(Compact Textbooks in Mathematics) Ad Meskens, Paul Tytgat (Auth.) - Exploring Classical Greek Construction Problems With Interactive Geometry Software-Birkhäuser Basel (2017) PDF

Compact Textbooks in Mathematics
Ad Meskens
Paul Tytgat
Exploring
Classical Greek
Construction Problems
with Interactive
Geometry Software
For further volumes:

http://www.springer.com/series/11225
This textbook series presents concise introductions to current topics in math-

ematics and mainly addresses advanced undergraduates and master students.
The concept is to offer small books covering subject matter equivalent to 2- or
3-hour lectures or seminars which are also suitable for self-study. The books pro-
vide students and teachers with new perspectives and novel approaches. They
feature examples and exercises to illustrate key concepts and applications of the
theoretical contents. The series also includes textbooks specifically speaking to
the needs of students from other disciplines such as physics, computer science,
engineering, life sciences, finance.
Ad Meskens
Paul Tytgat
Exploring
Classical Greek
Construction Problems
with Interactive
Geometry Software
Ad Meskens Paul Tytgat
Department of Education and Training Antwerpen, Belgium
Artesis Plantijn University College
Antwerpen, Belgium
ISSN 2296-4568
ISBN 978-3-319-42862-8 ISBN 978-3-319-42863-5 (eBook)
DOI 10.1007/978-3-319-42863-5
Library of Congress Control Number: 2017931078
Mathematics Subject Classification (2010): 51-01, 51-03, 01A05, 1A20, 97G40
Birkhäuser
© Springer International Publishing Switzerland 2017
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or
information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publi-
cation does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the au-
thors or the editors give a warranty, express or implied, with respect to the material contained herein
or for any errors or omissions that may have been made.
Cover design: deblik, Berlin
Printed on acid-free paper
This book is published under the trade name Birkhäuser

The registered company is Springer International Publishing AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
v
Preface
This book has been ten odd years in the making. Its first inception was a lecture
which I was asked to give at a refresher course for mathematics teachers. The text
evolved into a little booklet for the Dutch Zebra series (Epsilon publishers)1 . This
is a series of 60-page booklets aimed at high school pupils preparing for their final
exams. As the old saying goes, “Du choc des idées jaillit la lumière”, discussions
with the editors led to a text which had more potential than the allowed 60 pages.
What you have before you is a greatly expanded version of that booklet.
This book is not intended as a history of mathematics, nor is it a complete
overview of the problems at hand, the duplication of the cube, the trisection of
an angle and the squaring of a circle. Anyone looking for new historical insights
should consult other books. We only put forward our mostly educational view,
sometimes new, of constructions which have been published numerous times be-
fore. The history (and sometimes mythology) of mathematics is used as an intro-
ductory story to raise students’ interest in the problem.
On the other hand, neither is this a mathematics manual that observes a de-
ductively oriented organisation. It is a well-known fact that deductive organisation
is the last step in any given discovery process. Getting acquainted with a histor-
ical problem will show students that the pathway to a solution is not linear but
curved, sometimes even going backwards, full of mistakes due to lack of appropri-
ate mathematical language and symbolism. Indeed, if anything can be said about
the classical Greek construction problems, it is that they are a prime example of
this non-linear development. They have captured the imagination of mathemati-
cians for thousands of years, they have led them astray, they have led them into
new realms of mathematics until it turned out they were unsolvable with the re-
straints placed on them.
The following pages are excursions, at undergraduate student level, into these
three famous problems. A historian of mathematics faces a dilemma: citing and
subsequently explaining an ancient text or translating the text into modern math-
ematical terminology. The first approach alienates the modern reader, the second
betrays the ideas of the ancient author. The author of an educational exposé does
not face that dilemma. He is concerned more with heritage than with history per
se. Here you will find ancient problems written in modern terminology, making
them accessible to the mathematical community at large. By making the point
that the ancient mathematicians had fewer tools at their disposal the reader will
acquire even greater admiration for their accomplishments, possibly even finding
him/herself drawn to ancient texts and perhaps to exploring them in greater depth.
The history of mathematics is important for teachers of mathematics. It gives
them the opportunity to let their students see the motivation behind the introduction
of certain concepts, it makes clear that mathematics is a non-linear endeavour and
students get a glimpse of the people who created mathematics in all their human
1
Meskens and Tytgat (2015).
vi Preface
aspects. It also makes students aware that they often come up against the same
difficulties in seeking to grasp certain concepts as the inventors did.
We hope that many teachers will use these chapters as enrichment material in
their classes and enable their students to gain a deeper understanding of geometry.
The use of dynamic geometry software packages, enables them to explore geo-
metric relationships in an educational, enquiry-based fashion. Relationships which
most likely would have remained hidden in a classic pencil and paper approach.
By focusing on constructions and the use of Interactive Geometry Software
or IGS for short, the reader is confronted with the same problems that ancient
mathematicians once faced. The neusis construction is of particular interest to be
explored with IGS, as it lets readers discover a class of interesting curves. Read-
ers get to retrace the footsteps of Euclid, Viète and Cusanus amongst others and
then, by experimenting, discover geometric relationships that far exceed their ac-
complishments. Over 140 exercises guide readers through methods which were
developed to try and solve the problems. The exercises are at undergraduate stu-
dent level and only require a command of elementary Euclidean geometry and
pre-calculus algebra. These exercises are especially well-suited for students who
are thinking of becoming a mathematics teacher.
It may be argued that the constructions performed in IGS are not real con-
structions because IGS uses coordinate geometry. This is of course true, more or
less. Any construction, be it on paper, or using a computer program, constitutes
a particular case. IGS has the advantage that, using the Move tool, a multitude of
particular cases can be checked. While a geometric assertion may not be proved
using IGS it may be disproved, by finding a counterexample. IGS will not replace
the deductive approach, but it adds a feature to doing geometry: experimentation.
For some students, an animation showing that the medians of a triangle always
intersect in one point may be more convincing than the proof itself. For the more
mathematically-minded students, it is an invitation to search for that proof, show-
ing that what IGS suggests is indeed always true. IGS is a didactic tool of which
we are only seeing the beginnings of its possibilities as an incentive for engaging
in proper mathematics.
Writing a book like this is not possible without the help of many people and
institutions. I have the pleasure of thanking Epsilon Uitgaven and NVvW (Dutch
Association of Mathematics Teachers) for their permission to reuse parts of our
booklet in the Zebra series. I also wish to thank the editors of the Zebra series for
the useful feedback and the many valuable suggestions. Their comments made me
turn a classic text into an enquiry-based and experiment-oriented exposé. Thanks
are also due to VVWL (Flemish Association of Mathematics Teachers) for the use
of parts of articles which appeared in their journal Wiskunde en Onderwijs.
Jean Paul van Bendegem, Hiram Bollaert and Joost Bambust read through the
English manuscript and also gave many valuable suggestions. I could count on
Stephen Hargreaves for useful advice on English terminology.
I am indebted to Artesis Plantijn University College Antwerp for its support
in the form of a grant. The collegues and the staff of the department of Education
and Training have supported the writing of this book in various ways. The present
book would never have materialised without the support from various libraries,
with particular reference to the library of the Education and Training Depart-
vii
Preface
ment of Artesis Plantijn University College Antwerp, Erfgoedbibliotheek Hendrik

Conscience (Antwerp), Museum Plantin Moretus (Antwerp) and the Koninklijke
Bibliotheek Albert I (Brussels).
Rttaalkantoor corrected the English.
The longstanding collaboration with Paul Tytgat proved as invaluable as ever.
We are also grateful to Yola and David Coffeen of Tesseract – Early Scientific
Instruments for their permission to publish the photographs of the ellipsograph in
7 Sect. 2.4 (. Fig. 2.14).
Finally, I would like to thank my wife Nicole whose support, for over thirty
years, has been a continuing labour of love.
In one apocryphal story, King Ptolemy wanted Euclid to teach him mathemat-
ics, but he got bored very quickly with the theorems and proofs. He asked Euclid
whether there was a faster way of learning mathematics, to which Euclid replied
“Sire, in mathematics there is no Royal Road.” A similar story is told about Alexan-
der the Great and Menaechmus. You, reader, will not walk a regal road, but when
you have finished the book you will feel you have been given a royal gift.
Some preliminary remarks
The equivalent American style notation is given between brackets.

!
4 AB (AB) means the straight line through A and B
!
4 ŒAB (AB) means the ray with vertex A and through B
4 ŒAB (AB) means the line segment bounded by the end points A and B
4 kABk (AB) means the length of ŒAB
!
4 AB (AB or AB) means the vector represented by the line segment ŒAB and
with initial point A and terminal point B
4 !a (a) means the vector represented by the line segment ŒOA and with initial
point O and terminal point A
4 ]BAC means the angle with vertex A and bounded by the rays ŒAB and ŒAC
4 †BAC (m†BAC ) means the size of the angle ]BAC (i.e. the magnitude of
the smallest rotation that maps one of the rays onto the other)
4 A, O in a triangle, is the angle opposite the edge a
4 CA means the circle centered at A
4 rA means the radius of circle CA
4 4ABC Š 4DEF means that the triangles 4ABC and 4DEF are congru-
ent, i.e. their respective sides have the same length and their respective angles
have the same size.
In the text at hand, “to construct” means using geometric tools with which we
cannot measure. So we can use a straightedge and a compass, but we can neither
use a ruler nor a protractor.
This means we remain more or less faithful to the classical Greek concept of
geometric construction.
Diligent readers will notice that we sometimes add that the length of a line
segment is a, for instance. This usually simplifies the reasoning in a proof and
avoids cumbersome formulations. In these cases the construction is possible, even
though the intitial line segment itself is not necessarily constructible.
The bulk of the illustrations were sourced from books from three libraries,
which are referred to by the following abbreviations:
EHC – Erfgoedbibliotheek Hendrik Conscience, Antwerp

KBR – Koninklijk Bibliotheek Albert I, Brussels
MPM – Museum Plantin-Moretus/Prentenkabinet, Antwerp – Unesco-World Heritage
The text was typeset in LYX an editor for LATEX.

ix
Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Ad Meskens and Paul Tytgat
1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Interactive Geometry Software (IGS) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.3 Polar coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2 The Genesis of Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

2.1 Greeks and their Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2 Examples from Plato’s hierarchy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.3 Adrian van Roomen and Apollonius’ Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.4 Mechanical construction aids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.5 Neusis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3 Compass and straightedge constructions . . . . . . . . . . . . . . . . . . . . . . . . . . 27

3.1 Euclid’s Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.2 Compass and straightedge constructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.3 François Viète and Apollonius’ Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4 The Delian Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

4.1 Delos and its altar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
4.2 Wine gauging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.3 Doubling the cube with a neusis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5 Trisecting an angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.1 Nicomedes’ conchoid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.2 Archimedes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.3 The first Archimedean trisection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.4 The Archimedean spiral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.5 The Flemish Jesuits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.6 Hippias of Elis and the quadratrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
6 Squaring the circle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

6.1 Archimedes’ spiral revisited . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.2 Dinostratus’ quadratrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
6.3 Hippocrates’ lunes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
6.4 Franco of Liège, the demise of mathematics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.5 Nicolas of Cusa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
6.6 Archimedes’ approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
x Contents
6.7 Adriaan van Roomen and Ludolff van Ceulen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

6.8 Kepler’s use of infinitesimal methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
6.9 Gregory of St Vincent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
7 Constructible numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

7.1 Constructing numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
7.2 The theory of equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
7.3 Squarable lunes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
7.4 Squaring the circle is different . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
8 The Cinderella of regular polygons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121

8.1 The inconstructibility of the heptagon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
8.2 The relation with the trisection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
8.3 A neusis for the heptagon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Servicepart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
xi
List of Figures1
Fig. 1.1 The cubic close packing (drawings by Paul Tytgat) . . . . . . . . . . . . . . 3

Fig. 1.2 The construction of Euclid 1.1 in C.a.R. (Z.u.L.) . . . . . . . . . . . . . . . . . . . . 4
Fig. 1.3 Loci in GeoGebra, Cinderella and CaRMetal . . . . . . . . . . . . . . . . . . . . . 6
Fig. 1.4 Dürer’s conchoid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Fig. 2.12 Drawing an ellipse using two tacks and piece of rope . . . . . . . . . . 19
Fig. 2.13 Archimedes’ trammel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Fig. 2.14 Italian ellipsograph (trammel) ca. 1700 . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Fig. 2.15 Some of Frans van Schooten’s drawing devices to draw a
conic section . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Fig. 2.18 Performing a neusis (drawing by Paul Tytgat) . . . . . . . . . . . . . . . . . . . 24
Fig. 2.20 An instrument to draw conchoids with (drawing by Paul Tytgat) 25
Fig. 2.21 The branches of a conchoid (drawing by Paul Tytgat) . . . . . . . . . . 25
Fig. 3.1 The proof of Pythagoras’ theorem in Byrne’s Euclid (1847) . . . . . . 28
Fig. 3.9 The construction of a square equiareal to a rectangle ABCD
from Marolois’ La Géometrie . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Fig. 3.10 Viète’s solution to Apollonius’ tangency problem from Opera
mathematica (1646) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Fig. 4.2 A winegauger is taking a measurement with his gauging rod
(left), terminology (right, drawing by Paul Tytgat) . . . . . . . . . . . . . . . 47
Fig. 5.3 An instrument with which (a branch of) the conchoid can be
drawn (drawing by Paul Tytgat) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Fig. 5.4 Construction of the conchoid with which, for a given angle,
the trisection can be accomplished (drawings by Paul Tytgat) . 57
Fig. 5.7 Artistic impression of Archimedes’ claw (drawings by Paul Tyt-
gat) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Fig. 5.9 A trisection instrument based on Archimedes’ neusis (draw-
ing by Paul Tytgat) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Fig. 5.13 Antwerp’s Carolus Borromeus Church, designed by Francis-
cus de Aguilon s.j. and Pieter Huyssens s.j. (Photos Ad Meskens) 66
Fig. 5.14 Portrait of Gregory of St-Vincent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Fig. 5.15 Gregory of St Vincent’s generalisation of Pythagoras’ theorem 67
Fig. 5.16 Frontispiece to the second book of de Aguilon’s Opticorum
Libri Sex . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Fig. 6.5 In all of these cases Pythagoras’ theorem holds:
area B C area C D area A (drawing by Paul Tytgat) . . . . . . . . . . . . . 79
Fig. 6.15 Transforming curvilinear figures into rectilinear figures, from
Falco (1591) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Fig. 6.21 Ludolph van Ceulen and his approximation of . . . . . . . . . . . . . . . 96
Fig. 6.25 Gregory’s figure to explain the calculation of the area be-
tween two parabolae and between a parabola and a straight
line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
1
Apart from this list of figures all other figures are made in GeoGebra by Ad Meskens
xii List of Figures
Fig. 6.26 Frontispiece of Gregoy of Saint Vincent’s Problema Austriacum

on the squaring of the circle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Fig. 7.6 Part of Viète’s solution to van Roomen’s problem using chords 111
1 1
Introduction

A. Meskens, P. Tytgat, Exploring Classical Greek Construction Problems with Interactive Geometry
Software, Compact Textbooks in Mathematics, DOI 10.1007/978-3-319-42863-5_1
1.1 Introduction
For ages, mathematicians have challenged one another by posing problems. In most cases
these were generalisations or slightly adapted versions of problems that had already been
solved previously.
For instance, we know that every even number can be written as the sum of two uneven
numbers. The proof of this is nearly self-evident:
2n D .2n 1/ C 1 D .2n 3/ C 3 D : : :
Is it possible to place more rigorous demands on our question? For instance, is it possible
to write each even number as the sum of two uneven prime numbers? Let’s see, we have
20 D 13 C 7, 22 D 11 C 11, 24 D 13 C 11, 26 D 13 C 13, 28 D 17 C 11, . . .
Just pick an even number and you’ll see that you can always find two uneven prime num-
bers whose sum is your even number. Always?? That is the question! To date, no one has
succeeded in proving this so-called Goldbach’s conjecture!
Christian Goldbach (1690–1764) was a German mathematician, who corresponded with
many famous mathematicians of his time, such as Gottfried Leibniz (1646–1716), Leonhard
Euler (1707–1783), and Nicholas I Bernoulli (1687–1759). In a letter addressed to Leonhard
Euler dated 7 June 1742, he proposed that: “Every integer which can be written as the sum of
two primes, can also be written as the sum of as many primes as one wishes, until all terms are
units.” In the margin of his letter he wrote a second conjecture: “Every integer greater than 2
can be written as the sum of three primes.” In his reply of 30 June 1742, Euler reminded him
of a conversation they had had in which he had remarked that his conjecture followed from
the statement: “Every even integer greater than 2 can be written as the sum of two primes.”
The latter conjecture is now known as Goldbach’s conjecture1 .
Another example is Fermat’s Last Theorem. We know that in a right-angled triangle with
legs having lengths 3 and 4, the hypotenuse has length 5. Indeed 32 C 42 D 52 . We know a
multitude of triplets .a; b; c/ with a; b; c 2 N that satisfy the equation a2 C b 2 D c 2 . These
numbers are known as Pythagorean triplets.
Now would it be possible to find triplets that satisfy the equation a3 C b 3 D c 3 or
a C b 4 D c 4 or an C b n D c n (n > 2, n 2 N).
4
1
Euler; Koch (2007).
2 Chapter 1 Introduction
1
Greek mathematician Diophantus (3rd century AD) had proved, amongst other theorems
in number theory, that any square of a rational number can be decomposed into two other
squares of rational numbers (theorem II.8). When reading the 1621 edition of Diophantus by
Claude Gaspard Bachet de Méziriac (1581–1638), French mathematician Pierre de Fermat
(1601 or 1606/7–1665) made marginal notes. In the margin next to theorem II.8 Fermat wrote
that he had found a proof for the assertion that for all but trivial cases, no integer solutions
exist for the equation an C b n D c n (n > 2, n 2 N). “Alas”, he added, “this margin is too
small to contain the proof”. Nowhere among his papers the proof was found. For centuries,
mathematicians tried to find a proof. Only in 1994 did Andrew Wiles succeed in proving it.
The proof is incomprehensible to all but the brightest mathematicians. Wiles used the theory
of elliptic curves, a subject which at first glance is light years away from number theory2 .
Problems such as those cited above are the kind of problems that mathematicians love:
they are easy to pose, a secondary school pupil can understand them and . . . they are very,
very hard to solve.
Exercise 1
Show that .2pq; p 2 q 2 ; p 2 C q 2 / is a Pythagorean triplet for all p; q 2 N.
Show that .2n2 C 2n; 2n C 1; 2n2 C 2n C 1/ is a Pythagorean triplet for all n 2 N. Also show
that these triplets are a special case of the triplets in the first part of the exercise.
Exercise 2
Prove that .3; 4; 5/ is the only Pythagorean triplet containing three consecutive positive integer
numbers.
Exercise 3
Prove that no isosceles right triangle exists whose sides are integers.
Exercise 4
In the Treehouse of Horror VI episode of The Simpsons (the animated TV series), one can see the
equation 178212 C 184112 D 192212 . In another episode, entitled The Wizard of Waverley Terrace
one can see 398712 C 436512 D 447212 . Use a pocket calculator to calculate both sides of the
equations. The equations seem to hold.
Use elementary number theory to disprove these equations, without actually calculating the
values of both sides of the equation. (Hint: look at the divisibility of each term in these equations).
Exercise 5
The Kepler conjecture was a problem posed by the astronomer Johannes Kepler (1571–1630). It
says that no arrangement of equally sized spheres filling space has a greater average density than
that of the cubic close packing (a packing known from material science). . Fig. 1.1 (left) shows a
cubic close packing. A unit cell is the simplest structural unit which describes such a crystalline
structure. The unit cell for the cubic close packing is shown in . Fig. 1.1 (right). It is a cube with
one eight of a sphere at each vertex and half a sphere at each face. Show that the space taken up
by the parts of the spheres is approximately 74% of the volume of the cube and hence that the total
density is about 74%. The conjecture itself was proved in 2005 by Thomas Hales3 . Hint: replace
2
Singh (1997); Lanser (2000); Barner (2001); Meskens (2010), p. 166–169.
3
Hales (2005).
1.2 Interactive Geometry Software (IGS)
3 1
. Fig. 1.1 The cubic close pack-
ing (also called the face-centered
cubic), a crystalline structure (draw-
ings by Paul Tytgat)
the parts of the spheres in the unit cell by the equivalent number of spheres, calculate the volume
of these spheres and divide it by the volume of the unit cell. To calculate the volume of the sphere,
express the radius of the spheres in terms of the edges of the cube.
The same kind of problems exists in plane geometry. This is the story of three of these
problems which have haunted mathematicians for ages. They are old, they are easy to formu-
late and they are . . . very hard to solve. Yet with a little help, any mathematics student can get
to grasp them.
Consider a square. It is easy to construct, using only compass and straightedge, a square
which is twice as large as a given one. One needs only to draw the diagonal and consider this
as the edge of the new square. Would it be possible to construct a cube which has a double
volume of a given one, using only compass and straightedge?
It is easy to divide an angle into two equal angles by drawing, using only compass and
straightedge, the angle bisectors. Now would it be possible to divide an angle into three equal
angles, using only compass and straightedge?
It is possible, using rigid transformations equivalent to using only compass and straight-
edge, to transform a rectangle or a polygon into a square. But is this also true for a circle?
These problems are respectively referred to as the duplication of the cube, the trisection of
an angle and the squaring of the circle. We shall see how mathematicians tried to solve these
problems but, if limited to compass and straightedge, failed to do so.
Many of the illustrations in this book are made with GeoGebra, but they could just as easily
have been made with Geometer’s Sketchpad, Cinderella or any other Interactive Geometry
Software or IGS for short. Interactive Geometry Softwares are all interactive geometry, alge-
bra, and calculus applications, intended for teachers and students. They allow one to make
constructions with points, vectors, segments, lines, polygons, conic sections and functions.
All of these objects can be changed dynamically afterwards. Elements can be entered and
modified directly on screen. Most options can be activated by clicking an icon in the toolbar.
Sometimes these icons are grouped and can be accessed through a drop down menu. The
groups typically include a Point toolbox , a Lines toolbox , a Lines Properties toolbox
and a Curves toolbox . This last toolbox holds buttons for circles, circular arcs and
conic sections.
1
. Fig. 1.2 The construction of
Euclid 1.1 in C.a.R. (Z.u.L.)
Other toolboxes typically are a Transform toolbox , allowing you to translate, rotate
or mirror an object and a Measure toolbox , allowing you to measure the length of a line
segment or to determine the measure of an angle. Some of these programs also allow you to
check properties such as collinearity or perpendicularity.
Because of the similarity between these programs we are able to write a book using generic
terms to indicate icons, without referring to a specific program. Yet the similarity of names,
and in many cases also of icons, between the programs allows the reader to recognize all the
names of icons in his favourite IGS. We only use straightedge and compass constructions,
which are very basic geometric constructions, so they can all be carried out with the basic
tools of an IGS.
As an example, open your IGS, click on Create Point and click twice inside the
drawing window. You will see two points appear: A and B. Now select Segment , click
both points A and B. Click Circle with Centre and Point , click A and then B. Follow the
same procedure, but reverse the order of A and B. You will see two circles. Select Intersect
then click each circle – points D and C appear (. Fig. 1.2).
The triangles 4ABC and 4ABD are equilateral. What you have just done is your IGS’s
equivalent to Euclid’s first construction in book I of Elements (see 7 Sect. 3.1).
Exercise 6
Create an IGS file in which you draw a triangle 4ABC . You can do this in one of two ways. You
can draw three points and connect them with line segments. Or you can use the Polygon tool, draw
three points and close the polygon by clicking the first point.
1. Determine the intersection point(s) of the perpendicular bisectors of the sides of the triangle
4ABC .
2. Determine the intersection point(s) of the medians of the triangle 4ABC .
3. Determine the intersection point(s) of the altitudes of the triangle 4ABC .
4. Determine the intersection point(s) of the bisectors of each vertex of the triangle 4ABC .
In each of the above cases you will notice that the respective lines intersect in one point. Now for
each of the cases, using the Move tool, move any of the vertices of the triangle around. What do
you notice?
5 1
Exercise 7
Illustrate Napoleon’s theorem in IGS: if equilateral triangles are constructed on the sides of any
triangle, either all outward or all inward, the centroids of those equilateral triangles themselves
form an equilateral triangle.
Exercise 8
A well-known tale about a treasure island is this:
A pirate was stranded on a deserted island. Among the wreckage was a treasure, which he
buried. The island was home to one oak tree and two birches. The pirate walked from the oak tree
to one of the birches, then turned at a right angle and walked an equal number of steps. Here he
drove a spike into the ground.
He returned to the oak tree and repeated the procedure with the second birch. Again, he drove a
spike into the ground. He buried the treasure halfway between both spikes and then removed the
spikes.
A week later he was rescued.
Two years later he returned to the island only to find that the coastline had receded and the oak
tree had disappeared.
Create an IGS file for this problem and show how the pirate will be able to find his treasure
anyway (hint: choose three points (oak, birch1 and birch2) and for this configuration determine
where the pirate will bury his treasure). Now move the point representing the oak around. Prove
why this is the case for this configuration. Do configurations exist in which he will not be able to
find his treasure?
These exercises show the great advantage of an IGS: we can experiment and see a myriad
of cases by just moving one or a couple of points around.
We will now take a closer look at some concepts we regularly use in the exercises: sliders
and loci.
A slider is a tool you can vary a parameter continuously with. Not all IGSs have these
sliders, but it is not too difficult to construct one. A slider takes the form of a line segment on
which a point can move indicating its the value.
kAC k
First define a line segment ŒAB and a moveable point C on it. Calculate a D .
kABk
Moving C along ŒAB will let a vary between 0 and 1. If you wish your slider to vary between
kAC k
0 and simply multiply the ratio by W a D . You can even let your slider take on
kABk
negative values. Define a line segment ŒAB and let its midpoint be M . Let C be a point on
kAC k kAM k
ŒAB and define a D , then a will vary between and . In some IGSs
kAM k
some programming (“scripting”) may be necessary.
. Fig. 1.3 Left the locus generated with the command Trace, right the locus generated with the command Locus
in GeoGebra (top), Cinderella (middle) and CaRMetal (bottom)
The Locus and Trace tools have similar effects, but with the first option the IGS itself
draws the solution, whilst the second tool enables users to dynamically create the result. An
example will clarify this immediately.
Suppose a point, A (the mover), can move freely on a curve (a straight line, a circle, . . . )
called the directrix or road. Suppose the position of another point, B (the tracer), in some way
depends on the position of A. It is then possible to find the locus of B as A moves along the
directrix.
A first method to find this locus is to select Trace and drag A along the curve using the
cursor. B now traces out a line. This line is part of the locus of B. The effect is even more
spectacular if you use animation. A will now automatically move along its directrix, while B
traces out the locus.
We can also select Locus and indicate the mover A and the tracer B. The locus immedi-
ately becomes visible. In the following paragraphs, we will call this “the locus of B with A”,
meaning “the locus of B while A moves along the directrix”.
Let us take a look at an example. On a straight line l select two points A and C . Now select
a point D, not on l (for ease, select C on the negative x-axis, A on the positive x-axis and D
on the negative y-axis). Draw the straight line AD and the straight line lC parallel to AD and
1.3 Polar coordinates
7 1
. Fig. 1.4 Dürer’s conchoid (right: Dürer (1525), EHC H 202415)
through A. Draw the line lC perpendicular to CD and through C . Determine the intersection
B of lC and lA . The position of B is dependent on each of the positions of A, C and D. We
will call this drawing the initial configuration.
Click B and select Trace. Move C along the straight line l. We find a line which is (a part
of) the locus of D as C moves along l (see . Fig. 1.3, left).
Now return to the initial configuration. Select Locus, then select the mover C and the
tracer B, the locus of B appears immediately (see . Fig. 1.3, right).
Exercise 9
Putting A.a; 0/, C.t; 0/ and D.0; d / with t 2 R, determine the Cartesian equations of CD; lD
and lA . Determine the system of equations with which you can determine the coordinates of B.
Eliminate t from these equations. Prove that the resulting Cartesian equation of the locus of B with
C is:
y 3 C .x a/xy C d.x d /2 D 0 (1)
Exercise 10
Let A and B be points moving on a pair of perpendicular lines which intersect at O in such a way
that kOAk C kOBk is constant. On AB, mark points C and D at a fixed distance c from A. The
loci of points C and D is Dürer’s conchoid (see . Fig. 1.4). Although it is named conchoid, it is
not a true conchoid (see 7 Sect. 5.1). Create an IGS file in which you draw Dürer’s conchoid. The
geometric definition can be generalised analytically. Let A and B be on the x- and y-axis respec-
tively, the condition then becomes Ax C By D b, b 2 R. Define three sliders a, b and c. Define
A.a; 0/ and B.0; b a/. Determine the points C and D on AB for which kAC k D kADk D c.
Moving the slider a while Trace is on for C and D will show you Dürer’s conchoid. If you want
to use Locus, the mover, slider a, and the tracer, point C or D, the loci of C and D will appear
respectively. You can still move sliders b and c for other configurations.
Cartesian coordinates are not the only coordinates used in mathematics. One other coordinate
system is that of polar coordinates.
In polar coordinates we determine the position of a point P by a distance r to a given
point O, the origin or pole, and the angle between the positive x-axis or polar axis and ŒOP ,
1
. Fig. 1.5 Polar coordinates of a
point P
in counterclockwise direction. We call .r; / the polar coordinates of point P . By definition,

the origin has polar coordinates.0; / for any .
Using trigonometry, we can deduce transformation formulae to find the Cartesian coordi-
nates if polar coordinates are known and vice versa.
Take a closer look at . Fig. 1.5: in 4OPQ QO is a right angle. So we can use the formulae
for right-angled triangles.
Suppose P has Cartesian coordinates .x; y/ then in 4OPQ:
8
<kOQk D kOP k cos †QOP
: kQP k D kOP k sin †QOP
or
8
< x D r cos
(2)
:y D r sin
for short.
Similarly, in 4OPQ: kOQk2 C kQP k2 D kOP k2 or x 2 C y 2 D r 2 .
y
Furthermore tan D .
x
Or
8 p
< r D x2 C y 2
(3)
:tan D y :
x
We have now proved the transformation formulae for the first quadrant. It turns out that we
can use the very same sets of formulae (2) and (3) in the other quadrants.
y
In the first quadrant we can also use D arctan .
x
Exercise 11
Prove the transformation formulae (2) and (3) for the other quadrants.
A polar equation is an equation in r and . A solution to a polar equation is an ordered

pair .R; #/ that satisfies the equation if R is substituted for r and # is substituted for . The
graph of a polar equation is the set of all points with coordinates .R; #/ which satisfy the
equation.
Some IGSs allow you to work in polar coordinates, but for a number of reasons it may
also be more appropriate to keep working in a Cartesian coordinate system. In many cases the
9 1
polar equation can be expressed as r D f . /. First define a function r.t / D f .t / (each
in the polar equation is replaced by t ). Then let your IGS draw the curve defined by points
.r.t / cos t; r.t / sin t /, you can do this in a number of ways e.g. by defining the parametric
curve
8
< x D r.t / cos t
:y D r.t / sin t:
Here are two examples. The curve with polar equation r D 2 sin looks like this:
We obtain a circle with radius 1 of which the centre has Cartesian coordinates .0; 1/.
The curve with polar equation r D 2 C 4 cos shows a completely different picture:
This curve is called Pascal’s limaçon, named after Etienne Pascal (1588–1651), father of the
well-known mathematician Blaise. In French, “limaçon” means “little snail”.
Exercise 12
(a) Use the transformation formulae to obtain the Cartesian equation of the circle with polar equa-
tion r D 2 sin .
(b) Which curve is represented by the polar equation r D 2 cos ? Which Cartesian equation does
this curve have?
(c) Which curves are represented by r D 2a sin and r D 2a cos ?
Exercise 13
Create an IGS file in which you define a slider a. Draw the curve(s) with polar equation r D
2 C a cos . Let a vary between 0 and 10.
Exercise 14
Use the transformation formulae to obtain the polar equation of the lines with Cartesian equation
x D a, y D b, y D mx, y D mx C q.
11 2
The Genesis of Geometry

2.1 Greeks and their Geometry
Many elementary geometry books for secondary schools contain exercises in which a con-
struction with compass and straightedge is called for. In many cases, these are exercises whose
history goes back to Ancient Greece.
In the history of western civilisation, Greek Antiquity occupies a prominent place. This
civilisation is sometimes called the cradle of western civilisation, philosophy and mathemat-
ics. Greek civilisation can not be reduced to the Greek mainland however. At its zenith, it
stretched from Sicily across the eastern Mediterranean to Asia Minor (now Turkey) and the
Black Sea coasts (see . Fig. 2.1).
Next to Athens and Sparta, the main Greek cities of the era were Alexandria on the Nile
delta, Syracuse on Sicily and Milete on the coast of Asia Minor. Greek identity was defined
through common trade interests, not as a nation state. Politically, the polis or the city state was
the most important political entity.
Despite the impression to the contrary, our knowledge about the lives of Greek mathe-
maticians is limited to say the least. Moreover it is hard to separate fact from fiction as most
biographical information about them was written several decades, sometimes even centuries,
after their death. This applies even more to pre-Socratic thinkers who lived before the fifth
century BC.
The first Greek mathematician we can identify with some confidence is Thales of Milete
(ca. 624–548/45 BC). Thales lived in Ionia (on the coast of Turkey). It was there that classical
Greek science emerged, which would go on to have such a profound impact on our science.
It was in Ionia that the idea took root that complex, natural phenomena can be explained by a
set of basic rules.
Some of the theorems that are attributed to Thales include:
1. A circle is bisected by its diameter
2. The base angles in an isosceles triangle are equal
3. When two straight lines intersect then the opposite angles are equal
4. An angle inscribed in a semicircle is a right angle (Thales’ circle theorem)
5. If two intersecting lines are cut by a two parallel lines then two similar triangles are pro-
duced (Thales’ intercept theorem)
12 Chapter 2 The Genesis of Geometry
. Fig. 2.1 Greek colonisation about 500 BC (after Vermaseren (1977), p. 9)
Plato (Athens ca. 427 BC–ca. 348 BC) was a philosopher, as well as a mathematician. He
founded the Academy in Athens, the first institution of higher learning in the Western world.
Plato, Socrates and his most famous student, Aristotle, are seen as the founders of Western
philosophy and science. From the fourth century onwards, Plato’s philosophical ideas gained
ground. Their influence on mathematics became visible in geometry, in which hierarchies of
classes of constructions emerged.
The first class was the most abstract, the third and last was the most “earthly and mechan-
ical”:
1. constructions with only straight lines and/or circles, i.e. constructions with compass and
straightedge
2. constructions in which conic sections (parabolae, hyperbolae and ellipses) are used
3. constructions with other (mechanical) construction means
Of course these constructions need to be accomplished in a finite number of steps.
2.2 Examples from Plato’s hierarchy
In Plato’s hierarchy, compass and straightedge constructions are the most abstract. Here is one
example of such a construction.
To draw a hexagon inscribed in a circle, we observe the following procedure (see . Fig.
2.2):
» Select the width of the compass equal to the radius of the given circle
Select a point A on the circle
Place the compass point in A
Draw arcs which intersect the circle at B and C
Place the compass point, with the same width, in B and draw an arc
This arc intersects the circle at A and D
2.2 Examples from Plato’s hierarchy
13 2
. Fig. 2.2 The construction of a
regular hexagon inscribed in a circle
Place the compass point, with the same width, in D and repeat the above procedure
until an arc intersects the circle at C
The intersection points which you have constructed are the vertices of a regular hexagon
inscribed in the circle.
Plato’s second category contains constructions which can be carried out with the aid of conic
sections. A circle, which can be drawn with the aid of a compass, is a special type of conic
section. It is therefore natural to consider the conic sections as additional construction aids. To
show why they cannot be included in the first category, let us first take a look at the parabola.
A parabola is defined as the locus of all points having the same distance to a given line,
directrix d , and to a given point, focus F .
We can construct points of a parabola with this procedure (see . Fig. 2.3):
» Draw the perpendicular on d and through F

This line intersects d in D
kDF k
Draw straight lines di parallel to d at regular intervals, beginning at a distance
2
to d
Let the intersection of di with DF be Di
For each of these lines set the compass width at kDDi k, place the compass point in F
. Fig. 2.3 The construction of a parabola with focus F and directrix d

Draw arcs which intersect di

2 Let these intersection points be denoted by Pi;1 and Pi;2
Pi;1 and Pi;2 are points of the parabola
D1 is a special point, here P1;1 and P1;2 coincide. D1 is called the vertex of the parabola,
and is often referred to with the letter V . The line DF is called the axis of symmetry.
We notice that it is possible to construct individual points of the parabola with compass and
straightedge, but that it is not possible to construct the curve itself in this way.
Plato’s second category therefore is a natural extension of the first category.
Exercise 15
Create an IGS file in which you define two sliders a and b both ranging from 0 to 10. Draw the
straight line y D a and the point A .0; a/. Draw the straight line y D b and the circle centered at
A and with radius a C b and determine their intersections B and C . For B and C select Trace and
move the slider b. You will see a parabola appear. Alternatively click on Locus, select the mover,
the slider b, and the tracer, B and C respectively. You will now see a parabola being traced out.
What happens when you move the slider a (do NOT use Animation On)?
Exercise 16
Select a system of axes x and y, in which
p x coincides with DF and y intersects ŒDF pat the
midpoint. Let the coordinates of F be ; 0 , then the Cartesian equation of d is x D . Use
2 2
the formulae for the distance between two points and the distance of a point to a straight line to
prove that the Cartesian equation for the parabola in this configuration is y 2 D 2px.
2.3 Adrian van Roomen and Apollonius’ Problem
Some problems require ingenious methods to find a compass and straightedge solution, but can
be easily solved using conic sections. One such problem is Apollonius’ problem, for which
a solution with conic sections, i.e. belonging to Plato’s second category, is straightforward.
Later we will also discuss the compass and straightedge construction (see 7 Sect. 3.3).
Apollonius (ca. 262 BC–ca. 190 BC) is best known for his work on conic sections. His
book Conics is preserved partly in Greek with commentaries by Eutocius and partly in an
Arab translation by Thabit ibn Qurra. Apollonius’ other work has unfortunately been lost, but
we know of his exploits through Pappus of Alexandria’s commentaries1 . In his book ᾿Επαφαί
(Epaphaí, “Tangencies”) Apollonius posed, and solved, the problem which is now known as
Apollonius’ problem: find a circle which is tangent to three given circles2 . Three given circles
have at most eight circles which are tangent to all of them. Each solution encloses or excludes
the circles in a different way.
Adriaan van Roomen, also known as Adrianus Romanus (1561–1615) was a Flemish
physician and mathematician who solved the problem using conic sections. Van Roomen be-
came professor of mathematics and medicine first in Leuven then at Würzburg (1593). In 1604
he was ordained priest. He died in 1615 in Mainz en route to Leuven. He corresponded with
the foremost mathematicians of his age, Ludolff van Ceulen and François Viète among them3 .
1
On Apollonius and Conics see Heath (1961); Fried and Unguru (2012).
2
For a reconstruction of Apollonius’s solution see Heath (1981) vol. 2, p. 182–185.
3
Bockstaele (1976, 2009).
15 2
. Fig. 2.4 Adriaan van Roomen’s
challenge: to find a solution to a
45th degree polynomial equation
(van Roomen (1593), MPM 8 533)
In 1593 he published Ideæ Mathematica in which he challenged his contemporaries to solve

a 45th degree polynomial equation4 (see . Fig. 2.4):
45x 3795x 3 C 95 634x 5 1 138 500x 7 C 7 811 375x 9 34 512 075x 11

