INTRODUCTION

Those who assert that the mathematical

sciences say nothing of the beautiful or the good
are in error.

– Aristotle

There are at least four reasons to study geometry.

(1) GEOMETRY IS FULL OF WONDERS. At every level of
this science, from the most elementary to the most advanced, we are con-
fronted with the unexpected. Often, the seemingly possible proves impos-
sible, and conversely what at first seemed impossible turns out to be pos-
sible. An example from the most elementary level is the possibility of
one rectangle having more area than another one, and yet less total length
around its sides. At first, it sounds impossible to enclose a bigger yard
with less fence, or a smaller yard with a longer fence. (Incidentally, that
is why swindlers in ancient times used to sell land by perimeter instead
of by area.)

i
 C

B Now a more advanced example. Take any

g
quadrilateral ABCD, as ugly and random-
e
looking as you please. Cut each side in
A half at e, f, g, h, and – surprise!– the quad-
h
D
rilateral e f g h is a perfect parallelogram.

B
And a slightly more advanced example. If you
X take any triangle ABC and drop perpendiculars
C from each vertex to its opposite side, these three
A perpendiculars all meet at one point, X ...

And if you join each vertex of triangle ABC to

Y

C
the midpoint of the opposite side, these three
lines all meet at one point, Y ...
A

B Finally, if you draw perpendiculars from the

midpoints of the sides of ABC, they also all
Z
C meet at one point, Z. Now the surprise: in
A any triangle ABC, these three points X, Y, Z are
in a perfectly straight line. (Not only that, but XY is exactly double the
length of YZ!)

ii
There are many more surprises than these in geometry, but to get into
them would take us out of this introduction, and into the science itself.

(2) GEOMETRY IS BEAUTIFUL. There are at least three

sources of beauty in geometry. First, there are the figures themselves:
perfect things of their kind, without bump or wrinkle, such as a perfect
circle, or the five Platonic Solids. Symmetry and proportion, which are
universal principles of beauty in nature, architecture, poetry and music,
abound in geometric diagrams. There is also a beauty in the truths of
geometry themselves. For example, if you take any triangle you like (Δ
ABC ), and cut each of its angles into three equal parts, the six trisecting
lines you have drawn will meet each other inside the triangle at three
points (D, E, F). The beautiful thing is
A

that the three points D, E, F will be

the vertices of a perfectly equilateral tri-
angle. Such revelations are not only
E
F
surprising, but pleasing in their simplici-
D
ty and symmetry. In geometry, order
B C
pops up unlooked for; a beauty that we
do not make, but only discover.
The proofs of geometry can also be beautiful. The best geometric
proofs are adorned with a brilliance all their own in virtue of their inge-
nuity, clarity, universality, and rigor. Geometry, properly presented, yields
an experience of intelligible beauty, introducing minds to the special
pleasures attending insight and understanding.

iii
 (3) GEOMETRY IS FULL OF FUNDAMENTALS. Over the en-
trance to Plato’s Academy there hung a sign which read Let no one igno-
rant of mathematics enter here. Why did a school of philosophy desig-
nate mathematics as a prerequisite for admission? Plato saw that many
universal principles are most readily accessible to us through mathematics.
The geometrical science of proportion, for example, shows in a con-
crete way how some things can be known by proportions or analogies.
We can come to know an unknown quantity x if we see it in proportion
to other terms already known to us, say if x has to 4 the same ratio that
3 has to 2. Knowing 4, 3, and 2, and knowing the relationship between
3 and 2, we can come to know the mysterious x. This is a useful way
of getting at something to which we have no direct access, say if x were
a length we could not measure directly, like the height of an Egyptian
pyramid. There is no way to drop a plumb line from the peak of a pyr-
amid straight down to its base, but three other lengths that we can meas-
ure might form a proportion with the inaccessible height. Philosophers
and scientists, too, must sometimes find ways to investigate things not di-
rectly observable or imaginable, and one tool for this purpose is propor-
tion or analogy, the most fundamental use of which we find in geometry.
Geometry is also fundamental in another way. It is the science
most easily acquired by the human mind with rigor and exactness. In
geometry, one can settle disagreements. One can draw inescapable con-
clusions. This makes geometry an ideal entryway into the whole life of
the mind.

(4) GEOMETRY EXERCISES THE MIND. People exercise their

bodies to maintain their strength and health, and also because it feels

iv
good. There is such a thing as mental exercise, too, which both strength-
ens and exhilarates the mind. Studying geometry is among the best of
mental workouts, simultaneously exercising one’s imagination, memory, and
reason. In the course of a proof, the imagination must follow a line of
reasoning from one part of a diagram to another; it must flip, rotate, and
otherwise manipulate geometrical objects; it must interpret two-dimensional
diagrams of three-dimensional things; it must picture how the other parts
of a diagram are affected if one part is moved or changed. Memory also
gets a workout, since geometry is cumulative. Each conclusion must be
understood, and then used to establish later results, which in turn help to
establish still more advanced results. And geometry obviously exercises
reason. There is no reasoning more exact than a mathematical argument.
Geometrical objects are perfect subject matter for forming definitions and
proofs, proposing difficulties and finding resolutions, drawing distinctions,
finding examples … in short, for doing all the best things that human
reason can do. Thus geometry builds people’s confidence that reason can
find satisfying answers to serious questions.

For the above reasons geometry is justly recognized as an essential

element in the formation of every educated person and is worthy of life-
long study. Current books written on the premise that geometry is inter-
esting in itself are largely intended for advanced students or professional
mathematicians. They presuppose a mastery of elementary theorems. On
the other hand, geometry books which begin at the very beginning are
generally not written for enthusiastic readers, but for students who need to
pass an exam. Such introductions gloss over proofs (or skip them entire-

v
ly), emphasizing instead various formulas, exercises, and problem-solving
techniques.
This course is written for anyone motivated to study geometry for
the wonder and beauty of it, for readers disposed to contemplate theorems
as if they were works of art. And yet it begins at the very beginning.
To master it, you need no prior training in mathematics. In consequence,
this course represents a unique introduction to geometry. Readers interest-
ed in learning mathematics will find it better suited to their needs than
study manuals or high school geometry books because of its scope, its pu-
rity, and its rigor.
THE SCOPE OF THIS COURSE by far surpasses that of the typi-
cal introduction. This course covers most of the content of the thirteen
books of Euclid’s Elements, whereas typical introductions do not cover
material much beyond the first three or four books of Euclid. Written
most often for the high school level, they do not go deep enough into ge-
ometry to reach the most beautiful and exciting material accessible to rec-
reational mathematicians. Yet this course is not longer than the average
high school textbook, but actually shorter, since it does not multiply exer-
cises.
THE PURITY OF THIS COURSE should be refreshing to anyone
who loves geometry. Other introductions to the science, written so read-
ers can “get the right answer,” employ algebra, trigonometry, number
lines, a system of coordinate axes, and a host of other devices. Such de-
vices and techniques, though useful (elsewhere) and important to study
(elsewhere), have no place in a formal introduction to geometry intended
for those who wish to begin at the beginning and understand the reasons
for things. The impression is given that there is no geometry without

vi
these extras. The truth is that geometrical things can be known geometri-
cally, without recourse to algebra or trigonometry.
The proof of the Pythagorean Theorem given in this book, for ex-
ample, makes no use of algebraic operations. The theorem is demonstra-
ble on purely geometrical grounds. The proof given for this theorem in
many introductory books is an algebraic one that quickly leaves behind
the diagram altogether. The result is a very abstract and unmemorable
proof, the steps of which are not explicitly correlated with the right trian-
gle and the squares that the geometrical theorem is about. The purpose
of teaching the Pythagorean Theorem algebraically is to encourage profi-
ciency in applying it to problems. This denies students any real under-
standing of the theorem, however, and reinforces the idea that geometry
has no intrinsic worth or beauty.
RIGOR. Many introductory books use theorems they do not prove,
such as the theorem that if a cone and a cylinder stand on the same cir-
cle and have the same height, the volume of the cone is one third that of
the cylinder. Current high school textbooks including this theorem or a
formula based on it do not attempt even a sketchy proof for it. In this
course a complete proof is given for this theorem and for every other the-
orem covered. Once again, the implicit message of the textbook is that
understanding the theorem is not important, but only the use of a formula
which one should be willing to take on faith. This presumes an audience
uninterested in the reasons for things, or incapable of understanding them.
Like a novel, it is essential to read this book in the order in which
it is written, but unlike a novel, you can stop after any chapter or theo-
rem and come away with something completely understood. But enough
of introductions. On to the adventure of geometry.

vii
SYMBOLS

Here are the symbols and abbreviations I will use:

A = B A is equal to B
A > B A is greater than B
A < B A is less than B
AB ⊥ CD AB is perpendicular to CD
AB ║ CD AB is parallel to CD
60° sixty degrees
∠ABC angle ABC
rABC triangle ABC
rABC ≅ rDEF triangle ABC is congruent to triangle DEF
£AB the square on line AB
£ABCD the square with corners A, B, C, D
AB · CD the rectangle with sides of length AB and length CD
3A three times A
A : B = C : D A has to B the same ratio that C has to D
A : B > C : D A has to B a greater ratio than C has to D

Q.E.D. Short for Quod Erat Demonstrandum, a Latin expres-

sion meaning “that which was to be demonstrated,” a
customary way of marking the end of a demonstrative
theorem.

viii
Q.E.F. Short for Quod Erat Faciendum, a Latin expression
meaning “that which was to be done,” and a customary
way of marking the end of a construction or “how to”
theorem.

ix
 Chapter One

Triangles
and
Parallelograms
 DEFINITIONS

W 1. A SOLID is whatever has length, width, and

L depth.
H
A gold brick is a solid, having a length
HL, a width HW, and a depth HD.

2. A solid stops at its SURFACE (or surfaces); so a surface has length and width, but no
depth.
The top face of a gold brick is a surface, having a length HL, and a width HW.
But it has no depth, since it is only the
W face of the brick; if it had any depth, it
L
would not be the top face of a brick, but
H it would be a brick itself, even if a very
slim one.

3. When a surface comes to an end, it stops at a LINE (or lines); so a line has length, but
no width or depth.
One edge of the brick's top surface, such as HL, is a line, having a certain length.
But it has no width or depth. It has no depth, since it is an edge of a surface, which has
no depth. It has no width, since it is only the edge of the surface; if it had any width, it
would not be only the edge of a surface, but it would be a surface itself, even if a very
narrow one.
Today a finite line is often called a “line segment.” But since infinite lines don’t
come up too frequently, and it is tedious to say “line segment” every time a finite line is
meant, I will call a finite line simply a “line.”

4. When a line comes to an end, it stops at a POINT; so a point has no length, no width,
no depth.
One end of the edge of the brick's top surface, such as H, is a point, having no
length, width, or depth. It has no width or depth, since it is the end of a line, which itself
has no width or depth. It has no length, since it is only the end of a line; if it had any
length, it would not be only the end of the line, but it would itself be a line, even if a very
short one. Although it has no shape or size, a point does have one positive feature: its
location.

2
5. A STRAIGHT LINE is a perfectly uniform line. Every part of it is the same “shape”
as every other part, regardless of length, and different straight lines differ only in length,
location, and orientation. Every other kind of line is called a CURVED LINE.
One imaginative way to express the uniformity of a straight line is with the
following “thought experiment.” If you look at a straight line on end, it will look like a
point; that is, none of the line’s length will be visible. Another way to express the
uniformity of straight lines is like this: it is impossible for both endpoints of one straight
line to coincide with those of another, without the straight lines themselves completely
coinciding.
A straight line is sometimes defined as the shortest distance between two points.
We will see more about the “shortness” of straight lines in Theorem 17. In the
meantime, it is enough to note that the reason a straight line is so short and direct
compared to other lines is because of its uniformity.
NOTE: Since this book is only about straight lines and circles, and no other kinds
of lines, when I say “line” I will mean one of these, and which one I mean will be clear
by context.

6. A FLAT SURFACE is a perfectly uniform surface. Every part of it, regardless of

size, is the same “shape” all over and on both sides, with no difference of “terrain”
between one part and another. A flat surface can also be called a PLANE.
One imaginative way to express the uniform terrain of a plane is with the
following thought experiment. If you look at a plane on edge, it will look like a straight
line; that is, none of the plane’s width will be visible.

A 7. When two distinct lines in the same plane meet at a

point, the inclination of one to the other is a PLANE
ANGLE. The point at which the lines meet is called the
VERTEX of the angle.
Angle ABC is an example. Since the line AB has
an inclination to line BC on one side of itself, 1, but
B C also another inclination to (or away from) line BC on
the other side of itself, 2, we may speak of an “interior” angle 1, and an “exterior” angle
2. By “angle ABC,” I will mean inclination 1.

D
8. A RECTILINEAL ANGLE is an angle formed by two
different straight lines.
Angle DEF is an example.
E F

3
 A
9. When one straight line stands on another in such a way that
the two adjacent angles formed are equal to each other, each
angle is called a RIGHT ANGLE.
1 2 Thus if AB stands on CD making angle 1 equal to angle
C B D 2, then each of these angles is a “right” angle.
A

10. A straight line standing on another at right angles is said to

be PERPENDICULAR to the line on which it stands.
AB is perpendicular to CD, for example.
C B D

A 11. An OBTUSE ANGLE is a rectilineal angle greater than a

right angle. An ACUTE ANGLE is a rectilineal angle less than
E a right angle.
Angle CBE is acute, being less than angle CBA, which
C B D
is right. Angle EBD is obtuse, being greater than angle ABD,
which is right.

12. An obtuse angle and an acute angle are called SUPPLEMENTARY when they add
up to two right angles. Two acute angles are called COMPLEMENTARY when they add
up to one right angle.
Angles CBE and EBD are supplementary; angles CBE and EBA are
complementary.

13. A BOUNDARY of a thing is its limit, or where it stops.

For example, a sphere is bounded by one surface, a square by four lines, a
straight line by two points.

14. A FIGURE is something contained by its

boundary or boundaries – something which cannot
F G be entered or departed from without cutting across
its boundary or boundaries.
A straight line, such as FG, though it is
K
bounded by two endpoints, is not a figure, since it is
possible to pass through the line without passing
through the endpoints which bound it. But a
triangle, such as HKL, is a figure, since it is
impossible to pass into the triangle or out of it
H L without passing through one of the straight lines
that bound it.

4
 D 15. A CIRCLE is a plane figure contained by one curved
E
line whose every point is the same distance from a single
point inside it.
C
Figure ADEBFA is a circle, since it is bounded by
A
B
the one line ADEBFA, and all the points on that line, such
as A,D,E,B,F, are the same distance from C; that is, CA =
F CD = CE = CB = CF.

16. The single point inside a circle which is the same distance from every point along the
curved line bounding the circle is called the circle's CENTER. And the curved line
bounding the circle is called the circle's CIRCUMFERENCE.
C is the center of the circle ADEBF, and ADEBFA is its circumference.

D E 17. Any straight line drawn from the center of a circle and
stopping at the circumference is called a RADIUS of the
A
C
B circle. Thus, by definitions 15 & 16, it is evident that all the
radii of a circle are equal.
The plural of “radius” is “radii.”
F
Any straight line drawn through the center of a circle
and terminated at each end by the circumference is called a DIAMETER of the circle. It
is evident that any diameter of a circle bisects the circle, that is, cuts the circle into two
equal parts.
CD is an example of a radius of circle C, and ACB is an example of a diameter of
circle C.

18. A SEMICIRCLE is the figure contained by a circle’s diameter and the circumference
cut off by it.
The figure contained by ADEB and the straight line AB is an example of a
semicircle.

19. A RECTILINEAL FIGURE is a plane figure contained by straight lines only.

A TRIANGLE is a plane figure contained by three straight lines.
A QUADRILATERAL is a plane figure contained by four straight lines.
A POLYGON is any plane figure contained by more than four straight lines.

5
20. Among triangles, an EQUILATERAL TRIANGLE is one with all three sides equal.
An ISOSCELES TRIANGLE is one with only two sides equal. A SCALENE
D F TRIANGLE is one without any equal sides.
B H
ABC is an equilateral triangle, DEF
is an isosceles triangle, GHK is a scalene
A C
triangle.
G K

N Q
R S
21. Among triangles, a RIGHT
TRIANGLE is one containing
a right angle. An OBTUSE
TRIANGLE is one containing M L P
O
T
an obtuse angle. An ACUTE
TRIANGLE is one with all three of its angles acute.
LMN is a right triangle, OPQ is an obtuse triangle, RST is an acute triangle.
In a right triangle, the side opposite the right angle is called the HYPOTENUSE, and the
two sides containing the right angle are called LEGS.
NL is the hypotenuse of triangle LMN; MN and ML are its legs.

22. If two straight lines lie in one plane together, but never meet each other in either
direction however far they are extended, they are
A B
said to be PARALLEL to each other.
C D AB and CD are parallels.

E F
23. Among quadrilaterals, a SQUARE is a quadrilateral
with all four sides equal and all four angles right; a
H G RECTANGLE is a quadrilateral with all four angles right
K L but not all four sides equal; a RHOMBUS is a
N M
quadrilateral with all four sides equal but no right angles.
O P
EFGH is a square, KLMN is a rectangle, OPQR
is a rhombus.
R Q A PARALLELOGRAM is any quadrilateral
S T contained by two pairs of parallel straight lines.
Quadrilaterals that are none of the above will be called
V U TRAPEZIA.
X
Y STUV is a parallelogram, even though it has no
right angles and not all of its sides are equal. WXYZ is a
W Z trapezium.

6
24. Two straight lines are INCLINED TOWARD EACH OTHER if, when cut by a third
straight line, the sum of the two interior angles on one side is less than the sum of the two
exterior angles on that same side.
For example, if 1 + 2 is less than 3 + 4, then AB and
3
A
CD are inclined to each other toward the right.
1
B

2 D

C 4

7
 GEOMETRICAL POSTULATES

1. A straight line can be drawn from any point to any point.

A B For example, from A to B.

2. Any straight line can be extended continuously in a straight line in either direction
and as far as you please.

For example, AB can be extended to C,

A
B
C in such a way that ABC is one straight
line.

3. A circle can be drawn around any point as its center and with a radius of any
given length.

L
For example, around point P we may draw a circle whose radius
shall be any given length, such as PL.
P

4. All right angles are equal.

Not only are adjacent right angles equal to each other, such as 1
4 and 2, but even those that are not adjacent, such as 1 and 3.
3
1 2

5. Straight lines inclined towards each other eventually meet, when extended far
enough.

A 3
For example, let A and B be two straight lines that cut
1 across another straight line C, making angles 1 & 2
C
to the right of C. If 1 + 2 is less than 3 + 4, then A
and B must eventually meet on the right side of C at
2
B
4

some point.

8
 GENERAL PRINCIPLES

1. Things equal to the same thing are also equal to each other.

For example, 36 inches is equal to a yard, but 3 feet is also equal to a yard, so 36 inches
has to be equal to 3 feet.

2. When equals are added to equals, the wholes are equal.

For example, if Fred and Jack are both exactly five feet tall, and this summer each will
grow the exact same amount, then their heights will still be equal.

3. When equals are subtracted from equals, the remainders are equal.

For example, if two brand new pencils have exactly the same length, and then you
sharpen them down by exactly the same amount, they will still have equal lengths.

4. Things that can be made to coincide with each other are equal.

That is, if two things are such that neither one goes outside of the other even a little bit,
then neither is greater than the other, and so they must be equal. For example, if the
bottom surface of a box fits exactly on the surface of a table, nowhere hanging over the
edge or letting any part of the table's surface show, then the bottom of the box and the
surface of the table have equal areas.

5. Every whole is greater than any one of its parts.

The surface area of a whole lake is greater than the surface area of any part of it; fifty
dollars is greater than any part of fifty dollars, etc.

9
 THEOREMS

THEOREM 1: How to make an equilateral triangle.

If someone gives us a straight line AB, how can we make an equilateral triangle on top of
it so that AB is the base of the triangle?
As follows.

[1] Draw circle X around point A

C with radius AB (by Postulate 3).
X Z
[2] Draw circle Z around point B
with radius BA (by Postulate 3).
[3] These two circles obviously
A B
must intersect each other, namely
at points C and D.
[4] Join points A and C with a
straight line (Postulate 1).
[5] Join points B and C with a
D straight line (Postulate 1).

So we have made a triangle, ABC.

It is in fact equilateral.
Why? Because:
[6] AC = AB, since these two lines are radii of circle A (see Definition 17).
[7] BC = AB, since these are both radii of circle B (Def. 17).
[8] AC = BC, since each is equal to AB (Steps 6 & 7; see Common Notion 1).
[9] So all three sides of the triangle are equal to one another, making it an equilateral
triangle (Def. 20).

Q.E.F.

10
THEOREM 1 Remarks

1. Notice what we have done in this first theorem: we have made a perfectly
equilateral triangle without measuring anything. There is something surprising about
that. And who would have thought that circles are helpful for making triangles?

2. The equilateral triangle is the simplest rectilineal figure, and so it makes sense
that we should begin geometry by making it. It has the fewest number of sides, and they
are all the same.

3. Despite the simplicity of the equilateral triangle, it is a figure rich with surprising
properties, some of which we will discover in this book. Also, it is very useful for other
constructions, as we will see once we get to Theorem 7. We have already accomplished
something significant in making the equilateral triangle, right here in the very first
theorem!

THEOREM 1 Questions

1. Does an equilateral triangle appear to be right, obtuse, or acute?

2. Looking at the diagram, find a way to make a rhombus. Prove all four of its sides
are equal. (We have not yet done enough geometry to prove that its angles are not right
angles.)

3. Imagine that your compass became so rusty you can no longer adjust it, so that
you can make circles only of one size. If the unadjustable radius of your compass is not
equal to AB in the diagram, but was something less than AB, can you still use it to make
an equilateral triangle on AB? (Remember that you still have a straight-edge which lets
you extend straight lines as far as you like.)

4. Using a method of construction similar to that for the equilateral triangle, can you
find a way to make an isosceles triangle? What about a scalene triangle?

5. What happens if you make 3 more equilateral triangles, one on each side of
ΔABC ?

11
 THE SIDE-ANGLE-SIDE THEOREM

THEOREM 2: If in one triangle a side, the next angle, and the next side, are
respectively equal to a side, the next angle, and the next side in another triangle, then all
the corresponding sides and angles of the two triangles are equal, and they have equal
areas.

Imagine two triangles ABC and DEF such that

AB = DE
∠ABC = ∠DEF (angle ABC equals angle DEF)
BC = EF
Then the remaining corresponding sides and angles of these two triangles must also be
equal, and the two triangles must contain equal areas. To see this,

[1] Imagine moving rABC so that AB lies on DE; since they are equal, they will
coincide, so that A is on D and B is on E.
B E

[2] Since ∠ABC = ∠DEF, BC will fall

along EF, and C will fall on F (since BC =
C F
EF).
A D
E
[3] So A, B, C are sitting on D, E, F.
Then AC must coincide with DF; for if it
(B)
fell outside it, as the dotted line, then two
(C)
straight lines, namely AC & DF, would cut
F
(A) each other at two points and enclose a
D space (which is impossible).

[4] So AB, BC, AC coincide with DE, EF, DF. Thus the two triangles have been made
to coincide exactly, and so they are completely identical (Common Notion 4).

[5] Since the two triangles are identical, it follows that

AC = DF & ∠BAC = ∠EDF & ∠BCA = ∠EFD
that is, the remaining sides and angles are equal, and also the triangles have equal areas.

Q.E.D.

THEOREM 2 Remarks

1. This theorem is often called the “Side-Angle-Side” Theorem. If a side, the next
angle, and the next side in one triangle are equal to a side, the next angle, and the next
side in another triangle, then the two triangles are identical to each other.

12
 B E 2. This “Side-Angle-Side” Theorem is used
throughout geometry, and throughout this book. Since
the areas of the two triangles are equal, and also the
corresponding sides and angles, this theorem is useful
C F
for proving the equality of (1) lines, (2) angles, and
A D
(3) areas.

B E 3. An important principle is used in this theorem,

namely that Two straight lines cannot enclose a space.
Another way to say the same thing is that Two straight
F
lines cannot cut each other more than once. These
statements are obvious because of the perfect
A C D uniformity of straight lines; after cutting once they
cannot bend back to cut each other a second time.

4. The two triangles in this theorem are not only equal in area, but all their
corresponding sides and angles are equal. Such triangles are said to be congruent to each
other; in fact, any pair of figures which are both the same shape and the same size are
called “congruent”. The symbol for congruency is ≅. The squiggly line over the equal
sign means “similar”, so that the two symbols together mean “equal and similar to”, or
“the same size and the same shape”. So when you read “ΔABC ≅ ΔDEF”, this means
that all the corresponding sides and angles of triangle ABC and triangle DEF are equal,
and the two triangles have the same area.

THEOREM 2 Questions

1. When we are told which initial sides and angles are equal, exactly how do we
decide which of the remaining angles and sides in the two triangles are “corresponding”
sides and angles?

2. What if the two triangles in question are mirror images of each other? Then it is
not possible to slide one over on top of the other. So how would we prove the theorem?

3. What if ∠ABC = ∠DEF, AB = DE, but BC is greater than EF? What can we say
about the two triangles then?

13
THEOREM 3: In an isosceles triangle, the base angles are equal to each other,
and the angles under the base are equal to each other.

Imagine an isosceles triangle ABC in which AB

A
= AC. I say that the base angles, namely ABC
and ACB, those opposite the equal sides, must
be equal to each other, too. Moreover, if you
extend AB to D and AC to E, the angles under
the base must also be equal to each other,
namely DBC and ECB. To see it, just …
B C
[1] Cut off CE = BD by drawing a circle
around C with a radius equal to BD.

D E [2] Join BE. Join CD.

[3] AB = AC (triangle ABC is given as isosceles)

BD = CE (Step 1)
so AD = AE (the sums of equals are equal; C.N. 2)

[4] AC = AB (triangle ABC is given as isosceles)

[5] ∠BAC is common to triangles ADC & AEB

[6] ΔADC ≅ ΔAEB (Steps 3, 4, & 5: Side-Angle-Side)

[7] ∠ADC = ∠AEB (being corresponding angles of the equal triangles)

and DC = EB (being corresponding sides of the equal triangles)
and BD = CE (Step 1)
so ΔBDC ≅ ΔCEB (Side-Angle-Side)

[8] ∠EBA = ∠DCA (being corresponding angles in ΔADC & ΔAEB; Step 6)
but ∠EBC = ∠DCB (being corresponding angles in ΔCEB & ΔBDC; Step 7)
so ∠CBA = ∠BCA (the remainders of equals are equal; C.N. 3)
i.e. the angles at the base of ΔABC are equal to each other.

[9] And the angles under the base of ΔABC are also equal, namely ∠DBC & ∠ECB,
being corresponding angles of ΔBDC & Δ CEB (Step 7).

Q.E.D.

14
THEOREM 3 Remarks

1. Notice the plan of attack, or strategy of this theorem. It is to establish first that
the larger overlapping triangles (ADC and AEB) are identical or congruent, and then to
use information gained from that to prove the small overlapping triangles under the base
are identical or congruent (BDC and CEB). Once that is accomplished, we can derive
that the base angles are equal as well as those under the base by using corresponding
angles and by subtracting equals from equals. To understand more elaborate proofs it is
always helpful to figure out the overall strategy used.

2. “All isosceles triangles have their base angles equal” is an example of a

universal truth. Such truths apply to an unlimited number of cases. How many isosceles
triangles exist, have existed, or will exist? An unlimited number, and they all have their
base angles equal. We have just proven this, even though we could not possibly inspect
every individual isosceles triangle. When did Theorem 3 begin to be true? Never. It was
always true and always will be. Its truth did not depend on us proving it – all the proof
did was allow us to come to know it. Theorem 3 was true before any geometer existed.
Can the truth of Theorem 3 ever change? No. It is necessarily true and cannot be
otherwise. Where is Theorem 3 true? Does it apply only in the United States? No. It is
true everywhere. Geometry, and all other sciences, aim especially at understanding truths
of this kind – universal truths, which are not tied to time and place, and which can never
change but must always be so.

THEOREM 3 Questions

A
1. You can prove this theorem in another way by flipping the
triangle ABC over onto itself: since AB = AC, each will fit on top of
the other when we flip the triangle over, and so ∠ACB will coincide
exactly with the place where ∠ABC was. Therefore these two angles
are equal.
B C

A
2. In Theorem 7 we will learn how to bisect any angle. Imagine
that ∠BAC was already bisected for us by the straight line AM. Given
that AB = AC, can you see how to prove once more that ∠ABC =
∠ACB?
B M C

3. Use Theorem 3 to prove that all three angles of an equilateral

triangle are equal to each other.

15
THEOREM 4: If two angles in a triangle are equal to each other, then the sides
opposite them are also equal to each other, i.e., the triangle is isosceles.

A Imagine a triangle ABC, in which ∠ABC = ∠ACB. Then AB =

AC, too. To see this, assume for a moment that AB and AC are
D not equal – let's see if that is really possible. Suppose, for
example, that AB > AC. Then …

[1] From AB, the supposedly greater of the two sides, cut off BD
= AC and join CD.
B C

[2] Consider ΔDBC and ΔACB:

since BD = AC (we supposedly made it so)
and ∠DBC = ∠ACB (as given)
and BC is common to both triangles
thus ΔDBC ≅ ΔACB (Side-Angle-Side), and so their areas are equal.

[3] But that is ridiculous, since ΔDBC is a part of ΔACB, and the part is always less than
the whole (Common Notion 5).

[4] Since something impossible follows from the assumption that AB > AC, therefore
the assumption that AB > AC is itself impossible. So AB is not greater than AC.

[5] By the same reasoning we can show that AC is not greater than AB.

[6] Since AB and AC are such that neither is greater than the other, therefore AB = AC
after all.

Q.E.D.

THEOREM 4 Remarks

1. By Theorem 3, an equilateral triangle has all three of its angles equal. Does it
follow from Theorem 4 that, since an equilateral triangle has any two of its angles equal
to each other, it must therefore be an isosceles triangle? An equilateral triangle is
“isosceles” in the sense that it has at least two of its sides equal, but not in the more
restrictive sense that it has only two of its sides equal. We might restate Theorem 4 by
saying “If a triangle has two angles equal, then it has at least two sides equal, namely
those opposite the equal angles.”

16
2. Theorem 4 is the “converse” of Theorem 3. The “converse” of any statement is
the statement formed by switching its subject and predicate; for example the converse of
the statement
Every bachelor is a man who never married
is the statement that
Every man who never married is a bachelor.
Not all converses of true statements are themselves true. For example,
Every triangle is a figure = TRUE
Every figure is a triangle = FALSE.
That is why we must often prove the converse of a true statement. Still, sometimes this is
unnecessary. Looking back at Theorem 2, we learned that triangles having certain
corresponding parts equal will have all corresponding parts equal. The converse is also
true, but needs no proof, namely that if two triangles have all corresponding parts equal,
then they will also have some of their parts equal.
So Theorem 4 proves the converse of Theorem 3, since this is what each proves:
Theorem 3: An isosceles triangle is a triangle with two equal base angles.
Theorem 4: A triangle with two equal base angles is an isosceles triangle.

3. The point of proving the converse of a theorem (whenever its converse is true) is
to show that the property we showed belongs to some figure belongs only to that kind of
figure. In Theorem 3 we learned that ALL isosceles triangles have equal base angles –
but is having equal base angles unique to isosceles triangles (namely those with at least
two equal sides), or can scalene triangles have that property too? Well, in Theorem 4 we
learn that ONLY isosceles triangles have equal base angles. As soon as the base angles
are equal, the sides opposite them must also be equal.

THEOREM 4 Question

Use Theorem 4 to prove that any triangle with 3 equal angles is an equilateral triangle.

17
THEOREM 5: Triangles are rigid.

B C Imagine a standing square made out of 4 boards nailed

E F together with one nail at each corner: A, B, C, D. If you
gave AB a swift kick, then each nail would act like a hinge-
pin, and the square would slant over into a new shape such as
A D AEFD, where all the boards remain the same length, but they
are now at new angles to each other. So squares are
M N
2 3 "flexible".
1 4 Can we do the same thing with a triangle? Imagine
ΔABM, and suppose another triangle ABN could sit on the
same base AB, having different angles from ΔABM, and yet
AM = AN
A B
and BM = BN.

Is that possible? No! To see the impossibility, try to suppose it is possible ...

[1] Join MN.

[2] ∠1 + ∠2 = ∠3 (since AM = AN, ∠AMN = ∠ANM by Thm. 3)

[3] ∠2 = ∠3 + ∠4 (since BM = BN, ∠BMN = ∠BNM by Thm. 3)

[4] Now, by Step 3 we can substitute (∠3 + ∠4) in place of ∠2 anywhere we please,
since they are equal. So in Step 2 let's replace ∠2 with (∠3 + ∠4), which gives us

∠1 + (∠3 + ∠4) = ∠3

[5] So ∠3 = ∠3 + ∠1 + ∠4 (Step 4 rearranged),

which is to say that an angle is equal to itself plus some other angles, which is
impossible.

[6] Therefore the original supposition from which this absurdity follows is also
impossible, namely that the triangle ABM might be flexible, and we could keep all its
sides the same length but change its angles. So triangles are rigid.

Q.E.D.

18
THEOREM 5 Remarks

1. The kind of proof we just used for Theorem 5, like the proof for Theorem 4
before it, is called a “reduction to the absurd” (reductio ad absurdum or reductio ad
impossibile in Latin). It is also called “indirect proof.”
The strategy of such a proof is to assume the opposite of what we wish to prove,
show that something impossible would necessarily follow from such an assumption, and
conclude that therefore the original assumption which gave rise to the impossibility is
itself impossible. And so the thing we wish to prove is necessarily true – since its
opposite turned out impossible.
Theorem 4 was a perfect example of such a line of reasoning. We wanted to
show that any triangle with two equal angles is isosceles. Assuming the opposite, namely
that we could have a triangle with equal angles whose opposite sides were not equal, it
would follow that we could cut out a part of that triangle which is equal to the whole
triangle. But that’s impossible. So we also had to condemn the initial assumption. That
is, the sides opposite equal angles in a triangle must be equal, lest an absurdity follow.

2. Another way to state Theorem 5 is this: given three straight lines making up a
triangle, it is impossible to take them apart and put them together again and get a different
triangle. Try it.

C D
3. It is possible to make two triangles on top of
AB which have identical sides but do not coincide,
such as rABC and rABD. Yet they remain
congruent, having identical sides and angles, and it is
not AD that AC is equal to, but BD. The triangles are
mirror images of each other. Therefore we do not get
A B rADB by “tilting” rACB, but by flipping it over.

4. Theorem 5 has some applications in building things. If you look at most bridges
and frames for roofs, you find triangular braces everywhere – that's because triangles add
rigidity. If we use only rectangles, things can fold over, tilt, collapse!

19
THEOREM 5 Questions
P
1. What if we supposed triangle ANB fell
inside the original triangle AMB? Could we still
M prove the theorem then? Draw yourself a diagram:
1
join MN as before, extend AM to any point P and
2
extend AN until it meets MB at a point R. You will
want to use the part of Theorem 3 about angles
R
3 under the base of an isosceles triangle, since you are
N
4 given that AM = AN (making AMN isosceles) and
that BN = BM (making BMN isosceles, too). Begin
A B by writing out the angles which must therefore be
equal, and see if you can find something impossible.
If so, you will have proved that neither is it possible to squish triangle AMB down to a
smaller triangle ANB while keeping all the sides the same length.
M
2. How would the proof go if we assume N fell on MA or on
N MB?

A B

3. Five sticks are nailed together to make a pentagon, with only one nail at each
corner. How many cross braces are needed to make the figure rigid? In how many
different ways can this be done?

20
 THE SIDE-SIDE-SIDE THEOREM

THEOREM 6: If the 3 sides of one triangle are equal to the 3 corresponding

sides of another triangle, then the two triangles will also have their corresponding angles
equal (namely those between equal sides), and they will have equal areas.

A B Imagine ΔA, whose three sides are equal to the

corresponding sides of ΔB. Since all their sides are equal,
let them sit on the same base CD, and place them so that the
equal sides share an endpoint, namely C and D, so that
CA = CB
and DA = DB.

[1] Now, if points A and B do not coincide with each other,

then triangles will not be rigid; for then ΔBCD would have
its three sides all equal to the sides of ΔACD, and yet it
C D would slant a little to the right.

[2] Since triangles are rigid (Thm. 5), A and B must coincide.

[3] Since points A & B must coincide, therefore CA coincides with CB, and DA
coincides with DB.

[4] Since CA coincides with CB and DA with DB, therefore

∠DCA coincides with ∠DCB
and ∠CDA coincides with ∠CDB
and ∠CAD coincides with ∠CBD

[5] Since things which coincide are equal, therefore these two triangles have all their
corresponding angles equal, and are also equal in area.

Q.E.D.

THEOREM 6 Remarks

1. This Side-Side-Side Theorem is in the same family as Theorem 2, the Side-

Angle-Side Theorem. It is our second Theorem about the conditions required for two
triangles to be congruent to each other.

2. As with Theorem 2, it does not matter if the two triangles are mirror images of
each other – we have only to flip one of them over, and then the proof can proceed as it
does.

21
THEOREM 6 Question

Using this Theorem, prove that the two equilateral triangles made on the same straight
line (one above it and the other below it) are equal to each other.

THEOREM 7: How to bisect any rectilineal angle.

B Suppose someone gives you an angle ABC.

How can you cut it exactly in half? Easy:

[1] Pick any point D on AB, and make a circle around

B with radius BD, cutting off BE = BD.
[2] Join DE.
[3] Make an equilateral triangle DEF on DE (Thm. 1).
D E
[4] Join BF.

This line BF bisects angle DBE, that is, ∠FBD =

∠FBE. Why? Because …

A F C

[5] BE = BD (we made it so)

[6] FE = FD (they are sides of an equilateral triangle)

[7] BF is common to both ΔFBD and ΔFBE

[8] So the 3 sides of ΔFBD are equal to the 3 sides of ΔFBE, and so their corresponding
angles are equal (Side-Side-Side, Thm. 6).

[9] ∠FBD = ∠FBE (being corresponding angles in ΔFBD & ΔFBE).

Q.E.F.

22
THEOREM 7 Remarks

1. It follows from this Theorem that there is no smallest rectilineal angle, since no
matter how small it is, we can always use this theorem to bisect it and get two smaller
angles.

2. The way to trisect any random angle, that is, cut it into 3 equal angles, is not so
easy; in fact it is impossible if we limit ourselves to using circles and straight lines in a
plane. Some angles can be trisected without difficulty, but others, such as the angle of an
equilateral triangle, require more sophisticated tools than circles and straight lines.

3. Since we are cutting angles in half, we might as well mention here the mechanical
tool for doing this, the protractor, and the unit it uses, the degree.
Suppose you have a circle whose circumference has been divided into 360 equal
parts for you. Each part is one 360th of the way around a circle, and it is called a degree.
Degrees, accordingly, can be used to measure either the length of an arc around the circle,
or the angle drawn from the center standing on that arc.
A
If we go along the circumference of a circle from A
R
through R to B, and we have gone one quarter of the way around,
then we have gone through 90 of the 360 equal parts of the
D
C
B circumference, or 90°. We can say that the arc ARB is an arc of
90°, or we can say that the angle ACB is an angle of 90°.
To go halfway around the circle from B to D is to go 180°
(since that is half of 360°), and that is to open up an angle into a
straight line DCB. Since ∠ACB is 90°, ∠ACD must also be 90° (since together they
make 180°). Since they are adjacent equal angles and DCB is a straight line, each of
them is right. So a right angle is 90 degrees. Any angle more than that is obtuse, and
any angle less than that is acute.
A protractor is a simple hand tool used for measuring angles – it is basically a
circle (or semicircle) with degrees marked off along its circumference and numbered. In
geometry, a protractor can be handy for making accurate diagrams, even though we don't
need it to prove anything.

THEOREM 7 Questions

1. Will the construction and proof for Theorem 7 still work if we put the equilateral
triangle on top of DE?

2. Draw any triangle and, using the method in Theorem 7, bisect its three angles as
carefully as you can. What do you notice about the three bisectors?

23
3. Do you need to draw the sides of the equilateral triangle in order to draw the
bisector in Theorem 7? No. That is only for the sake of the proof. Then what is the
fewest number of steps needed in order to bisect an angle?

4. If 90° is a right angle, and 180° is two right angles, then give the complementary
and supplementary angle for each of the following angles: 30°, 45°, 37.5°, 27.368°.

THEOREM 8: How to bisect any straight line.

C Suppose you have a straight line AB you need

to bisect. How do you do it?
Like this:

[1] Make an equilateral triangle ABC on AB

(Thm. 1).

[2] Bisect ∠ACB with a line CD (Thm 7).

Now I say that the point D, where angle-bisector

CD meets AB, bisects AB. That is, AD = DB.
A D B Why? Because …

[3] AC = CB (being sides of equilateral triangle ACB)

[4] ∠ACD = ∠BCD (since CD bisects ∠ACB)

[5] CD is common to both ΔACD and ΔBCD.

[6] So by the Side-Angle-Side Theorem (Thm 2), the other corresponding sides and
angles of ΔACD and ΔBCD are also equal.

[7] AD = DB (being corresponding sides in ΔACD and ΔBCD)

Q.E.F.

24
THEOREM 8 Remarks

1. From Theorem 8 it follows that there is no smallest straight line, since no matter
how small it is we can use this Theorem to bisect it and get two smaller lines.

2. A rough way to bisect a length in practice is simply by measuring with a ruler or

tape measure and dividing the length in half numerically. You can also take a piece of
paper of the given length and fold it in half, giving the half length.

3. Unlike the problem of cutting an angle into 3 equal parts, the problem of cutting a
straight line into 3 equal parts is not very difficult. In fact, later in this book, we will find
a way to cut any straight line into any number of equal parts. Before we get there, see
whether you can come up with a way to do it yourself.

THEOREM 8 Questions

1. Looking back to the diagram for Theorem 1, can you see the fewest steps needed
in order to bisect a straight line?

2. Draw any triangle you like and bisect the three sides of it. Join each vertex of the
triangle to the midpoint of the opposite side. What do you notice about the three straight
lines you have drawn?

THEOREM 9: How to draw a line at right angles to any straight line from any
point on it.

R Suppose you have a straight line AB, and I

pick a random point P on it. How can you
draw a line from P at right angles to AB? As
follows.

[1] Pick any point C on AP, and draw a circle

around P with radius PC, thus cutting off PD =
A B PC.
C P D
[2] Make an equilateral triangle CDR on CD.

25
[3] Join PR.

We did it: PR is at right angles to AB. How do we know? Because …

[4] PD = PC (Step 1)

[5] CR = DR (being sides of equilateral triangle CDR)

[6] PR is common to ΔRPC and ΔRPD.

[7] So, by the Side-Side-Side Theorem (Thm. 6), all of the corresponding angles of
ΔRPC and ΔRPD must be equal.

[8] ∠RPC = ∠RPD (being corresponding angles of ΔRPC & ΔRPD)

[9] So ∠RPC and ∠RPD are adjacent angles, formed by one line RP standing on
another, AB, and they are equal to each other. Therefore they are right angles (Def. 9).

[10] Thus a straight line, PR, has been set up at right angles to the given line AB and
from the point P on it.

Q.E.F.

THEOREM 9 Remarks

1. Since we can now draw right angles, we can also draw right triangles. In fact, we
have drawn two right triangles in the construction for this Theorem, namely ΔRPC and
ΔRPD.

2. In order to avoid having to go through all the steps in this construction every time
a right angle is needed, carpenters and engineers use a tool called a carpenter's "square",
which is a tool shaped like a big right angle or letter "L". You can also use a protractor to
mark off angles of 90°. Of course, all such tools have to be constructed by employing
geometrical constructions such as the one we have given here.

26
THEOREM 9 Questions

R 1. Suppose you have a straight line CD that has been

bisected at P, so CP = PD. Imagine that three circles of
M N the same radius (namely CP or PD) have been drawn
around point C, point P, and point D.
The left circle cuts the middle one at point M, the
C
P
D
right circle cuts the middle one at point N.
Extend CM and ND until they meet at a point, R.
Join RP.
If you have drawn all of this carefully, and you
measure CM and MR, what do you notice? If you measure angle RPD, how many
degrees does it seem to be? See if you can settle the question with a proof.

2. What is the fewest number of steps needed to draw a line from a point P at right
angles to a line through P?

R 3. Prove that any point along the perpendicular bisector of a

straight line is equidistant from both ends of the line. That is, if
RB is perpendicular to CD and bisects it at B, and R is any point
along the perpendicular RB, prove that RC = RD.

C D
B

THEOREM 10: How to drop a perpendicular line to any straight line from any
point above it.

Now suppose you have a straight line AB, and I pick a random point P above it. How
can you drop a line from P which is perpendicular to AB? Easy.

[1] Choose any point D below AB.

[2] Draw a circle around P with radius
P
PD, thus cutting AB at G and E.

[3] Join GP; join EP.

A
G H E
B [4] Bisect GE at H (Thm. 8).
D
[5] Join PH.

PH is in fact perpendicular to AB. Why? Because …

27
[6] PG = PE (being radii of circle P)
[7] HG = HE (since we bisected GE at H)
[8] PH is common to ΔPHG and ΔPHE.
[9] So, by the Side-Side-Side Theorem (Thm. 6), all the corresponding angles of ΔPHG
and ΔPHE are equal.
[10] ∠PHG = ∠PHE (being corresponding angles of ΔPHG and ΔPHE)
[11] But these two equal angles, ∠PHG and ∠PHE, are adjacent angles formed by one
line, PH, standing on another, AB, and so they are right angles (Def. 9).
[12] Hence PH has been drawn perpendicular to AB (Def. 10).

Q.E.F.

THEOREM 10 Remarks

1. Theorem 10 is somewhat the opposite of Theorem 9, since in Theorem 9 the point

from which we had to draw a perpendicular was on the given line – here in Theorem 10
the point is above the given line.

P 2. What if P is not directly above the straight

line AB? Then we simply have to extend AB
A B
until it passes underneath P.

3. The practical way to draw a straight line perpendicular to AB from a point P is to

use a tool such as a carpenter's square, the tool shaped like the letter L or like a big right
angle. Put one leg of the square on AB and slide it along until the other leg touches point
P, and then trace the edge of the square from point P down to AB.

THEOREM 10 Question

Do we need to draw PG and PE in order to draw PH? No; they were drawn for the sake
of the proof that PH is perpendicular. So what is the fewest number of steps actually
needed to draw a perpendicular from a point to a straight line?

28
THEOREM 11: When one straight line stands on another one, the adjacent
angles add up to two right angles.

P
Obviously if PB stands upon CD at right angles, then
∠PBC + ∠PBD equals two right angles. Now what if
A
AB stands upon CD, but not at right angles to it? Will it
still be true that ∠ABC + ∠ABD = 2 rights? Yes.
2
To see it,
1 3
C D
B [1] Draw BP at right angles to CD (Thm. 9).

[2] Thus ∠PBC + ∠PBD = Two rights

[3] But ∠PBC + ∠PBD = ∠1 + ∠2 + ∠3

[4] So ∠1 + ∠2 + ∠3 = Two rights (Common Notion 1)

[5] But ∠1 + ∠2 + ∠3 = ∠ABC + ∠ABD

[6] So ∠ABC + ∠ABD = Two rights (Common Notion 1)

Q.E.D.

THEOREM 11 Remarks

1. Theorems 9 & 10 began the study of right angles; now we are investigating “two
right angles.”

2. This Theorem 11 is rather obvious even without a proof. If CBD is a straight line,
then BD must go through half of one full rotation to get to BC, which is 180°. It does not
make any difference how we divide up that 180° with another line such as AB; the two
angles into which the 180° has been divided must still add up to 180°.

3. We should note in connection with this Theorem that angles supplementary to the
same angle are equal to each other. For example,
if ∠X + ∠Y = two rights,
and ∠X + ∠Z = two rights,
then ∠Y = ∠Z,

29
being both supplementary to ∠X. This is obvious, since each is equal to two right angles
minus ∠X (recall that all right angles are equal, and that equals with equals subtracted
from them leave equal remainders).

4. Recall Postulate 5, which says that inclined straight lines eventually meet. By
definition, lines such as A and B are “inclined to each other” if 1 + 2 < 3 + 4. With the
help of Thm. 11, we can now define “inclined straight lines” another way.
We know that 1 + 3 = two rights
and that 2 + 4 = two rights
thus 1 + 2 + 3 + 4 = four rights
so if 1 + 2 is less than two rights,
it follows that 3 + 4 is more than two rights.
A 3 In other words, if 1 + 2 is less than two
1 rights, then these two inside angles are less
C than the outside angles 3 + 4, and therefore
B
2 the two lines A and B will incline to each
4 other toward the right, and eventually meet
there. So we can now restate our fifth
postulate like this: if two straight lines make less than two right angles on one side of a
third straight line, then they will eventually meet on that side.

THEOREM 11 Questions

1. What is ∠2 called in relation to ∠3?

2. What is ∠3 called in relation to ∠ABC?

30
THEOREM 12: If two adjacent angles add up to two right angles, then the two
lines other than the line common to both angles are in a straight line with each other.

Let ∠APB and ∠BPC be two adjacent

B
rectilineal angles beginning at the point P, and
suppose that
∠APB + ∠BPC = two rights.
Then AP and PC must be in a straight line with
2 X
1 3 each other.
A C
P
Why is that?

Simple. Suppose that PC is not the extension of AP in a straight line, then

[1] Extend AP in a straight line to some point X (Post. 2)

[2] Thus ∠1 + ∠2 = two rights (Thm. 11)

[3] But ∠1 + ∠2 + ∠3 = two rights,

That is, ∠APB + ∠BPC = two rights, since that is how they were given to us.

[4] So ∠1 + ∠2 = ∠1 + ∠2 + ∠3,
putting together Steps 2 & 3, since each side is supposedly equal to two right
angles.

[5] But that is crazy. For ∠1 + ∠2 is only a part of ∠1 + ∠2 + ∠3, and the whole
never equals a part of itself (Common Notion 5).
And since this impossibility follows from our initial supposition that PC is not the
extension of AP in a straight line, therefore that initial supposition is also
impossible.

[6] Therefore PC is the extension of AP in a straight line, i.e. AP and PC are in a

straight line with each other.

Q.E.D.

THEOREM 12 Remarks

1. Theorem 12 is the converse of Theorem 11. Theorem 11 showed that angles

adding up to a straight line must add up to two right angles; Theorem 12 shows that
angles adding up to two right angles must add up to a straight line.

31
2. Notice that we use Postulate 4 for the first time in this theorem. In Step 4 we say
that
∠1 + ∠2 = ∠1 + ∠2 + ∠3
on the grounds that each side is equal to two right angles. We are presuming that “two
right angles” is always the same amount, i.e. that all right angles are equal.

3. Notice that if from the equation in Step 4 we subtract ∠1 + ∠2 from both sides,
we get ∠3 = nothing! If that were true, then PX would coincide with PC, and AP would
be in a straight line with PC, which is what we set out to prove.

THEOREM 12 Question

Does it make any difference to the proof if someone says that PX actually falls below
PC?

THEOREM 13: When two straight lines cut each other, they make the vertical
angles equal to each other.

C B Imagine two straight lines AB and CD cutting

across each other at P.
4
The “vertical angles” in the diagram are equal to
1
P
3
each other, that is
1=3
2 2=4
Can it be proven? Of course.
A D

[1] ∠1 + ∠2 = two rights (Thm. 11).

[2] ∠3 + ∠2 = two rights (Thm. 11).

[3] ∠1 + ∠2 = ∠3 + ∠2 (Steps 1 + 2, Common Notion 1)

[4] Subtracting ∠2 from each side of Step 3, we have

∠1 = ∠3 (Common Notion 3)

[5] ∠2 = ∠4 is proven in the same way.

Q.E.D.

32
THEOREM 13 Remarks

1. Theorem 13 demonstrates a kind of symmetry: the angle on one side of two

cutting straight lines is a mirror image of the angle on the opposite side of the vertex.
The reason for this symmetry is clear enough: it is a result of the uniformity of
straight lines. Two straight lines cannot make one size opening on one side and another
size opening on the other side of the vertex. They tend toward each other on one side in
the same way that they tend away from each other on the other.

2. What about the converse? If we state the

A D total converse of Theorem 13, it is not true, namely
If two rectilineal angles are equal, then they are
1 2 formed by two cutting straight lines. Obviously,
two rectilineal angles can be equal and yet be far
B C
apart from each other. But the partial converse of
Theorem 13 is true, namely If two rectilineal angles
A D are equal and share a common vertex, and one leg
1 2
of one angle is in a straight line with the alternate
leg of the other angle, then the remaining legs are
B C also in a straight line with each other.

To be more concrete, look at the adjoining figure: If 1 = 2, and line A is in line with line
C, then line B must also be in line with line D.

On the other hand, if A is in line with line D, the adjacent leg in the other angle, it is not
true that B has to be in line with line C.

THEOREM 13 Questions

1. Prove the partial converse of Theorem 13 stated above in Remark 2.

2. Looking back at the diagram for Theorem 13, how many degrees does ∠1 + ∠2 +
∠3 + ∠4 equal?

33
THEOREM 14: If any side of a triangle is extended, the exterior angle is greater
than either of the interior and opposite angles.

A F Take any triangle – ABC will do. Extend any side,

say BC, to D. The angle ACD is called an “exterior
1
E angle,” and this theorem says
∠ACD > ∠BAC
2 and ∠ACD > ∠ABC.
B
D If you want proof, then …
C

G
[1] Bisect AC at E. (Thm. 8)
[2] Join BE and produce it far enough to cut off EF = BE.
[3] Join CF. Now,

[4] EF = BE (we made it so)

[5] EA = EC (we bisected AC at E)
[6] ∠AEB = ∠CEF (Thm. 13; they are vertical angles)
[7] So by the Side-Angle-Side Theorem (Thm. 2), the other corresponding angles of
ΔAEB and ΔCEF are equal.
[8] ∠1 = ∠2 (being corresponding angles of ΔAEB & ΔCEF)
[9] But ∠ACD >∠2 (since the whole is greater than the part)
[10] So ∠ACD > ∠1 (putting together Steps 8 & 9)
[11] So ∠ACD > ∠BAC (∠1 is ∠BAC)
And that is one part of what we wanted to prove.
[12] ∠ACD > ∠ABC is proved by bisecting BC, and by extending AC to any point G.
Then by the same argument as before,
∠BCG > ∠ABC
But ∠ACD = ∠BCG (Thm. 13; they are vertical angles)
Thus ∠ACD > ∠ABC.
[13] So the exterior angle ACD is greater than either of the opposite and interior angles,
namely ∠ABC and ∠BAC.

Q.E.D.

34
THEOREM 14 Remarks

1 1. Here we learn that

∠3 > ∠1
2 and ∠3 > ∠2.
3

2. In Theorem 27, we will learn that

∠3 = ∠1 + ∠2
which explains why ∠3 is greater than either one of them taken alone.

THEOREM 14 Questions

1. In Step 12 of Thm. 14, I said “by the same argument as before, ∠BCG > ∠ABC.”
Go through the actual steps. Make a diagram for yourself, bisecting BC this time.

2. Join AF in the original diagram or in your copy of it. What shape does AFCB
appear to be?

THEOREM 15: In any triangle, a greater side will have opposite to it a greater
angle.

A Look at triangle ABC. If it is given that

AC > AB,
it must also be true that the angle opposite AC is
D
2 greater than the angle opposite AB, that is:
∠ABC > ∠BCA.
1 3
C Why? Let's see …
B

[1] Since AC > AB, we can cut off AD = AB. Join BD.

[2] ∠1 = ∠2 (Thm. 3, since AB = AD)

35
[3] ∠2 > ∠3 (Thm. 14; since ∠2 is exterior to ΔBDC)

[4] ∠1 > ∠3 (Putting Steps 2 & 3 together)

[5] ∠ABC > ∠1 (Since the whole is greater than the part)

[6] ∠ABC > ∠3 (Putting Steps 4 & 5 together)

[7] That is, ∠ABC > ∠BCA (∠BCA is the same as ∠3)

Q.E.D.

THEOREM 15 Remarks

1. It follows from Theorem 15 that in any triangle:

The greatest angle is the one opposite the greatest side.
The least angle is the one opposite the least side.
We already know from Theorem 3 that in any triangle:
Angles opposite to equal sides are themselves equal.

2. These facts might lead us to think that angles in a triangle are somehow
proportional to their opposite sides, for example that if one side is double another side,
then the opposite angle is double the opposite angle. Unfortunately, that is completely
FALSE! To see this better, take a look at Question 2 below.

THEOREM 15 Questions

D 1. Using Theorem 15, prove that if

1
2 two triangles have two sides equal to two
A C sides, but one base greater than the other
B base, then the included angle is also
F
greater than the included angle. That is,
E

If A=C
and B=D
but E<F
Show that 1<2

36
 H
2. Draw an equilateral triangle GHK. Bisect GK at M and
join HM. Now just focus on ΔHMK. Clearly HK is double MK,
1
but are their opposite angles respectively double? That is, is ∠2
double ∠1? Prove it is not so.
2

G M K

THEOREM 16: In any triangle, a greater angle will have opposite to it a greater
side.

C Look at triangle ABC. If it is given that

2 ∠1 > ∠2,
B
1
then it must also be true that the side opposite ∠1 is greater
than the side opposite ∠2, which is to say that
AC > AB
Who says? Pure logic says . . .

[1] There are only three possibilities:

Either AC = AB
Or AC < AB
Or AC > AB

[2] Assume AC = AB, and let's see what happens.

[3] Then ∠1 = ∠2 (Thm. 3)

[4] But ∠1 > ∠2 (They are given to us that way)
[5] So AC cannot be equal to AB.

[6] Assume AC < AB, and let's see what happens.

[7] Then ∠1 < ∠2

since Thm. 15 says that in any triangle, a greater side stands opposite a greater angle.
[8] But ∠1 > ∠2 (They are given to us that way)
[9] So AC cannot be less than AB.

[10] Since AC is neither equal to AB (Step 5), nor less than AB (step 9), therefore AC is
greater than AB.

Q.E.D.

37
THEOREM 16 Remarks

1. From Theorem 16 we now know that, in any triangle:

The greatest side is the one opposite the greatest angle.
The least side is the one opposite the least angle.
We already know, from Theorem 4, that
Sides opposite equal angles are themselves equal.

2. Theorem 16 is the converse of Theorem 15. The strategy here is the process of
elimination. If there are only three possibilities, and we eliminate two of them, then the
third one must be true.

3. Notice there is no construction whatsoever in the proof for Theorem 16; it is a

matter of pure logic.

4. Even though angles are not exactly proportional to sides in a triangle, we now
know that the order of inequality among angles corresponds to the order of inequality
among their opposite sides.
If you have a protractor, but not a ruler, you can still tell which is the greatest side
in the triangle just by measuring the angles.
If you have a ruler, but not a protractor, you can still tell which is the greatest
angle in the triangle just by measuring the sides.

THEOREM 16 Questions

D 1. Using Theorem 16, prove that if two

2
1 triangles have two sides equal to two sides, but
A C
B
the included angle in the one triangle is less than
F the included angle in the other, then the base is
also less than the base. That is,
E

Given that A=C

and that B=D
but that 1<2
show that E<F
Use the process of elimination (either E = F or E > F or E < F), Theorem 6, and the
conclusion of Question 1 after Theorem 15.

38
 b 40 e
35 36 36
38
2. Given the lengths of some of the sides in each of these a c d f g
triangles, say what we know about the labeled angles. 37 22

k
51 51 120
35
3. Given the number of degrees in some of the angles of n r s
t h
25
these triangles, say what we know about the labeled sides. 100 35
m

THEOREM 17: In any triangle, any side is less than the sum of the other two
sides, but greater than their difference.

D Take any triangle ABC. Any two sides of it added

together must be greater than the third side, that is:
1 BA + AC > BC
and AC + BC > BA
A and BC + BA > AC

2 Proof:

C
[1] Extend BA to D so that AD = AC.
[2] Join CD.
B

[3] ∠BCD > ∠2 (since ∠2 is part of ∠BCD)

[4] ∠BCD > ∠1 (∠2 = ∠1, since AD = AC)
[5] So in ΔDBC, the side opposite ∠BCD is greater than the side opposite ∠1 (Thm.
16).
[6] BD > BC (being the sides opposite ∠BCD & ∠1 respectively)
[7] BA + AC > BC (BD = BA + AD, and AD = AC)
[8] Likewise we can prove that
AC + BC > BA
and BC + BA > AC
[9] Thus in rABC any side is less than the sum of the other two. It follows also that
any side must be greater than the difference between the other two. For example,
BA > BC – AC. Why? Because
BA + AC > BC (any two sides are greater than the third)
so BA > BC – AC (subtracting AC from both sides).

Q.E.D.

39
THEOREM 17 Remarks

1. We could also look at the theorem this way: if you had to race someone from A
to B, would you run first from A to C, then from C to B? No way. AB is the shortest
distance from A to B.

2. A more general version of this theorem is this: A straight line is the shortest
distance between two points. Every jointed or curved path is longer.
Obviously, that does not mean that the straight path is always the easiest to travel.
The shortest distance between two points on either side of a forest is a straight line, but
the quickest way through might be a curved path that takes us around the trees! Hence
the expression “as the crow flies”, which means I am telling you the straight line
distance, although the path you actually take to walk the distance might be curved and
therefore much longer. Crows don’t have to move around obstacles.

3. Although the straight line is the shortest distance between two points, there is
obviously no longest distance between two points. You can take as convoluted a path
from point A to point B as you like, and there will always be one even more convoluted
and longer.
This often happens in mathematics, that there is one kind of limit, e.g. a shortest
or smallest, but there is no limit in the opposite direction, e.g. a longest or greatest.

4. This theorem shows that there is a condition that must be fulfilled before any 3
straight lines will be able to make a triangle. Any two of those lines added together has to
be greater than the third. Otherwise, you can't make a triangle with those 3 straight lines.
For example, try to make a triangle whose sides are equal to 1, 2, and 3 unit lengths (say
1 inch, 2 inches, and 3 inches). Good luck!
This means that as soon as I give you two sides to make a triangle with, I have
placed a restraint on what you can use for the third side: whatever you use has to be less
than the sum of the two sides I already gave you, but greater than their difference. For
example, if I give you a side of 4 feet and a side of 7 feet, you cannot use 24 feet for the
other side, since that is bigger than 11 feet (the sum of the two sides I gave you). On the
other hand, you cannot use 2 feet, since that is smaller than 3 feet (the difference between
the sides I gave you). Any side you choose must therefore be greater than 3 feet and yet
less than 11 feet.

5. This condition is not true with regard to the angles of a triangle, i.e. it is not true
that any 2 angles in a triangle, taken together, have to add up to more than the remaining
angle. It is possible, for example, to have a triangle whose angles are 1°, 1°, and 178°.

40
THEOREM 17 Questions

1. Say which of the following triads of straight lines can be used to form a triangle,
and which cannot (say the lengths are all in inches):

(a) 3 4 5

(b) 1 2 3

(c) 1 2 1

(d) 4 5 6

(e) 1 10 11

(f) 1 1 1

(g) 3 2 3

(h) 7 9 8

2. If I give you a line 5” long, and another that is 6” long, the Theorem states that
any third line able to form a triangle with them must be between 1” and 11”; it must be
more than 1” long, and less than 11” long. What is the range of possible lengths for the
third side of a triangle with one side of 8 miles and another side of 11 miles?

3. Cut out six thin strips of paper with the following lengths in inches: ½, 1, 2, 6, 8,
10, 19. Measure exactly. Color the 10-inch strip red. Next make any triangle you like
using any three of these strips as sides.
I will now make a prediction: the triangle you made uses the red strip, and the
angle opposite to it is a right angle.
Why was I able to predict that you used the red strip? (The reason I know the
angle opposite it is a right angle you will learn in Theorem 37.)

41
THEOREM 18: The shortest distance from a point to a straight line is the
perpendicular from the point to the line.

P If P is a point and AB a straight line below it, what is the

shortest distance between P and AB? Clearly the shortest
distance between P and some point on AB is the straight
line between them, since the shortest distance between
any two points is a straight line (Thm. 17). For example,
PB is the shortest distance between P and B. And if R is
1 3 2
A B
a random point on AB, PR is the shortest distance
L R between P and R. But which among these sorts of
straight lines is shortest? The perpendicular line PL. PL is shorter than any other straight
line drawn from P to any random point R on AB. Here’s why:

[1] ∠1 > ∠2 (∠1 is exterior to ΔPLR; Thm. 14)

[2] ∠3 > ∠2 (∠3 = ∠1, since both are right angles)
[3] Therefore in ΔPLR, the side opposite ∠3 is greater than the side opposite ∠2
(Thm. 16), that is
[4] PR > PL.

Q.E.D.

THEOREM 19: How to make a triangle out of any three straight lines (provided
any two of them together are greater than the third).

Take any three straight lines X, Y, Z

C which are such that any two added
together make a line greater than the third
A F one. Can you make a triangle out of
D E
x y B z them? You bet.

[1] Call line X “DA”, and extend it out to

E as far as you need.

[2] Cut off AB = Y, and BF = Z.

[3] Draw a circle around A with radius AD.

[4] Draw a circle around B with radius BF.

42
[5] Since AD and BF add up to more than AB (given), the circles with radii AD and BF
will overreach each other along AB. So at least a part of AB is inside both circles at
once. Thus the circles around A and B overlap.

[6] Since BA and AD add up to more than BF (given), thus BD is greater than the radius
of circle B, and so point D lies outside circle B.

[7] Since AB and BF add up to more than AD (given), thus AF is greater than the radius
of circle A, and so point F lies outside circle A.

[8] Thus circle A and circle B must cut each other. For they overlap each other (by Step
5), and each passes outside of the other, since D on circle A lies outside circle B (Step 6),
and F on circle B lies outside circle A (Step 7).

[9] Call the point where circles A and B cut each other point C.

[10] Join AC and join CB, forming triangle ABC.

[11] AC = AD = X
AB = Y
BC = BF = Z

[12] So triangle ABC has been made, with its sides equal to X, Y, and Z.

Q.E.F.

THEOREM 19 Remarks

1. Theorem 19 is the converse of Theorem 17.

Thm. 17 says: If 3 lines can form a triangle, then any 2 of them is greater than the
remaining one,
Thm. 19 says: If 3 lines are such that any 2 of them is greater than the remaining
one, then they can form a triangle.

THEOREM 19 Question

Make a triangle with sides equal to 3”, 4”, and 5”. What do you notice about it?

43
THEOREM 20: How to make an angle equal to any angle.

Suppose an angle X is drawn on the wall, and I

A point out a straight line PR somewhere else on the
wall. Can you draw an angle beginning at P,
X
equal to X, and having PR as one of its sides?
B

Sure you can. Just . . .

Z

R
S P T V
[1] Pick points A & B at random on each of the
legs of ∠X, and join them.

[2] Extend RP, and

cut off PS = XA
cut off PT = XB
cut off TV = AB

[3] Make ΔPZT = ΔXAB (Thm. 19)

And so by the Side-Side-Side Theorem, the corresponding angles of these two
triangles must be equal.

[4] ∠ZPT = ∠AXB (being corresponding angles of ΔPZT & ΔXAB)

[5] But ∠ZPT begins at P, and PT lies along PR, and ∠AXB is the given angle X.

Q.E.F.

