

z) {x

z) (x

- z).
+
x

Form a table of the

coefficients of the first four

powers of x2

+ +

1.

Continue the table of coefficients of successive powers of a as far as the tenth power.
20.
21.

Find

(4 {x

3 y)2 and (4 x

3 y)^

22.
23.

+ 2y + 3z-4 u)^. Find (x + 2 + 3 z)3 also (x + 2 y Multiply (a + 2 6)2 by (a - 2 6)2.

Find
?/
;

3 z)\

24.

Find the

coefficients of x29

and of

x^^ in the

product

(aox2'

aix26

a26X

a^i)

{bcfic^^

bix^^

-\

6i8X

619).

THE FUNDAMENTAL OPERATIONS

25.

107

Find the

coefficients of x^, x^,

and

x* in the product

(2x6
26.

-3x5 +

4x*

7x3

+ 2x-5)(3x5-x3 + 2x2+ 3x-8).

:

Verify the following identities

1.

2.
3.

4.

+ y3 ^ z3) = 3(y + z){z+x){x + y). - ay)'^. (a^ + ft2) {x2 + 2/2) = (ax + byf + (6x - 62) (a;2 _ ^2) = (a^ + 5^)2 _ (j^; + ayy.^ (a2 (a + 6 + c)3 = a3 + 63 4. c3 + 3 a2 (6 + c) + 3 62 (c + a)
(X

2)3

(x3

+
27.

3c2(a

6)

6a6c.

Simplify the following powers

(2 a2x32/'')5,

(_x5y829)7^

(a26'"c3)2,
:

(a'"6"c2")".

28.

Simplify the following products

(

a62c3) (a36)2

ac3)5,

_
(

2 x'^y^f (axSyiija,

DIVISION
algebraic expressions the quotient of A divided by B, we shall mean the simplest form to which the fraction A / B can be reduced by the rules of reckoning.
of
Quotient.

Let A and

B denote any two

0.

319

which

is

not equal to

By

Formulas.

In making such reductions the following formu-

320

las are especially useful,

namely,

ac
1.

2.

=
a"

a""",

when
a

m>

ti
;

=
a"

_,^^ a"

>

when w
a

> wi.

3.

a
4.

a +b=d

d
rule,

We may
Two
expression

prove

1, 3,

and 4 by aid of the

253

expressions are equal if their products by

(iiot

any third

0) are equal.

108
For
in 1 the

A COLLEGE ALGEBRA
product of each member by be
is ac.

Thus,

bc =
be

~ be ac; and b

~ b
b

ac.

254, 252

Again, in 3 the product of each member of the first equaby b is a, and the products of each member of the second and third equations hj b are a and a respectively.
tion
Thus,

-^6 = -a;
b

and

(--\b = -^b=-a.
\

b/

298,254
a. -f- ^

Finally, in 4 the product of each

Thus,

member by
~ d

c?

is

"^^d =
d

b;

(^ -^^\d ^^^d +
\d
d/
d

b.

254, 252

The formula 2
Thus,
if

is

a particular case of the formula

a am
a"

1.

m > n,

= a-"
a"'"
a"

a".

256

Hence

a"

a"*".

by

321

Rules

for simplifying

A/B.

The formulas

1, 2,

and 3 give

us the following rules for simplifying

1.

A/B.

2.

Cancel all factors covimon to numerator and denominator. When numerator and denominator involve different jyoicers
letter (or expression^ as factors, cancel the lower subtract its exponent from that of the higher power. Give the quotient the -{-or sign, according as the

of the same

power and
3.

mimerator and denominator have the same or opposite

Thus,

signs.

^
ca-

6a5-2

ba\ and

-^ = _ J_ = _ 1.
a'

a"-2

a*

322

Rules for dividing by a monomial. From the definition of division and 320, 4, we derive the following rules.

To divide one monomial bij another, form a fraction by 1. writing the dividend over the divisor, and simplify.

THE FUNDAMENTAL OPERATIONS

2.

109

the dividend by the divisor,

Thus,

To divide a polynomial hy a monomial, divide each term of and add the quotients so obtained.

a^h'^c -^

6 alPd

8 a%'^c

Gad

4 a-c
,

Hb^d
of signs.

by cancelling the

common

factor 2 ab-

and applying the rule

h-

Again,

(ax^

4 a^x-)

ax

=
ax

= x^
ax

4 ax.

But when
regard (a

common with a and b, we cf "has no factor in b)/d as a simpler form of the quotient than

a/d +

b/d.
If

Division of a polynomial by a polynomial.

A and B
quotient

are
is

323

polynomials whicli have

the expression to which
cancelled.

common

factors,

the

A/B
x'^

reduces

when

these factors are

Thus,

if

^ = x2 -

?/2,

B=
2

+ 2xy +

y^,

the quotient

is {x

- y)/(x + y).

For

^_
B
x2

x2-y2

^
y-^

^x

y) (x

y)

x x

xy

(X

y)2

y y

In another chapter we shall give methods for finding the

factors

which are common

is

called long division

to two polynomials. considered in Chapter V.

The process

-^ b x c means j c, means a /be. In the chapter on fractions we shall consider complex expressions in which a number of indicated multiplications and divisions occur. In particular we shall find that

Complex expressions. while a -T- be, lilce a -h

Observe that a

324

(b

c),

ax(bxc-hd)=axbxc^d.
a-h(bxc-r-d)=a-^b-v-cxd.
In (1) the signs

(1)

(2)

x and -=- within the parentheses remain unchanged when the parentheses are removed but in (2) each X is changed to -;-, and each -f- to x
;
.

(6

-a){c-

SIMPLE EQUATIONS
the "

111

unknown

letter

" x

is

to satisfy.

It serves the purpose

of restricting x to values which satisfy this condition, being 4 and a; 6 are the same and true when the values of 3 a;

then only.
Similarly x
-\- ij

is

a conditional equation in the two

and, in general,

unknown

letters

x and

y,

When

the exjjressions

A
B

and

A=B
B

are not identically equal,

326

a conditional equation. This equation means : " A and And it restricts the are supposed to have equal values."
is

variable letters in
is true.

and

to

values for which this supposition

The

letters

whose values the equation A

=B

thus restricts

are called the xmknoivn letters of the equation. In what follows, the word '' equation " will ditional equation."
If the only letters in an equation are the

mean

" con-

unknown

letters,

327

as X, y,

z,

we

call it

known
Thus,

letters, as a,

a numerical equation but if there are also h, c, we call it a literal equation.

;

2x

32/

= 5isa numerical,

but ox

?*?/

c is

literal

equation.

A literal equation does not restrict the values of the known letters.
A and B are rational and integral with respect to unknown letters, the equation .1 = J5 is said to be rational and integral. But if A or B is irrational or fractional, the
If both

328

the

equation

is

said to be irrational or fractional.

is

No account
Thus,

taken of numbers or
c is

known

letters in this classification.

v2x + y/h =

both rational and integral.

simplest form,

In the case of a rational integral equation reduced to its 340, the degree of the term or terms of highest
is

329

degree

called the degree of the equation itself.

The degree

b is five. Thus, the degree of ax^ + 6x = c is two ; that of x^z"^ + y* is measured with respect to all the unknown letters, but

these letters only.

112
330

A COLLEGE ALGEBRA
first degree are often called simple or linear those of the second, third, fourth degrees are called

Equations of the
equations
;

quadratic, cubic, biquadratic

equations respectively.
letter, as x, restricts
a;

331

An
finite

equation in

07ie

unknown
its

to a

number

of values.

We say that

these values of x satisfy

the equation, or that they are

solutions or roots.

Hence

332

A root of
which,
if

an equation in x is any number or knoicn expression substituted for a;, will make the equation an identity.
_^

Thus, 1 and 2 are roots of the equation x^ and (-2)2 + (-2) = 2. 6 is a root of x + 6 = a for (a Again, a
;

for

P+ 1=2

6)

+ 6 = a.
it

333

equation may have no root tion which no number can satisfy.

Notes.
1.

An

for

may

state a condi-

finite number can satisfy the equation x + 2 = a; + 3. In every equation in x which has roots, x is merely a sym.bol for one In fact the equation itself is merely a disguised or other of these roots.

Thus, no
2.

identity, a substitute for the several actual identities

obtained by replacing

X by each root
Thus, x2

in turn.

and (-2)2

+ X = 2 is merely + (-2)^2.

a substitute for the two identities

P + 1 =2

ON SOLVING EQUATIONS
334

To

solve

roots, or to

an equation in one unknown letter prove that it has no root.

is

to find all its

The reasoning on which the process depends

in the following examples.

is

illustrated

Example
is

1.

Solve the equation

3x

4=x +

fi.

Starting with the supposition that x has a value for which this equation true, we may reason as follows
:

If

then
or

3x-4=:X + 6, 3x -4 + (-x + 4) = x + G + (-x +

2x =
x
10,
5.

4),

(2)

and therefore
Hence,
if

3x

= =

6,

then x

5.

(4)

114
conclude that c
is

A COLLEGE ALGEBRA
a
root
;

and

the process is reversible if

it is

made up of reversible
336
It
is

ste2)s.

certain value of x has been derived

important to remember that the mere fact that a from an equation by the

must be

rules of reckoning does not prove it to be a root. reversible to Avarrant this conclusion. Thus, from

The process

it

follows that

and hence

x - 2 = 0, - 2) (x - 3) = 0, that either x = 2,
(x

(1) (2)

253 253

or x

3.

(3)

But we have no right to draw the absurd conchision that 3 is a root For when x = 3 we cannot reverse the process, tliat is, divide (1). both members of (2) by x 3, since the divisor x 3 is then 0. On the other hand, when x = 2 we can reverse the process, since x 3 is then not but 1 and 2 ts a root of (1).
of
;

TRANSFORMATION THEOREMS
337
In the light of what has just been said we may regard any correct application of the rules of reckoning to an equation as a legitimate transformation of the equation and if such a
;

transformation
the roots of

we may conclude that it leaves the equation unchanged. Hence the following
is reversible,

theorems.

338

Theorem
leave
1.
its

1.

The following transformations of an equation

unchanged, namely
:

roots

Applying the rides of combination,

259, to each

member

separately.
2.

members, or subtracting
3.

Adding any expression which has a finite value to both it from both. Multiplying or dividing both members by the same constant
all

{not 0).

For

the rules of reckoning involved in these transformations are

also state the proofs of 2

reversible, 259.

We

may

and 3 as follows

SIMPLE EQUATIONS
If

115

A and B denote expressions in a;, the roots of the equation A = B numbers which substituted for x in A and B make A = B, ^ 332. But any value of z which maizes A = B and C finite will make A + C = B + C, and conversely, 249 hence the roots oi A = B are the same as those oi A + C B + C. Again, if c denote any constant except 0, any value of x which makes A = B will make cA = cB, and conversely, 253 hence the roots oi A = B are the same as those of cA = cB.
are
;

Thus, in

334,

Ex.

1,

the equations
a;

3x-4 = + 6, 3x-4 + (-x + 4) = x + 6 +

2x =
X
all

(1)

(-a;

4),

(2)
(3) (4)

10,
5.

have the same root, 5. Here (2) is derived from (1) by the transformation 2, (3) from the transformation 1, and (4) from (3) by the transformation 3.

(2)

by

The folloivlng transformations Corollary. leave its roots unchanged^ namely :

1. Transposing a term, with mem,ber to the other. 2.
3.
its

of an equation

339

sign changed,
occur

from one

Cancelling any terms that

may

tVi

both members.

Changing

the signs of all terms in both members.

For 3 is equivalent to multiplying both members by 1. And 1 and 2 are equivalent to subtracting the term in question from both members of the equation.
Thus,
if

from both members of x

x

we subtract we obtain
The
to transpose second.

a
a

-\-h

a
c

+ + +

6
b
a. (1)

(1)

(2)

effect of the subtraction is to cancel b in

both members of
first

and

a,

with

its

sign changed, from the

member

to the

By

aid of these transformations, 338, 339, every rational

340

integral equation in x may, without changing its roots, be reduced to the standard form
aoX"

+ aiX"-^

-\

h a_iX

+ a =

0.

116

A COLLEGE ALGEBRA

its

suppose such an equation reduced to this form when The like is true of rational is measured, 329. integral equations in more than one unknown letter.
degree
Thus, x2

We

3a:

= x2 4x +

can be reduced to the form

2 = 0.

Its

degree

is

therefore one, not two.

2.

341

Theorem

When

A, B, and

are integral, the equation

AC = BC
has the same roots as the two equations

A=B
For any value
or
of x

and C

0.

which makes

AC = BC

must make

either

=B

C=

and, conversely, any value of x which makes either

A = B or

C=

will

make

AC = BC,

251, 253.

In this proof it is assumed that A, B, C have finite values for the This is always true when, as is here supposed, values of x in question. A, B, C are integral but it is not always true when A, B, C are
;

fractional.

In particular, lohen

same

roots as the equations

and C are integral, the equation A = and C = jointly.

AC =

has the

two equations x

Thus, the roots of the equation x^ = 3 x are the same as those of the = 3 and x = 0, that is, 3 and 0.
Similarly the roots of (x

1) (x

two equations x

and x

2) = are = 0, that is,

the same as those of the

1

and

2.

342

Hence the

effect of

multiplying both members of an inte-

B by the same integral function C is to gral equation A introduce extraneous roots, namely, the roots of the equation
C=
0.

Conversely, the effect of removing the same integral

factor

C from both members

is to lose

of an integral equation certain of its roots, namely, the roots of C

AC = BC,

0.

the other hand, in a, fractional equation, it is usually the case that no extraneous roots are introduced when both members are multiplied by the lowest common denominator of all the fractions.

On

Thus,

if

members by x(2x not a root of x(2x

the equation be 1/x = 1 /(2x 1), and we multiply both 1 = x, whose root is 1. As 1 is 1), we obtain 2x
1)

0,

we have introduced no extraneous

root.

SIMPLE EQUATIONS
Corollary.

117
the

The

integral equation

as the equations

A=B
A
B

and

A" = B^ has A = H jointly.

same

roots

343

For A- = B^ has the same roots as A"^ B- = 0, 339. And since A^ - B^ = {A- B) {A + B), the equation A'^ - B- = has the same roots

= as the two equations same roots as the two equations

and A + B = 0, ^ 341, and therefore A B and A = B, ^ 339.

the

Thus, the roots of the equation (2x 1)2 = (x 2)2 are the same as those of the two equations 2x l = x 2 and 2 x 1 = (x 2), that
is,

and

1.

Hence the

effect of squaring

both members of the equation

344

to introduce extraneous roots, namely, the roots of the equation A B. Conversely, the effect of deriving from
is

=B

the single equation A i? is to lose certain of the B. roots, namely, the roots of the equation A
A"^

B'^

Since A^ - 5" =
it

(.4

- B) {A^~^ + A^'-^B

follows by the reasoning of are those oi A =^ B and .4"-^ +

Thus, since
x''

-\ 1- 5"~0' ^^S, 343 that the roots of .4" = iJ"

345

A^-^B
x

-\

h i?""'

=
=
1

jointly.

1=

(x

1) (x^

+
x-

1),

the equation x^

has the

same

roots as the equations x

and

jointly.

The theorems just demonstrated, 338-345, hold good for equations in more than one unknown letter if the word root be replaced by the word solution, 355.
Thus, by 339, the equation x + 2?/ 3 tions as the equation x 2?/ + 3 (2), that

346

=
is,

(1)

has

tlie

same

solu-

every pair of values of

X and y which

satisfy (1) will also satisfy (2),

and conversely.

same

When two or more equations have the Equivalent equations. roots (or solutions), we say that they are equivalent.
Thus,
3, the

347

equations

A =B

and
is

Again,

341, the

equation

AC = BC

+ C = B + C are equivalent. equivalent to the two equations

A ^ B and C = 0.
But
root 3.
xJ2

(1)

and x

For

(1) also

has the root

3 (2) are not equivalent although both have the 3, which (2) does not have.

118

A COLLEGE ALGEBRA
SOLUTION OF SIMPLE EQUATIONS

348

From
in one

the transformation theorems of 338, 339

letter, as x.
x, reduce it to the

we may

at once derive the following rule for solving a simple equation

unknown

To Then
1.

solve

a simple equation in

form ax

= b.

2. 3.

If If

= = i/" a
a,

0,=^ 0, the equation

0,

and b
b

9^ 0, the equation

has the single root b/a. has no root.

0, a7id

0,

the equation is

an

identity.
it is

If the equation has fractional coefficients, to begin

usually best

by multiplying both members by the lowest common denominator of these fractions. This process is called clearing
the equation of fractions.

We then reduce the equation to the form ax = h hj transposing the unknown terms to the first member and the known terms to the second, and collecting the terms in each member,
To
verify the result, substitute
it

for

x in the given equation,

2x
Example
1.

Solve

X
(4

x).

To
or

clear of fractions, multiply both

members by

the Led.,

6.

Then
Transpose and
Therefore
Verification.

= x - 6(4 - x), 4x-3x + 6 = x-24 + 6x. collect terms, 6x = 30. x = 5.

2)

4x - 3(x -

- = - 2
q.

(4

5).

Example

2.

Sol'^e

mx + n px +

p)x = q n. Transpose and collect terms, {m Hence if m ?t p, the equation has the single root
If

(q

n)/{m p).

If
satisfies
it.

m=p m =p

and q and q

-^

11,

it

has no root.
is

n,

it

an identity and every value of x

SIMPLE EQUATIONS
Example
Expand,
Cancel
x^,
3.

119

Solve
x^

(x

(a

+ a) (x + 6) = + 6) x + a6 =
(3a

(x

x-

a)^.

ax

a?.

and transpose and

collect terms.

Then
and therefore

6)x

a^

a6,

=
3a +
6

Sometimes a root of an equation can be found by

inspection.

349

The equation
tion, for it

then completely solved if it be a simple equacan have no other root than the one thus found.
is

Example.

Solve (x
is

a)2

(x

6)2

(a

6)2.

Evidently this
identity (6

a simple equation, and

(a

when x =
is 6.

it

reduces to the

a)2

6)2.

Hence

its

root

roots of an equation of the form AB = 0, in which A denote integral expressions of the fii'st degree in x, can be found by solving the two simple equations ^ = and

The

350

and

B 0, 341. In like manner, when A, B, C are of the BC and A'^ B'^ 0, degree, the roots of ABC be found by solving simple equations, 341, 343.

first

AC =
x

may

Example

1.

Solve (x
is

2) (x

+ 3)

(2

5) (3

2)

= =

0.

This equation

equivalent, 347, to the four equations

0, 2,

x-2 =
Hence
its

roots are
2.

+ 3 = 0, 2x-5 = 0, 3x + 3, 5/2, 2/3.

5

0.

Example

Solve 4 x2
lias

3 x2

7 x.

This equation

the

same roots as the two equations and 4x 5 = 3x + 7. X =

and
12.

Its roots are therefore

EXERCISE V
Solve the following equations.
1.

2. 3.

= 2x + (5-3x). (7 -5x) + 3)-4x(x-5) = 3x(5-x)-16. x(x - (x + 3) (x + 4) = 0. (x + 1) (X + 2)

15-

'

120
^

A COLLEGE ALGEBRA
X

5.
6.

x-2[x-3(x + 4)-5] = 3^2x-[x-8(x-4)]^ -2.

SIMPLE EQUATIONS
tlie

121

remaining unknown numbers in terms of x and to form a

single equation connecting the expressions thiis obtained. This equation is the statement of the problem in algebraic symbols.

We

solve

it

for x.

If the

be the value thus found for

values of the other
It

problem have any solution, it will x, together with the corresponding

thus found for x
is

unknown numbers. may happen, however, that the value

352

For the problem not an admissible solution of the problem. may be one which imposes a restriction on the character of
the

unknown numbers,

tion in

as that they be integers, and the equaX into which the statement of the problem has been

translated does not express this restriction. Having solved the equation in x, therefore,

we must

notice

whether the result

accept it that the problem

of the kind required before we If it is not, we conclude as a solution of the problem.

is

number

is

an impossible one.

Example 1. The sum of the digits of a certain number of two digits is 12. If we reverse the order of the digits we obtain a number which is 4/7 as great. What is the number ?
Here there are four unknown numbers, namely, the tens
digit,

the

units digit, the value of the number as it stands, and the value when the but all four can be readily expressed in terms of digits are reversed either units or tens digit.
;

Thus,

let

X
12 10 X

the tens digit.

the units digit,

Then

(12

x= x) = +
X

value of required number, value with digits reversed.

10 (12

x)

By

the remaining condition of the problem,

we have
(1)
less

Solving this

[10 + (12-x)]. equation we obtain x = 8, which being an integer

10(12

-x) + x=:

than
12

10, is

an admissible solution of the problem.

The

like

is

true of

X or

4.

Hence the required number

is

84.

Thus,

Notice that with a slight modification the problem becomes impossible. if we require that reversing the digits shall double the value of the
(1),

number, we have, instead of

10(12

the equation
(12

-X) + x:=2[10x +

-x)].

(2)

122

A COLLEGE ALGEBRA
is

And solving (2) we obtain x 32/9, which being fractional admissible solution of the problem.
353

not an

When dealing with a problem which has to do with certain magnitudes, as intervals of time, remember that the letters
used in stating the problem algebraically are to represent
not the magnitudes themselves, but the numbers which are Care their measures in terms of some given iniit or units.

must
of the

also be taken to express the measures of all magnitudes same kind, whether known or unknown, in terms of the

same

unit.

Example 1. A tank has a supply pipe A which will fill it in 3 hours, and a waste pipe B which will empty it in 3 hours and 40 minutes. K the tank be empty when both pipes are opened, how long will it be before
the tank
is full ?

Let X denote the number of hours required. Then 1 /x is the part filled in one hour when both A and B are open. But were A alone open, the part filled in one hour would be 1/3. And were B alone open (and water in the tank) the part emptied in one hour would be 1/3| or 3/11.
TT Hence

X
x

113 =
3

or

11

33

Therefore

= 33/2

hours, or 16 hours 30 minutes.

crew can row 2 miles against the current in a certain river in 15 minutes with the current in 10 minutes. What is the rate of And at what rate can the crew row in dead water ? the current ?

Example

2.

Let X

rate of current in miles per minute.

is

the rate of the crew against the current minute, in dead water it would be 2/15 + x.

As

2/15

in

miles per

And

as the rate of the crew with the current

it

is

2/10. or 1/5 in miles

per minute, in dead water

would be 1/5 -

x.

Hence
whence

-j!^

+
-\-

x,

x
-^^
3.

=
=

(miles per minute),

and

I (miles per minute).

Example and minute hands

Let X

At what time between two and

of minutes past

three o'clock do the hour

of a clock point in opposite directions ?

= number

two o'clock

at the time required.

SBIPLE EQUATIONS

123

Since the minute hand starts at XII it will then have traversed x minute spaces. The hour hand starts at II, or 10 minute spaces in advance of the minute hand, but it moves only 1/12 as fast as the minute hand. Therefore when the minute hand is at x minute spaces past XII, the hour hand is at 10 + x/12 minute spaces past XII.

But by the conditions of the problem, at the time required the minute hand is 30 minute spaces in advance of the hour hand.

Hence
or solving,

x
x

= Ao +

) + 30,
43j''j-

43/3-

minute spaces.
minutes after

Therefore the hands point in opposite directions at two o'clock, or I6j\ minutes before three o'clock.

Sometimes

in the statement of a
a, b,
a, h,

bers are denoted by letters, as will then be an expression in

c.

problem the known numThe value found for x c y^Yach. may represent an

354

admissible solution of the problem for certain values of these The discussion of the followletters, but not for others.

ing problem,
this point.

known

as the problem of cou?'iers, will illustrate

Example.
in the

Two

couriers

and

same

direction at the rates of

are traveling along the same road and n miles an hour respectively.
if

B
so,

is

now d when ?

miles in advance of A.

Will they ever be together, and

Let X

the

number

will then

of hours hence when they will be together. have traveled mx miles, and B nx miles and since
;

is

now d

miles in advance of A,

we have

mx = nx +
whence
and therefore
1.

d,

(1) (2)
(3)

{m,

n)x x

= =

d,

m n hours hence.
must be
positive
;

If

is

to overtake B, this value of x

and

since

by hypothesis

m> n.

d, ?n, n all denote positive numbers, Which corresponds to the obvious fact that

this requires
if

that

is

to overtake

B, he must travel faster than B does. 2. At the same time we can interpret the negative value which x takes if we suppose and B were together d/(n m) as meaning that

m<

hours ago.

124

A COLLEGE ALGEBRA
m
m

= n, we cannot, properly speaking, derive (3) from (1), since 3. If n wliich is 0. Bat we can derive the process involves dividing by differs at all from n, it matters not how little. And if (3) from (1) if

in (3)

we regard m as a variable, which while greater than n is continually n) becomes a variable approaching equality with n, the fraction d/{in which continually increases, and that without limit, 510. All of which

corresponds to the obvious fact that the smaller the excess of A's rate will never over B's, the longer it will take A to overtake B, and that

overtake
4.

if

his rate be the

if

Finally,

same as B's rate. we suppose both in = n and d

x.

0,

the equation

(1) is

satisfied
if

by every value of

Which corresponds
same
rate

to the obvious fact that

A and B

are traveling at the

and are now together, they

will

always be together.

EXERCISE VI
of the digits of a certain number of the order of the digits be reversed, the number
1.

The sum
is

two
is

digits

is

14.

If

increased by 18.

What
2.

the

number

By what number must

?

156 be divided to give the quotient 11 and

the remainder 2
3.

There are two numbers whose difference is 298. And if the greater What less, the quotient and remainder are both 12. are the numbers ?
be divided by the
4.

The tens

digit of a certain
1

number

of

two

digits is twice the units

digit.

be added to the tens digit and 5 to the units digit, the number obtained is three times as great as if the order of the digits be first reversed and then 1 be subtracted from the tens digit and 5 from
if

And

the units
5.

digit.

What

is

the

number

If 2

be subtracted from a certain

number and

the remainder be

multiplied by 4, the same result is obtained as if twice the number and half a number one less be added together. What is the number ?
is now four times as old as his son. If both he and his 20 years longer, he will then be twice as old as his son. What are the present ages of father and son, and how many years hence will the father be three times as old as the son ?

6.

father

son

live

7.

tank can be

in 2 hours,

tank

if it

filled by one pipe in 3 hours, and emptied by a second and by a third in 4 hours. How long will it take to empty the start full and all the pipes are opened ?

SIMPLE EQUATIONS
;

125

8. A and B can do a certain piece of work in 10 days but at the end of the seventh day A falls sick and B finishes the piece by working alone for 5 days. How long would it take each man to do the entire

piece,
9.

working alone

At what time between eight and nine o'clock do the hands watch point in the same direction ? in opposite directions ?
10.

of a

How
?

soon after four o'clock are the hands of a watch at right

angles
11.

In a clock which

is

not keeping true time

it is

observed that the

interval between the successive coincidences of the

is

hour and minute hands

66 minutes.
12.

What

is

the error of the clock (in seconds per hour) ?

Four persons, A, B, C, D, divide $1300 so that much as A, C I as much as B, and D | as much as

does each receive
13.
I

B
C.

receives f as

How much

A man

of the remainder

leaves | his property and .f 1000 besides to his oldest son ; and .flOOO besides to his second son I of the sum
;

remaining and $1000 besides to his youngest son. remain, what is the amount of the entire property ?
still

If

$3500

still

//^

14.

If 2

feet

increased by 100 square feet.

be added to both sides of a certain square, What is the area of the square

its

area

is

15. The height of a certain flagstaff is unknown ; but it is observed that a flag rope fastened to the top of the staff is 2 feet longer than the staff, and that its end just reaches the ground when carried to a point 18

feet distant
16.

from the foot of the

staff.

What

is

the height of the staff

purse contains a certain number of dollar pieces, twice as many If the total value of half-dollar pieces, and three times as many dimes.
the pieces
17.
is

$11.50,

how many

pieces are there of each kind ?

invests $5000, partly at 6% and partly at. 4%, so that the average rate of interest on the entire investment is 5J%. What sum does he invest at each rate ?
18.

A man

30
26

cts.
cts.

In what proportions should two kinds of coffee worth 20 cts. and a pound respectively be combined to obtain a mixture worth
a pound
?

and copper contains 2 parts of copper must be melted with this alloy to obtain one which contains 3 parts of silver to 7 of copper ?
19.

A pound

of a certain alloy of silver

silver to 8 of copper.

How much

126
20.

A COLLEGE ALGEBRA
If a certain quantity of

liquid, it contains

added,

it

30% of alcohol contains 20% of alcohol.

water be added to a gallon of a given if twice this quantity of water be How much water is added each time,
;

and what percentage

21.

of alcohol did the original liquid contain ?

A train whose rate of motion is 45 miles per hour starts on its trip

train

from Philadelphia to Jersey City at 10 a.m., and at 10.30 a.m. another whose rate is 50 miles an hour starts on its trip from Jersey City to Philadelphia. Assuming that the two cities are 90 miles apart, when will the trains pass each other, and at what distance from Jersey City ?
22.

If two trains start at the times mentioned in the preceding examand pass each other at a point half way between Jersey City and Philadelphia, and if the slower train moves | as fast as the swifter one, what are their rates, and when do they pass each other ?

ple

is now a distance equal to 50 of her leaps ahead of a fox pursuing her. How many leaps will the rabbit take before the fox overtakes her if she takes 5 leaps while the fox takes 4, but 2 of

23.

A rabbit

which

is

the fox's leaps are equivalent to 3 of her leaps ?

24. If 19 ounces of gold weigh but 18 ounces when submerged in water, and 10 ounces of silver then weigh 9 ounces, how many ounces of silver and of gold are there in a mass of an alloy of the two metals

which weighs 387 ounces

25.

in air

and 351 ounces

in

water

traveler set out on a journey with a certain

sum

of

money

in

his pocket

besides.

At

and each day spent \ of what he began the day with and $2 the end of the third day his money was exhausted. How
at the outset ?

much had he
26.

of a certain pyramid is a square, and the altitude of each of the triangles which bound it laterally is equal to an edge of the base. Were this edge and altitude each increased by 3 inches, the area of the

The base

pyramid would be increased by 117 square the pyramid ?

27.

inches.

What
of

is

the area of

The sum

of the digits of a certain

number

two

digits is a.

If

the order of the digits be reversed, the number is increased by b. is the number? Show that the solution is admissible only when

What

and when both 9a

28.

and 9a

9a^6
Is
if

b are exactly divisible

by

18.

Two
?

persons

there a time
so,

when

and B are now a and b years old respectively. was or when A will be c times as old as B, and
a, b, c,

when

Discuss the result for various values of

as in 354.

SIMULTANEOUS SIMPLE EQUATIONS

IV.

127

SYSTEMS OF SIMULTANEOUS SIMPLE EQUATIONS

SIMULTANEOUS EQUATIONS

conditional equation in two

unknown

letters, as

x and

y,

355

will be satisfied
letters.

We
is

The

like

pairs of values of these call eveiy such pair a solution of the equation. true of an equation in more than two unknown

by

infinitely

many

letters.

Thus, the equation 2 x

soever to

?/

(1) is satisfied if

we

give

X and the corresponding value of substitute any number, as h, for x and 3 26

identity 2 6

2x

to y.

any value whatFor in (1)

Thus, x
of
(1).

+ (3 - 2 6) = 3. = 0, y = 3; x =

for y,

and we have the true

1;

l,

\; x

2,

?/

are solutions

under consideration, the 2, and y any value whatsoever ; in other words, the equation x = 2 then has an infinite number And the like is true of any equation which involves but of solutions. one of the unknown letters.
Note.

equation x

When two unknown letters, x, y, are = 2 means that x is to have the value

356

It is therefore natural to inquire whether there may not be pairs of values of x and y which will satisfy two given equaSuch pairs usually exist. tions in these letters.

357

when x =

Thus, both the equations 2 x + ?/ = 3 and 4x + Sy = 5 are 2 and ?/ = - 1; for 2-2 + (- 1) = 3, and 4-2 + 3(-

satisfied
1)

= 5.
358

Simultaneous equations.
certain

Two

or

more equations involving

unknown letters are said to be simultaneous when each unknown letter is supposed to stand for the same number in
all

the equations.

Thus, the equations 2x + y = S (1) and 4x + 3?/ = 5 (2) are simultaneous if we suppose x to denote the same number in (1) as in (2), and

similarly.
It is

equations. in X and y.

not necessary that all the unknown letters occur in every one of the Thus, x = 2, y = 3 constitute a pair of simultaneous equations

128
359

A COLLEGE ALGEBRA

are

Generally speaking, the supposition that certain equations simultaneous is allowable only when the number of
is

equations
letters.

equal

to,

or less than, the

number

of

unknown

since x

Thus, the two equations x = 2 and x = 3 cannot be simultaneous, must denote different numbers in the two.

360

solution of a

of values of the

system of simultaneous equations is any set unknown letters which will satisfy all the

equations of the system.

Thus, x

2,

2/

= -lisa
2x +

solution of the system

3,

ix + 3y =

6.

361

To

solve a

solutions or to prove that

system of simultaneous equations it has no solution.

is

to find all its

362

The reasoning on which the

that described

process depends

is

similar to

and y we begin by supposing that x and y actually have values which satisfy both equations. On this supposition the equations may be
treated like identities and the rules of reckoning applied to them. By aid of these rules we endeavor to transform the

illustrated in 334, 335. Thus, in the case of a pair of equations in x

and

equations into one or more pairs of equations of the form X = a, y = b. If the process by which such a pair x = a, y = b has been derived is reversible when x, y have the values
a, 6,

we may
;

at once conclude

that
is

a, b is

one of the soluit

tions sought and the process reversible steps.

reversible if

consists of

The only new

363

principle involved in all this

is

the following

.-

Principle of substitution. Tf from the. supposition that all the given equations are actually satisfied it follows that the values of a certain pair of expressions, A and B; are the same, the one expression may be substituted for the other in any of the

equations.

SIMULTANEOUS SIMPLE EQUATIONS

Example.
Solve the pair

129
(1)
(2)

2x + y =
2/

12,
8.

the supposition that x and y actually have values which satisfy both equations it follows that the value of ?/ in (2) and therefore in (1) is 8.

From

Substituting this value, 8, for y in

(1),

we obtain
(3)

whence
Therefore,
if (1), (2)

2x+8 = =
a;

12,
2.

(4)
is

have any solution, that solution

2,
?/

But conversely, x process from (1), (2)

Thus,
(3)
1.

=
to

is

a solution of

(1), (2),

x = 2, y = 8. inasmuch as the

(^4),

(2) is reversible.

follows from

(4),

and then

(1)

from
is

(3), (2).

Note

This principle of substitution

a consequence of the several

364

rules of equality, 240, 253, 257, and of the general rule of equality, If a = 6, and b c, then a = c, ^ 261.

Thus,
follows
If
7/
:

we may prove our

8,

right to

make

the preceding substitution as

8

=
if

then

y4-2x =

+ 2x, or2x +
and
2

= 2x +

7/,

249.

And
Note

2X

y,

?7

12,

then 2 x

12, 261.

can be applied only when we have a right to suppose the given equations to be simultaneous. Thus, from x = 2 and x = 3 we cannot draw the absurd conclusion
2.

Of course

this principle

365

3,

because

we have no

right to suppose x

2,

3 simultaneous.

TRANSFORMATION THEOREMS
In view of what has just been said we may regard any correct application of the rules of reckoning to a pair of equations as a legitimate transformation of the pair and if
;

366

such a transformation be reversible, we may conclude that leaves the solutions of the pair unchanged.
tions in

it

Hence the following theorems, which hold good any number of unknown letters.

for equa-

Theorem 1. The solutions of a pair of equations remain unchanged when the trarisformutions o/ 338, 339 nre applied
to the

367

equations sej^arately.

130
For the solutions
such transformations.

A COLLEGE ALGEBRA
of the individual equations

remain unchanged by

Thus, the pair of equations 3 x has the same solutions as 3x

2 = 2y =
?/

1
1

and y and

2x
?/

= =

5 5

2 x.

368

Theorem

2.

The jjair of equatio7is j

X,

f(x,y)=0
f(x,X)=0.

has the same solutions as the pair

y
Here
y,

= X,

X denotes any expression in x alone (or a constant), /(x, y) any expression in x and and/(x, X) the result of substituting X for y in
/(x,
2/),

280.
is

The theorem

merely a special case of the principle of substitution.

Thus, the pair of equations y has the same solutions as y

= x + 2 and = x + 2 and

3x 2y
3
.x

2 (x

2)

1
1.

369

Theorems.

The pair of equations

A=
has the same
solntiotis as the

B,

C
C

=D
=
D.
as

pair
D,

A+C=B+
For

A=

B^

C=

I)

has

the
-\-

.same
C,

.solntions

A + C=B +
~

C,

C = D, % 338, and A + C = B A + C = B ^ D, G = D, % 303.

Thus, the pair
has the same solution as x

C= +
y

D
5

has the same solutions as

?/

(x

and therefore as

= = 2x =
?/)

and x

y
y

and x

= =

\
1,
.

and x - y =\.

370

Corollary.

Before applying the theorem of

3G9 we may,

equations
constants

without changing their solutions, multiply both the given that is, both members of each equation by any

we

please, except 0.

Hence

If k and

denote any constants except 0, the pair of equation},

A=

B,

C
C

=D
= D.

has the same solutions as the pair

kA IC = kB ID,

SIMULTANEOUS SIMPLE EQUATIONS

Theorem
equations
4.

131
371

When
*

-d

A, B, and _a

are integral, the pair of

= =
0,

has the same solutions as the two pairs

A = 0,
For Ali
i^

and B

0.
J.

^ has the

same

solutions as the

two equations
are the

and

jointly, 341.

Hence the solutions

of the pairs J.

of the pair

AB = 0,
0,

C=

same as those

0,

C=

and

iJ

jointly.

Thus, the solutions of xy

are that of

together with that of

= = =

and x
and x

+
+

y
y
?/

and x

= =

2
2,
2.

Two systems of simultaneous equations Equivalent systems. are said to be equivalent when their solutions are the same.
Thus, the pair of equations x
pair

372

3x +

2/

5,

4x +

3?/

10,

Again, the pair xy = 0, x + y = 2 and y = 0, x + y = 2.

2?/ = 5, 2x + ?/ = 4is equivalent to the both pairs having the same solution 1, 2. = 0, ?/ = 2 is equivalent to the two pairs x

ELIMINATION.
Elimination.

SOLUTION OF A PAIR OF SIMPLE EQUATIONS

To

eliminate an

unknown

letter, as x,

from a

373

pair of equations is to derive from this pair an equation in which X does not occur.

We proceed to explain the more useful methods of eliminating X ov y from a pair of simple equations in x and y, and of deriving the solution of the equations from the result.
Method
of

of substitution.

This method

is

based on the theorem

374

368.
Solve

Example.

32/
52/

3x +
Solving
(1) for

= =
= = =

3,
l.

(1) (2)

in

terms of

y,

x
3 (3

3
1.

?/.

(3)

Substituting 3

?/

for x in (2),

?/)

?/

(4)
(5)

Solving

(4),

2/

2.

Substituting 2 for y in (3),

= - 3.

(6)

132
Hence
For,
solution,

A COLLEGE ALGEBRA
the solution,

and the only one, of

(1), (2) is

= 3,

?/

2.

bj'

367,

3(J8, (1),
a;

the following pairs of equations have the

(2); (3), (2); (3), (4); (3), (5); (6), (5);
3,
?/

namely:

same and the

solution of

(6), (5) is

= 2.
directly

The same conclusion may be drawn

process from
(1),

from

362.

For the

(2)

to (5), (6)

is

reversible.
(1)

Verification.

-3 +

3-2

= 3,

(-

3)

1.

(2)

Here

(4)

was obtained by eliminating x by

substitution.

'

To eliminate an ^inhnown letter, as x, from a pair of equations by substitution, obtain an exjyression for x in terms of the other letter (or letters) from one of the equations, and then in the other
equation replace x by this expression.

375

The following example illustrates a sjDecial form of this method, called elimination by comparison.
Example.
Solve

x
X

+ +
?/,

5?/

7,

(1) (2)

67/=:8.

Solving both

(1)

and

(2) for

x in terms of

= l~by,

(3)

a;

Equating these two expressions for

Solving
(5)
(3),
(1), (2) is

x,

?/

2/

Substituting 1 for y in

= 8-G7/. = 8 6 y. = 1. = 2.

(4) (5) (6)

(7)

Hence the

solution of

2,

?/

1.

376

Method

of addition or subtraction.

This method

is

based on

the theorem of 369, 370.

Example.
Solve

2x-62/ =

7,

Multiply
Multiply

(1) (2)

by

3,

Subtract

(4)

by 2, from

(3),

Whence,
Substitute

- 1/2

iov y in (1),

2x

Whence,

Hence the solution

of (1), (2) is x

2,

2/

1/2

SIMULTANEOUS SIMPLE EQUATIONS

133

For hy 367, 368, 370, the following pairs of equations have the same namely (1), (2) (1), (5) (1), (6) (7), (6) (8), (6) and the solution of (8), (6) is x 2, y = 1/2.
solution,
: ;

Verification.

- 6(- 1/2) = 7,
subtraction.

(1)

+ 4((1), (2)

l/2)

= 4.

(2)

Here x was eliminated by

We

y by addition.
Multiply Multiply

can also find the value of x directly from Thus,

(1) (2)

by eliminating

by by

2,

4x

12 y
12 y

3,

9x +
13x
x
letter,

= =

14.
12.

(9)

(10) (11)

Add

(0)

and

(10),

=26.

Whence, as

before,

=2.

(12)

To eliminate an unknown

as x,

from a pair of simple

of x in the resulting

equatiotis by addition or subtraction, Tnultiply the equations by

numbers which will maize the

coefficients

Then subtract or equations equal numerically. as these coefficients have like or unlike signs.
Exceptional cases.

add according
377

Let
y.

A =0, B
Q,

denote a pair of simple

07ie

equations in x and show that this pair

Tlie preceding sections, 374, 376,

B=

Q has

solution

and but

one,

unless the expressions A and B are such that in eliminating x we shall at the same time eliminate y. This can occur in the

following cases only. 1. If the expressions

.1

and B are such that

.1
.-1

denotes a constant, we say that the equations are not independent.

= =

kB, where k

and

B=

Evidently
of

if

A =

kB, every solution of

0,

and

vice versa, so that the pair

= ^ =

is 0,

a solution

B=

Q has

infinitely

many solutions. = x + 32/-5 = Thus, let^=2x + 6?/-10 = (1), andB Here ^ = 2 2?, so that ^ = and B = are not independent.
if

(2).

that
(1)

to eliminate

x we multiply

(2)

Observe by 2 and subtract the result from

we
2.

at the

same time eliminate

y.

If

A and B

are such that

I

A = kB +

I,

where k and

denote constants,

not

0,

we

and

^=

say that the equations

^ =

are not consistent.

134

A COLLEGE ALGEBRA

In this case the pair A

values of x, y that
Thus, \etA

make 5

= 0, B = = will
(3), a.nd

has no solution

make A

for any

I,

not

0.

Here

= 2x + Qy-9 = A = 2 B + 1, so that ^ =
(3), (4)

and

B = x + 3y - 5 = (4). B = are not consistent. If we

y.

eliminate x from

we

shall at the

same time eliminate

378

Formulas for the

solution.
x,
c,

We may
(1)

reduce any given pair of

simple equations in

y to the form
a'x

ax-\-by

b'y

c',

(2)

where

a,

h, c, a', b', c'

denote

known numbers

or expressions.

By 377, the pair (1), (2) has one solution, and but one, ka and b' unless a constant k can be found such that a' kb,

and therefore

ab'

a'b

=k(ab

ah)

0.

To obtain

this solution, eliminate

y and x independently by
results are
(4)

the method of subtraction,

376.

The

- a'b) x = b'c~ be', (3) {ah' - a'b) y = ac' - a'c. (ab' a'b ^ 0, the solution of (1), (2) is Therefore, if ah' _ ac' a'c _ b'c be' "^^ y^
ab'

-a'b'

ab'-a'b

^^

These formulas are more ^

easily

remembered
.

if

written

y = = -^ ca' c'a ab' a'b be' b'c

Did we not know
ab'

(6) ^ -^

a'b -^ 0, the

in advance that the pair (1), (2) has a solution when argument here given would only prove that if the pair
it is

(1), (2)

has any solution,

(5).

SIMULTANEOUS SIMPLE EQUATIONS

rx/.3

135

5?/

3|,

L5x + 3y =
g

1.65.

M
10.
^
f

r2{2x + 3y) = 3(2x-3y) + 10,

4x-32/ =

4(6 2/-2x)4-3.

+ 2)(2/ + l) = (x-5)(2/-l), |(x = - 2/(8 -X). Lx(4 +

?/)

fax
lbx

+ +

by

ay

= =

aa^

+ 2a + b^, + 2b + b^.

az
11.

+
y

by

c,

px =
(x

12.

(a-6)x +
(a

^
I

qy.

= 2(a2-62), (rt + 6)?/ + 6)x + (a-6)2/ = 2(a2+62).

'

+
^

'

5,

jC-y
14.

_ x + 2y-5 _ y-3 _ y + 2x-5 ~

6 4

13.

lUx
1

-2y + Q-0.
V
f b
2/

x a
15.

w
f
6

_,
c

X a
16.

+
+

= 1+ X,
1

X
a'

y
b'

2_
c'

X V
L

+y.

17.

Show

that the following equation.s are inconsistent.

Ilx -2i,y
18.

10,

Gx - lOy =

15.

In Ex. 15 assign values to a, b, c, a', b', c' for which the equations are (1) not consistent, (2) not independent.

PAIRS OF EQUATIONS NOT OF THE FIRST DEGREE WHOSE SOLUTIONS CAN BE FOUND BY SOLVING PAIRS OF SIMPLE EQUATIONS
pair of equations which are not of the first degree with respect to x and y may yet be of the first degree with respect to a certain pair of functions of x and y. can then solve the equations for this pair of functions, and from the result it

379

We

is ofiten

possible to derive the values of x and y themselves.

2 5 Solve - H X 32/
9 -

Example

1.

1,

10 =

5.

y
1

Both equations are of the first degree with respect to 1 /x and Solving for 1/x and \/y, we find l/x= 1/3, \/y= 1/5.
X

/y.

Hence

3,

2/

5.

136

A COLLEGE ALGEBRA
2.

Example

Solve 3a;

6,

7x

=
I,

1.

Solving for x and y /x,

we

find x

y/x =

3.

Hence

a;

1,

3.

380

Given a pair of equations reducible to the form AB = 0, A'B' = 0, where A, B, A', B' denote integral expressions of the It follows from the theorem of 371 first degree in x and y.
all the solutions of this pair can be obtained by solving the four pairs of simple equations ^=0, A' ^=0, 0. B' 0, A' 0, B' 0; B 0; B

that

Example.

Solve
(x

x--2xy = 0, + y-l){2x+y-S)=0.
+ + x + 2x +
X
2X
;

(1)

(2)

This pair

is

equivalent to the four pairs

= 0, ^ 0, x-2y = 0, x-2y = 0,
a; a;

2/

?/

= =

0,

(3) (4)

0,

y-l=0, y -3 = 0.
we

(5)
(6)

Solving these four pairs

of
(1), (2),

(3), (4),
1

(5), (0)

obtain the four solutions

namely

x,

= 0,

0,

2/3, 1/3; 0/5, 3/5.

381

And,

in general, if

ABC

and A'B'C

m and

integral factors of the first degree in x solutions of the pair of equations 0,

?i

denote products of and y, all the

ABC

A'B'C

can be found by solving the mn pairs of simple equations obtained by combining each factor of the first product equated to with each factor of the second likewise equated to 0.

and consistent, we thus obtain mji

of solutions of the given epilations

If all these pairs of simple equations are both independent solutions, that is, the number
is

the product of their degrees.

EXERCISE Vin
Solve the following pairs of equations.

2x
2.

3y
(\

10x

5,

15x

in =

8.

SIMULTANEOUS SIMPLE EQUATIONS

9-

137

y X

-'-

2(3

-y)
X

r ,

3
)

+
X

= ^y-^
X
2)

1-

1.

2
{x
0,

4. 5. 6. 7.
8.

xy

0,

+ 2y-l){Sx-y +
3x
y)
8?/

0.

xy-y =

0.

x(x~y){x +

9.

10.

= 0, x+2y -5 = 0. = 0, (X - 2) - 3) = 0. (x-l)(y-2) = _ 1)2^ 2x + 3 2/-7 = 0. (2x + 2/)-2={x-3 + 5)2, (x + 2/)2 = 1. (x-52/ + 8)(x + 32/ + 5) = 0, (2x + + 5)(5x +
(2/

2/2

(a;

2/

?/

?/

14)

0.

GRAPHS OF SIMPLE EQUATIONS

Graph
of a pair of values of

IN
y.

TWO VARIABLES
It is

x and

convenient to

382

represent jo/?w of values of two variables, as x and y, by points in a plane.

lines,

In the plane select as axes of reference two fixed straight X'OX and Y'OY, which meet at right angles at the point O, called the origin; and choose ^
for

some convenient unit

lengths.

measuring

B
if

Then
follows
:

the given pair of vala,

I I
I

ues be X

h,

proceed as

(
1
I

to the right or left of 0, according as a is positive or

On X'OX and

x-

negative, measure off a segment, OA, whose length is \a\, the numerical

value of

a.

.Y'

Similarly on
is

Y'OY and above

or below 0, according as b

is

positive or negative, measure off a segment, OB,

\h\.

whose length

Then through A and B draw

respectively.

parallels to

Y'OY and X'OX

We

take P, the point in which these parallels

138

A COLLEGE ALGEBRA

intersect, as the point-picture, or graph, of the pair of values

a,

b.

= a, y b by the symbol (a, b). We call the number a, or one of the equal line segments OA or BP, the abscissa of P and b, or one of the equal segments OB or AP, the ordinate of P. And we call the abscissa
It is convenient to represent both the value-pair x
its

and

graph

and ordinate together the coordinates of P.

We

Y'OY the

also call A''OA' the x-axis or the axis of abscissas, y-axls or the axis of ordinates.
this

and

Observe that
correspondence,
pair
is

method brings the value-pairs of

the points of the plane
;

x,

2, v?ith
is

that

is,

y into one-to-one for each value-

(a, b)

there

one value-pair

(a, 6)

one point P, and reciprocally for each point P there found by measuring the distances of P from Y'OY

and X'OA" respectively, and

giving them their
priate signs.

appro-

of

In particular, the graph (0, 0) is the origin, that

a;-axis,

is a point on the and that of (0, h) is a point on the y-axis.

of (a, 0)

Example.
pairs
(4, 4),

Plot the value-

(-3,

3),

(-

5,

4), (3,

(-4,

0),

2).

Carrying

out. the construc-

tion just described for each

value-pair in turn, we obtain their graphs as indicated in

the

accompanying

figure.

Notice particularly of the coordinates.

how

the position of the graph depends on the signs

303

an equatk in x and y. If, as is commonly the equation ini x and y has infinitely many real solutions, there will usually be a definite curve which contains the graphs of all these solutions and no other points.
of
,

The graph

case, a given

We

call this

curve the graph of the equation.

SIMULTANEOUS SIMPLE EQUATIONS

139

But the graph of an equation may consist of more than one curve. Observe that we here include straight lines among curves.
Theorem.

of the

letters

The graph of every simple equation in one or both x and y is a strair/ht line.

384

On

this

account simple equations are often called linear equations.

readily convince himself of the truth of the theorem by selecting some particular Y " " equation and plotting a number of its
solutions.
(-L6)

The student may

Thus, take the equation y

=
3,

2 x

i.

When x=:0, = 4, we have

?/

1,

2,

'(0,4)

2,

0,

2,

And
(2,

0),

plotting these value-pairs (0, 4), (1, 2), 2) as in the accompanying (3,

(L2)

figure,

we

find that their

graphs

all

lie

in the

same

straight line.

X'i
:

jJ2^
(3*2)

We
1.

msiy prove the theorem as follows

When
a, or

the equation has the form

h.

'

Example.

Find the graph of x

is

2.

This equation

satisfied

by the value 2 of x and every value of y, 356. Hence the graph is a parallel to the ?/-axis at For this line conthe distance 2 to its right. tains all points whose abscissas are 2, and such
points only.

(x=2)

X^

'^

x a so, in general, the graph oi a parallel to the ^/-axis at the distance to the right or left according as a is |a|

And

is

of positive or negative and the graph lei to the x-axis at the i is a pa ?/

;

distance
as h positive or negative. In particular, the graph of
is
is

\h\

above or below according

is

the x-axis, and that of

ic

the y-axis.

140
2.

A COLLEGE ALGEBRA

When

the equation has the form y

Find the graph of y
is
;

= mx.

Example.

2z.

The graph
the point

(1, 2)

the right line which passes through the origin (0, 0) a,nd for this line contains every point whose ordinate is twice
its

abscissa,

and such points only.

in

And

so,

general,

the

p = mx is the right line which passes hrough the origin and the point (1, m).
3.

graph

of

When
-\- c.

the equation has the form

2/

= mx

Example.

Find the graph ot y

2x

3.

Evidently we shall obtain the graph of this equation if we increase the ordinate of every But that comes to the same thing point of the graph oi y = 2 x by S. as shifting the line y = 2x upward parallel to itself until its point of intersection with the
j/-axis is 3 units

above the origin.

general, the graph of a right line parallel to ?fta; and the graph of y meeting the 7/-axis at the distance \c\ from the origin,
in
is

And so, = mx + c

above or below, according as

tive or negative.

is

posi-

385

To

find this line.

As any two

of its

points suffice to determine a right line, we may find the graph of any equation, ax
in the following example.

+ by + c =

0,

as

Example.
First,

Plot the graph of 3

a;

0.

when y = 0, then x = 2. Hence we have only to plot the

xoill

Second,
points

when x =
and

0,

then y
that
is,

6.

(2, 0)

(0, 6),

the points

where the line

deternnne

meet the axes, and draw the line which these points

(see figure in 38()).

fails

This method

when

b,

mx.
2.

We

the equation has one of the forms x = a, then find the line by the methods explained in

384, 1

and

SIMULTANEOUS SIMPLE EQUATIONS

141
386

Graph of the solution of a pair of simultaneous simple equations. This is the point of intersection of the two lines which are the

graphs of the equations themselves

for this point, and this point only, is the graph of a solution of both

equations.
Thus, the solution of2a; 3y + and 3 X + 2/ - 6 = (2) is a; = 1, 2/

7=0
=
3.

(1),

And, as the figure shows, the graphs and (2) intersect at the point (1, 3).

of (1)

When
lines

consiste?it,

the given equations are not 377, 2, their graphs are A'-

387

which have no point in common,

thsitis,parallellines; when the equations are 7iot indej^endent, 377, 1,

their graphs are lines

which have

all their points in

common,,

that

is,

coincident lines.

2/

Thus, the equations ?/ = 2 x, 2 X + 3 are not consistent,

and the graphs of these equa^

tions, 384, 3, are parallel lines.

2/

Again, the equations?/ = 2x, = 6 X, which are not inde-

pendent, have the same graph.

The graph
tion of the

of an equa-

388

form

AB =
jointly
;

consists of the graphs of

and

B=

AB = Q are those of .4 = and B =

for the solutions oi
jointly, 341, 346.

Example.

Find the graphs of the equations

{4x
{X

+ y-7){Sx + 2y-11) = 0, - 2y + 5) {2x - 3y - 1) = 0,

(1)

(2)

and the graphs of their solutions.

142
The graph

A COLLEGE ALGEBRA
of (1) consists of the lines PQ and BS which are the graphs 7=0 and 3x4-2?/ 17=0 respectively.

of4x + y

The graph of (2) consists of the lines PS and QE which are the graphs ofx 27/-f5 = and 2x 3?/ 7 = respectively. The points P, Q, E, S in which the pair PQ, RS meets the pair
PS,

QR

are the graphs of the solutions of

(1), (2),

namely,

(1, 3), (2,

1),

(5, 1), (3, 4).

389

Graph

of

an equation of higher degree in x and y.

We

find a

number
and

of the solutions of the equation, plot these solutions, then with a free hand draw a curve which will pass
all

through

"near" enough

the points thus found. By taking the solutions together, we can in this way obtain a curve

which

differs

from the true graph

please.
it is

as little as

we

In work of this kind

con-

venient to use paper ruled into small squares, as in the accom-

panying

figure.

Example. Find the graph of the equation y = x^.

When

0,

1,

2,

3,
9,

4,

we have y = And when

0,

1,
1,

4,

16,

2,

3, 9,

4,

we have
Taking the

1,

4,

16,

side of a square as the

unit of length, plot the corresponding points (0, 0), (1, 1), (2, 4) ( 1, 1), A few of them suffice to indicate the general character of ( 2, 4) the graph, the curve in the figure, except between x = 1 and x = -I- 1.

wholly above the x-axis, extending upward indefinitely and it symmetrically with respect to the y-ax\s, the same value of y corre a. sponding to X = a and x
It lies
;

lies

SIMULTANEOUS SIMPLE EQUATIONS

EXERCISE IX
1.

143

Plot the following pairs of values of x and y.

(0, 0), (5, 0), (0,

7), (6, 2),

(-

7,

1),

(-

4, 3), (5,

9).

2.

Find the graphs of the following equations.

x

0,

?/

0,

2?/

0, 0,

3(/

7x + 32/-18 =
3.
a;y

+ x = 0, 3x-42/ =
1

z
24.

?/

0,

Find the graphs of the following.

0,

=
4.

(x

2/

- 3)

(X

2/)

0,

x2

0,

x2

2/2,

x^

2/2

0.

cal

Find the solutions of the following pairs of equations by the graphimethod and verify the results algebraically. rx
(1)

\
5.

+ 2/-3 = x-2?/ =
+
6)(x

0,
0.

= 0, f 32/ + 2x + 19 \ ^'l22/-3x + 4 = 0.
^A
of the following pairs.

Do

the

same with each

j'(x-42/
6.

32/

6)

=
5)

0,

J
0.

(y

2)x
2)2/

U3x + 2y-10)(2x-2/ +
Find the graphs
2/

l(2/-x +
V

= 0, = 0.

of the following

two equations.

=-

(ic

1)-,

'

SYSTEMS OF SIMPLE EQUATIONS WHICH INVOLVE MORE THAN TWO UNKNOWN LETTERS
Method
letters.

of solving a

system of n simple equations in n unknown pair of equations in three unknown letters will

390

ordinarily have infinitely

many

solutions.
;

Thus, the pair x = 2z, 2/ = z + l has infinitely many solutions for both equations are satisfied if we assign any value whatsoever, as 6, to z and the values 2 h and 6 + 1 to x and y.

But a system

of three simple equations in three

unknown

391

letters ordinarily has one, and but one, solution, be obtained as in the following example.

which may

Example.

Solve the system of equations

3x-22/ + 4z = 13, 2x + 5?/-32; = -9,

6X

(1) (2)

3y

2 2

7.

(3)

144

A COLLEGE ALGEBRA
:

Elimiuate z between two pairs of these equations, thus

Multiply
(1) (2)

by
by

3,

Multiply

4,

Add
Again,
(1) is (3) (7)

+ 8x + 20y17x + 14y
9x 6 y

12 2

39

(4)
(5)

122

=3
(6), (9),

=-36

(6)
(7)

Multiply
Subtract

by

2,
(8),

from

3x-22/ + 4z=13 12x + 6y + 42 = 14 9x + Sy =1

thus:

(8) (9)

Eliminate y between the resulting equations

Multiply
(6)

by

4,

68 x

by 7, Subtract (11) from

Multiply
(9)

63x
(10),

+ 56 y = + 56y=

12
7

(10)
(11) (12)

5x

=5
x
y

Hence
Substituting x
Substituting x

= =

1
1,

in (9),

we

find
(1),

1 in

we

find 2

= 1. = \. = 2.

- 1, Therefore, 362, (2), (3) has any solution, it is x = 1, y = But the process by which we have derived x = l, y = 1,2 = 2 2 = 2. from (1), (2), (3) is reversible. In fact, it may readily be traced backward step by step. Hence x = 1, y = 1, 2 = 2 is the solution of (1),
if (1),

(2), (3).

We may
of (1),
It is

also prove as follows that

l,

y=

1,

2 is the solution

(2), (3).

evident by 368 that x = 1, y = - 1, 2 = 2 is the solution of (12), We therefore have only to prove that the system (12), (9), (1) (9), (1). has the same solution as the given system (1), (2), (3).

posed to the

Let us represent the equations first members, thus

(1), (2), (3),

:

with the

known terms trans0.

^ = 0,
It will

(1)

B = 0,

(2)

C=
(9)

(3)

then follow from the manner in which

the equations
0,
(1), (9), (12)

and

(12)

were derived,

that

we may express
(1)

thus:
0.

A=0,
will

-^ + 2C =

(9)
x, y, z

19^ 4-16B-14C =
that

(12)

Evidently any set of values of

makes
14

J.

make

Conversely,
also 19 .4

^ = 0, - ^ + 2 C = 0, 19 ^ + 16 B when ^ = and A 2 C = 0, + 16 J5 - 14 C = 0, then B = 0.

-|-

then

C = 0. C=

= 0, B = 0, C =
;

and when

Hence the system

<9), (12),

(1), (2), (3)

has the same solution as the system

2.

(1),

namely,

x=l,2/ = -l,2 =

SIMULTANEOUS SIMPLE EQUATIONS

145
392

equations in the three

In the case just considered, from the given system of three unknown letters, x, y, z, we derived a
tivo

system of

equations in two letters,

cr,

ij,

and then from

this system, a single equation in one letter, x. And, in general, if we start with a system of n simple equa1 oi these steps, wg tions in n unknown letters and take n

shall arrive at a single equation in one of the letters, as x, of

the form ax

Then,
in

ufiless

h = Q. a = 0, the
may

which the value of x

letters

system has one, and but one, solution, is b/a and the values of the other
This

unknown

be found by successive siibstitutions in

the equations obtained in the process. proved as in the example.

may always

be

On
nitehj

the other hand, if a many solutions when

less laborious

=
ft

the system ordinarily has 0, and no solution when b

inji-

0.

This will be proved

in 394.

equations

method of solving a system of simple given in the chapter on deterviinants. In certain cases labor may be saved by special devices.
is

A much

393

Example.

Solve

Add And
we

(1), (2), (3), (4),

Hence

+ y + 2 = 8, + + M = 12, = 14, X + 2 + + z + u = 14, 3x + 3?/ + 3z + 3i( = x + +z+u=

x
a;

(1)

2/

(2) (3) (4)

tt

7j

48.
16.
(5)

?/

subtracting each of the equations obtain x = 2, y = 2, z = 4, m = 8.

(4), (3), (2), (1) in

turn from

(5),

B Q denote a system Let 0, C Exceptional cases. b of simple equations in x, y, z, and, as in 392, let ax denote the equation obtained by eliminating y and z.

A=0,

394

1.

If a

and

=
/>

0, it

will be found that one of the func:

tions A, B,

= kB +

be expressed in terms of the other two, thus and I denote constants. We then say IC, where

C may

that the equations

=:

0,

0,

are not independent.

146

A COLLEGE ALGEBRA
the identity

From
tion oi

A = IB

IC

it

B=

follows that every solu-

and C

is

a solution of

0.

Hence

if

B= A=

and C
0,

=
C

are consistent,

377,

2,

B=

the three equations

solutions.

0,

will

have

infinitely

many

Thus, consider the system of equations

^ = 3x-2y + 4z-13 =
i?=:2x

0, 0, 0.

(1) (2) (3)

+ 5y-32+9 = C=7x + 8?/-2z+5 =

(1)

Eliminating z between
,

and

(2),

3^1

+4 5 = 17 x +
(1)

14?/

-3 = 0.

(4)

Eliminating z between

and

(.3),

^ + 2C = 17x+ 14y-3 = 0.
Eliminating y between
(4)

(5)

and

(5),

2J.

4B-2C=
=

0-x

-0 =

0.

(6)

Here the
deriving
it,

final

we

+ i - 2 C 0, or C -\- 2 B. identity 2 And, in fact, we see on examining (1), (2), (3) that by multiplying by 2 and adding the result to A.

has the form x = 0, and in equation ax 5 = find that the expressions A, B, C are connected by the

=A

C may

be obtained

Hence the system

2.

(1), (2), (3)

has infinitely

many

solutions.

If

and

h =h 0, it will

be found that one of the func:

tions A, B,

C may

be expressed in terms of the other two, thus

A = kB -^IC
where
k,
I,

+ m,
0.

denote constants,

not

We then say that

the

are not consistent. equations A = 0, B = Q, C = !^rom the identity A kB -\- IC -\- m it follows that .4 = 0, B = 0, C = Q have no solution. For any values of x, y, z that

make S

and C

will

make A

m, not

.4

0.

Thus, consider the system of equations

^ = 3x- 2^ + 42-13 = 0, J?=2x + 52/-3z+9 = 0, C = 7x + 82/-2z+6 = 0.

(1) (2)

(3)

SIMULTANEOUS SIMPLE EQUATIONS

Eliminating z and y as above,

147

we

obtain
0.

2^ + 45-2C = 0x-2 =
Hence the
-4, i?,

6 = equation ax are connected by the identity

final

has the form

C = J. +
to

2 = 0,
And,

and

25+L
From

in fact,

on examining (1), (2), (3) we Hence the system (1), (2),

find
(.3)

tliis

be the case.

has no solution.

Systems
discussion

of

simple equations in general. the conclusion

:

the preceding

395

we may draw

simjde equations in n unknown Ordinarily a system of = n, infinitely many solutions has one solution when when n. n, no solution when
letters

m<

m m>

Whenever exceptions
377, 394.

to this

rule occur, two or

tions are connected hy identical 7'elations of the

more of the equakind described in

B=

In particular, a system of three simple equations, A = 0, 0, C = 0, in two unknown letters, x, y, has a solution

the form

when, and only when. A, B, C are connected by an identity of A = kB + IC and B = 0, C are consistent.
Thus, the system x
solution
;

\ (1),

1 (2),

3x - y = 10

(3)

has no

for the solution of (1), (2), namely, x

4,

2/

3,

does not

satisfy (3).

On the other hand, the system x-?/ = l(l),x + y = has a solution for (4) is satisfied by x = 4, y = 3. 3 X - 2/ - 9 = 2 (X - 2/ - 1) + (X + ?/ - 7).
;

7 (2), 3

x - y = 9 (4) But observe that

they

Let the student draw the graphs of meet in a common point.

(1), (2), (4).

He

will find that

EXERCISE X
Solve the following systems of equations.
'

x
2/

^2

+ y= +2= +X=

11, 13, 12.

2.

'x

x X

+ + 2 = 1, + 2y + 3z = + 3 + 7z=
?/
2/

4,

13.

3.

x-f22/-3z = 3, 3x3x - 5?/ + 7z = 19, [sx- 82/- llz = -

5x -22/
4.

X
X

13.

+ +

2/

7 z

= -33, = 13,
10.

32/

=-

148
'

A COLLEGE ALGEBRA
X +2?/
2 X
2/

2/

4z
1

4z = = 0, = 0.
2 z
9,

ll,
6.
.

'3x-5 =
(x

2(x-2),

+ l)(2/-l) = (x + 2)(2/-2) + 5, = [2x + 32/ + 2 6.

2/

6 __ 4 z
1

Z
1

32 _
y

-+

X
5,

8.

x Lx

4.

+ + + +

2
2
2/

2/

+ = 4, + M = 3, + M = 1, + 2 = 10.

z
'i{

4x 3z +
5?/
9.
<^

9,

+ z-4u = 32/ + u = 12, ,x + 2?/ + 3m =

Ix

ex
17,

&2/

10.

62/

+.az

az
8.

ex

= = =

m,
n.

11.

ax -{-by

= my = }i2, + cz =

d.

-c
14.

2x

32/

6z,

(x+22/ + 2-16)(3x-22/ + 20)=0.

Show
identity

that the following systems are not independent and find the which connects the equations of each system.

X
13.
2/

- = 3, - z = - 5, - X = 2.
2/

'3x

82/

+ 72 =
2/

10,

2x +

5y-3z = 12, - 2 = 80. 16 X + 9

PROBLEMS
396

The following problems can be solved by means of simple How equations in two or more unknown letters, a,s x, y, . many of these letters it is best to employ in any case will
depend on the conditions of the problem. But when a choice has been made of the unknown numbers of the problem which X, y, are to represent, and the remaining unknown
numbers,
if

ters, it will

any, have been expressed in terms of these letbe found that the conditions of the problem still

unused will yield just as many independent and consistent

equations connecting
fact, if
:r,

y,

as there are letters x,

i/,

In

they gave more than this

number

of equations, the

SIMULTANEOUS SIMPLE EQUATIONS

problem would have no solution
of solutions,

149

if less,

an infinite number

395.
of

The remark
of the problem

352 on the restrictions which the nature

the character of the

may impose on
also.

unknown

numbers applies here

Example
equal to the
1.

In a certain
of the first

number

of three digits, the second digit is

and third, the sum of the second and third digits is 8, and if the first and third digits be interchanged, the number is increased by 99. Find the number.
Let
X
the

sum

hundreds
is

digit,

tens digit, z

units digit.

Then

number

100 x

-{-

\0 y

z.

But, by the conditions of the problem,

we have
(1) (2)

x
2/

+ z = y, + z = 8, +
x

100 2

+
is

10?/

= =

100 a:
2,
?/

10?/
5,

Solving

(1), (2), (3)

we

find x

+z+ = 3.

99.

(3)

Hence

the

number
2.

253.

After walking a certain distance a pedestrian rests for continues his journey, but at | of his original rate, and on reaching his destination finds that he has accomplished the entire distance, 20 miles, in 6 hours. If he had walked 4 miles

Example

30 minutes.

He then

further at the original rate and then rested as before, the journey would have taken 5& hours. What was his original rate, and how far from the
starting point did he rest ?

Let X

from starting point

original rate in miles per hour, to resting place.

and

let

= number

of miles

actual journey, (2) the supposed journey,

Expressing in terms of x and y the number of hours taken by we have

(1)

the

2/12^-,^ 2
X

'

7x/8
i,

^'

|MJ
X
find

2^7x/8

16^^
]

'

^'

Solving

(1), (2)

for

y/x and 1/x, we

y/x = 3/2,

/x

1/4.

Hence x

6.

Example
water.

40%

of alcohol

3. Two vessels, A and B, contain mixtures of alcohol and mixture of 3 parts from A and 2 parts from B will contain and a mixture of 1 part from A and 2 parts from B will
;

150

A COLLEGE ALGEBRA
What
are the percentages of alcohol in

contain 32% of alcohol.

respectively
?

A and B

Let X and y denote the percentages of alcohol in Then by the given conditions we shall have
^ '

and

B respectively.

5
(1), (2)

100

100

'

Solving

we

find x

52/100, or 52%, y

22/100, or 22%.

EXERCISE XI
Find three numbers whose sum is 20 and such that (1) the first plus twice the second plus three times the third equals 44 and (2) twice the sum of the first and second minus four times the third equals 14.
1.

If the first number be divided 2. The sum of three numbers is 51. by the second, the quotient is 2 and the remainder 5 but if the second number be divided by the third, the quotient is 3 and the remainder 2. What are the numbers ?
;

digits from the following data (1) twice plus three times the second equals 37 ; (2) if the order of the digits be reversed, the number is diminished by 9.
3.
:

Find a number of two

the

first digit

4.

owes $5000 and

B owes

$3000.
;

could pay

all his

debts

if

besides his

own money he had | of B's and B could pay all but $100 How much money of his debts if besides his own money he had h of A's.
? 5.

has each

Find the fortunes of three men, A, B, and C, from the following data A and B together have p dollars B and C, q dollars C and A, What conditions nuist p, q, and r satisfy in order that the r dollars. solution found may be an admissible one ?
: ;

6.

sum

of

money

at simple interest

amounts

to $2556.05 in 2 years

and

to $2767.10 in 4 years.
?

What

is

the

sum

of

money, and what the

rate of interest
7.

sum of money partly in 4% bonds at par, partly in 5% bonds at 110, and his income from the investment was $050. at 80 and the 5% bonds at 110, his income from If the 4% bonds had been
invested a certain

A man

the investment would have been $100 greater.

8.

How much did

:

he invest

Find the area of a rectangle from the following data if 6 inches be added to its length and 6 inches to its breadth, the one becomes f of the other, and the area of the rectangle is increasetl by 84 square inches.

SIMULTANEOUS SIMPLE EQUATIONS

9.

151

money
left.

as

A 10. A

B as much money as B had then B gave A as much had left finally A gave B as much money as B then had then had $16 and B $24. How much had each origmally ?
gave
;

and

and C,

in 4| days.

together can do a certain piece of All three of them work at

work
it

in

5}-

days
it

for 2 days

when
take

A drops out and B and C finish it in If'j days. each man separately to do the piece of work ?
11.
circle

How

long would

Two

points

move

at constant rates along the circumference of a

whose length is 150 feet. When they move in opposite senses they meet every 5 seconds when they move in the same sense they are
;

together every 25 seconds.

12.
It

What

are their rates

trains

whose lengths are 240 yards and 200 yards respectively 25 seconds to pass one another when moving in opposite directions but were the trains moving in the same direction, it would take the faster one 3J minutes to pass the slower one. What are the rates of the trains in miles per hour ?
would take two freight
;

13.

Two

steamers,

and B, ply between the

cities

C and

D which

are

200 miles apart. The steamer A can start from C 1 hour later than B, overtake B in 2 hours, and having reached D and made a 4 hours' wait What are the rates of there, on its return trip meet B 10 miles from D.

and
14.

In a half-mile race

can beat

B by

20 yards and

C by

30 yards.

By how many yards can B

beat

15. and B run two 440-yard races. In the first race gives B a start of 20 yards and beats him by 2 seconds. In the second race gives B a start of 4 seconds and beats him by 6 yards. What are the rates of and B ?

A A

passengers together have 500 pounds of baggage. One pays If the bag$1.25, the other .$1.75 for excess above the weight allowed. gage had belonged to one person, he would have had to pay $4. How
16.

Two

much baggage
17.

is

allowed free to a single passenger

Given three alloys of the following composition A, 5 parts (by weight) gold, 2 silver, 1 lead B, 2 parts gold, 5 silver, 1 lead C, 3 parts To obtain 9 ounces of an alloy containing equal gold, 1 silver, 4 lead. quantities (by weight) of gold, silver, and lead, how many ounces of A, B,
:

and C must be taken and melted together

18.

A and B are alloys of silver and copper. An alloy which is 5 parts A and 3 parts B is 52% silver. One which is 5 parts A and 11 parts B is 42% silver. What are the percentages of silver in A and B respectively ?

152

A COLLEGE ALGEBRA
A

19. marksman who is firing at a target 500 yards distant hears the An observer distant 600 yards bullet strike 25 seconds after he fires.

from the target and 210 yards from the marksman hears the bullet strike 2yV seconds after he hears the report of the rifle. Find the velocity of sound and the velocity of the bullet, assuming that both of these
velocities are constant.

by two pipes, A and B, and emptied by a third and all the pipes be opened, the tank will be emptied in 3 hours if A and C alone be opened, in 1 hour; if B and C If A supplies 100 more gallons a minute alone be opened, in 45 minutes. than B does, what is the capacity of the tank, and how many gallons a
20.

tank

is

supplied

pipe, C.

If the

tank be
;

full

minute pass through each of the pipes

PROBLEMS ILLUSTRATING THE METHOD OF UNDETERMINED

COEFFICIENTS
397

We

proceed to consider one or two simple problems relating

to the subject matter of algebra itself. The inquiry may arise with regard to

some particular funcCan this ftmction be reduced to a certain specified form and, if so, what are its coefficients M'hen reduced to this form? The following example will illustrate the method of attacktion of the variables under consideration,

ing a problem of this kind.

Example. Can the expression x^ + a polynomial of the second degree in x ficients when reduced to this form ?

4z +

6 be reduced to the
if so,

form

of

1,

and,

what are

its coef-

The most
a{x + \)^ Hence,

general expression of the form in question may be written b{x + I) + c, where a, b, c denote constants. if the reduction under consideration is possible, we must have
x-^

or

x^

+ ix + + 4x +

= a{x + iy^ + h{x + \) + c = ax^ + (2 a + b)x + {a + b +

(1) will

(1)
c).

(2)

be an identity when, and only when, the coefficients of like powers of x in (2) are equal, that is, when a = 1, 2 o 4- b = 4, a + 6 + c = 6, or, solving for a, b, c, when a 1, 6 = 2, c = 3

By 284,

(2)

and hence

Hence

x^

-}-

4x

-f-

= (x +

1)2

2(x

1)

3-

SIMULTANEOUS SIMPLE EQUATIONS

153

sion of

Observe that we set the given expression equal to an exprestlie required form but with undeteniiined coefficients.
then lind that to make this supposed identity true, the

We

must satisfy certain conditional equations. And by solving these equations we obtain the values of the
coefficients
coefficients.

The following is a more general kind of problem including that just considered. Certain conditions are stated and the question is then asked, Does any function of a certain specified form exist
which
will satisfy these conditions, and, if so,

398

what are

its

coefficients ?

To
form

solve such a problem,

in question

we

construct an expression of the

coefficients.

cients are the

These coeffiand the given conditions yield the system of equations which they must
with undetermined

unknown numbers

of the problem

If this s^'stem of equations has a single solution, we if obtain a single function satisfying the given conditions the system has no solution, no such function exists if the
satisfy.
; ;

system has

infinitely

many

solutions, the

problem

is

indeter-

minate, there being infinitely many functions satisfying the It is here supposed that the function under given conditions.
discussion
is

ajinife expression,
If possible, find a

264.
in x, of the

Example. which has the value

X

polynomial
I

second degree,

when x =

and when x

3,

and the value 6 when

4.

in question must have the form axby the conditions of the problem

The polynomial
a

bx

c.

And

0,

9a +

36

+
is

Solving for a, b, c, we find a = Hence the required polynomial

2, 6

= 0, 16a + = 8, c = 6. 2 x^ 8 x + 6.
c

4&-f-c

6.

the problem been to find a polynomial of the first degree satisfying the given conditions, there would have been no solution had it been to find one of the third degree, there would have been infinitely many
;

Had

somtions.

154
399

A COLLEGE ALGEBRA

The method illustrated in the preceding sections is called the method of undetermined coefficients. It is the principal method of investigation in algebra and we shall often have occasion to
apply
>^

it

as

we

proceed.

EXERCISE

XII

1.

Express 3 x^ Express 4x2

x^

a;

5 as a

polynomial in x

2.

2. 3.

+ 8x + 7asa
ax^

polynomial in 2 x

3.

Find/(x)

=
=

6x

such that
/(l)
ex

/(-1)=11,
4.

= -5,

/(5)

6.

rind/(x)

ax3

ftx^

+
1)

+
1,

d such that
/(I)

/(O)
5.

=
=

5,

/(-{-

=
c

9,

/(2)

31.

rind/(x,

y)

ax

by
4,

-\-

Such that

/(O, 0)
6.

/(4, 4)

0,

/(I, 0)
1

6.

Find a simple equation ax + by are X = 3, 2/ = 1 and x = 4, ?/ = - 1.

7. Can a simple equation ax three solutions x = 3, 2/ = l; x =

two

of

whose solutions

+
4,

6y
2/

+c= = -l;

be found which has the = l, 2/ = l?

the straight line deter-

8.

Find the simple equation whose graph

(2, 3),

is

mined by the points

9.

4, 5).

Determine

so that the graph

of3x +

2/

will pass

through

the point

2, 3).

10. Find two simple equations, ax + 6y + 1 = (1), a'x + h'y such that both are satisfied by x = 2, y = 3 and also (1) by x and (2) by x = 3, ?/ = 7.

+\ =

(2),

7,

Plot the graphs of these equations.

11.

Find the equation

x^

6x2

ex

whose roots are

2, 1,

and

3.

12.

the solutions
13.

Find an equation of the form x^ + hxy + ex + x=l,?/ = 0; x = 2, 2/ = l; x = 2, y=l.

Express 3 x

%=

which has

2y

3 in the

form

where

a, 6,

a(x + y- l) + 6(2x-y and c denote constants.

2)

c{x

+ 2y-3),

THE DIVISION TRANSFORMATION

V.

155

THE DIVISION TRANSFORMATION

THE GENERAL METHOD

of

In 319 we defined the quctient Preliminary considerations. .1 by B as the simplest form to which the fraction A/ B

400

can be reduced by the rules of reckoning.

We are now to give a general method for finding the quotient

as thus defined,
letter, as x,
1.

when A and B are polynomials in the same and the degree of A is not less than that of B.

that

It is then possible that iJ is a factor of ^, in other words, can be reduced to the form

A
where Q
is

QB,
x.

(1)

an integral function of

We
that

then have
the quotient of
is

=
B

Q,

is,

^ by 5

is

the integral function Q;

and we say that A

Thus,

exactly divisible

by B.

if^=x3 + 4x2-2x-5 and i? = x2 + 3 x - 5, it will be found that x^ + 4x2 2x 5 = (x + 1) (x^ + 3x 5), an identity of the form (1), Q being x + 1. A = x3 + 4x2-2x-5 = x + 1. Hence B x2 + 3x-5
2.

But

We

cannot then reduce

show,

happen that B is not a factor of A. to the form QB; but, as we shall 401, we can reduce it to the form
it

will usually

A = QB +
where both Q and
of

R,
x,

(2)

is

are integral functions of less than that of B.

"^

and the degree

We

then have

+ ^'

156
that
tion,
is,

A COLLEGE ALGEBRA
the quotient of A by B is the sum of an integral funcand a fraction, B/B, whose numerator is of loiver

Q,

degree than its denominator. In this case we call Q the ijitegral part of the quotient, and

the remainder.

Thus, it A = x^ + 2x^ + Sx + 3 and reduce A to the form (2) by writing it

x3

B = x2 + 2x +
2)

2,

we can

at once

2i2
i? is

+ 3x +
x

= x(x2+2x +
is

(x

3),

where Q

is

x and

3,

which

of lower degree than B.

Hence

^ = x3 + 2x2 +
B
x2

2x + +
+

3x +
2

+
,

x2

x + 3 + 2x +

401

The
reduce

division
.4

transformation.

It

remains to show

how

to

to the

form QB

R, where

is

of lower degree than

and may be
is

plished

The process by which this is usually accom0. called the division transformation, or " long division."
2 x*

It is illustrated in the following example.

and B = x'^-x + l. and the problem is to find an integral function, Q, such that the remainder, R, obtained by subtracting QB from A, shall be of t\].e first degree at most and for if such a function, Q, be found, we shall have may be
Let
.4

+3

ic^

+ 4 ^2 + X - 2
is ttvo,

Here the degree of

QB =

R, and therefore

Since the degree of

greater than one, Q are cancelled when

is foiir

A = QB -^ R. and that of R is to

be not

must be such that the f7'st

we

subtract QB.

three terms of This suggests the follow-

ing method for finding Q.

THE DIVISION TRANSFORMATION

157

tract

Evidently we shall cancel the leading term of A if we subany multiple of B which has the same leading term that

The simplest multiple of this kind is 2x^B, where has. the multiplier 2 x^ is found by dividing the leading term of A, namely 2 x*, by the leading term of B, namely x^. Subtracting 2x'-B from A, as above, we have

A-2x''B^^x^ +

2x-

+ x-2.

(1)

cancel the leading term of the remainder (1) thus obtained, and with it the second term of ^, by a similar process. The quotient of 5 a;* by x^ is 5 a- and multiplying fi by 5 x
;

We may

and subtracting we have

.4

- (2

a;2

+5

ar)

5=

7 x2

- 4 x - 2.

(2)

Finally,
(2),

we

shall cancel the leading

it

term of the remainder

x^.

and with

the third term of A, by subtracting 7 B, where

the multiplier,

7, is

found by dividing 7 x^ by
5X

The

result is
(3)
it

.4

- (2 x^ +

+ 7) B = 3 X -

9.

The remainder

(3) is of the first degree,

and we obtain

by subtracting (2x"-^ + 5x + 7)fi from A. Hence the polynomials Q and R which we are seeking are

2 x2

5X

7 and

7?

3x

9.

And

writing the identity (3) thus,

A =(2x2
we have A
in the

form

+ 5x + 7)z;+(3x-9), QB + R, where the degree

of

is less

than that of B.

We

therefore have the following rule for finding Q and

are given.

when A and B

Arrange both A and B according to descendiyig poivers of x. Divide the leading term of A by the leading term of B ; the
quotient will be the first term of Q.
Multijjly

by this first term of Q, and subtract the product

from A.

158

A COLLEGE ALGEBRA

Proceed in a similar manner with the remainder thus obtained,

dividing its leading term by the leading term of B, and so on. Continue the process tintil a remainder is reached which is of lower degree than B. We shall then have found all the terms

of Q, and the final remainder will be

It is

K QB

or

K..

illustrated above.
in multiplication.

customary to arrange the reckoning in the manner We can then use detached coefficients, as

Example

1.

Given
4; find

J.

2 x^

6 x*

7 x^

8 x2

19 x

20,

and

5 = x2-3x +

Q and

B.

2-6 + 7 + 8-19 + 2-6 + 8

-1 + 8-1 +3-

20

THE DIVISION TRANSFORMATION

159

At the same time it is customary to call the operation by which Q and R are found "division," and also to call Q the
"

"

quotient
is

when R

not

instead of the " integral part of the quotient " even and we shall usually follow this practice.
;

^" does not then mean, as in 254, which multiplied by B will produce A, but finding, first, what multiple of B we must subtract from A to obtain a remainder which is of lower degree than B, and, second, what this remainder is. Compare 87. 3. The steps by which the integral expression A is reduced to the integral form QB + R may be taken whatever the value of X. Hence A and QB + R have equal values for all values of On the other X, even those values for which B is equal to 0.
But "dividing A by
finding an expression

hand, neither
Thus,
if

A / B nor Q
x2

+ R /B
X+
=
\

has any meaning

when

= 0.
(1)

A=x'^ +
401,

and

5=
-

x
1)

1,

we

find,

by

+
1

(x

2) (x

3,

and therefore
Here B = 3, which

X i = x +
when x
is

(2) ^ '

1.

true, but

3/0 =

Substituting 1 for x in (1) and (2), 3 + 3/0, which is meaningless.

we have

that

of A to the form QB + ^ is unique, there exists but one pair of integral functions Q and R (of which it is of a lower degree than B) such that
4.

The transformation

is,

A
QB +
But
but (Q

QB

+ R.
we should have
q')B = R'

For were there a second such

pair, say Q', E',

R= qB + R'

and therefore
R'

{Q-

R.

this is impossible, since

Q')

would be

of lower degree than

B
403

not of lower degree than B.

The

effect of

The following theorems

multiplying the dividend or divisor by a constant. will be of service further on.

c,

1. If we multiphj the dividend by any constant, as multiply both quotient and remainder by c.

we

F-or if

A = QB + R,

then

cA

= cQ-B + cR.

160

A COLLEGE ALGEBRA
we
divide the quotient by

c,

2. If we multi-ply the divisor by c, but leave the remainder unchanged.

For
3.

if

A = qB +
')n\dtiply both
c,

R, then

A = ^-cB-\- R.

If we

the remainder by

dividend and divisor by c, we multiply but leave the quotient unchanged.

R, then

For

if

A = QB +

cA

Q,-cB

cR.

4. If at any stage of a division tra7isformation we midtiply an intermediate remainder or the divisor by c, the final remainder,

if changed at

all, is
1

merely multiplied by
and 2 and
402, 1.

c.

This follows from

The student
particular case.
Thus, we
first

will do well to verify these

theorems for some

may
2

by
4

B=
+
6

X
1

verify the second theorem by dividing 1 and then by 2 4 x 2.

,

=4x2_).6x

B=

+ +

4-2
8
1

1
1

4-2
8

1
.-.

8-4
404

.-.

Q = 2x + R = b.

4,

8-4

Q= R=

x
b.

2,

Division by the method of undetermined coefficients.

We may
:

also find

Q and R when
1.

.4

and

are given, as follows

Example

Divide

^=

2x*

3x3

4x2

x-2byE

= x2-x +

l.

Since the degree of A is four and that of B is two, we know in advance that the degree of Q is two and that the degree of -K is one at most.

Hence

let

Q =

CqX^

c^x

Ci

and

R=

dox

di,

where such values are we shall have

2x*

to be

found for the

coefficients Cq,

ci, c^, do,

di that

3x3

4x2

X-2 = (CoX2 +
CoX*

CiX

Cj) (X2

1)

cfoX

di

++

Co X3
Ci

C0

THE DIVISION TRANSFORMATION

But
to

161

make

(1)

an

identity,
2.

we must

have, 284,

Co=

= 3+2 + ci = 3, .-.01= 3 + Co + C2 = 4, .-.02= 4 Co + Ci = 4 2 + 5 rfo = 1 - Ci + C2 = 1 - 5 + 7 C2 + Cio = Cl 1, - 2, =-2-7 Cj + di = di = - 2 - C2 Hence Q = 2x'^ + bx + 1 and jB = 3x - 9, as in 401. Example 2. Divide 6x5 + 13x* - 12x3- 11x2 + llx-2 by
Co
Ci

Co

.-.

= 5. = 7. = o. = - 9.
2x2 + x-

2.

denote polynomials in x with For literal coefficients, and suppose the degree of B to be 7?i. the division of .4 by B to be exact, the remainder R must equal
Exact
division.

Let

A and B

405

identically.

This requires that

all

the coefficients of

be

0.

277, and evidently these coefficients are functions of the coefficients of A and B. Hence

Since the degree of

is

??i

1, it

has

m coefficients,

In order that a

jjohjiioviial

viaij he exactly divisiJ)le

hy a

polynomial of the m.th deyi'ee, conditions. must satisfy

I>, tlie

coefficients

of

and

The following example

Example
divisible
1.

will illustrate this fact.

6 is
x-^

by x2

For what values of a and ax + 1 ?

x^

3x2

6x

2 exactly

We

have

x3

+ +

3x2
ax2

+ +

^^

+21 x2 + gx +
|x

+
a)
1)

(3

a)

(3(3

a)x2
g) x2

(6

+2 + (6 - l)x + (3 a - a"^) x + (3 - 1 - 3a + a2)x+ (a =

0,

Hence a and
6

must
a2

satisfy the two conditions

1 3a +
2.

0; whence a

and
Zx

3.

Determine I and 6. exactly divisible by x2 + x

Example

so that 2x3

3x2

+ m ^ay

be

A and B denote dividend and

powers of
x,

Let Dividend and divisor arranged in ascending powers of x. divisor arranged in ascending

406

and suppose that A does not begin with a term

162

.A COLLEGE

ALGEBRA

We may then obtain an begins with. A in terms by B by the process of If A is exactly cancelling leading terms described in 401. divisible by B, this result is the same as when A and B are
of lower degree than

integral expression for

arranged
divisible

in

descending powers
is

but

if

is

not

exactly

by B, the result
3x2

entirely different.

The following

examples will make this

1
1

clear.

+ 3x + + X
2x 2x

x3

THE DIVISION TRANSFORMATION

163

polynomial of as many terms and therefore of as high a degree

as

we

please.

Hence
denote polynomials arranged in ascending powers

T/'A and

of X, A not being exactly divisible by B nor beginning with a term of lower degree than B begins with, we can reduce the
qiiotient

of A. by

to the

form

B-^
and R' beginning with a
If the
still

+B'

where Q' and R' are integral functions arranged in ascending j^owers of x, Q' ending with as high a power of x as we please
higher power.
is n, we call Q' the quotient of n terms, and R' the corresponding remainder. "When the value of x is small (how small will be shown

number

of terms in Q'

by

to

subsequently)

we can make

the value of R' /B as small as

;

please by taking n great enough that is, we can find a polynomial Q' whose value will differ as little as we please from that of A / B. On this account the polynomials Q' are

we

sometimes called approximate

tion

integr-al expressions for

the frac-

A /B.
1

Thus "dividing"
1 1

by

to

n "steps," we obtain
h

= 1+X +

x2H

X"-!

+
1
1,

X"

If
tliat

we
1

give x

any value numerically

X"
if

less

than

4-

will differ in value as

we can choose n so little as we please from

1/18, so that 1 + x + x^ 1 + x + x- + x^ differs

1/(1

x).

Thus,

differs

from

from 1/(1^ x) by only 1/18. Similarly x) by only 1/54 and so on. /(I
;

1/3, then

xV(l -

x)

Quotients to n terms found by the method of undetermined coefficients. proceed as in the following example.

408

We

Example
Let

1.

Find the quotient

(3

x)/(l
aq,x"

2x2) to four terms.

X - X + 2 x2 = ao +
3

ttix

a^x^

(1)

164
Multiplying both

A COLLEGE ALGEBRA
members by
3
1

+
a;

2 x^

and collecting terms, we have

= ao +

tti

a.2

ao

THE DIVISION TRANSFORMATION

165

and R is of lower degree than B with respect to that letter. But the form of this expression will depend on what choice is

made of the
Example.
(1)

letter

of arrangement.

Divide 4x2

6x2/

!/2

by 2x

y.

Clioosing X as the letter of arrangement,

we have

4 x2 4 x2

+ +

6 xy 2 xy

2/2

+ 2 X +
2X

?y

2y

Hence
4x2

43^2/+

y2

4xy+2y2

_
(2)

+ 6xw + w2 Z ?^L^^2x+27/ -Ix + y

'

y1

2x + y

Choosing y as the
y

letter of

arrangement,

we have

y2
y2

+ 2x

1G6
11.
12.

A COLLEGE ALGEBRA
Divide 2 x2 Divide 2

13.

14. 15.
16.

x + 3?/-4. + ac + bbc hy 2a + b - oc. - c^ (a + 6) + abc by ab + 6c + ca. Divide a^ (6 + c) + (c + a) - 3)a;3 + (4 - a)x2 - 2 ax + 8 a by x^ - 3x + 4. Divide x* + (a

a"^

+ -

5 xy
b-

3y'^

- 5x + 13y - 12by

6c^
6'^

ab

Divide 8 x^

27 y^

by
?/*

2x

3 y, using detached coefficients.

Also X*

4 X2/3
o*6

17.
18.

Also 6

a5

+ - a362 +
3

by X

?/.

n a^b^ +

5 a6*

4b^ hy 2a^

ab

62.

Tiie dividend being 2 x"

Q and R, first, when x is when y is taken as that letter.

find
19.

xy^ + y^ and the divisor being 2 x + y, taken as "letter of arrangement," second,

Arranging dividend and divisor

in ascending

powers of
is

x, find

quotient to three terms and remainder and divisor 1 + x + 3 x2.

20.

when dividend

3x

5 x2

Also when dividend

is 1

3 x2 and divisor 1

3x

5 x2.

21.

By

the

method

of

undetermined
2
x).

coefficients, 408, find to

four

terms the quotient 1/(1

22.

Also the quotient (2

3x

4 x2) /

(1

2 x2) to four terms.

SYNTHETIC DIVISION AND THE REMAINDER THEOREM

410
proceed to explain a very expediSynthetic Division. method of making the division transformation, 401, a binomial of when the divisor has the form x h, that is, is
tious

We

the

degree whose leading coefficient is 1. Consider the result of dividing o-^ + (h^^
first
h.

2a^

by

UqX^

a^x"^

a.,x

-\-

a^

+ (n^b + ai)x + (a J? + cixb + a-) + ai)x^ + a^x (aj)'^ + aib)x (a J) + fti)x'^ as {ajj^ + /> + a^)x {ay + (>\b + 2)-^ - {(iJ>^ + (Ub"^ + Cob) =R a^b^ + a-J)^ -\-aj)-\-a^
rtpx^

(a^b

-\-

THE DIVISION TRANSFORMATION

The
coefficients of
Q,

167

and

are

Observe

t hat

the

first

of these coefficients
tliat

is

the leading

coefficient of the

dividend and

the rest
:

may

be obtained.

one afte r the other by the follow i ng rule

__

Multiply the coefficient last obtained by b and add the next unused coefficient of the dividend.
Thus,
ao6-

+
+

a-ih

a^

= =

{aoh
{aah-

aj) 6

+
+

a^,

and

aoh^

ai6-

a^h

as

aib

a2)h

a^.

This rule applies whatever the degree of the dividend maybe. For since the coefficient of the leading term of the divisor
is 1,

each

new

coefficient of

will

always be the same as the

Like that leading coefficient of the remainder last obtained. coefficient, therefore, it is found by multiplying the preceding coefficient of Q by J and adding a new coefficient of the dividend.

And

for a like reason

we

shall obtain R, if

we multiply

the last coefficient of dividend.

Q by

and add the

last coefficient of the

b and the diviHence, when the divisor has the form x dend the form a^x" + Oja;""^ -f + , we can find Q and R as follows, where Cq, Ci, c_i denote the coefficients of Q.

411

tto

a2
Cjfi

a_i
c_2b
^n-l>

\b

_ c^
Cq
Cl

_^

c_ib
-^

C2

We

first

write

down

the coefficients of the dividend in their

proper order and b at their right. Under a^ we write Cq, which we

know

to be the

same as
ai,

aQ.

We then multiply

Cq

by

b,

set the product c^b

set the

under

add,

and so obtain c^. In like manner we multiply aj, add, and so obtain c^.

Ci

by

b,

product

CjS

under

168

A collegp: algp:bra

And we continue thus, multiplying and adding alternately, until all the coefficients Uq, a^, a are exhausted.
Example.
Divide 3 x*

5 x^

4 x^

3x

by x

2.

We

have

3_5_4+3 _2U 2-4-2

3

1-2-1,-4
x2

Hence

Q=

3 x^

2x

and

E = - 4,

This very compact method is called synthetic division. The student should accustom himself to employ it whenever the
divisor has the form x

h.

112

Remarks on
the dividend

this method.
is

1.

In dividing synthetically when

a?

an incomplete polynomial, care must be taken

to indicate the missing

2.

powers of
h),

by

coefficients.

Since x

-{-

h :=

we may

divide synthetically by

a binomial of the form x

hhy

va.

It is only necessary to replace -{- b. the reckoning just explained.

Example
Here

1.

Divide x*

by x

1.

x4-l=x-( 1), and dividing by x 1 +0 +0 +0 - 1

-1

(
I-

1),

we have

1-1+1-1,
Hence
3.

+1

-1

+1
and

Q=

x*

x^

x-

E=

0.

To
:

divide by a binomial of the form ax

(3,

write

it

thus

a(x

(3

Then divide synthetically by x /3/a, and let Q and R represent the quotient and remainder so obtained. The quotient and remainder corresponding to the divisor

/ex).

ax-

13

will he
2.

Q/a and
3(x

li,

403,
18 x

2.

Example
Here 3x

Divide 3 x^
2

11 x^

by 3 x

2.

2/3), and dividing by x

2/3,

we have

-11

+18
--_6
12,

-3

|2/3

J 3-9

_8
6

THE DIVISION TRANSFORMATION

Hence the required quotient and the remainder is 5.
is

169

(3x2

_ 9x +
by x

12)/3, or z^

Sx +

4:

Example Example Example

3.

Divide 5 x^
Divide x^

x'

+ +

+
x

3.

4.
5.

6 x^

Divide 2 x^

3 x^

+ 6 by x + 3. + 8 x - 12 by 2 x 11

3,

The Remainder Theorem. When a poly 7iomial in si is divided b, a remainder is obtained which is equal to the result of iy X
substituting b fur x in the dividend
;

413

so that if f (x) denote the

dividend, f (b)

ivill

denote the remainder.

;

The demonstration of this theorem is given in 410 for it there shown that if we divide a^^ + a^x^ + a^ -\- a^ by X b, we obtain the remainder ajb^ + a^b^ + aj) + a^, and, in + by general, that if we divide f{x) = a^x"^ + ajx""^ +
is

b,

the remainder will be aj)"

aj)"'^
:

+ a,

or f(b).

The theorem may

If /(x)

also be proved as follows

be the dividend, x
401,

h the divisor,

(x)

the quotient, and

B.

the

remainder, then,

/(x)

<l>

(X) (X

6)

E,

where B, being of lower degree than x 6, does not involve x at all and therefore has the same value for all values of x. The two members of this identity havs equal values whatever the Hence value of X. In particular they have equal values when x = b.
f{b)

<p{b}(b-b)

R.
<p

But 6 Hence

<p

0; and since

(x) is

integral,

(6) is finite.

(6) (6

6)

0,

and therefore
/(6)

R.

The following example

illustrating the truth of the remainder

will serve the double purpose of theorem and of show-

414

ing that, when b and the coefficients of f(x) are given numbers, usually the simplest method of computing the value of f(b) is to divide the remainder thus b synthetically f(x) hy x

obtained being f(b).

170
Example
1.

A COLLEGE ALGEBRA
What
=: 5 X*
is

the value of
12 x3

f{x)
1.

20 x2

43 X

6,

when x

=4?

By

the
4*

method
12
.

of direct substitution
42

we have

/(4)

=5

43

- 20

- 43
-20
32
12

+ 6 = 1280 - 768 - 320 - 172 + 6 = 2&

we have

2.

By

the

method
5

of synthetic division

-12
20
8

-43
48
5,

6[4 20
26 =/(4).

Example
415

2.

Given /(x)

3 x^

x^

5 x2

8x

4.

Find, by syn-

thetic division, /(2),

.

/(-

2), /(4),

/(-

2/3).
a.

Corollary 1 If ^ (x) vanishes when x ^, and conversely. exactly divisible by x

= b,

is

then f (x)

is

For, by

hy X

413, f{b)

ft,

and the division

the remainder in the division of f(x) is exact when the remainder is 0.

Thus, /(x)
division.

= x3 - 3 x2 + 2
is

vanishes

when x =

1,

Hence f{x)
Again, /(x)

exactly divisible by x

1,

as

may

for /(I) be verified

= 1-3 + 2 = 0.
by actual

Example
value of

1.

= x" 6 is exactly divisible by x 6, since /(6) = 6" 6" = 0. If x^ + 3x2 jg exactly divisible by x 2, what is the
;,

m?
2^

We

must have
2.
is

2^

m=
6" is

0, or

m 20.
by x

Example
but not
if

Show
even.

that x"

exactly divisible

&

if

is

odd,

416

Corollary 2.

vanishes

when two of

If an integral function of two or rnore variables these variables, as x and j, are supposed

equal, the function is exactly divisible by the difference of these

variables, as x

y.

For the function may be reduced to the form of a polynomial in X with coefficients involving the other variables. By It is therehypothesis, this polynomial vanishes when x = y. fore exactly divisible by x y, 415.
Thus, x2 {y x) -f 7r (x z) + xp- (z 2) stituting y for X, we have y- (y

y) vanishes
(2

when x = y

for sub-

2/2

y)

2^ {y

y)

= 0.

THE DIVISION TRANSFORMATION

Hence
x^ {y

171
by x

We may
(y (y

z) + y"^ {z x) z^ (x y) is exactly divisible verify this conclusion by actual division, thus

y.

z)x2

(2/2

22)x

{y^z

z^y)

y
{y

2)a;2- {y^

-yz)x -(yz-z^)x + {yH-z^y)

-z)x-

{yz

- 22)

- (yzExample. Show that hy X y, y z, and z

z^)x

^ (y z

z^y)

(x
x.

y)^ +
f

{y

z)^ +

{z

x)^

is

exactly divisible

Theorem.
whe7i X

If a poli/tiomial
then f (x)
is

(x) vanishes

when x

b,

exactly divisible hy (x
is

a) (x

= a and also b).

we have
(1)

417

For since /()

by
ic

by hypothesis, f(x)
if

exactly divisible
<^i

a,

415, and

we
</>!

call the quotient

(x),

f(x) = (x
If in (1)

a)

(x),

where ^i (x)

is

integral.

we

set

b,

we have
(2)

f(b)

But by

= (b-a)Mb). hypothesis f(b) = 0,

when

and

=^ 0.
its

Therefore, since

a product vanishes one of

factors

must vanish, 253, it follows from (2) that ^i (b) = 0. But if ^1 (b) = 0, then <^i (x) is exactly divisible hj x

b,

415, and

if

we

call the quotient (ji2(^)

b)
<f)2

we have
is

^1 (x)

= (x

(x),

where ^2 (^)

integral.

(3)

Substituting this expression for ^i{x) in (1),

we have
(4)

f{x)

= {x-a)(x-b)is

4>2

(x),

which proves that f(x)

Continuing thus, we

exactly divisible

by

a)(x b). (x
is

may

prove the more general theorem

b,
c,

If f(x) vanishes for x

divisible by (x

= a,

,
.

a) (x b) (x c)
2X

then f(x)

exactly

418

Thus, 2 x3

3 x2

-3

vanishes

when x

1,

for 2

0,

and when x = - 1, for - 2 + 3 + 2 - 3 = 0. Hence 2x3 + 3x2 2x 3 jg exactly divisible by x2 1, as may be verified by actual division.

(x

1) (x

1),

or

172
Example L
X

A COLLEGE ALGEBRA
Find a polynomial /(x), of the second degree, which when x = 2 and when a; = 3, and the value 6 when

will take the value

4.
is

Since /(x)
417,
it

may

2)(x 3), of the second degree and is exactly divisible by (x be expressed in the form/(x) = ao{x - 2) (x - 3), where Uq
=
6,

denotes a constant.

And

since /(4)

we have
2) (x

Hence f(x)

3 (x

3)

= ao(4 - 2) (4 - 3), = 3 x'^ - 15 x + 18.

whence ao

3.