C 105 306 075x 13 232 676 280x 15 C 384 942 375x 17 488 494 125x 19
C 483 841 800x 21 378 658 800x 23 C 236 030 652x 25 117 679 100x 27
C 46 955 700x 29 14 945 040x 31 C 3 764 565x 33 740 259x 35
C 111 150x 37 12 300x 39 C 945x 41 45x 43 C x 45 D C
s r q
p
C D 2C 2C 2C 2
At one point an ambassador from the Low Countries boasted to the French King Henry IV
that van Roomen had listed all European mathematicians but did not mention any French
mathematicians. He also referred to van Roomen’s challenge. Henry IV had Viète summoned
and had him read van Roomen’s book. Viète recognised that the polynomial expresses the
45˛ ˛
length of a chord d45 D chord.45˛/ D 2 sin given that x D chord.˛/ D 2 sin .
2 2
He reported to the King that same day and gave him two solutions, of which one was
4
Bockstaele (1993).
. Fig. 2.5 Viète’s solution to van Roomen’s problem (Viète (1646), EHC G 4858)
. Fig. 2.6 Circle C touches circle

C1 internally and circle C2 externally
s r q p p
x0 D 2 2 C 2 C 2 C 3, the edge of a regular 96-gon (see 7 Sect. 6.7). He
claimed that he could calculate all other solutions as well (see . Fig. 2.5).
Viète published his solution in a book Ad problema quod omnibus mathematicis totius
orbis construendum proposuit Adrianus Romanus, Francisci Vietae responsum (1595). In this
book he challenged van Roomen to a new problem: this was nothing else than Apollonius’
problem. Viète maintained that, because all curves have a degree two it should have a compass
and straightedge solution (also see 7 Sect. 3.3). van Roomen was able to solve the problem,
but not with compass and straightedge. He used hyperbolae to solve the problem, his solution
thus belongs to Plato’s second category5 .
Suppose the circles C1 , C2 and C3 , centered at O1 , O2 and O3 and with radii r1 , r2 and
r3 respectively are given. Van Roomen wants to find the locus of the centre of all circles
which are tangent to C1 and C2 and the locus of the centre of all circles which are tangent
to C1 and C3 . In both cases C1 has to be touched in the same way by the tangent circle, i.e.
either internally or externally (see . Fig. 2.6). The intersection of these two loci then yields
the centre of a circle which is tangent to the three circles and hence a solution to Apollonius’
problem.
Exercise 17
Prove that if a circle C centered at O and with radius r is tangent to a circle C1 centered at O1 and
with radius r1 , then kO1 Ok D r1 C r.
Suppose a circle C centered at O and with radius r is tangent to C1 and C2 (see . Fig.
2.7). We notice that kO1 Ok kO2 Ok D .r1 C r/ .r2 C r/ D r1 r2 D constant. We
therefore have to determine the locus of all points for which the differences of the distances to
the centres of the circles is r1 r2 . This is a hyperbola H12 with foci O1 and O2 . All points
on this hyperbola are the centre of circles which are tangent to both circles C1 and C2 . We
can now do the same for C2 and C3 and we again find a hyperbola H23 . The intersections
5
Barbin and Boyé (2005), p. 10, 24–27, Bos (2001), p. 110–112.
17 2
. Fig. 2.7 The locus of the centres
of the circles tangent to two given
circles is a hyperbola. kO1 Ok
kO2 Ok D .r1 C r/ .r2 C r/ D
r1 r2 D constant
of these two hyperbolae yield two centres of tangent circles (see . Fig. 2.8). To see why not
all intersections yield a centre for a tangent circle we need to look at the branches of the
hyperbolae. For instance for H12 one branch yields all the centres of circles C which are
externally tangent to C1 and C2 – i.e. C contains neither C1 nor C2 – the other branch yields
the centres of circles C for which C1 and C2 are internally tangent to C (see . Fig. 2.9).
Constructing e.g. H13 yields other solutions to the problem (see . Fig. 2.10).
Exercise 18
Create an IGS file in which you solve Apollonius’ problem. Select three points A, B and C and
draw circles CA ; CB ; CC centered at these points and with chosen radii rA ; rB ; rC . Draw a line
through A and B on the one hand and B and C on the other. Determine the intersection points
of these lines with the respective circles. For the line through A and B, select one of the intersec-
tion points on each circle, IA and IB and determine their midpoint MAB (see . Fig. 2.11). A and B
are the foci of a hyperbola and MAB is on the hyperbola. Repeat this procedure for B and C . B and
C are the foci of the second hyperbola, and MBC is on the hyperbola. Determine the intersections
of the hyperbolae. Select one of the intersection points as the centre of a circle. Select the option
. Fig. 2.8 Three circles, in solid lines, are given. The hyperbolae which are the loci of the centre of circles tangent
to C1 and C2 on the one hand and C2 and C3 on the other are drawn. The intersection point A is the centre of a
circle externally tangent to the three circles (dotted line). The intersection point B is the centre of a circle for which
the given circles are internally tangent (dashed line)
. Fig. 2.9 In this figure intersection point C is used to construct the tangent circle. We clearly see that one circle
centered at C is externally tangent to two circles (dotted line), while another is internally tangent to two circles
(dashed line). The branch of H12 through C yields all the centres of circles C which are internally tangent to C1 and
C2 , the branch of H23 yields the centres of circles C which are externally tangent to C2 and C3 . Obviously a tangent
circle to the three circles cannot be tangent to C2 internally and externally at the same time
. Fig. 2.10 Another solution to Apollonius’ problem. The hyperbolae which are the loci of the centres of circles
tangent to C1 and C2 on the one hand, and C1 and C3 on the other, are drawn. The intersection point D is the centre
of a circle externally tangent to one circle and internally tangent to two circles (dashed line). The intersection point
E is the centre of a circle externally tangent to two circles and internally to one (dotted line)
. Fig. 2.11 MAB is the midpoint

of ŒIA ; IB . We can distinguish
between four line segments
for which the endpoints are
on the respective circles:

ŒIA ; IB ; IA ; IB0 ; IA0 ; IB ; IA0 ; IB0
2.4 Mechanical construction aids
19 2
Circle with centre and through point, draw the circle in such a way that it is tangent to one of the
given circles. You will notice that either the circle is tangent to only two circles or that it is tangent
to the three circles. Depending on which of the intersection points of the line with the circle you
have chosen the tangency will either be internal or external.
Exercise 19
Use the IGS file of Exercise 18 to identify all possible solutions to Apollonius’ problem. The
problem has a maximum of eight solutions.
Although conic sections cannot be constructed with compass and straightedge, they can be
constructed using mechanical means, methods from Plato’s third category. Consider for in-
stance an ellipse. An ellipse is defined as the locus of the points of which the sum of the
distances to two given points (the foci) is a constant. Figure . Fig. 2.12 shows how an ellipse
can be constructed with the aid of two drawing pins and a piece of rope.
Exercise 20
Create an IGS file in which you choose two numbers, a and c, c < a. Define a slider t ranging from
a c to a C c. Draw the points F 0 .c; 0/ and F .c; 0/. Draw the circles C1 , centered at F and with
radius t, and C2 , centered at F 0 and with radius 2a t. Determine the intersection points D and E
of the circles C1 and C2 . Determine the locus of D and E respectively with the slider t.
There are other ways to describe an ellipse with mechanical aids (see . Fig. 2.13 right and
. Fig. 2.14). One is Archimedes’ trammel for which we use a ruler, a set square, and a pencil
(see . Fig. 2.13 left):
» Draw two perpendicular lines x; y on paper; these will be the major (x) and minor (y)
axes of the ellipse
Mark three points A, B and C on the ruler such that kAC k is the length of the
semi-major axis and kBC k is the length of the semi-minor axis
With one hand, move the ruler across the paper, turning and sliding it so as to keep
point A on line y, and B on line x at all times
With the other hand, keep the pencil’s tip on the paper, following point C of the ruler
The tip will trace out an ellipse.
. Fig. 2.12 Drawing an ellipse us-

ing two tacks and piece of rope (van
Schooten (1659), KBR VH 8.040 A)
. Fig. 2.13 Archimedes’ trammel (right: van Schooten (1659), KBR VH 8.040 A)
. Fig. 2.14 Italian ellipsograph signed on the arm “Dominicus Lusuerg F. Romae 1700”. The Lusuerg family had some
of the most remarkable craftsmen building scientific instruments of the late 17th and early 18th centuries, such as
Dominicus (1669–1744) who built a wide number of mathematical instruments. Photos Tesseract – Early Scientific
Instruments (Hastings-on-Hudson)
Exercise 21
Prove the procedure of Archimedes’ trammel (hint: determine the coordinates of C when AB
makes an angle with the x-axis). Prove that ŒAB is the diameter of a circle C which passes
through the origin. If ŒOD is a diameter of C , prove that D traces out a circle as A moves along
the x-axis (the locus of D with A).
Exercise 22
Create an IGS file in which you mimic Archimedes’ trammel.
Quite a number of drawing devices with which conic sections can be drawn were proposed
by Frans van Schooten in his book Exercitationum Mathematicorum libri quinque (Leiden,
1657) (see . Fig. 2.15). Frans van Schooten was the grandson of Franchois Verschooten,
a Fleming who had fled the religious intolerance of the Spanish Netherlands in 1584 and
had settled in the Northern Netherlands, which was fighting for independence from Spain6 .
6
van Maanen (1987), p. 212–213.
21 2
. Fig. 2.15 Some of Frans van Schooten’s drawing devices to draw a conic section (van Schooten (1659), KBR VH
8.040 A)
His father, Frans van Schooten sr., was a professor at the Leiden engineering school. Frans
jr. followed in his footsteps. He championed the ideas of Descartes and made fundamental
contributions to analytic geometry.
Exercise 237
One of van Schooten’s devices can be seen in the figure (van Schooten (1659), KBR VH 8.040 A).
On a long slat KL a small slat AB, with length a, can rotate about A. Another slat BE can
rotate about B. On BE a point D is determined for which kBDk D kABk D a.
7
van Maanen (1987), p. 215.
. Fig. 2.16 The “construction”

2 of a tractrix. After a drawing in
Huygens’ manuscript on tractional
motion (University Library Leiden,
Ms Hug. 6, f64r)
Point D is attached to KL and can move along this slat. Put kDEk D b. As D is moved along KL
E traces out an ellipse.
Prove this assertion by finding the equation of the ellipse in a system in which KL is the x-axis
and A the origin.
Exercise 24
Create an IGS file in which you mimic van Schooten’s device of Exercise 23.
An archetypical example of a curve which is drawn mechanically is the tractrix (see . Fig.
2.16). In the seventeenth century, watches were rather heavy, and were usually kept on a chain
in a pocket. The problem of which curve the watch would observe if the end of the chain was
dragged along a straight line, was put by Claude Perrault (1613–1688) to Gottfried Leibniz
(1646–1716) in 1676. Perrault is best remembered as the brother of Charles Perrault, author
of such classic children’s tales as “Cinderella” and “Puss-in-Boots”. Trained as a physician, in
1666 Claude was invited to become a founding member of the Académie des Sciences, where
he earned a reputation as an anatomist.
The first known solution to Perrault’s problem was given by Christian Huygens (1629–
1695), who named the curve the tractrix from the Latin “tractus” which denotes something
that is pulled along. In Huygens’ presentation, it is not a watch but a cart of which the drawbar
is being drawn along a line. At the rear of the cart a pencil is attached which traces out the
tractrix8 .
Although solving this problem is just beyond the scope of this book we will give an outline
here for those who have a knowledge of differential equations.
Take a look at . Fig. 2.17. Suppose P .x; y/ is a point on the curve. Suppose C has co-
ordinates .x; 0/, suppose D is the other end of the drawbar ŒPD, moving along the x-axis.
The drawbar, which has length a, has to be tangentp to the curve. If the height is y, then using
Pythagoras’ theorem, we know that kCDk D a2 y 2 . Therefore the slope of the straight
y
line along the drawbar is p . On the other hand, calculus proves that the slope is
dy a y2
2
given by .
dx
8
Bos (1989), p. 10–12.
2.5 Neusis
23 2
. Fig. 2.17 The tractrix
The problem reduces to solving the differential equation
dy y
D p :
dx a y2
2
This is an equation which can be solved by separating the variables to produce the following
solution:
Za p 2 p !
a t2 a C a2 y 2 p 2
xD dt D ˙ a ln a y : 2
t y
y
Unfortunately, the nature of this problem, involving differentials, does not allow for a con-
struction to be created in an IGS using elementary techniques.
2.5 Neusis
A particular construction method belonging to Plato’s third category is a neusis (from the
Greek νεῦσις from νεύειν or “neuein”, meaning “inclining or pointing towards”; plural:
νεύσεις “neuseis”). In a neusis construction, a line segment ŒAB with length a is fitted be-
tween two curves, l, the directrix, and m, the captrix (from the Latin captare, meaning “to
catch”), in such a way that the straight line AB passes through a given point P , the pole of
the neusis (see . Fig. 2.18). Neusis is also called verging in English. To obtain this solution
one verges a ruler, on which ŒAB is marked off, through P in such a way that A follows the
directrix. The solution is obtained at the position where B is on the captrix. Note that the roles
of the directrix and the captrix can be interchanged. If B follows line m, then the solution is
reached when A is on l (see . Fig. 2.19).
When the line segment is fitted in between the curves, a solution for a particular problem
is found (e.g. the trisection of an angle, the duplication of the cube, . . . ). Implicitly, by using a
neusis we have defined a new curve: the locus of point B while A moves along the directrix.
The question is thereby reduced to finding the intersection(s) of this locus with the captrix m.
Note that we have already defined such a curve as “the locus of B with A” (see 7 Sect. 1.2).
. Fig. 2.18 Performing a neusis

2 (drawing by Paul Tytgat)
Exercise 25
Create an IGS file in which you select a pole P for a neusis. Draw a circle centered at O, not
coincident with P , and with radius 5. Draw a straight line a which intersects the circle. Select
a point A on a. Draw PA and determine the points B and C for which kABk D kAC k D 3.
Determine all positions for which either ŒAB or ŒAC can be fitted in between a and the circle by
dragging A along a. Remember the point A does not have to lie within the circle.
Exercise 26
Use the IGS file from Exercise 25 and draw the locus of B with A and of C with A. The inter-
sections of both branches of the locus give the positions for either B or C when the line segment
is fitted in between the straight line a and the circle. You can determine these positions by sliding
A over a. Let the position of the pole P and the orientation of the line a differ. Can you find an
orientation for which there are 2, 3, 4, 5, 6 intersections of the branches with the circle? Can you
find a configuration with only one intersection?
We can make an instrument to draw such curves. Take a ruler with a slot in the middle.
Put a cursor in the pole P and put the slot of the ruler over it. From the end of the ruler A
. Fig. 2.19 An example of a neusis. Suppose that given a pole P a line segment ŒAB has to be fitted in between
the straight line and the circle. In the left figure the straight line is used as directrix, in the middle one the circle is the
directrix. In the right figure (one of) the final position(s) is reached. If we use the circle as directrix A has to coincide
with the intersection of the straight lines. If the straight line is used a directrix, then B has to coincide with the
intersection of the line PA and the arc of a circle DF
2.5 Neusis
25 2
. Fig. 2.20 An instrument to draw conchoids with (drawing by Paul Tytgat)
. Fig. 2.21 D1 and D2 are branches of a conchoid generated by K (drawing by Paul Tytgat)
determine a point B for which kABk D a. Attach a pointer to B and a pencil to A. Let the
pointer B follow the directrix l and the pencil, at A, will draw the desired curve.
Of course, there are two points at a distance a from a given point. So again take the ruler
and from the end A1 mark off a point B at a distance a. Then from B, mark off another point
A2 at a distance a from B. Put the slot of the ruler over the cursor at P and attach a cursor to
B. Attach pencils to both A1 and A2 . Let B follow the directrix l. The pencils now draw two
branches of a curve, which we call a conchoid (see 7 Sect. 5.1).
For directrices with polar equation r D f . /, and the pole of the neusis in the origin, the
associated conchoids have a polar equation r D f . / ˙ a.
Consider the following situation: the pole P for a neusis is given, a line segment has
length a, the directrix is a circle with radius b, the captrix is a certain curve (not shown in
. Fig. 2.22). One of the end points A of the line segment is on the circle. Which curve will
the other endpoint B on ŒAP describe? In other words, what is its locus when A moves along
the circle? The intersection of this line with the captrix will solve the problem. In . Fig. 2.22
we can see examples of the locus of B with A. In the figure on the left, the pole of the neusis
P and the centre of the circle O coincide. The locus is a circle centered at P and with radius
R D b a. In the figure on the right, the pole and the centre do not coincide, the locus now is
a closed curve.
Exercise 27
Create an IGS file in which the origin is the pole P of a neusis. Draw a point A on the x-axis, which
is the centre of a circle with radius 5. Select a point B on the circle. Determine points D and E on
PB for which kDBk D kBEk D 3. Determine the loci of D with B and of E with B respectively.
Move A along the x-axis. What do you notice?
. Fig. 2.22 The locus of B with A generated by a neusis in which the directrix is a circle. Left if the pole coincides
with the centre of the circle, right if the pole and the centre do not coincide
Exercise 28
Create an IGS file in which the origin is the pole P of a neusis and in which the directrix is a circle
with radius 5 through the origin and with its centre on the x-axis. Define a slider with variable a.
Let a vary between 0 and three times the radius. Select a point B on the circle. Draw the straight
line PB. Determine points D and E on PB for which kDBk D kBEk D a. Determine the loci of
D with B and E with B respectively. Which kinds of curves are they?
Exercise 29
Determine the polar equations of the loci you found in the previous exercise. Hint: look at the
triangle with ŒBP as one edge and the diameter through P as another. Which kind of triangle is
this? Can you determine kPBk? Remember for the locus of D with B r D kPBk kDBk.
27 3
Compass and straightedge constructions

3.1 Euclid’s Elements
Of all mathematical treatises, none has had such a profound influence as Euclid’s Elements.
We know practically nothing about its author Euclid (Εὐκλείδης – Eukleídēs lived ca. 300
BC). The apocryphal stories told about him deal more with the nature of mathematics than
telling us anything about the man himself. In one story, a student asked him what profit he
could derive from mathematics. To which Euclid replied: “Give the man three obols, for he
wants to make a profit from everything.” In another story, King Ptolemy asked whether there
was a faster way of learning mathematics, to which Euclid replied: “Sire, in mathematics there
is no Royal Road”.
Euclid’s Elements is arguably the most important treatise of Greek Antiquity. It is divided
into 13 books (which we would call chapters), and sets out 120 definitions, 372 theorems, 93
problems, 19 porisms (corollaries) and 16 lemma’s. Euclid was not the first author to have
edited such an encyclopaedic work, yet his became the standard. Book I culminates in theo-
rems I.47 and 48, which deal with Pythagoras’ theorem and its converse.
Over the ages, numerous editions of Euclid have been published. In the school book edition
it survived well into the nineteenth century. Parts of our geometry course at middle school and
the lower grades of secondary education still deal with Euclid’s geometry, especially the first
book of Elements (see . Fig. 3.1).
Elements starts out with axioms and postulates. Postulates are the rules by which new
mathematical theorems can be derived. The Ancient Greeks made a distinction between ax-
ioms, which are universal truths, and postulates, which are basic truths for a particular science.
In geometry these postulates are:
1. It is possible to draw a straight line between two points (line segment between two points)
2. It is possible to extend any line segment into a straight line as far as one wants (line through
two points)
3. With a point as a centre and a radius it is possible to draw a circle (circle with centre and
radius)
4. All right angles are equal
28 Chapter 3 Compass and straightedge constructions
. Fig. 3.1 The proof of Pythagoras’ theorem in Byrne’s Euclid (1847). Byrne’s edition of the first six books of Eu-
clid’s Elements was an unusual, very appealing, as well as an educationally innovative book. Using coloured pictures
he attempted to present Euclid’s proofs with as little text as possible. (Byrne (1847), Wikimedia Commons, public
domain)
5. If a straight line is intersected by two other straight lines and if the sum of the interior
angles is less than a straight angle then the two lines will intersect.
The first three postulates describe the basic constructions: drawing a straight line and drawing
a circle. With these operations, all other constructions in Elements are performed. They also
impose a restriction on a construction: only compass and straightedge are allowed as aids
in the construction process. In Euclid’s view, a compass collapses when it is lifted from the
paper. As a consequence, we cannot transfer lengths with it. Fortunately, it can be proved
that all constructions that are performed with a collapsible compass can be performed with
a modern compass. In fact, this is what Euclid’s second proposition in Book I amounts to:
“To place at a given point (as an extremity) a straight line equal to a given straight line”1 .
Therefore we will not distinguish between these compasses.
The fourth postulate is not a construction but simply states the equality of all right angles.
This postulate is necessary because it can not be proved with the previous postulates2. Yet
Euclid needs this assertion for some proofs, e.g. of Pythagoras’ theorem.
1
Heath (1956) I, p. 244.
2
The actual study of the independence of axioms of one another and of the consistency of the set of axioms did not
start until the twentieth century. It uses the concept of propositional functions, which is outside the scope of this book.
See e.g. Eves (1995), p. 257 ff.
3.1 Euclid’s Elements
29 3
. Fig. 3.2 Playfair’s axiom. m and l cannot be both parallel to BC and intersect at A
The fifth and last postulate, also known as the parallel postulate has a long history, which
merits a book in its own right. Unlike the first four postulates, the parallel postulate is not
self-evident and has the appearance of a theorem. Moreover proposition I.17 (In any triangle
the sum of any two angles is less than two right angles) is the converse of the fifth postulate.
Over the ages many attempts were made to prove the parallel postulate using Euclid’s first four
postulates. Invariably some property, which turned out to be equivalent to the fifth postulate,
was assumed. Ultimately the search for a proof of the fifth postulate gave rise to the discovery
by Saccheri, Lobachevsky and Bolyai of what is now called non-Euclidean geometry3 .
We limit ourselves to Playfair’s modern formulation of the fifth postulate: In a plane, given
a straight line and a point not on it, at most one straight line parallel to the given line can
be drawn through the point. In fact in his Elements of Geometry (1795) Playfair claimed that
Two straight lines cannot be drawn through the same point, parallel to the same straight line
without coinciding with one another. In the second edition this had become Two straight lines
which intersect one another, cannot be both parallel to the same straight line4 . John Playfair
(1748–1819) was a Scottish scientist, mathematician and professor of natural philosophy at
the University of Edinburgh. He is now best remembered for his popularising of geologic
theories.
Playfair’s postulate in our formulation actually is a construction in Elements: To draw a
straight line through a given point parallel to a given straight line (book I, proposition 31).
To prove this construction, Euclid does not need the parallel postulate. However, the parallel
line constructed in this proposition is the only one passing through the given point and this is
a consequence of the parallel postulate. We use . Fig. 3.2 to see why. Suppose there are two
lines, m and l, through A which are parallel to BC . Then for either m or l and BC , the sum
of the interior angles on one side or the other of AD would be less than two right angles, and
therefore by the parallel postulate either m or l would meet BC . But this contradicts the fact
that we assumed both m and l were parallel to BC .
3
See e.g. Eves (1995), p. 219 ff. Non-Euclidean geometry is a geometry in which Euclid’s fifth postulate is replaced by
another postulate in which either at least two parallel lines through a given point are allowed or none at all are allowed.
For a didactical introduction see van Gulik-Gulikers (2005).
4
Playfair (1795), p. 7, Playfair (1814), p. 22.
. Fig. 3.3 Euclid I.1, the construc-

tion of an equilateral triangle
3.2 Compass and straightedge constructions
Euclid’s first theorem calls for a compass and straightedge construction (see . Fig. 3.3):
» Construct an equilateral triangle on a given line segment
In Euclid’s version the solution reads5 :
» Let AB be the given finite straight line.

Thus it is required to construct an equilateral triangle on the straight line AB.
With centre A and distance AB let the circle BCD be described
Again, with centre B and distance BA let the circle ACE be described;
and from the point C , in which the circles cut one another, to the points A, B let the
straight lines CA, CB be joined.
Now, since the point A is the centre of the circle CDB, AC is equal to AB
Again, since the point B is the centre of the circle CAE, BC is equal to BA
But CA was also proved to be equal to AB;
therefore each of the straight lines CA, CB is equal to AB.
And things which are equal to the same thing are also equal to one another;
therefore CA is also equal to CB
Therefore the three straight lines CA, AB, BC are equal to one another.
The construction can be carried out like this:
» Let the line segment be ŒAB

Position the compass point in A and set the width to kABk
Draw a circle
Now with the same width draw a circle centered at B
Let the intersections of the circles be C1 and C2 6
Either C1 or C2 can be chosen as the third vertex of the desired equilateral triangle.
Because we have kept the compass width at kABk it follows that kABk D kBC k D kCAk
and so 4ABC is equilateral.
5
Heath (1956), p. 240.
6
Although we assume, as Euclid did, that these circles intersect, this does not follow from Euclid’s postulates! Euclid
would have needed a postulate, similar to the parallel postulate, about the intersections of circles to prove this assertion.
In modern geometry the existence of these points is guaranteed by some sort of continuity postulate. See e.g. Eves
(1995), p. 259 ff.
31 3
. Fig. 3.4 The construction of a
line parallel to a given one
. Fig. 3.5 Proof of the construc-

tion of a line parallel to a given one
We can also construct a straight line parallel to a given line (see . Fig. 3.4).
» Select a point B on a given straight line a

Open the compass to a chosen width
Position the compass point in B and draw arcs which intersect at A and C
With the previous construction construct equilateral triangles 4AD1 B and 4BD2 C ,
D1 and D2 being on the same side of a
The straight line D1 D2 is parallel to a.
We have now described a procedure with which we can construct a parallel line. Is the line
really a parallel straight line? In mathematics we have to prove our assertions, therefore we
have to prove that the construction procedure is actually correct. Even though the construction
may be simple, this is not necessarily true for the proof.
For this construction we give two proofs (see . Fig. 3.5):
» Consider the triangle 4D1 BD2

From the construction it follows that this triangle is isosceles, because kBD1 k D kBD2 k
So †BD1 D2 D †D1 D2 B
and †BD1 D2 C †D1 D2 B C †D2 BD1 is a straight angle (sum of angles of a triangle)
or 2†BD1 D2 C †D2 BD1 is a straight angle (4)
On the other hand it follows from the construction that 4ABD1 and 4BCD2 are
equilateral triangles
So †ABD1 D †D2 BC ,
and †ABD1 C †D1 BD2 C †D2 BC is a straight angle
and so is 2†ABD1 C †D1 BD2 (5)
If two straight lines are intersected by a third straight line and the corresponding angles
of intersection with the transversal are congruent, then the two lines are parallel
From (4) and (5) it follows that †BD1 D2 D †ABD1
These are corresponding angles of intersection
So the straight lines are parallel.
An alternative proof is of a more algebraic nature:
» Suppose the magnitude of a straight angle is a

From the construction it follows that 4ABD1 and 4BCD2 are equilateral
1
From which †ABD1 D †BD1 A D †D1 AB D a
3
1
and †BCD2 D †CD2 B D †D2 BC D a
3
1
from which †D1 BD2 equals a
3
3 Now 4BD1 D2 is an isosceles triangle
from which †BD1 D2 D †BD2 D1
The sum of the angles of a triangle is a straight angle so
†BD1 D2 C †D1 BD2 C †BD2 D1 D a
1
or 2†BD1 D2 C a D a
3
1
from which †BD1 D2 D a
3
It is clear that the corresponding angles of intersection are equal, again proving the
correctness of the construction.
Exercise 30
In the construction of a parallel line we have used equilateral triangles with sides ŒAB and ŒBC .
Prove that if we use congruent isosceles triangles with base sides ŒAB and ŒBC the third vertices
define a parallel straight line.
Exercise 31
If a straight line AB and a point P are given, construct a straight line through P and parallel to
AB.
Another construction (6) of a straight line is this:
» Let AB be a given line

Draw a straight line l, not coincident with AB, through A
Draw a circle CA centered at A and with radius kABk
CA intersects l at C
Draw a circle CC centered at C and with radius kABk
CC intersects l at D
Draw a circle CD centered at D and with radius kCBk
CD intersects CC at E
Then CE k AB.
Remember that in the following exercises you can only use a compass and straightedge, not a
ruler or a protractor.
Exercise 32
Prove construction (6).
Exercise 33
If a straight line AB and a point P are given, use construction (6) to construct a straight line through
P and parallel to AB.
Exercise 34
Find a construction to divide a line segment into two equal parts.
33 3
. Fig. 3.6 Dividing a line segment
into three equal parts
Exercise 35
Find a construction to erect a perpendicular on ŒAB through a point C on ŒAB.
Exercise 36
Find a construction to erect a perpendicular on ŒAB through a point D not on ŒAB.
Exercise 37
If a straight line AB and a point P are given, use Exercises 35 and 36 to construct a straight line
through P and parallel to AB.
Parallel lines allow us to divide a line segment into any number of equal parts (Thales’
intercept theorem). We will describe the procedure for a division into three equal parts, but
the procedure can readily be generalised for any number (see . Fig. 3.6).
» Suppose we want to divide the line segment ŒAB into three equal parts
Select a point C , not on AB, preferably such that ]CAB is an acute angle
Select a point D1 on ŒAC
D1 can coincide with C , but this is not necessary
With the compass, construct the points D2 and D3
for which kAD1 k D kD1 D2 k D kD2 D3 k
So kAD3 k D 3 kAD1 k
Draw the straight line D3 B
Draw the straight lines parallel to D3 B through D1 and D2
These lines intersect ŒAB in P and Q respectively
and kAP k D kPQk D kQBk.
Exercise 38
1
Use the properties of similar triangles to prove that in the above procedure kAP k D kABk and
3
kAP k D kPQk D kQBk.
In the same fashion we can construct angles which obey particular demands.
For instance:
» An angle and a ray are given, construct, with the ray as one side, an angle which is
congruent to the given angle (see . Fig. 3.7).
. Fig. 3.7 The construction of an angle with a given ray equal to a given angle
» Let the ray be ŒAB and the angle ]KLM

Draw, with L as centre, an arc with radius equal to kLM k
The arc intersects KL at a point N
Draw, with the same radius, an arc centered at A
This arc intersects AB at a point C
Position the compass point in N and determine the length kNM k
Keep this compass width, position the compass point in C and draw an arc
Let the intersection of the arc centered at C and the arc centered at A be D
Then †BAD is equal to †KLM .
This can easily be seen: from the construction it follows that 4LMN and 4ACD are
congruent (SSS).
Exercise 39
Divide, by construction, an angle into two equal parts. Prove the construction!
Using these basic constructions we can find rectilinear figures which have the same area
as a given rectilinear figure. In most cases, we will “reduce” a rectilinear figure to a square by
doing so.
For instance:
» Given a rectangle, construct a square of the same area (see . Fig. 3.8).
» Suppose rectangle ABCD is given

Draw the ray ŒAB
Draw a circle centered at B and with radius kBC k
The circle intersects ŒAB at E, which is not on ŒAB
Determine the midpoint M of ŒAE (see Exercise 34)
Draw a circle centered at M and with radius kMEk
Draw a straight line BC , which intersects the circle at F1 and F2
We will proceed with F1 , but the argument is equally valid for F2
The square with edge ŒBF1 has the same area as the rectangle ABCD
The angle ]AF1 E is subtended by a diameter and is therefore a right angle
It is easy to show that 4ABF1 ; 4F1 BE and 4AF1 E are similar triangles
35 3
. Fig. 3.8 Construction of a
square equiareal to a given rect-
angle
kABk kF1 Bk
So we can deduce D ,
kF1 Bk kBEk
and kF1 Bk2 D kABk kBEk D kABk kBC k
(because kBEk D kBC k).
The formula expresses that the area of the square with edge ŒF1 B is equal to the area of
rectangle ABCD. For this reason, the number kF1 Bk is also called the geometric mean or
mean proportional of the two numbers kABk and kBEk.
Exercise 40
Create an IGS file in which you perform this construction. Define a slider a, ranging from 1 to 10.
Let B be the origin. Define a point A.a; 0/ on the negative x-axis. Draw a line, m, through A
and perpendicular to the x-axis. Select a point D.a; d / (d > 0) on m. Construct the rectangle
ADCB. Determine E on the x-axis such that kBEk D kBC k. With ŒAE as diameter, draw the
circle C , which intersects the y-axis at F1 and F2 . Draw the square BF1 GH . Draw the straight lines
DC and GH and determine their intersection point L. Determine the locus of L with D. The part
of the locus in the first quadrant is a parasitic part. The part of the curve in the fourth quadrant is
a part of a well-known curve. Which one? Can you determine the equation of this curve? Now use
Locus to draw the locus of L with B and move the slider a. What do you notice?
There are also other ways to obtain this result. Samuel Marolois (1572–1627), a Dutch
mathematician and military engineer, proceeded in following fashion7 (see . Fig. 3.9):
» Given a rectangle ABCD

Determine the midpoint F of the edge ŒAC
Position the compass point in A and with a compass width kAF k, draw an arc
The arc intersects ŒAB at E
Erect the perpendicular on AB in E
Position the compass point in B and draw an arc, with compass width kBF k
The arc intersects the perpendicular on ŒAB at G
The length of ŒGE is the length of the side of the required square with the same area as
the rectangle ABCD.
7
Marolois (1614–1615), plate 5, van de Weyer (2011), p. 51–54.

a square equiareal to a rectangle
ABCD from Marolois’ La Géome-
trie (1614), EHC G 48942. Marolois’
3 treatise on geometry was part of
his Opera Mathematica (1614) which
also contained one of the most
widely read treatises on perpective
of its day
Exercise 418
Prove Marolois’ construction.
b
Hint: use Pythagoras’ theorem. Put kABk D l, kAC k D b from which kAF k D , ...
2
Exercise 42
Consider two circles CA and CB which do not intersect. Neither circle is enclosed by the other.
Construct all possible straight lines tangent to both circles.
Exercise 43
Use the IGS file of Exercise 18. Let the file show only the circles CA ; CB ; CC with their centres
and the tangent external circle. Define a slider R varying from min .rA ; rB ; rC / to 0. Now re-
define the circles CA ; CB ; CC as having radii rA C R; rB C R; rC C R. Let R diminish from 0 to
min .rA ; rB ; rC /. What do you notice? Now choose a situation in which the smallest circle, say
CA , touches the tangent circle externally, while the others touch internally.
Redefine the circles CA ; CB ; CC as having radii rA C R; rB R; rC R. Let R diminish from 0
to min .rA ; rB ; rC / D rA . What do you notice?
3.3 François Viète and Apollonius’ Problem
The result of Exercise 43 was used by François Viète (1540–1603) to show that there actually
is a compass and straightedge solution to Apollonius’ problem.
François Viète was born in Fontenay-le-Comte in 1540. After an education at the Fran-
ciscan school of Fontenay, he enrolled at the university of Poitiers in 1558 to study civil and
canonical law. It took him just one year to obtain a baccalaureate and a licentiate’s degree,
after which he embarked on a successful career as a lawyer in his hometown.
From 1571 onwards, he held several posts both in Paris and in Brittanny. After a turbulent
period in 1587–88, Viète became a counsellor and cryptographer in the service of Henry IV.
As a cryptographer, he was able to decipher an intercepted letter to Philip II, King of Spain
and pretender to the French throne, which helped Henry thwart Philip’s military plans9 .
8
van de Weyer (2011), p. 51–54.
9
Delahaye (2005).
37 3
Viète introduced a method which, by way of an appropriate substitution, can be used to
solve second, third and fourth-degree equations10.
For the equation
x 2 C bx D c
make the substitution

b b
yDxC or x D y
2 2
then
2 2
b b
y 2 D x 2 C bx C , y2 D c C
2 2
which immediately gives the value of x.