THEOREM 20 Remark

Theorem 20 obviously enables us to do more than duplicate an angle; it enables us to

duplicate an entire triangle.

44
 THE ANGLE-SIDE-ANGLE THEOREM

THEOREM 21: If in a triangle two angles and the side joining them are
respectively equal to two angles and the side joining them in another triangle, then all the
corresponding sides and angles of the two triangles are equal, and the triangles have the
same area.

C F Imagine two triangles ABC & DEF, such

that

∠BAC = ∠EDF
AB = DE
∠ABC = ∠DEF
A D

Then the remaining sides and angles of these

two triangles must be equal, too. Why, you
B E
ask? Because:

[1] Pick up ΔABC in your imagination and place AB on top of DE.

Since AB = DE, A will sit on D and B will sit on E.

[2] Since ∠BAC = ∠EDF, therefore AC and DF will lie in line together.

[3] Since ∠ABC = ∠DEF, therefore BC and EF will lie in line together.

[4] Since AC lies on DF (Step 2),

and since BC lies on EF (Step 3),
the meeting point of AC & BC sits on the meeting point of DF & EF.

[5] That is, C sits on F.

[6] So if the base AB is placed on the base DE, then

A coincides with D
B coincides with E
and C coincides with F

[7] So the 3 sides of ΔABC coincide with the 3 sides of ΔDEF, and so all of their
corresponding sides and angles are equal and they also have equal areas (Side-Side-Side
Theorem).

Q.E.D.

45
THEOREM 21 Remarks

C F
1. This is our third triangle-congruence theorem. We now
have:
SAS (Side-Angle-Side)
A D SSS (Side-Side-Side)
ASA (Angle-Side-Angle)
B E

2. As with the other congruence theorems, it is possible for the given triangles to be
mirror images of each other. In that case, we must flip over one of the triangles to show
that they can coincide.

THEOREM 21 Questions

1. Prove Theorem 21 by reduction to the absurd. Start by assuming, if possible, that

EF > BC. That will make it possible to cut off a part of EF that is equal to BC. Find an
absurdity, and you will have found a proof.

2. Some Boy Scouts stood at the edge of a canyon at point E, across from a tree on
the other side of the canyon standing at T. They needed to measure the distance across
the canyon, namely the distance ET. But how could they
T
measure a distance across a canyon? Here is what they did.
They marked E with a stake, and from it sighted along a
straight stick, pointing it at T, putting the stick in line with ET.
Next, they walked out to some point P further along the edge
E
30 P S of the canyon, so that ∠TEP was a right angle, and then they
marked P with a tall pole. Then they continued in a straight
40 line to S until PS was the same distance as EP, and then
50
marked S with another stake. Then they walked away from the
K
D canyon along the line SD so that SD was at right angles to PS.
They stopped at the point D where the pole at P was in the
same line of sight with the tree at T across the canyon. Then they just measured SD, and
said that was the same length as the distance ET across the canyon. Prove they were
right.
Incidentally, how could they ensure EP is at right angles to ET? They could
stretch a string E in the same line of sight with T, say 40 feet long, then form a triangle
with EP being 30 feet long, and KP 50 feet long. Why angle KEP would be 90° we shall
see later.

46
 ANGLE-ANGLE-SIDE THEOREM

THEOREM 22: If two triangles have two angles equal to two angles, and a side
equal to a side (namely a side opposite an equal angle), then they will have all their
corresponding sides and angles equal, and they will also have equal areas.

[C] Suppose you have two triangles ABC and DEF such
that
4
C F ∠BCA = ∠EFD
∠BAC = ∠EDF
2 3 AB = DE
{C}
Then the remaining sides and angles of the two
1
triangles must be equal, too. To prove it:

A B D E

[1] Pick up ΔABC and place AB on top of DE.

Since AB = DE, A will sit on D and B will sit on E.
[2] Since ∠BAC = ∠EDF,
AC must lie along DF.
But is AC the same length as DF?
[3] Suppose, if possible, that AC is shorter than DF, and C lands at {C} along DF. Let's
see if that works . . .

[4] ∠1 = ∠2 (since ΔD{C}E is the same as ΔACB)

[5] ∠3 = ∠2 (since the original triangles are given that way)
[6] ∠3 = ∠1 (putting Steps 4 & 5 together)
[7] So exterior ∠1 is equal to the opposite and interior ∠3 in triangle E{C}F – which is
absurd, since the exterior angle of a triangle must always be greater than an interior and
opposite angle (Thm. 14).
[8] Since the absurdity in Step 7 follows from our assumption in Step 3 that AC < DF,
therefore that assumption is also absurd, and so AC is not less than DF.
[9] For the same reason, neither can AC be greater than DF, such that C would fall as
[C] along DF extended. For then exterior ∠3 would equal interior and opposite ∠4.
[10] Since AC is not less than DF (Step 8), and AC is not greater than DF (Step 9), thus
AC is equal to DF.
[11] Since AB = DE, ∠BAC = ∠EDF, and AC = DF (Step 10), therefore all
corresponding sides & angles of ΔABC & ΔDEF are equal and they also have equal areas
(Thm. 2).

Q.E.D.

47
THEOREM 22 Remarks

1. This is our fourth triangle-congruence theorem. Now we have

SAS SSS ASA AAS

Essentially, we need 3 pieces of information about a pair of triangles to determine

whether or not they are equal; we need to know either that
(a) two sides and an included angle are equal to two sides and an included angle,
or that
(b) three sides are equal to three sides, or that
(c) two angles and one side are equal to two angles and one corresponding side.

B 2. What happens if the two sides given as equal are not

F
corresponding sides? For example, what if AB = DF? Then, as we
C shall see later, although the triangles might not be congruent, they
D
will still be similar.
A

3. The difference between Theorem 21 and Theorem 22 is simply this: in Theorem

21, we are given two triangles having two angles equal to two angles, and the side
adjoining the equal angles equal to the side adjoining the equal angles. In Theorem 22,
however, the sides given as equal are not the ones adjoining the pairs of equal angles, but
another pair of corresponding sides.

THEOREM 22 Question

1. What if two sides and a non-included angle are given as

1 equal to two sides and a corresponding non-included angle? Should
D
there be an SSA theorem? As a hint, consider rABC, and suppose
a circle around center B with radius AB cuts AC at D. Compare
rABC with rDCB. Will it be possible to make any more
triangles with ∠1 and side BC, having another side equal to AB?
A B

48
THEOREM 23: If two straight lines are cut by a third making the alternate
angles equal, then the two straight lines are parallel.

Imagine you have three straight lines: AB,

A E B
CD, EF. Suppose EF cuts AB and CD in such
1
a way that
X
∠1 = ∠2
2
Then AB and CD are parallels.
C F D
Let's prove it.

[1] Suppose, if possible, that AB and CD are not parallel, but rather EB and FD meet
toward the right (when extended far enough) at some point X.

[2] Then they form a triangle EFX.

[3] ∠1 is the exterior angle of triangle EFX,

∠2 is one of the interior and opposite angles of triangle EFX.
And yet . . .

[4] ∠ 1 = ∠2 (they are given to us that way)

[5] And so the exterior angle of triangle EFX is equal to one of its interior and
opposite angles, which is absurd, since by Theorem 14 the exterior angle is
always greater.

[6] Since the absurdity in Step 5 follows from our initial assumption that AB and CD
are not parallel, but meet and form a triangle, therefore that assumption is absurd,
and we must say instead that AB and CD are parallel, and never meet so as to
form a triangle with EF.

Q.E.D.

THEOREM 23 Remarks

1. Here we begin the study of parallels.

2. It is amazing that we can prove that two straight lines never meet, even though we
can't actually extend them forever and check. We have managed a way around that!

3. The straight line EF is called a transversal in relation to the straight lines AB and
CD, since it cuts across them both.

49
THEOREM 23 Question

Prove Theorem 23 in another way by the symmetry of the diagram; is there reason to
think that what happens on one side of EF must also happen on the other side? And if so,
can you see how that will help prove the theorem? Recall that two straight lines cannot
enclose a space.

THEOREM 24: If two straight lines are cut by a third making interior angles on
one side add up to two rights, then the two lines are parallel.
E
Suppose AB & CD are cut by EF so that
∠2 + ∠3 = two right angles.
A
1 2
B Then AB is parallel to CD.

Why?
3
C D

[1] ∠2 + ∠3 = two right angles (given)

[2] ∠2 + ∠1 = two right angles (Thm. 11)

[3] ∠2 + ∠3 = ∠2 + ∠1 (Putting together Steps 1 & 2)

[4] ∠3 = ∠1 (Subtracting ∠2 from both sides of Step 3)

[5] AB is parallel to CD (Since ∠3 = ∠1; see Thm. 22)

Q.E.D.

THEOREM 24 Remarks

1. Compare Theorem 24 to our fifth postulate. Postulate 5

1
says that if 1 + 2 < 180°, then the two lines must meet out
2 toward the right somewhere. Theorem 24 says that if 1 + 2 =
180°, then the two lines do not meet anywhere.

50
2. Notice that the postulate is about lines that do meet, and this theorem is about
lines that do not meet.

P
3. It follows from this theorem that there cannot be two
perpendiculars from one point to one straight line, but only one.
Assume that you could have two perpendiculars, PR & PL, from
the same point P to the straight line AB. Then we have formed a
triangle PRL having two right angles in it, namely ∠PRL &
∠PLR. By Theorem 24, RP and LP have to be parallel! And so
either RP and LP do not meet (and therefore there is no point P
A R L B common to them), or one of those two angles is less than a right
angle.

THEOREM 24 Questions

1. What if 1 + 2 > 180°? What happens then?

2. Is it possible to prove Theorem 24

A E P B
2
independently of Theorem 23? Try bisecting EF in the
diagram at its midpoint M, then drop MP perpendicular
M to AB, and extend PM to T on CD. As with Theorem
23, think about the symmetry of the figure. If you can
3
C T F D
prove ∠MTF is a right angle (compare rEMP and
rFMT), then you can see that whatever happens on
one side of PMT must also happen on the other side of it.

THEOREM 25: If a straight line cuts two parallels, it makes interior angles on
one side add up to two rights, and it makes equal alternate angles.

E Let AB and CD be two parallel straight lines cut by

EGHF, making alternate angles 1 & 3, and interior
G
angles 2 & 3 on one side of EGHF.
A B
1 2
Then 2 + 3 = two right angles
and 1=3
3
C D Here’s how we know:
H

51
[1] Suppose, if possible, that 1 > 3.

[2] Then 1 + 2 > 3 + 2 (adding equals to unequals gives unequals)

[3] But 1 + 2 = two rights (Thm. 11)

[4] So two rights > 3 + 2 (putting together Steps 2 & 3)

[5] So AB and CD eventually meet on the side of angles 3 & 2, if AB and CD are
extended far enough. (Postulate 5, as restated in Thm.11, Remark 4)

[6] But that is absurd, since AB and CD are given to us as parallel straight lines, i.e. lines
that never meet no matter how far they are extended.

[7] Therefore our initial supposition, namely that 1 > 3, is also absurd.

[8] By the same kind of argument, we can show that 1 < 3 is also absurd.

[9] Since 1 is neither greater than 3 (Step 7) nor less than 3 (Step 8), therefore 1 = 3.

[10] But 2 + 1 = two right angles (Thm. 11)

[11] Thus 2 + 3 = two right angles (Since 1 = 3, by Step 9)

Q.E.D.

THEOREM 25 Remarks

1. This theorem is the converse of Theorems 23 and 24.

2. In this theorem, we make the first use of our 5th Postulate, namely in Step 5.

3. Notice the order of Theorems 23 & 24 & 25. In Theorems 23 & 24, we are given
lines that are disposed to a transversal at certain angles, and then we prove that they must
be parallel. In Theorem 25 we are given lines that are parallel, and then we prove that
they are disposed to a transversal at certain angles. We begin by giving ourselves what
we know we can have, namely two lines disposed at certain angles to a transversal; we
can construct that. But we do not know how to construct parallel straight lines (or even
that they exist) until after Theorems 23 & 24. That is why we wait until Theorem 25 to
give ourselves a pair of parallels, and ask what has to be true about them.

52
4. We have just proved that if AB & CD are parallel, then ∠2 + ∠3 must add up to
180°. That means that if ∠2 + ∠3 do not add up to 180° (but add up to something more
or less than that), then AB & CD are not parallel.
If they add up to less than 180°, then they meet out toward the right side, which is
what the fifth postulate says. If they add up to more than 180°, then they will meet on the
other side of GH, toward the left, since the angles on that side will then add up to less
than 180°.

THEOREM 25 Questions

X
1. Using Theorem 25, prove that only one line through a
P L point P can be parallel to any given line AB. Start by assuming
the possibility of having two parallels to AB drawn through the
A B same point P (call the other one PX), and join P to any random
R
point R along AB, making PR a transversal.

A
1

2. If A is parallel to B, and ∠4 = 35°, then how many

B
2 3 degrees is ∠1?
4

THEOREM 26: Straight lines parallel to the same straight line are parallel to
each other.

D Should you require proof for this, let A, B, C be three

straight lines such that
A
1
A is parallel to B
and B is parallel to C
2
B Then it must also be true that
3
A is parallel to C.

4
C

53
To prove it:

[1] Cut all three straight lines with another straight line D; in doing so you form
angles 1, 2, 3, 4.

[2] ∠1 = ∠2 (since A is parallel to B; Thm. 25)

[3] ∠2 = ∠3 (vertical angles are equal; Thm. 13)

[4] ∠1 = ∠3 (putting Steps 2 & 3 together)

[5] ∠3 = ∠4 (since B is parallel to C; Thm. 25)

[6] ∠1 = ∠4 (putting Steps 4 & 5 together)

[7] A is parallel to C (since ∠1 = ∠4; Thm. 23)

Q.E.D.

THEOREM 26 Remarks

1. Parallelism, according to Theorem 26, is similar to the relationship of equality.

Just as lines equal to the same line are equal to each other, so too lines parallel to the
same line are parallel to each other.

2. Notice this kind of statement does not apply to the relationship of

perpendicularity: it is not true that perpendiculars to the same line are perpendicular to
each other.
Some family relationships are like equality and parallelism, and others are more
like perpendicularity. For example:
Brothers to the same person are brothers to each other.
On the other hand, we cannot say that
Parents to the same person are parents to each other.
A

B
1
C
3. Another theorem similar to Theorem 26 is this: If the sides of
4
3 one angle are parallel to the sides of another angle, then the two
D
2
E angles are either equal or supplementary. Consider the case of
∠ABC and ∠DEF, where AB║EF and BC║DE. Join BE, and
F

54
suppose that each angle lies on opposite sides of BE. It is easy to see that ∠1 = ∠2, i.e.
that ∠ABC = ∠DEF. For:
1 + 3 = 4 + 2 (since AB║EF; Theorem 25)
and 3=4 (since BC║DE; Theorem 25)
thus 1 = 2 (subtracting the equal angles 3 & 4 from each side).
Obviously the converse of this theorem is not true, namely “If two angles are
equal, then the sides of one are parallel to the sides of the other”.

THEOREM 26 Questions

1. What true statement can we make about two straight lines that are perpendicular
to the same straight line?

A 2. There is an easy case of Theorem 25. Suppose A is

parallel to B, and C is parallel to B, and B is in between A and C.
In that case, see if you can prove that A and C are also parallel,
B
without using any prior theorems at all. (If we are instead given
C
that A║C and B║C, we need a proof such as the one we gave.)

A
K D 3. Show that if two straight lines cut, then so do their
4 X
parallels. Suppose AB ‖EF, and CD ‖GH, and suppose
2 that AB cuts CD at a point X. Prove that GH must also cut
C
L
B
EF.
E Start by joining GE and extending it. Since GE
3 passes through E, but EF is the only parallel to AB through
F E, BE must cut AB at some point, say K. Likewise GE
H must cut CD at some point, say L. You have now formed a
1
triangle KXL. Can you see how to proceed from there?
G

R
U 4. In Remark 3 above, we chose
G O one particular case of two angles
13
S
11
V W whose corresponding sides are parallel,
5 9
H
6
K
P 10
12 14
Z namely the case where each angle falls
L
7
T
15
on an opposite side of the line joining
Q
M
8
N X
16
their vertices. See if you can prove the
Y
theorem in some of the remaining
cases, and state when the angles are equal, and when they are supplementary to each

55
other. Consider, for example, what happens when (1) Each angle falls on the same side
of the line joining their vertices, as ∠GHK and ∠LMN, (2) Each angle straddles the line
joining their vertices, but in such a way that parallel lines are on the same side of it, as
∠OPQ and ∠RST, or (3) Two of the parallel lines lie on the same side of the line joining
the vertices, but the other two parallel lines lie on opposite sides of it, as in ∠UVW and
∠XYZ.

THEOREM 27: How to draw a line parallel to any straight line, and passing
through any point.

L P If I give you a straight line AB and a point P, can you

draw a straight line through P parallel to AB?
Absolutely.
A B
X

[1] Pick any point X on AB.

[2] Join PX, forming angle PXB.
[3] Draw PL so that ∠LPX = ∠PXB. (Thm. 20)
[4] That makes PL parallel to AB. (Thm. 23)

Q.E.F.

THEOREM 27 Remarks

1. In practice, for example in carpentry, this kind of construction is difficult to use

with accuracy. If ∠LPX is even slightly unequal to ∠PXB, then LP and AB will become
obviously not parallel fairly quickly.

THEOREM 27 Questions

A B
1. A more practical way to draw a line
X
parallel to another line AB through a point P is
to draw PA perpendicular to AB using a
P R carpenter's square, go out to the other end of
AB and draw BR also perpendicular to AB (and make it equal to AP in length), and then
join PR.

56
Prove that PR is parallel to AB by first assuming the opposite, namely that AB and PR do
meet, say out to the right at some point X. Use Angle-Angle-Side to show that the two
resulting triangles AXP and BXR must be equal – and say why that is impossible.

2. Now prove that quadrilateral ABRP is a rectangle.

THE TRIANGULAR ANGLE-SUM THEOREM

THEOREM 28: In any triangle the three angles added together equal two right
angles, and the exterior angle equals the two interior and opposite angles added
together.

A P Take any triangle you want, such as ABC, having

angles 1, 2, 3. Extend any side such as BC to any
2
point X forming exterior angle ACX.
I say that
4
3 1 5 (a) ACX = 2 + 3
B X (b) 1 + 2 + 3 = two right angles
C

[1] Draw CP parallel to BA (Thm. 27), thus dividing exterior angle ACX into the two
angles 4 and 5.

[2] 2=4 (since BA is parallel to CP; Thm. 25)

[3] 3=5 (since BA is parallel to CP; Thm. 25)

[4] 2+3=4+5 (putting together Steps 2 & 3)

Thus ACX = 2 + 3.

[5] 1+2+3=1+4+5 (equals added to equals make equals)

[6] 1 + 4 + 5 = two right angles (BCX is a straight line; Thm. 11)

[7] 1 + 2 + 3 = two right angles (putting together Steps 5 & 6)

Q.E.D.

57
THEOREM 28 Remarks

1. This theorem is actually quite surprising, if you think about it. Triangles can be
wildly different shapes from each other, and yet the three angles in any of them will
always add up to the same sum!

2. Recall Theorem 14, which says that ∠ACX has to be greater than ∠2 and also
greater than ∠3. Now we see, more precisely, that it is greater than either because it is
equal to the sum of both of them.

3. A corollary of this theorem, meaning “a side-result we get for free” is that Any
two angles of a triangle add up to LESS than 180°, because only all three together add up
to 180°. This corollary is the exact converse of the 5th Postulate.
Corollary: In a triangle, 2 angles add up to less than 180°
5th Postulate: 2 angles (on one side of a transversal) that add up to less than 180°
are angles in a triangle, i.e. the two outside lines forming the angles must eventually meet
and form a triangle.

4. Another corollary to the theorem is this: In any right triangle, the right angle is
the greatest angle. That has to be, since if there were another angle equal to or greater
than the right angle, the sum of the three angles in the triangle would be greater than
180°. Also, since the side opposite the greatest angle is the greatest, it follows that in any
right triangle the hypotenuse is the greatest side.
Yet another corollary is this: In any obtuse triangle, the obtuse angle is the
greatest. The same reason applies, and obviously the side opposite the obtuse angle must
be the greatest.

5. It also follows from this theorem that In any right triangle, the angles other than
the right angle are complementary to one another. Since the right angle plus the other
two equals two right angles, therefore the other two together equal one right angle, i.e.
they are complements of each other.

THEOREM 28 Questions

1. Draw any triangle on a piece of paper and label its angles 1, 2, 3. Next cut out the
triangle with a pair of scissors. Now tear off each angle of the triangle and line up the
three angles so that they all come to a single point. Do the three of them add up to a

58
straight line? See if you can come up with two reasons why this procedure does not
constitute a proof that the angles in every triangle add up to exactly to two right angles.

2. Using Theorem 28, you should be able to find the angle-sum of any rectilineal
figure, by cutting it into triangles and adding up the angles. Start with a quadrilateral:
what is the angle-sum of any quadrilateral? What is the angle-sum of any pentagon? Of
any hexagon? Can you find a general rule, so that given the number of sides in a convex
polygon, you can state the sum of its angles?
D

3. What if one or more of the sides of a polygon is

“dented in”? For example, take quadrilateral ABCD, where ∠ABC is
B “dented in”. Does that affect the sum of the angles in the figure?
C

A
R L 4. Prove Theorem 28 by drawing a straight line RAL through
2 point A parallel to the base BC.
3 1
B C

A A A
5. A
Place any straight edge (such as a
pencil) along BC, then rotate it through
∠BCA; from CA, next rotate the straight
B C B C B
B C C
edge through ∠CAB; from AB, finally
rotate the straight edge through ∠ABC. You have rotated the straight edge through the 3
angles of the triangle. Can you see that the straight edge has turned through exactly one
half a full rotation, namely 180°?

6. Using Theorem 17, prove that every angle in an equilateral triangle is acute.
Prove, further, that every angle in an equilateral triangle is 60°.
Can a scalene triangle be right? Acute? Obtuse?
Can an isosceles triangle be right? Acute? Obtuse?

7. Armed with Theorem 28, go back to Theorem 1 Question 2 and prove that the
figure is a rhombus.

8. In Theorem 19, we showed how to make a triangle given any three straight lines
(as long as any two of them summed up to more than the third line). Can we make a
triangle with any three angles? Yes, as long as the three of them add up to two right
angles.
Suppose ∠1, ∠2, ∠3 add up to 180°. Given a
straight line AB, can you make a triangle, using AB
1 2
as one of the sides, whose three angles are ∠1, ∠2,
3
and ∠3?

59
9. Since we now know that the angles of a triangle add up to two right angles, or
180°, it is possible to make a triangle whose angles are 30°, 60°,
B
and 90°, since these add up to 180°. Suppose triangle ABC is
just such a triangle. Prove that in such a triangle, AB is double
30
the length of AC, and furthermore that this triangle is actually the
E left half of an equilateral triangle. Start by making an equilateral
triangle on AC; since the angles of an equilateral triangle each
equal 60°, therefore ΔAEC will sit right inside angle BAC. And
A
60
C
since AE = AC, and AC < AB (since AB is the hypotenuse and
thus the longest side of triangle ABC), AE < AB, too, and thus E
will land somewhere between A and B. See if you can prove that it lands exactly in the
middle of AB, and you will be nearly done. (Hint: prove EC = EB first, using Thm.3.)

THEOREM 29: If two triangles have two angles equal to two angles, then the
remaining angle is equal to the remaining angle.

B Picture two triangles ABC & DEF which

are such that
E
2 1=4
2=5
5
Then it must also be true that
6 3=6
4
3 D
F Why? Well, because
1
C
A

[1] 1 + 2 + 3 = two rights (Thm. 28)

[2] 1=4
2=5 (the angles are given that way)

[3] 4 + 5 + 3 = two rights (putting together Steps 1 & 2)

[4] 4 + 5 + 6 = two rights (Thm. 28)

[5] 4+5+3=4+5+6 (putting together Steps 3 & 4)

[6] 3=6 (subtracting 4 + 5 from both sides)

Q.E.D.

60
THEOREM 29 Remarks

1. Just by applying Thm. 28, we see that given any two angles of a triangle, the third
can be determined. In fact, even if we do not know the two angles of a triangle
separately, but we know their sum, then we also know what the remaining angle must be.
For example, if ∠1 = 30°, and ∠2 = 50°, then since they add up to 80°, the
remaining angle, ∠3, must be 100° in order for all three of the angles to add up to 180°
(as Thm. 28 says they must).
And if ∠1 + ∠2 = 121°, then even if we don't know their values separately, we
still know that ∠3 must be 59°, in order for all the angles to add up to 180°.

2. Notice that we cannot say the same thing about the sides of a triangle, namely that
given any 2 of them, we can figure out the length of the third side. Suppose, for example,
that Side One = 5”, and Side Two = 6”. What is the third side? There's no telling. Could
it be 8”? Yes, since 5, 6, 8 are such that any two are greater than the third. Could it be
10”? Yes, since 5, 6, 10, are also such that any two are greater than the third. Could it be
3”? Yes, since 5, 6, 3 are also such that any two of them are greater than the third.

THEOREM 29 Questions

1. Are triangles ABC and DEF the same size? Are they the same shape?

2. If ΔABC has an angle of 35°, and another angle of 42°, then what is the value of
the remaining angle?

THEOREM 30: If one pair of opposite sides in a quadrilateral are both parallel
and equal, then the quadrilateral is a parallelogram.

Consider quadrilateral ABCD, in

A B
1 which AB & CD are both parallel
and equal. Then ABCD is a
parallelogram, meaning that AC &
BD are also parallel, and what’s
2
C more AC = BD.
D

On what grounds can we make such a claim?

61
[1] Join AD, forming ΔACD and ΔABD.

[2] ∠1 = ∠2 (since AB is parallel to CD; Thm. 25)

[3] AB = CD (these lines are given that way)

[4] AD is common to ΔACD & ΔABD.

[5] So, by the Side-Angle-Side Theorem, all the corresponding sides and angles of
ΔACD & ΔABD are equal.

[6] ∠CAD = ∠BDA (being corresponding angles of ΔACD & ΔABD)

[7] AC is parallel to BD (since ∠CAD = ∠BDA; Thm. 23)

[8] AC = BD (being corresponding sides of ΔACD & ΔABD)

Q.E.D.

THEOREM 30 Remark

We could also state Theorem 30 in another way: If two straight lines are parallel and
equal, then the straight lines joining their corresponding endpoints are also parallel and
equal.

THEOREM 30 Questions

1. Construct a parallelogram on a given straight line CD, and having a given point A
above CD as one of its corners.

2. Prove that every square is a parallelogram.

3. Prove that every rectangle is a parallelogram.

62
THEOREM 31: In any parallelogram, the opposite sides and angles are equal,
and the diagonal bisects the parallelogram.

B C If ABCD is a parallelogram, then by definition

5 3 AB is parallel to CD, and BC is parallel to
2 AD. But it will also be the case that
1
∠ABC = ∠ADC
4 6 ∠BAD = ∠BCD
A D AB = CD
BC = DA
And AC, the diagonal, will bisect the area of the parallelogram.

[1] 1=2 (since AB is parallel to CD; Thm. 25)

[2] 4=3 (since BC is parallel to AD; Thm. 25)

[3] AC is common to ΔABC & ΔADC.

[4] So, by the Angle-Side-Angle Theorem, all the corresponding sides and angles of
ΔABC & ΔADC will be equal, and the triangles themselves are equal in area.

[5] So the diagonal AC cuts the parallelogram ABCD into two triangles of equal area,
i.e. AC bisects the area of the parallelogram, and

[6] AB = CD (being corresponding sides of ΔABC & ΔADC)

[7] BC = DA (being corresponding sides of ΔABC & ΔADC)

[8] 5=6 (being corresponding angles of ΔABC & ΔADC),

that is, ∠ABC = ∠ADC.

[9] 1+4=2+3 (putting together Steps 1 & 2),

that is, ∠BAD = ∠BCD.

Q.E.D.

THEOREM 31 Remark

Theorem 31 is the partial converse of Thm. 30, which says that in a quadrilateral, if one
pair of sides is both parallel and equal, then the other pair is parallel and equal. 31 says
that in a quadrilateral, if both pairs of sides are parallel, then both pairs of sides are equal.

63
THEOREM 31 Questions

1. Prove that if ABCD is a quadrilateral, and AB = BC = CD = DA, then also

AB║CD and AD║BC. (In other words, prove that every rhombus is a parallelogram.)

2. Prove that if ABCD is a quadrilateral, and ∠ABC = ∠ADC and ∠BAD = ∠BCD,
then also AB║DC and AD║BC.
This, together with the solution to Q.1 above, proves the full converse of Thm. 31.

3. Prove that the diagonals of any parallelogram bisect each other.

4. Prove the converse: that if the diagonals of a quadrilateral bisect each other, then
the quadrilateral is a parallelogram.

5. Prove that the diagonals of a rectangle are equal.

6. Prove the converse, that if the diagonals of a quadrilateral are equal, then the
quadrilateral is a rectangle.

7. Prove that the diagonals of any equilateral quadrilateral are perpendicular to each
other; then prove the converse, that if the diagonals of a quadrilateral are perpendicular to
each other, the quadrilateral is equilateral (i.e. has all its sides equal to each other, and so
is either a square or a rhombus).

THEOREM 32: Parallelograms on the same base and in the same parallels have
equal areas.
A D C F
4
Imagine two parallels AF and BE that
1 3
contain two parallelograms ABED and
CBEF which stand on base BE. Then
2 ABED and CBEF have equal areas.
Surprised? Well, here's proof:
B E

[1] AD = BE (being opposite sides of a parallelogram)

CF = BE (being opposite sides of a parallelogram)
so AD = CF

[2] AD + CD = CF + CD (adding CD to both sides of AD = CF)

i.e. AC = DF

64
[4] AC = DF (Step 3)
AB = DE (being opposite sides of a parallelogram)
BC = EF (being opposite sides of a parallelogram)
so by the Side-Side-Side Theorem, ΔABC & ΔDEF are congruent triangles and so have
the same area.

[5] 1+4=4+3 (simply renaming the areas of the equal triangles)

[6] 1=3 (subtracting area 4 from each side of Step 5)

[7] 1+2=2+3 (adding area 2 to each side of Step 6)

[8] That is, ABED = CBEF

Q.E.D.

THEOREM 32 Remarks

1. Notice how parallelograms ABED & CBEF have the same area, but do not have
the same shape. One cannot simply be placed right on top of the other one so as to
coincide with it. This shows that things can be equal, or have the same quantity, without
being exactly identical in every way, e.g. without having the same shape.
The word congruent is commonly used to name the relationship between figures
that have not only the same size, but also the same shape, as we remarked after Theorem
2 (Remark 4). Congruent triangles, for example, would be those which are entirely
identical, according to the theorems of Side-Angle-Side or Side-Side-Side or Angle-Side-
Angle. You might say that “congruence” is a particular kind of equality, the most perfect
kind of equality.

2. Theorem 32 is a bit surprising. Parallelogram CBEF could be tilted toward the

right without limit, making it a wildly different shape from parallelogram ABED, and yet
the two would always have exactly the same area.

3. Notice it easily follows that parallelograms on

1 2 equal bases and in the same parallels have equal areas.
If parallelograms 1 and 2 are in the same parallels and
have equal bases, just slide 2 until its base coincides
with that of 1, and we are back to Thm. 32.

65
THEOREM 32 Questions

A C D F

1. Prove Thm. 32 in the case where C lands

between A and D.
B E

A C/D F
2. Prove Thm. 32 in the case where point C lands
right on point D.

B E

3. Prove the converse, namely that if parallelograms ABCD & EFGH have the same
area and their equal bases AB & EF are in a straight line, then their tops DC & HG must
D C L K
also be in a straight line. Start by assuming DC &
H
V
HG are not in a straight line, but, say, DC is higher
G
up, and extend DC to V, and EH & FG to L & K
forming a new parallelogram ELKF.
A B E F

THEOREM 33: Triangles on the same base and in the same parallels have equal
areas; and a triangle on the same base and in the same parallels as a parallelogram has
half the area of the parallelogram.

Let there be two triangles, ABC and ABG,

K G C L
standing on base AB and inside the same
set of parallels AB and GC.
Then ΔABC and ΔABG have the
same area (even if they don't have the
A B same shape). We prove it like this:

[1] Draw AK parallel to BG, making parallelogram ABGK.

[2] Draw BL parallel to AC, making parallelogram ABLC.

[3] ABGK = ABLC (Thm. 32)

66
[4] ΔABG = ½ABGK (Thm. 31)
ΔABC = ½ABLC (Thm. 31)

[5] ΔABG = ΔABC,

since by Steps 4 & 5 these two triangles are halves of two equal things, namely
the parallelograms ABGK & ABLC.

[6] Also, it is evident that ΔABC is equal to half of parallelogram ABGK, since it is
half its equal, namely parallelogram ABLC.

Q.E.D.

THEOREM 33 Remarks

1. Obviously, it follows that a triangle is half of any rectangle that stands in the same
parallels and has an equal base.

2. Notice that in Step 5 we use the principle that the halves of equal things are equal.

3. The converse for this Theorem can be proved in the same way that the converse
for Theorem 32 is proved (see Q.3 after Thm. 32). And, just like with Thm. 32, the two
triangles need not have the same base, but equal bases is enough.

4. What we call the “base” of a triangle is not any special side of it – any of the three
sides can be called its “base”. In this case we call the “base” the side which lies along
one of the parallels.

G F 5. Notice that since ΔABG and ΔDEF are

inside the same set of parallels, the
perpendicular dropped from G to the bottom
A B D E parallel will be equal to the perpendicular
dropped from F to the bottom parallel. For
that reason, these triangles can be said to have the same height or altitude. The “height”
or “altitude” of a triangle is the perpendicular drawn from its vertex to the base of the
triangle. And just as a triangle can have three “bases”, so too it has three “heights” (or
“altitudes”), namely one in relation to each base.

6. Obviously, the perpendiculars drawn from A and B to the

upper parallel will also be equal to the “height” of triangle ABG.
A B

67
THEOREM 33 Questions

1. If a rectangle has Side One = 3”, and Side Two = 4”, then what is the total area of
the rectangle in square inches? What operation does one perform on 3 & 4 to get the
right answer? Why is that the operation performed? Draw a diagram that shows your
procedure was right.

2. What if one side of a rectangle is 2 and the other is 2½? Will that change the
procedure? Draw a diagram of the rectangle, and divide it into squares showing that the
answer is still correct using the same procedure.
Accordingly, the formula for calculating the area of a rectangle is
A = bh
Where A means the area of the rectangle, and b means the length of its base, and h means
the height of the rectangle, and putting b and h right next to each other means to multiply
them. The result is the number of unit square areas in the rectangle.

3. Show that the formula for finding the area of a triangle is

A = ½bh
Where A means the area of the triangle, and b means the length of its base, and h means
the height of the triangle.

V 4. Look at triangle ABV. VH has been

dropped perpendicularly to AB extended.
VH = 8 feet
AB = 13 feet
A
B H
How many square feet of area is in triangle
ABV? Just apply the second part of Thm. 33.

68
THEOREM 34: In any parallelogram, the “complements” of any two
parallelograms about the diagonal are equal.

A H D Take a parallelogram ABCD and

6 2
join the diagonal AC. Pick any point
5
E
K K on AC, and draw
F
4
HKG parallel to AB
1 EKF parallel to AD
3
thus forming parallelograms EH and
B G C GF sharing the original diagonal as
their own diagonals, and also forming two “complementary” parallelograms EG & HF
(or 1 & 2). These complementary parallelograms must be equal in area, because . . .

[1] 5=6 (AK bisects parallelogram EH)

[2] 3=4 (KC bisects parallelogram GF)

[3] 5+3=6+4 (putting together Steps 1 & 2)

[4] 5 + 3 + 1 = 6 + 4 + 2,
since diagonal AC bisects the area of parallelogram ABCD.

[5] 1 = 2,
subtracting the equal areas 5 + 3 and 6 + 4 from each side of the equal areas in
Step 4.

Q.E.D.

THEOREM 34 Remarks

1. It is also true that if we join and extend GE, and then join and extend FH, and then
extend CA, these three straight lines will all meet each other in the same point! In fact,
that will still be true even if EF and HG do not pass through the same point along the
diagonal AC (just as long as they are each parallel to one of the sides of the original
parallelogram). We don’t have the tools to prove it, yet, though.

2. Notice that the little parallelograms EH and GF seem to be the same shape as the
original parallelogram BD.

69
THEOREM 34 Question

A G Y
Now prove the converse of Thm. 34. Let 1 & 2
E

X
2 (DVCH and GVFE) be parallelograms of equal
D F
V area drawn inside the opposite angles of the larger
1
parallelogram ACBE by drawing DVF║CB, and
C H Z B GVH║AC. Join AB. You should be able to show
that V lies on AB.
Start by assuming that V does not lie on diagonal AB, but instead DVF passes
through AB at another point X. Then draw YXZ║AC and see what happens when you
apply Thm. 34.

THEOREM 35: How to make a square.

E M If I give you a straight line such as AB, can you

build a square on it so that AB is one of the sides?
C No problem.
D K
2 3
[1] Draw AE perpendicular to AB, and make it
as long as you need (Thm. 9).
1 4
A B [2] Cut off AD = AB (by drawing a circle
around A, with radius AB).

[3] Draw DK parallel to AB (Thm. 27)

Draw BM parallel to AD (Thm. 27)
Call the intersection of DK and BM the point C.

[4] ABCD is a parallelogram (since AB║CD, AD║BC)

[5] 1 + 2 = two rights (Thm. 25)

but 1 = one right (we made AD ⊥ AB)
so 2 = one right

[6] But the angles opposite 1 & 2 are equal to them, since ABCD is a parallelogram
(Thm. 31),
so 3 = one right (being opposite ∠1)
and 4 = one right (being opposite ∠2)

[7] And the sides opposite AB & AD are equal to them, since ABCD is a
parallelogram (Thm. 31),
so DC = AB
and CB = AD

70
[8] But AB = AD (we made them so)

[9] So AB = AD = DC = CB (putting together Steps 7 & 8)

[10] So ABCD is equilateral (Step 9), and has all right angles (Steps 5 & 6), and so it
is a square (Def. 23)

Q.E.F.

THEOREM 35 Remarks

1. We have now made two regular rectilineal figures, the equilateral triangle and the
square. A regular rectilineal figure is one all of whose sides are equal and all of whose
angles are equal. We have made the 3-sided regular figure, and the 4-sided; we will learn
how to do the 5-sided, 6-sided, 8-sided, 10-sided, 12-sided, and 15-sided later.

Parallelogram
Rhombus 2. As you know, rectangles, rhombuses, and
squares are special kinds of parallelograms.
Moreover, a square can be considered as a special
kind of rectangle, namely one whose sides are all
Rectangle
Square equal. A square can also be considered as a special
kind of rhombus, namely one which has been
“squared up”.
Because a square is a parallelogram, its
diagonals bisect each other; because a square is a
rectangle, its diagonals are equal to each other; and because a square is equilateral (like a
rhombus), its diagonals are perpendicular to each other.

3. Make a square on two consecutive sides of your original square, and then add a
fourth square in between the two added squares. You now have one big square made out
of the four little squares.
Look at one of the little squares and compare it to the big square.
How do the lengths of their sides compare? How do their areas compare? Does
the area of a square increase in the same proportion as the length of its side increases?

THEOREM 35 Questions

1. Using a compass and a straight edge, what is the fewest number of steps in which
you can make a square on a given straight line as one of its sides? Count each circle or
straight line you make as one step. See if you can find the top two corners of the square in
5 steps.

71
2. As we will see later, among all quadrilaterals, the square has the greatest area for
a given perimeter, and conversely the least perimeter for a given area. We can say the
same thing about the equilateral triangle among all triangles. Proof will come later.

THE PYTHAGOREAN THEOREM

THEOREM 36: In any right triangle, the square on the hypotenuse is equal to
the sum of the two squares on the remaining sides.

ΔABC has a right angle at A. If we make a

H
square on each side, as ABFG, ACKH,
BCED, then the area of the square on
G
hypotenuse BC, namely BCED, is equal to
K
A the area of the other two squares combined.
F
Let’s prove it.

B C
[1] Drop AL perpendicular to DE, thus
forming rectangles BL & LC.
Join AD, join CF.

[2] Since ∠GAB & ∠BAC are both

right angles, therefore GAC is one straight
D E
L line (Thm. 12).

[3] FB = AB (ABFG is a square)

BC = BD (BCED is a square)
∠FBC = ∠ABD (each is ∠ABC plus a right angle)
so ΔFBC = ΔABD (Side-Angle-Side)

[4] 2ΔFBC = 2ΔABD (the doubles of equal areas are equal)

[5] 2ΔFBC = £ABFG,
since this triangle and parallelogram share a common base FB and stand within
the same parallels FB & GAC (Thm. 33).
[6] 2ΔABD = rectangle BL,
since this triangle and parallelogram share a common base BD and stand within
the same parallels BD & AL (Thm. 33).
[7] £ABFG = rect. BL (putting together Steps 4, 5, 6)
[8] £ACKH = rect. CL (by a corresponding argument)
[9] £ABFG + £ACKH = rect. BL + rect. CL
[10] rect. BL + rect. CL = £BCED (both parts of a whole equal the whole)
[11] £ABFG + £ACKH = £BCED (putting together Steps 9 & 10)

Q.E.D.

72
THEOREM 36 Remarks

1. This is the most famous theorem in all of geometry. The first general proof of it
is attributed to Pythagoras, a Greek philosopher who lived five centuries before Christ.
Legend has it he was so delighted by the Theorem that he went out and sacrificed an ox
in thanksgiving to the gods for so beautiful a truth.

2. In the diagram for this theorem, FC, AL, and BK all pass through one point! If
you draw a careful diagram, you can verify this for yourself – but we won’t burden
ourselves with a proof for it right now.

3. Numerically, the way to calculate the area of a square, as with any rectangle, is to
multiply the number of unit lengths in one side by the number of those same unit lengths
in an adjacent side. But since any two adjacent sides of a square are equal, it follows that
we need only multiply the numerical length of one side times itself to obtain the number
of unit squares in the whole square. The algebraic symbol for the product of any number
n times itself is n2. So if a is the hypotenuse of a right triangle, and b and c are its other
two sides, we can express the Pythagorean Theorem this way:

a2 = b2 + c2

THEOREM 36 Questions

E
3
4 1. Illustrate the Pythagorean Theorem by making a
5 2 1
G puzzle. If DEF is a right triangle, and DF the hypotenuse,
and FDKL the square on it, then extend KD to H on the
F D larger of the two other squares, and draw EG parallel to
DF. If you draw this carefully, and then cut out areas 1, 2,
3, 4, 5, you will see that they can be placed together to
make a square identical to DFLK.
L K

C F
2. We can now prove another “Triangle Congruence
Theorem”, although it is limited to right triangles. If we
have two right triangles ABC and DEF, and two sides of
one of them equal two corresponding sides in the other,
A B then the triangles are identical. This is true even if the
D E

corresponding sides are not around the right angle.

Let AB = DE, and BC = EF. Using the Pythagorean Theorem, show that these two
triangles are identical.

73
 b 3. See the two figures at left? Each is made by
a
c a
taking four identical right triangles with sides a,
b, c (c being the hypotenuse). Can you verify
=
b the Pythagorean Theorem by looking at these
c
two figures? In the left figure, we see that the
square on c is equal to the big square minus
four triangles. In the right figure, we see that the squares on a and b together are equal
to the big square minus four triangles.

4. Given the triangle at left, with legs of 6 and 8 containing a right

6
angle, how long is the hypotenuse?

5. Given the triangle at left, with one leg of 12 and the hypotenuse of
15
15, how long is the other leg?

12

13
6. Given that the hypotenuse of a right triangle is 13 units long,
and one leg is 12, how long is the other leg?
12

D C
7. ARB is a right triangle, AB its hypotenuse. AR = 4, RB =
3, and ABCD is a square. What is the area of the figure
R ARBCD?
4 3

A B

G H
E 8. EFG is a right triangle, EF its
T
hypotenuse. EF = 25, EG = 24, and
GFKH is a square. What is the area of
F K
triangle GTF?
X

25
9. XY = 15, XZ = 25. What is the area of triangle XYZ?
15

Z Y
74
THEOREM 37: If the square on one side of a triangle equals the sum of the
squares on the remaining sides, then the triangle is a right triangle.

A
Suppose you have a triangle ABC in which
£AB = £AC + £BC. Then the angle opposite
AB is a right angle, i.e. ∠ACB is a right angle.
Here's proof:

[1] Draw CN at right angles to AC, and cut

off NC = BC.
N C B
[2] Join NA.

[3] £AN = £AC + £NC (Pythagorean Theorem)

[4] £AN = £AC + £BC (By Step 1 NC = BC, so £NC = £BC)
[5] £AB = £AC + £BC (ΔABC is given this way)
[6] £AB = £AN (Putting together Steps 4 & 5)
[7] so AB = AN
[8] but NC = BC (Step 1)
[9] and CA is common to ΔANC & ΔABC
[10] so ΔANC ≅ ΔABC (Side-Side-Side)
[11] thus ∠ACN = ∠ACB (Corresponding angles of ΔANC & ΔABC)
[12] but ∠ACN is a right angle (We made it so)
[13] so ∠ACB is a right angle (Putting together Steps 11 & 12)

Q.E.D.

THEOREM 37 Remarks

1. Theorem 37 is simply the converse of the Pythagorean Theorem: so not only do

all right triangles have the Pythagorean property, but only right triangles have that
property.

2. It follows that the square on the side of an acute or obtuse triangle does not equal
the sum of the squares on the remaining two sides.

3. A triangle whose sides are 3 units long, 4 units long, and 5 units long is a right
triangle, and the longest side, 5, is its hypotenuse. This fact is often used in carpentry. If
you need to make a very large right angle, say at the corner of a house's foundation, then
a little carpenter's square won't do. You might be off (or the carpenter's square might be
off) of 90° by only a miniscule amount near the square, but by the time you extend
straight lines far from the square, you might have an appreciable difference between
where you end up and where a precise 90° angle would end up. You can get around this
by marking lengths along a string (with knots) that are 3, 4, and 5 units long; if you

75
stretch them tight and form a triangle with them (there is only 1 triangle you can form
with them!), then the angle between the sides of lengths 3 & 4 (feet or yards or whatever)
will be a right angle.

THEOREM 37 Questions

1. Using Theorem 37, prove that a triangle having one side 3 feet long, another side
4 feet long, and the remaining side 5 feet long is a right triangle, and that the 5-foot long
side is the hypotenuse.

2. Notice that 3, 4, and 5 fit the Pythagorean equation a2 = b2 + c2,

since 52 = 32 + 42

i.e. 25 = 9 + 16.

3, 4, and 5 are not the only whole numbers that do that. Any three whole numbers which
do that are called “Pythagorean Triples”. Here are all the Pythagorean Triples in which
no member is greater than 50:

3 4 5 14 48 50

5 12 13 15 20 25

6 8 10 15 36 39

7 24 25 16 30 34

8 15 17 18 24 30

9 12 15 20 21 29

9 40 41 21 28 35

10 24 26 24 32 40

12 16 20 27 36 45

12 35 37 30 40 50

In each case, the squares of the first two numbers, when added together, equals the square
of the third number. Obviously, some of these triples are based on others, e.g. 30, 40, 50,
is based on 3, 4, 5. Can you see any others that are based on 3, 4, 5?

76
 HOOK THEOREMS

(1) DESARGUES’ THEOREM.

V
As an inducement to further study, I will end
each chapter of this book with a theorem or two,
stated but left unproved. Each of these will have
some bearing on the material of the chapter to A
which it is appended, and in some cases it will be
possible to prove it by using the theorems of the C
chapter itself. But in most cases, it will simply
be a theorem of particular interest, illustrating
some of the wonders of geometry which grow B c Z
out of the elements covered in this book. I will a
call these “HOOK THEOREMS.” And as the
first of these “hooks”, I give you Desargues’ b
Y
Theorem: If two triangles, ABC, abc, in the
same plane, be such that the three straight lines
joining their vertices (aA, bB, cC) all intersect
in one point V, then the three points of X
intersection of their corresponding sides (X, Y,
Z) will lie in a straight line.

(2) SUM OF PERPENDICULARS FROM A POINT IN AN EQUILATERAL

TRIANGLE.

Here is another “hook theorem.” Take an equilateral triangle ABC, and pick any point P
inside it. Now draw the perpendiculars from P to the three sides: PQ, PR, PS. If you add
up those three lengths, their sum is equal to AT, the altitude of the triangle (the
perpendicular drawn from A to BC).
A

R
Q
P

C B
S T

77
(3) PAPPUS’ THEOREM.

Here is a third hook. If DF, KG be any two straight lines in a plane, and E, H be any
points on them, and we form three “X” figures by joining DH & EK (intersecting at L),
EG & FH (intersecting at N), DG & FK (intersecting at M), then L, M, N will lie in a
straight line.
F
E

N
M
L

K H G

78
Chapter Two

Squares
and
Rectangles
 DEFINITIONS

D C
1. A rectangle is said to be CONTAINED BY any two of
R its adjacent sides, and any two straight lines are said TO
CONTAIN a rectangle made with adjacent sides equal to
E F
them.
For example, rectangle R is contained by DE and EF.
And if A and B are two straight lines equal to DE and EF respectively, then R is
the rectangle which A and B contain.

(A + B) 2. This is not so much a definition, but more an

introduction of a way of to symbolize rectangles. Up till
now, we have designated a rectangle such as R by its four
(A - B)

corners, calling it “Rectangle DEFC”. But since every

rectangle is uniquely determined by the lengths of any two
of its adjacent sides, we can also use these to identify a
rectangle.
For example, we can designate DEFC by writing DE ⋅ EF, which means “The rectangle
contained by DE and EF.” Likewise, A⋅B means “The rectangle contained by lines A
and B.”
If we wish to name the side of any rectangle as the sum of or difference between
two straight lines, then we simply place parentheses around the sum or difference. For
example, (A + B)(A – B) means “The rectangle contained by (A + B) as one side, and by
(A – B) as the adjacent side.”

G 3. IDENTICAL RECTANGLES are those contained

by equal corresponding sides. We can also call such
H K rectangles CONGRUENT.
L For example, if GH = LM and HK = MN,
then GH ⋅ HK and LM ⋅ MN are identical
M N
rectangles, having the same shape and area.

79
 THEOREMS

THEOREM 1: The rectangle contained by any line L and the sum of two other
lines (A + B) is equal to the sum of the rectangles contained by L and A and by L and B.
A B
That is to say L (A + B) = L ⋅A + L ⋅B

L This is obvious just from the diagram.

Q.E.D.

THEOREM 1 Remarks

1. This simple theorem is not limited to just two lines. If we had a third line C, then
it is likewise true that the rectangle contained by L and (A + B + C) is equal to the sum of
the rectangles L ⋅A + L ⋅ B + L ⋅ C.

Y (X - Y)
2. This theorem is obviously true for subtraction as well,
i.e. L (X – Y) = L ⋅X – L ⋅Y
L

3. This theorem is a cousin to a very similar theorem about numbers, since the
number of square units of area in a rectangle is found by multiplying the numerical
measure of the sides. The theorem about numbers says Any number n multiplied by the
sum of any two numbers (m + p) is equal to the sum of the products n × m and n × p. As
with the rectangles, there can be more than two numbers in the sum. Thus if we add a
third number, r, it will be true that n (m + p + r) = n × m + n × p + n × r.

80
THEOREM 2: If through any point on the diagonal of a square two lines are
drawn parallel to each side of the square, then the four resulting figures inside are two
squares along the original diagonal and two identical complementary rectangles.
A B G
Take any square AGFD, and choose any random
point R along its diagonal DG. Now draw BRE
C H
R parallel to AD, and CRH parallel to DEF. I say that
CRED and BGHR are squares, and that ABRC and
RHFE are identical rectangles.

D E F [1] Look at rDAG and rDFG. Each is evidently

an isosceles right triangle, and because CR is
parallel to AG, rDCR is clearly an isosceles right triangle. Again, because ER is
parallel to FG, rDER is clearly an isosceles right triangle. Thus CRED is evidently a
square. For the same reasons, BGHR is also a square.

[2] Now CAB is a right angle, and AB║CR and AC║BR, so it is clear that ABRC is a
rectangle. Likewise RHFE is a rectangle. Moreover, since CR = RE (being sides of
square CRED) and BR = RH (being sides of square BGHR), therefore ABRC and RHFE
are identical rectangles.

Q.E.D.

THEOREM 2 Questions

1. What figures do we get if DR = RG?

2. What figures do we get if BE and CH are drawn parallel to the sides of the square,
but they do not intersect along either diagonal of the square?

3. What if AGFD were a rectangle? What four figures would we get? By drawing
two lines parallel to its sides, could we ever get a square? Could we ever get two
squares? Four squares?

4. What if AGFD were a rhombus? What four figures would we get?

5. Prove the converse, i.e. prove that if two little squares situated in opposite corners
of a big square touch at their corners, their common corner lies on the diagonal of the
original square (and the other two figures are identical rectangles).

6. Prove another form of the converse, i.e. prove that if two identical rectangles in
opposite corners of a square touch at their corners, that common corner lies on the
diagonal of the original square (and the other two figures are squares).

81
THEOREM 3: Given any two lines, the square on their sum is equal to the sum of
the squares on each plus twice the rectangle they contain.

D E F This theorem, while not fascinating in itself, is useful

for better upcoming theorems, and has the virtue of
being easily proved. Start with a figure like that in
L
R
M Theorem 2, namely a big square DFKG, its diagonal
GF, any point R on it, and LRM ║ DF, and ERH ║FK.

[1] A brief examination of the figure reveals that

G H K £DF = £LR + £EF + 2 DE · ER

[2] Now since LR = DE,

and since ER = EF,
we can substitute these lengths in the Step 1 equality, and say that
£DF = £DE + £EF + 2 DE · EF

[3] Which is to say that given any two lines DE and EF, the square on their sum
(£DF) is equal to the sum of the squares on them (£DE + £EF) plus twice the rectangle
they contain (2 DE · EF).

Q.E.D.

THEOREM 3 Remarks

1. This theorem has a cousin theorem about numbers, namely that The square of the
sum of two numbers is equal to the sum of the squares of each number plus twice their
product. That is, if n and m are our numbers, then (n + m)2 = n2 + m2 + 2nm.

D E F 2. This theorem was about the square on the sum of

two lines. Something similar is true about the square on the
L M
difference between two lines. Let DF and FE be the given
R lines, so that £DE is the square on their difference. Now
add a square HKZQ onto the figure. Then obviously
£LRHG + 2 DF⋅ FM = £DFKG + £HKZQ,
G H
K or £LRHG = £DFKG + £HKZQ – 2 DF⋅ FM
(subtracting 2DF⋅FM). Renaming these quadrilaterals by
Q Z their sides (or lines equal to their sides), we get
£DE = £DF + £EF – 2 DF⋅ FE.
In other words, given any two lines DF and FE, the square on their difference
(£DE) is equal to the sum of the squares on them (£DF + £EF) minus twice the
rectangle they contain (2 DF · FE). Q.E.D.

82
THEOREM 4: The difference between any two squares equals the rectangle
contained by the sum and the difference of their sides.

D E F N

Like Theorem 3, this theorem is not so much

interesting now as useful later. But it’s short! Lay
out the same figure as in Theorem 3, but this time
L
R M T
tack on rectangle FNTM, identical to DERL, or DE
· ER.
G H K

[1] Inspecting the diagram verifies that

£DF – £LR = £EF + 2 DE · ER
but DN · NT = £EF + 2 DE · ER
so £DF – £LR = DN · NT

[2] Since LR = DE
and DN = (DF + DE)
while NT = FM = EF = (DF – DE)
we can substitute these lengths in the equality of Step 1, and say that
£DF – £DE = (DF + DE)(DF – DE)

[3] In other words, the difference between any two squares (£DF – £DE) equals the
rectangle contained by the sum of their sides (DF + DE) and the difference between their
sides (DF – DE).

Q.E.D.

THEOREM 4 Remarks

1. The corresponding truth about numbers is that The difference between the squares
of two numbers is equal to the product of the sum of those numbers and the difference
between those numbers. That is, supposing the numbers are n and m, and n is the greater
of the two, then n2 – m2 = (n + m)(n – m)
For example 72 – 42 = (7 + 4)(7 – 4).

83
THEOREM 5: How to make a single rectangle equal to any two given rectangles.

B A M T
P Suppose you have two rectangles
with no length of any side common
to both, and yet you would like to
make one big rectangle equal to the
two of them. Easy. Put their corners
together as ABCD and DEFG in
such a way that CD and DG are in a
G L straight line. Extend BA out to P,
R
C D and CDG out to R, as far as needed.
E F K
Next,

[1] Join AG.

[2] Extend AG and EF to where they intersect at K.

[3] Extend FG up to M, and KL up to T, thus completing the large rectangle AEKT,
and the two rectangles along its diagonal, AMGD and GLKF.
[4] Now, DEFG = MGLT
since these rectangles are complementary parallelograms (Ch.1, Thm. 34).

[5] And MG = AD (being opposite sides of rectangle AMGD)

[6] Therefore MGLT is equal to DEFG in area, but has one side equal to a side of
rectangle ABCD. So if we slide MGLT over until it is right up against ABCD, the two of
them will form one rectangle together. And the area of that rectangle will be equal to
ABCD + DEFG.

Q.E.F.

THEOREM 5 Remarks

1. Notice that we can use this construction to add together as many rectangles as we
please into one big rectangle. Suppose, instead of just two rectangles, you had three,
namely 1, 2, and 3, and you wanted to make a big rectangle equal to the three of them.
Using the theorem, make a rectangle R which is equal to 1 + 2. Using the theorem again,
make a rectangle T which is equal to R + 3. Since R = 1 + 2, thus T = 1 + 2 + 3.

2. Notice that rectangles DEFG and MGLT have the same area but they do not have
the same shape. That leads us to expect a difference in their perimeters, or the total
length around their sides. In fact, it is quite possible for two different rectangles to
contain the same area, but with unequal perimeters.

84
 For example, consider two rectangles, one of which is 2” × 6”, the other of which
is 3” × 4”. Each has an area of 12 square inches, but the 2” × 6” rectangle has a perimeter
of 16 inches, whereas the 3” × 4” rectangle has a perimeter of only 14 inches.
6 A 2” × 5” rectangle also has a perimeter of
Area = 12
2
Perimeter = 16
14 inches, but it has less area than the 3” × 4”
rectangle, i.e. only 10 square inches of area.
Stranger still, a 1” × 7” rectangle has a bigger
4
perimeter than the 3” × 4” rectangle, namely 16
Area = 12 inches, and yet it has less area (namely 7 square
3
Perimeter = 14
inches). So if your yard is 30 yards by 40 yards,
and mine is 10 yards by 70 yards, then the fence
5 around your yard is shorter than the fence around
2
Area = 10 my yard, and yet your yard is bigger than mine.
Perimeter = 14

7
Area = 7
1
Perimeter = 16

THEOREM 6: If a perpendicular is drawn from a circle’s circumference to its

diameter, the square on it equals the rectangle contained by the segments of the diameter.

P
In any circle with center M and any diameter
AMB, choose any point along the circumference
and drop PR perpendicular to AB. I say that
£PR = AR ⋅ RB.
A M R B

The proof:

[1] Join PM.

[2] Now, £PR + £MR = £MP (Pythagorean Theorem)

so £PR = £MP – £MR (subtracting £MR from each side)

[3] Thus £PR = £MB – £MR (MP = MB, being radii of the circle)

[4] But £MB – £MR = (MB + MR)(MB – MR),

because of Theorem 4 above, the theorem about the difference of two squares.

[5] Thus £PR = (MB + MR)(MB – MR),

putting Steps 3 & 4 together.

85
[6] i.e. £PR = (AM + MR)(MB – MR),
since AM = MB, both being radii of the circle.

[7] But (AM + MR) is just AR,

and (MB – MR) is just RB.
So (AM + MR)(MB – MR) = AR ⋅ RB.

[8] Thus £PR = AR ⋅ RB (putting together Steps 6 & 7)

Q.E.D.

THEOREM 7: How to make a square that has the same area as any given
rectilineal figure.
B
Suppose you have a rectilineal figure ABCDE
with as many sides as you want (in this case,
1 five). How can you make a square that is equal
A C to it in area? The general procedure is to cut it
2 3 up into triangles, turn each triangle into a
E
D
rectangle, turn all the rectangles into one big
rectangle, and turn that big rectangle into a
J
F H B G P square.

1
[1] You can always divide a rectilineal
figure into triangles by joining its vertices
A M C
together, say into Δ1 and Δ2 and Δ3.
Q
[2] Now make a rectangle equal to each
triangle.
To do this, just take each triangle, such
K
O
R
as Δ1, and draw a line JP through its vertex
L parallel to its base AC. Next, draw AF and CG
P N
perpendicular to base AC, forming rectangle
ACGF. Since Δ1 and ACGF are on the same
base and in the same parallels, therefore Δ1 is half of rectangle ACGF (Ch.1, Thm. 33).
Bisect AC at M, and draw MH perpendicular to base AC, dividing ACGF into
two rectangles AFHM and HMCG. These rectangles are identical, since AM = MC and
since HM is a side common to both. Since they have equal areas, and since the two
together make up rectangle ACGF, either one of these rectangles is half of rectangle
ACGF.

86
 But that means each of these rectangles is equal to Δ1, which is also half of
rectangle ACGF. That is how to make a rectangle equal to a triangle. And so, equal to
Δ1, Δ2, Δ3 etc., we will have rectangle 1, rectangle 2, rectangle 3 etc.

[3] Now make a rectangle equal to rectangle 1 + rectangle 2, using Theorem 5 above.
Next, make a rectangle equal to that rectangle plus rectangle 3, again using
Theorem 5. The result is a rectangle equal to Δ1 + Δ2 + Δ3. Continuing this process as
often as needed, we can make a rectangle equal to any number of triangles added
together.

[4] Let KLNP be the finished product, namely a rectangle whose area is equal to
Δ1 + Δ2 + Δ3.

[5] Extend KL to R so that LR = LN.

Bisect KR at O and make a circle with center O and radius OR.
Extend NL to Q on the circle's circumference.

[6] Now £QL = KL ⋅ LR (Ch.2, Thm.6)

[7] But KL⋅LR = Δ1 + Δ2 + Δ3 (Step 4)

[8] So £QL = Δ1 + Δ2 + Δ3 (putting together Steps 6 & 7)

[9] i.e. £QL = ABCDE (ABCDE is composed of those 3 triangles)

So if we build a square on QL, it will be equal in area to the figure ABCDE.

Q.E.F.

THEOREM 7 Remarks

1. This theorem obviously does not enable us to make a square equal to any plane
figure, but only to a rectilineal figure. The construction requires us to resolve the given
figure into triangles, which we can’t do if we are given, for example, a circle.

2. To make a square equal to a given rectangle, we can skip right down to Step 5 in
the construction and start making our semi-circle. There is no need to divide the
rectangle into triangles.

87
THEOREM 7 Question

ABCD is an isosceles trapezium, which means B C

that AB = CD and BC║AD. Let BE and CF each

be drawn perpendicular to BC. Supposing that 18
30
AE = 10 feet long,
10
BE = 18 feet long, A E F D
EC = 30 feet long,
find out how long is the side of a square with an area equal to the area of trapezium
ABCD.
Start by noticing that BCEF is in fact a rectangle (which you can prove because
BC║AD). Then prove that ΔABE ≅ ΔDCF. Then use the Pythagorean Theorem to
determine the length of BC. Having done that, you will be able to calculate the areas of
all the triangles in the trapezium. Adding these together, you will have the area of the
trapezium itself, and taking the square root of that area, you should be able to find the
length of the side of the square equal to ABCD.

THEOREM 8: An isosceles triangle has less perimeter than any other triangle
under the same height and on an equal base.

This theorem, along with Theorems 9 and 10, is interesting in its own right, but I include
all of them here for the sake of the upcoming Theorem 11 ...
Suppose ABC is an isosceles
A
P triangle. Draw AP║BC. Let ARK be any
other triangle in the same parallels (slid
over, if necessary, until its vertex
coincides with vertex A of rABC), and
B K let its base RK be equal to BC – thus the
G R C
two triangles have equal areas.
Nonetheless, I say that the perimeter of
ΔABC is less than the perimeter of ΔARK.

[1] Cut off BG = RC. Join AG. Thus it is evident that rARG is also isosceles.

[2] Now GC = BC – BG (obviously)

but BC = RK (given)
and BG = RC (we made it so)
so GC = RK – RC
i.e. GC = CK

88
[3] Thus if we extend AC to E so that AC = CE as well, it is obvious that AKEG is a
parallelogram, since AC = CE and GC = CK.

[4] Now EK + AK > AE (since AKE is a triangle)

or GA + AK > 2AC (GA = EK, opp. sides of a parallelogram)
or GA + AK > BA + AC (BA = AC)
thus RA + AK > BA + AC (GA = RA; Step 1: rARG is isosceles)

[5] If we now add the lines RK and BC, given as equal, to either side of this
inequality, we get
RA + AK + RK > BA + AC + BC
which is to say that the perimeter of rARK is greater than that of rABC.

Q.E.D.

THEOREM 9: Given any quadrilateral which is not equilateral, how to make an

equilateral quadrilateral with the same area but less perimeter.

Suppose ABCD is a quadrilateral, but it is not equilateral. The following steps will make
an equilateral quadrilateral equal to it, but having less perimeter.
X

[1] Join any diagonal, say AC.

F C
Y Draw WBX and YDZ parallel to AC.
L
Draw FCG and EAG perpendicular to WX.
B
G
Thus EFGH is a rectangle.
K

[2] Bisect its sides at K, L, M, N. Join KM and

join NL thus cutting EFGH into four identical
M
E
rectangles, and forming at the same time the
A D obviously equilateral quadrilateral KLMN (whose
W
N sides are not joined in the diagram, in order to
H avoid clutter).
Z

[3] Now it is evident that rAKC is isosceles, and so too rAMC.

Thus rAKC = rABC, but has less perimeter (Ch.2, Thm.8)
and rAMC = rADC, but has less perimeter (Ch.2, Thm.8)
so quadrilateral AKCM = quadrilateral ABCD, but has less perimeter

[4] Again it is clear that rKLM is isosceles, and so too rKNM.

Thus rKLM = rKCM, but has less perimeter (Ch.2, Thm.8)
and rKNM = rKAM, but has less perimeter (Ch.2, Thm.8)
so quadrilateral KLMN = quadrilateral AKCM, but has less perimeter

89
[5] Putting together Steps 3 & 4, it is clear that quadrilateral KLMN equals
quadrilateral ABCD, but has less perimeter. And KLMN is equilateral.

Q.E.F.

THEOREM 9 Remark

It is of course possible that ABC should be isosceles, and thus AKC would have the same
perimeter as ABC, coinciding with it, and not less perimeter. But that does not affect the
proof, since ABCD has to have some unequal adjacent sides, being given as not
equilateral.

THEOREM 10: Any square of the same area as a rhombus has less perimeter
than the rhombus.

K L E M Let ABCD be a rhombus. Then the square

which has the same area as it must have a
G
B C lesser perimeter. To see why, simply

[1] Draw DG perpendicular to AD.

Clearly DG is less than DC, the side of the
rhombus, since DC is the hypotenuse in
A D
right triangle DGC. So extend DG to E so
that DE = DC, the side of the rhombus, and complete square DEKA. Extend AB and DC
to L and M on the uppermost parallel.

