Seventeenth Century Instruments for Drawing Conic Sections

Author(s): Jan Van Maanen

Source: The Mathematical Gazette, Vol. 76, No. 476 (Jul., 1992), pp. 222-230
Published by: Mathematical Association
Stable URL: http://www.jstor.org/stable/3619131
Accessed: 04-11-2015 09:38 UTC

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact [email protected].

Mathematical Association is collaborating with JSTOR to digitize, preserve and extend access to The Mathematical Gazette.

http://www.jstor.org

This content downloaded from 128.248.155.225 on Wed, 04 Nov 2015 09:38:35 UTC
All use subject to JSTOR Terms and Conditions
222 THEMATHEMATICAL
GAZE'I'E

Seventeenth century instruments for drawing conic

sections

JAN VAN MAANEN

Introduction
One of the consequences of Descartes' new approachto geometry
(1637) was an increased interest in instruments for drawing conic
sections. Conic section drawers,which is the name I shall give to instru-
ments for drawing conic sections in a continuous movement, had been
describedearlier,but the centralrole of conic sections in the geometry of
Descartes made them by about 1640 into a new topic of research. This
article sketches the background to this growing interest and then
describes some of the instrumentsdesigned by the Dutch mathematician
Van Schooten (1615/6 - 1660). It concludes with an applicationof this
materialin mathematicsteaching.
Conic section drawers revived in mathematicsin the course of the
17th century in connection with the new method, introduced by
Descartes in 1637, of solving geometric problems with the help of
algebra. Construction problems were fundamental to mathematics in
Greek antiquity.The execution of constructionswas governed by strict
rules: one should use only ruler and compass and should perform only
the actions which Euclid had declared practicablein postulates 1 to 3 of
his Elements.
c

A B*

FIGURE1.

An example of such a constructionproblem,belonging to the ancient

traditionof the "division of figures", is the following (see figure 1).
Given: a triangleABC. Required:to constructa line throughpoint P
on AB so that the triangleis divided into two partsof equal area.
A possible solution along pre-Cartesianlines goes like this. Draw
PC and draw a line through A, parallel to PC. The latter meets BC
extended at a point which is called D. Draw PD. Since PC and AD are

This content downloaded from 128.248.155.225 on Wed, 04 Nov 2015 09:38:35 UTC
All use subject to JSTOR Terms and Conditions
 CENTURYCONICSECTIONDRAWERS
SEVENTEENTH 223

parallel,area(AAPC) = area(ADPC), so area(ADPB)= area(AABC). Let

E be the mid-point of line segment BD. Since the medians of a triangle
divide it into two triangles of equal area, PE divides ADPB into two
parts of equal area, so area(APBE) = /? area(ADPB) = ?/2area(A ABC).
The conclusion is thatPE is the line that had to be constructed. In figure
2 this constructionhas been executed by rulerand compass.

A(^[^rf^ B

FIGURE2.

Clearly the crucial point is to find out which constructionsteps are

needed to solve the problem(the analysis stage). When these steps have
been discovered the proof of the correctness of the construction (the
synthesis stage) is usually easy.
What Descartes did was to show that algebra could be used to
analyse geometricalconstructionproblems. His analysis would have run
like this (the following is more or less a paraphraseof a passage from
Descartes Geometrie). Start by making a sketch in which you imagine
that the problem has already been solved. Represent the lengths of the
line-segments which appearin the figureby letters, the known lengths by
letters from the beginning of the alphabet,the unknown ones by letters
from the end of the alphabet.Try to express algebraicallyin two different
ways one of the quantitiesthat appearsin the figure, and put these two
expressions equal to each other. This gives you an equation, and the
solution of this equation relates the length of the unknown line-segment
(which has to be constructed) to the given lengths. The geometrical
interpretationof this relationyields the construction.
In our example (see figure 3) we make a drawing in which we
suppose thatthe line PU solves the problem. In the figure the lengths of
AB, AC and BC (c, b and a respectively) and the distances AP and PB

This content downloaded from 128.248.155.225 on Wed, 04 Nov 2015 09:38:35 UTC
All use subject to JSTOR Terms and Conditions
224 THEMATHEMATICAL
GAZE'ITE

(here called c-d and d) are known. Unknown are the lengths of BU and
CU, here called x and a-x, and the height h of APBU, which however
can be expressed directly: h = xsinfl. The subsequent steps speak for
themselves: area(AABC) = ?2acsinf, and area(APBU) = ?2dxsinf,. Since
C

A B

FIGURE3.

this latterarea must be half of the first area we have dx = I/2ac,which is

the equationrelatingthe unknownx to the given lengths a, c and d. It is
on the basis of this equationthatx is to be constructed;this can be done
using the equivalent proportionalityx : c = V?a : d, which can be
constructedwith two similartrianglesas in figure4.

FIGURE4.