For the third degree equation
x 3 C bx 2 C cx C d D 0;
he puts
b
xDy
3
yielding the equation
y 3 C py C q D 0
A second substitution
p
yDz
3z
which has become the classical substitution for solving third-degree equations, yields a
quadratic equation in z 3
p3
z3 Cq D 0
27z 3
p 3
) z 6 C qz 3 D0
3
from which
r q 3
q p 3
z3 D ˙ C :
2 3 2
Viète only uses the positive cube root of z, but it is easily demonstrated that the six solutions
for z yield three different solutions for x.
10
Meskens (2010), p. 157.
. Fig. 3.10 Viète’s solution to Apollonius’ tangency problem from Opera mathematica (1646). Left for two circles
externally tangent and one internally tangent to the tangent circle, right where one circle is externally tangent and
the other two internally tangent to the tangent circle (Viète (1646), EHC G 4858)
An analogous substitution solves a fourth-degree equation. It goes without saying that

finding these substitutions encouraged Viète and other mathematicians to search for similar
substitutions for higher-degree equations, an undertaking that was, for that matter, doomed to
failure.
Viète found a compass and straightedge construction for Apollonius’ problem (see . Fig.
3.10). Exercise 43 shows that the tangency problem can be reduced to finding a circle tangent
to two given circles and through a point. This problem can be further reduced to finding a
circle tangent to a given circle and through two points11.
Suppose three circles CA , CB and CC are given.
11
Bogomolny (a); Henry.
39 3
Let rC D min.rA ; rB ; rC /. Diminish the radii of the three circles by rC . We now have two
circles CA0 , CB0 and a point C .
Construct the external point of similitude P of circles CA0 and CB0 . The common tangent from
P touches CA0 at D and CB0 at E. Construct the circle CDE. Let C 0 be the second intersection
of CP and the circle CDE. Draw a circle which passes through C and C 0 which intersects
the circle CA at F and G, e.g. the circle CDE, then G D D.
Let H be the intersection of C C 0 and FD.
Construct the tangents HS and H T from H to CA .
Construct the perpendicular bisector of C C 0 . Let K and L be the intersections of the bisector
with AS and AT respectively.
The circles with centres K and L respectively and radii kKSk and kLT k solve the problem.
41 3
To find the tangent circle for the original problem expand the circles again to the given circles
and shrink the tangent circle by the same amount.
We have found the circle to which the three circles CA , CB and CC are externally tangent.
If we use L as the centre for the tangent circle, we again expand the radii of the circles and of
the tangent circle. We have now found the tangent circle to which the circles CA , CB and CC
are internally tangent.
43 4
The Delian Problem

4.1 Delos and its altar
After defeating the Persians at Mycale (479 BC), Athens became the centre of the Greek
political, commercial and intellectual world.
The first philosophical Athenian school was the Sophist school. Alongside grammar and
rhetoric, this school showed an interest in mathematics and in geometry in particular. One of
the problems that caught their attention was the so-called Delian problem.
According to legend, the island of Delos was hit by the plague. The local residents made
offerings to the local oracle and asked what they had to do to stop the epidemic. The oracle
answered that they should build a new cubic altar, with a size (D volume) double that of
the existing one. The length of the edge should be determined by compass and straightedge
methods only.
The inhabitants first created an altar with an edge double that of the original. The plague
did not disappear though. It became clear that to have a double volume the edge should not be
doubled, but multiplied by a number different from 2. But which factor should this be?
Suppose the length of the edge of the original cube is a, then the volume of the new cube
should
p bep2a3 . If the edge of the new cube is b then b 3 D 2a3 . If b D t a then t 3 D 2 or
t D 2. 3 2 is the third root or the cubic root of 2.
3
p The problem of doubling the cube is thus reduced to constructing a line segment which is
3
2 as long as the original edges.
The problem seems easy enough. It is possible to construct a line with a length equal
to any square root of a natural number, so why would this not be possible for a line with a
length equal to a cubic root? The problem proved tougher than expected. A first solution was
proposed by Hippocrates of Chios (fl. ca. 430 BC – not to be confused with Hippocrates of
Kos, after whom the hippocratic oath is named). Hippocrates was a merchant on the isle of
Chios. After suffering various misfortunes, he moved to Athens, where he became a leading
mathematician. He was the first to write a systematically organised geometry textbook called
Elements1 (not to be confused with Euclid’s Elements).
To duplicate a cube, the equation
x 3 D 2a3 (7)
1
Heath (1981) I, p. 182ff.
44 Chapter 4 The Delian Problem
needs to be solved. The equation was reduced by Hippocrates to finding two mean proportion-
als x and y which satisfy
a x y
D D : (8)
x y 2a
4 From these equations, we can deduce two equations
x 2 D ay and y 2 D 2ax: (9)
Note that the problem of solving a third degree cubic equation is reduced to finding two
mean proportionals by solving two quadratic equations. On the one hand we need to find the
geometric mean x of y and a on the other hand we need to find the geometric mean y of x
and 2a. We have already seen that we can construct a geometric mean if we know the two
numbers. This is not the case! In the first equation, we are unable to construct the line segment
with length x because we do not know length y. In the second equation, we are unable to
construct the line segment with length y because we do not know length x.
It seems to be a stalemate.
Exercise 44
Show that (8) can be written as the system of equations given by (9). Eliminate the parameter y and
show that the resulting equation is (7). Create an IGS file in which you define a slider a, ranging
from 0 to 10, and draw the curves which are given by the system of equations (9).
Which curves have you drawn? Determine the coordinates of their intersection(s). (Note that
when using an IGS you will only be given numerical approximations). Algebraically determine the
coordinates of the intersection for any value of a.
Menaechmus (ca. 350 BC) realised that Hippocrates’ problem could be solved using conic
sections. Little is known about Menaechmus, other than that he was a student of Eudoxus and
that his brother was Dinostratus (see 7 Sect. 6.2). He is credited with the discovery of conic
sections and some of their properties2 . Menaechmus saw that Hippocrates’ problem could be
reduced to finding the intersection of two parabolae. Unfortunately, this intersection cannot
be constructed using compass and straightedge alone. It is a construction from Plato’s second
category.
Exercise 45
Show that finding the mean proportionals of two numbers a and b,
a x y
D D
x y b
can be solved by finding the solution to either of the systems of equations

8
< ay D x 2
(10)
:bx D y 2
2
Heath (1981) I, p. 251–253, Coolidge (1945), p. 3–5, Fladt (1967), p. 5–7.
4.1 Delos and its altar
45 4
. Fig. 4.1 Plato’s construction of the mean proportional of a and 2a
or
8
< ay D x 2
(11)
:xy D ab
Which curves do these equations represent?
Other Grecian solutions belong to Plato’s third category. What they have in common is
that they make use of an instrument with moving parts. The following solution is said to be
Plato’s. It was used, in different guises, well into the seventeenth century (see . Fig. 4.1).
» Consider two rods, BD and AC , perpendicular to one another, which intersect at E.

Slide the rods over each other such that kEAk D a and kEDk D 2a. In the top left
figure all lines represent rods. The rods F G and GH are rigidly connected to each
other, so F GH is a carpenter’s square. LK is perpendicular to GH and L can slide
along GH remaining parallel to F G. The cross like rods have to be placed on the other
rods in such a way that A is on F G, D is on LK, the extension of DE passes through
G (B does not have to coincide with G) and the extension of AE passes through L
(again C does not have to coincide with L). This is done by appropriately sliding
p and
rotating the cross, while simultaneously sliding LK. Then kEGk is the length 3 2a of
the required cube.
Exercise 46 p
Use similar triangles to prove that the procedure is correct and that kEGk D 3 2a.
Exercise 473
Gregory of St-Vincent (see 7 Sect. 5.5) found a construction for two mean proportionals using
conic sections. To find the mean proportional between a and b, construct the rectangle ABCD
3
Sancto Vincentio (1647), p. 602.
with sides a and b. Choose a coordinate system for which the origin coincides with A and the
x- and y-axis coincide with the lines AB and AD. Then C has coordinates .a; b/. prove that the
intersection of the orthogonal hyperbola through C and with the coordinate axes as asymptotes
and the circumscribed circle to the rectangle intersect in a point whose coordinates are the mean
proportionals between the adjacent sides of the rectangle.
4
4.2 Wine gauging
The procedure of the previous 7 Sect. 4.1 was used well into the sixteenth century, albeit in a
slightly other version, by wine gaugers.
Wine gaugers were employees of the local tax offices4 . From the thirteenth century on-
ward, many European cities levied an excise on the consumption of wine. The most commonly
used container type for transporting liquids was the barrel. In order to be taxed correctly the
contents of the barrel had to be measured. This was the task of wine gaugers: they measured
the barrel and made the necessary calculations (see . Fig. 4.2). Usually the barrel was approx-
imated by a cylinder. To do so the diameters of the heads and the bulge were averaged. Let this
averaged diameter be d and the length of the barrel l, then the volume of the approximating
cylinder is
2
V D d l:
4
Gaugers could simplify their task by using special gauging rods, such as the cubic gauging
rod. This rod seems to have been primarily used in Austria. The use of this rod is based on
the property that for two similar bodies B1 and B2 , which have known volumes V1 and V2 and
diameters d1 and d2 respectively, the ratio of the volumes is equal to the ratio of the cubes of
the diameters:
V1 d3
D 13 :
V2 d2
To make a measurement, the rod is inserted diagonally through the bung hole towards the
lowest part of a head. With the correct graduation the volume can then be read immedi-
ately.
Suppose that a rod is divided into n equal parts of a certain linear measure, the division
points of which are simply called points. If the rod is introduced into the barrel in the appropri-
ate way and at the bung hole the fifth point can be seen, then the contents will be proportional
to 53 D 125 units of the cube of that linear measure. This means that the verge has to be
graduated with the numbers 1, 8, 27, . . . , n3 indicated
p at the 1st, 2nd, . . . , nth point. Other
3
points, in between the integer third roots such as 2, need to be constructed.
In the seventeenth century, tables of cubic roots were drawn up to simplify the construction
of a cubic gauging rod.
In one sixteenth century manuscript we find the following procedure, closely resembling
Plato’s construction5 (see . Fig. 4.3).
4
On winegauging see Folkerts (1974); Meskens (1994, 2013), p. 97–112, Grabiner (1996, 1998).
5
Coignet (1576–77).
4.2 Wine gauging
47 4
. Fig. 4.2 Left: a wine gauger is taking a measurement with his gauging rod. The man at the front is constructing
a quadratic gauging rod. By using a quadratic scale a gauger could immediately read the surface area of the cross
section at the bung hole. With this kind of rod it was still necessary to multiply this number (or an average based
on it) with the length of the barrel to find the volume (Mennher (1565), MPM A 3589), right: the parts of a barrel,
terminology (drawing by Paul Tytgat)
p
. Fig. 4.3 An instrument with which 3
n can be determined (Plato-instrument, drawing by Paul Tytgat)
. Fig. 4.4 Diagram of parts with which to make a Plato-instrument
» Build an instrument consisting of a carpenter’s square F GH and a slider LK which

slides over GH and is perpendicular to it
Draw a horizontal straight line x and a vertical straight line y, which intersect each
other at O
Select A on x (to the right of O), such that kOAk D a, in which a is the length of the
edge of the original cube
Select D on y (under O) such that kODk D 2a
Put the instrument on the drawing
Make sure that the inner edge of GF always passes through A
Turn and move the instrument and slide the slider LK along GH such that
the inner vertex L in on x,
the inner vertex G is on y,
the inner edge of LK passes through D
p with a pencil and denote them C (on x) and B (on y) respectively
Indicate these points
Then kOBk D 3 2a.
Exercise 48 p
Use similar triangles to prove that the procedure is correct and that kOBk D 3 2a.
Exercise 49
p
Select D on y and under O such that kODk D na. Prove that that kOBk D 3 na.
Exercise 50
Create an IGS file in which the construction of Exercise 9 in 7 Sect. 1.2 is made. Define a slider b
ranging from 0 to 10. Select A.a; 0/ and D.0; 2a/. Determine the Cartesian equation of the locus
of B with C . Determine the coordinates of the intersection with the y-axis. Explain the connection
of this exercise to the wine gaugers’ method described above.
It is easy to make your own Plato instrument by cutting it out of cardboard. Take the
necessary precautions when cutting the cardboard! The parts of the Plato instrument are shown
in . Fig. 4.4. Take a piece of cardboard (not too thick), and draw a carpenter’s square about
3 cm wide and a ruler also 3 cm wide. Towards one end, draw two lines parallel to the edge,
with a length slightly more than 3 cm. Cut the carpenter’s square and the ruler. On the ruler,
4.3 Doubling the cube with a neusis
49 4
p
. Fig. 4.5 Determining 3
2 with
a carpenter’s square only
cut along the lines. You can now slide the ruler over the carpenter’s square. If you have not cut
along the lines too far, the ruler should fit snugly and it should make a right angle with one leg
of the square. You can now solve the problem of the duplication of the cube yourself, using
this instrument.
It is not really necessary to have the whole instrument to solve the problem. In the same
manuscript we find a procedure using only a simple carpenter’s square (see . Fig. 4.5).
» Draw a rectangle ABCD, such that kABk D a, in which a is the length of the edge of
the original cube, and kAC k D 2a
Position the angle of the carpenter’s square on the elongated line segment ŒDC in such
a way that one leg passes through B
Let the intersection of the other leg with the elongated edge ŒBD be G
p until kAFpk D kDGk,
Move and turn the square
from which kAF k D 3 2kABk D 3 2a.
Exercise 51
Prove this construction!
Exercise 52
Create an IGS file in which you perform this construction. Let D be the origin. Define the points
A .2a; a/ and B .2a; 0/. Construct the rectangle ABDC . Choose a point G on the x-axis. De-
termine the point F on AC such that kAC k D kDGk. Draw the line BF and the perpendicular `
through G. Determine the intersection point H of BF and `. Determine the locus of H with G.
p
We can construct a line segment of length 3
2 using a neusis6 (see . Fig. 4.6):
» Draw a regular hexagon ABCDE for which the length of the edge is k. Draw the straight
lines BE and BD.
Use A as the pole of a neusis and by verging fit in a line segment of length k between
the two straight lines BE and BD. p
Suppose this is effected at G on BE, then kAGk D .1 C 2/k.
3
The proof draws on Menelaus’ theorem7 and the power of a point theorem.
6
Based on Delahaye (1997), p. 52.
7
Menelaus’ theorem uses directed line segments. Here, we only set out the version with lengths of line segments.
p
. Fig. 4.6 Constructing 3
2 with neusis
. Fig. 4.7 The power of a point P

with respect to a circle
Exercise 53
Create an IGS file in which you draw a straight line PQ and a circle with radius 5. Consider two
cases, one in which P is inside the circle and one in which it is outside the circle. Determine the
intersection points A and B of line PQ with the circle. Determine h D kPAk kPBk. Now, leaving
P stationary, move Q about. What do you notice for h?
The power of a point theorem states that for any chord going through a point P outside a
circle, the product of the distance from P to the points where the chord intersects the circle
is constant. This constant is called the power of the point P with respect to the circle, and is
written as h. Referring to . Fig. 4.7 this means that h D kPAk kPBk D kP C k kPDk D
kP T k2 . It follows from these relations that h D kOP k2 r 2 , in which r is the radius of the
circle.
Exercise 54
Two chords ŒAB and ŒCD are drawn in a circle with radius r. The line segments ŒAB and ŒCD
intersect at a point P (see . Fig. 4.8 (left)).
Show that 4APD 4CPB.
Deduce that kPAk kPBk D kP C k kPDk.
51 4
. Fig. 4.8 The butterfly theorem (see Exercises 54 and 55)
. Fig. 4.9 Menelaus’ theorem
Exercise 55
From a point P outside a circle with radius r two straight lines which intersect the circle in A, B
and C , D respectively are drawn (see . Fig. 4.8 (right)).
Show that 4APD 4CPB.
Deduce that kPAk kPBk D kP C k kPDk.
Prove that this product is equal to kOP k2 r 2 .
Is this expression equally valid for the situation of the previous exercise?
If a triangle 4AOG is cut by a transversal LBD then according to Menelaus’ theorem
kADk kOBk kGI k

D1
kDOk kBGk kIAk
(see . Fig. 4.9, left), which can also be written as
kAOk kOBk kGI k D kODk kBGk kIAk:
Applied to the triangle 4AOG in . Fig. 4.6, we find
2k k k D kkBGk kIAk:
Letting kBGk D x and kIAk D y, this yields
2k 2 D xy:
On the other hand, if we draw a circle centered at A and with radius k, we can apply the power
of a point theorem.
4 From the fact that the radius is k, we can deduce that O and B are on the circle.
Suppose the line GA intersects the circle at M and N (see . Fig. 4.9, right).
Then
h D kGBk kGOk D kGM k kGN k:
Now
kGBk D x and kGOk D x C k;

kGM k D k C y k D y and kGN k D k C y C k D y C 2k;
from which
x .x C k/ D y .y C 2k/ : (12)
Multiplying both sides by y 2 we find
xy .xy C ky/ D y 3 .y C 2k/ :
But xy D 2k 2 , so

y 3 .y C 2k/ D xy .xy C ky/ D 2k 2 2k 2 C ky D 2k 3 .2k C y/ :
Cancelling out the factor 2k C y on both sides we find that y 3 D 2k 3 or

p3
kLAk D y D 2k:
Exercise 56
Create an IGS file in which you construct a regular hexagon ABCDEF . Choose a point G on the
straight line BE. Draw a straight line AG. Determine points H and I on AG with the property that
kHGk D kGI k D kABk. Determine the locus of H and I respectively with G. The curve you see
is called Nicomedes’ conchoid (see 7 Sect. 5.1). Determine point I1 on BD for which the neusis is
accomplished. Determine the length of ŒAI .
Exercise 57
Show that equation (12) can be rewritten as

k 2 3k 2
.y C k/2 x C D ;
2 4
which is the equation of a hyperbola.

Determine the intersections of this hyperbola with the hyperbola xy D 2k 2 . Create an IGS file
in which k D 1 and determine the intersections graphically.
53 4
. Fig. 4.10 The cissoid of Diocles
Exercise 588
Let C be the circle centered at A.0; a/ and with radius a. Let t be the tangent line at T .0; 2a/. Let
B be a point of t and let C be the intersection of ŒOB and the circle C .
The cissoid of Diocles is the locus of all points P such that kOP k D kCBk (see . Fig. 4.10).
Show that the polar equation of the cissoid is r D 2a.csc sin / and that its Cartesian equation
y3
is x 2 D (hint: write csc of the polar equation as a fraction, then multiply the equation
2a y
with r 2 sin ).
Let Q be the intersection of the cissoid and thepline x C 2y D 4a. Let R be the intersection
p of
OQ and t. Show that the coordinates of R are .2a 3 2; 2a/. Deduce that kTRk D 2a 3 2.
Exercise 59
Create an IGS file in which you construct the cissoid of Diocles.
8
Aarts (2000), p. 194.
55 5
Trisecting an angle

The Greeks knew how to bisect an angle using compass and straightedge constructions. The
question whether an angle can be trisected, or divided into n equal parts, by compass and
straightedge methods then comes naturally.
The question is simple enough and seems to suggest a simple solution. Here too, appear-
ances are deceptive. It turns out that, like the duplication of the cube, the construction with
compass and straightedge is impossible. We can find a construction if we allow a marked ruler
and verging solutions.
5.1 Nicomedes’ conchoid
A solution to the trisection problem was proposed by Nicomedes (ca. 280 BC–210 BC), of
whose life we know nothing. His solution gives rise to a curve known as Nicomedes’ conchoid.
» Suppose the angle ]AOB has to be trisected (see . Fig. 5.1)

Suppose kOBk D a
Draw ŒBC ? ŒOA , draw BD k OA
Now draw a line segment OPQ, with P on ŒBC and Q on ŒBD , such that
kPQk D 2 kOBk D 2a
1
Then †AOQ D †AOB.
3
Exercise 60
Prove Nicomedes’ procedure for trisecting an angle.
Exercise 61
Create an IGS file in which you draw an angle ]AOB. Draw ŒBC ? ŒOA , draw ŒBD k ŒOA .
Now draw a line OE, with E on the line segment ŒBC . Draw a circle centered at E and with
radius 2 kOBk. Determine the intersection F of this circle with OE. Determine the locus of F with
E (see . Fig. 5.2).
What we have done in Exercise 61 is nothing more than a neusis with O as its pole, BD
as the directrix and BC as the captrix. The curve we have found as the locus of F with E is
called Nicomedes’ conchoid (see . Fig. 5.2 top, in which only a very small part of the curve
56 Chapter 5 Trisecting an angle
. Fig. 5.1 pNicomedes’ neusis to

construct 3 2
5 is visible). It actually is a part of one branch of a two-branched curve (see . Fig. 5.2 bottom).
The name “conchoid” derives from the Greek κόγχη meaning “shell”.
We can draw this branch of the curve using a neusis apparatus (see . Fig. 5.3).
This neusis apparatus consists of a T-square and a moving arm. The T-square consists of a
slotted arm K 0 L0 and a perpendicular arm O 0 C 0 with a cursor attached to O 0 . The other part
is a slotted arm G 0 N 0 .
» Draw the angle ]AOB which has to be trisected (see . Fig. 5.4)
Position the arm O 0 over O and turn O 0 C 0 in such a way that it lies over OA
Put a pencil in the slot of K 0 L0 and draw the line KL
Place the slot of G 0 N 0 over O 0 and turn the arm until it is over OB
. Fig. 5.2 Top: part of the conchoid found as locus of F with E (as point on ŒBC ), kEF k D 2 kOBk. Bottom: the
two branches of Nicomedes’ conchoid
5.1 Nicomedes’ conchoid
57 5
. Fig. 5.3 An instrument with
which (a branch of) the conchoid
can be drawn (drawing by Paul
Tytgat)
Call J 0 the intersection of the slots of G 0 N 0 and K 0 L0

Now move G 0 N 0 over O 0 until kG 0 J 0 k D b D 2 kOJ 0 k D 2a
Place the cursor B 0 at M 0 . Attach it to G 0 M 0 , but it should still be able to slide over
K 0 L0
In other words, the length kG 0 B 0 k always equals b when B 0 slides along K 0 L0
The pencil will draw a branch of Nicomedes’ conchoid.
Remove the neusis apparatus
Elongate ŒOB until it intersects the straight line KL at P
Draw the parallel with ŒOC through P
This straight line intersects the conchoid at Q
Draw the ray ŒOQ
1
Then †AOQ D †AOB.
3
c
The polar equation of the conchoid is r D ˙ b in which is the angle with the positive
0 0
sin
x-axis and c D kO C k. In Cartesian coordinates, the equation is .y c/2 x 2 C y 2 D b 2 y 2 .
. Fig. 5.4 Construction of the conchoid with which, for a given angle, the trisection can be accomplished (draw-
ings by Paul Tytgat)
. Fig. 5.5 Trisection of an angle by means of a conchoid
Exercise 62
kO 0 C 0 k
Use the properties of right-angled triangles to show that kO 0 B 0 k D , for any position of B 0
0 0
sin
on K L . Use this result to derive the polar equation of the conchoid.
Exercise 63
Show that the Cartesian equation of the conchoid is

.y c/2 x 2 C y 2 D b 2 y 2 :
Hint: use the transformation formulae from Cartesian to polar coordinates and note that you find
the polar equation.
We can trisect an angle in IGS using the following procedure:

Suppose an angle has magnitude ( rad was chosen in . Fig. 5.5). Select a point A
6
on the negative y-axis and a point B on the positive x-axis for which kOBk D kOAk tan
(making †OAB D ). Select a point C on the x-axis. Draw the straight line AC . Determine
kABk and draw a circle centered at C and with radius 2 kABk. Determine the intersection
points D and E of the circle and AC . Determine the loci of D and E respectively with C .
The two branches of the conchoid become visible. Draw a parallel with the y-axis through B.
Determine the intersection point F of this straight line with the branch of the conchoid defined
by D with C (i.e. the branch which does not pass through A). Unfortunately not all IGSs will
automatically indicate this point. However, in most IGSs, when the cursor is moved over the
intersection point both the straight line and the conchoid will be highlighted. Click for a new
point F at this spot. The angle ]OAF trisects ]OAB.
Exercise 64
Consider the branch of Nicomedes’ conchoid determined by E with C . A is a double point (crun-
ode) of this branch, i.e. the conchoid intersects itself at A. Using the Trace option, you will find that
the conchoid passes through A twice. Prove why!
5.2 Archimedes
59 5
. Fig. 5.6 Pappus’ version of
Nicomedes’ trisection
Pappus (IV.31) reduced Nicomedes’ procedure, which solved the problem by mechanical
means, to a solution which used only conic sections1 . He cleverly uses Thales’ intercept theo-
rem twice. With reference to . Fig. 5.6 draw a line segment ŒCF with lengh 2a and parallel
to ŒPQ, put kOC k D x, kP C k D y, kEQk D c, kBC k D d .
Then
kEQk c kOQk
D D because 4OEQ 4PBQ
kEBk x kOP k
and

kOQk kBC k d
D D because 4OCP 4QBP
kOP k kP C k y
So
x d
D ) xy D cd
c y
which is an orthogonal hyperbola with EO and EB as asymptotes.

The problem of trisecting an angle is thereby reduced to finding the intersection of an
orthogonal hyperbola with asymptotes EO and EB and a circle centered at C and with ra-
dius 2a.
5.2 Archimedes
Archimedes is undoubtedly the most important mathematician we know from Antiquity (287–
211 BC)2 . While Euclid is mainly known for one book, Archimedes’ fame rests on several
smaller treatises on a variety of mathematical subjects. He was born in Syracuse, a Greek
colony on Sicily. It appears that during his youth he went to (or was sent to) Alexandria,
where he seems to have received a thorough mathematical education. He later returned to
Syracuse, where he put himself at the service of the King.
Undoubtedly the most famous story about Archimedes is his discovery of how to deter-
mine the volume of an irregular solid and, from that, how to deduce the volumetric mass
1
Sefrin-Weis (2010), p. 146ff and p. 284ff.
2
On Archimedes see Ver Eecke (1921); Heath (1953) and for a reassessment Jaeger (2008); Paipetis and Ceccarelli
(2010).
. Fig. 5.7 Artistic impression of Archimedes’ claw (drawings by Paul Tytgat)
density of the solid. The problem posed itself when the Tyrant of Syracuse, King Hieron,
asked Archimedes to determine whether the crown he had been given by the goldsmith was
indeed made of solid gold. When Archimedes stepped into his bathtub, he noticed that the
water level was raised. This gave him the idea for what we now call Archimedes’ law. He was
so elated that he ran naked through the streets shouting: “I have found it!” (Eureka!).
During Archimedes’ lifetime, he also witnessed the Second Punic War, in which Syracuse
had sided with Carthage. The city was besieged by a Roman army under the command of
Marcellus. Archimedes became the symbol of the heroic, if futile, resistance against Roman
might. Popular lore credits Archimedes with the invention of the parabolic mirror with which
he managed to destroy at least part of the Roman fleet. Parabolic mirrors concentrate the
sunlight in one point, the focus3 . Roman ships coming in the range of these mirrors were
targeted on their sails, which are said to have caught fire.
Ships managing to sail close to the city walls were caught by Archimedes’ claw (prob-
ably a kind of lever, see . Fig. 5.7). The besieging army equally met with the effects of
Archimedes’ infernal machines.
How unlikely the story may be, it sheds light on the image of a mathematician to his
contemporaries. Well into the sixteenth century, mathematicians simultaneously served as
mathematicians, engineers, physicists and astronomers, and sometimes even as physicians.
The Romans held Archimedes in the greatest esteem. When Syracuse ultimately fell, Mar-
cellus ordered his troops to spare the life of Archimedes. Most probably he wanted to put
Archimedes’ formidable genius to good use at the service of the Roman army. However, when
a soldier found Archimedes, he was summoned to identify himself. Archimedes was too pen-
3
In German Brennpunkt, in Dutch brandpunt, literally burning point.
5.3 The first Archimedean trisection
61 5
sive about a mathematical problem and did not notice the soldier. The soldier slew Archimedes
who uttered his famous last words: “Do not touch my circles” (“Nōlı̄ turbāre circulōs meōs!”,
in Ancient Greek Μή μου τοὺς κύκλους τάραττε! M´ē mou toùs kúklous táratte)4 .
There are other versions of this story, including one in which Archimedes is on his way to
meet Marcellus with some of his contraptions.
Note the striking contradiction between a man organising, or at least making major contri-
butions to, the defence of the city on the one hand and a pensive man trying to prove a theorem
while soldiers are ransacking the city5 .
5.3 The first Archimedean trisection
» Suppose we have to trisect ]AOB (see . Fig. 5.8)

Draw a circle centered at O and with radius r, for which A and B are on the circle
Elongate the line segment ŒAO in the direction of A to O
Draw a line through B, which intersects the circle at C and AO at D, such that
kCDk D r
1
Then †ADB D †AOB.
3
Using the procedure described in 7 Sect. 3.2 (. Fig. 3.7) we can transfer the angle
]ADB to an angle having ŒOA as a leg
Thereby we have trisected the angle ]AOB.
Exercise 65
Use isosceles triangles to prove Archimedes’ procedure.
We can find the points C and D using a neusis instrument called the trisector (see . Fig.
5.9). It consists of two slotted arms and one short ruler. Both the slotted arms and the ruler
have two holes at the same distance r. The short ruler is mounted to both long arms with a
hinge on which it turns. The distance from the hinge to the cursor on the moving arm is equal
to the length of the short ruler. One end of a long arm is attached to a slot in the other arm, the
directing arm. This end can thus be moved along the arm.
To trisect an angle, we have to draw a semicircle of a radius equal to the distance between
the hinges. With one leg on the diameter of the semicircle and with O as vertex, draw the
angle you wish to trisect. Put the directing arm on the diameter. Move the other arm in such a
way that it passes through the intersection of the second leg of the angle and the circle.
The angle made by the two slotted arms is the desired trisector of the given angle.
The neusis we have performed with the trisector can of course be described geometrically.
Again consider the angle ]AOB which has to be trisected. Suppose A and B are on a circle
C centered at O and with radius r. Select B as the pole, the circle as the directrix and the
straight line AO as the captrix. Let the – other – intersection point of the line and the circle
be C . Determine the point D outside the circle, on BC , for which kCDk D r. Determine the
locus of D with C . We find a part of Pascal’s limaçon (see 7 Sect. 1.3 and . Fig. 5.10).
1
If the locus intersects the straight line OA at D0 , then †AD0 B D AOB.
3
4
Voza (2010).
5
We refer the reader to Jaeger (2008) for a very enlightening interpretation of these stories in the light of Roman
literary styles.
. Fig. 5.8 Archimedes’ neusis for a

trisection
Exercise 66
Create an IGS file in which you perform this neusis. Draw a circle with radius r, e.g. 5. Call the
intersection with the positive x-axis A. Determine the point B on the circle for which †BOA D .
Select a point C on the circle and draw the straight line BC . Draw a circle centered at C and with
radius r. Determine the intersections D and E of the circle and the straight line BC . You can now
determine the loci of D and E respectively with C . Let C move along the circle, you will meet
the situations in which D is on the negative x-axis or E is on the positive x-axis or D is on the
positive x-axis. In the first position measure the angle ]AD0 C , in the second the angle ]AE0 C
(if it exists) and in the last the angle ]AD1 C (remember angles are measured counterclockwise).
Reduce the magnitudes of these angles to the interval Œ0; 2 rad. What do you notice? E does not
exist for certain positions of B, which ones?
. Fig. 5.9 A trisection instrument based on Archimedes’ neusis (drawing by Paul Tytgat)
. Fig. 5.10 Left: the part of Pascal’s limaçon found as the locus of D with C . Right: the whole curve is found by
determining the locus of D and E resp. with C (with E being the other point on BC for which kCF k D r).
One can find the whole limaçon as the locus of D with C if one allows C to go around the circle twice
5.4 The Archimedean spiral
63 5
. Fig. 5.11 Archimedes’ trisection

using a conchoid for an angle of
3

rad (top) and for an angle of rad
6
(bottom)
Of course, as we have already noticed, the roles of the directrix and the captrix can be
interchanged. Select B as the pole and the line AO as directrix. Let the intersection of a line
through B and AO be D. Determine the point C on BD for which kDC k D r, then the locus
of C with D is a branch of Nicomedes’ conchoid (see 7 Sect. 5.1).
1
Let C0 be the point where the conchoid intersects the circle then †ADC0 D †AOB.
3
Exercise 67
Create an IGS file in which you perform this neusis. Draw a circle with radius r, e.g. 5. Call the
intersection with the positive x-axis A. Determine the point B on the circle for which †BOA D .
Select a point D on the straight line OA and draw the straight line BD. Draw a circle centered at
D and with radius r. Determine the intersections C and E with the BD. Determine the locus of
C and E respectively with D. Determine the intersections of this locus with the circle. You will at
most find five intersections, one of which is B. Determine †CDA or †EDA at C0; C1; E0 and E1
respectively. Determine †ADC0 , †ADC1 , †ADE0 and †ADE1 and multiply by 3. Reduce the
magnitudes of these angles to the interval Œ0; 2 rad. What do you notice?
5.4 The Archimedean spiral
A second Archimedean solution makes use of a curve which we call the Archimedean spiral
(. Fig. 5.12). The Archimedean spiral has an equation which reads in polar coordinates r D
k; k 2 R0 .
We cannot construct Archimedes’ spiral using compass and straightedge. What we can
do, like we have done for the parabola, is construct individual points (see . Fig. 5.12). Draw
a radius of length R on the polar axis. Construct, using compass and straightedge methods,
3 15
rays with initial point O which make an angle of ; ; ;:::; rad respectively with
8 4 8 8
the polar axis.
. Fig. 5.12 Construction of an Archimedean spiral

On the ray which makes an angle i rad with the polar axis draw a line segment with one
8
R
endpoint being O and having length i , i 2 f1; : : : ; 15g. The endpoints of the line segments
16
R
which are constructed thus are points of the Archimedean spiral for which k D . Start
2
out at O and draw a continuous line through these points, this line is an approximation of the
Archimedean spiral.
Exercise 68
3 15
Find a procedure to construct angles with magnitudes ; ; ;:::; rad.
8 4 8 8
To divide an angle into three equal parts using Archimedes’ spiral we use the following
procedure.
Consider the Archimedean spiral r D (i.e. k D 1). In particular consider the first turn,
i.e. 0 2 rad. Put one leg of the angle you want to trisect on the polar axis, with the
vertex in the pole. Let this angle be ]AOB, with ŒOA the polar axis. The ray ŒOB intersects
the spiral at C . This means that kOC k D †AOB (because r D ).
kOC k
Use the construction of 7 Sect. 3.2, . Fig. 3.6, to find a line segment with length .
3
kOC k
Put the compass point in O and draw a circle with radius . Let the intersection with
3
the spiral be D.
Then kODk D †AOD (because r D ).
kOC k †AOB
Now †AOD D kODk D D , which completes the construction of the
3 3
trisector.
This is an elegant and easy solution, but again not a compass and straightedge construction.
We can determine individual points of the spiral with compass and straightedge, but we cannot
construct the spiral using compass and straightedge alone.
With a little imagination we can see this construction as a neusis construction. We fit in a
line with a given length between a point, the pole of the coordinate system, and a curve, the
spiral. We can consider the point as a circle with radius 0. The spiral acts as the captrix, the
point is both pole and (degenerate) directrix. The curve which is traced out by the other end
point of the line segment is the circle, which we have drawn in the procedure above.
5.5 The Flemish Jesuits
65 5
Exercise 69

Draw an Archimedean spiral, or enlarge . Fig. 5.12 and divide an angle of rad and rad into
4 3
three equal parts. Try the procedure with an angle chosen at random.
Exercise 70

Create an IGS file in which you draw an Archimedean spiral. Divide an angle of rad and rad
4 3
into three equal parts. Try the procedure with an angle chosen at random.
Exercise 71
In the previous paragraph and in the previous exercises we have trisected an angle ]AOB using
the first turn of the spiral. Use the IGS file of the previous exercise. Determine the intersections of
ŒOB and the second and third turns of the spiral respectively. Perform the trisection procedure with
these points. Use the option Angle to measure the magnitude of the given angle and the constructed
angles. Multiply these last results with three and reduce these products to the interval Œ0; 2 rad.
What do you notice?
The Jesuits were an important part of intellectual life in early-seventeenth century Antwerp6 .
The Jesuit College of Antwerp was founded in 1574 and flourished by the beginning of the
seventeenth century. In 1615 plans for creating a school of mathematics were put into practice.
Franciscus de Aguilon was to be its first mathematics professor. By the end of 1615 Gregorius
a Sancto Vincentio or Gregory of St-Vincent (1584–1667), a student of Christopher Clavius,
arrived at Antwerp to help de Aguilon write the curriculum (see . Fig. 5.14). By the end of
1617 the school of mathematics had opened. De Aguilon unfortunately did not live to see it
open its doors.
Franciscus de Aguilon was born in Brussels in January 1567, son of Pedro, a secretary
to Philip II. Franciscus studied at the Colleges of Paris and Douai and became a novice at
Kortrijk. On 15 September 1588 he joined the Jesuit order and became a priest in 1596.
De Aguilon turned out to be not only an able mathematician but also a gifted architect.
In 1615 the first stone of the Ignatius Church, now known as Carolus Borromeus Church,
designed by de Aguilon was laid (see . Fig. 5.13). The impressive and richly decorated church
would be finished a mere seven years later. It was not the first church Franciscus de Aguilon
had built; he had already co-designed the Jesuit churches of Mons and of Tournai.
Gregory was born on September 8, 1584 in Bruges. He attended secondary school in
Bruges, studied philosophy in Douai and joined the Society of Jesus on October 21, 1605
at the Santo Andrea noviciate in Rome. After two years of noviciate, the Jesuit General had
a post in Sicily in mind for Gregory, but Christopher Clavius managed to have him stay in
Rome to study mathematics. On March 23, 1613 he was ordained as a priest in Leuven and
went on to hold several posts in Brussels, ’s Hertogenbosch and Kortrijk. From 1617 to 1621,
he was a mathematics teacher at the Antwerp Jesuit College.
To Gregory we owe a generalisation of Pythagoras’ theorem7 . In an acute-angled triangle,
erect the squares on the sides, draw the altitudes from each vertex and extend these into the
6
On the Jesuit mathematics school in Antwerp see Meskens (1997).
7
Sancto Vincentio (1647), Heath (1956), I p. 404, Ostermann and Wanner (2012), p. 351.
. Fig. 5.13 Antwerp’s Carolus Borromeus Church, designed by Franciscus de Aguilon s.j. and Pieter Huyssens s.j.
(Photos Ad Meskens)
. Fig. 5.14 Portrait of Gregory of

St-Vincent (Sancto Vincentio (1647),
EHC G 4869)
squares. These altitudes divide the squares into two rectangles. If you shade these rectangles
as seen in . Fig. 5.15, then the regions in the same shade of grey have the same area.
Exercise 72
Prove Gregory’s generalisation of Pythagoras’ theorem.
In 1613, Jan Moretus’ widow and sons published de Aguilon’s magnum opus, Opticorum
Libri Sex (see . Fig. 5.16). Despite the fact that the book is voluminous, it does not treat
the whole of optics, but just those phenomena which are related to direct rays of light: no
reflection, no refraction. The book was to be the first of a three volume series. de Aguilon’s
untimely death prevented him from finishing this work. He had started work on his other
books. Among Gregory of St Vincent’s manuscripts preserved in the Royal Library Albert I
in Brussels is a manuscript by de Aguilon with preparatory writings for his other books.
67 5
. Fig. 5.15 Gregory of St Vincent’s generalisation of Pythagoras’ theorem (right Sancto Vincentio (1647), EHC G
4869)
. Fig. 5.16 Frontispiece to the

second book of de Aguilon’s Op-
ticorum Libri Sex. In this book de
Aguilon defines the horopter plane
to explain stereoscopic vision. The
frontispieces were designed by Pe-
ter Paul Rubens and executed by
Theodore Galle (Aguilonius (1613),
EHC G 5050)
In the first part of this manuscript de Aguilon deals with trisections of an angle. In the
following paragraphs we will see some examples of de Aguilon’s methods8.
Exercise 73
Create an IGS file in which you draw a circle centered at O and with radius 5. Let A be the intersec-
tion of the negative x-axis with the circle. Let B be the point on the circle for which †AOB D .
Define a point C on the y-axis. Draw the straight line BC and determine a point D on the
circle for which kCDk D 5. Alternatively, determine a point P on BC , not on ŒCB , for which
kCP k D 5 and determine its locus with C . The intersection of the locus and the circle gives you
the desired point D.
Extend BC until it intersects the x-axis at E. Measure †AOB and †AEB. What do you
notice?
In the previous exercise we have found another way of trisecting an angle using a neusis
procedure9 . The proof is both easy and surprising.
8
Based on van Looy (1979), p. 67–76.
9
This method can already be found in Ibn-al-Haytham’s Optica (1011–1021). Ibn-al-Haytham (ca. 965 – ca. 1040) is
also known as Alhazen. See Hogendijk (1979).
. Fig. 5.17 de Aguilon’s neusis for the trisection of an angle
. Fig. 5.18 The conchoid associ-

ated with de Aguilon’s method in
solid line, the conchoid associated
with Archimedes’ method in dashed
line
Let C be a circle centered at O and with radius r. Keep all the names of points as shown
in . Fig. 5.17. To prove that this neusis construction is correct, we will prove that kDEk D r.
This means that if kCDk D r, then so is kDEk.

Let †OCE D ˛ , then †CEO D ˛.
2
4ODC is isosceles, so †COD D ˛ and †DOE D ˛.
2
Therefore 4ODE is isosceles and kDEk D kODk D r.
But we know that kCDk D r, so kDEk also equals r.
De Aguilon’s method is nothing other than Archimedes’ solution (see 7 Sect. 5.3) in a
different guise. Figure . Fig. 5.18 shows the conchoids resulting from the neusis using de
Aguilon’s and Archimedes’s method.
In another lemma, de Aguilon uses another neusislike procedure. This time, two line seg-
ments need to be of equal length, which is unknown at the start.
Let O be the centre of a circle C .
Let F be the point from which a straight line a is drawn.
69 5
. Fig. 5.19 Another solution by de Aguilon for the trisection of an angle
This line intersects the circle at B and the line AO at E.

