Académique Documents
Professionnel Documents
Culture Documents
First published Mon Nov 28, 2011; substantive revision Fri Dec 11, 2015
To speak of Ren Descartes' contributions to the history of mathematics is to
speak of his La Gomtrie (1637), a short tract included with the
anonymously published Discourse on Method. In La Gomtrie, Descartes
details a groundbreaking program for geometrical problem-solvingwhat he
refers to as a geometrical calculus (calcul gomtrique)that rests on a
distinctive approach to the relationship between algebra and geometry.
Specifically, Descartes offers innovative algebraic techniques for analyzing
geometrical problems, a novel way of understanding the connection
between a curve's construction and its algebraic equation, and an algebraic
classification of curves that is based on the degree of the equations used to
represent these curves.
Related Entries
1. The Background to Descartes' Mathematical Researches
When Descartes' mathematical researches commenced in the early
seventeenth century, mathematicians were wrestling with questions
concerning the appropriate methods for geometrical proof and, in particular,
the criteria for identifying curves that met the exact and rigorous standards
of geometry and that could thus be used in geometrical problem-solving.
These issues were given an added sense of urgency for practicing
mathematicians when, in 1588, Commandino's Latin translation of Pappus's
Collection (early fourth century CE) was published. In the Collection Pappus
appeals to the ancient practice of geometry as he offers normative claims
about how geometrical problems ought to be solved. Early modern readers
gave special attention to Pappus's proposals concerning (1) how a
mathematician should construct the curves used in geometrical proof, and
(2) how a geometer should apply the methods of analysis and synthesis in
geometrical problem-solving. The construction of curves will be treated in
1.1 and analysis and synthesis in section 1.2 below.
The ancients stated that there are three kinds of geometrical problems, and
that some are called plane, others solid, and others line-like; and those that
can be solved by straight lines and the circumference of a circle are rightly
called plane because the lines by means of which these problems are solved
have their origin in the plane. But such problems that must be solved by
assuming one or more conic sections in the construction, are called solid
because for their construction it is necessary to use the surfaces of solid
figures, namely cones. There remains a third kind that is called line-like. For
in their construction other lines than the ones just mentioned are assumed,
having an inconstant and changeable origin, such as spirals, and the curves
that the Greeks call tetragonizousas [square-making], and which we call
quandrantes, and conchoids, and cissoids, which have many amazing
properties (Pappus 1588, III, 7; translation from Bos 2001, 38).
A few examples will help clarify what is at stake here. The problem of
bisecting a given angle is counted among planar problems, because, as
detailed by Euclid in Elements I.9, to construct the line segment that divides
a given angle into two equal parts, we construct (by compass) three circles
of equal radius, and then (by straightedge) join the vertex of the angle with
the point at which the circles intersect (Euclid 1956, Volume I, 264265).
Notice here that, to generate the solution, curves are used to construct a
point that gives the solution to the problem: by constructing the circles, we
identify a point that allows us to bisect the curve. (When dealing with locus
problems, such as the Pappus problem, the curves that are constructed are
themselves the solution to the problem. See section 3 below.) The problem
of trisecting an angle, on the other hand, was considered a line-like problem,
because its solution required the construction of curves, such as the spiral,
which were not constructible by straightedge and compass. Perhaps most
famous among line-like problems is that of squaring the circle; for those who
deemed this problem solvable, the solution required the construction of a
curve such as the quadratrix, a curve that was proposed by the ancients in
order to solve this very problem (which is how the curve received its name).
Certainly, the generation of such curves could be described; Archimedes
famously describes the generation of the spiral in Definition 1 of his Spirals
and Pappus describes the generation of the quadratrix in Book IV of the
Collection. However, these descriptions were considered more
complicated precisely because they go beyond the intersection of curves
that are generated by straightedge and compass construction. For instance,
according to Archimedes, the spiral is generated by uniformly moving a line
segment around a given point while tracing the path of a point that itself
moves uniformly along the line segment. And, according to Pappus, the
quadratrix is generated by the uniform motions of two line segments, where
one segment moves around the center of a given circle and the other moves
through a quadrant of the circle. (Cf. Bos 2001, 4042 for the details of both
these constructions.) In a similar vein, the construction of conics was
considered more complicated: One of the accepted techniques for
constructing a conic required cutting a cone in a specified way, which again,
went beyond the consideration of intersecting curves that were
constructible by straightedge and compass.