Example 2. Find a polynomial /(x) of the third degree which will when x = 2 and when x = 3, and will take the value 6 when x = 1 and the value 18 when x = 4.
vanish

Reasoning as before we have /(x)

ao, Ui

(aox

ai) (x

2) (x

3)

where

are constants.

Again, since /(I)

6

6,

and /(4)

= -

18,
3),

we have
or or
ao 4 ao

18

= = =

(ao

ai) (1

2) (1

(4ao

ai)(4

-2)(4-3),
ao

+ +

ai

Oi

= =
x

3,
9.

(1)
(2)

Solving

(1)

and

(2),

we obtain
1) (x

Hence /(x)

(2x

2) (x

= 2, ai = 1. - 3) = 2x3 -

Ox^

6.

419

Theorem.

A polynomial f (x).

whose degree

is n,

cannot vanish

for more than n values of x.

For if f{x) could vanish for more than n values of x, it would be exactly divisible by the product of more than n factors of the form ic a, 41 8, which is evidently impossible
since the degree of such a product exceeds n.

420

Theorem. If we know of a certain polynomial f(x), whose degree cannot exceed n, that it will iJanish for more than n values of x, we may conclude that all its coefficients are 0.

For if the coefficients were not all 0, the polynomial could not vanish for more than w values of x, 419.

We
421
<^(x),

say of such a polynomial that

it

vanishes identically.

Theorem.

If tiro polynomials of the nth degree, f (x) and have equal values for more than n values of x, their

corresponding coefficients are equal.

THE DIVISION TRANSFORMATION

For and
also let
let

173

= a^x"" + aia;"-* h + ^i^;""^ H h^; ^(a;)= = f(x) (x) (x) = (ao - ^o) X- + (ai - b,) a;"-' + . + (- J).
f{x)
-\

^o-"

\p

<fi

Then

xf/

(x) is

whenever the values of f(x) and

<^

(x) are

the same, and by hypothesis these values are the same for more than 71 values of x.

Hence the polynomial

\j/(x)

= (ao

bo)x"

+(a

b),

whose degree does not exceed n, vanishes for more than n values of x, and therefore, 420, all its coefficients are 0.
Therefore Uq

whence
that
is,

bo = 0, a^ = b^,
=
2 x2

Ui

= 0, Oi = b^,
bi

-,

b =0, = b,
a.

the corresponding coefficients off(x) and

if

<^

(x) are equal.

Thus,
values

f{x)

+
6,

wheu x

2, 4,

and (x) = ax'^ we must have a = 2, 6

6j;

+ Sx c = 3, and
-\-

have equal
c

5.

EXERCISE XIV
1.

Divide x*

3 x^

x^

11 x

4 by x

4 synthetically.

2.
3.

Also 5x5

-6x*+ x2

8x3

7x2
2.

+ 6x + 3by

3.

Also 3x*

4.
5. 6. 7.
8.

Also3x3 + 16x2
Also 3x3 Also x^

+ - 13x by X
X

by 3x

1.

6x2

+
b

by 3x
-f-

1.

{a

c) x-

{ab

ac

-{

bc)x

abc by x

a.

Also 2x*

-x^y

Given/(x)

-7 x"y^ + 7 xy^ = 2x3- 5x + 3. By

/{-

10 y^ by x
the

2 y.

method
3),

of 414, find
6).

/(I), /(2), /(5),

9.

1),

/(-

/(-

By

aid of the remainder theorem, determiners so that

x3

may may

be exactly divisible by x

+ 7nx2 - 20x + 3.
I

10.

In a similar manner, determine

and
4).

so that 2 x^

x^

-f ix

+m

be exactly divisible by (x

2) (x

174
11.

A COLLEGE ALGEBRA
By
416, show that 2 n, also by a + 3 b.

Sbm + am
that

an

6bn

is

exactly divisible

by

m
12.

By 416,

417,

show

exactly divisible by (a
13.

b){b

a{b

a).

c)^

c) (c

b{c

- af +

c{a

b)^

is

when
14.

a;

Find the integral function of x of the third degree which vanishes = 1, 4, 2, and takes the value 16 when x = 2.

when
x

a;

Find the integral function of x of the third degree which vanishes = 2, 3, and takes the value 6 when x = and the value 12 when

l.

15.

Show

that

X'''

ax
x.

-ii-

and

x^

+ 5x +

cannot have equal

values for four values of

EXPRESSION OF ONE POLYNOMIAL IN TERMS OF ANOTHER

422
Let

A and B

denote two polynomials in

Q,

x,

of higher degree

than B.
Divide

Ahj B and call the quotient

A = QB +
R.

the remainder

then
(1)

not of lower degree than B, divide the quotient Qi, the remainder Ri then
If

is

Qhy B and

call

Q
if
is

Qi5

R^.

(2)

not of lower degree than B, divide Qi by Qi Similarly, and call the quotient Q2, the remainder R^; then
Qi

Q,B

R^.

(3)

Suppose that Q2

is

of lower degree than B.

We then

have

A = QB + R

by by by

(1)
(2)

= IQ,B + R^IB + R = \(Q2B + R^)B + R,\B + R = QoB^ + R.B'' + R^B + R,

where
all

(3)

the coefficients

Q^,

R2, Ri,

are of lower degree

than B.

THE DIVISION TRANSFORMATION

And,
in general, if

175
is

any polynomial A be given which

of higher degree than B, and we continue the process just described until a quotient is reached which is of lower degree

than B, we shall have

A=

Q,_,B'-

i?,_i'-i

---

R,B-\-R
than B.

where R, Ri, , Rr-i, and the final quotient,

Example. nomial in x^

Q,.-i
all

denote the successive remainders

lowei' deyree
'|

being of

Reduce

4 to the form of a poly1 with coefficients whose degrees are less than two.
x^

4 x*

x-^

-.'

Using detached

coefficients,

we may arrange

the reckoning thus

1-4+3-1+1+4
1

-5+2-1 -5-5-5

176

A COLLEGE ALGEBRA
EXERCISE XV

1.

By

the method of 422 express x*

z^

1 iu

terms of x^

4. i.

2. 3. 4.
5. 6.

Also 4 X* Also 2 x"

Also x5

+ -

2 x^

+ +

4 x2 2 x^

+
+

6 in terms of 2 ^

1.

3 x6

5 X*

x2

6 in terms of x^

x^

3.

+ x3?/2
+

+ +
4-

a;22/3

y^ in

terms of x^

xy
x

H-

y"^.

By the method
Also x5
3 X* 9 x2
3 x2

of 423 express 2 x^

8 x^ +
7 in

in

terms of x
2.

-3

6 x3

2 x2

3X

terms of x

7.
8.

Also x'
Also x'

+
+

27 X in terms of x

3.

1 in

terms of x

1.

VI.

FACTORS OF RATIONAL INTEGRAL EXPRESSIONS

PRELIMINARY CONSIDERATIONS

424
or

Factor.

Let A denote a rational integral function of one

variables.
di.

Any rational integral function of these variables which exactly divides A is called factor of A. Hence in order that a given function, F, may be a factor of
vl, it is

more

and necessary be rational and integral with respect to the variables of which A is a function.
sufficient
1.

That

2.

That A be reducible to the form A

GF, where

also

is integral.

Example

1.

Since 2x2

2xi/

= 2x(x

y),

both x and

x~y

are

factors of 2 x2

2 xy.

Example

2.

Since

3x2

2/

2^/2

V3 X + V2 V^

2/

and
3.

V3 X V2

are factors of 3 x2

^ (Vsx + V22/) (VSx - ^^2/), 2 ?/2.

( (

both

Example
-I-

Althoiigh x

Vy and Vx Vy
y.

factors of x

= Vx + y/y) Vx - Vy), we do not call y, because they are not rational

with respect to x and

FACTORS OF INTEGRAL EXPRESSIONS

Note
rational.
1.

177
425

The

coefficients of a factor

need not be either integral or

kind.

Therefore, since x^ y = {x + Vy) (j; V^), the expression x^ y, regarded as a function of both x and y, cannot be factored but regarded as a function of x alone, it has the factors x + V^ and x Vy. And the like may be said of other expressions which involve more than
;

the contrary, they may be In Ex. 2 they are irrational.

On

numbers or expressions

of

any

one

letter.

Note

2.

integral coefficients, it is not factor," like 2 in Ex. 1, in a

tion,

Except v?hen dealing exclusively with functions having customary to include a mere "numerical
''st

426

of the factors of a given integral func-

for

if,

as here,

function to be integers, be said to divide A exactly.

we do not require the coefficients of an integral any mere number (or constant) whatsoever may
is

For a

like reason, if

but we regard factor and include but one of them in a

is

cF

also a factor of

or that

Thus, in Ex. 1, it 2 X and y

c is any constant (not 0), and cF as essentially the same list of the factors of A. would be equally correct to say that 2x and x y,
;

a factor of A, and

x,

are the factors.

'B is

Theorem.

is

If is a factor of B, and a factor of A.

a factor of A, then

427

For, by

424,

A and B A

are reducible to the forms

= GB =

and

B = HF,

where

G and

H are

integral.

Hence
that
is,

A
is

GHF= GH-F,

a factor of ^,

424.

Prime, composite.

An

integral function

may have no

other

428

factor than itself (or a constant).

In that case we
call it conijjosite.

call it prime.

But

if it

have other

factors,

we

Thus, X

+ 2/2

and x

2y

are prime, but z^

yi

js

composite.

composite function, A, of the nth. degree, is the product of not less than two nor more than n prime functions, B, C, -.

429

These prime functions are called the prime

factors of A.

178
430
In what follows
1. 2.

A COLLEGE ALGEBRA
Ave shall

assume that

Two or viore of these prime factors may be equal, but A 3. can be expressed in only one way as a product of j^owers of its
different pjrime factors.

Any given function All other factors of

A has hnt one set of prime factors. A are products of these jJrime factors.

in 484,

These theorems, of which 2 and 3 are corollaries to 1, will be proved 485 for the case in which ^ is a function of a single variable, and they can be proved generally.

- 2 x^^i - xxyyy (x - 2 y), the prime factors of Thus, since x^y^ - 2x2;/* are x, x, y, y,y,x-2y. Its other factors, as x^, xy, and x^y^ Its different so on, are products of two or more of these prime factors. and it can be expressed in but one way as factors are

x,y,x-2y, prime a product of powers of these factors, namely thus:

x'^y^(x

2y).
is it

431

Factoring. factor a given function, A, completely "resolve it into its prime factors," that is, to reduce

To

to to

the form
functions.

A=BCD---,
We

where B, C, D,--- denote prime

But ordinarily we do not attempt to discover these prime endeavor first to resolve A into factors at the outset. of some tivo of its factors, as F and G, next to a
product
resolve

reached.

And

and G, and so on, until the prime factors are even the first step in this process may be
.4.

called "factoring"
Factoring
involves two
is

the reverse of multiplication.

:

main steps

(1)

law, in order to replace (a + tion of like terms in the result thus obtained.
(1)

A multiplication usually a number of applications of the distributive (2) the combinab)c by ac + be, and so on
;

we must separate the terms thus combined the diihculty of factoring consists and then
(2)

To
it

reverse the process, is in doing this that

apply the distributive

law in order to replace ac + be by (a + b) c, and so on. It must not be supposed that every composite function can be actually factored. Thus, while it can be proved that x& + ax^ + bx- + ex + d is it can also be proved that tlie factors of this expression cannot
composite,

be found by algebraic methods, that

is,

by applying, a

finite

number

of

times, the various algebraic operations.

FACTORS OF INTEGRAL EXPRESSIONS

179

EXPRESSIONS WHOSE TERMS HAVE A COMMON FACTOR

Expressions whose terms all have a common factor, monomial or polynomial, can be factored by a single application of the distributive law, namely
:

432

ab

-{-

ac

-\-

ad

-\-

^
+

(b

-\-

-{-

-\-

).

Example

1.

Factor 2 a^c

2 abc

4 ac^

6 acd.

All the terms have the factor 2 ac.

2 a^c

"

Separating

"
it,

we have

2 a5c

4 ac2

d)

6 acd

2 ac (a

2 c

3 d).

Example

+ b{d c). Both terms have the factor c d. Separating it, we have ~ d) + b{d c) = a{c - d) b(c - d) = {a b){c a{c
2.

Factor a{c

d).

Factors such as these should be separated at the outset.

Some
a

expressions which are not in this form as they stand

it

433

can be reduced to

by combining such of their terms

as

have

common
Example

factor.
1.

Factor ac

bd

ad

be.

Combining ac and ad, also be and bd, we obtain a(c a binomial whose terms have the common factor c + d.

d)

5(c

d),

Hence

ac

bd

ad

bc

{a

b) {c

d).

Observe that the parts into which we separate the expression in applying this method must all have the same number
of terms.
jExample
2.

Factor a^

ab

bd

ad

ac

cd.

must have either two groups of three terms, or three groups of two terms. Four of the terms involve a, namely, a^, ab, ad, ac, and To obtain groups the remaining two involve d, namely, bd and cd. which have the same number of terms we combine the term ad with the d-terms, and have
a'^

We

ab

ac

- ad-bd -

cd

a{a

c)

d{a

c)

= {a-

d)(a

b-ir c).

180

A COLLEGE ALGEBRA
EXERCISE XVI

Factor the following expressions.

1.

6 x*yH^

12 x^y^z
6

8 x^yS.

2.

3.

a6

- L
8.

A.
6.
8.

+ (n - 3) . mx - nx mn +
2 n^

n^.

5.

3xi/-2x-122/ +
x32/2

lOxy
x*

2/2

+ 6x +

By.

7. 9.

x^ys

2 x2y

- 2 xy^.

ac

M - (6c + ad).
ce

10.
12.

11.

ad

bcZ

+ ae + cd + 6e.

+ x^ + x2 + x. a2c - ahd - abc + a'^d. a^ + cd - ab - bd + ac +

ad.

FACTORS FOUND BY AID OF KNOWN IDENTITIES

434
In the second chapter we derived a number of special b) = a" b'^. If a given function, A, -\- b) (a

products, as (a

can be reduced to the form of one of these products, we can The following sections will write down its factors at once.
illustrate this

method

of factoring.

435

Perfect trinomial squares.

This name

is

given to expressions

which have one of the forms a^ 2 ab + b'^. can be factored by means of the formulas
:

Such expressions

= (a + b)(a-\-b) = (ai- by. = (a-b)(a-b) = (a- by. a^-2ab-\-b^

a''-\-2ab

b^

Observe

that

in

arranged) the middle term

perfect trinomial square (properly is twice the product of square roofs

of the exti-eme terms, and that the factors, which are equal, are obtained by connecting the principal square roots of the

extreme terms by the sign of the middle term.

To

extract the square root of the perfect square is to find

factors.

one of these equal

Example
This
is
1.

Factor 9x2

12x1/

+
2

4^2,

a per fect square, s ince 12 xy

v9x2

vij/^.
of the middle term

And
is

since

V9x2 =

3 x,

Vij^ =

?/,

and the sign

-, we have 9x2

\2xy

4y2

(3a;

22/)(3x

-2y) =

(3x

22/)2.

FACTORS OF INTEGUAL EXPRESSIONS

Example
2.

181

Factor a^

b'^

c^

2 ab

2 ac

2bc.

We
a2

can reduce
:

this to the

form

of a trinomial square

by grouping the

terms thus

2 a6

62

2 ac 4- 2 6c

c2 :=

(rt

=
Example
1.

(a

+ +

6)2

2 (a

6) c

c2

c)2.

3.

Factor the following expressions.

x2

14 X

49.

2.

9
x2 a2

3. 5.

9x2^/2

64 a8

+ 30x1/ + - 48 a* + 9.

25.

4.
6.

6 a + a2. - 4xy + 42/2 + 6x - 12 y + 9. + 62 + c2 - 2 a6 + 2 ac - 2 6c.

difference of
it,

reducible to

two squares. Expressions of this form, or can be factored by aid of the formula:

436

a-'-P=(a +
Thus,
X2

b) {a

b).

2/2

22

2/2

= X2 = X2 - (X +

(?/2

(2/

_ 2 2/2 + 22) - 2)2 - 2) (X - +

2/

2).

very useful device for reducing a trinomial expression to this form is that of making it a perfect square by adding a suitable quantity to one of its terms and then subtracting
this quantity

from the resulting expression.

x*

Thus,

x22/2

2/*

a;4

= =
Example.
1.

+ 2 x^y"^ + - ^^v"^ _ x2y2 (X2 + ^2)2 (x2 + 2/2 + xy) (x2 + 2/2
2/*

xy).

Factor the following expressions.

2.
5.

X*

?/.

6 a3

6 a62.

3.
6.

12
x*

aH^ - 75 axy^.

4.

25x2n_49x2m.

36x*-l.

Sx'^y'^

y*.

A sum
unit
i

of

two
1,

squares.

By making
sum

use of the imaginary

437

be b"^ can of squares a^ reduced to the form of a difference of squares and then factored
218, 220, a

= V

by

436, the factors being imaginary.

For since

P=-

1,

we have
a^

Hence

a^ + 6^ =

b^ = - (- b^) = - (iby. _ ^,-^)2 = ( + ib) (a - ib).

182

A COLLEGE ALGEBRA
seen, 219, 220,
i

As we have
ing
it

conforms to
to

all

the ordinai-j

rules of reckoning.

that

i^

One has only

any two
odd or

remember when employIn 308, 309,

1.

438

Sums and
First.

diiferences of

like powers.

310 we proved that

Whether

?i

is

everi,

a--bSecond.
a"

= {aWhen

b) (a"-'

+ a"-^ +
a^-'b

+ ab' + b"-').
b"-').

(1)

is even,

^.

= (a + b) (fl^-i n
is

- + ab"-^ ab^'^

(2)

Third.
a"

When

odd,

+ b"=(a +
b is b is + b" is

b)

(a"-^

a'^-H
:

+
h.

b"'').

(3)

Hence the following theorems

1. 2.

a"
a

always

divisible by

a,

divisible

by

a, -{a.

3. 4.

a"

divisible by

h when n h tvheji n
consists

is even.

is odd.

In every case the quotient

qu-i

of the terms
b"-i

a-2b-ab"-2

connected by signs which are all when a and hut are alternately

when a

b
'

is

the divisor,

+bw

the divisor.
a;2

Thus,

1.

2.
3.

8 o3

_ 1 = (X - 1) (x5 + X* + x3 + + X + 1). - 1 = (X + 1) (X5 - X* + X3 - X2 + X - 1). + 27 6c3 = (2 a)3 + (3 bc)^ = (2 a + 3 6c) [(2 a)2 - (2 a) (3 be) + (3 6c)2] = (2 a + 3 6c) (4 a2 - 6 a6c + 9 62c2).
x6 X6
2.

Example.
1.

Factor the following expressions.

64x8-1252/3.

27x^

1.

3.

Wx*-8ly*.
imme-

439

Tlie following theorems are an diate consequence of 438, (1), (2), (3) and 436.

When
1.

is

composite.

If n

is

divisible by a"

multiiile
b^.

of any integer,

p,

then a"

b"

is

exactly

FACTORS OF INTEGRAL EXPRESSIONS

Thus,
x^

183

-y^ = (x2)3 - (y2)3 = (X2 - 2/2) +

(a;4

a;2y2

,/4).

2.

7/*

is

an even multiple of any

-\-

integer, p, then a"

h" ts

exactly divisible by oP
Thus,
X6

b^.

2/6

(X3)2

(2/3)2

= (x3 +
3.

2/^)

(x3

- 2/3).
a"

an odd multiple of any integer, p, then 7/" exactly divisible by a^ + b^, whether n itself is odd or
is

is

even.

Thus,

X6

2/6

= =

(a;2)3

(2/2)3
(a;4

(X2

^2)

^ZyO

2/*).

4.

i/"

is

a power of

2,

then a"

+ b"

can be resolved into

of the second degree by repjeated use of the device explained in 436.

factors
Thus,
x8

^8

= x8 + 2 x*2/* + 2/^-2 x*2/* = (x* + 2/*)2-2x''2/* = (x* + 2/* + V2 x22/2) (x* +

X*

2/*

- V2 x22/2).

Again,
X*

y*

+ V2x2i/2 = =

and so on.

+ 2 x22/2 + 2/4 _ (2 - v^)x22/2 - (2 - V2) x22/2 + = (x2 + 2/2 + V2 - V2x2/)(x2 + y2 _ V2 - V2X2/),

(x2
2/2)'^

As each
o"

of these " quadratic

first

" factors can be resolved into two (imagi-

nary) factors of the

6" is

degree by 444, the complete factorization of always possible when n is a power of 2.

When
or a"

is

composite,

it is

b",

into

two

factors

best to begin by resolving a" + 5", whose degrees are as nearly equal as

jjossible.

It will

always be possible to factor at least one of

the factors thus obtained.

Thus, the factorization of x^ we have
X6

y^ given

under 2

is

the best.

Continu-

ing,

y6

(a;3

(x

+ 2/3) (X3 2/3) + y) (x2 -xy +

2/2)

(X

y) (x2

+ xy +

y^).

184
Example.
1.

A COLLEGE ALGEBRA
Factor the following expressions.
X*

y'^.

2.

x^

2/8.

3.

x9

y^.

440

The theorems
the form
integer p.
Thus,
a'"

of 438, 439 also apply to expressions of h" when and n are multiples of the same

m
= =

X6

yl*

{X2)3

(?/5)3

(X2

2/5)

(X*

+ x2!/5 +

2/10).

EXERCISE XVII
In the following examples carry the factorization as far as without introducing irrational or imaginary coefficients.
1.
is

possible

x3y/

20

x22/2 3.

+
x2

25 xy^.

2.

28

te2

63

ty^.

2/2

9 z2

4 x?/

12 yz

6 zx.

4.
5.

(7 a2

2 62)2

(2 a2

+
6. 8.

7 62)2.

(7x2
X*

4-4x-3)2-(x2 + 4x + 3)2.
x2

4(1
a*

62

a6)

a2.

7. 9.

+
+

1.

-& a-W +

6*.

a*

4 a2

16.

10.

9 x*

11.

4 (a6

cd)2

(a2 4. ^2

c2

+ Ibx^y^- + _ ^2)2.
14.
17.

16 y*.

12. 15.

576xV-9
xw

2/^^-

13.
16.

x^
x5

2/9.

x^'^

y^^
2/".

2^w.

32.

x'^

FACTORS FOUND BY GROUPING TERMS

441

in groups, all of common factor

Sometimes the terms of a polynomial in x can be combined which have some common factor, as F. This

is

then a factor of the entire expression.

Compare

433.
1.

Example

Factor x^

+ 3 x2 -

2x

6.

Noticing that the last two coefficients are equimultiples of the

first

two,

we have
x8

3 x2

2X

= =

x2(x
(x2

3)

2 (x
3)

2) (X

+ 3) = (X + V2) (X - V2) (X +

3).

FACTORS OF INTEGRAL EXPRESSIONS

Example
2.

185

Factor x3

2x2

+ 2x +

l.

Combining terms wiiich have

x3

like coeflBcients,
l)

we have
X (X

2x2

+ 2x +

= = =

x3

+ l + 2x(x + - X + 1) (X + (x2 - X + 1 -f 2X) (X2

1) 4- 2

1)

(X

1)

(X2

+X+

1) (X

1).

this can be accomplished by first separating one of the given terms into tw^o terms.

Sometimes

Example

3.

Factor x^

4 x^

5x

+
^2

6.

We

have

X3

4X2

+ 5X +

G =:X3

+ +

3x2

_|_

-)-

3X

-f

2X

=
=r

x2 (X
(x2

+
X

3)

X (x

3)

2 (X

3)

2) (X

3).

Consider also the following example.

Example

4.

Factor x*

2 x^

3 x2

1.

We
x*

have

2x3

+ 3x2 + 2x +

= x* + 2x3 + x2 + 2x2 + 2x + = (X2 + X)2 + 2 (X2 + X) + 1 = (X2+X + 1)2.

EXERCISE XVIII

Factor the following expressions.

1.

3. 5.

x*-x3 + x-l. x* - 2x3 + 2x x6

2.
1.

4.
?/.
6. 8.

x5-x3-8x2 + 8. x3 - 7x2 - 4x + 28.

x3 x*

x*ij-

^-y^

+
+

2 x2

3X

2.

y{. x5

+ 2x* + 3x3 + 3x2 + 2x + l.

4x3

10x2

12x

9.

FACTORIZATION OF QUADRATIC EXPRESSIONS

The quadratic x- + px -|- q, factored by inspection. sometimes possible, when /> and q are integers.
Since
{x
-\-

This

is

442

a) (x

+ h) =

oc"^

we

shall

know

the factors of x^
h,

numbers, a and

such that a

+ (a + b) x + ab, if we can x>x + + I v and ab = q.

-\5-

find

two

186

A COLLEGE ALGEBRA

Two such numbers always exist, 444, though they are seldom rational. But when rational they are integers, 454, and may be found by inspection, as in the following examples.
Example
1.

Factor

x"^

+ lSx +

42.

seek two integers, a and b, whose product is 42 and sum 13. As both ab and a + b are positive, both a and b must be positive. Hence among the positive integers whose product is 42 namely, 42 and 1, 21 and 2,
14 and
3, 7

We

and 6

we seek a pair whose sum

x2

is

13,
6).

and find

and

6.

Hence

13 x

42

(x

7) (x

Example
their

2.

Factor x^
b

13 x

22.
;

Here both a and

sum

11 and gers whose product is 22, we find x2 - 13x + 22 = (x - 11) Hence

Example
since a
3.

negative.

for their product is positive and Hence, testing as before the pairs of negative inte2
(x
;

must be negative

for

11

13.

- 2).

Factor x2
is

9x

22.

Here, since ab

6 is

negative, a and b negative, the one which

Hence we set negative. - 11 and 6 before, find a = Hence

x2
4.

22

= 2; for - 11 + 2 = - 9. - 9x - 22 = (x - 11) (x + 2).

2. 5.

must have opposite signs and numerically gi-eater must be 22 x 1 = 11 x 2, and, testing as
;

is

Example
1.

Factor the following expressions.

x2

4.

x2

+ 3x + 2. + X - 30.

x2-16x + x2 + 20 X +

15.

3.

96.

6.

x2-4x-12. x^ - 21 x + 80.

443

The quadratic ax^ + bx + c factored by inspection. This is sometimes possible, when a, h, and c are integers. By multiplying and dividing by a, we may reduce ax'' -\-hx-\-c
to
h (ax) -{- ac'\ / a, and then factor the the form \_(adcY bracketed expression with respect to ax by the method just

explained, namely, by finding two integers

ac,

whose product

is

and sum
1.

b.

Example

Factor 2 x^

3.

We have

2x2-|-7x

(2x)2

7(2x)

= <?^^M^^ll = (x +
2

3)(2x

l).

FACTORS OF INTEGRAL EXPRESSIONS

Example
2.

187

Factor

abx'^

{a^ -\-h")x-\- ab.

/ ,

We have
abx^

.o

ro^

oro

(a2

b^)x

ab

^ _
{abx

^^

ab

g-) (abx

6^)

ab

= {bx +
Example
3.

a) {ax

b).

Factor 16 x^
it is

72 x

63.

In this case

not necessary to multiply and divide by 16, for

we have

16x2

+ 72x-63 = =

(4x)2

+ 18(4x)-63

(4x

+ 21)(4x-3).
2. 4.

Example
1.

4.

Factor the following expressions.

3.
5.

6x2-13x + 6. 14 x2 + X - 3.
49 x2

105 X

44.

6.

+ 14x-3. + 21 X + 5. a6x2 - {ac -b'^)x5x2

18 x2

be.

The quadratic x^
pleting the square.

+ px +

q or ax^

bx

c factored

by comin

444

While the preceding methods apply

is

particular cases only, the following

perfectly general.

Since

(^"^9) =^^+/'^ +

't"

we can make

x^ -\- px a perfect square by adding --' ^^^^ the square of half the coefficient of x. This process is called comjileting the square of x^ px.

i^,

1.

We

shall not affect the value of

a;"

+ ^;x +

?? if

we both

add and

siibtract jo^/4.

But by

this

the expression into the difference then factor it by 436. Thus,

means we can transform between two squares and

x'^+px

+ q = x^+px + ^ + q
(

.(.-fft^2?S)(.tS-^E5>. (

188

A COLLEGE ALGEBRA
Since
ax^

2.

bx

-\-

= a{
y

x^

-\

x a a
-\

)>

we may
b /a for

2^

obtain the factors of this expression by substituting and c/a for q in (1). Simplifying the result, we

have
ax^ + bx

+c=
1.

a[x

+ -+

^^

+ -j[..

^^J-

(2)

Example

Factor
x^

We have

x"^ 6x + 2. - 6x + 2 = - 6x + - + 2 = - 3)2 - 7 = (X - 3 + V7) (X - 3 - V?).

x"S"^
3'^
(a;

Example

2.

Factor x^
x^

We have

8x

+ 8 x + 20. + 20 = x^ + 8 x + 42 - 42 + 20 = (X + 4)2 + 4 = (X + 4)2 - 4 i2 = (X + 4 + 2 (x + 4 - 2 i).

J)

Here we
this

sum

of squares, (x 4)2 4 into a difference by replacing 4 by

first

obtain a

sum

4,

and then transform

437.

i-,

The

factors

are imaginary.

Example

3.

Factor 3x2

5x +
1

1.

We have

3x2

-^

5x +

3 rx2

- -x +

^1

-(^-^f)(^-^f)
Example
1.

4.

Factor the following expressions.

23.
2. 4.

x2
x2

+ 10x +

3. 5.

+ 45. 2 x2 + 3 X + 2.
12 X

6.

x2-10x + + X + 1. x2 - 4 ax x2

24.

62 4.

8 a6.

FACTORS OF INTEGRAL EXPRESSIONS

Homogeneous quadratic functions
of
of

189
445

two variables.

The methods
form

442-444 are applicable

ax^

to quadratics of the

+ hxy + cy^.
+
x'^

Example

1.

Factor x^
x2

8x2/

14

2/2.

We

have

xy

+Uy^ =
= =

-8xy +
(4

16

y'^

2/2

(x-42/)2_22/2
[X

+ V2) y]
2.

[X

(4

- V2) y]

Example

2.
1.

Factor the following expressions.

x2

x?/

4 y-.

x2

xy

2/2.

Non-homogeneous quadratic functions of two variables.

Such

446

functions are ordinarily prime. But when composite, they may be factored as in the following example.

Example
If

1.

Factor

=x'^

+ 2xy -

8y'^

+ 2x +

Uy - S.

the product of two polynomials of the first Moreover its terms of the second degree, x2 + 2 xy 8 y'^, must degree. be the product of the terms of the first degree in these polynomials.
is

composite,

it

is

We
that

find

by inspection that

x^

2 xy

8 y^

(x

i y) {x
I

2 y).

Hence,

if

is

composite, there must be two numbers,

and m, such

we
x^

shall

have

+ 2xy

-8y'i

+ 2x + Uy -3 - 2y + m) 4:y + {X = x^ + 2 xy - 8 y^ + {I + in)x + {im -2l)y +

~{x +
I)

Im.

(1)

But

to
l

make

(1)

an identity, we must have,

285,

+m=
(2)

2 (2),
(3)

find
I

From
for

and

we

= 14 (3), 2 + 4 = 1, m = S. And
Z

??i

Im

=-

(4).

these values satisfy (4);

=-

3.

Therefore x2
Note.
are.
If,

+ 2x2/-8 2/2 + 2x+14 2/-3 =

(x

+ 4?/-

l){x

-2y + S).

The example shows how exceptional these composite functions 3, leaving A otherwise unchanged, we replace the last term,

by any other number, by Z = 1, m = 3.

becomes prime

for (4) will then not be satisfied

This method is also applicable to ho7}iogeneous quadratic functions of three variables.

190
Thus, to factor x^

A COLLEGE ALGEBRA
+ xy 2

+ 2xy - Sy"^ + 2xz + Uyz - Sz"^, we set - 3 22 = (X + 4 + Za) (x - 2 + 4- 2 xz + 14 7/z = 3. and then proceed as above, again finding = 1, Example 2. Factor 2 x^ - 7 xy + 3 y^ + 5 xz - 5 yz + 2 z^. Example 3. Show that x- y- + 2x + y 1 is prime.
x2

8 y2

2/

2/

mz)

??i

447

Polynomials

of the

nth degree.

polynomial of the second degree,

of factors of the first degree.
in

aox"^

We have shown that every + aiX + a^, is the product

The like is true of polynomials X of every degree, though no general method exists for
Every polynomial in
f (x)

finding these factors; in other words,

Theorem.

x,

of the nth degree,

h

aox"

+ aix-i

-\

a_ix

+ a
is,

product ofn factors of the first degree; that x /3^, such that n binomials, x jSi, x (3^, ,
is the

there are

f (X)

ao(x

/?i)

(X

^2)-

(x

- ^).

The
448

proof of this theorem will be given later.

Corollary.

y, of the

A homogeneous polynomial in two variables, x and nth degree, is the product of n homogeneous factors in
+
aixh/

X and

y, of the first degree.

Thus, the homogeneous polynomial aox^

a^xy'^

a^y^

may

be derived from aoX^

a^x"^

aox

as

by substituting x/y for x and

multiplying the result by

y^.

But by
if

447, aoX^

OiX^

a^x

03

= ao (x ^1) (x

we

substitute

x/y

for x in this identity

/Ss), and /So) (x and then multiply both members

by

y^,

we

obtain

aox3

aix~y

a^xij'^

azy^

= ao(x - /3iy) (x - ^^y) (x - PzV)-

EXERCISE XIX
Factor the following expressions.
1.

x2

14 X

48.

2. 4.
6.

x2

21 X

120.

.3.
5.

5x2-63x-22.
64x2-21x +
2.

16x2

12x2

+ 64x + 63. + 20x2/ - 82/'.

FACTORS OF INTEGRAL EXPRESSIONS

7.
9.

191

X*
x2

13 x2
3X

+
3.

36.

8.

x-^y

x'^y"-

3.

18 xy^.

11.

13.
15.

+ x2 - 4 xy - 2 y\ ate2 - (a2 + W) x - (a^ - 62).

x'^-8x;/

10.

3 x2
X-

2X

12.
14. 16.

6 ox

9 62

_
6x

18 ah.

^2

6d

dx

cx2

cdx.

+ 152/2 + 2x-

4?/-3.

x2+3x2/ + 22/2 + 32x -1-5^2+222.

APPLICATIONS OF THE REMAINDER THEOREM AND

SYNTHETIC DIVISION
On finding factors by aid of the remainder theorem. Let f{x) denote a polynomial in x. By the remainder theorem, 415, if h denote a number such that y(i) = 0, then cc is a factor of
ft

44S

f{x)'

We
~

can sometimes find such a number

Factor/(x)

by inspection.

Example.
"*"

"^

1 1
1

^
4,

L
X2

5x + 4. Since /(I) = 1 + 4 = 0, x 1 is a factor of /(x). 1, we obtain the quotient Dividing /(x) by x + X-4. - 1) (x2 + X - 4). jjgj^^g ^^^^ = (x

x^

-5

Note.

of the coefficients of /(x)

Observe that whenever, as in this example, the algebraic is 0, x 1 is a factor of /(x).

sum
450

If asked to factor a Polynomials with integral coefficients. polynomial, /(x), with integral coefficients, it is usually best to look first for any factors of the first degree with integral These ma}' always be found by coefficients that it may have.

aid of the following principles, 451, 452.

+ , with integral polynomial fix) = ao" + a^x''-'^ + may have a factor of the form x ~b, where h is an But if so, b must be a factor of a^, the constant term integer.

451

coefficients,

o/f(x).
Thus,
/(x),
let

/(x)

Oox^

+ =
4-

aix2

a^x

+
+

03.

If

is

to be a factor of

we must

have, 415,
/(6)
ao63

+
+

ai62

a^b

as

and therefore
Therefore, since O062
ai6

{a^ffl

aib

aa) 6

= 0, = az.
6 is a factor of 03.

ag denotes

an integer,

192
Hence
all

A COLLEGE ALGEBRA
such factors x

b may
3a;<

be found as in the follow-

ing example.
Example.
Factor
/(a;)

3a;6

ISx^

lli2

lOx

6.

The factors must have one

of the constant term, - 6, are 1, 2, 3, 6, of these values if x - 6 is to be a factor of /(x).

and b
test

We

these values of b as follows by synthetic division.

_ 3 _ 1,3 _ 11 _ 3+ 6+ 7+ 3 36^^ 1^ 4 3
3

10

4
6

Since /(I)

5^ 0,

1 is

not a factor.

01

_g, g_ 24. gl7^;;jl 2~ 6 9 0+6*

3 Q
2

013

We therefore begin by testing X ( 1) or X + L The division proves to be exact, the quotient being Qj = 3x* 6x3 7x2 4 X 6, and the remainder
H^^ce X + 1 is one factor of /(x), ^^^ ^^ '^ ^^^ product of the remaining
^'

factors.

We
X
X

have next to factor

Qi.

It also

+ +

1,

To
1,

the quotient being Q2 = 3x^ - Ox^ + 2x - 6. factor Q2, whose constant term is also 6,

proves to be exactly divisible by

vt^e

but in each case obtain a remainder which Hence none of these are factors. But testing x 3, we find that
2, 2,

test successively
is it

not

0.

divides

Q2 exactly, the quotient being Q3 /(x)

3x2

2.

Therefore

= (x+

l)2(x-3)(3x2

2).

452

(- a, with integral polynomial f{x) = a^x^ + a^x"-^ H coefficients, may have a factor of the form ax /3, where a and (i denote integers which have no common factor. But

if so,

a must

be a factor

of

ao,

and

fi

a factor of

a.

This

theorem includes that of

451.
03.

is to

Thus, let /(x) = flox' + aix2 + ojX + be a factor of /(x), we must have,

If

ax -

/3,

or

a (x -

/3/a),

415,

^ - + as = 0, a = 0. and therefore Oq/S^ + aiP-a + a2/3n-2 + asa^ (1) From (1) we obtain ao^^= (ai/32 + uo^a + 3a-2)Q-. (2) Therefore, since 01^32 + OoPcr + a^a"^ is an integer, nr is a factor of ao,8'But ct has no factor in common with 2. Hence a is a factor 492,

/(D

i3\ *

flo
,

i33

a*

^2

ai

^, a^

a2

/3'',

of ao, 492,

1.

Again from

(1),

(aoi32

ax^a

a2f^^)/3

=/3

azo^,
is

(3)
03.

whence, reasoning as before, we conclude that

a factor of

FACTORS OF INTEGRAL EXPRESSIONS

Hence
all

193

such factors ax

/3 may
5 x^

be found as in the

following example.
Example.
If

Factor f{x)

6 x*

3 x^

3x

2.

ax /3 is to be a factor of /(x), a must have one of the values 1, 2, 3, 6, and /3 one of the values 1, 2 therefore p/a must have one of the values 1, 2, 1/2, 1/3, 2/3, 1/6. We may test ax ^ for these various values of ^/a by dividing /(x)
;

by X /3/a

If the division is exact and Q denotes the synthetically. ,8 is a factor of f{x) and Q/a is the product of the quotient, then ax

remaining factors, 412, 3. 1, x + 1, x Testing x

6

2,

+ 2,

successively, vre find that

exactly.

none of

+ -

+ 3-3-2-1-1/2 - 1 -

them divides /(x)

But

x+

1/2

^^gg^ 2x

^j^g

quotient

4.

being Qi = 6 x^ + 2 x^ Hence 2 x + 1 is one factor of

/(x) and the product of the remaining factors is Qi/2 = 3x=' + x2 + x - 2.

We
ax
/3

is

next proceed to factor Qi/2. If to be a factor, ^/ a must have one

we

of the values 1, 2, 1/3, 2/3. But 1, x + 1, x 2, x + 2 are not factors, since already know that x 1/3, x + 1/3, we find that they are not factors of /(x). Testing x neither of them divides Qi/'2 exactly; but x 2/3 does, the quotient

being Q2

3x'-

+ 3x +
fix)

3.

Hence 3x
is

is

product of the remaining factors

(^2/3

x-

a factor of Qi/2, and the + x + 1. Therefore

(2

1) (3

2) {x2

1).

is

often becomes evident before a division by x completed that the division cannot be exact.

Note.

It

6 or

^/a

453

Thus, the reckoning here given

suffices to

prove that x
8 exactly
;

2 will not

divide 5 x^

2,

4 x-

for since

the

"divisor"

the last coefficient of

Q already

found,

namely

6,

and the unused

coefficients of the divi-

194
454

A COLLEGE ALGEBRA

h a^, It follows from 452 that a polynomial f (x) = x H whose leading coefficient is 1, the rest being integers, cannot vanish for a rational fractional value of x.

For

if

/(/3/a)

0,

therefore

a must be

then f{x) must be exactly divisible by ax - /3 pnd a factor of 1, which is only possible when a = ^. 1.
It follows

455

Factoring polynomials and solving equations.

from

350 that the problem of resolving a polynomial f{x) into its factors of the first degree is essentially the same as that oi
solving the equation f(x)

=
=

0.

Example

1.

Solve fix)

2 x*

+
x

a;3

17 x2

16 x

12

= 0.

By

452 we find that /(x)

(2

1) (x

2)2 (x

3).

Hence the equation /(x)

2x-l = 0,
Therefore the roots

= is equivalent to the four equations x-3 = 0. x + 2 = 0, + 2 = 0, are 1/2, - 2, - 2, and 3. of f(x) =

x

Example

2.

Solve x^

Transposing,
Factoring,

Hence the required

3 x2 = 10 x + 24. + 3 x2 - 10 x - 24 = 0. - 3) (x + 4) = 0. (x + 2) (x roots are 2, 3, and 4.

x^

EXERCISE
Factor the following expressions.
1.

XX

i3-7x +
X*

3.
5.

7.

2. x3 + 6x2 4- llx + 6. 6. + 35x2 _ 50x + 24. 4. x* - 2 x2 + 3 x - 2. 6. 2x3 - 5x2y - 2x2/2 + 22/8. 6x3-13x2-14x-3. 8. 4x6 - 41x* + 46x2 - 9. 2x*-x3-9x2 + 13x-5. 9. 6x5 + 19x* + 22x3 + 23x2+ 16x + 4. - 4. 10. 5x6 - 7 x5 - 8x* - x3 + 7 x2 + 8 X

10x3

Solve the following equations.

11. 13.

x2-4x-12 = x2-5x = 14.

x8-l =
0.

0.

12. 14.
24.

15. 17.

x8-9x2 + 26x =

16.

18.

6x2_7x + 2 = 0. + 6x = 2. x* + 2x8 - 4x2 - 2x + 3 = 10x8 - 9x2 -3x + 2 = 0.

x2

FACTORS OF INTEGRAL EXPRESSIONS

EXERCISE XXI
The following expressions can be factored by methods explained
present chapter.

195

in the

Carry the factorization as far as

is

possible without

introducing irrational or imaginary coefficients.

1.

6 xy
a3 (a

3. 5.

+ 15 X - 4 - 10. - 6) + h^ {b - a).
2/

2.

a"-bc

ac"-d

- obH +

hcd"^.

4.
6. 8.

a^

81 a6*.

a*6

a-b^

+ aW -

a.

3 abx^
(x2

6 axy

+ bxy-2 y\

7.
9.

3x6-192 2/6,
64x62/3_yi5.

+ x)3-8.

10.
12.

x'^-{a-b)x-ab.

11. 13. 15. 17.

19.

x2-3x"-18.
3x*
2 xc

+ 3x3-24x-24. - a2 + x2 - 2 a6 +
&2--5(f

14.
c2

X-X2 + 42. x^ - Ox^ + 8x2 3,2

72.

62,

iq
18.

(^2

20)

+ +

64.

a2-2a6 +
6x2

7a;2/

52/2

+ 56 + G. - 4x - 2?/.

x*

10 x2?/2
(a2

9 y*.

^20.
22. 24.

b^)x'^

a262.

21.
23.
25. 27.

4(x2 + M2/)2-(x2-2/2 + 2;2-u2)2.

1

14x2
xy^
(x2

+ 19x-3.

a*

+ 19 y - 66 - 18 a262c2 +
+
2/)''

2/2.

81 6<c*.

26. 28. 30.

+ 55 x'^y^ + 204 x^y. - 7 x)2 + 6 x2 - 42 x.

{y

8 (X

ftx

27 (X

2/)3.

{x

-2y)x^ -

- 2x)yK

29.
31.

x2
x5

+ -

a2
x''

_ -

a6

2 ax.

x^
6*

y^

(:x

yf.

X-'

2 x2

1.

32. 34.
36.

33.
35.
37. 39.

2x2 + 7x2/ + 32/2

+ 9x + 22/-5. 25

0^

+ 62 + + 4.

i.

x2-X2/-22/2 + 4xz-52/z + 322.

x2
(x2

4 a*
x^

+ +

3 a262

96*.

8 ax

40 a6

62.

38.
l)2.

x*

1.

41.

x3

+ 2x-l)2-(x2-2x + - ax2 - 62x + a62.

40.
42. 44.

(ax
x^

62/)2

(6x

a2/)2,

bx^

a^x

a36.

43. 45.
47.

a2-962+ 126c-4c2. x* - 2 x3 + 3 x2 - 2 X +
x2

1.

46.
48. 50.

(ax

4x5+4 x-37x3-37x2+9x+9.

+ 12a2 + 6a + 1. + by)"^ + (6x - ayf. x4-4x + 3.

8a3
15x3
2 x^

49.
51.

+ 5ax + 6a2- a6-62.

abcz^

(a262

c^) x

ahc.

52.

+ 29x2 - 8x - 12. - ax2 - 5 a^x - 2 a^.

196
53.

A COLLEGE ALGEBRA
(a

6) x2

2 ax

55.

x*-6x3 +
57.
58.

7x2

3x5
5x*
aJ^bc

+ (a + b). + 6x -8. - lOx* - 8x3 _ + +

24x3
ac-

54. 56.

x^s

yis.

4x3-3x-l.
iqx

3x2

8.

15x2

59.
60.

acd

-118x + 24. abd cd d^.

X*

^ + 2* - 2 x22/2 - 2 y2z2 _ 2 22x2,

VII.

HIGHEST COMMON FACTOR AND LOWEST COMMON MULTIPLE

HIGHEST COMMON FACTOR

456

Let A, B, denote rational, inteHighest common factor. gral functions of one or more variables, as x, or x and y. have no factor in common, we say that they are li A, B, prime to one another. But if they have any common factor,

they have one whose degree

is

highest

we

call it their highest

common factor
Thus, x2

(H.C.F.).

+ 2/2 and x + ?/ are prime to one another. But 4 xi/z^, 8 xz*, and 4 x'^yz^ have the common factors xz2, xz^, and their highest common factor is xz^.
457
Notes.
2.

x, z, z^, z', xz,

1. We here ignore common numerical factors. sometimes said of two or more functions which are prime to one another that their H.C.F. is 1. 3. The numerical value of the H.C.F. of A and B is not necessarily the greatest common divisor of integral numerical values of A and B. Thus, the H.C.F. of (2x + l)x and (x l)x is x. But when x = 4, the values of (2 x + l)x and (x l)x are 36 and 12, and the greatest common

It is

divisor of 36

and 12
\.

is

not

4,

but

12.

458

The H.C.F. of K, B, is the product of all the different common jmnie factors of K, B, , each raised to the loivest 2'>ower in irhich it occtirs ij2 any of these f met ions.
Theorem

The truth

of this theorem

is

obvious

if

we suppose each

of the functions A, B,--- expressed in the

form of a product of

HIGHEST COMMON FACTOR

powers of
that there
its
is

197
assume

different

prime factors and, as

in 430,

but one such expression for each function.

Thus, the different common prime factors of xyz^, xz*, and x^yz^ are X and z, and the lowest powers of x and z in any of these functions are x and z^. Hence the H.C.F. is xz^.

Observe that
than one

if

it

way

in

terms of

were possible its prime

to express a given function in

more

factors, the process described in the

theorem might lead of expressing A, B,

of highest degree.

to various results

corresponding to the various ways and there might be more than one common factor

Applications of this theorem.

When

be completely factored, their H.C.F. once by aid of the theorem of 458.

may be

the given functions can written down at

459

Example

Find the H.C.F. of x^y^

x^y"^

We
and

have

6 x*y^
X*
?/

- 6 x'^y^ + 9 x^y* and x*y - 9 xV. = x^y^ {x - S yY 9 x'^y'^ = x-y (x 3 y) (x + 3 y).

9x^y*

Hence the H.C.F.

is

x^y (x

3 y).

Example

2.

1.

Find the H.C.F. of the following. 2xVz^ Sx^y% and ix^y\

x^
x2

2.

3.

2/2,

x^
6,

2 xy

y^,

and

x^

x2

6X

8,

and

x^

+ +

t/^.

4.

x3-6x2 + llx -6 and 2x3-

9x2

x + 6. + 7x +

6.

prime factors of one of the functions A, B, are we can find by division or the remainder theorem known, which of them, if any, are factors of all the other functions. The H.C.F. may then be obtained by aid of 458.
If the

460

Example
f{x)

1.

Find the H.C.F. of

3X

= x2 we
^

and

<;&

(x)

inspection in ^(x), we find

By

have/(.c)

(x

1) (x

= x* - 3 x^ + 5 x2 - 8 x + 2). Testing x = 1
Hence the H.C.F.
is

5.

and x
x

=2

(1)

0,

but

<}){2) jt 0.

1.

Example
/(x)

2.

Find the H.C.F. of

x2

+ 4x +

and

(^(x)

x*

Sx^

9x2

+ 8x +

4.

Since /(x) = (x + 2)2, we must find not only whether x + 2 is a factor of Dividing <^ /xi bv a: ^ a (xjj but whether it is a factor once or twice.

198

A COLLEGE ALGEBRA
obtain Qi

(synthetically) we Qi by X + 2 we obtain

i^

Q^^x^ +
2)2.

+ 3x2 + 3x + 2 and Ri = + 1 and E2 = 0. Hence

dividing the H.C.F.

;

of/(x) and0(x)is

(x

Example
1.
x"^

3.

Find the H.C.F. of the following.

2. 3.

x2
(X

+ X-6 and 2 x^ + 7 x^ + 4 x + 3. + 5x + 6 and x*4- 6x3 + 13x2 + 16x + 12. - l)2(x - 3)3(3x + 1)2 and x< - Sx^ + x2 + 21x 2.

18.

461

Theorem

Let

and

and

denote two given integral functions, any two integral functions or constants. Then
is

A and B

every

common factor of A and B

let

a factor of

MA + NB.
B.

Por

F denote

common

factor of

A and

Then
where

A=
MA

GF

and

B=

HF,

Hence
where

G and H are integral. NB= M- GF + N HF= (MG + NH) F, +

A^ + NH
F is
a factor of

is

integral.

Therefore

MA + NB,

424.

462

By aid of this theorem the Applications of this theorem. in x, whose problem of finding the H.C.F. of two polynomials
of factoring a degrees are the same, may be reduced to that polynomial of a lower degree. single

Example
Hence,

1.

Find the H.C.F. of

^ = x2 + 2 x -

4
\.

and

B=

x^

3.

Subtracting

B from
x

A, we obtain

A B = x

461,

1 is

But, since Therefore

A does not vanish when x = 1, x A and B are prime to one another.

Find the H.C.F. of
2

the only possible

common

factor of

and B.

1 is

not a factor of A.

Example
.4
1.

2.

= 2x3-3x2-3x +
begin by multiplying

and

We

and

results having the

Then
3

subtracting 2
5 x2

same leading term, B from 3 -4, we obtain

B = 3x3 - 2x2 - 7x - 2. B by numbers which will give namely, ^ by 3 and B by 2.

X

^ - 2B=-

5X

10

=-

6 (x2

2)

=-

5 (X

1) (X

2).

Hence the only possible common factors of A and .B are x +

and x

2.

HIGHEST COMMON FACTOR

By
the remainder theorem

199

we

find that both are factors of

2).

and B.

Hence the H.C.F. of A and B is {x + 1) (x 2. Or we may add A and B, thus obtaining

J.

+S=

5x3

5x2

- 10x = 5x(x2-x-2) = 5x(x +

is

l)(x-2).

It is at

once evident that x

not a factor of
2.

or B, so that as before

we have

only to test x
3.

and x -

Example
1.

Find the H.C.F. of the following.

12

X*

-x^ + 3x2 _ 4x -

and

x*

x3

2.

6x3

25x2

^ 5x +

and 4x3

15x2

+ 2x2 + 3x - 2x + 8.

22.

Theorem
the

3.

related that

A = QB +
the

If the four mtegral functions A, B, Q, R are so R, the common factors of A and B are

463

same as
have

common factors of B and R.

We

A
from

= QB + R,

(1)
(2)

and therefore
It follows

A- QB= R.
(2),

by 461, that every common factor of A and Bis Si factor of R and therefore a common factor of B and R. And, conversely, it follows from (1), by 461, that every common factor of B and i2 is a factor of A, and therefore
a
of

common

factor of

A and

B.

Hence the common

factors of

A and B

are the

same

as those

B and

R.

The general method for finding the H.C.F. of two polynomials When one polynomial in x is divided by another, the dividend, divisor, quotient, and remainder are connected by Hence it follows from 463 that the identity A ^ QB -{- R.
in X.

464

The common factors of dividend and divisor are always same as those of divisor and remainder.
mials in X

the

By making use of this fact, the H.C.F. of any two polynomay always be found. The method is analogous to that employed in arithmetic to find the greatest common
divisor of

where

yl

two integers. It is described in the following rule, and 5 represent the given polynomials, A the one of
if

higher degree

their degrees are not the same.

200
465
Rule.

A COLLEGE ALGEBRA
Divide

K
B

hy

^ and
and

call

the qriotient

q and the
qi

remainder Rj.

Next divide
^

by

E.^

call

the

quotient

and the

remainder RgNext divide

E-i

new remainder by
I

by Rgj *'^^ *o on, continually dividing each the one last obtained, until a remainder is
x. is

reached which does not involve

If

this final

remainder

not 0,

and

have no common
the H.C.F. of

factor.

If
B.

it

is 0, the divisor

which yielded

it is

A and

For suppose, for the sake of definiteness, that the final Then according as (1) R^ = c, where c is R^denotes a constant not 0, or (2) R^ = 0, we shall have
remainder
(1)

(1)

In

+ 7?i q,Ri + i?2 R^ = q^Rz + c this case A and B

.4

qB

or

(2)

B =

B =

=qB + R^ qiRi + Rs = qzRo i?i

A

have no common factor.

For it follows from the identities (1), by 463, that A and have the same common factors as B and Ri B and Ri, as Ri and i?2 ^i and Ro, as i?2 and c.

Hence the
Ro and c, But as
factor.
(2)
all
c

pairs of functions

have the same

is

A and B, B and common factors.

c

R^,

R^ and Ro,

a constant (not 0), R2 and

have no common

Hence A and B have none.

In this case Rg
*'*

the H.C.F. of

and B.

of

For since 7?i i?i and /?2>

q2.R2j every factor of R^ is a common factor and R^ itself is the common factor of highest

degree.

But

as the

common

factors of ^1

and R2 are the same as

those of

A and

B, the factor of highest degree

and Ro is also the factor of highest degree common Hence R^ is the H.C.F. of A and B.

common to R^ to A and B.

HIGHEST COMMON FACTOR

Example
1.

201
X'

Find the H.C.F. of 2^

and

x^

+ 2x +

3.

Writiug divisors at tlie left of dividends, we have B = x'^ + x + l\x^ + x:' + 2x + S\x- q X a;^ + g^ +

Ri=

3|a;2
X?-

+ x+ l \x-2 =
+
3X
1

qi

- 2x +
-R2

-2x -6
=
+
x
7

As

the final remainder,

i?2, is

not

0,

x^

and

x^

x^

2x

have no

common
2.

factor.

Example

Find the H.C.F. of

x3

x2

2X
in

and
1,

x3

2 x2

3X

2.

Arranging the work as

B = x3 +

x2

we have + 2x + 2jx3 + 2x2 + 3x + 2|l x3+ x2 + 2x + 2

Ex.

Rx

x^

X
i?2

|x3

X3

+ x2 + 2x + + X2 2x +

2|x
2lx2
X?

+ x |x/2 +X

i?3=
Here the division by
numerical factor 2 in
i?2 is

i?2,

R% being we have H.C.F. = x

exact,

0.

Hence, discarding the

1.

Here

for the first time

we have an

tions of a

single variable

that

actual proof

for

f unc-

466

if

two integral functions

have any common factor, they have a highest common factor; for in 463, 465 it is not assumed as in 458 that an integral function can be expressed in but one way in terms of its

prime

factors.

Observe that in the proof in

1.

465

it is

shown that
list

467
E,2)

Every two consecutive functions in the have the same H.C.F. as A and B.
2.

A, B, Ei,

Every common factor of

method.

A and B

is

a divisor of the H.C.F.

of A and B.
Abridgments
of
of this

or

iJ

are obvious

1. If any of the prime factors by inspection, begin by removing these

468

202
factors

A COLLEGE ALGEBRA
and then
find the H.C.F. of the resulting expressions.

The result thus obtained, multiplied by such of the factors removed at the outset as are common to A and B, will be the H.C.F. of A and B, 458. The same course may be followed with any two consecutive since every two such functions in the list -1, B, R^, R^, functions have the same H.C.F. as A and B, 467.

Thus, A=x^ + x^ + 2x'^ + 2x and B = x* + 2x^ + Sx'^ -\- 2x obviously have the common factor x. Removing it we have x^ + x- + 2x + 2 and to be x + 1 (see a;3 -I- 2x2 + 3x + 2, whose H.C.F. we have just found and B is x(x + 1). 465, Ex. 2). Hence the H.C.F. of it cannot be a Again in Ex. 2, since x is a factor of Ri, but not of B, factor of the H.C.F. of B and i?i, and is therefore not a factor of the H.C.F. of A and B. Hence we may discard this factor x of Ri and divide B by the remaining factor x + 1, so lessening the number of divisions.

2.

divisor or dividend or
ical factor
;

In any of the divisions we may multiply or divide the any intermediate remainder by a numerfor this will affect the subsequent remainders

by

a numerical factor at most, 403, and therefore the H.C.F. not This device enables us to avoid fractional coefficients at all.

when the given

3.

coefficients are rational.

coefficients.

It is

advantageous to employ detached

1.

Example

Find the H.C.F. of

.3x3

^ = X* +
Multiplying
2

2x2

+ 3x +

and

J?

2x3

5x2

1.

.4

by 2 and using detached

6
1

coefficients,

we have

+ 5-1 -l|2 + 6+ 4+ 2 + 5- 1 -

+ +

2[l
2

1+ 5+
2

+ 10+
5 )5

14

+
+ +

2+ 5- 1-1
+
15 3

4[1
5

1+
Hence the H.C.F.
is

l|2
2

of

^
1.

and

+ 5-l -1 |2-1 +6+2 _i _3- 1

x2

3X

-1-3-1

HIGHEST COMMON FACTOR

This reckoning
2 2

203

may be more compactly

1

arranged, as follows;
1

+ 5+6+

204

A COLLEGE ALGEBRA
;

For (1), D^ is a factor of Zi and C and D^ is a factor of A and B. Hence, 427, D^ is a factor of ^, B, and C. And (2), every common factor of A, B, and C is a common
factor of
>!

and

C,

and therefore a factor of

D^, 467.

Hence

Dg

the highest common factor of A, B, and C. The same conclusion follows from 458.
is

Example.

Find the H.C.F.

oi
7

A = x^ +
+
2,

x^

x"

x^

2,

B=
By

2 x*

5 x^

2 x2

and

C=

3 x*

x2

2.

465,

we

find that the

(7

that the H.C.F. of Di and

H.C.F. oi is x - 1.

A
x

a,nd

B is Bi-

x-

2;

and

Hence the H.C.F.

of

A, B, and

is

1.

471

The polynomials in more than one variable. general problem of finding the H.C.F. of two such polynomials But the H.C.F. of is too complicated to be considered here.
The H.C.F.
of

two polynomials which are liomogeneons functions of two variables, as X and y, may readily be found by aid of the rule
given in

465.

EXERCISE XXn
Find the H.C.F. of the following.
1.

10
(a
2/*

x^yV,

4 x^yz^, 6 x'^yH^,

and 8

x*2/^z*u.

2.
3.

6)2 (a
2/2

6), (a

+
-

+
a2

and y^
2

+ 1,

h) {a

\.

6)2,

and

a?b

ahK

4.
5.
6.

a2

1,

and
ax

a^

+
1.

1.

x^
X*

and

x^

ax^

+ -

7.
8.
9.

x2
(X

+ 2/", and x^ + x'^y + x?/2 + y^. + 6, x2 + X - 2, and x2 - 14 x - 32. - 2) and 5 x* - 15 x-^ + 8 x2 + 6 x 1) (x

y*, x^

5 X

4.

x3
(x2

and
1)2 (X

x^

4 x2

4 x

5.

10. 11. 12. 13.

+
-

1)2

and

(x^ (x2

(X

l)2{x

2)2

and

+ 5 x2 + 7 x + 3) (x2 - 6 x - 7). - 3 x + 2) (2 x^ - 5 x2 + 5 x - 6).

x3

2 x3
z8

3 x2

11 X

and 4

+
x2

3 x2

9X
1.

2.

2 x2

2 X

3 and 2 x8

HIGHEST COMMON FACTOR

14.

205

3 x3
a;*

+
x3

2 x2

19 X

+
+

and 2

x3

+ +

x2

13

a;

15.

3 x2

a;

2 and 2 X*

3 x3

X'-

^ -

6;

3X

1.

16.
17.

3 x3 3 x3 6 x5

+ + -

13 x2
8 x2
7 x*

18. 19.
20.

+ 23 X - 21 and 6 x^ + x2 - 44 x + 21. - 4 X - 15 and 6 x* + 10 x3 - 3 x2 - 2 X + 5. - 9 x3 - 7 x2 + 3 X and x5 + 7 X* + 3 x3 + 7 x2 -

3 X.

21.
22. 23.

+ 7 x2 + X - 3 and 2 X* + 3 x3 + 7 x2 + 3 X + 9. 6x5-4x*-llx3-3x2-3x-l and4x* + 2x3- 18x2 + 3x-5. x5 - x3 - 4 x2 - 3 X - 2 and 5 x* - 3 x2 - 8 x - 3.

6 x<
3 x3

3 x3
x3

- x2 ax2

12 X 3X

+ 4, x3 - 2 x2 - 5 X + 6, - 3 a, x3 - x2 - 3 X + 3,
4
xy'^

and

7 x3 4-

19 x2

8X

- 4.

and

24. 25. 26.

7 x*y

- 6 xV _ 18 x2;/3 +
1) (x3

and 14 x^y

+ - 19 x^-y^ - 32 xy^ + 28
x^ x2

3x

3.

?/*.

X (x
4 x3

8 x2

+ 4 x2 + 4 X + 3) and (x - 1) (x + 3) (12 x^ + x2 + x - 1), - 3 X + 9 and (2 x2 - x - 3) (2 x2 - 7 x + 6).

LOWEST COMMON MULTIPLE

Lowest common multiple.
integral functions, A, B,---,

A
is

common multiple

of

two or more
is

472

an integral function which

exactly divisible by each of the functions A, B, . Among such common multiples there is one whose degree is lowest. call this the lowest common multiple (L.C.M.) of

We

A,B,....

Theorem
A, B,
,

A, B,

The L.C.M. of tioo or more integral functions, the product of all the different prime factors of each raised to the highest power in ivhich it occurs in
1.

473

is

any of these functions.

This follows from the fact that a common multiple of .4, 5, must contain every prime factor of each function J ,,
least as often as
it

at

factors mentioned in the theorem.

occurs in that function, hence all the And the common multiple

of lowest degree, that

is,

the L.C.M.,

is

the one which contains

no factors besides

these.

206

A COLLEGE ALGEBRA

Here, as previously, we ignore numerical factors and assume that an integral function can be expressed in only one way as a product of powers of its different prime factors.

474

Finding the L.C.M. by inspection.

into their prime factors,

we may

If we can resolve A, B, obtain their L.C.M. at once

by applying the theorem just demonstrated.

Example Here the
1.

Find the L.C.M. of Sx^y"z,

xy*z^, a,nd 2 x^yz^.

which it Hence the L.C.M.

different prime factors, each raised to the highest occurs in any of the functions, are x^, y*, z^.
is x^y^z'^.

power

iu

Example
"We have

2.

Find the L.C.M. of

2/2

xhj'^

xV -4x2/2 + 4 2/2=
is 2/2 (x

Hence the L.C.M.

(x

2)2 (x

- 4 xy- + 4 ?/2 and x'^y -iy. _ 2)2 and x22/ - 4 = (x -2)(x4-2). + 2).
?/ ?/

475
is

Theorem

2.

The L.C.M. of two

ttvo,

the product of the

integi'al functions, divided hy their H.C.F.

and B,

For

let

denote the H.C.F. of

A and

B,

and

let A-^

denote the quotients obtained by dividing

A and B by

and B^ D, so that

A = A^D and
Then
if

B= B^D.

M denote the L.C.M. of A and B, we have M= AiBiD= AB/D.

For evidently a common multiple of A and B must contain the product of all prime factors common to A and B, namely D, (2) the product of all prime factors of A not belonging to B, namely Ai, (3) the product of all prime factors of B not belonging to A, namely 7?i and the lowest common
(1)
;

multiple will contain no factors besides these.

476
is

Corollary.

The product of two

iiitegral functions,

and B,

equal

to

the product of their L.C.M.

and

their H.C.F.

477

General method for finding the L.C.M. of two polynomials in x. from 4G5, 475 that the L.C.M. of two such polynomials, A and B, may always be obtained by the rule
It follows
:

To find the L.C.M. of

and B, divide
by B.

hij

the H.C.F. of

A and

B,

and

multijjlij the result

HIGHEST COMMON FACTOR

207

Observe that this is equivalent to multiplying B by all those prime factors of A which are not already present in B.
Example.
X*

Find the L.C.M. of

3x3

2x'-

By

465,

we

find that the H.C.F.

Again (2x3 + 5x2 Hence the L.C.M. is

+ 5x- ~x-l. + 3x + 1. -x- i)/(x2 + 3x+ l) = 2x-l. - 1). (x* + 3x3 + 2x2 + 3x + 1) (2x
+ 3x +
1

and 2x^
x2

The L.C.M.

of

more than two polynomials in

x.

This

may

be

478

obtained by the following rule.

First find the L.C.M. of tivo of the polynomials, L.C.M. of the result and the third poli/nomial, and so final result tvill be the L.C.M. required.
7iext the

on.

The

This follows from the fact that each step in the process is equivalent to multiplying the L.C.M. last obtained by those prime factors of the next function which are not alreadypresent in that L.C.M.
3 x3 + 2 x2 + 3 x + 1, = 2x3-3x2 + 2x-3. As we have just shown, 477, Ex., the L.C.M. of A and B is Ml = (X* + 3 x3 + 2 x2 + 3 X + 1) (2 X - 1).

Example.

Find the L.C.M. of

^ = x* +

B=

2 x3

5 x2

1,

and C

have next to find the L.C.M. of Mi and C. we find that 2 x 1 is prime to C, and by 465 we find that the H.C.F. of x* + 3 x3 + 2 x2 + 3 x + 1 and C is x2 + 1.

We

By

division

Furthermore,

Hence the L.C.M.

Jlf

of

C/ {x- + 1) = Mi and C, and

2x

3.

therefore of

(X*

3 x3

2 x2

3X

1) (2

1) (2

A, B, C, is X - 3).

Observe that we do not multiply the factors of proceeding to find the H.C.F. of Mi and C.

Mi

together before

EXERCISE XXra
Find the L.C.M.
1.

of the following.
i^

3x (a

1,

9x2

and 9x2
(a

i.

2.

6) (aS

65)

and

b) (ofi

lf>).

208
3.

A COLLEGE ALGEBRA
a^
-^ a"^

a, a^

o^,
(X*

and

a^

a'.

4.
5. 6.

_ y3) (X (x3
x2

y)\

y*) (x
6,

yy, and
x^

(x2
3.

^2)2.

3X
(2/

2, x2
z)2,

5X

and

4x
{x

+
+

x2

y"--{z
92/2,
1

x)2,

and

z^-

y)^

7.
8. 9.

2x2
x^

+ ^xy x2

3x2

and

+ g^y - 3?/2, x^ - x2 + x - L

and 6x2

- Uxy + 3y\

2 a2x

10.
11.

8x3
x3
x6

+ 2 x2y + 3 y2x + 3 a"-y and (2 x2 - 3 a^^) + (2 a2 - 3 y-)x. - 18x2/2, 8x3 8x22/ - 6x^2, and 8x2 -2xy - 152/2,
?/
_|_

2/')

^^

y^i

and

x*

+
4

x22/2
5,

?/*.

12.
13.
14.

1,

3x3

5x2

- 3x
36x2

and

x*

1.

8x3 ^27, 16x4

x2
x2
(x2 (x3

81,

and 6x2
8
a"^,

4 a-, x3 2 X, x2

2 ax2

+ 4 a2x +
+
2
6,

15. 16. 17.

6x

2X

and

+ 5x _6. and x3 - 2 ax2 + 4 a2x x3 + rtx2 - 62x _ ab^.

aK

+ 3x + 2) (x2 + 7 X + 12) and (x2 + 5x + 6) (2x2 - 3x - 5). - 8)(27x3 + 1) and (2x3 + 5x2 + lox + 4) (x3 - x2 - x - 2).
4-

18. 19.

x3-6x2
x*

11 X

-6, 2x3 - 7x2 +

5,

7x- 2,
+ 4X

and 2x3

+ x2 -

13x

+ 6.

5 x2

4X

2 X*

x3
3.

10 x2

5,

and

2 X*
20.

2x*
X*

+ x3 + 7 x2 + 3 X + - x3 + 2x2 + 3x - 2,
+
3 x3

2x*

3x3

4x2

i3x

6,

and

x2

5X

6.

ON THE PRIME AND IRREDUCIBLE FACTORS OF FUNCTIONS OF A SINGLE VARIABLE

In the following theorems A and

denote polynomials in
to

x.

479

Fundamental Theorem.
tions,

and N, can

be

//"A is prime found such that

B, two integral func-

MA + NB =

1.

For if we apply the method of 465 to A and B, we final remainder a constant, c, different from 0.

shall obtain as

HIGHEST COMMON FACTOR

If we suppose, as in 405, that notation there explained, we have
1.

209
and use the

c is

the third remainder,

A =qB +
B
Ei
=5ijRi

Ri,
/?2,
c,

and therefore

4.
5.

2.
3.

= q2R2

+ +

6.

Bi^A qB, = B - giZ?i, c = i?i - (?2^2E2

Substitute in 6 the value of Eo given by 5, collecting the i?i and the B terms, and in the result substitute the value of Ri given by 4,
collecting the

and
c

terms.
~

We

thus obtain

= Ri QoR^ = (1 + qm) Ri - q^B = (1 + gi'Zs) A-{q + q. + qqiq^) B.

Divide both sides of this last identity by c, and for (1 + q\q-i)/c and + 52 + Q'hq'i) / c, which are integral functions since c is a constant, write and N. AVe obtain

(?

and are integral functions. where, as just said, And we may demonstrate the theorem in the same
stant remainder,
c, is

l=MA+ NB, N

way when

the con-

obtained earlier or later than the third division.

1,

Conversely, If

MA + NB =
B.
of

where

and

are integral,

480

then

is 2)riine to

For a common factor

461, and therefore of
1,

A and B would be a factor of which is impossible.

MA + NB^

The following theorems are some of the more important consequences of the fundamental theorem just demonstrated.
Theorem
divisible
1.

If

is

prime
is

to

B,

and

the product

AC

is

481

by B, then

divisible by B.

For since

is

prime to

i?,

we can

find

M and N,
-l,

479,

such that

MA
and therefore

+ NB =
B.

M AC + NC
AC and
If

B=C.
Hence
it is

But

is

a factor of both
2.
to

a factor of C, 461.

Theorem
it is

prime

prime to each of the their p)roduct, BC.

is

fu7ictions

B and

C,

482

210
For since

A COLLEGE ALGEBRA A
is

prime

to B,

we can

find

M and N,

479,

such that

MA
and therefore
Hence,
461.
if

+ NB=l,
it

MC A + NBC=C.
and
is

BC

had a common

factor,
is

would be contained
to C.

in C,

But

this

impossible, since

prime

483

Corollary.

If

is

prune

to

each of the functions, B, C, D,

and

so on,

it is

j^^'ime to their product,

B C D

For, as just demonstrated, is prime to And since

A is prime to BC BC and also to X,

it is

prime to the product

BCD
484

and so on.
3.

Theorem

composite function has one

and

but one set of

prime factors.
For
If
let

denote the given function and n

its

degree.

P is composite it has
;

some factor A.

If y1. in turn, is

composite,

it

Continuing thus, we mu.st ultimately come upon a prime function for the degrees of the successive functions P, A, B, cannot fall below 1. begin with the finite number n, decrease, and Let F denote this prime function. It is one of the prime factors of P,
has some factor B.

427,

and we have
if

M= F'M',

Similarly

P= FM,

where

is

integral.

is

composite, a prime function F' exists such that

and therefore

P = FF'M',

where M'

is

integral.

Continuing thus, we reach the conclusion that a series of prime functions F, F', F", exists, whose number cannot exceed n, such that

P = FF'- F"

Hence P has at least one set of prime factors. Moreover P can have but one such set of factors.

For, suppose that

P = FF'F"
where G,
so, it
(?',

= G-G'

G"
if

G",

also denote

prime functions.

Then G cannot
of P.

be prime to all the functions F, F', P",

to their

for,

would be prime

product P,

483,

whereas

it

is

a factor

Then G is not prime to F, for example. Suppo.se, therefore, that G and F have a common factor. But G and F are prime functions, and two prime functions can have no factor in x but themselves in common. Hence G differs from P by a numerical factor, as c, at most, and we have

G = cF.

HIGHEST COMMON FACTOR

But substituting this value of G in the identity FF'F" and dividing both members by F, we have

211

= GG'G"

F'F"

= cG'G"---,

from which it follows by a mere repetition of our reasoning that G' differs from one of the functions F', F", by a numerical factor at most.
Continuing thus, we reach the conclusion that the set of functions G, differs from the set F, F', F", at most by numerical factors G\ G'\ or in the order in which they are arranged.

Corollary.

way

composite function can be expressed in only one as a product of powers of its different prime factors.

485

This follows at once from the identity P = F- F' each set of equal factors in the product F F' F" ing power of one of these factors.

F"

if

we

replace

by the correspond-

Irreducible factors.

By

function with rational coefficients, we usually of lowest degree with rational coefficients.

the irreducible factors of an integral mean the factors

486

+ V2,

Thus, while the prime factors of (x 1) (x^ 2) are x the irreducible factors are x 1 and 3fl 2.

1,

Vi,
487

From

469,

it

the theorems just demonstrated and the theorem of follows that

reducible integral function with rational coefficients can be expressed in only one way as a product of powers of its different
irreducible factors.

DIGRESSION IN THE THEORY OF NUMBERS

Theorems analogous
for integral numbers.
to those just

demonstrated hold good

488

We
A
than

shall

employ the

letters a,
"

b,

and so

on, to represent

integers, positive or negative (not 0), and shall factor of a any integer which exactly divides a.

mean by a
489

prime number
itself

is

an integer which has no other factors

and

1.

212
490
a
If
is

A COLLEGE ALGEBRA
two integers, a and said to be prime to h.
h,

have no

common

factor except

1,

491
he

Theorem,

//"a is prime to b, two integers,

m and n, can always

found such that

For since a
is

ma +
6, if

nb

1.

prime to
divisor,

we apply

the usual
1

greatest

common

we

shall obtain

method for finding the as the final renaainder.

We

may deduce
we have

the theorem from this fact by the reasoning of 479. Thus, let a = 325, b = 116. Applying the method for finding G.C.D.,

116|325[2 232

ri=~93]ll6U.
93

i.e.

325
116

= = =

116

+
+

93, or 93

= = =

325
116
93

-2
1

116
93

(1)

r2=~23]93[4
92
rs

1-

93 23

23, or 23

(2)

93

1,

or

23

(3)

Hence, starting with (3), and substituting first the value of 23 given by (2), and then the value of 93 given by (1), we have
1

93

23

Therefore

Hence we have

= 5-93 -4. 116 = 5-325- 14-116. 5 325 + (- 14) 116 = 1. found two integers, m = 5 and m-325 + n- 116 = 1.

14,

such that

And
492

similarly in every case.

Example.
Corollaries.

Find integers

and n such that 223

m+

125 n

1.

fimdamental theorem we may derive for integral numbers theorems analogous to those derived for integral functions in 481-485, and by the same reasoning.
this

From

In particular we
1.

may prove
by
to
b.

that
the product ac
is divisible

If
7/"

a.

is

prime

to b, aiid

hy b,

then c
2.

is divisible

is

prime

b and

c,

then a

is

prime

to be.

composite number can be expressed in one loay, and but one, as a jyroduct of powers of its different prime factors.
3.

RATIONAL FRACTIONS
VIII.

213

RATIONAL FRACTIONS
REDUCTION OF FRACTIONS

Let A and B denote any two algebraic expreswhich B is not 0. The quotient of A by B, expressed in the form A / B, is called a fraction ; and A is called the numerator, B the deno?ninator, and A and B together the terms
Fractions.
sions, of

493

of this fraction.

When
fraction.

both

A and B

are rational,

A/B

is

called a rational

494

When
fraction
fraction.
;

both
but

if

A and B are integral, .4 /B is called a simple 495 A or B is fractional, A/B is called a complex
496

simple fraction is called a proper or an improper fraction, according as the degree of its numerator is or is not less than that of its denominator.

_
Thus,

x-y
X2 -f
2/2

and

2x2_3
x^

are proper,

2x2+1
X2

and

x^
X2

3
1

improper. ^ ^

An

variable can be reduced to the

improper fraction whose terms are functions of a single sum of an integral expression
fraction,

497

and a proper
expression.

400.
1
-

This sum

is

called a mixed
X
X2

_ Thus,

2x2
X2

+1= +1

X2

x3-3 = X2 + 1

+ +

.3
.

Allowable changes in the form of a fraction. the following theorem, 320, 1.

These depend on

498

ator

The value of a fraction remains unchanged tvhen its numerand denominator are multiplied or divided hy the same
lue may change the signs of both numerator and denominabeing equivalent to multiplying both numerator and denominator Changing the sign of the numerator or of the denominator alone
itself,

expression (not 0).

In particular,
tor, this

by

1.

will

change the sign of the fraction

320, 3.

214
If the

A COLLEGE ALGEBRA
numerator or denominator be a polynomial, changing
all its
its

sign

is

equivalent to changing the signs of

terms.
c

a
Thus,
If the

b
b

c
c

a
c

ab b+c

a
b

+bc ca

we may change
fraction.

numerator, denominator, or both, are products of certain factors, the signs of an even number of these factors but chang;

ing the signs of an odd

number
{b
(e

of

them
d) g)

will

change the sign of the

Thus,

(a
(e

b)

jc-d) _
{g

-/)

h)

-a){c-/) {h -

(b

(/-

-a)(d- c) - h) e) (g

499

Reduction of fractions.
factors

To

simplift/ a fraction is to cancel all

which are common

this being a change in the aifect its value, 498.

numerator and denominator, form of the fraction which will not

to its
is

When

this has been

done the fraction

said to be in its

lowest teryns, or to be irred%icible.

We

discover what these

common

factors are, or

show that

look there are none, by the methods of Chapter VII. first for common monomial factors and other common factors

We

which are obvious by inspection or which can be fotmd by aid of the remainder theorem, and when these simpler methods fail we apply the general method of 465.

The following examples

Example
, We
,

will illustrate

some of these methods.

1.

Simplify {aec
aec

(c
^^

ade) / {bde

^^

ebc).

have

ade = bde ebc

Simplify

ae

be {d

d)
c)

'-

=
+

a(c
b{c

d) -= d)

a
b

Example

2.

{x^

x^

G)/{x-

+ 3x +

2).

+ I and x + 2. numerator and denominator have any common factor, it must be one of these. Testing by synthetic division, we find that the numerator is not divisible by x -|- 1, but is divisible by x + 2, the quotient being x2 - X + 3. X2 - X -f- 3 X3 -I- X2 -f- X + 6
By
inspection, the factors of the denominator are x
if

Hence

Hence

x2-|-3x-t-2

RATIONAL FRACTIONS
Example
3.

215
6).

Simplify (x^

+ -

10) / (x3

Subtracting denominator from numerator,

x3

we have

10

(x3

6)

2 (X

2).

Hence, must be X Hence,

if

the numerator
2, 461.

and denominator have any common factor, it But the numerator does not vanish when x = 2.
ah'eady in
its

415, the fraction is

"'^

lowest terms.

Example

4.

^^

"

^^

^'^

^^^

Simplify
(a

"") +

^'^

(^

^)

b)(b

c) (c

a)

Here the only possible common factors are a b, b c, and c a. = 6 in the numerator, we have b-{b c) + b-{c b), or 0. Hence, 417, the numerator is divisible by a b. And we may show in the same way that it is divisible by 6 c and c a. Therefore the numerator is exactly divisible by the denominator. But the two are of the same degree, namely three, in a, 6, c. Their quotient must therefore be a mere number and since the a- terms in the two, c) and a- (6 c) respecwhen arranged as polynomials in a, are a- (6
Setting a
;

tively, this

number is 1. Hence the given fraction

Examples.

is

equal to

1.

Simplify (2x3+

13x2-6x + 7)/{2x* + 5xH8x2-2x + 5).

numerator and denominator

By
is

2x2

465, x+

we
1.

find that the H.C.F. of

2 x2

1,

we

dividing both .numerator and denominator by obtain

2 x3

And

13 x2

2x*

5x3

- 6X + 7 _ x + 7 + 8x2-2x + 5~x2 + 3x +
EXERCISE XXIV

5'

Reduce the following fractions

x5y3
jc3y2

to their lowest terms.

_ 4 a;3y5 _ 2 xhj^

(x^
'

y^) (x

'

(x3

y^) (X*

+ y) - y*)
2'

'

x2 x2

- 4 X - 21 + 2x-63' 3 x2 - 18 6x + 27 62 2 x2 - 18 62
(x2-25)(x2-8x +
(x2

3x2

_8x-3
+ 7x +

3x2
5x^
5x2
15)
10)
'

+ 6ax + a^ + 2ax-3a2'

9) (x2

15x2-46x + 35 10x2-29x + 2r

216

A COLLEGE ALGEBRA
(x

(X3
, ,

+ +

2/3)

(Z3

2/3)

X^

11,

(1

x;/)"-^

2/)2

-I

x2

1<2.

2x3 2x3

13.

+ 7x2_7x- 12 + 3x2-14x- 15
4

::

,^

14.

+ 2 Xy 2 "^^ ~ '"y 12 ?ix + 6 ny Gmx 3my 2nx + ny x3_8x2 + 19x-12 2x3- I3x2+ I7x + 12 +
^/i

2^

15.

2x*

x*4-x3 + 5x2 + 4x + + 2x3 + 14x2 + 12x + 12

(x2

lb.
x*
(g
"

x3_2x2-x-6
+ 3x3 + 8x2 + 8x + 8 - 6)3 + (b- c)3 + (c - a)3 - 6) (6 - c) (c - a) (a

c2)2

4 b2x2
52a;2

X*

4 6x3

'

c4

OPERATIONS WITH FRACTIONS

500
Lowest common denominator.

To add

or subtract fractious,

reduce them to equivalent fractions having a common, denominator.

we

first

Evidently the lowest

nators will be the
is

common multiple of the given denomicommon denominator of lowest degree. It

common denominator (L.C.D.)
of

therefore called the lowest

the given fractions.

Example.

Reduce

and

ah

to a lowest

common

denominator.

6c

ca

The L.C.M. of the given denominators is ahc. To reduce a /he to an equivalent fraction having the denominator

abc,

we must

multiply both

Similarly we of c/ah by c.

terms by a. must multiply both terms of h/ca by

its

6,

and both terms

Thus,

r = abc - = -r' -K^-T^' ca ahe ab abc be

a^
b
62
c c2

501

To reduce two or more fractions to eqtiivalent fractions having a lowest common denominator, find the lowest comtnon mnltiple of the given denomijiators. Then in each fraction rej^lace the denominator by this lowest

common
thtis

mxdtiple, and midtijjly the numerator by the introduced in the denominator

new factor

RATIONAL FRACTIONS
Addition and subtraction.

217
502

For fractions which have a common

is

denominator the rule of addition and subtraction

in the formula, 320, 4,

contained

abca-\-b
d

Hence

to find the algebraic

sum

of

two or more

fractiorf^,

If necessary, reduce them to a lowest common denommator. Connect the numerators of the resulting fractions by the signs

which connect the given fractions, and

ivrite the result over the

common denominator.
Filially, simplify the result thus obtained.

of one or

This rule applies when integral expressions take the place more of the fractions for such an expression may
;

be regarded as
It is

fraction whose denominator is 1. best to reduce each of the given fractions to

?i

its

lowest

terms, unless a factor wliich would thus be cancelled occurs in one of the other denominators.

Care should
lowest

bvi

common denominator
is

taken that the expression selected as the A actually is this denominator.

to treat factors like a

frequent mistake
in the lowest

and

a,

differ only in sign, as distinct,

and

to introduce both of

which them

common

denominator.

It is often better to

combine the given fractions by

1-

pairs.

Example

1.

Simplify

~b

a'-

6^

Here the lowest common denominator

1
1

is

a-

h"^,

and we have

2h
a^ b"^

_ a-h
a"^

a-b

b-

a'^

+ -

6
62
a'^

26

62

a-6 + a + 6-26
a2

2a -26
a2

62

62
its

+b

terms,

Observe that the denominator of the sum, when reduced to may be but a factor of the lowest common denominator.

lowest

218

A COLLEGE ALGEBRA
1
J.3

Example
Since the

2.

Simplify x
1

3x4-1
X''

1,

1
is

common
^

denominator is denominator is x^ 1.
first

and the second

(x

1),

the lowest

We

therefore have

~
1

_x

x3-3x + l_x3-x x + 1 x3-3x + l X2 - 1 X2 - 1 X2 - 1 X2- 1 x3-x + x + l-x3 + 3x-l X2 - 1

Simplify

3x
X2

Examples.
Here
it is

^ + ^ ^
1

2
^

^_^

^^^
pairs.

simpler to
1

combine the fractions by

x
2

Thus,
4

+
2

+ 2-(x-2) _ x2 4

X'-

x+1
4

_2-l-(^ + l)_ x2-l x-1

4

x2-l
12^

_ x2-l-(x2-4)_
(x2-l)(x2-4)
2x2

x2_4

x2-l
x2-l

x*-5x2 +
'

+ 3x-2 + Example 4. Simplify ^,^^,_,^ 2x3 + x2 + 3x-2 -1 aid of the remainder theorem, 415, we find that x
By
factor of the

is

a common

two terms

of the first fraction, but

is

not a factor of the second

denominator.

We therefore simplify the first fraction by cancelling x-1

+
;

in both terms, thus obtaining (x By 465, the H.C.F. of x^ x'

+ l)/(x3 + x^ + 2x). + 2x and 2x3 + x2 + 3x - 2 is x2 + x + 2 and x3 + x2 + 2x = (x^ + x + 2)x, 2x3 + x2 + 3x 2 = (2 x 1) (^2 .y ^ + 2). 1 Before reducing to a common denominator, we inquire whether 2 x AVe find that it is, and 3 x - 2. is also a factor of the numerator 2 x2 + fraction to (x + 2) / (x2 + x + 2). cancelling it, reduce the second 2x2 + 3x-2 x2-l - Hence ?r; + x4 + x2-2x 2x3-fx2 + 3x-2

7,

x+1
X (x2

x+2

+ X + 2) X2 + X + 2 2x _ x2 + 3x + l x2 _ ~ ~x+l+ + X3 + X2 + 2 X X3 + X2 + 2 X
603

The product of two or more fractions Multiplication. be found by applying the following theorem.

may

RATIONAL FRACTIONS

219

The product of two or more fractions is the fraction whose numerator is the product of the numerators of the given fractions, and
,
its

denorninator the ^jroduct of their denominators.

a
c

ac

bd bd
is

For the product

Thus,

of each

member by
and
b

ac (see 253).

bd

bd = ac;
its

--bd = ^b--d = ac. d

d
b

252,254

In particular, to multiply a fraction by an integral expression,

multiply

The

fraction

numerator by that expression. ac/hd should always be reduced to

its

its

lowest

terms before the multiplications indicated in and denominator are actually performed.
Example
,. We
,

numerator

1.

Multiply
x3

(x^

1 1

have

x3

X
1

+ -

+ 1) by = (x3 1)(X + (x3 + 1) (X

l)/(x3
-^^

(x
1) 1)

l)/(x
x2

1). 1

= + x2 2)

4- 1

Example

2.

Multiply
(
\
\

(x

2) / (x2

+
x2

x2

by

(x

+
X

2) /x.

We

have

^~ ^ \ ^ x-2\x + x-2/ X x2 X2 + X-2/

From

+ x-2
:

x-1
^"^

Involution.

503 we derive the rule

To raise a fraction to any given power, raise both numerator and denominator to that power.
Thus,

fy=^".
/a\" (-) = \b) ~h'b

For

_,

OCT

ton

factors
6

a-a--6

to
to

factors
factors

a"
6"

Example.

Find the cube of

(

ab'^c^

/efg^.

We have
Division.

To

numerator and denominator.

is

invert a fraction, as a /h, is to interchange its The fraction h I a thus obtained

505

called the reciprocal of a Jh.

220

A COLLEGE ALGEBRA

To divide one fraction by another, rmdtiply the dividend by the reciprocal of the divisor.
a
Thus,
c

a d

ad

Z^l^Z-^JT' be d b c b
of each
c c

For the product

rr,!,

member by c/d
a
b
,,;

is

a/6

(see 253).

Thus,

7^:;X:; = d d b

-and - = adc =
c

ad
be

a
b

bed

252,254,503

multiply

In particular, to divide a fraction by an integral expression, its denominator by that expression.

Example
{x2
1.

Divide

-X2/
x2

+
xy

2/2)

/(x2

X*

2/2)

by

(X*

x2y2
x"^

V*

C UitVC

2/^

+ x2y2 +

y4

+ yi) / {x* xy +
y"^

yi).

x*

y^

RATIONAL FRACTIONS

221

better to divide

In simplifying a complex fraction, A j B, it is sometimes ^ by at the outset by the rule of 505, sometimes better first to multiply both A and B by the L.C.M. Before taking either of of all the denominators in A and B.
these steps
it is

often best to simplify

'a

A and B

h
1

separately.

Examplel.

StapHfy
a

(|5

+
,

,\ l)

+ /(^^ + )
a
b

We

have

+ +

6 b b

a-\-h

\-

+
b

2a
a

a
a

2a

+ a +
b

a
b

_ 2a a b

+
2a

__a
a

+b +b b

Observe that when the terms of a complex fraction are simple fractions we may cancel any factors which are common to the numerators or to
the denominators of these fractions.

Thus, in the third expression above,

we may cancel 2 a.

Example ^

2.

Simplify (

\a-b
;

/ + by'\a +
J

4-

a-b^

We may proceed as in Ex. 1 but a simpler method is to begin by - b). We thus obtain multiplying both terms by (a + b) {a
a a
a
b

- b~
a

+
b

_ "

a{a
a{a

+ -

b) b)

b(a
b{a

b) b)

_ "

a"^
a'^

+
-

ab
ab

ba ba

+ b^ _ + V^

b
3.

Example

Simplify
b

Working from the bottom upwards, we have

a
,

_ ~^
e

a
,

cf

_ a{df-\-e) _ adf + ae ~b{df+e)^cf~bdf+be + cf'

df+e

Complex
fractions.

fractions like that in Ex. 3 are called continued

222

A COLLEGE ALGEBRA
EXERCISE

XXV

Simplify the following expressions.

2a-3&
o 2.

2a + Sb

4a2-962
1

X+1

+-

X2-1

+
X3

+
1

3.

_ ^ x2-3x + 2

1 ^

+-

x2_5a; +

+ -

x2-4x +
3)
(3

4-

(X

5.

1111
1) (X

x+1 ~ - x+2 + rr^

2)

x+3
x) (1

(2

X) (X

- X)

+ g)

b b

a
(a

"^

6)

(a

c)

(6

c) (6

"^

a)

(c

a) (c

"

6)

j/z (x

+
(X
1

zx (y
z)

a) x)
{z

xy{z

a)

(x

7/)

{y

-z){y -

x) (z

y)

3" -I

3-2x
+

8x*-33x 8x3-27

2x + 6 4x2 + 6x +
'

9'

9.

(x
{a

l)%(, + l)%(x, + iy-(x+l)(, +

(b + cY-a? b + c a
{c + af-b^ c a b
-\-

l)(x,

l)

+ 5)3-c3 a+ b c

x2-4
11.

x3-

3x2

-x + 6
.

3x2-14x-5 3x3-2x2 - lOx2x*-3x3-5x2 + 7x-2

14.

12.

x*-4x2-x + 2

...,',

2x* + 3x3-2x2-2x + l

13.

(a4-l)^(a--). \ a/ \ a*/

(i \a3

1+ a^

l)(a* a/

a').

15

16

x2 + x-6 x2-6x + 6 x2 + 7x + 12 x2-4x-5 x2 + 3x-4 x2-8x + 15 1 rx-1 1/x-l (x-2)(x-3) \-|1 [i 2\x + l /J J L X X L x(x + 1)

RATIONAL FRACTIONS
17.

223

19

+ ^\2to^-cx3^ - ex (a + x)2 /a + ~ + \ /a + a3 - 6-V Va - 6 \a - 6

"^
2 6
a>>
O'^

^g_

(^,_^,_,,^22/z).^-^
'

+^

x-y

a"^

o- \

a2

62/

20

111 X2/ z2/a;

'^

1_ +
+

1 _^

'^

a;

]_ +z
_

a ^^
6

&

1_1
a*
6*
1

a
z

6a\a6/
+
^

_b

/I

\2

22.

23.

x-2 X

x+

x-2
INDETERMINATE FORMS

Limits. Suppose the variable x to be taking successively the values 1/2, 3/4, 7/8, 15/16, and so on without end; then evidently x is approaching the value 1, and in such a

508

manner that the

remain
matter
less

difference 1

will ultimately

than every positive number that

small that number

we can

become and assign, no

how

may
, it

be.

We

indicate all this

by saying that as x runs through the never-ending sequence of

values 1/2, 3/4, 7/8, 15/16,

approaches 1 as limit.

x denote a variable which is supposed to be running through some given but never-ending sequence of values, and if there be a number a such that the difference a x will ultimately become and remain numerically less than every positive number that we can assign, tve say that x approaches this number a as limit.
in general, if

And

To
x

indicate that x
or a

is

approaching the limit

a,

we write

= a,

= lim x.

It will

be noticed that the word variable has here a more restricted

meaning than in 242. Whether or not a variable of the kind here under consideration approaches a limit depends on the sequence of values through which

224
it is

A COLLEGE ALGEBRA
supposed to be running.
Thus,
if

the sequence be

1, 2, 1, 2,

the

variable will not approach a limit. full discussion of variables and limits will be found in 187-205, which the student is advised to read, at least in part, in this connection.

509

Theorems respecting
variables x and
ij

limits.

In

203

approach

limits,

it is proved that if the then their sum, difference,

product, and quotient also approach limits, and

lim (x

+ y) = lim x + lim y. y) = lim x lim y. = lim x lim y. lim xy

lim
(cc

..

lim -

= lim X
t:

lim y
it

unless lim y
if

IT

rv

0.

From

these theorems

follows that

given rational function of x, and F(a') its value when x = a, then r(x) will approach F(a) as limit whenever x approaches a as limit, that is,
limF(a-)

F(x) denote any

F(a),

where ^^^ Fix) x= a

Thus,

is

read "limit of F{x), as x approaches a."

'

I(2x2-3x + x=a

l)

= 2a2-3a +

l.

510

Infinity. Evidently if x be made to run through the never ending sequence 1, 2, 3, 4, , it will ultimately become and remain greater than every number that we can assign.

variable x which will thus ultimately become

and remain
is

numerically greater than every number that we can assign said to approach infinity.

For the word

to indicate that
lira

infinity

we employ

the

symbol
a;

oo,

and,

is

approaching

infinity, write

== oo, or

00.

511

It is important to notice that oo, as thus defined, does not denote a definite number, and that the rules for reckoning with numbers do not apply to it. Illustrations of this will be found below.

Note.

KATIONAL FRACTIONS
The phrase "x
"X
is is

225

approaching infinity"

is

merely an abbreviation for

a variable which will ultimately become and remain greater than every number that we can assign." When, as is sometimes convenient, we write lim x = oo, of course the

word

limit

does not have the definite meaning given

it

in 508.
is

Theorem.

Given any fraction whose numerator

a constant

512

and

its

denominator a variable.

as limit, the fraction will If the denominator approach approach oo; and if the denominator approach co, the fraction
will approach

as limit.
1

Thus, consider the fraction

If

/x.
.01,

X approach by running through the sequence of values 1, .1, .001, , then 1/x will run through the sequence 1, 10, 100, 1000,

and

will therefore
if

approach

co.

X approach co by running through the sequence 1, 10, 100, 1000, , then 1/x will run through the sequence 1, .1, .01, .OOl,---,

And

and

will therefore

approach

as limit.

And

so in general.

Indeterminate forms.

A
=

rational

fraction

of

the

form

513

f(x)/<{>(x) has a definite value for any given value of x 0. But when <f>(x) 0, the fracexcept one for which <f>(x)

tion takes one of the forms

0/0

or a/0,

which are arithmet-

It is convenient, nevertheless, to ically meaningless, 103. assign a meaning to the fraction in both of these cases. The form 0/0. The fraction (x^ 1) takes the 1) / (x

514

form 0/0 when x

1.

Now,

excep)t

when x

1,

we can

divide

x"^

by x

1,

and have
This
is

(x^-l)/(x-l) =
true

1.

however

little

may

differ
1,

from

1.

Hence

if,

without actually giving x the value as limit, we shall have

lim
(a;2

we make

it

approach 1

1)

/(^

1)

= lim

(a-

1)

2.

Thus, while the rules of reckoning give us no meaning for 1) / (x 1) when x = 1, they enable us to prove that (x^

226
this fraction

A COLLEGE ALGEBRA
always approaches the definite limit 2 when x

approaches

1 as limit.

Now we have just shown, 509, that F(a) = l^^F(3c) whenever F(x) represents a rational function and F(a) has a meaning. It is therefore convenient when, as here, the rules of reckoning give us no meaning for F(ti), to assirjn to it the value ^^^^F(x)-, in other words, to make the formula

= ^^^^F{x) our definition of F{a). We therefore give {x- 1) / 1) the value 2 when x 1. 1) / {x 1) = x -^ 1 for all values of We then have
F{a)
(a;

{x"^

ic,

the value 1 included.

for a like reason, to every fraction which can be written form (x a)f{x)/(x a)<f>(x), where f(x) and (f>(x) are integral and is not divisible hj x ~ a, we assign the (x') value /()/^(ff) when x = a, and so have

And

in the

<f)

(x (x

a)f(x) f(x) a) (x) ^ (x)

cf> (j>

for all values of x, the value a included.

515

The form a/ 0.
0.

The

fraction 1 /a; takes the

1

form 1/0 when

1 by a value Moreover we

While we cannot divide

of X which differs as
little

by

0,

we can divide
0.

as

we

please from

as limit, 512, that if x be made to approach then 1 /x will approach (X). therefore assign to 1/0, and in general to a/0, when " value " a 0, the co, writing

have shown,

We ^

= "form

And

for a like reason, to every fraction of the

f(x)/(x-a)<f>{x),

where /(a-) and <l>(x) are integral and /(.r) is not divisible by X a, we assign the value " a: when x = a; our meaning
*'

RATIONAL FRACTIONS
being that the fraction will always approach to approach a as limit.
co

227

when x

is

made

Conclusion as to the values of a fraction. From 514, 515 we draw the following conclusions regarding a simple fraction

516

of the form f(x)/<f>(x). 1. If f(x) /<f>(x) is in

its

lowest terms,

it

will vanish for

values of x which
infinite

numerator f (x) vanish, and become for values of x which make its denominator (x) vanish.
its
(f>

make

For
both
2.

all

other finite values of x

it

has a value different from

and ooBut if /(a) and

4>{.^)

have the common factor x

?/i

a,
it

and
71

/(ic) contains this factor

times and

times, then /(a-)/<^(.r) will vanish for x

infinite

<l>(x)

contains

a when

m>

n,

become
both

and

oo

when /n < n, and have when m n.

a value different from

= 2, we have - 2)3 -2 ^ X+ 1 - 2) - 2)2 1 (X (x = U, =: = = = x-2 CO, x(x-2) U, x(x-2)2 CO, x(x-2)2 2 x+ 1 The form oo/oo. It is often important to know what limit
Thus when x
cc
-

(a;

'

517

the value of a fraction /(a;) / <^ (a?) approaches nitely increased, that is, when x == ao.

when x

is indefi-

Consider the following examples.

We have shown, when

,

512, that

1/x,

/x^,

0,

when x

oo.

Hence,

=
.

oo,

X2-X + 3 = 1-1/X + 3/X2 = 2x2 + x-4 2 + 1/X-4/X2 X + 2 1 + 2/x _ x2 + x + 5""x + l + 5/x x2 + x-7_x + l-7/x ~ 2x + 3 2 + 3/x
a;

1/2,

And

in general,

when

(aox^ +

a.x"'-'

= oo, the fraction + a,)/(boX" + b,x^-' +

;

b)

approaches the limit ao/bo, if, as in (1), the degrees of numerator and denominator are the same the limit 0, if, as in (2),

228

A COLLEGE ALGEBRA
;

the degree of the denominator is the greater the limit cc, if, as in (3), the degree of the numerator is the greater. And in each case the limit is called the " value which the
fraction takes

when x

=
oo

oo,"

that

is,

when

the fraction itself

assumes the indeterminate form

518

The forms

co

and

co /cc.

co.

take one of the indeterminate forms

rational function of x mayoo or co oc for some

But the expression can then be reduced particular value of x. to one which will take one of the forms 0/0, a/0, or co/oo already considered.
1.

Thus,
1,

(x2

1)

takes the form

(x^

co

when x

1.

But, except

when X =

we have

I)

=
1

and therefore

lim

r(a;2

1) '

="==11

x-lJ

lim ^_1 = lim ?izii = ^=1^(x +

^=ix-

1) '

2.

Hence we assign
1 2.

to the given expression the value 2,

when x

=
0.

1.

Again,

x(x
0,

+
^

takes the form

2)
^

cc

<x>

when x =
,

But,

except

when x

we have

X
lim r^
='='>Lx
"I

X (X

=
2)

X (x

and therefore

2)

lim

x(x

2)J

=^-0x{x

2)

lim_i_^l. ^-"x + 2 2
1/2 when x

Hence we assign

to the given expression the value

0.

519

General conclusion.
variable, as

(-''),

Therefore, if a given function of a single assumes an indeterminate form when x a,

proceed as follows

Reduce the exjiression to its simplest form, and then find what limit its value apjjvoaches wlien x is made to approach a as limit. Call this limit the value which the function has when X = a.
520
This method is restricted to functions of a single variable, as Note. For the reason that the method yields definite results is this the F{x). value of "j" F (x) depends solely on the value of a and not on the values which X may take in approaching a and the like is not true of functions
:

of

more than one

variable.

RATIONAL FRACTIONS

229

fraction

Thus, suppose that x and y are unrelated variables, and consider the x/y when x = and y = 0. and ?/ The limit, if any, which x/y will approach when x 0,

depends on the sequences of values through which x and y may run. as limit if it runs through For example, a variable will approach either of the following sequences
:

1/2, 1/3, 1/4,

If

(1);

1/2^ 1/3^ 1/42,

(2)

X runs through

the sequence 2, 3,

and y through (1), 1 /4, and approach 0. Therefore, if x and y are unrelated and y lutely indeterminate when x =

will run through (1), and y through (2), then x/y and approach co. But if x runs through (2), 4, then x/y will run through the sequence 1/2, 1/3,

variables,

we regard x/y
so in general.

as abso-

0.

And

Infinity in relation

to

the rules of reckoning.

If

we

take

521

infinite values of tlae letters into account,

we must

state the

rules of 249, 251, 253 as follows

1.
2.

a
If
li

3.

-\-

= 0, ac = he, c = h

unless a

then a
c,

-\-

then

= h, a = h,

unless
unless

c c

= oo. = or
=
cc.

oo.

important to keep these exceptional cases in mind when applying the rules to the solution of equations.
It is

Thus, consider the product

factor,

-x

1.

When

1,

the second
oo, it

1, is

but as the

first factor, 1

/(x^

i),

is

then

does

not follow that the product

is 0.

The product

is 1

/2

in fact, 518.

Infinite roots of equations.

Instead of saying, as

we have

522

been doing, that the equation a- + 2 = a- + 3 and other simple -x = b, have no equations which will reduce to the form
root,

we sometimes say that they have the root cc. For however small a may be, if not actually 0, ax = b has the root b/a. And if, keeping b constant and different
from

512.

its

as limit, b/a will approach oc, In other words, as arc = 6 approaches the form Ox =b, It is therefore quite in root b/a approaches the value oo.
0,

we make a approach

230

A COLLEGE ALGEBRA

when ax
is CO,

agreement with the practice explained in 515 to say that =^ b has the form Ox = b, it has the root oo.

