School of Mathematics and Statistics

Newcastle University

MMath Mathematics Report

Spherical Trigonometry
Through the Ages

Author: Supervisor:
Lauren Roberts Dr. Andrew Fletcher

May 1, 2015
Contents

1 Introduction 1

2 Geometry on the Sphere

Part I 4
2.1 Cross-sections and Measurements . . . . . . . . . . . . . . . . 4
2.2 Triangles on the Sphere . . . . . . . . . . . . . . . . . . . . . . 7
2.2.1 What is the Largest Possible Spherical Triangle? . . . . 7
2.2.2 The Sum of Spherical Angles . . . . . . . . . . . . . . 9

3 Menelaus of Alexandria 12
3.1 Menelaus’ Theorem . . . . . . . . . . . . . . . . . . . . . . . . 13
3.2 Astronomy and the Menelaus Configuration: The Celestial
Sphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

4 Medieval Islamic Mathematics 18

4.1 The Rule of Four Quantities . . . . . . . . . . . . . . . . . . . 18
4.2 Finding the Qibla . . . . . . . . . . . . . . . . . . . . . . . . . 20
4.3 The Spherical Law of Sines . . . . . . . . . . . . . . . . . . . . 22

5 Right-Angled Triangles 25
5.1 Pythagoras’ Theorem on the Sphere . . . . . . . . . . . . . . . 25
5.2 The Spherical Law of Cosines . . . . . . . . . . . . . . . . . . 27

6 Geometry on the Sphere

Part II 30
6.1 Areas of Spherical Polygons . . . . . . . . . . . . . . . . . . . 30
6.2 Euler’s Polyhedral Formula . . . . . . . . . . . . . . . . . . . . 33

7 Quaternions 35
7.1 Some Useful Properties . . . . . . . . . . . . . . . . . . . . . . 35
7.1.1 The Quaternion Product . . . . . . . . . . . . . . . . . 36
7.1.2 The Complex Conjugate and Inverse Quaternions . . . 36

1
CONTENTS 2

7.2 Rotation Sequences . . . . . . . . . . . . . . . . . . . . . . . . 37

7.2.1 3-Dimensional Representation . . . . . . . . . . . . . . 39
7.3 Deriving the Spherical Law of Sines . . . . . . . . . . . . . . . 39

8 Conclusions 42

Bibliography 44
 Abstract

The art of applying trigonometry to triangles constructed on the surface of

the sphere is the branch of mathematics known as spherical trigonometry. It
allows new theorems relating the sides and angles of spherical triangles to be
derived.
The history and development of spherical trigonometry throughout the
ages is discussed. The geometry on the surface of the sphere is introduced
to aid the understanding of fundamental proofs of spherical results. Ques-
tions such as “how big can we make a spherical triangle?” are answered
so the strange geometry the sphere yields can be visualised. How spherical
trigonometry was simplified from ancient to medieval times is investigated.
Crucial applications to astronomy and cartography are identified, such as
determining the coordinates of the Sun, finding the qibla, and calculating
distances on the surface of the Earth. The use of hyper-complex num-
bers, known as quaternions, and their properties are introduced to determine
spherical results, as well as briefly looking at the application of quaternions
to 3-dimensional rotations.
Chapter 1

Introduction

The face of mathematics has been continuously growing and changing since
the dawn of humanity. Things that we think of as simple, such as the con-
cept of a number, have in fact been under development since the Old Stone
Age. Cave men did not have the same advanced set of symbols and words to
represent numbers as we do, nevertheless they still needed to quantify how
much food to gather. As we have become more knowledgeable of the math-
ematics around us, we have accepted what those before us have discovered,
and developed it further. It is therefore not surprising that as time has gone
by, the topics we are interested in and what is taught in schools have changed
dramatically.
It is widely believed that for many centuries people around the world
thought that the Earth was flat. The belief of this myth is in itself a myth.
As far back as the famous Greek philosopher Pythagoras in the 6th cen-
tury BC, and Aristotle c. 340 BC, (Heath, 1932) scholars had evidence and
observations to confirm that the world is spherical. Although there were

Figure 1.1: Image from Tøndering.

Known as “The Woodcut”, this engrav-
ing first appeared in Flammarion (1888)
and depicts someone from the middle
ages peering out to the Heavens at the
edge of a flat Earth.

1
CHAPTER 1. INTRODUCTION 2

still people who thought the Earth was flat, the majority accepted this was
not the case. It was therefore apparent from an early point in time, any
observations across large enough distances that the curvature of the Earth
would need to be taken into account would need a good understanding of
the sphere. Since much of the world was still being discovered, maps were
continuously being updated, so accuracy was crucial for setting sail on the
open seas. Not only maps of the Earth, but maps of the skies above were
needed for navigation.
Spherical trigonometry mainly came about as a tool to help ancient as-
tronomers. Astronomy played a major part in everyday life, from time keep-
ing to navigation, long before there were theories and equations to describe
it. Since the earliest civilizations, when farming and trade were expanding,
connections were made between the phases of the moon and vegetation. This
caused people to monitor the moon by creating lunar calendars, but also in-
creased interest in the bodies around us. By simply looking up at the sky,
day or night, we see that the celestial bodies, such as the moon and stars
or the sun, move in circular arcs across it. Initially, after realizing that the
rising sun was the same each day, and that certain constellations appeared
on a nightly basis, it was thought that the celestial sphere rotated around
the Earth. Although we know this now not to be true, it was still clear
that a method of calculating the positions of each object, as well as their
trajectories, would involve extensive knowledge of the sphere.
The first major breakthrough for spherical trigonometry came in the first
century AD, by Greek mathematician and astronomer Menelaus. Properties
of the sphere had been known for thousands of years prior to this, but his
groundbreaking results lead to great exploration of the Heavens and paved the
way for future mathematicians to unearth further results. Astronomers relied
on Menelaus’ results for almost 1000 years. Medieval Middle Eastern scholars
were able to simplify these results, and what’s more discovered important
formulas, including the spherical Law of Sines. These findings were not
just vital to the development of spherical mathematics, religious followers
depended on their applications too.
Considering we are looking into the history of spherical trigonometry, the
question of why and when the trigonometric functions themselves came about
is one expected to arise. The non-existence of calculators meant trigono-
metric tables needed to be produced, and this was done to various degrees
of accuracy. Such tables were originally constructed to give the length of
chords.
Hipparchus of Nicaea, who lived around 150 BC, invented the chord func-
tion whilst trying to model the Sun’s orbit around the Earth. Following an-
cient belief that the universe rotated around the Earth, he had placed the
CHAPTER 1. INTRODUCTION 3

Earth at the centre of the orbit, however came to realize that this meant the
number of days between the spring and autumn equinoxes, and vice versa,
should be equal, but he knew this not to be true. Knowing how far to move
the Earth from the centre of the orbit in order to reproduce the observed
inter-equinoctial times required a new function, and so trigonometry was
born. The sine would later replace the chord in India around the year 500,
and soon after the rest of the world (Pingree, 1976).
The calculation of the table of sines was the main requirement in calcu-
lating a trigonometric table, since all other functions can be found from this.
This is an arduous task when done by hand, although it is one that can be
reduced by acknowledging the periodicity of the sine function and that it is
an even function between 0 and 180 ◦ .
The years leading up to the 20th century saw an explosion in mathemat-
ical developments. The invention of the logarithm as well as the derivation
of the spherical law of cosines left users of spherical trigonometry with a
complete set of tools. The subject remained popular in schools and univer-
sities until the beginning of the 1900s. World War I and II arose, where the
applications of spherical trigonometry to navigation for the military ensured
that it remained a very relevant and important topic up until the 1950s.
Since then it has rarely been explored, being replaced by its not so distant
younger sister: trigonometry in the plane. As we will find in the forthcom-
ing chapters, many of the planar results we know today are actually special
cases of their spherical counterparts. The advances of computers and soft-
ware leave little reason for people to apply spherical trigonometry to solve
real life problems, like finding the rising times of the sun, for example: the
answers are just a click away. It is therefore surprising, but also insightful,
to notice that computer graphics, animation and GPS all use results from
spherical trigonometry in their programs today. As we will see, the spherical
world yields some interesting and amazing results.
Note on historical sources: Statements regarding the history of math-
ematics and spherical trigonometry are not individually referenced. The
following sources were use for this material:

• Heavenly Mathematics: The Forgotten Art of Spherical Trigonometry,

Van Brummelen (2013);

• A Concise History of Mathematics, Struik (1967);

• The MacTutor website, O’Connor and Robertson.

Chapter 2

Geometry on the Sphere

Part I

The method of writing numbers as we do today, where place is an indication

of a number’s value, was brought into fruition c. 2100 BC in Mesopotamia.
This dates back to the Sumerian period and replaced a system that mirrors
Roman numerals, where a string of symbols must be added to give the value.
Interestingly, a sexagesimal system was imposed, i.e. each position repre-
sents 1, 60, 3600, and 60−1 , 60−2 , which we still use in part today for time.
Furthermore, we use this system for measuring angles, with a whole circle
representing 360 ◦ , each degree 60 minutes, and each minute 60 seconds. This
is transformed into coordinates on the surface of the Earth. The reason for
the use of base 60 may be unclear, but there is no doubt that its relation-
ship with the sphere and spherical applications is something of beauty. The
proofs in this chapter follow the methodology of Van Brummelen (2013).