In the Collection, Pappus does not offer a firm verdict on whether the conics
and more complicated curves meet the rigorous standards of geometrical
construction and hence, on whether they are admissible in the domain of
geometry. In the case of the conics, he relies on Apollonius's commentary
and reports the usefulness of these curves for the synthesis (or proofs) of
some problems (Pappus, 116). However, to claim a curve useful is quite
different from claiming it can be constructed by properly geometrical
methods (as we'll see more clearly below). Moreover, in the case of the
quadratrix, Pappus sets out the description of the curve in Book IV of the
Collection, and then immediately proceeds to identify the common
objections to the curve's description, e.g., that there is a petitio principii in
the very definition of the curve, without commenting on whether these
objections can be overcome. Thus, although it was known by the ancients
that conics and other complicated curves could be used to solve outstanding
problems, it was not clear to early modern mathematicians whether the
ancients considered these solutions genuinely geometrical. In other words, it
was not clear from Pappus's Collection whether these curves were
admissible in geometrical problem-solving and therefore, whether solid
problems (such as identifying the mean proportionals between given line
segments) or line-like problems (such as trisecting an angle and squaring
the circle) had genuine geometrical solutions.
Now analysis is the path from what one is seeking, as if it were established,
by way of its consequences, to something that is established by synthesis.
That is to say, in analysis we assume what is sought as if it has been
achieved, and look for the thing from which it follows, and again what comes
before that, until by regressing in this way we come upon some one of the
things that are already known, or that occupy the rank of a first principle.
We call this kind of method analysis, as if to say anapalin lysis (reduction
backward). In synthesis, by reversal, we assume what was obtained last in
the analysis to have been achieved already, and, setting now in natural
order, as precedents, what before were following, and fitting them to each
other, we attain the end of the construction of what was sought. This is what
we call synthesis (Pappus, 8283).
Prior to the end of the sixteenth century, mathematicians had already used
algebra in the analysis of geometrical problems, but the program Vite
details marks a significant step forward. On the one hand, in his Isagoge
[Introduction to the analytic art] of 1591, which was presented as part of a
larger project to restore ancient analysis (entitled Book of the restored
mathematical analysis or the new algebra), Vite introduces a notation that
allowed him to treat magnitudes in a general way. The literal symbols he
uses (consonants and vowels depending on whether the variable in the
equation was unknown or indeterminate, respectively) represent
magnitudes generally and do not specify whether they are arithmetical
magnitudes (numbers) or geometrical magnitudes (such as line segments or
angles). He can thus represent arithmetic operations as applied to
magnitudes in general. For instance, A + B represents the addition of two
magnitudes and does not specify whether A and B are numbers (in which
case the addition represents a process of counting) or geometrical objects
(in which case the addition represents the combination of two line
segments) (see Vite 1591, 1127; for the significance of Vite's new
algebra for early modern mathematics see Bos 2001, Chp. 8; Mahoney
1973, Chp. 2; and Pycior 1997, Chp. 1).
x
y
y
b
x2 = ay and
y2 = xb.
Solving (1) for y, we have y = x2/a, and by substitution into (2), we get y2 =
(x2/a)2 = x4/a2 = xb, which yields:
x3 = a2b.
Solving (2) for x, we have x = y2/b, and by substitution into (1), we get x2 =
(y2/b)2 = y4/b2 = ay, which yields:
y3 = ab2.
