L

1
8- A:<'l' J
UN T'f OF
c..,L NIA
SAr- OlbGO
d fl) the
Y I.IBH,\RY
01\I\'EHSITY OF CAl IFOH"IA
s \.\ mcco
KARL DYK
,.
;1 3 1822 01264 8036 '>
3 t1 -re{_ - - A'O A /:" .?(/I'
SCIENCE AND ENGINEERING LIBRARY
University of California, San Diego .....
1
.,

Please Note: This item is subject to
RECALL after one week.
DATE DUE
MAR 0
;.: . '
. .
.
SE 7 (Rev. 7/82) UCSD Libr.
ANALY'l"lO GEOl\IETRY
OF SPACE
BY
VIRGIL Pn.D. (GoTTINGEN)
Pnon:sson or :\I ATHE.\IATICS AT
UI' JVEJlSifY
AI(D
C. H. SJSAi\1, Pu.D. (CoRNELL)
AssiSTANT l 'non:SSO!l Of' AT THE
UNn"l;mnTY Of' I LUNOIS
NEW YORK
HENRY HOLT AND COMPANY
COPYiliOIIT, 1914,
BY
HENRY HOLT AND COMPANY
April, 1924
PRINT0 IN V, 8. A.
PREFACE
Ix this book, which is planned for an introductory course,
the first eight chapters include the subjects usually treated in
rectangular coordinates. 'l'hey presuppose as much knowledge
of algebra, geometry, and trigonometry as is contained in the
major rcquitement of the College Entrance Examination Board,
and as much plane analytic geometry as is contained in the
better elementary textbooks. In this portion, proofs of theorems
from more n.clvanced subjects in algt>bra are supplied as needed.
Among the features of this part are the development of linear
systems of pln.nes, plane cootdinates, the concept of infinity, the
treatment of imaginaries, and the distinction between centers
and ,-erti ccs of quadri c surfaces. The study of this portion can
be regarded as a first course, not demanding more than thirty or
forty lessons.
In Chapter IX tetrahedral coordinates are introduced by means
of liu<>ar trausformations, uuder which various inYariant proper-
ties are establishNL These COitrclinatcl> are used throughout the
next three chapters. The notation is so choscn that no atnbigu-
ity can arise between tetrahcllral ant) rectangular systems. The
selection of suuject matte1 is such as Lo be of greatest service fo1
further study of algebraic geomet1y.
In Chapter XIl l a more achanct>d knowledge of plane analytic
geomE'try is presupposed, bnt the part involving Pliicker's num-
bers tna)" be omitted without disturbing the continuity of the
subject. In the last chapter extcnsi,e use is made of the cal-
culus, including the use of partial differentiation and of the
element of arc.
The second part will require about fifty lessons.
ill
ARTlOLZ
l o Coordinates
20 Orthogonal projection
CONTENTS
CHAPTER I
COORDINATES
3o Direction cosines of a line
4o Distance between two points
So Angle between two directed lines
Go Point dividing a segment in a given ratio
7. Polar coordinates
8o Cylindrical coordinates
9o Spherical coordinates
CHAPTER II
PLANES AND LINES
10o Equation of a plane
llo Plane through three points
120 Intercept form of the equation of a plane
130 Normal form of the equation of a plane 0
140 Reduction of a linear equation to the normal form 0
150 Angle between two planes
16. Distance to a point from a plane
170 Equations of a line 0
18o Direction cosines of the line of intersection of two planes
190 Forms of the equations of a li ne
200 Parametric equations of a line
21. Angle which a line makes with a plane
220 Distance from a point to a line
2So Distance between two non-interseding lines
24o System of planes through a line
250 Application to descriptive geometry
v
PAC
I
3
5
6
7
8
10
10
11
12
13
13
14
15
16
li
19
w
20
21
22
23
24
25
28
Vl
CONTENTS
AP.TtrLF. PAGF.
26. Rundles of planes 2!l
27. l'lane coordinates 31
28. Equation of a point 32
29. Homogeneous coordi nate of the point and of the plane 33
30. Equation of the plane and of the point iu homogeneous coordinates 34
31. Equation of the origin. Coordinates of planes through the origin . 34
32. !'lane at infinity 35
33. Lines at. it1finity 35
34. Coordinate tetrahedron 35
35. System of four planes 36
CHAPTER Ill
TRANSFORMATION OF COORDINATES
36. Translation
37. Rotation .
38. Rotation and reflection of a x ~ > s
39. Eul er's formulas for rotation of axes
40. Degree unchanged by transformation of coordinates
CHAPTER IV
TYPES OF SURFACES
41. Imaginary points, lines, and planes
42. Loci of equations
43. Cylindrical surfaces
44. P1-ojectin:,: cylinders
45. Plane sections of surfaces
46. Coucs
47. Surfaces of revolution
CHAPTER V
THE SPHERE
48. The equation of the sphere
49. The absolute .
50. Tangent plane .
51. Anglo between two spheres
52. Spheres satisfying given conditions
53. Linear systems of spheres
64. Stereographic projection .
38
38
41
42
42
44
46
47
47
48
49
50
52
52
55
55
56
57
50
CONTENTS
CIIAPTTm VI
FORMS OF QUADRIC SURFACES
ARTICLE
55. Definition of a quadric
56. The ellipsoid
57. The hyperboloid of one sheet
58. The hyperboloid of two sheets
59. The imaginary ellipsoid .
60. The elliptic paraboloid
61. The hyperbolic paraboloid
62. The quadri c cones .
63. The qLtadric cyli nders
64. Summary .
CHAPTER VII
CLASSIFICATION OF QUADRIC SURFACES
65. Intersection of a quadric and a line
66. Dia111etral planes, center
67. Equation of a quadric referred to its center
68. Principal planes
69. Heality of the roots of the discriminating cubic
70. Simplification of the equation of a quadric
71. Classification of quarlric surfaces
72. Jnvarianrs under motion .
73. !'roof that J, J, and D are invariant
74. Proof that A is invariant .
75. Discussion of numerical equations .
CHAPTER VIII
SOME PROPERTIES OF QUADRIC SURFACES
76. Tangent lines and planes
77. Normal forms of the equatiou of the tangent plane
78 ~ o n n a l to a quadric
79. Rectilinear generators
80. Asymptotic cone
81. Plane sections of quadrics
82. Circular sections
83. Real circles on types of quadrics
84. Confocal quadrics
85. Confocal quadrics through a point. Ell ipt.ic coordinates
86. Confocal quadrics tangent to a line .
87. Confocal quadrics in plane coordinates .
VII
PAGE
63
63
65
67
68
69
70
71
72
72
74
75
77
78
j!)
80
tll
82
83
84
86
00
01
02
03
05
!)6
!)8
100
104
105
107
108
Vlll CONTENTS
CHAPTER IX
TETRAHEDRAL COORDI NATES
A RTICLY. PA.HR
88. Definition of tetrahedral coordinates 109
89. Uni t point 110
90. Equation of a plane. Plane coordi nates 111
91. Equation of a point 112
92. Equations of a line 112
93. Dual ity . 113
94. Parnmetric equations of a plane and of a poi nt 114
95. Parametric equations of aline. Range of points. Pencils of planes 115
96. Transformation of point coordinates 117
97. Transformation of plane coordinates 110
98. l'roj ecti ve transformations 120
99. Invariant points 121
100. Cross-ratio 121
CHAPTER X
QUADRIC SURFACES IN TETRAHEDRAL COORDINATES
101. Form of equation .
102. Tangent lines and planes
103. Condition thM the tangent plane is indeterminate .
104. The invariance of t he di scriminant
105. Lines on the quadric surface .
106. Equation of a quadric in plane coordinates
107. Polar planes .
108. Harmonic property of conjugate points
109. Locus of poi nts which lie on their own polar planes
110. Taul{ent COliC
111. Cnnjug:l.te lines to a quatlric
112. Self-polar tctrahr clnm .
113. Equation referre<l to a self-polar tetrahedron
114. Law of inertia
11 5. Hectilinear generators. Heguli
116. Hyperbolic coordinates. Parametric eqnations
117. Proj ection of a quadric upon a plane
118. Equati ons of t he .
ll9. Quadrics determined by three non-intersecting lines
120. T mm;versals of four skew lines
121. Tht> q n:ul ric cone .
122. l'rojcctiun of a quadric couc upon a plane
124
124
125
126
12!}
130
132
132
13:3
133
134
135
136
136
137
1:38
13!)
140
141
143
14:'1
145
CONTENTS .
CHAPTER XI
LINEAR SYSTEMS OF QUADRICS
AJtTlCLE
123. l'encil of quadrics .
124. The >..-discriminant
125. Invariant factors .
126. The characteristic .
127. Pencil of quadrics having a common vertex
128. Classification of pencils of quadrics
129. Quadrics a double plane in common
130. Quadrics having a line of vertices in common
131. Quadrics having a vertex in common
132. Quadrics having no vertex in common
133. Forms of pencils of quadrics .
IX
147
147
148
150
151
151
151
151
152
156
163
134. Line conjugate to a point 165
135. Equation of the pencil in plane coordinates 160
136. Bundle of quadrics 167
137. Representation of the quadrics of a bundle by tbe points of a plane 168
138. Singular quadrics of the bundle 168
139. Intersection of the bundle by a plane 160
140. The vertex locus ,f J70
141. Polar theory in the bundle 171
142. Some bundles 173
143. Webs of quadrics . 175
144. The .Jacobian surface of a web 1i5
145. Correspondence with the planes of space li7
146. Web with six basis points 177
147. Linear systems of rank r 180
148. Linear systen1s of rank r in plane coordinates 181
149. Apolarity 181
150. Linear systems of apolP.r quadrics . 186
CHAPTER XII
TRANSFORMATIONS OF SPACE
151. l'rojecthe metric . 188
152. Pole and polar as to the absolute . 188
153. r:quations of motion 100
154. Classitication of projccti1e transformations l!H
155. Standard forms of equations of projective transformations 1!15
156. Rirational transformations 1!16
157. Quadratic transformations 1!18
158. Quadratic inversion 201
159. Transformation by reciprocal radii 201
160. Cyclides 203
X CONTENTS
CHAPTER XIII
CURVES AND SURFACES IN TETRAHEDRAL
COORDINATES
I. ALGEBRAIC SuRFACES

161. Number of constants in the equation of a surface .
162. Notation
163. Intersection of a line and a surface
164. Polar surfaces
165. Tangent lines and planes
166. Inflexional tangents
167. Double points
168. The first polar surface and tangent cone
169. Class of a surface. Equation in plane col.\rdinates
170. The Hessian .
171. The parabolic curve
172. The Steinerian
II. ALGEilRAtC CuRn;s
173. Systems of equations defining a space curve
174. Order of an algebraic curve
175. Projecting cones
176. .\lonoidal representation
177. Number of intersections of algebraic curves and sur faces
178. Parametric cqnations of mtional curYes
179. Tangent lines and developable surface of a curve .
180. planes. Equation in plane coordinates
181. Singular points, lines, and planes .
182. The Cayley-Salmon formulas
183. Curves on non-singular quadric surfaces
184. cnbic curves
185. Metric classification of space cubic curves
186. Classification of space quartic curves
187. Non-singular quartic curYcs of the first kind
188. Rational quartics .
CHAPT ER XIV
DIFFERENTIAL GEOMETRY
I. Al'ALYTIC CUII\ 'ES
189. Length of arc of a space curve
190. The 1110ving trihedral
191. Curvature

206
207
207
208
20()
210
210
211
212
213
214
214
215
216
217
219
221
222
224
224
22G
226
2:28
2:30
234
235
238
242
245
246
248
CONTENTS
A.ltTtCL.
192. Torsion
193. The }'renet-Serret formulas
194. The osculating sphere
195. :\l inimal curves
II. ANALYTIC SuRFACEs
196. Parametric equations of a surface
197. Syst.ems of curves on a surface
198. Tangent plane. :Xormalline
199. Differential of arc .
200. ;\[inimal curves
201 . Angle between curves. Differential of surface
202. Radius of normal curvature. Meusnier's theorem
203. Asymptotic tangents. Asymptotic curves
204. Conjugate tangents
205. J>rincipal radii of normal curvature
206. Lines of curvature .
207. The indicatrix
A:<SWt:RS
hDEX
xi
P A t ~ P
21!1
250
251
252
254
255
255
257
258
259
259
201
261
202
263
206
269
287
ANALYTIC GEOMETRY OF SPACE
CHAPTER I
COORDINATES
1. Rectangular coordinates. The idea of rectangular coordinates
as developed in plane analytic geometry may be extended to space
in the following manner.
Let there be given three mutually perpendicular planes
(Fig. 1) XO } ~ YOZ, ZOX, intersecting at 0, the origin. These
planes will be called coordinat e planes. The planes ZOX, XOY
intersect in X' OX, the X-axis; the pbnes XO 1 ~ YOZ intersect
in Y' 0 J ~ the r:a.xis ; the
planes YOZ, ZOX intersect
in Z' OZ, the Z-axis. Dis-
tances measmed in the
directions X'OX, P 0 1 ~
Z'OZ, respectively, will be
considered positive; those
measured in the opposite
directions will be regarded
as negative. The coordi- Fro. 1.
nates of any point P are its distances from the three coordinate
planes. The distance from the plane YOZ is denoted by x, the
distance from the plane ZOX is denote< I by y, and the distance
from the plane XOY is denoted by z. These three numbers
x, y, z are spoken of as the x-, y-, z-coordinates of P, respect-
i,ely. Any point P in space has three real coordinates. Con-
versely, any three real numbers x, y, z, taken as x-, y-, and z-
coordinates, respectively, determine a pointP; for if we lay off a
distance OA=x on the X-axis, OB=y on the Y:axis, OC = z on
1
2
COORDINAT ES (CHAP. I.
the Z-axis, and draw planes through A, B, C parallel to the co-
ordinate planes, these planes will intersect in a point P whose
coordinates are x, y, and z.
It will fl-equently be more convenient to determine the point
P ' ~ h o s e coordinates are x, y, and z, as follows: Lay off the
distc'l.llce OA = x on t he X-axis (Fig. 2). From A la.y off the
distance AD= y on a parallel to the }':.axis. From D lay off the
Z distance DP = z on a parallel to
C the Z-axis.
'rhe eight portions of space
separated by the coordinate
planes are called octants. U the
coordinates of a point P aro u,
X b, c, the points in the remaining
octants at the same absolute
y B D distances from the coordinate
F1o. 2. planes arc (-a, b, c), (a, - b, c),
(a, b,- c), (- a,-b, c), (-a, b,- c), (a,- b,- c), (-a,- b, - c).
Two points are symmetric with regard to a plane if the line
joining them is perpendiculat to the plane and the segment
between them is bisected by the plane. They :ue symmetric with
regard to a line if the line joining them is perpendicular to the
given line and the segment between them is bisected by the line.
They are symmetic with reganl to a point if the segment be-
tween them is bisected by the point.
The problem of representing a figure in space on a plane is
considered in descriptive geometry, where it is solved in several
ways by means of projections. In the figures appearing in this
book a particular kind of parallel projection is used in which the
X-axis and the Z-axis are represented by lines perpendicular to
each other in the plane of the paper; the Y..axis is repesented by
a line making eqnal angles with t he other two. Distances
parallel to the X-axis or to the Z-axis are represented correctly
to scale, but distances parallel to the }':.axis will be foreshortened,
the amount of which may be chosen to suit the particular drawing
considered. It will usna1ly be convenient for the student, in
drawing figures on cross section paper, to take a unit on the
Yaxis 1/ v'2 times as long as the nui t on the other axes_
ART. 2) ORT HOGONAL PROJECTIONS 3
EXERCISES
1 . Plot the following points to scale, using cross section paper: (1, 1, 1),
(2, 0, 3), (- 4, - 1, -4), (-3, -4, 1), (4, 4, - 1), ( - i,2,3), ( -1, 6, -6),
<- -t, 2, S), {a, - -1 , - 1), (2, 1, - 3), (- 1, o, O), (4, - 2, 2), (O, o, 2),
(0, -1, 0), (- 3, 0, 0), (0, 0, 0).
2. What is the locus of a point for which x = 0?
3 . the locus of a point for which ._ = 0, y = 0?
4 . What is the locus of a poiut for which x = a, y = b?
5. Given a point (k, l, m), write the cotirdinates of the point symmetr ic
with it as to the plane XO I'; the plane Z OX; the the r-axis; the
origin.
:!. Orthogona l projections. The orthogonal projection of a
point on a plane is the foot of the per pendiculat from the point
to the plane. The orthogonal projection on a plane of a segment
PQ of a line* is the segment joining the }Jrojections P' and
Q' of P and Q on the plane.
The orthogonal projec:tion of a point on a line is the point in
which the line is intersected by a plane which passes through the
given point and is perpendicular to the given line. 'l'he or-
thogona,l projection of a segment PQ of a line l ou a second line
l' is the segmeut P'Q' joining the projections P' and Q' of P and
Q on l.
For the purpose of me<\smiug distances and angles, one direc-
tion along a line will l.Je regartlerl as positi,e and the opposite
direction as negative. A segmeut PQ on a directed line is
positive or negative according as Q is iu the positive Ol' nega-
tive direction from P. From this definition it follows that
PQ=- QP.
The angle between two intersecting directed lines l and l' will
be defined as the smallest angle which has its sides extending
in the positive dil'ectious aloug l and l'. We shall, in general,
roa,ke no convention as to whether this augle is to be considered
positive or negative. The angle between two non-intersecting
directed lines l and l' will be defined as equal to the angle be-
tween two intersecting lines 1n and m' having the same directions
as land l', respectively.
We shall use the word line throughout to mean a straight liue.
4
COORDlNA'rES [CHAP. 1.
TnEORJm I. The length of the 1'1ojeclion of a segment of a
d irectecl line on a second clirectecl line is equal to the length of the
given segment ?nultipliecl by the cosine of the Wtgle between the lines.
Let PQ (Figs. 3 a, 3 b) be the given segment on l and let P' Q'
be its projection on l'. Denote the angle between l and l' by 8.
It is required to prove that
P'Q' = PQ cos 8.
Through P' draw a line l" having the same direction as l. The
angle between l' and l" is equal to 8. Let Q" be the point in
FIG. 3u. FlO. 3 b.
which l" meets the plane through Q perpendicular to l'. Then
the angle P'Q'Q" is a right angle. Hence, by trigonometry,
we have
But
It follows that
P'Q' = P'Q" cos 8.
P'Q" = PQ.
P'Q' = PQ cos 8.
It should be observed that it makes no difference in this
theorem whether the segment PQ is positive or negative. The
segment PQ = r will always be regarded as positive in defining
cosines.
Tu1W1um II. Tlte }nojection on a (lirectecl line l of a broken
lil1e macle 1tp of segme11ts P
1
P
2
, P
2
P
3
, .. , P,._
1
P,. of dijJ"erent lines is
the sunt of the projections on l of its parts, an<l is equal to the pro-
jection 011 l of tlte straigztt line P
1
Pn-
ARTs. 2, 3) COSDJES OF A LINE 5
For, let P\, P'
8
, , P',.-1
1
P
8
, , P,,_
1
, Prtt respectively.
equal to P'tP'n ; t hat is,
PI,. be t he projections of P
1
, P
2
,
The sum of t he p rojections is
P'J)I
2
+ P'
2
P'
3
+ +P',._
1
P',. = P\P',..
But P'
1
P',. is the projection of P
1
P,. The theorem therefore foll ows.
CoROLLARY. If P
1
, P
2
, , P,. _
1
ate the vel"liCe/5 of a ptJiygrm, the
sum of the 1JI"Ojections on any directecl line I of the sl'gme111:;
P
2
P
3
, , P,._
1
P
1
fonned by the sides of the polygon i::; zero.
Since in this case P,. and P
1
coincide, i t follows that P\ and P',
also coincide. The suJU of the projections is consequently zero.
EXERCISES
1. If 0 is the origin and P any puint in l<pact!, show that the pl"Ojr:ctions
of the 0 P npn tlw C<>ordinate axes are equal tO the coi.ird iuates of P.
2. If the coiirdiuatts of P
1
an .(" Z 1 and of are y
2
ze, show that
the of the :<Pglllt>nL l't l'e upon the coordinate axt>s are equal to
X1. Ye- y,, Zz- Zit
3. H the leugths or tlu: projections of P1 P" upou the axes are respectilely
3, - i and the co()rdinates of 1'1 are (- -1, 3. the coordinates of P
2

4. Find the distance fron1 the urigiu to thP point ( -l, 3, 1::!).
5. Find the distance ftom the origin to the point (a, t,. r').
6. Find the of the angles made with the ax.,s IJy the line joining
. the origin to each of the following points.
( 1, 2, 0) (1, 1, 1)
(0, 2, 4) (1,- 4, 2)
3. Direction cosines of a line.
Let l be any d irected li ne in
space, and let l' be a line through
the origin which has the same
direction. If a, {3, y (Fig. 4)
are the angles which l' makes
'vith the co<:irdinate axes, t hese
are also, by definition ( Art. 2),
the angles which l makes with y
the axes. They are called the
(- 7, tJ, :!)
(.', y, z)
FIG. 4.
X
dhection angles of l and t heir cosines are call ed direction cosines.
The latter will be denoted by A, p., v, respectively.
6
COORDI::SATES
(CHAP. I. .
LE't P =(CI, u, c) be any point on l' in t he positi\e di rection from
the origin and let OP = 1'. Then, from trigonometry, we luwe
('
A= cos cc = - ,
r
b
J-1-= cos /3=- ,
,.
c
v=cosy=-
,.
But r is the diagonal of a rectangular parallelopiped whose edges
are
OA =a, OB = u, 00=C.
Hence, we obtain
In this equation, as in the formulas throughout the book, except
when the contrary is stated, imlicnted roots are to be taken with
the positive sign.
By substituting this value of,. in the above equations, we obtain
A = ccs /( = (( I
Vet2 + IJ2 +c2
c
v = cos )' =
vu?+b
2
+ 1-.2
Hy squaring each mem her of these equations and adding the
results, we obtain
A:! + I":! + v:! = 1, (1)
hence we hn ve the following theorem. .
The sum of the S'JitCit'es of the direction cosines of a
line is eq,wl to unity.
If All P.u VI and A2, l-'-21 v2 are the uirection eosiues of two like
directed lines, we have
If the lines are oppositely directed, we have
4. Distance between two points. Let P
1
=(.r
1
, Y11 z1)
1
P2-=(x2,Y2'
z
2
) be any two points in space. Denote the direction cosines of the
ARTS. 4, 51 AKGLE BE'l' '.\'EEN T'.\'0 DIRECTED LINES 7
line P
1
J>
2
(Fig. 5) by A, ,..., v and the length of the segment P
1
P
2
by <l. The projection of the segment P
1
P
1
on of the axes is
equal to the sum of the projections Z
of P
1
0 and OP
1
, that is
By ::;quating both members of t hese
equations, adding, and extracting the Y
square root, we obtain
EXERCISES
1 . Find the distance between (3, 4, - 2) and ( - 5, l , - 6).
llf!) X
(2)
2. Show thaL Lhe points (- 3, 2, - 7), (2, 2, - 3), and (- 3, 6, - 2) are
vettices of an isosceles triangle.
3 . Show that the points (4, 3, - 4), ( - 2, !1, - 4), and(- 2, 3, 2) are
vertices or an equilateral triangle.
4 . Express by an equation that the point (x, y, z) is equidistant from
(1, 1, 1) ami (2, 3, -1 ).
5. Show that + yz + = ,J is the eqnation llf a sphere whose center is
the and whose radius is 2.
6. Find the directitu cosiues of the line P
1
P
2
, given:
(a) P1 = (0, (\. 0),
(h) P1 :=(I, I, I),
r.::(I,-2,3),
T'2:=(2, 3, 5).
= (2, 2, 2) .
P2:=(.J, 2, -1).
7. What is known about tiJe direction of a line if (a) cos u = 0?
(b) COS(C=0andcos{j=0? (c) COSIC=l?
8. Show that tile points (3, - 2, i ), (G, 4, - 2), and (5, 2, I) are on a
line.
9. Find the direction cosines of a line which makes equal angles with the
coordinate axes.
5 . Angle between two directed lines. J,et /
1
and 1
2
he two
directed lines ha\'ing the di rection cosines .\., p.
1
, v
1
and .\
2
, p.
2
, v
2
,
respecthreJy. It is l'eqnired to find an expression for the rosine
of the angle between l
1
and 1
2
Through 0 (Fig. 6) draw two
8
COORDINATES [CHAP. l.
lines OP
1
and OP
2
having the same di-
rections as l
1
and l
2
, respectively. Let
OP
2
= r
2
and let the coordinates of_P
2
be
Yz = Jl!N,
The projection of OP
2
on OP
1
is equal
FtG. 6.
Hence
to the sum of the projections of the
broken line OJLVP
2
on OP
1
(Art. 2).
t
.I But OP
2
= 1
2
, = TzAz, Yz = l'zP-2
1
NP = Zz = r
2
v
2

Hence, we obtain
1'z cos (J = 1'zAtAz + 1'z!J-tP.z + Tzvt vz,
or cos e = .\
1
>.:! + ... .,.
2
+ v
1
v
2
(3)
'rhe condition that the two given lines are perpendicular is
that cos (J = 0. Hence we luwe the following theorem :
'fHEOHF.U. Two lines l
1
ancll
2
.with direction cosines .\
1
, p.
11
v
1
and
.\
2
, p.
2
, v
2
, 1espectively, me 11e1')1euclicular if
(4)
The square of the sine of fl may be found from (1) and (3).
Since sin
2
(J = 1 - cos
2
(J, it. follows that
sin
2
(J = (.\.
2
+ P.t
2
+ v.
2
)(.\z
2
+ rz
2
+ Vz
2
)-(.\t,\2 + P.tP.z +v.vz)
2
= (.\
1
p.z- .\
2
p.
1
)
2
+ (p.
1
v
2
- p.zv
1
)
2
+(v
1
,\z- vz.\
1
)
2
(5)
H. Point dividing a segment in a Z
given ratio. Let 1'
1
= (x .. !I" z
1
) and
p2:;: !f2, Zz) he tWO gi Vt'll )JOints
(Fig. 7). It is required to find tlw
point P =: (x, y, z) on the line P
1
P
2
such that P
1
P: P 1'
2
= 111
1
: 111
2
Let
.\, Jl, v be the tlir<'ct ion cosines of
the line 1'
1
P
2
Then (AI't. Th. I) we have
P
1
P ,\ = x- :t'
1
and PP
2
,\ = x
2
- x.
Hence
ART. G) PO!WJ' Dl \ ' l DH\U A ::>IW ... \ I EN'l'
On solving for x we obtain
Similarly,
X 11t2X 1 + lntX:!
= llt l + 11t2 0
1
_m:!Yt + 111tY2
y - + 1112 '
+ 11ttZ:!
::: = 1/tl + 11t2 0
9
(6)
It should be noticed that if m
1
and have the same sign, P
1
P
antl are measured in the same direction so that Plies between
1'
1
and P
2
If m
1
and 1n
2
have opposite signs, P lies outside the
segment P
1
P
2
By giving m
1
and 111
2
snita.ble \'alues, the coor-
tlinates of any point on the line P
1
P
2
can be represented in
this way. In particular, if P is the mid-point of the segment
P
1
P
2
, 11t
1
= m
2
, so that the coordinates of tht! mid-point are
y = YI +.112
1
2
EXERCI SES
z. + Z:!
Z=---
2
1. Find the cosine of the angle between the two lines whose direction
. 1 2 ll d 2 -1 5
cosmcs are --= -- -- an - - , --, --
v'H vi4' vao vao Y30
2 . Find the direction cosines of each of the coordinate axes.
3. The direction cosines of a line are proportional to 4, - 3, 1. Find
their values.
4. The direction cosines of two lines are proportional to 6, 2, - 1 and
-3, 1, - 5, respectively. Fiud the cosine of the angle between the liues.
5. Show that the lines whose direction cosines arc proportional to 3, 0,
2; - 2, 3, - 0; - 0, 2, 3 are mutually perpendicular.
6. Show that the points (7, 3, 4), (1, 0, 6), (4, 5, - 2) are the venices
of a right triangle.
7. Show that the points (3, 7, 2), (4, 3, 1 ), (1, 6, 3), (2, 2, 2) are the
vertices of a parallelogram.
8. Find tlte coordinates of the intersection of the diagonals in the paral-
lelogram of Ex. 7.
9. Show by two different methods that the three points (4, 13, 3),
(3, 6, 4), (2,- 1, 5) are collinear.
10 COORDIXA'l'ES [CilAP. I.
10. A line Ulakes an angle of i5 with the X-axis and 30 with tile r-axis.
How many positious may it hnsc? Find, for each position, the co:;ine of the
angle it makes with the Z-nxis.
11. Dl'lermiue t he coonl i uate!i of the iutcrsection of the medians of the
triangle with ,ertices at ( I, :!, :3), (:! , 3, I), (:{, 1, 2).
' 12. Pro,c that tho medians of auy meet in a point twice as far
from each \'t'rtcx as from the mid-point of the opposiw sitle. This point is
called the ceulrr 11J ymvily of the tri:mgle. '
, .
""' 13. Pro,c that the three straight lines joining the mid-poirlt..;; of oppo-
site tllgcs of any tctahcd ron meet in a point, and arc IJil\CCtcd by i t. This
point is called the t't' l!I N' !/l'avity of the tetrahed ron.
14. Show that the li iiCs joiuing each vertex of a tetrahedron to U1e point
of intersection of the medians of the opposi te face pass through the center of
gravity.
15. Show that the lines joining the middle points of the sides of auy
ltnndril:ucral form a
16. how the ratio ut
1
: m
2
( Art. ll) varies as P describes the liue
7. Polar Coordinates. Let OX, Or; OZ be a set of rectangular
axes and P any point in space. Let OP = p have the direc-
f.. z tion angles rt, {3, y. The position
I
Z l //
! 'Y /
of the line OP is determined by
a, {3, y and the position of P on
the line is given by p, so that the
position of the point P in space
is fixed when p, {3, y are
X known. These quantities p, a, {3,
y are called the polar coordinates
of P. As a, (3, y :ue direction
angles, they are not independent, since by equation (1)
__
Y/ 0 X
,/
,._-"-------...V
y Fw. l;.
cos
2
a + cos
2
f3 + cos
2
y = 1.
If the rectangular coonlinatcs of Pare x, y, z, then (Art. 3)
x = {I cos a, y = p cos {3, z = p cos y.
8. Cylindrical coordinates. A point is determined when its
dit'ccted distanco from a fixed plane and tbe volat coordinates of
its orthogonal projection on t.hat plane arc known. The:;e co-
ordinates arc called tht> cylindrical col!rdinates of a point. If the
ARTS. 8, OJ SPHERICAL COORDINATES
point P is refcrreu to the rectaugulat
axes x, y, z, and the fixed plane is taken
as z = 0 and the x-axi s for polar axis,
we way write (Fig. 9)
:c= p cos 8, y=psin 8, Z =Z,
in which p, 0, z are t he cylindrical coordi- y
nates of P.
11
z
p
X
p'
fro. 9.
0. Spherical coordinates. L<'t 0.\, OZ, and P be chosen as
in Art. 7, and let P' be the ort hogon:d proj<'ction of J> on the plane
XO Draw OP. The pllsition of P is defined by the tli stance
p, the angle cf> = ZOP whid1 t he !me OP makes with the z-axis,
aud the angle B (measnrell by the angle XOP' ) which t he plane
through P and t he z-axis lllakes with t he plane XOZ. 'J.'he num-
bers p, cf>, 8 are call ed the spherical coordinates of P. The length
p is called the radius vector, t he angl e cf> is called the co-lat.itnde,
Z and 8 is call ed t he longitude.
P If P = !/, z), then, from the figure
10),
OJ>' = p cos (!JO - cf>) = p sin cf>.
Hence x = p sin cJ> <:os 8,
X y=psin cJ> sin(J,
z = p cos .
On solving these equations for p, cf>, fJ, we find
cf> = arc ros - z _ ,
+ + z
2
8 = arc tan '!1. .
X
EXERCISES
1. What locus is defined by p = 1 ?
2. "'hat l OCUS iS de> fined by IC = ()0 ?
3. What locus is defined by II=
4. locus is defined by tf> = 15 ?
5. Trausfortn :-.-
2
+ + z2 = -1 t o : (a) polar co<5rdinntes, (b) spherical
cotirdinates, (c) cylinurical coordinate::.
6 . Transform + y
2
= :fl into co<5rdin:ttel\; into cylindrical
co<5rd i nates.
7 . Express the distance between two points in t erms of thei r polar
co<5rdinates.
CHAPTER II
PLANES AND LINES
l o. Equation of a plane. A plane is characterized by the
properties :
(a) It contai ns three points not on a line.
(b) It contai ns every point on any line j oining two points on it.
(c) lt does not contain all the points of space.
Tm:om:)t. The locus of the ]lOillts whose vounlinttleiJ a
linea1 e<JIWtion
Ax+By+ Oz+ D = O
'With 1ectl coe.t}icients a plane.
(1)
We shall prove t his theorem on the supposition t hat a -:1= 0.
Since .A, B, a a1e uot all zero, a proof for the case in which
a= 0 can be obtained iu a similar way.
I t is seen by inspection that t he coordinates ( O, O, - g}
(o, 1, - (1, 0, D)) satisfy equation. These
three points are not collinear, since no values of m., -rn
2
other than
zero satisfy the simul taneous equations ( .\rt . 6)

Let P
1
= !Ju Z
1
) ancl P
2
z
2
) be any tl'l.O points whose
coordinates satisfy (1). The coordinates of any point P on the
1 i ne P
1
P
2
are of the form ...
Y = lllt!lz + 11ti!lt ,
m1+m2
z = m1z2 +
7n + m2
The equation (1) is satisfied by the coordinates of P if

+ By
1
+ az
1
+ D)+ ?n
1
(.t Ll:t + By
2
+ az
2
+ D)= O,
but since the coordinates of 1'
1
and 1'
2
satisfy (1), we have
Ax
1
+ By
1
+ Oz. + D = O, AX:! + B!!z + Oz2 + D = 0,
hence the coordinates of P satisfy (1) for all values of m
1
and m
2

12
ARTS. tl , 12) FORJ\1 OF THE EQUA'riON 13
Finally, not all the points of space lie on the locus defined by
(1), since the coordinates ( 0, 0, - (D ""&
0
>) do not satisfy (1).
This completes the proof of the theorem.
11. Plane through three points. Let (x
11
?fu Z
1
), !/2 z2),
(.t3, ?13J z3) be the coordinates of three non-collinear poi nts. The
condition that these points all lie in the plane
Ax + 1Jy + Cz + D = 0
is that their coordinates satisfy this equation, thus
Ax1 + B!ll + Cz1 + D = O,
D=O,
Ax
3
+B!I3+ D=O.
The condi tion that four numbers .tl, B, 0, D (not a.IJ zero)
exist which satisfy t he above four simul taneous equations is
X y Z 1 77kaJJ;
.c. !h z. 1 0 . 0 = g, -<A.J
?h Z2 1=
X3 ?/3 Z3 1
This is the reqniretl equation, for it is the equation of a plane,
since it is of first degree in x, !/, z (Art. 10). The plane passes
t hrough the given points, since the cootdinates of each of the given
points satisfy the equation.
12. Intercept form of the equation of a plane. If a plane inter-
sects the X-, Y-, Z-axes in three points B, C, r espectively, the
segments 0.1, OB, and 00 are calle() the intE>rcepts of the plane.
Let .A, B, 0 all be distinct from the origin and let the lengths of
the intercepts be a, b, c, so that A =:(a, 0, 0), B =: (0, b, 0), 0 =:
(0, 0, c). The equation (2) of the plane determined by these three
points (Art. 11) may be reduced to
(3)
This equation is called the intercept form of the equation of a
plane.
14 PLANES AND LINES [CHAP. I I.
EXERCISES
1 . Find the equation of the plane through the points ( 1, 2, 3), (8, 1, 2),
(5, -I, :3) .
.,. 2. Find the e'(nation o[ tho plane through the points (0, 0, 0), ( l, 1, 1),
(2, 2, - 2). What are its intercepts?
3. ProYe t hat the four ( 1, 2, :3), (2, 4, 1), ( - 1, 0, J), (0, 0, 5)
lie in a plane. Fino the equation of the plane.
4. Determine k so that thP points ( 1, 2, - 1 ), (3, - 1, 2), (2, - 2, 3),
(1, - I, k) shall lie in a plane.
r 5. Find the point o intersection o[ the three planes. x + y + z = 6,
2 Z - y +2 X= 0, X- 2 y + 3 Z = 4.
13. The normal form of the equation of a plane. Let ABO
(Fig. 11) be any plane. Let OQ be drawn through the ori gin per-
Z pendicular to the given plane
c and intersecting it at P'. Let
the direction cosines of OQ be
.\, p., v and denote the length of
the segment Ol" by p.
X Let P= (x, y, z) be any poi nt
in the given plane. The projec-
tion of P on OQ is P ' (Art.
Draw OP and the broken line
O.lLVP, made up of segments
Fto. 11. 0,1[ = x, ;1/ N = y, and NP= z,
parallel to the."\.-, J:, and Z-axes, respectively. The projections of
OP and OJ!NP on OQ are equal (Art. 3, 'rh. II). The projection
of the broken line is .\.t + p.!J + vz, the projection of OP is OP' or p,
so that
.\:c + p.y + vz = ]J.
(4)
This equation is satisfied by the coordinates of every point Pin
the given plane. lt is not satisfied by the coordinates of any
other point. For, if P
1
is a point not lying in the given plane, it
is simi larly seen, since the projection of OP
1
on OQ is not equal to
p, that the coordinates of l'
1
do not satisfy (4).
lienee, (-l.) is the equation of the plane. 1t is called the normal
form of the equation of the plane. The number 11 in this equa-
tion is positive o1 negative, acr.ol'cling as P' is in the positive or
negative dil'ection from 0 on OQ.
ART. 141 REDUCTIOX OF THE EQUATION 15
14o. Reduction of the equation of a plane to the normal form. Let
.th+By+Gz+D=O (5)
be any equation of first degree with real coefficients. Tt is required
to reduce this equation to the normal form. Let Q =(A, B, 0)
be the point whose coordinates are the coefficients of x, y, z in t hi s
e'tn:ltiou. The direction cosines of the directed line OQ a re
(Art. :3)
A B 0
X= ---- p. = - v= (6)
v.l
2
+E- + cP + w +C
2
' v .. P +ff-+0
2
If we transpose the constant term of (5) to the othe1 member of
the ecptation, and divide both members by VA
2
+ B
2
+ 0
2
, we
obtain
A B

v . P + JJl + 0
2
v ..rP + B
2
+ 0
2
0 - D (
+ - z = 7)
.y::-12 + +
0
2 12 + n2 +
0
2
The plane determinetl by (7) is ill entical with that (]etermined
by (5) since the coordinates of a point will sati sfy (7) if, and only
if, they satisfy (:1). By subtit.nting f rom (6) in (I) and comparing
with (-l), we see that the locus of the equatiou is a plane perpen-
dicular to OQ and intersecting OQ at a point P ' whose distance
from 0 is
- D
1>= .
y_ L2+1J2+()2
(8)
In these equations, the radical is to be taken with the positive
sign. The coefficients of x, ?/, z are proportional to X, p.., v in such
a way that the tl i recti on <osi nes of the normal t o the pla.ue are
fixed when the signs of A, B, Care known. But t he plane is not
changed if its equation is multiplied by - 1, hence the position
of the plane alone is not suftlcient to determi ne t he direction of
the normal. In order to dtlfine a positi ve and a negative side of
a plane we shall first prove the following theorem :
TnEORE)I. T1w 7winl.s P
1
P
2
are on the same side or on
sicles of the phwe A.t + Ry + Oz + D = 0, accorcliug as their courcli-
nntes mt(ke ml'mi;er of the erJU(( /i on of tlll' Jdaue lt ace like or
u,li/,e
lG PLANES AND LINF.S [CHAP. II.
For, let P
1
=(Xt, y
11
z
1
) ,

y
2
, x2) be two points not lying
on the plane. The poi nt P :=(x, y, z) in which the line P
1
P
2
inter-
sects the plane is determiued (Art. 6) by the values of m
11
m
2
which satisfy the equation
+ B!h + Oz
2
+D)+ m
2
(AXt + Bt!t + Oz, +D)= 0.
If A.1, + fl!Jt + Oz, + 1J and .d.v
2
+ B!t
2
+ Oz
2
+ D h:we unlike
signs, t hen ?/1.
1
and m
2
have the same sign, and the poi nt P lirs be-
tween P
1
and P
2
If

+ B.'!t + Oz, + D r.nd + B,112 + Oz
2
+ lJ have the same sign, t hen the nHlllhers m.
2
have opposite
sigus, hence t he })Oiut Pis not between J>
1
aml P
2

When all the terms in the equation
A.1: + B!J +Oz + D =0
are t ransposed to the first membc, a point y
11
z
1
) will be said
to he on the posi ti vc side of the plane if A.v
1
+ B.lft + Oz
1
+ D is a
positi ''C num bcr; the point will be said to be on the negative side
i f this expression is a negative number. Finally, the point is on
the plano if the expression vanishes. It should be observed that
the ectnation Ill ust not be mu I ti plied by - 1 n.ftcr the positive and
negati ,e sidE's have been chosen.
Fro. 12.
1.'). Angle between two planes. The angl e
between two planes is eC),nal to the angle
between two directed normals to the planes;
hrnce, by Arts. !) and 14, we have at once
the following thc01em:
Tm;ottJm. The cosine of the angle (J be-
t ween ta:o plcwes
A.t:+By+ Oz +D=O,
A'x+ ll'y+ O'z + D'=O
is df!/iuerl by the eiJII (ttion
8
AA
1
+ BU' + 00
1
cos =
V ..d:! + + 0'1. VA':! + JJ'
2
+ 0
1
:!
(9)
In particular, the condition that the planes are perpendicular is
AA
1
+ UTl ' + 00' = 0. (10)
1.5, 161 DISTANCE TO A POINT A PLANE 17
The conditions that the planes are parallel are (Art. 3)
A. 13 C
.--t t = Bi = (5i "
(11)
The equations (11) are satisfied whether the normals have the
same direction 01 opposite directions. From the definition of the
between two planes it follows that in the first case the two
planes are parallel and iu the second case they make an angle of
180 degrees with each other. We shall say, however, that the
planes are parallel in each case.
16. Distance to a point from a plane. Let P = (xu !lu z
1
) be a
given point and Ax+ B!l + Cz + D = 0 be the erptation of a given
plane. The distance to P from the plane is equal to the distance
from the given plane to a plane through P parall el to it.
The equation
.l:c+ B!J + Cz- (.l.l;
1
+ B!J
1
+ 0z
1
)= 0
represents a plane, since it is of first rleg1ee with real coefficients
(Art .. 10). It is parallel to the given plane by ErJS. (11). Jt passes
P since the coonlinat<>s of P satisfy the equation. When
the equations of the planes are reduced to the normal form, they
become, rcspectivel .Y,
___ . 1 __ ,I;+ B !I
,/_.(2 + [ff + ()2 v. L2 + 8"2 + ()2
0 -D
+ ---z= '
v. L
2
+IP+C
2
v.-t
2
+11
2
+CY
A n
- -- --- X+
v71
2
+ Bl+0
2
v.P+IP+0
2

+ 0 z = Ax1 + il!h + Oz1
vA2+BZ+ 02 .Y. tz+-B2+ c
The second members of these two equations represent the dis-
tances of the two planes from the origin, hence the distance from
the first plane to the second, which is equal to the distanced toP
from the given plane, is found by sn Ltracti ng the former from the
latter.
18
The result is
PLANES AND LINES
d =AXt + B y t + Cz1 +D.
v' A:!+ B'!+ c:!
[CHAP. II.
(12)
The direction toP from the plane. along the normal, is positive
or negative according as the expression in the numerator of the
second membet is positive or nel;::ttive (Art.14), thn.t is, according
as P is on the positive or negative side of the pl:lne.
EXERCISES
1 . Reduce the et1uation 3 :r - 12 y - J z - 26 = 0 to the normal form.
2. Write the equation of a plane through the origin parallel to the plane
:r + 2y = u.
3. What is the distance from the plane 3 x + .J y - z = 5 to the point
(2, 2, 2)?
4 . Find the di;;tance between the parallel planes
2 :r- y + 3 z = 4, :c - y + 3 z + 5 = 0.
5. Which of the points (J, 3, 1), (1, - .J, ::), (3, r,, 2), (-1, 2,- 2),
(5, 4, G) arc on the same sitle of the pl:\ne 5 :c- 2 y - 3 z = 0 as the point
(I, a, - s)?
6. Find the coi>rdinates of a point in each of the dihedral angles formed
by the planes
3x+2y+5z-.J=0, x -2y-z +6 =0.
7. Show that each of the planes :!5 .r + :!!l y + Sz- .JS = 0 and 25 x
- S!l y + I 12 z + ll::l = 0 bisect a pair of l'ert ical dihedral angles formed by
the planes 5.r + 12 z + i = 0 and a y- 4 z- 0 = 0. Which plane bisects
tho angle in which the origin lies ?
8. Find lhe equalion of the plane which bisects th:.t angle formed by
the planes S;c-:ly+z-4=0, 2x+y-3z-:l=0, in which the point
( 1, 3,- 2) li es. 2 .. , ,
9. Fintl the equations of the plarH'S which bisect t.hc dihedral angles
formed by the planes A ,x + B1Y + r'1z + Dt = 0, + + C2z + = 0.
10. Finl the equation of the locus uf 1t point 1vhose distance from the
origiwls eqnal to its distance from t.hc plane !'l.r + y - 2:; = 11.
11. Write the equation of a plane whose from the poiut (0, I)
is :1, and which is perpeuliicular to the vccwr of tlte poi nt (2, - 1, - I).
12. Show that the planes 2.r.-y+z+S=0, x-y+4z=0, 3l+y
-2 z + 8 = 0, 4 .r- 2 y + ll z- 5 = 0, Ox+ 3 y- G z - 7 = 0, ancl 7 :r- 7 y
+ 28z- G == 0 bountl a parallclopipetl.
13. Write the equation of a plane through (1, 2, -1), parallel to the
plane x - 2y - z = 0, and find its intercrpt.o;.
ARTS. 17, 18) DIRECTION COSINES OF THE LINE 19
14. Find the equation of the plane passing through the points (1, 2, 3),
(2, - 3, 6) and perpendicular to the plane 4...; + 2 y + 3 z = 1.
15 .. Find the equation of the plaue through the point (1, 3, 2) perpen-
clicular to the planes
2 X + 3 y - 4 Z = 2, 4 X - 3 y - 2 Z = 5.
16. Show that the planes x+'ly-z=O, y +7 z - 2=0, 7 -2y-z
- -l = 0, x + 3 y + z = 4, and 3 x + 3 y - z = 8 bound a ljtladrilateral
pyramid.
17. Find the equation of the locus of a point which is 3 times as far
from tile plane 3 x - 6 y- 2 z = 0 ;ls from the plane 2 x- y + 2 z = 9.
18. Determine the value of m such that the plane nu + 2 y - 3 z = 14
shall be 2 units from tbe origin.
19. Determiue k from the condition til at x- ky + 3 z = 2 shall be peq>en-
dicular w 3J' + 4 y- :lz = 6.
17. Equations of a line. Let A
1
x + B
1
y + 0
1
z + D
1
= 0 and A
2
x
+ B2!1 + + D
2
= 0 be the equations of two non-parallel planes.
The locus of the two equations considered as simultaneous is a
line, namely, the line of intersection of the t\vO planes (Art. 10).
The simultaneous equations
A
1
x+ B
1
y + 0
1
z+D
1
=0,
Bi!J + 0
2
z + D
2
= 0
are called the equations of the line.
The locus represented by the equations of two parallel planes,
considered as simultaneous, will be considereu later (Att. 33).
18. Direction cosines of the line of intersection of two planes.
Let .\, p., v be the uirection cosines of the line of intersection of
the two planes
L,:.: . 1
1
x+B,y + 0
1
z+ D
1
=0,
L2 :.: Az-1: + B2y + 0-zZ + D
2
= 0.
Since the line lies in the plane L, = 0, it is perpendicular to the
normal to the plane. Hence, {Arts. 5, 14)
.\A
1
+ p.B
1
+ v0
1
= 0.
Similarly,
p.B2 + v02 = 0.
By solving these two equations for the ratios of.\, p., v, we obtain
A p. = v (13)
B,02- B20, 01A2- C'2A1 A2B1
20 PLANES AND LINES [CRAP. II.
The denominators in these expressions are, therefore, pwportional
to the direction cosines. In many ptoblems, they may be used
instead of the direction cosines themselves, bnt, in any case, the
actual cosines may be determined by dividing these denominators
by t he square root of the sum of their squares. It should be
observed that the equations of a line are uot sufficient to deter-
mine a positive direction on it.
19. Forms of the equations of a line. If A, J:L, v are the direction
cosines of a line, and if P
1
:=(x
1
, y
1
, z
1
) is any point on it, the
distance d frotn P
1
to another point P = (x, y, z) on the line satis-
fies the relations (Art. 'l)
Ad= x-xu J:Ld = !1- vel= z -Z
1

By eliminating cl, we obtain the equations
X- X
1
_ .lf- JIJ _ Z- Z
1
_,\ ___ J:L ___ v_'
(14)
which me called the symmetric form of the equations of the line.
Instead of the direction cosines themsehes, it is fteqnently
convenient to use, in these equations, three numbers a, b, c, prO-
portional, respectively, to.\, J:L, v. The equations then become
.c!- .l'a =!I- Ya = Z- Za.
Ct {J C
(15)
They may be reduced to the preceding forlll by dividing the de-
nominator of each member by Vct
2
+ b
2
+ c
2
(Art. 3).
lf the line (15) passes tluough the point P
2
=:(x
2
, y
2
, z
2
), the
coordinates of P
2
satisfy the equations, so that
On elimiuating a, b, c bet11een these equations and (15), we
obtain
X- X
1
= y- J/a = Z- Z1
z2-za
(16)
These equations are called the two-point form of the equations
of a line.
ART. 20] PARAl\IETJUC EQUATIONS OF A LINE 21
!!0. Parametric equations of a line. Any point on a line may
be delined in terms of a fixed point on it, the direction cosines of
the line, and the distance cl of the variable point from the fixed
one. Thus, by Art. 4
X= x
1
+ >..cl, y = y
1
+ p.d, z = Z
1
+ vel.
(17)
If >.., p., v are gi,en and (x., y
1
, z
1
) represents a fixed point, any
point (x, y, z) on the line may be defined in terms of d. To every
real value of d correspomls a point on the line, and conversely.
These equations are called parametric equations of the line, the
parameter I.Jeing the distance.
lt is sometimes conYeni<-nt to express the coordinates of a
point in terms of a parameter K which is defined in terms of cl by
a linear fractional equation of the form
l
- (( + (3K
( - ---,
y+8K
in which r(, (3, y, 8 are constants satisfying the inequality
((8- (3y * 0.
By substituting these valuE's of d in (17) and simplifying, we
obtain ecptations or the form
.t = Ht + b1k , If = a2 + z = "a + _!_Ia! (
1
S)
tt4 + 0
4
K Ct
4
+ 0
4
K ft
4
+ 0
4
K
1
in which a . ., u., etc., are constants. Equations (18) are called the
parametric equations of the line in tenns of the paxameter K.
It shonltl be observed that the llenominators in the second
memuers of equations (18) are all n.likP.. Each val ue of K fot
which a.+ u.K =# 0 determines a <IE'finite point on the line. As
<t
4
+ b
4
K approaches zero, thE' distance of the conesponcling point
from the origin inr.1enses without limit. To the value deter-
mined by " + u
4
K = 0 we shall say that there corresponds a
unique point which we shall call the point at infinity on the li ne.
EXERCISES
1. Find the pointS in which the following li nes pierce the coordinate
planes:
(a) :c+2y-3z=l, 3x - 2y+5z=2.
(b) 5x-3y+z=2.
(c) x + 2y-5=0,
22 PLANES AND (C'BAP. II.
2. Write the equationsof the line x+y-3z=(l; 2x-y+2z=i in
the symmetric form, the two-point form, the pammetric form.
3. Show that the lines I x + y- 3 z = 0, :! .t.- y + 2 z + 6 = 0, and 8 :r
- y + z = 1. 10 .. + y- -1 z + l = 0 are parallel.
4 . Write the equations of the line tluough ca. i I 3) and (- 1 I 5, G).
D<t<tmine its direction cosines.
5. Fiud thl' equation of the plane through the> point (2, - 2, 0)
and perpendicular to the li ne z = 3, y = 2 .r - .J.
6. Find the value of k. for which the lin!:!.'>'- !1 = Y .
1
= z +::land x-l
'lk k+l j 3
= Y + G =.: +
2
arc perpendicular.
1 k-2
7 . Do thE' points (2, .J, G), ( .J, 6, 2), (1, 8) lie on a line?
8. 1:or what \' alne of k are the points (k, - 3, 2), (:!, - 2, 3), (6, -1, I)
collinear?
9 . Is 1 here a value of k for which the points (k, 2, - 2), (2, - 2, k), mal
(- :!, 1, 3) are collinear?
10. Show that the line :r -
2
= Y + g = z-
3
lies in the plane 2 .t + y
:1 -1 I
- z+3 =0.
1 1 . In equations (18) show that. as I( approaches in!inity, the corr<'spond-
ing point approaches a ddlnite point as a limit. Does this limiting point lie
011 Ut e givl'n line :
1. Angle which a line makes with a plane. Given the plane
. l x + U!J + Cz + D = 0
and the l ine
- x
1
_ !I - !It _ z - z
1
_ t_L_ - - b- - -c- - .
The angle which the line makes with the plauc is the complement
of the angle which it makes with the non ual to the plane. 'l'he
direction cosines of the normal to the plane are proportional to
A, B, C and the tlirec:tion cosines of the line are JWOportionaJ'to
a, b, c, heuce the angle 0 between the plane and the line is de-
termined (:\1t. :)) by the formuln.
sin 6 =
+ [fl + ()2 v'(t2 + lJ2 + c2
rr..-1 + l1R + rC
-'------ (19)
ARTs. 21, 221 DISTANCE FROl\1 A POINT TO A LIXE 23
EXERCISES
1. Show that the plan<'s 2 x - 8 !I + z + 1 = O, 5 x + z - 1 = 0, 4 x +
n y- z - 5 = 0 have a line in common, and tine! its direction cosines.
2 . Write the equalions of a linr which passes through (5, 2, G) and is
parallel to the line 2 x - 3 z + y - 2 = 0, :r + y + z + 1 = 0.
3. Find thcanglewhich thelinex+y+2z=0,2x - y+2z-1=0
makes with the plane 3.r + (jz- .Jy + l = 0.
4 . J-'ind the Cl)llation Of the plane the point (2, -2. 0) and
pl'rpenclicnl:uto the line .c + 2 !1- 8z = 1. 2x- 3y + .J z = 0.
5. Find the equation of the plane letrnnincd by the parallel lines
x + 1 >J - 2 z :r - >/ + I z - I
-:l- = = )' -3- = '-:r- = -1-.
6. For what \'alue of k will the two lines x + 2 y - z + 3 = 0, 3 x- y +
2 z + 1 = 0 ; 2 .r- y + z - 2 = 0, .r + y- z + /,; = 0 intersect?
Ill' 7. Fincl the equation of thl' plane through the points (1, - I , 2) and
(:!, 0, I), parallel to the line .r + y - z = 0, 2 .r + y + z = 0. '
B. Show that the lines :r-
2
= Y +
1
= ....:._ anti ?:. -
3
= Y + .J = z +
2
:) -2 - 1 3 2
intersect, and find the cc1uation o the plane determined by them.
9 . Find the equation of th<' plane the poiut (a, h, c), parallel to
each of the !I-!!!= z- z,; = !1- 112 =
I, m, ltt m2 liz
10. Find the equation o[ the plane through the origin and perpendicular
to the line 3 ;r, - y + .J :: + r, = 0, .r. + y - z = 0.
11. Find the value of /,; for which the lines :r - :J = If +
1
= z -
2k k+l 5 I
,. _1 Y+5 z+:! I" I
-
3
- = -
1
- = k _
2
are perpem 1cu ar.
12. Find the values of k for which the plaues k.r- 5 y + (k + G)z + 3 = 0
and (k - l )x + ky + z = 0 are perpendicular.
13. Find the equations of the line through the point (2, 3, 4) which meets
the r-a.xis at right angles;
22. Distance from a point to a line. Given the line
L ' ,.
x - :>'1 = ?!-?It=
A p. v
and the point P
2
= z
2
) not lying on it. It is required to find
the distance betweeu the point and the line.
PLANES AND LINES !CRAP II.
Let P
1
= (xt, y., zt) (Fig. 13) be any
point on the line; let P be the foot of
the perpendicular from J>
2
on the line;
() the angle between the given line and
J--------x the line PtP
2
; let cl be the length of
the segment We have (Fig. 13)
y Fu;. 13.
P
2
P
2
= Pt Pl sin
2
() = cl
2
- d
2
cos
2
0.
The direction cosines of the line P
1
/'
2
are Xz ?lz ?It, Zz; zt,
from whi ch (Art.. r.)
()
3:. - II,, - lit z. - Zt
cos = A ---- + 1'- -'---' + II ----.
d d

J>Pl = rl2 - rl2co:;z () =(:!- .>"tlz +(!h - ?lt)z + (zz - z.)z
- (.A C"2 - .r,) + 1'- (!h - !It)+ v (z2 - zt))
2
(20)
:!:J. Dist ance between two non-intersecting lines. Gi vrn the
two liMs
= !I- !It = z - zt and .r - = !f- ?h = z- z2
At JJt lit ,\2 /'2 Vz
which do not intc rse<-t. lt i;; rc1p1ire<l to lint! tlae shor test dis-
tance between them. Let A, JL, v he the direct ion cosines of the
line on which the distance is measmed. Since thi s line is per-
penclieular to en,ch of the given lines, we have, by Art. 5,
Equations (-l) and (!1),
where () is t he angl e between
the gi,en lines.
The length rl of the requi red
perpentlit:nlar is cq nal to the
projection on the common per-
pendiculai of the segment 1'1'', Fro. 14.

and is equal to the projection of the hroken line l',l!SP' (Fig. 14).
ART:; . :!3, 211 S'l81'E:.f 01<' PLAXi':S TIIROUUH A LfXE 25
or
X
1
- ,\,
cl = y, - p., (21)
EXERCI SES
I d
. f h 1
1
. :r- 1 y - 3 z - 2
1 . Find t JC stance rom t e ong1n to t e me - - = --= --.
4 I
2 . Find the distance from ( I, I, I) w .r + y +:: = 0, 3.c - 2 y + 4:: = 0.
3 . Find the perpemlicular dist:tncc f.-om the point (- 2, I, 3) to the line

4 . \\'hat are the clire.-tion cosines nf tho line through the origin aml the
point of of the .r 2y-:: + 3 = 0, O.c - y + 2:: + I= 0;
2 .f - 2 y + ;3:: - 2 = 0, L - y - :: + ;j = 0.
5. l>etermin( the tlibtance of the point (I. I, I) to the line x = 0, y = 0
and th(' direction cosines of the line (lll which it is measUJed.
6 . Find the distance between the lims .r = Y +
2
= z -
1
and x....=_!
2 -2 I 4
_y-3_::+1
- :!-
7 . Fintl the equations of the line aloug which the lli>.lauce in F.x. ti
is
8 . Find the between the lines 2 .r + y- :: = 0, :r- y + 2:: = 3
and x + 2 y - :1:: = I, 2 x - 3 y + 4 :: = 6.
9. Exprel>s the condition that the lines .c- .t! = = z- z,, .r- .ct
/1 fill Ill /2
= Y- 1/J =::- ::
2
intersect.
')t2 lt2
-24. System of planes through a line. If
L
1
= ..d
1
.c + Bi!J + O,z + D
1
= 0,
L2 := + B:!J + + = 0
ate the equations of two intersecting planes, the equation 7.:
1

1
+
k
2
IJ
2
= 0 i:;, for all real values of 7.:
1
and 7.:
21
the equation of a plane
passing through the line L
1
= 0, = 0. For, k
1
L
1
+ k
2
!AJ = 0 is
ahvays of the first degree with real roefficients, and is therefore
the equation of a plane (Art. 10); this plane passes tluongh the
line L
1
= 0, L
2
= 0, since the of every point on the li ne
satisfy L
1
= 0 nnd L
2
= 0 and consecptently satisfy the equation
26 PLANES A:-\D [CIIAI' . IT.
kiLl+ = 0. Conversely, the Ci}Uation of any vlanc passing
through the line can be expressed in the form k
1
L
1
+ k
2
L
2
= 0,
since k
1
antl can be so choseu that the plane

+ k
2
L
2
= 0
will contain any point in space. Since any plane through the
given line is llcterminml by the line and a point not lying ou
it, the theorem follows.
To find the equations of the plane dl:'tcrminecl by the line L
1
= 0,
L
2
= 0, and a point P
1
not lying on it, let the coordinates of J>
1
he
y, z
1
) . [f P
1
lies in the plane k
1
l,
1
+ k
2
L
2
= 0, its coordinates
must sa.ti;;f_y the el}nation of the plrmP; t hus
k
1
(.-1
1
Y
1
+ B
1
!t. + (f
1
z
1
+D
1
) +

+ B211t +

+ 1J
2
) = 0.
On eliminating k
1
and k
2
bctweOJJ this equation and k
1
L
1
+ h.j,
2
= 0,
we obtain
0 = (.'1"1'1 + lJ2?ft + + D2)(Atx + Bty + 0 1z + D,)
- (A,.Jl + nl!h + Clzl + !JJ) (. 12'1' + B'l'l + 0Z + Dt),
as the equation of the pl:tne dctennined by the line L
1
= 0, L
2
= 0,
and the point P
1

It will be con\cuiont to write the above equation in the abbre-
viated form
Lz(.t'
1
)L
1
(.c) - Lt(.r
1
) f ,,(.c) = 0.
The tolality of planes passing tlllongb a line is callecl a pencil
of planes. Tho number k
2
/ k
1
which determines a plane of the
pencil is called the paramett>r of the pencil.
If, in t he er1tw.tion
J.:1L1 +

= 0,
k
1
and are givt'n such ,alnes that the coefficient of xis equal to
zero, the corrtspon<l ing plane is perpendicular to t he plane x = 0.
Since this plane contains the line, it intersects the plane x = 0 in
the orthogun:tl projrction of t he line D
1
= 0, L
2
= 0. Similarly,
if k
1
and k" are given such va.lnes t hat. the coeflicicnt of y is equal
to zero, the corresponding plane is pPrpcndicular to the plane ,1/ = 0
ancl will cnt the plane y = 0 in the projPction of /,
1
= 0, = 0 on
that plane; if the coolllcicnt of z is made to ,:-wish, the plane will
contain the projrction of the given line upon the plane z =0. The
three planes of the system k
1
T,
1
+ k
2
J,
2
= 0 obtained in this way
are callr<l the three projecting planes of the line L
1
= O, L
2
= 0 on
the coor<linatc plaucs.
ART. SYSTE:\ OF T HROUGH A LINE
Since distinct planes
passing through a line are
sufficient to clf'termine the
line. t\\'o projecting planes of
a line may always be em- .
ploycd to define the line. If L:::.... _____ _
z
0

the li nc is not )Xtrallel to the
plane z = 0, its projecting
planes on x = 0 and !J = 0 are
distinct n.nd the erptations of
Ftc. t;i.
the line may be reduced to the furm (Fig. 15)
,I'= mz + 11, !/ = liZ+ b.
27
X
(22)
If the line is parallel to z = 0. the value of k fot which the coeffi-
ciPnt of,. is mn.dP to nmish wi ll also rC'(luce the co<>fficient of !J to
z zero. so that the projPrting planes on
.1: = 0 and on !/ = 0 coincide'. This
proje(ting plane z = 1' and the projec-
ting plane on z = 0 may uow be chosen
to define the line. If the line is not
------------x parallf'l to the X-axis, the ef)uations
ur the line mn.y be reduced to (Fig. lG)
y
rw. tr.. X = Jl!J + C, Z = C.
Finally, if the line is parallel to the X-n.xis, its equations may be
reduced to (Fig. 1 7)
?!= u, z =c. (3-1)
If the planes L
1
= 0, = 0 are par-
allel but distinct, so that
z
y
then every <'qnation of the fonn k
1
! ,
1
+ k
2
[ ,
2
= 0, except whe n
_ = A
1
= BB
1
= ee
1
, defines a. plane parallel to the given ones.
kt Az 2 2
j Conversely, the eCJuation of any plane to the given ones
\.can be written in th.:l fol'm k
1
L
1
+

= 0 hy so choosi ng /,
1
: k
2

28 PLANES AND LINES [CBAP. II.
tlaat the plane will pass through a given point. In this case the
system of planes k
1
L
1
+

= 0 is called a pencil of parallel planes.

Two equations
L,:= A
1
.-c+B
1
y+ O,z + D
1
= 0,
L
2
.;=. A:l'1; + Bt!J + + Dz = 0
will represent the same plane when, and only when, the coefficients
A
1
, B
1
, Oh D
1
are respectively proportional t.o A
2
, B
2
, Ov D
2
; thus,
when
Aa B. 0, D1
A
2
= B
2
= 0
2
= Dz
These conditions may be expressed by saying that every deter-
minant of order two formed by any square array in the system
shall vanish.
In this case multipliers k
1
, k
2
can be fonnd such that the equa-
tion k
1
!-
1
+ 1.:0
2
= 0 is identically satisfied.
Conversely, if mul t ipliers 71:
11
k
2
can be found such that the pre-
ceding identity is satisfied, then the equations L
1
= 0, L
2
= 0
define the same plane.
EXERCISES
1 . Write the equation of a plane through the line 7 x + 2 y - z- 3 = 0,
3x- 3y + 2z- 5 = 0 perpendicular to the plane 2.r. + y- 2z::: 0.
2. What is the equation of the plane cleterrnine<l by the line 2 x - 3 y-
z + 2 = 0, x- y + l z = 3 ancl the point (3, 2, - 2)?
v 3 . Determine the eqnation of the plane passing through the line
'l: + 2z = 4, y- z = 8 and parallel to the line x -
3
=!I+
4
= z-
7
.
l I 2
1
4. Does the plane '.T + 2 y - z + = 0 have mor(' than one point in
common with the line 3x- ?I + 2 z + I = 0, 2 x- 3 y + 3 z- 2 = 0?
V 5. Determine the equations of the line through (I, 2, 3) intersecting the
two lines '.T + 2 y-3z=0. y- -l z=4 :llltl 2, - y + !3 z::::l, S '.T + y + 2z + I =0.
Application in descriptive geometry. A line may be repre-
sPntecl by the three orthogonal projections of a segment of the line,
each drawn to scale. Consider the XZ-plane (elevation, o1 vlrti-
cal plane) as the plane of the paper, ancl the xr:.plane as turned
about tlw .\-:1xis unti l it coincides with the XZ-plane. The pro-

ARTS. 2J, 26) BU:NDl;ES OF PLANES 29
jections in the xr-plane are thns drawn to scale on the same
paper as projections on the XZ-plane, but points are dist.inguished
by different symbols, as P', 1\. Q, __ _ ,
The X }:plane is caJled tho plan ---7j Q
or horizontal plane. Finally, let li 'fp -- --{p' l
the rZ-plaoe be turned about the . i' p ' I
Z-axis unti l it coinciil.es with the ! . ! j
XZ-plane, and let figures in the X
new position be drawn to scale. \
0
J
This is oalled tho om\ o pofilo '-.._ P
1
I I Q
plane. Thus, in the figure (Fig. _
18), a segment J>Q, wherein y 1
P = (7, 4, 8), Q = (13, !), 1:3), Fro 18.
ma.y be indicated by the three segments P'Q', P
1
Q
11
PPQr>.
ExA)IPI.E. Find the cqnations of the planes of lhe line
2 X + 3 y - I z = 5, - 4 y + 5 z = 6.
II ere, /,
1
= 2 x + 3 y - 4 z- 5, = x - 4 y + 5 z- 6,
k1L1 + = (2 k, + k2)x + (3 k 1 - 4 k2 )?1
+ (- ! kt + 5 k2)z + (- i) k1 - 6 k2) = 0.
If k
2
= - 2 k
11
the Clletlicienl of x disappears; thus the equation of the
plane projecting the given line on the plane .r = 0 is
I I y- H z + i = 0.
If
k 3 I ffi . f . I 1 . ' I 0 .
-"= -, t te coe ctent o y Yams t te prOJect tn:.: pane on y = 1s
kt .j
fonnd to be 11 x - z = 38.
Finally, if k
2
=
4
, the prnjectin:; plane nn z = 0 is found. Its equation
kt 5
is H x - y =HI.
EXERCISES
I
Find the c-4u<1tinns or the projecting plaHC:; uf each of the following lines:
1. :r + 2 y - z = 2 .r - !} y + z = r,,
2 .r + y + z = 0. :>' - y + 2 z = !},
3. .l' + !f + z = 1. ,,. - y + 3 z = -L
4 . It + R,y + r'1z t n
1
- 0, . l"f + + + = 0.
26. Bundles of planes. 'l'lu pln.nl' {,
3
= .la-l' + l3
3
y + 0
3
z + D
3
= 0 will belong to the peneillletenniuPd by the plaues L
1
= 0, L
2
= 0,
assumed distiuct., when thrE'e number:; kl' k
3
, not all zero, caa
be found such that the equation k
1
L
1
+ k
2
L
2
+ k
3
L
3
= 0 is ideuti
30 PftES AND LINES (CaAI' . ll.
cally satis fied for all val ues ar y, z. This condit ion requires that
t he four equations k
1
.1
1
+ k
2
A
2
+ k
3
A
3
= 0, k,B, + k
2
B
2
+ k
3
B
3
= 0,
k, O, + k
2
0
2
+ 7.:.
3
0
3
= 0, k, D, + k
2
D
2
+ k
3
D:. = 0 are satisfi ed by
three numbers k u k
2
, k
3
, not all zero; hence, that the f olll' equa-
tions
are all sat isfi ed, wherein we have wri t t en for brevity,
1.1, JJ, u.