2.1 Cross-sections and Measurements

Before we begin working on the surface of the sphere, we need to know the
properties lines and shapes have when drawn upon it. This raises some
important questions: How can we draw a straight line on the sphere? What
is the simplest shape that can be drawn onto the surface of the sphere? Both
of these questions can be answered with an example. It is recommended to
have a spherical object to hand in the forthcoming chapters to help visualise
the spherical geometry. A useful object to have is a Lénart sphere, in use
in Figure 2.1a. It comes with a spherical compass and spherical protractor,
both pictured, to draw triangles and shapes on the sphere.
If we take a plane and intersect the sphere at any point, we see that the

4
CHAPTER 2. GEOMETRY ON THE SPHERE PART I 5

(b) Proof of Theorem 1.

(a) A Lénart sphere.

Figure 2.1: Cross-sections of the sphere.

cross-section is always a circle, Figure 2.1a. This can easily be proved with
an application of Pythagoras’ Theorem using Figure 2.1b.
Proof. Take any cross-section of the sphere, centre O, by a plane. Drop a
perpendicular from a point A on the plane to O and extend lines from both
points to a point B on the edge of the cross-section. This implies that the
angle OAB is a right-angle, so applying Pythagoras’ Theorem we can write

AO2 + AB 2 = OB 2 .

We also know that as its endpoints are fixed, AO is constant. Since OB

is the radius of the sphere, this too is constant, which implies that AB is
constant. As B is an arbitrary point on the cross-section, this is true for any
point on the cross-section. We have therefore defined any cross-section to be
a circle.

Theorem 1. Every cross-section of a sphere by a plane is a circle.

This is also the simplest shape we can create on the sphere, as it consists
of just one side. Depending on where the plane intersects the sphere, we get
circles of different radii. Figure 2.1a shows circles with a radius less than
that of our sphere as small circles (black), but the most important circle for
us is a great circle (red).
Great circles are cross-sections of the sphere whose centres coincide with
the centre of the sphere. They have the greatest circumference and therefore
will take the role of straight lines upon the sphere. Clearly, each sphere has
an infinite number of great circles and small circles that can be drawn upon
it. Given that great circles are straight lines on the sphere, great circle arcs
CHAPTER 2. GEOMETRY ON THE SPHERE PART I 6

Figure 2.2: A great circle arc of

length θ.

are straight line segments. We can therefore say that since the shortest dis-
tance between two points in the plane is a straight line, the shortest distance
between two points on the sphere is a great circle arc.
Theorem 2. Great circles represent straight lines on the surface of the
sphere. Great circle arcs are line segments.
Now that we have the tools to construct shapes on the surface of the
sphere, we need to make a note about their measurements. Not only the
angles of a shape, which in our case will be triangles, but the lengths of the
sides are measured in degrees.
Take a sphere of radius 1, Figure 2.2 (for simplicity, all spheres we consider
will have unit radius). If we have an angle θ at the centre, O, we can project
this onto the surface of the sphere to give an arc length of θ. In a circle,
arc length = rθ. A great circle arc is part of the cross-section of the sphere
that is a great circle, so in Figure 2.2 the arc length is θ. This observation
allows us to define a rule for converting arc lengths into angles.
We start with a great circle and its pole, P , Figure 2.3a. A pole is the
perpendicular axis of a great circle, so if we drop two line segments from P
onto the great circle, they will each be of length 90 ◦ and meet the great circle
at right-angles. A bird’s-eye view in Figure 2.3b, with P at the centre, gives
a similar picture to Figure 2.2. If the angle between the two line segments
at P is θ, the distance between these two lines on the corresponding great
circle to P is θ. So we have our result.
Theorem 3. An angle between two lines on the surface of the sphere can be
converted into a length by moving 90 ◦ away along both lines and joining the
endpoints.
A practical example of the properties of the sphere we have seen so far is
the Earth. The equator and lines of longitude are great circles, and the lines
of latitude are small circles. The north and south poles are the two poles
to the equator, they are at a latitude of ±90 ◦ , respectively. Therefore, the
CHAPTER 2. GEOMETRY ON THE SPHERE PART I 7

(a) Line segments from the pole,

P, to the corresponding great cir- (b) A bird’s-eye view of the pole.
cle.

Figure 2.3: Converting between angles and lengths.

distance between two lines of longitude on the equator is equal to the angle
between them at the north (or south) pole.

2.2 Triangles on the Sphere

When first imagining a triangle, one might think of a shape with three
straight sides, be it equilateral, isosceles or scalene, with the sum of the
internal angles equal to 180 ◦ . When we move onto the surface of the sphere,
where the sides of a spherical triangle are great circle arcs, things get in-
teresting. Can you imagine a triangle with an internal angle greater than
180 ◦ ? Or a side as long as the circumference of a great circle? We need some
constraints to our spherical triangles; let us start with how big we can build
them.

2.2.1 What is the Largest Possible Spherical Triangle?

Since the sides of a triangle can be as small as we like, we need only consider
how large the perimeter of a spherical triangle can be.
Picture a triangle on the surface of the sphere, Figure 2.4, left. Gradually
increase the internal angles at each vertex, which will in turn increase the side
lengths so they remain great circle arcs; the triangle will begin to straighten
out until we have a great circle, Figure 2.4, right. Increasing the angles
beyond this point would begin to make the perimeter of the triangle smaller
again. We have therefore reached an upper limit for the perimeter of a
spherical triangle: 360 ◦ . We will now look at the reasons behind this limit
CHAPTER 2. GEOMETRY ON THE SPHERE PART I 8

Figure 2.4: Increasing the edges of a spherical triangle.

(b) A spherical triangle with ver-

(a) A spherical triangle with one
tices connected to the centre of
side > 180 ◦ and its compliment
the sphere.
(red, dashed).

Figure 2.5: Spherical triangles.

in more detail.
On the surface of a sphere, the longest straight line we can draw is a
great circle: length 360 ◦ . Intuitively, it is impossible to have three sides this
length, so we have to go slightly shorter. In Figure 2.5a, we see a triangle with
two short sides and one side greater than 180 ◦ . This is where we establish
our first constraint.
Where the side of a spherical triangle is greater than 180 ◦ , we replace it
with its supplementary side.
As this can be done with every possible triangle, it is unnecessary to
allow triangles with individual sides longer than 180 ◦ . This also keeps the
geometry on the surface of the sphere relatively simple. Our train of thought
then tells us that if each side is less than 180 ◦ , the upper bound of the
perimeter is 3 × 180 ◦ = 540 ◦ . Again this is not possible, if two sides are of
length 180 ◦ , the third would have to be zero, or very small.
We construct a triangle in Figure 2.5b where the angles at the centre
of the sphere, O, are equal to the side lengths. If we collapse the side OA
onto the plane below, the sum of the angles at O is AOB + AOC = BOC.
CHAPTER 2. GEOMETRY ON THE SPHERE PART I 9

Bringing the side OA back up to its original position we are increasing the
angles AOC and AOB, so we have AOB + AOC > BOC. This gives us our
second constraint.
The sum of any two sides in a spherical triangle must be greater than the
third.
This is equivalent to proposition 20 in Euclid’s Elements in planar maths,
that the third side of a planar triangle cannot exceed the sum of the other
two. The proof of this is an example of one that works for spherical geometry
as well as geometry in the plane.
Proof. We continue with the tetrahedron formed by joining the vertices of
the spherical triangle in Figure 2.5b to the centre of the sphere. Since the
three sides connected to the centre, O, are triangles, their angles must sum
to 3 × 180 ◦ = 540 ◦ . It can be shown, using the same argument for the
angles at one of the vertices, say A, as for the justification that two sides of
a spherical triangle must be greater than the third, that the sum of the two
angles OAB and OAC is greater than the angle BAC (similarly for B and
C). Therefore,

Perimeter of ABC = angles at O

= 540 ◦ − (angles at A + angles at B + angles at C)
< 540 ◦ − (A + B + C)
= 540 ◦ − 180 ◦
= 360 ◦ .

2.2.2 The Sum of Spherical Angles

Can an equilateral triangle consist of three right-angles? On the plane this is
impossible, but the curved surface of the sphere allows for strange geometry.
We turn our attention to the smallest and largest possible sums of angles in
spherical triangles.
As the length of a great circle arc tends to zero, the arc tends to a straight
line. Therefore the smallest possible triangle we can construct on the sphere
will resemble a planar triangle, whose sum of internal angles is 180 ◦ . This is
the lower bound of the sum of angles in a spherical triangle, as increasing the
side lengths will increase the size of the angles. We found that the largest
perimeter of a spherical triangle is 360 ◦ , whose sum of angles is 3 × 180 ◦ =
540 ◦ . This can be taken as the maximum sum of angles by imposing the
same constraint as used for the perimeter: where an angle is greater than
180 ◦ , we replace it with its complimentary angle. This argument suffices
CHAPTER 2. GEOMETRY ON THE SPHERE PART I 10

(a) Constructing the polar tri-

angle (red) of triangle ABC (b) The Polar Duality theorem.
(black).

Figure 2.6: The polar triangle.

for the upper bound, but we will look closer at the lower bound in order to
introduce the polar triangle.
To do this we go back to the 10th century in medieval Persia. At this
time, there was a profound development in mathematics in the Muslim world,
with many theorems and ideas being re-invented. It is here that Abū Nas.r
Mans.ūr ibn ‘Alı̄ ibn ‘Irāq constructed the first polar triangle.
Take an arbitrary triangle, Figure 2.6a. Mark the pole of each side by
moving 90 ◦ perpendicularly away from it (blue dashed lines), choosing the
pole on the same side as the original triangle. The point C 0 is the pole of
AB and we define A0 and B 0 in the same manner. Joining A0 , B 0 and C 0 (red
lines) forms the polar triangle of triangle ABC. Here we introduce the first
of two polar triangle results.