The most famous letter written to Beeckman in 1619 dates from 26 March of
that year. In this letter Descartes announces his plan to expound an
entirely new science [scientia penitus nova], by which all problems that can
be posed, concerning any kind of quantity, continuous or discrete, can be
generally solved (AT X, 156). As he elaborates on how this new science will
proceed, Descartes clarifies that his solutions to the problems of discrete
and continuous quantitiesthat is, of arithmetic and geometry, respectively
will vary depending on the nature of the problem at hand. As he puts it,
[In this new science] each problem will be solved according to its own nature
as for example, in arithmetic some questions are resolved by rational
numbers, others only by surd [irrational] numbers, and others finally can be
imagined but not solved. So also I hope to show for continuous quantities
that some problems can be solved by straight lines and circles alone; others
only by other curved lines, which, however, result from a single motion and
can therefore be drawn with new types of compasses, which are no less
exact and geometrical, I think, than the common ones used to draw circles;
and finally others that can be solved by curved lines generated by diverse
motions not subordinated to one another, which curves are certainly only
imaginary such as the rather well-known quadratrix. I cannot imagine
anything that could not be solved by such lines at least, though I hope to
show which questions can be solved in this or that way and not any other,
so that almost nothing will remain to be found in geometry. It is, of course,
an infinite task, not for one man only. Incredibly ambitious; but I have seen
some light through the dark chaos of the science, by the help of which I
think all the thickest darkness can be dispelled (AT X, 156158; CSMK 23;
translation from Sasaki 2003, 102).
Figure 1
Figure 1: As described in the text, four rulers (OA, OB, OC, OD),
which are hinged at point O with OA horizontal and OD at about 60 degrees,
the others between in order. Four rods of equal length a (HJ, FJ, GI, EI) are
attached to the rulers such that E, F, G, and H are a distance a from O on
rulers (OA, OB, OC, OD) respectively.Figure 2
Figure 2: Same as figure 1 except a curve, KLM, starting at K on ruler OA,
intersecting J on ruler OC and ending at M beyond ruler OD.
For instance, to solve the angular section problem, Descartes begins by
presenting an instrument that includes four rulers (OA, OB, OC, OD), which
are hinged at point O (figure 1). We then take four rods (HJ, FJ, GI, EI), which
are of equal length a, and attach them to the arms of the instrument such
that they are a distance a from O and are pair-wise hinged at points J and I.
Leaving OA stationary, we now move OD so as to vary the measure of angle
DOA, and following the path of point J, we generate the curve KLM (figure 2).
As Descartes has it, we can construct the curve KLM on any given angle by
appeal to the instrument described above, because the angle we are
trisecting plays no role in the construction of KLM. And once the curve KLM
is constructed, the given angle can be trisected by means of some basic
constructions with straight lines and circles. In this respect, the curve KLM
is, for Descartes, the means for solving the angle trisection problem, and
moreover, his treatment suggests that the construction can be generalized
further so that, by means of his new compass, an angle can also be
Figure 3: Mesolabe
Figure 3: demonstration of a mesolabe. The mesolabe consists of a set of
rulers. rulers YX and YZ are hinged at Y to form an adjustable angle XYZ.
Ruler BC is perpendicular to YX and fixed at point B, C is on YZ and movable.
Ruler CD is perpendicular to YZ and intersects YX at movable point D. DE is
perpendicular to YX and intersect YZ at movable point E.
EF is
perpendicular to YZ and intersects YX at movable point F.
FG is
perpendicular to YX and intersects YZ at movable point G.
GH is
perpendicular to YZ and intersects YX at movable point H. Points D, F, and H
trace out dotted lines. As the XYZ angle closes, all three points converge on
A.
While some of the results from this period are connected with the problemsolving program presented in the 1637 La Gomtrie, Rabouin (2010) points
out that it is still not clear whether Descartes discovered his methods for
solution using the techniques that are applied in 1637 (Rabouin 2010, 456).
As such, Rabouin urges us to resist the somewhat standard reading of
Descartes' early mathematical works according to which there is a linear
and teleological progression from the 1619 pronouncement of an entirely
new science to the groundbreaking program of La Gomtrie (a reading
found, for instance, in Sasaki 2003, especially 156176). According to
Rabouin, it is not until the early 1630s, when Descartes engages with the
Pappus problemwhat Bos also considers the crucial catalyst of
Descartes' mature mathematical researches (Bos 2001, 283)that he
returns to his 1619 project to craft a new science of geometry that is
grounded on a new classification of curves and problems. Following Rabouin,
it is at this point of his mathematical career that Descartes more clearly
sees just how crucial the interplay of algebraic equations and geometry
could be for a general program of geometrical problem-solving.