= . 1
2
B
2
0
2
, et c.
, ta Da ('3
These simnl ta.ueous ma.r he l'Xpressecl hy saying
that every determinant of onlet t hn'e f ormed by the elements
contained in any srpuu e array in tho syst em
A
1
B
1
G'
1
D
1
1
A z llz Gz IJ2
A3 na ('3 v 3
shall Yanish.
Convers<>ly, if these conditions a re satis fi ed, then three con-
stants k., k
2
, k
3
can be found suc h t hat; t he equat ion k.
1
L
1
+ k
2
L
2
+ kaLa= 0 is itl enticall y sat isfied, a nd the t hree pl anes L
1
= 0,
L
2
= 0,
3
= 0 bel ong to t he sa111 e peucil.
Let L
1
=: .A
1
. 1: + IJ,y + C
1
z + D, = 0,
L 2=: . 12-1: + lJ:!lf + Cc.z + D2 = 0,
L
3
=: A
3
.v + B
3
y + ('
3
z + D
3
= 0
be t he er1uations of t hree planes not bel onging to a pencil. 1f we
solve these t hree ecpw.Lion:; for (.v, y, z), we find f or t he
of t he point of in tersection of t he three planes, in case ' ..:l,B
2
C'
3
:;1= 0,
. I D,B20.1 1
' = - [ . I.Hz0a l '
lf . 1
1


= 0, b11t not a ll Lhc> !l ('tl' rlllinants in the Humet ntors
of atP zero, no SPt of vn.ht l'S of .1, ?f, z will satisfy all t hree
C'rptations. In thi:> case, t ho line of in tNscctiou of any t wo of t.hc
planes is parall el to the t hird. l'01. if / ,
1
= n a n1l = 0 i11tcrsect,


ARTS. 2G, 2i) PLAXE COORDIX.\TES 31
the direction cosines of their line of intersection are proportiOnal
(.Art. 18) to
The coutlition that this line is parallel to the plane L
3
= 0 is
(Art. :? 1)
A
3
(JJ,Cz -JJ2C
1
)+ D3(C
1
. l
2
- 0
2
. 1,) + C'
3
(..: 1
1
B2 - A zB
1
)= 0,
which is the condition

= 0. The proof for the
other lines and planes is found in the same wa.v.
If at. least one of the determin:utts

I D, B2C'
3
I, l.l,Dz031,
and I A
1
BtlJa is uot zero, the system of planes
k
1
L
1
+ k
2
J,
2
+ k
3
L
3
= 0
is called a bundle. If I A BCI * 0, all the planes of the bundle
pass through the point since the coordi nates of this point
satisfy the equation of every plaue of the bnudle. Conversely,
the el}nation of every plane passing thtongh the point (:!.1) can be
expressed in I his form. This point is callecl tlte vertex of the
bundle. If j.IBUI= 0, all the plaut>s of the bundle are parallel
to a fixE'd lim (such as J,, = 0, [,
2
= 0). In this case, the bundle
is railed a bundle.
Plane coordinates. The t>quation of any plane not passing
through the origin may be re,Jnce1l to the fonn
liX + (!f + lt'Z + 1 = 0. (:?G)
When the equation is in this form, the position of the plane is
fixecl when thl values of the cne!Hcilnts 11, r, w (not all zetu) :ue
known; aut! COJl\'etsely, if the positi()n of the plane (not pa:;sing
through the ol'igin) is kno'' n, the values of the coefficiE'Ilts ::ue
fixed. in<:e the numbers (11, c, It') determine a plane dE'finitely.
just as (.c, !I z) determine a point, \H' shall call the set of num
bers (u, ,., a) the coordinates of t he plane represented Ly E'quation
(::?G). 'flms, the plane (:>, n, ::?) will be mHierstood to mean the
plane whose Ct}Uation is 3 x + :i !I+:! z + 1 = 0. Similarly, the
equation of the plane (::?, 0, -1) is .r- z + 1 = 0.
If u, v, ware different from zero, they are the negative recipro
cals of the intC'rcepts of the plane (11, v, w) on the axes (Art. 13).
32 PLA::\ES A}JD LINES [CHAP. II.
If u = 0, the plane is parallel to the X-axis ; if ~ ~ = 0, v = 0, the
plane is parallel to the Xr-plane. The vanishing of t he other
coefficients may be interpreted in a similar way.
28. Equation of a point. If the poi nt (xu y., z
1
) lies in the
plane (26), the equation
1t."CI + 'll!}! + 1t"Zt + 1 = 0 (37)
must be satisfietl. I f x
1
, ?h, z
1
are considered fixed and u, v, w
variable, (27) is t he condition that the plane (u, v, w) passes
through the point (xu y
1
, z
1
) . For this reason, equation (27) is
called the equation of the point (x., y., z
1
) in pln.ne coordinates.
'l'hus,
is the equation of the point (1, - 5, 2); similarly,
3n+w+1=0
is the equation of the point (3, 0, 1).
If equation ( ~ I ) is multiplied by any constant different from
zero, the Joens of the equation is unchanged. Hence, we have
the following t heorem:
'rllEORJ:)J. '1'/te linear equatiou
.du+ Bv+ Ow+D=O (D=t=O)
l t {t' . (A R
0
) . I z t
1s t 1.e equu 1011 ~ 11 e powt U. D' D tn 1) ewe coorc ma es.
'l'hus, 'tt- 5 v- 3 w- 2 = 0 is the equation of the point
(
-1 r; 3)
T ' ~ , z
The coJHl ition that the coordinates (u, v, a) of a plaue satisfy
two linear equations
ux
1
+ 'ry
1
+ wz
1
+ 1 = 0,
is that the plane passes through the two points (4, y
1
, z
1
) and
(:r-z, y
2
, z
2
) and therefore through the line join i n ~ the two points.
The two equations arc called t he equations of tbe li:le in plane
eooJtlinates.
I . ( (.
ARTS. 28, 2!)) HO.MOGENEOU::i COORDI NATES 33
EXERCISES
,;
1 . Plot the following planes and write their er1nations: ( 1, 2, } ), (3, - !,
-!),(-l,t, -D
2. Find the volume of the tetrahedron bounded by the coordinate planes
and the plane (- l, -}, - !).
3. What are the coordinates of the planes whose equations are
i X+[) y .!:"z + I = 0, :r- () y + 11 z +.) = 0, \) .r- 4 = 0?
4. Find the angle which the plane (2, li, 5) makts with the plane
(- 1, ! 2).
...
5. Write the equations of the points (1, 1, 1) , n, (li, -2, I ).
6. What are the coiirdinntes of the points eqnation:s are
,

11'7. Find the direction cosines of the line
:} 11 - v + 2 w + 1 = 0, 1t + r; v + 2 - I = 0.
8. What Joens is determined by three simultaneous linear equations in
( u, v, to) ? " 1
""9. Write the equation satisfied by the coiinlin:tte,; of the planes whose
distance from the origin is 2. \\'hat. is the locus of a plane which satisfies
this condition ?
29. Homogeneous coordinates of the point and of the plane. lt is
sometimes convenient to express t he coordinates :c, y, z of a point
in terms of four numbers x', !!', z', t' l.>y means of the equations
x'
/!=X,
,,.
- =y
t' '
z'
- =Z.
t'
A set of four numbers (,;, y', z', t'), not all of whith arc zero, t hat
satisfy these e<luations a te said to he the homogeneous coordinates
of a point. If the coonlinates (:c', y', z', t') arc gi ven, the point
is uu1f]ltely dctel'luincc:l (for tlH' case t' = 0, compare Art.
but if (:t:, y, z) arc gi,cn, only Lhc ratios of the homogeneous
coordinates are determined, since (.t', !t', z', t') and (k:c', ky', kz', kt')
define the same point, k being au arbitrary constant, different
from zero.
Similarly, if the coordinates of a plane are (tt
1
v, tv), four num-
bers (u', v', w', s'), not all of which are zero, may be found such
that
v'
-=v,
s'
tv'
- =tV.
s'
34 PLA!'-:BS AND LINES
The set. of numbers (11', v', w', s') are called the homogeneous coordi-
nates of the plane.
Where no ambiguity arises, the accents will be omitted from
the homogeneous coordinates.
3 0. Equat ion of a plane and of a point in homogeneous coordinates.
If, in the equation
. .cl.t + By+ Oz + D = 0
( D =F 0, aucl A, B, 0 are not all
zero) the homogeneous coord i-
nates of a point are sub:;titute<l,
we obta.in, after multiplying by
t, the eqnation of the plane m
homogeneous coordinates
A.t+B!!+ Oz+ Dt=O.
The homogeneous coordinates
of this plane nrc (A, B, 0, D).
If, in the ecp1ation
An + Bv + Ow+ D = 0
( D =F O, :tml A, B, 0 a1e not all
zero) the homogeneous coordi-
nates of a plane are substituted,
we obtain, after multiplying hy
s, the equation of the point in
homogeneous coordinates
Att + Bv + Ow+ IJs = 0.
The homogeneous coonlinates
of this point are ( .. J, B, 0, D).
3 1. Equation of t he origin. Coordinat es of planes through the
origin. The necess:try :mel sufficient contlition that the plane
whose e1p1ation 11.1: + 1"!/ + wz + :;t = 0 shall pass through the
origin is s =0. " ' e sec then 0 is the equation of the
origin, and th:tt (11, e, w, 0) are thc homogeneous roortl inatrs of a
plane through the origin. s = 0, it follows fron1 Art. that
the non-homogeneon::; coordinates of sul'h a plane cease to exist.
3 2. The plane at infinity. Let (.r, y, z, I) be l11e homogeneous
coordinates of n point. If we :tssign fixed values (not all zero)
to :t'. y, z and allow I to var;v, the corresponding point will vary in
snc-h a way that, as t = 0, one or lllore of the non-lwmogeneous ro-
ordinates of the point incrl:!ascs without limit. If t = 0, the non-
homogeneous coordinates cease to exist, bnt it is assumed that
there still exists a corresponding point which is sni<l to he at
infinity. It is also :tssnmetl that two points at intinity coincide
if, and only if, their homogeneous cot>r<linatcs :uo proportional.
The equation of the locus of the points at infinity is I =0.
:Sinc.e this rqmttion is homogrueous of the fin;t degree in :t, .1/, z, t,
it will he s:t.i1l lhat t = 0 is the cqunlion of a plane. This plane
is called the plane at infi nity.
ARTS. 33, 34) COORDINATE TETRAHEDRON 35
33. Lines at infinity. Auy tinite plane is said to intersect the
plane at iufinit.y in a line. This line is ca,lled the infinitely dis.
taut line in the plane. The equations of the infinitely distant line
in the plane .A.t + B!! + Oz + Dt = 0 are Ax + By + Cz = 0, t = 0.
'l'he conclition that tll'O finite plaues Gtl'e z>ctrallel is
that they intersect the }Jla11e ut i11ji11il!J in the li11e.
If the planes are their eqnati ons may he \VI'itten in
the form ( r\.l't. l:"i)
Ax+ JJy+ Gz+llt =O, .A:c+By+ Cz +D't= O. (28)
It follows that they both pass through the line
.A.t +By + Oz = 0, t = 0. (29)
Comersely, the oqnati ous of any two tiui te planes through the
line (29) may be written in the form The plan('s a1e there.
fore parallel.
34 . Coordinate tetrahedron. The four planes wi10Se ec1uations
in point coordinates a1e
x= 0, y = 0, z=O, t=O
will be called the fom coonlinate planes in homogeneous coordi-
nates. Since the planes do not all pass through a common point,
they will be reg:udod as fo1mi ug a tet1ahcclron, called t il e coordi-
nate tetrahedron. 'l'he coordinates of t he verti ces of this tetra-
hedron are
(0, 0, 0, 1), (0, o, 1, 0), (0, 1, 0, 0), (1, 0, 0, 0).
The coordinates of the four faces in plane coo1dimttes are
(0, 0, 0, 1), (0, 0, 1, 0), (0, 1, 0, 0), ( l, 0, 0, 0).
The equations of the Yerticrs are n = 0, v = 0, H' = 0, = 0.
EXERCISES
1 . :Find the coordinates of the following poi uts and
planes:
(a) 7 x + 3 y + :3 z- -1 = 0, "' (cl) (l tt- 11 - 81o + 2 = 0,
(b) 10x - 3y+Hi=0, (e) u+v-w-7=0,
(c) {f)2 w+ l l = 0 .
.I- 2. Uetermint' Lbe coonJinaLeS of tbe iufinitely distant poillL ou the line
3 :c -tty + fJ t = IJ, t f - IU z + ;: t = 0.
,,
3G
PLANhS AND LINES [CHAP. H.
V3. Show that if L
1
(tc):=A1u+ Btv+ Ctto+Dts=O, and
+ B
2
v + C
2
to + Dzs = 0 are the equations of two points, tile equation of any
point on ll.e joining line may be written in the form k1L1 + kzLz = 0.
4. 8how thn.t the planes :r + 2 y + i z - 3 t = 0, x + 3 y + G z = 0, x + 4 y
+ 5 z - 2 t = 0 determine a parallel bundle. Fiud the eqnMion of the plane
of thr bumllr through the points (:.!, - 1, 1, I), 5, 0, 1).
35. System of four planes. The condi Lion that four given planr:s
L
1
=: A
1
x + B
1
y + C1;"+ Dll ,;, -o,
L2 =: B
2
y + 0
2
z + fJ,.t = 0,
L, = .

+ B
3
y + C3z + DJt = 0,
= + + + = 0
. ' ,,.J, of;
all pass through a. point)is tl1at four nu1nbe1s (.t, y, z, t), not all
zero, exist whit:h satls(y the four simnlta.neous equations. The
condition is, consequently, that the <leLerminaut
AI nl cl nl
Az Rz Oz D2
is equal to zero. If this condition is not satisfied, the four planes
a.re saitl to be independent. When the given planes are independ-
ent, four numbers /.:
11
/.:
2
, 1.:
3
, 1.:
4
can alw;tys be found such that the
equation
k
1
L
1
+ k
2
L
2
+ k
3
L
3
+ 7.'
4
1,
4
= 0
shall repnscnt a,ny given plane. For, let ax+ by+ cz +<I= 0 be
the CIJUation of the gi,cn plane. The two equations will repre-
S('nt tile sa1n e pl:tne if their coefilcirn ts arc proportional, that is,
if numbers k
1
, k, k
3
, k., not all zero, can be found such that
a= k
1
.:1
1
+ k
2
A
2
+ k
3
il
3
+
b = k
1
R
1
+ k
2
B
2
+ k
3
7J
3
+

c = k
1
0
1
+ k2Cz + k3C3 +
d = k
1
D
1
+ k
2
D
2
+ k
3
D
3
+

Since the planes are independent, the determinant of the coeffi-
cients in the second mem hers of these equatious is not zero, and
the numbers ku k
1
, 1.:
3
, can always be determined so as to satisfy
these equations.
ART. 35] SYSTEM OF FOUR PLANES 37
These results, together with those of Arts. 2.J., 26, may be ex-
pressed as follows: The necessary and sufficient condition that a
system of pl:tnes h:t\'C no point in common is that tlle
formed by their coefficients is of rank four; the planes beloug to
a bundle when the matrix is of rank three; the planes belong to
a penci l when the matrix is of rank two; finally, the planes all
coincide when the matrix is of rank one. We shall use the ex-
pression u rank of the system of planes" to mean the rank of the
matrix of coefficients in the equn.tions of the planes.
'
EXERCISES
1. Determine the nature o the following systems of plaues :
....-(a) :?x-5y+z-3t=0, x+y+h-5t = 0, :r+3y+6z -t = 0.
(b) 3:t+ 4y+ 5z - 5t=0, U:t+5y+!lz-10t=0,
- 5 l = 0. X - 11 + :? Z = 0.
(c) 2x+4y=0, 3.c+ -111 -2.r+:1t=0, :r=O.
( <l) 2 X + 5 11 -t- 3 Z = 0, j y - 5 Z + .J t = 0, .c - y + -I Z = 8 t.
L--2 . Show that the line _,. + :1 11- z + t = 0, 2 .r - 11 + 2 z - 3t = 0 lies in
the plane i x + i y + z- 3t = 0.
V3. Determine the conditions that the !Jlanes
,. = ty +liz, y = ar + cz, z = bx + ay
shall ha"e just one common point; a cowmou line ; are identical.
4 . Prove that the planes :! X - :3 11- 7 z = 0, 8 :r - 14 y - ];) z = 0,
8 x- 31 y- 33 z = 0 have a line in COlllmon, and find direction
5. Rhowthat the planes 4x-2z-2t=0, 4 x+-ly
- f> z = 0 belong to a parnllel lluau.lle.
Auy rectangular array uf numbers
.1
1
JJ
1
1'
1
/J
1
,1/
1
( .112
-"3 Us ' 's /Is .1/a
is called a matrix. Associr<ted with every matrix are other matrices obtained
by ;;oppressing one or more of the rows or one or more of thtl col11111ns of the
giYen matrix, or both ; in parth:ular, associated with ev1ry ;,otuare auatrix, that
is, one in wbicll the number or mws is to the numher of columns, is a de-
terminant whose elements are the elements or the matrix. Cotwersely, associat.,d
IVith e\ery determinant is a square matri x, formeol it s tleanents. " 'e shall
n..e the word rank tu the otder of the nou-vanishiu)( tleteruoinant uf high-
est order contained iu auy matrix. The rank or thP is
as the rank of the wnt rix roruaetl hy the l' lcu1euts uf a he tlet erauinaut.
CHAPTER III
TRANSFORMATION OF COORDINATES
The coordinates of a point, referred to two di fferent systems
of axes, are connected by certain relations which will now be
uetermiueJ. The process of changing from one system of axes
to another is called a t ransformation of coordinates.
:u;. Translation. Let t he coordiuates of a point Pwith respect
to a set of l'ect:wgnl:u axes OS, 0 1', OZ be (x, y, z) and with
respect to a sel of axes 0' X', 0' J'', 0' Z', parall el respectively
to the first set, be (.i, Jl, z'). If the coorJiuates of 0', rcfened
to the axes OX, 0 l", OZ :ue (It, k, l) we have (Fig. 19)
.c=.l:' +h, Y=?l +lo, z=z' +l. ( 1)
For, the projection 011 0 X of 0 P is cq ual t.o the stun of the prO-
jections of 00' and 0' 1' (Art. bnt tbe projection of OJ> is x,
y
z z' of 00' is h, and of 0' pis .l:'; hence
P x = x' + lt . The other for mulas are

,.,;: I
derived in a si milar wa.y. Since
tlH new axes can be obtained from
,/" .
the ol<l ones by moving the three
coordinate planes parall el to the
y
X-n.xis a distance h, then parallel
to the y:.axis a distance !.:, and
Frl'l. 1!1. parallel to the Z-axis a distance
l, without changing tll('i r tl i1ections, the t ransformation (1) is
called a t ranslation of axes.
3 7. Rotation. Let the coordi nates of a point P, referred to a.
set of rectangular axes OX, 0 Y, OZ, be x, y, z, acu referred to
a.uother rectangular system OX', 0 J"', OZ' having tht same origin,
be x', .'}', z'. Let x'= OJ/, Jl'= l /.1! ', z' =,1l' / '(l<'ig. and let
the direction cosines of 0 X', referrecl to OX, 0 OZ, bt> ;\., p.
1
, v
1
;
those of 0 Y' br. i\
2
, p.
2
, v", a.nd of OZ' be A
3
, JJ-3, v
3

3t:!
AnT. :ri) ROTATION
We shall show that
x = ,\lx' >-az',
y = P.lx' + + P.az',
Z = v
1
.t
1
+ + vaZ'
39
(2)
For, the projection of OP (Fig. on the axis OX is x. The sum
of the projections of 0 !/,
J/J/', and Jf'P is l\
1
.r'
+>-az'.
That these two exvres-
sions are equal follows from
Art. :t. The second ancl
third e(1uations are obtained
in a similat way.
The direction cosines of
OX, 0 ancl OZ, with re-
to the axes 0.\', 01"',
OZ' are >-u Jl.2, A.a; P1' p.z, fJ.ai
v
11
v
2
, v
3
, r<>spcctively. If we
z
Fw. 20.
project OP and OL = x, /,.1[ = y, and 1l[f' = z on OX', 0 Y', and
OZ', we obtain
,J:' = >..l.l; + P.I!J + viz,
y' = >..
2
x + p...!/ + vaZ,
z' = .\
3
.r + p.
3
,1t + v
3
z.
(2')
The of etnatious (:!) and arc expressed in con-
,enient fo1m by tuenns of the accompanying
x' y' z'
- - --- --- ---
:r
>..I ,\.! A a
--- - ----
!J P.l f.L2 JJ.a
--

--
z
VI va
I
Since >..
1
, p.
11
v
1
; Jl.
2
, p..
1
, v
2
; X
3
, P-3 v
3
a,re the <lirection cosines of
three mutually perpendicular lines, we have the six relations
).,12 + fJ-12+ Vl2 = 1,
At2 + fJo22 + y22 = 1,
>..a2 + P-32 + vat = 1,
A1A2 + /J.1!'2 + V1V2 = 0,
A2Aa + fJ.'l}J.a + v2va = 0,
Aa\1 + fJ.aP.I + vav1 = 0.
(3)
-
40 OF C'OOIWJ}JATES [CRAP. Ill.
\\\> h:wu :;et:ln that ,\
11
..\
2
, A
3
; fJ.
1
, P-3; v
11
v
2
, va are also the di-
rection cosines of three mutually perpendicula.r lines. It follows
that
..\1
2
+..\l+'Al=l,
fJ.I2 + J-'22+ fJ.l = 1,
vl2 + v22 + v32 = 1,
It witluext be shown that
,\lfJ.I + A2f1.2 + A3fJ.3 = o,
fJ.1V1 + J-'2112 + fJ.sVa = 0,
111..\1 + v2A2 + va.X.a = 0.
>..
1
= <(f1.211a - V'!.JJ.s), A
2
= <(fJ.aV1 - VsfJ.1), A3 = E(fJ.1v2- v
1
f1.2),
(4)
fJ.I = ( ( VzAa- A2va), 1"'2 = ( (v3'\l - Aavl), fJ.3 = ( (111A2 - XI liz), (5)
V
1
= E (Azp.a- JL2Aa), 112 = E (..\3/l.I-/J.J'\), v3 = < (..\1P.2- fJ.JAz),
where <= 1. From the first autl thi rd equations of the last
column of ( ..J.) we obtain
..\1 Az ,\s
fl.2113 - Vzl'-3 fJ.alll - llafJ.I /J.I 112 - Vtfl.z
If we denote the value of these fractions by<, solYe for >..
1
, A.
2
, and
>..
3
and substitute in the first of equations ('1), we obtain
E
2
V2fJ.a)
2
+ VJ.all1 - Vafl.t)
2
+ (JJ.Ivz-

= 1.
Since the lines 0 l r, a till OZ' are perpendicular, the coefficient of
<
2
is unity (Art. :i, E'l (i>)). It fo1lows that E
2
= 1 or<= 1. The
first thtee of ertuations (5) are consequently true. The other equa-
tions 111ay he verified in a simila.t way.
It can now be showu that
At ,\2 ,\a
fJ.1 fJ.z = E = 1.
v
1
v
2
v
3
(6)
For, expand. the uetermiuant by minot'S of the elements of the
fitst row, and substitute for the cofactors of ..\
1
, >..
3
their values
from (:i) . The value of the dcteminant reduces to
!(..\.z + + V)= 1 = (.
( E
lt will br shown in the next Article that if= 1, the system of
axes 0 - X'Y'Z' can be obtained by rotation- from 0-XYZ. If
=- 1, a rotation and reflection are uecess;uy.
ART. as1 RO'l'ATIOX REFLEC''l' IQ}J OF AXES 41
38. Rotation and reflection of axes. 1-I:wing given three mutu-
ally perpendiculat diJectcd lines, forming the trihedral angle
0-SI'Z (Fig. 21), and thrre other mutually perpenuieular directed
lines through 0, for111ing the t rihedral angle 0-S' Y' Z', we sbn.JJ
show that the t riheclral angle 0-S YZ can be revolved in such
a w:ty that OX :tntl OZ coincide in direction with OX' and OZ',
respeetivt'ly. 0 Y will then eoi neill e with 0 Y' or will be di-
rected oppositely to it.
Let .YY' he the lin<' of inter section of the planes XO} and
X'Ol. ''. Denote the angle ZOZ' by 8, the angle XO_V by, and
the angle 4\ '0 X' by .p. Let the z z
axes be rc,ol ved as a
rigid borly :tbont 0 7- th rough t he
angle cf>. so that OX is revolved
into tho position OX Denote the
new position of 0 }" by 0 1r
1
, so
that the angle = cp. The
trihedral anglr 0 STZ is t hus re-
vohecl into 0-XY,Z. Xow let
O-.:Yr
1
Z be re,ohcd about OY Y
through an angle 8, so that OZ fy'
takes a position OZ', :llHl a Fro. 21.
X'
X
position Ol.r
1
. Then t he angle ZOZ' =angle Y
1
01'
2
=8. The
trihedral angle 0-ST ,Z is t hus brought into t he position
O-...\'T
1
Z'. Finally, let the ttihechal angie in t.his l:tst posit ion
be re,ohecl about OZ' through an .p, so that O . .V is rt>vohed
into 0 X'. Hy the same oper:ttion 0} is revohed in to <Hl i recti on
0 Jlt'rpendicnbt to OX' ami to OZ'. It. eit her coi nC" itles
with (} l" o1 is oppositely directed. ln the fhst case t he t1ihechal
O-X rz has been I'Otatell into the t rihedr:tl 0 - X' 1'' Z '. In the
second <'ase the rotation must be followecl by changing t he direc-
tion of the 1'-ax:is. This latter operation is call ed reflection on the
plane y' = 0. l t cannot be <'II'<'Omplishcd by means of 1otati ons.
In case the tril tedral 0-XYZ can be rotated into 0-X' Y'Z',
the number ( (Art. 37) is positi ve; otherwi se, it is negative. Fo1,
during a continuous rotation of the axes, the value of ( (Eq. (G))
cannot change discontinuously. If, after the rotation, the trihe-
drals coincide, we have, in that position, A
1
= 1'-'! = v
3
= 1 rmd the
42 0r>' C:O(>RDINAT8S rTI.
othet cosines are zero, so that (Eq. (G)) t =f. If, however, at the
end of the rotation, 0 Y and 0 Y' are oppositely directed, ,\
1
=
v
3
= 1, /J'2 =- 1, and t =- 1.
3!). Euler's formulas for rotation of axes. Let the coordinates of
a. point Preferred to 0-X be (x, y, z), referred to O-.VY
1
Z be
(x
1
, ?ft, z
1
), referred to 0-NY
2
Z be (x
2
, y
2
, z
2
), and referred to
0- X'r'Z' be (x', y', z'), (Fig.
In the first rotation, through the angle cp, z remains fixed.
Hence, from plane analytic geometry,
z = Zu x = X
1
cos cp- !It s iu cp, y = x
1
sin cp + y
1
cos cp.
In the rotation through the angle 0, x
1
temains fixed. Hence
we have
x
1
= x2> ?It = y
2
cos 0- z
2
sin B, z
1
= y
2
sin 0 + z
2
cos B.
Finally, if 0-X' Y' Z' can be obtained from 0- XYZ by rotation,
temains fixell, and we have
= z', = x' cos oft - ?! sin oft, y
2
= x sin oft + y' cos oft.
On eliminating x
2
, !h, Xu !J
11
z
1
, the fina.l result is obtained,
namely:
x = x' (cos cp cos oft- sin cp sin of! cos (cos cp sin oft
+sin cp cos of! cos 0) + z' sin cp sin B.
y = x' (sin cp cos oft+ cos cp sin oft cos B)- y'(sin cp.siu oft
- cos cp cos ap cos 0) - z' cos cp sin 8.
z = x' sin oft siu () + y' cos oft sin 8 + z' cos().
[f 0-X' Y' Z' cannot he obtained from 0-X by rotation, the
sign of y' should be changed. These formulas :ue known as
Euler's formulas.
-&.0 . Degree of an equation unchanged by transformation of co-
ordinates. If in an equation J!'(x, y, z) = 0 the valnes of x, y, z
repln.cetl by thei t values in any transformation of axes the degree
of F <:.an not be made larger, since x, !J, z are replaced by 1 in ear ex-
ptcssions in x', y, z'. l3ut the degree of the equation cannot be
made smaller, since by returning to the original axes and tn the
eqnn.tion, it would be made largct, which was just to
be impossible.
ART. 40) EXERCISES 43
EXERCISES
1 . Transform the equation x,z - 3 yz + y
2
- 6 x + z = 0 to parallel axes
through the point (1, - I , 2).
Q. {..c
2. By means of equations (2) show that the expression x
2
+ y 2 + zZ is un-
chauged by rotation of the axes. Interpret geometrically.
3. Show that the lines x = !I = = ...1L = z ; = y = arc mu-
4 2 :! - 1 2 -3
wally perpeudicular. "'rite the equat ions of a transforUl:ltion of coordinates
to these lines aJS axes.
V4. Translate the axes in such a way as to remove the first degree terms
from the equation x
2
- 2 y
2
+ 6 z
2
- Hi x- 4 y- z + 37 = 0.
Show that the equatiou ax+ by+ cz + s = 0 may be re<luced to :r = 0
by a transformation of colinlinaw;.
/ 6. Find the equation of the locus 11 + lO yz + 0 z2 - 8 yz + I u- 12 Y!f
- 12 = 0 when lines through the origin whose direction cosines are t, i, j ;
i ! - j; - ;, -! arc takeu as new coordiuate axes.
7. Show that if 0-X' J"' Z
1
can be obtained from 0- .\TZ by rotation, and
if 0 r can be made to coincide with OX by a re,olution of !lO degree::,
counterclockwisE'. as viewed from the end of the Z-axis, then 0 P
can be re\olved into OX' by rotatiug couutcrclockwise through !JO degrees as
viewed from the podithe Z'-axis.
8. Deri ve from Ex. i a necessary and sufficient condition that O-X' 1"
1
Z
can be obtained frow 0-..r l"Z by rotation.
CHAPTER IV
TYPES OF SURF ACES
4 t. I maginary points, lines, and planes. In so lYing problems
tlmt arise in analytic gcometty, it frequently happens that the
values of some of the IIuantities x, !J, z which satisfy the given
comlitions ;ue imaginary. Although we shall not be able to plot
a poi nt in the sense of Art. 1, wlt en some or all of its cotirtlinates
arc imaginary, it will be comcnient to t efcr to any
tri a.1l of numhers x, !1 z, real or imaginary, as the coordinates of a
point. [f :til the coonlinates am real, t.he point is re:tl atHl is tle-
tcnnined by its coonlinates as in Art. 1; if :;ome or all of the
coonlin:ttes art> or eomplcx, the 1oint will be said to be
ima.gin:try. Similarly, a set of pl:we coonlina.lcs 11, v, w will de-
fine a real plane if all the coordinates are real; if some or all
of the cocitdinates are ima.ginary, the plane will be said to be
imaginary.
A linear erptat ion in .l'
1
!J, z, with coeffici ents real or imaginary,
will be said to define a pbne, aml a linear equation in n, 11, w,
with coefficients real m imaginary, will be said to define a. point.
The equations of any two di sti net pl:tncs, considered as simnl-
taneons, will be said to define a line. It fol lows that if (.1,, y
1
, z
1
)
y, Zz) a1e any twO points on the line, then t he COordinates
of auy other poiut ou the line r.:w be \\Titten in the form
k
1
.,
1
+ k zl'-1., etc. 'The line is also dete.-minecl by the equati ons of
any two distin('t points 011 it.
The line joining t.wo imaginary points is rea.! if it also contains
two real voi nts. If p a(n + ik, b + il, c + im) is an imagi nary
point, the poiut ]>' =(a.- lk, b- il, c - im), whose coonlinates
arc the resp('r,t.ivc ronjugates of those of J>, is called the point
conjugate to !'. Thr. line joining :tn_y two conjugate points is
real; thus tlw Pl)ll:-ttions of the lint> PP' arc l.t- ky + bk- al = 0,
(bm- <t,n)!J +(al- bk)z = 0. The line of intersee-
44
ART. 41) DlAGINARY POINTS. LINES, AND PLANES 45
tiou of imaginary planes is real if through it pass two distinct
real planes. The line of intersection of two conjugate planes
is ret
Fr "" the preceding it follo"'S that no imagina1y line can con-
tain more than one real point, and through an imaginary line
cannot pass more than one real plane. If a plane passes throngh1
an imaginary point and not through its conjngate, t he plane is
imaginary. If a point lies in an imaginary plane and not in
its conjugate, the point is i1nagin;uy.
One advantage of using the form of statement suggested in this
Article is that many theorems may be stated in more general form
than would otherwise be possible. \\'e mny for example,
that e\"Cl'Y Hne has two (distinct or coincident) points in common
with any gi ,en sphere.
With these assnm]'tions the preceding formulas will be applied
to imaginary elemt>nts as ''ell as to real ones. attempt wi ll be
made to ghe to such formulas a geometric meaning when imagi-
nary quantities are ill\ohcd.
In the following chapters, in all discuc;sions in which it is
necessary to distinguish between real and imaginary quantities,
it will be assumed, unless the contrary is stated, that given points,
1int>s, and planes, and the coefficients in the equations of given
snrfaces, are real.

EXERCISES
1 . Shnw that the point (2+i, 1 +3 i, i ) on the plane x-2 y + 5 z=O.
V 2. Find the coorrlinates of the points of intersection of the line whose
pammetri c equations are ( 20) x = 1 +
1
3
l cl, y =- 2 + d, z = 5- H d,
with the sphere x2 + y2 + z2 = 1.
\/" 3. Show that the line of of ,the. : + i1J = 0, ( l t i ).r 7
11



4. Find the coordinates of the point of intersection of the line through
(3, 2, -2) and ( 4, 0, 3) with the plane x+3y+ (1-2 i)z+ 1 =0.
II/ 5 . Find the equation of the plane de1errninerl hy the points (5 + i, 2, - 2
- i), (4 + 2 i, -I+ 2 i, 0), (i, 1 + 2 i. 1 + 8 i ).
6. Determine the points ill whi.:h the sphere (.r- 1)2 + 1P + (z = 1
intersects the X-axis.
46 TYPES OF SURFACES
I
[CHAP. IV.
42. Loci of equations. The locus defined by a single equation
mnong t he variables y, z is called a surface. A point
P;;; y
1
, z
1
) lies on the surface F = 0 if, and only if, Lhe coor-
dinates of l'satisfy the ec1uation of the surface. 'Ve have seeu,
for example, that the locus of a linear equation is a plane. More-
over, the locus of the equation
x2+ +z
2
=1
is a of radius unity with center at the origin.
The locus of the ren.l points on a surface may be composed of
curves and points, or there may be uo real points on the stll'face;
for example, the locus of the real points on the surface
x
2
+?l =0
is the Z-axis; t he locus of real points on the surface
xz +!l+zz=O
is the origin; the surface
xz + ?lz + z2 + 1 = 0
has no real points.
If the equation of a surface is mnlt.iplied by a constant diffe1ent
from zero, the resulting equatiou defi nes the same surface as be-
fore; for, if fi'= 0 is the equation of the slllface and k a constaut
different from zero, the coordinates of a point P will satisfy the
equation kF=O if, and ouly if, they also satisfy the equation Ji'=O.
The locus of two simultaneous equations is the totality of the
points whose coordiua.tcs satisfy both equations. If Jt'(x, y, z)=O,
.f(., ?J, z) = 0 n.re the equations of two surfaces, then the locus of
the si multaneous equations F= 0, f= 0 is the curve or curves in
which these smfa.ces intersect. Every poiut on the curve of in-
tersection IMLY be imaginary.
The locus of three simultaneous equations is the totality of the
points whose coordinates satisfy the three simultaneous equations .
.
EXERCISES
.1. Finrl the equation of the locus of n. point whos<' distance from the Z-a.xis
is twice i ts <list.'l.nce fou1 the _\"T-plam.
2. Discuss the defined hy the equation xz + zZ = y
2

-' 3. Fine! the equation of the Joens of a point the sum of the squares or'
whose clistances from I he poiuts ( 1, a, -:.!), (t>, --1, 2) is 10.
Atn:;. 43, 44) CYLINDERS 47
4. Fintl the of the Joens of a point which is three times as far
from G, 3) as froJU the (1. - l!, 4).
5. Find the C<Juatious of the Joens of a point which is 5 units from the
XY-plane and 3 units from the point (3, 7, 1).
V 6. Finrl the equations of Joens of a point which is eqnitlistanL from the
points (2, 3, i ), (3, -4, <i) , (4, :3,- 2).
7. Find the col.irdinatcs of points in which the line x =- 1. z = 2 in-
tersects the cy I i tH llt = -t x.
43. Cylindrical surfaces. It was :;een in Art. 42 that the locus
of a single equation F(x, y, z) = 0 is a. surface. We shall now
discuss the types of smfaces which arise when the form of this
equation is restricted in cert.aiu ways.
THEORE)I. If tlte equatioa of a involces onl!J two of tlte
coordi11ates y, z. the s1o:!itce '' cyliucln"crd swface wftose geuemt-
iny lines we par((llel to the ct.t;i:s wlw:se cou,clinute does not ct.ppew
in the equalirm.
Let.f(x, y) = 0 be an equation contaiuiug the variables x andy
but not containing z. If we consider the two equations y)=O,
z = 0 simultaneously, we have a plane curve .f(.t, ?!) = 0 in the
plane z = 0. If !J
1
, 0) . is a point of this cmve, .f(.v
1
, y
1
) = 0.
The coordinates of any point on the line x = '\ y = ?h are of the
form Xt, ?/1, z. But these cool'llinates s:ttisfy the equa.tiou .f(xll y
1
)
= 0 independently of z, be1we every point of the line lies on the
surface!(.,., y) = 0. lt is the1e'f<l,.e gcnerate1l by a. line moving pal-
a.llel to the Z-axis and always intersecting the cnrve.f(:c, y) = 0 in
the .X Y-plaue. The surface is consequently acyl ind rica! surface.
Lu the same way it is shown that (:1.', z) = 0 is the equation of a
cylindtical Stll'face whose generating ele111 ents are parallel to the
J':.axis, an1l that F(!f, z) = 0 is the' equation of a cylindrical sur-
face whose generating elements are parallel to the X-axis.
44. Projecting cylinders. A cyliuder whose elelllents are pel-
pendicular to a given plane antl intersect a given curve is called
the projecting cylinder of the given cmve on the given plane.
The equation of the projecting cylinder of the curve of intet
section of two F(x, y, z) = 0, .f(x, y, z) = 0 on the plane
z = 0 is independent of z (Art. 43). The equations of this cylin-
48 TYPES OJ" SURFACES ({'HAP. IY.
der may be obtained by eliminating z between t he equations of the
curve.
IfF and f arc polynomials in z, the el imination may be effected
in the following way, known as Sylvester's method of elimination.
Since the coordi nates of points on t he curve satisfy F= 0 and
.f = 0, they satisfy
F= 0, zF= 0, z
2
F= 0, ... , .f= 0, z.f= 0, z
2
.f = 0, ... ,
simul tnneously . I f we consider these equations as linear eqna-
t.ions in t he variables z, z
2
, il, , and eliminate z and its powers,
we obtain an etltmtion R(.r, y) = 0, which is t he e(]_uation requi1ed.
The foll owing exampl e will illustrate the method.
Given the curve
z
2
+ 3 .rz + ;l: + y = O, 2 z
2
+ z + .t' + y
2
= 0.
The e(]_nati on of its projecting cylinder on z = 0 is fonnd by elimi-
nating z between the giYen e(]_nat ions and
z
3
+ 3xz
2
+ (:c+y)z=O, 2 il + 3 z
2
+ (x + y
2
)z = 0.
The result is
1 3 .1; x+y 0
0 1 3x x +y
=0,
2 3
x+yz 0
0 3
x + yz
which si mplifies to
(y2 - 2 y - x)
2
= 9(1- 2 x)(.ry
2
+ x
2
-x-y).
The e(]_nations of the projecting cylinders on x = 0 and on y = 0
may be found in a simi lar maniH'I'.
4.1. Plane sections of surfaces. The equation of the projecting
cylimler of the S<'ction of a surface F(x, y, z)= 0 by a plane z = k
parallel to the X I':plaul' may be fouutl by pntting z = k in the
equation of the smface. The :;tlrtion of t.his cylintlcr F (x,]t, k )=O
by the plane z = 0 is parallel to t he section by z = k. Since paral-
lel sections of a cyli nder, by planes peq>CJHlicular to the elements,
a re congruent, we have the following t heorem:
TuP.omnJ. If in the equation of c' lee Jml z = k and COll-
sider thf' ,psolt the equntion of a Cllrce in the plane z = 0, this curve
is collfJ1'1tellt to the sectio11 the ''!flhe plane z = k.
ART. 46] CONES 49
-t.H. Cones. A surface such that the line joining an arbitrary
point on the surface to a fixed point lies entirely on the surface is
a cone. The fixed point is the vertex of the cone.
THEOIUDI. If tlte equation of a. Sluface is lwnwgeneous in x, y, z,
lhe surj1.1ce is ct cone wilh verte.1; at the ol"iyi11.
Letf(.t, y, z)= 0 he the equation of the f:nrface. Let f be ho-
mogeneous of degree 11 iu (.t, ?/, z), and let P
1
=:(.tu !Ju z
1
) be an
arbitrary point ou the surface, so thatj(.tu !f
11
z,)= 0. The origin
lies on the surface, since f(O, 0, 0)= 0. 'rhe coordinates of any
point P on the line joiuing P
1
to the origin are (Art. 6)
x = kx
11
y = ky., z = kz
11
where k = m
2
m, + ?n2
Rut the coordinates of P satisfy the equation, since
f(x, y, z)=f(k.t
11
k!tu y
1
, z,)=O
for every value of k. Thus, cve1y point of the line OP, lies on
the surface, which is therefore a cone with the vertex at the origin.
EXERCISES
1. Describe the loci represented by the following equations:
(a) x
2
+ = -t. .,, (cl) + lt = 1.
I !)
(b) = x.
(e)
I !)
(c) y =sinx.
x(.t.-l)tx -l)(x -3)=0.
2. Describe as fully as possible the locus of the equation 4 ,.z + yz = 25 z
2

3. Show that the section of the surface x
2
+ y
2
= !l z by the plane z = 1
is a circle. Find the coordinates of its center and the length of its radius.
4. Find the equ:nion of the projection upon the plane z = 0 of the curve
of intersection of the surfaces
yz + 1 = 0, + !P- l)z + 2 y = 0.
Is. Show that the section of the surface x2zl + a2y2 = r2z2 by the plane
z = k is an ellipse. Fin<l its seuJi-axes. By gi1ing k a series of 1alucs, dc-
tPrmine the fortll of the surface.
\16. Show that if the equation of a surface is homogeneous in x - h, y- k,
z -I, the surface is a cone 1vith vertex at (h, k, l).
7. By using show that the cylinder f(x, y, t)=O
"" 00 """;''"'a oono w;th
50
!CHAP. IV.
4 7. Surfaces of revolution. The sm face generated by revolving
:t plane cune a.bout !t line in its plane is callec.l. a surface of revo-
lution. The fi xed line is called the axis of revolution. Every
point of the revol ving curve tlcscribes a circll;l, whose pla.ne is per-
pendi cuhtr to the axis of rcvolutiou, whose center is on the a.xis
antl whose radius is the distance of the point from the axis.
To determine the equat ion of the surface generated by revolving
a given curve auout a given axis, take the plane of the given curve
for the XY-pl ane and the axis of r evolution for the X-axis. Let
the equation of the given curve in z = 0 be j (x, ?J) = 0. Let
P
1
= (.r., y., 0), Fig. 22, be a.ny point on the cur ve, so that f(x., y
1
) = 0
X
and let J> ;: (x. !f, z) be any point on the circle described by 1\.
Si uce the plane of the cin;)(> is perpemlicnlar Lo ihc X-axis, the
cttuation of this plane is .1: = .r
1
The cooadin:ttcs of the centca 0
of t he cirr lc an' 0 ;::. (.1;
1
, 0, 0); and the radius Of\ is ?J
1
The
distanc< !'rom a to p is
?h =Y(x
1
- .1:
1
)2 + (y- 0)
2
+(z- 0)
2
= V?J
2
+ z
2

On snustituting
in t.l1e ctpmtion f(.&., y
1
)= 0 we obtain, a.s the condition that the
point J> li os on the
f(.r, v.,/+z
2
)=0,
whicil is the dcsiactl equation.
In t.hc satnc wa.y it, ma.y be seen that the Ct ln:ttion of the sur-
face of revolution obtained by revolving the cnrvc f(:c, y) = 0 n.bout
the }':.axis is
\
ART. 47 ) EXERCISES 51
EXERCISES
\\'bat is tbe equ;ltiou of the surface generated uy rei"OIIing the circle
x2 + = 25 about the X-axi s? about the
V 2. Obtain the equation of the genet-ated by revoll"ing the line
2 .c + 3 y = 15 ahont the X-axis. Show that th<' surface is a cone. Fiud its
vertex. \Yhat is the equation of the section tnade by the !Jiane x = 0 :>
the equation of the cone by revo!ling the line about the
l is.
3 . Why is the resulting eqnatiou of t he same degree as that of the gen-
cune in 1-:x. I, but twice the of the given curve iu F.x. 2?
Formulate a general rule.
4. What i:; the equat iou of the !<urface generated by reYulling t he line
y = II about tbe X-axis ;lhOUL the
: 5. If tht' unc f(., !I)= 0 till' .t-axis at the point 0, 0) , de-
scribe the appearanc<' of the
f(.r,

+ .:')= 0 ll t'at the poiut (;;" 0, 0) .

6 . Find tht> equation of the s111face by revolving the following
curves abnlll tlw a11d abnm thE' !"-axis. Draw a figu re of each surface.
(u) 1.
-1 II
(/..) ;;_: _ u: = 1.

() =!:I.e. (e) y =si n x.
(!) Y = e.
CHAPTER V
THE SPHERE
48. The equation of t he sphere. The equation of the sphere
having its center at (a
0
y
0
, z
0
) and radius?' is
(1)
or
:t.z + y2 + z2- :t'o-1:- !lo.'lf - ZoZ + Xoz + Yoz + Zoz - ?.:! = 0.
Any equati on of the form
c1 ( x
2
+ y2 + z
2
) + 'J f v + Y?l + hz + k = 0, c' =/= 0
may be \Vl"itteu in the form
(
x+l.)z +(y+ f!.)2 +(z +!:.)2 =.f2 + rl + ft2-ak.
a n <L a
2
(3)
If p + g
2
+ h
2
- ak > 0, t his is seen, by conqxtring with (1), to be
a sphere with center at ( - !., - 1., -h) and radius
( I ((. CL
:!.t"'!. + !l + h
2
- ak.
It
If t he expression n111le1 the 1adical sign mnishes, the center is
the onl y real point ly ing on t he sphCIe, which in t hi s case has a
zero l'atlins, awl i:; calletl a point sphere. lf t he expression under
the r:ulieal is negat.i vr, no real point lies on the locus, which is
called an imaginary sphere.
4 0. The absolute. 'Ve shall now prO\' (> t he foll owing theorem:
TnEORE)t I. A ll SJ'lu>l'es il!le1sect thP , ,zane Ht bt the
sa1ne cwve.
In order to determine t he intersection uf the sphere and the
plane at infini ty, we first wri te the equation of the sphere in
coi.i l'lli nates:
+y
2
+z
2
) + 2 fTl + 2yyt + hzt + kt
2
= 0, <1 =1= 0.
!)2
ART. 49) 'l'HE ABSOLUTE 53
The equations of the curve of intersection of this sphere with the
plane at infinity are
(4)
Since these equations are intlependent of the coefficients a, j, g,
It, k which appear in the equation of the sphere, the theorem
follows.
The curve determined by equations (4) is called the absolute.
Since the homogeneous of a point cannot all be zero
(Art. 29), there are no real points on the absolutf'.
The equation of any surface of second uegrce which contains
the absolute may be written in the form
a + !/ + + (k:r + ly + mz + 1tt) t = 0.
If a =1= 0, this is the eqnation of a sphere (Att. 48). If a= 0, the
locns of the equation is two planes of which at least one is t = 0.
In the latter case also, we shall call the surface a sphere, since
its equation is of the second degree and it passes through the abso-
lute. \\'hen it is necessary to distinguish it from a proper sphere,
it will be called a composite sphere. With this extended defini-
tion, we have at once the following theorem:
II. Etery su,face of the secoml degree which contains
the absolute is a sphere.
Any plane
11x + vy + u:z + st = 0,
olher than t = 0, intersects the absolute in two points whose coor-
dinates may be fount! by solving the equation of the plane as
simultaneous with the equations of the absolute. Any circle in
this plane is the intersection of the plane with a sphere. Since
the absolute lies on the sphere, the circle must pass through the
two points in which its plane intersects the absolute. These two
points are called the circular points in the plane.
Evidently all the planes pa.rallel to the given one will contain
the same ci1cnlar points. The reason for the designation circu-
lar points is seen from the fact that any conic lying in any real
transversal plane and passing through the circular points is a
circle, as will now he shown. Since the equations of the absolute
are uot changed by displacement of the axes, it is no restriction
54 [CHAP. V.
to t.ake z = 0 for the e<1uation of the t ransversal plane. The
coordinates o( the points ill which t he plane z = 0 meets the
curve t = 0, + !f
2
+ z
2
= 0 arc (1, i, 0, 0), (1, - i, 0, 0). A conic
ill the pl:Lne z = 0 has an e<Jnation in hoH\Ogcneons coordinates
of the fonn
.A:t;
2
+ B!i + 2 + 2 G.t + 2 Fyt + Ot
2
= 0.
If the points (1, i, 0, 0), (1, - i, 0, 0) lie on t his curve,
.11 = B, II= 0.
13ut these are exactly the conditions that the conic is a circle.
Conversely, it follows at once that every circle i n the plane z = 0
p:tsses through the two ci1cular poi uts iu that plane. A conic
in an imaginary plane wi ll be dctiuetl n,s n, ci rcle if it passes
through the circular points of the plane.
Lf the two circular points in a plane coincide, the plane is said
to be tangent to tbe absolute. Such a plane is cn.llt>tl an isotropic
plane. The condition that the plane 1t.c + vy + wz + st = 0 is
isotropic is found, by imposing the condition that its intersections
with the absolute coincide, to be
(5)
This equation is the eqnation of the al>solute in phme coordinates.
EXERCISES
1 . " 'rite the equation of a sph<'re, given
(a) at (0, 0, 0) an<l radim: 1,
(b) center at (- 1, 4, 2) and radius G,
(c) center at (:J, 1, 5) and radius .J .
2. Deterllliue t.lte center and radius of each of tle following spheres:
(a) :r2 + y2 + + i x + 2 y + z + 5 = 0.
(b) + y2 + z2 + 2 .t + .J y - G z + 11 = 0.
4 (c) 2(.t2+y2+z2)-x-2y+5z+5=0.
(d) x2 + y
2
+ z'
1
+ fx = 0.
Find Lhe points of inLerscction of the absolute nnu the plane
;:.
2 X- y + 2 Z + 15/ = 0.
4. Find the coordinates of the points of intersection of the line x = - 2
+ d, y = 3 - i d, z =- 2 + iJ. with the sphere :t
2
+ y2 + z2 + 1 = 0.
5. Show tha.t ,e2 + y2 + z2 = 0 is the equation of a cone.
6. Find the distance of the point (I, 0, i) from the origin.
ARTs. 49-51] THE AXGLE BET"\\'EE}.[ TWO SPHERES 55
7. that the radius of the circle in which z = 2 intersects the
.c2 + yz + =I is imagina1y.
'1. .Ya. !'rove thaL, if (:r., y., z
1
) is any point exterior to the sphere (x-
+(Y - Yo)Z + (z- zo)Z = rz, the expr!'ssion (x,- +(!It - + (zt- zo)
2
- rz is the s4uarc of the segment on a tangent from Zt) to Lhl' point of
contact on the sphere.
, \
50. Tangent Plane. Let P= (.
11
!/" z
1
) be any poiut on the
sphere
+ yt + z2) + "2):-.; +:.! U!J + ltz + 7.; = 0.
The plane through P prrpendi<nl::n to tht> line joinin:;: P
to the center uf the sphere is tht.: tnngeut plane to t he sphere at /'.
It is required to find its CtJUation. The coonlinates of t he
(
r " h\ r
center are - . -'-,- }" The cquatioas o the l ine joining
n n 11
the center to Pare (A1-t. Hl)
x-.>,
f
- - - x,
a
y- !h
q

(l
The Cl)uation of the plane passing
through P antl perpendie;ular to this
line is
FIG. :!;l.
(
... + .c,)(:c-

+!h)('!- .'lt)+(h +z
1
)(z - z
1
)= 0.
11 ,a a
lf we expand thP membc1 of this N]ualion and athl to it
a(.1:
1
2
+!J
1
2
+z
1
2
)+"2/1:
1
+:!y!!t

which is equal to zNo

since the point (.1., y
1
, z
1
) lies u11 the sphere, we obtain
a(.r,.c+ y
1
y + z
1
z) +.f(:c +.t
1
)+!l(?ft + !J)+h(z
1
+ z)+k= 0, (li)
which is the required equation of the tangent plane.
;; t. The angle between two spheres. The between two
sphe1es at a poiut f>
1
on tht>ir cunc of intPr:>ectiou is tletined a.s
eqnal to the angle bctweE>n tl1e ta ngent planes to the at 1'
1

To deterllline the magnitude of this Itt the coordi nll.tes
of P, ue (.ta !Ia z,) and let the equaticms of tht' spheres be
+ !l 1- z
2
)+ "lfx +:! Y!l + k = 0,
a'(.c
2
+ y
2
+ z
2
)+ :.!f'x + g'!J + h'z + k' = 0.
56 'T' HE SPHERE [CHAP. V.
Thn equations of the tangent planes to these spheres at P
1
are
a(:t:
1
X + !h?l + z
1
z) + .f(x + Xt) + g(y + y
1
) + h(z + z
1
) + k = 0.
a'(X
1
X + ?J
1
?J + z
1
z) + .f'(x + x
1
) + g'(y + y
1
) + h'(z + z
1
) + k' = 0.
Since the angle (} between the spheres is equal to the angle
between these planes, we ltave (Art. l!i)
COS 0 = _... 1.1 t',.. I ,.. /I\.. ' I
(rtx1 + J)(a'.'J +.f)+ (tt!h + {T)(a'y + rl') + (azt + h) (a'z1 + h')
v'(a.r+ f)
2
+(a!l (az1 (a'y
1
(a'z1
Since (.v., !lu z
1
) lies on both spheres, this relation reduces to
3 [(' + 2 gg' +:.! hh'- ak' -a'k
cos (} = .. .. - (7)
2 .J.t"l + r/ + h
2
- ab/./'
12
+ + h'
2
- a'k'
Since this expression is int.lepeudent of the coordinates of P., we
ha,e the following t heorem:
'1'11 EOttE)I. TICO spheres intersect at the swne angle at allJJOints
of their Ctrve of intcrsectiou.
lf (} = 90 degrees, the sphetes are said to be orthogonal. The
condition that two spheres are orthogonal is
2ff" + 2gg' + 21th'- ak' - a'k = 0. (8)
52. Spheres satisfyi ng given conditions. The equation of a
sphere is homogeneous iu the five coefi1cients a, f, g, h, k. Hence
the sphere may he nHLtl e to satisfy folll' conditions, as, fm example,
to pass tlll'ongh four given points, or to intersect. fonr given
spheres at given angles. If the given conditions are such that
a= 0, the sphere is cotuposite (Art.
EXERCISES
1 . Prove that the point (- 3, I, - I) lies on the sphere x2 + ?12 + z2 + 6 x
+ 2-1 y + 8 z = 0 and write t he equation of the tangent plane to the sphere at
t hat point.
2. Find the angle of intersection of the !>pheres x,2 + y! + + x + 6 y
+ 2 Z + !J = 0, x2 + yZ + zZ + 5 X+ 3 Z +I = 0.
I
1
3. Fiud the ClJUation of tho sphere with its ceutcr at ( 1, :3, 3) and making
an angle of GO degrees with the sphere :r2 + !P + = L
J 4. Dctennine the uquation of Lhe sphere which thruugb the points
(0, 0, 0), (0, 0, 3), (0, 2, 0} , ( 1, 2, 1).
.\RTS. J2, 53] SYSTE:\lS OF SPHERES 57
5. Determine the of the sphere which through the points
(1, 3, :!), (3, :!, - S), ( - 1, :!, :3), (4, S, :!) .
v' 6. Write the of the sphere passing through the poiuts (2, 2, - 1 ),
(3, - 1, 4), ( 1, :J, - :l) orthogonal to the sphere
.t.2 + + zZ - 3 .r. + y + z = 0.
V,, Write the equation of the sphere inscribod In the tetrahedron :z: = 0,
y = 0, 6 :z; + 12 z + 3 = 0, 3 :z;- 12 y + 4 z = 0.
o3. Linear systems of spheres. Let
S = u (:If+ + z') + 2 fx + Z (1!1 + Z hz + k = 0,
S' = <t' i +z') + ?. j'x + '1(/y + Zll'z + k' = 0
be the equation of two spheres. The equation
>-18 + = 0,
OL' (a.\
1
+ u',\
2
)(x
2
+ Y
2
+ Z
2
) + '2 (j> ... l +.f'A2) x +'2 (y,\1 + g'>...2)y
+ 'J (II>... I+ h'.\2) z + k,\1 + k' = 0
also represents a sphere for all values of >...
1
and A
2
Eve1y
sphere of the system >.
1
:) + = 0 contains the curve of inter-
sectiou of S = 0 am! S' = 0 ( Art. In particular, if <tA.
1
=- a 'A
2
,
the l>phere .\
1
S + ,\
2
= 0 is composite; it consists of the plane at
infinity (whi('h intersects all the spheres of the system in the
absolute) and the plane
.Sc:ti'
Z("J- <{{') :c + '2 (u'y- !I+ '2 (H'h - 11lt') z + ak' = 0, (!J)
whi ch iulcrse<'ts al l the sphcn'S of the system in a fixed circle,
common to S = 0 and S' = n. The plane (!J) is call ed t he radical
plane of the gi Vt'll System of S!Jhen'S.
It will now Le shown t hat thP racl ical plane is the locus of the
Cl'nters of the spheres intersecting S = 0 and S' = 0 ort hogoually.
For this p01pose llt

0
(x'+ !l +z') + ZJ,;t+2go?J + '2 hoZ +ko =0 (10)
be the equation of a sphere. It will be 01thogonal to S if (Art.
51)
+ 2gr{J + 2 hJt - a,;.- uk
0
= 0,
and to S' if
58
'PHE SPHI!}RE (CHAP. \ ".
lf we el iminate k
0
between these two equations, we have
2(ct'f -ay')g
0
+2(a'h-ah')lt
0
- (a'k-ak')a
0
=0, (11)
which is exactly the cendition that the center (_h_, __f.&_, .JlL)
- -o - flo - Lo
of the orthogonal sphere lies in the radical plane (9). Con-
versely, if .fo, (!
0
, h
0
are given numbers which satisfy(!>), a valne
of ko can be found snc h that the conesponcli ug sphere (10) is
orthogomtl to every sphere of the system A.
1
S + A
2
S' = 0.
Again, if
S" = a" (x
2
+ y
2
+ z
2
) + '2.f".); + 2 u"!t + 2 h"z + k'' = 0
is a sphere whose center does not li e on the line joining the
centers of S ami S', every S!Jhe re of the system
AIS + A2S' + A3S" = 0
passes the points of intersection of tht! spheres S = 0,
S'=O, S" = 0.
Every sphCl't' of t he systent (12) determined uy values of
Au A
2
, A
3
for which
A
1
tt + + A
3
u" = 0
is composed of two planes of which one is the plane at infinity
anc1 t he other passes throu1,;h the line
C({').t + ag')!! + '.!(u'h- uh')z+ a'k- ak' = 0, (1:{)
11
'2 (ct''f -af")x+ 'J(ct"y-ay")y+'l(a" h-ult")z+a"k- ak" = 0. ; a''5-a.S
This line is called the radical axis of the system of spheres (12).
By comparing crptations (13) \\'ith (11) a.ml the ef]nation analo-
gous to (11) for ,s" = 0. it may be shown tliat the radical axis is
the locus of centers of the spheres which intersect all the spheres
of the system (1:3) orthogon:1.lly.
Now l et
S
111
=a'" ( . .;
2
+ !!
2
+ z
2
) + + 'Jg'"y + 2h"'z + lt" = 0
be tl1e ef]ua.tion of a sphere whose center is not in the plane de-
tcrminetl by the centt!rs of S = 0, S' = 0, S" = 0. The condition
th:1.t a sphere of the system
AIS + A2S' + ,\3S'
1
+ A.S"' = 0
is composite, is that A
1
A
2
A
3
nn<l A
4
satisfy the relation
A
1
ct + Ap! + A
3
et'' + A
4

111
= 0.
Ar<r>" ;)3, 54] STEREOORAPHIC PROJECTION 5!)
'!'he sphere orthogonal to all the spheres of the system is i n
this case uniquely determined by equations analogous to (lp ).
The center of this ol'thogonal sphere is called the radical center
of the system. Through the radical center passes one plane of
every composite sphere of lhe system.
EXERCISES
11. Prove that the center of any sphere of the system x
11
<; + ">-:8' = 0 lies
on the line joining the center of S = 0 to the center of 8
1
= 0.
j/2. Prove that the line joining the centers of thf> spheres S = 0 and 8
1
= 0
('") is perpendicular t.o the radical plane of the system X
1
S + X
2
8' = 0.
3. Show that the radical axis of the syRteut Xp'\ + + x
3
R
11
= 0 is per-
pendicular to the plane of centers of the spheres to the system.
1 4. Determine the equation of the system of sphereR orthogonal to the
system ">-tS + "A2S' +">-aS"= 0.
1s. Show that two point es are inclutleli 111 the system ">-
1
8 + = 0.
1 6. Show that auy !lphere of the system >-tS + :>.
2
8
1
= 0 is the locus of
a point. the mtio of whose distances from the centers of the two point
SJlheres of the system is con;;tant.
1 7. If S = 0, S' = 0, 8
11
== 0, 8
111
== 0, S
1111
= 0 are the equations of five
spheres which do not belong to a linear system of four or less terms, show
that the equation of anv sphere. in tipace can be expressed by the equation

5-1. Stereographic projection. Let () be n fixed point on the
surface of :1. sphere of radius r, and let -rr be the plane tangent to
the sphere at the opposite end of the diameter passing through
0. The intersection with rr of the liuc joiniug 0 to any point P
1
on the surface is called the
stereographic projection of
PI (Fig. 24).
To determine the equa-
tions connecting the co-
ordinates of P
1
and its
projection, take the plane
r. for the plano z = 0, a.nd
the diameter of the sphere
through 0 for Z-axis. The
equation of the sphere is "'-----+-----../
x
2
+ y
2
+ z
2
- 2Tz = 0.
Fro. 24.
-'L
60
THE SPHERE
The equations of the line joining 0 = (0, O, 21) to P
1
= (x
1
, y., z
1
)
on the sphere are (Art. 19)
X 11 Z - 2 ?'
- =-=---
x, y, z, - 2 ,.
To determine the coordinates (x, y, 0) of P, the point in which
OP
1
intersects -rr, we make the e1uations of the line simultaneous
with z = 0. On solvi ng fot x, y, z we obtain
2 n; 2 1' 11
X= - '>--
1
- , ?J = - ., --
1
- , Z = 0.
~ 1' - z, ~ 1' - z,
These equations can he solved for x., y
1
, z
1
by making use of the
f:tet that, since P, li es on the sphere,
x2
1
+ y\ + z
2
1
- 2 rz
1
= 0.
The resnlts are
-l r
2
x _ . . J . 1 ~ y _ 2 1(x
2
+ y
2
)
x, = ' 1!J - ' z,- . (14)
.1.:2 + y2 + 4 1.2 3p + '!/ + -l 1.2 x2 + y2 + 4 1.2
TlmOJtE,\t I . The xtel'eoyraphic ]>,ojection of a. cirde is a ci1cle.
Let the equation of the plane of the given ci rcl e on the sphero be
Ax+ By + Gz + D = 0.
The condition that P, lies on this circle is consequently
Ax
1
+ By
1
+ Cz, + D = 0.
If we substitute from (14) in thi s equation, we obtain as the
equation of the stereographic projection,
4 .th
2
.t + 4 Bl'
2
y + 2 01'(x
2
+ !/) + D(x
2
+ y
2
+ 4 r
2
) = 0, (15)
which represents a circle in the X }:plane.
In particular, if the plane of the given ci rcle passes through 0,
the stereographi c projection of the ci rcle is composite. The con-
dition that the plane
Ax+ By + Cz + D = 0
passes through 0 is
2?C+ D=O.
If this condition is satisfied, the eqnation of the circle of projec-
tion is, in homogeneous coorclinates,
t(Ax+ By+ Dl) = 0.
ART. M) STEREOGRAPHIC PROJECTIO}.T 61
The points of the line t = 0 con espond only to the point 0 itself.
The line
Ax+ By+Dt =0
is the line of intersection of the plane of the circle and the plane
of projection. We have consequently the following theorem:
Tnr.OHEH H. The c i r c l e . ~ on the $phel'e tc!tich puss throuah the
ce11ter of Jlrnjecl.ion we jlrojected .steJeoyraplticall!J into the lines in
which their Jllcmes intersect the 7'llane o.f projection.
The angle between two inter secting Clll' \'es is defined as the
angle between thci1 tangents at the point of intersection. We
shall prove t he following theorem :
TIIEOJU\)1 liT. The an(lle l1etn:een two inte1secting cun.:es on the
sphere is ef_lual to the angle betuPell their stl.'reograp!tic projections.
H will snfllce if we prove Lhe t. he01em for great circles. For,
let C'
1
and 0'
2
be any two cu,es whatever on the sphere having
a point P' in common. The great cirr les whose planes pass
through the tangents to 0'
1
ancl 0'
2
at P' are tangent t o 0'
1
and
C'
2
, respecti, ely, at P'. Let Ou 0
2
, and P
1
he the stereographi c pro-
jections of 0'., 0'
2
, antl P'. The stcreographi c projections of
the great ci rrles are tangent to C
1
and C
2
, respectively, at P
1
so
that the angle bet ween them is the angle bet ween 0
1
and
0
2
If, then, the theorem holds for great circles, it holds for all
intersecting cnrves.
The condition that a circle is a great circle is that its plane
A.1: + B!! + Gz+ D= 0
passes through the center (0, 0, 1') so that
The equation (15) of the stercographic projection reduces to
G (:l-.2 + y
2
) + 4 r(A.t + B!J - 1C) = 0.
The angle between two great circles is equal to the angle be-
tween their planes, since the tangents to the circles at their com-
62
TilE SPHERE [C'HAI'. \'.
mon points are perpendicular to the line of iut.erscction of their
planes. The angle 0 between the planes
.A.t: +By+ Cz - Cr=O
and A'x+ B'y + C'z- C'r = 0
is defined by the formula (Art. 15)
AA' + l3B' + CC'
cos 6= (16)
.y + ]32 + (Jl VA'2 + B'2 + 0'2
The tangents to the projections
C(xz + !J
2
) + 4 r(A.1; + B!J- 10) = 0,
C'(x
2
+ !l) + 4 r(.A'.v + B'y- 1C') = 0
of the given circles, at the point (x
11
y
1
) in which they intersect, are
(Cx
1
+ 21A).t + (C!h + 2 rB)!J + 2 1A.t
1
+ 2 rBy
1
- 4 r
2
0= 0,
(C'x
1
+ (O'y
1
+ 2?B')!I + 1 . .:l'x
1
+ 2 1B'!J.- 11
2
0' = 0.
The angle cJ> between these circles is giYen by the formula.
cos cJ> =
( Cx
1
( 0'.'1;
1
+ 21A') + ( C!h + 2 1R) ( C'y
1
+ 21'lY)
.Y( r.A)2+ (Oy. + '!, 1B)
2
v( C'.r
1
+ rA')
2
+ (C'y
1
+
13y expanding this expression and making nsc of the fact that
(x., y
1
) lies on both circles, we simplify the preceding equa-
tion to
cos cf>= A..- 1' + BB' _ CO' (l7)
Y Li2 + b'2 + 0
2
V .1'
2
+ B'
2
+ 0'
2
From (lG) and (17) we have cos(}= co:; cp. We may conse-
quently choose the angles iu such a way that 0 = cp, which proves
the proposition.
The relation cstabl ished in Theorem III makes stereographi c
projection of great importance in map drawing.
.'
... . . t
CfiAPTEH. VI
FORMS OF QUADRIC SURFACES
r;c;. Definition of a quadric. The locus of an equation of the
degree in .t, ,11. z is called a quadric surface. T n this chapter
l'l'rtain standard types of the equation will be considered. It will
be shown la.ter that the equation of any non-composite quadric
may, by a suitable transformation of coordinates, be reduced to
one oE these types.
5U. The ellipsoid. The locus of the equation
x1 1P z2
c2 = 1
is called tl1e ellipsoid. Since only t he second powers of the varia-
bles .1, .'/, z appear in the C<puttion, the smface is symmetrical as
to l'n.ch coordinate plane, as to each cooJ'lli nate axis, and as to the
origin.
The coordinates of the poi nts of intersection of the ellipsoid
with the X-axis are found by putting y = z = 0 to be (a, 0, 0).
Its inte1sections with the )':axis are (0, b, 0), and with the Z-axis
are (0, 0, c). These six points are call eel the vertices. The seg-
ments of the coordmate axes included betwPen the vertices are
called the axes of the ellipsoid.. The point of intersection of the
is eallecl the center. The segments from the center to the
\'Crtices are the semi-axes; their lengths are a, b, c. \Ve shall
Stt!Jpose t.he coordinatt> n.xes are so r hosen that ct > b > c > 0. The
segment joining the ,erticrs on the X-axis is then known as the
major axis ; that joining the vertices on the T"-n,;.:is as the mean
axis: that joining the Yertices on the Z-axis as the minor axis.
The section of the ellipsoid by the plane z = k is an elli pse
whose equations are
x2 y2

+ =l,
Z=k.
(33
64
FORMS OF QUADRIC SURFACES [CHAP. VI.
The semi-axes of this ellipse are a 11 _ ~ , b ~ l _ ~ . As 1 k 1 in-
V c2 c2
creases from 0 to c, the axes of the ellipses of section decrease.
If I k I= c, the ellipse reduces to a point. If I k I > c, the ellipse
of section is imaginary, since its axes are imaginary. 'l'he real
J:o'IC. 25.
part or the surface
therefore lies en-
tirely between the
planes z = c and
z = -c.
In the same man-
X ner, it is seen that
the plane y = k'
intersects the sur-
face in a real ellipse
if I k' I < b, that the
ellipse reduces to
a point if I k' I = b,
and that it becomes imaginary if I k' I > b. Finally, it is SE\en
t hat the section x = !.:" is a real ellipse, a point, ot an imaginary
ell ipse, acconling as I k" I is less than, equal to, or greater than a.
The ellipsoid, there-
fore, lies entirely
withi n the rectan-
gular parallclopiped
formed by the planes
x = a, y=b, z = c;
x =-a, y =- b,
z = - c, and has one
point on each of t hese
planes (Fig 25).
If a = b > c, the
ellipsoid is a surface of revolution (Art. 47) obtained by revolving
the ellipse
about its minor axis. This sul'face is called an oblate spheroid.
If u > b = c, the ellipsoid is the smface of revolution obtained
ARTS. 56, 571 THE HYPERBOLOID OF OXE SHEET 65
by revolving t}1e same ellipse abont its major axis. It is called a
prolate spheroid.
If a = b = c, the smface is a sphere.
5 1. The hyperboloid of one sheet. The surface represented by ,
the equation
is called an hyperboloid of one sheet. Jt is symmetri c as to each of
the coordinate planes, as to each of the coordi nate axes, and as to
the origiu.
The section of the by the plane z = k is an ellipse
whose equations are
x
2
1/
(
J.-2 + ( . k2) = 1, z = k.
(t
2
1 + C2) 0
2
1 +
..
This ellipse is real for every real value of k. The semi-axes are
which are the smallest when k = 0, and increase without limit as
I k I inc reases. 'For no value of k does the ellipse reduce to a
point.
The plane y = k' intersects the surface in the hyperbola
,
a2(1 - k'"i" ) - c2(1 - J.-,'2) - = 1, y = k .
lJ2 lJ2
If i k' I < b. the trMsverse axis of the hyperbola is the line
z = 0, y = k', and the conjugate axis is x = 0, y = k.'; t he lengths
of the semi-axes are a,/ t - 1 - 1.::
2
As I k' I increases from
b- lr
zero to b, the semi-axes (\('crease to zero. \ \' hen !7.:' I = b, the
x1 z2
equation cannot be put in the abo\e for m, hut becomes - --- = 0
0
2 c2
and the hyperbola is composite; it consists of the t wo lines
X Z X Z
c =0, y= b; y=b ;
66 FOR.l\IS OF QUADRIC SURFACES
when k' =- b, t ho hyperbola consists of the lines
xzo bx z o b
- + -= ' y =- ; ---= ' y = - .
a c a c
(CRAP. VI .
These four lines lie entirely on the surface. If I k' 1 > b, the
transverse axis of the section is x = 0, y = k' and the conjugate
axis is z = 0, y = k'. The lengths of the semi-axes are

. /1.:-'2 -
a\ bz - l, c\J[; - 1.
They incre.1.se wi thout limi t as k' increases.
The plane x = k" intersects t he surface in the hyper bola
If I k" I <a, the transverse axis of t his hyperbola is z = 0, x = k".
The section on the ]'lane x = ct consists of the two lines l
!! z _ . _ . !I z _ . _
+ - - 0, X - ct, - - - 0, X - a. ,.......!
u c u c
The section on the plane x = - ct cons ists of the lines
?'J Z II Z
'- + - = 0, X = - a j '
1
- - = 0, X = - C!.
u c u c
If I k" I >a, the line !I = 0, x = J.;" is t he t ransYerse axis and z = 0,
x = k'' is t he conj ugate
As I k" I iucreas<>s, thel<>ngth::;
of the se111i -axes innea:;c with-
out 1 i mit. The form of t he
smfaoeis indi(a1('<l in
Fm. 21i.
ARTS. 57,581 THE HYPERllOLOID 0 1 ~ TWO SFJEE'T'S 67
If ct = u, the hyperboloid is the surface of revolution obtained
by revolving the hyperbola
.t:2 z2
---;; = 1, y = 0
a
2
c
about its conjugate axis.
uS. The hyperboloid of two sheets. The locus of the equation
x2 1/2 z2
- - - - =1
a
2
u ~ c ~
is called an hyperboloid of two sheets. It is symmetric as to each
of the coordi nate planes, the coordinate axes, and the origin.
FIG. 27.
The plane z = k intersects the surface in the hyperbola
1, Z=k.
The trans\erse axis is y = O, z = k, for all values of k. 'l 'he
leugths of the semi-axes are a\ /1 + ~ ' h/1 + ~ - They arc
c2 " c2
smallest fork = 0, namely a, and u, and increase withont limit as
k I incrE>ases. The hyperbola is not composite for any real
value of k.
The plane ?/ = k' intersects the sul'face in the hyperbola
(
7,:12) = 1, y = 7.:'.
c
2
1+ -
l}
z2
68 FORMS OF QUADRIC SURFACES [CHAP. VI.
The transverse axis is z = 0, y = k'. The conj llgate axis is
x = 0, y = k'. If k' = 0, the lengths of the semi-axes are a and c;
they increase withont lin1it ask' increases.
The plane x = k" intersects the surface in the ellipse
Y ~ z ~
lJ2(k';2 - 1) + c2(k"2- 1) = 1, x = lc".
c ~ 2 a2
This ellipse is imagin:u-y if lk" I< a. If I k" I = a, the semi-
axes are ze j o; they increase without limit :t..'> k" increases.
If b = c, the hyperboloiu of two sheets is the surface of revolu-
tion obtained by revolving the hypmbola
r ?l
- - - = 1, z=O
u2 b2
about its transverse axis.
50. The imaginary ellipsoid. 'rhe surface defined by the equa-
tion
is called an imaginary ellipsoid. Since the sum of the squares of
three real numbers cannot be negative, there are no real points on
it.
ARTS. 59, tiO) THE ELLIPTIC PARAilOLOID 69
EXERCISES
1. By translating the axes of coordinates, show that the surface defined
by the equation 2 .1! + 3 y
2
+ 4 z
2
- 4 x - ti y + HI z + Hi = 0 is an ellipsoid.
Find the co!irdinMes of the center and the of the semi-axes.
2. Clas . .;ify and describe the surface + yz- .J x- :l y + 10 z = 20- z2.
r
3. Show that the surface 2 :x2 - 3 5 z = 7 - 2 y
2
is a smfacc of revo-
lution. Find the CI[Uations of tile generating curve .
.c/4. On tile hyperboloid of one sheet .t
2
+ z
2
= 1, find the 1.'411atinns
of the two lines which pas::; through the point (I, 0, 0); through (- 1, 0, 0).
5. Classify and plot the loci defined by the following equations:
(a) 9 .(.! + Hi yl + :!:i = I, (d) ::cJ + y!- 4 z
2
= 25,
(b) 4 .r2- 9yZ- = 25, (P) + .J yz + z2 = !l,
(r) 4 Ill !P + n = :!;;, (.() + .J y1 + !l z2 + 8 = 0.
HO. The elliptic paraboloid. The locus of the equation
is called an elli ptic paraboloid. 'L'he surface is symmetric as to
the planrs :r = 0 and !I = 0 but uot as to z = 0. 1 t passes through
the and lirs on the positi\e side of z=O if n is positive
aUll ou the nc.>gati re sitle if n is uegati n. In the following dis-
cussion it will bt assuntCd that n is positive. lf 11 is nc.>gati\e, it
is nc.>cessaty ouly to reflect Lhe surface on tlw plane z = 0.
The St><!Liuu of the p:uaholoid by the phuw z = 1.: is an ellipse
whose are and lrJ'.!. 11k, respect ively. lf k < o,
the elli pse is imaginaty. If k = 0, the ellipse reduces to a point,
the origin. As k increases, the semi-axes of the ellipse imrease
without limit.
The sc.>ctiou of the paraboloid by the plane y = k' is the
parabola
z! A k'
2
=:::11%- - y=k'.