[2] DEKA is a square with the same perimeter as rhombus ABCD, because they both
have four sides equal to AD.
But DEKA = ALMD (parallelograms on same base, under same height)
and ALMD > ABCD (whole is greater than part)
so DEKA > ABCD

[3] Now any square of less area than DEKA has less perimeter than it.
But the square equal to ABCD has to be of less area than DEKA (by Step 2).
Therefore the square equal to ABCD has less perimeter than DEKA,
i.e. less perimeter than ABCD.

Q.E.D.

90
THEOREM 11: A square has the least perimeter of all quadrilaterals with the
same area, and the greatest area of all quadrilaterals with the same perimeter.

PART ONE: Suppose you have a square S and any other

S A non-square quadrilateral A having the same area. Then
the perimeter of S is less than the perimeter of A. Proof:

[1] If A is a rhombus, then it is clear that square S has

R less perimeter, since the two of them have equal areas
(Thm.10).

[2] If A is not a rhombus, and thus is not equilateral, then make an equilateral
quadrilateral R with the same area as A but less perimeter (Thm.9). If R happens to be a
square, then since it has the same area as square S, it will be identical to it, and thus again
it is clear that S, just like its twin R, must have less perimeter than A.

[3] If R is a mere rhombus, then it is clear that square S has less perimeter than it,
since they have equal areas (Thm.10). But R was made with less perimeter than A.
Therefore S has still less perimeter than A. Q.E.D.

PART TWO: Now suppose you have a square S and any

S P other non-square quadrilateral P having the same
perimeter. Then the area of S is greater than the area of
P.

E Make square E equal in area to P (Thm.7). Since square

E is equal to P, it has a lesser perimeter than P (by Part
One above). But square S has a perimeter equal to that of P, and thus has a greater
perimeter than that of square E. But a square with a greater perimeter has a greater area.
So S is greater than E, and therefore is greater than P.

Q.E.D.

91
THEOREM 11 Remarks

1. We have seen that a rhombus can be made with the

same area as any non-equilateral quadrilateral, while using
less perimeter. It is easy to see that a rectangle can be made
with an area equal to any rhombus, while using less perimeter
(just take the rectangle on the same base and under the same
height and you will see it – remember that a hypotenuse is the longest side of any right
triangle). Now in Theorem 11 we see that the square (which is equilateral and right
angled) contains the maximum quadrilateral area for any perimeter, and also that it has
the minimum perimeter for any given quadrilateral area.
It can similarly be proved (in fact, more easily), using Theorem 8, that the
equilateral triangle has, among triangles, the maximum area for a given perimeter, and it
also has, among triangles, the minimum perimeter for a given area.
There is a lesson here: the more uniform the rectilineal figure is, the greater an
area it can encompass with a given perimeter, and the less perimeter it will need to
encompass a given area.

G D K
C 2. But what about comparing the
Q U
equilateral triangle and the square? Which
one of them makes “better” use of a given
perimeter? Which one holds more area for a
given perimeter? Let’s see. Say equilateral
A E B S R
triangle ABC and square SQUR each have a
perimeter of 12 inches.
Then SR = 3
and AB = 4.

Drop CE at right angles to AB.

Since CB is the hypotenuse in ΔCEB, thus CB > CE.
Make ED = CB, so now ED = 4.
Complete square AGKB.
Since ΔABD > ΔABC
thus 2ΔABD > 2ΔABC
so AGKB > 2ΔABC, since AGKB is equal to 2ΔABD, being a parallelogram on the
same base with rABD and under the same height. But AGKB, being a square whose
side is 4 inches, has an area of 16 square inches. Therefore ΔABC has an area less than
half of that, i.e. ΔABC is less than 8 square inches.
On the other hand, square SQUR has an area greater than 8 square inches; since
its side is 3 inches, it has an area of 9 square inches.
So when a square and an equilateral triangle have the same perimeter, the square
has a greater area.
There is another lesson here: the greater the number of sides in the uniform
rectilineal figure, the more area it can hold with a given perimeter.

92
3. If among rectilineal figures those which are more uniform and have a greater
number of sides can hold more area within a given perimeter, what plane figure would
hold the most area within a given perimeter?

THEOREM 12: How to extend a straight line so that the square on the original
line is equal to the rectangle contained by the whole new line and the extension.

Suppose you had a straight line AS, and you

G K want to extend AS to some point B so that
£AS = AB · BS
Here’s how to do it:
[1] Draw any square ASKG.
[2] Bisect AS at C.
D A C S B [3] Join CG; join CK.

[4] Now, AC = CS (each is half of AS)

and AG = KS (being two sides of square ASKG)
and ∠CAG = ∠CSK (each is a right angle in square ASKG)
so the corresponding sides of ΔCAG and ΔCSK are equal. (Side-Angle-Side)

[5] So CG = CK (being corresponding sides of ΔCAG and ΔCSK)

Thus the circle around C as center and with radius CG will pass through both G
and K. Draw it and call the points where it cuts AS extended “D” and “B”.

[6] Since SK is perpendicular to DB, the diameter of the circle, therefore

£SK = DS⋅BS (Ch.2, Thm. 6)

[7] But £SK = £AS (SK = AS, being two sides of square ASKG)
So £AS = DS⋅BS

[8] Now CD = CB (being radii of the circle)

and CA = CS (each is half of AS)
so DA = SB (being remainders of equals with equals subtracted)
so DA + AS = AS + SB (adding the same thing, AS, to both sides)
i.e. DS = AB

[9] But £AS = DS⋅BS (Step 7)

so £AS = AB⋅BS (DS = AB; Step 8)

Q.E.F.

Line ASB is cut at S in a famous ratio called “The Golden Section,” which we will learn
more about in Chapter 6. In the next chapter, we will need this construction to make a
regular pentagon.

93
“HOOK”: MARION’S THEOREM.

In 1993 Mathematics Teacher published this interesting theorem: Take any triangle
ABC, and cut each side into three equal segments. Then join each vertex to the two
section-points on the opposite side. This will form a hexagonal figure in the middle.
This hexagonal figure has an area exactly one tenth that of the whole triangle. Similar
theorems result from cutting the opposite sides into other odd-numbered portions.
A

D G
Y
M E

N Q
J S
T
V Z
O
P
R W

B C
K F

94
Chapter Three

Circles
 DEFINITIONS

1. EQUAL CIRCLES are those with equal radii.

2. A CHORD of a circle is any straight line joining two points on D

its circumference.
For example, KD is a chord.

A C

3. An ARC of a circle is any portion of its circumference.

For example, ARC is an arc.

4. The side of a circular arc on which lies the straight line joining its endpoints is its
CONCAVE side. The other side is its CONVEX side.
For example if FH, the straight line joining the G
endpoints of arc FGH, lies below FGH, then the underside
of arc FGH is concave, and its top side is convex. F H

5. A STRAIGHT LINE TANGENT TO A CIRCLE is one

which touches the circle at some point but does not go inside it
when it is extended in either direction.
For example, ATB is tangent to the circle at T.
A T B

95
 6. Two CIRCLES EXTERNALLY TANGENT TO EACH
OTHER are those which touch each other at some point on
their convex sides, but do not cut into each other there. A
CIRCLE INTERNALLY TANGENT TO ANOTHER N X
CIRCLE is one whose convex side touches the concave side
of another circle at some point without cutting out of it there.
For example, the two circles outside each other are
externally tangent at X, and the circle inside the other circle
is internally tangent to it at N.

G 7. A SEGMENT OF A CIRCLE is a figure

contained by any arc of the circle’s circumference
S and the straight line (called the BASE of the
segment) joining the endpoints of that arc.
For example, S is a segment of a circle, and
so is G.

D
8. An ANGLE IN A SEGMENTis the angle joining the
endpoints of a segment's base to any point on its
circumference.
For example, CDE is an angle in a segment.
C E

N
9. A rectilineal angle drawn inside a circle
is said to STAND ON the arc that it cuts off,
K and to be AT THE CENTER if its vertex
H
lies on the center of the circle, but AT THE
O
CIRCUMFERENCE if its vertex lies on the
L M circumference of the circle.
For example, angle HKL stands on
arc HL, and it is at the center of the circle. Angle MNO stands on arc MO, and it is at
the circumference.

10. A rectilineal figure is said to be INSCRIBED IN A CIRCLE C

when the vertex of every one of its angles lies on the circle's B
circumference. When this happens, the circle is said to be
CIRCUMSCRIBED AROUND A RECTILINEAL FIGURE. A D
For example, ABCDE is inscribed in a circle, and the
circle is circumscribed around it. E

96
11. A rectilineal figure is said to be CIRCUMSCRIBED H

AROUND A CIRCLE when every side of it is tangent to the

K
circle. When this happens, the circle is also said to be G

INSCRIBED IN A RECTILINEAL FIGURE.

For example, GHKLM is circumscribed around a circle,
and a circle is inscribed in it. L
M

12. A REGULAR POLYGON is one all of whose sides are equal and all of whose angles
are equal.
For example, a square, if we call it a polygon (often the name "polygon" is
reserved for rectilineal figures having more than four sides), is a regular polygon, since
all four of its sides are equal AND all four of its angles are equal. On the other hand, a
rhombus is not a regular figure, since not all of its angles are equal, although all its sides
are equal. Likewise a rectangle, although its four angles are all equal, is not regular,
since not all its sides are equal. A regular polygon must be both equilateral and
equiangular.

97
 THEOREMS

THEOREM 1: How to find the center of a circle.

D Up till now, we have been making our own circles, and so

we started with the center. But what if I make my own
circle, and then give it to you without showing you where
X the center is that I used to make it? Can you find the
M center? Here is the way to do it:

A C B [1] Pick any two points A and B on the circumference.

E
[2] Join AB, and bisect it at C.

[3] Draw CD perpendicular to AB, extend it to E, and bisect DE at M.

Then M is the center of the circle. Why? First of all, the center has to lie somewhere
along DE, for if it did not, but was somewhere else like the point X, then

[4] Join XA, join XC, join XB.

[5] Now AX = XB (X is supposedly the center of the circle)

AC = CB (we bisected AB at C)
CX is common to ΔACX and ΔBCX
so the corresponding angles of ΔACX and ΔBCX are equal,
because of the Side-Side-Side Theorem.

[6] So ∠ACX = ∠BCX (being corresponding angles of ΔACX & ΔBCX)

But these two angles are adjacent, and therefore they are both right angles.
So ∠ACX is a right angle.

[7] Thus ∠ACX = ∠ACD (since both are right angles; we made ∠ACD right)
But that is impossible, since ∠ACD is only part of ∠ACX.

[8] Therefore our initial assumption, namely that the center of the circle does not lie
on the line DE, is itself impossible. So the center of the circle does lie on DE
somewhere.

[9] Since the center of the circle is equidistant from D and E, it must lie at the
midpoint of DE, that is, at M.

Q.E.F.

98
THEOREM 1 Remarks

1. From this Theorem it is plain that Whenever a line bisects any chord of a circle at
right angles, the center of the circle must lie on that line.

2. Plainly, then, if the perpendicular bisectors of two different chords in a circle

intersect each other at one point, that point must be the center.

THEOREM 2: Any chord of a circle falls entirely inside that circle.

Let A and B be any two points on a circle. Then every point

C
between A and B lies inside the circle. Proof:

A R B [1] Choose any random point R along AB. Join CR.

[2] Thus ∠CRA > ∠CBR (∠CRB is exterior to rCRB)

but ∠CAR = ∠CBR (rACB is isosceles)
thus ∠CRA > ∠CAR
and so the sides opposite these angles in rCAR are unequal in the same order, that is
CA > CR

[3] So CA, the radius of the circle, is greater than CR. Therefore R lies inside the
circle. Thus every point along AB lies inside the circle.

Q.E.D.

THEOREM 2 Remarks

1. This Theorem is rather obvious for big chords drawn inside the circle, but what
about very small ones? The smaller the chord is, the closer it gets to the circumference of
the circle, and the less distinguishable from the circumference it becomes. Without this
Theorem, we might think that very small chords could actually coincide with a very small
part of the circumference, or that a very small portion of a circle is no different from a
straight line.

99
2. A corollary of this Theorem: no straight line can have more than two points in
common with the circumference of a circle. So a straight line cannot cut the
circumference of a circle more than twice.

THEOREM 3: If a straight line through the center of a circle bisects a chord not
drawn through the center, it cuts it at right angles.

Suppose AB passes through C, the center of a circle, and it

C also bisects chord DE at M. Then AMB is at right angles to
DME. Here's why:
D M E

[1] Join CD; join CE.

[2] Now CD = CE (they are radii)

and MD = ME (it is given that M bisects DE)
and MC is common to ΔCMD and ΔCME
so the corresponding angles of ΔCMD and ΔCME are equal (Side-Side-Side).

[3] Thus ∠CMD = ∠CME, being corresponding angles of ΔCMD and ΔCME.
But these two angles are adjacent; therefore they are each right angles.
Thus AMB is at right angles to DME.

Q.E.D.

THEOREM 3 Question

Why must we stipulate that the bisected line is not through the center?

100
THEOREM 4: If a straight line through the center of a circle cuts a chord at
right angles, it bisects it.

Suppose AB passes through C, the center of a circle, and

C
also cuts DE through M at right angles. Then DM = ME.
Here's why:

D M E

[1] Join CD; join CE.

[2] Now CD = CE (they are radii)

and ∠CDM = ∠CEM (because ΔCDE is isosceles)
and ∠CMD = ∠CME (they are given as right angles)
so the corresponding sides of ΔCMD and ΔCME are equal,
because of the Angle-Angle-Side Theorem.

[3] Thus DM = ME (being corresponding sides of ΔCMD & ΔCME)

Q.E.D.

THEOREM 5: If two straight lines in a circle cut each other, but their
intersection is not the center of the circle, then they do not bisect each other.

C
Imagine that AB and CD are drawn inside a circle, and
they cut each other at a point X, but X is not the center
of the circle. Then X cannot bisect both AB and CD.
Here's what happens if you suppose it does:
B
M
X [1] Assume CX = XD
and AX = XB
A D
[2] Find M, the center of the circle. (Thm. 1)

101
[3] Join MX.

[4] Now ∠CXM is a right angle, since M is the center and X is supposedly the
midpoint of CD (Thm. 3).

[5] But ∠BXM is a right angle, too, since M is the center and X is supposedly the
midpoint of AB (Thm. 3).

[6] So ∠CXM = ∠BXM,

that is, the whole is equal to the part, which is impossible.
So it is also impossible for AB and CD to bisect each other.

Q.E.D.

THEOREM 5 Questions

1. What if one of the lines is a diameter? Does that affect the proof?

2. Can one of the two lines (which cut each other not through the center) be bisected
by the other?

THEOREM 6: If two circles cut each other, they cannot have the same center.

C B
Conceive of two circles, circle A and circle B,
that cut each other at some point C. Then they
have different centers. Why?
M
A X
P

[1] Let M be the center of circle A. Join MC.

[2] Since circles A and B cut each other, each passes into and back out of the other, so
choose any point P on circle B that is outside circle A. Join MP.

102
[3] Since M is inside circle A (being its center), and P is outside circle A (that is how we
chose it), thus MP must pass out of circle A at some point, X. Thus MX is only part of
MP, and so
MX < MP
but MX = MC (these being radii of circle A)
so MC < MP

[4] Hence MC and MP, being unequal, are not radii of circle B. And thus M cannot be
the center of circle B. But M is the center of circle A. Therefore circles A and B have
different centers.

Q.E.D.

THEOREM 6 Remarks

The flip side of this Theorem is this: If two circles DO have the same center, then they
don't cut each other.

THEOREM 7: If two circles touch one another, then they cannot share the same
center.

Imagine circle A touching circle B internally at a

M X P point T. These two circles cannot possibly have the
same center. Why not?
A
[1] Let M be the center of circle A. Join MT.
B

[2] Since circle A touches circle B inside it, then at least some of circle B must be
outside of circle A. Choose any point P on circle B that is outside circle A. Join MP.

[3] Since M is inside circle A (being its center), and P is outside circle A (that is how we
chose it), thus MP must pass out of circle A at some point, X. Thus MX is only part of
MP, and so
MX < MP
but MX = MT (these being radii of circle A)
so MT < MP

103
[4] Hence MT and MP, being unequal, are not radii of circle B. And thus M cannot be
the center of circle B. But M is the center of circle A. Therefore circles A and B have
different centers.

Q.E.D.

THEOREM 7 Remarks

1. Notice that the proof of the Theorem is only about internally tangent circles.
What about externally tangent circles? Can they have the same center? Obviously not.
If they are outside each other, and touch each other at some point but don't cut into each
other there, then clearly they have different centers, because each circle's
center is inside it, and therefore outside the other one. Besides, nothing
prevents us from making an argument similar to that above for the case
of externally tangent circles.

2. Taking Theorems 6 and 7 together, we see now that If two distinct circles DO
have the same center, then their circumferences do not have any points in common at all.
Circles which have the same center are called "concentric" circles – like the circular
waves on the surface of a pond into which we have dropped a pebble.

THEOREM 8: If three lines drawn from one point inside a circle to three points
on its circumference are equal to each other, then that point is the center.

B
Imagine a circle with a point M inside it, and three
E
points along the circumference, A, B, C. If MA =
C MB = MC, then M has to be the center of the circle.
M Now the proof.

D
[1] Bisect BC at E; join ME.
A
[2] Bisect AC at D; join MD.

[3] Now AM = MC (we are given that the three original lines from M are equal)
and AD = DC (we bisected AC at D)
and MD is common to ΔAMD and ΔCMD
so the corresponding angles of ΔAMD and ΔCMD are equal (Side-Side-Side)

104
[4] Thus ∠ADM = ∠CDM, being corresponding sides of ΔAMD and ΔCMD. But
these two angles are also adjacent; therefore they are right angles.

[5] So MD is at right angles to AC (Step 4) and it also bisects AC (Step 2), and
therefore the center of the circle lies somewhere along MD, as we saw in the way to find
a center (Ch.3, Thm.1).

[6] But likewise we can prove that the center has to lie along ME somewhere, since
ME will also be at right angles to BC, and it bisects it.

[7] Therefore the center of the circle is a point common to both MD and ME. But M
is the only point common to ME and MD (since two different straight lines can never
have more than one point in common, because they cannot cut each other twice).
Therefore M is the center of the circle.

Q.E.D.

THEOREM 8 Remarks

1. Why 3 lines? If 2 lines drawn from one point to the circumference of a circle are
equal, isn't that enough to conclude that the point is the center? No. Just look back at the
diagram for this Theorem. AD = DC, and yet D is not the center of the circle. 3 is the
minimum number of equal lines required to conclude that we have found the center, and
4 is superfluous.
It often happens in mathematics (and elsewhere in life) that Three is enough. We
need at least 3 straight lines to make a rectilineal figure – two won't do. And things often
come in threes. For example, there are three basic relationships between two comparable
quantities: greater than, less than, equal to. And accordingly there are three main species
of triangle (equilateral, isosceles, and scalene), and three species of angle (right, obtuse,
and acute). If you are watchful for it, you will see, over and over again, that important
things often come in threes.

105
2. Can three equal lines be drawn to a circle from a point outside the circle? No. It
will be possible to draw two equal lines from it to the circle, but it will be impossible to
draw three equal lines to the circle. Assume for a moment
that a point P has been chosen outside a circle, and PE = PG
= PH. To see the impossibility of this, draw from center K K
the lines KE, KG, KH. Now, if PGK happens to be a single
straight line, then
E H
KG = HK (radii)
and GP = PH (they are assumed equal) G
so KG + GP = PH + HK
i.e. KP = PH + HK
which is impossible, since PHK is a triangle, and so side KP
has to be less than the sum of the remaining sides. P

On the other hand, if PGK is not a straight line, then

join PK. In this scenario, PGK is a triangle instead of a
K
straight line, and
KG = KH (radii)
and GP = HP (they are assumed equal) E H
and KP is common to ΔPGK and ΔPHK G
so ΔPGK ≅ ΔPHK (Side-Side-Side)
which is impossible, since then these two triangles would
have equal areas, whereas ΔPGK is only part of ΔPHK.
P

THEOREM 9: The circumferences of two distinct circles cannot have more than
two points in common.

Imagine two different circles, circle 1 and circle 2. The

2 circumferences of these two circles cannot have more than
two points in common. Why not?
P
[1] Suppose they have two points in common, namely A and
A
B. Now choose any third point P on the circumference of
M
circle 1.
1 B
[2] Join MA, MB, MP.

[3] Since circles 1 and 2 have common points at A and B, therefore they cannot have the
same center (Thms. 6 & 7). But M is the center of circle 1. Therefore M is not the center
of circle 2.

106
[4] Now if three equal straight lines from M fall upon the circumference of a circle, then
M will be its center (Thm.8). But M is not the center of circle 2 (Step 3), and hence three
equal straight lines cannot be drawn from M to the circumference of circle 2.

[5] Since MA, MB, MP are three equal straight lines from M (being radii of circle 1),
hence they do not all fall on the circumference of circle 2 (Step 4). But MA and MB do
fall on the circumference of circle 2 (given). Therefore MP does not.

[6] Therefore any random point P on circle 1 other than A and B cannot lie on circle 2.
Therefore circles 1 and 2 cannot have more points in common than A and B.

Q.E.D.

THEOREM 9 Question

So the maximum number of times two circles can cut each other is twice. Is it possible
for two circles to cut each other only once?

THEOREM 10: If he circumferences of two circles have two points in common,

then they cut each other at those two points.

A
Suppose the circumferences of circles 1 and 2 have two
points in common, namely T and P. I say the circles cut
C each other, rather than touch each other, at these two
points.
T P
M [1] Join TP and bisect it at M. Draw a perpendicular to
TP at M, cutting circle 1 at A and B, and circle 2 at C and
1 D.
B
2 [2] Since TP is a chord in both circles, therefore it lies
D inside both (Thm.2). Thus the circles overlap each other.

[3] Since TP is a chord in both circles, therefore its perpendicular bisector is a diameter
of both circles (Thm.1). Since TM is perpendicular to these diameters, therefore the
square on TM equals the rectangles contained by the segments into which M divides the
diameters (Ch.2, Thm.6), i.e.

£TM = AM · MB
and £TM = CM · MD
thus AM · MB = CM · MD.

107
[4] Now MB cannot equal MD, since then B and D would be the same point, and thus
circles 1 and 2 would have three points in common, which is not possible (Thm.9).
Therefore they are unequal. Suppose
MB < MD.
But then, since the rectangles are equal, it is necessary that
AM > CM.

[5] Since MA > MC (Step 4), thus point A on circle 1 lies outside circle 2.
Since MB < MD (Step 4), thus point D on circle 2 lies outside circle 1.
Thus each circle falls partly outside the other.

[6] Since circles 1 and 2 partly overlap (Step 2) and each falls partly outside the other
(Step 5), therefore each cuts into and out of the other. Since T and P are the only points
shared by their circumferences, therefore these are the points at which they cut one
another.

THEOREM 10 Remarks

A corollary to this Theorem, or another way of stating it, is that If two circles touch one
another, whether internally or externally, their circumferences have no point in common
other than the one point of contact.

THEOREM 11: If one circle is internally tangent to another, the straight line
joining their centers (when extended) passes through their point of contact.

E
Circle 1 is internally tangent to circle 2 at point C. Find A, the
D center of circle 2, and join AC. I say that the center of circle 1
1
P lies along AC.
C
A
[1] Inside circle 1, pick any random point P not on line AC.
Join AP and extend it until it cuts circle 1 at D and circle 2 at
E. Join PC.

[2] Now AP + PC > AC (since APC is a triangle)

but AC = AE (being radii of circle 2)
so AP + PC > AE
or AP + PC > AP + PE (AE is AP + PE)

108
[3] So PC > PE (subtracting AP from each side in Step 2)
but PE > PD (whole and part)
so PC > PD

[4] Thus PC and PD, being unequal, are not radii of circle 1, and therefore P is not
the center of circle 1. Thus no point which is not on AC can be the center of circle 1.
Therefore the center of circle 1 lies along AC.

[5] Hence the straight line joining center A and the point of contact, C, passes
through the center of circle 1. And therefore the straight line joining the centers of the
circles, when extended, passes through the point of contact.

Q.E.D.

THEOREM 12: If two circles are externally tangent, the straight line joining
their centers passes through their point of contact.

2
1
Circle 1 and circle 2 are externally tangent at
point C. Find A, the center of circle 1, and B,
C B
and join AC, extending it through circle 2
A
until it meets it again at point B. I say that the
D
E
center of circle 2 lies along CB.
P

[1] Inside circle 2, pick any random point P not on line CB. Join AP, which must pass
out of circle 1 at a point (D), and into circle 2 at another point (E).

[2] Now AC + PC > AP (APC is a triangle)

but AC = AD (these being radii of circle 1)
so AD + PC > AP
or AD + PC > AD + DE + EP

[3] So PC > DE + EP (Subtracting AD from both sides)

Thus PC > PE (Subracting DE from the lesser side)
And so, being unequal, PC and PE are not radii of circle 2, and therefore P is not the
center of circle 2. Thus no point which is not on CB can be the center of circle 2.
Therefore the center of circle 2 lies on CB.

[4] Thus the straight line joining center A and the point of contact, C, passes through
the center of circle 2 when extended. That is, the line joining the centers passes through
the point of contact.

Q.E.D.

109
 THEOREM 13: In any circle, equal chords are equally distant from the center.

Imagine a circle in which two equal chords, AB and CD,

have been drawn. Then the perpendiculars drawn to them
G
M from the center will be equal.
A Find the center, M, and drop ME at right angles to
AB, and MG at right angles to CD. I say that ME = MG.
E C Here's the proof:
B

[1] First, AE = EB,

since ME is drawn from the center at right angles to AB (Ch.3, Thm.4).

[2] So EB = ½AB
So too CG = ½CD

[3] But AB = CD (given)

so EB = CG (the halves of equals are equal)

[4] Therefore £EB = £CG (squares built on equal sides are equal)

[5] So £MB – £EB = £MB – £CG,

since this is subtracting equal squares from the same square.

[6] But £MB = £MC,

since the squares on equal sides are equal, and MB = MC, since both are radii of
the circle.

[7] So £MB – £EB = £MC – £CG,

putting together Steps 5 & 6.

[8] But £MB – £EB = £ME (Pythagorean Theorem)

and £MC – £CG = £MG (Pythagorean Theorem)

[9] Therefore £ME = £MG (Putting together Steps 7 & 8)

[10] So ME = MG,
since the sides of equal squares have to be equal.
So the perpendicular distances of equal chords from the center of the circle are equal.

Q.E.D.

110
THEOREM 13 Remarks

Why do we call the perpendicular distance between a point and a line the distance
between them? Because 1. there is only one perpendicular distance, and 2. it is the least
distance.

THEOREM 13 Questions

Prove the converse, using the same diagram. Given that ME = MG, i.e. that the
perpendicular distances of chords AB and CD from the center are equal, prove that the
chords themselves are equal.

THEOREM 14: The diameter is the longest chord inside a circle, and chords
closer to the center are greater than those further away from it.

B Imagine a circle with diameter AB, whose midpoint C is

C
therefore the center of the circle. Now take any other
A chord RN inside the circle. I say that AB > RN. Why?
R
N [1] Join CR; join CN.

[2] Now CR + CN > RN (since RCN is a triangle)

[3] But CR = AC
and CN = CB (since all these are radii of the circle)

[4] So AC + CB > RN (putting together Steps 2 & 3)

i.e. AB > RN

Q.E.D.

111
 P E
D That's the first part of the Theorem. For the next part,
take any other chord DE, whose perpendicular distance
N
from the center (namely CP) is greater than RN's
C perpendicular distance from the center (namely CG). I
say that RN > DE. Why?
G

[1] Join CD; join CE.

[2] Now £CG < £CP,

since we are given that CG < CP, and a square on a lesser line is a lesser square.

[3] Now whenever we subtract a lesser thing from something, it leaves a greater
remainder than if we subtract something greater from it. So if we subtract £CG from
£CD, it leaves a greater remainder than if we subtract £CP from £CD. That is,
£CD – £CG > £CD – £CP

[4] But £CD = £CR,

since CD = CR, being radii of the circle, and the squares on equal lines are equal.

[5] So £CR – £CG > £CD – £CP (putting together Steps 3 & 4)

[6] But £CR – £CG = £RG (Pythagorean Theorem)

and £CD – £CP = £DP (Pythagorean Theorem)

[7] So £RG > £DP (putting together Steps 5 & 6)

[8] i.e. RG > DP, since the side of a greater square is greater.
Thus 2RG > 2DP, since double a greater line remains greater than double the
lesser line.

[9] But 2RG = RN,

since CG is drawn from the center and at right angles to RN (Ch.3, Thm.4).
Also 2DP = DE, for the same reason.

[10] So RN > DE (putting together Steps 8 & 9)

Q.E.D.

112
 THEOREM 15: A straight line drawn at right angles to the end of a circle's
radius is tangent to the circle.

Conceive of a circle with center C and radius CT, and TG

drawn at right angles to CT.
C
I say that TG is tangent to the circle, i.e. that
although it touches it at T it falls entirely outside the
circle forever after, not cutting into the circle at all.
G
Here’s why.
P T

[1] Choose any random point P along TG other than T. Join CP.

[2] Looking at triangle CPT, since angle CTP is right, therefore ∠CPT is acute.

[3] Hence ∠CTP > ∠CPT

thus CP > CT (opposite a greater angle is a greater side)

[4] But CT is a radius of the circle. Therefore, since CP is greater than a radius of the
circle, P must lie outside the circle.

[5] But P is just a random point on TG other than T. Hence every point on TG other than
T itself lies outside the circle. Therefore TG is tangent to the circle at T.

Q.E.D.

THEOREM 15 Remarks

1. Will TG cut into the circle on the other side, namely to the right of CT? No.
Because ∠CTG is a right angle, therefore when we extend GT to the right, the angle
adjacent to it will also be a right angle, and so the exact same proof will apply on the
right side as well.

2. We could state the Theorem another way: a straight line drawn at right angles to
a circle’s radius has one and only one point in common with the circumference of the
circle, namely the endpoint of the radius to which it is drawn at right angles.

3. Notice that this Theorem gives us an easy construction for a tangent to any point
on the circumference of a circle: if the point is T, merely find the center of the circle
(Thm.1), say C, and join CT, and then draw a straight line at right angles to CT through
point T.

113
 THEOREM 16: Any straight line drawn at less than a right angle to the end of a
circle's radius cuts into the circle.

Conceive a circle with center C, radius CA, and a straight

C line AS drawn at any acute angle CAS. I say that AS cuts
through the circle.
B

P
S
A

[1] Drop CP perpendicular to AS (extended, if need be).

[2] Extend AP to B so that AP = PB. Join CB.

Since AP = PB
and CP is common
and ∠CPA = ∠CPB (both being right)
thus rCPA ≅ rCPB (Side-Angle-Side)
hence CB = CA
Thus, since CB is equal to the radius CA and is drawn from the center C, CB is also a
radius of the circle, and therefore B is on the circumference of the circle.

[3] But B lies along the line AS, and A is another point on AS on the circumference of
the circle. Therefore line AS (extended, if need be) joins two points on the circumference
of the circle, and therefore it is a chord (Thm.2).

Q.E.D.

THEOREM 16 Remarks

1. From this Theorem it follows that it is impossible to draw two tangents to a circle
touching it at the same point. For each and every point on the circumference of a circle,
there is one and only one tangent, namely the straight line drawn through it at right angles
to the radius drawn to it.

114
2. This Theorem is truly surprising. To see why, draw the tangent AT at right angles
to CA, and draw a line AS making CAS an acute angle as close to a right angle as you
like. By the Theorem, AS must cut into the circle at A
and out again at some other point (very close to A).
Now there is a space between the circumference of the
circle and the tangent AT all the way down to the point C
A. And a straight line has no thickness, and so in a way
it doesn't take up any space. And yet we still can't find S

room to stick a straight line like AS in between the

T
tangent and the circle without cutting into the circle! A

THEOREM 17: How to draw a straight line tangent to any circle from any point
outside it.

If I give you a circle M, and a point P outside it, can you draw a line from P tangent to
circle M? Sure. All you have to do is …
B
D
[1] Find the center, C.

[2] Join CP, cutting circle M at A, and draw another P A C

circle around C with radius CP.
M
[3] Draw AB at right angles to PAC, cutting circle P at B.

[4] Join CB, cutting circle M at D.

[5] Join DP. This line DP is in fact tangent to circle M. Here's why:

[6] AC = CD (being radii of circle M)

and CB = CP (being radii of circle P)
and ∠ACB = ∠PCD (being in fact the same angle)
so the corresponding angles of ΔACB and ΔPCD are equal (Side-Angle-Side)

[7] So ∠CDP = ∠CAB (being corresponding angles of ΔACB & ΔPCD)

[8] Thus ∠CDP is a right angle,

since it equals ∠CAB, which we made a right angle.

[9] But CD is a radius of circle M, and so DP, drawn at right angles to CD, is tangent
to circle M (Thm.15).

Q.E.F.

115
THEOREM 18: If a straight line is tangent to a circle, then the radius joined to
the point of tangency is perpendicular to the tangent.

Picture a circle with center C, to which

straight line TN is tangent at T. Then
C
∠CTN is a right angle. The proof:

[1] Extend NT to M. Since NT is

1 2
tangent to the circle, therefore NTM does
M T N not cut the circle, but lies outside the circle.

[2] If ∠2 were less than a right angle, then TN would cut the circle at T, and so NTM
would not be tangent to the circle (Thm.16). But NTM is tangent to the circle, and
therefore ∠2 is not less than a right angle.

[3] If ∠2 were more than a right angle, then ∠1 would be less than a right angle
(since together they add up to two rights), and so TM would cut the circle at T, and so
NTM would not be tangent to the circle (Thm.16). But NTM is tangent to the circle, and
therefore ∠2 is not more than a right angle.

[4] Since ∠2 is neither less than a right angle (Step 2), nor more than a right angle
(Step 3), therefore it is a right angle. That is, ∠CTN is a right angle.

Q.E.D.

THEOREM 18 Remarks

1. We can now show that the two tangents to a A

circle from any one point R must be equal. Let O be
the center of a circle, and join RO. Calling the tangents
RA and RB, draw the lines OA and OB. Now, ∠RAO R O

= 90°, and likewise ∠RBO = 90° (by Theorem 18). So

ΔRAO and ΔRBO are both right triangles, and the B
Pythagorean Property applies.
i.e. £RO = £OA + £RA (ΔRAO is a right triangle)
or £RO = £OB + £RA (since OA = OB, therefore £OA = £OB)
but £RO = £OB + £RB (ΔRBO is a right triangle)
thus £RA = £RB (since each is equal to £RO – £OB)
so RA = RB.

116
2. We can see from Theorem 18 that If a straight line is drawn at right angles to a
circle's tangent at the point of tangency, the line will pass through the center of the circle.
Use the same diagram as in Theorem 18. We are given that MTN is tangent at T, and we
want to show that the perpendicular at T passes through the center. Find center C and
join TC. Then by Theorem 18 TC is perpendicular to MTN. Therefore the line
perpendicular to MTN at the point of tangency passes through the center of the circle.

3. Prove that there are only two tangents to a circle from a given point outside it.

THEOREM 19: An angle at the center of a circle is double an angle at the

circumference if they stand on the same arc.

Consider a circle with center M, and an

angle AMB standing on arc AB, and another
angle ACB also on arc AB but having its
D vertex C at the circumference instead of at
M the center. Then, surprisingly, ∠AMB is
4 C
2 3 double ∠ACB. To see why, first join CM
1
and extend it to D.
6 5
A B Letting numbers designate angles,

[1] 4+2 = 3+1+5 (4 + 2 is exterior to rBMC)

but 5 = 3+1 (rBMC is isosceles)
so 4+2 = 3+1+3+1

[2] and 4 = 3+6 (4 is exterior to rAMC)

but 6 = 3 (rAMC is isosceles)
so 4 = 3+3

[3] thus (3 + 3) + 2 = 3 + 1 + 3 + 1 (Steps 1 & 2)

thus 2 = 1 + 1 (subtracting 3 from each side twice)
Hence angle 2 is double angle 1, i.e. ∠AMB is double ∠ACB.

Q.E.D.

117
 THEOREM 19 Remarks

B
M
1. Does it make any difference if CM goes between
the legs of the angles?

A D

E
2. The angle at the center and the angle at the
circumference do not actually have to be on the same arc in K
order for the theorem to hold. They need only be on equal
arcs. In a circle with center K let arc EF = arc HG. F
Choose a random point R on the circumference and draw
RH and RG. Then it is still true that H G
∠EKF = 2∠HRG
Start by drawing KH and KG. Now, since the arcs EF and HG are equal, therefore if pie-
piece EKF is rotated clockwise so that E is on G, then also F will be on H. Therefore KE
and KG will coincide while KF and KH also coincide. In other words, ∠EKF will
coincide with ∠GKH. Therefore
∠EKF = ∠GKH
but 2∠HRG = ∠GKH (Thm.19)
so 2∠HRG = ∠EKF

THEOREM 20: In a circle, angles at the circumference standing on the same

arc are equal.

2 Imagine a circle with center M, and let ARC be any arc of

1
its circumference. Now let ∠1 and ∠2 be any two angles
at the circumference standing on arc ARC. I say ∠1 = ∠2.
Join MA and join MC to find out why.
M

A C
[1] ∠1 = ½∠AMC,
since ∠1 is at the circumference, ∠AMC is at the center,
R and they stand on the same arc (Thm.19).

118
[2] ∠2 = ½∠AMC,
since ∠2 is at the circumference, ∠AMC is at the center, and they stand on the
same arc (Thm.19).

[3] Therefore ∠1 = ∠2,

since each is half of ∠AMC (Steps 1 & 2).

Q.E.D.

THEOREM 20 Remarks
R
E
1. As with the last theorem, this Theorem does not depend
on the angles at the circumference being on the same arc; they
need only be on equal arcs. So if arc EF = arc HG, and P and R K
P
are points chosen on the circumference, then
F
∠EPF = ∠HRG H G
Take center K, and draw KH and KG.
Now ∠HRG = ½∠HKG (Thm.19)
But ∠EPF = ½∠HKG (since arc EF = arc GH; Thm.19, Remark 2)
So ∠EPF = ∠HRG

2. The converse of this last Remark is also true: The arcs R

on which stand equal angles at the circumference are equal.

That is, E
if ∠EPF = ∠HRG P

then arc EF = arc HG. L

If you doubt it, assume for a moment that the arcs are not equal; F

suppose arc EF is greater than arc HG. Therefore a part of arc H G

EF will be equal to arc HG, say
arc EL = arc HG
then ∠EPL = ∠HRG (by Remark 1 just above)
but ∠EPF = ∠HRG (given)
so ∠EPF = ∠EPL,
which is impossible, since the whole never equals a part of itself. So it is impossible that
arc EF should be greater than arc HG. Likewise HG cannot be greater than EF.
Therefore the arcs are equal.

3. Theorem 20 is just as true if the two angles are not in the same circle, but instead
are in two different but equal circles 1 and 2. That is, if ∠X at the circumference of
circle 1 stands on arc A, and ∠Z at the circumference of circle 2 stands on arc B, and arc
A = arc B, then also ∠X = ∠Z.

119
THEOREM 21: The opposite angles of a quadrilateral inscribed in a circle add
up to two right angles.

A
Imagine a quadrilateral ABCD whose four corners lie B
2
on the circumference of a circle.
Then ∠ABC + ∠ADC = two right angles
and ∠DCB + ∠DAB = two right angles. 4
Join AC and join BD to begin the proof. 1 3
C

[1] First ∠1 = ∠4 D
(both stand on arc AB; Thm.20)

[2] And ∠2 = ∠3 (both stand on arc AD; Thm.20)

[3] So ∠1 + ∠2 = ∠3 + ∠4 (adding together Steps 1 & 2)

[4] But ∠1 + ∠2 + ∠DAB = two rights (DAB is a triangle)

[5] So ∠3 + ∠4 + ∠DAB = two rights (putting together Steps 3 & 4)

[6] i.e. ∠DCB + ∠DAB = two rights (∠3 + ∠4 is ∠DCB)

[7] But since the angles of any quadrilateral add up to four right angles (since it is
made up of two triangles, each of whose angles add up to two right angles), and since
∠DCB + ∠DAB equal two rights, therefore the remaining angles of quadrilateral ABCD
must add up to the remaining two right angles. That is, ∠ABC + ∠ADC = two rights.

Q.E.D.

THEOREM 21 Remarks

1. It follows from this that a quadrilateral whose opposite angles do not add up to
two right angles cannot be inscribed in a circle. Take two identical equilateral triangles
and put two of their sides together – you now have a rhombus one pair of whose opposite
angles are 60° and 60°. Such a quadrilateral cannot be inscribed in a circle – draw it for
yourself and confirm it with a diagram.
A quadrilateral that can be inscribed in a circle is sometimes called a "cyclic"
quadrilateral. "Cyclic" comes from the Greek word for "circle". Since a pair of angles
that add up to two right angles are called supplementary angles, we can restate Theorem
21 like this: In cyclic quadrilaterals, opposite angles are supplementary.

120
 A
2. The converse of Theorem 21 is also true, and it goes E
B
like this: If the opposite angles in a quadrilateral are
supplementary, then we can circumscribe a circle around
G
it. M
Suppose quadrilateral ABCD is such that its
opposite angles are supplementary. Bisect AB at E and BC C
D
at G. Draw lines perpendicular to AB and BC at E and G,
and where these two perpendiculars meet call M. Join MA,
MB, MC. A
B
Now ΔAEM ≅ ΔBEM (Side-Angle-Side) 2
so MA = MB
and ΔCGM ≅ ΔBGM (Side-Angle-Side)
so MB = MC D
1
Thus MA = MB = MC, and so the circle with center M and 3 C
radius MA will pass through A, B, and C. X

But will it pass through D, the other corner of quadrilateral ABCD? Yes, it must.
For suppose it did not, but rather D fell inside the circle. Then extend AD until it
cuts the circle at X. Join CX.
Now ∠3 + ∠2 = two rights (quadrilateral ABCX is inscribed in a circle)
but ∠1 + ∠2 = two rights (quadrilateral ABCD is given that way)
so ∠1 = ∠3, that is, an angle exterior to triangle CDX is equal to one of its interior
and opposite angles, which is impossible (Ch.1, Thm.14). So it is impossible for D to fall
inside the circle. Likewise it is impossible for it to fall outside the circle. Therefore D
falls on the circumference of the circle.

B
3. Theorem 21 has a cousin theorem about
quadrilaterals that are circumscribed around circles.
F
Consider quadrilateral ABCD, circumscribed around a
E
circle and so having its four sides tangent to the circle at
C
E, F, G, H. Although its opposite angles do not have to
A add up to two right angles, its opposite sides must have
the same sum, that is AD + BC = AB + DC.
H G

D
The proof is easy. Since AH is tangent to the circle at H,
and AE is tangent to it at E, therefore AH and AE are two tangents to the same circle
from the same point, and thus
AH = AE (Thm.18, Remark 1)
and HD = DG (Same reason)
and BF = BE (Same reason)
and CF = CG (Same reason)
so AH + HD + BF + CF = AE + DG + BE + CG
or (AH + HD) + (BF + CF) = (AE + BE) + (DG + CG)
i.e. AD + BC = AB + DC.

121
THEOREM 22: In any circle, the arcs cut off by equal chords are equal.

A R
Imagine a circle with any
center M and having two B
equal chords AB and CD in it M
cutting off arcs ARB and M
CLD. Then these two arcs (B) C
must be equal. To see the C
(A)
proof, draw these lines: MA,
D
MB, MC, MD. D L

[1] First, AB = CD (the chords are given as equal)

and MA = MC (being radii of the circle)
and MB = MD (being radii of the circle)
so ΔAMB ≅ ΔCMD (Side-Side-Side)

[2] Since ΔAMB and ΔCMD are congruent, therefore if we were to rotate the pie-
piece MARB clockwise so that A is on C, then also B would be on D, and so the arc ARB
would then coincide with the arc CLD (for if it fell outside as the dotted line, the radii
from M would not all be equal!).

[3] Since the arcs ARB and CLD can be made to coincide with each other, therefore
they are equal.

Q.E.D.

THEOREM 22 Remarks

1. The converse of this Theorem is also true. That is, In A R

any circle, the chords of equal arcs are equal. For instance,
if arc ARB = arc CLD B
then AB = CD.
M E
If possible, assume that
AB > CD C
then make DE = AB, which is easily done by making a circle
(not shown) around point D with a radius equal to AB. D L

122
Thus arc ECD = arc ARB (arcs of equal chords are equal; Thm.22)
but arc CLD = arc ARB (that is given)
so arc CLD = arc ECD,
which is to say that the part is equal to the whole, which is impossible. So AB is not
greater than CD. Likewise we can prove that neither is it less. Therefore
AB = CD.

2. Theorem 22 and its converse are also true about chords and arcs in equal circles:
* If arc A in one circle equals arc B in an equal circle, then their chords are equal.
* If chord A in one circle equals chord B in an equal circle, then their arcs are equal.

THEOREM 23: How to bisect any circular arc.

D
R Suppose you have any arc of a circle, ARB. How
do you cut it into two equal parts? Like this:
A B
C

[1] Join AB.

[2] Bisect AB at C.

[3] Draw CD at right angles to AB.

[4] Join AD; join DB.

[5] Now, ∠ACD = ∠BCD (since each is a right angle)

and AC = CB (since we bisected AB at C)
and CD is common to ΔACD and ΔBCD
so the corresponding sides of ΔACD and ΔBCD are equal (Side-Angle-Side)

[6] Thus AD = BD (being corresponding sides of ΔACD & ΔBCD)

[7] So arc AD = arc BD (Thm.22)

Q.E.F.

123
THEOREM 23 Remarks

Have we done enough geometry now to cut any circular arc into 3 equal parts? No – that
is a problem for more advanced geometry, and its simplest solution involves the use of
conic sections. It is impossible to cut an arc of 60° into three equal parts, for example,
using nothing but circles and straight lines, which are the only tools we are allowing
ourselves in this book.
How about 4 equal parts? Of course. We simply bisect the arc once, and then
bisect each of its halves.
How about 5 equal parts? No. That is even more difficult than 3.
Neither can we cut any given arc into 6 or 7 equal parts. But we can cut any
given arc into 8 equal parts simply by repeated bisection (likewise with 16, 32, etc.).

THEOREM 24: Any angle inside a semicircle is right.

R
Take any circle with diameter AB cutting it into two
semicircles, pick any point R along the circumference,
and then join AR and RB. Angle ARB is a right angle.
A B
To see it for yourself, start by picking any point T
on the other semi-circumference, and then join AT and
TB.
T

[1] Now ∠ARB = ∠ATB,

since they stand at the circumference and cut off equal arcs, namely half the whole
circumference (Thm.20, Remarks)

[2] But ∠ARB + ∠ATB = two right angles,

since they are opposite angles in a quadrilateral inscribed in a circle (Thm.21).

[3] So ∠ARB = half of two rights,

i.e. ∠ARB = one right angle.

Q.E.D.

124
 THEOREM 24 Remarks

G
X X
1
2 G
The converse of this Theorem is also
3 true, namely The circle whose
E
diameter is the hypotenuse of a right
F E F
triangle must pass through the vertex
of the right angle. If EGF is a right
triangle, and ∠EGF is the right angle,
then the circle having EF as its
diameter passes through point G. If possible, assume it doesn't, but instead G winds up
inside the circle. Then FG and the circle will intersect at some other point, X; join XE.

Thus ∠1 = one right angle (since it is an angle in a semi-circle; Thm.24)

but ∠2 = one right angle (since it is adjacent to ∠3, which is given as right)
so ΔEXG has two right angles in it, and thus its angle-sum is greater than two right
angles, which is impossible. So it is impossible for G to fall inside the circle. Likewise it
is impossible for it to fall outside it. Therefore G lies right on the circumference.

THEOREM 24 Questions
P

1. If M is the center of a circle, and MP is

perpendicular to the diameter, prove that ∠APB is a A
M
B
right angle (independently of Theorem 24). Now use
Theorem 20 to prove that all the angles at the
circumference standing on diameter AB are right.

2. Let L be any point on the circumference of a L

circle and CD its diamter, K its center. To prove 2
3
Theorem 24 in yet another way, start by establishing
that C
1 4
D
K
1 + 2 + 3 + 4 = 180°
and use the fact that ΔCKL and ΔLKD are both
isosceles.

125
 W 3. Prove that the angle in a segment of a circle
that is greater than a semicircle is acute, and that the
S T
angle in a segment that is less than a semicircle is
obtuse. Let ST divide a circle into two unequal
Z segments, i.e. one greater than a semicircle, and one
less. Find the center Z. Pick W and X at random on
each of the two arcs. Draw SW, WT, SX, and XT.
Draw the diameter TZY, and join WY. Start the
Y
proof by showing that ∠SWT is obtuse; then show
X
that ∠SXT is acute by using Theorem 21.

THEOREM 25: The angles between a tangent to a circle and any straight line
drawn through the circle from the point of tangency are equal to the angles in the
alternate segments of the circle.

D
Conceive of a circle with a straight line TPN
S
X tangent to it at point P, and any straight line
PS drawn through the circle and cutting it into
R two segments. Choose any point X on the arc
C
of one segment, and any point R on the arc of
the other. Then
∠SPN = ∠SXP
and ∠SPT = ∠SRP.
T P N

[1] Start the proof by joining center C to P and extending PC to D to complete

diameter PD.

[2] Thus PD is at right angles to TPN, since TPN is tangent at P and C is the center of
the circle (Thm.18).

[3] Now ∠DSP = a right angle (since it is an angle in a semicircle)

[4] So ∠SDP + ∠DPS = a right angle,

that is, the remaining two angles in ΔDSP must add up to a right angle, so that all three of
its angles together will equal two right angles.

[5] But ∠SPN + ∠DPS = a right angle,

since these two angles make up ∠DPN, and DP is at right angles to PN (Step 2).

126
[6] So ∠SDP = ∠SPN,
because according to Steps 4 and 5, these two angles are complementary to the same
angle, namely ∠DPS.

[7] But ∠SDP = ∠SXP,

since they both stand at the circumference and on the same arc SRP (Thm.20).

[8] So ∠SPN = ∠SXP (putting together Steps 6 & 7)

And this is the first part of what we wanted to prove.

[9] Now ∠SXP + ∠SRP = two rights,

since they are opposite angles of a quadrilateral inscribed in a circle (Thm.21)
So ∠SPN + ∠SRP = two rights (∠SPN = ∠SXP; Step 8)

[10] But ∠SPN + ∠SPT = two rights (they are adjacent angles)

[11] So ∠SPT = ∠SRP,

since each is the supplement of angle SPN (Steps 9 & 10)

Q.E.D.

THEOREM 26: If from a point on a circle's tangent a straight line is drawn

cutting through the circle, then the rectangle contained by the whole cutting line and the
part outside the circle is equal to the square on the tangent.

B Let PT be a straight line touching a circle with

center M at point T, and take any point P on the
S
M
tangent line. Draw any line PSA through the
circle. Then £TP = AP · PS.
To prove it, start by bisecting AS at B,
then join the following lines: MA, MB, MS,
MP, MT.
P
T

[1] Now, MB is at right angles to AS, since MB is drawn from the center and it
bisects AS (Thm.3). So MBS and MBP are right triangles.

[2] Also, MT is at right angles to TP, since MT is a radius and TP is tangent

(Thm.18). So MTP is a right triangle.

127
[3] Now, £TP = £MP – £MT (ΔMTP; Pythagorean Theorem)

[4] But £MP = £MB + £BP (ΔMBP; Pythagorean Theorem)

[5] So £TP = (£MB + £BP) – £MT (putting together Steps 3 & 4)

[6] But £MB = £MS – £BS (ΔMBS; Pythagorean Theorem)

[7] So £TP = (£MS – £BS) + £BP – £MT (Steps 5 & 6).

Reordering the terms, we can say
£TP = £BP – £BS + £MS – £MT

[8] But £MS = £MT (since MS = MT, being radii),

[9] So £TP = £BP – £BS (putting together Steps 7 & 8)

[10] Thus £TP = (BP + BS) (BP – BS),

because of Ch.2, Thm.4: the "difference between two squares" Theorem.

[11] Now AB = BS (since we bisected AS at B)

so AB + BP = BP + BS (adding equals to the same thing)
i.e. AP = BP + BS

[12] Thus £TP = AP (BP – BS) (putting together Steps 10 &

11)
i.e. £TP = AP · PS (BP – PS is PS)

Q.E.D.

THEOREM 26 Remarks

1. A straight line drawn from outside a circle that cuts E

the circle is called a "secant" (from the Latin secare, to cut
or divide). There is an interesting corollary to Theorem 26 A
about secants, namely that If two secants are drawn to a
circle from the same point, then the rectangles contained by
the wholes of each and by their parts outside the circle are T
equal. Consider secant PSA and secant PDE, and also S
D

draw in tangent PT. Then

AP · PS = £PT (Thm.26)
P

128
and EP · PD = £PT (Thm.26)
so EP · PD = AP · PS

2. The converse of Theorem 26 is also true. That is, if PSA is drawn through a
circle, and PT is drawn to the circle such that £PT = AP · PS, then PT is tangent to
the circle at T.
Proof: Two tangents from P can be drawn to the circle. A

And the square on each of them must equal rectangle AP · PS

(Thm.26). Thus the square on each tangent equals the square
on PT, and so each tangent from P equals PT. Since it is
impossible for three equal lines to be drawn to a circle from S

one point outside it (Thm.8, Remark 2), therefore PT must be

one of the tangents from P. Q.E.D.
P
T

THEOREM 26 Question

M
What if PSA goes through the center M? How will that
change the proof?

S
P
T

129
“HOOK”: THE MIQUEL POINT.

If ABC is any triangle, and E, D are any two points along AC, AB, and we join BE, CD,
we get a “dented triangle,” or a “Menelaos Figure.” There are four triangles in this
figure—two small ones and two larger, overlapping ones. If we circumscribe a circle
around each of these four triangles, the four circles will all cut each other at one point, p.

E B

p
C

130
“HOOK”: CENTER OF WEIGHT OF A TRIANGLE.

If you take any triangle ABC, bisect its sides, and join each vertex to the midpoint of the
opposite side, these three lines intersect at one point inside the triangle, Y. This point is
the center of weight of the triangle. If you were to make such a triangle out of some
rather uniformly dense material, and drill a hole through Y, then hang it on a small nail
through Y, you could rotate the triangle into any orientation and it would always be
balanced.
B

“HOOK”: COTANGENTS TO THREE CIRCLES.

Take any three circles in one plane, so long as no one of them is completely inside
another one. Then it will be possible to take them in three pairs, and for each pair there
will be a pair of straight lines tangent to both circles and which intersect—as DPG, DNK
are each tangent to the two circles PN, GK. The three resulting points of intersection,
namely D, E, F, all lie in a straight line.
H

G A F
K

L Q
W
R
M
C

B T V S
P
N
U E
O

131
Chapter Four

Polygons
THEOREM 1: How to inscribe in a circle a triangle that is equiangular with a
given triangle.

Take any triangle you like, and call its angles 1, 2, 3. Take any circle you like – call it K.
Can we put inside circle K a triangle whose angles are equal to 1, 2, and 3? Of course.

B
K
[1] Draw a tangent TPN through any
2 point P along circle K’s circumference.
N

1 3 A [2] Draw ∠TPA = ∠2.

P Draw ∠NPB = ∠1

[3] Now ∠1 + ∠2 + ∠3 = two right angles,

so ∠NPB + ∠TPA + ∠3 = two right angles.

[4] But ∠NPB + ∠TPA + ∠APB = two right angles

[5] So ∠APB = ∠3 (putting together Steps 3 & 4)

[6] Join AB.

[7] Now ∠TPA = ∠ABP (Ch.3, Thm.25)

But ∠TPA = ∠2 (Step 2)
So ∠ABP = ∠2

[8] Since ∠APB = ∠3 (Step 5)

and ∠ABP = ∠2 (Step 7)
thus ∠BAP = ∠1,
because whenever two angles in one triangle are equal to two angles in another, the third
angle equals the third angle.
Therefore ΔAPB, the triangle inscribed in circle K, is equiangular with the given
triangle.

Q.E.F.

131
THEOREM 1 Questions

1. We can now inscribe a triangle of any shape inside a given circle. Obviously we
cannot inscribe a triangle of any size inside a given circle – some triangles are too big or
too small to be inscribed in a given circle. In fact, once we are given the three angles
which the inscribed triangle is to have, and once we are given the circle to inscribe it in,
the size of the inscribed triangle is already determined.

2. Try to prove that If two triangles are inscribed in the same circle and they are
equiangular, then they will be congruent, i.e. all their corresponding sides will also be
equal and they will have the same area. Consider using the converses of Theorems 20
and 22 in Chapter 3.

3. Is it possible to inscribe a quadrilateral inside a given circle? If we are given a

quadrilateral and a circle, will we always be able to draw a quadrilateral inside the circle,
whose corners are all on the circumference, and whose angles are the same as the given
quadrilateral? What must be true about the quadrilateral for it to be possible?

THEOREM 2: How to circumscribe about a circle a triangle that is equiangular

with a given triangle.

K Say I give you a circle with

center M and a random
D
triangle ABC having angles
4 B
F 1, 2, 3. How can you
R
1
4 circumscribe another triangle
M 5 around circle M having
3 2
angles equal to 1, 2, and 3?
A C P
G
L You can, as follows:
5
E

[1] Extend any side of ΔABC both ways, say BC (to D and E), forming the exterior
angles 4 and 5.

[2] Choose point R at random on the circumference of circle M. Join MR.

Draw ∠RMF = ∠4
Draw ∠RMG = ∠5 (Ch.1, Thm.20)

[3] Draw lines at right angles to MF at F, to MR at R, and to MG at G.

Call the points where these three lines meet P, K, L.

132
[4] Since PFK is drawn at right angles to the end of radius MF, it is tangent to the
circle at F (Ch.3, Thm.15). Likewise KRL is tangent at R, and PGL is tangent at G.
Thus the triangle PKL is circumscribed about circle M (Ch.3, Def. 11).

[5] Now the angles of quadrilateral FMRK add up to a total of four right angles (since
it is composed of two triangles, each of whose angles add up to two right angles).
But ∠KFM + ∠KRM = two rights, since each of these is a right angle.
Thus the remaining two angles in quadrilateral FMRK must add up to two rights,
i.e. ∠4 + ∠FKR = two rights,

[6] But ∠4 + ∠1 = two rights (since they are adjacent)

[7] So ∠FKR = ∠1 (putting together Steps 5 & 6)

Likewise ∠RLG = ∠2

[8] Since ∠FKR = ∠1 (Step 7)

and ∠RLG = ∠2 (Step 7)
thus ∠GPF = ∠3,
since whenever two angles in one triangle equal two angles in another triangle, the
remaining angles are equal.
So triangle PKL, which has been circumscribed around circle M, is equiangular
with the given triangle ABC.

Q.E.F.

THEOREM 2 Questions

1. How do we know the lines drawn at right angles to MF, MR, and MG at F, R, and
G will meet, as is asserted in Step 3?

2. What about a quadrilateral? If I give you a random quadrilateral and a circle, will
you be able to circumscribe around the circle a quadrilateral whose angles are equal to
the given quadrilateral? What must be true about the quadrilateral for it to be possible?

133
 THEOREM 3: How to inscribe a circle in any triangle.

A
Now suppose I give you a triangle. Can you draw
inside it a circle to which all three sides are tangent?
E You can, like this:
G
D
[1] Take your triangle ABC and bisect ∠ABC and
∠ACB (any two angles will do).
Let the bisectors meet at D.
B C
F

[2] Drop DE perpendicular to AB.

Drop DF perpendicular to BC.
Drop DG perpendicular to AC.

[3] Now ∠BED = ∠BFD (both are right angles)

and ∠EBD = ∠FBD (BD bisects ∠EBF; Step 1)
and BD is common to ΔDEB and ΔDFB
so the corresponding sides of ΔDEB and ΔDFB are equal (Angle-Angle-Side)

[4] So DE = DF (being corresponding sides of ΔDEB & ΔDFB)

[5] In the same way, we can prove that

ΔDFC ≅ ΔDGC,
so DF = DG

[6] Thus DE = DF = DG (Steps 4 & 5),

so draw a circle with D as the center and DE as the radius, and its circumference
will pass through E, F, and G.

[7] Since AEB is at right angles to the endpoint of radius DE,

AB is tangent to circle D at E (Ch.3, Thm.15)
Likewise BC is tangent to circle D at F,
and AC is tangent to circle D at G.

So circle D has been inscribed in triangle ABC (Ch.3, Def.11).

Q.E.F.

134
THEOREM 3 Remarks

There is another important theorem lurking within Theorem 3, namely that The bisectors
of the three angles of any triangle all meet at one point. BD and CD bisect angles ABC
and ACB, and they meet at D. Join DA; if we can prove that it bisects angle BAC, then
all three angle-bisectors meet at one point.
First, DE = DG (proved in Step 6 of Theorem 3)
so £DE = £DG
so £AD – £DE = £AD – £DG (subtracting equals from the same thing)
but £AD – £DE = £AE (ΔADE; Pythagorean Theorem)
and £AD – £DG = £AG (ΔAGD; Pythagorean Theorem)
so £AE = £AG
i.e. AE = AG A
thus ΔADE ≅ ΔAGD by the Side-Side-Side
Theorem, since AE = AG, DE = DG, and AD is
common to both triangles. Therefore their E
corresponding angles are equal, so G

∠EAD = ∠GAD, D

which means that AD is in fact the bisector of angle

BAC. B C
F

So the 3 angular bisectors of a triangle all meet in one point. You can illustrate
this fact by carefully cutting out any triangle from a piece of paper and folding each of its
angles in half. Do the three creases all pass through one point?

THEOREM 3 Question

Prove that the bisectors of angles ABC and ACB must in fact meet, as is asserted in Step
1, and that they must meet inside the triangle, not outside it. To prove these bisectors
must meet, use the fact that any two angles of a triangle (such as rABC) add up to less
than 180°. To prove they must meet inside the triangle, assume for a moment that they
meet outside, and see whether they can really bisect the angles of the original triangle
anymore.

135
THEOREM 4: How to circumscribe a circle about any triangle.

If I give you a triangle ABC, how can you draw A

a circle that goes through A, B, and C? You
can, as follows.
E
[1] Bisect AB at D, and bisect AC at E (any D
two sides will do).
M
[2] Draw perpendiculars to AB and AC at D C
and E, and let them meet at M. B

[3] Draw AM, BM, CM.

[4] Now, ∠BDM = ∠ADM (both are right angles)

and BD = DA (we bisected AB at D)
and DM is common to ΔBDM and ΔADM
so the corresponding sides of ΔBDM and ΔADM are equal (Side-Angle-Side)

[5] Thus BM = AM (being corresponding sides of ΔBDM & ΔADM)

[6] Likewise we can prove that

ΔCEM ≅ ΔAEM,
so AM = CM

[7] Thus BM = AM = CM (Steps 5 & 6),

so draw a circle with M as center and AM as radius, and its circumference will
pass through A, B, and C.
So a circle has been circumscribed about triangle ABC (Ch.3, Def. 10)

Q.E.F.

THEOREM 4 Remarks

1. Notice that, unlike quadrilaterals, all triangles are “cyclic”; all triangles can have
a circle drawn through their three points.

136
2. There is an important theorem A
lurking within Theorem 4, namely this:
The three perpendicular bisectors of the
sides of any triangle all meet at one point. E
DM and EM meet at M, and these are the D
perpendicular bisectors of AB and AC. M
Drop MX perpendicular to BC, and if we
can show that it bisects BC then we will C
have proved that the perpendicular X
B
bisectors of all three sides meet at M.

First BM = CM (proved in Step 7 of Theorem 4)

so £BM = £CM
so £BM – £MX = £CM – £MX (taking the same square from both sides)
but £BM – £MX = £BX (ΔBMX; Pythagorean Theorem)
and £CM – £MX = £XC (ΔCMX; Pythagorean Theorem)
so £BX = £XC
thus BX = XC
Therefore MX is not only perpendicular to BC, but bisects it at X, and so indeed the three
perpendicular bisectors of the sides of ΔABC all meet at one point, M. This point M is
called the circumcenter of a triangle.

THEOREM 4 Questions

1. What happens if ∠BAC is a right angle? What does BC become in relation to the
circumscribed circle? Where will center M fall? What happens if ∠BAC is obtuse?
Where will M fall then?

2. Prove that the two lines drawn perpendicular to AB and AC at D and E must in
fact meet, as is asserted in Step 2. Start by joining DE, and you should see immediately
that the perpendiculars at D and E must make angles less than two right angles on one
side of DE.

137
THEOREM 5: How to inscribe a square in any circle.

M Given a circle, how do we inscribe a square in it?

B D

[1] Begin by finding its center, M. (Ch.3, Thm.1)

Draw any line through M, AMC; thus AC is a diameter.
Draw BMD at right angles to AMC; thus BD is another diameter.

[2] Draw AB, BC, CD, and DA, forming quadrilateral ABCD inscribed in circle M.

[3] Now, ∠BMA = ∠BMC (both are right angles)

and AM = MC (both are radii)
and BM is common to ΔABM and ΔBCM
so rABM ≅ rBCM (Side-Angle-Side)

[4] Thus AB = BC (rABM ≅ rBCM)

[5] We can prove in the same way that

rBCM ≅ rCDM,
so BC = CD,
and again we can prove in the same way that
rCDM ≅ rDAM,
so CD = DA

[6] So AB = BC = CD = DA (Steps 4 & 5),

and so quadrilateral ABCD is equilateral.

[7] Since BMD is a diameter, therefore ∠BAD is a right angle (Ch.3, Thm.24). For
the same reason every other angle of quadrilateral ABCD is a right angle.

[8] Since ABCD is both equilateral (Step 6), and also right angled (Step 7),
ABCD is a square.

Q.E.F.

138
THEOREM 5 Questions

1. Can you see how to circumscribe a square around a given circle?

2. Can you see how to inscribe a circle in a given square?
3. Can you see how to circumscribe a circle around a given square?

THEOREM 6: How to make an isosceles triangle whose base angles are each
double its peak angle.

X Recall that in every isosceles triangle the base angles are equal to each
other. But, for the sake of the upcoming Theorem 7, we want to make a
particular kind of isosceles triangle in which each base angle is double the
2X 2X peak angle X. Here’s how:

[1] Set out any straight line AS. Extend it to P so

that £AS = AP · PS (Ch.2, Thm.12). Draw a circle
with center A and radius AP. Mark off PT = AS by
drawing a circle around P with a radius equal to AS A
(not shown). Join AT. Now AT = AP, being radii of
circle A. Thus ATP is an isosceles triangle, and I say
that ∠TPA is double ∠TAP. S

N T P
[2] Circumscribe a circle about rAST. (Thm.4)
Since £AS = AP · PS (Step 1)
thus £PT = AP · PS (PT = AS; Step 1)
thus PT is tangent to the circumscribed circle (see Ch.3, Thm.26, Remark 2).

[3] Extend PT to N. Because PTN is tangent, therefore it makes angles with AT

equal to the angles in the alternate segments of the circle (Ch.3, Thm.25).
Hence ∠ATN = ∠AST

[4] Hence the supplements of these angles are also equal to each other,
i.e. ∠ATP = ∠TSP

[5] But ∠TPA = ∠TPS (they are actually the same angle)
thus rAPT is equiangular with rTSP (Ch.1, Thm.29)

139
[6] Hence rTSP is also isosceles, so
PT = ST
but PT = AS (Step 1)
so AS = ST
so rAST is also isosceles.