3. La Gomtrie (1637)
In late 1631, the Dutch mathematician Golius urged Descartes to consider
the solution to the Pappus problem. Unlike the geometrical problems that
occupied Descartes' early researches, the Pappus problem is a locus
problem, i.e., a problem whose solution requires constructing a curvethe
Pappus curve according to Bos's terminologythat includes all the points
that satisfy the relationship stated in the problem. Generally speaking, the
Pappus Problem begins with a given number of lines, a given number of
angles, a given ratio, and a given segment, and the task is identify a curve
such that all the points on the curve satisfy a specified relation to the given
ratio. For instance, in the most basic two-line Pappus Problem (figure 4), we
are given two lines (L1, L2), two angles (1, 2), and a ratio . We designate
d1 to be the oblique distance between a point P and L1 such that P creates
1 with L1, and we designate d2 to be the oblique distance between a point
P in the plane and L2 such that P creates 2 with L2. The problem is to find
all points P such that d1 : d2 = . In this case, all the sought after points P
will lie along two straight lines, one line to the right of L1 and the other to
the left of L1. (See figure 5 for Bos's presentation of the general problem.)
In the Collection, Pappus presents a solution to the three and four line
versions of the problem (i.e., the versions of the problem in which we begin
with three or four given lines and angles) as well as Apollonius's solution to
the six-line case, which relies on his theory of conics and the transformation
of areas to construct the locus of points (Pappus, 118123). However,
Pappus does not treat the general (n-line) case, and this is the advance of
the solution Descartes achieves in 1632, a solution published in La
Gomtrie, where he claims that, unlike the ancients, he has found a
method to successfully determine, describe, [and] explain the nature of the
line required when the question [of the Pappus Problem] involves a greater
number of lines (G, 22). And as Descartes reports to Mersenne in 1632, he
could not have found his general solution without the help of algebra:
I must admit that I took five or six weeks to find the solution [to the Pappus
Problem]; and if anyone else discovers it, I will not believe that he is
ignorant of algebra (To Mersenne 5 April 1632; AT I, 244; CSMK, 37).
Figure 5: The General Pappus Problem (from Bos 2001,Fig. 19.1, 273)
Problem: Find the locus of points P such that the following ratios are equal to
the given ratio :
: ad4d5
: d4d5d6
In general,
: dk+1d2k
: adk+2d2k+1
If, then, we wish to solve any problem, we first suppose the solution already
effected, and give names to all the lines that seem needful for its
construction,to those that are unknown as well as to those that are known.
Then, making no distinction between unknown and unknown lines, we must
unravel the difficulty in any way that shows most naturally the relations
between these lines, until we find it possible to express a single quantity in
two ways. This will constitute an equation, since the terms of one of these
two expressions are together equal to the terms of the other (G, 69).
Given a line segment AB containing point C (see figure 6), the problem is to
produce AB to D such that the product AD*DB is equal to the square of CD.
Let AC = a, CB = b, and BD = x, which yields AD = a + b + x and CD = b +
x. Thus, the problem to find BD such that AD*DB = (CD)2 is algebraically
equivalent to finding x such that: (a + b + x)*(x) = (b + x)2. Or, solving for
x, the problem is to find x such that, given a and b, x = b2 / (ab).
Figure 6:
A horizontal line with points A, C, B, and D marked on it in that order.
Distance a is from point A to C. Distance b from point C to B, distance x
from B to D.
Three solid, straight lines, FES which goes from the lower left to mid top,
EABG which goes from left to right, and TGH which goes from upper left to
mid right. Point C is below all three lines and from it dotted, straight lines,
CBRST (which intersects all three solid lines at points B, S, and T), CF, CH,
and CD which is a dotted line to point D to the left and down of C. There is
also a dotted, straight line DAR.
Descartes begins with consideration of the problem when we are given three
or four lines, which, borrowing from Guicciardini (2009), can be stated as
follows (see figure 7):
Having three or four lines given in position, it is required to find the locus of
points C from which drawing three or four lines to the three or four lines
given in position and making given angles with each one of the given lines
the following condition holds: the rectangle [or product] of two of the three
lines so drawn shall bear a given ratio to the square of the third (if there be
only three), or to the rectangle [or product] of the other two (if there be
four) (Guicciardini 2009, 54; based on G, 22).
