MA1004: GEOMETRY (YEAR 1)

LECTURE NOTES

TIMOTHY LOGVINENKO

1. Introduction
The word “geometry” comes to us from ancient Greek
gewmetrÐa = gew(“geo”, earth) + metrÐa(“metria”, measuring)
and as it suggests the science of geometry originates from the kind
of questions that preoccupied the humanity since times immemorial –
which one of two given patches of land is bigger? And in the beginning
it was all about taking a contemporary equivalent of a tape measure and
finding out. All measuring tools are subtly different, however, while all
scientists have the same inexorable tendency to abstractify the prob-
lems they contemplate. By the Hellenistic period geometry had well
established itself as a science about the principles of earth-measuring
(traditionally contemplated upon whilst sitting in the shade of an olive
tree). It studies the properties (shape, size, etc.) of points, lines and
other idealised versions of real world objects and the properties of their
positions relative to each other (distance, angle, etc.).
In this course, designed to serve as a gentle introduction to this
venerable old subject, we aim:
• To give an overview of plane Euclidean geometry, with focus on
proofs
• To introduce some basic notions of spherical geometry, empha-
sising its differences with Euclidean geometry.
• To practice drawing diagrams and use them as means to con-
struct proofs
• To develop intuition and visualisation in 3 dimensions
A rigorous mathematical proof requires one to logically deduce the
result you want to prove from the results you’ve already proven (lem-
mas and theorems) and the base set of assumptions you’ve started
with (axioms). Circa 300 B.C. Euclid of Alexandria, a famous greek
geometer, proposed in his immortal treatise StoiqeØa(“Stoicheia”, Ele-
ments) to rigorously deduce the whole existing body of results in plane
geometry from the following set of five axioms:
(1) Through any two points there passes a unique line
(2) It is possible to extend any line segment continuously in a
straight line to a larger line segment
1
2 TIMOTHY LOGVINENKO

(3) It is possible to draw a unique circle of any given radius around

any given point
(4) All right angles are equal to each other
(5) If a straight line crossing two straight lines makes interior angles
with them on the same side which add up to less that π then
these two lines if continued indefinitely will eventually meet on
that side.
He made a pretty good effort of it - “Elements” dictated the way geom-
etry was taught in Europe for centuries to come. But he didn’t get it
completely right. In the course of deducing 6 books worth of theorems
in plane geometry he made a number of implicit assumptions which
didn’t actually follow from these five axioms. Famously, in the very
first proposition of “Elements” a pair of circles “obviously” have to in-
tersect each other. Later on, there is a number of arguments based on
superimposing one geometrical figure upon another, a procedure which
has no rigorous foundation in the above axioms.
In 19th century it was realised that Euclid’s axioms are actually
insufficient to prove all the theorems in “Elements”. Finally, in 1899
a German mathematician David Hilbert proposed a set of 20 axioms
(21 originally, but one of them was later shown to follow from the
other 20) from which all known theorems of Euclidean geometry can
be rigorously deduced.
Due to the time constrains, we can’t afford to follow either Hilbert’s
rigorous or even Euclid’s semi-rigorous approach to plane geometry. I
shall therefore cut a number of corners when proving the theorems in
the course, appealing at times to intuition, and at times to the fact
that due to some small advancements in mathematics since the time of
300 B.C. we now have a number of handy tools, such as the machinery
of real numbers, at our disposal. You, however, should be able to prove
things in exercises and exam questions fully rigorously by appealing to
the axioms and the results which were proven during the lectures.
Acknowledgements: I am very grateful to David Mond whose
excellent set of lecture notes helped me to plan out this course and
served as a basis for these present notes.

2. Motivation. A comparison with the coordinate

geometry
Here’s a typical theorem of Euclidean geometry:
Theorem. Given any three points which are not collinear1 there is
a unique circle which passes through all three of them.

1Aset of two or more points is said to be collinear if there exists a straight line
which contains all of them. Note that by Euclid’s first axiom such line is necessarily
unique.
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 3

One can try and approach this theorem by the methods of coordinate
geometry. The unique circle of radius r centered at the point (p, q) is
given by the equation
(x − p)2 + (y − q)2 = r2 .
To ask that this circle passes through three given points (x1 , y1 ), (x2 , y2 )
and (x3 , y3 ) is to ask that
(x1 − p)2 + (y1 − q)2 = r2
(x2 − p)2 + (y2 − q)2 = r2
(x3 − p)2 + (y3 − q)2 = r2
so the problem reduces to solving the above system of equations in
three unknowns: p, q and r. A system of three linear equations in
three unknowns certainly has under certain conditions2, but this is a
system of quadratic equations!
This illustrates the main weakness of the coordinate geometry ap-
proach: it allows to turn any geometric problem into a bunch of equa-
tions to solve, but sometimes solving them may be harder than to solve
the original problem geometrically. Indeed, suppose we ask:
Problem. Given three lines, can we construct a circle tangent to all
three of them?

This is exactly the sort of question that Euclidean geometry helps

to solve very visually and elegantly.

3. Distances and angles

Question 1:What is “the distance between two given points”?
One approach is simply to ask our plane to come equipped with an
abstract function
d : Plane × Plane → R≥0
2its determinant has to be non-zero
4 TIMOTHY LOGVINENKO

which satisfies a number of axioms, such as the triangle inequality. This

function then serves as a definition of distance:
/
point p1 function d / distance between p1 and p2
/
point p2

For example, our usual notion of distance on the coordinate plane R2

yields the function
p
d((x1 , y2 ), (x2 , y2 )) = (x1 − x2 )2 + (y1 − y2 )2
This approach leads eventually to an abstract notion of metric spaces,
which can be pretty counter-intuitive to deal with.
We want to model our plane geometry, to some extent, on the real
world. We therefore would like our notion of distance to correspond
to an intuitive idea, that the distance between two points is the length
of the straight line joining them. By Euclid’s first axiom there is a
unique straight line passing through any two points, and therefore a
unique line segment joining them. We say that the distance between
two points is the length of this line segment.
Question 1 21 : How do we measure it?
Given a line segment it is easy enough to construct a segment of twice
the length. Or thrice, or four times the length. With a little ingenuity
we can also construct a segment of half the length. Et cetera. This sort
of thinking led men to develop the theories of integer, of rational and
then of real numbers. In a sense, real numbers were invented precisely
for us to be able to measure any distance along a line. Having this
machinery at our disposal makes our job much easier.
Let us fix, therefore, once and for all a line segment of “unit” length.
Then, given any line with a choice of the point 0 and a choice of the
“positive” direction along the line, the machinery above gives us a
one-to-one correspondence between the points on the line and the real
numbers
 
  
 Points on a line 

Real numbers ←→
R 
 -2 0 1 3.1415... 


This correspondence is precisely such that the length of the line segment
between the points 0 and 1 is our unit length, and the direction from
0 to 1 is our chosen positive direction.
Therefore, given any pair of points P1 and P2 in our plane, we can
define the distance between them as follows. First, we draw the unique
straight line through them. Then we choose P1 to be the point 0 and
the direction from P1 to P2 to be the positive direction. The real
number corresponding to P2 is then the distance between P1 and P2 .
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 5

We shall denote this by d(P1 , P2 ) or by |P1 P2 |, the latter reflecting that

it is the length of the line segment P1 P2 .
Question 2: Given three points A, B, C in the plane,
what is the angle ∠ABC, i.e. the angle between line
segment BA and line segment BC?

One might be somewhat puzzled by Euclid’s fourth axiom, which

asserts that all right angles are equal. To understand this, we have to
consider Euclid’s original definition of a right angle. Euclid said that
if a straight line intersects another straight line in a way which makes
two angles on one side equal then these equal angles are said to be right
angles.