[7] Now ∠TSP = ∠STA + ∠TAS (∠TSP is exterior to rAST)

or ∠TSP = 2 ∠TAS (∠STA = ∠TAS, since rAST is isosceles)
so ∠TPS = 2 ∠TAS
i.e. ∠TPA = 2 ∠TAP

Q.E.F.

THEOREM 6 Remarks

1. The triangle we have constructed is called The Golden Triangle, for reasons we
will understand later. Can you find another golden triangle in the construction diagram
other than rATP? Prove it.

2. What are the angles of a golden triangle in degrees? How many degrees is the
peak angle? How many degrees is one of the base angles? If we call the peak angle X,
each base angle is 2X, thus the sum of the angles in the whole triangle is 5X. But this
must be 180°. Therefore X = 180° ÷ 5, which is 36°. So the peak angle is 36°, and
each base angle is 72°.

THEOREM 7: How to inscribe a regular pentagon inside any circle.

A
Imagine you have a circle A, and you want
to draw a regular pentagon inside it (that is, 1
a five-sided polygon that is equilateral and B E

equiangular). Follow these steps:

T

[1] Start by making an isosceles triangle C D

T whose base angles are each double
its peak angle (Thm.6).

140
[2] In the circle inscribe rACD whose angles equal those of triangle T (Thm.1).
Thus ½∠ADC = ∠1
and ½∠ACD = ∠1

[3] Bisect ∠ACD with CE.

Bisect ∠ADC with DB.
Draw AB, BC, DE, EA, yielding pentagon ABCDE.

[4] Now, ½∠ADC = ∠ADB (we bisected ∠ADC with DB; Step 3)
but ½∠ADC = ∠1 (Step 2)
so ∠ADB = ∠1

[5] In the same way, we can prove that

∠BDC = ∠1
∠DCE = ∠1
∠ECA = ∠1

[6] And all these equal angles are at the circumference. Therefore the arcs on which
they stand are all equal to each other (Ch.3, Thm.20, Converse proved in Remarks),

i.e. arc AB = arc BC = arc CD = arc DE = arc EA

[7] Since all these arcs are equal, the chords joining their endpoints are all equal, too
(Ch.3, Thm. 22, Converse proved in Remarks). That is

AB = BC = CD = DE = EA.

Therefore the pentagon ABCDE is equilateral.

[8] Now ∠ABC stands on three of the five equal arcs, namely on
arc CD + arc DE + arc EA.
Likewise ∠BCD, ∠CDE, ∠DEA, ∠EAB each stand on three of the five equal arcs.
Therefore these five angles stand on equal arcs.

[9] And all five of these angles standing on equal arcs are at the circumference.
Therefore they are equal (Ch.3, Thm.20, proof in Remarks).
That is, ∠ABC = ∠BCD = ∠CDE = ∠DEA = ∠EAB.

[10] Thus the pentagon ABCDE, inscribed in the circle, is both equilateral (Step 7) and
equiangular (Step 9). Therefore it is a regular pentagon.

Q.E.F.

141
THEOREM 7 Remarks

The base angle of a golden isosceles triangle is 72° A

(Thm.6, Remark 2). Now make an isosceles triangle

whose peak angle is 72°. Make the equal sides CG and B E
GD of whatever length you like. Since 72° × 5 = 360°, G

thus we can place the peak angles of 5 identical

isosceles triangle of this kind around point G. Since the 5
triangles are identical isosceles triangles, their bases are C D
equal, and therefore the pentagon ABCDE is equilateral.
Again, since every angle of the pentagon, such as ∠ABC, consists of two of the identical
base angles, pentagon ABCDE is equiangular. Therefore a regular pentagon consists of 5
identical isosceles triangles whose peak angles are each 72°.

Clearly GA = GB = GC = GD = GE. The circle

O with center G and radius GA therefore circumscribes the
pentagon. Given a regular pentagon OPKLN, then, it will
be easy to circumscribe a circle about it. Choose any side
P N
M
KL, and draw angles 1 and 2 each equal to ∠GCD above.
Therefore the lines drawn must meet inside the pentagon
at a point, M, such that ∠KML = ∠CGD or 72°. Thus M
1 2 is the point common to the 5 identical isosceles triangles
K L composing pentagon OPKLN, and therefore M is the
center of the circumscribing circle.

THEOREM 7 Questions

G
1. How many degrees is each angle of the regular pentagon?
Start by recalling that ∠HGK is 72° in isosceles triangle HGK.

H K

2. How would you circumscribe a regular pentagon around a

circle? Start by inscribing one as we did in Theorem 7.
Describe the construction and prove that the result is a regular
pentagon.

3. Find a way to construct a regular pentagon on a given

straight line as its side.

THEOREM 8: How to inscribe a regular hexagon inside any circle.

142
 E F
Take any circle; call its center M. We want to make
a regular hexagon inside it, that is, we want to
inscribe in it a six-sided polygon all of whose sides M
and angles are equal. D A

[1] Start by choosing any point A on the

circumference, and join MA.
C B

[2] Make an equilateral triangle on MA; thus its vertex F will lie on the
circumference of the circle (Ch.1, Thm.1).

[3] Extend AM to D.
Make an equilateral triangle on MD; thus its vertex E will also lie on the
circumference of the circle.

[4] Extend FM to C, EM to B. Draw AB, BC, CD, EF.

[5] Since all three angles of an equilateral triangle are equal, any one of its angles is
one third of two right angles.
so ∠FMA = one third of two rights
and ∠EMD = one third of two rights.
and ∠EMF must equal one third of two rights also, since these three angles
add up to a straight line or 180°.

[6] Hence ∠EMF is also the angle of an equilateral triangle

But EM = MF (they are radii of the circle)
And therefore triangle EMF is also equilateral, and its side is the radius of the circle.

[7] Now ∠DMC = ∠FMA (being vertical angles)

and MC = MF (they are radii of the circle)
and MD = MA (again, they are radii)
so ΔDMC ≅ ΔAMF (by Side-Angle-Side)
And therefore triangle DMC is also equilateral, and its side is the radius of the circle.

[8] In the same way, we can prove that ΔCMB and ΔMBA are equilateral triangles,
too, whose side is the radius of the circle.

[9] Therefore the sides of hexagon ABCDEF are the sides of six identical equilateral
triangles, and so all of its sides are equal. And each of the angles of hexagon ABCDEF is
composed of two angles of an equilateral triangle, and so all of its angles are equal.
Therefore hexagon ABCDEF, inscribed in the circle, is a regular hexagon.

Q.E.F.

143
THEOREM 8 Remarks

It is amazing that the radius of a circle fits inside its circumference exactly six times.

THEOREM 8 Questions

1. How many degrees are in one angle of an equilateral triangle? How many
degrees are in one angle of a regular hexagon?

2. Prove that it is possible to cover a surface with identical regular

hexagons, without leaving any gaps and without any overlap of the 2

hexagons. Begin by looking at angles 1, 2, and 3. 1

3
We might put this little theorem this way: it is possible to use
regular hexagons as floor tiles.

3. With the same sort of reasoning, show that it is not possible to

cover a whole surface with identical regular pentagons – you will
necessarily have gaps. Consider angles 4, 5, and 6 in your reasoning.
We might say it is impossible to use regular pentagons as floor 5

tiles, since we leave gaps. Incidentally, can you identify the shape of 4 6

the gaps left out by the pentagons?

144
THEOREM 9: How to inscribe a regular decagon in any circle.

Take any circle M, and you can make a regular

decagon (a regular 10-sided polygon) inside it as A
follows. L B

K C
[1] Inscribe a regular pentagon inside the
circle, ACEGK (Thm.7). M
H D
[2] Bisect each of the five equal arcs cut off
by its sides, namely arc AC at B, arc CE at D, arc
G E
EG at F, arc GK at H, and arc KA at L.
F
[3] Draw AB, BC, CD, DE, EF, FG, GH, HK, KL, LA.

[4] Since the five arcs cut off by the sides of the pentagon are equal, their halves are
equal. And since those ten arcs are equal, the chords joining their endpoints are equal
(Ch.3, Thm.22 Converse), i.e.
AB = BC = CD = DE = EF = FG = GH = HK = KL = LA
Thus the decagon ABCDEFGHKL is equilateral.

[5] Draw MA, MB, MC, thus forming isosceles triangles AMB and BMC.

[6] Now, AB = BC (Step 4)

and MA = MC (being radii of the circle)
and MB is common to ΔMBA and ΔMBC
so ΔMBA ≅ ΔMBC (Side-Side-Side)

[7] Similarly we can prove that all ten of the isosceles triangles having a side of the
decagon as base and having center M as vertex are identical. And therefore they all have
the same base angle. But every angle of the decagon is equal to two of these identical
base angles (e.g. ∠ABC = ∠MBA + ∠MBC), and so all of the angles of the decagon are
equal.

[8] Since decagon ABCDEFGHKL, inscribed in the circle, is both equilateral (Step
4) and equiangular (Step 7), it is regular.

Q.E.F.

145
THEOREM 9 Remarks

1. Now we have succeeded in making regular rectilineal figures of 3 sides, 4 sides, 5

sides, 6 sides, and 10 sides. What about 7, 8, and 9 sides? The 7 and 9-sided regular
polygons require more sophisticated tools than straight lines and circles, but the regular
octagon is easy – just bisect the four arcs of a circle around the four sides of the inscribed
square. In fact, given any regular polygon that we can inscribe in a circle, we can always
make another with twice as many sides just by bisecting the arcs of the circle. We cannot
make the 11-sided regular polygon using only straight lines and circles. But the 12-sided
regular polygon we can do. Just bisect the six arcs of a circle around the six sides of the
inscribed regular hexagon. The 13 and 14-sided regular polygons transcend the powers
of our elementary geometry.

But we can do the 15-sided regular polygon!

If AB were the side of the regular 15-sided figure
A inscribed in a circle with center C, then ∠ACB
would be one fifteenth of 360°, namely 24°. So if
C we can make a 24° angle, we can cut the circle into
B
15 equal parts and have the figure we seek. How?
Well, the angle drawn from the center of a circle
and standing on one side of the inscribed decagon is
one tenth of 360°, i.e. 36°, so we can make that
angle. And we can also make the 60° angle, since that is the angle of an equilateral
triangle. So if we draw these two angles together with a common side, the difference
between them will be an angle of 24°. Voilà.

Obviously, we can also do the 16-sided, by starting with the regular octagon (the
“stop sign”) and bisecting.
Now what about the regular 17-sided figure? (This figure is called a regular
“heptadecagon.”) Believe it or not, we actually can make it using nothing but circles and
straight lines! This was not known until the great German mathematician Carl Friedrich
Gauss (1777 – 1855) discovered it. He also found the general formula for determining
when it is possible to make a regular polygon of a given prime number of sides using
nothing but circles and straight lines, and when not.

146
 A
B 2. Having mentioned the 12-sided regular polygon or
dodecagon, it is worth adding here an interesting little
D
C theorem about it: Its area is exactly 3 times that of the square
on the radius of the circle in which it is inscribed.
M
Given: Regular dodecagon ABC etc. in circle with center M.
Prove: The dodecagon’s area = 3 £BM.

Join AC, and where it cuts radius BM call D.

To begin, note that the dodecagon consists of 12 identical isosceles triangles such
as MAB. Thus

Area Dodecagon = 12 rMAB

Now by the regularity of the figure, AB = BC, and ∠ABD = ∠CBD, but BD is common
to triangles ABD and CBD. Hence they are congruent, and so the adjacent angles ADB
and CDB are right angles. Therefore AD is the height of rMAB, taking BM as base.
Since rMAB is half the rectangle of that height and on that base, i.e. it is half rectangle
MB · AD, therefore 12 triangles such as MAB equal 6 such rectangles. Therefore

Area Dodecagon = 6 MB · AD

But AC is double AD, and therefore MB · AC is double MB · AD, since if one

rectangle has the same height as another, but double its base, it will be twice as big. And
accordingly 6 rectangles such as MB · AD will equal only 3 rectangles such as MB
· AC. Therefore

Area Dodecagon = 3 MB · AC

And since AC cuts off one sixth of the circumference of the circle, it is the side of the
regular hexagon, which is equal to the radius of the circle. Therefore AC = MB, so that
the rectangle MB · AC is the same as the square on MB, or the square on the radius.
Hence

Area Dodecagon = 3 £MB.

THEOREM 9 Questions

1. How many degrees are in one angle of the regular decagon?

2. In the diagram for Theorem 9, what kind of triangle is triangle AMB? Compare it
to golden triangle AGC, and apply the theorem that says ∠AMC is double ∠AGC.

147
“HOOK”: THE 3-4-5 TRIANGLE AND THE CIRCLE.

B
If EP is 3 units of length, and PB is 4, and BE
is 5, then EPB is a “3-4-5 right triangle.” If we
inscribe the circle in it, center C, its radius will
have a length of 1 unit.

C
A

Also, if LMN is a 3-4-5 right triangle in which P K

E
LM = 4 and LN = 5, and if we draw a circle
whose diameter lies along LM and which is tangent to MN at M and also to LN at T, this
will be the equivalent of inscribing a circle in the isosceles triangle which is made up of
two triangles like LMN, symmetrical about LM. Let O be the center of this circle, and
join NO, cutting the circle at P & Q. Then PQN is cut in “mean and extreme ratio,” also
called “the golden ratio,” about which we will learn more in Chapters 5 and 6.

T
P

N
M

148
“HOOK”: BRIANCHON’S THEOREM.

Take any circle and draw a tangent to it, RS, of whatever length you please. From S,
draw another tangent, ST. Continue doing this so that you end up with a hexagonal
figure circumscribed about the circle, RSTVWZ. Join the “opposite” vertices of this
figure: RV, SW, TZ. These three straight lines will intersect in a single point.

Z
X

V
W

“HOOK”: CHORD PRODUCTS OF REGULAR POLYGONS.

In a circle with center O and a unit-long radius OR, draw a regular polygon of any
number of sides you like, N, using R itself as the starting vertex. If we now join R to all
the other vertices, we
get chords (such as B
RA, RB, RC, etc.).
Since we are calling
the radius “1,” it is C A
decided for us what
we have to call each
of the other chords. If
we then multiply all
the lengths of these O
chords together, what D R
will the result be? It
will always be N. For
instance, if our
polygon is a hexagon,
the chord-product will
be 6. If an octagon, 8. E G
And so on.
F

149
 Chapter Five

Proportion in General
 DEFINITIONS

1. A lesser quantity is said TO MEASURE a greater one if it goes into it exactly a whole
number of times.
For example, a foot measures a yard, since it goes into it exactly three times; but
it does not measure a meter, since three feet is less than a meter, but four feet is more
than a meter.

2. When a lesser quantity measures a greater one, the lesser is called A MEASURE of
the greater, and the greater is called A MULTIPLE of the lesser.
For example, a foot is a measure of a yard, and a yard is a multiple of a foot. If
B is a multiple of A, say 3 times A, then the shorthand notation for this is B = 3A, which
reads " B is three A", or " B is three times A".

3. We form EQUIMULTIPLES of two quantities whenever we multiply each of them

the same number of times.
For example, 12 and 15 are not only multiples of 4 and 5 respectively, but
equimultiples of 4 and 5, since 12 and 15 are three times 4 and 5 respectively.

4. Whenever two quantities are capable of exceeding each other by being multiplied
enough times, the relative size of one to the other is called a RATIO.
For example, a foot, when multiplied four times, exceeds a yard, and a yard, when
multiplied two times, exceeds four feet. So the relation of a foot to a yard is a "ratio",
namely the ratio of 1 to 3, and the yard also has a ratio to the foot, namely that of 3 to
1.
But a line does NOT have a ratio to an area, such as a rectangle. Even though
each is a kind of quantity, the line has no area at all, and so no multiple of it can ever
exceed the area of the rectangle. There is, then, no comparison of these two quantities
with respect to size.
If a quantity A has a ratio to another quantity B, the shorthand notation for their
ratio is A : B, which reads "A to B". The order of the terms in a ratio makes a
difference. For example, if A has to B the ratio of 1 : 2 , then A is HALF of B. On the
other hand, if A has to B the ratio of 2 : 1 , then A is the DOUBLE of B.

5. The first term in a ratio is called its ANTECEDENT, and the second term in a ratio is
called its CONSEQUENT.

149
6. Given four quantities in two ratios, taking CORRESPONDING MULTIPLES of them
means taking equimultiples of the antecedents, and again taking equimultiples of the
consequents.
For example, given quantities A, B, C, D in the ratios A : B and C : D, if we
take multiples of A and B such as 3A and 5B, then the corresponding multiples of C
and D are 3C and 5D.

7. One pair of quantities in a ratio is said to COMPARE THE SAME WAY as another
pair when each pair is the ratio of a greater to a lesser, or when each is the ratio of a lesser
to a greater, or when both pairs are equalities.
For example, the numbers in the ratio 5 : 3 and those in the ratio 7 : 4 compare
the same way, since 5 > 3 and 7 > 4.
Again, the numbers in the ratio 5 : 5 and those in the ratio 7 : 7 compare the
same way, since 5 = 5 and 7 = 7.
But the numbers in the ratios 6 : 4 and 9 : 12 do not compare the same way,
since 6 > 4 but 9 < 12.

8. If the multiples of two quantities always compare the same way as the corresponding
multiples of two other quantities, then the first two quantities have the SAME RATIO as
the other two.
For example, suppose you have two ratios A : B and C : D. And suppose their
corresponding multiples compare the same way, e.g.
5A > 3B and 5C > 3D
2A < 4B and 2C < 4D
and so on. If this is true for all their corresponding multiples, then A has to B the
same ratio that C has to D.

9. If one quantity has to a second the same ratio that a third has to a fourth, then the four
quantities are said to be PROPORTIONAL.
If quantity A has to B the same ratio that C has to D, the shorthand notation for
that proportion is A : B = C : D , which reads "A is to B as C is to D", or, if you
prefer, "the ratio of A to B is the same as the ratio of C to D".
The four terms in a proportion need not all be different. For example, three terms
X, Y, Z can be in proportion like this: X : Y = Y : Z.

10. By contrast, when it is possible to find a multiple of a 1st quantity greater than some
multiple of a 2nd, but the corresponding multiple of a 3rd quantity is not greater than the
corresponding multiple of a 4th, then the 1st has to the 2nd a GREATER RATIO than the
3rd has to the 4th.
For example, take 3, 5, 6, and 17 in that order. Let's multiply 3 and 5 by 4 and 2
respectively, and likewise multiply 6 and 17 by 4 and 2 respectively.

150
 Notice that 4(3) > 2(5), whereas 4(6) < 2(17). We took multiples of 3 and 5, and
then corresponding multiples of 6 and 17, but the two pairs of multiples do not compare
the same way. Since the multiple of 3 exceeded the multiple of 5, whereas the
corresponding multiple of 6 fell short of the corresponding multiple of 17, therefore 3 has
to 5 a greater ratio than 6 has to 17. The way to write this is 3 : 5 > 6 : 17.
Again, take 5, 2, 3 and 4 in that order. Multiply 5 and 2 by 8 and 6, and
also 3 and 4 by 8 and 6.
Now 8(5) > 6(2) but 8(3) = 6(4)
hence 5 : 2 > 3 : 4.

11. If one term in a ratio is not greater than the other, then it is either less than or equal to
it. The symbol for this is ≤. Again, if one term in a ratio is not less than the other, then it
is either greater than or equal to it. The symbol for this is ≥ .
Using these symbols, we can define “greater ratio” as follows.
If A : B > C : D,
then for some pair of numbers n and m it will happen that
nA > mB but nC ≤ mD.

151
 THEOREMS

THEOREM 1: Given any number of quantities, the sum of their equimultiples is

that same multiple of their sum. Again, given any number of quantities, the difference of
their equimultiples is that same multiple of their difference.

This is clear enough in examples.

Suppose you have two quantities which are able to be added

together, namely A and B. Take the same multiple of each one,
say 3 times, giving us equimultiples 3A and 3B.
Then the sum of these equimultiples is also 3 times the
sum of the original quantities. That is: A B

3A 3B
3A + 3B = 3(A + B)

This is because we can subtract A + B from 3A + 3B exactly

three times.
And, for the same reason, no matter how many original
quantities we have, and no matter what multiple we choose to
take of them all, the sum of the multiples will always be that
same multiple of the sum of the original quantities. So, for
example,

5A + 5B + 5C = 5(A + B + C)

and so on.

The same goes for the differences between unequal

G quantities. Suppose G is greater than K, and you take any
equimultiples of G and K, say 3G and 3K. Then
G-K K
3G – 3K = 3(G – K).

If we set out three G's, as in the diagram, and subtract one

K from each, there will be three remainders of G – K.

Q.E.D.

152
THEOREM 1 Remarks

1. Confirm this Theorem with some numerical examples:

3×6 + 3×5 + 3×4 = 3(6 + 5 + 4)

7×36 – 7×19 – 7×11 = 7(36 – 19 – 11)

2. Another very basic truth about multiples is this: If we multiply a quantity by one
number, and then the result by another number, it makes no difference which multiplier is
used first. For example, if we double A and then triple its double, we get 6 times A, and
again if we triple A and then double its triple, we still get 6 times A. In other words,
3(2A) = 2(3A).

THEOREM 2: Proportional quantities are proportional inversely.

Take any four quantities in proportion, say A : B = C : D.

I say they are also proportional inversely, that is B : A = D : C.

Here's why.

A B C D

[1] Take any multiples of A and B, say 5A and 3B.

Take the corresponding multiples of C and D, namely 5C and 3D.

[2] Since A : B = C : D, therefore

if 5A > 3B, then 5C > 3D
but if 5A < 3B, then 5C < 3D (Def. 8),
and that is true for any multiples of A and B and the corresponding multiples of C and D.

[3] Restating the information in Step 2, but putting B & D first in each case, we have
if 3B < 5A, then 3D < 5C
but if 3B > 5A, then 3D > 5C,
and that will be true for any multiples of B and A, and the corresponding multiples of D
and C.

[4] So taking any multiples of B and A, the corresponding multiples of D and C must
compare the same way. Therefore
B:A = D:C (Def. 8).

Q.E.D.

153
THEOREM 2 Remarks

Here is a numerical example of the Theorem:

since 3 : 4 = 6 : 8,
so too 4 : 3 = 8 : 6.

THEOREM 3: Two ratios that are the same with a third ratio are the same as
each other.

Take any three pairs of quantities A and B, C and D,

X and Y, which are such that
A:B = X:Y
and C : D = X : Y.
Then A : B = C : D as well.
A B C D X Y
Here's the proof.

[1] Take any multiples of A and B, say 5A and 3B.

Take the corresponding multiples of X and Y, namely 5X and 3Y.
Take the corresponding multiples of C and D, namely 5C and 3D.

[2] Suppose 5A > 3B.

Then, since A : B = X : Y, (given)
therefore 5X > 3Y. (Def. 8)

[3] Now C:D = X:Y (given)

But 5X > 3Y,
therefore 5C > 3D. (Def. 8)

[4] So if 5A > 3B, then 5C > 3D (Steps 2 & 3)

Likewise we can prove that
if 5A < 3B, then 5C < 3D.

[5] So, taking any multiples of A and B, it turns out that they must compare the same
way as the corresponding multiples of C and D.
Therefore A : B = C : D. (Def. 8)

Q.E.D.

154
THEOREM 3 Remarks

1. That two quantities equal to the same thing are equal to each other is a self-
evident axiom. But the fact that two ratios the same as a third ratio are also the same as
each other needs to be proved, because “same ratio” has a rather involved definition.

2. Notice that it is not necessary for all 4 quantities to have a ratio in order for the
Theorem to be true. For example, two lines may have the same ratio as two numbers (for
example, the ratio of being double), and two areas may also have that same ratio. Then it
will follow that the two lines and the two areas have the same ratios, even though neither
line has any ratio to either area.

THEOREM 4: No ratio can be greater than another ratio, and at the same time
less than it.

If possible, suppose that A:B > C:D

and yet at the same time A : B < C : D.
Watch what follows from each supposition.

A B C D
For some pair of numbers, say 7 and 3,

7A > 3B but 7C ≤ 3D (since A : B > C : D; Def. 11)

Again, for some pair of numbers, call them N and M,

NA ≤ MB but NC > MD (since A : B < C : D; Def.11)

Now, multiply both sides of the top two inequalities by N, and multiply both sides of the
bottom two inequalities by 7, and we get four new inequalities:

7NA > 3NB 7NC ≤ 3ND

and 7NA ≤ 7MB 7NC > 7MD
so 7MB > 3NB and 3ND > 7MD,
in each case taking the first and last terms among three unequal things.

Hence 7M > 3N and 3N > 7M,

because in each case the greater multiple (whether of B or of D) must have a greater
multiplier. But the number 7M cannot be both greater than and less than the number 3N.
So it is also impossible that one ratio should be both greater than and less than another
ratio.

Q.E.D.

155
THEOREM 4 Remarks

We can now see that just as any two comparable quantities must either be equal, or one
greater than the other, likewise any two ratios must either be the same or one must be
greater than the other. Consider any two ratios, A : B and C : D. Either all multiples of
C and D corresponding to multiples of A and B compare in the same way, or some
don’t. If all do, then the two ratios are the same (Def. 8). If some do not, then one ratio
will be greater, and by the present Theorem, the other must be less.

THEOREM 5: If the multiples of two quantities are equal, and the

corresponding multiples of two other quantities are also equal, then the four original
quantites are proportional.

Given: Quantities A, B, C, D.
Numbers n and m such that
nA = mB and nC = mD

Prove: A:B = C:D

A B C D

[1] Take any random multiples of A and B, say xA and yB, where x and y are any
numbers you please. Suppose first that xm > ny.

[2] Then xmB > nyB and xmD > nyD

since the greater multiplier yields a greater multiple of the same thing.

[3] But xmB = xnA and xmD = xnC

since we are given that mB = nA and mD = nC, and we have multiplied them all by x.

[4] Hence xnA > nyB and xnC > nyD

putting together Steps 2 and 3.

[5] Dividing all by n, we have

xA > yB and xC > yD
Likewise if we assume instead that xm < ny, then
xA < yB and xC < yD
and if we assume instead that xm = ny, then
xA = yB and xC = yD

So, taking random multiples of A and B, the corresponding multiples of C and D

must compare in the same way. Hence
A:B = C:D (Def. 8)

Q.E.D.

156
THEOREM 5 Remarks:

Verify the theorem concretely by finding a numerical example.

THEOREM 6: Equal quantities have the same ratio to any one quantity.

A 2A
Take any two equal quantities A and B, and
any third quantity C to which they have a B 2B

ratio. I say that A : C = B : C. C 3C

[1] Take any multiples of A and C, say 2A and 3C.

Take the corresponding multiple of B, namely 2B.

[2] Since A = B (given)

thus 2A = 2B (equimultiples of equal things are equal)

[3] So if 2A > 3C, then also 2B > 3C (since 2A = 2B)

but if 2A < 3C, then also 2B < 3C (since 2A = 2B)
and if 2A = 3C, then also 2B = 3C (since 2A = 2B)

[4] Thus any multiples of A and C must compare the same way as the corresponding
multiples of B and C. Therefore
A:C = B:C (Def. 8).

Q.E.D.

THEOREM 6 Remarks

1. We used the numbers 2 and 3 in this Theorem, but there is nothing special about
them – if you look at the proof carefully, you will see that no Step depends on those
specific numbers. We could have used general symbols for any two numbers (such as n
and m) just as easily.

2. With numbers, this Theorem is an especially trivial piece of obviousness, i.e.

equal numbers have the same ratio to the same number:
5 : 7 = 5 : 7.

3. It should also be clear that the same quantity has the same ratio to equal
quantities: If A = B, then C : A = C : B. For, by the present Theorem, A : C = B : C,
but by Theorem 2, these are proportional inversely, i.e. C : A = C : B.

157
THEOREM 7: How to find equimultiples of two unequal quantities such that the
multiple of the greater will be greater than some multiple of a third quantity, while the
multiple of the lesser will be less.

Suppose A > B, and C is any third quantity A

comparable to them. B
Is it possible to multiply both A and B by the
same number N, and C by some number R, such that C
NA > RC
but NB < RC ?

Believe it or not, no matter how close A and B are to being equal, no matter how small
the difference between them is, we can always do it. Here's how:

[1] There must be a first multiple of (A – B) that is greater than C.

Let it be 9(A – B).
So 9(A – B) > C.

[2] There must also be a first multiple of C that is greater than 9B.
Let it be 4C.
Since 4C is the first multiple of C that exceeds 9B,
hence 9B is greater than or equal to 3C
so C + 9B is greater than or equal to 4C (adding C to each)

[3] Now 9(A – B) > C (Step 1)

so 9A – 9B > C (Thm.1)
so 9A > C + 9B (adding 9B to each)
but C + 9B ≥ 4C (Step 2)
so 9A > 4C

[4] So 9A > 4C (Step 3)

but 9B < 4C (Step 2)

We did it!

Q.E.F.

THEOREM 7 Remarks:

1. As with Theorem 6, there is nothing special about the numbers 9 and 4 occurring
in this Theorem. Use n and m, or some other pair of numbers, if you prefer.

158
2. Let's try a numerical example. Let
A = 20
B = 19
C = 23
We want to multiply 20 and also 19 by the same number, so that the multiple of 20 is
greater than some multiple of 23, while the multiple of 19 is less.
Following what we did in the Theorem, (A – B) = 1. And the first multiple of this
that is greater than 23 is 24. That is Step [1]. Following Step [2], we next find the first
multiple of 23 that is greater than 19 × 24. Well,
19 × 24 = 456
and 20 × 23 = 460, so that is the first multiple of 23 that is greater than 19 × 24.
According to the Theorem, the multiples satisfying the requirements are:
24 × A = 480
24 × B = 456
20 × C = 460

3. Find a way to multiply 36 and 35 by the same number so that the multiple of 36
exceeds some multiple of 57, but the multiple of 35 does not.

THEOREM 8: Of two unequal quantities, the greater one has a greater ratio to
any other quantity than the lesser one has to it.
A
Let A and B be unequal quantities, A the greater one,
B
and C any other quantity comparable with them. Then
A : C > B : C. C

The proof is as follows.

[1] Since A > B, therefore we can find a multiple of A that is greater than some
multiple of C, while the same multiple of B is less (Thm.7). Suppose then that
5A > 3C
but 5B < 3C.

[2] Therefore it is possible to take a multiple of A that exceeds a multiple of C (i.e.

5A > 3C), while the corresponding multiple of B does not exceed that same multiple of
C (i.e. 5B < 3C). Therefore
A:C > B:C (Def. 9)

Q.E.D.

159
THEOREM 8 Remarks

1. Here is a numerical example. Since 7 > 3, then if we take any third number, such
as 5, it follows that 7 : 5 > 3 : 5.

2. Also, of two unequal quantities, any other quantity has to the lesser one a greater
ratio. Say A and B are unequal, A the greater, and C is any third quantity comparable to
them. Then C : B > C : A. Why?
Since A > B, we can find a multiple of A that is greater than some multiple of C,
while the same multiple of B is less (Thm.7). Say
3C > 5B
but 3C < 5A.
Therefore it is possible to take multiples of C and B (namely 3C and 5B), such that the
multiple of C is greater than that of B, while the same multiple of C is less than the
corresponding multiple of A. Therefore
C:B > C:A (Def. 10).

THEOREM 9: A first quantity has to a second quantity a greater ratio than a

quantity less than the first has to a quantity greater than the second.

Suppose A and B are any two quantities having a ratio,

and C < A
and D > B.
Then A : B > C : D.

Here's why: A B C D

[1] Since A > C, (given)

therefore it is possible to find some multiple of A that exceeds a certain multiple of B,
while the same multiple of C falls short of it (Thm.7). Suppose, then, that
5A > 2B
while 5C < 2B

[2] Since D > B, (given)

thus 2D > 2B (doubling each)

but 5C < 2B (Step 2)
so 2D > 5C

160
[3] So 5A > 2B (Step 1)
but 5C < 2D (Step 2)
And therefore it is possible to take a multiple of A that is greater than some multiple of B
whereas the corresponding multiple of C is less than the corresponding multiple of D.
Therefore A:B > C:D (Def.10)

Q.E.D.

THEOREM 9 Remarks

1. Here is a numerical example.

4 < 9
and 12 > 3,
so 9 : 3 > 4 : 12.

2. Also, if A > B and C is not greater than D, then

A : B > C : D. For if C is not greater than D, it is either
equal to it, or less than it.
First suppose that C = D. Then 3A > 3B, but
3C = 3D; and thus a multiple of A is greater than a
multiple of B, while the corresponding multiple of C is
not greater than the corresponding multiple of D.
Therefore A : B > C : D (Def. 10).
A B C D
Now suppose that C < D. Then 3A > 3B, but
3C < 3D, and so again A : B > C : D (Def. 10).

THEOREM 10: Quantities having the same ratio to the same quantity are equal.

Suppose two quantities A and B have the same ratio to C, that A

is A : C = B : C.
B
Then A = B.
C
Proof:

[1] If A were greater than B, then A:C > B:C (Thm.8)

But in fact A:C = B:C (given)
Therefore A is not greater than B.

161
[2] If A were less than B, then A:C < B:C (Thm.8)
But in fact A:C = B:C (given)
Therefore A is not less than B.

[3] Since A is not greater than B, nor is it less than B, therefore it is equal to B.

Q.E.D.

THEOREM 10 Remarks

1. This Theorem is the converse of Theorem 6.

2. This Theorem is an important tool for proving the equality of all kinds of
quantities. Whenever we can derive a proportion such as this: X : Z :: Q : Z, we can
conclude from this Theorem that X = Q.

THEOREM 11: Of two quantities, the one having a greater ratio to any third
quantity is greater.
A
Suppose you have three quantities A, B, C, and A : C > B : C.
B
Then A > B.
C

Proof:

[1] If A were equal to B, then A:C = B:C (Thm.6)

But in fact A:C > B:C (given)
Therefore A is not equal to B.

[2] If A were less than B, then A:C < B:C (Thm.6)

But in fact A:C > B:C (given)
which is impossible, since two ratios cannot be greater than each other (Thm. 4).
Therefore A is not less than B.

[3] Since A is not equal to B, nor is it less than B, therefore it is greater than B.

Q.E.D.

162
THEOREM 11 Remarks

Also, of two quantities, the one to which any third quantity has a lesser ratio is greater.
Suppose that C : B > C : A. Then it has to be that A > B. Why?
(1) If A = B, then C : B = C : A (Thm.6, Remark 3)
but C:B > C:A (given)
Therefore A is not equal to B.
(2) If A < B, then C : A > C : B (Thm.8, Remark 2)
but C:B > C:A (given)
and two ratios cannot be greater than each other (Thm.4).
Therefore A is not less than B.
(3) Since A is not equal to B (Step 1), and A is not less than B (Step 2),
therefore A is greater than B.

THEOREM 12: If the first term in a proportion is greater than the third, then the
second will be greater than the fourth (but if less, less, and if equal, equal).

Suppose A:B = C:D

and A > C.
Then also B > D.

Proof:
A B C D

[1] First try to assume that B = D.

So B = D (assumed)
But A > C (given)
so A:B > C:B (Thm.8)
or A:B > C:D (assuming B = D)
But A:B = C:D (given)
So B is not equal to D.

[2] Next try to assume that B < D.

So B < D (assumed)
But A > C (given)
so A:B > C:D (Thm.9)
But A:B = C:D (given)

163
 So B is not less than D.

[3] Since B is not equal to D (Step 1)

and B is not less than D (Step 2)
thus B > D.

[4] Similarly, we can prove that

if A < C, then also B < D,
and if A = C, then also B = D.

Q.E.D.

THEOREM 13: Any two quantities have the same ratio as their equimultiples,
taken in corresponding order.

Let A and B be any two quantities in a ratio, and take any

equimultiples of them, say 3A and 3B.
I say that A : B = 3A : 3B.

[1] Take any multiples of A and B, such as 2A and 5B.

Take the corresponding multiples of 3A and 3B,
namely 2(3A) and 5(3B).

[2] Now, if 2A > 5B A B 3A 3B

then 3(2A) > 3(5B)

since equimultiples of unequals are unequal in the same order.
Thus 2(3A) > 5(3B)
since the order of multiplication makes no difference (as we noted in the Remarks to
Theorem 1). In sum,
if 2A > 5B then 2(3A) > 5(3B).

[3] Likewise, if 2A < 5B then 2(3A) < 5(3B),

and if 2A = 5B then 2(3A) = 5(3B).

[4] So, taking any multiples of A and B, it turns out that the corresponding multiples
of 3A and 3B must compare the same way. Therefore A : B = 3A : 3B (Def. 8).

Q.E.D.

164
THEOREM 13 Remarks

1. It follows also that two quantities have the same ratio as their corresponding
measures. For example: A : B = ⅓A : ⅓B.
The reason is that A and B are equimultiples of ⅓A and ⅓B, and so, by the
present Theorem, the four of them are proportional.

2. Clearly, too, if A : B = C : D, then any equimultiples of A and B will be

proportional to any multiples of C and D. For example, 5A : 5B = 3C : 3D.
For 5A : 5B = C : D (Thm.13)
but 3C : 3D = C : D (Thm.13)
hence 5A : 5B = 3C : 3D (Thm.3)

3. Also, Corresponding multiples of proportional quantities are proportional.

i.e. if A:B = C:D
then 3A : 2B = 3C : 2D
(I chose 3 and 2 for my multiples, here, but any numbers would do).
To prove this, simply take any multiples of 3A and 2B, say 5(3A) and 6(2B).
Take also the corresponding multiples of 3C and 2D, i.e. 5(3C) and 6(2D).

Now if 5(3A) > 6(2B)

then 15A > 12B (simply by multiplying)
thus 15C > 12D (since A : B = C : D)
so 5(3C) > 6(2D)

So if 5(3A) > 6(2B) then 5(3C) > 6(2D).

In short, any random multiples being taken of 3A and 2B, the corresponding multiples of
3C and 2D must compare the same way. Therefore
3A : 2B = 3C : 2D.

165
THEOREM 14: Alternating the terms in a proportion makes a new proportion.

Suppose that A : B = C : D.
Then also A : C = B : D.

Here's why: A B C D

[1] Take any multiples of A and C, say 7A and 2C.

Take the corresponding multiples of B and D, namely 7B and 2D.

[2] Now, since A : B = C : D, (given)

then 7A : 7B = 2C : 2D (Thm.13, Remark 2)

[3] So if 7A > 2C, then 7B > 2D (Thm.12)

but if 7A < 2C, then 7B < 2D (Thm.12)
and if 7A = 2C, then 7B = 2D (Thm.12)

[4] So, taking any multiples of A and C, it turns out that the corresponding multiples
of B and D must compare in the same way. Therefore
A:C = B:D (Def. 8)

Q.E.D.

THEOREM 14 Remarks

1. Of course, the proportion will not alternate unless all four quantities are
comparable. For example, if two lines have the same ratio as two areas, we cannot
alternate the proportion, since then we would be saying that a line has to an area the same
ratio that another line has to another area, whereas, in fact, a line does not have any ratio
to an area at all.

2. The new ratios resulting from alternating the terms in the original proportion do
not have to be the same as the ratios in the original proportion. For example, given
2 : 4 = 3 : 6,
if we alternate that proportion, we get a new proportion:
2 : 3 = 4 : 6.
But these two new ratios are not the same as the two original ratios.

166
THEOREM 15: If any two ratios are the same as a third, then the sums of their
corresponding terms also have that same ratio.

Suppose that A:B = C:D

and also that A:B = E:F
Then A:B = C+E:D+F

A B C D E F

[1] Take any multiples of A and B, say 2A and 3B.

Take the corresponding multiples of C, D, E, F, namely 2C, 3D, 2E, 3F.

[2] If 2A > 3B,

then 2C > 3D (since A : B = C : D)
and 2E > 3F (since A : B = E : F).

But then 2C + 2E > 3D + 3F, (adding)

that is 2(C + E) > 3(D + F) (Thm.1)

[3] So if 2A > 3B, then 2(C + E) > 3(D + F).

Likewise if 2A < 3B, then 2(C + E) < 3(D + F).
and if 2A = 3B, then 2(C + E) = 3(D + F).

[4] Therefore, taking any multiples of A and B, we find that whenever they are
unequal, the corresponding multiples of (C + E) and (D + F) are also unequal and in the
same order. Therefore A : B = C + E : D + F.
Q.E.D.

THEOREM 15 Remarks

1. Obviously C D E F must be the same kinds of quantities so we can add them

together. As an example, let's use numbers:
1:2 = 3:6
and 1 : 2 = 5 : 10
so 1 : 2 = 3 + 5 : 6 + 10.

2. Another similar and important Theorem is this:

Given: A : B = C : D
Prove: A + B : B = C + D : D

167
Take any multiples of A + B and B, e.g. 2(A + B) and 3B.
Take the corresponding multiples of C + D and D, i.e. 2(C + D) and 3D.

Now, if 2(A + B) > 3B

then 2A + 2B > 3B (Thm.1)
so 2A > B (subtracting 2B from each side)
thus 2C > D (since A : B = C : D)
so 2C + 2D > 3D (adding 2D to each side)
i.e. 2(C + D) > 3D (Thm.1)

So if 2(A + B) > 3B then 2(C + D) > 3D.

Likewise if 2(A + B) < 3B then 2(C + D) < 3D.
and if 2(A + B) = 3B then 2(C + D) = 3D.

That is, taking any random multiples of (A + B) and B, they must compare the same way
as the corresponding multiples of (C + D) and D. Therefore A + B : B = C + D : D.

Q.E.D.

Notice that this Theorem does not depend on A, B, C, and D being all comparable
quantities. A and B must be comparable, and C and D also, but A and B might be
volumes, and C and D might be areas.

THEOREM 16: If any two ratios are the same as a third, then the differences of
their corresponding terms also have that same ratio.

Suppose that A : B = C : D
and also that A : B = E : F
then A : B = (C – E) : (D – F)

A B C D E F

We will start by proving that (C – E) and (D – F) have the same ratio as E and F, and
from there it will be easy to show they have the same ratio as A and B. Accordingly,

[1] Take any multiples of (C – E) and E, say 3(C – E) and 5E.

168
[2] Now, C : D = E : F,
as is clear from what is given. Therefore, alternately
C:E = D:F (Thm.14)
So if 3C > (5 + 3)E then 3D > (5 + 3)F (Def. 8)
i.e if 3C > 5E + 3E then 3D > 5F + 3F (Thm.1)
so if 3C – 3E > 5E then 3D – 3F > 5F (subtracting 3E and 3F)
so if 3(C – E) > 5E then 3(D – F) > 5F (Thm.1)

[3] By the same kind of argument,

if 3(C – E) < 5E then 3(D – F) < 5F
and if 3(C – E) = 5E then 3(D – F) = 5F

[4] So, taking random multiples of (C – E) and E, they must compare the same way
as the corresponding multiples of (D – F) and F. Therefore
(C – E) : E = (D – F) : F (Def. 8)

[5] Alternating this proportion, we have

(C – E) : (D – F) = E : F (Thm.14)

[6] But since E : F = A : B (given), it follows that

A : B = (C – E) : (D – F) (Thm.3)

Q.E.D.

THEOREM 16 Remarks

1. Here is a numerical example:

1 : 2 = 7 : 14
and 1:2 = 3:6
so 1 : 2 = 7 – 3 : 14 – 6

2. In Step 5 we alternated a proportion. How do we know that C, D, E, and F are all

comparable quantities, such that we can alternate the proportion? We must presume this
in what we are given: if they were not all comparable, then we could not even subtract E
from C, for example.

3. A similar and important Theorem is this:

Given: A : B = C : D.
Prove: A – B : B = C – D : D.

Take any multiples of A – B and B, e.g. 2(A – B) and 3B.

Take the corresponding multiples of C – D and D, i.e. 2(C – D) and 3D.

Now, if 2(A – B) > 3B

169
 then 2A – 2B > 3B (Thm.1)
so 2A > 5B (adding 2B to each side)
thus 2C > 5D (since A : B = C : D)
so 2C – 2D > 3D (subtracting 2D from each side)
i.e. 2(C – D) > 3D. (Thm.1)

so if 2(A – B) > 3B then 2(C – D) > 3D.

Likewise if 2(A – B) < 3B then 2(C – D) < 3D
and if 2(A – B) = 3B then 2(C – D) = 3D

That is, taking any random multiples of (A – B) and B, we find they must compare the
same way as the corresponding multiples of (C – D) and D.
Therefore A–B:B = C–D:D (Def. 8)

Q.E.D.

FORMING A PROPORTION "EX AEQUALI"

THEOREM 17: If the antecedents in one proportion are also the antecedents in
a second proportion, then a new proportion arises by taking the consequents of the first
proportion as antecedents, and the consequents of the second proportion as consequents.

Suppose A:B = C:D

and also A:E = C:F
then B:E = D:F

B D
A E C F

Take any multiples of B and E, say 2B and 3E.

Take the corresponding multiples of D and F, i.e. 2D and 3F.

[1] Now, if 2B > 3E

then A : 2B < A : 3E (Thm.8, Remark 2)
but A : 2B = C : 2D (since A : B = C : D; & Thm.13 Remark 3)
thus C : 2D < A : 3E

170
[2] So C : 2D < A : 3E (Step 1)
but A : 3E = C : 3F (since A : E = C : F; & Thm.13 Remark 3)
thus C : 2D < C : 3F
therefore 2D > 3F (Thm.11 Remarks)

[3] Thus if 2B > 3E then 2D > 3F (Steps 1-2)

Likewise if 2B < 3E then 2D < 3F
and if 2B = 3E then 2D = 3F

[4] That is, taking any random multiples of B and E, they must compare the same
way as the corresponding multiples of D and F.
Therefore B : E = D : F. (Def. 8)

Q.E.D.

THEOREM 17 Remarks

1. Here is a numerical example:

3:6 = 4:8
and 3 : 9 = 4 : 12
so 6 : 9 = 8 : 12

2. According to the Theorem, whenever two proportions have equal antecedents,

their consequents form a proportion (taken in proper order). We are actually forming a
proportion just by “dropping out” the 2 pairs of identical antecedents in the original two
proportions, leaving us with just the consequents. Ancient mathematicians called this
procedure “forming a proportion ex aequali.”

3. It is not necessary for the antecedents of the two proportions to be the same in
order for the Theorem to work; they need only be proportional. Thus
if A:B=C:D
and E:F=G:H
and A : E = C : G (i.e. the antecedents of the first two proportions are proportional)
then B : F = D : H (i.e. the consequents of the first two proportions are proportional).
If you don't believe it, here is the proof:
A : B = C : D (given)
A : E = C : G (given)
so B : E = D : G (Thm.17)
so E : B = G : D (Thm.2)

171
but E : F = G : H (given)
so B : F = D : H (Thm.17). Q.E.D.

4. If B and D happen to be comparable quantities (as opposed to being a line and an

area, for example), the proof for Theorem 17 is much simpler:
A:C = B:D (Alternate of the first given proportion)
A:C = E:F (Alternate of the second given proportion)
so B:D = E:F (Thm.3)
thus B : E = D : F (Alternating)

FORMING A PROPORTION BY TRACING THE LETTER "U"

THEOREM 18: If the extreme terms in one proportion are also the extreme
terms in a second proportion, a new proportion arises by taking the middle terms of the
first proportion as extremes, and the middle terms of the second proportion as middles.

Suppose that A : B = C : D
and also that A : E = F : D
then B:E = F:C

A B C D Let's adopt the usual method, and try to show that B : E = F : C

A E F D by proving that random multiples of B and E must compare the
same way as the corresponding multiples of F and C.

Take any multiples of B and E, say 3B and 5E.

Take the corresponding multiples of F and C, namely 3F and
5C.

[1] Now if 3B > 5E

then A : 5E > A : 3B (Thm.8, Remark 2)
but A : 3B = C : 3D (A : B = C : D; Thm.13 Remark 3)
so A : 5E > C : 3D
but A : 5E = F : 5D (A : E = F : D; Thm.13 Remark 3)
thus F : 5D > C : 3D

172
[2] So C : 3D < F : 5D (Step 1)
but F : 5D = 3F : 15D (Thm.13; multiplying by 3)
thus C : 3D < 3F : 15D

[3] So C : 3D < 3F : 15D (Step 2)

but C : 3D = 5C : 15D (Thm.13; multiplying by 5)
thus 5C : 15D < 3F : 15D
therefore 3F > 5C (Thm.11)

[4] Thus if 3B > 5E then 3F > 5C (Steps 1 – 3)

Likewise if 3B < 5E then 3F < 5C
And if 3B = 5E then 3F = 5C

[5] That is, taking any random multiples of B and E, they must compare to each other
the same way as the corresponding multiples of F and C compare to each other.
Therefore B:E = F:C (Def. 8)

Q.E.D.

THEOREM 18 Remarks

1. Here is a numerical example:

4 : 3 = 16 : 12
and 4 : 2 = 24 : 12
so 3 : 2 = 24 : 16

2. As a mnemonic device for remembering the proportion that arises out of the
original two proportions, think of it as forming a “letter U”: if you look at B, E, F, and C,
in the original two given proportions, you are tracing out a letter “U”. And that is the
order in which they form a proportion: B : E = F : C. So we might call this “The U-
Theorem”, and we might call the procedure of forming a proportion in this way “Tracing
the U”. Ancient geometers called this a “perturbed proportion.”

3. For the Theorem to work, the extremes need not be the same, but only
proportional in the order of the letter U. Thus
if A:B = C:D
and E:F = G:H
and A:E = H:D (i.e. extremes of the 1st two proportions are proportional)
then B : F = G : C (i.e. the middles of the 1st two proportions are proportional)

The proof for it is as follows.

173
 First A:B = C:D (given)
and A:E = H:D (given)
so B:E = H:C (Thm.18)
so E:B = C:H (inverting)
but E:F = G:H (given)
so B:F = G:C (Thm.18)

Q.E.D.

“HOOK”: THREE MEANS.

There are many different kinds of “means” or “middles” between any two comparable
magnitudes, e.g. between two line-lengths. Three of the most interesting are the
arithmetic mean (or average), the geometric mean (or mean proportional), and the
harmonic mean:

ARITHMETIC MEAN

Definition: ma – a = b – ma

Formula: ma = (a + b)/2

GEOMETRIC MEAN

Definition: a : mg = mg : b

Formula: mg = √ab

HARMONIC MEAN

Definition: (mh – a) : a = (b – mh) : b

Formula: mh = (2ab)/(a+ b)

174
There are many interesting theorems relating these means. Here is one of them:

THEOREM: The product of two extremes is equal to the product of their arithmetic and
harmonic means.

Given: extremes a and b,

their arithmetic mean: (a + b)/2
their harmonic mean: (2ab)/(a + b)

Proof:

Multiply the two formulas as given, and the (a + b) terms cancel out, and so do the 2s,
leaving just (ab), the product of the extremes.

Q.E.D.

PORISM 1: The geometric mean mg is the geometric mean not only between the
original extremes a and b, but also between the arithmetic and harmonic means
between a and b.

For ab = mamh (Theorem above)

so (√a·b)(√a·b) = mamh

so ma : mg = mg : mh

Q.E.D.
D

A C E B

a b

175
PORISM 2:

In the accompanying figure, if a and b are our extremes,

then CD is their arithmetic mean,
DE is their geometric mean,
DK is their harmonic mean.

Since CD = CA = (a + b)/2, the radius is the arithmetic mean.

And DE = √a·b = the geometric mean between a and b.

And DK is the 3rd proportional such that

CD : DE = DE : DK

i.e. it is the 3rd proportional from the arithmetic and geometric means between a and b.
But only the harmonic mean between a and b can be that (according to Porism 1
above). Hence DK is the harmonic mean between a and b.

Q.E.D.

Again, it is plain that the following figure exhibits the three means between a and b:

a b

176
 Chapter Six

Proportions
in Plane Geometry
 DEFINITIONS

1. SIMILAR RECTILINEAL FIGURES are B C F G

those with corresponding angles equal and the 1 2
sides about the equal angles proportional. E H
For example, quadrilateral ABCD is A
D
similar to quadrilateral EFGH, because all their
X Y
corresponding angles are equal (e.g. ∠1 = ∠2) 3
AND the sides about the equal angles are W
proportional when taken in corresponding
Z
order, i.e. AB : BC = EF : FG.
On the other hand, quadrilateral WXYZ, although equiangular with ABCD, is
NOT similar to it, because the sides about its angles have ratios different from the ratios
of the sides about the angles in ABCD. For example, ∠3 = ∠1, but AB : BC is not the
same as WX : XY.
Also, the sides of a rhombus are proportional to the sides of a square, since they
are all equal, but a rhombus is not similar to a square, since its angles differ from those
of a square.
From these examples, it should be clear that similar figures are the same shape,
but can have different sizes.

2. If ΔLMN is similar to ΔPQR, we can use the M

symbol ~ to designate that relationship. So “ΔLMN ~
ΔPQR” means “Triangle LMN is similar to triangle L
P

PQR”. N R

3. Quantities are CONTINUOUSLY PROPORTIONAL when they form a

proportion in which the consequent of the first ratio is the antecedent of the next ratio.
For example, 1, 2, 4 are continuously proportional, because 1 : 2 = 2 : 4.
Also, 3, 6, 12, 24 are continuously proportional, because 3 : 6 = 6 : 12 = 12 : 24.

4. To DUPLICATE a ratio means to find a third term in continuous proportion with

the two terms in the original ratio, and to form a new ratio of the first term to the third
term.
For example, to duplicate the ratio of 1 : 2, we find a third term in continuous
proportion with them, namely 4, since 1 : 2 = 2 : 4, and so 1 : 4 is the duplicate ratio
of 1 : 2.
Similarly, if we find two ratios in continuous proportion with the original ratio, as
in 1 : 2 = 2 : 4 = 4 : 8, then the extreme terms are said to have the triplicate ratio
of the original terms, i.e. 1 : 8 is the triplicate ratio of 1 : 2.

175
 C
5. Two figures have RECIPROCALLY PROPORTIONAL
D
SIDES if a side in the first figure is to a side in the second figure as
another side in the second figure is to another side in the first figure.
For example, the sides of rectangles C · D and E · K are E
reciprocally proportional if C : E = K : D.
K

6. A straight line is CUT IN MEAN AND EXTREME RATIO if the whole of it has
to the greater part of it the same ratio that the greater part of it has to the remaining part
of it.
For example, AB is cut in mean and extreme ratio at G if BA : AG = AG : GB.
A G B

AG is called the GREATER SEGMENT of AB, and GB is called the LESSER

SEGMENT.
Because of the beauty of this ratio (which is used in art and architecture and is
approximated by some parts of the human body and other things in nature), it is called the
GOLDEN RATIO. So a GOLDEN RECTANGLE is one whose sides are to each other
in the golden ratio.

176
 THEOREMS

THEOREM 1: Triangles having the same height are to each other as their bases.
Parallelograms having the same height are to each other as their bases.

Triangles DHE and EHG M H N

have the same height, HT.

I say that rDHE : rEHG
= DE : EG, that is, their
areas have the same ratio that A B C D E T G K L
their bases do.

[1] Extend the bases both ways.

Cut off any number of parts equal to DE, say DC, CB, BA.
Cut off any number of parts equal to EG, say GK, KL.

[2] Since AB = DE, therefore rAHB = rDHE (Ch.1, Thm.33)

Likewise rBHC and rCHD also equal rDHE.

[3] Since GK = EG, therefore rGHK = rEHG (Ch.1, Thm.33)

Likewise rKHL = rEHG.

[4] So rEHA = 4 rDHE, and EA = 4DE

and rEHL = 3 rEHG, and EL = 3EG
and, in general, when we multiply the base of either of the original triangles, the triangle
on that multiple base, and having height HT, also multiplies the area of the original
triangle the same number of times.

[5] Suppose rEHA > rEHL (i.e. 4 rDHE > 3 rEHG),

then also EA > EL (i.e. 4 DE > 3 EG),
since the base of the greater triangle will be greater, just as the bases of equal triangles
are equal, for triangles under the same height.
So if 4 rDHE > 3 rEHG, then 4 DE > 3 EG.
Likewise, if 4 rDHE < 3 rEHG, then 4 DE < 3 EG
and if 4 rDHE = 3 rEHG, then 4 DE = 3 EG

[6] And so whatever multiples of rDHE and rEHG we take, however they
compare, the corresponding multiples of their bases must compare the same way.
Therefore rDHE : rEHG = DE : EG (Ch.5, Def. 8)

[7] And since these triangles have the same ratio as their doubles (Ch.5, Thm.13), it
follows that the parallelograms DMHE and EHNG, being the doubles of these triangles
(Ch.1, Thm.33), also have the same ratio as the bases DE and EG.

Q.E.D.

177
THEOREM 2: If a straight line is drawn inside a triangle parallel to one of the
sides, it will cut the remaining sides proportionally; and if a straight line cut two sides of
a triangle proportionally, it will be parallel to the remaining side.

Take any triangle ABC. Draw PL parallel to BC. A

L
I say that AP : PB = AL : LC. C

Draw PC and BL.

P

[1] AP : PB = rAPL : rPBL (Thm.1) B

[2] rPBL = rLCP (same base, same height)

[3] AP : PB = rAPL : rLCP (putting together Steps 1 and 2)

[4] rAPL : rLCP = AL : LC (Thm.1)

[5] So AP : PB = AL : LC,
since each of these ratios is the same as rAPL : rLCP (by Steps 3 and 4), and since
two ratios the same with a third ratio are the same as each other (Ch.5, Thm.3).
And so the sides of the triangle have been cut proportionally. Q.E.D.

Again, suppose that in some triangle ABC the sides have been cut proportionally, such
that AP : PB = AL : LC. I say that PL is parallel to BC.

Draw PC and BL.

[1] AP : PB = AL : LC (given)
AP : PB = rAPL : rPBL (Thm.1)
so AL : LC = rAPL : rPBL (Ch.5, Thm.3)
but AL : LC = rAPL : rLCP (Thm.1)
so rAPL : rPBL = rAPL : rLCP

[2] Thus rPBL and rLCP have the same ratio to the same thing (namely to
rAPL).
Thus rPBL = rLCP (Ch.5, Thm.10)

[3] But rPBL and rLCP are on the same base LP. Since they are equal, therefore
they must also be under the same height (Ch.1, Thm.33 converse). Therefore P and L
must each be the same height above BC, i.e. PL is parallel to BC. Q.E.D.

178
THEOREM 2 Remarks

1. Notice we have just proven a theorem and its converse together.

X
2. Using the second part of this Theorem, you
can now easily prove the following interesting Q
theorem: If you bisect the sides of ANY quadrilateral W
WXYZ at the points Q, R, S, T, then QRST is a
R
parallelogram. Draw QR, RS, ST, TQ, WY, and XZ. T
Since WQ = QX and WT = TZ, therefore WQ : QX
= WT : TZ. Thus QT is parallel to XZ in rWXZ. Z
Finish the proof. S
Y

THEOREM 3: The line bisecting an angle of a triangle cuts the base into two
parts having the same ratio as the two sides.
Conversely, if the base of a triangle is cut into two parts having the same ratio as
the two sides, then the line joining the point of section to the opposite angle bisects the
angle.

Take any triangle ABC. Bisect ∠BAC with line P

AK. I say BK : KC = BA : AC.
Draw CP‖AK, cutting BA extended at P. 3

[1] Now ∠2 = ∠4 (CP‖AK) 1 2

and ∠2 = ∠1 (given) 4
so ∠1 = ∠4
but ∠1 = ∠3 (CP ‖AK) B K C
so ∠3 = ∠4

[2] Thus AP = AC (since ∠3 = ∠4 in Step 1)

[3] But BK : KC = BA : AP (since CP‖AK; Thm.2)

[4] So BK : KC = BA : AC (putting together Steps 2 and 3)

Q.E.D.

179
Conversely, take any triangle ABC, and suppose that BK : KC = BA : AC.
I say that AK bisects ∠BAC, i.e. ∠1 = ∠2.
Draw CP‖AK, cutting BA extended at P.

[1] Now BK : KC = BA : AC (given)

but BK : KC = BA : AP (since CP‖AK, Thm.2)
so BA : AC = BA : AP (Ch.5: 2 ratios the same as a 3rd are the same)
so AC = AP (Ch 5: 2 quantities with the same ratio to the same
quantity must be equal to each other).

[2] Thus ∠3 = ∠4 (angles opposite equal sides in a triangle are equal)

But ∠3 = ∠1 (since CP‖AK)
so ∠1 = ∠4
but ∠2 = ∠4 (since CP‖AK)
so ∠1 = ∠2

Q.E.D.

THEOREM 4: Equiangular triangles are similar.

X
Suppose ΔABC and ΔDCE are equiangular, i.e.
7
∠1 = ∠2
∠3 = ∠4 A

∠5 = ∠6 3
D
Then I say that the sides about their equal angles
are proportional when taken in corresponding 4
order, and thus rABC ~ rDCE. 1 5 2 6
B E
C

[1] Place BC and CE in a straight line.

[2] Extend BA and ED till they meet at X.

[3] Since ∠5 = ∠6 (given)

thus CA‖EX. (Ch.1, Thm.24 Remark 1)

[4] Since ∠1 = ∠2 (given)

thus CD‖BX (Ch.1, Thm.24 Remark 1)

180
[5] So ACDX is a parallelogram.

[6] And BA : AX = BC : CE (CA‖EX; Thm.2)

But AX = CD (opposite sides of a parallelogram)
so BA : CD = BC : CE
thus AB : BC = CD : CE (alternating the proportion)

[7] Now BC : CE = XD : DE (CD‖BX; Thm.2)

But XD = AC (opposite sides of a parallelogram)
so BC : CE = AC : DE
thus BC : AC = CE : DE (alternating the proportion)

[8] So AB : BC = CD : CE (Step 6)
and BC : AC = CE : DE (Step 7)
so AB : AC = CD : DE (Ch.5, Thm.17, ex aequali)

[9] So AB : BC = DC : CE (Step 6)
and BC : CA = CE : ED (Step 7)
and BA : AC = CD : DE (Step 8)
Therefore the sides about the equal angles in the two triangles are proportional.
Therefore rABC ~ rDCE.

Q.E.D.

THEOREM 4 Remarks

1. All we need is two angles, of course, because if two angles in a triangle are equal
to two angles in another triangle, they must be equiangular.

2. Can you prove the assumption in Step 2 that BA and ED must meet at some
point?
H
M
3. Notice that similar triangles need not be
oriented the same way; for example, one can be
“flipped” compared to the other, as rGHK and K N
L
rLMN. G

181
THEOREM 5: Triangles with proportional sides are similar.

Suppose rABC and rDEG have proportional sides, i.e.

AB : BC = DE : EG
BC : CA = EG : GD A
CA : AB = GD : DE 5 D

Then I say the angles contained by the 6

proportional sides are equal, i.e.
1 3 2 4
E G
B C
1=2
3=4
5=6
K

and thus rABC ~ rDEG. Here's an easy proof:

[1] Make rEGK on EG, such that it is equiangular with rABC.

Since rEGK is equiangular with rABC, therefore its sides are proportional to
the sides about the equal angles in rABC.
e.g. AB : BC = KE : EG
but AB : BC = DE : EG (given)
so KE : EG = DE : EG
thus KE = DE, since we learned in Ch.5 that quantities having the same ratio to the
same quantity must be equal to each other.

[2] Similarly, we can prove that

KG = DG.
And EG is common to both triangles.

[3] Thus rEGK ≅ rDEG (Side-Side-Side)

But rEGK is equiangular with rABC, by the construction in Step 1.
Thus rDEG is also equiangular with rABC.
But these two triangles are given as having the sides about those equal angles
proportional. Therefore they are also similar.

Q.E.D.

THEOREM 5 Remarks

There are 3 proportions given for this Theorem. Even if there were only two given, the
Theorem will still hold, because the third one would follow automatically from the other
two ex aequali, as illustrated in Theorem 4, Step 8.

182
THEOREM 6: Triangles with one angle of one equal to one angle of the other,
and with the sides about the equal angles proportional, are similar.

Suppose you have two triangles, B

K
rABC and rDEG, in which
1
∠1 = ∠2 E
7
and AB : BC = DE : EG. 2
Then rABC ~ rDEG. 8
5 3 6 4
The proof is nice and easy: A C D G

[1] Draw ∠7 = ∠1
Draw ∠8 = ∠3
Thus rKEG is equiangular with rABC.
Thus rKEG is similar to rABC (Thm.4), and so the sides about the equal
angles are proportional.

[2] So AB : BC = KE : EG
but AB : BC = DE : EG (given)
so KE : EG = DE : EG (these ratios are both the same as AB : BC)
so KE = DE (each has the same ratio to EG)

[3] But ∠7 = ∠2 (each being equal to ∠1)

and GE is common to rDEG and rKEG
so rDEG ≅ rKEG (Side-Angle-Side)

[4] But rKEG ~ rABC (Step 1)

Thus rDEG ~ rABC

Q.E.D.

THEOREM 6 Remarks

1. Notice that in Step 4 we take it as obvious that if a first triangle is similar to a

second one, then any triangle congruent to the first one is also similar to the second one.
It is also evident that if two triangles are similar to a third, then they are similar to each
other, since if they are both equiangular with the same triangle, they must be equiangular
with each other, and thus they must be similar to each other.

183
2. Here is a related Theorem: If two triangles B
have one angle in one equal to one angle in the other, 1
and another pair of corresponding angles both acute,
and a pair of sides in one proportional to a A
corresponding pair in the other, then the triangles are 2
similar. C

Given: rABC and rGBH sharing ∠1, with G 4

P
∠2 and ∠3 both acute, and 5

BA : AC = BG : GH 3
H
Prove: rABC is similar to rGBH

If possible, assume GH is not parallel to AC. Draw GP parallel to GH.

Thus rABC and rGBP are equiangular, and hence similar. Thus
BA : AC = BG : GP (rABC and rGBP are similar)
but BA : AC = BG : GH (given)
so BG : GP = BG : GH
so GP = GH (each having the same ratio to BG)
thus ∠3 = ∠5 (rGPH being isosceles)
now ∠4 = ∠2 (since rABC and rGBP are similar)
and ∠2 is acute (given)
thus ∠ 4 is acute
hence ∠5 is obtuse (being the supplement of ∠4)
but ∠3 = ∠5, and therefore rGPH has two obtuse angles, which is impossible.
Therefore our initial assumption is impossible, and it is necessary that GH be parallel to
AC. But then it follows that rABC is similar to rGBH. Q.E.D.

THEOREM 7: The perpendicular dropped from the right angle to the hypotenuse
in a right triangle divides it into two right triangles similar to each other and to the
whole.

B
Let ABC be a right triangle, AC its hypotenuse,
and drop BP perpendicular to AC.
Then rAPB ~ rBPC ~ rABC.
Is that just my opinion? No, it's a fact. Here's 1 2
why: A
P
C

[1] ∠APB = ∠ABC (they are both right angles)

and ∠1 is common to rAPB and rABC
so rAPB and rABC are equiangular
so rAPB ~ rABC (Thm.4)

184
[2] Now ∠BPC = ∠ABC (they are both right angles)
and ∠2 is common to rBPC and rABC
so rBPC and rABC are equiangular (Thm.4)
so rBPC ~ rABC

[3] So rAPB and rBPC are equiangular,

each being equiangular with rABC (Steps 1 and 2).
Thus rAPB ~ rBPC (Thm.4)

Q.E.D.