Theorem 4. The polar triangle of a polar triangle is the original triangle.

Proof. In Figure 2.6a, since C 0 is the pole of AB and A0 is the pole of BC,
B is 90 ◦ from both A0 and C 0 , so B is the pole of A0 C 0 . This also holds for
the other two sides in relation to A and B. The result follows.
An object is said to have a dual relationship with the original when taking
a transformation of the original and then applying the same transformation
to the result brings us back to our starting point. We can often learn a lot
from the dual of an object that we would not have known from studying the
original. This brings us to the Polar Duality Theorem.
CHAPTER 2. GEOMETRY ON THE SPHERE PART I 11

Theorem 5. The sides of a polar triangle are equal to the supplements of

the angles of the original. The angles of a polar triangle are equal to the
supplements of the sides of the original.

Proof. In Figure 2.6b, extending the arcs from A to points D and E on B 0 C 0

means we can say the angle at A is equal to DE, as A is the pole of B 0 C 0 .
Since C 0 is the pole of AD, C 0 D is also of length 90 ◦ , similarly for B 0 E since
B 0 is the pole of AE. Therefore

B 0 C 0 = B 0 E + C 0 D − DE = 180 ◦ − DE = 180 ◦ − A. (2.1)

Likewise for the other two sides of the polar triangle and the corresponding
angles of the original triangle opposite them.
The second part of Theorem 5 follows from noting that the angle B 0 =
EF . We know AE = CF = 90 ◦ , so

B 0 = AE + CF − AC = 180 ◦ − AC.

The result follows for A0 and C 0 too.

We can now complete the proof for the lower bound of the sum of angles
in a spherical triangle.
Proof. Since the perimeter of a spherical triangle is less than 360 ◦ , this holds
for a polar triangle, so we write

A0 B 0 + B 0 C 0 + A0 C 0 < 360 ◦ .

Thus, using Theorem 5 and Eq. (2.1),

(180 ◦ − C) + (180 ◦ − A) + (180 ◦ − B) < 360 ◦ ⇒ A + B + C > 180 ◦ .

Now that we understand some of the geometry on the sphere, we can

begin exploring theorems for spherical triangles upon it.
Chapter 3

Menelaus of Alexandria

When introducing this topic in Chapter 1 we mentioned Hipparchus of Nicaea,

the founding father of trigonometry. As Hipparchus lived over 2000 years ago,
there is not much that we can say about him. He was born in Nicaea, now
Iznik in modern day Turkey, in 190 BC and is believed to have died on the
island of Rhodes around 120 BC. His work as an astronomer and mathe-
matician is mainly known of from references to him in Claudius Ptolemy’s
Almagest. Another, slightly less elusive, character also praised by Ptolemy
lived around the first century AD, and is whom this chapter is about.
It is speculated that Menelaus was born around 70 AD in Alexandria,
Egypt, due to the scholars Pappus and Proclus referring to him as Menelaus
of Alexandria. He later made some astronomical observations in Rome ac-
cording to Ptolemy’s Almagest in 98 AD, and died around the year 130.
Menelaus was an astronomer and mathematician who produced many books
on geometry and trigonometry, however only one, Sphaerica, survives today
in its Latin and Arabic translations. Luckily, this is the most important for
us, as it concerns spherical triangles and their application to astronomy.
Although Sphaerica was not the first book on spherical trigonometry, it
was the most innovative. Up until this point, no astronomer or mathemati-
cian had the crucial tool of planar trigonometry as Menelaus did. This left
their works in vague territory where one arc was longer than another, but no
one could calculate by how much.
As the original Greek text of Sphaerica is lost, we are unsure of how
precise the translated copies are. It is likely that the translators altered some
of the text and historical content to suit their audiences, evident from the use
of the sine rather than the chord in the summary text. The main message
however is still there, which leads us to our first trigonometric theorem. The
proof of Menelaus’ Theorem A, Thābit’s proof of Theorem B, and information
on the celestial sphere are inspired by Van Brummelen (2013).

12
CHAPTER 3. MENELAUS OF ALEXANDRIA 13

Figure 3.1: The Menelaus configuration.

3.1 Menelaus’ Theorem

We begin with the Menelaus configuration, Figure 3.1. This construction
yields Menelaus’ planar theorem, which is stated in book II of Sphaerica,
but not proved. This suggests that the planar configuration and theorem
was already known. We will prove it for completeness and to use as an inter-
mediate theorem in projecting the Menelaus configuration onto the sphere.

Proof. In Figure 3.1, by extending a line from the point D, parallel to that of
KL, to a point X on the line AK, we highlight two sets of similar triangles,
illustrated.
Rewriting the ratio of the sides AK and KB and eliminating terms in-
volving X, we obtain the required result.
AK AK XK AT DL
= · = · . (3.1)
KB XK BK T D LB

To move Figure 3.1 onto the surface of the sphere, we place the planar
version behind its spherical counterpart, effectively cutting the sphere by
a plane in the shape of the Menelaus configuration, Figure 3.2. Here, the
planar configuration meets the surface of the sphere at points A, B and
D. Therefore, the lines and points falling in between these boundaries are
within the sphere, and those outside, point T , are outside of the sphere.
Menelaus’ planar theorem involves three straight lines and their ratios, so
to complete the transformation our aim is to ‘pop’ the straight lines onto
their corresponding arcs. That is, AKB → AZB, ADT → ADG, and
DLB → DEB. The first and third have line segments that fall within
CHAPTER 3. MENELAUS OF ALEXANDRIA 14

Figure 3.2: The Menelaus configu-

ration on the surface of the sphere.

(b) Proof for the partially exter-

(a) Proof for the internal lines.
nal lines.

Figure 3.3: Popping the line segments onto the sphere.

the sphere, but the second line segment to be projected onto the sphere is
partially outside of it. For this we will need two separate methods; let us
generalise these so as to use them for any arbitrary line segment. We will
first deal with the internal lines.
Proof. Figure 3.3a shows an arc of length a+b with the planar line underneath
it. This is part of a circle of radius 1, so the angles at the centre are a and b.
Using planar trigonometry, the two red lines in Figure 3.3a are of length sin a
and sin b, respectively, since we are dealing with two right-angled triangles.
As the two triangles about AC are similar, we can take the ratio of these
sides to get an immediate relationship between the straight line segment and
the arc:
AB sin a
= .
BC sin b
A similar proof follows for the partially external line segment using Fig-
ure 3.3b.
Proof. Here we have the length a going from the horizontal line at A to C
CHAPTER 3. MENELAUS OF ALEXANDRIA 15

and b going from the same horizontal line to B. Again the angles at the
centre are equal to the corresponding lengths, and by similar triangles we
yield the result
AC sin a
= .
AB sin b
Both of these results tell us that to pop the line segments onto the arcs on
the sphere, we need to take the sines of the arc lengths. Doing this to the lines
in Menelaus’ planar theorem, Eq. (3.1), we obtain Menelaus’ Theorem A:

sin AZ sin AG sin DE

= · . (3.2)
sin ZB sin DG sin EB
This is the first proposition in the third and final installment of Sphaerica,
but with Theorem A comes Menelaus’ Theorem B.
sin AB sin BD sin GE
= · . (3.3)
sin AZ sin DE sin GZ
Menelaus’ Theorem B was used by Ptolemy in the Almagest, but he did
not prove it. There are several ways of getting to the result, and since
Menelaus himself did not prove it, we are free to choose any of them. One
way would be to go back to the Menelaus configuration, Figure 3.4a.
Proof. This time, we extend a line from K to a point Y , parallel to DL in
Figure 3.4a, rather than parallel to KL from D. We can then manipulate
the new set of similar triangles to give
AB AB T K LT BD Y K LT BD LT
= · · = · · = · .
AK AK T L T K Y K DL T K DL T K
Replacing these line segments with the sines of the corresponding arc lengths
as before, we have Menelaus’ Theorem B.
We also could have gotten to this point by using Theorem A, Eq. (3.2).
This method was favoured by Thābit ibn Qurra, who used this proof long
after Menelaus’ time, in the 9th century. It goes like this.
Proof. Begin with the Menelaus configuration on the surface of the sphere,
Figure 3.4b. Extend arcs BA and BD until the two arcs meet at point X,
directly opposite B on the sphere. We now have the Menelaus configuration
XAZEGD to which we can apply Theorem A.
sin AZ sin GZ sin DE
= · .
sin AX sin GE sin DX
CHAPTER 3. MENELAUS OF ALEXANDRIA 16

(a) Planar Menelaus’ Theo- (b) Thābit’s proof of Theorem B.

rem B.

Figure 3.4: Proving Menelaus’ Theorem B.

As BX is a semicircle (for both arcs), AX = 180 ◦ − AB and therefore

sin AX = sin AB. Similarly, sin DX = sin BD. Making these substitutions
and rearranging eliminates our X term, and we are left with Theorem B,
Eq. (3.3).
These two equations were known by many as “regula sex quantitatem”
during medieval times, translated as “the rule of six quantities”. But how
was this strange configuration useful to astronomy?

