Dennis Duke

Six Easy Lectures on Ancient Mathematical Astronomy

Course website: www.scs.fsu.edu/~dduke/lectures

Department of Physics and School of Computational Science

Florida State University

Lecture 1
Where, When and Who
Almagest Books 1 and 2
the celestial sphere
numbers and angles (sexagesimal base-60)
obliquity and latitude and the related instruments
plane geometry and trigonometry, the chord tables
spherical trigonometry, circles on the celestial sphere

the Where: Ptolemys World A.D. 150

The world as our story begins. The East

and the West

the Greeks near their peak

Alexander the Greats empire

Strabos Geography (1st-2nd century B.C.)

Ptolemys World Map (1st century A.D.)

Most of what we have from antiquity was preserved and

transmitted to us by the Islamic societies of the 8th 13th
centuries A.D.

Who and When: Ancient Astronomers

Homer/Hesiod
Meton/Euctomen
Eudoxus
Aristotle
Heraclides
Callippus
Autolycus/Euclid
Aristyllus
Berosus

-750
-430
-380
-340
-330
-330
-330
-300
-300

Aratus
-270
Timocharis -260
Aristarchus -240
Archimedes -220
Eratosthenes -210
Apollonius -200
Hipparchus -130
Posidonius -100
Geminus
-50

Theon of Smyrna
120
Ptolemy Almagest
150
Theon of Alexandria 350

Relevant Famous People

Plato
Alexander the Great
Strabo
Pliny
Plutarch
Marinus of Tyre

-375
-330
10
70
100

philosopher
conquered Babylon
Geography
Natural History
Concerning Nature
The Face in the Moon
120 geography (Ptolemys source)

Later Famous Astronomers

(and Ptolemy influenced every one of them)
Hipparchus
Ptolemy Almagest

-130
150

Aryabhata (India)

500

al-Sufi (Islam)
al-Tusi/Urdi/Shatir
Ulugh Beg

950
1250
1420

Copernicus
Tycho Brahe
Kepler

1540
1570
1620

Newton

1680

Almagest, Book I begins:

and a bit later:

similarly (see the excerpts on the supplementary reading page):

Theon of Smyrna (about A.D. 120)
Strabo Geography (about A.D. 5)
Geminus (about 50 B.C.)
Hipparchus (about 130 B.C.)
Autolycus (about 300 B.C.), and Euclids Phenomena is similar
Eudoxus (about 320 B.C.)
Aristotle (about 350 B.C.)
Hesiod (about 750 B.C.)
Homer (about 780 B.C.)
It is fair to say that Ptolemy makes the best effort to give fairly cogent
arguments, usually astronomical, to support all of these assumptions.

For example:

Ptolemy is probably summarizing the winning arguments in an old debate, going

back as far as Aristarchus in about 240 B.C.:

Celestial Sphere

The oblique circle (the ecliptic, the path of the Sun, Moon and
planets)

Using a gnomon

= geographical latitude
= twice the obliquity of the ecliptic

Gnomons are also the basis of sundials:

How were these angles measured other than using a gnomon?

Ptolemy describes two instruments:

Expressing Numbers
Even today we measure angles in degrees,
minutes, seconds, and we also measure time
in hours, minutes, seconds.
In both cases there are 60 minutes per
degree or hour, and 60 seconds per minute.
Apparently this began in Babylon, no later
than early first millenium B.C. and probably
a lot earlier, since we have many 1000s of
surviving clay tablets covered with such
numbers.

Ptolemy also used this base-60 sexagesimal number format, at least for the
fractional part of the number. Thus he expressed the number 365 + 1 1 as
4 300

365 + 15 12 = 365 + 14 + ( 60 12 )
60 3600
60 3600 3600
= 365 + 14 + 48
60 3600
= 365;14,48

The integer part of the number was given in decimal.

With a good set of multiplication and division tables, which

everyone had, manual arithmetic was no harder for them than it is
for us.

Ptolemy used mostly plane geometry and trigonometry, with a

little spherical trig when he needed it, which was not often.
For plane trig he had only one construct the chord rather than
our sine, cosine, tangent, etc, and this was enough.

He also had good tables of the chord function, and was quite
capable of interpolation, just as we (used to) do it.

Ptolemy says that he will present a simple and efficient way to compute the chords, but he doesnt
actually say the table was computed that way, or even that he computed it. In fact, there is good
reason to think that it was not computed using his methods, or that he was the person who computed
it. Unfortunately, however, we have no evidence about who did compute it.

As we will see in Lectures 2 and 3, it is likely that Hipparchus also had a good command of
trigonometry, both plane and spherical, but he also probably had a simpler trig table. Most people
assume he also used the chord construct, but there is no evidence for this, and there is some reason to
think he used instead the sine.

360D 60/ D 21, 600

R=
=
 3438
2
2

D=

21600

 6875

Angle(degrees)
0
7
15
22
30
37
45
52
60
67
75
82
90
97
105
112
120
127
135
142
150
157
165
172
180

Chord
0
450
897
1341
1780
2210
2631
3041
3438
3820
4186
4533
4862
5169
5455
5717
5954
6166
6352
6511
6641
6743
6817
6861
6875

There is also no reason to think that Hipparchus invented trignonometry and tables, either chord or
sine. In fact, a work of Archimedes shows the explicit computation of about 2/3s of the entries in
Hipparchus (supposed) table, and computing the other entries would be straightforward.

