Geometry

CONTEMPORARY UNDERGRADUATE MATHEMATICS SERIES
Roberl J. Wisner. Editor !

,CE
MATHEMATICS FOR THE LIBERAL ARTS STUDENT, SECOND EDITION. Fred'
Richman, Carol Walker. and Robert J. Wisner
INTERMEDIATE ALGEBRA. Edward D. Gaughan
MODERN
ALGEBRA: A PRECALCULUS COURSE. James E. Hall
TRIGONOMETRY: CIRCULAR FUNCTIONS AND THEIR APPLICATIONS.
James E. Hall
GEOMETRIES
MODERN MATHEMATICS: AN ELEMENTARY APPROACH. THIRD
EDITION. Ruric E. Wheeler
A PROGRAMMED STUDY OF NUMBER SYSTEMS, Ruric E. Wheeler and
Ed A. Wheeler
FUNDAMENTAL COLLEGE MATHEMATICS: NUMBER SYSTEMS AND
INTUITIVE GEOMETRY, Ruric E, Wheeler
MODERN MATHEMATICS FOR BUSINESS STUDENTS, Ruric E. Wheeler
and W. O. Peepies
ANALYTIC GEOMETRY. James E. Hall
INTRODUCTORY GEOMi:TRY: AN INFORMAL APPROACH. SECOND
EDITION, Jame~ R. Smart
MODERN GEOMETRIES; James R. Smart
AN INTUITIVE APPROACH TO ELEMENTARY GEOMETRY. Beauregard JAMES R. SMART
Stubblefield
California State University, San Jose
GEOMETRY FOR TEACHERS. Paul B. Johnson and Carol H. Kipps
LINEAR ALGEBRA. James E Scroggs 5,
ESSENTIALS OF ABSTRACT ALGEBRA, Charles M. Bundrick and John J.
Leeson
AN INTRODUCTION TO ABSTRACT ALGEBRA. A. Richard Mitchell and 6,
Roger W. MitChel!
INTRODUCTION TO ANALYSIS. Edward O. Gaughan
.\11
DIFFERENTIAL EOUATIONS AND RELATED TOPICS FOR SCIENCE AND
ENGINEERING. Robert W. Hunt
n
A PRIMER OF COMPLEX VARIABLES WITH AN INTRODUCTION TO ·0
ADVANCED TECHNIQUES, Hugh J. Hamilton of
It
CALCULUS OF SEVERAL VARIABLES. E, K. McLachlan ,f ty,
PROBABILITY. Donald R. Barr and Peter W. Zehna s of
THEORY AND EXAMPLES OF POINT·SET TOPOLOGY. John Greever of
ale
AN INTRODUCTION TO ALGEBRAIC TOPOLOGY. John W. Keesee
BROOKS/COLE PUBLISHING COMPANY any
EXPLORATIONS IN NUMBER THEORY. Jeanne Agnew Monterey, California )Ve-
NUMBER THEORY: AN INTRODUCTION TO ALGEBRA. Fred Richman
A Division of Wadsworth Publishing Company: Inc.
MODERN GEO METRIES
,
This book was edited by Phyllis London and PREFACE
designed by Linda Marcetti. Tlte technical art
was drawn by Jolm Foster. Tlte book was printed
alfd bOlf.!,d by Kingsport Press, Kingsport.
'.
Tennessf!.l?:
QA47J "S!J
~~dt:lD ieomdnn {II,.) Jamn R. Smart.
1IIIIIImlllll~I~llmlllllllllllllllllllllllllllllllllll
00200612:67
128366
6,
til
© 1973 by Wadsworth Publishing Company,
Inc., Belmont, CaUfomia 94002. All rights
reserved. No part of 'this book may be In recent years, the traditional course in college geometry often
reproduced, stored in a retrieval system. or has been dropped, only to be replaced by other courses no more
transcribed, in any foml or by any means~ of
satisfactory. In some cases, the mistake has been made of throwing out
electronic, mechanical, photocopying, record- ty,
all of Euclidean geometry, both traditional and modern, whether it is of
ing, o'r otherwise-without the prior written of
permission of the publisher: Brooks/Cole continuing significance or not. The even worse mistake is sometillles
of
Publishing Company, Monterey, California. a made of assuming that students understand geometry simply because
ate
division of Wadsworth Publishing C-Ompany, '<:..
. they have taken some nongeometric course such as linear algebra.
:lIly
Inc. T believe that the traditional course in college geometry should
Ive-
ISBN: 0-81 85-005J-4 be replaced by an extensive course in modern geometry that includes
L.C. Catalog Card No: 72-79015 significant topics from modern Euclidean geometry. This text provides
Printed in the United States of America adequate material for such a two-semester or three-quarter course at the
1 2 3 4 5 6 7 8 9 10-77 76 75 74 73 ' upper-division college level. The scope is broad but nol all-inclusive.
Material is presented from foundations of geometry. finite geq.metry. the The various chapters of the text are largely independent. This
modern Euclidean geoOletl:y -~f '~on~ex'ity~ modern ~~Il~ge geometry and the arrangement of topics within each chapter allow for great
(inch.iding constructions), and transformations. There are separate flexibility in their use, according to the needs of lhe class and the
chapters on projective geo.111etry, 1l0n-Euc1idea.n geometry, inver.~!ve desires of the instructor. The teacher who wishes only minimal coverage
geometry•. and "geo'metric topology. Within the various chapters ~-;.~ such may use just the first sections of. a chapter. The taler sections may
new and Intriguing topics as caroms and Morley's theorem, as well:;as be used for a more complete and rigorous course if time allows.
fascinating geometry stemming from consideration of the g91den ratio. While the entire text pro.vides adequate material for a two-
Throughout, the text emphasizes practical and .up-ta-date applications semester or three-quarter course itl modern geometry. sections of it are
I
of modern geometry. The student should be aware 'that many of these app~?priate for use in shorter courses. Some suggestions for such use
topics are discussed in current professional journals and that con- are .given below.
temporary research mathematicians are seriously involved in. the extension
l. Two-quarter course in survey of modern geometry:
of geometric ideas.
All of text except for Chapters 4. and 5; also possibly omit
Most of the geometry included in this text is s.Yflth~Ji_c ..r£l.!tter Sections 3.6 and 7.8-7.10 in Chapters 3 and 7.
t~.1an. al~~!r.t~t;. but many important ideas from analytic geometry ,are
2. One-semester course in Euclidean geometry:
used as needed. An introduction to the analytic appr9ach is provided
All sections through Section 5.3.
for transformations and for projective and inversive geometry.
J. One-semester course in survey of modern geometry:
The text is not 1:1 sweeping survey. Instead, each major topic is
Finite geometries, Chapters 2,3,7,8.
developed so that the student can become quite knowledgeable about
4. One-semester course in modern geometry, not induding Euclidean
the procedures, concepts, and proofs involved in the various geometries.
geometry:
The chapters actually present concise courses on each kind of geometry,
Finite geometries, Chapters '6, 7, 8, 9.
not just summaries of what each type is about.
5. One-quarter course in Euclidean geometry:
The text is written for students who range widely in their
All sections through 5.3, p6ssibiy omitting Sections 1.5, 1.6, 2.6.
mathematical abilities. Much of the material is appropriate for those
3.6, and 4.6.
who have average or weak backgrounds in geometry. an the other hand,
6. One-quarter course in survey of modern geometry:
students with strong backgrounds wilJ also find much to interest them.
Finite geometries, Chapters 2, 3, 7. 8, possibly omitting Sections 2.6,
The text is planned for both majors and minors in mathematics. It is
3.6, and 7.8-7.10.
appropriate for students interested in mathematics from the liberal arts
7. One-quarter cqurse in modern geometry, not including Euclidean
standpoint and for those planning to be teachers of mathematics.
geometry:
M~ny 0"[ the first exercises in each set can be used orally: as
Finite geometries, Chapters 7, 8, 9.
the basis for classroom discussion; in that way the instructor can n1~ke
certain that fundamental concepts are unctC;;.t&lQod. Later exercises allow I would like to express my appreciation to James Moser, of
extensive practice in providing independent proofs. i i: the University of Wisconsin, Bruce Partner, of Ball State University.
.. The~texi'encourages independent investigation so that the stuaent John Peterson. of Brigham Young University. Demitrios Prekeges, of
will have experience in using geometric intuition, makIng conject#res, Eastern Washington State College, Curtis Shaw, of the University of
~lIld then p,+oving fonnni theorems. Advanced theorems are sometjmes Southwestern Louisiana, and Marvin Winzenread, of California State
inserted without proofifthey contribute to the development of a topic. University, Hayward, for reviewing the manuscript, and to the many
Because good teachers often prefer to bring in necessary re~iew serious college students who made worthwhile suggestions for improve-
topics when they are needed rather than in a single introductory ments while the material was being used in preliminary form.
chapter, the text incl-udes review at appropriate intervals. Students should
find a high school geometry text useful when a reference is needed.
CONTENTS
SETS OF AXIOMS AND FINITE GEOMETRIES
~
.I Introduction I
TI"Sets of Axioms for Euclidean Geometry 5
1.3 Introduction to Finite Geomelries II
1.4 Four-Line and Four-Point Geometries 14
Finite Geometries of Fano ami Pappus 19
i7Other Finite Geometdes 25
2 GEOMETRIC TRANSFORMATIONS 33 6 THE GEOMETRY OF INVERSION 197
2.1 Introduction to Transformations 33 6.1 Basic Concepts 197
2,2 Groups of Transformations 39 6.2 Additional Properties and Invariants under
2.3 Euclidean Molions of a Plane 45 Inversion 204
2.4 Sets of Equations for Motions of the Plane 51 6.3 The Analytic Geometry of Inversion 209
2.5 PropcI1ies of the Euclidean Group 58 6.4 Some Applications of Inversion 216
2.6 Motions of Threc·Space 66
2.7 Similarity Transformations 70
7 PROJECTIVE GEOMETRY 225
3 CONVEXITY 79 7.1 Fundamental Concepts 225

7.2 Postulational Basis for Projective Geometry 229
3.1 B~lSic Concepts 79 7.3 Duality and Some Consequences 233
3.2 Convex Sets and Supporting Lines 88 7.4 Harmonic Sets 239
3.3 Convex Bod~es in Two·Space 95 7.5 Projectivities 245
3.4 Convex Bodies in Tbree-Space 104 7.6 Homogeneous Coordinates 251
3.5 Conve, Hulls 108 7.7 Equations for Projective Transformations 255
3.6 Width of. Sel 114 7.8 Special Projeclivities 261
3.7 HeUy's Theorem mId Applications 120 7.9 Conics 265
7.10 Construction of Conics 271
4 EUCLIDEAN GEOMETRY OF THE POLYGON AND

CIRCLE 125 8 GEOMETRIC TOPOLOGY 277
J4.1 FundaOlental Concepts and TheOl-ems 125 8.1 Topological Transformations 277
\,4.2 Some Theorems Leading to Modern Synthetic 8.2 Simple Closed Curves 283
'. Geometry 133 8.3 Invariant Points and Networks 2139
,4.3 The Nine-Point Circle and Early Nineteenth Century 8.4 Introduction to the Topology of Surfaces 293
Synthetic Geometry (41 8.5 Euler's Formula and Special Surfaces 298
i4.4 Isogonal Conjugates 145
. \ 4.5 Recent Syntheti.; Geometry of the Triangle 150
. 4.6 Speci."11 Applications of Euclidean Geometry l54 9 NON-EUCLIDEAN GEOMETRY 305
,j9.1 Introduction t~. Hyperbolic Geometry 305
5 CONSTRUCTIONS 163 \./9.2 Ideal Points and Omega Triangles 312
'( 9.3 Quadrilaterals and Triangles' 315
5.1 The Philosophy .of Construc1ions 163 9.4 Pairs of Lines and Area of Triangular Regions 320
5.2 Constructible Numbers 1;69 9.5 Curves 326 ,
5.3 Constructions in Advanqed Euclidean Geometry 173 9,6 Elliptic Geometry 331
. 5.4 Constructions and Impossibility Proofs 180 9.7 Consistency of Non-Euclidean Geometry 335
5.5 Constructions by Paper Folding 187
5.6 Constructions with Only One Instrument 191
CHAPTER 1
Appendix 1: Hilbert's Axioms 341
Appendix 2: Birkhoff's Postulates 345
Appendix 3: Postulates from High School Geometry 347
SETS OF AXIOMS AND
Appendix 4: Examples of Notation Used in·Text
I .
351 FINITE GEOMETRIES
Answers to Selected hercises 353
Bibliography 363
Index 366
1 1 INTRODUCTION
The w"rd "geometry" clIn be interpreted liten,lIy lI$ "ellrth

measure," hUI Ihis literal meaning .'teems far removed from the various
modern geometries hl he explored in this text. The idea of earth
measure arises frOll1 the ancient, pre~Greek development or geometry.
These pfilctiLal Egyptian i.\nd Babylonian ttpplktltiolls (lr geometry in-
volved meaSurement to a great extent, tll1d they were not complicated
by formal proofs. For eXHmple, the properties of the right trinngle
were known to the extent thut it rope with knots {Figure 1.ll, held
tightly by three men so 1.\5 to form tl righl triangle, cOllld be used III
surveying.
SETS OF AXIOMS AND F'INITE GEOMETRIES 3
2 CHAPTER 1
The ancient Greeks of the period 500 D.C. to 100 A.D. receive
much of the credit for the development of demonstrative geometry of
the sort studied at the high school level. TIley recognize(r-l1re~beauty
of geometry as a discipline with a structure and, understood that the
proof of a theorem could be even more exciting than the discovery
of a practical application. The geometry of Chapters 2 through 5 is
essentially based on these early Greek contributions, though most of it
FIGURE 11
was developed much later.
Greek geometry, called Euclidean geom,etry because of the
. dence of earth measure beca~e monumental work of Euclid (300 B.C.), includes undefined terms, defined
During the Greek pertod, the s made a remarkably preCIse terms, axioms or postulates. and theorems. Almost every geometry studied
refined. About 230 D... C , Eratosthenes .d' .. I
to the famlhar s ory, in this text has the same sort of structure. Tn Euclidean geometry,
more . f th earth AcCOI tng . I undefined terms, which are arbitrary and could easily be replaced by
measurement of the SIze 0 e . solstice the sun shone direct y
t t the summer . ' other terms, normally include points, Jines, and planes; it would be
Eratosthenes knew th a a H r nd that at the same time, 10
into a well at Syene at noon .. 1 e d~~ north of Syene, the rays of the possible to develop Euclidean geometry using such concepts as distance
Alexandria, approximately 4~9r ml e: vertical (Figure 1.2). With! these and angle as undefined. Definitions of new words involve use of the
he undefined terms, The fact that definitions usually contain undefined terms
sun were inclined about 7.2 rom bl t find the diameter of the ,earth.
measurements, Eratosthenes w~s a .~h_:easurement aspect of geometry is standard in mathematics.
Interestingly enough. t le eal f h . portance of measurements Today, the words Qxi.QIt!. and postulate are used interchangeably.
. t because 0 t e 1m h
has been of recent mteres In by instruments placed on ~ e In the development of geometry, however, the 'word postulate was used
r great precision made by sate 1 es, G d tic Survey in producmg for an assumption confined to one particular subject (such as geometry),
o U S Coast and eo e
moon and by the .. while axiom denoted a "universal truth," a more general assumption
nauti~al and aeronautical charts, that applied to all of" mathematics. The axioms and postulates of Euclid
are stated in Section 1.2. The truth of axioms or postulates is ,not at
issue, These statements are beginning assumptior,s from which logical
consequences follow. They are analogous to the rules for a game. Since
the anathema,tical system to be developed depends on the axioms,
changing the axioms can greatly change the system, just as changing
the rules for a game would change the game.
T.llilarems are statements to be proved by using the axioms,
definitions, and previous theorems as reasons for the logical steps in the
proof. The theorems of geometry are valid conclusions based on the
axioms, A simple theorem typically is stated in the form of an jf~then
statement such as "If the sum of the measures o~r the opposite angles
of a quadrilateral is 180 (in degrees), then the ,'quadrilateral can be
inscribed in a circle," In logic, this theorem is, an ~mplicatiorl. The
hypothesis (the "if" part) is often identified as the statement p and the
Alexandria
conclusion (the "then" part) may be identified as the statement q. The
implication is "if p, then q," or p ~ q (p implies q). The proof of a
4B9 miles theorem means a proof of the entire implication, not just proof of
FIGURE 1.2
4 CHAPTER 1 SETS OF AXIOMS AND FINITE GEOMETRIES 5
the conclusion. A deductive proof begins with the hypothesis and leads Even this brief sketch of some of the major steps in the his lory
through a series of logical steps to the conclusion. Up to this point, of geometry should have cOllvinced YOlt that a discussion of modern
the proof has shown that the conclusion q is always true if the geometries must deal with many different kinds of geometry. It is the
hypothesis p is true. F Qf this reason, it is important to realize that acknowledgement of the diversity of mathematical systems deserving the
the deduction theorem from logic is involved whether or not it is title of geometry that distinguishes a book on 1110dern geometries from a
stated explicitly in the steps of the proof. traditional college geometry text of a quarter of a century ago, which
concentrated only on a (Jirecl extension of the Euclidean geometry of the
high school.
DEDUCTION THEOREM. If, under a given true premise p. the
conclusion q. is always trlle. then the entire statement p implies q is true.
EXERCISES 1.1
Since many examples of theorems and proofs will be given in
later sections, the concepts will become more meaningful with experience. (Answers to selected exerciseJ nre given at the back of the text.)
The "discussion above has provided a description of some of the 1. 3. Verify that the Pytllagorean theorem, liZ + b2 = c2 , holds for the sides of
basic premises and methods of synthetic geometry. Really significant the triangle ill Fig. 1.1. (." t"c.\..".lr~;""3k ~Q'i'b·I·C. \, .. ,,,~.: .f:; M<!Jej ')
advances over the synthetic geometry of the Greeks were made only with b. How many knots in the rope would be necessary to produce ll. trill.ngle
with shorter sides measuring 5 and 12 units?
the invention of analytic geometry (about 1637) and its subsequent use
as a tool in modern analysis. While analytic geometry is not the :\ 2. Use the measurements of Eratosthenes to find the approximate dinmeter of the
earth.
dominant theme of this text, coordinates of points are used as an
alternative to the synthetic approach when convenient. for Exercises 3-8, answer true or false; then ex.plain wh~\t is wrong with each
As 'the title Modern Geometries implies, the major emphasis is false statement.
on newer geometries that have been developed since 1800. The emergence 3. High school geometry owes more to the ancient Egyptians than to the
of modern algebra, with its theory of groups, and the introduction of ancient Greeks.
axiomatics into algebra paved the way for Felix Klein's classification 4. Euclid used the word posiulw/:, for all assumption confined to aue p,U"licuiar
of geometries in 1872. The basic concept of transformations needed to subject. .
understand this classification is discussed in Chapter 2. The latter part 5. The deduction theorem is useo because a conclusion, not an implication,
of the nineteenth century witnessed a revival of interest in the classical is to be proved.
geometry of the circle and the triangle, with the result that the Greek 6.. Analytic geometry was invented before the development of finite geometries.
geometry was extended by many signific..'\nt additions (Chapters 4 and 5). 7. The latter part of the nineteenth century witnessed ll. revival of interest
Projective geometry (Chapter 7) was invented about 1822; material on in the classical geometry of the circle and the triangle.
non-Euclidean geometry (Chapter 9) was in print by about 1830. 8. Traditional college geometry' of a quarter of a cenhll'y Ilgo included the
Inversive geometry (Chapter 6) was developed about the same time. study of more different geometries than are included today.
During the twentielh century, studies in the axiomatic founda-
tions of geometry and the finite geometries (Chapter 1), the geometry of··
convexity (Chapter 3), and geometric topology (Chapter 8) have all been 1.2 SETS OF AXIOMS FOR EUCLIDEAN
added to the great body of geometry that is relatively independent of GEOMETRY
analysis. ~1though they are outside the scope of this text, differential
geometry and vector analysis, to name but two areas of investigation, The importance of a wise choice of axioms was implied in the
could also: be covered in the study of modern geometry. prev~ous section, but the selection of axioms is no simple matter. A
SETS OF AXIOMS AND FINITE GEOMETRIES 7
6 CHAPTER 1
. for some 2000 years was how to provide an c. The need to state the fact that if a straight line enters a
major problem m geometry . E lidean geometry. Several such triangle at a vertex, it must intersect the opposite side.
adequate set of axioms for ?rdl~ary ,ue d. The need for statements about the order of points on a line.
" 1 sed lfl thiS sectIOn. d
: sets of axioms are d l~C1. S ve a famous set of five axioms an e. The need for a statement about the concept of betweenness.
Euclid, about 300 B.C., ga f. The need for a statement guaranteeing the uniqueness of a
five postulates as fonows: line joining two distinct points.
g. The need for a more logical approach. such as that of trans-
. Axioms (or common notions) formations (Chapter 21 which does not depend on the concept
. I equal to one
t to the same thmg are a so
1. Things that are equa of superposition. Euclid assumed that a triangle can be picked
another. . I the wholes are equal. up and put down in another place with all the properties
2. If equals are added to equa 5, 1 the remainders are equal. 'remaining invariant, yet no statement to this effect was made.
S
3 If equals are subtracted from equa h' equal to one·another. h. The need for a list of undefined temis.
. . 'd 'th one anot erare - Many modern sets of axioms for Euclidean geometry have been
. : 4. Things that comel e WI
s. T he whole is greater than the part. introduced to r~~<?dy the defec~s in Euclid. In a course on the
foundations of geometry. or on the foundations of mathematics, these
postulates are studied extensively. In general. these newer sets of axioms were
f ny point to any point.
1. A straight lin~ c~n ,be draw~ r~;o~uced continuously. in a straight mOl'e comprehensive than that of Euclid, but for this reason they appear
2. A .finite stralght hue can e more complex. it has been difficult to develop the beginning high
school geometry course without resorting to the device of introducing
line. ' . with any point as center and any
3. A circle may be described still more axioms to avoid. the proof of very . difficult theorems at the
distance as radius. beginning of the course.
4. All right angles are equal to one anl?ther.. such a way that the One of the first modern sets of axioms for Euclidean geometry
. 1 falls on two mes m was devised by Moritz Pasch in 1882. He is given credit for what is
5. If a transversa . f the transversal are less than two
interior angles on one. SIde 0 th t side on which the angles called Pascll's axiom: A line .entering a triangle at a vertex intersects
right angles. then the hnes meet on a the opposite side. Guiseppi Peano provided ~nother new approach in
are less than two right angles. 1889. Probably the most famous set of axioms for Euclidean geometry
'. nee that a segment can be extended was given by David Hilbert and was published in English in 1902.
postulate 2 means tn esse I. f Euclid (the parallel These axioms are given in Appendix 1. Hilbert used six undefined terms:
· The fiifith postu ate 0
indefinitely to form a Ime. . er 9 in the discussion of pOint, line, plane, between, congruent, and on. He gave his axioms in
postulate) will be considered further tn Ch~Pt five groups, and the grouping helps show how the logical difficulties
non-Euclidean geometry. . particular I mathematicians of Euclid were overcome, For example. note ·that Hilbert's first three
tu
Within the past hundred ye~rs h i ointed 'but various flaws axioms clarify Euclid's postulate about drawing a straight line from
· s of mathematics ave p
studying the foun d a t ton 'd II used other tacit, assump- any point to any point and that his axioms of congruence take care of
. f Euclid Euch actua y '. d
in the assmnpttons 0 . r 'tt ) Logical problems pomte out the logical defect of superposition.
tions {assumptions nbt stated exp lCI y, . Since the timeofHilbel't there have been many other sets of modern
have included: fi. t t ment about :the continuity of postuJates for Euclidean geometry. Some of these are by Oswald Veblen
a. T he need for a de 1l1te 5 a e (in 1904 and 19l!), Huntington (1913), Henry Forder (1927), G. D.
lines and circles. h . fini~e extent of a straight Birkholf (1932), and others. The axian1s of Birkholf (Appendix 2) are
b. The need for a statement about t e 10
of special signi"ticance because they emphasize the connections between
line.
8 CHAPTE~ 1 SETS OF AXIOMS AND FINITE GEOMETRIES 9
geometry al1d the rea! numbers, because they include c/istOllce and angle that associates the numbel.· I with il certain polygonal region
as undefined terms, and because they have been modified for incorporation and a unique positive real number with every convex
in most rec~nt experimental geometry courses, such as that of the School pOlygonal region; a statement that if two triangles arc con-
Mathematitfi Study Grotlp, a[ the secondary school leveL gruent, the respective triangular regions have the same urea;
One of the main reasons why BirkholT's system is so brief is the and the assumption that the measure of area of [I rectangular
power of tt,le first postulale. Since this assumption has the eITect of region is the product _of the measures of the lengths of its
assuming aU the properties of the real numbers, the order relations base and altitude.
for points depend only on theorems and definitions rather than on An example of a modern set of axioms for secondary school
additional postulates. geometry appears in Appendix 3.
Because modern geometry programs depend on the coordination Sets of axioms for Euclidean geometry, as well as for any
of analytic geometry with synthetic geometry lUore than was the case mathematical system, should have two important properties.
even at the time of Hilbert, axioms fox the real numbers and the
logical ass.umptions of algebra for relations are needed in geoflletry. L The set should be consistent. In a consistent set ofax.ioms, it is not
Se~s of aXIOms for modern geometry texts ordinarily contain specific possible to use tllemto···prove a theorem that contradicts any
aXIOms needed for each of the following purposes not readily apparent axiom or other theorem that has already been proved. All of the sets
as a result of the axioms of Hilbert: of axioms in this text are examples of consistent systems. The
a. To show the existence of a correspondence thnt associates a concept of consistency becomes more significant after the discllssion
unique number with every pair of distinct points. in Chapter 9 of the consistency of non-Euclidean geometry. An
b. To establish the measure of the distance between any two example of an inconsistent system could be one that inclUded both
points in the line as the absolute value of the dilference of of these axioms: a. 'Two distinct points determine exactly one line;
their corresponding numbers. b. Two distinct pojnts determine'ex<1ct1y two lines.
c. To stat~ the existence of a unique coordinate system for a 2. The set should be, COIm?!!!..te. It should be impossible to add a
line that assigns to two dislinct points two given distinct consistent, independet{t (see definition below) axiom to the set withOUl
i'eal numbers. introducing new undefined terms.
d. To formulate the ,logical assumption necessary for the
development of the theory of convexity (Chapter 3) by stating In the study of foundations of mathematics, it is· often important
that a line in a plane partitions the points of the plane not that the set of axioms have at least one additional property so that
on the ,line into two convex sets such that every segment none of the axioms can be proved from the remaining set of axioms.
~hut joins a point of one set to a point of the other Sels of axioms having this property are calle.d independent sets of
_intersects the line. axioms. All of the sets of axioms in the remaining sections of this
e. To include additional assumptions about congruence of chapter are examples of independent sets of axioms. The requirement
triangles-assumptions that eliminate the need for lengthy of independence is not always desirable at a more elementary level.
prools at an early stage in the text. The two most common Therefore, secondary geometry texts ordinarily have additional axioms
additional axioms are the assumptions that (1) congruence of that could be proved fro III the others. These axioms are included
'triangles follows from (.,"Ongruence of tw~ angles and the in- because they are convenient to -stale and use early in the course or
cluded side and that (2) congruence of triangles follows from because the proofs are too difficult for that level. The assumptions of
congruence of the three sides. Euclidean geometry, along with some theorems proved in an introductory
f. To postulate additional assumptions about fundamental con- course, are the basic· assumptions used in Chapters 2 through 5 of
cepts of area. These include the existence of a correspondence this text. However, you won't be expected to remember eHcll theorem.
SETS OF AXIOMS ANQ FINITE GEOMETRIES 11
10 CHAPTER 1
1.3 INTRODUCTION TO FINITE
I order to begin the study of the appiic~tion of specific
GEOMETRIES
n . etries it is helpful to mtroduce a type
sets of axioms to ~Ulld geom. Ie' a structure as possible. In the rest
.,I of geometry that dIsplays as sl.mp 1 for this purpose are finite From the standpoint of structure, Euclidean geometry is complex .
of this chapter, the geometries c l~sen The Euc1idean plane has an infinite number of points and lines in it,
t geometries, explained in the next sections. and a rich collection of theorems continue~ to increase over the years.
I By contrast, "miniature" geometr:ies have just a few axioms and theorems
I' EXERCISES 1.2
and a definite number of elements that can be named by a counting
number. These geometries are finite geometries. and they provide excellent
L Draw a figure to explain the wording of Euclid'S. fifth postulate. opportunities for .study of geometries with a simple structure.
All of the geometries studied in this text have a finite number of
Draw a figure to exp1am
. tl1e following axiomfi of Hilbert.
axioms and a finite number of undefined terms. Thus those features
2. Order Axiom 2\ do not make a geometry finite. Instead, a finite geometry has a finite
3. Order Axiom 4. number of elements-that is, points or lines or "things to work with."
For the geometries studied in this chapter, these elements can be
4" Order Axiom 5.
considered points and lines. It would seem that finite geometries are
5. Congruence Axiom 3.
thus inherently simpler than geometries with an infinite number of
6. Congruence Axiom 4.
points and lines, although that may not be your opinion when you first
1. The axiom of continuity. encounter them.
. 8- to, name the axioms of Hilbert that:
For ExerCises Historically, the first finite geometry to be considered was a
three-dimensional geometry. each plane of which contained seven points
8. State Pasch's axiom. .
9. Guarantee the uniqueness of a line joining two ~istinct pomts. and seven lines. The modernity of finite geometries is emphasized by
the fact that Fano explored this first finite geometry in 1892. It was
10. Deal with betweenness for points on a line. .
not until 1906 that finite projective geometries were studied by Veblen
For Exercises 11-16, state which axioms from Appendix 3 : . . and Bussey. Since that time, a great many finite geometries have been
It. Show the existence of a correspondence that associates a uOlque number wtth (ir-a"re being studied. Many sets of points- and lines that were already
every pair of distinct points. . . familiar figures in Euclidean geometry were investigated from this new
. b t en any two points III a hne. point of view. However, at the present time it is quite possible for a
12. Establish the measure of the dls\ance e we
13. State the existence of a coordinate system for a line. mathematics major to graduare without ever encountering finite geo-
metries, although it is also true that finite geometries are being used
14. Introduce the concept of convexity.
increasingly as enrichment topics and extension units at the high school
l5. Provide assumptions about congruence of triangles.
level. Finite geometries also find a practical application in statistics.
16. Provide assumptions about the area of plane regions. All of the finite geometries in this chapter have point and line
For Exercises l7-20, state wh~ther a set of axioms could be: as undefined terms. The connotation of line is not the same in finite
geometry as in ordinary Euclidean geometry,. however, since a line in
17. Complete but not independent?
finite geometry canoot have an infinite numbe!' or points on it.
18. Independent but not complete?
The first simple finite geometry to be investigated, called a
19. Independent but not consistent'l three-poilll geometry here for identification, has only four axioms:
20. C onsistent but not independent?
12 CHAPTER 1 SETS OF AXIOMS AND FINITE GEOMETRIES 13
Axioms for Three-PolM! Geometry THEOREM 1. L Two distinct lines are on exactly one point.
I, There exist exactly three distinct points in the geometry.
2. Two distinct points are,on exactly one line. By Axiom 4, two distinct lines are.on al least one point. Assume
3. Not all the points of the geometry are on the sfl,me line. that two lines. are on more than one point. If tW? dist~nct lines I anu JlI.
4. Two dislinct lines are on at least oue point. lie on poinls P and Q. then Axiom 2 is c01)tradicted, because points
P and Q would be on two distinct lines ..
Some immediate questions to consider intuitively before reading Theoren1 1.1 is proved by what is called an i,/dirt'ct lirgwlIt'lIf.
farther are the following: The theorem could be rewritten" in the form of an implication. If twO
a. What kinds of figures or models could be drawn to represent lines are distinct, then they are on exactly 0lne potrit. The assumption
the geornetry? was made that the conclusion was not true, and a contradiction _was
i
b. Ho'd rnany lines are in the geometry?
c. Can. any theorems be proved for the geometry?
i, reached, showing that the negative assumption was nol tenable. Thus,
the conclusion is valid, and, by the deduction theorem, Theorem L 1
d. What representations are possible for the geometry, other than is established. Indirect proofs, which are probably more elTective in
tho~e with points and lines? geometry than in algebra, will be used many times in this text.
e. Which properties of Euclidean geometry continue to hold in The ex-act number of lines in the three-point geometry can now be
the .three-point geometry, and which do not? determined.
Not all of these questions can be answered completely for each
finite geometry studied, but the questions do illustrate the nature of
THEOREM l.2. The three-point geometry has exaclly three
inquiry about a geom,etry based on an axiomatic system. It will be
itml1'ediately helpfu ( to give 11 partial answer to question a. lines.
The n~ite geometry of three points can be represented by many
drawings, four of which are shown in Figure 1.3. Verify the fact that all From Axiom 2, each pair of poipts is on exactly one line.
rour axioms hold for each figure. Each possible pair of points "is on a dis_tinct line, so lhe geolllet.r y has
at least three lines. Suppose there is it fourth line. From AXIOm I,
there are only the three points in the geometry, This fourth line
must also be" on two of the three points, but this contraciicts Axiom 2
and Theorem 1.1. There can be no more than three lines in the
geometry. .
While point and line have been used l:lS the undefined terms
in this first finite geometry, other words: could be substituted to give an
equally meaningful interpretation of tI~e structure. For example, fJ:~e
FIGURE 1 3 could be substituted for poillt, and row for line, so that the postulates
would reau as follows:
While the sets of points and lines in Figure 1.3 are such that u. There exist exactly three dislinct trees.
all the axioms of the three-point geometry hold, there is still the b. Two distinct trees are on exactly one row,
possibility that the geometry might have additional lines not shown. c. Not all trees are on the same row.
This matter is settled by proving two theorems. First, a compmisotl of d. Two distinct rows have at least one tree in common.
the wording of Axioms 2 and 4 leads to the need to determine Other interpretations could be found by lIsing pairs of .worl~s such <\s
whether two distinct lines might be on mOre than one point. beads and wires, students and committees, or books and IIbranes.
14 CHAPTER 1
The next finite geometry to be c '
In the finite geometry under consideration. it should be evident lines as undefined terms Th 'II .onsldered also has points and
. e ,0 owmg thr .
from an examination of the axioms and Figure 1.3 that such Euclidean characterize the geometry called f i ' ee aXIOms completely
tion. ' a our-lrne geometry here for identifica-
concepts as length of a segment, measure of an .ngle, and area-in
f.ct all concepts concerning measurement-no longer apply. In this
geometry, if a triangle is defined .s three distinct lines meeting by pairs Axioms for FOUI"- Line Geometry
in ,three distinct points, not all collinear, then on'e and only one
1. The total number of lines is four
triangle exists. The concept of parallel lines does not apply, if parallel
2.
3 Eachh pair
E . of. lines has exactly o' . In
ne pomt . common
lines are defined as two lines with no points in common, since each
. ac pomt IS on exactly two lines. .
two lines meet in a point. The fami.liar ideas of congruence also have
no meaning in this geometry. Even though you are far more familiar
Axiom "I is an existel1ce axiom b .
with Euclidean geometry than with finite geometry, thi's consideration of geometry is not the en t • .ecause J{ guarantees that the
. lp Y set of pomls Th tI .
whether the Euclidean properties hold in a new geometry will neverthe- mcidence axioms deaII'n 'Ih' , e 0 'ler
, g WI pomts on r . aXIOms are
less give added meaning to the familiar concepts. The introduction of Before reading on try to dra d' mes and hnes on points.
more significant finite geometries in the next sections wi1l provide satisfy all of the three axiol wAIlagrams of points and lines that wiH
of points. TIS, so, try to determine the total number
additional opportunities for the same kind of consideration.
.,,
THEOREM I ,.
3 TIle four-line geometry has exactly six points.
EXERCISES 1.3
For the Ihree-point geometry: By Axiom 1. there are six pairs of l'
I. Draw a pictorial representation differenl from those in Figure 1.3. obtained as the combination f f . h' meso The number six is
t ['" 0 om.t HlgS taken two at a tlllle. Thl'ee
2. Rewrite the ax.ioms, using the words bool, for IJoillt and library for line. no a 100S used for combinations are
3. Rewrite the axioms, using the words student for point al~d committee for line.
1'\.=) 4. Through a point not on a given line, there are how soan), lines parallel to C4 •2 • 4 C 2' or '4)
(2'
the given tine'1
~'5. Exactly hoW many points are on each linei The general formula for the combination of
time is 11 things taken r at a
"".,6. Must lines be straight in the Euclidean sense?,
,;:. 7. Could three lines all contain the same poiql? i, n!
(n - r)!'-!
fl..~ -:.!:J 8. Do an)' squares ,;xist'l I
~.. ~9. Prove that a Ii~e cannot contain three distinct points.
[ '1 ~BYuppose
Axiom 2," each pair of rmes Ilas
£
that two of these s,'

exact Iy one . t .
I
(i:
S x pomts are n t d' . pom 10 common.
,
contradiction 'of Axiom 3 b 0 lstmct. That would be a
1.4 FOUR-LINE AND FOUR-POINT I . • ecause each point wo Id b
two hnes. Also • by Ax"om 3 no other . u e on more than
GEOMETRIES eometry other than those six 0' I '. ~OIllt
n t le pau s of hues.
could exist in the
The three-point geometry of. Section 1.3 had a structure so
simple that it seemed of relatively little significance. The two related THEOREM 1.4. Each line of th e ~our- I"Ine geometry has exactly
finite geometries in this section are more rewarding to study, yet their three points on it.
sets of axioms are even briefer.

·16 CHAPTER 1 SETS OF AXIOMS AND FINITE GEOMETRIES 17
By Axiom 2, each line of the geometry has a distinct point in Axioms for Fow'-Poil1t Geometry
common with each of the other three lines, and all three of these
1. The total number pf points in this geometry is four.
points are on the ,given line. Suppose there were a fourth point on the
2. Each pair of points has exactlv one line in common.
given line. Then by Axiom 3, it must also be on one of th(; Qther line.~.
3. Each line is on exactly two points.
But this is impossible. because the other three lines already determine
exactly one point with the given line, and by Axiom 2, they can only
Two possible representations for this geometry are shown lJ1
determine one. Thus, each line of the geometry has exactly three points
Figure 1.5.
ouit.
Figure 1.4 shows two diagrams thal can be used to represent
tbis finite geometry of four lines and six points. An examination of
Figure 1.4 will lead to further inquiries about the four-line geometry.
For example, the following questions nre typical of those that should be
answered: ,I
FIGURE 1.5
One thing that should be noted in Figure 1.5 is that the lines
meet only where points are indicated, not just where tl1t~Y appear to
cross in the picture,. because these are not ordinary Euclidean lines. If
it is assumed that the four-point geometry and the four-line geometry
an: related in such a way that the plane dual of any vat1d theorem
in one geometry becomes a valid theorem in the other, then it is possible
FIGURE 1.4 to gain m.ore inrormation about lhe new geometry rather easily. Thus,
the plane duals of Theol:ems 1.3 and 1.4 become theorems ror the four-
a. Do en.ch lwo points of the geometry lie on a line? point geometry.
b. How 'many triangles exist in the geometry (all three sides
must be lines of the geometry)? THEOREM 1.5. The four-point geometry has exactly six tines.
c. Doesthe geometry have examples of parallel Hues (lines wilh
no point in common)?
These questions are considered in the exercises at the end of the section. THEOREM 1..6. Each point of -the rour-point geometry has
eXflclly lhree lines on it
The general concept of dualitv is c'onsidered in Chapter 7) but a
specific example is needed here in order to explain how the next set of .
axioms is formulated. The plane dual QL~_.it_atern~~. forrned by Lines exist in the four-point geometry that do not h ..lVC one of
exchanging the wordsnp.~i~.t.~:!..~~.tiug oth~J!E..~...s;a~=~h-~ges the four points in common, so these lines may be considered parallel.
ulfhe English as req~lired. :Writing the plane dual of each axiom for the Other properties of the four-point geometry are investigated in the
prevIoUs four-line geometry results in a four-point geometry. exercises.
1.5 FINITE GEOMETRIES OF FANO

18 CHAPTER 1
AND PAPPUS
For each of the finite geometries sO far, one axiom stated the
exact number of points on a line or gave the total number of points The original finite geometr f Fano was a three-dimensional
geometry, but the cross sectio f y 0
or lines for the geometry. Without this limiting axiom, the sets of
configuration yields a plane fi " armed by a piane passing through his
axioms might have resulted in geometries with an infinitude of points
and lines. Indeed, most of the axioms in sets for finite geometries are
\ 111 t e geometry al
that. can be studied here. The comp Iete set ofax',mus
It d
• so ea efollows:
F ano 's geometry'
also valid axioms in Euclidean geometry. For example, Axioms 2 and 3

for the finite geometry of three points hold in ordinary Euclidean Axioms!or Fano's Geometry
geometry, and even Axi01U 4 is true except when the lines are paralle\.
1. There exists at least 0 ne I'me.
. The situation is similar for other finite geometries. 2. Every
3. Not allline of the
points of geometr
tl y has exactly three points on it.
le geometry are a tl .
4. For two d' t" n Ie same Ime.
them. IS met pomts, there exists exactly one line on both of
,
I
,
I.
EXERCISES 1.4
Write the plane dual o[ the axioms [or the three-point geometry o[
5. Each two lines have at least one point in common,'
Section t.3. For finite

in previous this geometry
g " of F a
0,n '
pomt .
and Ime are undefined, As
eometlles, the meaning f .
For the four-line geometry: to the intuition. In finite . 0 011 m the axioms is left
2. Draw another representation [or this geOinetry different [rom those shown in
geometry, various expres~~~~e~:les bas well as ill ordinary Euclidean
example all of th n e used for the same idea. FOI'
Figure 1.4. • ese state the same relationship:
"" 3. Which axioms are also true statements in Euciidean geometry?
4. \1.ewrite the set o[ axioms for this geometry, using stu£i,"t [or point and
A point is on a line.
CU/lIlllirtee for line. The line contains the point.
q 5. Do each two points of the geometry He on a line?
,) 6. BoW many triangles exist in the geometry in which all three sides are The line goes through the point.
lines of the geometry 7 An almost immediate consequence of the axioms is:

_;;. 7. How many other lines are parallel to each Hne?
For the four-point gemnetry: common.THEOREM 17

.. Eaclt two· lines have exactly one point in
Draw another model [or this geometry different from those shown in
8.
Figure 1.5. e
Which axioms are also true statements in Euclidean geom try1
0;; 9. \1.ewrite the set of axioms for this geometry, using tree for point and rOW By Axiom 5, two lines h
The assumption that they have t avde:
A '
a~ .
least one pomt in common,'
wo 1st met poi t .
10. XlOm 4, because then the two d" . n s III common violates
for line.
Do each tWO lines of this geometry determine a point'? containing both of them. !Stmct pOl\lIs would have two lines
""ll. How m.ny othe< lines of the geometry are parallel to ench line?
For the geometry of F
more difficult to guess. Try to an~~s:he numper of points and lines is
I '" \2. Prove, without using the idea of duality, that the geometry indudes exactly
-1 13 .
.~ 14.
leading to the next theorem. g before readtng the development
I six. tines.
Prove, without using the idea of duality. that each point o[ the geometry
is on ex.actly three lines.
20 CHAPTER 1
SETS OF AXIOMS AND FINITE GEOMETR!ES 21
From Axioms land 2, there are at least three points in the THEOREM 1.8. Fano's geomelry consists of exactly seven points
geometry, while from Axiom 3 there is at least a fourth point, as and seven lines.
symbolized in Figure 1.6a. By Axiom 4, there must be lines joining
this rourth point and each or the existing points (Figure 1.6b), and
Note in Figures 1.6c and t.7 that lines in finite geometry may
by Axjoms 4 and 5, there must be lines joining points I, 6, 1, painls
appear to cross without actually having a point in common. They are not
J, 6, 5, and points 5, 2, 7 (Figure 1.6c). Thus the geometry or 'Fana
contains at leasl seven points and seven lines. Euclidean iines, and the distinct points 011 them need (0 be clearly
shown in drawings to avoid confusion.
4 4 Consider a rewriting of the set of axioms for Fano's geometry
• 4
with the word point replaced ~y student and the word lille by
committee. A table may be used to represent this finite system. For
example, Table 1.1 shows committees in vertical columns. Substitution
of numbers for the student names would show that Fano's geometry
~-~---~
2 3 itself could be represented by the same sort of table. You should be able
t') tb) to set up a correspondence between each column in this table and
(e)
FIGURE 1.6
each line in Figure 1.6c, noting that the students can b~ matched in
alphabetical order with the points in numerical order.
The fact that there are exactly seven points in Fano's geometry
can be established by an indirect argument, since the a<;slll11ption of an TABLE 1.1
eighth point leads to n contradiction, as will now be shown. Assume
that there is an eighth point, and consider for example, as in Figure 1.7, Cnml/lifter 1 C(Jlllllliffer 2 Commure) 3 Cmllmirree 4"""'
Alice Alice Alice Brad .'
the intersection of the line through points I, 8 and the line 3, 7, 4. Brad '1 O'I'~ Frank .- Dille .;
(The notation «line 3, 7, 4" means the line containing points 3, 7, and 4.) Cathy~l Ellief:) Greg Frank
Axiom 5 requires that lines I, 8 and 3, 7, 4 have a point of inter- CO/lfmi(lf!e 5 Committee 6 C(III""itlet! 7
section. The point of intersection required by Axiom 5 cannot be point Cathy Brad ~ Cathy
3, 7, or 4, since that would violate Axiom 4. Thus, it must be a ninth
Ellie .- Ellie .;.- Dl'lle -
Frank Greg . .-::. Greg
point, but that violates Axiom 2. The assumption of an eighth point
has led to a contradiction and must be rejected. The result is the
rallowing theorem: The connections between the theory of combinations and lhe
number of points and lines in finite geometries are profltabl~ to explore.
For example. it might appear at first glance that a qUick way to
determine the number of lines in .Fano's geometry would be to count
all possible combinations of seven things taken three at a time, since
each line has three points. But there are 35 combinations of seven
things taken three at a time. and there are only sevel~ lines. in the
geometry. A study of Figure 1.7 (or L6c) will make It po~slble to
reconcile this difference. Consider poinls 1 and 5 and the line lhey
2 3 determine for example. There are five other possible points, 2, 3, 4, 6,7,
FIGURE 1.7 that could be. matched with the two given points (0 rOfm lines, and all
22 CHAPTER 1
of these possibilities are counted in the 35. But only one of these points and lines that have long been familiar to mathematicians. This
points, point 4, is actually on the line through poinfs 1 and 5; hence, only familiarity is especially evident in the finite geometry arising from a
1 of the actual number of possibilities result in lines. Since ¥ = 7, there Euclidean-geometry theorem called the Theorem of Pappus. Figure 1.8
~re a total of sevel~ lines. Further insight can be gained by asking illustrates the theorem, which was discovered and proved by Pappus of
why the other four points, 2, 3, 6, 7, are not matched with points 1 Alexandria about 340 AD. The theorem is stated here without proof. The
and 5 to determine lines of the geometry. for any point such as I, lines in this theorem are considered to be the same as lines in ordinary
there are three Hnes of the geometry on it. These contain the other six Euclidean geometry.
points of the geometry {two on each line}. None of the pairs of points
on one of these lines can be matched with a point on another line
THEOREM 1.9. TI!eorem of Pappus: If A, B, and C are three
without contradicting Axiom 4. distinct points on one line and if A', R', and C' are three different
The concept of parallelism is not in evidence in Fano's geometry,
since each pair of lines has a point in common. .on the other hand, it
distinct points on a second line. then the intersections of
1'--1- -+-+ +--4 +--+
1
rc
and C7t\
All' and BAt, and Be' and CB' are collinear.
would be quite possible to give a new interpretation of paraliel for this .~='"
(\
finite geometry so that the concept could be I considered. in' a way
quite different from that used in Euclidean. geometry. For example,
suppose (using Figure 1.6c) that any two lines intersecting on:line 4. 7, 3 cr."
-
The notation AC' means "the line containing points A and
(A summary of the technical notation us~d here and throughout
\ \... are called parallel in Fano's geometry. Then lines 5, 2, 7 and 1. 6, 7, this text may be found in Appendix 4.) Poi~ts are called collinear if they
o
for example, are parallel in this interpretation l since they have point 7 lie on the same line. There are exceptions to the theorem if some of the
on line 4, 7, 3 in common. This example helps to emphasize the lines are parallel (in the ordinary Euclidean s.-::nse). but it is assumed
important role of arbitrary defini!ions in determining the structu~·e· of a here that the Hnes intersect as required in real points. The theorem of
geometry. Pappus is seen to involve nine distinct points, lying by threes on three
Some finite geometries, though quite modem, relate to sets of lines. There are nine points and nine lines in F~gure 1.8, and these may
be studied as a finite geometry with the familiar'Euclidean properties no
longer evident and with only the following axioms.
Axioms/or Finite Geometry of Pappus

1. There exists at least one line.
8 2. Every line has exactly three points.
3. Not all points are on the same line.
c' 4. There exists exactly one line through a point l~ot on a line that is
parallel to the given line.
E F 5. If P is a point not on a line, there exists exactly one point P' on
A
i the line such that no line joins P and r.
6. With the exception in Axiom 5, if P and Q are distinct points,
then exactly one line contains both of them.
c
Study Figure 1.8 to see that each line of the finite geometry
FIGURE 1.8 seems to have exactly two lines paralJel to it. Of course, this observation
.~
I,";. '.
24 CHAPTER 1 SETS OF AXIOMS AND FINITE GEOM ETAIES 25
"
does not cOll.stitute a proof. Axiom 4 is the familiar parallel axiom of TABLE 1.2
Euclidean geometry, in a somewhat different setting.
]
F F n I
The next theorem gives an additional property that can be A lB' D A A
I
proved from the axioms. B lA' E D B' C B II'
I C
I
iC F C' E B' e E
!
THEOREM l.lO. Each point in the geometry of Pappus lies on l
exactly three lines.
check~(t directly to
By Axioms 1 and 2. there exists a line with three points
..
:
I
cases in the table. For example, the ta.ble C<.'ln be
see that each poidt lies on exactly three hnes.
I
I,
1
A, B, C On it (Figure 1.9). By Axiom 3, there is a fourth pOint I
(point X in Figure 1.9) not 011 this Hne. Consider the total number of
EXERCI~ES 1.5
lines on X. which represents any point of the geometry. By Axiom 5,
X lies on lines meeting two of the points on the given line, say i
1. Rewrite the sqt\ ofax.ioms ~or the geometry of Fnno. using book for point
Band C. By, Axiom 4, there is exactly one line through X parallel to and libmry forpine.
Dc, so that there are at least three lines on X. But there cannot be a f\-<---)~ 2. °
Which ax.ioJl1~, in the geometry f f at10 are also true statements in
fourth line through X. By Axiom 5, there is no line connecting X and Euclidean geo!"netr y ?
A, and by Axiom 4, there is no other line through X not meeting ?) 3. Using Figure 1.6c, name all I he tnang . . the geometry of Fano having
Ies III
Bc. point 4 as oneiverlex. . '
,. . p. 116cif parallel lines are defined as those thaI mtersed on line

~s~~!"\ 11~~1::el ~U' the pairs. of parallel lines in the geo~netry tdo not use
i l- ,
x "7 4,
I
A 8
IS :
FIGURE 1.9
c
~5.
6.
7.
1,2,3 as a lnimber of the pan's).
for Fano's gqomelry, prove that each point lies on exactly three lines.
Rewrite the
f_~~~_~_
Which axion~s in the geometry of Pnpptls are also tnle statements III
Euclidean ge~t1letrY'1 . .
.
~et of axioms [or the geometry of Pappus, using the words

.
Although the disclission of the geometry of Pappus began whh 8. For ench point in Ihe ge?n1etry of Pappt~s. hOW. n;any o~her pomls 111 the
the complet1ad figure, it can be proved from the axioms alone (see geometry do hot lie on n hne through the gIven pom ?
Exercise Set 1.5) that the geometry of Pappus does in fact consist of 9. Prove that tllel'e are at least twO lines in the geometry of Pappus pnrallel
exactly nine-points and nine lines. Other versions of the axioms can be to a given li~'e.. . .
found using substitute words for points and lines, but it is also instructive to. Prove that t~e geometry of Pappus contains exactly n~lle I~omts.
Lo arrange the information in the form of a table. In Table 1.2, the t l. Prove that tt geometry of Pappus contains e,acfly rune hnes.
\,
notation fforil Figure 1.8 is used, with each vertical column representing
a line of the geq,metry. I
It is important to understand that a table such as Table 1.2 1.6 @THER FINITE GEOMETRIES
could be used, rather than a set of axioms, to give the initial I .
representation of a geometry. Note, however, that in this case Illost A o~ example of a modern finite geornetry lhat. is actt~a~ly
proofs are very simple, since they depend just On checking all possible a study of ;~ar'0US set of points from Euclidean geometry rs the .fi""e
,
:r
"'i
, SETS OF AXIOMS AND FINITE GEOMETRIES 27
I
26 CHAPTER 1 I
the polar of the point. In the finite geometry of Desargues, no Hne joins a
I
" conce ts must be introduced so that they
geometry of Desargues. Several h' P geometry Triangles ABC and pole and a point on a polar. This fact leads to the formal definitions,
can be used in the study of t IS ~ew . ; P This means that Wl1ich do not use the concepts of point or line of perspectivity.
AIB'C' of Figure 1.10 are perspectwe from pam . ) of the three
. t (the point of concurrency , ~
point P is the common pom I' Afil jjjf and CC. The line I in the finite geometry of Desargues is a polar of the
!~
d' g vertices Thus P ,es on , point P if there is no line connecting P and a point on I.
lines joining correspon m 't d' din lUore detail in Chapter 7,
· t Desargues' theorem, s U Ie /.
ACCOf.d mg 0 " oint are aJso perspective from a me. Tlle point P in tbe finite geometry of pesargues is a pole of the
two trtangJes perspectIve. [10m a P I' corresponding sides of the line I if t11ere is no point common to I and any line on P.
. I
If tnang es are p .
erspechve [rom a me,
. r
.' .
TI e fine of perspectJvlty JU I
. I t at pomts all th,s me. 1 . AlB The intuitive development has led to the axioms, but now the
tnang es mee. . S T For example, corresponding SIdes
Figure 1. 10 contams pomts R, , .. . axioms become the beginning ideas; nothing" can be proved from
and 7n' meet at T, a point on thIS Ime, Figure 1.10--proof must come fi'om the axioms themselves.
See if you can identify various poles and polars in Figure 1.10.
For example, if point T is taken as the pole, where are the polars?
This question can be answered by considering the lines through T
and noting that there is exactly one line in the figure that has no points
~ ~
in common with these lines. That line is PC', hence PC' is the polar.
Axioms/or (he Finite Geometry of Desargues

1. There exists at least one point.
2. Each point has at least one polar.
3. Every line has at most one pole,
4. Two distinct points are on at most one line,
5. Every line has at least three distinct points on it.
6, If a line does not contain a certain point, then there is a point
on both the line and any polar of the point.
T
;,"" The exact number of poles for a polar and the exact number
of polars for a pole are not stated expJicitly in the axioms. Study
FIGURE 1.10 Figure 1.11. Assume that line p is a polar of P, since Axiom 2-
. ure 1 to shows a total of ten labeled
A further study of Fig . ' h "line and three lines on p
"
points on ten Imes, W I ,
'th three pomts on eae
d I' re the elements in the •
. Tl ten pomts an ten mes a
each POUlt. "lese e further concept is necessary to under-
finite geometry of Desargues. On . a correspondence between
. . method 0 r settmg up ", ,
stand the aXlOms, a . . th point or perspect1v1ty for -+--~---------+----p
. . ded If a pomt 1S e
points and hnes IS. nee, ", the line of perspectivity for the same two A 8 c
tWO triangles, and If ~ h~e IS h I f the line and the line is called FIGURE 1.11
triangles, then the pomt 15 called t e po e 0
SETS OF AXIOMS AND FINITE GEOMETHIE.S 29
28 CHAPTER 1
q'" I _ I
guaranlees the existence of at least one. By definition, no line through ~.
-----.
q -
P contains anyone of the points (at least three) on p. But this also
says that there is no point common to p and any line on P. so
that P satisfies the definition of a pole of p. This information, along For the geometry of Fano.
with Axioms 2 and 3, leads to the following theorems.
q" + I - ~ = 1 -' - ~ = 7.
q - I I
THBOREM l.11. Every line of the geometry of Desargues has
exactly one pole. If lJ = 3. then PG(2.3) is a new finite geometry that is self-d~Hl1. fr~111
Theorem 1.13, the total number of points is 13. PGt2.3) IS a tillite
THt20REM 1.12. Every point of the geometry of Desargucs has geometry of 13 points and 13 lines. This geometry has the same. axioms
exactly one polar. as Fano's geometry, except that there are four points rather than three
on every line. .
Parallel lines exist in the geometry of Ocsargues, but their A dill"erent finite geometry, no longer self-d\1aI, can be obtallled
properties are dilTerent frol11 ordinary Euclidean parallels. For example, ~rom Fano's geometry by a modification of the last a~iom. Th: ne~v
note in Figure 1.10 that three ditTerent lines can be' drawn parallel to geometry, called r OlIllY'S geomefry. has the first four axIOms of hlilo s
line R. C. B through point A' but thut only one line can be drawn geometry along with the following substitute ttxioll1 for Axiom S.
parallel to line A, B, T.
It is worthwhile to study the axioms and the drawings for a 5. There is exactly one line on a point 1:1I1d not 011 <Illy point on a
geometry to make additional conjectures. Proofs of additional theorems line not containing the point.
for the finite geometry of Desargues appear as exerCises.
Several other finite geometries are discussed briefly in the rest Axiom 5 shows that the ordinary Euclidean concept of parallel
of this section so that the interested reader can develop some of their lines applies, since it means that exactly one parall~1 to :' given Hn,e
theorems on his own. The starting place for each could be a set of can be found passing through each point not on the gtv.en tllle. Youngs
axioms, although that approach may not always be the most productive geometry is a finite geometry of nine points and twelve Im~s. .
when intuition is involved. The interested student can make up finite geometnes of hiS own,
• fl. Fano's finite geometry has the special properly of being selF-clugl. although some of them may be of limited significrtl1ce. Various examples
" In other words, the plane dual of each true statement is a true statement
for the geometry. Actually, Fano discussed a particular set 0'[ Ilnite
geometries, each of which was self-dual. The general symbol for
geometries of this special type is PG(Il, q). The letters PG stand for
projective geometry. to be discussed in Chapter 7. The letter 11 is t1;e
number of dimensions, and q is the positive integral power of a prilne
, number. The geometry has q + 1 points on each line. Thus, Fano's
, \'
,
geometry of the plane is PG(2,2), since there are three points 0\1 each
line. The theory of projective geometry can be used to establish the
following useful theorem, assumed here.
Ih) lei
(a)
THEOREM 1.13. The general formula for the total number of FIGURE 1.12
points in PG(n, q) is
30 CHAPTER 1
16, The geometry
. consists of how many points and 1"0 w many I'llles 1
of simple drawings that can be used for a finite geometry are shown in
17. Each hne has how many other lilles parallel to it?
Figure 1.12. A few of the interesting properties are mentioned in the
For the geometry of Figure Lt2b:
exercises that follow.
18. Each Hne has how many points on it?
19. The geometry consists of how many points and how many lines?
EXERCISES 1.6 20. Each line bas how many other lines parallel to it?
For the geometry of Figure 1.l2c:

Exercises 1-8 concern the finite geometry of De:sargues.
1. Prepare a table to represent the geometry of Desargues, using the points as 21. The geometry consists of how many points and how many Jines?
named in Figure 1.10 and letting each colmnn of the table represent a line 22. Each line has how many other lines parallel to it?
in the geometry.
2. In Figure 1.10, name the pole of: ..-.
~ b. AS
a. RB'
3. In Figure L io, name the polar of:
b. point S t' (31
a. Point R r, \
(J
4. Using fjgure 1.10, name the pair of triangles in the Euclidean figure per-
..-.
spective from point T and from PC'. t~f- (~\. ~ As f\ '
5. Which axioms in -the geometry of Desargues are also true statements in
Euclidean geometry? ..-.
6. In Figure 1.10, name all the tines parallel to AS.
7, Prove that if point P is on the polar of point Q. then point! Q is on the
polar of point p,
8. Prove lhat two tines parallel to the same line are not parallel to each
other.
9. Prepare a table for the finite geometry PG(2.3); using the numbers 1-13
for the points and showing th,e points on a line as one column of the table.
10. For PG{2.3). there are how many lines on each point?
11. Which of these s.ymbols represent self~dllal geometries'l
b. PG(2, 5)
a. PG(2,4)
d. PG(2,7)
c. PG(2,6)
"i. 12. Find the number of points and lines, where possible, for the geometries of
Exercise 11.
13. Prove that Young's geometry includes at least nine points,
14. For Young's geometry. prove that two lines parallel to a third line are
I'
parallel to each other.
For the geomelry of Figure U2a:

15. Each line has how many points on it?
'. .1, CHAPTER 2
! {:
GEOMETRIC TRANSFORMATIONS
"t
"" .
. :/
..I" ' .•
2.1 INTRODUCTION TO
TRANSFORMATIONS
In high school geometry, it is customary to speak of rotation

and translation. Sorne triangles are proved congruent, and others are
proved similar. In other types of geometry to be studied, one talks of
the ideas of inversion and projection. All of these ideas are related lo
a very basic concept in geometry, that of a geollletric tramd"ormatioll.
In this - chapter, geometric transformations are defined and
iIlustraled. The major examples are the set of transformations of Euclidean
geometry. In addition. the study of transformations of similarity and
33
GEOMETRIC TRANSFORMATIONS 35
34 CHAPTER 2
. 'ased understanding of
finite sets of transformations contributes to mere
the concept. , d fi '( n of transformation. it is necessary
Before giving a precise e 111 10
to explain the idea of mapping.
FIGURE 2.2
A a set B is a pairing
t
DEFINITION. A mapping of a set an a f A ' p' ired with
d B 0 that each element 0 tS.l RecalJ that a function is a set of ordered pairs with no two
of elements of A an s h element of B is paired with at least different pairs having the same first element. It can be seen that the
exactly olle element of B. and eac ,
definition of mapping is equivalent to a typi'cal definition of function,
one element of A.
since in a mapping. having two pairs with the same first element
f ing whose ordered would mean that an element of A is paired with more than one
Figure 2.1 shows an example 0 a ~app t B element of B. It is customary to use mapping rather than function
. d ( b) A mappmg of set A onto se
pairs are (a,.b,). (G,.b,). an a3' ': b _ /(a) Here an . element b in geometry. however, when the sets being considered are sets of points.
may also b~ indicated by the notation u~der ~he m~ppi~g f. For a One of the essential differences between a mapping and a transformation
of B is the Image of an element a of A , d that each element of B is that reversing the elements in the pairs of a mapping does not
'{ . t B it is not reqUIre
mapping of set A lOCO s . e . . d be ma )ed onto a necessarily result in a mapping., while reversing'-'the elements in the pairs
be involved in the pairing. In this case A co;\ has a :~owledge of
h
proper subset of B. It is assumed that t e rea e
of a transformation also results in a transformation. Make sure you
understand why this is so.
;i:~~!i~;:': elementary concepts of set theory. Here are other examples of transformations to help make the
;i;r:::}:" concept clear.
a. A pairing of points on a number line indicated by x -i- 2x + 3.
~
For example, points with coordinates 1 and .5 or Z- and 7
would be paired. The arrow notation simply designates that
the second element is the image of the first under the
A B transformation. A summary of notation used in this text is
FIGURE 2.1
found in Appendix 4.
. IS
. 0 f par t'lCular importance in y
, I k' d of onto mappmg .
A speClft 111 • tails/ormatIOn
mathematics; this is a one-to,,:one onto mapplOg, or a r ...'
'ii)l( .'
;
ping / of A onto B
1 (x. v)
D~FINlll0N. A transformatIOn IS a map
such
that each element of ~ is the image of exactly one element of A~B
A. (x+2,y-l)
deuce exists between the sets of --~------------~X