So the fourth axiom effectively asserts that all straight angles, angles
formed by two halves of a straight line, are equal (being a sum of two
right angles). Equivalently, it asserts that all angles formed by going
around a point in a complete circle
6 TIMOTHY LOGVINENKO

are equal (being a sum of four right angles). Therefore any two angles
which are the same fraction of a complete circle are equal.
This answers the question above in a fashion which matches our
intuition - the angle between line segments BA and BC is uniquely
determined by the fraction of a circle around B which they cut out.
Question 2 21 : How do we measure it?
We could simply define the value of the angle to be this fraction.
However, above we’ve made a choice of a unit length and it gives us
for free a choice of a “unit” circle around every point in our plane –
the circle of radius 1. It is more convenient to define the value of the
angle between BA and BC to be the distance we travel from BA to
BC along the circumference of the unit circle around B. Since the full
circumference of a unit circle is of length 2π, this effectively defines the
value of an angle to be 2π times the fraction of a complete circle cut
out by it.
More precisely, if we have a unit circle with a choice of the point
0 and with a choice of a direction, clockwise or counter-clockwise3,
the machinery of real numbers gives us a one-to-one correspondence
between the points on the circle and the real numbers in the interval
between 0 and 2π:
 


 Points on a circle 


 


 π/2 


π/4

 

  
 

Real numbers ←→
π 0 
0 ≤ x < 2π 
 1 

 


 


 


 

3π/2
 

Therefore, given any three points A, B and C such that neither A

nor C coincides with B,

we first extend both BA and BC to straight lines

3By assuming, intuitively, that we have universal notions of “clockwise” and

”counter-clockwise” in our plane we make implicit use of notions of orientation and
of orientability, a rather sophisticated machinery of its own
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 7

and then we draw a unit circle around B and mark by A0 the point
where this circle intersects the line BA on the same side of B as A.
Similarly, mark by C 0 the point where the circle intersects the line BC
on the same side of B as C.

C'

A'

We now define the value of ∠ABC to be the real number corresponding

to C 0 if we choose A0 to be the point 0 on the circle and the direction
around the circle to be anti-clockwise.
An important point: ∠ABC is not the same thing as ∠CBA! The
definition above ensures that by ∠ABC we always mean the angle from
BA to BC anti-clockwise.

Similarly, ∠CBA is the angle from BC to BA anti-clockwise. And since

to go from BA to BC anti-clockwise and then from BC to BA anti-
clockwise is to come a full circle, it follows that ∠ABC + ∠CBA = 2π,
or in other words

∠CBA = 2π − ∠ABC.

Finally, note that as a complete circle is an angle of 2π, any right

angle is an angle of π2 and every straight angle is an angle of π.

Lemma 1. Let a and b be a pair of intersecting lines. Then the two

angles where we go counter-clockwise from a to b are equal.
8 TIMOTHY LOGVINENKO

b
β
α
α'
a

Proof. We have
∠α + ∠β = π (from a straight angle formed by line a)
0
∠β + ∠α = π (from a straight angle formed by line b)
and so we conclude that α = α0 . Q.E.D. 4

Definition 1. Given two intersecting lines a and b we denote by ∠ab
the value of either of the 2 equal angles where we go from a to b anti-
clockwise (cf. Lemma 1) and refer to it as the angle from a to b.
Definition 2. Two lines a and b are said to be perpendicular to each
other if they intersect at right angles, that is ∠ab = ∠ba = π/2. We
denote this by a ⊥ b.
4. Isometries and congruences
What does it mean to say that two geometrical objects are “equal”?
Our intuition tells us that two objects in a plane are equal if we can
move one on top of the other so that they match perfectly. Below
we make this notion precise by giving a mathematical procedure for
“moving one object on top of the other”.
Definition 3. A map from the plane to the plane is a rule which sends
each point in the plane to some other point in the plane. Given such a
map f and a point P in the plane we denote by f (P ) the point where
f sends P :
f : Plane → Plane
P 7→ f (P )
Before giving examples of this we need to establish the following
notion:
Definition 4. Given any line segment AB its perpendicular bisect
is the unique line which passes through the midpoint of AB and is
perpendicular to AB.
4
The acronym Q.E.D. which you may often see concluding a mathematical

íper êdei deĩxai

proof stands for “quod erat demonstrandum” which is Latin for “what was to be

OED
demonstrated”. Its origin lies in the Greek phrase , often abbreviated
, which means “what was required to be proved” and which concludes every
proof in Euclid’s “Elements”.
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 9

We have established in the previous section that there is a unique

line passing through a given point and making a specified angle with a
given line through this point. This ensures that a perpendicular bisect
exists and is unique. For those unconvinced by such abstractions, here
is a nice and explicit construction. Given a line segment AB draw two
circles centered at A and B, both of radius |AB|. Then draw a straight
line through the two points where these circles intersect.

A B

We will prove later on in the course that, as is evident on the diagram

above, the resulting line is the perpendicular bisect of AB.
We now proceed to give examples of maps from the plane to itself:

Example 1. (1) Define the reflection rl in l to be the map

which sends any point P to the unique point rl (P ) such that l
is the perpendicular bisect of the line segment P rl (P ). In other
words, we drop a perpendicular from P to l and then extend it
by the same distance again on the other side:

P l

rl (P)

(2) Given a point O in the plane define the dilation DO,2 with
centre O and scale factor 2 to be the map which leaves
the point O fixed and sends any point P 6= O to the unique
point D0,2 (P ) which lies on the continuation of the line segment
OP in the direction of P at the distance 2|OP | from O.
10 TIMOTHY LOGVINENKO

DO,2(P)

DO,2(Q) P
Q

For an arbitrary α ∈ R we define the dilation DO,α with

centre O and scale factor α analogously.
(3) Given a point O in the plane define the rotation RO,θ with
centre O and angle 0 ≤ θ < 2π to be the map which ro-
tates the plane about O through an angle of θ in an anti-
clockwise sense. More precisely, for every point P 6= O define
RO,θ (P ) to be the unique point P 0 on the circle with centre O
and radius |OP | such that ∠P OP 0 = θ.
Q RO,θ(P)

P
θ
θ
RO,θ(Q) O

(4) Given a vector ~v = AB~ (an oriented line segment) define the
translation T~v by ~v to be the map which translates every
~
point in the plane by AB:
B
v
A

v Tv (P)

P
E.g. if we work in coordinate plane R2 then given a vector
~v = (v1 , v2 ) define
T~v : R2 → R2
(x, y) 7→ (x + v1 , y + v2 )
Of course, an abstract map from the plane to itself is not guaranteed
to preserve geometrical figures. It can map a triangle to something
which doesn’t in the least resemble one. If we want to use maps to
make precise our intuitive notion of “moving one figure on top of the
other” we must demand for a map to preserve geometrical properties of
a figure, that is - lengths and angles. It turns out, that is sufficient to
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 11

demand that a map preserves distances, for then it necessarily preserves

angles:
Definition 5. A map f : Plane → Plane is said to be an isometry 5
or distance-preserving if for any two points A and B in the plane
we have
d (A, B) = d (f (A), f (B)) .
Question: Which of the four types of maps in Example 1 are isome-
tries?
Answer: Reflections, rotations and translations are. Dilations are
clearly not, with exception of dilations of scale factor 1 which are just
identity maps.
Lemma 2 (“Isometries preserve angles”). Let f : Plane → Plane be an
isometry. Then there are two possible cases:
(1) For any three points A, B and C in the plane we have
∠ABC = ∠f (A)f (B)f (C).

f(C)

f(A)

f(B)

We call such isometries orientation-preserving.

(2) For any three points A, B and C in the plane we have
∠ABC = ∠f (C)f (B)f (A).

f(A)

f(B)
f(C)
We call such isometries orientation-reversing.
Proof. In a nutshell, the proof goes like this: isometries preserve lengths,
therefore they take unit circles to unit circles. An angle was defined
in terms of a length of an arc going counter-clockwise on a unit cir-
cle. An isometry, having to preserve lengths, must take this arc to an
arc of the same length on the image circle. This new arc either still
5 The word ‘isometry’ is derived from Greek imetrÐa (isometria) which means
“of equal measure”.
12 TIMOTHY LOGVINENKO

goes counter-clockwise (Case 1 above) or it now goes clockwise instead

(Case 2 above). Q.E.D.
“But wait!”, a skeptical reader might exclaim here. “You only know
that isometries preserve distances between points. These are the lengths
of straight lines. Why do isometries have to preserve the lengths of cir-
cle arcs too?” This is because we can actually approximate any circle
arc by a collection of straight line segments and by increasing the num-
ber of used segments we can approximate with any necessary precision.
Settling this technical point takes up most of the proof below.
And so: just as in Section 3 consider a unit circle around B, let A0
be the point where it intersects line BA on the same side of B as A,
and similarly for C 0 :

C'

A'

_
By definition, the value of ∠ABC is the arc length | A0 C 0 |. As our

first and crudest approximation of this arc we take straight line segment
AC

C'

A'
B

So set d1 = |AC| and observe that as the shortest path between any
two points is the straight line joining them we must have

_
d1 < | A0 C 0 |.

Now subdivide this arc in two equal parts and approximate each of
the halves by the corresponding straight line segment:
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 13

C'

A'
B
_
Our second approximation of arc | A0 C 0 | is this two-segment path

and we set d2 to be its length, i.e. the sum of the lengths of its two
segments. Observe, again, that as the shortest path between any two
points is the straight line joining them we must have
_
d1 < d2 < | A0 C 0 |.