For all values of k' these are congruent. As k' iu-
creases, the vertices recede from the plane y= 0 along the parabola
'!/!.= 2nz x = 0.
b2 '
'v
70
FORM::> 01' QUADRIC [CHAP. VI.
FlO.
If a = b, the prrrn boloid is
the surface of revolution
generated by revohing the
r2
parabola :._ = 2 nz, y = 0
(/2
about the Z-a.xis.
() 1. The hyperbolic parab-
oloid. The surface defi ned
by the equation
The sections by the planes .c =
k" are the congruent parabolas
k"2
._ = 2 liZ - - , X = k".
IP ct
2
Their vertices llescribe the pa,.
rabol a
x! ,,
nz, y= 0.
a2
The form of the surface is in-
(licated by Fig. 28.
is called an hyperbolic paraboloid. The surface is symmetric as to
the planes x = 0 an<l 71 = 0, but not as to z = 0.
As before, let it be n.ssuwetl t hat n > 0. The plane z = k iuter.
sects the surface in the hyperbola
If k > 0, the line x = 0, z = k is the tmnsverse axis and y = 0,
z = k is the conjugate axis. 1f 7.; < 0, the n.xf's are intf'rchanged.
The lengths of the semi-axes increase without limi t as 11.:1 increases.
-::t
ARTS. ()1, 62) . THE QU.-\DRIC 71
\\'hen k = 0, the section of the paraboloid consists of the two
lines
z=O :: _ y_ =O, z= O.
ct b ' a o
FlO. W.
The sections of the surface hy t.he planes y ;:== k' are the con-
parabolas -t
. ,
k'2
:....= 2 11Z +-:;-, y = k'.
u- lr
The ,-ertices of these
parabolas describe the
parabola
1
'
112
L-- '> ll"" l:- 0
lP - - N' . - .
The sections by the
p 1 an e s .1 = k" a r e
congment )':traholas
whose ices de-
scribe the parabola
x2 ..,
-;; = - liZ, y = 0.
((
62. The quadric cones. The cone (Art . 4G)
rJP 112 z2
-+,!... --=0
a
2
c
2
I
72 FORMS OF QUADRIC SURI<'ACE::> (CHAP. VI.
is called the real quadric cone. its vertex is at the origin. The
section of the cone by the plane z = c is the ellipse
x2 y2
- + - =1, z =c.
az IJ2
The cone is therefore the locus of a. line which passes through the
origin and intersects this ellipse.
If a= b, the surface is the right circular cone generated by re-
volving the line :.!3 = z, y = 0 about the Z-axis.
a c;
The equation
represents an imaginary quadric cone. There are no real points
ou it except the origin.
63. The quadric cylinders. The cylinders (Art. 43) whose
equations are
are called elliptic, hyperbolic, imaginary, and parabolic cylinders,
respectively, since the sections of them by the planes z = k are
congruent ellipses, hyperbolas, ima.giuaty ellipses, and parabolas,
res peeti vel y.
64. Summary. The surfaces discussed will be enumerated
again for reference.
, : ~ ll z2
- + + - = 1. Ellipsoid.
(t2 1,2 c2
(Art. 56)
x ~ ?/ z
2
. +
6
.- -;, = 1. I fyperboloid of one sheet.
a- . c
(Art. 57)
~ _ lt. _ z ~ = 1. Hyperboloid of two sheets.
(12 IJ2 ('2
(Art. ,")8)
x_2 +?_P + z_2= -1.. J . ll' .d
. magwary e 1 ps01 .
a2 IJ2 c2
(Art. 59)
x} /
- + '- = 2 nz. Ellipt.ic paraboloid.
a2 b2
(Art. 60)
ART. 641 SlJM.MARY
x2 y2
- - ._ = ~ 11z. Hyperbolic paraboloid.
az IJ2
xz + 1_.1 z-_z2 -- 0. R l d .
ea qua nc cone.
a ~
0
2 c2
x2 y2 z2
-+- + -=0. Imaginary quadric cone.
a' uz ':l
xz y2 - 1 . z ? Q 1 . l' '
u
2
b2- , y = -1xc. uac r1c cy muers.
EXERCISES
Classify the following surfaces :
l . 4 z2- 6 xz + 2 y
2
= 3.
2. xl + 3 yz + 5 x + 2 y + 7 = 0.
3. ,.z + 8 yz + 4 x- 2 z = 0.
4. 4 xz + 4 y2- 3 z2 = 0.
5. 2 z
2
- x
2
- 8 yz - 2 x - 12 y = 15.
6. xz - 2 yz - li y - G z = 0.
73
(Art. 61)
(Art. 62)
(Art. 62)
(Art. 63)
7. Find the equatit1n of and classify the locus of a point which moves so
that (a) the Sllm of its distances, (b) the difference of its distances from t\\'O
lixed points is constant. Take the points (a, 0, 0).
8 . Find and classify the equation of the locus of a point which moves so
that its distance from (a, 0, 0) bears a constant ratio to its distance (a) from
the plane x = 0; (b) from the Z-axis.
9. Show that the locns of a point whose distance from a fixed plane is al-
ways equal to its distance from a fixed line perpendicnlar to the plane is a
quadric cone.
10. A line moves in such a way that three points fixed on it remain in
three fixed planes at right angles to each other. Show that any other point.
fixed on the line describes an ellipsoid. (Sue. Find the direction cosines of
the line in terms of the co()rdinates of the point chosen. and substitute in
formula 0), Art. 3.)
CHAPTER VII
CLASSIFICATION OF QUADRIC SURFACES
65. Intersection of a quadric and a line. 'l'he most general form
of the equation of a quatl ric snrfac.e is (Art. :15)
F(:t, y, z) = ax
2
+ uy
2
+ cz
2
+ 'J.f!JZ + :3 + 2 hxy
(1)
" 'e shall suppose, unless the contrn. ty is stated, that the coeffi
cients are nJI real, and that the coefficieut.s of the second-tlegree
terms are not all zero.
To determine tlte points of intersection of n given li ne (A1t. 20)
x = :t'
0
+ A?', !I = Yo + p.r, z = z
0
+ vr (2)
with the qn::ulrie (1), subst.itute the values of x, y, z f1o:n (2) in
F(x, y, z) ::mel arrange in powers of r. 'l'he result is
Qr
2
+ :? Rr + S = 0, (3)
in which
Q =: a.\
2
+ bp.
2
+ cv
2
+ 2fp.v + 2 gv>.. + 2Hp., (1)
R = (a:r
0
+ h!!o+flz
0
+l) .\+ (h.t
0
+b!Jo+ fz
0
+ m)p. + (gx
0
+fy
0
+czo+n)v
= !(aF A +.aF,.,. + aPv)
a.-c
0
oy
0
Dz
0
'
S = ?Jo, zo)
The roots in ,. of equation :ue the distances from the point
P
0
=. (x
0
, y
0
, z
0
) on the line to the points in which this line
inte1sects the quadric.
If Q * 0, equn.tion (:))is a quatlmtir. in ,.. If Q = 0, but Rand
S are not. both zero, (3) is st.iJI to be consideted a qun.dratic, with
one or more infinite roots. If Q = U = S = 0, (2) is satisfied for
all values of,. and the corresponding line l ies enti rely on the
quadric. \Ye hav(, consequently, the following theorems:
THEORIDI I. El'ei'!J line 1rhich does uot lie f1n <t rJitH!II quacll'ic
swface has two Oi' coincident) points in common with the
SlO:Jitce.
74
ARTS. 65, 661 DIA .. \l8TI1.\L PLAXES, 75
THEORE)I J I. If a gireu line has more than t!co points in common
with a given 'JIIaclric, it lies entirely on the quaclfic.
For, if (3) is s:ttisfied by more than t,,o values of 1, it is satis-
fied for aU values.
66. Diametral planes, center.
intersection of the line with
is called a chord of the qtHttlric.
Let P
1
;mel P
2
be the points of
the quadtic. The segment P
1
P
2
Tm:oRE;\t I. The loc11s nf the middle J'oint of a system of purallel
chor<ls of a quadric.; is a 7llane.
Let r
1
and r
2
be the roots of (3) so lhat P
0
P
1
= r
1
and P
0
P
2
= 1
2

The condition that P
0
is the middle point of the chord P
1
P
2
is
P
0
P
1
+ P
0
P
2
= 0,
or
Hence, from (4). we l1avc
(axo + hyo + [!Zo + 1).\ + (!..ro + U.lfo +.fzo + m)Ji
+ (y.to + .fUo + czo + n) v = 0. (5)
If, now, ,\, p., 11 are constants, but X
0
, y
0
, z
0
a re allowed to vary,
the line a system of pnxallcl lints. 'rhe locus of the
miudlc points of t he chords on these lines is given by (5). Since
(5) is linear in

y
0
, z
0
, this locus is a plane.
Such a plane is callctl n diametral plane. ,
TnEOt:Im I 1. A .dllt!te ml of a 'JU(Hhic have at least
one (finite ot injinitr).JIOh<l in common.
For all values of .\, p., 11 the plane (:"5) passes through t,}1e inter-
section of the planes
a:c + lty + (IZ + I = 0;
hx + b!J + .fz + 111 = 0,-
(f'l; + fy + t'Z + 11 = 0.
(6)
In discussing the locus determined by (6), it will be convenient
to put, for bre,,ity,
(( h
!I
' ((
h
!,J
a q l h g
D==. h b .r,
sr=
h b Jf-=. ltfm, L==. b
.r
m. (7)
g
J
c
,r/ f nl
y c n
.r
c n
'} I
QUADRIC SURFACES [CUAP. Vl l.
1 D =1= 0, the planes (6) intersect in a single finite poi nt
(Art. 26)
L
X=--,
lJ
Jf N
y = D' z =- ]}
(7)
Tf this point (x
0
, y
0
, z
0
) does not lie on the surface, it is called
the center of the quadriu. It is the middle poi nt of every chord
through it. If the point (x
0
, y
0
, z
0
) does lie on t he surface, it is
called a vertex of the quadric. In either case the system of planes
(:)) is a bundle with vertex aL
(
- !:. J1f - N)
D' D' D.
If D = 0, but L, 11, N are not all zero, the planes (6) intersect
in a single infinitely distant poi nt, the homogeneous coordinates
of which are found, by making (6) homogeneous and solving, to
be (L, -11[, N, 0). The system of planes (5) is a parallel bundle.
The quadric is, in this case, s:tid to be non-central.
If the system of planes (6) is of rank two (Art. 35), the planes
determine a line; the diamctral pl:l.nes (5) constitute a pencil of
pl:mcs through the line. If this line is finite and does not lie
on the quadric, it is called a line of centers ; if it is finite and does
lie on the quadric, it is called a line of vertices. If the system is
of rank oue, the diamettical pl::tnes coincide. If each point of this
plane does not lie on the quadric, it is called a plane of centers ; if
every point of the plane lies on the quadric, it is called a plane of
vertices.
ExAmu:. F ind the center of the quadric
z2 + 4 y2 - z2 + 4 ~ y + 4 yz + 2 zx + 2 x + 4 y - 2 z + d = 0.
The equations (G) for determining the center are
X + 2 y + Z + 1 = 0, X + 2 !J + Z + 1 = 0, X + 2 y - z - 1 = 0,
from which x + 2 y = 0, z + 1 = 0. This line is a li ne of centers unless
d =- I, in which case it is a line of vertices.
EXERCI SES
1 . Find the coordinates of the points in which the line x = 1 + ~ r,
l!
z = -I+!: intersects the quadric ,.2 + 3y2- 4 z2+4 z- 2y-
3
AnTs. 66, 67) QUADRIC RElt'ERRED TO ITS CENTER 77
Find which of the following quadrics have centers. Locate the center
when it exists.
2. z2- 2 yz + 6 z ~ + 12 xz - 11 = 0.
Y 3 . 2 x2 + y2 - zZ - 2 XZ + 4 X!f + 4 !fZ + 2 y - 4 Z - 4 = 0.
4. xy + yz + zx - :t + 2 y- z - 9 = 0.
5. 2 x
2
+ 5 yz + z2- 4 xy- 2 x- 4 y- 8 = 0.
1""' 6. z2 - xz - yz- z = 0 .
.1; 7. xz + y2 + z ~ - 2 yz + 2 xz - 2 xy - x + y - z = 0.
S . xz + 4 yz + z2- 4 yz- 2 xz + 4 xy + 10 x + 5 y- 7 z + 15 = 0.
9. Show that any plane which passes through the center of a quadric is
a. cliametral plane.
10. Let P
1
and P
2
be two points on an ellipsoid, and let 0 be it.s center.
Pro,e that if P
1
is on the diametral plane of the system of chords parallel to
0 P
2
, then P
2
is on the cliatnetral plane of the system of chords parallel to
OP1. - ~ .
- ~ . . ? A . <
() 7. Equation of a quadric referred to its center. If a quadric
has a center (x
0
, y
0
, z
0
), its equation, r eferred to its center as origin,
may be obtained in the following way :
If we tra.nslate the origin to the center by putting
x = x' + Xo, y = y' + y
0
, z = z' + z
0
,
the equation F(x, y, z) = 0 is transformed into
ax
12
+ by'
2
+ cz'
2
+ ?.f!/z' + 2 gz'x' + 2 hx'y' + 2(aXo + hy
0
+
gz
0
+l)x' + 2(hxo+ by
0
+ fz
0
+ 1n)y' + 2(gx
0
+fy
0
+ cz
0
+n)z' + S = 0
wherein, as in Eq. (4), S = F(xo, Jt
0
, z
0
) .
Since (Xo, y
0
, Zo) is the center, it follows from (6) that
ctXo + hyo + !7Zo+ I = 0,
hx
0
+ by
0
+.f'z
0
+ m = 0,
9Xo + l!Jo + I.'Zo + II = O,
(8)
so that the coefficients of x', y', z' are zero, and the equation has
the form (after dropping the accents)
ax
2
+ by
2
+ cz
2
+ 2 jyz + 2 gzx + 2 hxy + S = 0. '- (0)
The function S = F(x
0
, y
0
, z
0
) may be written in the form
S = F(:r
0
, y
0
, z
0
) =
Xo(aXo + hyo + gzo + l) + Yo(h.t,. + O!Ju + fzo + m) + Zo(9Xo +.fYo+Czo+
n) + lxo + m.lfo + nz
0
+d.
78 QUADRIC SURFACES [CHAP. VII.
Hence, from (8) we have
S = /.l'o + ?n!Jo + nz
0
+ cl. (10)
By el iminating x,.,, y
0
, Z
0
from (8) and (10) we obtain the relation
(L h I) l
h b .r ?It
s
!! f c n
1n u cl- S
This equation may be written in the form
h
a h
[/ l
(1,
!I
II b
.f S=
h b
f
?It
...
.r
'.!
r
c n
g c
l 1n n cl
Denote the right-hauc"i member of this equation by A. The
coctficient of S is D (Eq. 7). lienee
DS= t.,
or, if D =P 0,
(q)
If D =P 0 and t. = 0, it follows from (9) and (11) that the
quadric is a cone (Art. 4G). The vertex of the quadric is the
vertex of the cone. . ( 1
If t. = 0 and S =P O, then D = 0. Since (x
0
, y
0
, z
0
) was
assumed to be a finite point, it follows that L = M = N = 0 so
that the Rmface has a line or plane of centers.
If t. = 0 and S = D = 0, then from (9) the smface is com-
posite. Every point common to the component planes is a vertex.
Tbe dctermiuaut t. is called the discriminant of the given
quadric. 1 f t. = 0, the quadric is sn,id to be singular. If t. =P 0,
the quadric is non-singular.
(j8, Principal planes. A <liametral plane which is perpendic-
ular to the chords it bisects is call e<l a principal plane.
TnEORJDJ. l.f the coe.f)icicnts 7.Jt the eqwclion of a qtlCcdric arc
?'l'ftl, aud {f the quadric !loes uot hare the plttue at il(/illii!J as a com,- I
ponent, the quadric has ca least 011e I'Ntl,jinitf', J>rhuiJ1nlz1lane.
ARTS. 68, 09] THE DISCRil\IINATING CUBIC
The condition that the tliametral plane (5)
(aA + hp. + gy):c + (hA + bp. + .fy)y + (gA + f p. + Cy)z +
lA+ ?Itp.+uv=O
is perpendicular to the chords it bisects is (Art. 14)
aA + ltp. + gy = ltA + bp. + fy = f!A + fp. + cv.
A p. Y
79
(12)
If we denote the common value of these C..actions by k, equa-
tions (12) may be replaced by
(ct - k)A + hp. + gy = 0,
hA + (b- k)p. + fy = 0, (13)
gA + fp. + (c- k)v = 0.
'l'be condition that these equations in A, p., Y have a solution other
than 0, 0, 0 is '
ct-k
h
(!
h
b-k
f
y
f =0,
c-k
or, devel oped and nrrangl'd in powers of k,
(14)
J..:! - (ct + b + c)l,a +(ttb + bt + ca -/
2
-(!
2
- h
2
)k- D=O, (15)
where lJ has the n,eaning as i n (1). T his equation is call ed
the discriminating cubic of the quadric P(.r, y, z) = 0.
To each root, tlifl'ennt from zero, of the discriminating
cubic corresponds, on acrount of ( 12), ancl (:3), a real finite
principal plane. Otu theorem will conseqtll'ntly l>e proved if we
show that equation (15) has at least one real root different from
zero. T he proof will ue given in the next article.
G!l. Reality of the rol.lts of the discriminating cubic. 'Ye shall
first prove the following theorem :
THEOREM I. The 1oots of the disc1iminating are all real.
Let k
1
be any root of (1:'5) and let A.
0
, P-Q, Yo (not all zero) be values
of A, p., Y that satisfy (13) when k = k
1
If k is a complex number,
A.
0
, lAo v
0
lllay be Let
A
0
= A
1
+ i>..
1
1
, p.
0
= p.
1
+ ip.'
11
Yo= Y
1
+ iY'
11
where i = v'-=l;:.nd A
11
A
1
11
p.
11
p.
1
11
Y
11
v\ arc real.
80 QUADRIC SURFACES [CHAP. VII.
Substitute k
1
and these values of ''- 1 ~
0
, v
0
for k, >-, p., v in (13),
multiply the resulting equations by A.
1
- t'> .. 'u p.
1
- ip.'., v
1
- iv'u re-
spectively, and add. The result is
(,V + >-? +P.t2 + P.?+ Vt2 + v?)kt =(,\.2 + >-'.2)a+ (P.t2 + p.'.2)b
+ (v
1
2
+ v'
1
2
) c + 2 (p.
1
v
1
+ p.
1
1
'v'
1
)f + ~ (v
1
A
1
+ v't,\'1) g
+ 2(A.tP.t + >-'.p.\)h.
The coeHlcient of k
1
is real and different from zero. The number
in the other member of the egnation is real. Hence k
1
is real.
Since X
1
is any root of (15), the theorem follows.
THEOltJ.:.II II. ~ V o t all tlte 1oots of the discriminating culAc are
equal to zero.
The condition that all the roots of (15) are zero is
a+ lJ + c = o, ab + be + C(t- p- rl- h
2
= 0, D = 0.
Square the fitst member of the first ecptation, antl subtract twice
the first member of the second from it. The result is
tt
2
+ b
2
+ c
2
+ 2.[2 + 3 g
2
+ 2 h
2
= 0.
Since these numbers are rea.l, it follows t hat
ct=b = t:=f=g=!t=O;
but if these conditions are satisfied, the equation of the quadric
contnins no term in the second clegree in x, y, z, w bich is contrary
to hypothesis (Art. G:3).
'
. ..
70. Simplification of the equatiOnof a qu; dric. Let the axes be
transformed in such a way that a real, finite principal plane of the
quadric F(x, y, z) = 0 is taken as x = 0. Since the surface is now
symmetri c with respect to x = 0 (Art. G8), the coefficients of the
terms of first degree in x must all be zero. H ence the equation
has the form
ax
2
+ byZ + cz
2
+ 2fyz + 2'my + 2nz +<l = 0.
:Moreover, ct =1= 0, since otherwise x = 0 would not be a principal
plane (Art. 68).
Now let the planes y = 0, z = 0 be rotated about the X-axis
through the angle 8 uefined by tan 2 8 = .J..L_. This rotation re-
b-c
ARTS. 70, 71) CLASSIFICATION OJo, QUADRICS 81
duces the coefficient of yz to zero, and the eqnation has the form
+ b'y
2
+ + nt'.IJ + :3 n'z + (l' = 0, (16)
wherein n' * 0, but auy of the other coefficients may be equal to zero.
71. Classification of quadric surfaces. Since the equation of a
quadric can always be reducetl to the form (1G), a complete classi-
fication can be made by consitlering tho possible value:; of the co-
efficients.
I. Lrt bot.h b' and 1'
1
be different from zero. By t ranslation of
the a.xos 111 such a way that 0, - ,- , -
1
- IS the new origin,
. ( -m.' -n').
b c
the equation reduces to
+ + c'zt = <l".
If d" * 0, divide by <l" anrl put
d"
-, =a\
((
rl" cl"
_ _ 1.2 _ c2
I
I- v, )
I I'
the signs being so chosen tbat a, b, c arc real. This gives the fol-
lowing four types:
Ellipsoid. (Art. 56)
lfyperholoitl one sheet. (Art. 57)
Hyperboloid two sheets. (Art. 58)
Imaginary ellipsoid. (Art. 59)
.
If cl" = 0, the retlucccl forms arc
xz l z2
- + ,__ +-= 0. lmagiuary cone.
az IJ2 c2
(Art. 62)
':1!- ,
1
z zZ
- + --= 0. Real cone.
az IJ2 c2
(Art. 62)
II. l,et c' = 0, b' * 0.
If n' * 0, by a translation of axes, the equations may be re-
duced to
a'x
2
+ b'y
2
+ 2 11
1
z = 0.
82 QUADRIC SURFACES
This equation takes the form
x'
- +- = 2nz. Elliptic paraboloid.
((2 b2
[CBAP. VII.
(Art. 60)
or ?!. -!f.= 2 nz. Hyperboli c paraboloid. (Art. 61)
az bz
accordi ng as (t' and b' have the same or opposite s igns.
If n' = 0, the equation may be reduced to
ct
1
x
2
+ b'!/+ cl'' = p.
If d" =1= 0, this may be written in the
J1. 1 = 0. Qn:td;ic
b2
(Art. 63)
and if d" =0,
.,p
- ,_ = 0. Pair of inte>rsectiug planes.
nz bz
Ill. Let b' = c' = 0. Equation (tG) is in this case
a'x
2
+ 2m'y + 2 n'z + <I'= 0.
If 1n and narc not both zero, since the plane 2 m'y + 2 n'z + d'
= 0 is at right angles to x = 0, we may rotate and translat e the
axes so tlmt this plane is tbe new y = 0. The equn.tion of the sur-
face becomes
aP = 2 my. Paraboli c cylinder. (Art. 63)
If m' and n' arc both zero, we have,
if d' =1= 0, x
2
k
2
= 0. Two parall el planes.
if d' = 0, x
2
= 0. One plane counted twice.
72. Invariants uuder motion. A function of the coefficients of
the equati on of a smface, the value of which i:; ul,.lchanged when
t he axes arc rotated and tr:tnslatcd (A rt.s. 3(i and 37), is called an
invariant under motion of Lite gi,cn surface. It will be shown
that the expressions
J =.u+ b+c,
J =. be+ ca +ab -P- rf-11
2
,
I
n h yl
D=. h b f ,
,(j f c,
(t h
(I
h b
.r
?n
g
f
c n
,l ?n n
(l
ARTS. 72, 73) INVARIA"l\'CE OF /, J , AND D 83
fonned from the coeftlcients of the equation (1) of a quadric are
invariants under motion.
J J
J.'
73. Proof that I, J, and D are invariants. When the axes an
(Art. 3G), the coeflicients of the ter1ns in the second
degree in the equation of a quadric are unchanged. Hence I, J,
and D are unchanged.
Since the equations of rotation (Art. 3i) are linear and homo-
geneous in x, y, z, x', y', z', the d<:'gree of any term is not changed
by these transformations, so that a term of the ll!-st degree can-
not be made to be of the secowl, nor eon ''ersely. Suppose the
expressio11
f(:c, y, z) = ax
2
+ 2 hxy + by
2
+ 2 gxz + 2 f!lz + cz
2
is transform<'d by a rotation into
f'(x', y', z')= (,'x'
2
+ h'x'y' + b'y'
2
+ 2 y'x'z' + 2j'y'z' + c'z
12

Now consider the function
(.r, y, z):: f(x, y, z) - + y
2
+ z2).
The expression 1,.2 + !l + z
2
is the SC[nare of the distance of a point
(x, y, z) from t.he ori gin, aJHl will therefore remain of the same
form :<'
2
+ y'
2
+ z1
2
by the transformation of rotation ( Art. :ri).
If, .f(:c, y, z) is changed into f'(.e', ,1/, z'), y, z) will be
changed into
cp'(x', y' z') =f' (x', y', z')- + y'
2
+ z'2).
If k has s tch a value that cf> is the product of two linear factors in
x, y, z, then, for the same value of k, the expression ' will be the
product of two linear factors iu x', y', z'. The condition that cp is
the product of two factors is that its discr iminant vanishes, that
is
a-k h g
h b - k
f
=0,
(/
f
c-k
which, developed in powet's of k, is exactly the equation of the dis-
criminating cubic (Art. 68)
J...:l - I k
2
+ J k- D = 0.
84 QUADRIC SURFACES [CHAP. VII
Similady, the condition thn,t cp' is the produc:t of two linear fac-
t.ors is J._:J- I'k
2
+ J'k- D' = O,
where J', J', and D' are the expressions I, J, and D formed ftom
the coefficients of f'(x', y', z').
These two equations have the same roots, hence the coefficients
of like powers of k must be ptoportional. But the coefficient of k'
is unity in each, hence,
1'=1, J'=J, D'=D,
that is, I, J, Dare invariants.
From the theorem just proved the following is readily obtained:
When the axes are tTansfonnecl in such a u:ay that
the coe.Oicie11ts of xy, yz, ancl zx ate all zero, the coe.(ficients of??, y
2
,
ancl z
2
are the 1oots of the cliscriminatioy C'lbic.
l!'or, if the equation of the quachic has been reduced to
+ b'y
2
+ c'z
2
+ 2l'x + 2 1n'y + 2 11
1
z + cl' = 0,
the disctiminaLing cubic is
k
3
- (a'+ b' + c')k + (ct'b' + b'c' + c'ct')k- <c-'b'c' = 0.
The routs of this eqnation are (t', b', and c'. This proves the
proposition.
From the theorem just proved, the following criteria immedi-
ately follow:
If two roots of the discriminating cubic are equal and different
from zero, the rptathic is a surfac:c of revolution, and conversely.
Tf all tlll''e roots of the discriminating cubic are equal and
different from zero, the quadric is a sphete.
Lf A =1= 0, aud a root of the discriminatiug cubic is zero, the
qua.<hic is nOll-central.
If two roots of the discriminating cubic are equal to zeto, the
terms of secoud degree in the ecluation of the qun.dtic form a
perfect sq nare.
74. Proof that 1:t.. is invariant. It will first Le pi'Oved that A is
invariant under rotatiou. '!'he reasoning is similar to that iu
Art. 7:t Let
F(x, y, z)= a:t.'l + by
2
+ cz
2
+ 2 fyz + 2 gz:c + 2 hxy + 2lx + 2 my
+2 nz+d=0
ART. 7-11 PHOO.f<' THA'l' t:. JS INVARIANT 85
he the equation of the gi \cn quadri c. Let this equation be trans-
formed by a rotatiou in to
F(x', y', z') = a'x'Z + b'y'2 + c'z': + f'y'z' + 2 y'z'x' + 2 h'x'y' + Zl'x'
+ 2 m'y' + 2 n'z' + cl' = 0.
This rotation transforms the expression
y, z)= Fe.l;, ?/, z) - k(.1:
2
+ y
2
+ + 1)
into !1', z') = F'ex', y', z')- ke.1:'
2
+ y'
2
+ z'
2
+ 1).
The uiscriminants of atHl are, respectively,
(t-k h y
"'-k
h' g' l'
It u-k
.r
?n
and
h' u' -k
f'
m'
g
.r
c-k n g'
f'
c'-k n'
l ?It IL <l-k l' ?n' n' d' -k
The roots of the quartic equations in k obtained by equating these
tliscrimi nants to zero are equa.l; since a value of k which makes
<I>= 0 singular also makes <1>' = 0 singular and conversely (Art. 67).
Hence, since the coefficient of k
1
in each equation is unity, t he
constant terms are equal; that is, 11 = 11'. Hence, 11 is invariant
under rotation.
In order to prove that A is invariant under translation, let the
axes be translated to parallel axes through ex
0
, y
01
z
0
) . 'fhe equa-
tion of the quadric becomes (cf. Art. G7)
F'ex' , y', z') = ax'
2
+ uy'
2
+ cz'
2
+ 2 + 2 Y!t'z' + 2/tz'x'
+ 2 (a.1
0
+ htfo + rJZ
0
+ l) x' + 2 (hx
0
+ uy
0
+ fz
0
+ 1n) y'
+ 2 (y.t
0
+ fyo + cz
0
+ n)z' + S = 0,
where S = F (.t
0
, y
0
, z
0
). The cliscriminant of F(x', y', z') is
a h y axo+hYo+!lzo+l
h b f fzo+m
!I f c gxo+ (yo+czo +n
axo+hYu+Yzo+l hxo+byo+Jzo+m g.tu+luo+czo+n S
Multiply the first column by :l'
0
, the second by y
0
, the third by z
0
,
and subtract their sum from the last column. In the resulting
determinant, multiply the first row by x
0
, the second by y
0
, the
third by z
0
, and subtract their sum from the last row. Finally
diYiue the first row and column each by x
0
, the second row and
column each by y
0
, ancl the third row and column each by z
0

86 QUADRIC SURFACES [CHAP. VII.
The resulting determinant is Hence is inva-
riant under translation. is invariant unaer both transla-
tion and rotation, it is invariant under motio11 .
..:!.!!.:- Discussion of numerical equations. ln order to determine
the form aud position of a quadric with a equa-
tion, it is ad vi sable to tletennine the sta.ntlard form (Art. 71) to
which the equati on of the given quadric may be reduced, and the
position in space of the coordinate axes for which the equation
h:1.s this stn,nd:l.rtl form. .B'or this purpose the roots ku k
2
, k
3
of
the lliscri m i nating cubic and the value of the discriminaut
should first be computctl.
A. al l the roots ku k
2
, k
3
n, re different from zero, the three
principal planes mn,y be determined as in Art. 68. If these planes
are taken as coordinate planes, the eq nation reduces to (Art. 67,
11; Art. 73)
B. If one root k
3
is zero, two finite principal planes may be.
det ermined as before. Let these be taken as x = 0 and y = 0.
At least, one intersecLion of the new Z-axis with Lhe smface is at
infinity. If this axis does not lie on the surface, and does meet
the !>nrface in one finiLe point, the a,xes .should be transl:l.ted to
this point as origin. The equation of the s urface now has the
form

+ l.:2Jl + n"z = 0.
Since
"'
0 0 0
0

0 0

0 0 0
lllt J
0 0 n" 0
it follows that

n"= .
klk2
If the new Z-axis lies on the quadric, or if it has no finite point
in common with it, any point on the new Z-axis may be chosen for
origin and the equation takes the form
k . -c
2
+ kiJJ
2
+ s = 0,
ART. 751
DISCUSSION OF EQUATIONS 87
where (.Art. 67)
S = lx
0
+ my
0
+ nzo + d,
and (:r
0
, y
0
, Zo) are the old coon.linates of the new origin.
0. If two l'OOtS of ti re discriminating cubic :u-e zero, the terms
of the second deg1ee in the original equation form a; perfect square,
so that the equation of the surface, refened to the original axes,
is of tho form "
(u.t: + f3!J + yz)
2
+ 3 lx + 2 my+ 2 nz + cl = 0,
or (u.l: + {3y + yz + 8y + 2(l - a8).l; + 2(m- {38)y + 2(n- y8)z

If the planes a..l: + {3y + yz + 8 = O,
2(l - r'8)x + 2( - {38)y + 2( n. - y8)z + cl - 8
2
= 0
are not we may choose 8 so that they are perpendicular.
The first term of (17) is proportional to the square of the distance
of the point (x, y, z) from the plane
nx + (3y + yz+ 8= 0 .
. . The remaining terms of (17) are proportional to the distance to
the second plane. If these planes, with the appropriate value of
8, are chosen as x = 0, !J = 0, the equation reduces to
(1,
2
+ (32 + '/)y2 + Zv'(l- u8l+(m- {38)
2
+ (n- y8)
2
x = 0.
If the two planes are parallel, 8 may be so chosen that
l- lc8 = 0, ?n - {38 = O, n - y8 = 0.
The equation now becomes
("2 + (32 + y2)y2 + cl - 82 = 0,
whrrein a.t + {3y + yz + S = 0 is the new y = 0.
EX.\liPI. 1. Discuss the eqnation
,-2- 2 y2 + () + 12 :rz- JG .- 4 y- 36 z + U2 = 0.
The equations 1hc CE'nter are r + (i z- 8 = 0, 2 y + 2 = 0,
6 .r + ti z- 18 = 1), from which the coordinates of the center are (2, - 1, 1).
The itwariants are I= 5, J = - 4-1, D = (jQ, A = 1800.
li enee, the discriminating cubic is
J..-3- 5 k
2
- 44 k- GO = 0.
Its roots are k
1
= 10, k
2
=- 2, k
3
=- 3. The transformed equation is
1Qz2- 2 y
2
- 3z2 + 30 =0.
88
QUADRiC SUltl<'ACES [CHAP. VII.
The direction cosines of the new axes through (2, - l , 1) are found. as in
Art. 68, by giving k the values 10, - 2, - 3, to be
2 3 -3 2
- .- , o, .,-i o, 1, 0; ,- o, '- 13.
Vl3 vIa v 13 v
The surface is an hyperboloid of one sheet.
'
BxA)Jt>r,E 2. Discuss the quadric
11 :r} + 10 yt + G z2- 8 yz + -1 zx- 12 xy + 72 x - 72 y + 36 z + 150 = 0.
The discriminating cubic is
/,;3 - 27 k2 + 180 k- 324 = 0.
Its roots are 3. G, 18. A =- 3888. The surface is an ellipsoid.
The equations for finding tht:: center arc
11 x - (l y + Z z + ilfi = 0, -fi X+ 10 y- 4 z- 3G = 0,
i :r - -1 y + G z + 18 = 0.
The coortlinates of the center are ( - 2, 2, - 1) . Tbe direction cosines of
the axes are
!. , L ;. t. - L - J, J, - !
The equation of the ellipsoid refernd to its axes is
.r2 + (i + 18 z2 = 12.
EXA)IPLE 3. Discuss the quae! ric
3 :r2- y2 + :1 + 6 yz- .J z:r - i -"!1 - 14 x + -1 y + 20 z + 21 = 0.
The discrimin:1.ting cubi c is
1.;3 - .j 13 k + 1!) = 0.
Its root.<; arc approximately 1. :2, 5. 7, - 2. fl . A = 0. The surface is a cone.
The equations for finding tho ,ertex are
3 X - !J - 2 z - 7 = 0, - :r - !I + il z + 2 = 0, - 2 X + 3 'J' + 2 ;; + 10 = 0.
The col\rdiuates of the vertex <trc ( 1, -2, - 1). The direction cosines of
the axe!; arc approximately
.8 . .4, .:) ; .6, - 4, -.7; 0, .G, - .4.
The equation of the cone rl.'ferrt!l to its axes is approximately
1.2 . 2 + ;,,7 y!- 2.Vz2 = 0.
4. Discuss the quadric
.j x2 + y2 + z2 - 2 yz + 4 n - -t :ry - 8 x + 4 z + 7 = 0.
This equation may be writt en in the form
(2 x - y + z + o)2 = (8 + 4 o)x - 2 oy -(-t- 2 o)z- 7 + o2.
If o =- l , the planes 2 X - y + z - 1 = 0 and 4 :r + 2 y - 6 z - 6 = 0 are
perpendicular. If we take these planes as y
1
= 0 and x
1
= 0, the equation of
the suriace reduces to 6 y
2
::: .Jf)(j x. The surface is a parabolic cylinder.
ART. 75] DISCUSSION OF NUMERICAL EQUATIONS 89
EXERCISES
Discuss the quadrics:
1. 3 ~ t + 2 yt + zt - 4 xy - 4 yz + 2 .. 0.
2. ~ t - yt + 2 zt - 2 yz + 4 ~ z + 4 ~ y - 2 x - 4 y - 1 .. 0.
>""3. 6 yt + 8 z
2
+ 6 yz + 6 ~ z + 2 ~ y + 2 ~ + 4 y - 2 z - 1 - 0.
4. 4 ~ ~ + y
2
- 8 zt + 8 yz - 4 u + 4 ~ y - 8 ~ - 4 y + 4 z + 4 = 0.
6. 3 ~ , + 2 y' + 2 z.
2
- 4 yz- 2 z.x + 2 xy- G x + 2 y + 2 z- 12 .. 0.
6. 6 x
2
- 2 zt - 6 yz - 6 zz - 2 xy + 2 x + 4 y + 2 z .. 0.
7. 4 zt + 4 y
2
+ z2- 4 yz- 4 xz + 8 xy- G y + G z- 3 .. 0.
8. x
2
- yz + xz - xy + x + y + 2 z - 2 - 0.
9. 3 y
2
+ G yz + 6 xy - 2 :t + 2 z + 4 - 0.
10. 3 x' + 3 y' + zt + 2 yz + 2 xz - 2 xy - 7 ~ + y + 6 z - 7 = 0.
11. 3 ~ ~ - 5 y
2
+ 15 z
2
- 22 yz. + 14 xz- 14 :z:y + 2 x- 10 y + 6 z- 5- 0.
12. x' - y
2
- 2 z
2
- 4 yz + 2 zy - 2 y + 2 z - 0.
13. :t
2
- G yz + 3 ZL + 2 xy + x - 13 z .. 0.
14. x
2
- 2 y
2
+ z
2
- 4 z.x - 12 xy + 4 y + 4 z - 9 .. 0.
16. :t
2
+ 2 y' + 2 z ~ + 2 xy - 2 x - 4 y - 4 z - 0.
16. 3 x
2
+ y
1
+ z
1
+ yz - 3 z.x - 2 xy + 2 x + 4 y + 2 z 0.
17. For what values of c is the surface
.') xt + 3 y
2
+ cz
2
+ 2 xz + 15 - 0
a surface of revolution?
18. Determine d in such a way that
~ ~ + y
2
+ 5 z
2
+ 2 Y + 4 xz. - 4 xy + 2 ~ + 2 y + d .. 0
is a cone.
CHAPTER VIII
SOME PROPERTIES OF QUADRIC SURFACES
7H. Tangent lines and- planes. 1f the two points of i ntersection
of a line and a quadric coiucitlc at a point P
0
, the line is Cf\llecl a
tangent line aud P
0
the point of tangency. If the surface is sin-
gular, it is supposcu in t hi s deti ui t iou that P
0
is not a vertex.
TnJWJtE.,r. The of the li11es tangent to the quaclric at P
0
is
<t plww.
Let the equation of the quadtic be
P(.c, y, z) := a.c
2
+ by
2
+ cz
2
+ '2.f!Jz + 2yz.c + 2/txy
+ 2 I.e + my + :! nz + d = 0, (1)
and let t he eq nation of any I i ue I' c = ( !1.
0
, y
0
, z
0
) be
(Art. 20)
X = .1
0
+ A1, y = Yo + p.l', z = z
0
+ vr. (2)
Since P
0
li es on the quadric, /<'(.t
0
, y
0
, z
0
) = 0. 1 fence, one root of
rqnat iou Chfl.pter Yfl, whieh <letf'rlllines t he intersections of
the line (2) with the qnathic (1), is zero. The condition that a
second root is zrro is ll = 0, o1
A (cu
0
+ hy
0
+ !JZo +I)+ p. (lt.r
0
+ by
0
+ /z
0
+ m)
+ v (gx
0
+ f!lo + I'Zo + n) = 0. (3)
if we substitute in the val ues of ,\, p.. v from ('2), we obtain
(.c- :t'
0
)(<t.t'
0
+ hy
0
+ IJZo + /)+(!!- .1/o)(h.to + li!Jo + fz
0
+ ?1t)
+ (z- z
0
)((J!i.'o + f!lo + CZo + n) = 0, (4)
which must be satisfied by t he coonlinates of every point of every
line tangent to the qua.dric at P
0
Cou vcrsely, if (x, y, z) is any
point distiuct from J>
0
, whose coordinates satis fy (4), the line de-
tennined by (x, y, z) n.nd P
0
is to the surface at P
0

Since (4) is of the first iu y, z), it is the equation of a,
plane. This plane is eallotl the tangent plane at Jl
0

!)0
ARTS. 7G, 77) TA)JQE};T PLA, E FORM
The equation (-l) of the tangent pLtne may be simplifi t>tl. .:\Iul-
t iply out, transpose the constant trrms to the second member, and
add /.r
0
+ 111!/o + uz
0
+ tl to each llll'mber of the equation. The
secontl membrr is F(.l'
0
, !lo z
0
), whieh is equal to zero. since 1'
0
li e::;
on the quadric. The equation of the tangent plane thus reduc{'S
to the form
rc:c.r
0
+ U?/!lo + t'ZZ
0
+ .f(yz
0
+ zy
0
) + !f (z.>
0
+

+ h (.l:!fo + y.r
0
)
+ I (. + .t
0
) + m (y + !fo) + 11 (z + z
0
) + rl = 0. (.1)
This equation is easily rememhcr!'<l. Tt may be ohtainerl ftom
the equation of the quadric by replacing .r
2
, !/, z
2
by xx
0
, !IJk, zz
0
;
:l !/Z, 3 z.t, X!J by !JZo + Z!/o, z:r
0
+ xz
0
, + !/Xo; and x, :? y, :! z by
+ .r
0
, !I+ y
0
, z + z
0
, respectively.
..
i 7. Normal forms of the equation of the tangent plane. The eq ua
tion of the tangent plane to the central quadric
rl.l'z + byz + czZ = 1 (6)
at the point ( J'
0
, y
0
, z
0
) on it is
(1.!'.>'
0
+ U.'l!lo + fZZ
0
= 1.
Let the normal form of the equation of this plane (Art. 13) be
so that
AX + t!l + vz = ]>, (7)
II
- = CZo
]1
Since (x
0
, y
0
, z
0
) lies on t.he quadric, we have
a.1'
0
2
+ uy
0
2
+ tz
0
2
= 1,
from which
(8)
Conversely, if this equation is satisfied, the plane (i ) is tangent
to the quatlric (6).
By substituting the value of p from (8) i n (7), we have
/
,\2 vz
AX + p?J + . z = \ - + I!:.. + - ,
(I b c
which is called the normal form of the equation of the tangent
plane to the central quadric (6).
PROPERTIES OF QUADRIC SURI<'ACES [CnAP. Vlll .
It follows from (8) that the necessary and sufficient condition
that the plane
!lX + V!J + WZ = 1
is tangent to the quadric (6) is that
= 1.
a b c
I -
(9)
This equation ts called the equation of the quadric (6) in plane
coordinates.
Again, if
(10)
is the equation of a paraboloid (Arts. 60 and 61), it is proved in a
similar way that the normal forri1 of the equation of the tangent
plane to the paraboloid is
.\x + p.y + vz = - - - + 1!:..
n ('v 2)
2 v a b
and that the condition that the plane
nx + vy + wz = 1
is tangent to the paraboloid is
2 2 l)

a b n
(11)
(12)
Equation (12) is the equation of the paraboloid in plane collrdinates.
78. Normal to a quadric. The line through a point P
0
on a
quadric, perpendicular to the tangent plane at P
0
, is called the
normal to the surface at P
0

It follows from equation (4) that the equations of the normal
at Po to the quadric F(x, y, z) = 0 are
x - = Jl - ?fo = z - Zo
a.:t
0
+ hy
0
+ gz
0
+ l hx
0
+ by
0
+.fz
0
+ m gXo +!Yo+ cz0 + n
(13)
EXERCISES
1 . Show that the point (1, - 2, 1) lies on the quadric yz + z2 +
4 yz + 2 zx + xy - z + y + z + 12 = 0. Wri te the equations of the tangent
plane and the normal liuc at this point.
2. Show that the etuation of the tan:.tt'nt plane to a sphere, as derived in
Art. 76, agrees with the equn.tion obtained in Art. 50.
ARTS. 78, 791 REC'1'1L!NEAR GENERATORS 93
3. Pro,e that the normals to a central quadric m;2 + byz + cz2 = I, at all
points on it, in a plane parallel to a principal p\aue, meet two fixed lines,
one in each of the other two principal planes.
4. Prove that, if all the normals to the central ttuadric axz + hyz + cz2 = 1
intersect the X-axis, the quadric is a surface of revolution about the X-axis.
5. ]'rove that the tangent plane at any point of the quadric cone
11xz + by, + cz2 = 0 passes through the vertex.
6. Prove that the locus of the point of intersection of three mutually per-
pendicular tangent planes to the central quadric a Y ~ + by2 + cz2 = 1 is the
concentric sphere xz + yz + z2 =! +! +!. This sphere is call ed the di rector
a b c
sphere o( the given central qnadric.
7. Prove that through any point in space pass six normals to a t:iven
central quadric, and five normals to a given paraboloid.
79. Rectilinear generators. The equation of the hyperboloid
of one sheet
may be written in the form
or
or also
: : + ~ 1+;'!
a c b
--=--
1-?/ ,; z
b a c
(14)
(15)
Let the value of each member in ( 14) be denoted bye, so that
by clearing of fractions we ha,e
~ + z = ~ (1 + tt). (1- '')= e(=-!).
etc b b ac
(16)
For each value of e, these equations define a line. Every point
on such a line lies on the surface, since its coordinates satisfy
94 PROPERTIES OF QUADRIC SURFACE:::> (CHAt'. VIII.
(14). ?lloreover, through each point of the smface passes a li ne
of tho system (16) since the coordinates of each point on the sur-
face satisfy and consequently satisfy (16). The system of
lines (l o), in which the parametet, is called a regulus of lines
on the hyperboloid. Any line of the regulus is called a generator.
Similarly, by equati ng each member of (Hi) to.,, we obtain the
system of lines whose equati ons arc
1 +?t =
(tC b b nc
in which., is the parameter. This system of lines constitutes a
second regulns lying on tho stnfa.ce. The two regnli will be
called the regulus aml the 'I rogulns, rl'spectively. Through
every point P of the surface passes onr, and but one, generator
belonging to each regulus. Moreover, any plane tha.t contains a
generator of one regulus contains a generato1 of the other regulus
also. The equation of any plane t hrough a generator of the
regnlns, for example, may be written in the Conn (A1't.
+.:- (1 + ?') = .,[(1 - !!.) - e - :)]
ac b b (tc
this equation may also be written in the form
+: - .,(1 -I!.)= e[(1 + !!.)-., _:)]
(tC 0 0 (1, <:
it. follows that this plano also passes t hrough a generator of t he
., regulus. Every snrh plane is tangent to the surface at the
point of intersection I' of the gener;\tors in it., since e"ery line
in tho plane through P has its two intersections with the surface
roincident at P.
The eqna.tions of the reguli on the hyperboloid
xz yz
-:t+o-z2=1
are y (!; )
l-
3
-z ;
<lllrl
1/ (X )
1+3='1 2-z .
The poinL (2, 6, 2) lies ou the surface. Tho \'alucs of {and 11 which
ARTS. 79, SO] ASYMPTOTIC CONE 95
determine the generators through this point are = 1, 11 =- 3. Hence, the
equations of these generators are

2 3' 3 !! t 2 1l t 3 2
The equation o[ the plane determined by these lines is
3 X+ -1 y- 12 Z- 0 = 0.
This is the equation of the tangent plane at (2, 0, 2) (Art. 76).
It is similarly seen that the equation
x2 y2
- - =2 nz
(t2 b2
of the hyperbolic paraboloid may also be written in the forms


aud
2nz
a b
x_y_
(t b 1
2?lZ = X?j='Y/
- + '-
(! "
Ilcn<".e, on this surface n.lso, there is a and an ., regulus
The generators of the e regulus are pamllel to the fixed plane
X T V
- = 0; those of the ., regulus, to the fixed plane + = 0.
u b u b
Ry writing tl1e abO\e e'lnations in homogeneous coordinates, it is
seen that the line :'!: + '!1. = 0, t = 0 i11 the plane at intlnity belongs
(( b
X
to the regulus; and the line -:... = 0, t = 0 to thP YJ regulus.
(t b
Hence the plane at infiuity is to the paraholoitl.
The hyperboloid of one sheet and the hyperbolic paraboloid are
sometimes called mled quadri cs, since the reguli on them are rf>al.
It will be shown (A.tt. lli>), that on e\ery non-singular quadric
there are two regnli; but, on all the quadrics except these two,
the reguli are imaginary.
80. Asymptotic cone. The cone whose vertex is the center of
a given central quadric, and which contains the ctuve in which
96 PROPERTIES OF QUADRIC SURFACES [Cll AP. VIII.
the quadric intersects the plane at infinity, is called the asymp-
totic cone of the given quadric.
If the equation of the quadric is
ctx2 + lJy2 tz2 = tt,
the equation of its asymptotic cone is
. \
ax
2
+ by
2
+ cz
2
= 0.
For, this equation is the equation of a cone with vertex at the
center (0, 0, 0, 1) of the given quadric (Art. 46). Its cune of
intersection with the plane at infinity coincides with the curve of
intersection
ax
2
+ by
2
+ cz
2
= t
2
, t = 0,
of the gi ven surface wi th that plane.
EXERCISES
tJ:':' Show that the quadric xy = z is ruled. Find the equations of its gen-
erators .
2. Show that 2 + 5 y - x + 8 z = 0 is a ruled quadric.
. 3. Provo thnt, for all mlues of k, the line x + 1 = k y = - (k + l)z lies on
the r;urface yz + zx + xy + y + z = 0.
4. Prove that (y + mz) (7 + nz) = z represents an hyperbolic paraboloid
which the _\"-axis and the J"-axis.
5. Show that every genemtor of the asymptotic cone of a central qnadric
is tangent to the surface at infinity. :From this proptlrty derive a definition
of an asymptotic cone.
6. Show that every generator of the asymptotic cone of an hyperboloid of
oue sheet is parall el to a generator of each regulus on the surface.
81. Plane sections of quadrics.
'l'uF.Oin; )t f. Tlte seclio11 of tt Cfll <td d e by a Jinite planP, 1uhich i8
,wl a <:omponent of lite sw')(we, a conic.
l<'or, let 71' be any given finite plane, and let the axes be chosen
so that the equation of this plane is z = 0. Let the equation of
the qua<hic, referred to this system of axes, be
ax2+lJy2+cz
2
+2fyz+2 [Jzx+211xy+2lx +2 ?ny+2 nz+cl=O. (17)
If, when z = 0, (17) vanishes identically, the given quadric is
I
ART. 80]
PLANE SECTIONS OF QUADRICS 97
composite and z = 0 is one component; other wise, the locus
defined in t he Xl':plane by putting z = 0 in (17) is a conic.
II. The sections of a qucultic by a systent of pwallel
plcwes are si?nilat conics ami similatly placed.
Let the axes be chosen so t hat the equations of the given sys-
tem of parallel planes is z = k, and let ( l 7) be the equation of the
gi\en quadric. 'fhe equation of the pwjecting cylillller of the
section by the plane z = k is
ax
2
+ 2 hxy + by
2
+ 2 ( l + gk):c + 2 (m + fk)!J + ck
2
+ 2 11k + tl = 0.
The curves in \Vhich these cylindets intersect z = 0, and conse-
quently (Art. 45) the curves of which they arc the projections,
are similar anrl similarly placed, since the coctlicients of xy,
and y
2
in the above equation are indepenrlent of k."'
The equations of the section of the surface by tl1e plane at
infinity are found by making ( 17) homogeneous y, z, t and put-
ting t = 0. They are
ax
2
+ by + cz
2
+ 2}!JZ +2 gzx + Z hx!f = 0, t = 0.
The locus of t hese equations is called the infini tely distant conic
of the quadric. This conic consists of two lines if the first mem-
ber of the first equation is the product of two linear The
condition for factorability is
D=O.
EXERCISES

Find the semi-axes of the ellipse in which the plane z = 1 intersects

th ad ric z2 + 4 y2- 3 zl + J yz - 2 :r - 4 y = 1.
Show that the planes z = k inter:;ect the 'JUadric 2 :rZ- yz + 3 z2 +
4 2 yz + 4 z + 2 y = 0 in hyperbolas. Find the equations of the locus of
thc;nters of these hyperbolas.
3 . Show that the curve of intersection of the sphere :r) + yz + z2 = and
2 ,. ,
the elli psoid :!:.. + 1!:: + = 1 lies on the cone
a
2
b
2

(
..!.._ l )z2 + (..!..-l)yz + (.!.- .!.)z2 = 0.
al ,.2 1
1
2 rz cz rz
Find the \alues of r for which th is cone is composite. Show that each com
ponent of the composite cones intersects the ellipsoid in a circle.
Ci. Salmon, "Conic Sections," 6tb edition, p. 222.
9 PROPERTIES Olt' QUADRIC SURI<'ACES (CoAP. VIII.
82. Circular sections. We shall prove the following theorem:
'l' l mORE)f I. ThrOHf!h each 1ectl, jiuilc J>oiut i>l space six
p/wl(!s tdtich i11tersect ct uh:eu non-spherical IJIIadric
in circles. If this quadric is uol et xllljnce of revolution nor et J>Ctra-
bolic cylinrler, these plcwt>s arc i tco are real ancl.four
(lte iuwginary. If lite qwulric is a stn:fcue of retolution or a
bolic C!Jli11Cler, .four of the plcmes rtre real wul coincident aucllwo are
t!J
'l'wo proofs will be given, based on different principles.
Proof I. Since pn.rallel sections of a quadric are simi lar, it
\\'ill suffice if \t"C pro,e this theorem for planes through the origin.
The planes through any other point, parallel to the planes of the
circular sections through the otigin, also intersect the quadric in
circles.
Let the axes be chosen in such a. way that the equation of the
quadric is (Art. 70)
+ k'J]J
1
+ k
3
z
2
+ + 2 ?II!J + 2 nz + cl = 0, (18)
where k
11
k
2
, k
3
are the roots of the disCJiminating cubic (Art. 73).
The condition that a. plane intersects this <ptadric in a. eire!<> is
that its conics of intersection with the given quadric and with a.
sphere coinr.itle.
The curve of iutersection of the qnru.lric (18) with the sphere
k(x
2
+ !l + z
2
) + Zl.t 2 liZ+ d = 0 (19)
coincides with the iute1scction of either of these surfaces with
the cone

'rhis cone is composite if the first m<>mber of its e(pllttion is
factorable, that is, if k is cq ual to k., k
2
, ot k
3

lt follows that each of t he six planes
vJ.:, - k, '= vk,- k:!J
vkl - k: r = k, z
J.:, !I= vk, - 1.:, z
intersects the quadric (18) in a. conic which lies on the sphere (19)
a.nd is consequently a. <:irclc.
ART. 82] CIRCULAR
If k
1
> k
2
> k
3
, the six planes are distinct. The planes
Vk1 - k2 X = Vkz - k3 Z
are real. The others are imaginary.
99
If k
1
=

k
3
, the last fom planes coincide with z = 0. The
other two are imaginary. If k
1
= k
2
=/= 0, the quadric (18) is a.
surface of re\olution (Art. 73). If k
1
= k
2
= 0, it is a parabolic
cylinder (Art. iJ).
If the equation of the surface is in the form (17), and k
1
, k
2
, k
3
are the roots of its disc1iminating cubic, it follows from the uis-
CIISsion in Article i3, that equations of the planes of the
circular sections through t he origin are
a.v
2
+ b?/ + cz
2
+ '2 .f!JZ + Z gzx + Z h.t!J - l."t ( x
2
+ .11
2
+ z
2
) = 0,
a.t
2
+ by
2
+ cz
2
+ 2f!Jz + Z gz.t + 2 h.c:!J- k
2
( x
2
+ y
2
+ z
2
) = 0,
a:c
2
+ by
2
+ cz
2
+ '2jyz + '2 gz.t + 3 h.t!J- k
3
(:t.:z + ?l + z
2
)= 0.
Proof II. It was shown (Art. -l!)) that a plane section of a
quad1ic is a circle if it passes through the cilcular points of its
plane. The conic in which the l}uadric meets the plane at infinity
has four points of intersection with the absolnte. Any plane
other than the plane at infinity which passes through two of
these points will meet the quadric in a conic through the circular
points of the plane; hellCe the section is a circl e.
The coordinates of the points of intersection may be found by
making the equations
a.1:
2
+ &.11
2
+ cz
2
+ '2 .fyz + 2 gz.t + 2 !txy = 0, x
2
+ y2 + z
2
= 0
simultaneous. Since both equations have real coefficients and the
second is satisfied by no real values of the variables, it follows
that the four points Pu P
2
, P
3
, P. consist of two pairs of conjugate
imaginary points, or of one pair counted twice.
In the first case, let P" P
2
be one pai1 of conjugate points, and
P
3
, P
4
the other. The lines P
3
P
4
a.re real (Art. 41), while the
lines P
1
P
3
,

P
1
P
4
, P
2
P
3
are imaginary. The pairs of lines

P
3
P
4
; P
1
P
3
, P
2
P
4
; P,P
4
, P
2
P
3
constitute composite conics
passing through all four of the points P
1
, P
2
, P
3
, P .
ln the second case, let P
2
= P
4
and P, = P
3
. The lines P
1
P
2
and
P
3
P
4
coincide, and the lines P
1
P
3
, P
2
P
4
are tangents to both curves,
which have double contact with each other at these points.
100 PROPEllTII!;:-:; 01<' (J UADIUC SURFACES [CHAP. VIII.
In eithf't' case the equations of the lines P,Pk can be found as
follows. Through t he points of intersection of (17) and the abso-
lute passes a system of conics
ax
2
+ b!/+ez
2
+'2 f'lz +'J uzx+ 'J hxy- k(x
2
+y
2
+z2) = 0, t=O. (19')
A conic of this syste111 will consist of two straight lines through
the four points of intersection if its equation is factorable, that is,
u -k h g 1
h b-k f =0;
g f c- k
thus k must be a root of the discriminating cubic (Att. 73). Let
k., k
2
, k
3
be the roots of this equation. The eqnatious of the pairs
of I i nes are t hen
rt.-cz+&yz+ t::z
2
+'2.f!Jz+3 hxy-k,(.c
2
+!l+z
2
)=0, t=O, (20)
with simila1 expressions for k
2
ancl k
3
l'rom Art. 4 t it follows
that fo1 one of the oots k, the two :teton; of the first member of
the quat!ratic equation (20) are real, but the fn,ctors for each of
the others :we illl :tginary when the roots k, are all distinct.
If u, v are the two linear factors of t hen the line 1l = 0,
t = 0 will pass through oue pair of points and v = O, t = 0 will pass
throngh t he other. A plane of the pencil u + pt = 0 will cut the
qualric in a circle. Since (1, plane is determined by a line and a
point not on the liue, the theorem follows.
ln case two roots of the di scriminating cubic are equal and
different from zero, the qnMric is one of revolution; the two
roni cs in t he plane at iuliuity now have tlouble contact.
lf k
1
> k
2
> k
3
, the planes uetermiuetl by the secont! root are
rea. I.
83. Real circles on types of quadrics. The above results will now
be applied to the consideration of the real planes of circular section
for t he standard forms of the equation of the quadric (Chap. VI).
(a) For the ellipsoid
the roots of the discriminating cubic are l ja
2
, 1/b
2
, ljc
2

ART. 83lk RE;.AL J;IRCLES ON TYPES OF QUADRICS
I< -If,< 3
101
Let a > b > c > 0. Since parallel sections of the surface are
similar, it follows that the equations of the real planes of circular
section are
c Va
2
- b
2
x a v'b
2
- c
2
z + d = O, (21)
where cl is a real parameter.
The circle in which a plane (21) intersects the ellipsoid is real
if the plane intersects the ellipsoirl in real points, that is, if it is
not more distant from the center than the tangent planes parallrl
to it. The cond.ition for this is (Arts. 'j(j ancl l6) I d I;;; ucv'ct
2
- c'
1

If I ell > ac v'a
2
- c
2
, the circles are imaginary.
If i<l l=acv'a
2
-c
2
, the circles are point circles. The four
planes determined by these two values of cl are the tangent planes
to the ell ipsoid at the points
Each of these points is rallccl an umbilic.
The two systems of planes (21) are also the real planes of ci rcu-
lar section of the imaginary cone
x_: +u:+===O
a2
0
2
0
2
and of the imaginary ellipsoid
(b) The equations of the real planes of circular section of the
hyperboloids of one and two sheets
and of the real cone
where a > b > 0, are found to be
cv a
2
- b
2
y b Vlt
2
+ c
2
z + cl = 0.
102 PROPERTIES 01'' QUADRIC SURFACES [CDAP. VIII.
On the hyperboloid of one sheet :tnt! the real cone, the muii of
cittlcs :trl' r;:al for all values of d. On the hyperboloid of two
sheets. thl' citclcs are renl onl y if I tl l >be v b
2
+ c
2
The coordi-
nates of the umbilics on t he hyperboloitl of two sheets arc
(c) The real planes of citcular section of the elliptic paraboloid
1/
- + '- = 11z a > b > 0 n > 0
ct2 IJ2 ' '
aud the real or imaginary elliptic cylinders
+ .1/ = 1 a > IJ > 0
lt2 IJ2 '
are uetermined by
-.J tc
2
- u
2
.11 + uz + d = 0.
On the real ell iptic ryl intler, the rircles are real, and on the
imaginary cylinder they are imaginar_y, for all values of rl. On
the elliptic parnboloid, the circles arc real if <l <;

(cc
2
- &
2
) .
The coi)rdinatcs of the umhilics on t lw elliptic paraboloid ate
[o, uu 'J<c2- f,2, ( a2- lJ2) J
(cl) For the hype rboli <: paraboloid
and the hypel'bol ic cylinder
,,.'.! ,,'1 ')
-., - It = 1 ..
w IJ
the equations of the planes of the circular sect ions are
IJ.t ay + clt = 0.
'fhe ci rcl es in these planes are all composite. For, the planes
uy + clt = 0
ART. REAL ClRCLES TYPES OF QUADRICS 103
intersect these surfaces in the fixe<l infinitely distant. line
+ Cl!f = 0, t = 0
and in a rectilinca1 generator whi ch varies with tl.
the planes
intersect them iu the line
b.1 - ((!I = 0, t = 0
aud in \'ariablc generator.
Also on the pnrnbolic cylj nder
.1'
2
= 2 inJJt
Similarly,
the real ci rcl es arc all composite, since lhe planes J: = clt intersect
the surface in the fixed line x = l = 0, aiHl in a var iable gcnerato
" 'e have, therefore, the following theorem:
Tm:ORE)r II. 011 the lt!tJiPrf,IJ[ic 1)(1ml,olnid, the h!Jperbolic
cyliltder, aml the pwa!Jolic C!Jl iwler, lhP t elll ci rcu[(() sections we
composite. The com}lultent:-; ertdt circle we an inJinitely distctnt
line ancl a rectiline((r generator tddch iutasects it.
EXERCISES
1. Find the equations of the
1
circ\thu,

of t he surface
! x'+ 2y'! + + :3 yz + :rz = l. 'l.<l -r"
0
2. Find the equati l n:; of the r eal circnlar sections of the surface
2 .rY+ 5 y2 + a z2 + .j .ry = 1.
r 3 . FiiHI the radius of a circular section through the o rigin in Ex. 2.
'?:' Find the equatims of the plant's through ( I. - 3, 2) which in-
lci'S'ect the ell ipsoid 2 + + -1 = 1 in circles.
Find the conditions which tonst br satisfitld by the coetlici ents of the
ettaati()n F(.r. y, z) = 0 of a quadric if the planes z = k inlJ' J'Set it in circles.
6l !"how that the centers of the circles in Ex . . ; lie on a line. :Find the
C<Ju1\tions of this line.
,/
/ 7. Find the second system of r eal planes culling ci1cles from qualric
t m Ex. 5.
8. Find the conditions which be satisfied ly the coeflkients if the
plant> A.r. + By+ Oz + D = 0 intersects the: qnaclric F(r. y. z) = 0 in ciJcles.
9 . Find thr coiirdinat<'S of the cemer anrl the raclins of the circle in
which tlw plaue .r = 2 z + 5 intersects cone 3 ;r'! + 2 y2- 2 z2 = 0.
.a" _b... -f'" f
..
01<' QUADRIC [CHAP. Vlll.
I
{lo. Show that, for all vahtcs of X, the equation of the planes of the ci r-
cuhfr sections of the quadrics
(a+ X)x2 + (b + X)y2 + (c + X)zZ = 1
are the same. The quadrics of this system arc said to be concyclic.
84: Confocal quadrics. The system of surfaces represented by
the eqnation '
x2 112 z2
- -+-"-+-=1, a>b> c>O, (22)
n2 + k b
2
+ k & + k
in which k is a parameter, is called a system of confocal quadrics.
The sections of t he quadrics of t he system by the princi paJ planes
x = 0, y = 0, z = 0 arc confocal conics.
If k > - c
2
, the surface (22) is an ellipsoid; if - c
2
> k >- b
2
,
the surface is an hyperboloid of one sheet; if - b
2
> k >- tt
2
, the
sul'face is an hyperboloid of two sheets; if - a
2
> k, the surface
is an imaginary ellipsoid.
If k >- c
2
, but approaches - c
2
as a limit, t he minor axis of
the ellipsoid approaches zero as a limit, and the ell ipsoid ap-
proaches as a limit the part of the Xl"'-plane within the ell ipse
(23)
If - c
2
>k > - u
2
, the surface is an hyperboloid of one sheet.
As k approaches - c
2
, the surface approaches the par t of t he
XY-plane cxterio1 to the ellipse (2:J). Ask approaches - b
2
, the
smfacc approaches tha.t part of the XZ-plane which contains the
origin and is bounded by the hyper bola
(24)
If - b
2
> k >- a
2
, the surface is an hyperboloid of two sheets.
As k apptoaches - b
2
, the hypcrboloitl approaches that part of the
plane !I= 0 which does not contain the origin. Ask approaches
- a
2
, the real part of t he surfa.ce approaches the plane x = 0,
counted twice.
The ell ipse in the Xl':.plnne :wcl the hyperbola (2-l) iu the
XZ-planc arc called the focal conics of the system (22).
ARTS. 84, 85J
ELLIP'l'lC COUflDIXA1'ES 105
The vertices of the focal ellipse are
( va
2
- c', 0, 0).
The foci are
( Va
2
...::.. b\ O, 0).
On t he fo<al hyper bola the vertices :uc ( va
2
- b
2
, 0, 0) and the
foci arc ( V<t
2
- 1.P, 0, 0). Hence, on the focal conics, t he ver-
tices of each arc the foci of the other.
8.3. Confocal quadrics through a poi nt. Elliptic coordinates.
[. Three conforal ]>(ISS throuyh ecery poiut
P in !IJXIce. rl!ftl, onP of tltPse fJIICUldCii i!l WI one wt
hyperboloid of Otll' s!tPet, anclthe thircl cw hyperboloicl of tu;o
If P ::(.1
11
!h, z
1
) lies on a quadric of the system (22), the par:un-
etet k satisfies the equation
, (k + a
2
)(k + b
2
)(k + c
2
) - x
1
2
(k + b
2
)(k + c
2
)- ?h
2
(k + c
2
)(k + a
2
)
- z
1
2(k + a
2
)(k + b
2
)= 0. = l (
Since thi:; is a cubic equation in k, :utll each of its roots determines .
a quachic of the system thl'ongh P, there arc three quadrics of ._ "'.1'
the system which pass throngh P.
Let P be real. :. }
If k;;, first member of (2:i) becomes positive. .f( .,ofi)l
I f k =- c
2
, it is - z
1
2
(-c
2
+ a
2
)(- 1;
2
+ b
2
), which is ve. = 1 (
If k =- b
2
, it is - .111
2
( -lr + c
2
)(- b
2
+ a
2
), which is positive. + (- b".
If k =- ct
2
, it is - .1:
1
2
(- ct
2
+ b
2
)(- a
2
+ c
2
), which is negative. " + ( -"
Hence the roots of (25) are real. One is greater than - c
2
, one
lies between - c
2
and - b
2
, and the third between - b
2
and - a
2

Denote these roots by ku 1.:
2
, 7.:
3
Hence, we have
1.:
1
>- c
2
> k
2
>- b
2
> k
3
> - a
2

Then, of the three quadrics
X! ll zt
--+ -- + = 1,
u2 + k IJ2 + "' c' + kl
+ ___t_ + _ z_ 2 - = 1,
at + k2 fJ2 + k, c2 + k,
(26)
':1-a yt z2
--+- -+- - = 1
at + /,:3 bz + k3 c2 + 7.:3
"'f:)

(l'UAP. Yfll.
106
PROPlm'riES OF QUADR1C' SURFACES

which pass t hrough P, the first is an ellipsoid, t he second an hy-
1
berboloid of one sheet, and the third an hyperboloid of two sheets.
II. The th?'ee quadl'ics of (( C011focal system which n
... .
J)(tss though P intersect ecu:h othe1 at ?'ight a11gles.
l<'or, the equations of the t.1.ngent planes to the first two quad-
rics (26) are ' . ' ,
., X1X + _}jJ]J__ + Z1Z _
1

\l

+

+ r2 =

v
. .,? Y: r1 E
\} r These planes are at right angles if J
\ \, t. 1&- .
,! xl2 +- _ _ yl + 0
.
That this condition is satis- ;
1
.
fied is secu hy snhstituting
eoorclinatrs of P in (26), subj ;..o:
tracting the senoud N]natio;,t
frou1 thcfi r:st, ancl removi ng th&\
f:utot k
2
- k
1
, which was !'('en
to be d ilfcrcn t from zero. The
proof for the other pairs may
be obtained in the same way.
'I' he th r('e roots kll k
2
, k
3
of
eqnation (2il) are call ed the el-
liptic coordinates of the point P.
'l'o fi ntl the expressions for the
rcctangnlrtr coortlinat<.'s of P .i n
terms of the elliptin cool'llinatcs,
\\' C suhstitntt1 the coordiuates
(.1\. ?It, z
1
) of J' in allll sohc
fol' X
1
2
, !It zt The tesnlt is
(27)
ARTS. 85, 86)
QUADRICS TA).;GE}(T TO A 107
It is seen at once from these equations that /,., k
2
, a ncl 1.:
3
a re the
ell iptical coordinates, not only of P, but a.lso of t he points sym
metric with 1' as to the cotinlina.te planes, o.xes, and origin.
8H. Confocal quadrics tangent to a line.
TnEORK\I I . .Any fine louchr>!> I teo 'J.IWII'ics of n confocal
T he points of intersection of a given line with n qun,dric of t he
system determined by the equa.tion (Art. 6:))
t..K.t.i .l f" v ..

+ _______ + _v_
2
-) 1'2 + 2 (2L + !/of.A. + 1' f 7 .If
&+k
+ !lo' +_:l_- 1) = 0.
lt
2
+ k {} + k G
2
+ k
The condition that this line is tangent .... """ n)...o1r,t:)i
(
XoA + ?/oft

u Q...;.,..
a
2
+ k b
2
+ k c
2
+ k
- + _L + _v2_)(_5_ + ?/o: + ___!L_ -1) = 0.
&+k
" rhen expanded n,nd simplifictl, this equation reduces to
k2+ [(b2 + c2),\2 +(
1
.: + a=)tz +(112 + /,2) v2 -(Xop. - y
0
>..)2
- (!tov- ZoiJ.)z -(z
0
>. _ ,,.
01
,)2] /; + [&21;2,\2 + ,.2CL2p.2 + ,
1
:fJzvz
- (.l'of.A.- !lo,\)tz -(!fov- ZofJ.)IJz- (zo.\- .l1Jv)C12] = 0.
Since this equation IJttadr:ttic in k, the theorc111 follows.
II. l..f two t't)lt./OI'(([ 'JtWtlrics a lilte
1
lhe taugenl
( 'h at thr poiut8 toutacl wr> 111 ri!Jhl
, ; Let k
1
and he the p:uanwters of the quadrics, and l et
P' = (x', ?t', z'), P" =. !f
11
, z") be the points of tangeucy of the
with the given qnn,drics. The equations of the tangent
a.t P' am\ P" are (.\ rt. 7G), rrspecti vely,
IJ -:. :r'x 11't z'z :l'
11
.J: t"'l z"z
'It. - - +- - -+-- =1, --+---+--=1.

j y These planes are n.t right angles, if
x'.l"' u'/' z'z"
---- +-- --+ = 0 (28)
f (a'+ k,)(<' + !;) W + !,)(// + k,) (c' + k,)(c' + k,)
., OF QUADRIC SUR>'AC"S !C"" VIII.
i J "'Since the line through P' and P
11
is tangent to both quadrics, it
lies in the tangent planes at both points. Hence P and P" lie in
both planes, so that
x'x" + y'y" + = 1, x'.l:lf + ?!'J!'' + ..i..!!.!._ = 1. ,,
a
2
+k
1
b
2
+k
1
c
2
+k
1
ct
2
+k
2
b
2
+k
2
c
2
+k2
By subtracting one of these equations from the other, it is seen _., '
that (28) is sa.tisHed. The planes arc therefore at right angles. "t
87. Confocal quadrics in plane coordinates. The eqnation of tJ ..
system (22) in homogeneous plane coordinates (Art. 77) is /),-,..
4
,_;:
etznz + b2c2 + c2w2- sz + k(u2 ryz + w2) = 0. 1'f rl

Since this ec1uation is of the jlrst degree n k, we have the follow- ;"
ing l- +
1
IT'I u: t.W. J. __
e-u.t.
TnEomm. An nrbitretry ]>lane v
1
,

s
1
) is t<W[tenl to one
and only one l'onji,ccd sy:;lenL.;_ -
Jt :oO'IW'f"- (f--
The (imaginary) '\.Jl:\iie's whose homogeneous cootdiHates satisfy
the two cquat.ions (lvv o)

are exceptional. They touch all the quadrics of the system.
Hence, all the quadrics of a confocal system touch all the planes
common to the quadric k = 0 and the absolute.
EXERCISES
-:f l . Prove that the difference of the squares of the perpendicular from the
center on two parallel tangent planes to two ghen confocal quadrics is con-
stant. This ma.y be used as a definition of confocal quadrics.
V2. Prove that the locus of the point of intersection of three mutually per-
pendicular planes, each of which touches one of three given confocal quadrics,
is a sphere.
3. Write the equation of a quadric of the system (22) in ell iptic coordi-
nates. Derhe from (27) a set of parametric cr1uatious or this quadric, using
elliptic coordinates as p;uamuters.
1f 4. Discuss the system of confocal paraboloids A.- \ y(>

---- + ___= 2 nz + kn
2

k b'! + k
5. Discuss the confocal cones
+ ____jf:_ + = 0.
a2 + k bt + k c2 + k
s
,
l
CHAPTER IX
TETRAHEDRAL COORDINATES
88. Defi nition of tetrahedral coordinat es. It was pointed ont in
Art. 34 that the four planes x = 0, ?f = 0, z = 0, and t = 0, which
do not all pass through a point, may be consitlerecl as formi ng a
tetrahedron which was called the coordinate tetrahedron. We
shall now show that a system of coordinates may be set up in
which the tetrahedron determined by any four given non-goncur-
planes is the coordinate tetrahedron. A system of coordi-
nates so determined wi ll be called a system of tetrahedral
coonlinates.
Let the eqn:ttions of the four given non-concnrreut planes (re-
ferred to a given system of homogeneous coordin:ttes) be
.A,x + B,y + O;z + D.t = 0, i = 1, 2, 3, 4. (1)
Since these planes do not all pass through a point, the determinant
A, B, 0, D,
GUA r-0
T = !:
(} '(o, D I.
does not vanish.
Let the coonlinates (.t, y, z, t) of any point P in space be sub-
stituted in the first members of (1) and denote the values of the
resulting expressions by

respectively, so that
x, = A,:c + B
1
y + 0
1
z + D,t,
x2 = A2x + R2.'! + 02% + DJ,
Xa = .A
3
.l: + B
3
'! + O:JZ + DJ,,
:t
4
= A4x + B.y + 0
4
z + D
4
t.
(3)
We shall call the four numbers .l'u x
2
, x
4
determined by these
equations the t etrahedral coordinates of P. The four planes (1) are
called the coordinate planes. Their equations in tetrahedral coor-
dinates arc x
1
= 0, = 0. Xa = 0, :mel :r
4
= 0, respectively.
1::>1
110 'J'I<..:TR.\liJ<:l)flAL COOHDJ :\.\ [CIIAI'. LX.
Since the four planl's (1) do not all pass through a point, the
cotirdinatL::; .1
17
.r
2
, .r
3
, tannot all he zero for <tny point in space.
\\'hen y, z. t) arc givln, the values of ... ., .r
3
, .r, are uniquely
cleterm incd by l'ou vcn;ely, si nee t he determinant does
not vanish, equations can be sohcd fo .r, y, z, l so that, when
x
1
, .'t
2
, .1
3
, .1, arc gi ' 'en, one and only one set of values of y, z, t
ran be found. Si uce (x, !/, z, t) a.nd (k.,, ky, kz, kt) represent the
same point (Art. it follows from (3) that (.c., x
2
, X-J, .t
4
) and
(kx
1
, k.1:,) tcprescnt the s:une point, k being an arbitrary
constant, different from zero.
8H. Uni t point. A system of tetrahedral coordinates is not
completely determined when the positions of its coordinate planes
are known. For, since the equations
and
k(Ax+ B!! + Cz+ Dt) = 0,
A.t+BY+ Cz+Dt=O
k=FO,
represent t he same plane (Art. it follows that if k
1
, k
3
, k,
are four arbitrary constants differeJlt from zero, t.he equations
x', = k;(.A,x + B,y + C,z + D,t), 1:= 1, 2, 3, 4 (4)
define a system of tetraheclral coonl inates having the same coordi-
nate pla.nrs as (:'i) but sucl1 that
x', = J.,:r,, i= 1, 2, 0, 4.
The point whose tetrahedral coordi nates with respect to a gi,en
system ::ue all equal, that x
1
: .l'
2
: :t
3
: = 1: 1: 1: 1, is called the
unit point of the system.
Tm:oim't 1. Ail!/ point P, not. l!Jin') on a face o.f the coordinate
tetrahP.(lron, 11Hl!J be ltrken as nnit point.
For, by substituting the coii rdinates (x, y, z, t) of P in (4)
values of 1.'
1
, k
3
, k. may be found such that X
1
' = x
2
'= .'t
3
' = x/,
so that J> is the unit point.
Since the ratios k
1
: k
2
: k
3
: 1.:
4
are fixed when the unit point has
been chosen, we have the following theorem:
H. 'The of lPtralll' clral COo)'(linraes is dete1
m itll!rl whf'll t/11' coiil'tlinu/f' pllnlf's

= 0.