In 'other words, a one-la-one ~orresPOl~ f a transformation is shown
elements of A and B. A simp e examp e 0 FIGURE 2.3
in Figure 2.2.
".,-
36 CHAPTER 2 GEOMETRIC TRANSFORMATIONS 37
b. (.\". J') -+ (x + 2. J' - I) is an example of a translation, a type r (x. y) - (x + I. y + I)

of transformation that will be studied in more detail in a later
section. As shown in Figure 2.3, each point in the plane is
9 (x.yl - (x + 2. Y - 2)
paired with a point two units to the right and one lInit In this example. note that the order~d pair has elements thal <He
below the original point. points expressed using an x and J' coordinate in the Cartesian co-
The concept of transformation is important because of the fact ordinate system of analytic geometry.
that sets of transformations can be used to classify geometries. In high The results of f followed by y, (Un for the sample points (2,2)
school geometry. for example, lhe transformations allowed are rotation, and {I, 5) are shown in Figure 2.5. The combined effect of at is to cause
translation, and sometimes sitnilarity. Allowing other, more general types (x,J') to have the image (.\" + 3. J' - l).
of transformations results in other modern geometries, In a particular
geometry, the student studies properties of figures and their images under y
a set oftransformations.ltwwlclIlf propercies are those that do not chltnge,
In the stud); of geometry through transformations, the student is asked to
(1.~
notice close relationships between modern algebra and modern geometry.
Readers who have not yet studied abstract algebra will find necessary
concepts ex-plained in this text. (4.4)
'-'-;--'"
Iffis a transformation from A onto Band 9 is a transformation
(2.~
'., .t.
-:i.·~;:~ .
from B onto C. the product IT = gf is defined as the transformation
from A onto C such that hlP) = g[J(P)] for each point of A. NOle (5,1)
that the product fJi is defined in such a way that the transformation
x
on the right is performed first. 0
In Figure 2.4, the pairs in It are (a t. e3 ). (° 2 , ('2)' (03' (
1
),.
FIGURE 2.5
In special cases, the product of two transformations is the

idel1cicJ' tr(lllsjiJJ'1rwriol1. t. lat. lit). (° 2.(/2) ..... (If,,.lt,,). For the identity
g transformation, each element is its own image, The identity trans-
''-'.~.
formation leaves each point fixed. In this case, if I = g,/: as illustrated
a,
b,---::-"
r
'- ---.:::.......---
a ,......."- _
J
\
,......... ....... ~--
------ - ----.::;---~'----
h
a, a,
FIGURE 2.4
As an example of finding the product of two transformations. "

.' j consider the two defined as follows for all real points (x.y) . FIGURE 2.6
38 CHAPTER 2
8. Find the product/yo
. . .s the inverse tral1~fi}rmafi{}n of J. .U1 d'lcated by
In_ tFlgure 2.6, _t~le~ 9 .1_1 = 1. Recall that the inverse of a transfonna- 9. Find the product fir.
.r . so that f f - f"
.fl .' I t 'ans'ormation , so that the 10. Findf- I.
doing" the ongllla I "
tion has the effect 0 u~ d" 'se is the identity transformation. 11. Findg- ' .
. :tatlOn an Its mvel .
product 0 f n trans f01 n . h that (x V) has the Image
'f f . transformatIOn sue ( ".
For example, 1 IS a . f. tion such that (x,y) has the
(x + 4, y - 2), then.r - I IS a trans OHna
image (s - 4. Y + 2). 2.2 GROUPS OF TRANSFOHMATIONS
r :,
fn modern geometry, sets of transformations are ordinarily
EXERCISES 2.1 discussed rather than individual transformations.
. I 0 sets one with six elements and one . h el'gil" •
Wit
1 Draw a diagram sllOwmg w • r ,'on
. . that is not a trans onna I . DEFINITION. A group of tl'G/l.~r(Jrmat;oJls is a nonempty set S
then indicate a mappmg . ?
':;
'. 2. Which of the mappings shown III
. f"
Igu
re 2 7 are transformations.
. of transformations f such that:
a. fE S implies tllatf-' E S.
b. fE Sand gE Simply tllatfYE S,
c. f(g") = (/g)".
(b) The first of the properties above guarantees that the inverse for
(a)
each transformation is also an element in the set of transformations.
~ The second of the properties is the closure property. This means the
~
product of any two transformations in the set is also a transformatipn-
in the set. Stated another way, the operation of multiplication can
(e)
always be performed within a group of transformations without going
FIGURE 2.7 outside the group. The last of the properties is the associative property.
It is assumed here that, regardless of the grouping by parentheses, the
3.
. • .•
For the transformatIon mdlcated by x.y
( ) _ (x
.
+ 5.y - 3). give the image
definition of multiplication of transformations results in one transforma-
of these points: tion followed by another, and .f(g") and (fg)" imply the product of
a. (2.1) b. (0,0) e. (-3, -2) the same three transformations in the same order. for example, consider
For Exercises 4-7. let f f) (c,d). (e,h)} and g
= { (a, 1.
= {(h,i). (d,j). (h.klJ be the three transformations which have been defined as follows:
transformations.
4. Find the product gJ. J: x ..... x + 3.
5. Find the produet/q. g: x ...... x - 2,
6.. FindF'·
h:x - 2x,
7. findg-: J •
. r be the transformation such that (x.y) has :the image /(gll): x ~ (2.>; - 2) + 3 = 2x + I,
For ExerCises 8-11. let. . eh that (x y) has the image
(x -,5,), + 2) and g be the transronnatIon su . (/g)h:x ~ 2x + L
.
"": (x + 2,.1' - 3).
.,'
i,:
40 CHAPTER 2
A fourth<propeUy that needs to be established for every group R, Rel1ection about the axis through verlex I.
of transformations is the existence of the identity element as an element R, Reflection about' the axis through vertex 2.
R, Reflection about the axis through vertex 3. .
of the group. This is left as an exercise. In general. the commutative Rotation through an angle of 120<' counterclockw~se.
property fg = gf does not hold for transformations. For example. the R( 120)
Rotation through an angle of 240'" counterclock \VIse.
finite group to be introduced shortly is not a commutative group. If R(240)
the commutative property does hold for a group. that group is called a I Identity,
commutative or Abelian group (after the Norwegian mathematician N. H. To verify that the set of symmetries of an equilat~ral .triangle
Abel, 1802-1829), . .' ssary to verify that f E.S Hnpites that
constitutes a group, It IS nece ' . .
Many of the groups of transformations in geometry· are infinite f - 1 E S and that f E Sand 9 E S implies fg E S: The II1verse element fOl
groups-that is, groups with an infinite number of members. On the
each transformation is listed below.
other hand, examples of ~nite groups will help in developing the concept
of a group of transformations. It is important to remember that the f r'
elements of a group of transformations are transfonnations, not points. -'---
Consider the set of all symmetries of an equilateral triangle, as R, R,
shown in Figure 2.8. "Symmetries of the triangle" designates the reflections R, R2
R, R,
R(120) R(240)
3 R(240) R(120)
I I
TI verification of both facts can be accomplished by completing

a multipli::tion t~ble fot' the elements. The entries in. t1dl~ tabdie Fare
3 3 2 , [ , t' s in the order 111 lcale. or
R, R, found by perfortmng the trans onna IOn. .
(,' (b) (e) example, Table 2,1 show~. ~e[R(,12011:J'!' Recall that R(120) comes
3 2 1 first in this multiplication.
666
2 R,
(d)
3
R(20)
(e)
2 R(240)
(f)
3
1
..1"
1
R,
R,
TABLE 2.1
First Tnllls{ormarirm Peljormell
RJ
Rl ,
R,
R,
R(qO)
R(120)
I{l240)
R(240)
R(120) R(240) R, R,
\ FIGURE 2.8 R, 1
fl'
R(240) 1 R(120) ['~l~~'.~\ a,
a, , R' .--
about rhe axes of symmetry or rotation such that the new figure a(240) 1 R, a,
R, R, R(120)
coincides with the old. The result of each symmetry can be represented
by a renaming of the three vertices. Check each of the figures in Rll~O). a(120) a, (~:~.!!.l ' R, 1«240) 1
R, 1 n(t20)
Figure 2.8 to see that it corresponds to the following definitions of , a (240) R(240) R, R,
--
.,
the elements of the set of symmetries of an eqJ]ilateral triangle:
GEOMETRIC TRANSFORMA nONS 43
42 CHAPTER 2
The notation means that R 2 is first • TI' IS .In d'lcated by
115 .
Introduction of the permutation group symbols will ilIuslrate a
conunon agreement that ties the work on transformation groups more
closely to that found in modern algebra texts. Two roWS of numbers
are _used to define a transformation. The first row shows the original
G: ~: ~).
vertices and the second row shows the new position of those vertices.
Then the result on the bottom r
one of the equivalent rorms for R
,. JOe becomes the top line of
R, = I. c· 2.
3. D
Explanation for R t • Vertex 1 remains fixed,
vertex 2 moves to the place where vertex 3 was
originally. and vertex 3 moves 10 the place
G: ~: :).
The product has the top row frOi
~~. =~d asnth~2d:;:it::~ :;~~I;O;'oW
where vertex 2 was originally.
form of can be identified from the
R2 =. c·
3.
2.
2. ~) m tng the product of the tw
illustrated as in Figure 2,9.
.
0 transformations can also be
~)
2.
R,
=C' 2. 1.
= G: ~)
2. Explanation for R(120). Vertex 1. moves t.o the
R(120) 3. place where vertex 2 was originally. vertex 2
moves to the place where 3 was originally. and
1 was
vertex 3 moves to the place where
originally.
C· ~)
6=6
2.
R(240) = 3. I.
There are five other equivalent forms for each of the permuta-
tion group symbols. This is true because the elements on ·the first' 1 2 3 1
row can be arranged in any of six ways. For example. two other
equivalent forms for R 1 are FIGURE 2.9
3. 3. 1,
and ( 2, . Note the important convention b · '
( 2. 1,
FIgure 2.9. The second trans'o t' . 0 Sel ved 10 the first row of
• • {I rma 1011 IS perform d ' . .
Either the permutation symbols. or a series of pictures, can posItIons for the vertices TI h e usmg the ongmal
. , • lUS, t e vertex that I
be used to check the entries in the multiplication table. Three -examples was ongmally (1' is d - was w lere vertex 3
• h move to where vert 2 . .
vertex that was where t 2 ' . . ex . was onginally. The
follow to illustrate both methods. ver ex was ongmall (2)'
vertex 3 was originall d I y. ,IS moved to where
. . y, an t le vertex that - h
onglnally. (3), remains unch d B was were vertex I was
EXAMPLE I . ange. ecause of th'
mterpretation, the reader will robabl .IS somewhat tedious
~: ~) G: ~: 3)I
symbols a valuable shortcut. P y find the. use of permutation
DC: =
= R(l20)
.
'I.. .
];h.:.
44 CHAPTER 2
GEOMETRIC TRANSFQRM,6,TIONS 45
EXAMPLE 2
vertical 'order, the multiplication table is not symmetric ..tbOllt the
R(120)R,-= (3.l. 2.
J. 2
1)(1.3. 2.
2. ~) = C: ~: ~) R,
diagonal frol11 upper left to lower right.
Withi.n a group of transformations may be subgroups of trans-
fOrm.,ltions. A sul1gn!y.ILis a subset of ,\ group thal is itself H group.
For examp'le, 'o'~~'-such subgroup of the symmetries of an equilateral
EXAMPLE 3
triangle has element~J. R{ 120). and R(240). Verify frorn the multiplication
R(240)R( ;20) = (2.I. 3.

2.
l)(l.
3 2.
2.
3.
3)1 (I.l.
=
2.
2. D = I
table that the products are elements of a subgroup. Furthermore, observe
that this subgroup is 11 commutative group.
Other exarnples of finite groups of transformations nre the
symmetries of a square and the symmetries of an isosceles triangle.
Examples 2 and 3 are Wllstrated in Figure 2, to.
These are explored in the following set of exercises.
I'
,·6~6=6
EXERCISES 2.2
L Prove tlmt fI group of Inmsformntiolls includes the identity element.
1 23 2-1 3 2. Write an explanation for the permutation group symbols for R 2 • R 3 • nnd
R, R(2401·
For Exercises 3-6, nnd the products from the multiplicUlion table for the sym-
.A
(b~~~-L~
R(120) A R(240) A ~
metries of an equilateral triangle, then use the pernnHnti01l notulil;n to verify each
answer.
3. R(240)R, ";:) 4. R j R(240)
::::) 5. R j RJ =)6. R)Hl
1 2 3 1 1 2 ;::-) 7. 1s the set II, R j , R1 .R 3 1 of symmetries for the equilateral tri<l1lgle <l sub-
group'!
FIGURE 2 10 :::"\ H: Prepure a multiplication tt\ble for the symmetries '01' nil h;osceles triangle.
:::) 9. Verify the facl that the symmetries of an isosceles triangle form.t group.
The conclusion from the discussion of the symmetries of the ~.\ 10. List all the subgroups of the symmetries of all isosceles triangle."? ,
equilateral triangle is stated as a theorem. ~.01 L Prepare a multiplication table for the symmetries of n squme.· 1
12. Verify the fact thaI the symmetries of a square form a group.
. THEOREM 2. L The set of symmetries of the equilateral triangle ! 3. List ali the subgroups of the S)lnllnetries of t\ sqlHU'e. -\
IS a group of tra.nsformations.
The group of symmetries of an equilateral triangle is not a 2.3 EUCLIDEAN MOTIONS OF A PLANE
commutative group. As a specific example, R 1 R(240) = R 3 • but
The examples of tran~formations that are most ran~iJiar are those
~(240)R 1 = R/ Note that a consequence of this hi'ck of commutativity
used in high school geometry. Unlike the finite groups of transformations
Is that, when the elements are listed in the same horizontal and
studied in the previous sectio~, the sets of transformations in this section
t·
i,
46 CHAPTER 2
.' er of members. The detailed study in this secti?n In 1872, Felix Klein classified geometries by applying this
have an mfimte numb . E lidean geometry from a pomt definition: A geometry is the study of invariant properties of a set of
. Ives lookmg at uc
and the next two mvo . dar school approach. points under a group of transformations. Thus, Euclidean geometry is
of view different from the typlc~l s~c::fth~ transformations ofEudidean the study of invariant properties, such as angle measure and area, of
The essential chara~tenzatto d Th t is distance must be an sets of points under the group of Euclidean transformations,
·
geometry is that d.fstanCe IS preserve. a •
The .reader should ask himself) before continuing, what are the
invariant property. different kinds of Euclidean motions. The same question can be stated
differently by asking what types of transformations can be applied to a set
.
DEFINITION. A transformation I IS an -;c-;--
isometry of A onto B
of points in the plane so the distance between any two points is
,
if it preserves dIstances. F or an Y two points P,. P 2 of A. . always preserved. This question is answered in this section, but the
answer for the same question applied to analytic geometry is reserved for
I/(P ,)/(p,)1 = IP, P ,I, the next section. l' 0\
The first type of plane motion is a ~.tfa;ls7aiiolt:J Intuitively, a

I . IP t
Wlele P2 I denotes the distance between PI and P2' ~- .. --.....- .. ~- ..
translation sets up a correspondence between points and their image
I(P,) points so that each image is the same distance in the same direction
from the origimiJ point. Figure 2,12a shows a translation of segment An
in a direction represented by the vector AA'. Figure 2.12b shows a
P'~ translation of triangle ABC in a direction represented by the vector BB'.
P,
\
I(P,)
Check each of these observations about a translation.
B'
FIGURE 2.11
.' . . Fi ure 2.11. The symbol IP, P 21 means

This idea IS 111ustl ate~ In g
f the distance so t lat on
I Iy positive (undirected)
d
A,~~>'>B
......... -..: .. IE" ..............
the absolute va Iue 0 '.. distance is an undefine '"<-....;. ........
'defed here Sometimes, d B B' . C
distances need be conSI '. I tic geometry the distance A
term in synthetic geometry. but l,n ana Yd b (al (b)
.
between pomts ) d (x
(XI'}'1 an . l · J,,)
].
IS define
.
Y
FIGURE 2.12
d = )(-'2 - x,j'.+ (y, y,)2,

a. A segment is translated into a parallel segment.
.' en the new concept of isometry and b. All of the vectors connecting corresponding points are equal.
The close reiatlOnslup betwe . I I dded statement thal, if an c. The inverse of a translation is anotJie~ tran~Jatjon the same
T t is apparent WIt 1 t le a
a more fam1 Jar eoncep r . t the sets are j}E!!lItUi.c,..-or distance in the opposite direction.
isometry exists between two sets 0 pom 5,
d. As illustrated in Figure 2.13, the product of two trans.lations
col1g~t. , t the isometry that is studied is an is a translation. The vector for the product is the sum of the
...,......."" In ordinary Euchdean geo~e rY lft I other words the image of vecto,rs fo~ the two translatibns.
of points onto ltse. n • .
isometry 0 f a set . . the same plane. Isometnes e, The set of all translations forms a group.
, f I lane is another pomt III d
each pomt 0 t ,e.p ' E I"d geometry involves the stu Y Additional information about transJations can be found in the
of this kind are called motIOns. uc 1 ean
., next sections, where the discussion includes ("he use of coordinates.
of motions of the real plane.
fl',i
....
,J,[ 48 CHAPTER 2
'",'
( ('21
The second basic type of motion of the plane is a .~_~ The set of all translations and rotations is called the set of
The symbol R(O,o:) indicates a rotation through an angle of ex about rigid motions or displacements. Using a cardboard model of a triangular
the point 0, as in Figure 2.14. By cOl1vention, a cQuntercio(:kwise region makes it possible to illustrate rigid motions by sliding the model
rotation is associated with a positive angle. from one position to another in a rigid way without changing its size
or shape. . . .,._.. .{:t..'
The third basic example of motion of a plane is a0'eflectjo~9 A
reflection R, utilizes a fixed line 1. as in Figure 2.16. A poinl.l.nCns it~
P'
own image. Any other point P" is mapped into a point P' sucb that 1
«
), is the perpendicular bisec.tor of P P'. (The notation Pr means the
segment with P and P' as endpoints.) As the name reflection implies,
a sel of points and thei·r "images are reflections of each other; it is as if
line I were a mirror. Note that a set of points having 1 as a line of
a symmetry is mapped onto itself by a reflection about I. One half of the
figure is the image of the other half.
FIGURE 2.14
Figure 215 shows rotation of a segment and a triangle about a P C'

point. In both cases, the angle of rotation is L AOA'. Check these
observations about the motion of rotation.
P'
,'p~
J.
, FIGURE 2.16 FIGURE 2.17

'.~
V/~B Check these additional observations about reflections.

a. In general. a segment is not parallel to its image.
a
b. The inverse of a reflection is the same reflection.
(b)
c. The product of two reflections about a line is not a reflection.
FIGURE 2.15
d. As illustrated in Figure 2.t7. a reflection cannot be con-
sidered a sliding in the plane. It is necessary to H·ip over the
a. A segment is not necessarily parallel to its image. cardboard model ofthe triangular region to have it correspond
b. The inverse of a fptation is a rotation about the same point to the image.
and with the sani.e angle of rotation, but measured in the The three transformations given so far, translation. rotation, and
opposite sense. reflection, may be considered the three basic motions of the plane. But
c. The product of two rotations about the same point is another this set is not closed, and in fact one more transformation is necessary
rotation about that point. for closure. What this means is that some products of two of these
d. The set of all rotations about a fixed point is a group of transformations are neither a translation, a rotation, nor a reflection, but
transformations. yet another type of transformation. Specifically. the product of n reflection
b
'.
50 CHAPTER 2
deta!1. Section 2.4 develops the material us' .
and one of the other two transformations may not be an element of SectIOn 2.5 discusses the set of II . mg analytIc geometry, and
the set. The fourth type of motion of a plane is a:~;de . re~~~:ti9 general point of view. a motIons of the plane from a more
A glide reflection is the product of a reflection followed by a tl:ansiation (

parallel to the fixed line of reflection, as in Figure 2. 18, in which P is EXERCISES 2. 3
mapped into r'. I. For a translation is the measure
invariant?' of the angle between two rays an
jOp ~ 2. Could a translation be its own inverse? r-IC)

/
J. Explain what is mea n t by. t he Identity
. translation
4. Explain what is meant by the iden ,", .
I y rotation .
,, 5. Explain how a segment and its irna e '
than the identity_
.
g might be parallel for a rotation other
-.r
6. Desc~ibe the inverse of a rotation us' .. -
negative one_ 109 a posItive direction instead of a
-:g 7_ Give an . example in

re fl ectlon. sri LlO' r'-
Wh"ICh a segment a d ·t
n I
.
s Image are parallel under
FIGURE 2. 18
Describe the inverse of a gl,"de re fectton

l· .)
The exercises include questions that will require experimentation
Are a segmen t and'Its Image
. . .?
parallel under a glide reflecti ?' I/)
with drawings or cardboard models to deyelop additional information
about the glide reflection.
One geometric application of various motions of the· plane is in
Can a model r:
. lor a triangular region be
under a glide reflection by sli~_i!y~? "to on (,--,'.
. .
ma~e to cOincide with its image
simple geometric designs, such as those shown in Figure 2.19. that
fOl-m plane-filling repeated patterns. The Dutch artist M. C. iEsche~ has 2..4 SETS OF EOUATIONS FOR·
extended this idea to very intricate plane-filling patterns involving ·such MOTIONS OF THE PLANE
figures as fish, birds, horsemen, and reptiles. The interested reader will
enjoy studying these works of art in The Graphic Wor/{ of M. C. Escher. In this section, analytic geometr is ..
study
of motions of the I y . utIlIzed to approach the
pane as sets of linear equations • If A .IS any
X(x •x ) _ _ ........... TA{X) .

1 2
,-.,...._-- (XI + a l ,x 2 + a2 )
(bJ
(ot. e:::::=-___
(0) ,A (a,.8 2 )
FIGURE 2 . 19
x
The sOl1lewhat intuitive introduction to plane motions in this

FIGURE 2. 20
section is followed by twO sections developing the concepts in greater
point of the piane, the transformation I:, stich that T... (X) = X + A is a for a and b nny real numbers. The inverse of this transformation has
translation in the direction of 0 A. In Figure 2.20, the point TA{X) is the equations
l
I.r
:r:
i{
: i.
.',-
,It'
the image of! point X under the translation. Its coordinates are the
sum of the coordinates of points X and A. (Xl + "1' Xl + G2 ). Thus,
the notation X + A is used to indicate the addition of the corresponding
coordinates of points X and A.
x =
y ~
Xl -
y' - b,
a,
The concept of sets of equations for a translation should be

familiar from elementary analytic geometry. In Figure 2.21, for example, EXAMPLE. Find the image of (5, 2) under the translation with
equations x' = x + 7, y' = y + 3. The solution is (12,5).
y
P'(3,3)
In typical analytic geomettty textbooks, it is shown that the
de-fining equations for a rotation about the origin are as foHows.
LJ
P(1,2)
DEFINITION. A rotation about the
with equations of the form
origi~ is a transformation
"
;)
A
x' = xcosa - ysina.
y' = xsina + ycoso:.
FIGURE 2.21 EXAMPLE, Find the image of P(2, 3) under a rotation of 60·
~. . about the origin.
if P( 1. 2) has the image P'{3, 3). then the transl~tion may be represented The equations arc
by the equations
= xcos60° - ysin60°,
x' = x -I- 2, ,, X'
y' = x Sill 60° + ycos 60°,

y'~y+l.
Then
Also, using the nOlation of addition of coordinates of points,
(1,2) + (2, I) ~(3,3), hence A ~ (2, I~ Note of course that (2, I) also x' ~ (2)G) -(3)(f).
indicates the veclor PP: in the diagram.
DEFINITION. A translation is a transformation with equations

y' ~ (2{f) + (3)G)'
of the form.
The coordinates or P' are
x'=x+a, ,
.\
y' = y + b.
;;
;
,:
" GEOMETRIC TRANSFORMATIONS 55
54 CHAPTER 2
, EXAMPLE. Find the image oC (2
The equations of rotation about a point (11, k) are (me y = x. Since cos 20 = 0 and sin 20 =' ~~ under a reflection abollt the
x' _ II = (x - h)cosoc - (y - ")sinoc.

x' = y.
y' _ " = (x - h)sinoc + (y - I<)cos a.
y' = x,
Note how this transrormation may be considered the product of a hence,
;.. rotation and a translation. X' = 3,
The sets of equations for the reflection transformation are not as
familiar as those for the transformations of translation and rotation. y' = 2,
For the special case of re~!iQ.I.:t~J:!Out the x-axis, the equations arc and the point is (3 ) 2)• as sho~ld be expected.
X' = x,
y' = -yo can b The. general equatIOns_for
. reflection b
a out a ~!}.~.1.....= IIlX + b
.
e obtalned frorn the speci I
the or"
'
b a case of reflect'
Igm y a translation through d'
b --~
IOn a out a line through
The forms are also simple when the reflection is,about the y-axis.
A reflection about the line }' ~ mX, where III = tan () as shown d'Istance d from _ a tstance
Y - niX to l' = niX + b TI twice the e'
, P Ipendlcular.
d
in Figure 2.22, is a reflection about, the x-axis, combined with a lstance is explained in the dis " . le reason for twice the
CUSsion of TheOl"e m 2.5 .Ill' ' the next
y y
y=mx
P' y=mx
A
o
,P"
y = mx + b
FIGURE 2,22
rotation through an angle of 20 about the origin. This can be seen FIGURE 2,23
[rom Figure 2,22 because o[ the congruences L POH =
L p'OH and
LPOA = LP"OA. The equations [or this reflection are section, although the,,--,answer
S. ' can I be determined here I'C d '
lnee the transJation is in the dl' a.so m . d' . eSired
x' = xcos28 + ysin20, the e . recUon .I. '.,' n F'Igure 2.23,'
leated by 'I'
quatlOfls of the tr~Jlslation are
y' = xsin20 - ycos28,
x' = x + 2dcosq"
obtained by writing the equations for a rotation of 28 about the origin,
y' = Y + 2dsil1q"
then substituting - y for y.
and the desired equations for the general reflection are THEOREM 2.3. If a transformation has equations of the form
x' ~ :<cos28 + ysin2G + 2dcosq" x' = ax + by + c,

y' ~ xsin28 - ycos20 + 2dsinq,. y' ~ ±(-bx + ay) + d,
EXAMPLE. Find the image of point (2, 6) und~r a reflection

for a. b. c, dE Rand a 2 + b2 = 1. then it is a plane motion.
about the line such that 8 ~ 30", q, ~ 300", and d ~ 2.
X' = + y sin 60° +

x cos 60° 4 cos 3000
The equations for the inverse of a transformation can be found
by solving the set of equations for x and y. This was illustrated
y' = x sin 60° - Ycos 60° + 4 sin 3000 previously for a translatiQn. The following example is for a reflection.
x' ~ (1/2)x + (fiI2)y + 2
y' ~ (.fi/2)x - (lf2)y - 2.fi EXAMPLE. Find the inverse of the transformation with
For x ~ 2, Y ~ 6, equations
x' ~ 1 + 3fi + 2·~ 3 + 3.f{ x' = x cos 30° + Ysin 30°,

y' ~ .fi - 3 - zj3 ~ -3 - fi. y' = x-sin 30° - Y cos 30°.
The general equations for a glide reflection are derived from Multiply the members of the first equation by sin 30° and the
those for a reflection and those for a translation in a direction parallel members of the second by cos 30°.
to the line of reflection. The images can be found using t11e two sets of
equations given previously for these transformations. x' sin 30° = x sin 30° cos 30° +
y sin 2 30°
A study of the form of the sets of equations for the four motions y' cos 30° = x sin 30° cos 30° - Y cos 30
2 e
of the plane shows they may all be described in the same general way.
0
x' sin 30" - y' cos 30 = y
THEOREM 22. If a transformation is a plane motion, then it
has equations of the form Now mUltiply both members of the original first equation by
cos 30° and the members of the second by sin 30°.
x' = ax + by + c,
x' cos 30 e = X cos! 30e + y sin 30° cos 30 e
y' ~ ±(-bx + ay) + d,
y'sin30e = xsin 2 30° - ),sin30ecos30e
for a, b. c, dE Rand a 1 + b 1 = 1.
x' cos 30" + y' si~ 30 0
~ x
The converse of this theorem is also true, although the px:oof is

not given here. The converse of a theorem stated as an implication is As should be expected, a change of variables will make obvious
obtained by interchanging the if and then parts of the statement. The the fact that the inverse of a reflection is the original transformation
converse is: itself.
f!
EXERCISES 2.4 The proof of Theorem 2.4 consists of showing that there exists an
inverse for each element and that the product of any two elements is
Let the vector for a t:"nslntion be (3, - 5) and give
Ih e image for each another element of the set.
f 1. point under the translation. It is easy enough to verify that each element of the four types
j;
I: a. (0,0) b. (5,7) of motions named has an inverse element. For example, the inverse of a
c. (-1,-8) d. (-2, -1) translation is another translation in the opposite direction. The check [or
(5,9) what points have these points as closure with respect to mUltiplication is somewhat more complex. A proof
2. In the translation with vector
images'? using analytic geometry could be given, using The:orems 2.2 and 2.3 of the
b. (5,9) previous section. The synthetic proof is normally faciJitated- by showing
a. (0,0) of 4'5" aboul the
What is the image of the Point (3,4) under a rotation that motions of the plane may be discussed as pr?ducts of reflections.
- ) 3. 'g,·ol f 450 about the
on . (-I, 2) under a rotation
::;94 What point has the Image 0
THEOREM 2.5. Every plane motion is the producl of three or
fewer retlections and conversely.
origin'1 Cor the inverse 0 f a transformation of rotation about a
::7 point (Ide).
.
5. Derive the equations
.
.
b ut the line y = (J313~'(.

1) der a reflectIon a 0 . The theorem is proved in two parts. The first part is for trans-
Find the image of (3, un . b t the line with 0 = 60", lations and rotations, while the second part is for refiections and gJide
.
Find the Ullage o r (3. 7) under .a reflectIOn a ou reflections'.
0
A. - 330 and (I = 5: . consisting 0. f a I. Every translation and rotation is the product :of two reflections and
'/' - . 7) dec the glide reflection .
image of (~, lUI 0 d by a translation of three uOlts 1n a conversely.
reflection a b ou t the y~axis fo owe . ..,
8. What is the
positive direction parallel to the .v-axIs. . whose The two cases depend on whether the two fines of reflection
'.". intersect or are para-Uel.
.
For ExerCises 9- 12, find the inverse transformation for the transfonnatJon
a. If I, and 12 are any two fines of the plane intersecting at 0,
equations are given:
X' := x cos 30" - ysin30"
and if the angle from I, to I, is 0, then R(O, 20) ~ R , R ,.
9. x' = x - 5
~'=.:~x 3~''
I I
to. sin 30" + }' cos In Figure 2.24, Ihe rotation that transforms l'Q into P" Q" can
, y'=y+2
~' , = x cos 60" + ysin60" be considered the product of two reflections about I, and 1 "
11. x'=x 12. oX 0" 2
y' ~ -y
y' = x sin 60" - ycos6 b. If I, and I, are parallel lines, and if they are perpendicular
to a transJation vector OP and at a distance 10plf2 apart,
then t = Ril RS1 "
2.5 PROPERTIES OF THE EUCLIDEAN
I,
GROUP
,I. f a lane introduced in Sections
The set of Euclidean motIOns 0 p. ftections and glide
J
. f t anslations rotations, re '. d

2.3 and 2.4, co.nsIsts, 0 ~d" al p;operties of motions are consldere .
reHections. In thiS sectIOn. a Won
....,~::'---t---t----_ I,
. . . of the plane
THEOR EM 2.. 4 The four. basic Euclidean motions
.
constitute a grou P of tninsformatlons. FIGURE 2.24
60 CHAPTER 2
Itt Figure 2.25, the motion that transforms points A and B inlo Tn Section 23, experimentation led to the ract that some
A" and B" may be considered the product of two reRections about It motions could be- represented by sliding a cardboard model in the piane,
and 12 . while others could not. This matter is now reconsidered in the light of
Theorem 2.5.
DEFINITION. A plane motion is a fliJ'(!c( motion ir it is lhe

8 8·
product of an even number of reflections. It is an opposite motion if it is
the product of an odd number of reOections.
A A· A··
According to this definition, rotations and translations are direct
IOPI motions, whereas reflections and glide reflections are opposite tno,lions.
-2-
Intuitively. the difference between direct and opposite ulOtions can be seen
OP
by considering a triangular region cut out of cardboard, such ~s 6ABC
in Figure 2.26. This piece or cardboard can be moved in the plane
FIGURE 2.25
to represent any direct motion. For exarnpte, it can be translated to
the position of .6.A'B'C'. On the other hand. the paper must be turned
2. Every reflection or glide reflection is the product of three or fewer
over ir th~ movement is to represent an opposite motion. It is necess'HY.
reflections. This statement js trivial for a reflection itself. Since a glide for example, to turn the cardboard triangular region over for it to
reflection is the product of a reflection and a translation. and since a coincide with ,0,A"B·'C"'.
...
translation is the product of two reflections, a glide reflection is the
product of three reflections.
A A' A··
The proof of the converse of Theorem 2.5 is not completed.
The converse states that the product of three or fewer reflections is a ~ ~8"
motion of the plane. The type of motion depends on the relative 8 C 8' C' C"
position of the lines of reflection. The possibilities are summarized in FIGURE 2,26
Table 2.2.
The general equations for motions include a ± symbol. The
TABLE 2.2 plus sign indicates a direct motion, al1d the negative si'go' indicates an
opposite mo~i(:)fl. This may be verilled by a check or each individual
PmdUCl Ilf A. If the two lines of rei1ection are paral!e~ (hen the motion is
Two Reflectiolls a translation. type of equation for a motioll.
B. Hthe two lines of reflection are nonparallel, then the motion The product of direct and opposite motions is summarized in
is a rotation.
Table 2.3.
Product of A. If two of the lilles of reflection coincide, then the motion
Tllree Reflectio/ls is a reflection. TABLE 2.3
B. If the three lines of reflection are purallel, then the motion , - - - - r----,----·- .
is a reflection. _ Din:cr OI'I'!I$ill!
C. 1f two lines of reflection intersect at a point on the third.
then the motioll is a reflection. . . Din't"l Direc! Opposite
D. If two lines of reilection intersect at a point not On the [-----1.-.--- .----.
third, then the Illotion is a glide reflection. OPPI1Sih' Opposite Direct
---- 1.. _.... __ , . ___ •
62 CHAPTER 2
GEOMETRIC TRANSFORMATIONS
63
As one with sufficient algebraic background might suspect, the "
system of direct and opposite motions under multiplication is isomoqJhic . For each type of motion, the . '"
pOints Without being the I·de t·t POSSIbilIties for numbers of fixed
to the system of even and odd numbers under addition (or the integers n 1 y transform f
form of Table 2.5. a IOn are summarized in the
modulo 2). This call be seen by· <;omparing Table 2.3 with the addition
tables derived for the other two systems mentioned.
The use of Theorem 2.~ ilnd the concepts of direct and opposite TABLE 2.5
motions finally makes it possible to complete a table showing closure for
Theorem 2.4. Since yach entry in Table 2.4 shows two possibilities. more MotioIJ
Pflssibi/itiesjrJI" III1'a/'illlll Pllil!(.~
information about the beginning motions is necessary in order to classify translation
none
the product as a unique tyP'~ of motion. rotation
one point
reflection
•
d
glide reflection olle line
nOne
TABLE 2.4
- _.. .. - "'---""--"_' .
'GUt/e.'- . At most, one line of the plane . " .
R(;ftI(i"" TruIIslatiall Reflectioll Refte(·tillll that IS not the identity. In the light ~a~ ~em~m lnvan~nt in a motion
,. to ask about the minimum numb ~ t .IS dIScovery. It is significant
Rotf/Oon rotation rotation glide glide to uniqueJy determine a pJa . e.r 0 pOints and their images needed
0' or reflection
, reflection hr. ne motIon other th I ·d .
, tnlllsiation translation 0' P' t ~ 10JJowing theorem wjJJ establish that h an t .le 1 entlty. Proof of
reflection refiecli(lO pOlOtS. t e answer IS three noncoHinear
Trllllsi!Ir/(1/i rotation I"otatioll glide glide
0' 0' reflection reflection
tnUlslatiou translation or 0' THEOREM 2.6. A unique t·
determined by an isometry of one tria~Ogf~on t the plane is completely
reflection reflection Of
Rejlectioll
on 0 a second.
glide glide rotatiotf rotation
, reflection refiection 0' or
or or translation trauslntion Let bABe", 6A'B'C'· F·
reflection reflection there is exactly One m~tion of tl In I Igure 2.~7. It is necessary to show
le pane agreeing with tf)is isometry.
Glide glide glide rotation fotati<lU
Refie(·ti(J/I reflection reflection or or
~~
.. or or: translation translation
reflection reflection
-
Recall that a geometry consists of the study of properties in-

variant under a group of transformations. The group of p{~ne motions. 8 A'
8'
for example, are the transformations of ordinary Euclidean geometry. FIGURE 2.27
Basic to the investigation of other invariant properties is the question df
invariant points. Sometimes a point and its image are identical; and this Let P be any point of th
the isolUetry is direct then P" e . pJane other than A. B. or C. If
result affects the study of any other invariant properties. The effect of a
L B'A' P , '" L HAP and• IS uniquely det . .
A'P' = AP (Th elllllned by constructing
motion on the points of a plane is to change many points and possibly
to leave some invariant. from A'to P', rather than the . e SY~bol A'P' means the distance
angle is constructed in the op . segment ltself.) If it is opposite the
pOslte sense. •
64 CHAPTER 2
Let PI and P 2 be two points of the p1ane in Figure 2.28. DAA'B' ~ 6B'BA. because pairs of corresponding sides are congruenl
6AP,P, ;;; 6A'P,'P,',since AP, = A'P,', AP, = A'P ',and LP AP Thus, LAB' A' ;;; L S' AB, and ;lti and A'S' are parallel.
_ LP 1'A'P'
= 2'
TIlUS, P1 P 2 = PI' P2', so that a motion" is determined,
because distance between two points is preserved. EXAMPLE. Prove analytically that a line and its image are
parallel under a translation.
Let the equation of the line be ax + by + c = 0 and the
equations of the translation be x' = x + cI, )" = y + e. The equations
of the inverse are x = x' - iI, Y = y' - e. so that the imuge of the line
is a(x' - el) + b(y' - e) + c = 0, or ax' + b.... + (- "d - be + c) = O.
lilis equation and the original equation have the same slope, so that the
8 C'
FIGURE 2.28
lines are parallel.
The location of the points is virtually the only property not
an invariant in a motion. The realization that the group of motions
Since, for any given point P, there is only one point of the
allows very few changes in properties leads to the need Io investigate
plane P' stich that PA = P'A', PB = P'B', and PC = p'e, P' is the
more g~neral types of transformations that do not leave as mHny
,mage of P, and the motion is the only one for which 6ABC has as
invariant properties. One. such group, the similarities. is introduced in the
tts nnage D. A'B' C'.
last section of this chapter, following the section on motions in three-
The properties studied in elementary geometry. such as con-
gru~nce of segments, triangles, and angles, area of regions, and intersection
space.
of hnes, are ~ll properties of sets of points that are invariant under the
group of motIons. EXERCISES 2.5
Th~ proof of the invariance of a property under a group of
tran.sformatlon.s can be either synthetic or analytic. Both types of proofs 1. Drnw a figure for each case in T~lble 2.2.
are Illustraf.ed In examples. . 2. Verify, as stlggested, Ihl.1t the eqmHions for direct motions have a +
symbol and the equations for opposite mOl ions huve a - symbol.
EXAMPLE. Prove that a segment and its image are parallel 3. Draw a specific example showing that the product of a reflection and tl
under a translation.· rotation could be a glide reflection.

In Figure 2.29, let As be the segment and A'B' its image. 4. Draw it specific example showing that the product of n reflection and a
rotation could be a reRection.
A' 5. Draw a specific example showing that the product of two glide retlections
could be a translation.
6. Show analytically thai .\ translation has no invariant points.
A
7. Investigate analytically the results of setting x = s' und .1' = y' in the
equations of a rOlation.
, B'
8.. Use Theorem 2.6 to prove that, if three llollcollinel.lf points of a pinne
r~ll1ain invariant under"a motion. the motion is the identity.
9. Prove analyticallY thaI.. for a translation, the image of two intersecting lines
8
is two intersectipg lines l.Ind {he imag.e of the origim.1 point of inlen,ection
FIGURE 2.29 is the new point of intersection.
GEOMETRIC TRANSFORMATiONS 67
66 CHAPTER 2
DEFINITION. The rotation .
angle IX about a fixed line' . R{l. ex) III three-space thlOugh
10. Prove analytically Ihal the angle between IwO intersecling lines is an invarianl See F'Igure 2.31, 15 a rotatIOn in a plan .
e perpendIcular toanl.
under Ih. s" of all translations. '
2.6 MOTIONS OF THREE-SPACE
The motions of three-space are tess familiar than those of tWO- i

space. To discover the various -.
types
of motions, it is again essential
A
to see what types of transformations can be applied to a set of .points " 0

in three-space so that the distance between any twO points is preserved. B
TIlese transformations are introduced rather briefty in this section. <
The motions of three-space are six in number. The three simplest

morions are translation, rotation, and reflection, but these concepts must
FIGURE 2.31
be clarified for three-space.
The definition of translation in space is analogouS' to the
The image of point A not on I . , f-Io
definition of translation for the plane. determine a plane 1t perpendicular to I." a POUlt B su"h that ;<0 and OS
DEFlNITION. The Iransformalio n T. such thaI' T)X)
DEFINITION. A reflection R . '
X + A is a translation of the space in the direction 0 A. a~out a plane such that for Pad' ~ m three-space is a reflection
b -- • n Its image P' .
ISector of PP'. See Figure 2.32. ' 1t IS the perpendicular
for example, in figure 2.30, if A is the point (2, 4, 8) and X
is the point (I, 2, 3), then TA(X) = (3, 6, 11). The image point can p
easily be found by addition of coordinates. Verify the fact. that a
translation in three-space sends all the points of a plane into the
points of a parallel plane.
z
/ ---71
.,
1-1--........ TA (Xl p'
I A I I
I 'I I FIGURE 2.32

I I I \
I xl I I
I , I I I
I I I I . The complete analytic res .
I oori+l'.f--L.f--- y dimensions is not given but 5 . P entation of motions for threc
I /
.:-=-=J.. equations for a transla~ion i:~~al cases arc considered. The following
~
translations in a plane:' ree-space are analogous to those for

x
FIGURE 2.30
GEOMETAIC TRANSFORMATIONS 69
68 CHAPTER 2
While the complete proof is not given here, it is analogo.tls. to

x' ~
X + a,
the proof of Theorem 2.4 in that the anal~sis depends ~n c1asslfY,lIl g
y' ~
y + b, motions in three-space as direct or opPos1te and S.hOW111 g that e,\ch
z' ~ z + c. space motion call be expressed as the product of refle~tlOns about a ~lane.
Table 2.7 shows which motions are direct and which are OpPosite. It
For a rotation in space, if the fixed line is the z-axis and the also gives a description of the invarianl points for each.
plane is the xy-plane. then the image of a point in that plane can be
found using the rotation equations given previously for rotation in a plane.
TABLE 2.7
about the origin. for a point in a plane parallel to the xy-plane, the z
coordinate will remain constant. '1II'(lrilllll Pflilll,~
For a reflection about the xy-plane, the image of (a, b,c) is
Mmiml;1I Tlu'('e·Spa('f! ,
Di/"f.'('( "I' OPJI(1sih!
la, b, - cj, hence direct " line

rotation none
translation. direct
opposite ph\lle of reflection
x' x, reflection III lensl one point
~
rOlatory reflection opposite

direct none·
screw displacement none
y ~
y, glide rellect ion opposite
-
z' ~ -z.
Corresponding sets of equations can be written for reHections about the The table makes it possible to indicate the product, of any two
xz- and yz-planes. motions, but the results cannot be determined t~niquely Without m~re
In addition to the three simple motions, it is necessary to. define information. For example. the product of a screw dls~lac.ement and a g~lde
three additional types of motions which are products of the basic ones reflection is an opposite transformation, hence it IS either a refleclton,
so that the set of motions of three-space wHl be closed. Table 2.6 a rotatory reflection, or a glide reflection. .
. shows the definitions of the additional motions as products of two The study of Table 2.7 shows lhat a refle.ctlOo has an invariant
simpler motions. plane and that no other motion h as as ex t ensl ve a' set of invariant
points. The foHowing theorem is analogous to Theorem 2.6.
TABLE 2.6
THEOREM 2.8. A motion in space is uniquely determined by a J,.
Mofim! Explalllltio/f tetrahedron and its image.'
•. screw displacement
b. glide reflection
rotation and translation along axis of rotation
reflection and translation parallel to plane of reflection
c. rotatory reflection reflection and rotation with axis perpendicular to plane The proof is left as an exercise.
of reflection
EXERCISES 2.6
A complete analysis of the motions of three~space is not given
here. but the theorems are analogous to the theorems in two dimensions. ·
Let the vector for a trans 1atlOo
be (_ ,_, 1,7), and give thl.: linage for each
I.
point under the translation,
THEOREM 2.7. The motions of three-space constitute a group b. (2, 5,3)
a. (O, 0, 0)
d. (O, -4, 1)
of transformations. c. (-2, -3, -5)
;
!,
70 CHAPTER 2
geometry, however, the idea of ~f!!iJ~J... l?ut noncon~ruent figures is
3 4) what points have t hese images? introduced. Similarity is an example of a type of transformation in which
In the translation with vector (-1, - • •
2. b. (2, - I, 3) distance between points is not preserved. But distance between points is
L ~~~ . modified in a consistent way throughout the plane.
h type of O1otton 0 [ space.
3 Describe the inverse for eae d rotation of 450 in the
. ·t(2,45)un era ,
What is the image of the. pOlO I
DEFINITION. A plane similarity is a translonnation of the I:
4. = 4 about the y_a:us'l .
plane y • . e of the point under a reflection about the given plane onto itself such that I/(A)/(B)I = rIABI. r
where is some positive
For Exercises 5-8, give the Imag real number. See Figure 2.33.
plane. 6. ( _ 2. 4, 3), xz plane
5. (3,S,I),xyplane S. (_I,4,-2),x=3
7. (2, 3. 9), yz plane r the last three types of motions of space.
For this example, ,. = 2, so that AS is stretched by the :.1
9. Sketch an example of each 0 . r space remain invariant under a similarity transformation into a segment twice its length. For r = 1/3, i, l
p that if four noncoplanar pomts 0 it would have been shrunk into a segment one third of its length. For
10. rove , . . h identity.
r = I, the similarity transformation is an isometry. The number r is caIIed
motion, the motton 15 t e [ ' that is the product of:
. T' r r the type 0 mouon the ratio of similarity. ...." ... j
ll. Nalne the posslbl lUes 0
. d a translation. ,. .I
,. I
a. A rotatory reflection an .
b A rotation and a glide reflection. 2 THEOREM 2.9. The set of all similarities of the plane is a
. ft tion about the plane z = .
Derive the equations for a re ec transformation group.
12.
13. Prove Theorem 2.8.
,
Make a sketch showing a tetrahedron and its image for a: The inverse of a similarity with ratio r has the ratio l/r. TIle
l4. Translation.
ratio of a product of two similarities i~ the product of the. ratjps of the
two similarities. For example. if the first ratio of similarity is two and
15. Rotation.
the second ratio of similarity is three, then the ratio of similarity for the
16. ReHection. product transformation is six. Theorem 2.9 makes it possible to think of
the motions of a plane as a subgroup with r = 1 of the group of
SIMILARITY TRANSFORMATIONS . similarities of the plane.
2.7 In the previous section, it was proved that a unique motion is
d' d in the last three sections of
In the Euclidean moti~,nSdslpu Ie erties such as length, area, anQ._ determined if three non collinear points and their images are known.
. was preset ve. .
rop ' The same proof can be modified to apply to the more general group of
this chapter, QI§tance Even . m ' I 1
hIgh SC lOO
reserved under ISOInetnes. - similarities.
volume are a 11 P
(Al
THEOREM 2. 10. A plane similarity is uniquely determined when
a triangle and its image are given.
In Figure 2.34, b,A'B'C' is the image of b,ABC. The distances

A P'A', P'B', and P'C' are determined, and P' is a uniquely determined
(8l point for any given point P.
FIGURE 2.33
P' P'
/1'\
"<0
/! \
// ( \
/ / \
A' / I
! \ ',~\. / \
\
Q'
\
\
FIGURE 2.35
\
·d
responding segments under a homothety should be p.araHeL .That ~he '-;'j
"'.
8' ':1
property of parallelism is an invariant under a homo the tic transfortnal1OI1 -,.J.
FIGURE 2.34
is proved as the next theorem. ,i
An i~tuitive exploration of how to explain similarities as a series
of simple steps leading from a figure to its image indicates that there THEOREM 2.11. Th~ image of a segment under a hOlllOthety is a
must be both a c~ange in position and a uniform change in size (but parallel segment.
not a change in shape).
For example, in Figure 2.34, changes of position can result in
In Figure 2,35, OP'/OQ' = OP/OQ, and 60P'Q' - 60PQ,
trIarigle ABC being situated so that A and A' coincide, so tha~ AB
lies along iB', and so that AC lies along A'C". A change in size is then hence PQ 1Il'72, , ,
It is sometimes said that homothetic figures are both sImilar
necessary irB and B' are to coincide. Since motions of the plane can
accomplish the change-in position, the new transf~rmation needed is one
and similarly placed, since their corresponding sides are parallel. For ,.\
example, in Figure 2.36, 6ABC and 6A'B'C' are similar but not
to accomplish the .Ul'iiform change in size. The new type of transformation
IWIDothetic, whereas 6ABC and 6A"B"C" are homothettc, Also,
must be a special similarity in which the image of a set of points is a
LJA'B'C' and 6,A"B"C" are similar but not homothettc.
similar figure with corresponding sides parallel.
DEFINITION, A I,omotllety H(O, r) is a similarity such that, for
t;7'
B U
~
any point P, OP' = rOP where 0 is the center of the homothety and r 8
is its ratio.
A
/"'>.C AU C"
A'
The center of homothety is an invariant point, according to this FIGURE 2.36
definition. The ratio of the distances from the center to the image
point and to the original point is a constant, the ratio for the The property of parallelism is preserved under a hornolhety.
homothety, For example, Figure 2.35 shows a homothety with Another invariant property is the property of he.iUg Il cir~l~.
OP'/OP = 3/2, Remember that the ratio of similarity always compares
the image with the original.
The exploration of the need for a homothety showed that motions THEOREM 2.12. The image of a circle under a homothety is a
could result in an image in the correct relative _position; thus cor- circle. See Figure 2.37.
74 CHAPTER 2
P' I(A)
B B'
FIGURE 2,37
FIGURE 2.38
Let 0 be the center of homothety and P' and Q'. the images formations begins with the b '
homothety 'th 0 servatlOn
that the equations for a
of a point on the given circle and its center. respectively. Since WI center at the origin are
PQ 111"([ by Theorem 2,11, P'Q'IPQ = OQ'IOQ, or
X' = rx,
P'Q' = PQ'OQ' y' = ry,
OQ
But each segment in the expression on the right has a fixed length, S,ince a similarity is the product of ' "
the equatIOns for a similarity are of the form a molion and a homothety.
hence P'QI is a constant. For any position of point p, P' lies on a
circle with Q' as center.
So far it has been determined that a similarity !can be con-
x' = le(ax + by + c)
sidered the product of a homothety and a motion. Because. a hornothety y' = ±le[(-bx + ay) + d]
also can accomplish a translation, the foHowing stronger s~atement can fo r a, b c', d .~ Rand a 2 + b 2 = 1.
J
be established. SImIlarities in three

motions of thre space also form a group having the
e-space as a subg
THEOREM 2.13, A plane similarity that is not a motion is the analogous to those ~ t d' roup, The theory and definitions are
or wo ImenSlOns.
product of a hOll1othety and a rotation or a reflection.
THEOREM 2, 14.Aspacesl' 'Il . .
It is assumed here, and can be proved using the.orems from a tetrahedron and its image are give:' anty IS uniquely determined when
topology, that every plane similarity not a motion has exactly one
fixed point. In Theorem 2.13, the rotation is about the fixed point and
, THEOREM 2,15, The set of II ' .
the reflection is about a line -through the fixed point. (ormation group . ' a transR
a slnularities of space IS
111 Figure 2.38, let 0 be the fixed point and All any segment.
A' and B' are the images of A and B under a homothety with 0 as
center. Let j(A) and I(B) be the images of A and B under any THEOREM 2.16, A space . 'I '
space is the product of a I ' Simi anty that is not a motion of
similarity that is not a motion, 60A'B' - 60j(A)j(B), b.OA'B' can lOmothety ab ut '
10 atlOn about a line p~ssing through the fix °d I~S fixed point and "a
. t '
be mapped into OJ(A)I(B) either by a rotation through angle AOf(A) e pomt.
or by a reflection about the bisector of this angle, In the first case, the
similarity is direct, and in the second case the similarity is opposite.
For Theorem 2.16 to be true. it is necessary to reinterpret a
The analytic development of the equations for similarity tran5-
76 CHAPTER 2
homothety in a somewhat more general way by allowing the ratio r to be 6. In figure 2.37, compare the ratio of the radii of the two circles with the
either positive Ol: negative. The p}1Ysical interpretation of a negative ratio ratio of similarity.
7. Find the image of (3,5) under a homothety with center at (0,0) and with
is that a point and its image are on opposite sides of the center. of
homothety. ratio 3/4.
1 Consider, because of Theorem 2.14, any tetrahedron OABe 8. Find the image of (4, 3) under the similarity with equotions
J with the fixed point 0 as one vertex. It is assumed here, as for two x' = 2{3x - .I' + 5).
I dimensions, that a fixed point exists for any similarity. Let r be the
J" = 2(x + 3)' + 2),
ratio of similarity. Then the image of DABe under the homothcty
H(O,r). OA"B"C', is congruent to the image OA'BtC· under tht: Prove analytically that if two lines illtersect, their images under a sim\iarily
9.
similarity. If the motion connecting OA"B"e" and OA'B'C' is direct, also intersect.
then the similarity is the product of H(O. r), for r positive, and a to. 1s the set of homothelies with a given invariant point a group of lrHns-
rotation about some line through O. If the similarity is opposite, then formations?
r is negative. 1L Prove analytically that the property of being a parabola is an invariant
This chapter has enlarged the idea of Euclidean geometry by for the group of homotheties with a given invariant point.
showing how it can be considered the study of those properties that
are invariant under the group of -motions (or under the group of
similarities). Through this new approach, additional significance is
attached to the study of congruent and similar triangles. Even more
important, however, is the anticipation that more general transformations
exist (such as those given in Chapter 7, for example), with Euclidean
geometry as a special case.
The study of Euclidean geometry has been revitalized during
the present century. In addition to the approach through transformations,
a classification and detailed study of sets of points in Euclidean
geometry utilizing the modern concept of convexity is a subject of current
research. The geometry of convexity is the subject of Chapter 3.
EXERCISES 2.7
l. Find the length of the image of a 3-inch segment under a similarity with
ratio 4/3.
I ,. 2. Find the ratio of similarity if a 5~illCh segment has .a 6·inch image,

,I'· • 3. Name some properties preserved under all motions which are not preserved
I under all similarities.
I
i 4. What is the effect of a similarity 011 the area of a triangular region?
\' 5. Find the ratio of similarity if a square region with an area of 12 square
units has an .image with an area of 17 square units.
CHAPTER 3
CONVEXITY
3.1 BASIC CONCEPTS
The study of convexity in Euclidean geometry is a very modem

development. As is true with the approach through transformations. the
study of convex sets employs many basic ideas of modern mathematics
to provide added meaning to concepts that h~ve long been a part of
geometry.
The words "convex" and "convexity" arc common outside of
mathematics. Convex means curving outward. For example, a convex lens
bulges outward. Fortunately. the use of the word convex to describe
particular sets of points in mathematics is only a careful refinement
of this common meaning. Study Figure 3.1, parts a and b. to discover
79
80 CHAPTER 3 CONVEXITY 01
the basic difference between convex sets of points and those that are measure less than n) and its interior, and a spherical region. The empty sel
nonconvex. and a set consisting or a single point are both convex by agreement.
To show that a particular set is convex by definition is not
always easy. In addition to Theorem 3.1 below, analytic geometry is often
~ employed. You should assume here and elsew:here in this chapter. unles~
stated to the contrary, lhat the variables for coordinates of points are
~ elements of the set of real numbers.
~
The postulational system for a secondary school geometry
(a) Examples of
convex sets includes the assumption that a hair-plane is a convex set. For example,
see Appendix 3, Postulate 14. The determination of whether a set is
con~ex or not is facilitated by the following theorem,
THEOREM 3.1. The intersection of two convex sets is a convex

set.
(b) Examples of
non Convex sets
FIGURE 3.1
. F~r a conv:x set of points, such ;~s K in Figure 3.2<1, any

~egment wuh endpOI?ts A and B in the set lies wholly in the set This
FIGURE 3 3
I~ ~ot ,true fer all. palrs of points in the nonconvex set in Figure 3.1b.
f 01 eX~U1)ple, CD Includes points outside set K'.
In Figure 3.3. tet J( t and K1 be any two convex sets and
let set S be their intersection. for A. BE S. all poinls of AB are
elements of [( I because K t is convex. All points of AB are also elements
of K 2' because it is ,tiso convex. Hence, ABc (K I (\ 1\.1): thus,
K I (\ J( 2 is- convex. Theorem 3.1 may be generalized to more than two
sets, as in Exercise 3 of Exercise Set 3.1. The theorem may also be lIsed to
(a) (b) show the convexity of other COl11mon sets of pl'lints. For example, nn
FIGURE 32 angle {measure less than n} and its interior can be defined as {he inter-
section of l W,-) half-planes, so that Theorem 3.1 applies.
DEFINITION. A COI1l't!.'\ set is a set or points K such that ir An alternative approach is the use of an;'llytic geometry to
A. Be {(.lhen As c K. prove that sets of points are convex. This approach can even be used for a
half-plane.
The symbol c means "is a subset of." Although this definition
of convex set may seem restrictive, it is general enough to include many EXAMPLE. Show that S = {{x.y}: x > O~ is a plane convex set.
:~ tl~e ~:~)~1~1~ s~ts ?r points studied in Euclidean geometry. Examples IS is the set of points with cOl'lI-dinales {s,y} such that x> 0.)
e (\ ell cul.1I I eglOn. some polygol1i.ll regions, segments, an angle (with Let At'I'YI) and B(Sl'."2) be finy two points of S. as in
·",
CONVEXllY 83
82 CHAPTER 3
. > 0 and x 2 ;? Xl' Then, rm any poil:t

Figure 3.4. such that Xl > 0, . . \ 2 ' ;nd ;[B c S, so that S IS
iB x ;:,: x, ;:,: X l ' hence x, > 0
P(x/"Y ) 0 f A • 2 _ -
conveX by definition.
v
(a) (b)
FIGURE 3.5
an interval of real numbers. If the function is onewto-one, except

possibly at the endpoints. the curve is a simple curve, and if the points
corresponding to both endpoints of the defining interval are identical,
the curve is a closed curve. For this section, you will need to recognize
examples of these various types of curves from drawings so that the
equation need not be dealt with. Intuitively, a simple closed curve is
thought of as a curve that begins and ends at the same point but does
not C('OSS itself; thus there is only _one interior. {See Figure 3,6.) The
FIGURE 3 4 set of points on and inside a simple closed curve or an angle in a
plane is called a plane I'egion.
.' stematic study of convex sets in the
Credit for begmnmg the sy d H Minkowski. A
. . en to H Brunn an .
early twentieth century IS glV W Fenchel showed much progress.
1934 survey ,by T. BOl1nes~n .:n:ure 'Mathematics, Volume was thevn,
The Proceedmgs of Symposia 1 • n Convexity sponsored
. r the 1963 SymposIUm 0
publication resultmg rom. . Th' olume is an 'interesting
by the American Mathematical Socle~y. IS V
Curve Simple curve Closed curve Simple closed curve
reference text for the student of conve~ltY. '0 erties of convex sets, it is
Before exploring 1Tlore c~mp ex pi P mod. ern geometry that FIGURE 3.6
. t duce several bastc concepts rrom
necessary to 1fl 1'0 . S
will be useful both here and.In late~ c/h(ap)t~r ~ontintlous at x = a if and
Precise definitions of interior, exterior, and boundary points of a
Recall that a function ~. -:- _ x IS . set of points such as a pJane region depend on the concept of
e > 0 , there exists a asuch that
neighborhood of a point.
only 1' ,fgiven
'
\fIx) - I(a)\ < e if \x - "\ < J.