We now subdivide each of the two halves of the arc in two again
and obtain a four-segment approximation d3 . Then an eight-segment
approximation d4

C' C'

A' A'
B B
and so on, obtaining at n-th step an 2n−1 -segment approximation dn .
The resulting sequence of approximations satisfies
_
d1 < d2 < d3 < d4 < · · · < | A0 C 0 |

and it is not too difficult to show that as n grows these multi-segment

paths get arbitrarily close to the arc they approximate, in other words
_
| A0 C 0 | = lim dn .
n→∞

Our isometry f takes this whole construction to an identical con-

struction approximating the image arc from f (A0 ) to f (C 0 ) on the unit
circle around f (B). Since f preserves the lengths of straight line seg-
ments we must have dn = f (dn ) and therefore
14 TIMOTHY LOGVINENKO

_ _
| A0 C 0 | = lim dn = lim f (dn ) = | f (A0 )f (C 0 ) |.
n→∞ n→∞

There are now two possibilities. First one is that the image arc goes
counter-clockwise from f (A0 ) to f (C 0 ) on the unit circle around B, and
therefore its length is the value of ∠f (A)f (B)f (C). We have then
_ _
∠ABC = | A0 C 0 | = | f (A0 )f (C 0 ) | = ∠f (A)f (B)f (C)

which corresponds to the Case 1 in the statement of this lemma. The

other possibility is that the image arc goes clockwise from f (A0 ) from
f (C 0 ), i.e. it goes counter-clockwise from f (C 0 ) to f (A0 ) and therefore
its length is the value of ∠f (C)f (B)f (A). We have then
_ _
∠ABC = | A0 C 0 | = | f (A0 )f (C 0 ) | = ∠f (C)f (B)f (A)

which corresponds to the Case 2.

It remains only to show that this behaviour must stay the same for
all angles in the plane. In other words, if f is orientation-preserving
on one angle, it can’t be orientation-reversing on another angle and
vice versa. This is a simple continuity argument: the behaviour of
f can’t change if we shift each of the points A, B and C by some
distance which is small compared to the size of the triangle ABC.
This is because if the behaviour of f does change, then the point f (C)
would suddenly “jump” to the other side of line f (A)f (B) - and being
distance-preserving f can only shift f (A), f (B) and f (C) by the same
small distances we’ve shifted A, B and C by. But we can now keep
doing this again and again, until we’ve moved A, B and C to any three
points in the plane we like. Therefore the behaviour of f must be the
same for any triple of points in the plane. Q.E.D. 
We now see that isometries are the transformations of our plane
which preserve the properties of geometrical objects. We can therefore
make precise our intuitive notion that two objects are equal if we can
“superimpose one upon the other and match them up exactly”:
Definition 6. We say that two geometrical figures are congruent if
there exists an isometry taking one to the other. We denote this relation
by symbol ∼=. If we specify a vertex order on the figures then we demand
for an isometry to preserve it, e.g. 4ABC ∼ = 4A0 B 0 C 0 only if there
exists an isometry taking A to A , B to B and C to C 0 .
0 0

But now we run into a problem: clearly any two points in the plane,
or any two line segments of the same length, are “equal” and should
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 15

therefore be congruent to each other. But how do we show that, how

do we show that there exists an isometry taking one to the other? For
that matter, what is there to say that any non-trivial isometries (i.e.
isometries which are not just identity maps) exist per se? The reflec-
tion maps or rotation maps we’ve defined above - we haven’t yet got
enough facts established to rigorously prove that they indeed preserve
distances. Moreover, the usual such proofs (as we will see later on in
this course) use triangle congruence criteria, and those are established
with arguments along the lines of “these two line segments are of equal
length, therefore superimpose one against the other...” - in other words,
they implicitly assume existence of appropriate isometries.
This is one of the problems with Euclid’s five axioms we’ve mentioned
before. We need, in fact, two extra axioms telling us some basic facts
about an existence of isometries:
Axiom 6. Any two line segments of equal length are congruent: if
A, B, C and D are four points in the plane such that |AB| = |CD|,
then there exists an isometry taking A to C and B to D.
Observe that we can do this by a combination of a translation and
~ which by definition takes
a rotation. First we do a translation by AC,
point A to point C. Now the points C and A coincide, while B and
D lie on the same circle of radius |AB| = |CD| around them. We can
therefore do a rotation and match them up. Unfortunately, as discussed
above, we haven’t yet established that translations and rotations are
isometries.
Axiom 7. Give a line l and a point P there exists an isometry which
leaves l fixed and moves P to the other side of l.
Observe, again, that a reflection in l does precisely that. But we
haven’t yet established that reflections are isometries either.

5. Triangle congruences
For the clarity of the exposition we assume that all our triangles
are non-degenerate, that is - that their vertices are not collinear. An
enthusiastic reader is encouraged to try and extend the results of this
section to degenerate triangles, it is not very difficult.
Notation: In a triangle 4ABC we denote by ∠A the internal angle
of the triangle at the vertex A and by α the value of ∠A. Similarly,
we denote by β the value of ∠B, the internal angle at B, and by γ the
value of ∠C, the internal angle at C:
16 TIMOTHY LOGVINENKO

B
β

γ
α C
A
Similarly, in a triangle 4A0 B 0 C 0 we use ∠A0 , ∠B 0 and ∠C 0 to denote
its internal angles at A0 , B 0 and C 0 , respectively. We further use α0 , β 0
and γ 0 to denote the values of these angles.
Note, that in any 4ABC we have 0 < α, β, γ < π as internal angles
of a triangle are clearly less that a straight angle:
B

α C
A
Definition 7. A triangle 4ABC is clockwise (resp. anti-clockwise)
oriented if moving from A to B to C takes you clockwise (resp. anti-
clockwise) around the points in the interior of the triangle:
B B

C C
A A
Exercise 1. Verify that an isometry f : Plane → Plane is orientation-
preserving (resp. orientation-reversing) if and only if for every non-
degenerate 4ABC the orientation of 4f (A)f (B)f (C) is the same as
(resp. opposite to) orientation of 4ABC.
Note that if 4ABC is clockwise oriented, then
α = ∠CAB, β = ∠ABC, γ = ∠BCA
and if 4ABC is anti-clockwise oriented, then
α = ∠BAC, β = ∠CBA, γ = ∠ACB.
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 17

Lemma (Lemma 2 12 ). Isometries preserve internal angles of a triangle.

More precisely, let 4ABC and 4A0 B 0 C 0 be a pair of triangles in a
plane. If 4ABC ∼ = 4A0 B 0 C 0 then
α = α0 , β = β 0 , γ = γ 0

B A'
α' C'
β γ'

γ β'
α C
A B'
Proof. Let f be the isometry taking 4ABC to 4A0 B 0 C 0 . Observe,
that the angle f (∠A) to which f takes ∠A is either the internal angle
at A0 in 4A0 B 0 C 0 and then f (∠A) = α0 < π or the external angle at
A0 in 4A0 B 0 C 0 and then f (∠A) = 2π − α0 > π.
Now apply Lemma 2. If f is orientation-preserving, then the value
of f (∠A) is α and therefore f (∠A) has to be the internal angle at A0
as α < π. Therefore α0 = f (∠A) = α. If f is orientation-reversing
(this is the case depicted on the diagram) then f (∠A) = 2π − α, and
we see that f (∠A) has to be the external angle at A0 as 2π − α > π.
Therefore f (∠A) = 2π − α0 and we conclude that α = α0 . 
We now proceed to prove standard criteria for a pair of triangles to
be congruent:
Lemma 3 (SAS: Side-Angle-Side). Let 4ABC and 4A0 B 0 C 0 be a pair
of triangles in the plane. If
|BA| = |B 0 A0 |, |BC| = |B 0 C 0 | and β = β 0

B A'
C'
β

β'
C
A B'
then
4ABC ∼
= 4A0 B 0 C 0 .
Proof. Since |BC| = |B 0 C 0 | there exists by Axiom 6 an isometry f such
that f (B) = B 0 and f (C) = C 0 . By Axiom 7 we can also assume that
f (A) is on the same side of B 0 C 0 as A0 . By Lemma 2 12 the value of
∠f (B 0 ) in f (4ABC) equals β, the value of ∠B in 4ABC:
18 TIMOTHY LOGVINENKO

f(A)
A'
C'= f(C)

β
β'
B' = f(B)

By assumption β = β 0 , therefore the values of ∠B 0 in 4f (A)B 0 C 0 and

of ∠B 0 in 4A0 B 0 C 0 are equal. Since A0 and f 0 (A) lie on the same side
of B 0 C 0 this implies that lines B 0 f (A) and B 0 A0 actually coincide:

f(A)

A'
C'= f(C)

'
β=β

B' = f(B)

So f (A) and A0 on the same line through B 0 and on the same side of
B 0 C 0 . By assumption we also have |BA| = |B 0 A0 | and since isometries
preserve lengths, we further have |B 0 f (A)| = |B 0 A0 |. So f (A) and A0
lie also at the same distance from B 0 and must therefore coincide. So
f (A) = A0 , f (B) = B 0 and f (C) = C 0 . As f is an isometry, we conclude
that triangles 4ABC and 4A0 B 0 C 0 are congruent. Q.E.D. 

