A Course on Geometry:

After Euclid and His Descendants

Farhad Riahi

August, 2018
TO THE ARDENT CALLIGRAPHER
ii
Contents

1 On Symbolic Logic 3
1.1 Preamble . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 First Definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 The Symbols of Logic . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4 Tautologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.5 Implication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.6 Converse, Inverse, Contrapositive . . . . . . . . . . . . . . . . . . 15
1.7 Quantifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.8 Quantifiers and Mathematics . . . . . . . . . . . . . . . . . . . . 19

2 On Formal Axiomatic Systems 25

2.1 Definitions and Basic Concepts . . . . . . . . . . . . . . . . . . . 25
2.2 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30

3 On the Foundations of Euclidean Geometry 45

3.1 First Axioms and Distance Measurements . . . . . . . . . . . . . 45
3.2 Axioms of Betweenness . . . . . . . . . . . . . . . . . . . . . . . . 48
3.3 Angles and Triangles . . . . . . . . . . . . . . . . . . . . . . . . . 54
3.4 Axioms for Angle Measurements . . . . . . . . . . . . . . . . . . 57
3.5 Congruence of Angles . . . . . . . . . . . . . . . . . . . . . . . . 59
3.6 Congruence of Triangles . . . . . . . . . . . . . . . . . . . . . . . 60

4 On Neutral Geometry I 69
4.1 Congruence of Triangles . . . . . . . . . . . . . . . . . . . . . . . 71
4.2 Theorems on Perpendiculars in a Plane . . . . . . . . . . . . . . 74
4.3 Fundamentals of Spatial Geometry . . . . . . . . . . . . . . . . . 78

5 On Neutral Geometry II 89
5.1 Geometric Inequalities . . . . . . . . . . . . . . . . . . . . . . . . 89
5.2 Perpendicular Lines and Planes . . . . . . . . . . . . . . . . . . . 95

6 The Parallel Axiom 111

6.1 Parallel Lines in a Plane . . . . . . . . . . . . . . . . . . . . . . . 111
6.2 Quadrilaterals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117

iii
iv CONTENTS

6.3 Parallel Lines and Transversals . . . . . . . . . . . . . . . . . . . 121

6.4 Parallelism in Space . . . . . . . . . . . . . . . . . . . . . . . . . 123

7 On Area 137

8 On Similarity 155

9 On Circles and Spheres 169

9.1 Tangent Lines and Planes . . . . . . . . . . . . . . . . . . . . . . 169
9.2 Arcs, Angles, Secants . . . . . . . . . . . . . . . . . . . . . . . . . 174

10 Loci, Constructions, and A Medley ... 195

10.1 Loci . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
10.2 Constructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
10.3 Some Newer Theorems . . . . . . . . . . . . . . . . . . . . . . . . 210

11 On Arc Lengths and Sector Areas 219

11.1 Arc Length . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
11.2 Sector Area . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224

12 On Solids and Their Volumes 231

12.1 Prisms and Cylinders, Pyramids and Cones . . . . . . . . . . . . 231
12.2 Volumes of Solids . . . . . . . . . . . . . . . . . . . . . . . . . . . 235

13 Introduction to Non-Euclidean Geometry 249

13.1 On Neutral Geometry III . . . . . . . . . . . . . . . . . . . . . . 249
13.2 The Hyperbolic Axiom . . . . . . . . . . . . . . . . . . . . . . . . 259
13.3 On Elliptic Geometry . . . . . . . . . . . . . . . . . . . . . . . . 268

14 On Regular Polyhedra 275

15 Trigonometry 289
15.1 Right Triangle Trigonometry . . . . . . . . . . . . . . . . . . . . 289
15.2 The Unit Circle and (Plane) Trigonometric Functions . . . . . . 293
15.3 The Law of Sines, The Law of Cosines, and All That . . . . . . . 299

A Euclid’s Axioms 307

B Hilbert’s Axioms 311

C The Field of Real Numbers 315

D The Basic Similarity Theorem 321

E The Two-Circle Theorem 325

F Additional Exercises 329

CONTENTS v

G A Potpourri of Nontrivial Problems 339

H Mathematical Induction 347

vi CONTENTS
Preface

Among all human intellectual journeys, mathematics is unique in that each of

its statements will not be recognized valid unless supported by a proof. Thus,
the notion of proof, i.e. an irrefutable logical argument, permeates the entire
mathematical experience, and, at the high school level, Geometry constitutes
the discipline par excellence where this indispensible ingredient can be concep-
tualized by young students for use in subsequent more advanced courses.
If elementary Geometry deserves to be properly understood, then it should
be taught equally properly. A student of mathematics needs to learn to analyze
the structure of a proof critically and to write mathematical statements that
can be taken literally. Unfortunately, in recent years, high school instruction
in Geometry has become less extensive and rigorous and does not prepare stu-
dents for sound appreciation of the subject matter of Algebra and Calculus.
Consequently, many students of mathematics leave high school for college with
a conception of mathematical proof that is not much better than the under-
standing they had developed during their middle school years.
The purpose of the present text is to elucidate and establish on firm, per-
manent foundations the very notion of what constitutes a mathematical proof
through the vehicle of Geometry. It begins at the beginning with an introduc-
tion to the rudiments of logic—logical argumentation—that are indispensible
for elaborating rigorous mathematical proofs, and proceeds with a detailed de-
velopment of axiomatic Euclidean Geometry, with an optional brief excursion
into Hyperbolic Geometry. Attempt has also been made to include topics that
are accessible to high school students, are interesting in their own right and
stimulate a sense of curiousity and discovery in the young reader. By stressing
abstraction and rigor of the argument, it is hoped that this text will better pre-
pare students to participate in the mathematical experience that is unfolding
around us. In particular, a completely logical development of Geometry should
contain proofs of “intuitively obvious” and generally accepted theorems. This
course tries to adhere to this requirement to the letter, at the risk of boring the
impatient and mathematically inclined juvenile student. The benefit of enduring
such a boredom will become apparent when the study of more advanced math-
ematics is undertaken. The proposed exercises are also meant to be nontrivial,
that is, challenging and rewarding.
I would like to express my sincere gratitude to my colleagues, Dr. Robert
Sherry and Dr. Robert Vollrath, for a critical reading of the whole manuscript

1
2 CONTENTS

and for suggesting numerous valuable improvements. Three generations of

mathematically inclined Commonwealth School students (classes of 2006, 2007,
and 2008) have worked through consecutive revisions of this text, pointed to
many typographical errors, proposed more instructive examples and clearer ex-
planations of certain abstract concepts. Their enthusiasm and ebullient cu-
riousity have been a source of inspiration for me while boosting my morale. I
am particularly indebted to Eric Brotman, Aaron Littman, Anna Moss, Danielle
Boyda, Jenny Oberholtzer, Michael Cohen, Max Ehrman, Alice Kaufman, Carol
Moraff, and Meghan Short for their contributions to the amelioration of this
course. Needless to say, all shortcomings or errors remain my sole responsibil-
ity.
Chapter 1

On Symbolic Logic

1.1 Preamble
Aristotle deserves full credit as the founder of formal (or verbal) logic, that
is, a set of laws regulating human reasoning. Even though he restricted his
consideration almost exclusively to the syllogism, the most famous of which
is: “All men are mortal; all heroes are men; therefore all heroes are mortal,”
the Greek philosopher exerted a profound influence on all scientific thought for
more than two millennia. He set forth 14 such syllogisms and believed that
they summed up most of the operations of reasoning. During the Middle Ages,
theologians added five more syllogisms to Aristotles’s 14, and for hundreds of
years these 19 syllogisms were the foundation of the teaching of logic.
As a challenging warm-up, consider this syllogism taken from a textbook
on logic by Lewis Carroll, the mathematician author of Alice’s Adventures in
Wonderland:

- No kitten that loves fish is unteachable.

- No kitten without a tail will play with a gorilla.

- Kittens with whiskers always love fish.

- No teachable kitten has green eyes.

- No kittens have tails unless they have whiskers.

One, and only one logical conclusion can be deduced from this set of state-
ments. To deduce this answer, reword and rearrange the statements as follows:

- Green-eyed kittens cannot be taught.

- Kittens that cannot be taught do not love fish.

- Kittens that do not love fish have no whiskers.

3
4 CHAPTER 1. ON SYMBOLIC LOGIC

- Kittens that have no whiskers have no tail.

- Kittens that have no tails will not play with a gorilla.

The single conclusion is thus that green-eyed kittens will not play with a
gorilla.
These rewording and rearrangements of the original statements have been
carried in accordance with the Rules of logic, and it is the purpose of this
chapter to formulate and develop these Rules. We shall, however, not follow the
Aristotelian, that is, verbal, approach. Rather, we will introduce and study the
fundamental elements of the mathematical discipline known as symbolic logic
which is a modern, expanded version of classical (Aristotelian) logic.
With the rapid advancement and diversification of human knowledge and
the development of technology and business operations, problems of increasing
complexity arose which could no longer be solved by the conventional methods
of verbal logic. Although all the basic facts were clearly spelled out, their in-
terrelationships were so intricate that the human brain could hardly organize
them logically, that is, analyze them with words. A big turning point came in
1847 when George Boole (1815–1864) published a pamphlet entitled The Math-
ematical Analysis of Logic—Being an Essay Towards a Calculus of Deductive
Reasoning.
This title says it all. Boole’s fundamental contribution was to produce a
first operational framework for handling algebraic symbols representing classes
of objects or ideas and deducing more complex statements about them through
algebraic operations. Thus, Boole can be considered as the inventor of sym-
bolic logic which shall occupy us in this chapter. Today, the term Boolean
Algebra refers to an uninterpreted abstract algebraic structure that can be ax-
iomatized in all kinds of ways. “Uninterpreted” means that no concrete mean-
ings whatever—in logic, mathematics or the physical world—are assigned to the
structures, symbols and operations.
Until modern times, logic was considered a discipline of philosophy, but since
its birth, symbolic logic has enjoyed a steadily widening field of applications
with the major part of the work being in mathematics. In fact, the influence of
this tool on mathematics has been so great that many mathematicians consider
mathematics as only a branch of (symbolic) logic.
Modern logicians have rejected four of the 19 syllogisms formulated by Aris-
totle and his medieval followers, and reduced the remaining 15 to just five the-
orems. They have worked out the formal structure of logic algebra in the same
way that mathematicians have elaborated the formal structure of number alge-
bra and other mathematical disciplines through adoption of certain definitions
and postulates from which new results (theorems) are obtained through well-
defined operational Rules. Attempts have also been made to widen the scope
of Aristotelian logic through abandoning one of Aristotles most fundamental
Rules of logic, namely the Law of the Excluded Middle which requires that a
(logical) statement must be either true or false. In a new, multivalued logical
system, a statement may have, say, three values: true, false, or indeterminate.
1.2. FIRST DEFINITIONS 5

An example of an actual trivalued logical system is provided by the Scottish

trial system which allows three verdicts: guilty, not guilty, and “not proven.”
After this preamble, we now proceed with an elementary exposition of the
fundamental concepts and techniques of two-valued symbolic logic.

1.2 First Definitions

To begin with, logic may be defined as the science that evaluates arguments. We
encounter arguments on a daily basis, reading them in books and newspapers,
hearing them on radio and TV, articulating them when communicating with
friends, relatives, and other human beings. The aim of logic is to propose a
framework consisting of Rules and methods that we may use for evaluating the
arguments of others or for constructing our own.
An argument, as it occurs in logic, is a group of statements, several of which
(the premises) are claimed to imply, to provide support for or reason to believe
in, one of the others (the conclusion). The purpose of logic, as the science that
evaluates arguments, is therefore to develop methods that allow to distinguish
good arguments (the premises do support the conclusion) from bad ones (the
premises do not support the conclusion). Needless to say, the word “argument”
has a very specific significance in logic and does not mean, for example, a verbal
fight as one might have with one’s parents or teacher. Here, we will be primarily
concerned with mathematical arguments, usually called proofs.
We begin with the introduction of the concept of statement (or proposi-
tion), a term already used in preceding paragraphs, as meaning typically an
unambiguous declarative sentence. The following sentences are statements:

- Gold is a precious metal.

- George Washington was a Greek god.

The first statement is generally agreed to be true while the second is known
to be false. In two-valued logic, statements can only be either true or false. Any
statement which is simultaneously both true and false cannot be dealt with in
this logic and should be discarded. Truth and falsity are called the two truth
values of a statement. Thus, the truth value of the first statement above is true,
and that of the second statement is false. Unlike statements, many sentences
cannot be said to be either true or false. Questions, suggestions, commands, or
exclamations such as

- Who is coming for dinner?

- Lets go to the movies tonight.

- Take a left turn at the next intersection.

- By Jove!
6 CHAPTER 1. ON SYMBOLIC LOGIC

are not usually classified as statements and do not appear in logical cogitation.
In all that precedes, we have implicitly observed two of the basic require-
ments of logic, namely:
Requirement 1. In any logical discourse, a mutual understanding of the mean-
ings of words and symbols used in the discourse should be taken for granted.
Requirement 2. In any logical discourse, agreement on how and when one
statement “follows logically” from another (or agreement on certain Rules of
reasoning) should be taken for granted.
The Rules of reasoning referred to in Requirement 2 should obviously be
spelled out at the outset of the discourse, but to discuss them thoroughly and
exhaustively would be tantamount to changing the content of this textbook from
Geometry to Symbolic Logic. Instead, we will limit ourselves to the enumeration
of a few basic Rules of reasoning which have particular relevance to the context
of Geometry and are essential for an adequate appreciation of the material of
this chapter. This being said, the above two general requirements should be
borne in mind constantly in order to prevent the discourse from degenerating
into an undesirable argument of the kind mentioned above.
We now formulate the first Rule of our framework:
Rule 1. A statement is either true or false. No statement is both true and
false.
This Rule combines the Aristotelian Law of the Excluded Middle and the
Law of Non-Contradiction.
Statements, also called logical variables, will be denoted with lower case
Latin letters p, q, t, . . .
Thus, the value that any logical variable can take is either true or false.
A simple statement is one that does not contain any other statement as a
component. Here are some examples:
- The Earth orbits the Sun.
- The Boston Public Library is in Chicago.
- I hate school.
A compound statement is one that contains at least one simple statement as
a component. Here are some examples:
- It is not the case that President Lincoln wrote the Bible.
- The Boston Symphony Orchestra will perform Mozart and the Cleveland
Symphony Orchestra will play Beethoven.
- If you score at least a B in Calculus then you will be allowed to enroll in
Theoretical Calculus.
- It will rain or snow today.
- A triangle with sidelengths a, b, c where c > b > a is a right triangle if
and only if c2 = a2 + b2 .
1.3. THE SYMBOLS OF LOGIC 7

1.3 The Symbols of Logic

Compound statements containing connectives like “it is not the case that,”
“and,” “or,” “if . . . then . . . ,” and “if and only if” (also sometimes abbrevi-
ated by the unusual “iff”) may be represented by special symbols called logical
operators. The basic logical operators are conventionally accepted to be:
Operator Name Logical function Used to translate
− (or ∼) Minus (tilde) Negation Not, it is not the case that
+ (or &) Plus (and) Conjunction And, also, moreover
∨ Wedge Disjunction Or, unless
→ Arrow Implication If . . . then . . . , only if
= Equality Equivalence If and only if
In accordance with Requirement 1, when we say that the logical operators
may be used to translate logical expressions into ordinary language (English or
any other modern language), we mean that the operators convey a certain aspect
of their corresponding connectives in that language. The precise character of
this aspect will be spelled out in the subsequent paragraphs. For now, we
want to develop some familiarity with the logical operators through practice in
translation:
- Negation. If p stands for a specific statement, then “not p” or −p (or ∼ p)
stands for the negation of p. If p stands for “President Lincoln wrote the
Bible”, then −p stands for “It is not the case that President Lincoln wrote
the Bible.”
- Conjunction. The symbol p + q (or p&q) means “p and q.” If p means
“I found my dog” and q means “I returned home,” then p + q (or p&q)
means “I found my dog and I returned home.” The compound statement
p + q is called the conjunction of p and q.
- Disjunction (also called Alternation). The symbol “p ∨ q” means “p or q
or both,” that is, at least one of the two statements is true. The use of
“or” in English is ambiguous as displayed in “It will rain or snow today.”
Nobody would consider the speaker wrong if both rain and snow came
down.
It should also be clear that p + q and q + p as well as p ∨ q and q ∨ p
have the same logical content. Before tackling the remaining two operators, we
introduce an important concept which will allow for precise definitions of the
logical operators.
Let us consider the definition of the negation. This definition is given in a
truth table, an arrangement (tableau) of truth values that shows how the truth
value of a compound statement depends on the truth values of its components.
The truth table for negation shows how any statement having the form of a
negation (−p) depends on the truth value of the statement (p) that is negated:
p −p
True False
False True
8 CHAPTER 1. ON SYMBOLIC LOGIC

This table and, hence, the definition of the negation, is a consequence of

Rule 1.
At this stage, we should introduce Rule 2 which states that “two negatives
make a positive:”
Rule 2. The statement −(−p) means the same as p.
It is often convenient to use a mathematical notation for truth and falsity,
using “p is one” to mean that p is true, and “p is zero” to mean p is false. The
truth table then becomes a table of values for a function in which, in the case
of negation, p is the single logical variable:
p −p
0 1
1 0
Next, let us consider the definition of conjunction via a truth table. To
this end, we need
Rule 3. The statement −(p + q) means the same as −p or −q (that is, (−p) ∨
(−q)).
Using the truth table for negation and Rules 2 and 3, we can readily establish
the following truth table which concisely defines the conjunction (with T for
true and F for false):
p q p+q
T T T
T F F
F T F
F F F
We can also define the disjunction p ∨ q by listing the values of the logical
function p ∨ q when the variable p and q take on different values:
p q p∨q
1 1 1
0 1 1
1 0 1
0 0 0
One should keep in mind that we are using plus, minus, times, one, and zero
with meanings different from those used in Algebra. For instance, the first three
entries in the (p∨q)-value table would be false in ordinary number algebra while
the last entry would be true (0 + 0 = 0).
Similarly, the truth table (and definition) for conjunction is equally easily
established:
p q p+q
1 1 1
1 0 0
0 1 0
0 0 0
The use of truth tables helps when analyzing compound statements and can
clarify some of the ambiguities inherent to everyday language. For instance, if
1.3. THE SYMBOLS OF LOGIC 9

you hear “I must vote for him or lose my job” followed by “That’s not true,”
does the second speaker mean “You don’t have to vote for him or you don’t
have to lose your job” or “You don’t have to vote for him and you don’t have to
lose your job” or something different from both of these? The question is easily
answered if one uses “or” in the logical sense described above. Make up a truth
table, letting p stand for “I must vote for him” and q for “I must lose my job,”
and list the truth values of p ∨ q. Then fill the −(p ∨ q) column using Rule 3:
p q p ∨ q −(p ∨ q)
1 1 1 0
1 0 1 0
0 1 1 0
0 0 0 1
We see that −(p ∨ q) is true only when p and q are both false and the
required negation is therefore (−p and −q) “You don’t have to vote for him
and you don’t have to lose your job.”
In order to codify—and, hence, simplify—the syntax of logical statements,
we introduce the following definition:

Definition. When the truth tables of any two statements p and q are identical,
these statements are called logically equivalent and denoted with p = q.

In other words, p = q means that if p is true, then q is true AND if q is

true, then p is true. For example, in p + q = q + p, each side of this equality has
exactly the same logical content. In the sequel, we shall see that equivalence
can be expressed by the validity of an implication AND of its converse. Then,
one recognizes that the above definition effectively reduces the number of inde-
pendent logical operators to four. We shall encounter further reductions of the
number of independent logical operators in the subsequent development of the
structure of symbolic logic.
This last example suggests that we could express −(p ∨ q) in the form (−p) +
(−q) that is:

−(p ∨ q) = (−p) + (−q) (1.1)

Equation (1.1) says that the statement “(p ∨ q) is false” is equivalent to the
statement “p is false and q is also false.” To prove Equation (1.1), set up a truth
table for (−p) + (−q) to see whether it agrees with the truth table of −(p ∨ q):
p q −p −q (−p) + (−q)
1 1 0 0 0
1 0 0 1 0
0 1 1 0 0
0 0 1 1 1
The last columns of the two tables agree for each value of p and q, and hence
Equation (1.1) is established. Equation (1.1) is one of the two De Morgan’s
Laws, the other being

−(p + q) = (−p) ∨ (−q) (1.2)

10 CHAPTER 1. ON SYMBOLIC LOGIC

The proof of Equation (1.2) is similar to that of Equation (1.1) and is left as
an exercise. According to the second De Morgan’s Law, the negation of “Either
I will score an A in English or a B in Calculus” is “I will not score an A in
English and I will not score a B in Calculus.”

1.4 Tautologies
Some compound statements made up of variables p, q, . . . with the connectives
“and,” “or,” and “not” turn out to be true for all possible values of the logical
variables. Such propositions are called tautologies. As a first example, consider
the compound statement (p ∨ q) ∨ [(−p) + (−q)]. Its truth table is:
p q p ∨ q −p −q (−p) + (−q) (p ∨ q) ∨ [(−p) + (−q)]
1 1 1 0 0 0 1
1 0 1 0 1 0 1
0 1 1 1 0 0 1
0 0 0 1 1 1 1
Similarly, De Morgan’s Laws (1.1) and (1.2) hold for all values of p and q,
and are hence tautologies.
At the other end of the truth spectrum are compound statements which are
false in all cases. They are called inconsistent or contradictory. For example,
the statement p + (−p) is inconsistent, as shown by its truth table:
p −p p + (−p)
1 0 0
0 1 0
Most of the compound statements in ordinary use fall between the two
extremes of tautologies and inconsistent propositions. Compound statements
which are sometimes true and sometimes false are called consistent.
Finally, it is worthwhile mentioning an example for certain verbal paradoxes
which have no solution in a two-valued logical framework. Consider the state-
ment “All Rules have exceptions.” This statement denies itself, since by its own
assertion the statement must have exceptions and therefore cannot be true. Such
indeterminate statements are usually banned from the realms of conventional
logic because they violate Rule 1.
We have already seen that the logical operators of conjunction and disjunc-
tion are commutative. It is evident from their truth tables, i.e. from their very
definitions, that p ∨ q and q ∨ p, and similarly p + q and q + p, have the same
logical content and are, hence, equivalent. Using truth tables, it is easy to verify
that the associative law holds equally for both conjunction and disjunction:

p + (q + r) = (p + q) + r

p ∨ (q ∨ r) = (p ∨ q) ∨ r
In the algebra of numbers, multiplication is distributive over addition, that is
a(b+c) = ab+ac, but addition is not distributive over multiplication: a+(bc) 6=
1.5. IMPLICATION 11

(a + b)(a + c). In logic, however, it can be proved via truth tables that each of
the operations of conjunction and disjunction is distributive over the other:

p + (q ∨ r) = (p + q) ∨ (p + r) and p ∨ (q + r) = (p ∨ q) + (p ∨ r)

1.5 Implication
All mathematical theorems are conditional (compound) statements of the “if-
then” form:

If (hypothesis, premise), then (conclusion, consequence)

In other terms, such a conditional statement says that one statement (the
hypothesis) implies another (the conclusion). More generally, compound state-
ments of the form “If p is true, then q is true” are called implications and are
denoted p → q (read “p implies q”). The following examples are only a few of
the numerous instances where implications occur in daily human affairs and in
the communication of ideas of all kinds:

- If a toddler is read to frequently, he/she will develop a liking for books.

- If you enter my office once more, I will give you a beating.

- If (a − b)(a + b) = c, then a2 − b2 = c.

- If water is cooled below zero degree centigrade, it will freeze.

All these compound statements are in the “if-then” form, but note that an
implication p → q does not necessarily assert the truth of p or of q; it only
asserts that p and q are related in a certain way. Thus, the second example
above does not assert that “You will enter my office once more” nor does it
assert that “I will give you a beating.” In some of the examples above, the
validity of the implication can be established by logical reasoning alone while
in others, such as the last example, outside knowledge or experimentation must
be called on to verify the validity of the assertion made. We will return to this
point subsequently.
Now, part of the logical content of the implication p → q can be expressed
by stating that whenever we know that p implies q, and also know that p is
true, then we know that q is true. But this is not a complete definition because
we have not specified whether we will consider p → q to be true when p itself is
false. In fact, out of the four possible cases, only the one where the premise is
true and the conclusion is false is considered to be an invalid implication. Thus,
we are led to formulate

Rule 4. The statement −(p → q) means the same as p + (−q).

12 CHAPTER 1. ON SYMBOLIC LOGIC

Hence, the truth table for implication is:

p q p→q
1 1 1
1 0 0
0 1 1
0 0 1
Useful alternative definitions for p → q are:

(p → q) = −[p + (−q)] and (p → q) = (−p) ∨ q

which can be verified by truth tables and are quite useful in applications.
It is instructive to note that these two logically equivalent statements define
implication in terms of negation, conjunction or disjunction only. Thus, we are
left now with only three basic logical operations.
An important property of implication is that of transitivity: if p implies q
and q implies r then p implies r:

[(p → q) + (q → r)] → (p → r)
This property can be readily proved by a truth table.
We mentioned earlier that some implications can be shown to be valid by
logic alone. Such implications are, in fact, tautologies. Consider the example
“If it is not true that she has not eaten and not slept, then she has either eaten
or she has slept.” Let p stand for “She has eaten” and q for “She has slept.”
Then the statement becomes:

−[(−p) + (−q)] → (p ∨ q)
Comparison of the truth tables for the statements in the left and right hand
sides of this expression shows that these statements are indeed logically equiv-
alent, so that whenever the “if” part of the implication is true, the “then” part
must also be true, regardless of the meaning of p and q. Implications that are
not tautologies cannot be proved by logic alone but are nonetheless useful in
organizing information and drawing logical conclusions. In general, when the
truth of an implication p → q is known, then as soon as the truth of p is estab-
lished, the truth of q is established. One should not forget, however, that the
primary purpose of the statement “p implies q” is to express knowledge about
the relationship between p and q when we do not know yet whether p or
q individually is true or false.
This section on implication began with the statement that mathematical
theorems are statements of the form “if . . . then,” hence, implications. We
elaborate somewhat on this assertion by adding that in a given mathematical
system the only statements which are called theorems—from the ancient Greek
theorein, to look at, to observe—or sometimes propositions, corollaries, lemmas
are those for which a proof has been supplied. This leads us to present here a
preliminary definition of the concept of “proof” in mathematics. By definition, a
proof is a sequence of statements (i.e. an argument), together with a justification
1.5. IMPLICATION 13

for each, ending up with the desired conclusion. Only six types of justifications
are allowed in a proof:

1. “By hypothesis . . . ”

2. “By axiom (or postulate) . . . ”

3. “By theorem . . . ” (proved previously)

4. “By step . . . ” (a previous step in the argument)

5. “By definition . . . ”

6. “By Rule . . . of logic”

The concepts of definition and axiom (or postulate) will be introduced and
discussed in detail in the next chapter, but for now and in light of the preced-
ing remarks on the concept of what we have called mathematical proof, it is
imperative to introduce

Rule 5. No unstated assumption may be used in a proof.

Note that this does not mean that an argument is invalid if its premises
have not been established as true—merely that the conclusion of the argument
may be false, and so the assumption should not be used.
Last, and most importantly, it should be borne in mind that there is no rote
method for proving or disproving every statement in mathematics. This truth
will become increasingly evident as we progress in this course.
We may now use what we have learned so far to test arguments. As men-
tioned earlier, there are usually two ingredients in an argument: first, we must
prove that a set of assumptions or premises is true and then prove that these
premises imply the conclusion. We illustrate this process with the following
argument:
If the Headmaster coaches the School’s soccer teams, Ignaz will be assigned
to teach math. (Symbolize this by h → i).
If Ignaz teaches math, students will not score high in the SAT 1. (Symbolize
this by i → −s).
Therefore, if students score high in the SAT1, the Headmaster does not coach
the School’s soccer teams. (Symbolize by s → −h).
The word “therefore” is used to indicate the belief that the last statement
(s → −h) is a logical consequence of the preceding ones, and to test the validity
of this argument we must examine whether or not this is indeed the case. Hence,
we must verify the validity of the implication:

[(h → i) + (i → −s)] → (s → −h)

We can test this alleged implication by truth tables and find that it is, in
fact, a tautology so that the argument is valid. We can also treat the alleged
implication as a mathematical formula to be proved by application of established
14 CHAPTER 1. ON SYMBOLIC LOGIC

Rules and results. In our example, the transitivity property of implication

applied to the left-hand side, that is, to the premise (h → i) + (i → −s) yields
(h → −s). Furthermore, by virtue of Rule 3 and Rule 4, we have the equivalence

(p → q) = (−p) ∨ q
(mentioned earlier as one alternative definition for p → q) so that:

(h → −s) = (−h) ∨ (−s) and also (s → −h) = (−s) ∨ (−h)

Hence,

(h → −s) = (s → −h)
and therefore

[(h → i) + (i → −s)] → (s → −h)

which was the implication to be proved. Thus the argument has been proved
valid by the laws of logic alone, and is indeed a tautology.
For quick reference, we gather the logic laws established so far:

−(p ∨ q) = (−p) + (−q)

−(p + q) = (−p) ∨ (−q)

p ∨ (q ∨ r) = (p ∨ q) ∨ r

p + (q + r) = (p + q) + r

p + (q ∨ r) = (p + q) ∨ (p + r)

p ∨ (q + r) = (p ∨ q) + (p ∨ r)

(p → q) = (−p) ∨ q = −[p + (−q)]

[(p → q) + (q → r)] → [p → r]
It should be emphasized once more that the validity of an argument does not
establish the truth of the premises nor the truth of the conclusion. To prove
that a conclusion is true, one must prove that the argument leading to that
conclusion is logically valid, and also that the premises are true. The reason
for clearly distinguishing the premises of an argument from the conclusion is
that one may judge the validity of the argument without being prejudiced by a
belief in the truth or falsity of the premises or conclusion. Thus, the argument:
1.6. CONVERSE, INVERSE, CONTRAPOSITIVE 15

“If I am the richest man in the world, then I can speak all the world languages.
If I can speak all the world languages, then I will never die. Therefore, if I am
the richest man in the world, I will never die.” is a perfectly valid argument
although both its premises and conclusions are obviously false.

1.6 Converse, Inverse, Contrapositive

Given an implication p → q, three different implications can be constructed
from it:

- Its converse: q → p

- Its inverse: −p → −q

- Its contrapositive: −q → −p.

One of the most frequent mistakes in applying logic is the assumption that
if a given implication is valid, then its converse must also be valid. Regrettably,
this human weakness is often exploited by advertisers, politicians, and other
“persuaders” to advance their agenda. For a trivial example, consider the im-
plication “A universal health coverage plan costs money” and its converse “If a
health coverage plan costs money, then it must be universal.”
The following truth table should clarify the relationship between an impli-
cation and the three variations derived from it:
p q −p −q p → q q → p −p → −q −q → −p
1 1 0 0 1 1 1 1
1 0 0 1 0 1 1 0
0 1 1 0 1 0 0 1
0 0 1 1 1 1 1 1
Since the truth table does not contain any reference to the nature of the
statements p and q other than their truth or falsity, the conclusion we draw from
it will be correct for all possible examples of p and q . This table shows that
the contrapositive agrees in all four of its values with the corresponding values
of the original implication. Hence, the contrapositive is logically equivalent to
the original statement. The table also shows that the inverse and converse are
also logically equivalent to each other, but not to the original implication.

1.7 Quantifiers
So far, we have considered statements (simple or compound) as building blocks
from which arbitrarily complex composite statements can be constructed. Log-
ical deductions were made either by showing that the statement to be deduced
is a tautology, or by showing that it is the consequent (conclusion) of an impli-
cation whose antecedent (premises) is known to be true. Unfortunately, many
simple logical propositions, though valid, cannot be deduced by this procedure.
For instance, the intuitively acceptable statement “There is at least one math
16 CHAPTER 1. ON SYMBOLIC LOGIC

teacher. If all math teachers are vertebrates, then some math teacher at Com-
monwealth School is a vertebrate” cannot be proved to be valid by the methods
used heretofore (Try to prove it!). The reason is that, so far, a statement was
either true or false, whereas the premise and the conclusion of this implica-
tion are actually statements which relate to two different classes or categories
(vertebrates and teachers) and include all of one class and only part (some) of
the other. Such statements are called categorical statements. The classes in a
categorical statement are called the subject term and the predicate term, and
the statement asserts that either all or part of the class denoted by the subject
term is included in or excluded from the class denoted by the predicate term.
Here are some categorical statements:

- Light rays travel at a fixed speed in vacuum.

- Fundamentalists are not easy to argue with.

- Some drug addicts die from overdose.

- Not all human beings are diabetics.
- The movie star Arnold Schwarzenegger acted in the movie “The Termi-
nator.”

The first statement asserts that the entire category (all) of light rays is
included in the category of things that move at a fixed speed in vacuum, the
second that the entire class (all) of fundamentalist persons is excluded (not)
from the category of persons with whom it is easy to argue, the third that part
of the class of drug addicts (some) is included in the class of persons who die
from overdose. The fourth statement asserts that part of the class of human
beings (not all) is excluded from the class of diabetics, and the fifth that the
single individual denoted by the name “Arnold Schwarzenegger” is included
in the class of actors who acted in the movie “The Terminator.” Since any
categorical statement asserts that either all or part of the class denoted by the
subject term is included in or excluded from the class denoted by the predicate
term, it follows that here are exactly four types of categorical statements: if S
denotes the subject term and P denotes the predicate term, then

- All S are P
- No S are P

- Some S are P
- Some S are not P

Arguments depending on (usually) two categorical statements for premises

with a conclusion to be drawn from them are called syllogisms. The most
common technique for judging the validity of a syllogism makes use of Venn
diagrams (after the 19th century British mathematician John Venn). Venn
1.7. QUANTIFIERS 17

diagrams consist of two or more overlapping circles, each of them representing

the information content of the categorical statements appearing in a syllogism.
The size of these circles is irrelevant, but is usually chosen to be the same for
all of them. We consider first a categorical statement: it has exactly two terms,
and hence, two circles appear in its Venn diagram. One circle, S, is selected
to represent the subject term, that is, the members of the class denoted by
the subject term (if such members exist) are conceived as being located inside
circle S. The predicate term is dealt with similarly with a circle P and the
corresponding interpretation.

Thus, if an entity is in the region numbered 2 in the diagram above, this

entity belongs simultaneously to the classes of the subject term and of the
predicate term. An entity in region 4 does not belong to either of these two
categories. Shading an area means that the shaded area is empty. Placing an
X or a bar in an area means that at least one thing exists in that area. If no
marks appear in an area, nothing is known about that area: it may contain
members or it may be empty. With these conventions, we have the following
four possible Venn diagrams for each type of categorical statement:

The earlier mentioned implication with categorical premise (from the first
paragraph of this section) “There is at least one math teacher. If all math
18 CHAPTER 1. ON SYMBOLIC LOGIC

teachers (T ) are vertebrates (V ), then some math teacher is a vertebrate” can

now be deduced to be true, as shown by its Venn diagram below:

To examine the validity of a syllogism, we draw a set of overlapping circles

in such a way as to show all possible intersections, and then shade the areas
which are known to contain no members. For example, judge the validity of
the following argument (syllogism): “Every Commonwealth School alumnus is
either rich or famous, but some of the alumni are not rich. Therefore, not all
famous people are rich.”

Let A stand for the category of alumni, R for the class of people who are
rich, and F for that of famous people. Record the fact that every alumnus is
either rich or famous by shading the part of A which is neither in R nor in F .
Record the fact that some alumni are not rich by placing an X in the part of
A that is not in R and has not been previously shaded. This X then shows
that there exist famous people who are not rich, that is in F but not in R, and
justifies the required conclusion:

Now consider the argument: “Whatever interests me does not interest my

father, and whatever interests my mother does not interest my father. Therefore
whatever interests me interests my mother.”
1.8. QUANTIFIERS AND MATHEMATICS 19

The three circles are labeled I, F , and M to represent things that interest
me, my father, and my mother. The first and second premises are indicated by
horizontal and vertical hatching respectively. We see that the claimed conclusion
is not valid, because circles I and M have only a “small” area in common and
“large” disjoint areas.
Sometimes it is necessary to vary the technique slightly. Consider the argu-
ment: “All living beings are either vertebrates or invertebrates. Human beings
are not invertebrates, therefore human beings are vertebrates.”

Here, we have to begin with the class L of all living beings, and insert the
three overlapping circles representing human beings (H), vertebrates (V ), and
invertebrates (I) inside the larger area L, as shown. Then, we record the first
premise by hatching (horizontally) all the area in L which is outside of both
the V and I circles. The second premise is recorded by slanted hatching. The
conclusion is seen to be valid.

1.8 Quantifiers and Mathematics

Most mathematical statements contain variables, that is, quantities which can
take on any one from a set of allowed values. For instance, the Pythagorean
Theorem states that for any right triangle, if a and b are the lengths of the
legs and c is the length of the hypotenuse, then a2 + b2 = c2 . Here a, b, c are
variable (real) numbers, and the triangle whose sides they measure is a variable
20 CHAPTER 1. ON SYMBOLIC LOGIC

triangle. Now, mathematical statements being usually categorical statements, it

often becomes necessary to quantify the variables contained in such statements.
It turns out that variables can be quantified in two different ways. First, in a
universal way, as in the expressions:

- “For any x . . . ”
- “For every x . . . ”
- “For all x . . . ”
- “Given any x . . . ”
- “If x is any . . . ”

Variables can also be quantified in an existential way, as in the expressions:

- “For some x . . . ”
- “There exists an x . . . ”
- “There is an x . . . ”
- “There are x . . . ”

Again, let us emphasize that a statement beginning with “For every . . . ”

does not imply the existence of anything: the statement “Every red elephant
has polka dots” does not imply that red elephants exist!
If a variable x is quantified universally, this is usually denoted as ∀x (read:
“For all x . . . ”), and when it is quantified existentially, this is usually denoted
as ∃x (read: “There exists an x such that . . . ”). After a variable x is quantified,
some statement is made about x which we denote as p(x) (read: “Statement
about x . . . ”). Logic Rules for negating quantified mathematical statements
are:
Rule 6. The statement −[∀x p(x)] is logically equivalent to ∃x[−p(x)].
In words, the negation of the statement that statement p(x) holds true for
all x asserts that for some x, the statement p(x) does not hold. For example, to
negate “All rectangles are squares” is to assert that “There is a rectangle which
is not a square”.
Rule 7. The statement −[∃x p(x)] is logically equivalent to ∀x[−p(x)].
In words, to deny that there exists an x such that p(x) holds is to assert
that for all x statement p(x) is not true.
To conclude this chapter, we examine a certain procedure for proving math-
ematical statements which is used extensively in all mathematical disciplines,
including Geometry, and is called reductio ad absurdum, abbreviated RAA, or
proof by contradiction or indirect proof. In this method of proof for the impli-
cation h → c (h for hypothesis and c for conclusion), one begins by assuming
1.8. QUANTIFIERS AND MATHEMATICS 21

the inverse (contrary) −c of the desired conclusion to be true. This is called

the RAA hypothesis, to distinguish it from the actual hypothesis h. The RAA
hypothesis is a temporary assumption from which one derives, by logical rea-
soning, an absurd statement. Here, “absurd” qualifies a statement known to be
false. Such an absurd statement might deny the hypothesis of the theorem, or
the RAA hypothesis, or a previously proven theorem, or a statement accepted
to be universally true. Once it is shown that the negation of the conclusion c
leads to an absurdity, it follows by Rule 1 that the conclusion c must be valid.
To summarize:
Rule 8. To prove the implication h → c, assume the negation (−c) of the
statement c (RAA hypothesis) and deduce an absurd/contradictory statement
using the hypothesis h in the argument.
We shall encounter many examples of proof by RAA in our forthcoming
study of Geometry. For now, it is important to note that Rule 8 relies heavily
on Rule 1 (Law of Excluded Middle). Rule 8 has provoked much controversy
in mathematics and philosophy. Mathematicians of the constructivist school
(Brouwer, Stolzenberg, Bishop, to name a few eminent members) object to the
unqualified use of this Rule when statements about existence of mathematical
entities are involved. They contend that in order to prove that a mathematical
object exists, one must first supply an operational method for constructing it.
They conclude by claiming that it is not sufficient to assume that the object in
question does not exist (RAA hypothesis) and then derive a contradiction.
This concludes our informal excursion into the realms of symbolic logic. In
the next chapter, we examine the structure of certain mathematical systems
called Formal Axiomatic Systems, of which Geometry constitutes an example
par excellence. In our subsequent study of the axiomatic foundation of Euclidean
Geometry, the fundamental concepts and the Rules of logic, as developed in this
chapter, will constitute an indispensable tool.
22 CHAPTER 1. ON SYMBOLIC LOGIC

Exercises
1. Given that a means “I finished my work” and b means “I found my coat”
and c means “I went home,” state the following statements in words (in
as simple language as possible):

(a) a + b + c.
(b) −[a + (−b)].
(c) −[(−a) ∨ (−b) ∨ (−c)].

2. If p means “John quit watching TV” and q means “John’s math grade
improved,” rewrite the following statements as logical functions of p and
q:

(a) John quit watching TV and his math grade improved.

(b) John’s math grade dropped or he quit watching TV.
(c) John quit watching TV or else he didn’t.
(d) It is not the case that either John quit watching TV or his math
grade did not improve.

3. Construct a truth table for −[p + (−q)].

4. Using a truth table prove De Morgan’s Law: −(p + q) = (−p) ∨ (−q).

5. Make up a truth table for p + [(−q) + r].

6. By means of De Morgan’s Laws restate −[(−p) + (−q)] in simpler terms.

7. By means of truth tables, determine which of the following statements are

tautologies:

(a) −[p + (−p)].

(b) (p ∨ q) ∨ (−p).
(c) p + (p ∨ q).
(d) [p + (−q)] ∨ [(−p) + q].

8. Prove p + (p ∨ q) = p ∨ (p + q) = p.

9. Prove the distributive law: p ∨ (q + r) = (p ∨ q) + (p ∨ r).

10. Prove that the implication: [p + (p → q)] → q is a tautology.

11. Show by means of a truth table that p → q is equivalent to: −[p + (−q)].

12. Apply one of De Morgan’s Laws to −[p+(−q)] to obtain another statement

which is equivalent to p → q.
1.8. QUANTIFIERS AND MATHEMATICS 23

13. Fill the blanks in the truth table below and determine whether (p + q) →
(p ∨ q) is a tautology:
p q p+q p∨q (p + q) → (p ∨ q)
1 1
0 1
1 0
0 0

14. Test the validity of the following argument either by means of truth table
or by analysis:
“If it is not true that he has refused to study and has refused to take the
exam, then he has either consented to study or he has consented to take
the exam.”

15. State the converse, the contrapositive, and the inverse of each of the fol-
lowing implications:

(a) If x is a negative number, then the square of x is a positive number.

(b) If it has feathers and two legs, then it is a bird.

16. State the converse of the converse of an implication. Also state the con-
verse of the contrapositive of an implication, and the inverse of that con-
verse.

17. Test the equivalence of [(p + q) → r] and [(p → r) + (q → r)].

18. Which of the following implications is a tautology?

(a) p → q.
(b) (p → q) → (−q → −p).

19. Are the following arguments valid?

(a) If it rains, we will get wet. Therefore, if it does not rain, we will not
get wet.
(b) If x is even, then 2x is even. Therefore, if 2x is even, x is even.
(c) If x2 = y 2 , then x = y. Therefore, if x 6= y, x2 6= y 2 .

20. Label each of the operations that can be performed on an implication as

follows:
c: operation of replacing the implication by its converse
k: operation of replacing the implication by its contrapositive
i: operation of replacing the implication by its inverse
n: neutral operation of replacing the implication by itself.
24 CHAPTER 1. ON SYMBOLIC LOGIC

Let k · c mean “perform k and follow it by c”, and similarly for i · k , c · i ,

etc. Set up a table for the operation “·” with the elements i, c, k, and n,
and show that the successive performing of this operation does not yield
any new statement.
21. The following syllogism is taken from Lewis Carroll’s Alice in Wonderland.
Write the “if . . . then” statement on which the conclusion is based:

“In that direction,” the Cat said, “lives a Hatter; and in that
direction lives a March Hare. . . They’re both mad.”
“But I don’t want to go among mad people,” Alice remarked.
“Oh, you can’t help that,” said the Cat, “we’re all mad here.
I’m mad. You’re mad.”
“How do you know that I’m mad?” said Alice.
“You must be,” said the cat, “or you wouldn’t have come here.”

22. Draw a Venn diagram to determine the validity of the following argument:
“People who like yellow do not like purple, and some people who like purple
also like red. Therefore, some people who like red do not like yellow.”
23. (*) Introduce a new logical operator denoted by a vertical bar “|” as
follows:
If p and q are statements, then p|q means that p and q are not both true
and reads “p is incompatible with q.”

(a) Set up the truth table for p|q.

(b) Show by means of a truth table that (p|p)|p is a tautology.
(c) Now let t mean the statement (p|p)|p for conciseness. Express the
negation −p in terms of p, t, and the incompatibility operator |.
(d) With −p defined, show by means of a truth table that p ∨ q =
(−p)|(−q).
(e) Show by means of a truth table that since p|q is false only when p
and q are both true, therefore p + q is true only when it is NOT true
that p|q, i.e. p + q = −(p|q).

Having carried out the proofs of the preceding steps, and bearing in mind
that implication and equivalence have already been defined in terms of
negation and conjunction, one may now conclude that the entire edifice of
symbolic logic can be built with only one single basic logical operator!
Chapter 2

On Formal Axiomatic
Systems

2.1 Definitions and Basic Concepts

Prior to the discovery (or invention, as some mathematicians contend) of a non-
Euclidean Geometry in the second half of the 19th century independently by
Janos Bolyai and Nicolai Ivanovich Lobachevsky, it was believed that Math-
ematics should be modeled after Nature, and that its sole task consisted in
finding unique and necessary truths about the real world. Indeed, Calculus was
developed by Leibniz and Newton as a tool for solving mathematical problems
appearing in the scientific investigation of the physical world. Thus, mathe-
matical postulates were adopted from observation of Nature, and mathemat-
ical theorems were interpreted as representing Natural Laws. In particular,
Geometry (from the ancient Greek geometrein) began as the science of mea-
suring land, and continued to constitute an unequivocal description of physical
space. The non-Euclidean “revolution” forced mathematicians to revise their
traditional viewpoint of the purpose of Mathematics: evidently, Euclidean and
non-Euclidean Geometries could not each be a true description of physical space
for they contained many contradictory theorems. Since non-Euclidean Geom-
etry was indubitably a mathematical discourse, then Mathematics as a whole
could no longer be considered as a, possibly simplified, but nonetheless faith-
ful, representation of the physical world. Thus, a totally new viewpoint of
the nature and role of mathematics was initiated. The Euclidean edifice was
carefully scrutinized, and many flaws, hidden assumptions, fallacies stemming
from overreliance on intuition as well as on drawing figures, and other modes
of loose reasoning were discovered in the work of Euclid. In later Chapters,
we shall examine a few of these logical gaps in Euclid’s Elements. Thus, a
need for a logically satisfactory postulational (i.e., based on postulates) treat-
ment of Euclidean Geometry was felt. We shall not dwell here in the historical
reorganization of the Euclidean corpus, and refer to the works listed in the bib-

25
26 CHAPTER 2. ON FORMAL AXIOMATIC SYSTEMS

liography for a discussion of the “revolutionary” phenomenon already referred

to. Rather, since modern Geometry, in its very essence, is a logical abstract
discourse with a postulational framework, we shall proceed at once with the de-
scription of so-called Formal Axiomatic Systems, of which Geometry is a most
excellent example. To begin with, any logical discourse should be carried out
in intelligible language. In other words, and in accordance with Requirements
1 and 2 of the preceding chapter, the concepts of the discourse, the relations
among them, and the operations which can be performed on them should all
be defined explicitly, and agreed upon. A moment of reflection shows that such
definitions must necessarily be made in terms of other concepts, relations, and
operations. These, too, should be defined explicitly. If these are defined, it
must be again with reference to further concepts, relations, and operations, and
so on. In order to avoid a circular chain of definitions (because the vocabulary
of any modern language contains a finite number of words), the sequence of
definitions must be cut short. Thus, it becomes inevitable that one or more
concepts, relations, and operations remain undefined, that is, receive no ex-
plicit definition. These will constitute the primitive terms of the discourse.
It should be stressed clearly that such terms, when accepted once and for all,
will be freely used in the discourse without further question concerning their
actual meanings. Likewise, any effort to logically deduce the statements of the
discourse is bound to be trapped into a similar vicious circle from which there is
no escape but by accepting that one or more of the statements remain entirely
unproved. These will constitute the axioms or postulates of the discourse.
Euclid did attempt to define every single term used in his Elements, but
many of his definitions were typical examples of circularity. For example1 , Eu-
clid defines a line by stating: “A line is breadthless length.” Here, the words
“breadth” and “length” each require a definition in order to give meaning to this
definition. It can be shown2 that the pursuit of a logically satisfactory definition
of a line can be quite an exacting and fruitless undertaking. Again, Euclid de-
fines a point as being “That which has no part” and a straight line as “. . . a line
which lies evenly with the points on itself.” Leaving out the obscure property of
“lying evenly,” one cannot help but wonder how something which has “length”
is made out of objects which have “no part.” Although Euclid relied heavily
on common sense (there are five “Common Notions” in his Elements that are
accepted without question), this very reliance on common sense points to the
shortcomings of Euclid’s endeavor to define every term in his discourse. In the
case of, say, the point, he could have left it as well as undefined, couldn’t he?
We introduce now our definition of a formal axiomatic system as follows:
Definition. A formal axiomatic system is an abstract mathematical discourse
with the following constituents:
- Primitive terms (concepts, relations among concepts, operations that can
1 See The Thirteen Books of Euclid’s Elements, by Sir Thomas Heath, Dover Publications,

New York, 1956, or Appendix A

2 See, for example, The non-Euclidean Revolution, by Richard J. Trudeau, Birkhäuser,

Boston, 1987, pp. 22-33

2.1. DEFINITIONS AND BASIC CONCEPTS 27

be performed on such concepts), which are sometimes called irreducible

and are deliberately left as undefined terms and constitute the building
blocks of the system.

- Axioms or Postulates, which are statements about the primitive terms

and are deliberately left as unproved statements, and constitute the oper-
ational mechanism of the system.

- All other technical terms of the discourse (the defined terms) are defined
explicitly by means of the primitive terms and the axioms.

- All other statements of the discourse (the theorems) are logically de-
duced from the postulates.

A discourse conducted according to the pattern above is usually called a

discipline (or branch) of pure mathematics, an exercise in abstraction which the
eminent British mathematician and philosopher, Lord Bertrand Russell (1872–
1970) facetiously described by “(Pure) Mathematics is the subject in which
we never know what we are talking about, nor whether what we say is true.”
We don’t know what we’re talking about because the primitive terms have
been left undefined on purpose and all other terms of the discourse refer to
them. They can mean or be anything and do not necessarily have material
manifestations. We don’t know whether what we’re saying (read: proving) is
true because it does not necessarily have a natural representation in the physical
world, and we don’t even know what it means. But the very generality and,
hence, power of formal axiomatic systems lies in the freedom they allow for
interpreting the primitive terms and thus, for modeling—that is, for applied
mathematics. By attributing concrete meanings to the abstract primitive terms,
and turning them into the basic objects of some model of the abstract system,
the postulates become evident properties of those basic objects. This was the
view of the ancient Greek geometers. To them, Geometry was a study dealing
with the actual physical space in which the geometrical concepts of point, line,
etc. were idealizations of certain real physical entities, and in which postulates
are unquestionable statements about these idealizations. In the modern point of
view, however, Geometry is the purely abstract investigation of a certain formal
axiomatic system, and is devoid of any physical meaning or imagery. In this
sense, Henri Poincaré’s (1854–1912) claim that Mathematics is “the giving of
the same name to different things” acquires its full import.
One may wonder whether any arbitrary set of primitive terms along with
an arbitrary list of axioms relating to these terms constitute a formal axiomatic
system. The answer is “yes” in principle and “no” in practice. It occurs quite
frequently that a carelessly designed formal axiomatic system turns out to be
“vacuous,” that is, it does not yield any interesting theorem. Regrettably, one
never gets wind of such fruitless efforts because no one takes care to publish
them as a warning to newcomers in the field. In fact, in order to grow and
mature into a full-fledged discipline, a formal axiomatic system should exhibit
certain required and certain desired properties which we enumerate below.
28 CHAPTER 2. ON FORMAL AXIOMATIC SYSTEMS

The first and most elusive of the desired properties which has ever since
been the major preoccupation of (pure) mathematicians is that of “elegance.”
In many instances, mathematical elegance is synonymous with simplicity, econ-
omy, or ingenuity, all equally difficult to explain precisely. For instance, the
fewer primitive terms and the fewer postulates enunciated for a nontrivial (i.e.
with a “rich theorem content”) formal axiomatic system, the more “elegant”
this system. There are obvious limits to this reduction of the basic constituents
of a formal axiomatic system: a system with “too few” primitive terms serving
as building blocks and “too few” axioms to operate with these blocks can easily
turn out to be a sterile realm with a sparse implicational power (“meager the-
orem content”). Thus, a certain minimal richness (or complexity) of structure
is necessary in order for the formal axiomatic system to be worth the effort of
mathematicians. We shall return to this desired property later on.
The next, and most important, required property that a formal axiomatic
system should possess is that of consistency. A set of axioms is said to be con-
sistent if contradictory statements are not implied by the set. How can one be
sure that a given set of postulates will never lead to a contradiction? The short
answer is that, in general, one cannot. The only method devised so far for test-
ing the consistency of an axiom set is to try to construct a concrete model for
it. As mentioned earlier, such a model is obtained by identifying the primitive
terms of the formal axiomatic system in question with concrete objects in such
a way that the postulates of the system become true statements relating these
concrete entities. In this fashion, one may test the absolute consistency of
a formal axiomatic system, for if contradictory statements were implied by the
axioms, these would in turn reflect contradictory statements about concrete ob-
jects of the real world which is inadmissible. Unfortunately, it is rarely possible
to design such a concrete model for an arbitrary formal axiomatic system. The
best that one can hope to achieve is then to test the relative consistency of
the system by assigning to the primitive terms of the original formal axiomatic
system concepts of some other formal axiomatic system which one knows or
assumes to be consistent, in such a way that the interpretations of the postu-
lates of the original system become logical consequences of the postulates of the
model system. Thus, all one can say is that the formal axiomatic system A
is (relatively) consistent if the axiom set of the model system B is consistent.
The (relative) consistency of non-Euclidean Geometry has been established by
constructing a Euclidean model for it and by assuming that Euclidean Geome-
try itself is consistent. But the consistency of Euclidean Geometry deserves an
independent proof for its own sake.
Another property that is often required of a formal axiomatic system is
that of independence. An axiom is said to be independent if it is not a logical
consequence of the other axioms of the set. The entire axiom set is then called
independent if each axiom of the set is independent. The most famous case of
questionable independence is that of Euclid’s Fifth or Parallel Postulate: for
almost a millennium, mathematicians attempted to prove that this axiom is a
consequence of the other postulates. All such attempts were in vain, and all
proofs of the dependence of the Parallel Postulate on the other postulates were
2.1. DEFINITIONS AND BASIC CONCEPTS 29

flawed for they all relied on some subtle hidden extra assumption. With the
discovery (or invention?) of non-Euclidean Geometries and the proof of their
relative consistency with respect to Euclidean Geometry, the question of the
independence of the Fifth Postulate was definitely settled in the affirmative.
In this connection, we recall that it was the mathematicians’ obsession with
the question of the independence of Euclid’s Parallel Postulate that led to the
development of non-Euclidean Geometry and initiated the study of the structure
and properties of axiomatic systems which, in turn, shaped much of the modern
approach to mathematics.
The independence of an axiom can be tested by finding an interpretation
(model) of the primitive terms which verifies all the axioms of the set except
the axiom concerned. If such a model is found, then that particular axiom can-
not be a logical consequence of the other axioms because if it were, then the
model which converts all the other axioms into verifiable propositions should
convert this axiom too into a true proposition. It goes without saying that such
a test of the independence of an entire axiom set can be a rather lengthy proce-
dure but, in principle at least, it can be carried out. We hurry to say, however,
that the verification of the independence of an axiom set is by no means neces-
sary to the logical development of the mathematical discourse: a set of axioms
that are not independent can be qualified as being redundant in that it con-
tains statement(s) which can be treated as theorems instead of as axioms. To
cite such an example, after David Hilbert (1862–1943), the prominent German
mathematician of the early twentieth century, published his axiom set for Eu-
clidean Geometry, it was shown that two of these axioms were, in fact, implied
by the others. These two dependent axioms were removed from the axiom set
and re-baptized as theorems, and this issue did not invalidate Hilbert’s entire
axiomatic system. One may also point out that the mathematician’s concern
for the independence of an axiom set is related to the earlier mentioned desired
“elegance” (or “minimality”) for a set of axioms and primitive terms, because
mathematicians always want to elaborate their theories with a minimum number
of fundamental assumptions.
The last property that is sometimes required of an axiom set is that of com-
pleteness. In the same fashion that independence of an axiom set insures that
the set does not assume redundant axioms, the property of completeness assures
that it is impossible to add a new independent axiom which is also consistent
with the original axiom set (assumed to be already consistent) and which does
not introduce a new primitive term. In other words, any proposition expressed
in terms of the original system must be either provable or disprovable if the
system is complete. For instance, the original system of Euclid was incomplete
because it contained—as we shall see in the next chapters—statements which
could neither be proved nor disproved. To determine whether a set of axioms
is complete, one might try over and over again to find new independent and
consistent axiom(s), and failing to do so, conclude that the system might be
complete. This is obviously no proof of completeness for a given axiom set, be-
sides being a truly demanding exercise. Models, however, can be invoked once
more to investigate the completeness of an axiomatic system. Two models for
30 CHAPTER 2. ON FORMAL AXIOMATIC SYSTEMS

the same formal axiomatic system are called isomorphic if there is a one-to-one
correspondence between the basic objects (elements and relations) of one model
and those of the other such that when a proposition involving the basic objects
of one model is true (or false), the corresponding proposition in terms of the
corresponding basic objects of the other model is equally true (or false). If an
axiomatic system is such that any two models for it are isomorphic, the system
is called categorical. It can be shown that categorical axiomatic systems are
complete. It should be clear that such a categorical axiom system possesses
essentially one model only, and that all other models for this system are mere
“avatars” of (i.e., are isomorphic to) that single one. In other words, categorical
axiomatic systems yield only one mathematical discourse. It can be proved that
the modern axiomatic frameworks of Euclidean and non-Euclidean Geometries
constitute categorical systems. In this sense, it may be argued that Euclid et
al. were not too remote from mathematical rigor when they used diagrams to
support their arguments since any other model for the Euclidean system would
have been isomorphic to these conventional figures made out of dots, straight
lines, circles, etc. In practice, categorical systems do not have a wide range
of applicability: they describe essentially one and only one “thing.” On the
other hand, mathematicians have often discovered that an incomplete formal
axiomatic system formulated with one particular model in mind has found ap-
plication to totally different realms of scientific investigation that they had not
thought of. For instance, one formal axiomatic system called Group Theory can
be applied to the solution of algebraic equations, to the classification of crystal
types, to the description of the interactions between elementary particles, and
much more.
We have seen that although the independence and completeness of an ax-
iomatic system may be aesthetically satisfying (and render the system more “el-
egant”), they are by no means necessary from the viewpoint of practicality. On
the other hand, definitions formulated in an axiomatic system should be both
independent and complete. For example, the definition of an equilateral triangle
as a triangle with three congruent (equal) sides AND three congruent (equal)
angles is not acceptable because these two properties are not independent, and
the definition of a square as a quadrilateral with four congruent (equal) sides is
not acceptable either because this property does not define a square completely
(visualize the rhombus).

2.2 Examples
We now illustrate the preceding concepts with a few, necessarily simple, exam-
ples of formal axiomatic systems. We begin with an example3 that shows how
theorems are proved.

3 Adapted from The Nature and Power of Mathematics, by Donald M. Davis, Princeton

University Press, New York, 1993

2.2. EXAMPLES 31

Formal Axiomatic System I

The primitive terms are “pflug” (noun, plural: pflugs), “lork,” (noun, plural:
lorks) and “to frax” (verb, fraxed, fraxed). The axioms are:

Axiom I.1. There is at least one pflug.

Axiom I.2. For any two distinct lorks, there is exactly one pflug that fraxes
both of them.

Axiom I.3. Each pflug fraxes at least two lorks.

Axiom I.4. For every pflug, there is at least one lork which it does not frax.

The words “pflug,” “lork,” and “to frax” do not exist in the English vo-
cabulary, and, as such, make no sense (“We don’t know what we’re talking
about . . . ”) but the axioms tell everything one needs to know about the prim-
itive terms in order to operate with them. In particular, pflugs and lorks are
“things”, and fraxing is “something” that pflugs do to lorks. Note that Axiom
I.1 allows that there might be several pflugs, and that Axiom I.2 does not guar-
antee that there are two distinct lorks. Axiom I.2 also permits the same pflug
to frax two different pairs of distinct lorks. We now prove:

Theorem I.1. There are at least three lorks.

Proof. By Axiom I.1, there is at least one pflug. By Axiom I.3, there are at
least two lorks that it fraxes. By Axiom I.4, there must be another lork that is
not fraxed by the pflug in question. Hence, there are at least three lorks.

Note that dropping Axiom I.1 while retaining the other three axioms might
lead to a vacuous system: no pflugs to frax any lork.
To formulate and subsequently prove more complicated theorems, it is use-
ful to devise a model for this axiomatic system in order to be able to picture
pflugs, lorks and relations between them, much in the same way that one draws
diagrams of triangles, circles, etc. in Euclidean Geometry to visualize these
objects; and in the same way as with geometrical figures, our model cannot
constitute a part of the proof of a theorem. We represent pflugs by the symbols
Pi , lorks by the symbol Lj , where the different values of the subscripts i, j, . . .
are 1, 2, 3, . . . and identify the distinct pflugs and lorks. We further represent
fraxing by an arrow connecting a pflug to a lork. We must now verify that
Axioms I.1 through I.4 become true properties under our interpretation of the
primitive terms. By Theorem I.1, we know that there must be at least three
lorks L1 , L2 , and L3 . By Axiom I.2, there is one and only one pflug P12 which
fraxes L1 and L2 , exactly one pflug P13 which fraxes L1 and L3 , and exactly
one pflug P23 which fraxes L2 and L3 . However, at this stage, we dont know
whether P12 and, say, P23 are distinct. This question leads to the formulation,
and subsequent proof of the following theorem:
32 CHAPTER 2. ON FORMAL AXIOMATIC SYSTEMS

Theorem I.2. If there are only three lorks, then there are exactly three distinct
pflugs.

Proof. We know that P12 exists. Axiom I.2 does not prohibit P12 to frax L3
too, but Axiom I.4 does, since there are no more (nor less) than three lorks.
Therefore, the pair of lorks L2 and L3 must be fraxed by a pflug different from
P12 , namely P23 . Similarly, Axiom I.4 implies the existence of a third, different
pflug P13 . Thus, we have at least three distinct pflugs. But there cannot be a
fourth distinct pflug because if there were, it would have to frax at least two
lorks by Axiom I.3. But then there would be two distinct pflugs fraxing the
same pair of lorks, in contradiction to Axiom I.2 . This completes the proof of
Theorem I.2.

We note in passing that the second half of the proof of Theorem I.2 is a
typical RAA argument: we assume that there is a fourth pflug and derive a
contradiction with one of the hypotheses of the theorem. It is also noteworthy
that models can equally serve as guides for avoiding false statements, for a
proposition that is false in some model of an axiomatic system cannot be a true
theorem of that system. But, on the other hand, a proposition that is true
in a model for an axiomatic system does not necessarily correspond to a true
theorem of that system. For instance, we can certainly consider a 3-lorks-model
for axiomatic System I consisting of exactly three lorks. The proposition “There
are exactly three lorks,” however, is not a valid theorem of System I as seen
from a 4-lorks-model: it is easily checked that all four Axioms I.1 through I.4
are fulfilled so that the 4-lorks-model contradicts the above proposition.
2.2. EXAMPLES 33

Formal Axiomatic System II

The primitive terms are X, Y , and “to cajole”. The axioms are:

Axiom II.1. There are exactly two distinct X’s.

Axiom II.2. There are exactly three distinct Y ’s.

Axiom II.3. Each X cajoles exactly two Y ’s.

Axiom II.4. No two distinct X’s cajole the same Y .

This axiomatic system is inconsistent. For instance, Axioms II.1, II.3 and
II.4 imply that there are at least four Y ’s, thus contradicting Axiom II.2 which
is thus both true (as an axiom) and false (as a consequence of the other axioms).
In the same fashion, each of the axioms of System II can be proved to be false
using the other three axioms.
Thus, it is sometimes easy to prove that an axiomatic system is inconsistent
by using logic alone. On the other hand, we saw that the consistency of an
axiomatic system can also be established through a real world model. The
following example illustrates this method once more:

Formal Axiomatic System III

The primitive terms are “hydrant”, “dog” and “to mark”. The axioms are:

Axiom III.1. There are at least two hydrants.

Axiom III. 2. For any two distinct hydrants, there is exactly one dog that
marks them.

Axiom III.3. No dog marks every hydrant.

Axiom III. 4. Given any dog D and any hydrant H that this dog does not
mark, there exists exactly one dog that marks the given hydrant H but does
not mark any of the same hydrants as dog D.

To check the (absolute) consistency of this system, we construct a model

using the same symbolic representations as in the models for system I (letters
and arrows for the primitive terms).Then, we know that there are at least two
hydrants H1 and H2 (Axiom III.1). Axiom III.2 ensures that there is exactly
one dog D12 (hence, there is at least one dog, so far) that marks both H1 and
H2 . Axiom III.3 then ensures that there must be at least a third hydrant H3
that dog D12 does not mark. But then, Axiom III.2 requires the existence of
34 CHAPTER 2. ON FORMAL AXIOMATIC SYSTEMS

two more dogs, D13 and D23 , to mark the distinct hydrant pairs {H1 , H3 } and
{H2 , H3 }. So far, Axioms III.1 through III.3 are satisfied in our model. To fulfill
Axiom III.4, we are forced to introduce a dog D3x that marks H3 but does not
mark either of the hydrants H1 and H2 , and similarly another dog D2x which
marks H2 but does not mark H1 and H3 , and a last dog D1x that marks H1
but does not mark H2 and H3 . Yet, our model is not completed because Axiom
III.4 is not fulfilled for dog D3x and for hydrant H2 , since there are two dogs
(D12 and D2x ) that mark H2 but do not mark any of the hydrants marked by
dog D3x . Thus, we are forced to introduce hydrant H4 , and let dogs D3x and
D2x mark it, thereby identifying dogs D34 and D24 .

But dog D34 and hydrant H1 still cause a problem because of Axiom III.4,
since hydrant H1 is marked by dogs D12 and D1x , neither of which marks any
of the same hydrants as dog D34 . Therefore, we must ask dog D1x to mark
hydrant H4 , as shown. Now, our model is completed, i.e. all axioms are verified
as “realizable” properties of a real life model of six distinct dogs and four distinct
hydrants. We may then conclude that Axiomatic System III is consistent.