DEFINITIONS
. t of its domain.
. 's continuous at every pOUl
'f'l For Two Dimensions: The open circular neighborhood with radius r
A funclion is continuous I t 1. . F 3 5a is continuo liS, whereas
for example, the function tn 19ure . of a point P is the set of points inside a circle of radius r with P
as center. {See Figure 3.7a.} This definition can be written in symbols as
the function in Figure 3.5b is not. f quations of the form
. I . I,h of a set 0 e
A curve 15 t le gl a , d the domain of t N(p.r) = fA: IPAI < r).
) =- (·1) . for 1· and .9 continuous functIOns an
s= j .( l.y-g
oLinterior agrees with the common understanding of interior l.hat j~ used in

speaking of the interior of a simple closed curve or the Intenor of a
sphere.
(a) (b) DEFINITION. P is an exterior point of a set'S if and only if

FIGURE 3.7
lhere is a positive real number r 's-~ch thal N{P, r} c ~S.
For Tlwee Dilnensions: The ope/1 spherical neighborhood with The symbol ",S means "not S" or the complement of. S: A
radius r of a point P is the set of points ill space inside a sphere df point is an exterior point of a set if every point sutnciently. ne.ar 1t .IS a
radjus r with P as center. (See Figure 3.7b.} This definition can be member of the complement of the set. For lhe set of points Xl + yl. < t.
written in symbols as the exterior is x 2 + y2 > 1. Point B is an exterior point in Figure 3.8b.
A point that is neither an interior nor an exterior point. of a sel
N(P.r) ~ {A: IPAI < rl. is called n botllldqry pOint. The set of boundary pOints for a set IS called
lhe boulld£llJ~-'~"r" the set. Note that if P is a boundary point of S. then
A clQ~ed neighborhood in two or three dimensions includes the il is also a boundary point of the complement of S.
points on the circle Or sphere as well as those inside. The definition
of closed neighborhood, using set symbolism, is DEFINITION. P is a boundary point of set S if every neighbor-
N[P.r] ~ {A: IPAI ;;; r}. hood of P contains both points of S and of ..... S.
Both open and closed nejghborhoods are convex sets of points. F or the set x 2 + l < t. the boundary is x +- y2 = 1. Point ~
2
is a b9uudary point in Figure 3.8e. The boundary of.a half-plane IS

DEFINITION. P is an intel:iqr point of a set S if and only jf the line determining the half-plane. A simple closed curve IS the boundary
, there is a positive real number r such that NCP, t) c: S. of the set of points in its interior.
You should realize that the boundary points of II set mayor
may not be elements of the set. Another important conclusion is that,
This definition means intuitively that a point is an interior
point of a set if every point sufficiently near it is an element of the set.
For example, point A is an interior point in Figure 3.8a. This definition
(a) (b) (e)

FIGURE 3.3 FIGURE 3.9
CONVEXITY 87
B6 CHAPTER 3
to decide whether a point is an interior, an exterior. or a boundary Some bounded sets are .
cannot extend indefinitely F shown In Figure 3.1 I. Bou d d
. or example, the para~ I n e sets
point, it is necessary to consider the number of dimensions. For example, o a y = x 2 is not a
in figure 3.9, ((x,y,O): x' + y' ;;; l} has no interior points in three-
space. Each point of the set is a boundary point, since every spherical
neighborhood for a point of the set contains both points of the set and
points of its complement.
The classification of points as interior, exterior, or boundary
points for a set leads to useful classifications of the sets themselves.
FIGURE 311
DEFINITION. An open set has only interior points. bounded set beca use no cJrci
.
concept of a b e can enclose the
set ollnded set should not be co' d set of points. The
. nluse with Ihe boundary of a
An open neighborhood is an open set. as is the interior of any
simple closed curve.
y(.$2j..~1ERCISES 3.1
DEFINITION. A closed set contains all its boundary points.
(, 1.! WhicIl of these sets are convex?
Examples of closed sets are x 2 + y2 ~ 1. polygonal regions, and a, An angle.

c, ~nt~rior of an ellipsoid. b. A re.ctangular region.
a segment All It is important to observe that these definitions of open e. smgle point. d. StraIght line.
and closed sets. are not mutually exclusive, nor do they include all g. A triangular region, r. A ray.
possible sels of points. A set may be both open and closed or neither 2. Which of these sets are convex sets? h. A triangle.
open nor closed. An example of a set both open and closed is the
a. AA
b circuJargre io n with
. OIle point 0 tl
entire plane. Since this set has no boundary, it includes all its boundary . rectangular region with 0 n le boundary removed
points. At the same time, ali of its points are interior points. Examples c. A rectangular re io . ne vertex removed. ,-'
of sets that are neither open nor closed are given in Figure 3.10. removed. g n with one point, not a ve I
dA' rex, on the b
, Circular region 'Ih . . oundary
3 WI one mteno .
. Prove that th . r pomt removed
e mtersection of any coil . .
,4. :: ~~:I~:~:aO~~WO convex sets eve, a ::~::xo~e~07nve)( sets is a convex set
nvex set? i:,
Prove that a triangular region is a
Show analy" II convex set. p-' - U -, '
lea y that S = f( • ,'"
Sh
Gi'
ow analyticaUy that T = «. .
l",y):x>J}is'
_'X,y). x > 3 and,
a convex set. -'j -'
4\ .
.-\
FIGURE 3.10
D ve ~ JmenslOnal nei hb
8 definition for a one d" J > I IS a convex set
. L.' •. ' :'/
f·_1 -' j:, - '/'
escribe the interior b g orhood.
DEFINITION. A bounded set in two dimensions is one that is a • oundary, and eX[eri r.
a. ({x,},): x2 + )'1 :s;: IJ or or these plane sets.
subset of some circular region with a real number for radius. In three - b . {('( y)'}' > 11'
ion
dimensions, a bounded set is a subset of some spherical .reg with c. ((x,y):.v < IxU .,. x i
d. {IX 1'). l ' + 4~
,.. J' I}
real radius.
CONVEXITY 89
88 CHAPTER 3
e. ((.x,y): x and)' are integers}

L {(x.y): x and yare raliooalnumbersJ
to. Describe the inlerior, boundary, and exterior for these sets of points in space.
2
a. {(x.y,z): x 2 + y2 + Z2 ;$. I} b. ({x.y,O): x + }'2 ;£ 9}
c. {(x,J'.O): x > y} J. {(x,y,O):;'( = 2)
1L Classify these sets as open, closed, neither, or both.
FIGURE 3.12
a. The sets in Exercise 9. b. _ The sets in. Exercise 10.
c. A rectangular region with one vertex removed.
The proof of the following theorem results in an alternative
12. Which of the-sets in euch exercise are bounded?
definition for supporting lines of convex sets in two dimensions with
b. The sets in Exercise 10.
F
-, 13.
a. The sets in Exercise 9.
Prove that the boundary of a set of points is also the boundary of tht:
interior points.
I complement of the set.
.'
THEOREM 3.2. A line is a supporting line for a convex. set of
14. Prove thnl the complemenl of a dosed set is open.
points if it goes through at least one boundary point of the set but nO
15. What can you say about the complement of a set that is neither closed nor
interior points. and conversely.
open?
16. Give examples of (a) a bounded set that does not contain its boundary and
(b) a sel that contains its boundary but is not bounded. Theorem 3.2 includes two in).plications:
a. If a line goes through at. least one boundary point of the set
, but contains no interior points, theI~ it is a supporting line for
i- 3.2 CONVEX SETS AND SUPPORTING
the convex set of points.
" b. If a line is a supporting line for a convex set of points,
LINES
l: then it contains at least one boundary point of the set but no
interior points.
I,. The basic ideas of convex sets were introduced in the previous
section, along with the vocabulary needed to continue the-study. It is now
Associated with each implication are three other impiicalions.
the converse, inverse, and contrapositive. These are shown symbolically
possible to explore some additional concepts associated with convex sets.
(n this section, it is assumed that the geometry being considered is the as follows:
geometry of two-space. A concept especially important in discussing [J-q
Original implication
convex sets, but applicable to other sets of points also, is the concept q-+[J
Converse
of supporting line. ~p _ ~q (not p implies not 'I)
Inverse
Contra positive -q -+ ""P
DEFINITION. A supporting line for a set in two dimensions An implication and its contrapositive are logically equivalent.
with interior points is a line through at least one of the boundary So are the converse and the inverse. Furthermore, the converse of one
points of the set such that all points of the set are in the same closed implication of Theorem 3.2 is the other implication. Proving an «if and
half-plane determined by the line. only if" theorem is the same as proving a theorem and its cOllv.crse.
Proving p -+ q and q -* p results in proving the equivalence l' '1. .(-4
Sometimes, especially in geometry, it is easier to prove the

Figure 3.12 shows three examples of sets, with several supporting
inverse and/or contrapositive t,han some other form of the ~mplicH(ion.
lines for each.
CONVEXITY 91
90 CHAPTER 3
Any of these pairs of statements may be proved to prove an implication

and its converse: !
( tontrapOSitiVe)
Theorem) (contrapositive)
Theorem) \ Converse
( Converse ( Inverse 1n verse
ve (a) A tangent that is (b) A supporting lina
a supporting line not
For the first implication of Theorem 3.2, the easiest to,pro of thai is not a tangent
these pairs is that including the contrapositive and the inverse. FIGURE 3.16
lneerse In Convex geometry til

ContJ'apasilillt' 1f a line contains at least one depend. on t he notion of ' Iii"e concept of t. ngent, which does not
1f a line is not a supporting line for bo.undary point and als0 interior concept used in calculus The nblts~ l~ somewhat different from til
a convex set in two dimensions but points, then it is not a s~pporting cone. . egmnmg c o cept
n ·IS that of a tangente
contains a boundary point, then it line for the 'convex set in twO
conlains interior points
dimensions.
DEFINITION A I ent
all rays that: (a) have' bOU1;;:r c~ne of a convex set is the set of
(b) also pass through oth . y pomts of the set as
er POll1ts of the conv . endp OInts
. and
ex set or Its boundary.
A E 8 F
c/6.\,D
'!VII\~'
FIGURE 3.13
FIGURE 3.14
proof: If I is not a supporting line, Proof: Suppose I contains bound- (a) (b)
as it is not in Figure 3.13, then FIGURE 3.16
ary point A and interior point B,
points A and B of set K can be as in Figure 3.14. There exists a
found in different half-planes
neighborhood of B, N(B, e), con- ..... F'Igure 3.16 shows two t
formed by f. The intersection of and AD are the boundaries oftl angent cones. For Figure 3.160, A-C
taining interior points. But this ray AC) . In general th b lC tangent cone. (Th e notation AC -
AB and I is not empty. But C is an
interior point of set K, since K is
neighborhood includes points in tal1g~nt bo~nd:rie:~;daries
cone. The are not themselves rays ~;e;17;
convex and A and B are interior
both of the half-planes formed by I; In FJgure 3.16b, BE and BF . the ~angel1t cone are called semira
al e semJtangents and are colhnear.
. ngenls.
hence, I is not a supporting line.
points. See Exercise 14, Exercise
Set 3.2. DEFINITION . ,The umon . of tw ' .
The concept of supporting line is closely related to the more tangent to a convex set at a point. a collinear semi tangents is the
familiar idea of a tangent. from calculus, the intuitive idea of a tangent
to a curve is that of a line intersecting lhe curve at a' point and
t From Figure 3. 16,or from a c 'd
having the same slope as the curve at that point. As Figure 3.15 . angent, it should be clear h anSi eration of the definition of
t at there are tangents at ~ome boundary
shows, supporting lines and tangents are not necessarily the-same thing.
I'
,
I CONVEXITY 93
92 CHAPTER 3
is greater than re. then the original set is nol convex, contr<-lry to the
points of a convex set and not at others. It is possible to classil}' beginning /;lssumplion. See Exercise 12. Exercise Set 3.1.
boundary points of a convex set as regular or corner points on the
basis of whether there is or is not a tangent to the curve passing
A
through that boundary point. There is a tangent at every regular point.
In Figure 3.16, II is a corner point, whereas B is a regular point.
Other examples of corner points are the vertices of a convex polygonal
region. On lhe other hand. all of the points on a circle are regular
points on lhe boundary of the circular region.
One of the main reasons for introducing the concept of sup~
porting line is that it can be used to dislinguish a convex set frO!H one
FIGURE 3,18
," that is nonconvex.
I
THEOREM 3.3. The interior of a simple closed curve is a convex

r.
b. I S' is not a convex set, then there is not tl supporting
set if and only if through each point of the curve there passes at least line at every boundary point.
This statement is the inver'se of the implication in In), so
one supporting line for the interior. proving it will complete the proof of Theorem 3.3. (See Figure J,19.)
Since S' is not convex, an interior point C and a boundary
- ~ point B can
be found such that a boundary point A lies on Be. Be is not n
supporting line, since it contains interior points, Any other line through
A has D and C in opposite half-planes; hence il cannot be a supporting
line. There is no supporting line to St through A.
<al (bl
FIGURE 3.17
Figure 3.17 helps to explain intuitively the significance of

Theorem 3.3. If a set is not convex. then points (such as A) exist on
the boundary ,such that any line through A contains interior points. FIGURE 3 '19
Through any point A' on the boundary of a convex set, at least one
line can be drawn containing no interior points.
tr a line is a supporting line for a set in twO dimensions.
Proof: the dosed half-plane formed by the line and containing the set is called
a. If K is convex, then there is at least one supporting line a supporting fw/f-p/alle. This concept can also be used to give an
passing through each boundary point A . additional property of convex sets.
• ~ ~
Let AB and AC be the semitangents at point A in Figure 3.18.
If- the measurement of LBAC is n, then Be is a tangent and hence THEOREM 3.4. A plane closed convex set that is a proper
a supporting line. If the measurement of L SAC is less than n. then subset of a plane is the intersection of ,-,II its supporting half-planes.
~ ~
AB and AC are both supporting lines. If the measurement of L SAC
CONVEXITY 95
94 CHAPTER 3
uadrilateral and ils interior, EXERCISES 3.2

An example of this the~rem for a ~ rmed by the sides, is in
the. supporltng planes 0
showing four of For Exercises 1-4, lell whether the line is or is not a supporting line for the set.
Figure 3.20. Set Lilli:'
I. Circular region line through center
2. Circular region tangent
J. Square region line containing side of square
4. Square region line containing diagonal 'of square
For Exercises 5-8, describe the regular points and cornt;:r points for each of these
sels.
5. Circular region.
6. Square region.
FIGURE 3.20 7. Angle and its interior-measure less than 180".
8. Convex polygonal region.
of points in the corlvex set is 9. Use Theorem 3.4 to give a definition of:
4 tates that t Ile se t . fting
Theorem 3. S ~ • I 'te 'section of the suppa a. A triangular. region. h. A convex polygonall'egion.
f points m t le 10 I , :. I' to
identical to the set 0 tl at two sets are Identlca IS 10. Draw three nOtlconvex sets and show one point au the boundary of each
'ent way to prove 1 : t then
half-planes. A convem ' r I other since the sets '111US through which no supporting line passes.
1 . a subset 0 t le 1 K d the
prove that eae 1 IS " L 1 the convex set be an I L State the converse, inverse, and contrapositive of the second implication in
I ave exactly the same elements. e b K' By definition of sup- Theorem 3.2.
1 orting half-planes e .
intersection of the supp f K'
. ha If-p I',\fie,
sKis a subset 0 . 12. Prove that the interior angles of a convex. polygon have measurements less
portlllg than tr.
A IJ. Prove that a tangent cone to a convex set at a boundary point is itself a
• convex set.
14. Prove that if A is an interior point of convex set K and B is any other
point of K. then every point in ASIAB except for point 8) is an interior
point of K. Hint: Use Theorem 3.3.
15. Prove that if A and B are boundary points of a convex set K as well as
points of K. then either A1J(AB without the endpoints) is entirely in the
O~
interior of K or AB is entirely in the boundary of K.

FIGURE 3.21
, t in the complement of K. ~s in 3.3 CONVEX BODIES IN TWO-SPACE

N ow let A be a pom . f K say B IS the
d that some pomt o . . B'
3. 21. It can be prove . I' I to K at IS
figure AI the 5UpportlOg me In the previous section, the concept of supporting line gave
closest pOll1t of SlO A. s~, . the supporting half-plane of K added meaning to convexity. Many of the common sets of points from
perpendicular to AB. Then A ISK~o~.m e any p'oint it: the complement Euclidean geometry were found to be convex sets. These ideas can nOw
I therefore A ~ . llle . K'
determined b y. f K' KI C K hence K = . be used to define a special type of convex set calied a convex body.
of K is also in the complement 0 • - ,
CONVEXITY 97
96 CHAPTER 3
DEFINITION. A convex body is a convex set of points that is The contrapositive states that if K is nonconvex, then some line 'through
closed, bounded, and nonempty. an interior point of K does not intersect the boundary S in exactly
two points.)
In Figure 3.23, if K is nonconvex, interior points A and B
Many comlnon convex sets are convex bodies. Examples include
can be found such that C on AB is an exterior point. Then it can be
convex polygonal regions and Ijne segments. On the other hand, the
assumed that boundary points D and E for K exist on AB. The ray
interior of a parabola and the entire plane are convex sets that are not <->
convex bodies. Convex bodies always have interior points for the number
VA also has a third boundary point F, so AB intersects S in at least
of dimensions considered. three points.

The following theorem gives an imporlant characteristic of a
two-dimensional COnvex body.
THEOREM 3.5. A simpJe closed curve S and its jnterior form a

convex body K if and only if every line through an interior point of K
intersects S in exactly two points.
FIGURE 323
Proof:
a. If a simple closed curve S is the boundary of a convex Another theorem about a convex body and its boundary is
body K. then every line through an interior point of K intersects S in assumed here, although a topological proof has been provided by Verner
exactly two points.
Hoggatt, Jr.
-+ See Figure 3.22. Let A be an interior point of K. Any ray
AP intersects K in a segment AB. This is true because the intersection
THEOREM 3.6. The boundary of a convex body in two-space is
of two convex sets is a convex set. AB is a one-dimensional convex
body. But AB', where B' , A, and B are collinear, also intersects K in a a simple closed curve,
segment AB', and H' is the second of the two boundary points on the
line through A. Of course, the converse of Theorem 3.6 is not true, because
simple closed curves include a great variety of shapes in addition to
those that are convex. Another condition for determining whether 11
particular simple closed curve is the boundary ofa convex body is provided
p
by t.he next' theorem.
THEOREM 3.7. A simple closed curve S is the boundary of a

two-dimensional convex body K if and only if each closed polygon
FIGURE 3.22
T = f'o Pl'" PI' Po determined by successive points on and inscribed
in [( is the boundary of a convex polygonal region.
b. If every line through each interior point of K intersects the
simple closed curve S in exactly two points, then S is the boundary of a
convex body K. (It Celll be shown that the contrapositive of this second The significance of the theorem is illustrated in Figure 3.24. Any
implication. which is logically equivalent to the implication itself, holds. polygonal region inscribed in the simple closed curve of Figure 3.24a
CONVEXITY 99
98 CHAPTER 3
inscribed quadrilateral that is not the boundary of a convex region. Such a
. f'gure 324b however, some
1
lsedcurvem 1 •• quadrilateral is pictured as ABeD in Figure 3.26, but the details of the
will be convex. For t lC co. I the one shown are nonconvex.
orthe inscribed polygona I reglons • sue 1 as ' proof are left as Exercise to of Exercise Set 3.3.
(b)
(aJ
FIGURE 326
FIGURE 3.24
T' im lication involves showing that

« Two other theorems concern a special property of the length
The proof of the only b : has a supporting line at each
I of the boundaries of two convex bodies. The concept of length of a curve is
any inscribed polygon [or a convex 0 Y probably a familiar one, but it is restated here.
point. . . e side AB of an inscribed
Let 1 be the line contam1l1g 011 t"ABO contains only points of
. . ex AD c: K so t1a 1 . f"~AB DEFINITION. The length of a simple closed curve is the least
polygon. Since K IS conv, E'. 15 Exercise Set 3.2). 1
. . oints of K (xerCISe , T d upper bound of the length of all inscribed polygonal curves. Recall that
S or only mterlOf P . . interior points of an
. . of S then It can contam nO ..' the least upper bound of a set of real nu~bers is the smallest real
contalils only POUltS. d T !fe ABo contains mtenof pomts number that is greater than or equal to each number of the set. The
is a supporting line for both K a~ . (tioned into two curves lying
length of the boundary of a region is called. the perimeter of the
of K. as in Figure 3.25, then S 15 ~ar 1 f T lie in one supporting
. . f I All the vertices 0 b region.
on opposIte SIdes 0 . h . A and B would not e
.med by t since ot erWlse .
half-plane H determ , . I' f r H n K and thus IS a
. Tl n 1 is a support1l1g me 0 .
THEOREM 3.8. If K 1 and K, are convex polygonal regions with
successive vertices. Ie . . t of T has a supportmg
, ' f ' T and its intenor. Every P0111 1 I K I S; f( 2' then the perimeter of K 1 is less than or equal to the
supportmg Ime 0 1 . T is the boundary of a convex yo ygona
line passing through It. so that perimeter of K 2"
region.
Proof: Let the vertices of J( 1 be, in order. Po, PI- ... ! PII' and
let f/. i = 1..... n, be the line through PI and Pi-I' The typical picture
is shown in Figure 3.27 to help with this and the following notation.
Let S, be a polygonal region such that S, = H, n S,_ •. i = I. .... ".
where H, is the supporting half-plane of K ( bounded by II and So is
the greater region with boundary K,. In figure 3.27, S, = H, n So is
FIGURE 325 shaded. Since AB < (AC + CD + DB), the perimeter of S, < So' In
general,
. mellt can be accomplished by showing
The proof of the If state . t convex there exists an perimeter of S 1 ;'5 perimeter of Sl_ I .
if the set J( with boundary S IS no •
that.
CONVEXITY 10"1
100 CHAPTER 3
\ corresponding points. such as Band B'. There is a one-lo-one cor-

\ respondence between polygons inscribed in K1 and in K 2 For example.
<
BCD EF is paired with B' C' D' E' F'. Since each polygon S 1 of K 1 is con-
tained in the corresponding polygon Sz of K 2 • from Theorem 3.8, the
length of S 1 is less than or equal to the length of S2' The least upper
bounds of these sets of lengths are finite, with
'l
". least upper bound of S 1 ~ least upper bound of S2'
so the perimeter of K 1 is less than or equal to the perimeter of Kr

:', -'
One important modern application of the theory of convex
bodies (convex polygonal regions in this case) is in lineal· programming.
FIGURE 3.27
The. basic idea on which linear programming depends is lila.t the maximum
and minimum values of a linear function defined for aU points on a
Since the boundary of SII is the boundary of [(I' then convex polygonal region occur at vertices of the polygon. Although a
futI development of this idea is appropriate to a different mathematics
ie. perimeter of K 1 ;£ perimeter. of K 2" course, a brief explanation of the theory and an example are necessary
here in order to make the application meaningful.
Theorem 3.8 can be generalized from polygonal regions to convex To use a very simplified example, suppose that a company plans
bodies. its production for nlaximum profit and that they sell just two kinds of
items, for dit1erent prices. Other limiting restrictions in addition to the
THEOREM 3.9. If K I and K z are two-dimensional convex selling price must be considered-for example, the number of hours
bodies with K I ~ K 2- then required to manufacture each item and the availability of machines and
labor. The problem is to decide how many of each item to manufa?ture
.{: perimeter of K 1 & perimeter of K 2' for maximum profit.
from a mathematical point of view, the problem is to Ilnd
In Figure 3.28, let A be an interior point of K;. Each ray values for x and y that will maximize a linear function of the form
with endpoint A intersects the boundaries of K1 and K z in pairs of ax + by + c, where x and y can take on only certain values because of
the restrictions. The general theorem on which the application depends is
stated without proof:
1.t!Eq.R:E~ 3.10. If a function ax + by + c is defi.ned for each

point of a convex polygonal region. the maximum value occurs for the
I.' .
coordinates of one vertex and the minimum value occurs for the co-
ordinates of another vertex.
C'
o· EXAMPLE. In order to produce the first kind of item for a profit
FIGURE 328
of $50. the first machine must be used for one hour and the second
.'.,
•
102 CHAPTER 3
CONVEXITY 103
· d 0 f item for a
duce the seeon d k 1D
machine for two hours. In ~rder to p~~ used for one hour and the second For the example, the function giving the profit is! = 50x + 40y.
and the vaJues at the vertices of the polygon are:
~:~chine [or one·haU hour. Each Iflla:l~~nei;em
. fit of $40. the first maclune must . annot be operated more than
should be manufactured
12 hours a d ay. How many .e
0 at A (0,0), f= 0;
each day for a maximum profit? er of the first kind of item per day and at B (0, 12), f = 480;
If x represents the numb d k' d of item per day for a
II represents the number of the secen In at C (4,8), f= 520;
;naximum profit. tht'm at D (6,0), f= 300.
x ~ 0,
For a maxin-HIm profit, the company should make four of the
y ~ O. first item and eight of Ihe second. The profil will be $520 per day.
The inequalities. S tating the restrictions on the time for the

machines, are EXERCISES 3.3
x + y;;; (2,
For Exercises 1-8, indicate which sets are COnvex bodies. For each sel that is not a
convex body. explain why not.
2x + 1Y ~ 12.
I. Triangular region.
r' . . F'·19ure329asa
represented In .
Each of the four i~equa JtI~S IS f these four convex sets is the 2. Line.
closed half-plane. an~ the mtersectIon 0
3. Angle and its interior.
convex polygonal regIOn ABC D., 4. Circle.
5. Half-plane.
y
6. Nonempty intersection of two convex sets.
7. Ray.
B. Convex polygonal region.
9. Sketch two examples to show Ihilt 111eorem 3.7 does not necessarily hold
if the points PI are not successive.
to. Complele the proof of Theorem 3.7.
11. Sketch two examples to show thnt Theorem 3.8 does not necessarily hold
if the regions are nOllconvex.
J2. Describe all possible one-dimensional convex bodies.
13. Show that a plane convex body has a tangent at boundary point A if and
only if there is exactly one supporting line for the body al A.
14. Rework the example given for finear programming with the following
conditions: for the first item, the first machine must be nUl two hours
and the second machine three hours; the profit on Ihe first item is $90 per
FIGURE 3.29 item.
CONVEXITY 105
104 CHAPTER 3
as shown in figure 3.31a. The lines meeting at the pole are lines of
IS. A HIncher hils space for 300 cows. He wants no more Ilum 200 or one longitude and the circles are lines of latitude (Figure 3.31 b).
breed,. If he makes a profit of $40 Oil a Hereford and $50 on a Bluck For three-space, a tangent cone is actually the interior of a cone,
Angus, how many or each should he raise for <lll1axill1t11l1 profil'!
as in Figure 3.32. There is no direct analogue in three-space to semi-
tangents) since the boundary of the tangent is the cone ilself. l.r the
boundary cone of the tangent cone for a point A is a plane n, then 1t
3.4 CONVEX BODIES IN THREE-SPACE
is called the tangent pla1le at A.
. ," . Many of rile concepts and theorems for convex bodies in three
d~menslOns .are th~ san~e as. or closely analogous to those already
dIscussed fOI two d~menslOns III the last section.
. . DEFINITION. Plane 1T is called a supporring plane of a three-

dimensIOnal s~t S if ~tnd o,nly if 1t contains at least one boundary point
of Sand S lIes entirely 111 one of the closed half~spaces determined
by It •. (See .Figure 3.30.) The closed half-space determined by 1t which
conlalllS S IS called a supporting half-space of S. FIGURE 3.33
FIGURE 332
Supporting lines may also be defined for lhree-dimensional

convex sets, but the definition of supporting lines used for two dimensions
will not suffice, since a line does not determine unique half-planes or
half-spaces in space. Line 1 is called a supporting line of a lhree-
-. A dimensional convex set K if and only if I contains at least one
boundary point but no interior points of K. This definition guarantees
FIGURE 3.30 that I in Figure 3.33 will not be considered as a supporting line ror a
two-dimensioO<ll set considered in three dimensions. since the definition
Things from the physical world that suggest sets and supporting
applies only to three-dimensional sets.
plal~es ~re a ~aslcetball on a 0001' and a polar map. which is made' by
proJcctmg pOints onto a plane tangent to the globe at the north pole, Theorem 3.2 from Section 3.2 corresponds to the following
theorem in lhree-space. The proof is quite similar and is lei"t as an
exercise.
THEOREM 3.11. n is a supporting plane of the three-dimensional

set K ir and only if 1t contains boundary points but not interior points of
K.
The concepts of regular point and corner point can be extended

(a) (b) to three-sp<1ce. A boundary point of a convex set K is regular if and only
FIGURE 3 31
CONVEXITY 107
106 CHAPTER 3
The next theorem relates two-dimensIOnal
convex bodies. . . and three-dimensional
if K has a tangent plane at the point. The vertices of a convex
polyhedral region are examples of corner points, whereas each point 011 a
sphere, is a regular point for the spherical region. THEOREM 313
The next theorem is the three-space analogue of Theorem 3.3. d" . . Let / be a SUppOI ting I
ImenslOnai convex body K a t ' ine of the three-
Before stating the theorem, the concept of surface must be introduced.
A ,ur/llCe is the graph of a set of equations of the (..11'111 x = lIt) ..
!'
d~termined by I and an interior p~~~n~ and let " be the plane
dimensIOnal convex body that I I r K . Then 7C (\ K is a two-
.I' = flU), : = h(t). for./: g. and Ii continuous functions and the dOlnain of las as a supporting line.
t un interval of real numbers. A simple closed SUI/tIC!! has a single
interior. It partitions every spherical neighborhood into two disjoint ' . A is also a n Intenor
. . point of
ImenslOnal convex b d B 7C n K, so 7C n K is a two-
dK ' 0 y. ut suppose I i
sets jf the center is a point of the surface. II~. Smce B lies on I and B . s not a supporting line of
contams an interior point C of K IS a. boundary point of K II 7r I
THEOREM 3.12. A simple closed surface S is the boundary of a D of K
. . n 7C such that C lies b t .~. lere is a point
n 1[. as In Figure 3 35 TI .
convex set K if and only if through each. point of S there passes at Illtenor point of K d D . e ween A and D. Since A .
f' an E K. It follow h . . IS an
least one supporting plane of K. o K. ThIS means that I is not .s t ~t C JS an mterior point
assumption. a supportmg hne of K. contrary to the
Proc?l:
a. Proof of the inverse of the only if statement.
Points A and B can be found with A an interior pginl such
that A B contains a boundary point C if set f( is nonconvex, as in
Figure 3.34. No plane through A and B can be a supportil.lg plane,
K FIGURE 3.35
i
FIGURE 3.34 The last theorem of this section I'S a three~spa ce counterpart
of Theorem 3.4. :1
since A is an interior point. Any other plane through C separates A

i
from B and cannot be a supporting piane. Thus, there is no supporting subset oTHEOREM
f space) . "314 A three-dimensional
IS identical to tI
.
' closed convex set {a proper
plane of K through C. half-spaces . le mtersection of all I't S supportIng
.
,
b. The proof of the only if statement depends on the concept of
projection, to be explained in Chapter 7. It consists of showing that if This theorem is proved . to [he proof of
the curve is the boundary of a convex set. then there is a supporting Theorem 3 4 but th f. III a way analogous
. . e proo IS left as an exercise.
plane at each boundary point. This proof is not given.
EXERCISES 3.4
•
1. Give other examples from the physical world representing supporting planes
Set
and three-dimensional sets.
2. Describe the tangent cone for a vertex of a cubical solid. • •
3. Is every line in a supporting plane for a convex set a supporting line
for the set? Why?
• Which of the sets of points in Exercises 4-8 are three-dimensional convex

bodies? Convex
hun
4. Tetrahedral solid.
5. Sphere,
6. Half-space. FIGURE 3.36
7. Convex polyhedral solid.
8. Cubical solid.
hull, rather than the object itself, must be considered. For example, a
9. Prove Theorem 3. i 1. shoe box is rectangular, rather than the/shape of a shoe; a typewriter
10. Use Theorem 3.14 to give a definition for a tetrahedl1ll solid. cover is not usually the shape of a type~riler.
11. Prove Theorem 3.l4. The points on the boundary of a convex sel have previously
been das·sifted as corner points or regular points. In ol.·der to investigate
·additional relationships between sets and their convex hulls, it is necessary
3.5 CONVEX HU LLS to introduce a second classification based on whether a boundary point
is or is not an extreme point.
The last two sections advanced the theory of convexity by
concentrating On properties of convex bodies. In this section, a major
new idea is introduced-that of a convex hull DEFINITION. A point A of convex set K is caned a,n ext,'ellle
• \ . . ' ....,1\ .~,. poim of K if and only if there are no two points PI and P 2 of 1<
DEFINITION. The cOllvex hull (sometimes called the convex such that A is a point DrOp 1 p 2~
covel') of a set of-points S is the smallest convex set containing S.
In Figure 3.37, A. B, C, D, E are extreme points, whereas F and
Examples of convex hulls are shown in Figure 3.36. The. G are not. Every point all the boundary of a circular region is an
concepts of convex set and convex hull are fluther related by the
fonowing theorem, whose proof is left as an exercise.
,
0
A F
THEOREM 3.15. A set is convex if and only if it is its own
convex hull.
C B
In many problems that involve packing objects in convex con- E
tainers, the convex hull must fit within the container, so the convex FIGURE 3.37
CONVEXITY 111
110 CHAPTER 3
extreme point. The connection between extreme points and convex hulls notation . Let U be t he mtersection
. f .
half-planes_of_S wh ose boundanes
. c0t ' the filllte number of suppo r t'lllg
for some sets is made clear in the following two theorems. ~ple. P 1 P 2 and PP det .on am at least two points of S. For
pp ~
1 3 and P 6 P 7 do not.
24 ernunesuppor t"mg h alf-planes. whereas
THEOREM 3.16. A convex polygonal region S is the convex
Since U is convex and S c U I
hull K orits extreme points. proof involves showing that U c K ,t len K .c U. The rest of the
Suppose, as in Figure 3.40 . th t

The extreme points are the vertices. Since each pair of ex~reme of K at a minimum distance from"'; Tt ~ rt K and that B is the point
supporting line of Kat B. . le hne I perpendicular to AB is a
points belongs to K, each edge of S joining extreme points belongs to K.
Any line through an interior point of S intersects the boundary in two
points belonging to K. Then S c K. Since K is the smaUest convex set
containing the vertices, K c: S and therefore S = K.
THEOREM 3.17. The convex hull of a finite number of points in A

K
a plane is a convex polygonal region. Examples of convex hulls for a
finite number of points in a plane are shown in Figure 3.38.
I
FIGURE 3.40
At least one point P

could be Iocated paraJJel to• say
I h S must Ii e on I ; otherwise f
dl' of
subset of K ,an t ere wouJd be
. . on one side of l' lhat wo . a convex proper
defimtlOn of K as the convex hulL uld contam S, con:trary to the
FIGURE 3.38
Now think of line I being rotat ed
a outb
Pt. as'In Figure 3041.
Let K be the convex hull of the set of points S = (P I' .... p .. }
lying in a plane. If K is one. dimensional. then K is a liI1e segment
that may be considered a special case (If a convex polygonal region.
If K is two dirnensional. it is bounded and the boundary is a
simple closed curve T. Use a typical figure (Figure 3.39) to help with the
•
A
FIGURE 3.41
. "keeping• po'In t A In
. the same half- .
P, •. IS contamed in I. The h If- I
.
pl~ne until a second point P f S
a pane OpposIte point A "II 2 0
P,
. half-planes of U Th' WI then be one of the
. U . 18 means that A I' .
FIGURE 3.39
In . The conclusion is that K is ' w uch JS not in K. is
a subset of U and, finally, that
CONVEXITY 113
112 CHAPTER 3
points, the vertices. Similarly, the convex hull of a finite number of

K = U. The boundary of K consists of the union of line segments,
points in space is a convex polyhedral solid.
50 that f( is a convex polygonal region.
The idea of a convex cover can be used in yet another test to
determine whether or not a simple closed curve is the boundary. of a The final theorem in this section 15 also n three-spa~e
convex set. generalization of Theorem 3.18.
THEQREM 3. L8. Let S be an arbitrary finite set of points on a THEOREM 3.19. Let S be a finite set of point~ on a simple
simple closed curve T. T is the boundary of a convex set K if and only closed surface T. T is the boundary of a convex set if and only if" no·
if no point of S is an interior point of the convex cover ](' of S for all point of S is an interior point of the convex huH U of S for al.1 such
such sets S. sets S.
Figure 3.42 helps to make clear the meaning of this theorem.

For the convex set in Figure 3.42a, the convex hull of any finite number EXERCISES 3.5
L Describe or sketch in two dimensions the convex cover for each set listed
beloW:
Circle.
::t.
b. Tri<U1gle.
c. Two intersecting lines. d. Two noninlersectillg circles.
e. Parabola, f. Hyperbola. .
h. FOll!" distint;t points. no three of which are
D g, Angle.
collinear.
(a) (b)
2. Give examples involving practical applications of packing oNects in convex
FIGURE 3.42
containers,
3. prove Theorem 3.15.
of points, such as B, C, D, does not contain any other points On the
4. Can an exlreme poinl be:
boundary. For Figure 3.42b, however, the convex hull for EGHI contains
a. A regular poin11 b. A corner point1
boundary point F.
5. Complete the proof of Theorem 3.ts..
6. Give an example of a set of poinls thaI is dosed, hut whose t.:tl!lvex hull
hoof:
is not dosed.
a. If T bounds a convex set K, then. by definition of convex 7. Prove that a convex body is the convex huH of its extreme points..
hull, K' s; K. Therefore, all interior points of K' are also interior point:; of
8. Describe the convex cover of:
K. b. Two ske\'{ lines.
b. The proof of the converse consists of showing that if K is a. A dihedral angle.
d. Four distinct nOllcoplanar points.
not convex, then some member of S is an interior point of the convex hull c. A sphere.
9. Can the convex cover of any two~dimens.ion.nl set be three dimensional'!
of S (Exercise 5, Exercise Set 3.5).
I L Prove thal every supporting plane of the (..""t.lIlvex l."over of a closed bounded
The concept of convex hull is easily extended to sets in three
set contains al least one point of the seL
dime~si~ns. A convex polyhedral solid is the convexhull of its extreme
CONVEXITY 1 T5
114 CHAPTER 3
in a ·directio n 0 [ maXImum
. width If A
3.6 WIDTH OF A SET tlle. IWe I ti". 1 . is any p~i~t of n () S. then
. oug 1 A perpendicular 11: II1tersects 1£;' in a
pDm! B that lS a point of S. to 1£ and
A significant property of sets of points that leads to some
modern applications of convexity is the concept of width of a se.t.
DEfINITION. The perpendicular distance along a line 1 between

;.%,;'
1 1
Assume the contrary as in F
n S; butAC > AB.so·thatline~gure 3..45. Then there is a point
contrary to assumption. IS not tn a direction of maximum
l wo parallel supporting line:; of a two-dimensional set or two parallel
supporting planes of a bounded three-dimensional set is the width of the
set in the direction indicated by the line 1. A
In Figure 3.43, tV is the width of the set in the direction of I.

tn general, a set has different widths in different directions. Figure 3.44 s
shows two examples of width in particular· directions. for a two- n·-_-,,;~.I1.. __
dimensional set. The maximum width of a convex body is called the C B
di(/mefer. Note thal this agrees with the ordinary concept of dia~11eter of a FIGURE 3.45
circular region.
The proof of the next theore .
Exercise Set 3.6). . m IS left as an exercise (E xerclse
. I J.
THEOREM 3.21. Let n a '

I closed set S In
. a direction o (ndf1[ 'be parallel su pportmg' planes for
I n' n Sea h maXimum widtl TI
I c contain exactly one point. 1. len 1t n Sand
w,1
I
I
I
. . For sorne special cases the 'd .
dIrectIOns. Sets of this kind ar~ I WI th of a set is the same for alJ
FIGURE 3.44
~on~tant width does not need to bca led sets o~ constant width. A set of
FIGURE 3.43 . In FIgure 3.46. e convex, as. IS shown by tlle examples
For a set that is not a convex body. the concepts of diameter

and maximum width are more difficult to relate. The diame~er of any
set can be defined as the least upper bound of the dist"d.nces between
any two points of the set. Then, for a closed bbunded set; it can be
proved that the diameter is equal to the maximum width.
Several special properties are associated with the direction of FIGURE 3.46 ,,.
maxirnum width. Two of these are contained in the following two
theOl·ems. , The most convex set
COlllmon
(.... eglon In two dimensions E l ' . of constant Width is a .
first to t d ' u el, pnor to 1800, seems to I,ave bClrcullar
s u y convex sets of COIlS[(mt Width
. een
. t le
THEOREM 3.20. Let nand n' be parallel supporting thai we'Ie not clrcui.ar
CONVEXITY 117
116 CHAPTER 3 '
A second application of a Re111eaux triangle is in the gear for

regions. It was not until 1953 that Hammer and Sobyzk gave the first driving a movie film. The action desired is a brief, quick movement,
satisfactory general HCCOlUlt of ail such sets. followed by a momentary stoppage. See Figure 3.49. A third application
The simplest example of a convex set of constant width that is not
is the construction of a drill that lnakes a hole with straight sides.
a circular region is shown in Figure 3.47. Let LiABe be an equilateral This is possible because a Reuleaux triangle can be moved around (but
tdangle with the length of each side w. With each vertex as center,
not rotated) inside a square; hence the shape of the drill bit is a
construct the shorter arc (of radius w) joining the other two vertices. The Reuleaux triangle, and the bit moves in an eccentric path (Figure 3.50).
union of these three arcs is the boundary of a set of constant width w
called a Reulellux triangle, after an early contributor to the study of sets of
constant width. Note that a Reuleaux triangle is actually a region. rather
than a curve.
11
FIGURE 3.50
FIGURE 3.49
Some sets of constant width that are not circular regions also
do not have corner points. Let~ ABC be an equilateral triangle with
FIGURE 3.47 edge d. and ex lend each side <l distance 11 past the vertices, as in
Figure 3.51. With each vertex as center and with radius II. construct
An interesting difference between the Reuleaux triangle and a the arcs A'A", B'B", and ce". Then with each vertex. as center f!nd
circular region is that a Reuleaux triangle has corner points at A. B, with radius d + h, construct the arcs fl"B', B"C', and e" A'. The six arcs
and C. In recent years, the malhematical properties of the Reuleaux are the boundary of a set of constant width d + 2ft.
triangle have led to some extremely important applications. One such
application is in the cylinder of a Wankel engine. as shown in Figure 3A8. 8' B"
The Reuleaux tliangle inside the double elliptical chamber allows more
than one phase 'of operation of the engine to take place simultaneously.
This type engine is being used increasingly in modern automobiles. The
same type of efficient engine has been used in snowmobiles. For further
information, consult Scientific American, February, 1969.
FIGURE 3.51
Sets of constant widlh of still another type may be constructed

in a manner similar to that used for a Reuleaux triangle. Sets of this type
use regular po~ygons with an odd number of sides and are sometimes
FIGURE 3.48
CONVEXITY 119
118 CHAPTER 3
called Reule/lUX IJOlygons. A Reuleaux polygon with five sides is shown in

Figure 3.52. In general, any set of constant width may be considered
as the orthogonal trajectory of sets of intersecting lines. The cOncept of 8' A
orthogonal trajectory is illustrated in part for a Reuleaux triangle in
Figure 3.53. The orthogonal trajectory of a set of lines maybe thought
of as a cUl've intersecting each line of the set at right angles. Each
ray with endpoint A that intersects fjC does so at right angles. A FIGURE 3.54
similar statement may be made for R
using rays from Band ,.[lJ using
, The converse of this theOL'e
IS not given here. Theorem 322 am c~n also be proved, but the proof
rays from C. A body of constant width . od Its converse describe a
inc' . .
as containing I I ' .
convex
wI( out
leasmg the diameter a POints possible .h
, Consider all convex bodie .
I~ I~ easy
~n~u:;,:'t~~:' ~~!t ~e~;II:~;erb~Undary)
to find the perimeter (th: I Wlthl the same constant width w
for a circular region, of severa;
The f 11 . also the peflmeter of the R I . S 1tw. Iuterestingly
sit to oWing theorem, not proved in th~U eaux tnangle with width w
ua Ion. J5 text, describes tlle general.
FIGURE 3.p
FIGURE 3.52
THEOREM 323 B . . The perimeter of a plane
convex body of con t . t '. arbler s theorem.
Sets of constant width are still being studied 'intensively by s an Width w is nw.
leading research mathematicians. Several of the simpler _properties are
given here, but the interested reader may pursue the topic in recent Certainly, aU convex sets .th
h.ave the same measure of area ;I
the same constant width do not
professional journals. ~~cular region. 11le set with mi~im~~, set w~th maximum earea is the
corem. 'II owmg
alea IS named in th 10 .
THEOREM 3.22. If K is a convex body of constant width and if
A ~ K, then the diameter of K u
(Al is greater than the- diameter of K.
THEOREM3 .24. Blaschke L b
th with the least area is the Reu~e e esq~e theorem. The set of constant
'd
WI
aux tnangJe.
The intuitive meaning of this theorem is that other outside
points cannot be joined to a convex body of constant width without
increasing the width. The theorem probably seems self-evident, at least for but it dThe Reuleaux tdangle has some prop t'
oes not have a er les of a circul .
constant width h' center. In fact, the only c ar regIOn,
dimensi:~es
h avmg a center' Ollvex bodies f
a circular region.
sp erical regions in three clfcular regions in the plane anod
Proof: In Figure 3.54, suppose K has consta.nt width wand
~
C. One of the current areas - f. . .
A K. Let B be a point of K closest to II and let 1t be the plane' ". oncept of universal cover A .0 InvestIgation -.in convexity is the
through B perpendicular to AB.

Then" is a supporting plane of - that can be used t~ co umversal cover is defined as a plane
and BB' = IV. But then AB + BB' > BB' = w. This proof is written for other words, any set with a ~~r every set whose diameter is one
convex body in three-space, but it can easily be modified for m such a way that it is a SUbs:~:~era universal
of one can be located in th~
cover.
space.
CONVEXITY 121
120 CHAPTER 3
A set that will cover every convex set of diameter one will cover theorem and the wealth of related material constitute a su bstantinl
every set of diameter one. The smallest square universal cover is a unit portion of the recent discoveries in the geometry of convexity.
square. The smallest equilateral triangle that is a universal cover has an
indIde of diameter one. The general problem of the smallest universal THEOREM 3.25. rIelly's tlteorellI. Ld K = {K,. K1 .·:·.' KN } be
cover of any g;;ven shape has not been completely solved. N convex sets of points, N ~ It + 1. lying in 'u-space, 11- = 1. 2, or 3, so
that every J1 + 1 sets have a nonempty intersection. Then the intersection
of all the sets is not empty.
EXERCISES 3.6
Figure 3.55 is an illustration of Helly's theorem for fOlif convex.
, Find the minimum width and the diameter of each set in Exercises 1-4.
sets in a plane.
I' L Square region with edge 1 inch.
l' ! .,
l, .. 2, Rectangular region 2 inches by 3 inches.
3, Reuleaux triangle constructed on equilateral triangle with edge 1 inch.
4. Isosceles lfiangie with edges 7 inches and noncongruenl edge 3 inches.
5. Picture other nonconvex sets of constant width.
.. " 6. Draw a Reuleaux polygon of seven sides.
7. Construct a set of constant width ns in Figure 3.51, but begin with a
pentagon instead of a triangle.
8. Show that lhe perimeter of a Reuleaux triangle is 1tW. FIGURE 355
9, Find the maximum and minimum areas of plane sets of constant width
four, The proof is given for lhree dimensions. For N = 4. the
10, Find the length of the edge of the smallest equilateral triangle that is a theorem is trivially true. The proof can be completed by induction on N.
universal cover. If necessary, the reader should review the idea of mathematical induction
11. Prove Theorem 3.21. before continuing the proof. Assume the theorem true for ailY 4, ... , III
12. Give an example to show that Theorem 3.22 is not necessarily true for a convex sets of points and let K 1 •.• " Kill' K",+ 1 be convex sets, any four
nonconvex body of constant width. of which have a common point. Let K = Km ( j Km+ l' Since the theorem
13. Give an example to show [hat Theorem 3.22 is not necessarily true for a is assumed to be true for N = 5. K"" K",.t l' and any other three sets
convex body lIor of constant width. K j • K). K" have a point in common. Therefore, {K l' [(2'·.·. Km _ l • I<}
is a set of m convex sets, each four of which have a point in common.
By the induction assumption, the intersection of all these sets is not
3.7 HELLY'S THEOREM AND empty. This common point is also common to K I , .... K",. K",+l' since
APPLICATIONS each point of K is a point of both K", and Km+ I' The theorem holds
for 111 + 1 sets when it holds for up through 111 sets, and the theorem
The central theorem of this section is named for the Austrian is proved by induction.
,,"!
, mathematician Eduard Helly (1884-1943). Helly studied at the University Many applications of Helly's theorem depend on a more general
of Vienna and nt Gottillgen. His theorem was discovered in 1913 and form for a nondenumerable infinity of sets, stated here without proof.
published in 1923. Interestingly enough, I-Ielly, like Poncelet, who dis-
, , covered prqjective geometry, spent several years as a prisoner of the
Russians. If; 1938, 11e and his family moved to the United States. The THEOREM 3.26. For any collection of convex bodies in II-space,
CONVEXITY 123
122 CHAPTER 3
plane parallel to passing th
~e closed half-space H' with :::::a;he ,poi~ts
1[
of. S n Ii closest to 1[
if every Il + 1 of them have a common point, then all the ,convex
lOn 311/4 points of S but H' d Y 1[ an lying in H contains mor~
bodies have a common point. tradiction, since A was ,assumed
o eto sbenot . contain A. Th'IS IS . a con-
spaces containing more than 311/4 m t~e intersection or an half-
that the theorem does not explain h:: ~II~: po~nts. ~t should be obvious
One of the interesting applications of Helly's.theorem shows the
existence of points that behave somewhat like the center of symmetry The applications of Helly' th pOint A ,~ actually found.
of a set of points even when the set of points is not symmetric. ~nlY a few are listed here. A th:Ol'e~~rem ~re surprisingly numerous;
nvo1vmg volume, is the following t sl~l1Iar to Theorem 3.27 but
, sated WIthout proof. '
THEOREM 3.27. Let S = {P" ... ,P,} be any finite set of points.
in. space. Then there ·is a point A sud, that every closed half-space
THEOREM
having vo 328
Iume V. Then Let S be a bounded t
. . there f pomts
.
formed by a plane through A contains at least n/4 points of S. is . se 0 in space
As an illustration, suppose S consists of six points in space. The

~;:~-::a;/!ormed by a plane thrOug~ !~~~:rs~ct:ut
• III
that ev~ry closed
a set with volume
theorem says that there is a point A such that every closed half-space
formed by a plane through A contains at least 6/4 points of S. This
Helly's theorem is rathe
means each half_plane:must actually contain at least tWO points"f S, since
;~~:~t~ in~IUding the ideas of U~iv~~::II~o:~;~dd to. other geometric
;h~ 'l:s~~I~!e~o::lIal~~~Ct~i::~:Ittonal
Attemptin~ t:'::::f: ~~~r:s t~
it contains more than one. It is also important to point out that A does
not have to be a member of S. illustrate
understanding.
Proof: Since S is a finite set of points, there exists a spherical
I THEOREM 3.29. If each thre f .
enchosed In a circle of radius one th e ~I n POints in a plane can be
region B containing all the points P,. Consider the set of all closed
half-spaces containing more than 3n/4 points of S. Let H" H", H" H. sue a Circle. . en a n POlllts can be encfosed in
be any four of these closed half-spaces. In the study of set theory, it is
. The proof consists of sho wmg. .
proved that de~lred circle) exists that is not m that a point (the center of the
POlDtS. A typical figure' h _. o:e than one unit from each of tI,e
(H, n fl, n H, n H.), = H,' u H,' u fl,' v H;, JS sown 111 FIgure 3.56.
where H,' is the complement of H,. Since each Hi contains less than n/4
points of S, H,' v H,' v H; v H; does not contain all n of tlie points of
s. Then H1 n H 2 ("'\ Ii) {"\ H4 must contain at least one point of S.
Since every four of the closed half-spaces have a common point

of S. Helly's theorem (in the form of Theorem 3.26) can be applied to the
intersections of these half-spaces with B to conclude there' is a point A
common to all such half-spaces. FIGURE 3.56
This point A is the desired point in the theorem. If it were not,
then there would be a plane n through A that is the boundary of a Let a circle with center X b .
closed half-space containing less than 11/4 points of S. The opposite . 11 points, A, B, C. The three circles e ~tlumt circle enclosing three of the
. Wi 1 centers ABC, . aII contain
. X.
open half-space H will contain more than 311/4 points of S. Let n' be the'
CHAPTER 4
124 CHAPTER 3
By I-Ielly's theorem, since any three of the unit circles with the given EUCLIDEAN GEOMETRY OF THE
points as center have a nonempty intersection, all of the circles have a
nonempty intersection. A point in this common intersection will serve as
POLYGON AND CIRCLE
the center of a unit circle enclosing all the points.
THEOREM 3.30. lung's theorem. For 11 points of the plane, such

that each two are less than one unit apart, alt can be enclosed in a
circle of radius -/3/3.
THEOREM 3.31. Blaschke's theorem. Every bounded convex

figure of constant width onc contains a circle of radius 1/3.
THEOREM 3.32. Given 11 parallel line segments in the same

plane. if there exists a line that intersects each three of them, then there
is a line intersecting all the line segments.
The ge~metry of convexity is a rapidly expanding area of

modern geometry. Many of the familiar figures of Euclidean geometry
have been shown to possess new properties when they are considered
as convex sets. The chapter is deliberately open-ended in the hope that
the reader will want to seek recent research articles on convexity.
EXERCISES 3.7
1. Draw a picture with six convex sets in a plane to illustrate Helly's theorem.
2. Give an example in one dimension to show that Helly's theorem does llOt 4_1 FUNDAMENTAL CONCEPTS AND
hold for nonconvex sets. THEOREMS
3. Give an example in two dimensions to show that Helly's theorem does not
hold for nonconvex sets. In Chapters 2 and 3, much of the material of Euclidean geometry
4. Prove I-lelly's theorem for one dimension. was studied from two ditl'erent points of view. Chapter 2 emphasized
5. Prove Helly's theorem for two dimensions. the idea of Euclidean geometry as a study of t~:.i.X::':~~!!l~.Lp.r.Qp~[ties
6. For the eight vertices of a cube, does Theorem 3.27 apply for A thl~ of sets oIY~~~:~g!:.r Etichdean ~ similari!L.t.!...~~f.QLI!lru.i9~S. Chapter 3
'conc;ent'rated on the concept of convexity, one of the properties preserved
center of the cube?
by similarity transformations (as well as by somewhat more gen~ral
7. Stale and prove a theorem for the plane that is analogolls to 111eorem 3.27.
transformations). The present chapter begins with a survey of some
8. Verify Theorem 3.31 for a Reuleaux triangle of constant width one. fundamental concepts and theorems conc~rning polygons (particularly
9. Draw a picture to verify Theorem 3.32. triangles) and circles-concepts and theorems that have long been a part
125
to. Prove Theorem 3.28.
EUCLIDEAN G EOMETAY
- OF THE
126 CHAPTER 4 . POLYGON AND THE CIRCLE
of traditional Euclidean geometry. Then it progresses rapidly to Ithe vel:!lces, it is called th . 127
~.CJr~umcenter to show that °t '
circilnidi'di-tlic.'u -.
wealth of modern material that has been discovered since 1800 a~out In F'. mque circle containing all th' I IS the center of the
these basic figures of syntl]et).c Euclidean geometry (geometry thai does fl' 2 trl' angIe A'B'G" ,
tgure 4., lee vertices 0 f a Inangl
. '
o t le SIdes of the ad . . IS lormed by joinin . e.
not use coordinates). The last seclion includes three -;;;gtiificant ·';PP!"i;;··
~f E~c1idean
points of two s'd gmal tnangle. Since the segn .g the midpoints
tic",;s ihe geometry of the polygon and cirHe-the go'idell I es of a tri f lents Joinin g th e
perpendicular bisectors oflb a;g e are parallel to the third 'd mid·
ratio, tessellations. and caroms. tothesidesoftriauge I eSJ esoftriangleABC 'I 51 e, the
It is essential that the student of modern college geon~etry
I
ABIC'. Th'IS means that ;po areasoperpe
. n d'lcular.
understand several useful key concepts and theorems from elemelltary IS perpendicular to
geometry. Among these are the concurrence theorems, sometimes fbund

A
in high school geometry texts, that identify four significant points!con-
nected with a triangle.
THEOREM 4.1. The perpendicular bisectors ~f lhe sides of a

triangle are concurrent at a point called the circumcenter· ' !
C'---~l..........-":~
A' B
Theorem 4.1 is i11ustrated in Figure 4.1. It is significant :when
FIGURE 4,2
any three Hnes have a common· point in a geometry. because any
three lines generally intersect by pairs to determine thr1e distinct'points, li'C and so 0 n.. t h us, the per . .
ABC are the altit . pelldlCular bisectors of .
to the cl . udes of tnangle A'E'G' R . the s.des of triangle
bisectors :~ththat'dthe altitudes of a 'tri::geJ:JIl~ this statement leads
e 51 es of a I • ale the p .
(by Theorem 4. I). oat ler triangle and I erpendtcular
leDce are concurrent
. THEOREM42 T .
pomt called the orthocente,..
.. he altitudes of a triangle are concurrent at a
Figure 43 h
. sows triangle ABC a .
od Its orthocenter H . Th e lOur
'
FIGURE 4.1
A
rather than a single one. TIle proof of Theo~em 4.1 depends on s!lowing
~ the point of intersection of two of the perpendicular bisect~rs. say
B'O and 1':'0 in Figure 4.1, also lies on the perpendicular bisecto~ of the
third side. Point 0 is equidistant from A and C because it is on
Eo; . it is also equidistant from A and B because it is ·on Co.
Therefore. 0 is equidistant from Band C on the perpendicular bisector C~--..r;JtL.._::::1
o B
--_..
of SC. Since 0 is the point in a triangle equidistant.. from the--- three .
FIGURE 4.3
,I
128 CHAPTER 4 EUCLIDEAN GEOMETRY OF THE POLYGON AND THE CIRCLE 129
points A. B. C, and H constitute an orrhocenlric set of four points, so So far, four points of concurrency have been introduced. It is
named because each of the four points is the ortiW-center of the triangle natural to wonder whether the four new points constitute a significant
formed by the other three. set of points in their own right. The matter will be discussed in
Two additional theorems about concurrency involve the internal Section 4.3, but you may proHt from conjecturing about the location of
angle bisectors and the medians of a triangle. these points now.
lvfuch of the study of the triangle in Euclidean geometry involves
work with proportions, and ordinarily this concept is related to similar
THEOREM.4.3. The internal bisectors of the angles of a triangle
triangles. A basic proportion used as a tool in college geometry is the
meet at a point called the i~.~cent:T'
property that is connected with internal angle bisectors of the a~gles of a .
triangle, stated as Theorem 4.5.
The proof of Theorem 4.3 (the details are left as Exercise 7,
Exercise Set 4.1) depends on the fact that every point on the internal
THEOREM 4.5. ·The internal bise~tors of all angle of a triangle
bisector of an angle is equidistant from the adjacent sides of the angle.
divide the opposite side into two segments proportional to the adjacent
For example, in Figure 4.4a, if 1 is on the angle bisector of angle B.
sides of the triangle.
then lY =: lX. Since 1 is equidistant from all three sides of the triangle,
it is the center of the it/circle. a circle inscribed in the triangle. This
means thif1he three s{des -~(ii~e triangle are tangent to the incircle. In Figure 4.5, assume that AD is the internal angle bisector
,.:. ---==-.... of angle A of triangle ABC and thut CE is parallel to AD. Because
;
of the parallelism, L EC A ;;;:: L CAD;;;;; L C EA. This means that triangle
, \' A
EC A is iS0sceles and that EA ;;;; AC. Now think of lhe figure as formed
~- ~+.;:.-\- A ~
"-~'? by two transversals from B intersecting a paif of parallel lines, EC

~ .
and AD. so that the foHowing is a true proportion:
.~ "
x
CD DB CD DB
-~-
£;1 AB or CA ~ AB
so that
C~----~----~B
A' .
(,) (b)
FIGURE 4.4 as was to be established. Note lh<lt directed segments are nol employed
in this development.
THEOREM 4.4. The medians of • triangle meet at a point
called the centroid. E
~,
\ "
\ ' ..... A
\~B
The centr~!(tis.Jh.~. cel1~~': of gravity for a triangle. Recall that
Qledians join '"th;- vertex. and the· midpoi-;;r" of the opposite side of a
triangle. In Figure' 4.4b, .t"i:ia-ngles C BG··"'and GC' B' aFe similar with a
,.! ratio of similarity bf two to one. The completion of the details of the CD'
FIGURE 4.5
proof is left as an exercise. . . /[1:,
' ..
!. . ::---.p r\ _.
EUCLIDEAN GEOMETRY OF THE POLYGON AND THE CIRCLE 131
130 CHAPTER 4 I lift
for/externql bisectors
Ias a su rprlse etoIrs sections prior to 200 B.C. Assume. using the notation of Figure 4.7a. that
The proof of t he an alogous 1 theorem -
metimes comes PAIPB = c, a given constant. It must be shown that the set of aU
. Set 4 1 t so ! 1 f a tnang e
a:u!:~~:i:~ ~~~!:;:yCI;~at 'e~te:~:~:i~~C:~;si::e:~:1 e~i~ectorsi
locations for point P is a circle. There are two points on AB whose
~
the Cor- distances from A and B are the correct ratio without regard to directed
5 .
have properttes COlTespondmg •
to .
t r for example, IS the concept
_:
0f distances. These are indicated by points C and D in the figure. Then. in
h concept of mcen e t triangle APB, PC and PD are internaJ and external angle .bisectors of the
responding to t e t of Theorem 4.6.
excenter given in the statemcn ! angle at P. This can be shown for point C. for exampJe. since
. ftwo angles ofa!ttiangie AC PA

T HEOREM 4.6. The external ~\Sectorsl at a point call¢d the
0
-=-'
. f the third ang e CB PB
meet the I'nlernal bisector 0
excel1ter. ,i
, . il one
al bisectors of a tnang e.
Ofcourse. when talki~gabout:.~t~~n consider the sides I~S lines
must think of the sides a;i :~:e~~:
assume that £ lis the exc:~::~iS~~
I ather than segments. In g of of Theorem 4.6, llift as an £1 . the
arent in the pro h triangle' hence! :.
~::~~~: ::'~distant-- ternally
from all three ~i~:S ~: \h~ee si~es' of the Itriangle.
15
(,)
(b)
.' '-I. - - - '.- tangen .' FIGURE 47
center of a clrcle ex . I f the given trtangle ..
This circle .IS one 0 f tl1e ('"c"'C es 0 ...
But the internal and external angle' bisectors at a jvertex of a triangle
1
are perpendicular, so LCPD is a right angle. Triangle CPD is
inscribed in a semicircle. so P lies on a circle with CD as diameter.
Figure 4.7b shows a specific example of the cirde of ApolJonius,
for which P'A'/P'B' = 5/3. Each point 011 the circle is 5/3 as far from A'
as it is from B', and the points on the circle are the only points in the
plane with this property.
In the proof of Theorem 4.7 and elsewhere. note that even
FIGURE 46 though parallelism (perpendicularity) is a relationship usually defined for
· tors-the lines, use of the concept is extended to segments and rays without
d external angIe bIsec
One other property of inten~a~ a~2}-is needed to help prove the difficulty. For example, although it is technically imprecise to say that
. endicular tExel else
-
fact that they are pel p . AB is parallel (perpendicular) to CD. that expressioll is understood (0
! mean that lB. of which AB is a subset. is parallel (perpendicular) to
foHowing theorem. -
CD. of which CD is a subset.
a oint P to two fixed Two final theorems about the segments related to a circle are
THEOREM 4. 7. . If I the distancesr from P s for the point P
\I location somewhat connected to the previous theorem and also are useful in
. I gIven ratio, t le n the set 0 a proving more advanced theorems.
pOll1lS lave a . I ,cApolloni"s.
is a circle. ca IIe d the eire e OJ___
. ~ __ _.
the Greek mathe- THEOREM 4.8. Ifa quadrilateral is inscribed in a circle. then the
. ollonius is named a~ter
The circle of Ap t comprehensive treatment on conic, ',OI'p(lSJI.e angles are supplementary.
matician Apo llonius. who WfO e a
.~.
132 CHAPTE R 4
AC' A.D = (AB)2. Recall again thal the notation Ae indicates a number,
The term cyclic quadrilateral is also used for a quadrilateral the m(~asure of AC. Triangles ABC and ADB cue similar because
inscribed in a circle. Study F;g;;r~ 4.8. You should recall that the corresponding angles are congruent; therefore, AC/AB = AB/AD. from
measure of an inscribed angle in a circle is half that of its int~rcepted
which the theorem follows.
a~c. For example, the measure of angle B is half that of arc ADC.
Since angles Band D together intercept arcs with a total measurement
of 2u degrees, the angles are supplementary since the sum of their EXERCISES 4.1
measures is 1£.
A 1. Use the notation of Figure 4.3 to name all of the triangles wh~se
verlices are three or the given points and whose orthocenter is the fourth
point.
2. Where is the orthocenter of a righllriangle?
3. Under what conditions would the orthocenter of a triangle lie outside the
triangular region?
c 4. How many excenlers does a triangle have'!
5. Could antiparnllel segments also be parallel?
o 6. Prove that the segment joining t~e midpoints of twO sides of a triangle is
FIGURE 4.8
parallel to the third side.
7. Complete the proof of Theorem 4.3.
Opposite 'iides of an inscribed quadrilateral are sometimes called
8. Prove tlUlt the angle between the segments from the incenter to two
GlIlil'arallei with respect to the remaining pair of sides. The prefix Vi,rtices of a triangle has a measure equal to 90 plus one~half the
I "anti" suggests across from or opposite. In Figure 4.8, if angles D and A measure of the angle of the triangle a't the third vertex.
were supplementary. then DC and AS would be parallel. But instead, it is
\ the angle opposite or across from A that is supplementary to D. hence the
,
I
segments are antiparallel instead of parallel.
to. Prove that the exlernal bisector of an angle or a triangle divides the
opposite side (externally) into two segments proportional to the adjacent sides
of the triangle.
THEOREM 4.9.111e product of the lengths of the segments from I L Prove Theorem 4.6.
an exterior point to the points of intersection of a secant with a \ 12. Prove that the internal and external angle bisectors at a vertex of a triangle
circle is equal to the square of the length of the tangent from the are perpendicular.
point to the circle. n. Prove the converse of 111eorem 4.8.
14. Prove lhat two vertices of a triangle and the feet of the allitudes to the
Using the notation of Figure 4.9, the theorem says that sides adjacent to the third vertex can be inscribed in a circle. {The feet are
the points of intersection of the altitude with the opposite side of lhe
lriangle.)
c 0 4.2 SOME THEOREMS LEADING TO

A •
0
MODERN SYNTHETIC GEOMETRY
£
F In surveying the large body of geometric material developed
prior to the modern era, it becomes obvimis that some of the earlier
FIGURE 4.9
-.'1
EUCLIDEAN GEOMETRY
134 CHAPTER 4 OF THE POLYGON AND THE CIRCLE
135
.
theorems are more important than others in leading 1
'
to. later develop- CI CF (considered negative
ments. Several of these still-significant ideas are discussed in this section. IB = BE because I divides BH BE
Two theorems usually studied together, though proved in dHif~rent CB externally) HA = AD
ages, have great significance because they concern only collinearitiyiand Then
concurrence and hence are of value in the study of projective geometry
when distance is no longer an invariant. The first of these. Menelaus' _ AD CF BE
theorem, is credited to the Greek mathematician Menelaus, who lived in CF . BE' AD = :- I.
Alexandria about 100 B.C.