In Book One, Descartes applies his geometrical analysis to the four-line case
of the Pappus problem. He begins by designating two given line segments
(of unknown length) AB and BC as oblique coordinates x and y, respectively,
such that all other lines needed to solve the problem will be expressed in
terms of x and y.[6] Then, by considering the angles given in the problem
and the properties of similar triangles, he generates an algebraic expression
of the sought after points C in terms of the two unknowns x and y and the
known quantity z (where z designates the ratio given in the problem) (G, 29
30).
Importantly, the analytic method that Descartes uses in the four-line case is
generalized to apply to the general, n-line version of the Pappus Problem.
That is, Descartes' claim is that no matter how many lines and angles are
given in the problem, it is possible, by means of his analytic method, to
express the sought after points C in terms of two unknown quantities (in
contemporary terms, to reduce the problem to a polynomial equation in two
unknowns) (G, 33). As a result, for any n-line version of the Pappus Problem,
we can generate values for C and construct the sought after Pappus curve in
a pointwise manner by assigning different values to x and y. As Descartes
puts it,
find the value of the other from this equation. It is obvious that when not
more than five lines are given, the quantity x, which is not used to express
the first of the lines can never be of degree higher than the second.
The ancients were familiar with the fact that the problems of geometry may
be divided into three classes, namely, plane, solid, and linear problems. This
is equivalent to saying that some problems require only circles and straight
lines for their construction, while others require a conic section and still
others more complex curves. I am surprised, however, that they did not go
further, and distinguish between different degrees of those more complex
curves, nor do I see why they called the latter mechanical, rather than
geometrical. If we say that they are called mechanical because some sort of
instrument has to be used to describe them, then we must, to be consistent,
reject circles and straight lines, since these cannot be described on paper
without the use of compasses and a ruler, which may also be termed
instruments. It is not because the other instruments, being more
complicated than the ruler and compass, are therefore less accurate, for if
this were so they would have to be excluded from mechanics, in which
accuracy of construction is even more important than in geometry. In the
latter, exactness of reasoning alone is sought, and this can surely be as
thorough with reference to such lines as to simpler ones (G, 4044).
It is true that the conic sections were never freely received into ancient
geometry, and I do not care to undertake to change names confirmed by
usage; nevertheless, it seems very clear to me that if we make the usual
assumption that geometry is precise and exact, while mechanics is not; and
if we think of geometry as the science which furnishes a general knowledge
of the measurement of all bodies, then we have no more right to exclude
the more complex curves than the simpler ones, provided they can be
conceived of as described by a continuous motion or by several successive
motions, each motion being completely determined by those which precede;
for in this way an exact knowledge of the magnitude of each is always
obtainable (G, 43).
Consider the lines AB, AD, AF, and so forth, which we may suppose to be
described by means of the instrument YZ [Figure 8]. This instrument
consists of several rulers hinged together in such a way that YZ being
placed along the line AN the angle XYZ can be increased or decreased in
size, and when its sides are together, the points B, C, D, E, F, G, H, all
coincide with A; but as the size of the angle is increased, the ruler BC,
fastened at right angles to XY at the point B, pushed toward Z the ruler CD
which slides along YZ always at right angles. In a like manner, CD pushes DE
which slides along YX always parallel to BC; DE pushes EF; EF pushes FG; FG
pushes GH, and so on. Thus we may imagine an infinity of rulers, each
pushing another, half of them making equal angles with YX and the rest with
YZ.
Now as the angle XYZ is increased, the point B describes the curve AB,
which is a circle; while the intersections of the other rulers, namely, the
points D, F, H describe the other curves, AD, AF, AH, of which the latter are
more complex than the first and this more complex than the circle.
Nevertheless I see no reason why the description of the first cannot be
conceived as clearly and distinctly as that of the circle, or at least as that of
the conic sections; or why that of the second, third, or any other that can be
thus described, cannot be as clearly conceived of as the first: and therefore I
see no reason why they should not be used in the same way in the solution
of geometric problems (G, 4447).[7]
Figure 8: Mesolabe
A mesolabe same as Figure 3.