Then it is possible to rotate the line a into such a position that kBEk D kEOk.
De Aguilon uses this result to trisect an angle. Suppose an angle ]KAC has to be trisected.
Draw a circle with diameter ŒAC which intersects AK at F . Now find the intersection E of
AC and a line through F for which kBEk D kEOk.
1
Draw BO, which intersects the circle at G, then †CAG D †CAK.
3
Exercise 74
Prove that the above procedure indeed yields the trisection of an angle, i.e. that if kBEk D kEOk
1
then †CAG D †CAK.
3
In Gregory’s manuscripts kept at the Royal Library, we find a proposition which indicates
that he had come to grips with what we would call the limit of a geometric series10 .
Let ŒAB be a line segment one wants to divide into three equal parts.
According to Gregory11 :
10
van Looy (1979), p. 109.
11
Sancto Vincentio (1647), p. 111–112.
. Fig. 5.20 Trisection of an angle by infinitely many consecutive bisections
» If one takes away from ŒAB one half ŒAC , from what remains ŒCB again one half
ŒDB, from what remains ŒDC again one half ŒCE, from what remains ŒED again
ŒDF , from ŒEF again one half ŒEG and so on, then I claim that the end of this
sequence will be where ŒAB is trisected.
Gregory uses this procedure for angle trisection as well. Begin with an angle ]AOC (in
. Fig. 5.20 we used a right angle to be trisected) and construct the interior bisector b1 . For
the angle with sides OA and b1 construct the interior bisector b2 . Now construct the interior
bisector b3 of b1 and b2 , the interior bisector b4 of b2 and b3 , the interior bisector b5 of b3 and
b4 , and so on.
His procedure amounts to

1 1 1 1 1 1 1 1
C ::: D 1 C :::
2 4 8 16 2 2 4 8
n
1
1
1 2
D lim
2 n!1 1
1
2
1 1
D
2 1
1
2
1 2 1
D D
2 3 3
As we will see Gregory was a mathematician able to take bold steps into new realms.
5.6 Hippias of Elis and the quadratrix
71 5
. Fig. 5.21 The quadratrix as the
locus of X. The part of the quadra-
trix outside the circle is called the
parasitic part, these points satisfy
the equation but not the geometric
conditions (here 0 k R)
Hippias of Elis (late 5th century BC) found a method with which the trisection of an angle
as well as the quadrature of the circle could be solved. Hippias seems to have been a talented
man, sent by his fellow citizens on a diplomatic mission to Sparta. He led a peripatetic life
travelling from one city to another. Plato described him as arrogant and vain12 .
Hippias performed the trisection by means of a curve we now call the quadratrix. This
curve is the locus of the intersection of two straight lines which simultaneously perform a cer-
tain movement (one a translation, the other a rotation). The curve itself cannot be constructed
with compass and straightedge methods.
Consider the square ABCD. Let a line segment ŒAF , beginning at ŒAD, rotate with
constant angular velocity about A and at the same time let ŒD 0 C 0 , beginning at ŒDC , move
with constant velocity towards ŒAB in such a way that both reach ŒAB at the same time.
The locus of the intersection points X of ŒAF and ŒD 0 C 0 is the curve we call the quadra-
trix (see . Fig. 5.21).
If X 0 is the perpendicular projection of X on AB, then the equation
†XAB kXX 0 k
D (13)
†DAB kDAk
is always satisfied.
It is clear that if ŒAF rotates about A, then F will trace out a quarter of a circle.
Let R D kAF k D kADk, †BAF D †BAX D and suppose .x; y/ are the coordinates
of X
y 2R
Inserting these values in equation (13) gives D or y D .
=2 R
The quadratrix therefore is the locus of the straight lines with equations y D k .0 k R/
k
and the ray with vertex A making an angle D with ray ŒAB .
R 2
12
Heath (1981), p. 23–24.

points of the quadratrix. Divide the
line segment ŒAD in 2n equal seg-
ments (in the figure 23 D 8). For
each division point, draw the line
segment ŒAi Bi parallel to ŒAB.
Select the angle with ŒAB as one
leg and the first bisectrix as the
second leg as the first angle. The
bisectrix intersects the line segment
5 ŒA1 B1 in K1 , a point of the quadra-
trix. Determine the intersections Ki
with the corresponding line seg-
ments. These points are points of
the quadratrix
To trisect an acute angle ]BAF with the quadratrix we follow this procedure:
» Construct the square ABCD on ŒAB, with ŒAF inside ]BAD

Draw the quadratrix DE and call X the intersection with ŒAF
Through X draw the line segment ŒD 0 X parallel with ŒAB and with D 0 on ŒAD
1
Construct the point H on ŒAD such that kAH k D kAD 0 k
3
Draw ŒH Y parallel to ŒAB and Y on the quadratrix
1
From equation (13) it follows that †BAY D †BAF .
3
Exercise 75
2R
Show that the equation, in polar coordinates, of the quadratrix is r D .
sin
Exercise 76
Create an IGS file in which you use the equations of the respective straight lines and the radii to
draw the quadratrix as the locus of their intersection points. You will notice that the quadratrix, as
drawn by IGS, has a part outside the quarter circle. We call this part the parasitic part. The points
on this part of the curve satisfy the equation of the quadratrix, but they do not satisfy the geometric

condition. Here < .
2
Exercise 77
Create an IGS file in which you construct points of the quadratrix. First draw a square ABCD. Draw
the diagonal ŒAC . Draw the parallel with AB through A1 , the midpoint of ŒAD. The intersection
of this parallel and the diagonal is a point of the quadratrix (see . Fig. 5.22).
Now bisect the angles ]DAC and ]CAB respectively and find the midpoint of ŒA1 D and
ŒAA1 . Repeat this procedure in every step (the number of angles to bisect and midpoints to find
will double with every step).
Exercise 78
The following method was first proposed in nineteenth century France (see . Fig. 5.23).
Draw a line segment ŒPQ and divide it into three equal parts. Place the point of the compass
1
in one of the division points N or M , e.g. M , and draw a semicircle with radius kPQk. Draw
3
73 5
. Fig. 5.23 The tomahawk construction for trisecting an angle
a ray ŒNS , to the side of the semicircle, perpendicular to ŒPQ at N . The figure we have drawn is
called a tomahawk.
Suppose ]AOB is to be trisected. Place Q on the ray ŒOA , while the semicircle touches ray
ŒOB , in such a way that the perpendicular ŒNS passes through O. The angle ]AON trisects
]AOB.
The tomahawk itself can be constructed with compass and straightedge constructions, but the
trisection itself is performed by rotating and translating. Again this is not a compass and straight-
edge construction.
Prove that the tomahawk construction is correct.
Exercise 79
Show that:
sin 3˛ D 3 sin ˛ 4 sin3 ˛
cos 3˛ D 4 cos3 ˛ 3 cos ˛
4 tan ˛ tan3 ˛
tan 3˛ D
1 tan2 ˛
Exercise 80
Colin McLaurin (1698–1746) was a Scottish mathematician, most famous for his series. He also
found a way of trisecting an angle by means of a curve we now call McLaurin’s trisectrix. This
curve is defined as the locus of the intersections of two moving lines, one, l, through O and another,
a, through a point A.a; 0/ on the x-axis. Both lines rotate about O and A respectively with angular
velocities P and 3P . If the angle of l with the x-axis is , then the angle of a with the x-axis is 3
and the angle between a and l 2.
Prove that the polar equation of McLaurin’s trisectrix is
r a a
D ) r D .4 cos sec /;
sin 3 sin 2 2
deduce its Cartesian equation 2x.x 2 C y 2 / D a.3x 2 y 2 /.

To trisect an angle ˛, draw this angle with the x-axis as one leg and with A as its vertex.
˛
Determine the intersection B with the trisectrix. Draw the line OB. OB makes an angle with the
3
x-axis.
Exercise 81
Create an IGS file in which you draw a circle C centered at O and with radius 5. Let P be a point of
the circle. Let A be a point on the x-axis. Determine the slope m.D tan ˛/: Draw a line lA through
A with slope tan 3˛. Use the result of Exercise 79 to express the slope of the line through A in terms
of m. Determine the intersection Q of OP and lA . Determine the locus of Q with P .
Exercise 82
Giovanni Ceva (1647–1734) was an Italian mathematician best remembered for the theorem named
5 after him. He also found a method for trisecting an angle with a curve now called Ceva’s trisectrix
or Ceva’s cycloid. Consider a Cartesian coordinate system. Let C1 be a circle centered at O and
with radius R. Let B be a point on C1 . Let C2 be a circle centered at B and with radius R. C2
intersects the x-axis at O and D. Let C3 be a circle centered at D and with radius R. C3 intersects
OB a second time in P . When B moves along the circle, P traces out Ceva’s trisectrix. Prove that
the angle which DP makes with the x-axis is thrice that which OB makes with the x-axis. Prove
that the polar equation of Ceva’s trisectrix is r D 3R 4R sin2 D R.1 C 2 cos 2/ and deduce
its Cartesian equation.
Exercise 83
Prove that the following procedure also yields Ceva’s trisectrix. Draw a circle C1 centered at O and
with radius R. Let B be a point on C1 . Draw the perpendicular b to x and through B. Let D be
the mirror image of O with respect to b. Draw the straight line OB. Draw the perpendicular e to
OB and through D. Let P be the mirror image of B with respect to e. Then point P is a point on
Ceva’s trisectrix.
Create an IGS file in which you make this construction and determine the locus of P with B.
Exercise 84
Ceva’s trisectrix is one curve of a class of curves defined as r D 1 C a cos 2, sometimes called
botanic curves. Create an IGS file in which you define a slider a and draw some of these curves.
75 6
Squaring the circle

Constructively transforming a circle into a square is often referred to as squaring the circle.
In this case squaring means to find a square with the same area as a given figure. We still refer
to the square root of a number, meaning we are looking for the length of the edge of a square
with the given number as area.
“Squaring the circle” has become more or less a catch line for something which is im-
possible or unsolvable. Indeed, the problem cannot be solved with compass and straightedge
methods. But as with the other problems, it has a rich history. Although we have already men-
tioned that the quadratrix can be used to find a square with an equal area to a circle, the nature
of this problem is completely different.
We will now explain some solutions that were put forward to address the problem. As
in the previous chapters, we will encounter solutions to the problem which cannot be ac-
complished with compass and straightedge methods, but other geometrical constructions are
possible.
We now know that the area of a circle is given by S D r 2 . Now suppose that a square
p
with edge a has the same area. Then a2 D r 2 , from which a D r. The problem of
p
squaring the circle is thereby reduced to constructing a line segment with length , using
p
compass and straightedge methods. Because is the geometric mean of and 1, it suffices
to construct a line segment with length (see also 7 Sect. 7.1, Construction 3).
6.1 Archimedes’ spiral revisited
In 7 Sect. 5.3, we encountered a curve called Archimedes’ spiral, which we used to trisect an
angle. The same curve also allows us to square the circle (see . Fig. 6.1).
Suppose a circle has radius a. Draw the spiral with polar equation r D a . This spiral

intersects the perpendicular to x and through the pole at a point P , i.e. at r D a .
2
Draw a parallel to the x-axis through this point. Draw the tangents to the circle at the
intersections with the x-axis. These three lines and the x-axis define a rectangle with area

s D 2a a D a2 , which is the area of the given circle. Once we have constructed a
2
rectangle, it is easy to construct an equiareal square (see 7 Sect. 3.2, . Fig. 3.8).
Just like the case of the parabola, we note that only individual points of the spiral, but not
the spiral itself, can be constructed with compass and straightedge methods.
76 Chapter 6 Squaring the circle
. Fig. 6.1 Using the Archimedean

spiral to solve the circle quadrature
6.2 Dinostratus’ quadratrix
In a previous chapter we have encountered a curve called the quadratrix (see 7 Sect. 5.6).
Dinostratus noticed that Hippias’ trisectrix also allowed the circle to be squared, hence the
curve’s name Dinostratus’ quadratrix.
2r 2 kADk
Referring to . Fig. 6.2, the length of the line segment ŒAE equals D .

To rigourously prove this result we need to use calculus, because this is the result of a
limiting process.
If X is on the quadratrix and X 0 is the perpendicular projection of X on AB, then length
kXX 0 k and angle †XAB are equal to zero in the equation
†XAB kXX 0 k
D
†DAB kDAk
when X D X 0 D E.
Exercise 85
2R
Use the polar equation r D of the quadratrix and calculate the values of r for a given value
sin
of R and some very small angles .
Exercise 86
In a quarter circle centered at O and with radius kOSk D 1, an angle =]SOQ is drawn. Draw
the tangent line t to the circle in S. OQ intersects t at R. Through Q draw the perpendicular to
OS. Let the foot be P .
Use
area 4OPQ < area circle sector OSQ < area 4 ORS
to prove that
1
sin cos < < tan and cos < <
sin cos
6.2 Dinostratus’ quadratrix
77 6
. Fig. 6.2 The quadratrix
What happens to the left-hand side and to the right-hand side respectively if D 0 rad? What would

you conclude for ?
sin
Mathematically, we express the result of Exercise 86 as

lim D 1:
!0 sin
This result allows us to interpret the geometry of the quadratrix. We can now say (see . Fig.
6.2) that
2R 2R 2R
kAEk D lim D lim D :
!0 sin !0 sin
Dinostratus came to the same conclusion but on the basis of geometrical considerations. If
you accept this step, you can construct a line segment of length , which is essential for
squaring the circle. Needless to say that we cannot construct E using compass and straightedge
methods.
Suppose we have constructed E as shown in . Fig. 6.2. We will now proceed to construct
p 2
a line segment of length . In the following let R D 1, then kAEk D (see . Fig. 6.3

left).
» Determine the midpoint M of the line segment ŒAE,

1
then kMEk D

Draw a right-angled triangle 4EMF , right-angled in M , and choose kMF k D 1
Elongate the line segment ŒME to point G, for which kM Gk D 1
Draw a parallel straight line to FE through G, this line intersects the line FM at H
Draw a semi-circle of diameter kFH k (see . Fig. 6.3)
Elongate the line segment ŒME. Let the intersection with the semi-circle be K
A square with edge ŒKM will have an area S D .

using the quadratrix
Exercise 87
Prove that 4GMH 4EMF and deduce that
kMH k kGM k kMH k 1
D ) D ) kMH k D :
kMF k kMEk 1 1=
p
Show that kKM k D .
6.3 Hippocrates’ lunes
Hippocrates of Chios (see 7 Sect. 4.1) was one of the first mathematicians who thought he had
found a way to solve the problem of squaring the circle.
Hippocrates thought that if he could determine the area of curvilinear figures consisting of
arcs of circles by reducing them to a rectangle it would also be possible to determine the area
of a circle in the same fashion. This amounts to a compass and straightedge construction.
He succeeded in squaring certain curvilinear figures, now called Hippocrates’ lunes. Con-
sider an area of the plane bounded by two arcs of a circle, as seen in . Fig. 6.4. The shaded
concave regions are called lunes (because of their resemblance to a crescent moon), the con-
vex region is called a lens. It turns out that for five types of lunes whose area we are able to
determine using compass and straightedge methods, none of the lenses are squarable.
Classically, Pythagoras’ theorem reads: “[the area of] the square on the hypotenuse equals
the sum [of the areas] of the squares on the perpendiculars.” Actually it does not matter which
shape the figures on the sides of the triangle have, as long as they are similar Pythagoras’
theorem holds!
. Fig. 6.4 Two lunes (shaded area)

and a lens
79 6
. Fig. 6.5 In all of these cases Pythagoras’ theorem holds: area B C area C D area A (drawing by Paul Tytgat)
. Fig. 6.6 Archimedes’ arbelos
Exercise 88
Prove that Pythagoras’ theorem also holds if you replace square by circle or semi-circle: “[the
area of] the semi-circle on the hypotenuse equals the sum [of the areas] of the semi-circles on the
perpendiculars.”
Exercise 89
Prove that Pythagoras’ theorem also holds if the figures on the sides are similar (see . Fig. 6.5).
Exercise 901
In Greek, a shoemaker’s knife is called an arbelos. The word was used by Archimedes to describe
a plane figure bounded by three semi-circles, with reference to . Fig. 6.6 SCAB , SCAC and SC CB .
Draw the perpendicular to AB through C , intersecting the semi-circle SCAB at D. Then the area of
the arbelos ABCD is equal to the area of the circle with diameter CD.
Exercise 91
Create an IGS file in which you define a line segment ŒAB with C on it. Draw the semi-circles
SCAB , SCAC and SC CB . Construct the perpendicular to AB in C , intersecting the semi-circle SCAB
at D. Draw the circle with diameter ŒCD. Determine its area. Determine the area of the arbelos
using the formula
area arbelos D area SCAB area SCAC area SC CB :
What do you notice?

Determine the perimeters PAB ; PAC and PCB . Compare PAB to PAC C PCB . What do you
notice?
1
van Lamoen (2009).
6
. Fig. 6.7 Hippocrates’ lunes in an isosceles right-angled triangle
. Fig. 6.8 Hippocrates’ lunes in a

right-angled triangle
Exercise 92
Create an IGS file in which you draw a line segment ŒAC , in A erect a perpendicular l to the line
segment ŒAC . Find B on l such that kABk D kAC k. 4ABC is an isoceles right-angled triangle,
right-angled in A. Draw semi-circles on the sides as seen in . Fig. 6.7. Calculate the area of the
shaded parts in . Fig. 6.7. To do so, use:
1
area lune D area SCAB area quarter circle MAB C area 4ABC:
2
Determine the area of 4ABC and compare to 2.area lune/. You can move the vertices around.
What do you notice?
Exercise 93
Consider an isosceles right-angled triangle 4ABC , right-angled in A as in . Fig. 6.7. Use this
result to determine the area of the lune on one of the perpendiculars.
Show that the area of a circular sector with radius r and angle ˛ (expressed in radians) is equal
1
to S D r 2 ˛.
2
Show that the area of sector BMA is equal to the area of the semi-circle BA.
Exercise 94
Create an IGS file in which you draw two perpendicular line segments ŒBA and ŒAC . 4ABC is
a right-angled triangle, right-angled in A. Calculate the area of the shaded parts in . Fig. 6.8. Use
1
area lunes D area SCAB C area SCAC area SCBC C area 4ABC:
2
81 6
. Fig. 6.9 Other properties of lunes in a right-angled triangle
Exercise 95
In a right-angled triangle 4ABC , right-angled in A, draw semi-circles with the sides as diameter
as seen in . Fig. 6.8. Use Exercise 88 to show that the sum of the areas of lunes I and II (shaded
areas) is equal to the area of the triangle 4ABC :
area I C area II D area 4ABC
Exercise 962
In a right-angled triangle 4ABC, right-angled in A, draw semi-circles as in . Fig. 6.9. Prove that
in the left triangle we have
area III area I area II D area 4ABC
and in the right triangle
area I C area II C area III C area IV D area 4ABC:
Exercise 973
Leonardo’s claw is the grey portion of the circle in the middle figure of . Fig. 6.10. To obtain
Leonardo’s Claw, cut out a lens from the big circle. The lens is made up of two equal circular
segments with right angles as central angles. Now cut out a circle, with diameter along the axis of
the lens, which touches the lens externally and the big circle internally. Show that Leonardo’s Claw
is squarable and that its area equals the area of the square in its grasp.
The result of Exercise 93 means that we have calculated the area of a curvilinear figure
by means of a rectilinear figure. Moreover the curves which determine the lune are arcs of
a circle. This seems to open perspectives to do the same for a circle. Hippocrates explored
similar figures hoping to find a way to square the circle.
1
Consider an isosceles trapezoid with kADk D kDC k D kCBk and kADk D p kABk
(see . Fig. 6.12). 3
2
Alsina and Nelsen (2010), p. 141–142.
3
Alsina and Nelsen (2010), p. 157.
. Fig. 6.10 Leonardo’s Claw

6
. Fig. 6.11 Constructing an isosceles trapezoid
This trapezoid can be constructed (see . Fig. 6.11). Consider a line segment ŒAB. Open a
kABk
compass on p . Put the compass point in A and draw a circle C1 . Now place the compass
3
point in B and draw a circle C2 . Determine the midpoint M of ŒAB. Position the compass
kABk
point in M and draw a circle with radius p . Let the intersection points of the circle and
2 3
ŒAB be R and S respectively. Erect perpendiculars to ŒAB in R and S. The intersection
points, on the same side of ŒAB, of these perpendiculars with the circles C1 and C2 respec-
tively determine points C and D of the trapezoid.
Exercise 98
Show that the perpendicular bisectors on AD, DC and CB intersect at one point, O.
The result of Exercise 98 implies that the trapezoid can be circumscribed by a circle cen-
tered at O (see . Fig. 6.12).
Construct the triangle 4ABE which is similar to triangle 4ADO.
Draw a circle centered at E and with radius kABk. Together with an arc of the circum-
scribed circle, this circle, defines a lune AFBCD.
From the similarity 4ABE 4ADO it also follows that:
area 4ABE D 3 area 4ADO
and
area circular segment AB D 3.area circular segment AD/;
because the areas of circular segments are proportional to the squares of their bases.
83 6
. Fig. 6.12 Area circular segment AB D 3.area circular segment AD/
Add the area of the trapezoid to both sides:
area trapezoid ABCD C area circular segment AB

D area trapezoid ABCD C 3.area circular segment AD/
area trapezoid ABCD
D area trapezoid ABCD C 3.area circular segment AD/ area circular segment AB
The right hand side of this equation is nothing else than the area of the lune ADCBF .
Exercise 99
Show that
area circular segment AB D 3.area circular segment AD/
Exercise 100
Show that the area of the lune ADCBF is equal to the area of the kite-like quadrilateral AOBE.
Exercise 1014
If a regular hexagon is inscribed in a circle and six semi-circles are constructed on its edges (see
. Fig. 6.13), then the area of the hexagon equals the area of the six lunes plus the area of a circle
whose diameter is equal to one of the sides of a hexagon.
Hippocrates now considered an isosceles trapezoid with
1
kADk D kDC k D kCBk and kADk D kABk
2
(see . Fig. 6.14). Note that this trapezoid is also the half of a hexagon.
4
Alsina and Nelsen (2010), p. 140.
. Fig. 6.13 Lunes and the regular

hexagon
6
Because kAOk D kDC k and ŒAO k ŒDC Thales’ intercept theorem says that ŒAD k
ŒOC .
On the other hand the line segments ŒAO and ŒDC are parallel to one another as well,
from which kADk D kOC k.
Therefore A, B, C and D are on a circle centered at O and with radius kOAk.
Obviously this means that
area semicircle AD AD 2 1
D D
area semicircle AB AB 2 4
Because of the hypotheses we know that:
area semicircle AD D area semicircle DC
D area semicircle CB
D area semicircle EF
from which:
area semicircle AD C area semicircle DC C area semicircle CB C area semicircle EF
D area semicircle AB
We now subtract the areas indicated by IV, V and VI from both sides and find:
area semicircle EF C 3 area lune III D area trapezoid ABCD
so
area semicircle EF D area trapezoidABCD 3 area lune III (14)
. Fig. 6.14 Hippocrates’ lunes in

an isosceles trapezoid
85 6
Had Hippocrates succeeded in finding the area of lune III using a compass and straightedge
construction, he would also have been able to find the area of the semi-circle AD and thus of
the circle AD.
In the previous problems it was possible to square the lunes. Unfortunately for Hip-
pocrates, he was unable to square these lunes5 . We will return to this subject in 7 Sect. 7.3.
Exercise 102
Use Exercise 101 to prove equation (14).
Exercise 103
Use the formula for the area of a circular sector to determine the area of lune III. Express all angles
in radians!
Hippocrates explored yet another kind of lune, the concave pentagon lune6 . It is a lune
p a concave pentagon for which kCDk D
which canpbe constructed with a neusis. Consider
kDEk D 3a and kABk D kBC k D kEAk D 2a.
Construct a (semi) circle centered at A and with radius kABk. Erect the perpendicular bisector
QP on ŒAB. Hippocrates now uses a neusis to find a certain point E on the circumference of
the semi-circle.
A line through B will intersect PQ at a point D and the semi-circle at another point E. Rotate
this line about B until
p
2 3 2 6
kDEk D kEAk or kDEk D kEAk :
2 2
The line segment ŒDE is an edge of the concave pentagon ABCDE.
5
Postnikov and Shenitzer (transl.) (2000).
6
Shelburne (2008).
Draw a line b parallel to AB through E. Draw a circle C centered at B and with radius
kABk, which intersects b at C (i.e. construct the mirror image of E about PQ).
The vertices of the concave pentagon ABCDE define a lune, for which we will identify
its generating circles.
CBAE is an isosceles trapezoid, which has a circumscribed circle centered at O1 . Draw the
perpendicular bisectors on AE and BC and let their intersection be O1 . Draw the circle cen-
tered at O1 and with radius kO1 Ak. This circle also passes through B, C and E.
4EDC is a triangle, which has a circumscribed circle centered at O2 . Draw the perpen-
dicular bisectors on ED and DC , which intersect at O2 . Draw the circle centered at O2 and
with radius kO2 C k, which also passes through D and E.
p
In the concave pentagon
p AEDCB we have that kCDk D kDEk D 3a and kABk D
kBC k D kEAk D 2a.
Exercise 104
Show that 4ADB and 4EAB are similar.
Put a D kEAk D kABk and x D kDBk.
Show that
r r !
3 11 a
xD C :
2 2 2
Deduce that the neusis which Hippocrates used is not necessary.

87 6
We will now show that we can determine the area of the lune by compass and straightedge
methods.
First determine M , the intersection of AC and BO1 .
Exercise 105
Prove that AC ?BO1 .
It is clear that †EO1 A D †AO1 B D †BO1 C and †EO2 F D †FO2 C .

Note that †BAC D †ABC (base angles of isosceles triangle 4ABC ).
and †BAC D †F CA (alternate angles of parallel lines EC and AB).
Thus †F CD D †BCM .
In 4DF C we have that †FDC D †O2 DC is the complementary angle of †F CD.
In 4CMB we have that †MBC D †O1 BC is the complementary angle of †BCM .
Triangles 4O2 DC and 4O1 BC are isosceles. Furthermore we have proved that
†O1 BC D †O1 CB D †O2 DC D †O2 CD whence †BO1 C D †DO2 C (15)
Exercise 106
Use (15) to show that the areas of the circular sectors O1 CBAE and O2 CDE are equal.
Since the areas of both circular sectors O1 CBAE and O2 CDE are equal, the area of the lune
ABCDE equals the dartlike quadrilateral EO1 CO2 .
sector EO2 C D I C II
sector EO1 C D II C III
) 0 D I III
) I D III
Which proves the assertion.
Exercise 107
Look closely at Jacobus Falco’s figure (. Fig. 6.15 – from Falco (1591)). Prove that the area of
rectangle GM V T equals the area of the curvilinear figure QNPOSR. Prove that the way this
figure is constructed in all cases results in a constructive transformation into a rectilinear figure.
. Fig. 6.15 Transforming curvilinear figures into rectilinear figures kept the minds of many mathematicians occu-
pied over the centuries. The figure shows a sixteenth century example (from Falco (1591))
6.4 Franco of Liège, the demise of mathematics
During the Middle Ages, the study of mathematics was in dire straits. The knowledge of the
Roman agrimensores (surveyors) was still known, but the larger part of the Greek corpus was
forgotten.
Exercise 108
The following method to square the circle was used by the Roman agrimensores: draw two per-
pendicular diameters, divide the radii into four equal parts, elongate the radii with one fourth part.
The endpoints determine an edge of the square. Determine the length of the edge of the square.
Assuming the equality of the areas of the circle and the square holds, what is the value of ?7
An example of how low the study of mathematics had sunk is given by Franco of Liège (?–
ca. 1083) in his treatise on the squaring of the circle8 . Geometry, as it is depicted in Franco’s
treatise, had become an almost experimental science. Theorems involving equal areas are not
proven but are checked, sometimes even by weighing pieces cut out of paper or cloth.
Franco of Liège was one of the students of Fulbert of Chartres (ca. 960–1028). From 1066
onwards, Franco taught at the famous Cathedral School of Liège.
Another student of Fulbert’s, Radolph of Liège, corresponded with Regombold, the rector p
of the Cathedral School of Cologne.
p In one of these letters, the subject is the nature of 2.
While a line segment of length 2 is easily constructed, Radolph comes to the conclusion that
7
Smeur (1968), p. 16.
8
Smeur (1968); Smeur and Folkerts (1976a,b).
6.4 Franco of Liège, the demise of mathematics
89 6
. Fig. 6.16 Franco of Liège’s attempt at the quadrature of the circle
p
2 cannot be written as a number (i.e. as what we call a rational number), although sometimes
7 17
and are used as its value.
5 12
Exercise 109 p p
Suppose that p and q have no common factors. Prove that the assertion 2 D leads to a
q
contradiction. (Hint: a square is an even number if it is a quadruple of another square).
17
It is not unlikely that Franco knew of the correspondence. He shows that is not equal
p 12 p
to 2, but is larger. He succeeds in calculating a better approximation, not only for 2, but
also for other irrational roots. Without a proof, he asserts that these numbers are not rational.
In the following paragraphs, Franco does not prove the circle quadrature as we know it,
but he tries to show that the square root of some numbers are not rational and therefore cannot
be calculated exactly, but can only be constructed as the length of a line segment.
For the ratio of the perimeter of a circle to its diameter (the value of ) he uses 3 71 , without
questioning the correctness of the number. A circle with radius 7 therefore has a circumference
of 44 and an area of 154.
If the perimeter is divided into 44 equal parts, then these parts define circle sectors which
can be rearranged as a rectangle (see . Fig. 6.16).
The rectangle with sides 11 and 14 has the same area, 154, as the circle. To find a square
with an area equal to that of the rectangle one has to transform this rectangle into a square.
This should not be hard (see 7 Sect. 3.2, . Fig. 3.8) but Franco has no knowledge of the
construction of mean proportionals!!
In . Fig.
p 6.17, EH has the same length as the diagonal of a unit square EG, from which
kEH k D 2. AEF C is the square with edge 11, therefore kAEk D 11 and kEBk D 3.
Exercise 110
Find K and I on ŒCD such that kHKk D kBI k D kAH k. Prove that area HBIK p D
area HBDL. Calculate the distance between ŒHK and ŒBI and compare your result to 2.
p
Franco suggests, incorrectly, that the distance between ŒHK and ŒBI is 2. He then
moves the parallellogram KHBI in such a way that ŒBI coincides with ŒAH , which is
6
. Fig. 6.17 Franco of Liège’s attempt at the quadrature of the circle 2
possible because kBI k D kAH k. Transforming this parallellogram into a rectangle, is easy
by cutting away a triangle at one end and adding it to the other
p end.
Franco now has a square (he believes) with edge
11 C 2, which is not too bad an ap-
p p 2
proximation of 154 11 C 2 D 154:11 : : : . This square, again according to Franco,
has an area equal to that of a circle with radius 7.
6.5 Nicolas of Cusa
Cardinal Nicholas of Kues or Nicolas of Cusa (1401–1464), also known as Nicolas Cusanus,
was a German theologian and Prince of the Church9 . He studied in Heidelberg and Padua.
After his graduation, he rejected an offer made by the newly established University of Leuven.
He was involved in negotiations during the Council of Florence of 1439, which sought to
reunite the Eastern Orthodox Church and the Western Roman Catholic Church. In 1448 or
1449, he was made a Cardinal and bishop of Brixen in Tyrol by Pope Nicholas V. Nicolas
of Cusa wrote some deeply mystical works about Christianity. He was not only interested in
theology and philosophy, but equally in mathematics, astronomy and linguistics. His scientific
ideas are scattered across his philosophical treatises. He advocated a reform of the Julian
calendar, an idea which would lead to the Gregorian reform in the sixteenth century. Nicolas
was the first to describe the use of concave lenses to treat myopia.
In trying to solve the circle quadrature, Nicolas implictly put forward an approximation of
. His method is a tedious one, involving trigonometry. Moreover his prose is excursive and
prolix. Cusanus for instance still used a sexagesimal system to denote fractions.
He stated that10 :
» around the midpoint D shall be inscribed a circle EF G and a circle HI circumscribed;

the straight line DE shall be so drawn, that E is the midpoint between A and B; then
DB shall be drawn. Further, a straight line DK shall be drawn from D to the midpoint
9
Meuthen (1982).
10
As cited in Wertz Jr. (2001).
6.5 Nicolas of Cusa
91 6
. Fig. 6.18 Cusanus’ attempt at
the quadrature of the circle
between E and B. I maintain: DK is smaller than the radius of the circle isoperimetric
5
to the triangle, by one-fourth of the length of DK. [i.e. r D kDKk]
4
What he does is this (see . Fig. 6.18):
» Let 4ABC be equilateral (Nicolas does not state this explicitly).

Determine the centroid D of this triangle
Choose a side, e.g. ŒAB and divide it into four equal line segments
Let K denote the division point closest to B
Draw ŒDK
Divide ŒDK into four equal line segments
Elongate ŒDK and draw a circle centered at K and with radius one quarter of kDKk
Let the point L be the intersection of this circle and DK.
Then ŒDL is the radius of a circle with a circumference equal to the circumference of
the triangle.
Cusanus suspects he only gives approximations, as he himself admits: “In mathematical sci-
ence, every proposition which poses the exact equality between a circle and a square is
impossible [=false]”
This statement also makes clear Cusanus’ reasons for his interest in the circle quadrature,
in his search for a proof of God he uses geometrical analogies, e.g.11 :
» Whatever is not truth cannot measure truth precisely. (By comparison, a non-circle
cannot measure a circle, whose being is something indivisible.) Hence, the intellect,
which is not truth, never comprehends truth so precisely that truth cannot be
comprehended infinitely more precisely. For the intellect is to truth as [an inscribed]
polygon is to [the inscribing] circle. The more angles the inscribed polygon has, the
more similar it is to the circle. However, even if the number of its angles is increased ad
infinitum, the polygon never becomes equal [to the circle], unless it is resolved into an
identity with the circle.
11
As cited in Wertz Jr. (2001).
. Fig. 6.19 Construction lines for

Cusanus’ calculation of
Cusanus uses the qualitative difference in the nature of a circle and a polygon to argue that,
although the human mind is created in the image of God, and is therefore finite, it cannot
precisely attain the Truth itself, which is infinite12 . In the paragraphs below, we will only
highlight his mathematics and refer interested readers to philosophical treatises on Nicolas of
Cusa.
For the given configuration (see . Fig. 6.18) Cusanus first shows that the circle with the
same circumference as the perimeter of the triangle 4ABC has a radius smaller than that of
the circumscribed circle, but larger than that of the inscribed circle (see . Fig. 6.19).For the

1
isoperimetric circle, he approximates the ratio of the diameter to the circumference D
p
2 12 1575
by p .
6 2700
He asserts that his approximation cannot be closer “with even a part of a minute” (this is
1
closer than ).
60
Exercise 111
In . Fig. 6.18 put the length of the edges of the triangle equal to 1, kABk D kBC k D kCAk D 1.
5
Use trigonometry to calculate the length of ŒDK and deduce the length of kDLk D kDKk.
4
Use the fact that in an equilateral triangle the centroid, orthocentre and midpoint coincide. Deter-
mine the circumference of the circle with radius kDLk and of 4ABC . What can you conclude?
p
2 12 1575
How did Cusanus arrive at the ratio p ?
6 2700
He put kDBk D 60 whence kDEk D 30. Because 4ABC is equilateral †ABC D 60ı .
DB is the bisectrix, so †ABD D 30ı . In any right-angled triangle with one angle equal to
30ı , one perpendicular is half the hypotenuse.
Exercise 112
If kDBk D 60, calculate kEBk, kDKk and kDLk. The circumference of the triangle equals
6 kEBk, the diameter of the circle is 2 kDLk.
12
Wertz Jr. (2001).
6.6 Archimedes’ approximation
93 6
Exercise 11313
The following procedure for finding a circle whose area was equal to that of a given square was
described in the Indian Salvasutras. In square ABCD, let M be the intersection of the diagonals.
Draw the circle centered at M and with radius kMAk, let kMEk be the radius of the circle perpen-
1
dicular to the side ŒAB and cutting ŒAB in G. Let kGN k D kGEk. Then kMN k is the radius
p 3
2C 2
of the desired circle. Show that kAN k D kABk. Assuming the equality of the areas of the
3
circle and the square holds, what is the value of ?
6.6 Archimedes’ approximation
Archimedes was able to prove that the area of a circle is equal to the area of a right-angled
triangle with perpendiculars equal to the radius and to the circumference of the circle respec-
tively.
Suppose a circle has an area S which is larger than the area T of a right-angled triangle
with perpendiculars equal to the radius and to the circumference of the circle respectively. We
can then find an integer number n for which
S area inscribed 2n -gon < S T
1
The area of a regular inscribed 2n -gon is larger than 1 -part of the circle. It is then clear
2n
that we can find an n-gon with an area larger than the area T of the triangle.
In other words, T < area 2n -gon.
Now suppose that AB is an edge of the inscribed 2n -gon and ON a perpendicular from
the centre of the circle to AB (so N is the middle of AB).
It is clear that ON is smaller than the radius of the circle.
Because the perimeter of the inscribed polygon is smaller than the perimeter of the circle
we find that:

n 1
area 2 -gon D 2
n
kABk kON k
2
1
D .2n kABk/ kON k
2
1
< perimeter C radius D T
2
But this is a contradiction.
Similarly, we are able to prove that a circumscribed polygon exists for which area 2n -
gon < T , but this also leads to a contradiction.
From this, we can conclude that the area of the circle and that of the triangle are equal.
k
If k is the length of the perimeter of a circle with radius 1, then its area is .
4
Now the areas of circles are to each other as the squares of their diameters, so for a circle
with radius r we find that
SC .2r/2
D and SC D kr 2
k=4 12
Archimedes however was unable to calculate the value of k, but he did propose very good
approximations for what we now call .
13
Katz (1998), p. 41.
. Fig. 6.20 Left the inscribed and circumscribed equilateral triangles of a circle, right: the inscribed and circum-
scribed regular hendecagons or 11-gons
So if we know the value of k (D ), we may be able to find a way to construct a square
with the same area as the circle.
To approximate this number k, Archimedes used the fact that Spn < SC < SPn , in which
pn is the regular n-gon inscribed in the circle and Pn the regular n-gon circumscribed about
the circle.
Archimedes was able to calculate the areas of the inscribed and circumscribed 96-gon and
arrived at 3 10 < k < 3 71 or 3:1408 < k < 3:1429.
71 22
These are very good approximations and 3 71 or still survives as a value for in the
7
lower grades.
It is clear that the larger n becomes, the closer the areas of the inscribed and circumscribed
polygons will approximate the area of the circle.
Exercise 114
Create an IGS file in which you define a slider a ranging from 3 to 100 in steps of 1. Draw
a circle centered at A and with radius e.g. 5. Define a point B on the circle with coordinates
2 2
R cos ; R sin .
a a
Determine the intersection points C and D of this circle and the x-axis. Choose C on the
positive x-axis and define a regular polygon with ŒCB as the edge and with a vertices. You will
first see a triangle 4CBE.
Define the tangents in C , B and E respectively and determine the intersections F ,G and H of
the tangents, taken pairwise.
Define a regular polygon with edge ŒF G and a vertices. This is the circumscribed polygon
(see . Fig. 6.20).
Define the numbers
b D .circumference polygon1/=.2radius/ and c D .circumference polygon1/=.2radius/
[alternatively define .area polygon2/=.radius/2 and .area polygon2/=.radius/2 ].