Note Ihat the three .
vertex (say A) and . ratIOs may be named b . i...
THEOREM 4.10. Menelatls' theorem. If three points, one onleach clockwise d' . proceedmg around the t.' y startmg at a
"ectlOn) . Ii lIangle (s .
side of a triangle (extended if necessary) are collinear, "then the product, Ihe next vertex, th~ng~~'~1
rom a vert."x to a point (,f :;;V:~i
a counter-
of the ratios of division of the sides by the points is negative one same patter le next pOInt of di . . on, then to
. n may be used wh VIsIon, and so .
(_ t), if internal ratios are considered positive and e,xternal ratiCjs are as In Figure 4.11'• en aU three points of d·1",JSIOI1
.. . are externaJ,
011. ThIS
considered negative. i
. I
; AE.CF BD
A AC FB'DA =-1.
o
E
A
c··L-______________JL ____~t'
8 C~ _ _~_~:::::::".
B F
In case the line contai . '

Suppose; using the notation of Figure 4.lO, that points G. H. I vertex, the state ~ fling the three .
are the three given collinear points, one on each side of the triangle the del1~mjnato~::~ ~f the. theorem can be mOd;;~~t: goes through a
A Be. The following pairs of similar triangles can be observed:
The xerClse 3, Exercise Set 4 2) 0 aVOId.a .zero in
converse of M )' . .
for proving tlla t t Jlree pointsene
areaus theorem PI'OVJ'd es an important tool
eoUine
f',.ADG - f',.CFG, or.
f',.BEI - f',.CFI.
THEOREM 4.11. Converse of M
f',.BEH - f',.ADH. sidel:f
of t l '
r:tI~r~ Ofldivision of the Sj~es ~ylle:~'rt'er' rheo:·em. If the product
from the three pairs of similar triangles are derived the three pro- three p . nde d 1'f necessa _ . e pomts• one On each
lang e,exte
OInts are coIlinear. ' I y, IS negative one• IIlen the .
portions. since pairs of corresponding sides of similar triangles have the
same ratio. Also recall that the notation AG indicates the measure of AG,
which is a number. i Suppose, using the notation of fjgure 4.1 I,
that D, E, F are
three points on each side of the triangle and that the product of the THEOREM 4.13. Ceva's theorem and irs converse. Three lilles
~
ratios of division is negative one. Suppose, furthermore, that EF meets that join three points, one on each side of a triangle, to the opposite
An at a distinct point D'. It must be shown that D' = D. Using vertices are concurrent if and only if the product of the ratios of
Menelaus' theorem for E. F. and V', division of the sides is one.
AE CF BD'
l. The proof of the direct theorem is given here. The form of the
AC'FB'D'A=
theorem can be written with equal products rather than ra~ios lo avoid
But this means zero in the denominator if one of the three points coincides with a
vertex. Using the notation of Figure 4.12, Menelaus' theorem applied to "
!l
AE CF BD' AE CF BD <->
triangle ABD and CF shows that .
AG DC BF
or GD·CD··FA=-1. ,.
"
Since BA = BD' + D' A or BD + DA,
BD' + D'A BD + DA
D'A DA
and
D'A= DA; ~
For triangle ACD and BE,
therefore, D and D' are the same point, which means that D, E. Fare
collinear. " AE CB DG
EC' BD' GA =-1.
Theorem 4.11 can be used to prove several other theorems, one
of which is stated below but proved as an exercise.
MUltiplying left and right members of these equations results in
AG DC BF AE CB DG~
THEOREfv14. 12. The internal bisectors of two angles of a triangle -'-'-'-'-'-= I
GD CB FA EC BD GA
and the external bisector of the third angle intersect the opposite sides of
the triangle in three collinear points, or
AE CD BF
The theorem often paired with the theorem of Menelaus, Ceva's EC . DB . FA = t.
theorem, was discovered about 1678 by the Italian mathematician Ceva.
Menelaus' theorem deals with three collinear points, whereas the theorem Note thtlt the three ratios can be written quickly by following
of Ceva concerns three concurrent lines. the same pattern described for the theorem of Menelaus: start <It a vertex,
138 CHAPTER 4
example, credited to the E r
. , so on. It should
go to the point of division, then the next vertex, anb 1768), refers to a triangle an~g ~Sh ~athe~atician Robert Simson
pomt on Its circumcircle. (1687-
also be noted that similar triangles may be used to prove' Ceva's
theorem directly. without using the theorem of Menelaus (EXrrdSe 8,
. THEOREM 4.15. The three .
Exercise Set 4.2). . ' clfc.umcircle to. the sides of a ive p~rpend~culars from a point on the
The converse of Ceva's theorem, proved indirectly in a way colhnear points The r g n tflangle mtersect the sid . I
similar to Theorem 4.11, is a useful tool for provibg three lines con- Simson line. . me On which the three pomts
. " IS es
he 10 t 1ree
called the
current. For example, it can now be· used to give1very simpte proofs
that the medians and the internal bisectors of the angles of a trHmgle afe
. Use the notation of Fi
concurrent (see Exercises 9-11, Exercise Set 4.2). CIrcumcircle and'D" th ,eet
' gure
A more modern application of the converse of Ceva's theorem in f E Fe of t1 4.14, with P the p 010. t on the
o Theorem 4 8 poi t P Ie perpendiculars By th
ordinary Euclidean geometry is in the proof of the foHowing theorem: Similarly P A . , n 5 • D, A. E are c c l ' : n e converse
I~ (~e
third circle.' . C. Bare 011 another circle, and YP • . . Foare aoncircle).
yet a
THEOREM 4.14. The segments from the vertices of a vriangle to
the points of tangency of the incircle afe concurrent: In circle P, D, A, E LPDE ~ LPAE
LPAE ~ LPAC
This theorem was proved in the early nineteenth -century by
In circle p. A, C, B LPAC ~ LPBC
J. D. Gergo , a French mathematician, and the ,point of ooncurrency
nne
is called the Gergonne point. : I LPBC ~ LPBF
Using the notation of Figure 4. t3, since the two tangdnts from a
. In circle P ' D.B, F LPBF ~ LPDF
point to a circle are congruent, AE = AF, CE = CD, BD = 'BF, and
;OIOII~e this establishes the fact that L P DE ~ LPDF, points D• E, F are
lOear.
AE CD BF
EC·DB·FA=1.
so the three segments are concurrent by the converse of Ceva's

theorem.
FIGURE 4.14
FIGURE 4.13
For a given point P
determined . TIIe speCial
. case onh the circumcircleI one S'Imson Ii
Many other theorems about concurrency and coll~nearity have
. .
sJdered in the e '
XerCJses. w en P. coincides WJt con~
. h a vertex is ne IS
interested geometers throughout the history of geometry. One additional

EXERCISES 4,2 point of concurrency of three segments from the vertices. Use Ceva's
theorem to write three ratios whose product is equal !t.l one.
For Exercises I and 2, lise figure 4.15. For the triangle and the transversal 8. Prove Ceva's theorem directly, without using the theorem of Mel1ehlllS.
named in the excl-cise, write the product of the three ratios equal to negative one
by the theorem of Menelaus. 9. Use Theorem 4.13 to prove that the medians of a lriangle are concurrent.
o 10. Use Theorem 4.13 to prove that the internal bisectors of the angh!s of a
I
J F triangle are concurrent.
I,
Use Theorem 4.13 to prove that the external bisectors of two angles of II
triangle and the internal bisectors of the third angle are concurrenL
E
l2. Describe the location of the Simson line if the point on the circumcircle
is a vertex of the triangle.
AL---------~--~~C 4,3 THE NINE-POINT CIRCLE AND

B
fiGURE 4,15 EARLY NINETEENTH CENTURY
SYNTHETIC GEOMETRY
-'I
,0
L Triangle ADF and transversal E, D. C.
Renewed interest in dramatic extensions of the classical geometry
2. Triangle BCD and transversal A. e. F. of the triangle and the circle began in the. early nineteenth century.
3. Write nn alternative form of the theorem of Menelaus showing the product Probably the most significant of the advances at this time was the concept
of three segments equal to the negative of the product of three other segments.
oJ the nine-point circle. sometimes (:redited to the Gennan mathematician
What is the value of the proJuct if one of the points of intersection
coincides with a vertex of the triangle'l Feuerbach ih 1822.
5. Use the converse of Menelaus' theorem to prove that the external bisectors THEOREM 4.16. The midpoints of the sides of a triangle, the
of the angles of a triangle meet the opposite sides of the triangle in three: points of intersection of the altitudes and the sides, and the midpoints
collinear pohltS. of the segments joining the orthocenler and the vertices all tie on a circle
6. Altogether there are six bisectors of the angles of a triangle, three internal called the nine-poil!t Circle.
and three ex:ternal. Prove that these six: bisectors meet the opposite sides in six
poiltts that are on four lines, three on each line. (Compare this with -a
finite geometry studied in Chapter 1.) A nine-point circle is shown in Figure 4.17. 1n general. the
7. In Figure 4.16, tet A. B. C be lhe vertices of the triangle, with r. the circle intersects each side of the triangle in two distinct points. A
A /''>,
c'
\
E ._.' ,1 ".
--n
~" .
c0() r\ \
~.
. J
\f
l/-}·.
,.;, t.'f'
'1'\-' _.
./..'
!
fJ
C~~--------~----~F C~--~~~~----~B c
8 .D !:4'
.,;
"
J
~..--" _. -
11)9" ~<,-j!'!-1i-r-fc 7/
142 CHAPTER 4
. f ,1:'/
.£' i§'.{· \
k:_1
F
.,),
11' fI'
'- c Fn d7f/1. t:, 2
EUCLIDEAN GEOMETRY OF T.HE POLYGON AND . THE
CIRCLE 143
~llO~s thi~'at;er~: ~
G \ )
The proof of the theorem . ...

a. There is a circle through At. B'. C'. the midpoihts of the sides.
b. Show that points D. E, F are on this same circle.
c. ShoW that points G, H.I are on this same circle.
To prove that D is on the same circle as At, B', C. ~t is
possible to show that DB'e' A' is an isosceles trapezoid and hence,
inscribed in the circle. A'e' ~ DB' because AtC' connects the midpoints
of the sides of triangle A BC and therefore has a measure equal to, half
the 111easure of the base and 'i51f connects the vertex and the midpoint
of the hypotenuse of right triangle ADC. which implies that DB is hllilf the
'I FIGURE 4,18
hypotenuse.
Now consider the circle with I A' as diameter. point 8' is on
lhis circle. since angle I B' A' is a right angle; point C' is also on the . THEOREM 418 . . The center of th .
POInt of the segment whose endpoints . e nme-point circle is the mid-
circle. Therefore I lies on the same circle as A'B'C'.
center of a triangle. (Se e F'Igure 4.19.) al e the orthocenter and tIle CJrcuITI-
.
There is a wealth of additional information tabout the nine~

point circle and its relationships with other sets of points in Euclidean From Theorem 4.17 EH - - ,
geometry. One of the important ideas is the location 0,
the cente~ of the ?
parallel.. If 0' is the center 'of th = A • a~d th,e two segments afe also
nine-point circle in reference to several other points previously mentioned. off the dIameter n . Bllt OA'H EeisGme-pomt II Circle, it is the m·d .
I POlOt
This information depends 011 a theorem established by the G~rman Of 08 paraHelogram bisect each athe; ~tara elogram. and the diagonals
o H. . liS means Ihal 0' is Ihe fil·d POUlt
.
mathematician Euler in 1765.
THEOREM 4.17. The centroid ofa triangle trisects the 'fg01ent A

,I
joining the drcumcenter and the orlhoce nter: '
The line containing [he three points is called the Euler line.
"Trisect" as used here means that the distance along the line frbm the
circUlUcenter to the centroid is one~third of the distance along the same
line from the circumcenter to the orthocenter.
In Figure 4.18, let 0 be the circumcenter, G the centroid. and H
the orthocenter. The measure of AH is twice that of OA'. This is true
because tdangles ClJI and COA' are similar with a ratio of simjlarity of
2 to l, and IB = AH because ARB! is a parallelogram. Now ~riangles FIGURE 4.19
GH
GOA' and GHA are similar with a ratio of 1 to 2; therefore OG 1= 1 ,
AU the theorems about concur' .
and G trisects 0 H. . t~e .concurrence of three lines M' I~nce conSIdered so far relate to
It may be surprising to find that the Euler line also contains slgmficant in part beca use it . . Iquei s theorem, proved in 1838 '
eifel conSIders the • IS
the center of the nine-point circle, as indica led in the following es associated with any triangle. concurrence of sets of three
theorem.
144 CHAPTER 4 3. Prove that the segment connecting the vertex of the righl angle of a right
triangle wilh the midpoint of the hypntenuse has a measure half Ihat of the
i
.I THEOREM 4.19. If three points are chosen, one on each side
hypotenuse.
of a triangle, then the three circles determined by a vertex and the two
4. Prove that IB' A' is a right angle,
\,. points on the adjacent sides meet at a point called the M iquef point. 5. Complete the details of the proof ofTheorem 4.17 not included in the text.
\ 6. Prove that the radius of the nine-point circle is half that of the circum-
Using the notation of Figure 4.20, let D. E. F be the arbitrary
circh~.
points on the sides of triangle ABC. Suppose that circles with centers 7. Prove that the four triapgles formed by the points of an orthocentric
grm,p of points have the same nine-point circle. .
A 8. a. Draw a fignre showing an example in which the Miquel point is ontside
b. the triangle.
Modify the proof as necessary so that the three circles are conennent.
9. Show that, on the Euler line, the centroid and the orthocenter divide
internally and externally in the same ratiO the segment whose endpoints are
the cirCUOlcenter and the center of the nine-point circle.
to. Prove that the Miquel point is a point on the circumcircle if Ihe three points
on the sides of the triangle are collinear. s
J 11. Prove that if the Miqoel point is on the circnmcircle, then the three poinl
FIGURE 4.20
Oil the sides of the tdangl e (Ire collinear.
I and J intersect at point G. In circle J, LEGD and LECD are

supplementary, and in circle l, LFGD and LDBA are supplementary. In 4.4 ISOGONAL CONJUGATES
the rest of the proof, the notation m means the measure of the angle.
In this section and the next, three major contributions to the
\, Since
mLEGD + mLDGF + mLEGF ~ 360.
synthetic geometry of the triangle, all made within the past century, are
discussed. The first of these, credited to Lemoine in about 1873, has to do
180 _ 1l1LC + 180 - mLB + mLEGF ~ 360, with the concept of syrnmedians.
Consider Figure 4.21. If AE is Ihe bisector of angle A and if
\
mLEGF ~ nlLC + 1l1LB = 180 - mLA. LDAE", LFAE, then AD and AF are called isogoliO / (illes and one is
\ This means that LA and LEGF are supplementary, so A. F. G, E are

on a circle and all three of the circles are concurrent. It is possible that
A
the Miquel point could be outside lhe triangle, in which case the proof must
be modified slightly.
EXERCISES 4.3
FIGURE 4.21
l. A lriangle has how many: c. Mique! points? called the isogo"O/ conjugate of the other. 111e bisector of the angle is the
li. Nine.point circles? b. Euler lines?
bisector of the angle between twO isogonal conjugates. Nole that,
In the proof of Theorem 4. t 6 (for Exercises 2-4),
2. Prove that DB'C'A' is an isosceles trapezoid.
nts into which a symmedian divides the

7 Prove that the ratio of the segme . f the squares of the measures of inscribed angle, measured by haJf the intercepted arc. Circle 3 is tangent i):!
. ' I ' d of a triangle equals the rallo 0 'Ii

OppOSI e 51 e to AS al B, and BD subtends L DC B equal 10 L ABD, so this circle
the adjacent sides. ' . the iso onal conjugates of the also passes through D. The order chosen was counterclockwise, and
Prove that the altitudes of a tn~ngle me g point Dis caIJed the first or positive Brocard point.
8. circumradii to the vertices of the triangle. . h
In general, !llenL,aI~ JJ."'5?.. ~~.jstin~~~.JiTocar~ points, since the
Use Theorem 4.20 and Exercise 8 to give an altern~tive proof that t e order around the triangle can be eith'c·t clock wise or cQuntercJocicwise.
9. altitudes of a triangle are concurrent.
Furthermore, it can be proved that the two Brocal-d points are isogonal
conjugates of each other.
4.5 RECENT SYNTHETIC GEOMETRY

OF THE TRIANGLE DEFINITION. Points are isogonal conjugates if one is Ihe
intersection of lines that are isogonal conjugates of lines intersecting
A second major contribution to. the re~e,I:~l:e~::~;l, :~r~~:' at the other point
d with the BrocaI'd pomts an ,
triangle has to 0 d some theorems about these
named after Henri Brocard'fwh ~rovleee lth century. Like the Miquel
hO
Yet another theorem shows a connection between previous
. . I I'st part 0 t e 11Ine 1 . f
toptCS dUring t le cl • I oints of interst:{ctton 0 theorems about symmedians and the new theorem about the BrocaI'd
point, the BrocaI'd points are defined a~ specla P i points.
three circles. -
. I iv'th one side ofdera THEOREM 4.25. The Brocard points are on the Brocard circle.
THEOREM 4.24. The three Circles, eaCl .I l
. e I 'd and tangent to the adjacent sJde, taken lfl or
tnallgdl ahs fiagCU;~' me~t in a point called a Bl'ocard poillt. DEFINITION. The Bmcard circle is a circle whose diameter
aroun t e ,
has as endpoints the circumcenter and the symmedian point of a triangle.
·
In F 19ure 425
. , circle 1_ is tangent to Ac. drcle 2 is tangent to
Be. and circle 3 is tangent to AB. Tn Figure 4.26, let 0 and S be the circumcenter and the
symmedian point, -and Jet the perpendicular bisectors of the sides of the
c·=-'-----'!'-__-=~ 8
o
FIGURE 4,26
FIGURE 4.25
. triangle meet the Brocard circle again at Di. E', and P. It can be shown
Assume that circles I and 2 meet at D. L ABD ;l; LeAD that the lines AF'. BE', and CD' are concurrent at a Brocard point
.J like
_ L BCD, since Ihe angle between a tangent and a chord 1'1. Because right angles are inscribed in a semicircJe, SD' is
!,
153
EUCLIDEAN GEOMETRY OF THE POLYGON AND THE CIRCLE
152 CHAPTER 4
'I"ll less than 60 and X + Y -I- Z = 120. Points

,,. parallel to lIC, SF' is parallel to AC, and SE' is parallel to All, of the angles are e, '- I
' are determined by extending the sides of these lsosce es
: This means that the distances from S to the sides of the original A, B, ('
triangle are equal to the distances along the perpendicular bisectors triangle!i.
from the points D', E', F' to points D, E, F.
Beeanse of Theorein 4.23, A
A
DD' EE' FF'
CD EB FA
so that 6CDD' - 6BEE' - 6AFF'. As a result,
LBCD';,; LeAF';,; LABE'.
A comparison with Figure 4.25 shows that the three sides of the angles B
meet at a Braeard polnt G. Also, since LAF'F;,; LGF'O ;,; GD'O,
c (b)
_,1 (,)
,.' points F', D'. 0, G all lie on a circle, the Brocard circle, as was to be FIGURE 4.27
proved. For an equilateral triangle, the circumcenter and symmedi.an
"
point coincide so that the Brocard circle reduces to a single point. The measures of other angles at points D. E. F. . 'can be
The second major topic in this section is a theorem discovered de t .
. 'mformatton.
ed from the given . A ccording to ExerCise 8 0 f
about 1899 by Frank Morley. father of the author Christopher Morley_ er~m S t 4 1 the angle formed by the lines containing the segments
Its significance is based on the fact that it concerns trisectors of E~xerClse e ., 1 If J sure of
. t t two vertices is 90 plus one-la t le mea
from an mcen er 0
angles rather than bisectors. Angle bisectors of a triangle meet at a point,
the angle at the third vertex. In triangle C[ B"
but it seems that mathematicians have only recently considered what
happens to sets of three adjacent trisectors. "'LeIB = 180 - 2Z,
whereas "'LeDB = 360 - (X + Y + 2Z + 60).

THEOREM 4.26, Morley's tileoreln, The adjacent trisectors of the
angles of a triangle are concurrent by pairs at the vertices of an 360 _ (X + y + 2Z + 60) = 360 - (X + y + Z) - Z - 60
equilateral triangle. = 360 'C' 120 - Z - 60
= 180 - Z
Before reading further, you will find it profitable to use a
protractor and carefully draw several triangles of various shapes to = 90 + ~(180 - 2Z)
verify lhe reasonableness of Morley's theorem.
From this discussion, along with Exercise 8 of Exercise S~l 4.1,
Since 1900, many different proofs have been given for this
theorem, but the one here is an indirect approach, starting with an point D is the incenter of 6CIB. Similarly. F:-is the incenter of 6(lG~
. I . t r of bAHB Thus the three angles at ale
elJuilalenll lriangle Hnt! ending with lhe original given triangle.
Figure 4.27a shows what is lTIeant by adjacent trisectors. For
and E 1S t le lan;:nt~e three at 'A
congruent as
and
at B. These angles have
example, CD and BD are a pair of adjacent trisectors. In Figure 4.27b, measures

assume that DDEF is an equilateral triangle. Isosceles triangles DGE. 60 - y, 60 - Z. 60 - X.
DHF, and FIE can be determined such that the measures X. Y. Z
r'
I'
f
EUCLIDEAN GEOMETRY CF THE POLYGON AND THE CIRCLE 155
154 CHAPTER 4
The base angles of the isosceles triangle can be found in terms 9f the
~ LB
\
tri~ngle.
angles A. B. C of the original triangle; for example,. x = 60 - •
A 8 C
so that a triangle ABe can always be found similar to arty given
. I
FIGURE 4.28
EXERCISES 4.5 Let x represent the numerical

the previous proportion value of the golden ratio. from
1. Does a Brocard point always lie within a triangle? ,
2. Draw a figure localing the second Brocard point for the triangle shown in
Figure 4.25.
3. Prove that the two Brocard points ~re isogonal. conjugates.of each other'
4. Triangle D' E'F' in Figure 4.26 is called the first Brocard itriangle. PH ve that 1 = x2 - X.
this triangle is similar to the original triangle. - x' - x - I = O.

In Exercises 5 and 6, draw a figure showing the equilateral triangle of Morley's
theorem jf the measurements of two of the angles of the original triangle are: x= ±v"5
2 .
5. 90" and 40".
The positive value of x.
6. 60" and 20".
7. Prove that~ in Figure 4.27b, the segments i5i, FG. and EIlI :i;lre concm7rent.
8. lnvestigate the figure formed by the intersections of the !bur pairs of jadjacent
I +v"5
2
trisectors of the angles of a square. I· ~
9. lnvestigate the figure formed by the intersections of the four pairs ofjadjacent is the numerical value ·of the golden ratio. This .'.
fOUi"Msectors (or quadrisectors) of the angles of a square, if four-sebtors are number. An approximate decimal val' • of COUlse, IS an Irrational
defined to be the rays partitioning the angle and its i~terior into four cOO
w
1:
sl~own that the ratios of the (n + ute 1.62. Mathemat.icians have also
gruent angles and their interiors. Fibonacci numbers I I 2 3 5 8 13 ) 0 the nth lerm in the set of
. . • , , • , . ,21,. n appro h h
as a lrult. The reader should .. ac t e golden ratio
to I h express 3/2 5/3 8/5 d
see t at this statement is reasonabl ' , ,an so on as decimals
4.6 SPECIAL APPLICATIONS OF
EUCLIDEAN GEOMETRY
In Figure 4.29, ABeD is a g~lden rectangle, with AB = x the
Applications of synthetic Euclidean geometry to be 4iscussed

in this section include the golden ratio, tessellations, and caroms. Each
of these is of current interest to mathematicians.
With their sense of beauty and proportion, the Greeks;came to
regard certain shapes as more pleasing than otherSI and to build many
of their buildings in these shapes. The most famous example of this is G t7:-<:i-----J H
whal is called the golden ratio, the ratio of the lengthS of the sides K
of a rectangle with the most pleasing proportion. I

In Figure 4.28, suppose that AB/Be = AeIAB. In !his case, o L-------~FL::"'-lJ'""":::::::::.- c
point B is said to divide AC in extreme and meall ratio. and, if t~e length
FIGURE 4 29
of AB is one, the length of Ac is the golden ratio. !
156 CHAPTER 4
golden ratio and Be the unit segment. If square AEFD is removed, the ,6ACF _ ,6DFE, so AFIFD = AC/ED. Bul ED;;;; Be and Be;;;; AT,
remaining rectangle EBe F is also a golden rectangle. since ABCF is a rhombus. Thus,
AF AD
It is necessary to show thal CBIEB = x. Since CB = 1 and or FD = AF'
BH = x-I,
so AFIFD is the golden ralio. NOle lhal if the sides of a regular
CB
pentagon are one unit, then the length of the diago.nals is the
x BH
numerical value of the golden raOo.
Or, since Ell = EB. the proportion may be written as
8
A~----------~C
From the equation x 2 - x - I = 0,
x - 1.
x
Then
The second application of synthetic Euclidean geometry to be
disl::ussed is that of tessellations.
=X. DEFINITION. A tessellation is a pattern of polygons fitted

together to cover an entire plane without overlapping. Regular tessellations
This process of removing squares may be continued, and each . are those in which all of the polygons are congruent and regular, with
time the result is a smaller golden rectangle. As shown in the .figure,
common vertices.
the pattern suggests a spiral, called the golden spiral, whose equation
in polar coordinates is r = x 26/ n, where x is the golden ratio.
One additional application of the golden ratio, its connection Figure 4.31 shows regular tessellations with squares. equilateral
to a regular pentagon, may have been suggested earlier by the expression triangles. and regular hexagons. The fact that these are the only three
regular tessellations is eSlablished in the proof of :he following theorem.
x = 1 + .)5.
2
THEOREM 4.27. The diagonals or.a regular pentagon divide

each other in the golden ratio.
(c)
The notation of Figure 4.30 is used to illustrate Theorem 4.27 la) (b)
FIGURE 4.31
by showing lhal AFIFD is the golden ratio. In isosceles trapezoid ACDE,
CHAPTER 5
If)2. CHAPTER 4
EXERCISES 4.6
CONSTRUCTIONS
i. find the approximate value of the golden ratio correct to four decinial
j
. '. 2.
places.
Use fibonacci numbers to give a series of appro,imations to tbe golden ratio
i
until one is reached that is correct to three places past the decimal point.
3. In figure 4.29, if Be = l, find the measure of segment ft.
4. In Figl.lfe 4.30, prove that ABCf is a rhombUS.
5. Draw a tessellation that is not Ii regular tessellation.
6. Show by a drawing and a numerical explanation why regular octagons
cannot be used for a regular tessellation.
7. Use Heron's theorem to prove that. given the length of one side and the
n;easure of area. the triangle for which the sum of the o,ther twO sides is
smallest is isosceles. .
! J.
Draw figures similar to Figure 4.37 for the following circumstances.
o B
i 0
Cc=JD'
,
.A
.8
C D
F ' £
F £
F E
FIGURE 4.37
8. Where should ball A hit side CD in figure 4.37a in ord~r to carom and hit
baH m i "
9, BaH A in figure 4.37b is to carom off DE. then IT. and then hit ball 8.i
10. Ball A in Figure 4.37c is to carom off side B5. then DE. then IT. and 5.1 THE PHILOSOPHY OF
CONSTRUCTIONS
finally hit ball B,
Ii. Using the notation of figure 4.33. prove th.Hhe shortest path from;A to B.
by way of I, and then I,. is lhe palh shown. ' i :- The concept of
. at least as far as t
. lived abollt a
.
constructIOns in geometr
he work of Plato. the great y can be traced back
12. Complete lhe proof oCTheo ren>: 4.30. ' , " given credit fa century before the time of EUcli~r;ek t philosopher who
13. Complete the proof of Theorem 4.3\. ~pe~lfytng the use of the straigh;ed was Plato who is
t~S constr:~and
, as the only tw:
Plato specified ruments for performing the compass
modern com
used t
e use of dividers. or a colla . c IOns. Technically.
pass we are accustomed to Wh psmg compass. not the
.'
;. ,. , a construct one circle , e n the instrument had
Instrument could not b e moved ' t'!.e_. l11easurement was 10 t b
to construct an th . s ecause been the
o er cIrcle with the same
163
CONSTRUCTIONS 165
164 CHAPTER 5
for the constructions of the Greeks. The first two of these postulates
radius. This characteristic and the limitations of the unmarked straight
make it possible to construct any portion of a straight line through two
edge account for the fact that the transfer of distance was not postulated
points, whereas the third makes it possible to construct a circle if the
by the Greeks. It will be shown, however, that the divider (collapsing
compass) is equivalent to the modern compass, so the straightedge and the center and the measure of the radius are given.
An example of a simple constntction problem solved with lhe
compass are still the only two instruments allowed in the classical
modern compass and then with --the collapsing . compass will illustrate·
mathematical constructions of Euclidean geometf'Y; Constructions that
the difference in the methods that must be used. Figure 5.la shows
are almost trivial with a modern compass become morc complicated
the familiar method of finding the midpoint of ~.segment by consfrucl.iqn.
with the original compass. Nevertheless, the reader should check the
great variety of construction problems specified in t"he books of Euclid
all of which are possible with t11e unmarked straightedge and th;
collapsing compa3S. c
The concept of mathematical construction is difficult to explain
because the word "construction" is used in at least three ways:
A--t--- B A'I----IB'
1. To describe the geometric problem to be solved.
2. To describe the process of solving the problem.
3. To describe lhe completed drawing that results from solving the
problem.
o
(al (bl
FIGURE 5.1
The result of a construction is a drawing that shows certain relationships
among lines and c;:ircles. Philosophically, constructions may be explained
as methods for sOI.ving certain geometrical problems according to a fixed Note that Ac and Be must be congruent, that AD and BD must
set of rules, This will become clearer when other basic ideas have been also be congruent, but that ic and AD are not necessarily congruent.
presented. Figure 5.1b shows the same problem solved with a collapsing compass.
The concept of a construction was basic to the axiomatic In this case, the measure of A'B' is used for the radius since its
system of Euclid. as can be seen by reviewing his axioms and length can be determined using either endpoint as the center of the are,
postulates stated in Chapter 1. On the other hand, modern postulates whereas the arbitrary radius AC in Figure 5.1a cannot be reproduced
for Euclidean geometry represented by the sets in Appendices 1-3
again with the second endpoint B as the center of the circle.
make no mention of constructions, and it may be stated. that con-
structions are outside the strict axiomatic development of modern
. Euclidean geometry. Yet the concept of a construction remains significant THEOREM 5.1. The compass and the collapsing compass
in geometric thought. Constructions are studied in various modern (dividers) are mathematically equivalent.
geometries not only because they are interesting in their own right but
also because they provide applicatIons of other geometric (;oncepts.
The proof of this theorem consists of showing that a circle can be.
The problem in a construction is not simply that of drawing a
constructed with the coiIapsing compass, given the center and two other
figure to satisfy certain conditions but whether, by using a compass and
point8 that determine the length of the radius. The steps in performing
straightedge only. a theoretically exact solution can be obtained. The
this construction are stated, using the notation of Figure 5.2. The
drawings in construction problems are approximations, but the theory is
problem is to construct a circle with .center A and with radius Be.
exact. The first three postulates of Euclid provide the axiomatic basis
166 CHAPTER 5
. , CONSTRUCTIONS 167
Construct the circle with center A and passing through B.

th~ough the established points on
pomts I, J, K, L.
le. determining the four required
I. Construct the circle with center B and passing through A. These
2.
two circles meet at D and E.
3. Construct the circle with center E and passing through C.
4. Construct the circle with center D and passing through C.
5. The circles in steps 3 and 4 intersect again in a point F. and the
circle with center A and radius AF is the required circle.
FIGURE 5.3
. A second problem related to t

;~;!~en ratio. The problem is to ::dfi:st i~
to partitio:.. a segmelll
OfP::~;; i~ne
. as a gIven numerical v POlnt B on At; such tha
In the theory of harmonic sets use. of this construction i:
FIGURE 5.2 proJective geometry
( . F'19ure 5.4
,
The proof, using congruent triangles, that AF is congruent to Be
is left as an exercise.
It is assumed that you are familiar with the basic constructions
of Euclidean geometry. These include transferring a segment, bisecting a
segment, constructing a perpendicular to a line at a certai~ point,
constructing an angle bisector, copying an angle, constructing a triangle
given an angle and the two adjacent sides, construct;n g a triari~le gi~en o
the three sides. and constructing a Hne through a point para~lel to a (.) (b)
given line. It is also quite possible that you have i encounterJd some
FIGURE 5.4
of the more difficult constructions to be introduced later in the!chapter.
Several additional basic constructions afe introduced or reviewed
is an analysis figure, showi~
here so that they may be used easily. that Band G divide AB inter!a~;le completed solutions. It is assumed
Partition of a segment into n congruent segments. for n a A~ and JJF are parallel, then 6~;;d externally in the desired ratio If
~ ~
positive integer greater than one, is the.'ilrsl of these constructions and is
illustrated in Figure 5.3, for n = 5. The problem is to partition All
into five congruent segments. A""C is constructed through A at any _.
ratlO of division. Then
and of AE t C
t~:or.
an. the ratios of similarit . 6CDB and 6AEG
the palfs of similar triangles are
gIven ratio is also the t"
II;; .
CFG,
gIVen
o F' this k ra 10 of AE t
convenient angle. Then an arbitrary unit AD is chosen to determine · Illustrated in Fi ' rna es possible the followi 0 CD
. rail I I' gure 5.4b for a ratio of 5/2 TI ng construction
five congruent segments along A(;. Next H B is constructed, and the e IDes U . . nough A' d '
· pa . Sing an arbitrary unit find Ef an C. construct
four segments [51. l!J. PK. and rrr. are 'constructed parallel to HD. · ' sothatA'E'-5 - . and find
CONSTRUCTIONS 169
168 CHAPTER 5.
Df and F' so that CD' = C'F' = 2. Connect E' , D' and E', F' to locate the EXERCISES 5.1
required points B' and G',
In the problem of partitioning a segment into a given ratio, the For E>.efcises 1 and 2, explain how a collapsing compass CHit be llsed to:
known information might be given entirely in the form of segments. I. Bisect an angle.
For example, AC could be given along with a unit segment and a third 2. Transfer a segmenl.
segment whose length, in terms of the unit •. represents the given ratio. 3. Complete the proof of Theorem 5.1.
In this case, the unit segment could be ~sed for CD', whereas the
third segment would determine the length AB. A specific ex.ample of this In Exercises 4-6, show how to perform these basic constructions in. Euclidean
type of construction is given in Figure 5.5. The problem is to partition geometry:
A:C internally and externally in the given ralio. . 4. Construct a perpendicular to a lin.e at a certain point.
5. Construct an angle congruent to a given angle.
6. Construct a line through a point and parallel to a given line.
Given: Construction:
For Exercises 7-10, carry out the construction indicated.
A----C 7. Partition a given segment into. seven congruent segments.

8. Pan it ion a given segment internally and externnlly in the ratio of tlH~e to
--Unit two, given a unit segment.
Given ratio 9. Carry out the SHme construction ,1S in Exercise 8, but use II nHio of llll"t~e
o to fOUL
FIGURE 5.5 10. Partition a given segment internally and externally in the ratio of length
of two given arbitrary segments, neither of which is the unit segment.
The use of the basic constructions of this section leads to the

solution of more complicated construction problems. In a formal study· of
constructions. four distinct steps are required in the solution of any con- 5.2 CONSTRUCTIBLE NUMBERS
struction problem.
The weaknesses of Greek algebra and lhe strengths of. Greek
1. Analysis. In this step. the solver assumes that the con~truction has geomelry were all too apparent. Without an adequate notation, they
been performed. then analyzes the completed picture of the solution . used geometrk constructions to solve many ·of the problems J.1ow solved
to find the needed connections between the unknown elements in the algebraically-for example, the solution of some algebraic equations. nut
" one thing that this did accomplish was to focus attention on the con-
"
j. figure and the given facts in the original problem.
2. Construction. The result .of this step is the drawing itself. made nections between geometry and the various number systems of Higebra.
i-
;}
with straightedge and compass and showing the construclion marks. a study that has continued to have significance in both modern algebru
>,
3. Proof It is necessary to prove that the figure construded is actually and modern geometry.
A unit segment represents the number one. What other numbers
the required figure.
4. Discussion. The number of possible solutb;H1s and the conditions for can also be represented by segments, beginning with Ihis unit segment
and using only lhe straightedge and compass to construct other segments'?
any possible solution are explained in this step.
The answer to this question defines what is known as lhe set of
cCJ1lstrucribfe /lumhers. Figure 5.6 shows a geometric interpretation of the
In this text, it will not be necessary to carry through all four
steps in complete d~tail, although each will be illustrated. four ratiooal operations on whole numbers, as well as the construction of
170 CHAPTER 6 CONSTRUCTIONS 171
. the square root of a D.ADB - [';DCB.

. . b the process of extractmg. h three
some irrational numbels YTh iven information consIsts of t e
positive ratio~al number. e g so that
segments in FIgure 5.6a.
AB BD
a b BD = BC'
a b a~bG AB = (BD)'.
b
a+b
a .JAB = BD.
(e)
(b)
(a) Given The first four constructions of addition, subtraction, multiplica-
tion)nd division make it possible to construct a segment representing any
number in the field of rational numbers, given the unit .segment. The
construction of the square foot makes it possible to construct numbers in
~
extension fields having the field of rational numbers as a subset. Recall
that a field of numbers has the closure property for rational operations,
with division by zero excluded. Examples of extension fields of con-
J a + b.j'J..
;
;
-'
-' structible numbers are a + by'2. a + b.j3. for a and b
A 8 C rational numbers and the entire radicand positive in the third example.
A 1 8 C
~
a For example, the number 5 + J3 + J7 +../2 is a constructible
a b number, while ;y7 is not.: -
(e) (f)
(d) A somewhat more general approach to constructible numbers
FIGURE 5.6. can be studied from an algebraic viewpoint. Suppose that all of the
numbers in SOme number field F can be constructed.
. . 56b and c for addition and sub-
TIle diagrams III Ftgure
Id be self-explanatory. .
For multiplication. the proof of the
traction SIlOU d on the proportion THEOREM 5.2. TI,e use of a straightedge alone can never
construction in Figure 5.6d depen 5 yield segments for numbers outside the original number field.
1 a
b= AE
The equations for any two pairs of lines through distinct
pairs of points with coordinates (a. b). (c. d). (e,f). (g. h) in a field are
so that A E. = ab. . . .
f of the constructlOn ,'n Figure 5.6e also ,:1
For divisIOn, the plOO j',.
d-b
y - b = - - (x - a)
depends on a proportion. C - a
1 = -b, a and
_ Of AD = b'
AD a
h- J
. ual to Ja
in Figure S.6f likewise J' - J= - - (x - e).
The proof that B~h:t ~~ turn is derived from similar right g-e
depends on a proportIOn .
The point of intersection of these two lines has coordinates obtained
triangles in the figure.
'I 172 CHAPTER 5
CONSTRUCTIONS 173
I by rational operations on elements of the field F. hence the use of the

stnlightedge alone did not result in a number outside the original
extension field. The proof of the second part of the theorem involving
the intersection of lwo circles is left as an exercise. As a result of this
field. It is naturally assumed that division by zero is avoided. analysis, constructible numbers can be characterized as those. (hat can be
Now select an element 1I of F such that fir is not an element obtained through a sequence of extension fields of the type dIsclIssed.
of F. All numbers of the type b + c}7t (b and c also in F) can be
constructed with the use of the straightedge and compass. Numbers of
the form b + cJct themselves constitute a field, and this is an extension
lield of F. A single tlse of the compass cannot lead from i beyond
an extension field of F. however. Take a given unit segment and two other given segments to perform- the con-
structions in Exercises 1-5.
1. Find a segment whose measure is the sufl:t of the measures of lh~ two
THEOREM 5.3. A single application of the compass using givc!n segments.
numbers of a number field results only in elements of the ex.tension 2. Find a segment whose measure is the dill'erence of the two given seglllenis.
field b + c.ja, where a, b, and c are elements or
the original field, with 3. Find a segment whose measure is the prouuct of the measures of the two
a positive. given segments.
4. Find a segment whose measure is the quotient of the measures of the two
From an algebraic point of view, it is necessary to consider given segments.
5. Find a segment whose measure is the square root of the meJ:lsure of the
the intersection of a circle and a straight line and then the intersectIon
of two circles in order to prove the theorem. longer segment.
The intersection of a circle, .'(1 + y2 + ax + by + c = 0, and a For Exercises 6-lO, with a given unit seginent, construct segments for each .of
straight line, dx + ey + J = 0, with all coefficients in F. is given by these numbers in the set of constructible llumbers. Use the results of one. exercise
the solutions of for the next.
6. 3 7. .fi 8. 2 + .fi 9. /2 + .)3 10.)3 + )2 + .)3
x+,(-I-dX)'
e
+ax+)((-I-dS) +
e
t 1. Construct a segment for ,fiJ3:given a unit segment.
12. In Theorem 5.2, find the point of intersection of the two lines.
This equation may be written in the form 13. In the proof of Theorem 5.2, show llat I oum bers 0 f 11le form l1 + V~u !lre
C
closed under the four rational operations.
(1 [dJ') , + (2dI - bd +a) x+ (fl---+c

-
e
x
e
- bI ) e
14. Complete the proof of Theorem 5.3 for two inlersecting circles.
Each of the coefficients of this equation is an element of F, so they

may be indicated as g, h. k. so that gx 2 + hx + Ie = O. The quadratic 5.3 CONSTRUCTIONS IN ADVANCED
formula yields the solutions EUCLIDEAN GEOMETRY
In traditional college geometry, the construction of triangles

4gl<
X= played a dominant role. One reason for this was that constructions
2g
gave a ready application of many of the new concepts introduc~(L. In
Borh of these solutions are of the form b + cJot
for certain numbers some modern intuitive geometry courses, especially at the beglllnmg
in F. so that the use of the compass did not lead outside the secondary level. and iI\ college courses- for elementary tel1chers of
CONSTRUCTIONS
174 CHAPTER 5 " , 175
POl~ts satisfying given conditions ..
mathematics, constructions serve as a primary vehicle to present· con- the mtersection of two loci a 'dTradltI?nally, this point is desc 'b d
C .wormeanm" neas
cepts for the first time. onstruct a triangl' g paths."
ng
The reader of this section should get some idea of the variety
of construction problems and should also appreciate the use of con-
ana~SI,
opp 't h'"
e ~ IS an,ole, and the length of t: : ,the length of the side
e. gwen one [
YSIS gure IS shown in Fl' e a tuude to that side A
structions in the application of concepts alreadY studied in previous ,and B C '
are given, Since Band ' , ASSume that L BAC ' ADn
gureC 58a
chapters. Developing great skill in using constructions is not the desired mformation. t h e only remaining blcan . be Iocated from the " .
outcome, Often only a brief analysis of the problem will be presented, to Band C Th pro em IS to Iocat' gIven
but the ' e other conditions do not d ' e pomt A relative
that it ~ can be used together. One condition e~ermme A individually,
rather than the detailed construction itself. The proof and the discussion
witt ~ometimes be suggested. the 0 u,st b~ on a line parallel to BC an or the location of A is
COllstruction problem, Construct a triangle, given the length of pposIte Side, Vertex A ' d at a distance of AD f
one side of the triangle and the lengths of the altitude and the median can b ~ mllst also he 0 Ih ' rom
_'I' e ound from knowing the angle nd e Circumcircle, and that
to that side, " wry
aU.\l . cons t ructlon
. . shown in F'
IS an the 0 pposlte . .
SIde. This
The analysis figUl'e is shown in Figure 5.7a. From the given
determme A . an d '
the triangle can be Igure 58b " Th,e two conditions
information, right triangle AA'D can be construcle~' immediately since constructed.
two sides are known. Then points Band C can be located on 15A.
Each is a distance half the measure of the given side from the
determined midpoint A'. The actual construction is shown in Figure S.7h.
Triangle AA'D. and hence the required triangle, can always be constructed,
as long as the segment for the median is at least as long as the segment
for the altitude. There is only one possible solution~ Triangle .;,lA'D is
an example of an auxiliary triangle in a construction, An a~xiliary
triangle) usually a right triangle, is one. that can be constructed
(a) (b)
immediately from the given information.
FIGURE 5,8
Side , The auxiliary problem i

Median - - - - -
fight. In Figure 58b the I n th1s example is interesting' .
of the circle can" ang e FeE is the given an m Its Own
Altitude - - - - -
bisector of BC and bt~ located at the intersection of t~le,
so the center
'Th Ie perpendicular to CE at C e perpendicular
A e three element '. .
constitut:!~ e~~:lel ~;:hang!~
triangle, the Ie s consIstIng of the measur f
C~8
circle oPfPosite side, and the ra:iu: of a
" p e 0 a datum. e Cllcum-
o A'
(b) , DEFINITION, A datum is
(a) whIch determine the rema"10mg
. one. a set of n elements • any n - I of
FIGURE 5.7
The second example-illustrates the important idea of locating one Figure 5.8b shows onl
~ one.Itpart
elements constitute a d a t urn, SInce mustof bthe hproof [hat the three
of the vertices of the required triangle as the intersection of two sets of e s Own that each -two
determine the third. The CPIH':Cpt I..,f a unltllll i~ illlpurtnni in con~

slructions.. since two t!i\'cn elements will determine
~ , .
a - third
_." - -1'.:nt- if the
.,.c\e1l ~.
three (on:;litutc a dalulll.

The di:"t:u.s."illll of the pruhlem rcprcsC;11\cd \JY Figure :. ~ cPIl~i~lS
~)r deter1l1ining Ihe numher llf solutiuns. There could he I'XI,I lillcs ,.
pnnlllc1 to BC. ami each (.lr these might intersect the cirdl 1 \\-icc. (n
one ct\5C. however, the <lnglc would he the ~ltpp1cIllCnl PI' the !then
i.lltgk, 50 that case nHl~t he di"C:lI"ocd. This."lea\'e" the IH1;-·.1 1,ility ror
t ,vo, one, or no solUlil..l1ls, depending on lww many times the lin.:: intersects
the circ\lIncircle.
Often, the cUI1I.:cpl (If similar tri<ltlg.lcs can be lI~r:J in con-
FIGURE.S.IO
structions. This is illustrated in -the problem of fO/l.'ilrrlcrifl~1 t.f I riallH It',
uirefl {he meclsure f!( two anuks (m(l till' lenuth of Ihl' M:;{', 1"1 or fill"
fl." thc Orllwccnlcr, i~ detcrmincd from lhe het thal the' .
{Irint oll!!!e. Figure 5.9 i!-; the analysis figure. The measures (I! the three cClller \' . tl .I ~ nll1e~p0ll11
. ' f ,IS Ie 11m Iwinl of the segment joining {ile orllHlcenler 'lI1d
th: 1.:lr:·U11lt."Clllcr O. Since N i." nl."o the midpoint of AT where 1 is \he
A ~lIdP~ll~ll of IT.l nm! A' ~5 thc mitlpoint of the :<>itlc. point I CHIl be
cICfl11l11et!. then Ventx A can be rount! a known distance 011 Hi
beyond
l. One .loc\15 f~lr p{lilll<; Hand C is a line throug.h At perpendicuhr
c>~-----F~--~B
o l~l ~.l'. n.te scc~Hld IllclIs is the dn,:ul1lcirclc, ami (hi:::; call be d~tcrl11in;d
tcc.luse the radIUS 0.1 has hecn round.
. . :: SCCll l1(j example or a construction using H more i.u.lvnllced
U'll~Cpl Is Ihe nlllSlnWrio1i (!fa 'rialfule, nit'<'TI rhe lell.tfrh.<t of (Ire crltiflfc/e
nGURE 5!1 ~1lt'd~~111. (lIId 5Y'l/IH~i(/I/J"o1l1 (ht' ,<tam(' /'C'r('s. The annl)'::;is n'gt,,>c", I >"
111I I - II . !;; S HlWn
. gu~e ='. ...IV IS the altitude, ..IE is lhe symmeuian, and tiA' is
<\llglcs of a {rinnglc COI1;::.titutc a dfilulll. ~ince any {\\n dClermine the the tl1edl~lI1. The an::llysis and discU5Sioll ::ue left as an exercise.
third. Fnll11 this infvrl11:'lth.lll. a triangle .-lIfe'. ;::.imilar Itl Ik· required
triangle. call be c\1Il5lfll<..:lcd. This idea i;::. known :'IS the tri:!nglc being A
determined in -"pedes. A family of triangles is determined. amI I he required
,~
tri~lIlg!c can be fqund as one particular memher nf thb !:unily. The
required triangle is found hy laying nIT the ghen length pf 1he bisector
aitlilg. ,.I/)' to luc.lle point lJ anli then cnn~lruclill!.! tilt' l'an:dlcl III
("Ii through D, Therc is ;lIwilYs nilC 5\11111inll as hlllg (\5 lilt' .-tllll ~lf the
DE F A' 8
measures of the 1Wt1 given ang.les j<;: less than 1'(.
Cnnslructi\ln can innll\'c CI.Jm:cpts fn111l Euclitlean ~l:"'l1letry that FIGURE 5 11
arc more :'HhllI1CCt! than thusc u.<;ctl ~p rar in lhis ~(' tiPI!. One
cx:,ul1plc of this is 10 ('Oll:;;(rlfcl a lri(/II~lk. Ilinm fIll' Cin.:ltlll~('''(t'r. (II('
n'llft'r {{(lIe lIilll..'~f1ni"r circle. (l1Ifl til<' midl'uillf (!rOIlC sitlc.
The allal.ysis ilg.urc is ~dwwl1 in Fig.ure 5.10. Th;;- JllcHlitm of
A 0----1---7 B Draw the ac!Uaf c{.lllstruction figure corresponding to Figure 5.10. j.
J. Complete the discussion of the number of solutions ill Figure 5.13.
Fllr c;lch of the following construcliol1s:

D""'-_ _- + _ - _ J ) C a. Sketch an an:ll}"is fi!!ure.
h. Give the •.HlilIYSIS.
FIGURE 1) 12 c. DrieR)' discuss the number of possible solutions.
4. Cnnslruct a tri;:tnp.le. given the mellsure of one angle, the length of an
The nnl.llysis figure. Figure 5.12. shuws Ihat each of the t·.·'!') right adjucelll side. and the length of the altitude to Ihat side.
triangles BCD and ALJD can be ClHlstructed from the given inliq·mution.
5. Construct a triangle. given the measure of one angle. the length of (hal
A linal extunple is that of COlIstruclill!/ a circle ll"ith If I\IIOInI internal angle bise(:lor. and the lenglh of one ntljacent sitle_
raelitls tltClt is tanyenl, ta a uin'n lint! anel orthogollal to c/ vir ,,·It circle. 6. Construct a triangle. given the length of one side and the lengths of the
Two curves are ort/wf./cJllal if they meet at right angles. Th ..~ ;maiysis
figure is shown in Figure 5.13. the circle with center ais th.: required
circle, I is the given tangent, and the circle with center 0' Ig the
7.
medians In the other two sides.
Construct a tri<lIlf!.le. given the length of one side. {he length
altitude to a second side, and the circulllrndius.
or [he I
given circle.
8. Construct a triangle, given the length of one side. the length of (he median
to that side, and the circulllrndius.
9. Construct a circle with n given radius and tungenl to two given intersecting
lincs.
10. Ctlf}strucl a ("irde wilh 11 given radius UlIlgen( 10 :1 given line and tnngenr
It-I a given circle.
t I. Com;truct a triangle given the length of one side, the length of the median
FIGURE 5.13 12.

III that sidt. and the ratio of the lengths of the Iwo remaining sitles.
C(mstrllct II triung/e. gi\"C1I the measure uf one angle, the measure of the
(lrrosite side. und the radius (If Ihe incit·c1e.
I
The only thing to be determined is the It.H.:utiul1 of llh! center 13. Construct a Iri'lIIgle, given the measure of one angle, the lenglh of the
of the required circle. One condition is that the center lie "'II a Iinc internal bisectllr nf thaI angle, and Ihe wdills of the illcircic.
parallel to I at a ghren perpendicuinr distance ..10 from lint: 1. The 1-1. Construct a lfinngle. given Ihe mellsure uf vne tingle nnd the length of the
second condition is thal the cenler lie a kno\\"n di,,! ..IIKC of allitudes to the I\\n ;uJjncent sides.
00' = J(OB)2 + (ii(j;)i from O' so that its position cun I.'e Uxctl. 15. Cllostrucl a triang.le. gi\"cn the length of one side. the length of the l11edhm
Because the location of 0 depends on the interscctitlll llf a ·.:irde with to Ihnt side, and t~e length ~l( vne other mclli.lIt.
two parallel lines, 11 complete tlh:;'cll$~jon (If the problem "'llltL! h;:l\"e to 16. Construct 11 tri<tll.glc, given the measure of 1\\"(1 <Ingles and the length of
consider from four down to z.ero possible solutions.
17.
Ihe mediun to (he thirtl side.
Construct a triangle. gil'en the measure of one tll1gle. the lcngth of the
altiWde to the opposite side. and the ratio of the two adjacenl sides.
I
EXERCISES 5.3 IS. COmilruct a triangle, gil'en the ortilocenter. the ninc-point center, and the
fool or one aitilude.
L Finish the proof thal the measure of one angle of a triangle. til: IllCrlsure 19. Construct a triangle; given the lengths or the altitude. median, and symmedian
of (he opposite side, and the wdius of the circumcirde COllslitUIt" a ualum. rrom one \·erlex.
.-.~
..,
';1 '. '
CONSTRUCTIONS 181
100 CHAPTER 5·
lies in one of these extension fields leads to a contradiction. It can be

54 CONSTRUCTIONS AND
proved, although the algebraic details are omitted here, that if a number
IMPOSSIBILITY PROOFS
of the form a + bJc is a solution of a quadratic equation, then so is
While th.e Greeks were deeply interested in the philosophy of
a_ bJc. It has previously been shown that elements of extension fields
can always be written in the form a + b~. But the assumption of
constructions using the instruments specified by Plato, they also realized
two real roots for x 3 = 2 contradicts the known fact that two of the
that some problems were beyond their abilities to solve by these
means. Rather than give up, they invented new instruments that would cube roots of2 are nenreal. Thus, x is not a constructib1e number.
This completes the proof of the impossibility of solving the
make the constructions possible. However. not until the nineteenth
century did advances in the algebra of the real-number system make first of the Greek problems.
it possible to identify constructions that were impossible with the
straightedge and compass. THEOREM 5.4. The construction of doubling the volume of a
In the history of mathematics, three construction problems cube, cannot be performed by straightedge and compass alone.
became so famous that they are called the "Three Famous Greek
Problems." These are the problems of doubling the cube. of trisecting The second problem. that of trisecting a general angle, can be
the angle, and of squaring the circle. The algebra of the constructible
disposed of in a way similar to the first. Ordinarily, this problem is
numbers has provided the information necessary to prove that each
approached by showing that a single example, such as trisecting an
construction is impossible. angle of 60 degrees, is impossible, and hence the general case is
Various legendary stories account for the origin of the problem
of doubling the cube. One states that a king wanted to double the size
~-~_B
of the cubical tomb of his son. Another tells that the DeIians were
instructed by their oracle to double the size of the aILar erected to
Apollo in order to rid the city of a plague. In these problems, it is the c
volume of the cube, not the edge, that is to be doubled (see
a k:::..J:t1--O'1 A
Figure 5.14). a £
FIGURE 5.15
impossible. In Figure 5.15, suppose that mLBOA = 60 and mLCOA

FIGURE 5,14 = 20. II' the circle is a unit circle, then ~OD = cos L BOA and
OE = cos L COA. The problem of constructing the smaller angle is
If the otiginal edge is one unit and the required edge is x equivalent to that of finding DE, given OD. The trigonometric identity
units, then the original volume is one cubic unit and the volume of the relating the cosines of an angle and a second angle with one-third
doubled cube is two units. This means that x must be a real solution the measure of the first angle is
of the equation x' = 2. It was established in Section 5.2 that the set of
constructible numbers must be an element of some extension field of the cosO = 4COS'(~) - 3COS(~}
set of rational numbers. The assumption that the solution of x 3 = 2
CONSTRUCTIONS 183
182 CHAPTER 6
DEFINITION Al eb . .
t
If COS (0/3) = x, and cos 0 is a given constant (since 0 = 60°), then equation of the form . g rate numbers are solutions of an a IgebraIc
.
4x 3 - 3x - 1: = o.
• .
aI/x" + a
II-I"
,,1/-,1 + ... + Q o = 0
or >
with mtegnil coefficients, and with n=landan:;cO.
Sx' - 6x - 1 = O.
ll1is equation has no rational solutions, because, if alb is a 2. T. Lindemann proved in 1882 that th
rational solution, then a is a positive or negative factor :Of 1 and b number, so In
is not algebraic either. e number" is not an algebraic
is a positive or negative factor of 8; all possibilities can be checked
quickly by synthetic division to see that none is a solution.
If the equation has a solution of the form a + it hasbVc, THEOREM 5 6 Th
possible by Iileans of a ~t' . : ;nstruction of squaring a circle is im-
a _ b}C as another solution. Suppose these represent' the least- ralg te ge and compass alone.
inclusive extension field of which the solution is a member. The sum
of the three roots must be zero, the coefficient or" the Xl term, so that if As mentioned earlier h
r is the third solution.
thes~ three constructions by ;h~ :::~/:~at the Greeks could not perform
~ot eep them from finding solutio e straIghtedge and compass did
+ + (a + ~ 0,
~Ol;~~~i:! pro:a~~v:~~::
(a bJc) - bJc) r IIImathematics, inc1udin ns by other means. Man
2a + r = 0, as the result of attempts the theory of conics,
Two examples showing how the co solutl~ns for construction problems.
r = -2a. are .taken from the histo ry 0 f rnathemati
nstructlOns
de 0 can actually be perlormed
'
This contradicts the assumption that a + bJc
represented a number TI Vice or curve called the conchoid oj N' cs. ne of these is the use of a
5.16~1~~tlllg c:~~~:OdD.'.'.
le actual method of '. lcomedes. who lived ab
in the least-inclusive extension field of solutions. The conclusion is that illustrated in Figure an angle by using the
the solutions of 8x 3 _ 6x - 1 = 0 are not constructible numbers, so. e CUi ve through G is tJle conchOJd.
• For
1 IS
a
the ronowing theorem is established:
THEOREM 5.5. The construction of trisecting the:general angle G
cannot be performed by use of the straightedge and compass alone.
The third and most complex of the three famous

problems is that of squaring a circle. This means to fi~d the length
of one edge of a square that has the same measure of area as that of a
circle whose radius is known. Algebraically, if x is: the required
length of edge and if the length of the known radius is one, then·
x 2 = n. The length In
is the required measure of the :edge. Proving·
that this number cannot be constructed depends on the' following two' J'---~:::,ilo
statements, both of which are true but not proved here. (a)
(b)
1. All constructible numbers are algebraic. FIGURE 5.16

~ ~
fixed line CD and a fixed point 0 not on CD, the conchoid (which as shown, then the arc for any angle AOD has length kO. which is also
really consists of two branches, one 011 each side of CD) is the set of the length of r = OC. 11,e area of the circle is
points defined as follows: Consider the set of all lines through 0 and
intersecting CD. Take a fixed distance on the line beyond the point of
intersection. For example, let DB be one of the Hnes. with EX the
fixed distance. FOf a given ~int. line, and distance, the set of all points If OA and OB are perpendicular, then
X the fixed distance from CD along the rays from 0 is a branch of a
conchoid. I(
1: (4
AtB) .
= 21«OC).
If the cline (or rather the mechanical instrument) is placed as in
Figure 5,16, with CD perpendicular to OA and with X E ,= 2(EO), then
the given angle AOB can be trisected simply by locating point E on the line Tf a square of side x is to have the same size as the circle, then
on DB. drawing EV par~llel to DA. and connecting 0 to G, the point x' = 21r(OC),
011 the conchoid. oG is the trisector of the given angle. .
The reason why oG trisects the angle depends on the theory and x can be constructed.
explained in connection with Figure S.16b. GJ = 2EO by assumption.
If H is the midpoint of W, then EO, ill, GH, m are all congruent. In addition to the three famous problems, the equally interesting
Becaus6 of the isosceles triangles, mLEOH = mLEHO = mLHGE problem of constructing a regular polygon inscribed in a given circle
-\- mLHEG = 2mLHGE.But LFOI;; LHGEbecauseofparaliellines, has concerned mathematicians since the time of the Greeks. In this case,
so 2mLHGE = 2mLFOJ. The result is that LEOG has a measure the proof for the general problem was again algebraic and was provided
twice that of LFOJ. so 01 is the trisector. by the great mathematician Karl Gauss. At the age of 18, he solved the
previously unsolved problem of how to inscribe a regular polygon of 17
sides in a circle, using only a straightedge and compass. He also proved
the theorem that tells which regular polygons can and cannot be
c inscribed. This theorem is stated here without proof.
o
THEOREM 5.7. A regular polygon can be inscribed in a circle
by mean:; of a straightedge and compass alone if and only if the nUlnber
of sides, 11. can be expressed as 2-'=· PI' p.z .•. pJ.. for x a non-negative
A integer and each PI a distinct prime of the [orm 2:H ' + 1. for y ~ o.
Some of the regular polygons that are constructible, according

FIGURE 5.17 to the theorem of Gauss, are those with 3,4, 5, 6, 8, 10, 12. 15, 16, 17,
20, or 24 sides. The statement of the theorem gives no clue as to how
A second example of a curve from higher geometry used to to prove the possibility or impossibility in any specilic case without
construct a solution to one of the famous problems is the use of the reference to the general theorem. The specific cases for n = 10 and
spim/ of Arcltim£des to solve the problem of squaring the circle. The 11 = 7 are discussed briefly.
spiral of Arcbimedes is the curve OeD shown in Figure 5.17, with Figure SolS shows the analysis for the regular decagon. The
polar equation,. = I<fJ, for f( a given constant. If a circle has a radius k central angle AOS has a measurement of 36", and angles OAB and
CONSTRUCTIONS 187
186 CHAPTER 5
2cos30 = 2(4cos'O - 3 cos 8) = x' 3x
2cos48 = 2(2cos'28 - I) = 4(2cos 2 8 - 1)2 - 2
= (x 2 - 2)' - 2
Setting the two expressions equal yields a quartic equation in x .
. X4 - x3 - 4x 2 + 3x + 2 = O.
-which can be factored as (x - 2)(x' - 2x - I) = O. It is left as an

exercise to show why these factors do not give vaJues for x that can all be
constructed with straightedge and compass.
FIGURE 5.18
f 720 If BC is the bisector of L ABO, EXERCISES 5.4

ABO each have 1Ueasu;~e;~ ~re b~th isosceles. so AD ;;= ;rr ~ CO.
then triangles ABC an I. Apply the definition to show that J2 + .J2 is an algebraic number.
Also. 2. Is,y2 an algebraic number? Why?
6ABO ~ 6ACB, 3. Find an approximate decimal solution for the length of an edge of a cube
that has twice the volume of another cube with an edge 2 inches long,
and 4. Prove that if a number of the form .Q + bJc is a solution of x 3 + dx1
1 x + ex + f = o. so is a - bJc.
-=--'
x 1,- x 5. Explain why the equation 8x 3 - 6x - I = 0 has no rational solutions.
6. Explain how a mechanical device might be made that vould actually draw
__ O. The positive solution is
or x 2 +x - 1 one branch of a conchoid.
x =
J5 2- 1 ' 7. Show that the spiral of Archimedes can also be used for trisecting an
angle.
8. Use Theorem 5.7 to list those regular polygons with between 24 and 30
. au should recognize x as: the golden sides that can be constructed in a circle.
which is a construcuble number. Y
. 9. In tbe Quartic equation for the regular heptagon. explain why the first
ratIO. . , . the regular decagon can be <.::'onstructed factor cannot be used to find a solution,
As Theorem 5.7 mdlcl;ltes, (even sides) cannot. The
the regular heptagon s . I . 10. In the quantc equation for the regular heptagon, explain why the second
in a circle, whereas . t heptagon in - a eIre e 15
factor cannot be used to find a solution.
Proof that it is impossible .to constr.uc
~W_I
t.aon of a general angle. The
U
similar to the proof concermng . a length x = 2 cos 2n/7.
.
prob Iem 15 e quivalent to constructmg
5.5 CONSTRUCTIONS BY PAPER
21</7 = 0, then
FOLDING
30 + 40 = 360o ",d cos30 = c0540.
As discussed earlier in this chapter. the Greeks specified that
. . I'dentities can be used to express these cosines in :.
Tngonometrlc geometric constructions should be performed by straightedge and
terms of x.
188 CHAPTER 5 CONSTRUCTIONS '189
compass alone; nevertheless. they devised other instruments when some

problems proved impossible with the given instruments. Since that time,
other ways of accomplishing the same constructions have been devised.
One that may seem elementary, yet has occupied the attention of
mathematicians in recent years, is the approach through paper folding.
In actual practice, waxed paper is ordinarily used so that the crease will
, remain visible. A few of the basic constructions are illustrated in this FIGURE 5.21
; section so that you will catch the spirit of this work and thus be
Ii
l -~
able to contrast it with other construction methods in Euclidean
geometry.
F aiding a perpendicular ji-om a paillt to a line. Simply fold a
perpendicular to the line through the poillt by superimposing part of line
F aiding a straight line that is the perpeHdicu[m' bisector' of a given in on itself (Figure 5.22).
segment. Suppose An is given, as in Figure 5.19a. Fold the paper so that
~nt A is superimposed on point B and crease, as in Figure 5.19b.
CD is the required line.
\----i,O----"B
i
I.':. FIGURE 5.22

", E D
Although paper folding seems very simple. mathematicians have

(a) (b)
been able to prove the following rather startling theorem about paper-
FIGURE 6.19 folding constructions:
Folding the bisector oj a given angle (Figure 5.20). Fold the