In preparation for our forthcoming study of plane (that is, two-dimensional)

Euclidean Geometry, we examine in some detail a simplified version of a formal
axiomatic system called Incidence Geometry:
2.2. EXAMPLES 35

Formal Axiomatic System IV

The primitive terms are “point” (denoted by the letter P ) and “line” (denoted
by the letter l), and the relation “incidence” between a point and a line. We will
use the equivalent expressions “a point lies on a line” or “a line contains/passes
through a point” to mean “a point is incident with a line”. The axioms are:
Axiom IV.1. For every pair of distinct points, there exists a unique line that
contains these two points.
Axiom IV.2. For every line, there exist at least two distinct points incident
with that line.
Axiom IV.3. There exist three distinct points with the property that no line
passes through all three of them.
We first note that Axiom IV.3 implies that there are at least three distinct
points otherwise System IV would be vacuous. We also remark in passing that
Axiom IV.3 is the conjunction of the two statements “There exist three distinct
points” and “For every line, at least one of these three points does not lie on
that line”. Our rules of logic now tell us that each of these two statements is
true separately. Furthermore, Axioms IV.1 and IV.2 imply that there exists
at least one line, and that such line(s) contain points—at least two, possibly
more—and that all points do not lie on the same line. Next, some definitions
are in order. These explicit definitions may freely use the primitive terms of
the system but should be independent and complete, as explained earlier, and
should not introduce new primitive or undefined terms. We define:
Definition IV.1. Three points which lie on the same line are called collinear.
Note that this definition is compatible with the axioms and that Axiom IV.3
states that there exist (at least) three non-collinear points. The second half of
the previous sentence is obtained by a straightforward application of the rules
of logic developed in the preceding chapter. The first theorem of System IV can
now be stated:
Theorem IV.1. There exist at least three distinct lines.
Proof. By Axiom IV.3, there exist (at least) three distinct points P1 , P2 and P3
which are non-collinear. Then Axiom IV.1 implies that there exist unique lines
l12 , l13 , and l23 which contain the pairs of points {P1 , P2 }, {P1 , P3 } and {P2 , P3 }
respectively. Now, these three lines must be distinct by virtue of Axiom IV.3,
for if, for example, l12 and l13 were not distinct, they both would contain the
three assumed non-collinear points P1 , P2 , and P3 . This last RAA argument
completes the proof of the theorem.
Our next definition is:
Definition IV.2. Two distinct lines are called parallel if no point lies on both
of them.
36 CHAPTER 2. ON FORMAL AXIOMATIC SYSTEMS

One may now state the following—admittedly not very exciting—theorems

whose proofs are left as exercises:

Theorem IV.2. If l1 and l2 are distinct lines that are not parallel, then l1 and
l2 have a unique point in common.

Theorem IV.3. For every line there is at least one point not lying on it.

Theorem IV.4. For every point there is at least one line not passing through
it.

Theorem IV.5. For every point there exist at least two distinct lines passing
through it.

Our next definition is:

Definition IV.3. Three distinct lines are called concurrent if they contain one
point in common.

Theorem IV.6. Three concurrent lines have a unique point in common.

Theorem IV.7. There exist three distinct nonconcurrent lines.

We may wish to test the consistency of Incidence Geometry. To that end,

we devise a concrete model for the axiomatic System IV as follows: we interpret
the primitive term “point” as any member of the set of capital letters of the
Latin alphabet. Axiom IV.3 states that there are (at least) three such distinct
letters. Hence, we begin with only the three distinct letters A, B, and C.
“Lines” will be pairs of different letters: {A, B}, {A, C}, and {B, C}, with the
convention that the pairs {A, B} and {B, A} represent the same line (so-called
“unordered” pairs), etc. A “point” will be “incident with a line” if the letter
representing that point is a member of the pair of letters representing the line.
For example, point A is not incident with the line {B, C} but lies on line {A, C}.
We may now easily verify that the interpretations of Axioms IV.1 through IV.3
are correct statements. For instance, if P , Q are any two of the three letters A,
B, or C, then {P, Q} is the unique line that contains both letters P and Q, and
so on. The consistency of Incidence Geometry is thus established. Moreover,
since theorems are logical consequences of the axioms, and axioms, in turn,
are reflected as correct statements in the model, we conclude that Theorems
IV.1 through Theorem IV.7 (if proved to be valid) must correspond to true
statements when interpreted in our model. This verification is carried out easily.
For example, Theorem IV.5 is interpreted in the model by the statement that
if P is any of the three letters A, B, or C, then there exist two distinct pairs
(i.e. “lines”) {P, X} and {P, Y } where X and Y are not P , and X and Y are
different letters, which contain the letter P . This statement is evidently true in
our model. In fact, this “3-letter” model suggests a sharper version of Theorem
IV.5: since there are only two pairs {P, X} and {P, Y } with X, Y 6= P and
X 6= Y , we may consider proving
2.2. EXAMPLES 37

Theorem IV.5 bis. For every point there are exactly two distinct lines pass-
ing through it.
A moment of reflection, however, will show that we have devised a partic-
ular model containing exactly three “points” whereas the axioms guarantee
only the existence of at least three points. In this model, Theorem IV.5 bis will
be true, but we might equally have considered a model containing exactly four
letters (“points”) with the same interpretations of the primitive terms, and then
Theorem IV.5 bis would turn out to be NOT true, as can be seen easily. This
example reminds us that if one can find one model in which the interpretation
of a presumed valid theorem translates into an incorrect statement, then one
can be assured that no proof of that theorem is possible. This fact should be
compared with another one that is the source of the most common error among
early students of Geometry: one may intuitively test the validity of geometric
statements by drawing figures, but just because a drawing makes a statement
appear valid is not a proof for the correctness of that statement (“Believing
should not rely on seeing”). We have seen that the advantage of having several
models for a given formal axiomatic system resides in the fact that a statement
found correct in one model may very well fail to remain valid in another. Thus
models are useful operational tools (“laboratories”) for experimenting with for-
mal axiomatic systems.
To continue with the illustration of the properties of formal axiomatic sys-
tems and the preceding remarks, we observe that Incidence Geometry, as for-
mulated in System IV, is not a categorical (or complete) system, since it allows
for (at least) two non-isomorphic models, namely the “3-letters” model and the
“4-letters” model. We may, however, turn System IV into a categorical system
by adding one extra axiom (independent and consistent), for example:
Extra Axiom. There are at most three distinct points.
Then, along with Axiom IV.3, this system will contain exactly (no more, no
less) three points, and all its models will be isomorphic to the “3-letters” model
described above.
We may experiment further with models. One most instructive example
arises when we add the following axiom to those of Incidence Geometry:
Axiom IV.4. For every given line and every given point not lying on that line,
there exists a unique line passing through the given point that is parallel to the
given line.
This is the so-called Playfair version (after the 18th century British mathe-
matician John Playfair) of the already mentioned notorious Parallel Postulate
(or Fifth Postulate) of Euclid. In our “3-letters” model, it is immediately appar-
ent that no parallel lines exist at all. One says that this model has the elliptic
property. Thus, we conclude that no proof of the Euclidean Parallel Postulate
(i.e. Axiom IV.4) using the other axioms of Incidence Geometry is possible. In
the same vein, the converse statement “Any two lines contain a point in com-
mon” cannot be proved either from the axioms of Incidence Geometry alone,
38 CHAPTER 2. ON FORMAL AXIOMATIC SYSTEMS

because if one could prove it, it would reflect in a valid property of the model.
These considerations lead further to the conclusion that Axiom IV.4 is indeed
independent of the other axioms of Incidence Geometry, because it is impossible
to prove or to disprove it from the other three axioms. We have already seen
that Axiom IV.4 is not valid in the “3-letters” model. But if we consider now
the “4-letters” model introduced earlier, with exactly four letters A, B, C, and
D, and exactly six “lines,” namely the pairs {A, B}, {A, C}, {A, D}, {B, C},
{B, D} and {C, D}, then one can easily verify that the Euclidean Parallel Pos-
tulate (Axiom IV.4) does hold.
On the other hand, if one devises a “5-letters” model for Incidence Geometry
which contains exactly five “points” with the same interpretations of primitive
terms as the “3-letters” and “4-letters” models, then it can be verified that the
following statement about parallel lines will hold:
Axiom IV.4 bis. For every given line and every point not lying on that line,
there exist at least two lines containing the point and parallel to the given line.
One says that the “5-letters” model of Incidence Geometry has the hyperbolic
property.
In conclusion, Incidence Geometry can be extended to potentially very differ-
ent, more complex formal axiomatic systems by the addition of new independent
(and consistent) axiom(s), such as the parallel axiom of the Euclidean, elliptic
or hyperbolic types. Chapter Thirteen covers these matters in greater detail.
Finally, it is not too difficult to construct a model for each pair of the three
axioms of Incidence Geometry in which two of the axioms are verified but the
third is not. This establishes the independence of these axioms because it is
impossible to prove any one of them from the other two.
We conclude this chapter with a brief excursion into the realms of so-called
Finite Geometries. Incidence Geometry, as defined by the formal axiomatic
system IV, is called a Finite Geometry if its axiom set is complemented by an
additional axiom stating that the number of distinct points in the system is
equal to some finite integer. These “miniature” Geometries are instructive to
study because of their rather simple structures, in contrast to, say, Euclidean
Geometry which contains an “infinitude” of points and lines.
We have already implicitly encountered the 3-point Geometry as represented
categorically by our earlier “3-letters” model. Let us first consider the Four-Line
Geometry which is defined as an Incidence Geometry with exactly four “lines.”
The axioms are then:
Axiom 4-L.1. There exist exactly four lines.
Axiom 4-L.2. Any two distinct lines have exactly one point in common.
Axiom 4-L.3. Every point is on exactly two lines.
One can then prove that the Four-Line Geometry contains exactly six dis-
tinct points, and that each line passes through exactly three distinct points.
A new set of axioms can be formally generated from the axioms of the Four-
Line Geometry by substituting the word “point” for “line” and vice-versa, and
2.2. EXAMPLES 39

by carrying out the necessary changes in the English syntax. This procedure,
called plane dualization, yields the so-called Four-Point Geometry, and is defined
by the axioms:
Axiom 4-P.1. There exist exactly four points.
Axiom 4-P.2. Any two distinct points lie on exactly one line.
Axiom 4-P.3. Every line passes through exactly two points.
The plane duals of the theorems mentioned for the Four-Line Geometry
would then read: “The Four-Point Geometry contains exactly six distinct lines”
and “Each point of the Four-Point Geometry lies on exactly three lines”. There-
fore, there exist lines in the Four-Point Geometry that do not have one of the
four points in common. Hence, such lines are parallel. A simple model consist-
ing of dots to represent “points” and (not necessarily straight) lines connecting
such dots to represent “lines” shows that the Euclidean Parallel Postulate is
valid in the Four-Point Geometry, while the elliptic Parallel Postulate holds in
the Four-Line Geometry, as seen on the diagrams.

A somewhat less “trivial” Finite Geometry is the so-called Fano’s Geometry

(after the early 20th century Italian mathematician Gino Fano who studied
Finite Geometries extensively): Fano’s Geometry is constructed with the same
primitive terms (i.e. “point,” “line,” and “incidence”) as our earlier Incidence
Geometry but with more axioms, namely
Axiom F.1. There exists at least one line.
Axiom F.2. Every line passes through exactly three distinct points.
Axiom F.3. Not all points lie on the same line.
Axiom F. 4. For two distinct points, there exists exactly one line passing
through both points.
Axiom F.5. Every two lines have at least one point in common.
40 CHAPTER 2. ON FORMAL AXIOMATIC SYSTEMS

Although none of these axioms restricts explicitly the number of points (or
lines), we will nonetheless prove that Fano’s Geometry is finite. To this end,
and to illustrate once more the RAA method of proof described in Chapter One,
we begin with proving:

Theorem F.1. Every two lines have exactly one point in common.

Proof. By Axiom F.5, every two lines have at least one point in common. Sup-
pose that a given pair of lines has two distinct points in common (RAA hy-
pothesis). This assumption, however, violates Axiom F.4 because the two dis-
tinct common points would lie on two distinct lines. This establishes Theorem
F.1.

Next, we prove:

Theorem F.2. Fano’s Geometry contains exactly seven points and seven lines.

Proof. From Axioms F.1 and F.2, there are at least three points, while Axiom
F.3 requires the existence of at least one fourth point non-collinear with the
other three. To illustrate the further reasoning, it is convenient (although not
necessary) to have recourse to diagrams with dots representing “points” con-
nected together by “lines.” Thus, diagram (a) represents our current state of
knowledge. Now, by Axiom F.4, there must be lines containing this fourth point
and each of the other existing three, while Axiom F.3 requires the existence of
further points 5,6, and 7 on the lines containing points 1 and 4, 2 and 4, and
3 and 4, respectively, as shown on diagram (b). Finally, Axioms F.4 and F.5
require the existence of lines passing through points 1,6,7, and points 3,6,5,
and points 5,2,7, as represented on diagram (c). Thus, Fano’s Geometry must
contain at least seven points and seven lines.

To prove that this Geometry does not contain more than seven points and
seven lines, we revert once more to an RAA argument:
Assume there is an eighth point (RAA hypothesis), and consider, for in-
stance, the intersection of the unique line containing points 1 and 8 with the
line containing the points 3, 7, and 4, as indicated on the diagram. By Axiom
F.5, these two lines must have at least one point in common. This common
point, however, cannot be point 3, point 7 or point 4, since that would violate
Axiom F.4. Hence, that common point must be a ninth point, but that violates
2.2. EXAMPLES 41

Axiom F.2. Thus, the assumption of the existence of an eighth point leads to
a violation of the axioms, and must be rejected. The proof of Theorem F.2 is
now complete.

The foundations are now laid for the introduction of THE formal axiomatic
system par excellence, namely the system of Euclidean Geometry, and for the
systematic development of its superstructure.
42 CHAPTER 2. ON FORMAL AXIOMATIC SYSTEMS

Exercises
1. If p, q, r are statements, prove that the following set of four axioms is
inconsistent:

(a) If q is true, then r is false

(b) If q is false, then p is false
(c) r is true
(d) p is true.

2. Consider the following formal axiomatic system in which the primitive

terms are “bee” and “hive”, and “to contain”. The axioms are:

Axiom 1. Every hive contains at least one bee.

Axiom 2. Any two distinct hives have one and only one bee in common.

Axiom 3. Every bee belongs to two and only two hives.

Axiom 4. There are exactly four hives.

Show that this system is (absolutely) consistent.

3. For the formal axiomatic system of Exercise 2, prove the following theo-
rems:

Theorem 1. There are exactly six bees.

Theorem 2. There are exactly three bees in each hive.

Theorem 3. For each bee there is exactly one other bee not in the same
hive.

4. Prove that Axioms II.1, II.3 and II.4 of formal axiomatic system II are
inconsistent with Axiom II.2.

5. Use symbolic logic to express the axioms of Incidence Geometry.

6. Show that the Euclidean Parallel Postulate (Axiom IV.4 of Incidence Ge-
ometry) is a correct statement in the 4-point model for Incidence Geom-
etry.

7. Show that the hyperbolic Parallel Postulate (Axiom IV.4 bis) is a correct
statement in the 5-point model for Incidence Geometry.

8. Consider the following formal axiomatic system4 called “The Scorpling

Flugs”:
The primitive terms are “flug” and “to scorple.” The axioms are:
4 Adapted from The non-Euclidean Revolution, op. cit Ref. 2
2.2. EXAMPLES 43

Axiom SF. 1. If A and B are distinct flugs, then A scorples B or B

scorples A—the possibility of both happening is not excluded.
Axiom SF.2. No flug scorples itself.
Axiom SF.3. If A, B, and C are flugs such that A scorples B and B
scorples C, then A scorples C.
Axiom SF.4. There are exactly four flugs.

Prove the following theorems (use of diagrams to visualize the theorems

is “kosher”):

Theorem SF.1. If a flug scorples another, then it is not scorpled by the

other.
Corollary. Given two distinct flugs, either the first scorples the second
or the second scorples the first, but not both.

Theorem SF.2. If flug A scorples flug B and if flug C is distinct from

flug A, then A scorples C or C scorples B or possibly both.
Theorem SF.3. There is at least one flug that scorples every other flug.
Definition SF.1. A flug that scorples every other flug is called a pushy
flug.

Theorem SF.4. There is exactly one pushy flug.

Models for this formal axiomatic system can be devised by interpreting the
four flugs as the four integers 1, 2, 3 and 4, and the statement “a scorples
b” as the inequality a < b. Another model is obtained by interpreting the
four flugs as a man, his father, his father’s father, and his father’s father’s
father, and “a scorples b” would then mean “a is an ancestor of b.”
9. Invent/design your own Finite Geometry according to the following guide-
lines:

- It should have an evocative title.

- Primitive terms should be stated clearly.
- Axioms must be stated clearly.
- Axioms must be consistent.
- There should be a finite number of elements (either by virtue of an
explicit Axiom or—as in Fano’s Geometry—as a consequence of the
Axioms).
- Your Geometry should be neither trivial nor empty.
44 CHAPTER 2. ON FORMAL AXIOMATIC SYSTEMS
Chapter 3

On the Foundations of
Euclidean Geometry

3.1 First Axioms and Distance Measurements

Euclid’s Elements consists of 13 Books containing 465 propositions (or theo-
rems)1 . It opens with a list of 23 “Definitions” and introduces 5 “Axioms” (or
“Common Notions”) and 5 “Postulates”. We have already seen that Euclid’s
insistence in defining every technical term in his discourse led to a circularity
which can only be avoided by accepting to operate with primitive or undefined
terms. On the other hand, he seems to have made a distinction between his
“Common Notions” and his “Postulates” on the basis of his belief that Postu-
lates are purely geometric assumptions whereas Common Notions are equally
fundamental to other mathematical disciplines, such as arithmetic. Moreover,
in his elaboration of Geometry, Euclid makes specific use of certain tacit as-
sumptions which he did not bother to write down explicitly because they were
either “intuitively evident” (“seeing is believing”) to him or involuntary flaws
in his strict logical reasoning. Logical problems stemming from such unstated
assumptions include, among others:
- the need to state the fact that if a line enters a triangle at a vertex or a
circle at a point, it must intersect the opposite side of the triangle or the
circle at a second point;
- the need for a statement about the possibility to extend a line as needed;
- the need for a statement about the order of points on a line and the concept
of betweenness;
- the need for a purely geometric substitute for the concept of “superposi-
tion” in which geometric figures are moved from one place to another with-
out any deformation, i.e. with all their properties remaining unchanged.
1 See Appendix A or Ref. 1 of Chapter Two.

45
46 CHAPTER 3. ON THE FOUNDATIONS OF EUCLIDEAN GEOMETRY

Since the end of the 19th century, several modern axiom sets for a logi-
cally rigorous framework for Euclidean Geometry have been proposed by math-
ematicians such as Moritz Pasch, Giuseppi Peano, Oswald Veblen, and Garrett
Birkhoff, among others. The postulational treatment which has received the
widest acceptance was first proposed by David Hilbert in 18992 . Hilbert used
six primitive terms, namely “point,” “line,” “plane,” “incident on,” “between,”
and “congruent,” and grouped his axioms in five categories such that each cat-
egory addressed a class of logical flaws observed in Euclid’s arguments. For
instance, Hilbert’s axioms of congruence circumvent problems stemming from
Euclid’s questionable method of superposing figures to examine their congru-
ence. Consequently, Hilbert’s axiom set contains many more axioms than Eu-
clid’s. To illustrate this unavoidable increase in the number of axioms, consider
Euclid’s First Postulate which asserts, in essence, that “It is possible to draw
a straight line from any point to any point.” In order not to leave anything as
intuitively evident, Hilbert had to substitute three axioms for this single pos-
tulate to ensure the existence and the uniqueness of a line passing through
two points.
In general, the modern axiom sets proposed for a rigorous treatment of
Euclidean Geometry are far more abstract than Euclid’s original set, and it is not
appropriate to edify the entire Euclidean Geometry on such a purely geometrical
foundation in an introductory course at the high school level, even for inquisitive
students. The difficulty resides in having to invoke rather subtle and involved
arguments for proving the very first, painfully “evident” theorems—it is a fact
that theorems whose truth appears obvious are often quite arduous to prove.
Moreover, this approach might, in turn, obscure a vision of the “forest” for the
benefit of individual “trees.” In this course, instead, we will use an axiom set
which combines the ideas of Hilbert and Birkhoff, thereby making extensive
use of the properties of real numbers while providing a rigorous but simple
enough structure for use at the secondary school level. It should be noted
that it can be proved—but the proof is rather intricate—that Birkhoff’s two
basic axioms (the so-called Ruler-and-Protractor Axioms) although different in
their algebraic nature from the purely geometric Hilbert’s axioms, are actually
theorems of Hilbert’s formal axiomatic system3 . Thus, Birkhoff’s axioms and
Hilbert’s axioms lead to the same Geometry.
We proceed now with the step by step introduction of the primitive terms,
the set of axioms, first definitions and basic theorems which will constitute
the framework of our discourse4 . Since real numbers, their properties, and
the operations with them will be extensively used, it is imperative that the
reader maintains a solid command of the algebra of real numbers throughout
this course. This algebra is described in Appendix C. Throughout this text, the
end of a proof shall be indicated with the sign “ .”
2 Hilbert’saxioms are listed in Appendix B.
3 See,for example, Euclidean and Non-Euclidean Geometries, by Martin J. Greenberg,
W.H. Freeman & Co., New York, 1980, p. 99
4 Our axioms are a modified version of those proposed by the School Mathematics Study

Group in Geometry, Yale University Press, New Haven, 1960

3.1. FIRST AXIOMS AND DISTANCE MEASUREMENTS 47

The primitive terms are point, line, plane, and incidence. Congruence will
become a defined term.
The Axioms will be introduced gradually. We begin with the axioms of
plane Euclidean Geometry and will introduce those relevant to three-dimensional
space in later chapters.

Axiom 1. For any two distinct points there is exactly one line that is incident
with (also contains, passes through) both of them.

If P and Q are distinct points, the unique line that contains them will be
←→
denoted by P Q or sometimes by some other symbol such as l or L.

Axiom 2 (Distance Axiom). To every pair of distinct points there corresponds

a unique positive real number.

Definition 1. The distance between two points is the unique positive number
postulated by Axiom 2.

For the sake of convenience, we agree that the distance between a point and
itself is equal to zero. Notice that the notion of distance is defined for a pair of
points independently of their ordering, that is, of the order in which these two
points are mentioned. The distance between points P and Q will be denoted by
P Q or QP .

Axiom 3 (Ruler Axiom). For any line, there exists a one-to-one correspondence
(isomorphism) f between the points of that line and the real numbers such that:

(i) to every point P on the line there corresponds exactly one real number
f (P ),

(ii) to every real number r there corresponds exactly one point P on the line
with r = f (P ),

(iii) the distance between two points P and Q on the line is equal to the absolute
value of the difference of the real numbers r and s corresponding to the
two points: if f (P ) = r and f (Q) = s, then P Q = |r − s| = |s − r|.

While Axiom 1 implies the existence of a line if there are at least two points,
Axiom 3 guarantees that there are infinitely many points on every line. Axiom 2
allows for the choice of a unit for measuring distances between points according
to Axiom 3: select any two distinct points on any line and let their distance
be arbitrarily equal to one unit. Thus, Axioms 1, 2 and 3 together imply the
existence of a (straight) ruler with markings on it indicating the number scale.

Axiom 4 (Ruler Placement Axiom). Given two distinct points P and Q on a

line, it is possible to choose the one-to-one correspondence f between points on
the line and real numbers introduced in Axiom 3 in such a way that f (P ) = 0
and f (Q) > 0.
48 CHAPTER 3. ON THE FOUNDATIONS OF EUCLIDEAN GEOMETRY

It should be noted that Axioms 2, 3, and 4 together require the existence of

a universal isomorphism f between points on any line and the real numbers.
In other words, the distance P Q = |f (P ) − f (Q)| should not depend on which
particular line these two points happen to lie. The existence of this universal
one-to-one correspondence between points on any line and the real numbers is
sometimes considered to be a consequence of the somewhat more fundamental
assumptions of homogeneity and isotropy of space which essentially assert that
there is no “privileged” point among the infinitude of points on a line and that
there is no “privileged” line among the infinitude of lines in a plane or in 3-space.
A proof for the universality of this isomorphism in the Euclidean discourse will
be provided in the sequel.

3.2 Axioms of Betweenness

Now, Axiom 3(ii) guarantees the existence of many points on a line, and the
concept of “betweenness” which is strangely absent in the work of Euclid, and a
source for several of the logical flaws mentioned earlier, has to be introduced in
precise terms. For points lying on the same line, this concept can be stated on
the basis of intuition, but since “line” is a primitive term intuition alone may
lead astray. For example, it is easy to see that if three distinct points lie on
a closed curve, say a circle, then any one of them is between the other two or
none is between any two (depending on which arc one looks at). The following
definition ensures that a “line” is not “circular”, i.e. that the order of points
incident with a line is serial rather than cyclical:

Definition 2. Point B is between points A and C if

(i) A, B, C are distinct points incident with the same line, and

(ii) AB + BC = AC.

Needless to say, it follows from this definition that if point B is between A

and C then B is between point C and A. The next theorem uses this definition
to express the “order” of points on a line:

Theorem 1. Let P , Q, R be three distinct points on the same line, and let
p = f (P ), q = f (Q), and r = f (R) be the real numbers associated with these
three points. If p < q < r, then Q is between P and R.

Proof. The hypothesis p < q < r ensures that the real numbers q − p = f (Q) −
f (P ), r − q = f (R) − f (Q) and r − p = f (R) − f (P ) are positive. By definition
of absolute value: P Q = |f (Q) − f (P )| = |q − p| = q − p, QR = |f (R) −
f (Q)| = |r − q| = r − q, and P R = |f (R) − f (P )| = |r − p| = r − p so that
P R = (r − q) + (q − p) = QR + P Q, and that by Definition 2, Q is between P
and R.

The content of this first theorem is sharpened with the following theorem:
3.2. AXIOMS OF BETWEENNESS 49

Theorem 2. Given three distinct points incident with the same line, then
exactly one of these points is between the other two.
Proof. We only sketch the proof and leave the details to the reader as a straight-
forward exercise. First we show that one of the three points must lie be-
tween the other two by examining all six possibilities for ordering the three
real numbers p = f (P ), q = f (Q) and r = f (R) in increasing order, such
as p < r < q or r < q < p, etc., and applying Theorem 1 to each possibil-
ity. To prove the uniqueness of the point between the other two, we use an
RAA argument: suppose Q is between P and R and P is between Q and R.
Then by Definition 2, P Q + QR = P R and QP + P R = QR. The distances
P Q, QR, and P R being positive numbers, we have: P Q = P R − QR and
P Q = QR − P R = −(P R − QR) = −P Q. Since a real number cannot be both
positive and negative, it follows that P Q must be zero, implying that P and Q
are the same point. This conclusion, however, contradicts the hypothesis of the
theorem which requires three distinct points.
These admittedly elementary theorems clarify the notion of betweenness
rigorously and lead to
Definition 3. For any two distinct points P and Q, the segment P Q consists
of all points, including P and Q, that are between P and Q. The points P and
Q are called the endpoints of the segment P Q.
One remark is eminently important here: the segment P Q (a geometrical
object) and the distance P Q between its endpoints (an algebraic quantity) are
NOT the same entity, although they are related. Such careful distinctions will
be necessary throughout our discourse in order to avoid pitfalls similar to those
encountered in Euclid’s.
The preceding remark entails:
Definition 4. The length of segment P Q is the distance between its endpoints.
Next, we introduce the concept of ray which will be needed for the definition
of angles:
←→ −−→
Definition 5. Given two distinct points P and Q on line P Q, the ray P Q
←→
contains the segment P Q and all points R on line P Q for which it is true that
−−→
Q is between P and R. The point P of ray P Q is called the endpoint of the ray.
−−→
Definition 6. If point P is between points Q and R, then the rays P Q and
−→
P R are called opposite rays.
−−→
Theorem 3 (Point Plotting Theorem). Given ray P Q and the positive number
−−→
r, there exists exactly one point R on P Q such that P R = r.
Proof. By Axiom 4, we may assign the number zero to the endpoint P of ray
−−→
P Q and some positive number f (Q) to point Q. By Axiom 3(ii), there exists a
←→
unique point R on line P Q such that f (R) = r. Then P R = |f (R) − f (P )| =
|r − 0| = |r| = r since r is positive.
50 CHAPTER 3. ON THE FOUNDATIONS OF EUCLIDEAN GEOMETRY

An important particular point on a given segment P Q is the object of our

next definition:

Definition 7. A point M is called a midpoint of the given segment P Q if M

is between P and Q and P M = QM .

Definition 8. A line bisects line segment P Q if it intersects P Q at its midpoint.

Theorem 4. Every segment has a unique midpoint.

Proof. Based on Definition 2 and Theorem 3, on segment P Q we may choose

point M between P and Q such that P M + M Q = P Q and, moreover, P M =
QM = 21 P Q. This shows the existence of a midpoint M for the given segment
−−→
P Q. The uniqueness of M is guaranteed by applying Theorem 3 to ray P Q
1
with r = 2 P Q.

So far, we have been applying the first few axioms to discover the properties
of lines and relationships between points incident with one given line, but there
is no reason yet to accept that there are distinct lines in our Geometry. To
ensure that our Geometry is not linear (or one-dimensional), we must use the
primitive term “plane.” To begin with, and in compliance with Requirement 1
of Logic, we agree to define space as follows:

Definition 9. The collection of all points is called space.

Furthermore, we introduce two new notions:

Definition 10. Points P , Q, R, . . . are called collinear if they are all incident
with the same line.

Definition 11. Points P , Q, R, . . . are called coplanar if they are all incident
with the same plane.

To ensure that planes contain points (and hence, lines) and that space con-
tains planes (and hence, points and lines), we state:

Axiom 5.

(i) Every plane contains at least three non-collinear points.

(ii) Space contains at least four non-coplanar points.

This axiom entails immediately that our Geometry contains at least three
distinct lines and, since lines contain an infinitude of points, Axiom 1 ensures
that the plane containing the three non-collinear points contains an infinitude
of lines. In the same fashion, we shall conclude later that space contains an
infinitude of points, lines, and planes. For the time being, however, we must
clarify the relationship between the primitive terms “plane” and “line” and
“incidence of lines with a plane” with a new axiom:
3.2. AXIOMS OF BETWEENNESS 51

Axiom 6. If two distinct points are incident with a plane, then the line incident
with these two points is incident with that plane.
Intuitively, Axiom 6 ensures that if a segment of a line is contained in a
plane, then the entire line is contained in that plane and does not “wobble in
and out” of the plane.
In the same way that Axiom 1 states that two distinct points determine a
unique line and relates two primitive terms (point and line) together, the next
two axioms fulfill the same purpose for planes, lines and points:
Axiom 7. Any three distinct points are coplanar and any three non-collinear
points are incident with (determine) a unique plane.

Axiom 8. If two distinct planes intersect, then their intersection is a line.

A clever student might notice that Axiom 8 can be weakened. One need only
assume that if two planes intersect, then their intersection contains at least two
points. Axioms 1 and 6 will then guarantee the full statement of Axiom 8.
Alternatively, if we keep the full statement of Axiom 8, then we can change
Axiom 6 to a theorem. We leave this proof (using Axioms 8 and 5) to the
reader.
We will postpone further discussion of the axioms pertaining to lines and
planes in space to a later chapter and focus on the further development of plane
Euclidean Geometry, that is of relationships between points and lines which are
all incident with one single plane. At this stage, it has become imperative to
state the long due
Theorem 5. Two distinct lines can contain at most one point in common.
Proof. The proof is a straightforward consequence of Axiom 1 along with an
RAA argument and is left as an exercise.

Theorem 5 allows for the introduction of a technical “shorthand” term which

will be used extensively throughout this text: two distinct lines or distinct
segments intersect one another if they contain a point in common.
The next axiom, whose necessity was first recognized by Pasch, differs from
the axioms introduced so far since it involves points not all on the same line and
gives information about the plane as a whole. As mentioned earlier, Euclid failed
to acknowledge the importance of “order” (that is, betweenness) for points on a
line, and this omission led to serious lacunas in several of his proofs. Intuitively
speaking, that is, arguing again that “seeing is believing,” if two points lie on
different “sides” of a given line, then the line determined by these two points
by Axiom 1 must intersect the given line. Our axioms, however, cannot provide
a proof for this “obvious” fact, and one has no other option but to cast this
“evidence” in the form of a new axiom. To that end, we must first state precisely
what is meant by “side” of a line:
52 CHAPTER 3. ON THE FOUNDATIONS OF EUCLIDEAN GEOMETRY

Definition 12. Let l be a line, A and B distinct points that are not incident
with l. If segment AB contains no point incident with l, we say that A and B
are on the same side of l, and if segment CD intersects l, we say that C and
D are on opposite sides of l.

Note that the Law of the Excluded Middle (our former Rule 1 of Logic),
along with Theorem 5, imply that two distinct points are either on the same or
on opposite sides of a line, unless they are both incident with that line.
We may state now:

Axiom 9 (Plane Separation Axiom). For every line l and any three distinct
points A, B, C not incident with l, all lying in the same plane as l:

(i) if A and B are on the same side of l and B and C are on the same side
of l, then A and C are on the same side of l.

(ii) if A and B are on opposite sides of l and B and C are on opposite sides
of l, then A and C are on the same side of l.

It is convenient to call each side of a line l a half-plane bounded by l. Note

that with our Definition 12, the “dividing” line l does not belong to either
half-plane. We may now prove:

Theorem 6. Every line bounds exactly two half-planes and these half-planes
have no point in common.

Proof. For every line l, there exists a point A not incident with l (Axiom 5(i)).
There exists a point O lying on l (Axiom 3). Points A and O determine a unique
←→ ←→
line OA (Axiom 1). There exists point B on line OA such that O is between A
3.2. AXIOMS OF BETWEENNESS 53

and B (Reason?). Then A and B are on opposite sides of l (Definition 12), so

l has at least two sides. Now, let C be any point distinct from A and B, and
not incident with l. If B and C are not on the same side of l, then A and C
are on the same side of l (by Rule 1 of Logic and Axiom 9(ii)). Therefore, l
has exactly two sides, and if these two sides had a point C in common (RAA
hypothesis), then by Axiom 9(i), A and B would be on the same side of l, which
is a contradiction with our earlier result.

We have gone through the various steps of the argument leading to the
conclusion of Theorem 6 in detail, citing the reason which justifies each such
step, in order to exhibit the standard format of a rigorous proof. In most
textbooks, proofs are arranged in two columns with each numbered step in the
left-hand column and its justification in the right-hand column. Such a format
makes the analysis easier and, in particular, allows for a quick detection of errors
which might have crept in the argument sequence.
As a Corollary of Theorem 6, we may state that if two points lie in the
same half-plane determined by a given line, then the segment between these
two points lies entirely in the same half-plane and never intersects the line,
whereas if each point lies in a different half-plane, then the segment with these
two points for endpoints must intersect the given line.
We prove now the very important

Theorem 7 (Pasch’s Theorem). If A, B, C are distinct non-collinear points

and l is any line intersecting, say, segment AB in a point between A and B,
then l must also intersect either segment AC or segment BC. Moreover, if C
does not lie on l, then l does not intersect both segments AC and BC.

Proof. Either C is incident with l or it is not. If it is, then the Theorem holds:
line l intersects segments AC and BC at their endpoint C. Suppose now that
C is not incident with l. Since l intersects segment AB (hypothesis), then A
and B are on opposite sides of l (Definition 12). Since l has two sides (Theorem
6), point C must lie either on the same side of l as A or on the same side of l as
B (Axiom 9). If C is on the same side of l as A, then it is on the opposite side
from B and, consequently, l must intersect segment BC and does not intersect
segment AC. A similar argument holds when C is on the same side of l as B.
This completes the proof of Pasch’s Theorem.
54 CHAPTER 3. ON THE FOUNDATIONS OF EUCLIDEAN GEOMETRY

3.3 Angles and Triangles

We now introduce a new fundamental geometric object which, in a sense, plays
the dual role of a segment:
Definition 13. An angle with vertex A consists of a point A together with two
−−→ −→
(distinct) non-opposite rays AB and AC called the sides or legs of the angle
emanating from A.

−−→ −→
The angle formed by the non-opposite rays AB and AC with vertex A will
be denoted ]BAC or ]CAB or simply ]A. First notice that an angle ]BAC
can be equally defined by its vertex A and any two points D and E on its legs
−→ −−→ −→ −−→
AC and AB respectively, because by Definition 5, rays AC and AD are identical
−−→ −→
and so are rays AB and AE. Next, it is important to realize that an angle is
a “plane” object, entirely contained in a well-defined plane, namely the plane
determined by the three non-collinear points consisting in the angle vertex and
any two points incident with each of the two sides of the angle. This is an
immediate consequence of Axioms 7 and 6. Finally, it should be pointed out
that “straight” and “zero” angles are excluded from Definition 13.
−−→
Definition 14. If two angles ]BAD and ]CAD have a common leg AD and
−−→ −→
their other two legs AB and AC form opposite rays, these two angles are a
linear pair.

Definition 15. Let A, B, C be three non-collinear points. Then the points

lying on the segments AB, AC, and BC constitute the triangle ABC. The
points A, B, C are called the vertices of the triangle and the segments AB, AC,
BC are called the sides of the triangle.
A triangle will be denoted with the symbol 4ABC. It should be clear that a
triangle is also a “plane” figure entirely contained in the unique plane determined
3.3. ANGLES AND TRIANGLES 55

by the vertices (by virtue of Axioms 6 and 7). Furthermore, a triangle 4ABC
determines three angles ]BAC, ]CBA, and ]BCA which are called the angles
of the triangle 4ABC.
The concept of “betweenness” induces an order among points on a line. In
an analogous manner, points in a plane can be “ordered” as lying “inside” or
“outside” a closed plane figure, in particular, a triangle. The general definition
of the interior of a plane closed curve is quite intricate but that of an angle or
of a triangle is relatively straightforward:

Definition 16. Let ]BAC be an angle in plane Π. A point D lies in the

←→
interior of ]BAC if D is on the same side of line AC as point B and D is on
←→
the same side of line AB as point C. A point in plane Π which is not in the
−−→ −→
interior of an angle nor lying on one of the rays AB or AC lies in its exterior.

Thus, according to Definition 16, point E is not in the interior of angle

←→
]BAC because E and C are on opposite sides of line AB, and F is in the
←→
exterior of that angle because F is on the “wrong side” of both lines AB and
←→
AC.
−−→
Theorem 8. If a ray AB has its vertex A on line l but B does not lie on l,
−−→
then all points of ray AB, except for point A, lie on the same side of l as B.

Proof. We use an RAA argument:

−−→
Assume there is a point C on ray AB such that B and C are on opposite
sides of l. Then by the Corollary of Theorem 6, l must intersect segment BC.
←→
But since BC is contained in line BC, it follows from Theorem 5 that this
intersection of l with BC must be point A. Since A is distinct from B and C,
and A is a point of segment BC, it follows that A must be between B and C
−−→
(Definition 5). But this means that C is not on ray AB by the very definition
of a ray with vertex A, thus contradicting the RAA hypothesis.

The following theorems, each of which being eminently “evident,” will be

needed for the rigorous development of the foundations of Euclidean Geometry.
←→
Theorem 9. Consider angle ]ABC and a point P lying on line AC. Then P
is in the interior of ]ABC if and only if P is between A and C.
56 CHAPTER 3. ON THE FOUNDATIONS OF EUCLIDEAN GEOMETRY

←→
Proof. Suppose P on line AC is between A and C. Then by Theorem 8 applied
−→ ←→
to ray AC with line AB playing the role of line l, it follows that P and C are
←→ ←→
on the same side of BA. Similarly, P and A are on the same side of BC. By
Definition 16, P is in the interior of angle ]ABC.
←→
Conversely, suppose point P on line AC is in the interior of angle ]ABC.
←→
Then, by Definition 16, P and A are on the same side of line BC so C is not
←→
between A and P , and P and C are on the same side of line AB so A is not
between P and C. Therefore, P must be between A and C.

Theorem 10. Given angle ]ABC, let point D be between A and C and E be
−−→
a point on ray BC such that C is between E and B. Then each point F on ray
−−→
BD such that D is between B and F is in the interior of angle ]ACE.

Proof. By Theorem 9, D is in the interior of angle ]ABC so that A and D are

←→
on the same side of BC. Then Theorem 8 implies that points D and F which lie
−−→ ←→
on ray BD are on the same side of line BC too, so A and F are on the same side
←→ ←→
of CE. Now, line AC separates the plane into two half-planes (Axiom 9 and/or
←→
Theorem 6), and since segment BF intersects line AC at D, it follows that F
and B belong to different half-planes. Similarly, since segment BE intersects
←→
line AC at C, Axiom 9 and/or Theorem 6 imply that B and E lie in different
←→
half-planes. Consequently, E and F are on the same side of AC. By Definition
16, F is in the interior of angle ]ACE.

Theorem 11 (“Crossbar Theorem”). If point D is in the interior of angle

−−→
]ABC, then ray BD intersects segment AC.
3.4. AXIOMS FOR ANGLE MEASUREMENTS 57

←→
Proof. Let E be a point on line AB such that B is between A and E (i.e. E
−−→
is on the opposite ray of BA). Let Π1 and Π2 be the two half-planes bounded
←→
by line BC. Since D is in the interior of angle ]ABC, then A and D are on
←→
the same side of BC, say, in the half plane Π1 . By Axiom 9, E is then in the
−−→
half-plane Π2 . By Theorem 8, each point of ray BD, except the endpoint B,
lies in Π1 , but no point of segment CE does. Hence, segment CE does not
−−→ ←→
intersect ray BD, since B is not incident with EC. On the other hand, each
←→
point of segment EC, except point E, lies on the same side of line AB as points
−−→
C and D, but each point of the ray opposite to ray BD, except point B, lies on
←→ −−→
the opposite side of AB. Hence, the ray opposite to ray BD fails to intersect
segment EC too. However, Pasch’s Theorem (Theorem 7) applied to points A,
←→
E, C asserts that line BD must intersect either segment AC or segment EC.
←→
Consequently, line BD must intersect segment AC. Notice that each point of
←→
segment AC other than A lies on the same side of line AB as points C and
−−→
D (Theorem 8 and Axiom 9), but each point of the ray opposite to ray BD,
←→
except for point B, lies on the other side of AB. Hence, the ray opposite to ray
−−→
BD fails to intersect segment AC. The proof of the Crossbar Theorem is now
complete.
Definition 17. A point lies in the interior of a triangle if it lies in the interior
of each angle of the triangle. A point in the plane of the triangle but not lying
in the interior of the triangle nor on any of its sides is said to be lying in the
exterior of the triangle.
Clearly, Theorems 9, 10, and 11 can be adapted to express the relationships
between a line and a triangle as well. We shall use such versions of these
theorems frequently as we proceed.

3.4 Axioms for Angle Measurements

Having defined the concept of angle, it is now necessary to develop the tools
for comparing different angles in the same quantitative fashion that we can
compare segments by means of their lengths. It is well-known that angles can
58 CHAPTER 3. ON THE FOUNDATIONS OF EUCLIDEAN GEOMETRY

be measured with a protractor in the same way that lengths of segments can be
measured with a marked straightedge (ruler). Using a protractor, the measure
of an angle is the number of degrees (and fractions of degree) indicated by the
protractor. But protractors are concrete, material objects which do not have an
immediate bearing with our abstract definition of angles in terms of rays which
are, in the last analysis, defined in terms of primitive terms that have no obvious
concrete reality. Therefore, we must introduce further axioms concerning the
measurement of angles. Hence:

Axiom 10 (Angle Measurement Axiom). To every angle there corresponds a

real number strictly between 0 and 180.

Notice that according to Definition 13 and Axiom 10, there is no angle

whose measure is zero or 180: an angle whose measure is zero would involve two
identical rays with the same vertex, and one with measure 180 would require
two opposite rays. This, however, does not preclude the possibility of two angles
having the sum of their measures equal to 180 (see Axiom 12). Moreover, in the
same fashion that the unit for length measurement (Axioms 2 and 3) was left
unspecified—it can be one inch or one kilometer—our unit for measuring angles
could have been left a priori arbitrary. For historical reasons, however, the unit
for angle measurement, the degree of arc or simply the degree, is stipulated in
Axiom 10 which asserts that the measure of an angle has to be a real number
between 0 and 180. We shall indicate the measure of angle by writing either
m ]A = n or m ]A = n◦ .
−−→
Axiom 11 (Angle Construction Axiom). For a given ray AB and a half-plane
←→
on a given side of line AB and a real number r between 0 and 180, there ex-
−→
ists exactly one ray AC in the given half-plane with vertex A, and such that
m ]BAC = r.
−−→
Definition 18. If point D is in the interior of angle ]BAC, then ray AD is
−−→ −→
said to be between rays AB and AC.

Axiom 12 (Angle Addition Axiom). If D is in the interior of angle ]BAC,

−−→ −−→ −→
then the ray AD between rays AB and AC determines two angles ]BAD and
]DAC such that m ]BAC = m ]BAD + m ]DAC.
3.5. CONGRUENCE OF ANGLES 59

Definition 19. Two angles are supplementary if the sum of their measures is
180.

Axiom 13 (Angle Supplement Axiom). If two angles form a linear pair, then
they are supplementary.

Definition 20. A right angle is an angle with measure 90. Angles with measure
less (greater) than 90 are called acute (obtuse).

Definition 21. Two lines l and m are called perpendicular if they intersect at
−−→ −→
point P and if there is a ray P Q that is part of l and a ray P R that is part of
m such that ]QP R is a right angle.

Perpendicular lines l and m will be denoted l ⊥ m. Clearly, this definition

can be adapted to the perpendicularity of intersecting rays or segments. We
also have the following simple theorem that follows from Axiom 13. We include
it here for convenience even though it depends on definition 23 below.

Theorem 12. Line l is perpendicular to line m if and only if they meet at

congruent adjacent angles.

Definition 22. If the sum of the measures of two angles is 90, then the angles
are called complementary, and each is called a complement of the other.

3.5 Congruence of Angles

Definition 23. Angles having the same measure are called congruent angles.
Thus, angles ]ABC and ]A0 B 0 C 0 are congruent if m ]ABC = m ]A0 B 0 C 0 and
denoted by ]ABC ∼= ]A0 B 0 C 0 .

It is of the utmost importance to clearly distinguish between the notion of

“congruence” (denoted with the symbol ∼ =) and that of “equality” (denoted by
=). The relation m ]ABC = m ]A0 B 0 C 0 expresses the equality of two num-
bers, that is two algebraic quantities whereas the expression ]ABC ∼ = ]A0 B 0 C 0
describes a relationship between two purely geometric objects. Definition 23
provides a means for ordering such objects and comparing them. It should also
be noted that, in the light of this remark, the expression ]ABC = ]P QR
asserts that ]ABC and ]P QR are one and the same angle.
Axioms 10 through 13, along with Definition 23, lead to the following theo-
rems whose proofs are left as straightforward exercises:

Theorem 13 (Reflexivity of Congruence). Every angle is congruent to itself.

Theorem 14 (Symmetry of Congruence). If ]α is congruent to ]β, then ]β

is congruent to ]α.

Theorem 15 (Transitivity of Congruence). If ]α ∼

= ]β and ]β ∼
= ]γ, then
]α ∼
= ]γ.
60 CHAPTER 3. ON THE FOUNDATIONS OF EUCLIDEAN GEOMETRY

Theorem 16. All right angles are congruent.

Theorem 17. Each of two complementary angles is acute.
Theorem 18. If two angles are both supplementary and congruent, then both
are right angles.
Theorem 19. Supplements of congruent angles are congruent.
Theorem 20. Complements of congruent angles are congruent.
Definition 24. If two angles are such that their legs are pairs of opposite rays
with the same endpoint (or vertex), then these angles are called vertical or
opposite angles.

In the diagram, the angle pairs ]1 and ]3, and ]2 and ]4, are pairs of
opposite angles.
Theorem 21. Vertical angles are congruent.

Proof. Consider, say the vertical angles ]CAB and ]DAE. By Definition 24,
−→ −→ −−→ −−→
rays AC and AE are opposite and rays AB and AD are opposite. Then angles
]DAE and ]CAD are supplementary and so are angles ]DAE and ]BAE.
By Theorem 13, angle ]DAE is congruent to itself, and by Theorem 19, angles
]CAD and ]BAE are congruent, and thus have congruent supplements. It
follows then that angles ]DAE and ]CAB are congruent.

3.6 Congruence of Triangles

By now, it must have become clear that the building blocks of the Euclidean—
and, equally, for any geometric—edifice are lines (or segments or rays) and
3.6. CONGRUENCE OF TRIANGLES 61

angles. Triangles (see Definition 15) are the simplest constructs made from these
building elements, and we are naturally led to examine relationships between
such objects.
To begin with, since the definition of triangle involves segments, it is neces-
sary to introduce the notion of congruence of segments:
Definition 25. Two segments are congruent if they have the same length.
If segments AB and A0 B 0 have the same length, that is, if AB = A0 B 0 , then
the congruence of segments is denoted by AB ∼ = A0 B 0 .
It should be stressed once more that equality of geometric entities (here,
segments and angles) means that these objects are exactly the same. Otherwise,
one must revert to the notion of congruence which, in turn, is defined in terms
of equality of measurements of these entities. One immediate consequence of
the definition above is:
Theorem 22 (Reflexivity of Congruence). Every segment is congruent to itself.
Proof. Left as an exercise in semantics.
Equally “trivial” are:
Theorem 23 (Symmetry of Congruence). If AB ∼ = CD, then CD ∼
= AB.
∼ CD and CD ∼
Theorem 24 (Transitivity of Congruence). If AB = = EF , then
∼ EF .
AB =
Consider now two triangles 4ABC and 4XY Z. In order to relate/compare
these two triangles, it is natural to try to match up their vertices in order to
establish a correspondence between their sides and angles which will allow to
compare the triangles with the help of earlier axioms. Thus, the matching-up of,
say, vertices A ↔ Y , B ↔ Z, and C ↔ X leads to unambiguous correspondences
between the angles and sides of the two triangles: ]A ↔ ]Y , ]B ↔ ]Z, ]C ↔
]X, and AB ↔ Y Z, AC ↔ Y X, BC ↔ ZX. Needless to say, each different
match-up of vertices leads to different correspondences among the angles and
sides of the triangles concerned. We may now define the congruence of two
triangles:
Definition 26. Two triangles are said to be congruent if there exists a match-
ing of their vertices such that every pair of corresponding angles and every
pair of corresponding sides are congruent. We say 4ABC ∼ = 4XY Z if the
correspondence is such that A ↔ X, B ↔ Y , and C ↔ Z
Notice that congruence of two triangles implies six congruences between geo-
metric objects—three pairs of sides and angles—and, equivalently, six equalities
between algebraic quantities—lengths and angle measures. It is straightforward
to see that the properties of reflexivity, symmetry, and transitivity which hold
for the congruence of segments and angles are automatically transferred to the
congruence of triangles.
Now, according to Definition 26, to examine if two triangles are congruent
requires finding a matching-up of vertices which leads to the above-mentioned
six congruences. The following axiom simplifies this search considerably:
62 CHAPTER 3. ON THE FOUNDATIONS OF EUCLIDEAN GEOMETRY

Axiom 14 (S.A.S. Axiom). Given two triangles, if there is a correspondence

between their vertices such that two sides and the included angle of one triangle
are congruent to the corresponding sides and the included angle of the other,
then the two triangles are congruent.

This is the congruence that Euclid tried to prove by—reluctantly—moving

one triangle onto the other so that the corresponding congruent elements were
superposed, thus “proving” the coincidence and, hence, the congruence of the
remaining pair of sides and two pairs of corresponding angles. This argument
derives from the experience of drawing two triangles on, say, paper, cutting
out one and trying to place it on top of the other one to demonstrate their
congruence or lack thereof. This is, however, no proof in the framework of the
Euclid’s axiomatic system, because none of his postulates or common notions al-
lows geometric figures to be moved around without changing their shape or size.
Identifying (abstract) geometric objects with rigid bodies introduces, in turn, a
concept borrowed from Mechanics, which is a branch of Applied Mathematics,
and as such, unacceptable in pure mathematics.

Notice that Axiom 14 requires the existence of a particular correspon-

dence between the vertices of the given triangles with the stated property.
The concluded congruence of the triangles is then valid under this particular
matching-up. Such a correspondence is called a congruence for the triangles
concerned. Furthermore, in applying Axiom 14, it is very important that the
angles appearing in the statement of this axiom are those included by pairs of
congruent sides. If in two triangles only two pairs of corresponding sides and
a pair of corresponding angles are congruent, the triangles are not necessarily
congruent. as illustrated in the diagram.
3.6. CONGRUENCE OF TRIANGLES 63

A well-known first application of Axiom 14 is the elegant proof of Pappus

(ca. 300 A.D.) concerning the base angles of an isosceles triangle:
Definition 27. A triangle with two congruent sides is called isosceles. The
third side is called the base, and the angles whose legs are the base and each of
the two congruent sides are called the base angles of the triangle.
Theorem 25. If in triangle 4ABC the sides AB and AC are congruent: AB ∼
=
AC, then the base angles ]B and ]C are congruent: ]B ∼
= ]C.

Proof. We establish a correspondence of vertices between the given triangle and

itself: A ↔ A, B ↔ C and C ↔ B. Under this correspondence, we obtain
the correspondence of two pairs of sides and their included angles: AB ↔ AC,
AC ↔ AB, ]A ↔ ]A. By hypothesis, AB ∼ = AC and AC ∼ = AB (by Theorem
23) and ]A ∼= ]A (by Theorem 13). By Axiom 14, it follows then that 4ABC ∼ =
4ACB, so that BC ∼ = BC (which is trivial) and ∠B ∼ = ∠C which was to be
proved.
We can now provide one proof for our earlier assertion that the isomorphism
between points on lines and real numbers as stated in Axiom 3 must be universal
in the framework of (plane) Euclidean Geometry developed so far:
Suppose that we have two different isomorphisms between real numbers and
points on different lines, but that these isomorphisms are consistent with the
axiomatic system consisting of Axioms 1 through 14. This means that it must
not be possible to derive a theorem within this framework that contradicts
an axiom or one of the theorems proved so far. Consider now two triangles
4ABC and 4XY Z which are congruent by virtue of Axiom 14 according to
←→
AB ∼ = XY , AC ∼ = XZ, and ]A ∼ = ]X. Suppose further that the lines BC
64 CHAPTER 3. ON THE FOUNDATIONS OF EUCLIDEAN GEOMETRY

←→
and Y Z are equipped with isomorphisms f and g respectively which assign
different lengths to the segments BC and Y Z (see Axioms 2, 3, and 4). It
would then follow that sides BC and Y Z of the congruent triangles 4ABC and
4XY Z would not be congruent (see Definition 25), that is a contradiction with
Axiom 14 and Definition 26. We may therefore conclude that (plane) Euclidean
Geometry as laid out by means of Axioms 1 through 14 allows for essentially—
that is, up to the choice of the unit or scale—one single distance measurement
between points. In the same vein, we may assert that angle measurements are
equally uniquely determined up to the choice of the unit for angle measure: one
can equally restate Axiom 10 by requiring that the measure of angle be a real
number between zero and 200 or between zero and the number π = 3.14159 . . .
without creating a contradiction with the accepted axioms or proven theorems.
We close this chapter with the conclusion that allowing the distance and
angle measurements to depend on the “location” of points and lines in space
opens the door to a wealth of new geometries, but that is the realm of Differential
Geometry, and is of no concern to us here.
3.6. CONGRUENCE OF TRIANGLES 65

Exercises
1. Let f be an isomorphism between the points on a line and the real num-
bers, as required by Axiom 3, so each point P on line l corresponds the
unique real number f (P ). Let g be another isomorphism, defined by
g(P ) = −f (P ) + 3. Show that if Axiom 3 is valid with the isomorphism
f , it is equally valid with the isomorphism g.

2. How does Axiom 3 imply that a line is of infinite extent?

3. Towns A, B and C are collinear but not necessarily in this order. A and
B are 16 miles away from one another and B and C are distant 25 miles.

(a) Is it possible to determine which town is between the other two?

Which town is not between the other two?
(b) What is the distance between A and C?

4. Points A, B, C are fixed collinear points. Three different sets of real num-
bers are assigned to these points by three different functions f satisfying
Axiom 3: with one function, f1 (A) = −6 and f1 (B) = −2; with a second
function, f2 (A) = 4 and f2 (C) = −3; with a third function, f3 (B) = 4 and
f3 (C) = 7. Which point is between the other two? Evaluate the length
AB + BC + AC.

5. Suppose we defined the distance between points P and Q on a line by

P Q = [f (P ) − f (Q)]2 instead of the usual P Q = |f (P ) − f (Q)|. How
would this new definition of the length of a segment affect the contents of
Theorems 1, 2, 4?

6. Prove Theorem 5

7. Suppose we chose Theorem 5 as an Axiom and deleted Axiom 1 from our

set of Axioms. Can one then prove that two points determine a unique
line?

8. Consider the points on a line to which the following sets of real numbers
x correspond:

(a) x < 3
(b) x = 1
(c) x > 2
(d) x ≤ 1
(e) x = −3
(f) |x| ≤ 3
(g) |x| > 2
(h) |x| ≥ 0
66 CHAPTER 3. ON THE FOUNDATIONS OF EUCLIDEAN GEOMETRY

Which of the above sets constitute a ray? A point? A line? A segment?

9. Criticize the statement: “The top of the table is a plane.”

10. Can a point be in the interior of an angle of a triangle and in the exterior
of the triangle? Can a point be in the exterior of a triangle and not in the
interior of any angle of the triangle?

11. Given 4ABC, and a point P in the interior of angle ]BAC and also in
the interior of angle ]ACB. What can one conclude about point P ?

12. Consider a triangle 4ABC and a point P . P and A are on the same side
←→ ←→
of line BC, and P and B are on the same side of line AC.

(a) Is P in the interior of angle ]ACB?

(b) Is P in the interior of triangle 4ABC?

13. Prove the following (without using Pasch’s Theorem):

Theorem. If a line l intersects two sides of a triangle 4ABC in points D

and E, not the vertices of the triangle, then l does not intersect the third
side.

←→ ←→
14. Is the following statement always true? “If two lines AB and CD intersect
at point O, then ]AOC ∼ = ]BOD”?

15. Define a circle.

←→
16. Prove that in triangle 4ABC, line AB has no point in common with the
triangle except those on segment AB. Use an RAA argument.

17. Given two triangles 4ABC and 4P QR. Suppose that the triangles are
congruent under the following two different correspondences between their
vertices: A ↔ P , B ↔ Q, C ↔ R and A ↔ P , B ↔ R, C ↔ Q. What
can be said of triangle 4ABC?

18. If one of two supplementary angles has a measure of 50 more than the
measure of the other, what are the measures of these angles? (Write an
equation and then solve.)

19. The measure of an angle is five times that of its complement. What is the
measure of this angle? (Be sure to write an equation for this one, too.)

20. Is there a point in the plane of a triangle such that the point is neither in
the exterior nor the interior of the triangle and neither in the interior nor
the exterior of any of its angles?
3.6. CONGRUENCE OF TRIANGLES 67

21. Is the following statement equivalent to the Angle Construction Axiom

(Axiom 11)?
−−→
Given a ray AB and a number n between 0 and 180, there is exactly one
−−→
ray AX such that m ]XAB = n.
JUSTIFY your answer!
68 CHAPTER 3. ON THE FOUNDATIONS OF EUCLIDEAN GEOMETRY
Chapter 4

On Neutral Geometry I

We have introduced fourteen axioms for (plane) Euclidean Geometry so far.

In this chapter, a number of important and useful theorems will be proved
rigorously using the axioms of distance and angle measurements, betweenness,
and congruence. We will also begin the study of three-dimensional (or spatial)
Euclidean Geometry, and will introduce additional axioms necessitated by this
increase of dimension.
The corpus of theorems which can be proved using these axioms alone consti-
tutes the so-called Neutral Geometry. Such results remain valid independently
of the axiom of parallelism that can be chosen in addition to those already in-
troduced, be it Euclidean, elliptic, or hyperbolic. In other words, by developing
Neutral Geometry, one clarifies the import of a parallel postulate by identifying
the role played by this axiom in the geometric edifice. A surprisingly large num-
ber of theorems can be proved in Neutral Geometry, and in so doing, a solid
experience of the techniques of geometric proof can be acquired. In carrying
out the proofs of these theorems, figures and drawings will be used freely as a
visual aid for the information available and for suggesting steps in the logical
argument that might lead to the desired conclusion. Under no circumstance
can an observation taken from a figure be considered a justification for a logical
step in the argument: as mentioned earlier (see Chapter One), only those six
reasons enumerated are acceptable for a mathematical proof. Stated differently,
all valid geometric proofs should remain independent of figures and drawings,
and Geometry should remain “the art of reasoning correctly according to [po-
tentially] erroneous drawings.” It should be borne in mind constantly that a
diagram may be even slightly inaccurate or depict merely a particular case (e.g.,
drawing two isosceles triangles to illustrate the S.A.S. Axiom), and suggest a
fallacious argument. In due course, examples for blatantly wrong conclusions
drawn from erroneous figures will be studied for the sake of a forceful realization
of the dangers lurking in diagrams.

Definition 28. A triangle whose three sides are congruent is called equilateral.
A triangle whose three angles are congruent is called equiangular. A triangle

69
70 CHAPTER 4. ON NEUTRAL GEOMETRY I

with no two congruent sides is called scalene.

Using Theorem 25, it is an easy exercise to prove the following Corollary:

Corollary. Every equilateral triangle is equiangular.

−−→ −−→
Definition 29. A ray AD bisects or is a bisector of angle ]CAB if AD is
−−→ −→
between rays AB and AC (see Definition 18) and if ]BAD ∼= ]CAD.

−−→
Clearly, if AD bisects angle ]BAC, then m ]BAD = m ]CAD = 12 m ]BAC.

Theorem 26. Every angle has a unique bisector.

Proof. Given angle ]A, points B and C can be determined on each leg in such
a way that AB = AC (by virtue of Theorem 3). Let D be the midpoint of
segment BC: DB = DC (by virtue of Theorem 4). From Theorem 25, it
follows that ]ABD ∼ = ]ACD. The triangles 4ABD and 4ACD are then
→
−
congruent (S.A.S.), so that ]BAD ∼ = ]CAD, and ray A D bisects angle ]A.
This establishes the existence of an angle bisector. To prove its uniqueness, we
−→ −−→
use an RAA argument. Suppose there exists ray AE distinct from AD which
1
equally bisects angle ]A. Then we have: m ]EAB = m ]DAB = 2 m ]A. By
Axiom 11 (Angle Construction Axiom) applied in the half-plane of angle ]A
←→ −−→ −→
with line AB as edge, it follows that rays AD and AE must be one and the
same ray. The proof is now complete.

Definition 30. An angle bisector of a triangle is a segment whose endpoints

are one triangle vertex and the intersection of the bisector of this angle with the
opposite side.
4.1. CONGRUENCE OF TRIANGLES 71

Notice that this definition relies heavily on Theorem 11 (Crossbar Theorem)

and, hence, on Axiom 9 without which it would not have been possible to
prove (using the other axioms alone) that the bisector of an angle of a triangle
does indeed intersect the opposite side and that this point lies between the
vertices of the triangle. The diagrams shown illustrate the fact that “line” (and
consequently, “ray”) is a primitive term and that one should not let oneself be
guided by one’s sole visual intuition.

Definition 31. A median of a triangle is a segment whose endpoints are a

vertex of the triangle and the midpoint of its opposite side.

It is easily recognized that each angle bisector and each median of a triangle
is entirely contained in the interior of the triangle.

4.1 Congruence of Triangles

We can now prove a new criterion for the congruence of two triangles, known
as the Angle-Side-Angle (A.S.A.) Theorem:

Theorem 27 (The A.S.A. Theorem). Given two triangles, if there is a corre-

spondence between their vertices such that two angles and the included side of
one triangle are congruent to the corresponding angles and included side of the
other triangle, then the two triangles are congruent (the correspondence is a
congruence).

Proof. Given two triangles 4ABC and 4DEF such that ]A ∼ = ]D, ]B ∼ = ]E,
and AB ∼= DE. For a change, we present the RAA argument in the two-column
format as follows:
72 CHAPTER 4. ON NEUTRAL GEOMETRY I

Statements Reasons
1. If AC ∼
= DF , then 4ABC ∼ = S.A.S. (Axiom 14)
4DEF
2. Suppose AC 6= DF RAA hypothesis
3. There is a unique point X on Theorem 3
−−→
DF such that DX = AC
4. 4ABC ∼ = 4DEX S.A.S. (Axiom 14)
5. ]ABC ∼= ]DEX Corresponding parts of congruent
triangles
6. ]ABC ∼ = ]DEF Given
7. ]DEX ∼ = ]DEF Steps 5 and 6
−−→ −−→
8. Rays EF and EX are the same Step 7 and Axiom 11
←→
9. Point X and F are the same Theorem 5 applied to lines EF and
←→
DF
10. 4ABC ∼
= 4DEF Steps 4 and 9.

Theorem 28. If two angles of a triangle are congruent, the triangle is isosceles,
with the two sides opposite the congruent angles congruent.
Proof. Given 4ABC with, say, ]B ∼ = ]C. Consider again the congruence
A ↔ A, B ↔ C, and C ↔ B. Now, BC = CB (identity) and ]B ∼ = ]C
(given). Hence, 4ABC ∼ = 4ACB by Theorem 27 (A.S.A.) and it follows that
AB ∼= AC as corresponding parts of congruent triangles.
Theorem 29. Every equiangular triangle is equilateral.
Proof. Given triangle 4ABC with ]A ∼ = ]B ∼ = ]C. The sides opposite angles
]B and ]C are congruent by Theorem 28: BC ∼ = AC. Another application of
Theorem 28 to another pair of congruent angles leads to the congruence of their
opposite sides. Thus the triangle is equilateral.
The next criterion for congruence of two triangles is the following:
Theorem 30 (The S.S.S. Theorem). Given two triangles, if there is a cor-
respondence between their vertices such that all three pairs of corresponding
sides are congruent, then the triangles are congruent (the correspondence is a
congruence).
4.1. CONGRUENCE OF TRIANGLES 73

Proof.
Statements Reasons
−→
1. There exists ray AG such that Axiom 11 and Axiom 9
]CAG ∼ = ]EDF and B and G are
←→
on opposite sides of line AC
2. There exists a point E 0 on ray Theorem 3
−→
AG such that AE 0 ∼ = DE
3. 4DEF ∼ = 4AE 0 C S.A.S.
Note: So far, we have only “duplicated” triangle 4DEF on the other side
←→
of line AC, using the S.A.S. Axiom.
4. AB ∼= AE 0 Hypothesis and Step 2
∼
5. BC = CE 0 Hypothesis and Step 3
←→
6. Segment BE 0 intersects line AC Step 1 and Separation Axiom 9
in a unique point H
There are now four possibilities: point H may lie between A and C, or
coincide with A, or coincide with C, or A comes to lie between H and C (or C
comes to lie between H and A).
Suppose H is between A and C, as shown above. We then proceed:
7. ]ABH ∼ = ]AE 0 H Step 4 and Theorem 25
8. ]CBH ∼ = ]CE 0 H Step 5 and Theorem 25
9. m ]ABH + m ]CBH = Axiom 12
m ]ABC
10. m ]AE 0 H + m ]CE 0 H = Axiom 12
m ]CE 0 A
11. ]ABC ∼ = ]AE 0 C Steps 7, 8, 9, and 10
12. ]ABC ∼ = ]DEF Steps 3 and 11
13. 4ABC ∼ = 4DEF Hypothesis and S.A.S.
If H = A, then ]B ∼ = ]E 0
by Step 5 and Theorem 25. Therefore,]B ∼ = ]E
by Step 3, and it follows that 4ABC ∼ = 4AE 0
C ∼
= 4DEF by the S.A.S. Axiom
and Step 3. The same argument applies if H = C. Suppose now that A comes
to lie between H and C, as shown. Then m ]ABC = m ]HBC − m ]HBA by
virtue of Axiom 12, and similarly, m ]AE 0 C = m ]HE 0 C − m ]HE 0 A.
Then, ]ABC ∼ = ]AE 0 C ∼ = ]DEF by Steps 7, 8, 11, and 12 above, and the
proof is completed similarly.
74 CHAPTER 4. ON NEUTRAL GEOMETRY I

4.2 Theorems on Perpendiculars in a Plane

Perpendicular lines were introduced earlier in Definition 21.

Theorem 31. Given a line l and a point P on l in a given plane Π, there exists
in Π one and only one line m through P and perpendicular to l.

Proof. We have to prove the existence and the uniqueness of line m. Let
plane Π be identified with the plane of the illustrative drawing.

We first prove the existence of a line m containing point P and perpendicular

to l. Let Π1 be one of the half-planes in Π which have l for boundary, and let
Q be a point distinct from P on l. By Axiom 11 (Angle Construction), there
←→
exists a point R in Π1 such that ]QP R is a right angle. Let m be line P R.
Then, by construction, m is perpendicular to l and we have shown that there is
at least one line through P perpendicular to l.

To prove the uniqueness of line m, we show that there is at most one line
perpendicular to l through P . Suppose there were two lines m1 and m2 through
P and both perpendicular to l. Let Q be a point distinct from P on l, and
−−→
consider the rays P R1 and P R2 having common vertex at P lying in the same
half-plane Π1 with l for boundary (edge), and determined by the points R1 and
R2 on m1 and m2 , respectively. Then, by definition of perpendicularity, one of
the angles determined by l and m1 is a right angle, so that we may consider, say,
m ]QP R1 = 90. In an entirely similar fashion, we may write: m ]QP R2 = 90.
But this contradicts Axiom 11 which ensures the uniqueness of a ray in a given
half-plane and creating an angle of prescribed measure (here, 90) with a given
ray. Hence, the assumption that there are two perpendiculars m1 and m2 to
line l through P is false.
4.2. THEOREMS ON PERPENDICULARS IN A PLANE 75

It should be noticed that the condition stated in Theorem 31 that the line
m perpendicular to line l through point P must lie in a prescribed plane Π is
essential for the uniqueness of line m: if one allows line m to lie in an arbitrary
plane, it is easily visualized that there are more than one such line. We have thus
an example of an existence theorem with no corresponding uniqueness theorem.
The inverse situation, i.e. a uniqueness theorem with no corresponding existence
theorem can also occur, as provided by the following example: “Two distinct
lines intersect in at most one point.”