3.2 Astronomy and the Menelaus Configura-

tion: The Celestial Sphere
The most important use of Menelaus’ Theorems for ancient and medieval
astronomers was converting between different coordinate systems on the ce-
lestial sphere. The different lines of reference mean that there are several
different ways of locating stars and planets. The celestial sphere embodies
the Sun, the Moon, and all of the stars and planets visible from Earth. When
observing these bodies, we can quickly draw three conclusions: all move in
circles; their speeds are constant; and the Earth is at the centre of these
motions.
We can account these properties back to Aristotle through Ptolemy, as he
describes them in the Almagest. The axis of these rotations is not dissimilar
to our own; the North Pole is the north star, Polaris, which can be seen as an
extension of our own North Pole. We can also think of the celestial equator
CHAPTER 3. MENELAUS OF ALEXANDRIA 17

Figure 3.5: Application of

Menelaus’ Theorem.

as a projection of our own equator. Due to the infinitely large nature of the
universe, the Earth is modelled as a pin prick at the centre of the celestial
sphere. The horizon of the celestial sphere is dependent on where on Earth
you are standing. The dome of universe surrounding you is unique to where
you are; the point where this dome meets the surface of the Earth is the
horizon. When standing in the northern hemisphere, the altitude of the
North Star from the horizon is equal to the latitude of the observer.
If we were to record the position of the Sun at the same time everyday
for a year, we would find that it changes slightly each day; it follows the path
of the ecliptic, shown in Figure 3.5 along with the North Pole, P, and the
equator. The Sun moves approximately 1 ◦ across the ecliptic each day. As
the ancient Babylonians used the sexagesimal system, and 360 is a multiple
of 60, dividing the sphere into 360 parts makes sense and coincides with the
number of days in a year. The spring equinox, where the ecliptic intersects
the equator, occurs at the point à, the symbol for the zodiac sign Aries.
The two lines make an angle of ε between them, which is the obliquity of the
ecliptic. This is equal to the tilt of the Earth’s axis, now around 23.44 ◦ .
Looking at figure 3.5, we can see the Menelaus configuration P CBAàS,
where S is the Sun. We can apply Menelaus’ Theorem to this to find the
equitorial coordinates of the Sun, given by the distances àA and AS.
Chapter 4

Medieval Islamic Mathematics

We fast forward to the 10th century. Menelaus’ Theorems stood the test of
time for the rest of the first millennium, but something exciting was happen-
ing in the Muslim world. It is difficult to determine who came up with our
next result first; different proofs and publications of the theorem appeared
almost simultaneously.
There were many influential Arab scholars in the medieval Middle East,
one of whom we met in Chapter 2: Abū Nas.r Mans.ūr. He was born in 970 in
Persia, and died around 66 years later in Ghazna, now Ghazni in Afghanistan.
Abū Nas.r was a talented mathematician, the teacher of al-Bı̄rūnı̄, and also a
prince. His family ruled over Khwarazm, the region where Abū Nas.r studied
under Abū ’l-Wafā’. When civil war broke out in Khwarazm, it is likely
that Abū Nas.r fled the area, along with al-Bı̄rūnı̄. He was later said to be
employed at the court of Ali ibn Ma’mun (where al-Bı̄rūnı̄ was also employed)
whose state was soon to be taken over by the Ghaznavids, whose capital was
Ghazna. As well as discovering the polar triangle (Chapter 2), Abū Nas.r was
responsible for preserving and simplifying many of Menelaus’ findings. He
is most famous for his work alongside al-Bı̄rūnı̄. The proofs in this chapter
follow the style of Van Brummelen (2013).

4.1 The Rule of Four Quantities

Whilst Menelaus’ Theorem gives us a “rule of six quantities” and was used by
many generations of astronomers, it was not the easiest to remember. With-
out it, however, we would not have Abū Nas.r’s Rule of Four Quantities.
Abū Nas.r’s original work on the topic appeared in the Book of the Azimuth,
but the only surviving reference we have to this is through al-Bı̄rūnı̄’s Keys
to Astronomy.

18
CHAPTER 4. MEDIEVAL ISLAMIC MATHEMATICS 19

(b) In the plane.

(a) Proof on the sphere.

Figure 4.1: The Rule of Four Quantities.

Proof. Figure 4.1a shows a Menelaus configuration where B is the pole to

AG, giving us two embedded right-angled triangles. Applying Menelaus’
Theorem B, Eq. (3.3), to the diagram, we get
sin AB sin BD sin GE
= · .
sin AZ sin DE sin GZ
As BD and AB are of length 90 ◦ , the sines of these sides are equal to 1, so
sin DE sin EG
= . (4.1)
sin AZ sin GZ

Aside It is here we have our first example of the principle of locality, which
is where we we can make a direct comparison of a spherical result with
its planar case.
On the graph of sin x, as x → 0, sin x → x. (The exact value of sin x
actually only tends to x when using radians, but seeing as we are not
using numbers here we need not worry about it too much.) Applying
this to the arrangement in the Rule of Four Quantities, we get
sin DE sin EG DE EG
= → = ,
sin AZ sin GZ AZ GZ
which is equivalent to the well known relation between taking the ratios
of the sides of similar triangles in the plane, shown in Figure 4.1b.

Clearly the Rule of Four Quantities is much easier to apply than the
Menelaus Theorems, which allowed for huge advances in astronomy. The
celestial sphere is not the only sphere that directly concerned mathematicians
though, the surface of the Earth too is spherical.
CHAPTER 4. MEDIEVAL ISLAMIC MATHEMATICS 20

(a) At a glance. Image adapted

from Google Maps (map, 2014).
(b) Al-Bı̄rūnı̄’s determination us-
ing the Rule of Four Quantities.

Figure 4.2: Finding the qibla.

4.2 Finding the Qibla

This particular application of the Rule of Four Quantities is in cartography,
but it came about for a religious purpose. People of the Islamic faith pray
facing the direction of Mecca five times each day. In doing this, they are
actually facing the Ka‘ba, the religious building that all Muslims are required
to visit at least once in their lives, and the direction their bodies must face
when buried after death, amongst other acts that must be performed in this
direction (King and Lorch, 1992). This is known as the qibla.
The problem that arose for Muslims was how to calculate the qibla. At
face value, it seems as though we can draw a right-angled triangle on the map;
we know the location of both Mecca and the worshipper, so can calculate the
differences in latitude and longitude of the two. Unfortunately, this is not the
case. Turning our attention to Figure 4.2a, we have the qibla from Ghazna
highlighted on the map of the Middle East, with the right-angled triangle
constructed from the differences in coordinates of the two locations. The
issue with this is that the bottom side of the triangle is not a great circle arc,
it is an arc of a circle of latitude. Therefore, the image we are looking at is
not even a triangle.
Various approximate solutions to this problem were proposed, some even
taking the triangle in Figure 4.2a to be planar. Around the 10th century,
solutions using Menelaus’ Theorems began to surface. After the Rule of
Four quantities was derived, methods of calculating the qibla were simpli-
fied. The familiar figure of al-Bı̄rūnı̄ was able to solve this problem in four
CHAPTER 4. MEDIEVAL ISLAMIC MATHEMATICS 21

different ways. He was no stranger to applying his mathematical knowledge

to geographical problems. In his work Determination of the Coordinates of
Cities he explains how to find the circumference of the Earth using only
planar trigonometry. It is in this book that he also calculates the qibla. His
methodology was then applied to over 3000 places across the Muslim world
by Shams al-Dı̄n al-Khalı̄lı̄. The qibla for each of these places was recorded
in a table spanning sixteen pages, for any struggling worshipper to look up.
We will follow in the steps of al-Bı̄rūnı̄ and calculate the direction of
Mecca from Ghazna, using Figure 4.2b and the Rule of Four Quantities,
Eq. (4.1). Not only was Ghazna the location of al-Bı̄rūnı̄ at the time, it was
also the capital of the Ghaznavid Empire and so a place of great importance.
Consider a bird’s-eye view above Ghazna, G, from the celestial sphere.
The outer circle of the sphere in Figure 4.2b is the horizon of Ghazna, and the
line that connects the north point on the horizon to the south is the meridian
of Ghazna, passing through the North Pole, P . The point M is where Mecca
would be if we were directly above it, with its associated meridian the arc
going through points P M B. The line going through W , M and A connects
the west point on the horizon to the meridian of Ghazna. GD is the great
circle arc connecting Ghazna to Mecca.
We know the latitudes of Ghazna, ϕG = 33.58 ◦ , and Mecca, ϕM = 21.67 ◦ ,
and from their longitudes we can calculate their difference in longitude as
∆λ = 27.37 ◦ . From these values we can determine all necessary distances
that we need for the qibla.
We begin with the CAP M B configuration. We know the difference in
longitude of the two locations, which is equal to the angle between the two
meridians at P . Therefore, BC = ∆λ. We also know P M = 90 ◦ − M B, so

sin P M sin P B sin(90 ◦ − ϕM ) 1

= ⇒ = ,
sin M A sin BC sin M A sin ∆λ
which gives us sin M A = cos ϕM sin ∆λ and we get M A = 25.29 ◦ .
Using the same method for the W M ACB configuration, we find sin AC =
sin ϕM / cos M A, which gives us AC = 24.11 ◦ . From this we can deduce that
GA = GC − AC = 9.47 ◦ .
It is at this point that it is useful to think about what it is we want to
find. The qibla is a direction, not a length, so how does the Rule of Four
Quantities enable us to find this? It comes from looking out to the horizon.
Using configuration W M ASD, we find sin M D = cos M A cos GA, so
M D = 63.10 ◦ . We can now take the final step in our calculation, by finding
the distance DS from the configuration GM DSA. Doing so, we get sin DS =
sin M A/ cos M D, which tells us DS = 70.79 ◦ .
CHAPTER 4. MEDIEVAL ISLAMIC MATHEMATICS 22

This solves our problem; when standing in Ghazna, one must turn 70.79 ◦
west from the south in order to face Mecca.