THEOREM 5.B. All of the constructions of the plane Euclidean
paper with the crease through the vertex and with one side of the
geometry that can be performed by straightedge and compass can also be
angle superimposed on the other.
performed by folding and creasing paper.
r\
~
11 This remarkable theorem is based on several assumptioIl!:> that
I: can be found in the excellent reference on paper folding, Papel" Folding
for the Mathematics Class. by Donovan A. Johnson. These include, for
8 example, the assumption that paper can be folded in such a way lhat one
FIGURE 5.20 line ,?an be superimposed on another line on the same sheet of paper
and that the crease formed is in fact a straight line. Even the brief
Folding a line through a given paint ami pamllel to a given line. analysis of paper folding in this section gives additional insight into the
First fold a crease for Co perpendicular to the given line in nature of the approximations involved in ordinary constructions. The
(Figure 5.21), then fold a crease through the given point G for a line theory of paper folding is just as mathematical and just as exact as the
~ • +--4 +-+
EF perpendIcular to CD. That line is parallel to AB and j,s the theory of constructions with straightedge and compass, but the methods
required line. differ widely.
CONSTRUCTIONS 191
190 CHAPTER 5
. be used for a somewhat different. type of to. Prove that the paper-folding method of constructing lines tangent to a
Paper foldmg can . . d' college geometry. that of parabola is valid.
II studIed In or mary .
construction not usua Y I F' 5 23a illustrates how a senes
t parabo a .gure . F
,constructing tangents 0 a b . determined by folding the focus
5.6 CONSTRUCTIONS WITH ONLY ONE
of tangents to the parabola can e
INSTRUMENT
In every section of this chapter, including Section 5.5 on paper

B folding, the concepts introduced have been closely.related to the Greek
cB---;(
restrictions of using only the compass and the straightedge to perform
d constructions. A somewhat radical departure is to determine whether it is
d F really necessary to use both instruments or whether it might be possible,
using just one instrument, to perform all of the constructions that can be
performed with both. For students who have trouble.making constructions
with both instruments available, this may seem sheer folly, yet mathe~
maticians are interested in finding the truly minimum set of instruments
. necessary for making constructions.
(a) (b)
Historically, attempts to discover the constructions possible with
FIGURE 5.23
the collapsing compass alone came tirst. However, constructions using
the straightedge alone will be discussed first here. Though the mathe-
. . wh this construction works is exp.lained
matical interest in this problem comes from the middle ages, it was not
onto the directflx d. The leasor:vedY that the tangent BII: at -a pomt on
in Figure 5.23b. It can be. p . f FBC where F is the focus and C until the invention of projective geometry in the early nineteenth century
I . h angle blsectOi 0 L , that the theory was established on a firm foundation. The basic result is
the parabo a IS t e dicular from B to the directrix.
is the foot of the perpen . . stated in what is known as the Poncelet-Stejner construction theorem.
THEOREM 5.9. Ponce/et-Steiner construction theorem. All of the

EXERCISES 5.5 constructions that can be performed with the straightedge and compass
. d perform the following constructions for Exercises 1-9. can be performed with the straightedge alone, given a single circle and its
Usmg waxe paper, . f iven segment. center.
. I t I',ne that is the perpendicular bisector 0 a g .
l. Fold a stralg 1
2. Construcr (he bisector of a given angle.
. d parallel to a given line not As Theorem 5.9 indicates, it is not possible to perform all of the
3, Construct a line through a given pomt an constructions of Euclidean geometry using just the straightedge. But it is
through the point. . t possibJe if one circle and its center are also given. This implies that the
, I f point to a line not through the pam.
4. Construct the perpendlcu ar rom a f length of the radius and the midpoint between two ends of the diameter
. ts to a parabola. given the directrix an~ the ocus.
5. Construct ten tangen are known. Despite the use of the one circle, many of the' concepts in
6. Construct the tangent al a point on a circle. this theory are from projective geometry (a geometry that does not
1. Construct the incenter of a given triangle. include the circle and its radius as invariants, as will be seen in
Chapter 7).
8. Construct the circumcenter of a given triangle,
As explained in the previous paragraphs. the attempt to limit the
i
9. Construct the orthocenter of a given triangle.
instruments used in constructions to the straightedge alone was not 4. . 1e w ith center A' and radius A' A. intersecting the
Construct t 1le Clrc
totally successful, since a single circle and its center were needed. The circle of step 3 at the points E and F.
second alternative is to limit the instrument to the compass alone. 5. Construct the circles with centers E an d F and melitiS EA, These
Obviously, it is impossible to draw a straight line with the compass meet at A and the required point G.
alone, so it must be understood that a line is completely determined
if two points on it are found Or given. Constructions with the compass The roof that this rather elaborate construction .results in ~he·
alone are called Moh,.-Mascherolli constructions. C. Mohr published the , P 1S b ase d on the fact that A' and G nre TI
correct pomt flwel"se FOlllfS
t.
first known account of these constructions in 1672, although his book with respect to the circle with center A and radius AB. le geollle ty
was not well known to mathematicians until 1928 when it was re- of inverse points is discussed in the following chapter, but the proor
discovered. Meanwhile, the Italian mathematician Mascheroni, during the used here does not require a knowledge of that concept.
last half of. the eighteenth century, had independently discovered the Triangles ;I'EA and EGA are similar, so
following theorem:
AA' AE
or AA" AG - (AEl'.
AE ~ AG'
THEOREM 5.10. All constructions possible by use of the straight-
Because
edge and compass can also be made by use of the compass alone.'
AA' ~ 2AB ~ 2AE,
2AE' AG ~ (AEl',
For a discussion of this theorem, see H. Rademacher and
{. O. ToepIitz, The Elljoyment of Mathematics. 2AG = AE ~ AB,
The following example illustrates a construction performed
entirely with the compass. and G is the midpoint of AB, as was to be proved.
Find the midpOint of a given segment by use oj a compass alone. The proof of Theorem 5.10 consists of sl~owing how .to fi~ld th~
points of intersection of a straight line and a circle and the .pOl1lt.s ~
1. Consider AB. as in Figure 5.24. Construct the circle with B as center intersection of two straight lines with the compass alone, sl11c.e It IS
, " and BA as radius. obviously simple enough to draw a circle and to find the pomts 0 r
. t·o of two circles with the compass alone. The method of
mtersec t. n I . t ,{ '.I1·tersection or· a circle (IIu/ a line liar Ihrouglt
COllsrruct111g f le POl11 S OJ ') . ' t
the center .IS shown .111 F'tgme 25 Let Band C be the gwen pom s,
. 5...
with A lhe center of the given circle.
A'
FIGURE 5.24
B. .c
~ ~ ~
With AB as radius, mark off three arcs AC. CD. DA', locating
A' so that AA' is a diameter of the circle.
}, Construct the circle with center A and radius AB. FIGURE 5.25
CONSTRUCTIONS 195
194 CHAPTER 5
For Exercises 7-11 give the fO f

1.
Construct circles with centers Band C. the given points on the line, the indicated exerci;es. P 0 of the constructions performed previously in
passing through A and intersecting again in D. 7. Exercise 2.
Following steps 4 and 5 of the previous construction, find;a point E B. Exercise 3.
2. 9. Exercise 4.
such that AD' AE ~ ,.2 for ,. the radius of the original circle. 10. Exercise 5.
11. Exercise 6.
(This construction must be modified if D is inside the original
circle.)
3. The circle with center E and radius EA intersects the original
circle in the required points X and Y.
The proof that X and Yare the points of intersection of the

given line and circle is left as Exercise 1, Exercise Set 5.6.
This chapter has shown that constructions in Euclidean geometry
are important in that they provide additional information about concepts
previously studied. In the last section, the somewhat surprising con-
clusion i;; that constructions also provide one transition from,Euciidean
geometry to the geometry of inversion and to projective geometry, which
will be studied in the next two chapters. This study will involve other
relationships between these geometries and Euclidean geometry, including
a classification based on transformations.
EXERCISES 5.6
I. Complete the proof of the Mascheroni construction of the point of intersection

ofa circle and a line not through the center.
With the compass alone. perform the following Mol1r~Mascheroni constructions

and brieRy describe each step. It is assumed that a line is given by IlwO distinct
points on the line.
2. Given a circle and a point inside the circle but distinct from the center,
find a point collinear with the center and the given point such that tbe
product of the distances of the two points from the center is equal to the
square of the radius of the giv'en circle.
3. Find a distinct point C on lB such that AS = Be. if A and n are given
points.
4. Find the midpoint of a given arc of a circle.
5. Construct the perpendicular to a given line at a given point on the line.
6. Construct the parallel Lo a given line through a point in the plane but not
on the line.
CHAPTER 6
THE GEOMETRY OF INVERSION

I,
:t
'I
.'
i:
6.1 BASIC CONCEPTS
With the exception of the finite geometries of Chapter 1, the

previous chapters have been an expansion of the relatively familiar
geometry of Euclid. In the last four chapters of the text, new
geomc::tries with some properties unlike those of Euclidean geometry are
discussed. The first of these is the geometry of inversion.
The geometry of inversion begins with the definition of inverse
points..
DEFINITION. Two points P and Q are i,werse poillls with

197
198 CHAPTER 6 THE GEOMETRY QF INVERSION 199
, 'f OP' OQ = c, for c a

respect to a fi xed collinear point 0 on the Ime I represents the square of the radius of the circle of inversion. In this
positive realllumber. pairing of points, two special cases become apparent. A point on the
circle of inversion is its own inverse point. In other words. these points are
d' the same direction from, O. The seff-iJwerse. The second special case is that no point on the line is the
Bo th distances are measure ~ . Figure 6 I is the
circle with center 0 and rad'lUS v c• shown 111 •• inverse of the center of inversion, since division by zero does not result
in a real number. This last statement shows the need 'for a consideration
of inversion from a more general point of view.
A review of the definition of transformation given in Chapter 2
shows that inversion as defined so far is not a transformation. since
Q
there is an exception to the pairing. But if the inversive plane is created by
including on each line an inverse point for the center of inversion, then
the definition oflr<insformation is satisfied. A line in ~he inversive plane is
FIGURE 6.1 formed by including, in addition to the ordinary points. a special point
calJed an ideal point, defined to be the image of the center of inversion.
. , .r' . t From With the addition of the ideaJ point, each point on the fine of inversive
, TI poml 0 'is ca II ed the center OJ IIwaSUJI. d' t
circle of illverSlOll. 1e d h t II e product or the 15 ances geometry has a unique image under inversion. The line of inversive
. y conclu eta 1 .
these statements, one ma f I'nversion is equal to ,the squaJe geometry is not an ordinary Euclidean line. The ideal point in inversive
, ' I from the center 0 . .'
of two Inverse pom 5 . ' II d the radius oj moe} 51011. geometry is not a concept from Euclidean geometry but a new concept
. h ' -de of II1verSlOll. ea e
of the radIUs of .t e ell . wa 5 considered positive. ., created to guarantee a desired one-to-one correspondence of points.
In this chapter, distances ale al y I ofi'nversion is numencal Because of the fact that the center of inversion in the inversive
. t' 11 theconcep
In its simplest mterpreta 10 ~. . II I always have the same plane is to have a unique image in the inversive plane, then each line
.. . I f two dIstances 1a
in nature. conslstl11g mel e y 0 t 'I '5 easy' to see that if one of the through the center must have the same inverse point for it, and the new
.' .. statemen I 1 .
P roduct. From thIS mtUltive 'h d must move toward It. point created must lie"" on each line through the original point. The
f' tl e center t e seeon .
l)oints moves away [0111 1 "lh'ls arithmetic interpretation. inversive plane is illustrated symbolically in Figure 6.2, with 0 the center
, I will illuslra e , f
A numerical examp e . d' lance of two umts rom
, (' F' re 6 I) IS at a IS .r
Suppose that pomt P In 19u - ' _ duct called the cons(altt oJ
the poult.' 0 and Ihat the constanl pi 0 •
1
itwersiol1. is 9. Then
2'OQ = 9, 1
and
.' .
OQ = 9/2, '1---,*=:----1
"h nstanl is unchanged bUI thllt OP = I. '. 1

Now suppose th~t I e co " is I then the numerical values
Then OQ = 9, If the radius of mvelSlOn • b found from the other 1 1
. als and one can e h' 1
of the distances afe reClproc. ..' . 0" may be related to t IS
by inverting the number. The name mverslO FIGURE 6.2
fact. " k it ossible 10 pair the poinls on a

The concept of mverSlOn 11:3 es Ph I' with a constant that' of inversion and with I (the ideal point on the inversive plane) the
line -in reference to one fi xed poml on t e me 'inverse of the center on any fine through O.
It shoul<:J be understood
THE GEOMETRY OF INVERSION 201
200 CHAPTER 6
been expected, concerns the inverse c~Jrve f~r a straight line not pnssing
thoroughly that the inversive plane, because it includes the one ideal point
through the center of inversion in the mverSlVe plane.
that lies on each line through the center of inversion, has some
properties unlike the Euclidean plane with which you are familiar. 'The
transformation of inversion in the inversive plane is a new transformation THEOREM 6,3, The image of a straight line n?t through the
that can be studied in its own right. As usual, it is good to speculate . '
center 0 f mverSlOn, U
oder the transformation of inverSIOn, IS a circle
about these matt'ers before reading on. passing through the center of inversion.
it The invention of the transformation of inversion is sometimes
credited to L. 1., Magnus in 1831. But prior to this time, Vieta in the '
1n F19ure ., let P and P' and Q and Q' be pairs ofI inverse
63 .
sixteenth century and Robert Simson in the eighteenth century were points with respect to the cir~ of inversi.on with c~nter, O. T le given
aware of elements of the theory. Some mathematici~ns give the major lioe whose image is desired is PQ, and PO I:S perpendlculm to PQ,
credit to Steiner, but his work on the subject was not published. Many
mathematicians worked during the 1830s and 1840s to develop the general
theory further.
The basic properties of the transformation of inversion depend
on the peculiar properties of the inversive plane, a plane with one ideal
point. 1t should be understood that the inversive plane as used here
includes real points only and not the complex points that will be
introduced later.
THEOREM 6.1. The circle of inversion is an invariant under

the transformation of inversion.
FIGURE 6,3
This theorem folJows from the fact that each point of the circle of By the definition or inverse points,
inversion is its own inverse under inversion with respect to that circle.
The circle is pointwise invariant, according to the following definition. OP'OP' = OQ'OQ',
or
OP OQ'
, . DEFINITION. A set of points is pointwise invariant for a trans-

formation if every point is its own image.
,
OQ = OP"
' I'les th a t L.:>.
AOPQ - • L
A OQ' p' becallse they
-".
This proportlon lInp 'J.. •
THEOREM 6.2. The inverse of a line through the center of . Th A OQ'1' is a right triangle, with the
have an angle In common. us L.:>. . '
inversion is the same line. vertex of the right angle at Q'. If Q is considered a ~anabl~ P01~t 00, the
. r then Q' is a variable vertex or a right tnangle inSCribed m a
glve~ . Ill", I'th 7'\"i'iI the diameter of the circle. The proor of the converse
This line is not pointwise invariant because, in general, a second semlClfCIe. w VI' . ' . 1 nent
of Theorem 6.3 consists of reversing the steps tn the prevlOu.~ at gt I
distinct point on the line is the image of a given point on the line.
The fact that the inverse points and the center of inversion are collinear and is lert as an exercise. .' t'
The fact that the image of a line under inverSion IS some 11.nes
is actually a part of the definition of inverse points.
a circle rather than a line shows that the property of being a stnllght
A more significant theorem, and possibly one that may not have
202 CHAPTER 6 THE GEOMETRY OF INVERSION 203
line is not always an invariant property under the transformation of h THEOREM 6.4. The image under' .
iilVersiol1. This is a major difference between the geometry of inversion t rough the center of
inversion I' • I mverslOn of a circle not passing
sa CIrc e.
and ordinary Euclidean geometry. The relationship between circles and
Hoes under inversion -can be further extended by investigating the inverse
of a circle not passing through the center of inversion. Before reading
ahead. you should predict what this image wiIJ be.
FIGURE 6.5
Tllis section concludes with one .

.transformation ofinversion. Su ~the~ unportant property of the
Jnverse points with respect to pp~se, as ~n FIgure 6.5, that P and P' are
formation takes P to its inver: gJv~n cl~cJe with center O. The tran;-
FIGURE 6.4 . [ormation is applied again the ~ poml P. If the same inversive trans
• Image of P' i P Th -
prod uct of an inversion folio d b s .' e result is Ulat the
In Figure 6.4, let 0 be the center of inversion, with circle 0' The jnversion transformation ~e call~dthe ~ame Inversion is the identity.
~
the given circle. For any line OP intersecting the circle in two distinct trans/ormation a/period two b an mvolutory trans/ormation or a
h' . • ecause two sue' . •
points and passing through the center of inversion, if P' and Q' are the t e l?entJty. It should be observed that a . eess~ve apphcations result in
OWJllOverse. which provides another 0 d n InVerSIOn transformation is its
inverse points for P and Q. then correspondence. g 0 reason for the name given to the
Op·Op' = OQ'OQ' = c,
the constant of inversion. Also EXERCISES 6.1

OP' OQ = (OA)2 = k,
Complete the fOUowing table:
a constant equal to the square of the length of the tangent from 0 to
Distance fi'om
the given circle. celUer oj inversiOll Distance of
Radius oj circle inverse poine /rom
to original poiltt of inversion cenrer of inversion
1. 3 4 ?
2. J 3 ?
which is a constant. 3. 3
--J... The situation reduces to a familiar transformation, a homothetic 2 ?
4. 3
transformation with center O. This means that, when P is a point on the ?
5. 9/8
given circle, the image P' is a point on a circle that is homothetic 3(7 ?
to the given circle; thus elk is the homothetic ratio. This completes the: 6. 2/3 ? 5/6
proof of the next theorem. 7. ? 9/5 6(7
204 CHAPTER 6
8. Do each two points in the inversive plane determine a ullique line? r,

9. Describe the location of the inverse of any point inside the circle of c, T,
inversion.
10. When is the image of a circle under inversion a circle, and when is it not?
II. Prove the converse of Theorem 6.3.
12. What is the product of inversive transformations if the same transformation
is used as a factor an even number of times?
13. What are possible images under inversion for a sel of points consisting of FIGURE 6.6
two distinct intersecting lines?
14. Describe the inverse of a circle outside rhe circle of inversion and concentric secting curves are two intersecting curves .C I ' an~ C 2 '· The angle
to it. between their two tange.nts at the point of mtersecl10n has the same
IS. Suppose a curve intersects its inverse curve. Where are til(: points of inter- measure as between the two original tangents. .
section? The proof of Theorem 6.5 depends on provmg thai the. angle
16. In Section 5.6, the construction was given to find the midpoint of a given that a curve makes with a line through the center of mverslOl~ IS
...
~ segment by use of a compass alone. Use the ~oncept of inversion to prove congruent to the angle the inverse curve makes with the same lme.
that this construction results in the required point.
6.2 ADDITIONAL PROPERTIES AND

INVARIANTS UNDER INVERSION
In the ptevious section, it was found that the cirde of inversion

is pointwise invariant under inversion and that a line through the center A
of inversion is invariant, but not pointwise invariant. A set of points is FIGURE 6.7
pointwise invariant if each point is its own image under a trans-
formation. It was further found that coincidence of points and curves is In Figure 6.7, let curve PQ and· P'Q' be inverse curves with O. the
preserved under inversion. These facts provide the basis for the continued .
center of inverSIOn andWIt' h P•P' and Q• Q' pairs of inverse pOInls.
_ .
investigation of invariant properties under the inversion transformation, . t P P' Q' Q all lie on a circle (the proof of thIS
The four pom s • I •
One of the most useful of the invariants under inversion is indicated statement is left as an exercise), so opposite angles are supplementary.
in the next theorem.
Thus,
L OPQ ;" L P'Q'O.
THEOREM 6.5. The measure of the angle between two inter-
secting curves is an invariant under the transformation of inversion. Applying a fundamental conc~ from analysis, one may conclude
.. f '~OQ' OP as Q approaches P along lhe
that the limiting pOSItIOn 0 IS • . .
. .' I The tallgents FA and VA are the limiting pOSItIOns of ~
The meaning of this theorem is shown symbolically in Figure 6.6. ongma curve. .. f OP
.=~ P
~Q d p"-Q' Thus the limiting posJtJOn of the angle rom
The angle between two curves eland C 2 is defined to be the angle m . , 'Ad
between their tangents at the point of intersection. One of the two to the original curve is L OPB. which is ~ongruent to LOP . le
supplementary angles must be designated. The images of the two inter- limiting position of the angle from OP to the Image curve.
206 CHAPTER 6
It is important to observe that, although the measure of the

angle is preserved under inversion, the direction of the angie is reversed.
In Figure 6.7, the angle from OP to PB is measured in a clockwise .0'
direction, whereas th~ angle from OP" to P'A is measured in a O'
counterclockwise direction. For this .reason, inversion (like reflection) is A
another example of an opposite transformation. eO"

As a corollary to Theorem 6.5, two curves that intersect at right
angles have images that also meet at right angles. Orthogonal circles.
two circles intersecting at right angles. playa very important role in the
FIGURE 6.9
theory of inversion. A fundamental concept is included in the next
theorem.
0 be th
o" two In Figure 6.9, let
circles orthogonal
. If"
to ,'t de. CIrc e ~ mverslon with 0' and
an mtersectmg each th
.
THEOREM 6.6. A circle orthogonal to the circl'; ·of inversion is
B. It can be proved that point O' II" •. 0 er at A and
an invariant set of points (but not pointwise invaripnt)- under the it is a point from which con ~s co mear With A and B because
inversion transformation. 0' and 0". (The proof of thisg:~:~ tang~nts can be drawn to circle
from Theorem 6.6 A and B . ernent . IS left as an exercise.) Then
• are mverse pomts •
. Recall that the concept of inverse .-
In the last section of eha t 5 pomt was encountered briefly
p er on construction S I'
A f-_ _;-'-::'I--F---'.D IlOW a point inverse t ' . . pecu atlOn about
o a gIven pomt n . ht
could lead ,to discovery of an add'f lIg actually be constructed
lOnal
points that is implied in the n t t'l fundamental property of inverse
ex leorern.
FIGURE 6.8
. THEOREM 6.8. The inverse point for . .
lies on the Ime J'oining th . f. a pomt outside a circle
. ~ pomts 0 mtersect' f
In Figure 6.8. ,circle 0 is the circle of inversion and circle 0 ' the potnt to the circle. Jon 0 the tangents from
is any circle orthogonal to it. If a line through 0' intersects circle
0' at C and D, then OC'OD = (OE)2, for E a point of intersection
'of the two circles. But since lJE is also the radius iof the circle of
inversion, C and D are by definition inverse points with respect to the
circle of inversion.
For each point of fDE. the image under inversion is a point of o (--f-+-_"""=:::" p
f.CE, with the endpo~nts F and E invariant points. Thus, circle 0' is
a set of invariant points.
A second useful theorem about orthogonal circles under inversion 8
is the following: FIGURE 6.10
THEOREM 6.7. If two circles orthogonal 10 the circle of In Figure 6.10, P is th .

inversion intersect each other in two points, these: two points are and P' are inverse points beca:s:x~~r~al pomt. We can prove thar P
inverse of P lies on OP. 's known by definition that the
inverse points.
'"
208 CHAPTER 6
........
In Figure 6.ll, if P'Q' is the polar of P. then it must be shown
From similar right triangles OAP and OP'A. that the polar of Q' passes through P. If Q is the inverse point for Q'.
then all that is needed is to show that QP is perpendicular to ag.
(OA)' = op· OP', The two pairs of inverse points. P. P' and Q. Q'. lie on a circle. Because
LPP'Q' is a right angle, L Q'QP. the opposite angle in the inscribed
so P and P' are inverse points. quadrilateral, is also a right angle. The theory of poles and polars is also
From Theorem 6.8 and Figure 6.10, it is ·possible to devi!?e a important in projective geometry. Some extensions of the theory ~re found
n~ethod of constructing the point inverse to a given point, whether the in the following exercises. .
given point is inside or outside the circle. These construction problems
are included in the exercises.
EXERCISES 6.2
. DEFINITldN. In inversive geometry. the line through an inverse
l. List the possibilities for the images under inversion of two lines meeting at
pomt and perp'endicular to the line jo'ining the original point to the
right angles at " point other than the center of inversion.
ce~ter of the circle of inversion is called the polar of the original
2. In the proof of Theorem 6.5, prove thai P. P', Q'. Q all lie on a circle.
pOInt, whereas the point itself is called the pole of the line.
3. In the proof or Theorem 6.7, prove lhtll point 0 is collinear with points A
and B.
In Figure 6.10, ill is the polar of P and P is the pole of ill. 4. Given a point outside a given circle, constnlct its inverse point with respect
Recall that the words pole and polar were encountered in a different
to the circle.
setting in one of the finite geometries of Chapter 1. 5. Given a point inside a given circle. construct its inverse point with respect
In Figure 6.11, i~ P and P' ar~nverse points, then Fo: the circle.
1.0
is U,e polar of P and P IS the pole of P'Q'. From Figure 6.11, you 6. Prove thaI the polars of every point all a line are concurrent ill the pole or the
flll.ght surmise that the reciprocal relation between poles and polars
line.
eXists as stated forn,lally in the next theorem. 7. Construct the pole of a line not intersecting the circle of inversion.
S. Construct the pole of a line intersecting the cin:k of inversion in two
distinct points.
Q' 9. A triangle is self-polar with respect to the circle of inversion if each side_is
the polar of the opposite vertex. Ex.plain how to construct a triangle self-
polar to a given circle.
to. Describe the inverse of a circle with respect to a point on the circle.
oL....~-+--<1p· t 1. Through two points inside a given circle but not collinear with the center,
construct a circle orthogonal to the given circle.
6.3 THE ANALYTIC GEOM,ETRY OF

FIGURE 6.'1
INVERSION :'
. THEOREM 6.9. If a second point is on the polar of a first, For the Euclidean motions, it was possible to write a set of
With respect to a given circle of inversion, then the first is on the polar equations that clearly showed the one-lo-one correspondence. 1n that
of the second with respect to the same circle.
. . terms
expressed In 0f a set of simul- But
case, the transformatlOl1s were f t' of similarity was also
. t' 5 The trans onua Ion .
taneous Imear equa Ion: r I'
. II usmg sets 0 m e a r ·
equations The transformatIOn
d
described algebralca Y h d f an analytic point of view, an
. I be appraae e rom
of inversIOn can a so .. . . ht . to the nature of the
this approach will give add1t1onal mSlg m and if both sides are multiplied by x'. then x"(x' + y')' = x','" Taking
transformation. the square root of both sides shows that
y
or
PIx'. y) xrZ
x' = 2 2'
X +y
Similarly. the expression for y' can be found so that the next
theorem is proved.
THEOREM 6.10. The equations for the inversion transformation

FIGURE 6.12 of a point P(x,y) into P'(x',y') relative to the circle x' + y' = 1" are
The equations for inversion can be de:etopedd. usjn~ :~gu:: ~~~~

t the origin and with fa lUS r C
The circle with center a . t P and P' have coordinates
sidered the center ofinversion. Inverse pOUl S
(x. y) and (x'. y'). respectively. The equations in Theorem 6.10 show clearly that the image of the
center of inversion is the ideal point in the inversive plane, since
2
Since Op· OP' = 1". x + y' = O. and x' and y' do not have real values. They also show that
for points on the circle of inversion, x = x' and y = )".
l)x' + )" )(Jx" + y" )= ,.'.
or
(x' + y'),(x" + /') = ,.4. EXAMPLE. Find the inverse of (2,3) with respect to the circle
x2 + y' = 9.
Beeause P and P' are collinear with 0, x'ix = ill'. . x'y = xy'.
9(2) 18 • 9(3) 27
or X'2 y = X 2 y /2; therefo"i'e,
2 x'=·_-=-.
. 4+9 13 Y=lT =13
.'
. (x' + (x'/'
7 + ") =,..
y') y 4 The inverse point is
~(x, + y2)(X'Y" + y"y')= 1'4, (.!!!.

13
27).
13
Y
" Solving the equations in Theorem 6.10 for x and y provides
; . (x' + y')(x' + y') = I'4 • an analytic proof of the fact that the inverse transformation is its own
Y
THE GEOMETRY OF INVERSION- 213
212 CHAPTER 6
inverse. The result is that x and Xl can be interchanged and y and y'
can be interchanged so that
(3,0)
x-4=O
The requirement that the circle of inversion have ils center at the
origin is of course not mandatory. The substitution
FIGURE 6.13
x" = X + h. y" = y + Ie
illustrates the wide variety of possibilities. Find the image of the parabola
shows that more general forms of the equations for inversion are l . " t tl c'rcle- -Xl + yl = 4. (See
= 4x under lflverSlOn WIth respect 0 le I.
(x" - 11)1'2 , (y" _ k).,' Figure 6,1.4.)
X' =(,-x-;'-"----';-;!J)'"-+-7(y-;'-"---I"",)" y = (x"-h)'+(y"-I<)". y
It is instructive to use the analytic formulas to verify the

inversion theorems of the last two sections for particular numerical
examples. For ~xample, what is the image of the line x -* 4 = 0 under
inversion with respect to a circle with center at the origin and radius
three? (See Figuro 6.13.)
The equt.tion of the image is obtained by substituting
9x'
X'2 + ]1'2 FIGURE 6.14
for x in the equa~ion of the line. Thus, The substitution for x and y in the equation of the parabola
9x' _ 4 = 0 results in
X'2 + y'2 •
(4y')' _ 4 4-,'
9x' ~ 4
(X'2 + /2)2 - X'2 + y'2
X'l + y'2 •
This equation can be somewhat simplified, as follows:
9x' = 4(x'2 + y'2),
or
16y" = 16x'(x" + ),,'),
4x" - 9x' + 4y'2 = O. y'2 = x') + X'/2.
This is a circle through the origin, as expected. y.2. _ x'y'2 = x').

The equations of the transformation of inversion and its inverse ,'2
X'3
= ___ '
transformation also make it possible to find the equations for the )
. 1 - x'
inverse of curves not covered in previous theorems. An example
The inverse of the parabola in this case is not a ~ara~ola, ,DO,r eve~ a 1. Show that the product of their distances from the origin is one.
conic, so the property of being a conic is not an'InvarIant In inverSive 2, Show that they are collinear with the origin.
geometr~hen the study of inversion is extended ;rom the real Planet~~ Demonstrating that z and z' are inverse points is left as an exercise.
I
the campIex pane, a ..:other unexpected simplIficatlOn appears In . Inversion with respect to the circle with center at the origin
'" and radius in the complex plane is accomplished by the transformation
equation of the transformation.
r'
z'
Z·(c.d) It is somewhat surprising that the numerical work of finding an inverse

point in the complex plane is considerably simpler than is finding it in
the real pJane. For example, the inverse of 3 + 4;. with respect to the
Z(o.b) circle with center at the origin and radius two, is 4/(3 - 4i), which can
-t----~---4----R be simplified as follows:
o
(_4_)(3 ++
3 - 4i 3
4i) = 4(3 + 4i) =
4i 9 + 16
!2 + ~ i.
25 25
FIGURE 6.15
EXERCIS ES 6.3
I F' e 615 let the real and imaginary axes be given. If
n Igur ., .. d 'f ,_ c + di is the inverse
tl~e circle of inversion is a umt cIrcle, an I Z - . Use the equations for Im'ersion with respect to a circle with center at the origin
of z = a + bi with respect to this circle, then the followmg theorem to find the inverse of the following real points.
shows the relationship between z and z'.
Point Radills ojjllt'el'sioJl
THEOREM 6.1 L The transformation of points in .the complex 1. (3,4)
plane, with respect to the un~t circle with center at the on~m, has the 2. ( 1,1) 2
. z, =
equation l/z- whe,'e 11 is the conJ'ugate of z. If z = a + bl, then 3. (0,3) 3
1 1 a + bi 4. (7,1) 4
z' - - = - - - ' - - - .
-2 a-bia+bf
a + bi
For Exercises 5 and 6, find the inverse of the given point with respect to a circle
with center (2, 3).
j
5. (6,2) 2
To show that z and z' are inverse points with respect to the unit
6. (5,5) J
circle, it is necessary to:
216 CHAPTER 6
In Figure 6.16, OC' OD = (OB)', or

solving for x and y in the original equat~ol:s~erstve transformation by actually
7. Find the equations for the inverse of lh' .
OD OB
For
. h each set of .pom
. t s given
• . Exercises 8-13 find the image under inversion
In
OB = oc'
wll respect to the circle x 2 + y2 = 9. '
8. x-2=0: 9. x - 5 =0
10. x +y + 2 ~ 0 it. y2=4x
12. x 2 = }'
In lhe complex plane, find the inverse of II

with reference to a circle with ce t I .1: . .
pomts gJVen
.
III
Exercises I5-lB,
n er at t Ie ongm and with the given radius.
Point FIGURE 6.16

R(ulilts oIinversion
15. 2 - 5;
" 16. 3 + 7; By a theorem on proportions, if alb = cld, then
17. 1 - 2; 2
a + b c+d
18. 3+ Iii 3 ~=-;=d'
so that
OD + OB OB + OC
6.4 SOME APPLICATIONS OF
OD - OB OB - OC
INVERSION
or
Three of the most appal' t r .
been mentioned but they need toe: ~~p l~atJO.ns of inversion have already
areas. e Iste< agam before we proceed to new
so that the ratios of division of AB by C and D are equal (directed
1. The theory of poles and I" . segments have not been considered here). Note also that points A and B
Mascheroni constructior!:°e~S I~ used In pr~jective geometry. divide CD internally and externally in the sarn~ ra~io; tllltS, the circle of
2.
3.
as a 'basic technique in ord r
y constructIOns of Ibe inverse poim
The finite geometry of D er 0 c~ry out many other constructions.
inversion is the circle of Apollonius with respect to the two points C
and polar. ' esargues ( hapter 1) used th(~ concept of pole andD.
The fact that an inversion transformation sometimes transforms a
straight line into a circle, or a circle into a straight line, is the basis for
A new fact about inverse pomts
. physical applications of inversion in linkages that change linear motion
shows an applicatl'on to an
. into curvilinear motion or vice-versa. The harnessing of the tides, the
ear 1ler construction problem.
driving of a locomotive" and the production of electricity are all
illustrations of one lype of motion changed into another type of motion.
THEOREM 6.12. Two inver . . The mechanical problem of devising linkages to do this sort of thing has
divide the diameter on which th /e .POllltS WIth respect to a circle
same ratio. . ey Ie Internally and externally in the concerned engineers during the past hundred years. One such device is
218 CHAPTER 6
The more elementary theorem is illustrated . . F·

Peucellier's cell, whose construction is based on the theory of inversion. i t d fi III Igure 6 18a
whereas the Ilver e gure, representing Theorem 6.13, is shown., in
The device is illustrated in Figure 6.17.
A
Qed))
A C B E
(a) (b)
B
FIGURE 6.18
FIGURE 6.17
Figure 6.18b. It is very important to re r
The quadrilateral AP' BF is a rhombus, so Dc is the perpendicular concerned only with the ,.elatl· I .
. .
a Jze that the new figure is
ons ups among the i .
It IS not concerned with the reI t· ons h. mage sets of pOll1ts-
bisector of All
the new nor with areas ane! at'h Ip between the original figure and
o er extraneous t t b
Op· OP' = (OC - PC)(OC + CP') the relationship within the new fi u. tl a~ ors ut only with
new theorem. g le - 1at makes It possible to state a
= (OC)' - (PC)'
. l5E is the image of OC, whereas th . I . .
= (OA)' _ (AC)' - [(P A)' - (A e)'] IS the image of in
Th ,.. I . . e clrc e WIth dIameter DE
e CIt c e of mverslOn" .
IS mv~nant. !he fact that
t-t •
= (OA)' - (PA)' AB is tangent to the circle means that its i

to the circle of inversion since 1 . mage (a Circle) is. also tangent
But in the mechanical device, which consists of the six rods Figure 6.19a sh~ws th:": ~Isr ale preserved under inversion.
forming the rhombus as wen -as OIl and OB. the distances OA and an angle inscribed in a semicirci g. e f~rhthe theorem that states that
P A arc constant, and thus P and P' are inverse points. If point P e IS a fig t angle. Figure 6.1 9b shows
moves around the circle, P' traces a straight line.
Quite a different application of inversion is that of lfinding -
new theorems by inverting familiar ones. More precisely. the figure for a . B
@,
familiar Euclidean theorem, when inverted, may yield a new theorem by 4
suggesting the corresponding properties. Even further. the new theorem. ' A C
does not have to be proved from the beginning; it is true because of:. o 2
1
the proof of the original theorem and the accepted properties of inversion..
Examples will illustrate the technique. 3
One of the elementary theorems from geometry is the stateI11enl~, (a) (b) (e)
that if a line is perpendicular to a radius at its endpoint on the FIGURE 6.19
then it is tangent to the ci,rFle. A new theorem by inversion

frol11 taking the circle as the circle .ofi~version. It is stated as follows. the result of inverting the sets of oints i
the center of inversion with ~
.
. n Fl~ure 6. 19a.with respect to 0
Figure 6.19b, circle l' and Iin~ ;lven ~lrcJe. the Circle of inversion.
THEOREM 6.\3. The circle whose diameter is the radius of circle 3 as its image a d liC h ar~ mvanant sets of points, in
second circle is internally tangent to the second circle. ,n as CIrcle 4 as its image. The new
220 CHAPTER 6
: theorem already established. The classic example of this is in lhe proof of

theorem involves the conclusion that circles 3 and 4 will be orthogonal. :'
Feuerbach's theorem.
The formulation of the wording of the theorem is left as an exercise.
The choice of the center of inversion was arbitrary, and various· · I . The nine-point circle -of a
different theorems can be written if other centers of inversion are chosen. THEOREM 6.15. FeuerbaciI s t 1e01 em. ,',
I . . Ie and to each of the three excllcles·
As might be expected. choosing a key point of the figure itself as the triangle is tangent to t le mClre
center of inversion results in the greatest simplification, as in the previous of the triangle,
example. Figure 6.19c shows the result ofinversion with respect to a 'point
not on any of the given lines or circIt;'>. and l' be the centers of the inscribed
In Figure 6,20, let I
Verify the following statements with respect to Figure 6.19c: with A" the center of the side BL:, If
circle and anyone excircle,
1. The image of circle 0, a circle not through the center of inversion, is A

a circle (circle 1).
2. The image of the line lc is a circle (circle 2) that goes through the
center of inversion and is orthogonal to the image of circle O.
3.. The image of the Hne AB is a circle (circle 3) that passes through
the intersection of. circles 1 and 2 and also through .the center of
inversion.
4. The image of the line 1iC is a circle (circle 4) that passes through
the intersections of 1, 2 and 1, 3 and also through the center of
inversion.
The new theorem will involve four circles, and the conclusion will FIGURE 6.20
be that two ·of them are orthogonal. This new theorem, valid in
Euclidean geo~etry, is stated in somewhat complex wording, as follows: 'I Be II A' is also the midpoint
ID and l' E are radii perpendlClJ ar to ,len IJ15 . . . C E
of m. This can be shown as a result of the. fact that D ~ J '
, . I to Idf lhe perimeter of the lliangle
THEOREM 6.14. If circle 2 is orthogonal to circle 1, if circle 3 since both have measmes equa ,
passes through One intersection of circles 1 and 2 and also intersects minw) the length of side Ac. . . and jf
circle 1 in another distinct point, if circle 4 passes through the second Now if A' is taken as the ~enter of tn,:,er~l.on ,
intersection of circles 1 and 2 and the second intersection of circles 1 and 'D __ A' E is taken as the radius of inverSion. then the mcn de ~~nd the
A th onal to the CIrcle 0 f
excirde are both invariant, Sll1ce they are o~ ~~ ~ F the foot of the
0- •
3, and if circles 2, 3, 4 have another distinct point in common, then

circles 3 and 4 are orthogonal. inversion. The nine-point circle passes throug.l A a n : . erse of
. r. A B t A' is the center of inverSIOn and G IS the mv
al~ltude 10m . u .' I be established that G
F with respect to the circle or 11lVerslO~. ( t C31~ 'vides it eXlernaBy.)
In the discussion thus far, inversion has been used to get new
theorems from old. In general, the new theorems have been more com- t ' 'des DE internally in the same ratIo
(IVI . .
that F dt
I' tl ough G Because
'. f the nine-point Circle IS a me Ir .
plicated because some· of the straight lines have been inverted into Thus, the Image odd . [sion the angle between the image of the
circles. In the application of inversion called proof by inversion. the a~gles a,re p~'eserve u~ ~r ll1~e ruen~ to the angle that the tangent to the
procedure is somewhat reversed. Here the idea is to prove a more nme-po~nt ct.rcle and ~C IS co ~ll Be It can be shown lhut this tungent
nine..pomt c!fcie at A makes Wi 1 •
complicated theorem by inverting the figure and then using a simpler
222 CHAPTER 6
ite side of the orthic triangle (the triangle THEOREM 6.16. The image under inversion with respect to a
is parallel to the oppo~ 't d ) This line is antiparallel to BL. Also, sphere of a plane not through the center of inversion is a sphere passing
through the feet of the a tl u es.. . 11 I lines make congruent angles through the center of inversion. and conversely.
I n that tWO anupala e f I
it can be S lOW tween the other two sides o. t lC
with the bisector of the . a~gl: be internal tangent of circles 1 and
ther A special case of inversion with respect to a sphere is known as
inscribed quadrangle. But It IS t ,e 0. h BC that IT' does so it is the
l' that makes the same a1lgle whIt. oint circle TI~is means that stereographic projection and is very useful in map making. This melhod of
H.i h t' the imag~ of t e nme-p . projection is illustrated in Figure 6.22. Plane a is tangent to both the
tangent t a. IS . nt to both the incircle and the excircle.; In a
the nine-point circle IS t~nge. b hawn to be tangent to the bther sphere with 0 as center and the sphere with B as the center at the
. '1 ar way
511111 ( • tl,e nine-pomt cIrcle can e s ., common point A. Plane !X and the sphere with 0 as center are inverses
excircles. Ie of proof by inversion appears in Chapte~ 9 o~
~nother examp In this example, inversion is used in the POIncare
nOIl-Euchdean geoo:etr y. . fi d the sum of the measures of the
model for hyperbohc geomeh y to n
angles of a triangle. "th look at inversion in three
TI is section concludes WI a . .
. . ' Analogous to the circle of inversion is the sphere of InverSIOn
dimensIOns.
FIGURE 6.22
"\ with respect to the second sphere. The effect of this inversion is to
p
-----.\.- establish a one-to-one correspondence between the points on the sphere
" ,
o.........rl1 and on the plane. One of the important features of a map prepared in
this way is that all angle measures are preserved under inversion.
I
Stereographic projection is also used in obtaining the Poincare model
for hyperbolic geometry mentioned above.
The inversive geometry of this chapter has proved to be a very
FIGURE 6.21 useful geometry. Though the invariant properties are not 'aU the same as
those of EucJidean geometry. the two geometries do have many common
. . . 621 If P and P' are inverse points with respect to a concepts. Inversive and Euclidean gtiometry are distinct geometries in the
as shown III Flgm e . .
sphere of radius r, then sense that neither i~ a special case or the other. The next two geometries
to be studied, pl'Ojective geometry and topology, are generalizations of
op,oP' = r2. Euclidean geometry and include it as a special case.
.' . nsions can be seen to be a cross section pf the
Inversl~n m .two dIme. 'th the plane passing through the center of
three-dimensIOnal drawmg, WI" EXERCISES 6.4
the sphere.
l. Explain the special case of llleorem 6.12 if one of the points is the center
. Corresponding to inversion relationships between the line land tl~e of inversion.
circle in two dimensions are t~oser::::e~~nt:~e ~::tet~~~r~~~. sphere Ill .. 2. Use cardboard and papel' fasteners to prepare a woi'king model of
Peucelliel"s cell.
three dimensions. An example IS P
CHAPTER 7
224 CHAPTER 6
3. Invert the figure of a line perpendicular to a radius of a cirde at its end~ PROJECTIVE GEOMETRY
point on the circle, with respect to a point on the tangent, and state the
resulting theorem.
4. Invert the figure of a triangle inscribed in a semicircle with respect to one of
the endpoints of the diameter and state the resulting theorem.
For Exercises 5-8, write new theorems by inverting the figure.

5. If the opposite angles of a quadrilateral are supplementary, the quadrilateral
can be inscribed in a circle.
6. The line joining the centers of two intersecting circles is perpendicular to their
common chord.
7. If two circles :ue tangent to the same line at the same point, the line
joining the centers passes through the common poiut of tangency.
8. The altitudes of a [riangle are concurrent.
Exercises 9-t2 concern the proof of Theorem 6.15.

9. Verify that BD ;;: CEo
10. Establish that G divides l5E internally in the same ratio that F divides it
externally.
It. Show that the tangent to the nine-point circle at a midpoint of a side is
parallel to the opposite side of the orthic triangle.
12. Prove that two antiparallellilles make congruent angles with the bisector of
the angle bisector between the other two sides of the inscribed quadrangle.
D. Under inversion, with respect to a sphere, what is the image of a plane
through the center of the sphere',
7,1 FUNDAMENTAL CONCEPTS
The mathematician thinks of Euclidean geometry as a very special

case of more general geometries. In this chapter, the first of these m~re
. .. '. xplained The first secl10n
general geometnes, projectIVe geometry, 15 e '. 1 5
introduces some of the basic concepts, and the second sectIon cleve op
the subject rrom an axiomatic point of view. .. f
Special cases of projection were discussed III rel~tton to s~me 0
the finite geometries of Chapter 1. In ordinary analytiC ~eon:ettY, the
oj'ection of a segment onto an axis is used, As illustrated tn FIgure 7,1,
pr , d' I r AA' and
As is projected onto the .x-axis by droppmg perpen leu a ~ I
B' Although the lengths of As. and ;t'if are not the same. 10 genera,
B .' 225
PROJECTIVE GEOMETRY 227
226 CHAPTER 7
Last Supper by Leonardo D Y' . .

Figure 7.3. a InCI, shown wUhout complete details in
A
1'~ -0'!..::::",
-- B
I I
I I
I I
I I
I :
I I
I I
A· B·
FIGURE 7 1
there is a one-to-one correspondence established by the projection

between points on the two segments.
FIGURE 7.3
Another familiar use of the word "projection" is in connection
with a motion picture projector.. Here, the picture on the film· is
projected onto the screen, as shown·in Figure 7.2a.lt should be appal'ent Famous painters dudng the period of b .
to use the concept of pe' .. t .e RenaIssance attempted
that pictures on the film and on the screen will be similar to each other,
realistic TI,ey knew tha/:rectIVIt;. to ~ake their paintings look more
since the shape remains invariant while the size changes uniformly. For view lo~ked d I 10 'pos, IOn 0 the artist determined how the
,an t ley attempted to put on
actually saw. For e x . canvas what the eye
position to one Sidea~p:~~I~:~I~ °t~ser;e a rectangular table top from a
~
Ughl ~ Screen
Similarly, a circle viewed f • e op no longer looks rectangular.
rom an angle no longer 10 I
~oJ'k of the Renaissance painters can be said
. I
0 (.5 elreu ar. The
mcentive for the development f . . to have prOVided part of the
. 0 .proJectIve geometry.
(b)
(a)
Figures 7Aa and 7Ab show sketches ill wh,'cll the center of
FIGURE 7.2
Figure 7.2h, however, the picture on the screen will be distorted because
the film and the screen are not parallel. The figures are no longer
similar, and it is no longer apparent what properties of figures reniain r I
~
unchanged. For example, even the ratios of distances are no longer·
invariant.
Although the example in Figure 7.2b may at first seem completely
new, many people will be able to describe the figure as a perspectiIJity.
as the term is used in art. Artists talk about centers of perspectivity to
(a)
show that the lines in' a painting converge at a particular point that (b)
becomes the center of attention. A famous example is the painting df the FIGURE ]A
228 CHAPTER 7
1822 Most of the work on this book was done while POllcelet was."
perspectivity is not at the center. You should be able to locate this .'. . after the wars of Napoleon. Other advances III
point in each case. prts?ne.r 111 RUSSIa, ade during the nineteenth century. Karl von
In this text, geometries have been studied on the basis of their projective geometry wer~ m. etry could be developed without the
d h dhow projective geom . ."
invariant properties under a group of transformations. You should realize Stau ; :n;W;etrical basis for measurement. Felix Klein gav~ pro!ectIv~
that projective geometry can also be studied using this same approach, use 0 . I it deserves in the classlficatlOn 0
although historically the topic was not developed that way. geometry the pro111111ent P ace f 1872 Since that time postulational
It is important first to explain what is meant by a projective geometries in his Erlanger program 0 . b "tt' n and finite
transformation. Figure 7.5 shows several perspectivities in the same developments of projective ge~metry have een Wit e ,
drawing. The points on 11 and 12 are perspective. with A the center of projective geometries have been mtroduced.
EXERCISES 7.1
c 1. Write an equation relating the lengths of A'B and ;f1f in Figure 7.1.
2. Describe the location of the center of pe:rspectivity in
a Figure 7 Aa. b. Figure 7A b . . ..
. h sketch of a picture looking down a road with sets of .utlht y
3. Malke a ro.uthg side Make the lines in the drawing converge nl a pOIll! on
poesoneI er .
FIGURE 7.5 the horizon.
. 4-11 which of these properties, which are invariant in Euclidean
For E.xercises, .... t ?
geometry, also seem to be invanant III proJecuve geome .ry.
perspectivity. The points on l2 and 13 are perspective, with point B the
4. Measures of angles.
center of perspectivity. and the points on 1;, and 14 are perspective, with
'4
point C the center ofperspectivity. In general. the points on 11 and are 5. Collinearity.
I not perspective. Instead, they are defined to be projective. From this 6. Measures of area.
'.~ point of view, a projectivity is considered a chain of perspectivities. 7. Ratios of distances.
t;'., 8. Property of being a circle.
,
.i;
,~
DEFINITION. A projectivity is a finite sequence of per- 9. Property of being a square.
spectivities. 10. Property of being a triangle.
11. Property of being a rectangle. . .. (
. .' to Fi ure 75 in which several perspecttvIt!es resu t
In projective geometry, the invariant properties .for a projectivity 12. Make a drawmg sl.mdar. d
g. t back into points on the same line.
in the POllliS on a hoe bemg proJec e
are studied. While it is not obvious which properties may be preserved,
it should be evident that there is a one-to-one correspondence and that
collinear points remain collinear even though distances and relatiwi
7.2 POSTULATIONAL BASIS FOR
distances may change.
PROJ ECTIVE G EOM ETRY
This intuitive discussion has introduced several of the basic
concepts of projective geometry. Although it grew out of ideas dd at first, but the postulational basis for projectiv.e
developed earlier, the first real text in projective geometry, Traite des It .m~y se~~ ~om a mathematical point of view than is the baSIS
propri"es des figures by J. V. Poncelet (1788-1867), was published in geometry IS SImp .
230 CHAPTER 7
DEFINITION. A complete quadrangle is a set of 'ou .

for Euclidean geometry. The following axioms for projective geometry . .' r pomts
(vertices) in a plane, no t Iuee collmear• and the lin es JOllllllg
... these
are adapted from those stated by H. S. M. Coxeter.' This set of axioms is
vertices in pairs.
based on the undefined terms of point, Hne, and incidence. The
intuitive meaning of incidence is simply the idea of lying on or
containing. For example, AKiom 1 means that the point does not lie on lhe A complete quadrangle is shown i F·
quadrangle ABeD has three . f n 'gure 7.6. Complete
line and the line does not contain the point. Axiom 2 m~ans that every and Be a d ic d ~ pairS 0 opposite .sides. A'l! and CD AD
line contains at least three distinct points. or has these points on· it. . n an BD. Note that 0 . ·d ...
Opposite sides are two lines one d t . PdPosJte Sl es WIll mtersect.
Axiom 3 means that two distinct points lie on just one lirie. • e ermme by an t .
the other determined by th . . . y wo vertIces and
e remainIng vertIces of a com l
For example vertices A and Cd t . . pete quadrangle.
. ' e ermme one Side 0 tl "
Axioms for Projective Geometry the hne determined by the two . . . ' 5 le opposJte SIde is
remammg vertices, B al~d D.
L There exist a point and a line not incident.
2. Every line is incident with at least three distinct points. E
3. Each two distinct points are-incident with exactly one line.
4. If A, B. C. D are four distinct points such that ls meets CD. ·then
ic meets Bo.
5. If ABC is a plane, there exists at least one point not in the plane
ABC. D~--~~--~~~G
6. Each two distinct planes have at least two common points. ",
7. The three diagonal points of a complete quadrangle cannot be FIGURE 7.6
collinear.
8. If a projectivity leaves invariant each of three distinct points: on a The opposite sides of a com fete d
three points other than vert,.ces ( .p qua rangle meet by twos in
line, it lea yes invariant every point on the line. . pomts E F G of F· 76
t I1fe~ . pomts are the diagonal points of h' Igure .). These .
The vocabulary of Axiom 7 leads to new definitions. used III Axiom 7 TIle dott d I' . . t e complete quadrangle, as
. . e mes In FIgure 76 J'
diagonal triangle, whose vertices a' th d' . ~re t le sJdes of the
M lee lagonai pomts
any of the axioms for pro' t' .
DEFINITION. In projective geometry, a triangle consists df the those for Euclidean geo t d l~c l.ve geometry seem the same as
three noncollinear points called vertices.and the three lines (not segments) me ry. an thIS IS carre t 0 h .
same, however, and that is responsible for i ' c. . ne. t at IS not the
joining these vertices in pairs. charac[eristic structure is Axio 4 TI . .g vl~g. projectIve geometry its
In . l1S aXIOm JS Illustrated in Figure 7.7.
This definition, although not identical to that given in elementary

texts, is the one often employed in higher mathematics. It is necessary A
----- C
in projective geometry because the concept of line segment involves the
idea of betweenness, which does not appear, either explicitly or implicitly,
in the axioms of projective geometry.
I Reprinted by permission of the publisher, rrom PROJECTIVE GEOMETRY by D B

l. t S. M. Coxeter. © Copyright 1964 by Xerox Corporation through Xerox College
FIGURE 77
Publishing, successor in interest to Blaisdell Publishing.
232 CHAPTER i
',f
It is assumed in the drawint that is n CD is not empty. Axiom 4 EXERCISES 7.2

, guarantees as a result that AC n 1iD is not empty. Since iii and CD . a person should expect the postulational basis for
I determine a unique plane, the consequence of the axiom is that any two L ~:;~~~~::~o:~r~e~~o;e simpler than that for Euclidean geometry..
lines of a plane meet in a point. That is, there are no parallel lines • for projective geometry are also ax.loms for
in projective geometry. Each pair of lines in the plane determines a 2. Which of the eight axioms
ordinary Euclidean geometry?
unique point.
. I axioms for projective geometry are also
While ex.ploring the idea in the previous paragraph, it is important For Exercises 3-9, which of the elg 1t
to understand lhe development leading from Euclidean geometry to axioms lor the finite geometry n(lmed?
projective geometry in terms of the type of elements allowed. In J. Geometry of three points and three lines.
Euclidean geometry. every point of the Euclidean plane is an ordinary 4. Geometry of four points.
point. Two parallel lines have 110 point in common. In the extended
5. Geometry of four li~es.
Euclidean plane, parallel lines are said to meet at idedl points. Any two
6. Geometry of F ano.
lines meet at an ordinary point or at an ideal point. Saying that two
lines meet at an ideal poin~ in the extended Euclidean plane is just 1. Geometry of Pappus.
another way of saying that the lines are parallel. The extended 8. Geometry of Desargues.
Euclidean plane consists qf the union of the ordinary and ideal points 9. Geometry of Young. . l'
in lhe plane. The ideal points of the extended Euclidean plane are a . .' melry two distinct lines cannot have mOle t tan
10. Prove that, til proJecUvc geo ,
different concept lhan that of the ideal point of the inversive plane. one point in common.
In the projective planes and spaces of projective geometry, the In Exercises I! and 12, use the sel of points shown in fjgure 7.6.
ideal points lose their special nature, and parallelism is not an invariant. many of its
. 'd d the basic complete qua d rung Ie, how
t 1. If AEGC IS consl ere
It is correct to think intuitively of projective geometry as being derived
diagonal points are labeled?
from the geometry or the extended Euclidean plane through the . 'd . d the basic complete quadrangle, how many of its
elimination of any distinction between real and ideal points. If ideal 12. If AEBF IS consl Ole <
diagonal points are labeled? . r
points are not treated as special elements, the need for consideration of .. r' 1ete quadrangle are the vertiCes 0 a
special cases or exceptions in much of what has been said so far in this' 13. Suppose the four vel Uces a a comp What effect would this have on the
parallelogram in Euclidean geomelrY· '
\ section has been eliminated. For example, the points A, B. C. in
diagonal points?
. . I for more than one complete
Figure 7.5 could just as easily be ideal points as real points. In
4 Could a triangle be a dtagonal lnang e .
Figure 7.6, the opposite sides could intersect in ideal points or in real 1. quadrangle? Make a drawing to support yom answer. . .?
points. Parallelism has lost its significance, and projective geometry is ch line in affine or projective geometry contains how many Ideal pOints.
displayed as a general geometry with ordinary geometry as a very special 15. Ea
case. The axioms of projective geometry apply to points in the projective
plane and in projective space.
It should be mentioned that the geometry of the extended 7.3 DUALITY AND SOME
, Euclidean plane, which considers ideal points as special cases, is called
CONSEQUENCES
t~:n~o~:t:~~W:~a~:t~;!~~~,~:~1~~I~~i':~
ciffine geometry. In affine geomelry, parallelism is preserved; parallel
lines are transformed into parallel lines. The transformations of affine In Euclidea.n geometry:
geometry form 2, proper subgroup of the group of all projective trans- line, but two lines 10 the sam. p. metry on the other hand, this
" formations. The group of affine transformations has the group of (they may be parallel). In projective ge? or;hestatements"twopoints
exception has been eliminated. A comparison
similarities and the group of motions as proper subgroups.
,.j
234 CHAPTER 7
concurrent. The definition of a com I ' ,

determine a line" and "two lines determine a point," with respect to a taking the plane dual of the d fi " pte quadrdateral 's written by
e mtlOn 0 a complete quadrangle.
plane, shows that one may be changed to the other by interchanging
the words poine and line. This is an example of plane duality, one of the
interesting concepts encountered in projective geometry but o'ot in DEFINITION, A complete quadrilateral is ,.
three concurrent and the point r' a set of four lines, 110
Euclidean geometry, For sets of points in the projective plane (the plane of (see Figure 7.9). • s 0 mtersection of these Hnes in pairs
projective geometry), every statement remains true when the words point
and line are interchanged if certain other pairs of words, such as
collinear and concurrent. arc changed accordingly. For example, the;pla ne \ i,
dual of the expression "three collinear points" is "three concurrent \
\
lines."
The existence and validity of the concept of plane duality in
projective geometry can be shown to be a consequence of the axioms
themselves, Axiom 1 is its own plane dual. The dual of Axiom 2 is a
theorem that can be proved.
THEOREM 7,1. Every point is incident with at least three distinct

lines, FIGURE 7.9
Let A"B and C be any line and point (Figure 7,8) not incident, The properties of the co I
the pJane dual of th
'
mp e:e quadrIlateral can be studied as
as in Axiom 1..!rAxiom 3. points C and A determine a second line, e correspondmg p
quadrangle. The six points of ,'nt .
(' f
roper les 0 the complete
. ersectlOn of the r. . .
By Axiom 2, BC has a third' point D, and by Axiom 3, D and A
thre~ sets of opposite vertices. In Fi ure our ~Iven hnes are in
determine a third line, vertices, as are C and D and E an~ 7.9, A. and, ~ ~re opposite
vertices are diagonal fines and tl ~. The hnes Jommg opposite
diagonal fines is the diago;zal lrilal:.a~nlatera.1 ,for~ed by the three
customary to speak or a
t' I . In ploJectJve geometry. it is
both the vertices and th rdIan e as being self-dua~ since it includes
g
e Sl es, and to use d' I'
the complete quadrangle and th I lagona tnangle for both
Th d I f ' e comp ete quadrilateral.
e ua 0 Ax,om 7 implies th t th d "
cannot be-concurrehCThis should be e:s e oUed hnes I~ Figure 7.9
concurrent. then the three d,'a I ' Y to see, because If they were
, ' gona pOlOts of quad I AB
De colJmear, contrary t9 Axiom 7. rang~. CD would
FIGURE 7.B , ,~inally, the dual of Axiom 8 states h ' " ,
Invariant each of th.ree d' t' I' t at If a projectlvJty leaves
Ismctmesonaoit'l'
every line on the pOir1t Th . P n, It eaves Invariant
, e assumptJOn that th' d I'
Axioms 3 and 4 are, in effect, duals of each other, using the results in the completion f tl lS ua lS a true statement
interpretation of Axiom 4 as in Figure 7,7, The plane dual of Axiom 7 axioms of proJ'ective ge 0 1e argument that all the duals of the
ometry are true and th t th .
introduces new terminology that must be explained. This dual states duality may be used freely It ' a e concept of plane
. IS assumed that the plane dual of a
that the three diagonal lines of a complete quadrilateral are never
236 CHAPTER 7
definition or proved tlleorem is a good definition or another theorem and D' E' F' are perspective from line 1. since corresponding sides meet at
whose proof has been established without further work. points X, y, and Z on line I. . .
The concept of space duality is based on an interchange of the Unlike most theorems of geometry, Desargues' theorem IS easl~r
words point and plane. with the word line being self-dual in space. In to prov(~ for two triangles in different planes than for two triangles
the foHowing statements, the second gives the space dual for the first. in the same plane.
Any two distinct planes have at least two common points. Proof:
Space dual: Any two distinct points have at least two common a. Assume that the two given triangles are in differe~lanes,
planes. as in Figure 7.11. Since OB'C'BC lie in a plane, BC and B'C' must
One of the most fundamental, yet nontrivial. applications of the

concept of plane duality is involved in the proof of Desargues' o
theorem and its converse. Recall that tbis theorem was first encountered n'
in a finite geometry in Chapter 1.
THEOREM 7.2. Desargues' -c.heorem. If two triangles are per-

spective from a point, they are perspective from a line.
The theorem is named for the French maihematician Desargues

(1593-1662), wh,o anticipated the development of projective geometry FIGURE 7.11
many years before it was actually developed. His definition (Theorem 7.2)
of triangles perspective from a point and line is made dear by Figure 7.10. meet, and this point must be a point Z on· the line. of intersection
of the twoJ:!anes It and 1I'. Similarly, OA'CAC. detenm~e a plane. an~
lc and A'C meet at a point Y on the line of mtersectl0n of 1t and 1I.
Lines .In and E likewise must meet on the same line of intersectio~
of nand n', so l;ABC and l;A'B'C' are perspective from a line.
b. Assume that the two given triangles are in the same plane.
Let ABC and A'B'C' be any two coplanar triangles. as in Figure ?12H.
B' pers·pective from point O. Let poin~s C, D. E be points of intersectIOn of
pairs of corresponding sides (Figure 7.1.2b). We need to shoW that these
three points are collinear. .'
D'~ AF Let 0' and 0" be any two points colllI~ear WIth 0, but not ~n the
~
plane of the given triangles (Figure 7.12c). Connect 0' and the vertices of
triangle A' B'C' and 0" and the vertices of triangle 1B~. Si~ce the
x y Z · 0'0 d OA determine a plane, AO" and O'A' he III tillS plane
Imes an B" d
FIGURE 7.10 and must meet at a point Arl. -Similarly. O"B and O'B' meet. at. an
o"e and o'er meet at err. Triangles ABC and Air B"C" are 111 different
Two triangles such as ABC and A'Bre are perspective from point O. since planes and are perspective from 0". By DeSa~gues' theorem for, non-
corresponding vertices are collinear with O. Two triangles such as DEF coplanar triangles, they are perspective from a lme common to their l wo
238 CHAPTER 7 PROJECTIVE GEOMETRY 239
possible by using Desargues' theorem as an axiom that eliminates the

need for the space axioms.
A Several other remarks should be made about Figure 7.12. The
set of points and lines in the plane of the two given tdangles is called
the Desargues' configuration. It consists of ten points with three lines on
each point and ten lines with three points on each line. Observe that
this has already been used as the basis for a finite geometry in
(b)
(a) Chapter I. This is an example of a finite projective geometry. In the
Desargues' configuration are not only one pair of perspective triangles
but a total of ten pairs) each perspective from a different point and line.
B
A EXERCISES 7.3
D.L----.-:.~~ Write the plane dual of each statement.
(d)
1. If a and b are distinct lines of a plane, there is at least one point on both
(e). lines.
FIGURE 7.12
2. On a point are at least four lines.
3. A triangle consists of three noncolIinear points and the lipes joining them in
• C' d A" 8"C" are noncoplanar and perspective
, I
pairs.
planes. Also, trIangles A B an . f1 n the same common line. The
from 0', so they are als2.1crspectlve rO.l t on the common line of the Write the space dual of each of the next two statements.
. I
result IS t lat
in d A' Bf meet at a pom
an. -. d the plane of triangle A
';B- IIC"
. 4. TIlree planes not on the same line determine a point.
plane containing t~lven t~ng~sA~ eel on this same line' and the
Similarly. 8C and B'C' and AC an ":. DE 5. A plane is determined by two intersecting lines.
two given triangles are perspective from tlus hne, . 6. Prove the pJane dual of Axiom 8.
7. Prove the converse of Desargues' theorem directly for triangles in different
From the principle of planar duality. the plane dual ofDesargues' planes, without assuming the original theorem or using duality.
theorem has also been established. 8. Draw a figure to represent Desargues' configuration, then indicate the ten
pairs of perspective triangles, naming the center and axis of perspectivity for
each pair.
THEOREM 7.3. If two triangles are perspective from a.line. they.
"
are perspective from a point. 9. Prove that if three triangles have a common center of perspectivity, their
three axes of perspectivity have a common point.
10. Prove that the diagonal triangle of a quadrangle is perspective with each of the
Note that the plane dual of .p'~sarg~€?s1 theorem IS
four triangles whose vertices are three of the verlices of the quadrangle.
. . I but neither 15 It a umque case. '. .
converse. ThiS IS not usua • t d d to a study of perspective
n
Desargues' the~rem could bec~xa: e~tended study of projective.'
polygons other than tnangles. In Sll • theorem or its converse .. ' 7.4 HARMONIC SETS
ou would find Desargues . .
ge-ometry. Y f three Jines or the
1 t 1 lp prove the concurrence 0 Since a complete quadrangle consists of four points and six lines,
frequent y 0 1e . f' for projective geometry
of three points. An alternative set 0 aXIOms an arbitrary line of the plane that does nOl pass through any of the four
points are on the sides passing through the third diagonal point. A
special notation for harmonic sets emphasizes this distinction between the
two pairs of points in the set. For example, H(NI, FG) indicates the
harmonicse! composed of the four points N, I and F, G. F and G are paired
because they are the two diagonal points, whereas N and I are the points
of intersection of the sides of the quadrangle through the third diagonal
point The same harmonic set could be indicated by H(I N. FG).
H(NI,GF), or H(IN.GF).
vertices or any of the three dia onal 0 · ·
distinct points This set f ,g . p mts wIll me~t the SIX sides in six In the notation H(NI,FG). G is the harmonic conjugate of oF"
.
points. Figure' 7 13 h 0 SIX pOInts is called a quadrangular set of with respect to Nand 1. Each point of the harmonic set is the
. sows complete quadran I ABCD harmonic conjugate of the other member of its pair wil~l respect to the
quadrangular set of points E F G ge and the
used to show that each poin~ of • H.~. J. Desargues' theorem can be other pair of points. Since Axiom 7 specifies that the three dia"gonal-
uniquely if the other five k a qua ran gular set can be determined points of a complete quadrangle arc not collinear, a point and its
are llown The co c t ' . harmonic conjugate must be distinct points, so there are at least four points
however, to give added m ' . . n ep 15 Introduced here,
eamng to the specIal case of a d on each line in projective geometry.
set called a harmonic set of POillts, qua ran gular The four points of a harmonic set cannot all be located
independently. The statement of dependence is given in the following
DEFINITION. A harmonic set f . .. theorem.
a set of four collinear in .' 0 pomts, or a harmonic range, IS
diagonal points.
0:
of the sides of a com:l~te t~tco:sIStl~g the f~ur pojn~s of intersection
m rang e wIth a hne passmg through two
THEOREM 7.4. Tne harmonic conjugate of a point A with
respe:ct to two other given collinear points Band C is uniquely
determined.
Figure 7.14 shows three harmoni f'
complete quadrangle ABCD Th r . c sets 0 pOInts determined by In Figure 7.15, let B. A. C be any three given collinear points. l'
.'
.
pomts on t1 r . ' e lour pOInts NFIG are a Ilarmomc
. set of
. .1 I le me... pasSIng through the diagonal points F and G
SIml ar y, the four oi ts LEMG . B
diagonal points E aPndnG h are a harmonic set on the lille through
,w ereas the points EJ FK . .
011 the line through diagonal points E d F I are a harmomc set
four points of the h a r . . an . n each case, two of the
mOfllC set are dIagonal points, and the other two
L ~~-~~===~====-----
----
E
o
!: 0
FIGURE 7.15
,".I FIGURE 7.14
242 CHAPTER 7
A harmonic conjugate D of A with respect to Band C can be found

by constructing complete quadrangle EFGH so that Band C are two of
the diagonal points and A and D lie on lines through the third diagonal
point. This can be done by dlOosing any point F not on A13 and g
connecting F and points A, B, C. An arbitrary line can ·be. drawn
through B intersecting FA at H and Fe at G. Let CH intersect BF h/
/
at E. Then fG intersects Ajj at the required point D, ' /

The fact that D is unique is somewhat more difficul~ to prove. I
/
Assume, as in Figure 7.l5, that a second and distinct quadrangle

E F' G' fl' has been constructed as before, beginning with the choice of
FIGURE 7.16
F' as a point not on A13 and also distinct from F, It is required to
show that E'G intersects A13 at D, '
Triangles FGfl and F'G'H' are perspective from line )113, hence THEOREM 75 T
roiectl'vI'tl'es,
J .. he harmonic property is preserved under
nre perspective from a point J by the converse of Desargues' theorem. P
Triangles FEH and F' E'H' are also perspective from line A13 and are then
perspective from the same point J, so EE'J are collinear. This means , . ~ projectivity may be conside' '
that triangles FEG and F' E'G' are perspective from J and by Desargues' spectlvltles, so it suffices to show that til hIed a fimte sequence of per-
• J e armonk, .'
pl~pelty IS preserved
jJ
theorem are perspective from a line, This line is lB, and fG and 1 a sing e perspectivity. To prove that thO ,
IT must meet at a point D on is; thus the harmonic conjugate of A separate statements: IS IS so consIsts of proving two
with respect to Band C is a unique point. . a. The set of lines joinin an .