THEOREM 7 Remarks

Prove in particular that ∠ABP = ∠2, and that ∠CBP = ∠1.

THEOREM 8: How to cut a given straight line similarly to a given cut straight
line; also, how to cut off any fraction of a straight line.
B

Suppose you have a straight line AB, and also another straight C
line DE cut at some point K. How can you cut AB at a point C
so that AC : CB = DK : KE ?
Like this:
A H
G
[1] Draw AH at any angle to AB, and make AH = DE.

[2] Cut off AG = DK. D K E

[3] Join HB.

Draw GC‖HB.
Then C is the point we are looking for!

Proof:

[4] Since GC‖HB in triangle ABH,

thus AC : CB = AG : GH (Thm.2)
but AG : GH = DK : KE (since AG = DK and GH = KE)
so AC : CB = DK : KE

Q.E.F.

185
 L M N P
X X X
Suppose now you want to cut off some particular
fraction of AB, say two thirds of AB. How do
you do it? Lay out any straight line X, and lay it out three times in a straight line. Thus
LN is two thirds of LP.
Using the construction above, cut AB at a point C so that
AC : AB = LN : LP.
Since 2 : 3 = LN : NP
thus AC : CB = 2 : 3
i.e. AC is two thirds of AB, and so we have cut AB in the required fraction.
Similarly we can cut off any other fraction of straight line AB.

Q.E.F.

THEOREM 9: How to find a fourth proportional to three straight lines.

Suppose you have three lines A, B, C.

How can we find a fourth line D such that A : B = C : D ?
Like this:

[1] Place A and B in a straight line such as PEG, in which

PE = A
EG = B
A
[2] Draw PK = C at any angle to PEG. G
B
(B)
[3] Join EK. C E
Draw GH‖EK, cutting PK (extended) at
H. (A)

Then KH is the line D we are looking for;

i.e. A : B = C : KH. P
(C) K (D)
H

If you are skeptical, here's the irrefutable proof:

[4] Since EK‖GH in triangle PGH,

thus PE : EG = PK : KH (Thm.2)
but PE = A, EG = B, PK = C (Steps 1 and 2)
so A : B = C : KH

Q.E.F.

186
THEOREM 9 Remarks

We can now also find a third proportional to two given straight lines. For example, if we
are given A and B, we can find a third line D such that A : B = B : D. All we have to
do is use the same construction we just used, letting C = B, or PK = EG. For by the
construction
A : B = C : D.
But since C = B, then we have
A : B = B : D.

THEOREM 10: How to find the mean proportional between two straight lines.

If you have two straight lines A and B, how do

you find a straight line M such that (M)
A:M = M:B ?
(A) (B)
Like this:
D E F

[1] Place A and B in a straight line such as DEF, in which DE = A and EF = B.

[2] Draw a semicircle on DF as diameter, and draw EP perpendicular to DF, cutting

the circumference at P. Call EP “M”.

[3] Join DP. Join PF.

[4] Now, ∠DPF is a right angle (since it is in a semicircle).

And PE is perpendicular to DF (Step 2)
So rDEP ~ rPEF (Thm.7)
Thus the sides about the equal angles in these triangles are proportional
when taken in corresponding order.

[5] But ∠DEP and ∠PEF are equal angles (both being right) in these two
triangles. Therefore the sides about them are proportional, i.e.
DE : EP = EP : EF.

[6] But DE = A, EP = M, EF = B (Steps 1 and 2)

So A:M = M:B (substituting these terms in Step 5)

Q.E.F.

187
THEOREM 10 Remarks
P

1. How do we know in Step 5 that we have

taken the sides about the equal angles in
corresponding order? As follows. If we complete D F
E
the circle and extend PE down to G, clearly PE =
EG, since PEG is perpendicular to a diameter.
Thus rPEF ≅ rGEF (Side Angle Side)
G
so PF = GF
and so the arcs cut off by these chords are also equal,
i.e. arc PF = arc GF.
Therefore the angles standing on these arcs from the circumference are also equal,
i.e. ∠PDF = ∠GFP,
or ∠PDE = ∠EPF.
Thus these are corresponding angles in rDEP and rPEF, and so the opposite
sides PE and EF are corresponding sides. We already know hypotenuses DP and
PF correspond. Hence DE and EP, the remainding sides, correspond.

2. We have a nice corollary with this Theorem: The perpendicular dropped

from the circumference to the diameter of a circle is a mean proportional between
the segments into which it cuts the diameter. Likewise the perpendicular dropped
from the right angle of a right triangle to its hypotenuse is a mean proportional
between the segments into which it cuts the hypotenuse.

3. We already know that the square on EP is equal to the rectangle contained

by DE · EF (from Ch.2). Now we see that the side of a square equal to any
rectangle is a mean proportional between the two sides of that rectangle.

188
THEOREM 11: Equal and equiangular parallelograms have
reciprocally proportional sides; conversely, equiangular parallelograms with
reciprocally proportional sides are equal.
L C K

Let CBDK and EBAH be equiangular

Y Z
parallelograms (i.e. ∠EBA = ∠CBD, ∠BEH =
∠BCK, etc.). And let them also have equal E
B D
areas. I say that their sides are reciprocally
proportional, that is AB : BC = DB : BE. X
The proof is very simple:

H A
[1] Place EB and BD in a straight line.
Thus ∠EBC + ∠CBD = two rights
But ∠EBA = ∠CBD (given)
so ∠EBC + ∠EBA = two rights
and thus AB and BC are also in a straight line.

[2] Extend HE and KC till they meet at L, completing parallelogram ELCB, having
the same angles as the two other parallelograms.

[3] Looking at the areas of the parallelograms,

Z : Y = DB : BE (Thm.1)
But Z = X (given)
so X : Y = DB : BE
But X : Y = AB : BC (Thm.1)
So AB : BC = DB : BE

Q.E.D.

Conversely, let CBDK and EBAH be equiangular parallelograms in which AB : BC =

DB : BE. Then I say that EBAH = CBDK, i.e. they have the same area.

[1] Let them be placed as before.

[2] Now AB : BC = DB : BE (given)

but AB : BC = X : Y (Thm.1)
so X : Y = DB : BE
but Z : Y = DB : BE (Thm.1)
so X:Y = Z:Y
so X = Z
i.e. EBAH = CBDK

Q.E.D.

189
THEOREM 11 Remarks: L C K

1. Complete the parallelogram in ∠HLK by E

B D
extending HA and KD to G. Look familiar? The
diagonal LG passes through B and parallelograms
EBAH and CBDK are none other than the equal
“complements” about the diagonal. To prove it is easy
enough. H A G

Since AB : BC = DB : BE,
thus AB : BD = CB : BE (alternating the proportion)
so BA : AG = LE : EB (since BD = AG and CB = LE)
but ∠BAG = ∠LEB (by the parallels)
hence rBAG is similar to rLEB (Thm.6)
so ∠ABG = ∠ELB
and ∠EBA = ∠BEL (by the parallels)
so ∠ABG + ∠EBA = ∠ELB + ∠BEL (adding equals to equals)
Now if we add ∠EBL to both sides, we get
∠ABG + ∠EBA + ∠EBL = ∠ELB + ∠BEL + ∠EBL
But the right side of this equation is all the angles in rEBL, and so they add up to 180°.
So ∠ABG + ∠EBA + ∠EBL = 180°
from which it follows that LBG is a straight line.
You should be able to prove now that parallelograms EBAH and CBDK are
similar to each other and to HLKG.

2. This Theorem is about equal and equiangular parallelograms.

Obviously it is possible for parallelograms to be equiangular without A B

having equal areas, as A and B.

C D
Or again, they can have equal areas without being equiangular,
as C and D.

Or again, parallelograms can be equiangular without having R

proportional (or reciprocally proportional) sides, as a square S and a S

rectangle R that has a different area.

Also, nothing prevents parallelograms from having

A
B reciprocally proportional sides without being equiangular,
D C as in the case of a rectangle of sides A and D and a tilted-
over parallelogram in sides B and C, where A : B = C : D.

190
3. It is also true, conversely, that if two
parallelograms are equal in area and have G D F

reciprocally proportional sides, then they are

equiangular. Suppose ABNK and CBDF are equal A
B C
in area, but not equiangular, so that ABC is one
straight line, but NBD is not. Extend DB to E, and
complete parallelograms ABDG and ABEH. H
Clearly ABEH = ABNK K E N

so also ABEH = CBDF.

And since these last two are also equiangular, therefore, by the Theorem above,
AB : BC = DB : BE
Now since CBDF is equiangular with ABEH, but not with ABNK, thus ABEH and
ABNK are not equiangular.
So ∠AHE ≠ ∠BNK ( ≠ means “is not equal to”)
or ∠BEN ≠ ∠BNE.
Hence BE ≠ BN in rBEN.
Now since AB : BC = DB : BE
It follows that AB : BC ≠ DB : BN, since BE ≠ BN.
That is, the sides of ABNK and CBDF are not reciprocally proportional. Therefore,
equal but not equiangular parallelograms do not have reciprocally proportional sides, and
so equal parallelograms that do have reciprocally proportional sides must also be
equiangular. Q.E.D.

THEOREM 12: If four straight lines are proportional, then the rectangle
contained by the means is equal to the rectangle contained by the extremes.
Conversely, if two rectangles are equal, then their sides are reciprocally
proportional.

Let A, B, C, D be proportional straight lines, i.e. A : B = C : D.

I say that A · D = B · C, i.e. that these rectangles are equal in area.
A B C D
[1] Make rectangle A · D and rectangle B · C.
A
[2] Since both are rectangles, they are equiangular parallelograms.
D

[3] And also their sides are reciprocally proportional, since we are
given that B
A : B = C : D.
C

[4] Therefore they have equal areas (Thm.11), that is A · D = B · C. Q.E.D.

191
Conversely, let A · D = B · C, i.e. suppose these are any two rectangles having equal
areas. Then I say that A : B = C : D.

[1] Since both are rectangles, they are equiangular parallelograms.

[2] Also, they are equal (we are given that).
[3] Therefore their sides are reciprocally proportional (Thm.11), i.e.
A : B = C : D.

Q.E.D.

THEOREM 12 Remarks

If it happens that B = C, then B · C is actually a square. And since

A : B = C : D,
we can also say (since B = C in this case) that
A : B = B : D.
So, when a square is equal to a rectangle in area, its side is a mean proportional between
the sides of the rectangle. And, conversely, when three straight lines are in continuous
proportion, the square on the mean is equal in area to the rectangle contained by the
extreme terms in the proportion.

THEOREM 13: If two lines inside a circle cut each other, then their segments
contain equal rectangles.
A
D
L
X Let AB and CD cut each other inside a circle at X.
R I say that AX · XB = CX · XD, i.e. these
rectangles have the same area.
C
B

[1] Join AD. Join CB.

[2] Now, ∠CXB = ∠AXD (these are vertical angles)

and ∠XCB = ∠XAD (both stand on arc BRD from the circumference)
so rCXB and rAXD are equiangular

[3] Thus rCXB ~ rAXD (Thm.4)

192
 So the sides about their equal angles are proportional (when taken in
corresponding order).

[4] Since ∠XCB = ∠XAD, therefore the sides opposite these angles are
corresponding sides, i.e. XB and XD are corresponding sides.
Likewise ∠CBX = ∠ADX, therefore the sides opposite these angles are
corresponding sides, i.e. CX and AX are corresponding sides.

[5] Thus CX : XB = AX : XD, these being corresponding sides about the equal
angles CXB and AXD (Steps 3 and 4).

[6] So AX · XB = CX · XD (Thm.12).

Q.E.D.

THEOREM 13 Remarks

If AB and CD meet outside the circle, i.e. if X is outside the

circle, is the theorem still true? Is it still true that the
rectangles AX · XB and CX · XD have the same area?
C
Start by trying to prove that the triangles CXB and AXD are
B still similar.
D

193
THEOREM 14: How to make, on a given straight line, a rectilineal figure
similar and similarly situated to a given rectilineal figure.

C
Take any rectilineal figure ABCD, and L
any straight line EH. How can we make D K
a figure on EH similar to ABCD, in
which EH is the side corresponding to E H
AB? As follows. A B

[1] Divide ABCD into triangles (in this case two) ABD and DBC.

[2] Draw ∠HEK = ∠BAD.

Draw ∠EHK = ∠ABD.
So rEHK is equiangular with rABD.
So rEHK is similar to rABD (Thm.4)

[3] Draw ∠HKL = ∠BDC.

Draw ∠KHL = ∠DBC.
So rKHL is equiangular with rDBC.
So rKHL is similar to rDBC (Thm.4)

[4] Thus figure EHLK is equiangular with figure ABCD, each being composed of an
equal number of correspondingly arranged equiangular triangles.

[5] In order for EHLK and ABCD to be similar, though, the sides about their equal
angles must be proportional. Is that the case?
Obviously DA : AB = KE : EH (rEHK ~ rABD; Step 2)
And BC : CD = HL : LK (rKHL ~ rDBC; Step 3)

But can we say AB : BC = EH : HL ?

[6] Yes, since AB : BD = EH : HK (rEHK ~ rABD; Step 2)

And also BD : BC = HK : HL (rKHL ~ rDBC; Step 3)
Thus AB : BC = EH : HL (Ex Aequali, Ch.5, Thm.17)

Likewise CD : DA = LK : KE.

[7] Since EHLK and ABCD are equiangular (Step 4), and since the sides about their
equal angles are proportional (Steps 5 and 6), therefore they are similar to each other, and
AB corresponds to EH.

Q.E.F.

194
THEOREM 14 Remarks

Since we made a rectilineal figure similar to a given one by putting together triangles that
are similar to those into which the given figure was divided (and by arranging them
similarly), it is an obvious corollary to this Theorem that Any two similar rectilineal
figures can be divided into an equal number of similar triangles, similarly arranged.

THEOREM 15: If a third proportional is found to the sides of two squares, then
the first square is to the second square as the side of the first square is to the third
proportional.

Take any square ABDE, and place any other square G C

KBCG on it so that two sides are in line, as DB and
BC. Join AC, and draw CT at right angles to AC,
meeting AB (extended) at T. Complete the rectangle A T
DBTF. K B

Now, since ACT is a right triangle, and CB is

dropped from the right angle to the hypotenuse,
therefore AB : BC = BC : BT (Thm.10, Remark 2).
Thus BT is a third proportional to the sides of the two
E D F
squares. I say that
£AB : £BC = AB : BT.

[1] For, the square ABDE and the rectangle DBTF are parallelograms under the same
height, and so they are as their bases (Thm.1),
i.e. £AB : DB · BT = AB : BT
or £AB : AB · BT = AB : BT (AB = DB)

[2] But AB : BC = BC : BT (by construction)

so AB · BT = £BC (Thm.12 Remarks)

[3] Thus, if substituting £BC for AB · BT in Step 1, we have

£AB : £BC = AB : BT
And so the first square is to the second square as the side of the first square is to the third
proportional line.

Q.E.D.

195
THEOREM 15 Remarks

1. Here is a numerical example:

Square A has a side that is 2 feet long.
Square B has a side that is 4 feet long.
According to the Theorem, Square A is not to Square B as 2 to 4, but rather if we find X
such that 2 : 4 = 4 : X, then Square A is to Square B as 2 is to X. Of course, in this
case X = 8. So the square A is to square B as 2 is to 8, or more simply as 1 is to 4.

2. Unequal squares are never in the same ratio as their G E

H
sides. Suppose square ABGH is greater than square
BCDK. Line up sides AB and BC. If we extend HG and K D
CD till they meet at E, BCEG will be a rectangle of the
same height as the square on AB.
Hence £AB : BCEG = AB : BC
A B C
Accordingly, since £BC is less than BCEG,
£AB : £BC > AB : BC.
So the greater square always has to the lesser a ratio greater than the corresponding ratio
between the sides.

3. Looking back to Definition 4, we can also now say that Squares are to each other
in the duplicate ratio of their sides.

THEOREM 16: Similar triangles have the same ratio as the squares on their
corresponding sides. G B L
N E Q
Take any two similar triangles, ABC and
DEF, and build squares ACKH and DFPO
on corresponding sides AC and DF. I say A
R
C D
T
F

rABC : rDEF = £AC : £DF.

O P
[1] Drop BR perpendicular to AC. Drop
ET perpendicular to DF. H K
Since ∠BAR = ∠EDT
(since rBAC is similar to rEDF)
thus rBAR is equiangular with rEDT,
so BR : RA = ET : ED
but BA : AC = ED : DF (since rBAC is similar to rEDF)
hence BR : AC = ET : DF (ex aequali, Ch.5, Thm.17)

196
[2] Complete the rectangle of height BR, base AC, namely ACLG.
Complete the rectangle of height ET, base DF, namely DFQN.
Now ACLG : £AG = LC : CK (Thm.1)

Or ACLG : £AC = BR : AC (LC = BR, CK = AC)

and DFQN : £DF = ET : DF (for similar reasons)
but BR : AC = ET : DF (Step 1)
so ACLG : £AC = DFQN : £DF

[3] So ACLG : DFQN = £AC : £DF (alternating)

but ACLG : DFQN = rABC : rDEF (wholes are as their halves)
so rABC : rDEF = £AC : £DF

Q.E.D.

D
THEOREM 17: Similar rectilineal L
figures have the same ratio as the squares on their E R
M X
corresponding sides. S C Y K
G Z
A
Take any two similar rectilineal figures ABCDE T
H
and GHKLM.
B
I say that ABCDE : GHKLM = £AB : £GH.

[1] By the corollary to Theorem 14, we know we can divide ABCDE and GHKLM
into similarly arranged similar triangles, equal in number. For brevity, let the triangles in
ABCDE be called R, S, T, and let those in GHKLM be called X, Y, Z.

[2] Since S and Y are similar triangles,

thus S : Y = £AC : £GK (Thm.16)
But T : Z = £AC : £GK (Thm.16)
So T:Z = S:Y
But T : Z = £AB : £GH (Thm.16)
So S : Y = £AB : £GH
And likewise we can prove that every pair of corresponding triangles in the two figures
has the ratio of £AB : £GH.

[3] That is to say,

T : Z = £AB : £GH
and S : Y = £AB : £GH
and R : X = £AB : £GH
Thus R + S + T : X + Y + Z = £AB : £GH (Ch.5, Thm.15)
i.e. ABCDE : GHKLM = £AB : £GH

Q.E.D.

197
THEOREM 17 Remarks:

Looking back at Theorem 15, Remark 2, we can now see that similar rectilineal figures
are to each other in the duplicate ratio of their corresponding sides.

THEOREM 18: Figures similar to the same rectilineal figure are similar to each
other.

A
Suppose A is similar to X, and B is also similar to X.
X Then A is similar to B. Why? Because …
B

[1] The angles in A equal the angles in X (Since A is similar to X)

The angles in B equal the angles in X (Since B is similar to X)
So the angles in A equal the angles in B (since things equal to the same thing are
themselves equal).
So A and B are equiangular.

[2] The sides about the angles in A have the same ratios as the sides about the equal
angles in X.
The sides about the angles in B also have the same ratios as the sides about the
equal angles in X
So the sides about the angles in A have the same ratios as the sides about the
equal angles in B (since two ratios that are both the same with another ratio are the same
as each other).

[3] So A and B are equiangular and the sides about the equal angles are proportional.
Therefore A is similar to B (Def. 1).

Q.E.D.

198
THEOREM 19: If four straight lines are proportional, the similar figures
similarly situated on them will be proportional.

W
Y

A D

X
Z
B E

C F

Suppose you have 4 straight lines A, B, D, E and A : B = D : E.

Then place any similar figures W and X on A and B, similarly situated. Place any two
other similar figures Y and Z on D and E, similarly situated.
I say that W : X = Y : Z.

[1] Find a third proportional line to the straight lines A and B, namely C, so that
A:B = B:C (Thm.9)
Find a third proportional line to the straight lines D and E, namely F, so that
D:E = E:F (Thm.9)

[2] So A : C = £A : £B (Step 1 and Thm.15)

And D : F = £D : £E (Step 1 and Thm.15)

[3] But the similar figures are also in the ratios of these squares on their
corresponding sides (Thm.15),
so W:X = A:C
and Y:Z = D:F

[4] Now A:B = D:E (given)

and B:C = E:F (A : B = B : C, and D : E = E : F, Step 1)
so A:C = D:F (ex aequali, Ch.5, Thm.17)

[5] But each of the ratios in this last proportion is the same as one of the ratios of
similar figures (see Step 3).
So W:X = Y:Z

Q.E.D.

199
THEOREM 20: If two parallelograms share an angle and are about the same
diagonal, then they will be similar.

A K D
Take any parallelogram ABCD, and choose any
point P along its diagonal, completing
parallelogram AEPK. I say that AEPK is similar
E P to ABCD.

B C

[1] Now ∠AEP = ∠ABC (since EP║BC)

[2] And ∠KAE is common to both parallelograms.

[3] Since the opposite angles in any parallelogram are equal, it follows that the
remaining angles in AEPK and ABCD are correspondingly equal. So they are
equiangular.

[4] Now since ∠AEP = ∠ABC (Step 1), and ∠EAP is common to rEAP and
rBAC, hence these triangles are equiangular and therefore similar.
So AE : EP = AB : BC

[5] Since the opposite sides in any parallelogram are equal, it follows that the
remaining sides in AEPK and ABCD about their equal angles (and taken in
corresponding order) are proportional.
Thus AEPK is similar to ABCD. So all parallelograms about the diagonal and
sharing an angle with the whole parallelogram ABCD will be similar to the whole and
therefore to each other.

Q.E.D.

THEOREM 20 Remarks:

Conversely, if two similar parallelograms share an angle they will also share a diagonal.
Using the same diagram as in the Theorem, suppose that AEPK and ABCD are similar
parallelograms sharing the angle at A. I say that P lies on diagonal AC. From the
similarlity of the parallelograms, AE : EP = AB : BC. Since EP is parallel to BC, it
must cut AC at some point, say Q. And since EPQ is parallel to BC in triangle ABC,
therefore AE : EQ = AB : BC. From the two proportions, it is clear that EQ = EP, and
therefore Q and P are the same point. So P lies on AC.

200
THEOREM 21: How to cut a line in mean and extreme ratio.

G
Take any straight line AB. F
How can we cut it at a point K so that BA : AK = AK : KB ?
Thus: K B
A

[1] Draw square ABCD. M

[2] Bisect AD at M and join MB. D C

R

[3] Draw a circle with center M and radius MB (thus passing

through C, and cutting AD extended at F and L). L

[4] Draw square AFGK. Extend GK to R.

K is the point we sought. How do we know? As follows:

[5] Now £AB = FA · AL (AB is ⊥ to diameter FAL; Ch.2, Thm.6)

But FA = GF (being sides of £AFGK)
and AL = FD (each is a radius plus half of AD)
so £AB = GF · FD

[6] Or ABCD = FGRD (renaming them)

Now subtract AKRD from each and we have
KBCR = FGKA
or CB · BK = £AK (renaming them)

[7] Thus CB : AK = AK : KB (Thm.12)

or BA : AK = AK : KB (CB = BA)

Q.E.F.

THEOREM 21 Remarks:

1. Do you recognize the diagram? Compare it to Ch.2, Theorem 12. There we

extended a straight line AS to B so that £AS = AB · BS. By the present Theorem, we
see that ASB was cut in mean and extreme ratio ( BA : AS = AS : SB).

201
 A
2. We used Ch.2, Theorem 12 to make a “Golden Triangle” in
Ch.4, Theorem 6. This was the isosceles triangle TAP, whose base
angles were each double its peak angle. Recall that in the course
of making it, we made line ASP such that £AS = AP · PS ;
S
thus PA : AS = AS : SP, so that AP was divided at S in mean
and extreme ratio or the “Golden” ratio. But we had also made PT
= AS, so that AP : PT was also the Golden ratio. For this reason, T P
rTAP is called a “Golden Triangle.”

3. We used a Golden Triangle to construct a regular pentagon in Ch.4, Theorem 7.

So the Golden ratio is the key to producing that figure as well.

A P K B

4. Can we cut AB at another point besides K, say P, such that BA : AP = AP : PB?

No. It is impossible. K is unique. If possible, suppose P is on AK,
and BA : AP = AP : PB.
Thus AB · BP = £AP (Thm.12)
Now AB · BP > AB · BK (since BP > BK)
so £AP > AB · BK
but AB · BK = £AK (since AB : AK = AK : KB)
so £AP > £AK
Thus AP > AK, which is impossible.
It is true, of course, that if we cut off AP = BK, then AB : BP = BP : PA,
because P is simply the mirror image of K on the other side of AB’s midpoint. Thus it
does not divide AB in an essentially new way.

5. It should be clear, too, that any two lines cut in mean and extreme ratio are cut in
the same ratio. To cut any other line besides AB in mean and extreme ratio, we would
use the same construction again, and the similarity of all the figures in both constructions
would make it obvious that the lines are cut similarly.

F G T
6. Looking back at the diagram for this Theorem, complete
the rectangle FDCT. Notice that square FK and rectangle KC are
equal (Step 6), and they are parallelograms drawn in opposite
K
A B corners of FDCT and they are both equiangular with rectangle
FDCT. They also have a common corner K. Therefore K lies on
the diagonal of rectangle FDCT (Ch.1, Thm.34 Questions).
Accordingly rectangles AKRD and GTBK are similar to the whole
D R C rectangle FDCT and to each other (Thm.20).

202
7. Since AK : KB is the golden ratio,
and GK : KB = AK : KB,
hence GK · KB is called a “Golden Rectangle.”
Since, by the preceding remark, GK · KB is similar to AD · DR and FD · DC, these
are also golden rectangles.

8. Note that if we draw a square inside a Golden Rectangle on its lesser side, the
remaining rectangle is another Golden Rectangle. e.g. FDCT is a Golden Rectangle, and
when we take away square ABCD, what remains is BA · AF.
But BA : AF = BA : AK (since AF = AK),
and BA : AK is the Golden ratio.
Hence BA : AF is also the Golden ratio, and thus BA · AF is a Golden Rectangle. And
when we subtract square FGKA from it, we are left with Golden Rectangle GK · KB.

THEOREM 22: In right triangles the rectilineal

figure on the hypotenuse is equal to the sum of the similar A

(and similarly situated) figures on the other two sides. X

Y

Imagine a right triangle ABC, whose hypotenuse is BC. B C

Imagine further some rectilineal figure Z on hypotenuse
BC. If X and Y are similar to Z, and similarly situated on Z

AB and AC, then X + Y = Z.

Here's the proof:

[1] First, £AB : £AC = X : Y, since similar figures are as the squares on their
corresponding sides (Thm.17).

[2] Thus £AB : X = £AC : Y (alternating the proportion)

so £AB + £AC : X + Y = £AC : Y (Ch.5, Thm.15)

[3] Now £BC : £AC = Z : Y, since similar figures are as the squares on their
corresponding sides (Thm.17).
Thus £BC : Z = £AC : Y (alternating the proportion)

[4] So £BC : Z = £AC : Y (Step 3)

and £AB + £AC : X + Y = £AC : Y (Step 2)
thus £AB + £AC : X + Y = £BC : Z

[5] So £AB + £AC : £BC = X + Y : Z (alternating Step 4)

but £AB + £AC = £BC (Pythagorean Theorem)
thus X + Y = Z.

Q.E.D.

203
THEOREM 22 Remarks:

1. The diagram depicts rectangles, but they could just as well be irregular (but
similar) pentagons, or any other kind of rectilineal figure.

2. Do you think that the Theorem would hold for similar curvilineal figures, such as
semicircles?

3. This Theorem is in some ways a generalized version of the Pythagorean Theorem.

The Pythagorean Theorem proved a relationship between squares on the sides of a right
triangle, whereas this Theorem proves that the same relationship exists between any
similar rectilineal figures similarly situated on the sides of a right triangle.

“HOOK”: CEVA’S THEOREM.

If in any triangle XYZ lines drawn from the three vertices to the opposite sides (XR, YS,
ZQ) are concurrent at P, cutting the sides XY, YZ, ZX into a & b, c & d, e & f
(labeling all these clockwise from X),

then a c e
⋅ ⋅ = 1
b d f

and conversely.

c
b
R
Q
P
d
a

X Z
f S e

204
“HOOK”: THE EQUILATERAL TRIANGLE AND THE GOLDEN RATIO.

If the midpoints of two sides of an equilateral triangle inscribed in a circle be joined and
extended to meet the circle, this line is cut in mean and extreme ratio.

A B C

205
Chapter Seven

Numbers
 DEFINITIONS

1. When the same kind of thing is taken repeatedly, the result is a MULTITUDE.
For example, if you take a chair, and another chair, you have a multitude of
chairs. Again, if you have a pear, a banana, and an orange, although you do not have a
multitude of pears, you do have a multitude of fruits.

2. The UNIT is what is indivisible, constituting a multitude by repetition of itself.

For example, if you have a multitude of chairs, the unit constituting this multitude
is one chair. Suppose you have 24 chairs: then the repetition of a dozen chairs also
constitutes it, but a dozen is not the unit of this multitude, since the dozen is itself a
multitude constituted by repetition of something – a dozen is still itself a multitude of
chairs. The unit, on the other hand, is the elementary beginning of multitude; it is not
itself a multitude, and hence it is indivisible. The unit is like the atom of a multitude.
If we consider concrete multitudes such as our 24 chairs, we see that being a unit
can sometimes happen to a divisible thing: one chair can be sawn in half. But does this
belong to the chair because it is a unit in a multitude, or because it is a chair? If a chair
were divisible precisely because it is a unit, then every unit in every multitude would be
divisible. And that is false. What is the unit in a multitude of geometric points? One
geometric point. Is that divisible? No. Consequently, it does not belong to a unit as
such to be divisible, but rather one apple is divisible because it is an apple, and one line
is divisible because it is a line, and so on. So no unit is as such divisible.
Mathematics ignores what the quantities it studies are made of. For example, it
studies circles without considering what they might be made of, and so the circle studied
by the geometer is neither heavy nor light, neither hot nor cold, neither hard nor soft.
Circularity as such does not have such properties, although it might be found in things
having such properties. Similarly, mathematics studies multitudes without considering
what they might be multitudes of, and so the units it studies cannot be divisible: unity
itself does not have divisibility, although it might belong to something that is divisible.
Multitude might also be found in things that have position, orientation, and
location, such as apples. But differences in position, orientation, and location make no
difference with respect to multitude; 10 apples remain 10 apples regardless of changes in
their relative position or arrangement. Hence the mathematical unit in itself has no
position, location, or orientation.

206
 Nothing prevents us from choosing something divisible or something having
position or orientation as a kind of conventional unit to measure other things of like kind,
such as a unit length or a unit weight or a unit time. For example, we might lay out
Y
straight lines X and Y at right angles to each other at point
O, to use as references for designating any point in the plane.
Choose any point on X to the right of O and label it 1. Thus
the line length between O and 1 (call it “U”) is our chosen
-1 O 1 unit length. To distinguish the same length in the opposite
X
direction, we call it – 1 (“negative one”). Notice that U is
U
something divisible, having position, and also direction (it is
to the right of O). All this proves to be very useful elsewhere
in mathematics, of course, but note that U is not a pure unit.
It is not “the unit,” the “one” by which we count all things,
but a unit length, having an arbitrary length and orientation (none of which belong to
“one” as such). It is no more identical with “one” than “two apples” is identical with
“two.”

3. A NUMBER is a multitude measured by the unit.

Some multitudes might be infinite, but a number is a measured multitude, and
hence finite. It is a multitude that can be comprehended by counting.
Such things as “three and a half”, and other things such as “negative six”, and
still other things such as “the square root of two”, have come to be called “real
numbers” with reference to a conventional unit that is divisible and has position (such as
U in the discussion of the last definition). And yet – 6 and + 6 do not differ in
multitude, but only in direction (or some other extraneous thing). And 3 ½ divides its
own unit, which must therefore be a divisible thing, and not simply the unit considered
entirely apart from certain divisible things out of which a multitude might be made.
Prior to all such compound notions, applying the concept of number to various things
and combining it with such things as direction and length, there is the pure concept of
number as a kind of multitude. 6 is simply and purely a kind of multitude, but – 6 or +
6 is a multitude plus an additional idea of direction (or something similar). 3 is simply
and purely a multitude, but 3 ½ is a multitude of divisible things. This chapter is about
pure numbers.
Is 1 a number? If you ask me to hand you “a number of nails”, and I hand you
one nail, you might not be satisfied. One is not a multitude or a plurality, and therefore it
is not a number in the same sense that 5 or 6 are numbers. Still, since 1 is the beginning
and measure of all the numbers, and since it has a ratio to every number, it is also called
a “number.”

207
4. An EVEN NUMBER is a number divisible into two equal numbers.
For example, 2, 4, 6 are even numbers.

5. An ODD NUMBER is a number not divisible into two equal numbers.

For example, 3, 5, 7 are odd numbers.

6. A FACTOR of a number is any number that measures it, i.e. goes into it exactly some
number of times.
For example, the factors of 12 are 1, 2, 3, 4, 6, 12 (but not 5, 7, 8, 9, 10, 11). A
factor of a number is also called a “divisor” of it, since it divides it exactly, and it is also
called a “measure” of it since it fits into it exactly.

7. A PRIME NUMBER is a number with no factors other than 1 and itself.

For example, 2 and 3 are prime numbers, since neither has any factors other than
the number 1 and itself.

8. Two different numbers are PRIME TO EACH OTHER if they have no factor in
common except 1.
For example, 8 and 15 are prime to each other, having no factor in common other
than 1. But 12 and 15 are not prime to each other, since they share a factor besides 1
(namely 3).

9. A COMPOSITE NUMBER is a number with factors other than 1.

For example, 14 is a composite number, since it has factors besides 1, namely 2
and 7. Obviously no prime number is composite, and no composite number is prime,
since their definitions are opposed.

10. TO MULTIPLY one number N by another number M means to find the sum of as
many N's as there are units in M. The resulting number is called the PRODUCT of M
times N.
For example, to multiply 3 by 5 means to add together as many threes as there
are units in 5, i.e. 3 + 3 + 3 + 3 + 3 = 15. So 15 is the product of 5 times 3.
The symbolic notation for multiplying 5 times 3 is this: 5 × 3 = 15. We can also
write it this way: 5 · 3 = 15.

208
11. TO DIVIDE one number N by another number M means to count the number of
times M can be subtracted from N. The resulting number is called the QUOTIENT of N
divided by M.
For example, to divide 12 by 4 means to count the number of times 4 can be
subtracted from 12:
(1) 12 – 4 = 8
(2) 8–4 = 4
(3) 4–4 = 0
It can be subtracted 3 times. So 3 is the quotient of 12 divided by 4.
The symbolic notation for dividing 12 by 4 is this: 12 ÷ 4 = 3.
Obviously a greater number is not always exactly divisible by a lesser one, for
example 16 ÷ 3. 3 can be subracted from 16 five times, but one unit of 16 is left over.
Note that multiplication and division are opposite operations, and they undo each
other. If
(a) M×N = P
then by definition N fits into P exactly M times. But that means we can subtract N
from P exactly M times, i.e.
(b) P÷N = M
Now by equation (a) we know that P = M × N, so let us substitute M × N for P in
equation (b):
( M×N )÷N = M
which is to say that any number M multiplied by any number N, and then again divided
by N, leaves us with M once more. Or, by equation (b), M = P ÷ N, so let us substitute
P ÷ N for M in equation (a):
( P÷N )×N = P
so any number P divided by a number N, and then again multiplied by that number N,
leaves us with P once more (at least, so long as we suppose N goes into P some exact
number of times; it is also true more generally, but that goes beyond the point made
here).

12. A TRIANGULAR NUMBER is the sum of any number of consecutive numbers

beginning with 1; the last number added is called the BASE of the triangular number.
For example, 3 is the first triangular number, since 3 = 1 + 2, and 2 is its base. And 6 is
the second triangular number, since 6 = 1 + 2 + 3, and 3 is its base. Such numbers are
called “triangular” because the numbers which add up to them can be stacked on top of
each other, and the result is a triangular form.

209
13. A SQUARE NUMBER is the product of two equal numbers. The number which,
thus multiplied by itself, yields a square number, is the SQUARE ROOT of the square
number.
For example, 4 is a square number, since 4 = 2 × 2, and 2 is the square root of 4. And 9
is the next square number, since 9 = 3 × 3, and 3 is the square root of 9. The symbolic
notation 52 = 25 reads “five squared equals twenty five”, and the notation 5 = 25
reads “five equals the square root of twenty five”. Square numbers are called “square”
because their units can be arranged in a square pattern. Accordingly, the square root is
also called the “side” of the square, as in “3 is the side of the square number 9”.

14. A CUBE NUMBER is the product of three equal numbers. The number which, thus
multiplied by itself, yields a cube number, is the CUBE ROOT of the cube number.
For example, 8 is the first cube number, since 8 = 2 × 2 × 2, and 2 is the cube root of 8.
The symbolic notation 23 = 8 reads “two cubed equals eight”, and the notation 2 = 3 8
reads “two equals the cube root of eight”.

15. A POWER of a number is the product of that number times itself any number of
times.
For example, the fourth power of 5 is 5 × 5 × 5 × 5, and the notation 54 = 625
reads “five to the fourth power equals six hundred and twenty five”.

16. A FACTORIAL NUMBER is the product of any number of consecutive numbers

beginning with 1.
For example, 24 is a factorial number, since 24 = 1 × 2 × 3 × 4. The notation 4!
= 24 reads “four factorial equals twenty four”.

17. A PERFECT NUMBER is a number equal to the sum of all its factors less than itself.
For example, 6 is the first perfect number, since, other than 6 itself, all its factors
are 1, 2, 3, and it also happens that 6 = 1 + 2 + 3. With most numbers the sum of all
their factors (other than themselves) falls short or exceeds the number itself. For
example, the factors of 8 less than itself are 1, 2, 4 which add up to 7. The factors of 12
less than itself are 1, 2, 3, 4, 6 which add up to 16.

210
 PRINCIPLES OF NUMBER THEORY

1. No number is both even and odd.

2. If one is added to or subtracted from an even, the result is odd.

3. If one is added to or subtracted from an odd, the result is even.

A possible exception occurs if the odd number you start with is 1. Is 1 – 1 “even”?
Perhaps, but it is not an even “number” insofar as a “number” means “a multitude of
units.” No units at all, or zero, is not a multitude of units. Still, 1 + 1 is even.

4. Any two numbers can be added together, and the result is a number.
Any number can have numbers added to it without limit.

5. Any two numbers can be multiplied together, and the result is a number.
Any number can be multiplied without limit.

6. Any lesser number can be subtracted from a greater number, and the result is a
number.

7. Any number can be divided by any one of its factors, and the result is a number.

8. A number that measures two numbers also measures their sum.

For example, if 2 goes into 6, and it also goes into 10, then 2 also goes into 6 + 10.

9. A number that measures two unequal numbers also measures their difference.
For example, if 3 goes evenly into 15, and also into 9, then 3 also goes evenly into 15 – 9.

10. A number that measures a number also measures any of its products.
For example, if 5 goes into 35, then 5 also goes into 26 × 35, and also into 327 × 35, and
in general 5 must go into N × 35, no matter what number N is.

211
 THEOREMS

THEOREM 1: The unit has to any number the same ratio that any second
number has to the product of the two numbers.

Take any two numbers, say N and M. Then I say that the following proportion holds:
1:M = N: N × M
Start off with the ratio of 1 to M. Any equimultiples of these two have the same ratio that
they have to each other (Ch.5, Thm.13). So let's take each of them N times, giving us N
× 1 and N × M. Thus we have
1 : M = N × 1 : N × M.
But N × 1 is taking N once, which is just N. So we have, in fact,
1 : M = N : N × M.

Q.E.D.

THEOREM 1 Remarks

1. Let's just confirm this with a few concrete examples, for the sake of clarity.
2 × 3 = 6, and lo and behold, it is also true that 1 : 2 = 3 : 6.
Also, 4 × 5 = 20, but also it is true that 1 : 4 = 5 : 20.

2. Can you see how this Theorem is actually just a different way of stating
Definition 10?

3. We can formulate a similar Theorem for the division of numbers, namely that The
unit has to any number the same ratio which its quotient with a second number has to the
second number. That is, for any two numbes M and N, it will be true that
1 : M = N ÷ M : N.
Start by noting that M × ( N ÷ M ) = N (see Def.11). Thus, by Theorem 1, the above
proportion follows. Here is an example:
12 ÷ 4 = 3, and it is also true that 1 : 4 = 3 : 12.

212
THEOREM 2: The order of multiplication makes no difference.

Given: Numbers A and B

Prove: A × B = B × A

This is no big surprise. But it is still worth noting that “three fives” equals “five threes”;
although the result is the same, the process of adding together three fives is not identical
to the process of adding together five threes. So why must the result be the same?

[1] Well, 1:A = B:B×A (Thm.1)

so 1:B = A:B×A (alternating)
but 1:B = A:A×B (Thm.1)
so A:B×A = A:A×B

[2] Since B × A and A × B have the same ratio to A, therefore they must be equal.

Q.E.D.

THEOREM 2 Remarks:
3
3
3
1. The truth of this Theorem can be manifested visually: 5 rows of 3 is 3
necessarily at the same time 3 columns of 5, i.e. 5 threes must equal 3 3
3
fives, since they are just two different ways of looking at the same 3
3
number. 3
3

2. The Theorem proves the case for two numbers, but it is just as true for any number of
numbers we multiply together. The final product is the same regardless of the order in
which we multiply them. For example, 3 × 4 × 5 = 5 × 4 × 3. For, by the Theorem
4×5 = 5×4 (each is equal to 20)
but again, by the Theorem
3 × 20 = 20 × 3
and so
3 × 4 × 5 = 5 × 4 × 3.

213
THEOREM 3: For any three numbers A, B, C, as long as B is exactly divisible
by C, then (A × B) ÷ C = A × (B ÷ C).

Whether we multiply A × B first, and then divide the product by C, or instead divide B
by C first, and then multiply the result by A, we get the same number. For example,
(4 × 6) ÷ 3 = 4 × (6 ÷ 3).
or 24 ÷ 3 = 4 × 2
Must it always work out, regardless of the numbers we choose? Naturally. And here’s
proof:

[1] Let (A × B) ÷ C = K
thus A×B = K×C (multiplying both sides by C)

[2] Let B÷C = N

thus B = C×N (multiplying both sides by C)

[3] Now A×B = K×C (Step 1)

or A × (C × N) = K × C ( B = C × N; Step 2)
so A×N = K (dividing both sides by C)
or A × N = (A × B) ÷ C (K = (A × B) ÷ C; Step 1)
hence A × (B ÷ C) = (A × B) ÷ C (N = (B ÷ C); Step 2)

Q.E.D.

THEOREM 3 Remarks:

1. Such a Theorem is not extremely surprising, but it is extremely useful to know

when we can change the order of operations without affecting the final result. Verify the
Theorem yourself with other numerical examples.

2. Here is another similar rule: suppose you have four numbers A, B, C, D and
A is exactly divisible by C and B is exactly divisible by D. Then I say that
(A × B) ÷ (C × D) = (A ÷ C) × (B ÷ D)
For example (6 × 4) ÷ (3 × 2) = (6 ÷ 3) × (4 ÷ 2).
Why must this always work out?

[1] Let A÷C = M

and B÷D = N

[2] Since multiplication is the reverse of division, hence it is also true that
M×C = A
and N×D = B

[3] So (A ÷ C) × (B ÷ D) = M × N (see Step 1)

214
[4] But (A × B) = [M × C] × [N × D] (see Step 2)
Hence, dividing both sides by (C × D), we have
(A × B) ÷ (C × D) = ([M × C] × [N × D]) ÷ (C × D)
and since the order of multiplication makes no difference, we can write
(A × B) ÷ (C × D) = (M × N) × (C × D) ÷ (C × D)
But since on the right we first multiply and then divide by (C × D), this leaves us
with the number (M × N), and so
(A × B) ÷ (C × D) = M × N

[5] Comparing Steps 3 and 4, we see that

(A ÷ C) × (B ÷ D) = (A × B) ÷ (C × D).

THEOREM 4: If four numbers are proportional, the product of the means equals
the product of the extremes. Conversely, if among four numbers the product of the means
equals the product of the extremes, then they are proportional.

Given: A : B = C : D
Prove: A × D = B × C

We know that A:B = C:D (given)

and so A:C = B:D (alternating)
Now if we multiply the first two terms by B, we maintain the proportion, and again if we
multiply the last two terms by A, we maintain the proportion (Ch.5, Thm.13).
Hence B×A : B×C = A×B : A×D
so B × A : A × B = B × C : A × D (alternating)
but B×A = A×B (Thm.2)
hence B×C = A×D

Q.E.D.

Now, conversely,

Given: A × D = B × C
Prove: A : B = C : D

We know that A×D = B×C (given)

But since it is also true that B × A = A × B, we can say that
B×A : A×B = B×C : A×D
so B × A : B × C = A × B : A × D (alternating)
But equimultiples have the same ratio has the terms they multiply (Ch.5, Thm.13),
and so A:C = B:D
or A:B = C:D (alternating)

Q.E.D.

215
THEOREM 5: Beginning with any two unequal numbers, the last number
produced by reciprocal subtraction is a common factor of them both.

Of course you are wondering what “reciprocal subtraction” means. It’s not hard. Take
unequal numbers, say 9 and 7, and take the lesser from the greater:
9–7 = 2
Now compare the subrtracted number to the difference, and take the lesser from the
greater:
7–2 = 5
Now compare the subtracted number to the difference, and take the lesser from the
greater:
5–2 = 3
You get the idea. This process cannot go on forever, since numbers are finite and the
number from which we are subtracting is smaller every time. So we must eventually end
up with nothing. But the only subtraction that leaves us with nothing is when something
is subtracted from itself. Hence the last subtraction in such a process must be

[1] X–X=0

Now the Theorem states that X is a factor of both original numbers, whatever they may
be. Why? Note that after the first step, the two numbers in the subtraction for any step
are (a) the number that was subtracted in the previous step, and (b) the difference in the
previous step. So, given our last step, the second to last step must be

[2] Z–X = X

and so Z = 2X; so X is a factor of Z. Again, the third to last step must be

[3] Q–X = Z

(or else Q – X = Z; it doesn’t matter which). Thus Q = X + Z, or Q = 3X (since Z

= 2X by Step 2). So X is now a factor of Q as well. Accordingly, X must be a factor of
every number in the whole process, right back to our original numbers A and B.

Q.E.D.

216
THEOREM 5 Remarks:

1. Here is a numerical example. Start with 27 and 12:

27 – 12 = 15
15 – 12 = 3
12 – 3 = 9
9–3=6
6–3=3
3–3=0

So the last number produced was 3, and indeed it is a factor of both 27 and 12. Notice,
too, that the last step is of the form X – X = 0, and the second to last step is of the form
Z – X = X, as asserted in the proof.

2. What happens if we choose numbers that are prime to each other, which have no
common factor except the unit? Then the process must end by producing the unit, since
we proved that the process ends with a common factor of the two original numbers. Try
an example. Start with 9 and 4, which are prime to each other:

9–4=5
5–4=1
4–1=3
3–1=2
2–1=1
1–1=0

Notice that 1 showed up before the process finished – the Theorem does not forbid that,
but demands that if 1 is the only common factor of the two original numbers, the last
number in the process must be 1.

217
THEOREM 6: Given any pair of numbers that are prime to each other, to find
multiples of them that differ by a unit.

Suppose you have a pair of numbers A and B which are prime to each other. Can you
multiply each of them so that the multiples differ by 1? Consider 12 and 5, which are
prime to each other. By Theorem 5, we know that their reciprocal subtraction must
produce the number 1, at least at the end of the process if not sooner. Let’s go through
the steps, then:

[1] 12 – 5 = 7
[2] 7–5=2
[3] 5–2=3
[4] 3–2=1

Since we hit 1 (as we must), let’s stop on this last Step [4]:

3 – 2 = 1

Now, by Step [3] we see that 3 = 5 – 2. So replace:

(5 – 2) – 2 = 1

And by Step [2] we see that 2 = 7 – 5. So replace:

[5 – (7 – 5) ] – (7 – 5) = 1
And by Step [1] we see that 7 = 12 – 5. So replace:

[5 – { (12 – 5) – 5 } ] – [ (12 – 5) – 5 ] = 1

Now the expression is exclusively in terms of the original numbers 12 and 5.

Simplifying, we have

(5 + 5 + 5 + 5 + 5) – (12 + 12) = 1

That is, we have a multiple of 5 and a multiple of 12 which differ by 1. Since this
process does not depend on 5 and 12 in particular, but presumes only that the original
pair of numbers are prime to each other, we have found a way to multiply any pair of
numbers that are prime to each other so that their multiples differ by 1.

Q.E.F.

218
THEOREM 6 Remarks:

1. Using the process described in the Theorem, find a pair of multiples of 16 and 7 that
differ by 1.

2. Sometimes, of course, a pair of numbers that are prime to each other themselves differ
by 1. In fact, any two consecutive numbers are prime to each other, such as N and N +
1. Why? Because any number that measures them both must also measure their
difference, namely 1. But the only thing that measures 1 is 1. Therefore the only thing
that measures both is 1, i.e. N and N + 1 are prime to each other.

3. The Theorem stipulates that the original pair of numbers must be prime to each other.
Must we use numbers that are prime to each other, or can we use any pair of numbers?
We must use numbers that are prime to each other. If instead we start with a pair of
numers that have a common factor greater than 1, then it will be impossible to find
multiples of them that differ by 1. Proof:
Take any two numbers, A and B, and any unequal multiples of them, NA and
MB. Now any number that measures both A and B must also measure both their
multiples NA and MB. And whatever measures both NA and MB must also measure
their difference. Hence, if A and B have a common factor greater than 1, then since this
common factor must measure the difference between any unequal multiples of A and B,
it follows that such a difference must be greater than 1. In general all unequal multiples
of any two numbers must differ by at least the greatest common factor of those two
numbers. For example, since 4 and 6 have a greatest common factor of 2, all multiples of
4 and 6 that differ must differ by 2 or more.

THEOREM 7: Two numbers proportional to two numbers that are prime to each
other are equimultiples of them.

Given: A is prime to B
A:B = C:D

Prove: A and B measure C and D the same number of times.

[1] Since A is prime to B, therefore find multiples of them that differ by 1 (Thm.6).
Say nA – mB = 1

[2] Now A:B = C:D (given)

so nA : mB = nC : mD (corresponding multiples)
Subtracting the consequent from the antecedent in each case, we will still have a
proportion (Ch.5, Thm.16, Remark 3).
So nA – mB : mB = nC – mD : mD
or 1 : mB = nC – mD : mD (nA – mB = 1; Step 1)

219
[3] Hence mD = mB(nC – mD) (products of means and extremes)
thus D = B(nC – mD) (dividing both sides by m)
and so B measures D exactly (nC – mD) times.

[4] Now looking at the given proportion, and multiplying the means and extremes,
A · D = B · C
or A · B · (nC – mD) = B · C ( D = B · (nC – mD), by
Step 3)
and so A · (nC – mD) = C (dividing both sides by B)
and so A measures C exactly (nC – mD) times.

[5] Looking at Steps 3 and 4, we see that B measures D and A measures C, and the
same number of times in each case.

Q.E.D.

THEOREM 7 Remarks:

1. How do we know that we can subtract nC – mD? This will not be a number at all
unless nC is greater than mD. Look at Step 2: nA : mB = nC : mD. We know that nA
is greater than mB, since nA – mB = 1 (Step 1). Hence, by the proportion, it must also
be true that nC is greater than mD.

2. From this Theorem, it is evidence that numbers which are prime to each other are the
least numbers in their ratio, measuring all others that are in the same ratio with
themselves. For example, 5 and 6 are prime to each other, and all other numbers in their
ratio are equimultiples of them, such as 10 and 12.

THEOREM 8: If N is the greatest common factor of A × N and B × N, then A

and B are prime to each other.

For example, 3 is the greatest common factor of 6 and 15, i.e. of 2 × 3 and 5 × 3. And, as
the Theorem states, 2 and 5 have no common factor but 1. Now let’s prove it generally:

Given: Two numbers A × N and B × N, and N is their greatest common factor.

Prove: A and B are prime to each other.

If possible, suppose A and B have a common factor other than 1. Then say it is M, and
A = R×M
and B = S × M.

220
Thus A × N = R × M × N (substituting R × M for A)
and B×N = S×M×N (substituting S × M for B)
So M × N is now a common factor of A × N and B × N. But since M is supposedly
greater than 1, it follows that M × N is greater than N. Therefore M × N is a common
factor of A × N and B × N and it is greater than N. Which is absurd, since it is given that
N is the greatest common factor of A × N and B × N. Therefore our original assumption,
namely that A and B have a common factor other than 1, is impossible. Therefore A and
B have no common factor other than 1, and so they are prime to each other.

Q.E.D.

THEOREM 9: If four numbers are proportional, the greatest common factor of

the first pair measures each the same number of times that the greatest common factor of
the second pair measures each.

Given: A × N : B × N = C × M : D × M,
N is the greatest common factor of A × N and B × N
M is the greatest common factor of C × M and D × M

Prove: A=C
B=D

[1] First of all, we know that A : B = C : D, just by removing the identical multipliers
in the given proportion (cf. Ch.5, Thm.13).

[2] We also know that A is prime to B, since N is the greatest common measure of A
× N and B × N (Given, and Thm.8).

[3] Therefore C and D are equimultiples of A and B (looking at Steps 1 and 2, and
applying Thm.7).

[4] Likewise, C is prime to D, since M is the greatest common measure of C × M and

D × M (Given, and Thm. 8).

[5] Therefore A and B are equimultiples of C and D (applying Thm.7 to Steps 1 and
4).

[6] Thus A and B are equimultiples of C and D (Step 3), and yet C and D are also
equimultiples of A and B (Step 5). Which is impossible, except when A = C and B = D.
Therefore A = C and B = D.

Q.E.D.

221
THEOREM 9 Remarks:

Verify the Theorem in some numerical examples. Find any proportional numbers, and
divide out the greatest common factor of the first two numbers and again the greatest
common factor of the last two numbers, and see what you have left.

THEOREM 10: A prime number is prime to any number that is not its multiple.

Given: P is a prime number.

N is any number which is not a multiple of P.

Prove: N and P have no common factor but 1.

P has no factors but 1 and P (by the definition of a prime number).

N does not have P as a factor (given).
Therefore 1 is the only common factor of N and P.
Therefore N and P are prime to each other.

Q.E.D.

THEOREM 10 Remarks:

What about 1? Is that a “prime number”? If so, it is an unusual one, since every other
number is a multiple of it. Shall we say, then, that 1 is not prime to any number, since it
measures them all? But then again, 1 and 3 are prime to each other in the sense that
nothing except 1 measures both of them. Still, 1 measures 3, and that kind of thing does
not happen with other numbers that are prime to each other. Take any other pair of
numbers that are prime to each other, such as 3 and 5, and we see that neither one can
measure the other; if 3 measured 5, then they would have another common factor besides
1, namely 3. Hence, in the case of numbers other than 1, if two numbers are prime to
each other, then neither one measures the other.

222
THEOREM 11: If a number is prime to two numbers, it is also prime to their
product.

For example, 3 is prime to 4 and also prime to 5. Hence 3 is also prime to 4 × 5. Now
let’s prove it generally:

Given: P is prime to A and also to B.

Prove: P is prime to A × B.

[1] To see it, let F be any common factor of P and A × B. Suppose, then that
P = F×N
and A × B = F × M.

[2] Making a proportion out of these two equalities, we get

A×B : F×M = P: F×N

[3] According to Ch.5, Theorem 13, we can divide the first two by the same number,
and maintain the proportion. So divide the first two equal numbers by B:
A : (F × Μ) ÷ Β = P : F × N
Now let's alternate that proportion:
A : P = (F × M) ÷ B : F × N
Again, we can divide both numbers in the second ratio by F and maintain the proportion:
A : P = M÷B : N

[4] And since A and P are given as prime to each other, it follows that N is a
multiple of P (Thm.7).
But P=F×N (Step 1)
So P is also a multiple of N.
But the only way P and N can be multiples of each other is if P = N, and F = 1.

[5] Since F was a randomly taken common factor of P and A × B, and F must
equal 1, therefore 1 is the only factor common to P and A × B. Therefore P is prime to
A × B.

Q.E.D.

223
THEOREM 11 Remarks

1. For example, 3 is prime to 4 and also to 5. So 3 is also prime to 4 × 5.

2. It also follows that if P is prime to A and to B and to C, that P must also be prime to
the product of all three. For since P is prime to A and also to B, therefore, by the above
Theorem, it is also prime to A × B. Since P is prime to A × B and also to C, therefore, by
the above Theorem, it is also prime to A × B × C.
Obviously, this works for any number of numbers to which P is prime.

3. Does it follow from the Theorem that P does not measure A × B? Not necessarily.
Suppose P = 1. If P, A, B are 1, 3, 5, then 1 is prime to 3 × 5 because it has no
common factor with it but 1, which is itself. But it also measures 3 × 5. However, if P is
anything besides 1, then it will not measure A × B, since it would then have another
factor in common with A × B, namely P itself.

THEOREM 12: If a prime number measures a product, it also measures at least

one of its factors.

Given: The number A × B × C, a product which is divisible by prime number P.

Prove: P is a factor of at least one of the numbers A, B, C.

If P were not a factor of any of the numbers A, B, C, what would follow? Since none of
those numbers would be a multipl3 of P, then by Theorem 10 it would follow that P is
prime to each of them. Since P would be prime to each of the numbers A, B, C, then by
Theorem 11, it would follow that P is prime to their product A × B × C. But P is not
prime to that product (since it is given that it measures that product, and so P and A × B ×
C have at least P as a common factor). Therefore P cannot be prime to A and to B and to
C. Therefore P must measure at least one of them.

Q.E.D.

THEOREM 12 Remarks:

From this Theorem we can draw the following corollary: If a prime number measures a
square number, then it also measures its side. For suppose that the prime number P
measures the square number S × S. By the present Theorem, P must therefore measure
at least one of the factors of S × S, i.e. it must measure S, the side of the square number.
For example, if a prime number measures 6 × 6, it must also measure 6. 1, 2, and 3 are
the only prime numbers that measure 36, and all of them also measure 6.

224
 THE UNIQUE PRIME FACTORIZATION THEOREM

THEOREM 13: Any number is expressible as a product of prime numbers in

only one way.

For example, 24 = 2 · 2 · 2 · 3. And 24 has no other prime factors. Now let’s

show that every number is expressible in such a way.

Given: N, any number you like.

Prove: N is expressible as the product of certain prime numbers, and it cannot be
expressed as any other product of prime numbers.

Obviously, if N is prime, we can express it as 1 × N, where 1 and N are both prime

numbers. And it cannot be expressed as the product of any other numbers at all, since,
being prime, N has no other factors.
What if N is a composite number? Then it is the product of numbers other than
itself and 1, e.g. N = A × B. And if A and B are both composite, then we can express N
as a product of their factors, i.e. N = (C × D) × (E × F).
But since the factors get ever smaller as we do this, we cannot continue forever,
but all composite factors will eventually reduce to factors which have no further factors,
i.e. to prime factors. Thus N itself will be expressible as the product of prime numbers,
say, like this:
N = P1 P 2 P 3 P 4
Of course the order of the terms makes no difference, and some of these prime numbers
might be equal (as in 75 = 5 × 5 × 3). But the question before us is this: is it possible to
express N as the product of a different collection of prime numbers? For example:
N = R1R2R3 …
Even without comitting ourselves to how many primes are in this “new” expression for
N, we can prove that these primes must in fact be the same as those in the product
P1P2P3P4. For since N is divisible by P1, therefore (R1R2R3 …) is too. But since P1 is
prime, therefore it measures one of the primes in (R1R2R3 …), according to Theorem 12.
But since (R1R2R3 …) are all primes, each is measured only by itself and 1, and so the
prime P1 must be identical to one of the primes in (R1R2R3 …). Say, then, that P1 = R1.
We can prove, likewise, that P2 = R2, etc. Therefore all the primes in (R1R2R3 …) end up
being none other than the primes P1P2P3P4. Therefore N is expressible as a product of
certain prime numbers, and only those prime numbers.
We may say, then, that the “prime factorization” of any number is like its DNA
code, its fingerprint.

Q.E.D.

225
THEOREM 13 Remarks:

1. Since every composite number is expressible as a unique product of primes, but no

prime number is expressible as a product of composites, there is a kind of priority of
prime numbers over composite numbers. That is why they are called primes, from
primus for “first.”

2. Exercise: find the prime factorization of the following numbers …

12, 120, 85, 496, 2808.

THEOREM 14: All the factors of any number are: the primes in its prime
factorization, all their products, and 1.

Take any number N, and say its prime factorization is A × B × C.

Suppose F is some factor of N besides 1 and N itself. I say that F is either A, or B, or C,
or some product formed out of these numbers (i.e. either A× B or A × C or B × C or A ×
B × C).
Proof: Since F is a factor of N, then the prime factorization of F contains only
primes which are also in the prime factorization of N (otherwise N could be expressed as
a product of primes in more than one way, contrary to Theorem 13). But A, B, and C are
the primes in the prime factorization of N. Therefore F contains only A, or B, or C (or
some combination of them) in its prime factorization.
Since it is obvious that A × B × C does in fact have A, B, C (and any product
formed out of these three numbers) as factors, and since we have just shown that these
are the only factors, the Theorem is proved.

Q.E.D.

THEOREM 14 Remarks:

If two or more of the prime factors are the same, then some of the factors of the number
N will simply be powers of those prime factors. For example, suppose we take the
number 56, whose prime factorization is 2 × 2 × 2 × 7. What are the factors of 56?
According to the present Theorem, they are 1, 2, 22, 23, 7, 2 × 7, 22 × 7, 23 × 7.

226
THEOREM 15: There is always a bigger prime number.

Given: P is any prime number you like.

Prove: There is another prime number bigger than P.

Suppose someone thought the prime number 5 was the biggest prime number of all, so
that no number after 5 was prime. Let's prove him wrong.
Form the number 5! + 1 (i.e. “five factorial plus one”), which is the number
(5 × 4 × 3 × 2 × 1) + 1
If this number is prime, then since it is obviously greater than 5, we are done. On the
other hand, if it is composite, since every composite number is expressible as the product
of primes (Thm.13), so this number will be also. But since 2 goes evenly into (5 × 4 × 3
× 2 × 1), then it will not go evenly into (5 × 4 × 3 × 2 × 1) plus one, because the unit is
not divisible. For the same reason, (5 × 4 × 3 × 2 × 1) + 1 is not evenly divisible by 3, or
by 4, or by 5. Therefore whatever prime number does measure (5 × 4 × 3 × 2 × 1) + 1 has
to be greater than 5. Therefore there is a prime number greater than 5.
By exactly the same argument, we can always prove that there is a prime number
greater than any given prime number P, just by forming the number P! + 1. This means
that there is an unlimited multitude of prime numbers.

Q.E.D.

THEOREM 15 Remarks:

Note that sometimes P! + 1 is itself a prime greater than P. In other cases it is not
prime, but is still measured by a prime greater than P.
For example, if someone says 3 is the greatest prime number, we can refute him
by proving that 3! + 1 is either a greater prime, or is measured by one. In this case, 3! +
1 = 7, which is a prime number greater than 3.
But if someone says 5 is the greatest prime number, we can refute him by
proving that 5! + 1 is either a greater prime, or is measured by one. In this case, 5! + 1
= 121, which is not prime, but it is measured by 11, which is a prime greater than 5.
Also, notice that 11 is not the next prime after 5.
Accordingly, the proof shows that there must always be a greater prime, but gives
us no sure way of producing it. To this day, there is no known formula which generates
all the prime numbers. The only way to find the primes is to list all the numbers as far as
you like and cross out all the ones that have factors other than 1; the remaining numbers
are the primes. This method is called the “Sieve of Eratosthenes.”

227
THEOREM 16: Give me any number: I can always find that many consecutive
numbers none of which is prime.

We might wonder how primes are “spaced” as we take the next biggest prime, the next
biggest, etc. Are they more or less evenly spaced, or do they become “rarer”? They
become much rarer. For example, is it possible to find 99 consecutive numbers none of
which is prime? Yes indeed.
Just form the factorial number (99 + 1)!, or 100!
Now 100! + 1 is, of course, divisible by 1, but it might be prime.
Still, 100! + 2 must be divisible by 2, since 100! is divisible by 2,
and so is 2. Thus 100! + 2 is not prime.
Again, 100! + 3 is divisible by 3, and thus not prime.
Again, 100! + 4 is divisible by 4, and thus not prime.

etc., etc. all the way up to

100! + 100 which is divisible by 100, and therefore not prime.

So there you have 99 consecutive numbers none of which is prime. Likewise, we can
construct a billion consecutive numbers none of which is prime! Just begin with the
number (1,000,000,000 + 1)! + 2.

Q.E.F.

THEOREM 16 Remarks:

Notice that even though the prime numbers “thin out” as we climb upward in the
numbers, that does not mean the distances between consecutive primes steadily increases
as we go. That is because strange clusters of prime numbers show up – something still
not well understood to this day. For example, prime numbers which have only one
number between them are called “Twin Primes”, such as 5 and 7, or 29 and 31, or 197
and 199, or 821 and 823. As I write these words, it is still not known to the mathematical
community whether or not the “Twin Primes” run out, i.e. whether there is a highest pair
of twin primes, or if there is always another pair.
There are only 26 prime numbers less than 100, which are:
1, 2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37, 41, 43, 47, 53, 59, 61, 67, 71, 73, 79, 83, 89,
97. How many pairs of twin primes can you find in this list?

228
THEOREM 17: How to find the greatest common factor of any two numbers.

Given: Two numbers A and B, and A is greater than B.

Find: Their greatest common factor (or “G.C.F.” for short).

Repeatedly subtract the lesser from the greater till nothing remains (which must happen,
since numbers are finite):
A–B=C
C–B=D
B–D=E
etc. etc. ...
Z–X=X
X–X=0
Then X is the greatest common factor of A and B!
First of all, that X is a common factor of A and B is clear already from Theorem 6
(go back and look at it if you forget!). But how do we know that it is the greatest factor
common to A and B?
Take any number N which is a factor of both A and B. I will show you that it
cannot be greater than X.
Since N measures both A and B, therefore it measures their difference, which is A
– B, i.e. C. Since we now see that N measures both B and C, it must also measure their
difference, i.e. D. Since we now see that N measures both C and D, it must measure their
difference, E, etc. So N must measure every number in the whole process, including X.
Therefore any factor common to A and B must also measure X, and therefore must be
equal to or less than it. Therefore X is the greatest factor common to A and B.

Q.E.F.

THEOREM 17 Remarks:

1. If A and B are prime to each other, D will equal 1. If they are not prime to eacah
other, D will be some number greater than 1. So for any two numbers there is always a
G.C.F., even if it is only 1.

2. We can see from this that If a number measures two numbers, it will also measure
their greatest common factor.

229
3. Notice that in this Theorem we have found a way to discover the greatest common
factor of two numbers without factoring either number! Let’s try an example. What is
the greatest common factor of 36 and 81?
81 – 36 = 45
45 – 36 = 9
36 – 9 = 27
27 – 9 = 18
18 – 9 = 9
9–9=0
So 9 is their greatest common factor. Now try it with 176 and 132.