= 0, x
3
= 0, x, = 0 a11d
tltf' unit JIOilll ( I , I , 1, I ) lta11' 111'1'11 rhosl'n.
ARTS. 8!), 90] OF A Ill
EXERCISES
In the following problems, the equations in homogeneous coordinates of
the coordinate planes of the given system of tetrahedral coordinates are
:r - !1.+ 2t = 0,
3 X + 3 y + 2 Z + :.! t = 0,
:r + 2 y - 2 z + t = 0,
X - 3 y + Z + 2 t = 0.
The homogeneous coordinates of the unit point are (- 1, 2, - I, I).
1 . Find the tetrahedral corrdinates of the points whose homogeneons
rectangular coordinates are (.c, y; z, t), (0, 0, 0, 1), ( I, 1, 1, 1), (5, 1, -2, 1),
(:1, I, 1, 0), (0, 1, - 1, 0).
2 . Find the coordinates of the points whose tetrahedral
coordinates are (- 1, 1, -1, (1, :.!, - 1,- 5), (0, 0, I, 3), (.;,
1
, ,
2
, x
3
,
3. n rite the ecJwllion of the surface ,
1
+ 2 :r
2
- 2 :r. = 0 in rec-
tangular cotirdinn.tes. !'how that the locus is a plane.
4. "rite the ec1nation of the plan<' ;, .r + y +::- t = 0 in tetrahedral
coordinates.
5. Wriw the of thr smfact

+ = 0 in rectangular
' )-..
t " 6. 1 and :! poiut whose rectangnln cot;r,Jin:lles are
(.,, 1. - 2, 2) 1s takeu as nmt pomt.
f
7. \\'hy may not l!..Jluint lying in a face of the tetrahcclron of reference
be taken as nmt poiu?'!JI! .,...
.
90. Equation of a plane. Plane coordinates. From the equati on
m; + ty + lt:Z +sf= 0 (fi)
}'
"
l
of a plane in homogeneous rectang1tl:u coorclinatcs, the corre-
spomling t'quatious ill Let1ahedral coordinales <an be found by
solving equations (:l) for .t, y, z, l and substituti ng in (<3). The
),:'4:- tesulting equation is linear and homogeneous in .t., .t:i! .t
3
, of
:J. the form
(G)
'ft,. with constant coefficients 11., u
2
, tt
3
, 11,. Conversely, any eqnation
of the form (G) defines a plane. Fot, if x., x
2
,

x, arc replaced
by their values from (3), the resulting equation is

....
"
112
wherein
TE'rR.AHEDRAL COORDINATES
11. = A
1
tt
1
+ A
2
u
2
+ A
3
u
3
+ A
4
u
11
v = B
1
11
1
+

+ B
3
tt
3
+ B
4
n
4
,
w = 0
1
u
1
+ G
2
u
2
+ 0
3
u
3
+ 0
4
n
4
,
s = D
1
rt
1
+ D
2
u
2
+ D
3
u
3
+ D
4
tt
4

CBAP. IX.
(7)
The coefficients u
1
, 11
2
, u
3
, 11
4
in (6) n.re called the tetrahedral
coordinates of the plane (compare Arts. 27 and 29). It follows
from equations (7) and (2) that, if u
11
11
2
, 11
3
, u. (not all zero) are
given, the plane is definitely determined, and that, if the plane is
given, its tetrahedral coordinates (u
1
, 11
2
, 11
3
, tt
4
) are fixed except for
an arbitrary multiplier, different from zero.
91. Equation of a point. Let (x
11
x
2
, x
3
, x
4
) be the co<:itdinates
of a given point. The condition that a plane whose coordinates
are (11
11
11
21
u
3
, t1
4
) passes through the given point is, from (6)
(8)
This equation, which is satisfied only by the coordinates of the
planes which pass through the given point, is called the equation
of the point (x
11
x
2
, Xs, x,) in plane coiinlinates ( cf. Art. 28).
It shoultl be noticed that., in the equation (6) of a plane,
(11
11
11
2
, 11
3
, u
4
) are coust:wts antl (x
1
, x
3
, x
4
) are variables. In
the eqnat.ion (8) of a point (xu x
2
, x
3
, x
4
) are constants and
( tt
1
, 11
2
, 11
3
, 11
4
) are variables.
92. Equations of a line. The Joens of the points whose coordi-
nates satisfy two silll nl tanrous linear equations
tt\x
1
+ u'
2
x
2
+ u'
3
.l:
3
+ u',,x
4
= 0,
tt
11
1
x
1
+ tt"
2
x
2
+ n"
3
Xs + lt
11
4
:t'
4
= 0
(9)
is a line (Art. 17). The two simultaneous equations are called
the equations of the line in point coordinates.
Similarly, the locus of the planes whose coordinates satisfy
two simultaneous linear equations
x'
1
u
1
+ x'
2
u
2
+ X
1
3
11
3
+ X
1
4
?1
4
= 0,
x"
1
u
1
+

+ x'
1
3
?t
3
+ x"
4
tt
4
= 0
(10)
is a line (Art. 28). These two simul taneous equations are called
the equations of the line in plane coordinates.
ARTS. 91-93] DUALITY 113
EXERCISES
1. Write the equations and the coordinates of the vertices and of the
faces of the coordinate tetrahedron.
2. Write the equations in point and in plane cMrdinates of the edges of
the co<lrdi nate tetrahedron.
3 . Find the equations of the following points: (1, 1, 1, 1), (3, - 5, 7, - 1),
( - 1, 6, - 4, 2), (i, 2, -1, 6). , ,.
4 . Write the coorrl inates of the following planes:
x, + :r2 + X3 + = 0, i :r, - :1:2 - 3 :ra 4- = O, Zt + 0 .r2 - 5 X3 - 2 = 0.
5. Write the equations of the line Zt + x: = 0, x
3
- 7 x. = 0 in plane
coordinates.
Write the equations of two points on the line.
6 . Find the coordinates of the point of intersection of the planes (I, 2, 7,
3), (1, 3, (J, 0), ( 1, 4, 5, 2).
03. Duality. We have seen that any four numbers xl> X:J, x
4
,
not all zero, are the coordinates of a. point aud that any four num-
bers tt
2
, u
3
, 11
4
, not. all zero, are the cootdinates of a plane.
The condition that the point

x
3
, x.
1
) lies in the plane
( uu tt
2
, tt
3
, u
4
), or that the plane (tlu 11
2
, u
3
, u
4
) passes through the
point (xu x
2
, x
3
, :1.'
4
) is
tt1:t1 + tl2-'l:: + ti3X3 + U4X
4
= 0.
This equation remains unchanged if x
1
, Xa
1
:r
4
and 7t
1
, tt,, u
3
, tt
4
are interchanged.
The equations (9) anu (10) of a line are simply interchanged if
point and plane coordinates :ue interchanged.
From the abo\e observations, the following important pr inciple,
called tbe principle of duality, may be deduced; namefy, that if
we interchange x
1
, X:J, x
4
and u
11
tt
2
, u
4
in the proof of a
theorem concerning the incitlt>nce of points, lines, and planes, or
concerning point and plane coordinates, we obtain at once the
proof of a second theorem. 'l'he theorem so derived is called the
dual of the first. It is obtained from the given one by inter-
changing the words point and plane in the statement.
In the next two Articles we shall write side by side for com-
parison the proofs of several theorems and their duals.
The symbols (x), (x'), (u), etc., will be used a.s abbreviations for
(x
1
, x2, xa, x
4
), (x'
1
, x'
2
, x'
3
, x'
4
), (u
1
, tt
2
, tta, u
4
) etc., respectively.
114
TETRAHI':DRAL ('OORDINATF.S \CDAP. IX.
!l-l. Parametric equations of a plane and of a point.
Let (x'), (x"), be t hree
gi vcn llOll COlli ueaL' poin ts. The
equati on of t he plane detet-
minccl by them is found, by
the same mcthotl as that em-
]'loyed in AL't. 11, t o be
X t
!);2 x,
x'
x'2

x'
I
3
4
=0. (11)
t "

:1'" :t''
1 2 3
I
] ' Il l
I

2 x"'
3
, .. ,
.
Let (.!') be any point in t he
pl auc (ll ). F t um t he form
(11) of t.h c CfJ.n;ttioll of l he
plano it foll oll's tlt:t l there
exi st fum Jllllll UCrs J>, lu l2, !3,
not all zc ro, such that
Jl.l' ; = l
1
.t;
1
; + l
2
x", + l3.-c'".,
i = 1, ::, 4. (13)
I n part.icnlar, we have ]' =1= 0,
since otherwise it would foll ow
t hat (.t '
1
), and (.t"') a re
collinear ( Art. fl.i), whieh is
cont 1at'.)' lty poLhcsis. Con-
\'Ptsely, c,rry point (.c) whose
l'Wirtlinatcs nre cxpr!:'ssil>le iu
t.hc for m (1::), 7' =F 0 l ies itt
t hr pl an<' (11) ,; ince i ts roii t-
cli natcs satisfy the equat ion of
the phwc.
El}n:ttions (1::) arc call ed t he
para metri c ec1uations of t he
plane ( L 1), a nd lu 1
2
, 1
3
are C\alled
t he homogeneous parameters of
t he points of t he plane.
Let (n'), (u"), (tt"') be three
gi,cn non-collinear planes. The
equat ion of the poi nt deter-
Ill ined by them is found, by
the same method as that em-
ployed i u At't. 11, t o be
Ill u2 u4
u',
u'2 u'3
=0. (12)
?t"t
'lt , 2 11"
3
tt"4
I
N'"z 1t'"a ?1"' ..
Let (u) be :llly plane through
t he poi nt (1 From t he form
( I:.:') of the eq twt ion of the
point it foll o\\'s t.hat thete
exist font nuntbers p, /
11
l
2
, l
3
,
not all zero, sn( lt t lt at
}) II ; = falt
1
; + 1211
11
; + /3tt
111
;
1
i = 1, 3, 4. (14)
In part icul a r, we have Jl * 0
si nee othErwise it wouJ d follow
t hat (u"), and (tt"') are
coll ineat ( Art . 95), wltich is
cont ra ry to hy pothesis. Con-
' 'ersely, every pla,ne ( u) whose
cniir.liuatcs at'C expressible in
t he furm ( 14), 71 =1= 0 passes
t l he point (1 its
roiircl ina,tes satisfy the eq nation
of tlae point.
F[nations (14) are called t he
par:unct.ri c eC'juati ons of the
point (12), and l ., l
2
, l
3
arc call ed
t he paramet ers of
the planes throug h the point.
Anr,.;. HI, n:,J EQUATIONS 0[,' A 115
Tlu;: systPm of poiut;;
is saiti to form a plane field.
The equation of the points of
tuis plane fit'lcl is found uy sub-
stituting the \'alnes of x
11
'':!
.1:
3
, .1:. from (l:'i) in the cr1ua.tion
:::!.u,:c
1
= 0 of a. point. The re-
sulting cCJnation
is the CtJnn.tion, in plane co-
ordinates, uf the plane field
(
The system of plaues ( 1-t) is,
saitl t.o form a bundle of planes.
T he equation of the planes of
the bundl e is fonutl by sub-
stituting the \'al ues of u
11
u
2
,
11
3
, 11
4
from (14) in the equation
::i u
1
.1'
1
= 0 of a plane. T he re-
suiting equation
is the equatiou, in point coordi-
uales, of the bundle of planes
(U).
!).3. Parametric equations of a line. Range of points. Pencil
of planes.
If tx) is any
]JOillt 011 /hi' lilf' tli/NIItillel{
b,11 t11:0 (Jil'l'll ,7i,,/(Jw( JIIJi/1/S (/)
rwtl (.t"). 1,,.,,,.,, tle/l'm iuw!l
ottler tl1re in llu nut/d.
r .1:!
,1'3 .(',
I' :c'2
.l.'l3 J.l
( l.i)
1
x"
1
t"
0 2
:t:"
3
.(".J
i11 eqwd to zi'I'O.
For, the points (x), (.c'), (.<"
1
)
and auy fourth point (.c'") are
coplanar. Their coorclinates
consequently satisfy (11). Since
(Ll) is sati,;fietl for all v:tlues
of x"'
11
x"'
1
, .r'"
3
, .r"',, it follows
that the coefllcient of each of
these variables is equal to zero,
that is, that a 11 the detenn i-
nants'of order three in (l:'i) are
equal to zero.
J.f (n) is any
Jlwl' lhrou[!h tltl' liue tlPlenninecl
:n, /wf) uiPn Jll<wes (t
1
)
I (Ill/ ( 11
11
), (! 1'1'/'lf llell'nniurwl
tJrl/er llur''' iu lhP mall'iX
I,,
({:! l<a 11.
I
tt' 2
I
u', (16)
,, 1
(( 3
Ia",
II
?t"3 1". , 2
l!lflltLf liJ Zf!J'O.
For, the planes (a), ( u'), ( u")
and any fourth plane (11"') are
concurrent. T heir coordinates
conseql;ently satisfy Since
(12) is s::ttisfied for all values
of 11
111
" u"'
2
, n"'
3
, u"'
4
, it follows
that the coefficient of each of
these variables is equal to zero,
that is, that all the detenni-
nants of order three in (16) are
E' 1na l to zero.
116 TETRAHEDRAL COORDINATES !CHAP. IX.
Conversely, if the detenui-
nants of order three in (W) are
all equal to zpro, the points (a:),
(.c'), null (x'') are coll inear,
since they ::ue coplanar with
any fourth point (x"') what-
e,er.
It follows f1om the above
theorem that there exist three
numbers p, l
11
1
2
, not all zero,
such that
In particular, we have Jl * 0,
since otherwise the coonlinates
of the points (x') and (a:")
would be proportional so that
the points would coincide.
Equations (17) are called the
parametric eq nations of the line
determined by (x') and (:c").
The coefficients 1
1
aud 1
2
are
calle<l the lJOnlOgcneous pnram-
etets of the points on the
line.
'l'!Je system of points (17)
is said to form a range of
points. 'l'he equation of
the points of this range is
found, by substituting from
(17) in the equation
Iu
1
x, =0
of a point, to be
Conversely, if the determi-
nants of order three in (lti) are
all equal to zem, the planes (n),
( u'), and ( u") are colli near,
since they have a point in com-
mon with any fourth plaue (1/")
whatever.
It follows hom the above
theorem that ther e exist three
nu m hers J!, 1
11
l
2
, not all ze1o,
such that
J111
1
=l
1
u',+l
2
u",, i = 1,2,3,4. (18)
ln particular, we have 1) =1= 0,
since othmwise the coordi nates
of the planrs ( u') and ( u")
would be propottioual so that
the planes would coincide.
Eqnations (18) axe called the
parametric equations of the line
lleterm i ned by ( 1/) and (1/').
The coeflicients l
1
ancl 1
2
are
called the homogeneous param-
eters of the p lanes t brongh
the line.
The system of planes (18)
is said to form a pencil of
planes ( A1t. 2-l ). 'r he eq nation
of the planes of this pencil is
found, by subst ituting from
(18) in the equation
of a plane, to be
\
'
ARTS. 95, 96)
'r RAKS!-'0 R.:\1.\ T ION 117
EXERCISES
1 . Prove the following theorems analytically. State and prove their duals.
(a) A line and a point not on it determine a plane.
(b) If a line bas two poims in common with a plane, it lies in the plane.
(c) If two lines have a point in common, they determine a plane.
(cl) II three planes have two points in common, they determine a line.
2. Write the parametric equations of the plane determined l.Jy tbe points
(1, i, - 1, 3), (2, 6, 4, 1), ( 10,- I,- 3,- S) . Fi111l the coordinates or this
plane.
3 . Write the parametric equations of the point determi ned l!y the planes
(- 6, 3, 4, I), (7, - 6, ;1, 2), (H, - 4, - 3, I). Find the coordinates of
this point.
4 . Write the eqnation, in plane coordinates, or the field or points in the
plane Xt + X3- .c4 = 0.
Suo. First find the coordinates of three points in the plane.
5. Find the parametric ectuations of the pencil of planes which pass through
tbe t wo points Ut - ;) 112 + ::lu3- 11 = O, 7 111 +:? uz- U3 - u. = 0.
6. Prove that the points (1, 2, - 3, - 1), (3, - 2, 5, - 2), ( 1, - 6, 11, 0)
are collinear. Find the parametric equations of the line df:'tennined by these
points and the equation in plane coordinates of thu range of poims ou this line.
06. Transformation of point coordinates. Let (.t
11
x
3
, x
4
) be
the coordinates of a point referred to a given system of tetra-
hedral coordinates, so that
i = 1, 2, 3, 4,
in which the determinant of the coefficients*
A= I C!u a33 a . d * 0.
(19)
Let the coordinates of the same poiut, refer red to a second system
of tetrahedral coordinates, be
in which
i = 1, 2, 3, 4, (20)
The symbol I on 02;2 083 a.,.j will be used for bre\ity to denote
an o1a u
t he determinant a21 12'J (/23 Uz..,
OgJ lls-J sa (/84
04) a.s a.,.
118 TETRAHEDRAL CO(.)JWINA'J'BS [CHAP. I X.
lt is rf'r tnircd to the Pquation toHHcctiug Lhe two sets
of coordinates (.t., x
3
, .r
4
) and ( .1'
1
, x'
2
,

x'
4
) . For this purpose
:;ulve equati ons CW) for :c, !J, z, t. The results are
in whi ch A',k is tho cofact or of a',k in the dctermiuaut A'. Snh-
sLiLntf' these values of .t, !J, z, l iu (1!)) aml impl ify. The result
is of the form
wherein
X
1
= 1tu.t'
1
+ lt
1
zX' z + lt13/ 3 + 1Ct4.r'll
J'z = ''1t1:' + 1cz2x'z + lt23.r'3 + IC:wr'4,
X3 = + a 3,.r'z + + ltJ41;'4,
x
4
= 1c
41
.t
1
1
+ cc
42
.t '
2
+ a
43
x '
3
+ lt
44
X
1
4
,
(21)
A'tt,k = ct
11
A'u + <1
12
. l'k
2
+ a,
3
A'k
3
+ 11;
4
.tl'
44
, i, 1.: = 1, 2, 3, 4. (22)
'l'he determinant
I ((11
((12 ((13 ((14

((:!".! 1(23 ((!4
1(32 ((33 1(34
((41 ((42 a.3
((Jl
is called the determi nant of tlt e transformn.tion (21). This deter-
minant is differf' nt from zero, for if we substitute in it the values
of the 1t,k from we have at onre * t
The of two cJct etnliuant s of Imle r ftlltr
.A = I u
11
a :l'! 11
33
<
44
1 ami n =: I bu b:l' ! u
33
b,,l
is also a thte rminant of or<l er fout
in which
(.';::I c
11
r :l' ! C:J3 t.
1
.d,
<'ik = a ; t bk
1
+ a ;
2
bt-.! + ct;sl'tJ + a;.
1
/, k
1
, i, k = t, 2, 3, 4.
This Lhcutctn can eas i ly IJe \'Crifie<l by these Y:dues of l ' i k in a ami ex-
prussiu;.: r as the sum of tl etermiuants, of l!:t<'h hein;.: the JH'Odnct
of au element of A aucJ au uf JJ. Of the dc tl:rmin:\nts in the
s tun, forty ,:wish icl entic;dly, all tlte cle me nts of unc tolumn proportional
tn the clements of IUtotlwr. of t ht re mainint: has
B a:; a fa cto r. Whe u t he fat to t B is tctnovecJ, the res nlting expressiou is the
expansion of the <let erminant A .
t The cletcrmiuanl I A'lt . f'aa A'
44
I whose cl emeuts are the cofactors
u f the ele ment s of : 1' is t:IJnal l< . f' 3, as is Stcu inuuetliatcly by multiplyiug it l>y
A' IJy the prcccdlug rule, aml s implifying the res ult.
.-\RTS. !)(), !)j) TR.\XHFOIDL\ TIOX 119
Since' T=F 0, till' (?1) can IJC' sohcd for .r'., x'
3
, :r!
4
in
terms of x
1
, J.'
2
, .1
3
, ., The results an
7:r', = + + f33\"l'3 + f3n
1
'JJ
T.t'z = {3,2-rt + f3'n"r2 + + f3,!xu
'l'.1'3 = {3,3.1, + f3!a.l't + + f3,3xH
= {3, p, + + f33tra + p.,.r"
in which {3 is t hi' cof:wtor of '' iu the tlc!<'l'lllinant T.
(23)
The tr:tn!:>formations (!?1) ;tnd arP sa.ttl to be in\crsc to each
other.
97. Transformation of plane coordinates. l.Pt
(2 1)
b<' the l'l[tHltion of :t 1'1.!11 planr, rc>fetTI'd to t lte s_,stem of tct rn-
hcdral coordinates ch,termimcl by ( I !l). Let, the tlctnation of tilt'
same plnm, rdcncd to the (:!0), ht
11'
1
.'
1
+ 11
1
'!'l'
1
2
+ 11'
3
.1'
3
+ ''
1
..-'
1
= 0.
If, in (:.? 1), we: x
1
.:t''! .
3
, .1'
1
hy tlwit ,alues from (:! I ), we
ohtain. aftt'r the tPtms.
(tt
11
11
1
+ ":1llz r I(JIII3 +I( IIIII) 1'
1
1
+ ( 1(,/lt + 1(,.11. +I( 113 +

z
+(1(13
11
1 + tt.'3
11
+

+

+ ( 1(11
11
1 +""''It+
(!!li)
Sine' PC'] II at ion:; (:!.i) atHl (:!(i) :li'P I he rqnations of the same plant',
their tocOit" iPn t" :trP prnpnrt ioual, ltcncp
i= L ?, ::, I, (:!7)
whPr 1' =!= 0 as :1. fa,rm nf pruport l f we sol \'C CfJUations
(::?;') fnt u 1 " "J "" we
in whwh u =F 0 nnd the {3,k haxe thc samt' nwaning as in (:2:0:).
SinCL', wlwn .1'
1
, .
3
, .1'
1
arc suhjeLtrcl to :\ t tansformat.ion
llu 11
3
, 11
1
are snbjcdcd simultaneously lo Lhc Lrnnsfotm:ttion
(28), Lhe systems of variables (.1:) and (1t) are (ailed contragredient.
120
'rET RAilED RA L COORDINATES [CIIAI' . IX.
EXERCISES
1. Prove that the four planes determined by equating to zero the second
meml!ers of equations (23) arc the faces of the coordinate tetralledron of the
system (x
1
11 X
1
2, X
1
3, x'.).
2. State and prove the of the theorem in Ex. 1 for the second mem-
IJcrs of equations (27) .
3. ny of eqnat ious (2 t ) and (23) find the coilrclinates in each sys-
tem of the unit poi nt of the other system.
4 . Determine the oqulnions of a transformation of coordinates in which
the only change is that a ,Jifferent point chos(!n a,.c; unit point.
98. Projective transformations. Equations were rlerived as
the equations connecting t.he roortlinates of a gi \' Cil arbitrary point
refen etl to two of teta.hedral coiinlirt:ttes. We shall now
gi ve these equations another interpretation, entirely distinct from
tho preceding one, bnt equally important.
Let there be gi ven a system uf equations (2 l) with determinant
T not equal to zero. l .. et be a given point and let i ts coordi-
nates, in a given system of tetrahedral coonlinatcs, be (x'
1
, x'
2
, x'
3
,
x'
4
). By substituting t he coordinates of 1" in the second members
of we determine four numbers x
1
, x
2
, x
3
, x
4
, which we consider
as t he coordinates (in t he same system of C<lonliuates as those of
P) of a secotlll point P. To each point/'' in space conesponds, in
t hi s way, one and onl y uue poiut J>. )[oreo,er, when the coordi
nates of P are given, t he coi)rdinatcs of f>' are fixecl by so
that to each poi nt P cor respot11ls one and only one point P'. It is
useful to t hink of the point P' as :wtually changed into P by the
transformation (21) so t hat, by means of (2l), the points of space
change thl'ir positions.
A transformation dctcl'lninecl by a system of eqn:1.tions of the
type (21), with determinant T not equal to zero, is call ed a pro-
jective transformation. The projeeti vc tmnsformation (23) is
call ed t he i1nrcrsc of (:.!1). [f, by P' is transformed int.o P,
then, by Pis tra.nsforme(l into P'.
By (2 1), the points of t he pl:1.11e (u') are transformed into the
points of t he plane (11) determined by Equations (28) are
call ed t he equations of the trausfonuatiou (21) in plane coor
dinates.
ARTS. 98-100)
CROSS RATIO 121
00. Invariant points. The points which remain fixed when
operated on by a given projective transformation (21) are called
the invariant points of the transformation. To determine these
points, put xi= px', in (21). The condition on p in order that
the resulting equations
(cc
11
- p) x\ + n
1
zX
1
z +

+ ccHX
1
4
= O,
a
21
x'
1
+ (c(.!.! -J?)x'
2
+

+ tc
24
X
1
4
= 0,
ll:J1X
1
1 + tt3'ZX'z + ("33- Jl) x'3 + = 0,
a
41
x'
1
+

+ "aX'
3
+ ( a
41
- p) x'
4
= 0
have a set of solutions (not all zero) in common is that
ct
11
-p ((12 ((13 ((14
D(p)=
rez, a.!.!-P C'za C.Cz.a
=0.
f'al
((3'2 ((33 -p ((34
ftu
{(.j.j- ])
(29)
(30)
Let 1)
1
be a root of D(fl) = 0. lf ])
1
is substituted for pin (39),
the points (;c') whose coorJinates satisfy the resulting eqitations
are in.varil'mt points of the given transformation.
If D (1J
1
) is of rank three, equations (39) determine a single
in\'ariantpointwhenp =p
1
(Art.:-$5). If JJ(p
1
)isof rauktwo,
equations (29) determine a line when 7) = 7'
1
Each point of this
line is an invariant point of the transformation. If D(p
1
) is of
rank one, equations (29) determine a plane of invariant points
when p=p
1
If all the of D (p
1
) arc zero, every point
in space remains fixed. In this last case, the transformation is
called the ideutical transformation.
100. Cross rat io. The cross ratio of fom numbers k
1
, k
2
, T.:a, k
4
is defined by the equation
- kl - k2 . k3 - k2
lT- --. ---
kl- k4 k3- k.
The cross ratio of four collinear points Pu P
2
, P
3
, P., or of four
collinear planes 1r
1
, 1r
21
1r
3
, 1r
4
, is equal to the cross ratio of the
ratios of their homogeneous parameters (equations (17) or (18)).
If the parameters of the given points or planes are, respectively,
lu l2; l'u l'
2
; l"
2
; l"'
1
, l"'
2
, it follows that their cross ratio is
l .. l' -ll'. l".l' - l'J"
u= I I - ; _ I _ _ ._1_ ,
l,j."'1 - ll"z l",l"\ - l'"2l"1
122 TETRAHEDRAL COUH.DINATES [CHAP. IX.
If u= -1, the four given points or phtnes are said to be
harmonic.
An important property of the cross ratio is stated in t he follow-
ing theorem :
TnEoHElJ. 'rhe cross ratio offout zwints (o> planes) is equal to
tlte cruss ratio of any fonr ]JOints (or ]Jlcmes) i11to which they can be
l'rojected.
In the projective transf01mation (:?1), Jet the points (x') and
(.c:") of (17) be projected into (y') and (y"), respectively.
[1. follows that the poiut of the range ( 17) whose parameters are
l, and l
2
is projected into a voint (y) of the range determined by
(y') aml (y") sucb that
i = 1, 3, 4.
Since the paramcten; of the points are nnchangerl, the cross ratio
is unchangNL Simi larly for a se.t of fom planes throngh a line.
Conversely, two ranges of points, or pencils of planes, are pro-
j<'clive if t.be cross ratio of flny fom t'lcnwnts in the first is the
same as that of the corresponding dcnwnts in tile second.
EXERCISES
Let A:= (1, 0, 0, 0), II := (0, 1, 0, 0), C:= (0, 0, 1. 0), D:= (0, O, 0, 1),
E:=(l, 1, l, 1). Fintl the equations of a projective trnm;fonnation which
imerchanges tltcsc points as intlicattcl, dewrmine the roots of D ( p):: 0,
fiml the uf the invari:tnt elcmems when
(ct) A is tr:Ln.-fornwcl into -1. B into J:, C into C. D into E, E into D.
(b) A is into U. 11 into .-1, C into1 LJ, lJ into C. E into E .
(c) A is inlll fl, fl iiHI G, C into A. JJ into V, E inlO K
(d) A is trnnsfortncd into T:, R into C. C intu D, D into E, 1:..' into . 1.
.,/ 2 . that a pnljccti\'c tmnsfonnation can he found that will transform
live gilen point.'> A. n. 0. D, E. no [,our of which arP in one plane, into five
given p(lints A', f1
1
, C'. U
1
, nspectilcly, no four of whicb lie in one
plane. 1'how that t he tmnsformatioiJ h: then uniquely tlxctl.
v 3. A non-icli'ntical pmjectivc t ransformation that coincide!; with its own
inv<'rse is called nn involution. Find tbe condition tlutt the tmnsfol'lnation
(21) is an involution .
... 4 . Show thaL the transformations x
1
= .r'" X2 = X
1
2, X3 = X
1
3, x' 4
are iuvolnthms. Find the imariant points in cacb case.
AnT. 100) <.'ROSS ItATlO
123
5. If r, I" are :my two distinct corresl>Onding points io either involution
of Ex:. 4, prove the following staten1ent.s:
(a) The line rl" c.ontains two di:<tinct inmriant points Jl, .ll'.
(/) The points ( fl P' Jf.l/') are hannonic.
6. Find the itwariant points of the transformation :r
1
= x'z, xz = x'
3
,
:ra = :r-
1
4, = x
1
,. Show that the points of space are arranged in sets of four
which are interchanged among
7. Interpret. thl' ( Art . !!G) of a translation of axes a.s the
of a l>rojecthe transformation. Find the invariant Plements.
8 . Interpret the equations ( Art. l'!i ) of a rotation of axes as the equations
of a projective t.mnsformat.ion. Show how this transformation can be
effected.
9. Find the cross ratin of the four points on the line (17) whose param-
eters are (0, 1), (1, 1), ( 1, 5), (4, 3) .
CHAPTER X
QUADRIC SURFACES IN TETRAHEDRAL COORDINATES
101. Form of equation. Since the equation F(.v, y, z, t) = 0
may be transformed into an equation in tetrahedral coordinates
by means of equation (:3) of Art. 88, it follows that the equation
of a quadric smface in tetrahedral cooruinates is of the form
A= (!
11
X
1
2
+ a,.p;.z
2
+ 33'1:/ + a
44
:t'l + 2 a
1
2X
1
X.l + 2 a 1
3
X
1
X
3
+ 2 c<
14
x
1
.l:
4
+ 3

+ 2 a
24
x
2
x
4
+ 2 a
34
X
3
X
4
= 0. a.,k = ak,. (1)
Conversely, any eqnation of this form will represent a quadric
surface, since by replacing each x, by its value from (3), Art. 88,
the resulting ell nation P (x, y, z, t) = 0 is of the form discussed in
Chapters VI, VII, and VIII.
102. Tangent li nes and planes. Let (x) and (y) be any two
points in space. The coordinates of any point (z) on the line
joining (x) to (!!) are of tbe form (Art. 95)
z, =AX, + P.Yi> i = 1, 2, 3, 4.
If (z) lies on the qltadric A= 0, then
.\?A(:c) + 2 Ap.A(x, ?J) + p.
2
A(y) = O,
wherein
A(x, ?!) = A(y, x) = (au?h + a21Y2 + <<31?13 + a.JJI4).vJ +
(a21Y1 + + a<t>.!h + a.2?14)Xt + (a31?h + a32?J2+ <
1
si!/a + auY4)xa +
1
(a41Y1 + a2?12 + a.J?/3 + ct . .y4):c. =- - y; = .t,.
2 iJ:i'; iJy,
(2)
(3)
(4)
If (y) lies on A= 0, then A(y) = 0 and one root of (3) is A= 0.
If (!J) is so chosen that both roots of (3) are A = 0, we must have
A(x, y) = 0. If (x) is regardetl as vatiable, and y) is not
identically zero, the equation A(x, y) = 0 defines a plane. The
line joining any point in this plane to the fixed point (y) on the
t1uadric A lonches the surface at the poi nt (y) (Art. 7G). 'J'he
I ine is a tangent line and the plane A(.1:, y) = 0 is a tangent plane
to A = 0 at (y).
124
ARTs. 101- 103] INDETERMINATE PLANE 125
EXERCISES
1. Find the equation of the tangent plane to x
1
2
+ x
2
Z + 7
3
2
- a
2
x
4
2
= 0 at
tbe_point (0, 0, a, 1).
2 . Show that C"quation ( 4) vanishes identically if
A = ax1
2
+ b7z
2
+ c..t
3
2 = 0 and (y) = (0, 0, 0, 1).
3. Determine the coordinates of the points in which the li ne
x
1
+ 2 7
2
+ x
4
= 0, x
3
- 2 = 0 meets the surface x1xz+xzx3+4 X3
2
= 0.
4. Show that the line :c
4
= 0, x
1
- 3 :r
2
= 0 touches the surface
"" 7
4
2
- 3 r
1
z + 5 7z
2
+ r4(X1 + 5 xz) + X3.1"4 = 0.
103. Condition that the tangent plane is indeterminate. If
equation (4) is satisfied identically, the coeflicient of each x, must
vanish. T hus we have the four equations
au?/ a + :a!/z + a3t!l3 + (ln?h = 0,
a
1
:?J1 + Cf..r..!/2 + Cl:r>.!J
3
+ Cl.12!/ = 0,
013'!!1 + (lzl!h + Cl33!/3 + (143?/4 = 0,
a14Jh + ct2.J!h + <tal!h + a44?14 = 0.
(5)
If these equations are multiplied by y
11
y
2
, y
3
, y
4
, respecti\ely, and
the products ad<lccl, the result is .A(y) =0, hence if the coonliuates
of a point (y) satisfy all the equations (ii), the point l ies on the
surface A = 0. l'rom (3) it follows that the line joining any
point in space to a poinL (y) satisfying equations (5) wi ll meet
the smface .A = 0 in t,,.o coincident points at (!!) If (:t) is any
other poinL on the surface A, so that A(:c) = 0, it follows hom (0)
that every point on the line joining (.t) to (?J) l ies on the smface.
T he surface .A is in this case singular and (?J) is a Yerl<>x (Arts.
66 and 67).
Comersely, if .tl(a) = 0 is singular, with a vertex at (?/), the
two intersections with the surface of the line joini ug (!!) to any
point in space coincide at (y). 'l'he coeffici<>nt A(.1, y) is identi.
cally zero and the coonlinates of (?J) satisfy (5). Since these co..
ordinates are not all zero, it foll ows that the determinant
au (113
a,.
A. =:
(121 Cln (123 ((24
(6)
(131 (132 (
1
33 (/34
au
(143
all
'
126 QUADRIC SURF'ACES [Cu.1r. X.
COJn-etsely, if t. = 0, then fom nnmhe1:-> !/1> .'/
2
, y
3
, y
1
can be found such that the fom equations (i>) :trf' sat i:;fir>cl. The
point (y) li es on A(x) = 0 and in tile plane . 1(-l', !/)=0. The line
joining (?J) Lo any point (.1:) will hrwe two coinciJcnt points in
common with A(x) = 0 at (!1); that is, (y) is a vertex of the quadric
. L \Yc thus have the follo\ving theorem:
Tln:OJ:E)J. The necesscuy comlilion that a quctdric
snl:fa.ce i:> is that the detennioa11t u vani;;hes.
The determinant t. is called the discriminant of the quadric A.
l f it does not \'rtni sh, the CJ.Hadl'ic will be called non-singular.
Unless the contl'ary is stated, iL will be assumed throughout this
chapter that the !':urface is non-singular.
104. The invariance of the discriminant. l n Chapter VII cer-
tain invariants under motion were consid<'red. \Ye shall now
prO\'C the following theorem which will include that of Art. 74 as
a particular case.
'
I. Jj the equatirm of et quadric sw:face 1's suhjectetlto
a linew trcm:sjormatio11 (Art. !JG), the cliscrintinant o.f the
equation is eql{al to the Jnodnci of the cliscrimincmt of the original
equation cmcl the of the cletenni11ctnt of the
I 4
Let A(x) = L a
1
kx,.1k = 0 be the equation of a given q uadl'ic,
l=l PI
and let
define a linear transformation of non-vani shing determinant T.
lf these values of x, al'c substituted in A(x), the equation becomes
4 i
A'(.-,;')= "V "" ct' x' :t" - 0 et',c = a'k,.,
"-f .z.. k' ,. k- I
1=1 b:l
in which
ART. l().t)
Dl::>C:RI;\IlNANT
If we now }JUt
it follows that

,.,k =
u.=l
If we form the tlisrriminnnt. of .A'(x'), we may write
127
lt
11
1'
11
+

+ ct
31
r
31
+ tt
41
i"
41
1tul'12 + + l'31l'32 + lt41i'z

+ 1'221'
21
+ a
32
1'
31
+

u
12
7'
12
+ +

+ ...

This tlE'terminant may be expreSSE'\1 as the product of two dete1-
minants T and R (Art. !JG, footnote), tlms
Uu 1(1:! ((13 1(14
l'u ,.12 ,.13
1\41
((:?L 1'22 ((24 t21
,.23
1'24
'
((31 1'32 1'-.13
((31 ,.31 1"32 )'33
,.3.
1(41 (( ..
''3
((44 ,." 1'43
,.'"I
the columns in the first fartm associated wit.h the columns
iu the second to form the elements t>f the rows in the prodnct.
Similarly, the second factor may b<' expressed in the form
a11ct11 + C112tJ1 + a13tt31 + C/11'1
((211(11 + (I:!'J"JI + (t:!3/(J 1 + ((21(.(41
+ + + CLtl"42
Cl21lt12 + + + ftzi'42
which is the product of hy T, the elements uf the rows in thE'
first factor being mnlt.ipl iNl by thu elrment:; of the columns in
the second, hence
']'2.)..
()u account of this relation, the discriminant is said to be a rela-
tive invariant nnder linear transformation of tetrahedral coordi-
nates. )lmeo,er, the thcor<'m will now be proved.
Tu1:rH:f:'1 I L Auy silt mino, ?IW?J be expl'essecl as a liltear
.fiu,liuu u(lftt silt mino,s uf
128
QUADRIC WH.F'ACES [CHAP. X.
The method of proof will be sufficiently indicated by consider-
ation of the minor
This determinant, when written in full,
I
Ctu ?"n + + s1?'st + l'4t1"4t
ct ,z?"n + + + et4z''t
+ CtztT22 + "atr32 + " t?'421
ct,z?"t 2 + l'-r-?"22 + Ua2rs, + 42?"2
may be expressed as the sum of sixteen determinants, fom of
which vaui sh identi cally . The remaining ones mu.y be ananged
pairs, by combining t he determinant formed by the ith t erm of
the fi rst column and the kt h term in t he second with that formed
by the kth term in t he first column and the ith i n t he second.
Every such pair is equi ,alent to the product of u. second minor of
6. aud a second min or of T. If i = 2, k= 3, for example, we have
I az,7.2t
"z2T21
Ctst1"az l + I Ctst1's1
a2l"a2 ct szi'at
lt2tr22 1
" z2r22
I C'-!t Ctst I I f'zt
"at l
= (t2'l
,-.t?' s - 1'atr22
1(32 ((32
((22
=I:
((3t
1
. r21
( ( 3'.:! ?.:!:l
?"at I
7'32
In this way it is seen t hat every second minor of t::. ' is a lin ear
fnuction of the second minors of the det erminant R, the coeffi.
cients not containing t ;t
4
By replacing each ?';k by its value aml repeating the
m= l
same process, it may be seen that each second minor of R may be
expressed as a linear function of the second minors of t:.., the
coeffi cients not containing any a,k. Tl1C same reasoning may be
applied to t he first minors of t::.'. This proves the proposition.
As a rorollary we have t he further proposition:
THEORE)t Ill. '17t e m nk of tlt P. of the equation of
ct quadric su1:(acP Mt cltWt(Jell by any lhtear 1uith
non-ucw ishiny deleo ninant.
ARTs. 104, 10.3) LINES 0::-r THg QUADRIC SURFACE 129
For, it follows from 'fh. ll that the rank of A' is not greater
than "that of A. Neither can it be less, since by the inverse trans-
formation the minors of A may be expressed linearly in terms of
those of A'.
We mn.y now conclude: if the discriminant A is of rank four,
the quadric .tl(.t) = 0 is non-singular (Art. 103). lf A is of rank
thtee, A= 0 is a. non-composite cone, for if we take its vertex
(Art. 103) as the vertex (0, 0, 0, 1) of the tetrahedron of refer-
ence, the equation A= 0 reduces to
a
11
.-r
1
2
+

+ a
33
x/ + 2 a
12
x
1
x
2
+ a
13
X
1
X
3
+ 2 a23X2--t
3
= 0.
The line joining any point on t.hc surflLCe to (0, 0, 0, 1) lies on
the smface, which is therefore a cone (Art. 46). Since by
hypothesis A is of rank three, we have
hence the cone is non-composite. If A is of rank two, the quadric
is composite, for if we take two vertices as (0, 0, 0, 1) and
(0, 0, 1, 0), the equation reduces to
which is factor:tble. Since by hypothesis A is of mnk two,
rt
11
Ctz:!- a
12
2
is not zero, hence the two components do not coincide.
If A is of rank one, the e11uation ma.y be reduced to the form
.t
1
2
= 0, which represeuts a plane counted twice.
105. Lines on the quadric surface.
'l'HtWRE)t. The section a fJIW.dric sw:face nwcle b!! any of
its tange11t Jllanes consiflt.s of tu:o lines passing through the point
of tcmgem:y.
For, let (!/) be any point on a. quadric surface A= 0, and (z) any
point on the tangent plane at (y), so that .tl(y) = 0, A(?t. z) = 0.
If (z) is on the cutve of intersection of .A(.) = 0, .ll(.r, y) = 0,
then A(z) = 0 and U3) is iuentically satisfied, hence e''et-y poiut
of the line joining (y) to (z) lies on the surface. Since the sec-
tion of a CJnallric made by any plane is a conic {Art. 81) and
oue component of this conic is the line joining (y) to (z), the
residual component in the tangent plane is also a straight line.
1:30
QUADRIC SURFACES [CHAP. X.
The second lino also passes through (!/), si nce every line lying in
tl te tangent plane an<l passing th rough (!!) has two coiucident
points of intersection with the snrbce at (y).
106. Equation of a quadric in plane coordinates. Let t he plane

+ H
2
.1
2
+

+ 1t
4
X
4
= 0 (7)
be tangent to t he given qu:uhic A , and let (!J) be its point of
tangency. Si nee .'!) = 0 also the equation of the tangent
plane at ('!), the equation = 0 mnst di ffer from .t1(x, ?!) = 0
by a constant factor k (..Att. hence
Ctu?h + Ct z1?12 + Ct31?!3 + a.1!1 = knu
a12!h + tt.r.?h + + a.z?l = ku2,
ct 1:J!J, + ct2J!h + Ctai!J
3
+ a4J?J4 = k113,
a ?l + ll21!h + lla.1!h + a.u!l = k u .
1\Ioreover, since (y) l ies in the tangent plane, we have
1t
1
y
1
+ ll'!Jh + ?t
3
.1f3 + 11
4
?!. = 0.
(8)
(9)
On el iminating y
11
y
2
, y
3
, !J
4
allll k. between (8) a nd (0), we obtain
as a necessary condition that the plane (11) shall be tangent to t he
surface,
(Ill <lz1 (131 a., u,
(112
11
22
1132 ({42 1!2
4>(11) =
ct1a
((23 1( :13 II .a 113 = 0.
(10)
"t
((21 1/H 114
11
1
7(2
Ua u. 0
Conversely, if the coonliJLates of a plane (u) satisfy (10), and
i f also u =F 0, then the pl ane is tangent to the qnatlric . 1 = 0.
For, if (10) i:; satisHell. number:; ?Ju y
3
, y
4
k. not all ZPto,
can he found which satisf: (X) and (9). Tn particular. k * 0, fnr
othctwise, si nce t. * 0, il woultl follow hom (8) that ?It= !h =
Ya =!f.,= 0, conLrary to the hypotheses. Since u" u
2
, 11
3
, u
4
arc
not all zero, it follows from (S) Lhat !It, y_, !/
3
, y
4
arc 110t all zero,
ancl hence atc the coiinlin:tt(s of :\ point. By sull'ing (8) for
llu tt-z, u
3
, u
4
and substituting iu (!>), we obtain .l(y)= 0, h'ncc
the point (!J) lies on t he I]U:ttlric A. From (I) an1l (7) it
follows that the plane (T) is tangent to A at the point (y).
ART. 106) 131
The equation <I> (tt) = 0 is of the second dt>gree iu 11
1
, a
2
, 11
3
, 11
4
.
It is the equation of t.he qu:1<lric in plaue cuottlinates.
By duality i t fol lows that auy eq nation of the secOtHI tlegrce
in plane coonlinaLes, whose ,}istrilllinant is not ?.cto, is the equa-
tion of a quadric snrfa<"e in vl:uw toiir<l iuates.
is of muk thn'P, so that A= 0 is tlw C'qnation of a COJH',
the equation <I>( rt) = 0 retluces to (:Sk,u.JZ= 0, 'j.k,t; = 0 the
equation of the Yertex of the cone. I f is of mnk hss tban
three, <I>( a) = 0 vanishes identically. The equation <1>(11) = 0 was
iu fact derived simply by imposing the coudition that the section
of the quadric uy Lhe plane (u) should be composite.
EXERCISES
1. If the equation A (.') = 0 colltai11s hut three VO'lrin show that it
represents a singular qn:ulric.
2. Calcul:\te the cliscrilllinalll of ... + x
2
2- :r
3
2 = 0.
3. Show thn.t tlw tlisritllinant uf ol(u) = o contain,; the discriminant of
A(.t) = 0 as a factor.
4. r.inm . l (.t) = +

+ ,.,.
2
+ t/ ... = 0, tktenuine tho form or
the r>quation 1>(11) = 0.
5. When the equation <1(11) = 0 is !::inn. show how to obtai n the equation
..t(x) = 0.
6. Gi\cn A (.c)= a/
1
2 +

+ 2 u
3
.c
4
= 0, fiutl <1>(11) = 11.
7. F ine! the tl iscrimlnant nf
A' .r) = x1
2
- - .Ct.C.\ - X2Xa + ,..,. 1 + 72-'4 + .r J''1 = 0,.
and tll'tr>rminc the form 0.
s. niHn <l>(u)= u1"- 2 u,u
2
+ u
2
2
+ 2u
1
u
3
+ 2 u
1
u4- 2 ;- 2 11!" +
u32 + u12 + 2 11;)11 = 0, find .A(.t) = 0 and interpret
9. Find the two lines lying in the tangent pla-ne x
1
= 0 to tlot: quadric
10. \\'rite the. equatiou of a quadric pas.;ing through each venex of the
tetrahedrun o( reference.
11. Write Lhe equat ion of a quadric touching each of the coonliunte
planes ( use dual of metholl of Ex. 10).
12. Write the equat ion of a quatlric which touches each edge of the tetra-
hedron of reference.
13. \ \.hat is represented by the equation :Sa;
1
u;uk = 0 when the dis-
criminant is of rank thr!!e? of rank two? of rank one?
11. Shnw Lhat. thrOn!rh any line two planes can be drawn tangent to a
gi,en non-singular quadric.
132 QUADRIC SURFACES [CHAP. X.
107. Polar planes. " 'hen the coordinates z" z
21
z
3
, z. of any
point (z) in space are substituted in ...1(.t, z) = 0, the resulting
equation defines a plane called the polar plane of (z) as to the
qnatlic A.
Let(!!) be any point in the polar plane of (z), so that
.A(y, z) = 0. Since the expression
A(y, z) = A(z, Y)
is symmetric in the two sets of cooi'Clinates y
11
y
2
, J/s
1
y
4
and z
11
z
21
z
3
, z., it follows that (z) lies in the pola.r plane of (y). Hence we
have the following theorem:
Tm:OKE:'It. If the poiut (y) lie.s on the polmJ>lane of (z), then (z)
lies on the polar plane of(];).
Any two points (.v), (z), each of which lies on the pola.r plane of
the other, arc ca.lle1l conjugate points as to the quadric .A(x) = 0.
Dually, any two planes are said to be conjugate if each passes
through the pole of the other.
108. Harmonic property of conjugate points. We shall prove
the following theorem.
Tm:ORE;\1. Any t1uo conjugate J>Oints (.t), (y) an(l the two ]JOints
in which the line joining intersects the QIHtclric constitute a set
of ltaononic points.
The coordinates of the points (z) in which the line joining the
conj ugate points (x), (y) as to the qnallric A are obtained by
putting z, = A.x, + P..l/; and substituting these values in A(z) = 0.
The values of the ratio.\: p. are roots of the equation (Art. 102)
A.
2
.A(x) + 2 A.p.A(x, ?I) + p.
2
A(!I) = 0.
Since A (x, y) = 0, if one root is .\
1
: p.
11
the other is - ..\.
1
: p.
1
'l'he
coordinates of the points (x), (!J) and the two points of intersec-
tion ate therefore of the form
hence, the four points arc harmoui c (Art. 100).
Dually, any two conjugate planes (a), (v) and the two tangent
planes to the qnatlic through the!r line of determine
a set of harmonic planes.
ARTS. JO"i-110) 'l'ANGENT CONE 133
1 09. Locus of points which lie on their own polar planes. The
condition that a point (y) lies on its own polar plane .A(.v, y) = 0
as to A(x)=O is A(y, !!) = A(y) = 0, that is, t.hat the point lies on
the quadric. \\"e therefore hn.Ye the theorem :
T HEOI{IDt. 77te locus of ]Joint:> ?t:hich lie on thei, pulal' planes as
to c' quadric .A(.v) = 0 is the fJIIttdric it:se{l:
Since when (.IJ) is a point on A(x) = 0, A(x, !I) = 0 is the equa-
tion of the tangent plane to A(x) = 0 at (y), it follows tltat, the
polar plane of any point on the surface is the tangent plaue at
that point.
A point which lies on its own polar plane will be said to be
self-conjugate. Dually, a plane which passes through its own pole
will be said to l.H self-conjugate.
1 10. Tangent cone. If from a poi ut (IJ) nut on the quadric A
aU the tangent lines to the smfacc a.re drawn, these liues define a
cone, called th<' tangent cone to 1 from (tJ).
Tut::Oin; )J. 'J'he cone to u 'f'tWlJ"iC .fJ"om cwy JIOint not on
the sw:fate is a quml1ic cone.
Let (.t) be any point in space. The of the points
(z) in which the line joining (:c) to (!J) meets th<> qu:uhic A are of
the form
in whi<:b >..: p. are roots of the quallric E>qnatiou
A
2
A(x) + 2 Ap.A(x, y) + p.
2
.A(.q) = 0.
The two points of intersection will be coincident if
[ A(x, !t)JZ = .A(x).A(y). (11)
If now (y) is fixed and (x) is auy point on the surface defined Ly
(11), then the line joining (x) to (y) will be tangent to A= 0.
Since this equation is of the second degree in x, the theorem
follows.
The cmve of intersection of the tangent cone from (y) and the
quadric is found by considering (11) and .A(x)= 0 simultaneous.
The intersection is evidently defined by
[.A(x,y)J2 = 0, A(x)=O.
13-!
QUADRIC SURFACES [UliAP. X.
This Joens is t he coni1' of intersection of the quadric and the polar
pl:uw of the point (y), <'oun ted twict>.
If (y) is a point on the smface, then A(y) = 0 and the tangent
cone reduc-es to the tangent plane to .:l = 0 at (y), counted twice.
1 11. Conjugate lines as to a quadric. We shall now prove the
foll owing theorem.
Tt m<li:JDL The pol((r
1
1luue of every ]JOint the lhw joiuiltff any
two fficen ]JOi11ts (y), (z) throuyh tlte line of intersection of the
jJ(Jlw plcwes of (!J) wHl (z).
The polar planes of (y) and of (z) are .il(.t, y) = 0 and A(x, z)
= 0. 'l'he coordinates of any point of the liue joining (?J) n.ncl (z)
are of the fotm A.!t; +p. z,; and tl1e volar pl:lne of this point is
.l(.t, ,\y + p.z) = 0. ::>ince t his equation is l iueat in 1\y
1
+ p.Z,, it
may be rewritten in the form
A..A(x, y) + p. .tl(x, z) = 0,
which proves the t heorem.
From Art. 107 it follows that the point plane of every point of
the secontl line passes thtongh the fit::;t. 'l'wo such lines are
cal led conjugate as to the qnn.clric. lf frotn P, any point on the
quadric, the t ransversal to n.n.' pair of conj11gatl' litH'S is 1lrawn, it
will nwet t.he qunchic again in the h:wmonic coujngnte of Pas to
the points of inLCJsection with Lhe tnnjngate Jines, sinl'e its inter-
sections with these 1ines are conjngn.Le points (Arts. 107, 108).
EXERCI SES
1 . Determine t he cquation of the polar planr of ( 1, 1, 1, 1) as to the
quadric + :r2'! + "J"a
2
+ = 0.
2. Find the equation of the line conjugate to x
1
= O, x
2
= 0 as to the
quadric

+ "J"
2
2 + xJz + :,
1
2 = 0.
3 . l"how tll<\t nuy four poiuts on a line h:we the l':une cross ratio as their
four polar planes.
4. Find the cone to X1/2- x
3
.c = 0 fmw the point (1, 2, 1, 3).
5 . If a li lll llll'<ts a pt:ttlric in Panll Q, show that the tangcut planes at
1' nucl Q mP('L in the roujng:ue nf the line.
ARTS. Ill ll:ll S8LF-POLAR 'l'ETR.\Ifl:DnO::-.: 135
6. !'hnw thaL t)ll' quadric . .;

+ + 0, .r," -r + r}- l.r

1
2
= 0 are siHh thaL lh<' pol:w plane uf ('), 0, 0, I) h; the :-a me ft>r hllt-r-
pret Lhili fact E:tometrically.
7. \\'ritP the l'f)llalion of a qnadtic the line r
1
= 0, .r
2
= 0.
Jlow many cunditiuus docs this iutp>se llJI!III thl' Cinatiun :
8. \\'l'ite the l'i}Untiou of a qun,Jric contalniug the line :r
1
= 0, ;r
2
= 0 and
the line .t
3
0, r
1
=- 0.
9. Show tlmt any three lines, no Lwo of which inter:sect, pa<ses
one and but o11e t)untlric.
11 :.?. Self-polar tetrahedron. AssocialC'tl with c\'ery tetrahedron
P
1
P
2
1' J>, a tetralwtlron -::-
1
-z-
3
-. fol'llll tl hy the polar planes of
its \'Prliccs, r.
1
of P., of Pz, ""a of / ',, arul "" of P.. Con,erscly,
it follows from .\ rt. 101 that the plaHc 1'
1
P
2
J>
3
is the polar plane
of the point r.
1
r.z-;:-
3
, Ptr.
Two iC'ttn.hcdtn P
1
1'
2
P
3
P
4
, -;:-
1
"
2
"
3
.,.., snell thnl tho faces of each
are the pol:l.t planes of the ,erti<Ps of t.hc oihcl' as to a. giveu
quadric, ai'O C>nllctl polar reciprocal letrahC>dra. I f the two lctaa-
hedra coincide, :>O that the plaue -;:-
1
is iciPntieal with the plane
PzP
3
P
4
, etc., the tt>tralaedron is cnlhd a self-polar tetmhcdro11.
To determine a sllf-polnr tctralu<lmll dwosc nn.\' poi11t J>, uot
on and 1lctermine its poln1 plan .. r.
1
In this pola plane
choose nny point P
2
not on 1( .. ) nml t!Ptemttae its pol:u plnne ":
This plane pa:ssC's f'
1
( \l't. 101). On the line of inter-
section of -;:-
1
,..-, choose a third point 1'
3
not 011 A(x) nllll detPr1ni11e
its polar plane -;:-
3
The plane -;:-
3
passes P
1
:mel
Finally. ]l'L /'
4
be the point, of inll!rsf>cLioll of r.
1
r.zll'a Tlw polar
plane r.
4
of /'
4
passes through point.s l',P: /'
3
Henc.: ihC' tetra..
hedron P
1
/'
2
1', 1'. = ... ,-;:-,-;:-
3
-;:-
4
is a self-polar tetrahedron.
1 13. Equation of a quadric referred to a self-polar tetrahedron.
T/11 C11lfl .wf/idPnl crm<litirm tlutllltP t '1"'1
lion of CL 'JIHtcldc Ctntlttiuli rml!llhe of lhf' c";irdi,wles i" tlw.t
a :st{f.polw telraltulrou du,scu t (,a!tl'fl m11 1!( ,rj( ,l'nt'e.
1f the tetmhcdron of rE>ferl'nce is a sC>If-polar tetrahedron, the
polar plane uf the vertex (0, 0, 0, I) is .
1
= 0. Hut 1 hP e11 natio11
A(.c, !!) = 0 of the polar plane of (0, 0, 0, 1) is ll
4
,x
1
+

+ <1
43
:r
3
+ CI
11
X
4
= 0, hence (!
41
= a
42
=

= 0. Siuce the polar plane of
136
QUADRIC SURFACES [CHAP. X.
(0, 0, 1, 0) is x
3
=0, it follows further

since
the polar plane of (0, 1, 0, 0) is = 0, that et
12
= 0. Hut if these
conditions are all satisfied, then the polar plane of (1, 0, 0, 0) is
x
1
= 0, and the equation of the quadric has the form
Conversely, if the equation of a quadric has the form
the tet.ra hed ron of reference is a self-polar tetrahedron. Since
ll * 0, the coefficients a.,, arc all different from zero.
If the coefticients in the equation of a quadric are real numbers,
it follows from equatiou (t) that the pola,r plane of a real point is
a real plane, hence from Art. 88 the equation of the quadric can
be reduced to the form ::Sa;;X
2
= 0 by a real transformation of
coordinates, that is, one in which all the coefficicnt.s in the eqna,.
tions of transformation are real numbers.
By a suitable choice of a real unit point the equation of the
quadric may further be reduced to the form
1 14. Law of inertia. The eqnatiou of a quadric having real
coefficients may thus he reduced by a real transformation to one
of the three forms
(a)
(b)
(c)
Xt2 + Xz2 + Xa2 + xi= O,
Xt2 + + Xa7 - ;1:/ = O,
xt2 + ,,.2z- J:a2- xtz = 0.
Ttt EO REM. 17u? NJIIfllirm CW!fTcal non-singular quad1ic may be
?'Nlucerlby n 1-ecll to one cmd only one of the type!i (a),
(/1), (c).
A quadric of type (a) contai ns no real points, as the sum of the
squares of four real numbers can be zero only when all the num-
bers are 1.cro. If the ef]nat.ion is of type (b), the surface contains
real points, bnt no real lines, for a real line lying on the surface
would cut every r<'al plane in a rea.! point, but the section of (b)
by x
4
= 0 is the conic

+ x
3
2
= 0, which contains no real
ARTS. 113-1151 RECTILINEAR GENERATORS 137
points. If the equation of a quadric can ue educed to type (c),
the surface coutains real points and real lines. The Jiue
x
1
- .1:
3
= 0, ~ - x
4
= 0, for example, lies on the sul"face. Any
nal plane through it will intersect the quadric in this liue and
another real line. If the equation of a quadric can be reduced
to one of those forms by a real transformation, it can evidently
not be reduced to either of the others, since real lines and real
points remain real lines and real points.
The theorem of this Article is kuown as the law of inertia of
quadric surfaces. It states that the numerical difference between
the number of positive tems and the number of negative terms
is a constant fot any particular equation independently of what
real transformation is employed.
By a transformation which may involve imaginary coefficients
the equation of any quadric may be reduced to the form Sx;
2
= 0.
Fot this purpose it is necessary ouly to replace X; by ..3!....=_ in the
-Jai,
eqn<ttion };a;;X/ = 0 of Art. 113.
llr;. Rectilinear generators. Reguli. If in the equation
S./ = 0, the transformation
x
1
= :1!
1
+ x'
2
, x
2
= i(x't- x'
2
), x
3
= i(x'
3
+ :1!
4
), X
4
= (x'
3
- x'
4
)
is made, it is seen that the equation of any quadric can also be
written in the form
(12)
If the quadric is of type (c), its equation can be reduced to (12)
by a real transformation. In the other cases the transformation
is imaginary.
The line of intersection of the planes
(13)
lies on the quadric for every value of k
1
: k
2
, since the coordinates
of any point (y) on (13) are seen by eliminating k
1
: k
2
to satisfy
(12). Conversely, if the coordinates of any point (!/) on the
quadri c are subst.itnted in (13), a value of 1.:
1
: k
2
is determined
such tha.t the corresponding line (13) lies on the quadric and
passes throu:;h (y).
138 QUADRIC SURFACES [CHAP. X.
Xo two lines of the syste m (l:J) intPI'sect, for if k
1
x
1
=

aml k'
1
.r
1
= k'
1
x
4
= k'
2
x
3
are the two lines, the con-
dition t hat they inte1sect is
kt 0 7.:2 0
/,'\ 0 7.:'2 0
(k k' k k' )2- 0
0 k2
()
k
= - I 2- 2"1 -
t
0 k'
2
0 k'
t
Bnt thi s conclit ion i s not satisfied unless k
1
: k
2
= k'
1
: 7.:'
2
, that i s,
unless t hP t.wo lines eoilll'i\lc, hc nct' :
TnEniW\1. TlutJIIflh Pwh poiul 011 the IJIHtclric ( 13) JJasses one
wol but onr' the ( 1:1 ), lyinu entirely on the su,:f((ce.
A system of lines h:-wing this property is callell n, regulus
(AI'! . 'iH).
In the sa.me way it is "hown iu Lhe S,'\Stem of lines
(1-1)
is a. regulus on the same quadric Those two rcguli will
be cnllell the k-rcgulus a.ncl t he res pectively. It wa,s
seen Llmt no two lines or tht: same regulus intersect. It wi ll now
he shown Lha.t every lin<' of eaeh regulus intersects every line of
the other. Let
be a line of the 7.:-regnlns and
ll'ct = l2-r
4
lt.l'
3
= lz-'1:
2
be a line of the l-reguln:;. T he roml ition thnt these lines intersect
is that
k
t
0
0
z, 0
0 12
7.:2 0
0 kt
=0.
0 12
It
0
But t hi s !'I'JllMion is itl (nt ically; hence t he lines intersect
for all ' alucs of 7.. : k
2
all(\ Z
1
: 1
2

11 <>. Hyperbolic coordinates. Parametric equations. Each value
of tlac 1at io 1.:
1
: 1111 ir1uely ll eteruJ itws n. li of the 1.:-regulus; each
value of l
1
: unip1ely tlctRIna ines a line of t.he These
t.wo lines iu tmscct ; tilcil' point of intersection l i<s on I he quadl'iC;
AnTs. L1 5-l l'i) PROJECTION UPON A PLANE 139
through luis point passes nu other line of either regulus. Thus,
a pair of vn.lues k
1
: and l
1
: l
2
fixes a point on the surface.
Conversely, any point on the surf:we fixes the linu of each systE'm
passing through it. and consequently a pair of valur:; of k
1
: k
2
:mel
l
1
: !
2
These two nutu bers are called hyperbolic coordinates of the
point.
From eC}uations (1.J.) the relations bc.>tween the coortlinatc.>s
x
11
x
3
, of a point on the smface and the hyperbolic coonli-
nates k
1
: k
2
, 1
1
: l
2
:we
x
1
= k
2
l
2
, = k
1
lu :t'
3
= k}
2
, .r
4
= kl
1
(1.3)
These eq nations are called the parn.nwt tic eC} tHttions of the
quad1 ic (12). the equation of any tWn-singnlar quadric can
be reduced to the form ( 1:3) by a suitahlr clwic<' of tetrahedron of
reference, it follows that Lhe g<'ncral form of the
equation of a quadtic snrfnce, refcned to any system of tetra-
hedral coordinates, may be 1nitte n in tlt c Conn
117. Projection of a quadric upon a plane. Given a quadric
surface A and a plane tr. If eaclt point Pof connectc<l wit.h a
fixed point 0 on .1 bnt not on tr, the line OP will intersect 1r in a
point P', called the image of P. C01wersely, if any point P' in 1r
is gi,en, the ]toint P of wltich it is the image is \,he residu:tl point
in which 01' intersects .tl. If P clescribes a locus on A, will
describe a l()cu:; on tr, aud conversely. This process is callNl the
projection of A upon tr.
Through 0 pass two generators :It a111l :12 of .1, one of each
regulus. These lines intersect 1r in points 0" 0;, whi<h are
singnlat elements in the ptojection, sime any puint of ;t
1
has 0
1
for its image, and any point of [h has Ot for its image. The tan-
gent pl:tne to .d at 0 (.ontains the litws g
11
ff2, hence it intet-
sects the plane tr in the line 0,0
1
Any point P' of

will
he the image of 0. Tlte line 0
1
0
2
wi ll be rallc.>cl a singular line.
The tangent lines to A at 0 form a penri l in the tangent plane;
any line of this pencil is fixeu if its point of intPl'S('ction with
0,0
2
is known. [f n. ru1Te on .1 passes through 0, the point in
which its tangent cuts 0,0
2
willlJe saill to be the image of the
140
QUADRIC SURFACES [CnAt. X.
point 0 on that curve. The generators of the regulus to which fll
belongs all intersect g
2
; each, with 0, determinrs a plaue passing
through g
2
, and the intersections of these planes with ,. is a
pencil of lines passing through 0
2
Similarly for the other regn-
lns and 0
1
The two reguli on .A have for images the pencils of
lines in ,. with vertices at 0
1
, 0
2

118. Equations of t he projection. Let 0, 0
1
, 0
2
be three vertices
of the tetrahedron of reference; take for fourth vertex the point
of contact 0' of the other tangent plane through o.o2. If
0 =(0, 0, 0, l), o. =(0, 0, 1, 0), 0' =(1, 0, 0, 0), 02 = (0, 1, 0, 0),
the equation of the surface may be written
A= x
1
:t
4
- = 0.
Let eu

be the coordinates of a point in the image plane, re-
ferred to the t riangle of intersection of x
1
= 0, Xj! = 0, x
3
= 0 and
the image plane ,. or =0. Any point of the line joining
(0, 0, 0, 1) to (y
1
, y
2
, !Ja
1
y
4
) on .A wil I have coordinates of the form
ky
1
, ky
2
, ky
3
, ky
4
+ .\,
wherein k'i.cc,y, + a
4
A = 0 for the point in which the line pierces
the plane ,..
Moreover, since = ky
1
(i = 1, 2, 3) and y
1
y
4
- Y'lJia = O,
y. = ?12?13 = k
?h e.
Hence, a point (y) on A and its image a) in ,. are connected by
the equations
PY =

P?/2 = PYa = P!l = (16)

If = 0, then y
1
= 0, ?h = 0, y
3
= 0, so that any point of the line
(): 0
2
corresponds to 0. If = 0 and

= 0, all the y, vanish, but
if we allow a point to approach o. in 1T along the line e.- = 0,
then the corresponding point ou A is
P'!h = P!h = P'!h = P?h =
from which the factor can be removed. If now e2 is made to
Yanish, the point 011 . 1 <!!'fined hy
?/J = 0, !h = 0, ?fa - = 0.
AnTS. 117-119) QUADRIC THROUGH THREE LINES 141
Thus, to the point 0
1
correspond a.ll the points of the generator
g
11
but in SUCh manner that to a direction el- Te2 = 0 through 01
corresponds a definite point (0, 0, T
1
1) on g
1
To the line
e1- Te2 = 0 aS a WhOle COrreSpOndS the line
'!h - = o, Y3 - TY = 0,
that is, a generator of the regulus fh A plane section cut from A
by the plane 'i.u,x, = 0 has for image in 1r the conic whose equa-
tion is
ulelz + + u3ele3 + u4e2e3 = 0.
It passes through 0
11
0
2

EXERCISES
1. Prove that if the image curve 0' is a conic not passing through 0
1
nor
0
2
, then the curve 0 on A has a double "point at 0, intersects each generator
of each regulus in two points, and is met by an arbitrary plane in four points.
2. If C' is a conic through 0
1
but not Oz, then Cpasses through 0, inter-
sects each generator g
1
in two points and each generator U2 in one point; it
is met by a plane in three points.
3 . By means of equations (16), show that C of Ex. 1 lies on another
quadric surface, and find its equation.
4. By means of equations (lG), show that C of Ex. 2 lies on another
quadric, having a line in common with A. Find the equation of the surface
and the equations of the line common to both.
/
119. Quadric determined by three non-intersecting lines. Let
the equations of three straight lines l, l', l", no two of which inter-
sect, be respectively
'i.u,x, = O, I v,x, = 0; :Su',x, = 0, :Sv',.1:, = 0; I tt",x, = 0, I v",x; = 0.
It is required to find the locus of lines intersecting l, l', l".
Let (y) be a point on l" so that
:Su",y, = 0, 'i. v",!f, = 0. (1 i)
The equation of the plane tleterminr!l hy (?!) and lis
(18)
and of the plane clet.t>rminecl hy (?f) nnd l' is
(19)
142
QUADRIC SURFACES [CHAP. X.
The planes (18) and (19) intersect in a line which intersects l, l',
l". :Moreover, t he equations of every line whi ch intersects the
given li nes may be written in this form. If we eliminate y., y
2
,
y
3
, y
4
from ( 17), (18), ( 1!>), we obtain a necessary condition t hat
a point (x) lies on such a line. The equation is
tt
1
(vx) -v,(ux) u"2(vx) u
3
(v:t) -v
3
(ttx) 1tlvx) -tlvx)
tt',(v'x) - v' ,(u'x)
1t''t
v"
I
u"a
v"a
wherein (wl:) is written for :Su,x;, etc.
=0,
(20)
Since this equation is of the second degree, the locus is a quadric.
The skew lines l, l', l" all lie on it, hence it cannot be singu-
lar. The common transversals of I, l', l" belong to one regulus,
and l, l', l" themselves are three lines of the othe!: regulus.
IE x, = 0, = 0 is chosen for l', and 0, x
4
= 0 for l", the
equation becomes

0 0
0 0
x4 -x3
=0.
u, t/2 11
3
114
v, v2 'Va v,
If we write
thi s assumes the form
'l' he peucil of planes k
1
x
1
+ k
2
x
2
= 0 is associated with the pencil
k
1
x
3
+

= 0 in such a way that associated planes pass through
the same point of the line l. Two pencils of planes associated in
t liis way are projective (Art. 100).
'fhe locus of the intersection of corresponc1in!; planes of two
projective pencils of pla.nes whose axes do not intersect is a non-
singular quadric containing the axes of both pencils.
Dually, the Jines joinin!; the corresponding points of two projec-
tive ranges generate a qnatlric surface. The lines containi ng the
gtven ranges of points belong to the other regnlus of the quadric.
For this rea.son it is sometimes convenieut to consider the gener-
ators of one regulus as directrices of the other rcgnlus.
ARTS. 11!)-121) THE QUADRIC CONE 143
120. Transversals of four skew lines. Lines in hyperbolic posi-
tion. 'Ve can now soh'e the problem of determining the number
of lines in space 'vhich intersect four given skew lines l" l
2
, l
3
, l
4
by proving the following theorem:
THEORE.:\I. Four
cirlent) tmnsversals.
ct Tegulus.
skew lines have at least t11:0 (distinct or coin-
If the!J have more them two, they all belo11g to
Any three of the lines, as l., La, determine a quadric on which
l
11
l
2
, l
3
lie anc.l belong to one regulus. The common t ransversals
of /
1
, 1
2
, 1
3
constitute the generators of the other regnlus. The
line l
1
ei ther pierces this quadric in two poi nts Pu P
2
, or lies
entirel:l' on the Slll'face. [n the first case, through each of the
points P
1
, P
2
passes one generator of each regulus, hence one line
meeting 1
1
, l
3
But Pu P
2
are on l
4
, hence through P
2
passes
one line meeting all four of the given lines. In the second case,
z. belongs to the same regul us as l
1
, 4, l
3

Four lines which uelong to the same regul us are said to be iu
hyperbolic position.
EXERCISES
1. Write the equations of the quadric deterlllined by the lines
:r. 1 + :r.z = 0, X3 + .r4 = 0; 2 .c1 + .ca = 0, Xz + Xa - 2 X4 = 0;
z1 - Xz - xa + J:4 = 0, x1 + 2 x2 + 3 xa + 4 X4 = 0.
2. Find the equations of the two transversals of the four lines
X1 = 0, Xz = 0 j f.a = 0, X4 = 0 j X1 + Xa = 0, X2 + z. = 0 j
Z1 + J' -= 0, ,,.2- .!'a= 0.
3. \\'hen a tetrahedron is inseribE:ll in a quadric surface, the tangent
planes at its vertices meet the opposite faces in four lines in hyperbolic
position.
4. State the dual of the theorem in Ex:. 3.
5. Find the polar tetrahedron of the tetrahedron of reference as to the
general quadric A = 0.
121. The quadric cone. It has been seen (Art. 104) that the
surface A= = 0 represents a proper cone if and only
if the discriminant is of rank three. In this there is one
point (y) whose coortlinates satisfy the four equatious
an!h + rr
1
'!)h + ( t.,
3
!h + a,.I!J
4
= 0, i = 1, 2, 3, 4. (21)
The point (y) is the vertex of the cone.
144
QUADRIC SURFACES [CDAP. X.
The equation of the polar plane (Art. 101) of any point (z) with
regard to the cone is
On rearranging the equation in the form
z
1
+

z
2
+ (:Sct
3
,.-c,)z
3
+ (:Sal
1
x
1
) zl = O, (2:3)
it is seen that the coordinates of the vertex (y) will make the
coefficient of every coordinate z, vanish, hence:
THEORF:.\T I. The 1JOlar J)lnne of any poiut in space 'with 1egard
to ct quadric cone passes through the vertex. The polar plane of the
vefiex ilse{f is incleter miuate.
:\Ioreover, the polar plane of all points on the line joining any
point (z) to the vertex will coincide with the polat plane of (z),
since the coordinates of any point on the line joining the vertex
(!!)to the point (z) are of the form k
1
y, + kzZ,. On substituting
these values in (:.!3) and making nse of (21) we obtain (22) again.
In particular, if (z) lies on the surface, the whole liue (y) (z) is on
the surface; the polar plnne is now a tangent plane to the cone
along the whole generator passing t hrough (z). Hence:
'l'm:or:tm II. Every tanrtent plane to tlte cone ]Jasses through the
'l:ertex cmd touche:; the surfi.tce along a yenemtor.
lf the vettex of the cone is chosen as the vertex (0, 0, 0, 1) of
the tetrahedron of reference, then from (22), Cl
14
=au= a
3
l =a
41
=0,
hence the equation of the surface is illllependent of xl. Con-
versely, if the equation of a qnadric does not contain x
4
, then
A= 0 and the surface is a proper or composite cone with vertex
at (0, 0, 0, 1). The equation of any quadric cone with vertex at
(0, 0, 0, 1) is of the form
i, k= 1, 2, 3.
The equation of the tangent plane to 1\ at a point (z) is
(a
11
z
1
+ a
12
Z
2
+ a
13
z
3
)x
1
+(<t
21
z
1
+ a
2
zZ
2
+ ct
23
z
3
)x
2
+ + aaaz3):r, = 0.
ARTS. 121, 122] PROJECTION Ol<' A QUADRIC CONI<: 145
If a plane = 0 coincitlcs with this plane, then
a
11
Z
1
+ Ct
12
Z
2
+ Ct
13
Z
3
= luh