The proof of Theorem 7.4 includes a method of constructing pomts ofa harmonic set of po' t ~ : nonc~nJnear point to the four
the harmonic conjugate of a point with respect to two bther given . b. The set ofpoints o/~;e~:eactiannonlc set of.fines.
collinear points. This construction. as anticipated in Section 5~6. uses only ~et WJth any line not through the oi on of the four hnes of a harmonic
center of the pencil) is a harmo . p nt of .concurrency of the lines (the .,
a straightedge. . me set 0 f pomts
The dual of the definition of a harmonic set of points is the Smce the two statements are I .
sufficient. The first statem t ' pane duals. proving one will be
definition of a harmonic set of lines. en IS proved here, In FIgure ' 7,17, let
DEFINITION. A harmonic set of lines. or a harmonic pencil. is a o

set of four concurrent lines such that two of them are diagonal lines of a
A~
complete quadrilateral and the other two pass through the 'two vertices
lying on the third diagonal line,
C B 0
In Figure 7,16, let a, b, c, d be the sides of the quadrilateral, . FIGURE 7.17
with e,j: 9 the diagonal lines, One set of harmonic lines is H(eg,lli),
The main reason for the study of harmonic sets in pr,oiective
geometry is that the property of being a harmonic set is an in·vn,i.nH H(~B, CD) be any harmonic set of oi t '
TillS harmonic set impJies the . p n s with 0 any non collinear point
under the group of projective transfonnations. eXIstence of a comp Iete quadrangle OEFG'
PROJEC)'IVE GEOMETRY 245
244 CHAPTER 7
, {IYll H(AB CD} is a iWfllwnic
with A and B two diagonal points and C and D points on lines 2. Locate D by construction with a straightedge so < '
through the third diagonal point. But GF. lEo II and GE are the four set. , \I possible pairings obtained from
sides of a complete quadrilateral. lo
and I51l are diagonal lines, J. U
se Theorems 7.6 and 7.7 to wnte a
whereas 15C and I5D are lines through the other two vertices, F and D. H(CX. Dn.
,
. - D .. ', II e
uclidean sense, that B is the 11lidpomt of .<Ie. escll)e \
which lie on the third diagonal line. This implies the existence of the 4, Suppose. In the E , ' . te of 8 with respect to A and e. .I
harmonic set of lines H(I51l. DA.X.OD).
The symbol for a perspectivity, ABCP *
A'B'CD', indicates that
S is the center of perspectivity and AA'. BB'~ ... are corresponding points.
location of the harmonic conJuga
.'
For E-xelc.ses -:.0.
(AD. KG) 6, (;tcrM)
' 8 : LF BD)
7. (AB.NE)
. F
5 t' which of these <Ire harmonic sets 1tl Igl1l e"
. ( '.
. 7 14" Why"
'
9. (BC.JG)
.
The symbol ABeD A A'B'C'D' indicates a projectivity in which AA·.
BB· •... are corresponding points. In Figure 7.17, ABCD
GEHD ~ BAeD. so ABeD A BACD. This implies that:
*
GEHD and
5.
.•
II.
.
TI . 75 direclly without usmg dunlHY·
10 Prov" statement b under leOletll '
k
uni uely if the other five are nowl~.
d' I"
Prove thnl the sixth paml of a qua I angu ,\I se _
" '
,
t (If points Ci.\Il be deternll ned
pr:e that a line in projective geometry hns more thall fOllf dlstl1lcl \111I11{S,
THEOREM 7.6. If H(AB. CD). then Ji(BA. CD). t2.
that:
Also i1l Figure 7.17, ABeD 4 FOHC and FOHC * BADC. so
7.5 PROJ ECTIVITI ES
fi '( sequence of
" ivit has been presented as a 111 ~ .'
THEOREM 7.7. If Ji(AB. CD). then H(BA. DC). ~. ~)wJect , ,Y t' 'ty is a type of transformation, S111ce It IS a one-
perspectlVlt1eS, A pi oJec IVl,
-to-one onto mapping of pomt~, " , 'OJ'ectivity and the product
Beginning with three distinct points on a line, the harmonic , f a proJectlvlty IS a pi •
conjugate of a-ne of these points with respect to the other two can be The l1~v~~se .0 , ' t ' 'ty,thus the following theorem is true
of two proje.cuvlues IS a pi oJec IVI. • .
uniquely determined. For the set of four points. each set of three (Exercise 9. Exercise Set 7.5).
determines other harmonic conjugates. All the harmonic conjugates on a
line determined by such a sequence of steps are said to be harmonically t of all projectivities of the plane
related to the three original points, II THEOREM 7.8. The se
constitutes a group of transformations.
DEFINITION. The set of all points harmonically related to three
distinct collinear points is a harmonic nee of points, , ' ' the stud of properties of sets of points
Projecl1ve geomellY IS f y!)I'oJ'ective transformations, The
, . t u del' the group 0 ,
that are mvartan n . ' eOl11etry must wait until the
The concept of hai'monic net is fundamental in the development . h th tudy of projective g .
algebralcapproac to. e s . the ro'ective plane in the next sectIOn,
of a system of coordinates for ,Points in projective geometry (see introduction or coord,m~tes ro~ p. Jd rojectivities from a different
Section 7.6). Before this, however, It IS possible to stu y P" d n you is: What is the
, t' tl 'tt may have OCClIlle l . " ,
viewpo1l1t. A ques lOn 1< ' I determine a prOJectlvlty r
. . . r
mllumllm 111 orma 10n n
t' eeded to untque Y
, d h '. '1 nages how many pturs 0
. r
EXERCISES 7.4 , " .' ' lie pomts an t ell I < • • ,
Smce proJectlVIUeS InVO \ , I t ii pairs can be deternuned?
, t be gIVen so t 1a a ~ I
Given lhree points A. B. C in this order on a line: points ,;lI1d Images mus. .' I '\t is caned the jiuu/amelli£l
The answers to these questIOnS appeal m w 1<
1. Locate D by construction with a straightedge so that H{AC.BD) is a harmonic
seL theorem c?f projective geometry.
246 CHAPTER 7
THEOREM 7.9. Fundamental theorem. A projectivity between the arbitrary distinct line I through A'. Then points BTl and C" are
sets of points on two lines in a plane is determined by three determined on I so that ABC!..7i A'B"C". The mtersection
. of bc" and
<-+
collinear points and their images. B'Btl determines a point S' such that A'B"G" ~ A'B'C'.
A'B'C'.
" so ABC 71
The proof is indirect. Assume, as illustrated in Figme 7.18, tilat
ABCX 7\ A'B'C'X', and ABCX 7\ A'B'C'X", for X' and X" distinct c ..
points.
B'
x ..
c·
8'
x.
x ABC
A 8 x •
C FIGURE 7.19
FIGURE 7.1 B
Given any point X on A'B X" and X'
perspectivities already set up If tI' . . C':'
be found by using the
Then A'B'C' X' 7\ ABCX 7\ A'B'C'X".and A'B'C'X"7\ A'B'CX". fine, then one extra perspecti~H :e SIX ~tven pomt~ all He on the same
But in this last projectivity, points A'. B'. and C' are fixed. By Set 7.4. y IS reqUJred. See Exercise 4 of Exercise
Axiom 8 of Section 7.2, all the other points on the line must also be The corresponding points in a '"
fixed. The assumption that X' and X are distinct contradicts the
If
perspective from a point n -r a th projectlvJty are not, in general
, v re e correspo d' r ' .
axiom, so X' and X" must be identical. perspective from a line But pro' '" n mg mes In a projectivity
. '
While the fundamental theorem of projective geometry is very
. jectlvltles do have' .
somewhat analogous to centers d a pomt or a hne that is
r. I . an axes of perspecf't Th
important from a theoretical point of view, it does not yeOt provide a are uselll In Section 7.8. tVl y. ese concepts
constructive method for actually determining additional pairs of cor~
responding points in a projectivity. One such method ·is based on
THEOREM7.11.A roo ..
establishing a minil11pm sequence of perspectivities for the -projectivity. distinct lines determines : tf~C~IV~~Y between two sets of points on two
This sequence, while not unique, yields unique results because of the 1f
homology. and contains the inte'rse t' me called the axis. or axis of
c lOllS of the c ..
fundamental theorem. .0 f corresponding points. ross JOinS of all pairs
THEOREM 7.10. Three distinct points A. B, C on one line may

f
be projected into any three distinct points A', B , C' on a second line' DEFINITION. The cross" f '.
A, A' and B. B', are An- and M. Joms 0 two pairs of points, such as
by means of a sequence of at most two perspectivities.
To prove Theorem 7.1 I, it must be shown that the axis is

In Figure 7.19, let A, B, C. A', B'. C' represent the six
~
of perspectivity on AA' and unique and that all of the intersections of the
points. Choose an arbitrary center cross joins are
~.
p.
"
111
248 CHAPTER 7
given pairs of points A, A', B, B', C, C, The axis of homology can be

~oiUnear. This can be accomplished by shO\ying that the axis is
determined by the intersection of the crosS joins liC.
~ <--+ ~
A~
and
IIldependent of the choices of points_ +-I>
AB',A'B. Let X represent any fourth point on AB. Now XC' intersects
, r Let D. Figure 7.20, be the point of intersection of the two the axis of homology at a point X". and ext meets IT at the
glve~ll1es. Let E be the image of D when D is considered as a point .-.
required point X' on A'C'. Note that this construction uses only-a
on DA', and--Iet E' be the image of D considered as a point on l5A:
slraightedge.
The duals of the theorems already p.l'esenled in this section are
A' interesting, in their own right. The proofs given depend on the concept of .
I
I duality. even though the theorems could be proved directly.
E'
I·
I THEOREM 7,12. Dual of Ihejwu/allIenw/ theorem. A projectivily
I
I between the sets of lines on two points in a plane is determined by·
0 I
E, three concurrent lines and their images.
I A
FIGURE 7 20 THEOREM 7.13. Dual of Theorem 7.1 1. A projectivity between
two sets of lines on two distinct points determines a third point, called
lif andF~';;'he It:,~ pairs of points A. A' and D. E'. the cross joins are the center, or centel' of hO/1l010gy, which lies on the joins of the cross
.• W lie 1 meet at point £' on the axis of homology For
intersections of corresponding lines.
tlIleD
two 'pairsof points A'. AdD an .E. t Ile cross joins are A'E
....; and
tw • ~leetmg at E. Thus. the axis of homology always passes through the
? u~ages of th~ common point of the two lines. when the common Theorem 7,13 is illustrated in Figure 7.22. Lines a, b, (' and
~~ll1tIS, first consIdered as an element of one line and then of 'mother
1~aXIs of homology is thus seen to be independent of tIle'· f'
POints selected. paIrS 0
c
Axis A aX
FIGURE 7.22
C'
, FIGURE 7.21
X is the interse~on of a and [,' and Y
a'. b'. c' are given. Point
is the intersection or
d and b, so IT passeFthrough S. the center of
homology. Likewise, c and [,' intersect at l'V and c' and b meet at Z. so
' f The axis ' of
. homology may be used to construct . . additional
S lies on Wz. ' ~. .
pairs 0 corresponding
ex'un Ie I F . . t . . ' .
pom S In a proJectlvlty. as shown in the following The center of homology can be u~ed to construct additional
( p . n 19u1e 7.21. assume that a projectivity is determined by the
250 CHAPTER 7
. ..
pairs ofcOllespon mg
.
d is given. Ll11es
d' lines in a projectivity. In Figure 7.22, suppose that
.
' meet at P Lme a mee
d an d a ·
ts d' at point Q on
of a' b! c' is
~,~'
?S. The line connecting Q and the point of concurrency "
~
the. desired line d',
EXERCISES 7.5
K
,'.,, .
FIGURE 7.25 ~l' ..
For Exercises 1-3, use Figure 7.23.
7.6 HOMOGENEOUS COORDINATES
~8' T!lis section presents a brief analytic treatment of some of the

~ fundamental ideas of projective geometry, Since the projective plane
includes ideal points that are no longer considered as speciaJ elements, ..,;
the Cartesian coordinate system of ordinary analytic geometry will no
longer suffice.
A 8 X c Prior to a more formal presentation of the coordinate system of
FIGuRE 723 projective geometry. called homogeneous coordinates. it is important to see
intuitively how the two systems of coordinates are related. Figure 7.26a
d' II proof of Theorem 7.10 Lo establish two perspe~tivities
I. Use the met h 0 In Ie C d A' 8' C
o that a projectivity is established between A. B. an .' .
s . ., . h A A' B B' and F 00
2. Construct the axis of homology for the proJecUvlty Wit •••.
. I~
C C' as corresponding points. F'
3. F:nd the image of point, X in the projeclivity of Exercise 2. _ _.
4.
. . . ABC
Prove that three distinct points • • on
a line
,
can be projected
f
mto
I in of
any three distinct points A', B', C' 011 the same hne by means 0 a cIa
three perspectivities. . , 0(0.1) e(l,l)
:;"'::"':'.<.j...-+='-'-'-~oo
. , Set up a chain of three perspeclivities so that A and A.
C opy F ·Igure 724
5.
Band B'. and C and C' correspond, -;-;;:-:+-I-::;:-;c---
A(O.O) 8(1.0)
E A,L-!:-':::::::::::::~~
8'
E'
--~---:': ' : - - : - - (a) (b)

C' A B A' C 8' X
,
6. For the projectivity in Exercise 5, find the image of X. .. '.1 shows the framework of the ordinary Cartesian coordinate system, with
7. Copy Figure 7.25. Find the. poi~l 0 f h 0010 Iogy for the projectlvlty WI 1 points E and F indicated as "points at infinity" in a coordinate system
aa'. bb', and ce' as correspondmg lines. for affine geometry. Figure 7,26a may be projected into Figure 7.26b,
8. Find the image of line X in the projectivity of Exercise 7. with the corresponding elements indicated by primes. The ideal points
E' and F' are no longer special points, and IT is the ideal line.
9. Complete the details of the proof of Theorem 7.8. . d I'
The points in the first quadrant of Figure 7.26a are projected into the
10. Prove the plane dual of the fundamental theorem. without uSlOg ua Hy.
-:.,

252 CHAPTER 7
points in the interior of triangle A'ET. Already, it should be apparent

lhat the coordinate system for projective geometry is not one in which
distance is preserved in the usual sense.
The homogeneous coordinates must be defined in such a way that [or the point (x l ' YI) to lie on the line
they will distinguish among an infinite number of ideal points and yet a b
_ x + - y + 1 ~ 0,
not require any restrictions in the way these points are handled. The c c
substitution For a variable point x on the line X. the ~quation of ~he line. is
Xl X,
, b v
given YAtXl
+ X' 2-2x + X x = O. For example, the equatlon of the
3~3 . [I·'
X=-
X,
Y = -
x,
[or -'=3:f: 0 line [2, _ 5, 7J is 2x 1 - 5X2 + 7:<) = O. Dually, the equatIOn 0 t ~e pOl~lt
(3, 1, 2) j~ 3X 1 + X 2 + 2X 3 = O. points ~o ~10t have ~q~1:1tJOnS Il1
relates the ordinary Euclidean point with coordinates (x,y) to the Euclidean analytic geometry, but the duahty 10 the prOjectIve plane
corresponding point in the projective plane with homogeneous co- should have prepared yOll for their ;appearance in the study of homo-
ordinates (xl,x"x,). For example, (2,3) becomes (2,3,1); (3,4,5) is the
geneous coordinates. ..
same point as (3/5, 4/5). For ideal points, the restriction X3 :f: 0 must be It is noW possible to relate the forma1 defimtlOlls of hOl11?-
removed, and the stage is set for a more formal approach. geneous coordinates to Figure 7.26, recalling the in~uitive development III
terms of ratios. Study Figure 7.27 carefully, notmg the h?1110 geneous
coordinates of each poin~ as related to the ordinary coord mates .of the
DEFINITION, In analytic projective geometry, a point is an
corresponding points.
ordered triple (Xl' X 2 ' x 3 ), not all zero.
F'(O.1.0)
The notation (ax 1 ,ax 2 ,ax 3 ) represents the same point as

(x l .X 2 ,X)) for any nonzero a. For example, (3,1,2) and (6,2,4) name the
same point in homogeneous coordinates. A line in homogeneous co-
ordinates is defined dually as an ordered triple [X I ,X 2 ,X,J. with the
understanding that [aX l ' aX 2' aX 3] names the same line for any non- 0'(0,"L'~)=L=::=====~ "',~O)
A'
zero G. For exampJe,
(O,O,1) 8'(1,0.1)
FIGURE 7.27
[3, 2, 7] and [q, iJ
'gure 7 27 it is not difficult to determine the equations
are two names for the same line. F rom Fl . , . r .
A point x lies on a line X if and only if of the various lines and to relate them to the correspondtng mes m
Euclidean geometry. The relationships are as follows.
kE
For example, the point (2, 3,0) lies on the line [3,·- 2, OJ, since
(2 ,3) + (3 : - 2) + (0· 0) ~ 0,
The condition in the previous paragraph is a generalization of the
-
AT
x2 = 0
Xl = 0
x, = 0
condition for points lying on lines in Euclidean analytic geometry. If X3 Xl - X3 = 0
and X) are set equal to I, the condition reduces to X 1 Xl + X 2 X 2
x2 - X3 = 0
+ (1 'I) ~ 0, which is equivalent to the condition
254 CHAPTER 7
can be co~tinued for points with negative coordinates
These equations can be derived formally from the condition (-1,0, I) IS the harmonic con'u ate f ( . ' For example,
X 1 Xl + X"2 X + X 3 X3 = 0 for a point to lie on a line, by subslituting and (I 0 0) Not 11th . J g 0 1,0, I) With respect to (0 0 1)
2
• •. a e pomts on the line but II . • •
the coordinates of the two points on a line
...-and then so1ving rhe set of net determined by the three begi' : a those In the harmonic
nnmg pomts, can be constructed.
simultaneous equations. F Of example, for C' D',
X,+X,=O,
EXERCISES 7.6
X,+X,+X,=O.
1. HO~ can one teU an ideal point by looking at its coordinates?
Let X 3 = 1; therefore, 2. Wnte homogeneous coordinates for these points: .'
x, = -I, a. (3, 8) b. (2,;1) c. (I, - 4)

3. Write Cartesian coordinates for these points:
1
X, = 0,
a. (2,5, I) b. (- t, 3, 2) c. (-2,5, -3)
and the equation is -X2 + X3 = 0 or x 2 - X3 = O. For the lines given, Write
4. (5, 2, 3),three other names, using 1IOmogeneous cOordinates, for the point
the equations can also be derived intuitively simply by examining the
coordinates to note a pattern. 5. Does the point (t, I, 1) lie on the line [I, I, 1J1 '
It should be understood that the triangle in Figure 7.27 :includes
6. FlO,d the coordi~ates for the ideal point on the line [1, I, 1].
not the entire projective plane but only that part corresponding to the
7. Wnte the equation of the line [3, t, -2].
first quadrant. The points A'E' F' are vertices of what is caned the
fundamental triangle, whereas C. with coordinates (I, I, 1), is called the 8. Write the equation of the point (3, I, - 2).
9, Develop the equation for ifCi, Figure 7.27.
ut/it poiHt.
The procedure by which coordinates are found for additional 10. Find the point of intersection of the lines [1 I t] d [2
points in the projective plane is based upon the concept .of a harmonic 11. Fi d til ' " an • l, 2].
,n e equation of the line through (1, 1, I) and (2, 1, 2),
net. This procedure is illustrated for the line .\2 = 0 in Figure 7.28.
12. Give the coordinates of the harmonic .
(0, I, I) and (0, 3, I). conjugate of (0, I, 0) with respect to
~
For Exerc'Ises 13- 16,construct with a strai hIed e ' .
Xl = 0, given (0, 0, 1), (I, 0, I). and (I, 0, 0), g g the reqUired points on the line
13. (2,0, t)
.,!'-..
t4. (3,0, t)
t5. (- t, 0, I) 16. (t,o, 1)
(0.0.1) (1.0.1) (2.0.1) x,~o (1.0.0)
,:/
FIGURE 7.28
7.7 EQUATIONS FOR PROJECTIVE
The point (2,O, I) is defined as the harmonic conjugate of (0,0, 1) TRANSFORMATIONS
with respect to (1,0, I) and (1,0,0). The point can be constructed with
The set of simultaneous equations need d ' ,
a straightedge as explained in Section 7.4. The series of constructions transformation analytically is more diffic I e to express a projective
could be carried on to show that (3,O, 1) is the harmonic conjugate of of equations for transformatio , u t to develop than were the sets
, ns given so far-those ~ E I'd
(1,0, I) with respect to (2, 0,1) and (I, 0, 0). Also, H, 0, I) is the. harmonic motIOns, for example . Keep i n mID ' d the fact that. or u~ 1 ean
conjugate of (1, 0, 0) with respect to (0,0, I) and (1,0, I). The procedure in projective
256 CHAPTER 7
PROJfiCTIVE GEOMETRY 257
geom~try, straight lines are. transformed into straight lines so the t r

~quatlOns r~rproject.ive tninsformations is a set of three Simulta~:o~s The new point may be indicated by (dle + el + fi11), which shows that it
linear equations relating the homogeneous coordinates of a point to the has corresponding coordinates referred to points tI. e, J rather than the
homogeneous coordinates of its image. original vertices of the fundamental triangle, since (Ie, I, 111) could have
Three points (x x x) ( ) ( been indicated by the coefficients in the expression
!
a d I l h . 1. 2' - J . YI' )'2' Y3' Lt. Z2> Z3) are collinear if
n on Y I 1 ere eXist three numbers a.. b, c, not all zero such tl,at .\
aX+by.+ -OTI' . ' (["'1] + [1'1] + [m·I]).
i . '. CZ, - . lIS means that any pOlllt on a linc is a linear
Clol11bl~atlOn of any other two distinct points On the line. For exam!)le The set of linear equations given are the simultaneous equations
1e pOint (8 l3 23)'
t
, , . IS on t h e I'me wllh
. .
pOints '
(1 2 4) and (2 3 5) .
• , • • • since
,
sought lo represent the projective transformation. They provide a one-to~
(2)(1) + (3){2) + (- 1)(8) ~ 0, one correspondence of points in the projective plane. They preserve
collinearity of points. Although the intuitive development has not been
(2)(2) + (3)(3) + (- 1)(13) ~ 0, complete, the theorem is stated, with additional disclission following.
(2)(4) + (3)(5) + (- 1)(23) ~ O.
In this example, a ~ 2. b ~ 3. and c ~ - 1. THEOREM 7.14. A set of equations representing a projective

No three of the fOllr basic points (I 0 0) (0 1 0) (0 0 transformation is of the form
I· (t. 1, 1) are collinear T1 r • • , • , , , , ,1),
. . .. lese 100~r pOInts may be transformed by a pro- XI' = (/1 Xl + el'~\l + fl X 3 ,
JeC~lve transformatton ~nto four noncollinear points tI, e, f, and
"
(d f- e -I- f) whose coordmates are Xl' = 11 2 X 1 + e 2 ''(1 + j~X3'

(£I,. £I,. d,). X3' = d3 x1 + 1!3 X 2 +/3 x 3'
(e I ' e1 , e3 ).
It is tmderstood that the determinant of the coefficients is not
and zero. Also, because homogeneous coordinates are not unique but Iilay be
mUltiplied by a const.ant and still name the same number. the introduction
(d,' + e, + f,. d, + e, + f,. d, + e, + f,). of a parameter as a coefficient of the primed tenns is considered
pnlis transformation can be accOl11plished by the cOlTespondence below: carefully in a more detailed course.
The proof of one and the assumption of two additi011al theorems
Xl' = {[tXt + etx l + f 1 x 3 • will help establish that the equations of Theorem 7.14 do provide for a
proje.ctive transformation because they keep projective properties
Xl' = d2 x, + e,' . 2 +12 x 3' invariant.
X3' = d3 x 1 + e3 -x 1 + /3-"3'
. This set o~ equations transforms any other point (le, /, 111) into a THEOREM 7.15. A transfol1l1ation in the plane is a projective
POlI1t whose coordmates are given by the equations: transformation if it transfonl1s one line pt·ojeclively.
Xl' = title + ell + 11 tn,

1n Figure 7.29, let 1 and l' be the given lines wilh the projective
Xl' = ell/( + ell + /2111, transformation and J11 and m' any other pair of corresponding lines.
x,' ~ d, Ie + e, 1 + f, m. Let X and X' represent corresponding points on I.and 1', with A and B
any other two distinct points not on any of the four lines. Line
258 CHAPTER 7
r THEOR~M 7.17: A transformation that 'preserves the cross ratio.

X' of every four collinear POInts is a projective transformation.
8
_ Theorem 7.16 and Theorem 7.17 may be used for the line
y'
m Xl = O. The s~t of Simultaneous equations given in Theorem 7.14
transformsapomt,(O,x,.x,)onx,. = 0. into (e 1 X 2 + Jr1 x3· e 2 X 2 +J2 x3 •
)
eJ x, + f J X J . The four pomts (0, 0, I), (0, I, 1), (0, 1, 0), and (0, x , 'x )
X' have a cross ratIO of Xl/Xl' since . 2 3
FIGURE 7.29
X AY is transformed into X' BY', and since Y and Y' are projectively
(0, x,. x,) = (0, x, . I) Xl.
related, the following chain shows the projectivity connecting I and 1'.
It. can ~e verified that these four points are projected in(e four points
X~YAY'!LX'
7( 7( WIth thiS same cross ratio. Since the four sets of coordinates chosen
ma~ r~present any four points on the fine, the line is transformed
proJectively.
The use of Theorem 7.15 means that the homogeneous co~
ordinates may be considered for one dihlension only. Projective co· . . It has now been established that every projectivity in the
projective plane has equations of the form .
ordinates for three points on a line can be chosen so that they are
(Xl' x ). (Yl' )12)' and (Xl + Yl' X 2 + Y2)' Then the harmonic conjugate
2 •x'
1 -- QI X I + a1 x 2 + a3 x 3 •
of the third point with respect to the first two has the coordinates
(x, - y,. x - Y2)' For example, the harmonic conjugate of (4, 3) with
2
x 2' = bJ x 1 + b2 x 1 + b3 x 3 •
respect to (1,1) and (3, 2) is (-2, -1). x'-
3 - CtX, + C2 X 2 + Cl X 3 •
A generalization of the concept of harmonic set is the-idea of the
cross ratio of four points. with the stipulation that the determinant of the coefficients,
I ~: ~~ ~:I
DEFINITION. The cross rati,? r of four collinear points is the
number I' when the coordinates of the four points are writren in the C1 c2 c3
form (x,. x 2). (y,. Y2)' (x, + Y,.x 2 + y,).(rx, + y,.rx 2 + Y2)'
is. not zero. In line.ar a:~ebra, the concept of a matrix. a rectang ular
at ray of nu.mbers, IS utlhz~d to analyze sets of Unear e(luations. The
For example, the crosS ratio for A(2. 3). B(5. 6). C(7.9)' D(9. 12).
set of equatJOns for the projective plane has the matrix or"toefficients
with the pairing as in the definition, is 2, since 9 = (2' 2) + 5 and
12 = (2' 3) + 6. '
Two additional theorems, stated without proof, will make a
conclusion possible. (~: ~: ~:)
c1 Cl c3
This convenient notation makes it easy to symb o J'lze vanous. sub-

THEOREM 7.16. The cross ratio of four points is an invariant
under projection.
groups of the set of projective transformations pr.eviousJy introduced I
:11 PROJECTIVE GEOMETRY 261
:1,
260 CHAPTER 7
11
IiI 6. Verify that the cross ratio of the images is the same us the origin<11 cross
,) by writing the" matrix
followin . of tl1e ceeffi Clents.
. This is illustrated by the ratio in Exercise 5.
g matllces, with the determinant nonzero in all cases. by each
For Exercises 7-lO. what lype of lransformation, if any, is represented
"
"2 matrix of coefficients?
C
b,
0
b2 b,
0
Q')
C,
affine transformations
7. (2
-3' 2
3
0 0 :J
8.
GD
2
5
8
2
G0
10. 0
( Q,
\~ D
a2 I
±( ~(2) (±a1) a,)
b, similarity transformations 5
0
0 1 (at' + a,2 '" 0) 0
a,
( a, SPECIAL PROJECTIVITIES
±(~a,) (±a1) a,)
b, motions 7.8
0 1 (a l
l
:+ a/ = 1) The general theory of projectivities developed in Section 7.5 can
be both used and extended by considering special projectivities in one
greatly ~ltl:n~S~ ,of ma~rkes in ,connection with transformations can be
e H1 an appropriate course F . .. and two dimensions.
of matrix multiplicatioll ca b d - Ol example, the defillltlOl1 A one~dimensional projectivity renames the points on the saille
. ( neuse to find tl d
formations and the cone t f' le pro uet of trans-
find the in~erse of a transfoe:ma~io~~verse of a matrix can be used to line.
DEFINITION. A one_dimensional projeclivily is called dliptic,
parabolic, or Jiyperbolic if the number of invariant points is zero, one, or
EXERCISES 7.7 . two, respectively. If there are three invariant points, the projeclivity is
the identity transformation.
L Find numbers {/. IJ • cos( how that the Ime
. through (1 I, 1) and (3, 2, I)
a Iso passes through (7, 4, 1). ' The fundamentallheorem of projective geometry can be used to
For Exercises 2 and 3, find the harmonic conjugate of the (hird point with
slate that:
respect to the first two.
2. (3, I), (-2,5), (1,6) 1. A hyperbolic projectivity is determined when both invariant points
3. 1-1, -2), (3,2), (2,0) and one other set of corresponding points are given.
2. A parabolic projectivity is determined when its invarianl point and
Use the projective transformation two other sets of corresponding points are given.
XI' = Xl +- ·"1 + .\3'
This second statement can be sharpened significantly, however,
X 2' = XI - X2 + "3_
by proving the following theorem:
for Exercises 4-6.