4. Obviously, we can also find the G.C.F. of three numbers A, B, C simply by finding
the G.C.F. of A and B (say it is N), and then the G.C.F. of B and C (say it is M), and then
finding the G.C.F. of N and M (say it is G). G then has to be the GCF of the 3 numbers
A, B, C. For since the G.C.F. of the 3 numbers measures both A and B, it must also
measure their G.C.F., namely N. And since it measures both B and C, it must also
measure their G.C.F., namely M. And since it measures both N and M, it must measure
their G.C.F., namely G. So the G.C.F. of A, B, C must measure G. And therefore G is
the greatest common factor of the 3 numbers. Q.E.F.

THEOREM 18: How to find the least common multiple of any two numbers.

Given: Two numbers, A and B, and A is the greater one.

Find: Their least common multiple (or “L.C.M.”, for short).

[1] First, by Theorem 17, find the greatest factor common to both A and B. Say it is
N, and that
A=C×N
B=D×N

[2] Therefore
A : B = C×N : D×N (By the two equalities above)
thus A : B = C : D (Chapter 5, Thm.13)

Now since N is the greatest factor common to A and B, it follows that C and D
are prime to each other (Thm. 8).

230
[3] Now the product of the extremes is equal to the product of the means, so
A×D=B×C (Thm.4)
Notice that A × D is measured by A (namely D times), but it is also measured by B
(namely C times). Therefore A × D is a common multiple of A and B. I say that it is
also the least.

[4] For let A and B measure any other number, K, and say that
A×R=K
B×T=K
So that K is also a multiple of both A and B. I say that K is greater than A × D.

[5] Now since the factors of equal numbers are reciprocally proportional (Thm.7),
and A×R=B×T (Since they both equal K)
Thus A : B = T : R (Thm.7)
But A:B = C:D (Step 2)
so C:D = T:R
But C and D are prime to each other (Step 2)
Thus D measures R (Thm.7)
Thus D is less than or equal to R.
So A × D is less than or equal to A × R (multiplying both by A).
i.e. A × D is less than or equal to K.
And since, by supposition, K is a common multiple of A and B other than A × D,
therefore A × D is less than K.
Thus A × D is the least common multiple of A and B.

Q.E.F.

THEOREM 18 Remarks:

1. There is no greatest common multiple of two numbers, since all multiples of their
least common multiple will also be common multiples of them, and we can multiply their
least common multiple as many times as we like.

2. Let’s try a numerical example. Take 24 and 18. What is the least number that both
of them measure? Divide out their greatest common factor, namely 6, and we form the
proportion
24 : 18 = 4 : 3
Multiplying the extremes (or the means), we get 72, which is the least common multiply
of 24 and 18. Now try it with 36 and 15.

231
3. Notice that the product of two numbers is always a common multiple of them, but it is
not always their least common multiple. For example, 4 × 6 = 24, and 24 is indeed a
multiple of both 4 and 6, but it is not their least common multiple – rather, it is 12.
Other times, however, it is the least, e.g. 5 × 6 = 30, and 30 is the least multiple
common to 5 and 6. When does that happen? When the two original numbers are
prime to each other. Use the procedure for finding the L.C.M. of any pair of numbers
that are prime to each other (such as 5 and 6) and you will quickly see why this must be
so.

4. Can you see how to find the L.C.M. of any three numbers?

THEOREM 19: If the base of a triangular number is N, then the triangular

N ( N + 1)
number equals .
2

Consider the number 1 + 2 + 3 + 4 + 5 + 6 + 7 + 8 + 9 + 10.

By definition, that is a triangular number whose base is 10.
Now rearrange the addition by putting together pairs of numbers starting with the two
ends (1 and 10), and working your way in toward the middle (2 and 9 are next):
(1 + 10) + (2 + 9) + (3 + 8) + (4 + 7) + (5 + 6)
Notice that we have here five sums each equal to 11, so the whole sum is 5 × 11 = 55.
The pairs will all be the same, since we start out with 1 + 10, and in the next pair we add
1 to the first number (giving us 2), but we subtract 1 from the second number (giving us
9), and so on.
10
Since there are 10 original numbers being added, that is pairs, or 5 pairs, each
2
10(10 + 1)
of which equals 10 + 1. So the whole sum is .
2
What happens if we have an odd number of numbers, you ask?
Consider 1 + 2 + 3 + 4 + 5 + 6 + 7 = K.
This number K equals half of (1 + 2 + 3 + 4 + 5 + 6 + 7) + (1 + 2 + 3 + 4 + 5 + 6 + 7).
Now rearrange, as before, in pairs of numbers, and now K is half of
(1 + 7) + (2 + 6) + (3 + 5) + (4 + 4) + (5 + 3) + (6 + 2) + (7 + 1)
each of these pairs being equal to (7 + 1). And how many pairs are there? Since we
simply repeated our original numbers twice, there are 7 pairs, i.e. as many pairs as
original numbers. The entire sum, then, is equal to the number of these pairs times the
value of each pair, i.e. 7(7 + 1). But K is only half of that, and so it is equal to
7(7 + 1)
2
N ( N + 1)
So the formula works equally well with odd numbers.
2

Q.E.D.

232
THEOREM 19 Remarks:

1. Carl Friedrich Gauss, the great German mathematician, discovered this formula for
triangular numbers all by himself at the tender age of 6. His teacher assigned his class
this problem: add all the numbers from 1 to 100. The teacher, expecting the students to
be busy for a solid hour, was surprised when little Gauss approached him with the correct
answer right away: 5050. Gauss saw that 1 + 2 + 3 + … + 98 + 99 + 100 is equal to 50
pairs each equal to 101. So he simply multiplied 50 times 101 and got his answer.

2. It is possible to give a “visual proof” for this Theorem.

Let a triangular number, having a base equal to N, be
represented by filled circles placed in the form of a right N
N
triangle. If we place an identical triangular number of
empty circles against it, we complete a rectangle with sides
of N and N + 1. The whole rectangle contains N (N + 1)
circles. Accordingly, since the original triangular number
is only half this rectangle, it contains N (N + 1) ÷ 2 N
N
circles. Hence the triangular number of base N is
N ( N + 1)
.
2

THEOREM 20: The sum of two consecutive triangular numbers is a square

number, and the base of the larger triangular number is the side of the square.

Given: Two consecutive triangular numbers, namely

N ( N + 1) ( N + 1)[( N + 1) + 1]
and
2 2
which, by the last Theorem, are triangular, and which are consecutive
because they have consecutive bases N and N + 1.

Prove: The sum of these two numbers is a square number whose side is N + 1.

Our two triangular numbers are half of N(N + 1) and (N + 1)[(N + 1) + 1] respectively.
So their sum is half the sum of these two numbers, or
N (N + 1) + (N + 1) [(N + 1) + 1] divided by two,
which is N (N + 1) + (N + 1) (N + 2) divided by two.
Now, using Theorem 1 of Chapter 5, we can “factor out” the (N + 1) common to both
parts of our sum, which is to say that our whole sum is the same as
(N + 1) [N + (N + 2)] divided by two
which is (N + 1) (2N + 2) divided by two
Again, using Theorem 1 of Chapter 5, we can “factor out” the 2 common to both parts of
(2N + 2), which is to say that our whole sum is the same as

233
 (N + 1) (N + 1)2 divided by two
And since our whole sum is reached by multiplying by two but then dividing by two, we
can simply say that the sum of our two triangular numbers is
(N + 1) (N + 1)
which is, by definition, a square number whose side is N + 1.

Q.E.D.

THEOREM 20 Remarks:

1. As a kind of porism, we can say that any square number is equal to the sum of
consecutive numbers from 1 up to its square root and back down to 1. For example,
16 = 1 + 2 + 3 + 4 + 3 + 2 + 1
The reason is that 1 + 2 + 3 is a triangular number
and again 1 + 2 + 3 + 4 is the next triangular number,
and so, by the Theorem, their sum is equal to 42. But their sum is the same as the
numbers from 1 up to 4 and back down again to 1.

2. Can a number be both square and triangular? Yes. For example, 36 is both 62 and
also 1 + 2 + 3 + 4 + 5 + 6 + 7 + 8. How many numbers are both square and triangular?
An infinity of them, but we won't prove that here.

3. As you have probably guessed, it is possible to give a kind of

visual proof for this Theorem. Arrange any square number of dots in
a square pattern, and you will see that you can slice it into two
consecutive triangular numbers. For example, 42 equals the
triangular number on base 3 plus the triangular number on base 4.
If you draw lines diagonally through a square number of
things arranged in a square pattern, you can see that the longest
diagonal – 5 in the diagram – equals the side of the square and is the
common base of two identical triangular numbers. You can see why
25 = 1 + 2 + 3 + 4 + 5 + 4 + 3 + 2 + 1.

234
THEOREM 21: Every square number N2 is the sum of N consecutive odd
numbers beginning with 1; and the even number following the last odd number added is
double the side of the square.

Notice 1 is a square, for it equals 1 × 1,

and 1 + 3 is a square, for it equals 2 × 2,
and 1 + 3 + 5 is a square, for it equals 3 × 3.
Will this always work? Yes, and here's why:
Take any square number N2, having side N. This number is the sum of two
consecutive triangular numbers, say one with base N and one with base N – 1, which is to
say that
N2 = 1+2+3+4+ …+N (triangular number with base N)
+1+2+3+…+N–1 (triangular number with base N – 1)

Now add the numbers in pairs, each number in the top row to the one below it in the
bottom row, and we have

N2 = 1 + (2 + 1) + (3 + 2) + (4 + 3) + … + (N + N – 1)

Since in each case we are adding an even number and an odd one, the parentheses all
contain odd sums. And since, moreover, each member of each pair is always one more
than its corresponding member in the previous pair, the result is that each pair adds up to
two more than the previous pair. i.e. we are adding consecutive odd numbers. i.e.

N2 = 1 + 3 + 5 + 7 + … + 2N – 1

And the last number added is (2N – 1), and so the number after it is simply 2N, half of
which is N, the side of the square.

Q.E.D.

THEOREM 21 Remarks:

Since 2N is double the number of odd numbers from 1 to N, it follows that N is not only
the side of the square, but also the number of consecutive odds which add up to the
square. For example, 1 + 3 + 5 = 9, and the side of this square, namely 3, is also the
number of odds added.

235
THEOREM 22: The product of two square numbers is a square number (whose
side is the product of their sides).

Given: A × A and B × B
Prove: (A × A)(B × B) is a square

Since the order of multiplication makes no difference, therefore

(A × A)(B × B) = (A × B)(A × B),

i.e. (A × A)(B × B) is a square number whose side is (A × B).

Q.E.D.

THEOREM 22 Remarks:

For example, 5 2 × 9 2 = (5 × 9) 2
or 25 × 81 = 45 2 = 2025.
It is also true that the product of two cubes is a cube whose side is the product of their
sides. That is, (A × A × A)(B × B × B) = (A × B)(A × B)(A × B)
For example 2 3 · 5 3 = (2 · 5) 3
or 8 · 125 = 10 3

THEOREM 23: If a square times some number makes a square, the number is
also a square.

Given: (A × A) × N = ( B × B)
Prove: N is a square number.

Now either A is prime, or else it is the product of primes (Thm.13). Either way, A can be
expressed in terms of nothing but primes, so say
A = P1 P2 P3
Since A measures B × B, therefore all these primes measure B × B. But any prime that
measures a square number must also measure the side of the square (Thm.12 Remarks).
Therefore all these primes measure B. Hence P1P2P3 must be found in the prime
factorization of B. But that means that each B in B × B is measured by P1P2P3, or
more simply, by A. Since each B is divisible by A, it follows that B ÷ A is a number.
Now A × A × N = B × B (given)
so N = ( B × B) ÷ (A × A) (dividing both sides by A × A)
so N = ( B ÷ A) × ( B ÷ A) (See Ch.6, Thm. 4)
and since ( B ÷ A) is a number, therefore N is a square number. Q.E.D.

236
THEOREM 23 Remarks:

1. For example, consider 5 2 × N = 30 2

clearly N = 30 2 ÷ 5 2
or N = (30 × 30) ÷ (5 × 5)
so N = (30 ÷ 5) × (30 ÷ 5) = 6 × 6

2. It follows from this Theorem that No square number is double any other square
number. For if so, i.e. if Q × 2 = R where Q and R are both square numbers, it would
follow by the Theorem that 2 is a square number, which it certainly isn't.
Thus, in general, it follows that No square number is a non-square multiple of any
other square number. For example, no square number is 3 times another, or 5 times
another, or 6 times another, etc.

THEOREM 24: No prime number is to any other prime number as a square

number is to a square number.

Take any two distinct prime numbers N and M. I say that N and M do not have the same
ratio as any two square numbers.

[1] If possible, assume that

N:M = A 2:B 2
where A2 and B2 are the least square numbers having the ratio of N to M.

[2] Since N and M are distinct primes, they have no common factor but 1, and hence
are prime to each other. Therefore, according to the proportion above, N must measure
A2 (Thm.7). But since N is prime, it must also measure A, since any prime number that
measures a square number also measures its side (Thm.12, Remarks). Likewise M
measures B.

[3] So let A = K· N and B = L · M

Thus A2 = K2 · N 2
and B 2 = L 2
· M 2

[4] Then A2 : B 2 = K2 · N 2 : L 2 · M 2
But N and M have the same ratio as A2 and B2 (Step 1), so
N : M = K2 · N 2 : L 2 · M 2
Now dividing both antecedents by N, and both consequents by M, we have
1 : 1 = K2 · N : L 2 · M
From which it obviously follows that
K2 · N = L 2 · M
Since the factors of equal numbers are reciprocally proportional (Thm.4), thus
N : M = L 2 : K 2

237
[5] By Step 3 it is plain that L2 and K2 are less than A2 and B2. But by Step 4, we
see that L2 and K2 have the same ratio as N and M, whereas A2 and B2 are
supposed to be the least squares in that ratio. Which is impossible. Therefore N and M
cannot have the same ratio as two square numbers.

Q.E.D.

THEOREM 24 Remarks:

Citing Step 3, Step 5 asserts that K2 < A2 and L2 < B2. How is this clear from Step 3?
Because A2 = K2 · N 2 and B2 = L2 · M 2.
But what if N = 1 ? 1 is a prime number, after all. And in such a case
A2 = K2 .
All right, but then since N and M are distinct primes, M is not 1, but some number
greater than 1. So since B2 = L2 · M 2 it will still be true that
B2 > L2.
So the Theorem will hold regardless.

From this Theorem, then, it follows that no two square numbers have the ratio of
1 to 2, or 1 to 3, or 2 to 3, or 3 to 5, etc.

THEOREM 25: Two numbers whose factors are proportional have a mean
proportional between them.

Consider A ⋅ B and C ⋅ D
where A : B = C : D.
I say that A ⋅ B and C ⋅ D have a mean proportional number between them.

For A:B = C:D (given)

so A:C = B:D (alternating the proportion)
thus A⋅B : B⋅C = B : D (multiplying the 1st two terms by B)
and A⋅B : B⋅C = B⋅C : C⋅D (multiplying the last two terms by C)
i.e. B ⋅ C is a mean proportional between A ⋅ B and C ⋅ D.

Q.E.D.

238
THEOREM 25 Remarks:

1. Since B ⋅ C is equal to A ⋅ D (since we are given that A : B = C : D), it also

follows that A ⋅ D is a mean proportional between the two given numbers. Although it
is the same mean proportional, it is expressed in terms of different factors.

2. Any two square numbers have one mean proportional between them. For N × N and
M × M obviously have proportional factors: N : N = M : M. Thus the mean
proportional between them is the product of their sides, N × M. Confirm this with some
examples.

3. Also, any two cube numbers have two mean proportionals between them. Consider
A⋅A⋅A and B⋅B⋅B. The 2 means between them are A⋅A⋅B and A⋅B⋅B. Why?
Because A⋅A : A⋅B = A⋅A : A⋅B obviously, since these ratios have identical
terms. Now we maintain the proportion if we multiply the first two terms by A and the
last two terms by B, giving us
(1) A⋅A⋅A : A⋅A⋅B = A⋅A⋅B : A⋅B⋅B
Again A⋅B : B⋅B = A⋅B : B⋅B obviously, since these ratios have identical
terms. Multiplying the first two terms by A and the last two terms by B, we have
(2) A⋅A⋅B : A⋅B⋅B = A⋅B⋅B : B⋅B⋅B
Putting together Proportion (1) with Proportion (2), we see that A⋅A⋅B and A⋅B⋅B are
mean proportionals between the two original cube numbers.

THEOREM 26: Two numbers with a mean proportional between them have
proportional factors.

Given: X:N = N:Z

Prove: X and Z have proportional factors

[1] Find the G.C.F. of X and N, and call it G (Thm.17). Since G is a factor of both X
and N, therefore each is equal to some number times G, say
A ⋅ G = X
B ⋅ G = N

[2] And since G is the greatest factor common to X and N, it follows that A and B are
prime to each other (Thm.8). Now, from the two equalities in Step 1, it is clear that
A⋅G : B⋅G = X : N
thus A:B = X:N (dividing the first terms by G)
but X:N = N:Z (given)
thus A:B = N:Z (each is the same as the ratio X : N)

239
[3] But A and B are prime to each other, and therefore N and Z are equimultiples of
A and B (Thm.7). Say N and Z are each F times A and B, i.e.
A ⋅ F = N
B ⋅ F = Z

[4] Thus A ⋅ F = B ⋅ G (each is equal to N; Steps 1 and 3)

so A:G = B:F (means and extremes are proportional)
But these are the factors of X and Z, since
A ⋅ G = X (Step 1)
and B ⋅ F = Z (Step 3)
So that X and Z in fact have proportional factors.

Q.E.D.

THEOREM 26 Remarks:

It follows from this Theorem that two numbers whose factors are not proportional have
no mean proportional between them. For if they did have a mean proportional between
them, their factors would be proportional (by this Theorem). Find some numerical
examples to confirm this and other numerical examples to confirm the Theorem.

THEOREM 27: The product of two numbers with proportional factors is square.

For example, consider 12 and 48, whose factors are proportional, since 3 × 4 = 12, and
6 × 8 = 48, and since 3 : 4 = 6 : 8. Hence the product of 12 and 48 is square, and
indeed 3 × 4 × 6 × 8 = 576 = 24 2. Now let’s prove it generally ...

Given: A × B and C × D
A:B = C:D

Prove: A × B × C × D is a square number

[1] First, A:B = C:D (given)

[2] Thus A×D = B×C (product of means = product of extremes)

[3] Therefore (A × D) (B × C) is a square number, since by Step 2 these two

factors are equal. That is, A × B × C × D is a square number (since the order of
multiplication makes no difference).

Q.E.D.

240
THEOREM 28: If one number measures the square of another, then there is a
third number proportional to the two numbers; if not, not.

Given: A measures B2

Prove: There is a third proportional number N such that

A:B = B:N

Since A measures B2 some definite number of times, let that number be N, so that
A ⋅ N = B ⋅ B
Thus A:B = B:N (Thm.4)
So N is a third proportional to A and B. Q.E.D.

Given: A does not measure B2

Prove: There is no third proportional number N such that

A:B = B:N

If possible, suppose that

A:B = B:N
then A ⋅ N = B ⋅ B (Thm.4)
i.e. A does measure B2, which is absurd, since we are given that A does not
measure B2. So there is no number N such that A : B = B : N, unless A measures B2.

Q.E.D.

THEOREM 28 Remarks:

1. For example, do 5 and 6 have a third proportional number N, such that 5 : 6 = 6 : N?

No, since 5 does not measure 6 × 6 or 36. On the other hand, 4 and 6 have a third
proportional number N, since 4 does measure 6 × 6, namely 9 times. So 4 : 6 = 6 : 9.

2. If two numbers (neither of which is 1) are prime to each other, then there is no third
proportional to them. Say A and B are prime to each other. Therefore A is prime to B2,
since A is given as prime to B and therefore it must also be prime to the product of two
B's (Thm.11). Therefore A does not measure B2. So there is no third proportional
number to A and B (by the present Theorem). Therefore if two numbers are prime to
each other, they have no 3rd proportional.
Of course, 1 and 3 are prime to each other, and yet they have a third proportional,
namely 9. But that is because 1 is not only prime to every number, but measures every
number (see Thm.11, Remark 3). So 1 must measure the square of every number, and
therefore for 1 and any number X there will always be a third proportional.

241
THEOREM 29: If one number measures the product of two others, then there is
a fourth number proportional to them; if not, not.

Given: A measures B × C

Prove: There is a fourth proportional number to A, B, C

i.e. there is some number N such that A : B = C : N

A measures B × C some number of times (given). Say A measures it N times.

Hence A ⋅ N = B ⋅ C
thus A:B = C:N (Thm.4)
i.e. A, B, C have a fourth proportional.

Q.E.D.

Given: A does not measure B × C

Prove: There is no fourth proportional number to A, B, C

If possible, say A:B = C:N

thus A ⋅ N = B ⋅ C (Thm.4)
i.e. A measures B × C exactly N times, which is absurd, since it is given that A does not
measure B × C. So it is impossible to find a fourth proportional number to A, B, C.

Q.E.D.

THEOREM 30: If numbers are in a continuous proportion that begins from 1,

then all terms after the second term are consecutive powers of it.

Consider the numbers 1, 3, 9, 27, 81. They are in a continuous proportion beginning
from 1, since 1 : 3 = 3 : 9 = 9 : 27 = 27 : 81.
Also 9 = 32
and 27 = 33
and 81 = 34
That is, all the terms after 3 are consecutive powers of 3. Every continuous proportion
beginning from 1 must be this way. Here’s why:

242
Given: A, B, C, D, E etc. in continuous proportion beginning from 1,
i.e. 1 : A = A : B = B : C = C : D = D : E etc.

Prove: B = A2
C = A3
D = A4 etc.

[1] 1:A =A:B (given)

so B = A2 (products of means and extremes)

[2] But A:B = B:C (given)

so A · C = B · B (products of means and extremes)
or A · C = A · A · A · A (B = A · A, Step 1)
so C = A · A · A (dividing both sides by A)
3
i.e. C = A

and so on.

Q.E.D.

THEOREM 31: How to find the sum of N numbers that are in a continuous
proportion beginning from 1.

Recalling the last theorem, numbers in continuous proportion beginning from 1 look like
this: 1, A, A2, A3, A4 etc.
If we want the sum of these up to and including A4, just take the difference (A5 – 1) and
divide it by (A – 1). For example

1 + 3 + 32 + 33 + 34 = (35 – 1) ÷ (3 – 1)

Why does that work? For brevity,

let S = 1 + A + A2 + A3 + A4
now S (A – 1) = S · A – S (Ch.5, Thm.1)
or S (A – 1) = A (1 + A + A + A + A ) – 1 – A – A2 – A3 – A4
2 3 4

so S (A – 1) = A + A2 + A3 + A4 + A5 – 1 – A – A2 – A3 – A4
or S (A – 1) = A5 – 1
hence S = (A5 – 1) ÷ (A – 1)

Q.E.F.

243
THEOREM 31 Remarks:

1. Although the proof sums terms only up to A4, you can see why 1 + A + A2 + A3 + A4
+ A5 = (A6 – 1) ÷ (A – 1). In general, the sum of continuously proportional numbers
from 1 to AM is (AM + 1 – 1) ÷ (A – 1).

2. In particular, then, 1 + 2 + 22 + 23 + ... + 2X = (2X + 1 – 1).

THEOREM 32: If (2N – 1) is a prime number P,

then the product T = (2N – 1) (2N – 1) is a perfect number.

[1] To see this, recall that a perfect number is a number equal to the sum of all its factors
other than itself. Now what are all the factors of the product T? Since (2N – 1) is prime
P, and since (2N – 1) is a power of 2 made of nothing but a bunch of twos multiplied
together (and 2 is prime), therefore all the factors of T = (2N – 1) (2N – 1) are:
{a} 1, plus all the powers of 2 up to 2N – 1,
and {b} the products of P with 1 and with all those different powers of 2.
This we know from Theorem 14.

[2] As for the sum of {a}, namely of 1 and all the powers of 2 up to 2N – 1, this is
equal to 2(N – 1) + 1 – 1 (Thm.31, Remark 2).
So the sum of {a} = (2N – 1).

[3] As for the sum of {b}, namely the products of P with 1 and with all the powers of
2, this is equal to the product of P with the sum of 1 plus all those powers of 2 (Ch.5,
Thm.1),
so the sum of {b} = P (1 + 2 + 22 + 23 + ... + 2N – 1)
i.e. the sum of {b} = P [ sum of {a} ]
that is, sum {b} = P (2N – 1)

[4] Since all the factors of T add up to sum {a} + sum {b}, hence all the factors of T
add up to (2N – 1) + P (2N – 1) (Steps 2 and 3)
or (2N – 1) (1 + P) (factoring out (2N – 1); Ch.5, Thm.1)
N N
or (2 – 1) (1 + 2 – 1) (writing out the expression for P)
so (2N – 1) (2N) is the sum total of all the factors of T.

244
[5] So what is the sum of T’s factors that are less than itself? Nothing but the sum of
all its factors minus itself, i.e.
(2N – 1) (2N) – T
Replacing T with its whole expression, this is
(2N – 1) (2N) – (2N – 1) (2N – 1)
But 2 is double 2N – 1, since every power of two is double the previous power of two.
N

Hence 2N = (2N – 1 + 2N – 1). Substituting this expression for 2N, all the factors of T
less than itself add up to
(2N – 1) (2N – 1 + 2N – 1) – (2N – 1) (2N – 1)
But the product on the left equals the sum of (2N – 1) times each term in the parentheses
(Ch.5, Thm.1). So we have
(2N – 1) (2N – 1) + (2N – 1) (2N – 1) – (2N – 1) (2N – 1)
But this simply leaves
(2N – 1) (2N – 1)
which is T !

[6] So the sum of all T’s factors less than itself is equal to T itself. Therefore T is a
perfect number.

Q.E.D.

THEOREM 32 Remarks:

1. For example, 22 – 1 is prime, since it is 3. Therefore (22 – 1)(22 – 1) must be perfect.

Working it out, this product is (4 – 1)(21), or simply (3)(2), which is 6. And indeed all
the factors of 6 which are less than 6 itself are 1, 2, 3. And these add up to 6.

2. Perfect numbers are relatively rare. The first 10 of them are:

N=2 6
N=3 28
N=5 496
N=7 8128
N = 13 33550336
N = 17 8589869056
N = 19 137438691328
N = 31 2305843008139952128
N = 61 2658455991569831744654692615953842176
N = 89 191561942608236107294793378084303638130997321548169216
Note that all the values for N are prime. The next several perfect numbers are generated
by plugging in the following values for N in the perfect number formula: 107, 127, 521,
607, 1279, 2203, 2281, 3217, 4253, 4423, 9689, 9941, 11213, 19937, 21701. The last
few of these, when spelled out in all their digits, take up several sheets of paper. In 1981,
the largest known prime was of the form 2N – 1, namely 244497 – 1, and so 44497 is
also a value for N generating a perfect number.

245
3. There are some interesting facts about perfect numbers. First, it is unknown to this
day whether or not any odd perfect numbers exist. All the perfect numbers of the type
described in this Theorem are obviously even. It is known, however, that all perfect
numbers that are even are of the form described in this Theorem.

4. If you add the digits in a perfect number, and do the same with the resulting number,
and so on until you can go no further, the result is always 1. The only exception is with
the first perfect number, 6. But the next is 28, and its digits add up to 10, whose digits
add up to 1. The next perfect number is 496, whose digits add to 19, whose digits add to
10, whose digits add to 1. The next perfect number is 8128, whose digits add to 19,
whose digits add to 10, whose digits add to 1, etc.

5. It has been proved that there is an infinity of perfect numbers, but fewer than 50 are
known today.

6. Every number of the form (2N – 1)(2N – 1), including those that are perfect (namely
when 2N – 1 is prime), is a triangular number whose base is (2N – 1). So the prime
number which is distinctive of each even perfect number is also its triangular base.

7. No even perfect number is square. That is obvious, since its prime factorization is a
bunch of twos and one odd prime. There is no way to divide up those factors into two
equal factors.

8. No even perfect number measures any other. That is clear because if they are
different, their unique primes are different. But then if one measures the other, all its
factors will have to measure the other, also. And so its unique prime will measure the
other, and so the other will have that prime in its prime factorization, too, which is
impossible. For each even perfect number has only one odd prime in its factorization,
namely 2N – 1.

9. All even perfect numbers end their digits with either 6 or 8.

10. All even perfect numbers (except 6) have the same remainder when you divide them
by 6, namely a remainder of 4.

“HOOK”: THE ONLY SQUARE SQUARE-PYRAMIDAL NUMBER.

If you add consecutive numbers starting from 1, the sum is called a “triangular number.”
But if you add consecutive square numbers starting from 1, the sum is called a “square
pyramidal number,” because the units can be arranged in the form of a “square pyramid”.

Question: Is any square pyramidal number also a square number? Yes! The number
4900 = 702 = 12 + 22 + 32 + ... + 242 .

246
What is more amazing, 4900 is the only square pyramidal number which is also a square
number (other than the trivial example of 1).

“HOOK”: THE ONLY TRIANGULAR SQUARE-PYRAMIDAL NUMBERS.

Are any square pyramidal numbers triangular? Yes, namely these:

1
55
91
208335

And those are the only ones!

“HOOK”: NUMBERS EXPRESSIBLE AS SUMS OF CUBES IN TWO WAYS.

When G. H. Hardy pulled up in front of Ramanujan’s flat for the first time, in a cab
numbered 1729, Hardy remarked it was a singularly uninteresting number, to which
Ramanujan famously replied “Not at all, it is a very interesting number, since it is the
first to be expressible as the sum of two cubes in two ways,” meaning

1729 = 13 + 123 = 93 + 103

Apparently, this identity was found in Ramanujan’s notebooks dated earlier than this
meeting, so he did not simply come up with it on the spot.

247
It is perhaps unsatisfying that 1 is involved, since 1 is in a way a “cube number” only in a
degraded sense. So here are a couple other examples which don’t involve 1:

6507811154 = 313 + 18673 = 3973 + 18613

6058655748 = 613 + 18233 = 10493 + 16993

“HOOK”: FERMAT’S LAST THEOREM.

Is it ever the case that the sum of two square numbers is itself a square number? Yes.
For example:

32 + 42 = 52

This is plainly connected to the Pythagorean Theorem.

But what about cube numbers? Is it ever the case that the sum of two cube numbers is
itself a cube?

No! With the trivial exception of 0 (which, for the purposes of this book, is not
considered a “number” anyway), it is never the case that

a3 + b3 = c3

where a, b, c are all numbers, i.e. positive integers.

In fact, it is never the case that

an + bn = cn

where a, b, c, n are all integers, and n > 2.

This more general statement is Fermat’s famous last theorem, proved only in the 20th
century.

248
 Chapter Eight

Irrational Magnitudes
 DEFINITIONS

1. A MAGNITUDE is a quantity which is divisible forever.

For example, a line is a magnitude because it can be divided into parts, its parts
can be divided into parts, and so on without ever coming to an end. But a number (as
defined in Ch.7, definitions 2 & 3) is not a magnitude, because its divisibility ends with
its units, which are indivisible.

2. Two magnitudes are COMMENSURABLE if they have a common measure, but

INCOMMENSURABLE if they do not.
(The sense of “measure” in this definition is the one defined in Ch.5, Def.1.)
For magnitudes to be commensurable or incommensurable they must be
comparable in terms of greater than, less than, or equal to – e.g. two lines, or two areas,
not a line and an area.

3. Designating any straight line as our unit length, straight lines commensurable with it
are called RATIONAL LINES in reference to it; those incommensurable with it are
called IRRATIONAL LINES. And any areas commensurable with the square on the unit
line (i.e. the “unit square”) are called RATIONAL AREAS in reference to it; those
incommensurable with the unit square are called IRRATIONAL AREAS.

4. A FRACTION of a magnitude is any measure of it taken any number of times.

For example, “five eighths” of a straight line is a fraction of it, since that means
taking one eighth of it give times, and it is written

5
8

And “eight fifths” of a straight line is a fraction of it, since that means taking one fifth of
it eight times, and it is written
8
5

Though eight fifths is larger than the original line, it is still called a fraction (although
sometimes it is called an “improper” fraction).

248
 PRINCIPLES

1. Any two commensurable magnitudes have to each other the same ratio as some pair of
numbers, and any two magnitudes having the same ratio as some pair of numbers are
commensurable with each other.
For example, if a straight line measures another straight line 5 times, and yet
another straight line 8 times, then the two measured lines have the ratio 5 : 8. And if two
straight lines have the ratio 5 : 8, then they are commensurable.

2. No two incommensurable magnitudes have to each other the same ratio that any two
numbers have, and magnitudes not having the same ratio that any two numbers have are
incommensurable with each other.

3. Any two multiples of the same magnitude are commensurable with each other, and
have the same ratio as the multiplying numbers.
For example, 5A and 3A are commensurable, having A as a common measure,
and they have the same ratio as 5 and 3.

249
 THEOREMS

THEOREM 1: Any fraction of a magnitude is commensurable with it.

Given: Any magnitude M

Any fraction of it, F

Prove: F is commensurable with M

F M

Since F is a fraction of M, it is equal to some measure of M taken some number of times.

Say it is equal to 5 times the seventh part of M.
Then a seventh of M measures F (5 times)
and a seventh of M measures M (7 times)
so F and M have a common measure (namely a seventh of M)
thus F is commensurable with M

Q.E.D.

THEOREM 1 Remarks:

1. It does not matter whether F is greater or smaller than M. The proof still works in
exactly the same way.

2. Any fraction of the unit length is rational. For a fraction of the unit length will have to
be commensurable with it (by this Theorem), and any length commensurable with the
unit length is rational (Def.3).

3. Any fraction of the unit square is rational. For a fraction of the unit square will have
to be commensurable with it (by this Theorem), and any area commensurable with the
unit square is rational (Def.3).

4. No irrational line length is expressible as a fraction of the unit line, for then it would
be rational (Remark 2 above). So if we call a certain line length “1,” then it is impossible
to designate an irrational line (i.e. one that is incommensurable with the line we have
chosen to call 1) as a fraction, e.g. as “two thirds” or as “one-hundred-and-twenty-three
thousandths.”

250
THEOREM 2: Squares with commensurable sides have the same ratio as two
square numbers.
A B
Given: Square A and square B, which have
commensurable sides.
mK
nK
Prove: Square A has to square B the same
ratio as a square number to a square number.
mK

nK
Since the side of A is commensurable with
the side of B, let K, their common
measure, go n times into the side of A, and m times into the side of B.
Now divide the sides of A into n equal parts, and divide the sides of B into m
equal parts, each part therefore being equal to K, and complete the “grid” in each square.
The area of square A is now divided into n rows each containing n of the
equal squares, each with a side of K. So the area of A equals n × n such squares.
Likewise, the area of B equals m × m of those same squares, each with a side of K.

Thus A : B = n × n squares : m × m squares

that is A : B = n×n : m×m (Principle 3)

And so squares A and B have the same ratio as a pair of square numbers.

Q.E.D.

THEOREM 2 Remarks:

1. The proof reveals that if the sides of two squares are as the numbers n and m, then
the squares themselves are as the numbers n2 and m2.

2. It follows that squares with areas that do not have the ratio of any two square numbers
have incommensurable sides. For if their sides were commensurable, then their areas
would have the same ratio as a pair of square numbers.

251
THEOREM 3: Any square whose area is a non-square number of times the unit
square has an irrational side.

Let your unit line length be set out, and make a

square on it, which is thus the unit square (with 1
square unit of area). Now take any square whose
2 area is some number of times this unit square, but
choose a non-square number, say 2.
1
Since 2 is a non-square number, I say that the side
of the new square (having 2 times the area of the
1 unit square) is irrational.

For these two squares have the ratio of 1 : 2 (given).

But no two square numbers have the ratio of 1 : 2.
For, if possible, say 1 : 2 = N2 : M2
Then, multiplying the means and extremes in this numerical proportion,
we get M2 = 2N2 (Ch.7, Thm.4)
And so it seems that 2 times a square number (namely N2) makes a square number
(namely M2). From this fact, it follows that 2 is itself a square number, since only a
square number times a square number produces a square number (Ch.7, Thm.23). But 2
is not a square number, so our initial assumption that two square numbers can have the
ratio of 1 : 2 must have been false.
So no two square numbers have the ratio of 1 : 2.
Therefore our two squares, which do have the ratio of 1 : 2, do not have the ratio
of any two square numbers, either. Therefore our two squares have incommensurable
sides (Thm.2, Remarks). Thus the side of the square with 2 times the area of the unit
square, being incommensurable with the unit length, is irrational.
And the same is true for the side of a square with 3 square units of area, or 5, or 6,
or 7, or any non-square number of square units of area. They all have irrational sides.

Q.E.D.

THEOREM 3 Remarks:

1. Incidentally, how would you make a square with 2 times the area of the unit square?
How do you make a square that is double the area of a given square? How about triple?
How about 4 times? Can you see how to make a square that has any number of times the
area of a given square? Start by taking the given square the required number of times,
and putting them together into one rectangle. Now make a square equal to that rectangle.

2. Even if we do not use the unit square, but just any old square S, the proof works the
same way to show that the side of another square that is a non-square number of times the
area of S is incommensurable with the side of the first square.

252
THEOREM 4: The diagonal of a square is incommensurable with its side.

Take any square ABCD. I say that side AB is B C

incommensurable with diagonal AC.

Draw the square on AC, namely ACEF. If you extend

D
CD and AD, they pass through F and E, dividing square A E
ACEF into 4 triangles identical to triangle ACD. But
square ABCD is divided into 2 triangles identical to
triangle ACD.
Therefore £ABCD : £ACEF = 2 : 4 F
or £ABCD : £ACEF = 1 : 2
So these two squares do not have the ratio of a square number to a square number, since
no square numbers have the ratio of 1 : 2 (as we saw in Thm.3). And therefore the sides
of these two squares are incommensurable with each other (by Thm.2, Remarks). Thus
AB is incommensurable with AC.

Q.E.D.

THEOREM 4 Remarks:

This Theorem, even more clearly than the last one, shows that some
magnitudes are incommensurable. We can easily make a square – now
just draw the diagonal, and it will be incommensurable with the side of
the square. In the absence of proof, it might seem impossible for two
comparable magnitudes to be incommensurable. Given any pair of
straight lines, A and B, can’t we find a tiny straight line that fits
exactly into each of them some number of times? After all, there is no A B
limit to how small straight lines can get – there must be one small
enough to measure both A and B exactly ... right?
Wrong! Some magnitudes are incommensurable. We have just shown that
nothing measures both the side of a square and its diagonal.
Before reading this Chapter, you might have thought that all ratios can be
expressed numerically, as a ratio between two numbers. After all, there is an infinity of
numbers, and an infinity of numerical ratios. So given any two straight lines, such as A
and B, musn’t their ratio be expressible as a ratio between two numbers? Not if A and B
are incommensurable, as the side of a square and its diagonal are. Not all ratios are
numerical ratios, and all the numerical ratios are not all the ratios.

253
THEOREM 5: The height of an equilateral triangle is incommensurable with its
side.
A

Take any equilateral triangle ABC, drop AH

perpendicular to BC. Thus AH is the height.

I say side AB is incommensurable with height AH.

B H C

Proof: Since BH is half the length of the side of the triangle, thus AB = 2BH. Hence the
square on AB is four times the square on BH, i.e.

£AB : £BH = 4 : 1
so £AB : £AB – £BH = 4 : 4 – 1 (Ch.5, Thm.16, Remark 3)
i.e. £AB : £AH = 4 : 3 (£AB - £BH = £AH; Pythagorean)

Now 4 and 3 don't have the same ratio as any two square numbers. For if possible,
suppose that

4 : 3 = M2 : N2
thus 4N2 = 3M2 (Ch.7, Thm.4)

And since 4N2 is a square number (since a square number times square number yields a
square number), thus 3M2, its equal, is also a square number. But since only a square
number times a square number yields a square number (Ch.7, Thm.23), thus 3 must be a
square number. But that's impossible. Therefore, too, it is impossible that 4 and 3 should
have the same ratio as any two square numbers.
But since the ratio of 4 : 3 is not the ratio of any two square numbers, therefore
the ratio of £AB : £AH is likewise not the ratio of any two square numbers. Therefore
(by Thm.2 Remarks) it follows that their sides are incommensurable, i.e. AB is
incommensurable with AH.

Q.E.D.

254
THEOREM 6: The parts of a magnitude cut into mean and extreme ratio are
incommensurable.

Take any straight line AB and cut it

S in mean and extreme ratio at S.
A B
Hence BA : AS = AS : SB.

I say that AS and SB are incommensurable with each other.

If possible, suppose AS and SB are commensurable, and therefore have a numerical

ratio. Say AS : SB = n : m, where n and m are the least numbers in that ratio, and so
are prime to each other (Ch.7, Thm.7, Remark 2).

[1] Since AS : SB = n : m (assumed)

thus AS : AS + SB = n : n + m (Ch.5, Thm.15, Remarks)
or AS : AB = n : n + m (AS + SB is AB)
but AS : AB = SB : AS (inverse of given)
so SB : AS = n : n + m (each ratio is the same as AS : AB)
but SB : AS = m : n (inverse of the assumed proportion)
so m:n = n:n+m (each ratio is the same as SB : AS)

[2] Since m and n are the least numbers in their ratio, hence they measure the
numbers in the same ratio with them (Ch.7, Thm.7). So the antecedent measures the
antecedent in the proportion above, i.e. m measures n. But since m and n are prime
to each other, they have no common measure but 1. So m = 1.

[3] So 1:n = n:n+1 (replacing m with 1 in Step 1 proportion)

hence n × n = n + 1 (product of means = product of extremes)
But this is impossible for any number n.
Consider: 1×1 < 1+1
but 2×2 > 2+1
and 3×3 > 3+1
and 4×4 > 4+1
and 5×5 > 5+1
etc.

[4] Since it is impossible for n to be a number, it is impossible to express mean and

extreme ratio as the ratio between two numbers, n and m. Hence AS is
incommensurable with SB.

Q.E.D.

255
THEOREM 7: Two magnitudes each commensurable with the same magnitude
are also commensurable with each other.

Given: A and C are commensurable

A B and C are commensurable
B
Prove: A and B are commensurable
C

Since A and C are commensurable, they have a common measure, say W, that goes into
each of them some number of times. Say
A = 5W
C = 7W
Since B and C are commensurable, they have a common measure, say Z, that goes into
each of them some number of times. Say
B = 3Z
C = 4Z

[1] Now A : 5W = B : 3Z (A = 5W, and B = 3Z)

[2] thus 7 · 4 · A : 7 · 4 · 5W = 7 · 4 · B : 7 · 4 · 3Z

All we have done in this step is multiply every term in the proportion of Step 1 by 7 × 4,
i.e. by 28. Doing this does not destroy the proportion. A few inconsequential
rearrangements of the multipliers gives us

[3] 28A : 4 · 5 · 7W = 28B : 7 · 3 · 4Z

And since 7W = C (as we set out in the beginning), and 4Z = C, we can replace those two
terms in the proportion with C:

so 28A : 4 · 5 · C = 28B : 7 · 3 · C
or 28A : 20C = 28B : 21C
thus 28A : 28B = 20C : 21C (alternating the proportion)
i.e. A : B = 20 : 21

So A and B have to each other the ratio that a number has to a number, and so they are
commensurable.

Q.E.D.

256
THEOREM 7 Remarks:

1. In particular, if one magnitude is commensurable with another, then it is also

commensurable with any measure of it. That is, if M is commensurable with 35Q, then
M is also commensurable with Q. For
M is commensurable with 35Q (given)
Q is commensurable with 35 Q (since Q measures both)
thus M and Q must be commensurable (both being commensurable with 35Q)

2. We have seen many relationships which are transitive, i.e. which are such that if two
things have that relationship to a third thing, they also have it to each other. This is true
about the relationships of (a) equality, (b) congruence, (c) parallelism, (d) sameness of
ratio, (e) similarity, and now (f) commensurability.

THEOREM 8: Any magnitude commensurable with one of two incommensurable

magnitudes must be incommensurable with the other one.

Z
Given: A and B are incommensurable with each other
A
Z is commensurable with A

B Prove: Z is incommensurable with B

If Z were commensurable with B,

then A and B would both be commensurable with Z (it is given that A is),
and so A and B would be commensurable with each other (by Thm.7),
but A and B are not commensurable with each other (given),
and therefore neither can Z be commensurable with B.

Q.E.D.

THEOREM 8 Remarks:

Now it is also clear that if a straight line is commensurable with an irrational line, it is
irrational too.

257
THEOREM 9: The sum of two incommensurable magnitudes is incommensurable
with each of them – and the difference of two incommensurable magnitudes is
incommensurable with each of them.

Given: A is incommensurable with B A

B
Prove: A + B is incommensurable with A
A + B is incommensurable with B

If A were commensurable with A + B, then they have some common measure M going
into each some number of times, say

A + B = 5M
and A = 3M
thus B = 2M (subtracting equals from equals)

And so M also measures B, not just A. Thus A will be commensurable with B. But it is
given that it is not. Therefore neither can A be commensurable with A + B.
Likewise B cannot be commensurable with A + B.

Q.E.D.

Similarly with the differences, if A is incommensurable with B, then

A – B is incommensurable with A
A – B is incommensurable with B

For suppose, if possible, that A and A – B were commensurable, and so each is measured
by a common measure, M, some definite number of times, say

A = 12M
A – B = 8M
thus B= 4M (subtracting equals from equals)

And so again A and B would have to have M as a common measure, which is contrary to
what is given about them. Therefore A must be incommensurable with A – B.
Likewise B is incommensurable with A – B.

Q.E.D.

258
THEOREM 10: Either of two commensurable magnitudes is commensurable
with their sum, and also with their difference.

Given: A and B are commensurable.

A
Prove: A is commensurable with A + B
B
B is commensurable with A + B
A is commensurable with A – B
B is commensurable with A – B

Since A and B are commensurable, they have a common measure, M,which goes into
each some number of times, say
A = 7M
B = 5M
Thus A + B = 12M, and so their sum is also measured by M, the common measure of A
and B, and so A + B is commensurable with A and with B.
And A – B = 2M, and so their difference is also measured by M, and thus is
commensurable with A and also with B.

Q.E.D.

259
THEOREM 11: How to make an infinity of magnitudes, every one of which is
incommensurable with all the rest.

A
Given: Two incommensurable magnitudes to start with,
A and B, such as the side and diagonal of a square. B

Find: A way to generate from them an infinity of magnitudes every one of which is
incommensurable with every other.

If we make the following magnitudes:

A + 1B
A + 2B
A + 3B … etc.

I say that every one of these is incommensurable with all the rest. Consider, for example,
A + 5B and A + 12B. I say they are incommensurable.

For, if possible, suppose that

A + 5B is commensurable with A + 12B
then A + 5B is commensurable with 7B (i.e. with their difference; Thm.10)
then A + 5B is commensurable with B (i.e. with a measure of 7B; Thm.7)
then A + 5B is commensurable with 5B (i.e. with a multiple of B)

Now, since 5B is commensurable with A + 5B, it follows that 5B is also commensurable

with their difference, namely with A (by Thm.10).

thus A is commensurable with 5B

so A is commensurable with B

since B is a measure of 5B, and anything commensurable with 5B must also be

commensurable with anything that measures it (see Thm. 7, Remark).
But it is impossible for A to be commensurable with B, since it is given that they
are not commensurable. Therefore our initial supposition, namely that A + 5B and A +
12B are commensurable, is impossible. Therefore A + 5B and A + 12B are
incommensurable.
Likewise any two magnitudes in our infinite list are incommensurable with each
other.

Q.E.D.

260
THEOREM 11 Remarks:
1

1
Here is another way to set out an 2
1
infinity of lines all incommensurable 3
with each other. Set out the unit line
1, make another line equal to it at 4
1
right angles and join the endpoints.
Since this hypotenuse is in fact the 5
diagonal of the square on the unit
10 6 1
line, the square on that hypotenuse
9 7
equals two times the unit square. 8
Call this line side of square two. 1
1
Now draw another unit line at right
angles to the side of square two, and 1
1
join its endpoint to the other end of 1
the side of square two. By the
Pythagorean Theorem, the square on
this new hypotenuse equals the
square on 1 (i.e. 1 square unit) plus the square on the side of square two (i.e. 2 square
units), and so it is equal to 3 square units of area. Call this line side of square three.
Repeat the process and you will get the hypotenuses which are the sides of squares 4, 5, 6
etc., having 4 square units of area, 5 square units of area, and so on.
Looking at the hypotenuses, the squares on any two of them are as the numbers
under the radical signs, which are simply all the numbers (if the process is carried on
indefinitely). Now attend only to the hypotenuses with prime numbers. The squares on
any two of these will be as those prime numbers – but no two prime numbers have the
same ratio as two square numbers (Ch.7, Thm.24). So the squares on any two prime-
numbered hypotenuses will not be as a square number to a square number, and so those
hypotenuses themselves will be incommensurable (Thm.2, Remark 2). But we can make
as many such hypotenuses as there are prime numbers, and there is an infinity of those
(Ch.7, Thm.15). Therefore we can make an infinity of straight lines, each one of which is
incommensurable with all the others.

261
THEOREM 12: How to make an infinity of straight lines commensurable with a
given line, but falling between two given lengths.

R V P
A
I N

Suppose I give you a straight line G, and also two other lengths AR and AP differing
by however much or little you please. Can you find a way to make an infinite series of
different lengths falling between AR and AP, which are all commensurable with G?
Absolutely.

[1] First notice that however small RP may be, some multiple of it will exceed G.
Suppose 57 RP > G
Now divide G into more than 57 equal parts, say 5700 equal parts (Ch.6, Thm.8), and
call each one F. So G = 5700 F.
So now 57 RP > 5700 F
hence RP > 100 F
So if we now multiply F until it first exceeds AR, we will have to add almost another
100 F’s before we reach P. In other words, there are plenty of multiples of F between
AR and AP.

[2] So let AI and AN both be multiples of F falling between AR and AP in length.

Bisect IN at V.

[3] Since AI and AN are both commensurable with G (all three are multiples of F),
hence AI is sommensurable with AN (Thm.7)
so AI is commensurable with IN (their difference; Thm.10)
but IV is commensurable with IN (IV is half IN)
so AI is commensurable with IV (Thm.7)
thus AI is commensurable with AV (their sum; Thm.10)
but AI is commensurable with G (both are multiples of F)
so AV is commensurable with G

[4] So AV lies between AR and AP in length, and it is commensurable with G, the given
line. If we now bisect IV at Q, AQ will likewise lie between AR and AP and be
commensurable with G by similar reasoning, and their is no limit to how many lines we
can make like this.

Q.E.F.

262
THEOREM 13: Between any two irrational lengths there is an infinity of
rational lengths; and between any two rational lengths there is an infinity of irrational
lengths, each incommensurable with all the others.

Q RT
L P

[1] Set out the unit length U, and any two irrational lengths LQ and LP, with as small
a difference between them as you like. By Thm.12, we can construct an infinity of
straight lines falling between LQ and LP in length, but all commensurable with U, and
hence rational. So the first part of the Theorem is proved.

[2] Take any two rational lengths LR and LT (falling between LQ and LP if you
wish), with as small a difference between them as you like. Set out any irrational line D
(such as the diagonal of the square on U).
Consider the length D + U. Since D is incommensurable with U, hence their
sum is also incommensurable with U and hence irrational (Thm.9). But by Thm.12, we
can make an infinity of lines between LR and LT in length, all commensurable with D
+ U, and thus all irrational. Choose any one of these and call it X1.

[3] Now consider the length D + 2U. It must likewise be irrational, since D is
incommensurable with U and hence with all its multiples such as 2U, and so 2U is
incommensurable with their sum D + 2U (Thm.9). Therefore U, being commensurable
with 2U, is also incommensurable with D + 2U (Thm.8). Therefore D + 2U is
irrational.
Also, D + 2U is incommensurable with D + U (Thm.11). But again, using
Thm.12, we can make an infinity of lines between LR and LT in length, all
commensurable with D + 2U, all of them therefore irrational and incommensurable with
D + U. Choose any of these, and call it X2.
Because X1 is commensurable with D + U, but X2 is commensurable with D
+ 2U, it follows that X1 and X2 are incommensurable with each other.

[4] We can continue with D + 3U, and make X3 between LR and LT, another irrational
length, again incommensurable with both X1 and X2. And so on forever.

Q.E.D.

263
THEOREM 14: If a straight line is divided into two parts incommensurable with
each other, it is impossible to divide it into two other parts commensurable with the
original two parts.

B
A C
X

Given: AC is divided at B, and AB is incommensurable with BC

Prove: It is impossible to divide AC at another point X so that AX is commensurable

with AB and XC is commensurable with BC.

For let it be given that AX is commensurable with AB.

Then I say that XC must be incommensurable with BC.

For AX is commensurable with AB (given)

and so the difference between them, XB, must be commensurable with each of them,

so XB is commensurable with AB (Thm.10)

Now since AB and BC are incommensurable (given), and XB is commensurable with one
of them, namely AB, it follows that

XB is incommensurable with BC (Thm.8)

and so the sum of these two must be incommensurable with each of them (Thm.9),

i.e. XC is incommensurable with BC.

And so it is impossible to divide the line AC at any point other than B so as to get two
parts that are each commensurable with AB and BC respectively.

Q.E.D.

THEOREM 14 Remarks:

Obviously we could make X the same distance from A that point B is from C, and thus
AX would be commensurable with BC (being equal to it), and XC with AB (being equal
to it). But then we would not have really divided the line into two new kinds of parts.

264
THEOREM 15: It is possible to contain a rational area with irrational lines, and
also to contain an irrational area with rational lines.

Begin with the square on the unit line, the unit square. Form the
square on its diagonal. Since this diagonal is incommensurable
with the unit side (Thm.4), it is irrational. So the square on the
diagonal of the unit square is contained by four irrational lines.
And yet its area is rational, being double (and so commensurable
with) the area of the unit square. So it is possible to contain a
1 rational area with irrational lines.
H R
Begin again with the square on the unit line, and extend one
of its sides, AB, to D, so that DA = AB (and so DB is
called 2 in reference to the unit length). Now make
equilateral triangle DHB on DB. Thus AH is both the
height of rDHB, and the extension of one side of the unit
square (namely EA). Complete the rectangle AHRB. D B
A

Now, £EABC : rect. AHRB = EA : AH

(Ch.6, Thm.1)
E 1 C

But AH is the height of the equilateral triangle, and so it is incommensurable with its
side, DB (Thm.5). Thus height AH is also incommensurable with AB, half the side.
Thus AH is also incommensurable with EA, which is equal to AB. So EA is
incommensurable with AH, and thus EA : AH is not a numerical ratio.

Thus £EABC : rect AHRB is not a numerical ratio.

But rectangle AHRB is equal to the equilateral triangle DHB. Therefore

£EABC : rDHB is not a numerical ratio,

which is to say that these two areas are incommensurable with each other. But since
square EABC is the unit square, hence rDHB is an irrational area. And yet it is
contained by rational sides, each of its sides being double (and thus commensurable with)
the unit length. So it is possible to contain an irrational area with rational lines.

Q.E.D.

265
 Chapter Nine

Basic Solid Geometry

 DEFINITIONS

1. A SOLID is whatever has length, width, and depth.

P
2. You have a STRAIGHT LINE PERPENDICULAR TO A
PLANE if it is perpendicular to all the straight lines it stands
on in that plane. B
For example, PR stands on AB, and is perpendicular
R
to it. And if PR is perpendicular to all such lines in the plane A
passing through R, then PR is perpendicular to the plane.

3. You have PERPENDICULAR PLANES if every straight

A
line in one of them that is perpendicular to their intersection
is also perpendicular to the other plane.
S For example, let two planes intersect along SX. If
every straight line AB drawn perpendicular to SX in one of
B the planes is also perpendicular to the other plane, then the
X two planes are perpendicular to each other.

B
4. Consider a straight line AB that passes through some plane
at point A, and is not perpendicular to the plane, but leans over
somewhat. How much does it lean? If we choose any point B
along it, and BP falls perpendicular to the plane, then the
A P
INCLINATION OF THE STRAIGHT LINE TO THE PLANE
is angle BAP.

5. The INCLINATION OF A PLANE TO A PLANE is the

angle between two straight lines, one drawn in each plane,
A and both drawn perpendicular to the line of intersection and
from the same point on it.
B X For example, angle ABC is formed by two
S perpendiculars to SX, each of them drawn in one of the two
C planes. So angle ABC is the inclination of the planes.

267
 6. PARALLEL PLANES are those which never meet, no matter how far they are
extended.

B D
A 7. A PLANAR SOLID ANGLE is formed by three or more
planes meeting at a point.
For example, in a cube, one angle of it is formed by three
right-angled faces, namely BAC, BAD, CAD.

8. A SOLID FIGURE is a figure contained by one or more surfaces.

9. A SPHERE is a solid figure contained by one surface which is at all

points equidistant from one point within called the CENTER. If a
semicircle is rotated all the way around on its diameter once, the solid
figure it describes is a sphere. The center and diameter of the
semicircle are also the center and diameter of the sphere. Any straight
line drawn through the center of a sphere and stopping at the surface of
the sphere in each direction is a DIAMETER of the sphere.

10. A RIGHT CONE is a solid figure described by rotating a right V

triangle all the way around once about one of the sides forming the
right angle. The side about which the triangle was rotated is called
the AXIS of the cone, and the circle described by the other side of
the right angle is the BASE of the cone. The point at which the
axis meets the hypotenuse of the original triangle is the VERTEX
of the cone.
Note: this kind of cone is called a right cone because its
axis is at right angles to its base. There are other kinds of cones
X
called oblique cones, but since these will not come up in this book,
the simple term cone will always refer to a right cone.

A 11. A RIGHT CYLINDER is a solid figure described by rotating a

rectangle all the way around once about one of its sides. The side about
which the rectangle was rotated is called the AXIS of the cylinder, and
the two circles described by the two sides of the rectangle adjacent to
the axis are the BASES of the cylinder.
Note: this kind of cylinder is called a right cylinder because its
axis is at right angles to its bases. There are other kinds of cylinders
X called oblique cylinders, but since these will not come up in this book,
the simple term cylinder will always refer to a right cylinder.

268
12. SIMILAR CONES or SIMILAR CYLINDERS are those in which the axes and the
diameters of the bases are proportional.

13. A POLYHEDRON is a solid figure contained by four or more rectilineal plane

figures. Note: the plural for polyhedron is often written polyhedra. I prefer to say
polyhedrons.

14. SIMILAR POLYHEDRONS are those whose faces are similar, each to each, and
similarly arranged.
By “similarly arranged” I mean that if any two faces in one solid meet each
other, then the two correspondingly similar faces in the other solid also meet each other,
forming an edge; also, if a solid angle in one solid is convex, then the corresponding
solid angle in the other solid is also convex, but if concave, then concave.
SIMILAR AND EQUAL POLYHEDRONS are similar polyhedrons whose
corresponding faces are equal in size. These can also be called congruent polyhedrons.

15. A PYRAMID is a polyhedron contained by a plane and three or more

triangles drawn down to it from one point. The portion of the plane bounding
the pyramid is called its BASE, whereas the point is called its VERTEX.

16. A PRISM is a polyhedron contained by two congruent and parallel

polygons similarly oriented, and all the parallelograms joining their
corresponding sides. The two identical and parallel polygons are the BASES
of the prism.

17. A PARALLELEPIPED is a prism whose bases are parallelograms.

269
BASIC PRINCIPLES OF SOLID GEOMETRY

1. If any two points of a straight line lie in a plane, the whole straight
line lies in that plane.

2. If two planes intersect, their intersection is a straight line, and they

have no other points in common.

3. Any plane can be extended as far as we please in any of its

directions.

4. Any plane can be rotated about any straight line that lies within it.

270
 THEOREMS

THEOREM 1: Any three points not lying in a straight line lie only in one plane,
and every triangle lies only in one plane.
B

Consider any three points A, B, C which do not lie

in one straight line. It is impossible for all three of
these points to lie in more than one plane.

If possible, suppose A, B, C all lie in two distinct A

planes: plane Q and also plane Z. C

Now, since A and B both lie in plane Q, therefore straight line AB lies in plane Q (Princ.
1). And since A and B both lie in plane Z, therefore straight line AB lies in plane Z
(Princ. 1). Therefore plane Q and plane Z have line AB in common, i.e. they intersect
along that straight line. But then they have no other points in common, beyond those
lying in a straight line with AB (Princ. 2). Therefore point C, not lying in line with AB
(given), is not common to planes Q and Z. And thus it is not possible for A, B, C to lie
all in plane Q, and also all in plane Z.

Again, the whole triangle ABC lies only in one plane. For any plane containing
all of triangle ABC must also contain its three vertices A, B, and C. But we have just
showed that there is only one such plane. Therefore the whole of any triangle lies only in
one plane.

Q.E.D.

THEOREM 1 Remarks:

1. One point can have many straight lines passing through it, but any two points lie only
in one straight line. Similarly, two points can have many planes passing through them,
but any three points (if they are not in a straight line) lie only in one plane.
It is obvious that any two points can in fact have a straight line passing through
them. Is it also obvious that any three points can have a plane passing through them?
Yes. Say the points are A, B, C. Join AB, and pass any plane through AB. Now rotate
the plane around AB like a hinge until it hits C, and the result is a plane containing points
A, B, and C.

271
 So any three points do lie in one plane. But a fourth point might not lie in the
same plane.
2. Obviously, if you have 3 points in a straight line, there is an infinity of planes that
contain those 3 points. Pass any plane through the straight line containing the 3 points,
and this plane will contain all 3 points.

3. If it were not obvious enough by itself, it is now obvious that One and only one plane
can be drawn through a given straight line and a given point not on that straight line.
For example, only one plane goes through straight line AB and point C – otherwise, more
than one plane would contain the three points A, B, C.

4. If it were not obvious enough by itself, it is now obvious that One and only one plane
can be drawn through a given pair of intersecting straight lines. For example, only one
plane goes through the straight lines AB and BC – otherwise, more than one plane would
contain the three points A, B, C.

THEOREM 2: One and only one plane passes through any pair of parallel
straight lines, and any straight line joining any two points on the parallels also lies in
that plane.
B
P
Given: AB and CD, a pair of parallel straight lines, A
with P and R being random points on each of them.
D
Prove: One and only one plane passes through both
R
AB and CD, and PR lies in that plane. C

Put a pencil down on the table, and imagine it indicating a straight line going on forever
in both directions, say North and South. Now hold a pen over the pencil, but pointing
East and West. These two straight lines will never intersect each other, and yet we do not
call them “parallel.” Why? Because they are not in the same plane. It is especially
interesting that even in the same plane two straight lines can be so oriented that they will
never meet – there is in fact only one orientation you can give a straight line to make it
parallel to another. And thus “parallel” means not only “never meeting,” but also “in one
plane.” Therefore the first part of the theorem, namely that any two straight lines that are
parallel must lie in the same plane, is really self-evident. It is part of what “parallel”
means.
It is also clear that the parallels AB and CD lie only in one plane – it is not
possible for more than one plane to contain them both. For supposing it were so, then
two distinct planes would contain points A, B, C, even though these do not lie in a
straight line with each other, which is impossible (Thm.1). Thus it is impossible for more
than one plane to contain a given pair of parallel straight lines.
And since points P and R both lie in the plane containing the parallels AB and
CD, therefore the line PR lies in that plane, too (Princ. 1).

Q.E.D.

272
THEOREM 2 Remarks:

A pair of lines that never meet, but are not in the same plane as each other, are called
skew lines.

THEOREM 3: If a straight line is perpendicular to two intersecting straight lines

at the point where they meet, then it is perpendicular to the plane in which they lie.

Suppose AB and CD meet each other at M, P

and PM is perpendicular to both AB and CD.
Then I say PM is perpendicular to the plane
A C
passing through AB and CD.
G
M H
Cut off MA = MD = MB = MC.
Join AD, BC, AP, DP, BP, CP. D B
In the plane of AB and CD, draw GMH
through M at random, cutting AD and BC.
Join GP, HP.

I say that PM is at right angles to the line GMH drawn through M randomly in the plane.

[1] Since ∠AMD and ∠BMC are vertical and are contained by equal lines, hence
rMAD ≅ rMBC, so ∠MAD = ∠MBC.

[2] Now ∠MAG = ∠MBH (being the same as ∠MAD and ∠MBC)
but ∠AMG = ∠BMH (being vertical)
and AM = MB (we made them so)
so rMAG ≅ rMBH (Angle Side Angle)

[3] Again PA = PB (rMAP ≅ rMBP by S-A-S)

and PD = PC (rMDP ≅ rMCP by S-A-S)
and AD = BC (rMAD ≅ rMBC; Step 1)
so rPAD ≅ rPBC (Side Side Side)
so ∠PAD = ∠PBC

[4] Again ∠PAG = ∠PBC (being the same as ∠PAD and ∠PBC)
and PA = PB (rMAP ≅ rMBP)
and AG = BH (rMAG ≅ rMBH; Step 2)

273
 so rPAG ≅ rPBH (Side Angle Side)

[5] Now PG = PH (rPAG ≅ rPBH; Step 4)

and MG = MH (rMAG ≅ rMBH; Step 2)
and PM is common (to rPMG and rPMH)
so rPMG ≅ rPMH (Side Side Side)
so ∠PMG = ∠PMH
But these equal angles are adjacent. Hence PM is at right angles to GMH.

[6] Since PM is thus at right angles to any straight line drawn through M in the plane
of AB and CD, therefore PM is perpendicular to that plane.

Q.E.D.

THEOREM 3 Remarks:

1. If GH is drawn through M so that it does not cut AD and BC, then it will cut AC and
DB, and we use them for the proof instead.

2. A kind of converse to this Theorem is: All perpendiculars to one point on a straight
line lie in one plane. All the perpendiculars to PM drawn from M lie in the plane of AB
and CD.

3. Prove that the line drawn perpendicular to a plane from a point above it is the shortest
straight line that can be drawn from that point to the plane.

4. Prove that only one straight line can be drawn from a given point perpendicular to a
given plane.

THEOREM 4: If one of two parallels is perpendicular to a plane, so is the other.

Given: AB is parallel to CD, and AB is perpendicular A

to plane X. C

Prove: CD is also perpendicular to plane X.

B
D
Join BD.
Draw DE (in plane X) perpendicular to BD, and cut X
off DE equal to AB. E

274
Join BE, AE, AD.

[1] Now, rABD ≅ rEDB (SAS)

[2] Thus AD = BE (see Step 1)

but DE = AB (we made them equal)
and AE is common (to rABE and rEDA
so rABE ≅ rEDA (SSS)

[3] Now ∠ABE = ∠EDA (by Step 2)

but ∠ABE is right (since AB is given perpendicular to plane X)
so ∠EDA is right

[4] So ED is perpendicular to DA
but ED is perpendicular to DB (by construction)
so ED is perpendicular to the plane through DA and DB (Thm.3), i.e. the
plane containing points A, B, D.

[5] Now, there is only one plane containing points A, B, D (Thm.1), but the plane
containing parallels AB and CD (Thm.2) contains points A, B, D, and therefore the plane
containing points A, B, D is the same as the plane containing parallels AB and CD. Thus
ED is perpendicular to the plane of the parallels, i.e. to the plane containing triangle
BDC. Therefore ∠EDC is a right angle (see Def.2).

[6] So CD is perpendicular to DE (Step 5)

and CD is perpendicular to BD (∠ABD is right, and CD is parallel to AB)
so CD is perpendicular to two straight lines intersecting in plane X, and thus
CD is perpendicular to plane X (Thm.3).

Q.E.D.

THEOREM 5: If two straight lines are perpendicular to the same plane, they are
parallel.