+ <t-.
1
zZz + a
23
z
3
= lu
2
,
a31Z1 + a3zZz + ct33Za = ltta,
u. = 0.
l\loreover, the point (z) must lie in t he plane ::Su,x, = 0, hence
:Srt ,z, = 0. If z., z
2
, z
3
, l are eli minated from these eq uations, t he
resulting equations are
1 Ctu fl12 Ct13 ll1
a 21 C122 a-za Uz
11, = o, = 0, (24)
cta1 Ctaz C133
1
13
tt
1
11
2
11
3
0
which defi ne the cone in plane coordinates.
If the ''ertex of the cone A= 0 is at t he point (k), then <l>(u)
= ("'i.k, ,t,)
2
= 0 ( Art. lOfl). [ f k, =1= 0, the seclion of A = 0 by
the pln.ur = 0 is a conic whose equation in plane coordinatt>s
is obtained by equating to zero the first minor of <]>(u) c.orrespond-
ing to ct.. . The fi rst minor of any element Ct;o of tl1e principal
diagoual equated to zero, together with <1>(11) = 0, wii!, if k, =1= 0,
define the ghen cone.
122. Projection of a quadric cone upon a plane. Giveu a point
0 on a cone JC, bnt not at its vertex. To project the cone from
0 upon a plane.,. not passing throug-h 0, <"Onneet e\ery point P on
f{ with 0. The point P' in which OJ> cnts .,. is called the Jli'Ojtc-
tion of P upon .,.. Let (I be t he generator of Kthrough 0, :tnd 0'
the point in which pierct>s .,., Let l be t he line of intersection
uf .,. and the tangent plane along 9 The point 0 on I{ tone-
"POnds to any point of l, and to 0' in .,. correspond a ll the points
With these exceptions there is one-to-one col'l'esponcl ence
hetween the points of .,. and of fl. A cune defined on either
will uniquely cletcrminr a curve on the othrr.
Let ]{ be defi ned by x
1
.1'
3
-.
2
2
=0,.,. by .1;
3
=0, arul 0:=(0, 0, 1, 0).
If P' =

0, the coordinates of P = (.1

1
, .1
2
, x
3
, are
seen, as in Art. 108, to he connected with those of P' by t he
equations
146
QUADRIC SURFACES
EXERCISES
1 . Show that
4 xl
2
+ 6 ,.,,.,.z + 8 + !l xa
2
+ 12 xax +4 x,
2
= 0
re}Jresents a cone. Find the coorclinates of its vertex.
2. Find a vahte of k sudJ that the equation
x,
2
- 5 X1X2 + G xz
2
+ 4 xa
2
- kxax + :r4
2
= 0
represents a cone.
[CHAP. X.
3 . Write the equations of the cone of Ex. 1 in plane coordinates.
4 . repl:wu; by :r, anrl interpret the resulting equations.
5. rrove t.hat if the 1 wo of iutersectiou of a quadric and a tangent
plane coincide, the surface is a cone.
6. What Jocu;; on the Nne 1\ has for its projection in ,. a conic:
(a) not passing 0'?
(b) through 0', not touching l?
(c) touching l at 0
1
?
7. State some properties of the projecti on upon ,. of a curve on K which
passes k times through 0, has k
1
branches at the vertex, and intersects g in n
addit.ional points.
CHAPTER XI
LINEAR SYSTEMS OF QUADRICS
In this chapter we shall discuss the equation of a quadric sur-
face under the assnm ption that the coefficients are linear functions
of one ot more parameters.
123. Pencil of quadrics. lf
A = = O, B = ::.b,kx,xk = 0
are the equations of two distinct quadric surfaces, the system
A- >..B = Ab,k):c,xk = 0, (1)
in which A is the parameter, is called a pencil of quadrics.
Every point which lies on both the given quadrics li es on every
quadric of Lhe pencil, for if the coortlinales of a point satisfy the
equations A = 0, n = 0, they also satisfy the eqnatiou A- AB = 0
for eYery Yalue of >...
'fhrough any point in space not lying on the intersection of
A= 0, B = 0 passes one ancl but one qna<llic of the pencil. If
0;), is the given point, its coordinates must satisfy the equation (1),
hence
A (y)- >..JJ (J;) = 0.
If this value of A is substituted in (1), we obtai n the equation
.Ll(}J)B- B(JJ)A = 0
of the quadric of the pencil (1) through the point (y).
124. The A-discriminant. The condition that a quadric
A- >..B = 0 of the pencil (1) is singular is that its discriminant
vanishes, that is,
fLu - 'Abu a,2- 'Ab,2 Cll3- >..b,3
Ab,.
I a,k- >..b,kl =
a,2- >..b,2 .\b22 ((23- >..b23 (124- >..b24
=0. (2)
((13- >..b,3 (/o:!3- >..b23 (t33- >..b33 (134- >..b34
a
14
-.\b
14
(124 - >..b24 (134 - ,\b:J.l (144 - ,\b
147
148 LlNEAR SYS'l'E.MS OF QUADRICS [CBAP. XI.
This detenninant will he calleu the >--discriminant. If it is iden-
tically zero, the pencil (1) will be called a singular pencil. If the
pencil is not singular, equation (2) may be written in the form
(3)
If A * 0, this equation is of the fourth degree in A. If A = 0,
the equation will still be considered to be of the fourth degree,
with one or more infinite roots. It follows at once from equation
(;{) that in any non-singular pencil of quadrics there are four
distinct or coincident singular quadrics. If in (3), A is put equal
to zero, A' results. But from (2), this is the discriminant of
A= 0. Similarly, A is the discriminant of B = 0. Let {3,k be
the cofactor of b
1
k in A. From (2) and. (3) we obtain
- -t 0 = <tuf3u + rt'!'!f3zz + ... + a34{334'
If 0 = 0, A = 0 is said to be apolar to B = 0. Similarly, if 0' = 0,
R= 0 is said to be apolar to A= 0. A geometric interpretation
of this property will be given later (Art. 149).
1 !!o. Invariant factors. If the eq nations of the quadrics of a
non-singular pencil are transformed by a linear substitution such
that A= 0 is transformed into A'= 0 antl R = 0 into B' = 0,
then A - >..B = 0 becomes A' - AB' = 0. :\[oreover, if T is the
determinant of the transformation of coordinates, then (Art. 104)
I a',k- >..&' ,k I = I a,k - Ab,k I
From this formula we have at once
I. If (A-A
1
)
4
is a factor of I a,k- Ab,k 1, it is also a
./tclor I a' ik - Ab' ik I an<l conversely.
Hence the numerical valnc and multiplicity of every root of
the A-disctiminant is invariant under any linear transformation
of coordinates. 1\[oreovet, by a proof similar to that of Theorem
li, Art. 10+, we obtain the following theorem :
Tm:OREM II. E1.ery1 sth minor o.f the A-discriminant
n linear .fnnclion of the sth minurs of the original A-cliscrimilta11l
an<l co1we1sely.
From the two theorems I and II we obtai n nt once
ARTS. 12-l, 125] NV FACTORS 149
III. If ("-- .\
1
)" is a ji11:tor all lite stlt minors of
1 a,k - Ab,k I, tit en it is also a fa clot of etll the stlt of
I et',1 - Ab',k I ami c01nersely.
Let (.\- .\
1
)" be a factor of the .\-discriminant,
(.\- .\
1
)' of all its first minors,
(.\ - .\
1
)k of all its seco11d minors, etc.,
k, being tl,e highest exponent of the power of (.\ - .\
1
) that
divides all the sth lllinors, and k, being the first exponent of the
set that is zero.
Let also
L1 = ko- k1, L2 = k1 - k2, ... , L, = k,-1 (4)
From Theorem III we have:
I\'. 'I'hf' exp1essions
(.\ - .\1)", (.\ - .\ly.,
ate iwlepe/l(le1t of the choice of the tetrulti'Clron of ?'eference.
These exp1essions fue called invariant factors or el ementary
uivisors to the base A- ,\1 of t hr .\-diSCJ'iminant.
'Ve shn.ll next pro\'e thf' following- theol'f'lll:
Tnl:otu:)l \'. The e.t'}J'"tent uf ,.,wit int'tll'iwlt .fuctor is at least
tlltit !I
Let
when> F(.\) is not divi sible by (A- A.,).
Then
where /(.\) is not tlivisibiP hy (A- ,\1) Bnt the ueJivative of
I a,k- ,\h,t ' with rt'S)H't'l tu A. 1na.y h<' cxptessf'tl as a liuear function
of the lirst. minors/ and is consequently tli,risiblc by (,\- .\
1
)k
at lf'ast.
If the tlemcnls nf a df't e& oniuant )ui,rdl are functions of a variallle. follows
frum the d clinitinn of ;l llerh ati\'e thai the lerhatii'C Of the tletermiuautaS tO tile
\'ariallle onay uc oxpressctl as the suno of ll'tcrminants of the form
I u'b d I+ I ub',ll + IHbc'<ll + 1 ubctl' 1.
iu which "'
1
the d!' rhath'P uf 11
1
etc.
H thesu tletermiuants art> iu terms or the columns which contain lbe
Jcri\'Utivc. it fulluWS that the U( tbc giveu LJoote rmiuaut iS e xpi'CSSiiJJe &.S
a lineal runctioll uf its litst miuuts.
150 LINEAR SYSTEMS OF QUADRICS [CHAP. XI.
Hence
The proof in the other cases may be obtained in a similar way.
126. The characteristic. It is now desitable to have a symbol
to indicate the anangement of the roots in a gi,en A-discrilllinant.
There may be one, two, three, or four distinct roots. If k
0
= 1 for
any root At, then Lt = 1, and no other L , n.ppears for that factor.
If k
0
= 2, then Lt may be 1 or 2, according as the same factor is
contained in all the first minors or not. If all the exponents L ,
associate<! with the same root are enclosed in parentheses (LIJ
L
2
, ), and all the sets for n.ll tho bases in brackets, the config-
uration is completely defined. This symbol is called the charac-
t eristic of the pencil (1). E.g., suppose
I a.,k- AU;k 1 =(A - At)
2
(A- A
2
)
2
,
and thn.t is also a factor of all the first minors, but that
A- At is not. The characteristic is [2 (11)]. if A- At is also a
factor of al l firs t minors so that L
1
= 1, L
2
= 1 to the base A- Au
the symbol has the form [ (11)(11)].
From (4) it is seen that
1
+ [,
2
+ ... + L, = k
0
, tlmt is, that the
sum of the exponents fot any root is equal to the multiplicity
of that root. Si nee tlw snm of the multi pi ici ties of all the roots
is equal to four, we ha\'e the following theorem:
TmoREM. 'l'he sum of the I!XJlOnents in the clturaclen'stic is
always eiJU<d to four.
EXERCISES
l
a
1
I2-XU
1
!2 II
1
13-XU
1
!31
1. Express the minor
1
HI
1
"' ofla
1
;k-Xb
1
;klinterms
(( 23 - 1\V 23 (! 33 - 1\V 33
of the second minor:; of 1 a,k- Xb,k I
2 . Find the factors and characteristic of each of the following
forms:
1- X 0 0 0 0 X 0 0
(a)
0 0 X 0
(b)
X 0 1 0
0 X 1 0 ' 0 1 0 X '
0 0 0 X 0 0 X
0 0 X 0 0 X 0 0
(c)
0 0 0 X
(cl)
X 0 I 0
X 0 0 1
;
0 1 X 0
0 X 1 0 0 0 0 1-X
ARTS. 125-130) QUADRICS WITH LIXE OJ.' VERTICES 151
127. Pencil of quadrics having a common vertex. lf the .\-dis-
critninant is identically zero, the discussion in Arts. 12-1-126 does
not apply. Iu case all the l}nadrics have a common vertex, we
may proceed as follows. If the common vertex is taken as
(0, 0, 0, 1), the \':triable will not appeat in the equation. We
then form the .\-discriminant of order tiuee of the equation in
x
3
If this discrimiuant is not identically zero, we deter-
mi ne its invariant factors and a characteristic such that the sum
of the exponents is three.
Similarl y, if the quadrics have a line of vertices in common, wo
form the .\-discriminant of order two, a nd a corresponding charac-
teristi c ; if the quadrics have a plane of vertices in common, the
.\-discriminant is of order one.
128. Classification of pencils of quadrics. The principles de-
veloped in the precedi ng Arti cl es will now be employed to classify
pencils of quadrics and to reduce their equations to the simplest
forms. When the equation of the pencil is given, the clJarac-
teristic is uni quely determined. It will be assumed that for any
gi,en pencil A- .\B = 0, the .\-discriminant has been calculated
and the form of its characteristic obtained. For convenience, the
cases in which .-1 = 0 and B = 0 coincide wil l be included in the
classification, although in this case A- .\IJ = 0 does uot constitute
a pencil as defined in Art. 123.
Since any t wo distinct quadrics of a pencil are sufficient to
denne the pencil, we shall always suppose that the quadric B = 0
is so chosen that the .\-discriminant has no infinite roots.
129. Quadrics havi ng a double plane in common. By taking the
plane fo1 x
1
= 0, the equation reduces to
.\1X1
2
- .\X1
2
= 0,
A = .\1X1
2
,
and the characteristic is [1].
130. Quadrics having a line of vertices in common. Let :t
1
= 0,
be the equations of the liue of vertices. Every quadric
consist:; of a pair of planes passing through this li ne, and the
equation of the pencil has the form
.A - .\B = a
11
X
1
2
+ 2

+ an:t
2
2
- .\(b
11
X
1
2
+ 2

+ b'l'fX-l)=O.
152 Ll:--rEAR SYSTE11S 01<' QUADRICS ICBAP. XI.
Three cases appeat :
(a) The .\-discriminant has two distinct roots .\
1
,
(b) The .\-discriminant has a double root .\
0
but not every first
minor vanishes for .\ = .\
1

(c) The .\-discriminant is of rank zero for .\ = .\
1

In case (a), A- .\
1
8 is a square and A- is another square.
I"et the tetrahedron of reference be so chosen that
A- .\
1
B = x,
2
, A- .\
2
8 =

If we solve these equations for A and B, we may, after a suitable
cha.nge of unit point, write .A, Bin the form
A= A
1
X
1
2
+ .\
2
x
2
2
= 0, B = X
1
2
+ X
2
2

In case (b) we have the relation
(a
11
b
22
- ar!b
11
)
2
= 4 (a
11
b
12
- a
12
b
11
)(a
12
b
22
-


whi<h is the condition that A= 0, B = 0 have a common factor.
By calling this common fartm .t
0
and the other factor of B = 0
(which is by hypothesis distinrt from the first) 2 we may put
A- .\
1
8 = xt
2
, B = 2 Xt!t.z
Solving for A, B, \\'e hase
A= xt
2
+ 2 .\
1
x
1
:t;, B = 2xt.-v
2

In case (c), we ha\e A - At B := 0, hence we may write at once
.\
1
(x
1
2
+ xl) -.\(x
1
2
+ x
2
2
) = 0.
The invariant factors are.\- .\
11
.\ - .\
1

In this case we have then the following types:
[ll]
[:!]
[(11)]
A= AtX.2 +
A= 2 .\
1
.-c
1
X-_t +

.iJ = .\,(x.
2
+ :rl),
B =xt2 +X-.t2,
B= 2 X.X
2
,
lJ= xt2 +x/.
131. Quadrics having a vertex in common. Let the common
vertex be taken so t,hat the equation of the pencil contains only
thrE'e variables, x
11
X-_t, x
3
It will first be assumed that the .\-dis-
criminant is not identically zero.
Suppose !a,t- Abk, I= 0 has a.t least one simple root .\
1
The
expression A- At B is the product of two distinct linear factors,
hence the quadric .A- .\
1
B = 0 consists of two distinct planes,
ARTS. 100, 13lj VERTI..:X 153
both passing through the point (0, 0, 0, 1). Let the line of inter-
section of the phtnes be taken for x
3
= 0, x
2
= 0, so that the ex-
pression A- >..,B does not contain x
1
It follows that
u - >..,bu = 0, a,2- >..,b,2 = o, a,3 - >..b,3 = 0.
By means of those relations au, a
12
, u
13
can be eliminated from
the >..-discriminant. The result may be writt.en in the form
I
bu(At - X) bn(X,- X) bu(X1 -X) I
I Uik - bil, I - b,t(X, - X) a22 - Abz: a23 - Ab23
bu(>., - X) a23 - Ab23 a33 - Ab33
Since >.., was assumed to be a simple root of I a;k- >..b;k 1, it follows
that bu =1= 0. The equation of the pE'ncil now Las the form
>..
1
(bux
1
2
+ 3

+ 3 b
13
X
1
.1:
3
) + +

+ 2

- >..(b
11
x
1
2
+ :2

+ 3

+ b
22
4
2
+ 2 b
23
X,X
3
+ b
33
x
3
2)= 0.
If we make the substitution
_ + l>1zX2 + b,3x3
Y
1
-X, ----,
bu
then replace y
11
y
2
, y
3
by .c
11
x
2
, x
3
, the equation of the pencil takes
the form
A
1
X
1
2
+ cf>(x
2
, x
3
)- >..(x
1
2
+ j(x.l, x
3
))= 0,
in which cf>(x
2
, and f(x
2
, x
3
) are homogeneous quadratic func-
tions of x
2
, x
3
'fhe abo,e transformation may be interpreted
geometrically as follows: Since bu =1= O, the quadric B = 0 does
not pass through the !JOint (1, 0, 0, 0). The polar plane
bu:t'
1
+ U
1
zXz +

= 0
of the point (1, 0, 0, 0) as to B is consequently not a tangent
plane to Bat this point. The transfotmation makes this polar
plane the new x
1
, changes the unit point, and leaves x
2
= 0, x
3
= 0
unchanged.
The expression cf>(x
2
, x
3
)- Aj(x.,., x
3
) may now be classified ac-
cording to the method of Art. 130, and the associated functions
of Xu x.,., x
3
are obtained by adding >..
1
x
1
2
to cf>(x
2
, a
3
), x
1
2
to j(x
2
x
3
).
suppose that I Ct;k - >..u;k I = 0 has no simple root. It has,
then, a triple root which we shall denote by>..,. If >.. -A
1
is not
a facto1 of all the first minors, the quaJric A- >..
1
B = 0 consists
154 OF QUADRICS [CHAP. XI.
of two distinct planes. Let the tetrahedron of reference be
chosen in such n. way that Lhese two planes are taken as = 0,
:xa = 0, so that the equation of the quadric has the form
A - >..,B = 2 (
23
- "A
1
b
23
)x
2
x
3
= 0,
wherein a
23
- >..,b
23
=1= 0, bnt
a
11
- >..
1
bu = 0, il.,b
22
= 0, a
33
- >..,b3
3
= 0, a
12
- >..,b
12
= 0,
!lu - il.,b,3 = 0, and
1 a;k - >.b,k 1 m bl2(11., - >.) - >.,) a?.\ - Mz3
I
bu(A, - A) b,!("A, - "A) bu("A, - A) I
b,,(>., - "A) a2.\ - b33(>., - X)
Since (>..- >..
1
)
3
is a factor of this determinant and a
23
- >..
1
&
23
=1= 0,
it that u
11
= O, aml b,
3
b
12
= 0, that is, either b
13
= 0 or
b
12
= 0. Since it is simply a matter of notation whi ch factor is
made to vanish, let b
13
= 0. Then b,
2
=I= 0, since la,k - >..u,
1
1 $0.
Geometrically, this means that the plane x
2
= 0 touches B = 0
along the 0, x
3
= 0. 'l'he plane x
3
= 0 interl:lerts the cone
B = 0 in the l ine :t'
2
= 0, x
3
= 0 and in one line. By a
fmther changfl of eoord i uates, if necessary, the tangent plane to
B = 0 along this second line may be taken for x, = 0.
We then have
but since
A= >..,B + 3((123-
we may, by a suitabl e choice of unit point, write the equation of
the pencil in the form
A - >..B = >.., (3 x,x
2
+ x
3
2
) + 2 Xz-'t'
3
- >..(2 X
1
X
2
+ xt) = 0.
I >.. - >.., is also a factor of all the first, minors of the >..-di:;-
criminant, but not of all its serond minors, .:1- >..,B is a square
and represents a plane counted twice. This plane may be chosen
for Xz = 0 so that
A ->..,B =(a
22
- >..,bzz)xt
Since (>..- >..
1
)
3
is a factor of the >..-discri 111inant, we must also
have
ART. 131] SIKGLR \'ER1'EX 155
Geometrically, this condition expreRst>s that x
2
= 0 is a tangent
plane to the cone B = 0. \Ye may now write
B = 2

+ b
33
x
3
2
, A= >..
1
B + ( a::'l -

Hence, by a sui table choice of unit point, the equation of t.he
pencil may be reduced to
.\
1
(2 l'
1
.l'
2
+ xl) + - .\(!.l + .!'
3
2
) = 0.
If .\- .\
1
is also a factor of all the second of I ((ik- Ali;k
the equation of B = 0 is a mnltiplc of that of A= 0 and the equa-
tion of t he pencil may be written in t he form
+ .l'l + ,!'32)- >..(l} + .1:22 + .r32)= 0.
1'\'e have thus far supposed, in this Article, that the .\-discrimi-
nant did not ,.:UJish iclcntically. Jt may happen tl1at the deter-
minant ltt
1
k- Ab,k l is itlentirally zero even though the quadrics of
the pen('il do not ha,e a line of vertices in common. In t his case
every quadric of the pencil consists of a vair of planes. Let
.A = </>(X
11
B = .f(x
2
, x
3
). Si nce !a,k - Ab,kl is idt>ntieally zero,
it follows that
au(b22b33- b2a2)= 0, b3J{aurt22- C!a22) = 0,
and hence that a
11
= 0, b
33
= 0, as otherwise the quadrics would
ha, e a line of vert-ices in tommon, contrary to hypothesis.
By au obdons change of coonliuatcs, we may write t he equa-
tion of the prncil in the form 2 .r
1
.r.
2
- >..2 :r
2
x
3
= 0. This is called
the singula.1 case in three variables. Jts chararteri stic will be
denoted by the symbol ! 0 !- Collecting a.ll t he p1ccetling resnl ts
of the present Article, we have the following types of pencils of
quadrics with a. common vertex.
[111]

[1(11) J
[3]
[ (:31)]
[ (111 )]
Pi
AaXa
1
+ A1-l'2
2
+ A3.vl
+ 2 + x22
,\ . la2 + >..2(x:2 + :l'a'r)
!J'a.l'2 + X3
2
) + 2 :l':-1'3
,\J(!.l .t'
1
.t.J + x
3
2
) + x
2
2
>... (x.2 + :t22 + x32)
Alxax2
xl2 + x:2 + Xa2
Xa2 + X:!Xa
x.2+Xa2 +xl
2 Xa:t'2 +x3
2
+ .tl
:1.'a2 + .1:22 + Xat
2 :t'2''l:3
156
LINEAR SYSTEMS OF QUADRICS [CHAP. XI.
EXERCISES
1 . Determine the invariant factors for each pencil in the above table.
2. Determine the nature of the locus A == 0, B ::: 0 for each pencil in the
:tbove table.
3. Find the invariant factors and the characteristic oi the pencils of
cones defined by
( A= !1 x
1
Z + !1 ?'2
2
+ 4 7zXa - 2 :r1.r3- 6 x1:r2 = 0,
B = ,) :r
1
2
+ 8 :r.z
2
- 2 G :r1xa - H XtXz = 0.
(b) A= 5 x
1
2
+ 3 Xz
2
+ 2 ::ra
2
+ 4 Xz:rs- 2 x.:ra + 2 :r1xz = 0,
B = 0 :r
1
2 - x
2
Z + xa
2
-I :rz:r3 + 14 XtXa + 42 Xt:l:z = 0.
(c) A = 5 :r
1
2
- 5 Xz
2
+ xa
2
+ 0 XzXa + 10 XtXa - 4 :Tt Xz = 0,
B = 10 x
1
2
+ 2 :r
2
2 + 10 :r
4
2
- 10 + 24 XJXa- 16 x
1
x
2
= 0.
(!l) A = 2 Xt
2
+ 2 Xz
2
- 2 :TzX3- 2 YtX3 = 0,
R = Xt
2
+3 :rz
2
+ xa
2
- 4 xz:ra - 2 Xt:l"a = 0.
4. Find the form of the intersection of A= 0, B = 0 in of the pencils
of Ex. 3.
5. Write the equations of each of the peucils iu Ex. 3 in the reduced form.
132. Quadrics having no vertex in common. As in the preced-
ing case, we shall suppose, except when the contrary is statecl,
that I a.,k - >..b,k
1
is uot identically zero. If (A. - >..
1
) is a simple
factor of the A.-discriminant, then A- A.
1
B = 0 is t he eqnation of
a cone. Ry choosing its vet'tex as (1, O, 0, 0) and proceeding
exactly as in Art. 131, the eqnation may be reduced to the fom1
A
1
X
1
2
+ cp(Xz, x
3
, X
4
)- A(X.
2
+ J(xz, x
3
, X
4
)) = 0.
Hy this process the variable x
1
has been separated and the func-
tions x
3
, x
4
), f(x
2
,

x
4
) can he reduced by the methods of
Art. 131, not including tl1e singular c:tse.
The only new cases that arise arc those in which the roots of
a;
4
- A.b;k I = 0 are equn.l in pairs or in which all four are equal.
Consider ihst the case in whi ch there are two distinct double
roots A.
1
and >..
2
, neither of which is n. 1oot of all the first minors of
the A.-discriminant. The quadrics A- >..
1
B = 0, A- >._zB = 0 am
cones having tlistinct ve1ticcs. Let the vertex of the first be
taken as (0, 0, 0, 1) and tha.t of the second as (0, 0, 1, 0). The
equation of the former does not contn.in x
4
Hence, we ha \'e
nu- = 0, "2 - >-.b2 = 0, n3 - >..ba
4
= O,

A
1
U
44
= 0.
ARTS. 131, 132] NO VERTEX IN COMMON 157
When those values of a
14
are substitutecl in I a,k->..bik I = 0, X - X
1
is seen to be a factor. The condition that {X- X
1
)2 is a factor is
that either b
41
= 0 or that X - X
1
is a factor of the minor cor-
responding to a."- Xb.w Hut in the latter case X - X
1
is a factor
of all the first minors, contrary to hypothesis, hence u
44
= 0.
Proceeding in the same way with the factor X-X
2
, it is seen that
a
13
- X
2
b
13
= 0, a
23
-

= 0, a
33
- X.)J
33
= 0, (t
31
- = 0
and also that b
33
= 0. Hence the vertices of both cones lie on the
quadric B = 0. Let the tangent plane to B = 0 at (0, 0, O, 1) be
taken as :1:-.! = 0, and the tangent plaue to R = 0 at (0, 0, 1, 0) be
taken as x. = 0. Since B = 0 is non-singular, b
13
in the trans-
formed equation does not vauish, hence the plane x
2
= 0 intersects
the cone A- X
1
B = 0 in the line X
1
= x
2
= 0 and in another line.
Let the tangent plane along this secoud line be taken as x
3
= 0;
that is, make the transformation
Y1 = X., Yz = :r2,
2 b
13
{Az- X
1
)y
3
= (a
11
- X
1
b
11
)x
1
+ 2 (a
12
- X
1
b
12
)Xz + 2 b
13
(X
2
- X
1
)x
3
,
Y4=x4.
The equation of the cone has now tlte form
A - " n = -

+ ( a,a- x.b.a)X.Xa = 0.
Similarly, the plane .l-. = 0 intersects the cone A- X
2
1J = 0 in
the line .t
1
= 0, .1.
2
= 0 and in another line. )[ake a further trans.
formation by choosing the tangent pl:l.ne to A- AzB = 0 along
this line for the new x
4
, thus
Y = :t., ?h = Xz, ?Ia = Xa,
2 b
2
.{X
1
- Az)Y
4
=(c'2- A./,
12
).t. + (tt
22
-

+ 2 b
2
.{X
1
- Az)X
4

The equation of the second cone now has the form
A- >.zB = (a
11
- .\
2
b
11
)X
1
2
+ 2(n
24
- V'
24
)l'z-1;
4
= 0.
By a suitable choice of unit point the equation of the pencil may
be red need to
x.(x.
2
+ 2 + Xz(.tl +:? :r.Xa) - X(x.
2
+

+2 x . ra + X:!'t'a) = 0.
If the invariant factors are {X- x.), (X- XI), (A- \2)2, the quad-
ric A- X
1
B = 0 is a pair of d istj net pla11es ancl as before A - X
2
B =0
is a cone having its vertex on the qnacbic B = 0. Let the line of
158 LIN EAR SYSTEMS OF QUADRICS [CRAP. XI.
intersection of the two planes of A- A
1
B = 0 be taken as Xa = O,
x
4
= 0, and l et the vertex of A- AzB = 0 be at (0, 0, 1, 0) as before.
Since this vertex I ies on A- A
2
B = 0 and on B = 0, it lies on every
quadri<: of t he pencil, in particular, therefore, on A- >..
1
B = 0.
Thus, one of the planes of the pait constituting A- A.sB= 0 is the
plane x
4
= 0. The other may be taken as Xa = 0 so that
A->.., B = ( a34- A.slla4):l:aX4 = 0.
The pl:tne x
4
= 0 is not tangent to A- A.
2
B = 0, si nee otherwise the
discrimi"nant j nik - >..bikl would vani sh identi cally. Hence we may
choose for Xs = 0, and x
2
= 0 any pair of planes conjugate to each
other and each conjngate to x.
1
= 0 as to the cone A ->..
2
B = 0. The
equati on of the cone A- = 0 is now
A - >vlB= (au- A./Ju).ts
2
+ (a.u-

= 0.
these two eqna.tions we may reduce the eqnation of the pen-
cil to the form
,\z.t
3
X
4
+ A.s (.r
1
2
+ X2
2
+ x/) - .l:J.t:.s+''t
2
+ .1:2
2
+xl) = 0.
If(>..- >..
2
) is also a factor of all the first minors, so that the in-
vai ant factors are (>..- >..
1
), (A- A
1
), (>..- A
2
), (A- >..
2
), the quadrics
A - A
1
B = 0 and A - A.
2
B = 0 both consist of non-coincident planes.
These four plnnes do not. all pass through a common point, since
in that case all the quadrics of the pencil would have a common
vertex at t hat point, contmry to the hypothesis. We may conse-
quently take
A- >..,B = (((aa- A.sf.JJa).tl + (nH - A.sf.J44).t:/ = 0,
A- =(au- A.
2
bu).r
1
2
+ (a
22
-

= 0.
By a suitable choice of unit point the equation of the pencil as-
sumes the form
As (X
1
2
+ :tz
2
) + >..z(x
3
2
+ xl)- >..(xs
2
+ :t'l + Xa
2
+ x,
2
)= 0.
The remai ning c:tses to consider (l.re those in whi ch lcc,k- >..bikl
has a fomfoiLl factot (>.. - >..,)
4
Suppose first that >..- A.s is not a
factor of all I he first mino1s. The 1ptatlri(: A- A
1
B = 0 is a cone
with w rtrx on U = 0. Its vertex may he taken as (1, 0, 0, 0), and
the tangrnt plane t.o B = 0 at t his point as x
2
= 0. Since A-
>..
1
B = 0 is a cone with vertPx at (1, 0, 0, 0) we have
au- ,\sun = 0, a
12
- A.
1
b
12
= 0, " sa- A.sbsa = 0,

=0.
ART. 132) NO VERTEX IN CO;\ll\lON 159
Since (1, 0, 0, 0) lies on B = 0, we have b
11
= 0, and since the tan-
gent plane at (1, O, 0, 0) is x
2
= 0, it follows that b
13
= O, b
14
= 0.
The A.-discriminant now has the form
0 btz(..\.- .\) 0 0
o.2( ..\.- ..\. ) Clz2- '..\b'l:!. a23 -.\b2a a2 - 'Abu
0
a23- >..u2a a ~ - >..baa Cla - .\ba.t
0 '(44- .\b24
aa- .\us
a
44
- '..\b.
14
Since (>..- .\
1
)
4
is a factor and b
12
=1= 0, it follows that
l
bss b.,, (.\_ '..\
1
) 2 = l'lu- .\baa au- .\IJa,,.
IJ
3
, b,,
1
a
34
- '..\u
3
, a.,- '..\b
44
The section of the pencil of quadrics A- .\B = 0 by the plane
x
2
= 0 is the pencil of composite conics
(luX:J
2
+ a<tx} + 2 <:La
4
XaX4- .\(buXa
2
+ u ... r;} + 2 ba,XaX,)= 0, X2 = 0.
'fhe characteristic of this pencil of composite conics is [ 2]; it con-
sists (Art. 130) of pairs of lines through (1, 0, 0, 0) all of which
have one line g in common. The plane x
2
= 0 cuts the cone A-
.\
1
1J = 0 in the line g counted twice, and g is defined by one of the
factors of b
33
xl + 2 u
3
,x
3
x, + IJ
44
.t
4
2
, since it is common to all the conies
of the pencil. The tangent plane x
2
= 0 to B = 0 therefore con-
tains the line g and another line gt. Through the line !t', which
passes through the vertex of the cone A -"..\
1
B= 0, can be drawn
two tangent planes to Lhe cone. One of them is . ~ = 0. Choose
the other for .r;
3
= 0. The plane x
3
= 0 will touch the cone A-
.\
1
/3 = 0 along a line g". The plane containing the two gE>nerators
y, y" of the cone is next chosen as x, = 0. The equation of the
cone A - '..\
1
B = 0 now has lhe form
A - >...B = 2((1
23
- .\
1
b
23
)x
2
:t'
3
+(a,,- '..\
1
b,
4
)x
1
2
= 0.
'l'he plane :r
3
= ' 0 contains the generator g' of B=O, hence it is
tangent lo B = 0, and intersects B = 0 in a line (It of the other
regulus. The plane x,= 0 contains the generator u of B = 0,
hence meets the surface in another line (f
2
'l 'he lines ff, g' are of
opposite systems, hence g
1
, fh belong to different reguli and inter-
sect. The plane of g
11
g
2
may be taken as the plane x
1
= 0. The
qnadric B = 0 now has the equation
B = 2 b
1
zX
1
X
2
+ 2 b
34
X
3
x, = 0.
160
LlNEAR SYSTEMS QUADRWS [CHAP. XI.
By means of this equation and the equation of the cone A- >..
1
B
= 0 it is seen t hat the equation of the pencil may be reduced, by
a s uitable choice of unit point, to
>..
1
(2 x
1
x
2
+ 2

+ 2 .rt--u
3
+

>..(2

+ 2 x
3
x
4
) = 0.
Now suppose >.. - >..
1
is also a factor of all the first minors, but
not of all the second minors. 'l'he surfn.ce A- >..
1
B = 0 consists
of a pair of planes which may be taken for x
3
= 0 and x
4
= 0, so
that
A- >..
1
13 = 2( a
3
.
1
- >..
1
b
34
)X:Jx, = 0,
and A- >..B = 2(a
34
- A
1
b
3
.
1
)x
3
x
4
+ (>..
1
- >..) 13.
If the >..-discriminant is calculated and the factor (>..- >..
1
)2 re-
moved, it. is seen that i u order for I ct;k- >..b,k I to luwe the further
factor (>..- >..
1
)
2
the expression bub
22
- b
12
2
must vanish. Hence
b
11
.v
1
2
+ 2 b
12
.c
1
.1'
2
+ b
22
x.} either vanishes identically, or is a square
of a linea expression.
In t he first b
11
= 0, b
12
= 0, b
22
= 0, so t hat the line x
3
= 0,
.1
1
= 0 li es on the quadric B = 0. The plane x
3
= 0 passes through
this line antl intersects IJ = 0 in a second line g'. Similarly,
x, = 0 intersects JJ = 0 in :'l'
3
= 0 and in another line g". Anothet
tangent plane t hrough g' may be taken as x
2
= 0, and the plane of
g" and the second line in = 0 as x
1
= 0. 'l'he equation of B = 0
is
B = 2 bi3XtX3 + 2 = 0,
n.nd the equation of t he pencil may be reduced to the form
>..
1
(2 x
1
x
3
+ 2 :-t
2
x
4
) + 2

- ..\. (2 x
1
x
3
+ 2 = 0.
In case u
11
x
1
2
+ 2

+ b'!'!xi is n. square, not identically zero,
t be line x
3
= 0, = 0 tonC'hes /3 = 0 but does not lie on it. Let
the point of tangency be L1.ken as (0, 1, 0, 0) so that b
12
= 0,
b
22
= 0. If we now remove the factor (>..- >..
1
)
3
from the >..-dis-
criminant and then put>.. equal to >..
1
, the result is

This expressiou is equal to zero, since (>..- .\
1
)
4
is a f:l.ctor of the
>..-discriminant. But a
3

1
- >../1
34
* 0, as otherwise A would be
itlPn t ical with lJ j be nee either = 0 0 1' b,., = 0. Let the notn-
t.iou be sunh that b
24
= 0. '!'hen the section of tho qnarlric 13 = 0
by tlw plane X:J = 0 consists of t wo lines through (0, 1, 0, 0).
ART. 1321 NO VERTEX IX ] 6l
Let L be the harmonic conjugate of the 0, :r
4
= 0 wit!;
regard to these two lines, and let P be any point on the conic
!1'
4
= 0, B = 0. If the plane determined by P and Lis chosen for
a, = 0 and tangent plane to B = 0 at Pis taken 0, the
equation of B = 0 becomes
B =

+ 2

+

= 0,
and the equation of the pencil has the form
A,(.c,z + ,,.tz + X-.!l:a) +:! .l'a.l"s -A (.,.,2 + .r/ + 2 x2.r3) = 0 .
.Now suppose that A- A., is a factor of all the second minors,
but not of all the third minors, so that .A- A
1
B = 0 is a plane
r.ounted twice. Let this plane be ta.ken as .r
4
= 0.
A- A
1
/3 = (cr.
11
- A.
1
I1
4
J).1/ = 0.
By substituting these values in the A.-discriminant, it is seen that
the determinant jb,/J
2
/J
33
j must also vanish if A.- ,\
1
is to he
a fourfold root. This means that the section of the quadric
B = 0 by the plane x
4
= 0 consist.s of two lines, hence that x, = 0
is a tangent plane to B = 0. Let planes through these two lines
be taken as a,= 0, .r
2
= 0. The remaining generators in x, = 0
and in = 0 belong to opposite reguli and therefore intersect.
The plane determined by them is now to be taken as x
3
= 0. The
equation of B = 0 is 2

+ 2 b
3
,.1J
3
X.
1
= 0, hence the C1p1ation
of the pencil may be reduced to the form
A
1
(2 X
1
X
2
+ :t
3
.r
4
) + X
1
2
- X
1
X
2
+ 3 :oax
4
) = 0.
If finally A- A.
1
is a factor of all the third minors, the two
equations A= 0, B = 0 differ only by a constant facto1. If B = 0
is reduced to the sum of squares by refening it to any self-polar
tetrahedron, the equation of the pencil becomes
A
1
(x
1
2
+ x
2
2
+ .tl + x.
2
)- A X
1
2
+ x/ +x
3
2
+ .t})= 0.
Thus far it has been assumed that the A<liscriminantdid not iden-
tically vanish. X ow suppose I cr;k - A.b,kl = 0 so that all the qu:Hl-
rics of the penci l are singular. By hypothesis they do not havE'
a common Yertex. In the singular pe11cil two distinct composite
quadrics cannot exist, f01, if A=O, B=O were composite, we
could choose A = 2 x
1
x
2
, B = 2

since the quadrics of the pencil

162 LINEAR SYSTEMS OF QUADRICS (CHAP. XI.
do not have a common vertex. But the >..-discrimiuant of the
pencil A- >..B = 0 is not identically zero, contrary to hypothesis,
hence the pencil does uot coutain two distinct composite quadrics.
The quadrics .A= o, B = 0 mn.y therefore be chosen as cones.
Let Lbe vertex of A= 0 be taken as (0, 0, O, 1) and the vertex of
B=O as (1, 0, 0, 0).
T"et !/, g' be generators of A= 0, B = 0 which intersect, but
s nch that the tangent pl:wes along each of them does not pass
through the vertex of the other cone. The plane g, g' can be
taken as .'l'
3
= 0, the tn.ngcnt plane to A= 0 nJong 'J as x
1
= 0, and
t he tangent plaue to B = 0 along a' as

= 0.
The equations of the si ngu Jar q narlrics A= 0, B = 0 are now of
the form
A= a
11
.-ct
2
+ 2 a
12
x.x
2
+ 2 a.
3
x.x3 +

= 0,
B = b
3
:r,;/ + 2 b
24
.Vz.1;
1
+ 2 b
34
x
3
X
4
+ b
44
X.
1
2
= O,
and the >..-discriminant is
Ctu a.z aa
0
I a,k- >..b,. I =
a.z
0 0 - Ab24
a13 0 (/33- >..b3J ->..b34
0 - - Aa4 :- >..0,11
Rince t his expression Yanisbes identi call y, the coefficient of each
power of >..must be equal to zero. 'rhese conditions are a
11
= 0,
= 0, a
12
b
34
- u
2
p'13 = 0. The ln.st condition expresses that the
planes a
12
.l'
2
+ a
13
.t
3
= 0 and b
24
.t
2
+ ba.
1
X
3
= 0 are coincident. By
transfol'lning the equati on of this plane to = 0, the equation of
the pencil reduces to
2 .t
1
Xz + axl- >..(2

+ x
3
2
) = 0.
Thi s case is called the si ngular case in four variables. The char-
actel'istic will be denoted by the symbol [ !3! 1].
The determination of the invariaut factors and the form of the
r. hararieristic f or each of the above ]l'ucils is ldt as an exercise
fo1 the student. The properties of the Mrve of intersection will
be developrd in Chapter XI I r, but in each case the curve is
described in the following table for reference. The table includes
ouly those forms which do not have common double point.
ARTS. 132, 133] FORl\fS OF OF QUADRICS 163
133. Forms of pencils of quadrics.

lST'lC
( 1111]
[112]
[11(11)]
[13]
[1(21)]
(1(111)]

[2(11)]
FORYS OP A B
A = .\1x1
2
+

+ .\3X3
2
+ .\4xl
B = xl2 + :tl + x3z + x4z
A = .\lxl2 + '4t22 + 2 + x32
B = x
1
2
+ .1'
2
2
+ :.? -taX
4
A= .\1x1
2
+ .\t>Tl + .\3(.tl + xl)
B = :t
1
2
+ X
2
2
+ .l'
3
2
+ xl
A = .\
1
x
1
2
+ .\
2
(2

+ X
4
2
) + 2 X:s-1'
4
B = x1
2
+ 2 'Xz:tJ + xl
A = .\
1
X
1
2
+ + x.Z) + .r
3
2
B =

+

A= .\1x1
2
+ ,\z(xl + %
2
+ xl)
B = x
1
2
+xl + x
3
2
+ xl
A= .\
1
(x
1
2
+ 2 x
2
x
4
) +
.\z(.tl +:! .l.'t.1'3)
B =x
1
2
+ .c
2
2
+:? :>'
2
.1'
4
+ 2 X
1
X
3
I
A = .\
1
(x1
2
+ x
2
2
+ x
4
2
) + 2

B= x
1
2
+ x
2
2
+ 'J'l +
[(11)(11)] .A =>-.(x.
2
+ x}) + ,\z(xl + 'J"l)
B = x.2 + x2z + x32 + :tl
CURYE OF
en A= 0 A so B=z u
A general space
quartic of the
fi rst species.
A nodal quar-
tic.
Two conics
which interscctat
two distinct
voints.
A cuspitlal
'l n:ntic.
Two con i cs
which tonch each
other.
A conic counted
twice. At each
point of this conic
the quadrics are
tangent.
A generator and
aspacccnbic. The
generator aml the
cubic intersect in
distinct points.
T"o i n tersr.ct-
ing generators,
and a cunic which
intersects each
generator. The
three points of in-
tersectiou are dis-
tinct.
l'out gtnf'rators
whi(h intersect at
font points.
LINEAR SYSTEMS Of' QUADRICS (CRAP. XI.
C'llAftAO
TKISTIC
[4)
[(22)]
[(31)]
Sl)tiLlrl ED OF
A " .B
A = AI (2 + 2 %X,)
+ + x,
2
B = X
1
:t'
2
+ 2 x
3
x,
A= x
1
x
3
+ 2 + 2 Xa-'t
4
B = 2 X
1
x! + Z:t
2
x,
A= "-1(.1'1
2
+ :t:/ + 2 X-.!X3)
+ 2 ,l)3:t,
B = x
1
2
+xl +

[(211)] A = A
1
(2:r
1
:t:.!+

B = 2 :c
1
X-z + 2x
3
x,
[(1111)] A= A. t<.c
1
2
+

+ x
3
2
+ xl)
n = xl2+ ,1'22+ .1'32 + ';l:l
J] .tf =:] + ((,\:3
2
B = 2 + x
3
2
EXERCISES
C t.'R \ ' E OF I NTfmSr.CTtON OP
.A=O A>ll> B=O
A generator and a space
cubic. 'l'he generator
touches the cubic.
Three generators, one
counted twice. This
generator intersects
each of t he others.
Two intersecting gener-
ators and a conic
which touches t he
plane of t he generators
at t hcit point of inter-
section.
Two intersecting gener-
ators each counted
twice. 'l ' he quadri cs
touch at each point of
each generator.
'rhe rpmdrics coincide.
A conic and a generator
counted twice. The
vertices of the cones
all lie on t his geu-
CJ"atOI.
1. Derive the iuv:uiant factors vf each of the above systetns of quadrics.
2. Fintl the equations of each conic a ucl each rcctili ue:\l' generator of in-
of the quadrics of the above pencils.
3. Determine the invariant factors; find the equations of the cun e of
intersrction, autl write the equations in the reduced form of the pencils
detrrmined by
(11)
.tl = x1
2
- + + 2.r,z + UX3X = 0,
1: =

8 x
4
2-

-2xaX =0.
AnTs. 133, 13 -1 ) C01\'J UGAT E TO A 1G5
(b)
(c)
(cl)
A = .t1
2
+ .r2
2
+ 4 +xi!+ 4 ' :Xt + u >"2.<:a + 1 x'a = 0,
IJ = x22 + 3.:
3
2
+ + 2 X1J:a + 2 = 0 .
I = 2 . .,.a
2
+ 2x.
2
+ 2 ''"t - .ca = 0,
B :::: 4 Y12 - :r22 + .ra'
1
+ ::!

+ 2 x
1
.r2 + 2 X
1
:>'
4
+ 4 XaX
4
= 0 .
.1l = 3 .r1
2
+ 2 :r2
2
- :ra
2
- x,
2
+ 4 :r1xz- 2 xzx = 0,
lJ = :3

:r
2
2
- x
3
2- x.
2
+

:l xa.Y -

3x .r. = 0.
4. To what t,ype does a pl:lllCil of concentric spheres belong? A pencil of
tangent spheres?
134. Line conjugate to a paint. The eq nation of t he polar
plane of a poiut (y) with respect to any quadric of the pencil (1) is
"J.a,ky,xk - = 0.
As A varies, this system dclincs a pencil of planes (Art. 24). The
axis of the peucil, namely the line
'S. a,klf/l.'k = O, 'S.&,ky,.1k = 0
is said to be conjugate to the point(!!) as to the pencil of quadrics.
Let (J;) describe a line, two points of which are (!/) and (y'').
It is requited to find the locus of the conj ugate line. Since
?/; = P.?/'; + p.z?/
11
, i = 1, :.?, 3, 4
(Art. 95), the line conjugate to (Ji) "is, by definition,
.
As (y) describes the line joining (!/) to (y") the ratio p.
1
: p.
2
takes
all possible values. lf between these equations p.
1
: P-! is elimi-
nated, the resulting equation defines the qnadrir surface
From the 1nethod of development it follows (Art. 1 H) ) 1 hat. all the
lines of the system belong to one regulus (Art.. 11.3).
The volar planes, with respect to a given quatlric of the pencil,
of two fixed points (];'), (y'') on the given line intersect in the line
'S((,ty';xk- = 0, :i,(l,s}/';xt - A'SO;k!/',.tk = 0.
If between these equations A is elituinated, t he resnlting equation
defines the same quadric (.i). From Art. llil it follows that this
second system of lines constitutes the othet regulus on the smface.
HiG LTNBAR SYSTEI\IS OF QUADRICS [CHAP. XI.
135. Equation of the pencil in plane coordinates. Let A->..B= 0
be the equal ion of a not1-singul ar penci l of quadrics. The
ec1uation
au- >..lJu ((12 - ((13- >..b,3 (t,. - >..b,. ti l
a,2- >..&, 2 >..&22 ((.'!3- >..b24
112
Ctu- ,\/;13 ((23- AlJ23 CLs3 - >..b33 aa. - >..&34
1t
3 =0 (6)
((14- >..b,4 t1
24
- >..b
24
<t34- >..b34 ((44- >..b 114
1t, 'l/2 113 114 0
expresses the condition t hat the section of a quadric of the pencil
by a plane (u) is composite (Art. 10G). For a given value>.., of>..,
(G) is t1t e equation of t he qna<hic A->.., JJ = 0 in plane coordi-
nates, if it is non-singular. If A ->.., B = 0 is n. cone, (6) is the
eq nation of its vertex counted twice. 1f A - >.., B = 0 is composite,
(6) vanishes identicnJly.
Equation (G) is call ed the equation of the pencil in plane coor-
dinat::)S. Ana.ngrtl i n powers of >.., it is of the form
If <1>
1
(11) * 0, the equation is of the thitd d('gree in >... When (7)
is not identically zero, it will be said to be :t cubic in any case,
even if it has one or more i ntlnite roots. Hence we have the fol-
lowing t heorem:
'l'tt EOIH: .IL Eer1!J ];luur intersects three or
quaclrics of CL tiUII - :<iuy"l't,. JICII.cil i11 composite
The cocttlcient of each power of >.. in (7) is homogeneous :md of
the second degtee in 11
1
, u
2
, u
3
, 11
4
(if it is not identi cally zero), hence,
when equn.tell t.o zero, it llefinrs a. quadri c in plane coordinates.
Since the penril is non-singulrtr, we ma.y, wi t hout loss of general-
ity, assume that the quadrics A =0, antllJ = 0 arc non-singular
(Art. lZS). The equation <f>
2
( I()= 0 is seen, by putting>..= 0 in
(6), to be the equn.tion of A = 0 in plane coordinates. An analo-
gous statement holds for <t>, (u) = 0 and B = 0. The geometric
meaning of the other coefficients \vill bo discussed later (Art.
149).
ARTS. 135. 136) BIDJ' DLE OF QUADRICS 167
EXERCISES
l . "rite the equation in plane coordinates of the pencil of quadrics
:r,:- + .r3
2
+ 3 Xt
2
- 6 x,:r., + 4 .a.c4- i\(2 X2X4-+ :r1
2
+

+ Xa
2
) = 0.
2. Determine the equations o[ t.he three quadrics of the pencil of Ex. 1
which touch the plane x
4
= 0.
3 . Determiue equation (i) for the pencil
a(2 XtX2 + 2 X3X4) + i\(:! + X3:r4) = 0.
Show that (7) vanishes identically fnr each of the planes Xt = 0, xa = 0,
x. = 0, and interpret the fact gcomttrically.
13H. Bundle of quadrics. T C .l = ;1:;.1k = 0, n = .. rk = 0,
0 = = 0 are three given C[llllllrits which tlo not hclong to
the same pencil, the system deli ned by the equation
..\
1
A + ..\JJ + ..\
3
0 = 0, (8)
in which ..\
3
are parameters, is called a bundle of quadrics.
The three given quadrics A= 0, lJ = 0, 0= 0 intersect in at least
eight distinct or coi ncident points,* en.ch of whic.h pass
all the quadrics of the bundle. These eight points cannot be
taken at muclom, for in ordP.t' that a qnadric shall pass through
eight given points, the coordinates of each point mnst satisfy its
thus giving rise to eight liucat homogcnPous ccptations
among the coeffirients in the equation of the qnadric. Tf the
Pight ght>n points are chosen arbitlarily, the!>e eight CC[n:ttious n.re
inuependeut and the of quadrics determiuerl by tl1em is a
pencil.
[ t is seen that sr,cn given arbitr:trily chosen points determine
a bundle of quadrics passing them. :-lince all the ()liadrics
of the bundle have at lrast one fixed eighth point in common, we
ha,e the following theorem:
I. All the sm:fetCCR which 1'(tg.s fhl'Olt!lh set,en
illclPpemleut in though rt ji:cecl eighth point.
"'Three algebraic sufatcs whose equations are of m, n, p, l'espectively,
intersect in at least mnp distinrt or cniucicleut points. If they have more than
mnp points in comtiiOD, they hrwe one or more curves in rommon. For a
proof of this theorem see Salmon: Lessons Introd uctory to illodcrn Higher
Algebra, Arts. 73, 78. Wo shall assume the truth of tbis theorem.
168
LINEAR SYSTF-l\IS OF QUADRICS [CBAP. XI.
These points are eallecl eight associated points. Jf t.he coordi-
natrs of any fixe!l ;ubit rarily chosen point (y) are substituted in
(R), the condition that (y) li es 0 11 the ![Uadric fumishes one linear
relation :tmong the A; Hence t hrough (y) pass all t he quadrics
of a pencil and therefore a proper or composite quartic curve
lying on every quadri 0 of the prncil. This qnnrtic curYe passes
through the eight associated poiuts of the bundle.
lf (!!) is chnsen on t he line joiuing n.uy two of the eight asso-
ciated points, every CJimtlriu of tlt e pencil passi11g through it wi ll
contai u the whole line, since each quadric of the pencil contains
t hree points on Lhc line (Art. Gi), Th. I 1). The resitlual intersec-
tion is a proper or co111posi te cubi c cmve passing through the
other six of the associated points and cutting the given line in
two points.
137. Representation of the quadrics of a bundle by points of a
plane. Let .\
1
, .\
2
, .\
3
be regarded a:; the coordinates of a point in a
plane, which we shall call t he .\-plane. To each poi rtt of the .\-
plane corresponds n, del'initc set of valnes of t he ratios .\
1
: .\
2
: .\
3
and
hence a detlnite l[llatlr-ic of the bundle (1) ami conversely, so that
t he quadrics of t he bundl e atHl t he points of the .\-plane n.re in one
to oue corrcsporulcnce. To the points of n.ny stmight line in the
.\-plane conespontl the quatl ries of ::1. pencil contained in the bundle.
The line will be said to con espond to the peucil. Since any two
lines intersect inn point, it fol lows that any two peuci ls of qnathics
contained in t he bundle lr;Lve one quaclric in comuron.
138. Singular quadrics of the bundle. Those values of .\., .\
2
, .\
3
which sa.tisfy the equati on
l.\,o k + .\/.J;k + .\3c,k I = o (V)
will define si ngulat !!H:tdrics of t he bnndle. Unl ess special rela-
tions exist among the corftiuicnts "k b;k c,k, nonn of these cones
will be composite, for in t hat case all of t he tirst minors of (9)
Ill liSt van ish, t,hns givi ng ri se to three i ndcpentlen t conditi ons among
the.\., .\
2
, .\
3
, whiclt are not sntislictl for arbi t rar-y valnes of the
cocfficirnts. Tt follows fur-ther t hn.t, under t he same conditions, no
two cones contained in the bnntlle have t he snmE\ vertex. For, if
J( = 0, L = 0 were two cones haviug t he same vertex, then every
ARTS. 136-1391 sgCTION OF A B(J)l'"DLI<; 169
cone of the pencil )..
1
/{ + = 0 \VOuld have this point for a ver-
tex. By choosing this point as vertex (0, 0, 0, 1) of the tetrahe-
dron of reference, the pencil could be expressed in terms of the
three variables

x
2
, x,. Tlte discriminant of this pencil equated
to zero would be a cubic in X..
1
: whose roots defiue composite
cones which were shown above not to exist for arbitrary values of
a,., b,k, c,k.
I t follows from (9) that the points in the >-..-plane determined by
values of >-..
2
, >-..
3
which define cones of the bundle of (8) lie on a
quartic cmve 0
1
. Every point of this curve defines a cone of the
bundle, and roHYersely. Each cone has a vertex, allll it was jnst
shown that no two cones have the same Yertex. '\Ve have therefore
the following theorem:
TH EOJ{E)t. Tlte ?'ellices of the cones in a yeneml bnmlle
u space CUI'Ve .T. Tile of J ure itt one to oue coi'l'esponclence
tcith the point:; of lite Cll l' l!e 0
4
in lite A-plane.
The four points in which any line in the A.-plane intersects 0
4
correspond to the four singular quadrics of the pencil which cot-
respontls to the line. If P is any point on the qmutic curve, the
tangent line to 0
4
at P defines a pencil of c1uadrics in which one
singular quadric is counted twice; if the residual p()ints of inter-
section of the tangent line aud 0
1
are distinct from each other and
from the point of contact, the characteristic of the pencil is [311].
All the cptadrics of tlte pencil pass through the vert<?x of the cone
cones1londing to the point of contact.
13 9. Intersection of the bundle by a plane. If the quadrics of
the bundle (8) are uot all si ngular, the equation
Ru sl:! S13 .iu
s12

s!?3 ll2
s:/J 1133 .S3.c 113 =0, (10)
SH
Sz" S34 s44 7t4
111 11
2 113 114 0
wherein s,k = X..
1
a,k + + X..
3
c,k, is called the equation of the
bundle in plane coordinates. If the coordinates of a given plane
(n) are substantiated in (10), the result.ing equation, if it does not
vanish identically, is homogeneous of degree three in X..u )..
3
and
170 LINEAR SYSTE:\IS OF QUADRICS (CUAP. XI.
is consequently the equation of a cubic curve 0
3
in the ,\-plane.
Equation (10) is the condition that the section of the quadric
(.\
11
,\
3
) by the J:.llaue (n) shall be composite. Every such com-
posite conic in the plane (tt) has at least one double point. It wi ll
now be shown that the locus of the point of tangency to (tt) of t he
quadrics of the bundle which are touched by (1t) is a cubic curve.
The equation of ::u1y plane (tt) may be reduced to = 0 by a
suitable choice of coordinates. Let "-a.

be any set of values
of .\., .\
2
, ,\
3
whi ch satisfy (1 0) when we have replaced all u
2
, u
3
, each,
by zero and 11
4
by 1.
The section of the quadric X
1
A + X
2
B + X
3
C = 0 by the plane
= 0 is a composite conic having at least one double point (Yu y
2
,
y
3
, 0). The coordinates of (.II) must satisfy the relations
+ >..t-:ib,k!Jk + A
3
'i.c
1
.?h = 0, for i = 1, 2, 3.
If f rom these thrl'e equations X
1
, A
3
are eliminated, the resnl t is
the equation of the locus of the point of contact (?J). Since there-
sulting equation is of tlegree three in the homogeneous variables
y
1
, y
2
, ?h the locus is a cubic curve. It is called the Jacobian of
t.lHl net of conics in the given plane.
14 0 . The vertex locus J The order of a space curve is defined
as the number of its (mal and ima.ginary) intersections with a
given plane.
\Ye shall now prove the following t.heorem:
' f HF.ORK\1. 'l'he verte.""C locn8 .! of" ye11e1al bundle 1s of order si.'c.
For, the condit.ion th:1t the vertex of :1 cone of the bundle lies
in a given plane (n) is that the corresponding point in the ,\-plane
lies on each of the cmves (9) and (10). The theorem will follow
if it is shown that t hest> cmves ha,'e contact of just the first order
a.t each of t he eou1mon points so that their twelve intersections
coincide in pairs.
Let the plane he taken as x. = 0. The equati on of a cone
of the bundle having its vertex in this plane can be reduced to
:vz2 + x3z + = 0,
and that of the bullllle to the form
"-aA + + .\J(.t'2
2
+ .r
3
1
+ xl) = 0.
ARTS. 13\J- 141] POLAB. 'rliEORY IN A BUNDLE 171
The point in the A-plane corresponding to the cone is (0, 0, 1).
It lies on CiA) and on C
3
(A). It is to be shown that ClA), 0
3
(A)
haYe the same tangent at (0, 0, 1), but that they do not have con-
tact of higher than the first order. r n (0) put c2'l = Caa = c44 = 1
and all the other c,k = 0, and develop in powers of ,\
3
The re-
sult may be written in the form
(auAa + buA2)Aa3 + { I<Pu <Pa21 + I<Pu <Paa l+ I <Pu <Pa l} Aa2 + ... = O,
</>22 <Pa3 <Paa <Pa 44
wherein </>;k = a,k,\
1
+ u,kA2 = <Pk,.
Similarly in (10) put u
1
= 11
2
= u
3
= 0, = 1, c,k = O, and
develop in powers of A
3
The result is
(a
11
A
1
+b
11
,\z),\
3
2
+ {1!:: : :1 + 1:::
These curves both pass through the point (0, O, 1) and have the
same tangent a
11
A
1
+ b
11
,\z = 0 at that point. By making the two
equations simultaneous, it is seen that they do not have contact
of order higher than the first unless au>.., + bu>-.2 is u factor of
<Pu - q,,.:,
which is not the case unless particular relations exist among the
coefficients a,k, b,k.
141. Polar theory in a bundle.
'rnEORF.)J. 'fhf' polw planf's of a ]?oint (y) with 1egarcllo nlllhe
q!l(telrics of u btmrlle ])ass lhrotrJh a fi:cetl J'Oinl (y') .
For, the polar plane of the point (!/) with regard to a quadric
of the bnndle A
1
A + ,\
2
8 + A
3
0 = 0 has lhc equation
+ . t,!h + = 0.
For all values of ,\
11
,\z, ,\
3
this plane passes through the point (y')
of intersection of the three planes
= 0, = 0, "S.c,kxiyk = 0. (11)
From the theorem that if the polar plane of (y) passes through
(Jj'), then the polar plane of (!/) passes through (y), it follows that
all the points in space are arranged in pairs of points (y), (y')
172
LINEAR SYSTEl\lS 01<' QUADRICS [CuAr. XL
conjugate as to e''ety quadric of the bundle. Since the cootdinates
Yu "h y
3
, y
4
aud y'., y'
2
, '!/
3
, ?/
4
appear symmetrically in the eqnations

defining the correspondence between (y) and (y'), the correspond-
l!nce is called involutorial.
By sohing the equations defining the col'l'esponclence for y'.,
y'
2
, y'
3
, ?/
4
we ohtain
"'
- a2k?h

-ct3k.
1
h
CJ',Ih'= "i.b2kYk "i.b3k?fk
;5.bJi
"i.t;.!h "i.t:.kyk
and si mi Jar ex pressious for y'
3
!t'
4
lf we denote the second
members of the respective ef1uations by ,(!f), then replace both
y, and y', by x, and ,;',, respccti,ely, the equations defining the
involution may be written in the form
(12)
If (y) describes a plane . , = 0, the equation of the locus of
(y') may be obtained hy eliminating the coordi nates of (y) from
(11) and the equation ;u,y, = 0. The result is
111 112 !Ia
11,,


:Sa.kxk - zkXk
=0. (13)
"i.lluXk "i.b2k:t:k "i.b3kxk "i.b.kxk
"i.c,k.1'k "i.c3k.tk
Hence, if (y) descri be:; a plane, (y') describes a cubic surface.
Similarly, if (y') describes a plane, (y) describes a cubic surface.
If (.v') describes a line Z, the point (y) to which it cotTesponds
describes a cm,e of order three. For, conesponding to each
intersection of the Joens of (y) wit.h the plane :Sn,x, = 0 there is
a point of intersection of l and the cubic surface, image of the
plane. But l intetsects the surface ( 13) iu t,Jnce points, hence
= 0 intersects the locus of (y) in three points; that is, the
Joens is a curve of ordPr three. Rim ibrly, if (y) describes a
straight line, (y') will desrribe a cnnc of order three.
The vertex locus .J lies on the surfa.<:e for all positions
of tltc plane "i.tt,.'l:, = 0. For, let (?!') be any point on ./. Since
(y') is the vertex of n. cone belonging to the buudle, its polat plane
AnTs. 141, 1421 80.:\lE SPEClAL ilUNDLES 173
with respect to this cone is imletcnninatn (Art. 1:?1). Hence
there exisls a st>t of values of X-
1
, X-
2
X-
3
not all zero, for which this
plane is indPterminate. It follows that t he matrix
' :Sa,k.rt

I .. tk
:SbztJ't
:SI14k.l'*
:Scw1' :Sc
2
k.1

is of rank at most two. 'fhns, in the eqnat.ion of the cubic sur
face (1.'3), the coefficient of each t!; vanishes when the coonlinates
of any point J are substituted in it; hence til!\ equation is satislied
fo all \al ues of (u., u
2
, u
3
, 11
4
)
.Any two planes :St,x, = 0. :Sl,:t, = 0 intersect in a line; their
image smfaces inte rsect in a composite curve of order nine, consist..
ing of J and the cubic cmve, image of the I in e. If the point (y)
is the vertex of a cone belonging to the bundle, the th rre polar
planes of (!!) tlcteminrc\ by (1 1) belong to a pencil. T ... et l be t he
axis of this pencil. E\'Cry point of the line l corresponds to (y)
in the correspondence ( 11), since it is involutorial.
As (y) de:.nibes J, its cont>spontling line l describes a ruled
surface R. T he image of a cnuic surface :S't,</>, = 0 in the involu-
tion (12) is the plane "J. tt,!h = 0 and a residllal sndace of orde1
eight. .As this residual surface is the locus of l, we concl ude
that the ruled surface R is of orde1 eight.
142. Some special bundles. While it would lead beyond t he
scope of this book to give a complete classification of buncHes of
quad1irs, like that for pencils of quadrics as developed in
Arts. still it is desirable to mention a few particulal'
cases. It was seen (Art. 138) that in the general bllndle there
are no composite quadrics. But bundles containing composite
quadrics may be constructed; for example, the bunule
X., A + X.
2
B +

= 0
evidently contai ns the composite quadric X
1
:t:
2
= 0. If x
1
= 0 i nter-
sects the cmve of intersection of A= 0, B = 0 in four points, and
0 i ntersects it iu follr points, so that no component of the
curve lies in either plane x
1
= 0, = 0, then t hese two sets of,four
points constitute eight assor;iatP.d points.
174 LINEAR SYSTEMS OF QUADRICS [CHAP. XI.
very point of tl1e 1 ine x
1
= 0, x
2
= 0 is a vertex of a com-
posite cone of the bundle. The locus .T consists of this line and
of a residual curve of order five. The image curve 0
4
(.\) in the
.\-plane has a double point corresponding to the composite quadric,
as may be seen as follows. The equation of a
4
(.\)= 0 now has
the form
in which
2
, cp
3
, cp
4
do uot contain .\
3
Hence the point .\
1
= 0,
= 0 is a double point on a
4
(.\)= 0; it corresponds to the quadric
x
1
x
2
= 0. The points of a
4
(.\) are uow in one to one correspond-
ence with the curve of order five, forming one part of J, and the
douule point is assor.iated with the whole line x
1
= O, x
2
= 0.
Similarly, bundles of quadrics may be constructed having eight
associated double points lying on two, three, four, five, or six
pairs of planes. In the last case the equation of the bundle may
be written iu the form
.\
1
(x
1
2
- xl)+ xl) + .\
3
(X
3
2
- xl)= 0.
The eight associated points arc ( 1, 1, 1, 1). The curve
J consists of the six edges of a tetrahedron and a
4
(.\) is composed
of the four sides of a quadrilateral. Its equation is
+ .\2 + .\3) = 0.
In this case the equations (12) of the involution (y), (y') have the
simple form
1 CT
y,=-,
y,
in which u is constant.
i=1, 2, 3, 4,
Bundles of quadrics exist having a common curve and one or
more distinct common points. The spheres through two fixed
points furnish an example.
EXERCISES
1. Show that (1, 0, 0, 0), (0, 1, 0, 0) , (0, 0, 1, 0), (0, 0, 0, 1), (1, l, 1, 1),
( 1, -1, -I), (1, -1, I, - 1), (1, -1, - 1, I) a.re eight associated
points.
ARTS. 142-144) THE JACOBIAN SURFACE OF A WEB 175
2. l'rOI'e that if P is a gi1en point and l a given line through it, there
is one and only one quadric of the bundle to which l is tangent at P.
3. Determine the characteristic of the pencil of quadrics in a general
bundle corresponding to :
(a) A tangent to C
4
( X).
(/.1) A double tangent to C4(X).
(c) An inftaxioual tangent to C4 (X).
4 . What is the general condition under which C,(X) may have a. double
point?
5. Deterruiue the nature of the bundle
X2X3) +

+ X2
2
+ X3
2
- 4 x,
2
) X3
2
) = 0
and of the involution of corresponding points (y), (y').
6. If three quadrics have a common self-polar tetrahedron, tlie twenty-four
tangent planes at thei r eight in tersections all touch a quadric.
7 . Write the e\JU:'Ition of a bundle of quadrics passing through two given
skew lines and a given point.
8 . If four of the eight common tangent planes of three quadrics meet in a
)JOint, the other four all meet in a poim.
9. Show that the cubic curve, image of an arbitrary line, intersects the
locus of vertices J in 8 points.
10. Show that the surface R of Art. 141 couta.ins J as a threefold curve.
143. Webs of quadrics. If A= :Sa,kx,xk = O, B = = 0,
0= "Sc,k:t,xk = 0, lJ = -:$rl,kx,xk = 0 are four quadrics not belonging
to the same bundle, the linear system
(14)
is called a web of quadrics. Through any point in space pass all
the quadrics of a bundle belonging to the web, through any two
independent points a pencil, and through any three independent
points, a single quad ri c of the web.
144. The Jacobian surface of a web. The polar planes of a
point (y) with regard to the quaurics of a web form a linear system
>..,:Sa,tX,yk + + >..3:Sc,tX,Yk + = 0. (15)
If t he point (y) is chosen arbi trarily, this plane may, by giving >..u
>..
21
>..
3
, snitable values, be made to coincide with any plane iu
176
LINEAR SYSTEMS OF QUADRICS [CHAP. XI.
space, nnless there are particular relations among the coefficients
aw bw C;h <l;k Thus an plane is the polar plane of (y)
with regn.nl to some quadric of the web. 'here exists a locus of
points (y) whose polar planes with regard to all the qmtdrics of a
web pass through a fixed point (y') . This locus is called the
J acobiau of the web. Since the eqnations connecti ng (y) and
are sym mr.tl'ical, it follows that (y') n,lso I ies on the J acobiaJ'. A
pair of points (y), (y') snch that all the polar planes of each pass
tb tongh t he other a 1e called conjugate points on t he ,Tacohian.
'L'o dctermiue the equation of the .Jacobian, we impose the con-
dition that the four polar planes of (y)
pass through a poiut. The resnlt is
;!,(( 3i!/i "S.n.;y;
]{4=
'i.bt;Y;

'J.b3;!/i :il
1
4oYi
=0.
'i.Ct;?/;

:ic.,y,
(16)
;!.clli?h '21/3,Yi :ill.,y;
rrhe condi tion that a. poi nt (IJ) is the vertex of a. cone contained
in the web is that i t:; C'Oiirdinates satisfy the equations
A.
1
':!<t;k!h + A.
2
"S.b;k!h + A.
3
-:!c ik?li + A.
4
'5.d,k?f, = O, k = 1, 2, 3, 4 (17)
for some values of .\
11
A
2
, A.
3
, A
4

By eli minating A.
11
>..
2
, >..
3
, .\
4
we obtain equation (16). This gives
the theorem :
I. The Jacobian surface is the locus of the vertices of the
cones contained in the web of quadrics.
If from equations (17) we eliminate Yt, y2, y3, y, we obtain
tn l12 /13 tu
T(X) , t21 t22 l23 tu = O
0
t 31 t32 t33 t3, '
lu t,2 l;a t,.
in which t;k = X1a;k + X2bik + X3Cik + X,d;k = lk< Any set of values
of X1, X
2
, X3, X, for which T(X) = 0 determines a singular quadric of
the web. Conversely, the parameters of every singui!1r quadric in
(14) satisfy T(X) = 0.
ARTS. 144-146] " 'EB WITH SIX BASIS POINTS 177
Since T(X) is a symmetric determinant there are ten sets of values
of X
1
, Xz, X
3
, X, for which it is of rank two.* The ten corresponding
quadrics arc composite and each li ne of vertices lies on [(, = 0, hence
we have the theorem:
II. The Jacobian of the general web of quadrics contains
len lines.
145. Correspondence with the planes of space. The polar plane
of a fixed point (!!) with regard to any quadric Q of the web will
be called the associated pln.ne of (!I) as to Q. When Q describes
a pencil, its associated plane will describe a pencil; when Q de-
scribes a bundle, its associated plane will describe a bundle. The
quartic cune of intersection of two quadrics of the web corre-
sponds to the-line of intersection of their associated pl:tues, and to
every set of eight associated points of a. bundle of quadrics in the
web corresponds one point, the vertex of the bundle of associated
planes. Through any two points a straigllt line can be drawn,
hence through any two sets of eight associated points within the
web can be passed a pencil of qnatltics belonging to the web. Since
through any three points a plane can be passed, it follows that a
quadric of the wPb can be found which passes through any three
sets of eig:1t ass0eiated points in the web.
14(). Web with six basis points. The maximum number of dis-
tinct basis points a web can have withont having a basis cnrve is si x.
Let 1, 2, -1-, 5, 6 designate the six basis points of a web having
six basis points. All the (jlln.tltics of the web through an arbitrary
point P belong to a bundle, and hence have eight associated points
(Art. tafi) in common, the ei);hth point P' being fixed when 1, 2,
3, +, !i, 6 and Pare given. Between P= and P' = (t') exists
a non-1 i nea.r correspondence.
\\' p shall now ptove t.he following theorem :
Tu r-:oRF.:'IT T. In the of a 1ceb 1cith six clistinct 1lOints,
the ./IICObicw Slll:{Hce K. = 0 i.'i also the locus of 1lOints a) such thctt
w
Salmon: Lessons Introductory to Modem Higher Algebra, Lesson XIX.
The configuration of the lines on the Jacobian has been studied by Rcyc. Sec
Grelle's Journal, Vol. 86 (1880).
178
LINEAR SYSTE.MS OF QUADRICS !CRAP. XI.
In order to prove this we shall prove the following theorems:
TnEORF.M II. The quadl'ics of a bundle of the wel> which pass
through the vel'te.'C of a given cone of the 1veb have, at this ve1tex
1
11
cmmnon tcmrtent line.
THEORE:II III. Gom:e1sely, if all the quad1ics of a bundle have
a cmnmon tangent line at a given point, a cone belonging to the
bundle has its vertex at the point.
To prove Theorem II, let the vertex of the given cone be
(1, 0, 0, 0), so that its equation G = 0 does not contain x,. Let
A= 0, B = 0 be any two non-singular quadrics of the bundle
passing through the point, so that a
11
= 0, bn = 0. The erptation
of the tangent plane to the quadric .\,A+A:B+.\
3
0=0 at
(1, 0, 0, 0) is
.\
1
(
1
11=
2
+ + a
14
x
4
) + b12X2 + + b14x4) = 0.
But these planes all contain the line