Observe that if we regard x-|-2 = x + 3asa true equation whose root we are not driven to the absurd conclusion that 2 = 3. For since X = CO we have no right to infer that the result of subtracting x from both members is a true equation, 521, 3.

523

Infinite solutions of simultaneous equations.

In like manner,

instead of saying of a system of inconsistent simple equations, 377, 2, 394, 2, that it has no solution, we sometimes say
it has an infinite solution ; for from such a system we can obtain by elimination a single equation of the form Oa; = b, and, by 522, this equation has the root oo.

that

Thus,

we may say
1 (2)

that the pair of inconsistent equations y

x=

(1),

jc

has an infinite solution.

(1), (2) is

Observe that this pair

the limiting case,

asm = \, of the

pair

- mx =

(3),

-x =

l (2).

The

solution of the pair

(3), (2) is

l/(??i

1),

= m/{m

1),

both l/(m 1) and m/ (m 1) approach infinity. and when The same thing may be shown by the graphical method, 386, 387. For, when m=:l, the graph of (3) approaches parallelism with that of

m == 1,

of intersection of the two graphs recedes to an infinite (1), and the point distance in the plane.

EXERCISE XXVI
Assign the appropriate values to the following expressions.

3.

5.

x^-5x + 2. 6^ ^^^^ ^ ^ 2. ^:-f^+f when x = 1. x3-2x + l x2-6x + 8 x2 x2 1 , - 2ax + a2 when x=a. = 4. whenx = l. 2x + l x-' x2-(a + 6)x + ab (3x + l)(x + 2)^ ,whenx = -2. (x2-4)(x2 + 3x + 2) X3 _ x2 - X + 1 =
,

when X

1.

c3-3x2-|-3a;-l

RATIONAL FRACTIONS
3x^-x + 5 x^ + (2x^ + l)(x3-5)^ when 1^ _3^^ - 6) x2 + 1 X + 6X - 7 (X* + 1) (X -X 1 X 2 when x = 3. 8. X (X - 3) x2 _ 9
^
^

231
x

= oo.

2 x2

1 9.

2
1

X
x2
10.

-i

(x

when x

1.

1)

X
X

X
,
,

X2

x-1
X

-2 whenx = 2.
2

,, 11.

x-lx +
l 3x + -

X
r
1

when x
,

= oo.

X2

FRACTIONAL EQUATIONS
On
tion
solving a fractional equation. be transformed into one
its

Any

given fractional equais

524

may

plying both

by multimembers by D, the lowest common denominator

integral

which

of all its fractions.

We call
341,

this process clearing the equation

of fractions.
It

follows from
is

342 that the integral equation

all

which

thus derived will have

the roots of the given

equation, and, if it has any roots besides these, that they and so may readily be must be roots of the equation i

detected and rejected.

Example

1.

3 r, Solve ,

x
1

x(x

+ 13 = ^ 1) 0.

...

(1)

Clear of fractions by multiplying by

D = x (x
(x

1).

We

obtain
(2),

3 (x

1)

6x

Solving

Therefore, since 2 and the only root.

is

not a root of

= 0. x = 2. D = x (x - 1) = +
13)

(2) (3)
0, it is

a root of

(1),

Example

2.

Solve -

1 1)

X (x

xj^ ^ ^
1)

^^^

Clearing of fractions, 3 (x

6x

(x

5)

Solving

(2),

= =

0.
1.

(2)

(3)

232

A COLLEGE ALGEBRA

As 1 is a root of Z) = a;(x - 1) = 0, it is not a root of (1). In fact, when X = 1, the first member of (1) has the form 3 + 6/0-G/O; and by not 0. 518 we find that its value is 8,
Hence
(1)

has no root.

We
525

may sum up

the method thus illustrated in the rule

To
ing

solve a fractional equation, clear of fractions hy multiplythe lowest common denominator of all both members

by D,

the fractions. Solve the resulting integral equation.

The

roots

D
526

vanishes
Note.

are the

of this equation

except

those, if

any, for which

roots of the given equation.

:

We may

also establish this rule as follows

the represent the result of collecting all the terms of = will be the then given equation in one member and adding them fractions. integral equation obtained by clearing of

Let

N/D =

1.

If

N/D is
;

in

its

lowest terms, the roots of

its

the same

for a fraction in

lowest terms vanishes

iV/D = and iV = are when its numerator

iV
77is

vanishes, and then only, 516.

Thus, in '

524, '

Ex.

1,

x
^

13

x-1
2.

x(x-l)

8(x-2) = = ,f

x(x-l)

Here / 1) as the root of iV

is

in its

lowest terms and the root of

iV/D =

the

same

0,

namely,

2.

If

N/I)

is

not in

its loioest

terms,

N=

will

have roots which

N/D =
to

does not have, namely, the roots for which the factor
iV +5 = 8(x-l) = -

common

N and D vanishes.
3

Thus, in

524,

Ex.

2,

x
;

x-1

x(x-l)

x{x-l)

Here
is

N/D

is

not in

its

not a root of

N/D =

lowest terms, and the root of TV = 0, namely = 8, 514, 0; for when x = 1,

1,

N/D
X
;

Evidently

1. is

the general case and 3

2. is

exceptional.

~ Thus, consider the equation

xx-lx(x-l)
this

+ a =
-

U-

except

Here iV/D = [8x-{a + 3)]/x(x-l), and when u = 5 or 3.

is

in

its

lowest terms

RATIONAL FRACTIONS
3.

233

It

must not be inferred from what has

equation will never be satisfied by a root of

'

^=
X
(x

just

been said that the given which is also a root of

D = 0.
X
Thus, consider the equation
2
1

=
1)

0.

Here

N/B = (x
1.

l)2/x(x
ij

when X =

But observe that

N = {x
1) ==

1)

and, by

51G, this

l)'

expression vanishes has the root 1 a greater

number of times than

x (x

has this root.

In applying the rule of 525, care must be taken not to introduce extraneous factors in the expression selected as the
lowest
If

527

common

denominator.

any fraction in the equation is not in its lowest terms, begin by simplifying this fraction, unless the factors thus
cancelled occur in other denominators.

Before clearing of fractions it is sometimes best to combine certain of the fractions, or to reduce certain of them to mixed
expressions.
x2

Example

1.

Solve

6X

x2

11

x'--8x +

15

6x-2x2

Here the terms of the first fraction have the common factor x 5, and those of the second the common factor x. Cancelling these factors, we have 1 11 X X 1 X 11

x-3
Solving,

6-2x

or

x-3
10 x

= = =

2(x-3)

Clearing of fractions,

10

5x

22 x
8.

66.

Example

, 2.

-, Solve

x+lx+6 = x+2,x+5
x+2
1

x+7

x+3
-

x+6

Reducing each fraction

X+2
minus
signs,

Transposing so that the terms in each member

x+2

1111 1111
to a

mixed expression and simplifying,

r

x+7

x+3

x+6
may
be connected by

x+3

x+G

x+7

234

A COLLEGE ALGEBRA
of each
1

Combining the terms

member

separately,
1

_ ~
x2

x2

Clearing of fractions,
Solving,

+ 5x + 6 x^ + 13 x +

42

+ 131 + 42' = x^ + 5 x + 6. = -9/2.

it

The given equation may be solved by clearing stands, but that method is much more laborious. 528

of fractions as

it

solving such a system

tions

Simultaneous fractional equations. The general method of is to clear the several equations of frac-

and then

to find, if possible, the solutions of the result-

except
371.

ing system of integral equations.

those,
if

The

solutions thus found

any, which make denominators in the are the solutions of these equations, given equations vanish

if the equations are of the form described in 379, or they can be reduced to this form, they should be solved by the method explained in that section.

But

if

Example
y

1.

Solve the following pair of equations for x, y.

2/

xy

2x
x

4y

2y'^

xy

Clearing both equations of fractions and simplifying, we obtain

X
Solving

-y + 1 = 0,
x

2,

+ 2y-S = = S.

0.

Therefore, since none of the denominators in the given equations vanish when x = 2 and y = 3, these equations have the solution x = 2,

= 3.
Example
2.

Solve the following system for

x, y, z.

+
xy

5 6

yz
,

=
z

3
1

2 (z

x)

xz

0.

These equations can be reduced to the form,

1

379,

1_5
y
(j

1__2
z

1__1
x
2

Solving for 1 /x, 1/y, 1/z, we find Hence the required solution is x =

l/x=
2,

1/2, 1/2/
z

=
1.

1/3, l/z

= -l.

= 3,

RATIONAL FRACTIONS
EXERCISE XXVn
Solve the following equations for x.
^

235

'

6x 3x +

4a;

2x

7_ ~
5
1

2.

-^+ 3-15x 5x-l

4
1

6 4

3.

4.

,31
2x +
3
'

x-2

x-4

x2-6x +
8

x-5
+
2

2x2-7x-15
2

= 0.
3

5.

(x

+ l)(x-3)
x2

(x-3)(x +

2)

(x

2)(x

^
l)

6.

-1
x2-l
x

+ 4x-5
2x
5

+
x2

'

+ 6x +
5

.0
5

,j

3x+l
x
b{x
^^

5-6x
x + b , r o (X + a)
:

+ 9x-18x2
a
b

+a + + b)

= +rab

^x^ + 9'

l 1

x^-l
x

20.

10.

x2

X
11.

x-o
^^"^
x

x-1 + 2x + l ^^ ^ + 5x + 4 x2 + 3x-4 8 X 9_x + 7 x + 2

x-4~x + 8
+ +
Q
a:

s'

12

+ +
1

6
2

,16.

x3

x-2
X

_ + 10 x + 6 x 8~x + 9 x + 5' x3-2 15 = 4 X. x+ 2 x2-4

^

14

x-2
1

3x2
1

X2

+x^
X*

15.

+ -A_+ 2x22X 4x5 x3-8 +

ox +c
X
6x
1

1_^ x-2
-

16.

+d= q

+
,

6.

236
x2
17.

A COLLEGE ALGEBRA
+ 7x-8
x'^
1

+
X

+
::

S
h

Xax

2
b^

2x2-a;+7_ X + 3
z

4 X.

18.

X
19.

+ 2bx-2ab + a
,

26 +
^

-x"^

3c2

X
-

0.

2c
(X

(x

(^- 6) (x -

"^^

+
c)

(x-6)2
(x

+
a) (X

c)2

c) (X

a) (x

6)

20.

3x+2
X2

x-5
X2

x-2
X2

+
2

X
0.

a
21.

x-2

+6 = x2-4
X

Solve the following for x and y, or for x,

?/,

and

r3x +
22.

2/

-1 a;-2/ + 2 X + 9 _ X + ~ +4 +
y
2/

6
7

-2 x-3
y
2
x2

x-y _ y-4 x2-9~x + 3'

3

23.

3 3
^

= 2 X + xy 2y + xy 0,
9
^

xy x
24.

+y
yz

a.

x
25.
<

+ 2y y + z

+22/ + 2z
5
z

=
7

3,

+
zx
-\-

&,

- 3x
2

_Z

3x

= -1.
2?/

PARTIAL FRACTIONS
529
It follows

from

506 that every rational function

of

single variable, as x, can be reduced to the form of an integral function, or a proper fraction, or the sum of an integral function and a proper fraction.

For certain purposes

further and,

it

is

when a proper

fi'action

useful to carry this reduction A/B is given, find the

simplest set of fractions of which method depends on the following

letters A, B, P, Q,

A/B

is

the

sum.

The

theorems in wliich the

and so

on, denote rational integral func-

tions of X.

RATIONAL FRACTIONS
Theorem
tions
1.

237
530

The sum and

the difference of two proper frac-

A/B

and

C/D

are themselves proper fractions.

,

For
Since

A C = ADBC

B
A
is

BD

of lower degree than B,

AD

is

of lower degree

than BD.

And

since

is

of lower degree than D,

of lower degree than

BC

is

of lower

degree than BD.

Hence

AD BC
2.

is

BD.

Theorem

Let

and

V denote

integral functions,

and Kf^

531

and A'/B' proper fractions.

Ifl +
But I
A'/B'

A/B =

I'

A'/B', then 1
I
I' =

For, by hypothesis,

= 1' ajid A/B = - A / B. A'/B'

0),

A'/B'.

I' denotes an integral function (or A / B denotes a proper fraction (or 0).
= - A/B = A/B =

and,

530,

Therefore, since an integral function cannot be identically equal to a proper fraction, we have

/
or

/'

and A'/B'

0,

/=/'and
Theorem
3.

A'/B'.

denote a jjroper fraction whose denominator has been separated into two factors, P and Q, which are jxrime to one another. This fraction can be reduced to a sum of two proper fractions

Let

A/PQ

532

of the forms,

B/P
is

and C/Q.
prime to P, we can
find, 479,

For, since Q

functions
1

M and N, such that

il/Q

two integral

NP, and therefore A

Hence
If

PQ
A
and

= AMQ + ANP. AM AMQ + ANP = - AN =

,

PQ

(1) ^ ^
is

AM/P

AN/Q

are proper fractions, our theorem

already demonstrated.

238
If

A COLLEGE ALGEBRA

AM I F and AN / Q are not proper

of integral functions

fractions, reduce

them

to

sums

and proper fractions, and

,

let the

results be

AN ^/ + pand
B

C ^A- + -.

(2)

Substituting these expressions for

"we

AM / P and AN/ Q in (1),

^

have

PQ
But
since

^B P
A

C
Q
^

A/PQ, B/P, and C/Q

are proper fractions,

it

follows from (3), by 530, 531, that 7

A'

and

_B

C
q'

PQ^ P
as

was
Note.

to be demonstrated.

533

The

fraction

A/PQ
A =B PQ P

can be reduced to but one such

sum
(1) ^ '

B/P +

C/Q.

For suppose
where B'/P and

= P Q
C

B'

C
,

Then
But
wise

C/Q ?-^^ = ^-^, P Q

also denote proper fractious.

and therefore ^^

^'^

=C'-C.
(7 = 0.

(2) ^ '

(2) is

impossible unless

BB')

B'

=Q
P

and

For other-

(2)

would mean that {B

is

cannot be the case since than P, 481.

is

prime to

exactly divisible by P, and this and B B' is, of lower degree

534

Partial fractions.

We call the fractions B/P and

C/Q, whose

just proved, jxirtia I fractions of A / PQ. To resolve a given fraction of the form A / PQ into its parand C/Q, it is not necessary to carry out tial fractions

existence

we have

B/P

the process indicated in

532

we may apply

1)

the method of

undetermined
Example
fractions.
1.

coefficients, 397, as in the following

example.
two partial

Resolve

(2

x"^

\)

/ (x^

into a

sum

of

This
factors,

is

a proper fraction, and

1

and

x^

1,

its denominator is a product of two which are prime to each other.

RATIONAL FRACTIONS
Hence, 532, (2x2 + l)/(x' tions whose denominators are x

239

sum of two proper fracand x^ + x + 1 respectively. The numerator of the first of these fractions nmst be a constant, that of the second an expression whose degree is one at most. Hence we must have
1) is

equal to a

2 X-

+ 1 Z = a + 6x + c Z x3-l x-lx2 + x +

n) '
^

where

a, b,

and

denote constants.
clear (1) of fractions.

To

find a,

b, c,

We
or

obtain

2 x2

2x2

+ +
+

1
1

= a (x^ + x + 1) + (6x + c) (x - 1), = (a + 6)x2 + (a- 6 + c)x + (a - c).

powers of x are equal,

(2)

As

(2) is

an

identity, the coefficients of like

284.
(3)

Hence
or, solving (3),

?>

a
2x2

= 2, I,

a
6
1

6 + c = 0, = 1, c = 0.
X
h

l,

_ Therefore
.

X3-1
2.

X-1

Example

Resolve (5x

+X+1 + 4)/(x* + x^ +
X2

x2

x) into

sum

of

two

partial fractions.

General theorem regarding partial fractions.

of
1.

From

the theorem

535

532 we Let

may draw

A /PQR

the following conclusions. denote a proper fraction in which the three

factors of the denominator P, Q,

R are prime to one another. This fraction can be reduced to a sum of the form
4

PQR
where

B /P, D / Q, and

PQR E
/R

^B

denote proper fractions.

prime to QR, 482, A /PQR is the sum of two proper fractions of the form B / P -{- C/QR, 532; and since Q is prime to R, C/QR is itself the sum of two proper
For since
is

fractions of the

form

D/Q + E/R,

532.

The like is true when the denominator is the product of any number of factors all prime to one another. 2. Consider the proper fraction A / PQ^ in which P is prime to Q. By 532 it can be resolved into the sum

ABC
P

PQ^

Q"

240

A COLLEGE ALGEBRA
cannot apply the theorem of
since
532 to the fraction C/Q^, Q are not prime to one another. it

We
But

since the factors Q, Q,

is

422, to a polynomial in

of lower degree than Q^, Q of the form

can be reduced,

C
where

CiQ2

c^Q

Cs,

Ci, C^, and C^ are of lower degree than Q. Dividing each member of this identity by Q^, we have

Hence the given

fraction can be reduced to the

sum

A
where

=^+

C\
-T;^

+ TT +

C<,

C,

is

of lower degree than P, and C\, C^, C^ are of lower

Q.

degree than

And

so in general

when

a factor, as Q, occurs more than

once in the denominator. We therefore have the following theorem. Suppose that the denominator of a given proper fraction has

some occurring once, some, it may been separated into factors \ohich are all j^rime to one another. he, more than once The fraction itself can then he resolved into one and hut one

sum of proper fractions

occurs but once, there
is

in which (1) for each factor, P, which

a single fraction of the form B/P, and ivhich occurs r times, there is a group (2) for each factor, Q, + C,/Q^ where of r fractions of the form Ci/Q + Cj/Q^ H are all of lower degree than Q. , C, Ci, Cj,

536
.

Simplest partial fractions.

It can
is

be proved that ever}^ poly-

nomial in x with real

the product of factors of one or both the types x a and x^ + px + q, where a, p, and q are real, but where the factors of x^ + px + q have imagicoefficients

nary

coefficients.
it

Moreover

of a given proper fraction

follows from 469, 532 that, if the numerator and the factors into which its

RATIONAL FRACTIONS

241

denominator has been separated have real coefficients, so will the numerators of the corresponding partial fractions. Hence,

by

534,
'proper fraction
coefficients is

Every
have real

whose
to

nnmerator and denominator

a definite sunt of partial fractions related as follows to the factors x a and x^ q o/* px

equal

its

denomiiiator.
1.

For every factor x For every factor x

fractions of the form

fraction of the
2.

form A/(x

a occurring once there is a single is a real constant. a), ivhere

occurring r times there

is

a group

of

A^/(^

a)

A,/(x

_ a)2 +

+ AJ(x -

a)^

where Aj, Aj, A^ are real constants. 3. For every factor x'^ -|- px + q occiirring once there is a single fraction of the form (Dx + E)/(x^ + px + q), where D

and
4.

are real constants.

x'^

For every factor

group) of r fractions of the

px + q form

occurriiig r times there

is

(Dix -+- E0/(x2 + px where Dj, E^, Dj, E2,

q)

+ (D,x +

E,)/(x2

px

q)^

D^, E^ denote real constants.

The fractions here described are usually called the simplest partial fractions of the given fraction. They are best found by the method of undetermined coefficients.
Example
fractions.
-r,
<.

537

1.

Resolve
x2
(a;

into

its

simplest partial

By

536,

ro^

we have

1) (X

+ x-3 - 2) (x obtain

3)

= A + B x - 2 x - 1

C -

,_
3
(1) ^ '

where A, B, C are constants.

Clearing
x2
(1)

of fractions,

we

+ x-3 = ^(x-2)(x-3) + B(x-3)(x-l) + C(x-l)(x-2).

(2)

A, B, C by arranging the second member of (2) according to powers of x and equating coefficients of like powers but, since A, B, C are constants, the same results will be obtained by the following method, which is simpler.
find
;

We may

242
In
(2)

A COLLEGE ALGEBRA
letx
let
let

next
finally

x
x

= 1, = 2, = 3,
x'^

and we have

-1=A{- l)(-2),
3

and we have and we have

= jB( = C 2

1)

1,

.-.

.-.A^- 1/2 5=8

;

1,

.-.

C=

9/2.
9

Hence

+ x-S
3)

=
X

(X- l)(x-2)(xExample
2.

2(x-l)

x-2

2(x-3)

Resolve

x(x

+ 1 1)3

into its simplest partial fractions

By

636,

we have
x

x(x-l)3
and therefore
In next
(1) let

= - + x-1 + X
I)-'

(X

+
1)-

(x-l)3
1)

= J. (x -

let

= =

0,
1,

and we have and we have

+ J5x (x - 1)- + Cx (x = J. ( 1)3, A = 1 = D, D = 2.

.-. ;

Bx.

(1)

.-.

Substitute these values of

and

in (1),

transpose to the

first

mem-

ber the terms thus found, namely result. We obtain

(x

1)^

and 2x, and simplify the

x3-3x2 + 2x
Dividing both members of
(2)

= x(xby x
(x

1)'^

+ Cx(xwe have
C.

1).

(2)

1),

x-2 = B(xEquating
1

1)+
7J

(3)

coefiicients of like

powers of x
C,
.-.

in (3),

we have

jB

and

=-5+

Hence

x+1 =
x(x-l)3
(1)

11
1

x-1

1,2
\-

and C

=-

1.

(x-l)2
x,

(x-l)3
obtaining

Or we may arrange
X

according to powers of

= (^ + B)x^ -{3A + 2Bcoefficients of like

C)x2

+ {3A + B - C + D)x - A.
we have
l.

Equating

powers of

x,

A + B = 0, SA + 2B-C = 0, 3A + B-C + D=-[, -A =

And from
these equations

we

find, as before,

A=-l, B=l, C = -l, D = 2.

Example
fractions.
3.

Resolve
(a;2

5 x2

4X

4-

16

-a;

into

its

l)2(x-3)

simplest partial

The

factors of x^

being imaginary,

we
-:; a;2

have, 536,

5x2-4x4 16 _ - X + 1)2 (X - 3) (x2

where A, B, C, B,

Ax + B
(x2

_X+

Cx

1)2

-^ - X + 1 + X - 3*

+D

E are constants.

RATIONAL FRACTIOXS
Clearing of fractions,

243

6x2-4x4- 16^{Ax + B){x-S) + (CX + Z>) (X2 - X +

"We

1) (

3)

^(X2

-X+

1)2.

(1)

may

find

A, B, C, D,

by arranging

(1)

X and then equating coeificients of like powers

is

according to po\Yers of but the following method

simpler.

In

(1)

let

3,

and we have 49

= 49 E,

.-.

E=

1.

Substitute this value of

in (1), transpose the

term

(x2

-f 1)2

thus found to the

first

member,

by X

simplify,

and divide both members

3.

"We obtain
(x3

x2

5)

= ^x + 5 +
of (2)

Next divide both members

by

x2

(Cx x

+ +

D)
1.

(x2

1).

(2)

"We obtain
(3) ^ '

By
and
.

2x +
X2

= Ax + B + 1 X2 - X
+
therefore

^ ^ Cx + D.
(3)

531,

the fractional parts and the integral parts in

are equal.

Hence

Cx + D, and

C=
3
1)2

1,

D = 2,
x

2x~S=Ax+

B, and therefore

A = 2,
x2

^, Therefore

5x2-4x + 16 = - X + 1)2 (X - 3) (X2

2x +
(X2

-X+

= 3. +2 -X+1
j5

1-

When
(x

a, 423. Similarly when the denominator has powers of a; the form (a;^ +joa; + qY, the factors of x"^ -\- px + q being imaginary, we express the numerator in powers of x" + px + qExample. ^
fractions.

ay,

the denominator of the given fraction has the form it is best to begin by expressing the numerator in

538

^ Eesolve
,

x*

+ x''_8x2 + 6x-32 mto - 2)5 (x

.

its

.,

^.

simplest partial

By
a;4

423,

we

find
(X

+ a;3 - 8 x2 + 6 X - 32 =

Dividing both members by (x

a;4

- 2)* + 9(x - 2)3 + 22(x - 2)2 + 2)^, we have

+
22
,

I8(x

- 2) - 28.
28
(x

+ x3-8x2 + 6x-32
(x-2)5

x-2

-,

(x-2)2

^ + (x-2)3 + (x-2)* ^
:

18

- 2)^
539

If given an improper fraction, we may first reduce it to the simi of an integral expression and a proper fraction and then resolve the latter into its partial fractions.

244

A COLLEGE ALGEBRA
Apply
this

Example.

method

to the fraction

x3-2x2-6x-21 X- 4 X 5

have We
,

x3-2x2-6x-21 = x + x2_4x-5
,

+ +

7X-11

x2-4x-5
7x (X

=
and proceeding as above we
find

11

+
4

1) (X

5)

(x

7x-ll + 1) (X -

3
5)

+
1

EXERCISE

XXVm

Resolve the following into the simplest partial fractions whose denominators have real coefficients.
2
1.
.r,

4-1

6X

SYMMETRIC FUNCTIONS
IX.

245

SYMMETRIC FUNCTIONS

ABSOLUTE SYMMETRY AND CYCLO-SYMMETRY

Absolute symmetry.

In the expression x^
x"^ -\z"^ y"^ -\-

y^

z^

the

let-

540

ters X, y, z are involved in such a

manner that
is

them be interchanged
or
x"^ -\-

any two of transformed into an idenif

a-^

tically equal expression,

z^
-\-

namely, y"

-f z-, or

z'^ -{y'^

+ x^,
z"^

tf.

To

indicate this,

we say

that x^

y'^

+
is

is

syimnetric with respect to x, y, z. And, in general, a function of a certain set of letters

said

to be symmetric with respect to these letters when every interchange of two of the letters will transform the function into

an identically equal

functioi-.

Other examples of symmetric functions are

+ y) + z) {y + z) with respect to x, y, z, + a) (x + 6) (x + c) with respect to a, 6, c. On the other hand, x + y z is not symmetric for if we interchange and z we obtain x + z y, which is not equal to x + y ~ z.
{xy

+
b

XZ

yz)/{x
(x

(a;

and

We

call 2 x'^y

and 3 y^z terms of the same type with respect

z,
y'^z,

541

to the variables x, y,

namely,

x'^y

and

because the variable parts of these terms, can be transformed into one another by
x, y, z.

interchanges of pairs of the letters

And
a7i

so in general.

The

sufficient

and necessary condition that

z,

integral finic-

542

tion of certain letters, as x, y,

these letters is that all its

symmetric with respect to terms of the same type shall have the
be

same

coefficients.

This implies that if a symmetric function contains one term of a certain type, it must contain all terms of that type that is, all terms that can be derived from the term in question by
;

making every possible interchange

Thus,
if

of the letters.

be symmetric, we must have a = b = c. Again, if a symmetric function of x. y, z contains the term x'^y, it must contain all the terms x'^y -\- y-x -{ x^z + z^x -f yH + z'^y.
ax-

by-

cz^ is to

246
543

A COLLEGE ALGEBRA

This theorem will indicate the general form of a symmetric function of given degree with respect to a given set of letters.
Thus, the general form of a symmetric function of the first degree with respect to x, j/, z, u is a (x + y + z + it) + 6, where a and 6 denote
constants.

first, 1. 3.

and homogeneous functions of the Again, the most general symmetric second, and third degrees with respect to x, y, z are
a (X
a
(x^

+ + z). + 2/3 42/

2.

a (x2
y^-x

^2

22)

2-3)

5 (x2y

x^z

z^x

+ 6 {xy + xz + yz). + y-z + z^y) + cxyz.

544

the

On expressing a symmetric function. The notation 2^ means sum of all terms of the same type as x^ that is, if x, y, z are
;

the letters under consideration, ^x^

^x'y

= xhj + ifx + xH + z'x + rfz

= + if + z^. Similarly + zhj and so on.

x"^
;

selectgiven symmetric function may be represented by and writing the symbol 2 ing from its terms one of each type, before their sum.

Any

Thus,

S (2 X 545

x32/2)

= 2x + 2y + 2z- xV -

V^^^

- ^^^"^ -

^^^'^

V^^^

2^2/^-

writing out symmetric functions at length, it is best The to arrange the terms in accordance with some fixed rule. following examples will indicate a convenient rule for the

When

arrangement of

iyitegral

symmetric functions.
and by the

Suppose that the letters under consideration are a, 6, c, d, normal order of these letters understand the order a, 6, c, d.

We shall then write Sa6 and 2a6c as follows Sa6 ab + ac + ad + be + bd + cd, Xabc = abc
:

+abd + acd +

bed.

Observe that in each term we write the letters in their normal order. In forming 2a6 we take each letter a, b, c in turn and after it write each subsequent letter. The terms of ^.abe are derived in a similar manner

from those of 2a6.

We shall

arrange the terms of

I,a"'b",

when

-^ n,

as follows

Sa263

a263

b'^a^

a^c^

c^-a^

cW + d^c^
;

Observe that we keep the order of the exponents fixed then under the exponents we write the letters of the first term of 2a6 in both the orders ah and ba, and so on.

SYMMETRIC FUNCTIONS
In like manner

247

we may

write

(terms similarly derived from the remaining terms of ^abc).

general theorem regarding symmetry.

It follows

from the

546

symmetry, 540, that a symmetric function will remain symmetric when its form is changed by the rules of In particular, reckoning.
definition of

The sum, difference, product, and quotient of functions are themselves symmetric.

tivo

symmetric

By

aid of this theorem

we may

obtain the result of combin-

ing given symmetric functions by algebraic operations without It is only necessary to actually carrying out these operations.

compute the various typical terms of the

Example
1.

result.

Find (Sa)2

z=

(a

)^

Evidently the required result is a homogeneous symmetric function of the second degree consisting of terms of the two types a^ and 2 ab.

Hence

(Sa)2
2.

Sa^

+
7/2

2 Sa6.

Example
Hence

Find 2x2 ^x
.

(x2

z^)(^x

z).

Evidently this product

is

sum

of terms of the

two types

x^

and

x^y.

2x2 2x

Example

3.

Find

+ 23.2^ -x^ + ys + z^ + x-y + 2/2x + x22; + z^x + (2x)3 = {x + y +zf.

2x3
is

y^z

z'^y.

The required result degree with respect to

(X +
?/

homogeneous, symmetric, and of the third

x, y, z.

We

must therefore have,

22x

543,

+ 2)3 =

(x3 + ?/3 + 23) +

6 (x2?/

+ yH + x22 +
6,
c,

+ y"z + z;^y) + cxyz.

To

find the values of the constants a,

x, y, 2

values to
equations.

which
x
x

will yield equations in a,

assign any three sets of solve these 6, c, and

Thus, putting

= 0, we have 1 = a. = 0, we have 8 = 2 o + = 3a + 1, y -I, 2 = 1, we have 27 Finally, putting x Solving (1), (2), (3), we obtain a = 1, 6 = 3, c = 6. Hence (2x)3 = 2x3 + 3 "Lx^y + 6 'Zxyz.
Again, putting

= = =

1,

2/

1,

2/

= 0, = 1,

2 2

(1)

6.

(2)
-f c.

66

(3)

248
547

A COLLEGE ALGEBRA

ters

In the expression x^y Cyclo-symmetry. y^z z are involved in such a manner that if 0-, y,
2/,

+ ^ic
we

the

let-

replace x

by
x^y

y by

z,

and
y'^z

by

a;,

sion,

namely,
y, z,

z'-x

we obtain an identically equal expresTo indicate this, we say that x^y.

or cyclic, with respect to the
z.

+ y'^z + s^jj
aj,

is cyclo-synimetric,

letters

taken in the order

x, y,
is

And,

in general,

an expression

said to be cyclo-symmetric,

or cyclic, with respect to certain letters arranyed in a given order, if it is transformed into an identically equal expression when we replace the first of these letters by the second, the

second by the third, and so on, and the last by the first. Such an interchange of the letters is called a cyclic interchange.

548

Observe that the terms of x'^y -\- y'h + z-x are themselves arranged eyelidy ; that is, so that the first changes into the second, the second into the third, and the third into the first,

when we

Cyclic expresreplace x hy y, y by z, and z by x. sions are of frequent occurrence and reckoning with them is
greatly facilitated by arranging them cyclicly. Evidently every symmetric function is cyclic, but not every

549

cyclic expression is symmetric.

Its value Thus, x^y + y^z + z^x, though cyclic, is not symmetric. if x and y are interchanged. To make it symmetric we must add the group of terms y^x + zhj + x^z.

changes

550

As the example shows, a cyclic function will ordinarily not contain all the terms of a given type, but such of these terms as it does contain will have the same coefficients.
Theorem.

551

The sum,

difference, product,

and

quotient of two

cyclic functions are themselves cyclic.

This follows at once from the definition of cyclo-symmetry.

Example.
Find the product
is

{x'^y

y'^z

z-x) (x

z).

Evidently the product tains the terms x^y, x^?/'^,

Moreover it concyclic but not symmetric. x'^yz, each once, and terms of these types only.

Hence the product

x^y

is
y'^z'^

y^z

+ z^ + xhj^ +

z^x^

+ x'^yz + yHx +

z^xy.

SYMMETRIC FUNCTIONS
EXERCISE XXIX
1.

249

State the letters with respect to which the expression

X4

2/4

2^

4 (X3

2/3) (?/3

23) (X^

z'^)

is

symmetric.
2.

Write out in
I,a^b^,

full

the following symmetric functions of a,

Sla^/b,
'E.aWc^,
62)c3,
'Za'^b^c^,

b, c.

Sa36*,
b)c,

'L{a
3.

S(a +

S(a

+ 26 +

3c).

Show

that (a
b, c
;

6) (6

c) (c

a) is

cyclic but not symmetric with

c)'^{c

respect to a,
4.

also that (a

6)2(6

a)'^ is

symmetric.

Is
c,

(a

6)2 (6

c)2 (c

d)2 (d

a)^

symmetric with respect to

a,

6,

6.

Arrange the following

y2
(a

sets of expressions cyclicly.

x2,

22

y2^ x^

z^

cC^bc,

obiP,

acH,

b'^cd
c).
tZ

c) (6

a),

(a

c) (c

6),

(a

6) (6

6.

Write out

in full

the cyclic functions of a,

6,

c,

whose

first

terms are
a63c2,
7.

a{b-

c),

(b

2c){a

d),

a^/(a

6) (a

c).

Prove the truth of the following

Sa3 2a
.

identities.

2a*

2a36

2a6 2a

2a-'6

3 2a6c.

FACTORIZATION OF SYMMETRIC AND CYCLIC FUNCTIONS

By

aid of the remainder theorem and the principles just

552

explained it is often possible to factor a complicated symmetric or cyclic function with comparatively little reckoning.

Example

1.

Factor x^ (y

z)

-\-

y^ (z
z,

x)

+
-

z^ (x

y).

This function vanishes

X3 (2

when y =

for
23 (x
z)

is

2)

Z3

(z-x) +
2

= 0.
;

Hence the function

reason
it is

exactly divisible

exactly divisible

by

by y 2, 416 and for a like x and hy x y, and therefore by the

product (y x) (x y). 2) (2 Both dividend and divisor are cyclic and homogeneous, and their Hence the quotient must be degrees are four and three respectively.

250

A COLLEGE ALGEBRA

the form k{x

a;3(t/

a cyclic and homogeneous function of the first degree and therefore have Hence -\- y + z), where k denotes some constant.

-z) + y^(z-x) + z^{x-y)~k{y - z) {z - x)yx ~y,{z + y ^ z) (1) To find k, assign to x, y, z any set of values for which the coefficient
Thus, putting x

of k will not vanish.

2,

1,

0,

we have

&

Qk, ov k

\.

Or we may find k by equating the coefficients of like powers of x in the two members of (1) arranged as polynomials in x. Thus, the x^ term in the first member is x^ (y z), and in the second member it is kx^{y z), whence as before k = \. We therefore have
x^iy

-z) + y^{z-x) + z^{x-y) =

2.

{y

z) {z

x){x

y){x

z).

Example

Factor

(x

z)^

This function vanishes

when x

x^ ys = y, for

z^.

y + zf + xj^ -y^ -z^ = 0. Hence the function is exactly divisible by x + ?/ and for a like reason it is divisible by ?/ + z and by z + x, and therefore by (x + y) (y -^ z) (z + x). As the dividend and divisor are symmetric and homogeneous and of the fifth and third degrees respectively, the quotient must be of the form

{-y +

ft(x2

2/2

z'^)

+
{y

i(X2/

+
{x

2/z

+
y

zx), 543.
-\-

Hence

= ( +
Putting
Putting

2/)

+
1,

z) (z
2/

+ +
1,

z)^

x^

y^

z^

X) \k (x2

= x =

2,

2/

Solving for k and

(x

I,

= 0, 1, z = 0, we obtain Zb = bk + 21. we find = 5, = 5, and therefore have

z
A;
Z.

= =

+ y^ + z2) -\-l{xy + yz + we obtain 15 = 2 +

i

zx)].

A;

+ y + )5 - 5 - - z5 = 5 (x + + z) (z + x) (x2 +
?/5
2/) (2/

2/2

22 ^. a;y

2/z

zx).

Example
(x

3.
2/

Factor

z)3

(2/

This function vanishes when x

- (2/ 0. (2/ z)3 {y + zf Hence the function is exactly divisible by x or x and for a like reason it is divisible by y and by z and therefore by xyz. Since both dividend and divisor are of the third degree, the quotient is some constant, k. Hence
;

- (z + x = 0, for + zf - (z - 2/)3 z

x)3

2/)^

(x

2/

- zf.

(X

+ +

2/

Putting X
(X
2/

+ 2)3 - + z - x)3 - (z + X = 1, = 1, z = li '^e find k = + z)3 - + z - x)3 - (z + X (2/

2/

y)3
24,
2/)8

(X

+ +

2/

z)3

= kxyz.
=

and therefore have

(X

(2/

z)3

24x2/2.

SYMMETRIC FUNCTIONS
The method just explained is sometimes useful
cyclic fractional expressions.
a^
h^
c3

251
in simplifying

553

Example.

Simplify
(a

6) (a

c)

(6

c) (6
c.

a)

(c

-a){c- h)

This expression is cyclic with respect to a, 6, The lowest common denominator is 'h c) (c

a) {a

h).

On

reducing the fractions to this lowest

common denominator we

first numerator a^(6 the second and third numerators are

obtain as the

Hence, by cyclo-symmetry, c). b). h^{c a) and c'^{a Adding these three numerators and factoring the result, 552, Ex. 1,
(a

we have

&

c) (6

c) (c

a) (a

b).

Hence the given expression reduces

to a

c.

EXERCISE
Factor the following expressions.
1.

XXX

2. 3.

- y). x) + (x - x) + xy (x - ?/). 7/z {y z) + zx (z - zf + (2 - xY + (X - y)\ (y

x^{y
z)

y- (z

z'^

4.
6.

x{y

6.

7.
8.

xY + z{x y)^. 2)3 + y{z zY + 2/^ (z x)3 + z^ (x yY. X* (2/2 - 22) + yi (22 _ a;2) + 2" {x2 - 2/2). + z)-^ x^ 2/3 2^. (x +

x^ (y

2/

(2/

2)6
2/

(2

x)5
(2/

+
2

(x

yf.

9.

(x
(2/

+
-

+
(2/

2)5

+
(z

x)6

(z

2/)^

(x

2/

2)^-

10.
11. 12.

X (2/
x5
(2/

+ (X - y) (X + + 2)2 + (z + x)2 + z (x + 2/)2 - 4 X2/Z. - z) + 2/5 (z - X) + 2^ (x - y).

2)

2)3

X) (Z

X)3

2/)^.

2/

Simplify the following fractional expressions.

13.
(a

-b)(a

5^

+
c)

'^ (6

c) (6

+
a)
(c

'*

a) (c

6)

14.
(a

x+a -
6) (a

+
c)

x+6
(6

+
a)
(c

x+c
a) (c

c) (6

6)

252
, _

A COLLEGE ALGEBRA
a'^
:

16.

be
-

(a 16.

(a
17.

-h)(a(& + c)2 - 6) (a -

c)

V^ ca +- - (b c){b
(c
^

+
a)
[c

c2

ah

a){c

~ -

b)

a)^

(a
^

6)2
6) c2
(c

c)

(6

c) (6

a)
62

(c

a) (c

(a

a2

b){a

c) {X
:

+
a)
{b

c) {b

-; -

a){x

b)

7-

a){c~

b) (x

c)

X.
554

THE BINOMIAL THEOREM

To
obtain the product
k)

Structure of continued products.

(a

+b^c+

d) (e

+f -^ g) (h ^
+b+c+

d by each term of e+f+g, then multiply every product thus obtained by each term of A + ^, and finally add the results of these last
of a

we may multiply each term

multiplications. Hence we shall obtain one term of the product if we select one term from each of the three given factors and multiply And we shall obtain all the terms of these terms together.

the product if we make this selection of terms from the three given factors in all possible ways. Thus, selecting b from the first factor, g f-om the second, and k from the third, we have the term bgk of the product
;

and so

on.

term from a-\-b-{-c-\-d in four ways, g in three ways, and a term from h -\- k in two ways, the number of terms in the complete product is 4 3 2, or 24. And so, in general.
Since
select a
-\-

we can

a term from

e -\-f

The product of any number of i^olynomials is the sum of all the prodiicts that can be obtained by selecting one term from each factor and multiplying these terms together. and
terms, the second n, the third p, if the first factor has number of terms in the comjylete product before is mnp , like terms, if any, have been collected
so on, the

And

THE BINOMIAL THEOREM

253
555

multiplication.

This theorem supplies a useful check on the correctness of a It may be applied to a product in which like

terms have been collected, provided its terms represent sums of terms of like sign and without numerical coefficients, the
coefficient of a

term then indicating how many uncollected

terms

it

represents.

Thus, by our theorem, the product {a

3

c) {a

c)

should have

3 or 9 terms,

which are
b

all
-\-

of the
b'^

same

sign.

{a

c) (a

c)

a'^

+
1

c'^

+
1

2 ab

sents an uncollected product of should.

+ +2+

But, as we have shown, 2 ac + 2bc, and this repre2

2 or 9 terms, as

it

Similarly, the product (a + b) (a + b) (a + b) should have 2 2 2 or 8 terms. But this product when simplified is a^ + 3 a-b + Sab- + 6", which means, as it should, an uncollected product of 1 + 3 + 3 + 1 or 8 terms.

One should bear

this

theorem in mind when reckoning with

last chapter.

556

symmetric functions by the methods of the

Thus, the student has proved, p. 249, Ex. 7, 2a6 2a = 2a26 + 3 l,abc. To test this formula, suppose that only the letters a, b, c are involved. Then 2a6 has 3 terms, 2a has 3 terms, 2a2& has 6 terms, and 2a6c has term and 3-3 = 6 + 3-l, asit should.
;

Products of binomial factors of the

first degree.

The theorem

557

554 enables one factors of the form x

of

to obtain the
-}-

product of any number of

hy inspection.

Thus,

(x

h,) (x

+ i) {x + b^) = x^ + (bi + + ^3)

Z'a

x"^

+ (^ A + *i^3 + M3) X + hKK

we have the term
x^.

For, selecting x from each factor,

Selecting in all possible ways x's from two of the factors and a b from the third, we have the terms b^x'^, b^x"^, b^x^.
Selecting in all possible ways x from one of the factors and from the other two, we have the terms ^1^2^? b-fi^x, b^b^x. Selecting b's from all three factors, we have the term b^b^b^.

&'s

Observe that when the terms of the product are collected,

as in the formula, the coefficient of
letters
^1, b^, b^,
a-^ is

the stim of the three

the coefficient of x

is

the

sum of the products of

254

A COLLEGE ALGEBRA
is

every two of these letters, and the final term

the product

of all three. Hence these coefficients are symmetric functions of

as

b^, b^, b^,

was to be expected since symmetric with respect to bi,

(x -f bi) (x
62, b^.

^2) (^

^3) is itself

Observe also that since (x + ^1) (x + Sj) (^ + ^3) is symmetric with respect to bi, bo, b^, we may obtain the product by finding one term of each type, as x^, b^x-, b^b^x, bxbj)^, and
then writing out
all

558
{x

By

the same reasoning

the terms of these several types. we may prove the general formula

+ *i) {x + ^2) {x + b^)---(x + b,) = a;+ ix"-^ + 5oX"-2 H B, = + ^2 + where 5i = S^i b^, + + bj}^ + + b_ib^, Bi = -Zb^bi = b^b^ + = '$bJ)J>3 = bibj)^ + bib^bi ^ h ^_2^-i^> B3
\-

^-^

^3 -|

\-

b,b.,

that

is,

Bi

is

^1) ^2>

^n) a^i^d

the sum, and B is the product of all the letters the intermediate coefficients are B.2, the sum
:

of the products of every tivo of these letters; 3, the the products of every three ; and so on.

sum

of

Thus, we obtain one term of the product each time that

we

select i's

from

tJiree

of the factors

and

a;'s

from the

rest.

Making the
bib2f)sx"''^,

selection in all possible ways, we obtain the terms and their sum is B.^x"~^. bibibiX"'^, -,

Observe that, as indicated above, the coefficients

Bi, B2,

B
bi, b^, , b^.

are symmetric fimctions of the letters

559
(x

In like manner,

we have

5i) {x

b^ {x-bs)---(x- b) = - BiX"-'^ + 52.T"-2

a;"

\-

{-

lYE,,,

where

Bi, B^,

-B

have the same meanings as in

558. and

THE BINOMIAL THEOREM

255

and +, the the signs connecting the terms are alteiuately when n is even and last sign, that of ( 1)"5, being when 11 is odd.

We
all

the letters

obtain this formula by merely changing the signs of For this h-^, b^, b in the formula of 558.

leaves unchanged every B whose terms are products of an even number of b's, and merely changes the sign of every B whose terms are products of an odd number of b's.

Example.
1.

By

the

method

of 557-559 find the following products.

2.

3.

+ l)(x + 2)(x + 3). {x + a){x + b){x + c){x + d).

(x

(x+2)(x-3)(x +
{x

4).

4.

y){x

+ 2y){x

Sy){x

4y).

terms in the sums 2bi, Sbibo, . Let i, Wr,, denote the number of terms in 2/>i, "^b^bz, respectively. 1. Since 2&i + Kf evidently ni = n. ^i /*2

The number

of

560

of the other n

^i, b^j , ^ ^J each obtain n (n 1) products all told. But these n (n 1) products are the terms of "^bji^, each counted twice. Hence n^, the number of terms in S^i^g*

2.

If

we multiply each

of the n letters

1 letters in turn,

we

is

n(n

l)/2,

or

n
Thus, we have
&1&2,

n(n

1)

hh,

'

hb,,

b^bi, bzbs,

b2bn

There are n groups of these products, and n group, hence n{n l) products all told.

;', Mi,

&n&2>

Mn-l-

products in each
in the

But the product bibo here occurs twice, namely once and once in the form 62&1 and so on.
;

form

bib^

3. Again, if we multiply each of the ??2 terms of ^bjt^ by each of the n 2 letters which do not occur in that term,

we

obtain ng (n But these n^ {n 2) 2) products all told. products are the terms of '^b^bj)^, each counted three times. 2)/ 3, or Hence n^, the number of terms in '^b^b^bg, is 112 (n

^^-2

_ n( 7i-

l)(n-2)

256
Thus, we have

A COLLEGE ALGEBRA

hhh, iihh,

hb2b',

hhh, hhh,

, hh^n

bn-lbnbu bn-lbnb2,

, 6,-lMn-2.

There are no groups of these products, and n 2 products in each 2) products all told. group, hence n2 {n But the product 6162&3 here occurs three times, namely in three forms, And similarly every term of 2616263 here 6162 &3) ^1^3 ^2) ^2^3 ^1occurs three times, once for each of the three ways in which a product of three letters may be obtained by multiplying the product of two of

the letters by the remaining

4.

letter.

By

the same reasoning

n,

we can show
1)
(71

that

n,^^=
i

3 _n(n

2) (n

3)
'

1.2.3-4
to r factors
.

and, in general, that

n(n =

l)(n 2)
^^

2 3

r
letters 61, 62, 63, 64,

Thus, the numbers of products of four

two, three, four at a time, are
Til

taken one,

=4,

712

4-3 1-2

0,

713

3-2. =
+

1-2 -3

4,

714

= 4-3.2-17 =
1.2.3.4
558, namely,
---

1.

561

Binomial theorem.
(x + b,) (x +
b,)--- (x

If in the formula of

J)

.x

B.x'^-'

+ B^x'^-'- + +

+ B,

we

b^ by the same replace all the n different letters bi, b^, letter b, and x by a, the first member becomes (a b)". Again, since each of the n terms of B^ becomes b, and each

of the

Tij

terms of B^ becomes
''

b'^,

and so
!L_^

on,

we
*>

have,

560,

B no, B -"1 -nb n^ -VL^-^b^ B ^

,)
)

"

^^ ^^

-^

-tig

o ^3 , ....

Our formula therefore reduces

(a

to the following
''

+ by = a" +

'^
-L

a"-'b

^''
J.

a"-'b'

^ "^
where

iris's

"^

'

THE BINOMIAL THEOREM

1.
2.

257

The number of terms on the right is n + ! The exponents of a. decrease by one and those ofh increase by one from term to term, their sum in each term being n. 3. The first coefficient is 1, the second is n, and the rest of them may be found by the folloiving rxde :
Multiply the
the term
coefficient

of any term by the exponent of a in

the exponent of b increased by 1 ; the result will be the coefficient of the next term.

and divide by
is

This formula
theorem.

known

sion on the right being called the expansion of (a

as the binomial theorem, the expresby by this

Thus, '

{a ^

^ +

6)3 I

a3

^ +

=
Since a
-\-

a3

a62 + ^JL^ji 53 + ^1.2 ^1.2-3 + 3 aV) + 3 a62 + b^.

3 a^b

symmetric with respect to a and b, it follows from 542 that the terms of the same type in the expansion of (a + b)" namely those involving a" and b", a"~^b and ab"~^, must have the same coefficients. But these are and so on the first and last terms, the second and next to last terms, and in general every two terms which are equally removed from the beginning and the end of the expansion.
b \b

562

Hence the
on.

last

term

is

Z",

the next to last

is

Since the number of terms

1,

is nab"~^, and so there will be one

When there ?i is even, two when n is odd. two middle terms, they are of the same type and have the same coefficients. And by what has just been said, the coefficients of the terms which follow the middle term or terms are but in reverse order. the same as those xvhich precede them
middle term when
are
It

may

also readily be

shown that the

coefficients increase

563

up

to the

middle term or terms and then decrease, so that

the middle coefficient or coefficients are the greatest.

This follows from the rule of coefficients, 561, 3, since in the terms which precede the middle term or terms the exponent of a is greater than the exponent of b increased by 1, while in the terms which follow
it is less.

258
564

A COLLEGE ALGEBRA
the sign of b in the preceding formula and obtain
^ ^

By changing
simplifying,
(a

we

by

= a" - na^-^b +

a"-^^*

1T2T3
(2

*"

''

where the terms which involve odd powers of b have and those which involve even powers have + signs.
Example.

signs^

Find the expansion of

y^)^.

Substituting 2 x for a, and y^ for b, in the formula, and remembering that the last three coeflBcients are the same as the first three in reverse
order, 562,
(2 X

we have
(2 x)^

2/3)6

^ =

- 6 (2 x)52/3 +
6 (2
X)52/3

(2x)* (y^)^

^^^

(2 x)3 (y3)3

+ 15 (2 X)* {?/3)2 20 (2 X)3 (2/3)3 + 15 (2 X)2 (7/3)4 6 (2 X) (2/3)5 + (y3)6 = 64x6 - 192x52/3 + 240xV - 160x32/9 + 60x22/^^ - I2xy^^ + 2/I8.
(2 X)6

565

The general term. From 561 it follows that the term in the expansion of (a + b)" is

(r

l)th

n(n -^

l)(7i

2)

1.2.3...r

to r factors " ^

^-

This, with a minus sign before it when r of (a b)". l)th term in the expansion (r

is

odd,

is

also the

Example
Here n

1.

Find the eighth term

in the
7.

expansion of (x

yY^.
is

16 and r

8,

or r

Hence the required term

_ 16.15.14.13-12-11.10
1.2.3-4-5-6.7
Example
x2o ?
2.

^_

^^^

Does any term

term.

in the

expansion of

(x3

l/x)i2 contain
,

If so, find this

Let r

+
1

and

denote the number of the term. /x, vre must have

1

Then, since n

12,

x^,

^-r5r

12(a;3)

r/xr

x36- 4

-^

X20.

THE BINOMIAL THEOREM

This condition
is

259

satisfied

if

36
9

x^o

4r

20, or r

4.

Hence the fifth term contains

12 11
'-

x^",

and substituting

in the

formula we

10

find this term to be

= 495 x^o.

1-2. 3-4

EXERCISE XXXI
Expand
1.

the following by aid of the binomial theorem.

(3x
(2

2/)3.

2.
5.

{a

bf.

3.
6.

(1

2x2)7.

4.

1/x)*.
(1

(x-3/x)6.
8.

(x/y-y/x)^.

7. 9.

2 x2)4.
(1

(a2

+ ax- x^f.

Find the sixth term in

x/2)".

10. 11. 12. 13.

Find' the eighth term in (3 a

6) 12.

Find the middle term in

{a^

2 bcyo.

Find the two middle terms

in (1

x)^.

Find the
Find the

coefiicient of x^ in (1 coefficient of x* in (3 coefficient of x* in (1 coefficient of x^ in (1

14. 15.
16.

+ x)8, 2 x)^. +
x
x^)^.

Find the Find the

2 x)9

(1

- 2x)".

17. 18.
19.

Find the constant term Find the

coefficient of

in (x

l/x)i2.

x''-

in (2

/x)

i^.

Find
Find

{x

+
+
b

2y){x
2) (x

3y)(x

5y) by inspection.

20.
21.

(x
is

3) (x

4) (x

5)

by

inspection.

What
{a

the

number
c

of terms in the product

+ d){f+g +

h){k

l)

{m

+ n+p +

q)?

22.
1.

Find the sum of the

(i

coefficients in the following products.

2.

+ X2 + X3 + X4)3.
is

(1

+ X2)2 (1 + X +
?
3.

3x3)2,

23.

"What

the

sum

of the coefficients in the following

b, c,

symmetric

functions of four letters a.

1,

d when expanded
Sn*

Sa2 Sa.

2.

Sa6c.

Sa6 2a6c.

24.
25.

Show that the sum of the coefficients in the expansion of (a + 6)" is 2".
Show
that in the expansion of (a

coefficients is numerically equal to the

sum

6)" the sum of thg positive of the negative coefficients.

260

A COLLEGE ALGEBKA
XI.

EVOLUTION
It is possible

566

Perfect powers.

Given a rational function P.

a perfect nth ijower ; in other words, that a second If so, this rational rational function (2 exists such that P Q".
that
is

function Q will be an
function

-nth root of P.
:

A rational In the present chapter we consider the problem P being given, it is required to determine whether or not P is a perfect ?ith power, and, if it is, to find its wth root Q.

We
567 an

suppose n to denote a given positive integer. Roots of monomials. Let P denote a rational monomial
its

reduced to

Tith root of

If P is a perfect wth power, simplest form. P may be obtained by the following rule.
literal factors

Divide the exponents of the several

a7id multiply the result by the principal nth root cal coefficient ofV.

of P by n, of the numeri-

This follows at once from the rule for involution,

318.

Thus, (aW/C")"
a^b'^/c"'"

a^"6''' /<;"'",

of a'"'V''/c'"", 566,

and

it

is

318. Hence a^b'/C" is an nth root obtained by dividing the exponents of

by

n.

568

The

root thus obtained

is

called the principal nth root of

(compare 258). We shall mean this root wh^n we speak of tJie nth root of P, or when we use the P. symbol

Example

1.

Find the cube root of

3l

8 a^b^ /27 x^y^.

We have
Example
1.

8a36s__ 2ab^

3xy^

2.

Find the following

a*Jfi

roots.

64
-x

2.

v81xV2^'^-

V o,

,.

3.

6/32x^2/30

9
is

569

Roots of polynomials.

Consider the following examples.

Example

1.

Determine wliether or not 4x*

if it is,

ix^

I3x-

6x +

a perfect square, and,

find its square root.

EVOLUTION

261

K this is a perfect
form 2 x^
4a;*
-\-

square, evidently

it

px +

q,

where p and q are constants.

must have a square root of the We must therefore have

-4x3+ 13x--6x + 9 = (2x^ + px + qf

= 4x* + ijjx^ +
which requires,
284,

{p-

4 3)x2
:

+ 2pqx +
32

(f,

that

p and
Vi

q satisfy the equations

(2),
1,

4p = -4 (1), From (1) and

satisfy (3)

p^
(2)
;

+ Aq =
find

2p? =
g

-6

(3),

(4).

we

and (4) for 2 (- 1) 3 = - 6, and 3^ = 9. Hence 4 x* - 4 x-^ + 13 x- - 6 x + 9 is a perfect square and

scjuare root.
2.

p=-

3,

and these values of p and q

2 x^

i^ its

Example
If this

Find the cube root of

x6

6x5

21x*
it

44x3

63x2 f 54x

27.

We

be a perfect cube, must therefore have

will

have a cube root of the form x2+j9x + 5'.

x6

6x5

21x*

= x6 +
which requires,

+ 63x2 + 54x + 27 = (x2 + px + + 3(p2 + q)xi + (p3 + 6p3)x3 + 3 (i)2g + q1) x2 + ^pq2y^ + ^3^
+
44x3

g)*

3px5

284, that p and q satisfy the six equations 3p = 6 (1), g3^27 3(p2 + g) = 21 (2),... From (1) and (2) we obtain p = 2, q = ^. And these values
:

(6).

of

p and

will

be found to satisfy the remaining equations (3)(6). Hence x^ + Gx^ + + 54x + 27 is a perfect cube, and

its

cube root

is

x2

+ 2x +
the

3.

By

method

illustrated in these

examples

it is

always

possible to determine whether or not a given polynomial in x is a perfect nth. power, and, if it is, to find its wth root.

Let the polynomial be

a^x""

aiS-"'"^

,.

If this be

a perfect ??th power, its degree m, must be a multiple of n so = kn, where k is an integer and it miist have an th that root of the form ax'' + .4^., where (x denotes the AiX^~'^ +

, A,, are principal wth root of ao, and A^, call this root the principal nth root.

unknown

constants.

We

To determine whether
and
aox'"

a^x""

a,

has any such root,

to find this root if it exists,

we

set

aia:""i

...

+ =

(ax*

+ ^iX^- +

A^Y.

262
/

A COLLEGE ALGEBRA

reduce the second member to the form of a polynomial in x, and then equate its coefficients to those of the like powers of X in the first member. We thus obtain a system of nk equations in ^1, A^, , A^. The first k of these equations will A and this set of give a single set of values for .4 1, A^, values must satisfy the rest of the equations if aQX'" a^

^.

is

to be a perfect nth power.

Example

3.

Find the cube root of

8x6

12a;5

I8x*

13x3

9x2

_ 3^ +

i.

570

Square roots of polynomials.

square,
its

If a polynomial

is

a perfect

square root

may

also be obtained

by the following

method.

As

in the preceding section, let

denote a polynomial in x

= (a-\-b-{-c + -- -y. powers of x, so that P The problem is, knowing P, to find a, b, c, . Now, whatever the values of a, b, c,- may be, we have + bf = 0^ + 2 ab + = + {2 a + b)b, + cy=(a + by + 2 (a + b)c + c" {a + = a^ +(2 a + ^*)& + [2(a + -f c]c, c ia-\-h -^ dy = a" + i2a -^b)h -^\_2{a + b)+ cy + [2(a + i + c)+fZ]c?,
{a
b''

of even degree and arranged in descending powers of x. Let us suppose that P is a perfect square and that a, b,c, denote the terms of its square root arranged in descending

a"-

^-)

-\-

each

and so on, a new group of terms being added on the right with new letter on the left, namely, a group formed by adding the new letter to twice the sum of the old letters and multijilying
the result by the new letter. Therefore, since by hypothesis

P=

(a -f i

+c+

)'^,

we

have

P=

a"

+ (2 a +

b)b -\-[_2{a

^[2(a

+ b) + cy + b + c) + d-]d + .--,

EVOLUTION

263

where the leading terms of the several groups on the righi, namely, a^, 2 ah, 2 ac, 2 ad, -, are all of higher degree in x than any of the terms which follow them.

From
1.

this identity

we may

find a,

b, c,

as follows

Evidently a is the square root of the leading term of P. 2. Subtract a^ from P. As the leading term of the remainder, Ri, must equal 2 ab, we may find b by dividing this term by 2 a.

Having found b, form (2 a + b)b and subtract it from R^. the leading term of the remainder, R^, must equal 2 ac, we may find a by dividing this term by 2 a.
3.

As

4.
is

Continue thus until a remainder of lower degree than a

is 0,

reached.
If this final remainder

is,

as

was supposed, a perfect

square and

square root is a + i If this final remainder is not 0,

its

+c+
P
is

not a perfect square

but we shall have reduced

to the

form

P=(a + b + c +
that
is,

..-y+R,

which
It is

is

and an integral function of lower degree than a. convenient to arrange the reckoning just described as in
of a perfect square

to the

sum

the following example.

Example.

Find the square root of 4 x*

4 z^

13

a;2

6x

9.

P = 4x*-4x3.|,i3a;2-6x + 9 |2x'^-x + 3 = a + & + c

2a +
2(a
6

&)

a2 = 4 x* = 4x2-x - 4 x3 + -4x^+ + c = 4x2-2x + 3

13 x2
x2

6X 6X

+ +

12 x2

9
9

12x2

-6x +
is

= El = P - a2 = {2a + b)b = Ba = P - (a + 6)2 = [2 (a + 6) + c] c = i? = p _ (a + 6 + c)2

its

Since the final remainder

root
is

is 0,

2x2

3.

Compare

569,

a perfect square and Ex. 1.

square

divide
root.

Observe that as each new remainder Pj, R^, is found we its leading term by 2 a and so get the next term of the

Then

at the left of the

remainder we write twice the

part of the root previously obtained plus the

new term

of

264
the root.

A COLLEGE ALGEBRA

We

multiply this

sum by

the

new term

of the root,

subtract the result from the remainder under consideration, and thus obtain the next remainder.
Example.

Find the square root of 25 x*

is

40x3

46 x-

- 24x +

9.

571

This method

more
first

applicable to a polynomial P which involves than one letter, provided it be a perfect square.

We

powers of one of the letters, with coefficients involving the rest, and then proceed as in 570, it being understood that x now denotes the letter of
arrange
in descending

arrangement.

572

Approximate square

roots.

We may

also apply this

method

to a polynomial in x arranged in ascending powers of this will then be arranged in ascending letter. But a, b, c,

powers of

and the degrees of the successive remainders will increase. Hence, 570, 4, if P is not a perfect square but has a constant term, we can reduce it to the form
x,

P = (a^b + c +
that
is,

..-y

R<,

of a perfect square and a polynomial, R', whose lowest term is of as high a degree as toe jilease. For small values of x we can make the value of i^' as small
to the

sum

Hence carrying this reckoning far enough. we call a -\-h, a -{- h -\- c, the approximate square roots of P to two terms, three terms, and so on. It should be added that these approximate roots are found more readily by the method of 569.
as in this case

we choose by

Example L

Find the square root of

write

x to four terms.

By

569,

we

Squaring,

Hence,
or solving,

284,

Therefore the

Vl + x =\ \-px + qx- + rx' + 1 +x=l +2px + (p^ + 2q)x^ + 2{i:>q + 2p=l, p^ + 2q~0, pq + r = 0, r=l/16. p = l/2, q = l/8, x-/8 + x^/W. required result is 1 + x/2

r)x^

Let the student verify this by the method of 570, 57 L

Example

2.

Find the square root of 4

x'^

to three terms.

EVOLUTION
Square roots of numbers.
also derive the ordinary

265

From
method

the formulas in

570 we

573

of finding the square root

of a ymmher.

Example.

Find the square root

of 53361.

Let a denote the greatest integer with but one significant figure, whose Its significant figure will be the leading square is contained in 533G1, We find a as figure of the root and its remaining figures wiil be O's.
follows
:

Remembering that for each at tlie end of a there will be two O's at the end of a-, we mark off in 53361, from right to left, as many j^eriods of two figures as we can, thus 5'33'61. at the end of a, and the Each of the periods 61 and 33 calls for one
:

remaining period, 5, integer whose square

calls for the initial figure 2, since 2 is the greatest

is less

than

5.

Hence a

200.

Having found

a,

we proceed

a polynomial. This is denotes the second figure of the root multiplied by 10, and c the units The scheme at the right gives the reckoning as abridged in figure.

quite as when seeking the square root of indicated in the scheme below at the left, where b

common

practice.

a
'5' 33'

61 200
1

+ b + + 30+

c 1
5' 33'

61 231
1

4 00 00
2 a 2

=
==

a^

=
=

400 430

1
1

33 61 29 00

Ri

43

1
1

2 (a

2 (k

6)

+ 6) = +c=

460 4 61
461 4 61

= = =

(2a Ba
[2{a

+
+

b)b

33 29
4 61

461 4 61
b)

=R
a',
i?i

c]c

"We

first

subtract

the remainder

by 2

a,

then find the significant figure of 6 by dividing next find Ro hy subtracting (2 a + b)b from Ri,

by dividing R2 by 2 (a + b). simplest way of accomplishing all this, as indicated in the abridged scheme at the right, is to omit final O's and to bring down one period

and

finally c

The

Then, as each new remainder is obtained, we write at its twice the part of the root already found as a "trial divisor," obtain the next figure of the root by dividing the remainder by this trial divisor,
at a time.
left

and complete the

divisor

by

affixing this figure to

it.

We

then multiply

the complete divisor by the new figure of the root, subtract, and so obtain the next remainder. If too large a figure is obtained at any stage In the process, that is, a figure which makes the product just described

greater than the remainder in question,

we

try the next smaller figure.

4 47

EVOLUTION
Therefore, since by hypothesis

267
(a

P=

+ b + c-\---

)',

we

have

P=

a''

+ (3
a^,

a""

+ 3 ab + b^b + [3 (a-\-by + 3 (a + b)c + c^^c+

3
a^c,

where the leading terms of the several groups on the

namely,
3
a'b,

right,

, are all of

higher degree in x than

any of the terms which follow them.

From
1.

this identity

we may

find a,

i, c,

as follows

Evidently a is the cube root of the leading term of P. As the leading term of the remain2. Subtract a^ from P. der, Ri, must equal 3 a^b, we may find b by dividing this term by 3 a\
3.

from
4.

Having found b, form (3 a^ As the leading term Ri.

a^c,

equal 3

we may

find c

Continue thus until

and subtract it must by dividing this term by 3 a^. a remainder is reached which is of
-{-

3 ab

-\-

b^)

of the remainder, R2,

lower degree than a^. If this final remainder

is 0,

then

is,

as

was supposed, a

^ c perfect cube and its cube root is a If this final remainder is not 0, P is not a perfect cube, but we shall have reduced it to the form
.

+ + +

P=
where

(a

+b+c+
a^.

---y

+ R,
:

is

of lower degree than

It is convenient to arrange this reckoning as follows

Example.

Find the cube root of

x6

6x6

+
x6 x^

21x*

44x3

63x2

543;

27.
|x2

3 a2

3 X*

6 x5

21 X*

44 x3

63 x2

54 X

+ 2x + + 27
27

= 3x* 6x5 + 21x* + = 6 x3 + 4 x2 3x2 2 X + 3x* + 6x3 + 4x2 6x5 + 12x*+ 3(x2+ 2x)2 = 3x* + 12x3+ 12x2 9x* + 9x2 + 18x + 9 3(x2 +2x)3 + 32= 3 X* + 12 x3 + 21 x2 + 18 X + 9 9x^ +

3(x2)2 ( 2 x)2

44x3
8x3 36x3

63x2

+ 54x +

= =

Bi

+ +

63x2 63x2

+ +

54

a;

27 27

R2

30x3

54x

=B

268

A COLLEGE ALGEBRA
is 0,

Since the final remainder

x^

6 x^

54 x

cube and

its

cube root

is x'^

2x

3.

Compa''e

569,

+ 27 is a perfect Ex. 2.

Observe that as each new remainder R^, R^, is found, we its leading term by 3 a'^ and so get the next term of Then at the left of the remainder we write the sum the root.
divide
of three times the square of the part of the root previously obtained, three times the product of this part by the new term, and the square of the new term. multiply this

We

term, subtract the result from the remainder under consideration, and thus obtain the next remainder.

sum by the new

576

This method is also applicable to a polynomial which involves more than one letter, if it be a perfect cube (compare 571). The method may also be applied to a polynomial in x
arranged in asceiiding powers of this
lack a constant term.
letter,

if it

does not

we
577

If the polynomial is not a perfect cube, thus obtain approxiinate cube roots (compare 572).

Cube roots of numbers.

We may

also find the cube root of

a number by aid of the formulas of

Example.

575.

Extract the cube root of 12487168.

a
8 000 000

N = 12' 487' 168 [200 + 30 + 2 = 232

= 3a6= 62 =
3a2
8 (a

-ir

120000 4 487 108

18000

=
= =

i?i

= iV-a3

900
138900 4167_000 = 158700 320 168
1380
4
(3 a2
i?2

3 a6

62)

3{a

+ 6)2 + 6)c= c2 =

-ZV'

(a

6)3

160084 320 168

[3 (a

6)2

3 (a
6

=i? = iV-(a +
In order to find
a,

+ 6) c + + cf.

c?] c

the greatest

number with one

significant figure

whose cube
figures in

is

contained in N, we begin by marking off periods of three from right to left (also from the decimal point to the right
:

when

12' 487' 168. there are decimal figures in iV), thus Each of the at the end of a, and the remaining periods 168 and 487 calls for one

EVOLUTION
cube
period, 12, calls for the initial figure 2, 2 being the greatest integer is contained in 12. Hence a = 200.

269
whose

rest of the reckoning is fully indicated above. Observe that each new figure of the root is found by dividing the remainder last obtained by three times the square of the part of the root

The

already found ; thus, we find tlie significant figure of b by dividing Ei by 3 a"-^, and c by dividing R^ by 3 (a + b)'^. If too large a figure is thus obtained, we test the next smaller figure.

The process may be abbreviated in the the square root of a number.

same way

as that for finding

cubes

Approximate cube roots of numbers which are not perfect may also be found by this process (compare 574).
578
is

which

of polynomials. The fourth root of a polynomial a perfect fourth power may be obtained by finding the square root of its square root similarly the sixth root of

Higher roots

a polynomial which is a perfect sixth power by finding the cube root of its square root.

may

be obtained

It is also possible to develop special methods, analogous to those of 570, 575, for finding any root that may be required. But the general method of 569 makes this unnecessary. In

we have given the special methods for square and cube roots explained in 570, 575 only because of their historic interest and their relation to the problem of finding square
fact

and cube roots of numbers.

EXERCISE
Simplify the following expressions.

XXXn

125

aV2

\625c2d3

'^

'

By
4. 5.
6.

569 or

570 find the square roots of the following.

1.

X*

x2

-2x3 + 3x2-2x+ - 2 X* + 6 x3 - 6 X +

x6

9,

4 x6

12 x^y

9 x^y'^

4 x^y^

6 x^y*

y^.

7.

4x2-20z +

13

+ 30/x +

9/x2.

270
8.
9.

A COLLEGE ALGEBRA
49 -84a;
x8
{X2

-34x2 + 60x8 +

25x*.
5 x2

2 x^

x6

x*

6 x3

4X

4.

10.

11.
12.

+ 1)2-4X(X2-1). 4x* + 9x22/2 _ 12x32/ + +

2/2/X2

16x2

24x?/

16.

X2/2/2

+2+

2 X2

2 y2

x22/2.

Find approximate square roots to four terms of the following.

13.
1

-2x.

14.

4-x + 3x2.
+
1.

By
15.

569 or

575 find the cube roots of the following.

x6

3 x5

X*

7 x3

6 x2

3X

16. 17. 18.

27x12
8 x6

+ 27x10- 18x8 -17x6 +

36 ax5

6 x*

90 aH^

135 aH^

+ 3x2- i_ - 81 a^x + +
135 a*x2

27
7.

a^.

xVy^ +
x2.

2/Va;^

3x2/2/2

3y2/x2

+ 6x/y +

62//X

19.
1

Find the approximate cube root to

-X+
20.

three terms of the expression

By

569 or 578 find the fourth root of

x8

4x^

10x6

16x5

+ + +

i9a;4

16x3

iox2

_ 4x +

1.

21.

By

569 find the

-).

fifth

root of

xio

5x9

+ +

15x8

30xT

45x6

45x*

+ +

51 x^

30x3

15x2

+ 5x +

1.

22.

To make

x* to

Ox^

11 x2

ax

5 a perfect square,

what values

must be assigned

a and

h ?

Find the square roots

23.
26.

of the following

numbers.
25. 583.2225. 28.

27889.

24. 27.

2313.61.

4149369.

.00320356.

9.024016.

Find

approximate square roots of the following

numbers correct to the

third decimal figure.

29.
2.

30.

55.5.

31.

234.561.

Find the cube roots

32.

of the following numbers.

1800867.

33.

167284.161.

34.

1036.433728.

IRRATIONAL FUNCTIONS
XII.

271

IRRATIONAL FUNCTIONS. RADICALS AND FRACTIONAL EXPONENTS

REDUCTION OF RADICALS

Roots.
tive

In what follows the letters

a, b,

will denote post-

579

numbers

or literal expressions supposed to have positive

values.

Again, va will denote the principal wth root of a, that is, the positive number whose nth power is a in other words, the
;

positive

number which

is

defined by the formula (Va)"

a.

Finally,

when
a,

71

is

odd,

V a will denote
a.

the principal ?ith

root of

namely
is

And when we use

This

the word root

we

shall

mean principal

root.

a restricted use of the word root ; for any number whose Note. nth power equals a is itself an nth root of a, and there are always n such numbers, as will be proved subsequently.
Thus, since 2^
of 4.

580

4 and (- 2)2

4,

both 2 and

We

shall indicate the principal root 2

by Vi, the other root

2 are square roots - 2

by -Vi.
the
is odd and a is real, one of the nth roots of a is real and of same sign as a, and the rest are imaginary. When n is even and a is positive, two of the nth roots of a are real, equal numerically, but of contrary sign, and the rest are imaginary.

When n

When

is

even and a

is

negative, all the nth roots of a are imaginary.

In the higher mathematics va usually denotes any nth root of a, not, as here, the principal root only.

Radicals. Any expression of the form Va^ or h Va is called a radical; and n is called the index, a the radicand, and b the coefficient of the radical.

581

When
b

both a and

b are rational

numbers or expressions,

Va

is

called a simple radical.

5

Thus

vi

is

a simple radical whose index

is 3, its

radicand

4,

and

its

coefl&cient 5.

272
582

A COLLEGE ALGEBRA

The rules for reckoning Formulas for reckoning with radicals. with radicals are based on the following formulas, in which
m,
n,

denote positive integers.

4.

/ {ydf =

71/

^aJ'\

5.

m/ V Va =
n

r-

mn ,

Va.

not changed
to both

Observe in particular that, by 1, the value of a radical is if its index and the exponent of its radicand are
if

multiplied by the same positive integer or

y

any factor common

similarity of this thus, rule to the rule for simplifying a fraction is obvious. These formulas may be proved by aid of the definition
is

cancelled

Va.''

Va^.

The

(Va)" = a, the laws of exponents and the rule of equality, 261, 3,

(.")"

""",

(aby

a"b",

Two

positive numbers are equal if

any

like

powers of these

numbers are equal.

Thus,
1.

Va'"
(

Va""^, since their

np\h powers are equal.

For
2.

Va"'i^)"^

a'"''

and

(Va'")"-^

(aP'Y

a"'''.

'Vab
(

= Va
Va6)"

V^, since their 7ith powers are equal.

ab
;

For
Q
'

=
,

and

Va

N^ti)"

(Va)

Vft)"

ab.

\/-

=
Vb
/

^^

since their th

powers are equal.

^ For
4.

"fax"

(VJ=^;
(

J and /

(-)=

Va \ "

Va)"

=-.

Va)'"

\a"', since their nth.

powers are equal.

[(Va)"]'"

For

(Va"')"

a'";

and [(Va)'"]"

a"'.

IRRATIONAL FUNCTIONS
m
5.
/

273

Va
(

= Va,

since their
;

mnth powers

are equal.

(va) = a. and [^ y^a) Va)""' = a The following examples will show the usefulness
For

of these

formulas.
1.

-v^S

= V^ = V2.
VSc

2.

Vsab^ = VTF^ V^ab = 26 V2ab.

3.

^1^
(

VSc
de2
3
,

4-

V V V32
,

Vd^

X1&2/5

_ 10, x^^y^ = V 2 x^y. V32

.

5.

V2 a;2/2)2 = V(2 x2/2)2 = v'4 x^y^ = y ^4 x-y.

583

- On That form of a radical is regarded simplifying radicals. as simplest in which the radicand is the simplest integral Hence for simplifying radicals we have expression possible.

the following rules, which are immediate consequences of the formulas just demonstrated.
1.

If the radicand

common with
index.

be a poiverivhose exponent has a factor in the index, cancel that factor in both exponent and

Thus,
2.
is

V27 X'V = V(3 xy-^Y = Vsxy^.

aiiij

//

divisible by the index, divide the exponent

factor of the radicand be a power ivhose exjyoncnt by the index and

then remove the factor from under the radical sign.

Thus,
3.

Vl6x"2/9

V2%x3y8y

2 xy'^ Vx^y.
its

If the radicand be a fraction, multiply denominator by the simplest expression which

sible to

numerator and
render
it

tvill

pos-

remove the denominator frotn under the radical sign.

^,

Thus,

-yA-^-x

s/xw

14
\

= Vn
xyz
1
^

4 xyz.

Similar radicals. Eadicals which, when reduced to their simplest forms, differ in their coefficients only are said to be
similar.

584

Thus,

VTx^
2

and

VsTx^
9 x^y
"^''xy,

are similar

for their simplest

forms,

namely

Vxy and

differ in their coefficients only.

274
585

A COLLEGE ALGEBRA
the coefficient of a radical under the radical sign.
Vjfri, the coefficient of a radical
if its

On bringing
Since b

V^ =

under the radical sign index of the radical.

may be brought exponent be multiplied by the

EXERCISE
Reduce each of the following
1.

XXXm

radicals to its simplest form.

3.

Vl8.

2.

V588.

-^-272.
^^374.
11. 14.

4.

^-

1000.

5. 9.

V372.

6.

V372.
10.

7.

8.

V?7l6.

V25 asfeiocisde.

Vl28aW.
Va"62"c3.
16. 18.

v'Sx^z/^zis.

12. "^/25 a26*c6.

15.

13.

Va2" + >6'' + 2e<".

y).
d-b''.

Vx22/2
V'xG

a;'2;22.

V(x2
-t/a^^*

?/2)

(X

17.

xSyS.

O'^d^

19.

4t^. \32a62
a/ 1

20.

J-^

X2 21.

+
^

1
-

\a-h

<f^ + \9(x
-V

l)2

22.

23.

A/

24.

sign.

Bring the
25.

coefficients of the following

under the radical

27.

SaVs^.

26.

^^-L^a/"^'.

3ax% 1/27 a^x^

Show
28. 30.

that the following sets of radicals are similar.

29.

vTs, V50, and VTJ^.

V(x3

V2i, VT92, and Vs/Q.

^SyS

2/3)

(X

y)

and Vx*y2

a;22/*.

OPERATIONS WITH RADICALS

586
Addition and subtraction.

We

have the rule

To reduce the algebraic sum of two or more radicals to its simplest form, simplify each radical and then combine such of them as am similar by adding their coefficients.

IRRATIONAL FUNCTIONS
Example.

275
Vi/2.

Add Vl6 a^b, - Voo^ft,

3 V2, and

We

have Vl6 a2&

- Voa^ + 3 v^ _ 2 VT/2. = iaVb -SaVb + 3V2 -V2 = aVb + 2V2.

dissirnilar radicals

Observe that a sum of two

reduced to a single radical.

cannot be

for squaring, .-. either x =

Thus, we cannot have Vx + Vy=^ Va; + y except when a; or ?/ is we have x + y + 2 y/xy = x + y, .-.2 Vxy = 0, .-. xy =
or
2/

0,

0.

Reduction of radicals to a

common

index.

It follows

from the

587

we can always reduce two or more radicals to equivalent radicals having a common index. The least common index is the least common multiple of the given
formula
Va"'* that
indices.

V a'" =

Example.

Reduce Vo^ and VE^

to their least

common

index.
8, is 24.

The
6.

least

common
8,,

multiple of the given indices, 6 and

And

V a5 = U, Va2o

and V63

= Vfts.

24

We make the reduction to a Comparison of radicals. index when we wish to compare given radicals.
Example 1. Compare v 16, V6, and v3. The least common multiple of the given indices,
15 y

common

588

V 16 = V162 = V256
Therefore, since 256

30

30/

10
;

V6 = V& = V216
have

30,

15, 10, 6, is
30,
30,

30

and

30;

15

> 243 > 216, we

V3 = Vs^ = V243. , V 16 > v3 > 10/V6.

;

6,

Example

2.

Compare 2 Vs and ViT.

Bringing the coefficient of the first radical under the radical sign, 585, and then reducing both radicals to the common index 6, we have
2

Vs = Vl2 = \/l23 = \/l728

^41
2

= VlP = ^1681.
VJT.

Therefore, since 1728

> 1681, we have

Vs >

Multiplication and division.

/

From
71
y

the formulas

589

n r~

\/a-^b
we

= -\ab

ni

and
:

^a/^h = ^a/b

ni

derive the following rule

276

A COLLEGE ALGEBRA

if necessary

To multiply or divide one radical by another, reduce them Then to radicals having the least common index.
coefficients

find the product or quotient of their

separately.

and radicands

Example

1.

Multiply 4
2

Vxy by

Va:'-?/^.

We have 4 Vxy
Example
2.

V^- = 8 Vx^^
2

-v^V =

VxV = 8

a;?/

Vxy.

Divide 6 Vicy by
6

vxy.

We
590

have

Vxy /2 Vxy =

Vx^/ Vxy = 3 Vxy.

and
:

Involution.

From

the formulas

(V^)'

= Va^

'v'a'"'^

= Va"'
the

we

derive the following rule

To raise a radical of the form, Va^

to

mth power,

dancel

any

factor tvhich

may

be

common

to

and

the index of the

radical,

and then multiply

6,

the exponent of the radicand by

the remaining factor of m.

Example.

Raise 2 Vxi/2 to the 9th power.

We
(2

have

^xy^f = 29(V^)9 = 128(V^2)3 ^ i28

V^ =
/

128 xy^ V^.

591

Evolution.

From
n,

the formulas

m/ V >/ = Va

nmi

and Va'"^

np

= Va"

we

derive the following rule

To

any

radical of the form Va, cancel fiiid the mth root of a and the exponent of the factor tvhich may be common to

radicand and midtiply the index of the radical by the remaining

factor of n.
B

Example

1.

Find the sixth root of vx%*.

0/& V V Vx^y* = V Vxy'^ =
.

We have
Example
2.

\xy^

Find the cube root of 54 a v6. a V6

We

have

-^Uay^h = ^S^

^VTm = 3 vT^.

IRRATIONAL FUNCTIONS
Simple radical expressions.

277
592

By

we
call

shall

mean any expression which

a simjile radical expression involves simple radicals

Thus, Va + Vo is a simple radical expression. We such an expression integral when it involves no fraction with a radical in its denominator.
only.

By

the rules just given, sums, differences, products, and

powers of simple integral radical expressions can be reduced to algebraic sums of simple radicals. In 607 we shall show
that the like
is

true of quotients.
-\-

But ordinarily a
'vb,

root of

simple radical expression, as \ a a simple radical expression. Example

1.

cannot be reduced to

Multiply

3V6 + 2V5by2V3- VlO.

3

We hare
(3

Ve +

V5)

(2

Vs - VTo) = 6 VTs + 4 VTs = 8 V2 - 2 Vl5.

Veo -

Vso

Example

2.

Square

V2 + Vi.
2

We have
(V2 +
\/4)2

V2 V4 + vT6 =

V^4-2 v^.

EXERCISE XXXIV
Reduce the following
6/
1.

to their least

common
3,
2.

index.
i.

10,

IS,

V3, V3, and V3.

following.
y/s.

Va^,

V2aW, andV7 65.

6,

Compare the
3.

Vi

and 2

4.

V3, v^, and V^.

Reduce each
5.

of the following to a simple radical in its simplest form.

6.

V35 - VTJl.

10

- V5.
.

7.
.

4
2 2

- V'2.
V3
-^

8.

Ve VlO

VT5.

9.

v^
-v^S

a/qo

vTs.

10. 13.

3 \^.

11.
14.

V2-^/2--^.
Va^b^c^

12.

-!-

v^.
'

V35 V65 h- V9I

y^a^c\

15.
17.

V^

'V^.
.

16.

V^ - V^.

Va26c2

Va62ci.

278
18.

A COLLEGE ALGEBRA
Va

Va.
15
,

19.

Va/fe-Vo/d.

20.
21.

va&2

Vafts

-4-

(VaT59
22.

Vai26")
23. (2 Vx2/223)6,
26.

(Vl2)3.

{\/a2)8.

24.

VV^V v^.
V V2..^2.

25.

vV8.

^a366/c9.

27.

28.

V2V2.
2

29.

V2V2.
/-"/an , \nmp "~/2n -\n

30.

31.

Vva"

m/

32.

(V

Va)

Simplify each of the following as far as possible.

33.
35.

V12 + V75 - V48 +

3 3

V^.
Vti.

34. 36.
38.

Vl25 + VlTS - V28 + Vl/20,

VSOO-^

los

+ vTTi. - Vax^ -

37.
39.

V50 _ Vii + \/-24 +

Vax3

+ V6/ca + y/c/ab. V(a + 6)2c - Va2c - V^.

y/a/hc

6 ax2

9 ax

4 a2x2

4 a^x.

w Vx + 2/
41.
43.

\x
Ve.

w y
(

\x2 \x^

y''

(Vi + Vs+Ve)-

4S..

Vo + VlO + Vl4)

-=-

V2.

(V6+V5)(V2+Vl5).
(1

44.

V5 + 2V2 V5 - 2 Vi".
(Va + Va+l)(Va- Va + 1)-

45.

+ V3)3.

46.

FRACTIONAL AND NEGATIVE EXPONENTS

693

In many cases reckoning with radicals is greatly facilitated by the use oi fractional exponents. Thus far we have attached a meaning to the expression a" The rules for reckononly when n denotes a positive integer.
ing with such expressions, namely,
1.
n a" a"

= f.m + n
a'

2.

(a^Y

^"'"5

3-

(^)"

= *"*">

among the simplest in algebra. Can we find useful meanings for a", in agreement with these rules, when n is not a positive integer ?
are

It is therefore natural to

inquire

IRRATIONAL FUNCTIONS
The
definition &v/(i

279

Va^. Takfi_af_for_matance. wish, a meaning for this symbol which will be in agreement with the rules 1, 2, 3. But, to be in agreement with 1, we must have
if possible, to find

We

594

(a^y
that
is,

= a^.a^ = J^'^ =

a^

a,

We We

must mean either Va or Va. choose the more convenient of these two meanings, and
a^

define a^ as

va.

thus find that one of the conditions which

we wish
a^ as

a^

to satisfy suffices to fix its meaning. Similar reasoning leads us to define a^ as

Va,

Va^,

and
of

in general a* as Va**, that

p

is,

as the principal qth root pm

p

aP,

is

Observe that since a* = Vo^ = Vo^ = a'", the value of a not changed when ^9/3- is replaced by an equivalent fraction.
Thus, a^

a^

a^] also

a"^

a^

a*.

The

definition

a^=

1.

Again, to be in agreement with

aOa"'

l,^

595

we must have
and therefore
a

= a^+"' = a, = a^/a*" = 1.
596

We are
The
1,

definition

therefore led to define a'' as 1. a~^ Finally, to be in agreement with 1/a.

we must

have,

595

or' a'

and therefore

= a-* + = a" = 1, a~* = l/a'.

-'

We
It

are therefore led to define a~^ as 1/a'.

Thus, by definition, a-^

/a^,

a~^

/a^

/ \^aP.

remains to prove that the meanings thus found for a, a**, and a~* are in complete agreement with the rules of exponents.

Theorem
values of va.

1.

The and u.

lav; sJ^

Si

= sJ^^'^

holds good for all rational

597

280
Let p,
1.
q, r, s

A COLLEGE ALGEBRA
denote any positive integers.

Then

When m = p / q
-

and n
SI

7'/s,

we have,
gs,

582,

2.

a^-a' = y/aP Va'" = V a^" Va'" P ps + _ -<JaP' + = .'' = at \ When m = p/q and n = r/s, we have, by Case
7/
qr
r
qs
I

qsi

'f

1,

a "-a

'

1 = ^^-r =
a"
a'

=
r/s,

-^r
a

-'-+(--) "

'.

3.

When m =p /q
T> -

and n
Si

=
qsi

and p/q

> r/s,

we have

T '

ai-a

q, = <Ja^ / <J =
a''

-y/a'"

Va^"-'"'
4,

-r^ = a-^ = a" i-H)p.t

/ Wa"''

qs I

qr

">'

When
p

m=p/q
3,

and n

=
1

r/s,

and

p/q<r/s, we

have, by Case

1
9 ..

?+(-0

598

Theorem

2.

The law

(a"")"

a"^"

holds good for all rational

vahies of ni and n.

For, let
1.

denote any rational number.

71

Then

When
(a)

is
a"*

a piositive integer,
a'"

we

have,
'"

597,

to

factors

+ '" +

<''>"

^ ^mn^
we

2.

When n = p / q,
1,

where

and q are positive

integers,

have, by Case

p (a"")?

= V(a"')" =
7/
s,

7/

Va""'

mp

P m-.

3.

When n =
1, 2,

where

s is

any positive
-""

rational,

we have,

by Cases

(a^\-' ^ ^

11 = =
(a.'")"

a"'^->.

a""

IRRATIONAL FUNCTIONS
Theorem
values of n.
1.

281
599

3.

The law

(ab)"

a"b" holds good for all rational

Let ii=zp/q, where

p

p and

q denote positive integers.

p p
a6.

Then
(aby
2.

= -^(aby =
s,

-^a^b"

Vop

-Vb^

Let n

where

denotes

whether integral or

fractional.

any positive rational, Then, by Case 1,

'

{aby
Applications.

aflf

The following examples

will

illustrate the

600

and negative exponents. A complicated piece of reckoning with radicals often becomes less confusing
use of fractional

when

this notation is empl-^yed.

Example

1.

Simplify

^ a/^a. V a/
3)*
6

We

have
2.

\a/y/a-.{aa
4

(f(3)5
Z,

ai

= Va.

Example

Simplify

^ ah^ v a^b
Va-ft^

h-

^a-b^.

We have

Va6^

va^6

-^

a^b^

a%^ a~%~^
12 ^

Example

3.

Expand
(x^

(x'

y~^^.

We

have

y-iy^

(xS)3

+ 3(a;3)2y-i + 3xi(y-i)2 +
Z

(j^-i^s

x~

-\-

x^" + 3 x%~ ^
^

-\-

y~^.

Example

4.

Divide x

?/

by x^

x^y^

+
2

y^.

Arranging the reckoning as

in 401,
1

we have

x^

+
y

11
x^y"^

+ y^

x?w3

x'w' x^y^

Hence the quotient

is

x^y^

x'

3.

282

A COLLEGE ALGEBRA
EXERCISE XXXV

Express as simply as possible without radical signs

1.

'^.

2.

V^.
b

3.

aV^.

4.

Vb*

^.
7.

Express without negative or fractional exponents

5.

a'^.

6. 8.

c-i-5.

(d')-6.

(e-^)-i

Express with positive exponents and without radical signs

9.

a- 1/6'0-2.

10. 12.

x~^Vy^.
a;-2

11.

(l/Vx^)-4.

Vj^/2/-2 Vx^.

Express as simply as possible without denominators

a
be

6-1 c-2

a-i(6-i
a-2(6

+ +

c-i)
c)

6-

+ +

c-i

Reduce each
14.
17.

of the following to its simplest exponential form.

15.
18.

(3i)l
8-^.
(a^6)^a56*.

8li.

16.

(-27)?.
a^oThr-^^.
{ai)K

ah^K
a6-2/a-36.

19. 22. 25.

27.
29.

20. 23.

21.
24.

(a-i6-2c3)-2.

(-32aio)l

{_ a66-9)-3.

26. 28.

6-iv^--6-i

V6=^.

(a-W6^)3.
Va'^ (6c1)-2.

(8a-iV^125a3)-3.

30.
32. 34. 35.

Va-i%'a3.
[(a;'^)^]''.

31.

"v'a?

V(F^/ Vv'^ Va.

33.

(x**

+ ^yy!''+^)=^+!'.

(X'

r)/(x-i

y"'*).

Multiply x*

+
6^

x^y'

y^

by x'
6'.

x'j/^

y.

36.
37.

Divide a^

by a^
yh^)*.

Expand

(x'

39.

Find square root of

Find cube root of
x^

x^

4x^y^

40.

+ 3 x^

Simplify [{^ + e-=^)2 - 4)]i ^ ^xy + 6x%^ + 'I2y^ + 9x-^y*. + 6 x + 7 + 6 z-i + 3 x-2 + x-s.
38.

IRRATIONAL FUNCTIONS
THE BINOMIAL THEOREM FOR NEGATIVE AND FRACTIONAL EXPONENTS
If in the binomial expansion,

283

561,
ti(n
J.

(a

by

= a" + na"-^b + V

T) ^

a^b^

we

assign a fractional or negative value to n,

infinite, series
;

we
for

on the right a never-ending, or

coefficients n, It will be
first

shall have none of the

601

1) /2, will then be 0. n(n shown further on that ii b < a the sum
;

of the

terms of this series will approach the value of (a + b)" as limit when 7n is indefinitely increased in other words, that, by adding a sufficient number of the terms of this series, we

may

obtain a result approximating as closely as

we

please to

the value of (a

+ by.
is

This
tive

is

what

meant when

iu

is

said that the binomial

is

theorem holds good for (a

+ by

when n

fractional or nega-

and

<
1.

a.

Example
Putting
71

Expand
1/3, a

(8

x~^)^ to four terms.

8, 6

= x~^

in the formula,

we have

(8

+ x~*)^ = 8' -f ^

S-'x"^

+ 5l_^8-8(x~5)2
V
;

2-3
. =2

X
-f

X
288

5x

'
.

12

20736
expansion of
1

Example
(a^

2.

Find the sixth term

in the

/ (a^

-{-

x^)^ or

x^)-2.

Putting n
term, 565,

2,

o^, 6

= x^,

5 in the formula for the (r

l)th

we have
,
^
' ^

(-2)(-3)(-4)(-5)(-6) 1-2-3-4.6

,
'

^_

^ ^_,^^_

284
Example
Since
3.

A COLLEGE ALGEBRA
Expand
x

vT+x to

four terms.

Vl +

=
x

(1

x)^ we have n

|,

1, 6

x,

Hence

^\

\-

-\ ^
=
1

^x
X
2

^-

x2

+
.

2-3

X2

X3

16

The

result is the
4.

same

as that obtained in

572,

Ex.

1.

Example

Find an approximate value

(3^

of

v 10.

We
and

have

VlO =

i)^

(1

\)\

3(l.,)^.3[l.^.l.^(i;
2-3
216

Q-]
+
.00025

3888
.00462

.16666

3.1623 nearly

EXERCISE XXXVI
Expand each of the following to four terms.
1.

(1

x)i
i

2.

(a^

+ x-^)-l
_i
=)-*.

3.

4^(27

-2x)2. 3V?/)-6.

4.

(a^

x)"*.

5.

(a-i-6

8.

2
10.

+ 3x

^
-^(1

e.

(Vx +

9.

x)2

(^ Vi'y Wi +3

Find the tenth term

in (1
in

x)-3,

11.
12.

Find the seventh term

(x-2

2 y*)^.

Find the term involving x^

in (1 in

x^)*.

13.
14.

Find the term involving x-2

x~^

(2

x~^)-3.

By

the

method

illustrated in 601, Ex. 4, find

approximate values

of the following.
1.

V99.

2.

V^.

3.

V3I.

IRRATIONAL FUNCTIONS
RATIONALIZING FACTORS

285

the product of two given radieach of these expressions is called a rationalizing factor of the other.
Rationalizing factors.
cal expressions is rational,

When

602

Thus,

izing factor of

Va + V6) Va - V6) = a b. Va Vo, and vice versa.

(

Hence

Va + Vft

is

a rational-

It can be proved that every finite expression which involves The following simple radicals only has a rationalizing factor. sections will serve to illustrate this general theorem.

Every expresand integral with respect to v .c can be reduced to the form A -\- B V^, where A and B are rational and integral with respect to x and A -\- B Vic has, with
sion which
is

Rationalizing factors of functions of square roots.

rational

603

respect to x, the rationalizing factor merely changing the sign of \x.

Thus, 2(Vx)*

A B

'wx, obtained

by
this

+ 3x(Vx)3 may

be written 2x2
a;2

expression has the rationalizing factor 2

x'^

-)- 3^2 Vx. vx.

Hence

We may
which
is

obtain

a rationalizing factor

of

an expression

and integral with respect to any finite number of square roots, as Va;, Vy, V^, by repetitions of the For we shall obtain a result which process just explained.
rational

is

completely rational
its

if

we multiply

the given expression

rationalizing factor with respect to Va-, the product by its rationalizing factor with respect to V^y, and so on.

by

Example.

Find the rationalizing factor

1

We We
or

have
(1)

Multiply

by

obtain

Multiply

(3)

by

We
is

obtain

ofl+Vx + Vy + + V^ + Vx (1 + 2 ^y). 1 + Vy- Vx(l + 2 Vy) _ x (1 + 2 V^)2, (1 + V^)2 1 -x + y -ixy + 2V^(\ -2x). 1 - x + y - ixy - 2 Vy {1 - 2 x) - 4:Xy)" - 4 (1 - 2 x)2. (I -x + y
. .

Vxy.
(1)

(2)

(3)

(4) (5)
(2)

?/

Therefore, since (5)

is

completely rational, the product of

(1).

and

(4)

the rationalizing factor of

286
604
Rationalizing

A COLLEGE ALGEBRA
factors
of

binomial

radical

expressions.

The

rationalizing factor of an expression of the may be found as in the following example.

form

Va V6

Example.

Find the rationalizing factor of

v a + v6.

We have

Va + V6 =
+

a'

6^

(a^)'^

(63)5.

(1)

But, 438, (a2) a2 - 63, the quotient being

(a^)^ will exactly divide the

(a'^)^

rational expression
(63)1
(2)

(a^)^

(b^)'^

Hence

(2) is

the rationalizing factor of

(1).

605

On

rationalizing the denominator of a fraction.

^4

Any

irrational

/B, in which B involves simple radicals only, may be reduced to an equivalent expression having a rational denominator by multiplying both A and B by the
expression of the form
rationalizing factor of B.
j

Example

1.

Rationalize the denominator of 1/ Va^.

1
4,

We u have
-m-

1 =
3

a*
3
1

a^ V = ^a/a.
, '

Example

2.

Rationalize the denominator of

VxVx2

+ a^ + Vx2 + a2 - Vz2 _

a'^

a2

We have
Vx2"+
a2 a2

Vx2

+ Vx2 _ _ Vx2 -

a2

_
(

a2

Vx2 +

a2

Vz2 + a2 + Vx^ - a^ )2 - v'x2 - a^) Vx2 + a2 + Vx2 - a2)

(

_
606

x2

+ Vx* -

a*

cal expression

In computing an approximate value of a fractional numeriwhich involves radicals, one should begin by

rationalizing the denominator.

is

Much

unnecessary reckoning

thus avoided.
Example. Find an approximate value of
(1

+ V8) / (3 v^)
.
+ v^ .

which

is

correct to the third decimal figure.

Wehave

11^^ = 11+1^^^^ 3-V^ (3 -V2)(3-f-

V2)

2.414

...

IRRATIONAL FUNCTIONS

287
607

To divide one radical expresDivision of radical expressions. sion by another, we write the quotient in the form of a fraction and then rationalize the denominator of this fraction.
Example.
"We have 4 + 2V5 + V2 + V5
Divide 4

+ 2V5byl-V2+ V5.
3

(1

(4+2V5)(l +V2-V5) + V2 + V5) (1 + V2 - Vs)

+ 2V2-V5 + VT0 V2 - 1
^/^
^

^ (-3 + 2V2-V5 + VlO)(V2 + (V2- 1){V2 + 1)

l)

^^

^/^_

It follows from 592 and 607 that every General result. expression which involves simple radicals only can be reduced to an algebraic sum of simple radicals.

608

EXERCISE XXXVn
Find rationalizing factors of the following.
1.

\/^.

2.

y^cC^

V63.

3.

x^

x^

xl

4. 7.
9.

Va + Vbc. Vx + Vy - Vi _
x3 x^ 3
3

5.

vx + Vy +
8.

Vz.

6.

Vxy+Vyz+Vzx.
1.

Vii.

Vx + Vx +
11.

+ yK
+
,

10.
13.

Va 1 1
_|_

v62.

x* x^
1

- yK

12.

2/1

x^yK

14.

+
3

x^

1.

15.
18.

- Vs.

3
,

16.

+ V2 +

V3.
3
.

17.
3
I

+ \^.
1^

1.

V9 + V3 +
its

19.

VI2+V6+V3.
number

Keduce each

of the following to a fraction having a rational

or expression for
20.

denominator.
21.

-J.
V^4^
^-

^^.
a-V6
2^.

22.

^-^
2V3 + 3V3

23.
5 4.

..

^/^^ + ^^^
Vx +
?/

V62

a2

- vx
^

25.

L^i+^.
1_V2+V3

26.

I+V2 + V3 + V6

288
-\-

A COLLEGE ALGEBRA
xVy y Vx Vx + Vy + Vx+

y
1

Vs -

\/3

Find approximate values of the following expressions correct to the

third decimal figure.
29.

^.
^125

30.

^-2^.
V?

31.

^^^"^ V2 +V3

IRRATIONAL EQUATIONS
609

On
rule.

solving an irrational equation.

is

solving an irrational equation

The general method of described in the following

First, rationalize the equation.

Next, solve the resulting rational equation. in the given equaFilially, test all the solutions thus obtained tion and reject those which do riot satisfy it.
the denote the given equation, and For, let P rational equation obtained by multiplying both members of hy R, the rationalizing factor of P. By 341, the roots

PR =

P= of PR =

are those of

P=

and R

cover which of them are the roots of

in this equation.

= P=

jointly.

We

dis-

by testing them

Example.

Solve x

- Vx -

.5

0.

Multiplying both we obtain

or simplifying,

members by
(X

the rationalizing factor x

+ vx

6.

7)2

x^

Solving, by 455,

Subst ituting 9 for x in

- (X - 5) = 0, - 15x + 54 =; 0. x = 9 or 6. x - 7 - Vx- 5 = 0, we
a root.
6

have 9

- 7 - V9-5=0,
is false.

which
6
is

is

true.

Hence

is

But substituting 6, we have

not a root.
is

- V6 -

0,

which

Hence
obtained
is

But observe that

by equating the rationalizing factor

a root of the equation x - 7 + Vx for 6 7 + to

;

'- 5

= 0,

Vo^ 5 =

true.

IRRATIONAL FUNCTIONS

289

An

equation which involves the radical \ A

radical
in

may

be ration-

610

alized with respect to this

by collecting the terras

which involve "^A

the other, and then raising both

one member and the remaining terms in members to the n\h power.

By

repetitions of this process an

square roots only

from

equation which involves be completely rationalized. It follows 345 that this method is equivalent to that described in

may

609, but

it

involves less reckoning.

Solve

Example

1.

v Vx + a =

Vft

Cubing both members,

Transposing and squaring,
Substituting
tliis

Vx +

a x

= 6^ = (&^

a)2.

result in the given equation,

we^nd

it

to be a root.

Example

2.

Solve

Vx +

Transposing,
Squaring,
Simplifying,

+ Vx Vx x Vx + x +

= = = = = =

9.

4
4 5 5

81
5.

Vx + 0. - 18 Vx +

+x+

5.

Squaring,
Solving,

25.
20.

which

Substituting 20 for x in the given equation, Hence 20 is a root. is true.

1.

we have V25 + Vl6 =

9,

Notes.

Observe, as in Ex.

1,

that

respect to the

unknown

letter

only and

we rationalize an equation with make no attempt to rid it of radicals

611

which do not involve this letter. 2. Observe also that a n irra tional equation may have no root. Thus, the equation Vx + 5 Vx 4 = 9 has no root. For
attempt to solve
3.
it

if

we

we

shall merelj' repeat the

;

reckoning in Ex. 2 and

shall again obtain the result

We may
form

x = 20 and V25 Vl6 = 9 is false. add that the simplest method of rationalizing equation

of the
is

VI + V^ + Vc + V^ =
it

(or

V^ + V^ + Vc +

J5;

0)

to begin

by writing

thus

Va+^b = -Vc -Vd

{otVa

+Vb = -Vc -E)

will involve

and then to square both members. The resulting equation but two radicals and it may be rationalized as in Ex. 2.