Archimedes gets
which leads to

66 < sin1 7 D < 153 (equivalent to 0.03272 < sin1 7 D < 0.03274 )
8
8
2017 14
4673 12
310 < < 3 1
71

circumscribed

inscribed

Angle
a
c
a
3 6/8 153 2339 3/8 780
7 4/8 153 1172 1/8 780
11 2/8 169
866 2/8 70
15
153
591 1/8 780
18 6/8 571 1776 2/8 2911
22 4/8 169
441 5/8 70
26 2/8 744 1682 3/83793 6/8
30
153
306
780
33 6/8 408
734 3/8 169
37 4/8 571
937 7/8 2911
41 2/8 1162 1/81762 3/85924 6/8
45
169
239
70

circumscribedinscribed
Base
c
Base 3438
3438
11926
225
225
5975 7/8
449
449
358 7/8
671
670
3013 6/8
890
890
9056 1/8
1105
1105
182 7/8
1315
1316
8577 3/8
1520
1520
1560
1719
1719
304 2/8
1910
1909
4781 7/8
2093
2093
8985 6/8
2267
2267
99
2431
2431

columns 25 come from Archimedes, while columns 67 are just

a 3,438
c
notice that Archimedes is working entirely in sine and cosine, never chord
there is no doubt that Hipparchus was familiar with Archimedes work on this
about all we can conclude is that Archimedes, Hipparchus, or someone in between
might have computed the first trig table this way

We can, in fact, go even farther back into the very early history of trigonometry by considering
Aristarchus On Sizes and Distances, and we shall see that a plausible case can be made that his
paper could easily have been the inspiration for Archimedes paper. The problem Aristarchus posed
was to find the ratio of the distance of the Earth to the Moon to the distance of the Earth to the Sun
[as we will see in Lecture 4]. He solved this problem by assuming that when that the Moon is at
quadrature, meaning it appears half-illuminated from Earth and so the angle Sun-Moon-Earth is 90,
the Sun-Moon elongation is 87, and so the Earth-Moon elongation as seen from the Sun would be
3. Thus his problem is solved if he can estimate the ratio of opposite side to hypotenuse for a right
triangle with an angle of 3, or simply what we call sin 3.
Aristarchus proceeded to solve this problem is a way that is very similar to, but not as systematic as,
the method used by Archimedes. By considering circumscribed (Fig. 2 below) and inscribed
triangles (Fig 3 below) and assuming a bound on 2 Aristarchus effectively establishes bounds on
sin 3 as
1 < sin3D < 1
20
18

and, although he does not mention it, this also establishes bounds on as

3 < < 313

clearly not as good as Aristarchus got just a few years later.

Actually, the sine (not chord!) table that we suppose was used by Hipparchus
shows up clearly in Indian astronomical texts of the 5th and 6th centuries A.D.
For example, Aryabhata writes in The Aryabhatiya (ca. A.D. 500) verse I.10:
10. The sines reckoned in minutes of arc are 225, 224,
222, 219, 215, 210, 205, 199, 191, 183, 174, 164, 154,
143, 131, 119, 106, 93, 79, 65, 51, 37, 22,7.

and later he explains how to compute these in verse II.12:

12. By what number the second sine is less than the first
sine, and by the quotient obtained by dividing the sum of
the preceding sines by the first sine, by the sum of these
quantities the following sines are less than the first sine.

These are clearly not sines but rather the differences of adjacent terms in the
table of sines. The base is 3,438, just as Hipparchus used.

Many similar examples (to be seen in coming weeks) lead to what I call the
Neugebauer Pingree van der Waerden Hypothesis:
The texts of ancient Indian astronomy give us a sort of wormhole through
space-time back into an otherwise inaccessible era of Greco-Roman
developments in astronomy.

Thus the essentially universally accepted view that the astronomy we find in
the Indian texts is pre-Ptolemaic. Summarizing the prevailing opinion,
Neugebauer wrote in 1956:
Ptolemys modification of the lunar theory is of importance for the problem
of transmission of Greek astronomy to India. The essentially Greek origin of
the Surya-Siddhanta and related works cannot be doubted terminology, use
of units and computational methods, epicyclic models as well as local
tradition all indicate Greek origin. But it was realized at an early date in the
investigation of Hindu astronomy that the Indian theories show no influence
of the Ptolemaic refinements of the lunar theory. This is confirmed by the
planetary theory, which also lacks a characteristic Ptolemaic
construction, namely, the punctum aequans, to use a medieval
terminology.
This fundamental idea will be explored much further in coming lectures.

Ptolemys obliquity and latitude of Alexandria

Ptolemy uses 2 = 47D;42,30 but in reality he should have gotten about 47;21.
Now 21 is a fairly large error for this kind of measurement, about 2/3rd the size
of the Moon. What is not surprising is that Ptolemy made such an error, but that
he got exactly the same values used by Eratosthenes and Hipparchus, who
should have gotten about 47;27.
This kind of thing occurs frequently throughout the Almagest.