+



+ aHx
4
= 0, b
12
x
2
+ b
13
x
3
+ b
14
x
4
= 0,
which proves the proposition.
To prove Theorem 111, let x, = 0, x
2
= 0 be the equations of the
line, and (0, 0, 0, 1) the common point. we may then take
A= 2 a
14
X
1
X
4
+ cJ>(:x;, x
2
, 0,
B = 2 b
24
X
2
X
4
+ 1f!(x
1
, x
2
, x
3
)= 0,
G = 2 c
14
:t;X
4
+ 2

+ f(x
1
, x
2
, x
3
)= 0,
wherein </>, If!, f contain only x
1
, x
3

In the bundle
the quadric corresponding to .\
1
=- c
14
bw =-

.\
3
= a
14
b
14
is a cone with vertex at (0, 0, 0, 1) since the equation of the quadric
docs not conL1.in x
4

Since n.t the vertex of every cone two associated points coin-
cide, and conversely, at every coincidence is the vertex of a cone,
the proposition of Theorem I follows.
'J'he ten pairs of pl:tnes determinecl by the six ba1>is points
1, 2, 3, 4, 5, 6 taken in groups of three, as, for example, the pair
AnT. 146) WEB WITH SIX BASIS POINTS 179
of pl:tnes (1:?8), (.J5G), are composite quatldcs of the web. The
line of vertices uf each pair lies on Il
4
= 0. The snrface J{, = 0
also coutains the tifteeu lines joining the basis points by twos,
since through any poi nt of such a liue five lines can be drawn to
the six basis points, and a quad ri c co11e of the web is fixed by
these five lines.
If the bn:;is points are taken for vertices of the tetrahedton of
reference, the 1111it poiut, and the point u
2
, a
3
, C1
4
), the equation
of 0 is found to be
fl t X2X3.1"4 x. lit
1

x2 ((2 1
=0.
ct
3
x,x
1
x
2
((3 1

X

a, 1
This surface is known as t he Weddle surface.*
If in (17) the ,alnes of y., y
2
, y
3
, y
4
are eliminated, the resul ting
equation A(.\) = 0 of degree four in the X, will define those val ues
for which the eqnatiou X
1
A + >..
2
B + .\
3
0 + .\
4
D = 0 is a eoue of
the web. The vertex of this cone is a point Let X., >..
2
,
.\
3
, >... be eonsillered as the tetmhetlral coordinates of a plane. To
each plane (>..) correspo111ls a quatlric of the web (1-l) ant! con-
versely. .A linear equation with given coefficients <0.
1
+ b>..
2
+
c>..
3
+!I >..
4
= 0 dete nuines a point in the .\-space (Art. fli ). By
making this equation ami (11) simultaneous, we define a bundle
whose basis points are the points (x) whose coordiuates satisfy
the equations
A B 0 D
Ct o c cl
Of the eight associated points so determined, the given points
1, 2, 3, 4, 5, G are six. Either of the remaining points P=. a),
P' =. Cn will uniquely determine the other and also uniquely
determine the point (a, b, c, <l) in the .\-space. 'l'he equation
a.\
1
+ + c.\
3
+ = 0 thus defines a one to two correspondence
between the points of the .\-space and the points P antl P'. For
First discussed in tbe CaUluridge and Dublin Mathematical J ournal, Vol. 5
(1850), p. 69.
180
LINEAR SYSTEi\lS 01+' QUADRICS [CnAP. XI.
points of ){, P and J>' coincillc. The locus of the corresponding
point (a., b, c, cZ) is call ed the Kummer surface.*
'Vc have thus proved the foll owing theorem:
TnEolm;\r IV. The o.f the lVecldle s1o:fuce w1l the point:>
of the J{ummer swface are in oue to one correspondence.
EXERCISES
1 . Show that all the quadrics having a com111on !>elf-polar tetrahedron
form a web.
2 . Determine the J acobian of the web of Ex. 1.
3 . D!'tennine under what conditions the .Jacobian of a web will have a
plane as component.
4 . Find the Jacobian of the web definctl by the spheres passing through
the orif: in x = 0, y = 0, z = 0.
5 . Show that the .Jacobian of a web having two basis lims is inde-
terminate.
6 . Discuss the involution of conjugate points (y), (y
1
) for the web of
Ex. 4.
7. Show t.ltat the spheres cutting a given sphere orthog<Hl ::llly define a
web.
8 . Show that the equation of the quadric determined by the lines joining
the points ( 1, 0, 0, 0), (a., a
2
, a
3
, a,); (0, l , 0, 0), (0, 0, 1, 0); (1, 1, 1, 1),
(0, 0, 0, 1) is
x,.c1(a2 - a3)+ (a3rz - a2x3) + a,x2)= 0.
14 7. Linear systems of rank r. The liuem system of quadrics
>..1F1 + >..2F2 + + >..,F, = 0
1
(19)
wherein
I
= 1 . .. .,.
I I
is said to be of rank 1, if the matrix
Clu{l) Ct:r.!
0 1
Ct33
1
et3}
11
a ut
2
> a22 <
21
Ct33 m a3, (
21
(20)
First <liscusse<l by K Kummer lu the ilfonatsbericbte der k. preusslscben
Akademie der Berlin, 1803.
ARTS. 14G-119) APOLARITY 181
is of rank,., that is, if there does not exist a set of values of ..\
11
, ..\,
1
not all zero, such that the expression
..\1F1 + ,\2F! + ... + >..,/<',
is identically zero. All the quadries in space form a li near sys-
tem of rank ten, since the equation of any quadric may be ex-
pressed lincal'iy in terms of the ten quadrics, X
1
2
,

... ,

for
which thu 111attix is of rank ten.
All the quadrics in space whose coeOlcients satisfy 10- r
independent homogeneous linear equations fotm a linear system
of rank r. For, if 'S.b,/(<r,.rk = 0 is the equation of auy quadric
whose <ocllicicnts satisfy the given contlitions, then all the co-
efficients b,
1
c:ut be expressetllinea..Jy in terlllS of the coefficients
of,. quadrit$ uelongiug to the systelll. 'L'hus
b,k =

+ + >..,a,,cr1, i, k = 1
1
3, 4, (21)
wherei n
a re fixed quadrics belonging to the system.
Con verscly, l 0- r indepentlen t homogeneous linear conditions
may be found which are satisfietl by th<' coetlh:ients in the equa-
tions of the quadrics F
1
= 0, F, = O, , = 0, and consequently
by the coefficients in the equations of all the quadrics of the
linear system (19) of rank r.
148. Li near systems of rank r i n plane col)rdinates. T he system
of quadrics
>..l<f>l + + .. . + >..,<f>,= 0,
wherein <f>
1
::= ':!{3, k
1
'
1
u,u
4
, is call ed a l inear system of rank ,. in
plane coonlinates if there docs not exist a set of values >..
11
, >.., for which the given equation is s:ttisfied ideutically.
These systems may be discussed in the same manner as that
considered in the preceding at-ticle.
14 9 . Apolarit y. Let F= ':!n,kx,xl = 0 be the equation of a
quadric in point cotitdinates and <I>= "S.{3,
4
u,u, = 0 be the equation
of a qnadl'ic in plane coordinates. If t.he equation
J. a,tf3it = C111{311 + + (!33{333 + a44{3H + 3 fC1zf312 + 2 rc,af313 + 2 al4f3,.
+ + 2 rcuf3z4 + 2 CJ34{334 = 0 (22)
182 LINEAR SYS1' El\ lS OF QUADRICS !CHAP. XI.
is satisfied by the coefficients i u the ecJlta.ti ons of the two quadrics,
F = 0 is said to be apolar to <I>= O, and <I>= 0 is said to be a.polar
to F = 0. It shoul<l be noticed that in this definition the equa-
tion F= 0 is given in point coordinates, and that of <I> = 0 in
plane coordinates. It should also be noticed that if F= 0 and
<I>= 0 are two given apolar quadrics, and if

= 0 is the
equation of F = 0 in plane coordinates, and = 0 is the
equation of <1> = 0 in point coordinates, then it does not necessarily
follow that = 0 because = 0.
In order to show the significance of the condition (22) of
apolarity, we shall prove the following theorem:
THEOitEM I. The exJn-ession a,kf3i k is a ?'l'l<ltive invariant.
Let the coordin::ttes of space be subjected to the linear trans-
fonnation
X;= ailx'
1
+ ct
12
:J;'
2
+ a
13
x'
3
+ a
14
x'
4
, i = 1, 2, 3, 4
of determinant T:t= 0. The coordinates of the planes of space
undergo the transformn.tion (Art. 97)
1t
1
= .Ail1t
1
1
+ A
12
tt'
2
+ Ll
13
1t'
3
+ A
14
n'
4
, i = 1, 2, 3, 4.
The equation F (x) = 0 goes into

= 0,
wherein (Art. 104)
and <I>= 0 is transformed in -:5.{3'
1
11
1
1t' k = O, wherein
The proof of the theorem consists in showing (Art. 104) that
In the first member, replace a',k, by their values from the
above equations, and collect the coefficients of any term a
1
,.{3
1
,. in
the result. \Ve find
hence
which proves the proposition.
' l'he vanishing of this relative invariant 1nay be interpreted
geometrically by means of the following theorem:
ART. 1-19] APOLARITY 183
JI. IfF= 0, <I>= 0 are apolar quadrics, there exists
a tetmhed,-on se((-polar as to <l> = 0 ancl iuscriuecl in F = 0.
This theorem should be replaced by others in the following
exceptional cases in which no such tetr:thedrou exists.
(a) If F = 0 i's a plane counterl twice. Jn this case (22)
is the condition that the coordinates in this pl ane satisfy
<1>=0.
(II) If <I>= 0 is the equation of the tangent planes to a proper
conic a and if F = 0 intersects the pl ane of a= 0 in a line
counted twice, is the condi tion th:tt this line touches a.
\\'e shall consider first the special ca:>es (a) and (b) .
Then a.t = u;uk and (23) reduces at once to <I>= 0.
Jn case (b), the plane of C be taken as X
4
= 0 and the line
of intersection of P = 0 with :r
4
= 0 be taken as x
1
= x
4
= 0.
Then
<I>= f3uu,z + f3zzlll + f3J:J"l + 2 f3tz'ltllz + 2 f32a1tzUa + 2 f3at11altt =0,
and F = C/
11
.1:
1
2
+ 2 a
14
X
1
.t
4
+ :! Ct
24
Xc.X
4
+ 2 a
34
J"
3
X
4
+ 2 a,
1
,
1
xl = 0,
whe1e a
11
=1= 0. Hence (:?:?) reduces to {3
11
= 0, that is, to the con-
dition that :c
1
= .l'
4
= 0 touches C.
To prO\'e Theorem fl, excltHling rases (11) :tnl (b), we must
consider various cases. First suppose <I>= 0 is nou-singular.
Choose a point P
1
on F = 0, not on the intersection F = 0, <1> = o,
ami fincl its polar plane r.
1
as to <I>= 0. l n r.
1
take a point P
2
ou
F = 0, not on <I>= 0, and find its pola1 plaue r.
2
as to <1> = 0. On
the line r.
1
r.
2
choose a point 1'
3
ou F = 0, not on <I>= 0, and llnd
its polar plane r.s If the point of intersection of r.
1
, r.
1
, r.
3
is
called P
4
, th<>n P
1
P
2
P
3
P
4
= 7Tf7Tzr.
3
r.
4
is tak<>n for the tetrahedron
of 1ef1:'1'ence; we lllay. hy JH'OJlt'r choice of the unit plane, rcdnte
the CfJnatinn of <I>= 0 to 11
1
2
+ u/ + n/ + 11/ = 0. Equation
now has tl1e form n
11
a>J + (1
44
= 0. Sinre three of the
vertice.; P
1
P
3
were chosen on F = O, three coefficients a,
1
= 0,
hence the fourth must also vanish, which proves the proposition
for this case.
It should be observed that if F = 0, 1> = 0 define the same

LINEAR SYSTE.MS OF QUADRICS [CHAP. XI.
IJII:tdric, equation cannot be satisfied since t heir equations
may Le reduced simul taneously to
Now let <I>= 0 be the equation of t he tangent planes to a
proper conic c. Take the plane of a as x
4
= 0, so that
/314 = f3u = = f3H = 0.
If F= 0 is composite and x
4
is one component, equation (22)
is i1lentically satisfied. In this case we may take three vertices
of a t ri:wgle in x., = 0 self-polar as to t he conic a and any point
on F = 0 not on = 0 as vertices of a tetrahedron self-polar to
<I>= 0 and inscribed in F = 0. If F = 0 consists of x. = 0
counted twice, (22) expresses the condition that t he plane
belongs to <I>= 0, whether <I>= 0 is singular or not. This is the
exceptional case (a).
If x
4
= 0 is not a component of F = 0, (22) has the form
auf3u + + a33f333 + C112f31z + 3 C113/313 + 2 0,
which is the condit ion that the section 0' of F = 0 by the plane
x
4
= 0 is apolar to 0.
It foll ows by the theorem fo1 apolar coni cs analogous to
T heorem II t hat a triangl e t>:\ists which is inscribed in 0' a nd is
self-polar to 0. A tetrahed1on having the vertices of this tri-
angle for three of its ve1tices and n. fourth vertex on F= 0 but
not ou x
4
= 0 satisfies the condition of the t heorem ((lual of
Th. I, Art. ).
If <I> = 0 is the equation of two distinct points, (22) expresses
the condition t. hat these points are conjugate as to F = 0. This
i s also the condi tion that a tt>trahedron exists which is inscribefl
in F= 0 and is self-polar to <I>= 0. If <f) = 0 is the eqnation of
a point coun ted t.wice, cxpre>sses t hat t he point lies on
F = 0. T his is the dual of t he exceptional case (a).
In each of the aboYe cases, the tet rahe1li'On which satisfies t he
conditions of t he theorem can he chosen in an infinite nnmber of
ways, hence we h:we the foll owing t!tcot'Cill.
TnEORE:\I IlL If one tetr<tlledron e:cists tvhit:h in
F = 0 cmrl i.s sr:lfpolar as to <I>= O, then an iufiuite number of such
t-etrahedra exist.
AnT. 149] APOLARITY 185
By duality we have the following t heorems:
THEORF..ll IV. IfF= 0, <1> = 0 are apolar quadrics, there exists
a tetraherlron seU:polar to F = 0 a11cl circumscn"becl to <1> = 0.
TuEORF.)I V. Ii one tetl"aheclron e:cists which is circum11cribed to
<I> = 0 a11cl is selj-polw as to F = 0, then wt in.Jittite nwnue1 of such
tetrahedrct exist.
:Moreover, both the exceptional cases of Theorem II have an
immediate dual inteapretation; they will not be considered further.
With the aid of these results we can now give an interpretation
to the vanishing of the coefficients 0 and 0' of equation Art.
and of 'll
1
(u), of equation (7), Art.135. If B = 0 in (1)
is non-singular, let its equation in plaue cocil"dinates be 'i.{3,k1t,1tk=0.
Since {3,k is the first minor of b,k in the discriminant of B = 0, it
follows at once hom equation (::3) that 0' = Hence 0' = 0
is the condition that A = 0 is apolar to B = 0. If B = 0 is a cone,
it is similarly seen that 0' = 0 is the condition that the vertex of
the cone B = 0 lies on A= 0. If U = 0 is composite, 0' is iden-
tically zero, independently of A, since the discriminant of B = 0
is of rank two, hence all the coefficients {3
1
k vanish. An analogous
discussion holds for 0 = 0.
The surface w
1
(1t) = 0 (Art. 13;3) may be defined as the envelope
of a plane which intersects A= 0 in a conic which is apolar to
the conic in which it intersects D = 0. l'or particular singular
quadrics this definition will not always apply.
Let au arbitrary plane of w
1
(1t) = 0 be taken as x
4
= 0. It fol-
lows from equation (7) that
I ub:!"lu331 + I buanb33 1 + I bub22Ct33l = 0. (23)
Let the sections of L1 = 0, B = 0 by x
4
= 0 be 0, 0', respectively.
If CJ is not composite, it is seen by writing the equation of CJ in
line coordinates that (23) is the condition that 0 is apolar to 0'.
If CJ is a pair of distinct lines, (23) is the condition that their
point of intersection lies on 0. If (J is a liue counted twice, (23)
is satisfied identically for all values of a,k, since all the first minors
of the discriminant of 0' vanish.
An analogous discussion holds for w
2
(u) = 0.
186 LINEAR SYSTEMS OF QUADRICS [CHAP. XI.
1130. Linear systems of apolar quadrics. Since equation (22) is
linear in the coefficients ofF = 0, from Art. 147 we may state the
following theorem:
TuEownr I. All the quad1ics apolar to a git
1
en quad1ic form a
Unectr system of 1ank nine.
(;onversely, since the coefficients of t he equations of all the
quadrics of a linear system of rank nine satisfy a linear condition
which may be written in the form of equation (27), we have the
further theorem :
THF.OitE)1 n. All the quml1ics of any linea1 system, of mnk nine
a.re apolctr to ajixetl qna<lric.
From the condition that a plane counted twice is apolar to a
quadric (Art.l-!9), it follows that this fixed quadric is the envelope
of the double planes of the given linear systmn.
If a quadric F = 0 is apolar to each of 1 quadrics
<1>
1
= = 0, <1>
2
= '2.f3,t<
2
>n,1tk = O, ... ,
<1>, = 'J.{3ik(r)ll;llk = 0,
the coefficients in its equation satisfy the r conditions
'2.a,kf3,k<l> = 0, 'S.u,kf3. k<2) = 0, ... ,
';S.a,d3,/'' = 0.
(24)
It follows that if a quadric is apolar to each of the given quadrics,
it is apolat to all the quadrics of the linear system
:\1<1>1 + + + :\,<1>, = 0.
The conditions that this linear system is of rank 1 are equivalent
to the conditions that the conesponding equations (24) are inde-
dendent. Hence :
'I'HEOIUDI Ill. All the apolar to the qtutclrics of a linear
system of mnk r in plane fo1'1n a linear syste1n of rank
10 - r in point coordinates aiiCl dually.
EXERCISES
1. Find the equation or the quadric in plane coordinates to which all the
quadrics through a point are apolar.
AnT. 150] LINEAR SYSTJ<:.i\1S OF APOLAR QUADRICS 187
2. Bow many double planes are there in a general linear system of rank
se,en in point coordinates?
3. Show that all the pairs of points in a linear system of rank six in plane
coordinates lie on a quartic surface.
4 . Show that : ~ . l l the spheres in space fo1m a linear system and find its
rank.
5. Find the system apolar to the sptem in Ex. 4.
6. Show that a system of confocal (JUadrics (Art. St) is a linear system of
rank two in plane coordinates. Determine 1 he characteristic and the singular
quadrics of the system (Art. 133) .
7. Show that, if the matrix (20) is of rank r' < r, the system of quadrics
(19) is a linear system of rank r'.
CHAPTER XII
TRANSFORMATIONS OF SPACE
l ul . Projective metric. ln order to characterize a transfor-
ma.tion of motion, either trfl.nslation, or rotation, or both, ot a. tmns-
formation involving motion and reflection, as a special case of a
projPctive transformation, it will first be sliown under what cir-
cumstances orthogonality is preserved when a new system of
coordinates is chosen.
If the new axes can be obtained from the old ones by motion
and reflection, the plane t = 0 must evidently remain fixed, and
the expression x
2
+ !l + z
2
, which tlefincs the square of the dis-
tance from the point (0, 0, 0, 1) to the point (x, y, z, 1), must be
transformed into itself or inLo (:c - at)
2
+ (y- bt)
2
+ (z- ct)2,
according as Lhe point (0, 0, 0, 1) remains fixed or is transformed
into the point (a, b, c, 1). It will be shown that, conversely, any
linear transformation having this property is a motiou or a motion
and a refl ection.
1.')2. Pole and polar as to the absolute. We shall first point out
the following relation between the direction cosines of a line and the
coordinates of the point in which it pierces the plaue at infiuit.y.
THEOI:EM I. The lwnwf!eneoll.S coo,clinates of the 1>oiut in tolticlt
a line meets the plane at h(finity are ptoportional to the direction
cosiues o.f the line.
Tlte eq nations of a I ine th rongh the given finite point (.t
0
, y
0
, z
0
, lu)
aud having the direction cosines (A, ,..., v) are
lo-"C- xot l f1J- !Jo1 tr(k- zot. (l)
A p. v
The point (x, y, z, 0) in whi ch the line pierces the plane at infinity is
given by the equa.tions
t"1 _ln?f _ tr(k
- - - - ,
A p. v
from which the theorem follows.
188
AnTs. 1!) 1, 1!)2] POLF. AND POLAR 189
We shall now establish the following theorems concerning poles
and polars as to the absolute.
TIIJWJn:;ll II. The necessary and couclilion a pla11e
and a linr> are perpendicular is that the line at infinity in the plane is
the 7olar of the point at iujiuity on the line as to the absolute.
The absolnte was defined (Art. as the imaginary circle in the
plane at infinity defined by the equations
xz + !l + z
2
= 0, t=O. (2)
The polar line as to the absolute of the point (>., ,.,., v, 0) in which
the line (1) intersects the plane at infinity is
A.c + p.!J+vZ=O, t = O. (3)
The equation of any plane through this line is of the form

These planes are all perpentlicular to the line (1). Conversely,
the equation of any plane perpendicular to the line (1) is of the
form (J); the plane will therefore intersect the plane at infinity
in the line
THF.ORJm HI. The 1H'Cessmy and sujJicient condition that two
lines are ])erprmclicular is that tltei,. 71oints at are conju[!ate
as to the ctbsolute.
The condition that two lines are perpendicular is that each lies
in a plane pe1pcnclicular to the other, that is, that each intersects
the polar line of the point at infinity on the other as to t he absol ute.
Finally, since two :ue perpendicular if each contains a
line perpendicular to t he other, we have the following theorem :
TuEORE:Ir IV. The w11l condition that ttco
planes are perpenrlic11lar is thut their lines at infinity are coujugute
as to the absolute.
A tangent plane to the absolute is conjugate to any plane pass-
ing throngh the point of contact; i n pn-rticnlar, it is conjugate to
itself. It should be observed that the equation of a tangent plane
to the absolute cannot be reduced to the normal form, hence we
cannot speak of the direction cosines of such a plane.
HJO TRANSFORl\1ATlONS OF SPACE [CHAP. XII.
Consider the pencil of planes passing through any real line.
We may choose two perpendicular planes of the pencil as x = 0,
y = 0, and write the equation of any other plane of the vencil in
the form
y=mx.
The equations of the two tangent planes to the absolute which
pass through this line are y = i:c anti y =- ix. By using the
usual formula to obtain the tangent of the angle 4> between y = ix
and y = mx, we obtain
t
1n - i m - i 1 .
an '+' = --- = =- = -
l+im i
independent of m. For this reason tangent planes to the absolute
a1e called isotropic planes. The having its vertex at (a, b, c)
and passing through the absolute has an equation of the form
(x - a)
2
+ (y- W + (z- c)
2
= 0.
If we employ the formula of Art. 4 for the distance between
two points, we see that the distance of any point of the cone from
vertex is equal to zero. Fo1 this reason the cone is called a
minimal cone. Moreover, if P
1
and P
2
are any two points on the
same generator, since
VP
2
- VP
1
= P2Pu
we conclude that the distance between any two points on any line
that intersects the absolute is zero. For this reason these lines
are called minimal lines. They have no direction cosines (Art. 3).
lo3. Equations of motion. Let an arbitrary point P be refened
to a. rectangular system of coordinates x, y, z, t and to a tetrahe-
dral system x., x
3
, x
4
, with the restriction that x
4
= 0 is the equa-
tion of the plane at infinity t = 0. The equations connecting the
two systems of coordinates are
ux = .\X
1
+ .\'x
2
+ .\''x
3
+ hx
4
,
uy = p..'V
1
+ + p."x
3
+h'x
4
,
uz = vX
1
+ v'x
2
+ v"l'a + h"x
4
,
ut = x
4

Divide the first three equations of (r>) by the
(5)
last, member by
member, :mel replace the IIOithomugenl'Oils cuordinatPS by
t
A RTS. 153, 154] P ROJ F.C'l' I\' E 'l'RANi:WOR).l A'riO}:S lHl
I I I l XI t b I ,I ,I 1 f j > . t t'
x, y, z an( , c c., y xu .t x
3
I S any pmnt no 1u e
x4
plane at infinity, we shall prove the following theorem:
I . 'Plte most ge11erallinear transfomwtions of the foi'1n
(5) that will tran4onn the e.t'J>lessiou
x
12
+ y
12
+ z
12
into x'
1
2
+ x'
2
2
+

are t!te ?'olatiom; ancl ?'P)lections about the JJOi11t x' = 0', y' = 0,
Z
1
= 0.
If we substitute the values of X
1
, y
1
, z
1
in the expression
x'2 + y
12
+ z'Z, we obtain
(M:
1
1
+ >..
1
x
1
2
+ >.."x'
3
+ h)
2
+ (p.'t
1
1
+ p.
1
x
1
2
+ p..
11
X
1
3
+ h
1
)
2
+ (vX
1
1
+ v
1
X
1
2
+ v
11
X
1
3
+ h")
2

If this is equal to X
1
1
2
+ X
1
2
2
+

for all finite values of x ' u X
1
2
,
x
3
1
, we have the following relations
)..2 + J.l-2 + v2 = )._12 + J.l-12 + v12 = >.."2 + J.l-112 + v"2 = 1,
A>..
1
+ p.p..
1
+ vv
1
= A
1
A
11
+ p.
1
p..
11
+ v
1
v
11
= >..">.. + p.
11
p. + v " v = 0, (6)
It>.. + h
1
p. + h"v = 0, h>..' + h
1
p..
1
+ h"v
1
= 0, h>.." + h'p.
11
+ h"v " = 0.
Since the determinant 1 "Ap.
1
v
11
I is uot zero, it foll ows that
h = h' = h" = 0. The formulas (6) which do not contain h, h',
It" are exactly the relations among the coefficients to define a
rotation or a rotation and reflection about the origin (Art. 37).
This proves the proposition.
By similar reasoning we may prove the theorem:
THEORDt II. Tmus.fonnations that will ttansfonn
x
12
+ y'
2
+ z'
2
i11to (x', - a)
2
+ (x
1
2
- b)
2
+ (x
1
3
- c)
2
consist of motion ot of motion ami reflection. ,
154. Classification of projective transformations. The equations
of any projective transformation (Art. !:18) are of the form
kx', =

+

+ a
13
Xs + a
14
X
4
,
kX
1
2
= a
2
,x, + anXz + a
23
x
3
+ a2.
1
X4,
kx
1
3 = Cl3,x, + a3zX2 + a33xa +
kx', = a4,xl + acr2 + a4Jx3 + ((44x,.
(7)
192 OF SPACE [CHAP. Xll.
We shall now consider the problem of classifying the existing
types of such trausfonnations and of reducing their equations to
the simplest form.
The invariant points of the transformation (7) are determined
by those values of k whi ch satisfy the equation
a
11
- k
(( 12 ({13 at4
D(k) =
((:n (122 - k
a2J a24
=0.
(8)
aat (132
(t33- k (134
a.u (142 ((43 (14-t- k
The classification will depend fundamentally on the invariant
factors (Art. 125) of this determinant.
In equation (7), aud (.t') arc regarded as different points,
referred to the same system of <>oordinates. In order to simplify
the equations, we shall refer both points to a new system of
coordinates. To tlo this both (x) and (.t') :U'e to be operated upon
by the same transformation
We shall use the symbols (.v), (!!) to indicate coordinates of the
same point, referred to t\\'O lliffe1ent systems of coordinates, while
el}_nations between and (:c') or between (y) and (y') will define
a projective transformation between two different points, referred
t o the snme system of coordinates.
Let k
1
be a root of D(k) = 0. The four equations
(au - k,).t't +

+ a
13
Xa + et14X4 = O,
a2ti, + ( Cl22 - k,).rz + a23x3 + Uz4X4 = 0,
Cta1X1 + tt3:!XI + ( Ct33 - + C!34X4 = 0,
<141:!; + + + ( - k,)x4 = 0
are therefore consistent and uetermine at least one point invariant
under t he transformation.
Let (i = 2, 3, 4)
be t he equati ons of three planes passing through this invariant
point but not belonging to the same pencil, antllet
4
"if3tkxk = 0
k=l
ART. 15-!)
PROJ ECTI\' E TRANSFORMATIONS 193
be the equation of any plane not passing through the invariant
point. If now we put
i = 1, ~ ' 3, 4,
and solve the equations for the x,,
1
~ c , = :S. y,kyk, and put also x', = -:$y,ky'k
I= I
then substitute tl1ese values in the members of (7), the new equa-
tions, when solved for y'., will be of the form
y'. = klyl + b1'iJh + bl'J!h + bl..y4J
Y'z = b-rJh + b-n!h + bz.l!/
Y
1
3 = bszYz + b33?fs + b34Yv
y'.= b42Y2 + b4J!h + b .y .
Without changing the vertex (1, 0, 0, 0), the planes y
2
= 0, Jla = 0,
y
4
= 0 may be replaced by others by repeating this same process
on the last three equations; in this way we may replace the
coefficients b
32
, b
42
by 0; by a further application to the variables
y
3
, y
4
we may replace b
43
by 0.
Referred to the system of coordinates just found, the equations
of the projective transformation (7) are
:t'
1
= k
1
:t
1
+ C
1
zX2 + c
13
:t
3
+ c
14
:t
41
X
1
2 = Cz:Xz + C-nXs + Cz
4
X
41
x'3=
x'=
in which ezz, c
33
, C
44
are all roots of D(k)= 0.
(9)
Equations (9) represent the form to which the equations of any
projective transformation may be reduced. The further simplifi-
cation depends upon the ,-alues of the coefficients, that is, upon
the characteristic (Art. 1 ~ 7 ) of D (k).
If c
34
=I= 0 and c
33
=1= C
44
, make the further ttansformation
XI= y., Xz = ?/2, Xs = Y3 + Cs4Yl ' x. = Y
C44- c33
On making this substitution we reduce the equations of (9) to a
form in which the coefficient c
34
is replaced by zero.
194 TRANSFORMATIONS OF SPACE [CHAP. XII.
In any case, if i < k and c,, =F ckk> we may always remove the
term c,k by replacing x, by x, + c,kxk in both members of the
ckk- c;;
equation. If c .. = ckk and c,k =F 0, by a change of unit point, c,k
may be replaced by unity ; thus, if c
33
= c
44
and c
34
=F 0, by writing

=,?/
4
, we obtain the equations
Y
1
3 = + ?f,
?t'. = Cai!/4
These two types of transformations will reduce the equations to
their simplest form in every case in which D(k)=O has no root
of multiplicity greater than two.
If D(k)= 0 has one simple root k
1
and a triple root k
2
, the pre-
ceding method can be applied to reduce the equations of the
transformation to
x'
1
=k
1
x
11
x'2 = k2x2 + a23Xa + Clz4x.,
x'
3
= k
2
Xa +

x'
4
= k
2
x
4

In case a
24
= 0, the preceding method can be applied again; thus,
if a
34
=F 0, =F 0, each may be replaced by unity; if coefficients
a
23
, a
24
, a
34
are zero, the transformation is already expressed in its
simplest form. If a
24
= 0, either or both of the coefficients a
23
and Cf-a
4
, if uot zero, may be replaced by unity by a transformation
of the type jnst discussed.
If a
24
=F O, a
34
=F O, replace by the substitution
a 1'

a34
In the transforme1l eiJIIation, the new aN is zero. In the same
way, if a
24
=F 0, au = 0, but <1
23
=F 0, put
and the same result will be accomplished. Finally, if

0,
but a
34
= 0, a
23
= 0, put
(10)
in both memhers of the equ;Ltion. Now a
24
=0, and the complete
reduction cau be made as before.
AnTs. 154, 155) FORl\IS OF 'I'RANSI<'ORl\IAT IONS 195
If D(k)= 0 has a fourfold root ku equations (9) reduce to
x'
1
= k
1
x
1
+ a
1
2-'l'
2
+

+ a
14
x
4
,
x'z = k . rz +
11
2:rt3 + azX,
x'
3
= k
1
.r
3
+ a
34
x
41
x'
4
= k
1
x
4

By t ransformations analogous to those in the preceding case, the
coefficients a
13
, aw and a
24
may be reduced to zero, and the cof'ffi-
cients a
1
2J <t
23
, and a
34
to zero or to uni ty.
This completes the problem of reduction. The detel'lltinalion
of the locus of the invariant points and the characteristic of D(k)
in the various cases is left as an exercise for the student. The re-
sults are collected in the following table.
l ou. Standard forms of equations of projective transformations.
CnAHAf'TEIU"Tif"
[1111 J
[112]
[11(11)]

[1(21)]
[ 1(111)]
[22]
[2(11)]
[(11)(11)]
[4]
x'
1
=k
1
x.,
x'3 = k3x3,
:r'2=
x'
4
= k
4
x
4
x'. = k . '\1, a/2 = k2x2
X 3 = l.:3x3 + x., = k3
1
'
x', =k
1
x
1
,
x'3= ksx3,
x', = k,.t.,
.c'
3
= k
2
.t
3
+ :c
4
,
:l.", = k
1
x.,
x's = kiX3 + x.,
x', =k
1
x
11
x'
1
= k2''t:
3
,
x', = k
1
X
1
+ x
2
,
a/3 = kiX3 + x.,
x'
1
= k
1
x
1
+ x
2
,
x'3 = kt':1.'3,
a/
1
=k
1
x.,
X3 = k{Xs,
x'
1
= + rc
2
,
a/3 = k.x3 + x.u
x'2 = 7.; .l2
x'. = k3x<
x'
2
= h'zXz + x
3
x'.=k2x

af
4
=k
2
x
4
x'2 = 1.:2'-ct
x'2 = kzx.

x'. = kzx.
x'
2
=k
1
x
2
x'4 = kc.X
x'
2
= k
1
x
2
x'
4
= k
2
x
4
X
2
= k
1
:t'
2
+ X
3
a/
4
= k
1
x
4
LCH' ti S tH' {S\' AHIAST J'l)JSTS
Four distinct points.
Two distinct, two co-
incident points.
Two distinct points
and a line.
One di,tinct, three
coincident points .
A point and a line.
A point aud a plane. :..
'fwo pairs of coinci-
dent points.
Two coincident points
and a line.
Two lines.
Four coincident points.
196 TRANSFORMATIONS OF SPACE
(CHAP. Xli.
CllARAC'TittttSTlC EQU.\TlOSS
[(22)] x'
1
=k
1
x
1
+Xz, x'
2
=k
1
X
2
x'
3
= k
1
x
3
+ x'
4
=

[ (13)]
x't = .lf! =
x'
3
= k
1
x
3
+ ,1'
41
x'
4
= k
1
x
4
[(112)) x'
1
= k
1
X
11
x'2= k1x2
x'
3
= k
1
x
3
+ x
4
, x'. = k
1
x
4
[(1111)]
x'1 =k1:t:" 'Jf2 = k1x2
x'3 =k1Xz, x'
4
= k
1
x
4
EXERCISES
Locus OJ' hlVARIANT r olST8
A line.
A line.
A plane.
All points of space;
the identical trans-
fonnation.
1. In type [111 1) obtain the necessary ami suflicient condition that the
transformation obtained by applying the given transfonuation p times is the
identity.
2. In (1 ( 111)) show that the line joining any point P to its image P'
alrrays passes through the inmriant point.
3. In J.:x. 2, let 0 be the invariant point, anrl let a line PP' intersect
the invatiant plane in .v. Show that the cross ratio of O.lll'P' is constant.
This transformation is called perspectivity. If the points O.llPP' are har-
monic, it is called involution. .
4. In ((ll)(ll)) show that the line joining any point P to its image P'
meets both invariant lines, and that the cross ratio of P, P' and these point:;
of intersection is constant.
5. Discuss the duals of the types of transformations of Art. 155.
loG. Birational transformations. Besides the projective trans-
formations, we have already met (At-ts. lH, 1-H>) with certain
non-l inear transformations in which corresponding to an arbitrary
point (x) is a definite point ('Jf) and conversely. These are all
particular illustrations of a class of transformations which will
now be considered.
Let
(11)
be four rational integral functions of Xu x.
1
, x., all of the same
degree. \\' hen Xu Xz, x
4
are given, t he values of .1:' h x'
2
, x'
3
, x
4
are uniquely determined, hence corresponding to a point (x) is a
ART. 156)
BIRATIONAL TRANSFOR:.IATIONS 197
definite point (x). If the equations (11) can be solved. rationally
for Xu x2, Xa, x4 in terms of x'l! x3, X.,
(12)
in which all the functions t/t, n.re of the same degree, then to a
point (.ll) n.lso COJ"l'esponds a definite point ( . .:). In this case the
tmnsformation dcfinell by (11) is called birational; that uefincd by
( 1:!) is callou the inve1se of that defined by (11).
When the point (.e') the plane = 0, the cone-
spontling point (x) describes the smface
1t'
1
cp,(x) + u'2c1>kv) + 1t
1
3
cp
3
(.);) + u',cf>lx) = 0. (13)
This surface will be said to ronrspoml to the phne (n'). If the
tt'; arc thought of as p:nameters, we may say: cont>sponding to all
the planP.s of space arc the surfaces of a web delinctl by ( l3).
ln the same way it is seen that, eone::;ponding to the planes
:Su;.l:; =0 of the (.1:), are the surfaces of thu web
Three planes (u') which do not belong to a pencil have one antl
only one point in common, henoc thrce surfaces of the web (13),
which do not belong to a pencil, clotermine a uuiquc point (x)
common to them all, whose coordinfl.tes are functions of the coor-
dinates of (u').
This fact shows that in the case of non-linear t1ansformations
the web defined by (13) cannot be a linca1 combi nation of :ubi-
trary snrface:; of given degree. For if the cp, arc non-linear, any
three of them intersect in more tl1an one point, but it was just
that of the points of intersection there is just one point
whose coordinates depend upon tho particular surfaces of the web
chosen. The remaining intersections arc common to all the sur-
faces of the web. They are called the fundamental points of the
system (x) in the tranformation (11). When the coordinates of a
fundamental point are substituted in (9), the coordinates of the
correspondi ng point (x') all vanish. For the fundamental points
the correspondence is not one to one. The fundamental points of
(x') are the common basis points of the sul'fa<;es cp,(x') = 0.
198
'l'RANi:iFORMA'l'IONS OF SPACE [CUAP. XII.
157 . Quadratic t ransformations. We have seen (Art. 98) that
if the <f>
1
ae linear functions, the transformation (11) is projective,
and that no point is common to a.l l four planes <f>,(x) = 0. The
simplest non-linea.r transformations nrc those in which the </>,are
quadratic. We shall consider the case in which all the quadrics
of the web have a conic c in common.
Let the equations of the given conic be
f(x) =0.
Any quadric of the system

+ .\
3
x
3
+ .\
4
X
4
) + .\s.f(x) = 0
will pass through this conic. Among the quadrics of this system
those passing throngh an arbitrary point P define a web. Any
two quatlrirs 11
1
= 0, II
2
= 0 of this web intersect in a space
curve consisting of the conic c and a secontl conic c' which passes
through P. The planes of c and of c' constitute a composite
quadric belonging to the pencil determined by II
1
= 0 and H, = 0,
and the conics c, c' lie on every quadric of the pencil. Hence c, c'
int.crseet in two points, as otherwise the line of intersection of the
two planes would have at least three points on e''ery quadric of
the pencil, which is impossible.
Any third qu::ulric If.,= 0 of the web but not of the pencil
determined by lf
1
= 0, If
2
= 0 passes through c and P. The plane
of c' intersects 1/
3
= 0 in a conic c" passing through P and the
two points co111mon to c, c' and in just one other point. The posi-
tion of this fourth point of intersection df'pemls on the choice of
the bundle JJ
1
= 0, 11
2
= 0, 1!
3
= 0. We have thus proved that the
web of quadrics defined by a conic and a point P has the neces-
sary property mentionetl in Art. 15G possessed by the web deter-
mined by a bir:ttional transformation.
Let the equations of the conic c be
If P is uot on the plane :t
4
= 0, it may be chosen as vertex
(0, 0, 0, 1) of the tetrahedron of reference. The equation of the
web bas the form
.\
1
x
1
x
4
+ .\
2
XzX
4
+ >.
3
x
3
x
4
+ .\le
1
x
1
2
+ e.f1:
2
2
+ e
3
x
3
2
) = 0.
ART. 157] QUADRATIC TRA:i'\SFOR:\IATIONS 199
In analogy w1th equation (11) we may now put
x'
1
= X
1
X
4
, x'
2
= x'
3
=

x'
4
= e
1
X
1
2
+

+ e
3
x
3
2
(15)
The most general form of the transformation of this type mny be
obtained by replacing by any linear functions of them with
determinant.
In the derivation of equations (12) it makes no uifference
whether the conic c is proper or composite, hence three cases
arise, acc01ding as e
1
= ez = e
3
= 1 or <'
1
= <'
2
= 1, e
3
= 0 or e1 = 1,
e
2
= e
3
= 0. The equatious are
x'
1
=

x'
2
= :t;zx
4
, x'
3
= x
3
x
4
= x
1
2
+ X
2
2
+ x{ (a)

x'
2
=x,x
4
, x'
3
=x
3
x
4

(b)

x'
3
=:r
3
x
4

(c.)
Now let P approach a point J{ on the conic c. If c is com-
posite, suppose its factors are distinnt and that ]{ lies on only
one of them. In the limit, the line/\/' is tangent to all the quad-
rics of the web determined by c and P. But. the tangent to cat /{
is also tangent to all these q uachics at K. Hence t.he plane of
these two lines is a common tangent plane to all the quadrics of
the web at /\ = P.
Let P be taken as (1, 0, 0, 0), the common tangent plane at P
as .t:
2
= 0, and let the equations of the conic be reduced to = 0,
x
1
x, + e:c
3
2
= 0. The equation of the web has the form
A
1
Xz.'t
4
+

+ ,\
3
:cl + ,\
4
(x
1
x
2
+ ex
3
2
)= 0.
The two cases, according as e = 1 ot e = 0, give rise to the
transformations
x'
1
= X
2
X
4
, x'
2
= x3'-v
4
, x'
3
= X
4
2
, x'
4
= X
1
Xz + x
3
2
, (d)
x'
1
= XzX
4
, x'
2
= x
3
x., x'
3
= x/, x'
4
= X
1
Xz (e)
of this type.
Finally, let. c he composite and Jet the point JC which P ap-
proaches lie on both compoueuts of c. Since all the quadrics
th.rongh c have in this case the plane of c for common tangent
plane at 1{, the point P must apptoach c in such a way that the
line KP approaches the plane of c as a limiting position. The
conics in which the quadrics of the web are in tersecte<l h.v any
plane through Pand Jfhave two points in common at /\anrl onP
200
TRANSFORl\IATIONS OF SPACE [CHAP. XII.
at P. Hence in the limit, all these conics must have three inter-
sections coincident at ]{ = P.
Let the equations of c be x
4
= 0,

+ ex
3
2
= 0, and the coordi-
nates of P be (1, 0, 0, 0). The equations of the system of rank
five of quadri cs through c is
'A
1
x'"t
4
+ + 'A
3
x
3
x
4
+ >..,xl + >..,(x.; + e.'tl)= 0.
The section of this system by any plane throngh P, different from
x., = 0, will consist of a. system of conics touching each other at P.
The required web belongs to this system a.ud satisfies the condi-
tion that its section by any plane through P other than x
4
= 0 is a
system of conics having three intersections coincident at (1, 0, 0, 0).
The equations of the section by the plane x
3
= 0 are
'A
1
x
1
x
4
+ A:r'l:
2
X
4
+ >..,xl + Ar.Xz
2
= O, x
3
= 0.
All these conics touch each other at P. Let >..'
11
>..'
2
, >..'v >..', be the
parameters of one conic, and A
11
A
5
of another contained in
this system. The equations of the lines fom (1, 0, 0, 0) to the
two remaining intersections of these two conic are
('A
1
A's- 'A
5
.\'
1
).tl +('A,.\'
2
-

+('A
1
'A'
4
- A
4
A
1
1
)xl = 0.
One of these remai ning points is also at P if .\
1
.\'
5
- .\
5
.\'
1
= 0.
Hence all the quadrics of the web satisfy a. relation of the form
.\
5
+ k.\
1
= 0. J t is no restriction to put k = 1. It can now be
shown that the conics cut from the quarlrics of the web As+ .\
1
= 0
by any plane a,x, a
3
x
3
= 0 through P have three coinci-
dt'nt points in common at P.
The ecl'.:tLion of the web is

+ 'Alxl + e:ra
2
- x
1
x
4
) = 0.
The t.wo birational transformations defined by webs of quadrics
of t!J is type are
x' , = .l;
2
X
4
, :t/2 = :&;
3
.1"
4
, .1:'
3
= X
1
2
, x', = Xz
2
+ x
3
2
- X
1
X
4
(f)
.'t'
1
=.t2"t,,

.c'
3
=x/,

(fJ)
The inverse transformat.ions of forms (<t) (g) are also quadratic.
For this reason transformations of this type are called quadratic-
q na.dratic.
ARTS. 158, 159] RECIPROCAL RADII 201
158. Quadratic inversion. A geomet.,ic m<'thod of constructing
some of the preceding types of birationaJ transformations will
now be considered. Gi,en a quadric A and a point 0. Let P be
any point in space, and P' the point in which the polar plane of
Pas to A cuts the line OP. The transfonnation deli ned by hav-
in:; P' correspond to P is called quadratic inversion. If 0 does
not lie on the qunllric A= 0, let 0 = (0, 0, O, 1) and let the equa-
tion of .A = 0 be
If P=:(y., y
2
, y
3
, y.), the coorllinates of P' are
y'.=y.y., Y'2=Yi!J., Y'a=Ya.lfu Y 4= eYt
2
+ e'l!h
2
+ eaxl,
which include forms (ct), (b), (c). lf 0 lies on A, we may t.c1.ke
.A = :c,_2 + e
3
x
3
2
- :t'
1
X
4
= 0, 0 = (0, 0, 0, 1 ).
The coordinates of P' in this case are functions of y., y
2
, y
3
, y
4
defiued by (/) and (f!) The quadratic-quadnttic transformations
(a), (b), (c), (!), (y) can therefore be generated in this manner.
159. Transformation by reciprocal radii. lf, for the quadric
A= 0 (A1t. 158) we take the sphere
(lG)
and for 0 the center of t his sphere, the equations of the trans-
formation assume the form
x' = k'lxt, y' = k
2
yt, z' = k'zt, ( = 'J..a + y
2
+ :r!-. (17)
On account of the relation
OP OP'=k
2
(18)
existing between the segments from 0 to any pair of correspond-
ing points P, P', it is called the t ransformation by reciprocal radii.
Any plane not passing through 0 goes into a sphee passing
through 0 and the circle iu which the given plane meets the
sphere ( l G), which is called the sphere of inversion.
The fundamental elements are t he center 0, the plane at
infinity, and the asymptotic cone of the sphere of inversion.
202
[CRAP. Xll.
A plane ux + uy + cz + rlt = 0 not passing through the origin
(cl * 0) is transformed into a sphere
ak2xt + :,7-;
2
yt + cl.ht + d(.t
2
+ y
2
+ z
2
) = 0
. passing through the origin.
A plane passing thrnngh the origin is transformed into a com-
posite sphere consisting of the given plane and the plane at in-
finity. \\' e shall say t hat phmes tiHough the origin are trans-
formed into t.hemsdves.
A sphere
a( + y
2
+ z
2
) + 2 fct + Z gyt + 2 ltzt + mt
2
= 0 (19)
not passing through the origin (m * 0) is transformed into the
sphere
m(x
2
+ y
2
+ z
2
) + Jk
2
;ct + 2 gk
2
yt + 2 hlht + uk' = 0. (20)
The factor x
2
+ !l + z
2
can uc remo,eu from the transformed
equation. A sphere passing through the origin (m = 0) is trans-
formed into a composite sphere consisting of a plane and the
plane at infinity.
If any Slll-face passes through the origin, its image is seen to ue
composite, one factor being the plane at infinity. The plane at
infinity is the image of the center 0, which is a fundamental
point.
Iu particular, the sphere (l!l) will go into itself if ?It= ak
1
;
but this is exactly the condition that the sphere (19) is orthogonal
to the sphere of inversion, lwnce we may say :
'l'IIIWRF.)I I. The uhich are orthogonal to the sphere of m-
vel'.sion (JO i11t0 wlten lta11Sf0?'111ecl uy ?'eCiprocal ?adii.
We shall now prove the following theorem:
TI. A11gles are J)?'eserve(l wlten transfvrmecl by recipro-
cal 1adii.
Let A
1
.t + B
1
y + 0
1
z + D
1
t = 0, A
2
x + Bi!J + + D
2
t = 0
be any two planes. The angle 0 at which they intersect is de-
fined by the formnla (Art. 15)
cos O= A.:lz+B.Bz+O.Oz (21)
,/ (..(1
1
2
+ B
1
2
+

+ B
2
2
+ Ozt)
1.39, 1(>0] CVCLIDES
These planes go into the spheres
D
1
(x
2
+ y
2
+ z
2
) + A/.:
2
xl + B
1
k
2
!tl + 0
1
k
2
zt = 0,
D
2
(Jil + y
2
+ z
2
) + A
2
k
2
xt + B
2
k
2
yt + 0
2
k2zl = 0.
203
Since the angle of intersection of two spheres is the same for
every point of their cur\'e of intersection (Art. :)1) and both
spheres pass through 0, we may determine the angle at which the
spheres intersect by obtaining the angle between the tangent
planes at 0. These tangent planes are

+ B
1
!J + G
1
z = 0, A
2
x + Bz!t + 0zZ = 0,
hence the angle between them is defined by Since the angle
of intersection of any two snrfaces at a point lying on both is de-
fined as the angle between their tangent planes at this common
point, the proposition is proved.
160. Cyclides. Since lines are transformed by reciprocal radi i
into circles passing through 0, a mlctl surface will be transformed
into a Stll'f<tce containing an infinite number of circles. A quadric
has two systems of lines, hence its transform wi ll contain two sys-
tems of citcles, and every circle of each system will pass through
0. )[oreo\'er, the qumhic contains six systems of circular sections
lying on the planes of six patallel pencils (Art. 82). Hence the
transform will also contain six additional systems of circles, not
passing through 0, but so situated that each system lies on a
pencil of spheres passing 0.
By rotating the axes (Art. 07), we may rednr.c (Art. 70) the
equat.ion of any quatlt-ic not passing through 0 to the form
+ by
2
+ cz
2
+ t
2
+ lxl + 2 myt + 2 nzt = 0 (22)
without changing the form of the equation of the sphere of inrrr-
sion. By transforming this surface hy reciprocal radii, we obtain
+ ?/ + z
2
)2 + 2 k
2
(x
2
+ y
2
+ z
2
){1.1; + 111?1 + nz)t
+ k
4
(a.Cl + by
2
+

= 0.
This surface is called the nodal cycl irle; it contains the absolute
as a double curve and has a double point at the point 0."'
*A point P on a surface is called a double point or node whetl every line
through P meets the surface in two coinciolPnt points at P. A curve 011 a surface
is c:11lod :1 double cune wben every point of tho curve is a tlouble point of the
surface.
204 TRANSFORMATIONS 01<' SPACE (CHAI'. XII.
If the given quadric is a cone with vertex at P, its image will
have a double point at 0 and another at P'. The cirrles which
are the images of the generators of the cone pass thnmgh 0
and P
1

The equation of the cone may be taken as
a(x - .ft)
2
+ b(y - ut)
2
+ c(z - ht)2 = 0 (23)
and the equation of the tr:tnsform is
(ap + brf + ch
2
)(x
2
+ + z
2
)2 - 3 k
2
t(x
2
+y
2
+z
2
)(af.1: + bgy + chz)
+ k
4
(ax
1
+ by
2
+ cz
2
)t
2
= 0.
This surface has a node at the origin and at the transform
(.(, g, h, P+ + h
2
) of the vertex of the cone (23). It is called a
binodal cyclide.
If, in equation (23), b= a, so that the given quadric is a surface
of revolution, the tr<tnsformed equation may be written in the
form
[x
2
+ y
2
+ z
2
+ k
2
(lx +my+ nz)t + 1.,.4t
2
]
2
+(a- b)k
4
x
2
t
2
2
- k
4
(lx + 1ny + nz + k
2
t)
2
t
2
= 0.
It has a node at 0 and at the points in which the line x= 0,
3lx + 21n!J + 2 nz + k
1
bt = 0 intersects the sphere x
2
+ y
2
+ z
2
+ 21-a:t
+ 2myt+ 2 nzt + bk
2
t
2
= 0. It is called the trinodal cyclide.
Finally, if the cone (21) is one of revoluti on, the resulting
cyclide has four nodes, and is call ed a cyclide of Dupin. If the
center of inversion is within the cone, so that no real tangent
planes can be dra.wn to t he cone through the line OP, the surface
is called a spindle cyclide; if the center is outsiue the cone, the
resulting surface is called a horn cyclide.
The generating circles of a cone of revolution intersect the recti-
linear generators at right angles. Since both the l ines and the
circles arc transformed into circles ant! angles are preserved by
the transformation, we have the following theorem:
TJ T. Through euch poi11t o.f rt C!Jdirle of flttpin ]JaS!;
two drrlPs lyi11Y mtirely on the meet each
othn at riyltt
ART. 160) CYCLIDES 205
A particular case of the spindle cyclide is obtained by taking
the axis of the cone through the center of inversion. The result-
ing cyclide is in this case a surface of revolution. It may be
generated by revolving a circle about one of its secants. If the
points of intersection of the circle and the secant are imaginary,
the cyclicle is called the ring cyclide. It has the form of an
anchor ring. In this case all the nodes of the cyclide ate
imaginary.
EXERCISES
1 . If A consists of a pair of non-parallel planes and 0 is taken on one of
them, show that the quadratic inversion reduces to the linear transformation
defined in Art. 15.), Ex. :J as central involution.
2. Obtain the transform of the ellipsoid

c}
\\ith regard to the sphere ,.z + !lz + zZ = I. How many systems of circles
lie on the resulting surface? Show that four minimal lines pass through 0
and lie on the surface.
3. Show that the transform of the paraboloid ax2 + byZ = 2 z by reciprocal
radii is a cubic surface. Ilow many systems of circles lie on this surface?
How many straight lines ?
4. Discuss the trans form of a quadric cone by reciprocal radii when the
center of the sphere of inversion lies on the surface but is not at the vert-ex.
5. Show that a surface of degree n passing k times through tbe center of
in\ersion is transformed by reciprocal r:tdii into a surface of degree 2{ n - k), ' ,.
having tbe absolute as an (n - k)-fold curve.
6. Show that the center of an a rbitrary sphl're not transformed into
the center of the transformed sphere by r<:ciprocal radii.
7 . Given the transformation
X
1
1 = (Xl- =(:r, - :r',. = (.cl- rz).rl, r
1
4
= :r
2
.r,..
:Find the equations of the imer!;e and the IJ:t8is points
in (:t).
8. Given the transformati on
X
1
t = XtXz. ,.,'2 = :t2.:rz, "':s = :r'-t = .t.a(Xl +X'.!+ :ta).
Find the equations of the ilwerse transformation. the basis points
in the web of quadrics :>.
1
:r
1
xz + X2:r1' l + + x.x.(xl + .r
2
+ :ra) = 0.
CHAPTER XIII
CURVES AND SURFACES IN TETRAHEDRAL CoORDINATES
I. Algebraic Surfaces
16 t. Number of constants in the equation of a surface. The
loeus of the equation
!()
n! afJy6 0
X = :S Xa X4 = 1
tt !{3! y!S!
(1)
wherein ((. {3, y, S are positive integets (or zero) satisfying the
eq nation (( + (3 + y + S = n, is ca.lled an algebraic surface of
d<'gree n.
If the equation is arranged in powers of one of the variables,
as x
4
, thus
(2)
in which 11, is a homogeneous polynomial of degree i in Xa,
the number of constants in the equation can be readily calculated.
l!'or we may write
11; = <f.oxs' + <f.I:J:/- 1 + ... + q,,,
q,k being a homogeneous polynomial in x
1
, of degree k and con-
sequently containing k + 1 constants. The number of constants
in u, is t herefore
1 + 2 + ... + i + 1 = ( i + 1) ( i + 2) = ( i + 2) ! .
2 i! 2!
This uumbe1 of constants in u, is now to he summed for all inte-
gral va.lues of i from 0 to n. By in<lnction the snm is readily
found to be
1
+2 3+3-4+ ...
2 2 n! 3 !
which is the number of homoge1\eou:s in the equation
of the surface. T he number of intlependent couc.litions which
the surface can satisfy is one than this 01
(n + 3)! _
1
= 1<
3
+ (i 11
2
+ 11 n .
n! 3! 6
20G
ARTS. 161-163] IXTERSECTION WITH A LINE 207
162. Notation. It will be convenient to introduce the follow-
ing symbols :
!( ) = iJ{(x) + iJjfE_ +
1
+ O.ftx) =
" X - Y i) Y2 , Ya , Y4 ,
l; v.C
2
vX3 l1X
4


y.--
iJx,
(3)
wherein 1 z 1' < n and a, b, c, cl are positive integers (or zero),
satisfying the condition a+ b + c + cl = 1.
EXERCISES
Let j(:c) = ll4oooX
4
+ao4ooX2
4
+aooo:t3
4
+aooox'+O a2wox.
2
x2
2
+6ao22o:t2
2
Xa
2
+ 6 ao2o2X2:!zil + u + G

+ 6

1. Find tl.?f(.c) for r = 1, 2, 3,
2. Show that c.. =
3. Show that ..!.. .:..3 j(Y) = A,j(y).
31
4. Show that f (x) = .1,2 f(y) .
5. Show that AJ(.c) t!J.,2j(x) = 12/(.c); c.,af(.c) = 24f(x).
1 G3. Intersection of a line and a surface. If (y), (:e) are any two
points in space, the coordina.tes of any poi nt (z) on the line j oi ning
them me of the form z, = A!f, + p..t, (Art. 95). ff (z) lies on
f(x) = 0, then j(>..y + = 0. By Taylor's theorr m for the
expansion of a function of four vari ables, we have, since
f(y) a 0 for all positive integtal values of k,
f(>..y + ,.... '1;) = >..f(y) + + ...
+ >.,-p. f(y) + ... + p." f (?J) = 0.
r! n!
This equation may also be wri t ten in the form
f(>..y + ,....1;) = p." f(x) + p."

+ ..
+ p.->., f(x) + ... + A" = 0,
r! n!
which is equi,alent to the preceding one.
(4)
(5)
208
CURVES AND [CUAP. XIII.
lf these equations are identically satisfied, the line joining (y)
to (:t;) lies entirely on the surface. If they are not identically
satistied, they ate homogeneous, of degree n in >.., JJ. and conse-
quently deterllline n intersections of the line and the surface. If
we defiue the order of a surface as the number of points in which
it is intersecteJ by a line, we have the following theorem.
'l'HEOI":E;\t. The order of a surface is the degree of its equation
point coorclinate.s.
164. Polar surfaces. In (5) let the point (y) be fixed but let
(x) vary in such a way that the equation
is satisfied..
e:.; f (x) = 0 (6)
This equation defines a surface of order n-r calleJ the rth polar
surface of the fixed point (y) with tegard to the given surface
f(x) = 0. When r = n -1, the snrface (6) is a plane. It is
ca.! led the polar plane of the point (.II) as to f(x) = 0; when
,. = 11- the resulti ng quadric defined by (6) is called the
polar quadric, etc.
ln the identities ( 4) and \5) the coetlicieuts of like powers of
A, JJ. are equal, that is,
l. e:.; f(x) = 1 e:..- f(y).
r! (n- r)!
From this identity we ha,e the following theorem:
I. If lies on the 1th }Jolw of (y), then (y) lies on
the (n -1)th 1Jolw
If in ( 4), the two points (y), (x) are coincident, then
f(.\x + JLX) = (>.. + JJ.)"f(x) = >.. f(x) + .\"-
1
JJ.C:.J(x) + ...
+>.."-'JJ.' C:.z'f(x) + ....
,. !
By expanding (>.. + JJ. )" by the binomial theorem and equating
coefficients of like powers of >.., JJ. in the preceding identity, we
obtain
n'
C:.z' f(x) = f(x),
(11 - 1') !
ARTS. 164, 165] TAKGEKT LINES AND PLANES 209
which is called the geueml ized Euler theorem for homogeneous
functions. From this icleut,ity we h:we the following theorem:
Tm:oJHDr II. The locus of a JlOillt u:hidt lies on any one aMl
therefore on all ils otw polw sto;fi.tces is the giuen sn;:{ace f(.t) = 0.
From the definition of t:.;.f(.c) (Art. 162) it follows that if
s < r < 11,
l<'rom this irlf'ntity we h:we the theorem:
TnEORFm TIL The slit 11olar sn:for-P of the ( r - s)th zwlw
face of(!!) with to f (:c) = 0 coi11cide.s with the 1th 1)0/ar
fnce of (y) .
EXERCISES
l . Determine the coordinates of the points in which the line joining
( 1. 0, 0, 0) to (0, 0, 0, I) intersects the surface
.r, +:! ;rz'
1
- xa- x,
3
+ 4

+ i'o xz
2
.,3 - ll x1.>"2X3 = 0.
2. Determine n so that two intersections of the line joining ( 0, 1, 0, 0)
to (0, 0, 1, 0) with the snrfaco
:r1 + .r2
1
+ xa + :rt
1
+ axb
3
+ 2 (a - l).r2
2
.ra
2
+ 4 x2:r3
3
+ 6 x,:r2:r3.t1 = 0
coincide.
3. Show that any line throui:h ( 1, 0, 0, 0) has two of its intersections
with the snrfare
3 :r
2
'l,r
3
2 + + 6 .Y
1
'l.rl + 12

+ 24 :t
1
.t
2
.r
3
x = 0
coincirlent at (1, 0, 0, 0).
4 . Prove the theorems of Art. 161 for the surface of Ex. 3 by actual
differentiatiou.
165. Tangent lines and planes. A line is said to touch a sut-
face at a point P on it if two of i ts intersections with the surface
coincide at P. In equation (.t) let (y) now be a fixed point on
the given surface so that A!t) = 0. One root of ( ..j.) is now p. = 0,
and one of the inte1sections (x) coinrides with (y) .
The condition that a second intersection of the line (!t) (x) coin-
cides with (y) is that p.
2
is a factor of ( -l ) , that is, that. (x) 1s a
point in the r,lane
uJ(y) = x
1
+ + 3:
3
+ iJJ(y) = o. (7)
iJy
1
u.rh uy
3

All the lines which tonrh .f(x) = 0 at (y) liP. in the plane (7) and
every line through (y) in this plane is a tangent line. This plane
210 CURVES AND SURF ACES [CHAP. XIII.
is called the tangent plane of (y). T he plane (7) is also the polar
plane of (y); hence we have the theorem:
The polar plaue of a point P on the surface is the
tauyent 11lune to the sm:fc.tce at P.
From Art. 16-J., Theorem III it also follows that the tangent
plane to .fl..:c) = 0 at a point (y) on it is also the tangent plane at
(y) to all Lhe polar surfaces of (y) with regard to f(x) = 0.
1 HO. Inflexional tangents. A line is said to have contact of
the see0111l order with a surface at any point P on it if three of
its intersections with the surface coincide at P.
Let (y) be a given point on the surface, so that f(y) = 0. The
condition that the line (y)(z) has contact of the second order at
(y) is that p:
1
is a factor of (..J.), that is, that (z) lies on the tangent
plane and on the polar quadric of (!J). Hence (z) lies on the
intersection of
u..f(y)= o, u /f(?J) = o.
Since u,.f(y) = 0 is the taugent plane of the quadric t:..,
2
f(Y)= 0
at (!/), the conic of intersection of the tangent plane anti polar
qnathic consists of two lines, each of which has contact of the
second nrtlrr with .f(.t) =I) at the point (y). These two lines are
called the inflexional tangents to the smface at the point P. The
section of the sur face by an arbi trary plane thtough either of
these lines has an inflexion at (y), the given line being the inflex-
ional tangent.
1 H7. Double points. A point P is said to be a double point or
node on a surface if every li ne through the point has two inter-
sections with the surface coi ncident at P. lf (y) is a double
point on .f(x) = 0, equation (l ) has p.
2
as factot for all positions
of (z), that is, t:.. ,j(y) = 0 is an identity in z" z
2
, zJ, z
4
It follows
that if (y) is a double point, its cothtlinates satisfy the four
equations
CV(?t) = o, = o, a.f(Y) = o u.f(?t) = o. (8)
U?JI ayz U!Ja '
Conversely, if these conditions are satisfied, it follows, since
= uvfC!t ), that eq nation (-!) has the double root p.
2
= 0 and
ARTS. 165-168] FIRST POLAR SURFACE 211
(!f) is a. double point. Hence t he necessary and sufficient condi-
tion that f(x) = 0 has a double point at (y) is that the coordinates
of (y) satisfy equations (8). Unless the contrary assumption is
stated, it will be assumed thatf(x) = 0 has no double points.
EXERCISES
1 . Impose the condition that the point (0, 0, 0, 1) lies on the surface
f(x) = 0 and find the equation of the tangent plane to the surface at that
point.
2 . Determine the condition that the snrface of Ex. 1 has a donble point
at (0, 0, 0, 1).
3. Show that the point ( 1, 1, 1, 1) lies ou the surface of Ex. 1, Art. HH,
and determine the ettualion of the t:w!:ent plane at that point.
4. Find the equations of the inlh:xional tangents of the surface of Ex. 1,
Art. 164, at the point ( 1, 1, 1, 1).
5. Show that the lines through a double point on a surface f(x) = 0 which
have three intersections with the surface coincident at the double point form
a quadric cone.
6. Show that there are six lines through a double point on a surf;tce
f(x) = 0 which h:we four poi nts of iutersection with the surface coincident
at the double point.
7. Pro1e that the curve of section of a surface by any tangent plane has a
double point at the point of tangency, and the inflexional tangents are the
tangents at the double point.
168. The first polar surface and tangent cone. If in equation (7),
the coordinates x
1
, ~ . x
3
, x. ate regarded as fixed, and y
1
, y
2
, y
3
, !1
4
as variable, the l ocus of the eq uatiou is the first polar of the
point (x).
TnEonEM I. The .first polar s!11fr.tce of any J?Oint in space p a s s e . ~
tluough all the dollble point.< of the uhen swjace.
For, if .f(x)= 0 has one or more double points, the coordinates
of each must satisfy t he system of equations (8) and also (7).
TnEORE:II II. The poi1tts of tuugency of the tcwueut plrtnes to t!te
surface from a ]Joint (:c) lie 011 the cw Le of intersection of the given
surface ancl the first polar of (.v).
For, if (y) is the point of tangency of a. tangent pl ane to
f(x)= 0 which passes through the given point (x), the coordi-
212 CURVES AND SURFACES (CHAP. XIII.
1mtes of (y) ::;atisfyf(y)=O aml j.J(!J)=O. Conversely, if (JJ)
is a non-multiple point on this cmve, it follows t hat t he tangent
plane at (Jj) pn.sses t hrough the given point (x).
Since the line j oining (.t:) to any point (y) on the curve defined
in Theorem 11 lies in the tangent plane at (y), it follows that it is
a tangent line. The locus of these lines is a cone called the
tangent cone from (x) to the surtwe .f(x) = 0. To obtain the
equation of this cone we think of as fixed in (4) and impose
the condit ion on (!f) that two of the roots of the equation in A.: p.
s hall be coincident. Hence we have the following theorem :
III. The er1uation of the tangent cone .from any poi1tt
(x) is obU,inecl by equatiny the cliscdmillltnt to zem.
160. Class of a surface. Equat ion i n plane coordinat es. A point
(x) lies on the smface .f(x) = 0 if its coordinates satisfy the equa-
tion of the surface. Similarly, a pla.ne (u) touches the surface if
its coonlinates satisfy a certain relation, called t he equation of
the surface in plane coordinates. T he class of a surface is the
dual of its order; it is defined as the number of tangent planes to
the surface that pass throngh an arbitrary line and is equal to
the degree of the e<1uation of the surf:we in plane coordinates.
T11 EO The class n.f an cdyebmic stu:( ace o1cler n, having
o fl ouble poi11ts and no othe1 singularities, is n (n -1 )2- 2 o.
Let f(x)=O be of order n, a.nd let P
1
=(y), P
2
=(z) be two
points on an arbitrary line l . The point of tangency of every
taugent plane to.f(x)=O that passes through l lics on the s mfa.ce
f(:c) = o, ou the polar of (.11) and on the polar of (z), so that its
col>rdinn.tes satisfy the eq nations
.f(x)=O, !:1J(x)= 0, 1:1./(x)=O.
These smf:wes are of orders n, n -1, n- 1, respectively, and
have n(n- 1)
2
points in common; if .f(.t:) = 0 has no donble points,
each of these points is a point of of a plane through the
line l, tangent to the given surface. If f(x)= 0 has a. double
poiut, t1J(x) = 0 aml = 0, both pass through it, hence the
number of r<:'mainiug sections is reJuccd by t wo.
If the plane 1c
1
x
1
+ u
2
.t:
2
+ u
3
x
3
+ u,x, = 0 is tangent to f(x)= 0
ARTS. lGS-170] THE HESSIAN
213
at (y), then this plane and that defined by equation (7) must be
ic.lentical, hence
pu, = () f(J!) I p!l2 = () f(y)' pU3 = () f(y)' pU4 = () f(y)_ (9)
ay, ay
2
ay
3
ay.
)[oreover, (y) lies in the given plane and also on the given sur-
face, hence uly, + Uz!h + u3y3 + u.y. = O, f(y)= 0. (10)
If between (9) and (10) the coordinates of (y) are eliminated, the
resulting equation will be the equation of the given surface in
plane coordinates. If f(x)= 0 has double points, the resulting
equation will be composite in such a way that the equation of
each double point appears as a double factor.
EXERCISES
1. Determine the equation of the tangent cone to the surface
x,s + x
2
S +

+ x
4
3
= 0
from the point (1, 0, 0, 0).
2. Write the equation of the surface of Ex. 1 in plane coordinates.
3. Write the equation of the surface
X1X2X3 + T1l'3X4 + T1T2:r4 + X2T3X4 = 0
in plane coordinates.
4. Write the equation of thtJ surface .c
1
2.r
3
- .c
2
2-c. = 0 in plane coordinates.
170. The Hessian. The locus of the points of space whose
polar qnad1ics arc cones is called tho Hessian of the given sur-
face f(.c) = 0. 'l'he equation of the polar quadric of a point (x)
may be written in the fo1m
"'()2 f'( X)
a . N.Yl = o,
x, .tk
(11)
in which y
1
, !h y
3
, y
4
are the CUI'I'cn t cool'llinates. This quadric
will be a cone if its discriminant \'anishcs (Art. 103), hence if we
put for ure\ity - iJ2f(:c)
.r.k = :. !l - '
u.l',v.t"t
the equation of the Hessian may be written in the form
.fu /.2 /13 J..
}[ = ;;2 ;;2 /23 ;;. = 0.
.f,3 f 23 ha .fa.
J.. /34 fu
It is of order I (n-
(12)
214 CURVES AND SURPACBS [CnAP. XIII.
It will now be shown that the Hessian may also be defined as
the locus of double poi nts ou first polar s urfaces of the given
surface. The equation of the first polar of 0/) as toj(x) =O is
= 0.
iJ.t,
If t his surface has a double point, t he coordinates of the double
point make each of its first partial tlt>rivativPs vanish, by(8), thus
iJ
2
rt x \ [)2 ( fPJ iJ2f
Y1 + Yt __._ + Y3 - - + Y4 - -- = 0,
iJx
1
2
u.t:
1
Clx, iJ.l'
1
CI.c
3
u.I\OX
4
ri2 r iJ2f ;p r i)2 r
Y1 __::_,[__ + Y2 - + Y3 - - + Y4 - - - = 0, (13)

Clx
2
2




iPJ a2J a2 r iJ2f
+ y. --=0,
Clx
1
CI:ta iJx
2
CI.t
3
iJx
3
2
iJ3.
3
iJ.c
4
a2 r iJ2 r rJ2 r (12 r
-+y2 -- - + !Js -- - + !!4 -'- = 0.
axlux. a.t2u:r;4 a.iil.t4 ax4
2
The condi ti on that these equations in .1/u y
2
, Ya
1
y
4
are consistent
is that their determinant is equal to zero, that is, that (x) lies on
the Hessi:tn.
171. The parabolic curve. The curve of intersection of the
given surface with its Hessian is call ed the parabolic curve on the
surface.
' ruEolum. At any poh1t of the l'W'Ctbolic cmve the tu;o injle.tional
ta11gents to th e Sltljuce coi ncile.
For, let (.c) be a point on the parabolic curve. Since (x) lies on
t he Hessiau, its polar IJ\H:ttlric is a cone. This cone passes t hrough
(x) (Art. 1GJ). The inflexional tangents are the lines which the
cone has in common with its tangent plane at (x) (At't. 166).
'l'hese lines coincide (Art. 121).
172. The Steinerian. I t was jnst seen t hat t he polar quadric of
any point on the Hessian is a cone. Let (x) be a. point on fl, and
(y) the vertex of its polar quadric cone. As (x) describes ll, (y)
also describes a snrface, called the Steinerian of f(x) = 0. The
polar quatll'ic of (.t) is given by equation ( 11). 1f (y) is the ver-
tex of the cone>, its noorcliuates satisfy (1:3). The equation of the
AnTs. 170- t 73] ALGEBRAIC SPACE CURVES 215
Stcinerian may be obtained by eliminating Xu x
2
, a
3
, x. from these
fom equations (13). As Lhe equations (1:3) were obtained by im-
posing the condition that the fitst polar of (y) has a double point,
we may also define the Steinerin.n as the locus of a point whose
first polar surface has a double point (lying on t he Hessian).
EXERCISES
1. J>rove .that the Hessian and the Stcinerian of a. cubic surface coincide.
2. Prove that if a point of the Hessian coincides with its corresponding
point on the Steineria.n, it is a double poiut of the given surface, and cou-
versely.
3 . Determine the equation of the Hessian of the surface
lltXta + n2:r2a + aa . + + + .r2 + xa +

= 0.
4 . Determine the order of the Steinerian of a general surface of order n.
II. Algebraic Space Curves
173. Systems of equations defining a space curve. A curve
which forms the complete or partial intersection of two algebraic
surfaces is called an algebraic curve; if the curve does not lie in
a plane, it is called a space curve.
If a given curve 0 forms the complete intersection of two sur-
faces F'
1
= 0, F
2
= 0, so that the points of 0, and no other points,
liE' on both surfaces, then the ec1uations of these surfaces, consid-
ered as silllultaneons, will be called the equations of the given
curve.
If the intersection of the surfaces F
1
= 0 and F
2
= 0 is composite,
and 0 is one component, the eq nations F
1
= 0, F
2
= 0 are satisfied
not only by the points of 0 bnt also by the points of the residual
cnne. If a F
3
= 0 through 0 ca.n be found which has no
points of intersection with the residual curve except those on 0,
the simultaneous equations F
1
= 0, F
2
= 0, F
3
= 0 are satisfied only
by the points of 0 and are called the equations of the curve.
l f the surfaces F
1
= 0, = 0, = 0 through 0 have one or
more points in common which do not lie on 0, then a fourth sur-
face F, = 0 can be found through 0 which does not contain these
residual points, but may intersect the resitlual curve of F
1
= 0,
F
2
= 0 in other points not on F
3
= 0; in this case the simultanl!ous
216
CURVES AND SURFACES [CBAP. XIII .
cq uations F
1
= 0, F
2
= 0, F
3
= 0, F.= 0 a re cal led t he equati ons of
the curve. In thi s way a system of equations cau be found which
a re simultaneously satisfi ell hy points of G anu by no others.
As au illustration, consider the composi te intersection of the
quadric surfaces
:t
1
:t
3
-