4.3 The Spherical Law of Sines

As well as its extensive practical applications, the Rule of Four Quantities
was used to make further advances in spherical trigonometry. There was
plenty of controversy over who discovered the spherical Law of Sines, mainly
between Abū Nas.r and his teacher, Abū ’l-Wafā’. Al-Bı̄rūnı̄ scorned Abū
’l-Wafā’ in his Keys to Astronomy for taking credit for the theorem, but
this opinion may of course be biased, given al-Bı̄rūnı̄’s relationship with Abū
Nas.r.
Abū ’l-Wafā’ al-Būzjānı̄ was born in 940 in Būzjān, Iran, and died in
998 in Baghdad, Iraq. He first moved to Baghdad to work in the court of
the ruler of Iran and most of Iraq at the time, Adud ad-Dawlah, alongside
other respected scientists. Amongst Abū ’l-Wafā’s works was his very own
Almagest, the same title as that by Ptolemy. His version, however, was much
more comprehensive and covered exciting new ideas, such as applying the
tangent function and the inverse trigonometric functions; the secant, cosecant
and cotangent, to astronomy. Along with these innovative applications, Abū
’l-Wafā proved the spherical Law of Sines.
Proof. We begin with an arbitrary triangle on the surface of the sphere in
Figure 4.3a, with angles A, B and C, and sides of length a, b and c. If we
choose a vertex, say C, to drop a perpendicular from, we split our triangle
into two right-angled triangles, Figure 4.3b.
We now choose A and B to be poles, and draw their corresponding great
circle arcs by extending their adjoining sides to be 90 ◦ long, Figure 4.3c. It is
now possible to apply the Rule of Four Quantities, Eq. (4.1), twice, by using
the line CD to create two sets of embedded right-angled triangles. Doing so,
we get
sin CD sin AC
= ⇒ sin CD = sin A sin b,
sin EZ sin AZ
and
sin CD sin BC
= ⇒ sin CD = sin B sin a.
sin T H sin BT
Combining these two equations and rearranging gives us
sin a sin b
= .
sin A sin B
CHAPTER 4. MEDIEVAL ISLAMIC MATHEMATICS 23

(a) An arbitrary triangle (b) Dropping a perpendicu-

with acute angles. lar from C.

(c) Making A and B poles. (d) Obtuse-angled triangle.

Figure 4.3: Proof of the spherical Law of Sines.

Of course, we could have dropped the perpendicular from any of the

vertices in Figure 4.3a, which gives us the complete result:
sin a sin b sin c
= = . (4.2)
sin A sin B sin C

But what if the perpendicular we drop does not fall within the triangle?
If we have an obtuse angle in the triangle, the line dropped from the opposite
vertex does not fall within the boundaries of the triangle itself, Figure 4.3d.
We therefore need to take a slightly different route in proving the spherical
Law of Sines.
Proof. In Figure 4.3d, drop a perpendicular from the point C onto an ex-
tension of the line c. From here, define the pole and great circle arc for A
as before, but take the great circle arc of B to the opposite side, instead
extending c to the right. We again apply the Rule of Four Quantities, and
the result follows.
What is special about this result is that the spherical Law of Sines was
the first equation of its kind; up until now we have not known of an equation
CHAPTER 4. MEDIEVAL ISLAMIC MATHEMATICS 24

that gives us a relationship between sides and angles of a spherical triangle.

It would not be surprising if the spherical Law of Sines went on to transform
the work of scientists from this point onwards, being able to apply it to any
triangle, as opposed to purely right-angled triangles, must have been eye-
opening. Unfortunately, this was not the case. The Rule of Four Quantities
was still the main tool in any astronomers pocket, as the configuration was
somewhat more useful for the quantities they were trying to find. It was not
until the Renaissance in Europe that the spherical Law of Sines took off.

Principle of Locality Aside from the lacklustre set of applications of the

spherical Law of Sines, we can find its planar brother. Using the same
argument as in Section 4.1 for the Rule of Four Quantities, reducing
the sides of a spherical triangle reduces the spherical Law of Sines to
the planar Law of Sines:
a b c
= = .
sin A sin B sin C
Chapter 5

Right-Angled Triangles

It is not often that in every day life a triangle has an angle of exactly 90 ◦ .
John Napier, however, did not see working purely with right-angled triangles
as a limitation, realising that any oblique triangle can be split into two right-
angled triangles by dropping a perpendicular from one of the vertices—just
as we do in planar trigonometry today. Napier was a Scottish Protestant who
lived from 1550 to 1617. He was greatly interested in theology, the study of
the nature of God and religious belief, and attended St. Andrew’s University
at the age of just 13.
Despite his discoveries, Napier was not primarily a mathematician; he
was an inventor. The mathematical ideas he came up with were mainly tools
to help him find something else quicker. The logarithm is a prime example
of one of these inventions. Napier published Mirifici logarithmorum canonis
descriptio in 1614, a book dedicated to easing the difficulty of arduous mul-
tiplication and division of trigonometric functions, a task that dominated all
work with spherical trigonometric formulae, by adding them instead. His
idea was this: to create a system of two sets of numbers such that when
x = x1 + x2 , y = y1 y2 . He eventually managed this with the help of Henry
Briggs of Gresham College, London, finding y = 10x gave the correct result.
This innovation made the work of future astronomers and mathematicians,
including that of Johannes Kepler, much simpler. We follow the proofs in
Van Brummelen (2013) to see some interesting results in this chapter.

5.1 Pythagoras’ Theorem on the Sphere

At the beginning of the 12th century lived Jābir ibn Aflah. of Seville, Spain.
Jābir was not known for excelling in his scientific findings, however he did
rewrite Ptolemy’s Almagest in his most famous work, Correction of the Al-

25
CHAPTER 5. RIGHT-ANGLED TRIANGLES 26

Figure 5.1: Proving Geber’s The-

orem.

magest. Fortunately for Jābir, this was found by astronomers in late medieval
Europe, bringing his ten fundamental identities of spherical right-angled tri-
angles to light, known as Geber’s Theorem.
To find these spherical identities, we need to use planar trigonometric
results. To do this, we must express the spherical angles A and B in Figure 5.1
as planar angles between two line segments. Create a planar right-angled
triangle in the configuration in Figure 5.1 by dropping a perpendicular from
a point D on the line OB, then join D and E to a point F on OA. This
constructs angle A at point F and subsequently produces three more planar
right-angled triangles on the faces attached to the centre of the sphere, O.
From here, applying trigonometry at each vertex of the four right-angled
planar triangles, we obtain four results relating the sides and angles of right-
angled triangles on the sphere:

sin a = sin A sin c,

sin b = tan a cot A,
cos A = tan b cot c,
cos c = cos a cos b. (5.1)

The remaining 6 identities can be derived in a similar fashion. The last of

the above identities is particularly important.
Principle of Locality Using the Maclaurin series expansion of the cosine,
x2 x4
cos x = 1 − + − ...,
2! 4!
tells us that for small x we can approximate cos x by 1−x2 /2. Substitut-
ing this expression into each term of cos c = cos a cos b and rearranging,
we get
2 2 2 a2 b 2
c =a +b + .
2
CHAPTER 5. RIGHT-ANGLED TRIANGLES 27

This simplifies to c2 = a2 + b2 because as the sides of the triangle

diminish, the last term becomes infinitesimally small. This is the planar
Pythagorean Theorem, so we can deduce that

cos c = cos a cos b (5.2)

is the spherical Pythagorean Theorem.

Pythagoras’ Theorem in planar triangles has actually been known and in

use since the ancient Babylonians. Specific cases of using a 3,4,5 triangle in
building ancient settlements date back much further, but lacked a formula
to describe the relation.
Clearly the application of Napier’s logarithm to the multiple products of
trigonometric functions in Geber’s Theorem, Eq. (5.1), was indispensable,
minimising hours of work. Pierre Siméon de Laplace was quoted praising
the work of Napier long after his death saying “by shortening the labours,
[logarithms] doubled the life of the astronomer”.

5.2 The Spherical Law of Cosines

We have seen the relevance of Euclid’s Elements, Section 2.2.1, when dis-
cussing properties of both spherical and planar triangles. To look again at
Euclid’s work for inspiration on the Law of Cosines seems counter-intuitive,
seeing as the Elements was written in the 3rd century BC, before trigonom-
etry had been invented. Book II of the Elements, however, can be seen as a
statement about algebraic geometry, Proposition 13 about acute-angled tri-
angles is of particular interest. The statement translates as this, in reference
to Figure 5.2a:

In triangle ABC with an acute angle at C and a perpendicular

dropped from A onto BC (defining D), c2 = a2 + b2 − 2CD · BC.

Since BC = a and CD = bcosC, Euclid actually had the planar Law of

Cosines right under his nose, he just did not have the trigonometric functions
to express it as we do today.
Proof. An application of Pythagoras’ Theorem on the RHS right-angled tri-
angle of Figure 5.2a quickly gives us the required result.
We use the same principle for the spherical version.
CHAPTER 5. RIGHT-ANGLED TRIANGLES 28

(a) The planar Law of Cosines.

(b) The spherical Law of Cosines.

Figure 5.2: Proof of the Law of Cosines.