Use the slider to increase the number of edges.
What do you notice for the values of b and c?
6.7 Adriaan van Roomen and Ludolff van Ceulen
95 6
Exercise 115
If pn and Pn denote the perimeters of the regular n-gons inscribed in and circumscribed about the
same circle, and if an and An are their areas, show that
pn an
D cos and D cos2 :
Pn n An n
Archimedes’ method was revived in the late sixteenth century by Adriaan van Roomen (see
7 Sect. 2.3) in his book Ideæ Mathematica (1593)14 and later in In Archimedis (1597), in
which he published a value for to sixteenth decimal places D 3:1415926535897931. He
used Archimedes’ method for regular polygons having 230 edges, by consecutively doubling
the number of vertices of a regular polygon. His result would soon be superseded by the work
of his friend Ludolff van Ceulen (1540–1610).
In August 1593, Josephus Justus Scaliger (1540–1609) was appointed professor at the Uni-
versity of Leiden. The next year he published two books Cyclometrica Elementa duo (1594)
on the squaring of the circle and Mesolabium (1594) on the duplication of the cube. His work
was already being circulated in the form of manuscript copies, in which he claimed to have
solved the three classical geometry problems of Antiquity. For this, he was vehemently criti-
cised by François Viète. p
Scaliger claimed that the ratio of the circumference of the circle to the diameter was 10.
He also criticised Archimedes for using an arithmetical approach to the squaring of the circle.
Following Aristotle’s advice that one “cannot pass from one genus to another”, he was op-
posed to solving a geometric problem by means of arithmetic15 . The first to read his book was
Ludolff van Ceulen, who advised him not to distribute it. Van Roomen read the book in the
autumn of 1594 and in his correspondence he was very critical of Scaliger. He wrote a dev-
astating answer to Scaliger’s claims. In Archimedis was published in 1597. The book contains
the Greek text with a Latin translation of Archimedes’ On the measurement of the circle, an
Apology of Archimedes, refutations of Scaliger’s claims and Finnaeus’ and Ursus’ quadratures
of the circle.
Ludolff van Ceulen was a German-Dutch reckoning master (see . Fig. 6.21 left), who
went from Hildesheim to Antwerp, where he stayed for some time with his brothers. Around
1562, he settled in Delft, where he made his fame as a fencing and reckoning master.
Around the mid-1580s, Ludolff van Ceulen got involved in a dispute. At the request of his
collegue Adriaan Anthonisz, he proved that Simon van der Eycke’s squaring of the circle was
wrong. Van der Eycke had published a book in 1584 in which he claimed the value for the
69
ratio of the circumference of a circle to its diameter was 3 484 .
In his 1596 book Vanden Circkel (About the Circle), van Ceulen published his ten-year
research on the area of the circle and the areas of in- and circumscribed polygons (see . Fig.
6.21 right). In the same book, he also criticised Scaliger, without mentioning his name.
Van Ceulen was buried in the Pieterskerk in Leiden. On his tombstone, the 35 decimals of
he had calculated were engraved. The original has long disappeared, but a new commemo-
14
van Roomen (1593), f (**iv) v, also Bockstaele (1976, 1993, 2009).
15
Aristotelean dogma had it that one could not use algebra for solving geometric problems (Yu (2003), p. 44). “It
follows that we cannot in demonstrating pass from one genus to another. We cannot, for instance, prove geometrical
truths by arithmetic. [. . . ] Nor can the theorem of any one science be demonstrated by means of another science, unless
these theorems are related as subordinate to superior. [. . . ] ” (Aristotle, Posterior Analytics, Book I, section 7).
. Fig. 6.21 Left: portrait of Ludolff van Ceulen, right: Ludolph van Ceulen’s approximation of , a lower limit in the
upper half of the circle, an upper limit in the lower half of the circle (van Ceulen (1619), EHC G 4867)
rative stone reminds us of his feat. In German is sometimes referred to as die Ludolphsche
Zahl (Ludolff’s number).
Van Ceulen used the same method as van Roomen to calculate . Whereas van Roomen’s
1593 book puts forward “recipes” without formulae to calculate the edge, van Ceulen’s 1596
book explains the method in a symbolism which is still quite readable.
We will determine the area of an inscribed and circumscribed regular polygon. By con-
structing a 2n-gon from an n-gon we can give a better approximation of the area. We can then
construct a 4n-gon etc16 .
To begin with, draw an inscribed and circumscribed hexagon in a circle with radius 1 (see
. Fig. 6.22). The edge of the inscribed hexagon is 1 and the circumference is p6 D 6.
Exercise 116
1
Referring to . Fig. 6.22, if a hexagon is inscribed in a circle with radius 1, prove that kAC k D
2
and calculate kOC k ; kCDk and kEDk.
p
3
We find that kEDk D .
3 p
2 3 p
The edge of the circumscribed polygon thus is and its circumference is P6 D 4 3.
p 3 p
Now pn < 2 < Pn from which 6 < 2 < 4 3 and 3 < < 2 3.
Exercise 117
q r q
Prove that .2 sin ˛/ D 2
2
4 .2 sin 2˛/ and .2 sin ˛/ D 2
2 2
2C 4 .2 sin 4˛/2 .
Deduce a formula to calculate the length of the edges of a 2i n-gon.
Hint: In . Fig. 6.23 kABk D 2 sin 2˛ and kA0 B 0 k D 2 sin ˛.
16
Based on Wepster (2010a,b).
97 6
. Fig. 6.22 The inscribed and
circumscribed hexagon of a circle
It was this formula that allowed van Roomen and van Ceulen to “quickly” calculate the
edges of a 2i n-gon, in a circle with radius 1, knowing the edge of the n-gon and to improve
the approximation step by step. Puttingpthe edge of a regular hexagon equal to x, then the edge
p
of a regular dodecagon equals s12 D 2 4 x 2 .
Beginning with a hexagon, the formula for a 192-gon (D 25 6-gon) reads as
v v
u u s
u u r
u t q p
t
s192 D 2 2 C 2 C 2 C 2 C 4 x2
Exercise 118
In a circle with radius 1, prove that if Sn is the edge of the circumscribed regular polygon and sn
2sn
the edge of the inscribed polygon, that Sn D p . (Hint: calculate tan ˛)
4 sn2
Exercise 119
Use the van Roomen-van Ceulen formula to calculate, to eight decimals, the circumference of the
inscribed and circumscribed 24, 48, 96 and 192-gons.
. Fig. 6.23 The inscribed hexagon

and inscribed dodecagon of a circle
Exercise 120
Use a spreadsheet and make a table in which you iteratively calculate the values of the circumfer-
ence
qof inscribed and circumscribed polygons beginning with the hexagon. Define the cell A.i C 1/
p
as 2 4 A.i/2 : Define cell A .1/ as the edge of the inscribed hexagon (e.g. 1).
In another column do the same for the circumscribed polygons.
Calculate the circumferences of the inscribed and circumscribed polygons in the other columns.
What do you notice?
6.8 Kepler’s use of infinitesimal methods

6
By the beginning of the seventeenth century, mathematicians started to explore completely
new ideas concerning the sums of areas: infinitesimal methods.
Let’s start with a simple idea. Consider a rectangle, which we divide intofour new
rectan-
B B
gles, each having length L and width . Then the area of the rectangle is 4 L D LB.
4 4
Suppose we make the parts even smaller and divide the rectangle into n smaller rect-
B
angles. Then every rectangle has length L and width . The area then of course still is
n
B
n L D LB.
n
Now suppose n can become infinitely large and the rectangle is cut into infinitely many
rectangles, which have zero width. These rectangles have become lines.
We would write

B
lim n L D lim L B D LB:
n!C1 n n!C1
But infinitesimal calculus was still in its infancy, the limit concept would not be introduced
until well into the nineteenth century. Mathematicians were still trying to come to grips with
these new concepts. The fundamental question they had to answer was: “How can one add an
infinity of lines to find an area?” Sometimes they made astounding and correct discoveries,
sometimes they made fundamental errors.
One of the first to use such infinitesimal methods was the astronomer Johannes Kepler. He
put forward the following solution to the circle quadrature.
Consider a circle, with radius r and perimeter P , which is cut up, like a cake, into n sec-
tors. Take the sectors from the circle and arrange them as shown in . Fig. 6.24. This method
seems similar to Franco of Liège’s method (see 7 Sect. 6.4), but Kepler takes a very different
approach.
Now consider a very small interior angle for each sector, i.e. let n become very large.
Consider one such circular sector OAB and also 4OAB (see . Fig. 6.24). Kepler states that
the height h of 4OAB is nearly equal to the radius r of the circular sector OAB, h r.
P P
Similarly, the base of 4OAB is nearly equal to or kABk D xi .
n n
Now the area of the parallellogram, which is made up of the triangles, is
X X
1 X
n n n
1 1 1 2r
SD hxi rxi D r xi D rP D r D r 2 :
i D1
2 i D1
2 2 i D1 2 2
6.9 Gregory of St Vincent
99 6
. Fig. 6.24 Kepler’s transformation of a circle into a parallellogram
Which is the same result as that of Archimedes (see 7 Sect. 6.6), but instead of using a tri-
angle for comparison to the circle, Kepler uses a parallellogram. While Archimedes’ proof is
cumbersome, but based on a solid geometric and logical basis, Kepler’s proof is quite easy
but not rigourous even by the standards of his time. Kepler does not build his proof on a finite
number of sectors, he boldly states that there are an infinite number. In that case, he argues,
the difference between sector and triangle is negligible. He considers this kind of reasoning to
be legitimate.
The most famous mathematician to tackle the squaring of the circle problem with infinitesimal
methods was Flemish Jesuit Gregory of St-Vincent (see also 7 Sect. 5.5). In 1615, Gregory
was sent to Antwerp to teach mathematics at the Jesuit college. Here Gregory developed
methods similar to those of his contemporary Kepler. He was one of the first to use the sum
of rectangles to approximate the area between curves. He would then put the width of the
rectangles equal to zero, in effect making them into lines. Being the first mathematician to
venture into this territory, he had to build up the mathematics and its axioms involved in these
processes, sometimes erring along the way. In . Fig. 6.25 we can see how he uses rectangles
to approximate the area between two parabolae and between a parabola and a straight line.
By making the width of the rectangles infinitely small and by adding the “areas” of these
rectangles, he was able to calculate the area between the parabola and the straight line.
By 1625, Gregory thought he could solve the problem of squaring the circle, considered
as determining the exact area of a circle (not as a construction problem). However, before they
were allowed to publish any of their work, Jesuits had to seek permission from the Superior-
General in Rome. In Gregory’s case, the General forwarded his manuscript to Christopher
Grienberger s.j., the mathematics professor at the Jesuit College in Rome, the famous Col-
legium Romamum. Grienberger was not convinced by Gregory’s results and summoned him
to Rome. Gregory’s journey proved to be in vain, he was unable to convince Grienberger and
failed to get permission to publish. His book entitled Problema Austriacum would not be pub-
lished until 1647 (see . Fig. 6.26). By that year, Girolamo Cardano had already published a
book explaining a very similar method.
Grienberger had objected to Gregory’s method because it was not logically sound, i.e. it
failed to live up to the strict classical Greek exigences of accuracy. Incidentally, Gregory’s
. Fig. 6.25 Gregory’s figure to explain the calculation of the area between two parabolae and between a parabola
and a straight line (from Sancto Vincentio (1647), EHC G 4869)
. Fig. 6.26 Frontispiece of Gregoy

of Saint Vincent’s Problema Austri-
acum on the squaring of the circle.
Notice the angel holding a square
window and the sun’s rays passing
through it casting a circular spot on
the ground (from Sancto Vincentio
(1647), EHC G 4869)
treatise did contain a major flaw and a fundamental error. However, this error was not discov-
ered until the treatise was published in 1647. It was Christian Huygens, one of the greatest
mathematical minds of his generation, who found the error. It damaged Gregory’s reputation
considerably.
Yet we should not be too harsh on Gregory. He had boldly gone where no man had gone
before, seeking out strange new mathematical worlds. To Gregory we owe (the equivalent of)
Zb
1
the formula dx D ln b ln a. A formula he proved using his rectangle method17.
x
a
17
Actually, Gregory proved that the area under a rectangular hyperbola and one of its asymptotes (also acting, in our
a c
notation, as x-axis) is the same over the segment Œa; b as over the segment Œc; d if D .
b d
101 6
. Fig. 6.27 Using the rectangle
method to calculate the area be-
tween a straight line y D x and the
x-axis over the interval Œ0; 1
In the paragraphs below, we will use infinitesimal methods to determine the area between
a straight line and a parabola, a problem which Gregory also solved18.
Exercise 121
Use the method of induction to prove that
X
n
1
iD n .n C 1/ : (16)
iD1
2
Suppose we want to find the area under the straight line y D ax (a > 0) between x D 0
1 1 1
and x D b (> 0). Obviously this is a triangle and its area is S D b y.b/ D b ab D ab 2 .
2 2 2
We will now, as an example, use infinitesimal methods to arrive at the same result.
b
Consider our triangle. We divide the leg along the x-axis into n pieces, with length .
n
Draw rectangles as in . Fig. 6.27.
ib aib
Then the height of a rectangle with vertex at x D is y D .
n n
2
b aib iab
Therefore the area of this rectangle is Sr D D 2 .
n n n
The area of the triangle can be approximated by
X
n
iab 2
SD
i D1
n2
ab 2 X
n
D i
n2 i D1
ab 2 1
D n .n C 1/
n2 2
1 nC1
D ab 2
2 n
in which we have used the result of Exercise 121.
18
Gregory’s method was more complicated. For instance instead of using rectangles with equal bases, in his method
the bases are in a geometric progression.
. Fig. 6.28 Using the rectangle

method to calculate the area be-
tween a parabola y D x 2 and the
x-axis over the interval Œ0; 1
nC1
We notice that if n becomes very large 1.
n
nC1
Mathematically lim D 1, so
n!1 n
1 2
SD ab : (17)
2
If you are lucky your IGS will be able to calculate a Riemann sum. A Riemann sum is a math-
ematical tool you can approximate the area between a curve and the x-axis with. Basically it
is the sum of the areas of rectangles with a base on the x-axis and one vertex on the curve.
Depending on which vertex is on the curve, we distinguish between a left Riemann sum and a
right Riemann sum. In our example the right vertices are on the curve so we have calculated the
right Riemann sum for the area between the straight line y D x and the x-axis between x D 0
and x D 1. Some IGSs allow you to calculate these Riemann sums, in others you have to re-
vert to tricks, such as calculating the area of each rectangle in a list and using a spreadsheet to
calculate their sum. Draw n rectangles with base Œai ; bi and height f .bi / on the interval Œa; b
between the x-axis and the function y D f .x/, as in . Fig. 6.27 and . Fig. 6.28. Calculate
the sum of these areas, call this RightRiemannSum(f; a; b; n). LeftRiemannSum(f; a; b; n) is
similar: draw n rectangles with base Œai ; bi and height f .ai / on the interval Œa; b between
the x-axis and the function y D f .x/ and calculate the sum of their areas.
Exercise 122
Create an IGS file in which you draw a line y D x, by defining a function f .x/ D x. Define a
slider n which varies from 3 to 100. Now calculate RightRiemannSum(f; 0; 1; n). What is the value
of this sum when n D 3 and when n D 100? Compare this number to the result of 17 when a D 1
and b D 1.
Now define a slider a > 0 and define f .x/ D ax. Again define a slider n which varies from 3
to 100 and calculate RightRiemannSum(f; 0; 1; n). What is the value of this sum when n D 3 and
when n D 100? Compare this number to the result of (17).
103 6
. Fig. 6.29 The area between a
parabola y D x 2 and a straight line
yDx
Exercise 123
Use the method of induction to prove that
X
n
n.n C 1/.2n C 1/ 1
i2 D D n 2n2 C 3n C 1 :
iD1
6 6
Use this result to calculate the area under the parabola y D x 2 and above the x-axis, between x D 0
and x D b.> 0/. Use a similar method as the one above.
ib i 2 b2 b i 2 b2
Hint: the height at x D is y D 2 , so the area of a rectangle is Sr D 2 . Now
n n n n
2n C 1
summate all the rectangle areas. Hint: lim D 2.
n!1 n
Exercise 124
Create an IGS file in which you draw a parabola y D x 2 by defining a function f .x/ D x 2 . Define
a slider n which varies from 3 to 100. Now calculate RightRiemannSum(f; 0; 1; n). What is the
value of this sum when n D 3 and when n D 100? Compare this number to the result you found in
Exercise 123 when a D 1 and b D 1.
Define a slider t > 0 and calculate RightRiemannSum(f; 0; t; n). Deduce that the area between
t3
the x-axis and the parabola f .x/ D x 2 over the interval Œ0; t is equal to .
3
Exercise 125
Use equation (17) and the result of Exercise 123 to prove that the area between the parabola y D x 2
1
and the straight line y D x, between x D 0 and x D 1, equals (see . Fig. 6.29).
6
To square a circle one would have to calculate the area under a circle with radius R and
above the line x D 0. This cannot readily be done using Gregory’s rectangle method. It can
be done once we know the concept of Riemann sum, which produces an integral. In this case,
p with centre O and radius R and above the line x D 0 one has
to obtain the area under Ra circle
R
to calculate the integral R R2 x 2 dx.
Exercise 126
Create an IGS file in which you draw a semi-circle with radius 1 by defining a function f .x/ D
p
1 x 2 . Define a slider n which varies from 3 to 100. Now calculate RightRiemannSum(f; 0; 1; n).
What is the value of this sum when n D 3 and when n D 100? Compare this number to .
Although Gregory did not succeed in determining the area of a circle, his student Joannes
della Faille (1597–1652) did. Ironically, della Faille succeeded in determining the centroid of
an ellipse and deduced the area, and hence also the area of a circle, using classical geometric
methods19.
19
Meskens (2005).
105 7
Constructible numbers

7.1 Constructing numbers
In the previous chapters we have taken a look at some classical Greek geometric compass and
straightedge construction problems. These constructions are a consequence of Euclid’s first
three postulates:
1. It is possible to draw a straight line between points
2. It is possible to elongate any segment in a straight line as far as one wants
3. With a point as the centre and a radius it is possible to draw a circle
We will call a number a constructible if we can construct a line segment with length a in a
finite number of steps1 .
Axiom 1
If A and B are two different points, then a one-to-one correspondence exists between the set
of real numbers R and straight line AB, such that A corresponds to 0 and B corresponds
to 1.
The axiom tells us that if we choose two different points A and B, then we actually turn the
straight line into a ruler with ŒAB as unit line segment. Every point on the line will correspond
to a real number x such that if C is to the left of D then xC < xD .
Definition 1
If the line segment ŒAB is given and A and B correspond to 0 and 1 respectively, then a
point is called constructible if it can be constructed using compass and straightedge methods
in a finite number of steps. The corresponding number is called a constructible number.
Note that this gauge can be transferred to any straight line through A or B using Thales’
intercept theorem. After we have shown that parallel lines exist and can be constructed (see
7 Sect. 3.2) we can transfer this gauge to any straight line.
1
For a thorough algebraic exposé on constructible numbers as an extension field of the rational numbers, see e.g.
Lorenz (2006), p. 1–13, Finston and Morandi (2014), p. 93–102.
106 Chapter 7 Constructible numbers
. Fig. 7.1 Dividing a segment into

three equal segments
. Fig. 7.2 The construction of the

sum and the difference of two given
constructible numbers
In the following constructions, you will note that we can construct a multitude of numbers.
But can we construct all of them?
i Construction 1 Divide a line segment into n equal segments, or, if a number a is

a
constructible, then so is (see . Fig. 7.1).
n
We have already proved this construction in 7 Sect. 3.2 (. Fig. 3.6).
i Construction 2 If a and b are constructible numbers, then a C b and a b .a > b/ are also
constructible (see . Fig. 7.2).
» Let the length of ŒOA be a and let that of ŒOB be b

Draw the straight line OA and draw a circle with radius b and centre O
The circle intersects the line at two points C and C 0
Suppose C 0 lies outside ŒOA then kAC k D a b and kAC 0 k D a C b.
i Construction 3 If a and b are constructible numbers, then ab is also constructible (see
. Fig. 7.3).
» On a straight line through O determine the points C and D such that kOC k D 1 and
kODk D a
Draw a random straight line l through O not coinciding with CD
Determine A on l such that kOAk D b
Draw a line through A and C
Now draw a straight line through D parallel to AC , this line intersects l at B and
kOBk D ab as a consequence of Thales’ intercept theorem.
Exercise 127
Prove that if a and b are constructible numbers, then ab is also constructible. Deduce a method to
construct a2 , a3 and a4 .
7.1 Constructing numbers
107 7
. Fig. 7.3 The construction of the
product of two given constructible
numbers
Exercise 128
a
Prove that if a and b are constructible numbers, then is also constructible.
b
From the foregoing, it is obvious that all (positive) rational numbers can be constructed.
Can other non-rational numbers be constructed? Well, yes some can.
p
i Construction 4 If a is a constructible number, then a is also constructible (see . Fig. 7.4).
» Choose three collinear points A, B and C such that kABk D a and kBC k D 1
Draw the circle with ŒAC as diameter
The perpendicular to AC and through B intersects the circle at D
It is obvious that 4ADB 4DCB
kABk kDBk
so D
kDBk kBC k
a kDBk p
and D from which kDBk D a.
kDBk 1
Note that this construction is also the construction of the geometric mean of a and 1 (see
7 Sect. 3.2, . Fig. 3.8).
These five constructions are the basic constructions. All other constructions for construct-
ing numbers can be reduced to a composition of these five constructions. This means that
pn
all positive rational numbers and all numbers of the kind 2 a, with a constructible, are
constructible. Moreover all sums, differences, products, quotients and square roots of con-
s
structible numbers are constructible.
p
p 10
For instance, the number 2 C 8
7C p is constructible.
1 C 16 3
We will denote the set of constructible numbers as K.

the square root of a constructible
number
Exercise 129
If a and b are constructible numbers, find a way to construct
p
(a) ab
p
(b) a n
1
(c) .abcd / 4
Exercise 130
Some constructible numbers have strange shapes and yet are so simple.
Try these: p p p p
Prove that 4 4 2 3 C 97 56 3 is an integer number.
6 6
Prove that p C p is an integer number.
3 3 3C 3
7 Exercise 131
(a) Consider the quadratic equation x 2 bx C c D 0 (b; c > 0). Draw a line segment ŒAB of
length b. From the midpoint of ŒAB, drop the perpendicular bisector. Find point C on the
p
perpendicular bisector for which kM C k D c. Draw a circle centered at C and with radius
b
, this circle intersects ŒMB at D. Then x D kADk D kAM k C kMDk is a solution to the
2
quadratic equation. Prove this assertion.
(b) Consider the equation x 2 C bx c D 0 (b; c > 0). Draw a line segment ŒAB of length b.
From the midpoint of ŒAB drop the perpendicular bisector. Find point C on the perpendicular
p b
bisector for which kM C k D c. Draw a circle centered at A and with radius , this circle
2
intersects ŒAC at E. Then kEC k is a solution to the quadratic equation. Prove this assertion.
(c) Use this result to prove that you can divide a line segment of length a, using only straightedge
and compass, into two parts such that the difference of their squares is equal to their product.
Show that the longer segment is the mean proportional between the shorter segment and the
whole line. This ratio is called the golden ratio or golden section (sectio aurea).
Exercise 132
Let the coordinates of A, B, C and D be constructible numbers. Show that the coordinates of the
intersections of AB and CD, AB and a circle centered at C and with radius kCDk, and the circles
centered at A and C and with radii kABk and kCDk respectively, are also constructible.
With our understanding of constructible numbers, we can now take another look at the
Greek construction problems.
7.2 The theory of equations
The three classical problems we have focused on were solved, in the negative, in the nine-
teenth century. The duplication of the cube and the trisection of the angle were proved to
be impossible by Pierre Wantzel (1814–1848). He proved that a constructible number is the
solution to a polynomial equation of degree 2n with rational coefficients.
Theorem
p 1
2…K
3
109 7
The proof uses two theorems from the theory of equations, which we will not prove.
i Lemma 1 A cubic equation with rational coefficients, which does not have a rational
solution, does not have a constructible solution.
i Lemma 2 If the coefficients of the polynomial equation an x n C : : : C a1 x C a0 D 0 are

p
integers, then every rational solution to the equation can be written as with p a factor of
q
a0 and q a factor of an .
If we consider the general cubic equation a3 x 3 C a2 x 2 C a1 x C a0 D 0 with ai 2 Z, then

p
the last theorem states that every rational solution can be written as with p a factor of a0
q
and q a factor of a3 .
For example:
x 3 1 D 0, it is clear that if there is a rational solution it has to be x D ˙1. ˙1 are the
only factors of 1 and a3 D a0 D 1.
˙1
A rational solution therefore can be written as .
˙1
Note that x 1 D .x 1/.x C x C 1/ D 0, the equation has x D 1 as a solution, but no
3 2
other real (and therefore no other constructible) solutions. This does not contradict Lemma 1.
Lemma 1 does not guarantee extra constructible solutions if there is a rational solution. It only
states that the lack of a rational solution implies that there are no constructible solutions.
x 3 2x 2 5x C 6 D 0. a3 D 1 and a0 D 6. We only need to look for factors of 6, which
are ˙1; ˙2; ˙3; ˙6. After some trial and error, we find .x 1/.x C 2/.x 3/ D 0, having
solutions x D 1; x D 2; x D 3, and these obviously are constructible numbers.
2x 3 13x 2 30x 9 D 0. a3 D 2 and a0 D 9, we have to consider all fractions of which
the denominator is a factor of 2 and the nominator a factor of 9. The factors of 2 are ˙1; ˙2,
3
those of 9 are ˙1; ˙3; ˙9. After some trial and error, we find that x D is a solution to
p 2
8 ˙ 76 p
the equation. The other solutions are x1;2 D D 4 ˙ 19, and these are obviously
2
constructible. p
We know that 3 2 is a solution to the equation x 3 2 D 0. The factors of a3 D 1 are
˙1, those of a0 D 2 are ˙1; ˙2. Possible rational solutions are ˙1 en ˙2. None of these
numbers satisfies the equation. There are no rational solutions!!
As a consequence the equation x 3 2 D 0 has no constructible solutions.
The duplication of the cube is impossible if we can find at least one cube for which we
cannot find a cube twice as large using compass and straightedge methods. From the forego-
ing, it is clear that the edge of a cube which has double the volume as a cube of which the
edge has length 1 is not constructible. More generally, if a line segment p has a constructible
length a then it is impossible to construct a line segment with length 3 2a using compass and
straightedge methods.
We can treat the problem of the trisection of an angle in a similar way.
To see why, consider the following figures (in . Fig. 7.5). Let the length of the chord ŒCB,
which is subtended by 3˛, be d3 and let x be the length of the chord subtended by ˛.
sin 2 sin 3˛
Using the sine rule in 4ABC , which is right-angled in C , we find D from
2R d3
which d3 D 2R sin 3˛ and x D 2R sin ˛.
Now we know from the inscribed angle theorem that an angle # inscribed in a circle is
half the central angle 2# that subtends the same arc on the circle. So, referring to . Fig. 7.5
. Fig. 7.5 Determining an equation for the trisection of an angle using inscribed angles (left) and central angles
(right)
ˇ 3ˇ ˇ
right, ˛ D and d3 D 2R sin and x D 2R sin . This result can readily be generalised
2 2 2
7 nˇ
to dn D 2R sin .
2
We can now generalise this result for the division of an angle into n C 2 parts. Let R D 1,
let the chord subtended by the angle .n C 2/ which has to be divided into n C 2 parts be
denoted by dnC2 .D 2 sin .n C 2/ / and let x be the chord subtended by .
From the sum-to-product identities we know:
sin ..n C 2/ / sin ..n 2/ / D 2 sin n cos 2
which after some manipulations:
sin ..n C 2/ / D 2 sin n cos 2 sin ..n 2/ /

D 2 1 sin2 sin n sin ..n 2/ /
ˇ
Putting D we find
2

ˇ 2 ˇ nˇ .n 2/
sin .n C 2/ D 2 1 sin sin sin ˇ
2 2 2 2
x 2 d
dnC2 n dnC2
D2 12
2 2 2 2

1 2
dnC2 D 2 1 x dn dn2
2
Now put d1 D 2 sin .˛/ D 2 sin ˛ D x, which allows us to draw up a table with the
lengths of the chords in consideration of x D 2 sin ˛.
d1 D x
d3 D 3x x 3
d5 D 5x 5x 3 C x 5
d7 D 7x 14x 3 C 7x 5 x 7
::
:
111 7
. Fig. 7.6 Part of Viète’s solution to van Roomen’s problem using chords (Viète (1646), EHC G 4858)
. Fig. 7.7 Nicomedes’ solution for

the angle trisection
Viète noticed that the left-hand side of van Roomen’s equation was the expression for d45 , the
right-hand side therefore was the length of this chord (see 7 Sect. 2.3) (see . Fig. 7.6).
The expression for d3 can be written as
7 2 sin 3˛ D 3x x 3 ; (18)
which is a third degree equation.

On the other hand, if we put y D sin ˛ and c D sin 3˛ then
4y 3 3y C c D 0: (19)
Exercise 133
If y D sin ˛ and c D sin 3˛ then 4y 3 3y C c D 0, prove this assertion! Also show that this
equation is equivalent with equation (18). Hint: first prove that sin 3˛ D 3 sin ˛ 4 sin3 ˛ (see
Exercise 79).
Exercise 1342
. Fig. 7.7 shows Nicomedes’ solution to the trisection problem. Put kOC k D b, kBQk D x,
kOP k D y. Prove that
x 3 3x 2b D 0: (20)
y C 3a
Hint: first prove that kEQk D .
2
The trisection of an angle is impossible if an ˛ exists for which equation (19) has no
solution which can be written as a C b sin ˛, a; b 2 K.
Exercise 135
If a line segment is given that has a length sin ˛ (which does not need to be constructible), then ˛
can be constructed.
The trisection of an angle in general is impossible if at least one angle can be found which
cannot be trisected using compass and straightedge methods.

Consider the angle D 3˛ D rad.
6
1
Equation (19) now becomes 4x 3 3x C D 0 ) 8x 3 6x C 1 D 0.
2
The factors of 8 are ˙1; ˙2; ˙4; ˙8, those of 1 are ˙1.
2
Bunt et al. (1988), p. 118–119.
7.3 Squarable lunes
113 7
1 1 1
We have to check whether any one of ˙1; ˙ ; ˙ or ˙ is a solution to the equation.
2 4 8
Substituting the values into the equation shows that none of these are solutions. Therefore the
equation has neither rational nor constructible solutions.

An angle of magnitude rad cannot be constructed, and therefore the angle rad cannot
18 6
be trisected. Because we have found one counterexample, we can decide that the trisection of
an angle in general is not possible. Remember the question was whether we can trisect any
angle using compass and straightedge methods.
Exercise 136

Use equations (18) and (20) to prove that an angle of rad cannot be trisected.
6
Exercise 137

Show that an angle of rad can be trisected.
2
7.3 Squarable lunes
In 7 Sect. 6.3 we encountered lunes for which we were able to determine the area by con-
structing a polygon with an equal area. We will now investigate whether and how we can
construct other lunes, and whether we can express the area of these lunes as a constructible
number.
Consider . Fig. 7.8 (left) in which we need to determine the area of the shaded part.
The lune is bounded by two arcs of a circle. For each arc we will determine the centres of
their circles, calculate the area of the segments and subtract the triangles with the centres and
endpoints of the arcs as vertices (see . Fig. 7.8 (right)).
. Fig. 7.8 A lune defined by two angles

area CEDF D area CFDG area CEDG

area CEDG D area circular sector ACED area triangle 4ACD
2˛
area ACED D R2 D ˛R2
2
2ˇ 2
area BCFD D r D ˇr 2
2

1 1
area CEDF D ˇr 2 r 2 sin 2ˇ ˛R2 r 2 sin 2˛
2 2
1
D ˇr 2 ˛R2 r 2 sin 2ˇ R2 sin 2˛
2
To simplify the calculations, we will assume that

7
ˇr 2 D ˛R2 : (21)
Hippocrates used two assumptions to find constructible lunes:

1. Two circular sectors (with radius r and R) corresponding to the lunes’ arcs have the same
area
2. The central angles (2˛; 2ˇ) of the two circular arcs are commensurable.
These assumptions are equivalent to equation (21). It can be shown that these relations hold
for all squarable lunes. The proof however goes beyond the scope of this book3.
Exercise 1384
Show that Hippocrates’ assumptions are equivalent to equation (21). Express algebraically that 2˛
and 2ˇ, and hence also ˛ and ˇ, are commensurable. Let be the greatest common denominator
of 2˛ and 2ˇ. Express the areas of the two sectors in terms of and deduce that
˛ r2
D 2:
ˇ R
Using equation (21) we find that:
1 2
area CEDF D R sin 2˛ r 2 sin 2ˇ :
2
˛R2
Now from equation (21) we also have that ˇ D and it follows that
r2

1 ˛R2
area CEDF D S D R2 sin 2˛ r 2 sin 2 2 : (22)
2 r
R2
Let m D
r2
3
See e.g. Rothe (2014), p. 696 ff. The proof uses algebraic numbers and some fundamental theorems of number theory.
Algebraic numbers are real numbers which are the solution to a polynomial equation with integer coefficients.
4
Rothe (2014), p. 696 ff.
7.3 Squarable lunes
115 7
Now
kC Gk D R sin ˛ D r sin ˇ
R
) sin ˇ D sin ˛
r
R2 R
) sin 2 ˛ D sin ˛ (23)
r r
p
) sin m˛ D m sin ˛
we can rewrite equation (22) as

1 ˛R2
SD R sin 2˛ r sin 2 2
2 2
2 r
!
1 2 R2 ˛R2
D r sin 2˛ sin 2 2
2 r2 r
1 2
D r .m sin 2˛ sin 2m˛/
2
1
D r 2 .2m sin ˛ cos ˛ 2 sin m˛ cos m˛/
2
p p
D r 2 m sin ˛ 1 sin2 ˛ sin m˛ 1 sin2 m˛
p p p
D r 2 m sin ˛ 1 sin2 ˛ m sin ˛ 1 m sin2 ˛ (from (23))
p p p p
D r 2 m sin ˛ m 1 sin2 ˛ 1 m sin2 ˛ (24)
To know whether the lune is squarable, we need to be able to determine whether sin ˛ is
constructible, i.e. that we can draw a line segment with length sin ˛ using compass and
straightedge methods. It is obvious that this will only be the case for certain values of m.
A lune is squarable if for certain constructible values of m the equation sin m˛ D
p
m sin ˛ yields an expression for sin ˛ which can be constructed. If sin ˛ is constructible,
then so is expression (24).
It turns out that only five values for m will yield a constructible result5 . These numbers are
3 5
m D 2; 3; ; 5; .
2 2 p
If m D 2 then sin 2˛ D 2 sin ˛, this is Hippocrates’ isosceles right triangle case.
p
sin 2˛ D 2 sin ˛
p
) 2 sin ˛ cos ˛ D 2 sin ˛
p
2
) cos ˛ D _ sin ˛ D 0
2
p
2
) sin ˛ D
2
Obviously sin ˛ D 0 is not a valid solution.
5
See Postnikov and Shenitzer (transl.) (2000).
So
p
p 2
sin 2˛ D 2 D 1;
2
the area then is
p p !
p 2 p 2 1 2
S Dr 2
2 2 0 Dr D R
2
2 2 2
If m D 3 then
p
sin 3˛ D 3 sin ˛; (25)
7 this the isosceles trapezoid case.
We know sin 3˛ D 3 sin ˛ 4 sin3 ˛ (see Exercise 133)
So
p
sin 3˛ D 3 sin ˛
p
) 3 sin ˛ 4 sin3 ˛ D 3 sin ˛
p
) 3 4 sin2 ˛ D 3
p
) 4 sin2 ˛ D 3 3
s p s p
3 3 3 3
) sin ˛ D _ sin ˛ D
4 4
q
1 p
) sin ˛ D 3 3
2
Obviously the sine cannot be negative in this case, so
0 v 1
q
s p
u
u p
p 1 p Bp 3 3 t 3 3 3 C
S D r2 3 3 3B
@ 3 1 1 C
A
2 4 4
p q p q q !
3 p 2 3 p 1 p
D 3 3r 43C 3 49C3 3
2 2 2
p q q q
3 p 2 p p p
D 3 3r 3 1 C 3 5 C 3 3
4
p q q q
3 p 2 p p
D 3 3r 3 C 3 3 5 C 3 3
4
7.3 Squarable lunes
117 7
r
3 3 3
If m D then sin ˛ D sin ˛.
2 2 2
˛
Put ˇ D then
2
r
3
sin 3ˇ D sin 2ˇ
2
r
3
) 3 sin ˇ 4 sin3 ˇ D 2 sin ˇ cos ˇ
2
p
) 3 4 sin2 ˇ D 6 cos ˇ
r
3
) 3 sin ˇ 4 sin3 ˇ D 2 sin ˇ cos ˇ
2
p
) 3 4 sin2 ˇ D 6 cos ˇ
p q
) 3 4 sin2 ˇ D 6 1 sin2 ˇ (26)
) 9 24 sin ˇ C 16 sin ˇ D 6 6 sin ˇ
2 4 2
) 16 sin4 ˇ 18 sin2 ˇ C 3 D 0
p p
18 ˙ 132 9 ˙ 33
) sin ˇ D
2
D
32 16
q q
1 p 1 p
) sin ˇ D 9 33 ^ cos ˇ D 7 C 33
4 4
which are constructible numbers.

pimplies 3 4 sin ˇ > 0. Inserting the possible values for sin ˇ reveals
2 2
Now equation (26)
9 C 33
that if sin2 ˇ D then 3 4 sin2 ˇ < 0.
16
1p p
Therefore we only retain sin ˇ D 9 33, from which
4
q q q
1 p 1 p 1 p
sin ˛ D 2 sin ˇ cos ˇ D 2 9 33 7 C 33 D 30 C 2 33:
4 4 8
Also
p
˛ 1 cos ˛ 1 C 33
sin ˇ D sin
2
D 2
) cos ˛ D :
2 2 8
Exercise 139
p p p p p 1 p
Prove that 34 2 33 D 1 C 33 and that 38 6 33 D p 9 33 .
3
Exercise 140
Prove that the area of this lune is
q
1 2 p p
SD r 3 C 33 30 C 2 33:
32
Exercise 141
Prove that cos 3˛ D 4 cos3 ˛ 3 cos ˛ (see also Exercises 79 and 133).
Use this result to prove that
sin 5˛ D 16 sin5 ˛ 20 sin3 ˛ C 5 sin ˛ (27)
Hint: sin 5˛ D sin .3˛ C 2˛/ D : : : p

Use equation (27) to solve sin 5˛ D r5 sin ˛ for sin ˛ (this is the case m D 5).
5 5 5
Use equation (27) to solve sin ˛ D sin ˛ for sin ˛ (this is the case m D ).
3 3 3
˛
Hint: put ˇ D
3
In both cases you will find a biquadratic equation, in sin ˛ and sin ˇ respectively.
7 In the last exercise you have identified the two other squarable lunes. Now you can also
determine their area.
Exercise 142 (For the extremely brave)

In Exercise 79 and 133 you have expressed sin 3˛ in terms of sin ˛ and in Exercise 141 you have
done the same for sin 5˛.
Now use the fact that sin 45˛ D sin .5 .3 .3˛/// D sin .3 .3 .5˛/// to show that van Roomen’s
x
polynomial (see 7 Sect. 2.3) is equivalent to expressing sin 45˛ in terms of D sin ˛.
2
You can also use de Moivre’s theorem for complex numbers and expand the binomial with the
binomial theorem.
7.4 Squaring the circle is different ...
The third classical Greek problem, squaring the circle, could not be solved using Wantzel’s
methods. Algebraically squaring the circle is equivalent to solving the equation x 2 D 0.
We know that a quadratic equation with constructible coefficients yields constructible solu-
p
tions. This equation therefore leads to an infernal circle: to prove that and thus is
constructible, we need to know whether the constant term of the equation, , is constructible,
which is exactly what we need to prove.
It was suspected in the nineteenth century that was not an algebraic number. Algebraic
numbers are numbers which are the solution to a polynomial equation with integer coeffi-
cients. By definition, constructible numbers are also algebraic, but not all algebraic numbers
are constructible.p
For instance, 3 2 is an algebraic number, because it is the solution to the equation x 3 2 D
0, but as we have seen it is not a constructible number. however is not the solution to such
an equation. We say that is transcendental.
Although it was clear that transcendental numbers must exist, it was only in 1844 that
Joseph Liouville (1809–1882) succeeded in constructing such a transcendental number,
thereby identifying the first such number:
1 1 1 1
C 2Š C 3Š C : : : C nŠ C : : : D 0:11000100000000000000000100:: :
10 10 10 10
7.4 Squaring the circle is different . . .
119 7
He also tried, in vain, to prove that the numbers e and are transcendental as well. It was
Charles Hermite (1822–1901) who succeeded in proving that the number e was transcen-
dental. Building on his work, Ferdinand Lindemann (1852–1939) was able to prove that
p1 e x1 C p2 e x2 C : : : C pn e xn cannot equal zero if all xi and pi are distinct algebraic numbers.
Now it is proven in complex analysis that e i C 1 D 0. This implies that i is not algebraic.
But we know that i is algebraic, because it is a solution to the equation x 2 C 1 D 0, therefore
is transcendental. Since is not algebraic, it is a forteriori not constructible.
As is often the case in mathematics, the solution to one problem leads to a multitude of
other problems. E.g., is algebraic or transcendental? Since the nineteenth century, we have
learned that if a is an algebraic number, not equal to 0 or 1, and if b is an irrational algebraic
p p p 3
2
3
number, then ab is transcendental6 . For instance, 2 2 or 2 are transcendental numbers.
Because is not algebraic, the question on the nature of remains open.
Why don’t you have a go at the proof?
6
Gelfond (1934).
121 8
The Cinderella of regular polygons

8.1 The inconstructibility of the heptagon
Among the regular polygons, the regular heptagon most certainly is Cinderella. Contrary to a
triangle, a square, a pentagon or a hexagon, the heptagon is not constructible using compass
and straightedge methods. In what follows, we will guide you through the proof of this non-
constructibility1 .
Exercise 143
Consider . Fig. 8.1. Show that if †BAC D ˛, then:
†ACB D ˛
†CBD D †CDB D 2˛
†DCE D †CED D 3˛
†EDF D †EFD D 4˛
and so on, if the initial angle allows it.