= 0, x
1
x
4
- xl= 0.
1 t consists of a space curve anu t he I i ne = 0, x
3
= 0. The
s urface x
1
:t
4
- x
2
x
3
= 0 al so contains the space curve since it
contains every point common t o the quadrics except points
on t he line x
2
= 0,

= 0. These t hree surfaces arc sufficient
t o de fine the curve. The surface x
1
x
4
(x
1
- x
4
)

+ :t
2
x
3
x
4
= 0
a lso contains t he given curve. [t does not, however, with
t he two given surfaces constit ute a system whose equations
define the given cm vc. All t hree equations a re sat isfi ed, not
only by t he Clonlinates of the points of the curve, but by the
coordinates of t he point (1, 0, 0, 1) which docs not li e on the
cu rvc, si uec i t does not I ic on t he sm face .v
1
.c
1
-

= 0. The s ur-
face

+ .r
4
) - .t} - x
3
3
= 0 passes t hrough the curve but not
t hrong!. the poi nt ( 1, o, 0, 1). The r.nrve is t herefor e complete ly
uefi ned IJy rcgar<li ng t he fonr eq nations
x . l:a - = 0,
x
1
x
4
(x
1
- x
4
)- x
2
3
+

= 0,
as simul taneous.
Xa
2
= 0,
x
1
x
4
(x
1
+ x
4
)-

x.
3
= 0
17+. Order of an algebraic curve. Let FIL = 0, FIL. = 0 be two
surfaces of orders p., p.', respectively, and l et G be t heir ( proper
or composite) curve of intersecti on. Any pl ane that docs not con-
t ain G (ot <'l. component of i t) intersects G in p.p.' poiuts. F or,
a.ny such plane intersects FIL = 0 iu a cm ve of onler p., a.nu inter-
sects h',.. in a. curve of Ol'der p.'. These coplanar cm ves have
precisely

points in common.*
I t can in fact be shown that every al gebraic curve, whether
defined as t he complete inte .. section of two surfaces or not, is
intersecteu by a.ny two planes, neither of which contains the
curve or a component of it., in t he same number of points.t \Ve
'*Sec, e.f!., Fine: Alg('hra (1005), p. 5t!l.
t llalphcn : J our. tie l'ccole polytcchniq ue, Vol. 52 (1882), p. 10.
ARTS. 173- 1751 PROJECTING CONES 217
shall assume, without proof, the truth of this statement. The
number of points in which an arbitrary plane intersects an alge-
braic curve is called the order of the curve (Art. 140).
I
1 75. Projecting cones. If every point of a space cu1ve is
joineu by a line to a fixed point P in space, a cone is defined,
called the proj ecting cone of the curve from the point P. If the
point P n.t infinity, t he projecting coue from P is a cylinder
(Art. -14). Except in metrical cases to be discussed later we
shall make no distinction between cylinders and cones.
For an arbitrary point Pan arbitrary generator of the project-
ing cone intersects the curve in only one poi nt. It may happen,
however, for particnhtr positions of the point P, that every
generator meets the cntve in two or more points. 1f in
this case P does not lie on the curve or if P lies on the
curve and every generator through P iutersects the curve in
two or more points distinct from P, the curve is called a conical
curve.
Let P be a point not on the curve, such that an arbitrary
generator of the projecting cone flom P meets the curve iu just
one point. The order of the projecting cone is the number of
generators in an arbitrary plane through its vertex. Each gener-
ator contains one point on the Clll've, hence the order of the pro-
jecting cone is equal to the order of the cmve. If P is on the
curve, the order of the projecting cone is one less than the order
of the curve.
To ji.ud the equation of the Jll'Ojecting cone of the
simple m composite ctm;e cleji.nPcl U!! the complete 'iutersection of t-wu
Sttlfaces, .from a 1Je1tex Of the tetra/teclton vf refereuce, eliminate Ue-
the equations the 'tXtriaule which cloes not 't:ctni:;h at that
certex.
Let the equations of the given surfaces be = 0 and = 0
and let it be required to project the curve of intersection of these
surfaces from the point (0, 0, 0, 1) .
Let ('!/) be any point of space. The co6rdinates of any point
(x) on the line joining (0, 0, 0, 1) to (y) are of the form
X1 = >..yu = >..y2, Xs = >..ys, x. = >..y. + rr.
218 CUR\'ES AND SURFACES [ CHAP. XIII.
The points in which this line intersects F" = 0, F,... = 0 are de-
fined by
F,..(x) = F,..(>..y., >..y
3
, >..y
4
+ u)
= >..,..F"(Yu YZJ Y3, Y4+ 0,
F,...(x) = F,...(>..y
11
>..y
2
, >..y
3
, >..y, + u)
= AJi'F,..{y., Y2, YJ, y, +0=0,
(14)
respectively. The condition th;Lt the line intetsects both
in the same point is that these equations have a common root in
hence the equation of the projecting cone is obtained by elimi-
nating between these two equations (cf. Art. 44).
>..
If is elimi-
>..
nated from (14), y
4
is also eliminateJ au<l the resulting surface is
identical with that obtained by eliminating x, between the equa-
tions of the given smfaces.
If the curve of intersection is composite, the projecting cone is
composite, one component belonging to each component cmve.
A method fo1 determining the projecting cone from any point
Pin space may be deduced by similar reasoning, but the process
is not quite so simple.
EXERCI SES
1 . Show that the intersection of the surfaces
X1X2 - X3X1 + x,
2
- X2X3 = 0, ZtX3
2
- .nX2X4 + Xz(x,
2
- X2Xs) = 0
is composite.
2 . Hepresent each component curve of Ex. 1 completely by two or more
equations.
3. :Fiud the equation of the projecting cone of the curve


+ x
3
2
+ Xt
2
+ 2 XtX = 0, X4
2
+ 2 xzxt - Xt
2
+ 2 xs
2
= 0
from the point (0, 0, 0, 1).
4. Find the equation of the projectiug cone of the curve
Xt
2
+ 4 X3
2
- X4
2
= 0, Xt
2
- 2X:
2
+ Z:r3
2
- 3 '4
2
= 0
frotn the point (0, 0, 0, 1).
5. Find the equation of the projecting cone of the curve
Xt
2
+ 'X2
2
+ X3
2
+ x,z = 0, a1X1
2
+ a2x:
2
+ aa:ra
2
+ a,x,
2
= 0
from the point (0, 0, 0, 1).
ARTS. 175, 176] MONOIDAL ltEPRESENTATION 219
6. Show by means of elimination that, if (0, 0, 0, 1) does not lie on the
cune F .. = 0, F
14
=0, the order of the cone from (0, 0, 0, 1) is 1-'J.L',
pro,idl'd the curve is not couicnl from (0, 0, 0, l ).
7. Find the equation of the projecting cone of the curve
:ra
2
+ 2x2
2
- :ra
2
= 0, :r4
2
- :r2x3 + x,z = 0
from the point ( 1, 1, 1, 1).
176. Monoidal representation. If n. non-composite space curve
om of order m is defined as the comphte or parti:tl intersection
of two surfaces P
14
= 0, F ... = 0, other surfaces on whidt the curve
lies cn.n be obtained from the given ones by algebraic processes.
Among such surfaces we h:we alrcn.(ly discussed the projP.cting
cone ftom a given point P. We shall now show how to obtain
the equation of a surface which contains a .. aud has at p a point
of multiplicity one less than the order of the smface. Such a
surface is called a monoid.
In determi ning the equatiou of a monoid through 0,., we shall
assume that neither the complete intersection of 1<'.. = 0 and
F!J.. = 0 nor any component of it is a conical curve from P. We
sha.ll also assume t.ha.t 1' does not lie on this curve of intersection.
Let P be rhosen as (0, 0, O, 1) and let the equations FJJ. = O,
F ... = 0 be arranged in powers of (Art. 161).
F
14
= llrfti' + 11
1
X
4
14
-
1
+ .. + lt!J. = 0,
F ... = L'rr'C
4
14
' +

+ ... + V
14
= 0,
wherein u., v, are homogeueons functions of x., x
2
, x
3
of degree i.
Let the notation be so chosen that p.' > p.. The eqllation
V,P:,
4
"'-"F
14
- tt
0
F
14
= 0
contains x
4
to at most the power p.' - 1. The surfMc represented
by it passes through the CUl've 0,., since the equation is satistied
by the coordinates of every point whieh satisfy F
14
= 0 and F
14
= 0.
The equation
v,...F .. - u .. F ... = 0
is diYisible hy x
4
If this factor is removed, the resulting equation
is of degre<! at 111ost p.' - 1 in .c
4
, anti determines a surface which
passes through G ...
If either of these equations contains x
4
to the first but to no
220
CURVES AND (CHAP. XIII.
higher power, the surface determined by it is of the type required.
I ( not, the two equations cannot both be i ndepeudent of x
4
nor
cau they coincide, since in that case the curve F,. = 0, F,.. = 0
wonltl be conical from (0, 0, 0, 1).
Hy applying this same process to the two equations just ob-
tained, we may obtain two new ones which contain x
4
to at most
the power p.' - 2.
Continuing in this way with successive partial elimination, we
obtain finally an equation of the form
AI= .l:
4
<f>._
1
(xl> X-.!, x
3
) - <P.(x., x
2
, x
3
) = O,
in which <P.-
1
and <P. are homogeneous functions, not identically
zero, of degree n - 1 aud n, respectively, in Xu x
2
, Xa The surface
Jf = 0 is of order nand has an (n -1)-fold point at (0, 0, 0, 1). It
is consequently a monoid. 'fhe snrface <P. = 0 is a cone; it is
called the superior cone of the monoid. If n > 1, <P.-
1
= 0 is the
equation of anotller cone, called t he inferior cone of the monoid.
Let j,.(x
1
, x
2
, x
3
) = 0 be the equation of the projecting cone from
(0, 0, 0, 1). The equations
j,. (xl, X-.!, Xa) = 0, :t4<f>n-J (xu X2, Xa) - </>.(xu Xa) = 0
are said to constitute a monoidal representation of the curve a,..
The ad vantage of this representation is that the residual inter-
section, if any, of the two surfaces Jf = 0, f ... = 0 consists of lines
common to the t hree cones
],.. = 0, <Pn-1 = 0, </>,. = 0.
For, let P be a point common lo .f,. = 0, J[ = 0, but not lying on
0,., nor at the vertex (0, 0, 0, 1). The generator of f .. = 0 passing
through P intersects a,. in some point P'. Since this generator
has P, P' and n- 1 points at (0, 0, 0, 1) in common with J. lf = 0,
it lies entirely on the monoid (Art. 163). For every point of this
line, that is, independently of the value of x
41
the equation
x
4
<f>._
1
- <P. = 0 must be satisfied; hence the given generator lies
on the inferior cone and on the superior cone.
It follows at once from the aho\'e discussion that if any genera-
tor of .f.,.= 0 intersects 0,.. in two points P, Q, it lies entirely on
the tnon<>id :wd forms a part of thP. resiclual intersection. Such a
line is call('() a uoublc gE'nerator of the projecting cone, since, in
ARTS. 17G, 177] NUMBER OF INTERSECTIONS 221
tracing the cur\' e, the generator takes the position determined by
p 011 a .. and also the position, coincident with the first, determined
by Q. E1ery such line counts for two intersections of 1l f = 0
and/ .. = 0.
Each of these bisecants of the curve is said to determine an
apparent double point of a,. from (0, 0, 0, 1); the Curve appeal'S
from (0, 0, 0, 1) to have a double point on each of these li nes.
It can be proved* and will here be assumed that the nH;nber
of apparent double points of a given curve is the same for every
point not lying on it, except the vertices of the cones, if any, on
which a .. is a conica.l Clli'Ve. This number will be denoted by h.
\\'e shall uow show that if a point p which does not lie on a .. ,
nor on any line that intersects a .. in more than two points, nor a.t
the l"ertex of a cone (if any) of bisecants to a .. , is chosen for the
vertex, then the order of the monoid from P is at least half the
order of a,..
The complete intersection of the projecting cone f.,.= 0 and the
monoid x
4
<fl._
1
- <fl.= 0 is a curve of order m.n. The curve a.., is
one component of order m, and the h bisecants of a,. through
(0, 0, 0, 1) together form a component of order Zh. If the num-
ber of residual intersecting lines is denoted by k, then
?nil - Ill - It = k, k 0 0
The h bisecants of a .. anrl the k resiclual lines are
the intersection of </J, _
1
= 0, . = 0. I [Pnce
from which

(m- n)(n-1)

-1),
and
which proves the proposition.
com pouents of
1 7 7. Number of intersections of algebraic curves and surfaces.
Auy s1o:{we of r)lder J1. which cloes not contain a giten
non-composite CIUl!l' of or1le,- 111 iu/1'/".sPct.s it in ?ltJ1.1lOilds.
* ;:\ooether: Znr r;runolle;.!nn):: <l<r Th .. Mil' il er al;.!ehraisdten r.aumkuncn, Ab
hanrllnu;:en ller k. prcussisthcn Akademie tier \\"iss,.nschaftcn fur llil!2.
222 CURVES SURFACES (CHAP. XIII.
Let 0"' be the gi ven curve and F,. = 0 be the equation of the
gi ven surface. Choose (0, 0, 0, 1) not on F,. = 0, and let the
monoidal t'I}Uatious of 0,. be fm = 0, X
4
cf>. _
1
- cf>. = 0. The com-
plete inte rscct. ion of /,. = 0, .-c,cf>.-
1
- cf>. = 0 ronsists of 0,. and of
?n(n -1) lines t hrough (0, 0, 0, 1). As F,. = 0 does not pass
t hrough (0, 0, 0, 1), it cannot contain any of t hese lines. Hence
F,. = 0, .f,. = 0, 1ll. = 0 have no common component. They COII-
sequeutly inte rsect in mnp. points. Of mp.(n -1) points are
where t be residual lines intersect F,.= 0. The remaining mp. points
li e on 0,.. If 0,. has mp. + t points on F,. = 0, it lies on t.he sur-
face, for t he t hree snrfaces / ,. = 0, Jf., = 0, F',. = 0 have now
nw p. + I points in common, aud t he refore all contain a common
curve. Since the lines tlo not li e on F,. = 0, and .f., = 0, Jf . = 0
have no other component cmve except 0.,, it follows that 0,. must
li e on F',. = 0.
EXERCISES
Rhow that a plane or any proper quarlric is a monoid.
2. Wr1te the equation of n. monoitl of order three.
3. Show tlHtt the only cun e of order one is a line.
4 . Show that the only irreducible curve of order two is a conic.
5. Show that a composite curve of order t iVO exists which does not lie in
a plane. !low many apparent doubl e points has t his cune ?
6. Show that a bundle of quadrics pass through a proper space cubic curve.
7. Write a rnonoidal representation of a space cubic curve.
8. Show 1hat every irreducible curve of order four lies on a quadric
surface.
9 . Discu"s the of F.xs . G and 8 for the case of composiLe cubics
and COIII pOSitC
178. Parametric equations of rational curves. Since a s pace
cune is delined as the compl ete or part ial intersection of t wo
surfaces, t he roordinatcs of its points a re functions of a single
varia ble. The expressions for t he coonl iuates of a point as func-
t ions of a singl e variable may uot be rational. A curve which
}' Ossesscs t he property t hat al l its eu(irtlinates !'an be expressed
ali rational fuuciiuus of a single v;uialJle is call rll a rational curve.
By tll.'flni tion t!t e Cl}nat ions of snch a r mve cau be wri tten para-
lll etri call y in the form
,l' , = f.(t ) = ct
10
t"' + a,,tm-l + ... + u, ,., i = 1, 2, 3, 4.
AnTs. Iii, 1 i$J
EQUATIONS 223
Since the variables x, arc homogeneous, it is no restriction to
suppose that the polynomials /;(t) have no common factor. To
every value oft corresponds a unique point (x) on the curve, but
it may happen that more than one Yalne oft will define the same
point (x) on the curve. If, for example, the J.(t) can
be expressed in the form
f. (t)=:F;(</>(t), Y, (t)),
in which F', are homogeneous rational functions, of the same order,
of the two volynomials <j>(t), Y,(t), then ;;(t) will tleHnc the same
point for every value oft that satisfies the equation
<j>(t) = ;;y,(t),
where s is given. In this case the cootdinates of the points on
the curve are rational functions of s.
Comen;ely, it will now be slwwn that if to each point (x) of the curve
correspond n Yalues of t (n I) , then t may be replaced by a new variable,
in terms of which the correspondence between it and the point (x) on the
curve is one to one.
Let t
1
, t
2
, , t. ;til correspond to the same point (x). The expressious
i, k = 1' 2, 3, 4
vanish for t = t,, t2, ... , t., hence they have a common factor of order n,
whose coefficients contain t.,
<l>o(ll)t + <l>(t,)t-l + + <t>.(IJ).
If 11 is replaced by t2, the expression must have the s:tme factor, hence the
function
<Po(h)t" + 1(t2)l"-
1
+ + <l>n(t2)
bas the same roots. Similarly for t
3
, , t.. It follows that the ratios of the
codiicients
<f>o: ,: ... <l>n
have the same values for l!, t2, . , t.. These ratios cannot be constant for
every point (x) on the curve, since in that case t., ... , t. would be independent
of (x), contrary to hypothesis. If we now put
</>;(1) = s,
'(>k(t)
and eliminate t betweE'n this equation and x; = f
1
(t), the resulting equations
may be written in the form
iu which
x, = b,
0
sP + bils,.-J + ... + b,,.,
np=m.
224 CURVES AND SURFACES [CHAP. XIII.
' Vhen the correspondence between (x) and t is one to one, the
ordct of the curve x, ism. For, to each point of intersection
of the curve with an arbitrary pl ane = 0 corresponds a root
of the equation = 0, and conversely.
179. Tangent lines and developable surface of a curve. Let 0 be
the intersection of two algebraic surfaces ];' = 0, F' = 0 and let P
be an arbitrary point ou 0. 1'he line t of intersection of the
tangent planes to F = 0 and F' = 0 at P has two points in common
with each of the s urfaces coincident at P (Art. 165), and hence
with 0. The line is called the tangent line to the cnr ve 0 at the
point P. The locus of the t:1.ngent lines to 0 is a ruled surface.
This snrface is called the developable surface of 0. Its equation
may be found by eliminating the coordinates y
2
, y
3
, y
4
of P
between the ecpmtions of 0 and of the tangent planes, thns:
DF('I) "" DF'f1,\
F(y)=O, F'(y)=O, "'-r.t,
av, ay,
The intersection of the surfaces
Xt
2
+ :t3
2
+ ::r.
2
= 0, a1.r1
2
+ a2:t2
2
+ aa:ta
2
+ a.x.z = 0
is a quartic cune. Tho equation of the tlevelopable surface of this quartic
is obtained by eliminating !/1 y
2
, y
3
, y., the equations
Yt
2
+ !12
2
+ y3
2
+ !1
2
= 0, a1Y1
2
+ azyz
2
+ a3y3
2
+ a. y.
2
= 0,
X1Y1 + + :ray3 + XY = 0,
a1:r1Y1 + a2X2Y2 + a3.c3y3 + aX!J = 0.
If we write a,k for a; - ak, the result may be written in the form
4 alz<ttaa.zata( ftt:JX1
2
+
2
) ( a2t:t1 a2a:ta
2
+

- [ a23nu'ly 1
2
x
4
2
+

+ + azzaa 2-.c3
2
x.
2
+ aat azt
2
:czZx,
2
+ aza( +

=0.
The number of tangents to the rnrve 0,. which meet an
arbitrary line is calle<l the rank of the curve. From this defini-
tion it follows that the rank is equal to the ortler of the de,elop-
able surface. It is the same number for every line not on the
surface (At-t. 163).
180. Osculating planes. Equation of a curve in plane coordi-
nates. Every pl:1.ue through the tangent line to 0 n.t P contains
the line and has therefore two points iu common with 0 at P.
ARTS. 170. 1801 OSCULATING PLANES 225
Such a. planu is called a tangent plane. Among the tangent planes
there is one having three intersections with C n.t P. This plane
is called tlt e osculating plane to Cat /'. The numbet of osculating
pln,nes to C which pass through n.n arbitrary point in space is
called the class of C. This nnmbcr is the same for every point in
space.* If C is the intersection of P= 0 and F' = 0, we can
obtain two equations which must be satisfied by the roordinatrs
of the osculating planes of C by eliminating two of the ,n,ria.hles,
as X;t, .1
4
, between the el] nations F = 0, F' = 0, antl the eq nation of
the plane ::11
1
:r, = 0, then imposing the condition that. the resulting
homogeneous equation in the other two vatiables has a. triple root.
The two surfaces x1
2
+ = 0, x
2
2 + 2 x
1
x
3
= 0 intersect
in the line :r1 = 0, x
2
= 0 and a space cubic curve. If between the first equa-
tion and = 0 we eliminate x
4
, we find
l14X1
2
- 2 lltXt.t2- 2 1t2X2
2
- 2 llaXzXa = 0.
Now if we eliminate x
3
between this result and the second given equation,
we obtain
li4X1a- 2

+ = 0.
Finally, if this cubic has three e1ualroots, its first member must be a cube.
lienee
A system of two or more ef)nations in plane cootdinates (Art. 173)
which are satisfied by the coordinates of the osculating planes of
C. and by no others, is said to 1lefine the curve 0 in plane coot-
dinatcs. To a curve () dcfiuccl in this way may be a.pplied a. dis-
cussion dual to that given in Arts. li-l-17!).
EXERCISES
1 . Find a system of parametric e'luatious of the rational curve
XtXz- X3X4 = 0, = x
1
2
- x
2
2.
2. Write the equation of the developable surface of the cubic curve lying
on the surfaces
Xt
2
+2.'2XI=0, Xg
2
+27tX3=0.
3. Find two equations satisfied by the coordinates of the osculating planes
of the curve
See reference in Art. 176.
226 CURVES AXD SURI<'ACES [CoAP. XIII.
4 . Define the dual of the projecting cone of a curve and show how its
equation may be obtained.
5. Derive thCl dun! of a monoidaJ representat.ion of a curve.
6. Define the dual of an app:uent double point.
7. What is the .Iual of the rank of a space curve?
181. Singular points, lines, and planes. A point P on a curve
is called an a.ctual double point if two of the points of intersection
of C with any plane through P coincide at P. If the two tangent
lines to a at p :ue distinct, the point is called a node. If the two
tangents at P coincide, the point is called a cusp or stationary
point. Curves may have higher point singularities, for example,
a curve may pass through the same point three or more times, etc.,
but such singularities will not be considered here.
A plane is said to be a double osculating plane if it is the oscu-
lating plane at two points on the cnrve. A plane having four
points of intersection with the curve coincident at P is called a
stationary plane.
A liue is called a double tangent if it touches the curve in two
distinct points. If a tangent line has three coincident points in
common with the Clll'\'e, it is called a stationary or an inflexional
tangent. The point of contact is called a linear inflexion.
182. The Cayley-Salmon formulas. We shall now collect, for
the purpose of point.ing out certain relations existing among them,
the following numbers associated with a given space curve. We
shall assume that these numbcts are fixed when the cmve is given,
and are independent of the arbitrarily chosen plane, line, or point
t hat may be used to determine them.
Given a space curve a. J,et
nt =its order (Art. 1..JO).
n =its chss (Art. 180).
r =its rank (Art. 17fl) .
ll=the number of its nodes (Art.181).
h =the nnmber of its apparent do11ble points (Art. 176).
g =the number of lines of intersection of two of its osculating
planes which lie in a given plane (dnal of h).
G = nnmber of donble oscillating planes (Art. 181).
a= the number of its stationa1y planes (Art. 181).
ARTS 181, 182] 'I'IIE 227
f3 =the number of its stationary points (Art. 181).
v =the number of its linear inflexions (Art. 181).
w =the number of its actual double tangents (Art. 181).
x = the number of points lying in a given plane, through
which pass two distinct tangents to a.
y = the number of planes passing through a given point, which
contain two distinct taugents to a.
These nmnbers are connected by certain equations called the
Cayley-Salmon formulas; they are derived from the analogous
equations, known as Plucker's formulas, connecting the character-
istic numbers of plane curves. Let p. =order, v = class, S = num-
ber of double points, -r = number of double tangents, I(= number
of cusps, '=number of inflexions, of an algebraic plano curve.
Plucker's formulas are*
v = ,.,. (p.- 1) - :33 - 3 I( ;

'= 3 p.(p. - 2) - () s- 8 I( ;
1< = 3 v (v - 2) - G -r - 8 t .
Those in the second line are the duals in the plane of those in the
first line.
Let the given space cnrve a he projected, from an arbitrary
point P not lying on it, upon an n.rbitrary plane not passing
through P. The plane curve then has the following characteristic
numbers:
p. = 1n, v = r, S = h +II, -r =?I+ w, I(= {3, '= n + v.
By substituting in the PlUcker formulas, we obtain
r= ?lt(m-1) -'2(h+ll) -3{3;
n+v = 3 m(m- G(II +h)- 8 f3;

f3 = 3 1'(1'- 2)- G(?J + w)- 8 (n + v).
(15)
By duality in space, that is, by taking the section of the develop-
able surface by au arbitrary plane, we have
?'= n(n -1) -2 (O+g) -3 ((;
?n+v = 3 n(n- Z) - G(G + g)-8 cc;
n = ?'(1' - 1)- 2(:v + w)- 3(m +v);
u = t'(t- !:!) - G(x + w)- 8(m + v).
Salmon: Hightlr Plane (Jurves, 3<.1 editiou (1S7!1). Seep. GG.
(16)
228 CURVES AND SURFACES [CHAP. XHI.
Of these eight equatious, six are indepent.lent. One relation
exists among the first set of four, ant.l one relation among the
second set.
'l'he genus of a curve is the difference between the sum of its
apparent and actual double and stationary points and t he maxi-
mum numbe1 of uouble points which a non-composite plane cmve
of the same oruer may have. If the genus of t he space curve C
is denoted by p, we have
(Jl+h+,B)=(n-1)(n- 2) (G+g+ a)=
2 2
(? - J)(?- (y+ w+n+v)

2
) (x + w+m+v)
2
183. Curves on non-singular quadric surfaces. It has been seen
(Art. 115) that the equation of any non-singular quadric sur face
may be reduced to the form
(17)
and that through each point of the surface passes a generator of
each regulus of the two systems
x
1
- ,\x
4
= 0, x
3
- = O, (18)
X
3
- p.X
1
= O, x
2
- p.X
4
= 0. (18')
The coordinates of the point of intersection of the generator
"-=constant with.the generator p.=constant are (Art. 115)
pxl = "- p:c
2
= p.,. px
3
= AJJ-
1
pX
4
= 1. (19)
Consider the locus of the points whose parameters A, P- satisfy a
gi1eu equation f(>.., p.)=O, algebraic, and of degree k
1
in A and of
uegree 7.:
2
in P- 'l'he curve j(A, p.)= 0 meets nn arbitrary generator
P- =constant in k
1
points, and an arbitrary genetator A= constant
in 7.:
2
points. It will be designated by the symbol (k .. 7.:
2
]. The
order of the curve is k
1
+k
2
, since the plane determined by any two
generators of different reguli meets the curve in k
1
+ k
2
points on
these two lines, and nowhere else.
By replacing>.., p. in j(A, p.) = 0 by thei r values, we see from (17),
(18), (18') that the curve is the intersection of the two surfaces
l(x
1
, x
2
) = O, x
1
x
2
- = o.
::C4
ARTs. 182, 183] CUR\ E;:; ON QUADRIC SURFACES 229
The second is a monoid of order two (Art. 176) and the first is a.
cone with vertex at (0, 0, 1, 0), a(2- 1)-fold point on the monoid.
Thus, these equations consti tnte a particular monoidal tepresenta-
tion of the curve. The equations of the image (Art. 118) of the
given curve on t.hc plane X3 = 0 are

=0, x
3
= 0.
J \.c
4
X
4
The two generators to the quat! ric through the vertex of the cone
f = 0 meet the plane in t he points (1, 0, 0, 0), (0, 1, 0, 0). 'fhe
former is a. 1.:
2
-fold point on the plane rurve, and the latter a
1.:
1
-foltl point.
'1'm:om:.\l I. Ttco cunes of symbols [7.:
1
, k
2
], [7.:'
11
k'
2
] 011 the swne
?lOll-singular qumll"ic intersect in

+ kzk'
1
points.
Let C, 0' be the gi,en curves of symbols [7.:
1
, k
2
] , [k'
1
, 7.:'
2
], re-
spectively, and let the equation of the quadric be reduecd to the
fonn (17) in such a way tba.t the point (0,-0, 1, 0) does not lie ou
either curve, and that the generators x
1
= 0, X
4
= 0 ; x
2
= 0, x
4
= 0
through (0, 0, 1, 0) do not pass through a point of intersection of
the given cunes. l>roject the curves from (0, 0, 1, 0). 1.'heir
images on x
3
= 0 arc of orders k
1
+ k
2
, k'
1
+ k'
2
, respectively; they
intersect in (7
1
+ + points. Of these points, k
1
k'
1
coin-
cide at (0, 1, 0, 0) mHl k
2
k'
2
at ( 1, 0, 0, 0). They are t.he projections
of the points in which the curves meet the generators passi ng
tbwugh (0, 0, l, 0), the vertex of the projecting conP, ancl are
therefore apparent, not actual, intersections of the space curves.
The remaining
(k
1
+

+ k'
2
)- k/.:'
1
-7.:
2
k'
2
=

+
intersections of the plane curves aro projections of t hP. artual in-
tersections of the space cu nes, heuce tho theorem is )'I oved.
TuEORF:)I II. 'l'he 1111mber of ppment clouble }Joints of tl cnrue
of symbol [k
11
k
2
] O?t is
h = t(k
1
2
+ k
2
2
- k
1
- k
2
).
'l'hrough an arbitrary point 0 on the surface pass only two lines
which meet the cur ve in more thau oue voint, namely, the two
generators passing- throngh 0. The generator p. = tonstant
230 CURVES AND SURFACES [CBAP. XII I.
through 0 meets tl1e curve in k
1
points, consequently counts
for (k
1
-1) bisccants through 0. Similarly, the generator
,\ =constant, which passes t hrough (0, 0, 1, 0), meets the curve
in k
2
poi nts and c-ounts for

(k
2
-1) bisecants. The number of
:tppment double points is the sum of these two numbers.
184-. Space cubic curves.
Tm:onE:\1 l. Tltrottyh any .si:v given points in space, 110 four of
1chich lie in a plane, cwt be ]>etsse<l one Ctn<l only one cubic curve.
Let P
1
, , P.: be the given points. The five lines connecting P
1
to each of the rem:tining points uniquely determine a quadric
cone h:wing P
1
as Yertex. Simil:.uly, the lines joining P
2
to each
of the other points t1efiue :1. quadric coue h:wi ug P
2
as vertex.
These two cones intersect iu a composite curve of order fonr, one
component of which is the line J>J>
2
, since it lies on both cones.
'l'he residual is a curve of order t hree. This cur ve cannot be com-
posite, for if it wen"l, n.t least one component would h:we to be a
straight line common to both cones. But t hat would require that
the cones touch .each other along J't}>
2
, which would count for two.
'l'he l'esiclnnl intersection wouhl in t hat case be a co11ic passing
thronglt P
3
, ... , Dnt this is itnpossiulc as it was assumed that
tho poi11 ts P
3
, , P
6
tlo not li e in a planP. X o other cubic curve
can be passecl thl'Ough the gi,,eu points, for every such curve woultl
h:we seven intersertiouf; with the two roues (Lhe vertex counting
for twn). Il t>twe iL would lie un their cun 'e of intersection, which
is siun' lhe complete iuterscction is of order four.
'1'1u:o1:K.\I I r. A :;pHee rubic cut t:e lies 01t all the qrt<tdl'ics nJ u
bundle.
For, let J>l! ... , P
7
be seven given points of the curve. Every
quadric through t hese poiuts has ::! + 1 points in common with
the curve and consequently contaius the curve (Art. 177). Bnt
throngh the given points pass all the quatlrics of a bnntlle (Art.
1:16), which proves t he t heorem.
X ot all the quadrics of t his bnntlle can be singular, for if so, at.
Unless otherwise stat et!, it will be assn met! in the following discussion that
the curve is noucosnposite.
ART!';. IS.3, 18-J]
SPACE CUBIC CURVES 231
least one of them would be com positc (Art. l :H) and still contain the
curve. This is impossible, since the given curve is not a plane curve.
The symbol (Art. 183) of a. spac-e cubic cmve on a non-singular
quadric is [ 2, 1] or [ 1, since such symbols as (0, 3] and OJ
simply define three generators belonging to the same regul ns.
The forms off(>.., p.) corresponding to t hese symbols are
(ao'A
2
+ 2 a,>.. + a
2
)p. + + 3 b,>.. + b
2
) = 0,
(a'op.
2
+ 2 c,',p. + c''
2
)>.. + (b'np.
2
+:! b'
1
p. + = 0.
(20)

Conversely, e\ery irrccluciblo crptation of this fo1m will define
a cubic curve on the CJ.Uadric.
Since these equations have six homogeneous coefficients, five in-
depeudent linear conditions are sufficient to determine a curve of
either system. Hence through any five points on a given non-
singular quadric can be drawn two cubics, one of each symbol.
Some of these cubics may be compositP.
From the formula of Art. 183 it. follows that on a given non-
singular quadric two cubics having the same symbol intersect in
four poi nts, while two cubics having ditierent symbols intersect
in five poiuts.
TuEORDr II L Every space wbic ctm:e mtional.
Let the parametric equations of a. non-siugula.r IIU:t<hic through
the ghen cubic be reduced to the form (19). Tho equations of
the curve in >.., p. arc of the form (20) or (:!0'). In (:?0), lot 'A=t,
solve for p. in terms of t, and substitute the values of >.. and of p.
in terms oft in (19).
The resulting equations reduce to the form
x, = a,of.
3
+ a"t
2
+ + ct,
31
i = 1, 3, 3, 4. (21)
These are the parametric equations of the curve (Art.178). Since
the curve is by hypothesis of order three, to each value of t cor-
responds a definite point on the cnrve, anti conversely.
Since the cubic (21) does not lie in a plane, the determinant
I a,k I =1= 0. The parametric equations, referred to the tetrahedron
defined by
i = 1, 2, 3, 4,
are, after dropping the p1imes,
x
1
= f.3, x
2
= 1
2
, X
3
= t, x, = 1. (22)
232 CUR\'I<:S AXD SURF'ACI':S [CHAP. XIII.
From (32), t he inlcrsections of the curve with the plane :Sn,x,=O
are defined by the roots of the equation
?t,t
3
+ 112t
2
+ 113{ + 114 = o. (23)
The condition that the plane is an osculating plane is that the
roots of (23) are all equal (Art. 180). It follows that the coor-
dinates of the oscnhti ng plane at the point whose l>arameter is t
may be expressed in the form
11
1
=1, ?t
2
=-3t,

u
4
=-t'.
These equations are called the paarnet1ic equations of the cnbic
curve in plane coordinates.
The condition that the osculating plane at the point whose
parameter is t passes througl1 a given point (y) in space is that t
is a root of the equation
(24)
Since this equation is a cubic in t, it follows that the cubic curve
is of class three.
\ Ve shall now the following theorem:
IV. The points nf contw-t of the three oscttlating planes
In It CIIUi CIII'I.IC llii'IJ11'jh ctll ltt'Uitmt'!J poiut f' lie in a ])lillie passiii(J
th rongh 1'.
Let :Sa,.c, = 0 be the plane passing through the points of oscu-
ln.tion of the three planes through any given point
1' = (y). The parameters of the points of osculation of the three
oscnl::tting planes through (.11) are the roots of (24). The roots of
(24) m nst also satisfy the equation
hence
a,f!l + a,.p + a
3
t + n.
4
= 0,

y. -;; ?13 !h - !h
From these conditions it follows that ":!.a;y; = 0, so that (y) lies
in the plane of the points of osculation.
Hy the method of A 1t. 17!) the equation of t he developable sur-
face of the cubic curve is found to be
ART. 18-11 SPACE CUBIC CURVES 233
This is also the condition that equation (2-l) has two equal roots.
From this equation it follows that the rank of the cubic cmve is
four (Art. 179).
It was stateJ without proof in Art. 13:3 that t he basis curve of
a pencil of quadrics of charactetistic [22] is a cubic and a hi-
secant; it was a,lso stated that the basis cutve of a pencil o( char-
acteristic[+] is a cubic curve and a tangent to it. 'Ve shall now
prove these st.atements.
It was shown in Art. 132 t hat the [22] pencil of quadrics is
defined by the two surfaces

+ 2 x
2
x
4
= 0,

+ 2 x
1
:ca = 0.
These quadrics intersect in the line x
1
= 0, .t
2
= 0 and the space
cubic whose parametric equations be fomHl by putting = 1,
= 2 t in the equations of the surfaces, in the form
x
1
= 2 t, X
2
= - t
1
, x
3
=- tJ, x, = 1.
It intetsects the line x
1
= 0, = 0 in the two points (0, 0, 0, 1),
(0, 0, 1, 0).
Similarly, it was seen that a pencil of characteristic [ 4] is
tlefined by the surfaces
+ .r
3
:t'
4
= 0, 2 x
2
x
3
+ .t} = 0.
The basis curve of this pencil consists of Lhe cubic
x, = J' = 2 t
3
, Xa =- t, x. = 2 t
2
and of the line x
2
= O, x
1
= 0 which touches it at (1, 0, 0, 0).
If in the parametric equation (20) of a cubic we replace >.. by
and p. by we determine as the projecting cone f rom
x, .l'.
(0, 0, 1, 0) a. cubic cone with a donble generator. It follows that
the projecting cone of the cubic is intersectetl by a plane in a
nodal or cnspidal plane cubic cmve. We shall now prove the
convetse theorem.
THEOJ:IDT V. Any 110clal ut caspiclal plane curve the
projection of a space cttOic.
Let Lhe plane of the cubic be taken as x
3
= 0, and the node or
cusp at (0, 1, 0, 0). The equation of the curve is of the form
x
2
(Ct&-1:
1
2
+ !3 Ct
1
X
1
X
4
+ a.;cl) + b
0
x,
2
x, + 2 b,.c,x.
2
+ b
2
x} = 0.
234 CURVES AND SURFACES (CHAP. Xlll.
By dividing this equation by x/ aud replacing x,: x
4
by ..\, ~ : x ~
by p., we obta-in equation (20) of a space cubic curve of which the
given curve is the projection.
TJn;oRE)f VI. Any plane nodal cubic cwue has three points of
injlexion lyiny on a line.
If a space cubic is projected from any point (y) upon a plane,
the osculating planes from (y) will be cut by the plane of projec-
tion in the inflexional tangents of the image curve and the points
of osculation will project into the points of inflexion. F1om the
theorem that the points of osculation lie in a plane through (y) it
follows that the points of intiexion of the plane cubic lie on a line.
EXERCISES
1. Show that any space cubic curve and a line which touches it or inter-
sects it twice form the ba.sL<> curve of a pencil of quac.lrics.
2. Show thaL a composite cubic curve exists, through which only one
qun.dric snrface can pass.
3. Prove that the oscnlating planes to a cubic cu1ve at its three points of
intersection with a given plane (tt) intersect at a point in (u).
4 . Show that if a cnbic curve has an actual double point or a trisecant it
must lie in a plane.
5. Obtain all the Cayley-Salmon numbers for the proper space cubics.
6. Where mnst the vertex of the projecting cone be taken, in order that
the plane projection of a proper space cubic shall have a cusp?
7. Show that the projection of a space cubic upon a plane frou1 a point
on the curve is a conic.
8. Show that the cubic curve through the six basis points of a web of
quadrics determined by sis basis points lies entirely on the Weddle surface
(Art. 1-16).
9. Show that a cubic through any six of eight associated points (An.
13C) will have the line joining the other two for bisec:tnt (or tangent).
185. Metric classification of space cubic curves. The space cubic
curves are metrically classified according to the form of their
intersection with the plane at infinity. If the three intersections
:tre 1eal and distinct, the curve is called a cul.Jit'al hyperbola. It
has three rectilinear asymptotes and lies on t hree cylinders all of
which are hyperbolic. Tf the points at infinity are all real and
two are coi ncident, the cmve is called a cubical hyperbolic
A ltTS. 185, 186) SPACE QUARTIC CURVES 235
parabola. It has one asymptote, and lies on one parabolic cylin-
der and on one hyperbolic cylinder. If all three of the points of
intersection are coincident, the plane at influity is an
plane. The curve is called a cubical parabola. It has no recti-
linear asymptote and lies on a parabolic cylinder. Finally, two
of the points of intersection may be imaginary. The curve is
now called a cubical ellipse. It has one rectilinear asymptote
and lies on one elliptic cylinder. An interesting particular rase
of the cubical ellipse is the curve cal led the horopter curve on
account of its part in the theory of physiological optics. If one
looks with both eyes at a point Pin space, the eyes are turned so
that the two images fall on corresponding points of the retinae.
The locus of the points in space whose images fall on correspond-
ing points is a horopter curve through the point P.
186. Classificat ion of space quartic curves.
I. Every space qumtic curve lies on at leust one quacl-
ric surface.
For, through any nine points on the curve a. qHa.dric surface
can be passed. This surface must contain the curve, since it has
2 X -1 + 1 points in common with it {Att. 177).
If a quartic curve lies on two different quadrics A= 0, B= 0,
it is called a quartic of the first kind. A quartic of the first kind
is the basis cmve of a pencil A- >..B = 0 of quadrics. Not all
the quadrics of this pencil arc singular, since in every singular
pencil are some composite quadrics. Composite quadrics are im-
possible in this case, since the curve does not lie in a plane. The
symbol of the curve on any non-singular quadric on which it
lies is [2, 2), since each gcneraLor of one quadric will intersect
the other quadric defining the curve in two points.
A quartic having the symbol [1, 3] canuot lie on two different
quadrics, nor can it lie on a quadric cone, since every generator
would have to cut the cnrve in the same number of points. The
[1, 3] cune is called a quartic of the second kind.
It follows from Arts. 132 and 184 that except in the cases of
the characteristics [1111), [112), [13) , the basis curve of a pencil
See footuote of Art. 184.
236
CURVES AND SURFACES [CRAP. XIII.
of quadrics is composite. It will now be shown that in these
three cases the basis cnrve is not composite, that in the case
[1111] the basis curve has no double point, that in the case [112]
it has a noue, and that in the case (13] it !"las a cusp or stationary
point (Art. 181). That t he basis curve is not composite may be
seen as follows: If it were, oue component would have to be a
line or a conic. It cannot be a line, for the line would have to
lie on every qnadric of t he pencil, hence pa.ss through the vertex
of every cone contained in the pencil. From the equations of the
pencils having these ch:wacteristics (Art. it is seen that in
each case t here is at least one cone whose vertex does not lie on
t he basis curve. J,[oreovcr, ouc component cannot be a conic, for
the quadric of thu pencil determined by an arbitra1-y point f> iu
t he plane of the conic would contain the plane of the conic, and
hence be composite; but pencils having t hese characteristics have
no composite qtta1lrics. It will now be shown that the basis CLnve
of the penci I [11 J 1 J has no aetna! nmle or cusp. It will be called
the non-singular qnattic curve of the first kinrl. Suppose the
basis curve had a node a.t 0. The projecting cone to the curve
from 0 is of order two. 'l'he quadric of the peucil through an
arbitraty point P on the projecting cone contains the line OP,
since it has tluee points in common with it. This quachic and the
cone must coincide, since they have a. quartic curve and a straight
line in comllion. Hence the cone would belong to the pencil, but
this is i111possiblc, since no cone of the pencil [1111] has its ver-
tex on the b:tsis curve.
F1om the equa.tion of the pencil of characteristic [112] it
follows that the vertex (0, 0, 0, 1) of the cone
( .\1 -

+ ( .\2 - .\a)x./ + X
3
2
= 0
of the pencil lies on t he ha.sis curve. This point is an actual
double point 011 the curve, s ince every pla.ne through it bas two
points of intersection with the Cttrve coincident at that point.
A II the quad ries of tlte p<'ncil touch the 1 olrt ne x
3
= 0 at (0, 0, 0, 1);
every pl ane t htongh cithct of t he distinct lines (.\
1
- .\
3
).v
1
1
+
( .\
2
- .\
3
)x.
2
2
= 0, in which .c
3
= 0 iutorslCts t he cone has three in-
tersec:tions with the cutvc coiucideut a.t (0, 0, O, 1). These two
I i nes arc ta.ngcnts at the uouc.
AnT. 186) SPACE QUARTIC CURVES 237
Finally, the vertex of the cone
(A
1
- Az)Xt
2
+ !3 X3X4 = 0
of the [13] pencil is a double point on the basis curve. The tan-
gent lines x
1
= 0, x
3
= 0 coincide. The double point is a cnsp.
The parametric ecruation of a. crua.rtic of symbol [2, 2] has the
form
(au>-
2
+ Z a
1
A + + !3(u
0
,\
2
+2 b
1
,\+b
2
)p.+ co>-
2
+2 c
1
,\+ez=0. (25)
The quartic defined by (:!5) is the intersection of the quadic
X
1
X
2
-x
3
x
4
=0 (Art. 18:3) and the quadric
Cl.,j.1:
3
2
+ 2

+ a
2
x
2
2
+ 2 bo-1:
1
.v
3
+ -1 b
1
:J;Xz + 3 b
2
XzX
4
+ c
11
1:
1
2
+ 2 c
1
x
1
X
4
+ C?.''l:
4
2
= 0. (25')
If the quartic of intersection has a double point or cusp, we
may take the double point as (0, 0, 0, 1), and a cone with vertex
at that point for one of the crundrics passing through it. The
parametric equation (::!:'3) now has the form
(3 a
1
A. + a
2
)p.
2
+ 3(b
0
il.
2
+ 2 b
1
A)p. + c
0
,\
2
= 0. (2G)
If in (2G) we put A. = p), solve for t. and pnt the values of p. and
>. = pJ in ecruations (19), we obtain a set of parametric equations
of the singular crnnrtic curve of the first kind, of the form
x, = a,
1
l
1
+ a,
1
l
3
+ a,
2
t
2
+ a,
3
t + (Lw i = 1, 3, 3, 4; (27)
hence the nodal and cuspidal quartics ate rational.
A quartic of the second kind can be expressed parametrically
in terms of the parameter which appears to the thin! degree in its
parametric equation, hence the quartics of the second kind are also
rational. Rational curves will be discussed later (Art. 188).
II. Through a 'JIWrlic curve of the Sl'f'OIHl kinrl cmrl
(tllY two <!fits triseccmts can be passecl ct non-composite cubic swface.
For, through nineteen points in space a cubic smface can be
passed (Art. 161). Choose thirteen on the quartic cnrve, one on
the trisecant y, one on the trisecant r/, not on the curve, and four
others in spa.ce, not in a plane no 011 the quadric on which the
quartic lies. The quartic curve and the lines (I andy' must lie on
the non-composite cubic surface determined by these ninC'teeu
238 CURVES Al\D [CUAP. XIII.
points as well as ou the quadric containing the tegulus of tri-
secants, hence together they form the complete intersection of the
en hie and the quadric.
187. Non-singular quartic curves of the first kind. Two quartic
curves of the first kind lying on the same quaclric intersect in
eight points (Art. 18:>); these points are eight associated points
defining a bundle (Art. laG), since they lie on three distinct
quatlrics not having a curve in common.
Tho number of apparent double points of a non-singular quartic
0
4
of the first kind is two. For each bisecant of 0.
1
through au
arbitrary point Pis a generator of the quadric of the pencil hav-
ing 0
4
fm basis curve which passes through P. Conversely, each
generator of every quadric through 0
4
is a bisecant.
Of the Cayley-Salmon numbers we now have ?n=4, h=2,
/3 = 0, II= 0. It also follows from the defini tiou that G = v
= w = 0, hence from the formulas of Art. 182 we have
1n = 4, n = 12, 1 = 8, II= 0, h = 2, G = 0, fl = 38, ct = 16, {3 = 0,
v = 0, w = 0, X = lG, ?J = 8, }J = 1.
TnEORE:\1 I. TMo11gh auy bi.secant of a non-sin{tulctr space
quartic Ct11ve of the first ki11d can be clmwn fonr ta11gent plcwe.s
tv the cu1ve, be.sicles hewing their point of contact on the giveu
bisecant.
Let the given bisecant be taken as x
1
= 0, x
4
= 0 and the quadric
of the pencil containing it as

.t
3
x
4
= 0. I,et another quaclric
of the pencil be determined by (2!>'). Any plane of the pencil
!l;
1
= mx
4
intersects 0
4
in two points on

= 0, x
4
= 0 and in two
other points determined by the roots of the quadratic equation in
':l;-l(a
0
?iL
2
+ 2 a
1
1n + ct
2
) + 2

(b
0
1n
2
+ 2b
1
1n + b
2
)
+ xl(Com
2
+ 2 c,m + c2)= 0.
The planes determined by values of m which make the roots of
this eqnatiou equal are tangent planes. The condition on ?IL is
4(bo?n
2
+ 21J,nt+ IJ2)
2
-4(eto?n
2
+2 a,m+etz)( eom
2
+2 c,m+ez) =0. (28)
::iince t!tis equation is of the fourth degree, the theorem is
established.
ART. 187] QUAR'I'ICS 011' THE FIRST KIND 239
.ruEORElll II. .dn mbitrary tangent to a non-singulm quctrtic of
the first kind four other tangents at J>oints not on the C!U'Ve
This is a particular case of Theorem I, since a tangent is a
bisecant whose points of intersection with the curve coincide.
TuEOREli III. The cross mtiu of the fow tangent planes th?ough
<tny Oiseccmt is the smne nwnber for every biseccmt of the curve.
Two cases are to be considered, according as the two given
hisecauts intersect on or not. Let g, g' be two bisecants
through a point 1' on but not on the same quadric of
the pencil. Let the equation of the quadri c of the pencil through
C, which cont.'tins g be reduced to the fonn X
1
X
2
- x
3
x
4
= 0 in such
a way that the equations of [J are x
1
= 0, x
4
= 0 and the points of
intersection of g' with C, are (0, 0, 1, 0) (0, 0, 0, 1). In (25') we
now ha,e u
0
= 0, c
2
= 0, and also in (28 ). The points of inter-
section not on r( of a plaue x, = and C
4
are determined by the
roots of the equation
2(c
1
n
2
+ b
2
n)xl + (c.,n
2
+ .j. 0
1
11 + et
2
) X
2
X; + 2(b
0
n + Ct
1
) xl = 0.
The parameters nu n
2
, n
3
, n
4
of the four tangent planes are roots
of the equation
(c
0
n
2
+ 4 b
1
n + <t.z)
2
-16(b
0
n + ct
1
)(c
1
n
2
+ b
2
n)= 0.
The cross ratio of the fonr roots of this equation is equal to the
cross ratio of the roots of (28) (when a
0
= = 0), since the two
equations can be shown to have the same invariants.*
To prove the theorem when g, r( intersect at P on C
4
and
lie on the same quadric t hrough C
4
, conside1 any third bisecant
y" of C
4
through P. The cross ratios on g and on g' are each
equal to that on g".
This completes the proof of the first case.
' To prove the theorem when the two bisecants do not intersect
on c., consider a third bisecant connecting a point of intersecti on
on the first with a point of intersection on the second. The
cross ratio on each of the given lines is equal to that on the
transversal.
This cross rat io is cailed the modulus of the quartic curve.
Burnside ami Panton: Theory of Eqnat ions, :ld edition, p. HS, Ex. 16. It
will be found tbat I and J have t.he same values for each equation.
240 CURVES AND SU ltPACES [CUAP. XIII.
The projecting cone of 0
4
from a point on it is a cubic cone.
'fhe section of this cone made by a plane not passing through the
vertex is a enbic curve. Coll\ersely, any plane cubic curve is t he
projection of a space quartic cmve of the first kind. Consitler
the cu hie curYc in the plane x
3
= 0. Tt is no restriction to choose
the triangle of reference with the two vertices (1, 0, 0, 0), (0, 1,
0, 0) on the curve. Tlte most general cubic equation in x
11
Xz. x
4
,
but lacking the terms x
1
3
, x}, may be written in the form
2 a
1
xl:c
1
+

+ 2 bo-'l:
1
2
X
2
+ 4

+ 2 bzX:-1:/ + CoX
1
2
x
4
+ 2 c
1
x
1
x/ + CzX
4
3
= 0.
But this is exactly the result of projecting (Art. 175) from the
point (0, 0, 1, 0) the curve (25) for the case a
0
= 0, that is, when
the quartic curve passes through (0, 0, 1, 0).
From Theorem HI it now follows that the cross ratio of the
four tangents to any non-singular cubic curve from a point on it,
not counting the tangent at the point, is constant.
lt was seen that every noH-singnlar quartic lies on four quadric
cones whose vertices (Art. 133) are the Yertices of the tetrahedron
self.polar as to the pencil of qnadrir surfaces on which the curve
lies (Art. 112). Let t, t' be two distinct tangents of 0
4
which
intersect in a point P. The plane 1r determined by t, t' touches 0
4
in the points of contact T, T' of t, t', respccti vely. The following
properties will now be proved :
(1) The line l = TT is a generator of a qnadric cone on which
0
4
lies.
(2) Tbe plane 1r is a tangent plane to this cone along l.
(3) The point P lies in the face of the self-polar tetrahedron
opposite to the vertex through whi('h l passes.
The plane 1r cuts the pencil of qnarhic s urfaces on which 0
4
lies
in a pencil of conics touching each other at '1' and T. One conic
of this pencil consists of the line l counted twice, hence l is a.
gcnetator of a cone of the pencil an!l 1r is its tangent plane. :More-
over, l is the polat line of Pas to the pencil of conics, hence the
vertex of the cone and the point Pare conjugate ]10ints. Thus
Plies in that face of the self.polar tetrahedron which is opposite
the vertex of the cone.
lf 1r avproaehcs a stationary plane (Art 181), thl"n 'l', T, P
ART. 1, 71 QlJAR'l' ICS OF THE FIRST KI ND 241
approach coincidence, and the tangents I, t' both approach l. This
occms at every point in which intersects t he faces of t he self
polar tetrahedron. We have thus the following theorem:
THEonE)t V. The 1,oiut.s of contnct of the sixteen
J1la11es (<' = 1G) of a non-:;;ingtdctr qnartic curve of the jitst kind lie
in the faces of the common self-polar tetrahedron. The planes ue-
lOII[!illg to the points in each frr.ce through tlte opposite vertex.
Referred to the self-polar tetrahedron, t he equations of the
quartic are (Art. 133)
:rt2 + x.,_2 + + x42 = 0, ar:tt2 + az:l'z2 + a3x32 + a.xl = 0.
The equation of the developable was derived in Art. 179.
The section of the developable smface by the plane x
4
= 0 is
the quartic curve (a,k = n,- a.),

+ r!J4rtt2
2
xttz
2
+ n23(atzCtat +

= 0
counted twice. It is a double curve 0 11 t he developable. It is
the locus in the plane

= 0 of the points of intersection of ta,n-
gents to C,. A simil:u locus lies in each of the other faces of the
self-polar tetrahedton. Since the Cayley-Salmon number xis 16,
Lhe enti re locus of intersecting tangents to 0
4
is these four curves.
Since tho points of intersection of 0
4
with the faces of the
self-polar tetrahedron are tho points of contact of the sixteen
stationary planes. the coordinates of these points are
(..Jci;, vaau 0), va.u 0, -Ya;:),
(va3 0, Vats), (0, -va.;;, via;).
EXERCISES
1 . Find the locus of a. point P such that the two bisccants to C
4
from 1'
coincide.
2 . ll ow many generators of each quadric through C
4
are t.'\ngent to the
cune?
3 . By lhe method of Art. 180 find the equations of the stationary planes.
4. Show that any plane containing three points of contact of stationary
planes will pass through a fourth. How many distinct planes of this kind
arc there?
5. Find the locus of a point P such that the plane projection of C
4
from
1' "'ill be a curve with one double point and one cusp; t\\'O cusps.
242 CUR\ . .ES AND SURFACEH [CII AP. XIII.
I 88. Rational quartics. The parametric equations of any
ra,tional quartic may be written in the form
X; = CL;r/} + 1 Ctj
1
+ 6 Ct;zl
2
+ 4 (l ;al + Cl w i = 1, 2, 3, 4.
The parameters of t he points of intersection of the curve with
any plane :Su,x, = 0 are t he roots of the equation
Let t., t
2
, /
3
, 1
4
be t he roots of this equation. From the formulas
expressing t he coefficients in terms of t he roots we have at once
(l
1
+ 1
2
+ 1
3
+ 1
4
) :Sa,o'' + 4 :Scr,171; = 0,
(t
1
1
2
+ 1
1
f
3
+ l
1
t
4
+ 1
2
1
3
+ l.j
4
+ 1
3
1
4
) :Sct,
0
1l, - 6 :Src;z"; = 0,
(l
1
l ,j.
3
+ l
1
1
2
1
4
+ t
1
l
3
1
4
+

::::n,o''
1
+ 4 :SCt,
3
!i; = 0,
(29)
If we el iminate 11
1
: 11
2
: u
3
: n
4
from these four C!Jll:l.t ions, we obtain
as tbe conrli tion that 1
11
, / .
1
are the pammeters of four coplam\r
points, t he equation
.L1.
1
l
1
l
2
1
3
1
4
+ 3 .t1
3
(t
1
l;l
3
+ 1
1
1
2
1
4
+ l
1
laf
4
+

+ 2 .'1
2
(1
1
1
2
+ l
1
1
3
+ l
1
l
4
+ 1
2
1
3
+ ll
4
+ t
3
1.
1
)
+ 3 A
1
(1
1
+ 1
2
+ 1
3
+ 1
4
) + 1:.3 ..'lo = 0,
in which A
0
= I ct
11
a
22
a
33
Ct.
14
l, A1 = I CCo'l2zrt
3
aa441> etc. If 1
1
= 1
2
= t
3
= l
4
in t ho conesponding point will be a point of con-
tact of a stationary plane. H ence there ate four points of con-
tact of stationaty planes. These four poiuts are defined by the
equation
(31)
Ttl .If (t qnmlic {,;II)'/' I' has (L double zwinl, the pmamelei'S
of lite points of contact of the slalirmmy ]>lanes we harmonic.
Let J> be the tlouble point autllet 1
1
, lz be the values of the para-
meter at P. Since 1' is coplanar with any other two points on
the curve, equation (:30) is satisfi ctl intlep<'ml ently of the values
of l
3
and /
4
Thus lu l
2
must satisfy the condit ions
12 Ai
1
l
2
+ ..'1
3
(1
1
+ 1
2
) + 2A
2
= O,
3 A
3
t
1
t
2
+ A
2
(l
1
+ 1
2
) + A
1
= 0,
:3 A
2
t
1
l
2
+ 3 .A
1
(l
1
+ l
2
) + .d
0
= 0.
(32)
ART. 1!)8] RA1'10NAL QUARTICS 243
These <'Cptations are comvatible only when the determinant
vanishes, thus
1::? 3 .112
3 A
3
.11
2
3 At = 0.
2.11
2
3A, 12A
0
But this is the condition that the roots of (31) are harmonic.*
The condition that the double point is a cusp is t, = t
2
In this
case equations (3::?) are replaced by the quadratic equations
6 ..tiP + 3 A
3
t + ..:1
2
= 0, .A
3
t
2
+ 4 + 3 A, = 0,
AP + 3 A,t + 6 = 0.
But these arc the conditions that (31) has a triple root. Hence, on
a cuspid:d quartic, tluee of the points of contact of stationary
planes coincide at the cusp. There is in this case only one proper
stationary plane.
Three points on are collinear if their parameters t
11
t
2
, 1
3
satisfy (30) for all values of The necessary conditions are
1:2 + 3 AJ(tlt2 + 123 + 131,) + 2 A2(tl + 12 + 13) + 3 AI = 0, (33)
3

+ 2 A
2
(l
1
l
2
+ l.f-
3
+ 1
3
t
1
) + 3 A
1
(1
1
+ 1
2
+ 1
3
) + 1:.! .:1
0
= 0.
If the curve has a double jJoint, the parameters t
1
, t
2
of the double
point satisfy these conditions for every value of 1
3
If it does
not ha\e a doubl e point, the equations arc satistied, for any
given value of 1
3
, by the parameters of the other points on the
trisE>cant through t.
If the equations resulting from by pntting 1
1
= 1
2
= t
3
have
a common solution t', the curve has a lint>ar innE>x iou at the point
whose parameter is t'. The conditioa that these equations in t'
have a common solution is exa('tly the (Ontlition thn.t has n
doubl e root. In patticular, if (:H) is :t square, curve has two
distinct linear inAexioas.
EXERCISES
l . Obtain the Cayley-Salmou nuwbors for:
(a) the nodal quartic.
(b) the cuspiual quartic.
(c) the general quartic of the second kind.
(d) the quartic having a linear inlli!xion.
(e) the quartic having two line:tr inflexions.
When tile root$ of a quartic equation che invariant J vanishes.
Seo: Bnruside ancl Paut.ou: Theory of B<JU.ltions. 4tli edit.iou, \' nl. I, p. 100.
244 CURVES AND SURFACES (C'FIAP. XIII.
2. Show that every (1, k] curve on a quadric is rational and can have no
actual double point.
3. Show that e1ery rational quartic is nodal, cuspidal, or a quartic of the
second kinu.
4. Show that if a r:\Lional quartic does not have a cusp or a linear in-
fl exion, its pammetric equations can l>e written in the forw
~ r 1 = (t + 1)\ X:l = (t + a )t, X3 = t\ X4 = 1.
Find the values of a for which the curve is nodal.
5. Prove that if a quartic Lias a single linear inflexion, its equations can
he written in the form
:t't = t\ X2 = t
3
, Za = (t + 1 )\ X4 = 1,
and if it has two distinct linear inflexions, in the form
... , = tl, :r2 =t
3
, :ra = t, :1'4 = 1.
6 . Show that the equations of a cuspidal quartic can be written in the form
X1 = tl, :r2 = t
3
, :ra = t
2
, x4 = 1.
7. Show that the t:\ngents at the poin ts of contact of the stationary planes
of a rational quartic are in hyperbolic position ( Art. 120).
8. Show that through any point P on a rational quartic curve pass three
osculating planes to the curve besides the one at P, and t hat the plane of tile
points of contact passes through P.
9. Determine the number of generators of a quadric surface which are
tangent to a (1, 3) curve lying on it.
10. Determine t he number of generators of a quadric surface which are
tangent to a nodal quartic curve lying on it.
ll. Find the parametric equations in plane coordinates of the curves of
Ex. 5.
CHAPTER XIV
DIFFERENTIAL GEOMETRY
In this chapter we shall consider some of the properties of
curves and surfaces which depend on the form of the locns in the
immediate neighborhood of a point on it. Since the prope rties
to be determined involve distances and angles, we shall use rec-
tangular coordinates.
I. Analytic Curves
189. Length of arc of a space curve. 'l' he locus of a point whose
coordinates are functions, not all constant, of a parameter u
x=.fJ.(u), z=/
3
(u) (1)
is a. space curve. The length of arc of s uch a curve is defined as
the limit (when it exists) of the perim<>ter of an inscribed poly-
gon as the lengths of the sides uniformly approach zero. Curves
for which no such limit exists will be excluded f rom our discus-
sion.
By reasoning similar to that in plane geometry it is seeu that
the length of arc s from the point whose parameter is to the
point whose parameter is u is
"\., - + - + -
-l " /(d:c)z (<ly)z (dz)2
, cln da rln
(2)
'l'his equation defines s as a function of u. If the function so
defined is not a constant, equation (2) also defines u a.s a function
of s. I n this case we may write (1) in the form
x=x(s), y=y(s), z=z(s), (3)
in which s is the parameter.
Unle1.s the contrary is staten, we shnJl suppose that s is the
parameter in each case, and that. x, y, z are analytic functions of
245
DIF'FERE>J'l'IAL OEOME'l'RY [CHAP. XIV.
s in the inte1val under tonsideration. In the neighborhood of
(x(s), y(s), z(s)), to which we shall refer as the point s, we have
In which
_<l.l:
X - - ,
<h
It follows from ectuation that
x'2 + y'2 + z'2 = 1.
By differentiating equation (5) we obtain
x'x" + y'y" + z'z" = 0.
(4)
(6)
1\'e have thus far supposed that the second member of (2) was
uot a constant. If the second member of (2) is a constant, we
have
- + -- + -- - 0.
(
<7.;;)2 (dy)2 (<lz)2 _
du clu cl11
(7)
Curves for which this condition (7) is satisfied are called minimal
lnrves. They will be presently. It will be supposed,
when the contrary is st:ttell, that the curve unclt.> r consider-
is not a minimal curve.
f t HO. The moving trihedral. 'l'he tangent line to the curve at
the point P := (x, y, z) ou it may be detincd as the limiting posi-
tion of a secant as two intersections of the line with the curve
approach 1'.
From (-l) the equations of the tangent at Pare
X-x_ Y-y_Z-z
-x'- ,,, - - -z,-.
(8)
Let A., p., " be the direction cosines of the tangent, the direction
in which s increases being positive. Froll) (8) and (5) we have
A.=x', p. =y', v =z'. (9)
ART. 100)
THE l\10VIKG TRIIIEDRAL 247
The plane through P = (.t, y, z) perpendicular to the tangent line
is called the normal plane. Its c11uation is
x'(X- x) + y' ( Y- y) + z' (Z- z) = 0. (10)
The osculating plane at /' is the limiting position of a plane
through the tangent line at P and a point P' on the curve, asP'
approaches P. \Ye ::hall now determine the equation of the
osculating plane.
The equation of any plane through J> is
A(X- :r.) + B(J'- y) + C(Z - z) =0.
It contains the tangent (8) if
.A.r.' + JJy' + Cz' = O,
and will be satisfied for powers of tls up to the third (Eqs. (4)) if
.Ax'' + B.IJ" + Cz" = 0.
By eliminating A, B, C, we obtaiu, as equation of the osculating
plane at P,
X-:r
Y-?1 Z-z
x'
?!'
z' = 0. (11)
x" !!" z''
The line of intersection of the osculating plane and the normal
plane is called the pri ncipal normal. From (10) allll (11) its /
equations are found to be
X -x Y-11 Z -z
~ = -y.-:- = ----;}'"
(12)
If ~
1
, P.ll v
1
are the direction cosines of t.he principal normal, and
if we put
we have
.! = , 1 x''2 + ?/'2 + z'2,
p
"-1 = px'', p.
1
= py", y, = pz".
(13)
(14)
The plane through P perpendicular to the principal normal is
called the rectifying plane. From (12) its equation is
x"(X- x) + !!''(}"- y)+ z"(Z - z)= 0. (15)
248
[CHAP. XIV.
The intersection of the rectifying plane and the normal plane is
called the binormal. equations (10) and (15) its equations
are
X- x
y' z" - y" z'
r-.1' Z-z
--,---,7----'!.,-,' --: =
z' x" - z" x' x' y" - x" y'
(16)
If >.
2
, Jl-2, v
2
are the direction cosines of the binormal, we have the
relations
A
2
=p(y'z"- y"z'), JL2 = p(z'x"- z"x'), v
2
=p(x'Jt"- x"y'). (17)
The tri rectangular trihedral whose edges extend in the positive
di rections from P a.long tbe tangent, principal normal, and bi-
normal is called the moving tribedml to the curve at P =.(.1:, y, z).
From (9), (14), and ( 17), we have
A p. v
(18)
It follows at once (Arts. 37, aS) that the positive directions of
the coordinate axes can be bronght into coincidence with the
positive directions of the moving trihellral at the point P by
motion a.lone, without teflexion. :i\Ioreover, we h:l.Ve (Art. 37)
A = Jl.lvz- V1IJ.z
A
1
= Jl-211 - v
2
!L
1
Az = V
1
!L - !L
1
v
p. = v
1
Az - A
1
v
27
p., = v2A - A
2
v,
p.
2
= A
1
v - v
1
A
1
v = A,iJ-2 - A2!Lu
v
1
= A2!L - (19)
v
2
= Ap.
1
- A
1
p..
,191. Curvature. The curvature of a space curve is defined,
like that of a plane curve, as the limit, if it exists, of the ratio
of the measure of the angle botweC'u two tangents to the length
of arc of the curve bctweeu their points of contwt, as the points
approach coincidence.
Let 8 be the n.uglc between the ta,ngents to the curve at P and
P'. 'The direction cosines of the tangent at P ate x', y', t (9),
those at P' a.re
+x'tls + ... , y' + ?f'tls + ... , t + z"tls + ....
From Art. fj, we have
sin
2
t:.O =! (?/z''- y"i)
2
+ (tx"- x'z")
2
+ (x' y''- x"y')
2
l (As)
2
+ ... ,
ARTS. 191, 192) TORSION 249
the remaining terms alt containing higher powers of t:.s. From
(5) and (6) the coefficient of reduces to x'
2
+ y"
2
+ z"
1

Since
I
. si n t:.{)
1 tm--=,
t:.{)
we have, on account of {13),
<18 1
ds p
(20)
as the expression for t he curvatute at P. The reciprocal of the
cnrvn.ture is called the radius of cmvature.
If !. = 0 at a point P on the curve, the tangent at P has three
p
points of intersection with the curve coincident at P; hence Pis
/
a linear i,xion.
1!>2, Torsion. The torsion of a space curve is defined as the
limit, if it exists, of the ratio of the angle between two osculating
planes to the length of atc between their points of osculation, as
the points approach coincidence. The reciprocal of the torsion
is called the radius of torsion aml is denoted by <r.
In ortler to find the value of <r, let .!h be the angle between the
osculating planes at the points whose parameters ares and s + t:.s.
By a proee:;s similar to that of .Art. 191 we obtain
sinz = + (vzA.'2- A2/z)
2


+ ,
the remaining terms all containing higher powers of t:.s. By dif-
ferentiating (17) we have
A'z = i Az + p(y'z'"- z'y"'), fl'z = i flz + p(z'x"'- x'z'"),
p p
' - i + (x"y"' y'x"') V2- Vz p -
p
(21)
It follows that
x' y' i
p.zv'z-v,fl'z = p
2
J! x'
y" z"
x'" y"' i"
250
DU'FERENTIAL OEOME'rRY (CHAP. XIV.
1 f we substitute these value$ in the above expression for sin Ar,
pass to the limit, take the square root, and assign opposite signs to
the two members, we obtain the result
"
clr 1
2
xl '!II
zl
-=-=-p x" y"
z"! '
(22)
ds u
:If" y'"
z'"
which is the formula required. Expand the determinant of equa-
tion (22) in terms of the elements of the second row, replace
x", y", z" by thei r values from (1-t), and the cofactors of these
num bcr::; by their valnes from (21), and put .\
1
.\
2
+ p.
1
p.
2
+ v
1
v
2
equal
to zero, since the principal normal and binormal are orthogonal.
lly performing these operations we simplify (22) to the form
1 \ \1 I I
- = "t" 2 + Jl-11'- 2 + VtV 2'
CT
193. The Frenet-Serret formulas. The nine equations
>..1 P.l vi= !2,
p p p
(23)
A
1
t ;} P.
1
t =- (;+':). v
1
1
(24)
/z=/!:.1,
CT
are called the Frenet-Serret formulas.
I Vt
Vz= -,
CT
The fitst three follow at once by r eplacing A., p., v and .\
11
p.
11
v
1
by their valnes from (9) and (14).
To derive the last three, differentiate the identities
A2
2
+ p.2
2
+ vl = 1, ,\,\z + P.P.2 + VVz = 0
with respect to sand substitute for >..
1
, p.', v
1
their values from (24)
which we have just established. The results are
From these equations we obtain, after simplifying by means of (19),
ARTS. 193, 1941 THE OSCULA'rnm SPHERE 251
S being a factor of proportionality. 'l'o determine its value, sub-
stitute these values of >..'
2
, p..'
2
, v'
2
in (23). Since >..
1
2
+ p..
1
2
+ v/ = 1,
we find 8 = The last three equations of (2-1:) are thus established.
u
To find the values >..'
11
differentiate the idantity >..
1
= p..
2
v- v
2
p..
(19) and substitute for p..', v', p..'
2
, v'
2
their values from (2-!). Hy
(19) the result reduces t o the form >..'
1
= + The values
of p..
1
11
v'
1
are found in the same way.
104. The osculating sphere. The sphere which has contact of
the third oruer with a curve at a point P is called the osculating
sphere of the ctme at P. To determine the center and radius of
the osculating sphere at P=.(x, y, z), denote the coordinates of
the center by (xv .lh, z
2
) and the radius by R.
The equation of the sphere is
(X- X2)
2
+( Y- y
2
)
2
+(Z- z
2
)
2
= R
2

/
This equation must be satisfied by the coordinates defined by (4)
to terms in (As)
3
inclusive. From these conclitions we obtain the
following equations
(x- Xz)
2
+(y- y
2
)
2
+(z-zz)
2
= W,
(x- Xz)x' + (y- + (z- z
2
)i = 0,
(x - x,.)x" +(y- ?h)!/' +(z -Zz)i' + 1 = 0,
(x - Xz)x"' +(y- y
2
)y"' +(z- z
2
)i" = 0.
(25)
By solving the last three e.rpmtions for x - x
2
, ?J- y
2
, z- z
2
and
simplifying by means of (21), (22\, and (24) we Jind
.tz =X+ pAt- p'uAz
1
Yz = y + pp.
1
- p'up..z, Zz = z + pv
1
- p'uv
2
. . (26)
If we substitute these values of X:, y
2
z
2
in the first of ertuations
(25) and simplify, we obtain

TnEORE:.r. The condition thrtt 11 curve lies on a sphere is
p + u( <T
1
p' + up")= 0.
If a given curve lies on a S!lhere, the sphere is the osculating
sphere at all points of t he c11ne so that x,., y
2
, z
2
and R are con-
252 DWI<'J<:RI<:NTIAL GEOMET RY [CHAP. XIV.
stants. Conversely, if these quantities are constants, the curve
lies on a sphere.
To determine the con<lition that the coor-dinates of the center
are const:wt, diiTcrentiate e11uations and situplify by means
of (24). Siuce .\
2
, JN!., v
2
are not all zero, the condi tion is
p + u(u'p' +up")= 0.
By di ffercnti:Lting (27) we sec that R is also constant if this equa-
t ion is satisfied. This proves the proposition.
1 !)::>. Minimal curves. We have thus far excluded from discus-
sion t hose curves (Art. 189)
for which
x =.f.(n), y =.t;(n), z =fa(tt),
(28)
Such curves we called minimal curves. A few of their properties
wi ll now be deri vctl.
From (28) we may write
and
in terms of a parameter t.