Given: AB is perpendicular to plane X.

E
CD is perpendicular to plane X. A
C

G
Prove: AB is parallel to CD.
B D
Suppose, if possible, that AB is not parallel to CD.
K
Then since B, D, C are all in one plane, draw BE in
this plane parallel to CD. Therefore BE is X

perpendicular to plane X (Thm.4).

275
 Now, A, B, E are all in one plane. Let the intersection of their plane with plane X
be called GK.
Since BE is perpendicular to plane X, therefore ∠GBE is right.
Since AB is perpendicular to plane X, therefore ∠GBA is right.
Thus ∠GBE = ∠GBA, i.e the whole is equal to the part, which is impossible.
Thus our initial assumption was impossible – AB in fact is parallel to CD.

Q.E.D.

THEOREM 5 Remarks:

From this it is clear that You can't have two straight lines perpendicular to the same point
on a plane (except, of course, on opposite sides of the plane, i.e. one above it and one
below it).

THEOREM 6: How to drop a straight line perpendicular to a plane from a given

point above it.

Suppose P is the point above our plane X. Choose any P

straight line RM in plane X. Thus P, R, M are in one
and only one plane – drop PL perpendicular to RM in
that new plane (as we learned to do in Ch. 1).
B R
Now draw LA perpendicular to RLM in plane
T
X. Thus P, L, A are in one and only one plane. Drop A
L
PT perpendicular to LA in that new plane. X
E
M
I say that PT is perpendicular to plane X.

In plane X, draw BTE parallel to RLM.

Now RLM is perpendicular to plane PLT, since it is perpendicular to both PL and

LT by construction (Thm.3). Thus BTE, parallel to RLM, is also perpendicular to plane
PLT (Thm.4).
Thus BT is perpendicular to all lines through T in plane PLT (Def.2).
So BT is perpendicular to PT.
But LT is perpendicular to PT, by construction.
So PT is perpendicular to BT and LT, which both lie in plane X.
Therefore PT is perpendicular to plane X (Thm.3).

Q.E.F.

276
THEOREM 7: How to set up a straight line perpendicular to a plane from a
given point on it.
R T

Given: Point P in plane X.

Make: A straight line perpendicular to plane X at P.
X

Choose any point R at random above plane X, and drop RL L P

perpendicular to plane X (Thm.6).
In the plane of R, L, P draw PT parallel to RL.

Now RL is perpendicular to plane X (by construction)

and PT is parallel to RL (by construction)
so PT is perpendicular to plane X (Thm.4)

Q.E.F.

THEOREM 8: Any plane containing a straight line perpendicular to another

plane is itself perpendicular to that plane.

E C A H
Given: AB is perpendicular to plane X,
EHKG is a containing through AB
X
Prove: Plane EHKG is perpendicular to plane X.
G R B K

Choose any random point R on GK, the intersection of plane EHKG and plane X.
Draw RC perpendicular to GK in plane EHKG.
We already know that AB is also perpendicular to GK, since AB is perpendicular to all
straight lines through B in plane X.

Since AB is perpendicular to plane X (given)

and RC is parallel to AB (RC and AB, in one plane, are ⊥ to GK)
thus RC is perpendicular to plane X (Thm.4)

For the same reasons, any straight line (in plane EHKG) drawn perpendicular to GK will
be perpendicular to plane X. Therefore plane EHKG is perpendicular to plane X (Def. 3).

Q.E.D.

277
THEOREM 8 Remarks:

From this it is clear how to Drop a plane perpendicular to a T P

given plane from a given straight line above the given plane,
and how to Set up a plane perpendicular to a given plane
upon a given straight line in the given plane.
Given a plane and a straight line in it, to set up a
Z R
plane on that line perpendicular to the given plane: (1) pick
any 2 points R and Z on the given line, (2) set up ZT and RP
perpendicular to the given plane (Thm.7), (3) since ZT and N
RP are perpendicular to the same plane, therefore they are L
parallel (Thm.5), and thus are in one and only one plane
together (Thm.2), (4) since the plane containing them passes
through lines that are at right angles to the given plane,
therefore their plane is at right angles to the given plane S V
(Thm.8).
Given a plane and a straight line above it, to construct the plane which contains
that line and is perpendicular to the base plane: (1) pick any 2 points L and N on the
given line, (2) drop LS and NV perpendicular to the given plane (Thm.6). The rest of
the proof is the same as above.

THEOREM 9: If three straight lines are not all in one plane, and yet one of them
is parallel to the other two, then the other two are also parallel to each other.

A G
B
Given: AB, CD, EF are three lines not all
in one plane. R F
AB is parallel to EF. E

CD is parallel to EF.
D
Prove: AB is parallel to CD. K
C

[1] Choose R at random on EF.

Draw RG perpendicular to EF in the plane of parallels AB and EF.
Draw RK perpendicular to EF in the plane of parallels CD and EF.

[2] Now R, G, K are all in one plane (Thm.1).

And since ER is perpendicular to both KR and RG in the plane of R, G, K,
therefore ER is perpendicular to the plane of KRG (Thm.3).

278
[3] Now AG is parallel to ER (given)
and ER is perpendicular to plane K, R, G (Step 2)
so AG is perpendicular to plane K, R, G (Thm.4)

[4] But CK is parallel to ER (given)

and ER is perpendicular to plane K, R, G (Step 2)
so CK is perpendicular to plane K, R, G (Thm.4)

[5] Since AG and CK are both perpendicular to the same plane, namely the plane of
points K, R, G, therefore AG and CK are parallel to each other (Thm.5).

Q.E.D.

THEOREM 10: If one straight line is perpendicular to two planes, the planes
are parallel.

X Given: AB is perpendicular to
K
A plane X and to plane Z.

R Prove: Plane X is parallel to

Z
plane Z.
B
G

If possible, suppose planes X and Z are not parallel, but eventually meet each other – let
KG be the line of their intersection. Pick point R at random on KG.
Join AR.
Join BR.

[1] Now, A and R are both in plane X, and so line AR is in plane X.

And B and R are both in plane Z, and so line BR is in plane Z.

[2] Since BA is perpendicular to plane X (given), therefore any straight line in plane
X passing through A is at right angles to BA. But AR is in plane X (Step 1), and it passes
through point A. Therefore AR is at right angles to BA.
Thus ∠BAR is right.

279
[3] Since AB is perpendicular to plane Z (given), therefore any straight line in plane
Z passing through B is at right angles to AB. But BR is in plane Z (Step 1), and it passes
through point B. Therefore BR is at right angles to AB.
Thus ∠ABR is right.

[4] Thus ABR is a triangle two of whose angles are right angles – which is
impossible. Therefore our initial assumption was impossible, namely that planes X and Z
should meet. Therefore planes X and Z never meet – and so they are parallel.

Q.E.D.

THEOREM 11: If two intersecting lines in one plane are parallel to two
intersecting lines in another plane, the two planes are parallel.

X
B
A C Given: AB and BC intersect in plane X,
DE and EF intersect in plane Z,
AB is parallel to DE,
BC is parallel to EF.
E Z
D G F Prove: Plane X is parallel to Plane Z
H K

[1] Drop BG perpendicular to plane Z (Thm.6).

In plane Z, draw GH parallel to ED, and GK parallel to EF.

[2] Since BG is perpendicular to plane Z,

thus ∠BGH is right
and ∠BGK is right

[3] But GH is parallel to DE (we made it so)

and AB is parallel to DE (given)
so AB is parallel to GH (Thm.9)

Thus ∠ABG is right (since ∠BGH is right; Step 1)

[4] Again GK is parallel to EF (we made it so)

and BC is parallel to EF (given)
so BC is parallel to GK (Thm.9)

280
 Thus ∠CBG is right (since ∠BGK is right; Step 1)
[5] Therefore BG is at right angles to both AB and BC (Steps 3 and 4), which are two
lines intersecting in plane X. Therefore BG is at right angles to plane X (Thm.3). But
BG is at right angles to plane Z (we dropped BG at right angles to plane Z; Step 1).
Therefore planes X and Z have a common perpendicular, namely BG, and thus these two
planes are parallel to each other (Thm.10).

Q.E.D.

THEOREM 12: A pair of intersecting lines parallel to another pair of

intersecting lines in another plane will contain the same angle (or supplementary
angles).

X Given: AB and BC intersect in plane X,

Z
B
DE and EF intersect in plane Z,
T
E AB is parallel to DE,
BC is parallel to EF.
C
F

A
Prove: ∠ABC = ∠DEF.
D
Cut off AB = DE, and cut off BC = EF.
Join AC, DF, AD, BE, CF.

[1] AB and DE are parallel (given), and so they are in one plane.
But we have just cut off AB and DE equal to each other.
Therefore the lines joining their endpoints are also parallel and equal (Ch.1).
i.e. AD and BE are parallel and equal to each other.

[2] BC and EF are parallel (given), and so they are in one plane.
But we have just cut off BC and EF equal to each other.
Therefore the lines joining their endpoints are also parallel and equal (Ch.1).
i.e. BE and CF are parallel and equal to each other.

[3] Since AD is parallel and equal to BE (Step 1)

and CF is parallel and equal to BE (Step 2)
thus AD is parallel and equal to CF (Thm.9)

And so the lines joining their endpoints are also parallel and equal (Ch.1), i.e. AC is
parallel and equal to DF.

281
[4] Now AB = DE (we cut them off equal)
and BC = EF (we cut them off equal)
and AC = DF (Step 3)
thus ΔABC ≅ ΔDEF (Side-Side-Side)
so ∠ABC = ∠DEF

Q.E.D.

THEOREM 12 Remarks:

If we extend FE to T, then TE is parallel to BC, too, but ∠TED will not be equal to
∠ABC (unless ∠TED and ∠FED are both right angles). Still, ∠TED is supplementary to
∠DEF, and therefore also supplementary to ∠ABC.

THEOREM 13: If one plane intersects two parallel planes, the two lines of
intersection are parallel.
B
Given: Plane X is parallel to plane Z, each is cut X
by plane ABCD, namely at AB and CD. A

Prove: AB is parallel to CD.

Since planes X and Z never meet in any direction, C

a line contained in one can never meet a line Z
contained in the other. Therefore AB can never D
meet CD.
But since AB and CD are both in the one plane ABCD (given), therefore they are
non-meeting straight lines in the same plane, and therefore they are parallel to each other.

So AB is parallel to CD.

Q.E.D.

282
THEOREM 14: Straight lines cut by parallel planes are cut in the same ratios.

X A
C

Given: AB and CD are cut by three parallel planes

X, Y, Z, cutting them off at A, K, B and C, E, D.
K G
Y
E
Prove: AK : KB = CE : ED.

Z B D

[1] Join AD, AC, DB, GE, GK.

[2] Since A and C are both in plane X, thus AC is in plane X.

Since A and C are both in the plane of A, C, D, thus AC is in the plane of A, C, D.
Therefore AC is the line of intersection of plane X and plane A, C, D.

[3] Likewise EG is the intersection of plane Y and plane A, C, D.

GK is the intersection of plane Y and plane A, B, D.
DB is the intersection of plane Z and plane A, B, D.

[4] Thus AC is parallel to EG, being intersections of plane ACD

with the parallel planes X and Y. (Thm.13)

and GK is parallel to DB, being intersections of plane ABD

with the parallel planes Y and Z. (Thm.13)

[5] And so, since GK is parallel to DB (Step 4) in ΔABD, thus

AK : KB = AG : GD
and since AC is parallel to EG (Step 4) in ΔACD, thus
CE : ED = AG : GD
and since in these two proportions two ratios are the same as a third, it follows that they
are the same as each other, i.e.
AK : KB = CE : ED

Q.E.D.

283
THEOREM 15: The intersection of two planes each perpendicular to a third
plane is a straight line perpendicular to the third plane.
A
N B

Given: Planes A and B, both perpendicular

to plane X, and intersecting each other along
C PN, P being in plane X.
G
E P Prove: PN is perpendicular to plane X.
X
D

[1] Since plane A is perpendicular to plane X, and CD is their intersection, therefore

every line drawn in plane A perpendicular to CD is also perpendicular to plane X (Def.
3). Therefore the straight line drawn from P (in plane A), perpendicular to CD, is
perpendicular to plane X.

[2] Likewise since plane B is perpendicular to plane X, and EG is their intersection,

therefore every line drawn in plane B perpendicular to EG is also perpendicular to plane
X (Def. 3). Therefore the straight line drawn from P (in plane B), perpendicular to EG, is
perpendicular to plane X.

[3] Therefore there is a perpendicular to plane X standing on point P that lies in plane
A (Step 1), and again there is a perpendicular to plane X standing on point P that lies in
plane B (Step 2). But there is only one perpendicular to plane X standing on point P
(Thm.5 Remark). Therefore the line perpendicular to plane X, standing on point P, must
be a line common to planes A and B. But the only line common to them is their line of
intersection (Princ. 2), namely NP. Therefore the line perpendicular to plane X, standing
on point P, is NP.

So PN is perpendicular to plane X.

Q.E.D.

284
THEOREM 16: In a solid angle formed by three rectilineal angles, any two of
those angles together are greater than the third.
D

Let V be the vertex of a solid angle

made up of the three rectilineal angles
AVD, DVB, and AVB. I say that any
two of these together are greater than Q
the third. A K
B

[1] Drop DK perpendicular to the T

plane of AVB (Thm.6). In plane V
AVB, draw KT perpendicular to VB.
Join DT.

[2] Now since DK is perpendicular to AVB,

thus plane DKT is perpendicular to plane AVB (Thm.8).
So any line in plane AVB that is perpendicular to KT
(which is the intersection of planes DKT and AVB)
must be perpendicular to plane DKT (Def.3).
But VT is perpendicular to KT (Step 1).
Hence VT is perpendicular to plane DKT.

[3] Since VT is perpendicular to plane DKT (Step 2),

thus VT is perpendicular to every line through T in plane DKT (Def.2).
So VT is perpendicular to DT, i.e. ∠DTV is right.

[4] Now since DK is perpendicular to plane AVB, hence ∠DKT is right.

Thus DT > TK (since DT is hypotenuse in right rDTK)
So cut off TQ = TK.
Now ∠VTK = ∠VTQ (both are right; ∠VTQ is ∠DTV)
and VT is common (to triangles VTK and VTQ)
so rVTK ≅ rVTQ
so ∠QVT = ∠KVT

[5] Now ∠DVT > ∠QVT (the whole is greater than the part)
so ∠DVT > ∠KVT (∠KVT = ∠QVT, Step 4)

[6] So ∠DVB > ∠KVB (Step 5)

Similarly ∠DVA > ∠KVA
hence ∠DVB + ∠DVA > ∠KVB + ∠KVA (adding)
or ∠DVB + ∠DVA > ∠AVB

285
 So these two given angles are greater than the third. Since there was nothing special
about the two angles we chose among the given three, it follows the same way that any
two of them will be greater than the third.

Q.E.D.

THEOREM 16 Remarks:

1. A solid angle contained by 3 plane angles is called a trihedral angle.

D
2. What if K lands outside angle AVB?
Then the proof is identical up to Step 4,
where we said ∠DVT > ∠KVT. Now
extend KV through ∠AVB. N B
A
Thus ∠KVT = ∠NVB (vertical) Q
so ∠DVT > ∠NVB. V
And since ∠DTV is right, hence ∠DVT is T
acute (in rDTV), and so its supplementary K
angle, ∠DVB, is obtuse.
Hence ∠DVB > ∠DVT
so ∠DVB > ∠NVB (since ∠DVT > ∠NVB above)
and ∠DVA > ∠NVA by the same reasoning. And the remainder of the proof is the
same as in the Theorem.

3. To illustrate why this Theorem is true, draw any

V X Z W
angle XYZ on a piece of paper, and on each side of it
draw angles VYX and ZYW which together add up to
an angle less than angle XYZ. Cut out rVYW, and
fold along XY and YZ. Do triangles VYX and ZYW
form a solid angle with triangle XYZ? Do they meet
above the plane of rXYZ? What happens if ∠VYX +
Y
∠ZYW = ∠XYZ?

286
THEOREM 17: Any solid angle is contained by plane angles adding up to less
than four right angles.
B

Let's start once more with a “trihedral”

angle, an angle formed by three plane 3 4
angles, namely 7, 8, 9, all coming up to a
point D. (You must imagine that point D 7
D 8
is above the plane of this page.) I say that 2
7 + 8 + 9 is less than four right angles. A 1
9
5
6
Choose A, B, C at random along the legs
of the solid angle, and join AB, BC, CA, C
thus forming solid angles again at A and at
B and at C. Looking at the diagram, then, you must remember that you are looking down
like a bird at the peak of a solid pyramid – so ABC is the base of the pyramid, but angles
1 through 9 all lie in planes that rise up toward you from that base.

[1] Because A is a trihedral angle, thus 1 + 2 > ∠CAB (Thm.16)

[2] Because B is a trihedral angle, thus 3 + 4 > ∠ABC (Thm.16)

[3] Because C is a trihedral angle, thus 5 + 6 > ∠BCA (Thm.16)

[4] Adding together all these inequalities, keeping the greater things on one side,
1 + 2 + 3 + 4 + 5 + 6 > ∠CAB + ∠ABC + ∠BCA
but ∠CAB + ∠ABC + ∠ΒCA = two rights (triangle ABC)
so 1 + 2 + 3 + 4 + 5 + 6 > two rights

[5] Now angles 1 through 9, added together, equal all the angles in three triangles,
and so all together they add up to three times the angle-sum of a triangle, i.e. three times
two rights, i.e. six rights. So
1 + 2 + 3 + 4 + 5 + 6 + 7 + 8 + 9 = six rights

[6] Thus, if we subtract more than two rights from these nine angles, less than four
rights will remain. But 1 + 2 + 3 + 4 + 5 + 6 is more than two rights (Step 4). Therefore,
when subtracted from the nine angles, less than four rights remain, i.e.
7 + 8 + 9 < four rights.
So the three plane angles forming a trihedral angle must add up to less than four right
angles.

Q.E.D.

287
THEOREM 17 Remarks:

This Theorem is not limited to solid angles made

of three plane angles. Take any solid angle with E
vertex V formed out of n plane angles. Pass a D
plane through the legs of the angle, forming a V
A
polygon base and a pyramid with vertex V. The
polygon base will thus have n sides, and if we
pick a random point R inside it, we can divide it
into n triangles. B
C
Now the angle-sum of the polygon base
equals the angles of all those n triangles minus V
the angles around R, i.e. minus 360°. So the
angles of the polygon = (n ×180° – 360°).
Since every vertex of the polygon base is
also the vertex of a trihedral angle in the pyramid, E
D
hence very angle of the polygon must be less than
A
the two angles above it which form the angles at R
the foot of the pyramid. For example, ∠ABC <
∠ABV + ∠CBV (Thm.16). So all 2n angles B
C
about the foot of the pyramid add up to more than
the n angles of the polygon, i.e. more than (n ×180° – 360°). So let those angles at
the foot of the pyramid add up to (n ×180° – 360° + Z°).
Now the n plane angles forming the solid angle at V equal the angles in the n
triangular faces of the pyramid minus their 2n angles at the foot of the pyramid. So the
n angles forming solid angle V add up to
(n ×180°) – (n ×180° – 360° + Z°)
or 360° – Z°
So the n plane angles forming solid angle V add up to less than four rights.

288
THEOREM 18: If among three angles in a plane any two are greater than the
third, and they are made the peak angles of three isosceles triangles of the same leg-
length, then likewise for the bases of these triangles, any two together will be greater
than the third.
P
Given: Three isosceles triangles whose legs
1
2
3 are all equal, i.e. PA = PB = PC = PD, and
A D whose peak angles (1, 2, 3) are such that any
two are greater than the third.

B Prove: Any two bases of these triangles will

C be greater than the third.

For example, I say that AB + BC > CD.

[1] Join AC.

[2] Since AP = CP = DP (given)

but ∠APC > ∠CPD (given)
thus AC > CD (Ch.1, Thm. 16 Question 1)

[3] Now AB + BC > AC (triangle ABC)

and AC > CD (Step 2)
so AB + BC > CD

Since there was nothing special about AB and BC, the same proof works just as well to
show that BC + CD > AB, and again that AB + CD > BC. To show that AB + CD > BC,
just rearrange the triangles so that angles 1 and 3 are next to each other, and 2 is on the
outside.
So whenever three isosceles triangles of the same leg-length are formed with three
peak angles any two of which are greater than the third, likewise for their bases any two
of them together will be greater than the third. Q.E.D.

THEOREM Remarks:

A quick corollary follows from this Theorem: we can make a triangle out of lengths AB,
BC, CD, since any two of them are greater than the third. Thus we conclude: When
three isosceles triangles of the same leg-length are formed with three peak angles any
two of which are greater than the third, then it will be possible to make a triangle out of
the lengths of their bases. For short, call such a triangle a “base triangle.”
Obviously, this Theorem is simply a matter of plane geometry, but we will need it
for the upcoming Theorem 20, here in solid geometry, where we shall construct a solid
angle.

289
THEOREM 19: If the peak angles of three isosceles triangles with a common
leg-length L add up to less than four right angles, then L is greater than the radius of the
circle circumscribing their “base triangle.”

Again, this is a matter of plane geometry, but it is

crucial for the solid geometry in the next theorem. Start
with three isosceles triangles of leg-length L, with peak
L angles 1, 2, 3 adding up to less than 360°, and bases X,
C
1
Y, Z. Since they have the same leg length, L, if we
3
X
2 place their equal sides together and give them a
Z
common vertex, C, the circle of center C and radius L
Y will pass through the endpoints of bases X, Y, Z. Since
1 + 2 + 3 is less than 360°, hence the chords X, Y, Z do
not cut off the circle’s entire circumference.
U Now if the angles 1, 2, 3 are such that any two
are together greater than the third, we can make a
X Z triangle out of X, Y, Z (Thm.18). So suppose this
M condition is met, and make rTUV with sides equal to
V X, Y, Z. Circumscribe a circle about rTUV (Ch.4).
T Y Call its center M.
Obviously the chords X, Y, Z together cut off
the entire circumference of circle M. But these same chords together cut off only a
portion of the circumference of circle C. Therefore circle C is greater than circle M, and
so L (the radius of circle C) is greater than the radius of circle M.

Q.E.D.

THEOREM 19 Remarks:

The proof takes it as evident that if the same chord

length cuts off a greater portion of the circumference in
one circle than it does in another, the other circle is
greater than the one. For example, if KD cuts off an arc H

in circle G corresponding to ∠KGD, and an arc in

G
circle H corresponding to ∠KHD, and ∠KGD >
∠KHD, then circle H is larger than circle G. To see it, K D
compare isosceles triangles KGD and KHD. Since
∠KGD is greater than ∠KHD, the base angles of
isosceles rKGD must be less than those of rKHD, and so KG and DG must meet
inside rKHD. Hence the legs of rKGD are less than those of rKHD. So GK < HK,
which means circle G is smaller than circle H.

290
 THEOREM 20: How to make a solid angle out of three plane angles. Thus it is
required that they add up to less than four right angles, and that any two of them are
greater than the third.

Let our three given plane angles be 1, 2, 3. By Theorem's 16 1

2

and 17 we know that it is impossible to make a solid angle

out of them unless they meet the conditions that any two of 3
them are greater than the third, and they add up to less than
four right angles. So let them meet these conditions.
To make a solid angle out of them,

P R [1] Cut off any length PW along the

leg of angle 1, and make three
Q 3
K
1 isosceles triangles PWX, QXY, RYZ,
2 all having leg-length PW.
W X Y Z
V

[2] Thus a triangle can be made out of their bases

(Thm.18). So make triangle ABC with
AB = WX
and BC = XY
and CA = YZ. A
Draw a circle around triangle ABC, find
M
center M, and join MA. C
B

[3] Draw a semicircle on PW. Setting your compass to length MA, make a circle (not
shown) around center W, and where it cuts the semicircle call K. Thus WK = MA. This
can be done because MA is less than diameter WP (by Thm.19).

[4] Join PK. Thus ∠PKW is right (Ch.3).

Set up MV perpendicular to the plane of the circle (Thm.7), making MV = PK.

[5] Now MV = KP (we made it so; Step 4)

and MA = KW (we made it so; Step 3)
and ∠VMA = ∠PKW (both are right; Step 4)
so rVMA ≅ rPKW (Side-Angle-Side)
thus VA = PW
Likewise VC and VB are also each equal to PW, the common leg-length of our original
isosceles triangles.

291
[6] But AB, BC, CA are equal to the bases of our isosceles triangles WX, XY, YZ
(Step 2). So the three triangles standing on AB, BC, CA from point V are congruent to
the three isosceles triangles (SSS), and hence the three peak angles forming solid angle V
are equal to the given angles 1, 2, 3.

Q.E.F.

THEOREM 20 Remarks:

This Theorem is the converse of Theorems 16 and 17. In 16 and 17 we learned that any
trihedral angle must be made of plane angles which add up to less than four rights and
any two of which add up to more than the third one. But we were left wondering: are
there more conditions required for three plane angles to be able to form a solid angle, or
are those two conditions sufficient? Also, we might wonder this: the three angles must
be less than four right angles – but do they in fact have to be less than three right angles,
too? Or is it enough for them to be less than four right angles? This Theorem answers all
those questions: as soon as the three plane angles are such that they are less than four
right angles (by whatever amount you like), and such that any two of them are greater
than the third, we can make them into a solid angle. Those conditions are not only
necessary, but sufficient.

THEOREM 21: If a solid is contained by three pairs of parallel planes, the

opposite faces are congruent parallelograms (i.e. the solid is a parallelepiped).

G Suppose solid AH is contained by three pairs

B of parallel planes, namely BE and CK, and
BH and AK, and BD and GK. I say that each
E
H
pair of opposite faces, such as ABCD and
A C EGHK, are identical parallelograms.

K [1] Since AB and CD are the intersections

of plane AC with the parallel planes BE and
D
CK, therefore AB is parallel to CD (Thm.13).

[2] Since BC and AD are the intersections of plane AC with the parallel planes BH
and AK, therefore BC is parallel to AD (Thm.13).

292
[3] Since AB is parallel to CD (Step 1)
and BC is parallel to AD (Step 2)
thus ABCD is a parallelogram.
Likewise the remaining 5 faces are parallelograms.

[4] Join AG, DH.

[5] Since AD is parallel to EK (because ADKE is a parallelogram)

and GH is parallel to EK (because GHKE is a parallelogram)
thus AD is parallel to GH (Thm.9)

[6] And thus A, D, G, H are all in one plane (Thm.2). And their plane intersects the
parallel planes BE and CK at AG and DH, and therefore AG is parallel to DH (Thm.13).
But AD was just proved parallel to GH (Step 5), and therefore AGHD is a parallelogram.

[7] So AG = DH (opp. sides in parallelogram AGHD)

and AB = DC (opp. sides in parallelogram ABCD)
and BG = CH (opp. sides in parallelogram BGHC)
so rABG ≅ rDCH (Side-Side-Side)

[8] But ABGE is just two of rABG, and DCHK is just two of rDCH, similarly
arranged. Therefore
ABGE ≅ DCHK.
Likewise the other opposite parallelograms containing the solid are congruent to each
other.
Therefore if a solid is contained by 3 pairs of parallel planes, then its six faces are
three pairs of congruent parallelograms, and such a solid is called a parallelepiped.

THEOREM 22: If a parallelepiped is cut by a plane parallel to one of its pairs

of opposite faces, the two resulting parts have to each other the same ratio as the bases
on which they stand.

P Given: Parallelepiped A + X, cut by a plane

L at PLN parallel to one pair of its opposite
faces, thus dividing it into two
B A X Y Z parallelepipeds, namely A and X.

Prove: volume of A : volume of X = area

N of base of A : area of base of X

293
[1] Place a solid B, identical to A, right next to it, and a solid Y, identical to X, right next
to it. And thus multiply solids A and X however many times you like. Say you double
A, and triple X.

[2] Because of the identical shape and size of solids A and B, it is clear that the base of
the whole solid A + B is double the base of A.
Likewise the base of the whole solid X + Y + Z is triple the base of solid X.

[3] Now, because they lie inside the same parallels and have identical angles, if the solid
A + B is equal in volume to the solid X + Y + Z, this can only be because they stand on
equal bases, i.e. the base of A + B must be equal to the base of X + Y + Z.
But if the solid A + B is bigger than solid X + Y + Z, then A + B must stand on a
bigger base than X + Y + Z does. And if the solid A + B is smaller than solid X + Y + Z,
then A + B must stand on a smaller base than X + Y + Z does.

[4] Therefore, whatever multiple we take of solid A (and therefore of its base), and
whatever multiple we take of solid X (and therefore of its base), the multiple solids must
compare the same way as the corresponding multiple bases.

[5] Therefore solid A : solid B = base of A : base of B (Ch.5, Def.8)

Q.E.D.

THEOREM 23: Parallelepipeds standing on the same base and having the same
height are equal (i.e. they have the same volume).
K P
Given: Parallelepipeds AE and ME, both A M
standing on base BCE and having their
G N
tops in the same plane. D L
3
1
Prove: AE and ME have the same B 2
volume.
E
First, suppose solids AE and ME not only C
have their tops in the same plane, but also
that some other pair of their faces lie in the same plane, say CG and CN lie in the same
plane – and therefore also the parallel faces BK and BP lie in the same plane on the
opposite side.

294
But CDA and EGK do not coincide with CLM and ENP (if they did, the two solids would
coincide entirely).

I say that solids AE and ME have the same volume.

[1] For since CDGE and CLNE are both parallelograms, therefore
DG = LN (each is equal to CE)
so DL = GN (subtracting part LG from both sides)
but DC = GE (in parallelogram CDGE)
and CL = EN (in parallelogram CLNE)
so rDCL ≅ rGEN (Side-Side-Side)

[2] Now AD is parallel to BC (ABCD is a parallelogram)

and ML is parallel to BC (BCLM is a parallelogram)
so AD is parallel to ML (Thm.9)
thus ADLM is a parallelogram.
Clearly KGNP is also a parallelogram, and it is congruent to ADLM.
And, because they are opposite faces in the parallelepipeds,
AC and KE are congruent parallelograms
and MC and PE are congruent parallelograms

[3] Clearly, then, the two triangles and three parallelograms containing prism 1 are
congruent with and arranged similarly to the two triangles and three parallelograms
containing prism 3 (Steps 1 and 2). And thus they can be made to coincide and therefore
have equal volumes.

[4] So prism 1 = prism 3 (Step 3)

so solid 1 + 2 = solid 2 + 3 (adding solid 2 to each side)
i.e. solid AE is equal to solid ME.

P
T
K Next, suppose that solids AE and ME
N have only their tops and bottoms in the
Z
G S
A
M same planes, and the front face of ME,
L namely CRSE, does not lie in the same
plane as CDGE, the front face of solid
D R AE.

B E AE and ME are still going to be equal

in volume.

295
Let MRST be the top face of solid ME, in the same plane as ADGK, the top face of solid
AE.

[1] Extend RM to Z on AK, and ST to P on the extension of AK.

Extend LG to N.
Join ZB, LC, NE, and P to X, the back corner of base BCE (which, to avoid
cluttering up the diagram, I have not drawn).

[2] Now ZLNP is a part of the top plane, and the top plane is parallel to base BCEX
(given). Thus
plane ZLNP is parallel to plane BCEX.

[3] And CLNE is a part of the face plane CDGE, which is parallel to the back plane
BAKX (in solid AE). But BZPX is a part of the back plane. Thus
plane CLNE is parallel to plane BZPX.

[4] And ZLCB is a part of the side plane MRCB, which is parallel to the opposite
side plane TSEX (in solid ME). But PNEX is a part of that opposite side plane. Thus
plane ZLCB is parallel to plane PNEX.

[5] Therefore the solid contained by planes

ZLNP and BCEX
and CLNE and BZPX
and ZLCB and PNEX
is contained by 3 pairs of parallel planes (Steps 2 – 3).
Therefore that solid, namely ZE, is a parallelepiped (Thm.21), and it stands on base
BCEX and under the same height as the two given solids)

[6] Since solid ZE has its face CLNE in the same plane as CDGE, the face of solid
AE, therefore solid ZE = solid AE, by the first part of this Theorem.

[7] Again, since solid ZE has its face ZLCB in the same plane as BMRC, the face of
solid ME, therefore solid ZE = solid ME, by the first part of this Theorem.

[8] Therefore solid AE = solid ME (each being equal to solid ZE; Steps 6 and 7).

Therefore, no matter what, when two parallelepipeds have the same base and stand under
the same height, they have the same volume.

Q.E.D.

296
THEOREM 24: Parallelepipeds which are of the same height and on bases of
equal area are equal.

Conceive two parallelepipeds, AV X

V
and TX, with the same height and
with bases ABCD and QRST C Q R
B
having the same area. I say the
solids have the same volume. A D

[1] Let’s take the simplest case B C W T S

first: let the sides of these solids
all be perpendicular to their bases –
thus CV and RX are perpendicular D Q R
A E
to the bases and CV = RX (because
the heights are the same). Because
the walls of these solids are thus all
standing at right angles to the
bases, we can imagine the solids
like two buildings, and just look at L G T S
K H
their “floor plans,” namely their
bases ABCD and QRST.

Now, to prove that AV = TX …

[2] We place a solid identical to TX in line with AD, that is, letting DEGH (identical
to base QRST) be its base, we place DE in a straight line with AD. Complete
parallelogram CDEW in the base plane, and build a “building” on it with the same height
again as the solids on ABCD and DEGH.

[3] Extend CD to where it meets GH extended, namely at L, and complete

parallelogram EDLK in the base plane, and build another “building” on top of it with the
same height once more.

[4] Now, there is an undrawn rectangle standing straight up on DE (coming up at you

out of the page) which is a wall for the building on DEKL; but it is also a wall for the
building on DEGH. Since there is no absolute up and down in geometry, this wall can
also be thought of as a base of each of these two solid buildings, and both are under the
same height, i.e. both their tops lie in the plane standing on LKHG. Therefore they are
equal in volume (Thm.23).
So The solid on DEKL = the solid on DEGH

[5] Notice that the buildings on DEKL and CDEW together make up one big
parallelepipedal building, since they are in line with each other. Therefore, by Thm.22,
building on DEKL : building on CDEW = area of DEKL : area of CDEW,

297
 [6] But, looking just at the parallelograms in the base plane,
DEKL = DEGH (both stand on DE, and are in the same parallels)
but DEGH = QRST (we made DEGH identical to QRST)
and QRST = ABCD (given)
so DEKL = ABCD

[7] So, substituting ABCD for DEKL in the proportion from Step 5, we have:

building on DEKL : building on CDEW = area of ABCD : area of CDEW,

But also by Theorem 22, we have

building on ABCD : building on CDEW = area of ABCD : area of CDEW

Since we have two ratios the same as a third ratio, they are the same as each other, i.e.

blding on DEKL : blding on CDEW = blding on ABCD : blding on CDEW.

Notice in this proportion the buildings on DEKL and ABCD both have the same ratio to
the building on CDEW. From this, it follows that they are equal. Thus

building on DEKL = building on CDEW.

[8] Now solid on DEKL = solid on ABCD (Step 7)

but solid on DEKL = solid on DEGH (Step 4)
so solid on ABCD = solid on DEGH
but solid on QRST = solid on DEGH (we made it thus in Step 2)
so solid on ABCD = solid on QRST

Therefore the solid AV is equal in volume to the solid TX.

[9] Now what if the solids on ABCD and QRST, although having their tops and
bottoms in the same planes, yet have their walls tilted in different ways? Will they still
be equal? Yes.
Just build the solids on
F J
V X those same bases whose walls are
perpendicular to the bases, having
their tops also in the same top-
C
R plane as the “tilty” solids. Then,
by Theorem 23, each upright solid
A D T S
is equal to the tilty solid whose
base it shares. But, by the proof we just gave, the two upright solids are equal to each
other – since they stand on equal bases and between the same parallel planes. Therefore
the tilty solids are equal, too.

Q.E.D.

298
THEOREM 25: Parallelepipeds of the same height are to each other as their
bases.
1 2 V 3 Z
T X

F B C
G P

E K A D Q

Given: Parallelepipeds 1 and 2 of the same height, standing on bases EFGK and ABCD.

Prove: Solid 1 has to solid 2 the same ratio that base EFGK has to base ABCD.

[1] Extend base ABCD so that parallelogram DCPQ, while having the same angles as
parallelogram ABCD, nonetheless has the same area as EFGK.

[2] Complete the parallelepipedal solid on DCPQ by extending the planes of solid 2,
and by capping it off with plane QXZP parallel to plane DTVC. Thus we have solid 3,
and solids 2 and 3 together form one big parallelepiped.

[3] Now solid 3 : solid 2 = DCPQ : ABCD (Thm.22)

[4] But solid 3 = solid 1,

since they stand between the same parallel planes, and have bases of equal area
(Thm.24). Substituting solid 1 for solid 3 in the proportion from Step 3, then, we have:
solid 1 : solid 2 = DCPQ : ABCD

[5] But DCPQ = EFGK, by Step 1. Substituting EFGK for DCPQ in the
proportion, we now have
solid 1 : solid 2 = EFGK : ABCD,
which is what we sought to prove.

Q.E.D.

299
THEOREM 25 Remarks:

You might be wondering how we

F L G M N accomplish Step 1. How do we extend
the base ABCD with a parallelogram
B E DCPQ that is equiangular with ABCD,
P
C
but equal in area to EFGK?
Since that all takes place in the
base plane, it is a matter of simple plane
A
geometry, and Chapter 1 gives us all we
R D Q
need:

[1] Place EFGK on BC so that K is on point C.

[2] Draw LER parallel to AB and CD. Join RC. Extend RC and FG until they meet at
N. Extend DC to M. Complete parallelogram DMNQ. Extend BC to P.

[3] Parallelogram DCPQ is clearly equiangular with parallelogram ABCD. But it is

equal to parallelogram EFGC in area,
since DCPQ = ECML (complements in parallelogram RLNQ)
and EFGC = ECML (in the same parallels and on the same base)
so DCPQ = EFGC

THEOREM 26: Similar parallelepipedal solids are to one another in the

triplicate ratio of their corresponding sides.

G B

Q H Given: Similar parallelepipeds AB and CD, with

sides AE and ED being a pair of corresponding
sides.
K

D
E
Prove: Solid AB : solid CD is the ratio triplicate of
A
L AE : ED.
M
C

300
[1] Place solids AB and CD so that they have a common corner at E, and the
corresponding sides AE and ED lie in a straight line. Thus the corresponding sides LE
and EK will also line up (since ∠LED = ∠KEA in the similar solids).

[2] In angles HED and HEK complete parallelepiped EG.

In angles HED and HEL complete parallelepiped LQ.

[3] Because of the similarity of the solids, AE, KE and HE are proportional to ED,
EL, and EM. Hence
AE : ED = KE : EL = HE : EM

[4] Now, because parallelograms under the same height are to one another as their
bases (Ch.6, Thm.1), it follows that:
AE : ED = AK : KD
and KE : EL = KD : DL
and HE : EM = HD : DM.
Because of Step 3, the first in each of these pairs of ratios are all the same ratio.
Therefore the second in each of these pairs of ratios are also all the same,
i.e. AK : KD = KD : DL = HD : DM.

[5] But since parallelepipeds under the same height are to each other as their bases
(Thm.25), it follows further that:
AK : KD = solid AB : solid EG
and KD : DL = solid EG : solid LQ
and HD : DM = solid LQ : solid CD
Because of Step 4, the first in each of these pairs of ratios are all the same ratio.
Therefore the second in each of these pairs of ratios are also all the same,
i.e. solid AB : solid EG = solid EG : solid LQ = solid LQ : solid CD

[6] Since that proportion is continuous, and contains four terms, therefore the first has
to the last the triplicate ratio of the first to the second, i.e.
solid AB : solid CD is the triplicate ratio of solid AB : solid EG.

[7] But, as we saw above in Steps 5 and 4,

solid AB : solid EG = AK : KD = AE : ED.
Therefore solid AB : solid CD is the triplicate ratio of AE : ED.

So similar parallelepipeds have to each other the triplicate ratio of their corresponding
sides.

Q.E.D.

301
THEOREM 26 Remarks:

The most important instance of this, of course, is with cubes. All cubes are similar
parallelepipeds, and so it follows that they are to each other in the ratio triplicate of their
corresponding sides.
For example, suppose you had a pair of cubes, and the side or edge of one was
double the side or edge of the other, i.e. their sides were in the ratio of 1 : 2. Then what
is the ratio of their volumes? It will be 1 : 8, since
1 : 2 = 2 : 4 = 4 : 8,
and thus 1 : 8 is the ratio triplicate of 1 : 2.
This Theorem should make you wonder about the ratios of other kinds of similar
solids, such as curved ones. Do spheres have to each other the triplicate ratio of their
diameters?

THEOREM 27: If the sides of opposite faces in a parallelepiped are bisected by

two planes, then the intersection of these two planes bisects (and is bisected by) the
diagonal of the solid.
D K F
Given: Parallelepiped BE, with diagonal O U P
CH. Planes QOPR and MKLN bisect the L
E
C
edges at Q, O, M, K, R, P, N, L. SU is the
intersection of these two cutting planes. T

Prove: SU and CH bisect each other. M H

B
R
Q S
A N G
[1] Join CU, UF.

[2] It is easily seen that OULC and UPEL are parallelograms.

Thus OU = CL
and UP = LE.
but CL = LE (given)
thus OU = UP
but OC = PF (being halves of the equal sides DC and EF)
and ∠UOC = ∠UPF (each is equal to ∠PEL)
so rUOC ≅ rUPF (Side-Angle-Side)
thus ∠OUC = ∠PUF.

302
[3] But OUP is a straight line, and therefore CUF is also a straight line, since the
vertical angles OUC and PUF are equal.
Likewise ASH is a straight line.
And since AC and FH are equal and parallel lines, ACFH is a parallelogram.

[4] Thus SU lies in the plane of parallelogram ACFH, since it joins points U and S
which lie on its opposite sides. Thus CH and SU must meet, say at T.

[5] Now CU = UF (since rUOC ≅ rUPF; Step 2)

and AS = SH (since similarly rSQA ≅ rSRH)
Thus SU joins the midpoints of the opposite sides in parallelogram ACFH. Therefore SU
bisects the diameter of ACFH, namely CH, and also is bisected by it.

Q.E.D.

THEOREM 27 Remarks:

C U F

1. If it is not perfectly clear why the line joining

the midpoints of a parallelogram's opposite sides must T
bisect and be bisected by the diagonal, consider the
following. Let ACFH be a parallelogram, and let CU
= UF, and AS = SH. A S H

Now CU = SH being halves of the opposite sides of a parallelogram,

and ∠HCF = ∠CHS since CF is parallel to HA
and ∠CUS = ∠HSU since CU is parallel to SH
so rCUT ≅ rHST (Angle-Side-Angle)
so UT = TS
and CT = TH Q.E.D.

2. Obviously, this Theorem is true about cubes in particular – if the sides of a cube
are bisected by two planes, the intersection of those planes will bisect the diagonal of the
cube, and be bisected by it.

3. In parallelepipeds other than cubes, the four diagonals can be unequal to each
other. But that doesn't make any difference to this Theorem – take any diagonal you like,
the proof did not require that we choose a special one.

303
THEOREM 28: If a triangular prism lies on one of its parallelogram sides, and
in this position has the same height as another triangular prism lying on its triangular
base, and if the parallelogram is double the triangle, then the prisms will have the same
volume.
N Z
Imagine a prism with triangular
R P
bases ABM and DCN, lying on M O
L
one of its parallelogrammic sides D T
ABCD, and another prism with
C K
triangular bases EGK and OLP, A E

lying on EGK, which has half the

B G
area of ABCD.
Now if we further suppose that ABCD and EGK lie in the same plane, and also
that OLP and MN lie in the same plane, then I say that the prisms will have the same
volume.

[1] Complete the parallelepiped AR contained by the angles ADC, ADN, NDC.
Complete parallelogram EGKT, and
Complete the parallelepiped GZ contained by the angles GKT, GKP, PKT.

[2] Since ABCD is double triangle EGK in area, and EGKT is also double triangle
EGK in area, therefore ABCD = EGKT.

[3] But that means that solids AR and GZ stand on equal bases. And yet they also
have the same height, since it is given that the height of the prisms is the same, and we
made the parallelepipeds to have that same height. Therefore AR and GZ have the same
volume (Thm. 24).

[5] Since AR and GZ are the same volume, therefore also their halves have the same
volume. But the triangular prisms are obviously their halves. Therefore the two prisms
are equal in volume, too.

Q.E.D.

304
THEOREM 28 Remarks:

We assumed in this Theorem that each prism is obviously half the volume of the
parallelepiped of which it is a part. Why is that obvious?
Consider the prism contained by triangles OLP and EGK. It makes up a
parallelepiped by being combined with another prism, the one contained by triangles
OZP and ETK. Now EGKT and OLPZ are parallelograms, and so are OZTE and all the
other faces of the parallelepiped.
Thus rOLP ≅ rOZP
and rEGK ≅ rETK
and LPKG ≅ OZTE
and OLGE ≅ ZPKT
and, of course, OPKE is a common face for both prisms.
So the two prisms are contained by an equal number of congruent and similarly
arranged faces. Therefore they are congruent and contain equal volumes.
Does that mean that these prisms can coincide? Not necessarily.
Consider your right hand and your left hand. Even if they were perfectly
symmetrical, and of a ghostly quality so that they could pass through each other, they
would not be able to coincide with each other and form one self-same hand. A right hand
simply can't be a left hand!
Now, can the two prism halves of a parallelepiped be like that? Can they be
perfect mirror images of each other, and yet not be able to coincide? Yes. It is almost
impossible to represent this in a two-dimensional diagram in a clear and convincing way,
so the best thing to do is to make a pair of such prisms. It is best not to use paper, since
that is too flimsy – you need something more rigid like cardstock or a manila folder.
Transfer the diagrams below onto a piece of manila: each consists of a square, a rhombus
with angles of 60° and 120° (it is made of two equilateral triangles), and two isosceles
triangles with peak angles of 105° (i.e. 60° + 45°) placed at the bottom corners of the
square. The legs of the isosceles triangles are equal to the sides of the square.
After you have transferred the diagrams, cut out the two figures along the solid
lines. Next, with all the labeling face up on the table, fold up the triangles and square
along all the dotted lines. Bring together the edges marked with the same letters, such as
“A”, and tape them together. When you are done, you will have two triangular prisms,
each with one open face. If you place the square faces down on the table and turn H and
Z toward you, you will see that the prisms are symmetrical, but, like a right hand and a
left hand, cannot be made to coincide. Their corresponding faces can be made to
coincide one at a time, but not all of them simultaneously. If you pick them up in your
hands, and place edges X and Z together, and in that position bring together the two open
faces of the prisms, you will be holding a parallelepiped.
What makes the equality of these two prisms obvious, then, is not that they could
be made to coincide. Rather, like your two hands, it is their perfect symmetry – one is a
perfect mirror image of the other.

305
 H

A B

105
105
A
LEFT

C D

105
105
D
RIGHT

306
“HOOK”: TRIANGULAR SECTIONS OF A CUBE.

If you are given a cube and a triangle abc, will it be possible to slice the cube with a
plane so that there will be formed a triangular facet which is similar to abc? Not if abc
is right or obtuse. But if abc is acute, it can always be done.

Z
K
E
A B
a D
b

C G

307
 Chapter Ten

Volumes and Areas

 DEFINITIONS
1. A SERIES is a multitude of quantities produced in an order determined by some
repeated process. It is an INFINITE SERIES if the process producing it can always be
repeated to produce the next quantity.
For example, the numbers 2 4 8 16 constitute a series because they are
quantities produced in this order by the repeated process of doubling. They are part of
an infinite series because there is no end to how many times we can repeat the process of
doubling, and so we can always say which number comes next in the series.

2. An infinite series is said to APPROACH a quantity that is not a member of it if, every
time we are given an assigned amount, it is possible to find a member of the series
differing from the quantity by less than that amount.
For example, take a series of lengths beginning with one foot:

First Term: 1

1
Second Term: 1 +
2

1 1
Third Term: 1+ +
2 4

1 1 1
Fourth Term: 1 + + + etc.
2 4 8

These lengths constitute an infinite series, since we can always form the next new term by
adding half of the last thing added in the previous term. Now, two feet, or 2, is not a
member of this series – no matter how far we go in it, every term will always be less than
two feet long. But each new member of the series gets us closer to two feet, and in fact
the terms come as close to two feet as you want (as we will prove in the Theorems). For
example, take one millionth of an inch – there will be a term in the above series that falls
short of being two feet long by LESS than a millionth of an inch. And so this series is
said to approach two feet. It never gets there, but it gets closer than any assigned
difference.

3. A GROWING series is one each new member of which is greater than the one before
it. A SHRINKING series is one each new member of which is less than the one before it.
Note: you can also have oscillating series, e.g. where all the odd terms are
growing but all the even terms are shrinking. It is not within the scope of this book to
exhaust all the kinds of series that are possible.

308
 THEOREMS

THEOREM 1: If we subtract half from the greater of two unequal magnitudes,

and always half again of what remains of it, by this process we will eventually have left a
magnitude less than the lesser one set out.
A

B
Recall from Chapter 8 that a magnitude is
any quantity which is infinitely divisible, Z
such as a line, or a surface, or a solid. Now
consider any two unequal magnitudes, A and B, and suppose A is the greater one. What I
wish to prove amounts to this:

1 1 1
The process A – A – A – A etc. will eventually leave a remainder that is
2 4 8
less than B.

[1] Since B has a ratio to A (A being greater than B), therefore some multiple of B is
greater than A. Thus if we double B, and then double its double, and double this again,
etc., we will eventually arrive at a magnitude Z that is greater than A.

[2] Suppose, then, that Z > A

and Z = 8B
8B being a number of B's reached by repeatedly doubling B.

[3] Now A < Z (Step 2)

1 1
so A < Z (obviously)
8 8

1
so A < B (B equals one eighth Z by Step 2)
8

1
[4] But A is a part of A that remains after repeatedly halving A, and halving what
8
remains, etc. That is,
1 1 1 1
A = A – A – A – A
8 2 4 8

[5] Therefore, by repeatedly halving A, and halving its half, etc., we must eventually
arrive at a remainder that is less than B.

Q.E.D.

309
THEOREM 1 Remarks:

1 1 1
1. In Step 3 we mentioned of A, and also of Z. of Z poses no problem,
8 8 8
because that is just B – we started with B, and got Z by taking B 8 times. But with A we
actually started with A, not with something which, taken 8 times, equals A.
1
So how do we make of A? Do we know how to find exactly one eighth of any
8
random magnitude? No. But this theorem does not assume we know how to do that – it
1
only says that if we can take half and half again as often as we like (and thus leave , or
8
1 1
, or , etc.) then we will by this process eventually leave a remainder that is less
16 32
than any given magnitude.

2. Even if we don't know how to take exactly half of any magnitude, as long as its
parts always have a ratio to each other we can easily take more than half of it. Divide it
at random, and take the larger piece (if they happen to be equal, then you did divide it in
half, after all).
And clearly, if repeatedly taking just half of what remains of A gets us to a
remainder less than B, then all the more quickly will repeatedly taking more than half of
what is left get us to a remainder less than B. So suppose we have a growing series of
magnitudes Q, R, S, T etc., and another greater magnitude X. Suppose further that each
new term in the series takes up more than half of the difference by which X exceeded the
last term, i.e. X – R is less than half of X – Q, and X – S is less than half of X – R
etc. Then the series approaches X.

3. Can you prove that repeatedly taking one third of A, and then one third of what
remains of A etc., will eventually leave a remainder less than B? Remember, we are not
subtracting one third of A each time (we could only do that 3 times, and we would have
nothing left): we are subtracting one third of each new remainder, so what we are
subtracting is always getting smaller. Do you think this will get us to less than B? How
about repeatedly taking one millionth of what remains each time? Will that get us down
to less than B, regardless of how small B is? Believe it or not, the answer is yes. But I
leave it to the reader to find proof – it is a digression from the goals of this book.

310
THEOREM 2: If some multiple of magnitude W is greater than magnitude Q,
and a series approaches W, then the same multiple of some term in that series is also
greater than Q.

W
...
A B C T (We are bothering with this only because
we need it for the next Theorem. Don't
W W W worry, though, it's easy.)

3W
Q _
Q

Suppose we are given that 3W > Q, and we are also given a series of magnitudes A, B,
C etc., which approaches W (i.e. there is no limit to how close the terms in the series get
to being equal to W, even if none of them actually equal it).
I say that there is some term T in the series A, B, C such that 3T > Q.

[1] Take any part of the difference between 3W and Q which is less than a third of
that difference, and call it d.
Thus 3d < 3W – Q (we chose d this way)

[2] Since A, B, C etc. approaches W (given), therefore there is always a term in that
series differing from W by less than any specified amount. So take any term T in the
series which differs from W by less than d.
Thus W – T < d

[3] So 3W – 3T < 3d (multiplying both sides by 3)

but 3d < 3W – Q (Step 1)
so 3W – 3T < 3W – Q
thus 3T > Q

Q.E.D.

THEOREM 2 Remarks:

1. We chose 3W and then took less than a third of 3W – Q. There is nothing special
about 3 and one third. The exact same argument would work for 5W, as long as we also
take less than one fifth of 5W – Q, etc.

311
2. In Step 1 we “take less than a third of 3W – Q.” How do we do that? As a matter
of fact, as with Theorem 1, we will not be applying this Theorem except to magnitudes
that we can do this with. So all this Theorem needs to assert (for our purposes) is that
whenever we can find a way to take less than one third of the difference (or whatever
fraction, depending on the multiple of W we start with), there must be a term in series A,
B, C such that 3T > Q.

3. Obviously, if A, B, C is a growing series, then it is likewise true for every term

after T, such as V, that 3V > Q. And if A, B, C is a shrinking series, then the Theorem
would hold for every term before T.

THEOREM 3: If two growing series approach two quantities, and the

corresponding terms in the two series always have the same ratio, then the two quantities
they are approaching also have that ratio.

X
A B C
...

Z
c
a b
...

Given: Series A, B, C … etc. is a growing series approaching X,

Series a, b, c … etc. is a growing series approaching Z,
and A : a = B : b = C : c etc.

Prove: X : Z = A : a

Take any multiple of X and A, say 5 times each: 5X and 5A.

Take any multiple of Z and a, say 3 times each: 3Z and 3a.

Suppose that 5X > 3Z

[1] Since 5X > 3Z,

therefore in the series approaching X it is possible to take a term T so that
5T > 3Z (Thm.2)

[2] Let t be the term in the series approaching Z that corresponds to T.

Thus T : t = A : a (given)
Now since the series approaching Z is growing toward it (given), therefore every member
of it is less than Z, and therefore
t < Z

312
[3] Thus 3t < 3Z (multiplying both sides of the inequality by 3)
but 5T > 3Z (Step 1)
so 5T > 3t

[4] Now T:t = A:a (Step 2)

thus 5A > 3a (since 5T > 3t, Step 3)
Therefore if 5X > 3Z then 5A > 3a.
Likewise if NX < MZ then NA < Ma,
and if QX = RZ then QA = Ra.

That is, any random multiples of X and Z must compare the same way as the
corresponding multiples of A and a. Therefore
X:Z = A:a (Ch.5, Def.8)

Q.E.D.

THEOREM 3 Remarks:

I chose to speak about growing series in this Theorem not because there is something
special about them, but because we will be looking only at growing series in this Chapter.
The Theorem, in fact, would also apply as well to other kinds of approaching series.

THEOREM 4: Similar polygons inscribed in circles are to each other as the

squares on the diameters.

Let ABCD and abcd be similar polygons, B

b
each inscribed in a circle. Let diameters C
c
be drawn through corresponding points in
the polygons, say points b and B, making A a
diameters BM and bm.
d
D m
I say that ABCD : abcd = £BM : £bm. M

313
[1] Join AC, ac, CM, cm.

[2] Because similar polygons are made of similar triangles (Ch.6, Thm.14, Remarks),
therefore rABC is similar to rabc
so ∠BAC = ∠bac

[3] But ∠BAC = ∠BMC (both stand on BC; Ch.3, Thm.20)

and ∠bac = ∠bmc (both stand on bc; Ch.3, Thm.20)
so ∠BMC = ∠bmc
But ∠MCB = ∠mcb (both are right; Ch.3, Thm.24)
so rBCM is similar to rbcm,
since they have two angles equal to two angles (Ch.6, Thm.4).

[4] Thus rBCM : rbcm = £BC : £bc (Ch.6, Thm.16)

but ABCD : abcd = £BC : £bc (Ch.6, Thm.17)
so ABCD : abcd = rBCM : rbcm
but £BM : £bm = rBCM : rbcm (Ch.6, Thm.16)
so ABCD : abcd = £BM : £bm

Q.E.D.

THEOREM 5: Circles are to each other as the squares on their diameters.

R
D
C E d
c e Take any two circles. Call their diameters
T
AB and ab. I say that the circles are to
A B a b each other in the same ratio as the squares
on their diameters, i.e. as £AB to £ab.
m g
M G
k
K

[1] Bisect the semicircumferences in the circle AB at D and K, and inscribe square
ADBK. Now bisect the quarter circumferences in circle AB at C, E, G, M, and inscribe
the regular octagon ACDEBGKM. Thus we are making a series of inscribed regular
polygons, each having twice the number of sides as the one before it.

314
[2] Each new polygon takes up more than half of what was left of the circle by the
last polygon. For example, the square ADBK leaves 4 pieces of the circle such as the one
contained by straight line AD and arc ACD. But the octagon removes ΔACD from that
piece, which is more than half of it – because ΔACD is half of the rectangle ATRD (Ch.1,
Thm.33), and since this rectangle is greater than the circle's segment ACD, it follows that
rACD is more than half segment ACD. So each polygon in the series leaves some
amount of the circle left over, but the next polygon takes up more than half of that
leftover. Therefore the series of polygons is approaching the area of the circle (Thm.1,
Remark 2).

[3] Likewise the series of regular polygons inscribed in circle ab, made by repeatedly
bisecting arcs, is approaching the area of circle ab. And just as the square in circle AB is
similar to the square in circle ab, so every polygon in circle AB is similar to the
corresponding polygon in ab.

[4] But similar polygons inscribed in circles are as the squares on their diameters
(Thm.4). Therefore all the corresponding polygons in the two series have the ratio of
£AB : £ab. And therefore the two magnitudes approached by these two series also
have that ratio (Thm.3). That is
circle AB : circle ab = £AB : £ab.

Q.E.D.

THEOREM 5 Remarks:

1. In Step 2 I asserted that rectangle ATRD is more than the segment ACD. How do
I know that? I form the rectangle by drawing TCR tangent to point C. See if you can
complete the proof by doing the following:
(a) Prove that TR is parallel to AD. (This allows you to complete a rectangle
between TR and AD.)
(b) Argue from the fact that TCR is tangent that segment ACD of the circle must
be less than that rectangle.

D C
2. From this Theorem we can show how to make a L
square equal to an interesting curvilinear figure called a S
“lunule,” or little moon. Q
(i) Consider a circle with center M, diameter AMB, and CM A B
M
drawn perpendicular to AMB. Join AC and circumscribe a
circle about triangle ACM.

315
(ii) Since ∠AMC is right, therefore AC is the diameter of the circle ADCM (Ch.3). So
circle ADCM : circle M = £AC : £AB (by the present Theorem)
(iii) But £AC is the square ACBE, and £AB is the square on its diagonal, so
£AC : £AB = 1 : 2
thus circle ADCM : circle M = 1 : 2 (because of Step ii)
(iv) Thus circle M is double circle ADCM,
so one quarter circle M = one half circle ADCM
i.e. quadrant MAC = semicircle ADC
or areas S + Q = areas S + L
thus area Q = area L (subtracting S from each)
i.e. rAMC = lunule L

3. Since we can make a square equal to any rectilineal figure, we can thus make a
square equal to the lunule L. This is called quadrature, or squaring an area. You can see
why the quadrature of this and other kinds of lunules gave the ancient Greeks great hopes
of finding a way to make a circle equal to a square. The various attempts at the
“quadrature of the circle” over the centuries make up an interesting part of the history of
mathematics. The end of the story came in modern mathematics (only about a couple
hundred years ago), with a proof that it is actually impossible to construct a square equal
to a circle using nothing but straight lines and circles (the tools we are using in this
geometry book). Worse than that, even if you allow yourself all kinds of curves
described by algebraic equations, you still cannot make a square equal to a circle!

THEOREM 6: Any pyramid on a triangular base is divisible into two congruent

pyramids (each similar to the whole) and two equal triangular prisms.

TWO CONGRUENT PYRAMIDS SIMILAR TO THE WHOLE

V
Conceive a pyramid on triangular base ABC,
vertex V.

[1] Bisect its six edges at D, K, E, L, T, M. Join

DE, EK, KD, DT, TL, LD.
E
D

K
[2] Since the sides of the pyramid’s triangular
faces are all cut proportionally by these lines
T B joining the midpoints, therefore these lines are
A parallel to the edges of the pyramid. For example,
M DE is parallel to AB.
L

316
 So rVDE is similar to rVAB
but rDAT is similar to rVAB
so rVDE is similar to rDAT
And the corresponding sides AD and DV are equal,
so rVDE ≅ rDAT.

[3] Similarly rVDK ≅ rDAL.

[4] Now AL = DK (Step 3 congruence)

and AT = DE (Step 2 congruence)
and ∠LAT = ∠KDE (since KD║LA and AT║DE, Ch.9 Thm.12)
so rLAT ≅ rKDE

[5] Similarly rDLT ≅ rVKE

[6] Hence all the triangles containing pyramid VDKE are congruent to those
containing pyramid DALT. So these pyramids are congruent. And since they are
contained by triangles similar to the triangular facees of the whole pyramid on base ABC,
hence they are similar to the whole pyramid.

TWO EQUAL TRIANGULAR PRISMS

[7] Join KM, MT.

Now DK║LC (since DK║AC)
but DL║KC (since DL║VC)
so DKCL is a parallelogram

[8] Likewise LTMC is a parallelogram.

[9] So DK is parallel and equal to LC (opp. sides of DKCL)

and TM is parallel and equal to LC (opp. sides of LTMC)
so DK is parallel and equal to TM
hence DKMT is a parallelogram

[10] Since the opposite sides of parallelograms are equal,

thus DL = KC
and LT = CM
and TD = MK
so rDLT ≅ rKCM

[11] Since rDLT and rKCM are congruent, and their corresponding sides are sides
of parallelograms, therefore they contain a prism on LTMC as base.

[12] Likewise rDKE and rTMB are congruent, and their corresponding sides are
sides of parallelograms, so they contain a prism on rTMB as base.

317
 [13] Now rTMB is half parallelogram LTMC, since they stand on equal bases CM
and MB, and in the same parallels LT and CB. Hence the two triangular prisms standing
on these bases, being under the same height (since the plane of DKE is parallel to the
plane of CLTB), are equal. (Ch.9, Thm.28)

[14] So the whole pyramid is composed of the two congruent pyramids and these two
equal triangular prisms

Q.E.D.

THEOREM 7: When a pyramid on a triangular base is divided as in the last

Theorem, the two equal prisms are more than half the whole.

V [1] Reproduce the diagram of Theorem 6, and join

LM. Can you see that KLCM and ETMB are pyramids
congruent to VDKE and DALT? (For clarity I leave
E
some lines undrawn.)
D

K [2] The triangular faces of the whole pyramid had

B
their sides bisected, as at D, E, and T. Joining these
T
A midpoints divides each triangular face into four congruent
M triangles, each similar to the whole face.
L

[3] So rBMT ≅ rALT

but rALT ≅ rDKE (Thm.6, Step 4)
so rBMT ≅ rDKE
again rETB ≅ rVDE for the same reasons,
and rEMB ≅ rVKE for the same reasons again.
V

V
E

K
T D E
A B
B

L M
M
A B
C T
C

318
[4] Since the corresponding sides in these congruent triangles are equal,
hence MT = DK
and TE = DV
and EM = VK
so rTEM ≅ rDVK (SSS)

[5] Hence all the triangles containing pyramid ETMB are congruent to those
containing pyramid DALT,
so ETMB ≅ DALT.
Likewise KLCM ≅ DALT.
Now pyramids ETMB and KLCM are only parts of the two equal prisms, and so
pyramids VDKE and DALT, equal to those pyramids, are less than the two prisms. Since
VDKE and DALT, together with those two prisms, exhaust the whole pyramid, hence the
two prisms are more than half of it.

Q.E.D.

THEOREM 8: If two triangular pyramids under the same height are divided
each into their two prisms and two pyramids (as in the last Theorem), then the two prisms
in one are to the two prisms in the other as the base to the base.

Suppose you have two V v

pyramids whose triangular
bases ABC and abc are in the
same plane, and also their D K d k
vertices V and v are in a
parallel plane. Let them each E e
be divided, as in Theorem 6, A
N
a
n
c
C
into two congruent pyramids
each similar to the whole and T R t r

two equal prisms. B b

[1] Now the prism on TBRN with top edge DE is half the parallelepiped on that base
and with its top in the plane of DEK (Ch.9, Thm.28, Remarks). So too, the prism on
tbrn is half the parallelepiped on tbrn. So
prism TBRN : prism tbrn = parallelepiped TBRN : parallelepiped tbrn

319
[2] Since the original pyramids are under the same height, therefore the planes cutting
their edges in half are also of the same height, i.e. planes DEK and dek. Therefore the
parallelepipeds just mentioned are under the same height, and therefore they are to each
other as their bases (Ch.9, Thm.25). And therefore, the prisms which are their halves are
also as their bases. So
prism TBRN : prism tbrn = TBRN : tbrn

[3] And likewise the prism on NRC (with top DEK) is to the prism on nrc (with top
dek) as the base NRC is to the base nrc. That is,
prism NRC : prism nrc = NRC : nrc

[4] Now if we join TR, then rABC is divided into four congruent triangles ATN,
NTB, BRN, NRC. Likewise if we join tr, then Δabc is divided into its four congruent
triangles.
So TBRN : rABC = tbrn : rabc (both are the ratio 2:4
thus TBRN : tbrn = rABC : rabc (alternating)

and NRC : rABC = nrc : rabc (both are the ratio 1:4)
thus NRC : nrc = rABC : rabc (alternating)

[5] Now prism TBRN : prism tbrn = TBRN : tbrn (Step 2)

But TBRN : tbrn = rABC : rabc (Step 4)
so prism TBRN : prism tbrn = rABC : rabc

Also prism NRC : prism nrc = NRC : nrc (Step 3)

But NRC : nrc = rABC : rabc (Step 4)
so prism NRC : prism nrc = rABC : rabc

[6] So prism TBRN : prism tbrn = rABC : rabc (Step 5)

but prism NRC : prism nrc = rABC : rabc (Step 5)
so prisms TBRN + NRC : prisms tbrn + nrc = rABC : rabc

since the sums of things in the same ratio remain in the same ratio (Ch.5, Theorem 15).
And so the two prisms in the one whole pyramid are to the two prisms in the other whole
pyramid in the same ratio as the bases of the whole pyramids.

Q.E.D.