Definition 32. In a given plane, the perpendicular bisector of a segment is

a line in that plane which is perpendicular to the given segment through its
midpoint.

Since the midpoint of a segment is unique (Theorem 4), Theorem 31 ensures

that in each given plane containing a given segment, the perpendicular bisector
of the segment is unique. We have here another instance of an existence and
uniqueness theorem that is valid only in a prescribed plane. The next theorem
is quite useful in many proofs:

Theorem 32. In a given plane, the perpendicular bisector of a given segment

consists of all points that are equidistant from the endpoints of that segment.

Proof. We first prove that if a point P is on the perpendicular bisector m of a

given segment AB in a given plane Π, then this point is equidistant from the
endpoints A and B of that segment. To that end, notice that if P happens to
be the midpoint M of segment AB, then obviously P A = P B.
Suppose that P is distinct from M . Then, P M ∼ = PM, MA ∼ = M B, and
]P M A = ]P M B, so that by the S.A.S. Axiom, 4AM P ∼
∼ = 4BM P . There-
fore, P A ∼
= P B and hence, P A = P B.
We must also prove that if a point P in the given plane Π is equidistant
from the endpoints A and B of a segment AB in P , then P must lie on the
perpendicular bisector m of segment AB which lies in Π. Again, if P lies on
←→
line AB, then P must be the (unique) midpoint M of segment AB and, thus,
←→
lies on the perpendicular bisector of this segment. If P does not lie on line AB,
←−→
consider the (unique) line P M : P M ∼ = PM, PA ∼ = P B (hypothesis), M A ∼ =
M B, so that by the S.S.S. Theorem, 4P M A = 4P M B. Hence, ]P M A ∼
∼ =
←−→
angP M B and, by definition, line P M is perpendicular to segment AB through
76 CHAPTER 4. ON NEUTRAL GEOMETRY I

←−→
its midpoint M . Therefore, P M is the perpendicular bisector of segment AB
←−→
in the given plane Π. By Theorem 31, P M is equal to m, and P lies on m.

In the next theorem, we prove a result similar to the conclusion of Theorem

31 when the given point does not lie on the given line. Now, the given line and
the given point not lying on it determine (at least) three non-collinear points and
hence, by Axiom 7, determine a unique plane. Therefore, the entire argument
leading to the proof of the next theorem will be conducted in that well-defined
plane. First, we prove the following uniqueness theorem:

Theorem 33. Given a line l and a point P not lying on l, there exists at most
one line through P perpendicular to l.

Proof. Suppose there were two distinct lines m1 and m2 containing P and per-
pendicular to l. Let Q1 and Q2 be the feet of m1 and m2 , respectively, on l. Since
m1 and m2 are distinct, it follows that Q1 and Q2 are different points (Why?).
−−→
On the ray opposite to, say, ray Q1 P , choose point R such that Q1 P ∼ = Q1 R
(Theorem 3). Then Q1 Q2 ∼ = Q1 Q2 , Q1 P ∼
= Q1 R and ]P Q1 Q2 ∼ = ]RQ1 Q2 (by
hypothesis), so that 4P Q1 Q2 ∼ = 4RQ1 Q2 by virtue of the S.A.S. Axiom. It
follows then that ]RQ2 Q1 ∼ = ]P Q2 Q1 and that ]RQ2 Q1 is a right angle, since
]P Q2 Q1 is a right angle (by hypothesis). But this contradicts Theorem 31 and,
therefore, there exists at most one line m1 (or m2 ) through P perpendicular to
l (in the plane determined by P and l).

Next, we prove the following existence theorem:

Theorem 34. Given a line l and a point P not lying on l, there exists at least
one line through P perpendicular to l.
4.2. THEOREMS ON PERPENDICULARS IN A PLANE 77

Proof. For a change, we use the double-column format:

Statements Reasons
1. Let A and B be arbitrary points Axiom 3
on l
−→
2. There exists ray AQ such that Axiom 11
]BAP ∼ = ]BAQ with Q and P on
opposite sides of l
3. There exists a point X on ray Theorem 3
−→
AQ such that AP ∼ = AX
4. X and P are on opposite sides Axiom 9: X and Q are on the same
of l side of l while P and Q are on op-
posite sides of l
5. Segment P X intersects l at Definition 12
point Y
6. 4P AY ∼ = 4XAY Steps 2, 3 and S.A.S. Axiom
7. ]AY P ∼ = ]XY A Corresponding parts of congruent
triangles
8. ]AY P is a right angle Definition 20
←→
9. P X ⊥ l Definition 21

Notice that the diagram accompanying the above proof corresponds to one
of three possible cases, the other two being when point A lies between points B
and Y or when point A is identical with point Y . These two latter cases can
be dealt with in a manner entirely similar to those occurring in the proof of the
Theorem 30 (S.S.S. Theorem) and, therefore, should be left as exercises.
Theorems 33 and 34 can now be merged into a single statement as follows:

Theorem 33/34. For any given line and any given point not lying on that line,
there exists a unique line through the given point perpendicular to the given
line.
78 CHAPTER 4. ON NEUTRAL GEOMETRY I

As a corollary, we may then assert:

Corollary. In any triangle at most one of the angles can be a right angle.
Proof. Suppose that in triangle 4ABC, angles ]A and ]B are both right.
Then this would imply that there are two lines through vertex C perpendicular
←→
to line AB which contains the opposite triangle side AB, thus contradicting
Theorem 33.
Definition 33. A triangle one of whose angles is a right angle is called a right
triangle. The side opposite the right angle is called the hypotenuse (and the
other two sides are sometimes called legs).

4.3 Fundamentals of Spatial Geometry

It is now appropriate to leave the plane and begin soaring in the three dimensions
that appear to be available to our subjective awareness. In this Chapter, we
will introduce additional axioms which are necessary for an equally rigorous
foundation of spatial or three-dimensional (3-D) Euclidean Geometry, similar
to our treatment of plane Euclidean Geometry. Our preliminary considerations
will continue to remain “neutral”, that is, free from any assumption concerning
the parallelism of lines and planes.
We begin with the reminder that “plane” is a primitive term and that a
plane contains an infinitude of points and lines (by virtue of Axioms 1 through
6), and that space contains an infinitude of planes. The justification of these
assertions is left as an exercise.
Theorem 35. If a line not contained in a given plane intersects that plane,
then the intersection is a unique point.

Proof. Let line l be not contained in plane Π and let l intersect Π at point P .
Suppose there is a second point Q—distinct from P —lying on l and contained
in Π. By Axiom 6, the line determined by points P and Q which are incident
with plane Π is entirely contained in Π. But this line must be identical with
line l (Axiom 1) thereby contradicting the hypothesis which states that l is not
contained in Π.
For the sake of completeness, we formulate and prove the following theorem
which we have already used implicitly (where?):
4.3. FUNDAMENTALS OF SPATIAL GEOMETRY 79

Theorem 36. A line and a point not incident with that line determine a unique
plane.
Proof. The existence and uniqueness of the plane are guaranteed by the fact
that the line contains (at least) two distinct points which, together with the
given point, constitute a triplet of non-collinear points which in turn, by virtue
of Axiom 7, determine a unique plane.
Theorem 37. Any two intersecting lines determine a unique plane.
Proof. Let lines l1 and l2 intersect at the unique point P . There exists point
Q1 distinct from P on line l1 (Axiom 3). Then, Q1 is not incident with l2
(Theorem 5). By Theorem 36, Q1 and l2 determine a unique plane Π. Since
plane Π contains points P and Q1 , it contains the entire line l1 (Axiom 6).
So far, we have proved the existence of at least one plane containing both
intersecting lines l1 and l2 . Suppose now that there is another plane Π0 , distinct
from plane Π0 , which also contains lines l1 and l2 . Then, planes Π and Π0 both
contain point Q1 and line l2 which is in contradiction with Theorem 36. Hence,
Π = Π0 .
Recall from Axiom 8 that if two planes intersect then their intersection is a
line. Combine this with Axiom 6 and we can deduce that if two distinct planes
contain two distinct points in common, then they contain the line determined
by those two points in common.
Spatial separation is entirely similar to the separation of a plane into two
half-planes by a line contained in that plane. We first define the different sides
of a plane:
Definition 34. Let Π be any plane and A and B any two distinct points not
incident with Π. If segment AB contains no point in common with Π, we say
that A and B are on the same side of Π, and if segment AB intersects plane
Π, we say that A and B are on opposite sides of Π.
The collection of all points not contained in a plane and located on the same
side of that plane is called a half-space. That any arbitrary plane separates
(three-dimensional) space into two disjoint half-spaces bounded by that plane
is established by the following spatial analog of the Plane Separation Axiom
(Axiom 9). Note that the proof of this theorem is somewhat complicated. The
reader can, instead, choose to take the theorem as an axiom and omit the proof.
Of course, the existence of the proof demonstrates that if taken as an axiom,
the axiom must be redundant.
Theorem 38 (Space Separation Theorem). For every plane Π and any three
points A, B, C not incident with Π:
(i) if A and B as well as B and C are in the same half-space, then A and C
are in the same half-space;
(ii) if A and B as well as B and C are in opposite half-spaces, then A and C
are in the same half-space.
80 CHAPTER 4. ON NEUTRAL GEOMETRY I

Proof. First, we assume the opposite of assertion (i), that A and B as well as
B and C are in the same half-space and that A and C are in opposite half-
spaces. Thus, by Definition 34, we know that segment AC intersects plane Π,
say at point P . Consider a plane Π1 containing the three given points (Axiom
7). Point P is also in Π1 (Axiom 6), and thus the two planes intersect and the
intersection is a line (Axiom 8), say l, which contains P . Thus, in Π1 , A and
C are on opposite sides of l. However, neither segment AB nor segment BC
intersect Π (Definition 34), so neither segment can intersect l, and thus in Π1
both A and B as well as B and C are on the same side on l. This contradicts
the Plane Separation Axiom (Axiom 9) as we’ve already established that A and
C are on opposite sides of l.
To prove assertion (ii), assume that A and B as well as B and C are in
opposite half-spaces of Π. Again, let Π1 be a plane containing A, B and C.
Since segment AB intersects plane Π (Definition 34) and this intersection must
be in plane Π1 (Axiom 6), the two planes must intersect and their intersection
is a line (Axiom 8). Again, call this line l. Since segments AB and BC both
intersect Π (Definition 34) and since these entire segments lie in plane Π1 (Axiom
6), both segments must intersect Π at points also in Π1 and thus must intersect
line l. So, in plane Π1 , we have A and B as well as B and C on opposite sides
of line l (Definition 12), and therefore, by Axiom 9, A and C are on same sides
of l in Π1 . Finally, by Definitions 12 and 34 and the fact that l is incident with
Π, A and C are on same half-planes of Π.
Note that if two points lie in different half-spaces determined by a given
plane, the segment determined by those two points must intersect the plane in
a (unique) point. It is also worthwhile mentioning that in three dimensions while
a plane is the boundary of exactly two half-spaces, a line is on the boundary of
infinitely many half-spaces.
4.3. FUNDAMENTALS OF SPATIAL GEOMETRY 81

Exercises
Present your proofs in the double-column format unless stated otherwise.

1. In the diagram shown, B is the midpoint of segment AF , RB ∼

= HB, and
]RBA ∼ = ]HBF . Prove that ]R ∼ = ]H.

2. Prove that in the diagram shown, if the line segments AD and BC bisect
each other at E, then AB ∼= CD and AC ∼ = BD.

3. In the figure shown, ]ABW ∼

= ]RHQ and point F is the midpoint of
segment BH.

Can you prove that 4W BF ∼

= 4QHF ? EXPLAIN

4. In the diagram shown below, AR ⊥ RX and BR ⊥ RY , AR = RX and

BR = RY .

Prove AB ∼
= XY .
82 CHAPTER 4. ON NEUTRAL GEOMETRY I

5. In the figure shown below, F A = F D and AB = DC.

Prove that 4F BC is isosceles.

6. In the plane diagram shown, AC = AB and CD = BD.

Prove: ]ACD ∼ = ]ABD.

7. Give a paragraph proof rather than a double-column proof of the following:

The equilateral triangle 4ABC is given. The midpoints P , Q, R of its
sides are the vertices of triangle 4P QR. Prove 4P QR is equilateral.
4.3. FUNDAMENTALS OF SPATIAL GEOMETRY 83

8. Prove that if the median AM of triangle 4ABC is perpendicular to side

BC, then 4ABC is isosceles.

9. Prove that if the bisector of angle ]F in triangle 4F GH is perpendicular

to the opposite side GH, then 4F GH is isosceles.

10. In the diagram shown, CB bisects GF and ]CGE ∼

= ]EF B.

Prove that GF bisects CB.

−→ −−→
11. Given ]AOB with OA ∼ = OB and points P and Q on rays OA and OB,
respectively, such that AQ ∼
= P B.

Can one prove that OP ∼

= OQ? Justify your answer!

12. Prove the following theorem:

“In an isosceles triangle, the angle bisector of the angle opposite to the
triangle base is perpendicular to this base.”

13. In the figure below: BQ ∼

= T S, ]B ∼
= ]T and ]Q ∼
= ]S.

Prove: RX ∼
= RY .
84 CHAPTER 4. ON NEUTRAL GEOMETRY I

14. Prove that the median from the vertex of an isosceles triangle to the
triangle’s base is the bisector of the vertex angle.
15. Prove that if in a triangle the median and the angle bisector from the
same vertex coincide, the triangle is isosceles.
16. A circular argument is camouflaged in the proof of the following theorem.
Can you unmask it?
Theorem. The base angles of an isosceles triangle are congruent.

Proof. Given 4ABC with AB ∼

= AC.
Statements Reasons
1. AB ∼= AC Given
2. AC ∼= AB Symmetry of congruence
∼
3. BC = CB Reflexivity of congruence
4. 4ABC ∼ = 4ACB S.S.S.
5. ]B ∼= ]C Definition of Congruent Triangles.

17. In the figure shown:

]HAR ∼ = ]BAR and ]HRF ∼
= ]BRF .
Prove that AF ⊥ HB.
4.3. FUNDAMENTALS OF SPATIAL GEOMETRY 85

18. In triangle 4HAF , points B and W are on sides AF and AH, respectively,
and F W ⊥ AH, HB ⊥ AF , and AW ∼ = AB.
Prove: F W = HB.

19. In triangles 4ABC and 4HRW , the medians AF and HQ are congruent
and AC ∼ = HW , AB ∼
= HR.
Prove that these two triangles are congruent.

20. Prove that the bisectors of a pair of corresponding angles of two congruent
triangles are congruent.

21. In the diagram shown, HA = HB, AF bisects ]HAB and BF bisects

]HBA.
Prove: AF = BF
86 CHAPTER 4. ON NEUTRAL GEOMETRY I

22. If two medians of a triangle are perpendicular to their respective sides,

then the triangle is equilateral.

23. In the figure shown:

AQ ⊥ RS, RQ ∼
= SQ, and RC ∼
= SC.

Prove: ]RCA ∼
= ]SCA.

24. The congruent triangles 4P AB and 4QAB lie in different planes but
have the side AB in common.

Prove that if X is any point between A and B, then 4P XQ is isosceles.

25. Prove that the bisector of any angle of a scalene triangle cannot be per-
pendicular to the opposite side.

26. A quadrilateral is a plane figure formed by four coplanar distinct points

such that no three of them are collinear, and such that the segments AB,
BC, CD and DA intersect only in their endpoints. (See diagram.)
4.3. FUNDAMENTALS OF SPATIAL GEOMETRY 87

In a given quadrilateral, one diagonal bisects two non-adjacent angles.

Prove that it bisects the other diagonal. See Definition 46.

27. In the figure shown:

RC = SC, Q is the midpoint of segment RS, and ]RCA ∼
= ]SCA.
Prove: AQ ⊥ RS.

28. Points P and Q are on opposite sides of plane Π . Let M be the intersection
of segment P Q with Π. Determine whether the following statements are
true or false. JUSTIFY their truth by referring to appropriate theorem(s)
or their falsity by supplying a counterexample.

(a) If l is a line in Π perpendicular to P Q, then P and Q are on opposite

sides of l in the plane determined by P and l.
(b) If l is a line in Π containing point M , then P and Q are on opposite
sides of l in the plane determined by P and l.
(c) If l is any line in Π, then P and Q are on opposite sides of l in the
plane determined by P and l.
88 CHAPTER 4. ON NEUTRAL GEOMETRY I

(d) P and Q are on opposite sides of any plane containing point M but
←→
not containing line P Q.
Chapter 5

On Neutral Geometry II

5.1 Geometric Inequalities

We continue our development of Neutral Geometry with theorems about the
comparison of segments and angles with one another in two or three dimensions.
To begin with, we define the exterior angles of a triangle:

−→
Definition 35. If point D is on ray AC of triangle 4ABC such that vertex C
is between A and D, then the angle ]BCD is an exterior angle of 4ABC.

Clearly, each triangle has six exterior angles which are grouped in three pairs
of opposite, hence, congruent angles, as shown.

Definition 36. The angles ]A and ]B of triangle 4ABC are called the remote
interior angles of the exterior angles ]BCD and ]ACE.

Theorem 39 (Exterior Angle Theorem). In any triangle, an exterior angle is

larger than either of its remote interior angles.

89
90 CHAPTER 5. ON NEUTRAL GEOMETRY II

Proof.
Statements Reasons
1. Let point E be the midpoint of By Theorem 4 there is such a mid-
BC point
−→
2. Let F be a point on ray AE Theorem 3
such that E is between A and F
and AE = EF
3. ]BEA ∼ = ]F EC Opposite angles (Theorem 21)
4. 4BEA ∼ = 4CEF Steps 1, 2, 3 and S.A.S.
5. m ]B = m ]ECF Corresponding parts of congruent
triangles
6. m ]BCD = m ]ECF + Axiom 12 (Angle Addition Axiom)
m ]F CD
7. m ]BCD = m ]B + m ]F CD Steps 5 and 6
8. m ]BCD > m ]B Step 7 and algebra
In a similar fashion, m ]ACX > m ]A (Proof left as an exercise). Since
]ACX ∼= ]BCD, it follows that m ]BCD is greater than m ]A and m ]B.

Corollary. In a right triangle, the other two angles are acute.

Proof. If m ]A = 90◦ , then m ]BCD > 90◦ . Since m ]BCA+m ]BCD = 180◦
(Axiom 13), it follows that m ]BCA = 180◦ − m ]BCD, so that m ]BCA <
90◦ . In a similar fashion, one proves that m ]CBA < 90◦ .

Theorem 40 (The S.A.A. Theorem). Given two triangles, if there is a corre-

spondence between their vertices such that two angles and a side opposite one
of these angles of one triangle are congruent to the corresponding angles and
side of the other triangle, then the correspondence is a congruence.
5.1. GEOMETRIC INEQUALITIES 91

Proof. In triangles 4ABC and 4DEF , let ]A ∼

= ]D, ]B ∼
= ]E, and AC ∼
=
DF .

Statements Reasons
−−→
1. On ray AB, let X be a point so Point Plotting Theorem
that AX = DE
2. 4AXC ∼ = 4DEF S.A.S.
3. m ]AXC = m ]DEF Corresponding parts
4. m ]AXC = m ]ABC Step 3 and hypothesis
5. Suppose point X is the same as Assumption
point B
6. AB ∼ = DE Steps 5 and 6
7. 4ABC ∼ = 4DEF A.S.A.
8. Suppose X 6= B and X is be- Step 5, Law of the Excluded Mid-
tween A and B or B is between A dle
and X
9. In either case, one of ]AXC Definitions 35 and 36, Theorem 3
and ]ABC is an exterior angle of
4BXC and the other is a remote
interior angle.
10. m ]AXC 6= m ]ABC Theorem 39
11. Contradiction Steps 9 and 4
12. X = B
13. 4ABC ∼ = 4DEF Steps 12 and 2

We have already noticed that a side-side-angle congruence theorem cannot

be proved in general. There is, however, a particular case for which an S.S.A.
congruence is valid:

Theorem 41 (The Hypotenuse-Leg Theorem). Given a correspondence be-

tween two right triangles, if the hypotenuse and a leg of one triangle are con-
92 CHAPTER 5. ON NEUTRAL GEOMETRY II

gruent to the corresponding parts of the other triangle, then the correspondence
is a congruence.

Proof. The right triangles 4ABC and 4DEF are given with BA ∼ = ED and
∼ −−→
BC = EF . On the ray opposite to DF , select point Q such that DQ = AC.
Then 4EDQ ∼ = 4BAC by the S.A.S. Axiom, and hence EQ = BC. In the
isosceles triangle 4EQF , ]EQD ∼
= ]EF D. Therefore, in the right triangles
4DEF and 4DEQ, we have EQ ∼ = EF , ]EQD ∼ = ]EF D and also ]EDQ ∼ =
]EDF . Hence, by Theorem 40 (S.A.A.), 4EDQ ∼ = 4EDF ∼ = 4ABC.

The next theorem asserts an “obvious” fact:

Theorem 42. If two sides of a triangle are not congruent, the angles opposite
these sides are not congruent, and the larger angle is opposite the longer side.

Proof. In triangle 4ABC, suppose AB > AC. Using the Point Plotting The-
−→
orem, select point D on ray AC such that AD = AB. Then in the isosceles
triangle 4ABD: m ]ADB = m ]ABD. Since C is between A and D (The-
orem 1), it follows from Theorem 9 that C is in the interior of angle ]ABD,
so that BC is in the interior of ]ABD, and by the Angle Addition Axiom
we have: m ]ABC + m ]CBD = m ]ABD. Since m ]CBD > 0, it follows
that m ]ABD is larger than m ]ABC. Therefore: m ]ADB > m ]ABC
(Theorem 25). Now, ]ACB is an exterior angle of 4BCD, and by Theo-
rem 39, m ]ACB > m ]ADB. Hence, m ]ACB > m ]ADB > m ]ABC, i.e.
m ]C > m ]B, which was to be proved.

It happens that the converse of Theorem 42 is equally true:

Theorem 43. If two angles of a triangle are not congruent, then the sides
opposite these angles are not congruent, and the larger angle is opposite the
longer side.
5.1. GEOMETRIC INEQUALITIES 93

Proof. In triangle 4ABC, assume that m ]B < m ]C. Then there are three
possibilities for the lengths of the sides AB and AC: either AB > AC or
AB < AC or AB = AC. If AB = AC, then 4ABC is isosceles and by
Theorem 25, ]B ∼ = ]C which contradicts the hypothesis. If AB < AC, then by
Theorem 42, m ]C < m ]B which also contradicts the hypothesis. Therefore,
the only remaining possibility is AB > AC which was to be proved.

Of particular importance in applications is the next theorem:

Theorem 44. Given line l and point P not lying on l, the shortest segment
with one endpoint at P and the other on l is the perpendicular segment through
P to l.

Proof. Let Q be the foot of the perpendicular from P to l, and let R be an

arbitrary point on l. Take point S on l such that Q is between R and S. Then
]P QS is an exterior angle of triangle 4P QR and therefore, by Theorem 39,
m ]P QS > m ]P RQ. Since m ]P QS = m ]P QR = 90◦ , we have m ]P QR >
m ]P RQ, and by Theorem 43, it follows that P R > P Q. Since R was an
arbitrary point on l, it follows that the length of the perpendicular segment
from P to l is shorter than any segment joining P to a point on l.

Definition 37. The distance of a point not lying on a line to that line is the
length of the perpendicular segment from that point to the line. The distance
between a line and a point lying on it is defined to be zero.

Theorem 45 (The Triangle Inequality). The sum of the lengths of any two
sides of a triangle is larger than the length of the third side.

−−→
Proof. Take point D on ray CB such that B is between D and C and DB = BA.
Then, CD = CB + BD = AB + CB. Furthermore, since B is in the interior of
angle ]CAD, it follows that m ]DAB < m ]DAC (Why?). On the other hand,
in the isosceles triangle 4BDA, m ]BDA = m ]BAD. Hence, m ]BDA <
m ]DAC, and by Theorem 43, we have AC < DC, that is, AC < CD or
AC < AB + BC, which was to be proved.
94 CHAPTER 5. ON NEUTRAL GEOMETRY II

The next theorem can be considered to be a generalization of Theorem 43

to two triangles:
Theorem 46. If two sides of a triangle are congruent respectively to two sides
of another triangle, and the included angle of the first triangle is larger than the
corresponding included angle of the second triangle, then the opposite side of
the first triangle is longer than the corresponding opposite side of the second.

Proof. Consider the two triangles 4ABC and 4DEF with AB = DE and
AC = DF and m ]BAC > m ]EDF . Use Axiom 11 (Angle Construction
−→ ←→
Axiom) to construct ray AQ with Q on the same side of line AC as vertex B of
4ABC, such that m ]QAC = m ]EDF . On ray AQ, use the Point Plotting
Theorem to determine point K such that AK = DE. Construct triangle 4AKC
in the interior of ]BAC. Then, by S.A.S., 4AKC ∼= 4DEF . Now, bisect angle
]BAK and let this angle bisector intersect segment BC at M (Justify this
intersection!). Then 4ABM ∼ = 4AKM (S.A.S.), so that BM = M K. The
Triangle Inequality (Theorem 45) applied to triangle 4CKM yields: CK <
KM + M C or CK < CM + BM i.e. CK < BC. Since CK = EF , it follows
that EF < BC which was to be proved.

The converse of Theorem 46 is also valid:

Theorem 47. If two sides of one triangle are congruent respectively to two
sides of a second triangle, and the third side of the first triangle is longer than
the third side of the second triangle, then the included angle of the first triangle
is larger than the included angle of the second triangle.
The proof of this theorem is similar to that of Theorem 43 and is left as an
exercise (Hint: use Theorem 46 and S.A.S. to exclude two unwanted possibili-
ties).
Definition 38. The perpendicular segment joining a vertex of a triangle to the
line that contains the opposite side is called an altitude of the triangle or the
altitude from that vertex.
5.2. PERPENDICULAR LINES AND PLANES 95

Every triangle has three altitudes. Notice that the foot of an altitude does
not necessarily lie between the two other triangle vertices, and that the length
of an altitude from a triangle vertex is always shorter than the lengths of the
two triangle sides meeting at that vertex (Theorem 44).

5.2 Perpendicular Lines and Planes

We now proceed with the study of the perpendicularity of lines and planes in
three-dimensional space.
Definition 39. A line l and a plane Π are perpendicular if they intersect at
point P and if every line in Π through P is perpendicular to l.
In symbols: l ⊥ Π.

It should be noted that given a line l intersecting a given plane Π, there

exists always at least one line contained in Π and perpendicular to l. The key
word in Definition 39 is “. . . if every line in Π . . . ”
In order to determine whether a line intersecting a plane is perpendicular
to the plane, Definition 39 is of no great help: one obviously cannot verify
that all lines contained in the plane through the intersection point are indeed
perpendicular to the given line. One therefore needs a perpendicularity criterion
that is more practical in applications. To that end, we first prove two lemmas:
Lemma. Given two points P and Q not lying on line l, if each of two points
A and B on l is equidistant from P and Q, then every point of l is equidistant
from P and Q.
In this lemma, points P , Q and line l do not necessarily lie in the same
plane (compare with Theorem 32).
96 CHAPTER 5. ON NEUTRAL GEOMETRY II

Proof. Suppose X is a point on l such that, say, B is between A and X. By

assumption, AP = AQ and BP = BQ, and also AB = AB, so that 4P AB ∼ =
4QAB (S.S.S.). Hence, we have ]P AB ∼ = ]QAB. With our choice of X,
we may then write: ]P AX ∼ = ]QAX. It follows then that4P AX ∼ = 4QAX
(S.A.S.). Therefore, P X = QX. The cases where X is between A and B and
A is between B and X are dealt with in a similar fashion.

Lemma. If each of three non-collinear points in a given plane is equidistant

from two points not contained in that plane, then every point of the plane is
equidistant from these two points.

Proof. Let Π be the plane determined by the three non-collinear points A, B,

C. By the first Lemma, since A and B are each equidistant from points P and
←→
Q, every point on line AB is also equidistant from P and Q. A similar result
←→
is true for every point on line BC. Now, let X be a point of Π, not lying on
←→ ←→ ←→ ←→
AB or BC, and let Y be a point on AB on the opposite side of BC than X.
The Plane Separation Axiom (Axiom 9) ensures the existence of such a point
←→
Y . Let Z be the intersection of segment XY with line BC. Since Y and Z are
←→
equidistant from P and Q, the first Lemma asserts that every point of line ZY ,
in particular, point X, is equally equidistant from P and Q, which is what we
wanted.

We may now prove the following perpendicularity theorem:

5.2. PERPENDICULAR LINES AND PLANES 97

Theorem 48. If a line is perpendicular to each of two intersecting lines at their

point of intersection, then it is perpendicular to the plane determined by these
two intersecting lines.

Proof. Let l1 and l2 be two lines intersecting at point A and let Π be the plane
determined by these two lines. Let line l be perpendicular to each of l1 and l2
at A, and let l3 be any line contained in Π through A. Choose an arbitrary
point P on l and let point Q be the point of l on the opposite side of Π from
P and such that AP = AQ (Justify the existence of point Q). Let points B1 ,
B2 , B3 be on lines l1 , l2 , l3 respectively. Then, in the plane determined by lines
l and l1 , l1 is the perpendicular bisector of segment P Q (Definition 32). By
Theorem 32, B1 is equidistant from P and Q. In a similar way, B2 is shown
to be equidistant from P and Q, and point A is evidently equidistant from P
and Q. By the second Lemma, it follows that point B3 is also equidistant from
P and Q, and so are all points of line l3 . Now, in the plane determined by l
and l3 , Theorem 32 asserts that l3 is the perpendicular bisector of segment P Q
and, hence, l3 is perpendicular to l. Since l3 was chosen arbitrarily, it follows
that l is perpendicular to all lines contained in Π through A. Therefore, l is
perpendicular to plane Π, which was to be proved.
Now, in analogy with plane Euclidean Geometry in which there exists a
unique line through a given point perpendicular to a given line, we prove that
there exists a unique plane through a given point perpendicular to a given line.
As in the two-dimensional case, we begin with the simpler case and proceed to
the general case.
Theorem 49. Given a line and a point on that line, there exists (at least) a
plane through the given point perpendicular to the given line.

Proof. Let P be a point on line l. Let R be a point not on l and let Π1 be

the (unique) plane determined by R and l (Theorem 36). Choose point Q not
98 CHAPTER 5. ON NEUTRAL GEOMETRY II

contained in Π1 and let Π2 be the plane determined by l and Q. In planes Π1

and Π2 there exists (unique) lines l1 and l2 , respectively, which pass through P
and are perpendicular to l (Theorem 31). By Theorem 48, line l is perpendicular
to the plane determined by l1 and l2 . This establishes the existence of a plane
through P perpendicular to line l.

The next theorem ensures that all lines perpendicular to l at P are contained
in the plane perpendicular to l at P whose existence is guaranteed by Theorem
49:

Theorem 50. If a plane and a line are perpendicular, then the plane contains
all lines perpendicular to the given line at its intersection with the given plane.

Proof. Let line l and plane Π be perpendicular at P , and let m be a line perpen-
dicular to l at P . Then m and l determine the plane Π0 which intersects plane
Π along line n (Axiom 8). Since n is contained in Π, n is perpendicular to l at
P . On the other hand, both lines m and n lie in Π0 and both are perpendicular
to l at P . By Theorem 31, m and n must be the same line. Since n lies in Π,
it follows that m is contained in Π.

We are now in a position to prove the uniqueness of the plane perpendicular

to a given line whose existence was established in Theorem 49:

Theorem 51. Through a given point on a given line, there exists at most a
plane perpendicular to the given line.

Proof. By Theorem 50, all lines perpendicular to the given line at the given
point are contained in a plane through that point and perpendicular to the
line. If there were another plane through the point and perpendicular to the
line, then the intersection of these two different perpendicular planes would be
a single line and the second plane would not contain all lines perpendicular to
the given line at the given point. This contradicts Theorem 50 and, thus, the
two perpendicular planes are the same.

Theorem 52. Given a line l and a point P not on l, there exists a unique plane
through P perpendicular to l.
5.2. PERPENDICULAR LINES AND PLANES 99

Proof. We reduce this case to the previous one by considering the (unique) plane
Π determined by the line l and point P (Theorem 36). In plane Π, Theorem 33
and 34 guarantee the existence of a unique line through P and perpendicular to
l. Let H be the foot of this perpendicular. Theorems 49, 50, and 51 guarantee
the existence and uniqueness of a plane through H and perpendicular to l, which
←→
necessarily contains line HP and, hence, point P.

The next theorem is the three-dimensional analog of Theorem 32, and its
proof is left as an exercise:

Theorem 53. The perpendicular bisecting plane of a segment consists of all

points equidistant from the endpoints of the segment.

In other words, if Π is the perpendicular bisecting plane of segment P Q,

then every point in Π is equidistant from P and Q, and if point X is equidistant
from P and Q, then X is contained in Π.

Theorem 54. Two lines perpendicular to a given plane are coplanar.

Proof. Let Π be a plane and lines l1 and l2 be perpendicular to Π at points

A and B, respectively. Let M be the midpoint of segment AB and l the line
in Π which is the perpendicular bisector of AB. Choose points P and Q on
←→
opposite sides of line AB such that M Q = M P . Now, choose point C different
from A on l1 . Then 4AM P ∼ = 4AM Q (S.A.S.), so that AP = AQ. Since l1 is
perpendicular to Π, ]CAP ∼ = ]CAQ, m ]CAQ = 90◦ , and, hence, 4CAP ∼ =
4CAQ (S.A.S.), so CP = CQ. By Theorem 53, points A and C lie in the
perpendicular bisecting plane Π0 of segment P Q. Hence, l1 is contained in Π0 .
In a similar fashion, one proves that l2 is contained in Π0 . Therefore, l1 and l2
are coplanar.

The following four theorems guarantee the existence and uniqueness of a line
through a given point and perpendicular to a given plane:

Theorem 55. Through a given point in a given plane, there exists (at least)
one line perpendicular to the plane.
100 CHAPTER 5. ON NEUTRAL GEOMETRY II

Proof. Let Π be a plane and P a point in Π. By Axiom 5(i), there exists point
←→
Q in Π, and line P Q is entirely contained in Π (Axiom 6). Let Π0 be the
←→
(unique) plane perpendicular to line P Q at P (Theorems 49 and 51). Planes
Π and Π intersect along line m containing P (Axiom 8). In plane Π0 , there
0

exists a unique line l perpendicular to m at P (Theorems 33 and 34). Since

←→
Π0 is perpendicular to line P Q at P and l contains P and lies in Π0 , l must be
←→ ←→
perpendicular to P Q (Definition 39). Now, l is perpendicular to two lines, P Q
and m, at their intersection P . Hence, by Theorem 48, l is perpendicular to the
plane Π determined by these two lines.

Theorem 56. Through a given point in a given plane, there is at most one line
perpendicular to the given plane.

Proof. Suppose there were two distinct lines l1 and l2 perpendicular to the
given plane Π at point P in Π. Then l1 and l2 determine a unique plane Π0
(Theorem 37) which intersects plane Π in a unique line m containing P (Axiom
8). In Π0 , we would then have two different lines perpendicular at P to m, thus
contradicting Theorem 31.

Similar existence and uniqueness theorems hold for the perpendicular line
to a given plane through a point not contained in that plane:

Theorem 57. Through a given point not contained in a given plane, there
exists (at least) one line perpendicular to that plane.
5.2. PERPENDICULAR LINES AND PLANES 101

Proof. Let plane Π and point P not contained in Π be given. Let Q be an

arbitrary point in Π and m the line through Q perpendicular to Π (Theorems
55 and 56). If P is incident with m, then m is the desired perpendicular. If P is
not incident with m, let Π0 be the plane determined by P and m (Theorem 36).
The intersection of planes Π and Π0 is line n through Q. In Π0 , there exists a
unique perpendicular through P to line n (Theorems 33 and 34). Let H be the
foot of this perpendicular. By Theorems 55 and 56, there exists a unique line l
perpendicular to plane Π at H. Lines m and l are both perpendicular to plane
Π. Therefore, by Theorem 54, m and l are coplanar, so l lies in Π0 . Now, in Π0 ,
←→
lines P H and l are both perpendicular to n at H. By the uniqueness of this
←→
perpendicular, it follows that P H and l are the same, so that l contains point
P and is the desired perpendicular to Π.
Theorem 58. Through a given point not contained in a given plane, there
exists at most one line perpendicular to that plane.
Proof. The proof is identical with that of Theorem 56, except for obvious re-
placements (e.g., “from point P ” instead of “at point P ”, etc.).
102 CHAPTER 5. ON NEUTRAL GEOMETRY II

Exercises
1. The figure shown illustrates the following statement:
An exterior angle of a quadrilateral is greater than each of the remote
interior angles.
Is this a valid statement? Explain.

2. Prove the following theorem:

The sum of the measures of any two angles of a triangle is less than 180◦ .
Refer to the diagram shown in your proof.

3. Prove the following theorem:

The base angles of an isosceles triangle are acute.
Hint: Base your proof on the statement of Exercise 2.

4. Prove the following statement:

If two altitudes of a triangle are congruent, the triangle is isosceles.

5. In the figure shown below: ]QKH ∼

= ]F BH and AQ = AF .
Prove: QB = F K
5.2. PERPENDICULAR LINES AND PLANES 103

6. If in the figure shown, AF is the shortest side and CB is the longest side
of the quadrilateral ABCF , prove that m ]F > m ]B.

7. Prove the following statement:

If the base of an isosceles triangle is extended, a segment which joins the

vertex of the triangle with any point on this extension is longer than one
of the congruent sides of the triangle.

8. Prove that the sum of the lengths of the diagonals of the quadrilateral
shown below is less than the perimeter of the quadrilateral.
104 CHAPTER 5. ON NEUTRAL GEOMETRY II

9. Let A, B, C be points, not necessarily distinct. Prove that AB+BC ≥ AC

and that AB + BC = AC is true if and only if B is on the segment AC.
10. Prove that the shortest polygonal path from one point to another is the
segment joining them.

In other words: given n points A1 , A2 , . . . , An , prove: A1 A2 + A2 A3 +

. . . + An−1 An ≥ A1 An .
11. Given two coplanar segments AC and BD intersecting at P , prove that if
X is any point in the plane of the four points A, B, C, D, other than P ,
then:

XA + XB + XC + XD > P A + P B + P C + P D
5.2. PERPENDICULAR LINES AND PLANES 105

12. Given a line m and two points P and Q on the same side of m, find the
point R on m for which P R + RQ is as small as possible.

13. In isosceles triangle 4RAF with RA = RF , B is a point on side AF such

that m ]ARB < m ]BRF .
Prove: AB < BF .

14. Given triangle 4ABF with median RB and m ]ARB = 80◦ .

Prove: m ]A > m ]F .

15. In 4ABC, BC > AC and Q is the midpoint of side AB. Is ]CQA acute
or obtuse? Justify your answer.

16. Prove the following theorem:

If a median of a triangle is not perpendicular to the side to which it is
drawn, then the lengths of the two other sides of the triangle are unequal.
106 CHAPTER 5. ON NEUTRAL GEOMETRY II

17. Given 4ABC with AB > AC and F C = DB, as shown below, prove that
F B > CD.

18. Prove the following theorem:

The length of the longest side of any triangle is less than half its perimeter.

19. Given the isosceles triangle 4ABF with F A = F B, AB < AF , and H on

←→
line AF so that F is between A and H.

Prove that triangle 4ABH is scalene.

20. In the triangle shown below, ]C is a right angle and m ]B = 2(m ]A).

Prove: AB = 2CB
5.2. PERPENDICULAR LINES AND PLANES 107

21. Prove the following statement:

The sum of the distances from a point inside a triangle to the ends of one
side is less than the sum of the lengths of the other two sides.
22. Given are three collinear points A, B, and X, as shown. Points A and B
are each equidistant from P and Q. Prove that X is also equidistant from
P and Q. Does the proof require that Q be in the plane of A, B, X, and
P ? (Don’t use the lemma.)

23. In the figure shown, F B is perpendicular to plane Π, and in 4RAB, which

lies in Π, BR = BA.
Prove ]F AR ∼ = ]F RA.
108 CHAPTER 5. ON NEUTRAL GEOMETRY II

24. A cube may be defined—somewhat abstractly—by stating that it consists

of twelve congruent segments such that any two intersecting segments are
perpendicular. In the cube shown, BR = BL. Prove that KR = KL.
(You may assume that the points that look coplanar are coplanar.)

25. Prove this theorem:

If a line l intersects a plane Π, there is at least one line m contained in Π
such that m ⊥ l.
26. Point P is in the plane Π determined by points A, B, C; P is equidistant
from A, B, and C. Line l is perpendicular to Π at P . Prove that every
point X of l is equidistant from A, B, C.

27. Line l is perpendicular to the plane of the three points A, B, and C.

Point P of l is equidistant from A, B, C. Prove that every point of l is
equidistant from A, B, C.
5.2. PERPENDICULAR LINES AND PLANES 109

28. Prove the following theorem:

If from a point A not contained in plane Π, a perpendicular AB and the
oblique segments AF and AH are drawn, meeting the plane at unequal
distances from the foot B of the perpendicular from A, then the segment
which meets the plane at the greater distance from B has the greater
length.

29. Segment AP is perpendicular to the segments P Q and P C. Segment P Q

←→
is perpendicular to line BC at Q. Prove that segment AQ is perpendicular
←→
to line BC.

30. Prove that Euclidean space is “three-dimensional” by proving that each

−−→ −→ −−→ −→
of four rays AB, AC, AD, AE with common vertex A cannot be perpen-
dicular to each of the other three.
110 CHAPTER 5. ON NEUTRAL GEOMETRY II
Chapter 6

The Parallel Axiom

6.1 Parallel Lines in a Plane

So far, we have been concerned with intersecting lines and planes. In this
chapter, we focus on lines and planes that have no point in common.
We already know that two lines can intersect in at most one unique point,
and that if two lines intersect, then they lie in one unique plane. Hence, if two
lines are not coplanar, they cannot intersect.
Definition 40. Two lines which are not coplanar are called skew.
Definition 41. Two coplanar lines which do not intersect are called parallel.
Two segments are called parallel if the two lines containing them are parallel.
Parallel lines l1 and l2 are denoted l1 k l2 with a similar notation for parallel
segments.
It is absolutely crucial to keep in mind that parallel lines are by definition
coplanar. It is easily seen that there is only one plane containing a pair of
parallel lines: if lines l1 and l2 are parallel, and Π1 is one plane containing these
two lines, then by Theorem 36, any arbitrary point P on, say, line l2 together
with line l1 define a unique plane Π2 which must then contain both l1 and l2 .
Hence Π1 and Π2 are the same plane.
The following theorem provides a convenient criterion for the parallelism of
two lines:
Theorem 59. Two coplanar lines (or segments) are parallel if they are both
perpendicular to the same line.
Proof. Consider two lines l1 and l2 lying in the same plane Π and both per-
pendicular to line l contained in Π. If lines l1 and l2 do not intersect, they are
parallel by Definition 41. If they intersect, say, at point P , then there would be
two lines through the same point P perpendicular to the same line l, which is
prohibited by Theorem 31 if P lies on l and by Theorem 33 if P does not lie on
l. Hence, l1 and l2 must be parallel.

111
112 CHAPTER 6. THE PARALLEL AXIOM

A legitimate question which one must ask is whether parallel lines exist at
all. It is well known that a flying elephant can be defined as an elephant with
two (large) wings on its back, but it is an entirely different matter to provide
an example for the beast that one has just defined. In the context of parallel
lines, suppose we have a line l and a point P not lying on l. There are infinitely
many lines incident with P and lying in the (unique) plane determined by P
and l. How many do not intersect l? The following important existence theorem
guarantees that there is at least one such line:

Theorem 60. Given line l and point P not incident with l, there exists at least
one line through P parallel to l.

Proof. By Theorems 33 and 34, there exists a unique line m through P perpen-
dicular to l. By Theorem 31, there exists—in the plane determined by P and
l—a unique line n through P perpendicular to m. Then, Theorem 59 asserts
that the coplanar lines n and l are parallel.

So we know that there is at least one parallel to a given line through a

point not lying on that line. Could there be more than one? It might appear
quite natural to try to prove that the answer is negative. For more than two
thousand years, Euclid, his predecessors, and scores of eminent mathematicians
who followed in their footsteps, tried very hard to prove this assertion from
the other axioms of Neutral Geometry. Those who believed to have succeeded
invariably made use of an extra assumption, and in every case, the hidden
assumption was eventually ferreted out by some later thinker. We shall devote
a whole chapter to some of these attempts (because they are instructive) and to
the foundations of Non-Euclidean Geometries, in particular that of Hyperbolic
Geometry. It was in the nineteenth century that it was definitively established
that the uniqueness of a parallel through a given point to a given line not
containing that point can NOT be proved from the axioms of Neutral Geometry
and, hence, must make the object of an independent axiom.
Before stating the “modern” version of the Parallel Axiom (see Appendix I
for Euclid’s version, the notorious Fifth Postulate), we introduce some defini-
tions and prove some additional theorems which are useful in the study of the
parallelism of lines.
6.1. PARALLEL LINES IN A PLANE 113

Definition 42. Given two coplanar lines, a transversal of these lines is a line
which intersects them in two distinct points.
It should be pointed out that in the definition of transversal, the two coplanar
lines that the transversal intersects may or may not be parallel, and that if those
two coplanar lines intersect, the transversal cannot be concurrent with them.
Definition 43. Given (coplanar) lines l1 and l2 and a transversal l intersecting
l1 and l2 at P and Q, respectively, let A be a point on l1 and B a point on l2 on
the opposite side of l than point A. Then angles ]AP Q and ]BQP are a pair
of alternate interior angles formed by the transversal and the lines l1 and l2 .

The following theorem is easily proved:

Theorem 61. If two lines are cut by a transversal, and if one pair of alternate
interior angles are congruent, then the other pair of alternate interior angles are
also congruent.
Proof. Left as an exercise.
Theorem 62. If two (coplanar) lines cut by a transversal form a pair of con-
gruent alternate interior angles, then the two lines are parallel.
114 CHAPTER 6. THE PARALLEL AXIOM

Proof. Let l be a transversal of the two coplanar lines l1 and l2 , intersecting them
at P and Q, respectively. Let a pair of alternate interior angles be congruent.
There are then two possibilities: either l1 and l2 are parallel, which is what we
wanted to prove, or l1 and l2 intersect at point R. Choose point S on, say, l1
on the opposite side of l from R. Then ]SP Q is an exterior angle of triangle
4P QR, and ]P QR is one of its remote interior angles. By Theorem 39, we
conclude m ]SP Q > m ]P QR. But by hypothesis, one pair of alternate interior
angles is congruent, and Theorem 61 asserts that the other pair of alternate
interior angles must also be congruent. Thus, m ]SP Q = m ]P QR which
contradicts the previous inequality. Therefore l1 and l2 cannot intersect.

We note in passing that the converse of Theorem 62 cannot be proved

without the Parallel Axiom and, thus, could be construed as an equivalent
formulation of this axiom. But for now, we proceed with the examination of
angle relationships when parallel lines are cut by a transversal.

Definition 44. When two (coplanar) lines l1 and l2 are cut by a transversal l,
if angles ]a and ]b are a pair of alternate interior angles and if angles ]b and
]c are opposite (vertical) angles, then angles ]c and ]a constitute a pair of
corresponding angles.

Clearly, when two lines are cut by a transversal, four pairs of corresponding
angles are formed. The following two theorems are often useful and their proofs
are left as exercises:

Theorem 63. If two (coplanar) lines are cut by a transversal and if one pair
of corresponding angles are congruent, the other three pairs of corresponding
angles are also congruent.

Theorem 64. If two (coplanar) lines are cut by a transversal and if a pair of
congruent corresponding angles is formed, then the two lines are parallel.

The same remark concerning the converse of Theorem 62 applies to the

converse of Theorem 64.
We are now poised to introduce the Euclidean—as contrasted with the Non-
Euclidean—Parallel Axiom in a formulation proposed by the British mathe-
matician John Playfair (late 18th century):

Axiom 15 (Euclidean Parallel Postulate). Through a point not incident with

a given line there is at most one line parallel to the given line.
6.1. PARALLEL LINES IN A PLANE 115

Since Theorem 60 ensures the existence of at least one line parallel to a

given line through a point not lying on the given line, Axiom 15 asserts that
this parallel line is unique.
We may now prove the converse of Theorem 62:

Theorem 65. If two parallel lines are cut by a transversal, then each pair of
alternate interior angles are congruent.

Proof. Let lines l1 and l2 be cut by the transversal l3 at points P and Q, respec-
tively. Assume that angles ]a and ]b are not congruent (RAA hypothesis).
Then, by the Angle Construction Axiom (Axiom 11), there exists a (unique)
line l through P such that pairs of alternate interior angles formed between l, l2
and l3 are congruent (see Theorem 61 too). Since by hypothesis angles ]b and
]c are not congruent, l1 6= l. On the other hand, l1 and l2 are parallel and by
Theorem 62, l and l2 are also parallel. Therefore, there are two distinct lines
through P which are parallel to line l2 , a contradiction with Axiom 15. Hence,
]a ∼= ]b which was to be proved.

In an entirely similar manner, one proves the following useful theorems whose
proofs are left as exercises:

Theorem 66. If two parallel lines are cut by a transversal, then each pair of
corresponding angles are congruent.

Theorem 67. If two parallel lines are cut by a transversal, then pairs of interior
angles (e.g. ]a and ]b or ]c and ]d) on the same side of the transversal are
supplementary.

Theorem 68. If two lines are cut by a transversal and a pair of interior angles
on the same side of the transversal are supplementary, then the two lines are
parallel.
116 CHAPTER 6. THE PARALLEL AXIOM

Theorem 69. In a plane, two lines parallel to the same line are parallel.

Theorem 70. If a line in the plane of two parallel lines is perpendicular to one
of them, it is also perpendicular to the other.

As an additional exercise, we show that if Theorem 70 is chosen as an axiom,

then Axiom 15 (the Parallel Axiom) can be proved:

Suppose line m and point P not incident with m are given. Assume further
that distinct lines l1 and l2 pass through P and that l1 is parallel to line m. We
prove that l2 cannot be parallel to m, a result which will restrict the number of
lines through P parallel to m to a single line. Let t be the line through P per-
pendicular to m (Theorems 33 and 34). By Theorem 70, line t is perpendicular
to line l1 . Suppose now that l2 is also parallel to line m. Then line t will be
equally perpendicular to l2 . Since both l1 and l2 cannot be perpendicular at the
same point P to line t (see Theorem 31), it follows that either l2 is not parallel
to m or that l1 and l2 cannot be distinct, which is what we set out to prove.
The first major consequence of the Parallel Axiom is the following theorem:

Theorem 71. The sum of the measures of the angles of a triangle is 180◦ .

Proof. Given triangle 4ABC, let l be the line through vertex B parallel to
side AC. Let angle ]x, ]y, ]z, ]x0 , ]y 0 be as shown, and let D be a point
←→ ←→ ←→
on l on the same side of line AB as vertex C. Since AC is parallel to BD,
←→
A is on the same side of BD as C. Hence, point C is in the interior of angle
]ABD (see Definition 16 for the interior of an angle!). Therefore, by the Angle
Addition Axiom (Axiom 12): m ]ABD = m ]z + m ]y 0 . Moreover, by Axiom
13 (Angle Supplement Axiom): m ]x0 + m ]ABD = 180◦ . Hence, we have:
m ]x0 +m ]z+m ]y 0 = 180◦ . But from Theorem 63 we know that m ]x = m ]x0
and also m ]y = m ]y 0 for the pairs of alternate interior angles formed by the
←→ ←→ ←→
parallel lines AC and l with the transversals AB and BC, respectively. By
substitution, we obtain:
6.2. QUADRILATERALS 117

m ]x + m ]y + m ]z = 180◦
which was to be proved.
Three important corollaries follow from Theorem 71:
Corollary. Given a correspondence between two triangles, if two pairs of cor-
responding angles are congruent, then the third pair of corresponding angles are
also congruent.

Note that the first corollary applies to correspondences between triangle

vertices which are not congruences as well as to congruences.
Corollary. In a right triangle, the acute angles are complementary.
Corollary. In a triangle the measure of any exterior angle is equal to the sum
of the measures of the two remote interior angles.
The proofs of these corollaries are straightforward and left as exercises.

6.2 Quadrilaterals
So far, we have been dealing mostly with triangles. In the following, we define
the next more evolved species of plane objects.
Definition 45. Let A1 , A2 , A3 , A4 be four coplanar points such that no three
of them are collinear and such that the segments A1 A2 , A2 A3 , A3 A4 and A4 A1
intersect only in their endpoints. Then the set of these four points (vertices)
and four segments (sides) is called a quadrilateral and denoted A1 A2 A3 A4 .
118 CHAPTER 6. THE PARALLEL AXIOM

Note that the order in which the vertices of a quadrilateral are labeled is
essential. For instance, ABCD does not denote a quadrilateral in the sense of
Definition 45 because the segments BC and AD intersect in a point which is not
an endpoint of either segment. The same four points, however, constitute a bona
fide quadrilateral when written in the order ABDC, as shown. Moreover, notice
that figure P QRS does form a bona fide quadrilateral according to Definition
45.
These considerations lead us to a further classification of quadrilaterals but
before doing so a few definitions concerning the elements of a quadrilateral are
necessary:
Definition 46. In a quadrilateral:
Opposite sides are two sides that do not intersect.
Opposite angles are two angles which do not share a common side.
Consecutive sides share a vertex as a common endpoint.
Consecutive angles share a side.
Diagonals are segments with two nonconsecutive vertices for endpoints.
Definition 47. Quadrilateral ABCD is called convex if each pair of its opposite
sides, e.g. AB and CD (and equally BC and DA), is such that CD is contained
←→
in one of the half-planes bounded by the line AB and AB is contained in one
←→
of the half-planes bounded by the line CD.
According to Definition 47, quadrilateral P QRS is not convex since, for
instance, side SP is not contained in one of the half-planes bounded by the line
←→
QR containing its opposite side QR. We shall be exclusively concerned with
convex quadrilaterals. A few “obvious” properties of convex quadrilaterals are
enumerated below:
Definition 48. A point lies in the interior of a quadrilateral if it lies in the
interior of each angle (see Definition 16) of the quadrilateral.
Theorem 72. The diagonals of a convex quadrilateral are contained in its
interior and intersect each other.
6.2. QUADRILATERALS 119

Proof. Consider the convex quadrilateral ABCD. By Definition 47, vertices A

←→
and D are on the same side of BC and vertices C and D are on the same side
←→
of AB. Therefore, by Definition 16, vertex D is in the interior of angle ]ABC,
−−→ −−→
and hence ray BD is in the interior of angle ]ABC. Similarly, ray DB is in
the interior of angle ]ADC. Therefore DB is in the interior of quadrilateral
ABCD. In a similar fashion, one proves that diagonal AC is contained in the
interior of the quadrilateral. Furthermore, by the Crossbar Theorem (Theorem
−−→ −−→
11), rays BD and DB must intersect AC and, consequently, the diagonals BD
and AC intersect and their intersection is in the interior of the quadrilateral
ABCD.
One can also prove that the interior of a convex quadrilateral is a convex set
of points in the sense that if P and Q are any two points in the interior of the
quadrilateral, then the entire segment P Q is contained therein.
We continue with the taxonomy of convex quadrilaterals and state some
useful theorems relating to them.
Definition 49. A trapezoid is a (convex) quadrilateral in which only one pair
of opposite sides are parallel. A parallelogram is a quadrilateral in which both
pairs of opposite sides are parallel. A rhombus is a parallelogram all of whose
sides are congruent. A rectangle is a parallelogram all of whose angles are right
angles. A square is a rectangle all of whose sides are congruent.
The proofs of the following theorems are easy and left as exercises:
Theorem 73. In a parallelogram, each pair of opposite sides are congruent.
Theorem 74. In a parallelogram, each pair of opposite angles are congruent
and any two consecutive angles are supplementary.
Theorem 75. In a parallelogram, the diagonals bisect each other.
Conversely, the following theorems identify the subspecies of a parallelogram:
Theorem 76. If both pairs of opposite sides of a (convex) quadrilateral are
congruent, the quadrilateral is a parallelogram.
Theorem 77. If two opposite sides of a (convex) quadrilateral are parallel AND
congruent, the quadrilateral is a parallelogram.
Theorem 78. If the diagonals of a (convex) quadrilateral bisect each other,
the quadrilateral is a parallelogram.
Theorem 79. If a parallelogram has one right angle, it is a rectangle.
Theorem 80. The diagonals of a rhombus are perpendicular bisectors of each
other.
Theorem 81. If the diagonals of a (convex) quadrilateral are perpendicular
bisectors of each other, the quadrilateral is a rhombus.
120 CHAPTER 6. THE PARALLEL AXIOM

The next theorem is frequently used in geometric proofs:

Theorem 82. The segment joining the midpoints of two sides of a triangle is
parallel to the third side and half as long as the third side.

Proof. Let D and E be the midpoints of sides AB and BC, respectively, of

−−→
triangle 4ABC. Let F be the point on ray DE such that E is between D and F
and DE = EF (Point Plotting Theorem). Since angles ]x and ]y are congruent
and BE = EC, it follows that 4EF C ∼ = 4EDB (S.A.S.) and hence, ]v ∼= ]w
←→ ←→
(corresponding parts). It follows that AB and F C are parallel (Theorem 62).
Since AD = DB and DB = F C (corresponding parts of congruent triangles
4DBE and 4F EC), we have AD = F C by transitivity. By Theorem 77,
quadrilateral ADF C is a parallelogram so that DE is parallel to AC. Finally,
since DF = AC (Theorem 73) and DE = EF (corresponding parts of congruent
triangles 4DBE and 4F EC), it follows that DE = 21 AC.

Corollary. A segment through the midpoint of one side of a triangle that is

parallel to a second side intersects the third side through its midpoint.

Another important theorem concerns the sum of the measures of the angles
of a (convex) quadrilateral:

Theorem 83. The sum of the measures of the angles of a convex quadrilateral
is equal to 360◦ .

Proof. Let ABCD be a convex quadrilateral. Let diagonal, say, BD divide

the quadrilateral into the two triangles 4ABD and 4CBD (use Theorem 72).
−−→
Label the various angles formed as shown. On ray BA select point X such that
vertex A is between X and B. Then angle ]XAD is an exterior angle of 4ABD
and m ]XAD = 180◦ − m ]A = m ]B1 + m ]D1 (Corollary 3 to Theorem 71).
6.3. PARALLEL LINES AND TRANSVERSALS 121

−−→
Similarly, we pick point Y on ray BC such that vertex C is between Y and B
and obtain : m ]Y CD = 180◦ − m ]C = m ]B2 + m ]D2 . Adding these two
relations yields: 360◦ − m ]A − m ]C = m ]B + m ]D (Angle Addition Axiom)
or m ]A + m ]B + m ]C + m ]D = 360◦ which was to be proved.
Theorem 83 can be generalized to an arbitrary convex polygon. A convex
polygon can be defined in a manner entirely similar to the definition of a convex
quadrilateral (or tetragon). It can then be proved that the measures of the
angles of a convex n-gon (polygon with n vertices or sides or angles) add up to
(n − 2)180◦ . A rigorous proof of this generalization of Theorem 83 relies on the
method of mathematical induction and will be omitted for the sake of brevity.
It is interesting to note in passing that weaker versions of Theorems 71
and 83 can be proved without using the Parallel Axiom, that is, in the
framework of Neutral Geometry: the Saccheri-Legendre Theorem (see Chapter
Thirteen) asserts that the sum of the measures of the angles of any triangle is
less than or equal to 180◦ . Consequently, by an argument similar to the one used
in the proof of Theorem 83, the sum of the measures of the angles of a convex
quadrilateral can be proved to be at most equal to 360◦ . We shall return to these
considerations when we examine the foundations of Non-Euclidean Geometry.
Before proceeding with the study of parallel lines cut by transversals, we
state one more theorem. The proof is left to the reader.
Theorem 84. If lines l1 and l2 are parallel and if P and Q are any two points
of l1 , then the distances of P and Q from l2 (see Definition 37) are equal.
This theorem justifies the following:
Definition 50. The distance between two parallel lines is the distance of any
point on one line from the other line.

6.3 Parallel Lines and Transversals

Let us now consider several parallel lines cut by a transversal. If lines l1 , l2 ,

l3 are parallel and intersected by a transversal t at points A, B, C, respectively,
then we say that l1 , l2 , l3 intercept segments AB and BC on the transversal t.
This notion of intercepts can be clearly generalized to arbitrary many parallel
lines. We now prove:
Theorem 85. If three parallel lines intercept congruent segments on a transver-
sal, they intercept congruent segments on any other transversal.
122 CHAPTER 6. THE PARALLEL AXIOM

Proof. Let transversal t1 intersect the three parallel lines l1 , l2 , l3 at points A,

B, C, respectively, and let AB = BC. Let t2 be any other transversal cutting
l1 , l2 , l3 at D, E, F , respectively.
Assume first that t1 and t2 are not parallel and that A is distinct from D,
as shown. Let t3 be the transversal through A parallel to t2 , intersecting l2
and l3 at points G and H, respectively, and let t4 be the transversal through B
parallel to t2 , intersecting l3 at I. Let angles ]x, ]y, ]z, ]w be as shown in the
figure. Then, ]x ∼ = ]z (Theorem 66). Since t3 k t4 (Theorem 69), then ]w ∼ =
]y (Theorem 66). By hypothesis, AB = BC and, hence, 4ABG ∼ = 4BCI
(A.S.A.), so that AG = BI (corresponding parts). Furthermore, AGED and
BIF E are parallelograms (Definition 49), so that AG = DE and BI = EF
(Theorem 73). It follows by transitivity that DE = EF .
Next, if the two transversals t1 and t2 are parallel (and clearly A and D are
distinct), then the quadrilaterals ABED and BCF E are parallelograms, and
since AB = BC, it follows that AB = DE = BC = EF (Theorem 73). Finally,
if A and D are the same point, then t2 is actually the same as t3 and we have
already proved the congruence of the intercepts AG and GH. This completes
the proof of the theorem.

By repeated applications of Theorem 85, one proves the following Corollary:

Corollary. If several parallel lines intercept congruent segments on one transver-

sal, they intercept congruent segments on any other transversal.

In Appendix IV we shall prove a profound and non-trivial generalization of

Theorem 81, but for now, we prove the following important theorem which has
applications in Mechanics:

Theorem 86. The medians of a triangle are concurrent at a point (called the
centroid of the triangle) whose distance from a triangle vertex is two-thirds the
length of the median from that vertex.
6.4. PARALLELISM IN SPACE 123

Proof. Let AD and BE be two medians of triangle 4ABC. Then AD and BE

are contained in the interior of 4ABC and intersect (why?) at a point P in
the interior of 4ABC. Now, use the Point Plotting Theorem (Theorem 3) to
−−→
locate points G and H on ray BC such that BG = GD = DH = HC and let l1 ,
←→
l2 , l3 = AD, l4 , l5 be five lines through B, G, D, H, C, respectively, parallel to
←→
AD. Then, by the Corollary to Theorem 85, l3 , l4 , l5 divide AC into congruent
segments, and so, the midpoint E of AC lies on l4 . In the same vein, l1 , l2 ,
l3 , l4 divide BE into three congruent segments BI = IP = P E. Therefore,
BP = 23 BE.

In the same fashion, introduce lines parallel to median CF of triangle 4ABC,

intercepting congruent segments on side AB. If P 0 is the intersection of medi-
ans BE and CF , then we find that BP 0 = 32 BE. The Point Plotting Theorem
(Theorem 3) then guarantees that P and P 0 are the same point, so that the three
medians are concurrent. Since CF passes now through point P , it is easily rec-
ognized that CP = 23 CF and that also AP = 32 AD (Corollary to Theorem 85).
This completes the proof of the theorem.

6.4 Parallelism in Space

We now turn briefly to the relationships between parallel lines in three-dimensional
space.

Definition 51. A line and a plane or two planes which do not intersect are
called parallel.
124 CHAPTER 6. THE PARALLEL AXIOM

Parallel lines and parallel planes should be carefully distinguished: as the

figures show, parallel planes Π and Π0 contain non-intersecting and non-parallel
(skew) lines, and intersecting planes contain non-intersecting parallel lines.

Theorem 87. If a plane intersects two parallel planes, then it intersects them
in two parallel lines.

Proof. Let plane Π intersect the parallel planes Π1 and Π2 along lines l1 and l2
(Axiom 8), which lie in the same plane Π. But lines l1 and l2 cannot intersect
because they are lying in the parallel planes Π1 and Π2 , respectively. Therefore,
l1 and l2 are parallel.

Theorem 88. If a line is perpendicular to one of two parallel planes, it is also

perpendicular to the other plane.
6.4. PARALLELISM IN SPACE 125

Proof. Let line l be perpendicular to plane Π1 which is parallel to plane Π2 . In

plane Π2 , let point A be not on l and let Π be the (unique) plane determined
by A and l. By Theorem 87, plane Π intersects planes Π1 and Π2 along the
parallel lines l1 and l2 , respectively. Since l is perpendicular to plane Π1 , it
is perpendicular to line l1 and, hence, by Theorem 70, l is perpendicular to
l2 . Now, take another point A0 in plane Π2 not incident with both l and l2
and repeat the preceding argument. We thus obtain two lines in plane Π2 each
perpendicular to l at the point of intersection of l with Π2 . So, l is perpendicular
to Π2 by Theorem 48.

Theorem 89. Two planes perpendicular to the same line are parallel.

Proof. Let planes Π1 and Π2 be perpendicular to line l which intersects them

at points P and Q, respectively. Now, if Π1 and Π2 were not parallel, they
would intersect. Let R be a point common to these two planes. Then l would
←→ ←→
be perpendicular to both lines RP and RQ which are contained in planes Π1
and Π2 , respectively. This, however, contradicts Theorem 33. Therefore, planes
Π1 and Π2 cannot intersect and are, hence, parallel by Definition 51.

An immediate consequence of this theorem is:

126 CHAPTER 6. THE PARALLEL AXIOM

Corollary. If each of two planes is parallel to a third plane, they are parallel
to each other.
Proof. Let plane Π1 be parallel to plane Π and plane Π2 be also parallel to
plane Π. Consider a line l perpendicular to plane Π. By Theorem 88, l is
perpendicular to both planes Π1 and Π2 , and by Theorem 89, planes Π1 and
Π2 must be parallel.
Theorem 90. Two lines perpendicular to the same plane are parallel.
Proof. By Theorem 54, two lines perpendicular to the same plane are coplanar.
If their intersections with the given plane are points P and Q, then the two
←→
(coplanar) lines would be perpendicular to the same line P Q and, hence, be
parallel (Theorem 59).
Two corollaries follow from this theorem:
Corollary. A plane perpendicular to one of two parallel lines is also perpen-
dicular to the other.

Proof. Consider the parallel lines l1 and l2 with, say, l1 perpendicular to plane
Π. Let A be any point on l2 not contained in plane Π and let l3 be the line
through A perpendicular to Π (Theorems 57 and 58). Then by Theorem 90, l1
and l3 are parallel. Hence, by the Parallel Axiom (Axiom 15), l3 = l2 and so, l2
is perpendicular to plane Π.
Corollary. If each of two lines are parallel to a third line, they are parallel to
each other.
Proof. Let l1 be parallel to l2 and l1 be parallel to l3 . Let Π be a plane perpen-
dicular to l1 . Then, by the first corollary, Π is perpendicular to both l2 and l3
and, hence, by Theorem 90, l2 and l3 are parallel.
This second corollary ensures the full (i.e. three-dimensional) transitivity of
line parallelism.
6.4. PARALLELISM IN SPACE 127

Theorem 91. Two parallel planes are equidistant.