A couple points are worth emphasizing. First, Descartes presents the more
complex curves generated by his compass as described by motions that can
be as conceived as clearly and distinctly as the motions required to
construct the more simple circle. And because of the clear and distinct
motions needed for their construction, these curves are legitimately
geometrical. That is, consistent with Descartes' general criterion for
constructing geometrical curves, these complex curves can be used in the
solution of geometric problems. Second, we see that although Descartes
takes care to distinguish the concerns of geometry from those of mechanics,
he does not steer away from the construction of curves by means of
instruments. Although instrumental constructions are mechanical
constructions, they can nonetheless give rise to geometrical curves
precisely because the motions of the instruments are clearly and distinctly
conceived. That the motions are generated by instruments does not render
the resultant curve non-geometrical. (For more on the use of instruments in
La Gomtrie, see Bos 1981.)
Descartes explicitly names the spiral and quadratrix as those curves whose
construction must be conceived of as described by two separate
movements whose relation does not admit of exact determination. Later in
Book Two he clarifies why such descriptions fail to be clearly and distinctly
conceived:
geometry should not include lines that are like strings, in that they are
sometimes straight and sometimes curved, since the ratios between
straight and curved lines are not known, and I believe cannot be discovered
by human minds, and therefore no conclusion based upon such ratios can
be accepted as rigorous and exact (G, 91).
Given these remarks, the fundamental problem with the spiral, the
quadratrix, and lines that are like strings is that their construction requires
consideration of the ratio, or relation, between a circle and straight line.
Consider the spiral. As we saw above, its construction involves two uniform
motions: the uniform rectilinear motion of a point along a segment and the
uniform circular motion of the segment around a point. These two motions
must simultaneously be considered in order for the moving point's path to
describe the spiral, and this, for Descartes, is what is ultimately problematic.
The human mind can think about simultaneous rectilinear and circular
motions, but it cannot do so with the clarity and distinctness required to
meet the exact and rigorous standards of geometry. (This claim is not
without its problems, which will be discussed in section 3.3 below. For a
comparison between Descartes' criterion for the construction of geometrical
curves and the views put forward by Pascal, see Jesseph 2007.)
I could give here several other ways of tracing and conceiving a series of
curved lines, each curve more complex than any preceding one, but I think
the best way to group together all such curves and then classify them in
order, is by recognizing the fact that all the points of those curves which we
may call geometric, that is, those which admit of precise and exact
measurement, must bear a definite relation to all points of a straight line,
and that this relation must be expressed means of a single equation (G, 48).
The same point is made later in Book Two, where Descartes emphasizes that
no matter how we conceive a curve to be described, provided it be one of
those which I have called geometric, it will always be possible to find an
equation determining all of the curve's points (G, 56). He reiterates that
geometric curves can be classified according to their equations but also
points out that within a specific class, a curves' simplicity should be ranked
according to the motions required for construction. For instance, although
the circle belongs to the same class as the ellipse, hyperbola, and parabola,
these latter curves are equally complex whereas the circle is evidently a
simpler curve and will thus be more useful in the construction of problems
(G, 56).
As in Book One, Descartes uses the Pappus Problem to illustrate the power
of his geometrical calculus, where in Book Two, his aim is to show how his
algebraic classification of curves makes it easy to demonstrate the solution
which [he has] already given of the problem of Pappus (G, 59). The specific
goal here is to establish that the curves which solve the general Pappus
Problem are legitimately geometrical curves, i.e., to show that the Pappus
curves meet the exact and rigorous standards of geometrical construction
that he has just laid out. Descartes' discussion of the Pappus Problem in
Book Two begins as follows:
or four lines the equation which serves to determine the required points is of
the second degree. It follows that the curve containing these points [i.e., the
Pappus curve] must belong to the first class, since such an equation
expresses the relation between all points of curves of Class I and all points
of a fixed straight line. When there are not more than eight given lines the
equation is at most a biquadratic, and therefore the resulting [Pappus] curve
belongs to Class II or Class I. When there are not more than twelve given
lines, the equation is of the sixth degree or lower, and therefore the required
curve belongs to Class III or a lower class, and so on for other cases (G, 59).