A warning:
side

angle

side
SAS is a criterion for congruence of triangles.
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 19

sid
e

side
angle

ASS is not.
Do not confuse the two.
Lemma 4. Let l be a line in the plane. The reflection rl in l is an
isometry.
Proof. Let A and B be a pair of points in the plane. Set A0 = rl (A) and
B 0 = rl (A). Set LA and LB to be the midpoints of AA0 and BB 0 . By
definition of rl the line l is perpendicular to AA0 and BB 0 and passes
through LA and LB . Finally, set O to be the intersection of AB 0 and l.
Case 1: A and B are on the same side of l
A

LA LB
O
l

B'

A'
We have |LB B| = |LB B 0 | and ∠LB both in 4BOLB and in 4B 0 OLB
equals π/2. Therefore 4BOLB ∼ = 4B 0 OLB by SAS. Consequently
|OB| = |OB 0 | and ∠LB OB = ∠B 0 OLB
Similarly, 4AOLA ∼
= 4A0 OLA by SAS and we have
|OA| = |OA0 | and ∠AOLA = ∠LA OA0
By construction AB 0 and l are straight lines intersecting at O. Hence
∠AOLA = ∠B 0 OLB by Lemma 1, and so
∠LB OB = ∠B 0 OLB = ∠AOLA = ∠LA OA0 .
Let us denote the common value of these four angles by α. Since
∠LA OLB is a straight angle we have
∠BOA + 2α = ∠AOLA + ∠BOA + ∠LB OB = π
∠A0 OB 0 + 2α = ∠LA OA0 + ∠A0 OB 0 + ∠B 0 OLB = π
20 TIMOTHY LOGVINENKO

and we conclude that ∠BOA = ∠A0 OB 0 = π − 2α.

We have now established that |OA| = |OA0 |, |OB| = |OB 0 | and
∠BOA = ∠A0 OB 0 . Therefore 4AOB ∼ = A0 OB 0 by SAS and hence
|AB| = |A0 B 0 |. Q.E.D.
Case 2: A and B are on the opposite sides of l
Exercise! (Hint: On the diagram above

∠A0 OB = ∠A0 OB 0 + ∠B 0 OB = π

and therefore A0 OB is actually a straight line). 

Lemma 5 (ASA: Angle-Side-Angle). Let 4ABC and 4A0 B 0 C 0 be a

pair of triangle in the plane. If |BC| = |B 0 C 0 |, β = β 0 and γ = γ 0

B A'
C'
β γ'

γ β'
C
A B'
then
4ABC ∼
= 4A0 B 0 C 0 .

Proof. Since |BC| = |B 0 C 0 | by Axioms 6 and 7 there exists an isometry

f such that f (B) = B 0 , f (C) = C 0 and f (A) and A0 lie on the same
side of B 0 C 0 .

f(A)
A'
C'= f(C)

β
β'
B' = f(B)

The next step is just like in Lemma 3: by assumption β = β 0 , so B 0 f (A)

and B 0 A0 make the same angle with with B 0 C 0 . Since f (A) and A0 lie
on the same side of B 0 C 0 , it follows that they actually lie on the same
line B 0 A0 through B 0 :
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 21

f(A)

A'
C'= f(C)
γ' γ
'
β=β

B' = f(B)
Similarly, by assumption γ = γ 0 and therefore f (A) and A0 also lie on
the same straight line C 0 A0 through C. But now f (A) and A0 lie on both
B 0 A and CA0 . Since two distinct straight lines can intersect at most one
point A0 and f (A) must coincide. We have therefore f (A) = A0 , f (B) =
B 0 and f (C) = C 0 . Since f was taken to be an isometry, we conclude
that triangles 4ABC and 4A0 B 0 C 0 are congruent. Q.E.D. 
Lemma 6. Let 4ABC be a triangle in the plane. Then |AC| = |BC|
if and only if α = β.
C C

A B A B
Proof. “If ” direction: Suppose α = β. Then 4ABC ∼ = 4BAC by
ASA, and therefore |AC| = |BC|.
“Only if ” direction: Suppose |AC| = |BC|. Then 4ABC ∼
= 4BAC
by SAS, and therefore α = β.
Q.E.D. 
Definition 8. If a triangle has two sides which are both of the same
length, it is called an isosceles triangle. If all three sides of a
triangle are equal, it is called an equilateral triangle6.
Exercise 2. Show that in an equilateral triangle all angles are equal
to each other.
6The word ‘isosceles’ is derived from Greek Êelàc (isoskeles) which means
“of equal legs”. The word ‘equilateral’ is derived from Latin ‘aequilateralis’ which
means ‘of equal sides’.
22 TIMOTHY LOGVINENKO

Lemma 7. Let 4ABC be an isosceles triangle with |AB| = |CB|. The

perpendicular bisector of AC then coincides with the bisector of ∠B.
B

A C
Proof. Exercise! (Hint: Extend the bisector of ∠B until it intersects
AC. Then use triangle congruence to show that it makes right angles
with AC and cuts it in half.) 
Lemma 8 (SSS: Side-Side-Side). Let 4ABC and 4A0 B 0 C 0 be a pair
of triangle in the plane. If |AB| = |A0 B 0 |, |BC| = |B 0 C| and |CA| =
|C 0 A0 |

B A'
C'

C
A B'
then
4ABC ∼
= 4A0 B 0 C 0 .
Proof. By assumption |BC| = |B 0 C 0 | and so by Axioms 6 and 7 there
exists an isometry f such that f (B) = B 0 , f (C) = C 0 and f (A) and A0
lie on the opposite sides of B 0 C 0 .

A'
C'=f(C)

B' =f(B) f(A)

 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 23

By assumption |C 0 A0 | = |CA| and since f is an isometry |CA| =

|f (C)f (A)|. Therefore 4A0 f (A)C 0 is isosceles and by Lemma 6
∠A0 = ∠f (A0 ) in 4A0 f (A)C 0 .
Similarly, the assumption |A0 B 0 | = |AB| implies that 4A0 f (A)B 0 is
isosceles and
∠A0 = ∠f (A0 ) in 4A0 f (A)B 0 .
Adding the above two equalities together we get
(∠A0 in 4A0 B 0 C 0 ) = (∠f (A0 ) in 4f (A)B 0 C 0 ) .
Since also |A0 B 0 | = |f (A)B 0 | and |A0 C 0 | = |f (A)C 0 | we have by SAS
4A0 B 0 C 0 ∼
= 4f (A)B 0 C 0 .
But 4f (A)B 0 C 0 is the image of 4ABC under isometry f , i.e.
4f (A)B 0 C 0 ∼
= 4ABC.
We conclude that 4A0 B 0 C 0 and 4ABC are congruent. Q.E.D. 

6. Circles and triangles

Lemma 9. For any two points A and B in the plane the perpendicular
bisector of AB is the locus7 of points equidistant from A and B.

A B

Proof. Let O be the midpoint of AB. Let P be a point in the plane.

(1) Suppose P is equidistant from A and B.
P

A O B
7The word ‘locus’ is a Latin word meaning “place” (plural: loci). In geometry it
means a collection of all points sharing a specified property.
24 TIMOTHY LOGVINENKO

Then 4AP O = 4BP O by SSS. Therefore ∠P OA = ∠BOP

and since ∠P OA and ∠BOP add up to a straight angle, it
follows that they are both right angles. So OP is perpendicular
to AB and by construction O is the midpoint of AB. Therefore
OP is the perpendicular bisect of AB. Q.E.D.
(2) Suppose P lies on the perpendicular bisect of AB.
P

A O B
Then 4AP O = 4BP O by SAS. Therefore |AP | = |BP | i.e. P
is equidistant from A and B. Q.E.D.

Theorem 10. In any 4ABC the perpendicular bisectors of its three
sides are concurrent, i.e. meet at a point. The point O where they meet
is the centre of the unique circle which passes through A, B and C.

A
lAC

C
lAB O

l BC

Proof. Let lAB , lBC and lAC be the perpendicular bisectors of AB, BC
and AC, respectively. Then by Lemma 9

lAB = { the locus of points equidistant from A and B }

lBC = { the locus of points equidistant from B and C }
lAC = { the locus of points equidistant from A and C } .