First construct(to producea length /2a) the mid-pointM of BC, and

draw PM. In APBM the lengths ?/2aand d occur; c is the length of AB.
So one has to construct a triangle which has a side AB and which is
similar to A PBM. This is easily done by drawing through A a line
parallel to PM, which intersectsBC in U. Then AABU and APBM are

This content downloaded from 128.248.155.225 on Wed, 04 Nov 2015 09:38:35 UTC
All use subject to JSTOR Terms and Conditions
SEVENTEENTH
CENTURYCONICSECTIONDRAWERS 225

similar, so BU: AB = BM: PB, thatis BU: c = 1?2a d. We conclude that

BU satisfies the proportionalityx: c = 12?a: d, which implies that PU is
the requiredline throughP which divides AABC into two parts of equal
area.
For Descartes this procedure was a new and powerful tool for
investigating whether a construction could be executed by ruler and
compass. For, if the analysis leads to an equation of the first or second
degree, and if a real solution exists, then a rulerand compass solution is
possible (a fact proved explicitly by Descartes). If the resultingequation
is of third or fourth degree, there might be a ruler and compass con-
struction,but it will exist only in special cases. In general, said Descartes,
one will have to use other means to constructa solution. Referring to
the ancient Greek notion of "solid" problem which could be constructed
with the help of conic sections, he proved that, if x satisfies an equation
of thirdor fourthdegree, x can be constructedgeometricallyif one allows
not only circles and straightlines as constructioncurves but also conic
sections.
This approach, however, caused a serious problem. Circles and
straightlines can be drawn in a continuous movement (at least that was
postulatedby Euclid: the straightrulerexists only as a mental concept of
course), but can conic sections be drawn continuously ? They can be
constructedpoint-by-point,but if one wants to intersect a conic section
with other constructioncurves, a point by point construction does not
suffice. So the admissionof conic sections as constructioncurves is prac-
ticable only if there are instrumentsthat can draw them in a continuous
movement. And this explains the growing interest in conic section
drawersaroundthe middle of the 17th century.

The conic section drawersof Van Schooten

In the GeometrieDescartes did not say anythingabout conic section
drawers. Frans van Schooten (1615/6-1660), the Leiden professor of
mathematics,who filled in so many gaps in the Geometrieand who was
the great propagatorof Cartesian mathematics, studied conic section
drawersto make up for this deficiency of Descartes. The magnum opus
of van Schooten was the Latin edition of Descartes' Geometrie (1649,
second enlarged edition 1659-61), which contained, in addition to the
Latin text, commentaries by van Schooten and the results of invest-
igations by studentsof van Schooten into the power of the new analytic
geometry. Van Schooten's first publicationof his own creative work was
his treatise on conic section drawersDe organica conicarum sectionum
in plano descriptione, tractatus, (Leiden 1646, reprinted in Exercit-
ationum Mathematicarumlibri quinque, Leiden 1656/7). Figures 5, 6
and 7 are takenfrom the treatise. They show threeconic section drawers.

This content downloaded from 128.248.155.225 on Wed, 04 Nov 2015 09:38:35 UTC
All use subject to JSTOR Terms and Conditions
226 THE MATHEMATICALGAZEI'IE

The instruments are depicted as seen from above, lying on a

horizontal plane on a sheet of paper. The instrument in figure 5 is
operated by moving the ruler GI along the ruler EG, perpendicularto it.
The rhombusBFGH (BF = FG = GH = HB) has hinges in B, F, G and H
and is fixed to the plane by a pin in B, and to rulerGI by a pin in G. At F
a slotted ruler is attachedto the rhombus. A pin H goes throughthe slot,
so that the slotted ruleralways forms the diagonal of BFGH. In D, where
the rulers FH and GI (which is also slotted) intersect, a style, passing
through the two slots, draws a curve. Triangles BHD and GHD are
congruent, for they have HD as a common side; and furthermoreBH =
HG since BFGH is a rhombus) and angle BHD = angle GHD (for in the
rhombus the diagonal FH bisects the angle H both internally and
externally). From the congruence it follows thatBD = GH, so D is at the
same distance from B as from the ruler EG, and the curve drawn by the
style is a parabola.

C.

FIGURE 5. Tractatusp. 74.

In figure 6, A is a point on the ruler KL. AB and BD are of equal

length. AB rotates aroundA, and BD is hinged at B. The instrumentis

This content downloaded from 128.248.155.225 on Wed, 04 Nov 2015 09:38:35 UTC
All use subject to JSTOR Terms and Conditions
SEVENTEENTHCENTURY CONIC SECTION DRAWERS 227

operated by moving D along the ruler KL. At the fixed point E a style is
placed, which draws an ellipse.

FIGURE6. Tractatusp. 26.

The functioning of the third instrument (figure 7) may now be clear.
One should know that CD = GF and DG = CF. The style, held by the
hand, draws a hyperbola, of which the vertices (E and K; CD is made
equal to EK) and the distance between the foci (CF) are given.

FIGURE7. Tractatusp. 57.