Hint: this configuration is a generalisation of the configuration used for the trisection of an
angle shown in . Fig. 5.8.
We call this configuration Viète’s ladder. Viète used this configuration to calculate the
edge of a regular heptagon.
Exercise 144
.n 2/
Show that in a regular n-gon, the internal angle in a vertex is equal to rad.
n
Let ABCDEF G be a regular heptagon with edge equal to 1.
Determine †BAG and from this deduce a value for †BAC .
Show that †BAC D †CAD D †DAE D †EAF D †FAG.
Draw the diagonals from A and let
kAC k D kAF k D x
kADk D kAEk D y
1
Inspired by Ostermann and Wanner (2012), p. 162–164.
122 Chapter 8 The Cinderella of regular polygons
. Fig. 8.1 Viète’s ladder
Cut the pentagon ABCDE. Fold the figure along diagonal AD. Turn and fold along diagonal
AC . You will find a figure as seen in . Fig. 8.2. The right figure is Viète’s Ladder2 .
To make a distinction we have added an accent to the letters in the right figure.
It is obvious that 4ADE Š 4A0 D 0 E 0 .
Exercise 145
Show that 4ACF 4C 0 B 0 D 0 and 4ADE 4D 0 E 0 C 0 .
y 1
Deduce that kB 0 D 0 k D and kC 0 E 0 k D
8 x y
Use these results to show that
y 1
y D1C and yDxC : (28)
x y
Exercise 146
Show that, if you eliminate y from these equations, you obtain
x 3 x 2 2x C 1 D 0: (29)
Which equation do you obtain when you eliminate x from the system of equations?
Use Lemma 2 in 7 Sect. 7.1 to identify the possible rational solutions to equation (29). Are
these numbers in effect solutions to the equation? Which conclusion can you draw?
The line segments ŒAC and ŒAD cannot be constructed using compass and straightedge.
So the regular seven-pointed stars ACEGBDF and ADGCFBE cannot be constructed with
the aid of only compass and straightedge. The edges of these stars are also the diagonals of the
regular heptagon. This means that the regular heptagon ABCDEF G cannot be constructed
using just compass and straightedge either. Should this be the case, then the seven-pointed
stars could be constructed, which we have proved to be impossible.
The conclusion seems strange. The edge with length 1 is constructible, so why cannot the
heptagon be constructed? Because you need two parameters to construct a regular polygon,
the length of the edge and the measure of the internal angle at a vertex.
2
A similar “folding” method was used by Arabic mathematician Abū’l-Jūd. In answer to a question posed by Al-
Bı̄rūnı̄, he proved that the edge of a heptagon is not equal to half that of an equilateral triangle inscribed in the same
circle. Heron of Alexandria had already stated that half the edge of an equilateral triangle inscribed in the same circle
as the heptagon is a very good approximation for the edge of the heptagon. See Hogendijk (1987); Berggren (2007),
p. 580ff.
8.1 The inconstructibility of the heptagon
123 8
. Fig. 8.2 Folding a regular heptagon into an equilateral triangle
. Fig. 8.3 Archimedes’ pseudo-

neusis to determine 2 cos
7
Exercise 147

Show that cos is not constructible.
7
5
rad nor angle
From the latter assertion it follows that neither angle rad is con-
7 7
structible using compass and straightedge. On one leg we would be able to determine a line
segment with length 1. We can then construct the perpendicular on the other leg. In doing so,

we have constructed a right-angled triangle with edges cos and sin . But we have shown
7 7
that cos cannot be constructed, so neither can rad.
7 7
Exercise 1483
The following pseudo-neusis method was attributed to Archimedes by the Arabs (see . Fig. 8.3).
Create an IGS file in which you perform this operation. Draw a unit square ABCD, draw the
diagonal ŒAC . Let E be a point on ŒAC . Draw line DE and determine the intersection points F
and G of DE with ŒBC and with AB respectively. Use the Polygon tool to define the triangles
4DEC and 4FBG. Move E along ŒAC until area 4DEC D area 4FBG. Determine the length

kAGk and compare this length to 2 cos .
7
Knowing the value of cos allows us to construct a heptagon with an edge of given length.
7
3
Hartshorne (2000), p. 270, Martin (1998), p. 136.
8.2 The relation with the trisection
As unlikely as it seems, there is a connection between the construction of the heptagon and
the trisection of an angle.
Exercise 149
1
Consider the equation (29): x 3 x 2 2x C 1 D 0. Use Viète’s substitution x D t C (see
3
7 Sect. 3.3) and prove that the resulting equation is
7 7
t3 t C D 0: (30)
3 27
We know that cos 3˛ D 4 cos3 ˛ 3 cos ˛ (see Exercise 136).

Putting z D cos ˛ we find that
8 3 1
z 3 z cos 3˛ D 0 (31)
4 4
This is an equation of the same form as equation (30).

If we put t D nz equation (30) becomes
7 z 7
z3 C D 0: (32)
3 n2 27n3
We now want to determine n in such a way that the coefficients of equations (31) and (32) are
equal. p
7 1 3 2 7
Then 2 D ) n D and
3n 4 3
1 7
cos 3˛ D
4 27n3
47
1
) cos 3˛ D
p !3
27
2 7
3
p
1 7
D p D
2 7 14
We know the value of cos 3˛, using the result of Exercise 135 we know that an angle 3˛ can
be constructed. To find z D cos ˛, we need to determine ˛, which means we have to trisect
3˛. We can trisect 3˛ by any of the constructions described in 7 Chapter 5. Obviously it is
this step which prevents a compass and straightedge construction. Once we know z, we can
backtrack to t and x again using compass and straightedge methods.
8.3 A neusis for the heptagon
125 8
. Fig. 8.4 A neusis for the heptagon
8.3 A neusis for the heptagon

It is possible to construct an angle of rad using neusis. Try performing this neusis in an IGS
7
file (see . Fig. 8.4). Draw a square ABCD with the length of the edge equal to the length of
the edge of the regular heptagon. Draw the diagonal ŒBD. Draw an arc of a circle C centered
at B and with radius kBDk. Define the perpendicular bisector m on ŒDC . Choose a point
E on m. Draw the straight line AE. Define a circle centered at E and with radius kABk.
Determine the intersection points F and G of the circle and the straight line AE. Determine
the locus of F with E.
Which kind of curve is this?
Let H be the intersection point of the locus and the circle. Let the line segment ŒEF D
ŒEH in this position be ŒEN FN . ŒEN FN is the edge of the regular heptagon, in the desired
orientation.
Define the perpendicular bisector of ŒEN FN and determine its intersection point O with
straight line m. Point O is the centre of the circumscribed circle. Draw this circle. Draw
the circle centered at EN and with radius kEN FN k .D kABk/. Determine the intersection
5
point I of this circle and the circumscribed circle. Angle ]FN EN I has a magnitude rad
7
(alternatively, you can mirror ŒEN FN about m, the image of FN is I ). It is now easy to
determine the other points of the heptagon, as we know the circumscribed circle.
Repeat the procedure, but instead of choosing E on the straight line, choose E on the
circle C . Which kind of curve do you obtain as locus of F with E?
Exercise 150
Use the method of Exercise 144 in 7 Sect. 8.1 to show that the enneagon (9-gon or nonagon) is not
constructible. Why is it not possible to decide whether the hendecagon (11-gon) is constructible
using this method? Can you determine the lengths of the diagonals in these regular polygons?
8
127
Servicepart
Solutions – 128
References – 178
Index – 184

A. Meskens, P. Tytgat, Exploring Classical Greek Construction Problems with Interactive Geometry Software, Compact
Textbooks in Mathematics, DOI 10.1007/978-3-319-42863-5
128
Solutions
Exercise 1
2
4p 2 q 2 C p 4 2p 2 q 2 C q 4 D p 4 C 2p 2 q 2 C q 4 D p 2 C q 2 :
For the second part of the exercise put p D n C 1 and q D n.
Exercise 2 Let .n 1; n; n C 1/ be a Pythagorean triplet. Then
.n 1/2 C n2 D .n C 1/2 ) n2 4n D 0 ) nD0_nD4
Only for n D 4 do we have three consecutive positive integers: .3; 4; 5/.
Exercise 3 Let the sides of the isosceles triangle be given by the triplet .a; a; b/ 2 N 3 , then
p
a2 C a2 D b 2 , 2a2 D b 2 ) 2a D b:
Therefore, if a 2 N then b … N.
Exercise 4 In the equation 178212 C 184112 D 192212 , it is clear that the left-hand side is
uneven (because 1782 is even and 1841 is uneven) and the right-hand side is even (because
1922 is even). So the equation cannot hold.
In the equation 398712 C 436512 D 447212 both 3987 and 4365 are divisible by 7, and
hence 398712, 436512 and 398712 C 436512 are divisible by 7, while 4472 is not and therefore
neither is 447212.
Exercise 5 We divide the cubic close packing into unit cells, such as the one in the figure. We
see that a unit cube consists of half a sphere at each face and one eight of a sphere at each
vertex.
1 1
The unit cube therefore consists of 6 C 8 spheres or the equivalent of 4 spheres. We
2 8
now try to determine how much space they take up relative to the unit cube. Suppose the edge
129
Solutions
of the sphere has length a and the radius of the spheres is r. We see that the length of apface
diagonal of the cube equals four radii of the spheres. The face diagonal ŒAC has length
p 2a,
p 2a
because it is the hypotenuse of the triangle 4ABC . Therefore 4r D 2a or r D . The
4
volume of one sphere then is
p !3 p p
4 3 4 2a 4 2 2 3 2 3
Vs D r D D 3 a D a :
3 3 4 3 4 24
The volume of the cube of course is Vc D a3 . The volume taken up by the spheres relative to
the cubes is
p
2 3 p
4Vs 4 a 2
Vrel D D 24 D 0:74:
Vc a3 6
This means that 74% of the volume of the cube is filled with spheres.
Exercise 8
Let C be the oak tree and P and Q the birches. For P there are two points which satisfy
conditions P1 and P2 , for Q these are Q1 , and Q2 . Let M11 be the midpoint of P1 and Q1 ,
M12 of P1 and Q2 , M21 of P2 and Q1 and M22 of P2 and Q2 . We notice that as we move
C around M11 and M22 remain stationary, while M12 and M21 move about. So the property
only holds if one angle is made clockwise, the other counterclockwise. If both right angles are
made in the same direction the property does not hold.
We assume the pirate is clever enough to choose the solution which is drawn in the right-
hand figure. Select the x-axis through Q and P and the y-axis, perpendicular to x, through C .
It is easy to prove that 4COP Š 4PBP1 and 4COQ Š 4QOQ 1 . Let
the coordinates
of C , Q and P be as shown in the figure, then P1 p C cy ; p cx and Q1 q cy ; q C cx .
Therefore

p C cy q cy p cx C q C cx pq pCq
M11 ; or M11 ; ;
2 2 2 2
which is independent of cx and of cy . Therefore M11 will remain stationary. The proof for M22
is analogous.
The coordinates for P2 are P2 p cy ; cx p yielding

p q 2cy q p C 2cx
M21 ;
2 2
which is neither independent of cx nor of cy and therefore will not remain stationary when C
is moved around. An analogous calculation can be made for M12 .
130 Solutions
Exercise 9
d d t
AC W y d D xI lC W y D .x a/I lA W y D .x t /:
t t d
The coordinates of D are found by determining the intersection of lA and lC . We can solve the
d
cartesian equation of lC for t to find that t D .x a/. Inserting this value in the cartesian
y
equation of lA gives the desired equation.
Exercise 12 (a) Multiply both sides of the equation with r to obtain r 2 D 2r sin . Using the
transformation formulae this becomes:
x 2 C y 2 D 2y , x 2 C y 2 2y D 0 , x 2 C y 2 2y C 1 D 1
, x 2 C .y 1/2 D 1:
Which is the cartesian equation of a circle centered at .0; 1/ and with radius 1.
(b) Again multiply both sides of the equation with r to obtain r 2 D 2r cos . This now
gives us the cartesian equation .x 1/2 C y 2 D 1, which is the equation of a circle centered
at .1; 0/ and with radius 1.
(c) The first equation is the equation of a circle centered at .0; a/ and with radius a, the
second of a circle centered at .a; 0/ with radius a.
Exercise 13 These are all limaçons. Depending on the value of a you find a curve which has an
inner loop, or one which is bulging in.
The curves with equations r D 2 C a cos and r D 2 a cos are each others mirror
image with respect to the straight line perpendicular to the polar axis and through the pole.
For those who want to prove this: consider r1 D 2 C a cos 1 and r2 D 2 a cos 2 ; in which
we have introduced indices to be able to make a distinction, then:
8
<x2 D .2 a cos 2 / cos 2
:y D .2 a cos / sin :
2 2 2
Now put 2 D 3 , 3 is nothing more than the supplementary angle of 2 . Then

x2 D .2 a cos . 3 // cos . 3 / D 2 cos . 3 / a cos2 . 3 /
D 2 cos 3 a cos2 3 D .2 C a cos 3 / cos 3 :
This last expression is of the same kind as x1 D r1 .1 / cos 1 whence x2 D x1 .
131
Solutions
Exercise 14 The straight line with cartesian equation x D a has polar equation r cos D a,
a
this is often written as r D or r D a sec . Analogously the line with cartesian equation
cos
y D b has polar equation r D b csc .
y D mx represents a straight line through the origin. Using the transformation formulae
we find
r sin D mr cos ) sin D m cos ) m D tan ) D arctan m
i h
for some value 0 in ; yielding the equation D 0 .
2 2
In general the line with cartesian equation y D mx C q has polar equation
q
r sin D mr cos C q ) r .sin m cos / D q ) rD :
sin m cos
x2
Exercise 15 The equation of the parabola is x 2 D 2ay or y D . A smaller magnitude of a
2a
gives the graph a more closed appearance, sharply curved at the vertex.
Exercise 16 A point A.x; y/ is on the parabola P if the distance of A to d is equal to the

distance of A to F . Or:
ˇ r
ˇ p ˇˇ p 2
A2P , d.A; d / D d.A; F / , ˇx C ˇ D x C y2
2 2
ˇ
ˇ p ˇˇ2 p 2
, ˇx C ˇ D x C y2
2 2
p2 p2
, x 2 C xp C D x 2 px C C y2 , 2px D y 2
4 4
Exercise 17 Let circles C and C1 be tangent in T , then the tangent line t in T to C is also the
tangent line in T to C1 . Now OT ?t and O1 T ?t , so OT k O1 T , so OT and O1 T coincide.
Therefore kO1 Ok D r1 C r.
Exercises 18 and 19 If we denote an external tangency to the first circle as 1E, and an internal
tangency as 1I, we find the following possibilities:
1E2E3E; 1E2E3I; 1E2I3E; 1E2I3I; 1I2E3E; 1I2E3I; 1I2I3E; 1I2I3I:
Some configurations of the initial circles do not allow all 8 solutions.
132 Solutions
Exercise 21
It is obvious from the figure that

8
ˆ
ˆ x D .a b/ cos C b cos
ˆ
<
D a cos
ˆ
ˆ
:̂y D b sin
which are the parameter equations of an ellipse with major axis a and minor axis b.
4AOB is a right-angled triangle. In a right-angled triangle the midpoint of the hy-
potenuse, here ŒAB, is also the centre of the circumscribed circle. Therefore O is on the
circle for which ŒAB is a diameter. Obviously M is the midpoint of both ŒOD and ŒAB
and ]AOB is a right angle. Therefore AOBD is a rectangle. If A .x; 0/ and B .0; y/ then
D .x; y/. Whence
8
< x D l cos ˛
:y D l sin ˛
which reduces to x 2 C y 2 D l 2 , which is the equation of a circle centered at O and with

radius l.
Exercise 22 Select a point A on the y-axis, draw a circle centered at A and with radius a b
and determine the intersection B with the x-axis. On the ray ŒAB determine the point C at a
distance b of B and draw the locus of C with A.
Exercise 23 Put A .0; 0/, †DAB D ˛ and E .x; y/. 4BAD is isosceles, so †BDA D
†DAB D ˛. It is obvious that B .a cos ˛; a sin ˛/. Because D is the mirror image about
the perpendicular on the x-axis and through B we find that D .2a cos ˛; 0/. The coordinates
for E therefore are:
8 8 x
< x D b cos ˛ 2a cos ˛ D .b C 2a/ cos ˛ ˆ
<cos ˛ D
b C 2a
)
:y D b sin ˛ :̂ sin ˛ D y
b
whence
2
x y 2 x2 y2
C D1 , C D 1:
b C 2a b .b C 2a/ 2 b2
133
Solutions
Exercise 24 Draw a circle centered at the origin A and with radius a (say 3), choose a point D
on the x-axis and draw a circle centered at D and with radius a. Determine the intersection
points of the circles, B and C . Draw the ray ŒBD , draw a circle centered at D and with radius
b and determine the intersection point E with the ray ŒBD . Now determine the locus of E
with A.
Exercises 25 and 26 You can find 2 to 6 configurations for which the line can be fitted in. If
pole P is on the straight line a, there is a degenerate solution if the length of the chord which
is subtended equals 3 (D r). In the figures below, we set out a few possible configurations
with the intersections of the locus with the circle.
Exercise 27
134 Solutions
In general we find two curves. When the pole of the neusis is on the circle, we see that both
loci merge into one curve, Pascal’s limaçon (figure under left).
Exercises 28 and 29 Let the second intersection of the diameter through P and the x-axis be Q.
We know 4PBQ is a right-angled triangle, with right angle in B (Thales’ circle theorem).
Let the angle of ŒPB with the x-axis be . Therefore
kPBk D kPQk cos D 2R cos D 10 cos :
For the locus of D with B, we find that r D 10 cos a, and for the locus of E with B that
r D 10 cos C a. These are the polar equations of limaçons.
Exercise 30 In the proof we have never used the property that the triangles are equilateral. We
did use the fact that kBD1 k D kBD2 k. For this assertion it suffices that we have congruent
isosceles triangles.
Exercise 31 We follow a procedure which is similar to that of the construction of a parallel

straight line to a given straight line.
Draw a circle C centered at P and of a chosen radius, larger than d.P; AB/. Let the
intersections of the circle and AB be C1 and C2 respectively. Draw a circle centered at C2
and with radius kC1 C2 k. Let the second intersection of this circle and AB be C3 . Draw circles
with radii kP C1 k and centered at C2 and C3 respectively. Let the intersection of both circles
(on the same side of AB as P ) be P1 . PP1 is parallel to AB.
The proof runs as follows:
Let the magnitude of a straight angle be a. It is clear that 4P C1 C2 Š 4P1 C2 C3 , more-
over because of the construction these are equilateral triangles, so
†P C1 C2 D †P C2 C1 D †P1 C2 C3 D †P1 C3 C2 : (1)
We know from the construction that C1; C2 and C3 are collinear, so
†P C2 C1 C †P C2 P1 C †P1 C2 C3 D a: (2)
In triangle 4C1 P C2 we have that
†C2 C1 P C †C1 P C2 C †P C2 C1 D a: (3)
From (1), (2) and (3) we have †P C2 C1 D †C1 P C2 .
From the SAS property 4P C1 C2 Š 4P C2 P1 , so †P C2 C1 D †C2 PP1 , which proves
the construction.
An alternative construction:
Draw a circle C centered at P and of a chosen radius, larger than d.P; AB/. Let the
intersection of this circle and AB be C1 and C2 . Draw a circle with the same compass width
and the compass point in C1 . This circle intersects AB at C3 . With the same compass width
and the compass point in C3 , draw an arc which intersects the first circle C , on the same side
of AB as P , at Q. Then PQ is parallel to AB. Find the proof yourself.
Hint: kP C1 k D kC1 C3 k D kC3 Qk D kPQk from which 4P C1 C3 Š 4P C3 Q.
Exercise 32 From the construction it is immediately clear that 4ACB Š 4CDE, therefore
†BAC D †ECD. If two straight lines are intersected by a third straight line and the cor-
responding angles of intersection with the transversal are congruent, then the two lines are
parallel. Therefore AB k CE.
135
Solutions
Exercise 33 Draw AP and draw a circle C1 centered at A and with radius kAP k. C1 intersects
AB at C . Draw a circle C2 centered at P and with radius kAP k. C2 intersects AP a second
time at D. Draw a circle C3 centered at D and with radius kP C k. C3 intersects C2 in Q. Then
PQ k AB.
Exercise 34 Let ŒAB denote the line segment. Draw a circle C1 centered at A and with radius
kABk and a circle C2 centered at B and with radius kABk. The two circles intersect each
other at two points C1 and C2 . Draw the straight line C1 C2 . This line intersects ŒAB at the
midpoint, and is therefore called the perpendicular bisector of ŒAB. The distance of any point
P of this line to A always equals the distance of P to B.
We now prove the construction.
We know that the triangles 4ABC1 and 4ABC2 are isosceles, so kAC1 k D kBC1 k D
kAC2 k D kBC2 k. Construct the bisectrix of †AC1 B in C1 (this is the line C1 C2 – check this!)
and let D be the intersection of the bisectrix and ŒAB. Now consider the triangles 4AC1 D
and 4BC1 D. We know that kAC1 k D kBC1 k and both triangles have C1 D in common as
an edge. Moreover because C1 D is the bisectrix †AC1 D D †BC1 D. From the congruence
property SAS we can conclude that 4AC1 D Š 4BC1 D. In other words kADk D kDBk,
so D is the midpoint of ŒAB.
To prove that C1 C2 ?AB we can easily see that †C1 AC2 D †C1 BC2 and †AC1 B D
†AC2 B, from which we conclude that AC1 BC2 is a rhombus. In a rhombus, the diagonals
intersect at right angles, so AB?C1 C2 .
Exercise 35 Place the point of the compass in C . Open the compass at will. With this opening
draw a circle, which intersects the straight line AB at A1 and B1 . Obviously C is the midpoint
of ŒA1 B1 . Construct the perpendicular bisector on ŒA1 B1 . This line passes through C and is
perpendicular to AB.
The proof of this construction is a corollary of the proof in Exercise 34.
Exercise 36 Place the point of the compass in D, choose an opening of the compass in such a
way that the circle will intersect AB. Let the points of intersection of the circle and AB be A1
and B1 respectively. Then kDA1 k D kDB1 k D R, so D is on the perpendicular bisector of
ŒA1 B1 . Construct this perpendicular bisector.
The proof of this construction is a corollary of the proof in Exercise 34.
Exercise 37 Construct l through P , such that l?AB, construct m through P such that m?l,
then m k AB.
Exercise 38 We know that 4AD1 P 4AD3 B and 3 kAD1 k D kAD3 k. From the similarity
of the triangles it follows that
kAD1 k kAD3 k kAD1 k 3 kAD1 k
D and D ;
kAP k kABk kAP k kABk
which proves the equality. The other equalities are corollaries.
Exercise 39 Consider an angle ]ABC . Place the compass point in B and describe an arc of
a circle with a chosen radius. The arc intersects the straight lines AB and BC at D and E
respectively. With the same compass width draw circles centered at D and E respectively. Let
136 Solutions
the second intersection of these circles be F (the first one is B). Then 4BDF Š 4BEF
(SSS) and †DBF D †FBE. BF bisects ]ABC .
Exercise 40
x2
The locus of L with B is part of a parabola with equation x 2 D ay or y D . A smaller
a
magnitude of a gives the graph a more closed appearance, sharply curved at the vertex.
Exercise 41 Put
kABk D kCDk D l and kAC k D kBDk D b;
then
b b
kAF k D kF C k D kAEk D and kEBk D l :
2 2
In the right-angled triangle 4FAB we have
2
2 2 2 b b2
kBF k D kFAk C kABk D l C 2
D l2 C :
2 4
Since point G is on a circle centered at B and with a radius kBF k we find that
s
b2
kBGk D l 2 C :
4
In the right-angled triangle 4GEB we have

2 2 b22 b 2 b2 b2
kGEk D kBGk kEBk D l C l 2
D l2 C l 2 C lb D lb:
4 2 4 4
kGEk equals the area of the rectangle, so GE is the edge of the
We see that the square of p
square we are looking for. lb is sometimes called the geometric mean of l and b.
Exercise 42 There are two pairs of solutions.

Let CA and CB be the given circles and let rB < rA . Construct the circle CAB centered at A
and with radius rA rB . Determine the midpoint M between B and A. Draw an arc of a circle
CM centered at M and with radius kMAk. Let C be one of the intersections of this arc with
137
Solutions
the circle CAB . Draw the straight line AC , which intersects CA at D. Draw the line BC . BC
is at right angles with the radius CA, because of Thales’ circle theorem in CM . Therefore BC
is tangent to CAB . Construct the parallel line t to BC through D. Repeat the procedure with
the other intersection of the arc of the circle CM with the circle CAB .
To find the other pair of tangents, consider the circle CAB centered at A and with radius rA CrB
and repeat the above procedure.
Exercise 43 In both cases you will notice that the radius of the tangent circle increases or
diminishes with the same amount as the radii of the circles CA , CB , CC .
a x y
Exercise 44 D D is shorthand for two equations:
x y 2a
a x x y
D and D
x y y 2a
a y
( D can also be used as an alternative for one of these equations). It is obvious that these
x 2a
equations can be transformed into x 2 D ay and y 2 D 2ax. The two equations x 2 D ay and
y 2 D 2ax represent parabolae, with vertex in the origin, but with the y- and x-axis as axes of
symmetry.
We find the intersection by solving the system of equations
8
< x 2 D ay ) x D y
2
2 2
x
a ) D 2ax:
: 2 a
y D 2ax
p
x D 2a. The intersection points of the parabolae
3
pfor xpyields
Solving this last equation
3 3
therefore are .0; 0/ and 2a; 22 a .
a x y
Exercise 45 D D is shorthand for a system of two equations, e.g.
x y b
a x x y
D and D or
x y y b
a x a y
D and D :
x y x b
The first system yields equations (10) which represent two parabolae, the second system yields
equations (11), which represent a parabola and an (orthogonal) hyperbola.
138 Solutions
Exercise 46 4LED; 4GEL and 4AEG are all right-angled triangles, with right angle in E.
Moreover
†ELD D †EGL D †GAE and †EDL D †GLE D †AGE:
Therefore
4LED 4GEL 4AEG
and
kEDk kELk kGEk 2a y x
D D or D D ;
kELk kEGk kEAk y x a
which can be rewritten as equations (8), which were solved in Exercise 44.
Exercise 47 The equation of an orthogonal hyperbola having the coordinate axes as asymptotes
is xy D k. The equation of the orthogonal hyperbola passing through C therefore is xy D ab.
The equation of the circle circumscribed about the rectangle is

a 2 b 2 a2 C b 2
x C y D :
2 2 4
Therefore we have to solve the system of equations
8
<x 2 ax C y 2 by D 0
: xy D ab
Which yields the equation:

a2 b 2 ab
x 2 ax C b D0
x2 x
) x 4 ax 3 ab 2 x C a2 b 2 D 0
3
) x ab 2 .x a/ D 0
p
3
) x D ab 2 _ x D a
p p
3 3
Therefore the intersections are .a; b/ and ab 2 ; a2 b . The latter coordinates are the mean
proportionals between a and b. p p
Choosing a D 1, b D 2 we find
3
22 ; 3 2 .
Exercise 48 When the slats are in their final position, we see three similar right-angled triangles
4AOB; 4BOC and 4COD.
Then
kOAk kOBk kOC k a kOBk kOC k
D D or D D ;
kOBk kOC k kODk kOBk kOC k 2a
these are the same equations as equations (8), which were solved in Exercise 44. From these,
after some arithmetical manipulations, we find kOBk3 D 2a3 , which is what we had to prove.
139
Solutions
Exercise 49 Repeat the procedure of the previous exercise with kODk D na.
Exercise 50 We use equation (1) (see 7 Sect. 1.2) and put c D 2b, then
y 3 C .x b/xy 2b.x b/2 D 0:
The y-axis is intersected for x D 0, so

p
y 3 C 0 2b.0 b/2 D 0 , y 3 2b 3 D 0 , yD
3
2b:
The configuration of the straight lines when D is on the y-axis is the same as those of the slats
of the winegauger’s instrument.
Exercise 51 Again, we turn to use the properties of similar triangles. We know that
kBAk kEDk kGDk

4BAF 4EDB 4GDE ) D D
kAF k kDBk kDEk
from which
kDEk2 D kDBk kGDk and kAF k kEDk D kBAk kDBk
kDBk2 kBAk2
) D kDBk kGDk ) kDBk kBAk2 D kAF k2 kGDk :
kAF k2
Now
kDBk D 2 kBAk and kGDk D kAF k
from which
p
2 kBAk3 D kAF k3 )
3
2 kBAk D kAF k :
Exercise 52
The locus of H with G is a curve with equation
y 3 D .x C 2a/ .xy a .x C 2a// :

140 Solutions
Exercise 54
]AP C and ]BPD are vertically opposite angles, therefore †AP C D †BPD.
]CAB is subtended by the circular arc CB.
]CDB is subtended by the circular arc CB. Therefore †CAB D †CDB D †PDB.
4AP C and 4BPD have two congruent angles therefore 4AP C 4BPD.
In similar triangles the ratios of the corresponding sides are equal, so
kPAk kPDk
D
kP C k kPBk
) kPAk kPBk D kP C k kPDk
Exercise 55
Obviously †APD D †BPD.

]ADC is subtended by the circular arc AC .
]ABC is subtended by the circular arc AC .
Therefore †ADC D †ABC .
4PAD and 4P CB have two congruent angles, therefore 4PAD 4P CB.
In similar triangles, the ratios of the corresponding sides are equal, so
kPAk kPDk
D
kP C k kPBk
) kPAk kPBk D kP C k kPDk
We now consider a ray through the centre O, which intersects the circle at E and F respec-
tively.
141
Solutions
Now kPEk D kOP k R and kPF k D kOP k C R.

We know that for any intersection points A and B of a straight line and the circle
kPAk kPBk D kPEk kPF k
D .kOP k r/ .kOP k C r/
D kOP k2 R2
Alternatively:
kPAk kPBk D kP T k2
D kOP k2 kOT k2
D kOP k2 R2
The formula is not valid if P is in the inside of the circle, however then
kPEk kPF k D R2 kOP k2
(proof as an exercise).
Exercise 57
x .x C k/ D y .y C 2k/
, x C kx C k 2
2
D y 2 C 2ky C k 2

k 2 3k 2
, xC C D .y C k/2
2 4

k 2 3k 2
, .y C k/2 x C D
2 4
2
k
xC
2 .y C k/2
, D 1
3k 2 3k 2
4 4
142 Solutions
The
p intersections
of this hyperbola with the hyperbola xy D 2k 2 are .k; 2k/ and
3
p
3
22 k; 2k .
Exercise 58 The polar equation of the circle C is r D 2a sin (see Exercise 11), the equation
of the
i straight line t is r D 2a csc , the line OP is given by D for some value of
h
2 ; . Thus in polar coordinates, we find C .2a sin ; / and B .2a csc ; /. Hence
2 2
kCBk .D kOP k/ D 2a.csc sin /:
Let kOP k D r and D and the polar equation of the cissoid becomes
r D 2a.csc sin /:
y3
Using the transformation formulae, we find the cartesian equation x 2 D :
2a y

1
r D 2a.csc sin / ) r D 2a sin
sin
) r 3 sin D 2ar 2 2ar 2 sin2
2
) x C y 2 y D 2a x 2 C y 2 2ay 2 ;
which yields the desired result.

Substituting x D 4a 2y into the equation of the cissoid yields:
p
y3 2 3 4a
.4a 2y/ D2
) 4 .2a y/ D y
3 3
) yD p :
2a y 1C 3 4
And
4a
xD p :
1C
3
4
The slope of OQ is
p
2 3 4a
p p
3
1C 3 4 4 1
mOQ D D D p
4a 2 3
2
p
1C 3 4
1
and its cartesian equation y D p 3
x. It immediately follows that the coordinates of R, the
p 2
intersection of OQ and t are 2a 3 2; 2a . kTRk is equal to the horizontal distance to the
y-axis and thus to the absolute value of the x-coordinate of R.
Exercise 59 Draw a line t with equation y D 2a, now draw the point A.0; a/ and the circle
C centered at A and with radius a. Choose a point B on t and draw the straight line OB.
Determine the instersection point C of OB and C . Draw the circle C2 centered at O and with
radius kCBk. Determine the intersection point P of C2 and OB. The locus of P with B is the
cissoid of Diocles.
143
Solutions
Exercise 60
Let R be the midpoint of ŒPQ. Let †POC D ˛, then †PQB D ˛ (alternate angles).
Let †OP C D ˇ, then †BPQ D ˇ (opposing angles). 4BRQ is isosceles, in a right-
angled triangle the length of the median on the hypotenuse is equal to half the length of the
hypotenuse.
Therefore †QBR D ˛ (equal base angles in an isosceles triangle) and †BRP D 2˛
(exterior angle is equal to the sum of the opposing interior angles).
Furthermore 4BRP and 4BOR are isosceles.
Now
†BOC D †BOP C †POC D †BOR C †POC D 2˛ C ˛ D 3˛:
Therefore the line OPQ is the trisectrix of ]BOC .
Exercise 62 is the angle ]L0 B 0 G 0 . Depending on the position of the slat O 0 B 0 †O 0 B 0 C 0 D

or †O 0 B 0 C 0 D . In the first case
kO 0 C 0 k D kO 0 B 0 k sin ;
in the second
kO 0 C 0 k D kO 0 B 0 k sin . / D kO 0 B 0 k sin :
To find the polar equation we have to determine the length r D kO 0 P k in which P is a point
of the conchoid. It is clear that
c
kO 0 P k D kO 0 B 0 k C kB 0 P k D C b:
sin
Exercise 63 Use the transformation formulae

8
< x D r cos
:y D r sin
which yields
.r sin c/2 r 2 D b 2 r 2 sin2 , .r sin c/2 D b 2 sin2

, r sin c D ˙b sin
c
, rD ˙b .if sin ¤ 0/:
sin
144 Solutions
Exercise 64 This branch (as well as the other) is defined as the locus of a neusis with pole
A and the x-axis as the directrix. For each point on C on the x-axis, there is a point on the
straight line AC for which kEC k D 2 kABk. Draw the circle centered at A and with radius
2 kABk. The circle intersects the x-axis at C1 and C2 . For these points, the point on AC at a
distance 2 kABk is A itself. Therefore A is twice on the conchoid.
Exercise 65 Because kDC k D kCOk D kOBk D r, 4DCO and 4COB are isosceles. From
which †ODC D †COD and †OCB D †CBO. Now the magnitude of an outer angle of a
triangle equals the sum of both opposing inner angles, from which
†OCB D †ODC C †COD and

†AOB D †ODC C †CBO D †ODC C †OCB D †ODC C †ODC C †COD
D 3†ODC D 3†ADB:
Exercises 66 and 67 If you want to trisect an angle you will find angles with magnitude
2 4
; C ; C rad in the intersections. Multiply these angles by 3 and reduce them to
3 3 3 3 3
Œ0; 2 and you will find .
Exercise 68 On the x-axis, construct a perpendicular through O. For the construction, we

confine ourselves to the first quadrant, the constructions in the other quadrants are analogous.
Construct the bisector of a right angle. Now bisect the angles defined by the bisector and the
x-axis and the perpendicular respectively.
Exercise 71 If you use Archimedes’ spiral to trisect an angle you will find angles with mag-
2 4
nitude ; C ; C rad, using the first, second and third turn respectively. Multiply
3 3 3 3 3
these angles by 3 and reduce them to Œ0; 2 and you will find .
Exercise 72
All shaded triangles in the figures have the same area, which proves Gregory’s generalisation.
The method used here is the same method as the one Euclid used when proving Pythagoras’
theorem.
Consider the triangles in figures (a) and (b). They have the same area because they have the
same base ŒAB (the edge of the square) and they have the same altitude kDEk. The triangles
in figures (b) and (c) are congruent: kDAk D kABk, kALk D kAC k, †DAC D †BAL
(SAS-property). The triangles in figures (c) and (d) have the same area because they have the
same base ŒAL and the same altitude kLKk.
145
Solutions
Now repeat this procedure for the rectangles in the other shades of gray.
Exercise 74
kBEk D kEOk so 4BED is isosceles.