I do not like to have to speak well of myself, but because there are few
people who are able to understand my Geometry, and since you will want
me to tell you what my own view of it is, I think it appropriate that I should
tell you that it is such that I could not wish to improve it. In the Optics and
the Meteorology I merely tried to show that my method is better than the
usual one; in my Geometry, however, I claim to have demonstrated this.
Right at the beginning I solve a problem which according to the testimony of
Pappus none of the ancients managed to solve; and it can be said that none
of the moderns has been able to solve it either, since none of them has
written about it, even though the cleverest of them have tried to solve the
other problems which Pappus mentions in the same place as having been
tackled by the ancients (To Mersenne, end of December 1637; AT 1, 478;
CSMK, 7778).
Recall that in addition to his emphasis on the precise and exact motions
that can be used to describe legitimately geometrical curves, Descartes also
claims that these curves can be conceived of as described by a continuous
motion or by several successive motions. As such, we would reasonably
expect that the geometrical construction of these curves should not proceed
pointwise in the manner of Book One, where Descartes constructed the
Pappus curves by solving the equations to which the problem had been
reduced. However, when Descartes treats the five-line Pappus Problem in
Book Two, he in fact offers a pointwise construction of the Pappus curve. He
then remarks that the pointwise construction of this geometric Pappus
curve is importantly different from the pointwise construction of nongeometrical, mechanical curves:
Figure 9
A vertical line AC and horizontal line AGB (going to the right). Three
hozontal parallel lines above AGB, (not labelled but consider them L1, L2, L3
from top to bottom). Three lines AE, AD, AF radiate from A between AC and
AGB. AE intersects L1 at point p1, AD with L2 at point p2, AF with L3 at
point p3 . Curve CG connects p1, p2, p3. Curve CB intersects L1 where AD
also intersects L1.
You ask me if I think that a sphere which rotates on a plane describes a line
equal to its circumference, to which I simply reply yes, according to one of
the maxims I have written down, that is that whatever we conceive clearly
and distinctly is true. For I conceive quite well that the same line can be
sometimes straight and sometimes curved, like a string (To Mersenne, 27
May 1638; AT 2, 140141; translation from Mancosu 2007, 118).
Whether Descartes had the hidden agenda that Mancosu suggests, the
explicit claims used to define the program of problem-solving presented in
La Gomtrie point to the limitations of Descartes' mathematics. As we have
seen above, Descartes' primary focus is on a standard for geometry's
exactness of reasoning that is bound to clear and distinct motions for
construction and to the finite equations to represent curves so constructed.
Consequently, there is no room in the program of La Gomtrie to use
infinitesimals in the construction of curves or to treat curves represented by
infinite equations. As such, Descartes eliminated from his geometrical
program the very elements of mathematical and geometrical reasoning that
made it possible for Newton and Leibniz to develop the calculus come the
late seventeenth century. Nonetheless, given how quickly Descartes honed
his mathematical skills and how quickly he developed his innovative
program for geometry, it would not be too risky to accept Descartes' selfassessment and maintain some confidence that the calculus would have
been in his reach had he considered the infinitesimal and the infinite:
Bibliography
Beeckman, Issac, 16041634, Journal tenu par Isaac Beeckman de 1604 a
1634, 4 volumes, introduction and notes by C. de Waard. La Haye: M.
Nijhoff, 19391953.
Bos, Henk J.M., 1981, On the representation of curves in Descartes'
Gomtrie, Archive for History of Exact Sciences 24: 295338.
, 2001, Redefining Geometrical Exactness: Descartes' Transformation of
the Early Modern Concept of Construction. New York, Berlin, Heidelberg:
Springer-Verlag.
Descartes, Ren, 1637, The Geometry of Rene Descartes with a facsimilie of
the first edition, translated by David E. Smith and Marcia L. Latham. New
York: Dover Publications, Inc., 1954. [cited as G followed by page number]
, 1985, The philosophical writings of Descartes. Two volumes, translated
by John Cottingham, Robert Stoothoff, and Dugald Murdoch. Cambridge:
Cambridge University Press. [cited as CSM followed by volume and page
number]
, 1987, Exercices pour les lments des solides: Essai en complment
d'Euclide: Progymnasmata de solidorum elementis. Edition critique avec
introduction, traduction, notes et commentaries par Pierre Coatabel. Paris:
PUF.