Proof. We have to show that the lengths of all segments perpendicular to the
two planes and having each of their endpoints in one of the planes are equal.
Let P Q and RS be two segments perpendicular to the parallel planes Π1 and
Π2 , as shown. Theorem 90 asserts that P Q and RS are parallel and, hence,
coplanar. Let Π be the plane determined by these two parallel lines. Then,
by Theorem 87, the intersections of Π with Π1 and Π2 are parallel lines which
means that the quadrilateral P QRS is a parallelogram (in fact, a rectangle).
Therefore, P Q = RS (Theorem 73), which was to be proved.
We close this chapter with two definitions which are frequently used in three-
dimensional Geometry:
Definition 52. Two non-coplanar half-planes intersecting along a line form a
dihedral angle. The line common to the half-planes is called the edge and the
two half-planes are called the faces or sides of the dihedral angle.

It can be easily proved that the intersections of a dihedral angle by planes

perpendicular to its edge form congruent angles (proof left as an exercise).
This leads, in turn, to the following definition:
128 CHAPTER 6. THE PARALLEL AXIOM

Definition 53. The measure of a dihedral angle is the measure of the angle ob-
tained by intersecting the dihedral angle with a plane perpendicular to its edge.
The measure of the angle between two half-planes is the measure of the dihedral
angle determined by the two half-planes. Planes that form right dihedral angles
are called perpendicular.

Immediate consequences of these definitions are the following corollaries

whose proofs are left as exercises:
Corollary. If a line is perpendicular to a given plane, then any plane incident
with this line is perpendicular to the given plane.

Corollary. If two planes are perpendicular, then any line in one of them per-
pendicular to the line of intersection of these two planes is perpendicular to the
other plane.
Finally, we define the (measure of the) angle between a line and and a plane.
To that end, we first introduce the following definitions:
Definition 54. The projection of a point onto a plane is the foot of the per-
pendicular from the point to the given plane.
Note that by Theorems 57 and 58 this projection is a unique point.
6.4. PARALLELISM IN SPACE 129

Definition 55. The projection of a line onto a plane is the set of points in the
plane which are the projections of the points of the given line onto the given
plane.
Theorem 92. If a line is not perpendicular to a plane, its projection onto the
given plane is a line.

Proof. Let l be a line not perpendicular to plane Π. If l is incident with Π, then

each point of l is its own projection. Hence, the line and its projection onto Π
are one and the same line. If l is not incident with Π, let P be a point of l and P 0
its projection onto Π. Let Π0 be the plane determined by the intersecting lines
←−→
l and P P 0 . Since planes Π and Π0 have the point P 0 in common, they intersect
along a line l0 (Axiom 8). Let R be a point of l and let R0 be its projection onto
Π. If R = P , then R0 = P 0 and so, R0 lies on l0 . If R is distinct from P , then
the lines RR0 and P P 0 are coplanar (Theorem 54). Since by Axiom 7, the plane
Π0 containing points P , R and P 0 is unique, R0 is contained in both planes Π
and Π0 . Therefore, R0 is incident with l0 since l0 , being the intersection of Π and
Π0 , contains all points common to these two planes. This, however, does not
guarantee that the projection of l onto P constitutes all of l0 ! We must still show
that if S 0 is any point of l0 , then there exists a point S on l whose projection
onto P is S 0 . To this end, note that the line n through S 0 perpendicular to P is
coplanar with line P P 0 and so lies in Π0 . Therefore, n must intersect l (why?)
at some unique point S so that S 0 is indeed the projection of S onto Π. This
completes the proof of Theorem 92.
We may now define the angle between a line and a plane:

Definition 56. The angle between a line and a plane is the angle between the
line and its projection onto the plane.
130 CHAPTER 6. THE PARALLEL AXIOM

Exercises
1. Give an indirect proof of each of the following statements:

(a) In a plane, if a line m intersects one of two parallel lines l1 and l2 , it

also intersects the other line.
(b) In a plane, if a line m intersects only one of two other lines l1 and l2 ,
then the two lines l1 and l2 intersect.

2. Show that if Theorem 70 is assumed as an axiom, the Euclidean Parallel

Axiom (Axiom 15) can be proved as a theorem:
If l1 k m and l1 and l2 contain point P , then l2 cannot be parallel to m.

3. The quadrilateral ABCD is given with angles as labeled in the figure.

Prove that s◦ + r◦ = t◦ + u◦ .

4. In the figure shown below, triangle 4ABC is a right triangle with m ]CAB =
90◦ and QB = QA.
6.4. PARALLELISM IN SPACE 131

Prove: QB = QC.

5. In triangle 4ABC, ]C is a right angle, AS = AT and BR = BT .

Prove: m ]ST R = 45◦ .

6. Would the following information about a quadrilateral suffice to prove it

is a parallelogram? a rectangle? a rhombus? a square? Consider each
piece of information separately.

(a) Both pairs of its opposite sides are parallel.

(b) Both pairs of its opposite sides are congruent.
(c) Three of its angles are right angles.
(d) Its diagonals bisect each other.
(e) Its diagonals are congruent.
(f) Its diagonals are perpendicular and congruent.
(g) Its diagonals are perpendicular bisectors of each other.
(h) All sides are congruent.
(i) All angles are congruent.
(j) All sides and angles are congruent.
(k) Both pairs of its opposite angles are congruent.
(l) Each pair of its consecutive angles are supplementary.
(m) A diagonal bisects two angles.
(n) Each diagonal bisects two angles.
(o) Every two consecutive angles are supplementary.
(p) Every two consecutive sides are congruent.
132 CHAPTER 6. THE PARALLEL AXIOM

7. The quadrilateral ABCD is a parallelogram. On diagonal AC, points P

and R are chosen so that AP = RC.
Prove that the quadrilateral DPBR is a parallelogram.

8. In the isosceles triangle 4RST , P is an arbitrary point on the triangle base

ST . Through P parallel lines are drawn to the sides RS and RT meeting
them in X and Y , respectively. Prove that P XRY is a parallelogram and
that its perimeter is equal to the sum of the lengths of the legs of the
isosceles triangle.

9. In the figure shown, ABCD is a parallelogram whose diagonals AC and

BD intersect at point Q. An arbitrary line is drawn through Q meeting
sides DC and AB of the parallelogram at E and F , respectively.
Prove that segment EF is bisected by Q.

10. The median of a trapezoid is the segment whose endpoints are the mid-
points of the non-parallel sides of the trapezoid. Prove that the median of
a trapezoid is parallel to the bases and its length is equal to half the sum
of the lengths of the bases.
6.4. PARALLELISM IN SPACE 133

←→ ←→
Hint: Draw line DQ meeting line AB at point R.

11. In the figure, AD < BC, DA is perpendicular to AB and CB is perpen-

dicular to AB.
Prove that m ]C < m ]D.
Hint: Select point C 0 on BC such that BC 0 = AD.

12. Triangle 4ABC is isosceles with AC = BC. Point P is an arbitrary point

on base AB. Through P , perpendicular lines are drawn to legs AC and
BC, meeting them at X and Y , respectively. Prove that the P X + P Y is
constant and independent of the choice of P .
134 CHAPTER 6. THE PARALLEL AXIOM

Hint: Draw P Q perpendicular to, say, the altitude BT from vertex B and
show that P X + P Y is equal to BT .

13. Prove that the sum of the lengths of the perpendiculars drawn from any
point in the interior of an equilateral triangle to the three sides is constant
and independent of the choice of the point.
Hint: Consider a triangle altitude, draw a segment perpendicular to this
altitude from the interior point, and use Problem 12.

14. Square ABCD is given. Points K, L, M , N divide the sides as shown,

with a and b being the lengths of the indicated segments.
Prove that KLM N is a square.

15. In triangle 4ABC, CM is a median from vertex C and segment BQ

bisects CM at point P .
Prove that Q is a trisection point of AC, i.e. AQ = 2QC.
−−→
Hint: On the ray opposite to ray CB, select point R such that CR = CB
−−→
and show that BQ is contained in a median of 4ABR.

16. In a parallelogram ABCD, prove that the segments connecting opposite

vertices, say B and D, to the midpoints X and Y of the opposite sides
AD and BC, trisect the diagonal AC.
Hint: Through, say, vertex C draw line l parallel to one of the segments
and stare at the figure.
6.4. PARALLELISM IN SPACE 135

17. Quadrilateral ABCD is a parallelogram with diagonals AC and DB.

Points P and R are on AC such that AP = CR < 21 AC.
Prove: DP BR is a parallelogram.

18. In triangle 4ABC, median AM is congruent to segment CM . Prove that

4ABC is a right triangle.
19. Draw a small sketch to illustrate the hypothesis of each of the following
statements and determine whether the conclusion is true or false. In each
case, JUSTIFY your answer!

(a) If a line is perpendicular to one of two parallel planes, it is perpen-

dicular to the other.
(b) Two lines parallel to the same plane may be perpendicular to each
other.
(c) Two planes perpendicular to the same segment may intersect each
other.
(d) If a plane intersects two intersecting planes, the lines of intersection
may be parallel.
(e) If two planes are both perpendicular to each of two parallel lines,
the segments intercepted between the two planes on these lines are
congruent.
(f) If two planes perpendicular to the same line are intersected by a third
plane, the lines of intersection are parallel.
(g) If a line is incident with a plane, a perpendicular to the line is per-
pendicular to the plane.
(h) If a line is incident with a plane, a perpendicular to the plane at some
point of the line is perpendicular to the line.
(i) If two lines are parallel, every plane containing only one of them is
parallel to the other line.
136 CHAPTER 6. THE PARALLEL AXIOM

(j) If two lines are parallel, every line intersecting one of them intersects
the other.
(k) If two planes are parallel, any line incident with one of them is parallel
to the other.
(l) If two planes are parallel, any line incident with one of them is parallel
to any line in the other.
(m) If a plane and a line not incident with the plane are both perpendic-
ular to the same line, they are parallel to each other.
(n) If a plane and a line not incident with the plane are both parallel to
the same line, they are parallel to each other.
(o) If two planes are parallel to the same line, they are parallel to each
other.
(p) Two planes perpendicular to the same plane are parallel to each
other.
(q) Two lines perpendicular to the same line at the same point are per-
pendicular to each other.
(r) A plane perpendicular to one of two intersecting planes must intersect
the other.

20. Show how to determine a plane containing one of two skew lines and
parallel to the other.
JUSTIFY your construction!
21. Given two skew lines, show how to construct a line perpendicular to both
of them. JUSTIFY your construction!
−→
22. Plane Π and point A contained in Π are given. Ray AQ is not incident
−→ −→ −→
with Π. Let ray AR be the projection of ray AQ onto Π and let AP be
any ray with vertex A and contained in Π.
Prove: m ]QAR < m ]QAP .
−→
Hint: Let Q0 be the projection of Q onto Π. On ray AP choose X such
0
that AX = AQ .
Chapter 7

On Area

In our study of triangles and their relationships, we were, so far, exclusively

concerned with the congruence of corresponding triangle sides and angles. In-
tuitively speaking, however, the concept of congruence—between triangles or
polygons—contains more information than merely the equality of side lengths
and of angle measures: two congruent geometric shapes have also the same “con-
tent”. This recognition leads to the concept of area of a plane geometric figure,
beginning with the simplest, namely the triangle. A precise purely geometric
definition of area is, however, as tricky as that of the length of a segment or
the measure of an angle. We shall therefore adopt the same axiomatic approach
which allowed us to assign real numbers to purely geometric entities, i.e. to
segments and angles, and then to use these numbers to study the relationships
between these objects.
We begin with broadening the definition of a triangle as a subset of the plane
consisting of three non-collinear points (vertices) and the three segments with
pairs of these points as endpoints (sides). To this end, recall that the interior
of a triangle has already been precisely defined (see Definition 17) and then
introduce the following definition:

Definition 57. A triangular region is a set of coplanar points consisting of the

vertices, the sides, and all interior points of a triangle. If two (coplanar) triangles
are congruent, then the triangular regions they define are also called congruent.
Two coplanar triangular regions are said to be non-overlapping if they have
no interior point in common but may share common vertices or sides (or parts
thereof). A polygonal region is the union of a finite number of (coplanar) non-
overlapping triangular regions called a triangulation of the polygonal region.

Note that according to this definition, we are exclusively concerned with

coplanar regions, and, hence, our entire study of the concept of area will remain
confined to one single plane. Although this is not an essential restriction, it
simplifies our elaboration of the theory of area. Some examples of polygonal
regions are displayed below (one of them has a triangular “hole” in it):

137
138 CHAPTER 7. ON AREA

It goes without saying that polygonal regions need not to be convex and that
although they constitute a fairly large class of plane shapes, there are simple
and important plane regions which are not polygonal: an example is provided
by the region formed by a circle together with its interior points. Moreover, it
is also intuitively evident that, quite generally, a given polygonal region can be
triangulated in the sense of Definition 57 in several—in fact, infinitely many—
different ways. Finally, it can be shown that the intersection and the union of
any two (coplanar) polygonal regions are again polygonal regions. We are now
ready for an axiomatic definition of the area of a polygonal region:
Axiom 16 (Area Measure). There exists a function α defined on triangular
regions with the following properties:

(i) To each triangular region ∆, α assigns a strictly positive real number α(∆)
called the area of the triangular region ∆.
(ii) If two triangular regions ∆1 and ∆2 are congruent, then their areas are
equal: α(∆1 ) = α(∆2 ).
(iii) If two triangular regions ∆1 and ∆2 do not overlap, then:

α(∆1 ∪ ∆2 ) = α(∆1 ) + α(∆2 )

The positive-valued function α is called an area function or area measure

(for triangular regions). Property (i) is the positivity of the area function,
property (ii) expresses its invariance under congruence, and property (iii)
reflects its additivity.
Using Axiom 16, one can extend the domain of definition of an area function
to include polygonal regions by means of the following intuitively “evident”
theorem:
Theorem 93. For any polygonal region R with some triangulation by non-
overlapping triangles ∆1 , ∆2 , . . . , ∆n such that R = ∆1 ∪ ∆2 ∪ . . . ∪ ∆n ,
the area of this region is given by the sum α(R) = α(∆1 ) + α(∆2 ) + . . . +
α(∆n ). Furthermore, this area is independent of the triangulation used
to calculate it.
Proof. The proof of this important theorem is both subtle and tedious and shall
be omitted in this elementary treatment of Geometry1 .
1 For a proof, consult, for example, Foundations of Geometry, by K. Borsuk and W

Szmielew, North-Holland Publishing Co., 1960, p. 281

 139

Axiom 16 guarantees the existence of one area function. It can be shown that
there is, in fact, essentially only one such function, as stated in the following
highly nontrivial theorem:

Theorem 94. Any multiple of an area function by a positive constant is also

an area function and any two area functions are proportional.

Proof. We omit the proof of this theorem too because it is rather difficult and
lies partly in the domain of real analysis2 .

The net upshot of Theorem 94 is that, as was the case with the length of
a segment or the measure of an angle, the “unit of area measurement” can be
specified at will. It is customary to choose this unit in association with the
unit chosen for length measurement: whatever unit of length we choose, we use
the corresponding “square unit” to measure areas. But this convention does
not yet provide an operational method for actually calculating the value of the
area function for a given triangular—and, subsequently, polygonal—region. We
therefore introduce another axiom to “fix” the area measurement procedure:

Axiom 17 (Unit of Area). The area of a rectangle is the product of the length
of its base by the length of its altitude.

Henceforth, we shall indulge in a minor abuse of language for the sake of

brevity and shall simply write “base” or “altitude” in lieu of “length of base”
or “length of altitude.” This admittedly imprecise language should not lead to
misunderstandings since it will always be apparent from the context whether we
mean a segment (i.e. a set of points) or the length thereof (i.e. a real number).
Regarding Axiom 17, it would seem more natural to choose the unit of area to
be the area of a square and set it equal to the square of its side length. This
choice, however, would necessitate the proof that the area of a rectangle is the
product of its base and its altitude. Such a proof can indeed be carried out
but turns out3 to be rather lengthy, especially when the lengths of the sides
of the rectangle are irrational numbers, whereas if we know how to calculate
the area of a rectangle, then the computation of the area of a square follows
automatically since a square is merely a particular case of a rectangle.
It should also be mentioned clearly that our Euclidean axiomatic system as
described by Axiom 1 through Axiom 15 would allow for the proof of Axioms
16 and 17 in the form of existence and uniqueness theorems. Such a proof,
however, is too lengthy and elaborate to be incorporated in our elementary
development of Euclidean Geometry4 .
We can now proceed with the calculation of the areas of a few simple polyg-
onal regions.
2 For a proof, see, for example, Elementary Geometry from an Advanced Standpoint, by E.

Moise, Addison-Wesley Publishing Co., 1963, p. 345

3 See E. Moise, op. cit., p. 165
4 The interested reader may consult any advanced text on Geometry, for instance, Com-

panion to Euclid, by R. Hartshorne, Berkeley Mathematics Lecture Notes, Vol. 9, American

Mathematical Society, 1997, Chapter 5.
140 CHAPTER 7. ON AREA

Theorem 95. The area of a right triangle is one-half the product of its legs.

Proof. Let 4P QR be a right triangle with right angle at R and P R = b,

←→
RQ = a. Let R0 be the intersection of the parallel through vertex Q to RP with
←→
the parallel through vertex P to RQ. The quadrilateral P RQR0 is then a rect-
angle and we have: 4P QR ∼ = 4P QR0 . By Axiom 16, α(4P QR) = α(4P QR0 )
and α(P RQR ) = α(4P QR) + α(4P QR0 ) = 2α(4P QR). By Axiom 17:
0

α(P RQR0 ) = ab, whence we obtain α(4P QR) = 21 ab.

Theorem 96. The area of a triangle is one-half the product of any side by the
altitude to that side.

Proof. Let ∆ be an arbitrary triangle. Choose one of its sides and let b and
h be the lengths of that side and of the corresponding altitude. If the foot of
the altitude coincides with an endpoint of the side considered, triangle ∆ is a
right triangle and Theorem 95 determines its area α(∆) = 12 bh. If the foot of
the altitude lies between the endpoints of the side considered, then this altitude
triangulates ∆ into two (non-overlapping) right triangles ∆1 and ∆2 with a
common leg h and respective legs b1 and b2 with b1 + b2 = b, as shown. By
Theorem 95, α(∆1 ) = 12 b1 h, α(∆2 ) = 21 b2 h and by Axiom 16, we can write:
α(∆) = α(∆1 ) + α(∆2 ) = 12 (b1 + b2 )h = 21 bh. Finally, if the foot of the altitude
falls “outside” the base considered, a similar argument using Theorem 95 and
Axiom 16 leads to the desired result. The details of this proof are left as a
straightforward exercise.
 141

From the proof of Theorem 96 it should be clear that the three possible half
products of bases by corresponding altitudes, that is the three numbers 12 b1 h1 ,
1 1
2 b2 h2 , 2 b3 h3 are equal because each of them expresses the area of the triangle
considered. The following corollary is readily proved:

Corollary. If two triangles have equal altitudes and equal bases, they have
equal areas. If two triangles have equal altitudes, the ratio of their areas is
equal to the ratio of their bases. If two triangles have equal bases, the ratio of
their areas is equal to the ratio of their altitudes.

The area of a polygonal region can now be calculated: by Definition 57, this
region can be somehow triangulated into a finite number of non-overlapping
triangular regions whose areas can be calculated by Theorem 96. Theorem
93 ensures that the area calculated for this polygonal region via Axiom 16 and
Theorem 96 does not depend on the particular triangulation used. In particular,
we can easily prove the following simple result:

Theorem 97. The area of a parallelogram is the product of any side (base)
and the corresponding altitude.

Proof. Triangulate the parallelogram P QRS into two congruent non-overlapping

triangles by drawing the diagonal QS. Choose side P Q for base with length b
and let SH be the altitude to this side from vertex, say, S and let SH = h. A
straightforward application of Theorem 96 together with Axiom 16 then yields:
α(P QRS) = bh.
142 CHAPTER 7. ON AREA

Notice that if we had drawn the altitude to the chosen base from vertex
R instead of from vertex S, or if we had considered another side b0 and its
corresponding altitude h0 , as shown, the area of the parallelogram would still
be equal to bh = b0 h0 (justify!)

Theorem 98. The area of a trapezoid is one-half the product of its altitude
and the sum of its bases.

Proof. Divide the trapezoid by a diagonal into two triangles, as shown. The
altitudes of these triangles are equal to h (why?) and their areas are 21 b1 h and
1 1
2 b2 h. Axiom 16 and Theorem 93 then yield: α(trapezoid) = 2 (b1 + b2 )h.

The theory of area developed so far leads to the proof of the best-known
theorem of Euclidean Geometry, namely the Pythagorean Theorem. There are
numerous proofs for this most famous theorem—among others, Napoleon the
First and General James A. Garfield each proposed a new proof thereof—but
we prefer to reproduce an original Greek proof:

Theorem 99. In a right triangle, the square of the hypotenuse is equal to the
sum of the squares of the legs.
 143

Proof. Consider a right triangle ∆ with side lengths a, b, c with, say, a < b < c.
Construct a square ABCD with side length a + b (prove that this construction
is possible!) and consider four congruent right triangles (S.A.S.) with legs a and
b, as shown. Each of these four triangles is itself congruent to the original right
triangle (why?) and their hypotenuses are equal to c. Then, the quadrilateral
P QRS is a square with side length c (why?). Hence, by Axioms 16 and 17
and Theorem 95, we have: α(ABCD) = α(P QRS) + 4α(∆), or, equivalently,
(a + b)2 = c2 + 4( 21 ab) or a2 + 2ab + b2 = c2 + 2ab. Therefore, a2 + b2 = c2 which
was to be proved.

The converse of the Pythagorean Theorem is also valid:

Theorem 100. In a triangle, if the square of one side is equal to the sum of
the squares of the other two sides, then the triangle is a right triangle with the
first side for hypotenuse.

Proof. Let 4ABC be a triangle with c2 = a2 + b2 , as shown. Let 4A0 B 0 C 0 be a

right triangle with right angle at C 0 and legs A0 B 0 = a, A0 C 0 = b (show that
triangle 4A0 B 0 C 0 can be constructed!). Let the hypotenuse A0 B 0 of 4A0 B 0 C 0
be equal to d. Then, by Theorem 99, d2 = a2 + b2 , i.e. d2 = c2 and hence, d = c
since lengths of segments are positive numbers. Then, 4ABC ∼ = 4A0 B 0 C 0
∼ 0
(S.S.S.) and ]C = ]C as corresponding parts.

Two useful corollaries to the Pythagorean Theorem follow:

144 CHAPTER 7. ON AREA
√
Corollary. A right triangle is isosceles if and only if its hypotenuse is 2 times
as long as its leg.
Corollary. The hypotenuse of a right triangle is twice as long as the shorter
leg if and only if the measures of the acute angles of the triangle are 30◦ and
60◦ .

The proof of the first corollary is a straightforward consequence of the

Pythagorean Theorem while that of the second is easily obtained from a mere
examination of the figure shown.
According to Theorem 96, to calculate the area of a triangle when the lengths
of its sides are given, it is necessary to choose one side and compute the length
of the altitude to this side. The following theorem, attributed to Heron of
Alexandria (around 50 A.D.), provides an explicit formula for the area of a
triangle with given side lengths:
Theorem 101. Let a, b, c be the lengths of the sidesp of triangle 4ABC. Then
the area of this triangle is given by α(4ABC) = s(s − a)(s − b)(s − c) where
s is the half-perimeter of the triangle: s = 21 (a + b + c).

Proof. Consider the altitude AN of 4ABC and let AN = h, BN = x. Then

N C = |a−x|, and in 4ABN : h2 = c2 −x2 , while in 4ACN : h2 = b2 −(a−x)2 .
Since (a − x)2 = |a − x|2 , one has: c2 − x2 = b2 − (a − x)2 = b2 − a2 − x2 + 2ax
1
or x = 2a (a2 − b2 + c2 ). Next recall that h2 = c2 − x2 = (c − x)(c + x), so that
2 1
h = 4a2 [(a + b + c)(a − b + c)(a + b − c)(b − a + c)]. Since 2s = a + b + c, then:
a − b + c = 2s − 2b, a + b − c = 2s − 2c, b − a + c = 2s − 2a and consequently:
h2 = 4a12 [(2s)(2s − 2b)(2s − 2c)(2s − 2a)] = a12 [(4s)(s − a)(s − b)(s − c)]. Then
p
α(4ABC) = 21 ah = s(s − a)(s − b)(s − c).
Finally, and for the sake of completeness, we mention an intuitively evident
generalization of Theorem 93:
 145

Two (convex) polygons with the same number of vertices are called congruent
if there is a one-to-one correspondence between their vertices such that corre-
sponding angles are congruent and corresponding sides are congruent. Using
triangulation and Theorem 93, it is then easily proved that congruent polygons
have equal areas.
146 CHAPTER 7. ON AREA

Exercises
1. Prove that a median of a triangle divides the triangle into two triangu-
lar regions each having an area equal to one-half the area of the initial
triangular region.

2. In the figure shown, CQ = QD. Prove that α(4ABC) = α(4ABD).

3. If ABCD is a square with side length s, calculate the area of the star-
shaped polygonal region shown in terms of s and b. The sides (and the
vertex angles) of this polygonal region are congruent.

4. Prove that the diagonals of a parallelogram divide it into four triangular

regions which have equal areas.

5. Prove the following theorem:

If the diagonals of a convex tetragon (quadrilateral) ABCD are perpen-
dicular, the area of the tetragonal region is equal to one-half the product
of the lengths of the diagonals.
 147

6. Would the Theorem of Problem 5 remain valid if the quadrilateral consid-

ered were not convex, such as the quadrilateral shown in the below figure?
Justify your answer.

7. If two medians AD and BE of triangle 4ABC intersect at G, prove that

α(4AEG) = α(4BDG). Determine the ratio of the of the area of 4BDG
to the area of 4ABC.
Hint: Use the other median CF .

8. Assume that AB is a fixed segment in plane Π and P is an arbitrary point

←→
in Π not incident with line AB. Determine the location (locus) of all
points X in Π for which α(4XAB) = α(4P AB). Determine the location
of all points X in space for which the latter equality holds.
148 CHAPTER 7. ON AREA

9. A line cuts a rectangular region into two regions of equal area. Prove that
this line must contain the intersection of the diagonals of the rectangle.

10. With right angles and segments as marked in the figure, find AY , AZ and
AB. If this construction pattern is pursued with BC = 1 and m ]CBA =
90◦ , what would be the length of AC? If this construction is continued in
the same manner, what emerging pattern can one identify?

11. In 4ABC, AB = 14, BC = 15, AC = 13. Find the length hc of the

altitude from vertex C to AB. Find the length ha of the altitude from
vertex A to BC.
 149

12. In triangle 4ABC, angle ]B is obtuse and AB = 6, BC = 14, AC = 18,

as shown. Determine the length of the altitude hc from vertex C to side
AB.

13. In the rhombus ABCD, AC = 6 and BD = 4. Determine the length of

the perpendicular from any vertex to either one of the opposite sides.

14. If the lengths of the legs of a right triangle 4ABC are a and b, as shown,
calculate the length of the altitude h to the hypotenuse in terms of a and
b.
150 CHAPTER 7. ON AREA

15. In triangle 4ABC, angle ]B is a right angle, AC = CD, and AB =

BC = 1. Determine the measure of angles ]ADC and ]DAB.

16. Triangle 4ABC is isosceles with CA = CB. The medians AP and BQ

are perpendicular to one another at point S. Let d be the length of SP .
Determine the area of the polygonal regions ASB, ABC and QSP C in
terms of d.

17. The four faces of the regular “pyramid” (tetrahedron) ABCD are congru-
ent equilateral triangles with side length equal to 2. If R and S are the
midpoints of sides AB and CD, respectively, prove that segment RS is
perpendicular to both AB and CD.
 151

18. In the figure shown, AB is perpendicular to plane Π. Triangle 4BF H lies

in plane Π and angle ]HF B is a right angle. Furthermore, m ]F HB =
30◦ and AB = BH = 6. Determine the lengths and the measures of as
many segments and angles of the figure as you can.

√
19. In triangle 4ABC, m ]A = 30◦ , AC = 4, AB = 3 3. Calculate the
length of side BC and determine whether 4ABC is a right triangle.

20. In the figure shown, DF is perpendicular to F B, CF is perpendicular to

F A, F D = F C, DB = CA.
Prove that triangle 4F AB is isosceles.

√
21. The area of an equilateral triangle is 9 3. Find the length of its side and
of its altitude.
152 CHAPTER 7. ON AREA

22. The figure shown represents a cube. Consider the plane determined by the
vertices A, C and F . If the length of the side of the cube is 9, determine
the length of AC, the measure of angle ]F AC and the area of 4F AC.

23. The figure shown represents a regular tetrahedron: its faces are congruent
equilateral triangles with side length equal to e. Segment N M joins the
midpoints N and M of sides AB and CD, respectively and is perpendicu-
lar to both of these sides (see Problem #17). Calculate the length of N M
and of the altitude AH of the tetrahedron.

24. Given any two arbitrary coplanar rectangles, how can a single line be
drawn which will separate each rectangular region into two regions of
equal area?

25. ABXY is a square with side length 6 in the plane Π. Rectangle ABCD
is the projection of the square ABXY onto plane Π0 . The measure of
the angle between planes Π and Π0 is 60◦ . What is the area of rectangle
ABCD?
 153

26. Square ABCD is given with points E and F as shown so that EC is per-
pendicular to F C. If α(4CEF ) = 200 and α(ABCD) = 256, determine
BE.

27. If W , X, Y , and Z are midpoints of the sides of square ABCD, as shown,

determine the ratio of the area of this square to the area of the tetragonal
region P QRS.

28. Two (convex) polygons with the same number of vertices are called congru-
154 CHAPTER 7. ON AREA

ent if there is a one-to-one corespondence between their vertices such that

corresponding angles are congruent and corresponding sides are congru-
ent. Use triangulation and Theorem 93 to prove that congruent polygons
have equal areas.
Chapter 8

On Similarity

The notion of similarity is a familiar one: two shapes are similar if they look
“alike” but do not have the same size. For example, two segments, two circles,
or two equilateral triangles look similar. But two, say, isosceles triangles or
two rectangles are not necessarily similar. Thus, similar figures are related to
one another by certain specific relationships. As usual, the precise definition
of similarity will be given first for the simplest of all geometric shapes, namely
triangles, and then generalized to more complex figures.

Definition 58. If a one-to-one correspondence can be established between the

vertices of two triangles 4ABC and 4A0 B 0 C 0 such that corresponding angles
are congruent and the lengths of corresponding sides are proportional, then the
correspondence is a similarity and the two triangles are said to be similar. We
write 4ABC ∼ 4A0 B 0 C 0 when the correspondence is as indicated.

It goes without saying that the correspondence between vertices should

match the correspondence of sides. According to the above definition, the cor-
respondence A, B, C A0 , B 0 , C 0 between the vertices of triangles 4ABC and
4A0 B 0 C 0 shown, will be a similarity if:

(i) corresponding angles are congruent, i.e. ]A ∼ = ]A0 , ]B ∼= ]B 0 , ]C ∼

=
]C 0 (in fact, only two of the three pairs of corresponding angle need be
congruent, since Theorem 71 on the angle sum of triangles guarantees
the congruence of the third pair), and

155
156 CHAPTER 8. ON SIMILARITY

(ii) the lengths a, b, c and a0 , b0 , c0 of corresponding sides are proportional, i.e.

a b c
a0 = b0 = c0 or any other expression equivalent to this proportionality.

The ratio of the lengths of two corresponding sides of two similar triangles
is called their ratio of similarity (or similitude). Notice that both requirements
stated in Definition 58 are necessary in order to extend the concept of similarity
to polygonal shapes. The following examples show that two figures satisfying
only one of the above conditions may be very dissimilar:

For triangles, however, we shall see shortly that only one of the two re-
quirements of Definition 58 is sufficient to guarantee the similarity of the two
triangles.
The study of similarity begins with the following theorem:
Theorem 102. If a line parallel to one side of a triangle intersects the other
two sides (or their extensions) in distinct points, then it cuts off segments whose
lengths are proportional to the lengths of the corresponding sides.

Proof. We prove the theorem only for the case where the intersections of the
parallel to one side with the other two sides lie between triangle vertices, and
leave the other case as a straightforward exercise.
Consider triangle 4ABC and let points D and E be on sides AB and AC
←→
such that DE is parallel to BC. In triangles 4ADE and 4BDE consider AD
and BD as the bases and the perpendicular from E to AB as their common al-
titude. By the Corollary to Theorem 96, we can write: α(4BDE)/α(4ADE) =
 157

BD
AD . Repeat with triangle 4AED and 4CED to obtain α(4CED)/α(4ADE) =
CE ←→ ←→
AE . Now, since DE and BC are parallel, points B and C are equidistant from
←→
DE and, hence, the altitudes of triangles 4BDE and 4CDE from vertices
B and C, respectively, to their common base ED are congruent, so that by
the Corollary to Theorem 96, α(4BDE) = α(4CDE). It then follows that
BD CE AB AC
AD = AE , and consequently, (BD + AD)/AD = (CE + EA)/AE or AD = AE ,
which was to be proved.
The converse of Theorem 102 is also true:
Theorem 103. If a line intersects two sides of a triangle (or their extensions)
and cuts off segments whose lengths are proportional to the lengths of the cor-
responding sides, then this line is parallel to the third side of the triangle.

Proof. In triangle 4ABC, let D be a point on side AB between vertices A and

B, and let E be a point on side AC between vertices A and C. The second
case where points D and E are located on the extensions of sides AB and AC is
AB AC ←→ ←→
again left as an exercise. Suppose AD = AE and that BC is not parallel to DE.
←−→ ←→
We may then introduce the (unique) parallel BC 0 to DE through, say, vertex
←→ 0
B (Axiom 15), intersecting AE in C 0 . By Theorem 102, we have AD AB
= AC
AE .
0
AB AC
But by hypothesis AD = AE , so that by transitivity, AC AC
AE = AE and therefore
AC = AC. By the Point Plotting Theorem, it follows that C and C 0 are the
0
←→ ←→
same point and BC is parallel to DE.
We are now ready for the proof of our earlier assertion concerning sufficient
criteria for the similarity of triangles.
Theorem 104 (A.A.A. Similarity Theorem). If there exists a correspondence
between two triangles such that corresponding angles are congruent, then the
triangles are similar.
158 CHAPTER 8. ON SIMILARITY

Proof. Let there be a correspondence between the vertices of the triangles

4ABC and 4DEF such that ]A ∼ = ]D, ]B ∼ = ]E and ]C ∼ = ]F . Let points
−−→ −→
E and F be selected on rays AB and AC such that AE 0 = DE and AF 0 = DF
0 0

(Theorem 3). By the S.A.S. Axiom, 4AE 0 F 0 ∼ = 4DEF so that ]AE 0 F 0 ∼ = ]B

←− →
0 0 ←→ ←−→
0 0 ←→
and E F and BC are parallel or coincide (Theorem 64). If E F and BC coin-
cide, then 4ABC ∼ = 4DEF and, trivially, DE AB AC
= DF = BC
EF = 1 so that these
←− →
0 0 ←→
triangles are also similar. If E F and BC are parallel, then by Theorem 102,
AB AC 0 0
we have AE 0 = AF 0 , and since, by construction, AE = DE and AF = DF ,
AB AC AB BC
it follows that DE = DF . The proportion DE = EF is proved in exactly the
AB AC
same manner, leading to DE = DF = BC
EF and thus completing the proof of the
theorem.

Notice that the A.A.A. Similarity Theorem should be called the A.A. Sim-
ilarity Theorem since by Theorem 71, the angle sum of a triangle is 180◦ , and
the congruence of two pairs of angles of two triangles entails the congruence of
the third pair. Furthermore, it is easy to show that if a line parallel to a side
of a triangle intersects the other two sides in distinct points, then it cuts off a
triangle similar to the given triangle.

Theorem 105 (S.A.S. Similarity Theorem). If there is a correspondence be-

tween the vertices of two triangles such that the lengths of two pairs of corre-
sponding sides are proportional, and the included angles are congruent, then
the triangles are similar.

Proof. In triangles 4ABC and 4DEF , let ]A = ∼ ]D and AB = AC . Using

DE
−−→EF −→
the Point Plotting Theorem, select points E 0 and F 0 on rays AB and AC,
respectively, such that AE 0 = DE and AF 0 = DF . Then, AE AB
0 = AFAC
0.
←− →
0 0 ← → ∼ 0 0
By Theorem 103, E F is parallel to BC and therefore, ]B = ]AE F and
]C ∼= ]AF 0 E 0 (corresponding angles). Hence, 4AE 0 F 0 ∼
= 4DEF (S.A.S.)
and consequently, ]AE 0 F 0 ∼
= ]E ∼
= ]B and ]AF 0 E 0 ∼= ]F ∼= ]C. Since by
hypothesis, ]A ∼ = ]D, the A.A.A. (or A.A.) Similarity Theorem guarantees
that 4ABC ∼ 4DEF .

Theorem 106 (S.S.S. Similarity Theorem). If there is a correspondence be-

tween the vertices of two triangles such that the lengths of corresponding sides
are proportional, then the triangles are similar.
 159

AB AC
Proof. Assume that in the two triangles 4ABC and 4DEF , DE = DF = BC
EF .
0 0 −−→ −→ 0
Once more, choose points E and F on rays AB and AC such that AE = DE
←− → ←→
and AF 0 = DF . Then AE AB AC
0 = AF 0 , so that by Theorem 99, E F
0 0
and BC are
parallel, and hence, ]AF 0 E 0 ∼ = ]C. Therefore, triangles 4ABC and 4AE 0 F 0
are similar and E 0 F 0 = AE 0 . It then follows from the hypothesis that E 0 F 0 =
BC AB

EF and thus 4AE 0 F 0 ∼ = 4DEF (S.S.S.), so that ]AE 0 F 0 ∼ = ]E ∼ = ]B and

0 0 ∼ ∼
]AF E = ]F = ]C (by transitivity). The A.A. Similarity Theorem then
implies that 4ABC ∼ 4DEF .

An interesting property of the bisectors of the internal and external angles

at a vertex of a triangle is contained in the following theorem:
←→
Theorem 107. If the bisector of angle A of triangle 4ABC meets BC at point
CD AC
D, then DB = AB . If the external bisector of angle A of triangle 4ABC meets
←→ 0 CD 0 AC
BC at point D , then BD 0 = AB .

Note that if 4ABC is isosceles with AB = AC, then the bisector of the
external angle at vertex A is parallel to the opposite side BC and the above
theorem becomes empty.

Proof. We prove the second assertion first. From vertex B construct the line
parallel to the external bisector AD0 of angle ]A meeting side AC at F . Then,
by the remark following the proof of Theorem 100, 4CF B ∼ 4CAD0 , so that
CA CD 0 0 AF BD 0
CF = CB and hence, (CA − CF )/CF = (CD − CB)/CB or CF = CB . On
the other hand, ]AF B ∼ = ]EAD0 (corresponding angles), ]EAD0 ∼ = ]BAD0
0
(AD is the bisector of the external angle at vertex A), and by transitivity,
]AF B ∼ = ]ABF and0 triangle 4AF B is isosceles with AB = AF . Now, the
AF 0 0
proportion CF = BD
CB is equivalent with the proportion BD /(BD + CB) =
BD 0 AF AB
AF/(AF + F C) i.e. CD0 = AC = AC . To prove the first assertion, we consider
the parallel through vertex B to the internal bisector AD of angle ]A and prove
CD AC AC
in the same manner that DB = AE = AB .
160 CHAPTER 8. ON SIMILARITY

In short, this theorem states that the bisector of an angle of a (scalene)

triangle divides the opposite side in the ratio of the other two sides of the
triangle. By reversing the steps of the preceding argument, one easily proves
the converse of Theorem 107.
Right triangles can be divided into similar triangles which are useful for
computational purposes:

Theorem 108. In any right triangle, the altitude to the hypotenuse divides
the triangle into two right triangles which are similar to one another as well as
to the original right triangle.

Proof. In triangle 4ABC with right angle at vertex C, let CD be the altitude
to the hypotenuse AB. Using the notation of the figure for brevity, we have:
m ]a + m ]b = 90◦ and m ]a + m ]b0 = 90◦ , so that ]b ∼ = ]b0 . Moreover,
m ]C = m ]d = 90◦ . Therefore, by the A.A.A. Similarity Theorem, 4ACD ∼
4ABC. The other half of the theorem is proved in the same manner.

Definition 59.
√ Given two positive real numbers a and b, the geometric mean
of a and b is ab.

Corollary. In a right triangle, the length of the altitude to the hypotenuse is

the geometric mean of the lengths of the segments into which it divides the
hypotenuse. Moreover, the length of either leg of the triangle is the geometric
mean of the length of the hypotenuse and the length of the segment of the
hypotenuse adjacent to that leg.
AD CD
Proof. By Theorem 108, 4ADC ∼ 4CDB. Therefore, CD = BD i.e. CD2 =
√
AD·BD so that CD = AD · BD is indeed the geometric mean of AD √ and BD.
Also 4ADC ∼ 4ABC, so AD AC = AC
AB or AC 2
= AD · AB or AC = AD · AB.
2
Finally, the
√ similarity of triangles 4BDC and 4ABC yields BC = BD · BA
or BC = BD · BA.

Theorem 109. The ratio of the areas of two similar triangles is the square of
the ratio of any two corresponding sides, i.e. of their similarity ratio.
 161

0 0
Proof. Let 4ABC ∼ 4A0 B 0 C 0 . Then, with reference to the figure, aa = bb =
c0
c = k with the positive number k being the ratio of similarity of these tri-
angles. Then, a0 = ka, b0 = kb, and c0 = kc. Let BD with BD = h be
the altitude from B to AC and, similarly, let B 0 D0 be the altitude from B 0
to A0 C 0 with B 0 D0 = h0 . Since 4ABD and 4A0 B 0 D0 are right triangles and
0 0
]A ∼= ]A0 by hypothesis, then 4ABD ∼ 4A0 B 0 D0 and hh = bb = k. Then:
α(4ABD) = 21 bh and α(4A0 B 0 D0 ) = 12 b0 h0 = 12 (kb)(kh) = k 2 [ 21 bh]. Therefore,
α(4A0 B 0 D0 )/α(4ABD) = k 2 , which was to be proved.
Definition 58 can be extended to polygons in a natural way:
Definition 60. Two polygons with the same number of vertices are similar
if there is a one-to-one correspondence between their vertices such that

(i) corresponding angles are congruent, and

(ii) the lengths of corresponding sides are proportional.

It is not difficult to prove that Theorem 109 remains valid for similar poly-
gons.
Many of the theorems that we have enunciated and proved so far were more
or less known to ancient Greek geometers. We conclude this chapter with a
theorem that is a relatively recent addition to the corpus of Euclidean theorems
and is due to the Italian mathematician Giovanni Ceva (around 1678 A.D.).
This theorem is worth mentioning both for its simplicity and for the elegance
of some proofs which can be produced with its help.
Definition 61. A cevian of a given triangle is any segment which connects a
triangle vertex to an arbitrary point on its opposite side.
Theorem 110 (Ceva’s Theorem). In a triangle 4ABC, three cevians AY , BZ,
AX BY CZ
CX are concurrent if and only if XB Y C ZA = 1.

Proof. Suppose that the three cevians AY , BZ, CX are concurrent at point P .
Through vertices, say, B and C draw lines parallel to AY meeting BZ and CX at
162 CHAPTER 8. ON SIMILARITY

E and D, respectively. Then, 4AXP ∼ 4BXD and 4AZP ∼ 4CZE (why?).

AX AP
Therefore, BX = BD and CZ CE
AZ = AP . On the other hand, 4BP Y ∼ 4BEC
BY
and 4CP Y ∼ 4CDB, so that BC = YCE P
and YBC BD
C = Y P . Multiplication of the
respective sides of these four proportions yields:
AX CZ BY BC AP CE Y P BD
=
BX AZ BC Y C BD AP CE Y P
which after simplification (since all segment lengths are positive numbers) gives
AX BY CZ
the desired result: XB Y C ZA = 1.

To prove the converse, suppose that in triangle 4ABC, the three cevians
AX BY CZ
AY , BZ, CX satisfy XB Y C ZA = 1. Let point P be the intersection of the
←→
cevians CX and BZ. Consider line AP and let its intersection with BC be Y 0 .
If Y 0 and Y are the same point, then we are done. Suppose that Y and Y 0 are
AX BY 0 CZ
distinct points. Then by the preceding argument, XB Y 0 C ZA = 1. Hence, using
0
AX BY CZ AX BY CZ BY BY 0
the hypothesis, XB Y C ZA = XB Y 0 C ZA , that is: Y C = Y 0 C and consequently,
Y C = Y 0 C. It follows from the Ruler Axiom (Axiom 3), that Y and Y 0 are
the same point and, hence, AY and AY 0 are the same line. Therefore, the
three cevians AY , BZ, CX are concurrent. This completes the proof of Ceva’s
Theorem.
One immediate corollary to Theorem 110 is that the medians of a triangle
are concurrent (why?). Another application of Ceva’s Theorem is the elegant
proof of the concurrence of the internal bisectors of the angles of a triangle:
Theorem 111. The internal bisectors of the angles of a triangle are concurrent
at a point called the incenter of the triangle.

Proof. In triangle 4ABC, let AX, BY , CZ be the internal bisectors of the

angles ]A, ]B, ]C, respectively. Then, by virtue of Theorem 107, we can
 163

AZ CA BX AB CY BC
write: ZB = CB , XC = CA and YA = AB . Multiplying these three proportions
together yields:
AZ BX CY CA AB CB
= =1
ZB XC Y A CB CA AB
Therfore, Ceva’s Theorem ensures that the three cevians AX, BY , CZ are
concurrent.
164 CHAPTER 8. ON SIMILARITY

Exercises
1. Prove that the triangle whose vertices are the midpoints of the sides of a
given triangle is similar to the given triangle.

2. Three lots extend from Packard Street to State Street, as shown. The
side lines make right angles with State Street, and the total frontage on
Packard Street is 3600 . Find the frontage of each lot on Packard Street.

3. Segments BR, CS, DT are perpendicular to BD. Using the notation of

the diagram, prove: x1 + y1 = z1 .

4. In the figure shown, DB is perpendicular to AC and DQ = BQ = 2AQ =

1
2 QC.

Prove that AD is perpendicular to DC.

 165

5. In the diagram shown, W S and LQ are medians of 4W RT and 4AM L,

respectively. Furthermore, RW RT WS
AL = AM = LQ .
Prove that 4RW T ∼ 4ALM .

6. In the figure shown, RA and F B are perpendicular to AB, and RH is

perpendicular to AF .
Prove: HR · BF = BA · HA.

7. In the quadrilateral RST Q shown, RS is parallel to QT and the two

triangles 4QXR and 4T XS are similar.
Prove: QR = T S.

8. Prove the following theorem:

In similar triangles, the lengths of corresponding medians have the same
ratio as corresponding sides.
166 CHAPTER 8. ON SIMILARITY

9. Prove the following theorem:

In similar triangles, the lengths of corresponding altitudes have the same
ratio as corresponding sides.

10. It is known that a S.S.A. congruence criterion is in general not valid for
triangles. Is the following statement true or false? Explain.
If a correspondence exists between the vertices of two triangles such that
the lengths of two sides of one triangle are proportional to the lengths of
the corresponding two sides of the other, and the angles opposite a pair
of corresponding sides are congruent, then the triangles are similar.

11. Triangle 4EDF is isosceles with DE = DF . Triangle 4ABC is such

that E and F lie between A and C, CB is parallel to ED, and A, B, D
are collinear.
What statements concerning similarity and proportion can be made about
triangles 4ABC, 4ADE and 4ADF ?

12. In the parallelogram ABCD shown, G is an arbitrary point on side DC.

−−→ −−→
Ray BG intersects the diagonal AC of the parallelogram at E and ray AD
at F . Prove that EB is the geometric mean of EG and EF .

13. How long must a side of an equilateral triangle be in order that its area
shall be twice that of an equilateral triangle whose side is 10 cm?
 167

14. If similar triangles are drawn on the side and on the altitude of an equi-
lateral triangle, so that the side and altitude are corresponding sides of
the triangles, prove that their areas are in the ratio of 4/3 to each other.

15. In the right triangle 4ABC, CH is the altitude to the hypotenuse AB.
If α1 , α2 , α3 represent the areas of triangles 4ABC, 4ACH, 4BCH,
respectively, then α1 = α2 + α3 or aa12 + aa31 = 1. Use this relation and the
similarity of those triangles to concoct another proof of the Pythagorean
Theorem.

16. A triangular lot 4ABC has sides with lengths 130 ft., 140 ft. and 150 ft.
The length of the altitude from C is 120 ft., as shown. A fence (shown as
a dashed line) is to be erected perpendicular to AB so that the area of the
lot is equally divided. How far from A along side AB should this fence be
set?

17. In the two triangles shown, ]A ∼= ]R, AC = b, AB = c, and RS = k.

Find a in terms of b, c, and k.

Hint: Let D be the foot of the altitude from C and x, y, h as indicated in

the figure. Express a2 in terms of h and y, and express h and y in terms
of x, b, c, and use the similarity 4ADC ∼ 4RST to express x in terms
of b and k.
168 CHAPTER 8. ON SIMILARITY

18. Let ma be the length of the median to the side BC of triangle 4ABC, and
let BC = a, AC = b, AB = c. Prove that m2a = 12 (b2 + c2 − 12 a2 ). Next,
let mb and mc be the lengths of the other two medians of the triangle.
Use the previous relation to prove:

3 2
m2a + m2b + m2c = (a + b2 + c2 )
4
19. In the right triangle 4ABC with CF the altitude to the hypotenuse and
lengths as indicated in the figure, find the values of x, y, and w.

20. Join the vertices of 4ABC to a point R outside the triangle. Through an
←→ ←→
arbitrary point X on AR, draw line XY parallel to AB meeting BR at Y .
←→ ←→
Draw line Y Z parallel to BC meeting RC at Z. Prove: 4ABC ∼ 4XY Z.
21. The perimeter of 4ABC is 15 cm. The bisector of ]A meets BC at X,
and the bisector of ]C meets AB at Z. If BX = 3 cm and CX = 2 cm,
find AZ.
Chapter 9

On Circles and Spheres

9.1 Tangent Lines and Planes

In this chapter we introduce and study geometric objects that are not made up
of planes, lines, rays, and segments. Such objects are usually called “curved”
and we shall restrict our investigations to the simplest and, in a sense, most
“perfect” curved entities, namely, to circles and spheres.

Definition 62. A sphere is the set of all points in space whose distance from
a given point, called the center, is equal to a given positive number, called
the radius of the sphere. A circle is the set of all points in a given plane whose
distance from a given point, called the center, is equal to a given positive number
called the radius of the circle. Spheres or circles which share a common center
are called concentric.

An immediate consequence of this definition is that the intersection of a

sphere with any plane through the center of the sphere is a circle with
the same center and radius as the sphere. Such circles are called great circles
of the sphere and we shall see that they play a fundamental role in our later
study of the so-called Elliptic Geometry which is one type of a Non-Euclidean
Geometry (cf. Chapter 12).
Given a sphere or a circle with center P and radius r, if Q is any point of
the sphere or circle, the segment P Q is a radius of the sphere/circle and Q is
its outer end. Note that the word “radius” is used in two ways: the radius of a
sphere/circle is the positive number r, whereas a radius of a sphere/circle is a
segment P Q.

Definition 63. If P and Q are any two points of a sphere/circle, then the
segment P Q is a chord of the sphere/circle, and the line containing a chord
is called a secant of the sphere/circle. A diameter is a chord containing the
center of the sphere/circle and its length is twice the radius. The interior of a
sphere/circle is the set of all points in space/plane of the sphere/circle whose

169
170 CHAPTER 9. ON CIRCLES AND SPHERES

distance from the center is less than the radius. The exterior of a sphere/circle
is defined accordingly.

It follows from these definitions that a point is either in the interior of a

sphere, or on the sphere, or in the exterior of the sphere. Similarly, a point in
the plane of a given circle is either in the interior of the circle, or on the circle,
or in its exterior (see Law of Trichotomy, Rule 10 in Appendix C).

Definition 64. Given a circle and a line in the same plane, if the line and the
circle have one and only one point in common, then the line is called a tangent
to the circle and the common point is called the point of tangency or point of
contact.

According to this definition and Definition 63, if the line in Definition 64

intersects the circle in more than one point, it may be called a secant of the
circle.

Consider a circle with center P and a line l in the plane of the circle. Let F
be the foot of the perpendicular from P to l. It is apparent that if F is in the
exterior of the circle, then so are all points of l, and that if F is in the interior
of the circle, then l is a secant with some points in the interior and others in the
exterior of the circle. Finally, if F is on the circle, then the line l is a tangent to
the circle. These intuitively evident observations lead to the statement of the
following theorem:

Theorem 112 (The Circle-Line Theorem). Given a line l in the plane of a

circle with center P and radius r, let F be the foot of the perpendicular from
P to l. Then either

1. Every point of l is outside the circle, or

2. F is on the circle and l is tangent to the circle at F , or

3. F is in the interior of the circle and l intersects the circle in exactly two
points which are equidistant from F .
9.1. TANGENT LINES AND PLANES 171

Proof. We first show that if F is outside the circle, then (1) holds. In this case,
P F > r and by Theorem 44, P F is the shortest segment joining P to l. If Q is
any other point on l, then P Q > P F , so that P Q > r and Q is in the exterior
of the circle.
Next, if F is on the circle, then P F = r. Let Q be another point of l. By
Theorem 44, P Q > r and Q is in the exterior of the circle. Thus, line l and the
circle (are coplanar and) have only the point F in common, and by Definition
64, l is tangent to the circle at F . This proves (2).
Finally, suppose that F is in the interior of the circle and let P F = s. Then
s < r. If Q is both on l and on the circle, then triangle 4P F Q is a right
√ triangle,
and by the Pythagorean Theorem: r2 = s2 + F Q2 . Hence, F Q = r2 − s2 and
any point Q common to l and the circle must satisfy this relation. Conversely,
if any point Q on l satisfies the previous relation, it will be at distance r from
P (why?). Now r2 − s2 > 0 because r > s. By the Euclidean Completeness
2 2
Axiom √ (see Appendix III), the positive number r − s has a positive square
2 2
root r − s . By the Point Plotting Theorem, there are exactly two points on
l, one on each of the two rays along l emanating from F and, by the preceding
argument, these are the only two points common to l and the circle. Note that
if l passes through the center P of the circle, then P and F are the same point
and P Q = F Q = r, so that l contains a diameter of the circle and there are
again two points common to l and the circle, as before.

The Line-Circle Theorem is noteworthy because to prove it we had to use

the Euclidean Completeness Axiom for the first time. If the plane “had holes
or gaps” in it, then the line could “hit” one of those “missing points” and the
172 CHAPTER 9. ON CIRCLES AND SPHERES

theorem would fail. In a purely geometric (synthetic) approach, Theorem 112

should be taken as a separate axiom. Several important facts follow as corollaries
to Theorem 112. They are enumerated below and their (straightforward) proofs
are left as exercises:
Corollary 1. Every line tangent to a circle is perpendicular to the radius with
the point of tangency as endpoint.
Corollary 2. In the plane of a circle, any line perpendicular to a radius at its
outer end is tangent to the circle.
Corollary 3. Any perpendicular from the center of a circle to a chord bisects
the chord.
Corollary 4. The segment joining the midpoint of a chord to the center of a
circle is perpendicular to the chord.
Corollary 5. In the plane of a circle, the perpendicular bisector of a chord
passes through the center of the circle.
Corollary 6. If a line in the plane of a circle contains a point in the interior of
a circle, then it intersects the circle in exactly two points.
Definition 65. Circles which have equal radii are called congruent.
The distance of a chord from the circle center is naturally defined as the
distance of the center from the line containing that chord.
Theorem 113. In a given circle or in congruent circles, two chords are congru-
ent if and only if they are equidistant from the center.
Proof. The proof is left as an easy exercise.
Finally, the relationship between two coplanar circles can be classified in-
tuitively as:
Definition 66. Two coplanar circles are disjoint when the circles have no point
in common; intersecting if they have (at least) one point in common; and inter-
nally or externally tangent according to whether their centers lie on the same
side or on opposite sides of a common tangent line, as shown below.

We now move on to the relationship between planes and spheres and, in

a sense, generalize Theorem 112 to higher dimensions. To that end, we first
introduce the analogue of Definition 91:
9.1. TANGENT LINES AND PLANES 173

Definition 67. If a given sphere and a given plane have exactly one point in
common, then the plane is called a tangent plane and the common point is called
the point of tangency or point of contact. If the plane intersects the sphere in
more than one point, it is called a secant plane.

Theorem 114. Given a sphere with center P and radius r and a plane Π, let
F be the foot of the perpendicular from P onto Π. Then either

1. Every point of Π is in the exterior of the sphere, or

2. F is on the sphere and Π is tangent to the sphere at F , or

3. F is inside the sphere and Π intersects the sphere in a circle with center
at F .

Proof. The proof of (1) follows almost word for word the corresponding proof
of Part (1) in Theorem 112, using an obvious consequence of Exercise 29 of
Chapter Five instead of Theorem 44. The proof of (2) is also almost identical
(with a similar adjustment) to that of Part (2) in Theorem 112, and is left as
an exercise. Lastly, if F falls in the interior of the sphere, let the distance of
P from Π, i.e. the length of segment P F , be x. Then x < r and there exists
(at least) one point Q in Π which is at distance r from P (why?). Hence, Q
is a point common to both Π and the sphere. Since every line in Π through F
←→
is perpendicular to line P F , triangle√4P F Q has a right angle at F , and the
Pythagorean Theorem yields F Q = r2 − x2 and F Q is a positive number by
virtue of the Euclidean Completeness Axiom. Since Q is any point common to
Π and the sphere, then every other point common to Π and the sphere must
satisfy the same relation as point Q does.√ Therefore, such points must lie on
the circle in Π centered at F with radius r2 − x2 . We still have to show that
all points common to the plane and the sphere lie on this same circle, since one
might conceive points on the circle that are not common to the plane and the
sphere. To settle this issue,
√ suppose that point Q lies on the circle in plane Π
with center F and radius r2 − x2 . Then, as before, m ]P F Q = 90◦ , so that
P Q2 = x2 + (r2 − x2 ) = r2 and P Q = r. Therefore, Q lies on the sphere and
every point of the circle is indeed both in Π and on the sphere.

The following corollaries are again immediate consequences of Theorem 113

and their proofs are as easy as those of the Corollaries to Theorem 112:

Corollary 1. A plane tangent to a sphere is perpendicular to the sphere’s

radius with outer end at the point of tangency.
174 CHAPTER 9. ON CIRCLES AND SPHERES

Corollary 2. A plane perpendicular to a radius of a sphere at its outer end is

tangent to the sphere.

Corollary 3. A perpendicular from the center of a sphere to a chord bisects

the chord.

Corollary 4. The segment joining the center of a sphere to the midpoint of a

chord is perpendicular to the chord.

Corollary 5. The perpendicular from the center of a sphere to a secant plane

passes through the center of the circle in which the plane intersects the sphere.

Corollary 6. If a plane intersects a sphere in a circle, the segment joining the

center of the sphere to the center of this circle is perpendicular to the plane.

Corollary 7. If a plane intersects a sphere in a circle, then the line perpendic-

ular to the plane through the center of this circle passes through the center of
the sphere.

9.2 Arcs, Angles, Secants

For the remainder of this chapter we confine ourselves to the further study of
circles only.
The axioms introduced at the beginning of this course allowed for the com-
parison of two segments (via their lengths) or two angles (via their measures),
and for the construction of various points on a given ray (via the Point Plotting
Theorem), as well as the addition or subtraction of angles (via relevant Angle
Axioms). We now would like to be able to extend these operations to arcs of
circles too. To that end, several definitions are required.

Definition 68. In a given circle, a central angle is an angle whose vertex is the
center of the circle.

Let A and B be the points where the sides of the angle intersect the circle
with center P so that the central angle is ]AP B.

Definition 69. The minor arc AB ˜ is the set consisting of points A and B
together with all points of the circle that lie in the interior of angle ]AP B. The
major arc AB˜ is the set consisting of A and B together with all points of the
circle that lie in the exterior of angle ]AP B. If A and B are the endpoints of
9.2. ARCS, ANGLES, SECANTS 175

a diameter AB of the circle, then the two arcs AB ˜ consisting of points A and
←→
B together with all points of the circle that lie on a given side of line AB are
called semicircles. In all cases, A and B are called the endpoints of the arc.
Thus, an arc is either a minor arc or a major arc or a semicircle. But an arc
with endpoints A and B and denoted AB ˜ is always ambiguous, for even on one
and the same circle there are always two arcs with endpoints A and B. In an
intuitive sense, a major arc is “bigger” than a minor arc. Whenever necessary,
one may resolve this ambiguity by taking an arbitrary third point X of the arc
considered and denote that arc by AXB.
˘ In the diagram, AXB ˘ is a minor arc,
AY
˘ B is its corresponding major arc, and CAB
˘ and CY ˘ B are semicircles.

The relation between minor and major arcs is quantified in the next defini-
tion.

Definition 70. The degree measure m AXB

˘ of an arc AXB
˘ is defined in each
case as follows:

1. If AXB
˘ is a minor arc, then m AXB
˘ is the measure of the corresponding
central angle ]AP B.

2. If AXB ˘ = 180◦ .
˘ is a semicircle, then m AXB

3. If AXB ˘ = 360◦ − m ]AP B.

˘ is a major arc, then m AXB
176 CHAPTER 9. ON CIRCLES AND SPHERES

According to this definition, an arc is minor or major depending on its degree

measure being less than or larger than 180◦ . Henceforth, m AB˜ shall be called
the measure of arc AB, for short.
˜
The next theorem states that the measure for arcs is additive in the way we
expect it to be:

Theorem 115. If AB
˜ and BC ˜ are arcs of the same circle, having only the point
B is common, and if their union is an arc AC,
˜ then m AB ˜ + m BC ˜ = m ABC.˘

Proof. There are five cases to consider and the proof of each case is easy.

1. If arc ABC
˘ is a minor arc, then the theorem follows from the Angle
Addition Axiom.

2. If ABC
˘ is a semicircle, then the theorem follows from the Angle Supple-
ment Axiom.

3. If ABC
˘ is a major arc and the endpoints A and C are on opposite sides
of the diameter containing the endpoint B, then with reference to the
diagram: m AB
˙ 0 C = m AB
¯0 + m B¯ 0 C = u◦ + t◦ by case (1), and m B¯0C +

m CB
˜ = mB ˙0 CB = t◦ +s◦ = 180◦ by case (2). Similarly, m B
¯ 0 A+m AB
˜=
m B AB = u + r = 180 . Hence, s = 180 − t and r = 180 − u◦ .
˙0 ◦ ◦ ◦ ◦ ◦ ◦ ◦ ◦

Now, since ABC

˘ is a major arc, by Definition 69
9.2. ARCS, ANGLES, SECANTS 177

˘ = 360◦ −m ]AP C = 360◦ −(u◦ +t◦ ) = 180◦ −u◦ +180◦ −t◦ = s◦ +r◦ = m BC+m
m ABC ˜ AB,
˜

thus completing the proof of case (3).

4. ABC
˘ is a major arc and the endpoints A and C are on the same side
of the diameter that contains the endpoint B. The proof of this case is
similar to that of case (3) and is left as an exercise.

5. ABC
˘ is a major arc, and one of the arcs AB,
˜ AC
˜ is a semicircle. Suppose
AB is a semicircle, as shown. Then: m ABC = 360◦ − t◦ = 180◦ + 180◦ −
˜ ˘
t◦ = 180◦ + s◦ = m AB˜ + m BC.˜ This completes the proof of Theorem
115.

Definition 71. An angle is inscribed in an arc if:

178 CHAPTER 9. ON CIRCLES AND SPHERES

1. each of the two endpoints of the arc lie on one of the two rays of the angle,
and

2. the vertex of the angle is a point, but not an endpoint, of the arc.

According to this definition, angle ]ABC is inscribed in the dotted major

arc ABC
˘ or in the minor arc ABC, ˘ respectively, as shown.
The figures below illustrate various configurations for the vertex and sides
of an angle with respect to a circle and give a general idea of the concept of
the arc intercepted by an angle. In each of the cases, the angle intercepts the
indicated dotted arc.

In the third case, the angle intercepts the dotted arc and arc AXB.
˘
The precise definition of the idea conveyed by the figures above is as follows:

Definition 72. An angle intercepts an arc if:

1. the endpoints of the arc lie on the sides of the angle,

2. each side of the angle contains at least one endpoint of the arc, and

3. except for its endpoints, the arc lies in the interior of the angle.

With this definition at hand, we may state the next important theorem:

Theorem 116. The measure of an inscribed angle is half the measure of its
intercepted arc.

Proof. Let angle ]A be inscribed in an arc BAC

˘ of a circle with center P ,
intercepting the arc BC.
˜ We consider three different cases:
9.2. ARCS, ANGLES, SECANTS 179

1. One side of angle ]A contains a diameter of the circle, as shown. Let

m ]ABP = x◦ , m ]BP C = y ◦ , and m ]AP B = z ◦ . Then: m ]A + x◦ +
z ◦ = 180◦ and y ◦ + z ◦ = 180◦ . Since A and B lie on the circle, P A =
P B and triangle 4AP B is isosceles, so that m ]A = x◦ . Consequently,
2m ]A + z ◦ = 180◦ or 2m ]A = 180◦ − z ◦ = y ◦ = m BC,˜ and m ]A =
1
2 m BC.
˜

2. Suppose that B and C are on opposite sides of the diameter through

the angle vertex A. Then, m ]A = m ]BAD + m ]CAD. By case (1),
m ]BAD = 12 m BD˜ and m ]CAD = 1 m DC.
2
˜ By Theorem 115, m BD ˜+
1
m DC = m BC. Therefore, m ]A = m BC.
˜ ˜
2
˜

3. Finally, suppose that B and C are on the same side of the diameter
through vertex A. With reference to the figure, we have: m ]BAC =
y ◦ = m ]BAD − m ]CAD = z ◦ − x◦ = 21 m BD ˜ − 1 m CD
2
˜ by case (1).
1
Therefore, m ]A = 2 m BC.
˜

The proof of Theorem 116 is now complete.

180 CHAPTER 9. ON CIRCLES AND SPHERES

Two very useful corollaries follow from Theorem 116:

Corollary. An angle inscribed in a semicircle is a right angle.

Corollary. Angles inscribed in the same arc are congruent.

The proofs of these two corollaries are rather straightforward. For instance,
in the figure, the three inscribed angles ]x, ]y, and ]z are congruent because
they intercept the same arc BC.
˜

Theorem 117. Two chords in the same circle or in congruent circles are con-
gruent if and only if the corresponding minor (or major) arcs have the same
measure.

Proof. Consider two congruent circles with centers P and P 0 and radius r. Let
AB and A0 B 0 be two congruent chords. Triangles 4P AB and 4P 0 A0 B 0 are
congruent (S.S.S.) and so ]AP B ∼ = ]A0 P 0 B 0 . Consequently, m AB
˜ = mA ˘0B0

by virtue of definition 70. The converse statement is equally easy to prove and
is left as an exercise.

Theorem 118. If an angle has its vertex on a circle with one side a secant ray
and the other side a tangent ray, then the measure of the angle is one half the
measure of the intercepted arc.
9.2. ARCS, ANGLES, SECANTS 181

Proof. Consider angle ]RQS with vertex Q on a circle with center P and side
−→ −−→
QS tangent to the circle while side QR is a secant. Assume that ]RQS is acute
(the case where the angle is obtuse is treated in a similar manner), and the
angle measures x◦ , y ◦ , z ◦ are as indicated in the figure. In the isosceles triangle
4P QR, z ◦ = 180◦ − 2y ◦ so that z ◦ = m ]QP R = m QR ˜ = 180◦ − 2y ◦ . By
the first corollary to Theorem 112, m ]P QS = 90 and, hence, x◦ = 90◦ − y ◦ .
◦

Therefore, x◦ = 12 z ◦ which was to be proved.