290
612

A COLLEGE ALGEBRA
To
solve a system of such

Simultaneous irrational equations.

equations we

may

first

rationalize each equation, then solve

the resulting rational system, and finally test the results thus obtained in the given system. But if the equations are of the form described in 379, they should be solved by the method there explained.

Example

1.

Solve

Vx

+ V?/ + 5 = Vx + v^, x + 2y=n.

5y

(1) (2)

Squaring
or

(1),

x-5 +y + 5 + 2 Vxy + 5 x Vxy + 5a;-5?/

25

25

+ 'Z\'xy,
(3)

Squaring
Solving

(3)

and simplifying,

(4), (2),

Sub.stituting x
is true.

Hence
2.

= 4. x 9, = 9, = 4 in (1), we have Vi + V9 = V9 + Vi, x = 9, y = 4 is tlie solution of (1), (2).

?/
?/

= = =

Vxy.
5.

(4) (5)

which

Example

Solve

Vx +
2 V'x

6
6

+ 2/Vy = + 6/\/7/ =

4,
9.

(1)
(2)

Solving for

Vx +
(3)

And from
and
(2).

we

/ Vy, we find Vx + 6 = 3, l/Vy = 1/2. obtain x = 3, y i, which is the solution of

and

(3)
(1)

EXERCISE
Solve the following equations for
1.

XXXVm
=
2.

x.

X*

4.

2.

x~^

3.

3.

x^

8.

4.

(V2x-l)^ = V3.

5.

a/2

+ VS + Vx =
X

6.

V^ + Vftx + V^ =
Vx +
Vx +
4

d.

7.
9.

V4 x2 +

10

2X

1.

8.

+ Vx
X
-I-

-f

11

7.

V4X + 5+ Vx+1- V9x+10=0.

Vx-'^

10.

-6 =

0.

11.

+ 3x -

- Vx-^-x- 1 = 2.

12.

Vx +

Vx + 2 + Vx - 2 = Vx +

+ Vx 14.

1.

13.

:^^+^ Vx + 3
-v^ 6

2.

--i

Vx+1

Vx-1

^+-^^ = Vx2-1

0.

IRRATIONAL FUNCTIONS
Solve the following for x and
^
f

291

y.

Vx +
y/y

15.

+ Vy-2 = Vx + 5 + Vy + e, -X- Vs - X + Vy - 3.
17

^^

lb.

-j

r3Vx-22/-Vx + 2/-4 = 3, t Vx^ 22/ + 2Vx + y-4 = 8.

that

17.

Show Show

Vx +

+ Vx + 6 + Vx +
h

+ Vx +

will reduce

to a rational equation of the first degree.

18.

that

Vax +

rational equation of the first degree

+ Vcx~+~d Vex +/ = will if Va + Vc Ve = 0.

reduce to a

QUADRATIC SURDS
Surds.

Numerical radicals
is

like

V2^and Vo,

in

which the
are called

613

radicaud
surds.

rational bnt the radical itself

is irrational,

surd
is

as its index

is called quadratic, cubic, and so on, according two, three, and so on.

is

Theorem 1. The product of two dissimilar quadratic surds a qtiadratic surd.

614

simplest forms, their radical factors are

Suppose that when the surds have been reduced to their Va and Vi. The

product of

Va

and

V6

is

'wab,

and this

is

a surd unless ab

is

a perfect square. But ab cannot be a perfect square, since by hypothesis a and b are integers none of whose factors are square numbers, and at least one of the factors of a is different from every
factor of
Thus,
b.

V2 Vs = Ve, Ve Vis = Voo =

.

3 Vio.

Theorem

2.

The sum and

the difference of two unequal quad-

615

ratic surds are irrational numbers.

This

is

obvious
let

when the surds

are similar.

Hence

Va

and

V6

denote dissimilar surds.

292
Suppose,
if

A COLLEGE ALGEBRA
possible, that

Va + Vo =
(1)

c,

(1)

where

c is rational.

Squaring both members of

2

and transposing,
(2)

which
c^

is

impossible

Va^ ^c'-a-h, since 2 Vo^ is irrational,

where

614, while

h is rational.
3,
i/"

616

Theorem

+ Vb = c + Vd,
b

Vb

and

Vd

are

surds, then a

and

^.

For, by hypothesis,

Vo

Vrt

Vt^ impossible unless Vo since otherwise Vi V(^ would be irrational, to c a, which is rational.

But

this

is

=c

a.

and

0,

615,

and equal

Hence
617

d and a

c.

Square roots of binomial surds.

(

V^ ^yY =

-\- ]/

We have 2 Vxy.

Hence if a + 2 VT* denote a given binomial surd, and we can find two 2>os'itioe rational numbers x and y such that
X
then
-\-

^=

a and xy

h,

\x +

'Vy will be a square root of a

will be a square root of will be binomial surds.

+ 2 Vi and v^ Vy

2 Vi, and both these square roots

When
Example

such numbers

x,

y exist they may be found by

inspection.
1.

Find the square root

form
a

of 37

20

20 v3.

Reducing But

to the

Hence

V37 2.

Vs = 37 - 2 V300. = 25-12 and 37 = 25 + 12. V3OO = V25 - Vr2 = 5 - 2 V3.

2 \^,

37

300

Example

Find the square root of 13/12

12
13
- ~ = + a /_13 +
,-,

+ V5/6.
13

We

13
;

V30
6

have

2 V.30

"*"

\6

12

12

IRRATIONAL FUNCTIONS
Since 30

293
2

10

3 and 13

10

3,

we have Vi3 +

V30 = VlO + V3.

Hence
Note.

^ /l3 Xl

+2 V30 = V10 -=V3 = +

12

V12O + V36
12
:

V12

V30
-\
.

We may

obtain formulas for x and y as follows

618
(1)

By hypothesis
and

Vx + Vy = v +

2 Vft,

Vx-V^ = Va-2V6. - 4 Multiplying (1) by (2), x-y = Va2

6.

(2)
(3) (4)

But
Solvmg
(3), (4),

y x

= =

a.

y^

square.

Observe that these values are rational only when a^ 4 6 is a perfect Hence in this case only is the square root of a + 2 Vft a binomial surd.

EXERCISE XXXIX
Find square roots of the following.

1.9+ V56.
4.

2.

20

V96.

3.

32-2

Vl75.

1+2(a

-.

5.

7-3V5.
8.

6.

8V2 + 2V3O.

7.

+ Va2-62).

Vab

Simplify the following.

9.

"V^17

12 V2.

10.

V 9 + 4 V4 + 2 V3.

IMAGINARY AND COMPLEX NUMBERS

Complex numbers. Since all even povs^ers of negative numbers are positive, no even root of a negative number can be a real number. Such roots are imaginary niirabers.
Definitions of the imaginary numbers and of the operations by which they may be combined are given in 217-228, which the student should read in this connection.

619

According to these definitions,

1.

The symbol

V 1

is

called the unit of imaginaries.

294
2.

A COLLEGE ALGEBRA
Symbols of the form
ai,

where a

is real,

are called pure

h are real, are

imaginaries.
3.

Symbols of the form a

hi,

where a and

called complex numbers.

4.

Two complex numbers

and
If
a-

are equal when,

and only when,

their real parts

their imaginary parts are equal, so that

+ hi = c + di,

then a

and

= d.

The sum, difference, product, or quotient of two complex numbers is itself a complex number (in special cases a real number or a pure imaginary) which may be found by applying The the ordinary rules of reckoning and the relation i^ = 1.
5.

any positive integral power of a complex number, to n factors. since by definition (a (a-\- bi) {a hi) biy
like is true of

+
3

i)

Example

1.

Add
5

5
3

We

have
2.

+ +

+ +

and 2

i.

(2

4
i

(5

2)
i.

(3

4)

i.

Example

Subtract 6
3
2
i

+
(6

from 3
i)

We have
Example
3.

+
3

2
i

+2 = (3 - 6) +
1

(2

2)

=-

3.

Multiply 2
(2

+
(1

by
i)

i.

We have
Example
4. (l 5.

i)

=2+3 +8 + = 2 + 3 + 8i i

12

^2

12

=-

10

11

1.

Expand

(1

i)2.

We

have

i)2

= l+2i +
2/i) i

i2

= l4-2i-l = = = +
+

2i.

Example

Find
(x

real values of x,

y satisfying the equation

(x yi) (1
i)-

+ +

+
+

Carrying out the indicated operations,

we have
2/)

(2/

2)

(X

4)

(X

(X

y)i.

Equating the

real

and the imaginary parts, 619, 4, -{y -\-2) = x-\-y and x + 4 = x - i/,
x

or, solving,

6,

4.

In 238-241 we have given a method for representing complex numbers by points called their grajths, and rules for
obtaining from the graphs of two complex numbers the graphs

IRRATIOXAL FUNCTIONS
of their
to Exs.

295

sum and
1, 3, 4.

product.

Let the student apply these rules

Two complex numbers like a + hi Conjugate imaginaries. and a bi, which differ only in the signs connecting their real and imaginary parts, are called conjugate imaginaries. The product of two conjugate imaginaries is a positive real number.
Thus,
(a

620

621

bi) {a

hi)

a^ hi)

hH^
(c

a2

Ifi.

Hence a

fraction, as (a

di),

may be reduced
its

to

622

the form of a complex number by multiplying both by the conjugate of its denominator.
Example.
Divide 5

terms

+ 7iby2

4i.
(' -^
(2
'^

We

have

'-^ =
2

')
i)

(^

-4 -18 +
20

(2

+ +

^') 4
z)

34i

9
10

17.
10

1 it follows that The powers of i. From the equation i^ 1 or 1, and the odd powers the even powers of i are either

623

either i or
Thus,
i^

i.

i"^- i

i; i^
i"

i^

i-

i"^

1
-^

and so

on.

any given value of n, divide n by 4. Then, according as the remainder is 0, 1, 2, 3, the value of t is 1, j, 1, i. Thus, i^* = (i*)^ = 1 i^ = i-* i = i; and so on.
find the value of
for
;

To

Even

roots of negative numbers.

The number
for

two square

roots
2)"'^i^

( 2 ly
root,

= (

= 4. We

2i and 2i;

and write

V 4 = 2i and

2^i^ A, and (2 iy select 2 i as the principal square

4 has

the

624

V 4 = 2
=

i.

Similarly the principal square root of any given negative a is \ai, that is, v a number Vai.

From
if

this definition of principal square root it follows that b are any two negative numbers, then and

625

V a V b =
For

^\/ab.

V a V 6 = Vai

Vbi

= i^y/aVb^ y/ah.

296

A COLLEGE ALGEBRA

negative numbers

Thus, while the product of the principal square roots of two a, b, is one of the square roots of their product ab, it is not, as in the case of real numbers, the principal
square root of this product.

When
mind

reckoning with imagiuaries,

is

it is

this modification of the rule

Va Vb

important to bear in ^ab. All chance

of confusion

avoided

if

at the outset

we

replace every

symbol

V a
1.

by Vai.

Example

Simplify

We have

V^
2.

V^ V^)5 {V^y. V^)^ V^)" = V2 Vs i)5( V5

i)^

= V2- (V3)S= 1125 VsOi.

Example
Multiply 2
(2

(V5)Tii3

+
(1

V^ by

+ V-1.
(2

We have
626
numbers.

V^)

+ ^^) =

/) {I

i)

=-

bi.

The higher even

roots of negative

numbers are complex

This will be proved subsequently.

Thus, one of the fourth roots of

(1

is 1

+
-I-

for

i)4

i^

i3

i^

=-

4.

627

As will be proved farther complex numbers. on, all roots of complex numbers are themselves complex numSquare roots of
bers.

We may
have
if

find their square roots as follows.

We

{Vx iVyy^ = x - y 2iy/xy.

a
6t

+ denote a given complex number in which Hence, and we can find two positive numbers x and y such that
X
then

h is positive,

-y = a,
will
bi.

(1)

and

Vxy =

b,

(2)

Vx + i V^
may

be a square root of a

bi,

and

vx i Vy

will

be a

square root of a

We
By
and

find such

numbers x and y as

follows.
(3)

hypothesis

Multiplying

(3)

by

(4),

Vx + i Vy = Va + bi, V^ _ i Vy = Va bi. x + y - VoM^-

(4)

(6)

IRRATIONAL FUNCTIONS
But, by
(1),

297
(6)

z-y = a.
(5)

Hence, solving

and

(6),

and

= a+ Va2 +
+
i.

62

And

both these values are positive since Va^

Jfi

> a.

Example.

Find a square root

a

Here

=-

and Va2

Hence
Therefore

= {1

1 + 4 Vs = V(- 1)24. + 9)/2 = 4 and y =

of

b'^

(4
(1

V5)2

9.

9)/2

5.

V_

Vs i =

+ Vs

i.

EXERCISE XL

298

A COLLEGE ALGEBRA
XIII.

QUADRATIC EQUATIONS
may
-{-

628

General form of the equation.

one unknown

letter, as x,

Every quadratic equation in be reduced to the form

-\-

ax^

bx

c =^

i),

where
If,

a, h,

and

denote
6

known numbers.

as

may happen,
;

0,

the equation

is

called a pure

quadratic

\l b =^ 0, \i is

629

The
ax"^

roots found

by

called an affected quadratic. The roots of the equation inspection.

^ are those particular values of x for which There are two c vanishes, 332. the polynomial ax"^ -\- bx of these roots.

bx

+c=

If
ax"^

the factors
bx
-{-

of ax"^

-{-

bx

-\-

are known, the

roots

of

Q are also known, for they are the values of x If for which the factors of ax"^ + bx -\- c vanish, 253, 341. the factors are x a and x (3, the roots are a and ^.
c

Example

1.

Solve the equation xx-

We

have

{x

The factor x + S vanishes when x when X = 2. Hence the roots are

.

+ x 6 = 0. + 3) (x - 2). = and the

.3,

factor x

2 vanishes

3 and

2.

Example

2.

Solve

abx"^

{a^

b-)

+ (a^ -

b-)

0.

Factoring, by 443,

[ax
(a
x'^

(a

b)] [bx

Hence the

roots are

b)/a

- (a - 6)] = and (a - b)/b.

0.

In particular, since

q =(x

the pure quadratic x^ q = are Vy and V^. Again, since ax^ + bx = (ax + b) x, the roots of a quadratic of the form ax^ -\- bx = are b /a and 0.
Thus, the roots of 4 x2
are

Vy)

(x

V^), the roots of

9 are

3/2 and -

3 /2

the roots of 2 x2

-x=

and

/2

the roots of 5x2

^^

q and

0.

630

Conversely, to obtain the quadratic whose roots are two given numbers, as a and ji, we form the product {x a) (x ft)

and equate

this product to 0.

Thus, the quadratic whose roots are or 3 x2 + 5 X - 2 = 0.

and

/3

is (x

+ 2)

(x

- 1/3) = 0,

QUADRATIC EQUATIONS
Example
1.
1.

299

Solve the following quadratics.

0.
2.

x2

+ 2x-8 =
2.

2x2-7x +

0.

3.

(2x-l)(x-2) =

x2

2.

4.

(x-l)(x-3) = (2x-l)2.
3.1/4,0.
-{-

Example

Find the quadratics whose roots are

1.-2/3,-3/2.
be factored
ax^
-{-

2.a,-a.
But ax^
in

General formula for the roots.

;

bx

+ may
c

always

631

for, as
G

was shown

444,

bx

+
X

= a\

-b-^

^p _ 4 a(r\
;^

r
\

-b- v^rri^-i

Therefore, since the roots of

ax"^

+ bx-Jt- c =
-\-

(1)

are the values of

x for which the factors of ax^

bx -^ c

vanish, these roots are

x
or, as

b+

-\/b-

- 4ac

za

and X

we

usually write them.

300
Example.

A COLLEGE ALGEBRA
Solve 3 + 56 x - 220 = 0. = 28, and substituting in (3) we - 28 V282 + 3.220 ^.
a;2

Here 6/2

have
10

x=
633

that

IS,

or

22.

given quadratic may also be solved by applying directly the process of completing the square, 444, as in the But since this method involves needless following example.

Any
it

to

reckoning,

it

is

not to be recommended

except

when the

formula of
Example.

631 has been forgotten.

Solve

3z2-6x +

0.

Transposing the

known term and

first

dividing by the coeflBcient of x*,

x2-2x = -2/3.
Completing the square of the

member,

x2-2x + 1 = 1/3.
Extracting the square root of both members,
a;

= V3/3,

whence x

(3

V3)/3.

634

The methods
equation which

just explained enable one to solve 2aij fractional yields a quadratic when cleared of fractions.

But

see 524-527.
1.

Example

Solve

x+1

x+2
X

x+3

x+4
11

Clearing of fractions and simplifying,

Solving,

+ lOx + = -5 V3
2x2

= 0.

Both of these values of x are roots of the given equation, for they cause none of its denominators to vanish.

Example ^

2.

Solve

X2-1

^^ + X2-X + ^^ = ^^ +
X2

0.

x(x2

Clearing of fractions by multiplying by the lowest common denominator 1), and simplifying, we obtain

3x2

+ 2x

0,

whence x

or

5/3.

cannot be a root of the given equation, since its first two denominators vanish when x = 1. Hence 5/3 is the only root of this

But

equation.

QUADRATIC EQUATIONS
EXERCISE XLI
Solve the following equations.
1.

301

3.

x2

i2+2x = 35. = 10 X - 18.

2x2
x2

2.

4x2
9x2

-4x
+
6X

=3.

4.
6.

= 0.

5.

+ 3x-4 = 0.
252

(2x-3)2 =
12x2
15x2

8x.

7.
9.

+ 9x-

0.

8.

8x2

-82x +
1

207

0.

lo.
12.
2) (X

+ 56x - 255 = 0. - 80x - 64 = 0.

i)x

11.

x2-3x- + V3 = 0.
(x

x2-(6 +

2i

= 0.

13.

2)2(x

7)

(X

3) (x

6).

14.

16.

x+22x ^ +-A_^1 4x + 4 8 2(x2-l)

2x-l_^3x

-^ + ^+^2 =

2.

15.

^1 + 1=
X

"

x~l
I

17.

_i
2x +
l

4x-2

1-4x2

^
2x

=!
8

^3_

x-2
4

+ 1^5x-U_ x-4 x-3

1

19

_^+i
x(x-2)

2x-2

L_ + JL = Q.

20.

X 21

4
a;

4{x
22.

+ + 2)(3x-l) +
3

2x

17x
4)

3(3x-l)(x +

6(x

+7 + 4)(x +
0.

^^
2)

^ l_ _^+L_ + x2-2x-3 + ^i +x-l = 2x2 2x2-7x

+
3x2
c2x2

23.
25.

+ {9a- l)x-3a = 0. + c (a - 6) X - a6 = 0.
6 acx
62)x2

24. 26.

x2

- 2ax + 4 ax

a2

62 =; o.

x2

4 a2

62

= q.

27.
28.

x2
(a2

+ -

a2(9 c2 2 (a2

62)

= o.
a2

62)x

62

0.

29.

l/(x-a) + l/(x-6) + l/(x-c) =

0.

30.

(x-)^-(x-6)2 ^J_a6_^^^ o) (x 6) a2 62 (x

302

A COLLEGE ALGEBRA
EXERCISE XLH

1.

Find two consecutive integers whose product

is

506.
is

2.
3.

Find two consecutive integers the sum of whose squares

481.
is 91.

Find two consecutive integers the difference of whose cubes

4.

Find three consecutive integers the sum of whose products by

is

pairs
5.

587.
:

Find a number of two digits from the following data the product of the digits is 48, and if the digits be interchanged the number is
diminished by
6.

18.

The numerator
itself

of a certain fraction exceeds its

its

and the fraction

7.

exceeds

reciprocal by 24 / 35.

denominator by 2, Find the fraction.

cattle

Having

lost 4 of

dealer bought a certain number of steers for $1260. them, he sold the rest for $10 a head more than they
entire transaction.

cost him,

and made $260 by the

How many

steers

did he buy ?
8. A man sold some goods for $48, and his gain per cent was equal to one half the cost of the goods in dollars. What was the cost of the goods ? 9.

If

$4000 amounts to $4410 when put at compound interest for two

years, interest being

compounded annually, what

is

the rate of interest ?

inherits $25,000, but after a certain percentage has been 10. deducted for the inheritance tax and then a percentage for fees at a rate

A man

one greater than that of the inheritance tax, he receives only $22,800.

What

is

tl;ie

rate of the inheritance tax

man bought a certain number of $50 shares for $4500 when they 11. were at a certain discount. Later he sold all but 10 of them for $5850 when the premium was three times the discount at which he bought them.

How many
12.

shares did he buy

The

circumference of a hind

fore wheel

by 8 inches, and

in traveling 1

wheel of a wagon exceeds that of a mile this wheel makes 88 less

revolutions than a fore wheel.

Find the circumference of each wheel.

13. A square is surrounded by a border whose width lacks 1 inch of in being one fourth of the length of a side of the square, and whose area in inches square inches exceeds the length of the perimeter of the square by 64. Find the area of the square and that of the border.

QUADRATIC EQUATIONS
14.

303
is

The corners

of a square the length of

whose side

2 are cut off

in such a

way

that a regular octagon remains.

What

is

the length of a

side of this octagon ?

15.

vintner draws a certain quantity of wine from a full cask con-

Having filled up the cask with water, he draws the taining 63 gallons. same quantity as before and then finds that only 28 gallons of pure wine remain in the cask. How many gallons did he draw each time ?
16. A man travels 50 miles by the train A, and then after a wait of 5 minutes returns by the train B, which runs 5 miles an hour faster than the train A. The entire journey occupies 2| hours. What are the rates

of the
17.

two trains

Had pedestrian walked 6 miles in a certain interval of time. the time been 1/2 hour less, the rate would have been 2 miles per hour
greater.

Required the time and

rate.

18. pedestrian walked 12 miles at a certain rate and then 6 miles Had he walked the entire farther at a rale 1/2 mile per hour greater. distance at the greater rate, his time would have been 20 minutes less. How long did it take him to walk the 18 miles ?

19.

From

the point of intersection of

two

straight roads

which cross

and B', set out simultaneously, on the one at right angles, two men, road at the rate of 3 miles per hour, B on the other at the rate of 4 miles
per hour.
20.
If

After

how many

hours will they be 30 miles apart

on the roads just described, but at the rates of 2 and 3 miles per hour respectively, and A starts 2 hours before B, how
long after
21.
If

and

B walk

starts will they be 10 miles apart ?

from a height of a

feet a

body be thrown

vertically

upward with

per second, its height at the end of t seconds The corresponding formula is given by the formula h = a -\- ht \(i f^. when the body is throvra vertically downward i?. h a bt \Q t-.

an

initial velocity of h feet

(1)

If a

body be thrown

vertically

upward from

the ground with an

initial velocity of

of 16 feet ?
(2)

32 feet per second, when will it be at a height of 7 feet? Will it ever reach a height of 17 feet ?
is

body

thrown from a height

with an

initial velocity of

48 feet per second.

of 64 feet vertically downward When will it reach the

height of 36 feet ? (3) If a body be dropped from a height of 36 feet, the ground ?

when

will

it

reach

304

A COLLEGE ALGEBRA

XIV.

A DISCUSSION OF THE QUADRATIC EQUATION. MAXIMA AND MINIMA

The discriminant.
Let a and
ft

635

Character of the roots.

denote

the roots of ax"

The

+ c = 0, so that, 631, b 4 ac b + -\/b- 4 (ic a = p= za 2a radicand i^ 4 ac is called the discriminant of

-{-

bx

-y/b'^

>

ax"^

-\-

bx

-\-

c =^ 0.

When
a,

the coefficients

a, b, c are real,
siffii

the character of the roots

p
1. 2.

is

indicated by the

of the discriminant.

Thus

When b^ 4 ac is jJositioe, the roots are real and distinct. When b^ 4 ac is 0, the roots are real and equal. When
b'^

3.

4 ac

is

negative, the roots are conjugate imagi-

naries.
It should also be observed that
1.
2.

When ^^ 4 ac = 0, then ax'^ + ia; + c is a perfect square. When a is positive and c is negative, the roots are always

real, since 63.

4 ac

is

then positive.

If a,

b, c

are rational, the roots are rational when,

is

and

only when,

i^

4 ac
Show

a perfect square.
a;^

Example

1.

that the roots of

6''

6x +
6)-

10
1

=
10

are imaginary.

They

are imaginary since

2.

4 ac

= (

4.

Example
equal
?

For what value of

32

are the roots of inx"

+ 3x +

We

must have
3.

7n

0,

that

is,

m=

9/8.

Example
0,

If possible, factor y^

xij

2x^

llx

\2.
it

Arranging the polynomial according to powers of y and equating we have y'^ + {x + l)y - (2x2 - 11 x + 12) = 0.
Solvuig,
.
.

to

y
y

(X
5^

1)

V9x2-42x +

49
.

that

is,

4,

or y

2X

-\-

S.

A DISCUSSION OF QUADRATIC EQUATIONS

305

Hence, 681, y'^ + xi/ - 2x'^ + U x + y - 12 = {y - x + i) {y + 2 x ~ S). Observe that the factorization is possible only because the radicand 9 x^ _ 42 X + 49 is a perfect square.

Relations between roots and coefficients.

If

a and

(3

denote

636

the roots of ax^

-i-

bx
-\-

-\-

c
-\-

0,

we

have,

631,

ax^

bx

a (x

a) (x

/3).

Dividing both members of tliis identity by a and carrying out the multiplication in the second member, we have
x^
-i

X =
b
c
-\

x^

(a +

B)x

aB.

Since this
in its

is an identity, the coefficients of two members are equal, 264, that is,

like

powers of x

a
This

;8

/a and a^

/ a.

may also be proved by adding and multiplying the values of a and ft given in 631. Therefore, since a, /3 are
the roots oi x^
-{-

bx / a

-\-

/a

0,

we have the theorem

/n any quadratic of the form x^

+
to

px

the coefficient the roots,

of X

tvith its

siff7i

changed

is

equal

the

sum of

and

the constant term is equal to the product of the roots.

sum

Thus, in the quadratic 6x- + x = 2, that is, x^ + x/6 of the roots is 1/6, and their product is 1/3.

1/3 =

0,

the

Example

1.

Solve 9x2

lOx

+
,

0.

Obviously one of the roots

the product of the roots
is

is 1

for 9

10

+ 1=0.
is

1/9, the other root

1/9

-^ 1

Therefore, since or 1/9.

Example 2. 3x2 + 8x + 5
Let a and

Find the equation whose roots are

three times those of

0.

/3

denote the roots of 3 x^

Then

a+

/3

+ 8 x + 5 = 0. = - 8/3 and a/3 = 5/3.

/3)x

x2_ (3a +

Hence the required equation is = x2-3(a + 3/3)x + 3a-3i3

9a/3

= x24-8x +

15

= 0.
637

of the roots. The expressions a + /8 and symmetric functions of the roots a, jS, 540. All other rational symmetric functions of a and /3 can be expressed

Symmetric functions

aft are

306

A COLLEGE ALGEBRA

a ^ and a^, and. rationally in terms of these two functions, therefore rationally in terms of the coefficients of the equation.
For every such function can be reduced to the form of an integral symIf an metric function or to that of a quotient of two such functions. term of the type kaP^P + ', it must integral symmetric function contaijis a

But 5427>K^herefore kaPpP {ai + ^i). readily be showrHixsuccessive applications of the binomial theorem that ai + /39 can be expresSed4n terms of powers
also contain the

term kaP + i^p,

it

aP^p

(a^)P,

and

may

ot

-\-

p and

a/3.

Thus, since {a
Similarly

we

Example. and a^/S + a^"^

+ ^Y=a^ + 2 n-^+/32, we have a2 + /S^ = (a + /3)2 - 2 a/3. + /33 = (a + /3)3 -Zap{a + /3). The roots of x- + px + g = being a, express 1 / a + 1 /
lind a^
jS,

/3

in

terms oi
l//3
a/33

p and

q.

Wehave l/a +
and
a^iS

= (a + ^)/a^= -_p/g, = ^^ (,v^ + ^2) = a/3[(a+i3)2-2 a/3] = g (p2_2 q).

that, instead of being constants, the

638

Infinite roots.

Suppose

coefficients of ax^

+ Z'x + c =

are variables.

We

can then

show

that
co

if
;

approach

as limit, one of the roots will a approaches and if both a and h (but not c) approach 0, both
oc.

roots will approach

For the formulas

for the roots are

5

a=

+ V62 2a

4 ac
1

/3

-b - V62 -

4 ac

Multiply both terms of the fraction

2a a hy b

\''b-

4 ac

and both

terms of the fraction ^ by

2c

+ V6-
--,

4 ac.

We
b

obtain

a =
b

2c
/3

+ V&2 _
if

4 ac

V62 _
b.

4 ac

By
and

203, 205,

Therefore

if if

= 0, then Vb'^ -Aac = = a = 0, then a = c /b and and b = 0, then a = oo and a =

a
/3

co,
/3

==

00.

It is

customary

to state these conclusions as follows, 519:

One root of ax2 4- bx -f c ishes, and hoth roots become

vanish simultaneously.

becomes infinite ivhen a van-

infinite

when a and b

(but not c)

A DISCUSSION OF QUADRATIC EQUATIONS 307

Maxima and minima.
Let
?/

be a function of x^

278.

It

639

may happen
value, m,

that, as

x increases, y will increase to a certain

to a certain value, in\

call

and then begin to decrease, or that y will decrease and then begin to increase. We then a maximum value of y and m' a minimum, value.
2/

Thus,
being

(x

if

1)2

4 has a

minimum
less

value

when x =
1

1,

this value

4.

For

x start from a value

than

and

increase, (x

1)^

will first decrease to

and then
(x

increase.

Similarly

?/

1)^

has a

maximum

value, 4,
c

when x =

1.

Every quadratic trinomial

has either a

ax'^

ix

with real coefficients

640

maximum

or a

minimum

value,

which may be

found as in the following examples.

Example
1.

Find the maximum or minimum value

the square,
3,
?/
x"-^

of

?/

x^

+6x

7.

By completing
Example
shall
2.

Hence when x =

has a

7 = (x + 3)^ 10. minimum value, namely

x

16.

Divide a given line segment into two parts whose rectangle

have the greatest possible area.

Let 2 a denote the length of the given segment, x and 2 a of the parts, y the area of their rectangle.

x the lengths

Then

(2

x)

2 ax

x^

a^

(a

x)^.

Hence y has a maximum value when x = a, that is, when the given ment is bisected and the rectangle is a square whose area is a"^.

seg-

The maximum and minimum values of quadratic trinomials and of certain more complex functions may also be found by
the following method.
Example.
2/

641

Find the

maximum and minimum

x,

values,

if

any,

of

= (4x2-2)/(4x-3).
Clearing of fractions and solving for

we have
1) (y

^ *~ y Vy^
2

3y

~ ^

V(y 2

2)

By hypothesis, x is restricted to real values. Hence y can only take values for which the radicand (y 1) (y 2) is positive (or 0), that is, the value 1 and lesser values and the value 2 and greater values.
It

follows from this that 1

is

maximum and

2 a

minimum

value

of y.

308

A COLLEGE ALGEBRA

For observe that as y increases to 1, the two values of x, namely 3 2/ + 2)/2 and {y + V?/2 _ -^y ^ 2)/2, respectively increase and decrease to 1/2. Hence, conversely, as x increases through 1/2, y first increases to 1 and then decreases.
{y

- V2/2 _

642

Variation of a quadratic trinomial.

Given y

= ax^ +
we

Sx

c,

where a

is

positive.

By completing

the square,

obtain

Hence y has a minimum value when x

mum

value being (4c

increases from
to {^ac
let

As X

+ 00

b^)/4:a
x2

h'^)/4:a. jT 00 to + co,
3

'

= b /2 a, this minifrom

?/

will first decrease

and then increase to

00.

Thus,

(X

1)2

4.

As X

increases from
00 to

00 to

+00,?/

first

decreases from

4 and then increases

from 4 to Moreover
that
is,

00.

when ?/ = when x = 1 or

x^
3.

?/

0,

Until X reaches the value

tive
;

1,

is

posi-

it

then remains negative until x

again becomes positive.

3,

when

it

When
(-1, 0)

x=..._3, -2, -1,

0,

1,

2,3,4,

5,...

we have
?/= 12,
5,

0,-3,-4,-3,0,5,12,....
?/

We may obtain the graph of = x^ 2 x 3

by plotting these pairs of values and passing
""
,

,
(1-41 '~
^
''

a curve through them, as in 389. Observe that to the zero values of y there

correspond the points where the graph cuts the x-axis, and that to the mininmm value of y there corresponds the lowermost point of the graph, which is also a turning point of this curve.

EXERCISE XLHI
1.

For what values of

are the roots of

{ni^

(?n-|-

2)

x2-

mx + 1=0 equal?
when

2.

What
?n

are the roots of

?

m) x^ + 3 mx

-2 =

m=-

1 ?

when

EQUATIONS SOLVABLE BY QUADRATICS

3.

309

If possible, factor

3x-

+ 5xy can xq

2yy"^

4.
5.

For what values of

- 5x + 4y - 2. + mx + by 6 he
a and a and
/3,

factored

The

roots of x^

+ px +
a
in

being

express (a

/3)2,

a*

/3*,

and a/^

(3/

terms of

p and

g.

6.

The

roots of 2

x'^

3x

+
2

being

p, find the

values of

a/^'^
7.

i3/a:2

and a^^

a/S^.

The

roots of x-

+x+
jS
;

being
;

whose roots are

8.
1.

a,

l/a, 1/^3

2 a, 2 P;

a and j3, a+

find the equations

1, ^3

1.

Find the
3.

maximum and minimum

2. 5.

values of the following.

3. 6. l

a;2_8x +
x/{x2
9.

2x2

-X + 4.

+ 4x-x2.
+
1)/

4.

l).

l/x + l/(l-x).

(x

(2x2-1).
;

Find the greatest rectangle that can be inscribed

in

a given circle

also the rectangle of greatest perimeter.

man who is in a boat 2 miles from the nearest point on the 10. shore wishes to reach as quickly as he can a point on the shore distant If he can row 4 miles an hour and 6 miles from that nearest point.
walk 5 miles an hour, toward what point should he row
11.
?

What

height will a

body reach

if

thrown

vertically

the ground with an initial velocity of 48 feet per second,

it

upward from and when will

reach this height?

See

p. 303,

Ex. 21.

XV. EQUATIONS OF HIGHER

DEGREE WHICH CAN BE SOLVED BY MEANS OF QUADRATICS

Equations which can be factored.

Given an integral equation

643

we can resolve A into factors of the first or second degrees, we can find all the roots of ^ = by equating the several factors of A to zero and solving the resultis equivalent For ii A = BC , then ^ = ing equations. toB = 0, C = 0, , jointly, 341.
in the

form A

0.

If

Example

1.

Solve

x"*

By

436,

X*

+ +

x2

x2

+ +

= =

0.

(x2

1) (x2

1).

310
Hence
x*

A COLLEGE ALGEBRA
+
,

x^

=
.

is

equivalent to the two equations

1

x2

+X+

and

x2

0.

Solving these equations,

,.

= -l-i-iVS'liiVs or
5 x^

Example

2.

Solve x*

x^

12

0,

Factoring by the method of

a;4

451,

we

find that

x3 x^

Hence

x*

_ 5x2 - 7x + 12 = (X - 1) (X - 3) (a;2 + 3x + 4). 5 x^ 7 x + 12 = is equivalent to the three equations x-l = 0, x-3 = 0, and x^ + 3 x + 4 = 0,
1, 3,

whose roots are

and ( 3

V7)/2.

Example
^
1.

3.

Solve the following equations.

6x3-

11x2

+ 8x-2 =

0.

2.

X*

5x3
0,

x2

+ llx +

0.

644

function of
for

bu + Equations of the type au^ x. If the roots of au'^

a are a and
/8,

=
is

where u denotes some

c

-\-

bu

-\-

when

solved

this equation

equivalent to the two

equations u

a and u
au^
-{-

-\-

/?,

for, 631,

a(u ft). a) (ti Hence to solve au^ + bu + c =^ for x, we have only the two equations u = a and u = (3 for x.
Example
Hence
1.

bii

to solve

Solve 3x<
x2,

10x2

Solving for

x2

= 0. = 2/3

or

4.

x
2.

= V(3/3

or

2i.

Example

Solve x^
x',

Multiplying by
Solving for x^

Hence

+ 3 - lOx"^ = 0. x^ + 3 x^ - 10 = 0. x^ = 2 or 5. x = 2V2or 5i Vs. + 3 x + 4) (x2 +

3x)2
3x

Example

3.

Solve (x2

+ +

5)

= =

6.

We

may reduce

this eijuation to the

(x2

form
3x)
14
0.

+
+
1,

9 (x2

Solving for x2

3 x,
x2

we

obtain the two equations

3X

=2,

2,

and

x2

+
i

3x

=-

7,

whose roots are

and { 3

Vl9)/2.

EQUATIONS SOLVABLE BY QUADRATICS

Example
4.

311

Solve (x

1) (x

2) {x

3) (x

4)

120.

multiplying together the first and fourth factors, and the second and third, we reduce the equation to the form

By

(x2

+ 5x + +
5

4) (x-

+ 5x +

6)

120,
3.

which may be solved

in the

same way

as the equation in Ex.

Example

5.

Solve x*

10 x^

31 x^
first

30 x

+
5

=
0.

0.

By completing
Solving for x^

the square of the

(x2

two terms, we obtain

5 X)

x)'^

6 (x2

5x,
x2

we obtain

the two equations

1,

whose roots are (- 5

+ 5 X = 5 and x^ + 5 x = V5)/2 and ( 5 V2T)/2.

r2 4- 2 X x2

Example

6.

Solve 8

^ -

x2

+
1

-=:

11

= 0.

x2

2X

tiply

Observing that the second fraction is the reciprocal of the first, we mulboth members of the equation by the first fraction, thus obtaining
\

Solving for (x^

x^-1 J + 2x) /(x^

x2
x2

\x2-l/
1),
,

we
1

obtain the two equations

+ 2x =
1

^x2 and

x2

+ 2x -1

3
-,

v?hose roots are

1/2

and 3, 1/5.

All the values of x thus found are roots of the given equation since they cause none of its denominators to vanish.

Example
1.

7.

Solve the following equations.

18

3x*
(X

-29x2 +
a) (X

0.

2.

x<

-6x3 +
a*.

8x2

+ 3x =

2.

3.

2 a) (x

6)

3 a) (x

4 a)

24

4.

(4 x2

2 x)/ (x2

(x2

0)/

(2 x2

x)

0.

These are equations which remain Reciprocal equations. unchanged when we replace x by l/x and clear of fractions.
If

645

we arrange
x,

powers of

the

the terms of such an equation in descending first and last coefficients will be the same,

also the second

and next

to last,

and so on

or each of these

pairs of coefficients will have the

same absolute values but

contrary signs.

312
Thus,

A COLLEGE ALGEBRA
2x*

and

jc-2x*

+ 3x3 4-4x2 + 3x + 2 = + 4x3-4x2 + 2x-l =

are reciprocal equations.

to the quadratic

Reciprocal equations of the fourth degree may be reduced form and solved as follows.
1.

Example

Solve 2x*

3x3

4x2

- 3x +

0.

Grouping the terms which have like reduce the given equation to the form

coeflBcients

and dividing by

x^,

we

2(.
Since x^

4 = 0. i)-3(x + l) +
2,

/x-

(x

+ 1 /x)- +

we may reduce

this equation to the

form

2(x
Solving for X

^)-3(x + ^)
1

0.

1/x, we obtain the two equations

and x
(3

X
i,

13 =
2

-,

whose roots are

i,

and

V7)/4.

Every reciprocal equation of odd degree has the root 1 or 1 and if the corresponding factor a; 1 or x 4- i he sepaHence rated, the "depressed" equation will also be reciprocal reciprocal equations of the third and Jifth degrees can be solved
;

by aid of quadratics.
Example
2.

Solve 2 x'

3 x2

3x

0.

to

Grouping terms, Since both terms of the two equations

X

2 (x^

1)

3 (x-

4- x)

0.

this equation are divisible

by x

1, it is

equivalent

and

2-x2

5x

0,

whose roots are

1, and
Solve x^

2,

1/2.

Example

3.

Grouping terms,
Dividing by x

- 5 x* + x' - 9 x2 + 5 x - 1 = 0. 1) 5 x (x^ 1) + 9 x2 (x 1) = 0. (x^

to

whose

roots

1, we find that this equation is equivalent - 1 = and x* - 4 x* + 5x2 - 4 x + 1 = 0, are 1, and (1 i V3)/2, (3 V6)/2.
X

EQUATIONS SOLVABLE BY QUADRATICS

Example
1.

313

4.

Solve the following equations.

a;3-2a;2

+ 23.

=0.

2.

x* X^

4x^

x5

x-*

X^

+ 5x^ - 4x + = 0.

0.

Binomial equations. Tliis name is given to equations of the x" + a = 0. They can be solved by methods already a can be resolved into factors of the first or given when x"

646

form

second degrees.
Example
Since x^
to the
1.
1

Solve x^

0.
1),

(x

1) (x2

+x+ = =
32

the equation x^
x^

is

equivalent

two equations
X

and
1

Solving,

X
2.

or
0.

+ x + 1 = 0. {- 1 i V3)/2.
vfe obtain y^

Example

Solve x^
32

=
is

From x5 By 438,

=
1

0,

643, y^
2/

by setting x
1

= V32 y -2y,
?/3

0.

equivalent to the two equations

and
1,

2/4

+
j

2/2

2/

+ 1 = 0,
2

Solving,

2/

(-

V5 + Vio
z

V5)/4,

or

(-1
Hence
x

V5- Vl02 V5)/4.

1

2y=z2, (-

Vs + i ViO
i

V5)/2,

or

(-1
By

V5- ViO
2/"

2 V5)/2.

the method here employed every binomial equation x"

can

be reduced to the reciprocal form

0.

Example
1.

3.

Solve the following equations.

0. 2.

x3

X*

+1=

0.

3.

x6

0.

These examples
n nth
roots.

illustrate the

theorem
1 is

Everij nianher has

647

any number which satisfies the equation x^ = l\ and in Ex. 1, we found three such numbers, namely 1, ( 1 + i V3)/2, and ( 1 t V3)/2. Irrational equations. If asked to solve an irrational equation,
Thus, a cube root of
But, as will be ordinarily begin by rationalizing it, 609. illustrated below, certain equations admit of a simpler treat-

648

we

ment than

this.

Whatever method

is

used, care

must be taken
letter

to test the values obtained

for the

unknown

before

accepting them as roots of the given equation.

314
Example
1.

A COLLEGE ALGEBRA
Solve

Transposing,

Squaring and Squaring and

Solving,

V2x -3 Vox - + V8 x - 5 = 0. V2 x 3 + Vs x 5 = Vs x 6. simplifying, V(2 x 3) (3 x 5) = 1. 6x2 igx + 14 = 0. simplifying, X = 2 or 7/6.

(3

Testing these values of x in the given equation, we find that 2 but that 7/6 is not a root.

is

a root

We may also rationalize the given equation by the method of G03. We thus discover that 4 (6x'^ 19x + 14) is identically equal to (V2x-3 - V5x - 6 + V3x- 5)(V2x - 3 + Vox - 6 - Vs x - 5). (V2X-3 + Vsx - 6 + V3x- 5)(V2x-3 - Vsx -6 - VSx -5).
which the product 6x2 _ 19a; + 14 right vanishes for one of these Hence there is no value of x values, 2; the second, for the other, 7/6. for which the third or fourth factor cau vanisli.
Tliere are but

two values
first

of x for

can vanish.

The

factor on the

Example

2.

Solve

V4x +

+ V] 2 x +

= V24 x +

10.

Some equations which involve

to the

a single radical can be reduced

form of a quadratic with respect to this radical. then begin by solving for the radical.
Example
3.

We

Solve 2 x2

6x

Vx^

3x

-5 = 0.

Observing that the x terms outside the radical are twice those under the radical, we are led to write the equation in the form

Solving for Vx2

- 3 X - 1) - 5 Vx2 _ 3 X - 1 -3 = 0. 3x 1, we obtain the two equations Vx2 - 3 X - 1 = 3, and Vx2 _ 3x - 1 =- 1/2.

2 (x2
of these equations

The second
Talue.

must be rejected since according

to the

convention made in

579 a radical of the form

Va

cannot have a negative

Squaring the

first

equation,

we
5

obtain
0,

x2_3x-l =
whose roots are
Testing 5 and are roots.

and

2.

2 in the given ecjuation,

we

find that both of

them

Example

4.

Solve 2 x2

14 x

Vx2

10

+ 18=0.

EQUATIONS SOLVABLE BY QUADRATICS

315

Sometimes an equation may be reduced to a form in which, both members are perfect squares or one member is a perfect
square and the other
is

a constant.
x

Example

5.

Solve ix-

+ 2x Vsx^+x =
X

9.

We may
The
first

write this equation in the form

3 x2

2X

Vs x'^ +

x2

9.

member is a perfect square, and both members, we obtain the two equations

extracting the square root of

Va x2 +

Solving these equations, x

= =

3,
1,

and

V3 x- +

9/2, or (5

Testing these results,

given equation.

we

find that only 1

+ x = - 3. V97) /4. and 9/2 are

x

roots of the

Sometimes

all

the terms,

when properly grouped, have a

+
10 a^

common

irrational factor.
6.

Example
Here the

Solve

Vx^ -

ax

first

two terms and

also x

Vx^ + ax Ga^ = x - 2 a. 2 a have the factor Vx

is

2 a.

Separating this factor,

we

find that the equation

equivalent to
2 a.

Vx 2a =
we

and

Solving these equations,

Testing,

Vx 5 a Vx + 3a = Vx x = 2a, 10 a/3, or 6a.

2a and 10 a/3

find that only

are roots of the given

equation.

Example

7.

Solve

V3x2- 5x-

12

- V2x2-llx +

15

3.

If one or more of the terms of the equation are fractions with irrational denominators, it is often best to rationalize these denominators at the outset.

Example

8.

Solve

Rationalizing the denominator in the

- 5. (VJ + Vx-3)/{Vr - Vx 3) = 2x first member and simplifying,

Vx2

we have

3X

Solving,

= 2 X - 6. = 3 or 4.
of the given equation.
1

Testing,

we
9.

find that both 3

and 4 are roots

Example

Solve

Vx -

- Vx +

1)

Vx -

+ Vx +

1)

- 3.

316

A COLLEGE ALGEBRA
EXERCISE XLIV

Solve the following equations.

1.

4a;*
(x2

-17x2 +

18
9)

= =

0.

2.
4.

3x?(2x2
x<

4x^

=
x2

7.

3.
5.

4) (x2

7 x2.

-x-3)(3x2 + x -2)2=0
2 x3

+ -

x3

x2

3 X

0.

6.

0.

2X

= 0.

7.
8. 9.

(3x2-2x + l)(3x2- 2xx*

7)

12
0.

12 x3

4 X*
X*
X*

4 x3

+ 33 x2 + 18 X - 28 = - x2 - X - 2 = 0.
2 x2

10. 11.
12. 13.

2 x3
x3

x5

x5(1

15.
17. 18.

+ 36 x3 = 0. + x)3 = (1 - x)3.
11 X*

+ -

2 X2

X + 1 = + 1 = 0. - 36 x2 +
2

0.

11 X

-1=

0.

243

14.

(2x(X-

1)8

= 1.
0.

16.

2)*

-81 =

(a (a X2

+ x)3 + - x)* + 3X +

(6
(6
1

+ -

x)3
x)*

(a (a

=
x2

+ -

6
?))

+
(a

2 x)3.

2 x).

^^

4x2
20.

+ 6x-l
x2

_3

4x2

+ 6x-l _^^^^ + 3x + l

x2 + - = a2 + -l.
a2

21. 22.

3x2
4 x2

-2x-5V3x2-2x + 3 + 9 = 0.

24.

25.

- 2 X - 1 = V2 x2 - X. 23. V3 - x + V2 - x = Vs - 2 x. V2X + 3 + V3X-5 - Vx+1 - V4x-3 = 0. x2 - X + 1 X -J-Vx + Vx _ Vl - X = 1. 26. ?

Vx +
|2x
3

27.

^
30.

- Vx2 + 3x = 0. 6 /x - 2

28.

v^ -6Vi + 6-v^ =

0.

'^^-"^ = x-3. Vx - + Vx +
1 1

31.

V5x2 - 6x +

- V5x2 + 9x -

= 6x -1.

SIMULTANEOUS QUADRATIC EQUATIONS

32.

S17

^^^2 - VSx + 3 = V2X-1

(x

^
+

0.

33.

^ + ^^^2^ =
3.

2.

34.
35. 36.

(X

+ a)5 + (X + 6)' + (X + - 1) (X - 2) (X - 3) =
a)2

c)'
.

0.

(X

4 (X

rt)

Vx =

a-

4 a

\^.

XVT.

SIMULTANEOUS EQUATIONS WHICH CAN BE SOLVED BY MEANS OF QUADRATICS

A PAIR OF EQUATIONS IN X, Y, ONE OF THE FIRST DEGREE, THE OTHER OF THE SECOND

pair of simultaneous equations of the form

649

f(x,
cf)

(x,

= y) = y)

ax^
a'x

+ bxi/ + f/ + + b'y -f =
c'

dx

+ ei/ +/= 0,

may

be solved as in the following example.

Solve y2

Example.

x2

2x

+
X

2 y
2/

+ -

4
7
?/

= 0,
= 0. = 2x-7. = 0. = 13/3 or 3. = 5/3 or - 1.

(1)
(2)

From

(2),

(3) (4)
(5)

Substituting
Solvins: (4),

(3) in (1),

3 x^

22 x

39

X
(3), (1), (2)

Substituting in

(6)

The

solutions of

are the pairs of values

2/

13/3,

= 5/3;
(2);

3,

= -l.

(7)

For, 368, 371, the following pairs of equations are equivalent:

(1),

(2)

to (4),

(2);

(4),

(2)

to {bi

(5),

(2)

to (7).

We may

indicate the solutions (7) thus: 13/3,

to

5/3;

3,

1.
y.

Care

must always be taken

group corresponding values of x and

tions.

Ordinarily such a pair of equations will have tivo finite soluBut if the group of first-degree terms in ^ (x, y), namely

650

318

A COLLEGE ALGEBRA

a^x Vy, is a factor of the group of second-degree terms in f{x, y), namely ax^ bxy + cy-, while <^ {x, y) itself is not a factor of f(x, y), there will be only one finite solution or no

such solution.
be infinitely

And
Solve

if ^ (x,

y)

is

a factor of f(x,

y),

there will

many

solutions.
y^

Example

1.

x^

2x

+ 2y +
y

= 0,
0.
0.

(1)

Eliminating
If wt2

y,
0, (3)

(m2

mx = - 1) x2 + 2 (m + 1) x + 4 =
finite roots,

(2) (3)

5^

has two

and

But ify- mx is a factor of y- - x-, that and (3) does not have two finite roots.
Thus,
if

(1), (2)

two
1,

finite solutions.

is, if ??i

then

in^

-1=0
=
0,

m = 1,

(3)

reduces to x

=
if

0,

which has but one

1,

the other being infinite, 638.

And

m=-

finite root,

(3)

reduces to 4

both roots being infinite, 638. Hence if (2) has the form y - x = 0, the pair (1), (2) has but one finite And if (2) has the form y + x = 0, the solution, the other being infinite. pair (1), (2) has no finite solution, both solutions being infinite.
finite root,

which has no

Example

2.

Solve

y"^

Eliminating

y,

a;2

+ 2x + 2y = 0, y + x = 0. - x^ + 2 x - 2 x = 0.
x"^

(1)

(2)
(3)

But (3) is an identity and is satisfied by every value of x. Hence every pair of numbers x = a, y = - a is a solution of (1), (2). The reason for this result is that y + x is a factor ot y'^ - x- + 2x + 651

2y.

When

A, B,

AB ^
and

0,

This principle and 649 enable us to solve two integral equations when(x, y)= /(x, y) = 0, ever /(x, ?/) can be resolved into factors of the first or second
0,

5=

C= 0, C =

are integral functions, the pair of equations is equivalent to the two pairs .4=0, C

371.

cf>

degrees and
Example.

<^ (x,

y) into factors of the first degree.

Solve x^

+ xy- is

Hx

0,
0.

(1)

{2x-y){x + y-l) =
This pair of equations
X

(2)

equivalent to the four pairs

X
a;2

= =

0, 0, 0,

2/2-5

x2

-f- ?/2

= =

0,

= 0, - 1 = 0, 2x-2/ = 0, - 1 = 0. X -h
2X
2/

(3)
'

(4) (5)

2/

(6)

SIMULTANEOUS QUADRATIC EQUATIONS

Solving the pairs
(3), (4), (5), (6),

319
be

0,0;

0, 1;

1,2; -1,

-2;

2,

we find -1; -1,2.

the solutions of

(1), (2) to

pair of integral equations in x, y can be solved by means of quadratics only when it has one of the forms described in 649, 651 or when an equivalent pair which has one of these

652

forms can be derived from

it.

x + \ = 0^ y = x"^, Thus, the pair of equations of the second degree, ycannot be solved by quadratics. For there is no simpler method of x + 1 = 0, an equasolving this pair than to eliminate ?/, which gives x* tion of the fourth degree which cannot be solved by quadratics.
The preceding sections illustrate the truth of the following important theorem
:

653

A pair

of

inter/ral equations f (x,

y)

0, <^ (x,

y)

= 0,

whose

degrees are

and n

respectively,
xy-

has
x

mn
-

solutions.

Thus, the pair x^ or 6, solutions, 651.

-bx =

0, (2

y) (x

1)

has 3

2,

See

381 also.

It should be added, however, that if the groups of terms of highest degree in f{x, y) and <^(a;, y), but not f{x, y) and ^ {x, //) themselves, have a common factor, there are less than

654

mn finite
which
is

solutions.

common
<^ (.r,

to the groups of

Thus, for every factor of the first degree terms of highest degree in
is

/(r, y) and

y) there

at least

one

infinite solution

for

every such factor which is also common to the groups of terms of next highest degree there are at least two infinite solutions;

and so

on.

If /(x, y)

and

^(.r, y)

themselves have a

common

factor, there are infinitely

many

solutions.

- y2 _ i = o (2) cannot Thus, the pair x^ -xy"^ + xy - y"^ -y = (1), x2 have more than three finite solutions for there are 3 2, or 6, solutions all told, and at least one of these is infinite since x -\- y is a common factor of the groups of terms of highest degree in (1) and (2), namely X" xy'^ and x^ ?/-, and at least two others are infinite since x y is a common factor of the groups of highest and next higliest degree in (1), (2),

namely

x'

xy^,

x"^

y"^

and xy

y'^,

0{x

y).

320

A COLLEGE ALGEBRA
EXERCISE XLV

Solve the following pairs of equations.

i2x-Sy =
4.

'

b.

L.3x-5?/
3

'

2.

\3x + 6y =

l.

3x^-3xy -y"--4x-8y + 3x-y-S = 0.

fx^-

0,

5y"~

5.

-Sx-7y = 0,
0.

1 x + 3y =
fx^
-i

r2x^-xy-3y = 0, [jx-6y-i =
f2x + 3y =
8.
-J

+ 3xy + 2y"--l = + y = 0. rl/7/-3/x = l, n/xy -l/y-^ = l2..

x
.x^-

0,

37,

ll/x + l/y = U/i5.

rx^
'

l{3x
_^^

jj
(X

+ y^-8 = 0, + 1)'^ = (y -If.

+ xy + 2 = 0, + 7/){2x + y-\) = 0. ^x'^^xy~2y^- + y = 0, l{x-2y)(x + y-3) = 0.

y"^

13.

Determine

m so that the two solutions of the pair

m
x2

4iX

0,

= mx
14.

shall be equal.

Determine

and

so that both solutions of the pair

.cy

7/2 _|_ a;

0,

?/

vix

shall be infinite.
15.

By

the

method
2/2

of 650, Ex. 2,

show that 2x

7/

+ 4isa

factor

of 2 x2
16.

xy

10 X

12.

Show

that the

finite solutions,

p.iir xy = \, xy + x and that the pair x-y + xy

+ =

y
1
,

=
x^y

has not more than two

y2

2 has not

more

than four

finite solutions.

PAIRS OF EQUATIONS WHICH CAN BE SOLVED BY FACTORIZATION, ADDITION, OR SUBTRACTION

655

When both equations are linear with respect to some pair of functions of x and y. begin by solving the equations for this pair of functions by tlie methods of 374-376.

We

Example!.

Solve

2x2

82/2

=_

53,

(1)
^2)

3x2

y2

=111.

SIMULTANEOUS QUADRATIC EQUATIONS

Solving for
x^,
y"^,

321

we

obtain x"

whence,

= 25, = 5,
is

?/2

= 36, = G. =
6,

By
y

367-372, the pair (1), (2)

equivalent to the four pairs x

6: x
;

6;

x= 5,
2.

?/

6; x

5,

?/=

Hence the

solutions of

(1), (2)

are 5, 6

= 5, = 6. 5, 6 5, 6
?/
;

5,-6.

Example

Solve the following pairs.

.5x2

3/7/2=

120.

When
possible

when

This is always one of the equations can be factored. the equation in question has the form.
ax^

656

bx]/

cj/^

=i=

0,

and, in general,

when

it is

reducible to the form.

au-

+
a;

ft -f- c

=
2

0,

where u denotes a function of

Example
1.

x, y.

Solve x2

y2

_
5

^^

_
G

x2

xy

2/2

= ^
?/,

0,
0.

(1) (2)

Factoring

(2)

by solving

for x in terms of

x
or

X
Solving
(1),

= =
all

22/,

(3) (4)
(1), (2),

?/.

namely

4,

2;
2.

-2/5, - 1/5;
Solve 2 x2

(3)

and

(1),

(4),

we obtain
;

the solutions of

3, 1

-3/5, 2 y2
_|_

1/5.
3y
2/2

Example

+
:

4 xy

3^

_
+ -

2 5

3 x2

32

= =
= =

0,
0.
0.

(1)
(2)

We may
Solving,

write

(1)

thus

2 (x

2/)2

3 (x

y)

2
?/

x
x

or

+ +

]/2,
2.

(3)

2/

=-

(4)
(1), (2),

Solving

(2),

(3)

and

(2),

(4),
;

we obtain

all

the solutions of

namely

1,

/2

3/29, 23/58

3, 1

- 41 /29, x
X
1

17/29.

Example

3.

Solve the following pairs.

f

+ X2/ - 6 = 0, U2_5a; + 6 = 0.
r

x2

x'-'

x-y _ 26
x

2.

-y

'

2/2

-2x2 =

1.

322
657

A COLLEGE ALGEBRA
the given equations

When

may

be combined

by

addition or sub-

an equation which can be factored. This is always possible when both the given equations are of the
traction so as to yield

form ax^

+
1.

^xy

cy^

d.

Example

Solve

6 -c2

xy

?/2

5 x2

a;y

y2

= =

55, 49.

(1)

(2)

We combine
Multiply
Multiply Subtract
Solve
(5)

(1)

and
7,

(2)

so as to eliminate the constant terms.

(3) (4)
(5)

(1) (2)
(4)

by

42 x2

by 8, from

40 x2
(3),

for X,

or

- 7 x?/ - 14?/2 = 392. - 8 xy - 8 = 392. 2 x2 + xy - 6 2/2 = 0. x = 3 ?//2, X = - 2 y.

2/2

(6) (7)
(1), (2),

Solvmg

(2), (6)

and

(2),

(7),
;

we

obtain
2

all

the solutions of

namely 3 V35/IO,

V35/5

V2T/3, T V21/3.

And, in general, we obtain an equation which can be factored when the given equations are of the second degree, and can be combined by addition or subtraction so as to eliminate (1) all
terms of the second degree (2) all terms except those of the second degree (3) all terms which involve x (or y) or (4) all
; ;

terms which do not involve x (or

Example
2.

?/).

Solve 2 x2

4 xy

2 x

?/

0, 0.

(1)
(2)

3x2

Gx?y

-x +

3y =

(1)

Here we can eliminate all terms of the second degree by multiplying by 3, and (2) by 2, and subtracting. We thus obtain
4X

9y

0.

(3)
all

Solving

(2), (3),

we

obtain

3. 2;

2,

14/9, and these are

the

finite solutions of (1), (2).

See

654,
0,
0.

Example
Here

3.

Solve

- 3 xy + 2 ?/2 + 4 x 4- 3 y _ 1 = 2 x2 xy + y2 + 8 X + 2 y - 3 =
x2
r>

(1) (2)

the terms which involve x can be eliminated by multiplying We thus obtain (1) by 2 and then subtracting (2).
all

3y2
Solving
(1), (3),

+ 4y +
7

0.

(3)
(1), (2),

we

obtain

all

the four solutions of

namely
1.

1/3,

- l/3j -

lG/3,

1/3; (-

V57)/2,

SIMULTANEOUS QUADRATIC EQUATIONS

Consider the following example Example
4.

323

also.
0,
0.

Solve

x'^

xy

+
2/2

2y^

xy
Multiply
(2)

2/

+ 3x = +1 =

(1)
(2)

by 2 and add
{x

to (1).

We

obtain
(3)
(4)
(5)
(1),

2y)^

3{x

Solving
or

(3),

2y) + 2 x + 2y
x

= 0. = -l, + 2y = -2.
all

Solving

(2),

(4)

and
1

(2),
;

(5),

we
3

obtain
(1

the solutions of

(2),

namely

- 3
5.

V2,

T ^2

^5,

^^5) /2.

Example
r2x'^
^

Solve the following pairs of equations.

^

1x2

+ xy + 5y=0, + 2/2 10 = 0.
_,.

rx-^+y"--\S=0,
'

2/

Xxy + y - x = -l.
658

When
factored.

we

obtained by eliminating x or y can be pair of equations of the second degree can eliminate x or ij by the following method. The result-

the

equation

From any

ing equation will ordinarily be of the fourth degree and not solvable by means of quadratics. But if we can resolve it
into factors of the first or second degrees, so obtain the solutions of the given pair.

we can
^

solve

it

and

Example

1.

Solve

lOx^

-^

5y^

27 x

x2

y2

~ iy + - 3X 5

5
2/

0,
0.

(1)
(2)

First eliminate y^

by multiplying

(2)

by

and subtracting the

result

from

(1).

We

obtain
2/

5x2-12x +
Substituting
(3) in (2),

0,

or y

=- 5x2 + 12x-5.
6

(3)

5x*
Factoring, by 45l,
(X

-24x3 + 40x2 -27x +

1) (X

0.

(4)

x
2,

2) (5 x2

(9

9X

3)

= 0.

(5)
(6)

Solving

(5),

643,
?/

Substituting

(6) in (3),

= -

1, 2, 1,

V21)/10.

(7 dz

3 V2r)/10.

(7)
(1), (2).

The

pairs of corresponding values

2.

(6), (7)

are the solutions of

0,

Example

Solve

x2

x?/

2x2

X2/

+ 2 2/2 + 3 x - 3 y = - 2/2 + I - 2 + 3 =
2/

0.

324

A COLLEGE ALGEBRA
EXERCISE XLVI

Solve the following pairs of equations.

x^ +
7x2
2/2
j/2

Sy"-

-2
X2/

2/2

= Sl, = 10.
6
2 5

r36/x2
1^1/2/2

1/2/2

_4/j.2
i/2

= 18, = 8.

+ _
-

+ -

y
x2

= 0, = 0. =
0,
2/2

fx2+
13x2

17

-3x + 22/-39 = 0, = 0. xy + 10
2/2

14 X'

.2

4 xy

4X

2/

3.

7.

x2+5x2/-2x + 32/ + l = 0, { 3x2 + i5a;y _ 7x + 82/ + 4 = 0. = 98, x2 - 15 xy - 3 2/2 + 2 X + 9 -3 = -21. 5x2/ +

2/
2/"^

2/

^^

+ 3x2/ -42/2 = 25, + 24x2/ -31 2/2 = 200. -x2-3xy + 3 = x22/2, = 12x2^2. .7x2 -10x2/ + 4
2x2
15x2
2/2 2/2

(rx(x
'

3?/)

1x2 -52/2
.x2

= 18, = 4.
2/^ ?/2

^^

'

1x2
rx'-^

+ -

X2/
X2/

+ +

= =

38,
14.

^^

x2-x2/
X2/

2/2

21(x-2/),

^g

20.

?/p2 + y_8 = 0, 15 L yL?/2 + 15:X - 40 =

0.

PAIRS WHICH CAN BE SOLVED BY DIVISION

659
In solving a pair of equations
to combine
it is

sometimes advantageous
;

them by multiplication

or division

but care must

then be taken not to introduce extraneous solutions nor to lose actual ones (see 362, 342).

660

If given a pair of the

form

AB = CD

(1),

B=D

(2),

where

A, B, C,

by

= CD, B = D which is evidently equivalent to the two pairs A = C, B = D and D = 0, ^ = 0. We may obtain the pair A = C, B = D by dividing
D
each
if

denote integral functions of in (1), thus obtaining the pair AD

x, y,

we may

replace

member of (1) by the corresponding member of (2); but we then merely solve this pair A C, B = D,wg lose some

SIMULTANEOUS QUADRATIC EQUATIONS

325

of the solutions of (1), (2), except, of course, when either B or i) is a constant, so that the pair B has no solution. 0, D

Example
Dividing
Solving
2, 1

1.

Solve

x*

(1)

by

(2),

= x^ + xy + y'^ = x^ - xy + y^ = +
x-y-

y*

21,
7.

(1)

(2)
(3)
(1), (2),

3.

(2), (3),

we
2
;

obtain

all

the finite solutions of

2,

2.

1,

1,

namely

2.

See
x^

654.
2/^

Example
Dividing

Solve
x^

xy

(1)

by
(2)

(2),

= - 3 (x + = X + 1. x-y = -Zy.

1) y,

(1)
(2)

y'^

(3)

The pair (1), the pair x- -\- xy

has the same

finite solutions as

the pair

(2), (3)

and

+ 1 = (5) jointly. And the solutions of - 2/3, 1/3; - 1, (1 i V3)/2. (2), (3) and (4), (5) are 2,-1; = x, Example 3. Solve (x + y)" (1) X2_^2 = _6y. (2) + y) / {x - y) = - x/Q y. Dividing (1) by (2), (3) Clearing of fractions, x^ + 5 xy + y^ = o. (4) - 2 9/4, - 3/4. The pair (2), (4) has the four solutions 0, 0, 4, +
y-

(4),

l^x

The process by which

X,

(4)

was derived from

or 9/4,

(1), (2) is

reversible

when

y have the values

v^alues 0, 0.

3/4, but not when they have the Hence this reckoning only proves that 4,-2 and 9/4,

4,-2

(1), (2), 362. is a solution of (1), (2); but it obvious by inspection that 0, should be counted only once as a solution, not twice as in the case of Tliis follows from the fact that (1), (2) can have but three finite (2), (4).

-3/4

are solutions of

It is

solutions, 654.

It

may
x

also

tution y

tx (5).

We
yield

obtain

And

(5), (1'), (2')

0,

be shown thus In (1), (2) make the substi= x (1'), (1 - f^)x'^ = - 6tx (2'). (1 + ty^x"^ y = once and but once.
:

EXERCISE XLVn
rx3-r/'
1
'

63,

\x-y = S.
+ x22/2 + 2/4 = 931, 1x2 + x?/ + ?/2 = 49, f{x + y)2{x-y) = 3xy+6y, la;*--l/2 = x + 2.
rx*

p3+2/ 98, lVx + Vy =

r(x

2.

+
-

2/)(x2-2?/2)=_70,
2/)

\
^

(X

(x2

2/2)

14.

rx^

- 3xy +
-2/2-_

2y^
5y.

6z,

1x2

326

A COLLEGE ALGEBRA
SYMMETRIC PAIRS OF EQUATIONS

661

A pair of equations in x, y is said to be symmetric remains unchanged when x and y are interchanged.
Thus, the following pairs,
(a)

if it

and

(6),

are symmetric.
^^^
'
'

,^ r2x2

22/2
a;j/

IxV ^.

+ 3x + 3y = + 1 = 0.
of

0,

rx2
l2/2

= 2x + 32/, = 2 y + 3a;.
which

Symmetric pairs are

two types, those

like (a) in

the individual equations remain unchanged when x and y are interchanged, and those like ih) in which the two equations change places when x and y are interchanged.

662

Symmetric pairs
metric
equations

of the first type.

is
a-

The simplest

+ =
?/

a,

xy

b.

solved as in

649, but the following

is

pair of symThis pair may be a more symmetric

method.
Example.
Solve

Square

(1),

x^

Multiply Subtract

(2)
(4)

by 4, from

(3),

x"^

Hence
or

= 5, = 6. + 2 xy + = 25. 4 xy = 24. - 2xy + y^ = \. x y I, x-y = ~l.

x

?/

(1)
(2)
(3)

xy
y'^

(4)
(5).

(6) (7)

From
663

(1), (6),

3,

and from

(1), (7),

2,

3.

If given a

more complicated pair of symmetric equations,

and then solve
v.

we may transform each equation

and xy,

637,

for these functions

into an equation in x y or in the

;

given equations
solve for
ti

and

we may set x = k -\- v, y = u v and then The second method is essentially the same

as solving the given equations for x

=u

-{-

V,

y
1.

V,

we have

ti

+y = (x
a;

and x

y;

for, since

-\-

y) /2, v
2/

=(x

y) /2.
(1)

Example
In

Solve

x2
(1)

and

(2) for

2 x^ + 5 xy + 2 y^ + + +1= + 4xy + + i2x + 12y + 10 = x^ + y2 substitute (x + yY - 2xy.

?/2

0,

0.

(2)

SIMULTANEOUS QUADRATIC EQUATIONS

Collecting terms, 2{x
(X

327
(3)

2/)2

Eliminating xy,
Solving,

or

Hence, from
or

(3),

+ y)- + xy + (x + y) + 1 = 0, + 2xy + 12 {x + y) + 10 = 0. - 10 (x + - 8 = 0. S{x + y)^ x + y = i, x + y = -2/3. xy = - 37, xy = -U/9.

?/)
,

(4)
(6)

(6)
(7)

(8)
(9)

And
X,

solving the pairs

2/

(6), (8)

and

(7), (9)

for x, y,

we have

= 2ViT, 2tV41; (_12V3)/3, (-1t2V3)/3. x* + y* = 97, Example 2. Solve x + y = 5. x = u + v, y = u v. In (1) and (2) set - u)* = 97, We obtain (u + 1;)* + (u
and
Eliminating
Solving,
w,

(1)

(2)

(3)
(4)

2u- 5.
16 v*

600

1)2

151

= 0. = l/2
(6)

(5)

or

iVlEl/2.

(6)

Substituting

=:

V,

y
X,

=u =
2/

= 5 / 2 (4) and ~ V, we obtain

2,

the four values

of v in the formulas

3;

3,

2; (5

iVT51)/2, (5TivT51)/2.
/?

Evidently
pair,

if

=
a

a,

is

= p,

\^

another solution.
;

one solution of a symmetric Unless ex = ft, these

two solutions are different

y = sponding values oi X
values for x
a,

^
y,

as for

x = p, y ^ and namely a
-\-

but xy and x

+y
a,
ft

have the same

and the

corre-

a, differ

only

in sign.

Hence

the values of xy or x

y derived from a symmetric

pair will be less numerous than the values of x or y, that is the degree of the equation in xy or x -{- y derived from the
pair by elimination, as in Ex. 1, will be less than the degree As for of an equation in a; or y similarly derived would be. c must be the equation in ay, if e is one of its roots,

another root.

Hence

this

powers of x

~ y,

equation will involve only even

as in Ex. 2, or only

odd powers with no

constant term.

328
664

A COLLEGE ALGEBRA

The methods just given are applicable to pairs of equations Note. which are symmetric with respect to x and y or some other pair of functions of x and y. Tlius, x* + 2/* = a, x -y = b may be written X* + (- y)* = a, x + (- y) = b.

665

Symmetric pairs

of the second type.

Such a pair may some-

times be solved as follows.

Example.
Solve
x'^

y^

Adding

(1)

and
(2)

(2),

x^
(1),

Subtracting

from

x^

+ -

2/^

7/3

= lx + Sy, = Ty + 3x. = 10{x + y). = 4 (x - y).

(1)
(2) (3)
,

(4)

By
and

341, (3) is

equivalent to the two equations X + 2/ = 0,

X-

(5)

-xy +
X

y-

= = =

10.

(6)

Similarly (4)

is

equivalent to the two equations

2/

0,
4. (6), (8),

(7)
(8)

and

x"

+
(1

xy
;

y"

And
-2,

solving

(5), (7)

(5), (8)

((3)^(7)

we

obtain

0,

2,

Vl0, Vl0; (_l^Vr3)/2.

2;

Vl3)/2, {\^Vvd)/2; (-lVl3)/2,

EXERCISE XL VIII
Solve the following pairs of equations.

= 5, JX + \xy + 36 = 0.
2/

7.

10.

12.

14.

SIMULTANEOUS QUADRATIC EQUATIONS

329

SYSTEMS INVOLVING MORE THAN TWO UNKNOWN LETTERS

A system of three equations in three unknown letters can be solved by means of quadratics when one of the equations is of the second degree and the other two of the first degree also
;

666

when

possible to reduce the system to one or more equivalent systems each consisting of one equation of the second
it is

The like is true of a degree and the rest of the first degree. system of four equations in four unknown letters, and so on. If A, B, C are integral functions of degrees 7n, n,p in x, y, z,

and no two of them have a common factor, the equations will have vmp solutions. But some of :=0, B = 0, C

these solutions

may

be infinite.
z-

Example

1.

Solve

+ 3x -2?/ + 2z +

xy
a;

z 2

= = =

0, 0,
0.
z,

(1) (2) (3)

Solving

(2), (3)

for x

and y

in

terms of
x
2/

= -(4z + 2)/5, = (-2 + 2)/5. =

0.

(4) (5)

Substituting in (1) and simplifying,

7 22

2z

57
2

(6)

Solving

(6), (4), (5),

=-3

or 19/7.

Hence, from

x,

330

A COLLEGE ALGEBRA
EXERCISE L

Solve the following systems of equations by any of the methods of the

present chapter.

r7x^-6xy =

8,

rt

\2x-Sy = b.

"' ^

-y -a,
=
?/

Ic ,x-2/

3.
1.

xy
fX

(62-a2)/4.

+ = a2 +
i.

62,

2/

X
6.

+
a

=
+

+
6
-\-

6,

x^

5.

1x2

2/2^0.
1

1-1=12.
X2/
2/2

= 1.
a

^x

+ +

_1001
'

1-02

62

x2

7^^y^~ l25
1

2/2

17 = XJ?,
4

X2
8.

?/2

+
+

n
5

=2.
xy
^y-

rX2/

10.

a (X

2/)

& (

- 2/) =
0,

11.

40x?/
x2

= 21{x2-2/2) = 210(x +
3
X2/

2/).

1-4x2-25 2/2
12

2/2

=
2/2

9,

l2x2-10y2_32/ =
7

13.
4.

x2x/2/

13x2/

21

= -9.

14

j'a;2-

2/2-29

0,

lx2-6x2/ + 92/2-2x+62/ = 3
H'

15.

2 (x2
X2/2

x22/
6.

16, LX2/2 xy

1x22/

= X2/2 =
^^

a,
6.

2//X = 65/28, + 2/2) + (X - = 34. rx = a (x2 + 2/^), 12/ -6(x2 + 2/2)

2/)

-{
21
23.
x2

(x

2/)/(x-2/)

=
2/)

5/3,

(2x + 32/)(3-2

= 110a2.

r3(x3- 2/3) = LX - y = 1.
21 (x
2/)

13x2/,

'X* r^"

-r ?/*: </"
2/

"")

Ix +

22
X + = 21,
7.

a.

+ 2/2 =
x2

X2/

=
2/2
2/

2/.

24.

x2

+ = 10X2/, + + x2 + 2/2 _ 68. - X2/ + = 3 a2 = x2 2/

2/-

y\

25

+ X2/ +

{ X + Vxy + =
2/

26

4x2-32/2 = r4x2XT/ 1x2/ = 0.

J

12(x-2/),

27.

x2 + 2/2 = X + + 20, { X2/ + 10 = 2(x + 2/).

x2

28
2/2

+ 4x-32/ = 0, + lOx- 92/ = 0.

29.

28 (pfi r28(x6

Ix

+ y5) = =
:

61 {x3

+ y^),

30

- x/y = a, rx2/ \xy - y/x = 1/a.

SIMULTANEOUS QUADRATIC EQUATIONS

31.
(X

331

+ 1)3 + (y _ + = 2.
?/

2)3

19,

f 2/2

33.

- xy - yz = 3, + iy + z = U, x-2/ + 2z = 0.
x
{y
(z

34.

-l

+ = 8/3, + y2 ^34/9. (x + y + z + u = 0, 3x + z + M = 0, 32/ + 2z = 0, X2 + 2/2 + 2t{ = 5.

rx' x2
2/

35.

+ z)(x + + X) (X + l{x + y){x +

?y

+ z) = 10, + z) = 20, y + z) = 20.

y

36.

+ 2/2 + 22- 6, + yz + zx = If ^2x + y -2z = -3.

'X2

xy

EXERCISE
1.

LI
is

The

difference of the cubes of

is 8.

two numbers

218 and the cube cf

their difference

Find the numbers.

is

2. The square of the sum of two numbers less their product and the difference of their cubes is 189. What are they ? 3.

63,

is 11, and the product by one whose numerator and denominator exceed its numerator and denominator by 3 and 4 respectively is 2/3. Find the

The sum

of the terms of a certain fraction

of this fraction

fraction.
4. Separate 37 into three parts whose product is 1440 and such that the product of two of tliem exceeds three times the third by 12.
5.

The diagonal
than
it is,

of a rectangle

is

13 feet long.

If

each side were 2

it is.

feet longer

the area would be 38 square feet greater than

What

are the sides ?

6. The perimeter of a right-angled triangle is 36 inches long and the area of the triangle is 54 square inches. Find the lengths of the sides.

7.

The hypotenuse

of a right-angled triangle

is

perpendicular sides by 3 and 24 inches respectively.

the triangle.

longer than the two Find the sides of

8. Find the dimensions of a room from the following data its floor a rectangle whose area is 224 square feet, and the areas of two of its side walls are 126 and 144 square feet respectively.
:

is

rectangle is surrounded by a border whose width is 5 inches. of the rectangle is 168 square inches, that of the border 360 square inches. Find the length and breadth of the rectangle.
9.

The area

332
10.

A COLLEGE ALGEBRA
In buying coal A gets 3 tons more for $135 than B does and pays than B pays for 5, Required the price each pays

$7

less for 4 tons

per ton.
certain principal at a certain rate amounts to $1248 in one year Were the principal $100 greater and the rate li times as great, the amount at the end of 2 years would be $1456. What is the principal and what is the rate ?
11.

at simple interest.

12.
all.

A man

The

children receive

leaves $60,000 to his children and grandchildren, seven in of it, .which is $2000 more apiece than the J

grandchildren get.
children,
13.
less

How many

children are there and

?

how many grand-

and what does each receive

his usual rate a
it

At

man can row

15 miles downstream in 5 hours

rate, his

time than

takes him to return.

Could he double his

time

downstream would be only 1 hour less than his time upstream. is his usual rate in dead water and what is the rate of the current
14.

What
?

Three men A, B, C together can do a piece of work in \ hour, 20 minutes. To do the work alone it would take C twice as long as A and 2 hours longer than B. How long would it take each man to do the

work alone ?
15.

Two

bodies

and

are

moving

at constant rates

direction around the circumference of a circle

makes one

circuit in 2 seconds less

and in the same whose length is 20 feet. time than B, and A and B are

together once every minute.

16.

What

are their rates ?

On two straight lines which meet at right angles at the points A and B are moving toward at constant rates. A is now 28 inches and B 9 inches 2 seconds hence A and B will be 13 inches apart, from
;

and 3 seconds hence they and B moving ?

17.

will

be 6 inches apart.

At what

rates are

Three men A, B, and C set out at the same time to walk a cei'tain A walks 4| miles an hour and finishes the journey 2 hours before B. B walks 1 mile an hour faster than C and finishes the journey
distance.
in 3 hours less time.

What

is

the distance ?
start sinniltanoously

tively

from P and Q respectoward each other. When they meet A has traveled 12 miles farther than B. After their meeting A continues toward Q at
18.

Two

couriers

A and B

and

travel

the

same

rate as before, arriving in i- hours.

in 7^ hours after the meeting.

What

is

Similarly B arrives at the distance from P to Q?

SIMULTANEOUS QUADRATIC EQUATIONS

GRAPHS OF EQUATIONS OF THE SECOND DEGREE IN
Examples
of such graphs.

333

X,

Y
667

of the second degree in x,

The graph of any given equation y may be obtained by the method

^x.
2

illustrated in the following examples.

Example

1.

Find the graph of

?/,

y2/

(1)

Solving for

Vx.

(2)

negative, y is imaginary ; when x is when x is positive, y has two real values which are equal numeri0, t/ is Hence the graph of (1) lies entirely to the cally but of opposite signs. right of the y-axis, passes through the origin, and is symmetric with respect to the x-axis.
(2) it

From
;

follows that

when x

is

When
we have

x=:0, 1/4,
?/

1/2,

1,

2,

3,

4,...,

0,

1, V2, 2,

V2,

Vs, 4,..-.

We

solutions

obtain the part of the graph given in the figure by plotting these 1), and passing a curve through the (0, 0), (}, 1), (],

points thus found.

Com-

pare

389.

It

touches the
called a

i/-axis.

This curve
parabola.
It

is

consists of
to

one "infinite branch, "here

extending
die right.
indefinitely

points
2/2

Example 2. In what is the graph of = 4x (1) met by the

graphs of y

?/=x
1.

l{3),

2/

= x Z (2), = x + 3(4)?
(1),
;

The
1,
(1,

solutions of

(2)

are

points
2.

- 2 9, G. 2), (9, 6),

(3)

Hence,
as
is

386, the

graphs of

(1), (2)

intersect in the

shown

in the preceding figure.

Tlie solutions

of (1), (3) are

equal,

namely

1,

2;

1,

2.

Hence

meets the graph of (1) in two coincident points at (1, 2). Tins means that the graph of (3) touches the graph of (1) at (1, 2), as is
the graph of

indicated in the figure. 3. The solutions of

2

V2

i.

l2V2i, (1), (4) are imaginary, namely Hence, as the figure shows, the graphs of (1), (4) do not meet.

334
Example
3.

A COLLEGE ALGEBRA
Find the graph
2/2-2x2/
of
2/

Solving for

y,

we have

+ 2x2-6x + 2 + = X 1 Vi X ?/

0.

(1)
(2)

x2.

The values
x(4
or

Hence the graph of (1) lies between the lines x = and x = 4. X = and when x = 4 the values of y given by (2) are equal: namely 1,-1 when x = 0, and 3, 3, when x = 4. Tliis means that the
4).

x), is

of y given by (2) are real when the radicand 4 x x^, or and 4 (or is positive (or 0), that is, when x Ues between

When

graph of
the point
line

(1)
(0,

tmdies the

line

at

1)

and the
See Ex.

line
2, 2.

=4
The

at the ptiint

xpoints of tangency, for when 4x vanishes, (2) gives the same values for
y that y = x \ gives. For each value of x between and 4 the equation (2) gives two real and distinct values of ?/, obtained by increasing and diminishing the value
of X

(4, 3).

passes through these

by that

of

V4 x

x^.

Hence

for each of these values of x there are

the line y = x \ the value of x in question by

two points of the graph of (1). They are most readily obtained by drawing and then increasing and diminis hing its ordinate for
tlie

value of
1,

v4 x
2,
1,

x^.

Thus, when

=
=-

0,
1, 1,

3, 4,
2, 3,
3.

we have
and

for the line

?/=
?/

0,

for the graph of (1)

figure

V3,

2,

V3,

shows the oval-shaped curve which the points thus found It is called an ellipse. determine. By solving (1) for x and applying the method of 641, we may show that the highest and lowest points of the curve are (2 + V 2, 1 + 2 V 2)

The

and

(2

V2,
4.

2 a^).
?/-

Example

Find the graph of

x-

+ 2x +

2 y

0.

(1)

Solving for y and factoring the radicand in the result,

?/

=-

V(x + l)(x-3).

(2)

The radicand (x + 1) (x 3) vanishes when x 1 and when x = 3, and in both cases (2) gives equal values of ?/, namely 1, 1. This

SIMULTANEOUS QUADRATIC EQUATIONS

means that the graph of (1) touches the line x and the line x = 3 at the point (3, 1). Tlie
these points of tangency.

335

=
line

1 at

the point
1

(1,

1)

passes through

3) is positive when and only when x < 1 (x + 1) (x For every such value of x the equation (2) gives two real and distinct values of y and therefore two points of the graph of (1). These points may be obtained by drawing the line y 1 and then increasing and diminishing its conThe radicand
or X

>

3.

stant ordinate

by the value of
indicated
(1)

V(x

l)(x Hence, as

3).
is

in

the

figure, the

graph of

consists of

two

infinite branches, the one touch 1 and extending ing the line x = indefinitely to the left, the other

touching the line x

3 and extend-

ing indefinitely to the right. This curve is called an hyperbola. There are two straight lines called
asymptotes,

which the infinite

this

branches of

touch at

infinity.
x,

and y

hyperbola tend to touch, and which they are said to These lines are the graphs of the equations y = x 2 which we obtain as follows. Compare 650, Ex. 1.
(1)

Eliminating y between

and the equation y

we

obtain

(m"^

1) x^

2 (vie

?/i

1)

(c-

= mx + c, 2 c + 4) =
0,
0.

(3)
0.

(4)

Both roots of
if

(4)

are infinite, 638,

mis, if

that

m=
y

1,

1= = 2,
or

and mc
or
if

+m+1= m = 1, c =
if (3)

Hence both solutions

of (1), (3) are infinite

has one of the forms

(3")
(1) in

= x-2

(30
(3')

=-

x.

Therefore the graphs of

and

(3")

each meet the graph of

two

infinitely distant coincident points.

Example
that

5.

Solving for y,
is,

Find the graph of we have ?/ = 2 x

?/

?/2

3 X or

?/

4 xy + 3 x2 + 6 x Vx^ 2 x + 1, = X + 2.

?/

0.

(1)

(2)

Hence the graph

of (1) consists of the pair of right lines y

= 3x

and

=X+

2.

tion (2) gives

1 = 0, the equaExcept when the radicand vanishes, that is, when x it two real and distinct values of y. But when x 1 =

336
gives

A COLLEGE ALGEBRA
two equal values
of y,

namely

3,

3.

Hence the

line

meets the graph of (1) in two coincident points at (1, 3). cannot mean that the line x - 1 = touches the graph of means that the points coincide in which the line x 1 =
lines,
2/

x 1 Of course
meets
tlie

=
this
It

(1) at (1, 3).

= 3x
6.

and y

2,

two which
(1).

together constitute the graph of

Example

Find the graphs, and

their mtersections, of
x2 + ?/ ^ 25, XV16 + 2/V9-2.
(1>

(2)

The graph
center
is is

at
5.

radius
ellipse.

whose the origin, 0, and whose Tlie graph of (2) is an

of (1)
is

circle

These curves intersect at the four

(4,

3),

points (4, 3), (-4, these points being the graphs of the solutions of
7.

3),
(1),

(-4, -3),
(2).

Example

Find the graphs, and their intersections, of

xy xy

y
2/

From

(1)

we

obtain

- 2 = 0, + 3 = 0. y = 2/ (x -

(1)

(2)
3).

(3)

Here there

is

one real value of

y,

and hence one point

2, 3,

of the graph,

for each real value of x.

When

= -3o, -1,

0,

1,

2|,

_ t, - |, _ l, - 2, - 8, oo, 8, 2, 0. we have y = 0, And plotting these solutions, we obtain an hyperbola whose two

3^, 4, oo,

infinite

branches are indicated by the unbroken curved lines in the figure, and whose asymptotes are y = (found as in Ex. 4), and x 3 = (since,

when X =

3,

then y

(x>).

In a similar manner, we find the graph of (2) to be the hyperbola indi-

cated by

tlie

the asymptotes y

dotted curve and having = and x + 2 = 0.

(1),

The equations

(2)

have the

1, ?/ 1, single finite solution x the remaining three solutions being infinite, 654.

The hyperbolas which are the graphs of (1), (2) meet in the single

finite

point

(1,

1).

But since they

SIMULTANEOUS QUADRATIC EQUATIONS

have the

337

common asymptote y =

0,

they are regarded as having two

infinitely distant coincident intersections at (ac, 0) ; and since they have o 2 =: 0, they are regarded as the parallel asymptotes x and x

having one infinitely distant intersection at

(0, oo).

General discussion of such graphs.

obtained in the preceding examples,

conclusions.

Generalizing the results we reach the following

668

Suppose any equation of the second degree in with real coefficients, as

ax^
If b
is

x,

y given,
'

+ 2 hxy +
by

bif

2 gx
y,

2fy

+c=

0.

(1)

not

0,

and we solve for

we have
(2)
be).

where

R = (A^ _

= -(hx+f)VR, ab) x^ + 2 (/if ~bfj)x+ (/^

(2) we obtain two real values of y for each value of x which the radicand R is j)ositii)e. Corresponding to these two values of y there are two points of the graph which may be found by drawing the line

From

for

by=-(JiX-^f)
and then increasing and decreasing its ordinate for the value of X in question by the value of '\li /b. See 667, Exs. 1, 3, 4.

The form
factors of R.
1.

of the graph depends on the character of the

When

(hf

byy
is

- (h^ -

ab)

(f ~

be)

0.

a perfect square, 635, and the first member of (1) can be resolved into factors of the first degree, If these factors have real coefficients, the graph 635, Ex. 3.

In this case

i?

of (1)
2.

is

a pair of right lines.

See

667, Ex.

5.

When

(hf

In this case,

reduced to the
a and
^8

- (h^ - ab) (f^ - be) > 0. bg)^ ab = 0, the radicand R can be unless form R = (h^ ab) (x a) (x /S) (3) where
h"^

are real and

a<

/3,

635.

338
If h^

A COLLEGE ALGEBRA

ab

<

0,

the product (3)

is

positive

when and only

when X

between a and /3. Hence the graph of (1) will be and x ^ = 0, a closed curve lying between the lines x a = See It is therefore an ellipse (or circle). it touches. which
lies

667, Ex. 3. ^j If 7^2

> 0^
or

when

x<a

x>

the product (3) is positive when and only Hence the graph will consist of two ft.

infinite branches, the

one touching the line x

fi extending to its left, the other touching the line x and extending to its right. It is therefore an hyperbola. See

and

667, Ex.
If

4.

W -ab = 0, we have
0,

R=

2 (hf -bg)x+ (/^

- be), where

x:> (f^ be) /2(hf bff). of one infinite branch lying entirely to one side of the line It is therefore 2 (A/ hff)x + {f^ be) = 0, which it touches.
2i

Jif-^hgi^

and

this

is

positive

when and only when we have Hence the graph will consist

parabola.
3.

See

667, Ex.

1.

- (Ir - ab) (f^ - be) < 0. brj)^ are conjugate imaginaries, In this case the roots of 72 = we can reduce R 635, and if we call them X + fxl and X to the form R = {h'' - ab) [(a; - X)^ + /x^], (4). If h^ ab>0, the product (4) is positive for all values of x.
When

(hf

ixi,

Hence
which

lie

1. therefore an hi/perbola. Thus, v/^ x" If h^ ab< 0, the product (4) is negative for all values of X. Hence the graph of (1) will be wholly imaginary. Thus,

the graph of (1) will consist of two infinite branches It is on opposite sides of the line &y [hx +/)

x^

+f+

0.

In the preceding discussion it is assumed that b^O. Biit it b = 0, while a ^ 0, and we solve (1) for x instead of y, we and b = 0, the If both a = arrive at similar conclusions.
graph of
(1) is

straight lines of
to the y-axis.

an hyperbola, as in 667, Ex. 7, or a pair of which one is parallel to the a;-axis, the other

SIMULTANEOUS QUADRATIC EQUATIONS

EXERCISE
LII

339

Find the graphs of the following equations.

1.

y2
x2
2/2

__ + -

8 X.

2.

x2

_
2/2

9.

3.

(y

_ +
0.

a;)2

= = =

x. 0.

4.
6. 8.

2 X2/

2/2

= 8.

5.

?/2

_ -

4 ^y
2 xy
x"^

2x2
2 x2

10. 11.

2 xy + 1 = 0. + 3 2/2-4x + 62/ = 0. + 4 X2/ + 4 2/2 + X + 4 ?/

7.
9.

2/2

y"^

1 = - 3x +

+ -

3 x2

4x

0.

=
0.

0.

4x2-12x2/4-

92/2

+ 3x -6y =

Find the graphs of the following pairs of equations and their points of
intersection.

12.

fxy
^

l, '

Ux-5?/ =
r 2/2

13.
2.

X2/

2 x2

2 X

rx2- 2/2 = 1, {x'-xy + x = - 2 - 2 = 0,

-^

14.
0.

fx2

2/2

3,

ly^=2x.

2/

L2/2-X2/-2x2 +
^g
17.

0.

+ |(x-2 2/)(x t (X - 2 (X
2/)

2/)

2/)

Find the graph

+ x-32/ = 0, + 2 X - 6 = 0. of x2 + 2/2 6x
2/

22/

+ l=0 and +
y)

its

points of

intersection with the axes of

18.

x and

y.

Show
axes.

that the graph of {x

y)-

2{x

touches the x

and y
19.

Show

that the line

2/

= 3 x + 5 touches the

graph of 16 x2 + 2/2 16 =
3 touch the graph of

at the point
20.

(-3/5,

16/5).

For what values of

will the line

= mx +

x2

2 y2

6 ?
c will

21. For what values of graph of 3 x2 - 2/2 + X = ? 22.

of the

the line

7x iy +
1 =:

touch the

Show

that the lines

2/

graph oi xy

2 y^

= +

and x
6

2y +

are the asymptotes

0.

23.

Find the asymptotes of the graph of the equation 2x2 + 3x2/-2 2/2 + x + 22/ + 2 = 0.

24.

a parabola

For what values of \ ? an hyperbola ?

is

the graph of x2

\xy

2/^

x an ellipse ?

340

A COLLEGE ALGEBRA
XVII.

INEQUALITIES
absolute

669
"

Single

inequalities.

An

inequality
all
is

is

one

like

y''

>
;

which holds good for

real values of the

letters involved

a conditional inequality

one like

cc

>

670

which does not hold good for all real values of these letters, but, on the contrary, imposes a restriction upon them. The principles which control the reckoning with inequalities
are given in or sign

261.

From

these principles

it

follows that the

connecting the two members of an inequality will remain unchanged if we transpose a term, with its sign
changed, from one member to the other, or if we multiply both members by the sa^me positive number but that the sign > will
;

>

<

be changed to <, or vice versa, by the same negative number.

if

we multiply both members

Example

1.

Prove that

We have
that
is,

a2

and therefore

+ 6^ > 2 ah. - 6)2 > 0, - 2 a6 + 62 > 0, cfi + 62 > 2 ah.

a^
(a
a"^ -\-

Example

2.

Prove that
a2

h62

-\-

d^

We have

52

> 2 ah,

c2

> ah > 2 6c,

62

-\-

he
c2

+ ca. + a2 > 2 ca.

and
6c

Adding the corresponding members

dividing the result by
2,

of these three inequalities

we have

a'^

Example
restriction

3.

Solve the inequality

+ c2 > a6 + 3x + 5>x + 11, +

> 6, > 3. < 0.

ca.
is,

that

find

what

it

imposes on the value of

x.

Transposing terms,

2x

whence

x
4.

Example

Solve
(x

x2

2 x

- 3 < 0.
3)

Factoring,

1) (x

To

satisfy this inequality

negative.

one factor must be positive and the other Hence we must have x > 1 and < 3, that is, 1 < x < 3.

671

Simultaneous inequalities.
qualities of the

A
hy

system of one or more

ine-

form

ax

+c>

INEQUALITIES
may
be solved for the variables
is

341

method which

x, y by a simple graphical based upon the following consideration

:

Draw

385.

the straight line which is the graph of ax + bi/ -\- c = 0, Then for all i)airs of values of x, y whose graphs lie

on one side of this line we shall have ax -\- by -\- c> 0, and for all pairs whose graphs lie on the other side of the line we shall have ax -\- by -\- c < 0.
Thus,
?/i

let

{xu
c)

l/i)

he a point on the graph oi y

{mx +

c)

so that

(inxi

= 0.

Then,

if

^2
c)

the line, we have y^ {mxi lies above the line, we have

+
7/3

< 2/i so < 0, and

{mxi

tliat
if 7/3

=
A:2

the point (xi, 2/2) lies below > 2/1 so that the point (xi, 2/3)

-f c)

> 0.

Example.
ki

Solve the simultaneous inequalities

fco

Find the graphs of

= X - 2y + K0, ki =
A-2

+ y-5<0,
0, ks

k3

= 2x -

1>0.

0,

0,

as indicated in the figure.

The

inequality

<

is

satisfied

by those pairs of values of x, y whose graphs lie on the side of the line toward the origin for when ^2 = X = 0, 2/ = 0, we have ko = 5, that
;

< 0. It may be shown in a similar manner that the inequalities ki<0 and kz>0 are satisfied by the paira of values of x, y whose graphs lie on
is

the sides of the lines ti remote from the origin.

^1

0,

/cs

A>

Therefore the given inequalities

< 0, ko< 0,

ks

>

are satisfied

the pairs of values of x, by the three lines.

by y whose graphs

lie

within the triangle formed

EXERCISE
imequal positive numbers.
1.

LIII
a, b, c

In the following examples the letters

are supposed to denota

Prove that a/f)

+ &/n> 2.
b) (a^

2.
3. 4.

Prove that

(a

b"^)

>

(a2

+
+

62)2.

Prove that a^
Prove that

+
+

6^

>

a'^b

062.

a'^b

b"a

b^-c

d^b

d^a

+ a^o 6 abc.

342
5.
6.

A COLLEGE ALGEBRA
Prove that
Solve x
a'^ + b^ + c^ > 3 abc. + 7>3x/2 - 8.

7.
8. 9.

Solve2x2
Solve (X
Solve

4x>x-^

1) (X

3) (x

+ 6x + 8. - 6) > 0.
X

?/-x 2<0,

3<0,

+ l>Oby

the graphical method.

10.
11.

Also y Also x

x>0, +
2/

2x<0.
2/

+ 3>0, 2/-2x-4<0,

+ 2x + 4>0.
all

12. 13.

Prove that x^
Solve x2

2/2

+ 2x + _ 1 < 0,

>

is

true for

values of x.

2/2

4X

<

by a graphical method.

XVIII.

INDETERMINATE EQUATIONS OF THE FIRST DEGREE

Given any equation of the

673

Single equations in two variables.

form

ax
a, b, c

-\-

01/

where
factor.

We

denote integers, of which a and h have no common seek an expression for all pairs of integral values

of X and y which will satisfy this equation; also such of these pairs as are positive as well as integral.

673

Theorem

1.

All equations ax

by

of the kind just

described have integral solutions.

prime to one another, by the method explained two integers p and q, positive or negative, such that ap -{ hq \ and therefore a (pc) + b (qc) = c, and this proves that x = pc, y = qc is & solution of az + by =: c.
h are in

For since a and

491

we can

find

674

Theorem 2. If x such an equation ax

=
-f-

x,

= y^
c,

be one integral solution of

by

all

of

its

integral solutions are

given by the formulas

Xo

+ bt,

= yo -

at
to

when

all possible integral values are assigned to

mDETERMINATE EQUATIONS
First,

343

Xo

bt,

yo

at

is

always a solution of ax

hy

c.

(1)

For, substituting in
or, simplifying,

(1),

a (Xo

-\-bt)

b {yo

oxq

= + byo =
at)

c, c,

which

true since, by hypothesis, x = Xo, y = yo is a, solution of (1). Second, every integral solution of (1) is given hy x = xq + bt, y = yo
is

at.

For

let

Xi,

yi denote any second integral solution.

Then

axi

byi

and axo
b{yi
is

whence, subtracting,

+ byo = c, yo) =

a{xi

Xo).

(2)

From
and
Xi

(2) it

follows that 6

a factor of the product of the integers

is

Xo.

Therefore, since b

prime
call

tained in Xi

Xo, 492, 1,

and
Xo

if

we

must be exactly conthe quotient t', we have

to a,
it

Xi

= W, =

or Xi

Xo

bt'.

(3)

And

substituting (3) in

(2)
2/1

and simplifying, we
2/0at'.

also

have
(4)

673, 674 it follows that every equation ax + by = c of the kind just described has infinitely many integral soluWhen a and b have contrary signs there are also tions. but when a, and b have the infinitely many ^:(os/f iye solutions

From

675

same sign there

no such solution.

is

but a limited number of such solutions or

Thus, one solution of 2 x

3y

18

is

3,

2/

4.

Hence the general solution isx = 3 + 3i, y=:4 2<. The positive solutions correspond to ^ = 1, 0, 1, 2 and are
3,

x,

1/

0,

6, 2

9, 0.

The theorem

of

674 enables one to write down the general

676

integral solution of an equation of the kind under consideration as soon as a particular solution is known. particular solution may often be found by inspection. Thus, one solution

= 4. A particular solution may a; + 3 ?/ = 12 is a; = 0, y always be found by the method indicated in 673 also by the method illustrated in the following example.
of 10
;

Example.

Find the integral solutions of

19

2/

213.
x,

(1)

Solving for the variable with the smaller coefficient, here

and reducing,
(2)

we have
x

= ^li:ii^ = 30-22/ +

'-^.

344
Hence
integer.
if

A COLLEGE ALGEBRA
x is to be an integer when y is one, (3 5 2/) / 7 = u. Call this integer u, so that (3
5

5?/)/ 7 must be an

Then
Treating (3) as

7 u

3.

(3)

we have

just treated

(1),

we have
...

3-7m ^5
Set (3

3-2u
5
to
v.

m)/ 5, which must be integral, equal

2

Then
Treating
(5)

5u

3.

(5)

as

we have already
M

treated (1)

and

(3),

we have
(o)

=
2

=:l-2v + ^-
2

When v-\
integral value

the fractional term

(1-d)/2

vanishes and u has the

1.

Substituting u

=-

1 in (4),

we

obtain y

2.

= 25. Substituting y = 2 in (2), we obtain x Hence the general integral solution of (1) is
x

25

19,

2/

= 2-7*.
to
<

respectively,

There are two positive solutions corresponding namely x, y = 6, 9 25, 2.

:
;

=-

and

Observe that in the fractional terms of

of the coefficients of y, m, v,

(2), (4), (6)

1,

the numerical values

are merely the successive remainders occurring in the process of finding the greatest common divisor of the given coefficients 7 and 19. We finally obtain the remainder,

namely

5, 2,

The like will factor. 1, because 7 and 19 have no common be true if we apply the method to any equation ax + by = c in which a and b have no common factor. Hence the method will always yield a solution of such an equation. it is seldom necessary to complete the reckoning above But in
or coefficient,
practice
indicated.
will

make

(3

= Thus, having obtained (4), we might have observed that u - 2 u) /5 integral, which would have at once given us y
=
25,

1 =2

and therefore x

by

(2).

677

Observe that an equation ax + bi/ = c with integral coefficients of which a and b have a common factor, as d, can have no integral solution unless d is also a factor of c. For if x and ax + by and therefore y were integers, d would be a factor of of c. Thus, 4 X + 6 ?/ = 7 has no integral solution.

INDETERMINATE EQUATIONS
Simultaneous equations.
trate a

345
will
illiis-

The following example

678

method

for finding the integral solutions, if there be

any, of a pair of simultaneous equations in three variables

with integral
Example.

coefficients.
of
22, 28.
(!)
(2)

Find the integral solutions

3x + 6y -2z = 5x + 8y -(iz =
First eliminate z

and simplify the resulting equation.

2
a;

We

obtain

5y

=
=

19.

(3)

Next

find the general solution of (3), as in 676.

"We obtain

x
(4) in (1)

=7+

^t, y

2t.

(4)

Next substitute

and simplify the

result.
(5)

We We

obtain

2z - St =

5.

Next

find the general solution of (5).

obtain

=1-

Su,

=-

2u,

(6)

where u denotes any integer whatsoever.

Finally substitute
t

We
which

obtain
is

= I 2u in (4) and simplify. - 10 m, y = 3 + 4 u, z = 1 - 3 u,

(7)

The
y

the general solution required. only positive solution is that corresponding to m

0,

namely x

2,

S,z

1.

tion

Observe that the given equations will have no integral soluif either of the derived equations in two variables has

none,

677.
in a similar

We proceed
example
integral

manner

if

given three equations in

fonr variables, and so on. Single equations in more than two

illustrates a

variables.

The following

679

method
a

solutions

of

of obtaining formulas for the single equation in more than two

variables with integral coefficients.

Example.

Find the integral solutions

x

of

5x + Sy + 192 =
5

50.

(1)

Solving for X,
Set
(3

10

-y-Sz-

^11^.

(2)

4 z)/5,

which must be

integral, equal to u.