Proof. Copying the planar configuration, including the perpendicular dropped

from A, onto the sphere, Figure 5.2b, we apply the spherical Pythagorean
Theorem, Eq. (5.2), to both right-angled triangles, giving
cos c = cos h cos(a − x) and cos b = cos h cos x.
Rearranging for cos h and applying the cosine subtraction law to cos(a − x),
we can equate these two statements
cos c cos x = cos b(cos a cos b + sin a sin b).
We want to eliminate terms involving x in order to generalise our formula
to any ABC triangle. Dividing through by cos x, we obtain a tan x term in
our equation. To get rid of this final term involving x, we turn to the third
of the fundamental trigonometric identities, Eq. (5.1); cos A = tan b cot c.
Putting this in terms of the right-angled triangle on the LHS of Figure 5.2b
and substituting for tan x, we obtain the spherical Law of Cosines
cos c = cos a cos b + sin a sin b cos C. (5.3)

As with every major theorem on the sphere so far, the planar version of
the spherical Law of Cosines is not far away.
Principle of Locality Section 5.1 uses the Maclaurin series to relate the
cosine function to an expression for small x; cos x ≈ 1 − x2 /2. We also
know from Section 4.1 that for small x, sin x reduces to x, approxi-
mately.
Applying these results to the spherical Law of Cosines, we have
c2  a2  b2 
1− = 1− 1− + ab cos C.
2 2 2
CHAPTER 5. RIGHT-ANGLED TRIANGLES 29

Thus, we have the planar equivalent

c2 = a2 + b2 − 2ab cos C.

This particular relation between the three sides of a triangle and one of
the angles lends itself well to problems on the surface of the Earth and on the
celestial sphere. Most textbooks on the subject require students to calculate
distances on the Earth, such as popular sea routes, like that of the Titanic.
The trick is to draw the great circle arc between the start and end point of
the journey, the arc CB in figure 5.2b, then connect both at the North Pole,
A. Doing so gives us two of the side lengths of the triangle and the angle
between them, which can then be put into the spherical Law of Cosines to
obtain the distance required.
When the situation arises that we know two angles in a triangle and the
side between them in the plane, the planar Law of Cosines does not help.
Instead we find the third angle and apply the planar Law of Sines. We cannot
do this on the sphere as the sum of the internal angles of a spherical triangle
is not one size fits all. We can however use the Polar Duality Theorem
(Theorem 5, Section 2.2.2).
This allows us to convert between sides and angles on the sphere, which
is easily done to the spherical Law of Cosines to give us a formula relating
the three angles and one side of a spherical triangle:

cos(180 ◦ − C) = cos(180 ◦ − A) cos(180 ◦ − B)

+ sin(180 ◦ − A) sin(180 ◦ − B) cos(180 ◦ − c).

This gives us the Law of Cosines for Angles

cos C = − cos A cos B + sin A sin B cos c.

This eased the worries of mathematicians wanting to find the remaining

angle when a side and two angles were known. We now have a complete set
of spherical trigonometric tools for working with sides and angles of spherical
triangles.
Chapter 6

Geometry on the Sphere

Part II

The application of spherical trigonometry to the Earth and Heavens consisted

mainly of two things: calculating the length of sides of spherical triangles to
find distances between points of interest; and finding the angle between two
sides to calculate the direction we must face to find them. Something of little
interest to astronomers and cartographers was the area of the triangles they
were working with. As mathematicians, a thorough look at any topic is the
only way to approach it, as interesting results may be lying just beneath the
surface. We will therefore fill the final gap in our knowledge by finding the
area of spherical triangles and polygons, Part II of our geometrical quest.

6.1 Areas of Spherical Polygons

The simplest shape we can draw on the surface of the sphere, as discussed
in Section 2.1, is a circle as it consists of just one side. In particular, a great
circle has been of most interest to us, as it takes the role of a straight line on
the sphere. Although this is not a polygon as such, it is a triangle with three
angles of 180 ◦ , which is a polygon. A great circle splits the sphere into two
hemispheres, both with area half of that of the surface area of the sphere. In
our case of the unit sphere, this is 1/2 · 4πr2 = 2π.
Almost as obscure as a one sided spherical polygon is a spherical polygon
with just two sides. This occurs by joining two great circle arcs of length 180 ◦
at their antipodal points—two points directly opposite one another on the
sphere, Figure 6.1a. The lune formed when doing this has an area depending
on the size of the angle, θ, between the two semicircles, which is relatively
easy to find. From Todhunter and Leathem (1949), the ratio of θ over 360 ◦

30
CHAPTER 6. GEOMETRY ON THE SPHERE PART II 31

(b) The area of a spherical trian-

(a) The area of a lune.
gle.

Figure 6.1: Calculating the area of a spherical polygon.

is equal to the ratio of the area of the lune over the area of the sphere, so
4π · θ πθ
area of lune = = . (6.1)
360 ◦ 90
The area of a spherical triangle was not discovered until the 17th century
by Albert Girard. Girard was a French born mathematician who was forced
to move to the Netherlands because of his Protestant faith, something that
saddened him for much of his life. He made large contributions to the areas
of algebra, arithmetic and trigonometry, including being the first to use the
“sin”, “sec” and “tan” abbreviations in his Trigonométrie in 1629. This is
where we also find his work on spherical triangle areas. His idea was this, in
reference to Figure 6.1b (Van Brummelen, 2013):
Proof. Extend the three sides of triangle ABC into their respective great
circles. A0 , B 0 and C 0 are the antipodal points of A, B and C, so the triangle
A0 B 0 C 0 is the same as the original triangle ABC. The great circle of BC is
the outer circle of the sphere. We now have several triangles and lunes on the
surface of the sphere. Each lune comes from extending the adjoining lines of
the vertices of the original triangle to their antipodal point, so we have 12 in
total—4 at each point. Focusing on the lune from each point that includes
the original triangle, we add the triangles involved to get

3 · ABC + A0 BC + AB 0 C + ABC 0 .

Another pair of equal triangles in this configuration is A0 BC and ABC 0 ,

so we can swap the former for the latter in the above equation. Doing so
CHAPTER 6. GEOMETRY ON THE SPHERE PART II 32

Figure 6.2: Finding the area of a

spherical polygon.

gives us the 4 triangles that make up the front hemisphere of Figure 6.1b, so
3 · ABC + A0 BC + AB 0 C + ABC 0 = 2 · ABC + 2π. Since this is still the sum
of three lunes, we can set it equal to the sum of their areas, using Eq. (6.1),
π
2 · ABC + 2π = (A + B + C),
90
which gives us the area we require
π
area of triangle = (A + B + C − 180 ◦ ).
180
So the area of a spherical triangle is proportional to the amount it’s angles
exceed 180 ◦ —the sum of the angles in a planar triangle.
In the plane, to find the area of a polygon we can divide it up into
triangles. The same applies to polygons on the sphere. Here we will extend
the formula for the area of a triangle.
Take the arbitrary polygon in Figure 6.2 and divide it into triangles by
picking a random point inside it. If the polygon has n sides, its area is n
times the area of a triangle (Van Brummelen, 2013)
π
area of polygon = (sum of triangles’ angles − n · 180 ◦ ),
180
π
= (sum of polygon’s angles + 360 ◦ − n · 180 ◦ ),
180
π
= [sum of polygon’s angles − (n − 2) · 180 ◦ ]. (6.2)
180
Similar to the spherical triangle, the area of a spherical polygon is pro-
portional to the amount it’s angles exceed (n − 2) · 180 ◦ . This is unsurprising
as the sum of the angles of a planar polygon is equal to (n − 2) · 180 ◦ .
CHAPTER 6. GEOMETRY ON THE SPHERE PART II 33

Figure 6.3: Legendre’s

proof of Euler’s Polyhedral
Formula.

6.2 Euler’s Polyhedral Formula

Leonhard Euler was a Swiss born mathematician that lived between 1707 and
1783. He is probably the most productive mathematician of all time, not just
the 18th century, making developments in almost all areas of mathematics.
His father was a Protestant minister, who introduced Euler to mathematics
before sending him to university in Basel, Switzerland, to get some further
education before doing a degree in theology. It was here that Euler discovered
his love and talent for mathematics, owing his father’s change of mind that
he could instead do a mathematical degree to Johann Bernoulli, who had
previously studied with Euler’s father and worked at the University of Basel.
Euler wrote an average of three pages of mathematics per day throughout
his life, including the time after he became blind. Amazingly, it is during
this time that Euler developed his Polyhedral Formula.
A convex polyhedron is a 3-dimensional shape whose sides are made up
of polygons, with none denting inwards. There are five (and no more) reg-
ular polyhedra, each of whose sides are made up purely of identical regular
polygons, but there are many more irregular polyhedra, where the sides can
be made up of a mixture of different regular or irregular polygons.
Theorem 6. For any polyhedron, with V vertices, E edges and F faces,
V − E + F = 2.
This is Euler’s Polyhedral Formula. Euler’s proof of this theorem for all
polyhedra was not considered rigorous enough when he initially wrote it,
but there are now a total of 19 proofs. The first came from a completely
unexpected area of mathematics, in Book 7 of Éléments de Géométrie enti-
tled The Sphere by Adrien-Marie Legendre. We will use Legendres proof as
shown in Van Brummelen (2013).
Proof. Legendre began his proof by projecting a polyhedron onto the sphere,
as demonstrated in Figure 6.3 with the projection of the cube. The polyhe-
CHAPTER 6. GEOMETRY ON THE SPHERE PART II 34

dron is at the centre of the sphere, with a light source at its centre casting
shadows of its edges and vertices onto the surface of the sphere. The edges
are great circle arcs, and form a spherical polyhedron.
Using the area of a spherical polygon, Eq. (6.2), we can sum all of the
faces of the spherical polyhedron to give the surface area of the unit sphere:
X π
[sum of polygon’s angles − (n − 2) · 180 ◦ ] = 4π,
180
which we can expand to give
X
sum of all angles − [n · 180 ◦ − 2 · 180 ◦ ] = 720 ◦ .