Let O be the unique point where lAB and lBC intersect. Then, by
above, O is equidistant from A and B, and also O is equidistant from
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 25

B and C. Therefore
|OA| = |OB| = |OC|
and as, in particular, |OA| = |OC| point O must also lie on lAC . More-
over, the circle of radius |OA| and center O clearly passes through A,
B and C.
Suppose there exists another circle passing through A, B and C and
let O0 be its center. But then O0 is equidistant from A, B and C, so
it belongs to each of the lines lAB , lBC and lAC . Since any two lines
intersect at no more than one point, point O0 must coincide with O.
Q.E.D. 

With this geometric picture in mind, we can now easily solve the
problem of finding the circle passing through three given points in
coordinates:
Exercise 3. Let A = (x1 , y1 ), B = (x2 , y2 ) and C = (x3 , y3 ) be any
three points in R2 .
(1) Find the gradient of AB and of BC.
(x2,y2)

Δy
Δy gradient =
Δx
(x1,y1)
Δx
(2) Find the midpoints LAB and LBC of AB and AC.
(3) Recall that the equation of a line of gradient k passing through
a point (a, b) is
y = kx + (b − ka)
and recall that perpendicular lines have gradients whose prod-
uct is −1. With this in mind, find the equations of perpendic-
ular bisects lAB and lBC of AB and AC.
(4) Find the point O where lAB and lBC intersect. Find the distance
|OA|. These are by Theorem 10 the center and the radius of
the requisite circle.
(5) We should check that |OA| = |OB| = |OC|. For this, check that
the formula for |OA| which you’ve obtained in the previous step
should be invariant under any permutation of indices. That
is - swapping xi with xj and yi with yj shouldn’t change the
formula. Why is it enough to check that?
Definition 9. A line l is tangent to a circle C at point P if it
meets C only at P .
26 TIMOTHY LOGVINENKO

O
C

Lemma 11. Let l be a line and P a point not on l. Then there exists
a unique line through P which is perpendicular to l, and the shortest
distance from P to l is along this line.
Proof. Existence: Let Q be any point on l. Let P 0 be the unique
point such that
∠l, QP = ∠QP 0 , l and |QP | = |QP 0 |.
Then draw the straight line segment P P 0 and let O be the point where
P P 0 intersects l. We then have 4OQP ∼= 4OQP 0 by SAS
P P

O
Q l Q l

P' P'
and therefore
|OP | = |OP 0 | and ∠P OQ = ∠QOP 0 .
By construction the angle ∠P OP 0 is straight, so we have also
∠P OQ + ∠QOP 0 = π
and it follows that
π
∠P OQ = ∠QOP 0 =
2
i.e. line P P 0 is perpendicular to l.
Recall now that by construction
∠OQP = ∠P 0 QO.
So if P Q is also perpendicular to l, then ∠P QP 0 = π, i.e. P QP 0 is in
fact a straight line. But then O and Q must coincide, as otherwise lines
P P 0 and l would intersect in two distinct points. Since Q was taken to
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 27

be any point on l, it shows O is the unique point on l such that P O is

perpendicular to l. This demonstrates that the line P P 0 is the unique
line through P perpendicular to l.
Finally, observe that if Q and O are distinct, then we must have
2|P Q| = |P Q| + |QP 0 | > |P O| + |OP 0 | = 2|P O|
as the shortest distance between two points is the straight line. Since
Q was taken to be any point on l, this shows that the distance from P
to O is indeed smaller that from P to any other point on l. Q.E.D. 
Definition 10. The distance from a point P to a line l is 0, if
P ∈ l, and is |P O| for the unique O ∈ l with P O ⊥ l, otherwise.
P

O l
I.e. it is the shortest distance from P to l along a straight line.
Lemma 12. Through any point P on a circle C there exists a unique
line tangent to C. It is the line perpendicular to OP , where O is the
center of the circle.

O
C

Proof. Let l be the unique line which passes through P and is perpen-
dicular to OP . Let P 0 be any point of l distinct from P .

P
P'
l

O
C
28 TIMOTHY LOGVINENKO

By Lemma 11 we have |OP 0 | > |OP | and therefore P 0 can’t lie on C

since |OP | is the radius of C. Therefore l intersects C only at P , i.e. l
is tangent to C at P .
On the other hand, let l0 be any line through P which is not perpen-
dicular to OP . Then, by Lemma 11 there exists P 0 ∈ l0 , distinct from
P , such that OP 0 ⊥ l0 . Let P 00 be the reflection of P in line OP 0 .

l'
P

P'
O
C
P''

Then line OP 0 is the perpendicular bisect of P P 00 . By Lemma 9 it

is then the locus of all the points equidistant from P and P 00 . In
particular, O is equidistant from P and P 00 , and so P 00 lies on C. But
by construction ∠P P 0 P 00 is a straight angle, i.e. P 00 ∈ l. So l0 meets C
in two points, P and P 00 , and is not therefore tangent to C.
Thus the unique line through P tangent to C is the line perpendicular
to OP . Q.E.D. 

Lemma 13. Let O, A and B be any three non-collinear points in the

plane. The bisector ∠AOB is the locus of all points within ∠AOB
equidistant from lines OA and OB.
B

Proof. Let P be any point within ∠AOB. Drop perpendiculars from

P onto OA and OB: let A0 ∈ OA and B 0 ∈ OB be the points such
that P A0 ⊥ OA and P B 0 ⊥ OB. Then distances from P to OA and
OB are |P A0 | and |P B 0 |, respectively.
(1) Suppose P is equidistant from OA and OB, i.e. |P A0 | = |P B 0 |.
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 29

B
B'

O P

A' A
Then 4P B 0 A0 is isosceles, and therefore by Lemma 6 we have
∠A0 B 0 P = ∠P A0 B 0 .
But then
π π
∠OB 0 A0 = − ∠A0 B 0 P = − ∠P A0 B 0 = ∠B 0 A0 O
2 2
and therefore 4OB A is isosceles with |OA0 | = |O0 B 0 |. We
0 0

conclude that 4OB 0 P ∼= 4OA0 P by SSS and therefore

∠A0 OP = ∠P OB 0
i.e. OP is the bisector of ∠AOB.
(2) Suppose OP is the bisector of ∠AOB, i.e.
∠A0 OP = ∠P OB 0 .
Then forget about point A0 for a moment, and let B 00 be reflec-
tion of B 0 in line OP .
B
B'

O P

A' A
B''

Since the reflection in OP leaves O and P fixed, it sends 4OP B 0

to 4OP B 00 . So these two triangles are congruent and we have
∠P OB 0 = ∠B 00 OP
π
∠P B 00 O = ∠OB 0 P =
2
|P B 00 | = |P B 0 |.
30 TIMOTHY LOGVINENKO

We then have
∠A0 OP = ∠P OB 0 = ∠B 00 OP
and therefore straight lines OA0 and OB 00 actually coincide. But
then points A0 and B 00 must also coincide, by uniqueness of the
perpendicular from P onto the line OA0 B 00 (Lemma 11). Hence
|P A0 | = |P B 00 | = |P B 0 |
i.e. P is equidistant from lines OA and OB.
We conclude that OP is the bisector of ∠AOB if and only if P is
equidistant from OA and OB. Q.E.D. 
Theorem 14. Let 4ABC be a triangle in the plane. The bisectors lA ,
lB and lC of angles ∠A, ∠B and ∠C in 4ABC are concurrent. The
point O in which they intersect is the center of the unique circle which
is contained in 4ABC and is tangent to all three of its sides.
A

lA
C
O lC

lB

B
Proof. Let O be the point where lA and lB intersect. By Lemma 13:
lA = { the locus of all points inside ∠A equidistant from AB and AC }
lB = { the locus of all points inside ∠B equidistant from AB and BC }
lC = { the locus of all points inside ∠C equidistant from BC and AC }
So lA ∩lB is the locus of all points inside ∠A∩∠B = 4ABC equidistant
from AB, BC and AC. But lA ∩ lB consists only of point O. So O
is the unique point within 4ABC equidistant from all three sides of
the triangle. In particular, O is equidistant from BC and AC. It
must therefore also lie on lC . So the three bisectors lA , lB and lC are
concurrent at O.
Let C be the circle whose center is O and whose radius is the distance
r from O to the three sides of the triangle. By Lemma 11 each side
of 4ABC has only one point whose distance from O is r – the unique
point where the perpendicular from O falls onto the side in question.
So C meets each of the sides of 4ABC at precisely one point, i.e. it is
tangent to each of them.
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 31

For the final claim, let there be a circle C 0 whose center O0 lies within
4ABC and which is tangent to all three sides of 4ABC. By Lemma
12 the radius from O0 to any of the three points where C 0 touches a side
of the triangle is perpendicular to the side in question. So, by definition
of distance from a point to the line, the distance from O0 to each of
the sides of 4ABC is the radius of C 0 . Hence O0 is equidistant from
all three sides of 4ABC, i.e. O0 is actually the point O constructed
above and C 0 is the circle C. 