346
Then
Solving for y,
Set (2 m

A COLLEGE ALGEBRA
3y

42

= 5 w. =u-

(3)

-\

(4)
u.

z)/3, which must be integral, equal to

Then
Substituting Substituting
(5) in (4),

z-2u-Zv.
y

(5)
(6)

= =
10

-{

Av.

(5)

and

(6) in (2),

- 6u +
u, v

V.

(7)

The formulas

(5), (6), (7), in

which

may have any

integral values

whatsoever, constitute the general solution required. Substituting u = 2, v = \ in the formulas (5), (6),
positive solution of (1),

(7),

we

obtain a

namely x

3,

2, z

\.

EXERCISE LIV
Find the general integral solutions of the following
integral solutions.
1.
;

also the positive

3.
5.

6x-17 = 18. 16x + 39?/ = l.

2/

2. 4. 6.

9.

49x-27y = 28. 2x + 52/-8z = 27, j'2x I3x + 2y + z = ll. 4x + 32/ = 2z + 3.

1043.
to a

43x-12y = 158. 72x + 23y = 845. 47x-97i/ = 50L

j'5x

22/
7 2

l3y10.

= =

42,
2.

2x + 3y + 4z =

17.

11.

3x + 7y =
12.

Find the number of the positive integral solutions of the equation

Reduce the fraction 41/35 whose denominators are 5 and 7.

13.

sum

of two positive fractions

A man
all

buys calves at

.$7

a head and lambs at $6 a head.

does he buy of
each'.'

He

spends in
14.

$110.

How many

first part,

Separate 23 into three parts such that the sum uf three times the twice the second part, and five times the third part will be 79.
5, 7,

15.

Find the smallest number which when divided by

4, 0, 8.

9 will give

the remainders

16. Two rods of equal length are divided into 250 and 253 equal divisions respectively. If one rod is laid along tlie other so that their ends coincide, which divisions will be nearest together?

RATIO AND PROPORTION

347

XIX.

RATIO AND PROPORTION.

RATIO AND PROPORTION

VARIATION

In arithmetic and algebra it is customary to extend Ratio. the use of the word ratio, 215, to numbers and, if a and b denote any two numbers, to define the ratio of a to i as the
;

680

quotient a /b.

(Compare
a to
i is

216.)

denoted by a /b ovhj d'. b. In the ratio a:bwe call a the antecedent and b the conseqxient. Since ratios such as a 5 are fractions, Properties of ratios.
ratio of
:

The

681

their properties are the properties of fractions.

Hence
its

The value of a
Thus, a:b

ratio is not

changed when both of

terms

are multiplied or divided by the

same number.

= ma mb = a /n
:

b/n.

On
is

the other hand, except when a b, the value of a:b changed when both terms are raised to the same power, or
the same
b,

when

number

is

added to both.

In particular,

If a, adding

and
to

are positive, the ratio a:b is increased by b decreased, when a > b. both a and b ivhen a

<

^ For
and

a
b
;

+m
-\-

a
b

= mlb a) -^^ b(b + m)

m {b

a) /bip +

m)

is

positive or negative according as a

:

<6

or a

> 6,
682

Proportion.

When
a, b,
c,

the ratios a

and

d are equal, the

four

numbers

d are said

to be in proportion, or to be

projiortional.

This proportion

may

be written in any of the ways

a/b
It is

= c/d,
is to i

or as
:

a:b

= c:d,
d,

or a

::

d.

read " a

c is to <Z."

In the proportion a
the extremes,

=
c

the terms a and d are called

and

and

the means.

Again, d

is

called the

fourth proportional to

a, b.

and

c.

848
683
Theorem.

A COLLEGE ALGEBRA
In any proportion the product of the extremes
is

equal

to

that of the

means ; that
:

is,

If
For from a/h
Example.

a b

d,

then ad

be.

= c/ d we
first,

obtain ad

be

by merely clearing of fractions.

of a proportion are

The

.second,

and fourth terms

1/2, 3, and 5 respectively; find the third term.

Calling the third term x,
1

/2

5.

Hence
or, solving for X,

5-l/2=:-3 x, x = 5/6.

684
to

Conversely, if the 2}roduct of a first pair of numbers be equal that of a second pair, the four numbers will be in prop)ortion
in

when arranged
let

any order which unakes one of

ad
bd,

the pairs

vieans and the other extremes.

For,

be.

Dividing both members by

we have a/b c/ d.
and
c

Hence
(2)

a:b = c:d
turn,

(1)

Similarly by dividing both we obtain

a-.e

members
and
and

of

= ad =
:

a
be

6.

by

cd, a6,

and ae

in

-.b

= b:d =c a
:

(3)
(5)

d:c 685

= b:a

(7)

and

= = b:a =
6
:

c,

(4) (6) (8)

C'.a

d:b,
d:c.

G83, G84 arranged in

Allowable rearrangements of the terms of a proportion. From it follows that if a, b, c, d are in proportion when any one of the orders (l)-(8), they will also be in

proportion
particular,
1.

when arranged

in

any other of these

orders.

In

In any jyroportion the terms of both ratios

if

may

be inter-

changed.
Thus,
2.

a:b = c:d, then

c.

In any proportion either the means or the extremes

a

may

be interchanged. Thus,
if
-.b

Old, then a:c

b:d.

RATIO AND PROPORTION

The transformations
nation respectively.
Other allowable transformations of a proportion. c: d, we may conclude that If we know that a:b
1

349
alter-

and 2 are called inversion and

686

1.

-{-

-\-

d.

2.

3.

a-\-b: a
:

= =

4,

ma

tub

= nc

nd.
6.

b = c d d. d. c-{-d: c 5. ma nb = mc nd.

a"

^z"

c"

d".

For
ad

in 1

take the product of the means and extremes and

that
is

we have
Hence

6d

= 6c + 6d,

ad

be,

which

is

true since a

-.b

d.

1 is true, 288.

The

truth of 2-6

may be proved

in a similar

manner.

The transformations
rtjspectively.

of

a:b

are called composition, division, and comjiosition

c:d into the forms 1, 2, 3 and division

Example.

Solve x2

By

3,

Hence
or by 4,
5,

Solving

(1)

and

(2),

+ 2x + 3:x2-2a;-3 = 2x2 + x-l:2x2-x + l. 2x2:2(2x + 3) = 4x2:2(x-l). x^ = 0, (1) 1 2 x + 3 = 2 x - 1. (2) x == 0, 0, or - 7/3.

: :

Theorem.

In a

series

of equal ratios any antecedent

all the antecedents is to the

is to its

687

consequent as the

sum of
a\:hi
ni
:

sum of

nil the consequents.

Thus,
then

if

bi

a-2 .b2

as:

63,
:

ai

ao

as

61

62

&3-

For have

let r

denote the

common

value of the equal ratios.

We

then

ai/bi

Hence
or,

Oi

adding,

ai

mu * Therefore

ai

+ ao + 03 + 02 + a3 = +h 61 +
?>a

= r, a2/b2 = r, a3/bs = r. = rbi, 02 = rio, as = *&3, = r(bi + 60 + 63).

r

= at
W

350
Example
then
1.

A COLLEGE ALGEBRA
If

{b

y"^

c)yz

x^

z'^

= =

y.
0.

{c

a)zx

{a

h)

xy,

y,

For multiplying the terms of the first ratio by x, those of the second by and those of the third by z, and then applying our theorem, we have
x^
(6

_
xyz
(c

y"~

c)

_
xyz
(a
?/2

z2

_
xyz
0.

a;2

y2

_|_

g2

a)

h)

which evidently requires that x^

2-

in the proof just given w^ill be found useful in dealing with complicated problems in proportion.

The device employed

2.

=x = x3 + 2 2/3 xy^. Set a/h = x/y = r, so that a = rb and x = ry. Then /a62 = (rW + 2 6^) /y^? = (a^ + 2 = {rhj^ + 2y^)/ ry^ = and (x3 + 2 2/3) / X2/2
Example
Prove that
aS
if

?/,

then

2 6^

a62

ft^)

(r^ {r^

+ 2)/r, + 2) / r.
are said to

688

Continued proportion.

The numbers

a, h,

c,

d,

b -.c -. be in continued proportion if a:h c:d If three numbers a, h, c are in continued proportion, so that a:b b c, then b is called a mean jjroportional to a and c,

and

c is called
b, c

a third proportional to a and

tVi

b.

i/a,

are

continued proportion, then h^

= ac.

For since a-.b

b-.c,

we have

6^

^c, 683.

EXERCISE LV
Find a fourth proportional to 15, 24, and 20; a third proportional to 15 and 24 a mean proportional between 5 a%'^ and 20 aft^ a mean proportional between V'l2 and VtS.
1.
; ;

2.

If 3

2/

=
+

5
2/-

2/,

find x :y
0,
fijifi

also
2/
;

+
2/

y x
:
:

y.

3. 4.

If

2x2

5 xy

6?/

=
z

also
6'2/

x.

If

ax

+
:

cz
:

then
5.

X
If 62 j^

= =

and
be'

a'x
:

+
ca'

+
c'a

c'z
:

0,

is

b'c

ab'

a'b.

d,

then

aft

cd

mean

proportional between a?

c^

and
6.

(^2

If

(pfi

62)

cd

{c-

+ ^2) a6,

then either a:6

c:(Z or

a:6 = d:c.

RATIO AND PROPORTION. VARIATION

7.

351
d.

If

=
^
b

rf,

then

Va + \^
a^
b^

^^a

-{-

= Vc
b

i-

Vd Vc +
:

8.

If^
a

= ',then^- +
c !, a2

^V = 3i^+A+^'. + +
c^

(a

c)3

9.

If the

numbers

) ;

i>i,
:

b^,

b,,

/i, Zoi

)' are all posi-

the ratio lia^ + hao in value to the greatest

tive,

hbi + l-zbo + and least of the ratios oi

ia

lb is
:

intermediate

6i, a-2

60,

&

10.

If

b:k =
0.

c:l

Iz

a: m, and

a, b, c

are unequal, then

+ +m=
I

11.

If

X
y2

mz ny = y
22
61

-.nx

-.ly

mx, then

Ix

+ my +
is

nz

and x2
12.

+
:

= =
1^

0.

If ai

ao
:

62

03

^3,

then each of these ratios

1

equal to

(hal

hal

+ halY (hK + W, +
x"^

hbl)n.

13.

By

aid of 686 solve each of the following equations.

ax
ax

(1)
x"^

a
a

2 x2 2 X-

+
2

rt

2x3-3x2 + 2x +
(2)

3x3 3x3

2x3-3x2-2x-2
A
and

-x2 + lOx - 26 -x2-10x + 26

14. 15.

Separate 520 into three parts in the ratios 2:3:5.

Two

casks
:

are filled with two kinds of sherry

:

mixed

in

A in the

ratio 3

5, in

in the ratio 3

7.

What amount must

be taken

from each cask to form a mixture which kind and 12 gallons of the other kind ?

shall consist of 6 gallons of

one

VARIATION
One independent
related
variable.

If

that

however

their values
is

remains constant, then y

two variables y and x are so may change their ratio said to vary as x, or y and x are

689

said to vary proportionally. More briefly, y is said to vary as x when y / x where c denotes a constant. The notation y (X)x means " y varies as cc."

c,

ov

= ex, = ex
690

and

If given that y varies as x, we may at once write y if also given one pair of corresponding values of

x and

352
y,

A COLLEGE ALGEBRA

we may find c. The equation connecting ?/ and x is then known, and from it we may compute the value of y which corresponds to any given vahie of x.
of y

Example. when x
"We have

If

y varies as
?

x,

and y
y

12

when x

2,

what

is

the value

20

=
is

ex,

and, by hypothesis, this equation

satisfied
c

when y =
is c

12,

2.

Hence
Therefore

12

Hence when x

20 we have y

= = =

2,

that

6.

(>x.
(]

20

120.

691

Instead of varying as x

itself,

may

vary as some function

of X, for example as x"^, or as a; if y varies as 1/x, we say that

+
y

In particular, or as 1 /x. varies inversely as x.

1,

Given that y is the sum of a constant and a term which varies inversely as x also that y = \ when ic = \, and y = 5 when x = 1. Find the equation connecting x and y.

Example.

By

a + b/x, where a and b are constants. hypothesis, y Since this equation is satisfied byx = -l,y = l, and by x
,

1,

5,

we have
Hence a

a r and 5

+
,

i,

o.

3, 6

2,

and the required equation

variable.

is

2/x.
//

692

More than one independent

Let x and

denote

variables which are independent of one another. If a third variable z varies as the product xy, so that z cxy, we say that z varies as x a7id y Jointly ; and if z varies as the quotient c X y, we x/i/,so that z / say that z varies directly as x and

inversely as y.
Tiius, the area of a rectangle varies as the lengtlis of its base

tude jointly

and altiand the length of the altitude varies directly as the area and

inversely as ihe length of the base.

693
is

Theorem.

If irhcn x

is

constant z varies as

x,

then

constant z varies as y, and when y when both x and y vary, z varies

as the product xy. For

^ii
y-2
;

select

any three pairs

of vnlnes of

x and

?/,

such as

Xi, yi
z,

X2,

2/2

and

let Z\, Zo, Zs

denote the corresponding values of

so that

VARIATION
^1,

353
(1) (2)

Vu

zi

Xl, ^2, 23,

X2, ^2, Z2

(3)

are sets of corresponding values of the three variables. Then since the value of x is the same in (1) as in (2), and, esis, for any given value of x, z varies as y, we have, 689,
21/2/1

by hypoth(4)

23/2/2-

Similarly since

1/2

is

common

to (2)

and

(3), vpe

have
(6)
(4)

Z3/Xl=Z2/X2.
Multiplying together the corresponding members of
Zl/-ClUl

and

(5),

Z2/X2y2;

(6)

Therefore corresponding values of z and xy are proportional

z varies as xy, 689.

that

is,

EXERCISE LVI
1.

If

y varies as
7 ?

x,

and y

=~
x^,

when x =

5,

what

is

the value of y

when X
2.

If

y varies inversely as

and y

when x

2,

for

what values

of X will y
3.

S?

X-

Given that y is the sum of a constant and a term which varies as y = I when x = 1, and y = when x = 2. Find the equation connecting x and y.
;

also that

4.

If

and
5.
6.

=
If If

2,

y varies directly as x- and inversely as 2*, and y = 1 when x what is the value of y when x = 3 and 2 = 1 ?
y varies as
x,

show

that x-

y- varies as xy.

X,

show
7.

the square of y varies as the cube of 2, and 2 varies inversely as that xy varies inversely as the square root of x.
3

The wages of
for

men

for 4

weeks being $108, how many weeks will

men work
8.

$135?
of a circular disc varies as its thickness

The volume
its

and the square

face jointly. Two metallic discs having the thicknesses 3 and 2 and the radii 24 and 36 respectively are melted and recast in a
of the radius of
single disc having the radius 48.
9.

What

is its

thickness

right-circular cone whose altitude is a is cut by a plane drawn far is the plane from the vertex of the cone parallel to its base. when the area of the section is half that of the base ? How far is the plane

How
it

from the vertex when

divides the cone into

two equivalent parts ?

354

A COLLEGE ALGEBRA

XX.
694
of

ARITHMETICAL PROGRESSION
This name
derived,
is

Arithmetical progression.

given to a sequence

from a given number a by repeatedly adding a given number d, tliat is, to any sequence which may be written in the form

numbers which may be

a,

-{-

d,

-\-

2d,

+ (n

T)d.

(I)

the difference between every two consecutive terms of (I), it is called the comvion difference of this arithmetical progression.

Since d

is

Thus,
2,

2, 5, 8, 4,

11

is

an arithmetical progression

in

which d

1,

7 is

an arithmetical progression in which d

= 3, = 3.

and

695

The nth term.

Observe that in

(I)

the coefficient of d in each

2 c?, a is one less than the number of the terms a, a c?, of the term. Hence the general or mth term is a (' l)f^ and if the entire number of terms is n and we call the last

term

I,

we have the formula

l

= a + {n-l)d.

(II)

Example. The seventh term of an arithmetical progression is 15 and its tenth term is 21; find the first term a and the common difference d; and if the entire number of terms is 20, find I.

We

have

a
(Z,

Solving for a and

Hence

+ 6d = a = =
l

15 and a
3,

+ 9d =
41.

21.

2.

19,2

696

The sum.
be written
term,
I

Evidently the next to the last term of (I) d, the term before that, I 2 d, the

may
first

(71
if

l)d.
(I),

Hence,

S denote the sum of the terms of

1-

we have
l)d^, l)d2.

S S

= a-\-(a + d) + (a + 2d)-\ = +(l - d) + (I - 2 d)+

I

[a + (?t - (71 +[1

ARITHMETICAL PROGRESSION
Adding the corresponding members obtain 2 S = n(a + I). Therefore
S
of these

355

two equations,

we

^(a

l).

(Ill)
of six terms

Example. Find the sum of an arithmetical progression whose first term is 5 and whose common difference is 4.
Since n

6,

we have

+
(5

Hence
Applications.

S=
If in
a,

25)

= 25. = 90.
697

of the live

numbers

an arithmetical progression any three S are given, the formulas (II) I, d, n,

and (III) enable us to hnd the other two. The only restriction on the given numbers is that they be such as will lead to
positive integral values of n.

Example.

Given

iZ

1/2,

=
3
-

3/2, S a
-i

= - 15/2

find

a and

n.

Substituting in

(II), (III),

n
^

(1)

15 16

n/

3\ 3

,.

Eliminating
Solving
(3),

a,

n^

- 7n -

30
/I

= =

0.

(3)

10 or

3.

obtain a
is

The value n = 3 is inadmissible. Substituting n = 10 in (1), we = 3. Hence n = 10, a = 3, and the arithmetical progression - 3, - 2|, - 2, - 11, -1,-1, 0, I, 1, U.
Arithmetical means.
If three

numbers form an arithmetical

is

698

progression, the middle of the other two.

number

called the arithmetical

mean

The arithmetical mean of any two numbers a and b

half of their sum.
For if X be the arithmetical mean of a and an arithmetical progression.
6,

is

one

then the sequence

a, x, b

is

Hence
and therefore

a
x

= =

x,
6)

(a

/2,

356

A COLLEGE ALGEBRA
all

In any arithmetical progression

the intermediate terms

first

may

be called the arithmetical means of the

It is

and
"

last

always possible to insert or interpolate any number of such means between two given numbers a and b.
Example.
Interpolate four arithmetical mean.s between 3 and
5.

terms.

"

asked to find the intermediate terms of an arithmetical progression in which a = 3, I b, and 7i == 4 + 2 or 6. Substituting i = 5, a = 3, n = 6 in (II), we have
5

We are

5d, whence d

2/5.

Hence

the required

means are

32, 3|, 4i, 4f.

EXERCISE LVII
1.

Find the twentieth term and the sum of the

first

twenty terms of

3, 6, 9, ; of 2.

-3, -li,

0, .

Find a formula for the sum

G, .

to

n terms

of

1, 2, 3,

of

1, 3, 5,

of 2, 4,
3.

Find the sum of the

first

n numbers

of the

form 6r

1,

where r

denotes

or a positive integer.
of ten terms

4. Find the arithmetical progi-ession and whose eighth term is 2.

5.

whose

fifth

term

is

Insert five arithmetical

means between
;

and 2

6.

Given

Ji

=
= =

10, a
1,
I

7.
8.

Given n Given n

12,
2,
7,
i

9.

Given a

=
=

= 4/3 find and S. 0, = -1, d = - 5/3 find a and S. a = - 5/3, = 311 find d and S. = - 23.^, .S = - 559 find n and d.
d
I
;

10. 11.

Given n

a
cZ

=
=

3/7, S
1

45
9?
;

find

d and

/.

Given a
Given
?i

4,

/5,

find
;

n and

S.
I.

12. 13. 14. 15.

16.

Given

Given Given

= 9, d = - 4, .S = 135 find a and n = 10, = - 2, S = 115 find a and d. d = 5, = - 47, S = - 357 find n and a. a = - 10, d = 7, S = 20 find n and
Z
;

I.

Show
c),

that

if a-, b-, c^

are in arithmetical progression, so also are

b).

1/(6

l/(c

a),

l/(n

GEOMETRICAL PROGRESSION
17.
if

357
is divisible

Show

that the

sum

of

any n consecutive integers

by

n,

n be odd.
18.

terms

is

Find an arithmetical progression such that the sum of the first three one half the sum of the next four terms, the first term being 1.
is

Their sum 19. Three numbers are in arithmetical progression. and the sum of their squares is 83. Find these numbers.
20.

15

Find the sum of

9.

all

positive integers of three digits

which are

multiples of
21.
If

sum
end

at simple Interest at 4%, to of 11 years?

a person saves $130 a year and at the end of the year puts this how much will his savings amount at the

22.

Two men A and B

set out at the

miles apart to walk toward one another. miles an hour, while B walks 2 miles the

same time from two places 72 If A walks at the rate of 4

first

hour, 2^ miles the second

will

hour, 3 miles the third hour, and so on,

when and where

they meet?

XXI.

GEOMETRICAL PROGRESSION
This name
is

Geometrical progression.

given to a sequence

699

numbers which may be derived from a given number a by repeatedly multiplying by a given number r, that is, to any sequence which may be written in the form
of
a,

ar, a)", , ar"~^.

(I)

ratio of the geometrical progression (I) and say that the progression is increasing or decreasing according as r is numerically greater or less than 1.
call r the

We

common

sions in

8 are increasing geometrical progres4, respectively; while 1, 1/2, 1/4, 1/8 is a decreasing geometrical progression in which r 1/2.
Thus,
1, 2, 4,

8 and

1,

which

and

2, 2

Observe that the exponent of r in each term one less than the number of the term. Hence in a geometrical progression of n terms whose first term is a and

The nth term.

is

700

of (I)

whose

ratio is

r,

the formula for the last term

I

is

= ar"-i.

(II)

358
701

A COLLEGE ALGEBRA
Let S denote the sum of the geometrical pro-

The sum.
gression
(I).

Then
and

S
rS

= a + ar + ar^ = ar ar^ +
-\

-{-

ar"-^ + ar"-^ + a^"~^ + ar^~^ +

\-

ar".

Subtracting the second of these equations from the a ar'\ Therefore obtain (1 r)S

first,

we

s=

^J^.
it

(Ill)

In applying this formula to an increasing geometrical progression

we may more conveniently write

5

thus:

From

(II)
:

written thus
Example.
terms, find
I

= a(?'-l)/(r-l). we obtain rl = ar". Hence (III) may also be S = (a rl) /(I r), ot S = (rl a) / (r 1).
2,

In the geometrical progression

4, 8,

16,

to eight

and

<S.

Here
Hence, by
and, by >
'

a
(II),
i

= =

2,

= 2, and n = - 256, 2)' =

8."

^ (III),

"

8=

2^^^^^=-'^. 3 l-(-2)

702

If in a geometrical progression any three of the five numbers a, I, r, n, S are given, the formulas (II) and Moreover these two numbers (III) determine the other two.
Applications.

can be actually found by methods already explained, except when the given numbers are a, n, S or I, n, S. If one of the

unknown numbers

is n, it

must be found by inspection

but

this is always possible if admissible values have been assigned to the given numbers, since n will then be a positive integer.

Example

1.

Given r

3,

6, -S

= =

728

find a

and

I.

Substituting the given values in (II) and (III),

i

we have

36

243

a,

and 728
a

a^^^ = 364a.
3-1
486.

Solving these equations,

2,

GEOMETRICAL PKOGRESSIOX
= 6, n = 5, = 2/27 find r and S. = 6r*, whence r* = 1/81, or r = 1- (1/3)5 242 Therefore, by (III), if r = 1 /3, then S = 6 27 1-1/3
Example
2.

359

Given

By

(II),

2/27

1/3.

and

if

=-

1/3, then

S-6 l-(- 1/3)5

l-(-l/3)
in

122
27

Hence there are two geometrical progressions and 1 = 2/27.

which a

= 6, n =

5,

Example

3.

Given a

= -S,

={a

46875,

S
{I

Substituting in the formula

S =

rl)

= - 39063 find r and r), 701, we have

;

n.

Therefore, by

39063
(II),

= nl + i^^IL^

1-r

whence

=-

5.

_ 46875 = - 3(-

5)-i, or

(- 5)-i
5)6.

= 15625.

But by factoring 15625 we find 15625 = 5^ = (Hence n 1 = 6, that is n = 7.

Example

4.

Given a
93

=
1

3,

5, -S

93

find r
r2

and

I.

By

(III),

3^^ = 3(1 +
r
r*

+
0.

r3

r4).

Hence

r^

r^

30

problem involves solving an equation of the fourth degree when a, n, S are the given numbers, to find r we must solve an equation of the degree n 1. In this particular case, however, we may find one value of r by the method of 455. It is 2. Substituting r = 2 in (II), we have Z = 3 2* = 48.
this
;

Thus

and, in general,

Geometrical means.

If three

numbers form a geometrical

is

703

progression, the middle of the other two.

number

called the geometrical

mean
a

The geometrical mean of any two numbers a aiid b square root of their product.
For if X be the geometrical a geometrical progression.

is

mean

of a

and

6,

the sequence a, x, h

is

Hence

x/a =

/x and therefore x
all

Vab.
the intermediate terms
first

In any geometrical progression

may

be called the geometrical means of the

and

last terms.

360

A COLLEGE ALGEBRA
insert

We may

any number of such means between two given

h,

numbers a and
Example.
It is

as in the following example.

Insert fuur geometrical

means between 18 and 2/27.

required to find the intermediate terms of a geometrical progression in which a = 18, Z = 2/27, and Ji = 4 + 2 = 6.
Substituting the given values in
(II),

we have
r

2/27
Hence the means are

18

r*,

whence

1/3.

6, 2,

2/3, 2/9.

704

Infinite decreasing geometric series.

the form

^^^.

_^ ^^,,

_^ ^,,._i

We call an ^
. .

expression of
^^^

.^

supposed continued without end, an infinite geometric series. By the formula (III), the sum of the first n terms of (1) is

a(l-.)/(l-r).
Suppose that
increased,
/"

<

numerically.

Then, as n

is

indefinitely

will

a(l /) will approach a /(I ?"")/(l r) as limit. this limit the su))i of the infinite series (1). Hence, if
the

approach

as limit,

724, and therefore

AVe

call

S denote

sum

of (1),

we have

Example
Here

1.

Find the sum of

a

Hence

1 + 1/2 4 1/4 + = 1 and r = /2. S = 1 / [1 - 1/2] = 2.

1

1/8-1

Example

2.

Find the value of the recurring decimal .72323

2.3
2.3
'-

The part which

the

recurs

may

be written

1-

1000

100000

, '

and, by ^
'

(2), ^

"

sum

of this infinite series

is

.023

23
or

.01

990

Adding

6.1
.7,

the part which

does not recur, we obtain for the value of the ^ given decimal

358

495

EXERCISE
1.

LVIII
first five

Find the

fifth

term and the sum of the

2,

terms of the

geometrical progression
2.

6,

18, .
first

Find the fourth term and the sum of the

4, 6, 9,
.

four terms of the

geometrical progression

GEOMETRICAL PROGRESSION
3.

361
:

Find the sums of the following

infinite

geometric series

12_6+3-...;l-i + i----;f + l + A +
4.

---.
:

Find the values of the following recurring decimals

.341341

.0567272
.03, r

8.45164516
I

5. 6.

Given a Given n
Given a Given n

== = = =
7, 1

= =

10,
i

?i

6
;

find

and

S.

48,

7.
8. 9.

/ 16, r

2,
Z

= 3/4 =8
i

find r find

and

S.
S.

n and

5, r

= 3, =

81

find a
;

and

S.
I.

Given a

54, r

10. 11.
12. 13. 14. 15.
16. 17.

Given n Given a Given

7i

=: 4,

= 1/3, 6' = 80| = - 3, -S = - 408

l

find
;

n and
and

find r

I.

=-9/16, = - 16/9, S = = 0, r = - 2/3, S = 665/216

781/144;
;

find
I.

n and

r.

find a

and
a.
r.

Given r Given
?i

= =

3/2,
4,
5,
Z

301,

-^

831; find n and

= =

54/25, S
48,

=
;

544/25
find a

find
r.

a and

Given n

S = 93

and

Find the positive geometrical mean of a^/b and b^/a.

Insert three geometrical

means between

and

405.
is

18.

The
is

term
19.

third term of a geometrical progression 3/8. Find the seventh term.

and the sixth

the

first

Find a geometrical progression of four terms in which the sum of and last terms is 133 and the sum of the middle terms is 70.

20.

sum

shall

Find three numbers in geometrical progression such that their be 7 and the sum of their squares 91.

Three numbers whose sum is 36 are in arithmetical progression. 43 be added to them respectively, the results are in geometrical Find the numbers. progression.
21.
If 1, 4,

22.

There are four numbers the

first

three of which are in arithmetical

The sum of progression and the last three in geometrical progression. the first and fourth is 16 and the sum of the second and third is 8. Find
the numbers.
23. What distance will an elastic ball traverse before coming to rest if be dropped from a height of 15 feet and if after each fall it rebounds to 2/3 the height from which it falls ?

it

362

A COLLEGE ALGEBRA
XXII.

HARMONICAL PROGRESSION
This name
is

705

Harmonical progression.

given to a sequence of

numbers whose reciprocals form an arithmetical progression, that is, to any sequence which may be written in the form
1/a, l/(a
Thus,
1,

d),

l/(a

2rf),

l/[a+(ft-l)(/].

1/2, 1/3,

1/4 and 3/2, 3/4, 3/6, 3/8 are harmonical

are in harmonical progression, then

progressions.

Example. c =2 a b

Prove that
:

if

a, 6, c

c.

Since 1/a, 1/6, 1/c

is

an arithmetical progression, we have

1/6.
that

l/6-l/a = 1/cHence
c (a

6)

(6

c),

is

c.

706

To find any particular term of an harmonical progression, we obtain the term which occupies the same position in the
corresponding arithmetical progression and invert
it.

Example. Find the tenth term of the harmonical progression 3/5, 3/7, 3/9,..-.

By
is

695,

6/3, 7/3, 9/3,

the tenth term of the corresponding arithmetical progression is 23/3. Hence the tenth term of 3/5, 3/7, 3/9,

3/23.

707

Harmonical msans.
gression, the middle

If three

numbers are
is

in

harmonical pro-

number

called the harmonical

mean

Again, in any harmonical progression all the intermediate terms may be called the harmonical means
of the extreme terms.

of the other two.

Example 1. Find the harmonical mean of a and 6. K this mean be x, then 1/a, 1 /x, 1 /6 is an arithmetical progression. Hence 1/x - 1/a = 1/6 - 1/x, or 2/x = l/a + 1/6.
Therefore

2 a6 / (a

6).

is

Prove that the geometrical mean of two numbers a and 6 also the geometrical mean of their arithmetical and harmonical means.

Example

2.

Let A, G, and 77 denote respectively the arithmetical, geometrical, and harmonical means of a and 6.

HARMONICAL PROGRESSION
Then
Henco

363

A=

'^^, G
2

= V^b, H "' ~ =
ah

2 ah

AH =
2

= G\

b b are positive,

Example

3.

Prove that when a and

A > G> H.

Wehave

H ='l^ - -^^
2

= i^^. + b 2 {a + b)

Therefore, since (a

2,

And

since,

by Ex.

A>G>n.
1.

6)2/2 (a + b) is positive, we have G is intermediate in value to A and H,

A>H.
we have

EXERCISE LIX
Continue the harmonical progression 3/5, 3/7, 1/3 for two

terms.
2.
3.
4.

Find the harmonical mean of 3/4 and

Insert four harmonical

5.

means between 10 and

of

15.

The second and fourth terms

4.

and

5.

an harmonical progression are 4/5

Find the third term.

of two numbers Find the numbers.
is

The arithmetical mean

is

and

their harmonical

mean
6.

15/4.

The geometrical mean

is

mean
7.

16/5.

of two numbers Find the numbers.

if

is

4 and their harmonical

Show
c),

that

a, b,

are in harmonical progression, so also are

b).

a/(b

b/{c

a),

c/{a

Three numbers are in harmonical progression. Show that if half of the middle term be .subtracted from each, the results will be in geomet8.

rical progression.
9. Show that if x is the harmonical mean between l/{x-a) + l/(x-b) = 1/a + 1/6.

a and

b,

then

meets bisector of the vertical angle C of the triangle at D, and the bisector of the exterior angle at C meets are in harmonical progression. produced at E. Show that AD, AB,
10.

The

ABC

the base

AB

AB

AE

11. The point P lies outside of a circle whose center is 0, and the If tlie line PO meets the tangents from P touch the circle at T and T. is the harmonical mean circle ut A and B and TT' at C, show that

PC

between

PA

and PB.

36-1

A COLLEGE ALGEBRA

XXIII.

IMETHOD OF DIFFERENCES. ARITHMETICAL PROGRESSIONS OF HIGHER ORDERS. INTERPOLATION

ARITHMETICAL PROGRESSIONS OF HIGHER ORDERS

708
If in any given sequence of Differences of various orders. numbers we subtract each term from the next following term,

we

obtain a sequence called the first order of differences of the new sequence in a similar given sequence; if we treat this of the given manner, we obtain the second order of differences

sequence
Thus,
if

and so

on.
l^,
1,

the given sequence be

2%
8,

.S^,

we have

given sequence
first

27, 64, 125, 216, , 91, -,

differences

7,19,37,61,
12,
6,

second differences
third differences

18, 24, 30,

6,

6,

The fourth and

all

subsequent differences are

0.

709

This name is given Arithmetical progression of the rth order. to a sequence whose rth differences are equal, and whose
subsequent differences are therefore
Thus,
1^, 2^, 33, 43,
.

0.

is

an arithmetical progression of the third order,

for, as just

shown,

its

third differences are equal.

An
of
its

ordinary arithmetical progression, 694, is of the ^rsi order, each first differences being the common difference d.
order.

710

The nth term of an arithmetical progression of the rth Given any arithmetical progression of the rth order
!,

a^,
d,.

3)

di,

*)

^H + ij

)
its

y^)

and

let

c?i,

d^,

denote the

first

terms of

successive

orders of differences.

We
d^,

are to obtain a formula for a in

n.
is

terms of

a^, d^, d^,

and

The

first

order of differences of (1)

a^

!,

as

a2,

a^

as,

On + i

(2)

METHOD OF DIFFERENCES
The
first

365
its first,

term of

(2) is di,

and the

first

terms of

for the first differences differences are d^, d^, second, of (2) are the second differences of (1), and so on.

term of

Hence when we have found an expression for any particular from it an expression for the corre(1), we can derive
:

sponding term of (2) by applying the rule

Replace
a^, di,
d^,

by

d^,

d<^,

d^

(3)
d^.

Now
follows
:

since

d^

a^

!,

with this formula for

2>

we have a^ a^ -{we may reckon out

Starting
a^,

Og,

as

We

have
(3),

02

= 01+
^i

di
di
-{-

Hence, by

2 =
= Og = ^4 =
^3

d^

Adding,
Hence, by
(3),

+2 +
3

r/i

a^

di
Oi
c?i

Adding,

+ d^ + 2 dz + ds + 3 + d^
(/o

and so on

indefinitely, the reckoning, so far as coefficients are concerned, being precisely the same as that given in 311 for Therefinding the coefficients of successive powers of a + b.
fore,

by

561,
(?t

we have the formula

= ! +

1) di
,^

^ (-l)(.-2)
Example.

^
do

^ fa-l);-(-^) ,_.
1^,

(I)

Compute
ai

the fifteenth term of

(h

2-^,

by

this formula.

Here, 708,

1,

7,

Hence

a^ =

14

= 12, 141S ^

ch

= +

dr

6.

12

141S12 '-^-^ 2-3

3375.

Sum

of the first n

terms

of

an arithmetical progression
.

of the

711

rth order.

Let

S,,

denote this sum, the sequence being

!, Oo, Og,

a.

o4.j, ,

(1)

and

di, d^,

d,.

having the same meanings as in

710.

366

A COLLEGE ALGEBRA
the sequence of which (1)
:

Form
ences,
0,

is

the

first

order of differ-

namely
rti

!,

+
is

Then
term
are

arithmetical
is 0,
6?i,

+ a^ + as,---, a^ ao + a, (2) + l)th term of (2), and since (2) is an progression of the ( + l)th order whose first
a2,

ai

-\-

-\-

the

(?i

and the
d^,

first

terms of
have, by

its

several orders of differences

!,

d,.,

we

(I),

Example.

Find the sum

of the first fifteen

1,
(Zi

terms of P,

2^, 3^,
6.

Here, 708, n

15,

ai

=
^ 7

7,

^2

=
, 12

12, ds

d,.

Hence
712

Si5

= ,^,15.14 15H
2

15

14

13

2-3

15 -14 -13 -12

2-3-4

G = 14400.

Piles of spherical shot.

1.

To

find the

number

of shot

when

the pile has the form of a triangular pyramid.

The top course contains

next
1

shot, the next lower course

2 shot, the of

+
if

3,

and so
1, 3, 6,

on.

Hence,

there are n courses, the

10, 15,

numher

of shot

is

the

sum

n terms

of the sequence

differences of this sequence are 2, 3, 4, differences are 1, 1, 1,

first
.

The

5,

and the second

in

Hence which tti

1, 3, 6,

is

1,

di

an arithmetical progression of the second order

2, (^2

1.

Therefore, by

(II), .S

+ M_2Lzi}
1-2

+ Mn-\){yi-2)
1.2.3

_
Thus,
2.
in

n(n

+ l)(n +
1.2-3

2)

a pile of twenty courses there are 20

21

22/6

1540

.shot.

To

find the

number

of shot
base.

when

the pile has the form

of a

pyramid with a square

...
.

Enumerating the shot by courses

12, 22, 32, 42,

as before,

we

obtain the sequence

The first differences are 3, 5, 7, and the second are 2, 2, Hence 12, 22, 32, is an arithmetical progression of the second order
, .
.

in

which

ai

1, cZi

3,

1/2

2.

METHOD OF DIFFERENCES
Therefore, by (II), S.

367
2

+ ^^^il 3 + -(--^n--^) _ 7i(n + l)(2n + l) ~

n
.

1-2-3

Thus, when n
3.

20, the pile contains 20

21

41

/6

2870 shot.

To

find the

number

of shot

gular base and terminates at

when the pile has a rectanthe top in a row of 2^ shot.

p,

Again enumerating the shot by courses, we obtain tlie sequence 2(p+l), 3(p + 2), 4(p + 3),.... The first differences are p + 2, p + 4, p + 6, and the second differ

ences are

2, 2,

Hence

p, 2 (p
in

1),

3 (p

second order

which Oi
(II), S,.

is an arithmetical + 2), = p, di = p + 2, and d^ 2.

progression of the

Therefore, by

= _ ~

np

+ ?^i^> (p +

2)

+ ^i^-^n--'^)

n(?i

l)(3p

+ 2n-2)
shot
is

1-2-3
Thus, when n = 20

andp = 5,

the

number of

20

21

53/6 = 3710.
713

tion of the

An examinatheorem respecting arithmetical progressions. formula for the nth term of an arithmetical

progression of the rth order, 710, (I), will show that if we carry out the indicated multiplications and arrange the result

according to descending powers of

n,

we can

reduce

it

to the

form

where the

coefficients

b^, ^1,

^r

^^^ independent of n.

Thus, when r

=
a

2,

we have
ai

+
.

(n
.

l)di

-^

1-2

-d^
di

^2

n2+

{di

3 '-d2)n

(ai

d^).

the rth order, Oj, a.^, a^, of a certain polynomial

Therefore the terms of any arithmetical progression of are the values for 71 = 1, 2, 3,
,

^v''^

^i?^'""'

with respect to n

is r.

We

are going to

+ br whose degree show conversely that

368
714
Theorem.

A COLLEGE ALGEBRA
If
(f)(x)

denote any rational integral function of

the ith degree, as

the sequeiice of numbers ^(1), ^(2), <^(3),

setting

obtained by

1, 2, 3,

progressio7i of the

rtli

successively in ^(x), order.

is

an arithmetical

Here the given sequence

</.(!),

of

numbers
/>(3),

is

.^(2),

<^(4),...

(1)

and we

are to prove that all of its rth differences are equal. Evidently the first differences of (1), namely

<^(2)-<^(l), <^(3)-</.(2), cA(4)-<^(3)...,

are the values of

(2)

But ^{x
nomial in
For, by

X.

be reduced to the form of a poly1) </>(a^) Ji^ay Its degree is r 1. Call this polynomial ^i {x).

<^ (a;

1)

<^ (cr)

for

a;

1, 2, 3,

.the

binomial theorem,

561,

we have
(6oX'-

0(x

l)-<?i(x)

= 6o(x + ])'-f6i(x + l)'-i + = 6oX'- + r6oX'-i + hM' + = rhox'-'^ +

^

6iX'-i

--.)
-

(M''+M'"-^ +

Similarly,
^l

if

we
</>l

write
(-^O

{X

+ 1) -

<^2

G^),

<^2

(--K

1)

<^2 (')

</>3

(a^),
a;

and
2, 3,

so on, the values of

cj>.,(x),

^3(.r), ,

^,. (.r)

for

1,

will be the second, third,

is

But

(f>,.(x)

rth differences of (1). a constant and the ?'th differences of (1) are
,

For the degree of <^2(^) is (r therefore equal. 1) 1, or 3 and finally that of <{>r(x) is r 2 that of ^3(.r) is r

r,

or

0.

For example,
<Pi (x)

if

(x)

^2{x) 03
(x)

= 2 (X + 1) = 6{x + l)2 + = 12(x + 1) +

+ 1, we have - (2 x3 - X + 1) = 6x2 4. Gx + 1, 6(x + l) + l- (6x2 + 6x + l) = 12x + 12, 12 - (12x + 12) = 12.

2 z^

(X

1)

mp:thod of differences
Hence the values
are the
firKt,

369

of

6x2

+ 6x +

1,

l2x

12,

12 for x

1,

2, 3,

2x'

second, third differences of tlie corresponding values of and the third differences are equal, all being 12.

Thus, for X

1, 2, 3, 4, 5,

we
15,

find
62,

2x3-x + l= 6x2 + 6x + l =
12

2,

125, 24G, , 121, 181,- ,

60,
12,

(1) (2)

13,

37, 73,

12x4-12=24,36,48,

72,

(3)
(4)

12,

12,

12,

12,-...

And by comparing
Corollary
1.

(1), (2), (3), (4),

first,

are, as they should be, the

find that (2), (3), (4) actually second, tliird differences of (1).

we

vth powers of consecatioe integers arithmetical prof/ression of the xth order.

The

form an

715

For

1'",

2'',

3'',

rth degree

(x)

x''

are the values of the rational integral function of the '. for x = 1, 2, 3,

The products of the corresponding terms of two arithmetical progressions, the one of the vth order and the other of the sth order, form an arithmetical progression of
Corollary 2.

716

the (r

s)^/i

order.
of a rational integral function of the rth degree by one a rational integral function of the (r -f s)th degree.

For the product

of the sth degree
is

EXERCISE LX
1.

Find the twentieth term and the sum of the

1, 2, 4, 7,

first

twenty terms of

the sequence
2.

Find the eightieth term and the sum of the

3, 8, 15, 24, 35,

first

eighty terms of

the sequence
3.

Determine the order of each of the following arithmetical pro-

gressions.
(1) (3)
3,

0,

1,

0,

3, ,

(2)
(4)

10, 38, 88,

2,

166, 278, 430, ,

285, 204, 140, 91, 55, ,

20, 90, 272, 650,

1332,

..
the

Also find the eighteenth term of (1), the twentieth term of twelfth term of (3), and the tenth term of (4).
4.

(2),

What
?

is

the order of
of its first

3,

4, 3

5,

What

is its

nth

term

n terms ? What is the order and what the nth term of

the

sum

4 22, 2 6

32,

3 8 42,

370
5.

A COLLEGE ALGEBRA
Find the number of shot
in a triangular pile of fourteen courses. shot are there in the lowest course ?
fifteen courses six courses

How many
6.

If

how many
7.
if

from a square pile of shot remain ?

be removed,

How many How many

sliot are there in

,5

the uppermost course contains

8.

a rectangular pile of twelve courses shot?

shot are there in a triangular pile whose lowermost

course contains 253 shot ?

9. The number of shot in a certain triangular pile is four sevenths of number in a square pile of the same number of courses. How many
?

the

shot are there in each pile

10.

shot are there in a rectangular pile whose top contains 9 balls and whose bottom course contains 240 balls?
11.

How many

row

Show
Show

that 13
that

+
+

2-

+
+

+ +

?x^

(1

+
{n

n)2.

12.

\*

2*

n* ^ ~

30

^){2n

\) (3

n^

+ Sn-

1).

13.

What
we

is

the order
?ith

and what the sum

is ?i2

of the first n

terms of the

progression whose
14. in
If

term

1 ?

n {n

1)

()i

2)

/6

write

down

the arithmetical progressions of the

respectively

first

order
to

which d

1, 2, 3,

and then sum each progression

one, two, three, four,

terms,

we

obtain the following .sequences of

numbers, called respectively the triangular, quadi angular, pentagonal,

numbers

1,

3,

6,

10,

1,

4,

9,

16,

1,

5,

12, 22,

Show

that in the

A;tii

of these sequences the ?ilh

term and the sum

of the first
respectively.

n terms are n{kn

2)/2 and

n(ji

1)

{kn

S)

/O

15. Show that the order of an arithmetical progre.ssion of any order is not changed by adding to its terms the corresponding terms of an arithmetical progression of a lower order.
16.

Show

that

if

in

for X successive terms of

a polynomial of the nth degree. f{x), we substitute any aiithnietical progression of the first order,
;

we obtain an arithmetical progression of the nth order and, in general, if we substitute for x successive terms of any aritlimetical progression of the rth order, we obtain an arithmetical progression of the ?irth order.
that

METHOD OF DIFFERENCES
INTERPOLATION

371

Interpolation. Suppose that y is known to depend on x in such a manner that for each value of x between a and h, y has a definite value. Suppose also that the values of y which

717

Then from these known values

correspond to certain of these values of x are actually known. it is possible, by a process

b.

called interpolation, to derive values of y corresponding to

other values of x between a and

This process is employed when the general expression for y in terms of x is unknown, or if known is too complicated to be conveniently used for reckoning out particular values
of
y.
:

Briefly stated, the process is as follows we set y equal to the simplest integral expression in x which will take the given

values and then derive the values of y which we seek from this Of course the values thus obtained will ordinarily equation. be only approximately correct.

Method

of

undetermined

coefficients.

We may
that y

proceed as in

718

the following example.

Example.
find y

For x

2, 3, 4, 5 it is

known

b,

4,

7,

34

when x = 5/2.

Since the simplest polynomial in x which will take given values for four given values of x will ordinarily be one of the third degree, we

assume that
2/

6o

bix

hnx^

?)3x',

and then

find the coefficients

6o} &i, &2) &3

as follows.
5

Since y

Since y

Since

Since

2/

=b =4 1 y = - 34

when x =
when x = when x = when x =

2, 3, 4,
5,

1, &i

34

= 6o + = 60 + = 60 + = 60 +
2, 62

Solving these equations, 60

=
+
2/

+ 4 &2 + 9 ^o 4 61 + 16 62 5 61 + 25 62 = 4, 63 = 1.
2 6i

3 61

+ + + +

63.
63.
63.

27 64

125 b.

Hence
Therefore,

?/

=1-

2 x

4 x2

x^.

when

= 5/2 we

have

= l-5 + 25-

125/8

= 43/8.

372

A COLLEGE ALGEBRA
in general, if r
?/

And
values

?/i,

y2,

x^^y respectively,

, we
h,

+ 1 values of ij are known, say the = x^, x^, , Vr+v corresponding to x assume that

y
find
bo,

h,x

hx^

+ K^%

(1)

bi,

,
Xi,

employ
719

(1) as

by the method just illustrated, and then a formula for computing y for values of x interb,.

mediate to

a-o,

a;^

i-

Method

of differences.

When

Xi, Xo,---, x^+t,

are consecutive

integers, the

formula (1) of
,

718,
(.X

may

be reduced to the form

2/

2/1

Xi) (X \ H + (^ - ^i)"i ^ 172 - xi) ---(x - X,) (x

/
7

Xo)

"^

+
where
di, ck,

1.2.-.r
denote the
iji,
?/2.

'^''

.2\ ^^

, ^^^

first

terms of the successive

+v

orders of diiferences of

; Vr

For since
values otbo
ical

Xi, X2,

x^ +

are consecutive integers, the corresponding

bix

6,.x^

namely

?/i,

progression of the rth order,

ITU.
in

y.2,

by substituting n
,

1, 2,

2/r + 1, form an arithmetHence we may also obtain 2/1, the formula, 710, (I), namely

^2,

2/

yi

{n-l){n-2) ^^2 + ^ ^, iw/ 4-(n-l)di +

+ (n-l)---(n-r)
I

1.2...

But setting n = 1, 2, 3, same results as setting x =

Therefore the second
values for r

in this

formula

Xi, X2, X3,

will give identically the in (2). 1, Xi 2, Xi, Xi

member
x.

of (2)

and that of

(1),

718,

have equal

values of

But both are

2, 3, 4, 5, let
2/4

of the rth degree.

Hence they

are identically ecjual, 421. Thus, as in 718, for x =

2/1, 2/2,

=
7

5, 4,

2/3,

5,

-1,

- ^^
10.

7,

34,

We

have

First differences

1, - H, 27.

Second differences
Third difference
Substituting in
ds
(2), Xi
2/

10,
6.

=-

6,

we have

which may be reduced

= 2, X2 = 3, X3 = 4, 2/1 = 5, di = - 1, ^2 = - 10, - (x - 2) - 5 (x - 2) (x - 3) - (x - 2) (x - 3){x - 4), = 1 - 2 x + 4 x^ - x^, as in 718. to

2/

METHOD OF DIFFERENCES
Example.
v'33

373
3,1748,

Given

3.2075; find
2/1,
2/2,

Vao = V3U3.
2/4

3.1072,

V3l =

3.1414,

V32 =

and

2/3,

==3.1072,

3.1414,
.0342,

3.1748, 3.2075.
.0334,
.0327.

First differences

Second differences
Tliird difference

.0008, .0001.

.0007.

Substituting in
da

(2) Xi

30, X2

31, X3

32,

?/i

3.1072, di

.0342,

=-

.0008, da

.0001,

and x

31.6,

we have

3.1072

(1.6) (.0342)

+ ILMl^ (_
.05472

.0008)

+ li:^M:M:iji) _
(

.0001)

3.1072

.000384

.0000064

3.1615 +

The formula (1) of 718 Lagrange's formula. reduced to the following form, due to Lagrange
:

may

also be

720

_ ^^

C-^'

- ^2) (x - a-a)
^2) (^1
~"

(a;

x,^,)
^r + l)

(^1

^3)

(^1
3^^4,1)

+ ^2
_l_
_

{x

Xi)

(a*

(a;

^3)

(a;

x-j)

(x

X2)

(x

x^)

For the right member of (3) is an integral function of x of ?'th degree and its values for x = x^, Xo, a-,.^! are t/i, Thus, if we set x = x^, every term except the y-2i ) Vr+ifirst vanishes and the first term reduces to ?/i. Hence, 421, the right member of (3) and that of (1), 718, are equal for ? + 1 values of x and are therefore equal identically.
the

Tlius, as in 718, for x

stituting in (3),

=
(X
(2

2,

3, 4,

5,

let

=
5)

5,

4,

7,

34.

Sub-

we

obtain

_ ^~
^

3) (X

4) (X

-3)

(2

-4)

(2

-5)
(
^

(x-2)(x-4)(x-5) _
(3-2) (3-4) (3-5)
will

(x-2)(x-3)(x-5) _ " x-2) (x-3) (x-4) ^^

(4-2) (4-3) (4-5)
2x

(5-2) (5-3) (5-4)'

which

reduce to y

=1

4 x^

x^,

as in 718.

374

A COLLEGE ALGEBRA
EXERCISE LXI

1.

For X

=-

3,

2,

- 1,

it is

known
/2.
12,

that

?/

=-

20, 6, 0, 4

find y

when X = 5/2,

also

when x =

2. Given that /(4) = 10, /(6) /(x) and then compute /(12).

=-

/(7)

=-

20, /(8)

=-

18

find

3.

Giventhat252 = 625, 262

= 676,

272=729;

find 26.542

by the method

of differences.
4.

Given that

23

8,

3-3

27, 43

64,

53

125; find 4.83 by the

method
5.

of differences.

Given that 1/22

.04546,

1/23

.04348,

1/24

.04167,

and

1/25
6.

.04; find 1/23.6

by the method of

differences.

Given that
20.8566,

V435 =
ences.
7.

V432 = 20.7846, V43 3 = 20 .8087, V434 = 20.8327, V436 = 20.8806 find V435.7 by the method of differ;

By

degree whose values for x

aid of Lagrange's formula find the polynomial of the third = 2, 0, 4, 5 are 5, 3, 2, 4.

XXIV.

LOGARITHMS

PRELIMINARY THEOREMS REGARDING EXPONENTS

721

Theorem

1.

If ^ denote any real number greater than

p

1,

and

p,

q denote positive integers, then a^

1,
.-.

For a >

aP

> 1,

'I,
.-.

Va''

>

>

1.

'J

1,

.-.

a'l

> 1,

261.

722

Theorem

and

r,

s r

//"a denote any real number greater than any two rationals such that r > s, then a"" > a'*.
2.
.-.

1,

For

- s>0,
3.

a''-*>l,

.-.

a'--'-

a^>a^,

.-.

a'->a% 721, 261.

lii^fi

723

Theorem

//"a

>

and n

be integral, then
1

a"

00.

For since a > 1, we may write a Then a" = (1 + d)'', and since (1 +
Therefore, since
^''"

=
00,

d)"

+ d, where d is positive. > 1 + Jid, 561, wehavea>l+nd.

we have
^''"

(1

nd)

a"

00.

LOGARITHMS
Theorem
For
let

375

4.

//O
/6,

<a<
6

1,

and n

he intearal,

1
n

a"

0.

724

where
/ /n=

> 1,

since a

< 1.

Then lii n = oo

^^"^ 6

0,

li 512, since = oc 6
n
lii^i

, 723.
1

Theorem
1.

5.

i/'ii

Ze

integral,

v^ =

7i

li^ii^
1

=X

a"

1.

725
is

When

> 1, we have
.-.

> 1,

721,

so that a"

=1+

dn,

where d

some positive number dependent on n. Then a = (1 + d)", a>l + ?id, d <(a

.-.

l)/n.
lip^

li Therefore, since (a
1

- l)/n =

0, 512,

we have

0.

Hence 1 a n = 00
2.

7i

li|" (1 =X

When 0<a<l,

Then

11.
let

a"

+ d) = 1. = 1/6, where 6>1, since a<l. 1 & = 1, since 1 6" = 1, by 1.

a
1

Theorem 6. If h he a rational number and x he a variable which apjjroaches b through rational values, then ^^^ a'' = a^. 1. The theorem holds true when h, the limit of x, is 0. For in this case we can select a variable n which takes integral values \/n<x<\/n and that when only and such that we shall always have
X

'26

then n = co. Then a^ will always

1
i_

lie

between a" and

a
i

",

722,

and since

lim
2.

an=

lim a

1,

725,

we have 1

a^

a\

The theorem holds

true

For since a^

a^

a^-*,

when 6 t^ 0. we have 1 a^ x=

aP

1'"^ a:= b

a^-*

a^,

by

1.

Theorem
able
tvJiicli

7.

If h be an irrational number and x he a vari-

727

apj)roaches b through rational values, then aj^ ivill b, and the value of this limit is approach a limit as x independent of the values ivhich x takes in approaching b.

The reasoning

is

the

same whether a > 1 or a <

1,

but to

fix

the ideas

we
X

shall suppose that a

sequences of rational values through which approaching b as limit. From among them select some particular increasing sequence, and represent x by x' when supposed to run through this sequence. Then as x' == b, the variable a^' continually increases, 722, but remains finite less, for instance, than a'', if c denote

There are

infinitely

> 1. many

may run

in

376

A COLLEGE ALGEBRA
b.

any rational greater than

this limit L.
It

Hence

a'-''

approaches a

limit, 192.

Call

this same limit i if x only remains to prove tliat a^ will approach b through any other sequence of rationals than that through which But a^ = a^' a^-^' and therefore lim a-^ = lim a^' lim a* *' = L, x' runs.

approach

since lim a^-^'

1,

726.

728

Irrational exponents. employ the symbol a'' to denote the limit which a^ will approach when x is made to approach Hence by a*, when b through any sequence of rational values.
b is irrational,
'

We

we

shall

mean

^^^^^ a"".

x=

729

Having thus assigned a meaning to

a""

we can

readily prove that ^^^^ a^

a*

when x is irrational, when x approaches b

through a sequence of irrational values.

For let x\ x, x" denote variables all of which approach b as limit, x' and x" through sequences of rational values and x through a sequence of irraIt then follows from 726, 727 tional values, and such that x' <x< x" = *. that a^ lies between a^' and a-^", and therefore since ^i^^ a^' = j,;^^ a^"
.

that

a^

a*.

730

Theorem
exjjonents.

8.

The laws of exponents are valid for irrational

For
only.
1.

let b

approach
a''

and c denote irrational numbers, and x and y variables which and c as limits. We suppose x and y to take rational values

a"

a'>

'^.

For since

W^a^'

a^ + i>,

we have

lim

a^'a'J

But
and
2.
(o*)'^

lim

a'^ay

lim a^ + y

= =

lim a^ lim ay
a^"" (-+
y^

= lim a^ + >. = a^a", = a* +

''.

203, 728

203, 728

a*"^.

For

(a^)^

Hence Hence
3.
(aft)'-

^i (a^)"
^'

= =
=

a^^^"^
a""'!

or
or

(a'')"

*"

"^^S

y=c^

(a'")"

^""

y=c

'"',

(a'')'^

= aK

728, 729

a'-b'^.

For

{ab)y

Hence
That
is,

\\m

{ab)!i

= = = (a6)<-

"&'.

V\m

Wbv =

Vim av

Ximbv

203

a'b^.

728

LOGARITHMS
LOGARITHMS.
Logarithms.
or

377

THEIR GENERAL PROPERTIES

a,

Take

number

of reference.

any positive number except 1, as a base We have shown that every real power
In a

731

of a, as

a'^,

denotes some definite positive number, as m.

subsequent section we shall show conversely that every positive number, m, may be expressed in the form a^, where /a is real.
to the base a and If a*^ = m, we call /* the loyarithin of Hence the logarithm of uirepresent it by the symbol log7. to the base a is the exponent of the power to which a must be raised to equal m, that is 0^""'" = m.
Thus, 3*

732

81,

.-.

log381

2-3

1/8,

.-.

-3 =

log.2l/8.

Since a"

=
>

1,

we always have

logl

0;

and since

a^

a,

733

we always have loga

= 1.

When

1, it

follows from

increase in the
;

number

a/^ = m, by 722, that to any there corresponds an increase in

734

also that if its logarithm /i rithm /x is positive, and that

m
if

is

greater than 1,
lies

its

loga0, its

between

and

logarithm
Again,

/x is

negative.

> 1, we have, 723, = oo, and lim a-'^ = lim 1 /a'^ = 0. We therefore say, when a > 1, that logco = x, and log0 = 00.
when a
lim a^

735

Theorem 1. The logarithm of a product to any base sum of the logarithms of the factors to the same base. m = af^, that is = loga m, For let n = a'', that is v = logafi. and mn = a'^a" = + ", Then that is, logam= + v = logam + logan.
fj.

is

the

736

a'-'

/j.

Theorem

2.

the dividend

The logarithm, of a quotient is the logarithm of minus the logarithm of the divisor.

737

For
that

if

m=
rn/n
loga rn/n

a^^

and n

a",

we have
is,

= =

a'^/a"

a*^-",

fj.

loga i

log

378
738

A COLLEGE ALGEBRA
3.

Theorem

The logaritlun of any power of a numher

viultijylied

is

the

logarithm of the numbe)'

by the exponent of the

power.
For
if

m = ai^,
ni'"

we have
that
is,

logaJn""

= (a'^y = a'"", = r/x = r logatn.

739

Theorem 4. The logarithm of any root of a number is the logarithm of the number divided by the index of the root.
For
if

we have
that
is,

m = a", v m Va*^ = a*, s loga vm = ^/s = (logam)/s.

s
.

?i

740

is due to the properLogarithms of numbers to the base 10 have been computed and arranged in tables. If we avail ourselves of such a table, we can find the value of a product by an addition, of a quotient by a subtraction, of

The

practical usefulness of logarithms

ties established in

736-739.

a power by a multij^lication, and of a root by a division.

Thus,

log

6-

l^^

= log Vo + =
(log 5)

log

Ve -

log 325

1 736
3.

737

/7

(log 6)

/8-

25 log

738,739

have only to look up the values of log 5, log 6, and log 3 given in the table, then to reckon out 25 log 3, and finally to look up in the. the value of (log 5) /7 + (log 0) /8 table the number of which this value is the logarithm.

Hence, to obtain the value of

7,- 8^ v5 V()/3-5, we

EXERCISE LXII
1.

Find log24, log42, logv/28, log5G25, logs 729, logio.OOl, log2l/64,

log2.125,
2.

logaV^,

logs 128, loga^'a'^.

If logio2

.3010 and logio3

.4771, find the logarithms to the

base 10 of 12, 9/2, V2, Vq.

3.

Express logaOOO^ in terms of loga2, loga3, and logo 5.

LOGARITHMS

379

4. Express the logarithms of each of the following expressions to the base a in terms of log6, logaC, logad.
(1)

bh'^/dK
logs

(2)
'

-v^a-Vft'^^VftWa"^31 / 18.

5.

Prove that

V 81 "^^729
X

9"

:=:

6.

Prove that loga

+ Vx'-i - 1 = - Vx'-^ - 1

log,, (x

+ Vx2 - 1).

COMMON LOGARITHMS
For the purposes of Computation of common logarithms. numerical reckoning we employ logarithms to the base 10. Tliese are called common logarithms. In what follows log m
will

741

mean

logi?.
;

We have 10 = 1, .\ log 1 = 102 ^ 100, also log 100 = 2, 10-2 = .01, ..log.01=-2, .
.-.

10^

10"'

= =

10,

.-.

log 10
.1

.1, .-.

log

= =-

742

gers

Hence, for the numbers whose common logarithms are we have the table
:

ifite-

The numbers
their logarithms

.001,

.01,

.1,

1,

10,
1,

100,
2,

1000,

3,

2,

1,

0,

3, .

rical

Observe that in this table the numbers constitute a geometprogression in which the common ratio is 10, and the

logarithms an arithmetical progression in which the

difference
is 1.

common
743

The numbers

common

logarithms are rational, for

in this table are the only rationals whose all fractional powers of

10 are irrational. But, as we proceed to show, every positive number has a common logarithm, and the value of this logarithm
as

may be obtained correct to as many may be desired. If we extract the square root of 10, the
and

places of decimals

square root of the

result thus obtained,

so on, continuing the reckoning iu

380
each case to the
Jable
:

A COLLEGE ALGEBRA
fifth

decimal figure,

we

obtain the following

10' 10* 10'

10^"*

= = = =

3.16228,
1.77828, 1.33352,
1.15478,

= 1.07461, = 1.03663, 10'=' = 1.01815, 10"^ = 1.00904,

10^'

10^"

10"

10"^*

= 1.00451, = 1.00225, 10=^" = 1.00112, 10"'" = 1.00056,

and so

on, the results

obtained approaching 1 as limit as

we proceed (compare

725).

The exponents 1/2, 1/4,

on

the left are the logarithms of the corresponding numbers on the right.

744

By

aid of this table

we may compute

the

common

loga-

rithm of any number between 1 and 10 as in the following

example.
Example. Find the common logarithm of
4.26.

Divide 4.26 by the next smaller number in the table, 3.16228. The quotient is 1.34719. Hence 4.26 = 3.16228 x 1.34719.

Divide 1.34719 by the next smaller number in the table, 1.33352. The quotient is 1.0102. Hence 4.26 = 3.16228 x 1.33352 x 1.0102.

Continue thus, always dividing the quotient

smaller
If
g,i

last

obtained by the next

shall obtain

number

in the table.

denote the quotient in the nth division,

we

method an expression for 4.26 in the form taken from the table and g, the result being
4.26

of a product of

by this n numbers

= =
it

3.16228 x 1.33352 x 1.00904 x

10^

x g
n terms

108

lO^h

=
1

10^

+ s + 2h o

^^

n terms also always a part of the infiwhose sum is 1, 704, Ex. 1. Hence nite series 1/2 + 1 /4 + 1/8 + it approaches a limit which is some number less than 1, 192. Represent this limit thus 1/2 + 1/8+1 /250 + For each quotient as limit. Again, as n increases, q approaches 1 lies between the divisor used in obtaining it and 1, and as the process is
increases, the
increases.

As n

But

exponent 1 /2 remains less than

---

/8

1/256

to

1,

since

it is

continued the divisors approach 1 as limit. to + + + 10= ^ =Jo 4.26

Hence

l'.'"

tcrn,B^_^

= 10^ + s + sis +

and there-

fore log 4.26

l'/2"+

1/8 +

/256

.6294 ....

LOGARITHMS

381
745

From the common logarithms of the numbers between 1 and 10 we may derive the common logarithms of all other positive numbers by the addition of positive or negative integers.
Example.
1.

Find the common logarithms

of 42.6

and

.426.

We

have

382
749 In like manner,

A COLLEGE ALGEBRA
if in

in its integral part

left,

we

shift the decimal point

10"'^,

a number n which has but one jBgure jx places to the

that

is,

if

we multiply n by

we add
1

/x

to the

char-

acteristic of log n.

log .0426

Thus, log .426 2.6294, and so on, 745.

it

=-

log 4.26

1.6294,

In practice we find
characteristics
1, 2,

in the

convenient to write these negative form 9 10, 8 10, , and to

before the mantissa, and the place the positive part 9, 8, 10 after it. Thus, instead of T.6294 we write 9.6294 10.

Hence the
If

rule

n<l,

the cliaracteristic of log

is

negative.
to

To obtain

it,

subtract from 9 the

the right of the decimal point in n, then write the result before the mantissa,
O's

number of

immediatelg

and

10 after

it.

Thus, log .00426

If

7.6294

10, log

.000000426

3.6294

10.

more than nine but

less

than nineteen O's immediately follow the

decimal point, subtract their number from 19 and write the result before the mantissa and 20 after it and so on.
;

Example.

Given log 2

.3010, find the

number

of digits in 2^.

750

A table of logarithms.

The accompanying

table, pp. 384, 385,

all

contains the mantissas of the logarithms of

numbers of

three figures computed to the fourth place of decimals and arranged in rows in the order of their magnitude, the decimal

points before the mantissas being omitted. From this table we may also derive mantissas for numbers
of

more than three

figures
is

by aid of the principle:

is

When

a number

changed by an amount which

itself,
is

very small

in comparison with the number

rithm of the number number.

the change in the loganearly propoHional to the change in the

Numerical results obtained by aid of this table are not to

be trusted beyond the fourth figure. When greater accuracy is required we must use tables in which the mantissas are

LOGARITHMS

383

The student will given to more than four places of decimals. find it easy to procure a five-, six-, or seven-place table.
To
find the

logarithm of a number from this

table.

We proceed
;

751

as in the following examples.

Example

1.

Find the logarithm of .00589.

N
is

look up the first two significant figures, 58, in the column headed in the table, then run along the row to the right of 58 until the column

We

reached which

is

headed by the third

it) is

figure, 9.

We

there find 7701.

This (with a decimal point before

acteristic is 7

the mantissa sought.

The char-

10,

749.

Hence

log .00589

7.7701

-10.
46.

Example The mantissas

2.

Find the logarithms of 8 and

of these logarithms are the

460 respectively.

Hence, proceeding as in Ex.

log 8

same as those 1, we find

1.6628.

of 800

and

.9031,

log46

Example 3. Find the logarithm of, 4673. The mantissa is the same as that of log 467.3.
:

It

must therefore

lie

between mant. log 467 and mant. log 468. From the table we find mant. log 467 = 6693 and mant. log 468 = 6702, and the difference between these mantissas is 9. Thus if we add 1 to 467, we add 9 to mant. log 467. Hence if we add .3 of 1 to 467, we should add .3 of 9, or 3 approximately, to mant. log 467. Hence mant. log 467.3 = 6693 + 3 = 6696, and therefore
log .4673

9.6696

10.

Observe that until the characteristic

is

introduced

we omit

the decimal point which properly belongs before the mantissa. The method illustrated in Ex. 3 for finding the mantissa of
the logarithm of a number of more than three figures described as follows
:

may
to

be

Fro7n the table obtain m, the mantissa corresponding

first three figures, also d, the difference between

the

and

the next

greater mantissa.
Multi2Jly d
bij

decimal

2)oint before

the remaining part of the number with a it, and add the integral part of the product
is .5

(increased by 1 if the decimal part

or more) to m.

384

A COLLEGE ALGEBRA

LOGARITHMS

385

386
752

A COLLEGE ALGEBRA

To find a number when its logarithm is given. We have merely to reverse the process described iu the preceding
section.

Find the number whose logarithm is 5.9552 10. row marked 90 and iu the column marked 2. Hence the required sequence offigures is 902. But since the characteristic is 6 10, the number is a decimal with

Example

1.

We

find the mantissa 9552 in the table in the

5,

number

or 4, O's at the right of the decimal point, 749. is .0000902.

2.

Hence the required

7.5520.

Example
Looking

Find the number whose logarithm

is

in the table

we

the mantissas 5514 and

find that the given mantissa 5520 lies between 5527 corresponding to 356 and 357 respectively.

The

lesser of these mantissas, 5514, differs from the greater, 5527, by 13 and from the given mantissa, 5520, by fi. Thus, if we add 13 to the mantissa 5514, we add 1 to the number 356. Hence if we add 6 to the mantissa 5514, we should add 6/13 of 1, or .S

approximately, to 356. Hence the required sequence of figures is 3565, and therefore by the rule for characteristic, 748, the required number is 35650000.

therefore have the following rule for finding the sequence of figures corresponding to a given mantissa which is not in the table
:

We

Find from

the table the next lesser mantissa

corresponding figures, and d, the difference between next greater mantissa.

Subtract

m, the three m and the.

'

m from

the given ma.ntissa

to

and divide

the remainder

hy

d,

annexing the resulting figure

the three figures already

obtained.

753

Cologarithms.

The cologarithm

of a

number

is

the logarithm

of the reciprocal of the number.

= log 1 /? = log 1 log = log m, 733, Since colog we can find the cologarithm of a number by merely changing the sign of its logarithm. But to avail ourselves of

737,

the table
positive.

we must keep

We

the decimal parts of therefore proceed as follows:

all

logarithms

LOGARITHMS
Example
"We have
1,

38T

Find colog

89.2.

log

and

log 89.2

= =

10

10

1.9504

Hence

colog
2.

89.2= 8.0496-10

Example

Find colog. 929.

log 1

We
and

have

=
=

10

10
10 .0320

log 929
.

9.9680

Hence

colog. 929=

Hence we may find the cologarithm of a number from its logarithm by beginning at the characteristic and subtracting each figure from 9 until the last significant figure is reached,
which figure must be subtracted from 10. do or do not afiix 10 according as 10
to the logarithm.

To
is

this result
is

we

not or

affixed

In this

way when the number has not more

obtain
its

than three figures we

the table.

may

cologarithm directly from

Computation by logarithms.
serve to
ucts,

The following examples will show how expeditiously approximate values of prodquotients, powers, and roots of numbers may be obtained

754

by aid of logarithms (compare

Example
1.

740).
5.26.

Find the value of .0325 x .6425 x

Log
But

(.0325

X .6425 x

5.26)

log .0325

log .6425

log 5.26.

log .0325

8.5119

10

log
log
log of product Therefore the product

.6425= 9.8079-10

5.26=

.7210

Hence

=
is

19.0408

20

9.0408

10

.1099.

Example

2.

Find the value

of 46. 72/. 0998.

Log But

(46.

72/. 0998)

log 46.72

log 46.72

log

Hence log of quotient

Therefore the quotient
is

- log. 0998. = 11.6695 -10 .0998 = 8.9991 - 10 = 2.6704

468.2.

-;

388

A COLLEGE ALGEBRA
write log 46.72, that
its
it.

We
to

is

1.6695, in the

form 11.6695

10 in order
is

make

positive part greater than that of 8.9991

10 which

to be

subtracted from

Example

3.

Find the value of 295 x .05631

-^

--

806.

Log (295 X .05631 But

806)

log 295

log .05631

colog 806.

log 295 log .05631

= 2.4698 = 8.7506-10
7.0937

colog

806=

Hence log

of required result

=
is

18.3141

-10 - 20 =

8.3141

10

Therefore the required result

.02061.

Example 4. Find Log (.7929)6 = 6 X

But

the sixth power of .7929.

log. 7929.
log .7929

= =

9.8992

10

Hence
Therefore (.7929)6

log (.7929)0

59.3952

60

9.3952

10

.2484.

Example

5.

Find the seventh root of .00898.

Log
But

-s/.

00898

(log .00898)

7.

log .00898

= =

7.9533

- 10
-10

7)67.9533-70
Hence
log V. 00898

9.7076

Therefore V. 00898

.510.

Observe that when as here we have occasion to divide a negative logawe add to its positive and negative parts such a multiple of 10 that the quotient of tlie negative part will be 10.
rithm by some number,

Negative munbers do not have real logarithms to the base 10 since all real powers of 10 are positive numbers. If asked to find the vahie of an expression which involves negative
factors,

we may

first find

the absolute value of the expression

by logarithms and then attach the appropriate sign to the

result.

Thus, if the given expression were 456 x ( - 85.96), we should first find the value of 456 x 85.96 by logaritiims and then attach the sign to the
result.

LOGARITHMS
EXERCISE LXni
Find approximate values
1.

389

of

llie

following by aid of logarithms.

390
Example.

A COLLEGE ALGEBRA
Find the logarithm of .586
to the base 7.

Log. .586 ^'

^"g^"-^^^
logioT

= 9.1619-10 ^ _
.8451

2321
8451

^_

^^^^^

We reduce 9. 7679
756 757
is

10 to the

form

of a single negative

number, namely

.2321,

and perform the

i

final division

by logarithms.

When

a the formula gives log^a

l/log^b.
is

The only base besides 10 of which any actual use a certain irrational number denoted by the letter
is

made
whose

approximate value
connection.

2.718.

Logarithms to

this

base are
in another

called natu7^al logarithms.

We shall consider thejn

758

the

Equations in which Exponential and logarithmic equations. unknown letter occurs in an exponent or in a logarithmic

expression

may sometimes
1.

be solved as follows.

Example

Solve the equation 132^+5

(2

=
x

14^+".

Taking logarithms of both members,

o Solv ng, ^'
,
.

5) log

13

(x

7) log 14.

7 log 14 2

5 log 13 5

=
14

2.4532
1.0817

= +

2.268.

_^

2 log 13
2.

-log

Example

Solve the equation log

Vx

21

\ log x

= 1.

By

736, 739,

we can reduce
log

this equation to the

10.

form

Hence
Solving,

Vx(x-21) = 1 = log x2 _ 21 x = 100.

X

25 or

4.

Example

3.

Solve the equation x^^og*

2 (log x)^
log x

=
+

10 x.
1.

Taking logarithms,
Solving for log
x,

Hence

= = x =

log x
1

or

1/2.
1/VlO.

10 or

759

Compound Suppose that a sum of P dollars is put at compound interest for a period of n years, interest being compounded annually and the interest on one dollar for one
interest.

year being r. Then the amount at the end of the

first

year will be
it

P + Pr

or

P (1 +

r),

at the end of the second year

will

LOGARITHMS
I

391

be P(l + r) (1 + r) or P(l + ry, and so on. Hence, if denote the amount at the end of the wth year, we have

A = P (1
If interest be
if

ry.

quarterly,

= P (1

compounded semiannually, A and so on. -\- r/4)'*"

;

= P (1 +

-{-

r/2)-";

We

call

we can

find

P the 2)resent worth P by means of the

of

oi A.

li A, n,

and r be given,
(1

formula

P=

^4

r)~".

Example 1. Find the amount pound interest.

$2500 in eighteen years at 4% com-

We

have

log^ =

log 2500

18 log 1.04

3.7039.

Hence

A = $5057,

approximately.

Example
of $120
is

sum compound

At the beginning of each of ten successive years a premium 2. What is the worth of the paid on a certain insurance policy. of these premiums at the end of the tenth year if computed at 4%
interest ?
is

The required value

that
is,

by
^

* 701,

120 x 1.04 x

120 [1.04 +(1.04)2 + (1.04)W

^^

..

^(i.o4)io],
1

1.04-1
-=-

By

logarithms, (1.04)io

1.479.

Hence the required value is 120 x 1.04 x .479 puted by logarithms, gives $1494, approximately.
Annuities.

.04;

and

this,

com-

sum

of

money which

is

to be paid at fixed

760

intervals, as annually, is called an annuity. It is required to find the present worth of

an annuity of

dollars payable annually for n years, beginning a year hence, the interest on one dollar for one year being r. The present worth of the first payment is ^ (1 r)~\ that

of the second

and so on. Hence the present worth of the whole is, 701,

payment

is

(1

r)~^,

lit;

"^

(1

+
ry

"^

(1

+ ry\

^
7L^

~
(i

If the annuity hejwrpetual, that is, if n cc, then (1 r)" and the formula for the present worth reduces to A /r.

+ ry\' = oo,

392

A COLLEGE ALGEBRA

Example. What sum should be paid for an annuity of $1000 payable annually for twenty years, money being supposed to be worth 3% per

annum ?
The present worth, P,
is 1

.03

(L 03)20 J

By

logarithms,

we

find that (L03)2o

1
-1

1.803.

Hence P

1000 r, 1 .03 L

1000 X .803
.03

1.803J

X 1.803

^,,,, = $14845, approximately. ^^

'

"^

EXERCISE LXIV
1.

Find

logs 555, log7.0463, logioo47.

2.

Solve the following exponential equations.

(1) 3^

729.

(2)

a^-

+2

a^^.

(3)

213^

516-^+4.

3.

Solve the following logarithmic equations.

(1)

log X

log (X

3)

1.

(2) log x2

(3)

log

(1-2 xY -

log (3

x)3

6.

(4) xios^

+ log z = 2. = 2.
5% compound

4.
.

Find the amount of $7500

in thirty-five years at

interest, the interest

being compounded annually.

5. Find the amount of $5500 in twenty years at the interest being compounded semiannually. 6.

3% compound

interest,

Show

that a
it

sum

of

years and that

years at 5%
7.

will increase

money will more than double itself in fifteen more than a hundredfold in ninety-five

compound
sura will

interest.

What

amount

to

$1250

if

put at compound interest at 4%

for fifteen years ?

8.

A man
all

Invests $200 a year in a savings

deposits.

annum on

What

will be the total

bank which pays S\% per amount due him at the

end of twenty-five years ?

9. What sum should be paid for an annuity of $1200 a year to be paid for thirty years, money being supposed to be worth 4% per annum ? What sum should be paid were this annuity to be perpetual ?

10.

and
c

denote the length of the hypotenuse of a right-angled triangle the lengths of the other two sides, b = V(c + a){c ~ a). Given 586.4, a = 312.2, find b and the area of the triangle, using logarithms.
If c a, b

PERMUTATIONS AND COMBINATIONS

11.
s

393

If

ft,

6,

denote the lengths of the sides of a triangle and

(rt

Knd
12.

a) (s c)/2, the area of the triangle is ^s{s the area of the triangle in which a 410.8, b = 424, c

b) (s = 25.68.

c).

of

Find the area of the surface and the volume of a sphere the length whose radius is 23.6 by aid of the formulas S 4^ 7tr'\ V = i ffr^/3, assuming that tt = 3.1416.

XXV.
of

PERMUTATIONS AND COMBINATIONS

c,

Definitions of permutation and combination.

letters to be given, as a, b,

k,

Suppose a group denoting objects of

without regard to

761

any kind.

Any
order,

set of r of these letters, considered

is

more

called a combination of the n letters v at a time, or, briefly, an v-combination of the n letters.
shall use the

We

symbol C" to denote the number of such

d are
(J^
is,

combinations.
Thus, the'2-combinations of the four letters
ab, ar, ad, be, bd, cd.
a, &, c,

There are

six of these combinations, that

Ci

6.

in a definite order in a

the other hand, any arrangement of > of these Ji letters row is called a permutation of the n letters, r at a time, or, more briefly, an T-pe7'rmitation of the
letters.

On

We

shall use the

symbol

P".

to denote the

number
d are

of such

permutations.
Thus, the 2-permutations of the four
ab,
ac,

letters a, b,

c,

ad, be, bd, cd,

ba, ca, da, cb, db, dc.

There are twelve of these permutations, that is Pg = 12. Observe that while ab and ba denote the same combination, they denote
different permutations.

a. b,

In what has just been said it is assumed that the letters -, k are all different and that the repetition of a letter

394

A COLLEGE ALGEBRA
is

within a permutation or combination

the contrary
is stated.

not allowed.

will be the understanding throughout the chapter except

This where

762

have already preliminary theorem. apply the following principle, 554

:

We

had

ocf^asion to

ways, and if, when it has If a certain thing can be done in been done, a certain other thing can be done in n ivays, the entire number of ways in which both tilings can be done in tlie order
stated is

mn.
:

We reason thus
there are n
first

Since for each

way

of doing the first thing

ways of doing both things, for in ways of doing the thing there are mn ways of doing both things.

More generally, if a first thing can be done in m ways, then a second thing in n ways, then a third in p ways, and so on, the entire number of ways in which all the things can be done
in the order stated
is in

-n-p

-.

Example. formed with the

How many numbers

digits
1, 2, 3,

of three different figures each 9 ?

can be

We may clioose any one of the nine digits for the first figure of the number, then any one of the remaining eight digits for its second figure, and finally any one of the seven digits still remaining for its third figure. Kence we may form 9 8 7, or 504, numbers of the kind required.

763

The number

of r-permutations of

n different

letters.

By

the

reasoning employed in the preceding example that this number Pj! is given by the formula

we

readily prove

w (

1) (n

2)

to r factors.

(1)

forming an r-permutation of n letters we may choose any one of the n letters for its first letter, then any one of the remaining n \ letters for its second letter, then any one of
in

For

the n

2 letters still remaining for its third letter, and so on. Hence, 762, the entire number of ways in which we maj choose its first, second, third, rth letters, in other wordf

PERMUTATIONS AND COMBINATIONS

the entire

395

number
n

of

ways

in

which we may form an

(?i

tation with the

letters, is

1) (n
t^jB

2)

?'-permuto r factors.
c,

Thus, the numbers of permutations of

three, four, five at a time are

letters a, 6,

d, e

one, two,

Pi

5,

P|

4,

P-^

3,

P^

2,

P^

1.

Evidently the
is
11

(r

factor in the product. n (n 1) (ji Hence the formula (1) r \. 1), or n

/-th

2)

may be
(2)

written

P = n{n- 1) (n -v2) ...(ii-r + 1). = n, the factor n r \ is n n + \, When r V) n. 2 -1, ov 1 -2 we have PI = n{n 1) {n
-\!

or

1,

and
con-

The

n is called factorial n and is denoted tinued product 1-2 by the symbol n or \iu Hence the entire number of orders in which we can arrange n letters in a row, using all of them in each arrangement, is given by the formula
Pl

= nl
later, 775,
1.

(3)

For a reason which will appear symbol 0!

'^'

the meaningless

is

assigned the value

Example 1. How many different signals can be made with four flags of different colors displayed singly, or one or more together, one above
another
?

There

will

be one signal for each arrangement of the

flags

taken

1, 2, 3,

or 4 at a time.

Hence the number

is

Pj

P^

P3

P\, or 64.

Example
taken
(1)
all

2.

Of the permutations

of the letters of the

word fancies

at a time,

How many
first

begin and end with a consonant

place may be filled in 4 ways, then the last place in 3 ways, then the intermediate places in 5 ways. Hence the required number is 4 -3 -5!, or 1440.
!

The

(2)

How many
in the

have the vowels

in 4

The vowels may be arranged

sonants

odd places

even places ? even places in 3 ways, the conways, and each arrangement of vowels may
in the

in the

be associated with every arrangement of consonants.

Hence the required

number

is

!,

or 144.

396
(3)

A COLLEGE ALGEBRA
How many
c is

do not have c as their middle letter ? the middle letter in (i of the permutations, for the remaining letters may be arranged in all possible orders. Therefox'e the number of the permutations in which c is not the middle letter is 7 6 !,
Evidently
! !

or 4320.

Example
Example

3.

Show
If

that

P^

Pl,

and that

P'g"

P^.

4.
5.

P\"

127 P-g", find n.

Example

How many
In

passenger tickets will a railway company

?

need for use on a division on which there are twenty stations

Example
taken
all

G.

how many of the permutations of the

e,
i

letters a,
?

e, i, o,

u, y,

at a time, do the letters a,

7.

stand together

Example
ments

With

the letters of the

word numerical how many arrange-

each can be formed in which the odd places are -i occupied by consonants ?
of five letters
V,

Example
form P^'

8.

Show

that with the digits 0,

1, 2,

it is

possible to

Pg numbers, each of which has four different

9.

figures.

Example
Example
if

How many numbers

In

all

told

can be formed with the

?

digits 3, 4, 5, 7, 8, all the figures in

10.

each number being different

in a

how many ways can seven boys be arranged

is

one particular boy

not permitted to stand at either end of the

row row ?

764

Circular permutations.

The number

of different orders in

which n

different letters can be arranged about the circum-

ference of a circle or any other closed curve is (re 1)!. For the relative order of the w letters will not be changed if we shift all the letters the same number of places along the
curve.

Hence we
11

shall take account of all the distinct orders

of the

of the letters fixed in position and the r(;maining n 1 then arranged in all possible orders. ])Ut these n 1 letters can be arranged in (ii 1)!
letters if

we suppose one

orders,

7G3, (3).
7
!,

Thus, eight persons can be seated at a round table in

orders.

or 5040,

Example
n

1.

Show

that the

number

of circular r-permutations of

different letters is

P^/r.

PERMUTATIONS AND COIMBINATIONS

Example
of 180,
.

397

2.

Taking account of the

if

into coincidence with itself

show that (n
)i

1)

fact that a circular ring will come revolved about a diameter through an angle !/2 different necklaces can be formed by stringing

together

beads of different colors.

Example 3. In how many ways can a party of four ladies and four gentlemen be arranged at a round table so that the ladies and gentlemen

may occupy

alternate seats?
of different letters

Permutations

when

repetitions are allowed.

765

With

11

different letters

we can form
if

n''

arrangements or per-

mutations of r letters each,

a permutation.

allowed to repeat a letter within

For in forming a permutation of this kind we may choose any one of the n letters for its lirst letter, and then again, since repetitions are allowed, any one of the n letters for its second letter, and so on. Hence, 702, the entire number of to ways in which we can form the permutation is n -n- n
r factors, or
??'".

Thus, with the digits

1, 2, 3,

we can form

all

told 9^, or 729,

numbers

of three figures each.

1.

Example

How many numbers

In
is

of one, two, or three figures each

can be formed with the characters

1, 2, 3, 5, 7 ?

Example
boys
if

2.

how many ways can

three prizes be given to seven

each boy

eligible for e\ ery prize ?

The number
different.

of

n-permutations of n letters which are not

all

766

Let us inquire how many distinguishable permutations can be formed with the letters a, a, a, h, c{l), three of
alike, all the letters being

which are

used in each permiitation.

Compare these permutations with the corresponding permutations of the letters a,

a',

a", b, c(2), all of

which are

different.

take any one of the permutations of (1), as abaca, and, leaving b, c undisturbed, we interchange the a's, we get nothing new. But if we treat the corresponding permutation of (2),
If

we

namely

aba'ca'\ in a similar

manner, we obtain 3

distinct per-

mutations, namely aba'ca", aba"ca', a'ba"ca, a'baca", a"haca', a"ba'ca. Hence to each permutation of (1) there correspond

398

A COLLEGE ALGEBRA
(2).

3! permutations of (2) is 5 !, 763, (3).

of (1)
is 5!-=- 3!.

The number of the permutations of Therefore the number of the permutations

the reasoning here emjjloyed we can prove in general that the number of distinguishable w-permutations of n letters

By

of which

are alike, q others alike,

i

and so

on, is given

by the

formula
N-.

''

2)\q\---

Example

1.

In

how many
?

different

ways can the

letters of the

word

independence be arranged

Of the 12 letters in this word 4 are Hence the required result is 12 !/4

e's,
!

3 are n's, 2 are d's. 2 !, or 1,663,200.

word Antioch be arranged without changing the relative order of the vowels or that of the consonants ?

Example

2.

In

how many ways can

the letters of the

From

the proof just given

it is

arrangements is the same as it would be and the four consonants were the same.

evident that the required number of if the three vowels were the same

Hence

it is

!/3

!,

or 35.

Example

3.

How many
?

functions of five

terms has each of the following symmetric variables x, y, z, w, v, namely, 'Ex^y'^z, 'Zx^y'^z^, Zx^yzu,

and Zx^y^zhi^

terms of Zx^y-z once each if, leaving the expowe write under them every 3-permutation of the letters x, y, z, u, v. Hence the number of the terms is P^, or 60. If we apply the same method to "Ex^y-z^, we obtain the term x^y-z^ twice, once in the form x^y-z- and once in the form x^z-y'^. Similarly every term is obtained twice, once for each of the orders in which its letters under the equal exponents can be written. Hence the number of
nents
3, 2, 1 fixed in position,

We

shall obtain all the

terms

in

Zx^-Z"

is -P;]/2,

or 30.
!,

Similarly Ex-^yzu has or 30, terms.

P]/3

or 20, terms,

and Ex^y^z-u^ has P^/2

!2

!,

Example 4. In how many ways can five jjennies, six five-cent pieces, and four dimes be distributed among fifteen children so that each may
receive a coin
?

In a certain district of a town there are ten streets running north and south, and five running east and west. In how many ways can a person walk from tlie southwest corner of the district to the northeast corner, always taking the shortest course ?

Example

5.

PERMUTATIONS AND COMBINATIONS

The number
ber, C", is
of r-combinations of

399
767

n different

letters.

This num-

given by the formula

C"^

^ (n = P"^-i-rl = -^
,

77

1)
^-

(n
f

r + 1) =^

^-

(1)

For evidently

if

we were

to

form

all

the r-combinations and

were then to arrange the

letters of each

combination in turn

in all possible orders, >\e should obtain all the r-permutations. But since each combination would thus yield rl permutations, 763, (3), all the combinations,

C" in number, would

yield rl

x C" permutations. Hence r x C^ = P'^, and therefore C^

!

P"^ -^

rl
c,

Thus, the numbers of combinations of the three, four, five at a time are

letters a, b,

d, e one,

two,

r.._5

^1

r5_5-4

^1-2

^,,_ ^

5.4.3
1-2.3

,_ 5.4.3-2
*

5.4.3.2-1
"

1.2. 3. 4

1.2.3.4-5

This expression for C". is the coefficient of the (r + l)th term in the expansion of (a + b)" by the binomial theorem, 565.

This was shown in

560 by an argument which

is

merely
768

another proof of the formula (1). If in the expression just obtained for C"

numerator and denominator by (n symmetrical formula

C"

r)

we multiply both
obtain the more

!,

we

'"

r\{n-r)\

(2^ ^ ^

From this formula (2) it follows that the number of the /-combinations of n letters is the same as the number of the
(?i

769

For T" "-'

?')-combinations. = (^n

r)\{7i

{n
'-

=
r)']\

{ii-r)\r\

C".

This also follows from the fact that for every set of r-things
taken, a set of n
Thus, C\\ found in this

r things
.

is left.

- CV =

14

13/1

91.

Observe

how much more

(1).

readily

Cjj is

way than by

a direct application of

400

A COLLEGE ALGEBRA

Example 1. There are fifteen points in a plane and no ttoee of these Find the number of triangles which points lie in the same straight line. can be formed by joining them.
Evidently there are as

many

points taken three at a time. is, 15 -14 -13/1 2- 3, or 455.

Hence the number

triangles as there are combinations of the of triangles is C'3, that

Example

2.

In

how many ways can

a committee of three be selected

from ten persons

(I)

so as always to include a particular person

A?

(2)

so

as always to exclude

A?

from the (1) The other two members of the committee can be chosen remaining nine pensons in Cl], that is, 9 -8/1 2, or 36, ways. (2) The entire committee can be chosen from the remaining nine
persons in C^, that
is,

9- 8-

7/1 2a,
e,

3,

or 84, ways.

Example

3.

With the vowels

i,

and the consonants

6, c, d, /,

how many arrangements

of letters can be

made, each consisting of two

vowels and three consonants ?

The vowels for the arrangement can be chosen in C^ ways, the consonants in C3 ways then each selection of vowels can be combined with Hence every selection of consonants and the whole arranged in 5 ways. the required result is Cg C3 5 !, or 7200.
; !

Example 4. Tn how many ways can among three persons A, B, C ?

A's books can be selected in
C's in
C[;,

eighteen books be divided equally

C]'^

ways, then B's in C^^ ways, then

required result
is

or

1,

way.

Hence,

7G2, the

C'g**

C'^?

Cji,

or 18!/(6!)-i.

sets of 6

how many ways the 18 books can be distributed into three books each, we nnist divide the result just obtained by 3 !, which for here the order in whicli the three sets may chance gives 18 /(e) !)3 3 to be arranged is immaterial.
find in
!
!

To

Example

5.

With the
?

different sele(^tions

letters of the word mathematical how many and how many different arrangements of four letters

each can be made

As

the letters are not

all

different

we cannot
f(uC",.

obtain the required results

P'^.
I.

by single applications
Uv.ncii

of the formulas
;

and

TIk' letters are a, a, a

??(,

in

t,t; h,

e, i, c,

we may

classify
:

and then enumerate the possible

selections

and

arrangements as follows

PERMUTATIONS AND COMBINATIONS

401

Those having three like letters. 1. Combining the 3 a's with each of the seven other letters in turn, we or 28, arrangements. obtain 7 selections and 7 4 !/3 Those having two pairs of like letters. 2. 2 or 18, such arrangements. There are 3 such selections and 3 4 / 2 3. Those having two letters alike, the other two different.

!,

!,

Of such selections there are 3 Cg, or 63 of arrangements there are 63-4I/2!, or 756. 4. Those having four different letters. Of selections there are 0% or 70 of arrangements, 70 4 or 1680. Hence the total number of selections is 7 + 3 + 63 + 70, or 143 of
; ;

!,

arrangements, 28

18

756

1680, or 2482.
Cjl, C^9,

Example

6.

Find the values of

If

and

Cj^.

Example
Example Example

7.

CJ

CI',

find n.

8.

If 2

q=5
the

q,

find

ji.

9.

How many
lie in

planes are determined by twelve points,

?

no

four of which

same plane

Example 10. How many parties of five men each can be chosen from a company of twelve men ? In how many of these parties will a particular

man

A be

included

Fjom how many

will

A be

excluded ?

Example 11. Of the parties described in the preceding example how many will include two particular men A and B ? How many willinclude one but not both of them ? How many will include neither of
them
?

From twenty Republicans and eighteen Democrats how 12. many committees can be chosen, each consisting of four Republicans and
Example
three Democrats ?

Example 13. With five vowels and fourteen consonants how many arrangements of letters can be formed, each consisting of three vowels and four consonants?
Example 14. In how many ways can a pack of fifty-two cards be divided equally among four players A, B, C, D ? In how many ways can the cards be distributed into four piles containing thirteen each ?

Example

15.

How many

numbers, each of

five figures,

can be formed

with the characters

2, 3, 4, 2, 5, 2, 3, 6, 7 ?

402
770

A COLLEGE ALGEBRA

Total number of combinations.

561,

bers,

we set a we obtain

=i=
C\

If in the formula for {a + &)", and then subtract 1 from both mem-

Q+
,

+
of

C:

2"

1.

Hence the

total

number

combinations of n different

n at a time, in other words, the total number of ways in which one or more things may be chosen from n things, is 2" 1.
things taken one, two,

This may also be proved as follows Each particular thing can be dealt with in one of two ways, that is, be taken or left.
:

Hence
2 2

the total
-.to

number

of

ways
762.
left,

factors, or
all

2",

in

which

the things are

n things is Therefore, rejecting the case we have as before, 2" 1.

of dealing with all

Example

1.

How many

different

sums

of

money can be paid with one

dime, one quarter, one half dollar, and one dollar ?

Example 2. By the reasoning just illustrated, show that the total number of ways in which one or more things can be chosen from p + g + things of which p are alike, q others alike but different from the p things,

and so on,
^.^

is (j)

Example

3.

+ 1) (g + !) 1. How many different sums

of

money can be

paid with

two dimes,

five quarters,

and four half dollars?

771

In the expression for C,", namely l)/r!, the r factors of the numerator Hence for decrease while those of the denominator increase.
Greatest value of C.

n(n

V)-

(?i

a given value of n the value of C" will be greatest when the next greater value of r will make
{n

r 4-

l)/r

<

1.

From this
r

it

= n/2

readily follows that if n be even, C" is greatest when" and if 7i be odd, C" is greatest when r l)/2 (^n

-\-l) /2, the value of C" being the same for these two values of r, 769.

or r

=(n

Example.

What

is

the greatest value of C';? of C'/?

PERMUTATIONS AND COMBINATIONS

403
772

how many ways we

Let us inquire in Combinations when repetitions are allowed. can select three of the four digits 1, 2, S, 4
111, 112, 124,

when repetitions are allowed. As examples of such selections we may take

illustrating respectively the cases in

which

all three,

two, none

of the digits are the same. If to the digits in 111, in 112,

^

respectively,

we obtain

123, 124,

and in 124 we add 0, 1, 2 and 136, which are three of the

3-combinations tvithout repetitions of the digits 1, 2, 3, 4, 5, 6. And a little reflection will show that if we make out a complete list of the selections like 111, 112, 124, arranging the digits in each so that no digit is followed by one of less value,

and then

to the digits in each selection

add

0, 1, 2,

we

shall

obtain once and but once every one of the 3-combinations 1) or 6 digits 1, 2, 3, 4, 5, 6. without repetitions of the 4 (3

The number of the latter combinations number which we are seeking. The same reasoning may be applied

is

Hence 0% is the

to the general case of r-combinations, with repetitions, of the n numbers 1, 2, , n. And since the numbers may correspond to n different things of any kind, we have the theorem
:

The number of the x-comblnations loith repetitions ofn different things is the same as the number of the r-comblnations tvithout
repetitions o/ n

n(n

-f

l)---(n

r 1 different + r - l)/r!.
-{-

things, namely,

C"+r~S ^^

Example 1. How many different throws can be made with four dice ? As any one of the faces marked 1, 2, 3, 4, 5, 6 may turn uppermost in the case of one, two, three, or four of the dice, the number of possible throws is the number of 4-combinations with repetitions of 1, 2, 3, 4, 5, 6,
namely, C'^j^ or 126.

Example 2. How many terms has a complete homogeneous polynomial of the rth degree in three variables x, y, z? Evidently it has as many terms as there are products of the rth degree whose factors are x's, y's, or z's. Hence the number is = C7'+2 = C"+2 = C^ +
;:-i

(^

1) (r

2)/2.

404
773

A COLLEGE ALGEBRA
of combinations.

Formulas connecting numbers

The correspond-

The following relations are of special ing algebraic identities. interest and importance.
C

C'V'

C",!].

(1)

Por we may distribute the ^--combinations of 7i letters into those which contain some particular letter, as a, two classes, and those which do not contain this letter. We shall obtain all

the combinations of the

(r

first class,

once each,

if

l)-combination of

the remaining n
;

we form every
and then add
is C'"z}.

letters

to each of these combinations

hence their number

The combinations
the remaining n
(jm+n

of the second class are the r-combinations of

1 letters
;

hence their number

is C"~'^.

c-;!

-f c,"!^

CI

c.'n^

QH

\-

<^'T

<",-i

C';..

(2)

For take any group of m -f n letters and separate it into two groups, one of ??i letters, the other of n letters. We shall
take account of
each, if
all

the r-combinations of the

???

+ n letters,

once

we classify them as follows. They consist of The (a) The r-combinations of the letters of the i-group. number of these combinations is C"'. 1 letters of the (b) The combinations which contain m-group and one le.tter of the ?? -group. As we can choose the 1 letters in C^!!!i ways and the one letter in C" ways, the number of combinations of this kind is C^'"i C". 2 letters of the (c) The combinations which contain r and two letters of the n-group. As we can choose m-group
1

the r 2 letters in C^2 ways and the two letters in ways, -^.nd so the number of combinations of this kind is C,.l2 Qon, until last of all we reach the ?--combinations of the letters

of the w-group, of which there are

Thus, Cl

C^..

8i and Cl

C'lC^

C^C^

C^

10

-f

40

30

-|-

84.

774

and (2) we replace the several expressions in terms of m, n, r, 767, (1),

If in (1)

symbols C by their we obtain* formulas

connecting

vi,

n,

r.

The

proofs just given only

show that

PERMUTATIONS AND COMBINATIONS

these formulas hold good But in fact, so far as

405

when m,

n, r

denote positive integers.

algebraic identities,

and n are concerned, they are true holding good for all values of these letters.
we have
2)

This

may

be shown by algebraic reduction.

(1),

Thus, in the case of

n -1
'

nn-i

'-^

(n

1) (n

(n

r)

(n

1)

(n

2)

(n

1)

l-2.-.r
(n
1) (n
1
.

2)

(?i

1 )

l-2...(r-l) n r'

(r

1) 1)

['-^]
^
^

n (n

1)

(n

-r+

1.2.--r

But it is not necessary to make such a reduction to prove that these formulas are true identities. Thus, when expressed
in

terms of m,

n,

r,

each

member

of (2) denotes an integral

function of vi and n whose degree with respect to each of these These two functions must be identically equal, letters is r.
since otherwise, were we to assign some particular integral value to m, thus making them functions of n alone, they could not be equal for more than r values of w, 421, whereas it has

already been shown that in reality they would be equal for integral values of n.

all

EXERCISE LXV
1.

If there are three

four from
2.

roads leading from P to Q, two from Q to i?, and to S, by how many routes can a person travel from P to iS?
a

In

how many ways can

seats ?

company

of five persons be arranged in

six

numbered
3.

If eight

runners enter a half-mile race, in how

many ways can

the

first,

second, and third places be

won
all

4.

In

men and
5.

how many ways can a in how many ways can

four-oar crew be chosen from ten oarsthese crews be arranged in the boat ?
soldiers

three

From a company of one hundred men can be chosen ?

how many

pickets of

6. Five baseball nines wish to arrange a schedule of games in which each nine shall meet every other nine three times. How many games must be scheduled?

406
7.
all

A COLLEGE ALGEBRA
In

how many ways can

the digits

1, 2, 1, 3, 2, 1, ?

5 be arranged,

the digits occurring in each arrangement

8.

Of the permutations of the

letters in the

word factoring, taken

all

at a time, (1)
(2)
first

how many

how many
9.

begin with a vowel and end with a consonant? do not begin with/? (3) how many have vowels in the

three places ?

In

how many
ct,

retain the order

o, i ?

of the permutations just described do the vowels In how many do the consonants retain the order

In how /, c, i, r, )i, ^ ? retain these orders?

10.

many do both
word

the vowels

and the consonants

With the
?

letters of the

resident

how many
first,

permutations of

five letters

each can be formed in which the

third,

and

fifth letters

are vowels
11.

In

how many ways can

a baseball nine be selected from fifteen

candidates of

whom
?

six are qualified to play in the outfield only

and nine

in the infield only

12.

In

how tiany ways can two numbers whose sum

1, 2, 3, 8, 9,

is

even be chosen

from the numbers

13.

10

How many

numbers

of one, two, or three figures can be

7 (1)
?

formed

with the digits

(2)

1, 2, 3, 4, 5, 6,

when

the digits

may

be repeated?

when they may not be repeated

14.

How many

odd numbers, each having

1, 2, 3, 4, 5,

five different figures,

can be

formed with the

15.

digits

3000 and 8000

16.

How many odd numbers without repeated digits are there How many of these are divisible by 5 ? ?
how many ways can how many ways can
a person invite one or more of

between

In

five friends

to dinner ?
17.

In

fifteen apples

be distributed

among

three

boys so that one boy

18.

shall receive six, anotlier five,

and another four?

negative signs be

In

how many ways can

?

six positive

and

five

written in a row
19.

How many numbers

of four figures each can be

7 ?

formed with the

characters
20.

1, 2, 3, 2, 3, 4, 2, 4, 5, 3, 6,

seven

From fifteen French and twelve German books eight French and German books are to be selected and arranged on a shelf. In how
this be

many ways can

done ?

PERMUTATIONS AND COMBINATIONS

21.

407

From
done
In
?

a complete suit of thirteen cards

five are

to be selected

which
this be

shall include the king or queen, or both.

In

how many ways can

five

22.

how many ways can

four

men

be chosen from

Americans
(2)

and
least

six

Englishmen so as one Englishman ?

to include (1) only

one Englishman?

at

23.
lines is

How many

met by another

parallelograms are formed when a set of ten parallel set of twelve parallel lines ?