THEOREM 7.18. A parabolic projectivity is determined when its
4. Find the images of the poinls (0, 0•I)•(100)
SF' • >,> 110•I)• and(30 5)
invariant point and one other set of corresponding points are given.
. 'md the cross ratio of the four original points in Exercise 4.• , .
s.~~c/{
/A
~ 8 8'
FIGURE 7.30
C" A
~
S'
8
A'
X Y I
!
FIGURE 7,31
In Figure 7.30, let a parabolic projectivity be given, with A the . I~ Fig~re 7.31,. le.t ABX be any three given distinct collinear I
invariant point and Band B' any other pair of corresponding points. i
~~~nt;; WIth A B' Y theIr tmages. in a projectivity that interchanges A ,I
An arbitrary point S can be chosen so that ABB' ~ ACe, where A, C, e . These SIX pomts de~e~~me a uniq,ue projectivity that can be
perspectivity S' can be chosen so that ACe *

are on a line I through A not containing S. A second center of
AB'·C". The image of any
other point D on A'B can be found by using the two perspectivities
represented by these perspectlVltIes: ABXY ~ EDXG ~ FCY D
so'ABXY BAYX TI'
.
pomts X Y and
.. . A G jf BAYX,
7 i . ' . us proJectIVlty interchanges all arbitrary pair
IS an mvoJution. s 0
f
..1
established. From what has been said so far, it has not beep proved . Some special one-dimensional transformations have b
that the projectivity established is indeed parabolic rather than h),perbolic. ;entlOned, "fnd one-dime.Hsional transformations are still invol~ed in :~~
That is, for some point D on line AB. the two perspectivities might IS~USS~o~ 0 the projective geometry of the plane. A two~dimensio 1
result in D as the image, giving a second invariant point. For this to proJe~tIvlty .transforms everyone-dimensional set projectively Tha't isnaa
happen, D. S, and Sf must be collinear, a situation that can be avoided
simply by inserting the restriction that Sand S' lie on the -same line
~~:~~I~:~S~~~~I rrojectiv: transfor~lation may involve every ~oint of the
, . me 10 t e plane IS transformed into another line 0
through A. that there IS a projectivity established by the points on the two lines. s
Some special projectivities are periodic in the sense that they
will result in the identity transformation after a finite number of
repeated applications. four line:~7~Rc~~p~~~0. A twdo:ldimensional projectivity that leaves the
f'. e qua n ateral mvanant IS the identity trans-
lormatlOn.
DEFINITION. A pl'ojectivity of period n is one that must be
repeated n times before first resulting in the identity transformation. As illustrated in Figure 7.32 for quadrilateral a. bed
. , . th e SIX
.
DEFINITION. An involution is a projectivity ofperiod:two.
It can be seen intuitively that a one-dimensional involution simply

interchanges pairs of points. TIle following theorem establishes a minimum a
condition for a projectivily to be an involution.
d
THEOREM 7.19. A one-dimensional projectivity that exchanges
one pair of distinct points is an involution. FIGURE 7.32
264 CHAPTER 7
EXERCISES 7.8
vertices lie three 011 each line, and these vertices are invariant
There is a projectivity between two corresponding sides, and every point. 1. Explain why a reflection is an eXi:lmple of an involution.
on these sides must be invariant since three pairs are. Any other linc· 2. Give an example of a rotation that is of period
meets the sides of this quadrilateral in invariant points and every point
a. Two. b. Three. c. Four.
Show how the image of any other point on the line ca~ be cons.tructed. i,f
on it must be invariant, so the transformation is the identity_
Special types of two-dimensional projectivities include one 3. the invariant point and one other set of corresponding pomts are glVen fm <1
relating two given perspective triangles, called a perspective colJil1eatioll. one~dimensionai parabolic projectivily.
That is, the two triangles are images under a projectivity and are also h . t on the tine c.lI1 be constructed
perspective. The point and line of perspectivity are the center and axis 4~ ~tt~: i~~::i~:~ pi~l:I:~e a~~ ~:~ :t\l:;·.S~~:r
corresponding points are given _~or
of the transformation. The special case for which the center lies on tho a one-dimensional hyperbolic projccllvlty.
. needed to determine a
axis is called an elatiolt, while aU other transformations of this type are 5. State and prove fI theorem about the in rormation
hWl101ogies, A special case is a homology in which the harmonic unique one-dimensional involution. ..
conjugate of the center with respect to pairs of corresponding points 6. State the plane dual of TIleorem 7.20 and pl'ove it by writing the dun I of {he
is on the axis, This transformation is called a harmonic homology and proof of Theorem 7.20.
is illustrated by the example in Figure 7.33. Triangles ABC and A'B'C' 7. Sketch an example of an elation.
are perspective from point S and from line t. Furthermore, the Sketch a homology that is not a harmonic homology,
harmonic conjugate of S with respect to pairs of corresponding vertices 8. . h I Y Construct the
is on line l. For example, S' is the harmonic conjugate of S with respect 9. 1n F ·Igure 734
. , let S and I determine a harmomc omo og .
images of points A. B. C.
to A and A'.
s· 8
A
• • s c•
•
FIGURE 7.34
A
FIGURE 7.33
7.9 CONICS
You should recall from elementary geometry that the various
'b d as sections of cones of
If the center and axis are given, the pairs of corresponding conics of Euclidean geometry can a.II b e descrt e . . r' W
points in a harmonic homology can be determined, since they are nappes as illustrated in Figure 7.35. From a dJfferel~t pomt 0 vll~e ,
two , . b ' t d tnto an e Ipse
harmonic conjugates with respect to the center and a collinear point on . 7 35 shows that a CIrcle can e proJec e
the axis. This type of transformation is of period two, since repeating it f;f~~:e 7·35a) a parabola (Figure 7.35b), ora hyperbola (Figure 7.35C\
will result in the identity. In fact. it can be established that every two- The pro~erty' o-r being a conic is an invariant under the group 0
dimensional transformation of period two is a harmonic homology.
266 CHAPTER 7
projective transformations, but the property of being an ellipse, a ~nd. ~ ,line conic, wi.th the connections between them and the pro-
parabola, Of a hyperbola is not. Jectlvltles
th thatf determme
.. them emphasized. In Figure 736
.a, A and B are
e centers 0 two projectively related pencils with C D E F . ts a f
. t . I '" pam
m'thersectJOn of" corresponding lines. In Figure 736b
' a
• an d b are I'mes
Wl two projectively related ranges of points. A,A'; B B" C C" '.
of corresponding points. ' , • ale pans
(b) (e)
(a)
FIGURE 7.35
A B
Just as there is more than one way to define conics in
Euclidean analytic geometry, so more than one possibility exists in c·
projective geometry. The first definitions chosen here are credited to
Jacob Steiner and are used because they emphasize dearly the relation~
ship between conics and the projectivities already studied. In the (a) (b)
,,
.,.J
definitions. pencils oJ lines are sets of concurrent lines, and ranges of FIGURE 7.36
points are sets of collinear points.

Two theorems stated and proved for point conics indicate
DEFINITION. A point conic is the set of points that are additional properties.
intersections of corresponding lines in two projectively related pencils of
lines in the same plane. The centers of the projective pencils must be . ' . THEOREM 7.2t. The centerS of the pencils of lines in the
distinct, and the sets cannot be perspective. pi 0JcctlvIty definmg the point conic are also points of the conk.
DEFINITION. A line conic is the set of lines that join . In Figure 7.37, let A and B be the given centers. If ill .
corresponding points in two projectively related ranges of points in the
same plane. The two lines must be distinct, and the set cannot be
considered ,as o~e of the lines in the penciJ with A as center then
correspondmg hue m is one of the lines in the pencil with B as
i:: .;'; .
perspective.
m
It would be well to reflect on these remarkable definitions. /..... ..... - -- .... ......
,
Observe that several previous figures in this chapter have includ~d points
/
)( )(' \
on a conic, although the text has not called attention to the fact. A f>C------=::>I B
\ I
While a complete reconciliation of these definitions with the intuith'e
ideas of what a conic is might be extremely difficult, it is not hard to " ........... _---//
see that the definitions do provide for the fact that no three points
of a point conic are colline~r. Figures 7.36a and b shoW a point conic FIGURE 7.37
268 CHAPTER 7
X are conjugate paints, since the polar of one passes lhrough the other.
center. The_two lines intersect at B, a point on the conic. Similarly.
Similarly, x and a are conjugate lines. Note that the setting for the
A lies on AB, considered as a line through B, and on its corresponding
words pole and polar here is somewhat different rrom that used berore.
line I through A.
In Figure 7.38, conjugate points A and X are distinct. "A
polarity is called a hyperbolic l'0farity if a point call be self-conjugate. '
THEOREM 7.22. The lines corresponding to the common line of
the two pencils of lines determining a point conic are the tangents
at the centers of the'two pencils of lines.
DEFINITION. A tangent to a point conic is a line in the plane

of the conic having ~xactly one point in commOn with the conic.
In Figure 1-37, each line x through A has two points in commOn

with the conic, A and the intersection of x and its corresponding Ijne.
But by Theorem 7.21, if x' is An considered as a line through B.
then x and x' intersect at A. so x must be the tangent I. FIGURE 7.39
A second approach to defining conics in projective geometry

requires the introduction of several new concepts. Up to this point, A self-conjugate point lies on its polar. Figure 7.39 shows varioLis lines
projectivities have been defined in such a way that they pair points with in a hyperbolic polarity, with the self-conjugate points indicated. These
points and lines wi!h lines. It is possible to generalize the definition to concepts can now be used to give an alternate definition for a conic.
include correspondence between points and lines. The original type of
projectivity is a collineation, and the newer type now introduced is a DEFINITION. A point conic-is the set of self-corresponding
correlation. The set of equations representing a correlation would have points in a hyperbolic polarity.
line coordinates for the primed _letters but otherwise would be the same
form as for a collineation. In a co-rrelation, an element and its plane dual
can correspond. Thus, the image of a complete quadrangle is a complete The section concludes with twO major theorems, plane duals of
. " quadrilateral. each other, giving an additional property of any conic.
The projective correlation of particular interest here is a correla-
tion of period two, called a polarity. A polarity pairs a point called the THEOREM 7.23. Pascal's tl,eol"el11. If a simple hexagon is
pole and a line called its polar, such as A and a in Figure 7.38. inscribed in a point conic, the intersection of the three pairs of ?pposite
Any point X on C/ is paired with a line x on A. Points such as A and sides are collinear.
DEFINITION. A simple lJexagon is a set of six points in a

plane, no three collinear, and the lines joining them in a particular
order.
A
Note that this definition allows hexagons in which the opposite
x
sides intersect within the conic.
FIGURE 7.38
Let ACBEFD in Figure 7.40 represent any. inscribed simp~

hexagon. The pairs of opposite sides are lc and ·IfF. CB and rD. •
and BE and DA. Let the points of intersection be P 1- P'1' P 3' It is
required to show that these three points are collinear. /
/ b
8
c c
A P,
P,
Pascal line
FIGURE 7.41
D
EXERCISES 7.9
G
1. What would be the effect of dropping 'he ,e",.·e'.·on that the pencils
FIGURE 7.40 cannot be perspective in the definition of a point conic?
2. Poi~t out figures in this chapter, pdor to Figure 7.35 in which 't
Two points A and B can be chosen as centers of pencils of COIllC could be located. • pOlO S on a
lines in a projectivity determining the conic. There is also a projectivity 3. State the plane dual of Theorem 7.21.
established on IfF and Fri. This means that if AD and IfF meet at G. 4. State the plane dual of Theorem 7.22.
and BEand Fri meet at H. then PI EFG 7i P, HFD. These two sets of
5. Write the e~uati?n of the line that is the image of the point (I,
points have their common element F self-corresponding, and they are the correlatIon WIth equations 1, 1) under
related by a perspectivity (see Exercise 8, Exercise Set 7.9). The center of ~;.
,-,'!
this perspectivity is P3' which implies that PI and ,P'1 lie on n line X l' = Xl + X2 + Xl'
through p 3' as was to be proved. .. X 2' = Xl - Xl + Xl'
The six vertices of a simple hexagon determine sixty different X 3' = XI + Xl - x3 ·
hexagons, found by connecting the points in different orders. Each oE these 6, Prove that ~he line. joining two self-conjugate points in it olarit cann
hexagons in turn has a different Pascal line. The sixty Pascal lines be a self-conjugate line (passing through its pole), P y ot
associated with six given points on a conic are known as Pascal's 7. In Figure 7.38, let the intersection of a and X be point Y At' I
mystic hexagram. s~ch that ea~ vertex is the pole of the opposite side is called' a se,~~a:Ta;
Pascal's theorem was proved by Blaise Pascal (l623-166~), but tnangle. Is triangle AX Y necessarily a self polar triangle?
M
p
the plane dual was not proved until much later, by C. J. Brian chon 8. , Prove that !f two sets of points in a projectivity on two lines have their
(1785-1864), after the development of the concept of duality. comm,on PolU~ self-corresponding, they are related by a perspectivity.
9. Explam why SIX points, no three collinear, determine sixty hexagons,
THEOREM 7.24. Brianchon's theorem. Ifthe six lines of a,simple 10. Prove Brianchon's theorem directly,
hexagon are lines of a line collie, then the three lines connecting pairs
of opposite vertices of the hexagon are concurrent.
7.10 CONSTRUCTION OF CONICS
. .1' ..
Figure 7.41 illustrates the theorem, The point of concurrency is , ~he the.orY,ofSection 7.9 has provided the basis for constructions
called the Brianchon point. of comes In projectIve geometry. using only a straightedge, The fact that
272 CHAPTER 7 .
a
conics have been defined in terms of projectivities means that any number
of points or lines on a conic can be constructed using only a straight- B
edge, if enough informatjon is given to determine a projectivity. This b b'

procedure is analyzed through examples. a'
A kc'---~--"J c
d'
EXAMPLE. Given five points on a point conic. construct other d
points. D
FIGURE 7.43
I'
EXAM PLE. Conslruct additional lines of a line conic. given nve

lines, no three concurrent.
8
c
F b
FIGURE 742 D'

a
Let A. B. C. D. E in Figure 7.42 be the five given points. d

Choose any two of these, say A and D. as centers of the pencils of lines Axis
to determine the projectivity_ Pairs of corresponding lines meet at B. C. E. {
The lines connecting the intersections of the cross joins determine FIGURE 7.44
S. the center ofholTIology.
Let f be another lille through A. Since fe' and e1' must meet
.
L t the five given I'mes be a, /, . c. d. e in Figure 7.44., Two
at S. f' can be constructed. The intersection off and /' is the required
point F. e . d b
r tl se say a an . can be chosen ':lS <
the lines to contalll
_.
the
o le. d' . ts l'n the proJ'ectivity while the other three deterl11ll1e
correspon !fig pom
'
" r ' 0 r·
. ts The intersectlOllS 0 pmrs CIO
ss
three pairs of correspon dIllg pom . . .
EXJ?M PLE. Given four points on a conic and the tangent at one
of them, construct another point of the point conic. J'oins determine the axis .of h0l11ology
F
I, _
b Fe and C F' must meet on lle <lXIS
I . .
If points A. B. C. D. and tangent a arc given as in Figure 7.43, Choose any POlllt on . ....-.- ~ . '.
so that F' can be detenmne. . d TI lei1 F F' is 'another hne of the COI1lt.:.
the projectivity can be determined by the three pairs of lines shown.
274 CHAPTER 7
..
The theorems of Pascal and Brianchon, which include degenerate 3. Locate five new given lines in Figure 7.44 and use the method described in ','.:
forms in which all six given points are not unique, provide an alternative the text to construct a sixth line on the conic. iiI
way of constructing additional points or lines on a conic. 4. Construct another line of a conic, given four lines and the point of contact on
one.. -
Ii
-j!
,I
5. Use Brianchon's theorem to find one additional line of a line conic if five
EXAM PLE. Find one additional point on the same conic as the
lines are given.
five points given in Figure 7.45.
6. Use Pascal's theorem with a degenerate hexagon to construct one more point Ii
7.
on a conic if four points and the tangent at one of these are known.
Given five points o~ ~ point ~onic, lise any method with a straightedge alone
~l
to construct ten add1tlOnai pomts on the conic. ,Ii
il
Ii
i)
B
.Ii
E
c
D
FIGURE 7.45
Let A. B. C. D. E. be the five given points, with I any line

through E not intersecting,one of the given points. The sixth point F
will be located on I so that A. 8. C. D. E. F will be points of an
inscribed hexagon.
The given information is elfough to determine two of the points
on the PascHI line and hence the line itself. Let
r-t . ~.f-t
Ja and 15E meet at P ['
and let Be and I meet at P2' Then CD and AF must meet at a point
~
p 3 011 P t P 2 so that F is the intersection of AP 3 and 1.
This section concludes the presentation of projective geometry as

a more general geometry that includes Euclidean geometry as a special
case.. In the next chapter, a much more general geometry is introduced,
and projective geometry will be jllst a special subgeometry.
EXERCISES 7.10
L Locate five new given points in figure 7.42, and use the method described in
the text to construct a sixth point on the conic.
2. Locate new given points and a given line in Figure 7.43 and use the method
described in the tex110 construct a sixth point on the conic,
CHAPTER 8
GEOMETRIC TOPOLOGY
O'~;?'
. ~. ~
8.1 TOPOLOGICAL TRANSFORMATIONS
From an intuitive (but nol totally correct) point or view, topology

has been thought of as "rubber sheet" geometry. Visualize sets of points
on a thin sheet of rubber. as in Figure 8.1. The sheet may be
stretched and twisted. but no! torn or placed so that l wO distinct points
actually coincide. 1ntuitively. it can be seen that lhe image of a circle
could be an ellipse, a triangle, or u polygon, for example. Straighl
lines are not necessarily changed into straight lines. Most of the cornmon
properties of Euclidean geometry are no longer preserved under the set of
topological transformations. Before YOll read ahead. it would be worlh-
277
GEOMETRIC TOPOLOGY 279
278 CHAPTER 8
onto S2 is continuous
• if for every point of S 2 alld eaClI POS!'t'Ive
1--/-_1/=1
number E t.h~re IS a p~sitive number b such that the image of any point
of.S I that IS In the neighborhood of point A with a radius of fJ is in the
neighborhood
.' . ' of A with a radius of v. Ti,l'S I'd ea may be
of the image 0
stated briefly In the folJowlng notation:

FIGURE 8.1
f[N(A,o) n S,] c N[j(A).e].
while to try to discover some of the properties that are preserved in This relationship is further clarified by a study of Figure 8.2.
topology.
Much of the basic vocabulary of topology has already bc::en
introduced, especially in Chapter 3 on convexity. Historically. topology s,
and convexity are closely connected and share m'any of the same s,
fundamental concepts, such as neighborhood of a point and interior,
exterior, and boundary points. FIGURE B.2
Topology is one orthe modem geometries created within the past
century. Outstanding names in the history of topology indude A. F.
Moebius (1790-1868), J. B. Listing (1808-1882), and Bernhard Riemann
If the preceding paragraph seems particularly difficult you may
profit .from the detailed study or review of the concept of ~ontinuity
;,;
(1826--1866). The study of topology continues 10 grow and develop, with fo~nd m a modern ~alculus text. Intuitively, a continuous transformation
'j
some American mathematicians in the forefront. Courses in topology are ta (es one set of pomts sufficiently near each other into another set of
common at the graduate level. and some are being introduced at the pomts located near.eac~ other. Physically. the tearing apart of a surface
undergraduate level. Intuitive concepts from topology are frequently used could take close pomts mto points at great distances.
as enrichment activities in both secondary and elementary schools. f ~ translation in Euclidean geometry is a very simple example
The introduction of an intuitive idea of topology has shown the o a contmuous transformation. The values of e and bare e I .
need to state and .discuss a formal definition of the subject. Topology is the . t . . qua. SJIlce
. _ pam s I.n any' Circular region have images in another congruent
a branch of mathematics in its own right, but the definition and
:1~u~a3r regIOn. Another example of a continuous transformation is
discussion given here will be limited to the geometric aspects of . : + h - l. whereas an example of a transformation that is not
topology. continuous IS x -+ tan x.
. Topology may be considered as a generalization of both
DEFINITION. Topology is the study of those properties of a Euc!ldean geometry and projective geometry, since the' group of pJane
set of points invariant under the group of bicontinuous transformations lllotlOns and the group of projective transformations are both proper
of a space onto itself. subgroups of the group of topological transformations. A topological
tI·ansformatlOn IS sometimes called a homeomorphism.
DEFINITION. A transformationJis bicontilluOI« if and only if
f andf- I are both continuous. THEOREM 8.1. The set of topological transformations of a
space onto Itself IS a group of transformations.
The concept of a C01f(inuous transformation, used extensively in

calculUS, is reviewed briefly. since its definition makes use of the Questions about the proof of Theorem 8.1 can be found in
topological concept of neighborhood of a point. A transformation of S I
Exercise Set 8.1.
280 CHAPTER 8 .
It was suggested earlier in the section that you think about An example of a simply connected set is shown in Figure 8An,
which properties are preserved under the group of topological transforma- whereas a multiply connected set is shown in Figure 8Ab.
tions. This section concludes with the exploration of one of these
properties that is fundamental.
A basic concept in topology is that of a connected set of points.
DEFINITION. A set is connected if and only if any two points (a) (b)
of the set can be joined by some curve tying wholly in the set. FIGURE 8.4
All convex sets are connected, since any two of their points can What is meant by being continuously deformed to a single
be joined by segments in the set. The requirement for a set to be point can be made clearer by considering a closed curve, like a rubber..
connected is a much looser requirement than for a set of points to be band, lying entirely within the region in Figure 8.4n. The band could be
convex in the following sense: if all the curves in the definition of a shrunk without any point going outside the regIOn. On the other hand.
connected set mllst be segments, then the set is convex. Figure 8.3a suppose the rubber band was in the region in Figure 8Ab, but wrapped .
shows examples of connected sets that are not convex, whereas Figure 8.3b around the hole. There would be no way to shrink it to a point
shows a set that is not connected. without going through points in the hole. A multiply connected set, such
as the one in Figure 8Ab, can be converted into a simply connected
set by making one cut as shown in Figure 8.5. Verify this intuitively
by considering the rubber band situation again. However, a connected
set with two holes, as shown in Figure 8.6a, requires two cuts {such as
those in Figure 8.6b) to convert it into a simply connected set.
(a) (b)
FIGURE 8.3
The property of being a connected set is not itself a topological

invariant. Not ali connected sets of points are topologically equivalent.
FIGURE 8.5
In other words, it is not always possible to find a topological
tra~sformation such that any two given connected sets of points are
images of each other. There are different types of connectivity, however.
and each specific type of connectivity is a topological invariant. This
means that a set of points and its image for any topological trans·
formation have the same type of connectivity. The analysis of types of
connectivity in this section is confined to two-dimensional sets of points (a) (b)
with interior points in a plane. FIGURE 8.6
DEFINITION. In general, ir /1 - 1 noninlersecting cuts. fII~e

DEFINITION. If a set is simply connected, any dosed curve in
neede.d to convert n set in~o a simply connected set, the domam IS
the set can be continuously deformed to a single point in the set. A
connected set that is riot simply connected is multiply connected. n-wply cOl1l1eC[ed.
GEOMETRIC TOPOLOGY 2B3
2B2 CHAPTER B , I·
8. Give a practical example of a flat object from the physical world resembling . ,
:..···1·
a set of points with connectivity of degree:
.p.'
a. One. b. Two. c. Five.
9. Draw two other sets of points topologically equivalent to those of Figure 8.7. I
to. Which of these properties seem to be topological invariants? .;
:'-~
FIGURE B.7 a. Intersection of curves. b. Cross ratio. !1
c. Midpoint. d. Convexity. "-'
For example, the degree of connectivity is tl~ree ~f two, cuts ,are "i
necessary. The degree of connectivity of a set of pomts IS an l~vanant
under the group of topological transformations. For examp~e. Flgu~e 8.7
i:
8.2 SIMPLE CLOSED CURVES
""1'
shows two plane sets of points that are topologically e~Ulvalen~, both ., :.
~ 1
.(
with degree of connectivity of four. These two se~s of p~mts are u~ages Section 8.1 introduced topological transformations and gave an
:u;
of each other for some topological transformaHon. Thts fi:st examp~e example of degree of connectivity as a topological invariant. In this !.'-:
of a topological invariant, degree o~ conne~tivity, is so ~bVlOUslY baSIC geometric approach to topology, informal arguments are necessary more
as to appear to have no significance m Euc1ldean geomeuy. often than in a regular course in topology that', uses analysis
extensively.
The concept of simple closed curve was used in Chapter 3 on
EXERCISES 8.1 convexity. The definitions of Section 3.1 should be reviewed if necessary.
The reason for extending the study of simple closed curves in this
and one
Give one additional example of a continuous tra~srormation chapter is that the property of being a simple closed curve is a
I. additional example of a transformation that is not contmuous. '·i
topological invariant. The image of a simple closed curve under a
Exercises 2 and 3 concern the proof of Theorem 8.1, topological transformation is a simple closed curve.
transformation is always a In general, proving that one curve is the image of another under
2. Explain why the inverse of a bicontilluous
bicontinuous transformation.
topological transformations is difficult. A special case is given to illustrate
3. Outline the proof of the fact that the product of two bicontinuous trans~ the method.
formations is a bicontinuous transformation.
f I motions that are oot
4. Name some invariants under the g:roup~ 0 pane THEOREM 8.2. A simple closed curve that is the boundary of
topological invariants. a two-dimensional convex body is the bicontinuous image of a circle.
5. Write a definition of convex set as a special kind of conne~ted set.
6. Which of the following sets of points are always conoecte~ sets? In Figure 8.8, K is the convex body. K contains an interior
a. Line. b. Circle.
c. Polygon. d. Angle.
e. Hyperbola.
r. Reuleaux triangle.
g. Two concentric circles. h. Half-plane.
i. Open set. j~ Bounded set.
7. Which of these sets are always co,nnected sets?
b. Hyperboloid of two sheets.
n. Sphere.
d. Tetrahedron.
c. Ellipsoid.
e. (x, y, x I x > yl· FIGURE B.B
r. {x, y. z I x. y. and z are rational numbers}.
284 CHAPTER 8 GEOMETRIC TOPOLOGY' 285
point A. and there is a circle with center A lying wholly within K. by moving a pencil over a piece of paper, subject only to the restriction
Each ray with endpoint A intersects the circle and the boundary of the that you cannot crosS a previous path and t~lat you mu~t return (0 the
convex. body in exactly one point each, so there is a transformation starting point. The result might be a c.ol~ph~ated drawmg. suc~~ ~s the
of. the boundary onto the circle. This transformation must be shown to one .m F' . tgure 8. 10. It is hard to dlstlOgUlsh between nllet 101 and
be continuous. Le~ Band C be corresponding points, one on the
boundary of K alld the other on the circle. For s any positive
number, a number (; can be found small enough so that the segment
joining any point of N(B.o) to A will intersect the circle in a point
of N(C, e). Then the' transformation is continuous.
For the transformation to be bicontinuous, the inverse trans-
formation mllst also be continuous. In Figure 8.9, e is any positive
number representing the radius of the neighborhood of point B. on the FIGURE'S.10
boundary of the convex body.
exterior points, let alone to visualize a simple interior. A significa~lt

theorem about simple closed curves proved in topology, but assumed tl1
earlier chapters and in elementary geometry, is called the Jordan curve
theorem.
D
THEOREM 8.3. Jordan curve tlreorem .. Any simple closed curve
in the plane partitions lhe plane into three disjoint connected sels
K
such that the set that is the curve is the boundary of both the other
FIGURE 8.9
sets.
The Jordan curve theorem is proved in this section only for the
On ray AB. it is possible to find a point D with neighbor- following special case in which the simple 'closed curves are polygons.
hood N(D. a) in the interior of both K and N(B. s) and an exterior
point E with neighborhood N(E. Pl in the exterior of K but in the
interior of N(B. e). for some positive numbers IX and p. THEOREM 8.4. Any simple closed polygon ill the plane
Now. [} can be chosen small enough for N(C, 0) that any ray partitions the plane into tlH'ee disjoint connected sets such lhat the set
with endpoint A passing through a point of N(C. J) will also pass that is the polygon is the boundary of the olher two sets.
through both N(D. a) and N(E. (J). Since one of the neighborhoods is
interior to K and one is exterior, a boundary point of K lies on the Let OX be some fixed ray in the plane, not parallel to one of
ray and also lies in N(B. e). It follows that the inverse. transformation. the sides of the polygon. Figure 8.11 shows a typical ~xampte. Any
meets the definition of a continuous function. point P .of the plane may be regarded as the endpol11t of ~ r.a y
Some proofs of the fundamental theorem of algebra depend on parallel to OX. In Figure 8.11, three such' rays. are sh~wn, WIth
an al)plication of topology using ideas very.similar to those in the proof d' P P' and P" The number of points of mtersectlOn of the
en pomts . ' . t' I If the
of Theorem 8.2. rays and the polygon are one, three, and two, respec lve y.
Consider tile intuitive notion of tracing a simple dosed curve number of intersections is odd, the endpoint P is said to have an odd
i
286 CHAPTER 8
GEOMETRIC TOPOLOGY 287 I
I
p
j
D
'I I·
I'
I
A
FIGURE 8.13
.I
FIGURE 8.11
set S or S' can be joined by a polygonal curve that does not intersect !I
arity' if the number of intersections is even, the endpoint has an even the polygon. ff P P' does not intersect the polygon, the statement is
. 'Thus , P and P' have an odd paritYt while pIt has an even
PparIty, obvious. Otherwise, let X be the point of intersection closest to P,
and let X' be the point of intersection closest to p'. as shown in
parity, To find the parity of each point in the planet it is necessary Figure 8.13. Let P" be a point of PP' near P' and with the same
o consider the possibility of a ray passing. through, one ~r, more parity as P'.
:ertices of the polygon. Two P?ssible r~~tio;~hlPS ~fl~dJI~~e;:g~~ee~.~~~ It is possible to trace a polygonal path with sides close to the
sides of the original polygon until a point pm on PP' near P is
the vertex of intersection are 11lustrate m. tgure . . .
reached. pm is between P and X. and the polygonal path from P" to
pm is between P and X. and the polygonal path from PrJ to pm is
entirely composed of points with the same parity.
The original polygon is the boundary of both set S and set S'.
If A is any point of the polygon, then there are points Band C on either
side of A. but arbitrarily close, such that their parity is different.
Since every neighborhood of A contains points of both Sand S', the
polygon is the boundary of both Sand S'. Points in the exterior of
(s) (h)
the polygon have an even parity, and points in the interior have an odd
FIGURE 8.12 parity.
The following two theorems, whose proofs are left as Exercises 6
the vertex is not counted as an intersection. For Figure 8.l~? th~ and 7 of Exercise Set 8.2, give additional properties of simple closed
. ounted The vertex is not counted when the two a Jacen curves and their intersections with one-dimensional convex: sets in the
vertex. IS c . 'd f tl but it is counted
sides of the polygon are on the same Sl e 0 le ray, . same plane.
when they are on opposite sides of the ray. th
Let set S be the set of all points in the plane not on . e
THEOREM 8.5. For one point A in the interior and one point
polygon such that the parity is odd, and let S' be the set ?f all P01~t:
. tI,e plane not on the polygon such that the panty is evel~. e B in the exterior of a simple closed curve S, As
n S is not empty.
111 ' be any segment not intersectll1g
PP
• the po Iygon. Then every pomt on
· . I . S or S' ' '. ~;
P.P' I,as the same parity. and the segment Iles enUre y m THEOREM 8.6. Every ray with an endpoint in the interior of
,
Now it can be shown that any two pom . ts P . P' of the same a simple closed curve intersects the curve.
288 CHAPTER 8 GEOMETRIC TOPOLOGY 269
. Th.e Jord;in curve theorem is involved in various puzzle-type 1. Prove Theorem 8.6.
plOblems In topology. One of these is illustrated in Figure 8.t4a. 8. In Figure 8.15, u..n Cllrves be drawn connecting the three p"irs of points with
the same numbers in stich ~\ way that the curves do not intersect el.\ch
other or any of the other lines given in the figme? Exphlin why or why not.
la) Ib) FIGURE 8.15

FIGURE 8.14
9. Pasch's axiom for Euclidean geometry states that a line in the plane of a
The
-,. problem is to draw curves connecting the three pairs . 0 f pOInts
. triangle which intersects one side of the triangle ul a poillt other thun lhe'
with the same numbers in such a way that the curves do not intersect vertex also intersects a second side. Prove Pasch's axiom as It (hem'em in
~ach o~her or an~ of tbe other lines given in the figure. The problem is topology, using the theorems of this section.
lInposslble to solve, as shown by Figure 8.14b. The given lines and the
l~aths drawn from I-I and 3-3 result in a shaded region representing a
~lIn~,le c1o~ed cUI~ve and its ,interior. One of the points labeled 2 is an 8.3 INVARIANT POINTS AND
Intel10r pOl~t nne. the other IS an exterior point, so any curve connecting NETWORKS
them !TItist mtersect the boundary. .
The study of invariants under the group of topological trnns-
formations is continued in niis section.1n the study of motions of a plnne,
EXERCISES 8.2 invariant points were common, but for more general transfonnalions, they
are more uncommon. One of the simplest examples of a theorem about
I. Which of thes!! sets of points are t opo IOglCU
. II y equivalent to any simple invariant points under topological transformations is the Brouwer fixed
cIosed curve? point theorem for a circular region. This theorem is narned for
a. Triangle. b. Hyperbola. L. E. J. Brouwer, a famous twentieth-century Dutch mathematician.
c. Two concentric circles. d. Boundary of Reuleaux triangle.
2. Which of these plane regions are topologically equivalent any simple
[0 THEOREM 8.7. Brouwer fixed point tlwort!lII. If the points of a
closed curve and its interior?
circuhfr region undergo a continuous transformation so that each image
tl. Tritlnguh~r region. b. Circular region. is a member of the set, then there is at least one fixed point.
c. Plane reglol1 with degree of connectivity olle.
d. Plane region with degree of connectivity two.
3. In Figtll"e. 8.11, explain what happens when vertices C, E of the polygon lue A simple example of Theorem 8.7 is a circle rotated about its
encollnteled by rays drawn parallel to OX with points on P' P as endpoints. center, with the center the fixed point. The theorem can be proved , !"
4. Do. the points in the interior of n simple closed curve have odd indirectly. Assume that there is nO fixed point. The transi'ormution may
~~1 m_ be visualized by associating a vector with each point. as in Figure 8.16(\.
5. Is the exterior of a simp
. Ie c Iosed polygon simply connected
. or multiply The initial point of each vector is the original point. and th~ terminal
connected? point is its image.
6. Prove Theorem 8.5. All of the vectors for points on the boundary of the region
290 CHAPTER B
srnaJier ones results in a continuous change in the transformation vector

3
index. 111is means that the changes cannot take place, since the vector
l\Y index may assume only integral values. No matter how sman the
concentric circle is, the index of the vectors of the transformation will
~ always be one. But this is impossible; the vectors on a sufficiently

smalJ circle will all point in approximately the same direction as the
vector at the center of the circle. since the transformation is continuous.
(b) (e)
(a) If the index of the vectors remains an integer and becomes smaller than
FIGURE 8.16 . one, it wiJI be zero. Thus, the assumption of a vector for every point
has led to a contradiction. so there is at least one fixed point.
. . F· 8 16b Suppose that the points on A recent application of invariant points and topology is in the
point into the cIrcle,. as 111. 19ure . I' k ise order around the circle.
the boundary are consl?ered m a c?un;~r~~c 8~6c shows that the vector
or
field oceanography. In an attempt to explain the theory of continental
The use of a vector diagram as l~ g.. '1 osition It can be shown
drift, a mathematical model is used to illustrate what happens when two :1
Its ofl~ma. ~ of th~ vectors (dot the
rigid plates on a sphere spread out from a ridge crossed by fracture
turns around and comes back t?
.that the total algebraic change ~~ the ~~I:~::n (the index of the vectors
zones. The plates must revolve about an invariant point called the IJOle
of spreading, The rotation also takes place around an axis of spreading.
:e~::)~;.n;;i:~:nU~~ :~:;: Pb~l:~:s~~ering the tangent vectors .for all which passes through the pole of spreading and the center of the earlh.
points on the circle, as in Figure 8.17. For more information, read "The Origin of the Oceans," by Sir Edward
BuJiard, in the September, 1969, Scientific American.
Another application ofthe concept of a fixed point theorem from
topology is in one proof of the fundamental theorem of algebra. Recall
that this theorem ·states that any polynomial equation with complex
coefficients has a root in the field of complex numbers.
The second example of a topologicai invariant in this section is
related to the idea of a network. The word "network" is used to
describe a connected set of vertices, segments, and portions of curves
such as those shown in Figure 8.18. These may be thought of as a set
FIGURE 6.17
The tangent vectors make one complete revolution, an~ jf the

vectors for the points on the circumference ~urn through so~e dllT~r;:~
e the difference is a mUltiple of 2n, ThIs means the vec ors 0
:;a~siormation must turn completely around the tangents at least o~ce,
But since the vectors tum continuously, at some time the. tra~SfO[~~~:o: FIGURE 8.18
nt vector must point in the same directIon.
vector and the tange , . 'd h ' Ie
i possible since all the transformation vectors pOlOt mst e t e CIre .' d
m . Ie concentrie to the given circle and con tame
• any Clre of vertices with paths connecting them. Networks such as these furnish
For
within it the index of the transformation vectors mu~t also be one. examples of what are known as tracing puzzles. Can the networks be
, . Iraced without lifting the pencil from the paper and without repeating
This is true because passmg continuous1y from one circumference to
29Z CHAPTER 8
a portion (tha.t is, llot more than a finite set of isolated points) of the a particular network, the number of paths leading to a vertex is not
path? Try to trace the networks in Figure 8.18a-c before rcading the changed by a topological transformation. For example, the two networks
next paragraph. in Figure 8.19 are topologically equivalent.
The network in Figure 8.18a canDot be -traced, whereas those in
Figures 8.18b and c can be traced. Try to decide what is the essential
difference between networks that can be traced and those that cannot E:<ERCISES 8.3
before you read the next paragraph.
The number of paths leading to each vertex in a network is 1. Could the continuous transformation in Theorem 8.7 be a gl~de reft~clio.n?
crucial in trying to decide whether or not the network can be traced Why?
If a vertex has an even number of paths to it (an even vertex), then 2. Outline briefly the major steps in the proof of Theorem 8.7.
these may be used in pairs in going to and leaving the vertex. If there For Exercises 3-7, tell whether or not the networks in Figure 8.20 can he
are an odd number of paths to a point (an odd vertex), they cannot traced,
be "sed in pairs, and it is necessary to begin or end at that point. If the
network has two or fewer vertices with an odd number of paths, it can
be traced. The network will be a curve in this case. If there are no
vertices with an odd number of paths, then the network can be traced
by beginning and ending at the same point. The network will be a closed .
curve in this case. Now check Figure 8.lSd. This network has more FIGURE 8.20
than two vertices with an odd number of paths, so it cannot be
traced. The i~lformation about tracing puzzles is summarized in the 8. Which of the networks in Exercises 3-7 are curves'!
following table. 9, Which of the networks in Exercises 3-7 are closed curves?
to. Which of the networks in Exercises 3-7 are simple closed curves?
TABLE 8.1
NUll/bel' of Vertices
with Od!' Number Netw(jrlcCml Nt/me of
8.4 INTRODUCTION TO THE
of Paths Be Trawd Nefworf( TOPOLOGY OF SURFACES
more than two no nor a curve
two or fewer yos curve The topology of Sections 8.1-8.3 was the topology of the plane.
zero yes closed curve The last two sections include concepts and theorem.s from the topology
of surfaces in three dimensions. 111is section introduces some basic
theorems for simple closed surfaces, analyzes a topological invariant of
The essential characteristics of a network in terms of the number surfaces, and gives an example of an unsolved problem in topology,
of even and odd vertices is a topological invariant. In other words, for The following theorem is the Jordan curve theorem stated for
three-spat."C.
THEOREM 8.8. Any simple dosed surface in three-space

partitions the space into three disjoint connected sets such that the ~
FIGURE 8.19
set that is the surface is the boundary of both of the other sets.
...
294 CHAPTER 8
The following theorem and corollary can also be proved -ror Figure 8.22. Similarly. many common surfaces such :as an ellipsoid
analogous to the theorem proved for and a convex polyhedron are also of genus zero.
simple closed surfaces; they are
the simple closed curve. \
THEOREM 8.9. For one poml . A m . 11e

I m . tel'ior and
. one point '.I'
.
B in the exterior of a sl'n1ple closed surface S, All () S IS not empty
(see Figure 8.21).
(a) Genus zero (b) Genus one (c) Genus two

FIGURE 8.22 .,.;"1
"'j
Other surfaces exist besides those with a genus of zero.

Figure 8.22b shows a (o ..us with genus one, and Figure 8.22e shows a
surface with genus two. For the figure of genus one, a second cut would
separate the surface into two unconnected parts. Similarly. for the
figure of genus two, a third cut would separate the surface into un.
connected parts. . ,~
FIGURE 8.21 The genus of a surface is another example of an invariant

under any topological transformation. Also, Mny surface of a particular
The following corollary is true becau~e the ~il11ple dosed surface genus may be changed into any other surface of the same genus by a
is bounded; hence the ray must have an extenor pomt. topological transformation. The effect of the two previous statements is
that two surfaces with the same genus are topologically equivalent.
THEOREM 8.10. Every ray with an endpoint in the interior of Figure 8.23 shows several topologically equivalent surfaces of genus one.
a simple closed surface intersects the surface.
The property of bei~g a simple closed, surface is an invarianl

under the group of topological transformations ,Ill, three-space. . < ,_
Corresponding to the study of. conneCtivIty for plane regtdns IS

the analysis of the genus of a surface.
FIGURE 8.23
DEFINITION. The genus of a surface is the largest num~er of

llonintersecting simple closed curves that can be drawn on the sl!rface
An interesting and unsolved topological problem closely
associated with networks, with simple closed curves, and with invariance
without separating it into two unconnected parts.
concerns the coloring of maps. The map-coloring problem has been such
a common enrichment topic in mathematics that many students may be
, zero, since any slmple
For example, the genus of a sp Ilere JS . familiar with it to some extent; nevertheless, it is a current topic of
closed curve separates it into two unconnected parts, as shown m research in topology.
296 CHAPTER B
2 As was pointed Ollt at the beginning of the discussion, the

map-coloring problem is related to other topological concepts. The
boundaries of the countries can be any simple closed curve, so the exact
shape is unimportant. The map-coloring problem changes if the genus
of the surface changes. Theorem 8.11 applies to any 'surface of gel)US
zero, but not to a torus, for example. Special cases of the four-color
FIGURE 8.24 . map problem, for up to some certain number of countries, have been
proved, but the general problem continues to interest mHny topologists.
.F~r two-dimensional maps such as those shown in Figure 8.24
or for slmllar ones on a sphere, only four colors are necessary if each
two countries with a COmmon boundary must have a different color. It is EXERCISES 8.4
~ssllmed that two countries may have the same color if the boundaries
Intersect in . a s~ngle point. This is the case for the two countries 1. Prove Theorem 8.9.
numbered 1 In Figure 8.25.
2. What is the genus of a simple closed surface?
3.. Name or describe olher surfaces of genus zero, one. and two besides those
mentioned in this section.
2
4. Show that the map in figure 8.26 can be colored with not more than foul'
colors,. according to the restrictions of the map~coloring problem.
2
FIGURE 8.25
St.rangely enough, the general statement that any map can be

colored Wit? four colors has not been proved, although Moebius pro-
posed provIng the statement as early as 1840. About 50 years late
FIGURE 8.26
the following theorem was proved. r,
5. What is the minimum number of colors required ttl color each map in
. THEOREM 8.11. Every map on a sphere can be colored accord- F.igure 8.24 according to the rules of the map~colorillg problem'!
l11g to the rules [or the map-coloring problem by using at most "five colors. 6. Draw a map with 12 countries Ihnt requires only three colors, yet stays.
within the map~coloring requirements.
7. Dr.:lw a map with 12 countries tim! requires only two colors 10 fulfill the
The ~ollntries in Theorem 8.11 are assumed to be simply
map~coloritlg requirements.
connec.ted r~glOns. The rules for the map-coloring problem are that two'
8. Explain why it is important that the countries on a llU1P in lhe map·
countrIes wIth a Common boundary must have a different color but that
coloring problem be simply connected regions.
th~y may have the same color if the boundaries intersect in a single
pom!. 9. Explain why it is important in the map-coloring problem that countries with <\
single point in common can have the same color.
..
~ :.
299 CHAPTER 8
What is the relationship between the number of vertices faces
8.5 EULER'S FORMULA AND SPECIAL and edges for these polyhedra 7 Take various other exam~les of
SURFACES polyhedra, if necessary, to arrive at Euler's formula on your own before
This section continues the study of the topology of surfaces by continuing.
introducing Euler's formula for surfaces of different genus and qy
suggesting special surfaces with strange properties unlike those of ordinaf:Y THEOREM 8.12. For a simple closed polyhedron, V + F
Euclidean geometry. = E + 2. where V is the number of vertices, E the number of edges,
The development of the subject of topology has come during and F the number of faces.
the past one hundred years, but it did have its beginning before that.
One of the earliest ideas that is actually to"pological in nature was first To prove the formula, think of cutting out one of the faces
discovered by Descartes, then rediscovered by Euler in 1752. It has since of the polyhedron so tile remaining surface can be stretched out flat by a
gone by the name of Euler's formula. Euler'S formula relates the number topological transformation. The network of vertices and edges in the
of faces, vertices, and edges of a simple polyhedron. Recall that; a plane will have the same number of vertices and edges as in the original
polyhedron is a dosed surface consisting of a number of faces, each of polyhedron, but there is one less face, since one has been removed.
which is a polygonal region. If the surface has no holes in it and can Figure 8.28 shows an example, using a cube.
be deformed into a sphere by a continuous transformation, it is a simple
polyhedron. A simple polyhedron has a genus of zero.
The Greeks showed particular interest in five polyhedra, those
that were regular, with congruent faces and angles. The five regular
polyhedra are shown in Figure 8.27. The names and some of the
FIGURE 8.28
m~GJ
. .- t'. . . No:", triangulate the plane network by drawing diagonals con-
ne.ctm~ vertlces until a triangular decomposition has been achieved In
-- -- : thIS trIangulation,
· I the value of V - E + F is not changed ' sI'noe d lawmg
. '.
(d) (0) eacI1 dlagona adds one edge and one face.
(.) (b) (0)
Once the trian~ulation has been compJeted, the triangles may
FIGURE B.27
~e removed one at a tIme untiJ a single one remains. (This process is
Illustrated for the cube in Fjgure 8.29.) Some of the triao?les will have
numbers of vertices, faces, and edges·' are listed in Table 8.2. The
completion of the table is left as Exercise I, Exercise Set 8.5.
---"
~
, ----, \
\\ /' / '\ [2][7

TABLE B.2
V F E
\
..... -...
\
\ ---.- ,
Figurf! Number Nalllf! oj P(lfyhf!!/nm
4 4 6 FIGURE 6.29
8.21a tetrahedron
octahedron 6 8 12
3.27b 1 1
cube 8
8.27e
1 12 1 edges on the boundary of the network. First, remove any edge of a
·8.27d dodecahedron
icosahedron 1 7 7 boundary triangle not an edge of another triangle.
8.21e
300 CHAPTER 0
THEOREM 8,13, If a closed surface of genus" is partitioned

1. Uthe tdang!::: has only one edge on the boundary, removing that one into regions by a number of vertices joined by curved arcs, then
edge reduces both E and F by one, so V - E + F is not changed, V _ E + F = 2 - 2n tF represents regions and E the at:cs. of the
2. If the triangle has two edges on the boundary, removing them . TI b 2 - 2n is called the Etller c11l11"£IcteI"lS[fc. Note
network). le num er f f s
reduces V by one, E by two, and F by om', so V - E + F is that this reduces to the ordinary Euler formula for a sur ace. 0 genu
unchanged.
zero.
By a continuation of this process, the boundary can be changed
until all that is left of the triangulation network is a single triangle The final topological topic to be introduced .is ~hat of 'a._ ve?
with three edges, three vertices, and one face. For this trinngle, s ecial kind of surface unlike those studied up to thiS. tune. Ordll~a1.y
V - E + F = t. But since V - E + F is an invariant in this process s~rfaces have two sides. but Moebius was the first to discover lha~ It 1.S
of removing triangles, the formula V - E + F = 1 holds for the original ossible to have surfaces with only one side. For a 8m-face of one s_lde. It
plane network, The formula V - E + F = 2 applies to the original is possible to move from any point on the surface to any other Without
polyhedron, which had one lUore face. going over the edge,
Euler's formula V - E + F _= 2 is an example of a topological
property of a figure, since it is unclHlllged under a topological trans-
formation. or course, the polyhedron call be transformed into any
simple closed surface so that the system of vertices, faces, and edges
becomes a network of points, regions, and paths On the surface.
Euler's formula does not hold unless the genus of the surface is FIGURE 8.31
zero. Furthermore, an agreement on how to count the faces is necessary.
For example, the set of points in Figure 8.30 has genus one. To count ln Figure 8.31, which shows a surface ?f two sides: you c~uld
the faces requires an agreement that face AGFD is counted but face not draw a line from A to B. for example. WIthout CrosSlOg ovel an
IJ KL is then subtracted so that the total number of faces is eight.
edge,
Check {o see {hat Euler's formula does not hold for this surface.
The simplest one-sided surface is called a Mo.ebitls ~(rip. It is
'
pictured in F 19ure 8)?
.-' A representation of a MoebIUs strIP can be
G F
I I J
I LN (:
"Kl I ;1
A f-----i-t-+--+-!-__={
0 c
H)---l-+_--I-+_
I I 1M I I
/ I J---+-JN FIGURE 8.32
/ ,/ II
I il. ___ ..Y
/ P 0
. f Give the paper a half-twist
8 c formed from a rectangular stnp 0 paper. . .d . d
and tape the ends together. A Moebius stnp has only one SI e dll
one edge. You can trace a curve from anyone pOil:t on t!le :ut~face
F1GURE 8.30
. 'th t crossing an edge. It IS fnscmattng to

to any other POll1ts WI au 1'1 t
It is interesting to realize that the concept of the Euler formula , {'tl a Moebius strip to see that it behaves un I {e a wo-
ex penmen WI 1 ,
can be generalized for any Closed surface, regardless of genus.
302 CHAPTE R 6
k,- 7. Draw a figure. simiJar [0 Figure S.30, but with a genus of two. Find the
sided surface. For example, if a Moebius strip is cut down a line
number of vertices, faces, and edges, then detennine the Euler· characteristic.
through the middle, it remains in one piece.
A second example of a one-:sided surface is a Klein bottle, S. By constructi~g a ~odel. ~erify that if a Moebius strip is eut down a line
i. through the mIddle, It remains in one piece.
shown in Figure 8.33. This intersecting surface has no inside or outside.
i.
9. W!lat is the result when a Moebius strip is cut lengthwise, beginning one
tlurd of the way from the edge, rather than down the middle?
to. Suppose a. surface is constructed by giving a strip of paper a full twist
befor~ glumg the ends: Is this a Moebius strip? What happens when the
band IS cut down the middle lengthwise?
11. ~n~wer the same questio~s ;ts in Exercise 10, but assume th.e strip of paper
FIGURE 6_34 given one and a half tWists before gluing the ends.
FIGURE 8.33 IS
12. A Moebius strip is a set of points in n-space, for equal to what num~r1
An unusual application of one~sided surfaces for women's dresses appeaied
It
13. A Klein bottle is a set of points in n-space, for It equal to what number?
in Jean 1. Pedersen's article, "Dressing up Mathematics," in The
Mathematics Teach.". February. 1968. A possibly more practical applioa-
tion of the Moebius strip is shown in Figure 8.34. The shaping of a belt
in the form of a Moebius strip connecting two wheels allows the belt
to wear out at the same rate everywhere, n.ot just on one side as in the
usual arrangement. However, it also reverses' the direction of revolution.
The property of being a two-sided or a one-sided surface is a
topological invariant. For example, a Moebius strip cannot be changed
to a two-sided surface by a topological transformation. A related
application of topology that should be mentioned briefly here is the
classification of various kinds of knots. Two knots are topologically
equivalent if one can be deformed into the other in a continuous way.
EXERCISES 8.5
1. Complete Table 8.2 showing the number of vertices, faces, and edges for:lhe
five regular polyhedra.
2. Verify Euler's formula for each of the five regular polyhedra.
3. Verify Euler's formula for a:
a. Hexagonal right prism. b. Octagonal right prislll.
4. Make drawings similar to Figures 8.28 and 8.29 for a tetrahedron.
5, Find the Euler characteristic for the set of points in Figure 8.30.
6. Whal is the numerical value of the Euler constant for a surface of:
a Genus zero? b, Genus one1
c. Genus two? d. Genus three?
CHAPTER 9
NON-EUCLIDEAN GEOMETRY
9.1 INTRODUCTION TO HYPERBOLIC

GEOMETRY
The term non-Euclidean. geometry is used in a very restricted

sense. It does not designate any geometry that is not identical to the
geometry of Euclid. Non-Euclidean geometry dmers from the ge01uelry
of Eudid because it substitutes another altermltive for his so-called
firth postulate. Thus, projective geometry and topology are not examples
of non-Euclidean geometry in this technical sense.
Recall from Chapter 1 that the form of the firth postulate as
stated by Euclid was considerably more complex than the form of the
other postulates and axioms: "If a transversal falls 011 two lines in such
305
306 CHAPTER 9 NON-EUCLIDEAN GEOMETRY 307
a way that the interior angles on one side of the tr~nsversal ~re less these people succeeded. Shortly after 1800, mathematicians such as Carl
than two right angles, then the lines meet on that SIde on Wh~l the Friedrich Gauss began to realize that the fifth postulate could never be
angles are less than two right angles." According to Euclid, AD- and proved from the others. because it was indeed an independent postuJate
Be meet to the right in Figure 9.1 if the sum of the measures of i~ the set of Euclidean postulates, not a theorem. Attempts to prove
i!
L DAB and L ABC is less than 7[ radians. " the fifth postulate by denying it had already produced strange theorems
that had to be accepted as valid if some other substitute postulate
A
was actually possible.
D
As in the discovery of calculus, more than one person shares the
credit for the actual discovery of non-Euclidean geometry. Though Gauss
was aware of the significance of the subject, he did not publish any
matenal. The first account of non-Euclidean geometry to be published
was based on th: assumption that, through a point not on a given line,
m~re than one hne can be drawn parallel to a given line in the plane.
B C
ThIS type of geometry, called hyperbolic non-Euclidean geometry, was
FIGURE 9.1 dIscovered mdependently by a Russian, Nikolai Lobachewsky (1793-
1856), and, at about the same time, by a Hungarian, Johann Bolyai
Note that the word "paratIel" does not appear in the fifth (1802-1860). The results were published about 1830. A second type of
postulate. . non~Euclidean geometry, elliptic geometry, is introduced briefly in
The wording of Euclid's fifth postulate that 15 most commonly Section 9.6.
used is called Play/air's axiom: Through a point not on a given line, The development of hyperbolic geometry in this chapter is based
exactly one line can be drawn in the plane parallel to the given line. on all the assumptions and undefined terms of modern Euclidean
The word "parallel" as used here means not intersecting or having no geometry except for the following substitution for the parallel postulate,
Euclidean point in common. Playfair's axiom and the original fifth ldentlfied as the characteristic postulate of hyperbolic geometry.
postulate are logically equivalent. This means -that either one can be used,
along with the other assumptions of Euclidean geometrYt to prove the CHARACTERISTIC POSTULATE. Through a given point C, J
second. ~ot on a gIven hne is. passes more than one line in the plane not ·'.i
Non-Euclidean geometry provided Einstein with a suitable .model Intersecting the given line.
for his work on relativity. While it also has applications in differential
geometry and elsewhere, it is worthwhile for other reasons as well. The The relationship described in the characteristic postulate is
real understanding of the concept of postulate sometimes comes only pictured in Figure 9.2. if it is assumed that CD and CE are two
when a person begi~s with postulates that are not self-evident. In non- distinct lines through C and that neither intersects All ~I ,
'! ~
Euclidean geometry, it is ·generally not possible to rely on intuition
or on drawings to the same extent as is true for Euclidean ge9metry. D
Finally, non-Euclidean geometry, unlike projective geometry or topology, c
is significant in that it is something other than a generalizatlOn of
Euclidean geometry. Non-Euclidean geometry does not include ordinary £
geometry as a special case.
From the time Euclid stated his postulates. about 300 B.C.,
I •
mathematicians attempted to show that the fifth postulate was actually A B
a theorem that could be proved from the other postulates. None of FIGURE 9.2
303 CHAPTER 9 NON-EUCLIDEAN GEOMETRY 309 -..:Aj
': 'nl~
There are an infinite number of lines passing through C and
. ;;;.i!
The, idea of changing the parallel postulate may seem extremely .. ",Ii
st~ange. but there are many finite geometries where the parallel the interior of angle DCE in Figure 9.2. For Figure 9.3, let a ".\Il!
Ji!:~11
be the
postulate of Euclid does not hold. For example, several of the finite
geometries of Chapter 1 have an axiom stating that each two lines
-
perpendicular from C. and assume that eG. anyone of these illterio~·
~
lines, does intersect AB. This means that CE must also intersect FC.
~
ti,.
I'jll"
intersect at a point. The finite geometry of Desargues has a peculiarity by the axiom of Pasch. which continues to hold in hyperbolic ge(;mlctry
';~;il' .
·!t ~I~;:;
concerning parallels; in that geometry, each point can have three lines since only the axiom of parallels from Eucli~ean geometry has been
through it parallel to one particular line, the polar of the point. replaced.
There is no finite geometry that has all th(~ axioms of Recall that the axiom of Pasch states that a line entering a
triangle at a vertex intersects the opposite side. In Figure 9.3, CE ",\,1 '"
hyperbolic geometry. but some do have the characteristic postulate. 'A
parlicular example is the geometry of thirteen points and twenty-six. enters triangle CFG. The fact that CE intersects An
contradicts the
.~~~:
d.!i'..I"~
lines, represented by the table below, in which each set of three points +-+ -- • .......
assumption that it is parallel to AB. so CG cannot lntersect AB. ,,,~n"'
:
lies on a line.
A,H,C
A,D.E
B.D.F
8, E.l
C.D,G
C,E,l D,l,K
TABLE 9.1
E,G,L
~_G_._M
F,T,)
E
c o
11
Ii,F.fI
A.G,)
A,I,L
B,G.H
B.1.M
B,K,L
C.f,L
C.H.K
C,I,M
D,L.M £,H,M
D_._H_._I__E
__._F_.K
____ ___G_._C, ""']
"j
;I~r'
A,K.M
A F 8 x
F1GURE 9.4
Returning to hyperbolic geometry and its characteristic postulate,
you should pause to consider some of lhe consequences of this change In Figure 9.4, the set of all lines in a plane passing through
in postulates before reading the development of theorems and proofs for C is partitioned into two subsets, those that do intersect in
and those
lhe new plane geometry.
that do nOt. Because of the assumption of the one.:.to-one correspondence
between sets of real numbers and sets of lines through a point, which
_ THEOREM 9.1. Through a given point C. not on a given line was retained from Euclidean geometry, it is known that this partitioning
~ ~
,~
~l~' ,
AB, pass an infinite number of lines not intersecting the given line. is brought about by two different lines, shown as CD and CEo These
two lines must be either the last lines in either direction that do
c
o
-
intersect As or the first lines in either direction that do not intersect
AB. The assumplion that there is a last intersecting line, suy ~ ex.. for
example, is immediately contradicted by the fact that olher points on AB
;',j:
~~,
.-.~,~\
E '>w'
to the right of X also determine intersecting lines. Then CD Hnd CE ":1
are the first lines that do not intersect lB.
~:~;';.,;
'.
;",',
DEFINITION. In hyperbolic geometry, the first lines in either
direction through a point that do not intersect a given line are parallel
.1
A F 8 G ".:l;j
""t!i~
F1GUFlE 9.3 /ines.
k.~
", 'i!{;I!
NON-EUCLIDEAN GEOMETRY 311
310 CHAPTER 9
fact prior to using the fifth postulate in Book I oJ his Elements.
DEFINITION. AU other lines through a point not int:r5e~ting
Furthermore, the angle of parallelism cannot be obtuse, as in Figure 9,6b,
, ' I tl the two parallel Hnes are lIonil1terSectUlg Imes.
the gIven hue, ot "ler nm I because then it would have a nonintersecting tine CA within the angle.
This would contradict the fact that the parallel is defined to be the first
According to these technical definitions, there are exactly :1 wo noncutting line.
.. II I through C to JR. CD is called the right hand pm'allel and The angles of parallelism in hyperbolic geometry are neither
pal(\ e 5 FeD nd FeE are angles
CE is the left hand par"llel. The angles a right nor obtuse, so they must be acute. This illustrates a consequence
oj parallelism fOf- the distance Fe. of the characteristic postulate that is radically different from the statement
for Euclidean angles of parallelism whose measures have a sum of n.
THEOREM 9.2. 111e two angles of parallelism for the s",no
distance are congruent and acute.
EXERCISES 9,1
C
D
E /
,, L Prove Playfair's axiom, assuming Euclid's fifth postulate in the original form.
/
/ 2. Prove the. original statement of Euclid's fifth postulate, assuming Playfair's
/
/ axiom.
/
,/ It can be proved that each of the statements in Exercises 3-8 is equivalent
/
G to Euclid's fifth postulate. RewQrd each sentence so that it becomes a valid
H A F B
statement in non-Euclidean geometry.
FIGURE 9,5
], If a straight line intersects one of two parallel lines, it will always intersect
the other.
and L FeD in Figure 9.5 are angles of
Assume t Ilat L FeE FeD 4. Straight lines parallel to the same straight line are always parallel to one
arallelism for CF. but are not congruen,t. Assume next that L _ • another.
P , I L FCG congruent to
for example. is greater. Then there IS, an ,an g e, . _ FG 5. TIlere exists one triangle for which the sum of the measures of the angles
L FC E and such that EG is in the mtenor of L FC D. If F H ~ . ' is 1C radians,
- A FC H so L FC H ;;: , L FeE. 111is is a contradIctIon
then I::. FCG = '-" . - FCD - LFCE. 6. There exists a pair of similar but noncongruent triangles.
because CE has no point in COl1un<?l1 With AB. Hence, L =
7, There exists a pair of straight line.s the same distance apart at every
point.
=r
A 8. It is always possible to pass a circle through three noncolline.ar points.
9. For which of these finite geometries of Chap~er I dOt:s Euclid's fifth
postulate always hold?
a. Geometry of Pappus. b. Fano's geometry. "
F F c, Four-line geometry. d. Geometry of Des8rgues.
(a) (b) 10: for the thirteen-point finite geometry of this section, lUuue· all the lines
FlGURE 9,6 through point A that do not have a point in common with line BDF.
11. Without using an axiom of parallelism, prove that if a transversal of two
The angles of parallelism cannot be right. angles, a~ in Figure 9::~ lines makes the alternate angles congruent, then the two givell lines do not
',nes shown are nonintersectmg. Euclid proved t intersect.
because the two I
12. In hyperbolic geometry. through a given point not on a given line, exactly The development of the properties of pan'llellines in hyperbolic
how many lines can be drawn in that plane that are pantllel to the given line?
geometry is continued by considering the omega triangle. a three-sided
figure as in Figure 9.8, with one ideal vertex. Though not a triangle
in the ordinary sense, an omega triangle does have some of the same
9.2 IDEAL POINTS AND OMEGA
TRIANGLES properties as a triangle with three ordinary vertices.
In hyperbolic geometry, as in the geometry of the extended THEOREM 9.3, The axiom of Pasch holds for an omega triangle,
Euclidean plane and in projective or inversive geometry. the inclusion whether the line entei's at a vertex or at a point 110t a vertex,
of ideal points is an important idea. In hyperbolic geometry, two
parallel lines do not have an ordinary point in common, but they are
In Figure 9.9, let C be any interior point of the omega triangle
said to meet at an ideal point.
A BQ. Then Be and lc intersect the opposite side because 00 is the
first noncutting line through B for iB and ;I'D is the first nonclitting
DEFINITION. An ideal poillt in hyperbolic geol11<:try is the point line through A for BE. If a line en enters the omega triangle through
of intersection of two para lie! lines. the ideal vertex, it intersects lB. because of Pasch's axiom applied to
ll.ABD. The second part of the proof is left as Exercise 5 of Exercise
A Set 9.2.
E
o· o
8 c
FIGURE 9.7
In Figure 21, the right and le~ hand parallels shown through FIGURE 9 9
point A to line BC meet that line in ideal points G (omega) and
G'. It can be proved (see Exercise 1, Exercise Set 9.2) that G and Q' Euclid proved, without the use of the fifth postulate, that an
are distinct points. An ordinary line in affine geometry has exactly One exte.rior angle. of a triangle has a measure greater than either opposite
ideal point, but a line in hyperbolic geometry has two distinct ideal points. interior angle, and the statement also holds for hyperbolic geometry_ The
familiar relationship of measures of exterior and opposite angles is
8r--_______ modified only slightly for omega triangles.
THEOREM 9.4. For any omega triangle ABQ, the measures of

l1 the exterior angles formed by extending AB are greater Lhan the
measures of their opposite interior angles.
A ' - - - - - ' - -_ _
FIGURE 9.8 This theorem nlay be proved indirectly by eliminating lhe other
,
NON·EUCLIDEAN GEOMETRY 315
314 CHAPTER 9
that AB;; A'B' and that LBAQ;= LB'A'Q'. Assume that one of the
c c angles, say S ABD. is greater than LA' u'n', Some point C can be
located on~n such that L ABC ~ L A' 8'0.'. If C' is located on An-
AI~- __ --J.D~
so that A'C';= AL'. then 6ABC;= 6A' B'C'. But this means that
L A'S'C' ~ LAIB'Q', which is a contradiction.
n . A second set of con~itions for the congruence of omega triangles
sLh.------ IS stated here, but the proof IS left as Exercise 7 of Exercise Set 9.2.
s~------------- £
. THEOREM 9.6. Omega triangles ABQ and A' B'Q' are congruent
If the pall' of angles at A and AI are congruent and the pair of angles
two possibilities. Suppose, ill Figure 9.lO, that IIl(LCAQ) < IIl(L.4BQ).
at Band B' are congruent.
Then a point D on in can be found such that L CAD ;= L ABD.
But this is impossible, since L:1ABD is an ordinary triangle and the
exterior angle cannot be congruent. to an opposite interior· angle.
EXERCISES 9.2
Suppose next, as in Figure 9.11, that L CAQ;= L ABU Let D ~be the
midpoint of AB. let lJE be perpendicular to Bn and let FA = BE.
I. EX!JJain why a line in ·hyperbolic geometry mllst contain two distinct ideal
Then 6FAD;= 6EBD, FDE is a straight line, and LDFA is a right pomts.
angle. But the angle of parallelism for the distance EF cannot be a right
angle because of Theorem 9.2. which means that this is a contradiction
2. Sk~tch three omega tri?ngles. all with the same ideal ver~ex and each two of
which also have an ordinary vertex in common.
and the assumption of congruence of angles must be rejected.
3. Sketch a three-sided figure with two ideal vertices.
Congruence of omega triangles is somewhat simpler than that of
ordinary triangles, since less information is required. One set of con- 4. Sketch a three-sided figure with three idenl vertices.
ditions for congruence is given in the-following theorem. 5. Prove that a .line intersecting a side of an omega triangle at a point other
than a verlex mtersects a second side.
6. Prove that the sum of the measures of the two angles at ordinary vertices
THEOREM 9.5. Omega triangles ABQ and A' B'Q' are congruent of an omega triangle is less than 11:.
if the sides of finite length are ~ongruent and if a pair of corresponding
angles at A and A' or Band B' are congruent.
8. Prove that the angle of parallelism is constant for a given distance.
9. Prove that, as the distance jncreases, the angle of parallelism decreases.
The theorem is proved here by assuming that the remaining pair
10. Prove that if the two .angles at ordinary vertices of an omega triangle are
of angles at an ordinary vertex are not congruent. _then arriving-on
cOl1gr~ent: th.en the h~e rrom the ideal vertex. to the midpoint of the
the basis of this assumption-at a contradiction. In Figure 9.12. ,assume OpPosite Side IS perpendicular to (hat side.
IL Prove the converse of the statement in Exercise to.
s B'
,
" ,,
,, 9.3 QUADRILATERALS AND TRIANGLES
,,
It ~hould not. be .surprising to find that omega triangles have
A c A' SOI11~ pe~uhar p~'opertIes, S1nce they are not identical to any sets of points
studIed 111 EuclIdean geometry. It is even rnore signifidant, however, (0
FIGURE 9 12
316 CHAPTER 9 NON· EUCLIDEAN GEOMETRY 317
consider how the co~cept of omega triangles, developed in Section 9.2. which does not depend on the fifth postulate, continues to hold in non-
leads very logically t~, the formulation of seemingly strange theorems for Euclidean geometry.
ordinary triangles and quadrilaterals in hyperbolic geometry_ Some of the properties of the Saccheri quadrilateral are unlike
Among the attempts to prove Euclid's fifth postulate, the most any for sets of points in Euclidean geometry.
productive were those using the indirect method. ·By adopting a
contradictory postulate and reaching valid conclusions based upon it,
THEOREM 9.8. The summit angles of a Saccheri quadrilateral
mathematicians were aClually developing non-Euclidean geometry. even
are congruent and acute.
though they remained unaware of the significance of their work.
Girolamo Saccheri (1667-1733), in his attempt to prove the fifth
postulate, made use of a set of points now called a SlIcchel'i The congruence or the summit angles is a consequence of the
quadrilateral. A Saccheri quadrilateral has two right angles and two congruence of the pairs of triangles in Figure 9.l4. The significanl
congruent sides, as shown in Figure 9.13. Ali is called the base and fact that the summit angles are acute is a consequence of established
properties of the omega triangle. In Figure 9.15, m( L ECl'l) > m( L C /)l'l),
0 c
o c E
f
A
}
8
t1
FIGURE 9.13
A 8
'CD is called the summit of the quadrilateral. The two congruent segments FIGURE 9.15
are the sides. The next theorem shows that some properties of the
corresponding figures in Euclidean geometry continue to hold in since L fen is an exterior angle for omega triangle C DO.. Since
hyperbolic geometry. LADl'l,,, LBCl'l.IlI(LBC£) > Ill(LtlDC). But LADC =
LBC/), and
therefor" m(LBCE) > Ill(LBCD), so LBCD is acute.
THEOREM 9.7. The segment joining the midpoint of the base l. H. Lambe.t (1728-1777), like Saccheri, attempted lo prove the
and summit of a Saccheri quadrilateral is perpendicular to both. fifth postulate by an indirect argument. He began with a quadrilateral
with three right angles. now called a Lambert quadrilateral, shown in
In Figure 9.14, 6DAE 6CBE. so 6DFE =

6CF£, and the = Figure 9.16.
proor rollows. Note that Euclid's work on congruence of triangles,
0
"-
"-
"-
"-
"-
"-
"-
F
/
/
/
/
/
/
/
C
D FIGURE 9.16
A E B THEOREM 9.9. The fourth angle of a Lambert qmldrilatemi is

FIGURE 914 acute.
318 CHAPTER 9 NON~EUCUOEAN GEOMETRY 319
The proof of Theorem 9.9 follows from the obsel;vatioll that, in For example, if the angle sum were 191[/20. the defect would be
Figure 9.14, EFCB and EFDA are Lambert quadrilaterals. 1[/20.
Theorem 9.9 is needed to prove the following even more
significant theorem, one that serves to clearly distinguish between
THEOREM 9.12. The sum of the measures of the angles of any
hyperbolic and Euclidean geometry.
convex quadrilateral is less than 2n.
THEOREM 9.10. The sum of the measures of the angles of a

You have found, in general, that the theory of congruence of
right triangle is less than n.
triangles in hyperbolic geometry is much like the theory of congruence
in Euclidean geometry. One significant difference results from an
In Figure 9.17, let b.ABC be any right triangle, with 'D the application of Theorem 9.12.
midpoint of the hypotenuse. DE is perpendicular to Be. Line ;tt is
THEOREM 9.13. Two triangles are congruent if the three pairs

A F of corresponding angles are congruent.
The proof is by contradiction. Assume, as in Figure 9.18, that

o
6ABC and ,6,ADE have three pairs of corresponding angJes congruent.
As a result, quadrilateral BCED has the sum of the measures of its
1L.--£?---"'-"-8
C E angles equal to 21e. but this is a contradiction of Theorem 9.12. This
FIGURE 9.17 means the assumption of the existence of the similar but noncongruent
triangles must be rejected for hyperbolic geometry.
constructed so that L FAD", L EBD and AF '" BE. Then b.AFD
'" b.BED. This means that LAFE is a right angle. and LADF A
'" L EDB, so E. D. F is a straight line. The consequence IS that (lCEF
is a Lambert quadrilateral with the acute angle at A. The two angles at
A. L CAB and L BAF, have the sum of their measures equal to the
sum of the measures of L CAB and L CHA. so the sum of the
measures of the three angles of ,6, ABC is less than n:
The following two theorems are corollaries of Theorem 9.10.
The proofs are left as Exercises 8 and 9 of Exercise Set 9.3. o E
FIGURE 9.18
THEOREM 9.1l. The sum or'the measures of the angles of any

One consequence of Theorem 9.13 is that, in hyperbolic geometry, I
triangle is less than 1[, ,I
the shape and size of triangles are not independent. All triangles of the
same shape are necessarily of the same size. Similar but noncongruent
DEFINITION. The difference between 1[ and the angle sum is figures do not exist in hyperbolic geometry. The existence of a pair of
called the defect. such triangles is equivalent to Euclid's fifth postulate.
EXERCISES 9.3 DEFINITION. Two nonintersecting lines are said to meet at i.\
gamma point (r). Another name for this point is an ulu'a-iclelll point.
1. What is the maximum number of angles of a Saccheri quadrilateral that
could be congruent to each other?
Recall that nonintersecting lines do not include the parallel lines
2. III hyperbolic geometry. why can there be no squares or rectangles? ~ ~ I
through a point ~ a line. F~r ex.!mple, if i1C and AB are lhe p~rall~ 5
3. Show that, for a figure such as Figl1l~ 9.19, if AD> Be, then
through A to D£. then AF, AG, and AH are three of the Illfimte
m(LBCD) > m(LADC~
number of nonintersectil1g lines through /1. Each of these lines has a
gamma point in common with DE (Figure 9.20).
4.
:0: FIGURE 9.19
For a Lambert quadrilateral, which is longer. a side ,tdjacent to the acute

c
A
>~==--G
B
F
angle or the side opposite? Prove that your answer is correct. o E

5. Which is longer, tile base or the summit of a Saccheri quadrilateral? FIGURE 9.20
6. Prove that your answer {Q Exercise 5 is correct.
7. Let a triangular region be partitioned into two triangular regions by a Besides having a gamma point in common. nonil1tersecling lines
segment through a vertex. Compare the defect of the original triangle with have. perhaps surprisingly. one of the properties of parallel lines in