Exercise 4 (Optional; hard!). In 3-dimensional space:

(1) What does it mean to say that a line through a point on a plane
is perpendicular to the plane in question? Is such line unique?
(2) How would you define a distance between a point and a plane?
(3) What does it mean to say that a sphere is tangent to a plane?
(4) Can you think of how to find a sphere tangent to 4 given planes?

7. Spherical Geometry
Nowhere8 in the results we’ve proved so far we’ve made use of Eu-
clid’s 5th axiom or of a number of statements equivalent to it, e.g.
“through every point not lying on a given line there passes exactly one
line parallel to the given one” or “the interior angles of every triangle
in the plane add up to π”.
This is because it is perfectly possible to have a consistent geometri-
cal theory where this is not true. It is worth, therefore, differentiating
those results whose proof doesn’t require the use of Euclid’s 5th from
those results whose proof does. Euclid himself must have been aware of
this, on some level, for in “Elements” he proves as many Propositions
as he can without using the 5th axiom before first invoking it in the
proof of his Proposition 29.
In this section we look at one of these non-Euclidean geometries -
the geometry of the surface of a sphere.
Problem: Evidently, there are no straight lines on the surface of
the sphere. Idea: In Euclidean geometry the straight line joining two
points A and B is the shortest path from one to the other. On the
surface of a sphere the shortest path between two points is a path
which goes along one of the geodesics.

Definition 11. A great circle or a geodesic on a sphere is the

intersection of this sphere with a plane passing through its centre.

8Almost nowhere.
32 TIMOTHY LOGVINENKO

O B

Clearly a great circle is determined uniquely by the plane which

carves it out on a sphere. On the other hand, in 3-dimensional Eu-
clidean space through any three non-collinear points there passes a
unique plane. So, conversely, any great circle determines uniquely the
plane through the center of a sphere which carves it out.
Definition 12. Two points on a sphere are called antipodal if they
lie on the opposite ends of a diameter of the sphere.

B
O
A

Exercise 5. Another way of defining a great circle would be to say,

that it is a circle which lies on the sphere and whose radius equals to
the radius of the sphere. To check this, show that any circle lying on
a sphere of radius r has radius ≤ r, and that the equality is achieved
only if the center of the circle coincides with the centre of the sphere.
Hint: A sphere of radius r is the locus of points whose distance from
its centre is r. If we have a circle lying on this sphere, take any diameter
of this circle and then use the triangle equality to get an upper bound
on its length. When is this upper bound achieved?
Lemma (S1). Through any two non-antipodal points on a sphere there
passes a unique great circle.
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 33

O B

Proof. Let O denote the centre of the sphere. If two points A and B
are not antipodal, then AB doesn’t contain O, i.e. A, B and O are
not collinear. Therefore there is a unique plane passing through O, A
and B, and this plane carves out on the sphere the unique great circle
which passes through A and B. 

Given any two points A and B on the sphere, we now have two
notions of distance:

O
A B

(1) Ambient notion: the distance from A to B is the length of a

straight line joining them, i.e. the length of the shortest path
from A to B in the 3-dimensional Euclidean space containing
the sphere (“Tunneler’s distance”).
(2) Intrinsic notion: the distance from A to B is the length of the
smaller (if they are non-equal) of the two arcs of any great circle
joining A and B, i.e. the length of the shortest path from A to
B along the surface of the sphere (“Sailor’s distance”).
We adopt the second one of these notions. There is not much of
a difference between the two, actually: one uniquely determines the
other via a simple trigonometric formula.

Definition 13. A map f : Sphere → Sphere is an isometry if it pre-

serves distances between points.
34 TIMOTHY LOGVINENKO

Note that if an isometry of the ambient 3-space fixes the centre of

the sphere, then it must take any point on the sphere to another point
on the sphere, i.e. it restricts to an isometry of the sphere.
Example 2. (1) Any rotation about an axis passing through the
center of the sphere yields an isometry of the sphere.
(2) Any reflection in a plane passing through the centre of the
sphere yields an isometry of the sphere.
(3) Translations have no fixed points, therefore no translation of
the ambient 3-space restricts to an isometry of the sphere.
Lemma (S2). If AB and CD are arcs of great circles which have the
same length, then there exists an isometry of the sphere taking AB to
CD.
Proof. Exercise. Hint: First find an isometry which takes a great circle
containing AB to a great circle containing CD. Then spin the latter
around. 
As noted before, there is no notion of parallel lines in spherical ge-
ometry - every pair of great circles meet. Another interesting point:
the sum of angles in a spherical triangle is never π.
Question: How do we measure an angle between two spherical lines?
Idea: Measure it in the tangent plane:

l2 l2
t2

O l1 O l1
t1

Fact: In 3-space, for any given line segment AB all the lines through
B which are perpendicular to AB form together a unique plane. This
plane is called the plane through B perpendicular to AB.
A

B
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 35

Lemma (S3). Let S be a sphere in a 3-dimensional space, let O be its

center and P be a point on the surface of S. Then the plane through
P perpendicular to OP is the unique plane tangent to S at P (i.e.
the plane which meets S only at P ).

Proof. Let πP be the plane through P perpendicular to OP . Suppose

πP meets S at some point P 0 6= P . Then let C 0 be the great circle
carved out on S by the plane OP P 0 :

plane OPP'

O P

C'
P'

By construction, line P P 0 meets C 0 in two points – P and P 0 . On the

other hand, line P P 0 lies in the plane πP perpendicular to OP and is
therefore itself perpendicular to OP . By Lemma 12 this makes P P 0
the unique line tangent to C 0 at P , i.e. P P 0 meets C 0 only at point P .
This is a contradiction, and so πP meets S only at P .
Conversely, let π 0 be any plane tangent to S at P . Let l be any line
through P in π 0 . Let Cl be the great circle carved out on S by the
plane containing l and O. By assumption, π 0 is tangent to S at P .
Therefore any line in π 0 through P is tangent to any great circle on S
through P . In particular, line l is tangent to Cl at P . By Lemma 12,
line l is then perpendicular to the radius of Cl at P , i.e. l ⊥ OP . Since
l was taken to be an arbitrary line through P in π 0 , we conclude that
all lines through P in π 0 are perpendicular to OP . So π 0 is the plane
through P perpendicular to OP . Q.E.D. 

Definition 14. Let AB and AC be two arcs of two great circles lB

and lC on a sphere S. Let πA be the plane tangent to S at A and let
tB and tC be the tangent lines carved out on πA by the planes of lB
and lC respectively. The spherical angle ∠BAC between lB and lC is
defined to be the planar angle ∠tB , tC between tB and tC in plane πA .
36 TIMOTHY LOGVINENKO

lC
tC
C
O lB
B
tB
A

NB: In 3-dimensional Euclidean geometry there is a well-defined

notion of an angle between a pair of intersecting planes. In terms of
that notion, the spherical angle between two great circles on a sphere
is simply the angle between the planes containing them.
Example 3. Cut the sphere with three planes through its center which
are perpendicular to each other. By above, the interior angles of the
resulting spherical triangle are each equal to π2 . The sum of the interior
angles in this triangle is therefore 3π
2
.

8. Plane geometry: the parallel postulate

We now return to plane geometry and proceed to prove the results
which require the use of:
Euclid’s 5th Axiom (The Parallel Postulate): If a straight
line falling on two straight lines makes the interior angles on one of
the sides which are less than two right angles in total, then the two
straight lines, if produced indefinitely, meet on that side.
α+β<π
α
β
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 37

This is, in fact, equivalent to:

Playfair’s Axiom: Given a line l and a point P not on l there is
a unique line through P parallel to l.
Proof. Euclid’s 5th Axiom ⇒ Playfair’s Axiom:
Let l1 and l2 be a pair of lines and l3 be a third line falling upon
them. Denote the internal angles l3 makes with l1 and l2 as follows:

α' α
l1
β' β l2
l3
Since
(α0 + α) + (β 0 + β) = π + π = 2π
there are three possible cases
• α + β > π and α0 + β 0 < π
• α + β < π and α0 + β 0 > π
• α + β = π and α0 + β 0 = π
We therefore see that Euclid’s 5th axiom implies that l1 and l2 are
parallel if and only if α + β = α0 + β 0 = π. In other words, two lines
are parallel if and only if a line falling on them makes on either side of
itself the internal angles which sum up to π.
Let now l be any line and P a point not on it. Drop a perpendicular
P A from P onto l, i.e. let A be the unique point of l such that P A ⊥ l.
Let now l1 be any line through P .