24.
line,

except

Given n points in a plane no three of which lie in the same straight m which all lie in the same straight line. Show that the
of lines obtained

number
25.

by joining these points

of bracelets that

is

C'^

C +

1.

Find the number

together five like

26.

can be formed by stringing pearls, six like rubies, and five like diamonds.
ten persons be arranged at two round

In

how many ways can

each table
?

tables, five at

27.

In
of

game
28.
for

how many ways can six ladies and five gentlemen arrange a lawn tennis, each side to consist of one lady and one gentleman ?

In how many ways can fifteen persons vote to fill a certain office which there are five candidates ? In how many of these ways will the vote be equally divided among the five candidates ?
29.

boat crew consists of eight

side only

row on the stroke

men two of whom are qualified to and one on the bow side only. In how many

ways can the crew be arranged ?

How many baseball nines can be chosen from eighteen players of ten are qualified to play in the infield only, five in the outfield only, and three in any position?
30.

whom
31.

Show
all at

that the a time,

taken

number when two

6
!

of permutations of six different letters of the letters are excluded each from a
!

particular position,
32.

is

!.

How many

combinations four at a time can be formed with the

letters p, q,

r, s, t, v,

when

repetitions are allowed

33.
34.

How many
How many
Show

different

throws can be made with

five dice ?

terms has each of the symmetric functions S*y%%.,

the

IfX'^y'^z'^u, I,x^y^z'^u^v,

number

of the variables being ten ?

of terms in a complete homogeneous function of the nth degree in four variables is {n 1) (n + 2) (n + 3) /S!
35.

that the

number

408

A COLLEGE ALGEBRA

XXVI.
775

THE MULTINOMIAL THEOREM

Let a

Multinomial theorem.
nomial, and n a positive
^

+b+

k denote any

poly,

integer.

Then
Kl

al pi

"where the

sum on
/3,

values of a, such that a

the right contains one term for each 'set of n, , k that can be selected from 0, 1, 2,

being understood that when a = 0, a! -s to be replaced by 1, and the like for (3, , k. For (a -\- b -\+ k)" denotes the continued product

/S

+ K = n,

it

(a

-{-

b -{-'

k) (a-\- b

-\-

1-)

to

factors,

and each of the

form

partial products obtained by actualty carrying out this multiplication, without collecting like terms, has the

a letter from the first parenthesis, times a letter from the second, times a letter from the third, and so on. But since the letter selected from each parenthesis m^ay be
:

any one of the letters a, b, , k, a list of the products as thus written would also be a complete list of the 7i-perniutations of

the letters a,b,---, k when repetitions are allowed. a,ft,---,K denote any particular set of numbers 0, 1, ,

And
?i

if

sum

is n,

there will be in the

factors are a's,
/?

list as

many
/>'s,

products in
of

whose which
A-'s,

a of the

of

them

, k

them

as

there are n-permutations of n letters of which a are alike, /3 others are alike, and so on, namely, 7i /n- /3 And k !, 76G.

since each of these products

is

equal to a%^

k",

their

sum

is

n'
.

a
j3l

0'

al

Ki
is

The binomial theorem

Thus, the expansion of {a
five

a particular case of this theorem.

b

e)*

consists of terms of the

!
!

a* with the coefficients 4 !/l 1 1 1 I, or types abed, a^bc, a'^b'^, 24; 4!/2!l!l!, or 12; 4!/2!2!, or 6 4!/3!l!, or 4; 4!/4!, or 1, terms of the same type, we have respectively. Hence, uniting
!

a%

(a

(i

e)*

2a4

2a6

Sa'-ft^

12 -Ea-bc

24 Eabcd.

PROBABILITY
Example. Find the
coefficient of x^ in the

409
expansion of (2+3x4-4x2)8.
8!
is
or !/3
!

2''3^4>' x3 + 2 v, !7 are those for which 8 (1), and the terms required where a + ^ + 7 = 5 (2). complete list of the solutions of (1), (2) in positive |3 -f- 27 Hence the required 4, 3, 1; 5, 1, 2. integers orO's is a, )3, 7 = 3, 5,

The general form

of a term of this expansion

coefficient is

'

23

35

3151

-^

2*

33

4!31v

-^
5!2!

25

42,

or 850,752.

EXERCISE LXVI
1.

Give the expansion of {a

type.

+
6

+
c

-\-

d)3,

collecting terms of the

same
2.
3.

Also the expansion of (a

d)^.

Find the

coefficients of a^ft^c^d, a^b^c*,

d)i2.

and

a^hH"^ in the

expansion

of (a
4.
5.
6.

+c+

Find the
Find the

coefficient of ah'^c^d^ in the

coefficient of a^h^c in the coefficient of x^ in the coefficient of
x''

expansion of (a

dy^.

expansion of (a
(1 (1

36

2 cf.

Find the Find the

expansion of expansion of

7.

in the

+ x + x2 + x^)!". x + 3x2)9.

XXVII.

PROBABILITY

SIMPLE EVENTS
Consider any future event which, if given a an opportunity to happen, must happen or fail to happen in one of a limited number of ways all equally likely, that is, ways so related that there is no reason for
Probability.

776

trial,

that

is,

The turning expecting any one of them rather than any other. of the ace uppermost when a die is thrown is such an event.
there

For one of the six faces of the die must turn uppermost, and is no reason for expecting any one face to turn rather
than any other.

410

A COLLEGE ALGEBRA

can happen or

Calling all the equally likely ways in which such an event fail the possible cases with respect to the event,

the ways in which it can happen the favorable cases, and the ways in which it can fail the unfavorable cases, we say
:

The probability or chance of the event is the ratio of the number of favorable cases to the entire number of possible cases, favorable and unfavorable.

Hence if m denote the number of possible cases, a the number of favorable cases, and p the probability, we have by
definition

p
m

a I m.
when a
die
is

Thus, the probability that the ace will turn up = 6 and a = 1. 1/6 for here
;

thrown

is

five balls, three

Again, the chance of drawing a white ball from a bag white and two black, is 3/5.
1.

known to

contain

777

Corollary
is 1
;

If an event

is certain to

happen,
;

its probabiliti/

if
its

it is

certain to fail, its probability is

j^'t'oper

in every other

case

probability is a positive

fraction.

the event is certain to happen, there are no ways in can fail; hence a and a/m 1. If the event is certain to fail, there are no ways in which it can happen hence a = and a/m = 0. In every other case a is greater
if
it

For which

=m

than

and

less

than

7n,

so that

a/m

is

a positive proper

fraction.

778

Corollary 2.

If the probability that an event

ivill

p, the probability that it will not hajjjjen is 1

happen

is

p.

For if a of the possible cases favor the occurrence of the a cases favor its non-occurrence. event, the remaining Hence the probability that the event will not happen is

m =

(m
779

a) Odds.

/m =

If the

a/m, number
and

certain event is a

1 p. of favorable cases with respect to a the number of unfavorable cases is b,

we

say,

when a

> b,

that the odds are a to 6 in favor of the

PROBABILITY
event
;

411

when

when a

b,

b a, that the odds are b to a against the event; In the first that the odds are even on the event.

>

case the probability of the event, namely, a / (a

b), is

greater

than 1/2; in the second

is

it is

less

than 1/2

in the third it

equal to 1/2.

Thus, if a ball is to be drawn from a bag containing five balls, three white and two black, the odds are 3 to 2 in favor of its being white, and
3 to 2 against
its

being black.

Expectation,

li of

a certain

sum

money M,

denote the chance that a person will win the product Mp is called the value

780

of his expectation so far as this

sum

is

concerned.
is

Thus, the value of the expectation of a gambler who throws an ace with a single die is 12 x 1/G, or $2.

to

win $12

if

he

Examples

of probability.

bility, 776, care

must be taken

In applying the definition of probato reduce the possible cases

781

The following examples will to such as are equally likely. illustrate the need of this precaution.
Example
1.

If

that the result will be two heads?

two coins be tossed simultaneously, what is the chance two tails? one head and one tail?
:

We
the
result

first result,
is

might reason thus There are three possible one the second, one the third hence
;

cases,
tlie

one favoring chance of each

1/3.

false, since the number of equally likely For if we name the coins and B possible cases is not three but four. tail, B tail ; head, B head respectively, the equally likely cases are And since one of these cases favors head, B tail; tail, B head.

But our conclusion would be

the result two heads, one the result two tails, and two the result one head and one tail, the chances of these results are 1/-1, 1/4, and 2/4
respectively.

Example two dice ?

2.

"What

is

the chance of throwing a total of eight with

face of one die

i,

Here the number of equally likely possible cases is 6 0, or 36, for any may turn up with any face of the other die. We have a total of eight if the faces which turn up read 2, 6 or 3, 5 or 4. But there are two ways in which 2, 6 may turn up, namely 2 on the

412
die

A COLLEGE ALGEBRA

A and 6 on the die B, or vice versa. Similarly there are two ways in which 3, 5 may turn up. On the contrary, there is but one way in which Hence there are five favorable cases. Therefore the 4, 4 can turn up. chance in question is 5/36.
Example
dice
if

3.

What

is

the chance of throwing a total of eight with three

at least one die turns ace

up

The number

We
or

of equally likely possible cases is 6 6 6, or 216. have a total of eight if the faces which turn up read 1, 1, 6 or

1, 2,

1, 3, 4.

But
6

1, 1,

numbers

1, 1,

may

turn up in 3 !/2 !, or 3, ways, 766 for the be distributed among the three dice in any of the

may

orders in which

1, 1, 6 can be written. Similarly 1, 2, 5 and 1, 3, 4 may each turn up in 3 !, or 6, ways. Hence there are 3 + 6 + 6, or 15, favorable cases. Therefore the chance in question is 15/216, or 5/72.

Example
balls.

4.

An

urn contains

six white,

four red, and two black

(1) If four balls are drawn, what is the chance that all are white ? There are as many ways of drawing four white balls as there are 4-combinations of the six white balls in the urn, namely C^. Similarly, since the urn contains twelve balls all told, the total number of possible drawings is C'4. Hence the chance in question is C^l/ C^^, or 1 /33. (2) If six balls are drawn"what is the chance that three of them are white, two red, and one black ? The three white balls can be chosen in Cg ways, the two red balls in C* ways, the one black ball in Cf ways. Hence the number of ways in which the required drawing can be made is C3 Cg Cj. The total number of Hence the chance in^^uestion is C'l- C^- C{/ C^g, possible drawings is C'^.

or 20/77.

Example
(1)

5.
is

Three cards are drawn from a

suit of thirteen cards.

the chance that neither king nor queen is Aside from the king and queen there are eleven cards.

What

drawn ? Hence there are

Therefore C'g sets of three cards which include neither king nor queen. the probability in question is C\J / C]^, or 15/20. (2) What is the chance that king or queen is drawn, one or both?

This event occurs when the event described

in (1) fails to occur.

Hence

the probability in question is 1 - 15/26, or 11/26, 778. (3) Wliat is the chance that botli king and queen are drawn?

We obtain
if

we combine each

every set of three cards which includes both king and queen of the remaining eleven cards in turn with king and
in question
is

queen.

Hence the chance

ll/C'g, or 1/26.

PROBABILITY

413
782

On the various meanings of probability. 1. The fraction ajm, which we have called the probability of an event, 776, means nothing so far as the actual outcome of a single trial, or a small number of trials, of the event is concerned. But it does indicate the frequency with which the event would occur in
the long run, that
series of trials.
is,

in the course of

an indefinitely long

Thus,

if

one try the experiment of throwing a die a very

great number of times, say a thousand times, he will find that as the number of throws increases the ratio of the number of

times that ace turns up to the total number of throws approaches the value 1/6 more and more closely.
life is

There are important classes of events the duration of one to which the definition of 776 does not apply, it being impossible to enumerate the ways, all equally likely, in which the event can happen or fail. But we may be able to determine the frequency with which events of such a class have
2.

occurred in the course of a very great number of past trials. If so, we call the fraction which indicates this frequency the Like 1/6 in the case of probability of an event of the class.
the die,
class
it

may

indicates the frequency with which events of the reasonably be expected to occur in the course of a

very great number of future trials. Thus, if we had learned from the census reports that of 100,000 persons aged sixty in 1880 about 2/3 were still living
in 1890, we should say that the probability that a person now sixty will be alive ten years hence is 2/3. 3. But we also use the fraction a Jm to indicate the strength

of our expectation that the event in question will occur on a s'luijle trial. The greater the ratio of the number of favor-

able cases to the

number

of possible cases, or the greater the

frequency with which, to our knowledge, events of a similar character have occurred in the past, the stronger is our expectation that this particular event will occur on the single trial

under consideration.

414

A COLLEGE ALGEBRA
speak of the probability, in this sense, of any kind Thus, before a game between two football and B, we hear it said that the odds are 3 to 2 in

We may

of future event.

teams, A favor of A's winning, or that the probability that will win is 3/5. This means that the general expectation of A's winning is about as strong as one's expectation of drawing a white ball

from an urn known

white.

to contain five balls three of

which are

EXERCISE LXVII
1. The probability of a certain event is 3/8. Are the odds in favor of the event or against it, and what are these odds? What is the probability that the event will not occur ?

2.

The odds
is

What
3.
is

the chance of his winning the

are 10 to 9 in favor of A's winning a certain game. game ? of losing it ?

The odds

are 5 to 3 in favor of A's winning a stake of $G0.

What

his expectation ?
:

4. The French philosopher D'Alembert said "There are two possible cases with respect to every future event, one that it will occur, the other that it will not occur. Hence the chance of every event is 1 /2 and the

definition of probability is meaningless."

5.

How should

he be answered

An
If

urn contains sixteen balls of which seven are white, six black,
red.
is

and three
(1)

a single ball be drawn, what

is

the chance that

it

is

white?

black? red?

two balls be drawn, what (2) one white and one red ?
If
(3)

the chance that both are black?

what is the chance that one white, one black, one red ? (4) If four balls be drawn, what is the chance that one the rest not ? two white and the other two not ?
If

three balls be drawn,

all

are red?

none red

is

white and

(5)

If ten balls

be drawn, what
?

is

the chance that five are white, three

black,
6.

and two red

What
What

is

the chance of throwing doublets with two dice

with

>

three dice?
7.
is

Show

that this

the chance of throwing a total of seven with two dice? is the most probable throw.

PROBABILITY
8.

415

What
?

is

two dice
9.
J

of tlirowing one ace

letter
is
is

the chance of throwing at least one ace in a throw with and but one ?

One

banter.
10.

What

the chance that the

taken at random from each of the words factor and same letter is taken from each ?

box contains nine tickets numbered 1, 2, 9. If two of the drawn at random, what is the chance that the product of the numbers on tliem is even ? odd ?

tickets be

Vil.
1, 2,

(3)

If five tickets be drawn from this box, find the chance (1) that and 3 are drawn (2) that one and but one of 1, 2, and 3 is drawn that none of these numbers is drawn.
;

12.

If
is

four cards be

drawn from a complete pack

of fifty-two cards,
?

what

the chance that they are ace, king, queen, queen, and knave of the same suit?

and knave

ace, king,

lis. What

is

the chance that a

hand

at whist contains four

?

trumps and

three cards of each of the remaining suits

14.

What

is

the chance of a total of five in a single throw with three

dice ?
15.

of a total of less than five ?

If eight

persons be seated at a round table, what

sit

is

the chance that

two particular persons

together

'

COMPOUND EVENTS. MUTUALLY EXCLUSIVE EVENTS

Independent events.
of

Two

or

more events

are said to be inde-

783

pendent when the occurrence or non-occurrence of any one

them
rest.

is

the

not affected by the occurrence or non-occurrence of In the contrary case the events are said to be

interdependent.
Thus, the results of two drawings of a ball from a bag are independent is returned after the first drawing, but interdependent if the

if

the ball

ball is not returned.

Theorem 1 The prohahilitij that all of a set of independent events will occur is the produc t of the probabilities of the single
.

784
.

events.

'"^

For consider two such events whose probabilities are i/mi and a^]m^ respectively.

416
Tlie

A COLLEGE ALGEBRA
number
event
of equally likely possible cases for and against for and against the second vx^, and since

the

first

is ?i,

the events are independent any one of the m^ cases may occur with any one of the m^ cases. Hence the number of equally
for likely possible cases

and against the occurrence of both

the same reasoning, a^a^ of these Therefore the probcases favor the occurrence of both events.
events
is ^riimz.

And by

ability that both events will occur

''

is
711

~
>

that

is

-^

>

as

im2

nil

^2

was to be demonstrated. The proof for the case

of

more than two events

is similar.

applies only to events of the kind described in 776, but for the reasons indicated in 782 we may apply the theorem itself to any kind of future event, as
in Ex. 2 below.

The demonstration

Thus, the chance of throwing ace twice in succession with a single die
is

1/6 X 1/6, or 1/36.

Again, the chance of twice drawing a white ball from a bag which contains five white and four black balls, the ball first drawn being returned
before the second drawing,
is

5/9 x 5/9,

or 25/81.

785

If the probability of a first event is pi, and if nfter this event has happened the probability of a second eveiit is P2, the j^robability that both events will occur in the order Theorem
2.

stated is piP2.

And

similarly for more than

tivo events.

This theorem

may

be proved

in

the

same manner

as

Theorem

1.

It evidently

includes that theorem.

Thus, after a white ball has been drawn from a bag containing five white and four black balls, and not replaced, the chance of drawing a

second white
ball

ball is

4/8.

when

the one
1.

first

Hence the chance of twice drawing a white drawn is not replaced is 5/9 x 4/8, or 5/18.
the chance that ace will turn up at least once in
?

Example

"What

is

the course of three throws with a die

Ace will turn up at least once unless it fails to turn in every throw. The chance of failure in a single throw being 5/6, the chance of failure in all three throws is 5/6 x 5/6 X 5/6, or 125/216. Hence the chance of
at least

one ace

is 1

125/216, or 91/216.

PROBABILITY
Example The chance
problem
2.

417
problem
is

The chance

that
it

A
is

will solve a certain

3/4.
?

that

will solve
if

2/3.

What

is
it

the chance that the

will be solved will

both

and B attempt

independently

The problem
A's
is

be solved unless both

failure

1/4 X 1/3, or 1/12.

is

solved

1/4, of B's failure 1/3. Therefore the chance that the problem will be --11/12.
is

and B fail. The chance of Hence the chance that both fail

There are two purses, one containing five silver coins and 3. one gold coin, the other three silver coins. If four coins be drawn from the first purse and put into the second, and five coins be then drawn from the second purse and put into the first, what is the chance that the gold coin is in the second purse ? in the first purse ?

Example

The chance that

into the second
left in the
is

the gold coin

is

C\/ C% or 2/3,
is

781,

taken from the first purse and put Ex. 5. The chance that it is then
is

second purse
it

^ / C\^ or 2/7.

both drawings chance that it

is

in the

second purse
is

Hence the chance that after 2/3 x 2/7, or 4/21. The

is

in the first purse

4/21, or 17/21.
is

Example

4.

If eight coins

that at least one of

them

will

be tossed simultaneously, what turn head up ?

the chance

If are hunting quail. Example 5. Four men A, B, C, and gets on the average one quail out of every two that he fires at, B two five out of every seven, out of every three, C four out of every five, and wliat is the chance that they get a bird at which all happen to fire

simultaneously
6.

An urn A contains five white and four red balls. A secExample ond urn B contains six white and two black balls. What is the chance of drawing a white ball from A and tlien, this ball having been put into B, of drawing a white ball from B also ? 'o
Example
B, 5/6.
alive ?
7.

The chance that

is

What

A will be alive five years hence is 3/4; the chance that five years hence both A and B will be
dead,

alive,

B dead ? A

alive ?

both dead

Mutually exclusive events.

related that but one of

If two or more events are so them can occur, they are said to be

786

mutually exclusive.
Thus, the turning of an ace and the turning of a deuce on the same throw of a single die are mutually exclusive events.

>

-1

418
787

A COLLEGE ALGEBRA
3.

Theorem

The prohahility that some one

extents

of mutually exclusive
'probabilities

will

occur

is

or other, of a set the sum of the

of the single events.

For consider two mutually exclusive events A and B. The possible cases with respect to the two events are of three kinds, all mutually exclusive, namely, those for which (1) .4
happens, B fails (2) .1 fails, B happens (3) A fails, B fails. Let the numbers of equally likely possible cases of these three kinds be I, ni, and n respectively. Then
;

(a) ^ ^

The chance that

I

either

6v

happens ^^
I

is
I

For there are

(b) ^ ^

}n

n possible and

+m
1

+m+
r7i)

favorable cases.
;

The chance

of the single event ^

is

+ (m +

For since A never happens except when B fails, the I cases in which A happens and B fails are all the cases in which A happens, and the m + n cases in which A fails and B happens
or both

A and B

fail are all

the cases in which

fails.

(c) Similarly the chance of the single event

B is

m + (^ + ^0
sum

But

^l^L_ = ^_^i + ^+ m + + m n + (? n) (^
I -\-\-

-^

?i)

Therefore the chance that either

ov

B happens

is

the

of the chances of the single events A and B. The proof for more than two events is similar.
Thus, if one ball be drawn from a bag containing four white, five black, and seven red balls, since the chance of its being white is 1/4, and that of its being black is 5/16, the chance of its being either white
or black
directly
is 1/4 + 5/16, or 9/16. Of course this from the definition of probability, 776.

result

may

be obtained

In fact that definition

may

be regarded as a special case of Theorem

3.

Care must be taken not to apply this theorem to events which are not mutually exclusive.
will

Thus, if asked, as in 785, Ex. 2, to find the chance that a problem be solved if both A and B attempt it, A's chance of success being

PROBABILITY

419

3/4 and B's 2/3, we cannot obtain the result by merely adding 3/4 and 2/3, since the two events A succeeds, B succeeds are not mutually excluThe mutually exclusive cases in which the problem will be solved sive. are A succeeds, B fails A fails, B succeeds A succeeds, B succeeds.
:

The chances of these cases are, 784, 3/4 x 1/3 or 3/12, 1/4 x 2/3 or 2/12, 3/4 x 2/3 or 6/12; and the sum of these three chances, or
11/12,
is

the chance that the problem will be solved.

1.

Example

and a second urn

An urn A contains ten balls three of which are B contains twelve balls four of which are white.
random and a
?

white,
If
is

one
the

of the urns be chosen at

ball

drawn from

it,

what

chance that the ball

is

white

We
(2)

exclusive events

are required to find the chance of one of the following mutually and then drawing a white ball from it; (1) choosing
:

choosing B and then drawing a white ball from it. The chance of choosing A is 1 / 2, and the chance when A has been chosen of drawing a white ball is 3/10. Hence the chance of (1) is

1/2 X 3/10,or3/20.

Similarly the chance of

is

(2) is 1

/2 x4/12,orl/6.

Therefore the chance in question

3/20

1/6, or 19/60.

Example
to have

2.

What

is

the value of the expectation of a person

who

is

any two coins he may draw at random from a purse which contains five dollar pieces and seven half-dollar pieces?
of his expectation so far as it depends on drawing two dollar $2 x C'^/C'| = $2 x 5/33 = .30; on drawing two half-dollar pieces, $1 x C^/ Cj" = $1 x 7 /22 = .32 on drawing one dollar piece and one half-dollar piece, $1.50 x 5 7/ Cf = $1.50 x 35/66 = $.80. Hence the total value of his expectation is $.30 -|- $.32 -|- $.80, or

The value
is

pieces

$1.42.

Example
at a time
balls

3. Two persons A and B are to draw alternately one ball from a bag containing three white and two black balls, the drawn not being replaced. If A begins, what chance has each of

being the

first

to

draw a white

ball ?

The chance that A succeeds in the first drawing is 3 / 5. The chance that A fails and B then succeeds is 2/5 x 3/4, or 3/10, for when B draws, the bag contains four balls three of which are
white.

The chance that A

or 1/10, for

fails,

B fails, and A then succeeds is 2/5x1/4x3/3,

all

when

draws, the bag contains three balls

is

white.
is

Therefore A's total chance

3/5

-f

1/10, or 7/10, and B's

3/10.

420
Example
chances of
4.

A COLLEGE ALGEBRA
A
and B
In the drawing described in Ex. 3 what are the respective if the balls are replaced as they are drawn ?

On

the

first

their chances
will be the

round A's chance is 3/5, B's 2/5 x 3/5, or 6/25; and on every later round, of which there may be any number,
as these.

same

Hence
"^

their total chances are in the ratio

is

3/5 6/25,
:

or 5

that

is,

A's total chance

5/7, B's 2/7.

Example 5. In a room there are three tables and on them nine, ten, and eleven books respectively. I wish any one of six books, two of which If a friend are on the first table, three on the second, one on the third. select a book for me at random from those in the room, what is the
it is

chance that

one of those
of

wish

running horses enters for a certain race two ^^ What horses whose chances of winning are 1/2 and 1/3 respectively. is the chance that he will obtain the stakes ?

Example

6.

An owner

Example
which
is

7.

and

B throw
the one

alternately with

two dice

for a stake

to be

won by

who

first
if

their respective chances of

winning

throws a doublet. throws first?

What

are

788

The following theorems Repeated trials of a single event. are concerned with the question of the chance that a certain
event will occur a specified number of times in the course of a series of trials, the chance of its occurrence on a single trial

being known.

789

Theorem 4. If the jjrohahUlty that an event tvill occur on a single trial is p, the probahilitij that it ivill occur exact/// r times in the course ofn trials is C"p'"q"~'', where q 1 p.

For the probability that

it

will occur on all of

ular set of r trials and fail on the remaining n 1 - />, 784. 2f{l -py~'', ov 2'''(f-'', if 7

any

partic-

r trials is

But since there are

??.

trials all told,

we may

select this

particular set of r trials in C" mutually exclusive.

ways which, of
C'^.p''q"~'',

course, are

Hence the probability

Thus, the chance that ace a single
die, or that

in question is
will turn

787.

out of
is

five
(1)2

up exactly twice in five throws with dice thrown simultaneously two and but

two

will turn ace up,

C^

(^y, or 625/3888.

PROBABILITY
Observe that
sion of
(j}

421

C^p'if''' is the

y)"

by the binomial theorem

term containing/)'' in the expanlor C" = C'l^.

7
.

Theorem
at least
r

r+

5. The probability tliat such an event will occur times in the coarse of w trials is the sum of the first 1 terms in the exjjansion of (^ -\- q)", namely,

790

pn

C-Jp^-'q

C5p-2q2

C^'l.pV^
it

For the event will occur

times or exactly any the terms ^j", Cip"~^q,
r

at least r times if

occurs exactly
r,

number

of times g)\'eater than

and

C",./''y"~'"repres'^'it

of the occurrence of the event exactly n, respectively, 789.

^
[

the probability 1, , r times

five tlirows

Thus, the chance that ace will turn up at least twice in the course of with a single die is
U)'

5 id' f

10 (l)MD^

10 (D^

(p\

or ^^,%%

in which A's skill is win as many as three such games in a se'. of five ? A's chance of winning a single game is 2/3, of losing 1 /3. Hence the chance that A will win as many as three of the five games is the sum of the

Example 1. be drawn and

Two

persons

and

are playi;*g a game which cannot twice B's. Wliat is the chance that

will

first

three terms of (|
2.

J)5,

that

is,

Q)^

5 {V}* i
1

10 (|)3
is

{i-)2,

or 64/81.

Under the conditions of Ex. Example will win three games before B wins two ?

what

the chance that

The chance in question is that of A's winning at least three of the first four games played and this chance is (?)* + 4(1)^ J, or if. And, in general, the chance of A's winning m games before B wins n is
;

the

same

as the chance of A's

winning at

least

of the first

m+

games played.

Example 3. Ten coins are tossed simultaneously. What is the chance that exactly six of tliem turn heads up ? that at least six turn heads up ? Example
4.

If

that exactly three turn ace

four dice be thrown simultaneously, what is the chance up? that at least three turn ace up?
in

Example
that

5.

Under the conditions stated

least four of the five

Ex.

what

is

the chance

will

win at
6.

games played ?
is

Example

Under the same conditions what

the chance that

A will

win four games before

wins one

422

A COLLEGE ALGEBRA
J

EXERCISE LXVni
five

1.

bag contaihs three white,

Mack, and seven red

balls.

On

the understanding that one ball is drawn at a time and replaced as soon as drawn, what are the chances of drawing (1) first a white, then a red, then a black ball ? (2) a wldte, red, and black ball in any order whatsoever ?
2. What is the chance of obtaining a white ball in the first only of three successive drawings from this bag, balls not being replaced ?

e
3.

What

is

the v jlue of the expectation of a person

at
ra^

who

is

allowed to

draw two coins

dom from
(t

a purse containing
?

five fifty -cent pieces,

four dollar pieces,

4.

ai

three five-dollar pieces

tF'it a certain door is locked is 1/2. The key to one of a'liunch of eight keys. If I select three of these keys at random and go to the door, what is the chance of my being able to open it ?

The chance
is

the door

5. There are three independent events whose chances are 1/2, 2/3, and 3/4 respectively. -^Vhat is the chance that none of the events will occur ? that one and but one of them will occur ? that two and but two

will occur ?
6.

that

all thret will

occur

in

Find the odds against throwing one of the a single throw with two dice.
7.

totals seven or eleven

What
8.

What are the odds against throwing a total of ten with three dice ? are the odds in favor of throwing a total of more than five ?

numbered
is

Three tickets are drawn from a case containing eleven tickets 11. AVhat is the chance that the sum of their numbers 1, 2, twelve ? What is the chance that this sum is an odd number?

9.
if

Two
is

seven

gamblers A and B throw two dice under an agreement that thrown A wins, if ten is thrown B wins, if any other number is
to be divided etjually.

thrown the stakes are

10.

Compare

their chances.

The same two gamblers play under an agreement

that

is

to

win if he throws six before B throws seven, and that B is to win if he throws seven before A throws six. A is to begin and they are to throw
alternately.
11.

Compare

their chances.

Three gamblers A, B, and C put four white and eight black balls into a bag and agree that the one who first draws a white ball shall win.

PROBABILITY
If

423

when

they draw in the order A, B, C, what are their respective chances the balls drawn are not replaced ? when they are replaced ?

12.

What

is

the worth of a ticket in a lottery of one hundred tickets

having
13. balls

five prizes of $100, ten of 50,

and twenty

of

.$5 ?

into

A bag A contains five balls one of which is white, and a bag B six none of which is white. If three balls be drawn from A and put B and three balls be then drawn from B and put into A, what is the
is

chance that the white ball

14.

in

A?
balls

The bag

contains

contains n balls b of which are white.

ball

the same that

by drawing a single ball it would be if

a of which are white, and the bag B Is the chance of obtaining a white from one of these bags chosen at random
all

the balls were put into one bag

and a

single ball then

15.

drawn

In a certain town
first.

January January
16.

What

is

five deaths occurred within ten days including the chance that none of the deaths occurred on

first?

on the average two persons out of three aged sixty live to be seventy, what is the chance that out of five persons now sixty at least three will be alive ten years hence?
If

17.

boy

is

lems set him.

If eight

able to solve on the average three out of five of the probproblems are given in an examination and five
is

are required for passing, what

the chance of his passing ?

if

person with two dice, a dollar until he throws seven.

19.

18.

is

to receive a dollar
if

he throws seven at the

first

throw

he throws seven at the second throw, and so on

is

What

the total value of his expectation ?

wins on the average three games out In playing tennis with B, What is the chance that he will win a set from B by the score of four.
of six to three ?

What

is

the total chance of his winning a set from B,

the case of deuce sets being disregarded ?

20.

Under

chance has
against
21.

the conditions described in the preceding example what of winning a set in which the score is now four to two

him ?

Two

gamblers
.$32.

player has staked

and B are playing a game of chance and each They are playing for three points, but when A has
one they decide
to stop playing.

gained two points and they divide the $64 ?

How

should

424

A COLLEGE ALGEBRA
XXVIII.

MATHEMATICAL INDUCTION
A
number
of

791

Mathematical induction.

the

formulas cou-

tained in recent chapters may be established by a method of It is illustrated in the proof called mathematical induction.

following example.
Example.
Prove that the sum of the
first

n odd numbers

is n^.

We

are asked to

show
1

that

-..

+ (2n-l) =

n2.

(1)

We see by
k, so

inspection that

Suppose that we
that

true for certain values of n, as 1 or 2. have thus found it true when n has the particular value
(1) is

1
is

...

(2A;

1)

A;2

(2)

known
A;

to
1,

be true.

or 2

to

Adding the next odd number, nan\ely, both members of (2) and replacing A;^ + 2 A; +
l

2
1

(A;

1)

1,

by

[k

1)^,

we

obtain

...

(2A:

l)

(^-

l)2.

(3)

But (3) is what we get if in (1) we replace n by A: + 1. We have therefore shown that if (1) is true when n has any particular value A;, it is also true when n has the next greater value k + \. But we have already found by inspection that (1) is true when k has the particular value 1. Hence it is true when n = l + l, or 2 hence when n = 2 + 1, or 3 and so on through all positive integral values of n, which is what we were asked to demonstrate.
;

true for n

formula involving n has been found and we can demonstrate that if true for n = k it is also true for n = k -{- 1, we may conclude that it is true for all positive integral values of n. For we may reason: Since it is true when n = 1, it is also true when n = 1 + 1,

And,

in general, if a

or 2

hence when w

1,

or o

and

so

on through

all posi-

tive integral values of n.

As another

illustration of this

method we add the following

proof of the binomial theorem.

For small values of n we
(a

find

by actual multiplication that

C^a^'-'^h^

6)

a"

C'i'a"-'i)

Cj'.a"

'

6'-

(1)

THEORY OF EQUATIONS
Multiplying both
(a

426

members
i

of (1)

by a

-\-

b,

we

obtain, 773, 1,
C';.

6)

a +

+ + +

C'l

a"b

C!;

an-ib^-...

art-r

+ ljr

CI

C" + ^a( + i)-'-6'-+

(2)

But (2) is the same as (1) with n replaced by n + 1. Hence if (1) is true when n = A;, it is also true when n = k = 1. It is therefore true when (1) is known to be true when n or 2 therefore when n = 2 + 1, or 3 and so on.
; ;

+
?i

1.

But
1

1,

Since the formula C" + C^"i = C"\^' can be proved independently of the doctrine of combinations, 774, the proof of the binomial theorem nere given is independent of that doctrine.

EXERCISE LXIX
Prove the truth of the following formulas, 701, 712, by the method of mathematical induction.
1.

2. 3.

4.

+ ar2 + + 32 + 13 + 23 + 33+ .. 1 + 3 + 6 + +
ar
12

+
+

ar -1

=
{n

(1

r")

(1

r).

2-'2

+ 1) (2 n + 1) /O. + 713 112(^ + 1)2/4. n(n + l)/2!=:n(n + l)(n +

n2

= =

2)/3!.

XXIX.

THEORY OF EQUATIONS

THE FUNDAMENTAL THEOREM. RATIONAL ROOTS

The two standard forms
degree in
x.

of

the

general equation of the nth

792

Every rational integral equation involving a single unknown letter, as x, and of the nth degree with respect to that letter, can be reduced to the standard form
a^x" + aiX"-^ H a_iX + a = 0. When the coefficients a^, !,, a are given numbers,
\-

(1)
(1) is

called a numerical equation, but when they are left wholly undetermined, (1) is called the general equation of the Tith

degree.

426 The

A COLLEGE ALGEBRA
final coefficient is often called
call

the absolute term.

an equation of the form (1) complete or incomplete according as none or some of the coefficients ai, a^,---, a are 0. Observe that in a complete equation the number of the terms

We

n + 1. In what follows, when all the coefficients Oo, !, a are real numbers, we may and shall suppose that the leading one
is

fto is positive, and when they are which have no common factor.

rational, that they are integers

By dividing both members of (1) by a^ second standard form

X-

we reduce

it

to the

Kx--^

j^^a;

which the leading For many purposes

in

coefficient is 1,

(2) is

and so on. the more convenient form of the

and

+ j^ = 0, h^ = ai/a^,

(2)

equation. In the present chapter it is to be understood that /(a-) denotes an equation of the form (1) or (2).

=
are
"'

793

Roots

of equations.

The

7^oots

of the equation f(x)

the values of x for which

332, 333.
It is

the

polynomial f(x) vanishes,

call the roots of

sometimes convenient to

794

the equation the roots of the polynomial. From the definition of root it follows that when is one of the roots of /(a-) is also that an equation/(a-) all of whose coefficients are positive can have no positive

root,

and that a complete equation f(x)

whose

coefficients

are alternately positive and negative can have no negative

root.

Thus, 2x3
2
a;3

x2

x2

cannot vanish when x

Q cajj jjave no positive root since the polynomial is positive and 2 X'' x^ + 3 x - 1 =
;

can have no negative root since 2x3 X is negative.

x^

+ Sx

cannot vanish when

795

Theorem
by X

divisible by

root of f (x)= 0, then f(x) is exactly conversely, iff(x) is exactly divisible 0. h, then b is a root of f (x)
1.

// b
b;

is

and

THEORY OF EQUATIONS
For, by

427

bis f(b).
is

413, the remainder in the division of f{x) by this remainder But when i is a root of f{x)

f{b)

0,

793, so that f(x)

is

exactly divisible by x
a;

b;

and conversely, when f(x) is exactly divisible by remainder f{b) is 0, so that 6 is a root of f{x) =
Example.
1

^',

the

0.

Prove that 3
9

is

a root of /(x)

-2 +0 -9[3
3

Dividing x^
ically, 411,
is 0.

= x^ - 2 x^ - 9 = 0. - 2 - 9 by x - 3
x'^

synthet-

3
3,

we
3
is

find that the remainder /(3)

0=/(3)

Hence

a root of /(x)
is

0.

If & is a root of
\)j

b,

f(x) =

0,

so that f{x)
</>

exactly divisible

796

and we

call the

quotient

{x),

we have

f(x)

= (x-b)<f>(x).
are the values of x
;

Hence the remaining roots of f(x) = for which the polynomial (x) vanishes
<f)

in other words, they

(x)

are the roots of the depressed equation

(f>

0,

341.

Example.
1 1 1

Solve the equation x^

3 x^

5x

= 0.

_3 +5
_
2

311 ~
3

ing x3

We see by inspection that 1 is a root, and divid- 3x- + 5x - 3 by X - 1, we obtain the

2x

Z2

+3, "~0
1, 1

depressed equation x^

0.

The

roots

of this quadratic,

found by

631, are 1
I

given equation are

Vi, and

V2. V2.
i

Hence the

roots of the

We

shall

assume now and demonstrate

later that evert/

roof.

797

rational integral equation f

(x)=

has at least one

795 we deduce the following theorem, often called the fundamental theorem of algebra.

From

this

assumption and

Theorem

2.

Every equation of the nth

degree, as
a

798
0,

f (x)

aox"

aix-i

+ a_ix +

has n and but n

roots.
is

By
Call

it

797 there
)8i.

Then f{x)

a value of x for which f{x) vanishes. - ^1, 795, is exactly divisible by a:

aQx''-'^.

the leading term of the quotient being

Hence
(1)

= {x-p,){a,x^-' + ). f{x)

428

A COLLEGE ALGEBRA
the same reasoning, since there is a value of which the polynomial a(^x"~^ + vanishes,
a-,

By
/3o,

call it

for

we have
(1),

a,x"''

(a;

^2) {a,x"-'13,)

f(x)

= (X == a, (X

(X

+ .). Therefore, by - 13,) (oa;"-- + )

{X- p,)---{^?

(2)

Continuing thus, after u divisions we obtain

f{x)

(30

Pn)-

(3)

have thus shown that n factors of the first degree exist, p,, x fi,,- ,^ A* of which/(a;) is the product namely, x and by 419, /(./) can have no other factors than these and
;

We

their products. But since a product vanishes

when one

of its factors vanishes

and then only, X = fti, or /3.2, numbers Pi, (3,,

799

it

follows from (3) that f{x) vanishes

or
/?,

and then only.

Hence,

793, the

when n
it

-,

/3

are roots of the equation /(a-)

and

has no other roots than these.

From this theorem it follows that the problem of solving an is equation f{x) = essentially the same as that of factoring the polynomial f(x). Also that to form an equation which
shall

have certain given numbers for

its roots,

we have merely
x,

to subtract each of these

numbers

in turn

from

and then to

equate to
Example.
It is (X

the product of the binomial factors thus obtained.

Form
2)
(.c

the eqviation whose roots are

2,

1/2,

/2)

(cc

1)

(a;

0)

0,

or 2 x*

3x5

1, 0. - 3 a;2 + 2x = q

800

Multiple roots.

Aj
13,

we

If A- '"^y be equal. call /? a )iiiiltipje root.

yS is

Observe that two or more of the roots /3i, two or more of them are equal to

And

according as the number of

in general, r, we call y8 a , double root, a triple root, in general, a root of order r. A simple root may be described as a root whose order r is 1. Evidently it follows from 798 that
tlie

roots equal to

two, three,

The condition that

f (x) be divisible by (x

(3 lie

py

a root of order
but not by (x

0/

f (x)

is

that

/8)'

+^

THEORY OF EQUATIONS

429

When we say, therefore, that every equation of the nth. degree has n roots, the understanding is that each multiple times. It is of course not root of order r is to be counted
/

true that every equation of the th degree has

roots.

different

Thus, x^
since x^

3 x2 + 3 X 1 = 3x2+3x l = (x
the
rational

is

1)^,

an equation of the third degree each of its roots is 1.

of

but

'_

On
f(x)

finding

roots
.

numerical

equations.

Let

801

a^yX"

ttix""^

a,,

denote an equation with

integral coefficients, and let b denote an integer and b/c a, rational fraction in its lowest terms. It follows from 451,

795 that if b is a root of /(a-) = and from 452, 795 that if b /c

of o
b

0,
is

then

i is a factor of a
is

a root, then S

a factor

and

c is

a factor of

a^.

Hence, in particular,

if a^

Ij

/c cannot be a root unless c = 1, that is, unless b /c denotes the integer b. Hence the following theorem, 454
:

An
ai, ,

+ a 0, where ajX''"' equation of the form x" a denote integers, cannot have a rational fractional

root.

It follows from what has just been said that all the rational roots of an equation with rational coefficients can be found by a limited number of tests. These tests are readily made by

803

synthetic division.
Example.

Find the rational

3 x5

roots, if any, of the equation

8 X*

x2

12 X

Tlie only possible rational roots are 8

1,

+ 4 = 0. 2, 4, 1/3, 2/3, 4/3.

see

12

14
2,

4 [2 4
[2

We
it is

by inspection that

is

not a root.

Testing 2, we find that a root and obtain the depressed

2
I

equation

3x^-2 x3-4x2-7x-2=0.
is

- 1/3

We

find that 2

a root of this

3x'

4x2

+ 4x +

0.

depressed equation also and obtain the second depressed equation This equation can have no positive root since

430
all its

A COLLEGE ALGEBRA
Testing 1, we find that it is not a root a root and obtain the third depressed

terms are positive, 794. 1/3, we find that it equation x^ + x + 1 = 0.

Testing

is

Hence the
remaining

roots,

Its rational roots of the given equation are 2, 2, 1/3. found by solving x'^ + x + 1 0, are ( 1 i v3)/2.

803
if

The reckoning involved in making these tests will be lessened one bears in mind the remark made in 453 also the fact
;

that a

number known not

to be a root of the given equation

cannot be a root of one of the depressed equations, finally the following theorem
:

796

and

If b

is 2)ositive

and

the sigyis of all the coefficients in the

b synthetically are plus, f (x) result of dividing f (x) iy x can have no root greater than b ifh is negative and the signs
;

just mentioned are alternately plus

and minus,
b.

f (x)

can

have no root algebraically

less

than

For

it

follows from the nature of synthetic division that in

both cases the

in the result

effect of increasing b numerically will be to increase the numerical values of all coefficients after the first

coefficient, that

without changing their signs, so that the is, the remainder, cannot be 0.

final

Example
2
2

1.

Showthat 2x3+3x2 4 a; + 5 =
Dividing by x
cients

has no root greater than

1.

+3 4 +5[1
5
1

_
2
If

only.
1.

obtain positive coeflSHence there is no root greater

1,

we

+5
we

+1,

than

divide by x

2,

we

obtain a result with larger coefiicients,

all

positive,

namely, 2
2.

10, 25.
3.x''

Example
3 3

Showthat
2
I

+ 4x2-3x + l = +

has no root less than -2.

2,

+4 - 3 +1 6 +4 2

If

Dividing by x

we

obtain coeflBclents

+1,-1
divide by x

which are alternately plus and minus. there is no root less than 2.
3,

Hence

we

we obtain

coefficients

with the same signs as those

just

found but numerically greater, namely, 3

12,

35.

804

add that any number which is known to be algeis called a braically greater than all the real roots of/(cc)=

We may

THEORY OF EQUATIONS
known
is

431
is

superior limit of these roots, and that any number which to be algebraically less than all the real roots oi f(x)
called an inferior limit of these roots.

Thus, we have just proved that

2a;^

is

of

+ 3a;2_4x + 5 = and 3a;3 + 4x2-3x + l = 0.

that

is

a superior limit of the roots of an inferior limit of the roots

EXERCISE LXX
1.

Form

the equations v?hose roots are

a,

2.

-b,a + b. (2) 3, 4, 1/2, 3 is a triple root of the equation Show that x4 + 8x3 + 18x2 -27 = 0.
(1)

1/3,

0.

3.

Show

that 1

and 1/2 are double roots

4 x5

of the equation

23 x?

33 x2

17 X

0.

the method of 803 find superior and inferior limits of the real roots of x^ 5 X* 5 x' + 4 X- 7 X 250 = 0.
4.

By

5.

Show

that 2 x*

3 x^

4 x^

10 x

has no rational root.

Solve

Each
them.
6. 8.

of the following equations has

one or more rational roots.

x^

x2

14 X

24
l

= =

0.

7. 9.

x^

2 x2

25 x

3x3-2x2 + 2x +
X*

0.

2x*

7x3

2x2

+ 50 = 0. - x = 0.

10. 11.
12.

4 x3

8 x2

8X

=
6

0.

2x*
3x*
x5

+ +

7x3

+ 4x2- 7x-

=
6

0.

11x3

9x2

+ 11X+
71 x2

=
+

0.

13. 14. 15.

9 X*

2 x3

81 X

70
4

=
0.

0.

2x5
x5

-8x4 +

7x3

+ 5x2- 8x +

16.
17.

12 x*

+ 3x*- 15x3- 35x2 + 54x + 72 = - 32 x3 + 13 x2 + 8 X - 4 = 0.

10x3

0.

x5-7x4 +
2 x<

18.

17 x3

25 x2

+ 18x2- 27X-27 =0. + 74 X - 120 = 0.

432
19.

A COLLEGE ALGEBRA
4 x5
x5

x''

6 x2

13 X

20. 21.

8x*

.3x'5

80x2

+ = 0. _ y2x + 240 =

0. 0.

2 x5

22. 23.
24. 25.

6x*

+ -

11 X*

89x3

+ 25 x2 + 1(3 x + 4 = + 359x2 _ 254x + 48 = 0.

2.1

X-

10 x<

3Gx*
12x5
2 x6

+ +

41 x^

46 x2

20 X

=
6

0.

108x3
20 X*

+ 107x2- 43x +
29x3

0,

77x2

+ 69x +

18

0.

26.
27. 28.

+
+ -

7 x5

8 x*

7 x3

2 x2

-14 x -

12

0.

2x6
5 x

11x3
7 x5

+ 24x* + 22x3 - 8x2 _ 33X - 18 = - 8 X* - x3 + 7 x2 + 8 x - 4 = 0.

0.

RELATIONS BETWEEN ROOTS AND COEFFICIENTS

805
Relations between roots and coefficients.

whose
form,

roots are

^Si, /^o,

/3 is

"When an equation reduced to the second standard

792, (2), the identity in

a;"

798,
-\

(.3),

becomes
h

h^x"-'^

h^x"-j80 {X

h^x"'^

\-

= {x-

p,) (x-fi,)... (X

p).

Carry out the multiplications indicated in the second ber and arrange the result as a polynomial in x, 559.

memThen

equate the coefficients of like powers of x in the two members, We thus obtain the following relations between the 284. coefficients h^, lu, and the roots ^y, ji^,--, (S,,: h,^

-^i = A + A + A + --- + h = M. + A/Sa + + A/?3 + + A,-! A,> -h = A/82/83 + M:P, + A,-2 A,-l/8n,
)8,.

(1)
(2)

(3)

(-l)''&

/3,/3,/3,...A

(n)

where the second members of (2), (3), represent the sum of the products of every two of the roots, of every three, and so on,

THEORY OF EQUATIONS

433

and the sign before the first member is plus or minus according as the number of the roots in each term of the second member Hence the theorem is even or odd.
:

Theorem.

In every equation reduced

X"

to

the form
f-

806

bix"-i

box"-^

-\

b =

0,

the coefficient bi of the second term, with its sign changed, is equal to the sum of all the roots; the absolute term b, with its sign changed or not according as n is odd or even, is equal to
the product of all the roots ; and the coefficient b^ of each intermediate term,, loith its sign changed or not according as r is

odd or even,
the roots.

is

equal

to the

sunt of the products of every r of

Before applying this theorem to an equation whose leading coefficient is not 1 we must divide the equation by that If the equation be incomplete, it must be rememcoefficient.

bered that the coefficients of the missing terms are

0.

Thus, without solving the equation 3 x" G x + 2 = 0, we know the Reduced to the proper following facts regarding its roots /3i, p^, l^sform for applying the theorem, the equation is x^ + x^ 2 x + 2/3 = 0.
Hgiicg
/3i
*

/32

^3

0,

/3i,'3o

^1^3

^2/33

=-

2,

/3i^2/33

=-

2 /3.

find the

If all but one of the roots of an equation are known, remaining root by subtracting the sum of the

we can known
if

807

roots

its sign changed from b^, or by dividing b, with odd, by the product of the known roots.

is

Example.

Two
is

of the roots of 2 x^

3 x^

23 x

12

are 3 and

4.

What

the remaining root?

root
is

The remaining

- 3/2 -

[3

+ (-

4)]

= - 1/2

or again,

it is

6-3(-4) = -l/2.

When
cients.

relation, a corresponding relation

the roots themselves are connected by some given must exist among the coeffifind this relation

808

To

we apply

the theorem of

806.

434
Example
1.

A COLLEGE ALGEBRA
Find the condition that the roots
of x^

+ px^ +

gx

+r=

shall be in geometrical progression.

Representing the roots by

a / p,
h
/3

a, a/3,

we have

(X
/3

I-

a+

a/3

= p,

ct'^

a^

a-/3

q,

ex a a^ =

r.

The

third equation reduces to a*

r,

whence a
0.

r.

Dividing the second equation by the

result,

first,

and simplifying, we have

2.

q^

ph
+
8 x^

substituting

a=

V r in the.

Example
that
it

Solve the equation x'

5x

50

0,

having given

has a double root.

/3,

Representing the roots by a, a,

2

we have
2 a/3

a+ ^=-

8,

a2

5,

a2/3

60.

= - 1 /3, ^ = - 22/3. The values a = 5, = 2 satisfy the equation a2/3 = a = 1/3, = 22/3do not .satisfy this equation. Hence the required roots are 5, 5, 2.
a=5,
/3

Solving the

first

and second
/3

of these equations for

a and
50,

/3,

we

obtain

and a

but the values

/3

809

Symmetric functions
roots to

of

the roots.

The expressions

in the

which the several

coefficients are equal, 805, are

sym-

metric functions of the roots, 540. It will be proved in 868 that all other rational symmetric functions of the roots can be

expressed rationally in terms of these functions, and therefore rationally in terms of the coefficients of the equation.

Example

1.

Find the sum of the squares of the roots of the equation

5

2x3-3x2- 4xa^

0.
)3,

Calling the roots a,

7,

we have

^ + y^ = (a + ^ + y)^ - 2(ap + py + ya) = (3/2)2 + 4=6^.

2.

Example
is

If the roots of x'

px^

+
?

gx

are a,

/3,

7,

what

the equation whose roots are ^y, ya, a/3

If p', q', r'

denote the coefficients of the required equation,

we have

-P' = ^1 + 7a + a/3 = 7, g' = /37 7a + 7a a/3 + a/3 ^7 = a/37 (a + + 7) = (- r) (- p) = - r' = Py ya ap = (a/37)2 = r*.

/3

ri),

Hence the required equation

is

x^

qx"^

+ prx r* = 0.

THEORY OF EQUATIONS
EXERCISE LXXI
1.

435

Two
The
;

of the roots of 2 x^

- 7 x2 +

10 x

are 1

find the

third root.
2.

progression
(1)
3.

roots of each of the following equations are in geometrical find them.

0.

8x3-14x2-21x4-27 =
The
find them.

(2)

x^

x2

+ 3x +

27

0.

roots of each of the following equations are in arithmetical

;

progression
(1) x3

6x2

+ 7x if

0.

(2) x3

9x2

+ 23x -

15

0.

4.

Show

that

one root of x^

+ px2 +

qx

he the negative of

another root, pq
5.

r.

Find the condition that one root of

x^

+ px- +

qx

shall

be

the reciprocal of another root.

6.

Solve X*

4.x3

10x2

12x

0,

having given that

it

has

two double
7.
its

roots.

Solve the equation 14x3

13x2 - 18x +

0,

having given that

roots are in harmonical progression.

8.

that two of

Solve the equation x* - x^ - 56 x2 + 36 x + 720 = 0, having given 3 and that the difference between its roots are in the ratio 2
:

the other two roots

9.

is 1.

If a, p,

7 are the roots of x^

+ j9x2 +

qx

0, find

the equa-

tions

whose roots are

(1) (3)
(5)

-a, -p, -y.

1/a,
a^,
If a,

(2)
(i)

ka, k^, ky.

l//i,

1/7.

a +

k,

+ k,y +

k.

p\ y\
/3,

(6)

_
x2

l/a2, -1//32,

I/72.

10.

+ 1 = 0, find the values of (2) a3 + /33 + 7^(1) a2 + ^1 + 1 / nr/3. (4) a^2 + ^^^2 + ^.^,2 + ^^i 4. ^^a + a27. (3) 1 / ^7 + 1 / 7a + of x^ 2 x2 + x 3 = 0, find the values of 11. If a-, 7 are the roots
7 are the roots
-y2.

of 2 x^

4x

/3,

(1)

a//37
(/3

i3/7

7//3

(2)
(4)

a/3/7
(^2

/37/

+ 7/^a2) (a2

(3)

7) (7

a) (

/3).

^2) (^2

+ ^\

436

A COLLEGE ALGEBRA
TRANSFORMATIONS OF EQUATIONS

810

Some important transformations. It is sometimes advan = into another tageous to transform a given equation f{x) in some given relation to the roots equation whose roots stand The transformations most frequently used are of f(x) = 0. the following
:

811

roots are those

into another To transform a given equation f (x)= o/f (x)= with their signs changed.

ivhose

The required equation

number, as
ing
y8,

for x in

/8

for y in

/(

For substituting any is /( y) = 0. f(x) gives the same result as substitutHence, \if{x) vanishes'when x = p, y).
?/

f(-- y) will vanish when - ^ is a root of f(x) = 0,

/3;

that

is, if

/3

is

a root of

= 0. Therefore every root of f{x) = 0, with its sign changed, has no other roots and /( y) = is a root of /( y) = are of the same and /(- y) = than these, since f(x) =
/(y)
;

degree. If the given equation is

a^x"

a^x-"-^

+ a^x"-^

-\

= 0, = 0,
=

the required equation will be

o(- vT

(-

y)""'

^2(-

2/)"--

o^n

which on being simplified becomes

oy"

iy""'

(>2i/"-'

+ (-

!)"

0.

Hence the required equation may be obtained from the given one by changing the signs of the terms of odd degree when n is even, and hjj changing the signs of the terms of even degree, including the absolute term, when n is odd.

We may use x instead of y for the unknown letter in the for /( y) = 0. transformed equation, and write /( x) =

THEORY OF EQUATIONS
Example.

437

Change the
4x5

signs of the roots of

9x3

6x2

- 18x +

= 0.

Changing the signs of the terms of even degree, we have 4x5 - 9x3 - 6x2 - 13x - 6 = o.
In fact, the roots of the given equation are
of the transformed equation are
1

/2, 3/2,
2,

i-

2,

i,

and those

1/2,

3/2,

To transform a given equation

roofs are those ofi(x)

(x)

into another

whose
k.

812

each multijilied by some constant, as

if
/?,

The required equation is f(y / k) = 0. For when X = P:f{u/k) will vanish when y /k =
y
=:

/(a;) vanishes

that

is,

when

k(i

(compare

811).

If the given equation is

a^x"

+ aia;"-i + a^x^-^

-\

1-

= 0,

the required equation will be

which when cleared of fractions becomes

a,^f

kaiy--^

+ k'^a^rf-^ +

k^a,,

0.

Hence the required equation may be obtained by multiplying the second term, of the given equation by k, its third term by k^, and so on, taking account of missing terms if any. When A- != 1 this transformation reduces to that of 811.
Example.
divide

Multiply the roots of

2.
is

x-*

+ +

2x3
4 x3

=
48

by

2.

Also

them by
first
is

The

of the required equations

x*

8x

0,

and since
0.

dividing by 2
x*

same as multiplying by 1/2, the second is x3-x/8 + 3/16 = 0, or 16x* + 16x3-2x + 3 =

the

The following example

of the transformation

illustrates

an important application

813

now under

consideration.

Example. Transform the equation 36x3 another whose leading coefficient is 1 and
integers.

18x2

+ 2x +

into

its

remaining coefficients

Dividing by 36,

we have
x3

x2/2

+ x/18 +

l/4

0.

(1)

438

A COLLEGE ALGEBRA
k,

Multiplying the roots by

x3

A:xV2

fc2x/18

+fcV4 =

0.

(2)
all

"We see by inspection that the smallest value of k which will cancel the denominators is 6. And substituting 6 for k in (2), we have
x3

3 x2

2X

54

= 0,
by

(3)

which has the form required.

roots of the given equation
(1).

The

roots of (3) each divided

6 are the

814

To transform a given equation f(x)

roots are the reciprocals

of those of

f (x)

= =

into another whose

0.

The required equation \sf{\/y) = 0. For when X = /3, fi)-/>j) will vanish when 1/y =
If the given equation is
aoX"

if

f(x) vanishes

p, that is,

when

aix"-^

-\

+ a_iX + a =

0,

the required equation will be

o/y'

l/y"'

+
"

+ an-l/y +

= =

0,

which when cleared of fractions becomes

y"

-i//"

iy

f'o

0.

Hence the required equation may be obtained by merely

reversimj the order of the coefficients of the given equitation.
Replace the roots of 2 x*

Example.
rocals.

x^

3x

= +
2

by

their recip-

Reversing the coefficients,

we have 4x* Sx^

x^

0.

815

An
is,

equation like 2 x^
this

a;"^

a;

2
is

0,

which remains
to
it,

unchanged when

transformation
its

api^lied

that

when the

order of

coefficients is reversed, is called a

If ^ is a root of such an equation, reciprocal equation, 645. Hence when the degree of the l/)8 must also be a root.

equation is even, half of the roots are the reciprocals of the The same is true of all the roots but one when other half.
the degree
is

which

is

its

own

odd; but in this case there must be one root reciprocal, that is, one root which is either

THEORY OF EQUATIONS
1 or

439
and

- 1.

Thus, one root of2a;''

+ 3a;2-3x-2 = 0isl

the other roots are

2 and

1/2.
816

it follows that when and the leading coefficient vanishes, 07ie of the roots becomes infinite; when the tivo leading the roots become infinite; and so on. coefficients vanish, two of

From

the nature of this transformation

coefficients

an equation has variable

Example
becomes
Applying

1.

infinite

Show when

that one of the roots of mx^

3a;2

2x + l=0
(1) (2)

vanislies.

814 to

mx^
x*

we

obtain
If the roots of (2) are

+ 3 x'- - 2 x + 1 = - 2x2 + 3x + m =
^3,

0,
0.

/3i, /Sg,

those of
if

By
roots

806,
/3i,

/3i/32i33

m.

Hence,

are l/i3i, 1/^32, \ / ^z. as limit, one of the approach

(1)
/3i,

/32, /Ss

must

also
(1),

corresponding root of

approach namely, 1

as limit,
//3i,

and if this root be must approach 00, 512.

the

Example
become

2.

infinite

Show that two of when m vanishes.

mx^
x^

the roots of mx^

vi^-

=
(1) (2)

Applying

814 to

we have
Also,
^iftft
It follows

+ m'^x'^ + x + 1 = + x^ + jn^x + m =
^2, ft,

0, 0.

If the roots of (2) are

/3i,

those of

(1)

are l//3i, l/jSa, l/ft.

=and

iSiiSa

^ift

+
jSi,

ftft

= m\
.,

,.

(3)
/3i,

from

(3)
0,

that

if

m approach 0,
these roots be

two
00.

of the three roots

ft,

ft, ft

must

also

approach

if

the corresponding roots

of (1), namely, 1/ft, 1/ft,

must approach
f

To transform a given equation

roots are those of f (x)

(x)=

into another

whose

817

0,

each diminished by some constant,

as k.

The required equation is/(.?/ + A;) = 0. For if /(x) vanishes when X = fi, f{y + k) will vanish when g + k = (3, that is, when ?/ = y8 k.
If the given equation is

f(x)

a^pc"

aix"-^ H

[-

a_-iX

= 0.

the required equation will be

f(g

k)

a,(g

+ ky +

a^{y

k)"-'

+ a =

0,

440

A COLLEGE ALGEBRA

which, when its terms are expanded by the binomial theorem and then collected, will reduce to the form
</>(y)

=
Ci

^oy"

^1^""'

c,-^y

= o,

where

= Cq

(,,

?ikci^

i)

and

so on.

laborious.

from f(x) is usually very much more expeditious, at least when the coefficients oif(x) are given rational numbers. li X = y -\- k, then y =z x k, and we have
This method of obtaining
(j>(}/)

The following method

is

that
Co(x

is,

k)"

-\

h c_i (x

k)

aoX H

\-

a_iX

If both

members

of this identity be divided

by

a;

+ a. k, if

k, and again the quotients thus obtained be divided by x so on, the successive remainders yielded by the first member, namely, c, c_i, , will be the same as those yielded by the

second member.
follows: Divide
f

Hence we may obtain ^

(x) ^y x

k,

(y)

from f(x) as

by X

k,

Cn-u

Ci,

and so on. The and the final quotient

Diminish the roots of 2
Substituting y

divide the quotient thus obtained successive remainders will be c,

will be

(compare

423).

The

divisions should be performed synthetically.

a;''

Example L
First method.

7 x^

3x +
3 (y

=
+
4)

by

4.

2 x3

7 x2

+ 4 for x, we have - 3x + 1 = 2 (y + 4)3 - 7 (y + 4)2 = 2y^ + \ly'^ + ?,-y + 5.

Arranging the reckoning as

Second method.

in 423,

we have

1[4
4
5
.-.

Ca

= =
= =

5.

.-.

C2

37.

2,

17

.-.

ci

17

and

Cq

2,

Hence, as before, we find the required equation to be

2y''

17y2

+ 37y +

o.

THEORY OF EQUATIONS
Example
2.

441
3 =:

Increase the roots of x^

is

4 x^

by

4.

To
- 4.

increase the roots by 4

the

same thing

as to diminish

them by

Hence

ing 2/ be 2/3

required equation may be obtained either by substitut4 for X or by dividing synthetically by 4. It will be found to
tlie

2/2

17

?/

0.

By aid of 817 we can transform a given equation into another which lacks some particular power of the unknown
letter.

818

Example 1. Transform the equation x^ 3x2 + 5x + 6 = another which lacks the second power of the unknown letter.
Substituting x

into

must have Sk
x3

= y + k, we have y'^ + {Sk 3)y'^ + -. Hence we 3 = 0, that is, k 1. And diminishing the roots of - 3x2 + 5x + 6 = by 1, we obtain x^ + 2x + 9 = 0. 5x- + 8x 1 = into Example 2. Transform the equation
x*^

another which lacks the

Substituting x

first

power

of

tlie

unknown

letter.

+ k, vfe have - 5) 2/2 + (3 - 10 + 8) + = 0. 2/3 + (3^ - 10 + 8 = 0, that is, = 2 or 4/3. Hence we must have 3 5 x2 + 8 x 1 = by 2, we obtain Diminishing the roots of x^ x3 + x2 + 3 = 0.
y
A:2
A;

2/

fc2

A;

fc

= by k, we obtain If, when we diminish the roots of f(x) an equation <^(a:)=0 whose terms are all positive, k is a = 0, 804. For in superior limit of the positive roots oi f(x) Hence any this case <^(.r)=0 has no positive root, 794. may have become negative when positive roots that f(x) =
diminished by
k.

819

They

are therefore less than k.

process of synthetic division is such that it is possil)le by inspection and trial to find the smallest integer k for which I'n most cases this will be positive. all the terms of <^(a-)=

The

can be accomplished with comparatively little labor. We may obtain an inferior limit of the negative roots of

f(x) = For if

by finding a superior limit of the roots of /( x)

a superior limit of the roots of /( a;)= an inferior limit of the roots of f(x) = 0, 811.
A;

0.

is

0,

then

A;

is

442
Example.
tion /(x)

A COLLEGE ALGEBK.^
Find superior and inferior

X*

6 x3

14 x2

48 X
trial

all

limits of the roots of the equa-

121

0.
A:

We

find

by inspection and
(x)
if

that neither of

nor

A;

2 will give

a transformed equation In fact, that k = 3 will.

obtain
<^

=
+

whose terms are

the roots of /(x) 68 = 0. Hence 3

positive,

but

we diminish
78 x

=
is

by

3,

we

(x)

x*

6 x^

14 x^
0.

a superior

limit of the roots of /(x)

The equation /(find

x)

= =

is

x*

6x^

+ Ux^ - 48x -

121

0.

We
roots.

by inspection and trial that 3 is a superior limit of its positive Hence 3 is an inferior limit of the negative roots of /(x) = 0.

820

On

rational transformations in general.

If

we eliminate x
x,

between

the

equations f(^x)

=
in

and

wq

obtain

y( ?/)=0. We have shown /( y)= are connected with

relation y rem that if

= x.
we

that the roots y of the roots x oif(x) by the

811

an illustration of the general theoand any properly eliminate x between /(>') =

is
(fi

This

equation of the form y = (x), where <^ (x) is rational, we shall obtain an equation F(y)= whose roots are connected with those oif(x) = by the relation y 4^{x), so that if the
roots oif(x)
<I>(J3.^,

are ^i,

</)(^).

those of F(y) = are cf> (/?i), The transformations of 812, 814, 817
p.,,
/3,

afford further illustrations of this theorem.

In the
is

first

of

these transformations the equation y z= (^(x) second it is ?/ = 1 /a-, and in the third it is 3/
as in these cases, // <^ of X is readily effected.

= kx, in tlie = k. When,

y
a;

(a-)

can be solved for

x,

the elimination

Example
roots of X*

1.

Find the equation whose roots are the squares of the

qx

+ px2 +

0.

In this case the relation y

(p{x) is

= x^.

y/y for

And substituting Solving y = x'^ for x, we liave x = VyX in the given equation and rationalizing, we obtain
y^

+ (2q- p2) y2 + +

(g2

_2pr)y -r^ =

0.

Example 2. equation whose

If the roots of x^

px^

+ gx +

are a,

j3,

7, find the

roots are ^7, ya,

a/3.

We
a/3 in

first endeavor to express each of the proposed roots /37, ya, terms of a single one of the given roots a, /3, 7 and the given

THEORY OF EQUATIONS
coefficients p, q,
r.

443

This

is

readily done, for, since

^y

= a^y/ a =

r/ a,

ycx

= a^y / =
jS

r/ ^,

r = a/37, we have a^ = a^y /y = r /y.

Hence each root y of the required equation is connected with the corre r /x. sponding root X of the given equation by the relation y r/x for x, we have x = r/y. Solving y =

And
fying,

substituting y^

we have

qy'^

r/y +

for x in x*

+ px^ +
which
is

qx

i:>r?/

7-2

0,

+ r = and simplithe equation required.

EXERCISE LXXII
1.

Change

the signs of the roots of x^

3 x*

2 x^

2.

Multiply the roots of 2 x*

+
10

x^

4 x2

6x

+ =

6x

by

+7= - 2.

0.

Also

divide
3. 4.

them by
In 5 x^

3.

X* +

3 x^

9X

+
x^

=
x*

replace each root by

its

reciprocal.

Diminish the roots of 2

1.

3 x^

by

2.

Also increase

them by
5.

Transform the equation x* into another whose coefficients are

Transform the equation 3 lacks the term involving x*.
6.

x^/S

x2/4

+ x/25 -

1/48

=
1.

integers, the leading one being

x*

36 x'

+x

into another

which

7.

Transform the following into equations which lack the x term.

(1)

x3

0x2

+ 9x + +
/32,

10

=
x

0.

(2)

x3

x2

0.

8.

If the roots of x*

x^

=
r

are a,

^3,

7, 5, find the

equation

whose roots are

9.

a2,

72, 52.

If the roots of x*

+
/3

3 x^

equation whose roots are

10.
If the roots of x^

+ +

2 x2
5,

are a,
5,

/3,

y,

5,
tt

find the
/3

a+ 7+

a +

^3

5,

7.

px-

qx

are a,

/3,

7, find the

equa-

tions

whose roots are

^7
(1)

a^
1

/37
1

ya
(<i)
i3

a
1

^
1

11.

If the roots of x^

+7 + 2x2 + 3x + 4 =
''/3
i32.

7+aa+i3
7),

are a, ^, y^ find the equa-

tions
(1)

whose roots are

^2

+
+

y2^

(3)

By

-, ytx+

111
yi

0-2,

a:2

(2)

ex

(^

+
a

^ (7
,

),

7 (
,

^)-

a-/3

-.

(4)

444
12.

A COLLEGE ALGEBRA
Find superior and
inferior limits of the real roots of the following

equations.
(1) x* (3) X*

+ 3x3 -13x2 -6a; + 28 = 0. (2) 2 x^ - 120x2 - 38x + - 26 x^ + 60x2 _ - 29x2 + 50x + 12 = 0. (4) 2x5 - 14 x3 + 44 x2 + 28 X - 92 = 0. (5) ji 77x2-50x-110=.0. (6) 3x-35x3 +

27

92

= =

0. 0.

IMAGINARY ROOTS.
821 Theorem.
cients.
-//"

DESCARTES'S RULE OF SIGNS

If a

it

ib is
if

= denote an equation with real coeffif (x) has imaginary roots, these occur in pairs; that is, a roof, a ib is also a root.
Let
a root of fix)
;

= 0, then f(x) is divisible by and we shall prove that a ib is a root (a ib), or, if we can show that f{x) is also divisible by a; what comes to the same thing, if we can show that f(x) is
For
4- ih is

X (^a

ib),

795

divisible

by the product \_x (a + ib)][x (a This product has real coefficients, for, since i^

ib)^. = 1,

lx-(a +

ib)^ [.r

(a

ib)^

Since the polynomials f(x)

the

common

factor x

(a +

= {x - )' + b^ = x--2ax+ (a' + b^). and x^ - 2 ax + (a^ +

b'^)

have

ib),

factor.

cannot be x (a + ib), since whereas the highest common factor of two polynomials with real coefficients must itself have real coefficients, 469. Hence the highest common factor - 2 ax + (a^ + b"") in oif{x) and x^-2ax+ (a"- + b") is x^
ov
x"^

This highest

common
b'^).

they have a highest common factor must be either x (a + ib)

it

2ax + (2 +
imaginary

But

this has

coefficients,

other words, f(x) is divisible hy to be demonstrated.

Example.
this equation.

x"^

2 ax

+ (a^ +
is

b-),

as

was
Solve

One

root of 2

x''

T)

x^

46 x

87

i.

Since
the roots

is

+ 5 i is a root, 2 5 iis also a root. But 5/2, 806. Hence the third root is _5/2-(-2 + 5i-2 - 5i) = 3/2.

the

sum

of all

THEORY OF EQUATIONS

445
822

Corollary 1. Eoery polynomial f (x) xi'itli real coefficients is the product of real factors of the first or second degree.

For

to

each real root

c of

real factor

f{p:)
;

there corresponds the

to each pair of imaginary

roots a
factor

a;2

+ ib, a ib of f(x) = there corresponds _ 2 ax + (a^ + oif(x), 821.

b'-)

oif(x),

795

and

the real

Corollary 2.

The product of those factors o/f (x) which

imaginary
roots of

corre-

823

spond which

to the

f(x)=
x.

is

a function of x

is positive
it

for all real values of

For

(x ay b', squares, is positive for all real values of x.

Corollary 3.
is

may

be expressed as a product of factors of the form 821, and every such factor, being a sum of

Every equation

icith real coefficients ichose

degree

824

odd has at

least one real root.

even,

For the number of its imaginary roots, if it have any, is 821, and the total number of its roots, real and imagiHence at least one root must be real. nary, is odd, 798.
them
Thus, the roots of a cubic equation with real coefficients are either real, or one real and two imaginary.
all

of

By
if

the reasoning employed in

821

it

may

be proved that

825

+ "Vo is a root of a given Vi is also a root; it cients, a

a

equation with rational coeffibeing understood that a and b

are themselves rational, but Vi^ irrational. be an equation whose Irreducible equations. Let <^ (x) coefficients are both rational and real. say that this

826

We

equation is irreducible if ^{x) has no factor whose coefficients are both rational and real (compare 486).
Thus, x2
i-

and

x-

are irreducible equations, but

is

not irreducible.

Theorem.

Let

are both rational

equation of the

be any equation and real, and let ^(x)= same or a lower degree.

f(x)=0

ichose coefficients
be

827

an

irreducible

446

A COLLEGE ALGEBRA

be a root of f (x) If one of the roots of ^ (x) are roots of f (x) = 0. of the roots of <f>(x)=

0,

then

all

This

f(x)
4> (a-)

be proved by the reasoning of 821. For if = have the root c in common, f(x) and <^ (x) have the common factor x c, 795, and therefore a

may

and

common factor which is either x c, some factor of which contains x c, ov (x) itself. But since, by hypothesis, (f>(x)=z is an irreducible equation, (fi(x) is the only one of these factors Avhich has real and
highest
(f>

(x)

(jt

rational coefficients such as the highest and (f> (x) must have, 469.

common

factor oif(x)

and

Therefore <^(x) is itself the highest common factor of f(x) <li(x); in other words, /(a-) is exactly divisible by
^(ir).

Hence f(x) may be expressed in the form f(x) = Q<j> (x), where Q is integral, and from this identity it follows that
f(x) vanishes whenever <f)(x) vanishes; in other words, that, is a root of every root of <f,(x) = f(x) = 0. Permanences and variations. In any polynomial f(x), or equation/(a-) = 0, with real coefficients a. 2ier7?ianence or continuation
of sign
sign,
is said to occur wherever a term follows one of like and a variation or change of sign wherever a term fol-

828

lows one of contrary sign.

Thus, in x^- x* - x^ + 2 x"^ + ^x -\ = permanences occur at the terms - x^ and 3x, and variations at the terms - x*, 2x2, and - 1.

829

f (x) is exactly divisible bt/ x h, where b is the coefficients ofi(x) are real, the quotient (f>(x) will have at least one less variation than f (x) has.

Theorem.

If

positive

and

from the rule of synthetic divided by x - b, the coeffiquotient are positive until the first negative coefficient of f(x) is reached. If then or later one of them
division, 411, that cients of the

For since

b is positive, it follows

when /(a;)

is

becomes negative or zero, they continue negative until the next positive coefficient of f(x) is reached, and so on. Hence <^ (x) can have no variations except such as occur at the same

THEORY OF EQUATIONS
or earlier

447
is,

terms of /(x).

But

since the division

by hypoth-

the last sign in sign in /(if), and therefore

esis, exact,

<^ (.r)
<f>

must be contrary to the last (x) must lack the last variation

in /(a-).
1

+1
2

-2
+_6

1+3
1

+4

quotient being first two variations

-10 -1 +12 -4[2 Thus, f(x) = x6 + x^ - 2 x* -10x3 -x2 + 12x - 4 isex+ 8 - 4 - 10 + 4 2, the 2 5 +2, actly divisible by x 2 x^ 5 x + 2. Observe that the (x) = x^ + 3 x* + 4 x'
of/
(x)

are reproduced in 4>{x), but not the third.

-1
2

T+1+3
disappear in

+1 +2 +

-7
G

+2\2 2

Again, /(x)
exactly

x*

divisible

by x

1, being 0(x) = x'' + xonly one of the four variations of /(x) is reproduced in 0(x), and we have an illustration of the fact that wlien inter)nedlute variations of /(x)

+ x2 - 7 x + 2 is 2, the quotient + 3 x 1. In this case x^

<p{x),

they disappear in pairs.

rule of signs).

Theorem (Descartes's

A71 equation f (x)

cannot

830

have a greater number of positive roots than it has variations, nor a greater number of negative roots than the equation f ( x) = has variations.

A]

For

let

if

we

hy X
since

divide f(x) by x fS^, the quotient thus obtained and so on, we obtain a final quotient ^(.r) which (3.,,
r less variations

jSi, /So,

/8,.

denote the positive roots of f{x)

0.

has at least

^ (x) cannot have

roots.'!

less

than f(x) has, 829. Therefore, than no variations, f{x) must have
at least as

at least r variations, that

is,

many

as f(x)

has

positive
2.

of/(.r)=0 become the positive roots And, as just demonstrated, /( x) = cannot have more positive roots than variations. Hence = cannot have more negative roots than /( x) = f(^x)

The negative
x)

roots

of

/(

0,

811.

has variations.

x^ x^ + x 1 = cannot have more Thus, the equation /(x) = x^ than three positive roots nor more than one negative root. For f(x) has three variations, and /( a') = 0, that is, x^ + x^ + x' x 1 = 0, has
one variation.

448
831

A COLLEGE ALGEBRA A

Corollary. cotnplete equation cannot have a greater number of negative roots than it has jjernianences.

For when f{x) = is complete its permanences correspond one for one to the variations of /( x) = 0, since of every two consecutive like signs in /(-x) = one is changed in /( x) 0,

81L
then

+ x*-6x3- 8x2-7x + l =0, (1) - X* - 6 x^ + 8 x2 - 7 X - 1 = 0. (2) In (1) we have permanences at the terms x*, 8_x2, 7x, and at the 7 x,; we have variations. 8 corresponding terms of (2), namely, cannot Since (1) has tivo variations and three permanences, /(x)
Thus,
if

/(x)
X)

/(-

= =

is

x5
x5

is

x**,

x'-,

have more than two positive roots nor more than three negative

roots.

832

Detection of imaginary roots.

In the case of an incomplete

equation
roots

we can frequently prove the existence of imaginary by aid of Descartes's rule of signs.

he an equation of the nth degree ^uhich has no Let f (x) zero roots, and tet v and v' denote the number of variations in

= and must have at

f (x)

x)

least

= respectively. The equatio7i (v + v') imaginary roots.

f (x)

For /(x) more than

than V

=
v'

negative roots,

cannot have more than v positive roots nor 830, and therefore not more
roots all told.

v' real

The

rest of its

n roots must

therefore be imaginary.

This theorem gives no information as to the imaginary roots of a complete equation, since v f' is equal to n in such an

equation.
Example.

Show

that x^
isx'^

In this case/(x)

=
')

Hence n

(u

+ x^ + 1 = has four imaginary roots. + x^ + 1 = 0, and/( x) = is x^^ x^ 1 = 0. (0 + 1) = 4, so that there cannot be less than

four imaginary roots. But since there are five roots all told and one of them is real, the degree of x^ + x^ + 1 = being odd, 824, there cannot be more than four imaginary roots. Hence x^ + x- + 1 = has exactly

four imaginary roots.

THEORY OF EQUATIONS
EXERCISE LXXIII
1.

449

One One

root of 2 x*

x^

5 x-

13 x

= +
2

is 1

is

t.

Solve this

equation.
2.

root of 2 x*

11 x^

17

x-^

10 X

V2.

Solve

this equation.
3. Find the equation of lowest degree with rational whose roots are 5 + 2 i and 1 + v 5.

coefficients

twc

of

4.
5.

Find

tlie

irreducible equation one of

whose roots

is

v'2

i.

conclusions regarding the roots of the following equations can be drawn by aid of Descartes's rule and g 8o2?
(1)

What
x^

(3) x^ (5) x"

(7) x5

+ + + -

0.

(2)

x*

2 x^
x5

+ +

x2

+
1

=
0.

0.

(4) X* (6) (8)

x^

X--

-X+ = 0.

1 = 0. + x2 - X + 1 = 0. x" + X* - x2 - 1 = 0. x"" - x2 + X" + X + 1 = 0.

x2

2 x'

as

6. Show that many positive

a complete equation roots as variations,

all of whose roots are real has and as many negative roots as

permanenc:es.
7.

Given that

all

the roots of x^

3 x*

+
1

5 x^

35 x2

54 x

72

are real, state

8.

how many

are positive and

how many

negative.

Prove by Descartes's rule that X-"

has no real root.

What

conclusions can be
X2'+1
9.

drawn by

aid

of this rule regarding the roots of

+ 1=0? X2"-1=:0?

X2'' + 1-1

0?

Prove that an equation which involves only even powers of x with positive coefficients cannot have a positive or a negative root.
10.

Prove that an equation which involves only odd powers of x with

0.

positive coefficients has no real root except

11.

that the equation z^ + px + q = 0, where tive, has but one real root, that root being negative.

Show

p and

q are posi-

12. Show that an incomplete equation which has no zero roots must have two or more imaginary roots except when, as in x^ 3 x"^ + 1 = 0, the missing terms occur singly and between terms which have contrary signs.
13. Show that in any equation f{x) = with real coefficients there must be an odd number of variations between two non-consecutive contrary signs, and an even number of variations, or none, between two

non-consecutive like signs.

450

A COLLEGE ALGEBRA

14. Prove that in the product of the factors corresponding to the negative and imaginary roots of an equation with real coefficients the final term is always positive, and then show that if this product has

any variations
15.

their

number

is

even.

Prove that when the number of variations exceeds the number of positive roots, the excess is an even number.
16.

Show

that x*

x^

x-

=0

has either one or three posi-

tive roots

and one negative

root.
is

17. Show that every equation of even degree who.se absolute term negative has at least one positive and one negative root.

LOCATION OF IRRATIONAL ROOTS

833
f

Theorem
(^xj

1.

7/'f(a)

and f(b) have contranj

b.

signs,

a root of

lies heticeen

a and

This may be proved as in the follov^ing example. statement of the proof will be given subsequently.
Example. and 2.
Prove that f{x)

A general
1

x^

3x +

has a root between

The

sign of /(I)

is

minus and that

of /(2)

is

plus.

By computing we

the values of /(x) for x

1.1, 1.2, 1.3,

2,

successively,

find two cmsecutive tenths

between

and

which /(x) has the .same signs as for x = 1 /(1. 5) =- .125 is miuu.s, and /(l.O) = .290 is plus. By the same method we find tiDO consecutive hundredths between 1.5 and 1.6, namely, 1.53 and 1.54, for which /(x) has the same signs as for x = 1 and X = 2 for /(1. 5:5) = - .008423 is minus and/(1.54) = .0,32204 is plus. This proce.ss may be continued indefinitely. It determines the two never-ending sequences of numbers:
;
;

namely, 1.5 and 1.6, for and x = 2 resiDectively for

(a)

1, 1.5,

1.53, 1.532,

(b)

2, 1.6, 1.54,

1.533,

-,

the terms of which approach the .same limiting value, 192, 193. this limiting value c. It is a root of /(.r) = 0, that is, /(c) = 0.
if

Call

run througli either of the sequences of For, by 509, values (a) or (b), /(x) will approach f(c) as limit. But since/(x) is always negative as x runs tiirougli the sequence (a), its limit /(c) cannot be positive and since /(.c) is always positive as x runs through the .sequence (b), its limit /(c) cannot be negative. IIence/(c) is zero.
x be
to
;

made

THEORY OF EQUATIONS

451
834

Theorem 2. If neither a nor \) is a root of f (x)= 0, and an lie between a and b, f odd number of the roots of f(x) = (a) and f(b) haae contrary signs; hut if no root or an even number of
roots lie between

a and

b, f (a) ujid f (b)

Aawe the same sign.

Conversely, if f(a) and f (b) liave contrary signs, an odd but (/" f (a) number of the roots of f (x) lie between a and b

and

(b) liare the

lie

same

sign, either
b.

no root or an even number

of roots

between a a7id

< b and that of the roots of /(a-) = between ^2) divisible by (x ^i)
Suppose that a

ySi, yS,,

/3,.

is

a complete
if

list

and

b.
/3,.),

(a;

(x

Then/(a;) ^18, and

is exactly-

we

call

the quotient

<^

(x),

we have
/30 (x

/(,r)

= (x-

/?,)

...(X-

/3,.)

<t>

(x).

(1)

Substituting first a and then b for x in (1) and dividing the first result by the second, we obtain

f(a)
f{b)

^ a- pi

(3.2

(f>

(3,

<f>(a)

b-(3i'b-/3,"'b~(3,.'
<}>

^^^ ct>(by

For In the product (2) the factor (a) / (b) is positive. and <^ (b) have the same sign, since otherwise, by 833, = and therebetween a and b there would be a root of (x)
<^

(a)

<f)

fore,

by

(1),

a root of f(x)

in addition to the roots

(3i,

(3---,/3,..

On
and

the other hand, each of the r factors (a (3i)/(b so on, is negative, since each of the r roots ^1, /So, ,

/3i),

/S^

is

greater than a and less than

b.

Therefore, when r is odd, f (a) /f(b) is negative, that is, f(a) and f(b) have contrary signs but when r is even or zero, f{fi)/f(b) is positive, that is, /() and f(l)) have the
;

same

sign.

Conversely,

when /() and fQ)) have contrary

it

signs, so that
;

f(a) /f(b)

follows from (2) that r is odd and when f (a) and f(b) have the same sign, it follows that r is even or zero.
is

negative,

452
835

A COLLEGE ALGEBRA

Observe that in the proofs of the preceding theorems, 833j, 834, no use has been made of the assumption that every equation /(a-) has a root. Notice also that in applying these

theorems a multiple root of order r is to be counted as r simple roots. From 834 it follows that as x varies from a to b, f(x) will change its sign ias x passes through each simple root or multiple root of odd order of /()= which lies between a and b, and that/(x) will experience no other changes of sign than these.
Thus,
to
5,
if

/(x)

(X

2) (x

3)2 (x

4)3,

and x be made

the sign of /(x) will be plus between x = 1 and x 2 and x 4 and x = 5. 4, and plus between x

2,

from 1 minus between

to vary

836

Location of irrational roots.

it is

By

aid of the theorem of

833

usually possible to determine between what pair of consecutive integers each of the fractional and irrational roots of

a given numerical equation

Example.

lies.

Locate the roots of /(x)

x*

6 x^

x2

12 x

= 0.

Descartes's rule of signs, 830, this equation cannot have more than three positive roots nor more than one negative root.

By

To locate the positive roots we compute successively /(O), /(I), /(2), until three roots are accounted for by 833 or until we reach a value of

X which
/(O)

is

a superior limit of the roots, 803.

/(I)

Thus, using the method of synthetic division, as in

/(6)

= - 6, = 102.

2,

/(2)

=-

414,

we

find

10, /(.3)

=-

42, /(4)

=-

70, /(u)

= - 46,
2,

Hence,

8.33,

one root

lies

and

the third between 5

and

6.

between and 1, another between 1 and There cannot be more than one root

in

any of these intervals, since there are only three positive roots all told. Making a similar search for the negative root, we have /(O) = - 6, /(-l)=-10>/(-2) = 38. Hence the negative root lies between -land -2.

The mere substitution of integers for x in f(x) will of course not lead to the detection of all the real roots when two or more
of

them

lie

will be considered in
is

between a pair of consecutive integers. This case 844 and again in 8G4, where a method

given for determining exactly

how many

roots lie between

any given pair of numbers.

THEORY OF EQUATIONS
EXERCISE LXXIV
Locate
1. 3.

458

the real roots of

each of the foUovviiig equations.

2 x3
x3

3 x2

9X

+
5

0.

2.

x^

3 x^

2X

0. 0.

4.
6.

2 x3
x3

- ix -2 = 0. + 3 x2 - 10 X - 15 =
x-

0.

5.

x3- 4x2-4 x +
x3

12

7.
9.

5 X

19

0.

8.

x*

+ 13x2 + 54x + 71 =0. - 95 = 0. +

Sx^

x*-8x3 + 14x2 + 4x -8 =
X*

0.

10. x*

x2

- 13x - 7 = 0.

11.

-11x3 + 32x2-

4x- 46 = 0.
+
is

12.
13.
is

x5

+
its

2 X*

16 x3
tliat

24 x2

48 X

32

0.

Assuming
term

when x

very large numerically the sign of /(x)

that of
(1)

of highest degree,

show

that
6

in

+ Every equation x" + 6iX"-i + which ?i is even and b,i is negative, has

with real

coefficients,

at least

one positive and one

negative root. (2) The four roots of the equation

^2 (x

6) (x

c)

+
+

^2

(x

Hi- (x
a,

- c) (x - a) a) (x 6) b,

X (x

a) (x

6) (x

c)

=
being

lie

between oo and assumed that a, 6, c,

a and
I,

k,

and c, c and co respectively, are real and that a<b<c.

b

it

14. Show that every equation of the form x^ + (x 1) {ax 1) = 0, where a>3, has two roots between and 1, namely, one between \/a and 1 1/a and one between 1 1/a and 1. 15.

Show

that x^

(x

1) (2
1

1)

(ax
1

between

and 1/a, 1/a and

2/a,

1) = 0, where a> 5, has roots 2/a and 1.

COMPUTATION OF IRRATIONAL ROOTS

Horner's method.
Positive roots.

There are several methods

irrational roots of

837

by which approximate values of the

ical

numer-

The most expeditious of equations can be computed. these methods is due, in its perfected form, to an English mathematician named Horner. It may best be explained in
connection with an example.

454
Example.
1.

A COLLEGE ALGEBRA
Find the positive root of /(x)

x-^

x^

15 x

59

0.

By

the

3 and 4. form 3. ^yd

method of 836, we find that the required root lies between Hence if it be expressed as a decimal number, it will have the

y, 5,

2.

denote its decimal figures. Diminish the roots of /(x) =

(1)

by

3.

We

obtain the transformed equation

0(x)

2x3

19x^

+ 45X-41 =

0,

^2)

which has the root .^yd and 1.

Testing x
1^

lying between

.1,

.2,

.3,

that 0(.6) is and ^(.7) is +. Hence the root of (2) lies between .6 and .7, that is, ^ is 6, and the root of (1) to the first decimal figure is 3.6.

in <p{x),

we

find

2+19
V2
20.2
1-2

+45
12.12

41 1^ 34.272

3.

Diminish the roots of

obtain

(2)

by

.6.

We
f

57.12
12.84

6.728

(x)

2x5

+ 22.6x5 +69.96X-

6.728

0,

(3)

THEORY OF EQUATIONS
In the example we found the
transformed equation, that
first figure

455
839

of the root of each

successive decimal figures of the root of the given ec[uation, by the method of substitution. But the first figure of the root of each transformed equation
is, tlie

from the second on may ordinarily be found by merely dividing the absolute term of the equation, with its sign changed, by the This is called the method of trial divisor. coefficient of x.
Thus, consider the second transformed equation in the example
^p{x)

2x^

22.6a;-

69.90x

6.728

0.

(3)

This equation is known to have a root c wliich is less than .1. The second and higher powers of such a number c will be much smaller than
c itself.

Thus, even

(.00)'- is

stituted in (3), the first

but .0081. Hence, were c known and subtwo terms of the resulting numerical identity
22.6 c2
in

2 c3

+ 69.96 c -

6.728

=
last

would be very small numbers

Therefore
equation
c is

comparison with the

its first

two.

not likely to differ in

69.96 x

figure

from the root of the

(3')

6.728

=
is,

obtained by discarding the x^ and x^ terms in (3). But solving (3'), we have x = 6.728/69.96 = .09 +, that above, that the first figure of the root of (3) is 9.

we

find, as

This method cannot be trusted to give the first figure of the root of the fJrst transformed equation correctly. But it will usually give at least some indication as to what that figure is

and

so lessen the

number

of tests that

need

to be

made

in

applying the method of substitutions. Occasionally the method fails to give correctly the first figure of the root of even the
second transformed equation. But in such a case the error is readily detected in carrying out the next transformation for if the figure is too large, a change of sign will occur in the
;

absolute term of this next transformed equation, 838 if it \s too small, the first figure of the root of this equation will be of too high a denomination.
;

We may

avoid the troublesome decimals which occur in the

first

840

transformations after the

by multiplying the roots of each

transformed equation by ten,

812, before

making the next

456
transformation.

A COLLEGE ALGEBRA
This may be done by affixing one zero to the

second coefficient of the equation in question, two zeros to its third coefficient, and so on. We then treat the iigure of the root

employed

in the next transformation as if

first

it

were an

integer.

Thus, the

transformed equation in the example in 2x3 + 19x2 + 45x- 41 = 0,

it

837 was
(2)

and we found that

had a root of the form

.6

+.

Multiplying the roots of (2) by 10, we obtain 2 x3 + 190 x2 + 4500 X - 41000

0,

(20

which has a root of the form 6

Diminishing the roots of (2') by 6, the reckoning differing from that above given only in the absence of decimal points, we have
2 x3

226 x2

G990 x

6728

0,

(SO

whose roots are ten times as great as those of

2 x3

22.6 x2

69.96 x

6.728

= 0.
(3')

(3)

The method
as above, .09

of trial divisor gives .9 as the root of (3).

as the root of

and therefore,

841

We may now arrange the reckoning involved in computing the root of 2 x^ + x^ lox 59 = to the third decimal figure as
follows
:

59 [3.693 18

-41

THEORY OF EQUATIONS
tenth, thus, .9, .3.

457

Observe that here each figure obtained by the trial divisor method is a Had the last coefficient of the second transformed equa-

tion been

672 instead of

for the next two figures. been 3.609 instead of 3.69,

The and

6728, we should have had 672/6996 = .09 root as far as computed would then have

before performing the next transformation

we should have
by
100. that
is,

multiplied the roots of this second transformed equation we should have affixed two zeros to its second coefficient,

four

to the third,

and

six to the fourth.

This process

may

be continued indefinitely.

But we soon

842

encounter very large numbers, and after a few decimal figures of the root have been obtained we can find as many more as are likely to be required, with much less reckoning, by the
following contracted method.
reckoning above given

The

last

transformed equation
2
a;'5

in the

is

2314 x^

740706 x

247082

0.

(4)

Instead of affixing zeros to the coefficients in order to multiply the roots of (4) by 10, we may substitute x/10 for x in (4), 812, thus obtaining
.002 x^

23. 14

X'i

74076.6 x

247082

0.

(4')

Ignoring the decimal parts thus cut off from the coefficients as being too small to affect the next few figures of the root, but adding 1 to the corresponding integral part when the decimal part is .5 or greater, we

have the quadratic

23 x^

74077 x

247082

0.

(4")
:

We may

then continue the reckoning as follows 23

74077

247082

|.

003332

69

222438

458

A COLLEGE ALGEBRA
off figures as

and thus reduce (5) to the simple equation 7422 x 24644 = 0. The next two figures of the root, namely, 3, 2, are then obtained by merely dividing 24644 by 7422 by a contracted process which consists in cutting off figures at the end of the divisor instead of affixing zeros at the end of
the dividend.

Before performing the next transformation we cut

before

843

Negative roots.

To

find a negative irrational root of f{x)

0,

find the corresponding positive root of

/( x)=0 and
x^

then

change

its sign.

Example.

Here/(
of /(x)

x)

Find the negative root of /(x) = = is x^ X'^ lOx 9 = 0.

x^

lOx

0.

Its positive root,

found by

Horner's method, is 4.03293 approximately. = is 4.03293 approximately.

Hence the negative root

844

If the given equation has two roots Roots nearly equal. lying between a pair of consecutive integers, they may be found as in the following example.

Example.

Find the positive

roots,
1,

if

any, of /(x)

= x3 + x2 10x+9 = 0.

/(2) = 1, /(3) = 15, and the reckoning shows that 3 is a superior limit of the roots, 803. Hence, 834, either there is no positive root, or there are two such roots both lying between

We

find that/(0)

9,

/(I)

and

1,

or between 1 and 2, or between 2 and

3.

But /(I) and /(2)

differ less

from
3.

than /(O) and /(3) do.

1

may

expect to find them between

Hence, if two roots exist, we and 2 rather than between and 1 or

between 2 and

We

therefore diminish the roots of /(x)

(x)

= +
1

by

1,

obtaining

x3

4 x2

5X
if

which has two roots between 1 and 2.

and
for x
is

f(x)

= =

0,

has two roots between

Computing the values

of

+ and

(/)(.3) is

(x)

also that ^(.7)

.2

= .1, .2, .3, and <^(.8)

we

find that

<f)

(.2) is

is

and another between .7 .8. By Horner's method we find that these roots are .25560 and .77733 approximately. Hence /(x) = has the two positive roots 1.2556 and 1.77733.
.3

has a root between

and

+. and

Hence ^{x)

845

On

locating large roots.

is

In case the given equation f{x)

has a root whi(;h

ing method

greater than ten, we may employ the followfor finding the figures of its integral part.

THEORY OF EQUATIONS
To
obtain the

459

compute the values of f(x) for = 100, 200, , and so Thus, if we found that/(400) and /(500) on, applying 833. had contrary signs, so that the root lay between 400 and 500, To find the remaining figures, the first figure would be 4.
first figure,

10, 20,

or, if

necessary, for x

successive transformations of the equation, as when Thus, in the case just cited we finding the decimal figures. should diminish the roots of /(a-) by 400 and so obtain an and 100. If we having a root between equation <^(a-)=

make

found that

this root lay

7.

between 70 and

80, the

second figure

of the root v/ould be

<fi(x)

We

should then diminish the roots of

=
9,

root between

having a by 70 and so obtain an equation i}/(x) = and 10. If we found that this root lay between
should have shown that the integral part of the

8 and

we

root of f{x)

was 478.
If asked to find uil the real

On

solving numerical equations.

846

roots of a given numerical equation f(x) 0, it is best, at least when the coefficients are rational numbers, to search first
for rational

roots

by the method of

802.

This process

will yield a depressed equation <^(x)= whose real roots, if locate these roots by the method of any, are irrational.

We

833, 844, 845,

and then

find their

approximate values by

Horner's method.
It may be added that a fractional root may also be found by Horner's method, exactly when the denominator involves only the factors 2 and 5, approximately in other cases.

EXERCISE LXXV
Compute the
1.

roots indicated below to the sixth decimal figure.

3

x^ 4-

X
2

=
20

root between 1 and

;

2.
3.

2. 3.

x^ x^

root between 2
2
;

and

4.
5.

+ 10 X 3 x^ + 5 X 40 = x3 + 10 x2 4- 8 X +
6 x2

root between

and
3.

1.

root between 2
r=
;

and

120

root between 2 and

3.

460
6.

A COLLEGE ALGEBRA
2 x3
x^ x^
X*
x'*

7.
8.

9.

10.

11.

x^

_ 9X + 1 = root between 1 and 2. 5x 1 = root between 1 and 2. root between 5 and 6. 2 x2 23 X + 70 = root between 3 and 4. 10 x2 4 X + 8 = root between - 2 and 6 x3 + 12 x2 - 11 X - 41 =
a;2
;

x2

3.

3 x2

4X

13

two roots between 2 and

3.

Find
12.
14.

to the third decimal figure all the roots of the following equations

x3
x^

3 x2
3X

4X

+
0.

10

0.

13.

x3
X*

+ +

x2

2X

0.

15.

5 x3

x2 -

13 X

0.

16.

By

applying Horner's method

to the equation x^

17

compute

Vl7

to the fourth decimal figure.

17.

By
By

the

same method compute 2

V3 and V87

each to the third

decimal figure.
18.

aid of 845

and Horner's method

find to the second decimal

figure the real root of x'

19.

x^

2500

0.

By

aid of 844 locate the roots of x'

all

20.

Find

the roots of Sx^

+ 5 x^ - 6 x + 1 = 0. - Ux^ -x2 + 9x-2 = 0.

TAYLOR'S THEOREM.
847
Derivatives.

MULTIPLE ROOTS
n,

Multiply any monomial of the form ax" by

the exponent of x, and then diminish that exponent by 1. obtain n(ix"~^, which is called the derivative of ax", or,
precisely, its derivative

We
more

with respect to
a,

x.

In particular, the

derivative of a constant

that

is ax'^, is 0.

of the derivatives of the terms of a polynomial f(x) is called the derivative of f(x), or, more precisely, its ,/irst
derivative,

The sum

and

is

represented by
of /'(.r)
is

/"'(x).

The derivative
f(x), and

called the second derivative of

is represented by f"(x), and so on. Evidently every polynomial f(x) of the nth. degree has a series of n derivatives, the last of vvhich, f^"^ (x), is a constant.

THEORY OF EQUATIONS
Thus,
if

461
4,

f{x)

we have

/'

{x)

/" (x)
/"'{x)

/""(x)

= x* - 8 + 4 x2 - X + = 12 x^ - 24 + 8 x - 1, = 36 x2 - 48 X + 8, = 72x-48, = 72.
.S

a;3

x'-

All the subsequent derivatives are 0.

the

Observe that the second, third, derivatives oi first, second,

If in

derivatives of /(x) are

Taylor's theorem. replace x by x -\- h,

f(x)

f (x).
a^,x"

a^x"-'^

-\

we

848

we obtain

f(x

-f

h)

a,(x
(x

70" (x

By expanding

//)",

+ i (x-\-hy-' + + + h)"'^, and so on, by

a.

the bino-

mial theorem and then collecting terms, we can reduce this We shall show expression to the form of a polynomial in h.
that the result will be

f(x

J>)=f(x) +f'(x) ^ +f"(x) ^'

+f"(:x)

^,

(I)

where /' (x), f" (x),

This identity
theorem,
is

are the successive derivatives of f(x).

called Tai/Ior^s theorem.

For when the result of expanding (x- + //)'" by the binomial 561, is multiplied by a constant a and written in the form
a (x

//)'"

= ax'" -f max'" -^

+ w (m 1) ax'"' ^ h^ + 77i(m - !)(?, 2)ax'"-^ H

^

each of the coefficients

max'"'^, m(u/
is

l)ax"'~^,

m(jn

1)('"

2) ax'""',

the derivative of the one which immediately precedes Hence, if we arrange the expansion of each term of

it.

f{x

//)

in this form, the

of the leading terms in these several expansions will be f(x) the sum of the second terms will be
;

=a,{x sum

+ hy +

a, (x

7/)"-'

462

A COLLEGE ALGEBRA
of the leading terms, or

h times the sum of the derivatives

f'(x)h\ the sum of the third terms will be A^/2! times the

sum
and

so on.

of the derivatives of the second terms, or /"(a;)/i^/2! In other words, we shall have

f{x
Thus,

+ h)= f(x) + f (X) k+f'(x)^f +

if

--.+ /") (X) ^"

f{x)

we have

/(x

h)

= ciox^ + aix2 + a2X + = ao{x + h)^ + ai(x + +

3aox2

a',

h)-

ao (x

^)

Uz

849

850

THEORY OF EQUATIONS
that/' (a;)
is

463
is

by (x
f'{x)
it

ay
is

not divisible hy x a. but not by (x a)^,

Again, iif(x)
it

divisible
(1)

follows from

that

/(a) =/'()
general,
ii

=0
is

but/"(a)^0, and
by
a*

divisible

therefore from (2) that

a but not by (x

f(x)

divisible

by (x

a)''

a^. but not by (x

And

in

/''"'

but follows from (1) that /(a) =/'() /<'-!> (a) ^0, and therefore from (2) that f'(x) is divisible by )'". a)''^^ but not by {x (x Therefore, by 800, we have the following theorem.

= (a)

a)'''^^,

is 7iot a root of i' Theorem. J simjyle root of f(x) (x):=0 ; is a simple root but a double root of f (x) 0, aiid, q/f'(x) ts a root of in general, a inultiple root of order r of f (x)

851

order \

\ of f (x)

'

0.

roots of /'(x)

Thus, the roots of /(x) = 3x2 - 2x

= -

x"
8

x2

~ 8x +

12

are 2, 2,

3,

and the

are 2,

4/3.

therefore have the following method for discovering the Seek the highest 0, if there be any. common factor oif(x) a,ndf'(x) by the method of 465. If

We

852

multiple roots of f(x)

that /() and /'(a;) are prime to one another, But if we find that /(a;) and has simple roots only. f'{x) have the highest common factor <^(a;), then every simple root of ^ (.) = is a double root of 0, every double root f{x)

we thus hud

f(x)

of
if

(f)

(x)
(.r)

is

a triple root off(x)

= 0,

and

so on.
is

</)

(x

by (a* ay and f{x) by (x

is

divisible

For,

850,

ay, then /' (x)

divisible

by

+^ a)'"

Observe that
of

F(a-)=0
Example.

if the quotient /(a;)/ <^ (a-) be F(x), the roots are those of /(a-) 0, each counted once.

Find the multiple

/(x)

roots,
x-^

if

any, of the equation

4

highest

and /' (x) to be <^ (x) = x^ 3x 2. The roots of ^(x) = may be found by 802 and are 1, 1, Hence /(x) has the triple root 1 and the double root 2, that is,

Here /'(x)

= x5 - X* - 5 + x2 + 8 X + = Sx* - 4x3 - ISx^ + 2x + 8, and

factor of /(x)

0.

by

4G5 we find the

2.

common

its

roots are

1,

1,

1,

2, 2.