We conclude this chapter with a discussion of the properties of the lengths

of tangent and secant segments in a circle.
We know that there exists exactly one tangent line to a given circle through
a given point on the circle and that there is clearly no tangent line to the circle
through a point in the interior of the circle (why?). When a point is in the
exterior of the circle, then one might justifiably ask about the existence and
number of lines through that point and tangent to the circle. The intuitive
answer to this question is two, as shown on the diagram. To prove this fact, one
may proceed as follows:

Let M be the midpoint of segment P Q, where P is the center of the given

circle and Q is the given point in the exterior of the circle. Draw the circle C 0
with center M and radius M P = M Q. Now, if C 0 intersects C in two points A
←→ ←→
and B, as suggested by the figure, then the lines QA and QB are tangent to the
circle C at A and B, because the angles ]QAP and ]QBP are each inscribed
in a semicircle and are, therefore, right angles. Then, the second corollary to
←→ ←→
Theorem 112 implies that QA and QB are indeed tangent to circle C at A and
B, respectively. This argument, however, does not prove that there are exactly
two tangent lines through the external point Q to circle C and, moreover, is
based on the assumption that circles C and C 0 intersect in two points. The
182 CHAPTER 9. ON CIRCLES AND SPHERES

rigorous proof for the existence of exactly two tangents to a given circle from a
given point outside the circle is the content of the following theorem:
Theorem 119 (The Two-Circle Theorem). Let C and C 0 be coplanar circles of
radius a and b, respectively, and let c be the distance between their centers. If
each of the three positive numbers a, b, c is less than the sum of the other two,
then the circles intersect in two points and these common points lie on opposite
sides of the line of circle centers.
Proof. The proof of this theorem is lengthy and, hence, relegated to Appendix
E.
Let us then apply the Two-Circle Theorem to our problem at hand. In the
preceding figure, let the radius of the given circle C be a and the radius of circle
C 0 be b. Then the distance between the centers of C and C 0 is M P = c = b.
Since Q is in the exterior of circle C, P Q > a, so that a < 2b. Then b + c = 2b
and thus, a < b + c. Furthermore, b < a + c because b = c and a > 0. Therefore,
Theorem 119 applies and C and C 0 intersect in two points A and B, and there
are at least two tangents to circle C from the exterior point Q. We still have to
prove that there are exactly two such tangents. To that end, suppose that the
two different circles C and C 0 intersect at three distinct points U , V , and W
and consider the triangle 4U V W which is inscribed in both circles C and C 0 .
Hence, its vertices are equidistant from the centers P and P 0 of these circles. By
Theorem 32, both P and P 0 must then lie on the perpendicular bisectors of, say,
sides U V and V W of triangle 4U V W . But these two perpendicular bisectors
intersect at a unique point and, consequently, P = P 0 which contradicts the
hypothesis that circles C and C 0 are different.
We may now summarize the preceding argument based on the Two-Circle
Theorem in the following theorem:
Theorem 120. From a given point in the exterior of a given circle, there are
exactly two tangent lines to the circle.
The length of the segment joining the given external point to the point
of tangency on the given circle is the tangent segment from that point to the
circle. Similarly, a secant line through an external point intersects a given circle
in two points and the lengths of the segments joining the external point to these
intersections on the circle are called the secant segments from the point.
Theorem 121. The two tangent segments to a circle from an external point
are congruent and form congruent angles with the line joining the external point
to the circle center.
9.2. ARCS, ANGLES, SECANTS 183

Proof. Triangles 4P QR and 4P QS are right triangles by the first corollary

to Theorem 112, and are congruent by the Hypotenuse-Leg Theorem (Theorem
41).

Theorem 122. If two secant lines through an external point Q intersect a given
circle at points R, S and T , U , respectively, then the products of the lengths of
corresponding secant segments are equal: QR · QS = QU · QT

Proof. Triangles 4SQU and 4T QR have angle ]Q in common, and ]RSU ∼ =

]U T R because these inscribed angles intercept the same arc RU
˜ (Theorem 116).
QS
Hence, 4SQU ∼ 4T QR (Theorem 100) and QT = QUQR or QS · QR = QU · QT ,
which was to be proved.

Notice that Theorem 122 implies that once the product QS ·QR is calculated
for one secant through an external point Q to a given circle, then any other
secant through that point to the same circle would yield the same product of
secant lengths. This constant product of secant segment lengths is called the
power of point Q with respect to the given circle, and plays an important role
in more elaborate studies of the relationships between points, lines, and circles.

Theorem 123. If QT is a tangent segment from the external point Q to a

given circle, and if a secant through Q intersects the circle in points R and S,
then QR · QS = QT 2 .

Proof. Angle ]RST is inscribed in the circle, so m ]RST = 21 m RT

˜ by Theorem
1 ˜. Now, triangles 4QRT and
116. Also, by Theorem 118, m ]RT Q = 2 m RT
4QT S share angle ]Q, and hence, are similar (Theorem 104). Therefore,
QR QT 2
QT = QS , i.e. QT = QR · QS.
184 CHAPTER 9. ON CIRCLES AND SPHERES

Theorem 124. If point Q is in the interior of a circle, and secants through

Q intersect the circle in points R, S, and T , U , respectively, then QR · QS =
QT · QU .

Proof. By Theorem 116, m ]S = m ]T = 21 m U

˜ R. Also, ]SQU ∼
= ]T QR, so
QS QU
that 4QSU ∼ 4QT R (Theorem 104). Consequently, QT = QR or QR · QS =
QT · QU .
9.2. ARCS, ANGLES, SECANTS 185

Exercises
1. In a circle with center P a chord AB is parallel to a tangent and intersects
the radius to the point of tangency at its midpoint. If AB = 12, determine
the radius.

2. In circle with center O, AB is a diameter and AC is any other chord with

←→ ←→ ←→
one endpoint at A. If CD is the tangent at C, and DO is parallel to AC,
←→
prove that DB is tangent to the circle at B.

3. Prove that the line of centers of two tangent circles contains the point of
tangency.

4. In the figure shown, P is the center of a circle and m ]AEP = m ]DEP .

Prove: AB = CD.
186 CHAPTER 9. ON CIRCLES AND SPHERES

←→
5. AB is a diameter of a circle with center O. Line CD is tangent to the
←→ ←→
circle at T . Lines AC and BD are perpendiculars from A and B to the
←→
tangent CD.
Prove: CO = DO.

6. Prove that the intersections of a sphere with planes equidistant from the
center of the sphere are congruent circles.

7. In the figure shown, plane Π intersects a sphere with center at O. F is

the foot of the perpendicular from O to plane Π. A and B are two points
of the intersection of the sphere with Π such that AF is perpendicular to
BF .
If AB = 5 and OF = AF , find the radius of the sphere and the measure
of angle ]AOB.

8. Given that plane Π is tangent to a sphere at point T , let Π0 be any plane

other than Π which contains T . Prove that plane Π0 intersects the sphere
and the plane Π in a circle and in a line which are tangent to one another.

9. Show that two great circles of a sphere intersect at the endpoints of a

diameter of the sphere.

10. Two great circles are said to be perpendicular if they lie in perpendicular
planes. Show that, given any two great circles, there is one other great
circle perpendicular to both.
9.2. ARCS, ANGLES, SECANTS 187

11. The center of an arc is the center of the circle of which the arc is a part.
Show how to determine the center of a given arc.
12. Given a sphere and three points on it, can one always determine the center
and the radius of the sphere? How many points would one need? Why?
˜ and m ]C = 45◦ .
13. In the figure shown, P is the center of arc AC,
Prove that BP is perpendicular to AP .

14. In the figure shown, m AB

˜ = m BF
˜ . Determine all triangles that are
similar to triangle 4BHF .

15. In the figure shown, the two circles are tangent at A and the inner circle
passes through the center O of the outer circle. Prove that any chord of
the outer circle with one endpoint at A is bisected by the inner circle.
188 CHAPTER 9. ON CIRCLES AND SPHERES

Sets of points lying on a circle are called concyclic points. Three non-
collinear points are always concyclic (why?). If the vertices of a quadrilat-
eral lie on a circle (i.e. the quadrilateral is inscribed in a circle), then these
four points are concyclic. The next problem states a necessary condition
for four (coplanar) points to be concyclic:

16. Prove that the opposite angles of an inscribed quadrilateral are supple-
mentary.
A sufficient condition for four (coplanar) points to be concyclic is the
subject of the following problem:

17. Prove that if a pair of opposite angles of a quadrilateral are supplementary,

then the quadrilateral can be inscribed in a circle.

18. In the diagram shown, AB is a diameter of the inner of two concentric

circles, both with center O, and AC and BD are tangent to the inner
circle. CO and DO are radii of the outer circle. Prove that CD is a
diameter of the outer circle.

19. XY is the common chord of two intersecting circles. AB and DC are two
segments cutting the circles as shown in the figure and containing points
X and Y , respectively.
9.2. ARCS, ANGLES, SECANTS 189

Prove: AD is parallel to BC.

20. In the figure shown, ACB

˘ is a semicircle and CD is perpendicular to AB.

Prove that CD is the geometric mean of AD and BD.

21. ABCD is an inscribed square. E is any point on arc DC,

˜ as shown in the
figure.
Prove that AE and BE trisect angle ]DEC.

←→ ←→
22. In the figure shown below, CP and AQ are tangents and P Q is a diameter
of the circle. If m P
˜ B = 120◦ and the radius of the circle is 3, determine
the length of AP .
190 CHAPTER 9. ON CIRCLES AND SPHERES

23. Two circles are tangent, either externally or internally, at point H. Let u
be any line through H meeting the circles again at M and N . Prove that
the tangents at M and N are parallel.

←→ ←→
24. P T and P R are a tangent and a secant of a circle, respectively. B is the
midpoint of arc P
˜ R.
←→ ←→
Prove that B is equidistant from lines P T and P R.

25. Prove the following theorem:

The measure of an angle formed by two secants of a circle intersecting in
the interior of the circle, is one-half the sum of the measures of the arcs
intercepted by the angle and its opposite angle.
In symbols, prove: m ]DEB = 21 (m DB
˜ + m AC)
˜
9.2. ARCS, ANGLES, SECANTS 191

26. Prove the following theorem:

The measure of an angle formed by two secants of a circle meeting in
the exterior of the circle is one-half the difference of the measures of the
intercepted arcs.

27. Verify that the theorem of problem 26 remains valid if the words “two
secants” are replaced by “a secant and a tangent” or by “two tangents”.

28. The diagonals of the inscribed quadrilateral ABCD intersect at P .

Prove: AP · P C = P D · P B.

←→
29. Line AD is tangent to a circle at A and secant BD intersects the circle at
B and C.
192 CHAPTER 9. ON CIRCLES AND SPHERES

Prove: BD · CD = AD2

30. In the figure shown below, quadrilateral ABCD is inscribed in a circle.

←→ ←→ ←→ ←→
Lines AD and BC intersect at P , and lines AB and DC intersect at Q.
−−→ −→
P V and QS are the bisectors of angles ]AP B and ]AQD, respectively.
Prove that P V is perpendicular to QS.

31. In the figure shown, CD is a tangent segment to a circle at D and AC is

a segment of a secant which contains the center of the circle. If CD = 12
and CB = 4, determine the radius of the circle.

32. Show that is not possible for the lengths of the segments of two intersecting
chords to be four consecutive integers.
9.2. ARCS, ANGLES, SECANTS 193

33. Prove that if two circles intersect, the common secant bisects either com-
mon tangent segment.
In symbols, prove: AM = M B and CN = N D.

34. Prove that the common internal tangents of two circles and the line de-
termined by their centers are concurrent and that the common tangent
segments of the common internal tangents are congruent.
194 CHAPTER 9. ON CIRCLES AND SPHERES

35. The radii of two circles have lengths 22 and 8, respectively, and the dis-
tance between their centers is 50. Determine the length of their common
external tangent segment.

−→
36. In the figure shown, AP is tangent to the circle at A. If AP = P X = XY ,
P Q = 1 and QZ = 8, determine AX.

37. Arcs AB,

˜ BC, ˜ are 120◦ arcs on a circle, and P is a point on arc
˜ and CA
AB, as shown.
˜
Prove that P A + P B = P C
Chapter 10

Loci, Constructions, and a

Medley of Post-Euclidean
Theorems

10.1 Loci
As indicated by its title, this chapter covers rather disparate topics in Euclidean
Geometry. Its purpose is to demonstrate the wealth of the mathematical struc-
ture we have developed so far, and to suggest that more striking theorems
pertaining to this apparently “elementary” mathematical discipline may still
remain to be discovered.
We begin with the concept of a locus (plural: loci). The question asked
under this rubric is to characterize and/or to describe the set of all points
which share a certain property or are subject to a specific condition or set of
conditions. Consider, for instance, the question of determining the locus (set) of
all points in a given plane which are equidistant from two given points A and B
incident with that plane. We answered this question first in Theorem 31 where
we proved that a point on the perpendicular bisector of segment AB in the
given plane is equidistant from A and B AND that any point in the given plane
and equidistant from A and B must lie on this perpendicular bisector. Both
arguments were required to determine uniquely and categorically the desired
locus. In order to further clarify the concept of locus, consider that any arc of a
circle of given radius r and center P contains points “at distance r from P ” but
does not contain all points sharing the property of being at a prescribed distance
r from a given point P : such points, if not restricted by further conditions, lie
on the surface of a sphere of radius r and center P . We summarize these ideas
in the following definition:

Definition 73. The locus of a condition (or set of conditions) is a set of points
such that every point in that set satisfies the condition (or set of conditions)

195
196 CHAPTER 10. LOCI, CONSTRUCTIONS, AND A MEDLEY . . .

AND every point which meets the condition (or set of conditions) belongs to
this set.

The following few examples are well-known loci:

- A sphere is the locus of all points which are at a given distance from a
given point.

- The perpendicular bisecting plane of a given segment is the locus of all

points which are equidistant from the endpoints of a given segment.

- A circle with diameter AB is the locus of the vertices of all right angles
each of whose legs contains one of the two points A and B as an endpoint.

A fundamental locus is provided by the following theorem:

Theorem 125. The bisector of an angle, except its vertex, is the locus of points
in the interior of the angle equidistant from the angle legs.

−−→ −−→
Proof. Given angle ]BAC, let AD be its bisector. If P is on AD and distinct
from A, P is in the interior of angle ]BAC. Let P M and P N be the (unique)
←→ ←→
perpendiculars from P to AB and AC, respectively. Then 4P M A ∼ = 4P N A
(S.A.A.) and, hence, P M = P N as corresponding parts. Conversely, let P be
in interior of angle ]BAC such that P M and P N are the perpendiculars from
P to AB and AC, respectively, with P M = P N . Then, P 6= A (definition
of the interior of an angle) and 4P M A ∼
= 4P N A (Hypotenuse-Leg Theorem).
Therefore, ]P AM ∼ = ]P AN as corresponding parts, and, by definition, P must
−−→
lie on the bisector AD of angle ]BAC.

Theorem 126. The perpendicular bisectors of the sides of a triangle are con-
current at a point called the circumcenter of the triangle.
10.1. LOCI 197

Proof. Let L1 , L2 , L3 be the perpendicular bisectors of the sides AB, AC,

BC of triangle 4ABC, respectively. L1 and L2 intersect in a unique point P
←→ ←→
because if L1 and L2 were parallel, then AB and AC would be parallel too. By
Theorem 32, AP = BP and AP = CP because P is on both L1 and L2 . Then,
Theorem 32 asserts that P must be incident with L3 . Therefore, P is on all
three perpendicular bisectors of the triangle. This completes the proof of the
theorem.

The preceding theorem may be paraphrased by stating that the locus of

points equidistant from three non-collinear points is the intersection of the
perpendicular bisectors of the sides of the triangle whose vertices are at the
given three points.

Corollary. There is exactly one circle through three non-collinear points.

Another issue which was examined in the preceding chapter (Theorem 119)
and in Appendix E can now be recast in the following corollary:

Corollary. Two distinct circles intersect in at most two distinct points.

Proof. If the two distinct circles had three points in common, then these points
would be either collinear or non-collinear. If these points were collinear, then
the line containing them would have to intersect either circle in three points,
which contradicts the Line-Circle Theorem (Theorem 112). If the three points
were non-collinear, then there would be two distinct circles through them in
contradiction with the preceding corollary.

A famous early nontrivial locus is the circle of Apollonius named after the
Greek mathematician Apollonius of Perga (ca. 260–190 B.C.) who wrote a
treatise on conic sections, and is given credit for naming these curves ellipse,
hyperbola and parabola.
The problem consists in determining the locus of all points in a given plane
whose distances from two given points is in a given ratio.

Let A and A0 be two given (fixed) points and let P be any point such that
0
PA 0
P A = k or P A = kP A where k is the given ratio. When k = 1, the locus is
clearly the perpendicular bisector of the segment AA0 . Suppose k 6= 1. Let the
←−→
internal and external bisectors of angle ]AP A0 meet line AA0 in A1 and A2 ,
respectively. Through A0 draw the parallel A0 E to A1 P and the parallel A0 F to
A2 P , respectively. Then, as in the proof of Theorem 107, F P = P A0 = EP , and
198 CHAPTER 10. LOCI, CONSTRUCTIONS, AND A MEDLEY . . .

AA1 AP AP AA2 AP AP
A0 A1 = EP = A0 P , A0 A2 = FP =
so that A1 and A2 divide the segment
A0 P
←−→
AA0 internally and externally in the ratio k1 and their location on line AA0 is
independent of the position of P . Since angle ]A1 P A2 is a right angle
(why?), it follows that P must lie on the circle with diameter A1 A2 . Thus, all
points P satisfying P A0 = kP A lie on this circle which is called the Apollonius
circle. ←−→
Conversely, if A1 and A2 are determined on line AA0 by their property of
dividing segment AA0 internally and externally in the ratio k1 , and P is a point
on the Apollonius circle, we have: EP AP
= AAA0A
1
1
= k1 = AAA0A
2
2
AP
= F P . Hence,
EP = F P and, P being the midpoint of segment F E, is the circumcenter of
the right-angled triangle 4EF A0 (why?). Therefore, A0 P = EP and AAP 0P =
AP 1 0
EP = k , i.e. the distances from A and A of any point of the Apollonius circle
satisfy the prescribed ratio P A0 = kP A. This completes the determination of
the desired locus.

Theorem 127. The three altitudes of a triangle are concurrent at a point called
the orthocenter of the triangle.

Proof. Triangle 4ABC is given. Through each of its vertices draw a line parallel
to the opposite side to determine the triangle 4DEF , as shown. Then, in the
parallelograms ACBD and ABCE, BC = AD = AE, and the altitude from A
of triangle 4ABC is the perpendicular bisector of side DE of triangle 4DEF .
By the same argument, the other two perpendicular bisectors of the sides of
triangle 4DEF are the altitudes of triangle 4ABC from vertices B and C. By
Theorem 126, these perpendicular bisectors are concurrent and, therefore, the
altitudes of triangle 4ABC are equally concurrent.

The vertices of a triangle and its orthocenter constitute a so-called ortho-

centric set of four points because each one of the four points is the orthocenter
of the triangle formed by the other three (why?).
We have already proved (Theorem 111) that the bisectors of the internal
angles of a triangle are concurrent at the triangle incenter which is, thus, the
locus of all points in the interior of a triangle equidistant from the three triangle
sides. It can be proved (see Exercises) that the external bisectors of two angles
of a triangle and the internal bisector of the third angle are also concurrent at
a point called an excenter of the triangle. Since the incenter of a triangle is
equidistant from the triangle sides, it is the center of the incircle, a circle in the
10.2. CONSTRUCTIONS 199

interior of the triangle which is tangent to its sides. Similarly, an excenter is the
center of an excircle in the exterior of a triangle and tangent to one side and to
the extensions of the other two.

A relatively recent discovery of yet another remarkable point in the interior

of a triangle is due to the French mathematician Joseph-Diaz Gergonne (1771–
1859):

In triangle 4ABC, let G be its incenter and D, E, F be the points of

tangency of the incircle with the triangle sides. Then, by Theorem 121, AE =
AF , CE = CD, and BD = BF . Now, consider the cevians AD, BE, CF :
AE CD BF
EC DB F A = 1, and by Ceva’s Theorem (Theorem 110), the segments AD, BE
and CF must be concurrent at the Gergonne point.

10.2 Constructions
We next proceed with a summary excursion into the realms of constructions with
ruler and compass. The Greek mathematicians investigated at length which ge-
ometric objects and figures could be constructed by means of two ideal instru-
ments, namely the unmarked ruler and the collapsible compass. We should not
fail to observe that our axiomatic treatment of neutral geometry has also been
essentially based on two ideal instruments: an ideal ruler which is marked on its
edge so that one can measure the lengths of arbitrary segments exactly, and an
ideal protractor which allows equally exact angle measurements (see Axioms 2,
3, 4, 10, 11). Greek geometers, however, were not interested in the actual mea-
surement process and considered that geometric objects should be constructed
exclusively in terms of equally purely geometric simpler elements, such as angles
and segments. We shall review only a few of such standard constructions, and
propose a few more in the Exercises, in order to stress the difference between
our approach to Geometry and that of the founders of this discipline. Before
doing so, a couple of remarks are necessary:
200 CHAPTER 10. LOCI, CONSTRUCTIONS, AND A MEDLEY . . .

- The ruler and compass at hand are ideal in the sense that they allow for
ideal lines and circles to be drawn.

- The ruler is unmarked and all one can do with it is to draw a line through
two given points. One cannot use this ruler to measure or even compare
two segments.

- The compass is collapsible and all one can do with it is to draw a circle
with given center and passing through one given point. In other words,
one cannot use the compass as a divider, and “as soon as one lifts the
spike off the paper, the compass collapses”

- A line is given when at least two of its points are given.

- Most constructions rely on the intersection properties of lines with one

another, lines with circles, and circles with circles. To justify such con-
structions, one is free to use the theorems proven in the corpus of Euclidean
Geometry.

Construction 1. To construct the perpendicular bisector of a given segment.

The endpoints P and Q of segment P Q are given. Draw circle C1 with

center P , containing point Q, and circle C2 with center Q, containing point P .
If P Q = a, then trivially, each of the three numbers a, a, a is less than the
sum of the other two, and the Two-Circle Theorem (Theorem 119, Appendix
V) guarantees that C1 and C2 intersect in exactly two points R and S lying on
←→ ←→ ←→
opposite sides of P Q. Let T be the intersection of P Q with RS. Since RS is a
chord of both circles C1 and C2 , T is in the interior of these circles and T P < a,
T Q < a. Consequently, T is between P and Q. Next, 4RSP ∼ = 4RSQ (S.S.S.),
so that ]1 ∼= ]2 and hence, 4T RP ∼ = 4T RQ (S.A.S.). Therefore, ]5 ∼ = ]6 so
←→ ←→
that RS is perpendicular to P Q and T P = T Q. Thus, RS is the perpendicular
bisector of P Q.
Note that Construction 1 can be used to determine the midpoint of a seg-
ment.

Construction 2. To construct the perpendicular to a given line through a

point on the line.
10.2. CONSTRUCTIONS 201

Line L and point X on it are given. Let P be any point on L, and draw
the circle with center at X containing P . This circle intersects L in exactly
one other point Q (why?), and X is the midpoint of segment P Q. Next, use
Construction 1 to construct the perpendicular bisector of segment P Q which
will provide the desired perpendicular to L through X.
Construction 3. Given three points P , Q, R, to construct a rectangle P QST
such that P T = P R.

←→
Construct the perpendicular L1 to P Q at P (Construction 2). Draw the
circle with center P , containing R and intersecting L1 at T and T 0 . Next,
←→
construct L2 perpendicular to P Q at Q and L3 perpendicular to L1 at, say, T
(Construction 2). Lines L2 and L3 intersect at S (why?). Now, each pair of
opposite sides of the quadrilateral P QST are parallel, and three of its angles
are right angles. Therefore, P QST is the desired rectangle with P T = P R.
Construction 4. To construct a point on a given ray such that the segment
with endpoints at this point and at the ray vertex is congruent to a given
segment.
202 CHAPTER 10. LOCI, CONSTRUCTIONS, AND A MEDLEY . . .

−−→
Let segment P Q and ray AB be given. Construct rectangle P AT U with
P U = P Q (Construction 3). Then AT = P Q. Draw the circle with center A,
−−→
containing point T . This circle will intersect ray AB in a (unique) point C with
AC = P Q, as required.
Construction 4 allows us to “carry over” any given segment onto a given
ray and is thus equivalent with our Point Plotting Theorem. This Construc-
tion, moreover, eliminates all difficulties stemming from the collapsibility of the
Greek compass, and shows that a ruler and a collapsible compass provide in com-
bination an effective pair of dividers. We may, therefore, use a non-collapsible
compass freely without sacrificing the rigor of the original Greek intentions, as
we will from now on.

Construction 5. To copy a given angle on a side of a given ray.

Angle ]A and a ray emanating at vertex D are given. With A as center,

draw an arc of circle intersecting the legs of angle ]A at B and C. Construct
point F on the given ray such that DF = AB (Construction 4), and draw
the circle C with center D and containing F . With F as center construct the
circle with radius BC (we may now use a non-collapsible compass!) intersecting
−−→
circle C at points E and E 0 (Two-Circle Theorem). Construct rays DE and
−−→0
DE . Then 4DEF ∼ = 4ABC (S.S.S.). Thus: ]EDF ∼ = ]BAC ∼ = ]E 0 DF , as
desired.

Construction 6. To bisect a given angle.

Draw an arbitrary (arc of) circle with center at the vertex A of the given
angle, intersecting its legs at B and C. Construct circles with centers at B and
C and radius r > 12 BC. By the Two-Circle Theorem, these circles intersect in
10.2. CONSTRUCTIONS 203

←→
exactly two points, one on each side of line BC. Let P be the intersection on
−→
the side opposite to A and construct ray AP . Then 4BAP ∼ = 4CAP (S.S.S.)
and ]BAP ∼ = ]CAP as corresponding parts.

Construction 7. To construct a parallel to a given line through point not lying

on the line.

Draw a line through the given point P to an arbitrary point Q on the given
line L. Construct angle ]QP S congruent to angle ]P QR (Construction 5) with
←→ ←→
points R and S on opposite sides of P Q. Line P S is the desired parallel to L.
The justification of this construction is straightforward and left as an exercise.
We now show how to carry out all of the elementary algebraic operations
in the Greek spirit, that is, with ruler and compass. In other words, if two
segments with lengths a √ and b are given, we show how segments whose lengths
are a + b, a1 , ab, ab , and a can be constructed with our ideal instruments. In
order to execute these constructions, either a segment with unit length should
be provided or an arbitrary segment should be chosen as unit segment.
The construction of a segment with length a + b is trivial: on any line, use
the (non-collapsible) compass to lay off a segment P Q with length a and then
a segment QR with length b such that Q is between P and R. A segment with
length a − b is obtained when segment QR with length b is laid off as before but
with endpoint R between P and Q.
1
Construction 8. To construct a segment with length a

Lay off a segment AB with length a on any line L1 (with a non-collapsible

compass!). Lay off a segment AC with unit length on any line L2 6= L1 through
←→
A. Construct point D on L1 such that AD = 1, and draw line DE parallel to
←→
BC (Construction 7) intersecting L2 at E. Then, 4ACB ∼ 4AED (why?)
and hence, AD AE 1
AB = AC i.e. a = AE so that AE is the desired segment.
204 CHAPTER 10. LOCI, CONSTRUCTIONS, AND A MEDLEY . . .

Construction 9. To construct a segment with length ab.

Lay off a segment AB of length a on any line L1 , and segments AC and AD

←→
of lengths 1 and b, respectively, on any line L2 6= L1 through A. Draw line DE
←→
parallel to line BC through D meeting line L1 at E. Then 4ABC ∼ 4AED
AE
and we have AB = AD
AC , i.e.
AE
a = b or AE = ab. Thus AE is the desired
segment with length ab.
Clearly, a segment with length ab can be constructed by constructing first a
segment of length 1b and then a segment with length a( 1b ), as already shown.
√
Construction 10. To construct a segment of length a

Construct segments P Q and QR on any line such that Q is between P and

R, P Q = 1, and QR = a. Bisect segment P R at M (Construction 1) and with
M as center, draw a circle with radius M P = M R = (a+1) 2 . Construct the
perpendicular to P R at Q (Construction 2) meeting the circle at two points,
one of them being S, as shown. We have then 4P √ QS ∼ 4SQR (why?), hence,
QS QR a 2
PQ = QS or QS = QS , i.e. QS = a or QS = a, as desired.
It requires little thought p to realize that now an infinite variety of segments
√ √
with lengths a + b, a+b √ ,
c
a + b, etc. can be constructed by combining
the four fundamental operations of addition (and subtraction), multiplication,
division, and extraction of square rootq as needed. In particular, segments with
p√ √ »p√ √ »p√ √ √
n
lengths a = 4 a, a = 8 a, a = 16 a, . . . , 2 a, where n =
2, 3, 4, . . ., can be constructed with ruler and compass. Geometric entities which
can be constructed with unmarked ruler and compass are called constructible,
and the foregoing shows that segments whose lengths can be constructed through
successive applications of the operations described above are constructible. By
abuse of language, numbers which correspond to the lengths of such segments
are also called constructible. A famous example of a constructible segment is
10.2. CONSTRUCTIONS 205

the side of a regular 17-gon inscribed in a circle of unit radius whose length
equals

√ √ √ √ √
q p » p p
34 − 17 − 34 − 2 17 − 2 17 + 3 17 + 170 − 26 17 − 4 34 + 2 17
4
There are, however, many—in fact, an infinitude—of segments which are
not constructible. One famous example is the problem of the duplication of
a cube with given edge of length a, that is, to construct a segment b such
that the volume of a cube with edge length b equals a3 . It turns out that it
is impossible to construct with unmarked ruler and compass, a segment √ of
length b such that b3 = 2a3 . In other words, a segment of length ab = 3 2—or,
equivalently, ( ab )3 = 2—is not constructible. Two other equally “impossible”
and equally notorious construction problems were the squaring of the circle, i.e.
the construction of a square whose area is exactly equal to the area of a given
circle, and the trisection of an angle. Of course, some angles—such as a right
angle—can be trisected with unmarked ruler and compass, but it can be proved
that it is not possible to divide with our instruments an arbitrary angle—in
particular, an angle with measure 60◦ —into three congruent angles by rays in
its interior. For some two thousand years, these three problems preoccupied
the best minds until it was mathematically proved in the late 19th and early
20th centuries that ruler-and-compass solutions of these problems did not exist.
The proof of these impossibilities requires some knowledge of advanced algebra
and would take us too far afield1 . Instead we include a method for trisecting
an arbitrary angle with a marked ruler and compass. For our purpose, it is
sufficient that the ruler carries only two marks on its edge.

Let an angle with vertex B be given. Draw the circle with center at B and
radius equal to the distance between the two marks on the ruler. This circle
intersects the legs of the given angle at A and C, as shown. Next, place the
ruler so that its edge passes through C, and slide and rotate it until one marked
point lies on the circle at Q and the other marked point comes to lie at P on
−−→
the ray opposite to ray BA. In terms of the angle measures indicated on the
figure, we have: v ◦ = u◦ so that w◦ = 2u◦ = x◦ . Therefore, z ◦ = x◦ + u◦ = 3u◦ .
Thus, angle ]BP C = 13 m ]ABC. Two applications of Construction 5 in the
1 The interested reader may consult, for example, Elementary Geometry from an Advanced

Standpoint, by E. Moise, Addison-Wesley Publishing Co., 1963pp. 227–241.

206 CHAPTER 10. LOCI, CONSTRUCTIONS, AND A MEDLEY . . .

interior of angle ]ABC yields two rays emanating from B and trisecting the
given angle.
Needless to say, ruler-and-compass construction problems vary greatly in
difficulty and some may require considerable experience and ingenuity for their
solution. A few interesting (and amenable) construction problems are suggested
in the Exercises to this chapter.
Another intractable—albeit less famous—construction problem which at-
tracted the attention and the efforts of mathematicians from antiquity to the
19th century was the problem of dividing a circle into a given number, say n,
of congruent arcs, end to end.

Joining the successive division points on the circumference by chords pro-

duces a polygon with n sides. If segments are drawn from the center of the
circle to the vertices of this n-gon, a set of n isosceles triangles are obtained.
◦
In each such triangle, the vertex angle has measure 360 n because the measure
◦
of the intercepted arc by the central angle is in each case equal to 360 n . There-
fore, these isosceles triangles are all congruent (S.A.S.). Hence, the measure of
the angles of the inscribed n-gon is twice the measure of the base angle of the
isosceles triangles. Thus, the n-gon has all its sides and all its angles congru-
ent and is, therefore, regular and inscribed in the circle. It is not difficult to
show that every regular n-gon can be inscribed in a circle: since n > 2, the
real number 360n is less than 180 and by the Angle Construction Axiom (Axiom
◦
11), there exists an angle having measure 360 n . This angle can then be copied
n times side by side with the vertex of each copy at the center of the circle to
determine n congruent end-to-end arcs whose endpoints are the vertices of the
desired inscribed n-gon. The actual problem referred to earlier is, however, that
of the constructibility of a regular (inscribed) n-gon with unmarked ruler and
compass alone. In the following, we shall briefly examine this question.
First, notice that if a regular n-gon can be constructed, then a regular 2n-gon
can be constructed: Construction 1 allows for the construction of the midpoints
of each side of the initial n-gon and the rays with vertex at the center of the
circumcircle of the n-gon through these midpoints divide each arc subtended by
the sides of the n-gon into two congruent arcs (why?). Then the chords with one
10.2. CONSTRUCTIONS 207

endpoint at a vertex of the n-gon and the other endpoint at these subdivision
points generate a regular, inscribed 2n-gon.

Next, it is easy to construct an inscribed square: construct a diameter per-

pendicular to an arbitrary diameter of the given circle (Construction 2), and
the endpoints of these two diameters yield the vertices of an inscribed square
(why?). Subdividing the arcs subtended by the sides of this square into two
congruent arcs, as described before, allows for the construction of a regular
inscribed octagon, which in turn would yield, by the same method, a regular
inscribed 16-gon, and so on. Thus, if m ≥ 2 is any natural number, a regular
inscribed 2m-gon can be constructed. On the other hand, a regular 3-gon—an
equilateral triangle—can be constructed (see Exercises to this chapter). Hence, a
regular hexagon (6-gon), a regular dodecagon (12-gon), and in general, a regular
3 · 2m -gon can be equally inscribed in a given circle through ruler-and-compass
constructions.

We shall now show how to construct a regular inscribed pentagon so that, by

the same argument, regular polygons with 5·2m sides also become constructible.
The central angle at the vertex of each of the ten congruent isosceles triangles
formed by the center of the circumcircle and the vertices of a regular inscribed
decagon is 36◦ . In the figure, we have chosen the radius of the circumcircle to be
one unit for convenience. Then, m ]A = m ]ABO = 72◦ . The bisector BD of
angle ]ABO can be constructed (Construction 6). Then, AB = BD = OD = x
and AD = 1−x, as shown. Since 4OAB ∼ 4ABD (see also Theorem 107), the
bisector BD divides the opposite side into segments that are proportional to the
adjacent sides of triangle 4OAB, i.e. x1 = 1−x x
or x2 = 1 − x (or x2 + x − 1 = 0).
The problem of the construction of a regular inscribed decagon is thus reduced
to that of the division of a segment of unit length into two segments with lengths
x and 1 − x such that x2 = 1 − x.
208 CHAPTER 10. LOCI, CONSTRUCTIONS, AND A MEDLEY . . .

To carry out such a partition of the unit segment, construct two perpendic-
ular diameters AA0 and EE 0 in the unit circle. With the midpoint C of radius
OA0 as center, draw a circle with radius CE intersecting radius OA √
at point
D. Then, CE 2 = OE 2 + OC 2 , i.e. CE 2 = 1 + 41 = 54 or CE = 25 = CD.
√ √
Consequently, OD = CD − OC = 5−1 and ED2 = OE 2 + OD2 = 5−2 5 .
2 √
It is easily verified that a segment with length 5 can be constructed, since
√
5 is the length of the hypotenuse of √a right triangle with side lengths 1 and
2, and that the constructible number 5−1 2 satisfies the equation x2 = 1 − x.
Hence, OD is the side of a regular inscribed decagon and a regular decagon is
constructible. By joining every other vertex of this regular decagon, a regular
inscribed pentagon is constructed. It is not difficult to show that the intersec-
tion B of a circle with center at D and radius DO with the unit circle yields
an isosceles triangle 4OAB with BD the bisector of angle ]ABO which is the
36◦ − 72◦ − 72◦ triangle we started from.

Finally, if two regular polygons with n and m sides, respectively, can be

constructed, then this is equivalent with the ruler-and-compass possibility of
◦
360◦
dividing the (unit) circle into congruent arcs with measures 360
n and m◦ , re-
spectively. Therefore, one can construct arcs with measures k times 360n and
◦
p times 360
m ◦ where k and p are natural numbers. Consequently, an arc with
360◦ 360◦
measure k 360n + p m = (km + pn) mn is constructible. Now, it is well-known
that any two natural numbers have a greatest common divisor: if m and n are
10.2. CONSTRUCTIONS 209

natural numbers, then there are integers (positive, negative, or zero) k and p
such that this greatest common divisor can be written as the sum km + pn. If
the natural numbers m and n are relatively prime, i.e. have no common divisor
except one, then there exist integers k and p such that km + pn = 1. Therefore,
if regular polygons of n and m sides can be constructed, and if m and n are
◦
relatively prime, then an arc with measure 360 mn can be constructed, leading
to the constructibility of a regular inscribed polygon with nm sides. Thus a
3 · 5-gon = 15-gon can be constructed.

Summing up, a regular inscribed n-gon can be constructed if n is a natural

number of the form 2m , 2m · 3, 2m · 5, or 2m · · · 3 · 5. For instance, to construct
the central angle of a regular inscribed 15-gon, construct first the central angles
◦
of the equilateral triangle and of the regular pentagon which have measures 360 3
◦ ◦ ◦
◦
and 360 360 2 1
5 , respectively; then the angles 2( 5 ) and [ 5 − 3 ]360 i.e.
360
15 can
be constructed. On the other hand, the central angle of a regular nonagon has
◦
◦ 360◦
measure 3609 = 40 , but although an arc with measure 3 can be constructed,
we know that the ruler-and-compass trisection of this angle in order to obtain
an arc with measure 3600/9 is not possible. Thus the regular nonagon is not
constructible.

But what about other regular n-gons, in particular the missing heptagon?
The short answer is contained in a remarkable result, (essentially) due to Gauss,
which states as a sufficient condition that a regular n-gon is constructible if n is a
prime number of the form 2µ + 1 where µ = 2m for some positive integer m ≥ 0.
For m = 0 and m = 1, one obtains the equilateral triangle and the regular
pentagon, respectively. The case m = 2 leads to the regular 17-gon which was
proved by Gauss in 1796 to be constructible. The regular heptagon, however, is
not constructible, although endowed with a prime number of sides, because the
natural number seven is not of the requisite form 2µ + 1 with µ = 2m . It was
conjectured by the first-rank French mathematician Pierre de Fermat (1601-
1665) that all numbers of this particular form are prime, and therefore such
numbers are called Fermat numbers which have turned up on many unrelated
occasions in the history of mathematics. For m = 3 and m = 4, the Fermat
numbers are 257 and 65,537 which are indeed prime and, consequently, the 257-
gon and the 65,537-gon are constructible. For m = 5, however, Euler showed
that the number 232 + 1 has 641 as a divisor, and is thus not prime. Fermat
numbers with m = 6, 7, 8, and 9 are known to be not prime.

At any rate, on the basis of the foregoing, it can be asserted that a sufficient
0
condition for the constructibility of a regular n-gon is that n = 2s · (p1 )r · (p2 )r ·
00
(p3 )r · · · where s ≥ 1 is a natural number, the exponents r, r0 , r00 , . . . take on
the values 0 or 1, and p1 , p2 , . . . are distinct primes of the form 2µ + 1 with
µ = 2m , m ≥ 0. It is not established that this condition for the constructibility
of a regular n-gon is also necessary.
210 CHAPTER 10. LOCI, CONSTRUCTIONS, AND A MEDLEY . . .

10.3 Some Newer Theorems

The remainder of this chapter is devoted to the presentation of a number of
elementary, yet remarkable, “post-Euclidean”, i.e. relatively modern, contribu-
tions to the corpus of classical theorems of Euclidean Geometry.

Theorem 128. The three perpendiculars from a point on the circumcircle to

the sides of a given inscribed triangle intersect the sides at three collinear points.

The line on which these three points lie is called a Simson Line of the trian-
gle, after the British mathematician Robert Simson (1687–1768) who discovered
it.

Proof. Let P be a point on the circumcircle of triangle 4ABC, and let D, E, F

be the feet of the perpendiculars through P to sides AB, AC, BC, respectively.
Then, besides the obvious quadrilateral P ABC, the quadrilaterals P DBF and
P DAE are concyclic because each of them has a pair of opposite suplementary
angles (see Exercises 16 and 17 to Chapter Nine). Therefore, in the circle
through P , D, A, E: m ]P DE = 12 m P ˜ E = m ]P AE and hence, ]P DE ∼ =
]P AE. But angles ]P AE and ]P AC are the same, so that ]P DE ∼ = ]P AC.
Next, in the circle through P , A, B, C, a similar argument yields ]P AC ∼ =
]P BC and since angles ]P BC and ]P BF are the same, ]P AC ∼ = ]P BF . A
similar argument with the circle through points P , D, B, F leads to ]P BF ∼ =
]P DF . By transitivity: ]P DE ∼ = ]P DF , i.e. points D, E, F are collinear.

Theorem 129. The centroid, the circumcenter and the orthocenter of a triangle
are collinear, and the centroid trisects the segment joining the circumcenter to
the orthocenter.

The line containing these three points is called the Euler Line of the triangle
after the Swiss mathematician and physicist Leonhard Euler (1717–1783) who
proved the above theorem.
10.3. SOME NEWER THEOREMS 211

Proof. Let G, H, O be the centroid, the orthocenter, and the circumcenter of

−−→
triangle 4ABC, respectively. Let A0 be the midpoint of side BC. Let ray CO
meet the circumcircle at I. Then the right triangles 4COA0 and 4CIB (CI
0
0
is a diameter of the circumcircle!) are similar and OA 1
IB = 2 or IB = 2OA .
Next, the quadrilateral AHBI is a parallelogram, since AH and BI are parallel
(both are perpendicular to BC) and BH and AI are also parallel because BH
is perpendicular to side AC and m ]CAI = 90◦ . Hence, AH = BI or OA AH
0 = 2.
AG AH
On the other hand, GA0 = 2 since G is the centroid of the triangle. Thus, OA0 =
AG ∼ 0
GA0 . Since ]HAG = ]GA O (alternate interior angles), it follows that triangles
4HGA and 4OGA are similar (S.A.S. similarity). Consequently, ]AGH ∼
0
=
]A0 GO and therefore points H, G, O are collinear so that angles ]AGH and
]A0 GO are vertical angles. Finally, the ratio of similitude of triangles 4HGA
and 4GOA0 being 2, it follows that HG = 2OG.

Another significant theorem which dates back to 1822 and was discovered
independently by the German Karl Feuerbach (1800–1834) and the two French-
men Charles Brianchon (1785–1864) and Jean-Victor Poncelet (1788–1867) is
the so-called Nine-Point Circle Theorem:

Theorem 130. The midpoints of the sides of a triangle, the feet of its altitudes,
and the midpoints of the segments joining its orthocenter to its vertices lie on
a circle.
212 CHAPTER 10. LOCI, CONSTRUCTIONS, AND A MEDLEY . . .

Proof. Let A0 , B 0 , C 0 be the midpoints of sides BC, AC, AB of triangle 4ABC,

respectively. Let D, E, F be the feet of the altitudes meeting at the orthocenter
H. To begin with, there is a (unique) circle through A0 , B 0 , C 0 . Next, we
show that point D is on this circle. The quadrilateral A0 DB 0 C 0 is an isosceles
←−→ ←→
trapezoid because B 0 C 0 is parallel to BC (why?) and A0 C 0 = 12 AC while
in the right triangle 4ADC, DB 0 is the median to the hypotenuse, so that
DB 0 = 21 AC = A0 C 0 . Hence, the four points A0 , D, B 0 , C 0 are concyclic (see
Exercises 16 and 17 to Chapter Nine) and D lies on the same circle as A0 , B 0 ,
C 0 . A similar argument establishes that points E and F lie on that same circle.
Now, let J, K, L be the midpoints of segments AH, BH, CH, respectively. We
prove that J lies on the circle through A0 , B 0 , C 0 . Consider the circle with JA0
←→ ←→ ←→
as diameter. In triangle 4AHC, B 0 J is parallel to CH (why?). Hence, B 0 J is
←→ ←−→ ←→
perpendicular to AB. In triangle 4ABC, A0 B 0 is parallel to AB (why?), and
←→ ←−→
consequently, B 0 J is perpendicular to A0 B 0 , i.e. ]JB 0 A0 is a right angle and
thus, B 0 lies on the circle with JA0 as diameter. Similarly, one proves that C 0
lies on that same circle so that J lies on the unique circle that contains A0 , B 0 ,
C 0 . In the same fashion, one shows that points K and L lie on the circle through
A0 , B 0 , C 0 . This completes the proof of the Nine-Point Circle Theorem.

The Nine-Point Circle has several interesting properties, some of which are
enumerated here without their proofs—which are not particularly involved—for
the sake of brevity:

- The center of the Nine-Point Circle bisects the segment joining the ortho-
center and the circumcenter of the triangle. Thus, surprisingly, the center
of the Nine-Point Circle lies on the Euler Line of the triangle.

- The radius of the Nine-Point Circle is half the radius of the circumcircle
of the triangle.

- The Nine-Point Circle is tangent internally to the incircle and externally

to the excircles of the triangle.

In the same vein, we may mention the little-known Spieker Circle which
is the incircle of the triangle formed by the midpoints of the sides of a given
triangle. It can be shown that the Gergonne point, the incenter, and the center
of the Spieker Circle of a triangle are collinear.
Our last entry in the list of post-Euclidean theorems is a result due to the
Anglo-American mathematician Frank Morley (1860–1937) who published it in
1899:

Theorem 131. The three points of intersection of the adjacent trisectors of

the angles of a triangle form an equilateral triangle.
10.3. SOME NEWER THEOREMS 213

In the diagram shown, triangle 4ABC is given. The adjacent trisectors AR0
and CP 0 of angles ]A and ]C of this triangle meet at Q, and similarly, those
for angles ]C and ]B, and those for angles ]B and ]A meet at P and R,
respectively. Morley’s Theorem asserts that triangle 4P QR is equilateral.

Proof. The proof of this theorem is broken down into several steps and proposed
in the Exercises to this chapter.
214 CHAPTER 10. LOCI, CONSTRUCTIONS, AND A MEDLEY . . .

Exercises
1. Describe the set of points at a given distance from

(a) a given line.

(b) a given plane.
(c) each of two intersecting planes.

2. Describe the set of points equidistant from

(a) two parallel lines.

(b) two parallel planes.
(c) two intersecting planes.
(d) a plane and a line perpendicular to it.

3. Indicate whether each statement is true or false.

(a) Given a line L and a plane Π, there is always a plane

i. containing L and perpendicular to Π.
ii. containing L and parallel to Π.
(b) Given two non-intersecting lines in space, there is always a plane
containing one and
i. parallel to the other.
ii. perpendicular to the other.
4. Describe the set which contains the vertices of all isosceles triangles having
the given segment AB for base.
5. Determine the locus of all points which are equidistant from three non-
collinear points.
6. What is the locus of points which are equidistant from two given points
and at the same time equidistant from two given parallel planes?
7. Prove the following theorem:
−−→ −→
Given angle ]DAE and B, C points on rays AD and AE, between A
and D and A and E, respectively, then the bisectors of the angles ]BAC,
]DBC, and ]BCE are concurrent.
10.3. SOME NEWER THEOREMS 215

8. Given the three lines determined by the sides of a triangle, prove that
there are exactly four points in the plane of the triangle each of which is
equidistant from all three lines.
9. If the vertices of a quadrilateral are concyclic (i.e. lie on a circle), the
quadrilateral is called cyclic. Prove that the perpendicular bisectors of the
four sides and of the two diagonals of a cyclic quadrilateral are concurrent.
10. On a circular lake, there are three docks, A, B, C. Draw a diagram indi-
cating the locus of the points on the lake which are closer to A than to B
or C.

11. A and B are fixed points in a plane Π. C and D are any points in Π such
that ABCD is a parallelogram with BC = 21 AB. Determine the locus of
the midpoint M of side CD.
12. A and B are fixed points in a plane Π and l is any line in Π through A. P
is a point in Π such that l is the perpendicular bisector of segment BP .
What is the locus of P as line l takes all possible positions?
In the following construction problems, justify your construction, deter-
mine whether the desired construction is always possible, and discuss the
number of distinct solutions, if any.
13. Construct a triangle 4ABC, given the lengths of two sides AB and AC,
and the length of the median from vertex A to side BC. (This one is
trickier than it seems.)
14. Construct a triangle, given the length of one side, and the lengths of the
altitude and of the median to that side.
15. Construct a rectangle, given the lengths of one side and of one diagonal.
16. Construct a triangle, given the measure of one angle, the length of an
adjacent side, and the length of the altitude to that side.
17. Construct a triangle, given the length of one side and the lengths of the
medians to the other two sides.
216 CHAPTER 10. LOCI, CONSTRUCTIONS, AND A MEDLEY . . .

18. Given a circle and a point in its exterior, construct the tangents from that
point to the circle.

19. Construct a triangle, given the length of one side, the circumradius, and
the length of the altitude to a second side.

20. Construct a circle with given radius and tangent to two given intersecting
lines.

21. Construct a circle with given radius, tangent to a given line and tangent
to a given circle.

22. Construct a triangle, given the measure of one angle, the length of the
internal bisector of that angle, and the radius of the incircle.

23. A segment of length a is given as the median to one of the two congru-
ent sides of an isosceles triangle in which the medians to these sides are
perpendicular to each other. Construct the triangle.

24. Given a circle C tangent to a line m at K, and point M on line m.

Construct a circle tangent to C and also tangent to m at a M .
Hint: Analyze the diagram shown.

25. Construct a common external tangent to two given coplanar circles.

26. Given a triangle 4ABC in which each angle has measure less than 120◦ ,
construct a point P in the interior of the triangle such that m ]AP B =
m ]BP C = m ]AP C.

27. Given two parallel lines l and m at distance d from each other. Determine
the set of all points P in the plane of these lines such that the distance
from P to l is k times the distance of P from m, where k is a given positive
number.

28. Given square ABCD with M and N the midpoints of sides BC and CD.
If AM and AN meet the diagonal BD at P and Q, prove that P and Q
trisect BD, but that m ]BAM 6= 30◦ .
10.3. SOME NEWER THEOREMS 217

29. Devise a ruler-and-compass construction for trisecting a right angle.

30. Given two parallel lines l and m and a transversal n, are there any points
which are equidistant from l, m, and n? Justify your answer.

31. Describe the location of the Simson Line if the point on the circumcircle
is a vertex of the triangle.

32. Examine the Simson Line for a right triangle. Repeat for an equilateral
triangle.

33. To prove the Morley Theorem, we proceed in several stages as follows:

(a) Let 4P QR be an equilateral triangle. Let α, β, γ be three angle

measures satisfying α + β + γ = 120◦ , and α < 60◦ , β < 60◦ , γ < 60◦ .
Construct three isosceles triangles 4P QR0 , 4P RQ0 , 4RQP 0 on the
sides P Q, P R, RQ, respectively, of triangle 4P QR, with base angles
having measures γ, β, α, respectively as shown on the figure above.
Determine the measures of angles ]ARP 0 , ]AQP 0 , ]BP Q0 , ]BRQ0 ,
]CP R0 and ]CQR0 . Then find the measures of angles ]QAR,
]QCP , and ]P BR.
(b) Next prove the following Lemma:
Lemma. In any triangle the measure of the angle subtended by the
bisectors of two angles is equal to 90◦ plus one-half the measure of
the third angle.
218 CHAPTER 10. LOCI, CONSTRUCTIONS, AND A MEDLEY . . .

(c) Find the measures of angles ]AQC, ]CP B, ]BRA.

(d) Find the measures of angles ]AQ0 C, ]CP 0 B, ]BR0 A.
Now, triangle 4P Q0 R is isosceles and triangle 4P QR is equilateral.
Hence, the bisector of angle ]P Q0 R is also the median as well as
the altitude of triangle 4P Q0 R from vertex Q0 . It is therefore the
same as the median or altitude or bisector from vertex Q of triangle
4P QR. Hence, Q is located on the bisector of angle ]AQ0 C, and
by the ”converse” of the above Lemma, Q is the incenter of triangle
4AQ0 C (why does the ”converse” work?). Similarly, P and R are
the incenters of triangles 4BP 0 C and 4AR0 B.
(e) Find the measure of angle ]BAC and compare it with the measures
of angles ]QAC and ]RAB.
It follows then that QA and RA trisect angle ]BAC and that m ]BAC =
3(60◦ − α) and hence, α = 60◦ − 13 m ]BAC. Similar results hold for
β and γ. This completes the “backward” proof of Morley’s Theorem:
we began with an equilateral triangle 4P QR, and built up a gen-
eral triangle which was afterwards identified with the given triangle
4ABC by choosing the angle measures α, β, γ as found above in
terms of the angles of triangle 4ABC.
(f) What values of α, β, γ will make the triangle 4ABC equilateral?
Repeat for a right-angled isosceles triangle.
Chapter 11

On Arc Lengths and Sector

Areas

11.1 Arc Length

So far, we have introduced the measure of an arc of circle and examined the
relationship between this measure and that of the corresponding central or in-
scribed angles, but we have not yet a mathematical definition for the length
of an arc of circle in analogy with the length associated with a line segment.
In our axiomatic approach, we postulated that a unique positive real number
is associated with every line segment and called that number the length of the
segment. It would be quite expeditious to proceed similarly with arcs of circle,
and to postulate that a certain positive real number which would be called the
arc length can be attributed to each such arc. This correspondence between
arcs of circle (i.e. purely geometric entities) and their lengths (i.e. purely alge-
braic quantities) should then possess certain intuitively desirable properties: it
should be one-to-one, that is each given arc should have a unique length, and it
should be additive in the sense that the length of the union of two arcs of the
same circle having only one endpoint in common should be equal to the sum of
the lengths of those arcs. Finally, the length of an arc of circle should evidently
depend on the arc radius and on the measure of its central angle. The crucial
problem with this approach is that we have no mathematical definition for the
circumference of a circle. We must, therefore, adopt a different and more subtle
approach which exploits the well-known geometry of line segments and allows
for a precise definition of the concept of arc length with the stated properties.

219
220 CHAPTER 11. ON ARC LENGTHS AND SECTOR AREAS

Given arc AB
˜ of a circle C, let A0 = A, A1 , . . . , An = B be a sequence of
points on this arc in the order from A to B. For each pair of consecutive points
Ai−1 and Ai , draw the line segment Ai−1 Ai , as shown. The union of these line
segments constitutes an inscribed broken line Bn with n vertices, and the sum
sn = A0 A1 + A1 A2 + . . . + An−1 An of the lengths of these segments is denoted
sn and called the length of the inscribed broken line Bn . Note that the arcs
A
˚ i−1 Ai and, consequently, the chords Ai−1 Ai need not be congruent. It is now
intuitively evident that as the number n of points on arc AB ˜ is increased, the
lengths of each chord Ai−1 Ai decreases while the length sn of Bn increases and
the inscribed broken line becomes increasingly “close” to the actual arc AB. ˜
Thus, sn provides an approximation of the length of the arc AB and this ap-
˜
proximation is “better and better” as the number n of vertices of Bn increases.
These considerations suggest only an idea and are not precise enough to consti-
tute a definition, because they involve non-mathematical notions, notably the
notions of “close” and “better and better.” To cast this idea in rigorous terms,
we need some elements of Calculus for which, however, we have no room in this
elementary treatment of Geometry. We shall therefore accept the mathematical
facts and results presented below without justification and proof, and use them
freely. Needless to say, our presentation will nonetheless remain at best sketchy,
but, we hope, sufficient for our purpose.
We consider sets of real numbers which contain infinitely many elements.
The precise definition of an infinite set will not be given, but one may consider
the set of all natural numbers or the set of all real numbers between zero and
one as typical examples of infinite sets.
A set S of positive real numbers is said to be bounded above if there is a
positive real number b, called an upper bound for S, such that every element of
S is less than or possibly equal to b. Clearly, if b is an upper bound for S then
any real number larger than b is also an upper bound for S. Now, one of the
fundamental concepts of Calculus is that of the least upper bound or supremum
(abbreviated: sup) of a set that is bounded above: it is conceivable that for
a given set S of positive real numbers that is bounded above, there exists,
among the myriad of such upper bounds, an upper bound which is smaller than
all. The Completeness Axiom for real numbers—which has already manifested
itself in a weaker form in this course, namely in the content of the Euclidean
Completeness Axiom—asserts that every set of real numbers that is bounded
above has indeed a unique least upper bound. It should be mentioned in
11.1. ARC LENGTH 221

passing that these concepts and facts are also valid for sets of real numbers
that are bounded below : if a set of real numbers is bounded below, i.e. if every
member of the set is larger than some real number, then the set has a unique
greatest lower bound or infimum (abbreviated: inf). Note that depending on
the bounded set S, its least upper/greatest lower bound may or may not belong
to S. For instance, if N is the set of all natural numbers, then N is bounded
below by 1 but is not bounded above. Therefore, sup(N) does not exist but
inf(N) = 1. If S is the set of all real numbers x satisfying 0 < x < 1, then S is
bounded both below and above and inf(S) = 0, sup(S) = 1 and neither inf(S)
nor sup(S) belongs to S, whereas if S 0 is the set of all real numbers satisfying,
say, 0 < x ≤ 1, then sup(S 0 ) belongs to S 0 and inf(S 0 ) does not.
If S is a set of positive real numbers that is bounded above and if k is some
positive real number, let kS be the set of positive real numbers obtained by
multiplying every element of S by the number k. Clearly, the set kS is equally
bounded above. It can be proved that sup(kS) = k sup(S), a result that will be
used subsequently.
We are now prepared for a precise definition of the length of an arc of circle
and, ipso facto, of the circumference of a circle. The following preliminary
Lemma is needed:

Lemma. Let triangle 4P QR be isosceles with P Q = P R. If S is a point on

−−→ −→
ray P Q with Q between P and S and T is a point on ray P R with R between
P and T , then ST > QR.

Proof. We may assume without loss of generality that P S < P T , as shown. Let
−→
U be the point on ray P R such that U is between R and T and SU is parallel to
SU
QR. Then 4P RQ ∼ 4P U S so that QR = PP Q
S
> 1, and therefore SU > QR.
Since ]1 is a base angle of the isosceles triangle 4P U S, ]1 is acute and, hence,
]2 is obtuse. Therefore, m ]3 < m ]2 and by Theorem 43, ST > SU .

We now return to the earlier broken line Bn inscribed in arc AB

˜ with length
sn . Henceforth, let S be the set of the lengths of all broken lines that can be
222 CHAPTER 11. ON ARC LENGTHS AND SECTOR AREAS

inscribed in the given arc AB.

˜ Clearly S is a set of positive real numbers and
contains infinitely many elements.

Lemma. The set S is bounded above.

Proof. Consider any square containing the entire circle of which AB ˜ is an arc,
in its interior, as shown. Let D be the center of this circle. Let A = A0 ,
A1 , A2 , . . . , An = B be an arbitrary sequence of points on AB ˜ in the order
0 −−→
from A to B. Let Ai be the (unique) intersection of ray DAi with the square.
Then, for each i, 1 ≤ i ≤ n, triangle 4DAi−1 Ai is isosceles, and by the first
Lemma, Ai−1 Ai < A0i−1 A0i . Thus, the length sn of the broken line inscribed in
arc AB
˜ with vertices A0 = A, A1 , . . . , An = B is always less than the length
A0 A01 + A01 A02 + . . . + A0n−1 B 0 which is a finite number because it is a fraction
of the perimeter of the square. Clearly, this argument does not depend on the
radius of the circle nor on the side length of the square, and remains valid for
any value of n, no matter how large. Therefore, the set S of all sn is bounded
above.

We may now define the length of an arc of circle in terms of the mathematical
concepts introduced above in relation to the given arc AB:
˜

Definition 74. The length s of the arc of circle AB

˜ is s = sup(S).

That this definition is not vacuous is guaranteed by the second Lemma which
ascertains that the set of the lengths of all possible broken lines which can be
inscribed in arc AB
˜ is bounded above and by the Completeness Axiom which
asserts that this set has a unique least upper bound.
The circumference of a circle can now be defined rigorously by inscribing an
n-gon with vertices A0 , A1 , . . . , An−1 , A0 in the given circle and by considering
the perimeter pn of this n-gon, that is, pn = A0 A1 + A1 A2 + . . . + An−1 A0 . Let
P be the set of the perimeters of all possible inscribed polygons. Then it is easy
to see that the set P is again bounded above by the perimeter of the square
used in the second Lemma. Then the following definition is justified:

Definition 75. The circumference p of a circle is p = sup(P ).

11.1. ARC LENGTH 223

In their very essence, Definitions 74 and 75 provide a unique positive real

number for the length of an arc of circle and for the circumference of a circle.
It can then be proved that these numbers do not depend on the particular
way the inscribed broken line or the inscribed polygon is chosen and on how the
number of their vertices has to be increased in order to make them “closer” to
the arc or to the circle. Furthermore, the following theorem can be proved:

Theorem 132 (Additivity of Arc Lengths). Let AB ˜ and BC ˜ be arcs of the

same circle with only the endpoint B in common. If s1 , s2 , s are the lengths of
arcs AB,
˜ BC ˜ and ABC,
˘ respectively, then s = s1 + s2 .

Proof. We omit the proof of this theorem because of its rather technical nature.

We can now prove the theorem which establishes the existence of the famous
real number π which is defined as the ratio of the circumference p of a circle of
radius r to its diameter 2r:
Theorem 133. Let C and C 0 be circles with radii r and r0 and circumferences
0 p p0
p and p , respectively. Then 2r = 2r0 .

Proof. Without loss of generality, we may assume that circles C and C 0 are
concentric with common center D, as shown. Let Ai−1 Ai be the ith side of
a polygon inscribed in C. To each such polygon, there corresponds a similar
−−−−→ −−→
polygon inscribed in C 0 whose vertices are the intersections of rays DAi−1 , DAi
with circle C 0 , as i runs through the vertices of the original polygon. Then
A0i−1 A0i DA0i r0
4DAi−1 Ai ∼ 4DA0i−1 A0i and therefore, we can write: Ai−1 Ai = DAi = r .
p0 0
Hence, if the perimeters of these polygons are pn and p0n , then pnn = rr . Now,
let P be the set of the perimeters of all polygons inscribed in C and P 0 be
0
defined similarly. Then, with our previous notation, P 0 = rr P , and by the
0 0
result mentioned earlier (using k = rr ), it follows that sup(P 0 ) = rr sup(P ) or
0 p0 p
p0 = rr p i.e. 2r 0 = 2r , which was to be proved.

Theorem 133 shows that the ratio of the circumference to the diameter is the
same for all circles and can, justifiably, be designated by a “universal” symbol
π. The immediate corollary to Theorem 133 is the well-known formula p = 2πr
for the circumference of a circle with radius r.
224 CHAPTER 11. ON ARC LENGTHS AND SECTOR AREAS

The number π was long suspected to be irrational, but the proof thereof had
to wait until 1761 when the German mathematician Johann Heinrich Lambert
(1728–1777) produced the first rigorous proof for the irrationality of π. Some
rational approximations which have been used for π throughout the ages are 3,
3.14, 22/7, 3.1416, 355/113.
In analogy with the proof of Theorem 133, one may prove the following
theorem:
Theorem 134. Let AB ˜ and A ˘0 B 0 be arcs of the same degree measure, in circles
0
of radii r and r and with lengths s and s0 , respectively. Then rs = rs0 .
0

Proof. Left as an exercise.

Definition 76. The ratio rs of the length s of an arc of circle to its radius r is
the radian measure of the arc.
The ratio rs is clearly a pure number, being the quotient of two lengths.
Moreover, Theorem 134 asserts that the radian measure of an arc depends only
on the degree measure of the arc, or, equivalently, only on the measure of the
central angle which intercepts the arc, and is independent of the radius of the
circle to which the arc belongs. Thus, if the radian measure of an arc of length
s and radius r is the number j, then, by definition, s = rj. It is immediately
recognized that an arc of circle of radian measure 1 has a length equal to the
radius of the circle, and that a full circle has radian measure 2π. Thus, a useful
correspondence can be established between the degree measure and the radian
measure of arcs of circle: 180 degrees correspond to π radians, 90 degrees to π2
radians, and so on. Conversely, one radian corresponds to the degree measure
180
π ≈ 57.2957795 or approximately 57.3 degrees.
Thus, an arc of degree measure q ◦ and radius r has the length 180
π ◦
◦ q r and a
180◦
central angle with degree measure π intercepts an arc whose length is equal
to the radius of the circle.

11.2 Sector Area

We move on to the definition and calculation of the area of circles and circular
sectors.
11.2. SECTOR AREA 225

Regrettably, due to the rather restricted definition of the concept of area

as developed in Chapter Seven on the basis of Axiom 16, our approach to the
notion of area of circles and circular sectors will have to remain even more
sketchy than the considerations which led to the definition of the length of an
arc of circle.
Definition 77. A circular region (or disk ) is the union of a circle and its
interior. If AB
˜ is an arc of circle with center C and radius r, the union of all
radii CP where P is on AB˜ is called a circular sector K with radius r and
boundary arc AB.
˜

In the same fashion that we abbreviated “the area of a triangular region” to

“area of a triangle” in Chapter Seven, we shall say “area of a circle/sector” as
short for “area of a circular region/circular sector” for the sake of brevity.

Let arc AB
˜ with center C and radius r be given. Inscribe a broken line
Bn in arc AB such that all the n chords constituting Bn are congruent with
˜
common length bn (see Chapter Ten for the possibility of inscribing regular
polygons in a circle). Let again A0 = A, A1 , . . . , An = B be the vertices
of Bn . All the isosceles triangles 4CAi−1 Ai are congruent, so their altitudes
from the circle center to the chords Ai−1 Ai are equally congruent. Let an be
the length of this common altitude (called apothem of the regular inscribed
polygon). Then, the area of each of these n triangles is 21 an bn , and the area
α(Kn ) of the union Kn of these n triangular regions is 21 nan bn . Let s be the
length of arc AB.
˜ We want to know what happens to bn , an , and α(Kn ) as the
number n of vertices of Bn increases without bound and becomes “as large as
one pleases”. Now, nbn is the length sn of the inscribed broken line Bn and
by Definition 70, sn is always less than the length s of arc AB,
˜ i.e. nbn < s
or bn < ns . Thus, as the number n of vertices of Bn increases, bn decreases
and can be made as small as one “pleases” by letting n become “sufficiently”
large. For instance, if the arc length s is equal to, say, 10 units and we want
226 CHAPTER 11. ON ARC LENGTHS AND SECTOR AREAS

bn to be, say, smaller than 10−10 = 1/1010 unit, that is, if we want bn < 10−10 ,
−10
then take n so large that 10 n < 10 i.e. let n > 109 . This procedure can be
−50
repeated to make bn smaller than 10 or smaller than 10−1001 and so on by
taking a large enough value for n. In other words, by increasing n we can make
bn as small as we wish. On the other hand, it is reasonable to accept that as
n increases, the length sn of Bn approaches the arc length s. Finally, in the
isosceles triangle 4CAi−1 Ai , the triangle inequality yields: r < an + 21 bn or
r − 12 bn < an . Since an < r for all values of n, we have the string of inequalities:
r − 12 bn < an < r, and it is intuitively acceptable that as the number n of the
vertices of Bn becomes increasingly large, then bn becomes increasingly small
and so the difference r − 21 bn gets closer to r, while an remains “sandwiched”
between r and a number that becomes closer and closer to r. Hence, an must
approach r as n increases. Therefore, the area α(Kn ) = 21 nbn an will approach
the value 12 sr. All these heuristic considerations can be made perfectly rigorous
with the tools of Calculus, and lead to the following theorem which we enunciate
without its proof:
Theorem 135. The area of a circular sector with radius r and boundary arc
of length s is equal to 21 sr.
An immediate consequence of Theorem 135 is that the area of a disk of
radius r (that is, when s = 2πr) is given by the well-known formula πr2 . If the
length s of the boundary arc of the circular sector is expressed in terms of the
radian measure φ of the central angle which intercepts it, then the area of the
circular sector reads 21 r2 φ. Similarly, if the degree measure of the central angle
which intercepts the boundary arc of the sector is q ◦ , the area of the sector is
π ◦ 2
360◦ q r .
11.2. SECTOR AREA 227

Exercises
1. Determine the radian measure of a central angle whose degree measure is:
60◦ ; 120◦ ; 135◦ ; 40◦ ; 36◦ ; 15◦ .
Give your answers in rational multiples of π.

2. Determine the degree measure of a central angle whose radian measure is:
π π π 7π 2π
4 ; 6 ; 2 ; 12 ; 5 .

3. Determine the length of an arc of circle of radius 2 inches whose central

angle has
5π
(a) a radian measure of 8
(b) a degree measure of 175◦ .

4. Determine the measure in degrees and in radians of the central angle of

(a) a minor arc of radius 14 cm whose length is 7 cm

19π
(b) a major arc of radius 6 cm whose length is 2 cm.