†EOB D †EBO, now ]EOB is a central angle and ]FBG is an inscribed angle.
_ _
†FOG D 2†FBG D 2†EOB ) F G D 2AB.
_ _ _ _
Now AB D GC and F G D 2GC .
Therefore †CAF D 3†CAG.
Exercise 75 The quadratrix is defined as the curve for which

†BAX kXX 0 k
D :
†BAD kDAk

Put kDAk D R, we know †DAB D . If †XAB D and kAX k D r then
2
kXX 0 k D r sin :
The equation becomes
r sin
D ;
=2 R
which yields the desired equation.
Exercise 75 Draw a circle with radius R (e.g. 5). Draw a point D 0 on the y-axis and draw the
y kAD 0 k
horizontal line D 0 C 0 . Since D it follows that D , therefore the slope
=2 R
2R
0
kAD k
of the line AF is given by m D tan and its equation is given by y D mx.
2R
Determine the intersection point X of D C and AF . Determine the locus of X with D 0 , this
0 0
locus is the quadratrix.

146 Solutions
Exercise 78
The names of the points are indicated on the figure. Draw the radius ŒMD. We know that the
tangent segments ŒOD and ŒON have the same length because 4ONM Š 4ODM . Also
4ONA Š 4ONM .
This implies that †AON D †NOM D †MOD, from which
†AOD D †AOB D 3†AON:
Exercise 79
sin 3˛ D sin .˛ C 2˛/ D sin ˛ cos 2˛ C cos ˛ sin 2˛ D : : : D 3 sin ˛ 4 sin3 ˛
cos 3˛ D cos .˛ C 2˛/ D cos ˛ cos 2˛ sin ˛ sin 2˛ D : : : D 4 cos3 ˛ 3 cos ˛
sin 3˛ 3 sin ˛ 4 sin3 ˛
tan 3˛ D D
cos 3˛ 4 cos3 ˛ 3 cos ˛
3 sin ˛ 4 sin3 ˛ sin ˛
3˛
3 3 4 tan3 ˛
D cos D cos ˛
4 cos3 ˛ 3 cos ˛ 1
3 2 C 4
3
cos ˛ cos ˛
sin ˛ 1
3 2˛
4 tan3 ˛ 3 tan ˛ 1 C tan2 ˛ 4 tan3 ˛
D cos ˛ cos D
1 3 .1 C tan2 ˛/ C 4
3 2
C 4
cos ˛
3 tan ˛ tan3 ˛
D
1 3 tan2 ˛
Exercise 80
147
Solutions
From the left figure it is obvious that, using the sine rule,
r a r a
D ) D
sin . 3 / sin 2 sin 3 sin 2
which, using the results of Exercise 79, yields:
r a
D
3 sin 4 sin
3 2 sin cos
r a
) D
3 4 sin2 2 cos
!
a 3 4 sin2
) rD
2 cos cos
!
a 3 4 1 cos2
) rD
2 cos cos

a 1
) rD 4 cos
2 cos
a
) r D .4 cos sec /
2
To find the cartesian equation, we use the transformation formulae

a 1
rD 4 cos
2 cos
a r
) r2 D 4r cos
2 cos

a r2
) r2 D 4r cos
2 r cos

a x C y2
2
) x Cy D
2 2
4x
2 x
2
) 2x x C y 2 D a 4x 2 x 2 C y 2

) 2x x 2 C y 2 D a 3x 2 y 2
Exercise 81 See figure in Exercise 80 right.

The equation of the straight line through O is y D mx, in which m D tan .
The equation of the straight line through A.a; 0/ is y D n.x a/, in which
3m m3
n D tan 3 D
1 3m2
(see Exercise 79).
The intersection of these lines is given by the system:
8
<y D mx
:y D 3m m .x a/
3
1 3m 2
148 Solutions
Eliminating m from these equations yields:

y y 3
3
y D x x 2 .x a/
y
13
x
3yx 2 y 3
3
3
) y D x3 x .x a/
x 3y 2 x
x3
3yx y 3
2
) yD 3 .x a/
x 3y 2 x
3
) y x 3y 2 x D 3yx 2 y 3 .x a/

) yx x 2 3y 2 D y 3x 2 y 2 .x a/
) 2x 3 C 2xy 2 3x 2 a C y 2 a D 0

) 2x x 2 C y 2 D a 3x 2 y 2
The latter equation is the equation of MacLaurin’s trisectrix.
Exercise 82
4OBD and 4BDP are isosceles. The configuration is the same as in the proof of Archimedes’
trisection (see proof of Exercise 65).
r R
D
sin . 3 / sin
R sin 3
) rD
sin
3 sin 4 sin3
) r DR
sin

) r D R 3 4 sin2
To find the cartesian equation, we use the transformation formulae

r D R 3 4 sin2

r 3 D R 3r 2 4r 2 sin2
2 p
x C y2 x 2 C y 2 D R 3 x 2 C y 2 4y 2
2 p
x C y2 x 2 C y 2 D R 3x 2 y 2
2 3 2
x C y 2 D R2 3x 2 y 2
149
Solutions
Exercise 83
Draw circles C1 centered at O and with radius R and C2 (not shown in the figure),
centered at B and with radius R. C2 intersects the x-axis in D. 4OBD is isosceles
.kOBk D kBDk D R/ Therefore the median through B and the altitude a on ŒOD and
through B coincide. Therefore D is the mirror image of O with respect to a. Draw circle
C3 centered at D and with radius R. Let P be the intersection of C3 and OB. Draw the
perpendicular e to OB and through D. 4BDP is isosceles so P is the mirror image of B
with respect to e.
Exercise 84 Some of the curves which you can find.
2r 2r
Exercise 85 The smaller you choose , the closer D will be to the number .
sin
Exercise 86
kOP k D cos I kQP k D sin I kRSk D tan

1 1
area 4OPQ D kOP k kQP k D cos sin
2 2
1
area circular sector QOS D kORk D
2 2
1 1
area 4ORS D kORk kRSk D tan
2 2
150 Solutions
From the figure it is obvious that

area 4OPQ < area circular sector QOS < area 4ORS
1 1
) cos sin < < tan ) cos sin < < tan
2 2 2
Dividing these inequalities by sin yields
1 1
cos < < tan D
sin sin cos
If D 0 both the left and right-hand sides are equal to 1, which would tempt one to conclude

that D 1, which is of course incorrect, because one cannot divide by 0.
sin

We can however say that if 0 then 1 or mathematically lim D 1.
sin !0 sin
Exercise 87 4GMH 4EMF because GH and AF are parallel lines by construction.

Parallel lines intersect other straight lines at the same angles. From the similarity we deduce
that
kMH k kGM k kMH k 1
D ) D ) kMH k D :
kMF k kMEk 1 1=
In any right-angled triangle the altitude on the hypotenuse is the mean proportional of the two
segments of the hypotenuse.
Hence
p
kFM k kMH k D kKM k2 ) 1 D kKM k2 ) kKM k D :
Exercise 88 Pythagoras’ theorem for a right-angled triangle 4ABC , right-angled A, states:

kBC k2 D kBAk2 C kAC k2 .
The area of the semi-circle on BC is

kBC k 2
SCBC D ;
2
the area of the semi-circles on BA and AC are resp.

kBAk 2 kAC k 2
SCBA D and SCAC D :
2 2
Now

kBAk 2 kAC k 2
SCBA C SCAC D C D kBAk2 C kAC k2
2 2 4 4
2 2
2
D kBAk C kAC k D kBC k D SCBC :
4 4
Exercise 89 The ratio between the areas of similar figures is equal to the square of the ratio of
corresponding lengths (e.g. a diameter) of those figures. Here we take the lengths of the sides
as corresponding lengths and obtain:
A B C b2 c2 b2 C c2
D 2 D 2 ) B CC D AC 2A D A D A:
a2 b c a 2 a a2
151
Solutions
Exercise 90
We use the semi-circular version of Pythagoras’ theorem as proved in the previous exercise.
In figure (a) we have that A C A1 C A2 D B1 C B2 , in figure (b) B1 D A1 C C1 and in
figure (c) B2 D A2 C C2 .
Substituting the values for B1 and B2 in the first equation yields
A C A1 C A2 D A1 C C1 C A2 C C2 ;
which simplifies to
A D C1 C C2 D C:
1 2
Exercise 93 To show that the area of a circle sector equals S D r ˛, we use the rule of three:
2
2 ˛
2
D :
r S
In the left-hand side of the equation we have used a special sector: a full circle.
The area of the circle sector BMA is equal to
p !2
1 2 1 2
S D kBM k .†BMA/ D kBAk D kBAk2 ;
2 2 2 2 8
which is equal to the area of the semi-circle on ŒBA.
The area of the lune on AB is given by:
area lune D SCAB sector BMA C 4BMA
kABk2 1
D kBAk2 C kBM k2
8 8 2
1
D kBAk2 :
4
Exercise 95 Referring to . Fig. 6.8 we know from the previous exercise that
SCAB C SCAC D SCBC ;
We deduct areas III and IV from both sides (these are the areas defined by the perpendiculars
and the semi-circle on the hypotenuse):
SCAB C SCAC D SCBC
area III C area IV D area III C area IV

area I C area II D area 4ABC
152 Solutions
Exercise 96
Referring to the figure, let T1 denote the area of the right-angled triangle to the left of the alti-
tude on the hypotenuse, and T2 that of the area of the triangle to the right, so that T1 CT2 D T .
Again, we turn to use Pythagoras’ semi-circular theorem.
In the left triangle II C a C T1 and I C b C T2 are the areas of the semi-circles on the legs of
a right-angled triangle, so they sum to the area of the semicircle on the hypotenuse A C a C b,
from which
II C a C T1 C I C b C T2 D III C a C b;
or
T1 C T2 D III II I;
which had to be proved.

In the right-angled triangle, we use the results of Exercise 95 and find that
.II C III/ C .I C IV/ D T1 C T2 D T:
Exercise 97 Let R denote the radius of the original circle and S, C and L the areas of the
circular segment to the left of the dashed chord, the white circle and Leonardo’s Claw, respec-
tively.
Then
R2 R2
SD ;
4 2
i.e. the difference between the area of the circular sector and the isosceles triangle (see fig-
ure (a)). p
R 2R
The altitude in the isosceles triangle is equal to p D .
2 2
153
Solutions
p p
Therefore the thickness of the lens is 2R 2R D 2 2 R. Subtracting from the di-
p
ameter ofpthe large circle, we find that the diameter of the small circle is 2R; therefore its
2R
radius is (see figure (b)).
2
The area of Leonardo’s Claw becomes:

1 R2 R2
L D R2 C 2S D R2 R2 2 D R2
2 4 2
p
The diagonal of the square equals 2R, therefore the side of the square is R and its area R2
(see figure (c)).
Exercise 98 The trapezoid ABCD is isosceles and DC k AB. Let the perpendicular bisectors
be mAD , mDC and mCB and the midpoints MAD , MDC and MCB . The bisector mCD is an
axis of symmetry (mirror reflection) of the trapezoid (check for yourself that †A D †B and
deduce that mCD is an axis of symmetry). mAD intersects mDC at a point O. The mirror image
of mAD is a line l through O. The mirror image of MAD with respect to mDC is MCB . So line
l passes through MCB . The property “at right angles” is invariant under a mirror reflection, so
image mAD ?image AD or l?CB. This means that line l is the perpendicular bisector mCB
of CB. Therefore the perpendicular bisectors mAD ; mDC and mCB intersect at one point O,
which is the centre of the circumscribed circle.
Exercise 99 4AOD 4AED ) †AOD D †AEB. Let †AOD D ˛, then
area circular segment AD D area circular sector AOD area 4AOD

1 1
D kOAk2 ˛ kOAk kODk sin ˛
2 2
1
D kOAk2 .˛ sin ˛/
2
154 Solutions
On the other hand
area circular segment AB D area circular sector AEB area 4AEB

1 1
D kEBk2 ˛ kEBk kEAk sin ˛
2 2
1
D kEBk2 .˛ sin ˛/
2
3
D kAOk2 .˛ sin ˛/
2
D 3 area circular segment AD
Exercise 100
Because 4ABE 4ADO, †AOD D †AEB and †AOB D 3†AOD we have
†AOB D 3†AEB
(all angles are expressed inpradians).

Furthermore kAEk D 3 kAOk.
Therefore
1 3
area circle sector AOD D kAOk2 .†AOD/ D kAOk2 .†AEB/
2 2
and
1 1 p 2
area circle sector AEB D kAEk2 .†AEB/ D 3 kAOk .†AEB/
2 2
3
D kAOk2 .†AEB/
2
sector AOB D I C II
sector AEB D II C III
) 0 D I III
) I D III
155
Solutions
Exercise 101 The radius of the small circles is half that of the large circles, therefore four times
their area equals the area of the large circle.
Exercise 102 From Exercise 84 we deduce that
area semi-circle AD D area semi-circle DC D area semi-circle CB D area semi-circle EF
and
4 area semi-circle AD D area semi-circle AB
which we can also write as
3 area semi-circle AD C area semi-circle EF D area semi-circle AB
or
area semi-circle EF D area semi-circle AB 3 area semi-circle AD

semi-circle EF D area trapezoid AB C 3 area VI 3 area semi-circle AD
semi-circle EF D area trapezoid AB .3 area VI 3 area semi-circle AD/
semi-circle EF D area trapezoid AB 3 area lune III
Exercise 103 Referring to . Fig. 6.14
S D area semicircle CB area VI

D area semicircle CB .area circular sector COB area 4COB/

kOBk 2 kOBk2 kOBk2
SD C sin
2 2 2 3 2 3
p
3
D kOBk2 kOBk2 C kOBk2
8 6! 4
p
3
D C kOBk2
24 4
156 Solutions
We notice that the area of the lune is a function of . Therefore we do not know, and neither
did Hippocrates, whether we can construct this lune with compass and straightedge because
we do not know whether we can construct a line segment of length (see 7 Chapter 7 and
esp. 7 Sect. 7.3 and 7 Sect. 7.4 for an explanation).
Exercise 104 We will first suppose that the pentagonal figure is already constructed. From this,
we will deduce the lengths of certain line segments, which will then allow us to actually draw
the pentagonal figure.
E, D, B are collinear, therefore †EBA D †DBA. Furthermore 4ADB and 4EAB are
isosceles. Two isosceles triangles are similar r
if they have equal
r base angles.
3 3
We have a D kEAk D kABk, kDEk D kEAk D a and x D kDBk.
2 2
kDBk kEAk
From 4ADB 4EAB it follows that D .
r kABk kEBk
3
Now kEBk D x C a so
2
r ! r
x a 3 3
D r )x xC a Da )x C
2 2
ax a2 D 0
a 3 2 2
xC a
2
which has solutions
r r !
3 11 a
xD ˙ ;
2 2 2
of which in this geometrical context

r r !
3 11 a
xD C
2 2 2
satisfies the condition of being positive.

A line segment with length x!can easily be constructed. Draw a circle centered at B and
r r
3 11 a
with radius x D C . One intersection of this circle and the line PQ is the
2 2 2
vertex D.
157
Solutions
Above, we have only used quadratic equations, the solutions to which are constructible
with compass and straightedge (see Exercise 131). Therefore we do not need a neusis.
Exercise 105
4AO1 B Š 4BO1 C ) †ABO1 D †O1 BC ) †ABM D †MBC:
Furthermore kABk D kBC k and †CAB D †ACB (4ABC is isosceles)
) 4AMB Š 4CMB
) †AMB D †BM C and ]AM C is a straight angle
) ]AMB and ]BM C are right angles
Exercise 106 Let

†DO2 C D ˛: (1)
p
The sector O2 EDC has a radius 3a and an interior angle †EO2 D D 2˛, therefore
1 p 2
area O2 CDE D 3a 2˛ D 3a2 ˛:
2
p
From (1) and (15) †BO1 C D ˛. The circular sector O1 CBAE has radius 2a and
†EO1 C D 3˛; therefore
1 p 2
area O1 CBAE D 2a 3˛ D 3a2 ˛:
2
Exercise 107 Falco constructs a curvilinear figure in which the two curves QNP and RSO are
a translation of each other along the other (straight) edges. A consequence is that the piece
which looks to have been cut from the rectangle is added on the other side.
The total area therefore remains unchanged. In Falco’s figure, PNM and OSV have the same
area. OSV has been “cut” from the rectangle and added again on the top. The same goes for
the figures QGN and RT S. The curvilinear figure RQP V therefore has the same area as the
rectangle T GM V .
Exercise 108
158 Solutions
5
It is clear that half a diagonal of the square is equal to R, therefore to find an edge E we can
4
apply Pythagoras’ theorem
2 2
5 5 5 p
E2 D R C R ) ED R 2:
4 4 4
If we assume that the area of the square and the circle are equal then
25 2 25
S D E 2 D S ) R D R2 ) D D 3 18 :
8 8
p p
Exercise 109 Suppose 2D
, and p; q 2 N. Furthermore p and q have no common factors,
q
in other words the fraction is irreducible. If we square both sides of the equation we find
p2
D2 ) 2q 2 D p 2 :
q2
So p 2 is even, and so is p. We can write 2p 0 D p. The equation becomes
.2p 0 /2
D2 ) 2q 2 D 4p 02 ) q 2 D 2p 02 :
q2
Then q 2 is even, and so is q. We can write 2q 0 D q. The equation becomes

p 2p 0 p0
D D :
q 2q 0 q0
p
But we started out with the assumption that was irreducible, so we have a contradiction. In
p q
other words, 2 can not be written as a rational number.
Exercise 110
HBDL is a rectangle, KHBI is a parallellogram. For a parallellogram, the area is produced

by base height. For KHBI ŒHB is a base and ŒBD the height, from which:
p
area KHBI D kHBk kBDk D 3 2 11 D area HBDL: (1)
159
Solutions
On the other hand, we can also consider BI as the base. In which case
area KHBI D kHZk kKH k

p
D kHZk 11 C 2
p
D 11 3 2 from (1)
p p p
p 3 2 11 2 7 11 5 2 2
3 2
kHZk D 11 p D 11 D
11 C 2 121 2 7 17
p
D 1:409 : : : 2
2 2 2
Exercise 111 Let kABk D kBC k D kCAk D 1 also kCEk C kEBk D kCBk from which
p
3
kCEk D :
2
4ABC is equilateral, so the respective altitudes, medians and angle bisectors coincide. The
centroid therefore is also orthocentre, circumcentrepand intersection of the angle bisectors.
1 3
Since D is the centroid kDEk D kCEk D .
3 6
1
We also know kEKk D kKBk D .
4
7
So kDKk2 D kDEk2 C kEKk2 D and
48
r p p
7 21 5 5 21
kDKk D D and kDLk D kDKk D :
48 12 4 48
According to Nicolas of Cusa kDLk is the radius of the circle isoperimetric to the triangle.
So 2 kDLk D 3 and
p p
3 144 144 21 62 21
D D p D D :
2 kDLk 10 21 210 105
Exercise 112 4ABC is equilateral, so the respective altitudes, medians and angle bisectors

coincide and †EBD D .
6
In 4DEB:
p
3 p
kEBk D kDBk cos D 60 D 30 3 and
6 2
1
kDEk D kDBk sin D 60 D 30:
6 2
1 1
Because kEBk D kABk and kEKk D kKBk D kABk:
2 4
kEBk p
kEKk D kKBk D D 15 3:
2
160 Solutions
4DEK is right-angled in E, from which

p
kDKk2 D kDEk2 C kEKk2 D 1575 D 7:225 and kDKk D 15 7:
Furthermore
5 75 p
kDLk D kDKk D 7:
4 4
According to Nicolas of Cusa
75 p p p
1 2 kDLk 7 75 7 5 7
D D 2 p D p D p
6 kEBk 6 30 3 12 30 3 24 3
At first glance, this expression does not resemble Cusa’s own result in 7 Sect. 6.5, but after
some manipulation we find:
p p p p p
2 12 1575 2:5 7 225 2:5 15 7 5 15 7 5 7
p D p D p D p D p
6 2700 6 3:900 6 30 3 12 30 3 24 3
which is, of course, the same value.
Exercise 113
p
2
Let kABk D d . Using Pythagoras’ theorem, it is clear that kMAk D d and therefore
2
p
2 d
kEGk D d :
2 2
So p p
21 d 2C2
kMN k D dC D d:
6 2 6
If the circle and the square have the same area, then
p !2 0 p 12
2 6 2 2
2C2 6
d D d2 ) D p D@ A 3:08831 : : :
6 2C2 2
Exercise 114 The values of b and c converge to a number with the same decimals as and
b < < c.
161
Solutions
Exercise 115
Let the radius of the circle be r and suppose we have an inscribed regular n-gon pn and a
circumscribed n-gon Pn . In triangle 4Oab we have that
2
kOak D kObk D r and †aOb D :
n

Since 4Oab is isosceles, 4Oma is right-angled, †aOm D therefore
n

kamk D r sin and kabk D 2r sin :
n n
The total circumference thus is

pn D 2nr sin :
n

Triangle 4OF G is isosceles and kOf k D r therefore kf Gk D r tan and the circumfer-
n

ence is Pn D 2nr tan .
n
It immediately follows that

pn 2nr sin
D n
Pn D cos n :
2nr tan
n
Similarly we find for the area:
1
An D n r 2r tan D nr 2 tan
2 n n
and

an D n r cos 2r sin D nr 2 cos sin
n n n n
from which

an nr 2 cos sin cos
D n n D n D cos2
An 1 n
nr 2 tan
n
cos
n
162 Solutions
Exercise 116
p
1 3
We know that kAC k D D sin , from which kOC k D D cos and so
2 6 2 6
p
3
kCDk D 1 :
2
We see that 4OCA 4ODE from which
kAC k kEDk
D
kOC k kODk
so
1 p
2 kEDk 1 3
p D and kEDk D p D :
3 1 3 3
2
Exercise 117
q p
2 4 .2 sin 2˛/2 D 2 4 4 sin2 2˛
D 2 2 cos 2˛

D 2 2 1 2 sin2 ˛
D 2 2 C 4 sin2 ˛
D .2 sin ˛/2
Using this formula:
q
.2 sin 2˛/ D 2 4 .2 sin 4˛/2
2
(33)
q
.2 sin ˛/2 D 2 4 .2 sin 2˛/2
s
q
D 2 4 2 4 .2 sin 4˛/ 2
r q
D2 2C 4 .2 sin 4˛/2
163
Solutions
This is the basis for a proof by induction.

v v
u u s
u u r
u t q
t
.2 sin ˛/ D 2 2 C 2 C 2 C : : : 2 C 4 .2 sin 2n ˛/2
2
with n 1 2s under the square roots.

The proof of the induction step is essentially the same as proving equation (33).
Alternatively, one can also write:
v v
u u s
u u r
u t q
˛ 2 t
2 sin n D 2 2 C 2 C 2 C : : : 2 C 4 .2 sin ˛/2
2
The edge of a regular n-gon inscribed in a circle with radius 1 equals 2 sin ˛, in which ˛ is the
central angle subtending the edge as its chord.
Exercise 118 With reference to the figure of Exercise 116, we see that:
1 1
kAC k kEDk sn Sn
D , s 2 2 D 2 ;
kOC k kODk 1 1
1 sn
2
2sn
which after simplification yields Sn D p .
4 sn2
Exercise 120 Use B.i/ D ROOT.2 ROOT.4 .B.i 1//O2//
2i n length edge perimeter

6 1 6 3
12 0:51763809 : : : 6:211657 : : : 3:105829 : : :
24 0:261052384 : : : 6:265257 : : : 3:132629 : : :
48 0:130806258 : : : 6:2787 : : : 3:141032 : : :
96 0:065438166 : : : 6:282064 : : : 3:141452 : : :
192 0:032723463 : : : 6:282905 : : : 3:141584 : : :
384 0:016362279 : : : 6:283115 : : : 3:141590 : : :
Exercise 121 We will use the method of complete induction.

We notice that the assertion holds for small values of n.
X
1
1
If n D 1 then i D 1 and 1 .1 C 1/ D 1:
i D1
2
X
2
1
i D1
2
X
3
1
i D1
2
164 Solutions
Now suppose the assertion holds for n D k, will the assertion also hold for n D k C 1?
X
kC1 X
k
1
iD i C .k C 1/ D k.k C 1/ C .k C 1/
i D1 i D1
2

1 k C2
D k C 1 .k C 1/ D .k C 1/
2 2
1 1
D .k C 2/.k C 1/ D .k C 1/ ..k C 1/ C 1/
2 2
This implies that the assertion holds for all positive integers.
Exercise 123 We will use the method of complete induction.

We notice that the assertion holds for small values of n.
X
1
123
If n D 1 then i 2 D 1 and D 1:
i D1
6
X
2
235
i D1
6
X
3
347
i D1
6
Suppose that the assertion holds for n D k, will the assertion also hold for n D k C 1?
X
kC1 X
k
k .k C 1/ .2k C 1/
.i C 1/2 D i 2 C .k C 1/2 D C .k C 1/2
i D1 i D1
6

.k C 1/ .k .2k C 1/ C 6 .k C 1// .k C 1/ 2k 2 C 7k C 6
D D
6 6
.k C 1/ .k C 2/ .2k C 3/
D
6
.k C 1/ ..k C 1/ C 1/ .2 .k C 1/ C 1/
D
6
which concludes the proof.
The area under the parabola is approximated by:
X n
i 2b3 b3 X 2
n
b 3 n .n C 1/ .2n C 1/
D i D
i D1
n3 n3 i D1 n3 6
b 3 n C 1 2n C 1
D
6 n n
So
b 3 n C 1 2n C 1 b3 b3
lim D 12D :
n!1 6 n n 6 3
165
Solutions
1
Exercise 125 Inserting a D 0 and b D 1 in equation (17) yields an area between the straight
2
1
line and the x-axis. Inserting b D 1 in the results of Exercise 123 yields an area between the
3
parabola and the x-axis. Therefore the area between the parabola and the straight line equals
1 1 1
D .
2 3 6
Exercise 127
Obviously
4OAC 4OBD
kOAk kOBk
) D
kOC k kODk
b kOBk
) D
1 a
) kOBk D ab
which proves the assertion.
To construct a2 , a3 and a4 , draw two intersecting straight lines l and m, let kOAk D 1 and
kOBk D a. Draw a circle with centered at O and with radius a. The circle intersects m
in P . Draw line AP and the parallel line through B. The intersection of this parallel line
with m is point Q. Now by the construction kOQk D a2 . Draw a circle centered at O and
with radius a2 , this circle intersects line l in C . Draw a parallel to AP through C , this line
intersects m in R. Now by the construction kORk D a3 .
To construct a4 , draw the line AQ and the parallel to it through C , this line intersects b
at T . Now by the construction kOT k D a4 .
166 Solutions
Exercise 128
Select points D, A and B in such a way that kODk D a, kOAk D 1 and kOBk D b. Connect
B and D, draw a parallel l line to BD through A. Then kOC k D a=b in which C is the
intersection point of l and OD.
Exercise 130
p p p 2
42 3D 32 3C1 D 31 ;
p p p 2
97 56 3 D 49 56 3 C 48 D 7 4 3 and
q q
p p p p
4 4 2 3 C 97 56 3 D 4 3 1 C 7 4 3 D 3:
p p
6 6 6 3C 3 6 3 3
p C p D C D 6:
3 3 3C 3 93 93
b p
Exercise 131 (a) We know that kABk D b, kAM k D kMBk D and kM C k D c.
2
b
By construction we also know that kCDk D . Therefore kCDk2 D kCM k2 C kMDk2 or
2
p
b2 b 2 4c b 2 4c
kMDk2 D c D and kMDk D :
4 4 2
So
p p
bC b 2 4c b b 2 4c
kADk D and kDBk D ;
2 2
which are the solutions to the quadratic equation x 2 bx C c D 0.
167
Solutions
b p
(b) Again we know that kABk D b, kAM k D kMBk D and kM C k D c. Let E be
2
the point on ŒAC for which kAEk D kAM k.
p
b 2 C 4c
By construction we also know that kAC k D , therefore
2
p p
b 2 C 4c b b 2 C 4c b
kEC k D D ;
2 2 2
which is a solution to the quadratic equation x 2 C bx c D 0.
The second solution is negative and therefore cannot be constructed.
(c) To divide a line segment into two segments, denote the parts as x and a x. The
condition can be written as the equation
x 2 .a x/2 D x .a x/ or x 2 C ax a2 D 0;
which can be solved using the method of (b). To show that the longer segment is the
mean proportional between the shorter segment and the whole line, we rewrite the equa-
tion x 2 C ax a2 D 0 as
x ax
x 2 D a2 ax , x 2 D a.a x/ , D ;
ax x
which proves the assertion.
Exercise 132 If the coordinates of the points are constructible, then the cartesian equations of
the straight lines through them have constructible coefficients. Let y D m1 x C q1 be the
equation of AB and y D m2 x C q2 the equation of CD. The intersection of these lines is
given by the system of equations:
8
<y D m1 x C q1
:y D m x C q
2 2
The solution to this system are rational expressions of constructible numbers, and therefore
these are also constructible
8 q2 q1
ˆ
<x D m m
1 2
:̂y D m1 q2 m2 q1
m1 m2
168 Solutions
Let x 2 2ax C y 2 2by C c D 0 be the equation of the circle centered at C and with radius
kCDk. The coefficients in this equation are constructible because they are rational expressions
of constructible numbers.
To find the intersection with AB, we need to solve the system
8
< y D m1 x C q1
(34)
:x 2 2ax C y 2 2by C c D 0
Substituting the value of y from the first equation into the second equation yields the equation
ux 2 C vx C w D 0: (35)
This is a quadratic equation in which u, v and w are rational expressions of constructible

numbers and which therefore are constructible. The solution is either a rational expression
or a rational expression which contains a square root. So this equation yields a constructible
number.
The coordinates of the intersections of two circles (if they exist) are given by the system
of quadratic equations:
8
<x 2 2a1 x C y 2 2b1 y C c1 D 0
:x 2 2a x C y 2 2b y C c D 0
2 2 2
Subtracting the second equation from the first equation yields:

8
<2.a1 a2 /x 2.b1 b2 /y C c1 c2 D 0
: x 2 2a2 x C y 2 2b2 y C c2 D 0
This system is of the same kind as the system (34), therefore the conclusion is the same.
Exercise 133 See also Exercise 79
sin 3˛ D sin .˛ C 2˛/ D sin ˛ cos 2˛ C cos ˛ sin 2˛ D : : : D 3 sin ˛ 4 sin3 ˛:
or, using the assumption, c D 3y 4y 3 .
d3 D 3x x 3 ) 2 sin 3˛ D 3 .2 sin ˛/ .2 sin ˛/3

) 2 sin 3˛ D 6 sin ˛ 8 sin3 ˛
) sin 3˛ D 3 sin ˛ 4 sin3 ˛:
Exercise 134
169
Solutions
From Exercise 65 we know that

a D kOBk D kPRk D kQRk :
Moreover 4PBQ, 4P CO and 4BEQ have two congruent angles in common, the right
angle and †POC D †PQB D †BQE.
Consequently
4PBQ 4P CO 4BEQ;
which yields the continued proportion
kBQk kCOk kEQk
D D :
kPQk kPOk kBQk
4ORB is isosceles, so
1 1
kERk D kORk D .y C a/ ;
2 2
and
1 y C 3a
kEQk D kERk C kRQk D .y C a/ C a D :
2 2
Whence
x b y C 3a
D D :
2a y 2x
Put a D 1 then b D cos 3˛ and
8
<xy D 2b
) x 3 3x 2b D 0
: x2 D y C 3
Exercise 135 Suppose the line segment ŒOA has a length sin ˛. Draw a circle centered at O and
with radius 1. Erect the perpendicular to ŒOA at A. This perpendicular intersects the circle
at B. Then †ABO D ˛.
Suppose †ABO D ˛ is given. Draw a circle centered at B and with radius 1. This circle
intersects leg AB at C . From C , construct the perpendicular on BO, let the foot be D. The
triangle 4BCD is a right-angled triangle in which the legs have length cos ˛ and sin ˛.

Exercise 136 d3 D x 3 3x, if ˛ D rad and R D 1 then 1 D x 3 3x ) x 3 3x1 D 0.
6
Potential solutions are ˙1 neither of which satisfies the equation, therefore the equation has
neither rational nor constructible solutions. p
3 p 3
x 3x 2b D 0, now b D a cos
3
D a. Select a D 3 then b D and
6 2 2
x 3x 3 D 0, with potential solutions ˙3, neither of which satisfies the equation, therefore
3
the equation has neither rational nor constructible solutions.
1 1
Exercise 137 We know that sin D and cos D . These are rational numbers and thus
6 2 3 2

constructible. Exercise 135 allows us to construct the angles rad and rad. Therefore a
6 3
right angle can be trisected.
For a right angle equation (19) becomes: 4x 3 3x C 1 D 0. It is clear that x D 1 is a
solution. All solutions to this equation are by consequence constructible.
170 Solutions
Exercise 138 Hippocrates used two assumptions to find constructible lunes:

(1) Two circular sectors (with radii r and R respectively) corresponding to the lunes’ arcs
have the same area
(2) The central angles (2˛, 2ˇ) of the two circular arcs are commensurable
˛ n r2
We will show that D D 2.
ˇ m R
Let 2˛ be the central angle of the arc centered at A.
Let 2ˇ be the central angle of the arc centered at B.
It is obvious that ˇ > ˛.
Let be the greatest common denominator of 2˛ and 2ˇ which, by assumption (2), are
commensurable.
˛ n
Therefore ˛ D n and ˇ D m for some m > n 1 and D .
ˇ m
Subdivide SA D ACED into n congruent sectors of central angle and centered at A.
Subdivide SB D BCFD into m congruent sectors of central angle and centered at B.
Now by assumption (1), the two circular sectors have the same area. Therefore the n
smaller sectors centered at A and with radius R taken together and the m smaller sectors
centered at B and with radius r taken together have the same area.
R2
Now the area of one of the n sectors is given by , while one of the m sectors has
2
2
r
area .
2
Therefore
nR2 m r 2 n r2
D ) D 2:
2 2 m R
From which we conclude that
˛ n r2
D D 2 and ˇr 2 D ˛R2 :
ˇ m R
We can prove that this relation holds for all squarable lunes.
Exercise 139
q q r
p p p 2 p
34 2 33 D 33 2 33 C 1 D 1 C 33 D 1 C 33:
q q q
p p 1 p
38 6 33 D 27 6 33 C 11 D p 81 18 33 C 33
3
r p
1 p 2 9 33
D p 9 33 D p :
3 3
171
Solutions
Exercise 140
r q r p p r !
3 1 p 3 34 2 33 3 1 p
S Dr 2
30 C 2 33
1 30 C 2 33
2 8 2 8 2 64
r q r p r !
3 1 p 3 1 C 33 3 1 p
Dr 2
30 C 2 33 1 30 C 2 33
2 8 2 8 2 64
r q 0r p s p 1
1 3 p 3 1 C 33 1 38 6 33 A
D r2 30 C 2 33 @
8 2 2 8 8 2
r q
21 3 p 1 p p 1 p
Dr 30 C 2 33 p 3 1 C 33 p 9 33 see Exercise 139
8 2 8 2 3
q
1 p p
D r2 30 C 2 33 12 C 4 33
128
q
r2 p p
D 30 C 2 33 3 C 33
32
Exercise 141 See also Exercises 79 and 133
cos 3˛ D cos .˛ C 2˛/ D cos ˛ cos 2˛ cos ˛ cos 2˛ D : : : D 4 cos3 ˛ 3 cos ˛

sin 5˛ D sin .2˛ C 3˛/
D sin 2˛ cos 3˛ C cos 2˛ sin 3˛

D 2 sin ˛ cos ˛ 4 cos3 ˛ 3 cos ˛ C 1 2 sin2 ˛ 3 sin ˛ 4 sin3 ˛
::
:
D 16 sin5 ˛ 20 sin3 ˛ C 5 sin ˛
If m D 5 then
p
sin 5˛ D 5 sin ˛
p
16 sin ˛ 20 sin ˛ C 5 sin ˛ D 5 sin ˛
5 3
p
16t 5 20t 3 C 5 5 t D 0
p
16t 4 20t 2 C 5 5 D 0 _ t D 0
The last equation is a biquadratic equation with solutions:

p p p p
20 ˙ 4 5 C 4 5 5˙ 5C4 5
t D
2
D
32 8
p p p p
5 5C4 5 5C 5C4 5
Only t D
2
is a valid solution because t D
2
> 1.
8 8
So
s p r
p q
5 5C4 5 1 p
sin ˛ D D 10 2 5 C 4 5
8 4
172 Solutions
5
If m D , then
3 r
5 5
sin ˛ D sin ˛
3 3
r
5
sin 5ˇ D sin 3ˇ
3
r
5
16 sin5 ˇ 20 sin3 ˇ C 5 sin ˇ D 3 sin ˇ 4 sin3 ˇ
3
4 p p
16 sin4 ˇ 15 15 sin2 ˇ C 5 15 D 0 _ sin ˇ D 0
3
The biquadratic equation has solutions
4 p 4p p p p p
15 15 ˙ 60 C 6 15 15 15 ˙ 60 C 6 15
sin ˇ D
2 3 3 D
32 24
Now

sin ˛ D sin 3ˇ D sin ˇ 3 4 sin2 ˇ
s p p
p p p p !
15 15 ˙ 60 C 6 15 15 15 ˙ 60 C 6 15
D 34
24 24
s p p
p p p p !
15 15 ˙ 60 C 6 15 15 15 ˙ 60 C 6 15
D 3
24 8
Exercise 142
sin .5 .3 .3˛/// D 16 sin5 .3 .3˛// 20 sin .3 .3˛// C 5 sin .3 .3˛//

5 3
D 16 3 sin 3˛ 4 sin3 3˛ 20 3 sin 3˛ 4 sin3 3˛

C 5 3 sin 3˛ 4 sin3 3˛
3 5
D 16 3 3 sin ˛ 4 sin3 ˛ 4 3 sin ˛ 4 sin3 ˛
3 3
20 3 3 sin ˛ 4 sin3 ˛ 4 3 sin ˛ 4 sin3 ˛
3
C 5 3 3 sin ˛ 4 sin3 ˛ 4 3 sin ˛ 4 sin3 ˛
!
x x 3
x x 3 3 5
D 16 3 3 4 4 3 4
2 2 2 2
!
x x 3
x x 3 3 3
20 3 3 4 4 3 4
2 2 2 2
!
x x 3 x x 3 3
C5 3 3 4 4 3 4
2 2 2 2
::
:
173
Solutions
Remember that .i sin ˛/2k 2 R.
sin 45˛ D Im .cos 45˛ C i sin 45˛/

D Im .cos ˛ C i sin ˛/45
! !
X45
45
D Im k
.i sin ˛/ cos 45k
˛
k
kD0
! !
1 X
22
45 2kC1
D .i sin ˛/ cos 442k
˛
i 2k C 1
kD0
!
X 22
45
D .1/k sin2kC1 ˛ cos442k ˛
2k C 1
kD0
!
X 22
45 22k
D .1/k sin2kC1 ˛ 1 sin2 ˛
2k C 1
kD0
::
:
" j ! !#
X
22 X 45 22 k x 2j C1
D .1/ j
j D0
2k C 1 j k 2
kD0
D 45x 3795x 3 C C x 45
Exercise 143 We know †BAC D ˛, 4ABC is isosceles so †BAC D †ACB D ˛.