464
Observe that
that

A COLLEGE ALGEBRA
f(x)
f'{x)

= =

and that

F{x)
if

- 2)2, - 2) (5 X - 4), = (x + 1) (x =f{x,)/<p{x)

(x
(X

+ +

1)^ (x

1)" (X

2).

853

We may add that

have a root
highest
in

any
it

ttvo equations /(r)

and

i/^(x)

common,
factor

may
x^

be discovered by finding the

xli(x).

common

oif{x) and
x*

Example. Solve/(x) one of its I'oots is the negative

3 x^

0,

having given that

of another of its roots.

and the Evidently the two roots mentioned are common to /(x) = x) = x* + x* Sx^ 4 x 4 = 0, and may therefore be equation /{ obtained by finding the highest common factor of /(x) and/( x). By 465 we find this highest common factor to be x^ 4. Hence
the roots mentioned are
2,

2.

Dividing /(x) by x^

4 and solving the

x resulting depressed equation x^ roots of f(x) are (1 i Vs) /2.

0,

we

find that the other

two

EXERCISE LXXVI
1.

Find the

first,

second,

derivatives of 2x5

4 x*

4-

x^

20 x.

2. 3.

Given /(x)

~x*

2x^-\-\, find/(x

h)

by Taylor's theorem.

Using the formula 849, (II), express (1) x* + x^ - 32 in powers of x - 2 (3) (x^^ + 1) / (2) x5
;

+
(x^

in
1)

powers of in terms

of X
4.

The following equations have multiple

(1)

roots.

Solve them.
0.

5.
6.

= x3-3x -2 = 0. (2) 9x3 + 12x2 -llx + 2 = 0. (3)4x^ + 12x2 + 9 = 0. (4) x*-4x3 + 8x + 4 = 0. (5) 2x* -12x3 + 19x2 -Gx + 9 - x3 - 4x2 - 3x - 2 = 0. (6) x6 - 2 x3 - x2 - 4 X + 12 = 0. (7) x< - x - 2 x3 + 2 x2 + X - 1 = 0. (8) x6 = 0. (9) 3x5-2x* + 6x3-4x2 + 3x- 2 Show that x" a" = cannot have a multiple root.
If the

equation x^

12x +

a=0
+

has a double root, find

0x2

a.

7.

Determine a and

b so that 3 x^

may have

a triple

root,

and

find this root.

THEORY OF EQUATIONS
8. 9.

465

Show

that x*

qx"^

+s=
tliat x^

cannot have a

triple root.

Find the condition

px- +

may

laave a

double root.
?

10. 11.

What

is

the

form oif{x)

if it is

exactly divisible by /'(x)

The equations x* + x'' + 2 x^ + x + 1 = have roots in common. Solve both equations.

12.

and

x*

x^

The equation

the roots of x*
13.

x^

x^ 20x IG = has a root v?hich is twice 3x 1 = 0. Solve both of these equations.

one of

Show

that

if

a cubic equation with rational coefficients has a

multiple root, this root must be rational.

14.

Show that if an

coefficients has a multiple root, this root

equation of the fourth degree/(x) = with rational must be rational unless /(x) is a

perfect square.
15.

Prove that

if

a
0,

is

a root of /(x)

0, of

order

r, it is

a root of

all

the equations /'(x)

f"(x)

0,

/('-i)(x)

0.

VARIATION OF A RATIONAL INTEGRAL FUNCTION

Theorem.
inff

Let
x,

f (x)

denote a polynomial arranged in ascend-

854

and let b denote the numerical value of its leading coefficient and g that of its numerically greatest coefficient. The leading term of i(x') ivill be numerically greater
powers of
than the sum of the remaining terms for are numerically less than b/(b + g)Eirst, let /(a-)
let x'

all values

of x which

b^x

boX^

so that b

= [b^],

and

denote the numerical value of

b^x

x.

Then
gx'

^a-^"

is

numerically less than (or equal

x''^

to)

gx''^

or g (x'

x'<

1, it is less

x') is less than b when x' <b / (b g). so that b =\b-^\. We + b^x^ -f b^x^ Second, let/(ir) = then have \b^x'^ + b^x^ < |6ix| when \b2X + b^x"^ + \< \bi\, that is, when x' < b / (b + g), and so on.
But
gx' / (1
-{-

than gx' / (1

+ -

), 235,

and therefore when

x'),

704.

ftia:
-\

-\

Thus,

if

x'<5/(5 +

9),

/(x) = that

5x + 3x2 - 9x*, ^e have is, when x'<5/U.

|3x2

9x*|<|5x| when

466
855
Theorem.

A COLLEGE ALGEBRA
Let
f (x)

denote a polynomial arranged in descend^

and let a denote the numerical value of its leading coefficient and g that of its numericalbj greatest coefficient. The leading term o/f(x) ivill be numerically greater than the sum of the remaining terms for all values of x which
ing jiowers of x,
are numerically greater than (a

g)/a.
h a, so that a
x.
ic"

For let/(a-)
let x'

a^x"" +

ajic"-'

-^

= |ao|,
1-

and

denote the nuinerical value of

h
1-

Hence
But, that

We have a^^" + a^x"-^ H H a^x" > a^x"

^
|
\

(a

+ a^/x

-\

a"/x").

1-

when
|
|

a \>\ai/x-\

\-a/x^\.

854, ||>|ai/a; H

a/x"\

is,

when
7)/

x'

>

when l/x'<a/(a
2]

g),

(a

+ g) / a.
when

Thus, if/(x)

3x3

+ x2- 7x +

'>(3

3,

that

is,

2, we have |3x3|>|x2 - 7x + when x'>10/3.

From this theorem it evidently follows that the number (a + g)/a is greater than the absolute or numerical value of any root of the equation = 0, whether the root be real or imaginary. f{x)
856

If & is a root of f (x) = 0, the values of{(x) and have contrary signs when x is slightly less than a, and the same sign when x is slightly greater than a.
Theorem.
f'(x)

For express /(.r) and /'(a-) in powers of x then divide the lirst expression by the second.
simple root, so that f{a) reduced to the form

a,

849,

and

but/'()

^ 0,

the

When a is a result may be

The numerator and denominator of the fraction on the right are polynomials in a; a. Hence for all values oi x a which are small enough to meet the requirements of 854 their signs wull be those of their common term and the

leading

f (a),

fraction itself will be positive. then be the same as that of a;

according as

a;

<a

or

ic

> a.

sign of f(x)/f'{x) will a and therefore minus or plus But when the sign of f(x) ff (x)

The

THEORY OF EQUATIONS
is

46?

of f(x)/f' (x)

minus, /(a*) and/' (a;) have contrary signs, and when the sign is plus, /(a-) and/'(a:;) have the same sign.

Wlien a
f(x)
^

is

a multiple root of order r

/'-''>

we have,

850,

f'(x)

a) {a) /r\ + terms involving {x 1)! + terms involving {x /^'H*)/(*'

a)

from which the theorem follows by the same reasoning as when a is a simple root.
Rolle's theorem.

Between two consecutive

roots of f (x)

there

is

ahvai/s a root o/'f'(x)^0.

;8i

For
a

let

and

^S.,

number Then

slightly greater than

be the roots in question, and let c denote and d a number slightly less )8i
/B^.

than

p., so

that

(Si

<c<d<

same sign as f(c), 856, and /(c) has the same sign as f(d), 834 but f(d) has a different sign from that of /' (d), 856. Hence /' (c) and /' ((/) have contrary Therefore a root of f'(^x)0 lies between c and d, signs. that is, between ft^ and ^o, 833. is 2 x - 3 = 0. The Thus, if f{x) = x2 - 3 X + 2 = 0, then /'(x) =
f'(c) has the
;

roots of /(x)

=
2.

are 1

and

2,

the root of /'(x)

=
+
9

is

3/2, and 3/2

lies

between

and

Example. Prove that /(x) = x^ + x^ between 1 and 2. (Compare 844, Ex.)

lOx

has two roots

Since /(I) = 1 and./"(2) = 1, there are two roots or none between 1 and 2. If there are two roots, /' (x) = must also have a root between 1 and 2, and this root must lie between the two roots of /(x) = 0. 10 = But /'(x) = 3x- + 2x has a root between 1 and 2, for 5 and /' (2) = 0. Solving, we find that this root is 1.5 approxi/' (1) = Moreover /(1. 5) = .375 is minus. Therefore, since both /(I) mately. and /(2) are plus, /(x) = has two roots between 1 and 2, namely, one between 1 and 1.5, and another between 1.5 and 2.

Theorem. If the variable x is increasing, then, as it passes through the value a, the value o/f (x) is increasing if i' (a,) > 0,
but decreasing i/f'(a) but f"(a) i/'f'(a)=

< 0. ^ 0,

(a) is

maximum
f"(a)

value o/f (x)

0.

when

f"(a)

<

0,

minimum

value

when

>

468
For by

A COLLEGE ALGEBRA
849 we have

- a)+/"()(x of 12 + .... /(x)-/(a) =/'() (X is a polynomial in x a, and for all The second member values of x which make x a small enough numerically to
!

meet the requirement

of

854, the leading term will control the

We

sign of the entire expression and therefore that of /(a-) /(a). Then shall suppose x restricted to such values.
1.

If

has the same sign as (x a). Therefore, since x a changes from minus to plus as x passbt. through a, the same is true of f(x)f(a'), that is, /*(.t) is then increasing from a value less than f{a) to a value greater than f{n).
2.

f{a)

>

0,

f{a){x

a),

and therefore f{x)-f{a),

If /' (a)

<

0,

/' {a) (x

fl)

and {x
1,

a)

have contrary
that f{x)
is

Hence, reasoning as in decreasing as x passes through

signs.
3.

we conclude
the sign

a.

If

and therefore that of /"(a); for a)'^ is positive whether x < a or x > a. Hence when (x /"() < 0, we have/(a') <f(a) just before x reaches a and also just after x passes a, which proves that /'(a) is a maximum
that oi

/'()= f" (a)(x

ay /2

but/"()^0,

oif{x)-f{a)

is

value of /(cc), 639. And the same manner when/"(fl) f(a) is a minimum, value of /(a-).

>

0,

we may show

that

It
is

may
2).

be added that

if

/"(a)=0 but/"'(a)^0,

/(a)

not a

maximum
And,

or

minimum

Ex.

in general, if all

value of f(x) (see 859, the derivatives from the first

to the rth, but not the {r is a maximum or minimum

l)th, vanish

when x

a,

f{a)
r
is

when
+ 9x
?

is

odd, but not

when

even.

Example.
values of/(x).

Ls/{x)

x^

Gx'^

increasing or decreasing as x,

increasing, passes through the value 2

Find the

maximum and minimum

8).

We find/'
is

(.x)

3x2

- T2x +
is

3(x

1) (x

Hence/' (2)
2.

=-

negative.

Therefore /(x)

decreasing as x passes through

THEORY OF EQUATIONS

469

We have/' (x) = when x=l and when x = 3. Moreover/" (x) = G x 12, Hence is negative and/" (3) = 6 is positive. and therefore /"(I) = Lis a minimum /(I) = 3 is a maximum value of /(x), and /(3)=
value.

Let us now consider how the value of a Variation of f (x). polynomial f(x) with real coefficients varies when x varies oo to + cc. (jontinuously, 214, from
2x2 x + 2. Example 1. Discuss the variation of /(x) = x^ The roots of /(x) = are - 1, 1, 2, and/(x) = (x + 1) (x - 1) (x - 2). oo, /(x) = co when x is between oo and 1, Hence, when x = = 1, /(x) ?-0 when x is between 1 and 1, /(x) is negative when x = 1, /(x) = when x is between 1 and 2, /(x) /(x) is positive when x when x is between 2 and oo, /(x) is when x = 2, /(x) = is negative

859

when x co, /(x) positive The roots of /'(x)

;

approximately. is positive, but

negative.

= co. = 3x2-4x -1 = are (2 V7)/3, or - .2 and 1.5 When x < (2 - V7) / 3 and when x > (2 +^7) / 3, /' (x) when x is between (2 V7)/3 and (2 f V7)/3, /'(x) is =
/(x)
is

Therefore,
to

858,

to
(2

(2 V7)/3, is continually (2+V7)/3, and is again

oo continually increasing as x varies from decreasing as x varies from (2 V7)/3

continually increasing as x varies from

+ V7) / 3
It

to

00.

follows from this, 639, that /(x) has a maximum value when V7) /3, and a minimum value when x = (2 + V?) /3. This is in 4 is negative when x = (2 V7)/3, agreement with 858, for/"(x)= 6 x

(2

and positive when x = (2 + v7)/3. The variation of /(x) will be exhibited to the eye if we put y f(x) and then construct the graph of this

equation by the method of 389. We thus obtain the curve indicated

in

the

accompanying

figure.

The

points A, B, C at which the curve cuts the X-axis are the graphs of the
roots

-1,

1,

2of/(x)=:

0.

The por-

tions of the curve above the x-axis

correspond to positive values of /(x), those below to negative values. The uppermost point on the curve between and B corresponds to the maximum value of /(x), the lowermost point

between

and C

to the

minimum

value.

4T0
As X
from an
varies

A COLLEGE ALGEBRA
from

infinite distance

to oo the corresponding point on the curve moves below the x-axis upward and to the right through

A to the maximum point, then downward through B to the minimum point,

then upward again tlirough
If

above the x-axis. term of /(x), the graph of y =/(x) would be shifted vertically upward and the points B and C would at first approach coincidence in a point of tangency and then disThe corresponding roots of /(x) = would at first become equal appear. and then imaginary.
to
infinite distance

an

we were gradually

to increase the absolute

Example 2. Discuss the variation of /(.x) = x* 2 x-^ -f- 2 x 1. The roots of /(x) = are - 1, 1, 1, 1, and/(x) = (x -)- 1) (x - 1). Hence when x = oc, /(x) = oo when x = 1 /(x) = when x < 1 and when x >1, /(x) is positive when x is between 1 and 1, /(x) is
;
, ;

negative.

Here /' (x)

/'(x)
is

are

between

1/2

= 4 x^ - 6 x2 + 2 = 2 (2 x + 1) (x - 1)2, and the roots of 1 /2, 1, 1. When x < 1/2, f (x) is negative when x
;

and

and

also

when x > 1, /' (x)

is

positive.

Therefore,

858, /(x) is continually decreasing as x varies from co to 1/2, and continually increasing as x var es from 1/2 to 1 and from 1 to oo. Hence /(x) has a minimum value when

1/2, but
is

it

has neither a maxi 858.

mum nor a minimum value when x = 1.

This
in

For /" (x) and /'" (X)

X X

= 12 x2 - 12 x = 12 x (x - 1) = 24 X - 12. Hence when

;

agreement with

1 /2, /" (x) is positive bi;t when /" (X) = and /'" (x) it 0.

The graph

of y

=f(x) has the form

indicated in the accompanying figure. The point where the curve merely

cuts the X-axis corresponds to the root 1 of /(x) = 0, and the point B where

the curve both touches and crosses the

X-axis corresponds to the triple root 1. The lowermost point of the curve corresponds to the minimum value of Its coordinates are 1/2, 27/16. /(x).

860

in these examples, so in general, iif(x) is of odd degree, oo leading coefficient being positive, when x varies from to +00, /(a-) increases from oo to the first maximum value,
its

As

then decreases to the

first

minimum value, and so

on,

and

finally

THEORY OF EQUATIONS
increases from the last

471
It is possible,

minimum

value to

+ oo.

however, that there are no maximum or minimum values, for the equation /'(a:) = 0, being of even degree, may have no real

The graph of y =f(x) extends from an infinite distance below the x-axis to an infinite distance above the cc-axis. It once at least. crosses the a*-axis an odd number of times,
root.

On

the other hand,

if

f{x)

is

of even degree, f{x) begins by

decreasing from

minimum value and ends by increasing from the last minimum value to + oo. In this If case the graph of ij =f{x) need not cross the a; -axis at all. it does cross the axis, it crosses an even number of times.

oo

to the first

In most cases we can obtain a sufficiently accurate representation of the graph of y =f(x) by the method of 889, which consists in assigning a series of values to x, computing the
corresponding values of y, plotting the pairs of values of x, y " thus found, and passing a '' smooth curve through all these Such a curve will indicate roughly where the true points.

graph crosses the

points
lie.

a;-axis

and where

its

maximum and minimum

But

to obtain the points of crossing with exactness,

we must

solve the equation /(a;)

positions of the

maximum and minimum

0; and to obtain the actual points, we must solve

0. the equation /' (a-) To every multiple root of /(.r) there corresponds a point of tangency of the graph with the If the order of the multiple root is odd, the graph X-axis. also crosses the a;-axis at this point.

EXERCISE LXXVII
1.

Discuss the
its

variation

of

finding

maximum and minimum

- 2)2 = a;3 - 3a;2 + 4, /(x) = (x + 1) (x values if any, and draw the graph of

y=f{^).
2.

Treat in a similar manner each of the following functions.

(1)

(3) x3

(5)
(7)

- X + 1. - 12x + 14. X3 - 3X2 + 5. (x2 + X + 1) (X +

2 x2

(2)
(4)
(6)
2).

(8)

+ 1) (X - 2) (2 X - 1). - 5x2 + 3x + 9. + 1)2 _ 2)2. - 1) (x 2) (x + 3). x(x

(X

x3

(j;

(j;

-I

472
3.

A COLLEGE ALGEBRA
Find the graphs
of each of the following fractional equations
a;

plotting the points corresponding to

1,

by

1/2,

0,

1/2,

1,

, 4.

(1).=

(j;

't:'\. - 2) (X - 3)

i^)y(X

"^'-'^

1) (X

3)

STURM'S THEOREM
361 be any equation which has Sturm functions. Let f(x) = no multiple roots, and let/i(a;) be the first derivative of /(x). Divide f(x) by fi(x) and call the quotient qi, and the remainder, with its sign changed, /g (^)' Again, divide /i (a;) by fo{x), and call the quotient q.^, and
its sign changed, /s (a;). so on, modifying the ordinary process of finding the highest common factor of f(x) and fi (x) in this respect only the sign of each remdlnder is changed, and care is taken to

the remainder, with

And

make no

Since /(a;)=

other changes of sign than these. has no multiple roots and therefore /(cc) and

fi(x) have no common factor, 851, we shall finally obtain a remainder which is a constant different from 0, 465. Call this remainder, with its sign changed, f,^.

The sequence

of functions

consisting of the given polynomial, its first derivative, and the several remainders in order, each with its sign changed, is called a sequence of Stiinn or a sequence of Sturm functions.

862

Relations

among

the

Sturm

functions.

These functions

are,

by

definition, connected

by the series of identical equations

(1)

f{x)^q,f,{x)-f{x),

Mx) =
/2 (x)

= =

q,fX^)-f{x), qsfs (x) -/, (x),

?m-l/m-l(^) "/n-

(2)

(3)

fm-ui^)

(^

- 1)

THEORY OF EQUATIONS
From
1.

473

these equations

we conclude

that

Two

consecutive functions cannot vanish for the

same

value of

X.
if

Thus,

from
is

(2).

c,\t follows both/i(a;) and /a (a;) vanish when x that fz{x) also vanishes; therefore, from (3), that
is 0.

fi(x) vanishes; and therefore, finally, that/,,

But

this

contrary to hypothesis.
2.

When

for a certain value of x one of the intermediate

functions

fj (x), i^

(x),

-,

f^-i

(a;)
it

vanishes, the

fmctions ivhich

immediateli/ 2^recede

and folio iv

have opposite signs.

Thus,
it

if

/2(c)=0,
/i
a,

follows from (2) that

(c)

= f (c).
be

Sturm's theorem.

Let

and h

any two real numbers neither

863

of

a root of f (x) = 0. The difference between the number of variations of sign in

wli ich
is
r / \ n
^

the sequence

r
,
f m

f(a), fi(a), 2 (a),

and that in the sequence

m f(b), fi(b), f,(b), ...,f
is

the

number of roots o/f (x)

fix

=
b,

which

lie

between a and

b.

To

the ideas, suppose a <.

6,

and suppose x to vary con-

tinuously from a to

214.
b,

As X

varies

from a

to

the sign of the constant f^ remains

unchanged, and the only changes whrchr are possible in the signs of the remaining functions, and therefore in the number
of variations of

^n in the sequence
occur
0,

are such as

may

f,n, f(^): fi(x), Ai^), when x passes through roots of the

equations f(x)
1.

fi(x)

0,

and so

on,

835.

But

The mimber of variations in the sequence is neither increased nor diminished when any function except the first^
f (x),

changes

its sign.

474

A COLLEGE ALGEBRA
e is

Suppose, for instance, that

/z
(35)

a root of f^ (x)

and that

changes from plus to minus as x passes through c. Since c is a root of /g (*) = 0, it cannot be a root of either

f^ (x)
lies

number h

And if Ave take a positive or fs (x) 0, 862. so sniall_.that__rio r oot of fijx)_^_0 or of fsix)^
c

between

7* and

c,

or

between

g_

and
its

-\-

h,

neither of

the functions fi(x) or f3(x) will change from c to c h, 835.

sign as x varies

7^,

c, f^^ix) and/3(.r) have opposite signs, 862. Suppose that fi(c) is plus; then fz(c) is minus, and we have the following scheme of the signs of f-^{x), f^ix), /^(x) for values of x between c h and c -\- h
:

But when x

+ =

flix)

f2{x)

fslx)

=c

THEORY OF EQUATIONS

475

In other words, the sequence has a variation between f{x) and/i(.r) just before x reaches c, and this variation is lost as

X passes through

c.

Hence the sequence

tion as X varies

of

Sturm functions never gains a

b.

varia-

But, on the other hand, it loses a variation each time that x passes through a root of. f{x) 0,

from a

to

and then only.

Therefore the difference between the number of variations
in

f{a), fi(a), f.(a),

and in
is

f(b), f,{h), f,{h), .,/.

the

number

of single variations that are lost as

through the roots

other words, it demonstrated.
If

oif(x)=

is

x passes which lie between a and b. In the number of these roots, as was to be

we

rnultijAe voots,

which has apply the method of 861 to an equation f{x) = we obtain a sequence/ (x),/i(x), ,/,(x), (1) the last term

Divide all the highest common factor of all the terms, 465. thus obtain a sequence of the form 0(x), the terms of (1) by /,(x).
of

which

is

We

01

(x),

1, (2)

which, as

is

easily

shown, possesses

all

the properties on
(x)

which Sturm's theorem depends.

that
is
is,

Hence the number

of roots of

0,
6,
1

852,

the

number

the difference between

of different roots of /(x) 0, between a the number of variations in <^(a), <pi{a),

is

and

and in <p{b), 0i(6), , 1. And this difference between the number of variations in /(a), /i(a),

/i

(^)i

by fm

{a)

fm (b) and /,

for multiplying the sequences </> (a), variations. {b) respect ively will not affect their

same as that /,() and in /(6), 1 1 and (6),

the

Applications of Sturm's theorem.

Sturm's theorem enables

real roots a given
to

864

one to find exactly how numerical equation has.

many
It

different

also

enables one

find

how

many

of these roots lie between any pair of consecutive integers and therefore in every case to solve the problem of locatBut this method of locating roots is very ing the roots. laborious
fails.

and

is

used only when the simpler method of

836

476
Example
1.

A COLLEGE ALGEBRA
Locate the roots of x^

3 x^

4x

+1=

0.

- 4x + 1 and/j (x) = 3x2 + 6x - 4, Here/(x) = x-' + 3x2 Arranging the computation of the remaining functions as in have

468, 3,

we

3+ 6-

Hence
/(x)
/i(x)
/2(x)

/3
3

= x3 + 3x2- 4x + = 3x2 + 6x-4, = 2x 1, =L

1,

15

defined in

Observe that the fi (x) and /s here obtained are not the fi (x) and /s 861, but these functions multiplied by positive constants* We should have lessened the reckoning had vFe divided /i(x) by

/2(x)

2(x

.5)

synthetically.

It

is

often best to use the synthetic

is

method in the final division. 1. "When X is very great numerically, the sign of a polynomial Hence the following table. of its term of highest degree, 855.
X X

that

=
=

f{z)

/i(x)

Mx) h

THEORY OF EQUATIONS

.477

The only property of the final function / of which any use made in the proof in 863 is that its sign is constant. Hence if, when computing the Sturm functions oi f(x)= in order to find the number of roots between a and b, we come upon a function j^ (x) which has the same sign for all values of X between a and h, we need not compute the subsequent
is

For it follows from the proof in 863 that the functions. required number of roots will be the difference between the number of variations in f(a), ,fp{a) and in f(b), ,fp(b).
Example
Here
/i
3.

(a;)

Sx^

How many real roots has f{x) = x^ + x'^ + x + l = 0? + 2 x +1, and, since 2^ < 4 3, this is positive for

all

real values of x, fi35, 823.

The

signs of /(x),

Hence we need not compute /o(x) and/3. - 00 are -, +; for x = 00 they are fi{x) f or x =
has one real root.

+.

Hence /(x)

EXERCISE

LXXVm

By
1.

aid of Sturm's theorem find the situation of the real roots of the

following equations.
x3 x3

G x2

5X

+
0.

13

= =

0.

2.

x3 x3

4 x2

10 X

41
8

0.

3.

+ 5x +

4.
0. 0.
0.

5.

a;3_a;2_i5a;4.28

6.

7.
9.

2x4-3x2 + 3x-l = x-i - 12 x2 + 12 X - 3 =

8.

0. + x* - 4x3- 5x2 +18x + 20 = 0. x* - 8x3 + 19x2 -12x + 2 = 0.

3x2

+ 8x +

10.

X*

2 x3

6 x2

8X

0.

By
11. 13.

aid of Sturm's theorem find the

number

of the real roots of each

of the following equations.

4x3
x"

0.

0.

12.
14.

X*

+1=

X*

+ x3 + x2 + X + 1 = 0. - 6 x3 + x2 + 14 X - 14 =

0.

15,
roots.

be an equation of the nth degree without multiple Let /(x) Show that the condition tliat all the roots of /(x) = be real is that

there be n

and that
16.
all

+ 1 terms in its sequence of Sturm functions /(x), /i (x), -,/ the leading terms of all these functions have the same sign.

By

the roots of the cubic x3

-I-

aid of the theorem in Ex. 15 prove that the condition that + px + g = be real and unequal is that

4p3

27 q2 be negative.

478.

A COLLEGE ALGEBRA
SYMMETRIC FUNCTIONS OF THE ROOTS

865

Theorem
are

If the roots of f (x)

that f (x)

^^

^,,

,^,,so

= x" + bix"-^ = (x - ^0 (x - ^.^

-\

h b

(x

= - ft,),

Thus, suppose that n

.

= 3, so that = (x-ft,)(x-l3.;)(x-l3,). f(x)

(1)

Substituting x

+ h for x in (1), we have - /30 + A] [(x - ft,) + h-\ [(X - ft,) + A], f(x + h) = [(x
can reduce each
h,

(2)

We

nomial in

the

first

member of (2) to the form member by Taylor's theorem,

of a poly 848, the

second by continued multiplication, as in 5ob. Since (2) is an identity, the coefficients of like powers of h in the two polynomials thus obtained must be equal, 284.

But since f(x

it

+ h)=f(x) +f(x)h^
is
ft.,)

cient of h in the first polynomial

is

{x

- ft.^ {X -

Hence

+ {x(X fix) = (xft.;)

fts)

(x

- ft,) + (x- ft,) (x (^

f'(x).

848, the coeffiIn the second

ySo).

ft.,)

- A) (^ - A)
^^
so on.

+ ix-fti)(x-ft.,)
x-ft,x-ft,x-ft./ ft?^ =zf(x)/{x ft^, and

since (x

ft,)

(x

This reasoning is applicable to an equation of any degree n. In the general case there are n factors in the second member of (2), and when this member is reduced to the form of a polynomial in }l, the coefficient of h is the sum of the products x y8, taken ?i 1 at a of the binomials x ft,, x ft,,
,

time, 558.

That one of these

be written /(a;) /
(ic

i)roducts

fti

may

ft,),

which lacks the factor and so on.

THEORY OF EQUATIONS
Thus,
if

479
2) (x

/(x)
/'

we have
and
(x

2) (x

= x^ - 6 a;2 + 11 X = 3 x2 - 12 x + 1 - 3) + (x - 3) (x - 1) +
(x)
1

=
-

(X

1) (x

3),

(x

1) (x

2)

3 x2

12 x

11.

Theorem
equation

2.

The sums of

f (x)

an of can be expressed rationally in terms of its

like jjowers
all the

roots of

866

coefficients.

Thus, suppose that the equation

is

f{x) = Let
a,
ft,

ic^

b^x"

b.x

&3

0.

(1)

y denote the roots of (1) and let

s^, S2,

s^

have

the meanings
Si

(3

y,

sj

'

/3'

y',

s.

a'-

yS'-

y^

are to prove that s^, s^, in terms of the coefficients ^1, ^>o,
1.

We

can be expressed rationally

b^.

By

tlie

preceding tlieorem,

865,

we have
X

^^
Since f{x)
is

a
.r

a-,

fi
a?

y
and x

divisible

by

;8,

y,

each of

the fractions in (2) represents a polynomial in x which can be found by the rule of 410. Applying this rule and then

adding the

results,

we have
0^2
^-i)

+ {a + ^ + (' + ^^1^ + ^2) fix) /(x-(3)==x-' + (fi-^b,)x+ {P' + b,fi + b,) - y) = x^ + (y + b,)x-{-(y' + b,y + b,) f(x)/(x f (x) =3x'+ (s, + 3b,)x+ {s, + b,s, +
f{x) / (x

a) =

b,)

(3)

But by

definition, 847,

we

also have

f'(x)=3x^-

2bix

b2.

(4)

Equating the coefficients of like powers of x in the two

expressions (3) and (4) and solving for
Sj
s^, s^,

we have
(5) (6)

&i

&i,

..

Si

S2

^i-si

+ 3 62 =

Z>2,

.'.

= bi, 2 &2S2 = ^1

480
2.

A COLLEGE ALGEBRA
From
a,

the values thus found for

Si

and

s^
:

we can

obtain

the values of Since

/3,

s^, s^,

successively as follows y are the roots of (1), we have

a

/3'

+ b^a^ + ha + ^ = 0, + Ih(3'+hP + b, = 0, = 0. y3 + i,^y^ + ijy +

i>^

(7)
(8) (9)

Adding

these identities,
^3

we

obtain
0,

^182

+ boSi + 3 ig =

(10)

which gives s^ rationally in terms of bi, b^, b^, s^, s.2 and therefore, by (5), (6), rationally in terms of bi, b^, b^. Next multiply the identities (7), (8), (9) by a, yS, y respecWe obtain tively and add the results.
Si

biSs

b^Sz

b^Si

0,

(11)

which by aid of
i>i,

in (5), (6), (10) gives s^ rationally

terms of

b^, b^.

And
by
a"^,

in like
/?-, y^,

manner,
/3^,

if

we multiply we
Se

(7), (8), (9)

by a^, multiplications add

S5

y^,

and so on, and after each

obtain identities

respectively series of

results,
^32

^l4

^2^3

0,

^1^5

^'2-^4

^'3^3

0,

which give

S5, Sq,

rationally in

terms of

b^, bn, b^.

similar reasoning the theorem of any degree n. equation f{pc)

By

may

be proved for an

If a, 1. 7 denote the roots oi x^ - 2 x- + ^ x + 2 = 0, find Sl/an=l/a + l/^ + l/7, 2 l/a2 = 1/^2 4. 1/^2 4. 1/^2^ 2 1/a'' = l/a3 + 1/^3 + 1/73. Applying the transformation x = \/y, and dividing the transformed y + / 2 = 0. equation by the coefficient of y^, we have y^ + 2 y'^ For this equation, by substituting 61 = 2, 62 = 1, &3 = /2 in the formulas (A), (6), (10) above, we obtain Si = 2, S2 = 6, S3 = 31/2. - 2, S l/a^ = 6, S l/aS = _ 31/2. Therefore, 814, S 1/a =

Example

/3,

THEORY OF EQUATIONS
Example show that
Si
2.

481
b^x"^

For the equation /(x)

2

=
Ss

x*

6ix^

63X

64

0,

Si

+ +

61

61S3

= 3 61, So 4- 61.S1 + 4 62 = + 62S2 + &3S1 + 4 64 = 0,

Si, S2, S3, S4

62,

61S2

85 4- &1S4

+ +

62S1

62S3

+ +

4 63
&3S2

= 63, + &4S1 =

0,

and compute

in terms of

&i, 62* &3) ^4.

The preceding formulas also show that s^, s^, s^, are inteyraZ functions of the coefficients of the equation when, as in is 1. (1), the leading coefficient
Theorem
2.

867

Every rational symmetric function of the

ca)i be

roots

868

of an eqiiation f(x)=
of
its coefficients.

expressed rationally in terms

Let the roots of /(cc)

be

or,

/?,

y,

v.

Every rational symmetric function

1,a^p', Ita^p^-f,

of a,

(3,

can be
2a:'',

expressed rationally in terms of functions of the types

and so on, 544. Hence it is only necessary to prove our theorem for functions of these several types. This was done for the type 2<;i:'' = s^ in 866, and we shall now show that Str''^, and so on, can be expressed rationally
in

terms of functions of this type

1.

s^^.
-\
.

The type %aPp'

{a"

a''^'

+
)

ji"a''

The product
is

/?'

(a'

+^

(1)

the

sum

of the

two symmetric groups of terms

aP + n j^ pp + 1 ^
\ ,

(2)
(3)

aPpJ

+ p'\x'>^

and (2) are rational functions of the coefficients, Hence (3), or 2a''/8^, which may be obtained by sub 866. tracting (2) from (1), is also such a function. Since (1) is s^s^ and (2) is s^+g, we have the formula
(1)
2a''iff^

But

= V,-P +

-z-

(4)

When p = q,
becomes 2

the terms of (3) are equal in pairs and (3)

'S^a^pp.

We

then have 2

Set''/?"

= s^ s^^.

482
2.

A COLLEGE ALGEBRA
The type ^a^^^Y
(a''/3''

a'^jS^'f
-\

/3''aY H

The product
is

+ /S"a''

) ('"

/T +

y-"

(5)

the

sum

of the three symmetric groups of terms aP + '-p'J

+ fiP + '-a" ^ ^ pp^l+r ^ppi aPp''Y + (i^'a^Y

+r
-\

(6) (7)

J^

(8)

But we have already shown that (5), (6), and (7) are rational Hence (8), or 2a''/3V5 which functions of the coefficients. be obtained by subtracting the sum of (6) and (7) from may
(5), is also

such a function.
the group (8) becomes 2 '%aPj^Y'i

When jj = q,
it

when^ = q = r,

becomes 6
3.

"^aP/S'^Y'-S.a''l3''y't',

The types

and so

on.
coeffi-

We may prove
cients

that these are rational functions of the

in 1

by repetitions of the process illustrated begin by multiplying 2a''j8''y'" by a" + ^8" + y"
Example.

and

2.

We

Show

that

EXERCISE LXXIX
1.

For the equation

Ctl,

aox^

aix^

a^x

a3

=
qx

find S3

and

s^ in

terms

of Uo,
2.

2i

ots-

If a,

/3,

7 denote the roots of x^

a/S^ in

+ px^ +

r =: 0, find

21/ a^,

S 1 /a^, and S
3.

terms of p,

q, r.

Find the equation whose roots are the cubes of the roots of
0.

x3-2x2 + 3x-l =
4. If a,
/3,

find the values of the following

x^ + 3 x + 4 = 0, 7 denote the roots of the equation x^ symmetric functions of these roots by the
(2)

methods
(1)

of 866, 867.
Si, S2, S3, Si.

Sa'^.

(3)

2as^7.

(4) 2a3i327.

(6)Sl/a4.

(6)

Sa2^/7.

CUBIC

AXD BIQUADRATIC EQUATIONS

483

XXX. THE GENERAL CUBIC AND BIQUADRATIC EQUATIONS

In the preceding chapter we have shown Algebraic solutions. that the real roots of a numerical equation can always be
869

found exactly or approximately, and it is possible to extend the methods there employed to the complex roots of such equaAs hardly need be said, these methods are not applitions. To solve such an equation we must cable to literal equations.
obtain expressions for its roots in terms of its coefficients. say that an equation can be solved algehraically when its roots can be expressed in terms of its coefficients by apply-

We

ing a finite

number

of times the several algebraic operations,

namely, addition, subtraction, multiplication, division, involution and evolution.

We have already proved, 631, that the general quadratic equation has such an algebraic solution, and we are now to prove that the like is true of the general cubic and biquadratic But general equations of a degree higher than four equations.
cannot be solved algebraically. In 646 we showed that the equaCube roots of unity. tion x^ = 1 has the roots 1, (- 1 -f t V3)/2, (- 1 - i V3)/2.

870

Hence each
will be

of these numbers is a cube root of unity. The third found to be the square of the second. Hence if we
w^.

represent the second by w, we may represent the third by 1 =.0 lacks an x^ term, we have, 805, 1 -f w + w^ Since x^

= 0.

Similarly every number a has three cube roots, namely, the a. If one of these roots be three roots of the equation a;^

V a, the other two will be w Va and w^

The general
cubic.

a.

Cardan's formula.

By

the method of

818

871

every cubic equation can be reduced to the form

x^
in
-{-

px

+q=

(1)

which the

a;'^

term

is

lacking.

484
In
(1)

A COLLEGE ALGEBRA
put x

= y + z.

(2)

We
or

obtain

y'

+ ^' + (3y.t+p)(y + )+? =

0.

(3)

As the variables y and z are subject to the single a second condition upon them. (3), we may impose

condition

We

suppose
(3), y^

^yz
-\-

and therefore, by

z^

From

(5), (4),

y^

+ p = 0, + q = 0. + z' = -q,
y^z^^-j?l21.

(4)

(5)
(6)

and from

(7)

Therefore,

636, y^ and z^ are the roots of a quadratic equa^2

tion of the form

+ y_ ^3^27 =

0.

(8)

Solving (8) and representing the expressions obtained for the roots by A and E respectively, we have

y%

These equations
namely,

(9) give three values for

y and three for

s,

870,

= Vl,

3/

3/

3/

(10)

a)Vl, w^VZ,

But by

(4),

= V^, yz = p/3,

0,-^, 0)2-^.

(11)

and the only pairs of the values

of y, z in (10), (11)
y,

which
-y/li
;

satisfy this condition are

,

v'J,

w v'j

w'-'

^/^

w^ -v^,

w V^.

Substituting these pairs in (2),

(1),

we

obtain the three roots of

namely,

where

.=_|,.Vfr|,B=-|_V^|-

(12)

CUBIC
Example
1
.

AND BIQUADRATIC EQUATIONS

6x^

-ISS

Solve x^

+ 6x q

0.

By

818 we find that the substitution x

given equation to the form y^

y^

+ py +

roots of this equation, the above formulas, are

The

= 4- 2 will transform the = 0. We thus obtain 6 y 6 = 0. found by substituting!) = 6, g = 6 in

j/

v^4 + V2, Vi w + V2 0)2, Vi w2 Hence the roots -of the given equation are
2

+ V2 w.
3
3

+
2.

3^3 V4
+

V2, 2 +

V4 0, + V2 ^2,

+ V4 ai2 + V2 w.
6x

Example

Solve the equation x^

x"^

0.

When /> and q are real, the charDiscussion of the solution. acter of the roots depends on the value of y^/4 -\- p^ Jll as follows
:

872

1.

7/"

q^/4

+ p^/27 >

0,

one root
are real.

is

real, ttvo are

imaginary.

For

in this case

A and B

Hence

x^ is real,

and

x^,

X3 are conjugate imaginaries,

2.

870.

If q^/4

For

and
by

ccg

= 0, all the roots are real and two equal. = 2 V^/2, in this case A = B = q /2. Hence = = (w + or) V///2 = V^/2, since w + w^ = 1,
-f-

p'/27

a-j

3/

r-r

3/

a;,,

870.

3.

If (^ /^

p'/27

<

0, (7^^

the roots are real

and unequal.

This
l>ut

may be proved by Sturm's theorem (see p. 477, Ex. 16). when q"^ / ^ -\- p^ /27 < 0, A and B are complex numbers,

and though the expressions for a-j, X2, x^ denote real numbers, they cannot be reduced to a real form by algebraic transformations. This is therefore called the irreducible case of the
cubic (see

885).
is

The expression <7'^/4 +^>^/27 the cubic x^ +2^^ + ? = 0, since

called the discriminant of

873

its

vanishing

is

the condition

that two of the roots be equal (compare

635).

486
874

A COLLEGE ALGEBRA

The general biquadratic. Ferrari's solution. By the method of 818 every biquadratic equation can be reduced to the form
,

X*

-\-

ax^

+ bx + c =

0.

(1)

a view to transforming the first member of (1) into a difference of squares, we add and subtract xhi ?*^/4, where

With

u denotes a constant wliose value

obtain
X*
_

is

to be found.

We

thus

or

xhi u'^/4 + ax^ + bx c 0, - [{u -a)x^-bx+ {u" I ^ - c)] = 0. (x^ + u/2y +

x'^ti

4- ?/^/4

-\-

(2)

To make

the second term a perfect square,

we must have

or

M^

b' = ^{n-a){ii'J^ -c), au'^ 4 cu + (4 ac &^) = 0.

(3)

Let
is

i denote

one of the roots of this cubic

(2),

in u.

When m

replaced by Uy in

the second term

is

the square of

V?<i

aX

b /2 Vmi
x'

a,

and

(2)

becomes equivalent to the

two quadratics

+ V^^T^ ^ + f^' \
-^

, 2

,-

Vmi

a^ J

0.

(4)

x'^-V^^;:^ax+('-^

\^

+ jt==.\=0.
2 "Vi

aJ

(5)

We may
and
(5).
1.

therefore obtain the roots of (1) by solving (4)

'^

Example
Here a

Solve x*
6

1,

4, c

+ =
M^

x'^

+
u2

4x

.3

0.
Is

3,

so that the cubic (3)

12 w

x^

28

=
x

0.

One

of the roots of this cubic

x2

is 2,

and setting mi

and

= 2 in = 0.

(4), (5),

we have

Solving these quadratics,

we

obtain x

Example

2.

Solve the equation x*

= ( 1 V5)/2, (1 -t i 4 x'' + x^ + 4 x + 1 = 0.

vll)/2.

As

taken as the u^ in (4) and

the cubic (3) has three roots any one of which may be the method (5), it would seem that

CUBIC

AND BIQUADRATIC EQUATIONS

487

above described might yield 3-4 or 12 values of x, whereas the given equation (1) can have but four roots. But it is not
difficult to

the values of the four roots of

prove that the choice made of u^ does not affect (4), (5) combined, but merely

the manner in which these roots are distributed between (4)

and

(5).

Reciprocal equations.

We may discover by inspection whether

It follows

875

or not a given reciprocal equation, 815, has either of the roots 1 or 1, and if it has, derive from it a depressed equation

^ (Vr) = which has neither of these roots. 815 that this depressed equation <^ (a;) =
aocc^"'

from

must be
ao

of the

form

+ aia;2'''- +
'

a,a;"'

+ a^x +

0,

(1)

is, degree must be even, and every two coefficients which are equally removed from the beginning and end of ^{x) must be equal. We proceed to show that by the substitution z x -{-\ Jx

that

its

this equation ^ (j-) may be transformed into an equation in z whose degree is one half that of It will then <f)(x)= 0.

follow that

if

we may

find the roots

(1)

the degree of (j>(x)= be not greater than eight, by aid of 631, 871, 874.

Dividing

by

x"'

and combining terms, we have

X"

But by carrying out the indicated reckoning we

find that

y(-;)-(-+.-^-)^<>
and
if

=x+
x^

1/x, and

in (3)

we

set

1, 2,

succes-

sively,

we have

+ x^ = z' - 2, \

x'

+ \ = z'-3z, x^
(4)

cc*

+ ^, = z*-4.z' + 2,-..,

488
that

A COLLEGE ALGEBRA

cc^ l/x^ an expression of the ^th is, we obtain for Substituting these several expressions in (2), we degree in z. obtain an equation of the ?/ith degree in z, as was to be demon-

strated (compare

645).

Example L
This
is

Solve 2x8

-xT_

12 x^

14x5

_ HxS +

12x2

0,

a reciprocal equation having the roots 1 and 1, Removing the factor x^

1.

2x6
Dividing by
x^,

a;5

lOx*

13x3

10x2

0.

Hence by
or
Solving,

(4),

2 (z^

3z)

(22

2)

lOz

13

2z3-z2_i6z +
z

15

= =

0, 0.

= l, -3, or 5/2. x + l/x = l, -3, or 5/2, Hence and therefore X = (1 iV3)/ 2, (-3V5)/2, 2, or - x^ + x* + ^2 - x + 1 = 0. Example 2. Solve x"
876

1/2.

Every binomial equation

a;"

may

be reduced to the

x Vay, the result reciprocal form by aid of the substitution 1 of this substitution being y" (compare 646, Ex. 2).

+ =

877

amplitude.

Expression of a complex number in terms of absolute value and In the accompanying figure, P is the graph of the

complex number a
in 238.

-\-hi,

constructed

as

The length
sent
it

of

OP

is
hi,

Vo-^
239.

h\ the

ahsolute value of a -f

Repre-

by

r.

Let 6 denote the circular measure of

the -X the angle XOP, that is, the length of arc subtended by this angle on a circle of
unit radius described about
as center.

We call 6 the amplihide of a + hi. We call the ratio b/r the sine of

6,

and write b/r

= sin

$.

CUBIC

AND BIQUADRATIC EQUATIONS

0,

489

We call

the ratio a J r the cosine of

and write a/r

r sin 0,

= cos 6.

We

thus have

a
a
-\-

and therefore

bi

= r cos 0, r (cos
and a

b
-[-

i
r.

sin 6).

When
cosO

0,

then

Hence

sin

= 0,

and
878

1.

The

circular

measure of 360

is

tt,

this being the length

r,

of a circle of unit radius.

Hence a point P given by

;

is

given equally by
r,

r,

niTT,

where

^ + 2 tt by m denotes any

?,

tt

and, in general, by

integer.

Hence we say that

ii is ^

the general value of the amplitude of a

2 mir.

Theorem.

The

absolute value of the product of two comptlex

879

numbers
tude
is

the product of their absolute values; the sum of their amplitudes.

is

and

its

ampli-

For

r (cos 6

sin 6)
0'

r'

(cos

6'

sin

6')

=r

r'

[(cos 9 cos
i

sin 6 sin

6')
6')']

+ (sin 6 cos $' + cos d sin = r r'[cos (0 + 0') + i sin (6 + 0')^,

since
it is

proved in trigonometry that

cos ($
sin (6

+ +

$')
(9')

= cos 6 cos ^' = sin cos ^' +

in

sin $ sin

5', ^'.

cos 6 sin

The construction given

Corollary
1.

240

is

based on this theorem.

By

repeated applications of
r'

879 we have

880

r (cos e

sin 6)

(cos

6' ^'

sin

6')

r" (cos 6"

sin 6")

J, ^',^.'f

[cos(^

^"

+ ...) + sin (^ +
i

^'

Corollary 2.
in 880,

we
:

r' r" ^" , and Setting r obtain the following formula, known as Demoivre's

+ ^" + ^ = e' =

)].

881

theorem
^

[?'(cos 6

-\-

sin 0)]"

= r" (cos nO + i sin nd).

490
882
Corollary 3.

A COLLEGE ALGEBRA
For a quotient we have the formula
.
.

r (cos ^

-^ r
For

+ i sin ^) = r^ ,^ ^ ,^ , l / /j 7 sm ^) y' [cos {6 -e') + i sin (e - e')]; + ~ O')']- r' (cos 6' i sin 6') ^ [cos (6 $') + sin {6 = r(cos^ + isin^). 879
..
.
.

(COS ^

-\-

883

Corollary 4.
,

The n

ni\i

roots of a

7^ n If Vr (cos 6+1 sni a. r" cos 6)

I

complex number are given by d + 2kir\ 0_2hTr

.

(-

sm

?i

1).

when
For

A;

is

assigned the n values

0, 1, 2,
.
.

(w

r
\_

?"

V
\

cos

2k'jr
,

n
i(sin 6

\- 1

e sm + 2^-7^^~l"
11

yj
t

= r [cos (^ + 2 A-tt) +
884

+ 2 k-K)'] = r(cos ^ +

sin 6).

881, 878

The n wth roots of r(cos 6 Binomial equations. the w roots of the equation a;" r(cos 6 + i sin ^) in particular, the n roots of the equation a;" r real and therefore 6 is 0, are

+ i sin 6) are = 0. Hence, = 0, where r is

,

Vr(cos 2

kir fn

sin 2 kir/ii), k

0, 1,

?i

1.

are Thus, the roots of the equation x^ 1 = cosO + isinO, cos2 ;r/3 + i sin2 7r/3, cos4 7r/3

isin4 7r/3,

which may be proved equal to

1,

(-l + iV3)/2,

(-l-iV3)/2.

885

In Trigonometric solution of the irreducible case of the cubic. the irreducible case of the cubic x^ 0, 872, 3, the px q For expressions .4 and B, 871, are conjugate imaginaries.

+ =

since in this case

q'^/4:

+^^^/27

is

negative,

we have

-^-i-V-M'^-l-W-M?
Hence the expressions for ^4 and B in terms of absolute value and amplitude, 877, will be of the form

=r (cos 6 + i sin 6), B ~ r (cos 6 i sin 6),

(1)

CUBIC

AND BIQUADRATIC EQUATIONS

491

where
\ J.

A.

//

//

(2)

and

r
jp

and q are given we can find the value of Q from tliat of cos B by aid of a table of cosines. In the formulas for the roots of a;^ + ^x + g- = 0, 871, (12), substitute the expressions (1) for A and 5, and the The results when expressions for w and w'^ given in 884.
simplified are
jTi

When

o =2

i
/'

^ COS -

o
r

a-2

o 2

r^

cos

+
o

27r
>

ccj

r'

cos

+ 47r
o

And,
to

and

Q being

known by

(2),

these formulas enable us

cosines.

compute the values of the roots by aid of a table of

Example.

Here

(^

/^

x + l/3 = 0. + p^/27 = 1/108, so that we

Solve x^
(2)

have the irreducible case.

Substituting in the formulas

r-1/ V27,
Hence by
Xi

cos ^

=-

and simplifying, we find V3/2, and therefore d = 150. and cosines we obtain

aid of tables of logarithms

.

= -^ cos 50 =
V27
Xs

7422

x^

-^ cos 170 = V27

1.

1371

-^-^ cos 290

.3949.

V27

EXERCISE LXXX
Solve equations
1. 3.

10 by the methods of 871 and 874.

2.

x3-9x-28 = 0.
x3 x3
X* x*

x3

9x2

3X

== 0.

4.
6.

x''

+ 9x -8 = - 6 = 0.
7

0.

5.

7. 9.

+ 9x-l = 0. + 6 X + 1 = 0. + 12x-5 = 0.
3x2
x2

3x='-9x2 + 14x +
X* x*

0. 0.

8.

4 x3

x2

+1=
x

10.

8x3

+ 12x2-l]
3

0.

11.

Solve

3x6-2x6 + 6x*-2x + 6x2-2x +

0.

492
12. 13. 14.

A COLLEGE ALGEBRA
Solve 2x Solve 6 x"

- 9x7+18x6 - 30x5 + 32x4 - 30x^+18x2 - 9x 2 = 0. + - x6 + 2 x5 - 7 X* - 7 x3 + 2 x2 - X + 6 =r 0.

in z

Find the cubic

on which, by

875,

the solution of x^

-1=
t- c
=--

depends.
15.

Find the condition that

all

the roots of x^

3 ax^

3 6x

be

real.

16.

and

Write down the trigonometric expressions for the roots of x^ ^ 1=0,

of x^

+1=

0.

17.

Solve the following irreducible cubics.

(1)

x3

3 X

0.

(2)
is

x3

_ 6x -

0.

18.

In a sphere whose diameter

inscribed.
If

base

is

a right prism with a square the volume of the prism is 27, what is its altitude ?

V3

19.

The volume

entire superficial area

altitude.

of a certain right circular cylinder is 50 tt is 105 tt/ 2. Find the radius of its base

and
and

its
its

The altitude of a right circular cone is 6 and the radius of its base In this cone a right circular cylinder is inscribed whose volume is four ninths that of the cone. Find the altitude of the cylinder.
20.
is 4.

XXXI.

DETERMINANTS AND ELIMINATION

DEFINITION OF DETERMINANT

886

Inversions. Odd and even permutations. When considering the permutations of a set of objects, as letters or numbers,

we may
normal

fix

upon some particuLar order of the objects as the

order.

Any

many

inversions as

it

given permutation is then said to have as presents instances in which an object is

the permutation of its inversions

followed by one which in the normal order And precedes it. is called odd or even according as the number is odd or even (or 0).

Thus, if the objects in normal order are the numbers 1, 2, 8, 4, 5, the permutation 45^,12 lias the eight inversions 43, 41, 42, 53, 51, 52, 31, 32. Hence 45312 is an tven permutation.

DETERMINANTS AND ELIMINATION

493
887

Theorem. If two of the objects in a permutation are interchanged, the number of inversions is increased or diminished by

an odd number.
two adjacent objects are interchanged, the number of is increased or diminished by 1. Thus, compare ApqB (1) and AqpB (2), where A and B denote the groups of objects which precede and follow the interchanged objects p and q. Any inversions which may occur in A and B and any which may be due to the fact that A, p and q precede B are common to (1) and (2). Hence the sole difference between If joj' is (1) and (2), so far as inversions are concerned, is this an inversion, qp is not, and (2) has one less inversion than (1) but if p)(l is not an inversion, qj) is, and (2) has one more For
if

inversions

inversion than (1). But the interchange of any two objects may be brought about by an odd number of interchanges of adjacent objects. Thus, from ^ja^^y we may derive qahiJ by five interchanges of

adjacent

letters.

We

first

interchange

2^

with each following

letter in turn, obtaining successively apbq, abpq, abqp,

and we

then interchange q with each preceding letter, obtaining aqbp, There is one less step in the second part of the process qabp. than in the first part, because when it is begun q has already been shifted one place in the required direction. Had there been ^l letters between p and q, there would have been ft + 1
steps in the
(/x 4-

first

1)

/A,

or 2

part of the process and is always odd. /x 4- 1,

yu,

in the second,

and

Therefore, since each interchange of adjacent objects changes the number of inversions by 1 or 1, and the sum of an odd

number
theorem

of
is

numbers each of which

demonstrated.

is

1 or

1 is odd, our

It will

Thus, if in 21457368 (1) we interchange 4 and 6, we get 21657348 (2). be found that (1) has^?;e inversions and (2) eight, and 8 5 is odd.

Of the n\ permutations of n objects taken all at a time, For from any one of 763, half are odd and half are even.

888

494

A COLLEGE ALGEBRA

these permutations we can derive all the rest by repeated As thus obtained, the permutainterchanges of two objects. tions will be alternately odd and even, or vice versa, 887.
are odd

Therefore, since 7i is an even number, half of the permutations and half are even.
!

889

In what follows

we

shall

with subscripts, as a^, a.^, h^, b^, , and so on. Having chosen any set of such symbols in which all the letters and
subscripts are diiferent, arrange and then find the sum of the
letters

have to do with sets of

letters

them in some particular order number of inversions of the

and of inversions of the subscripts. If this sum is even, it will be even when the symbols are arranged in any other order if odd, odd. For when any two of the symbols are interchanged the inversions of both letters and subscripts
;

are changed by odd numbers, 887, and therefore their sum by an even number. In particular, the number of inversions of the subscripts when the letters are in normal order and that of the letters when the subscripts are in normal order are both odd or both
even.
Thus,
in a^hzCi the

number

of inversions of the subscripts

is

in c^a^hs the

sum
is

of the

number number of

of inversions of the letters

is two; two; in hzaCi the

inversions of the letters and that of the subscripts

four.

890

Definition of determinant.

We may

in general n-

numbers

in the

arrange any set of 2^, form of a square array, thus

:

3^,

! 02

DETERMINANTS AND ELIMINATION

495

With the elements of such an array form all the products that can be formed by taking as factors one element and but one from each row and each column of the array.
(letters)

In each product arrange the factors so that the row marks are in normal 07'der, and then cotmt the inversions of

the

column marks

(subscripts).

If their number be even (or

0),

give the product the plus sign; if odd, the minus sign. The algebraic sum of all these p)lus and minus products

is

called the detei'minant of the array, and is rejjresented by the array itself with bars written at either side of it.

Thus,

496
893
894
895

A COLLEGE ALGEBRA

The products above described, with their proper signs, are called the terms of the determinant.

To expand a determinant
length.

is

to write

out

its

terms at

The diagonal

of elements

aj, h^, Cg,

is

called the leading

diagonal, and the product of the determinant.

a-Jj^c^,

is

called the leading term

The leading term enclosed by bars, thus, \ai used as a symbol for the determinant itself.
896

b^ Cz

is
],

often

The number of the terms of a determinant of the nth order n and half of these terms have plus signs and half have minus signs.
is
!,

Por, keeping the letters in normal order, we may form n\ permutations of the n subscripts, 763, and there is one term of the determinant for each of these n\ permutations, 890.

Furthermore half of these n\ permutations are even and

half are odd,
Thus, for n

888.

we have
It

or 6 terms

for

4,

we have

or 24.

897

must be remembered that the letters and subscripts are mere marks of row and column order. Any other symbols which will serve this purpose may be substiOther notations.

tuted for them.

an

Thus, the elements of a determinant are often represented by a single letter with two subscripts, as 023, the
first

indicating the

row and the second the column.

Thp

symbol

023 is

read " a two three," and so on.

898

A rule for expanding a determinant of the third order. To obtain the three positive terms, start at each element of the first row in turn and, so far as possible, follow the direction
of the leading diagonal, thus
aib^c^, 2^3^i>
:

(^abiC^.

DETERMINANTS AND ELIMINATION

497

To obtain the three negative terms, proceed in a similar ^manner but follow the direction of the other diagonal, thus
:

3
1

Tims.

= 5(-l)(-l) + 3(-3)2 + 2(-l)4 -5(-3)4-3(-l)(-l)-2(-l)2=:40.

third.

This rule does not apply to determinants of an order higher than the Tims, it would give but eight of the twenty-four terms of a deter-

minant of the fourth order.

EXERCISE LXXXI
Expand
the following determinants.
r
1

V
s

P
q
r
s

<1

p
s

498

A COLLEGE ALGEBRA
PROPERTIES OF DETERMINANTS

899

Theorem 1. The value of a determinant is not changed if its rows are made coluvins, and its columns rows, without changing
their relative order.
ai ^2 On

Thus,

900

901

902

DETERMINANTS AND ELIMINATION

For any term of
(1)

499

with factors arranged in the order of

the rows of (1) may be transformed into a term of (2) with factors arranged in the order of the rows of (2) by inter-

changing

its first

and

last factors

and conversely.

But

this

interchange will increase or diminish the inversions of the subscripts in the term by an odd number, 887, and therefore,
since the
(2), it will

normal order of the subscripts change the sign of the term.

is

is

123 in both

(1)

and

Thus, UibsCi
of
(2).

a term of

(1)

and

Cib^ao is the

corresponding term

Cibsa-z

subscripts present two inversions, while in they present one inversion.

in

For

UobsCi the

Example.
determinants
Corollary.

Verify the preceding theorem by expanding each of the

(1)

and

(2).

If two of the rows (or columns) of a determinant are identical, the determinant vanishes.

903

For
902,

let

denote the value of the determinant.

An
;

inter-

change of the identical rows must leave

it

will change

into

is,

unchanged

but,

by

D.

Therefore

D^- D, that
b
c e

i)

0,

or Z)

0.

a a d
Thus,
b
c

abf

aec

dbc

aec

abf

dbc

0.

f
If
the
all

Theorem
multiplied plied by k.

4.
bt/

the elements of a
k, the

row

(or column')
is

are

904

same number, as

determinant

multi'

For of the elements thus multiplied "by k one and but one occurs as a factor in each term of the determinant, 890.

The evaluation

of a determinant

may

often be simplified by

applying this theorem.

Thus,

6 8 15-20
3

2
5

-3
= 2-5

1 1

1 1

4-1

3-4 1 = 2-5-3.4. 3 4-1

480.

-1

500
905

A COLLEGE ALGEBRA

Corollary. If the corresponding elements of two columns (or rows^ are pi'oportional, the determinant vanishes.

ra a d

Thus,

906

907

DETERMINANTS AND ELIMINATION

4
7

501

Thus,

5-4
-2
5
7.

since

4,7,7

2(5,

-4,

2)

+ 3(-

2,

5,1).

Theorem

integral functions of some variable, as x, vanishes a,. the determinant is divisible bi/ x

If a determinant whose elements are rational when x = a,

For the determinant when expanded may be reduced to the form of a polynomial in x. And since this polynomial vanishes when X = a it is divisible by a; a, 415. The factors of a determinant may often be found by aid of
this theorem.
1

Example.

Show

that

502
1

A COLLEGE ALGEBRA

4.

Prove that

5.

Prove that

DETERMINANTS AND ELIMINATION

503
911

Corollary. If e denotes the element in the ith row and Vth column of A, the sum of all the terms of A which involve e is

(-

+
l)'
''eA,.

For we can bring e to the position of leading element without disturbing the relative positions of the elements which lie outside of the row and column in which e stands, namely, by
first interchanging the row in which e stands with each preceding row in turn and then interchanging the column in which e stands with each preceding column. In carrying out these successive interchanges of rows and columns we merely

change the sign of the determinant

times,

902.
shall

Hence,

if

2 (i 1) or i -\-k 1) (^ A' denote the determinant in its final

form,

we

have
A'

= (this

+ ^1)'

A = (- ly + ^-A.
which involve e is For the minor of

By
eA'g.

910, the
in

sum

Hence

of all the terms in A' + sum is

l)'

^eA^.

e in

is

the same as

its

minor
ai 62 cs

in A'.
(Z4

Thus,
i

in the case of

A=
|

the element
1

(Z3,

for
:

which

= 4,

3,

may

be brought to the leading position as follows

a\ a^ as at

dz di d^ di as ai a^ a^

h h
C3

62 C2

h
Ci
(3)

Ci

By
first

interchanging the fourth row of (1) with the third, second, and in turn, we obtain (2) before which we place the minus sign because
is, i

of the three, that

1,

interchanges of rows.
of (2) with the second

Then by interchanging the third column first in turn, we obtain (3) before which we
before
(2)

and

place the

same

sign as that

because of the two, that

is, fc

1,

of ds in (1) is the same as its of all the terms of (1) which involve d^ is

The minor

minor
ds

interchanges of columns. in (3). Hence the sum

|ai 60 Ci\.

Theoreifi.

determinant

may

he expressed as the
its

products of the elements of one of

sum of the rows or columns by their

913

504

A COLLEGE ALGEBRA

comxAevientary minors, with signs which are alternately plus

and minus,

or

minus and plus.

Thus, in the case of a determinant of the fourth order = ! ^2 t'g di we have

I I

2A^

az^^

a^^^

For each term in the expansion of A contains one and but

one of the elements

sum

of all the terms

is

which involve

a^, a 2> as, a.. And, by 910, 911, the "Sj which involve a^ is aiA the sum of all cloA. and so on. t2tA2,
,

In like manner,

A=-

b,\^

b^\^

hz\^
<'Ac^

= i^oi 913
Cofactors.
It
is

h\ +

+ h\^
diAi^,

and so

on.

sometimes more convenient to write the preceding expressions for A in the form

A=

i.4i

^2-4 2

-i-

asAs

a^A^

and so
call
-^"1

on,
-^21

where A^

A^, yl2

Aj,

and so

on.

We

then

1,

the cofactors of

a^, a^,

Thus, in
ai

914

DETERMINANTS AND ELIMINATION

Bordering a determinant.

605
expressed

Any determinant may be

For, by

as a determinant of a Iiigher order.

912,

we have

tt

tZ 2

OE' c

bi

bz

ba

506
is !, 0, 0,

A COLLEGE ALGEBRA

. Hence this determinant is equal to a^ times its minor, which is the determinant of the (n l)th order (3). Observe that each element of (3) is obtained from the cor-

responding element of the minor of a^ in (1) by multiplying

that element by a^ and from the result subtracting the product of the corresponding elements in the first row and the first

column of

(1).

Thus, by two reductions of the kind just described,

2

we have

2-1

13-2
2
1
1

2-1

3-2-2

DETERMINANTS AND ELIMINATION

But

507

(1)

and

905.

And

(4) vanish, since their cohimns are proportional, simplifying (2) and (3) by aid of 902, 904, and

adding them, we have

^l?2

508
3

A COLLEGE ALGEBRA
2 6 4
5

6
3.

9
12

DETEIIMINANTS AND ELIMINATION

509

! ^2 3

510
If

A COLLEGE ALGEBRA
A and
all

the determinants \k

h^

C3\,

\ai

Cg],

\ai h^

m\

vanish, the equations (2), (3), (4) impose no restriction on the values of iCi, x^, x^. In this case the given equations (1) are not

independent.

which

(2),

(3),

This follows, by 394, from the manner in (4) were derived from (1), unless all the

minors A^, A,

may

And if all the minors vanish, it vanish. readily be shown that the three equations (1) differ only by constant factors, so that every solution of one of them is a solution of the other two.
These results are readily generalized for a system of n equa71

tions in

unknown

letters.

921

Homogeneous
equations in

linear

equations.

When

= =
I

7n

0,

the

equations (1) of

x^, x^,

919 reduce to a system of homogeneous x^, namely.

ajxi

+ a^x^ + ^3X3 =
I

(1)
^1^1
^22*2

~r ^3'^3 ^^

'

and the equations

(2), (3), (4) of

919 become
Ax^

Aa-i

0,

A:r2

0,

0.

(2)

Evidently the equations (1) have the solution Xi

= X2 = Xs = 0,

and

it
0.

follows from (2) that this

if

is

the only solution unless

A=

But

A=

0,

the equations (1) are satisfied by

Xi

rAi, Xo

rAo, x^

rA^,

(3)

where r

may

denote any constant whatsoever.

For, substituting these values in (1) and simplifying,

we have

UiAi

a^A^

a3'43

0,

b^A^

CiAi

C2.42

C3A3

+ b^Az = 0,

+ b^As =

0,

and these are true other two by 914.

If

identities, the first

one because

A=

0,

the
:

The same thing may be proved

as follows

we solve the second and third of the equations (1) for Xi and X2 in terms of x^, we obtain Xi/ Ai = X2/A = X3/ A^, or, if

DETERMINANTS AND ELIMINATION

r denote the value of these equal ratios, x^
a-3

511

rA^.

And

as just

shown,

if

A=

0,

vA^, rvlj, a-g these values will also

satisfy the first of the equations (1). From this second proof it follows that
a-i
:

when A

=
:

.Tj

X3

= .li

J2

-^ 3

^1

-62

i?3

Ci

Cj

C'3,

that

is,

the minors

of corresponding elements in the

rows of
922

It is assumed that these minors are not 0. are proportional. From the system of three non-homogeneous equations in x, y

+ ^s = a^x + = byx + h^y + C^X + C2?/ + C3 =

(1,21/

/>3

(1')

we may
Hence

derive the homogeneous system (1) of

921 by suba

stituting X

= a-i/cTg, y = x^/x^ and clearing of fractions. A = is the condition that the equations (1') have
EXERCISE LXXXIV

common

solution.

Solve the following systems of equations by determinants.

2x + 3?/-52 =
1.

3,

x-2?/ + z = 0, ^ Ux + y + 3z = 7.
(

2x + 4y-3z 3x - 8y + 6z
!

= = =
+ + + -

3,
1,

4.

az

-\-

by

cz

d,
4.

3.

a?x

IP-y

(f-z

d^.

f2x- 4y + 3z 3x-2?/ + 62 5x + 8?/ + nz - 32 .X - 10

2/

= -3, = -l, 3 = 9, = 2. 7
4 5

Show
them

that the following systems of equations are consistent, and solve for the ratios x-.y-.z.
x

+ 2y-2 =

0,

3x-t/ + 4z =
,4x-|-?/
7
.

0,
0.

<!

ttoX
.

+ 3z =

asx

+ +

hoy
&32/

-f (A-a2

(A-as

+ +

^^2) 2

^^s) z

= =

0,

0.

For what values

of X are the following equations consistent ?

4X

2/

Xx,

Zx - Ay +
X
-\-

1 z

1 y

Qz

= =

\y,
\z.

512

A COLLEGE ALGEBRA
RESULTANTS

933

By the resultant of two equations /(x) = and meant that integral function of the coefficients of /(x) and ^ (x) whose vanishing is the necessary and sufficient
Resultants.
(^

(x)

is

condition that/(x)

and 0(x)

have a common

root.

Thus, the resultant of aoXao6^

aiX
aih

+
-|-

ag a2

!&

a2

for

when
6.

a^ffi -f

= = 0,

(1)

and

.x

=
(1)

(2)

is

the equations

and

(2)

have the

common

root

924

any two equations f(x) = 0, <^ (x) = may be obtained by eliminating x by the following method due to

The resultant

of

Sylvester.

To

fix

the ideas,

let
ttox'^

f(x)
cf,

= = (x)

-r aix^

b,x^

+
1,

b,x

+ a^x + + ^2 = 0.
(2)

Og

0,

(1)

(2)
x,

Multiply (1) by x and

and

by

x"^,

and
0, 0,

1 successively.

We

obtain

bfp:*

+ bix^ + b^x^ 0, b^x^ + bix'^ + b^x 0, = 0b(fX^ + ^ix + &2

five

These may be regarded as a system of

linear equations in the five quantities x*, x^, x"^, x, 1. 921, they cannot have a common solution unless
^0 ^1
rt ^'2
fli
^'2

homogeneous Hence,

'^3

*^

^3
0.

Z>

/>0

/^l

(3)

^'i,

^>i

b.
b^
bo_
,

b^

Hence

common
fifth

(3) is the necessary condition that (1) and (2) have a For to the root. It is also the sufficient condition.

column of

add the

first

four columns multiplied by

x*,

DETERMINANTS AND ELIMINATION

x^, x^,

513

determinant,
/(ar),

X respectively. We thus transform D into an equivalent 907, whose fifth column has the elements a-/(.T),
(x),
X(f>

x^^

(x),

cji

(x).

Hence,

if

/xi, fx^, fj-si p-i, /^s

denote the

cofactors of the elements of the fifth

S

column of D, we have,

Q1

"^
'

D = {fi^x +
from

fio)f(x)

(fi.sx'^

4-

fJ^iX

/xs) <^

(x)

It follows

this identity that each factor

(fj-sx'^

= 0. /3 oi f(x)

must be a

factor of

fj-iX

+ /aj) ^(a^),

and therefore, since

/(x) is of the third degree and /xgo;^ + fiiX + fi^ is of only the second degree, that at least one factor x (3 oif(x) must be
a factor of
^(j^"),

in

f(x)

must be a root of (/)(x)= 0, It is here assumed that the minors

other words, that one of the roots of 795.

fx-i,

fx^r'-fH-i ^^^

^ot

all zero.

If the minors of all the elements of

are 0,

it

can be proved
root.

that/(.T) If Xi, Ao,

and

<^(.r)

have more than one common

921 that
^^ Ai
'*

As denote the cofactors of the elements of

any

925

row

of D,

it

follows from

when

D=
I

',

',

X X

Ao

A3

A4

A5,

whence x = Ai/Ao and f(x) = (cr)

<f)

= =

A2/A3

A4/A5.

Therefore

when

have a common

root, the value of this

root

is

and when the degrees of f(x) = and n respectively, the resultant D will be a determinant of the (in + n^ih order whose first n rows consist
In the general case
(li(x)

Ai/Ao.
are

926

of the coefficients of f{x) and zeros and rows consist of the coefficients of <i>{x)

whose remaining in and zeros, arranged

as ii} 924, (3). Hence in the terms of D the coefficients of f(x) enter in the degree of ^ (x) and vice versa. Example. By the method just explained show that the equations and x + 1 = have a common root and find this root, x'^ + 3z + 2 =

.13 2

Here

D = 110 =1 + 2 3 = 0,
1

so that there

is

common
of

root.

The values of the cofactors of 1 and 3 in the first row and 1. Hence the common root is 1 1, that is, 1.
:

are 1

514
927
X,

A COLLEGE ALGEBRA

By the preceding method either of the unknown letters y may be eliminated from a pair of algebraic equations of
cf>

the form f{x, i/)=Q,

Example.
Solve

(x,

y)

= 0.
j/^

x2

x
?/

2 x2

?/2

= =

0,
0.

(1)

(2)

We may
1,

regard

(1)
(2)

and

1,

2y",

and

(2) as quadratics with the coefficients

in x, (1)
2, 0,

with the coeflBcients

7/2

Zy.

Hence the

result of eliminating
1

is

DETERMINANTS AND ELIMINATION

Let
a^
0.

515

= <f}(x)
let

and ^S^. any root of denote any root of f(x) = There will be 77in differences of the form a^ fi,^;
(3/,)

n (a^

denote their product.

= Evidently IT (a^ y8^.) dition that one of the roots

Moreover, since
11 (a^

is

a^
is

(3/^)

the necessary and sufficient conbe equal to one of the roots (3^-

the roots a- and of the roots

yS^.,

a symmetric integral function of it is a rational integral func-

tion of the coefficients of f(x)

and

<^(a;)

= 0,
of

867, 868.

Hence,
</,

if

B (/,
923,

</>)

denote the resultant

f(x)

and

(x)

= 0,

we have

R (f,
The product
11 (a-

<f>)

= U (, -

/?,).

^^)

may

be written

(i

- A) ('^1 - A)
-

(i

(^n),

(2-/8i)(2-y32)---(a2-/8),
/5l)

{<^>:,n

('l-,,.

/32)

("-,

- ^).
(a;

But since </> (x) (x of the factors in the first

and so
on.

/3i) (x
row
is

^82)

(3),

the product

^(^i), in the second

row

<^(a2).

Hence

Again, since f(x)

= (x

n-j)

(x

a,^)
is

-{x

of the factors in the first

column

column (

l)"'/(i32),

and so on.

( l)'"/(/8i), Hence

a), the product in the second

n {a, - A) = (When
f{x)
that

i)"'"/(A) -/(A)

-/(A,).

the given equations have the form

oX"'

...

+ , = 0,

<t>{x)

box-

^...

= 0,

the leading coefficients are not 1, the product of is, the factors in the first row is (j>(ai)/l)Q, and so on; and the product of the factors in the first column is ( l)"'/(/?i)/a(

when

and so

on.

Hence, in this case, to

make

11 (a^

/3^)

an integral

516

A COLLEGE ALGEBRA

function of the coefficients of /(^)

and

<^(a;)

we must

multiply

it

by

ao^'u".

We

then have

929

In the resultant of a pair of equations

the coejficients of f (x)
vice versa.

(x)=
(}>

0, <^ (x)

=
and

enter in the degree of

(x)

0,

to the first degree. involving the coefficients of /(a;) = thus have another proof that the determinant D described

For the product ^ (ai) (^ {a^) <^ (a-,) contains factors, to the first degree each involving the coefficients of <^ {x) = and the product /(/3i) -fifi^) fW} contains n factors, each

We

in 924,
is,

926

is

that

D^ R(f,

the resultant of f(x)

<f>).

and

</>(a;)

0,

that

930
<j>

The sum of
(x)

the subscripts of the coefficients

(f,
<f>)

ofi(x)=

and

in each term of E.

is

mn.

4>(x)

For, by 812, if we multiply each coefficient of f(x) and by the power of r indicated by its subscript, we obtain

two equations
/i (x)
cf)i

= a^x"" + raiX"'- + r^aox'"= boX" + rbyX"-^ + r%x"-^ (x)

^

^
-|

\-

r'"a
r^b^

-\

f-

= 0, = 0,

whose roots are r times the roots of f(x) = and (x) = 0. Each term of R (f, 4>i) will be equal to the corresponding term of R (f (j>) multij^lied by a power of r whose exponent is the sum of the subscripts of the coefficients off(x) and (f>(x) which occur in the term. Hence our theorem is demonstrated But if we can show that in every term this exponent is mn. since there are m?i factors in the product EI (a, (3/^), we have
(f>

R (f,

4>,)

alb'l

n (/-a, -

7'/3,)

= r- R (/,

<^).

detp:rminants and elimination

Discriminants.
is

517
\-a^z=0
931

The discriminant oif(x)

= a^pf^

-\

that integral function of the coefficients of /(x) whose vanishhave ing is the necessary and sufficient condition that/(a:')

a multiple root (compare 635, 873). If D denote the discriminant of /()=

0,

then

D = R{f,f')/a,. has a finite multiple root when and For, by 851, f(x) and /' (x) have a finite root in common. only when f(x) By 928, the condition that/(a') = and /' (x) = have a root
in

common

is

R(^f,f')=

0.

But

a^

is

a factor of
in the
a^,

R (/,/'),

as

may
of

be shown by expressing
924.

Hence R

(/, /')

R (f, f)
when
to /(a-)

determinant form
0.

But the root

which
816.

in this case is

common
co is

and/'(x)

is oo,

Therefore, since

not a multiple root of /(a-)

unless both a^ and a^ vanish,

we have D
ao

-f^

(/>/') /^o-

Thus, for aox2

UiX

0,

D=

518
same degree
''1
\

A COLLEGE ALGEBRA
in Wj, x^^ x^.
I

Thus, x^
"

+ xy + y

-{-

becomes

^l'''^"2-'^3

^2-^3

3^3

in

Observe also that a homogeneous equation of the th degree determines n finite values a-2, x^ which is not divisible by x^
Thus, from x|

x^/x^

of the ratio x^ fx^. 1 or 2,

3 x^x^ + 2
(x,

a;|

we obtain

934

Let

f{x, y)

and

<^

y)=0

and 7i respectively. denote two equations whose degrees are If they involve the terms x'" and x'\ then by substituting
x

Xi/xg,

X2/xs,

terms,

we can reduce them

F{x ^(x
x^, Xs) x.

clearing of fractions, and to the form

collecting

= = Xs)

aoXT
boxl

+ +

iaT

-'

b.x'r'

where each of the

coefficients Uq, !,

+ ,
=
(3

+ +

0,

(1)

bn

=0,

(2)

bo, b^,

denotes a

homogeneous fimction of Xo, x^ of the degree indicated by its subscript. Hence R, the resultant of (1) and (2) with respect to
Xi, is

By

a homogeneous function of x^, Xg of the degree mn, 930. 928 the necessary and sufficient condition that (1) and

(2) be satisfied

by the same value of

Xi is that
(3)
i?

R=
If

0.

not divisible by Xs, then finite values of Xc,/x2 or y, 933.

is

is satisfied

by

mn

If

these values, the equations f(x, j8) 0, <^ (x, common root, and if this root be a, then x

denote any one of ^)= have a

a,

fi

is

solution of /(, y) 0, <l>{x, y) (compare 927). Moreover it can be shown that to each simple root of 72 there thus

corresponds a single solution of f(x, y) = that to a multiple root of order r oi R

r solutions of f(x, ?/) 0, <f> (x, y) of them equal. Hence f(x, y) 0, solutions.

0,

<f>

=
y)=

(x, y)

and

there correspond

=
(f>

0, all different or

some
finite

(x,

have 7nn

If

R
fi

is

mn

divisible by

xJ^,

then

R=
y,

finite values of

x^/x^ or

is satisfied by only and therefore f(x, y)= 0,

DETERMINANTS AND ELIMINATION

<^ (^x,

519

have only itui ^ finite solutions. But since and y = x^/xs, when x^ = either x ox y or both Xi/X'i X and y are infinite. We therefore say in this case that = have /x infinite solutions. f(x, y')= 0, (x, y)
y)

<f)

by a substitution of the form y = y' -\- ex, into equations of the same degrees which have these terms. By what has just been proved the transformed equations in x, y' will have mn solutions. But if x = a, y' = ^ be any one of these solutions, then x = a, y = ft -\- ca is a, solution ofJ\x, y) also have
7iin

If the given equations f(x, y) 0, and x" terms, we can transform them,

<^

{x,

}/)

lack the

ic"*

0,(f> (x,

y)

= 0.

Hence /(a-,

y)

0,<f> {x,

y)

solutions.

In the preceding discussion it is assumed that R does not vanish identically. If R does thus vanish, f(x, y) and <^(cc, y) have a common factor and therefore f(x., y'^ = 0, (f> (x, y) =

have

infinitely

many

solutions.
:

We
tively
(fi

therefore have the theorem

a7id n resjiecIf i(x, y) and <^(x, y) are of the degrees and have no common factor, the equations f (x, y) = 0, (x, y)0 have mn solutions.

EXERCISE LXXXV
1.

By the method of

924, 925

show

and 2x3
2. 3.

x2

9x

have a

common
aiX

that the equations 6 x2+5 x-6=0 root and find this root.
a2

Form

the resultant of aoX-

+
ex

=
d

and

^oX^

+ =

biX
1.

60

0.

Find the resultant of ax^

bx'^

and

x"

4.

By

the

method
q

of 931 find the discriminants of the equations

(1) x^ 5.

+ px +

0.

(2)

ax'

By

aid of 931
this root.

show that

x^

x2

8x

+ 6x2 + c = 0. - 12 -- has a

double root

and find
6.

Solve the following pair of equations by the method of 927.

x2

3 x?/

2/2

16 X

x2

xy

2 y2

5X

- 28 = 5y =
?/

0,
0.

520

A COLLEGE ALGEBRA

XXXII.

CONVERGENCE OF INFINITE SERIES

DEFINITION OF CONVERGENCE

935

Infinite series.

If

u^, u^,

??,

denotes any given never-

ending sequence of numbers,

-"i

187, the expression

h H

''2

is

called an infinite series (compare

704).

For

?/i

?/o

we may write

2?/,

read "

sum

of

ti

to

infinity."

The

series 2?< is called real

real, positive

when

all its

when all its terms Vi, Vo, are terms are positive. In what follows

we

shall confine ourselves to real series.

series is often given
if
11

A
Thus,

- v7i/(ft +

by means
the series

of
is

1),

aformula for its nth term Vl/2 + V2/3 + V3/4

-\
.

m.

Sometimes such a formula is indicated by writing the first three or four terms of the series. Thus, in 1/2 + 1 3/2 4 + 1 3 5/2 4 6 4 we - l)/2 4 have M = 1 3 2 ji. (2 n

936
71

Convergence and divergence.

Let
---,

.S'

denote the
S-^

sum
S

of the first

terms of the series u^

in general
71
.S'

and
Ui

= v^
S,^

+ Vo + + Vo

so that
-{-

= u^,

u^

+ v^,

-\-

As

increases,

u^, ?<i

+
.S'

no

"s,

will take successively the values i, , and one of the following cases must

present

itself,

namely

will approach

some

finite

number

as limit,

or
or

S^ will approach infinity,

S,^

will be indeterminate.

In the
gen,t,

first

and lim

case the series Ui -f 2 is said to be converis called its suth. In the second and third .s;^

cases the series

is

said to be divergent.

When Ml +
lim
5,

Mj

is

convergent

by S and write S

= Ui + Ui +

we may

represent , that is,

its

sum,

we may

CONVERGENCE OF INFINITE SERIES

521

regard the series as merely another expression for the definite

number

S.

is convergent Thus, the geometric series 1/2 + l/i + 1/8 + 1/16 + and its sum is 1. For here, as n increases S takes successively the values

1/2, 3/4, 7/8, 15/16,

limit.

and, as is proved in 704, it approaches Observe that here, as in every convergent series, lim w = 0.

as

The
values

series 1
1, 2, 3,

+1+1+

is

divergent.

For
oo.

jS

takes successively the

and therefore approaches

1

The

series 1

l+
-.

the values

1, 0, 1, 0,

It is therefore

---is divergent. For S takes successively indeterminate.

:

We
An

therefore have the following definitions

infinite series is said to be convergent

937
the

its first

n terms approaches a
Otherwise
it is

finite limit
to

when as n is

sum of

indefinitely

increased.

said

he divergent.

The

limit of the

series is called the

sum of the first n terms of a sum of the series.

Hitherto

convergent

This

is

new

use of the word smn.

sum has meant

the

result of a finite

means

of additions performed consecutively; here it the li7nit of such a result. It must therefore not be assumed that

number

the characteristic properties of finite sums, namely, conformity to the commutative and associative laws, always belong to these infinite sums
(see 941, 961).

gent, a finite

In determining whether a given series is convergent or divernumber of its terms may be neglected.
of the neglected

938

For the sum

If
III

terms

will

have a

definite finite value.

+ 2 +
any

(1) is

a convergent series having the

cui

sum

S,

939

and
is

c is

finite

number, then

+ c2 +
if

(2) is a con-

vergent series having the sum cS.

(2).

But

(1) is divergent, so

For if the sum of the first n terms of n terms of (2) is cS and lim CiSi = c lim
;

(1)
/S

be S, the

sum

of the first

cS.

The sum of a convergent series will not be changed if its terms are combined in groups without changing their order.

940

622
Thus,
if

A COLLEGE ALGEBRA
,

and gi, g^,--- denote the sums the given series heui + u^-] two terms, its next four terms, and so on, the series gi + g2 will have the same sum as Wi + ifo + For if M denote the last term in the group g,,,, we have
of
its first
-\

Pi

f/2

\-

Om

Ui

U2

Mn,

and the two members

therefore
yi

is

of this equation approach the indefinitely increased.

it

same

limit as in

and

In the same manner

may

be

shown

that a divergent positive series

remains divergent when

its

terms are grouped.

941

We may therefore introduce parentheses at will in a conIt is also allowable to remove them unless, as vergent series. in the following example, the resulting series is divergent.
The convergent
(11^

series 1/2

+
.

1/4
.

+
1^

1/8

936,
it

may

be written

1)

-I-

(1|

1)

(11

1)

-I-

..

But here

is

not allowable to

remove parentheses since

IJ

1+

li

1+

is

divergent.

942

It

is

sometimes possible to find the sum of a series by the

removal of parentheses.
Thus, the

sum
Sn

of 1/1

+
-

1/2

1/3

+
1)

is 1.

For

+ - + 2-3 1-2
-

+
71(71

1223
Hence
Example.

71

71

S=

lim S

lim (1
\
71

=L
is 1

Find the sum

of the series

whose nth term m

/n (n +

2).

943

Remainder

after n terms.

If the series i

''2

(1)

is

convergent, that portion of the series which follows the nth.

?/ + term, namely, ?( + Let R denote the 938.
i
.>

(2), will also

be convergent,
is

sum
(1).

of (2).

It

called

the

remainder after n terms of

Evidently lim
Ji,^

0.

CONVERGENCE OF INFINITE SERIES

POSITIVE SERIES
Theorem
1.

523

A positive

series Uy

<2 -\-

is

convergent

if,

as

944

increases, S

remains always

less

than

som.e finite

number

c.

For since the series is positive, S^ continually increases as n increases. But it remains less than c. Hence, 192, it
approaches a limit.
Therefore,

937, the series

is

convergent.

Theorem

2.
-|-

Let u^
^2

and
1.

let

a^

convergent.

The

(1) denote a given positive series, (2) denote a positive series known to be series^l) is convergent in any of the cases:

+ U2 +

945

When

each term of (1)

is

less

than the corresponding term

of

(2).
2.

When

Herm of
3.

(2) is less When in (1)

is

the ratio of each term of (1) to the corresponding than soine finite nnmbei* c.
tlie

2)receding term
1.
?'i

less

ratio of each term to the immediately than the corresponding ratio hi (2).

For

let
J

i'

denote the sum

let

of

the

first

n terms of
-I

+ "2 H If Ux < i,
^1 4- W2

a.nd.

U2

< 2.
is

A denote the sum of the series ^i + 2 we shall always have < A. Hence

S'

convergent,

944.

2.

For

if

< <
c,

c,

then Ui

<

ca^, u^

< cag?

a\

a^

Therefore, since cai

Ux

+ ra, H
1.

is

convergent,

939, the series

+
6.

W2

is

convergent, by

ror

II

< < <

)
)

Ux

ax
^l^

W2

*2

f^h

^3
lis

Uo

then
It follows
u^/a^, Us/as, fore t<i W2

Wo ^4 Us <,<;<,.
a^ ax
as Og a^ as

from these inequalities that each of the ratios Thereis less than the finite number Ux/ax.

is

convergent, by

2.

524
It follows

A COLLEGE ALGEBRA
from 938 that the same conclusions can be any one of the relations 1, 2, 3 holds good for all but number of the terms of the series (1) and (2).
Prove that
1

drawn
a finite

if

Example.
is

+ +

1/2

1/2

+ +

1/2

(1)

convergent by comparing
1

it w^itli

the convergent geometric series

1/2

1/2

1/2

(2)

the second is less than the corresponding convergent, by 1, Second. The ratios of the terms of ( 1 ) to the corresponding terms of (2) are finite. Hence (1) is convergent, by 2. namely, 1, 1, 2/3, 2 2/3 4,
(1) after
is
,

methods First. Each term of term of (2). Hence (1)

of the

by each

1, 2, 3.

Third.

The

ratios of the terms of (1) to the immediately preceding

terms, namely, 1/2, 1/3, 1/4,

in (2), namely, 1/2, 1/2, 1/2,

are less than the corresponding ratios

-.

Hence

(1) is

convergent, by

3.

946

Theorem

3.

Let Ui
b,

+ Uo

-|

(1) denote

a given positive

series,
to

and

let bj

(2) denote a positive series

known

be

divergent.

The

series (1) is divergent in

any of

the cases:

1. When each term of (1) is greater than the corresponding term of (2). 2. Wlien the ratio of each term of (1) to the corresponding term of (2) is greater than some positive numher c.
3.

When

preceding term

in (1) the ratio of each term to the immediately is greater than the correspo7iding ratio in (2).

The proof

of this theorem,

which

is

similar to that given in

947

945, is left to the student. Test series. The practical usefulness of the preceding tests, 945, 946, evidently depends on our possessing test series

known

to be convergent or divergent.
is

The most important

the geometric series a ar -\- ar^ -\which has been shown, 704, to be convergent when r

of these test series

-,

<

1,

and which

is

serviceable test series

obviously divergent when r > is the following.

1

1.

Another very

948

The when p

series

+ 1/2p +

1/3p

-\

\-

> 1,

1/nP H

is

convergent

divergent

when p <

1.

CONVERGENCE OF INFINITE SERIES

1.

525
1/2'',

p>

1.

Combining the two terms beginning with

the four terms beginning with 1/4^, the eight terms beginning with 1/8^, and so on, we obtain the equivalent series, 940,
i

+ U7; + -^ +

ii;

i^

+ .7^+^

(1)

Evidently each term of (1) after the corresponding term of the series

first is less

than the

that

is,

less

than the corresponding term of

+
is 1-

|+^+
p>
1,

---'

^' l

2^ri

But
2.

since

and therefore 1/2^"^

^T^, < 1, the

---.

(3)

geometric

series (3)
2^

convergent.

Hence (1) is convergent, 945, 1. Combining the two terms ending with 1/4, the

four terms ending with 1/8, the eight terms ending with 1/16, and so on, we obtain

^Ml-d<l-l-'^^l>
Evidently each term of (4) after the second
the corresponding term of the series
is

(4)

greater than

that

is,

greater than the corresponding term of

^

+ 2 + 4 + 8 + "-'^-^ + 2 + 2 + 2 +
divergent.

12

111
is

---

^^^

But
3.
is

(6) is

Therefore (4)

divergent,

946,
-|

1.

p <1.

In this case the series 1

its

1/2^

1/3^'

divergent since

terms of the series 1

terms are greater than the corresponding + 1/2 + 1/3 + , which has just been

proved to be divergent,

946,

1.

526
949

A COLLEGE ALGEBRA
The following exam-

Applications of the preceding theorems.

ples will serve to illustrate the usefulness of the theorems of 945, 946.

Example
It is

1.

Show

that 1/1

1/2

1/3

4 H

is

convergent.

convergent because its terms after the first are corresponding terms of the convergent series 1/2^ + I/32

less

than the

I/42

945,1.
is divergent. Example 2. Show that 1 + 1/3 + 1/5 + 1/7 + The ratios of the terms of this series to the corresponding terms

of the

divergent series 1 1), are 71/(2 n

divergent, 946,

+
all
2.

1/2

1/3

1/4

namely,
1

1,

2/3, 3/5, 4/7,

greater than 1/2.

Hence

1/3

1/5

is

Example
or
is it

3.

Is the series in
?

which m
^
1

{2n
1

+
2

l)/{n^

+
_

n) convergent

divergent

Here

^
11,,-^'"'^^= n^ + n
n^

+ l/n ^ + l/n"

Hence the ratio of u to 1/n^ is (2 an expression which is finite for all values of n, and which approaches the finite limit 2 as n increases. But l/n^ is the 7ith term of the convergent series
1

+ 1/n n^ 1 + 1/n^' + l/ii)/(l + l/n'^),

1/2-

1/3- H

Therefore the given series

is

convergent,

945, 2.

950

method employed in Ex. 3, it may be proved that if u has the form ?/ =^f(n)/(j>(n), where f(n) and <^(?2.) denote

By

the

integral functions of n, the series

is

convergent
;

when

the

degree of
that
it is

<^

(n) exceeds that of /(?/)

by more than 1

otherwise,

divergent.
1.

Example

Show

that the following series are convei'gent.

(3)

11
a(a

b)

{a

b){a

+
2b)
{a

1
1

* .....
-|-

b) (a

36)

Example

2.

Show

that the following series are divergent.

V2

V3

Vi

+ 2 V2

+ 3 V3

I+4V4

CONVERGENCE OF INFINITE SERIES

527

Example 3. Write out the first four terms of the series in which u has each of the following values and determine which of these series are
convergent and which divergent.
(1) ^ '

=
(n

2n-l
+
1)

._^

(n

(2) ^ '

Un

V;^
ri-

.,

2)

(3) ^ '
l

= n2-(n-l)2 '-. n^ + (n + 1)3

^^

*' Theorem 4. The positive series u^ M2 convergent if the ratio of each of its terms to the immediately preceding term

951

is less

than some number

in i

which

itself is less

than

1.

+ u^ + Uz + (1) the ratio of each term to the immediately preceding term is less than the corresponding
For
ratio in the geometric series i

+ v-^r + i?-^ +

(2), since in

is always less than r, while in (2) it (1) the ratio in question 1. Therefore But (2) is convergent since ? is equal to r.

<

3. (1) is convergent, by 945, If the ratios above mentioned are

equal to 1 or greater than

1,

the series

is

divergent

for in this case lim

0.

Corollary. If as n increases the ratio u + i/Uj, approaches a 1, divergent definite limit X, the series is convergent when A

953

<

when X
1.

>

1.

For

if

<

1,

take any

number
X,

r such that X

<r<

1.

Then, since lim(?/ + i///) = we shall always have n^^/n,^

fore
2.
i^n
?/

after a certain value of

<r

X,

189,

n and there-

Hence the series is convergent, 938, 951. r. + i/ If X 1, after a certain value of 71 we shall always have

<

>

+ i/^'n

>

1-

Hence the

series

is

divergent,

951.

the series is + i/ (m + i/u) 1 and lim(w_^i/?/) 1, no divergent; but when + i/'' conclusion can be drawn from the theorem of 951.
1

When

>

and lim

= 1,

<

Example

1.

Show

3 that 5

3
5

10

3 5

5-7
,^ 10

,^

15

is

convergent.

5 n, and The 7zth term of this series is 3 5 7 (2 n + l)/5 10 15 the ratio of this term to the term which precedes it is (2 n + l)/5 n. - 2/5, which But since (2 n + l)/5 n = 2/5 + 1/5 n, lim (2 n + l)/5 n

is

<

1.

Hence the

series is convergent.

528

A COLLEGE ALGEBRA
2.

Example
,
.

When
?

is

X being positive

u + i

_
l

and

tlieref ore

convergent, 6 2 t2 i 4- S t3 + ^ i + zx + dx X" + 1/n + nx" x" + i(l + 1/n) + l/ft + (i + l)x" + -JLtl _ lim

1 1

\-

1 ^

-I- 3-

-4-

Hence the

series

is

convergent when 1/x

1
1
-\

<
'^

1,

that

is,

when x > L

Q
1

Example
Example Example

3.

Show

that 1 1

'

1-4

4-7
1
'-

1-

is

convergent.

4.

Show that ~

22525
-\

13
1

13
*

-|

\-

is

convergent. ^

5.

When
"When

is

- H
1

convergent, x being positive

Example
positive ?

6.

is

+ xi+x
1

+x

convergent, x being

953

Series in which lim (Uj, ^ ^/Un) =1. ratio u^ + i/n can be reduced to the
*n

In a series of form

this kind the

+ i/n

where lim

= (o-/n)

0.

= 1/(1 + a-/?i)> We proceed to show

that

if,

as

becomes and remains greater than some number which is itself greater than 1, the series is convergent but that if a ultimately becomes and remains less than 1, the series is divergent.
increases, a ultimately
;

1.

have

<T

For suppose that after a certain value of > 1 + a, where a is positive.

!l^Li

n,

which we may

call k,

we

Then

=
1

^-

Un

<
1

a/n

+
+
I

(1

when n^k.

But we may reduce

Un +
In
(1)

this inequality to the

a)/n form
i],

i<~[nUn
k

{n

l)u +
I

when n>fc.

(1)

set

=
1

k,

+
+2

1,

+
Ujt

successively,

and add the resulting

inequalities.
MA-

We
+

obtain
rii

+
+

+
I

i<-[kUk

{k

+l)u/,+
of the
less

,].

(2)
I

It follows

from

(2) that as

increases the

sum

first

terms of
finite

the positive series Uk + i

Uk + 2

remains always

than the

CONVERGENCE OF INFINITE SERIES

number kuk/a, which proves
that this series

529
There-

is

convergent, 944.

is convergent, 938. fore the complete series Wi + Mj + 2. Suppose that when n > A; we have < 1

Then
But

^^^^^

=
1

^-

Un
is

>

a/n

+
^
1

when n>k.

1/n,

the ratio of the corresponding terms of the divergent 1/(1 + 1/n) for l/(n + 1) ^ 1/n = 1/(1 + 1/n). series 1 + 1/2 + 1/3 +

Hence

tlie

given series Ui

ug

is

divergent, 946,

3.

If a remains greater than 1 but approaches 1 as limit, the preceding test will not determine whether the series is conver-

But in this case " can be reduced to the form a = 1 + y8/, where lim /8/?i = and if /? remains less than some finite number h, the series is divergent.
gent or divergent.
;

For since

/3

< 6, we have

1 1 1 ~ ^ ^ + <:r/n + 1/n + /3/n2 1 + l/,i + b/n'^' But 1/(1 + 1/n + h/n-) in turn is greater than the ratio of the corre sponding terms of tlie divergent series 1/(1 6) + 1/(2 6) + 1/(3 h)-\

Mn + l u

_ ~

For
(n

1
=

+ l)-&

n-h
n

smce

111
=
6

n-h
(n-6) +
n
1

=
l
1

+
-\

l/(n
6
1

62

<
6) l

l/n

+ ft/n^

=
b/n

1 n

(-
n^

n^

Therefore the given series ui

It follows

U2

is

divergent, 946, 3.

which

?/

+ i/ can be reduced to the

from the preceding discussion that a form

series in

954

is

convergent when

a'

>

1,

divergent

when
its

a'

a<^

1.

For dividing the denominator of M4-i _ nP + anP-i +

this fraction

by

numerator, we have

Un

ni>

a'nP

_ ~
1

(a'

- a)/n +

Pn/n'^'

where ^

is finite.

Example.

Prove that the " hypergeometric series"

a((T

g-^
1-7
is

+ l)/3(^+l)
/3

a(a + l)(a +

2)^(;3

1-27(7 + 1) convergent when 7 a

1-2 -37(7

1) (7

> 0,

divergent

when 7

+ 2) + 2) a ^ 0.
+
l)(^
)3

630

A COLLEGE ALGEBRA
EXERCISE LXXXVI

Determine whether the followhig

1

series are

convergent or divergent.

2+1
2

+
22

+1

23+1

+
4.

1-2
1-

1-3

1.2-3 1.3.5

f-'

4
-\

2.3
1

3-4
1
3

+
4
1

5.

V3
i
2
8.

V3

+T- +
V3
2-4.6
4-7 10
4

6.

i+
4
.

+
a2

+2

+
4
7

+
2

fi

ji

10

(3

+
1)

+
+

3 6 3 4

4
1

4
1 2-

4^
+
1.3 2T4

2_^

+
+

2.3-4--.

(71

1)

4.5.6.

(n

S)

9.

13.5. -(2n-l) +
2.4.6...2n

Write out the first four terms of the series in which u has the following values and determine whether these series are convergent or divergent.

n
10. u

n{n
12.

+1 + 2)
1

.. 11.

Vn
u

VnF+1
-n

Un

= Vjl2 +

VnF+l +
Determine whether the
series in

n
which w + i/u,i has the following
u +

values are convergent or divergent.

j3

Wn +

i_
2

2n
14.

Un

n+ 3

3n^-2n^ n'5 + Ji2 + 1

For what
15.
,

positive values of x are the following series convergent?

3 5
1

3-6 5.8

3.6.9
5-8.11
X-2

16.
1

+
1

+
X2
1

+
X3
1

X"*

+
,

+
X*

17.

Show
is

a
that
1

+
,

a(a
'

\)

a(a

1-2

l)(a

2)

1-2.3
Vv,,

+
,

is

divergent

when a
18.

positive.

If for all

values of n
Ui

we have
is

< r, where

is

positive
it

and

less

than

1,

show that

+
r^

Un

+
r''

convergent by comparing

with the

convergent series r

CONVERGENCE OF INFINITE SERIES

531

SERIES WHICH HAVE BOTH POSITIVE AND NEGATIVE TERMS

General test of convergence.
series of

By

definition, 937,
if 5'

an infinite
approaches

955

a finite

is convergent any kind u^ + u^ + limit as n is indefinitely increased.

But, 195, 197, S^ will approach a limit if the sequence of values through which it runs as n increases, namely, S^, S^,
Ss,

number
which
less

possesses the property that for every given positive 8, however small, a corresponding term S^ can be found

differs
8.

than

numerically from every subsequent term S/^.^^ by If this condition is not satisfied, .S' will not

approach a
Since

limit, 198.

and

we have

.S-^.^^

= Ui + = i + + ^. + ,S\.^^ - S = u^^^ + 7/^.+,

Sf.
-\- u,.,
-I

n,,

-j

h A.+p,

-\

\-

u^^^.
:

Hence the following general

test of convergence

convergent if for every given j^ositive number 8, however small, one can find a term u^ such that the sum of any number of the terms after w^ is numerinfijiite

Any

series Ui

Ug

is

ically less

than

8 ; in other words,

such that

|% +
for
it is

^^k

2H

H%+p1<

all values

of

p.

If the

series does not possess this property,

divergent.
in particular,

vergent unless lim

cannot be cona series + 2 + = 0. But this single condition is not sufficient for convergence. We must also have lim (?< + ti+i) = 0, + + + = 0, and so on. lim(i< -f

Hence

?<i

?/

,)

^
/

is divergent altliough lira u = lim \/n = 0. 1/3 + For in this series the sum of the k terms which follow the term \/k is

Thus,

1/2

always greater than

1
-I

1/2.
-1

A;

+ iyi: +
Hence
A;

1-

111
Thus,

>

k-\-k

2k

tI

2k

1-

,,, to terms,
+

i.

.^1,1 > 2k
e.

or 2

cannot be so chosen that Uk+\

is

Uk + ic'is

less

than every

assignable number, and the series

divergent (compare

94:8, 2).

532
956
is

A COLLEGE ALGEBRA
1.

Corollary

A series ivhich has both positive and negative terms

is

convergent if the corresjiotiding positive series

convergent.
let
its

given series, and u'l + u'n 4(2) be the same series with the signs of all Then negative terms changed.
let

For

Ui

+ 2 +

(1)

be

the

|%+i

I't+o.

-\

+ ifk+p\ <

li'k+i

+ 'a-+2 H
we can make

h u\+p.

Hence,

if

by taking k great enough

the same will be true of ["t + is convergent if (2) is, 955.

+ %+^J.

Therefore (1)

957

also shows that a series with imaginary terms is convergent if the series whose terms are the absolute values of u^, ?/, , 232, namely, the series ''i + "2 + is convergent.
111

The preceding demonstration

>

Thus, i/l
is

iV22

i-VSS

is

convergent since

+ 1/22 +

I/32

convergent.

958

Corollary 2.

series tvhose terms are alternate! >/ positive

is

and

negative

is

convergent if each term

it,

numericaUg

less

than the

term which i^recedes

and if the

limit of the nth term is 0.

For

let

positive.

the series be i ffj + "3 Using the notation of 955,

where

a^,

a<,,

are

we here have

We
in the

can write

%. + i
(a,.

form
form

and

in the

a^ +

- + ^ h(-l)^-'%.+p - a^ + 2) + (,. ^^-a^^^^ - (i. + - i + 3)

x.
2

(1) (2) (3)

Since

a^.^.,

>

cik+i^

%+3

->

parentheses in (2) and (3) is (2) that (1) is positive, and from (3) that (1) is algebraically less than o^. + i, and therefore from (2) and (3) combined that

"1 each of the expressions in Hence it follows from positive.