5. A segment of a circle is the region bounded by a chord and an arc of the

circle, as shown. Determine the area of a segment of a circle of radius 6
cm when the measure of the central angle is 120◦ . Repeat when the radian
measure of the central angle is π4 .

6. The side of a square is 12 cm long. What are the circumferences of its

inscribed and circumscribed circles?

7. In the figure shown below, square XY ZW is inscribed in a circle with

center O, and square ABCD is circumscribed about this circle. The di-
←→ ←→
agonals of both squares lie in AC and BD. Given that square P QRS is
formed when the midpoints P , Q, R, S of AX, BY , CZ, DW are joined,
is the perimeter of this square greater than, or less than, or equal to the
circumference of the circle with center O? Let OX = 1 unit and justify
your answer by computation.
228 CHAPTER 11. ON ARC LENGTHS AND SECTOR AREAS

8. Given a square whose side is 10 inches long, what is the area enclosed
between its circumscribed and inscribed circles?
9. An equilateral triangle is inscribed in a circle. If the side of the triangle
is 12 ft long, what is the radius of the circle?
10. Two concentric circles with common center P and radii r and R with
r < R are given. AC is a chord of the larger circle and is tangent to the
smaller circle at B. Determine the area of the shaded annulus (ring) in
terms of AC.

11. In a sphere whose radius is 10 cm, sections are made by planes whose
distances from the center are 3 cm and 5 cm. Which section will have the
larger area? Justify your answer.
11.2. SECTOR AREA 229

12. In the figure shown, ABCD is a square in which E, F , G, are the midpoints
of segments AD, AC, CB, respectively. AF ˜ and F ˜ C are circular arcs with
centers E and G, respectively. If the length of the side of the square is s,
determine the area of the shaded region.

13. An archery target is constructed in the following way:

−−→ −−→
Rays OM and P N are parallel. A circle with center O and radius r equal
−−→
to the distance between the rays is drawn intersecting OM at Q. QA
is drawn perpendicular to QM , and a circle with center O and radius
r1 = OA is drawn. This process is repeated by drawing perpendiculars at
R and S, and circles with centers O and radii r2 = OB and r3 = OC are
constructed.

Determine r1 , r2 , r3 in terms of r and show that the areas of the inner

circle and the three rings labeled b, c, d are equal.

14. In the figure shown below, semicircles are drawn with each side of the
right triangle 4ABC as diameter. The area of each region is indicated by
a lower case letter. Prove: r + s = t.
230 CHAPTER 11. ON ARC LENGTHS AND SECTOR AREAS

15. A continuous belt runs around two wheels of radii 6 and 30 cm. The
centers of the wheels are 48 cm apart. Determine the length of the belt.

16. In the figure shown, ABCD is a square whose side is 8 inches long. With
the midpoints of the sides of the square as centers, arcs of circle are drawn
tangent to the diagonals. Find the area enclosed by the four arcs.
Chapter 12

On Solids and Their

Volumes

12.1 Prisms and Cylinders, Pyramids and Cones

The theory of volume will be treated in a style analogous to that of Chapter
Seven (on area measurement). In other words, we shall base our derivations on
postulates and shall not attempt to prove the existence of a volume function
from the fundamental axioms of Euclidean Geometry.
In three-dimensional space, the natural counterparts of the basic two-dimensional
convex polygonal regions introduced in our theory of area would be convex poly-
hedral regions, i.e. sets of points in space whose boundaries are line segments
(“edges”) and convex polygons (“faces”), but, since we are interested in cones,
cylinders, and spheres along with polyhedra like prisms and pyramids, we will
have to adopt a slightly more general approach and consider certain bounded
convex sets of points in space as fundamental regions for our theory of volume.
Henceforth, a set of points shall be called bounded when the distance between
any two points of the set is less than some positive number.
We begin with a number of formal definitions.

Definition 78. Let Π1 and Π2 be two parallel planes, L a line which intersects
Π1 in one point, and K a convex (bounded) set in Π1 which does not intersect
L. For each point P in K, let P P 0 be the line segment parallel to L with P 0 in
Π2 . If the set K is a (convex) polygonal region/circular region (disk) D in Π1 ,
then the union of all such line segments P P 0 is called a prism/circular cylinder
with directrix L. The polygonal region/disk D in Π1 is the (lower) base of the
prism or circular cylinder while the set of all points P 0 in Π2 is called its upper
base. The distance h between Π1 and Π2 is the altitude of the prism/circular
cylinder. If L is perpendicular to Π1 , then the prism/circular cylinder is called
a right prism or circular cylinder.

231
232 CHAPTER 12. ON SOLIDS AND THEIR VOLUMES

It goes without saying that non-circular cylinders, i.e. cylinders whose

(lower) bases are ellipses or other convex regions enclosed in smooth closed
curves, are perfectly conceivable but the circular cylinder is by far the most
common species and the only one considered in this course. Henceforth, cir-
cular cylinders shall be abbreviated to cylinders, and lower bases of prisms or
cylinders shall be merely called bases for the sake of brevity. On the other
hand, prisms are classified according to the polygonal shape of their bases: a
triangular prism has a triangular region for lower base, and so on. If a prism
has a parallelogram for base, then it is called a parallelepiped. A cross-section
of a prism or cylinder is its intersection with a plane parallel to the plane of its
base, provided that this intersection is not empty.
The precise definitions of pyramids and (circular) cones are similar to those
of prisms and cylinders in several respects, and many of the terms already
introduced in Definition 78 and thereafter carry over with obvious adjustments
and shall be used without formal definitions.

Definition 79. Let a convex polygonal region/disk be given in a plane Π and a

point V not in Π. The union of all line segments P V where P is a point in the
polygonal region/disk is called a pyramid/(circular) cone with the polygonal
region/disk for base and the point V for vertex. The distance h from V to Π is
the altitude of the pyramid or cone. If the center of the base disk of a (circular)
cone is the foot of the perpendicular from V to Π, the cone is called a right
(circular) cone.

Since only cones with disks for bases will be considered, (circular) cones shall
be abbreviated to just cones, while pyramids will be classified according to the
shape of their bases: triangular pyramids, etc.
12.1. PRISMS AND CYLINDERS, PYRAMIDS AND CONES 233

Theorem 136. The cross-sections of a triangular prism are triangles congruent

to the prism base.

Proof. Let the triangular region 4ABC be the base of a prism with directrix L
and let a plane parallel to the plane of 4ABC intersect AA0 , BB 0 , CC 0 in D, E,
F , respectively. By Definition 78, AD and BE are parallel hence coplanar, and
←→ ←→
by Theorem 87, the intersections DE and AB of two parallel planes by a third
plane are parallel. Therefore, ABED is a parallelogram, so that AB = DE.
Similarly, AC = DF and BC = EF . Hence, 4DEF ∼ = 4ABC (S.S.S.).

The generalization of this theorem to prisms with (convex) polygonal regions

for base is straightforward since every such region can be triangulated (see
Definition 57) and along with a triangulation of its base, the prism will be cut
up into the union of non-overlapping triangular prisms for which Theorem 136
is valid. In particular, it follows that the upper and lower bases of a prism are
congruent polygons.

Theorem 137 (Prism Cross-Section Theorem). All cross-sections of a prism

have the same area.

Proof. A triangulation of the polygonal region in Π1 which is the base of the

prism cuts up the prism into non-overlapping triangular prisms. By Theorem
136, each triangle in the base is congruent to the corresponding triangle in the
cross-section, and these pairs of triangles have equal areas. Then, the additivity
of the area measure (Axiom 16(iii)) ensures that the areas of the base and of the
cross-section are equal too (see also the closing remark at the end of Chapter
Seven).
234 CHAPTER 12. ON SOLIDS AND THEIR VOLUMES

In an entirely similar manner, one proves the following theorem:

Theorem 138 (Cylinder Cross-Section Theorem). All cross-sections of a cylin-

der are congruent disks and have the same area.

The following terms are commonly used in relation to prisms and we merely
enumerate them below for the sake of completeness, using the notations of
Definition 78:
A lateral edge of a prism is a line segment AA0 where A is a vertex of the
prism base. A lateral face of a prism is the union of all line segments P P 0 where
P is a point of a given side of the prism base. The lateral surface of a prism is
the union of all its lateral faces. The total surface of a prism is the union of its
lateral surface and its bases. A formal proof that the lateral faces of a prism
are parallelograms (or rectangles, in the case of a right prism) requires a careful
examination of separation properties and is both long and involved. We shall
therefore accept this fact without proof.
Two more analogues of the Prism Cross-Section and Cylinder Cross-Section
Theorems (Theorems 137 and 138) for cones and pyramids are next:

Theorem 139 (Cone Cross-Section Theorem). The cross-section at altitude k

of a (circular) cone with altitude h whose base is a disk of radius r is a disk of
radius r0 = h−k
h r.

Proof. Consider a cone with vertex V , circular base D with radius r and center
C in plane Π, and altitude h. Let Πk be a plane parallel to Π, on the same
side of Π as V , at distance k < h from Π. Let A and A0 be the feet of the
perpendicular from V onto Π and Πk , respectively. Let C 0 be the intersection
of segment V C with plane Πk , as shown. Finally, let B be any point on the
boundary circle of the base disk D and B 0 be the intersection of V B with Πk .
Since Π and Πk are parallel, any plane that intersects Π and Πk intersects
them in two parallel lines. Hence, A0 B 0 is parallel to AB and B 0 C 0 is parallel
12.2. VOLUMES OF SOLIDS 235

to BC, and 4V A0 B 0 ∼ 4V AB and also 4V B 0 C 0 ∼ 4V BC. Consequently,

V A0 0 0 0

VA =
h−k
h = VV BB = CCBB
, so that we have C 0 B 0 = h−k
h CB. Therefore, if
0 0 h−k
CB ≤ r, then C B ≤ h r, and conversely. It follows that the cross-section
Dk of the cone with plane Πk is a disk of radius r0 = h−k
h r.

Thus, the ratio of the area of the cross-section at altitude k to the base area
of a (circular) cone with altitude h is equal to ( h−k 2
h ) .

A similar result can be proved along almost identical lines—with minor

adjustments—for triangular pyramids and, more generally, for pyramids with
(convex) polygonal bases through triangulation of the bases and cutting-up the
pyramids into non-overlapping triangular pyramids. The details of the proof of
this statement are left as a recommended exercise. Again, the ratio of the area
of a cross-section at altitude k of a pyramid with altitude h to the base area of
the pyramid equals ( h−k 2
h ) .

12.2 Volumes of Solids

We are now ready to introduce the axiom governing the existence and properties
of a volume function for a sufficiently broad class of solids. In the following,
we shall restrict ourselves to the category V of convex, bounded sets of points
in space. Then, the union and the intersection of finitely many such sets are
again convex and bounded. In our elementary theory of solid mensuration,
this restriction will appear to have no effective importance: it only serves to
exclude certain subtle “pathologies” and, for all practical purposes, we shall
identify these sets with the familiar “solids” we have already defined, i.e. cones,
pyramids, spheres, etc.
Axiom 18 (Volume Measure). There exists a function v defined on the class
V of convex, bounded sets of points in space with the following properties:
(i) To each convex, bounded set M ∈ V , v assigns a non-negative real number
v(M ) called the volume of M such that
(ii) v is monotonic: if M, N ∈ V and M is entirely contained in N , then
v(M ) ≤ v(N ),
236 CHAPTER 12. ON SOLIDS AND THEIR VOLUMES

(iii) v is additive: if M, N ∈ V and v(M ∩ N ) = 0, then v(M ∪ N ) = v(M ) +

v(N ),
(iv) the volume of a rectangular parallelepiped is the product of its base area
and its altitude, and
(v) v obeys Cavalieri’s Principle: Given two convex, bounded sets K and K 0
of points in space (“solids”) and a plane Π, if, for every plane intersecting
K and K 0 and parallel to Π, the two intersections have equal areas, then
K and K 0 have the same volume.

In the figure shown, the two “solids” K and K 0 are elements of V and Π
is a “horizontal” plane. Property (v) asserts that if every cross-section D of K
and cross-section D0 of K 0 by the same horizontal plane have equal areas, i.e.
α(D) = α(D0 ), then v(K) = v(K 0 ). Cavalieri’s Principle is the key to the actual
calculation of volumes. Clearly, if v is a volume measure satisfying properties
(i), (ii), (iii), then multiplying v by a positive constant yields another volume
measure. Thus, property (iv) serves to define a “unit of volume” in the same
manner that Axiom 17 fixed a unit of area measure. It should be stressed that
the existence of a volume measure with the properties (i) through (iii) listed
in Axiom 18 can be proved from our axioms of Euclidean Geometry but this
proof is technically very difficult and requires advanced mathematics, and that
properties (iv) and (v) will nonetheless have to retain their status of independent
necessary axioms.
Before proceeding with the computation of the volume of some fundamental
solids, we need a preliminary result:
Theorem 140. If M is a convex, bounded set contained in a plane, then
v(M ) = 0.
Proof. First notice that being bounded, M is contained in the interior of some
rectangular region R in the same plane that contains M . Let α(R) = ab (Axiom
17) where a and b are the lengths of the sides of this rectangle. Then, for any
positive real number h, the bounded set M is contained in the interior of a
rectangular parallelepiped K with v(K) = abh (Axiom 18(iv)). Now, the height
h of K, along with v(K), can be made as small as one pleases: given any positive
number  (no matter how small), v(K) can be made yet smaller than this  by
12.2. VOLUMES OF SOLIDS 237


choosing h < ab . Therefore, for every  > 0, v(M ) ≤ v(K) <  (Axiom 18(iii)).
Since v(M ) ≥ 0 (Axiom 18(i)), it follows that v(M ) must necessarily be equal
to zero.

Theorem 141. Let K be a right triangular prism with altitude h and the right
triangular region T for base. Then v(K) = hα(T ).

Proof. Construct another right prism K 0 with a right triangular base T 0 in such
a way that the union T ∪ T 0 is a rectangular region and the union K ∪ K 0 is a
rectangular parallelepiped with v(K ∪ K 0 ) = abh (Axiom 18(iv)) where a and
b are the lengths of the sides of the right triangular region T , as shown. By
construction and by virtue of the Cavalieri’s Principle, v(K) = v(K 0 ). On the
other hand, Theorem 140 implies that v(K ∩K 0 ) = 0 because K and K 0 intersect
each other in a plane rectangular set only. Hence, v(K ∪ K 0 ) = v(K) + v(K 0 ) =
2v(K) (Axiom 18(iii)). Consequently, v(K) = 12 abh = hα(T ), which was to be
proved.

Notice that Cavalieri’s Principle ensures that the volume of a triangular

right prism K with altitude h whose base is an arbitrary triangular region T
with area α(T ) is also equal to hα(T ): consider a right prism K 0 with the same
altitude as K and a right triangular region T 0 for base such that α(T ) = α(T 0 ).
Then, by Theorem 137, Cavalieri’s Principle applies to the prisms K and K 0 and
Theorem 141 implies v(K) = v(K 0 ) = hα(T ). The generalization of Theorem
141 to arbitrary prisms is carried out in two steps:

Theorem 142. Let K be a right prism with altitude h and a (convex) polygonal
region B as base. Then, v(K) = hα(B).
238 CHAPTER 12. ON SOLIDS AND THEIR VOLUMES

Proof. The base B of the prism can be triangulated into the union of a finite
number n of triangular regions Ti , i = 1, 2, . . . , n. Consequently, the prism K
is the union of n triangular right prisms K1 , . . . , Kn , each with a triangular
base Ti and the same altitude h. By Theorem 141 and the remark following
the proof of that Theorem, v(Ki ) = hα(Ti ) for each i, i = 1, . . . , n. Since each
prism Ki intersects the others in plane bounded regions, Theorem 140 implies:
v(K) = v(K1 )+v(K2 )+. . .+v(Kn ) = hα(T1 )+hα(T2 )+. . .+hα(Tn ). Therefore,
v(K) = h[α(T1 ) + α(T2 ) + . . . + α(Tn )] = hα(B) by virtue of the additivity of
the area measure (see Chapter Seven).

Theorem 143. The volume of any prism is the product of its altitude and its
base area.

Proof. Let K 0 be a right prism with the same base area and the same altitude
h as the given prism K. By Theorem 142: v(K 0 ) = hα(B 0 ) = hα(B), where B
and B 0 are the bases of K and K 0 , respectively, as shown. Then, by Theorem
137 and the Cavalieri’s Principle, it follows that v(K) = v(K 0 ) = hα(B).

Theorem 144. The volume of a cylinder with altitude h whose base is a disk
of radius r is equal to πr2 h.

Proof. Let K be the given cylinder, and let L be any prism with base in the
same plane as the base of the cylinder, with the same base area πr2 , and the
same altitude h as K. Then, by Theorem 143, v(L) = πr2 h. By Theorem 138
(Cylinder Cross-Section Theorem), all same altitude cross-sections of K and L
have the same area. Then, by Cavalieri’s Principle, v(K) = v(L) = πr2 h.

We now turn to the calculation of volumes of pyramids and cones, and begin
with a preliminary result on triangular pyramids:

Theorem 145. Two pyramids with the same base area and the same altitude
have the same volume.
12.2. VOLUMES OF SOLIDS 239

Proof. Let the common altitude and the common base area of the two pyramids
be h and a, respectively. By the Pyramid Cross-Section Theorem (i.e. the
“pyramid version” of Theorem 139), the cross-sections at altitude k of these
pyramids have equal areas, and by Cavalieri’s Principle, the two pyramids have
equal volumes.

Theorem 146. The volume of a pyramid is one-third the product of its base
area and its altitude.

Proof. We first prove the theorem for a triangular pyramid L with base area a
and altitude h. Consider a right triangular prism K with the same base area
and the same altitude as L, and cut up this prism into three triangular pyramids
K1 , K2 , and K3 , one of them being the original one. as shown below:

Let K3 be the triangular pyramid with the same base and altitude as L
and having the edge AD perpendicular to the triangular base 4DEF of the
right prism K. Now, 4BCF ∼ = 4BEF (why?). If we construe K1 and K2 as
triangular pyramids with vertex A as “top vertex,” then K1 and K2 have equal
240 CHAPTER 12. ON SOLIDS AND THEIR VOLUMES

base area and the same altitude, because the altitude of each of them is the
perpendicular distance from A to the plane of the four vertices B, C, F , and E.
Hence, by Theorem 145, we have v(K1 ) = v(K2 ). Next, regard K2 and K3 as
triangular pyramids with vertex F as “top vertex.” Since 4ABE ∼ = 4ADE and
K2 and K3 have the same altitude (distance of F from the plane of the vertices
A, B, D, and E), v(K2 ) = v(K3 ). Now, the three pyramids K1 , K2 , K3 intersect
each other only along bounded planar sets (namely 4ABF and 4AF E) which
have no volume (Theorem 140). Therefore, v(K) = v(K1 ) + v(K2 ) + v(K3 ) =
3v(K3 ) = 3v(L). Consequently, we obtain: v(L) = 31 v(K) = 13 ah. Next,
consider a pyramid P with a polygonal base of area a and altitude h. Let L be
a triangular pyramid with base area a and the same altitude h. By the Pyramid
Cross-Section Theorem, cross-sections at the same altitude of P and L have
the same area, and Cavalieri’s Principle asserts that v(P ) = v(L) = 31 ah. This
completes the proof of the theorem.
Theorem 147. The volume of a cone is one-third the product of its base area
and its altitude.

Proof. Given a cone with base area a = πr2 and altitude h, consider any pyramid
with its base in the same plane as the base of the cone, with the same base area
a, and the same altitude h. Then, the same-altitude cross-sections of the cone
and of the pyramid have the same area (why?), and by virtue of the Cavalieri’s
Principle, the cone and the pyramid have the same volume, that is 13 ah.
Our last application of the Cavalieri’s Principle is the calculation of the
volume of a sphere:
Theorem 148. The volume of a sphere of radius r is 43 πr3 .
12.2. VOLUMES OF SOLIDS 241

Proof. Let S be a sphere of radius r plus its interior, and let Π be a tangent
plane to the sphere. In Π take a circle of radius r and consider a right cylinder
C with this circle as base and altitude 2r, lying on the same side of Π as the
sphere and such that the perpendicular from the center of the sphere S onto
Π passes through the center of the cylinder base. Thus, sphere S is inscribed
in the interior of cylinder C. Let L be the set of points of C which do not
belong to S, that is, v(L ∩ S) = 0. Then, C = L ∪ S and v(C) = v(L) + v(S)
(Axiom 18(iii)) or v(S) = v(C) − v(L) = 2r(πr2 ) − v(L) or v(S) = 2πr3 − v(L)
(Theorem 144). Thus, the problem of calculating the volume v(S) is reduced
to the calculation of the volume v(L). To that end, we strive at finding a solid
whose volume we know and which has the same equal-altitude cross-sectional
areas as L, so that Cavalieri’s Principle can be applied.

Now, in a “vertical” cross-section of S and C by a plane perpendicular to Π

through the center of S, the cylinder appears as a rectangle and the sphere as
a circle inscribed in that rectangle, as shown. The “horizontal” cross-section of
L (by a plane parallel to Π) at altitude k above the
√ center of the sphere is then
a ring with outer radius r and inner radius s = r2 − k 2 , by the Pythagorean
Theorem. Therefore, the area of the “horizontal” cross-section of L at altitude
k above the center of the sphere is ak = πr2 − πs2 = π[r2 − (r2 − k 2 )] = πk 2 .

Next, consider the circular double-cone L0 with the two bases of the cylinder
C as upper and lower bases and vertex at the center of the sphere. A “vertical”
cross-section of this double-cone is shown and a “horizontal” cross-section of L’
at altitude k above its vertex is a disk of radius k and has area πk 2 . Thus,
v(L) = v(L0 ). On the other hand, v(L0 ) = 2( 31 πr2 )r (Theorem 144), so that
242 CHAPTER 12. ON SOLIDS AND THEIR VOLUMES

v(L) = 23 πr3 . Therefore, we obtain: v(S) = 2πr3 − 23 πr3 = 43 πr3 , which was to
be proved.
For reference, we state—without proof—that the surface area of a sphere of
radius r is equal to 4πr2 , i.e. four times the area of a great circle of the sphere.
12.2. VOLUMES OF SOLIDS 243

Exercises

1. Prove that two non-adjacent lateral edges of a prism are coplanar, and
that the intersection of their plane with the prism is a parallelogram.

2. Prove that the area of the lateral surface of a right prism is the product
of the perimeter of its base and the length of a lateral edge.

√
3. The sides of a cross-section of a triangular prism have lengths 3 cm, 3 3
cm, and 6 cm. Determine the angle measures and the area of another
cross-section of this prism which is at altitude 1 cm higher than the given
cross-section.

4. Prove the Pyramid Cross-Section Theorem:

The cross-section of a triangular pyramid is a triangular region similar to

the base of the pyramid, and the ratio of the area of the cross-section at
altitude k to the area of the base is ( h−k 2
h ) . (Note that h should be the
full altitude and k should be the altitude of the frustum. The diagram is
wrong.)
244 CHAPTER 12. ON SOLIDS AND THEIR VOLUMES

5. Given are two pyramids, one triangular and the other hexagonal, with
equal base areas and the same altitude 6 inches. The area of the cross-
section of the triangular pyramid 2 inches above the base is equal to 25
square inches. What is the area of the same-altitude cross-section of the
hexagonal pyramid?

6. A regular pyramid is a pyramid whose base is a regular polygonal re-

gion such that the foot of the perpendicular from the pyramid’s vertex is
equidistant from all the vertices of the base. Prove that the lateral faces
of a regular pyramid are congruent isosceles triangles, and that the area
of the lateral surface of a regular pyramid is one-half the product of the
perimeter of the base and the altitude of one lateral face.

7. Prove the following theorem:

In any pyramid, the ratio of the area of a cross-section to the area of the
base is equal to ( ab )2 where a is the length of a lateral edge of the smaller
pyramid and b is the length of the corresponding edge of the original
pyramid.
12.2. VOLUMES OF SOLIDS 245

8. A lump of metal submerged in a rectangular tank of water 20 cm long and

8 cm wide raises the water level by 4.6 cm. What is the volume of the
lump?

9. If one edge of the base of a large regular hexagonal pyramid is 12 feet and
the altitude of the pyramid is 9 feet, what is the area of the lateral surface
and the volume of this pyramid?

10. A plane bisects the altitude of a pyramid and is parallel to the plane of
the pyramid base. Determine the ratio of the volumes of the solids above
and below the plane.

11. A monument has the shape of an obelisk—a square pyramid cut off at a
certain height by a plane parallel to its base and capped with a second
square pyramid, as shown. The vertex of the second pyramid is 2 meters
above its base and 32 meters above ground level. If the original pyramid
had not been truncated, it would have been 60 meters tall. Find the
volume of the obelisk if each side of the base, at ground level, is 4 meters
long.
246 CHAPTER 12. ON SOLIDS AND THEIR VOLUMES

12. State and illustrate a principle, corresponding to Cavalieri’s Principle,

having the conclusion that two plane regions have equal areas.

13. A drainage pipe is a cylindrical shell 16 feet long. The inner and outer
diameters of the pipe are 5 and 5.6 feet, respectively. Determine the
volume of clay necessary to manufacture this pipe.

14. A certain cone has a volume of 27 cubic centimeters. Its height is 5 cm.
A second cone is cut from the first by a plane parallel to its base and two
cm below its vertex, as shown.

Determine the volume of the second cone.

15. In the figure shown, one is looking down upon a pyramid, whose base is
a square, inscribed in a right circular cone. The altitude of the cone (and
of the pyramid) is 36 inches and a base edge of the pyramid is 20 inches.
Find the ratio of the volume of the pyramid to that of the cone.
12.2. VOLUMES OF SOLIDS 247

16. A right circular cone stands inside a right circular cylinder with the same
base radius r and the same altitude h. In terms of r and h, find an
expression for the volume of the region enclosed between the cone and the
cylinder.

17. If a plane parallel to the base of a cone (or pyramid) cuts off another cone
(or pyramid), then the solid between the base and this plane is called a
frustum.
A frustum of a cone has lower radius 6 inches, upper radius 4 inches, and
a height of 8 inches, as shown. Find its volume.

18. An ice cream cone is 5 cm deep and has a top diameter of 2 cm. Two
hemispherical scoops of ice cream, each with a diameter of 2 cm, are placed
on top of the cone. If the ice cream melts entirely into the cone, will it
overflow?

19. In the figure shown, the sphere with radius r is inscribed in the interior of
the right circular cone. The measures of the angles between the altitude
of the cone and the sphere radii to the points of tangency are indicated.
Find the volume of the cone in terms of r.
248 CHAPTER 12. ON SOLIDS AND THEIR VOLUMES

20. Half the air is let out of a spherical rubber balloon. If the balloon retains
its spherical shape, how does its final radius compare with its original
radius?
21. The city engineer who is six feet tall walked up to inspect the new spherical
water tank. When he had walked to a spot 18 feet from the point where
the tank rested on the ground, he bumped his head on the tank. If the
city uses 10,000 cubic feet of water per hour, determine how many hours
one full tank would last.
22. A cylindrical container with radius 12 cm and height 20 cm is full of water.
If a sphere of radius 10 cm is lowered into the container and then removed,
what volume of water will remain in the container?
23. Find a method for calculating the area of the lateral surface of a right
circular cylinder with base radius r and height h.
24. The altitude of a right circular cone is 15 inches and the radius of its base
is 8 inches. A cylindrical hole of diameter 6 inches whose axis coincides
with the altitude of the cone is drilled through the cone, leaving a solid
shown in the figure. What is the volume of this solid?
Chapter 13

Introduction to
Non-Euclidean Geometry

13.1 On Neutral Geometry III

In this chapter we develop the fundamentals of a plane, i.e. two-dimensional,
Geometry in which the Euclidean Parallel Postulate (Axiom 15) is replaced by
its logical negation while retaining all the preceding axioms (Axioms 1 through
14) which underlie Neutral Geometry. In doing so, we shall endeavor to show
that a perfectly consistent, albeit counterintuitive, Non-Euclidean Geometry
can be elaborated.
As mentioned earlier (see Chapter 6), by the first half of the 18th century
the problem of proving Euclid’s Fifth Postulate from the axioms of Neutral
Geometry had become both notorious and vexatious. It had become increasingly
accepted that such a proof was impossible, but this was only an opinion which
did require a proof. It turned out that this very unproven idea made the eventual
discovery (or conception?) of a new Geometry contrary to Euclid’s an inevitable
logical possibility. We shall only briefly sketch the very last historical stage of
the genesis of this new Geometry and refer to the bibliography for suggested
further readings.
The great German mathematician and physicist Carl Friedrich Gauss (1777–
1855) seems to have been the first, around 1813 and after some twenty years
of fruitless efforts to prove the Fifth Postulate from Neutral Geometry, to
have acquired a clear insight of a consistent Geometry—which he called Non-
Euclidean—in which the Parallel Postulate is replaced by its negation. Gauss
did not publish his findings for fear of “the outcry from Boeotians,” but Non-
Euclidean ideas were already in the air, as evidenced by the contributions of
a few contemporary German mathematicians, such as C.F. Schweikart, F.L.
Wachter, and F.A. Taurinus. Eventually, the Hungarian Janos Boyai (1802–
1860) and the Russian Nicolai Ivanovich Lobachevsky (1793–1856) indepen-
dently published their works on Non-Euclidean Geometry around 1830. In this

249
250 CHAPTER 13. INTRODUCTION TO NON-EUCLIDEAN GEOMETRY

chapter, we will restrict our attention to the particular Geometry invented (or
discovered?) by Gauss, Bolyai, and Lobachevsky, which is called Hyperbolic Ge-
ometry. Specifically, Hyperbolic Geometry is defined by the axiomatic system of
(plane) Neutral Geometry augmented by the negation of the Euclidean Parallel
Postulate (Axiom 15). Needless to say, following the work of the German math-
ematician G. F. Bernhard Riemann (1826-1866), many other Non-Euclidean
Geometries were introduced and studied, and some of them play a fundamental
role in contemporary theoretical physics1 .
To begin with, we introduce a few nontrivial theorems from Neutral Geom-
etry which were mostly proved by the Italian priest Girolamo Saccheri (1667-
1733) in his fruitless efforts to prove that Euclidean Geometry is the only logi-
cally possible consistent Geometry. In his little opus entitled Euclides ab omni
Naevo Vindicatus (Euclid Freed of Every Flaw), Saccheri made what he thought
were some false assumptions concerning a certain quadrilateral, and by reason-
ing indirectly from these assumptions, he tried to develop contradictions which
could, in turn, be used to prove Euclid’s Parallel Postulate. Instead of this,
Saccheri ended up laying the foundations on which Hyperbolic Geometry would
be later edified.
In the sequel, definitions, theorems, lemmas and corollaries that specifically
pertain to Hyperbolic Geometry will be preceded by the letter H in order to
stress their Non-Euclidean, that is, hyperbolic character.
The first important and insufficiently known property concerning the sum
of the measures of the angles (hereinafter, “angle sum” for short) of a triangle
in Neutral Geometry is contained in the following Lemma:

Lemma. Given a triangle 4ABC, there exists a triangle 4A1 B1 C1 which has
the same angle sum as 4ABC and in which, say, m ]A1 ≤ 21 m ]A.

Note: Needless to say, corresponding angles ]A and ]A1 are not special in
any way.

Proof. Let E be the midpoint of side BC and let point F be chosen on ray
−→
AE such that AE = EF and E is between A and F (Point Plotting Theorem).
Then, 4BEA ∼ = 4CEF (S.A.S.) and hence, their corresponding angles are
congruent. We show that 4AF C is the triangle 4A1 B1 C1 we are seeking. We
have: m ]A2 = m ]AF C, m ]F CB = m ]B, and
1 A very readable first article on this subject is The Curvature of Space, by P. Le Corbeiller,

reprinted in Mathematics in the Modern World, Readings from Scientific American, W.H.
Freeman & Co., 1968
13.1. ON NEUTRAL GEOMETRY III 251

m ]A + m ]B + m ]BCA = m ]A1 + m ]A2 + m ]B + m ]BCA

= m ]A1 + m ]AF C + m ]F CB + m ]BCA
= m ]CAF + m ]AF C + m ]F CA

Thus, 4ABC and 4AF C have the same angle sum. Now, observe: m ]A =
m ]A1 + m ]AF C. In this equation, one of the terms on the right-hand side,
i.e. m ]A1 or m ]AF C must be less than or equal to 21 m ]A on the left-hand
side. If it happens that m ]A1 ≤ 12 m ]A, we are done, and if not, just re-label
the other two vertices of 4AF C as B1 and C1 .

Intuitively speaking, this first Lemma states that we can replace a given tri-
angle by a “slender” one without altering its angle sum. Moreover, the nontrivial
content of this lemma cannot be stressed enough, because in Neutral Geometry
we cannot assume that the angle sum is constant for all triangles: this is a Eu-
clidean theorem whose proof depends heavily on the Euclidean Parallel Axiom
(Axiom 15).
It can be now anticipated that starting with a given triangle, we can con-
struct a sequence of non-congruent triangles, all having the same angle sum as
the initial triangle, and having an angle with arbitrarily small measure. This is
the message of the Saccheri-Legendre Theorem, but we need yet another lemma:

Lemma. The sum of the measures of any two angles of a triangle is less than
180◦ .

Proof. Consider triangle 4ABC and select, say, angles ]A and ]B. Consider
−−→
ray CB and select point D on it such that B is between C and D. Then,
]ABD is an exterior angle of 4ABC, and by the Exterior Angle Theorem
(Theorem 39), m ]ABD > m ]A. But m ]ABD = 180◦ − m ]B (why?), so
that by substitution, we obtain 180◦ − m ]B > m ]A or 180◦ > m ]A + m ]B,
which proves the Lemma for the angles considered. Clearly, one can repeat this
argument for any other pair of angles of the given triangle.

Theorem 149 (Saccheri-Legendre). The angle sum of any triangle is less than
or equal to 180◦ .

Proof. Suppose the contrary (RAA argument). Then there exists a triangle
4ABC with angle sum 180◦ + p◦ , where p is some positive real number. Ap-
plying the first Lemma, there exists a triangle 4A1 B1 C1 with the same angle
252 CHAPTER 13. INTRODUCTION TO NON-EUCLIDEAN GEOMETRY

sum (that is, 180◦ + p◦ ) as 4ABC and such that m ]A1 ≤ 12 m ]A. By ap-
plying the first Lemma to 4A1 B1 C1 , one obtains a second triangle 4A2 B2 C2
with the same angle sum 180◦ + p◦ and such that m ]A2 ≤ 21 m ]A1 ≤ 14 m ]A.
Continuing in this fashion, one constructs a sequence of triangles 4A1 B1 C1 ,
4A2 B2 C2 , . . . , 4An Bn Cn , each with the same angle measure 180◦ + p◦ as
4ABC, and such that m ]An ≤ 21n m ]A. Clearly, we can choose the natural
number n as large as we wish, in particular, so large that m ]An ≤ p◦ . Since
m ]An + m ]Bn + m ]Cn = 180◦ + p◦ , it follows that m ]Bn + m ]Cn ≥ 180◦ .
This, however, contradicts the second Lemma and concludes the proof of the
theorem.

One immediate consequence of the Saccheri-Legendre Theorem is the follow-

ing corollary:

Corollary. The sum of the measures of two angles of a triangle is less than or
equal to the measure of their remote exterior angle.

Proof. In triangle 4ABC, m ]A + m ]B + m ]C ≤ 180◦ by Theorem 149.

Hence, we have: m ]A + m ]B ≤ 180◦ − m ]C. The right-hand side of this
inequality is, by definition, the measure of the remote exterior angle of angles
]A and ]B of 4ABC.

The Saccheri-Legendre Theorem can be generalized to a convex quadrilat-

eral with the result that the angle sum of a convex quadrilateral does not exceed
360◦ . For the proof, dissect the quadrilateral by one of its diagonals into two
triangles and apply the Saccheri-Legendre Theorem to each of them. Addition
of the corresponding inequalities yields the desired result.
We now introduce the following definition:

Definition 80. The defect of a triangle is equal to 180◦ minus the angle sum
of the triangle. The defect of a triangle ∆ shall be denoted δ(∆).

Clearly, congruent triangles have equal defects, and the Saccheri-Legendre

Theorem asserts that the defect of a triangle is a non-negative number (posi-
tivity of the defect). In Euclidean Geometry there are no “defective” triangles
(Theorem 71), and our main objective is to show that if one defective triangle
exists, then all triangles are defective. Equivalently, the contrapositive version
of the statement we want to prove would assert that if one triangle has an angle
sum equal to 180◦ , then so do all other triangles. Bear in mind that we are
NOT proving that one such triangle exists nor do we assert the contrary: we are
merely investigating the consequences of the hypothesis that one such triangle
might exist.

Theorem 150 (Additivity of the Defect). Let triangle 4ABC be an arbitrary

triangle and let D be a point on, say, side AB. Then δ(4ABC) = δ(4ACD) +
δ(4BCD).
13.1. ON NEUTRAL GEOMETRY III 253

−−→
Proof. Since ray CD is in the interior of ]ACB (Theorem 9), m ]ACB =
m ]ACD + m ]DCB (Angle Addition Axiom). Since ]ADC and ]BDC
are supplementary, δ(4ACD) = 180◦ − m ]A − m ]ADC − m ]ACD, and
δ(4BDC) = 180◦ − m ]B − m ]BDC − m ]DCB. Adding these two equations
yields:

δ(4ACD) + δ(4BDC) = 360◦ − m ]A − m ]B − 180◦ − m ]C = δ(4ABC)

Corollary. The angle sum of 4ABC is equal to 180◦ if and only if the angle
sums of each 4ACD and 4BCD are equal to 180◦ .

Proof. If δ(4ACD) = δ(4BCD) = 0, then δ(4ABC) = 0 by the preced-

ing theorem. Conversely, if δ(4ABC) = 0, then δ(4ACD) + δ(4BCD) =
0, and since the defect is a non-negative number, it follows that necessarily
δ(4ACD) = δ(4BCD) = 0.

Thus, the defect function δ which assigns a non-negative number δ(∆) to

each triangle ∆ has the characteristic properties (positivity, invariance under
congruence, and additivity) of an area function (see Chapter 7). Thus, if we
can prove δ(∆) > 0, we will have demoted our earlier Axiom 16 which guarantees
the existence of an area measure to the rank of a theorem without affecting our
theory of area measurement since the defect function provides, in fact, one area
measure with the requisite properties. Needless to say, if one adopts such an
approach to area measurements, our earlier Axiom 17 should still be retained
as an independent axiom in order to “fix” the area unit and will have to be
renumbered as Axiom 16.
Next, recall that a rectangle is defined as a quadrilateral whose four angles
are right angles. Hence, the angle sum of a rectangle is 360◦ . Of course, we
don’t know whether rectangles exist at all in Neutral Geometry, but the following
theorem sheds some light on this question:

Theorem 151. If a triangle exists whose angle sum is 180◦ , then a rectangle
exists. If a rectangle exists, then every triangle has angle sum 180◦ .

Proof. We prove the theorem in six steps:

254 CHAPTER 13. INTRODUCTION TO NON-EUCLIDEAN GEOMETRY

Step 1. We construct a right triangle having angle sum 180◦ as follows:

Let 4ABC be the given triangle with zero defect. If 4ABC is a right
triangle, we are done. So, assume that 4ABC is not a right triangle. By
the second Lemma, at least two angles of this triangle, say, angles ]A
and ]B, must be acute (why?). Let CD be the altitude from vertex C
(Theorem 33 and 34). The foot D of this altitude is between vertices A
and B. For, assume the contrary, that is, assume that, say, vertex A lies
between D and B. Then, in triangle 4CDA, the measure of the exterior
angle ]CAB would be less than the measure (90◦ ) of the remote interior
angle ]CDA, contradicting the Exterior Angle Theorem (Theorem 39).
In a similar fashion, one excludes the possibility that B lies between A
and D. Hence, D lies between vertices A and B. It follows then from
the Corollary to Theorem 147 that each of the right triangles 4ADC and
4ADB has zero defect.
Step 2. From a right triangle with zero defect we construct a rectangle as
follows:

Let 4CDB be a right triangle with zero defect and right angle at vertex
D. By Axiom 11 (Angle Construction Axiom), there exists a unique ray
−−→ ←→
CX on the opposite side of line CB from vertex D such that m ]DBC =
m ]BCX, and by the Point Plotting Theorem (Theorem 3), there exists
−−→
a unique point E on ray CX such that CE = BD. Then, 4CDB ∼ =
4BEC (S.A.S.) and hence, 4BCE is also a right triangle with zero defect
and right angle at vertex E. Furthermore, by hypothesis, m ]DBC +
m ]DCB = 90◦ so that, by substitution: m ]ECB + m ]BCD = 90◦
and m ]DBC + m ]EBC = 90◦ . Then, the Alternate Interior Angle
←→ ←→
Theorem (Theorem 62) implies that lines CE and DB are parallel and so
←→ ←→
are lines CD and BE. Therefore, points D and B are on the same side of
←→
line CE so that point B is in the interior of angle ]ECD, and similarly,
point C is in the interior of angle ]DBE. Hence, by Axiom 12 (Angle
Addition Axiom), we conclude that m ]ECD = m ]EBD = 90◦ and the
quadrilateral CDBE is a rectangle.
Step 3. From one rectangle we construct arbitrarily large rectangles:
13.1. ON NEUTRAL GEOMETRY III 255

Suppose a rectangle ABCD exists, and XY is any given segment. We

first show that there exists a rectangle with one side longer than XY . To
that end, we use ABCD as a “building block.” Construct a quadrilateral
D2 C2 CD congruent to ABCD so that corresponding sides C2 D2 and BA
←→
are on opposite sides of line CD. A way of achieving this is to extend
BC through C by its own length to point C2 and similarly to extend AD
through D by its own length to D2 , using the Point Plotting Theorem.
Then, D2 C2 CD is a rectangle (why?). Moreover, the sets of points B,
C, C2 and A, D, D2 are each collinear since there is a unique perpen-
dicular to CD at C and D, respectively. Thus, ABC2 D2 is a rectangle
with AD2 = 2AD. In like manner, we construct a sequence of rectangles
ABC3 D3 , . . . , ABCn Dn with AD3 = 3AD, . . . , ADn = nAD. We then
choose n so large that nAD > XY and rectangle ABCn Dn satisfies the
stated requirement. Next, suppose a rectangle ABCD exists and XY ,
ZW are given segments. By applying the preceding method, one shows
that one can construct a rectangle with adjacent sides larger than XY
and ZW , respectively.

Step 4. If one rectangle exists, then there exists a rectangle with two adjacent
sides congruent to pre-assigned segments XY and ZW :

First we construct a rectangle P QRS such that P Q > XY and P S > ZW

by the procedure of Step 3. Next, we cut down this rectangle to fit our
−−→
requirements as follows: There is a point Q0 on ray P Q such that P Q0 =
XY and Q0 is between P and Q (Point Plotting Theorem). From Q0 drop
←
→
the perpendicular to line RS with foot R0 . Quadrilateral P Q0 R0 S is a
rectangle since angles ]P , ]S, and ]R0 are right angles. Moreover, angle
]P Q0 R0 must be a right angle too, because if m ]P Q0 R0 > 90◦ , then the
angle sum of quadrilateral P Q0 R0 S 0 would exceed 360◦ in contradiction
with the remark following the proof of the Saccheri-Legendre Theorem,
256 CHAPTER 13. INTRODUCTION TO NON-EUCLIDEAN GEOMETRY

and if m ]P Q0 R0 < 90◦ , then m ]QQ0 R0 > 90◦ and the angle sum of
quadrilateral QQ0 R0 R would exceed 360◦ which is again impossible. Thus,
m ]P Q0 R0 = 90◦ and P Q0 R0 S is indeed a rectangle. In the same way,
−→
there is a point S 0 on ray P S such that S 0 is between P and S and
P S = ZW . Drop the perpendicular from S 0 onto line Q0 R0 with foot R00 .
0

Then the preceding argument can be used to show that the quadrilateral
P Q0 R00 S 0 is the rectangle desired.
Step 5. We now prove that if a rectangle exists, then all right triangles have
zero defect:

We first show that every right triangle is congruent to a second right

triangle obtained through splitting a rectangle by a diagonal. We then
show that the second triangle has angle sum 180◦ .
Let 4ABC be a right triangle with right angle at vertex B. By Step
4, there exists a rectangle A0 B 0 C 0 D0 with A0 B 0 = AB and B 0 C 0 = BC.
Then, 4ABC ∼ = 4A0 B 0 C 0 (S.A.S.) and 4ABC and 4A0 B 0 C 0 have the
same angle sum. Let p◦ and q ◦ be the angle sums of 4A0 B 0 C 0 and
4A0 D0 C 0 , respectively. Now, p◦ +q ◦ = 360◦ , and by the Saccheri-Legendre
Theorem, p◦ ≤ 180◦ . Suppose p◦ < 180◦ : then q ◦ > 180◦ in contradiction
with Theorem 149. Hence, p◦ = 180◦ , i.e. δ(4ABC) = 0.
Step 6. If one rectangle exists, then every triangle has an angle sum of 180◦ :
This is easily proved by applying the procedure used in Step 1 to de-
compose an arbitrary triangle into two right triangles (each of which has
zero defect) via an altitude and use the Corollary to Theorem 150. This
completes the proof of Theorem 151.

Thus, if one triangle exists with angle sum 180◦ , then every triangle has
angle sum 180◦ . We may also prove the following corollary:
Corollary. If one triangle has an angle sum which is less than 180◦ , then the
angle sum of every triangle is less than 180◦ .
Proof. Suppose the angle sum of triangle 4ABC is less than 180◦ , and let
4P QR be an arbitrary triangle. By the Saccheri-Legendre Theorem, the angle
sum of 4P QR cannot exceed 180◦ . If the angle sum of triangle 4P QR equals
180◦ , then by Theorem 151, the angle sum of triangle 4ABC must also equal
180◦ , in contradiction with our hypothesis. Therefore, the angle sum of triangle
of 4P QR must be less than 180◦ , which was to be proved.
13.1. ON NEUTRAL GEOMETRY III 257

At this point, we observe an important feature not implicit to the Saccheri-

Legendre Theorem: a Neutral Geometry is “uniform” in the sense that either
all of its triangles have angle sums equal to 180◦ or else they all have angle sums
less than 180◦ . The first type of Neutral Geometry corresponds evidently to Eu-
clidean Geometry, whereas the second possibility leads to Hyperbolic Geometry
as we shall see soon.
In order to prepare the ground for the development of Hyperbolic Geometry,
we still need one further tool from Neutral Geometry:

Definition 81. Convex quadrilateral ABCD is a Saccheri quadrilateral if m ]B =

m ]C = 90◦ and AB = CD. Side BC is the base, sides AB and CD are the
legs, and side AD is the summit, and angles ]A and ]D are the summit angles
of the Saccheri quadrilateral.

Theorem 152. The summit angles of a Saccheri quadrilateral are congruent.

Proof. Let JOEL be a Saccheri quadrilateral with base LE. Draw JE and
LO. Then, 4LJE ∼ = 4LOE (S.A.S.). Hence, JE = LO and 4JLO ∼= 4JOE
(S.S.S.). Therefore, ]LJO ∼
= ]JOE as corresponding parts.

Consequently, either the summit angles of a Saccheri quadrilateral are both

acute, or obtuse, or right angles. Clearly, the latter case holds true in Euclidean
Geometry. Also, the summit angles cannot be obtuse because this would lead to
a quadrilateral with angle sum in excess of 360◦ . We shall see that in Hyperbolic
Geometry, the summit angles of a Saccheri quadrilateral must be acute.

Theorem 153. If the base angles of a quadrilateral are right angles and its legs
have unequal lengths, then the measures of the summit angles of the quadrilat-
eral are unequal and the larger angle is opposite the longer side.
258 CHAPTER 13. INTRODUCTION TO NON-EUCLIDEAN GEOMETRY

Proof. In quadrilateral HOSA, let OH < SA and m ]O = m ]S = 90◦ . The

−→
Point Plotting Theorem allows us to choose point E on ray SA between S and A
such that SE = OH. Then HOSE is a Saccheri quadrilateral and by Theorem
152, m ]OHE = m ]SEH. Since point E is in the interior of angle ]OHA
(why?), the Angle Addition Axiom (Axiom 12) implies: m ]OHA = m ]OHE+
m ]EHA, so that m ]OHA > m ]OHE, and hence, m ]OHA > m ]SEH.
Finally, the Exterior Angle Theorem (Theorem 39) applied to triangle 4HEA
implies that m ]SEH > m ]A, that is, m ]OHA > m ]A, which was to be
proved.

The converse of the preceding theorem is also valid:

Theorem 154. If the base angles of a quadrilateral are right angles and the
measures of its summit angles are unequal, then the legs of the quadrilateral
have unequal lengths and the longer leg is opposite the larger angle.

Proof. Consider the quadrilateral RU T H with m ]H = m ]T = 90◦ and,

say, m ]R > m ]U . Then either U T > RH, in which case we are done, or
U T ≤ RH. If U T < RH, then Theorem 153 implies m ]R < m ]U which con-
tradicts the hypothesis. If U T = RH, then RU T H is a Saccheri quadrilateral
and by Theorem 152, m ]R = m ]U which is again in contradiction with our
hypothesis. Therefore, U T > RH.
13.2. THE HYPERBOLIC AXIOM 259

13.2 The Hyperbolic Axiom

We are now prepared for the study of the fundamentals of Hyperbolic Geometry
which we have already defined as the Geometry that is obtained when Neutral
Geometry is augmented by the logical negation of the Euclidean Parallel Axiom:
H-Axiom 1 (Hyperbolic Parallel Axiom). There exists a line l and a point P
not on l such that at least two distinct lines parallel to l pass through P .

The first hyperbolic theorems are immediate consequences of H-Axiom 1:

H-Theorem 1. There exists a line that is wholly contained in the interior of
some angle.

Proof. By H-Axiom 1, there exist a line l and a point P not on l such that (at
least) two parallels m and n to l pass through P . Lines m and n separate the
plane into four disjoint regions, namely the interiors of angles ]AP B, ]AP B 0 ,
]BP A0 , and ]A0 P B 0 , where P is between point A and A0 on m and between
points B and B 0 on n. Let Q be any point on l. Since l does not intersect m
and n, Q cannot be incident with m or n. Thus Q is in the interior of one of the
−−→
four angles cited, say, angle ]BP A0 . Since l does not intersect the legs P B and
−−→0
P A of this angle and one of its points is in the interior of the angle, it follows
that l is entirely contained in the interior of angle ]BP A0 .
H-Corollary. If H-Axiom 1 applies to line l and point P , then there are in-
finitely many lines parallel to l through P .
Proof. Using the diagram of H-Theorem 1, let R be any point in the interior
←→
of angle ]AP B. Then, line P R (excluding point P ) is wholly contained in the
interiors of angles ]AP B and ]A0 P B 0 and cannot intersect l which is entirely
contained in the interior of angle ]A0 P B. Since P is not incident with l by
260 CHAPTER 13. INTRODUCTION TO NON-EUCLIDEAN GEOMETRY

←→
hypothesis, it follows that line P R is parallel to l. Since point R was arbitrary,
the H-Corollary is proved.

H-Theorem 1 is a “positional” (i.e. non-metrical) property of the hyperbolic

plane as contrasted to the Euclidean plane where only part of a line can be
contained in the interior of an angle. In the Euclidean plane, a line through an
interior point of an angle must meet at least one of the legs of the angle.
We proceed now with the proof of the H-theorem which guarantees the
existence of a triangle with angle sum less than 180◦ . To this end, we still need
a lemma from Neutral Geometry:

Lemma. Let l be a line and P a point not incident with l and Q a point on l.
←→
Then, on each side of line P Q and for any positive real number , there exists
a point R on l such that m ]P RQ < .

Proof. Let  > 0 be any positive real number, no matter how small. Choose
←→
one side, say, the right-hand side of line P Q and let R1 be a point on l on
←→
the chosen side of P Q such that QR1 = P Q (Point Plotting Theorem). Then
triangle 4P QR1 is isosceles and m ]QP R1 = m ]QR1 P = b1 . Then, in tri-
angle 4P QR1 , by the Corollary to the Saccheri-Legendre Theorem, we have:
m ]QP R1 + m ]QR1 P = 2b1 ≤ m ]P QS. Thus, b1 ≤ 21 m ]P QS. Next,
construct the isosceles triangle 4P R1 R2 with R1 R2 = P R1 by the same pro-
cedure as for 4P QR1 , and conclude similarly that if b2 = m ]R1 P R2 =
m ]P R2 R1 = m ]P R2 Q, then b2 ≤ 21 b1 ≤ 14 m ]P QS. By repeating this
←→
process n times, we obtain a point Rn on l on the chosen side of line P Q such
that bn = m ]P Rn Q ≤ 21n m ]P QS. We may now choose n so large that
1
2n m ]P QS < , and the proof of the Lemma is complete.

H-Theorem 2. There exists a triangle whose angle sum is less than 180◦ .
13.2. THE HYPERBOLIC AXIOM 261

Proof. Let l be a line and P a point not on l satisfying H-Axiom 1. We construct

a line m parallel to l through P in the usual manner: we drop the perpendicular
P Q from P to l and construct the parallel m to l through P as the perpendicular
to P Q at P . By H-Axiom 1, there is (at least) another line n through P parallel
←→
to l. One of the angles that line n makes with line P Q must be acute (why?).
Choose point X on n such that m ]QP X < 90◦ . Let Y be a point on m on the
←→
same side of P Q as point X. Then, if m ]XP Y = a◦ , m ]QP X = 90◦ −a◦ . Now
←→
let R be a point on l on the same side of P Q as point X such that m ]P RQ < a◦
(the third Lemma guarantees the existence of such a point R). Then, in triangle
4P QR, m ]Q = 90◦ , m ]P RQ < a◦ , and m ]RP Q < m ]XP Q = 90◦ −a◦ . By
addition, one obtains: m ]Q+m ]P RQ+m ]RP Q < 90◦ +a◦ +90◦ −a◦ = 180◦ .
Thus, 4P QR has angle sum less than 180◦ and the proof is complete.
−→
In the above proof, one may object that ray P R may not lie in the interior
−→ −−→
of angle ]QP X. To take care of this, consider that rays P R and P X are
←→
distinct (H-Axiom 1) and both lie on the same side of line P Q by construction.
←→
Consequently, one of them has to lie in the interior of the angle formed by P Q
−−→
and the other (why?). Suppose ray P X fell in the interior of angle ]QP R.
−−→
Then ray P X would meet segment QR (Crossbar Theorem) and hence, line l,
−→
thus contradicting H-Axiom 1. Therefore, ray P R must lie in the interior of
angle ]QP X.
We may now enunciate the next H-theorem:

H-Theorem 3. The angle sum of every triangle is less than 180◦ .

Proof. By H-Theorem 2, there exists a triangle with angle sum less than 180◦ .
By the Corollary to Theorem 151, it then follows that the angle sum of every
triangle is less than 180◦ .

In contrast with Euclidean Geometry, it can be shown2 with the help of some
elementary tools from Calculus, that there exists an “H-triangle” whose angle
sum is any number between 0◦ and 180◦ .

H-Corollary. The angle sum of every (convex) quadrilateral is less than 360◦ .

H-Corollary. There are no rectangles in Hyperbolic Geometry.

2 See, for instance, E. Moise, op. cit., p. 324 and the proof of the continuity of the defect

function.
262 CHAPTER 13. INTRODUCTION TO NON-EUCLIDEAN GEOMETRY

The proofs of these two H-corollaries are straightforward and left as exercises.
Using H-Theorem 2, we can “universalize” H-Axiom 1. The Parallel Axiom
of Euclidean Geometry (Axiom 15) asserts that for every line and every point
not on that line, the uniqueness of parallel lines holds. Its logical negation,
the H-Axiom 1, states that for some line and some point not on that line,
uniqueness of parallels fails to hold. Thus, it is legitimate to ask whether the
H-Axiom holds for some line(s) and some point(s) not on those lines but fails for
other lines and other points not incident with those lines. In the next theorem,
we show that this definitely embarrassing circumstance cannot occur. In fact,
Hyperbolic Geometry acquires a logically consistent and non-trivial structure
on the same footing as Euclidean Geometry through the following theorem:

H-Theorem 4 (Universal Hyperbolic Theorem). If H-Axiom 1 holds for some

line l and some point P not on l, then it holds for every line and every point
not incident with that line.

Proof. Let l be any line and P any point not on l. Construct the by-now
familiar line m parallel to l through P as the perpendicular through P to the
←→
perpendicular line P Q from P to l. Let R be any other point of l different
from Q. Erect the perpendicular t to l through R (Theorem 31) and drop the
←→
perpendicular P S from P to t (Theorem 33 and 34). Now, line P S is parallel
←→ ←→
to l since both l and P S are perpendicular to t. We claim that m and P S are
distinct lines. Assume, on the contrary, that S lies on m. Then the quadrilateral
P QRS is a rectangle. But this contradicts the second corollary to H-Theorem
3, and establishes the validity of H-Axiom 1 for the arbitrary line l and point P
not on l, provided that the hyperbolic parallel postulate holds for one particular
line and one particular point not on that line.

As a consequence of H-Theorem 4 and the corollary to H-Theorem 1, one

may assert that in Hyperbolic Geometry, for any line and any point not lying
on that line there exist infinitely many lines parallel to the given line through
the given point.
We now prove that similar triangles cannot exist in Hyperbolic Geometry,
except for the trivial case of congruent triangles:

H-Theorem 5. Two triangles are congruent if their corresponding angles are

congruent.
13.2. THE HYPERBOLIC AXIOM 263

In other words, A.A.A. is a valid H-criterion for the congruence of triangles

in Hyperbolic Geometry.

Proof. Assume on the contrary that there exist triangles 4ABC and 4A0 B 0 C 0
which are similar but not congruent: ]A ∼ = ]A0 , ]B ∼ = ]B 0 , ]C ∼ = ]C 0
0 0 0 0 0 0
but AB 6= A B , AC 6= A C and BC 6= B C . Now consider the triples
{AB, AC, BC} and {A0 B 0 , A0 C 0 , B 0 C 0 } of sides of these two triangles. One of
these two triples must contain two segments whose lengths are greater than
those of the two corresponding segments of the other triple. Without loss of
generality, we may assume that AB > A0 B 0 and AC > A0 C 0 . Then there exist
points B 00 on AB and C 00 on AC such that AB 00 = A0 B 0 and AC 00 = A0 C 0 .
Consequently, 4AB 00 C 00 ∼
= 4A0 B 0 C 0 , so that ]AB 00 C 00 ∼= ]B 0 ∼= ]B and also
00 00 ∼ 0 ∼ ←→
]AC B = ]C = ]C, by hypothesis. This implies that BC is parallel to
←− −→
B 00 C 00 by Theorem 64 (congruence of corresponding angles of two lines cut by
a transversal). Therefore, the quadrilateral BB 00 C 00 C is convex (why?) and
m ]B + m ]BB 00 C 00 = m ]C + m ]CC 00 B 00 = 180◦ . Consequently, the angle
sum of the quadrilateral BB 00 C 00 C equals 360◦ . This, however, contradicts the
already cited consequence of the Saccheri-Legendre Theorem and completes the
proof of the theorem.

To sum up, in Hyperbolic Geometry it is impossible to “magnify” or “shrink”

a geometric shape, e.g. a triangle, without distortion. In other words, in a
hyperbolic world photography would be inherently surrealistic. A remarkable
consequence of H-Theorem 5 is that in Hyperbolic Geometry, the length of a
segment can be determined by means of the measure of an angle: for instance,
an angle of an equilateral triangle determines the length of the triangle side
uniquely. This state of affairs is expressed in lapidary terms by declaring that
“Hyperbolic Geometry has an absolute unit of length,” but this is another story
whose narrative would lead us beyond the scope of this modest introduction to
the fundamentals of Hyperbolic Geometry.
After these first hyperbolic theorems, we resume our investigation of the
structure of Hyperbolic Geometry with a brief comparison of the Euclidean
concept of parallelism with the Hyperbolic one. In Euclidean Geometry, an
important characteristic of parallel lines—which has misled several attempts to
prove the dependence of the Fifth Postulate on those of Neutral Geometry—is
that they are equidistant everywhere. More precisely, given two Euclidean
parallel lines l and l0 and points A, B, C, . . . on l, one may drop the (unique)
264 CHAPTER 13. INTRODUCTION TO NON-EUCLIDEAN GEOMETRY

perpendiculars AA0 , BB 0 , CC 0 , . . . from these points onto l0 , and the Euclidean

Parallel Axiom will guarantee that AA0 ∼ = BB 0 ∼
= CC 0 ∼= ...
The following H-theorem highlights another fundamental difference between
Euclidean and Hyperbolic Geometry:

H-Theorem 6. In Hyperbolic Geometry, if l and l0 are any distinct parallel

lines, then any set of points on l equidistant from l0 contains at most two points
in it.

Proof. Assume that there is a set of three points A, B, C on l equidistant from

l0 . Let A0 , B 0 , C 0 be the feet of the perpendiculars from A, B, C onto l0 . Then,
the quadrilaterals A0 B 0 BA, A0 C 0 CA and B 0 C 0 CB are Saccheri quadrilaterals.
Theorem 152 asserts that the summit angles of a Saccheri quadrilateral are con-
gruent, thus: ]A0 AB ∼ = ]B 0 BA, ]A0 AC ∼ = ]C 0 CA and ]B 0 BC ∼ = ]C 0 CB.
Then transitivity and Theorem 19 imply that the supplementary angles ]B 0 BA
and ]B 0 BC are congruent and, hence, are right angles. Therefore, these Sac-
cheri quadrilaterals are all rectangles. But rectangles do not exist in Hyperbolic
Geometry (H-Corollary 2 to H-Theorem 3). This contradiction shows that the
set of three points A, B, C cannot be equidistant from l0 .

The previous H-theorem implies that at most two points at a time on l can
be equidistant from l0 . It does not forbid the possibility that pairs of points
(A, B), (C, D), . . . exist on l such that each pair is equidistant from l0 . For
instance, the pairs of points (A, B) and (C, D) on l are equidistant from l0 but
AA0 is not congruent to CC 0 . It is important to realize that H-Theorem 6 allows
for another possibility, namely that there is no pair of points on l equidistant
from l0 .
13.2. THE HYPERBOLIC AXIOM 265

A diagram for this possibility might look as shown: the points on l are
at different distances from the parallel l0 , i.e. l “moves away” from l0 in one
direction and “approaches” l’ in the other direction without meeting it. These
preliminary considerations show that in Hyperbolic Geometry, different pairs of
parallel lines behave quite differently: given two hyperbolic parallel lines, either
there exist two points on one line equidistant from the other line, or no two
points on one line are equidistant from the other line. We now examine these
two cases in some detail.
H-Theorem 7. In Hyperbolic Geometry, consider a pair of points A and B
equidistant from line l0 and on the same side of l0 . If l is the line determined by
A and B, then l and l0 have a common perpendicular segment whose length is
the shortest distance between l and l0 .

Proof. Let points A and B on l be equidistant from l0 . Then, the quadrilateral

A0 B 0 BA is a Saccheri quadrilateral. Let M be the midpoint of AB and M 0 the
midpoint of A0 B 0 . We know that the summit angles of a Saccheri quadrilateral
are congruent (Theorem 149), so that ]A0 AB ∼ = ]B 0 BA. Hence, 4A0 AM ∼ =
0 0 ∼ 0
4B BM (S.A.S.) and A M = B M as corresponding parts. It follows then that
4A0 M 0 M ∼ = 4B 0 M 0 M (S.S.S.) and, consequently, ]A0 M 0 M ∼ = ]B 0 M 0 M as
0 0 0 0
corresponding parts. Since ]A M M and ]B M M are supplementary angles,
they must be right angles, so that M M 0 is perpendicular to the base A0 B 0 .
But we also have ]A0 M M 0 ∼ = ]B 0 M M 0 and ]A0 M A ∼ = ]B 0 M B, so that by
0 ∼ 0
the Angle Addition Axiom, ]AM M = ]BM M . This, in turn, implies that
these latter supplementary angles must be right angles. Therefore, M M 0 is
also perpendicular to the summit AB of the Saccheri quadrilateral ABB 0 A0 .
Next, consider the quadrilateral A0 M 0 M A. It has three right angles and since
in Hyperbolic Geometry no rectangles exist, m ]M AA0 < 90◦ . Theorem 154
implies that AA0 > M M 0 , i.e. M M 0 is shorter than AA0 . The proof of the
theorem is now completed by observing that

- if A and A0 are points on l and l0 , respectively, and AA0 is not perpendic-

ular to l0 , then clearly, AA0 > M M 0 , and
- if AA0 is perpendicular to l0 , then again Theorem 154 asserts that AA0 >
M M 0.

H-Corollary. In Hyperbolic Geometry, if l and l0 are parallel lines for which

there exist a pair of points A and B on l equidistant from l0 , then l and l0 have
a common perpendicular segment that is the shortest distance between l and l0 .
266 CHAPTER 13. INTRODUCTION TO NON-EUCLIDEAN GEOMETRY

This H-Corollary suggests that in Hyperbolic Geometry, the “perpendicular

at closest approach” between two parallel lines is an “axis of symmetry” for
these lines. This is formalized in the following theorem:

H-Theorem 8. In Hyperbolic Geometry, if lines l and l0 have a common per-

pendicular segment M M 0 , then they are parallel and M M 0 is unique. Further-
more, if A and B are points on l such that M is the midpoint of AB, then A
and B are equidistant from l0 .

Proof. The Alternate Interior Angle Theorem (Theorem 62) guarantees that l
and l0 are parallel. Now, if l and l0 had another common perpendicular seg-
ment N N 0 other than M M 0 , then quadrilateral M 0 N 0 N M would be a rectangle
which cannot exist in Hyperbolic Geometry. Next, let M be the midpoint of
segment AB on l. Drop the perpendiculars AA0 , BB 0 onto l0 . Then, 4AM 0 M ∼ =
4BM 0 M (S.A.S.), and hence, AM 0 ∼ = BM 0 and ]AM 0 M ∼ = ]BM 0 M . It fol-
lows that ]A0 M 0 A ∼ = ]B 0 M 0 B (Angle Addition Axiom) and, consequently,
0 0 ∼
4AA M = 4BB M 0 (A.A.S.). Therefore, AA0 ∼
0
= BB 0 as corresponding
parts.

It follows from the preceding H-theorems, that in Hyperbolic Geometry,

if two parallel lines have a (unique) common perpendicular, they “diverge”
indefinitely on both sides of this common perpendicular.

In order to establish the existence of the second species of H-parallel lines

which do not have a common perpendicular, the following intuitive consideration
is helpful. Start with any line l and a point P not incident with l. Drop
the perpendicular P Q from P onto l and let m be the perpendicular through
P to P Q. Then, m and l have the common perpendicular P Q, are parallel,
←→
and by H-Theorem 8, pairs of points on m situated symmetrically about P Q
are equidistant from l. By the Universal Hyperbolic Theorem (H-Theorem 4),
there exist other lines through P parallel to l, but we cannot say whether any
−→
one of them belongs to the second species of parallel lines. Let P S be one
−→
ray of m, and consider various rays emanating from P and lying between P S
13.2. THE HYPERBOLIC AXIOM 267

−−→ −→
and P Q. Some of these rays, such as P R, will intersect l and others, such
−−→ −−→
as P Y will not. Then, P Y will be parallel to l and ]QP Y will be acute
←→
(why?). Let point R on l be on the same side of P Q as point Y and view
←→
P R as a “generic intersector” of l through P . Now let R recede endlessly on
l from Q. Then, ]QP R will increase endlessly while remaining acute, since
m ]QP R < m ]QP Y . The Completeness Axiom for the field of real numbers
(Rule R14 of Appendix III) then guarantees that m ]QP R will approach a
−−→
certain limiting value and, thus, determine a certain limiting ray, say P X that
−−→
does not intersect l in the following precise sense: any ray between P X and
−−→ −−→ −−→ −−→
P Q intersects l, whereas any other ray P Y such that P X is between P Y and
−−→ ←→ −−→
P Q does not intersect l, so that P Y is a parallel through P to l. The ray P X is
called a left limiting (or asymptotic) parallel ray to l through P . Similarly, there
exists a right limiting (or asymptotic) ray parallel to l through P on the opposite
←→
side of P Q. These heuristic considerations are made precise in the following H-
theorem whose proof requires a continuity argument using the aforementioned
Completeness Axiom—hence, some fundamentals of Calculus—and is omitted
for the sake of brevity3 :
H-Theorem 9. For every line l and every point P not on l, let Q be the foot of
−−→ −−→
the perpendicular from P onto l. Then, there are two unique rays P X and P X 0
←→
on opposite sides of P Q that do not meet l and have the property that a ray
−−→ −−→
emanating from P intersects l if and only if it is between P X and P X 0 . These
limiting rays are symmetrical with respect to P Q, that is ]QP X ∼ = ]QP X 0 .