This content downloaded from 128.248.155.225 on Wed, 04 Nov 2015 09:38:35 UTC
All use subject to JSTOR Terms and Conditions
228 THEMATHEMATICAL
GAZEII'E

The proof that the second and third instruments produce conic
sections is left to the reader (a well-known Cartesianphrase!). For the
second instrumenta possible approachmay be derived from the problem
that is quoted below. It was set in a leaving certificateexaminationof the
grammarschool where I teach. [In the Dutch system half of the final
markis determinedby a test set by a governmentcommittee, which is the
same for all schools of the same type; the other half of the markdepends
on tests that are set by the schools themselves]. The subject was "conic
sections", but at an earlier stage the pupils had already studied plane
curves, given by a one-parameterrepresentation,and subjects from the
historyof mathematics. The problemis as follows:

For a long time mathematicianstreated conic sections with a

certain suspicion since there were no instrumentsby which you
could draw them as accuratelyas, for example, a straightline with
the use of a ruler,or a circle with the use of a pair of compasses.
In the 17th century the Leiden professor of mathematicsFrans
van Schooten (1615/6-1660) studied this problem. As a result he
describes in his book Exercitationum Mathematicorum libri
quinque (Leiden 1657), among other things, some methods for
drawing an ellipse mechanically. One such method is the well-
known one with string(see figure 8).

H. . hA .thtB..BA

i... .
*"..,
.............Oo****. ?

FIGURE8.

a) Explain why the hand with the pencil in E draws an ellipse.

A second instrumentworks as follows (see figure 9). A bar AB
of fixed length a rotates aroundA; at B a second bar BE is hinged
to it, on which there is a point D which moves along the x-axis
such that BD = BA = a.

This content downloaded from 128.248.155.225 on Wed, 04 Nov 2015 09:38:35 UTC
All use subject to JSTOR Terms and Conditions
SEVENTEENTH
CENTURYCONICSECTIONDRAWERS 229

L
K'

"
0"*
?.. . *... ... .".*

FIGURE9.
The length BE is called b (a and b are positive). Van Schooten
claims that the pencil in E describes an ellipse if D moves along
the x-axis.
b) Let A be the origin. If we take it for grantedthat the set of
points E is indeed an ellipse, what is the equationof thatellipse ?

I . . I x
A D

FIGURE 10.

Finally we shall prove thatthe set of points E really is an ellipse.

Again let A be the origin and let E be (x,y) as in figure 10, so for
convenience take x > 0 and y > 0. As above AB = BD = a and let
DE = b. Every direction of line BE with angle 0 determines a
point E(x,y).

This content downloaded from 128.248.155.225 on Wed, 04 Nov 2015 09:38:35 UTC
All use subject to JSTOR Terms and Conditions
230 THE MATHEMATICALGAZLETIE

c) Express the coordinates of B, D and E in a, b, and 4)

successively.
d) Derive from c) an equationthat is satisfiedby the coordinates
of the point E; formulatea conclusion.
e) Would this instrumentof van Schooten's be as suitable for
drawingconic sections as the compass is for drawingcircles ?

Not an easy problem,the candidatessaid, but a nice one !

JAN VAN MAANEN

University of Groningen and University of Utrecht, Mathematical

Institute,Budapestlaan6, P.O. Box 80.010, 3508 TA Utrecht

So who is perfectly useless ?

Recently we read in this newspaper how two "remarkablenew numbers" had been
discovered by Atomic Energy Authority scientists using a super computer at the Harwell
laboratory. The first was the biggest prime number,227,832 digits long, nearly enough to
fill six pages of The Daily Telegraph. The other, which is a new "perfect number"was
455,633 digits long, able to occupy more than 12 pages of The Daily Telegraph. People
complain aboutthe City prices and LeicesterUniversityclass lists, but if we startedprinting
all these digits even our most faithfulreaderswould complain.
I wonder how much money we taxpayers paid for these completely useless pieces of
information. I myself, recently playing with a calculator discovered the only number
which, like the numberone, is its own squareroot but which, unlike the number 1 is not its
own square. The answer is 0.9 recurring,but I do not expect to be awardedgreat sums of
public money for this discovery, which I freely give to the world; nor do I demand the
overthrowof governmentand its takeoverby a gang of lower class hooligans.
AuberonWaugh in The Daily Telegraph,April 4 1992, sent in by George Duckworth. The
Editor thanks the son of the authorof Decline and Fall for a nearly equally amusing piece
which inadvertantlycontains some good discussion points when introducingthe binomial
theorem.Why does the calculatorsay that 0.99999992 is 0.9999998, but that 40.9999999 is
0.9999999 ? Does 1.0000001 have a similar property? What does "0.9 recurring"mean
anyway ? The Timesdid a bit betterwhen reportingthe new primenumber,but still doesn't
get things quite right:

Scientists at Harwell have discovered the largest known prime number ... prime numbers
are believed to go on for ever, a propositionoriginallymade by Euclid.
From The Times,26 March 1992 sent in by John Deft.

An AuberonWaughInteger
Nigel Backhouse observes that 10'451 + 7496 is 139.9999970 : certainly an integer on
AuberonWaugh's calculator.

This content downloaded from 128.248.155.225 on Wed, 04 Nov 2015 09:38:35 UTC
All use subject to JSTOR Terms and Conditions