Since the number of polygons is equal to the number of faces of the spherical
P
polyhedron and each edge of the polyhedron is counted twice, we have [n ·
180 ◦ − 2 · 180 ◦ ] = 2E − 2F · 180 ◦ . All of the angles are at vertices, so the
sum of the angles is V · 360 ◦ . Thus,

V · 360 ◦ − 2E + 2F · 180 ◦ = 720 ◦ ⇒ V − E + F = 2.

Clearly as this holds for spherical polyhedra it holds for planar polyhedra
too, since the former is a generalisation of the latter.
Chapter 7

Quaternions

For many years, algebra was thought of as an extension of normal arithmetic,

obeying the laws of commutativity, associativity and closure (a property that
says numbers stay in the same set they are originally in under operation, e.g.
if a and b are real numbers, so is a + b) under algebraic operations. The 16th
century saw the introduction of complex numbers by Bombelli, an Italian
algebraist, with the algebra of complex numbers not being cemented into
mathematics until the work of Gauss in the 19th century. We can interpret
the complex number z = a + ib, where a and b are real numbers, as the
vector (a, b) in the plane. The magnitude of z is the length of the vector
from the origin and the angle between the vector and the x-axis is the angle
of z. Multiplication of complex numbers can be thought of as a rotation in
the plane.
William Rowan Hamilton was particularly interested in these rotations
and wanted to define a similar relationship for 3-dimensional space. In 1843,
he opened up the world of algebra with his introduction of the quaternion,
a hyper-complex number of rank 4 that violated the law of commutativity
under multiplication. Whilst walking one morning along the Royal Canal in
Dublin with his wife, Hamilton had a stroke of inspiration; something that
he was so excited by, he carved it into the bridge they were passing by:
i2 = j2 = k2 = ijk = −1. (7.1)
This was the invention of the quaternion, a complex number consisting
of three complex components and four real numbers.

7.1 Some Useful Properties

The tree of quaternions has many branches; we will only look at the relevant
properties needed in order to apply quaternions to the sphere. A good ex-

35
CHAPTER 7. QUATERNIONS 36

planation of quaternions in all their glory, and the source for the information
in this chapter, is Quaternions and Rotation Sequences: A Primer with Ap-
plications to Orbits, Aerospace and Virtual Reality, Kuipers (1999), as well
as Quaternions and Rotation Sequences, Kuipers (2000).
We denote a quaternion with a lower case letter, say q, and write it
as an element of R4 : q = (q0 , q1 , q2 , q3 ), where the components q0 , . . . , q3
are real numbers. The vectors i, j and k denote the orthonormal basis of
R3 , so the quaternion q is the sum of a scalar part, q0 , and a vector part,
q = iq1 + jq2 + kq3 . The scalar part of a pure quaternion is zero.

7.1.1 The Quaternion Product

Addition of quaternions follows the usual algebraic rules of adding each of the
corresponding components, as with multiplying a quaternion by a scalar—
each element is multiplied by the scalar. The difference occurs when taking
the product of two quaternions. Along with Hamilton’s Eq. (7.1), there are
3 more equations implied by this to make note of:
ij = k = −ji,
jk = i = −kj,
ki = j = −ik.
Keeping these in mind, the product of two quaternions, say p and q, with
some algebra, can be written as
pq = (p0 + ip1 + jp2 + kp3 )(q0 + iq1 + jq2 + kq3 )
= p0 q0 − (p1 q1 + p2 q2 + p3 q3 ) + p0 (iq1 + jq2 + kq3 ) + q0 (ip1 + jp2 + kp3 )
+ i(p2 q3 − p3 q2 ) + j(p3 q1 − p1 q3 ) + k(p1 q2 − p2 q1 ).
This can be written more concisely by using the notation of the cross and
dot products of two vectors, p and q, given that p = p0 + p and q = q0 + q;
pq = p0 q0 − p · q + p0 q + q0 p + p × q. (7.2)
This is known as the quaternion product.

7.1.2 The Complex Conjugate and Inverse Quaternions

Similar to the complex conjugate of any complex number, the quaternion
has its own complex conjugate. For the quaternion q = q0 + q, we define its
complex conjugate as
q ∗ = q0 − q = q0 − iq1 − jq2 − kq3 .
CHAPTER 7. QUATERNIONS 37

After a lot of algebra, it can also be shown that the complex conjugate of a
product of quaternions is the product of the complex conjugates of the two
quaternions in reverse order. That is (pq)∗ = q ∗ p∗ .
The norm of a quaternion q is the scalar denoted by N (q) or |q|. This
represents the length of the quaternion, defined as
√
N (q) = q ∗ q, or N 2 (q) = q ∗ q.

As with ordinary complex numbers, if a quaternion has a norm of 1 it is

said to be a unit quaternion or normalized.
Combining the ideas of the complex conjugate and the norm of quater-
nions, we can define the inverse of a quaternion. By the definition of an
inverse, we have
q −1 q = qq −1 = 1.
Multiplying this by the complex conjugate,
q∗
q −1 qq ∗ = q ∗ qq −1 = q ∗ ⇒ q −1 = .
N 2 (q)

Therefore, if q is normalized its inverse is equal to its complex conjugate:

q −1 = q ∗ . To take advantage of this property, all quaternions discussed from
here on will be unit quaternions. We ensure this is the case by writing

q = q0 + q = cos θ + u sin θ, (7.3)

where u = q/|q|, since sin2 θ + cos2 θ ≡ 1 for all θ, so N (q) = 1.

7.2 Rotation Sequences

Now that we have defined the basis of what a quaternion is and the necessary
properties to apply them to spherical trigonometry, we can begin to relate
them to rotations in the R3 space. Since the product of complex numbers,
which consist of the sum of a scalar and one imaginary part, results in a rota-
tion of a vector in the plane, it makes sense that a rotation in 3-dimensional
space will require a scalar part and three imaginary components: the product
of quaternions.
By introducing the unit quaternion, Eq. (7.3), we introduced the cosine
and sine of θ in order to ensure a length of unity, which gives us the perfect
opportunity to relate θ to the angle of rotation. Kuipers (1999, p. 127–132)
does just this, stating that the unit quaternion acting on a vector represents
CHAPTER 7. QUATERNIONS 38

a rotation of the vector by an angle of 2θ about q, the vector part of q. From

this, we can say
θ θ
qz,θ = cos + k sin
2 2
represents a rotation of θ about the z-axis. Similarly for the y and x axes,
we have
θ θ
qy,θ = cos + j sin ,
2 2
θ θ
qx,θ = cos + i sin . (7.4)
2 2
Note the presence of θ in the subscript of q and θ/2 in the equation. Here
we have introduced a new notation, where the subscript of q tells us the axis
of rotation as well as the angle. This will come in handy soon.
To denote a rotation sequence we will adopt another new notation, shown
in Figure 7.1a. This represents a clockwise rotation about the z-axis by an

(a) A single rotation. (b) A rotation sequence.

Figure 7.1: Rotation notation.

angle of α, which we write using quaternions as qz,α . It takes us from the

original coordinate frame to a new one. A sequence of rotations is then
a string of these symbols, read from left to right, Figure 7.1b, with the
corresponding sequence of quaternions also written in this order: qz,α qy,β .
This is an open rotation sequence, as the final coordinate frame is not the
same as the initial frame.
A closed rotation sequence is one that, after transformation, takes us back
to the original coordinate frame. Intuitively, an obvious way to do this is to
do the inverse of the previous rotations, Figure 7.2. Rotating the coordinate

Figure 7.2: A closed rotation sequence.

frame by an angle of α about the z-axis, followed by an angle of β about the

y-axis, to recover the original frame we must rotate by −β about y then −α
CHAPTER 7. QUATERNIONS 39

about z. Let us check this with quaternions. The inverse of a sequence of

unit quaternions is equal to the complex conjugate. Taking the conjugate of
a product reverses the order, so (qz,α qy,β )∗ = qy,β
∗ ∗
qz,α = qy,−β qz,−α .
Closed rotation sequences can be started at any point in the sequence, as
long as the order of rotations is maintained. It is because of this property that
we can write a string of quaternions representing a closed rotation sequence
equal to the identity. In the example of Figure 7.2, we write
∗ ∗
qz,α qy,β qy,β qz,α =1 (7.5)
and, starting from the middle of the sequence,
∗ ∗
qy,β qz,α qz,α qy,β = 1.
It is of great use to us to be able to split a rotation sequence up. This
can be done very easily by multiplying a sequence on the right or left by the
inverse of the last quaternion. Continuing with our closed loop example, we
∗
can multiply the LHS of Eq. (7.5) by qz,α , and the RHS by qz,α , giving
∗ ∗
qy,β qy,β = qz,α qz,α ,
which we know is true as, for this example, both sides are equal to 1.