From t hese equations we deduce
d!f <lz
clu
du.
1 - t
2
- i(l + ( 2) = -.
--z 2
If we denote the value of these fractions by cf>(n), solve for
, 1.1 cil1 tlz d h ( ) t bl
- , -=- , - an mtegrate, assnmmg t at cf> u JS 1n egra e, we
dn d!t
find that theequationsof a mini mal cun'e m:ty be written in the form
z= J tcf>(1t)dn,
(29)
in which t is a constant or a fm1ct ion of u. [f t is constant, the
Joens (29) is a liuc. For, let k be defined by k = J "cf>(!t)du.
.\nT. 103] CURVES 253
ln terms of k we obtain
1 - t
2
i
X = k + X
11
y = z(l + t
2
)k + y
1
, Z = tk + Z
11
wherein xu y., z
1
a re constants of integration. The l ocus of t he
point (x, y, z) is the minimal line through the point (xu Yu z
1
)
x - :r1 Y-?lt z - z
1
1 - t
2
= +_fl = -t-.
2 :t
The equ:ttion of the locus of the minimal lines through n,ny point
(xu Yu z
1
) is found by S<tnaring the terms of these equations and
adding numet-ators n,nd denominatos, rcspecti,ely, to be the cone
(x- x
1
)
2
+ ?I- ?!t)2 +(z- zt)
2
= 0,
having its ,ertex at (xu !Ju z
1
) and passing t he absolute.
This is identical with the f'qnation of t he poi nt sphere (Art. 48).
If t is not constant, but a function of v, we may take t as t he
parameter. u = .p(t), and let cp(u)rlu = cp(.p(t))t/t'(t)clt be re-
placed by F(t)dt. Equn,tions (2fl) have the form
x = 1 - t
2
)F(t)clt, y = + t
2
)F(t)clt, z = J tF(t)tlt. (30)
Let f(t) be defined by = F(t). By integrating equations (30)
by parts we ha. ve
X= (1- t
2
) d:f(t) + t _ ((t) +X
2 <lt
2
ell
1
'
(31)
z =t cl2.f{t2 - c(((t) + z
clt
2
dt
11
x
11
y
11
z
1
being constants. The equations of any l!ODrectili near
minimal cmve may ue expressed in this form.
EXERCISES
1 . The cuno
:t=acos, y=asin, z=a<t>
is called a circular helix. Find the parametric equations of the curve in
terms of the length of nrc.
254 DIFFERENTIAL CEOUETRY (CI:!Al' . XIV.
2. At an arbitrary point of the heli x of Ex. 1 fintl the direction cosines
of the tangent, principal normal, and binormal. Also lind tho valnes or p
and tT.
3 . .Find the radius of the osculating sphere at an arbitrary point of the
space cubic x = t, y = z =
4. Show that the equations of a curve, referred to the moving trihedral of
a point P on it, may be written in the form
s2 (tlo) s,
!!=z,- tl.-! ()p2+, z=-6ptT +
s being the length of arc from P.
5 . Di scuss the equations (31) of a minimum curve in each of the follow-
ing cases:
(a) f(t) a quadratic function oft.
(b) f(t) a cubic function oft.
II. Analytic Surfaces
HHS. Parametric equations of a surface. The locus of a point
(x, y, z) whose coordinates are analytic real functions of two in-
dependent real variables u, v
x =J.(u, v), y=J.;(u, v), z=.fa(u, v),
(32)
such that not every determinant of order two in the matrix
oft oh gfj
a lt iht au
g6 gfj
(33)
av ov {) v
is identically zero, is called an analytic surface. The locus de-
tined by those values of n, v for whi ch the matrix (33) is of tank
less than two is called the J::teobian of the surface. Points on
the Jacobian will be excluded in the following discussion.
'l'he reason for the restriction (33) is illustrated by the follow-
ing example.
EX.\)II'Ll:: . Consider the locus
x =u+ u, y=(u+ u)2, z =(u + v) 3.
For any vah1e t , an.v pair of values u, v which satisfy tbe equation
" + u = t define thr> point (t . t l). Tlw Incus of the equations is a space
cnllic curw . Iu this example the 1natrix (3!1) is of rauk one.
ARTS. 196-198] NOTUfAL 235
The necessary and suffic-ient condition that n, ,. enter;;,;;, .h in
such a way that .c, !/, z can be expressed as functions of one vari-
able is that the matrix is of rank less than two.
197. Systems of curves on a surface. If in (32) u is given a
constant value, the resul ting equations define a curve on the sur-
face. 1f 1t is given different values, the corresponding curve
describes :1. system of curves on the surface. Similarly, we may
determine a system of curves v = const. The two systems of
curves u = const., v = coust. are called the parametric curves for
the gi,en equations of the surface; the vatiables u, v are calleu
the curvilinear coordinates on the surface.
Any equation of the form
cp(u, v) = c
determines, for a giYen Yalue of c, a curve on the surface. The
parametric equations of the cur ve may be obtained by solvi r1g
for one of the variables and substituting its value in terms of
the other in (3:!). If we now give to c dif'flrent values, equation
(3-1) detetmines a system of cmves on the surface.
If cp(tt, v) = c, .p(u, v) = c' are two distinct systems of cun es on
the surface, such that
a.v _ aq, a.;$
0
au iJv iJv iJu '
by putting cp(u, t) = u', .; (tt, v) = v' and solving for u, v \Ve may
express x, y, z in terms of u', v'. This process is called the trans-
formation of cmvilinear coonlinates.
l!lS. Tangent plane. Normal line. The tangent plane to a sur-
face at a point P on it is the plane determined by the tangents
at P to the curves on the surface through P.
The equations of the tangent lines to the curves u = const.
and v = const. at P=. (x, y, z) = (u, v) are (Art. 190)
X - X r-1! Z-z
--=-- =--,
ax a.l/ az
av uu av
X- x Y- 'I Z-z
a;-= uy. =az-
au au uu
250
DH,I'ERENTIAL GEOME'rRY [ CHAP. XIV.
The plane of these two lines is
X-x Y-y Z - z
ax a!, az
au au a1L =0. (35)
ax
ay az
av av av
Let v = cf>(n) be the equation of any other curve on the surface
through (11, v) . . The equations of its tangent lines are
X-x Y-y Z-z
a.t +ax clef> az + az <let>'
an av <lll att av clu an av dtt
This line lies in the pbne (35) independently of the form of
cf>(n), hence (35) is the equation of the tangent plane.
The notmal is the line perpendicular to the tangent plane at
the point of tangency. Its equations arc
X-x Y-y Z-z
ay az - az ay = az ux - ux az = ux ()y- ay ax.
att av alt 01) uu Ul) au av aiL av au av
We shall denote the direction cosines of the normal by X, ji., v.
Their values are
(36)
wherein
D
2
= (a.!/ az _ az ay)2 + (az iJ.t _ ax az)
2
+(ax ay _ ay ax)
2
(S
7
)
au av alt av au av au av OIL av au av
If D = 0, the tangent plane (35) is isotropic (Art. 152), and the
formula for determining the di rection cosines of the normal fails.
\Ve shall limit our discussion to the case in which D =t- 0.
The equation of the tangent plane may be written in the form
X.(X- x)+ j:t(Y- y)+ v(Z-z)= 0.
ART. 190] OF ARC 257
109. Differential of arc. Let (11, v) = 0 be the equation of a
curve on the surface (32). The differential of the length of arc of
this curve is given by the formula (Art. 18!))
<lf = + + clz
2
,
in which
ox ax u11 iJ11 az az
d:c=-dn+-dv, dy=-'- du+-'- 11 dz=-dtt +-dv,
au av iJ!t av ' att av
and the differentials du, clu satisfy the equation
adu+ adv=0.
au av
If we substitute these ''alues for d.t, dy, clz in the expression for
cls we obtain
(38)
in which
E = - +- +-'
(
ax)2 (a
11
)2 (az)2
au an an
(3!3)
(
ax)
2
(a.11v (az)2
G = av + a v) + av
Since the expression <j>('u, t) does not cuter explicitly in the equa-
tion the cx1wession for cls has the s;une form and the coeffi-
cients B, F, G have the same values for n.ll the curves passing
through P, bnt the value of clv: clu (]epends upon the curve chosen.
The coefficients E, F, () are called the fundamental quantities of
the first order. From (37) and (3!3) it follows that
D
2
= EG-F
2

Let 0 be a curve on the' surface through (tt, v) and let ds be the
element of arc on 0. The direction cosines >..,

v of the tangent
to 0 are
>.. = dx = ax + ax dv, fl.= dy = uy cl1t + ay dv,
ds au cls av ds cls an ds av ds
v = clz = (}z d n + az d v .
f)s i) u. ds d1J cls
258 GEOMETRY {CRAP. XIV.
If we replace ds by its val ue from (38), divido numerator and
denominator of each equation by clu, and replace dv: du by k,
we have
ax +k ax
A._ au av
- v E + Fk + Gk
2
'
a.11 +kay
au au
p. = ---===,...,....---===
vE+"J Fk + Gk
2
'
(40)
az + k az
aiL iJv
v= .
v E + :! Fk + Gk2
It follows from these equations t hat at a given point on the sur-
face the tangent line to a. curve passing t lll'ongh the point is
uniquely determinecl when the value of the ratio clv: du = k is
k
ax t fi 1 1 t... ( )
nown, smce -, -, e c., are xec w 1en ue pomt 11, v IS gtven.
alL au
200. Minimal curves. Each factor of the expression
Edn
2
+ 2 Fcludv + Gclv
2
,
when equated to zero, cl elermines a system of cmves 011 the sur-
face. Let 4> (11, v)cltt + tb(u, v )dv be such n. facto1-. Hy equating to
zero and integrating we ohtain an equation of the form f(u, v,)= c,
in whi ch cis a constant of intE'gration, which determines a system
of r unps on t llP sul'facc.
The two systems of cm,<>s clelermined in this way are minimal
curves (Art. l !lii), s ince the differential of arc of every cmve of
each system satisfies the eonclition
cls
2
= Edu
2
+ 2 Fcluclv + Grl v
2
= 0.
This c>quation cleterminr.s, at (n, v), two \'itlncs of the ratio
dv: llu = k whit"11 d<'fine two ima.ginary tangC'nts to minimal
curves. ThC' two tangents coincide at poiuts for whi ch D= 0.
l11 t he su<creding discussion we sha.ll assume that minimal
curves are excl uded.
ARTS. 200-202] RADIUS OF NORl\L\L CURVATURE 259
201. Angle between curves. Differential of surface. 'l'he angle
between the tangents to the curves 1 = coust., v =con st. is deter-
mined from Art. 198 by the formula (Art. 5)
F . . vEG-P
2
D
cos w = --, from winch sm w = = --
vEG EG vEG
The curvilinear quadrilateral whose vetticcs are llcterllli ned by
(1c, v), (n + .lu, v), (u, v + (u + 11 + is :lpproximately
a parallelogram such that the lengths of the adjacent sides are,
from (38), v Edlt, v Gdv, and the included angle is w.
Hence we have in the limit for the differenti al of surface
clS =sin wv EGcluclv = Ddudv.
Let G, 0
1
be two given curves on the sur face throngh a point J>.
\\"e shall denote the of 11, c, son G IJ.\ cl11, ell, cls and
the lliffereutials of 11, t, son 0' by Su, Sr, Ss. The direction co-
sines .\, p., v of the tangent to (}are cletmminetl by teplacing kin
(.tO) by clv: clu; siruilarly the di rection cosines .\
1
, p.
1
, v
1
of the
tangent to 0
1
arc llcternrined by replacing 1.; by Sv: ou.
If(} is the angle betw<.'en the tangents to G and C
1
at (11, v),
8
1
1
I Etla81t + P(tl1t8u + cli'Su) + O<lvSo
COS = .\.\ + p.p. + vv = <lsSs ---
(41)
From (41) we have at ouce the fol lowing tlreor<.'lll:
Tu t::IJHtDI. The coudition that two <lirectious cll'letlllillecl by lit e
l'fllio:j cit: cla, ov: Sn al'e orthoyonal iii
EclaStt + F(rluSv + cluStt) + GclvSv = 0. ( 4:?)
202. Radius of normal curvature. Meusnier's theorem. Let .p
be the angle whi ch the principal normal to G makes with the
normal to the smface. Let .\
11
p.
11
v
1
denote the direction cosines
of the principal normal and cis the differential of arc along G.
\\' e have, from (14)
COS I/J=A
1
A+p.
1
p.+v
1
v=p .\ - +fA- - ' +v- ,
- - - ( - d
2
x - cl
2
1J - d
2
z)
d.s
2
ds
2
ds'l
p being t he radius of curvature of Gat(, v).
260 DIFFERENTIAL GEOMETRY [CBAP. XIV.
h ' 1 f d
211
cPz S b h I
w1t sum ar cxprosstous or - , - u st1tute t ese va ues
c/f;2 l(s2
for the second derivatives in the equation for cos.;. Since the
normal to the smface is perpendicular to the tangents to the
curves 1t = const., v = const., we have the relations
\ i J . ~ ; + _ iJy + _ iJz _
0
\ CJ.t + _ CJy +- iJz _
0 " - p. - v -- , "- p.- v - - .
iJa iJu o1t iJv iJv iJv
If we replace ds by its value from (38), the equation for cos .; may
be reduced to
wherein
cos .p _ J,r1u
2
+ 3 1lfcluclv + Xclv
2
p - Eclu
2
+ 2 Fcluclv + Gciv
2
'
(43)
(44)
The quantities L, JI, N are called the fundamental quantities of the
second order fo1 the gi von surface.
The second membe1 of equation (H) <lcpenus only ou (u, v) and
the ratio <lv : <lu = k. Consider the plane section of the SUIface
determinetl by the nmmal to the surface and the tangent to 0.
Such a section is called a normal section. Let the radius of
cmvaturc of this normal section at (u, v) be denoted by R. From
(--13) we have
and hence
1 IAtt
2
+ 2 .lf<luclv + Nrlv
2
R = E<lu
2
+ :t F<lwlv + G<lv
2
'
Rcos op=p.
(45)
(46)
The result expressed in equation (46) may be stated. in the follow
i n ~ form, known as l\Ieusnier's theorem:
ARTS. 202- 204] CONJUGATE 261
TuEOREM. The ceuter of curvature of any JiOint of a cnrve on a
surface is the projection on its oscula.ting plane of the center of CltiTCt
lure of tlte nonnal section tangent to the C/lrve at the JiOiJtt.
203. Asymptotic tangents. Asymptotic curves. The two tan-
gents to the given stnface at (u, v) defined by t he equation
Ld1t
2
+ 2 J f cluclv + Nd ,p = 0
are called the asymptotic tangents at P.
From (45) we have at once the following theorem:
(47)
I. If the cwve 0 on the sm:face is tangent to an asymp-
totic tangent at (u, v), then either the plane to C coincides
1vith the tangent plane to the surface or 0 has cc linew in.fle:r;ion at
(u, v).
The two systems of curves clefi ne<l by the factors of (47) are
calle<l the asymptotic curves of the surface. They have the prop-
erty that theit tangents ate the asy rn ptotic tangents to the sur-
face. \Ye have the further theorems :
Tn EOJUnt II. If a straight liue lies on a sm:fuce, it coincides
with an tanrtent at each its ]}()ints, luuce the line is em
ic cune.
III. The osculctting plane at each point of ct ?eal
Clsym ptotic curve, not et straight line, coincides with the tangent }Jlaue
to the sUI:{ ace ({l tltctt point.
20-!. Conjugate tangents. The equations of the tangent planes
at P= (x, y, z) and at P' =(x y + z + on t he surface
are (Art. 198)
X(X- :z:)+ fi(Y- y) + v(Z- z)= o,
x - + Y- y- t:. y)
(48)
Let P' approach P along a curve whose tangent at P is deter-
mined by k = clv: clu. We shall now determine the limiting posi-
tion of the line of intersection of the planes. If we subtract th<
262 DIFFERENTIAL GEOMETRY [CHAP. XIV.
first of equations ('l8) from the second, member by member, aud
pass to the limit, we have
dA(X -x) +d;:L(Y -y) + dv (Z-z) + Adx+ jj.dy+ vdz = 0.
But Adx + p.dy + vclz = 0, sin,.e the normal to the surface at P is
perpendicular to every tangent at P. Hence the limiting position
of the line of intersection passes through P, since it lies in the
tangent plane at P antl in the plane dA(X- x) + ctjj.( Y- y)
+ dV(Z- z) = 0 through P. Let the point (X, Y, Z) on the line
of intersection be denoted by X= X+ OX =X+ ax ou +ax ov, etc.
au av
(Art. 199). lYe have
d>:.(i:Jx ou + i:Jx oo) + clp.(ay ott+ ay ov) + czv(az ou +azov) = 0.
a t ~ iJv au av dlt au
If we replace \ p., ii by their values from (36) and simplify, this
equation reduces to
Lrluou + M(duov + dvon) + Ndvov = 0, (49)
which determines ov: ott linearly in terms of clv: clu.
Since equation (Hl) is symmetric in dv: d ~ t and ov: Ott, it follows
that if a point P" apptoaches Pin the direction determined by
ov: ou, the limiting position of the line of intersection of the tan-
gent planes at P and P" is determined by clv: clu.
'fwo tangents determined by dv: clu, ov: ott which satisfy (49)
are called conjugate tangents.
TnEOitmr. Tlte necessary and S1{/fident condition that a tangent
coincides with its conjugate is that it is an asym]Jtotic tangent.
For, if in (49) we put ov: ou = dv: du, we obtain (47). Con-
versely, if clv: cltt satisfies (47) and ov: ott is conjugate to it, then
clv : du = ov : ou.
20u. Principal radii of normal curvature. In order to determine
the maximum and minimum values of R in equation (4f)) at a
given point (u, 'll) put dv: dtt =k and differentiate ll as a function
of k. The derivative vanishes for values of k determined by the
equation
(FN- GJif)k
2
-(GL-EN)k+ (EJII- FL) = 0. (50)
ARTS. 205, 206] LINES OF CURVATURE 263
If this equation is not identically satisfied, the two roots k
11
k
2
are
real antl distinct, since the part under the radical may be expressed
as the sum of two squares.
(GL- ER)2- 4(FN- G,lf )(EJl f - FL)
:= 4 -FL )
2
+[ER- GL - '2: (EX- FL)]
2

One root will determine the tangent ,zv: (lit s nch that th<' normal
section through it will have a maximum ra<lius of curvature R
1
and
the other will determine the normal section having the mininmm
radius of curvature Jl
2

The tangents at(tt, v) determined by the roots of (50) are calleu
the tangents of principal curvature, and the correspouding tadii
R
11
R
2
are called the principal radii of curvature. To dett>tmine
the val ues of R
1
and R
2
we have ftom (45) and (50)
L + kM Jlf + LV 1
E+kF= F+kG=u:
By eliminating k between these equations, we obtain the quad-
ratic equation
(LV- .1!
2
)1?!- (E:S- 2 FJf + GL)R + .EG -F
2
= 0, (51)
whose roots are ll
1
and R
2

The expression..!._ +.!.. is called the mean curvature of the sur-
R1 .
face at (11, v); the expression Rl . -.!, is called the total curvature
I )J,z
of the surface at (u, v). From (51) we have
..!._+ 1 =EN- 2_!'Jf + GL
R
1
R
2
EO - F2 '
(52)
1 LY- Jf
2
RtRt = EG- F
2
206. Lines of curvature. If in (50) we put k = <l v : du, we obtain
(EJJ- FL)du
2
- (GL- + GJf )dt? = 0. (;33)
The two factors of this equation detetmine two systems of cnrvt>s
called lint>s of cmvatnre of the surface. If the two dirt>ctions at
264 DIFFERENTIAL GEOMETRY [CHAP. XIV.
(u, v) of the lines of curvature are deuoted by rlv: clu and 8v : ou,
then, from (53)
GL-EN clvov EJlf- FL
duOu = F ... V- GJf cluSu = F.LV-G.J.ll'
from which
Ecluoa + F,clvotc + duov) + Gclvov = 0, (54)
Lcluou + .ilf(clvou+ cluov) + Nclvov = 0.
From the first of these equations we have, by ( 41), the following
theorem :
THEORE)t I. The two lines of curvr.ttttre at a point on the sm-
ficce we orthogonal.
From the second equation we have, by (49), the further theorem:
T u EORf:) II. The tangents to the lines of cun;attcre at a ]JOint
on the Stttjace ate conjugate clirectioi!S.
Conversely, if two systems of curves on the surface are orthog-
onal and conjugate, their equations sat.isfy (53) and (54), hence
they are lines of curvature.
The normals to the surface at the points of a given curve a on
it generate a ruled surface. The ruled surface is said to be de-
velopable if the limit of the ratio of the distance between the
normals to two points P, P on a to the arc PP' approaches zero
as P' approaches P.
It should be noticeu that in particular a cone satisfies the con-
dition of being a. de\elopable surface. A cylinder is regarded as
a limiting case of a. cone, and is included among developable
surfaces.
Tm:oRE)t III. The condition tlwt tlt e normals to ct smface at the
points of a curve on it tlescribc et developable is that the cu1ve is ct
line of Clt1'uatme.
Let P = (.1:, y, z) and P' = (x + y + z + L\z) be two
points on the given curve a. 'he equations of the normals at
P and Pare (Art. 20)
X= X + lr, y = y + Ji;, z = z + vr,
X = x + +(A+ Y = y + L\y + "ji.L\)r',
z = z + + ('V +
ART. 200} LINES OF CURV 265
The ratio of the distance to the arc (Art. 23)
Al Ax{l:iAv- vAP,) + XAv)+ P,AA.)
as= As y (/:iAv _ yj.p_)2 + ( vAA. - -A. AV)2 + _ p.j.A_)1
Di\ide numerator and denominator of the second member of this
equation by As2, and pass to the limit as 0. Using the dif-
ferential notation to indicate lim Ax: As, etc., we have
I
. Al
110 -
4,..1 As
vclji) + dy("Ycl""i..- ""i..clv) + :- (
55
)
-Y (jj.dv- vd/1-)
2
+ (vrlA.- Ad.-)
2
+ (M/1-- p.clA-)
2
Both numerator and denominator of the second member of this
equation vanish for those values of >.., ji., v which satisfy the
equations
and the limiting value of the rati o is indeterminate. The de-

nominator cannot vanish for any other values of A, Ji, ii.
Since A
2
+ + V2 = 1,
we have, by differentiating,
AdA. + p.rl"jj + vdv = O,
which reduces, under the condition that clA = kA, etc., to
k(X
2
+ J1.
2
+ Vl) =k= 0.
Since k = 0, we have ciA. = clP. = clv = 0. Hence the normal to
the surface has a constant direction for all points of the curve 0.
The surface generated by the normal is in this case a cyli nder.
If the denominator of (5.J) is not zero, the condition that the
surface generated by normals to the surface along 0 is a develop-
able is that the numerator of the second member of (55) is zero,
that is, that
clx(Jiclv- vclii) + dy(vdX- Xrlv) + clz(XdjL- p.cU..) = 0.
If we substitute for A.,-;;., ii their values from (36) and for clx, ely,
d th
1 ax l ax l d I . .
z eu va ues ,- r. ,, + ,-c t, etc., we can r<' uce t liS equatwn to
vu vV
(53), which proves the theorem.
266 DTFI<'J!:RBNT!AL GEO.\ l ETRY [CHAP. Xl V.
The i ndicat rix. Let tiLe liues of cnrva.ture be chosen for
parametric curves. Iu (54), clv = 0 and on = 0, but clu =1= 0, ov =1= 0,
hence P = 0, Jl = 0.
Let 0' be a cune making an angle () with 11 = cons. and let R
be the radins of uorma,l curvature in the direction of 0'. Along
u = cons., cl.s = VGdv, hence from (-H),
_ r--;-clv
cos e = v u - ,
cls
sin 8 = v E clu.
ds
From ( 45) and (52) we now It ave I he formula
1 cos
2
8 sin
2
()
- = - + - .
ll R
1
R
2
This equation is known as Euler's formula for the radins of
cnnatme of normal sections. It is intimately connected with the
shape of the surface about P.
J,et the surface be referred to the tangents of principal curva-
ture and normal at P rts X, Z axes.
Let .r, y be taken as parameters. The equation in x, y, z has
the form
z = (az)x + (az)Y + !(a
2
:)x
2
+( iPz )xy + !(a
2
z)Y
2
+ ....
ax ay 2 axay 2 ay
2
Since z = 0 is the equation of the tangent plane at the origin,
(
az) = 0 and (az)= 0. Since the X and r axes arc the tangents
ax iJ?J
of principal curvature at the origin,
(
azz) 1
CJ.t.'l = H/
- - =0,
(
CJ'!z)

(
CJ2z) 1
a!!z = R2'
hence, neglecting powers of x and !I higher than the second, t.he
approximate equation of the surface for points near (0, 0, 0) is
11
2
2 z = ;_ + .i!-,
R
1
R
2
(.36)
If 1.. and l_ are both different from zero, the surface defined by
R
1
R2
(56) is a paraboloid. J f one of them is zero and the other finite,
the smface is a parabolic cylinder. If both are zero, the surface
is the t...1.ngent plane to the given smface. This last case will not
be consiclcrccl further.
ART. 207] THE INDICATRL'(
267
'!'he section of lhe quadric (.56) by a plane z =cons. is railed the
indicatrix of the given sm-facc at a point P.
If R
1
anu R
2
base the same sign, the secti on is an ellipse for a
plane on one side of the tangent plane, and is imaginary for a
plane on the other side. I n the neighborhood of P the surface
lies enti rely on one side of the tangent plane. Such a point P is
calleu an el li ptic point on the surface.
If R
1
and R
2
ha,e opposite signs, the paraboloid (56) is hyper-
bolic and the section by any plane z =eons. on either side of the
tangent plane is a rea.! hype1bola. '!'he point P is in this case
called a hyperbolic point on t.he surface.
If ..!... or _1_ is zero, the section z =cons. consists of two paral-
H1 Rz
lclliues for a plane on one side of the tangent plane, :1nd is im-
aginary for a plane on the other side. It follows from (i'12) that at
such points

and from (47) that the two asymptotic

tangents coincide. '!'he point Pis in this case called a parabolic
point on the surface.
In all three eases, the directions of the asymptotic tangents to
the surface at a point P arc the directions of the asymptotes of
the indicatrix. At an elliptic point t he asymptotic tangents are
imaginary; at a hyperbolic point they are real and distinct; at a
parabolic point they are coincident. :.\foreover, conjugate tangents
on the surface are parallel to conjugate diameters on the indica-
trix. The asymptotic tangents are self-conjugate.
EXERCISES
1 . Find the equation of the tangent plane and the direction cosines of the
normal to the surface x ="cos v, y = u sin v, z = at the point (1t, v) .
2. Determine the differential equation of the asymptotic lines on the sur-
face defined in Ex 1.
3. Show that the parametric cur\'eS in Ex. 1 are orthogonal.
4. Find the lines of curvature on the surface x = a(u + v), y = b(tt- v),
l; = 1111.
5. Pro>e that if E: F: G = L : .Y for every point of a surface, the
surface is either a sphere or a plane.
.
l
.
:t
.,
2. The YZ-plane.
ANSWERS
Page 3 . Art. 1
s. The Z-axis.
4. A line parallel to the Z-axis through (a, b, 0).
6. (k, l , - m), (k, -l, m), (k,- I, - m) , ( -k, l, -m), ( -k, -l, - m).
Page 5. Art. 2
3. (- 1, 1, 0). 4. 13.
6 . ....!... ~ 0 1 1 1 - i
v5' V5' ' ...!3' va' ..;3; ..;Sl)'
6
-,
..;8l)
0, _1_, ~ . _ 1_, -4' _2_;
V5 V5' VTt v'2l v'2i
Page 7. Art. 4
1. v'89. 4. V(x- J )2+ (y-1)2+(z-1)2= v(x- 2)2+ (y- 3)2+(z-4)2.
6
. (a) z __ , __ 3 __ , __ 5 __ . (b) _1_, 1 1 .
...!38 ...!38 ...!38 ...!3 ...!3' ...!3
(c) 3 , 4_ , -4
...!41 vH ...!41
7. (a) Parallel to the rZ-plane.
(c) Parallel to the X-axis.
(b) Parallel to the Z-axis.
9 . ....!..., ....!_, 1
...!3 ...!3 ...!8
Page 9 . Art. 6
1.
vl05
14
2. 1, o, 0; o, 1, 0 ; 0, 0, l.
4 -3
3
' ~ :!tl' v'2tl
4.
-11
...!1435
8. (i, i> 2). 10. Two. .Y7a=!.
2
Page 11. Art 9
1. Sphere of radius 1, center at origi n.
2. Cone of revolution, with .-'t'-axis for axis.
2GO
11. (2, 2, 2) .
270
ANSWERS
3. l'lane through Z-axi s, making a ngle of with X-axis.
4. Cone of revolntion, with Z-axis for axis.
5. (a) p = 2; (11) p = 2; (c) p2 + z2 = 4. 6. = 45", p2 = z2.
Page 14. Art. 12
1. 3x+ 4y+2 z= l7.
3. - 4 X + 3 y + Z = 5.
2 . . r- y = 0. X- anti }-- intercepts zero.
4. k = 2. 5. ( - 3, 4, 5).
Page 18. Art. 16
1. r\ x - H ?I - r
3
z = 2. 2. x + 2 y = 0.
3.
7
4. _2__. .5. (4, 3, l); ( l, - 4, 3).
hi; v'l4
7. 25 X+ 3!l y + 8 Z- 43 = 0. 8. :) X - Y - 2 Z - 6 = 0.
9
. A1x + B111 + Czt + D! = + + C2z +
+ + Ct
2
..! B2
2
+ C2l
10. J.l(x2 + yz + z!)=(3.r + y- 2z- 11)2.
11. 2x-....!f-Z+3+8VU=0. 13. x- 2y -z+2=0; - 2,1,2.
14. 21 .t. .:_ !J y - 22 Z + 63 = 0. 1.5. 3 X+ 2 y + ;J Z- 15 = 0.
17. 11 :r- y + 1G z- 63 = 0 and 17 x- 13 y + 12 z- 03 = 0.
18. rn = 6. 19. k = -l
Page 21. Art. 20
1. ( a) (o. C! o, U o}
(b) (o. -!)5' u2' 0, -:'l1' o).
(c) (o
20
) (s o -a) (
2
!! 1 o).
'2' 4 ' '' 2' 7'7'
4.
x-3 _y- 7 z-3 4 2 -a
5. 'X+ 2y + 2 = o.
z- = -il;
v2u'
V2l)'
V2U
6.
k=
7. Yes. 8. k::-2. 9. No. 11. Yes.
Page 23. Art. 21
1. -1 1 6
3 va ' 3 v'3 ' 3 va
2. 2 X + y- 3 Z + 6 =0, :r. + y + Z - 13 = 0.
3. arc sin
16
4. x + 10 y + 7 z + 18 = 0.
v2u v;o
ANSWEH.S 271
5. 8 X + !J - 2t) Z + 6 = 0. 6. k =- l. 7.
x + 2z = 6.
' X - (t y - b
z - cl
8.
3 X - y + 3 Z - 7 = 0.
9.
I
'
!ltj n
1
=0.
12 1112 n2
10. 3 :r - ; y - 4 z = 0. 11.
k =
12. k = 2 and k = 3. 13. 2x- z=O, y=3.
Page 25. Art. 23
1.
\ :n'
2. vs. 3. 0.
4. _7_. 0, -8.
vi vTf3
5. v2;
1
-,
v'2
7. 15 X+ 43 = 0, J:t !f = 6 Z + 13.
v2' o.
6.
v5
8.
vi.t.
2
I
X1 - X2 lJ /21
9. Yl- Y2 1111 m2 = 0.
Z1 - Z2 Itt n2
Page 28. Art. 24
1. 61 X - 52 y + 35 z - !l:3 = 0. 3. X + 5 !/ - 3 z - H = 0.
2. 12 ,r- I i y + S z + 4 = 0. 4. Yes.
5. 7 X+ 12 y- 13 z + 8 = 0, X - 3 y + 4 z- 7 = 0.
Page 29. Art. 25
1. i y- 10 Z- 3 = 0, 7 X- Z- 22 = 0, 10 7- y- 31 = 0.
2. y - z + 2 = o, .r. + z = 1, 7 + y = - 1.
3. y- Z = 0, X+ 2 Z = 4, X+ 2 y = 4.
4. ( A1B2- + (. l1 AzCt)z + (A1D2- AzD1) = 0.
(B1A2- B2A1)x + (R1C2 - R2C1)z + ( B1D2- R2D1) = 0.
( CIA2 - C2ilt)x + ( c,B2- C2Bl)Y + ( CtD2 - C2DI) = 0.
Page 33. Art. 28
. z y z y z
1. X + 2 y + S + 1 = 0 j 3 X - Z - 4 + ) = 0 j - X + 4 - S + 1 = 0.
2. 5. a. (7, 5, -t); -
5
r., ); o. o) .
4. arccos
3
VTii.
v:.t9o
5. U + 11 + 1.0 a - 1, 6 1L - 3 V + W + 3 a 0, 6 U - 2 V + W + 1
8
0,
(
'>
3
. (- 1 - 2). ( - 1) I 1 I
6
- -
1
- >. 3'
0
' T '
0
'
0
'
7
' vs' ...;s' v3
8. A plane. 9. 4( u'' + + 1n
2
) = 1. A sphere.
272
ANSWERS
Page 35. Art. 34
1.
(- 7 -3
(a) 4' 4'
-/)
(cl)
(9 -1
2' 2'
-23}
(b) c - 1 )
s' 5'
0

(e) (.=..! =-!
7 I 7 1
(c) ( --;
1
, o, o} (!) ( 0, 0,
2. (- 10, 151 - 21 0). 4. 7 X + fJ !/ + 54 Z - 59 t = 0.
Pa ge 3?. Art. 35
1. (a) Parallel bundle. Rank 3.
(d) Parallel bundle. Rank !).
(b) Rank 4. ( c) Rank 4.
3. The determinant -
b a
is of rank 3; of rank 2; of rank l.
- 1
- 6\) 5 - Hl
4
' v'3867 I v'3807' v'3867
Page 43. Art. 40
1. x
2
- 3 yz + yz- 4 x - 8 y + 4 z + 4 = 0.
x'

2 z
1
3. X= -+::..;z_+--
1
v'2T v'ti v'i4
4 x' y' z'
Y=---"-+-
v'2T v'6 v'I4

v'T! v'o v'f4
4. New equation is xz - 2112 + 6 z2 = 49.
Translation is x = x' + 8, y = y'- 1, z = z' + 2.
6. 3 z2 + 6 y
2
+ 18 z2 = 12.
Page 45. Art. 41
2. (28 6i, -68i, 6=F24i) 4. (17-4i,
13 13 13 6
6. ( 13 + 9 i) ;e + (3 + 4 i)y + (16- 7 i)z = 23 + 64 i .
6. (1 iv'3, 0, 0).
P age 46. Art. 42
6 + 8 i )
5
1 - 4 i '
1. x2 + y2 = 4 z2. 3. xz + y? + z2 - 7 x + !I + 30 = 0.
4. 8(x2 + y2 + z2)- GS x + 48 y- GO z + :!75 = 0.
ANSWERS 273
5. z=&, (x-3)"
2
+(y-7)"l+(z-1)
2
=9.
6. 2 X- 14 ?J - 2 Z + 1 = 0, 4 X - 18 Z + 38 = 0. 7. (- 4, 4 i, 2).
Page 49. Art. 46
3. Center at (0, 0, 4); radius G. 4. x2 - y
2
= 1.
Page 51. Art. 47
5. r,
a
1. x2 + y2 + zl = 2&. 2. 9(y2 + z2) = (15- 2 x)2. Vertex ('}, 0, 0);
:t = 0, 9(y2 + z2) = 225; 4(xZ + = 9(5- y)2.
4. y2 + z
2
= a
2
; y = a.
6. 1; = 1.
4 9 9 4 9 4
( b) ?:..
2
- =: = 1. + =: = 1.
b
2
' (,'! a2
(c) y2+z2=8x; y
1
=G4(.c2+z2).
(<l) (x2 + y
2
+ 5F = !()- 4 x
2
; xz +(y- 1)2 + z2 = 4.
(e) y2 + zZ = si n2 x; y =si n v'x2 + z2.
(/) y2 + z2 = eb; y = e''z'-+-'.
Page 54. Art. 49
(b) ;c2 + y2 + z2 + 2 x- 8 y- 4 z = 16.
(c) 2y-10z+ 14 =0.
2. (a) Center (-I, -1, "'
34
.
2 2 2
(b) Center ( - 1, - 2, 3); radius 0.
( )
C (I 1 - 5) d" i YlO
c enter :j"' ii'
4
- ; ra ms -
4
-
(d) Center( - l, 0, o); radius/.
:.! 2
3. (- 4 3i,2Gi,G,O).
4. (2 2 i v'2 1 =F 2 i v':.!' - 2 v'Z i).
3 ' 3 3
6
" o.
1. y = J.
Page 56. Art. 52
2. Arc cos 0"0. The spheres have no real point in common.
2
3. x
2
+ y2 + z2 - 2 x - G y- G z + 10 = 0 am\ z2 + yz + z2- 2 x- 6 y
- 6z - 6 = 0.
274 ANSWERS
4. xz + + z2 + x - 2 y - 3 z = 0.
5. 2 x - 3 ?I + z + 6 = 0. The sphere is composite.
6. 10(x2 + yl + + 71 x - 08 y - 8!1 z - 185 = 0.
7. 4232(:t'! + y2 + z2) - 276 X+ :"!70 y + I !132 z + 225 = 0.
Page 69. Art. 59
1. Center (I, 1, - 2); semi-axes v;o,
2. Sphere; center (2,
3
1
- s); radius v'iii');j.
2 :l
3. y = 0
1
2 xz = 3 z
2
+ 5 z + 7. Rotated abont the Z-axis.
4. :r=1, y=z; X= I, y=- z; X=-1, y=z; X=- 1, y=-z.
5. ('t) Ellipsoid. (b) Hyperboloid of t\vo sheets. (c) Hyperboloid of
one sheet. (ll) Hyperboloid of revolution of one sheet. (e) Ellipsoid.
(f) Imaginary ellipsoid.
Page 73. Art. 64
1. Hyperboloid of one sheet. 2. Imaginary cylinder.
3. Elliptic paraboloid. 4. Heal cone. 5. Hyperboloid of two sheets.
6. Hyperbolic paraboloitl.
8. (a)
(b) (I- ,.z)xZ +(I- ,.z)y2 + z2- 2 ax+ az = 0.
Page 76. Art. 66
(
8 ..JfOl) 5 'F v' 10\l
1. - a . I 3
- 1 j vlO!I) .
(j
3. (- !1 -ls )
4.(- 1,2,-1).
2. (01 o, 0).
5. (1, 1, 0).
6. Vertex (0, - 1, 0).
8. Non-central.
7. Plane of centers 2(x - y + z) - 1 = 0.
Page 89. Art. 75
1. Hyperboloid of two sheets. Center (0
1
0, 0). Di'rection cosines of axes
j, - !1 i1 - ! ; !1 !1 i 5 :t
2
+ 2 y
2
- z
2
+ 2 = 0.
2. Hyperboloid of one sheet. Center ( 1, j
1
- i). Direction cosines of axes
2 + zv'f> V5 - 1 I)+ v'6 2- 2v6 - vs - t
2V15+4v6' 2V15+4v5' 2Vt5+-tvo; 2Vl il- .J v5
1

5- , I[> . -3 .J 2
2v' 16- -1 vo' V2fi' v'2u' vt!
5 + v'f) 2 + ii- ,Jf., . 3 2 10
z ='?!
ANSWERS 275
. (14 - 5 -2)
3. Hyperboloid of one sheet. Center
11
' u ' U
Direction
1
1 2 3 1 -5
cosines of axes, Vl4 ' v'
14
' v'
14
; v'S5 ' v'35 '
3 . 3
v'35' v'10 '
o, vw
llx
2
+ 4 -
11
V ( 0 0) D
. . . f 2 1
4. Real cone. ertex, l, , . 1rect10n cosmes o axes, -v-' v' - '
- 1 2 l 1 - 2 5 5 5
0; v'6' v6' v'6' v'3o' V30' V3o' 5x2 + 2 yz- 10 z1 - 0.
Ell
. . b l . I N . . (
2 10 11
) D' t. .
6. 1pt1c o 01c. 1 ew or1gm, 3, 3 ,
3
. 1rec 10n cosmes
2 1 - 1 1 - 1 1 0 - 1 - 1
of axes, _ r-:, _ 17: , _
1
-; _
1
- , _
1
- , _
1
- ; , _
1
- , _
1
- G x
2
+ 3 y
2
=
v G v6 v6 v3 v3 v3 v2 v2
2v'2 z.
6. Hyperboloid of two sheets.
1 -5 3
cosines of axes, v'
35
' v'
35
' v'
35
'
2 x
1
- 5 y
2
+ 7 z
2
+
12
0.
35
Center,

-
5
1
,
1 2 3
v'1-1' v'f.i' v'14'
22)
35 .
-3
v'iQ'
Direction
0, _ 1_
v'IO
7. Parabolic cylinder. New origin on 2 x + 2 y - zz - 1 0. 2 x - y
?. ?
0 D
. - f 2 - 1 2 . 2 2 2 . - I
; - = 1rect10n co:smes o axes, 3 , 3 , 3' 3' 3 ' 3 ,
-, :::.. 5y
1
+Gx= O
3 3 .
8. Two real intersecting pluucs. Line of ver tices, x + z - I - 0,
x - y + 2 - 0. Direction cosines of axes, 0,
-1 -1 1 3x2 1
11
2
v'3' v'3' v3' 2 -2 =
0

9. Hyperbolic pamboloid. Xew origin, (I , 0, - 1). Direction cosines
of axes _I_ _l_ . _I_ ::1 _I_ . _I_ 0 - 1
' V'6' V'6 ' v'6 ' V'3 ' v'3' v'3 ' ' ' v'2. G x2 - 3 y2
2v'2 z.
10. Ellipti c paraboloid. New origin, (0, - 1, I ). Direction cosines of
1 l 1 l - 1 1 1 -2
axes, v3' v'3 ' V'3 ; v'2' v'2 ' 0 ; vii' vii ' v'(; . 3 .r.2 + -1 y2 =
av'6 z.
11. Hyperbolic cylinder. Line of centers 3 x - 7 y + 7 z + 1 = 0, or
2 I. - 2 y + 4 Z + 1 = 0. 7 X + 5 y + 11 Z + 5 = 0.
X + 3 y + Z + 1 = 0.
D
. . . f 1 - 1 2 . 1 3 1 . - 7
cosmes o axes, v'G, v'6, V'/3 , v'
11
, v'li , v'
11
' v'6'6 ,
v'66' V'6G 24 X
1
- 11 y
1
- 5 - 0.
276
ANSWERS
12. ll ypcrbolic paraboloid. Origin
(
35 13
-
1
) Direction cosines
72' 72' 7-i.
of axes
1
v'7-
2 v7 - a . -1
vf28-10v'7' vf28-t0v'7' vf28 -!Ov'7' vf28 + 10,!7'
v'7 + 2 v'i + 3 . 1 - 1 1
vi28+ tov'7' vi28+ 10 v1' vs' vs' vs
( - I +v'7)%2 -(1 +v7)y2 = 1v'3z.
13. Hyperboloid of one sheet. Center(:!. -
7
, ~ ) Direction cosines
(j 4 18
of axP.s .21, - .U5, .60; .01, .41, .10 ; .30, - .G4, - .08.
3.09 x ~ + 1.59 y2 - 3.G7 z2 = f-i-.
14. Hyperboloid of one sheet. Center (.!, ~ , -
26
). Direction cosines
15 5 15
of axes -.77, .56, .28; .14, - .31, .94; .63, .76, .13. 6.17x2+ .ily2
- 6.88 z2 = lf5'.
15. Ellipsoid. Center (0, 1, 1) . Direction cosines of axes
2 1+v'fi O -2 - 1+v'5 OOOl
v'10+2v5' vl1o +2v5' vl10-2vs' v'I0-2v5' ' '
3 + VS x'2 + 3 - VS yZ + 2 z2 = 4.
2 2
16. Ellipsoid. Center (-
9
, =.2, - 0). Di rection cosine!! of axes .83,
2 2
- .33, - ..14; .20, .95, - .22; .49, .Q7, .87. 4.20 z2 +.59 y2 + .20 z2 = A,f-.
17. ~ . 5 2 i.
-29
18. 28'
Page 92. Art. 78
1 x +10y-3z+22=0, x -l=Y+
2
=z-l.
. 1 10 -3
Page 96. Art. 80
1. Y=E, z =Ex; X=1J, Z=1JY
1. '5, v'fi.
2
Page 97. Art. 81
2. ;c + z + 1 = 0, y + z - 1 = 0. s. a, b, e.
ANSWERS 277
Page 103. Art. 83
1. x+y-z=(landx-y+Zz=p. 2. x - (2V6)y=d.
3.
1
4. y+3 v"ii(z - 2)=0. 5. a=b, h =O.
,a
6. a:r + yz +I= 0, ay + fz + m = 0. 7. 2 gx + 2fy +(c - a)z = cl.
8. (b-k)A2+(a-k)TP-2ltAB =0,
(c- k)02 - 2jBO = 0,
tct- k)02 +(c- k).t!Z -2 g0.1 = 0,
k being a rooL of the discriminating cubic.
9. (-1,0, - 3). vao.
2
Page 108. Art. 87
3. k, =cons. i = 1, 2, 8, 4. .For parametric equations, substitute this value
of k;, in Eqs. (27).
Page 111. Art. 89
1. (- 6 X+ u y - 12 t, X+ 2 y- 2 z + t, u X+ G y + 4 z + 4 t, - .t: + 3 y
(- 12,1,4,-2), (18,-6,-16,1).
(12, -3, -28, 1), (3, :.!, 1, 2).
2. ( - 373, li!l, U:.!, 283), (-500, 181, H5, (- 153, 01, 38, 107) ,
(- 37 :Ct- !J(l X2- !l X3 + 150 11 Xt + 24 :t2- 3 :1'3 - 60 8 Xt + 48 Xt
- U .c
3
- 36 x., 31 Xt + 60 x2 + 3 J'3 - 108 x.).
3. 15 X+ 5 y + 11 Z + 16 t = 0.
4. 11)7 Xt + 468 X2 + 57 x
3
- 702 :r
4
= 0.
!i. 6 .c2- 15 + 2 z2 + 3 yz- zx - 3 xy + 17 xt + 0 yt- 6 zt + 10 t2 = 0.
6. (22:t - 22y+441, 12x+24y-24z+12t, 33x+33y+22z+22t,
60:r-198y+U6z+132t), (22, II, 11, 00), (22, 12, .:iS, 33), (:1:3, au, 44, a3),
(22, IS, ii, (22, -48, -11, 2G-l), (- !17121, sr..t27, 22804, uGS51),
(20uW7,- 115487, - u-1316, - 20508I), (- I85H25, 7118I, 42570, 128403),
(8I4 .C!- 6012 Xt - 207 X3- 61776 X4
1
- 242 XI + 1728 X2- 99 X3 + 23760
-li6x1 + -198 xa + l.J2.:i6..:,, - 682:tt+4320:tt+09x
3
+42768x
4
).
Page 113. Art. 92
1. Vertices: 111 = O, (1, 0, 0, 0); 112 = 0, (0, 1, 0, 0);
11
3
= 0, (0, 0, 1, 0); u
4
= 0, (0, 0, 0, 1).
!-'aces: Xt = 0, (1, 0, 0, 0); :t2 = 0, (0, I, 0, 0);
:rs = O, (0, 0, I, 0); :r
4
= 0, (0, 0, 0, 1).
2. Xt = 0, X2 = 0 j u
3
= 0, 1l4 = 0.
x1 = 0, :ra = 0; 112 = 0, 11, = 0.
XI = 0, !r4 = 0 j 1lt = 0, 113 = 0.
278
ANSWERS
Xz = 0, X3 = 0 j U! = 0, 1/4 = 0.
Xz = 0, X4 = 0 j 1t1 = 0, 113 = 0.
X
3
= 0, x
4
= 0; 111 = 0, 112 = 0.
s. 11) + 112 + 113 + 114 = 0, 3 11) - 5 112 + 7 113 - 1t4 = 0,
- u
1
+ 0 1t2 - 4 113 + 2 11
4
= O, 7 111 + 2 112 + 4 u
3
+ 6 11
4
= 0.
4. ( 1, 1, 1, 1), (7, -1, - 3, 1), (I, 9,- 5, 2).
5. 111- 1t2 = 0, 7 ua + 11
4
= 0. 6. ( - 0, I, 1, 0).
Page 117. Art. 95
2. p.r1 = !1 + 2lz + 10 /3,
J>Xz = 7 /1 + 5/z- l3, ( 170, - 17G, 40, 363).
px3 =- l1 + 4 lz - 3 Z3,
p:r4 = 31, + lz- SZ3.
3. p111 =- u It + 7 lz + 6 !3,
puz = 31,- 5/z - 4 Z3, (21, 32, 1, 5).
pua = 4 It + 3 lz - 8/3,
pu4 = /1 + 2 /2 + la.
5. pu,=l,+7lz, p1tz=- 5l,+2lz, Plla=3 l, -lz, ptt4 =-l,-/z.
6. px,=l,+3l2, pxz =2 l,-2lz, p.r
3
=- 3l,+5lz, px4 =-l,-2lz.
II ( 11) + 2 tl2 - 3 113 - 114) + 12(3 1t! - 2 112 + 5 113 - 2 114) = 0.
Page 120. Art. 97
S. ( I(Jl + tt12 + tt13 + lt14, l-'21 + ttz2 + tt2a + Ct24, a31 + Uaz + aas + fta4,
U41 + IG42 + 1(43 + U44). (P11 + P21 + fJa1 + P41, fhz + fJ22 + fJ32 + fJ42,
fJ13 + (323 + f333 + fJ43o (314 + fJu + fJH + fJH)
4. X1 = k,:r,', xz = kzxz', X3 = ka:ra' , X4 = k4x/.
Page 122. Art. 100
1. (a) x1 = x1
1
- xi, xz = Xz' - x/, X3 = xa'- x' X4 =- xl.
D(JJ) = (1 + p)( 1 -]J)a. Invariant points are ( I, 1, I, 2) and all the
points of :t4 = 0.
(b) X1 = xz', xz = X1
1
, xa = x/ , X4 = xa
1
D(p) = (p
2
- 1)
2

Every point on each of the lines
X! + Xz = 0, X3 + X4 = 0 j X1 - Xz = 0, X3 - X4 = 0.
(c) X! = X3
1
, Xz = X1
1
, X
3
= Xz
1
, X
4
= X4
1
D( p):; ( 1-p)
2
(p
2
+ p + 1).
The points ( 1, w, w2, 0), (1, wz, w, 0), w8 = 1, and every point of the line
X! =Xz = X3.
(cZ) X1 = - x.', Xz = X1
1
- x/, xa = xz'- xi, X4 = xa' - X4
1

D(Jl) :;p4 + p8 + pZ + p + l. (0, 1 + O, - 0Z (1 + 0), - 02), qs =I, 0 =I= 1.
3. l(ik =cons. i, k = 1, 2, 3, 4.
fhi
ANSWERS 279
4. In case x
3
= .r
3
', the poi nt (0, 0, 0, 1) and all the poi nts of the !>lane
=0.
Itt case x
3
=- x
3
1
, every point of each of the lines xa = 0, = 0; x
3
= 0,
0.
6. (1 , 1, 1, 1), (1, - 1, 1,- 1) . (1, i, -1, - i), ( 1, - i, -1, i).
7. All the points in the phwe at iufinity.
9. H
1. :ta - = 0.
Page 125. Art. 102
s. 2 iv'Ef, - 3 =F iv'12i, 8, 4).
2. tJ. = - 1.
Page 131. Art. 106
4. ttt2 + u22 + 1ta2 + = 0.
a b c cl
6. bcu
1
2
+ + 2 abuaUt = 0.
7. J. =f. <l>(tt) = uzt- ltt llz + ttatta- Uztta - + UzUt - 2 Uall t = 0.
8. A(x)::O.
9. .r1 = O, xa - X-! = 0 ami Xt = 0, :ta + = 0.
10. a a= 0. i = 1, 2, 3, 4.
12. a,
1
2 =a,. au, i, k = 1, 2, 3, 4.
18. A conic ; two distinct point.<; ; two coincident poiuts.
Page 134. Art. 111
1. Xt + X2 + Xa + Xt = 0. 2 .. ra = 0, X4 = 0.
4. (2 XI+ X2 - 3 Xa - :t
4
)
2
+ .j (X1:t2- XaXt) = 0.
7. x1:::u,xt- xz:::u :r, = 0. Three.
8. + aat:tt.ri + azaxz.ca + = 0.
Page 141. Art. 118
3. aux1
2
+ azzxz
2
+ aaa:>-a
2
+ 2 aaz:ttXz + 2 azaX:f.ta + 2 Uta:rt:ra = 0.
4. au.ca
2
+ a2z:t2
4
+ 2 auxaxz + 2 et2a.cz:ta + 2 = 0 .
.ca = 0, xz = 0.
Page 143. Art. 120
1. 8x1
2
+ xz
2
- 5 xa
2
- 2 Xt
2
+!) :ta:tz+5 Ttxa+ 18 XaXt + 13 xz:rt- i xax4=0.
2. i(.r.t + :ta)- (xz + Xt) = 0, i(xa + :r
4
)- (:tz - xa) = 0, antl
i(.ra + :ta) + (xz + Xt) = 0, i(xa + Xt) + (xz - :ta) = 0.
5. Equations of faces

= 0, i = 1, 2, 3, 4.
280
AXSWERS
Page 146. Art. 122
1. 2. k=4.
S. 72 + 36 ur + 23 M 1111t2 = 0; 2 113 - 3 u. = 0.
6. (a) A quartic curve with double point at 0.
(I!) A cubic curve pn.ssing 0.
(c) A plane section of JL, not passing through 0.
Page 1 50. Art. 126
l . l<lit- a;z - Mil I lu,,
a .. t- )..lJ.,
1
a.,
1
- )../1.,
1
1c.,,
2. (11) ).. - 1, ).., (1(21)].
(I!) )..f (4].
(c)
(d) X3, ).. - 1
Page 1 56 . Art. 131
2. ( 111]. Four distinct lines.
[21]. Two distinct and two coincident lines.
(1(1 1)]. Two pairs of coincident liues.
(3]. Three coincident aml one llistiuct line.
( (21 )]. !''our coincident lines.
( ( Ill )]. A qu:ulric cone.
{3}. A pl:111e and a line.
S. (a) X- 1, X- !, X - t
(li) X+ t" ).. - ! 1 X-!
(r) (X -l)Z, ).. - !
(d) (X -1)3.
4. (a) .Fonr distiuct lines.
( Ill ].
( 1(1 1)].
(21].
[:1).
(b) Two pairs of coincident lines.
(1') Two distinct and two coincident lines.
(tl) Three coincident lines and one uistinct line.
:r,.. X t
5. (a) + _r: (:1:1
2
+ .t2
2
+ = 0.
2 3
(/.I) ::: L + + :132)- )..(zl2 + .C2'! + .r32) = 0.
11 5
(c)

+ 2 .c2.c3 + xr- X(:t1
2
+ 2 .C'!L3) = 0.
(d) 2 + x
3
2
+ 2 ::C!X
3
- X(2 X1X2 + X3
2
) : 0.
Page 164. Art. 133
((22)].
(31].
2. [ 11 ( 11 )).
(I(:H)j.
[ 1( 111)).
vl\, - X3:r, Vl\a - X2t2 = 0, x,
2
+ x2
2
+ xa2 + x42 = 0.
Vl\2 - l\
1
3'1 .ra = 0, :r,
2
+ 2 :rz.l'3 + 'J.,
2
= 0.
La= 0, X2
2
+ :1:3
2
+ = 0.
ANSWEHS
281
(22]. :r, = 0, :r2 = 0.
(:.!( II )). :r, +i:r3=0, 7
4
=0; Jt - i.lz=O, ,., =0; J'3=:0, :ra
2
+xz2+J'
4
2
=:0.
[(11)(11)] . :r
1
+ i:t2 = 0, X3 + ix = 0; Xt + ixz = 0, J'3- ix = 0;
,., - i.l'z = 0, Ya + ix" = 0; .c, - i.c2 = 0, J'1- i:r4 = 0.
(4]. Yz=O,:r
4
=0.
[ (22)]. :r-
2
= 0, .l'
3
= 0; ,., = 0, .. , = 0; .r
3
= 0, :r
4
= 0; the last one
tnunwcl twice.
[ (31) ].
( (211)).
[l3}1).
:r1+i:Y4 =0, :ra=O; :Ya-i.r, = O. 'a=O ; .r4=0, .ra
2
+2J'2:r3=:0.
,,.
1
= o ..
3
= n ; x
1
:.:: n, .r
4
== o.
Xz = X3 = 0; .ft - rl - 0, :! .r1r 1 + = 0.
3. ( 11) (X- !)2, (X - !)2. 7t- .rz f), 'a + ::! '4 = ll i
Xt - 7z- V:l(.-'3 + 2 :1'4) =: II, '\
1
l ( .rt + ''2) +:! ''3 + .r4 :;: 0 j
:r,- .rz + VS(:r3 + 2.c4) = 0. ' .:( J't + xz)- 2 "3- :r, = 0.
X1X3 + Xz.' + :! Ja.r
4
- X(2 .r1:r3 + 2 Jz:r 4) = 0.
(b) X -1, X- 1, (X+ 1)2. Xt + .r3 = 0, 7t- Xz + i:r
4
= 0; J't + X3 =: 0,
. , - .l'z - ix = 0;
..r
1
+ x
3
+ 4 7z = 0, x2
4
+ (:r, - 7z)" - 2.J Xz
2
- JG .rz:r3 = 0.
x a
2
+ xz
2
+ :r3;r,,- X(J'tJ + :rz
2
+ .>'4
2
+ :2 .ra.l'-t ) = 0.
(c) X+3, X-1, i\-l,X - 1. :rt+2x
3
+x
4
=0, ;;.r,
2
-:tz
2
+6.r3!
+ 4 Xt.T3 + 2 XtX2 :;: 0.
- 3 .i'tz + :rzZ + :r32 + x.,z- X(:r,z + :r2: + :r3z +:r, 2)= 0.
(<Z) X- 1, X-l,X2. :ra+:rz=O, Xt+.r3+.r,=0; J t +:rz=: O,
Jt- Xa- :r, = 0; 7z+:r = 0, 3 .ra
2
+ Xz
2
- :t3
2
+ 4 :r,:rz + 2xzJ3 = 0.
,.,z + 7z2 + :r.,z- X(.rt2 + :rzz + x4z + 2 :r3x4) = 0.
4. [I (111)). [2( 11 )].
Page 167. Ar t . 135
1. (X3 + 2 X2 + 4 i\ + J)u
1
z +(3 X2 + i X-

+ 2 xz + !l X +6)u
3
2
+(X
2
- l){X- l)t,
2
- G(X:!- A)t<t112 + 12(>. + l rtttll3 + G(i\
2
- l)llt ll 4
+ 4( i\Z- X)1121t3 + 2 X(i\- I )
2
112114 - .J (X2 .:.:"t) IIJll4 = 0.
2. 2 xzZ- 3 x,z + ll xa:r
4
+ :l :rz:r - 4 :r3:r
4
= 0, twice.
2 :r1
2
+ 2 :r3
2
+ 3:r
4
2- ll.rt:r4 +:! :rzY< + 4 Ja:r =:0.
3. 2(tttll2 + tta>t,)i\3 + { >tzZ- G l'!lltll2 - G ffll311
4
);o..2
+ (fl azu,rtz + U a
2
113114 - 2 a>tz
2
Ji'. + a2uz2 - 2 a
3
u,uz- 2 nht
3
u
4
= 0.
Page 174. Art. 142
3. (a) (211]. (b) [22]. (c) (31].
4. All the quadrics of the bundle touch a fixed line at a fixed point.
5. The quadrics touch :r:
1
= 0, :rz- 2 :r
4
= 0 at (0, 2, 0, 1 ), and :r
3
= 0,
x2 + 2 x, = 0 at (0, 2, 0, - 1); they have four hasis points in the plane
x2 - X3 = 0, at the points
282
ANS\\"ER:>
(2, 2, 2, .J3), (2, 2. :?, - , r:J), ( - :!, 2, :!, V:i), (- 2, 2, 2, - .J3).
!!'1 = .J y'. = .J !!I'J4 <:! !/3 -
Y
1
a = I !/1!/3!/4

= !/1 (:! + :t !12!/3- 112

2
).
1. ;>" ( tt:r )( u..c') [ ( u
111
x )( 11
11
.
11
) - ( u" :r) ( u
1
".r')]
+ X2( u
11
x) ( u
11
:r') [ ( u'x) ( lt.') - (ux) ( u
1
.r
1
)]
+ X3[ ( u
1
x) ( u
111
.t') ( u:r.')( u
11
.r
1
) - ( ux) ( u
11
.r) ( u
1
x
1
) ( tt
111
x') ] = 0.
(For nolalion, sec Art. ll!l. )
Page 180. Art. 146
2. !/IYzYaY
4
= 0. 3. The plane counted twice is a quadric of the web.
4. (xz + yz + zZ)t2 = 0.
6. Any point on :r2 + y
2
+ = 0 is conjugate to any point on t = 0.
Page 187. Art. 1 50
1. = 0 . 2. 8. 4. s.
.5. 10
2
) + Xz(v
2
- to
2
) + >-a>lll + X.1n1o + = 0.
6. (1111). (a
2
- + (b'!- = sz;
((tz- c2)v2 =
(a2- bZ)uz + (aZ- c2)v! + sz = 0;
uz + vz + tG2 = 0.
Page 196. Art. 155
1. k1 P = k2P = k
3
P = k
4
P.
Page 205. Art. 160
(x
2
y
2
z
2
)
2. (x- + y- + z-)- = ,(! + p. + c" t.
Eight.
3. 2 z(:rZ + yz + zZ) = (axz + btP}t. Eight. Fifteen.
7. Xt = :r2
1
(.r1
1
+ .r/)(:rz' + '4'), xz = xz'.r4'(:r.1' + x/),
:r3 = .rz'.r4'(:rz' + .rl), .r4 = .r3'.r/(:r2' + .r/).
( I, 0, 0, 0), (0, 1,0,11); the line :r.
1
= o, .r
2
=II. Tnuch at (0, 0, 0, 1).
8. :Tt = :rt
1
:T3
1
(.t1
1
.r.z' + ':2
1
.Ta
1
+

.z = . ,'.,./(.rl'xz' + xz'x3' + X3
1
X1
1
),
X3 = .r-2':ra'(.r1'.rz' + .Yz
1
.r.a
1
+ /3
1
.1'.1
1
), '4 = /t
1
:rz'.r3'x/.
( 1, 0, 0, 0), (0. I, 0, 0), (I, 0, 0, 0). Fou l' coincitlcnt at (0, 0, 0, 1 ).
Page 207. Art. 162
1. t:.,f(x) = Yt(a4oOOXta + 3 azwoX1T2
2
+ :1 llzozoXt:r.3
2
+ 3 CIW()Z:T'1X4
2
)
+ .J ?/2(::) fi2Zoo.rt
2
.rz + ''o4oo.r1 +:} flozzo . .,.z.r3
2
+ S aowz:r.2.r4
2
)
+ .J ?la(:l

+ :)

+ rtoo4ori + :3
+ 'I ... , + J + :l +

AXSWER::>
j,
11
'!f(.t) = 12 + + +

+ 12 + -r + uozoz.rZ)
+ 12

+ + +
+ 1:? + + floo::.t3Z +

+ 18 YtY:tlzzooYt.tz + Hi !/t!/all:o:o 't'3 -t It\
+ -1 8 + 4:< + !la!lt' loozz:L3.r4
c..$f(x) = :?l(y.
3
ttoooXt + !12
3
ttooo.z 1- +
+ 3 + 3 !ft!lzzllzzoo.rt + :; !1
2
113'12020Y3
+ !{ !/t!/3
2
(1zo2Q.I' t + 3 !/ 1
2
!/fii:OOZ' + ;)
+ 3 Y:
2
y3aoz:o:T-3 + + !l r,t
+ 3 YzY
2
aozo:x: + :; + :l
C. ;J( r) =: 21/(y).
Page 209. Art. 164
1. (1, 0, 0, 1). (1, 0, 0,- 1), (-1, 0, 0,- 1).
2. 1, 2:J5!l + 131 Vf7 :!3-'i!l- 1:{1 '\ 1i
:;;o :)7fi
Page 211. Art. 167
1. llOOQn = 0, Clt.,lu- l'Xl + (/J)tOn t::Cz + 1/mt .. - t.f3 = 0.
2. riUUO = 0, UtO.ln- t = 0, = 0, floot u-t = 0.
3. :!(.11- :r
3
) + 5(x
2
- r
4
) = 0.
283
4. 2(.rl- l'a)+ x
4
) = 0, .J Yt + 3:! .r: -

'-'104:?(:rz -7) = 0.
Page 213. Art. 169
1. (xi"+ + z.
3
r = 0.
3. + u
2
1 + u
3
! + u
4
l = 0.
2. 111! + + ..._ = 0. 4. llt
2
u
3
+ = 0.
Page 215. Art. 172
3
_ _1_ + _!_ + _ 1_ + _ I_ + I = O.
ltJI'I az.rz 113.r3 a.x.t a6Vt + .rz + 73 + :r4)
P age 218 . Art. 175
2. Yt =0. 1'4'l.- Xz7s = 0: :rz - Y3 = 0. :r:-.r = 0;
.lt.l'3 + .l4! - :1':73 = 0, J3
2
- .l:.t1 - .t:
2
+ = 0,
.1.f + 1'21'3 + 2 7:.14 + X3.l4- .lt.l2 = 0.
3. (.1'1
2
+ .tz
2
- Ya!)Z- -l (:rt - X:)(7t
3
+ .ti' +

2 .ltX3
2
) = 0.
4. .t,: + .t-? + ,') .13'). = o.
S. (at - a.).rt2 +(a:- + (a
3
- a.)x3z = 0.
7. 2 7zl + 73
2
+ 2 .t
2
- 2 :r:.l'3)
2
- 2(7t-!} :e: + 2 :r4)((r
1
z + .r
4
2
- :r1 + .J rz- 2.ra)
-(Ttl + 2 :til- .1'3
2
) (2 .r1 - :rz- '3 + 2 .r.J)] = 0.
284
ANSWERS
Page 225. Art. 180
1. :r1 = t(t
2
-1), :r2= -1, :ra -1)
2
, x, = t.
2. ( 4 J
3
:t
4
- :r,:l%!)
2
- 4(x:f! + 2 :r txa)(:r,
2
+ 2 'zx,) = 0.
3. 12(u
2
2
- u
3
2
)(u2
2
- u,
2
)- 12(11t11z- ua11t)
2
+ (uz + 2uz2- tta2- 1142)2 = 0.

+2uz2- 11
3
2- u
4
2)
+ I8(u1
2
- 4 ttzZ + 2113
2
+ 27tt
2
)(71J112- 113114)
2
+ 2ur - 113
2
- n.
2
)3 = o.
Page 234. Art. 184
S. m = 3, n = 3, r = 4, II= 0, h = 1, G = 0, g = 1, ct = 0, fJ = 0,
v = 0, w = 0, :r = 0, y = 0, p = 0.
6. On the developable of the given cune.
Page 241. Art. 187
1. The four quadric cones on which C
4
lies.
2. Eight. Four of each regulus.
4. Hl stationary planes.
24 planes tangent to C
4
at e:\ch of two stationary points.
!lG planes t:\ngent at one aud passing through two other points
116 planes through fom tlistinct stationary points.
5. The developable surfacE' of C
4
The four quartic curves i n which the
faces of the self-polar tetrahedron intersect the developable surface.
Page 243. Art. 188
1. (a) 7n = 4, n= 6, r= ll. ll = 1, h =2, G=O, [1=6, (t = 4,
,, = 0,
w = O, :r = o. !I= 4, p=O.
(b) m=.t, n=4,
)" '=: :),
lf=O, h=2, G=O, (I =2,
(t = 1,
t: = <l, w= O,
.r = ?I = 2, p=O.
(r) 111 = 4, n =0, T = li , ll = 0, h =C!, G=O, fl =a,
It = 4,
v=O, w =0,
,. =li.
!I =4, p=O.
(<l) m = .t, n = 5, r = fl, 11 = 0, 4=3, r: = 0, (/ =4, (t = 2,
v= I, w = 0. :r =5, !I= 4, p =0.
( 1' ) tn = 1, n=.t, r=H, 11= 0, h =3, G=O, [/ = 3, (t = 0,
v=2, w =0, :t = 4, y= 4, p=O.
4. -1, 2, ! 9. Fonr. 10. Four. Two of each regulus.
11. 111 = t3- 3 tZ- 2, 112 = .J t (t + 1)2, 11
3
= - t\ u, = tS(t + 1)2.
'l = l,uz =-2t, u
3
= :?t\ 11,=-'
1
.
fJ = o,
fJ = 1,
fJ = o,
fJ = 0,
fJ = 0,
285
Page 253. Art . 195
1. y=asin - s-, z= .}!__ .
a v i av':! i
2. Tangent --
1
- ....!....
1
vi a v:!. "' :! a i .J2
l'ri ucipal normal - cos , - 0.
a../2 tn i
Binormal ....!... si n -
11
-, -....!.... rO!\
1
2 a Vi .J2 a, 2 vi
,. = 2 a, = - 2 a.
3. R = ..!_(l+HZ+!Jtl)!(30(1+ 4 tZ+Ht
1
)+(-186t
7
+561 t
5

12 l+!lt!+0t1
4. (a) No curve. (b) A c11bic cur,c.
Page 26?. Art. 207
1. 2 u cos v x + 2 u sin v u = c + ,,z,
2 u cos v 2 u sin v - 1
- - --, t --
v' I + 4 u"! V I + 4 v' l + 4 u!
2. fluZ + II
1
<lvZ = 0.
4. (u + +a! + + V:V! + a I+ IP) =c.
11 + vul +a!+ fji = + vv! +a!+
Absolute, 53
Angle, 3
between two lines, 3, 22
between two planes, 22
Apolar, 182
Axis, radical, 47
of revolution, 50
major, mean, minor, 63
Binormal, 248
Bundle of planes, 31, ll5
of quadrics, 167
parallel, 31
Burnside and Panton, 239
CcntC'r, 76
of ellipsoid, 63
radical, 59
Characteristic, 160
Class of a curve, 225
of a surface, 210
Cone, 49
asymptotic. 06
minimum, 100
projecting, 217
quadric, 72
tangent, 212
points, 132
planes, 132
point aud line, 165
Contrap;redicnt, 119
Coordinates, 1
curvilin!'ar, 255
cylindric:ll, 10
clliiJtic, lOG
homogeneous, 33
hyperbolic, 139
plane, 31
polar. 10
spherical, ll
ll'trahedral, 109
CorrP$pondcnce, 120
involutorial, 172
I NDEX
The numbers refer to pages.
Cross ratio, 121
Curvature, 248
mean, 263
total, 263
Curve, 46
215
asymptotic, 261
minimum, 252
parametric, 255
space, 215. 245
Cusp, 226
Cy<'lidC', 203
binodal, 204
Dupin, 204
horn, 204
nodal, 203
ring,
spindle, 204
Cylinder, 40
elli ptic, 72
hyperbolic, 72
imaginary, 72
paraiJOiic, 72
projecting, 47
Direction, 3
cosines, 5
Discriminant, 78, 126
Discriminating cubic, 79
Distaucc. 4, 7
between two lines, 24
between a point and line, 23
from a !>lane to a point, 17
Double !>Oint of a curve, 226
"I>!Jarent, 221
of u surface, 203, 210
Duality, 113
Ellipse, cubical, 235
Ellip$oid, 63
imaginary, 68
Equat ion of plane, 12
of point, 32
287
288
EquationR of a litH'. W
par:mwtrir, 13'!
Euler, 42
Fal'tOrs, invariant, 140
Fi<'ld, plnnr, 115
Fine, 216
Formulas, Euler's, 42
Freuc-t-Serrct, 250
GPnl'rator, 91
Genus. 228
HalphPn, 216
Harmonic. 122
2 12
Horopter, 235
Hypcroob. ruhirnl, 234
Hyperboloid of onr !!hrct. 65
of two "hPI'ls, 67
lmagP, 1:l!l
Indcp<'ndt'ul 36
Indicntrix, 2!l7
InOcxiou. litwar, 22G
lnlcri'<'JllS, 1!1
Invariant
point" I
rdativl',
undrr motion, R2
Ju,rrsiou, quadrati<-, 201
Jn,olution, 122
J sotropic pluncs, 54
Jacobian of a nl't. 170
of a we-h. 176
1"0
surfarc, 11'-0
Lnw of iucrtia, 136
J.ini'S, 13-1
minimum. 1!10
normal, 255
of ccu lt'rs. 7(;
of l'or vaturc, 26.1
of vt'rtircs. 76
Matrix, 37
2CO
Monoid, 2lU
Node, 226
Nocther, 221
Normal, 92
lNDEX
Normal form, 13
principal. 2-17
Octant, 2
Ordt>r of curv<', 170
of surface, 20
Origin, I
cubical. 235
Paraholoid, !'liipti<-, 69
hyrwrhol il', 70
Pnramc-t<'r, 21
Parametric E'<Jnations, 21
Prnril of planes. 26. 115
of quadrics, 147
PI'TSJWClivity, 196
Plane, 13
nt infinity, 71\
dinmctral, 75
douhiC' Oll<'nlati ng, 226
fund:1mcntul. 73
normal, 1!32, 208
of cl'n 76
principal, i S
rudic:tl. 57
rNtifyinl(, 217
133
226
tangent, 2t0
Plane:.. roordinatc,
l!lO
l>rOjC'ctinl(. :!6
Point, ut infinity. 21
stationary, 226
Points. asociat<'d. 168
circular, .33
conjugal<'. 132, 153
rlliJ>tir, 2G7
fundnm'ntal, 197
hy1>1'rholic, 267
imaginary, 44
pnrabolil', 267
Sl'lf-conjul(atc, !33
Polar rcciprO<:Jl figurt'S, 135
tetrahedra, 135
hyl>rrholic, 143
Projection, orthogonal, 3
quadric on a plan(', 13!1
quadric cone on a plane. 149
,;lrrcographic, 50
Quadric COD(', 7Z
78
78
surface, 63, 124
Quadrics. confocal, 104
Qu:trtic curve, 235
first kind, 242
non-singular, 238
Quartic curve, rational, 240
second kind, 237
Radii, reciprocal, 201
Radius of curvature, 219
of torsion, 249
Hange of poi nts, 115
Rank of curves, 224
of determinants, 37
of a matrix, 37
Reflection, 41
ltcgulus, 94, 138
Reye, 77
Rotation, 38
!';almon, 107, 177, 227
Section, circular, 98
Remi-axis, 63
Sphere, 52
director. 93
imaginary, 52
osculating, 251
point, 52
oblate, G4
prolate, 65
Steincrian, 214
:;urface, 46
algebraic, 206
INDEX
Surface, developable, 225
of revolution, 50
polar, 208
quadri c, 63
Tangent, 209
double, 226
iJlflcxional, 226
stationary, 226
Tangents, asymptotic, 261
conj ugate, 262
inflexional, 210
Tetrahedron, coordinate, 35
self-polar, 135
Torsion, 248
Transformation, bir::.tional, 197
of roordinatcs, 3S
projective, 120
Translation, 38
Umbilic, 101
Unit plane, 110
point, 110
V crt ex of bundle, 31
of quadric, 70
V crt ices of ellipsoid, G3
Web, 176
Weddle, 179
:>urfucc, 179
289
SE 16
SCIENCE AND ENGINEERI NG
LIBRARY
niversity of California,
San Diego
DUE
UCSD Libr.