320
THEOREM 8 Remarks:

If we now divide the small pyramids on DEK and dek in the same way, the sums of their
corresponding prism-parts will be as their bases DEK and dek. So
2 prisms in pyramid DEK : 2 prisms in pyramid dek = DEK : dek (this Thm.)
But since rDEK is identical to rNRC, and rdek is identical to rnrc, therefore
DEK : dek = NRC : nrc
Therefore, putting together these two proportions,
2 prisms in pyr. DEK : 2 prisms in pyr. dek = NRC : nrc
But, as we saw, NRC is one fourth of ABC, and nrc is one fourth of abc. So
2 prisms in pyr. DEK : 2 prisms in pyr. dek = rABC : rabc
So even the little prisms in the leftover little pyramids also have to each other the ratio of
the original base ABC to the original base abc. Likewise, if we now take the little
pyramids into which DVEK and dvek are divided, and divide them, the equal prisms in
them will be as DEK to dek, and hence as ABC to abc. And therefore all the prisms so
made in pyramid ABC have to all the prisms so made in pyramid abc the ratio of rABC
to rabc.

THEOREM 9: Pyramids on triangular bases and under the same height are to
each other as their bases.
V X

Consider a pyramid with vertex V on G K N

L
base ABC, and another pyramid with
vertex X on base DEF, and suppose H M

their bases are in the same plane, and A

P
C D F
their vertices are in a parallel plane. S

I say that O Q R T

B E

pyramid V : pyramid X = ABC : DEF.

[1] Divide each pyramid into two congruent pyramids, similar to the whole, and two
equal triangular prisms, as before. In each pyramid, the 2 prisms take up more than half
the whole (Thm.7). And so if we now divide the smaller leftover pyramids in the same
way, such as pyramids VGHK and GAOP, their prisms will also take up more than half
of them. And they, too, will have two left over pyramids, and so on. So, continuing in
this way we will have a series of magnitudes approaching pyramid V, namely:

321
 1. The 2 prisms in pyramid V
2. The 2 prisms in pyramid V plus the 4 prisms in the two leftover pyramids
3. The 2 prisms in pyramid V plus the 4 prisms in the two leftover pyramids,
plus the 8 prisms in the four leftover pyramids etc.
In each step of this process, we take what we had before and add more than half of what
remained of pyramid V. Thus this series is approaching pyramid V (Thm.1, Remark 2).
So, too, there is a corresponding series of prisms approaching pyramid X.

[2] Now since, at any step in the series, all the prisms in pyramid V have to all the
corresponding ones in pyramid X the same ratio as ABC to DEF (Thm.8, Remarks),
therefore the magnitudes approached by these two series of prisms also have that same
ratio (Thm.3). That is
Pyramid V : Pyramid X = ABC : DEF.

Q.E.D.

THEOREM 9 Remarks:

In particular: Pyramids under the same height which are on equal triangular bases are
equal. For they must be as their bases (by this Theorem), which are equal.

THEOREM 10: Pyramids of the same height are to each other as their bases.

V X

Didn't we just do this? No – we

limited the last Theorem to pyramids
with triangular bases. Now we are C H
saying: the bases don't have to be 1
B
2 5
triangles. They can be any two D K
G
polygons at all. 3 A 4

E
L

Take any two pyramids V and X whose bases (ABCDE and GHKL) are in the same plane
and whose vertices are in a parallel plane, and it will follow that
Pyramid V : Pyramid X = ABCDE : GHKL.

322
[1] Pick any vertex on base ABCDE, say A, and join AC, AD, dividing the base into
triangles 1, 2, 3, and thus dividing pyramid X into three triangular pyramids on 1, 2, 3.
Similarly, divide base GHKL into triangles 4, 5, thus dividing pyramid X into two
triangular pyramids on 4, 5.
For shorthand, let P1 mean “The Pyramid on triangular base 1,” let P2 mean “The
Pyramid on triangular base 2,” and so on with the other triangular bases.

[2] Now P1 : P2 = Base 1 : Base 2 (Thm.9)

And so, adding each consequent to its antecedent,
P1 + P2 : P2 = Base 1 + Base 2 : Base 2 (Ch.5, Thm.15, Remarks)
And the consequents in this proportion are P2 and Base 2.

[3] But P3 : P2 = Base 3 : Base 2 (Thm.9)

And the consequents in this proportion are also P2 and Base 2.
Therefore, putting this proportion together with that in Step 2, and leaving out the
identical consequents, we have a proportion among the antecedents:

P1 + P2 : P3 = Base 1 + Base 2 : Base 3 (Ch.5, Thm.17)

And thus, adding each consequent to its antecedent (Ch.5, Thm.15), we have

P1 + P2 + P3 : P3 = Base 1 + Base 2 + Base 3 : Base 3

[4] Similarly, we can prove that all the triangular pyramids in Pyramid X have to any
one of them the same ratio as the whole base to the triangular base of that one, i.e.

P4 + P5 : P4 = Base 4 + Base 5 : Base 4

[5] But the consequents of this proportion and in the proportion from Step 3 form a
proportion, since
P3 : P4 = Base 3 : Base 4 (Thm.9)
And therefore the antecedents of the two proportions also form a proportion, and in the
same order (Ch.5, Thm.17). That is,
P1 + P2 + P3 : P4 + P5 = Base 1 + Base 2 + Base 3 : Base 4 + Base 5
which is the same as saying
Pyramid V : Pyramid X = ABCDE : GHKL.

Q.E.D.

THEOREM 10 Remarks:

Again, note the particular case that Pyramids under the same height and on equal bases
are equal.

323
THEOREM 11: Any triangular prism is composed of three triangular pyramids
of equal volume.

B
E

Let ABC and DEK be the triangular bases of

a prism. I say that it is divisible into three
equal pyramids having triangular bases.
C
K
[1] Join BD, BK, KA.

[2] Thus the prism is divided into three pyramids, namely

one with base DEK and vertex B (DEK*B for short)
one with base DKB and vertex A (DKB*A for short)
one with base AKB and vertex C (AKB*C for short)

[3] Now DEB*K = DAB*K

since they stand on equal bases (DEB and DAB each being half of parallelogram
ABED), and share a vertex, K (Thm.9, Remarks).
But DEB*K is the same as pyramid DEK*B
and DAB*K is the same as pyramid DKB*A
so DEK*B = DKB*A

[4] And DAK*B = AKC*B

since they stand on equal bases (DAK and AKC each being half of parallelogram
ACKD), and share a vertex, B (Thm.9, Remarks).
But DAK*B is the same as pyramid DKB*A
and AKC*B is the same as pyramid AKB*C
so DKB*A = AKB*C

[5] Thus DEK*B = AKB*C (Steps 3 and 4)

And so the three pyramids into which the whole prism is divided are equal to each
other.

Q.E.D.

324
THEOREM 11 Remarks:

1. Accordingly, any triangular pyramid is one third

of a triangular prism sharing its base and under the same
height. Using the same diagram above, consider pyramid
ABC*K, standing on base ABC. This pyramid is part of
the prism contained by ABC and DEK. But imagine
some other prism, Prism X, also standing on base ABC,
whose top triangle is somewhere else in the plane of DEK E
(just shift DEK over a bit within the same plane, and you D

get such a prism). So Prism X shares base ABC with

A
prism DEK, and also has the same height as it. Then it C
must be equal to it. For if we double triangle ABC into a
parallelogram, and complete the parallelepipeds on it, B

prism DEK and Prism X are half of these parallelepipeds,

which are clearly equal to each other (Ch.9, Thm.23).

But pyramid ABC*K is one third of prism DEK (by this Theorem).
So pyramid ABC*K is one third of Prism X, too.

2. More generally, though, any pyramid on any polygon as its base is one third of
the prism sharing that same polygon base and under the same height.
Take any pyramid, say on base ABCDE. Divide its base into triangles. Thus the
pyramid, too, is divided into three pyramids, one on each triangular part of the base. But
a prism standing on base ABCDE will likewise be divided into three prisms, one on each
triangular part of the base.
So if the prism has the same height as the pyramid, each triangular pyramid will
also have the same height as each triangular prism. And therefore, by the above Remark,
each triangular pyramid will be one third of the triangular prism sharing its base, and thus
all the triangular pyramids will be a third of all the triangular prisms, i.e. the whole
pyramid on ABCDE is one third of the whole prism on that base and under the same
height.

3. Since pyramids under the same height, regardless of what kind of bases they stand
on, are to each other as their bases (Thm.10), and since the prisms standing on their bases
and under the same heights are as those pyramids (being triples of them), it follows also
that Prisms standing under the same height are to each other as their bases.

325
THEOREM 12: Any cone is one third the cylinder sharing its base and height.

P Let circle ABCD be the base of a cone, and also of

A a cylinder having the same height (we are talking,
E G recall, about right cylinders and cones). I say that
the cone is exactly one third the volume of the
M cylinder.
D B
[1] Inscribe a square ABCD in the base circle,
and by bisecting the arcs inscribe a regular octagon
L K AGBKCLDE.
C

[2] As we saw in Theorem 5, each successive polygon (made by bisecting more arcs)
takes up more than half of what remained of the circle from the previous polygon, since,
for example, rADE is more than half of the segment of the circle contained by AD and
arc DEA.

[3] But since prisms under the same height are as their bases, therefore the right
prism on rADE is half the right prism on rectangle DMPA, each one having the height
of the cylinder. And since the portion of the cylinder standing on the segment contained
by AD and arc DEA is inside the right prism standing on rectangle DMPA, therefore it is
less than that prism. Therefore the prism on rADE is more than half that portion of the
cylinder. Thus the prisms standing on ABCD, AGBKCLDE, etc., constitute a series of
prisms each taking more than half of what was leftover of the cylinder by the previous
prism, and so this series of prisms approaches the volume of the cylinder (Thm.1
Remarks).

[4] Again, since pyramids under the same height are as their bases, therefore the
pyramid on rADE is half the pyramid on rectangle DMPA with that same vertex. And
since the portion of the cone standing on the segment in AD and arc DEA is inside the
pyramid on rectangle DMPA, therefore it is less than that pyramid. Therefore the
pyramid on rADE is more than half that portion of the cone. Thus the pyramids
standing on ABCD, AGBKCLDE, etc., constitute a series of pyramids approaching the
volume of the cone.

[5] But the pyramids on those polygon bases always have to the prisms on the same
bases the ratio of 1 to 3 (since any pyramid is one third of the prism sharing its base and
height). Therefore also the magnitude approached by the pyramids has to the magnitude
approached by the prisms the ratio of 1 to 3 (Thm.3). That is, the cone is one third of the
cylinder.

Q.E.D.

326
THEOREM 12 Remarks:

R V T
If we take a cross-section of the cone and cylinder through their axis,
we see that the cross-section of the cone is triangle AVC, and that of
the cylinder is rectangle ARTC. The cross-section of the cone is half
that of the cylinder. So
Base of Cone : Base of Cylinder = 1 : 1
Cross-Section of Cone : Cross-Section of Cylinder = 1 : 2
A C
Volume of Cone : Volume of Cylinder = 1 : 3

THEOREM 13: Cones of the same height are to each other as their bases.

Consider two cones of the same height, one on

base A, the other on base B.
A B
[1] Inscribe successive polygons in each as
before, beginning with a square, then doubling
the number of sides each time by bisecting the
arcs of the circles.

[2] Since the corresponding polygons in the two circles are always similar, therefore
they always have the ratio of the squares on the diameters (Thm.4).

[3] And since the pyramids on these similar polygons as bases, having the same
height, are to each other as their similar bases (Thm.10), therefore they are also always in
the ratio of the squares on the diameters (by Step 2).

[4] Now we saw in Theorem 12 that the series of pyramids in each cone approaches
the volume of the cone, and therefore the cones also have the ratio of the squares on the
diameters (Thm.3).

[5] But the squares on the diameters of the circles are in the same ratio as the circular
bases themselves (Thm.5). Therefore the cones are in the same ratio as their bases.

Q.E.D.

327
THEOREM 12 Remarks:

Since each cone is a third of the cylinder standing on its base and under the same height,
it likewise follows that Cones of the same height are to each other as their bases.

THEOREM 14: If a cylinder is cut by a plane parallel to its bases, then the two
resulting cylinders are to each other as their axes.

B A O X Y Z

Imagine a cylinder with axis AX, and suppose it is cut by a plane parallel to its bases,
resulting in two cylinders, one with axis AO, another with axis OX. I say that
cylinder AO : cylinder OX = AO : OX.

[1] Take any multiple of cylinder AO (and thus of its axis), say double it,
as cylinder BO.
Take any multiple of cylinder OX (and thus of its axis), say triple it,
as cylinder OZ.

[2] Clearly, if cylinder BO = cylinder OZ

then axis BO = axis OZ

and if cylinder BO > cylinder OZ

then axis BO > axis OZ

and if cylinder BO < cylinder OZ

then axis BO < axis OZ

[3] Therefore, taking any multiples of cylinders AO and OX, they will always
compare the same way as the corresponding multiples of their axes.
Therefore

cylinder AO : cylinder OX = AO : OX

Q.E.D.

328
THEOREM 15: Cylinders on equal bases are to one another as their heights.
Likewise for cones.

A
Q

X Z

Take any two cylinders on equal bases, one with axis (and thus height) AX, another with
a greater axis, BZ. I say that

cylinder AX : cylinder BZ = AX : BZ.

[1] Cut off ZQ = AX.

Pass a plane through Q parallel to the bases of cylinder BZ, thus completing
cylinder QZ.

[2] Since cylinder QZ has the same height as cylinder AX, and their bases are equal
circles, therefore cylinder QZ is congruent to cylinder AX.

[3] Cylinder QZ : Cylinder BZ = QZ : BZ (Thm.14)

[4] But Cylinder QZ = Cylinder AX (Step 2)

and QZ = AX (Step 1)
So let's substitute these in the last proportion from step 3:

[5] Cylinder AX : Cylinder BZ = AX : BZ

Q.E.D.

[6] As for cones, let AX and BZ be cones. Then if we complete the cylinders on their
bases and with heights AX and BZ,
Cone AX : Cone BZ = Cylinder AX : Cylinder BZ
since the cones are each a third of the cylinders (Thm.12).
But cylinders are to each other as their axes (Step 5 above), and so
Cone AX : Cone BZ = AX : BZ.

Q.E.D.

329
“HOOK”: SPHERICAL RINGS.

Take any pair of spheres, however unequal in size. Next inscribe a right cylinder in the
smaller of the two, as BCDE. Excluding the “caps” of the sphere, such as BAE, and also
the cylinder itself, what remains is a sort of ring with a bulging exterior face. Next,
inscribe a cylinder in the larger sphere, having the same height as that in the smaller (so
that GH = BC). Thus we will have formed a “ring” in the larger sphere, too. Now the
fun part: The volumes of the two “rings” are the same.

B E G L

C D H K

330
Chapter Eleven

The Five
Perfect Solids
 DEFINITIONS

1. A PERFECT SOLID is a convex polyhedron whose solid angles are equal and whose
faces are congruent regular polygons.

2. A TETRAHEDRON is a perfect solid contained by 4 equilateral triangles.

3. A CUBE is a perfect solid contained by 6 squares.

4. An OCTAHEDRON is a perfect solid contained by 8 equilateral triangles.

5. An ICOSAHEDRON is a perfect solid contained by 20 equilateral triangles.

6. A DODECAHEDRON is a perfect solid contained by 12 regular pentagons.

331
 THEOREMS

In this Chapter we will make the 5 perfect solids, prove that there are only 5 of them, and
learn some of their relationships and properties. We must begin with 7 preliminary Theo-
rems of plane geometry, however, which also happen to be of some interest in them-
selves.

THEOREM 1: If a line cut in mean and extreme ratio has added to it its greater
segment, the result is a whole line cut in mean and extreme ratio whose greater segment
is the original line.

Let AB be a straight line cut in mean

and extreme ratio at S, and let AS be
its greater segment. Extend it to Z so
that ZA = AS. I say that ZAB is also
cut in mean and extreme ratio, and AB
Z A S B is its greater segment.

[1] Clearly £AB = AB(AS + BS) (AS + BS = AB)

thus £AB = AB ⋅ AS + AB ⋅ BS (Ch.2, Thm.1)

[2] But AB : AS = AS : BS (given)

so AB ⋅ BS = £AS
since the square on a mean proportional line equals the rectangle contained by the ex-
tremes (Ch.6, Thm.12).

[3] Thus £AB = AB⋅AS + £AS (Steps 1 & 2)

i.e. £AB = AS(AB + AS) (Ch.2, Thm.1)
But since the side of a square is a mean proportional between the sides of any rectangle to
which it is equal (Ch.6, Thm.12), thus

(AB + AS) : AB = AB : AS

[4] But AS = ZA (we made it so)

thus AB + ZA : AB = AB : ZA
i.e. BZ : AB = AB : ZA

That is, ZAB is cut in mean and extreme ratio at A, and AB is the greater segment.

Q.E.D.

332
Remarks

Conversely, if QRT is cut in

mean and extreme ratio at R, and
RT is the greater segment, and if
we subtract the lesser segment
QR from the greater RT, cutting
it at S, then RST will also be cut Q R S T
in mean and extreme ratio. That (X)
is, the original greater segment
RT is now cut in the golden ratio at S, and RS, which was the original lesser segment, is
the new greater segment. To see it, we can just use the Theorem above. Cut RT in the
golden ratio at X so that RX is the greater segment of RT. Then if we add RX to RT, RT
will be the greater segment, and RX the lesser, by the above Theorem. But RT is the
greater segment and RQ is the lesser, since that is how the line is given. Therefore RX =
RQ. And we cut off RS equal to RQ, so that RS = RX also. Thus RS does in fact cut RT
in the golden ratio, and it is the greater segment of RT, just as RX was.

THEOREM 2: In a regular pentagon, any diagonal is parallel to the opposite

side, and the angle of the pentagon is one and one fifth of a right angle (or 108°).
A

B E
Given: ABCDE is a regular pentagon.
Prove: BE is parallel to CD, and ∠ABC = 108°.
P

C D

[1] Because of the regularity of the figure, it is clear by Side-Angle-Side that

ΔBCD ≅ ΔEDC (BD & CE are not drawn)

[2] Thus these triangles are equal in area.

But they stand on the same base.
Therefore they are in the same parallels. (Ch.1, Thm.33, Remark 3)
Therefore BE is parallel to CD.

[3] Let P be the center of the circumscribed circle. Thus the 5 triangles such as APB
are congruent (Side-Side-Side).

333
[4] Now the total angle-sum of these triangles is 10 right angles (i.e. 2 rights for each
triangle times 5, the number of triangles). If we subtract the angles around P, i.e. 4 right
angles, the remainder is the sum of the pentagon's angles, namely 6 right angles.
Since the 5 angles of the pentagon are all equal, each one is a fifth of 6 right an-
gles, or one and one fifth of a right angle.

[5] Since a right angle is 90°, therefore a fifth of a right angle is 18°, and so one right
angle and a fifth is 108°.
Thus the angle of a regular pentagon is 108°.

Q.E.D.

THEOREM 3: In a regular pentagon, if two diagonals cut each other, they cut
each other in mean and extreme ratio, and their greater segments equal the side of the
pentagon.

A
1 Given: Regular pentagon ABCDE. Diagonals AC
and EB cut each other at S.
2 3
E B
4 S Prove: AC and EB are cut in the golden ratio at S,
and ES = EA.

First, circumscribe a circle about ABCDE (Ch.4,

5 6 Thm.7 Remarks).
D C

[1] Now, by the previous Theorem, EB║DC and AC║ED, so that DESC is a paral-
lelogram. Thus
∠5 = ∠4 (opp. angles in a parallelogram)
but ∠3 = ∠4 (vertical angles)
so ∠5 = ∠3
or ∠EAB = ∠3 (∠5 = ∠EAB since the pentagon is regular)

[2] Notice ∠ABS is common to rEAB and rABS. Also, ∠EAB = ∠3.
So two angles in rEAB are equal to two angles in rABS.
Therefore rEAB is similar to rABS (Ch.6 Thm.4),
i.e. EB : BA = BA : BS

334
[3] Now since ∠1 and ∠6 are each at the circumference, standing on two of the 5
equal arcs, therefore
∠1 = ∠6 (Ch.3, Thm.21)
but ∠2 = ∠6 (since ES║DC)
so ∠1 = ∠2 (making rAES isosceles)
thus AE = ES

[4] So AE = ES (Step 3)
but AE = BA (being sides of the regular pentagon)
so BA = ES
Therefore, substituting ES for BA in our proportion from Step 2, we have

EB : ES = ES : BS

That is, ESB is cut in mean and extreme ratio at S (Ch.6 Def.9), and ES is the greater
segment. And we saw in Step 3 that ES is equal to EA, the side of the pentagon. Thus it
is clear that in a regular pentagon, the side has to the diagonal the golden ratio.

Q.E.D.

THEOREM 4: The sides of the hexagon and decagon inscribed in the same cir-
cle, when added together, make a whole straight line cut in mean and extreme ratio.

Given: Straight line BCD, with BC being a side

of the regular decagon in circle ABC, CD being
E equal to the side of the regular hexagon in circle
B A
3 1 ABC (and thus it is equal to the radius, so that
EC = CD, Ch.4 Thm.8).
C
Prove: BCD is cut in mean and extreme ratio at
F C.
2

Take the center of the circle, E (Ch.3, Thm.1).

D
Join EC, join EFD.

335
[1] Since ∠BEC is one central angle of the regular decagon, of which there are 10 in
the whole decagon, and 5 in each semicircle, therefore ∠CEA contains 4 such angles.
Thus 4∠BEC = ∠CEA
but 2∠ECB = ∠CEA (∠CEA is exterior to isosceles triangle ECB)
so 2∠ECB = 4∠BEC

[2] Thus ∠ECB = 2∠BEC (the halves of equals are equal)

but ∠ECB = 2∠CDE (∠ECB is exterior to isosceles triangle ECD)
thus 2∠BEC = 2∠CDE

[3] So ∠BEC = ∠CDE (the halves of equals are equal)

i.e. ∠1 = ∠2
but ∠3 is common to ΔBEC and ΔEDB
thus ΔBEC is similar to ΔEDB (Ch.6, Thm.4)

[4] So DB : BE = BE : BC (ΔBEC is similar to ΔEDB)

i.e. DB : CD = CD : BC (CD is equal to the radius BE)
i.e. BCD is cut in mean and extreme ratio at C, and CD, the side of the hexa-
gon, is the greater segment.

Q.E.D.

Remarks

K
1. Since BC and CD are related as lesser and greater segments B E

of a line cut in mean and extreme ratio, so too are BC and BE

(since BE = CD). Therefore if we subtract from the greater C

segment BE a part equal to the lesser segment BC (say BK),

then BKE will also be cut in the golden ratio (by the Remark
after the first Theorem of this Chapter), and BK, equal to BC (the side of the decagon),
will be the greater segment. Thus in any circle, the greater segment of the radius, when it
is cut in mean and extreme ratio, is equal to the side of the inscribed regular decagon.

2. Looking back at the diagram for the Theorem, one can see that CE bisects angle
BED as follows.
CD = CE (given)
so ∠CED = ∠2 (since rDCE is isosceles)
but ∠1 = ∠2 (Step 3 in the proof of the Theorem)
thus ∠CED = ∠1
i.e. CE bisects ∠BED, and thus CF is also an arc cut off by a side of the deca-
gon, and BC = CF.

336
THEOREM 5: In one circle, the square on the side of the regular hexagon plus
that on the side of the decagon equals the square on the side of the pentagon.

L
A Let ABCDE be a regular pentagon in a circle with
G center M. Bisect arcs AB, CD at G, T. Thus we
F
get arcs of the decagon, and AG is a side of the
decagon. Bisect arc AG at L, join ML, cutting
B E
AB at F. Join GB. Now I say that the square on
M
the side of pentagon ABCDE is equal to the
square on the regular hexagon in circle M, plus
the square on the side of the regular decagon, that
is
C D
T
£AB = £BM + £AG

[1] ∠BMF (or ∠BML) is at the center and stands on 1 ½ decagon arcs. Thus it is
double an angle at the circumference standing on that same arc (Ch.3, Thm.20), and equal
to an angle at the circumference standing on double that arc, i.e. standing on 1 ½ penta-
gon arcs. But ∠BAM (or ∠BAT) stands on 1 ½ pentagon arcs, and thus
∠BMF = ∠BAM.

[2] But ∠ABM = ∠BAM (since BM = MA)

thus rABM is similar to rBMF
so that AB : BM = BM : BF (Ch.6, Thm.4)
And since the square on a mean proportional line is equal to the rectangle contained by
the extremes (Ch.6, Thm.12), therefore
AB · BF = £BM

[3] Now rGMF ≅ rAMF (Side-Angle-Side)

so GF = AF
thus ∠AGF = ∠GAF
but ∠GAF = ∠GBA (since AG = GB in rAGB)
so rAGF is similar to rAGB (Ch.6, Thm.4)
thus AB : AG = AG : AF
thus AB · AF = £AG (Ch.6, Thm.12)

[4] Now £AB = AB · BF + AB · AF (Ch.2, Thm.1)

So, replacing these rectangles with the squares shown equal to them in Steps 2 and 3,
£AB = £BM + £AG

Q.E.D.

337
Remarks

1. From this fact, by the converse of the Pythagorean Theorem, it follows that BM and
AG, if placed at a right angle to each other, will form a right triangle whose hypotenuse is
equal to AB.
D
2. So if we cut the radius MC of a circle in mean and ex-
treme ratio at K, MK being the greater segment, and if
E K C ABCDE is a regular pentagon inscribed in the circle, then
M MK = the side of the regular decagon in the circle
(Thm.4, Remarks)
MC = the side of the regular hexagon
A B and £MK + £MC = £AB (by the present Theorem).

3. Also, since in one circle the side of the decagon is to the side of the hexagon in mean
and extreme ratio, i.e. as the greater segment is to the whole (Thm.4, Remarks), and since
the side of a regular pentagon is to its diagonal in mean and extreme ratio, i.e. as the
greater segment is to the whole (Thm.3), therefore, in a single circle,

Side of decagon : Side of Hexagon = Side of Pentagon : Diagonal of Pentagon.

THEOREM 6: If an equilateral triangle is inscribed in a circle, the square on its

side is three times the square on the radius.
A
Imagine an equilateral triangle ABC in a circle with
center D and radius DA.
I say that £AB = 3 £DA.

D Extend AD to E. Join BE.

B C [1] Since arc BEC is one third of the circum-

K
ference, thus arc BE is one sixth of the circumfer-
ence, so BE is the side of the hexagon, and BE =
E
DA.

[2] Since AE is a diameter, thus ∠ABE is right (Ch.3, Thm.25), and so

£AB = £AE – £BE (Pythagorean Theorem)
thus £AB = 4£DA – £BE (AE is bisected at D, so £AE = 4£DA)
thus £AB = 4£DA – £DA (BE = DA, Step 1)

[3] i.e. £AB = 3 £DA.

Q.E.D.

338
 Remarks

1. Notice that DE is bisected at K. Why?

Because BE = the radius of the circle (Step 1 in the proof above).
Thus BE = BD = DE.
Thus ΔBDE is an equilateral triangle.
Likewise ΔCDE is an equilateral triangle.
Thus ΔBDC ≅ ΔBEC by Side-Side-Side.
Thus ∠DBC = ∠EBC.
Thus ΔDBK ≅ ΔEBK by Side-Angle-Side.
Thus DK = KE.

2. It is impossible to make a 1-2-3 triangle in terms of the

A
lengths of the sides, since the sides of lengths 1 and 2 would
not, added together, be greater than the side of length 3 (see
G Ch.1 Thm.17). But it is possible to make a triangle the
D
squares on whose sides have the ratios of 1, 2, 3. How? Let
equilateral ΔABC be inscribed in a circle with radius BD.
B C On AB as diameter, describe a semicircle, and place BG in it
equal to BD. Join AG. Thus ∠AGB is a right angle.

So £BG + £AG = £AB

But £AB = 3 £BD (by the present Theorem)
So £BG + £AG = 3 £BD
Thus £AG = 3 £BD – £BG
i.e. £AG = 2 £BG (since £BD = £BG)
but £AB = 3 £BG (since BG = radius BD).
So if we call £BG "1" square unit of area, then we must call £AG "2," and we must call
£AB "3."
Notice, also, that since £AG = 2 £BG, it follows that AG is equal to the diago-
nal of the square on BG. So we could have made ΔAGB that way, namely by placing the
side and diagonal of a square at right angles to each other, and joining the hypotenuse.
But now we know something else about this 1-2-3 triangle, namely that if its hypotenuse
is equal to the side of an equilateral triangle inscribed in a circle, then its shorter leg is
equal to the radius of that circle.

339
THEOREM 7: The square on the diagonal of a regular pentagon, plus the
square on its side, equals five times the square on the radius of the circumscribing circle.

Take a regular pentagon AGBKC, with AB

G K one of its diagonals, and let D be the center of
D the circumscribing circle. I say that £BA +
£AC = 5£BD.

F
A C
E

[1] Join BD, extend it to E. Since AGBKC is a regular pentagon, thus diameter BDE
bisects AC at F, and so DF is perpendicular to AC (Ch.3, Thm.3).
Join EA. Thus EA is the side of the decagon.

[2] Now BE = 2DE

so £BE = 4£DE

[3] But £BE = £BA + £AE (∠BAE is right)

so £BE + £DE = £BA + £AE + £DE (+ £DE on both sides)

[4] But £BE = 4£DE (Step 2)

so 4£DE + £DE = £BA + £AE + £DE
i.e. 5£DE = £BA + £AE + £DE

[5] But the square on the side of the decagon plus that on the side of the hexagon
equals the square on the side of the pentagon,
i.e. £AE + £DE = £AC (Thm.5)

[6] So 5£DE = £BA + £AC (Steps 4 & 5)

or £BA + £AC = 5£BD (DE = BD)

Q.E.D.

340
Remarks

For the sake of brevity, let's use some shorthand in this remark:

R = the radius of a circumscribing circle.

E = the side of the equilateral triangle inscribed in it
P = the side of the regular pentagon inscribed in it
D = the diagonal of the regular pentagon inscribed in it

(a) Now £E = 3 £R (Thm.6)

so 5 £E = 15 £R (multiplying both sides by 5)

(b) But £P + £D = 5 £R (the present Theorem)

so 3 £P + 3 £D = 15 £R (multiplying both sides by 3)

(c) Thus 3 £P + 3 £D = 5 £E (Steps a & b)

Or, spelled out in words, Three times the square on the side of the regular pentagon plus
three times the square on its diagonal is equal to 5 times the square on the side of the
equilateral triangle inscribed in the same circle.

THEOREM 8: How to construct a tetrahedron, and contain it in a sphere.

V [1] Draw an equilateral triangle ABC

in the base plane. Circumscribe a circle
around it (Ch.4, Thm.4). Take its center
M. Join MA, MB, MC.
K

L
[2] Set up MP perpendicular to the
B base plane.
A Thus angles PMA, PMB, PMC
M
are all right angles.
C

341
[3] In the plane PMA, using A as your center and AB as your radius, draw a circle,
cutting MP at V. (Since AB is greater than AM, your radius will hit MP at some point V
above M).
So VA = AB.

[4] Join VB, VC.

Since ∠VMA = ∠VMB (both are right)
and MA = MB (both are radii of the base circle around ABC)
and VM is common
thus ΔVMA ≅ ΔVMB (Side Angle Side)
so VA = VB

[5] Likewise we can show that VA = VC.

So VA = AB (Step 3)
and VA = VB = VC (Step 4)
so the lines VA, VB, VC, AB, BC, CA are all equal to each other.
So the 4 triangular faces of the pyramid are actually 4 congruent equilateral triangles, and
therefore we have made a tetrahedron. Now to contain it in a sphere ...

[6] Bisect AV at K.
In plane AVM, draw KL perpendicular to AV, hitting VM at L.
Thus ΔLVK ≅ ΔLAK (Side Angle Side)
so LV = LA.

[7] But ∠LMA = ∠LMC (both are right)

and MA = MC (both are radii of the base circle around ABC)
and LM is common
thus ΔLAM ≅ ΔLCM (Side Angle Side)
so LA = LC
and LC = LB likewise.

[8] Therefore LV = LA = LB = LC (Steps 6 – 7). So L is the center of the sphere

containing the tetrahedron, and VL is the radius of it.

Q.E.F.

342
Remarks

1. If it wasn't clear by itself, what I mean by “the sphere containing the tetrahedron” is
the sphere whose surface passes through the 4 corners of the tetrahedron.

2. To get a good look at this thing, it's an excellent

idea to make it in three dimensions. A good material
for three dimensional models is the manila that file-
folders are made out of. Reproduce the accompanying
diagram of an equilateral triangle divided into four
equilateral triangles onto a piece of manila. Make your
version bigger than the little diagram. Cut it out, and
then fold along the three sides of the middle equilateral
triangle. Raise up the three outer equilateral triangles
all to one point, tape the edges together and … you
have a tetrahedron.

THEOREM 9: If a tetrahedron is contained in a sphere, the square on the diam-

eter is one and a half times the square on the tetrahedron’s side.
P

Begin with the same construction as before, in

V
Theorem 8.

[1] Now since both are right, K

∠VKL = ∠VMA L
and ∠AVM is common
so rVKL is similar to rVMA A
B
M
so VL : VK = VA : VM
thus £VL : £VK = £VA : £VM C

[2] But VA is the side of an equilateral triangle, and the radius of the circle around it
is equal to MA. Therefore
£VA = 3£MA (Thm.6)

[3] Now £VA – £MA = £VM (Pythagorean Theorem)

so 3£MA – £MA = £VM (£VA = 3£MA, Step 2)
or 2£MA = £VM

[4] Since £VA = 3£MA (Step 2)

and £VM = 2£MA (Step 3)
thus £VA : £VM = 3 : 2

343
[5] Thus £VL : £VK = 3 : 2 (Steps 1 & 4)
so 4£VL : 4£VK = 3 : 2
i.e. the square on 2VL (the diameter of the containing sphere) is to the square on 2VK
(the side of the tetrahedron) as 3 to 2.

Therefore the square on the diameter of the containing sphere is one and a half times the
square on the side of the tetrahedron.

Q.E.D.

Remarks

In Step 1 we assumed that if four straight lines are proportional, then the squares on them
are also proportional. Let's prove that. Suppose A, B, C, D are four straight lines, and
A:B = C:D
Then I say that £A : £B = £C : £D.

(a) First, make a 3rd proportional straight line, X, to A & B, so that

A : B = B : X. (Ch.6, Thm.9)
Thus A : X is the ratio double of A to B, and therefore
£A : £B = A : X (Ch.6, Thm.15).
Again, make a 3rd proportional straight line, Z, to C & D, so that
C:D = D:Z (Ch.6, Thm.9)
Thus £C : £D = C : Z (Ch.6, Thm.15).

(b) Now A:B = C:D (given)

but A:B = B:X (Step a)
so C:D = B:X

(c) Now C:D = B:X (Step b)

but C:D = D:Z (Step a)
so B:X = D:Z

(d) Now B : X = D : Z (Step c)

but B:A = D:C (inverse ratios of given proportion)
so A:X = C:Z (Ch.5, Thm.16)
thus £A : £B = £C : £D,
since these squares are as A : X and as C : Z by Step (a).

Q.E.D.

344
THEOREM 10: How to construct an octahedron, and prove the square on the
diameter of the containing sphere is two times the square on the side of the octahedron.

[1] Make a square ABCD.

D Join AC and BD, making X the
X C center of the circle circumscribing the
A
square.

B
[2] Set up XV perpendicular to the
plane of ABCD, cutting off XV = XA.

[3] So ∠VXA = ∠AXB (both are right)

and VX = XA (we made it so in Step 2)
and XA = XB (Step 1)
so ΔVXA ≅ ΔAXB (SAS)
so VA = AB

[4] Likewise VB, VC, VD are all equal to AB.

Therefore the four faces of the pyramid are equal equilateral triangles.
So, if we make another pyramid R just like pyramid V on the other side of base ABCD,
we will have a solid VR contained by 8 equal equilateral triangles, having 6 solid angles
each made of 4 plane angles of 60°. Thus we have our octahedron.

[5] And since VX = XA = XR, therefore X is the center of the sphere containing it,
and AC is the diameter.

[6] And since AC is the diagonal of square ABCD, therefore

£AC = 2£AB
and so the square on the diameter of the containing sphere is two times the square on the
side of the octahedron.

Q.E.D.

345
Remarks

1. You can make this solid figure out of manila using the accom-
panying pattern.

2. Notice that the octahedron is just two identical pyramids joined at their common
square base.

B 3. If we put two tetrahedrons together, would we

get a new perfect solid? It would be a solid con-
tained by 6 equal equilateral triangles, as in figure
ABCDE, but not all 5 of its solid angles would be
identical. Compare the solid angle at B to the one at
A, for example. The solid angle at B is made of 3
A C
faces (∠ABD, ∠DBC, ∠CBA), and the one at A is
made of 4 faces (∠EAC, ∠CAB, ∠BAD, ∠DAE).
D
Not only that, but the angle at A is “squished,” since
∠BAE is much greater than ∠CAD, making solid
angle A narrower in one direction and wider in an-
E
other. Thus it would not be possible for a solid,
made with angle A, to be of uniform convexity. Thus solid A is different from the angle
of the octahedron, because each angle of the octahedron stands on a square, and (in the
figure for Thm.10) ∠DVB = ∠AVC, since both are right angles (which is easily seen
since all triangles such as AXV are right isosceles). Hence the solid angle of the octahe-
dron is uniform.

346
THEOREM 11: How to construct a cube, and prove the square on the contain-
ing sphere's diameter is three times the square face of the cube.

H
[1] Make a square ABCD in the base
E G
plane.
F

[2] Set up BF perpendicular to the plane

ABCD (Ch.9, Thm.7).
D
Complete the squares ABFE and
A BCGF (Ch.1, Thm.35).
C Drawing parallels, complete the paral-
lelepiped contained by these 3 squares, and
B you have a cube.

[3] Join EC, CA.

[4] Now EC = AG (being the diagonals of rectangle EACG)

and AG = BH (being the diagonals of rectangle ABGH)
and BH = DF (being the diagonals of rectangle BDHF)
Thus all four diagonals of the cube are equal to each other, namely
EC = AG = BH = DF.

H G
[5] Now bisect EF & AB, HG & DC, and
M pass a plane through these midpoints.
E Bisect EH & FG, AD & BC, and pass
F
a plane through these midpoints.
Call the intersection of these 2 planes
K
ML. Each of the 4 diagonals of the cube is
bisected by K, the midpoint of ML (Ch.9,
D
C Thm.27). Therefore the 4 diagonals of the
A
L cube all pass through K, and K is equidistant
B from the 8 corners of the cube.

Therefore K is the center of the containing sphere, and EC is its diameter.

[6] In right triangle EAC,

£EC = £EA + £AC (Pythagorean Theorem)
but £AC = 2£AB (since AC is the diagonal of £AB)
i.e. £AC = 2£EA (since AB = EA)
thus £EC = £EA + 2£EA
i.e. £EC = 3£EA

[7] So the square on the sphere's diameter is three times the square on the side of the
cube, which is one face of the cube. Q.E.D.

347
Remarks

To make a model of a cube is easy enough. As with the tetra-

hedron, copy the accompanying pattern of squares, cut it out,
and fold up along the dotted lines, joining edges with tape.
Now you have a cube.

THEOREM 12: How to make an icosahedron.

The icosahedron is a bit more complicated, so we are going to build it in separate stages.
STAGE ONE: we will build the top and bottom "caps" of the icosahedron, each
being a pyramid built on a regular pentagon as base with 5 equilateral triangles as walls.
STAGE TWO: we will build the midsection or “drum” of the icosahedron, con-
tained by two pentagons and 10 equilateral triangles.
STAGE THREE: we will put these 3 parts together, assembling our icosahedron.

STAGE ONE: The "caps" of the icosahedron

D
[1] Draw a circle with center M, inscribe
regular pentagon ABCDE (Ch.4, Thm.7).
Draw MV perpendicular to the plane of
V
E
the pentagon.
C Cut off MV equal to the side of the dec-
agon (Ch.4, Thm.9) inscribed in the circle cir-
M cumscribing pentagon ABCDE. Of course, we
already know that MA, being the radius of that
circle, is equal to the side of the hexagon in-
A B
scribed in it.

[2] Since MA is the side of the hexagon, and VM is the side of the decagon, and since
MA and VM are drawn at right angles to each other, it follows that their hypotenuse, VA,
is equal to the side of the pentagon in the circle, namely AB (this Chapter, Thm.5).
So VA = AB.

[3] But MB, MC, MD, ME are also sides of the hexagon (i.e. radii of the same circle),
and therefore the hypotenuses VB, VC, VD, VE are also all equal to AB, the side of the
pentagon.

348
[4] Therefore ABV, BCV, CDV, DEV, EAV are all equal equilateral triangles (SSS).
And that is one of the "caps" to the icosahedron. We will need one for the top, and one
for the bottom.

STAGE TWO: The "drum" of the icosahedron.

D
E C
[5] Using the circle circumscribing
ABCDE, and using the radius of that cir-
A B cle (MA) as a height, complete a cylin-
der with circle ABCDE as the top of it.

[6] In the base circle of this cylinder,

L U K first draw a pentagon WXYUI identical
I Y to ABCDE and with each of its sides
N H parallel to each of the sides of ABCDE.
W X Thus ABXW is a rectangle, and AW and
G
BX lie on the cylinder's surface, parallel
and equal to its axis, and perpendicular to the circles of the cylinder.

[7] Bisect arc WX at G.

Thus GX = side of the decagon in the base circle.
But BX = height of the cylinder = radius of the base circle (Step 5)
i.e. BX = side of the hexagon in the base circle.
But BX is perpendicular to GX, since BX is perpendicular to the whole base plane (Step
6; see Ch.9, Thm.3). And therefore BXG is a right triangle, whose legs BX and GX are
equal, respectively, to the side of the hexagon in the base circle, and the side of the deca-
gon in the base circle. Therefore the hypotenuse BG is equal to the side of the pentagon
in the base circle (this Chapter, Thm.5).
i.e. BG = AB.
Likewise AG = AB. Thus ABG is an equilateral triangle.

U [8] Now, bisect arc XY at H, arc YU at K, arc UI at

L K
L, arc IW at N. Obviously GHKLN is another regular
Y
pentagon, with its sides equal to WX or AB. It will also
I
follow, just as in Step 7, that
ΔBGH, ΔBHC, ΔCHK, ΔCKD, ΔDKL, ΔDLE,
N H
ΔELN, ΔENA, ΔANG are all equilateral triangles, each
W X
having sides equal to AB.
G
[9] Thus we have a solid contained by two identical
regular pentagons, namely ABCDE and GHKLN, and by 10 equilateral triangles. The
pentagons are in parallel planes, but the sides of the bottom pentagon are not parallel to
those of the top one (but rather to its diagonals).

349
 STAGE THREE: Assembly of the icosahedron.

V
[10] We have made a "cap" in Stage One,
and added beneath it a "drum" in Stage Two.
Now just add another "cap" identical to the
D
E C
one made in Stage One underneath the
A
drum, with vertex Q.
B
Thus we have a complete solid con-
tained by 20 equal and equilateral triangles,
all having a side equal to AB, the side of the
L K pentagon with which we began. And every
solid angle on it is formed by 5 plane angles,
N H
G each of 60° (the angle of the equilateral tri-
angle). Thus the icosahedron is made.
Q

Q.E.F.

1
2

3 5
4 6
Remarks
7
8
The accompanying pattern can be used to make an icosahe-
9 dron. Draw your own larger version of it. Then cut it out,
10
and fold along the lines. Tip: triangles 1 through 5 will
11 make one "cap," just as triangles 16 through 20 will make
12 the other "cap." Triangles 6 through 15 will make the
13
"drum."
14

15 17
16 18

19
20

350
THEOREM 13: How to contain an icosahedron in a sphere.
V

D
E C Now let’s see how to wrap a sphere around our icosa-
A B hedron. First “imagine away” all the edges of it, re-
taining just the 12 vertices ABCDE and GHKLN and
L K
V and Q. That simplifies things!
N
G
H [1] Now, join VQ and bisect it at S.
I say that point S is the same distance from
Q every vertex of the icosahedron.
Choose any vertex, C. I say SC = SV.
V
[2] Clearly, if M is the center of circle ABCDE,
E D and J is the center of circle GHKLN, then MJ is the
M C axis of the cylinder, and VQ lies in line with it. And
A B
because our “caps” are of identical heights,
thus VJ = MQ.
S
L K [3] Also, VQ, being in line with the axis of the
J cylinder, is perpendicular to the planes of the circles,
N
G
H and therefore to every line in them that it cuts. Thus
VQ is at right angles to MC.
Q

[4] Since VM = side of decagon in circle ABCDE (made thus in Step 1 of Thm.12)
and MJ = side of hexagon in circle ABCDE (made thus in Step 5 of Thm.12)
thus VMJ is cut in mean and extreme ratio at M (Thm.4 of this Chapter)
i.e. VJ : JM = JM : MV

[5] i.e. MQ : JM = JM : MV (VJ = MQ, Step 2)

but JM = MC (height of cylinder = radius)
so MQ : MC = MC : MV

[6] So MC is a mean proportional between MQ and MV.

But MC is also at right angles to VQ (Step 3).
Therefore V, C, Q all lie on the circle having diameter VQ and thus center S (by Chapter
6 Thm.10). Therefore
SC = SV
Similarly we can prove that SB, SH, SL etc. are all equal to SV.
Therefore S is equidistant from every vertex of the icosahedron, and therefore it is
the center of the sphere containing the icosahedron.

Q.E.F.

351
Remarks

C 1. Containing the icosahedron in a sphere helps us to

B see its uniform convexity. It is also helpful, in under-
standing the figure’s radial symmetry, to notice that
every vertex in it can be regarded as the top of a “cap”
S F
with a regular pentagon as base. e.g. H can be seen as
H the top of a “cap,” QGBCK being a regular pentagon.
Let’s prove this quickly. Since S is the center of the
sphere, SB = SG, and since they are sides of the ico-
G K
sahedron, BH = GH, and SH is common. Thus
rSBH ≅ rSGH. So if we drop a perpendicular to
Q SH from B, and another from G, it is clear they will
hit SH at the same point, call it F. Likewise the per-
pendiculars to SH drawn from C, K, and Q will all land on F. But all the perpendiculars
to SH through F lie in one plane (Ch.9, Thm.3 Remarks). Therefore Q, G, B, C, K all lie
in one plane together, and thus define a pentagon. That it is equilateral is already clear,
since QG etc. are all edges of the icosahedron. And since a plane through a sphere cuts
out a circle, therefore the plane common to Q, G, B, C, K cuts out a circle on the surface
of the sphere containing the icosahedron, and therefore these 5 points all lie on one circle.
Now the equal sides of pentagon QGBCK each cut off equal arcs of this circle (Ch.3,
Thm.23), making it clear that QGBCK is also equiangular.

2. Notice that the diameter of the containing sphere is VQ,

and VQ = VM + JQ + MJ
or VQ = 2VM + MJ (VM = JQ)
i.e. the diameter of the containing sphere is equal to twice the side of the decagon in pen-
tagon ABCDE plus the side of the hexagon.

V
3. Let's compare the radius of the circle around the icosahedron's
M C
pentagon (MC) to the diameter of the containing sphere (VQ).
Recall, first, that we made the height of the icosahedron's midsec-
S
tion equal to the radius of the circle around the pentagon, i.e. MJ
= MC. And since S bisects MJ, it follows that MS is half of MC.
J

352
 Now £SC = £MS + £MC (Pythagorean Theorem)
so £SC = £(½ MC) + £MC (MS is equal to half of MC)
so £SC = ¼ £MC + £MC,
since the square on half of MC is one quarter of the square on the whole of MC.
5
Thus £SC = £MC (adding)
4
or 4 £SC = 5 £MC (multiplying both sides by 4)
i.e. 4 £SV = 5 £MC (SV = SC, being radii of the sphere)
or £VQ = 5 £MC (since VQ = 2SV)
So the square on the diameter of the containing sphere is equal to five times the square on
the radius of the circle around the icosahedron's pentagon.

THEOREM 14: How to make a dodecahedron.

D D
E E C E C
A A B A B
B P

L L M
O

N N G N G
K K F

H H
R

[1] Since the angle of a regular pentagon is 108° (Thm.2), therefore 3 of these is still
less than 4 right angles (i.e. less than 360°). So we can make a solid angle out of three of
them (Ch.9, Thm.20). Let it be done: ∠EAB, ∠BAN, ∠NAE are each 108°, the angle of
a regular pentagon, and they contain a solid angle at A.

[2] Cut the legs off equally, and complete the regular pentagons in each of our 3 an-
gles, namely ABCDE, ABGHN, ANKLE (Ch.4, Thm.7 Remarks)

[3] Now EC is parallel to NG (since each is parallel to AB by Thm.2)

but EC = NG (being diagonals in congruent pentagons)
Thus the lines joining their endpoints are also equal and parallel (Ch.1, Thm.30),
i.e. EN and CG are equal and parallel.

[4] Thus rGBC ≅ rNAE (Side-Side-Side)

and so ∠GBC = 108°.
By the same reasoning again, we can show that
∠KNH = 108°
and ∠LED = 108°

353
[5] Complete new pentagons in these angles, namely GBCMF, KNHSR, LEDPO.

[6] Now I say that points S, H, G, F are all in one plane.

For, by reasoning similar to that in Step 3,

SG║LE (since SG║KA, and KA║LE)
and GH║LD (since GH║NB, and NB║LD)
Therefore the plane of intersecting lines SG and GH is parallel to the plane of intersecting
lines LE and LD (Ch.9, Thm.11). That is, plane SGH is parallel to plane LED.

Again, reasoning in the same way,

FH║DE (since FH║CA, and CA║DE)
and GH║LD (already shown just above)
Therefore the plane of intersecting lines FH and GH is parallel to the plane of intersecting
lines DE and LD (Ch.9, Thm.11). That is, plane FGH is parallel to plane LED.

But there can be only one plane through straight line GH parallel to the plane of
LED (Ch.9, Thm.2 Remark). Therefore the planes of SGH and FGH, both parallel to the
plane of LED and both passing through GH, are in fact the same plane. Thus S, H, G, F
all lie in one plane parallel to the plane of LEDPO.
And SH = HG = GF, since they are sides of our pentagons,
And ∠SHG = ∠HGF = 108° (by the same reasoning as in Step 4),
So we are free to complete the pentagon in SHGF, namely SHGFT.

[7] Likewise T, F, M, Q are now all in one

D
plane, and we can complete a pentagon
E C
TFMQU. Again, by the same argument, R, K,
A P B
L, O are all in one plane, and we can complete
L M pentagon RKLOX.
O Q

[8] By reasoning similar to that in Step 6,

K
N G F we can prove now that points D, C, M, Q, P are
X U all in one plane. Therefore DCMQP is a penta-
H gon. And since PD = DC = CM = MQ (all be-
R T ing sides of our other congruent pentagons), and
S since ∠PDC = ∠DCM = ∠CMQ = 108° (by the
same reasoning used in Step 4), it follows that if we only join PQ, DCMQP is yet another
regular pentagon.

[9] Likewise XOPQU is a regular pentagon (reasoning the same way as in Step 8).

354
[10] Now RS is parallel to KH (in pentagon KNHSR)
and KH is parallel to EB (since KE = BH, & both parallel to AN)
so RS is parallel to EB
but EB is parallel to DC (in pentagon ABCDE)
so RS is parallel to DC
and likewise every side of RSTUX is parallel to a side of ABCDE. Therefore RSTUX,
like ABCDE, is a regular pentagon (and it is in a plane parallel to ABCDE).

[11] Therefore we have made a polyhedron contained by 12 equal regular pentagons,

all of whose solid angles are made of 3 plane angles each equal to 108° (or the angle of
the regular pentagon). Therefore we have made a dodecahedron.

Q.E.F.

Remarks

1. To make a model of a dodecahedron, reproduce and cut out

the accompanying figure (preferably enlarged) twice. Fold
each up into an open "cup," and place the two together to form
a dodecahedron.

2. In Step 8 I said that DCMQP are all in one plane, and that this can be proved similarly
to the reasoning in Step 6. Since it is complicated, you might want to see it all done out.
Well, all right then! Here goes.

First PD is parallel to OE (in PDELO)

next OE is parallel to RN (since EN = OR, & each is parallel to KL)
so PD is parallel to RN

and DC is parallel to EB (in ABCDE)

but EB is parallel to KH (since EK = BH, & each is parallel to AN)
so DC is parallel to KH

So PD, DC are parallel to RN, KH. But RN, KH intersect in the plane of pentagon
KNHSR. Therefore PDC are in a plane parallel to the plane of KNHSR (Ch.9, Thm.11).

Now QM is parallel to UF (in QMFTU)

and UF is parallel to RH (since HF = RU, & each is parallel to ST)
So QM is parallel to RH

And MC is parallel to BF (in CMFGB)

and BF is parallel to NS (since NB = FS, & each is parallel to HG)
so MC is parallel to NS

355
So QM, MC are parallel to RH, NS. But RH, NS intersect in the plane of pentagon
KNHSR. Therefore QMC are in a plane parallel to the plane of KNHSR.
Since PDC and CMQ are all in a plane parallel to KNHSR, and only one plane
through C is parallel to the plane KNHSR, therefore D, C, M, Q, P are all in one plane.

THEOREM 15: There is a cube hidden in every dodecahedron.

D
E
C
A B
Take the dodecahedron we just made, and
L
M join EC, CG, GN, NE.
I say that ECGN is a square.
N
K G
H F

R
T
S

[1] Bisect EC at L, AB at K, NG at V. Join LK, KV, VL.

C
[2] From the symmetry of the regular pentagon, it is
L
E obvious that
A K B NAKV ≅ GBKV
so ∠AKV = ∠BKV
Thus ∠AKV is right (since ∠AKV & ∠BKV are equal
G
and adjacent). Again, ∠AKL is right by a similar ar-
N
V gument.

[3] Thus AK is at right angles to both KL and KV, and therefore

AK is perpendicular to the plane of LKV (Ch.9, Thm.3).
But NV is parallel to AK.
Thus NV is also perpendicular to the plane of LKV (Ch.9, Thm.4).
So ∠NVL is a right angle (Ch.9, Def.1)

[4] But LV is parallel to EN (since EL = NV, and they are parallel)

so ∠ENV is also a right angle (Ch.1, Thm.25).
i.e. ∠ENG is right.
And since ECGN is a parallelogram (for EC = NG and they are parallel), it is now clear
that it is also a rectangle. But since its sides are all diagonals of congruent regular penta-
gons, its sides are also equal. Therefore ECGN is a square.

356
[5] Likewise NGTR is a square for all the same reasons, and it is equal to ECGN,
since they share a common side NG.
And thus, in fact, where two faces of the dodecahedron meet, such as at AB or
SH, they meet above a square. Since there are 6 pairs of faces on the dodecahedron,
therefore there are 6 equal squares that can be traced out along its surface by joining the
diagonals of its pentagons, and these 6 squares each share a side with four others, i.e. they
form a cube.

Q.E.D.

D
E C
Remarks A B

L M
With equal reason, ACFH is a square, and
if we pair off the faces differently, we get
another cube. There are, in total, five cu- K N G F
bes hiding in the dodecahedron, namely
one for every diagonal in pentagon H
ABCDE. R T
S

THEOREM 16: The diameter of the sphere containing the dodecahedron is the
diagonal of the cube hidden in it.

D
E C Consider the dodecahedron we have already
A B made. We saw in Theorem 15 that every diago-
M
nal in pentagon ABCDE is the side of a cube
L
hidden in the dodecahedron.

K N G F
[1] Thus RC is a diagonal of such a cube, with
H AC as its side. But LF is also a diagonal in it,
R T
and they bisect each other (Ch.9, Thm.27).
S

357
[2] But LF is also a diagonal of the cube with BD as a side, and SD is another diagonal
of that cube. Therefore LF and SD also bisect each other (Ch.9, Thm.27).

[3] And since LF has only one midpoint, which is also the midpoint of RC (Step 1), it
follows that RC, LF, SD all bisect each other.

D [4] Continuing in this way, we will find that

E C all the diagonals of all five cubes share the
A B same midpoint. Call it J. Thus J is equidistant
from all the corners of the cubes, i.e. from all
L M
the vertices of the dodecahedron. Therefore J
is the center of the sphere containing our do-
N G
decahedron, and the diameter of that sphere is
K F
any diagonal of one of the hidden cubes.
H
R T Q.E.D.
S

THEOREM 17: There are only 5 perfect solids.

[1] The fewest number of sides a face of a perfect solid can have is 3 sides, i.e. an
equilateral triangle. For there is no rectilineal plane figure with 2 sides or 1.

[2] But the most is 5 sides, i.e. a regular pentagon. For the angle of a regular hexa-
gon, having 6 sides, is 120° (double the angle of an equilateral triangle). And the fewest
number of faces which could form a solid angle is 3. But 3 of these angles of the hexa-
gon would add up to 360°, or 4 rights, and no solid angle can be made out of plane angles
that add up to 4 rights (Ch.9, Thm.17). So no solid angle can be made out of 3 regular
hexagons, and much less could any be made out of more than 3 regular hexagons.
And any regular polygon of more than 6 sides would have an angle more than
120°. And so no solid angle could be formed out any number of such angles, either. So
the greatest number of sides that can be found on the face of a perfect solid is 5.

[3] Therefore a perfect solid can be formed only out of

(a) Equilateral triangles
or (b) Squares
or (c) Regular pentagons

[4] The fewest number of equilateral triangles that form a solid angle is 3, and this is
the angle of the TETRAHEDRON.

[5] The next is 4, and 4 equilateral triangles form the angle of the OCTAHEDRON.

358
[6] The next is 5, and 5 equilateral triangles form the angle of the ICOSAHEDRON.
But it is not possible to form a solid angle out of 6 or more equilateral triangles,
since 6 × 60° = 360°, or four rights.

[7] The fewest number of squares that form a solid angle is 3, and this is the angle of
the CUBE. But 4 or more right angles cannot form a solid angle, since 4 × 90° = 360°, or
four rights.

[8] The fewest number of pentagons that form a solid angle is 3, and this is the angle
of the DODECAHEDRON. But 4 or more cannot form a solid angle, since the angle of
the regular pentagon is 108°, and 4 × 108° = 432°, which is more than four rights.

[9] Therefore these 5 are the only possible perfect solids.

Q.E.D.

Remarks

It is astonishing that when we limit ourselves to the use of only straight lines and circles,
we cannot make all the regular polygons in the world (we cannot, for example, make a 9-
sided regular polygon with only straight lines and circles), but we can make all the per-
fect solids!

THEOREM 18: A comparison of the five perfect solids regarding the number of
their vertices, edges, and faces.

For once, we will not need to prove anything, but only count. If you count the number of
vertices, edges, and faces in each solid, you can verify the entries in this table:

FIGURE VERTICES EDGES FACES

Tetrahedron 4 6 4
Cube 8 12 6
Octahedron 6 12 8
Dodecahedron 20 30 12
Icosahedron 12 30 20

359
 Notice that the cube and the octahedron have the same numbers, but in reverse order, and
the dodecahedron and icosahedron also form a symmetry that way. The tetrahedron’s
numbers are symmetrical by themselves.
This symmetry has an interesting consequence. In one way, we got a cube out of
a dodecahedron in Theorem 15. In another way, using the symmetries above, we can also
get an octahedron out of a cube. If we find the centers of the circles circumscribed about
each square face of the cube, the 6 resulting “centers” of the square faces are also the 6
vertices of an octahedron. This is clear from the uniformity with which those six points
are spread out from each other.
This works in reverse, too. Consider the eight faces of the octahedron, which are
equilateral triangles. If we find the “centers” of these triangles, namely the centers of the
circles circumscribed about them, we get 8 points spread apart in a uniform way, i.e. we
get the 8 vertices of a cube.
If we find the 4 centers of the 4 equilateral triangular faces of the tetrahedron, of
course we get the 4 vertices of another smaller tetrahedron.
If we take the 20 centers of the equilateral triangular faces of the icosahedron, we
get 20 points uniformly spread out, i.e. we get the 20 vertices of a dodecahedron. And if
we take the 12 pentagonal faces of the dodecahedron, finding the center of the circum-
scribed circle about each one, we get 12 points uniformly spread out, i.e. we get the 12
vertices of an icosahedron.
We can also get a tetrahedron out of a cube in much
B
the same way as we found a cube hidden in a dodecahe-
dron. Pick a vertex T on a cube. Across the three squares
meeting at T, draw their diagonals from T, namely TA, TB,
T
TC. And AB, BC, CA are three more diagonals on the re-
maining square faces. Now since the squares are all equal,
therefore these six diagonals are all equal, and therefore
C they contain four equal equilateral triangles. Therefore
A
ABCT is a tetrahedron. The other four vertices of the cube
are the vertices of another tetrahedron.

We can also get an octahedron out of a tetrahedron. Bisect

G
the six sides of a tetrahedron, EGHK, at L, M, N, O, P, Q.
Joining these six midpoints across each face of the tetrahe-
M
dron, it is easy to see we have made eight equal equilateral
L
triangles, and therefore L, M, N, O, P, Q are the vertices of
Q an octahedron.
P H One last note. The Swiss mathematician Leonhard
E Euler (1707 – 83) showed that if V is the number of verti-
O
N ces in a polyhedron, F the number of its faces, and E the
number of its edges, then V + F = E + 2. You can veri-
K fy this in the case of the perfect solids with the table above.

360
 THEOREM 19: If an icosahedron and a dodecahedron are contained in the
same sphere, the circle circumscribing the pentagonal face of the dodecahedron is equal
to the circle circumscribing the triangular face of the icosahedron.

M R

Given: ABCDE is the face of a dodecahedron.

D K
GHK is the face of an icosahedron.
MR is the diameter of the sphere con-
E C
taining both perfect solids.

G H
A B
Prove: The circles around ABCDE and GHK
T
are equal.
P S

[1] Join EC. So EC is the side of the cube inscribed in the sphere (Thms.15 - 16).
Therefore £MR = 3£EC (Thm.11)

[2] Let PS be the radius of the circle around “the pentagon” of the icosahedron, that
is, the pentagonal base of one of its “caps,” which is a pentagon with side equal to GK.
Therefore £MR = 5£PS (Thm.13, Remark 2)

[3] Thus 5£PS = 3£EC (Steps 1 & 2)

[4] Now cut PS in the golden ratio at T, with PT the greater segment (Ch.6, Thm.22).
Since the diagonal of a regular pentagon is to its side as a whole line is to the
greater segment of itself when it is cut in mean and extreme ratio (Thm.3 Conclusion)
Thus EC : CD = SP : PT
Therefore £EC : £CD = £SP : £PT (Ch.11 Thm.9, Remark)
So 3£EC : 3£CD = 5£SP : 5£PT (Ch.5, Thm.12)

[5] Now, taking the last proportion from Step 4, we can add the antecedents to the
consequents and still have a proportion (Ch.5, Thm.14, Remark 2). That is,
3£EC : 3£EC + 3£CD = 5£SP : 5£SP + 5£PT

[6] But in Step 3, we saw that 5£PS = 3£EC. So in the proportion of Step 5, we can
replace 3£EC with 5£PS. Let's do it:
5£SP : 3£EC + 3£CD = 5£SP : 5£SP + 5£PT

361
[7] Now PS is the radius of the circle in which GK is a side of the regular pentagon
(that is how we made PS, in Step 2). And PS is therefore the side of the hexagon in that
circle. And since PT is the greater segment of PS when it is cut in the golden ratio, PT is
the side of the decagon in that same circle (Thm.4 Remarks).
Therefore £PT + £SP = £GK (Thm.5)
or 5£PT + 5£SP = 5£GK (both sides × 5)

[8] Using this to simplify the proportion from Step 6, we get

5£SP : 3£EC + 3£CD = 5£SP : 5£GK
And, looking at this new proportion, it is obvious that
3£EC + 3£CD = 5£GK

[9] Now, in one and the same circle, 3 times the square on the diagonal of the in-
scribed pentagon plus 3 times the square on the side of that pentagon equals 5 times the
square on the inscribed equilateral triangle (Ch.11, Thm.7 Remarks). But EC is the di-
agonal of a pentagon, and CD is its side, so if Z is the side of the equilateral triangle in-
scribed in circle ABCDE, then
3£EC + 3£CD = 5£Z
But we have just shown that
3£EC + 3£CD = 5£GK (Step 8)
And therefore it follows that
GK = Z
That is, GK is equal to the side of the equilateral triangle inscribed in circle ABCDE. But
GK is the side of the equilateral triangle inscribed in circle GKH. Therefore circles
ABCDE and GKH must be equal.

Q.E.D.

THEOREM 20: When the five solids are all inscribed in the same sphere, their
order, from longest edge-length to shortest, is: tetrahedron, octahedron, cube, icosahe-
dron, dodecahedron.

For the sake of brevity, let's use a little shorthand:

Diam = the diameter of the containing sphere
Tet = the edge of the Tetrahedron
Oct = the edge of the Octahedron
Cube = the edge of the Cube

362
 3
[1] Now £Diam = £Tet (Thm.9)
2
So ⅔ £Diam = £Tet (double both, then one third both)

[2] But £Diam = 2 £Oct (Thm.10)

So ½ £Diam = £Oct

[3] So, since the square on the side of the tetrahedron is two thirds the square on the
diameter (Step 1), whereas the square on the side of the octahedron is only half the square
on the diameter (Step 2), it follows that
£Tet > £Oct
thus Tet > Oct
i.e. the side of the Tetrahedron is greater than that of the Octahedron.

[4] Now £Diam = 3 £Cube (Thm.11)

Therefore it is clear that the square on the side of the cube is only one third the square on
the diameter of the sphere, and so the side of the Tetrahedron is greater than that of the
Cube.
A
[5] And we know by Theorems 15 and 16
that the diagonal of any pentagonal face on the
G C
dodecahedron is the side of the cube inscribed in
the same sphere with the dodecahedron. And
K we know from Theorem 17 that if we draw a
circle around a pentagonal face of the dodeca-
B L hedron, the side of the equilateral triangle in-
scribed in that circle is the side of the icosahe-
E H dron inscribed in the same sphere with that do-
decahedron.

Let AGEHC be one pentagonal face of our dodecahedron. Draw a circle around
it, with center K, and inscribe equilateral triangle ABL in it. Join KA, KG, KB, KE.
Thus AE, being a diagonal of the dodecahedron’s pentagonal face, is the side of
the cube inscribed in our sphere, and AB is the side of the icosahedron.
Since ∠AKB stands on one third of the circumference from the center of the cir-
cle, therefore ∠AKB = 120°. But ∠AKE stands on two fifths of the circumference from
the center of the circle, and therefore ∠AKE = 144°. And yet both stand on less than a
semi-circumference, and therefore it follows that
AE > AB
i.e. the side of the Cube is greater than the side of the Icosahedron.

[6] But ∠AKG is only one fifth of 360°, i.e. 72°, whereas ∠AKB is 120°. Therefore
AB > AG
i.e. the side of the Icosahedron is greater than the side of the Dodecahedron.

Q.E.D.

363
Remarks

The order of the 5 solids in this Theorem also happens to put them in order of (1) in-
creasing surface area, and (2) increasing volume, when all are inscribed in the same
sphere. Two more interesting Theorems, whose proofs I leave to the reader, are these:

(a) When inscribed all in the same sphere, the surface of the Icosahedron is to that of the
Dodecahedron as the edge of the Cube is to the edge of the Icosahedron.

(b) When inscribed both in the same sphere, the volumes of the Dodecahedron and Ico-
sahedron have the same ratio as their surface areas.

“HOOK”: A SQUARE SECTION OF A TETRAHEDRON.

Given a tetrahedron ABCV, can you see how to cut it with a plane so that the section is a
square?

B
H
A
P
R G

364