P
α' α l1

l
A
By above l1 is parallel to P if and only if α0 = α = π2 , i.e. if l1 ⊥ P A.
Since there is only one line through P perpendicular to P A, we conclude
that there is only one line through P parallel to l. Q.E.D.
Playfair’s Axiom ⇒ Euclid’s 5th Axiom:
Exercise (hard!). 
Lemma 15. Let l1 and l2 be a pair of parallel lines. Let l3 be a line
falling on l1 and l2 . Let the internal angles l3 makes with l1 and l2 be
as marked on the diagram
38 TIMOTHY LOGVINENKO

α' α
l1
β' β l2
l3
then
α0 = β and α = β 0 .
Proof. Since l1 is parallel to l2 we must have by Euclid’s 5th axiom
α + β = α0 + β 0 = π.
But since α makes a straight angle with α0 , we have also
α + α0 = π
and therefore α0 = β and α = β 0 . Q.E.D. 
Lemma 16. Let 4ABC be a triangle in the plane. Then its interior
angles sum up to π, i.e.
∠A + ∠B + ∠C = π.
Proof. Let l be the unique line through C parallel to AB. Let α0 be the
angle l makes with AC that is adjacent to ∠C in 4ABC. Similarly,
let β 0 be the angle l makes with BC that is adjacent to ∠C.
C
α' β'
γ

α β
A B
We have α0 + γ + β 0 = π as the corresponding angles make up a straight
angle. On the other hand, by Lemma 15 we have α = α0 and β = β 0 .
We conclude that α + β + γ = π. Q.E.D. 
There is an alternative way to prove Lemma 15, which is more gen-
eral and gives a formula for the sum of interior angles in any plane
n-gon:
Theorem 17. The sum of interior angles in a plane n-gon adds up to
(n − 2)π.
Proof. Label the vertices of the n-gon by A1 , . . . , An in such a way that
starting at A1 and travelling anti-clockwise along the circumference one
encounters first A2 , then A3 , and so on.
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 39

A2 A1

A3

A4 A5

Now imagine placing a pencil along A1 A2 so that its tip is pointing

towards A2 . Now move pencil along the circumference of the n-gon,
turning it once it reaches vertex A2 in such a way that it would then
point along the next edge - A2 A3 . Repeat this until the pencil would
arrive back to its starting point. Let us now sum up all the angles
the pencil was turned through on its way, counting each angle with a
plus sign if the pencil was turned anti-clockwise and with a minus sign
otherwise. Since the pencil went once around the interior of the n-gon
anti-clockwise and is now back where it started and facing the same
way, the sum total is 2π.
On the other hand, at each vertex Ak we’ve encountered one of the
two possible situations:
• Case 1: ∠Ak < π

αk Ak

to Ak+1 Ak
interior
of the n from Ak-1
-gon
In this case, the pencil was rotated anti-clockwise through an
angle of αk and we have:

∠Ak + αk = π.

• Case 2: ∠Ak > π

from Ak-1 interi
o r of t h

Ak
Ak
to Ak+1 αk
n

e n- go
In this case, the pencil was rotated clock-wise through an angle
of αk and we have:

∠Ak − αk = π.
40 TIMOTHY LOGVINENKO

We see therefore that, taking account of the sign, the contribution

to the total sum at each vertex Ak is π − ∠Ak . Hence
Xn Xn
2π = (π − ∠Ak ) = nπ − ∠Ak
k=1 k=1

and so finally
n
X
∠Ak = (n − 2)π.
k=1
Q.E.D. 
Exercise 6. Where and how does the proof of Theorem 17 make use of
the parallel postulate? Hint: what does this “moving pencil” argument
mean in the rigorous language we’ve been trying to develop?
Lemma 18. Let ABCD be a parallelogram (a quadrilateral whose sides
are pairwise parallel) with AB k CD and AC k BD.
C D

A B
Then
|AB| = |CD|, |AC| = |BD|, ∠A = ∠D and ∠B = ∠C.
Proof. By Lemma 15 we have
∠BAD = ∠CDA and ∠ADB = ∠DAC.
It follows that ∠A = ∠D in ABCD. It also follows that 4ABD ∼
=
4DCA by ASA. And therefore
|AB| = |CD|, |AC| = |BD| and ∠B = ∠C in ABCD.
Q.E.D. 
Exercise 7. Show that the point where the diagonals of a parallelo-
gram intersect is the midpoint of each of the diagonals.

9. Area
Definition 15. The area of any geometrical figure is uniquely defined
by the following:
(1) The area of a rectangle is the product of the lengths of its two
adjacent sides.
(2) If two figures are disjoint or meet only along their edges, then
the area of their union is the sum of their areas.
(3) Congruent figures have equal areas.
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 41

NB: Essentially, items (1) and (3) could be replaced by saying that
any square of side 1 has area 1, i.e. it’s all about a choice of a unit.
The main defining property of area is (2).
Lemma 19. The area of a parallelogram is its base times its height,
where the “base” is the length of any side of the parallelogram, and the
“height” is the distance from either of the remaining two vertices to the
line through the chosen side.
Proof. Drop perpendiculars CC 0 and DD0 from C and D onto line AB.
In other words, let C 0 and D0 be points of AB such that CC 0 ⊥ AB
and DD0 ⊥ AB.
C D

A C' B D'
“base” = |AB|, “height” = |CC 0 | = |DD0 |
It follows from Lemma 15 that ∠C 0 AC = ∠D0 BD. By construction
∠CC 0 A = ∠DD0 B = π2 and since the sum of interior angles in a triangle
is π (Lemma 16) we must also have ∠ACC 0 = ∠BDD0 . Finally, as the
opposite sides in a parallelogram are of equal length (Lemma 18) we
have |AC| = |BD| and therefore 4ACC 0 ∼ = 4BDD0 by ASA. Hence
area(ABCD) = area(C 0 D0 CD) = |C 0 D0 | × |CC 0 | = |AB| × |CC 0 |.
Q.E.D. 
Lemma 20. The area of a triangle is a half of its base times its height,
where the “base” is the length of any side of the triangle, and the
“height” is the distance from the remaining vertex to the line through
the chosen side.
Proof. Choose side AB as the “base”, then the “height” is the distance
from C to AB. Draw a line through B parallel to AC, a line through
C parallel to AB, and let A0 be the point where they intersect.
C A'

A B
By same argument as in Lemma 18 we have 4ABC ∼
= 4A0 CB. Hence
area(ABCA0 ) = area(ABC) + area(A0 CB) = 2 × area(ABC).
42 TIMOTHY LOGVINENKO

On the other hand, by Lemma 19 the area of parallelogram ABCA0 is

its base times its height. Choosing side AB as the base, we have
area(ABCA0 ) = |AB| × dist(C, AB).
It follows that the area of 4ABC is 12 |AB| × dist(C, AB). Q.E.D. 
Theorem 21 (Ceva’s Theorem). Let 4ABC be a triangle in the plane,
and let A0 , B 0 and C 0 be points on BC, AC and AB respectively. If
AA0 , BB 0 and CC 0 are concurrent at some point P in the interior of
the triangle
C

B' A'
P

A C' B
then we have
|AC 0 | |BA0 | |CB 0 |
× × = 1.
|C 0 B| |A0 C| |B 0 A|
Proof. The area of a triangle is a half of its base times its height
(Lemma 20) and so we have:
1
area(4P AC 0 ) = |AC 0 | × dist(P, AC 0 )
2
1
area(4P BC 0 ) = |C 0 B| × dist(P, C 0 B)
2
Since A, C 0 and B are collinear, we also have
dist(P, AC 0 ) = dist(P, C 0 B)
and therefore
|AC 0 | area(4P AC 0 )
= .
|C 0 B| area(4P BC 0 )
Similarly we obtain
area(4CAC 0 ) |AC 0 |
= .
area(4CBC 0 ) |C 0 B|
It is also evident from the diagram that
area(4CAP ) = area(4CAC 0 ) − area(4P AC 0 )
area(4CBP ) = area(4CBC 0 ) − area(4P BC 0 ).
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 43

Therefore
area(4CAP ) area(4CAC 0 ) − area(4P AC 0 )
= =
area(4CBP ) area(4CBC 0 ) − area(4P BC 0 )
|AC 0 | |AC 0 |
|C 0 B|
area(4CBC 0 ) − |C 0 B|
area(4P BC 0 ) |AC 0 |
= =
area(4CBC 0 ) − area(4P BC 0 ) |C 0 B|
Similarly we obtain that
|BA0 | area(4ABP ) |CB 0 | area(4CBP )
0
= and 0
=
|A C| area(4CAP ) |B A| area(4ABP )
and so finally we have
|AC 0 | |BA0 | |CB 0 | area(4CAP ) area(4ABP ) area(4CBP )
0
× 0 × 0 = × × = 1.
|C B| |A C| |B A| area(4CBP ) area(4CAP ) area(4ABP )
Q.E.D. 
Exercise 8. Prove the converse to Ceva’s Theorem. That is, prove
that if 4ABC is a triangle in the plane, and if A0 , B 0 and C 0 are points
on BC, AC and AB respectively, then
|AC 0 | |BA0 | |CB 0 |
× × =1
|C 0 B| |A0 C| |B 0 A|
implies that AA0 , BB 0 and CC 0 are concurrent.
Hint: Let P be the intersection of AA0 and BB 0 . Let C 00 be the
point where the line through CP intersects AB. What can we say
|AC 00 |
about |C 00 B| ?