†CBD D 2˛ (in a triangle the exterior angle equals the sum of the opposite angles).
Now consider 4BCD. 4BCD is an isosceles triangle so †CBD D †BDC D 2˛.
We can now repeat the procedure with 4BCD and 4CDE and so on.
Exercise 144
Divide the regular polygon into triangles as is done for the heptagon in the figure. It is clear
2
that for these isosceles triangles the vertex angle is rad. The other angles are base angles
n
and are equal to ˇ.
174 Solutions
2
The sum of the angles of a triangle is a straight angle so ˇ C ˇ C D rad, whence
n
2 .n 2/
2ˇ D D rad.
n n
The base angle is half of this result, but the interior angle of the polygon is equal to the sum
of the base angles of two adjacent triangles. Therefore we can conclude that the interior angle
.n 2/
is equal to rad.
n
The angle BAG is an interior angle of a regular heptagon, so

5
†BAG D rad.
7
The angles †BAC; †CAD; †DAE; †EAF; †FAG are angles inscribed in a circle, subtend-
ing a chord of length 1.
Therefore they subtend equal arcs and are therefore equal, from which
5
†BAC C †CAD C †DAE C †EAF C †FAG D rad,
7
5
and 5†BAC D rad, so †BAC D rad.
7 7
0 0 0 0 0 0
†B A C D rad, so †B C A D rad.
7 7
Now in 4B 0 C 0 D 0 :
2
†D 0 B 0 C 0 D †C 0 D 0 B 0 D rad (exterior angle triangle).
7
On the other hand in 4ACF :
†ACF C †CFA C †FAC D rad (sum of the angles of a triangle)
and
†ACF D †CFA (isosceles triangle),
so
3 7 3 4 2
2†ACF D D D rad or †ACF D rad.
7 7 7 7 7
175
Solutions
Exercise 145 Two isosceles triangles with equal base angles are similar so
4ACF 4C 0 B 0 D 0 :
kAC k kCF k
We now know that 0 0
D x, so D x, and, using kCF k D kADk D y, we find
kB C k kB 0 D 0 k
that
y
kB 0 D 0 k D :
x
The proof for 4ADE 4D 0 E 0 C 0 is analogous.
y
Exercise 146 From the construction we know that kADk D kA0 D 0 k so y D 1 C .
1 x
On the other hand kAEk D kA0 E 0 k so y D x C
y
8 y
ˆ
<y D 1 C x )yx D x C y
1
:̂y D x C ) y 2 D xy C 1
y
x
Solving for y in the first equation yields y D , inserting this result into the second
x1
equation yields:
x 2 x x2 x2
Dx C1 ) D C1
x1 x1 .x 1/ 2 x1
) x D x .x 1/ C .x 1/
2 2 2
) x 2 D x 3 x 2 C x 2 2x C 1
) x 3 x 2 2x C 1 D 0 .29/
y2 1
If we want to eliminate x we solve the second equation for x and find x D , and insert
y
this result into the first equation:
2
y 1 y2 1 y3 y y2 1 y2
y D Cy ) D C
y y y y y
) y3 y D y2 1 C y2 ) y 3 2y 2 y C 1 D 0
Possible rational solutions for equation (29) are x D ˙1.

Inserting into the equation yields:
for x D 1W 1 1 2 C 1 D 1 ¤ 0I
for x D 1W 1 1 C 2 C 1 D 1 ¤ 0:
The equation has no rational solution and therefore no constructible solution. Neither ŒAC
nor ŒAF can be constructed using compass and straightedge methods.
Exercise 147 In 4CBA draw the altitude from B. Call the foot H , then 4ABH is right-angled
in H .
Moreover, because 4CBA is isosceles
x
kCH k D kHAk D :
2
176 Solutions

Furthermore †BAH D †BAC D rad so
7
kAH k
cos .†BAH / D
kBAk
therefore
x=2 x
cos D D
7 1 2

Now x is not a constructible number so neither is cos .
7

Consequently (see Exercise 133) an angle of magnitude rad cannot be constructed using
7
compass and straightedge methods.
Exercise 150
Fold the enneagon in the same fashion as we did with the heptagon. Thales’ intercept theorem
then leads to the equations
z 1
kAC k D x; kBDk D ; kCEk D 1; kDF k D ;
x x
kADk D y; kAEk D kAF k D z;
z 1
and 1 C D y, x C 1 D z and y C D z.
x z
Eliminating x and y from these equations yields:
z 3 3z 2 C 1 D 0 (36)
If this equation has rational solutions these are z D ˙1. Obviously neither is a solution to
equation (36), therefore there are no constructible solutions to the equation.
If z is not a constructible number, then neither is x. Using the same procedure as in the

proof of the inconstructibility of the heptagon, we find that angle rad is not constructible
9
and therefore the enneagon is not constructible using compass and straightedge methods.
Fold the hendecagon in the same fashion as we did with the heptagon and the enneagon.
This gives us the following relations:
z w y 1
y D 1C ; zDxC ; wDyC and w D z C
x y z w
177
Solutions
Eliminating x, y and z yields
w 5 3w 4 3w 3 C 4w 2 C w 1 D 0: (37)
This is a fifth degree equation, so we can draw no conclusions about the constructibility of the
roots. Lemma 1 only allows us to draw conclusions about the constructibility of solutions to
cubic equations.
Solving equations (36) and (37) gives us the length of the longest diagonal. Unfortunately
no general algorithm exists to solve a fifth degree equation algebraically, so we have to revert
to numerical methods.
For the enneagon z D 2:879 : : : , for the hendecagon z D 3:513 : : :
178
References
Primary sources1
Franciscus Aguilonius. Opticorum libri sex philosophis iuxtà ac mathematicis utiles. Jan I Moretus,
Antwerpen, 1613. EHC G 5050.
Aristotle (Trans. G. R. G. Mure). Posterior analytics. URL http://classics.mit.edu//Aristotle/posterior.1.i.html.
Retrieved 25-4-2015.
Oliver Byrne. The First Six Books of the Elements of Euclid. William Pickering, London, 1847.
http://commons.wikimedia.org/wiki/Category:Byrne
http://publicdomainreview.org/collections/the-first-six-books-of-the-elements-of-euclid-1847/
http://babel.hathitrust.org/cgi/pt?id=gri.ark:/13960/t9766b543;view=1up;seq=7.
Michiel Coignet. Est de Michaelis Coigneti 1576, 1576–77. part of the manuscript written 1603 and later;
Bibliothèque Nationale Paris, Ms Néer 56.
Albrecht Dürer. Underweysung der Messung. s.n., Nürenberg, 1525. EHC H 202415.
Leonhard Euler. Correspondence, letter 765 & 766. URL http://eulerarchive.maa.org/. Retrieved 24-8-2014.
Jacobus Falco. Jacobus Falco Valentinus, miles ordinis Montesiani, hanc circuli quadraturam invenit. Petrus
Bellerus I, Antwerpen, 1591. MPM 8 499.
Johannes Kepler. Ausszug auss der Uralten MesseKunst Archimedis. Johann Planck, Linz, 1616. EHC G
86948.
Samuel Marolois. Opera mathematica, ou: Œuvres mathematiques traictans de geometrie, perspective,
architecture, et fortification. H. Hondius, Hagae-Comitis (Den Haag), 1614–1615. EHC G 48942.
Valentin Mennher. Practicque pour brievement apprendre a ciffrer, & tenir livre de comptes. Gillis I Coppens
van Diest, Antwerpen, 1565. MPM A 3589.
John Playfair. Elements of Geometry. Bell & Bradfute and G.G. & J. Robinson, Edinburgh-London, 1795.
URL http://lib.ugent.be/catalog/rug01:001104161. University of Ghent.
John Playfair. Elements of Geometry (4th ed.). Bell & Bradfute, Edinburgh, 1814. https://books.google.be/.
John Playfair. Elements of Geometry (9th ed.). Bell & Bradfute, Edinburgh, 1836. https://books.google.be/.
Gregorius à Sancto Vincentio. Opus geometricum quadraturae circuli et sectionum coni, decem libris
comprehensum. Joannes et Jacobus Meursius, Antwerpen, 1647. EHC G 4869.
Ludolph van Ceulen. De circulo et adscriptis liber. Colster, Lugdunum Batavorum (Leiden), 1619. EHC G
4867.
Adriaan van Roomen. Ideæ mathematicae pars prima, sive Methodus polygonorum. Jan I van Keerberghen,
Antwerpen, 1593. MPM 8 533.
Frans van Schooten. Mathematische oeffeningen begrepen in vijf boecken: waer by gevougt is een tractaet
handelende van reeckening in speelen van geluck, door Christianus Hugenius. Gerrit van Goedesbergh,
Amsterdam, 1659. KBR VH 8.040 A.
Frans van Schooten. Eerste [- vijfde] bouck der mathematische oeffeningen. Gerrit van Goedesbergh,
Amsterdam, 1659–1660. KBR VB 4.820a B.
François Viète. Opera mathematica. Bonaventura et Abraham Elzevir, Lugdunum Batavorum (Leiden), 1646.
EHC G 4858.
1
Abbreviations used to refer to libraries.
EHC – Erfgoedbibliotheek Hendrik Conscience, Antwerp

KBR – Koninklijke Bibliotheek Albert I, Brussels
MPM – Museum Plantin-Moretus/Prentenkabinet, Antwerp – Unesco-World Heritage
179
References
Bibliography
J.M. Aarts. Meetkunde. Facetten van planimetrie en stereometrie. Epsilon Uitgaven, Utrecht, 2000.
Roger C. Alperin. Trisections and Totally Real Origami, 2004. URL arXiv:math/0408159v1 [math.HO] 11
Aug 2004.
Claudi Alsina and Roger B. Nelsen. Charming Proofs; A Journey into Elegant Mathematics, volume 42 of
Dolciani Mathematical Expositions. MAA, 2010.
Kirsti Andersen. The Geometry of an Art: The History of the Mathematical Theory of Perspective from Alberti
to Monge. Sources and Studies in the History of Mathematics and Physical Sciences. Springer Science &
Business Media LLC, New York, 2007.
W.S. Anglin. Mathematics: a Concise History and Philosophy. Springer-Verlag, New York – Berlin, 1994.
Eric Bainville and Bernard Genevés. Constructions using Conics. The Mathematical Intelligencer, 22(3):
60–72, 2000.
Adnan Baki, Temel Kosa, and Bulent Guven. A comparative study of the effects of using dynamic geometry
software and physical manipulatives on the spatial visualisation skills of pre-service mathematics teachers.
British Journal of Educational Technology, 42(2):291–310, 2011.
Évelyne Barbin and Anne Boyé. François Viète, Un mathématicien sous la Renaissance. Vuibert, Paris, 2005.
Klaus Barner. How Old did Fermat become? NTM International Journal of History & Ethics of Natural
Sciences, Technology & Medicine, 9:209–228, 2001.
J. Lennart Berggren. The Mathematics of Egypt, Mesopotamia, China, India, and Islam: A Sourcebook,
chapter Mathematics in Medieval Islam, pages 515–676. Princeton University Press, Princeton –
Woodstock, 2007.
Frits Beukers. Pi, volume 6 of Zebra. Epsilon Uitgaven, Utrecht, 2000.
Paul Bockstaele. The correspondence of Adriaan van Roomen. Lias, 3:85–129 and 249–299, 1976.
Paul Bockstaele. A Challenge to the Mathematicians of the University of Leuven as a New Year’s Gift for
1639. In M. Folkerts & J.P. Hogendijk, editor, Vestigia Mathematica, pages 15–28. Rodopi, 1993.
Paul Bockstaele. Between Viète and Descartes: Adriaan van Roomen and the Mathesis Universalis. Archive
for the History of Exact Sciences, 63(4):433–470, 2009.
Alexander Bogomolny. Francois Viete’s Reduction of CCC to PCC from Interactive Mathematics Miscellany
and Puzzles, a. URL http://www.cut-the-knot.org/Curriculum/Geometry/GeoGebra/CCC-Viete.shtml.
Retrieved 8-8-2014.
Alexander Bogomolny. Delian problem from Interactive Mathematics Miscellany and Puzzles, b. URL http://
www.cut-the-knot.org/do%20you%20know/Delian.shtml. Retrieved 17-11-2013.
Benjamin Bold. Famous Problems of Geometry and how to solve them. Dover Publications, New York, 1982.
Karl Bopp. Die Kegelschnitte des Gregorius a St Vincentio in vergleichender Bearbeitung, volume XX of
Abhandlungen zur Geschichte der mathematischen Wissenschaften. Teubner Verlag, Leipzig, 1907.
Henk J.M. Bos. Recognition and Wonder: Huygens, tractional motion and some thoughts on the history of
Mathematics. Tractrix, 1:3–20, 1989.
Henk J.M. Bos. Johann Molther’s Problema Deliacum, 1619. In M. Folkerts & J.P. Hogendijk, editor, Vestigia
Mathematica, pages 29–46. Rodopi, 1993.
Henk J.M. Bos. Redefining Geometric Exactness. Springer-Verlag, New York, 2001.
Lucas N.H. Bunt, Philip S. Jones, and Jack D. Bedient. The Historical Roots of Elementary Mathematics.
Dover Publications, New York, 1988.
Karine Chemla, editor. History of Science, History of Text, volume 238 of Boston Studies in the Philosophy of
Science. Springer, Dordrecht, 2004.
Constantinos Christou, Nikos Mousoulides, Marios Pittalis, and Demetra Pitta-Pantazi. Proofs through
Exploration in Dynamic Geometry Environments. International Journal of Science and Mathematics
Education, 2(3):339–352, 2004.
Julian L. Coolidge. A History of the Conic Sections and Quadric Surfaces. Clarendon Press, Oxford, 1945.
Gregg De Young. Translating Playfair’s Geometry into Arabic: Mathematics and Missions. In Nathan Sidoli
and Glenn Van Brummelen, editors, From Alexandria Through Baghdad: Surveys and Studies in the
Ancient Greek and Medieval Mathematical Sciences in Hounour of J.L. Berggren, pages 503–527. Springer
Verlag, New York – Heidelberg, 2014.
180 References
Jean-Pierre Delahaye. Le fascinant nombre . Belin-Pour la Science, Paris, 1997.

Jean-Pierre Delahaye. Viète et les codes secrets. In Barbin and Boyé (2005).
David Dennis. Historical Perspectives for the Reform of Mathematics Curriculum: Geometric Curve Drawing
Devices. Master’s thesis, Cornell University, 1995.
Heinrich Dörrie. 100 Great Problems of Elementary Mathematics. Dover Publications, New York, 1965.
Howard Eves. From Five Fingers to Infinity, chapter Omar Khayyam’s Solution of Cubic Equations, pages
302–303. In Swetz (1994), 1994.
Howard Eves. College Geometry. Jones and Bartlett Publishers, Boston – London, 1995.
Julie Fink and Nicholas Molbert. Sectioning Angles using Hyperbolic Curves. Rose-Hulman Undergraduate
Mathematics Journal, 14:183–194, 2013.
David R. Finston and Patrick J. Morandi. Abstract Algebra; Structure and Application. Undergraduate Texts
in Mathematics and Technology. Springer International Publishing Switzerland, Cham – Heidelberg – New
York - Dordrecht – London, 2014.
Kuno Fladt. Geschichte und Theorie der Kegelschnitte und der Flächen zweiten Grades. Ernst Klett Verlag,
Stuttgart, 1967.
Menso Folkerts. Die Entwicklung und Bedeutung der Visierkunst als Beispiel der praktischen Mathematik der
frühen Neuzeit. Humanismus und Technik, 18/1:1–41, 1974.
Michael. N. Fried and Sabetai Unguru. Apollonius of Perga’s Conica : text, context, subtext. Brill, Leiden,
2012.
Aleksandr O. Gelfond. Sur le septième Problème de Hilbert. Bulletin de l’Académie des Sciences de l’URSS.
Classe des sciences mathématiques, VII:623–634, 1934.
Stephen Gersh. Rewriting the Proslogion; Nicholas of Cusa’s Transformation of Anselm of Canterbury’s
Proof of the Existence of God. Epekeina, 1:65–89, 2012.
Judith V. Grabiner. The Origins of Cauchy’s Rigorous Calculus. MIT Press, Cambridge, Mass., 1981.
Judith V. Grabiner. A Mathematician among the Molasses Barrels: MacLaurin’s unpublished Memoir on
Volumes. Proceedings of the Edinburgh Mathematical Society, 39:193–240, 1996.
Judith V. Grabiner. ’Some Disputes of Consequence’: McLaurin among the Molasses Barrels. Social Studies
of Science, 28:139–168, 1998.
René Grothmann. The Geometry Program C.a.R. International Journal of Computer Discovered Mathematics,
1(1):45–61, 2016.
Iris Gulikers and Klaske Blom. ’A Historical Angle’, a Survey of Recent Literature on the Use and Value of
History in Geometrical Education. Educational Studies in Mathematics, 47:223–258, 2001.
Thomas C. Hales. A Proof of the Kepler Conjecture. Annals of Mathematics, 162(3):1065–1185, 2005.
Vagn L. Hansen. Shadows of the Circle. World Scientific, Singapore – River Edge, NJ – London, 1998.
Robin Hartshorne. Geometry: Euclid and Beyond. Springer-Verlag, New York, 2000.
Thomas L. Heath. The works of Archimedes. Dover Publications, New York, 1953.
Thomas L. Heath. Euclid; The Thirteen Books of the Elements (3 vols). Dover Publications, New York, 1956.
Thomas L. Heath. Treatise on Conic Sections: with Introductions, including an Essay on the Earlier History of
the Subject. Heffer, Cambridge, 1961.
Thomas L. Heath. A History of Greek Mathematics (2 vols). Dover Publications, New York, 1981.
Philippe P. A. Henry. La solution de François Viète du problème d’Adriaan van Roomen. URL http://egg.epfl.
ch/~phenry/VieteRoomen.pdf. Retrieved 18-8-2014.
Jan P. Hogendijk. On the Trisection of an Angle and the Construction of a Regular Nonagon by means of
Conic Sections in Medieval Islamic Geometry. Preprint 113, University of Utrecht, Department of
Mathematics, 1979.
Jan P. Hogendijk. How Trisections of the Angle were transmitted from Greek to Islamic Geometry. Historia
Mathematica, 8:417–438, 1981.
Jan P. Hogendijk. Abu’l-Jud’s answer to a question of al-Biruni concerning the regular heptagon. In D.A. King
and G. Saliba, editors, From Deferent to Equant: a volume of studies in the ancient and medieval Near East
in honor of E.S. Kennedy, volume 500 of Annals of the New York Academy of Sciences, pages 175–184.
New York Academy of Sciences, New York, 1987.
Markus Hohenwarter, Judith Hohenwarter, Yves Kreis, and Zsolt Lavicza. Teaching and Learning Calculus
with Free Dynamic Mathematics Software GeoGebra, 2008. URL http://tsg.icme11.org/document/get/666.
181
References
Mary Jaeger. Archimedes and the Roman Imagination. The University of Michigan Press, Ann Arbor, 2008.
Victor J. Katz. A History of Mathematics: an Introduction. Addison-Wesley, Reading, Mass., 2nd edition,
1998.
Victor J. Katz, editor. The Mathematics of Egypt, Mesopotamia, China, India, and Islam: A Sourcebook.
Princeton University Press, Princeton-Woodstock, 2007.
James King and Doris Schattschneider, editors. Geometry Turned On: Dynamic Software in Learning,
Teaching, and Research, number 41 in Notes, Washington, DC, 1997. The Mathematical Association of
America.
Dick Klingens. Kwadreerbare maantjes, 01 2008. URL http://www.pandd.nl/downloads/kwadreerbaar.pdf.
Wilbur R. Knorr. The Ancient Tradition of Geometric Problems. Dover Publications, New York, 1986.
Wilbur R. Knorr. The Ancient Tradition of Geometric Problems. Dover Publications, New York, 1993.
Helmut Koch. Der Briefwechsel von Leonhard Euler und Christian Goldbach. Birkhäuser Verlag, Basel, 2007.
Ulrich Kortenkamp, Axel M. Blessing, Christian Dohrmann, Yves Kreis, Paul Libbrecht, and Christian
Mercat. Interoperable interactive geometry for europe – first technological end educational results and
future cchallenge of the intergeo project. Proceedings of CERME 6, January 28th-February 1st 2009, 2010.
URL http://ife.ens-lyon.fr/editions/editions-electroniques/cerme6/.
Steven G. Krantz. The Proof is the Pudding; The Changing Nature of Mathematical Proof. Springer Science
& Busines Media LLC, New York, NY, 2011.
Colette Laborde. Dynamic Geometry Environments as a Source of Rich Learning Contexts for the Complex
Activity of Proving. Educational Studies in Mathematics, 44(1):151–161, 2000.
Peter Lanser. De laatste stelling van Fermat, volume 7 of Zebra. Epsilon, Utrecht, 2000.
J. Dennis Lawrence. A Catalog of Special Plane Curves. Dover books on advanced mathematics. Dover
Publications, New York, NY, 1972.
Falko Lorenz. Algebra, Volume I: Fields and Galois Theory. Universitext. Springer, New York, NY, 2006.
Arman Margaryan and Nerses Aramian. Compass constructions, 10 2014. URL http://arxiv.org/abs/1410.
3042.
Ramón Marrades and Ángel Gutiérrez. Proofs produced by Secondary School Students learning Geometry in
a Dynamic Computer Environment. Educational Studies in Mathematics, 44(1):87–125, 2000.
George E. Martin. Geometric Constructions. Springer-Verlag, New York, 1998.
J. P. McCarthy. The Cissoid of Diocles. The Mathematical Gazette, 25(263):12–15, February 1941.
Ad Meskens. Wine gauging at late Sixteenth and early Seventeenth Century Antwerp. Historia Mathematica,
21:121–147, 1994.
Ad Meskens. The Jesuit Mathematics School in Antwerp in the early Seventeenth Century. The Seventeenth
Century, 12:11–22, 1997.
Ad Meskens. Joannes della Faille s.j.: Mathematics, Modesty and Missed Opportunities, volume 53 of
Belgisch historisch instituut te Rome. Bibliotheek. Belgisch Historisch Instituut te Rome, Brussel-Rome,
2005.
Ad Meskens. Travelling Mathematics: the fate of Diophantos’ Arithmetic, volume 41 of Science Networks.
Historical Studies. Birkhäuser Verlag, Basel, 2010.
Ad Meskens. Practical Mathematics in a Commercial Metropolis: Mathematical Life in late 16th century
Antwerp, volume 31 of Archimedes. Springer Science & Business Media B.V., Dordrecht, 2013.
Ad Meskens and Paul Tytgat. Met passer, liniaal en neusislat. Number 41 in Zebra. Epsilon Uitgaven,
Amsterdam, 2015.
Erich Meuthen. Nikolaus von Kues, 1401 – 1464: Skizze einer Biographie. Aschendorff, Münster, 1982.
Dermot Moran. Platonism at the Origins of Modernity: Studies on Platonism and Early Modern Philosophy,
chapter Nicholas of Cusa (1401–1464): Platonism at the Dawn of Modernity, pages 9–29. Springer
Netherlands, Dordrecht, 2008.
Pier Daniele Napolitani. Between Myth and Mathematics: the Vicissitudes of Archimedes and his Work.
Lettera Matematica, 1:105–112, 2013.
R.W.D. Nickalls. Viète, Descartes and the Cubic Equation. The Mathematical Gazette, 90:203–208, 2006.
John J. O’Connor and Edmund F. Robertson. The MacTutor History of Mathematics archive. URL http://
www-history.mcs.st-and.ac.uk/. Retrieved 25-8-2014.
182 References
Alexander Ostermann and Gerhard Wanner. Geometry by Its History. Undergraduate Texts in Mathematics.
Springer Verlag, Dordrecht, 2012.
Stephanos A. Paipetis and Marco Ceccarelli, editors. The Genius of Archimedes – 23 Centuries of Influence on
Mathematics, Science and Engineering, volume 11 of History of mechanism and Machine Science,
Dordrecht – Heidelberg – London – New York, NY, 2010. Springer Science & Busines Media B.V.
Elena Pastuchová, Jana Galanová, and Alfonz Smola. Application of atypical curves for the formation surfaces
and reflectors. Journal of Electrical Engineering, 53:336–338, 2002.
George M. Phillips. Mathematics is not a Spectator Sport. Springer Science & BusinesSpringer Inc., New
York, NY, 2005.
M.M. Postnikov and Abe Shenitzer (transl.). The Problem of Squarable Lunes. The American Mathematical
Monthly, 107:645–651, 2000.
Jürgen Richter-Gebert and Ulrich H. Kortenkamp. The Cinderella.2 Manual. Springer Verlag, Berlin –
Heidelberg, 2012.
Franz Rothe. Several Topics in Geometry, 01 2014. URL http://math2.uncc.edu/~frothe/. Retrieved 15-2-2015.
Ken Saito. Doubling the Cube: A New Interpretation of Its Significance for Early Greek Geometry. Historia
Mathematica, 22:119–137, 1995.
C. Schoy. Drei planimetrische Aufgaben des arabischen Mathematikers Abu’l=Jud Muhammad ibn al=Lith.
ISIS, 7:5–8, 1925.
Heike Sefrin-Weis. Pappus of Alexandria. Book 4 of the Collection. Springer Verlag, London, 2010.
Brian J. Shelburne. The Five Quadrable (Squarable) Lunes, 04 2008. URL http://www4.wittenberg.edu/
academics/mathcomp/bjsdir/TheFiveLunes120408.pdf. Retrieved 26-8-2014.
Eugene V. Shikin. Handbook and Atlas of Curves. CRC Press, Boca Raton – New York, NY – London –
Tokyo, 1995.
Simon Singh. Fermat’s Enigma: the Quest to solve the World’s Greatest Mathematical Problem. Walker, New
York, 1997.
Matthijs H. Sitters. Sybrandt Hansz Cardinael; rekenmeester en wiskundige. Verloren, Hilversum, 2008.
Alfons J.E.M. Smeur. De verhandeling over de cirkelkwadratuur van Franco van Luik van omstreeks 1050.
Koninklijke Vlaamse Academie, Brussel, 1968.
Alfons J.E.M. Smeur and Menso Folkerts. A Treatise on the Squaring of the Circle by Franco of Liège, of
about 1050. I. Archives Internationales d’Histoire des Sciences, 26:59–105, 1976a.
Alfons J.E.M. Smeur and Menso Folkerts. A Treatise on the Squaring of the Circle by Franco of Liège, of
about 1050. II. Archives Internationales d’Histoire des Sciences, 26:225–253, 1976b.
John Stillwell. The Four Pillars of Geometry. Springer Science & BusinesSpringer Inc., New York, NY, 2005.
John Stillwell. Mathematics and Its History. Undergraduate Texts in Mathematics. Springer Science &
Busines Media LLC, New York, NY, 2010.
Dirk J. Struik. From Five Fingers to Infinity, chapter Omar Khayyam, Mathematician, pages 297–301. In
Swetz (1994), 1994.
Frank J. Swetz, editor. From Five Fingers to Infinity. Open Court, Peru, Ill., 1994.
Nikolaj Tschebotaröw. Über quadrierbare Kreisbogenzweiecke. I. Mathematische Zeitschrift, 39(1):161–175,
1935. ISSN 0025-5874.
C.J. Vamvacas. The Founders of Western Thought – The Presocratics, Boston Studies in the Philosophy of
Science, volume 257 of Boston Studies in the Philosophy of Science, chapter Thales of Miletus (ca. 625 –
546 B.C.), pages 2 – 33. Springer Science+Business Media B.V., Dordrecht, 2009.
Glen Van Brummelen. Heavenly Mathematics: The Forgotten Art of Spherical Trigonometry. Princeton
University Press, Princeton, 2012.
Glen Van Brummelen and Michael Kinyon, editors. Mathematics and the Historian’s Craft; The Kenneth O.
May Lectures. CMS Books in Mathematics. Canadian Mathematical Society – Societé mathématique du
Canada, New York, NY, 2005.
Delphi van de Weyer. Opera Mathematica: La Géometrie, S. Marolois. Master’s thesis, Vrije Universiteit
Brussel, Brussel, 2011.
Iris van Gulik-Gulikers. Geschiedenis van de niet-Euclidische meetkunde, volume 21 of Zebra. Epsilon,
Utrecht, 2005.
183
References
Floor van Lamoen. Passen en meten met cirkels: de arbelos van Archimedes, volume 30 of Zebra. Epsilon
Uitgaven, Utrecht, 2009.
Herman van Looy. Chronologie en analyse van de mathematische handschriften van G. a Sancto Vincentio.
PhD thesis, K.U. Leuven, 1979.
Jan A. van Maanen. Aangeslibd land, kegelsnedetrekkers en oneindige bierglazen. In Thomas Coppens,
editor, Voordrachten en werkgroepen van jhet vijfde congres van de Vlaamse Vereniging Wiskunde Leraars
Neerpelt juli 1987, volume 8 of Monografieën van de VVWL, pages 203–220, Wilrijk, 1987. VVWL.
Paul Ver Eecke. Les œuvres complètes d’Archimède. Desclée de Brouwer, Paris, 1921.
B.A. Vermaseren. Atlas der algemene en vaderlandse geschiedenis. Wolters-Noordhoff, Groningen, 1977.
Cettina Voza. The Death of Archimedes: a Reassessment. In Paipetis and Ceccarelli (2010), pages 507–514.
Steven Wepster. In de ban van de cirkel. Euclides, 85/3:98–100, 2009.
Steven Wepster. Ludolph van Ceulen (1540–1610), Meester der rekenmeesters. Pythagoras, 49(3):12–15,
2010a.
Steven Wepster. Hoe van Ceulen insloot. Pythagoras, 49(3):26–28, 2010b.
Steven Wepster and Marjanne de Nijs. Meester Ludolphs koordenvierhoek, volume 31 of Zebra. Epsilon
Uitgaven, Utrecht, 2010.
William F. Wertz Jr. Nicolaus of Cusa’s On the Quadrature of the Circle, 2001. URL http://www.
schillerinstitute.org/fid_97-01/012_Cusa_quad_circ.html. Retrieved 26-8-2014.
Jiyuan Yu. The Structure of Being in Aristotle’s Metaphysics, volume 52 of The new synthese historical
library: texts and studies in the history of philosophy. Kluwer, Dordrecht, 2003.
184 Index
Index
A – ellipse 12, 19, 22, 104, 132

– hyperbola 12, 16–18, 46, 52,
F
Abū’l-Jūd 122 59, 100, 137, 138, 142 Fermat, Pierre de 2
agrimensores 88 – limaçon 130, 134 – Fermat’s Last Theorem 1
Al-Bı̄rūnı̄ 122 – McLaurin’s trisectrix 73 Fontenay-le-Comte 36
Alexandria 11, 59 – Nicomedes’ conchoid 52, Franco of Liège 88–90, 98
Anthonisz, Adriaan 95 55–57, 63 Fulbert of Chartres 88
Antwerp 65, 66, 95, 99 – parabola 12–14, 44, 63, 75,
Apollonius of Perga 14
– Apollonius’ problem 14,
99–103, 131, 136, 137, 164,
165
G
16–19, 36, 38 – Pascal’s limaçon 9, 61, 62, Goldbach, Christian 1
– Conics 14 134 – Goldbach’s conjecture 1
– Tangencies 14 – tractrix 22, 23 Grienberger s.j., Christopher 99
Archimedes 59–61, 79, 93–95, Cusanus, Nicolas 90–92, 159,
99, 123 160
– arbelos 79 H
– Archimedes’ spiral 63, 64,
75, 144 D Hales, Thomas 2
Henry IV, King of France 15,
– Archimedes’ trisection 63,
148 de Aguilon s.j., 36
Aristotle 12, 95 Franciscus 65–69 Hermite, Charles 119
Asia Minor 11 – Opticorum Libri sex 66, 67 Heron of Alexandria 122
Athens 11, 12, 43 Delft 95 Hildesheim 95
della Faille s.j., Joannes 104 Hippias of Elis 71, 76
Hippocrates of Chios 43, 44,
B Delos, island 43
– oracle 43 78, 81, 83, 85, 86, 114, 115,
Dinostratus 44, 76, 77 156, 170
Bachet de Méziriac, Claude
– quadratrix 71, 72, 75–78, – Elements 43
Gaspard 2
145 – Hippocrates’ lunes 78, 80, 84
Bernoulli, Nicholas I 1
Diocles Huygens, Christian 22, 100
Bolyai, János 29
Bruges 65 – cissoid 53, 142
Brussels 65, 66 Diophantus 2 I
Dürer
C – conchoid 7 Ibn Qurra, Thabit 14

Ibn-al-Haytham 67
Ceva, Giovanni 74
Clavius s.j., Christopher 65
E J
curve equation
– Archimedes’ spiral 63–65, – quadratic 37 jesuit church
75, 144 Euclid 27–30, 43, 59, 105, 144, – Antwerp, Carolus
– botanic curve 74 vii Borromeus 65
– Ceva’s trisectrix 74 – Mons 65
– Elements 4, 27–29, 43
– cissoid of Diocles 53, 142 – Tournai 65
– parallel postulate 28–30
– conchoid 7, 25, 56–58, 63, jesuit college
Eudoxus 44
68, 143, 144 – ’s Hertogenbosch 65
Euler, Leonhard 1
– Dinostratus’ quadratrix 71, – Antwerp 65
Eutocius 14
72, 75–78, 145 – Brussels 65
– Dürer’s conchoid 7 – Douai 65
185
Index
– Kortrijk 65 neusis 23–26, 49, 50, 52, 55, Sicily 11, 59, 65
– Paris 65 56, 61–64, 67, 68, 85, 86, 125, slider, definition 5
134, 144, 157 Socrates 12
K Nicomedes 55, 59, 112
– conchoid 52, 55–57, 63
Sparta 11, 71
Syracuse 11, 59, 60
Kepler, Johannes 2, 98, 99 non-Euclidean geometry 29
– Kepler conjecture 2
Kortrijk 65 P T
Thales of Milete 11
Paris 36
L Pascal, Blaise 9
– circle theorem 11, 134, 137
– intercept theorem 11, 33, 59,
Pascal, Etienne 9 84, 105, 106, 176
Leibniz, Gottfried 1, 22
Perrault, Claude 22 The Simpsons (TV-series) 2
Leiden 95
Philip II, King of Spain 36, 65
– Engineering school 21
Philip III, King of Spain 36
– University 95
Leonardo’s Claw 81, 82, 152,
Pirate’s dilemma 5 V
Plato 12, 44–46, 71
153 Playfair, John 29 van Ceulen, Ludolff 14, 95–97
Leuven 14, 65 Ptolemy, King of Egypt 27, vii – Vanden Circkel 95
– University 14, 90 Pythagoras van Roomen, Adriaan 14–16,
Lindemann, Ferdinand 119 – Pythagoras’ theorem 22, 27, 95–97, 111, 112, 118
Lobachevsky, Nikolai 29 28, 36, 65–67, 78, 79, 144, – IdeæMathematica 15, 95
locus of B with A; definition 6 150–152, 158, 160 – In Archimedis 95
– Pythagorean triplet 1, 2, 128 van Schooten, Frans 20–22
M – Exercitationum
Mainz 14
R Mathematicorum 20
vander Eycke, Simon 95
Marolois, Samuel 35, 36 Radolph of Liège 88 Viète, François 14–16, 36–38,
– Opera Mathematica 36 95, 111, 112, 121, 124
– Ad Problema 16
McLaurin, Colin 73
– trisectrix 73
S – Opera Mathematica 38
Menaechmus 44, vii – Viète’s Ladder 121, 122
Saccheri, Girolamo 29
Menelaus Salvasutras 93
– Menelaus’ theorem 49, 51
Milete 11
Sancto Vincentio s.j., Gregorius W
a 45, 65–67, 69, 70,
99–101, 103, 104, 144 Wantzel, Pierre 108
N – Problema Austriacum 99,
100
Wiles, Andrew 2
Wine gauger 46, 47
Napoleon’s theorem 5 Scaliger, Joseph Justus 95 Würzburg 14

(Compact Textbooks in Mathematics) Ad Meskens, Paul Tytgat (Auth.) - Exploring Classical Greek Construction Problems With Interactive Geometry Software-Birkhäuser Basel (2017) PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(Compact Textbooks in Mathematics) Ad Meskens, Paul Tytgat (Auth.) - Exploring Classical Greek Construction Problems With Interactive Geometry Software-Birkhäuser Basel (2017) PDF

Uploaded by

Copyright:

Available Formats

Compact Textbooks in Mathematics

For further volumes:

This textbook series presents concise introductions to current topics in math-

Library of Congress Control Number: 2017931078

Mathematics Subject Classification (2010): 51-01, 51-03, 01A05, 1A20, 97G40

Cover design: deblik, Berlin

Printed on acid-free paper

This book is published under the trade name Birkhäuser

ment of Artesis Plantijn University College Antwerp, Erfgoedbibliotheek Hendrik

The equivalent American style notation is given between brackets.

EHC – Erfgoedbibliotheek Hendrik Conscience, Antwerp

The text was typeset in LYX an editor for LATEX.

2 The Genesis of Geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

3 Compass and straightedge constructions . . . . . . . . . . . . . . . . . . . . . . . . . . 27

4 The Delian Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43

6 Squaring the circle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75

6.7 Adriaan van Roomen and Ludolff van Ceulen . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95

7 Constructible numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

8 The Cinderella of regular polygons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121

Fig. 1.1 The cubic close packing (drawings by Paul Tytgat) . . . . . . . . . . . . . . 3

Fig. 6.26 Frontispiece of Gregoy of Saint Vincent’s Problema Austriacum

Ad Meskens and Paul Tytgat

© Springer International Publishing Switzerland 2017

1.2 Interactive Geometry Software (IGS)

. Fig. 1.4 Dürer’s conchoid (right: Dürer (1525), EHC H 202415)

y 3 C .x a/xy C d.x d /2 D 0 (1)

1.3 Polar coordinates

in counterclockwise direction. We call .r;  / the polar coordinates of point P . By definition,

A polar equation is an equation in r and  . A solution to a polar equation is an ordered

The Genesis of Geometry

Ad Meskens and Paul Tytgat

© Springer International Publishing Switzerland 2017

2.1 Greeks and their Geometry

. Fig. 2.1 Greek colonisation about 500 BC (after Vermaseren (1977), p. 9)

Of course these constructions need to be accomplished in a finite number of steps.

2.2 Examples from Plato’s hierarchy

» Draw the perpendicular on d and through F

. Fig. 2.3 The construction of a parabola with focus F and directrix d

Draw arcs which intersect di

2.3 Adrian van Roomen and Apollonius’ Problem

In 1593 he published Ideæ Mathematica in which he challenged his contemporaries to solve

45x 3795x 3 C 95 634x 5 1 138 500x 7 C 7 811 375x 9 34 512 075x 11

. Fig. 2.6 Circle C touches circle

. Fig. 2.11 MAB is the midpoint

2.4 Mechanical construction aids

. Fig. 2.12 Drawing an ellipse us-

. Fig. 2.16 The “construction”

The problem reduces to solving the differential equation

. Fig. 2.18 Performing a neusis

. Fig. 2.20 An instrument to draw conchoids with (drawing by Paul Tytgat)

Compass and straightedge constructions

Ad Meskens and Paul Tytgat

© Springer International Publishing Switzerland 2017

3.1 Euclid’s Elements

. Fig. 3.3 Euclid I.1, the construc-

3.2 Compass and straightedge constructions

» Construct an equilateral triangle on a given line segment

In Euclid’s version the solution reads5 :

» Let AB be the given finite straight line.

The construction can be carried out like this:

» Let the line segment be ŒAB

. Fig. 3.5 Proof of the construc-

» Select a point B on a given straight line a

in counterclockwise direction. We call .r; / the polar coordinates of point P . By definition,

A polar equation is an equation in r and . A solution to a polar equation is an ordered