Either of the congruent angles ]QP X and ]QP X 0 is called the angle of
parallelism at point P with respect to l. Clearly, this angle is always acute,
for if it were right, it would follow that there is a unique parallel through P
to l, thus contradicting the Universal Hyperbolic Theorem (H-Theorem 4). It
can be shown that as P varies, the measure of the angle of parallelism takes on
all possible values between 0◦ and 90◦ . The formula discovered independently
by Bolyai and Lobachevsky for the measure φ of the angle of parallelism at
a point P at distance d from a given line l is mentioned here for the sake
of completeness, but first we need another remarkable result from Hyperbolic
Geometry. It can be proved that the area of a H-triangle is proportional to
π 2
its defect, with proportionality constant ( 180◦ )k where k is a positive constant
which depends on the unit of area measure, i.e. on whichever H-triangle is
3 For a proof, consult, e.g., The Non-Euclidean Revolution, by Richard Trudeau, Birkhäuser

Boston, 1987, Chapter 6

268 CHAPTER 13. INTRODUCTION TO NON-EUCLIDEAN GEOMETRY

chosen to have area equal to one. This surprising result shows that in Hyperbolic
Geometry, there is an upper limit to the possible area a H-triangle can have, even
though there is no upper limit to the lengths of the triangle sides. In fact, since
the angle sum of triangle cannot be negative, it follows from H-Theorem 3 that
the defect of an H-triangle is a positive number that cannot exceed 180◦ . Thus,
the area of an H-triangle is at most equal to πk 2 . Now, the Bolyai-Lobachevsky
d
formula for the measure of the angle of parallelism reads: tan( φ2 ) = e− k where
π ◦
φ is measured in radians, i.e. φ = ( 180 ◦ )φ , k is the already mentioned constant

whose square relates the area of a hyperbolic triangle to its angle defect, and
e ≈ 2.718 . . . is the base of the natural logarithms.

To sum up, it can be proved that in Hyperbolic Geometry, there are exactly
two kinds of parallels to a given line through a point not incident with the given
line l. The first type consists of parallels m such that l and m have a common
perpendicular, and the parallel m diverges from l on both sides of this common
perpendicular. The second kind of parallels consist in lines m that approach l
asymptotically in one direction and diverge away from l in the opposite direction.
Such parallels have no common perpendicular. There are exactly two limiting
parallel rays to l through P and infinitely many lines through P that do not
enter the region between the limiting rays and l and are, hence, parallels of the
first kind to l. Each such line is divergently parallel to l and admits a unique
common perpendicular with l: for one of these lines the common perpendicular
will pass through P , but for all other lines, the common perpendicular will pass
through points other than P .

13.3 On Elliptic Geometry

We have completed our summary introduction to Hyperbolic Geometry and
have, hopefully, shown that a perfectly consistent, albeit counterintuitive, Ge-
ometry can be elaborated from Neutral Geometry augmented with the Hy-
perbolic Parallel Axiom. We may now naturally inquire whether other Non-
Euclidean Geometries might exist. Let us first recall that we have developed
Hyperbolic Geometry by retaining the corpus of standard Neutral Geometry,
that is the axiomatic system determined by our Axioms 1 through 14, and by
adding the negation (H-Axiom 1) of the Euclidean Parallel Axiom (Axiom 15).
Thus, we may assert that standard Neutral Geometry can be either Euclidean
or Hyperbolic. The question which then arises is whether there exists a variant
of Neutral Geometry which is both Non-Euclidean and Non-Hyperbolic. There
may be some kind of “Neutral Geometry” which is “partly Euclidean and partly
Hyperbolic.” The short answer to this question is “yes”: there are many—in
fact, infinitely many—other Non-Euclidean theories of Geometry. The first and
13.3. ON ELLIPTIC GEOMETRY 269

best known of these was proposed by the German mathematician Bernhard

Riemann in 1854. In this theory, Euclid’s Parallel Axiom is replaced by the
following postulate:

Riemann’s Parallel Axiom. There are no parallel lines.

Since the existence of parallel lines is a theorem of standard Neutral Geom-

etry (see Theorem 60), it follows that Riemann’s Axiom is inconsistent with
the axioms of standard Neutral Geometry. A careful analysis of the proof of
Theorem 60 shows that if Riemann’s Parallel Axiom is assumed, then one has
to relinquish either of two fundamental axioms of standard Neutral Geometry:
either one discards the Plane Separation Axiom (Axiom 9), or one has to give up
the axiom which asserts that two points determine a unique line (Axiom 1) and
allow two distinct lines to intersect in two points. Either of these two options
leads to an interesting, rather simple, and perfectly consistent Geometry called
Single Elliptic Geometry and Double Elliptic Geometry, respectively.
At first, Elliptic Geometries may appear rather strange but faithful and
illustrative models using Euclidean concepts can be devised for them. The
following Euclidean model for the plane Double Elliptic Geometry is a famous
example:
Let S be the surface of a Euclidean sphere (in 3-dimensional space). Great
circles are the intersections of S with Euclidean planes through the sphere center.
The elements of the model are defined in Euclidean terms in the roster below:
Double Elliptic Geometry Euclidean Model
Point Point on the surface of S
Line A great circle of S
Plane The surface of S
Segment An arc of a great circle of S
Distance between two points The length of the shortest arc of
great circle joining two points
Angle Spherical angle formed by two
great circles
Measure of angle Measure of spherical angle
In this model, the Plane Separation Axiom is satisfied: every great cir-
cle (“line”) separates the surface of the sphere into two disjoint hemispheres.
Riemann’s Parallel Axiom is also valid, since two distinct great circles meet
in exactly two points. This model, however, should not be construed as an
argument for reducing Riemann’s Double Elliptic Geometry to just standard
Euclidean Geometry on the surface of a sphere (i.e. Euclidean Spherical Ge-
ometry) with new names: in Riemann’s Geometry, we have a new abstract
theory of “straightness” which contradicts Euclid’s. Riemannian lines cannot
be represented faithfully by Euclidean lines in a Euclidean plane, but can be
represented faithfully by great circles on a Euclidean sphere. It can be proved—
and accepted heuristically by examining the preceding model—that in Double
Elliptic Geometry, the angle sum of any triangle is greater than 180◦ , and that
the angle sum of a quadrilateral is greater than 360◦ . The theory of similarity
270 CHAPTER 13. INTRODUCTION TO NON-EUCLIDEAN GEOMETRY

is equally “degenerate” as in Hyperbolic Geometry, that is, an A.A.A.-Theorem

is valid for the congruence of triangles.

We conclude this chapter on the rudiments of Non-Euclidean Geometries

with a few final remarks. Looking back over the geometrical theories we have
briefly examined, we might wonder which one is “correct.” It might be worth-
while mentioning that as early as 1827, Gauss reported on the largest plane
triangle to have been measured up to that time, with vertices on three moun-
tain tops in Germany (one side length was close to 66.50 miles long!). Gauss’
finding was that, within experimental error, the angle sum of the triangle was
180◦ . The question here, however, is not that of the empirical validity of these
various geometries—we have accepted, so far, a hyperbolic geometry for the
four-dimensional spacetime in which we live—but that of logical consistency.
It has been proved by Felix Klein, Henri Poincare, and others that if Euclidean
Geometry is consistent, then no proof or disproof of Euclid’s Parallel Postulate
from standard Neutral Geometry will ever be found, i.e. this postulate is in-
dependent of our Axioms 1 through 14 of Neutral Geometry. A similar result
is valid for the Riemannian Geometries. The truth that we wish to stress is
that all three Geometries we have discussed are equally perfectly valid theories
of “straightness” of lines which do not completely agree on the properties of
straightness. From a purely mathematical point of view, it is manifestly un-
fair to declare a theory incorrect because it is counterintuitive to the notions
we were brought up with. Two hyperbolic parallels that have a common per-
pendicular diverge and are clearly “curved” Euclidean lines, and a Riemannian
line is “obviously” curved, since no Euclidean line does close up on itself. The
real purpose of this chapter was to convince the reader that many “strange”
but nonetheless utterly rigorous and logically consistent theories of space and
its measurement can be conceived and elaborated. We cannot “prove” any one
of them to be “right,” we can only surmise that if mankind is in possession of
any objective truth at all, then there must surely be some element of truth in
mathematics.
13.3. ON ELLIPTIC GEOMETRY 271

Comparison Table for Euclidean And Non-Euclidean Plane Geometries:

Euclidean Hyperbolic Riemannian
Two distinct at most one at most one one/two point(s)
lines intersect
in
Given line l one and only at least two no lines through P
and point P one line lines parallel to l
not on l there
exists
A line is is is not separated
into two
parts by a
point
Parallel lines are equidis- are never do not exist
tant equidistant
If a line inter- must may or may intersect the
sects one of not other line
two parallel
lines, it
Two distinct are parallel are parallel intersect
lines perpen-
dicular to the
same line
The angle equal to less than larger than 180◦
sum of a
triangle is
The area of a independent proportional proportional of its angle
triangle is to the defect to the excess sum
Two triangles similar congruent congruent
with congru-
ent angles are
272 CHAPTER 13. INTRODUCTION TO NON-EUCLIDEAN GEOMETRY

Exercises
Which of the following statements are correct?

1. The negation of Euclid’s Parallel Axiom states that for every line l and
every point P not on l there exist more than one line through P parallel
to l.

2. It is a theorem in Neutral Geometry that if lines l and m meet on a given

side of a transversal t, then the sum of the measures of the interior angles
on that given side of t is less than 180◦ .

3. The Crossbar Theorem implies that a ray emanating from the vertex A of
triangle 4ABC and interior to ]A must intersect the opposite side BC
of the triangle.

4. It is a theorem in Hyperbolic Geometry that for any given segment there

exists a square having this segment as one of its sides.

5. Every Saccheri quadrilateral is a convex quadrilateral.

6. In Hyperbolic Geometry, if 4ABC and 4DEF are equilateral triangles

and ]A ∼
= ]D, then the triangles are congruent.
7. In Hyperbolic Geometry, given a line l and a fixed segment AB, the set
of all points on a given side of l whose distances from l are equal to AB
are on a line parallel to l.

8. In Hyperbolic Geometry any two parallel lines have a common perpendic-

ular.

9. In Hyperbolic Geometry, if three angles of a convex quadrilateral are right

angles, then the fourth angle must be obtuse.

10. In Hyperbolic Geometry, some triangles have angle sums less than 180◦
and some triangles have angle sums equal to 180◦ .

11. In Hyperbolic Geometry, if l and m are parallel lines, then there exist at
least three points on m that are equidistant from l.

12. In Hyperbolic Geometry, if m is any line parallel to line l, then there exist
two points on m that are equidistant from l.

13. In Hyperbolic Geometry, if P is a point not lying on line l, then there are
exactly two lines through P parallel to l.

14. In Hyperbolic Geometry, if P is a point not lying on line l, then there are
exactly two lines through P perpendicular to l.

15. In Hyperbolic Geometry, parallelism of lines is transitive: if l is parallel

to m and m is parallel to n, then l is parallel to n.
13.3. ON ELLIPTIC GEOMETRY 273

16. In Hyperbolic Geometry, if line m contains a limiting parallel ray to line

l, then l and m have a common perpendicular.
17. In Hyperbolic Geometry, if lines l and m have a common perpendicular,
then there is exactly one point on m that is closer to l than any other
point on m.

18. In Hyperbolic Geometry, if line m does not contain a limiting parallel ray
to line l, and m and l have no common perpendicular, then m intersects
l.
19. In Hyperbolic Geometry, the summit angles of a Saccheri quadrilateral
are right angles.

20. Every valid theorem of Neutral Geometry is also valid in Hyperbolic Ge-
ometry.
21. In Hyperbolic Geometry, opposite sides of any parallelogram are congruent
to each other.

22. In Hyperbolic Geometry, there exists an angle and there exists a line that
lies entirely within the interior of that angle.
23. Prove that in Hyperbolic Geometry, the summit angles of Saccheri quadri-
lateral are acute.

24. State a new H-theorem relating the measure of an exterior angle of an

H-triangle to the measures of its remote internal angles.
25. In Hyperbolic Geometry, prove that the summit of a Saccheri quadrilateral
is longer than its base. (Hint: Join the midpoints of the base and the
summit, show that this segment is perpendicular to the summit as well as
to the base, and use Question # 23)
26. In the Double Elliptic (Riemannian) Geometry, the angle sum of any tri-
angle is larger than 180◦ . Prove that in this Geometry, the summit angles
of Saccheri quadrilateral must be obtuse.
27. The Saccheri-Legendre Theorem (Theorem 149) asserts that the angle sum
of any triangle cannot exceed 180◦ . Therefore, this angle sum is for all
triangles either equal to 180◦ or less than 180◦ . The following argument,
which is NOT based on the Euclidean Parallel Axiom, purports to prove
that there exists a triangle whose angle sum is 180◦ :
274 CHAPTER 13. INTRODUCTION TO NON-EUCLIDEAN GEOMETRY

Since the angle sum of any triangle does not exceed 180◦ , let 4ABC be
a triangle whose angle sum is greatest. If there are several such triangles,
choose one at random. Let this greatest angle sum be denoted σ. Thus,
the angle sum of any triangle will be less than or equal to σ. Now, using
the notation of the figure shown, we have: m ]1 + m ]2 + m ]6 ≤ σ and
m ]3 + m ]4 + m ]5 ≤ σ, so that m ]1 + m ]2 + m ]3 + m ]4 + m ]5 +
m ]6 ≤ 2σ. But by assumption, m ]1 + m ]2 + m ]3 + m ]4 = σ, and
moreover, m ]5 + m ]6 = 180◦ . Consequently, σ + 180◦ ≤ 2σ or σ ≥ 180◦ .
Since σ cannot be greater than 180◦ , it follows that σ = 180◦ .
Can you detect the flaw in this “proof”?
Chapter 14

On Regular Polyhedra

Interest in polyhedra (plural of polyhedron) runs through the whole spectrum

of human intellectual activity, from the toddler who plays with cubes through
architects who design giant tessellated domes to mathematicians who study the
subtleties of polytopes, the generalization of polyhedra to spaces with higher
than three dimensions. Regular and semi-regular polyhedra are observed in
Nature: in crystals, in viruses, in hexagonal honeycombs made by bees, in
antique man-made pyramids, and so on. The theory of polyhedra has deep
connections with, and implications for, almost every branch of Mathematics:
their study has led to profound and far reaching developments in Combinatorial
Geometry, Group Theory, Topology, to cite but a few examples. This chapter
should be construed as an invitation to visit a few venerable monuments of the
theory of polyhedra, in particular, the regular, convex polyhedra to be defined
below.
In a vague but intuitive way, a polyhedron is a two-dimensional surface in
three-dimensional space determined by the intersection of four or more planes.
The plane sections of this surface are polygons called the faces of the polyhe-
dron, their common sides are its edges, and the points where edges meet are
its vertices. A uniform polyhedron has the same arrangement of polygons at
each vertex, and a regular polyhedron is a uniform polyhedron with congruent
regular polygons for all its faces.
The simplest regular polygon is the equilateral triangle and the simplest
regular polyhedron is the tetrahedron whose faces are four congruent equilateral
triangles. Thus, in the regular tetrahedron, each vertex is surrounded by three
congruent equilateral triangles. It is instructive to slice a tetrahedron along
edges so as to lay it out flat on a plane: the sum of the measures of the face
angles which meet at each vertex is then 180◦ . If a fourth equilateral triangle
is introduced at each vertex, this angle sum increases to 240◦ , and one obtains
a vertex of the regular octahedron (eight faces). A fifth equilateral triangle
would lead to a face angle sum of 300◦ at each vertex, and the corresponding
polyhedron is a regular icosahedron (twenty faces). A sixth triangle gives a face
angle sum of 360◦ and one can see that no polyhedral vertex arises: everything

275
276 CHAPTER 14. ON REGULAR POLYHEDRA

stays flat. The next regular polygon is the square. Since the minimum number
of faces that can meet at a polyhedron vertex is three, then three congruent
square faces at each vertex yield a face angle sum of 270◦ which corresponds
to the cube. Adding a fourth square brings this angle sum to 360◦ and again
no polyhedron is obtained. With regular pentagons, the minimum number of
three faces meeting at each vertex yields a face angle sum of 324◦ (why?),
corresponding to the regular dodecahedron (twelve faces) and four pentagonal
faces would make the face angle sum at a vertex to exceed 360◦ . Finally, with
regular hexagons, the required minimum number of three faces meeting at each
vertex is already too many (why?). Hence, no regular polyhedron exists with
only regular hexagons for faces and, a fortiori, the same holds true when the faces
are congruent polygons with more than six sides. These heuristic considerations
show that the number of regular (convex) polyhedra in 3-space is limited to
apparently five, in sharp contradistinction to the plane where regular polygons
of an arbitrary number of sides exist.
The regular polyhedra were known as the five “Platonic” solids, and there
is a tradition which assigns their knowledge to the Pythagoreans (5th century
B.C.). The reference to Plato (427–347 B.C.) stems from the dialogue Timaeus,
where Plato incorporates his knowledge of the five regular polyhedra in his philo-
sophical system. It should be mentioned, however, that man-made dodecahedral
objects have been unearthed which date back to well before Pythagorean times.
In addition to the five Platonic solids, Archimedes (ca. 287–212 B.C.) is said to
have described in a manuscript that is now lost, what we today call the 13 semi-
regular or Archimedian solids. A semi-regular polyhedron is one all of whose
faces are regular (convex) polygons, though not necessarily of the same type,
which meet in the same order at all vertices of the polyhedron (i.e. the polyhe-
dron vertices are “alike”). By some miracle, Pappus (who flourished in the 4th
century A.D.) gave an account of the lost book of Archimedes and described
explicitly these 13 semi-regular solids in terms of their various polygonal faces
and the number of edges meeting at a vertex. We shall examine these solids
briefly later on. Polyhedra were well-known to Euclid, and in Book XIII of his
Elements, he shows how to construct the five Platonic solids and even claims
that there are no others.
The study of polyhedra went into sleep for some 15 centuries until the Renais-
sance when the renewed interest in Plato resuscitated them: Albrecht Duerer,
Leonardo da Vinci, Simon Stevin, Luca Pacioli, and many other artists, archi-
tects, or scholars discovered and rediscovered the regular and semi-regular solids,
as well as other polyhedral forms. Johannes Kepler (1571–1630) was the first
to prove in his Harmonices Mundi that the species of semi-regular polyhedra
consists of the 13 Archimedian solids plus the two infinite families of so-called
semi-regular prisms and anti-prisms: in a semi-regular prism, the bases
are regular n-gons and the lateral faces are squares, whereas in a semi-regular
anti-prism, the bases are regular n-gons while the lateral faces are 2n equilat-
eral triangles. Since there are infinitely many regular n-gons, these two families
contain an infinite number of members.
 277

Now, in order to develop the theory of regular polyhedra on a firm math-

ematical basis, we need a precise definition for a polyhedron which would be
restrictive enough to bar the occurrence of a number of unwanted “pathologi-
cal” polyhedra. The following definition is an adaptation to our needs of the
currently accepted definition of a polyhedron:
Definition 82. A polyhedron P is a family of a finite number of plane convex
polygons called faces, each with a finite number of sides called edges, with the
following properties:

(i) Each edge of one face is an edge of exactly one other face
(ii) The family of polygons is connected, that is, for any two edges E and E 0
of P , there exists a chain E = E0 , F1 , E1 , F2 , E3 , . . . , Fn , En = E 0 of edges
Ej and faces Fj−1 of P in which each face Fj is incident with the edges
Ej−1 and Ej .

The vertices of the polygonal faces are called the vertices of P .

In other words, a polyhedron is a surface made out of a finite number of
plane convex polygons that are connected (“hinged”) to one another in such a
way that one can move from a point on a face or on one edge to a point on
an another face or edge by crossing a finite number of faces and edges without
having to go through a vertex. This definition excludes a number of undesirable
polyhedral surfaces from the particular class we wish to examine, namely:

- Polyhedral surfaces where more than two faces meet at one edge. For
instance, two pyramids which share a common edge, as shown: edge SC
is common to the triangular faces 4SAC, 4SF C, 4SBC and 4SCE.
278 CHAPTER 14. ON REGULAR POLYHEDRA

- Polyhedral surfaces which are not connected, in the sense that they contain
faces which meet at one vertex and have no edge in common. In the
example shown, one cannot move from a point on, say, face 4SAB to a
point on, say face 4SEF without passing through the common vertex at
S.

We shall actually restrict our study to the subclass of convex polyhedra:

Definition 83. A polyhedron is convex if it lies entirely on one side of the

plane of any of its faces.

This further restriction eliminates all polyhedral surfaces which contain “cav-
ities” or “tunnels.” A cube containing another cube entirely in its interior, or a
“picture frame”, is an example of this category of unwanted polyhedral surfaces.
It should be noted that a convex polyhedron can be deformed continuously, i.e.
without tearing or puncturing, into the surface of a sphere, whereas a toroidal
polyhedron, like the “picture frame” or a “cavernous” polyhedron cannot.

Definition 84. A regular polyhedron has faces which are regular congruent
polygons and vertices where the same number of faces (or edges) meet.

It is worthwhile mentioning that, similar to regular polygons whose incircle

and circumcircle are concentric, three concentric spheres can be associated with
a regular polyhedron: one through all its vertices, one tangent to all its faces ,
and a third one tangent to all its edges.
We are now ready to prove rigorously that there are exactly five regular poly-
hedra in 3-space. In 1750, the Swiss mathematician Leonhard Euler (1707–1783)
wrote to his German colleague Christian Goldbach of his discovery that the
 279

number V of vertices of any polyhedron less the number E of its edges plus the
number F of its faces equals two: V − E + F = 2. This is the famous polyhedral
formula which carries Euler’s name although its discoverer could not prove it.
The first proof of this formula is due to the French mathematician Adrien-Marie
Legendre (1752–1833) whom we have already met in collaboration with Saccheri
(Theorem 146), and a better known proof thereof, using graph-theoretical ideas,
belongs to another French mathematician, Augustin-Louis Cauchy (1789–1857).
Since then, many mathematicians have contributed to the theory of polyhedra.
One generalization led to the theory of polytopes which are the analogues of
3-dimensional polyhedra in higher dimensions, and the French mathematician
Henri Poincare (1854–1912) extended Euler’s formula to n-dimensional (Eu-
clidean) convex polytopes: instead of points, edges, and faces, one defines el-
ements with zero dimension, one dimension, two dimensions, and so on up to
(n − 1)-dimensional elements. Denoting the numbers of such elements for a
convex n-dimensional (Euclidean) polytope by N0 , N1 , N2 , . . . , Nn−1 , Poincare
proved the formula: N0 − N1 + N2 − N3 + . . . + (−1)n−1 Nn−1 = 1 − (−1)n
which reduces to Euler’s formula for n = 3. For n = 2, a convex 2-dimensional
polytope is just a convex polygon the number N1 of whose sides is equal to the
number N0 of vertices, hence N0 − N1 = 0 as predicted by the formula, and for
n = 1, one obtains a segment for which N0 = 2 (two endpoints) and N1 = 1
with N0 − N1 = 1.
It turns out that in (Euclidean) spaces of five or more dimensions, there are
only three types of regular convex polytopes, namely the higher-dimensional
analogues of the tetrahedron, the cube, and the octahedron. But in (Euclidean)
four-dimensional space, there are six kinds of regular convex polytopes:

- the 4-dimensional pentahedron (5-tope) where five 3-dimensional tetrahe-

dra form the 3-faces (N0 = 5, N1 = N2 = 10, N3 = 5),
- the 4-dimensional cube (8-tope) where eight 3-dimensional cubes form the
3-faces (N0 = 16, N1 = 32, N2 = 24, N3 = 8),
- the 4-dimensional 16-tope where the sixteen 3-faces are again 3-dimensional
tetrahedra (N0 = 8, N1 = 24, N2 = 32, N3 = 16),
- the 4-dimensional regular 120-tope whose 3-dimensional faces are dodec-
ahedrons,
- the 4-dimensional regular 24-hedron with octahedral 3-dimensional faces,
and
- the 4-dimensional regular 600-tope with tetrahedral 3-dimensional faces.

These six polytopes are difficult to visualize in 3-dimensional space, but com-
puter graphics can provide a great deal of information about their appearance.
We proceed next with the proof of Euler’s formula V − E + F = 2 for
convex (Euclidean) 3-dimensional polyhedra. We shall use some tools from
graph theory, and begin with the following definition:
280 CHAPTER 14. ON REGULAR POLYHEDRA

Definition 85. A network on a surface consists of a finite set of points called

nodes, line segments, and regions on the surface satisfying the following condi-
tions:

1. There are nodes at the ends of line segments.

2. Nodes exist only at the ends of line segments.

3. Each line segment contains the nodes at its ends.

4. The only points which can be common to two line segments are nodes.

5. Line segments do not intersect themselves.

6. The regions are exactly the parts into which the line segments divide the
surface.

In the example shown, the points A, A1 , A2 , . . . , A9 are the nodes of the net-
work, the lines a1 , a2 , . . . , a14 are its line segments, and the regions Q1 , Q2 , . . . , Q6
are its regions. Note that in this example, we have not restricted the line seg-
ments of a network to be “straight” line segments. Also, the surface carrying
a network may equally be “flat”, i.e. a Euclidean plane, or “curved” (without
holes and cusps), such as the surface of a sphere.
A polyhedron provides a natural example of a network: its vertices are
the nodes, its edges are the line segments, and its faces are the regions of the
network. A closed n-gon is another example of a network consisting in n nodes
(the vertices of the n-gon), n line segments (the sides of the n-gon), and two
regions, namely the interior and the exterior of the closed n-gon.

Definition 86. A prime network is the simplest possible network on a surface

and consists in a single point (node) on the surface. A prime network has no
line segments and only one region, the entire surface.

Definition 87. A network is connected if it is possible to go from one of its

nodes to any other node by moving along its line segments. If a network is
connected and if k is the least number of line segments of the network needed
to provide a path from a given node A to a given node B, then we say that B
is k steps away from A.
 281

In the examples above, all networks are connected, and in the first example,
node A is 4 steps away from node A9 .
We shall prove Euler’s formula for networks on convex surfaces and then
apply it to the classification of regular convex polyhedra.
Let n denote the number of nodes of a network, s the number of its line
segment, and r the number of its regions. For the network of the first example
above, n = 10, r = 6, and s = 14, so that, indeed, 10 − 14 + 6 = 2. For the
polygonal network provided by a convex k-gon: n = s = k and r = 2, and again,
k + 2 − k = 2. In particular, for a prime network: n = 1, s = 0 and r = 1 so
that the formula n − s + r = 2 is equally valid.

Definition 88. For any network with n nodes, s line segments, and r regions,
the number n − s + r is called the Euler characteristic of the network.

Starting from a connected network, more complex connected networks can

be constructed by the following two basic operations:

- In augmentations of the first kind, one starts from a node A of a connected

network K and create a new line segment AB having no point in common
with any other line segment of K except for its endpoint A, and so that
its other endpoint B does not belong to K.

- In augmentations of the second kind, two nodes C and D of a connected

network K are connected by a new line segment having no point in com-
mon with any other line segment of K, except for its endpoints.

In the example shown, line segment AB is an augmentation of the first kind

which generates a new connected network K 0 which has one line segment (AB)
and one node (B) more than K but the same number of regions. Thus, the
Euler characteristic of network K 0 is the same as that of the initial network:
(n + 1) − (s + 1) + r = n − s + r. In that same example, the new line segment
CD is an augmentation of the second kind: it lies entirely in a region R of the
network K and divides it into two new regions R1 and R2 . Hence, the new
network K 00 obtained from K by this augmentation has an Euler characteristic
n − (s + 1) + (r + 1) = n − s + r which is again the same as that of the original
network K.
We have thus established the following lemma:

Lemma. Augmentations of the first or second kind do not alter the Euler
characteristic of a connected network on a convex surface.
282 CHAPTER 14. ON REGULAR POLYHEDRA

Next, we prove:
Lemma. Any connected network on a convex surface arises from a prime net-
work by means of successive augmentations of the first and/or second kind.

Proof. Let K be some connected network on a convex surface and let A be one
of its nodes. Denote by K0 the prime network formed by A. Consider all line
segments of the network K emanating from A, i.e. the line segments a1 , a2 , a3 ,
a4 . We obtain a1 , a2 , a3 from K0 through successive augmentations of the first
kind and a4 through an augmentation of the second kind. One obtains a new
network K1 with nodes A, A1 , A2 , A3 and line segments a1 , a2 , a3 , a4 , and the
nodes of K1 other than A are each one step away from A. Next, we carry out
another sequence of augmentations of the first and second kind by adding to K1
those line segments of the network K for which one or both endpoints are nodes
of K1 but which themselves do not belong to K1 : a new connected network K2
is obtained from K1 containing the new lines a5 , a6 , a7 , a8 and the nodes of
K2 which do not belong to K1 are those nodes of the original network K which
are two steps away from A. Through a third sequence of augmentations of both
kinds, we add to K2 those line segments of the network K which do not belong
to K2 but have one or both endpoints belonging to K2 (line segments a9 , a10 ,
a11 ) and obtain the connected network K3 , and the nodes of K3 , which were
not already contained in K2 , are three steps away from A. Thus, by means
of successive augmentations of the first and second kind, one generates a chain
of connected networks K4 , K5 , . . . which contain all nodes of the network K
which are 4, 5, . . . steps away from the initial node A. Now, if K contains s line
segments, then all nodes of K are at most (i.e. within) s steps away from A.
Hence, by repeating the above procedure at most s + 1 times, one obtains the
entire network K, and the lemma is proved.
Euler’s formula for the natural network of a convex polyhedron is a special
case of the following theorem:
Theorem 155. The Euler characteristic of any connected network on a convex
surface is 2.
Proof. The Euler characteristic of a prime network is 2. The second Lemma
guarantees that any connected network K on a convex surface can be obtained
from a prime network by a finite number of successive augmentations of the first
 283

and second kind, none of which alters the Euler characteristic of the preceding
connected network. Therefore, the Euler characteristic of the final network is
also equal to 2, which was to be proved.
It should be stressed that the requirement of convexity is essential for the
proof of Euler’s formula for connected networks: this proof relies on the two
Lemmas which are valid for connected networks on convex surfaces only.
We now apply Theorem 155 to the identification of all possible regular convex
polyhedra in 3-space.
We have already seen that the “natural” network of vertices, edges, and faces
of a convex polyhedron constitutes a connected network on a convex surface.
Thus Euler’s formula is valid for this particular network and n − s + r = 2 is
true provided that we identify n with the number of vertices of the polyhedron,
s with the number of its edges, and r with the number of its faces. Since we are
restricting ourselves to regular polyhedra, every face of such a polyhedron has
the same number, say p, of edges and every vertex has the same number, say
q, of edges emanating from it. We note in passing that every convex uniform
polyhedron is characterized by its so-called Schlaefli symbol {p, q}. If a uniform
polyhedron is regular, then every edge has the same length.
The equation n − s + r = 2 is a typical Diophantine (after the Greek math-
ematician Diophantus of Alexandria who flourished around 250 AD) equation,
whose solutions are only natural numbers.
Since every edge has two vertices, then nq = 2s. Since every edge is common
to exactly two faces, then rp = 2s. Substitution into Euler’s formula yields
s[ p2 + 2q − 1] = 2 or equivalently, p2 + 2q − 1 = 2s . We must now determine all
natural numbers p, q, s which satisfy this equation.
Clearly, for a polyhedron in 3-space, p ≥ 3 and q ≥ 3 (why?). Next, observe
that p ≤ 5, for if p ≥ 6 while q ≥ 3, then p2 + 2q − 1 ≤ 13 + 23 − 1 = 0 which
contradicts the relation p2 + 2q − 1 = 2s . Hence, p can take on the values 3, 4 and
5 only. If p = 3, then it is easily seen that the requirement 2s > 0 leads to q < 6,
so that q too can only assume the values 3, 4 and 5. Similarly, one shows that if
p = 4 or p = 5, then q can only take on the value 3. p2 + 2q −1 ≤ 1/2+1/2−1 = 0.
It is also easily seen that for q = 4 or q = 5, one has p = 3.
There are therefore only five possible cases:

1. p = 3, q = 5. Then, 2s = 25 + 23 − 1 = 15
1
i.e. s = 30. Then, n = 2s 60
p = 5 =
12 and r = 2s 60
q = 3 = 20. This is the case of the icosahedron: twenty
congruent equilateral triangular faces.
2. p = 3, q = 4. Then, one calculates: s = 12, n = 6 and r = 8. This is the
case of the octahedron which has eight congruent equilateral triangular
faces.
3. p = 3, q = 3. Then s = 6, n = 4 and r = 4 which corresponds to the
tetrahedron with four congruent equilateral triangular faces.
4. p = 4, q = 3. This choice yields the cube with s = 12, n = 8, and r = 6.
284 CHAPTER 14. ON REGULAR POLYHEDRA

5. p = 5, q = 3. Here, s = 30, n = 20 and s = 12, which corresponds to the

dodecahedron which has twelve congruent regular pentagonal faces.

There are no other regular convex polyhedra in 3-dimensional space. Notice

that we have actually identified the uniform polyhedra rather than the regular
ones, since we did not use the fact that the edges of a regular polyhedron are
congruent: for the bona fide values of p and q as determined by Euler’s formula,
we have only the above five uniform convex polyhedra, up to stretching or
shrinking.
We conclude this chapter by mentioning that Euler’s deceptively simple for-
mula for the number of vertices, edges, and faces of a convex polyhedron turns
out to have a wealth of implications, even though it does not say anything about
the metric properties of polyhedra, e.g. the measures of their polyhedral angles
or the lengths of their sides, and raises a number of equally deceptively simple
but intricate questions. For example, one may assign to every polyhedron a so-
called f-vector (V, E, F ) which consists in an ordered triple of natural numbers
V , E, and F satisfying Euler’s formula, and then ask to identify all polyhedra
which correspond to a given f-vector, i.e. polyhedra with V vertices, E edges,
and F faces. In spatial dimensions higher than three, this problem has been
only partially solved to date.
Finally, regarding the intrinsically interesting—and intriguing—generalization
of Euler’s formula to non-convex polyhedra, we content ourselves with mention-
ing that there are non-convex polyhedra with Euler characteristic equal to 2,
and refer the interested reader to the select bibliography for this chapter.
 285

Exercises
1. The figure shown represents a 3-dimensional solid that has six equilateral
triangles for its face.

(a) Is it a uniform/regular polyhedron?

(b) Does Euler’s formula apply to it? Justify your answer.

2. The base of a pyramid is a convex n-gon. Check whether Euler’s formula

is valid for this polyhedron.

3. Calculate the f-vector (V,E,F) and the Euler characteristic of a semi-

regular prism and a semi-regular anti-prism whose bases are convex n-
gons.

4. One way of constructing new polyhedra out of given ones consists in cut-
ting off corners as shown in the case of a cube.

(a) Does Euler’s formula apply to this “mutilated” cube? Justify your
answer.
(b) Suppose Euler’s formula applies to a given polyhedron before one
corner with n edges meeting at it is cut off. Show that Euler’s formula
also applies to the polyhedron thus obtained.

Five of Archimedes’ 13 semi-regular polyhedra can be obtained from the

five Platonic solids by applying this truncation procedure in such a way
286 CHAPTER 14. ON REGULAR POLYHEDRA

that the new faces are again regular polygons while the portions of the
former faces that are left also form new regular polygons.

For instance, the sequence of the polyhedra shown above begins with a
cube and ends with a regular octahedron while passing through a semi-
regular truncated octahedron with 14 faces, 6 of which are congruent
squares and 8 are congruent regular hexagons.
5. Another way of constructing new polyhedra is to “roof over” faces of a
given polyhedron, as shown for the “roofed over” cube where a pyramid
with square basis has been added to one of the cube faces.

(a) Show that Euler’s formula applies to this “roofed over” cube.
(b) Determine the f-vector of the polyhedron obtained by “roofing over”
one face of a polyhedron with f-vector (V, E, F ).
(c) What happens if two different faces of a polyhedron with f-vector
(V, E, F ) are “roofed over”?

Stellation, which is a special way of “roofing over,” is another method for

constructing (non-convex) regular polyhedra from some regular polyhe-
dra. A stellated polyhedron is obtained by extending the planes of each
face of a convex polyhedron until these planes intersect to create a new
polyhedron. For instance, the Kepler’s urchin (“cui nomen Echino feci”)
or Kepler-Poinsot star polyhedron is obtained by stellating a regular do-
decahedron and consists of 12 star pentagons. It has 32 vertices, 90 edges
and 60 pentagonal faces. For this regular non-convex polyhedron we still
have V − E + F = 2.
287
288 CHAPTER 14. ON REGULAR POLYHEDRA
Chapter 15

Trigonometry

Literally, Trigonometry means “measurement of triangle(s).” This discipline, is

actually a computational scheme which allows for the numerical determination
of angle measures and lengths in triangles, and appears to have been origi-
nated in Ancient Greece by Hipparchus of Bithynia (around 100 B.C.). The
first extant work on Trigonometry is contained in the Almagest, a treatise on
astronomy produced by Claudius Ptolemy (ca. 100–178). In turn, Hindu math-
ematicians developed further calculational techniques and compiled tables of
sines (especially, the Surya Siddhanta, written around 400) to solve right trian-
gles. Unfortunately, none of these works has survived except in Arabic transla-
tions. Major developments of modern Trigonometry are due almost exclusively
to Muslim mathematicians, and culminate in the systemization of plane and
spherical trigonometry by the Persian al-Tusi (1201–1274). In this chapter, we
shall be concerned with plane Trigonometry only, beginning with measurements
in a right triangle and continuing with trigonometric functions.

15.1 Right Triangle Trigonometry

Throughout our discussion, we shall adhere to the convention which labels a
vertex of a triangle—and hence, the corresponding triangle angle—with an up-
per case Roman letter and its opposite side with the same lower case Roman
letter. Thus, in 4ABC, side BC which is opposite angle ]A shall be labeled
a, and so on.
Let the right triangle 4ABC have a right angle at vertex A. Each of the
two acute angles ]B and ]C has the hypotenuse BC for one leg, while the
other leg of each angle will be called its adjacent side in the triangle: AC is the
adjacent side of angle ]C and AB is the adjacent side of angle ]B.
Definition 89. The sine of the measure of angle ]C is the ratio ac of the length
of the triangle side opposite angle ]C to the length of the hypotenuse, and is
denoted sin C. Similarly, sin B is the ratio ab of the length of the triangle side
opposite angle ]B to the length of the hypotenuse.

289
290 CHAPTER 15. TRIGONOMETRY

Definition 90. The cosine of the measure of angle ]C is the ratio ab of the
length of the triangle side adjacent to angle ]C to the length of the hypotenuse,
and is denoted cos C. Similarly, cos B is the ratio ac of the length of the triangle
side adjacent to angle ]B to the length of the hypotenuse.

In the above definitions, a certain sloppiness in the notation should be

pointed out: the sine and cosine are quantities (real numbers) associated with
the measure of an (acute) angle of a right triangle and should be written
sin(m ]C) or cos(m ]B), but in view of the Angle Measurement Axiom (Ax-
iom 10) which assigns a unique real number between 0 and 180 to each angle,
it has become customary to identify the geometric object (angle) with its nu-
merical measure.
Several important remarks follow from Definitions 89 and 90. First, the
sine and/or cosine are only defined for acute angles in right triangles, and,
being ratios of lengths, are unitless positive numbers. In the next section, we
will extend the definition of these trigonometric objects to arbitrary angles.
Moreover, since the hypotenuse a is the longest triangle side, the ratios ac and
b
a are necessarily less than one, so that the sine and cosine of acute angles are
positive numbers strictly less than one. Next, it is immediately recognized that
according to their definitions, for any pair of complementary angles ]C and
]B, sin C = cos B and cos C = sin B. Finally, the sine or cosine of an acute
angle does not depend on the particular triangle used to calculate it: given any
acute angle ]X with measure α, construct two arbitrary right triangles each
having an acute angle with measure α. Since any two such triangles are similar,
it follows that the ratios defining the sine and cosine of their acute angles will
be the same (why?). Thus, every acute angle with measure, say, 37◦ will have
a sine equal to approximately 0.601815 and a cosine equal to approximately
0.789635, as provided by a common hand-held scientific calculator.

Theorem 156. For any acute angle with measure α, sin2 α + cos2 α = 1, where
sin2 α and cos2 α stand for (sin a)2 and (cos a)2 , respectively.

Proof. Construct a right triangle 4ABC with right angle at vertex A and,
say, m ]B = α. Then, by definition, sin B = ab and cos B = ac . Hence,
2 2 2 2
(sin B)2 + (cos B)2 = ab 2 + ac 2 = b a+c
2 = 1 by virtue of the Pythagorean Theo-
rem. Similarly, (sin C)2 + (cos C)2 = 1.

Theorem 156 expresses the fundamental identity of trigonometry.

Besides the sine and cosine, several other trigonometric quantities are as-
sociated with an acute angle. If 4ABC is a right triangle with right angle at
vertex A:

Definition 91. The tangent of (the measure of) angle ]C is the ratio of the
length of the triangle side opposite angle ]C to the length of its adjacent triangle
side, and is denoted tan C. Thus, tan C = cb = (c/a) sin C
(b/a) = cos C . Similarly,
b sin B
tan B = c = cos B .
15.1. RIGHT TRIANGLE TRIGONOMETRY 291

The cotangent, secant, and cosecant of the acute angle ]B (or ]C) in the
right triangle 4ABC, are defined by cot B = tan1 B , sec B = cos1 B , and csc B =
1
sin B , respectively, with similar relations for angle ]C.

Clearly, all these trigonometric quantities are again unitless positive num-
bers. In particular, the tangent and cotangent of acute angles can take on any
positive value (in contradistinction with their sines and cosines which are al-
ways less than one), since in a right triangle, the legs of the right angle can have
any length independently from one another. On the other hand, the secant and
cosecant of an acute angle are always larger than one.

Theorem 157. For any acute angle with measure α: 1 + (tan α)2 = (sec α)2
and 1 + (cot α)2 = (csc α)2 .
2 2 2
sin α 2 sin α cos α+sin α
Proof. From Definition 91 : 1+(tan α)2 = 1+( cos α ) = 1+ cos2 α = cos2 α =
1
cos2 α by virtue of Theorem 156. The second identity is proved in an entirely
similar manner.

Note that we have again adhered to conventional notation and written tan2 α
or csc2 α in lieu of (tan α)2 or (csc α)2 .
In order to apply trigonometric techniques to the determination of angle
measures or lengths in triangles (that is, in order to “solve” triangles in common
parlance), one needs the values of the trigonometric quantities associated with
all acute angles according to Definitions 89, 90 and 91. Now, Theorem 156 and
Definition 91 show that the knowledge of one trigonometric quantity for a given
acute angle is sufficient for the determination of the values of all other quantities
pertaining to that same angle. For instance, if the measure of an acute angle is a
and,
√ say, cos α is known, then one can calculate sin α via Theorem 156: sin α =
1 − cos2 α where the positive square root is considered. Note that the quantity
1−cos2 α is always positive since cos α is always less than one for an acute angle.
sin α
It is then easy to calculate tan α as the ratio cos α , etc. Thus, for practical
calculations, it is necessary to know the value of one trigonometric quantity
for all acute angles. Before the widespread availability of hand-held scientific
calculators, the values of, say, the sine of acute angles were tabulated at small
intervals. Nowadays, the scientific calculator provides such values conveniently
and with a high degree of accuracy. It should be noted, however, that scientific
calculators use decimal degrees to denote fractional parts of degrees, whereas
historically, fractions of degrees were expressed in sexagesimal base as minutes
of arc and seconds of arc, using the prime and the double prime notation,
respectively. Thus, one minute of arc, i.e. 10 , equals 60 1
of one degree and one
00 1 1
second of arc, that is 1 , equals 60 of one minute of arc or 3600 of one degree.
For example, an acute angle with measure 37 degrees and 13 minutes and 54
seconds is written 37◦ 130 5400 . To convert this angle measure into the decimal
degree form used on a scientific calculator, consider that 130 = (13)( 60 1
) degree
◦ ◦
and 5400 = (54)( 3600 ) degree so that 37◦ 130 5400 = 37◦ + 60 + 3600 ≈ 37.23167◦ .
1 13 54

Conversely, the measure 52.25806◦ can be expressed in degrees, minutes and

seconds of arc as follows: 52.25806◦ = 52◦ + 0.25806◦ . Since 1◦ contains 60
292 CHAPTER 15. TRIGONOMETRY

minutes of arc, 0.25806◦ contain (0.25806)(600 ) i.e. 15.25806 minutes of arc,

and, similarly, (0.25806)0 contain (0.25806)(6000 ) i.e. 29.016 seconds of arc.
Thus, 52.25806◦ ≈ 52◦ 150 2900 .
It should also be mentioned that the scientific calculator determines the
(acute) angle measure—in decimal notation—whose sine or cosine or tangent
is given. In this regard, the following conventions should be kept in mind: if,
say, the tangent of a certain acute angle with unknown measure is given to
be equal to 17.43217, then this angle measure is denoted tan−1 (17.43217) or
arctan(17.43217), and the inverse (INV) or second (2nd) key on the calculator
yields its value as 86.716814◦ .
With these preliminaries in mind, we may proceed with some typical exam-
ples of trigonometric calculations involving right triangles.
Example 1. In triangle 4ABC, m ]A = 58◦ 400 , m ]B = 34◦ 230 1000 , and
c = 8 cm. Determine the lengths of sides BC and AC.

Solution. Drop the altitude AH from vertex A onto side BC to create two
right triangles 4ABH and 4ACH. Clearly, m ]C = 180◦ −(54◦ 400 +34◦ 230 1000 )
i.e. m ]C = 180◦ − (92◦ 630 1000 ) = 180◦ − (93◦ 30 1000 ) or m ]C = 86◦ 560 5000
(why?). Furthermore, m ]B = 34◦ 230 1000 ≈ 34.3861◦ and m ]C = 86◦ 560 5000 ≈
86.9472◦ . Then, in the right triangle 4ABH: sin B = AH c or AH = c sin B =
(8)(sin 34.3861◦ ) ≈ (8)(0.564767) ≈ 4.51813 i.e. AH ≈ 4.520 cm. And, in
the right triangle 4ACH: sin C = AH AH 4.520
b so that b = sin C or b ≈ sin 86.9472◦ ≈
4.520 BH
0.99858 ≈ 4.526 cm. Finally, cos B = c or BH = c cos B so that BH ≈
(8)(cos 34.3861◦ ) ≈ 6.602 cm and cos C = CH b or CH ≈ (4.526)(cos 86.9472 )
◦

so that CH ≈ 0.241 cm. Then, a = CH + BH ≈ 6.843 cm.

Example 2. Determine the measure of the angle between an edge and a face
of a regular tetrahedron.

Solution. Let SABC be a regular tetrahedron: its four faces are congruent
15.2. THE UNIT CIRCLE AND (PLANE) TRIGONOMETRIC FUNCTIONS293

equilateral triangles (see Chapter 14). Let s be the length of an edge of the
tetrahedron and SO the altitude from vertex S onto face 4ABC: O is the
centroid of the equilateral triangle 4ABC. The angle between an edge, say,
SC and face 4ABC is then the angle ]SCO (see Definition 56). Now, in the √
equilateral triangle 4ABC, the length of the median (and altitude) CM is s 2 3
√
(why?), so that CO = 23 CM and CO = s 3 3 . On the other hand, in the right
2 2
» √
triangle 4SOC, SO2 = SC 2 −OC 2 = s2 − s3 = 2s3 , so that SO = s 23 = s √23 .
SO
Hence, in the right triangle 4SOC, we can write either: tan ]SCO = OC i.e.
√ √ √ √ √
tan ]SOC = s√32 + s 33 = 2, or sin ]SCO = SO SC =
s√ 2
3
+s = √2 , or
3
√ √
cos ]SCO = OC s 3
SC = 3 + s = 3 .
3

In either case, the calculator determines the measure of angle ]SCO to be

approximately equal to 54.73561◦ or 54◦ 440 800 .

15.2 The Unit Circle and (Plane) Trigonometric

Functions
We now extend the earlier definition of trigonometric quantities which were
introduced for acute angles only (using right triangles), to arbitrary angles with
measures larger than 90◦ . To this end, we need several preliminary concepts
and definitions.
The unit circle is the circle centered at the origin O of a Cartesian (rectan-
gular) x-y coordinate system whose radius equals one unit. Conventionally, the
positive x-axis is chosen “horizontally” to the right and the positive y-axis is
chosen “vertically” upward. These coordinate axes divide the x-y plane into four
quadrants numbered counterclockwise from I to IV starting from the positive
x-axis.

In Geometry, angles were defined essentially as the union of two non-collinear

rays with a common vertex, and their measures were real numbers strictly larger
than zero and less than 180 degrees. In Trigonometry, it turns out that in view of
its applications, it is convenient to expand this geometric definition of angle by
allowing its measure to assume positive and negative values without restriction.
Thus, we first introduce the concept of a directed angle as an ordered pair
of rays with common vertex, where one ray is called initial leg (or side) of the
angle, and the other ray, called the angle’s terminal leg (or side), is obtained
from the initial ray through a rotation around the vertex. Clearly, the rotation
involved in this definition of directed angle, can occur either in the clockwise
or in the counterclockwise direction. Furthermore, it is no longer required that
294 CHAPTER 15. TRIGONOMETRY

the rays that determine the directed angle be non-collinear. Directed angles
that have the same initial leg and the same terminal leg are called coterminal
directed angles. When the initial leg of a directed angle is placed along the
positive x-axis and its vertex at the origin of a Cartesian coordinate system, the
directed angle is said to be in standard position. Henceforth, we shall consider
directed angles in standard position only, and drop the adjective “directed” for
the sake of brevity.

The measure of an angle is defined as follows: when in standard position, the

initial and terminal legs of the angle—which is now a central angle—intercept
an arc on the unit circle. The measure of this arc (see Definition 70) is then the
measure of the angle with the convention that this angle measure is positive
if the rotation from the initial to the terminal leg is counterclockwise and
negative otherwise. In the diagrams, two coterminal angles with positive mea-
sures α1 and α2 , and two coterminal angles with negative measures β1 and β2
are shown. It should be clear that starting from an angle in standard position,
one may construct infinitely many angles that are coterminal with it through
rotations of the terminal leg of the given angle in the counter—or clockwise
directions through as many number of full revolutions as desired. In this fash-
ion, one obtains angles whose measures are the sum of the measure of the given
angle plus or minus a whole multiple of 360◦ , since a full circle corresponds to
a central angle with measure 360◦ (cf. Definition 70). In order to avoid such a
plethora of coterminal angles—which is actually unnecessary in this introduc-
tory account of Trigonometry—we agree that the measure of an angle should be
restricted to any positive or negative real number between −360◦ and +360◦ .
Finally, we recall another scale for angle measure which is almost exclusively
used in Mathematics and was introduced earlier in Definition 76: one radian
is the measure of a central angle with vertex at the center of a circle of radius
r whose legs intercept an arc of length s equal to the radius r of the circle.
Since the length of a full circumference of radius r is 2πr, it follows that a full
circumference (a full revolution) corresponds to a central angle of 2π radians.
In other words, 360◦ and 2π radians are two measures in different scales for
◦
◦
a full circumference. Hence, one radian corresponds to 360 2π ≈ 57.29578 , and
180◦ and 90◦ correspond to π and π2 radians, respectively. Consequently, to
convert the degree measure (in decimal form) of an angle into radians, one must
π
multiply that degree measure by the factor 180 ◦ radian and to convert the ra-

dian measure of an angle into degrees, one must multiply that radian measure
◦
by the factor 180
p degree. Needless to say, the radian measure of angle, being
15.2. THE UNIT CIRCLE AND (PLANE) TRIGONOMETRIC FUNCTIONS295

the ratio of two lengths, is also a unitless quantity. Henceforth, the measure of
angle α will be denoted α if measured in radians, and α◦ if measured in degrees.
Thus sin 1 is the sine of an angle with measure 1 radian whereas tan 47◦ is the
tangent of an angle whose measure is 47◦ . In general, when no scale of angle
measurement is specified, radian measure is implied.

Consider an angle with measure α in standard position. The terminal leg of

this angle intersects the unit circle at point P . Depending on the value of a, P
will lie in one of the four quadrants. From P drop perpendiculars P Q and P R
onto the x- and y-axis, respectively, to find the coordinates u and v of point P :
the intercepts OQ and OR on the x- and y-axis determine u and v, respectively.
Clearly, depending on the location of point P in one of the four quadrants, its
coordinates u and v will take on any value between -1 and +1, including these
end values.
We are now in a position to extend the definition of trigonometric quantities
already introduced for acute angles of right triangles to angles with arbitrary
measures. Keeping the preceding notation in mind, we define:

Definition 92. Consider an angle with measure θ in standard position and let
u and v be the coordinates of the intersection of its terminal leg with the unit
sin θ u
circle. Then: sin θ = v and cos θ = u. Furthermore, tan θ = cos θ = v.

Thus, −1 ≤ sin θ ≤ 1, −1 ≤ cos θ ≤ 1, and the value of tan θ can be any real
number (why?).
This definition allows for a convenient pictorial depiction of trigonometric
quantities. To this end, we introduce the (unconventional) concept of “directed”
segments in standard position in analogy with directed angles in standard po-
sition according to the following (somewhat lengthy) definition:

Definition 93. A “directed” segment in standard position is a geometric seg-

ment which

- is either along or parallel to one of the coordinate axes,

- has one (initial) endpoint on the other coordinate axis, and

- has a “value” which is a real number according to the following convention:

296 CHAPTER 15. TRIGONOMETRY

If the “directed” segment is along or parallel to the x-axis, and if its

second (terminal) endpoint is in the first or fourth quadrant, then its value
is positive and equal to the numerical length of the geometric segment,
whereas this value is negative and equal to the negative numerical length
of the geometric segment if the terminal endpoint is in the second or third
quadrant; and similarly, if the “directed” segment is along or parallel to
the y-axis, and if its second (terminal) endpoint is in the first or second
quadrant, then its value is positive and equal to the numerical length of the
geometric segment, whereas this value is negative and equal to the negative
numerical length of the geometric segment if the terminal endpoint is in
the third or fourth quadrant.

With these conventions in mind, it is easily recognized that the sine and
cosine of an angle, as defined in Definition 92, can be pictured by the “directed”
segments P Q and OQ, respectively.

A pictorial representation of the tangent of an angle can be also obtained

by the following procedure: let T be the intersection of the terminal leg (or its
extension) of the angle under consideration with the tangent line at A (with
coordinates 1 and 0) to the unit circle. The similarity ratio of the similar
PQ AT
triangles 4OP Q and 4OAT , yields OQ = OA = AT since OA = 1.

Thus, the tangent can be represented by the “directed” segment AT because,

by definition, the tangent of an angle is the ratio of the sine to the cosine of
that angle, and the sine and cosine are represented by the “directed” segments
P Q and OQ, respectively. The following diagrams summarize the preceding
considerations and illustrate the sine, cosine and tangent of angles in standard
position with terminal legs in each of the four quadrants:
15.2. THE UNIT CIRCLE AND (PLANE) TRIGONOMETRIC FUNCTIONS297

Bearing in mind that the hypotenuse of the right triangles 4OPi Qi , i =

1, 2, 3, 4, has length equal to one, one recognizes that the extended definitions
of the sine, cosine, and tangent according to Definition 92 agree with those of
Definitions 89, 90 and 91 when the angle with measure α is acute. Moreover, in
each of these right triangles 4OPi Qi , i = 1, 2, 3, 4, the Pythagorean Theorem
asserts that OQ2i +Pi Q2i = 1, i.e. sin2 θi +cos2 θi = 1 which is the generalization
of Theorem 156 to arbitrary angles with measures between 0 and 2π.
The following important trigonometric values are readily verified:

sin 0 = 0, cos 0 = 1, tan 0 = 0;

π p π
sin( ) = 1, cos( ) = 0, but tan( ) is undefined (why?);
2 2 2

sin π = 0, cos π = −1, tan π = 0;

3π 3π
sin( ) = −1, cos( ) = 0, but tan(3p/2) is undefined (why?);
2 2

sin(2π) = 0, cos(2π) = 1, tan(2π) = 0.

Definition 92 together with the consistent use of the unit circle leads to the
derivation of a number of important properties for the trigonometric quantities.
298 CHAPTER 15. TRIGONOMETRY

To begin with, it is easily seen that as the measure θ of the central angle increases
from 0 to π2 , sin θ increases from 0 to 1, while cos θ decreases from 1 to 0, and
tan θ increases without bound from 0 to arbitrarily large positive values. As θ
continues to increase from π2 to π, sin θ decreases from 1 to 0, cos θ decreases
further from 0 to -1, and tan θ increases from arbitrarily small negative values
to 0. For θ between π and 3π 2 , sin θ decreases further from 0 to -1, cos θ increases
from -1 to 0, and tan θ continues to increase without bound from 0 to arbitrarily
large values. Finally, when θ takes on values from 3π 2 to 2π, sin θ increases
from -1 to 0, cos θ increases from 0 to 1, while tan θ increases from arbitrarily
small negative values to 0. The values of the sine, cosine, and tangent of some
important acute angles can be calculated readily using elementary Geometry:
In a 30◦ − 60◦ − 90◦ triangle, it is well-known that the shorter leg of the right
√
angle is half as long as the hypotenuse, and the longer leg of the right angle is 23
times as long as the hypotenuse. In this case,√either Definition 89 or 92 yields: √
sin 30◦
sin 30◦ = cos 60◦ = 12 and cos 30◦ = sin 60◦ = 23 , so that tan 30◦ = cos ◦ = 3
3

◦ sin 60◦
√ 30
and tan 60 = cos 60◦ = 3.
√
Similarly, the legs of an isosceles right triangle have each a length equal to 22
√
times the length of the hypotenuse, so that sin π4 = cos π4 = 22 and tan π4 = 1.
Next, the following fundamental identities can be readily established through
inspection of the unit circle representations of sine and cosine via “directed”
segments: for any angle with measure θ such that 0 ≤ θ ≤ π:

sin(−θ) = − sin θ, cos(−θ) = cos θ

sin(π − θ) = sin θ, cos(π − θ) = − cos θ

sin(π + θ) = − sin θ, cos(π + θ) = − cos θ

π π
sin( − θ) = cos θ, cos( − θ) = sin θ
2 2
π π
sin( + θ) = cos θ, cos( + θ) = − sin θ
2 2

sin(2π − θ) = − sin θ, cos(2π − θ) = cos θ

15.3. THE LAW OF SINES, THE LAW OF COSINES, AND ALL THAT299

For example, consider the acute angle θ and its supplementary obtuse an-
gle π − θ. Since 4OP 0 Q0 ∼ = 4OP Q (why?), the “values” of both “directed”
segments P 0 Q0 and P Q will be the same positive number equal to the com-
mon length of segments P 0 Q0 and P Q, and consequently, sin(π − θ) = sin θ.
Similarly, the “directed” segments OQ0 and OQ will have equal and opposite
“values,” thus leading to cos(π − θ) = − cos θ.
The verification of the other identities listed above can be carried out equally
easily by similar methods, and is left as exercises.
Finally, it should be borne in mind that there are always two distinct angles
with measures between 0 and 2π whose sine, or cosine, or tangent have the same
value. To illustrate this fact, suppose we want to determine the measure θ of
the angle(s) whose, say, tangent has the value 0.6 : tan θ = 0.6. The scientific
calculator’s short answer is tan−1 0.6 ≈ 30.96376◦ . The unit circle represen-
tation, however, shows that there are two angles having a tangent equal to
0.6: an acute angle with measure 30.96376◦ and and obtuse angle with measure
180◦ + 30.96376◦ = 210.96376◦ .

The same holds true when the sine or the cosine of an angle is specified.
The scientific calculator always supplies the smallest angle measure whose sine,
cosine or tangent has a prescribed value.

15.3 The Law of Sines, The Law of Cosines, and

All That
Theorem 158 (Law of Sines). In any triangle 4ABC with sides of lengths a,
b, c, the following relation is valid: sina A = sinb B = sinc C 1 .

Proof. Let CH be the altitude from, say, vertex C onto side AB of the arbitrary
triangle 4ABC. Let CH = h. Then in triangle 4ACH: sin A = hb or h =
1 First proved in generality by the Persian mathematician al-Biruni (973–1048)
300 CHAPTER 15. TRIGONOMETRY

b sin A. Similarly, in triangle 4CHB: sin B = ha or h = a sin B. Consequently,

a sin B = b sin A or sina A = sinb B . Notice that sin A and sin B cannot be equal to
0 since no angle of non-degenerate triangle can have a measure equal to 0◦ or
180◦ . In a similar manner, one constructs the triangle altitude from, say, vertex
B onto side AC (extended, if necessary), and shows that sina A = sinc C . This
completes the proof of the Law of Sines.

Theorem 159 (Law of Cosines). In any triangle 4ABC with sides of lengths
a, b, c, the following relations are valid: a2 = b2 + c2 − 2bc cos A, b2 = a2 + c2 −
2ac cos B, and c2 = a2 + b2 − 2ab cos C. 2

Proof. We prove only the first of the three relations stated in the theorem. Let
triangle 4ABC be given with an acute angle at, say, vertex A. If one of the
angles of the triangle is obtuse, then both other angles are necessarily acute,
and we can choose any one of them as being angle ]A. Let CH be the altitude
from vertex C onto side AB, as shown. Then, in triangle 4ACH: AH = b cos A
and CH 2 = b2 − AH 2 = b2 − b2 cos2 A. In the same fashion, in triangle 4CHB:
CH 2 = a2 −HB 2 = a2 −(c−AH)2 = a2 −c2 −b2 cos2 A+2bc cos A. Consequently,
b2 − b2 cos2 A = a2 − c2 − b2 cos2 A + 2bc cos A or a2 = b2 + c2 − 2bc cos A, which
was to be proved. The other two relations can be proved in exactly the same
manner.

Notice that if m ]A = 90◦ , then cos A = 0 and the Law of Cosines becomes
a = b2 + c2 which is the Pythagorean Theorem applied to a right triangle with
2

right angle at vertex A. Thus, the Pythagorean Theorem is a special case of the
more general Law of Cosines.
The Law of Sines can be used to prove the so-called important Addition
Formulae:

Theorem 160 (Addition Formula for Sines). For any pair of angles with mea-
sures α and β with 0 ≤ α ≤ π and 0 ≤ β ≤ π: sin(α + β) = sin α cos β +
cos α sin β.

2 First proved by al-Battani (ca. 858–929) who also produced tables of tangents at 1◦

intervals.
15.3. THE LAW OF SINES, THE LAW OF COSINES, AND ALL THAT301

Proof. First consider the case of two acute angles α and β, such that 0 ≤ α+β ≤
−−→
π. Construct angle ]X with measure α + β. On ray XW , choose point H such
that XH = 1 unit and construct the perpendicular at H to XW meeting the
legs of ]X at Z and Y , as shown. Then, m ]XZH = π2 − α and since XH = 1,
we have 1 = XH = ZX sin[ π2 − α] = ZX cos α. Thus, ZX = sec α. In the same
ZH
fashion, one shows Y X = sec β. On the other hand, tan α = XH = ZH and also
YH
tan β = XH = Y H. Now, the Law of Sines applied to triangle 4XY Z yields:
ZY XY ZH+HY sec β tan α+tan β
sin(α+β) = sin( π −a) , that is, sin(α+β) = cos α , or equivalently,
2 sin(α+β) =
1
cos α cos β .
Hence, we obtain: sin(α + β) = (tan α + tan β)(cos α cos β) which
gives sin(α + β) = sin α cos β + cos α sin β.

When the angle measures α and β are each obtuse, then the preceding
construction can still be carried out with XH 0 = 1 and leads to the triangle
4XY 0 Z 0 shown. Now, however, m ]XZ 0 H 0 = π2 − (π − α) = −( π2 − a), so
that XH 0 = 1 = XZ 0 sin[−( π2 ) − a)] = XZ 0 [− sin( π2 − a)] = −XZ 0 cos α and
XZ 0 = − sec α. Notice that since α > π2 , both cos α and sec α are negative,
and in the end, − sec α is in fact a positive number expressing the length XZ 0 .
Similarly: XY 0 = − sec β. On the other hand, in triangle 4XZ 0 H 0 : Z 0 H 0 =
sin(π−α)
XH 0 tan(π − α). But tan(π − α) = cos(π−α) = −sin α 0 0
cos α = − tan α, so that Z H =
π
− tan α. Again, tan α itself is negative because α > 2 , and consequently, − tan α
is a positive real number. In the same manner: Y 0 H 0 = − tan β. The application
of the Law of Sines to triangle 4XY 0 Z 0 then yields:

Z 0Y 0 XY 0 XY 0 − sec β
= π = =
sin[(2π) − (α + β)] sin[−( 2 − a)] − cos α − cos α
Moreover, Z 0 Y 0 = −(tan α + tan β) and sin[(2π) − (α + β)] = − sin(α + β).
302 CHAPTER 15. TRIGONOMETRY

Therefore, all negative signs cancel out and one obtains anew sin(α + β) =
sin α cos β + cos α sin β. Thus, the Addition Formula for Sines is proved for all
angles α and β with measures between 0 and π.
Corollary (Subtraction Formula for Sines). For any pair of angles with mea-
sures α and β with 0 ≤ α ≤ π and 0 ≤ β ≤ π: sin(α − β) = sin α cos β −
cos α sin β.
Proof. Apply the Addition Formula to the angles α and −β and use the iden-
tities sin(−β) = − sin βand cos(−β) = cos β.
Theorem 161 (Addition Formula for Cosines). For any pair of angles with
measures α and β with 0 ≤ α ≤ π and 0 ≤ β ≤ π: cos(α + β) = cos αcosβ −
sin α sin β.
Proof. Use the Subtraction Formula for Sines applied to the angles with mea-
sures π2 − α and β: since cos(α + β) = sin[ π2 − (α + β)] = sin[( π2 − a) − b], we
can write: cos(α + β) = sin( π2 − a) cos(−β) − cos( π2 − a) sin(−β) = cos α cos β −
sin α sin β, because of the identities sin(−β) = − sin β and cos(−β) = cos β.
Corollary (Subtraction Formula for Cosines). For any pair of (acute) angles α
and β: cos(α − β) = cos α cos β + sin α sin β.
Proof. The proof is entirely similar to the proof of the preceding Corollary to
Theorem 157, and is left as an exercise.
Theorem 162 (Addition Formula for Tangents). For any pair of acute angles
tan α+tan β
α and β: tan(α ± β) = 1∓tan α tan β .

Proof. Left as an exercise.

The most important application of the Addition Formulae is the derivation
of the so-called Double Angle Formulae:
Theorem 163 (Double Angle Formulae). For any angle with measure α: sin 2α =
2 sin α cos α and cos 2α = cos2 α − sin2 α.
Proof. Apply the Addition Formulae for Sines and/or Cosines to the angles α
and β = α.
Triple Angle Formulae can also be derived by the same methods: for instance,
it is not difficult to prove the identities cos 3α = 4 cos3 α − 3 cos α and sin 3α =
3 sin α − 4 sin3 α.
Finally, the Addition Formulae for Sines and Cosines can be used to prove
the so-called Product-to-Sum and Sum-to-Product Formulae:
1
sin α sin β = [cos(α − β) − cos(α + β)]
2
1
cos α cos β = [cos(α − β) + cos(α + β)]
2
15.3. THE LAW OF SINES, THE LAW OF COSINES, AND ALL THAT303

1
sin α cos β = [sin(α + β) + sin(α − β)]
2
and
α+β α−β
sin α + sin β = 2 sin( ) cos( )
2 2
α+β α−β
sin α − sin β = 2 cos( ) sin( )
2 2
α+β α−β
cos α + cos β = 2 cos( ) cos( )
2 2
α+β α−β
cos α − cos β = −2 sin( ) sin( )
2 2
The proofs of all these relations involve straightforward algebraic manipula-
tions of the Addition Formulae and are left as exercises.
304 CHAPTER 15. TRIGONOMETRY

Exercises
1. If the hypotenuse of a right triangle is 42 cm and the measure of one acute
angle is 37◦ 100 , what are the lengths of the other two sides of this triangle?
2. Determine the measures of the acute angles of a right triangle with sides
5 inches, 12 inches, and 13 inches.
3. A boy is flying a kite and has let out 300 ft of string. If the string makes
an angle of 35◦ with the ground, how high is the kite?
4. Calculate the area of a regular pentagon inscribed in a circle of radius 12
cms.
5. The longer base of a trapezoid is 21 mm long, and the other three sides
are each 12 mm long.
Determine the measures of the angles of this trapezoid.
6. In isosceles trapezoid ABCD, the longer base CD is three times longer
than the shorter base AB. Let DB be a diagonal and m ]CDB = 30◦ .
Determine the measure of angle ]CDA.
7. A square pyramid has base edges 8 cm long and lateral edges each 12 cm
long. Determine the angle between a lateral face and the base.
8. In the figure shown, the angles of elevation to the top of the lighthouse
from boat A and B are 10.2◦ and 11.3◦ , respectively. If the two boats are
exactly 15 meters apart, how tall is the lighthouse?