7.2.1 3-Dimensional Representation

Rotating a vector of constant length, with fixed origin, about a coordinate
frame in 3-dimensional space maps out a circle, so when rotating normalized
quaternions we are effectively drawing great circles on the surface of a unit
sphere. The rotation sequence in Figure 7.1b looks like Figure 7.3a in 3-
dimensional space. It is a transformation of the coordinate frame XY Z to
xyz. Rotating about the Z-axis by angle α we arrive at the frame x0 yZ, then
a rotation of β about the y-axis takes us to the final coordinate frame of xyz.
Here we have drawn two great circle arcs of length α and β.
It can be shown that the sequence in Figure 7.4 maps out an arbitrary
great circle arc, shown in Figure 7.3b (Kuipers, 1999, p. 216–218). Here λ
and L can be thought of as the longitude and latitude of the start and end
points of the great circle arc: in this example, the arc goes from A at (λ1 , L1 )
to B at (λ2 , L2 ).

7.3 Deriving the Spherical Law of Sines

We are now ready to show how the use of quaternions and rotation sequences
can give us spherical trigonometry results. We begin by mapping out a
CHAPTER 7. QUATERNIONS 40

(a) A rotation of α about the Z- (b) Great circle path from A to

axis followed by a rotation of β B.
about the y-axis.

Figure 7.3: 3-dimensional rotation sequences.

Figure 7.4: A great circle rotation sequence.

spherical triangle, Figure 7.5, shown on the surface of the unit sphere in
Figure 7.6 (full details are shown in Kuipers (1999, p. 237–241)). This

Figure 7.5: The rotation sequence for a spherical triangle.

produces the quaternion closed loop sequence of

∗ ∗
qx,b qz,δ qx,a qz,β qx,d qz,γ = 1,
which we can appropriately pre and post-multiply to give us
∗ ∗ ∗
p = qz,β qx,d qz,γ = qx,a qz,δ qx,b = r.
As we know how to write each quaternion given its axis and angle of rotation,
Eq. (7.4), we can work out the components of p and r by expanding the
CHAPTER 7. QUATERNIONS 41

Figure 7.6: The spherical triangle of

the rotation sequence in Figure 7.5.

following, using the quaternion product, Eq. (7.2):

∗
p = p0 + ip1 + jp2 + kp3 = qz,β qx,d qz,γ
 β β  d d  γ γ
= cos + k sin cos − i sin cos + k sin
2 2 2 2 2 2
and
∗ ∗
r = r0 + ir1 + jr2 + kr3 = qx,a qz,δ qx,b
 a a  δ δ  b b
= cos − i sin cos − k sin cos + i sin .
2 2 2 2 2 2
We are now just a few steps away from obtaining our result. As p and r
are equal, each of their components must be equal, so we can write p0 = r0 ,
etc. With some algebra of the above equations for p and r, we get
d α+β δ a−b
p0 = − cos sin = cos cos = r0 , (7.6)
2 2 2 2
d α−β δ a−b
p1 = sin sin = − cos sin = r1 , (7.7)
2 2 2 2
d α−β δ a+b
p2 = sin cos = − sin sin = r2 , (7.8)
2 2 2 2
d α+β δ a+b
p3 = cos cos = − sin cos = r3 . (7.9)
2 2 2 2
Multiplying Eq. (7.7) by Eq. (7.9) and simplifying, we obtain
sin d sin α + sin d sin β = sin δ sin a + sin δ sin b. (7.10)
Similarly, multiplying Eq. (7.8) by Eq. (7.9), we get
sin d sin α − sin d sin β = sin δ sin a − sin δ sin b. (7.11)
Adding Eq. (7.10) and Eq. (7.11), then subtracting the two, we achieve our
goal: the spherical Law of Sines
sin d sin a sin b
= = .
sin δ sin α sin β
Chapter 8

Conclusions

It is clear to see that the role of spherical trigonometry in the development

of astronomy and Earth related measurements, as well as knowledge of the
sphere, was crucial. As we have journeyed through ancient, medieval and
modern times, we have touched upon some of the important applications of
spherical trigonometry results, from calculating the position of the Sun to de-
termining the direction of Mecca from any point on the surface of the Earth.
Without the inquisitive minds of people who lived thousands of years ago,
such as Hipparchus, trigonometric functions and knowledge and exploration
of the heavens would not be as advanced as they are today.
We have seen how spherical trigonometry has grown and simplified over
time, beginning in the first century with Menelaus’ Theorem, a difficult the-
orem with six quantities to memorise and a less than obvious configuration
to look for. Although it was revolutionary in gaining information of the
positions of stars and planets, the invention of the Rule of Four Quantities
from medieval 10th century mathematician Abū Nas.r simplified the work of
astronomers greatly. This also lead to an efficient and accurate method of
calculating the direction of Mecca, the qibla, from Ghazna by the work of
al-Bı̄rūnı̄, which gave the rest of the Muslim world access to the qibla from
their locations too. The Rule of Four Quantities is arguably the most impor-
tant result of spherical trigonometry, as it also gave us the derivation of the
spherical Law of Sines by Abū ’l-Wafā’—a result that finally related sides
and angles of spherical triangles, a property results prior to this lacked. One
thing all of these relations had in common was their extensive use of the sine
function, requiring users of such relations to trawl through tables of sines to
find their values, to then spend large proportions of time multiplying and
dividing them—an arduous task without a calculator. This was the case up
until the 16th century when Napier invented the logarithm in order to make
calculations simpler. The first tables of logarithms in fact were not of pure

42
CHAPTER 8. CONCLUSIONS 43

numbers, but of logs of sines of angles. Finally we met the spherical Law
of Cosines, a flexible result relating three sides and an angle of a spherical
triangle, or conversely the spherical Law of Cosines for Angles, giving us
the opposite. This result was ideal for navigation, enabling us to calculate
distances on the surface of the Earth when our information is limited.
We then explored a completely different topic; Hamilton’s quaternions.
The ability to obtain spherical trigonometry results from modern areas of
mathematics, as well as the publication of these, shows how important and
versatile spherical trigonometry is.
It is equally important to note that the applications of spherical trigonom-
etry did not die out when technology took over; it is still in use in sophisti-
cated computer programming today.
In order to create a sampling algorithm in computer graphics, the Law of
Cosines for Angles is applied. The problem arises from the angle subtended
by a spherical polygon, and can only be solved by breaking the polygon up
into spherical triangles. The use of stratified sampling, a measure of reduc-
ing variance, is applied to maintain a uniform distribution of random sam-
ples within the spherical triangles, giving the image a smoother appearance.
(Arvo, 1995)
Of particular interest to us is animation, which combines the use of
quaternions for their spherical applications to create 3-dimensional images.
Spherical linear interpolation (SLERP) of quaternions is used to obtain the
orientations given by a string of quaternions in order to rotate objects. This
links directly to the rotation sequences we saw in Chapter 7. The use of
quaternions and the SLERP approach, as opposed to linear interpolation,
means that animations maintain a constant speed throughout their dura-
tion, as opposed to speeding up in the middle when using linear interpolation.
(Barrera et al., 2004)
The application of spherical trigonometric results that were invented hun-
dreds of years ago to things we see on a daily basis, such as computer graphics
or animated films, shows how relevant the subject still is. The development
of spherical trigonometry throughout the ages therefore plays a larger part
in 21st century life than it is given credit for. Although it is now a fully de-
scribed subject, with limited room for further development, the applications
of spherical trigonometry are truly extensive.
Bibliography

Screen shot from google maps. https://www.google.co.uk/maps/@27.

175966,52.6476179,4z, December 2014. Accessed December 2014.

James Arvo. Stratified sampling of spherical triangles. In SIGGRAPH ’95

Proceedings of the 22nd Annual Conference on Computer Graphics and
Interactive Techniques, pages 437–438, New York, New York, 1995. ACM.

Tony Barrera, Anders Hast, and Ewert Bengtsson. Incremental spheri-

cal linear interpolation. In Stefan Seipel, editor, The Annual SIGRAD
Conference; Special Theme—Environmental Visualisation, pages 7–10,
Linköping, Sweden, November 2004. Linköping University Electronic
Press.

Camille Flammarion. L’atmosphère: météorologie populaire. Hachette, Paris,

1888.

Thomas L Heath. Greek astronomy. Courier Corporation, 1932.

David A. King and Richard P. Lorch. Qibla charts, qibla maps, and related
intruments. In J. B. Harley and D. Woodward, editors, The History of
Cartography: Cartography in the Traditional Islamic and South Asian So-
cieties, volume 2, chapter 9, pages 189–205. University of Chicago Press,
1992.

Jack B. Kuipers. Quaternions and Rotation Sequences: A Primer with Ap-

plications to Orbits, Aerospace, and Virtual Reality. Princeton University
Press, Princeton and Oxford, 1999.

Jack B. Kuipers. Quaternions and rotation sequences. In Ivaı̈lo M. Mladenov

and Gregory L. Naber, editors, Geometry, Integrability and Quantization,
September 1–10, 1999, Varna, Bulgaria, pages 127–143. Coral Press, Sofia,
2000.

44
BIBLIOGRAPHY 45

John J. O’Connor and Edmund F. Robertson. The mactutor history of math-

ematics archive. http://www-history.mcs.st-andrews.ac.uk/index.
html. Accessed April 2015.

David Pingree. The recovery of early greek astronomy from india. Journal
for the History of Astronomy, 7:109–123, 1976.

Dirk J. Struik. A Concise History of Mathematics. Dover Publications, Inc.,

New York, 3rd edition, 1967.

I. Todhunter and J. G. Leathem. Spherical Trigonometry. Macmillan and

Co., Limited, London, 11th edition, 1949.

Claus Tøndering. Announcing the sky. http://www.webexhibits.org/

calendars/. Accessed March 2015.

Glen Van Brummelen. Heavenly Mathematics: The Forgotten Art of Spheri-

cal Trigonometry. Princeton University Press, Princeton and Oxford, 2013.