Exercise 9. The medians of a triangle are the lines joining each vertex
to the middle of the opposite side. Show that the three medians of a
triangle are concurrent.

10. Triangles on a sphere

We assume without proof that the surface area of a sphere of radius
r is 4πr2 .
Definition 16. An area contained in a sector between two great circles
is called a lune.

O
44 TIMOTHY LOGVINENKO

It is clear that the area of the lune is proportional to the corre-

sponding angle between the great circles. Since the full sphere, which
corresponds to an angle of 2π, has the surface area of 4πr2 , we conclude
θ
that a lune of an angle θ has area 2π 4πr2 = 2θr2 .
Theorem 22. Let 4ABC be a triangle on a sphere of radius r. Then
area(ABC)
∠A + ∠B + ∠C = π + .
r2
Proof. Any spherical triangle has an antipodal twin - the triangle carved
out by the same great circles, but whose vertices are antipodal to the
vertices of the original triangle. Clearly, the map which sends every
point of a sphere to its antipode is an isometry. So any triangle is
congruent to its antipodal twin. On the diagram below we’ve marked
the antipodal twin of 4ABC - triangle 4A0 B 0 C 0 .

C
B' A'
O
A B
C'

Altogether, the great circles AB, BC and AC divide the sphere into 8
triangles which make up 4 pairs of antipodal twins. We write:
T = 4ABC and T 0 = A0 B 0 C 0 , TA = 4A0 BC and TA0 = AB 0 C 0
TB = 4AB 0 C and TB0 = A0 BC 0 , TC = 4ABC 0 and TC0 = A0 B 0 C
and we write t for the area of T , tA for the area of TA , t0A for the area
of TA0 , et cetera. Let us now “unwrap” the picture above onto a flat
plane by cutting out 4A0 B 0 C 0 and making it the triangle “at infinity”:
A'

' TA
TC
C TB'

B
T
TB TC
B' A C'
'
TA
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 45

We can see that T and TA make up a lune of angle ∠A. The area of a
lune of angle ∠A is 2r2 ∠A and therefore
t + tA = ∠A · 2r2
and similarly
t + tB = ∠B · 2r2 and t + tC = ∠C · 2r2 .
We conclude that
3t + tA + tB + tC = (∠A + ∠B + ∠C)2r2 .
On the other hand
4πr2 = surface area of the sphere = t + tA + tB + tC + t0 + t0A + t0B + t0C .
As antipodal twins are congruent, we have t = t0 , tA = t0A , etc. Hence
t + tA + tB + tC + t0 + t0A + t0B + t0C = 2(t + tA + tB + tC )
and so finally
t + tA + tB + tC = 2πr2 .
We conclude that
2t = (3t + tA + tB + tC ) − (t + tA + tB + tC ) =
= (∠A + ∠B + ∠C − π)2r2
and therefore
t
∠A + ∠B + ∠C = π + .
r2
Q.E.D. 
Corollary 23. In a non-degenerate spherical triangle (a triangle with
non-zero area) the sum of interior angles is strictly greater than π.

11. Similarity
Definition 17. Let 4ABC and 4A0 B 0 C 0 be a pair of triangles in
the plane. We say that 4ABC is similar to 4A0 B 0 C 0 , and write
4ABC ∼ 4A0 B 0 C 0 , if
∠A = ∠A0 , ∠B = ∠B 0 and ∠C = ∠C 0 .
B'

A C'

C
B
A'
46 TIMOTHY LOGVINENKO

Lemma 24. Let 4ABC and 4A0 B 0 C 0 be a pair of triangles in the

plane. If 4ABC ∼ 4A0 B 0 C 0 then
|AB| |BC| |AC|
= = .
|A0 B 0 | |B 0 C 0 | |A0 C 0 |
Proof. Let us prove that
|AB| |BC|
0 0
= 0 0 ,
|A B | |B C |
the other two equalities are proved analogously. Since ∠B = ∠B 0 there
exists an isometry f which takes B to B 0 , takes C to some point C 00 on
the line B 0 C 0 on the same side of B 0 as C 0 and takes A to some point
A00 on the line B 0 A0 on the same side of B 0 as A0 .
B' = f(B)
C'' = f(C)
C'

A''= f(A)

A'
By assumption ∠BCA = ∠BC 0 A0 and therefore line C 00 A00 is parallel
to the line C 0 A0 . Therefore dist(C 0 , C 00 A00 ) = dist(A0 , C 00 A00 ) and, as
the area of a triangle is its base time its height, (Lemma 20)
1
area(C 0 C 00 A00 ) = |C 00 A00 | × dist(C 0 , C 00 A00 ) =
2
1 00 00
= |C A | × dist(A0 , C 00 A00 ) = area(A0 C 00 A00 ).
2
Using Lemma 20 again we obtain
area(C 0 A00 B 0 ) 1
2
|A00 B 0 | × dist(C 0 , A0 B 0 ) |A00 B 0 | |AB|
= 1 = =
area(A0 B 0 C 0 ) 2
|A0 B 0 | × dist(C 0 , A0 B 0 ) |A0 B 0 | |A0 B 0 |

area(A0 B 0 C 00 ) 1
2
|B 0 C 00 | × dist(A0 , B 0 C 0 ) |B 0 C 00 | |BC|
= 1 = = .
area(A0 B 0 C 0 ) 2
|B 0 C 0 | × dist(A0 , B 0 C 0 ) |B 0C 0| |B 0C 0|

We therefore conclude that

|AB| area(C 0 A00 B 0 ) area(C 0 A00 C 00 ) + area(A00 C 00 B 0 )
= = =
|A0 B 0 | area(A0 B 0 C 0 ) area(A0 B 0 C 0 )
area(A0 A00 C 00 ) + area(A00 C 00 B 0 ) area(A0 C 00 B 0 ) |BC|
= 0 0 0
= 0 0 0
= 0 0 .
area(A B C ) area(A B C ) |B C |
Q.E.D. 
 MA1004: GEOMETRY (YEAR 1) LECTURE NOTES 47

Exercise 10. Prove the converse to Lemma 24. That is, prove that if
4ABC and 4A0 B 0 C 0 are a pair of triangles such that
|AB| |BC| |AC|
= =
|A0 B 0 | |B 0 C 0 | |A0 C 0 |
then
4ABC ∼ 4A0 B 0 C 0 .
Theorem 25 (Pythagoras Theorem). Let 4ABC be a triangle in the
plane such that ∠B is a right angle,
C

A B
then
|AB|2 + |BC|2 = |AC|2 .
Proof. Drop a perpendicular BD from B onto AC.
C
D

A B
Then since the sum of interior angles of a triangle is π we have
π
∠BAD + ∠DCB =
2
π
∠BAD + ∠DBA =
2
π
∠CBD + ∠DCB =
2
and so
∠BAD = ∠CBD and ∠DCB = ∠DAB.
By definition of similarity of triangles we have
4ABD ∼ 4ABC ∼ 4BCD.
Therefore by Lemma 24 we have
|AD| |AB|
=
|AB| |AC|
and consequently
|AB|2 = |AD||AC|.
48 TIMOTHY LOGVINENKO

Similarly,
|BC|2 = |DC||AC|
and we conclude that
|AB|2 + |BC|2 = (|AD| + |DC|)|AC| = |AC|2 .
Q.E.D. 
Email address: LogvinenkoT@cardiff.ac.uk

School of Mathematics, Cardiff University, Cardiff, CF24 4AG, UK