9. In triangle 4ABC, AB = AC = n inches and angle ]A is acute. Express

the area of this triangle in terms of n and sin A.
10. The lengths of two adjacent sides of a parallelogram are 4 inches and 6
inches. Find the area of this parallelogram if the angle between the two
sides has measure π6 radian.
11. Given a triangle with a = 22 inches, b = 12 inches and m ]A = 42◦ , find
the remaining side and angles.
12. Prove that there is no triangle satisfying a = 15.2 ft, b = 20 ft, and
m ]A = 110◦ .
15.3. THE LAW OF SINES, THE LAW OF COSINES, AND ALL THAT305

13. Given a triangle with a = 10 cm, b = 4.5 cm and m ]C = 110◦ , determine

the remaining side and angles.

14. Find the values of the following trigonometric quantities without the help
of a calculator:

(a) sin( 7π
4 )
(b) cos(− 5π
6 )
(c) tan(−210◦ )
(d) tan 135◦

15. If cos A = 32 , determine the values of sin A and tan A without a calculator.
sin2 A
16. Prove that for any angle ]A: cos A + 1+cosA =1

17. Verify the identities:

(a) (tan2 x + 1)(cos2 x − 1) = − tan2 x

(b) tan2 β + 4 = sec2 β + 3
(c) sec θ − cos θ = sin θ tan θ
cos(−θ)
(d) 1+sin(−θ) = sec θ + tan θ

18. Solve the following two equations and give their solutions in radians (no
calculator allowed!).

(a) tan2 x = 3
(b) 2 sin2 u + sin u = 1

19. For 0 ≤ α ≤ π, draw a unit circle representation with “directed” segments

that demonstrates the identities

(a) cos(π + α) = − cos α

(b) sin( π2 + α) = cos α

20. Solve the following five equations and give their solutions in radians (no
calculator allowed!).

(a) 2 sin2 t + sin t − 1 = 0

(b) sin2 t + 2 sin t − 1 = 0
(c) sin y + cos y = 0
(d) 4 sin2 x tan x = tan x
(e) sin x = sec x
tan α+tan β
21. Prove the Addition Formula for Tangents: tan(α + β) = 1−tan α tan β
306 CHAPTER 15. TRIGONOMETRY

22. Prove the Triple Angle Formula: sin 3α = 3 sin α − 4 sin3 α

23. Prove the Sum-to-Product Formula: cos α cos β = 21 [cos(α−β)+cos(α+β)]

24. Prove the Product-to-Sum Formula: sin α + sin β = 2 sin( α+β α−β
2 ) cos( 2 )
Appendix A

Euclid’s Axioms of
Geometry

The formal axiomatic system developed by Euclid consists in 23 Definitions, 5

Postulates, and 5 Common Notions, as follows:

Definitions
1. A point is that which has no part.

2. A line is breadthless length.

3. The extremities of a line are points.

4. A straight line is a line which lies evenly with the points on itself.

5. A surface is that which has length and breadth only.

6. The extremities of a surface are lines.

7. A plane surface is a surface which lies evenly with straight lines on itself.

8. A plane angle is the inclination to one another of two lines in a plane

which meet one another and do not lie in a straight line.

9. And when the lines containing the angle are straight, the angle is called
rectilinear.

10. When a straight line set up on a straight line makes the adjacent angles
equal to one another, each of the equal angles is right, and the straight
line standing on the other is called a perpendicular to that on which it
stands.

11. An obtuse angle is an angle greater than a right angle.

307
308 APPENDIX A. EUCLID’S AXIOMS

12. An acute angle is an angle less than a right angle.

13. A boundary is that which is an extremity of anything.

14. A figure is that which is contained by any boundary or boundaries.

15. A circle is a plane figure contained by one line such that all the straight
lines falling upon it from one point among those lying within the figure
are all equal to one another;

16. And the point is called the center of the circle.

17. A diameter of the circle is any straight line drawn through the center and
terminated in both directions by the circumference of the circle, and such
a straight line also bisects the circle.

18. A semicircle is the figure contained by the diameter and the circumference
cut off by it. And the center of the semicircle is the same as that of the
circle.

19. Rectilinear figures are those which are contained by straight lines, trilateral
figures being those contained by three, quadrilateral those contained by
four, and multilateral those contained by more than four straight lines.

20. Of trilateral figures, an equilateral triangle is that which has its three sides
equal, an isoceles triangle that which has two of its sides alone equal, and
a scalene triangle that which has its three sides unequal.

21. Further, of trilateral figures, a right-angled triangle is that which has a

right angle, an obtuse-angled triangle that which has its three angles acute.

22. Of quadrilateral figures, a square is that which is both equilateral and

right-angled; an oblong that which is right-angled but not equilateral; a
rhombus that which is equilateral but not right-angled; a rhomboid that
which has its opposite sides and angles equal to one another but is neither
equilateral nor right-angled. And let quadrilaterals other than these be
called trapezia.

23. Parallel straight lines are straight lines which, when being in the same
plane and being produced indefinitely in both directions, do not meet one
another in either direction.

Postulates
Let the following be postulated:

1. To draw a straight line from any point to any point.

2. To produce a finite straight line continuously in a straight line.

 309

3. To describe a circle with any center and distance.

4. That all right angles are equal to one another.
5. That, if a straight line falling on two straight lines make the interior angles
on the same side less than two right angles, the two straight lines, if
produced indefinitely, meet on that side on which are the angles less than
the two right angles.

Common Notions
1. Things which are equal to the same thing are also equal to one another.
2. If equals be added to equals, the wholes are equal.
3. If equals be subtracted from equals, the remainders are equal.
4. Things which coincide with one another are equal to one another.

5. The whole is greater than the part.

310 APPENDIX A. EUCLID’S AXIOMS
Appendix B

Hilbert’s Axioms for

(Plane) Euclidean
Geometry

“One must be able to say at all times—instead of points, lines and

planes—tables, chairs and beer mugs.”
David Hilbert (1862–1943)

In 1899, Hilbert proposed in his “Grundlage der Geometrie” (Foundations of

Geometry) a formal axiomatic system for Euclidean Geometry with the aim to
provide Geometry with content and to establish it on a mathematically rigorous
basis. We shall present here only a summary review of Hilbert’s axioms for plane
Euclidean Geometry and refer to the definitions in the main text as needed.
The purpose of this Appendix is to show how a purely geometric (i.e. non-
algebraic) formal axiomatic system can be developed which does not contain
the flaws of Euclid’s system and leads to all the theorems of standard Euclidean
Geometry. Hilbert’s axioms (far more numerous than Euclid’s) deal with the
primitive terms “point,” “line,” “plane,” “incidence,” and “congruence,” and
are necessary in order to provide unequivocal answers to fundamental existence
questions, such as whether a line contains more than two points, whether a
plane contains more than two lines, etc. It should be pointed out that if one
considers these existence questions as pedantic and/or irrelevant to the very
edifice of Geometry with its numerous beautiful results, then one may discard
Hilbert’s system as a mere elaboration on solid “common sense” and be content
with Euclid’s framework.
Incidence Axiom 1. For every point P and and every point Q distinct from
P , there exists a unique line that is incident with P and Q.
This Axiom is equivalent with Euclid’s Postulate 1 and implies the existence
of at least two points.

311
312 APPENDIX B. HILBERT’S AXIOMS

Incidence Axiom 2. For every line l there exist at least two distinct points
incident with l.
This axiom ensures that lines contain at least two points.
Incidence Axiom 3. There exist three distinct points with the property that
no line is incident with all three of them.
Incidence Axiom 3 asserts that there are at least three points.
Betweenness Axiom 1. If point B lies between points A and C, then A, B,
C are three distinct points lying on the same line and B is between A and C.

Betweenness Axiom 2. Given two distinct points B and D, there exist points
A, C, and E lying on the same line through B and D such that B lies between
A and D, C lies between B and D and D lies between B and E.
This Axiom ensures that the line incident with two points contains other
points and does not end at either of those two points.

Betweenness Axiom 3. If A, B, and C are three distinct points lying on the

same line, then one and only one of them is between the other two.
This Axioms that a line is not “circular.”
Betweenness Axiom 4. For any line l and any three points A, B, C not lying
on l:

(i) If A and B are on the same side of l and B and C are on the same side of
l, then A and C are on the same side of l.
(ii) If A and B are on opposite sides of l and B and C are on opposite sides
of l, then A and C are on the same side of l.

This is the Separation Axiom introduced in the main text as Axiom 9 and
guarantees that the geometry determined by these axioms is two-dimensional.
The next axiom group uses the primitive term “congruence”:
Congruence Axiom 1. If A and B are distinct points and if P is any point,
then for each ray emanating from P there is a unique point Q distinct from P
such that AB ∼= P Q.
This Axiom asserts that one can “move” segment AB to superimpose it on
segment P Q.
Congruence Axiom 2. If AB ∼ = CD and AB ∼ = EF , then CD ∼
= EF and
moreover every segment is congruent to itself.
This Axiom replaces two “Common Notions” of Euclid, namely that seg-
ments congruent to the same segment are congruent to each other and that
segments that coincide are congruent.
 313

Congruence Axiom 3. If B lies between A and C and P lies between Q and

R and AB ∼
= QP and BC ∼
= P R, then AC ∼ = QR.
This replaces another one of Euclid’s “Common Notions”, namely that if
congruent segments are added to congruent segments, the sums are congruent.
Congruence Axiom 4. Given an angle ∠BAC and given any ray emanating
−−→ −→
from a point P , say ray P Q, then there is a unique ray P R emanating from P
←→
on a given side of line P Q such that ]QP R ∼ = ]BAC.
This Axiom states that a given angle can be “laid off” on a given side of a
given ray in a unique way.
Congruence Axiom 5. If ]A = ∼ ]B and ]A ∼
= ]C, then ]B ∼
= ]C and
moreover every angle is congruent to itself.
Congruence Axiom 6 (SAS). If two sides and the included angle of one
triangle are congruent respectively to two sides and the included angle of another
triangle, then the two triangles are congruent.
We now proceed with a very sketchy account of the so-called Continuity
Axioms which are needed to fill in a number of gaps in Euclid’s proofs.
Axiom (Archimedes’). If AB and CD are any segments, then there exists a
−−→
positive integer n such that if segment CD is laid off n times on the ray AB
emanating from A, then a point E is reached where n · CD ∼ = AE and B is
between A and E.
This Axiom means that one may arbitrarily choose a segment as unit of
length and that then every other segment has a finite length with respect to
this unit.
More subtle but equally necessary is:
Axiom (Dedekind’s). Suppose that the set of all points on a line l is the union
of two non-empty subsets Σ1 and Σ2 , such that no point of Σ1 is between two
points of Σ2 and vice-versa. Then there exists a unique point O lying on l such
that O is between two points P1 and P2 if and only if one of these two points is
in Σ1 and the other is in Σ2 and O 6= P1 and O 6= P2 .
Although awkward in its formulation, the Dedekind Axiom is essentially the
converse of the line separation property which states that any point O lying on a
given line separates all the other points on that line into those to the “left” of O
and those to the “right” of O. The above Axiom asserts that any separation of
points on line l into “left” and “right” is produced by a unique point. Loosely
speaking, Dedekind’s Axiom ensures that there is no “hole” or “gap” in a line.
In terms of our axioms, this is tantamount to the statement that for any point
O on line l and for any positive real number r, there exist unique points P+ and
P− on l such that O is between them and segments OP+ and OP− have length
r. Without Dedekind’s Axiom there would be no guarantee for the existence of
a segment of length, say, π.
314 APPENDIX B. HILBERT’S AXIOMS

It can be shown that Archimedes’ Axiom follows from Dedekind’s Axiom

and the other Hilbert’s Axioms. Using Dedekind’s Axiom, one may prove a
number of so-called Continuity Principles which have been introduced by various
mathematicians to fill certain gaps in some of Euclid’s original proofs, among
others:

- The Circular Continuity Principle which asserts that if a circle has one
point inside and one point outside another circle, then the circles intersect
in two points.

- The Elementary Continuity Principle which ensures that if one endpoint

of a segment is inside a circle and the other endpoint is outside, then the
segment intersects the circle.

The last axiom is:

Parallelism Axiom 1. For every line l and every point P not lying on l there
is at most one line m through P such that m is parallel to l.

Euclid’s Fifth Postulate asserts, in addition, that at least one line through
P is parallel to l. However, this “at least” statement can be proved from the
other axioms, therefore, Hilbert used the “at most” in his formulation of the
Parallel Axiom.
This completes our presentation of Hilbert’s axioms for plane Euclidean
Geometry.
Appendix C

The Field of Real Numbers

The purpose of this Appendix is to provide the reader with the bare mini-
mal knowledge of the real numbers and their properties that is necessary for a
proper appreciation and application and application of the axioms underlying
our approach to Euclidean Geometry, as laid out in the main text. Thus, this
excursion into the realms of the algebra of real numbers must remain cursory,
with almost no proofs for assertions made and a few precise definitions. The
subject matter of this Appendix is alaborated in Algebra 2 and developed in
full rigor in Theoretical Calculus. We shall take for granted a sound knowledge
of the algebra of integers, that is the set of positive and negative whole numbers
and zero and of the operations of addition, subtraction, multiplication, and divi-
sion of these numbers, along with the properties pertaining to these operations
and the ordering of integers (inequalities).
A rational number is a quotient m n of two integers m and n where n is not
zero. Clearly, rational numbers can be positive, negative, or zero, e.g. 23 or
− 11
7 . Since there are infinitely many integers, there are infinitely many rational
numbers. This notion of “infinitude” should be accepted on an intuitive basis
and will not be explored here. Notice that integers are themselves rational
numbers with one for a denominator. Operations with rational numbers are
also assumed to be well-known: one may freely add, subtract, multiply and
divide two rational numbers—again, with the exception of division by zero—
and obtain new rational numbers. The natural ordering of integers induces, in
p
turn, an ordering of rational numbers: given two rational numbers m n and q ,
p p
the inequality m m
n > q means (in case of integers) that the difference n − q is a
positive rational number.
One important observation is the fact that there is always a rational number
p
between any two rational numbers, no matter how close: let m n and q be two
m p m p
positive rational numbers and suppose that n is less than q : n < q (which is
the same as pq > m n ). We construct the number

1 mp (mq + np)
[ ]=
2 nq 2nq

315
316 APPENDIX C. THE FIELD OF REAL NUMBERS

p
and claim that this is a rational number between m n and q . The rationality of
this number follows from the observation that the numbers mq and np are inte-
gers, and hence their sum is an integer, and that 2nq is another integer not equal
to zero, so that the ratio of mq + np and 2nq is indeed a rational number. Next,
p
consider the original inequality m n < q : cross-multiplication yields: mq < np.
Multiplying both sides by the positive number q gives: mq 2 < npq and adding
the integer npq to both sides of this inequality gives: mq 2 + npq < 2npq which
can be written as the cross-product of the inequality: (mq+np) 2nq < pq . But this
1 mp p
last inequality is exactly 2 [ n q ] < q which shows that our constructed num-
ber is indeed less than pq . The proof that our number satisfies the inequality
1 mp m
2 [ n q ] > n is similar, and we have constructed a rational number between
two given rational numbers. The case where the rational numbers considered
are both negative can be dealt with along the same lines. Finally, the rational
number zero is trivially between any negative and any positive rational number.
This construction of a rational number between two given arbitrarily close ra-
tional numbers shows that there are, in fact, infinitely many rational numbers
between any two given rational numbers: one can continue dividing each such
intermediary rational number by two ad infinitum and obtain an infinitude of
intermediary rational numbers.

There are, however, numbers that are not rational in the above sense. In
order to show the existence of such “irrational” numbers, let us recall the def-
inition of the square root of a positive rational number: √ if N is a positive
rational
√ number, then the square root of N , denoted N , is a number such
that ( N )2 = N , i.e. such that the square of that square root√is equal to the
original number. With this definition in mind, we prove that 2 cannot be a
rational number. The
√ proof reproduced below is credited to Euclid. One begins
by assuming that 2 is a rational number, that is that it can be expressed as
the quotient of two integers, and then, by showing that this assumption leads
to a contradiction, one concludes that the initial assumption is untenable. This
is the method of indirect proof or proof
√ by contradiction, also known as RAA
(see Chapter 1). Assume then that 2 = ab where a and b are integers. If a and
b have any common factors, they can be divided out so that the quotient ab is
reduced to its lowest term which we shall write as pq : p and q are again integers
which have no common factor (are relatively prime). Then we have ( pq )2 = 2.
If we square out the left-hand side and multiply out, we obtain: p2 = 2q 2 . The
right-hand side of this equation is an even number (because it contains the factor
2), therefore the left-hand side must also be an even number. But you cannot
get an even number by squaring an odd number (why?), therefore p itself must
be an even integer. It is hence possible to write p in the form p = 2r where r
is another integer. Substituting this value of p in the preceding equation yields
4r2 = 2q 2 or 2r2 = q 2 . Since the left-hand side is an even number, q 2 and thus
q itself must be an even number. But then both p and q will have the common
factor 2, which
√ contradicts our assumption that p and q have no common factor.
Therefore, 2 is not a rational number which was to be proved.
 317

Thus, there is at least one irrational number. It does not take much reflection
to convinve oneself that there are, actually, many more: the product of any
rational number with this irrational number is itself irrational, and, in general,
the sum and product of two such irrational numbers are, in general, again
irrational and so on. It can also be proved that the square root of a positive
integer is either an integer or an irrational number. The cube root of, say,
100 and the well-known number π are other examples of irrational numbers.
In fact, the set of irrational numbers is infinite, and in a vague sense, there
are “more” irrational numbers than rational numbers. With a mathematically
precise definition of irrational numbers at hand, it can be proved that between
any two rational numbers, no matter how close, there is always an irrational
number, and, in turn, between any two irrational numbers, no matter how close,
there is always a rational number. One such abstract and rigorous definition of
an irrational number is sketched here—without much elaboration—for the sake
of its very close connection with one of Hilbert’s Axioms, namely Dedekind’s
Continuity Axiom1 :
We know that there are rational and irrational numbers, the latter being
temporarily defined negatively as non-rational numbers. Suppose the universe
of numbers is partitioned in two distinct classes L (for “low”) and H (for “high”)
by means of a cut, so that every rational number in L is less than every rational
number in H, and no number may belong to both L and H. Suppose further that
there is a unique number which can achieve this partition. If L has a largest
rational number or if H has a smallest rational number, the cut is that rational
number. If L has no largest rational number and H has no smallest rational
number, the cut is an irrational number. The set of all rational and irrational
numbers constitutes the set of real numbers. In view of the preceding remark
and definition of an irrational number, the set of real numbers constitutes a
continuum in which there is a real number as close as one wishes to any other
real number, either smaller or larger than it. In figurative language, the set
of real numbers contains no “holes” or “gaps” in the same way that points on
a line are “packed so tightly that there is no room between adjacent points”
(whatever this statement means!).
We now introduce the rules governing operations with real numbers. Real
numbers can be added, subtracted, multiplied, and divided—as usual with the
exception of division by 0—and the results are again real numbers. This is
usually stated as the closure of the set of real numbers under these operations.
Operations with real numbers are subject to the following rules (axioms):
R 1. Addition of real numbers is associative: if a, b, and c are real numbers,
then a + (b + c) = (a + b) + c.
R 2. There is a unique real number 0 (zero) such that for any real number a:
a + 0 = a.
R 3. For every real number a there is a real number −a such that a + (−a) =
(−a) + a = 0.
1 See Appendix B.
318 APPENDIX C. THE FIELD OF REAL NUMBERS

R 4. Addition of real numbers is commutative: if a and b are real numbers,

then a + b = b + a.

R 5. Multiplication of real numbers is associative: if a, b, and c are any real

numbers, then a(bc) = (ab)c.

R 6. There is a unique number 1 (one) such that for any real number a: 1a =
a1 = a. Moreover, 1 6= 0.

R 7. For every non-zero real number a, there exists a real number a−1 (or 1
a)
such that aa−1 = a−1 a = 1.

R 8. Multiplication of real numbers is commutative: for any real numbers a

and b, ab = ba.

R 9. For any real numbers a, b, and c, the distributive law holds: a(b + c) =
ab + ac.

Rules R1 through R9 define the field of real numbers equipped with the
operations of addition and multiplication. Notice that subtraction of two real
numbers a and b is actually defined through addition: a−b = a+(−b). Similarly,
division is defined through multiplication: the division of real number a by the
real number b 6= 0 is actually ab = ab−1 .
The field of real numbers is furthermore endowed with an order inherited
from the natural ordering of the integers (and, ipso facto, of rational numbers):
there is a relation, denoted by the symbol > (or, equivalently <), between pairs
of real numbers with the following rules (axioms):

R 10 (Law of Trichotomy). For every pair of real numbers a and b, exactly one
of the three relations is true: either a > b, or b > a, or a = b.

R 11 (Transitivity of Order). For any real numbers a, b, and c, if a > b and

b > c, then a > c.

R 12. For any real numbers a, b, and c, if a > b, then a + c > b + c.

R 13. For any real numbers a and b, and for any real number c > 0, if a > b,
then ac > bc.

Notice that the relation a > b is equivalent to a − b > 0.

The last Rule (R14) governing the ordered field of real numbers is that of
completeness which finds its intuitive expression in our earlier remark that the
set of real numbers is a continuum without “holes” or “gaps.” The precise state-
ment of this last axiom is contained in Dedekind’s Axiom cited in Appendix B.
Many consequences of this axiom, however, can equally be used as statements
expressing the completeness of the set of real numbers. But all this belongs
to the realm of Calculus, and we shall no longer linger on it here. One conse-
quence of R14, however, shall be needed in Chapters Eleven and Thirteen of
this introductory course and we proceed with stating it here as follows:
 319

The Euclidean Completeness Axiom. Every positive real number has a

positive square root.
This postulate will ensure that circles will intersect lines and intersect one
another, as expected intuitively. We
√ define the square root of zero to be zero,
and agree that√the statement x = a means x2 = a AND x ≥ 0. Obviously, if
a ≥ 0, then − a is another square root of a.
Based on the preceding purely algebraic first 13 axioms, one can prove many
well-known theorems which are frequently used in algebraic operations with real
numbers. We enumerate some of these theorems without proofs below:

- For any real number a, −(−a) = a.

- For any real number a, 0a = 0.
- For any real numbers a and b, a(−b) = −(ab)
- For any real numbers a, b, and c, if a + b = c, then a = c − b.
- For any real numbers a and b, if ab = 0, then either a = 0 or b = 0 or
both.
- For any real numbers a, b, and c, if c 6= 0 and ac = bc then a = b.
- For any non-zero real numbers a and b, (ab)−1 = a−1 b−1 .
- For any real numbers a, b, c, and d, if a > b and c > d, then a + c > b + d.
- For any real numbers a and b, if a > b, then −b > −a.
- For any real numbers a, b, and c, if b > 0 and a + b = c, then c > a.
- For any real numbers a, b, and c, if a > b, then a − c > b − c.

We conclude this Appendix with a brief review of the concept of the absolute
value of a real number:
Definition. For any real number a, the absolute value of a is denoted by |a|
and defined by: |a| = a if a > 0, |a| = −a if a < 0, and |a| = 0 if a = 0.
Clearly, the absolute value is a short hand notation for “converting a real
number into a positive real number with the same magnitude.” Thus, |2| = 2
and | − π| = π but |a + b| is generally NOT equal to |a| + |b|, as asserted by the
“Grand Daddy” of all inequalities, namely the important Triangle Inequality,
which we state without proof:
Theorem. For any real numbers a and b: |a+b| ≤ |a|+|b| with equality holding
when a and b are both positive or negative or when one of them is zero, and
with the strict inequality holding when a and b have opposite signs.
In practice, the absolute value of the difference of two real numbers is the
magnitude of their separation, i.e. their “distance”.
320 APPENDIX C. THE FIELD OF REAL NUMBERS
Appendix D

The Basic Similarity

Theorem

In this Appendix, we prove the following nontrivial generalization of Theorem

81 of Chapter 6:

The Basic Similarity Theorem. Let L1 , L2 , L3 be three parallel lines cut

by the two transversals T and T 0 at points A, B, C and A0 , B 0 , C 0 , respectively.
B0 C 0
If B is between A and C, then BCAB = A0 B 0 .

Before expounding the details of the proof of this theorem, let us state clearly
B0 C 0
that ratios segment lengths such as BCAB , A0 B 0 , etc., are no longer required to
be rational numbers, as is usually assumed in elementary proofs of the Basic
Similarity Theorem: these ratios can be rational as well as irrational numbers.
For the proof, we shall need an already mentioned result which follows from
the Completeness Axiom (see Appendix C) and will be stated as the first lemma
(without proof), and two simple lemmata:

Lemma. Between between any two real numbers, there is at least one rational
number.

321
322 APPENDIX D. THE BASIC SIMILARITY THEOREM

Lemma. Let two parallel lines L1 and L2 be cut by distinct transversals T and
T 0 at points A, B and A0 , B 0 , as shown. Let P be a point on T between A and
B and m be the parallel through P to L1 (and L2 ) intersecting T 0 at P 0 . Then
P 0 is between A0 and B 0 .

←−→
Proof. Points A and A0 are on the same side of m (since AA0 is parallel to
m) and similarly, points B and B 0 are on the same side of m. By hypothesis,
A and B are on opposite sides of m (since segment AB intersects m at P ).
Hence, A and B 0 are on opposite sides of m, as are A0 and B. By the Plane
Separation Axiom, A0 and B 0 must be on opposite sides of m. It follows that
←− →
A0 B 0 must intersect m in a unique point X which is exactly P 0 . Consequently,
P 0 is between A0 and B 0 .

Lemma. Let x and y be any real numbers. Suppose:

1. that every rational number less than x is less than y, and

2. that every rational number less than y is also less than x.

Then x = y.

Proof. If x 6= y, then either x < y or y < x (Law of Trichotomy in Appendix C).

Suppose x < y. Then, by the first lemma, there is a rational number pq (with p
and q integers) such x < pq < y, i.e. pq is less than y but not less than x. This
p0
contradicts (2). Similarly, if y < x, there exists a rational number q0 such that
p0
y< q0 < x. Again this contradicts (1). Therefore, x = y.

We may now proceed with the proof of the Basic Similarity Theorem.
 323

0 0
Proof. For the sake of brevity, let BC B C
AB = x and A0 B 0 = y. Let p and q be two
positive integers. First divide segment AB into q congruent segments, end to
end, as shown. In other words, use the Point Plotting Theorem to select points
−−→
A0 = A, A1 , . . . , Aq = B in this order on ray AB such that the length of each
−−→
segment Ai Ai+1 is AB q . Next, lay off on ray BC a sequence of p segments,
each with the same length AB q . Let the end points of these segments be in
order B0 = B, B1 , . . . , Bp . Then, use the Parallel Axiom to draw lines parallel
to, say, L1 through points A1 , A2 , . . . , Aq , B1 , B2 , . . . , Bp . Let the intersections
of these parallel lines with the transversal T 0 be A01 , A02 , . . . , A0q , B10 , B20 , . . . , Bp0 .
BB
By construction, A1 Ai+1 = Bj Bj+1 so that ABp = pq . By Theorem 81, we also
B0 B0 p
have A0 BP0 = p/q. Now, suppose BC AB
AB = x > q . Then, p · q < BC. Therefore,
BBp < BC and BP lies between B and C, as shown, and by the second lemma,
0 0
Bp0 lies between B 0 and C 0 . Therefore, B 0 Bp0 < B 0 C 0 and hence, p · A qB < B 0 C 0
0 0p p
or B C BC
A0 B 0 > q (we have reversed the steps which led from q < AB for Bp lying
p p
between B and C). Consequently, if q < x, then q < y. In an exactly similar
fashion one shows that if pq < y, then pq <. It then follows from the third lemma
that x = y, which was to be proved.
324 APPENDIX D. THE BASIC SIMILARITY THEOREM
Appendix E

The Two-Circle Theorem

In Chapter 9 we stated and used the Two-Circle Theorem (Theorem 115) with-
out proof. In this Appendix, we provide a rigorous proof of this intuitively
“evident” theorem.

The Two-Circle Theorem. Let C and C 0 be coplanar circles of radius a and

b, respectively, and let c > 0 be the distance between their centers. If each of
the three numbers a, b, c is less than the sum of the other two, then circles
C and C 0 intersect in two distinct points lying on opposite sides of the line of
centers.

For a proof of this theorem we need a preliminary Lemma. If a, b, c are the

lengths of the sides of a triangle, then by the Triangle Inequality (see Appendix
C), each of these numbers is less than the sum of the other two. Our Lemma
proves the converse of the previous assertion:

Lemma (The Triangle Theorem). Given three positive numbers a, b, c, if each

of these numbers is less than the sum of the other two, then there exists a
triangle having side lengths a, b, c.

325
326 APPENDIX E. THE TWO-CIRCLE THEOREM

Proof. We may always arrange the given three positive numbers in decreasing
order, say, a ≥ b ≥ c. Let BC be a segment with length BC = a. We must
determine a point A such that AB = c and AC = b. Let us first try to figure out
the location of a point which fulfills our requirements. Thus, suppose 4ABC
is found with the prescribed side lengths, and let D be the foot of the altitude
from A. Then D is between B and C because we have assumed that BC is the
longest side of this triangle. Let BD = x so that CD = a − x, and let AD = y.
Then the Pythagorean Theorem yields: y 2 = c2 − x2 and also y 2 = b2 − (a − x)2 .
Therefore, we may write: c2 − x2 = b2 − (a − x)2 or a2 + c2 − b2 = 2ax. Hence,
2 2 2
x = a +c2a−b . Note that x > 0 because by assumption a ≥ b, i.e. a2 ≥ b2 and
p
c2 > 0. Consequently, y = (c2 − x2 ) which we know to exist by the Euclidean
Completeness Axiom (see Appendix III) since x < c (why?). Thus, if x and y
2 2 2
satisfy the relations y 2 = c2 − x2 and y 2 = b2 − (a − x)2 , then x = a +c2a−b and
p p
y = (c2 − x2 ). The converse of this statement is also true: if y = (c2 − x2 ),
2 2 2
then trivially y 2 = c2 − x2 , and if x = a +c2a−b , then 2ax = a2 + c2 − b2 or
equivalently 2ax − x2 = a2 + c2 − b2 − x2 or

c2 − x2 = b2 − a2 + 2ax − x2 = b2 − (a − x)2

so that y 2 = b2 − (a − x)2 . Now that we have acquired a better insight into what
sort of triangle we have to look for, let us start all over again. Three positive
numbers a, b, c with a ≥ b ≥ c are given and each of them is less than the sum
2 2 2
of the other two. Let x = a +c2a−b . Then again, x > 0. With this choice of x,
the number c − x is positive, i.e. c > x because:

a2 + c2 − b2
c−x=c−
2a
2ac − a2 − c2 + b2
=
2a
b2 − (a − c)2
=
2a

and since a < b + c, that is, a − c < b, it follows that b2 > (a − c)2 .
We can now construct the desired triangle:
 327

Let BC be a segment with length a. Use the Point Plotting Theorem to

−−→ 2 2 2
determine point D on ray BC such that BD = x = a +c2a−b . Point D will lie
between B and C, since, by assumption, a ≥ b ≥ c, i.e. a2 ≥ b2 ≥ c2 > c2 − b2
2 2 2 −−→
or 2a2 > a2 + c2 − b2 , and hence, a > a +c2a−b = x. On ray DE emanating from
√
D and perpendicular to BC, determine point A such that AD = y = c2 − x2
(Point Plotting Theorem). It then follows that x2 +y 2 = c2 and (a−x)2 +y 2 = b2 .
But AB 2 = x2 + y 2 and (a − x)2 + y 2 = AC 2 . Therefore, AB = c and AC = b
since b and c are positive numbers. Thus 4ABC satisfies our requirements and
the Lemma is proved.

With this Lemma at hand, the proof of the Two-Circle Theorem becomes
quite easy:

Proof. We are given circle C with center P and radius a and a second circle
C 0 with center M and radius b. The distance P M between the centers is c,
and each of the three positive numbers a, b, and c is less than the sum of the
other two. By virtue of the Lemma, there exists a triangle 4RST with RS = a,
ST = b, and RT = c. Let A be a point in the plane of circles C and C 0 on
←−→
one side of line P M such that ]AP M ∼ = ]R (Angle Construction Axiom) and
AP = a = RS (Point Plotting Theorem). Then, 4RST ∼ = 4P AM (S.A.S.),
so that AM = ST = b. Thus, A is on both circles C and C 0 . Now let B be a
←−→
point in the plane of the two circles on the opposite side of line P M from A,
such that ]BP M ∼ = ]R and BP = a = RS. Then 4RST ∼ = 4P BM (S.A.S)
and B is also on both circles. Consequently, circles C and C 0 intersect in at
least two points A and B on opposite sides of their center line. That these
two circles cannot intersect in more than two points has already been shown in
Chapter 9, but in light of the preceding argument, we may supply an alternative
328 APPENDIX E. THE TWO-CIRCLE THEOREM

proof thereof: the center line of two circles divides the plane of the circles into
two disjoint half-planes (Plane Separation Axiom). Hence, if there were a third
point X common C and C 0 , then X would have to lie in one of these two half-
−−→ −→ −−→ −−→
planes, and either rays P X and P A or rays P X and P B would be the same by
virtue of the Angle Contruction Axiom. Next, since P X = P A = P B (Point
Plotting Theorem), this implies that either points X and A or points X and B
must be the same, and there cannot be a third point on both circles C and C 0 .
This completes the proof of the Two-Circle Theorem.
Appendix F

Additional Exercises and

Practice Problems in
Geometry and
Trigonometry

First, some practice problems in Geometry:

←−−→ ←−−→
1. Lines EAB and DCB intersect at B and BC = CA = DA, as shown.
Prove that m ]DAE = 3m ]B.

2. M P is the perpendicular bisector of the side AC of the isoceles triangle

4ABC in which AB = AC. Prove that m ]AP B = 2m ]B.

329
330 APPENDIX F. ADDITIONAL EXERCISES

3. In triangle 4ABC, AB = BC and P and Q are points on sides BC and

AB, respectively, such that AC = AP = P Q = QB. Determine m ]B.

4. In triangle 4P QR, P Q = QR. K is any point on QR, and P R is extended

←→
to point T so that RT = KR. Segment T K is extended to meet line P Q
at S. Prove that m ]QST = 3m ]T .

5. The bisectors of the angles at vertex B and C of the equilateral triangle

4ABC meet at D. Through point D lines are drawn parallel to sides AB
and AC of the triangle, meeting side BC at E and F , respectively. Prove
that BE = DE AND that BE = EF = F C.

6. In triangle 4ABC, AB = AC. Side BA is extended to a point X and

XN is drawn perpendicular to BC, meeting AC at Y , as shown. Prove
that AX = AY .

7. The bisectors of angles ]B and ]C of triangle 4ABC meet at G. GQ

and GR are the perpendiculars from G to sides AC and AB, respectively.
Prove that GQ = GR.
Hint: Draw GP perpendicular to side BC.

8. In triangle 4ABC, AC = BC. Y is an arbitrary point on side BC.

Through Y a parallel is drawn to AC and point X is chosen on this
parallel such that Y X = BY , as shown. Prove that ]ABX is a right
angle.
 331

←→
9. P is a point on line AK. Segments AQ and KR are perpendicular to
AK and AQ = P K and AP = KR. Prove that triangle 4P QR is right
isoceles.

10. In triangle 4ABC, m ]B = 3m ]C. The perpendicular bisector of side

BC meets side AC at point T . Prove that AB + BT = AC.

11. Triangle 4ABC is isoceles with AB = AC. M is the midpoint of side BC.
M X is the perpendicular from M onto AB, and M Y is the perpendicular
frm M onto AC. Prove that XY is parallel to BC.

12. Triangle 4XY Z is isoceles with XY = XZ. The altitude from X meets
side Y Z at A, and a line through A parallel to XY meets XZ at B. Prove
that B is the midpoint of side XZ.

13. ABCD is a quadrilateral with AD = BC, and the angles at the vertices
A and C are right angles. Prove that AD is parallel to BC and that AB
is parallel to DC.

14. In triangle 4ABC, the perpendicular bisector of side BC and the bisec-
tor of angle ]A meet at point P . M is the midpoint of BC. From P
←→ ←→
perpendicular lines are dropped onto lines AB and AC meeting them at
points Y and X, respectively, as shown.
Prove that CX = BY .

15. Perpendiculars are drawn from the vertices B and D of the parallelogram
ABCD to the diagonal AC, meeting it at E and F , respectively. Prove
that the quadrilateral BEDF is a parallelogram.
332 APPENDIX F. ADDITIONAL EXERCISES

16. ABCD is a parallelogram with diagonals meeting at O. X and Y are the

midpoints of AO and OC, and points P and Q are on sides AB and DC,
respectively, such that the three points P , O and Q are collinear. Prove
that the quadrilateral P Y QX is a parallelogram.
17. What is the figure formed by the bisectors of the angles of a rectangle?
JUSTIFY your answer!
18. ABCD is a parallelogram. The bisectors of angles ]A and ]B meet at
point P . Prove that P is equidistant from AD and BC.
19. O is the point of intersection of the diagonals of a square ABCD. K is a
point on side AB such that AK = AO. Determine the measure of ]KOB.
20. Think 3-dimensional and determine whether each of following statements
is always true or always false:

(a) The vertices of a triangle determine a unique plane.

(b) If a line contains two points of a triangle, the line lies completely in
the plane of the triangle.
(c) A set of four points is contained in a unique plane.
(d) If two lines are not skew, they determine a unique plane.
(e) If two planes are perpendicular to the same line, the two planes are
parallel.
(f) If two planes are perpendicular to the same plane, the two planes are
parallel.
(g) Through a given point on a given line, there is one and only one line
perpendicular to the given line.
(h) Through a given point on a given line, there is one and only one plane
perpendicular to the given line.
(i) Through a given point not on a given line, there is one and only one
line perpendicular to the given line.
(j) Through a given point, there is one and only one plane perpendicular
to a given plane.
(k) If a line intersects one of two parallel planes, it intersects the other
plane also.
(l) If a line intersects one of two parallel lines, it intersects the other line
also.
(m) If each of two planes is parallel to a given line, the two planes are
parallel.
(n) If each of two given lines are parallel to a given plane, the two lines
are parallel.
(o) If a line is parallel to a plane, it is parallel to every line in that plane.
 333

(p) Through a given point not in a given plane, there is one and only one
plane parallel to the given plane.
(q) If two lines intersect and each one is parallel to a given plane, then
the plane determined by the two lines is parallel to the given plane.
(r) Two intersecting lines cannot both be perpendicular to a given plane.
(s) If a given line meets a plane Π at point P and makes congruent angles
with each of two lines contained in Π and passing through P , then
the given line is perpendicular to the plane Π.
(t) If a given line meets a plane Π and is not perpendicular to Π, there
is no line in Π which is perpendicular to the given line.
334 APPENDIX F. ADDITIONAL EXERCISES

Now, for something harder:

1. The triangle 4XY Z has a right angle at vertex Z. A is a point on side XZ

and B is a point on side Y Z. Perpendiculars from A and B to side XY
meet XY at M and N , respectively. Prove that AM · BN = XM · Y N .

2. The diagonals of a trapezoid intersect at K. Prove that the ratio of the

distances of K from the bases of the trapezoid is equal to the ratio of the
lengths of the bases.

3. D is a point on the base BC of triangle 4ABC. A line through B parallel

to DA meets CA extended at F , and a line through C parallel to DA meets
◦
BA extended at E. If BA 3
AE = 2 , AD = DC = 6 units, and m ]ACB = 60 ,
calculate the lengths BD, F A and EC.

4. P QRS is a parallelogram with P Q > P S. The bisector of ]P meets QS

at A and RS at B. Prove that P S · P A = P Q · AB.
AP
5. P is a point on the side AB of triangle 4ABC such that AB = 13 . Q is a
CQ 1
point on side BC such that CB = 3 . AQ and CP intersect at X. Prove
that AX 3
AQ = 5 .

6. P and Q are points on sides AB and AC, respectively, of triangle 4ABC

such that P Q is parallel to BC. The median AD meets P Q at M . Prove
that P M = M Q.

7. X is the midpoint of side BC of triangle 4ABC. Y is the point on AC

such that YAYC = 12 . AX meets BY at Z. Prove that BY
ZY
= 14 .

8. A, B, C, D are points on a straight line with AB = BC = CD. On

BC as base an equilateral triangle 4BCP is constructed. Prove that
AP 2 = AB · AD.

9. AD is a median of the triangle 4ABC, and E is a point on AD such that

AQ
AD = 3AE. BE extended meets AC at Q. Find the ratio QC .

10. ABCD is a parallelogram. P is a point on side AB and Q is a point on

side CD, and P Q meets AC at X. An arbitrary line through X meets
BC and AD at R and S, respectively. Prove that the triangles 4P XS
and 4QXR are similar.

11. In triangle 4ABC, AB = AC and m ]A = 36◦ . CP bisects angle ]BCA

and meets AB at P . Prove that the triangles 4BP C and 4BCA are
similar and that BP · BA = BC 2 = AP 2 .

12. Triangle 4ABC has a right angle at vertex C. The bisector of angle ]C
meets side AB at D, and DE is the perpendicular from D to side AC.
1 1 1
Prove that BC + AC = DE .
 335

13. ABCD is a quadrilateral such that the extensions of sides BA and CD

meet at right angles. Prove that BD2 + AC 2 = AD2 + BC 2 .
14. Triangle 4ABC has a right angle at vertex B and AB > BC. Show how
to find a point P on side AB such that AP 2 − BP 2 = BC 2 .
15. A cube has edges of length 8 units and stands with its face ABCD in a
horizontal plane. If M is the midpoint of an edge of the other horizontal
face, find the distance of M to side CD.
16. A triangular pyramid in which all the edges are congruent (and all faces
are congruent equilateral triangles) is called a regular tetrahedron. The
perpendicular from any vertex of the tetrahedron to the opposite face is
called an altitude of the tetrahedron.

(a) Find the length of an altitude of a regular tetrahedron having edges

n units long.
(b) Show that any two altitudes intersect, and that each divides the other
into segments in the ratio 1:3.
(c) What can one deduce about the four altitudes?
336 APPENDIX F. ADDITIONAL EXERCISES

Some trigonometric identities:

sec x sec x+1
1. 1−cos x = sin2 x

tan A−sin A tan A sin A

2. tan A sin A = tan A+sin A

sin C+cos C−1 cos C

3. sin C−cos C+1 = sin C+1

sin2 φ+2 cos φ−1 1

4. 2+cos φ−cos2 φ = 1+sec φ

5. tan λ + cot λ = 2 csc 2λ

6. sin y cos3 y − cos y sin3 y = 1

4 sin 4y

7. 1 + tan C tan C2 = sec C

cos(A+B) 1−tan A tan B
8. cos(A−B) = 1+tan A tan B

cos2 x
9. (1−sin x)2 = (sec x + tan x)2
cot x+csc y
10. tan x+tan x sec y = cot x cot y
1−csc u
11. sin u−1 = csc u

12. cos4 t − sin4 t = cos2 t − sin2 t

1+sin x 1−sin x
13. 1−sin x − 1+sin x = 4 tan x sec x
sin x tan x
14. 1−cos x − sec x = 1
1 1
15. 1+cos x + 1−cos x = 2 csc2 x

16. tan2 x − sin2 x = tan2 x sin2 x

sin x 1+cos x
17. 1−cos x − sin x =0
sin x sin x
18. cot x+csc x − cot x−csc x =2
 337

Now, for some more loosely organized trigonometric problems:

1. If the measures of angles B and A are 165◦ and 75◦ , respectively, calculate
the exact value of sin A + sin B and cos A − cos B.

2. In triangle 4ABC, the ratios of the sides a, b, c are a : b : c = 2 : 4 : 5.

To the nearest degree determine the measure of the largest angle of this
triangle.
(b+c+a)(b+c−a)
3. Prove that in triangle 4ABC: 1 + cos A = 2bc .

4. A vertical tree stands on a slope that is inclined at 10◦ with the horizontal.
When the angle of elevation of the sun measures 25◦ , the shadow of the
tree down the slope is 40 meters long. How tall is this tree?

5. Triangle 4ABC is equilateral with sides that are 18 cm long. Lines AD

and AE are drawn trisecting the vertex angle at A and meeting side BC
at points D and E. Determine the lengths of the segments BD, DE, and
EC.
cos A cos B
6. In triangle 4ABC, prove that if b = a , then the triangle is either
an isoceles or a right triangle.

7. In triangle 4ABC, a = 5 units, b = 8 units, and c = 18 units. Calculate

the measure of angle ]A.

8. In triangle 4ABC, a = 12 units, b = 7 units, and the measure of angle

]C is 63◦ . Calculate the length of c.

9. Each of two diagonals from the same vertex of a regular pentagon is 50

units. Find the length of the pentagon side.

10. One side of a parallelogram has length 40 units. An adjacent side of length
22 units makes an angle of 46◦ with one of the diagonals. Find the length
of this diagonal.
b2 sin A sin C
11. Prove that the area of triangle 4ABC is given by 2 sin B .

12. In triangle 4ABC, a = 7 units, b = 6 units, and the measure of angle ]C

is 24◦ 200 . Calculate the measure of the smallest angle of this triangle.

13. In triangle 4ABC, a = 14 units, b = 18 units, and the measure of angle

]A is 35◦ . Calculate all possible lengths of c and all possible measures of
the other two angles of the triangle.

14. From two points 6,000 meters apart in the plane of the base of a hill, the
angles of elevation of the summit measure 19◦ 100 and 20◦ 300 . If the points
are on opposite sides of the hill but in the same vertical plane with the
summit, what is the height of this hill?
338 APPENDIX F. ADDITIONAL EXERCISES

15. If the hypotenuse of a right triangle is 42 units and the measure of one
acute angle is 37◦ 100 , what are the lengths of the other two sides of this
triangle?
16. Determine the measures of the acute angles of a right triangle with sides
5 units, 12 units, and 13 units.

17. A boy is flying a kite and has let out 300 ft of string. If the string makes
an angle of 35◦ with the ground, how high is the kite?
18. A quadrilateral has sides a = 72 units, b = 58 units, c = 69 units, and
d = 82 units. Determine the lengths of the diagonals of this quadrilateral
if the measure of the angle between the sides with lengths a and b is 68◦ .
Determine the measure of the other angles of this quadrilateral.

And for dessert, some trigonometric equations:

Solve the following equations and give all their solutions in either radians or
degrees (no calculator allowed!).

1. tan2 x = 3
2. 2 sin2 u + sin u = 1
3. sin2 t + 2 sin t − 1 = 0

4. sin y + cos y = 0
5. 4 sin2 x tan x = tan x
6. sin x = sec x
7. 2 tan 3w = sec2 3w
Appendix G

A Potpourri of Nontrivial
Problems

1. If ABCD is a parallelogram, PROVE that vertex D is in the interior of

angle ]ABC.

2. PROVE that the diagonals of a parallelogram intersect.

3. Prove that the sum of the measures of the angles of a triangle is less than
270◦ . Assume that one of the angles of the triangle is obtuse.

4. In 4ABC, AC > AB. Prove that the length of any segment from vertex
A to a point D on side BC between vertices B and C is shorter than AC.

←→ ←→
5. Lines P R and QS are parallel. The bisectors of angles ]RP Q and ]P QS
meet at T . Prove that angle ]P T Q is a right angle.

6. Triangle 4ABC is isosceles with AC = BC. Y is a point on side BC

−−→ ←→
between B and C. Let ray Y Z be parallel to line AC, as shown, and X
−−
→
be a point on ray Y Z such that BY = Y X. Prove that ]ABX is a right
angle.

339
340 APPENDIX G. A POTPOURRI OF NONTRIVIAL PROBLEMS

7. Triangle 4ABC is isosceles with AB = AC. M is the midpoint of BC.

Perpendiculars are dropped from M onto sides AB and AC, meeting these
sides at X and Y , respectively. Prove that XY is parallel to BC.

8. The sum of the measures of the interior angles of a regular polygon is 4

times the sum of the measures of its exterior angles. How many sides has
this polygon?

9. ABCD is a square. Point X is on side AB and point Y is on side BC

such that AX = BY . Prove that DX is perpendicular to AY .

10. O is the point of intersection of the diagonals of square ABCD. K is a

point on side AB such that AK = AO. Find the measure of angle ]KOB.

11. P QRS is a parallelogram. SP is extended to point T such that P T = SP

and SR is extended to point V such that RV = SR. Prove that the points
T , Q, and V are collinear.

12. The equilateral triangles 4ABX and 4BCY are constructed on sides AB
and BC of triangle 4ABC, respectively. If AY meets CX at point P ,
prove that m ]AP X = 60◦ .

13. In triangle 4ABC, m ]A = 80◦ . The bisectors of the exterior angles at

vertices B and C meet at point X. Find the measure of angle ]BXC.

14. In triangle 4ABC, the measure of angle ]A is 60◦ and angle ]B is right.
The bisector of angle ]A meets side BC at X. The altitude from vertex
B meets side AC at Y and AX at E. Prove that XE = 2EY .

15. The bisectors of angles ]B and ]C of triangle 4ABC meet at point K.

Through K a line is drawn parallel to side BC meeting AB at E and AC
at F . Prove that EF = BE + CF .

16. ABCD is a parallelogram. The perpendiculars from A and C meet the

diagonal BD at P and Q, respectively. The perpendiculars from B and D
meet the diagonal AC at X and Y , respectively. Prove that the quadri-
lateral XP Y Q is a parallelogram.
 341

17. In triangle 4ABC, m ]A = 36◦ and angle ]C is right. M is the midpoint

of AB and N is the foot of the altitude from vertex C on AB. Find the
measure of angle ]M CN .

18. P and Q are the midpoints of the sides AB and BC, respectively, of the
parallelogram ABCD. P Q meets DB at point X. Prove that DB = 4XB.

19. Triangle 4ABC has a right angle at vertex C. M is the midpoint of side
AB. A line through C parallel to AB meets the line through A parallel
to M C at K. Prove that M K is the perpendicular bisector of AC.

20. The line drawn through vertex B of triangle 4ABC parallel to the bisector
of angle ]C meets AC extended at X. The line through vertex B parallel
to the bisector of angle ]A meets CA extended at Y . Prove that XY is
equal to the perimeter of 4ABC.

21. AB is a diagonal of a face of a cube. XY is the diagonal of the oppo-

site face, skew to AB. If M is the midpoint of AB, prove that AB is
perpendicular to the plane spanned by the points X, M and Y .

22. A regular tetrahedron is a solid whose faces are four congruent equilateral
triangles. Prove that the perpendiculars from the vertices to the opposite
faces of the tetrahedron are coplanar.

23. Triangle 4ABC is isosceles with AB = BC. BN is a line segment per-

pendicular to the plane of 4ABC. The perpendicular bisector plane of
segment BN intersects AN at P and CN at Q.
Prove that AP QC is an isosceles trapezoid.

24. In triangle 4ABC, AB = AC, and P and Q are the midpoints of sides
AB and AC, respectively. BC is extended to point D such that CD = BC
and P D meets AC at R. Prove that QR = AC 6 .

25. The medians AD and BE of triangle 4ABC meet at G. X is the midpoint

of AG and Y is the midpoint of BG. Prove that AG = 23 AD.

26. The median P S of the trapezoid ABCD meets the diagonals of the trape-
zoid at Q and R. Prove that P Q = RS and that QR is equal to one-half
the difference of the lengths of the trapezoid bases AB and CD.

27. ABCD is a trapezoid in which AB is parallel to DC and AD is perpen-

dicular to AB. If X is the midpoint of BC, prove that AX = DX.
342 APPENDIX G. A POTPOURRI OF NONTRIVIAL PROBLEMS

28. * Triangle 4ABC is isosceles with AC = AB. P is an arbitrary point

on side BC between vertices B and C. The perpendiculars from P onto
sides AC and AB meet these sides at F and E, respectively. Prove that
P F + P E is a constant number independent of the choice of P .
29. * The midpoints of the sides AB, BC, and AC of triangle 4ABC are F ,
G and H, respectively. If BE is drawn perpendicular to AC, prove that
the angles ]F EG and ]F HG are congruent.
30. * Triangle 4ABC is isosceles with AB = AC and m ]A = 120◦ . If the
base BC is trisected by points D and E, prove that triangle 4ADE is
equilateral.
31. * ABCD is a parallelogram. From D a perpendicular is dropped on
diagonal AC meeting it at R. The parallel from B to AC meets DR
extended at N . AN intersects BC at P and DN meets BC at Q. Prove
that P is the midpoint of BQ and that AR = BN + RC

32. * Triangle 4ABC has a right angle at vertex A and AB > AC. Let
AH and AM be the altitude and the median, respectively, from A to the
hypotenuse BC. Prove that m ]M AH = m ]C − m ]B. Use this result
to show how to construct a right triangle 4ABC, given the length of its
hypotenuse BC and the difference of the measures of its two acute angles.
33. Triangle 4XY Z has a right angle at vertex Z. A is a point on side
XY and B is a point on side Y Z. Points M and N are the feet of the
perpendiculars from A and B to side XY , respectively.
Prove that AM · BN = XM · Y N .
34. D is a point on the base BC of triangle 4ABC. A line through B parallel
to DA meets CA extended at F , and a line through C parallel to DA meets
◦
BA extended at E. If BA 3
AE = 2 , AD = DC = 6 units, and m ]ACB = 60 ,
calculate the lengths BD, F A and EC.
35. P QRS is a parallelogram with P Q > P S. The bisector of angle ]P meets
QS at A and RS at B.
Prove that P S · P A = P Q · AB.
AP
36. P is a point on side AB of triangle 4ABC such that AB = 13 . Q is a
CQ 1
point on side BC such that CB = 3 . AQ and CP intersect at X. Prove
that AX 3
AQ = 5 .
 343

37. X is the midpoint of side BC of triangle 4ABC. Y is the point on AC

such that YAYC = 12 . AX meets BY at Z. Prove that BY
ZY
= 14 .

38. A, B, C, D are collinear points with AB = BC = CD. On BC as base an

equilateral triangle 4BCP is constructed. Prove that (AP )2 = AB · AD.

39. AD is a median of triangle 4ABC, and E is a point on AD such that

AQ
AD = 3AE. The extension of BE meets AC at Q. Find the ratio QC .

40. In triangle 4ABC, AB = AC and m ]A = 36◦ . The bisector of angle

]BCA meets AB at P . Prove that the triangles 4BP C and 4BCA are
similar and that BP · BA = (BC)2 = (AP )2 .

41. * Triangle 4ABC has a right angle at vertex C. The bisector of angle ]C
meets side AB at D, and DE is the perpendicular from D to side AC.
1 1 1
Prove that BC + AC = DE .

42. ABCD is a quadrilateral such that the extensions of sides BA and CD

meet at right angles.
Prove that (BD)2 + (AC)2 = (AD)2 + (BC)2

43. Triangle 4ABC has a right angle at vertex B and AB > BC. Show how
to find a point P on side AB such that (AP )2 − (BP )2 = (BC)2

44. A cube has edges of length 8 units and stands with its face ABCD in a
horizontal plane. If M is the midpoint of an edge of the other horizontal
face of the cube, find the distance of M to side BD.

45. A triangular pyramid in which all the edges are congruent (and all faces
are congruent equilateral triangles) is called a regular tetrahedron. The
perpendicular from any vertex of the tetrahedron to the opposite face is
called an altitude of the tetrahedron.

(a) Find the length of an altitude of a regular tetrahedron having edges

s units long.
(b) Show that any two altitudes intersect, and that each divides the other
into segments in the ratio 1:3. What can be deduced about the four
altitudes?

46. ABCD is a parallelogram in which the bisector of angle ]C meets side AB

at its midpoint M . Prove that DM bisects angle ]ADC and determine
the measure of angle ]DM C.

47. AD and BE are altitudes of triangle 4ABC. M is the midpoint of side

AB. Prove that the triangle 4M ED is isosceles.

48. AD is an altitude of the equilateral triangle 4ABC. DN is the perpen-

dicular from D to side AC. Prove that AN = 3N C.
344 APPENDIX G. A POTPOURRI OF NONTRIVIAL PROBLEMS

49. Prove that the lines joining the midpoints of opposite sides of a convex
quadrilateral bisect each other.

50. ABCD is a convex quadrilateral in which AB = CD. P , Q, R, S are the

midpoints of segments AD, AC, DB, BC, respectively. Prove that P S is
perpendicular to QR.

51. The medians BE and CF of triangle 4ABC meet at point G and are
perpendicular to each other. Prove that AG = BG.
−−→ −−→
52. In the diagram shown, BN is perpendicular to both rays AX and BY .
−−→ ←→
Point Q is on ray BY and point P is chosen on line AQ between A and
Q such that P Q = 2AB.
Prove that m ]QBA = 3m ]QAX.

53. D, E, F are the midpoints of sides BC, CA, AB of triangle 4ABC, and
G is its centroid. EF meets AD at point M . What fraction of AD is
M G?

54. M is the midpoint of side AB of triangle 4ABC. The line through M

parallel to side BC meets the bisector of angle ]B at point X. Prove that
angle ]AXB is a right angle.

55. Triangle 4ABC is inscribed in a circle. The tangent at A to the circle is

←→
drawn and line XY is drawn parallel to this tangent meeting side AB at
X between A and B and side AC at Y between A and C. Prove that the
quadrilateral XY CB is cyclic.

56. Triangle 4ABC is an inscribed isosceles triangle with AB = AC. P and

Q are arbitrary points on arc BC,
˜ and AP and AQ meet BC at points R
and S, respectively. Prove that the points P , Q, R, and S are concyclic.

57. Triangle 4ABC is an inscribed triangle. The extensions of the altitudes

BE and CF of the triangle meet the circle at P and Q, respectively. Prove
that P Q is parallel to EF .

58. A, B, C are points on a circle such that AB = AC. P and Q are points
on arc AC
˜ such that CP = P Q, and the extensions of BP and AQ meet
at point R. Prove that AB = AR.
 345

59. Two circles are externally tangent at A. A common external tangent meets
the circles at P and Q. Let QR be a diameter. Prove that the points P ,
A, and R are collinear.

60. Triangle 4ABC is inscribed in a circle. The altitude AD to side BC is

extended to meet the circle at E. Point H is chosen on AD between A
and D such that HD = DE. Prove that BH extended is perpendicular
to AC.

61. Triangle 4ABC is inscribed in a circle. DE is the diameter perpendicular

to side BC. Prove that AD and AE are the bisectors of the interior and
exterior angles of the triangle 4ABC at vertex A.

62. The inscribed circle of triangle 4ABC is tangent to the sides AB, BC, AC
at points X, Y , Z, respectively. Another circle is tangent to AB extended
at U , to BC at V , and to AC extended at W . Prove: BV = CY .

63. In the diagram shown, A is a point on the tangent at T to the circle such
that T A = 15 units. The perpendicular at A to this tangent meets the
circle first at point B. If AB = 9 units, determine the radius of the circle.

←−−→
64. In the diagram shown, XP Y is a tangent to the left-hand circle, and
←−−−→ ←−−−→
P CDA and P EF B are straight lines such that AB is parallel to line
←−−→
XP Y .
Prove that CE and F D are parallel.
346 APPENDIX G. A POTPOURRI OF NONTRIVIAL PROBLEMS

65. Triangle 4ABC has a fixed side AB. If BC < AC, what is the locus of
vertex C?
66. Show how to find a point which is equidistant from two given parallel lines
←→ ←→ ←→
P Q and RS and lies on a given third line AB. Is it always possible to find
such a point?
67. A, B, C are fixed collinear points with B between A and C, and AB = 2
units. A point P is such that m ]P BC = 2m ]P AB. What is the locus
of P ?
68. AB is a fixed diameter of a circle with radius 3 units. C is an arbitrary
point on the circle. Find the locus of the point of the centroid of triangle
4ABC as C moves round the circle.
−−→ −−→
69. Rays OX and OY are perpendicular. A segment AB with length 6 units
−−→ −−→
has its endpoints A and B on rays OY and OX, respectively. Find the
−−→
locus of the midpoint of segment AB as its endpoints move on OX and
−−→
OY .
70. The fixed point O is at distance 3 units from the fixed plane Π. What is
the locus of all points which are at distance 4 units from O and at distance
1 unit from Π?
71. P is a fixed point in the exterior of a fixed sphere with center O. The
segment OP meets the circle at point A. If C is any point on the sphere
other than A, show that P A < P C.
72. Line l is parallel to plane Π and at distance 5 units from it. What is the
locus of all points which are at distance 6 units from l and at distance 1
units from Π?
73. A and B are fixed points in the fixed plane Π. Points C and D are any
points in Π such that ABCD is a parallelogram with BC = 21 AB. What
is the locus of the midpoint P of CD?
74. A and B are fixed points in the fixed plane Π, and l is any line in Π
containing point A. P is a point in Π such that l is the perpendicular
bisector of segment BP . What is the locus of P as l takes all possible
positions in Π?
75. In triangle 4ABC, AB > AC. Show that there is exactly one point in
the interior of angle ]A which is equidistant from vertices B and C and
←→ ←→
also from the lines AB and AC. Is it possible for this point to lie in the
interior of triangle 4ABC?
−−→ −→ −−→ −→
76. AB and AC are fixed rays. P and Q are arbitrary points on AB and AC,
respectively. The bisectors of the angles ]BP Q and ]CQP meet at R.
What is the locus of R as P and Q take all possible positions?
Appendix H

Mathematical Induction

There is no largest natural number because when we add 1 to a natural number

k we obtain the natural number k + 1 which is larger than k. Nevertheless,
starting with the number 1, one can reach any natural number in a finite num-
ber of steps, passing from k to k + 1 at each step. The method of proof by
mathematical induction which can be used to establish the general validity of
certain relations/formulae which depend on a natural number is based on this
fact. Here is how this method works.
Suppose that A(n) is some assertion involving the natural number n. For
instance, A1 (n) could mean that the sum of the first n natural numbers is equal
to n(n+1)
2 , in symbols: 1 + 2 + 3 + . . . + (n − 1) + n = n(n+1)
2 . Another assertion
A2 (n) could mean that the sum of the squares of the first n natural numbers
3 3
is larger than n3 : 12 + 22 + . . . + n2 > n3 . Yet a third example A3 (n) is the
assertion that the number of distinct diagonals of a convex n-gon is given by the
formula n(n−3)2 for n ≥ 3. All these assertions could be verified by choosing an
arbitrary value for n and checking explicitly that the stated assertion is indeed
true for that value of n. But this verification does not constitute a proof of the
validity of the assertion for ALL values of n.
Now suppose that assertion A(n) is verified by direct calculation for the
value n = 1, i.e. A(1) is true. Assume further that the assertion has been
proved for a particular value of n, say for n = k, i.e. assume that A(k) is
true. If, using the validity of A(k) one can prove that the assertion remains
valid for n = k + 1, i.e. if one can prove that A(k + 1) is true under the
assumption that A(k) is true, then the assertion A(n) would be true for all
natural numbers n. Namely, since A(1) has been verified directly, one concludes
that A(2) is also true. Knowing that A(2) is true, one concludes that A(3) is
true, and so on. Since every natural number can eventually be reached by a
series of steps of this sort, A(n) is true for all natural numbers. It is important
to understand the pattern of this proof. First, one proves the validity of the
assertion A(n) for n = 1. Next, one shows that IF the assertion is true for
one natural number (induction assumption), THEN it is also true for the next

347
348 APPENDIX H. MATHEMATICAL INDUCTION

natural number (induction step). From this pattern, one concludes that the
assertion is true for all natural numbers.
The idea of mathematical induction can be illustrated in many ways. For
instance, imagine a row of standing dominos, numbered consecutively starting
from 1, that extends, say, to the right with without end. Suppose the dominos
are closely arranged, so that if one of them, say the one labeled k, falls to the
right, it will knock over the next one, labeled k + 1. Then anyone can visualize
what would happen if domino 1 were toppled to the right. It is also clear that
if a farther domino, say the one numbered j, were knocked over to the right,
then all dominos to its right, labeled j + 1, j + 2, . . . would equally fall.
Let us apply this method to the proof of the assertion A2 (n): 12 + 22 + 32 +
3
. . . + n2 > n3 .
3
First, A2 (1) is evidently true, since 12 > 13 . Next, assume A2 (j) is true for
3
some natural number j > 1: 12 + 22 + 32 + . . . + j 2 = j3 (induction assumption).
Then, adding (j + 1)2 to both sides of the previous inequality yields:

j3
12 + 22 + 32 + . . . + j 2 + (j + 1)2 > + (j + 1)2
3
j3
= + j 2 + 2j + 1
3
1
= (j 3 + 3j 2 + 6j + 3)
3
But (j +1)3 = j 3 +3j 2 +3j +1 and clearly j 3 +3j 2 +6j +3 > j 3 +3j 2 +3j 2 +1,
so that we can write: 12 +22 +. . .+j 2 +(j +1)2 > 31 (j 3 +3j 2 +6j +3) > 31 (j +1)3 ,
implying that A2 (j + 1) is true (induction step). Hence, A2 (n) must be true for
all natural numbers.
We now prove by mathematical induction that the number Dn of distinct
diagonals of a convex n-gon is given by the formula: Dn = n(n−3) 2 , for n ≥ 3.
For n = 3 (a “convex” triangle), the formula yields D3 = 0 which is true
since a triangle has no diagonal. We can even verify that for n = 4 (convex
quadrilateral), the formula gives D4 = 2 which is the correct number of diagonals
of a convex quadrilateral. Assume next that the number of distinct diagonals of
a convex k-gon is Dk = n(n−3)2 (induction assumption). We then prove below
that under this assumption, the number of distinct diagonals of a convex
(k + 1)-gon is given by Dk+1 = (k+1)[(k+1)−3]
2 . To this end, consider a convex
k-gon, and pick an arbitrary point in the exterior of this convex k-gon. Note
that the Plane Separation Axiom and the definition of convexity determine
the interior and exterior of a convex polygon unambiguously. Label this point
as the (k + 1)st vertex of a convex (k + 1)-gon obtained by augmenting the
number of vertices of the original convex k-gon by one vertex. We now focus on
this (new) (k + 1)st vertex: connecting it to its two nearest neighboring (old)
vertices yields two sides of the (k + 1)-gon whereas connecting it to the other
remaining k − 2 (old) vertices creates (k − 2) new diagonals that did not exist
in the original k-gon. Furthermore, connecting those mentioned two nearest
 349

neighboring (old) vertices together also creates a new diagonal for the (k + 1)-
gon. Thus, the correct number of new diagonals in the convex (k + 1)-gon is in
fact (k − 2) + 1 = k − 1, and the total number of diagonals of this (k + 1)-gon
becomes:

k(k − 3)
Dk+1 = Dk + k − 1 = +k−1
2
k 2 − 3k + 2k − 2 k2 − k − 2
= =
2 2
(k + 1)(k − 2)
=
2

so that indeed Dk+1 = (k+1)[(k+1)−3]

2 , which was to be proved.
To conclude this Appendix, a couple of exercises are suggested:
n(n+1)
1. Prove assertion A1 (n): 1 + 2 + 3 + . . . + n = 2 .
2. Use mathematical induction to solve Exercise # 10 in Chapter 5 rigorously.
3. Prove by mathematical induction that the sum of the measures of the
interior angles of a convex n-gon (n ≥ 3) is equal to (n − 2)180◦ .

4. What is wrong with the following “proof” by mathematical induction that

all blonde girls have blue eyes?
It is obvious that there exists (at least) one blonde girl with blue eyes.
Assume that k blonde girls have all blue eyes (induction assumption), and
consider a set of k + 1 blonde girls. Remove one girl from this set. Then,
by the induction assumption, all the remaining k blonde girls have blue
eyes. Remove another girl from this set and replace her with the first
removed girl. Then again, all remaining k blonde girls will have blue eyes.
Repeat this procedure until by exhaustion the k + 1 sets of k blonde girls
will be shown to consist of blonde girls with blue eyes (induction step).
Therefore, the eyes of the girls in any set of k + 1 blonde girls must by
blue.