the defect of the two smaller triangles formed.
Euclidean geometry: they have a common perpendicublr.
8. Prove Tneorem 9.11.
THEOREM 9.14. Two nonintersecting lines have a common
to. Stale and prove a lheorem giving miuimum conditions that must be known perpendicular.
if two Saccheri quadrilaterals are to be congruent
It. State and prove a theorem giving a different set of minimum conditions
than your answer to Exercise 10. In the analysis figure shown in Figure 9.21. it can be seen thal
1! I
9.4 PAIRS OF LINES AND AREA OF
TRIANGULAR REGIONS
Two parallel lines in hyperbolic geometry meet in an ideal point

because the concept of ideal point was defined in that way. While the DEC
proofs are not included in this text. it is possible to prove thaI if one line FIGURE 9 21
is parallel to a second, then the second is parallel to the first, and that
if two lines are both parallel to a third in the same sense. they are also AB and CD, on two nonintersecling lines. call be though of as l1~e base
parallel to each other. and summit of S~\ccheri quadrilateral ABeD; hence EF. connet.:lmg lhe ·.1
One mOre type of point must be created for u complete midpoints of the base and summit, is the common perpendicul~ The
discussion of the set of points in hyperbolic geometry. problem is reduced to finding two congruent segments. such as AD and
BC. both perpendicular to one of the given nonintersecting lines, since nature of a set of points with each member the same distance from a
the common perpendicular can then be found. given Jin~ (see Secti~n 9.5). Notice also in this connection that defining
paraIJeJ hnes as two hnes everywhere equidistant is equivalent to assuming
the fifth postulate.
The fact that there is no square in hyperbolic geometry means
tha~ a method of measuring the area of a pJane region must be
de:ls~d that do~s not depend on square units. Congruent triangles
eXIst III hyperbolIc geometry. and the theory of area can be based on
o C H this concept in a way very similar to the modern theory of area in
Euclidean geometry.
FIGURE 9.22
Assume, as in Figure 9.22, that AD and Be are perpendicular .DEFINITION. Two polygons are called equivaielll if they can
to the same line, but not congruent. Let ED ~ BC, and L FED b~ partItIOned Into the same finite number of pairs of congruent
~ L FBe. If BV ~ EF and CH ~ DI, then quadrilateral BGHC can be tnangles.
shown to be congruent to EFID. and GH is perpendicular at H; thus,
BC and GH are the required sides for the Saccheri quadrilateral. Note . For example, polygons ABCD and CDEF in Figure 9.23 are
that the proof of the existence of point F on line AB has not been eqUIvalent but not congruent. Two polygons both equivalent to another
included here.
Because of Theorem 9.14, all of the lines perpendicular to a given
E
line can be said to have the same ultra-ideal point in common. It is A'_7_-,Br---iF
left as an exercise to show that two lines in hyperbolic geometry
cannot have two distinct common perpendiculars.
With the inclusion of the previous information about the .gamma
point and the common perpendicular for nonintersecting lines, Table 9.2
o C
can be given summarizing the relationships between pairs of lines inter-
FIGURE 9.23
secting at each kind of point in hyperbolic geometl'y.
polygon are also equivalent to each other. Recall that in hyperbolic

TABLE 9.2
geometry. the difference between n: and the sum of the measures of a
Type of Poi"c Camillo" tel V (friar/on ill Distallce triangle is defined as the defect of the triangle. The connection between
Two Lilies the defect and equivalence of triangles is made clear in the foIJowing
theorem.
Ordinary point Lines diverge from their point of intersection
Meal point Lines converge in the direction of parallelism
and diverge in the opposite direction
_._-_ _ - - -
U Ilra.-ideal point
..
Lilies. diverge from their common perpendicular THEOREM 9.15. Two triangles are equivalent if and only if
they have the same defect.
You may find it interesting to observe that in none of the cases If two triangles are equivalent, they can be partitioned into a
are the two lines always equidistant. You may speculate about the finite number of pairs of congruent triangles. The defect of each of the
original triangles is equal to the sum of the defects of the triangles' is equivalent to Saccheri quadrilateral BUe. Point [( on [] is a point
in the partitioning; hence, the original defects are equal. See Exercise 7, located so that KC = J:Df. and IT '" Df. Triangles ABC and LBC
Exercise Set 9.4. can be shown to have the same defect and to be equivalent. Triangles '·r· .
.. ::!
Now suppose that two triangles have the same defect. If they LBe and DEF also have the same derect ane? a pair of congruent
also have a pair of corresponding sides congruent, they can be shown sides, hence are equivalent. Since 6.DEF and 6.ABC are both equivalent ;11
to be equivalent to congruent Saccheri quadrilaterals and hence to the same triangle, they are equivalent to each other.
equivalent to each other. 1n the proof of Theorem 9.15, the significance of the defect of a
triangle was that it made possible the equivalence of trinngle~ t<?
A H
Saccheri quadrilaterals that could be proved congruenl.- Since the
equivalence of triangles depends on the defect, it is possible to define the
~
VE
G
C
D'S]
F
8
K;:]
J7--------------
M measure. of area of a triangle in hyperbolic geometry as fa!. where d is the
defect and 1< is a posilive constant the same for all triangles in
hyperbolic geometry. The value of!c depends on the particular triangle
chosen to have a unit ,;u-ea, It should be evident lhat the measure of area
FIGURE 9.24 increases as the defect increases. In other words, the larger the triangle,
the smaller the sum of the measllres of the angles.
. For example, in Figure 9.24, let triangles ABC and JNI have the Oddly enough, in hyperbolic geometry, triangles do not becon1e
same defect and congruent sides lJC and fl, Triangle ABC is equivalent larger and larger without limit. The triangle with three ideal vertices,
t~Saccheri ~uadrilateral BeGF, where D and E are the midpoints of
AC and All (see Exercise 8, Exercise Set 9.4). Triangle HI} is o
/~
equivalent to Saccheri quadrilateral lJ NM, where K and L are the
midpoints of m
and m. But the two Saccheri quadrilaterals are
congruent because they have congruent summits and congruent summit
angles.
0" , • O·
Fimllly. suppose that two triangles have lhe same defect but no
pair of congruent sides. Let triangles ABC and DEF be any t~o such FIGURE 9.26
triangles, as in Figure 9.25, with DF > Ae.
sllOwn in Figure 9.26, is the triangle of maximum area. and its angle
A
D Slim could be considered zero.
L
I
I
I EXERCISES 9.4
I
,
I L Explain why two distinct lines cannot have more thun one common
I
I perpendicular.
E~--------'" F 2. Where are tWO nonintersecting lines closest logether?
8 c
FIGURE 9.25 3. How many gamma points are on each line in hyperbolic geometry·r
4. Do two lines in hyperbolic geometry always diverge from their point of
If G and Ii are the midpoints of AS and AC. then 6ABC intersection?
.,
5. Prove that two parallel Jines converge continuously in the direction: of

because ,6,ABO is isosceles. This leads to a generalized definition of
parallelism. corresponding points.
6. Prove that two polygonal regions e(luivalent to the same polygonal region
are equivalent to each other.
.
P love 1I t the defect of a triangle is equal to the sum of the defects DEFINITION. Two points, one on each of two lines, are called
7. 18 .. I . h d' fo
of the triangles formed by partitioning the ongma, uSlOg 1 e me Ian f, m corresponding points if the two lines form congruent angles on the same
one vertex. side with the segment whose endpoints are the two given points.
8. Prove that 6.ABC. Figure 9.24, is equivalent to Saccheri quadrilateral
DeGF.
Thus, A and B in Figure 9.27a afe corresponding points because
9. Why. in Figure 9.24, do the two Saccheri quadrilaterals -have congruent
of the congruence of angles DAB and OBA. Points C and B, or C and
summit angles?
A. similarly can be shown to be corresponding points. In fact. any
10. Make a sketch to show how a hexagon and a triangle could be equivalent.
two points on a cirde are corresponding points. An ordinary circle
11. Show how. starling with any triangle, a triangle of maximum area can be
may be defined as the set of all points corresponding to a given point
drawn containing the given triangle in its interior.
(nol the center) on a pencil of rays with an ordinary point as center. A
pencil of rays is all rays in a plane with a common endpoint. Because
of the fact that the tangent to a circle is perpendicular (orthogonal)
to the radius at the point of contact, a circle can be considered the
In hyperbolic geometry, much of the Euclidean theory of circles orthogonal trajectory of a pencil of rays with an ordinary vertex.
must be modified because it depends on the fifth postulate. For example, Two new curves, considered generalized circles in hyperbolic
the measure of an inscribed angle is no longer equal to half the geometry, may now be defined.
measure of its intercepted arc. At the same time, the definition of
circles and those properties relating to circles that do not depend on the
DEFINITION. A limiting eU/"i'e (see Figure 9.27b) is the set of all
fifth postulate, -such as some having to do with perpendicularity, are
points corresponding to a given point on a pencil of rays with an ideal
parts of hyperbolic geometry. . . point as vertex.
Defining ideal and ultra-ideal points so that each .two hnes 10
hyperbolic geometry have a point in common leads to apphcatlOns that
are generalizations involving circles. ..,' A limiting curve may be considered the orthogonal trajectory of
For any two points A and B on an ordmary clrcle, ~s' tn a pencil of rays with an ideal vertex.
Figure 9.27a, L DAB:;;; L DBA, where D is the center of the Cll'ele,
DEFINITION. An equidistant curve (see Figure 9.27c) is the set of
all points corresponding to a given point on a pencil of rays with a
common perpendicular.
(a)
Q
tb)
ttJ
J K
(e)
L
An equidistant curve may be considered the orthog::mal trajectory
of a pencil of rays with a common perpendicular.
A liiniting curve in hyperbolic geometry has many of the
properties of an ordin'ary circle. For example, a line perpendicular to a
chord at its midpoint is a radius. Theorem 9.16 states a se~ond common
FIGURE 9.27 property.
THEOREM 9.16. Three distinct points on a limiting curve so that omega triangle A' B'n' is congruent to omega triangle ABO.
uniquely determine it. The result is that LABQ~· LA'B'!}.', and B' and A' are corresponding
points with B' on the second limiting curve. In the same way, it can be
shown that points C. D', ... on the limiting curve can be found so that
In Figure 9.28, the perpendicular bisectors of AB and BC, for il'C' ;;;; BC. CD' ;;;; CD, and so on. A corollary or Theorem 9.17 is lhe
A, B, C on a limiting curve, determine the unique point H, the center statement that congruent chords intercept congruent arcs and congruent
of the limiting curve. It should be emphasized that a part of the given arcs are intercepted by congruent chords for limiting curves.
condition was that the three points were on some limiting curve. This is The second new type of curve to be studied in hypei:bolic
not necessarily the case for any three points chosen at random. For
:~.,.
geometry is the equidistant curve. Let A, B. C be any three poit~ts all
example, in Figure 9.28, three points A, B. C lie on an ordinary circle an equidistant curve, as shown in Figure 9.30. The common perpelld.Jcul~r•.
if the' perpendicular bisectors at D and E meet at a real point. .1
~(IJ
o o E F
:1
The following theorem states a property of limiting curves that OF. j's ;:alled the baseline. Quadrilateral ABED is a Saccheri quadrilnteral,
is unlike any property of circles. since A. and B are corresponding points and L BAD -= L ABE. The
name "equidistant curve" is justified because every point on the curve is
THEOREM 9.17. Any two different limiting curves are congruent. the same perpendicular distance from the baseline.
All the points in the plane the same distance from the baseline
actuai1y lie on an equidistant curve of two branches, as shown in
Let ABeD and A'B'C'D' in Figure 9.29 be any two limiting
Figure 9.31. In Exercise SeI9.5, properties of the figure ABCD that are
curves. Suppose that points A, B. C, D are given and that A' is any
surprisingly like those of a parallelogram are investigated. The equidistant
curve has many other properties, a few "of which are stated here. Though
S' C'
,~ \IT! , A B
.j
...-' ["-
,- , 1
,I : ·I~
!
o . j
}
I
FIGURE 9.29 "
V
point on a second limiting curve. If L B' A'h' is constructed congruent
1 '-
0 C
to L BAn. and if A' B' is congruent to AB, then point B' is located
j FIGURE 9.31
1
1
j
]
330 C HAPTE R 9 NON-EUCliDEAN GEOMETRY 331
it is not as fundamental as the limiting curve in the analytic treatment 10. Compare the properties of ABeD in Figure 9.31 with those of a parallelogram
or non-Euclidean geometry, it is simpler to study. in Euclidean geometry. .
Three points on an equidistant curve determine the curve
uniquely, since the baselines can be determined. Not all equidistant
curves are congruent-only those whose points are the same distance 9.6 ELLIPTIC GEOMETRY
from the baseline. For congruent equidistant curves, congruent chords
intercept congruent arcs, and congruent arcs are intercepted by congruent The first five sections of this chapter have included a look at
chqrds. Given one line, an intersecting line diverges from the point of many of the major ideas of hyperbolic non-Euclidean geometry. Not
intersection; in addition, parallel lines and llonintersecting lines are at.· a long after the development of hyperbolic geometry, the German
nonconstant distance. It is only the equidistant curve in hyperbolic mathematician Riemann (1826-1866) suggested a geometry, now called
geometl)' that has the equidistant property characteristic of parallel lines elliptic. based on the alternative to the fifth postulate, which states
in Euclidean geometry. that there are no parallels to a line through a point On the iine.
The statement that a circle can. always be found that passes
through three noncollineal' points is equivalent to the fifth postulate or
CHARACTERISTIC POSTULATE OF ELLIPTIC GEO-
Euclid. In hyperbolic geometry. on the other hand, three noncollinear METR Y. Any two lines in a plane meet at an ordinary point. .i,
points may lie on a circle, a limiting curve, or on one branch of all
equidistant curve.
Various finite geometries from Chapter 1. such as the three- Ii
point geometry and Fano's geometry, satisfy the requirement of this ,i
EXERCISES 9.5 postulate, although they do not satisry all the postul.,es or elliptic
geometry.
,;i
I. In hyperbolic geometry,.is the measure of an inscribed angle less thall i.or It is also necessary to further modify the postulational system
greater than half the measure of its intercepted arc? of Euclid by replacing the statement about infinitude of a line with the
2. Why are any two given points on an ordinary circle corresponding points?, milder statement that a line is boundless in elliptic geometry. An
3. Prove that the segments of radii between ~ny pair of limiting curves with intuitive idea of the meaning of the word boundless is that the line
the same ideal center are congruent. cannot be enclosed by a circle lying in the same pJane. That is,
4. Could a straight Hne intersect a limiting curve in three distinct points? boundless means unbounded, as it was used in Chapter 3.
Why? If Figure 2J2 is in elliptic geometry, with EH and n
both
5. Could a straight line intersect one branyb of an equidistant curve in three perpendiCUlar to CG. these two lines meet at some pOint A because
distinct points? Why?
A
6. Show how to construct the baseline' of an equidistant curve, given three
points 011 one branch of the curve.
7. Prove that any segment connecting two poillts, one on each branch of an
equidistant curve, is bisected by the baseline.
8. Explain how to construct the baseline of an equidistant curve given three
points, not all of which are on rhe same branch of the curve.
9. Show that three ditlereot equidistant curves pass through the vertices of a
triangle. with twO vertices 011 one branch and the third vertex on the other B
bnlllch.
FIGURE 9,32
. I
of the characteristic postulate that any two lines interse.ct. In elliptic The use of a model also helps to explain whut it means when a
geometry, it is customary to use curves to show straight lines. U line is called boundless. A great circle on the sphere, representin.g i.l
i5E: '" IT. then 6AEF '" 6AED. and ;ill is also perpendicular to line in elliptic geometry. cannot be enclosed by a curve On the sphere.
eG. By an extensio11 0!..11is argument, it can be shown that every line There 'js no way to "get around" the great circle from a point on one
thr~h A intersects CG at right angles. Point A is called a pole side orit to a point on the other without intersecting the great circle.
of CG and the line is the polar for point A. Here, the distance from In elliptic geometry, there are no parallel or nonintersecting
A to any point on CG is a constant. Recall other uses of the words lines, since any two lines meet. However. there are quadrilaterals and
pole and polar to show a relationship between a point and a line in triangles that have some properties analogous to those encountered in
various geometries encountered in this text. hyperbolic geomelry.
In Figure 9.32, if At: '" EB. then 6BEF '" 6AEF. and A. F.
and B are also collinear. This means that B is also a pole of eG, and THEOREM 9.18. The segment joining lhe midpoinl or the base
that two lines intersect in two pojnts. 1t is assumed here that A and B and summit of a Saccheri quadrilateral is perpendicular to both the base
are distinct points. although it is also possible to consider them as and the s!..lmmit.
identical (in single elliptic geometry). The commOll polar of the two points
of intersection is the unique common perpendicular to the lWo lines.
The proof of this theorem is identical to the proof of Theorem 9.7
Interestingly enough, two straight lines, such as ADB and AFB, enclose a
region in elliptic geometry. This region is called a digOl1 or biangle. for hyperbolic geometry.
Fortunately, the elliptic geometry of the plane can be explained
conveniently by comparison with a familiar model, the earth and lines THEOREM 9.19. The summit angles of a Saccheri quadrilaleral
of longitude on its surface. In Figure 9.32, think of A and B as the are congruent and obtuse.
north and south. poles, and CG as the equator. Be very careful to
observe that the geometry on the surrace of a sphere is not non- In Figure 9.34, let ABCD be any Saccheri quadrilateral, with
Euclidean but instead provides a three-dimensional model for two- o and 0' the poles of EF. the line joining the midpoints of the base
dimensional elliptic geometry.
and summit.
In Figure 9.33, let Hi and KJ be lines in elliptic geometry
K
/-"'--------,
A E 8
J
FIGURE 9.34
The fact that the summit angles are congruent comes from the
FIGURE 9.33 congruent triangles used in the proof of Theorem 9.18.
To show that the summit angles are obtllse, it can first be
suc~hat H and I are lhe poles of KJ. and K and J are the poles established that their complements are acute. If, as in Figllre 9.34, X
of HI. In elliptic geometry, the distance from any line to its pole is lies on BO' and is Ihe pole of Be. then BX > BO·. since BO' < EO'.
COl1stant-the same for all lines. Furthermore, a line is or finite length.. This means that LBCX is a right angle, so LBCO' is acute and its
and the length is four times the distance from the pole to the line. supplement, LBCD. is obtuse.
'i
1
The following theorems are consequences of Theorem 9.19, and 8. Verify the fact that the proof of Theorem 9.18 is the same as that for
the proofs are len as Exercises 10-12 of Exercise Set 9.6. Theorem 9.7.
I
9. In elliptic geometry. where are two given lines farthest apart?
THEOREM 9.20. [n elliptic geometry, a Lambert quadrilateral 10. Prove ll1eorem 9.20.
has its fourth angle obtuse, and each side of this angle is shorter than II. . Prove Theorem 9.21-
the side opposite. 12. Prove Theorem 9.22.
13. Prove that similar but noncongruent triangles cannot exist in elliptic
geometry.
triangle is greater than 1t.
9.7 CONSISTENCY OF NON-EUCLIDEAN
GEOMETRY
CJuadrilateral is greater than 2n.
The question of which kind of geometry is the "right" geometry
to fit the physical universe is one that may never be answered.
You should find it worthwhile to return to the sections on While it might seem a simple problem to solve, there are formidable
hyperboJic geometry to see which oCthe concepts not already mentioned difficulties. For example, it might seem that physical measurement of the
can be used in elliptic geometry. For example, there is no angle of angles of triangles would easily settle the question of whether the sum
parallelism in elliptic geometry. Circles do exist in elliptic geometry. is or is not equal to n. But physical measurement always involves
They can be described as the set of all points a fixed distance from a errors. Besides, it is known that in non-Euclidean geometry the angle
given point. Since each point on a circle is also the same distance from sum depends on the size of the triangle. The small part of the universe
the polar for the given point, a circle can be considered as an in which we live may not be big enough to contain triangles with a
equidistant curve in elliptic geometry. defect (or excess) large enough to be measured. Space traveJ may someday
settle the question.
For most practical purposes, it makes little dilTerence in our lives
whether the universe is Euclidean or not. Euclidean geometry provides
EXERCISES 9.6 a simple model to use in most practical applications, such as engineering.
Einstein's general theory of relativity asserts that physical ;;pace that is in
1. Name two other geometries from Chapter I for which the characteristic
the neighborhood of any kind of matter is best described by the
postulate of eUiptic geometry holds.
postulates of elliptic geometry. Non-Euclidean geometry may have new
2. In ordinary Euclidean geometry, is a line bo.undless?
applications as man explores more of the universe. In many cases, it could
3. A line of latitude that is not the equator of a sphere represents what be used instead of Euclidean geometry without making much difference.
other. concept in the model of elliptic geometry shown in Figure 9.337
From a purely mathematical point of view, the truth of an
4. In elliptic geometry, if the distance from :-a line to its pole has a measui"e
axiomatic system is not what must be investigated. The important
of two, what is the measure of the length of a line in that geometry?
thing to determine is whether or not the system is consistent, as
5. When 111ny two points not always determine a unique line in elliptic
discussed briefly in Chapter 1. In other words, do the axioms of non-
geometry?
Euclidean geometry lead to valid conclusions, without any contradictions.
6. Do limiting curves exist in elliptic geometry? Why?
Actually, the mathematician is interested in what is called relative
7. What is the maximum measure of lhe third angle in a triangle in elliptic consistency. He needs to be sure that non-Euclidean ~eometry is as
geometry, if two of lhe angles are right angles? consistent as Euclidean geometry 01·· as the algebra of real numbers.
,
NON.EUCLIDEAN GEOMETRY 337
336 CHAPTER 9
Beltrami is given the credit for first proving the consistency. of Poincare and the Klein models for hyperbolic geometry is illustrated
non-Euclidean geometry, in 1868. The proof of the relative consistency of a in Figure 9.36. On a plane tangent to the sphere, Klein's model is
non-Euclidean geometry consists of finding a model within Euclidean shown as the interior of the circle C' congruent to the great circle e" of the
geometry that, with suitable interpretations. has the same postulational sphere.
structure as the non-Euclidean geometry. Then any inconsistency in the
;l-
non-Euclidean geometry would mean there is also an inconsistency in
Euclidean geometry.
It has already been mentioned that the lines of longitude on a
sphere provide a three-dimensional model for lines in plane elliptic
geometry. More generally, great circles in the model correspond to lines,
and each line has two poles associated with it. Lines of latitude are ., j
models for equidistant curves.

Two-dimensional models in Euclidean geometry for hyperbolic :1
i
geometry include those suggested by Felix Klein and Poincare. The
Klein model is illustrated in Figure 9.35. Points on the circle represent
ideal points. Sec'ants of the circle represent hyperbolic lines, with points The points on C' and its interior are projected onto the bollom
on the chords interior to the circle representioj ordin.ary points. In
-is -
Figure 9.35. and CD are intersecting lines AD and CD are parallel
lines, and AD and BC are nonintersecting lines.'
hemisphere of the sphere by a projection from an ideal point. Each
chord, sllch as a, is projected into the arc a' of a circle orthogonal
to C". Now, the points on the bottom hemisphere are carried by a
stereographic projection (considered in Section 6.4 as an application of
A A::-----;:'" 0
E
inversive geometry) into the points on circle C and its interior. The arc
a' is projected into an arc a" orthogonal to circle C in the tangent plane. ..
Poincare's model for ordinary points in hyperbolic geometry is illustrated
by- the interior of C. The ordinary points on a line are illustrated by
C~-----~B the points on an arc, such as a", interior to C.
According to the development above, lines in Poincare's model
are represented by arcs of circles orthogonal to a give~ circle. }E
Fi"ure 9.37, AB and fG represent intersecting lines, AB and Be
FIGURE 9.35
re;resent parallel lines, and iIi and fiE represent nonintersecling lines.
1
One advantage of the Klein model is that straight lines are
B
represented by portions of straight lines. Two major problems in con-
nection with the Klein model soon become apparent, however. One is
that some interpretation of distance must be given so that a segment such
as AD in Figure 9.35 has the characteristics of a line of infinite
I
length. The second major problem is that some interpretation of angle
must be given to avoid having the sum of the measures of the angles
li E
c
of a triangle equal to 1[. j

The Poincare model eliminates the second problf:m named in the
previous parag.raph, but not the first. The relationship between the
I fiGURE 9.37
,
1
j
·1
NON-EUCLIDEAN GEOMETRY 339
338 CHAPTER 9
The ingenious way in which distance is defined in Poincare's show the correct relative positions of the set of points in (a) and its
model is shown in the following formula, which refers to Figure 9.38. inverse (b), since this relative position is unimportant in the proof.
Actually the sets of points would overlap if drawn correctly. but such a
AC drawing would be confusing. thus making it harder to interpret relation-
CB ships within the inverted figure.
CD = !dog, AD .
DB
where AC, CB, AD. and DB are lengths of segments, not arcs, and k
A'
is a parameter. Check to see that this rormula does yield the correct
8'
results when the distance between two points is infinite (such as for D
and B in Figure 9.38) or is zero-that is, if the two points coincide. ,,
Also note that the concept of cross ratio used in the rormula is familiar
from the related concept studied in the projective geometry of Chapter 7.
,
-~o
\ ,
(a) (b)
FIGURE 9.39
The relatively brief discussion of consistency of non-Euclidean

geometries in this section has used several concepts from earlier chapters.
Each of the many geometries presented in this book has importance in its
own right, but each also contributes to a better understanding of the
other geometries and to a fuller appreciation of the meaning of the
FIGURE 9.38 expression "modern geometries." Other modern geometries, especially
those more dependent on an a,nalytic approach, remain to be explored
Fortunately, the measure of angles in Poincare's model may be by the interested reader. .
defined in the usual way for Euclidean geometry. This is true because
angles are preserved under a stereo graphic projection. Inversion is used as
a key tool to show that the geometry of points and lines in Poincare's
model- has the same postulational structure as the geometry of the EXERCISES 9.7
hyperbolic plane. For exampJe. a proof by inversion is u·<ied to prove that, ..
"
Exercises 1-4 refer to lhe Klein model.
in. the Poincare model. the sum of the measures of the angles of a r.-
t. Draw a hiangle with three ordinary points as vertices.
triangle is less than n.
In. Figure 9.39a, triangle ABC. which represents any triangle in the 2. Draw an omega triangle.
Poincare model, is inverted into 'A'B'C', shown in Figme 9.39b, by an 3. Draw a triangle with exactly two ideal vertices.
inversion with center 0, the inverse point of point B. with the fixed 4. Draw a triangle with three ideal vertices.
circle as the circle of inversion. It is apparent from the shape of A'B'C',
Exercises 5-8 refer to the Poincare model.
since arc A'e bends inward, that the sum of the angle measures is
less than n. It is important to note that Figure 9.39 does not actualJy 5. Draw an example of an angle of parallelism.
340 CHAPTER 9 APPENDIX I •r
(
6. Draw"an omega triangle.

,.
7. Draw a triangle with exactly two ideal vertices. HILBERT'S AXIOMS
8. Draw a triangle with three ideal vertices.
9. In Figure 9.38, lind the distance AD, using lhe formula given.
to. In the proof by inversion about the measures of angles, explain in detail how
Figure 9.39b was derived from Figure 9.39a.
1,
,I
..,
i
The axioms in Appendix 1 are reprinted from David Hilbert,
\
The Fouudqtiol1S of Geometry. La Salle: Open Court Publishing
Company, 1950, by permission of Open Court Publishing Company.
A new tenth edition of The Foumlations of Geometry was published
by Open Court Publishing Company in 1971.
GROUP L AXIOMS OF CONNECTION
I, 1. Two distinct points A and B always completely determine a

straight line a. We write AB = a or BA = a.
1,2. Any two distinct points of a straight line completely determine
341
342 APPENDIX 1 HILBERT'S AXIOMS 343
that line; that is, if AB = a and AC = a, where B "" C. then is which does not intersect the line Ct. This straight line is calJed the
also BC = a. parallel to a through the given point A. ,i I
I, 3. Three points A. B, C not situated in the same straight line
always completely determine a plane 0::, We write ABC = 0:..
1,4. Any three points A, B, C of a plane 0:, which do not lie ill the GROUP IV. AXIOMS OF CONGRUENCE
same straight line, completely determine that plane.
1,5. If two points A, B of a straight line a lie in a plane a, then IV, 1. If A. B are two points on a straight line a, and if A' is a point
upon the same or another straight line 0', then, upon a given side
every point of a lies in 0:.
1,6. If two planes a, IJ have a point A in common. then they have. of A' on the straight line 0'. we can always find one and only one
at least a second point B in common. '
point B' so that the segment AB (or BA) is congruent to the
1.7. Upon every straight line there exist at least two points, in every segment A'B'. We indicate this relation by writing AB A'B'. =
plane at least three points not lying in the same straight line, Every segment is congruent to itself; that is, we always have
and in space there exist at least four p.oints not lying in a .plane. AB =
AB.
rV,2. If a segment AB is congruent to the segment A'B' and also to
the· segment A"B". then the segment A'B' is congruent to the
GROUP II. AXIOMS OF ORDER segment A"B"; that is, if AB = A'B' and AB = A"B", then
A'B' == A"8".
II, 1. If A. B, C are points of a straight Hne and B lies between
IV, 3. Let AB and BC be two segments of a straight line a which
A and C. then B lies also between C and A.
have no points in common aside from the point B, and, further-
III 2, If A and C are two points of a straight line, then there exists
more,let A' B' and B'C' be two segments of the same or of another
at least one point B lying between A and C and at least one point
straight line a' having, likewise. no point other (han S' in common.
D so situated that C lies between A and D.
Then, if AB '" A'B', and BC '" B'C', we have AC '" A'C'.
II, 3. Of any three points situated on a straight line, there is always
IV,4. Let an angle (h, Ie) be given in the plane a and let a straight
one and only one which lies between the other two.
line a' be given in a pJane a', Suppose also that, in the plane 0:',
II,4. Any four points A, B, C, D of a straight line can always be so
a definite side of the straight line a' be assigned. Denote by It'
arranged that B shall lie between A and C and also between
a half-ray of the straight line a' emanating from a point 0' of
A and D, and furthermore, that C shall lie between A and D
t~is line. Then in the plane 0:' there is one and only one half-ray
and also between Band D.
/c such that the angle (h, k) or (/(, h) is congruent to the angle
n, 5. Let A. B, C b.e three points not lying in the same straight line
(11', "') and at the same time all interior point-s of the angle
al~d let a be a straight line lying in the plane ABC and not
(h',k') lie upon the given side of d. We express this relation by
·passing through any of the points A. B, C. Then. if the straight
line a passes through a point of the segment AB, it will also ...,'
means of the notation L(h, k) =
L(h', /c'). Every angle is congruent
i. to itself; that is
pass through either a point of the segment Be or a point of ,the
segment AC. L (h, k) '" L (h, k), I
i
L(h,I,) = L(k, h).
GROUP III. AXIOM OF PARALLELS IV,5. If the angle (/t, Ie) is congruent to the angle (h', Ie') and to the angle
(It", Ie:'), then the angle (It', Ie') is congruent to the angle (It", Ie");
III. In a plane a there can be drawn through any point A, lying that IS to say, if L (h, Ie) = L (It', Ie') and L (It, Ie) =. L (It", Ie") then
outside of a straight line a, one and only one straight line L (It', Ie') '" L (h", Ie"). .
344 APPENDIX 1 APPENDIX 2
IV. 6. If, in the two triangles ABC and A'B'C, the congruences
AB", A'B', AC '" A'C', LBAC '" LB'A'C' hold then the con-
I BIRKHOFF'S POSTULATES
gruences LABC ;;;; LA'B'C' and LACB '" LA'C'B' also h~ld.
1
V.
GROUP V. AXIOM OF CONTINUITY
Let A 1 be any point upon a straight line between the arbitrarily

I
chosen points A and B. Take the points A21 A l • A 4 •••. so that
At lies between A and A 2 • Al between At and A31 A3 between
A2 and A 4 • etc. Moreover, let the segments AA I' A 1 A 2' A 2 A 3.
A3 A 4 •· " be equal to one another. Then, among this series of
points. there .-always exists a certain point An such that B lies

between A and All'
These axioms are reprinted from O. D. Birkhol-T, "A Set of

Postulates for Plane Geometry (based on scale and protractor)," Annals
olM mliematics, Vol. 33, 1932, by permission of the AUlw/s of M olitematics.
1. Postulate of line measure: The points A. B •... of any tine 1 can be

put into (1, 1) correspondence with the real numbers x so that
Ix. - xAI = d(A. B) for all points A, B.
II. Point-line postulate: One and only one straight line 1 contains two
given points P, Q (P '" Q).
Ill. Postulate of angle measure: The half-lines I, m, . .. through any point
o can be put into (1.1) correspondence with the re(~l numbers a
345
(mod 2,,) so that, if A i' 0 and B i' 0 are points on I and In,
respectively, the difference Gill - 0, (mod 2n:) is LAOB. Further- POSTULATES FROM HIGH SCHOOL
more, if the point B on m varies continuously in a line ,. not
containing the vertex O. the number am varies continuously also.
GEOMETRY
IV. Postulate of similarity: If in two triangles, 6ABC, 6A'B'C'. and
for some constant k > 0, d(A', B') = kd(A, B), d(A', C') = kd(A, C)
and also LB'A'C' = ±LBAC, then also d(B'.C')=kd(B,C),
LC'B'A' = ± LCBA, and LA'C'B' = ±LACB.
The postulates in Appendix 3 are reprinted from H. R. Pearson

and J. R. Smart, Geometry. Boston: Ginn and Company, 1971, by
permIssIOn of Glfln and Company.
1. Space exists and contains at least two distinct points.

2. If two points are distinct, then there is exactly one line which
contains them.
3. Every line is a set of points and contains at least two distinct
points.
4. No line contains all of the points of space.
5. If a point is in a line and another point is not in that line,
~ then the two points are distinct.
347
.!i.
348 APPENDIX 3
POSTULATES FROM HIGH SCHOOL GEOMETRY 349
6. If three points are distinct and noncollinear, then there is exactly

one plane which contains them. 19. If I is the interior of L ABC. R is the set of interior points
~ ~
7. Every plane is a set of points and contains at least three distinct of all rays between BA and Be. H is the set ABle n BCfA. and
noncollinear points. S is the set of interior points of segments which join a point in
8. No plane contains all of the points of space. Be.
SA and a point in then I = R = H = s.
9. If two distinct planes intersect, then their intersection is a line. 20. If for a one-to-one mapping of the vertices of one triangle pnto
10. H two distinct 'points of a line lie in a plane, then every point the vertices of another (not necessarily distinct from the first), two
of the line lies in the plane. sides and the included angle of one are congruent to the correspond-
11. There exists a correspondence which associates the number One ing two sides and the included angle of the other triangle, then'
witl~ ,an arbitrarily chosen, pair of distinct points and a unique the two triangles are congruent. .
posilive real nurnber with every other pair of distinct points. 21. If for a one-to-one mapping of the vertices of one triangle onto.
12. There is a one-to-one mapping of the real numbers onto the t he vertices of another (not necessarily distinct from the first). two
points in a line such thal 0 and I are mapped onto the points angles and the included side 'of one triangle are congruent to thf!
o and ~. r~spectiv~ly. ~he measure of the distance between any corresponding two angles and the included side of the other, then
two POints III the hne IS the absolute value of the dift'erence of the two triangles are congruent.
their corresponding numbers. 22. If for a one-to-one mapping of lhe vertices of one trinngle Ol1to
13. If P and Q are distinct points in line Land IJ and q are distinct the vertices of another (not ·necessarily distinct from the first), the
real numbers, then there is a unique coordinate system for L three sides of one triangle are congruent to the corresponding three
which assigns P the coordinate p, and Q the coordinate q. sides of the other, then the two triangles are congruent.
14. Any line in a plane separates the points of the plane which are not 23. If a point is not in a given line, then there is no more than one
points of the line inlo two sets such that line containing the point and paraliel to the given line.
t. Each of the two sets is a convex set, and 24. There exists a correspondence which associates the number one
2. Every.segment which joins a point of one set to a point of the with an arbitrarily chosen polygonal region and a unique positive
other Intersects the line. . . real number with every polygonal region.
15. ~ny plane separates the pOints of space which are not in the plane 25. If the polygonal region R is the union of two polygonal regions
1I1tO two sets such that R t and R 2 , whose interiors do not intersect., then relative to a given
1. Each of the two sets is a convex set, and unit area, the measure of the area of R is the sum of the
2 Every.segment which joins a point of one set to a point of the measures of the areas of R t and R z•
other Intersects the plane. 26. The measure of the area of a square region is the square of the
16. There exists a 'correspondence which associates with each angle in measure of the length of its side.
space exactly one real number IT such that 0 < Ii < 180. 27. If two triangles are congruent, then the triangular regions bounded
17. For every. point 0 and every closed half-plane whQse edge contains by the triangles have the same area.
0, there 15 a one-to-one mapping of the real numbers n, where 28. There exists a correspondence which associates the number one
o a. II ~ 180, onto the set of all rays in the closed half-plane with an arbitrarily chosen geometric solid and a unique positive real
haVing 0 as their endpoint. number with every other geometric solid.
18. F?r any angJe. ABC there exists exactly one ray-coordinate system 29. If the polyhedral solid S is the union of two polyhedral solids
WI~~S th-e zero-ray stich that for any point X in the C-side 51 and S2 whose interiors do not intersect, then the measure of
of SA, DX corresponds to a real number 11, 0 < 11 < 180. the volume of S lS the sum of the meaSures of the volumes of
5{ and 52'
30. If a polyhedral solid whose boundary is a rectangular parallel-

opiped has B. h. and V as the measures of the area of its base, EXAMPLES OF NOTATION USED
its altitude, and its volume, respectively, then V = BI1.
31. Given two geometric solids S I and S2 and a plane R, if every
IN TEXT
plane parallel to R which intersects either SI or S, ~Iso intersects
the other. and if the intersections are regions with equal area, then
the volumes of S 1 and S 2 are equal.
32. If two polyhedral solids are congruent, they have equal volumes.
Addition of coordinates of points X+A

Angle LAOA'
Closed neighborhood N[P, r]
Congruence lY-;;,/X
Distance between two points AB
Element and image under mapping x-+x+2
Element of E
Harmonic set of points H(AB, CD)
j
Ideal point in hyperbolic geometry n
.'
-..; Intersection n
Inverse transformation I-I
'"
Line As
.~ 351
..
.~:
352 APPENDIX 4
Measure of angle /ilL ABC

Open neighborhood N(P,r) ANSWERS TO SELECTED
Permutation symbol
Perspectivity
G~ *D
ABCD A'S'C'D'
EXERCISES
Projectivity ARCD 7\ A'R'C'D'

Proper or improper subset K' £; K
Ray AD
Reflection R,
Rotation R(O, ,,)
Segment PP'
Segment without endpoints PP'
Segment without one endpoint J5P' or PP""
Set-builder notation (x, y) I x > 3, x, YE R)
Similar to 6ABC ~ 6DEF
Subset c
Ultra-ideal point 1
Vector AA'
EXERCISES l.l
I. a. }1 + 4 2 = 52; b. 30 1. ~7,787 miles 3. False 4. True
S. False 6. True 7. True 8. False
EXERCISES 1.2
8, II,5 9, I, 1 and I, 2 10, II, 1-4 II. II 12, 12 13, 13
14, 14 and 15 15, 20,21,22 16. 24,25,26,27 17, Yes 18, Yes
19. Yes 20. Yes
EXERCISES [.3
2. (I) There exist exactly three distinct books in the system. (2) Two distinct
353
354 ANSWERS TO SELECTED EXERCISES ANSWERS TO SELECTED EXERCISES 355
books are in exaclly one library. (3) Not all the books in the system are in Ihe 4. {(a.i).(c,JJ.(e.kj) 5. Not defined
same library. (4) Two distinct libraries have at least one book in common. 6. {(b. 0). (d, c). (h, e)l 7. {(i. b). (), d). (/<, II)}
3. (1) There exist exactly three different students. (2) Two different students are 8. (x,y)~ (x _ 3.y - I) 9. (x.y)~ (x - 3.y - I)
on exactly one committee. (3) Not all the students are on the same committee. 10. (x, y) ~ (x + 5.}, - 2) II. (x. y) ~ (x - 2, y + 3)
(4) Two different c~mmittees have at least one student in common.
4. NOlle S. Two 6. No.7. No S. No EXERCISES 2.2
3. RI 5. R(240) 6. R(120) 7. No
EXERCISES 1.4
8.
1. (1) There exist exactly three distinct lines in the geometry. (2) Two distinct R, I
lines are on exactly one point., (3) Not all the lines of the geometry are on Ihe
same point. (4) Two distinct points are on at least one lina R, I R,
3. None
4. (I) The total number of committees is four. (2) Each pair of committees has I R, I
exactly one student in commOn. (3) Each student is on exactly two committees.
5. No 6. Four 7. None 9. Axiom 2
10. R I • 1; I 13. One subgr?up with eight members; three subgroups with
10. (l) The total number of trees is four. (2) Each pair of trees has exactly
four members; five subgroups with two members; and one subgroup with one
one row in common. (3) Each row contains exactly two trees. member.
II. No 12. One
EXERCISES 2.4
EXERCISES 1.5
I. a. (3, -5); b. (8,2); c. (2, -13); d. (I, -13/2)
l. (l) There exists at least olle library. (2) Every library has exactly three books
2. a. L-5, -~; b. (0,0) 3. (-)2/2,7)2/2)
in it. (3) Not all books are in the same library. (4) For two different books,
4. (../2/2,3../2/2) 5. x - " ~ (x' - h)cos. + (y' - k)s;n •.
6. C+ ;.fi. 3.fi - 7)
there is exactly one library containing both of them.. (5) Each two libraries have
at Jeast one book in common. .y - k~ -(x' - h)sina + (y' - k)cos.
2. l, 3, 4 3. 4, 5, 6; 4, 5, 3; 4, 5, 2; 4, 5, 7; 4, 1, 2; 4, I, 3; 4, I, 6; 2
4, 1,7; 4.2,3; 4,2,7; 4,6,3; 4,6,7
5,2,7; 4,7,3 and 5,6,3
4.4,5, I and 7, 6, 1; 4,6,2 and 7. (17,/32 -. 3. 3,/32+ 3) 8.(-5,10) 9.x=x·+5; y~y'-2
10. x = x'cos30 + y'sin30 y = -x'sin30 + y'cos300
D
6. «() There exists at least olle row. (2) Every row has exactly three trees in.' it. D
;
D
(3) Not all trees are in the same row. (4) There exists exactly one row containing 11. x: = x'; y = -y' 12. x = x'cos60° + y'sin60";
a tree not on a row that contains no tree of the: given row. (5) If a tree is 110t y = x' sin 60" - )I' cos 60"
on a row, there exists exactly one differenf'tree in the row such that the thlo
trees do not have a row in COmmon. (6) With the exception in Axiom 5, exactly
one row contains each pair of distinct trees. EXERCISES 2.5
7. 1,3,4 8. Two
6. The result is that a and b in the general equations for the translation are
~ero, .so the. translation is the identity. 7. Only the center of rotation is an
EXERCISES [.6 mvanant pomt. 9. Let the lines be y = 0 and ]I = ntx.
2. a. A; b. B' 3. a. A~; b. Pi 4. BRB' and ASA' 5. 1. 3. 4. 5
6. lif. fiP.1fR 10. Four n. ll. b. and d 12. a. 21; b. 3l; d. 57 EXERCISES 2.6
15. Four 16. Ten, five 17. None 18. Two 19. Four, four -~
20. One 21. Twelve, six 22. One I. •. (-2,1,7); b. (0,6,10); c. (-4,-2,2); d. (-2,-3,8)
2. a. (1,3, -4); b. 13,2, -I); 4. (-3.j2i2,4, 7.ji/2)
5. (3,8, -I) 6. (-2, -4,3) 7. (-2,3,9) 8. (7,4, -2)
EXERCISES 2.1
ll. a.. RefIecti~n, rOlato,: reflection, or glide refIectionj b. Reflection, rotatory
2. c 3. a. (7, -2) b. (5, -3) c. (2, -5) refiectJOn, or glide reflecuon J2. x' = x, y' = ]I. =' = 4_ :
ANSWERS 1 U :,tLt; ..... ' eu '-,~L., ...... , .... ~~
356 ANSWERS TO SELECTED EXERCISES
EXERCISES 2.7 EXERCISES 3.5

1. 4 inches 2. 6/5 3.' Length and area 4. The measure of area is 1. u. Circular region; b. Triangular region; c. Entire plnne; e. Parabolic region;
multiplied by the square of the ratio of similarity. 5. .fii/6 f. Entire plane; g. Angle and interior; h. Triangular region or qundrangtllar
6. They are equal. 7. (9/4, 15/4) 8. (28,30) 10. Yes region 2. Packing a carton of books; gelling a large, irregularly shaped
package lhrough a door 4. ,a. Yes; b. Yes 6. One bra.nch or a hyperb~la
and the point of intersection of the two asymptotes S. a. The angle, Rnd Its
EXERCISES 3.1 interior; c. A spherical region; d, A tetrahedral region or olher speCial wIses
1. b, c, d, e. r. g 2. a, b 4. Yes; no 8. The open one~dimensjonal 9. No
neighborhood wilh radius,. of a point P is the set of interior points of a segment
of length 2,- with P as midpoint. 9. e. The empty set. the set itself, the
complement of the sel. f. The empty set, the entire plane. the empty set. EXERCISES 3.6
10. c. The empty set, {(x.y,O):x G: yl. the complement of the boundary_ d. The 1. 1 inch1...j2 inches 2. 2 inche~ fo inch~ 3. I inch. 1 inch
empty set. the set itself, the complement of Ihe set. 11. a. Closed; open; 4. 3ji87/14,7 9.4",8(" - )3) 10. ",3
open; closed; closed; neither b. Closed; closed; neither; closed c. Neither
12. a. a. d; b. a, b 15. It is neither closed nor open. 16. a. The interior
of a circular region. b. A parabola and its interior. EXERCISES 3.7
6. Yes 7. Let S = {Pl ..... ?"} be any finite set of points in a plane. Then
EXERCISES 3.2 there is a point A such that every closed half-plane formed by a plane through:
1. No 2. Yes 3. Yes 4. No 5. Allregular 6. Vertices are corner

A contains at ieasln/3 points of S:
points. 7. Vertex is corner point. S. Verlices are corner points. 9. a. A
triangular region is tbe intersection of the three supporting half-planes determined EXERCISES 4.1
by the sides of the triangle. b. A convex polygonal region is the intt:rsection
of the 11 supporting half-planes determined by the sides of the polygon. 1. ABC. H; ACH, B; AHB. C; CHB. A i.
At the verlex of lhe right angle
11. If a line contains at least one boundary point of the set but no intedor 3. For an obtuse triangle 4. Three 5. Yes
points, it is a supporting line for a convex set of points. If a line lS not a
supporting line for a convex. set of points (but does contain at least one
boundary point), then it contains interior points. U a line contains at least one EXERCISES 4.2
boundary point of the set and also contains interior points. then the line is not a AC BD FE CA BF DE AE CF BD
2 _·_·_~_I 7.-·-·-~
supporting line for a convex set of points. . 1. CB' DF' EA ~ -1 . AD FD EC EC FB DA
12. An altitude
EXERCISES 3.3
1. Yes 2. No, not bounded 3. No, not bounded 4. No, not convex
EXERCISES 4.3
5. No, not bOlll1ded 6. No, not always bounded 7. No, not bounded
8. Yes 12. Line segment 14. Three of the first and siK of the second 1. fl. One; b. One; c. An infinite number
15. Two hundred Black Angus and one hundred Hereford
EXERCISES 4.4
EXERCISES 3.4 6. 3/4; 3/5; 4/5
1. A median 3. At the centroid 4. No 5. Shorter
1. Door handle and door, book on n shelf 2. It fills an octant in space. with
the vertex at the origin and three sides lying on the reference planes. 3. No. i
H must pass through a boundary point of the convex set. 4. Yes 5. No i EXERCISES 4.5
6. No 7. Yes S. Yes 10. A tetrahedral solid IS the intersection of the I 9. The four points are
four closed half-spaces containing the sides of the tetrahedron and the vertex not 1. Yes 8. The four points are vertices of a square.
on the side. vertices of a square.
EXERCISES 4.6 6. (27/13,18/13) 8. 2x' + 2y' - 9.< ~ 0 9. 5x' + 5y' - 9x ~ 0

I. 1.6180 2. 55/34 '" 1.618 3. I/x' or 5 - 3x 6. The interior angles 10. 2x'+ 2y' + 9x + 9y ~ 0 II. 36x' + 36xy' - 81y'~ 0
2 - 5;
have measures of 135, and this number is not a factor of 360. 12. 9y' + 9x'y - 8Ix' ~ 0 13. x' + y' ~ 81/16 15. ~
16 3 + 71 4 - 8i 27 + 991
EXERCISES 5.2 . 58 17. - 5 - 18. 130
12. Ie - ,,)lfo - eh) - 10 - .)(I>e - ad)
Ie - a)lll - fl - Ig - e)(d - b)
EXERCISES 6.4
1.1 - 1» [Ie - o)lfg - eh) - 10 - e)lbe - ad)] be - ad
--
Ie - a) Ie - a)(h - J) - (g - e)(d b)
+ --
e- 0
1. The other point would be the ideal point. 3. A tangent to a circle passes
through the center of a circle orthogonal to the circle at the point of tangency.
4. A circle circumscribed abollt a right triangle has the hypotenuse as a diameter.
EXERCISES 5.3 13. The same plane
3. There are 4, 3, 2, 1, or no solutions. 4. 1 solution S. 1 or no solution

6. 1 or no solution 7. 1 or no solution 8. 2, t, or no solutions EXERCISES 7.1
9. 4 solutions 10. 4, 3, 2, I. or no solutions 11. 2, I, or no solutions
1. A'B' = ABcosO 2. a, Top center; b. Off to the left 4. Not invariant
12. I solution 13. 2, I, or no solutions 14. 1 solution 15. ( or no
S. Invariant 6. Not invariant 7. Not invariant 8. Not invariant
solution 16. or no solution 17. [solution IS. 1 or no solution
9. Not invariant 10. Invariant 11. Not invariant
19. I or no solution

1. x is a solution of .\A - 4Xl + 2 = O. 2. Yes, it is a solution of ."(3 - 2 = O. I. There are fewer invariant properties. 2. I, 2, 3, 5, 7 3. I, 3 4. I, 3
3. ",2.5198 5. 111e only possibilities are ±I, ±I/2, ±I/4, ±I/B, and each of 5.1,2 6. [,2,3,4,7 7.1,2 8. I, 2 9. 1,2 II. Two 12. Two
these can be checked by synthetic division to show it is not a solution. S. None 13. Two of them would be ideal points. 14. Yes 15. One
9. This makes the cosine of an acute angle equal to 1. 10. The equation has
no rational solution.
EXERCISES 7.3
1. If A and 8 nre distinct points of a plane, there is at least one line on both
EXERCISES 6.1
points. 2. On a line are at least fOllr points. 3. A triangle (trilateral)
I. 16/3 2. 3 3. 4/3 4. 1/3 5. 8/49. 6. fi/3 7. 189/50 consists of three nonconcurrent lines and the points of intersection of each pair.
8. No 9. Outside the circle of inversion 10. When the circle does not pass 4. Three points not on the same line determine a plane.
> 5. A point is determined
through the center of inversion; when the circle does pass through the center. by two illtersecting lines.
of inversion. 12. The identity 13. Two intersecting circles, two intersecting
lines, an intersecting line and circle 14. A circle inside and concentric to the
circle of inversion 15. On the circle of inversion EXERCISES 7.4·
3. HIXC. OY); H(CX. YO); H(Xe. YO) 4. A point at infinity 5. Yes
EXERCISES 6.2 6. Yes 7. Yes 8. Yes 9. Yes
1. Two orthogonal circles, a circle and a line through the center 10. A line nol
passing through the point EXERCISES 7.6
l. The last coordinate is zero. 2. a. 13, 8, .1); b. 12, 3/4, I); c. (I, -4, 1)
EXERCISES 6.3 3. a.12, 5) b.(-I/2,3/2); c.(2/3, -5/3) 4.15/3,2/3,1); (10,4,6); (20,8,12)
5. No 6. (I, -I, 0) . 7. 3x, + x, -
2x, ~ 0 8. 3X; + X, - 2X, ~ 0
1. 13/25,4/25) 1 (2, 2) 3. 10, 3) 4. 156/25, 8/25) 5. (16/17, -4, 17) 9. ", - x, ~ 0 10. iI.O, -I) 11. x, - x, ~ 0 12. 10.2. 1)
EXERCISES 7.7 to a circular region. 9. More than four countries with" common point would
I
1.-2,3,1 2. (5, -4) 3. (-4, -4) 4. (I, I. -I); (I, I, I); (I, 1,0); each require a separate color.
(4,4, -I) 5. 5/3 7. Similarity 8. 1'!ooe 9. Affine 10. Motion
EXERCISES 8.5
EXERCISES 7.8
I. By TOWS: 6, 12; 20, 30; 12, 20, 30 5. 0 6. 2; 0; -2; -4 9. Two
1. The product of a reflection and the same reflection is the identity. interlocking pieces 10. No. Two interlocking pieces 11. No. Two illter-
2. 3. Rotation of 180°; b. Rotation of 120°; c. Rotation of 90° 5. A unique locking pieces 12. Three 13. Four
one-dimensional involution is determined by two pairs of corresponding points.
6. A two-dimensional projectivity that leaves the fouf vertices of a complete
quadrangle invariant is the identity transfOimation. EXERCISES 9.1
3. If a straight line intersects one of two parallel lines, it witl 110.t always
EXERCISES 7.9 intersect the other. 4. Straight Jines parallel to the same straight line' are not
always parallel to each other, 5. There does not exist one triangle for which
1. It could be a degenerate conic. 2. Figures 7.5, 7.19, 7.21, 7.22, 7.23, 7.25 the stun of the measllres of the angles is 7t radians. 6. ll1ere does 110t exist a
3. The lines of the ranges of points in the projectivily defining the line conic pair or similar bUI noncongruent triangles. 7. 111ere does not exist a pair of
are also lines of the coniG. 4. The points corresponding to the common point straight lines the same distance apart at every point. 8. It is not always possible
of the two ranges of points determining a line conic are the points of contact of to pass a circle through three noncollinear points. 9. a
the two ranges of points. S. 3x 1 + x 2 + Xl = 0 7. Yes 9. There are 10. AGJ; AIL; AKM 12. Two
720 permutations of six things taken six at a time. Half of these are duplicates
caused by reversing the. order of points. Of the remaining 360, there are six
different names fo. each of 60 hexagons. EXERCISES 9.2
1. Otherwise, two distinct lines could be both parallel and intersecting.
EXERCISES 8.1
1. A rotation; x -+ cotangent X 2. TIle inverse also has an inverse that is a EXERCISES 9.3
con~inuous transformation. 4. Length of a segment; area of a region; number 1. Two 2. The sum of the measures of the angles would be 2n. 4. Adjacent
of Sides of a polygon 5. A convex set is a connected set ror which the curve side S. Summit 7. TIle defect of the original trjungle is equal to the sum
joining two points can always be a segment. 6.. a, b, c, d, r. h 7. a. C, d, e of the defects of the two smaller triangles.
8. a. Penny; b. Washer; c. Button 10. a
EXERCISES 9.4
EXERCISES 8.2
1. There would exist a rectangle. 2, At their common perpendicular 3. An
l. a, d 2. a, b, c 3. Vertex C is not counted. Vertex E is counted. infinite number 4. No 9. The Slim of the measures of the summit angles
4. Odd S. Simply connected 8. Yes is equal to the sum or the measures of lhe three angles of the original lriungies.

1. No. Images would not always be in the circular region. 3. No 4. No 1. Less than 2. The angles determined by the radii and the segment joining
5. Yes. 6. No 7. Yes 8. 5,7 9. 5 10. None the points are congruent. 4. No S. No 8. Connect the midpoints of two
of the segments joining pairs of points.
EXERCISES 8.4
2. Zero 3. Zero: prism; One: teacup; Two: dining room chair with arms
EXERCISES 9.6
S. Four; foqr 8. The theorems only apply to countries topologically equivalent t. Four-line geometry;PG (2. 3) 2. Yes 3. Circle 4. 8 S. When they
f
1
362 ANSWERS TO SELECTED EXERCISES
are the two poles of a line. 6. No, there are no ideal points. 7. Less than 21t
9. At the polar of their common points BIBLIOGRAPHY
EXERCISES 9.7
9. Infinite 10. Atjf is the image of circle A. B. ire
is the image of circle
H, C. Circle A', C' is the image of circle A. C. Each of the three images is
orthogonal to the image of the fundamental circle.
Adler, Claire F., Modern Geometry, Second Edition. New York: McGraw-Hili, 1967.
Barry, Edward H., Introduction to GeometJ';cal rrl/ns!ormatiolJs. Boston: Prindle,
Weber & Schmidt, 1966.
Benson, Russell V., Euclidean Geometry and Convexity. New York: McGraw-Hili,
1966.
Blumentha~ Leonard M., A Modem View of Geometry. San Francisco: W. H.
Freeman, 1961.
Bullard, Sir Edward. 111e Origin of tile Oceans. Scienrific American, September 1969.
Vol. 221 (3),
Chinitz, Wallace, Rotary Engines. Scientific American. February 1969. Vol. 220(2).
Chrestenson, H. E.. MappirJgs of the Phme. San Francisco: W. H. Freeman. 1966.
Courant. Richard. and Robbins, Herbert. What is Mathematics? London: Oxford
University Press, 1951.
363
BIBLIOGRAPHY 365
364 BIBLIOGRAPHY
Wolfe, Harold E., JlIlroductiol1 ro NOII-Euclidea/l Geometry. New York: Dryden

Coxeter, H. S. M., IfOlroducfiolllO Geometl'Y. New York: Wiley, 1961.
Coxeter, H. S. M., Projectiue Geometry. New York: Blaisdell, 1964. Press, t 945.
Wylie, C. R.,"FoulldationsofGeollletry. New York: McGruw-HiII, 1964.
Davis, David R., Modern College Geometry. Reading, Mass.: Addison-Wesley, 1954.
Yaglonl, 1. M., and Boltyanskil, V. G., COIfllex Figures. New York: Holt,
. Dorwart, Harold L., The Geometry 0/ Incidence. Englewood Cliffs, N.J.: Prentice-
Hal, [966. Rinehan & Winston, 1961.
Eggleston, H. G., Conuexity. Cambridge: Cambridge University Press, 1966.
Escher, M. C, Tlte Gmp/tic Work of M. C. Escher. New York: Ballantine, 1971.
Eves, Howard, A Suruey o/Geometry, VoU. Boston: Allyn and Bacon, 1963.
Fishback, W. T., P"ojecliue anci Euclidean Geometry. New York: Wiley, 1962.
Gans, David, Trans/ormations and Geometries. New York: Appleton-Century-Croft,
[969. I
Graustein, William c., Introduction to Higher Geometry. New York: Macmillan,
[930.
Hilbert, David, The Foundations o/Geometry. La SaUe, III., Open Court, 1950.
Johnson, Donovan A., Paper Folding fin' the M (ftiJematics Class. National Council
of Teachers of Mathematics, 1957.
Johnson, Donovan A., and Glenn, W. H., Topology, rhe Rubber Sheet Geometl·Y.
New York: McGraw-Hili, 1960.
Klee, Victor L. {Ed.}, Convexity. Proceedings of the Seventh Symposium in
Pure Mathematics of the American Mathematical Society, Providence, R. L,
[963.
Levi, Howard, Topics in Geometry. Boston: Prindle, Weber & Schmidt, 1968.
Levy, Lawrence S., Geometry: MOliern MlIll1e17wlics via tile Euclideal1 Plane.
Boston: P"dndle, Weber & Schmidt, 1970.
Lyusternik, L., Convex Figures and Polyhedra. New York: Dover, 1963.
Meserve, Druce E., "Fundamental Concepts of Geometry. Reading, Mass.: Addison-
Wesley, 1955.
Meserve, Bruce E.,. and luo, Joseph A., Fundamentals of Geometry. Reading,
Mass.: Addison-Wesley. 1969.
Moise, Edwin E., Elemel1!QI'Y Geometry /rom all Advanced Standpoint. Reading,
Mass.: Addisoh-Wesley, 1963.
Patterson, E. M., Topology. New York: Interscience, 1959.
Pearson, H. R., and Smart, 1. R., Geometry. Boston: Ginn, 1971.
Pedersen, Jean J., Dressing Up Mathematics. The Matltemacics Teacher, Febn.iary
[968, Vol. 61(2). .
Perfect, Hazel, Topics in Geometry. London: Pergamon Press, 1963.
Pervin, W. J., Fot/It.-fations a/General Topology. New. York: A~demic Press, 1964.
Pratt, M. M., Finite Geometries. Unpublished Master's Thesis, San Jose State
College, San Jose, California, 1964.
Prenowitz, Walter, and Swain, Henry, Congruence and Motion in Geometry.
Boston: Heath, 1966.
Rademacher, H., and Toeptitz, 0., The Enjoyment of Mathematics. Princeton:
Princeton University Press, 1957.
Tuller, Aunita, A Model'll Intl'oliuCliol1 co Geometries. Princeton: Van Nostrand,
[967.
Valentine, F., COl/uex SefS. New York: McGraw-Hill, 1964.
INDEX INDEX 367
Closed curve, 83 Convex sel(s): (continued)

Closed neighbourhood, 84 definition ·of, 80
Closed set, 86 intersection of, 81
Closure property, 39, 62 Convexity:
Collapsing compass, 163-166 axiom on, 8
COllinear, 234 in postulational system, 81
Collineation, 268 relationship with topology, 278
Combinations, 15 symposium on, 82
Commutative group. 40 Corner point, 92, 105-106
Compass, 163-166, 192-195 Correlation. 268-269
Complement, 85 Corresponding points, 327
Complete, 9 Coxeter, H. S. M., 230
Complete quadrangle. 231, 240 Cross intersections, 249
Complete quadrilateral, 235-236 Cross joins, 247-248
Complex plane. 214-216 Cross ratio, 258-259
Abelian group, 40 Babylonian geometry. 1-2 Conchoid of Nicomedes, 183-184 Cube. 298
Absolute value, 46 Barbier's theorem. 119 Conclusion. 3 Cmve(s):
Adjacent trisectors. 152 Base, 316-317 Concurrence theorems, 126-128 closed, 83
Afline geometry, 232 Baseline. 329 Concurrent. 234 in hyperbOlic geometry. 326-331
Affine lransformations. 260 Beltrami. E.• 336 Congruence: meanmg of, 82-83
Algebraic numbers. 182-183 Biangle, 332 of limiting curves. 328 simple. 83 .
Altitudes, 127 Bicontinuous transformation. 278-279 of omega triangles. 314-315 Cyclic quadrilateral. 132
American Mathematical Society, 82 Birkhoff, G. D" 7-8, 345 of sets, 46 Datum, 175-176
Analysis, 168 Bisecting angle. 188 of triangles, 8, 319 Oa Vinci. Leonardo, 227
Analysis figure, 167 BJaschke¥Lebesque theorem, 119 Conics. 265-275 Decagon. 185-186
Analytic geometry, 4. 209-215 Blaschke's theorem, 124 Conjugate. 214 Deduction theorem, 4
Angle(s): Bonnesen. T., 82 Conjugate points. 269 Deductive proof. 4
in elliptic ¥eometry. 334 Boundary, 85 Connected set, 280-282 Defect, 318-319, 323
of plmlileilsm. 310 Boundary point, 85 Consistency. 9, 335-340 Defined terms. 3
of Iriangle, 318-319 Bounded set, 86--87 Constant of iuversion. 198 Degree of connectivity. 282
undefined term, 8 Boundless. 331 ! Constant width, 115-119 Desargues, G., 236
Antiparallel. 132 Brianchon. C. I., 270 ; Constructible numbers, 169-173 Oesargues' configuration, 239
Apollonius.130-131 Bl'ianchon point. 270-271 ! Construction(s}: Desargues' finite geometry. 26-28
Area: Brfanchon's theorem, 270-t7l, 274 . auxiliary, 175 Desargues' theorem. 236-237
axioms on. 8-9 Brocard. Henri, 150 in axioms of Euclid, 164-165 Descartes, R.• 298
in hyperbolic geometry, 323-325 Brocard circle, 151-152 with compass. 192-195 Diagonal lines. 235
Associative property, 39 Brocard points, 150-152 conics, 271-275 Diagonal points. 231
Alixiliary construction, 175 Brouwer, L, E. J., 289 I in Euclidean geometry. 163-195 Diagonal triangle. 231
Auxiliary triangle, 1/4
Axiom(s):
of Birkhoff, 7-8, 345-346
Brunn. R .• 82
Bussey. II
I
Brouwer fixed point theorem, 289-290 examples of problems, 174-179
with one instrument, 191-195
par.er folding, 187-190
Diagomli tdlaterul. 235
Diameter:
of earth, 2
of Euclid, 6 Caroms. 158-161 , plulosophyof. 163-169 orsel.114
existence. 15 Cartesian coordinates, 251-1252 step in construction problem, 168 Digon.332
Fano's geometry. 19 Center: . , with straightedge, 191 Direct motion, 61
of Forde!", 7 of gravity, 128 II Continuity. 82 Discussion. 168
four·fine geometry. 15 of homology, 249, 272 Continuous transformation, 278-279 Displacements. 49
fOufwpoint geometry. 17 of homothety, 72 Continuously deformed, 281 Distance:
geometry of Desargues, 27 of inversion. 198.219-220 Conlrapositive. 89 fonnula in analytic geometry, 46
geometry of Pappus, 23 of perspective collineatioh. 264 COnverse: in Poincare model, 338
high school geometry. 347-350 of perspectivity. 227-228, of Ceva's theorem. 137-138 preserved in Euclidean geometry. 46
of Hilberl. 7, 341-344 Centroid, 128. 142 : of Menelaus' theorem, 135-136 undefined term. 8
of Huntington. 7 Ceva, 136 ! of theorem. 56, 89 Dividers, 163-166
incidence, 15 Ceva's theorem, 136-137. lrt6-147 Convex, 79 Dodecahedron, 298
modern geometry texts, 8-9 Characteristic postulate, 30f7, 331 Convex. body: Doubling the cube, 180-181
part of Euclidean geometry, 3 Circle(s): definition of. 96 Duality, 16,233-236; 249
of Pasch, 1, 309. 313 of Apollonius, 130-131. 217 perimeter of. 100 Earth measure. 1-2
of Peano, 7 geometry of, 126-161 In three·space, 104-108 Egyptian geometry, 1-2
in hyperbolic geometry, 326
projective geometry, 229-233
three~point geometry. 12
of Veblen. 7
image under homothety, 73-74
of inversion. 198
nine~point, 141-143
j in two¥space, 96-103
Convex cover (see Convex hulls)
Convex hulls. 108-113
Einstein, Albert, 306
Elation, 264
Elements. II
Axis: . . . j Convex polygonal regions. 99-100 Elliptic geometry. 307, 331-340
of homology. 247, 273 . secants and tangents, l32-133
of perspective collineation. 264 Circumcenter, 126, 142 ; Convex quadlilateral, 319 Elliptic projectivity. 261
Convex. set(s): Equations:
of spreading, 291 Circumcircle, 127 connected sets, 280 for glide reflection, 56
366
j
.]
INDEX 369
368 INDEX
Hyperbolic projectivity, 261 Linear programming. 101-103
Equations: (wlllinl/eel) Fundamental theorem: Hypotheses, 3 Linkllges, 217-218
for inversion transformation, 211 of algebra. 291 Icosahedron, 298 Listing. J. B., 278
for motions of plane. 51-58 of projective geometry, 245-246 ldeal point, 199.232.312 Loci. 175 .
for .motions in three-space, 68 Fundamental triangle. 254 Iden tical sets, 94 Logically equivalent, 306
for projective transformations. 255-261 Gamma point. 321 Identity element. 40 Magnus. L. 1., 200
for reflection, 54-56 Gauss, Karl, ISS, 307 Identity transfonnation, 37 MAp-colouring problem. 295
for similarity transformations. 75 Genus. 294-295 Implication. 3, 89-90 Mapping. 34
Equidistant curve, 327. 329-330 Geometric topology. 271-303 Impossibility proofs. 180-187 MnschefQni, L., 192
Equilateral1riangle: Geometric transformation (see Incenter. 128 MHthematic."l.1 construction, .164.
symmetries of, 40-45 Transformation) Incidence. 230 Matrix. 259-261
tessellation, 157-158 Geometry: Incidence axioms, 15 Maximum value. lOt
Equivalence, 89 amne. 232 Incircle, 128 Maximum width, 114-115
Equivalent polygons, 323,-324 convexity. 79-123 Inconsistent system, 9 Measure of area, 325
Equivalent triangles, 323-325 definition of, 47 Independent sets, 9 Medians. 128
Eratosthenes. 2 demonstrative. 3 Index, 290 Menelaus, 134
Erlanger program. 229 elliptic, 331-340 Indirect argument. 13 Menelaus' theorem, 134 .
Escher, M. C., 50, 158 finite, 11-31 Interior poinl, 84-85 Midpoint of segment, 192-193·
Euclid, 3, 6, 306 history of. 4 Inll:mal angle bisectors, 128-129 . Miniature geometries (see Finile
Euclidean geometry. 3 inversion. 197-224 Intersection, 81 geometry) .
Euclidean group, 58-65 literal meaning of, I Into mapping, 34 Minkowski. H .• 82
Euler, C, liS, 142, 298 non-Euclidean. 305-340 Invarianl{s): Miquel's theorem. 143-144
Euler characteristic, 301 polygon and circle, 125-161 group of motions, 62-65 Model:
Euler line, 142 projective, 225-275 inversion. 204-206 for elliptic geometry, 336
Euler's formula. 298-300 Gergonne. J. D., 138 meaning of, 36 for hyperbolic geometry, 332-333
Even parity, 285-286 Gergonne point. 138 molions in three-space, 69 Modern geometry, 4-5
Even vertex, 292 Glide reflection. 50, 56. 68-69 topology, 289-291. 294--295. 300-302 Moebius, A. F .• 278. 296, 301
Ellcircies, 130 Golden ratio. 154-157, 186 Inverse, 89 Moebius strip, 301-303
Existence axiom. 15 Golden rectangle, 155-156 _Inverse element. 59 Mohr, C., 192
Extension fields, 171-173 Golden spiral, 156 In'verse points. 193, 197-198,206--207 Mohr-M~scheroni constructions, 192-195
Exterior angle. 313-314 Great circle. 333 Inverse transformation, 38, 57 Morley, Frank. 152
Exterior point, 85 Greek geometry, 2-3 Inversion: Morley's theorem. 152-154
External bisectors, 130 Greek problems, 180-183 analytic geometry of, 209-215 Motion(s):
Extreme and mean ralio. 154 Group: applications of, 216--224 definition. 46
Extreme point. 109 Abelian, 40 in complex plane. 214-216 direct, 61
Extremum problems, 160-162 bicontinuous transformations, 278-279 equations for transformation. 211 equations in three-space, 68
Fagnano, J. F., 161 Euclidean, 58-65 geometry of, 197-224 matrix ror. 260
Fagnano's problem, 161 motions in three-space, 68··69 in three dimensions, 222-224 opposite, 61
Fano, II projective transformations, 245 Inversive plane. 199 rigid,49
Fano's geometry, 19-22 similarities in three-space. 75 Inverting theorems, 218-220 sets ofequatious for. 51-58
Fenchel. W., 82 similarity transformations, 71 Invohltion, 262-263 of three-space, "66-70
Feuerbach, K. W., 141 transforma tions, 39-45 Involutory transformation, 203 types of, 47-50
Feuerbach's theorem, 221-222 Hal~-plane, 81 Is~)gonal conjugates, 145-149 Multiply connected, 280-281
Fibonacci numbers, 155 Hammer, 116 Isometric sets, 46 Mystic hexagram. 270
Fifth postulate, 6, 305-306 Harmonic conjugfHe, 241 Isometry, 46. 63 Neighborhood. 83-84, 278-279
Finite geometry: Harmonic homology, 264 Isomorphic. 62 Network, 291-293
of Desargues, 26--28 Harmonic net, 244, 254 10·rdan curve theorem, 285-286, 293--294 Nicomedes, 183
of Fano. 19-22 Harmonic pencil, 242 hm~'s theorem. 124 Nine-point circle, 141-143,221-222
four-line. 14-16 Harmonic properly, 242-24:1 Klein. Felix. 47, 229. 336 Nonconvex sels, 80
rour-point, \6-18 Harmonic range, 240 Klein bottle, 302 Non-Euclidean geometry, 305-340
history of, II Harmonic sets, 239-245 Klein model, 336 Noninlersecling lines, 310, 321-322
meaning of, II Harmonically related, 244 Knots, 302 N~tuply connected, 281
non-Euclidean, 308 Helly, Eduard, 120 Lambert, J. H.. 317 Oceanography. 291
of Pappus, 23-25 Helly's theorem. 120-124 Lambert quadrilateral, 317-318, 334 OCI~hedron, 298
projective geometries. 28-29 Heron's theorem. 159-160 Least tIpper bound, lOt Odd parity. 285-286
three-point. 11-14 Hexagon. 269 Left hand parallel. 310 Odd vertex, 292
of Young, 29 Hilbert. David, 7, 34 I Lemoine. E., 146 Omegu lriangle, 313-315
First Brocard point, 151 Homeomorphism, 279 Lemoine point, 147 On. 19
Forder. Henry, 7 Homogeneous coordinates. 251-255 Length of curve, 99 One-sided surface, 301-303
Foundations of mathel!1atics, 9 Homology, 264 Limiting curve, 327-329 One-to-one correspondence, 34
Four-line geometry, 14-16 Homothetic transfonnation, 202 Limiting position. 205 Onto mapping, 34
Four-point geometry, t6--18 Homothety.72-75 Lindemann, T .• 183 Open neighbourhood, 83-84
Function: Huntington, E. V.• 7 Line, 3, ll, 23 Open set, 86
continuous. 82 Hyperbolic geometry. 307-331 Line conic, 266 Opposite molion. 61
definition of, 35 Hyperbolic polarity, 269
-
370 INDEX INDEX 371
Opposite sides, 2J I Polarity. 268 Rotatory refieclion. 68-69 Tessellation, 157-158
Opposite Inlnsformalion. 206 Pole. 26-27, 208-209, 268. 332 Rubber sheet geometry, 277 Tetrahedron, 69, 298
Orthocenter. 127, 142 Pole or spreading, 291 Saccheri. Girolamo. 316 Theorems:
Orthocentric set, 128 Polygons. 323-324 Saccheri quadrila~ral, 316-317. 324, 333 inverting, 218-220
Orthogonal circles, 206 Polyhedron, 298 School ~athem<lhcs Study Group, 8 parI of Euclidean geometry, 3
Orthogonal curves, 118 Poncelet. J. -V". 228-229 Screw displacement, 68-69 Three ramous Greek problems 180-183
Orthogonal trnjectory.• 118,327 Poncelet-Steiner construction theorem, 191. Secants. 132-133 Three-p?,int ¥eometry, 11-14 •
Pulrs of lines. 320-323 Positive Brocard point, 15 I Self-dual. 28, 235 Topological··mvarianl 282 292
Paper folding, 187-190 Postulates (see Axioms) Self-inverse, 199 Topology, 277:":303..' •
Pappus, 23-25 Pre-Greek mathematics. 1-2 Semilangents, 91 Torus, 295
Parabola: Projection, 223. 225-226 Set; Tracing puzzles, 291-293
inverse of. 213-214 Projective geometry, 28-29, 225-275 bounded, 86---87 Transformation(s) :
flIngCllls to. 190 Projective plane, 232 ! closed, 86 bicontinuous, 278-279
Parabolic projectivity, 261 Projective transformation, 228, 255-261 of constant widlh, 115-119 continuous, 278-279
Parallel lines: Projectivities. 228. 245-250, 261-265 open, 86 definition of, 34
ill Fa no's geometry. 22 Proof: width of, 114-·t~O equations for plane motion. 56
firth postulate, 306 of implication, 3-4 Sides. 316 examples of, 33, 35-36
in four-point geometry. J7 by inversion. 220-222. 338-339 Similarity transfonnations 70-7-7 260 groups of, 39-45
in geometry of Desargues, 28 step in construction problem, 168 Simple closed curws, 97, i83-289 identity, 37
in geometry of Pappus, 24 Properties. 36 Simple closed polnon, 285 inverse, 38
under homothety. 73 Proportions. 129 S!mple closed surfa:e, 106, 294 of inversion 199-200
in hyperbolic geometry, 309. 320 . Pythagorean theorem. 5
Quadrangle. 231
Stmple curve. 83
Simple hexagon, 269
2:03
involutory,
by paper folding, 188-189 isometry, 46
Parity. 285-286 Quadrangular set, 240 Simple polyhedron, 298 matrices for 259-261
Partition of segment. 166-167 Quadrilateral: _ Simply connected, :'80-281 period of, 2en
Pascal. Blaise, 270 cyclic, 132 j Simson, Robert, 139, 200 product of, 36-37
Pascal line, 270 in hyperbolic geometry, 315-318 Simson line. 139 projective, 228 245
Pascal's mystic hexagrtlm, 270 inscribed in circle. 131-132 Single elliptic geometry, 332 similarity, 70-77
P<lscai's theorem. 269-210. 274 Radius of inversion. 198 Sobyzk, 116 Tr~llslation, 47, 52-53. 66;279 .
Pasch. Moritz, 7 Ranges of points, ~66 Space duality, 236 Tnangle(s) :
Pasch's lIx.iOO1. 7 Ratio: Species, 176 angles of, 318-319
Peano. Guiseppi, 7 golden. 154-157 Spiral: auxiliary. 174
Pendl of lines. 266 of similarity. 71 of Archimedes, 184-185
Rational numbers, 170 deti~ition for projective georiietry, 230
Pencil of rays, 327 golden. 156 eqUivalent, 323-325
Pentagon. 156-157 Rect~mgle. 155-156 Square. 157-158 geometry of. 126-161
Perimeter. 119 Reflection(s): Squaring the circle, 180--183 maximum area. 325
Period, 203. 262-263, equations for, 54-56 \ Steiner. Jacob. 266 perspeeHve frem a line 26
Permutation group symbols, 42-44 in plane, 49-50 , Ster70graphic projection. 223. 337 perspective from a poi~t 26
Perpelldicuia r. 189 product of. 59--60 j Stl<ughtedge, 163
properties of r~ght triangle. 1-2
Perpendicular bisectors, 126-127, 188 symmetries of triangle, 41 I Subgroup, 45 Reuleaux, 116--117
Perspective collineation. 264 in three·space, 67 ' Subset. 80 ~n three-point ~eometry, 14
Perspective from line. 26. 237 Region. 83 Summit, 316
Reguhll' decagon. 185-186 Tr~ang~dar decomposition, 299-300
Perspective rrol11 point. 26, 236-237 Summit angles, 317 Tr!secung the angle, 180-182
Perspeclivity, 226-227 Regular heptagon. 186-187 Superposition. 7
Regular hexllgolls, 157-158 TrISectors of angle. 152
Peucellier's cell. 218 Supporting half-plane, 93-94 Truth. 3
rhme: Regular pelliagon. 156-157 Supporting half·spaces, 107
of inversive geometry. 199 Regular point. 92. 105-106 Supporting line, 88-90 105 Ultra-ideal point. 321
'Regular polygon. 185--186 Undefined lerms -3
undefined in Euclidean geometry. 3 Supporting plane. 104 • Unit point, 254 •
Plane du;dity, 16,234-236 Regular polyhedra, 298 Surface, 106
Unit segment. 169
Plane region, 83 Regular tessellation, 157-158 Surface ~f one side, 301-303
Plane similurily, 71 Relative consislency, 335 Symmedlan, 146-149 U.S. Coast and Geodetic Survey 2
Reuleaux polygons. II g ! Universal cover lt9-120 •
Plane-filling pHttems. 50 Symmedian point, 147 Valid. 335 '
Plato. 163 ReuleauxJriangie, 116-117 ! Symmetries. 40-45
Playfair"s axiom, 306 Rhombus. 218 . Veblen. Oswald, 7, 11
Symposium on ConvexilY 82 Vector. 47
Poincare, H .. 336 Riemann. Bernhard, 278. 331 j Synthetic geometry, 4 ' Vieta; F,. 200
Poincare model, 222-223. 336-340 Right hand pnraJlel. 310 ' Tacit assumptions. 6-7
Right triangle, 1-2 Von Staudt. Karl. 229
Point: Tangent, 90-92., 132-133. 190,268 Wankel engine 116
analytic projective geometry, 252 Rigid molions, 49 Tangent cone, 9 I, 105 Width ofa set'114-120
Euclidean geometry, 3 Rotation(s): Tangent vectors, 290 Young's geom~try, 29
finite geomeh·ies. II definition of, 53
Point at infinity, 251 equations for. 53-54
Point conic. 266. 269 in plane. 48
Pointwise invariant, 200 symmetries of triangle, 41
Polar, 26-27. 208-209. 268. 332 in Ihrce-space,,67

Geometry

Uploaded by

Copyright:

Available Formats

Geometry

Uploaded by

Document Information

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

Geometry

Uploaded by

Copyright:

Available Formats

What topics does the book cover?

What topics does the book cover?

What is the structure of the book?

What is the structure of the book?

CONTEMPORARY UNDERGRADUATE MATHEMATICS SERIES

Roberl J. Wisner. Editor !

~~dt:lD ieomdnn {II,.) Jamn R. Smart.

SETS OF AXIOMS AND FINITE GEOMETRIES

3 CONVEXITY 79 7.1 Fundamental Concepts 225

4 EUCLIDEAN GEOMETRY OF THE POLYGON AND

The w"rd "geometry" clIn be interpreted liten,lIy lI$ "ellrth

that two of these s,'

sets of axioms are even briefer.

1.5 FINITE GEOMETRIES OF FANO

also valid axioms in Euclidean geometry. For example, Axioms 2 and 3

Section t.3. For finite

lines of the geometry 7 An almost immediate consequence of the axioms is:

For the four-point gemnetry: common.THEOREM 17

Axioms/or Finite Geometry of Pappus

,. . p. 116cif parallel lines are defined as those thaI mtersed on line

~et of axioms [or the geometry of Pappus, using the words

Axioms/or (he Finite Geometry of Desargues

For the geometry of Figure 1.l2c:

For the geomelry of Figure U2a:

In high school geometry, it is customary to speak of rotation

deuce exists between the sets of --~------------~X

b. (.\". J') -+ (x + 2. J' - I) is an example of a translation, a type r (x. y) - (x + I. y + I)

In special cases, the product of two transformations is the

As an example of finding the product of two transformations. "

TI verification of both facts can be accomplished by completing

First Tnllls{ormarirm Peljormell

R(240)R( ;20) = (2.I. 3.

L Prove tlmt fI group of Inmsformntiolls includes the identity element.

The first type of plane motion is a ~.tfa;ls7aiiolt:J Intuitively, a

.' . . Fi ure 2.11. The symbol IP, P 21 means

d = )(-'2 - x,j'.+ (y, y,)2,

Figure 215 shows rotation of a segment and a triangle about a P C'

, FIGURE 2.16 FIGURE 2.17

V/~B Check these additional observations about reflections.

A glide reflection is the product of a reflection followed by a tl:ansiation (

jOp ~ 2. Could a translation be its own inverse? r-IC)

-:g 7_ Give an . example in

Describe the inverse of a gl,"de re fectton

X(x •x ) _ _ ........... TA{X) .

The sOl1lewhat intuitive introduction to plane motions in this

The concept of sets of equations for a translation should be

y' = x Sill 60° + ycos 60°,

DEFINITION. A translation is a transformation with equations

x' _ II = (x - h)cosoc - (y - ")sinoc.

x' ~ :<cos28 + ysin2G + 2dcosq" x' = ax + by + c,

EXAMPLE. Find the image of point (2, 6) und~r a reflection

X' = + y sin 60° +

x' ~ 1 + 3fi + 2·~ 3 + 3.f{ x' = x cos 30° + Ysin 30°,

The converse of this theorem is also true, although the px:oof is

b ut the line y = (J313~'(.

. f t anslations rotations, re '. d

DEFINITION. A plane motion is a fliJ'(!c( motion ir it is lhe

Recall that a geometry consists of the study of properties in-

under a translation.· rotation could be a glide reflection.

2.6 MOTIONS OF THREE-SPACE