'

numerically less than a^_^^. since lim a = 0, we can choose h so that a^^i<8. Therefore Oi ag + 3 is convergent, 955.
(1) is

But

CONVERGENCE OF INFINITE SERIES

533
959

is

Absolute and conditional convergence. convergent real series said to be ahsohiteli/ convcnjent if it continues to be convergent when the signs of all its negative terms, if any, are changed
;

conditionallij convergent if it

becomes divergent when these

signs are changed.

1 1/2 + 1/4 1/8 + is absohitely convergent since the is convergent. + 1/2 + 1/4 + 1/8 + But 1 - 1/2 + 1/3 - 1/4 + , which is convergent by 958, is only is divergent. conditionally convergent since 1 + 1/2 + 1/3 + 1/4 +

Thus,

series 1

Theorem.
bi/

themselves

In an absolutely convergent series the j^osltiue terms form a convergent series, and in like manner the

960

be

negative terms by themselves. And if the sums of these two series P and N. respectively, the sum of the entire series is P

But in a conditionally convergent series both the series of positive terms and the series of negative terms are divergent.
For
infinite
let Ux

^2

be a convergent series which has an

of positive and negative terms. Of the first n terms of this series suppose that^^ are positive and q negative. Then if S denote the sum of all n terms, Pp

number

the

sum

of the

j^

negative terms,

we

positive terms, shall have S^

and

N^ the N^. p^

sum

of the q

When

is

indefinitely,

and since

indefinitely increased both p) and q will increase S,^ will approach the finite limit S, one

of the following cases must present itself, namely, either (1) both Pp and N^ will approach finite limits which we may call

and N, or

both P^ and N^ will approach infinity. In the first case lim 5 = lim (P^ N^ = lim P^
(2)
is,

lim

203, that

P N.
is

N^,

The
iV,

series is absolutely convergent.

In

fact, after the

change of the signs of the negative terms the

sum
Por

of the series

P+

In the second case the series

if
.S"

is

denote the sum of the

first

conditionally convergent. n terms of the series

obtained by changing the signs of the negative terms, lim S' = lim (P^ + N^) = oo.

we have

534
961

A COLLEGE ALGEBRA

The terms of a condltionalhj convergent series may Corollary. be so arrancjed that the simi of the series will take any real
value that

may

be assigned.

Eor, as just shown, in a conditionally convergent series the positive terms by themselves and the negative terms by themselves each constitute a divergent series the limit of

whose wth term

is 0.

Hence, for example, if we assign some positive number c, and then, without changing the relative order of the positive terms or that of the negative terms among themselves, form 5 by first
adding positive terms until the sum is greater than c, then negative terms until the sum is less than c, and so on indefinitely, the limit of this S, as
?i

is

indefinitely increased, will be

c.

Hence

the commutative law of addition does not hold good

for a conditionally convergent series.

EXERCISE LXXXVII
1.

Determine whether the followmg

series are

convergent or divergent.

^''3
2.

3-5.7 _ 3 5 7 9 3j_5 3.6'^3.6-9 3-6-9.12

for

For what real values of x are the following what values are they divergent ?
(1)

^'l_x
X
^
'

Ill
h
l

series convergent

and

+ 2x
X'

l-3x
Z5
1

+
l

1 h

H-(-l)"Jix
X2n-1

absolutely convergent, and Oi, ai, Os, denote any sequence of numbers all of which are numerically less than some finite number c, prove by the method of 956 that the series
3.

If

Wi

H-x2 1 + 2x4 + 1*2 + ^<3 +

3x6

nx2''

is

aiMi
4.

a<2,U2

azUz

is

also convergent.

If

S denotes
sums
S.

the

sum

of a series of the kind described in 958,

show

that the
less

oi, ai

Oo, ai

ao

-\-

a^,

are alternately greater and

than

CONVERGENCE OF INFINITE SERIES

CONVERGENCE OF POWER SERIES
Power
series.

535

This name
a^x

is

the form a^ variable but

Uq,

a^pc^

given to any series which has + :" + (l)j where cc is a

962

ao, !,

are constants.

The values
if

of x and

Ui,

By
jao]

be real or imaginary. 957, the series (1) is convergent

may

the positive series

+ \aix\ + \a2X^\ +
is

+ \a.^x"\ +

(2) is convergent.
is

When

(2)

convergent we say that

(1)

(compare 959). Whether (1) is depend upon the value of x. Hence the importance of the
following theorems.

absolutely convergent convergent or divergent will

Theorem

If when x

=b

every term of ao

ajX

+
c,

is

963

numer'ically less than some finite positive x|<|b| the series is absolutely convergent.

number

when

For since

a,fi"

<c
|

for every n,

we have

(;"

= |ai"|of |ao|

<

for every n.

Hence each term

+ [^i^I + 1^2^^ +
|

(1) is less

than

the corresponding term of

x -

X
c

(2).

But
that

(2),

being a geometric
\x\

series,

converges when

\x/b

< 1,

is,

when
l

< \b\. And

+
2x^

when

(2) converges, so does (1),

945,

1.

Thus,

+ 2x +
1
.

x'^

---

converges
is

when

|x]

< 1.

Corollary

If

a,Q -\-

aiX

+
|x|

convergent

when x

b, it is

964

absolutely convergent

when

< |b|.

is

This follows immediately from 963. For since ao + i^ -l convergent when x = b, all its terms have finite values when

b.

Corollary'' 2.

7/* ao

also divergent

when

+ [x| > |bj. +

ajX

is

divergent

when x

= b,

it is

965

536
For were a^

A COLLEGE ALGEBRA

is

966

to converge for a value of x which + a^x + numerically greater than b, it would also converge for x = b, 964. Limits of convergence. It follows from 964, 965 that if

all positive values of x for which converges and to a class A^ all for which it diverges, each number in A^ will be less than every number in Ao. Hence, 159, there is either a greatest number in A^ or a least in A^. Call this number X. It represents the Ihnit of the series being convergence of a^ + a-^x + absolutely convergent when [a-] < A, divergent when |cc| > A.

we
Gq

assign to a class A^
a^x ^

Thus, in both x + xV2 + xV3 + (!) and x + x'VS^ + x^/2,^ + (2) the limit of convergence X is 1. Observe that (1) diverges and (2) con= X = 1. It is possible to construct a series in which verges when X X = for example, the series x + 2 x^ + 3 x^ + .

is more frequently called the radius of the circle of convergence. For if we picture complex numbers by points in a plane in the manner described in 238 and draw

Wliat we have called the limit of convergence

a circle whose center

ao + OiX + the circle, and the circle, 239.

at the origin and whose radius is X, the series will converge for all values of x whose graphs lie within it will lie without diverge for all values of x whose
is

graphs

967

Theorem

2.

If in
fi,

ao -f ajX H
fi is

a definite limit
For, by

then

the ratio a/a + j approaches the limit of convergence.

\ [

952, the series

1

\a^\

^-\a^x\ ^

converges
|.r|< lim a n +
a..

when

lim

*n +

-^

ax"
|c^,)|

<

1,

that

is,

when

Similarly

+ |ffix| +
3-5
5-10

diverges when

> \x\
l)

lim

..

'rt

+1

Example
.

1.

Find the limit of convergence of the

3 -X
5

series

X2

+ 3.5--.(2n +
5- 10- -.5/1
,

omce
a +
i

=
2n +
3

=
2

we have u =

lim
2

= -.
+ 3/Ti
2

3/ft

CONVERGEXCE OF IXFIXITE SERIES

Example

537

2.

Find the

limits of

convergence of the series

+ 23x-3 + 22a;-2 + 2x-i + 1 + a;/3 + ccVS^ + x^/3^ + . Here 1 + x/3 + x-/'S'^ + is a geometric power series in x which converges when |x|<3, for o/a + = 3. On tlie other hand, 2x-i + 2'^x-^ + 2^x-^ + is a geometric power

series in

x-i or 1/x which converges when lx-ij<l/2, and therefore when |x|> 2. Hence tlie given series converges when 2 <|x|<3.

Example
x/(l

3.

For what
x)

real values of

x will the series

.

+ 3 xV(l + x)3 + converge ? This is a power series in x/(l + x) which converges when |x/(l + x) |< 1, for lim a/a + = lim n/{n + 1) = 1. But |x/(l + x)|< 1 for all positive values of x and for negative values which are greater than /2. Hence the series converges when x > 1/2.
+
2

xV(l

x)2

The binomial, exponential, and logarithmic series. We proceed to apply the preceding theorem to three especially important

968

power
1.

series,

The exponential series,

X
1

990, namely,
x"

x^

nl
x.

is

convergent for
For here

all finite

values of

=
a+i
n\
lim

-.

(n
(ji

+
1)

=n+
1)!

1.

Hence
2.

= lim

x, that

is, /i

= oo.

The logarithmic
-i

series,

992, namely,

o
1,
1
;

n
divergent

is

convergent when \x\<

when

|x|

> 1.

^or here
I.

3i

=
1

=
1

+
n

a +

n
lim

+
n

Hence lim

~ =
ftn

=1,

lim

+l

+ -")=:11/

1,

that

a is, '^ '

1.

The

series converges

when x =

958, diverges

when x

1,

948.

538
3.

A COLLEGE ALGEBRA
The binomial
series,

namely,

where

is

not a positive integer,

|x|

is

convergent when \x\<.

1,

divergent
For here
a

when

>

1.

_7n(in
i

\)

(m

1)

m(m
lira

I)

(??i

n)

a +

12--n
lira

Hence

=
a +
1

lira

VI

l-2---(n+l)

mn
fj.

-t11

X 1001, Ex.

When
When

the series converges

1

if

= 1, tkat is, = 1. m/n 1, diverges if ?h< 1 (see m>

1

2).

a;

the series converges

I,

if

m > 0,
a we
l)(

diverges

if

m < 0.

For when x the form

=
1
I

by setting
"^"^^-^^
I
I

m=
(^

may

reduce the series to

^)
I

1-2

1-2-3

Evidently from a certain term on all the terms are of the same sign, so that the test of 954 is applicable, 956.

But here
Hence,
or, since
954,

- =
u +
i

+ n~l =
n
if

+ {a-l)
^

the series converges

(a

it

1)

>

1,

that
if

is, if

m,

it

converges

if

in

> 0.

But

diverges

(a

a > 0, 1) < 1,

that

is,

if

m < 0.
EXERCISE LXXXVIII

Determine the
1.

limits of

convergence of the following

niH^/S
!

series.

+mx + m2xV2! +
+
3(2x)3
2)

.
...

2.

2(2x)2

+
,

2(2x)*
TO (ni
!^

3.

mx +

m (m 5
2!

x2 +

+ 3(2x)5 + 2) (m 4)
'-

x3

3!

For what

real values of

will the following series

converge

3x
4.

1/ 3x
\--(

\2

+
X

2^x4-4/
/

)+-( 3Vx
/

1/ 3x

\8

)+ + 4/
\3
'
"

\2

x2TT
6.

x^Tl / Vx^Tl / + (3x)-2 + (3x)-i + (3x)-3

V

"^

+ 2x +

(2x)2

(2x)8

OPERATIONS WITH INFINITE SERIES

XXXIII.

539

OPERATIONS WITH INFINITE SERIES

SOME PRELIMINARY THEOREMS

When
sum
is

which we may represent by = a^ + a^x + In what follows when f(x), writing f(x) we write f(x) = a^ + i^ + , we apsume that a^ + (ii^ + has a limit of convergence A whicl is greater than 0, and

a given power series a^ a definite function of x

+ a^x +

is

convergent,

its

969

suppose that

|a-|

<

A..

Given that ^ (x) = aiX + ajX^ -{-..., and that when X has the positive value b every term of ^(x) is numerically less than some finite positive number c.

Theorem

970

|^(x)|

If any positive number 8 be assigned, hotvever small, then < 8, whenever |x| < b8/(c + 8).

For, as was

shown

in the proof in
2

963,

when

ja;|<5,

\<i>{x)\<c

and therefore

<
'

,,

that

is, '

\x/b\

< 7^^,^-|^|
8,

704

Hence
that
is,

\(i>(x)\< 8 ^ '^'

when

c\x\ '- ^.
1x1

<

when
i^
f (x)

Corollary,

s* c + = ao + aiX H
I^:^! '
'

<

th en

li^ f (x)
)

= a = f (0).
200.

971

For, as just shown,

li]i^

{a^x

a^x^

-\

0,

Theorem

2.

//'

the series ag
it

+ ajX +
have

vanishes for every

0,

972

value of X for which

converges, then ao
at once

For setting x

= 0,

we

Hence
for every value of

a^x

+ a^x^ +

a^x^

-|-

= 0, ai = a^ = 0. =

(1)
',

x for which

it

converges.

.>''

540
If
a;

A COLLEGE ALGEBRA

0,

we may
ai

divide (1) throughout by

x.

Hence

a^x

ag.x'^

=
;

(2)

for every value of

for

a;

a^ =^

But it 0. we could choose x

x for which it converges, except perhaps for were follows from this that Oi =
so small (without

making

it

0) that

\a^x

+ a^x^ + !< |i|,

we have
0.

970, and such a value of x

would not
be

satisfy (2), as

just

shown

it

must.
it

Hence
that a^

a^
0,

cts

= 0,

And by
and
a

the same reasoning

may

shown

so on.
6x'

The

like is true of

ex

dx'

which the exponents of x are positive and

the reasoning just given applies to
all

different
series.

and of every series in from one another for

;

such

vanishes for every value of x for that ao + aix + converges contains more than is required for the proof that For the reasoning above given shows that if ^i, ao = 0, ai = 0, -denote any given never-ending sequence of numbers such Pni Piivanishes when x = /3i, ^2, that lim /3 = 0, and if ao + otiX + A.,

The hypothesis
it

which

then ao = 0, ai be rational.

0,

In particular, the numbers

/Si,

^2,

may

all

973

= bo + bix + bzx'^ H Theorem 3. 7/* ag + aix + agx'^ -| for every value of y. for which these series converge, the coefficients = bo, a^ = bi, a.^ = bj, of the like powers ofiL are equal ; that is, a.^
and
so on.

Tor subtracting the second series from both members of the given equation, we have, by 974,
(ao

^o) + (i ^1) + (2 ^2)

which the given
0,

a;^

= =

for every value of x for

series converge.
a^

Hence,
that
is,

972, a^

flo

S^ = 0, i b^ = = ^i = ^u fh =
^0)

b^

0,

^2)

This theorem
cients.

is

called the theorem of undetermined

coeffi-

x cannot be expressed in more than one way as a power series in x (compare 421).
It asserts that a given function of

OPERATIONS WITH INFINITE SERIES

OPERATIONS WITH POWER SERIES
Since

541

many

functions of x can be defined by means of power

is

series only, it

with such

important to establish rules for reckoning These depend upon the following theorems, 974, 976, which we shall demonstrate for infinite series in
series.

general.

Theorem
converge
(Uj

1.

If the
the

series

Uj

U2

(md
the

v^

Vj

+ +

974

and have

sums S

ajid

respectively,

the series
T.

Vi)

+ (u2 + Vo) +
203,

converges

and has

sum S

For,

lim

= lim (i = S + T.

+ ( + v)] \_{u^ + v,) + {n. + v^) + + U2^ ) + lim (y^ + v^ v^

\-\

\-

The like is true of the series obtained by adding the corresponding terms of a.ny Ji.7iite number of infinite series. Hence the rule for adding any finite number of functions defined by power series in x is to add the corresponding terms of these
series, that is, the

975

terms which involve like powers of

x2

x.
.

Thus,
then

if

/(x)

=1+
/(x)

= X + 2 x2 + X + + (x) + 9i(x) = l + 2x + 3x2 + 4x3 + ...

and
is,

3 x3

when

the given series converge, that

when |x|<

1.

number of series whose sums are and we add the corresponding terms of these series, we ordinarily obtain a divergent series, even when the series But in the case described in the is convergent. +- 7' +following theorem we obtain a convergent series by this process, and its sum is 5 + T +
If there be given an infinite
T,
S,

.S'

Let Ui + U2 + denote a convergent series each of ivhose terms is the sum of an absolutely convergent series,

Theorem

2.

976

namely,

Ui

u^P

u<'>

(1),

U2

u^P

+ uT +

(2), ....

542

A COLLEGE ALGEBRA

denote the sums of the series obtained Again, let Uj', U2', their absolute values, by replacing the terms of (1), (2),

by

so that

and

so on.

series Ui' is convergent, the several series U2' obtained by adding the corresponding terms of (1), (2), ,

If the

namely, the

series u<J)

u<p

-\

u^ +

u^^)

-j

and

so on^

are convergent, shall have

and if

their

sums

be denoted by Vi, Vj, ,

we

Ui
For
let

Uo
...

+
by

U3

= Vi + V2 + V3 +
. .

us represent the remainders after n terms in the

R']^, R<^^,
(!>
-,

series (1), (2),

. .

943, so that
ci)

U,

(i)

7.(1)^

U,

= u^'p + (|) +
series

+ ) + R^^\
i<*f
-|
,

Each
and

of the
is

column

u'-\^

fp

?<<f

-\

convergent since each of its terms is numericallyless than the corresponding term of the convergent series
Ui + U^' -{-, 945, denoted by V V
If

so on,

1.

Let the sums of these series be

V, R,,

we add the corresponding terms of these n we obtain the original series Ui + U^ since n is finite, we have, 975,
series,
-\
.

column

Therefore,

U,+

C^2

---+ f4

---=

Fi+

72.

To prove our theorem, therefore, we have only to show that when n is indefinitely increased lim R = 0. But if the remainder after k terms in R'^]^ + 7?^;' + be denoted by 6<*>, we have R = R^l^ -j- R^^J h R^t? + S^n^

-\

Let

denote any positive number,

R'^]^

it

matters not

how

small.

Since each term of

7?<;f

-\

(a) i? numerically less than

OPERATIONS WITH INFINITE SERIES

the corresponding term of Ui
-\-

543

U^'

-(b),

the remainder

after k terms in (a) is numerically less than the corresponding remainder in (b). But since (b) is convergent we can so choose

k that the latter remainder will be less than 8/2. can so choose k that tvhatever the value ofn have 5^*' < 8/2 numerically.

Hence we
be,

may
u^\'

we

shall

But
w^f

again,
u^:^

since each of the

row

series

+ ^P +

>

is

convergent, as n increases each of the k

will ultimately become and remain R^]^, R^l\ 7?*^^^ numerically less than 8/2 k, and therefore the sum of these + jR^J,'^ will become and remainders, namely, ii^,^ + -R',f + remain less than (8/2 k) k, or 8/2.

remainders

Therefore, as
will

?i

increases, /?

7?'P

R^;^

-\

\-

R^^^

S^';^

ultimately

become and remain numerically

less

than

8/2

8/2, or

8.

Hence

liui iZ

= 0,

200

and therefore

as

was

to be demonstrated.

series

Ui

U^

each of whose terms

is

itself

an

infinite series is called a doubly infinite series.

Thus, consider the series

x/{\

x)

xV(l

x)2

a;V(l

x)3

(1)

which converges for all real values of x which are greater than 1/2 1/2). (also for imaginary values of x whose real parts are greater than Is it possible to transform (1) into a power series in x, that is, into a series which will converge for any value of x except ? When |x|<l, each term of (1) is the sum of a power series which loay
be obtained by the binomial theorem,
988.

Thus,
--

+ x)-i = -XV(1 +X)2=-x2(l + x)-2= XV(1 + X)3 = X3 (1 + a;)- 3 ^

x/{l

x)

(1

-x2 + x3 - x* + -X2 + 2X3-3X* +

x3

3 x*

(2)

Replacing each term of the

first
,

of these series
is

by

its

absolute value,

we

obtain

|x|

jx^l

[x^] 4-

whose sum

|x|/(l

)x|).

544

A COLLEGE ALGEBRA
we
is
. .

Treating the remaining series in a similar manner,

obtain series

whose sums are |x2|/(l Hence the series Z7i' +

\X\/{1

|x|)2, \x^\/{l

W+
1/2.

Us'

-]

Ix|)^ and so on. of our theorem

|X3|/(1

here
.,

\X\)

|X2|/(1

|XI)2

_ lx|)3 +

which converges when

|x|

<

Therefore, when |x|<l/2, the power series obtained by adding the - 2 x^ + 4 x^ - 8 x* + corresponding terms of the series (2), namely, x converges and is equal to the given series (1) that is, when x < 1/2 we have - x7(l + x)-^ + xV(l + x)3 = X - 2x2 + 4x3 - 8x4 + . x/(\ + x)

977

Commutative law valid

are

for absolutely convergent series.

We

in a position to demonstrate that the terms of an absoat pleasure without lutely convergent series may be rearranged the sum of the series. changing

now

1.

We may
let ui 4- M2

rearrange the terms so as to form any other

single infinite series out of them.

let Ui

and (1) denote any absolutely convergent series, denote the same series with its terms rearranged. of Again, let 8 denote the sum of the first n terms of (1), and S'm the sum the first m terms of (2). Assign any value to n then choose m so that the first n terms of (1) are
For

u./

(2)

to

be found
first

among

the
(2)

first

terms of

(2);

and
the

finally
first

choose

so that
(1).

the

terms of

are to be found

among

+ p terms of

Then S; - S is made up of terms in S +p S, that is, of terms the sum u + 1 + + 2 + + +/. Hence |S,' -S|<|u + ]| + |h + 2| H + |Un + p|+ |Mn + /)|) = But since (1) is absolutely convergent, lim (|tt + i| +

in

0-

Therefore lim |S^

2,

S|

0,

that

is,

lim S'm

lim S,

We

may break

the series up into any

number

(finite or

infinite)

of series the terms of each of

in the original series.

which occur

in the

same order as

For we can recover the

original series from every such set of series of the theorems of 974, 976.
Thus,
if

by applying one

we form one
indices,
974,

series out of the

which have odd

and another out

of those

terms of Wi + M2 + Ws + which have even indices,

we

liave,

by

Wl

"2

U3

U4

(Ui

W3

"6

("2

W4

"6

OPERATIOXS WITH INFINITE SERIES

Or
Ui
U2 M4
again, arrange the terms of Ui

545
:

u^

Ua

as follows

In this scheme there are an infinite number of

+ + +

U3
Ws Ms

columns, each forming an infinite

series.

+ +

Me M9

The sum

of Ui
tlie

Mo

+ Ms +

is

equal to the

sum

of

Mio

the terms of

scheme added by rows, 940. the sum by rows is equal to the sum by columns,

And
976.
.

Hence

Mi

+ M2 +

M3 H

(mi

M2

+ M4 +

(M3

M5

) 4

And
3.

similarly in every case.

Every possible rearrangement of the terms of

111

^2

ih-\
2.

may

be had by combining 1 and

series.
a;

Products of power
defined, when
<^ (a-)
I I

If the functions/(a3)

and

(f>

(a*)

are

978

< X, by the power series/ (x)=a(^-{- a iX-\ (1), = ig + bix + (2), their product /(cf) (x) will be defined, when \x\ < A, by a power series derived from (1) and

<f>

(2)

by the ordinary rules of multiplication (compare

f{x)

314).

Thus,

ao

aix

546

A COLLEGE ALGEBRA

sions for y,

repeatedly multiplying (2) by itself, we obtain expresin the form of power series in x which y"^, y^, converge when (2) converges. If we substitute these expres-

By

sions in the terms aiy, ^2^^

of (1),

we
2

obtain a series of
H
)

the form

-!

(^o

+ ^i^ H

+ 2 (^o +

bf^b^x

(3),

and
are
(o

this,

when
)

collected,

the terms which involve like powers of x becomes a power series in x of the form

i^o H

+ (i^i + 2 azMi
values
of

-1

)^-\

(4).

This
as
(1)

final

series will converge

all

and have the same sum

ji?*,,!

for

x such that

|Jia-|

< A.

For

in this case the condition of

infinite

976

is

satisiied

by the
being
con-

doubly
|o|

series

(3),

the

series

C/j'

Uo'

+ |^il(K^| + i^i'^i +)+ ' which < ^ verges when l^o] + l^r^I +

by

hypothesis

980

Quotients of power series.

fraction

whose numerator and

denominator are power

(tto

series, as

ajx

-1

)/(/>

b^x H

),

may be transformed into a power series which conwill converge for all values of x for which a^ + a^^x -f
where
b^ =^ 0,

verges and
For
let

\bix\

+ [b^x^l -{-<
y
1

\bf,\.

biJ:

Then
6o

111
+
box'^
-\
.

(1)

bix

62x2 H

bo

ij

60

-f-

y/bo

,2

bo\
since,

60

Oo

by hypothesis and 2-32, |y|<|5ia;| + \b-zx"\ + <|6o|' We In (2) replace y by its value (1) and then apply 979. transform (2) into a power series in x which converges when
\bix\

shall thus

\box'''\-\

<|6o|.

Multiply this power series by Uo + ciix + a^jfi +, 978. The result will be a power series in x which will converge and be equal to the given fraction for all values of x for which a^ -\- a\X 4converges

and

15ix|

162x21-1- _-<|&ol-

OPERATIONS WITH INFINITE SERIES

547

The quotient series may be obtained to any required term by the process of cancelling leading terms described in 406, or by the method of undetermined coefficients, 408.
Example.
terms.

Expand

(1

+ 2^2

4-

.)/(i

x^

to

four

Using detached

coefficients,

we have

1+2+4+8+ 1+1+1+1+
1

1+1+1+1+

+ +

3
1

+ +

7
1

+ +

---

548
Example
2.

A COLLEGE ALGEBRA
Expand
(1

x)/(x2

4.

4 x^) in increasing powers of

1

x,

We

have

l-x
x2h-4x3

11-x

(1 x'^ 1 + 4 X x^ x-2- 5x-i + 20

5x + 20x2-80x3+
80 X
2X

Example
2 x2
x3

3.

Expand

(2 x^

3)/(x3

+
X

4) in

powers of 1/x.

_ ~

1
'

2
1

2X

+ +

1/x

3/x2

2/x

4/x3

I/2 X\

_?_A

X2"'/X

^-?_A_^
X2~x3"

982

Reversion of

series.

From the
it is

defining y in terms of x,

equation y = a^x + aa*^ + ? possible to derive another of the

form X

x in terms of y. , defining l\y b^ij^ cess is called the reversion of the given series a^x

The
a^x^^

pro

It will be

observed that this series lacks the constant term

is

a^,

and the understanding

if

that i

0.

It can be

proved that

has a limit of convergence greater than 0, aiX -f a^piy^ the like is true of the reverted series biy ^2^^ H

+
.

Example.

Revert the series y x

2x~

Zx^

---.
(1)

Assume

biy

boy-

63?/- H

from the given equation by the method Computing ?/2, y^, and substituting the resulting series in (1), we have
X

of 978,

6iX

2 61

(2)

361

462

63,

...

(3)

By
y

the

same

method,
,

from

an
a^

ax- + or y Uq -\- a^x derive another of the form x =

b^

(y

equation of the forin (lix + a^x- +.-, we can

(7)

+- bo

(y

And from an
983

equation of the form y = a^xr +can derive two others of the form x b^y^ + b^y

a^^

0.30;^

we

b^y^-

An algebraic equation of Expansion of algebraic functions. the form f(x., y) = which lacks a constant term is satisfied

when X

and y

0.

Hence,

if

we suppose /(x,

y)

OPERATIONS WITH INFINITE SERIES

549

solved for y in terms of x, one or more of the solutions must be expressions in x which vanish when x vanishes. It can be proved that these expressions may be expanded in series
in increasing powers of

x which have limits of convergence

In ordinary cases these series may be obtained greater than 0. to any required term by the method illustrated in the following examples.

Example

1.

The equation
the equation
2/2

?/2

?/

2a;

lacks a constant term.

Find the expansion for the value

of

y which vanishes when x

2X

0.

When

0,

2/

(1)

becomes y^ + y = 0. Since one and but one of the roots of this equation is 0, one and but one of the solutions of (1) for y in terms of x vanishes
wlien X vanishes.

Suppose that when this sohition is expanded powers of x its first term is ax'^, so that y = ax^ +

in a series of increasing

(2)

Substituting

(2) in (1),
a2a;2fi

we have

h ax'^

2x

0.

(3)

an identity, the sums of the coefficients of its terms of like degree must be 0. Hence there must be at least two terms of lowest degree and since /u is positive, these must be the terms 2 x. Therefore /^ = 1 and a 2 = 0, or a = 2. ax'^ and We therefore assume that
Since by hypothesis
(3) is
;

2/

=
c

2X

6x2

+
(4

cx^ H

^20

Substituting Hence 4 + 6

(2')

in (1),

we

obtain
0,

0,

46 +

,
is

6)x2 + (4 6 and therefore 6

Therefore the required solution

4 x2

= + c)x3 + = 4, c = 16, + 16 x^ +

0.

Example

2.

satisfy the equation

Find the expansions of the values of y in terms of x which xy + x2 = and vanish when x = 0. y'^
the equation
2/3

When
becomes

=
=

0,

xy

x2

=
0.

(1)

y"^

0, all

three of whose roots are

Hence we may expect to

find three expansions of the kind required. Let ax^ denote the leading term in one of these expansions, so that
2/

ax'^

+
+

(2)

Substituting

(2) in (1),

we have
ax*^

a%3fi

+1

x2

0.

(3)

550
By

A COLLEGE ALGEBRA
the reasoning of Ex.
1,

and 2 must be equal and

ix

less

at least two of the exponents than any other exponent of x

/n

3/i,

/a

1,

in (3).
/u,

= 1/2. This is an admissible value of Setting 3/x = ^ + 1, we find = 1/2, both 3 and m + 1 are less than 2. since when + 1 = 2, we find = 1. This also is an admissible value of Setting = 1, both + 1 and 2 are less than 3^. since when 2/3. But this is not an admissible value Setting 3/i = 2, we find = 2/3, 3/a and 2 are greater than m + 1. of since when
/i
;u /x

fi,

/^

/x

ij.

ytt,

p.

Hence

fx.

When
it

/u

must have one of the values 1 or 1/2. ax~ + = 1, (3) becomes a'^x^ +
.
.

follows that

x^

0,

from which
from

+1=
(3)

0,

or a

=
0,

1.
.

When
which
it

II

1/2,

becomes a%'^

-j-

_
a

ax?
yii

follows that a^

or, since

0,

that a

x^ = 0, =L

We therefore assume that the required = x^ + 6x + cx^ H y = X + 6x- + cx^ +

2/

solutions are of the

form
.

= x^ + 6x +

cx^ H

substituting these expressions for y in (1) coefficients as in Ex. 1, we obtain

2/

And

and determining the

1

o o = X + x2 + 3x3 +
, ,

...,

2/

X 3x? X = X^ ----+,?/=: -X--^ + 2

1
,

3x? +
-

..

In this method

it is

assumed that
is
oa-^,

if

the leading term of one

of the required expansions

the expansion will be in

is is

powers of

a-'.

In exceptional cases this

not true and the

method

general. Having found the leading term ax^ of an expansion as in the examples, x*^ set y It becomes an equa(a v) in the given equation.

fails.

But the following method

x. From this equation find the leading term of the expansion of v in powers of x, and so on.*

tion in V

and

Thus, in Ex.

2, setting

x"^

2/3

(1 + v) in - xy + x2 =

(1)

and

.simplifying,

we have
v^

Sv^

+ 2v +

x^

-0;

(2)
,

whence v= X-/2H

To

find

and therefore 2/=x- (1 x-/2 + ) = x' x/2H the next term, set v = x^ ( 1/2 + v') in (2), and so on.
,

* For a fuller iliscnssion of the inctlioils of thi.s section and the use in connection with tlictii of Newton's iiaralli'loL;r;ini see Clirystal's Alr/ehra, II, also Fro.st's Curve Tracini/ and Johnson's Curve Tracing pp. 349-371
;

OPERATIONS WITH INFINITE SERIES

Taylor's
ja;|

551

<

A,

If f(x) = a^ + a^x + a^x^ theorem. and we replace x hy x -{- h, we obtain

4- h)

when

984

f(x
It follows

ag

aj

(a;

+
|

h)
a:;
]

-\-

a^^x
A
|

+ hy +

from

976 that when

< X we may

transform

this series into a

(a

+ hy,

h by expanding (a + hy, by the binomial theorem, and then collecting terms

power

series in

which involve like powers of h. 848 it may be shown that the

fi^x

By

the method employed in

result will be

+ h)=f(x)+f(x)h+f"{x).~ +
(x),

...+f''^(:x).^

+ ...,

where /'

f" (x),
first,

denote the sums of the series whose

terms are the

second,

derivatives of the terms of the

given series a^

+ 2^^ + namely, = ai 2 aoX + 3 a^x^ + f (x) and /" (x) = 2 a^ + S -2 a^x +

+
UiX
i

-\-

>

so on.

we replace x by a and h by where |a + |a5 a] < A, we obtain the expansion oif(x') in powers of x a, namely,

If in the preceding identity

a,
I

985

f(x)=f(a)+f'(a)(x-a)+..-+r(a)^^^^ +
From
f(x)
this
last

....

expansion and

971
if

it

follows that

if

986

a^-{-

UiX

when

|a;|

< A,

and

[rt|

<

A,

then

]}iy(x)=f(ay
EXERCISE LXXXIX
1.

Show
Show

that
that

(1

X
x

X-

+
+

) )^

= =

+ 2x +
+ 3x + +
1

Sx"^

+
-^

4x^

".

2.
3.

(1

x"

6x^

lOx^

-.

Show

that

(1

x2

x^

)/(!

+ x- +
UiX

---)

= l-x + x^+

-.

4.

Assuming

that

(1

2 x^)'

aoX^

find ai, Og,

as, Ui

by squaring the given equation and applying

973.

552
5.

A COLLEGE ALGEBRA
By
(1)

a similar

method

find the first four terms of the expansions of

(2)
(l

(8-3x)5.

+ x-x2)i.
powers of x

6.

Expand each
2

to the fourth term

by the method of the example

x
1

of the following fractions in ascending in 980.

+ x- 3x2+ 5x3 1 + 2 X + 3 x2

+ 5x2-x3 - X + x2 - x3
powers of x
coefficients.

7.

Expand each
3x2
(1)
1

of the following fractions in ascending

to the fourth term

by the method of undetermined

x

+
+
x3^
x2
^^^

5x*

x3

2 X*

3 x5
fifth

8.

Expand each of the following fractions to the

of the first

term by the

method

example

in 981

and indicate the

limits of con-

vergence of the expansions.

(1)

(x2
9.

^^-^^ - 4) (X -

'^ + ^
(2) ' '

3)

(2x

3)(x

1)2

Expand each of the following fractions to the fourth term in descending powers of x. For what values of x will the first of these
expansions converge
?

2 x2
10.

15

'

'

x^

x3

x2

Revert each of the following series to the fourth term.

(1)
2/

= X + x2 + x3 + X* +
+

(2)

2/

a;

|A

+^-^+ o 4

11.

From

series for
12.

= 1 + X + x2/2 2/ x in powers of y 1.
x2

x^/Z

derive to the fourth term a

From y =
?/^.

3 x* derive to the fourth term a series for x in

powers of
13.

By

the

method
0.

of 983 find the first three terms of the expansions

of the values of y in terms of x

which

satisfy the following equations

and

vanish

when x =
(1) x2

2/2

2,

3X

0.

(2) x3

2/3

x?/

0.

14.

By
"^

aid of the theorem of 976

2 X2

show

that
X2 X3
^

X
1

- X2 +

3x3
1

"^

_ ~
(1

x3

X)2

(1

x2)2

(1-X3)2

THE BINOMIAL SERIES

553

XXXIV.

THE BINOMIAL, EXPONENTIAL, AND LOGARITHMIC SERIES

series.
7)1

The binomial

When m

is

a positive integer,
''

m
(^(i
is

(m

1)

?"
.

!)('l-'Z'd

2)

x^

(1)

a finite or terminating series and its sum is ( 1 + a; )"*. is not a positive integer, (1) is an infinite series, but one which converges, that is, has a sum, when |a'|<l,

When m

'

We proceed to demonstrate that if m has any rational 968. value whatsoever, this sum is (1 + .r)'". The series (1) is a function of both x and m, but since we
are

now concerned mainly with

it

its

relation to

we

shall

represent
that

by </>(?). For convenience let Wy denote the coefficient of x^ in

?,.

= VI (ni
if vi

1)

{yn

r+

(1), so

I)//-

!.

Then

and n denote any two numbers, we have

= 1 + m.^x + ninx"' + TTizX^ + = 1 + n^x + tioX' + n^x^ + (n) = 1 + {yn + n)ix + (m + n).,x^ + <^ {m + n) We can prove that (w) (ii) = {ni + n). For when < 1, so that (2) and (3) converge, we <i>{m)-4>{n)=l + nu r + vu + i
^ (w)
<^

(-)
(3)

.
have

(4)

</>

<f}

<^

|.r|

554

A COLLEGE ALGEBRA
repeated applications of
(6),

By

we have
<f>

<i>(m)-cf>(n)-<f,(p)= (f>(m

-{-

v)

(p)

4>(m

-\-

p),

and so on, for any

finite

number

of factors of the

form ^(w))
all

<^('0' *^(i^)' <^('/)'

We
988

are

now prepared

to prove the binomial

theorem for

rational values of the exponent m, namely:

Theorem. of the series

.

//'

he

any rational number whatsoever,

1) ^

the

sum
,

,^(m)=l

iw

jx+ \
(1

m(m
,^

x^

.,

l)(m -^ + m(m y::^

2)

^x^+---,

when\x\<
Notice

1, is

+
1

x)"".

first

of all that
-m

when

7n

the series reduces to

1,

and that when

the series reduces to 1

1

+
+

a".

Hence
1.

<^(0)=
7n be

and ^(1)=

a-.

(1)

Let

a positive integer.
<j>

Then

(m)

= = =

<^

(1

+1+
<f)

to
to

w
?w.

terms)
factors

<t)(l)

(1)

[cf>{i)r=a

+ ^r,

by(i),

(2)

which proves the theorem

2.

for a positive integral exponent.

Let

be any positive rational fraction j?/;?.

[<^

Then

(p/Q)y

= = =

4>

U'l'l)

<^

{pI'1)

to

(^

factors

^ {p/'l
<^(/0

+ P/<1
p

to ? terms)

= (l+a-)^
^O''-

by

(2).

Therefore

= (1 + <^(;V'?)

(2)

For it follows from the equation [^ (/'/'?)]'' = (^ + ^Y ^"^ 986 that the values which <f>(p/q) takes for all values of x such that |a-l < 1 must be the corresponding values of one and the same (/th root of (1 + a*)''.

THE BINOMIAL SERIES

Moreover
p

555
qih. root,

this root
is

must be the principal

namely,

the only one of the qth. roots of (1 xf (1 which has the same value as ^ {p/q) when x 0.
;

for this

3.

Let

Since

we have

m be any negative rational number = (- 5 + s) = (0) = (s) (- s) = l/<^ (s) = 1/(1 + x)' <^ (- *) = (l+a-)

+ xy

s.

</,

</>

<^

<^

1,

by

(1)

by (3)
(4)

which proves the theorem

It is not difficult to

for

any rational exponent.

extend the theorem to irrational values

of the exponent.

Example.

We

have

(1

Expand + 2x +

(1

2x
=:,

+
[i

a;^)^

in ascending

powers of

x.

3x2)3

(2x

3x2)]3

=l+ +
.

i-(2x

3x2)

+ ilzil
3x2)3

(2

3 x2)2

^
^

3M^^(2X + o
'

...

2x

5x2

68x3

The expansion converges when 2|x| + 31x2|<l; therefore when 9|x2| + 6|x| + l<4; therefore when 3|x|+l<2; therefore when |x|<l/3.
Corollary.

If to.

is

rational

and \x\

<

[a,],

we have

989
-\

= (a + x)
For (a

infm 1) a + ma-^x + ^-^

a"'
f

on ^a.'^-^x^

a;)"*

1 -f

m(m
7H(

1) x^

~\

[.

(1)

= a"" + ma"'- lic-j

where
(1)

\-
7Ti

1^ ^-a"'-^x^-\

(2)

and therefore

(2) converge if \x/a\

<

1,

or

\x\

< \a\.

556
990
The exponential
1

A COLLEGE ALGEBRA
series.

We

liave already

+ VI + ^'72! + x'/3\ +
all finite
its

+
968.

shown that the

x"/nl

(1)

converges for

values of x,
ic

Let

denote
e

sum when

1,

so that
...

= 2.71828- . 1/2! 1/3! We are to prove that the sum of (1) for any real value of x
l

+l+

is e'.

For

let

/(a;)

denote the sum of

!

(1), so
!

that

f(x) = 1
/(2/)= 1

+ a'/l + a-72 + x^S + + y/1 + r/2! + i/V3! +

+
...

x'^/n

..-,

+ yV;i!

+....

Then by the

rule for multiplying infinite series, 978,

/(^)-/(y)=l+(a:

+ 2/)+

X |. 2!

y. + ^y + 2! ^ f'

+ (^ + ^^ + Tl^ + 2! l"^12
V3!

+
f-')

= -11/ + y) + 1 +(^
I

(^ y)^ ^^-2j

+
,

(-^

+ y)' 31^ +

From

this result

it

follows that
y)

/(^) -/(y) /(-)

=/(^ +

f{z)=f(x + y +

z),
e,

and so

on.

Hence, observing that /(O)

successively, precisely as in
1.

and /(I)
vi,

we may prove

988, that

When

ic

is

a positive integer

/(m) = [/(!)]'" =
2.

6-.

When

is

a positive fraction j^/q,

3.

When

a;

is

a negative rational

s,

THE LOGARITHMIC SERIES

Therefore when x
e"
is

557
that
is,

rational
!

we have f(x) = e%

+ X + x72 + a'V3! +
is

+ x^/n + .
!

(2)

Moreover (2) nent X. For if

also true for irrational values of the expo-

made

to

approach

denote any given irrational number and x be h as limit through a sequence of rational

values, for all these rational values of x

we have f{x)

e""

and

therefore ^^^^f(x)

li"i

e^

But li^/(x) =f{b),

fore /(6)~=
e",

li^ gx 986, and

that

is,

+ ^ + *V2 C-\
(2) is also

^ =

g&^

728.

There-

eK
is,

The second member of sum, when x is imaginary.

e'

a convergent series, that

has a

Hence
x.
!

(2)

may
!

be used to define

e-^

for imagi-

nary values of the exponent

Thus, by

definition,
h i"/n\

=1+i+

iy2

iV3

.
991

Series for a^.

Let a denote any positive number and x any

real

number.

e^iSe, 730. e^^^e", 732, we have a^ Therefore, substituting x log^a for x in the series,

Since a

990, (2),

we have
a'^

=l+x

log^a

+ x\\og,ay/2 +
!

+ x" (\og,ay/nl + -.

It can be proved, as in 968, 1, that this series converges

for all finite values of x.

The logarithmic

series.

If in the series just obtained for

y,

a""

992

we

replace a by 1 -f a and x by

we have

(l+xy = l + \og,(l + x)-i/ + [\og,(l + x)fyy2\ + .-..(l)

But by the binomial theorem,

988,

when

|a;|

<

1,

By carrying out the indicated multiplications and collecting terms we can transform (2) into a power series in y. The coefficient of y in this series will be
'

"

(-l)x^ x\ + 472 +-(-!)(- 2)x^ +->or.-- + -3--.. 12.3

x''
,

558
Equate

A COLLEGE ALGEBRA
this to the coefficient of y in (1).

We
-.-.

obtain, if
(3)

\x\<\,
'

log,(l

a;)

= x-a;V2 + a;73-a;V4 +
when \x\<
1.

This series

is

called the logarithviic series.

In

968 we

proved that

it

converges

In the proof just given we have assumed that the series (2) remains equal to (1 + xyj after it has been transformed into a power series in y.

But this follows from 976 when and y respectively, the series TJi
I \

|a;|<l.
-\-

For

if x'

and

y'

denote

|x|

Uo

of 976, corresponding to

^'^'"
^

+ y','^y'^y'^^\'^^y'^y'^'^^y'-'^)x'^ + 1.2.3 1-2

convergent when
x'

..., '

and

this series is

< 1,

its

sum

being

(1

x')-"', 988.

the truth of the binomial theorem only for rational values of the exponent, it may be observed that (3) follows from (1) and
(2)

As we have proved
even when y
of 972).
is

restricted to rational values (see the

remark at the

end

Example.
"We have

Show

that

^\"^

("

e.

logyi +

^y = nlog,(l+-)=n(l- +
1

...)
("l

= l--^ + + -V =
e^

....

Hence ^' logef

For limu

V=

1,

and therefore

e.

limeio?e

ei''"(ioge),

726-729, 731.

993

Computation

of natural logarithms.
e

The logarithms
:

of

num-

bers to the base

are called their natural logai-ltJims. of natural logarithms may be computed as follows

A table

We

have

and therefore

= x- x'']'! + x^J?> - xV4 + log, {i + x) = x~]2 x^/3 x^]^ log, (1

a:)
a;

,
.

(1)
(2)

Subtract (2)

from

(1).

Since
a;

log, (1

+
-\-

a;)

- log, {\-x)=- log, 1 +

2(a;

j-^^'
\
(3)

we

obtain

log,

X?
-^

+ ^+-j.

x^

THE LOGARITHMIC SERIES

In (3)
set

559

:;

=
1

and therefore x

=
1

7;

We

obtain

n+l_^f
or log^(7i

>
^^^

+
^

1)

= ^^^ +

+ + 3(27 + 1)^ 5(27TT7"^-/ = 1 in (4), Setting w - .6931 log,2 = 2(1/3 + 1/3* + 1/5 35 + = 2 in (4), Setting = 1-0986 = log, 2 + 2 (1/5 + 1/3 5" + log, 3
(,277+1

71

and so on
rithnis of

to

any integral value of

n.

Modulus.

By

nnmbers

Hence the loga755, logri log,?i/log,a. to any base a may be obtained by multi-

994

plying their natural logarithms by l/log^a. We call l/\og^a or its equivalent, 756, loge, the mochdus of the system of
logarithms to the base
a.

system of covfimon logarithms

In particular, the modulus of the . is logigg = .43429

EXERCISE XC
1. 2.

Compute loge4 and

lege 5
!

each to the fourth decimal

4/5
!

figure.

Show

that e-i

2/3

6/7

3.

Show by

multiplication that

4.
5. 6.

Show

Show

bmomial
7. 8.

l - xV2 + xV4 - xV6 + . = x- xV3 + xV5 - xV7 + Show that the (r + l)th term in the expansion of (1 x)-" by n(n + 1)- --(n + r - 1) theorem is

that (e^

that (&^

+ -

e-

'^)/2

'

e- '^)/2

the

^^

x'".

r!

Find the term

in the

expansion of

(8
(1

x)^

which involves

x*.

Find the term in the expansion of

x')~^ which involves

x^.

560
9.

A COLLEGE ALGEBRA
For what values
X

(12

x2)3 in
(1

of x will the expansion of (9-4a;2)i and of ascending powers of x converge?

10. 11.

Expand
Find the
x.

2 x^)^ in powers of x to the term involving

x*.
'

first

in

powers of
12.

three terms of the expansion of (8 + 3 x)^ (9 2 x)~ For what values of x will the expansion converge ?

Find the limiting values of the following expressions.

(1)

lim

'^'

e^^-e-^' 3x

U^
e^.

(i

13.

Prove that

^(\+ ^X' =
1
^ x=0 (e'^

^^''-=0(i

+ a;^)^ _ (i_ a;2)i + 3x)^-(l + 4x)i'

14. 15.

Prove that

x)^

e^.

Expand logc(l 4- x + x^) in powers of x to the term For what values of x will this expansion converge ?
16.

involving

x*.

Show

that

ioge=^?^^-v^^^^y+V'^^^y----.
7l2

17.
18.

Show
Show
n

that logg
n2

=
1

1
1

1
\-

[-.

n2

2 n*

3 n6

that

+1

2(n+l)2

3(n

l)3

2n2

Sn^

XXXV. RECURRING SERIES

995
in which Recurring series. A series a^ + n^x + Ogo;^ + every r + 1 consecutive coefficients are connected by an identity

of the

form

where pi, p^, , Pr are constant for all values of n, is called a recurring series of the xth order, and the identity is called its
scale of relation.

Thus, in

3x

5 x^

7 x3

(2

71

1)

x"

(1)

every three consecutive coefficients are connected by the formula

a for

- 2a_i +
3

5-2-3 + 1 =

0,

7-2-5 +

a_2 = 0; (2) = 0, 2n + 1- 2(2n -1) + 2n - 3 = 0.

ix

a<ixP-

562

A COLLEGE ALGEBRA
the example
it

Fiom

terms are given the series

will be seen that ordinarily may be continued in one

when 2
way

as a

recurring series of the rth order, and that when 2 r are given it may be continued in infinitely many

1 terms

ways

as a

recurring series of the (r

l)th order.

998

The generating function of a recurring power series. Every recurring power series of the rth order is the expansion of a proper fraction whose denominator is of the rth degree (comThis fraction is called the generating function pare 996).
of the
series.

It is the

sum +

of the series

when

the series

is

convergent.
Thus,
let

ao

a^x

^2x2

ax"

(1)

be a recurring series of the second order

'ivhose scale of relation is

Set

S=ao+ aix+ aox^H ..pxS= j^aoX + paix^H

..qx'^S=
qaax"^-]

+ pan-i + +
\-

qar,

_2

= 0.
+ pa-iX'

(2)

a_ix-i
2)ffl_2X"-i

qa-3X''-^+ qan-2X" + qan-^X^+^

..{l

+ px + qx^)Sn = ao+{ai +

pao)x

-\-{pan-i

qa-2)x'>-

+qa-ix"+^

the remaining terms on the right disappearing because of (2). When (1) is convergent, as n increases Sn will approach S, the of (1), as limit, and x" will approach 0.

sum

Therefore
that
is,

(I

+ px +

qx'^)

S=

ao

+
1

{ai

+ poo) x,
^
'

g^ao + {m+pao)x^
+ px +
gx2
of a recurring

999

power series. This may be generating function is known, by the method illustrated in the following example.
obtained,

The general term

when

the

Example. Find the generating function and the general term of the a_i 2 a_2 = and whose first two recurring series whose scale is a terras are 5 + 4 x.

Here
Therefore, by

p=

1,

=
1

2,

ao

5,

ai

4.

998, (3),

S =

^~\ x2 X 2

=
,,

(1

^ x) (1 + ^j/

,, , 2 x)

(1)

RECURRING SERIES
Separating
(1)

563

into its partial fractions, 537,

^~-^
(l

+ x)(l-2x)

1+x

= l-2x
-

2(l

a;)-i

+ 3(l-2x)-i.

+ (_l)na;n + ...]. But if |X|<1, 2(1 + X)-' = 2[1-X + X2 And if |x|<l/2, 3(l-2x)-i = 3[l + 2x + 4x2 + ... + 2"x + ].. Therefore the general term is [(- 1)2 + 3 2"]x''.

EXERCISE XCI
1.

If the first

three terms of a recurring series of the third order are

is

_ 3x +
2.

5x2 and the scale of relation

+ 2a_i

a_2

+ 3a_3 =
in

0,

find the fourth

and

fifth

terms.

Find the scale of relation and two additional terms

:

each of the

following
(1) (2)

2
1

(3)
3.

+ 2 x2 - x3 - 3 x* + + 4 x2 + 7 - 26 X* + -3x + 0x2 -10x3 + 15x* -21x6 +

3X
.

5X

x-'^

..

Find the generating function and the general term of each of the
:

following
(1) (2)

+ X + 5 x2 + 7 x3 + 17 X* + + 7x + 17x2 + 43x3 + l]3x* +

4.

Prove

that in a recurring series ao

is

whose

scale of relation
tto

function

(ai

is

of the third order, + diX + + pa_i + ga_2 + ra_3 = 0, the generating + pao) X + (a2 + pcn + 9o) x^ 1 + j)X + gx2 + rx^

5.

By

aid of the preceding formula find the generating function

and

the general term of the series

1
6.

+ 2x +
a

11x2 _i.24x3

85 x*

+ 238x5+
+
(a

..

Show that Show Show

+ +
1

(a

ci)

(a

2 d) x2
its

3 d) x^

is

a recur-

ring series of the second order, and find

7.

generating function.

that 12

22x

+
2

32x2

+
+

4flx^

is

a recurring series of the

third order whose generating function

8.

is (1

+
4

x)/(l
5 x^

x)^.

that

3x

4 x2

is

recurring

series of the third order,

and

find its

sum when

convergent.

564

A COLLEGE ALGEBRA

XXXVI.
1000
Infinite products.

INFINITE PRODUCTS
This name
is

given to expressions of the

2)---(i

form

n(i
in

,)

= (1

4-

ai)(i

.),

which the number of the factors is supposed to be infinite. Such a product is said to be convergent or divergent accordas
(1

ing

a-i)

(1

a^

(1

+
is

a)

approaches or does not

approach a
1001
Theorem.

finite limit as

indefinitely increased.
a^

If

all

the

numbers

are positive,

the

infiiiite

product

n (1 +

a^) is

infinite series 2a, is

convergent or divergent according as the convergent or divergent.

is

First,

suppose that Sa^

since 1
(1

convergent and has the sum

is

S.

Then

a-

<

e"^

when x

positive, 990,
e"^

we have
that
is,

a^) (1

+ a^) (!+ a) < < ei + + + n <

e"^---

e"-,

(>s^

Hence, as n increases, (1 + a^) (1 + Oo) (1 + ) increases It therefore but remains less than the finite number e'^.

approaches a limit,

192, that

is,

11

(1

a,.)

is

convergent.

Second, suppose that la,, is divergent. In this case lim (^i + Oo + + ) = co.

But

(1

i) (1

a)

(!

+
-\-

fl)

>

1 4(ai

Oo

+
co,

Therefore lim (1

Oj) (1

a^)

(1

a)

+ ).
is,

that

n (1

-t- a,.)

is

divergent.
l/nf)
If
is

Thus,

n (1 +

convergent when

p>

1,

divergent

when p < 1.
all

Example L
less

Sa^

is

a divergent positive series whose terms are

than

1,

show that

11 (1

a^)

0. \

Since a^

<

and
aj) (1

Hence But

(1

af. < - az)

1, (1

we have

lim

(1

Therefore

n (1 -

a,)

+ ay) (1 = lim (1

ar< 1/(1 + a,) numerically. + a^) (1 + ag) (1 + a). + ch) (!+ a) = co. - ay) (1 - 02) - a) = 0. (1
a,)

<

1/(1

INFINITE PRODUCTS
Example
verges
if

565

2.

Show

that

when x =
if

the binomial series, 987, con-

m + 1 > 0,
X

but diverges

m + 1 < 0.
l)

When
l

=
,

the binomial series becomes

+ + "M^-^)+... + "^('^^-l)---(^-" 1-2 l-2-.n

we have
u + \

....

^ ' (1)

In this series

m-n +
n
have
|

__ J

m+l
'

Un

Hence,

But
For
that

if

if m + 1 < 0, we m + 1> 0, after

u,, + \/Ua > 1, and (1) is divergent, 951. a certain term the series will be of the kind
\

described in
if

958 and will therefore converge.

r denote the first integer greater than

is

m+
= 0.

I, it

follows from (2)

when n>r, u + i/w

(1) after

negative and numerically less than

the terms of

Ur are alternately positive

(1)
1

1. Hence and negative and decrease

numerically.

Therefore

converges

if

limu

But

m
Un +
1

7n

(-)-0-^)(>-"^)-(-'^)and
it

follows from Ex.

as limit.

that as n increases the product on the right

approaches

EXERCISE XCII
1

1.

ci,

Show

3 . that 2
*!,

5
'^

17

,5 10 and
4 9

17

26

16

16

25

are convergent.

2.

For what positive values

?

of x are the following infinite products

convergent

x3\ 32/

<)

"(^ +!-:)=(
that lim
<'l<'

-D('-f )('-!;) +
^)---'

3.

Show
or

ni<^

n)

, according as

a>

a<o.

566

A COLLEGE ALGEBRA

XXXVII.
1002
Continued fractions.

CONTINUED FRACTIONS
This name
,

is

given to expressions of the

form a
written.

+d

or

+
,

+
,

as they are usually

We

shall consider simple continued fractions only.

a, A

These

have the form

or

+
,

,
;

where

a^

is

a positive integer

-.

and a^, a-s, The numbers

are positive integers. are called the first, second, !, Oo,

partial

quotients of the continued fraction. infiAccording as the number of these quotients is finite or
nite, the fraction is called

terminating or no7iterminatlng.

1003

Evidently every terminating simple Terminating fractions. continued fraction has a positive rational value, for it may be

reduced to a simple
Thus,
2

fraction..

3^^

4 - = 30 -;

13
3^'

be converted Conversely, every positive rational number may This will be into a terminating simple continued fraction.
evident from the following example.
Example.
Applying
29)67(2
58
integers to 67

Convert 67/29 into a continued fraction.

tlie

method for finding the greatest common and 20, we have

ai

divisor of

two

9)29(3
27

a2

67 ^ 29
a3

9 _ = 2+;^ = 29

+ -.
29/9

(1)

2)9(4

=
...f
a4

3+?=3 + ^.
2"

(2)

1)2(2

=^+

^^^

CONTINUED FRACTIONS
Substituting
67
(2) in (1),
1

'

667

and

(3) in

the result, 3
1

we have
-,as
2
1

=+2
29

=2 +
1

required. ^

4+1
2
Since 29/67

1 -^

67/29,

we

also have
1 1
1

29_

67~2 +
Convergents. are called the
1

2"

1
Tlie
first,

fraction

:r'i"l

Uo

?!

+ 4- ^5 o + Oo "3
c-2

11

1004

second, third,

convergents of the fraction

Oi H 2

3 H

When

aj is 0, the first

convergent

is

written

Theorem 1. Each odd convergent is less and each even con- 1005 vergent is greater than every subsequent convergent.
This follows from the fact that a fraction decreases when
its

denominator increases.
Thus,
1.

ai<at +
! H

2.

11

>ai
1

-\

since

.11
>
a2
,

a^

a2

02

3.

ai

-\

02

<ai-\
1

since ^2 H

.11
+

> a2

as

az-\

02 H
as

as

by 2

and so

on.

Reduction of convergents.

On reducing

the

first,

second, third, 1006

convergents of aj H

fractions,

we

obtain
1
.,

+ + +

to the

form of simple

^'l^2

1
,

<'\^l2^!\

^3
,

....

/^N

aj

"2*^3

(1)

568

A COLLEGE ALGEBRA

denote tne numerators, and q-^, q^, q^, Let pi, P2, pz, the denominators of the convergents as thus reduced, so that

Ih
<1\

= i? I5
a.^,
?i

Pt
<l-l

= i2 + = 2,

^,

Pz

= a^a^a-z + ^i + = 23 + 1, qs

3,

(2)
(3)

Since

a^,

a.3,

(3) that as

are positive integers, it follows from (2), increases j9 and q continually increase, and
00 if

that they approach

the given fraction does not terminate.

(3) it will be
qs

By examining
i>3

(2)
3i>2

and

found that

+ Pi and

03*72

qi.

(4)

This
1007

is

an illustration of the following theorem.

Theorem 2. The numerator and denominator of any convergent are connected with those of the two preceding convergents by
the formulas

Pn

a^Pa-l

+ Pn-2>

qn

anq _ i

q_2.

For suppose that these formulas have been proved to hold good for the kih. convergent, so that
Vk

(^kPk

-l+lh-2,
Pk
qk
__^

<lk

^klk -l

7k-2,

(1)
^2)

(hPk-\
(^k'lk-i

+ Pk-2 + 'Ik-i
be derived from the

The

{k

l)th convergent

may

Hh

by

merely replacing a^ by %. + l/% + i) Pk-\i Pk-2J 'ik-v 9k -2 do not involve

Pk+i
ik +
1

1004.

a^., it

Therefore, since follows from (2) that

^ + ^/ak + ^)Pk-i +Pk-2 + 1/X- +l)'Jk-l + 9k-2 ak+i(fhPk-i + Pk-2) + Pk-i ^ ak + iPk+Pk-i ^ ak + i(ak7k-i + 9k-2) + qk-i O'k+iUk + lk-i
((^k
{('k

by a)'

that

is,

Pk+

-\Vk+Pk- V

<ik

+ i=-- t + 1 Ik
7),,

q^

We have thus proved that if the formulas = a^q_i + q_2 hold good for any particular

+ 1k-i= a^_

+Pn-t>

convergent,

CONTINUED FRACTIONS

569

But we have they hold good for the next convergent also. already shown that they hold good for the third convergent. Hence they hold good for the fourth, hence for the fifth, and
so on to every convergent after the thh-d (compare

791).

Example.
Since 3
92

Compute the convergents

3/1 and 3

of 3

2+3+4+5
3,

1/2

7/2,

we have pi =

p2

7,

qi

1,

2.

Hence i53 = 3 and gs = 3

.

+ 3 = 24, j)4 = 4 + 1 = 7, 54 = 4
1

24
7

, ^, Therefore the convergeuts are -

^37

+7= +2=
24
7
,

103, ps
30, 55
>

=5 =5

103

30

+ 24 = 539, + 7 = 157.

103

539
157

30

Theorem

3.

The numerators and denominators of every two 1008

consecutive conver(je7its are connected by the formula

Pnqa-l

Pn-iqn

= (-

!)"

The formula holds good when n 1006, we have poqi 2^\1'i (i2 + 1) ~ f'1'^2 Moreover we can prove that if the formula holds good when n = k, it also holds good when n = k + 1.
For
pi^+iqj,

= 2, For, by = 1 = ( 1)^

-ihik+i

= (((k+iih- +ih-dik -Pki^k+iqk + qk-i) = -{Pkqk-x-Pk-xqK)1007 = (+ '.

Hence,
tl^en

if

p^q^ -i-Pk-iQk^i- l)^

p^ + iqk- Pk^k +
is

!)'

Therefore, since the formula 1 or 3, therefore for w 2

true for

= 2,

it is

true for
to

1 or 4,

and so on

any

positive integral value of n.

Corollary
1.

Eueri/ convergent
q

-p^^/q^

is

an
it

irreducible fraction. 1009

For

if p,^

and

had a common
1

factor,

would follow from

factor
is

the relation ^'7_ jo_j2^ =( 1)" that this divisor of ( 1)", which is impossible.

570
1010
Corollary 2. the fornmlas

A COLLEGE ALGEBRA
For
the differences between convergents

we

have,

1.

Pn qn

Pn^l
qn-1

(-1)
qnqn-1
is

"' 2

E^ qn

j^'
qn-2

(~^)"^^
qnqn-2
rela-

The
tion

first
1

formula

JMn -

- Pn -!'/=({'^nPn 1

an immediate consequence of the

!)"

The second follows from the

Pnin -2-Fn-1<ln=

fact, 1007,
^In

1008, that

+ Pn - 2)

-^-pn-'l iS^n'ln -X^^n-^

=
The theorem
1011
of

(i>-iy-2-2^n-2g'n-l)

= (-

~ 1)"
^.

1005 may be derived from these formulas.

Theorem 4. The nth com-ergent of a nonterminating simple continued fraction approaches a definite limit as n is indefinitely
increased.

form a For, by 1005, the odd convergents Pi/qi, Ps/qs, never-ending increasing sequence, every term of which is less than the finite number p<i,fp_. Hence, 192, a variable which

runs through this sequence will increase toward some number

as limit.

even convergents

Similarly a variable which runs through the sequence of will decrease toward some p^^/q^^ Pi/q^j
fj.

number

as limit.

But/A-=A,since/*-A=

lim

[i^-P?jn^~]

= Vim

(-1)''"'
5'2ni9'2m-l

Therefore a variable which runs through the complete will approach sequence of convergents Pi/qi,P2/q2, Ih/qs, X as limit.

1012
is

By

the value of a nonterminating simple continued fraction

CO

meant the number l\m (p/q)' It follows from 1003 that n= this number is always irrational. The value of a terminating fraction is that of its last convergent,

1004.

CONTINUED FR ACTIONS

571

In the statements of the following theorems, 1013, 1014,

the understanding
is

that

when

the fraction terminates, " conlast one.

lies

vergent"' means any convergent except the

Corollary
I.

The value of a simple continued fraction

Tlie difference between the value of
tltat

1013

between the values of every two consecutive convergents.

Corollary 2.

a continued 1014
less

fraction and

l/^nqn + i

nth converijent is numerically ^i^ greater than a + 2/qnqn + 2of

its

than

For

let

A denote the value of the fraction, and to

Ji

fix

the

ideas suppose that

is

odd.

We

then have

^'
'la

< ^^^^' < A < ^^^^</u

1005, 1013

+2

'Li

Hence
and

-^"
?

<^-^^^
? +
!

-^S
?

.-.

< -^
'/,/In

1010,1
1010,2
rela-

),^I!ji-^I!2Li_lji, 'In '/ 'In + 2

i

,.,^Jbi^.
Qnln +
i

Evidently l/iq +
tion y
_^2

<

'^-/'il,

and by making use of the

it

+ + 2A/7 + 2^ !/'/(// +
+ oy +
1

'i,

1007,
\)-

may

'L,

Hence

readily be shown that the difference between

X and p/q

is less

than l/yl and greater than l/'i(q

q +
to

i)-

Each convergent is a closer approximation value of the fraction than is any preceding convergent.
Corollary 3.

the 1015

For, by 1014, if A. denote the value of the fraction, the difference between X and 7?,yy is numerically less than l/qqn + n

while the difference between X and 2^n-i/q-\ and l/qq + 1 < greater than (/ + ,/y _ ,y +
^
;

is
((

numerically
+ i/y _ ,7,, +

since

q-i<

('n

+ \qn^

1006.

Tlie convergent Pn/<ln ^^ ^ closer approximation to the value of the fraction than is any other rational fraction

Corollary 4.

1016

whose denominator does not exceed

q.

ajb is a closer approximation to the value of the fraction than p^/q,, is, it must also, 1015, be a closer

For

if

?-

CONTINUED FRACTIONS
In general,
if

573

x denote the value of a pure recurring fra.ction

i

with the period

+
!

+
i

'

we

have,
1

1007,

re

+
i

(tk

+x

= Pk^+Pk-i
q^x

'

y^._i

and therefore
Since
it

+ (?< _ Ih) ^ -Ih-i^ 0. the absolute term p^. _i of this quadratic

q^x^

is is

has one and but one positive root, and this root

negative, the value

of the fraction.

Again,

if

y denote the value of the mixed recurring fraction

a^

-f-

i
i

a,.-\- a,

+1 H

1-

rt,

i.

we

find the value

x of the recurring part

as above,

and then

have,

1007,
2/

13r^+ 7^-1
q^x

a^

+x

q^

On converting
Every positive

irrational

numbers

into

continued

fractions.

1018

irrational

number

is

the value of a definite

nonterminating simple continued fraction which may be obtained to any required partial quotient by the following
process. If h denote the

number

in question, first find !, the greatest

integer less than b. Then h tional number greater than 1.

less

a-^ -\-

1/ii,

where
b^ is

J)i

is

some

irra-

Next
l/^2>

find Un, the greatest integer

than

b^.

Then

^^

where

some

irrational

number

greater than
1 +=
b-^

1.

Continuing thus, we have

b^

ai

a^^

11 =
(?o

...

a^-]

,11
Oo

<^2

3 H

It can be proved that when J is a quadratic surd the continued fraction thus obtained is a recurring fraction.

574
Example.

A COLLEGE ALGEBRA
Convert vll into a continued fraction.

The

greatest integer less than

VTl
3

is 3,

and,

603,

Vll =
The

+ {VII-3) =

+ -I
Vll +
3 3)/2

i
+
3)/2

(1)

greatest integer less tlian

VTI +
2

is 3,

{Vll and
3

^5i_3=3 + ^^ii:^ = 2
2

3+^l = +
(Vll
3
3)

_l_.
VTT+3
?

(2)

The

greatest integer less than

3

Vll +
6

is G,

and

VTl +

+ (Vn -3) =

=^
Vll +
3

=6 +
(Vli

(3)

3)/2

The last fraction in (3) is the same as the last in (1). Hence the steps from (3) on will be (2), (3) repeated indefinitely that is, tlie partial quotients 3 and 6 will recur. Hence, substituting (2) in (1), and (3) in the
;

result,

and so on, we have Vll

...

^3 + +

1019

given irrational number can be expressed in only one icay as a simple continued fraction. This follows from the fact
that two nonterminating simple continued fractions cannot be equal unless their corresponding partial quotients are equal.

For if a + a = c + 7, where a and c denote positive integers and a and 7 denote positive numbers which are less than 1, then a = c, since otherwise it would follow from a c = y a that an integer, not 0, is
numerically
less

than

1.

Hence,
Ci, C2, Cs,

if

aiH
a2

+
as H

Ci

-\

C2

denote positive integers,

,

=
1020

11
+
cs H

.-.

ao H '

11
...
+

where

Cg

ai, ag, as,

>

we ^ have
1

a\

Ci,

.-.

11
"2

("2

C2

as

Cg

.'.0-1

C2,

and so on.

we compute the continued fraction to which a given irrational number h is equal as far as the wth partial quotient, we can find its ?ith convergent j'nlln^ ^^^ ^l^is rational fraction
If
Vnl^ln will
1/y,^,

express h approximately with an error less than 1014. Moreover p^/qr, will be a closer approximation
is

to h than

any other rational fraction whose denominator

g-,,,

does not exceed

1016.

COXTIXUED FRACTIOXS
Thus, the
3 10 -, 1 3
first

575
3
,

four convergents

ofvll =

111
6

are
1

3 H

63
,

19

190
,

^ and

190

60

60

/ .^^ expresses Vll with an

error less than

602

^^

Given 1021 any equation of the form ax + htj = e, where a, b, c denote If integers of which a and h have no common factor, 672.
Solution of indeterminate equations of the first degree.

we convert a/b

into a continued fraction, the last convergent

of this fraction will be a/b itself, and if the convergent next This fact to the last be j^/o^, "^e have aq bp =-kl, 1008. makes it possible always to find a pair of integral values of x

which satisfy ax -{ij in the following example.

and
Example.

b//

c.

The method

is

illustrated

Find an hitegral solution

onr.

As

in 1003, Ex.,

we

find

=2 + 1111 + + +
93 ^ 4
_
1

of 205 x

+ dSy =
_
8
_.

1.

The convergents, found ^

'

2 9 11 as in 1007, Ex., are -, -, * 4 1 5

97
,

44

205 93

Hence
or,

205 44

93

97

multiplying by

7,

205 (- 44

7)

93(97

7)

= - 1, = 7.

Therefore x

=-

The general
Similarly

308, y = 679 is a solution of 205 x + 93 y = 7. solution is x = - 308 + 93 1, y = 679 - 205 1, 674.

=-

we may show

that 205 x

93 y

has the solution x

= 308,

679.

EXERCISE
Compute the convergents
1.

XCIII
:

of the following
2.

3+1 4

1. 5

1
1

^
1-

1
+
10

1.

12

Convert each of the following into a continued fraction. Eor each of the last three compute the fourth convergent and estimate the error made in taking this convergent as the value of the fraction.
3.

!2. 12

576

A COLLEGE ALGEBRA

Convert each of the following into a recurring continued fraction and fifth convergents and the corresponding errors for the first four of them.

compute the

11.

Vl7.
Vl05.
3 Vs.

12. 16.

V2t5.

13.
17.

V6.
Vl9.
21.

14. 18.

V.38.

15. 19.

I/V23.
20.

V7I.
l)/(

(Vl0-2)/2.

(V2 +

V2 -

1).

rind the values of the following recurring

i

fractions.

22.
1

i
-..
i

+ 2+3 + +
1
3

23.

i
2

24.

---

3+{ + 5 + 2+.. 4
^

25.

1 4

26.

1
2

...
1

+2 + 1 +

27.

Show Show
Show

that

Va^ +
Vu^

-a +
=
a

2a + 2a +
a

28.

that that

+-

+ 2a +
c)

+ ia +

29.

ill
a

_ ~
+

{abc

+ a-h +

y/(abc

c)^

2 (a6

30.

Convert the positive root of x^

+ 1) +x-1=
...

into

a continued

fraction. 31.

Show
Show

that

^"=^ + -i
9

32.

that

11
+
made

qi

3i?2

=
,

111
q-2qz

L+

llll)l'.

qn-iqn
.

a2
33.

as H

qiq2

q-iqz

qsqt
less

What

rational fraction having a

denominator

than 1000 will

most nearly express the

Estimate the error
34.

ratio of the diagonal of a square to its side ? in taking this fraction as the value of the ratio.
will express
tt

Find the simplest fraction which with an error which is less than .000001.
35.

3.14159265-

error
36.
37.
38.

Compute the sixth convergent of made in taking it as the value of e.

Find an integral solution of 127 x Find an integral solution of
Find the general integral
2.35

2.71828

and estimate the

214 y

=
=
-

6.

412 y
x

10.

solutiiTn of 517

323 y

81.

PROPERTIES OF CONTINUOUS FUNCTIONS

577

XXXVIII.

PROPERTIES OF CONTINUOUS FUNCTIONS

FUNCTIONS OF A SINGLE VARIABLE

If the variable y depends on the variable x in 1022 Functions. such a manner that to each value of x there corresponds a
definite value or set of values of y,

we

call
is

In what follows when we say that y

y Zs function of x. a function of x and

write y =f(x), we shall mean that it is a one-valued function; in other words, that to each value of x there corresponds but

one value of

y.

And f{a)
a function of x
x"^

will denote the value of

x.
it

y which

corresponds to the value a of

Evidently
in X, as
jr

is

if

when y =

\.

But a

relation

be equal to an algebraic expression which defines y as a function

of X may be one which cannot be expressed by an equation. Thus, y is a function of x if ?/ is 1 for all rational values of x and 1 for all other

values of
equation.

x.

But

this relation

between y and x cannot be expressed by an

call y a function of x even when there are exceptional values of x which the given relation between y and x fails to determine y, 1024. Sometimes y is defined as a function of x only for a certain class of values of x or only for values of x which lie between certain limits. Tims, the equation ?/ = x + 2x2 + 3x^ + , by itself considered, determines y for those values only of x which are numerically less than 1.
foi

We

Let f(x) denote a given function 1023 Continuity of a function. of x. a, say that /(a;) is continuotis at a, that is, when x

We

if

what follows the notation ^^^f(x) = f(a) means that f(x) will approach f(a) as limit whenever x approaches a as limit, that is, no matter what the sequence of values may
Here and
in

^^^ /(a) has a definite finite value, and if f{x)=f(a). In the contrary case we say that f(x) is discontinuous at

a.

be through which, x runs in approaching a as limit. In the case of a function y defined by a given equation 1024 y=f{x) it may happen that the expression /(x) assumes an

578

A COLLEGE ALGEBRA
when x

indeterminate form

a,

513-518.

The equation
i/

y =f(^x) by itself considered then fails to define when x = a. But if l^^/(ic) has a definite finite value, we assign this as the
value of f((i),

519,

which makes f{x) continuous

oo,

at a.

If
is

li;/(a;)=

00,

we

assign to fUi) the value

a.

515; f(x)

then discontinuous at
nate,

we

Finally, if ^^^ f(^) is indetermihave no reason for assigning any single value to

/(a).

^i^ Evidently we can assign none for which f{^) =/(^)In this case also f(x) is discontinuous at a.
1. Thus, every rational function/(a;) is continuous except perhaps the denominator of some fraction occurring in /(a;) vanishes.

r
; '

when

For example, consider the function /(x)

This function
is

(x

or

1.
X

continuous except when x- 1 For if a is not 1 or 1, /(a) = (a

l)/(x2

1). = 0, that is, when l)/(a- 1) has a

definite finite value

and

lifi

/(x) -/(a),

509.

When
form
x
0/0.

1,

the expression (x

But] /(x)

l)/(x2 1) assumes the indeterminate - l)/(x2 - 1)] = J [l/(x + 1)] = 1/2, [(x
we make
for

and by assigning

to /(I) the value 1/2

/(x) continuous

when

= l. When
2.

1,

/(x)

is

discontinuous
:

/(x)

oo.

Consider the following function

1
.,
^

/W - -\ +
2^

3 _
1

3/2"^
i

1/2

_]
i -

3
1

2^+1

1/2^

1/2

Here/(0) has the indeterminate form co/co, 517. But if we write /(x) in the second form and then make x approach
1

through posifjue values, we have lim 2-^ = co, and therefore lim/(x) = 1. If we write /(x) in the third form and then make x approach through
negative values,
Finally,
positive
if

we have lim 2 ^ = oo, and therefore lim/(x) = 3. we make x approach through values which are alternately
for

and negative, /(x) will not approach any limit. Hence /(x) is discontinuous at 0. No value can be assigned to/(0) which
''

]/(x)=/{0).

1025

Prom
follows,

the definition of continuity in 189, that

V

1023

it

immediately

PROPERTIES OF CONTTXUOUS FUNCTIONS

The
sufficient

579

and necessary condition that f (x) he continuous that f (a) have a definite finite value, and that for every positive number S luhich can he assigned it shall be possible to
at a
is

find a corresponding
|f

j^ositive

number

such that
|x

(x)

(a)|

<8

whenever

a|

<

e.

Thus in the neighborhood of a value of a-, as a, at which f{x) is continuous, very small changes in the value of x are accompanied by very small changes in the value of fix), and the change in the value of x can be taken small enough
to

of f{x) in an interval containing a value of x at which fix) is discontinuous. See the examples in 1024.
as

make the corresponding change we please. This is not true

in the value of /(cc) as small

Theorem
tinuous at
also of f

1.

If both of the functions

the

f (x)
</>

a,

same

is
(j>

true of f

(x)/^ (x) unless

is

= (a)
a,

(x)

(x)

and </)(x) are con- 1026 and f(x) <^ (x)^

0.

Ifi(x)

continuous at

the

same

is true

of vf

(x).

This follows immediately from the definition of continuity and the theorems of 203-205, according to which lim \f(x) + ^(^)] = lim/(a;) + lim ^{x), and so on.
at a, 1023,

Real functions.
that
is,

In what follows x will denote a real variahle, 1027

one which takes real values only, and f{x) will denote a real function of x, that is, one which has real values when x
is real.

Number intervals. The practice of picturing real numbers by 1028 points on a straight line, 134, 209, suggests the following convenient nomenclature.
Let us call the assemblage of all real numbers between a and b, a and h themselves included, the number interval a, b, and represent it by the symbol (a, b'). Moreover, it being understood that a Kb, let us call a and b the left and right extremities of the interval (a, b). Also, if

580
c

A COLLEGE ALGEBRA
say that
(a, h) is

= (a + i)/2, let us
(a, c)

divided at
on.

into the

two

equal intervals
Thus,
(1, 7) is

and

(c,

b)]

and so

divided at 4 into the two equal intervals

(1, 4), (4, 7);

and

at 3

and 5

into the three equal intervals

(1, 3), (3, 5), (5, 7).

1029

We

interval (a, b) interval.

say that the function f{x) is continuous throughout the if it is continuous for every value of x in this

1030

Theorem
a7id f (a)
Xfj

2.

Ifi (x)

is

and f (b) have contrary signs,

at f (Xy)

continuous throughout the interval (a, b), there is in (a, b) a number

such

til

0.

To

fix

Divide

the ideas, suppose that /(a) is + and that/(&) is into any number of equal intervals, say into the (a, b)
.

two equal intervals (a, c) and {c, b). Iif{c) =0, our theorem is proved, c being Xq. But iif(c) ^ 0, it must be true of one of the intervals (a, c) or (c, b) that /(a?) is + at its left extremity and at its right. Thus, if f(c) is this is true of (a, c), and if /(c) is +, it is true of (c, b). Then Select this interval and for convenience call it (a^, bi). is + and /(b^) is -. /(ai) Deal with this interval (a^, b^) as we have just dealt with and so on indefinitely. We shall either ultimately come (a, b),
,

upon an interval extremity

the Xq sought, or

for

which f(x)

0,

which

is

then

we

shall define a never-ending sequence of

intervals within intervals,

(a, b),
(ai, b^), (a^, ho),
, (r7, b),

,
are

such that

f(a), f{aC), f{a^),

f{b), f{h,), f(b,),

,
. .

f(a),
.

and
It follows

.,f(b),

are -.

from 192, 193 that as 7i increases a and b approach the same number as limit. For remains less than b and never decreases, and ^ remains greater than a and never a) = lim (b a)/2" = 0. increases, and lim (b,^
Call this limit
Xq.

Then /(xq) =

0.

PROPERTIES OF CONTINUOUS FUNCTIONS

For since/(a;)
is

581

continuous at

x^,

lim/(o)

= lim/(^) =f(x^.

since /() is always positive, its limit /"(xo) cannot be negative, and since f{h,^ is always negative, its limit f(x^

But

cannot be positive.
Thus,
if

Therefore
a;V2
!

/(j;'o)
!

is 0.

fix)

=1-

a;V4

x^/Q

shown

that /(I) is positive and /(2) negative. vanish for some value of x between 1 and 2.

!+, it may readily be Hence this f{x) will

Simpler illustrations of the theorem will be found in 833, 836.

Maximum
2 11

and

minimum
11
is
...

values.

Superior and inferior limits.

:

1031

Consider the following infinite assemblages of numbers

11

/A^

2 21

23

27

... (V,\

In (A) there

number

and in (B) there no greatest number.

;

a greatest number, namely 2, but no least is a least number, namely 2, but

On

among

the other hand, while there is no least number in (A), the numbers which are less than those in (A) there is a

namely 1. Similarly among the numbers which are greater than those in (B) there is a least, namely 3. The like is true of all infinite assemblages of finite numbers,
greatest,

that

is,
h.

of

numbers which
In other words,

lie

between two given

finite

numbers

a and

Theorem 3. Let ai, a^, , a, (A) denote any infinite 1032 assemblage of finite numbers. Then 1. Either among the different numbers in (A) there is a
greatest or
is

among

the

numbers greater than those in (A) there

a
2.

least.

or

among

Either among the different numbers in (A) there is a least the numbers less than those iti (A) there is a greatest.
1 assign all
all

To prove
a class
/?2,

and
i?i.

to a class

numbers greater than those in (A) to other real numbers, including those in (A), Since each num})er in E^ will then be less than

every number in R^, there will be, 159, either a greatest numwhich means either a greatest ber in Ri or a least in R2,

/"

582

A COLLEGE ALGEBRA
the different numbers in (A) or a least

among

among

the num-

bers which are greater than those in (A). By similar reasoning 2 may be proved.

1033

among the different numbers of an assemblage there is we call that number the maximum number of the assemblage if a least, its minimum number. The superior limit of an assemblage is the maximum number, if there be one. If not, it is the least number which is greater than every number in the assemblage. The inferior limit of an assemblage is the minimum number, If not, it is the greatest number which is lesa if there be one. than every number in the assemblage. An assemblage like 1, 2, 3, 4, which contains numbers greater than every assignable number is said to have the supeIf

a greatest,

rior limit 00.

4,

Similarly an assemblage like

is

said to have the inferior limit

if

1, 2, 3,
cc.

Evidently,
either \
is its

an assemblage has a
or
differ

finite superior limit X,

maximum number

we can

find in the assem-

blage

numbers which

from X as

little as

we

please.

1034

" By the " values of /(cc) in (a, h) we shall mean those which correspond to values of x in (a, h). And if this assem])lage has a maximum or a minimum value, we shall call it the abso-

Tnaximum or yninimum. value of f(x) in (a, b). The maxiand minimum values defined in 639 may or may not be the absolute maximum and minimum values.
lute

mum
1035

Theorem
(a, b),
it

4.

If

f (x)

is

continuous throughout

the interval

has an absolute
(a, b).

maximum and an

absolute

minimum

value in

For since the values oif(x) in (a, b) are finite, 1023, they have finite superior and inferior limits. Call these limits X and {x respectively. We are to demonstrate that in (a, b) there is a number K ^^^^^ ^ number Xi such that f(x^ = /*. Xq such that f(^o)

PROPERTIES OF CONTINUOUS FUNCTIONS

583

As the proofs of these two theorems are essentially the same, we shall give only the first of them. Divide (, b) into any number of equal intervals, say into
two such
intervals. Evidently A will be the superior limit of the values of f(x) in at least one of these half intervals. For convenience call this half interval (ui, bi).

(a, b),

Deal with the interval (i, and so on indefinitely.

b^)

as

we have

just dealt with

We

thus obtain a never-ending sequence of intervals within

(a,b), {ai,b^), (a^yb^),

intervals,

(a, b),

in

each of which A

is

the

superior

limit

of

the

values

of f(x).

As 71 is indefinitely increased, a and 5 approach the same number as limit (see 1030). If we call this limit .r^, then /(.r,,) = A.
For
if not,

difference
so that

since both /(x^) and A denote constants, their must be some constant, as a, different from 0,

X-f{Xo)=a.

(1)

(a, b)

Since f(x) is continuous at x^, we can make the interval so small that for every value of x in (a, Z) we have,

1025,

\f(x)-f(x,)\<a/2.

(2)

And
(a, b),

since A

we

the superior limit of the values of f(x) in can choose x in (a, b^^) and (2) so that, 1033,
is

A -f(x)

< <

a/2.

(3)
(3) that (4)
is

But

it

will then follow

from

(2)

and
a.

A -/(xo)

Therefore, since (4) contradicts (1), (1)

false

that

is,

\ /(.To)

0, or

= /{xq),

as

was

to be proved.

584
1036

A COLLEGE ALGEBRA

If f (x) is continuous throughout the interval (a, b), Corollary. will have in (a, b) every value intermediate to its maximum and minimum values in this interval.
it

For

let c

function /(.t)
If/(.r|,)

denote the value in question and consider the c, which is continuous in {a, h), 1026.

and/(xi) denote the absolute

{a, b),

maximum and

mini-

c is + and c is /(Xq) f{xi) Hence, 1030, between Xq and x^ there is a number, call it x^, such that/(j-2) c 0, ov f(xn) = c, as was to be proved. 1037 Oscillation of a function. By the oscillation of f(x) in (, h)

mum

values of f{x) in

is

meant the difference between the superior and

h).

inferior limits

of the values of /(.r) in [a,

1038

Theorem

5.

Let f(x) be continuous throughout

(a,

b).

If

any positive number a be assigned, however small, it is possible to divide a finite member of equal inter^'als in each of (a, b) into which the oscillation of iix) is less than a.
For divide (o, A) into any number of equal intervals, say into two such intervals, each of these in turn into two equal

The process must ultimately yield interintervals, and so on. vals in each of which the oscillation oifix) is less than a.
if not, there must be in {a, b) at least one half interval which the oscillation of /(.t) is not less than a in this, in turn, a half interval in which the oscillation of f(x) is not less than a and so on without end. Let this never-ending sequence of intervals within inter-

For

in

vals be

(a,b), (i,

Z>i),

(ao,b^), , (a,b),

,
it

lim b,, and, as in 1030, let lim a Since f(x) is continuous throughout
lute

.r,,.

(a, b),

has an abso-

maximum and

an absolute

intervals (a,

b), (a^, bi),

minimum

value in each of the

1035.
/(/8)

(a, b), ,

Let f(a) denote the absolute

absolute

maximum and
(a, b).

the

minimum

value of /(.r) in

PROPERTIES OF CONTINUOUS FUNCTIONS

Then, by hypothesis, /() -/(A.)> lim f(/3)^ and therefore ]im/(<r)
^^
a.

585

But this is impossible. For since a and /3 are and lim a = lim b = x^, we have lim a- = lim /S =
Therefore, since /(.r) is continuous at x^, lim/(a-) that is, lim/(n-) 0, .". not > a. lim/(/3)

in (a, b),
a-^.

= lim/(/3);

FUNCTIONS OF TWO INDEPENDENT VARIABLES

Functions of two variables. We say that the variable u is a 1039 function of the variables x and y when to each pair of values of X and y there corresponds a definite value or set of values of u.
shall confine ourselves to the case in which to each pair of values of x, y there corresponds a single value of u. The notation u f(x, y) will mean that u is a function of

We

X and y, and f(a, x= a and y = h.

Thus, u
is

h) will

mean

the value which u has

when

a function of x

when X = 1, ?/ = 2, we have u /(I, 2) = 1 4 + 1= 2. The note at the end of 1022 applies, mutatis mutandis, here

and y

if

n =/(x,

t/)

x^

_ 2y +

1.

Here,

also.

Let f(x, y) denote a given 1040 Continuity of such a function. function of x and y. say that /(a;, y) is contimwus at a, b,

We

that

is,

when x
\i

value and

= b, ii f(a, b) has a definite finite f(x, y) will always approach /(, b) as limit when
a and y

X and y are made to approach a and b respectively as limits. In the contrary case we say that f(x, y) is discontinuous at
a, b,

that

is,

when x

a and y

b.

From
The
ous at

this definition

and

189,

it

immediately follows that

1041

sufficient
a,

and necessary
(a,

coyidition that f (x, y) be continu-

is

that f

every jjositive number

to

b) have a definite finite value, and that for S tvhich can be assigned it shall be possible
t

find a corresponding positive number

(x,

such that
e

|f

y)

f (a, b)|

<

whenever

|x

a] <

and

[y

b|

<

e.

586
1042

A COLLEGE ALGEBRA
1.

Theorem
f (x,

If both of the functions

a, b,
tlte

(x,

y) and

<}>

(x,

y) are

continuous at

y)

</)(x,

true of f (x, y) <^(x, y) tmless <^(a, b) y), also ofi(x, y)/</>(x, y),
is

same

and
0.

If

f (x, y) is

continuous at

a, b,

the

same

is

true of

Vf (x,

y).

1043

This follows immediately from Number regions. In what follows

1040 and 203-205.

it is

to be understood that

X and

ij

denote real variables, and f(x, y) a real function of

these variables (compare

1027). 382, pairs of values of x and y may be Evidently, if employing this pictured by points in a plane. method we draw the lines wh. :h are the graphs of the equa-

As

is

shown

in

tions X := a, X

^ h,

c,

d,

384, the rectangle bounded

by these
X,

lines will contain the graphs cf all pairs of values oi such that a<x<h, c<y<d. With this rectangle in mind,

we

shall call the assemblage of all such pairs of values of x,

;

the number region (a, b 1044 say that f(x, y)

c,

dy

We
b
;

(a,

c,

d) if

it is

is continuous throughout the regio-A continuous for every pair of values of x, y

in this region.

1045

Theorem
c, d), it

2.

has a

maximum and

IfHy^, y) ^' continuous th roughout the regio7i(a,,hf a minimum value in this region.

Since f{x, y) is continuous throughout the given region, its values within this region have finite superior and inferior limits, 1032, 1040. Call these
limits X

and

/u..

77

77,

Cn
c,

are to prove that in (a, b d) there is a value pair .To, y

;

We

^^^^ ^i"'^" '^ that/(a:'o, ^o) lar reasoning will show that there

such

is

also a value pair

x-^,

y^

such

that/(.ri, yi)

/A.

EFGH

For construct the rectangle which pictures the number region (a, b c, d), 1043.
;

PROPERTIES OF CONTINUOUS FUNCTIONS

By the values of /(.x, y) in of f(x, y) corresponding to
''

587

EFGH"
all

we

shall

mean

the values

pairs of values of x, y in

(a,

c,

d).

EFGH into four equal rectangles as in the figure. Evidently A will be the superior limit of the values oi f{x, y) Call this quarter in at least one of these quarter rectangles.
Divide
rectangle E^F^GiH^. Deal with the rectangle EiFiGiHi as

we have

just dealt

with

EFGH, and

so

on

indefinitely.

We

thus obtain a never-

ending sequence of rectangles within rectangles,

EFGH, E,F,GJI
in each of

E^F^G^H^, .,

(1)

which A

is

the superior limit of the values of

f(^, y)-

Let

a denote the abscissa of E^,

and

c its

ordinate.

As
and

is c

proved in
If lim a

1030,

when n
limits.
c^

is

indefinitely increased a

approach definite

= Xq

and lim
A

== y^, then/(a;o, tj^

y)

= X.
(2)

For

if not, let

f{x^,

a.

Since f(x, y) is continuous at .Tq, ?/, we can so choose E^F^G^H^ that for every pair of values of x, y in this rectangle

we

have,

1041,
(3)

\f{x,y)-f{xy,)\<a/2.

And

since A

E^F^G^H^, we

the superior limit of the values oif(x, y) in can so choose x, y in E^F^G^H^ and in (3) that
is

- f(^;

!/)

<

a/2.

(4)

From

(3)

and (4)

it

then follows that

- fi^o,

y^

<

(5)

Hence (2) is false (5) contradicts (2). A, as was to be demonstrated. f(xo, yo)
But

and therefore

588

A COLLEGE ALGEBRA
THE FUNDAMENTAL THEOREM OF ALGEBRA

We
1046

are

now

gral equation has a Theorem.

b, c,

in a position to prove that every rational interoot, 797. proceed as follows.

We

1 + bz'" + cz'" + H Given k denote constants, real or complex, and

^

^ (z) =

1-

kz", ivhere

a complex
1.

variable;

it

is

always

z that j)ossible so to choose

|<^(z)|<

For

let the expressions for z

and

b in

terms of absolute

value and amplitude,

z

877, be
(S).

Then

= p (cos 9 + i sin 6), b = \b\- (cos (3 + i sin bz"' = p"'\b\- [cos (mO + /3)+i sin (md + 13)^ 879,881
77id

First choose 6 so that

(3

tt.

(1)

Then
since

bz""

= p^l^l- (cos tt +
cos
tt

sin ir)

and

= p'"\b\, sin tt = 0.

877,

878

Next

choose p so that,
+i
1,|p

854,
(2)

...+17,1p<|^,|p'<1.

If Zq denote the value of z

which
|

corresponds to the values

\

of 6

and p thus chosen, then

<f>

(z^)

<

1.

For since

= (l - p'"\b\) + <f,(z,) - p'"\b\ + \c\p"'^' +

Given the function

cz^

+'

+ kz^,

we

have,

235,
...

\<f>(z,)\<l

+\k\p", ..<!, by

(2).

1047
if

Corollary.
f (z)

f (z)

az"

aiz"

'

h a

does not vanish

when

= b,

v)e

can always choose

z so

that\i{z)\<\i(h)\.

For in f(z) set z = b -\- h and develop by Taylor's theorem, It may happen that certain of the derivatives f'{z), vanish when z = b; but they cannot all vanish f"{z), and so on
848.

PROPERTIES OF CONTINUOUS FUNCTIONS

since f^"\z) not vanish.

589

= n\aQ.
f{b

Let /'" (z) denote the

first

one which does

Then
and therefore

h)

=f(h) +f"(h)
l

+f'%b) ^,
fib)

^^^ = + ^.^ + ...+^.^. m\

J{b)

f{h)

n\

The second member

h of the form h that \f{l> + h)/f{b)\
Theorem.
z exists

of the last equation is a polynomial in considered in 1046. Hence we can so choose

Given

f (z)

< 1 and therefore \f(b + /0| <|/(^)| = az" + ajZ" ~ + + a ; a value

^

of 1048

for which f (z) vanishes.

/(,~) set z

For in

=x

-{-

iy,

where x and y are

real,

and

having expanded ao(x + binomial theorem, collect

likewise all

to the form f(z) = ^ (x, y) + /i/^ (.r, y), where <^ (x-, y) and (x, y) denote real polynomials in x, y, and therefore have, 232,
ij/

a^^x + iy)"~^, by aid of the all the real terms in the result, and the imaginary terms. We may thus reduce /(5;)
iy)'\

By
roots
less

855,

we can
0, if

find a positive

number, as

c,

such that the

oif(z)=
c
;

there be any, are all of

them numerically

than

and

if c'

than

c for all

values of

x,

c/ V2, evidently |,v|, or (x^ c' <x<c', y such that

c', c') the expression region ( c', r' [_(f>(x, yy i}/(x,yyy is a continuous function of x and y, It therefore has a minimum value in this region, 1042.

But

in this

number

+ y^)^, is less <y <c'.

c'

1045, say
If z,

make

Therefore |/(o)I = 0, since otherwise, |/(s;)|<|/(o)|. 1047, we could so choose z that |/(2)| < |/(.~o)f. Hence |/()|, and therefore /(), vanishes when z = Zq]
is,

= Xq, y = y^. then |/(^)| - \_4>i.^o. l/oY + ^(^o, 3/o)']^ = 0. For since |/(^o)| is the minimum value of |/()], we cannot
when x

x,

iy,,

that

Zq is

a root of the equation /(s)

= 0.

INDEX
Numbers
refer to pages

Abscissa, 138 Addition of integral expressions, 93 of numbers, 10, 19, 35, 50,71, 72 of radicals, 274
of rational expressions, 217 of series, 541

Commensurable, 37 Commutative law of addition,

22, 35, 54, 74, 534, 544

11,

of multiplication, 14, 23, 35, 54,

74

Completing the square,

187,

300

complex number, 488 Angle, circular measure of, 488

Amplitude
of

Condition, necessary, sufficient, 93 Constants, 79

Annuities, 391

Approximations, 48,

55,

453

Continuity of functions, 577, 585 of real system, 46

Assemblage,

infinite, 3

Associative law of addition, 11, 22, 35, 54, 74, 521

of multiplication, 14, 23, 35,
54, 74

Convergence of infinite series, 520 absolute and conditional, 533

limits of, 530
tests of, 523, 531

Convergents of a continued fraction, 507

Asymptote, 335

Converse, 92

Base of power, 39 of system of logarithms, 377 Binomial theorem, 256, 283, 554
Binomials, products of, 102, 253 Biquadratics, 112, 480

Coordinates, 138

Correspondence, one-to-one,
Cosine, 489 Counting, 9

Cube

root.

See Roots

Cardan's formula for cubic, 483 Chance, 409

Clearing of fractions, 118, 231 Coefficient, 80
Coefficients, detached, 99

Cubics, 112, 483 irreducible case of, 485, 490

Cyclo-symmetry, 248

Degree of equation, 111

of polynomial, 87 of product, 98

undetermined, method of, 152 undetermined, theorem of, 172,

540 Cologarithms, 386 Combinations, 393
591

Density of rational system, 34 of real system, 46

Derivatives, 460 Descartes's rule of signs, 447

592
Determinant, 494 bordering a, 605 cofactors of, 504 diagonals of, 490 elements of, 494 evaluation of, 505 minors of, 502
order
of,

A COLLEGE ALGEBRA
Equations, fractional, 111, 231, 300
identical, 89

inconsistent, 133, 146

indeterminate, 342, 575

integral, 111

495
of,

interdependent, 133, 145 irrational. 111, 288, 313

irreducible, 445
linear, 112, 139
literal,

products

506

properties of, 498

terms

of,

496
of,

111

Differences,

method

364

logarithmic, 390

Discriminant, 517
of cubic, 485
of quadratic, 304

numerical, 111, 429, 459 quadratic, 112, 298

rational. 111

Distributive law, 14, 23, 35, 54, 74

Divergence of

infinite series,

520

reciprocal, 311, 438, 487 roots of. See Roots

Divisibility, exact, 28, 155, 101

simple,

ll:i,

118

Division of integral expressions, 107

of numbers, 27, 35, 54, 73, 489 of radicals, 287 of rational expressions, 219 of series, 546

simultaneous simple, 127, 143, 508

simultaneous, of higher degree, 135, 317, 514

Division, synthetic, 166 Division transformation, 155

simultaneous symmetric, 326 solution of, 112, 1.8, 483

solution
coeffiof,

by factorization,
of, 114, 129,

by aid

of

undetermined

194, 309, 318

cients, 160, 163

transformation

436

Errors of approximation, 55
Elimination, 131, 143, 317
Evolution, 39, 56, 70, 83, 200, 276.

by determinants,
Ellipse, 334

508, 514

490
Expectation, value
of,

411

Equality,

3, 8, 32,

34, 45, 72

algebraic and numerical, 18, '^5 rules of, 13, 15, 24, 36, 54, 57

Exponents, integral, 39 irrational, 376 laws of, 57, 270, 376

rational, 279 Exprcs.sions, algebraic, 85

Equations, binomial, 313, 490 biquadratic, 112, 486

complete, 426, 448
conditional, 110
cubic, 112, 483

and infinite, 85 and fractional, 85 rational and irrational, 86

finite

integral

depressed, 427 equivalent, 117, 131

Factor, 14, 176

highest

exponential, 390

common, 196

INDEX
Factor, irreducible, 211
Infinitesimal, 63
Infinity as limit, 224, 229
Interest,

593

prime, 177, 208, 212 rationalizing, 285

Factorial n, 395
Factorization, 178, 240
Ferrari's solution of biquadratic, 486

compound, 390

Interpolation, 371
Inversions, 492

Involution, 39, 56, 76, 82, 105, 276,

Fractions, 32, 213

489 Lagrange's formula of interpolation, 373 Length, 26, 37, 66 Limit of variable, 58
Limits,

continued, 566

improper, 213
irreducible, 37
partial,

236

proper, 213
reciprocal, 219

recurring, continued, 572

superior and inferior, assemblages, 582

of

Functions, 88, 571, 58^5 defined by power series, 539

Logarithms, 39,^377
characteristic of, 381

expansion

of, 371, 548,

551

common, 379
mantissa
of, of,

381

integral, 85

modulus

559

rational, 86

symmetric, 245

natural, 558
of algebra,

Fundamental theorem
427, 588

table of, 384

Maximum,

307, 467, 582

Mean, arithmetical, 355

Graphs of ei-jnations, 139, 333 of numbers, 27, 38, 66
of variation of functions, 469

geometrical, 359 harmonical, 362

Measure, 26, 37, 65

Groups of things,
equivalence
finite

Minimum,
3

307, 467, 582

of, 1

Multinomial theorem, 408

Multiple, lowest common, 205 Multiplication of integral expressions, 97
of numbers, 14,20,35,52,72,489 of radicals, 275 of rational expressions, 218

and

infinite,

Homogeneity, 87, 99 Horner's method, 453

Hyperbola, 335
Identities, 89

of series, 545

Imaginaries, conjugate, 295 Incommensurable, 65

Number,

cardinal, 2, 10

Indeterminateness of rational functions, cases of, 223

Induction, mathematical, 424 Inequalities, solution of, 340
Inequality.

complex, 71 fractional, 33
imaginary, 70 integral, 18
irrational,

See Equality

46

594
Number,
natural, 6 negative, 18 positive, 18
rational, 34
real,

A COLLEGE ALGEBRA
Quadratics, 112, 298, 304 simultaneous, 317

45

Number

intervals, 579 regions, 586

of,

Radical expressions, simple, 277 Radicals, 271 index of, 271

similar, 273

Numbers, theory
Odds, 410
Ordinal,
7, 33,

211

simple, 271

Radicand, 271
45
Ratio, 09, 347 Rationalization, 285

Ordinate, 138 Origin, 137

Oscillation of a function, 584

Remainder theorem, 169

Resultants, 512
properties of, 514 Rolle's theorem, 467

Parabola, 333
Parentheses, rule
of,

95

_Roots of equations, 112, 426 extraneous, 116

infinite, 229, 306,

Part (of group), 3

Permanences of

sign,

446

imaginary, 444, 448 439

Permutations, 393

irrational,

453

odd and even, 492

Polynomials in
x,

location of, 452, 458, 475

multiple, 428, 463

87

products of, 103 Powers. See Exponents and Involution

perfect, 260

number

of,

427

rational, 429

superior and inferior limits

430, 441, 466

of,

Power

series, 535
of,

symmetric functions
535

of,

305,

convergence
products
quotients reversion
of,
of, of,

545

546
548
of,

434, 478 Roots of integral functions, 260 cube, 266

square, 262

transformation
Probability, 409

545

Roots of numbers, cube, 268,. 483

principal, 271,

39, 56, 76

Products, continued, 252 infinite, 564

Progressions, arithmetical, 354
arithmetical, of higher order,

square, 265, 292, 295, 296

trigonometric

expression

of,

490
Scale, complete, 6

364
geometrical, 357

harmonical, 362 Proportion, 347

continued, 350

natural, 17

Sequence of numbers, 58
regular, 60

INDEX
Series, alternating, 632

5%
of

Subtraction

rational

expres-

binomial, 538, 553

sions, 217

doubly

infinite,

543

of series, 541

exponential, 537, 556

geometric, 360

Surds^291 Symmetry, absolute, 245

cyclic,

hypergeometric, 529 infinite, 520

logarithmic, 537, 557
recurring, 560 Sign, rules of, 95

248

Taylor's theorem, 461, 551

Term, 86
absolute, 426

Simultaneous,
tions Sine, 488

127.

See

Equa-

Transformation of equations, 114, 129, 436 Transposition of terms, 115

Value, absolute or numerical, 18, 75, 488
Variable, 58, 79

Solutions of systems of equations,

128
infinite, 230,

318

integral, 342

number

of,

517

Square root. See Roots Sturm's theorem, 472

Substitution, principle of, 128 Subtraction of integral expressions, 93
of

continuous, 69 Variation, 351

of integral functions, 308, 469

Variations of sign, 446

Zero, 17 as limit, 63

numbers,

16, 19, 35, 51, 72

of radicals, 274

operations with, 19, 25, 31

RETURN

TO^^

CIRCULATION DEPARTMENT 202 Main Library

GENERAL LIBRARY

U.C.

BERKELEY

BDODflblSfll
V

'~f

V
y^i #r-^

^3

4
:

THE UNIVERSITY OF CALIFORNIA LIBRARY