For the geographical latitude, Ptolemy writes:

and later in Almagest 5.12:

Actually, the latitude of Alexandria is between 31;13and 31;19,depending on

exactly where Ptolemy worked (probably closer to the more northern limit).
Ptolemys value 30;58follows exactly from an equinoctial shadow ratio of 5/3,
and was probably also a value he inherited from some old tradition.

Spherical trigonometry solves problems related to circles on a sphere.

A particular problem is to compute the angles between the ecliptic, the equator,
and the horizon. Another is to compute the time required for a given segment of
the ecliptic to rise or set above or below the horizon. Another is to compute the
length of the longest (or shortest) day at any given geographic latitude.

from Almagest Book 2.6, for the parallel of the Tropic of Cancer:

and some parallels further north:

so Ptolemy is systematically computing what the shadow lengths will be at a

sequence of geographical longitudes from the equator to the arctic circle.

This had been going

on for centuries. In
about 200 B.C.
Eratosthenes had
managed to determine
the circumference of
the Earth.
Strabo, writing about
A.D. 5, gives and
interesting account of
the work of both
Eratosthenes and
Hipparchus in this
area (see the
supplementary
reading).

Eratosthenes is said to have measured the angle as 7 1/5 degrees, and took the
distance from Syene to Alexandria as 5,000 stades, giving
CEarth = 360
5,000 = 505,000 = 250,000 stades
7 15

which he rounded to 252,000 stades to make it divisible by 60 (and also 360).

Ancient Astronomy
Lecture 2
Course website: www.scs.fsu.edu/~dduke/lectures

Lecture 2
Almagest Book 3
the length of the year
the length of the seasons
the geometric models
the length of the day
the background
lost episodes in solar history

 Book 3 of the Almagest is about the Sun.

The Sun is first in Ptolemys logical structure,

followed by the Moon, then the fixed stars, and finally

the wandering stars (planets).
Ptolemy says he is following Hipparchus theory of the

Sun (a claim confirmed by Theon of Smyrna).

Probably nothing in Book 3 is original with Ptolemy,

apart from the four equinox and solstice observations

in Almagest 3.1 and 3.7.

What are the major questions to be answered?

1. What is a year?
2. Is the length of the year constant?
3. What is the length of the year?
4. What are the lengths of the seasons?
5. How does the speed of the Sun vary throughout the year?
6. What kind of geometrical model would account for the phenomena
(observations)?
7. How does the length of the day vary?

1. What

is a year?

There are several choices:

(a) return to the same star on the ecliptic (sidereal year).
(b) return to the same declination (e.g. the same place on the equator).
(c) return to the same speed (anomalistic).
(d) return to the same latitude (distance from the ecliptic).
Ptolemy, and probably Hipparchus before him, chose option (b), usually called
the tropical year, since you could define it as the time it takes for the Sun to
return to a tropic circle, i.e. a solstice (summer or winter). Ptolemy actually
measures relative to the vernal (spring) equinox.

What are the major questions to be answered?

1. What is a year?
2. Is the length of the year constant?
3. What is the length of the year?
4. What are the lengths of the seasons?
5. How does the speed of the Sun vary throughout the year?
6. What kind of geometrical model would account for the phenomena
(observations)?
7. How does the length of the day vary?

23. Is the length of the year constant, and how long?

Ptolemy says that Hipparchus measured the number of days between
successive equinoxes, first autumn:

then spring:

so Hipparchus finds that with a few exceptions the year is 365 days.
Further, the exceptions could easily be observation uncertainties, so Ptolemy
finds no reason to doubt that the year length is constant.

Getting a precise year length.

1 days (probably
Ptolemy says that Hipparchus found a year length of 365+ 14 300
from the interval between the summer solstices in 280 B.C. and 135 B.C.)
Ptolemy then says

Ptolemy gives the dates for the autumn equinoxes of 132 and 139
and the spring equinox and summer solstice of 140, all most
carefully observed, and compares them to the fall equinox of
-146, the spring equinox of -145, and the summer solstice of -431:
(1)
(2)
(3)
(4)

-146/9/27 midnight to 132/9/25 2 pm

-146/9/27 midnight to 139/9/26 7 am
-145/3/24 6 am
to 140/3/22 1 pm
-431/6/27 6 am
to 140/6/25 2 am

in each case you count the number of intervening days, divide by

1 days.
the number of years, and the year length is 365+ 14 300
1 , about 6 minutes shorter.
The correct value is about 365+ 14 133

What are the major questions to be answered?

1. What is a year?
2. Is the length of the year constant?
3. What is the length of the year?
4. What are the lengths of the seasons?
5. How does the speed of the Sun vary throughout the year?
6. What kind of geometrical model would account for the phenomena
(observations)?
7. How does the length of the day vary?

45. the length of the seasons and the varying speed of the
Sun
Early History of Season Lengths
based on a 365 day calendar (except for Hipparchus)
Democritus (460 B.C.)
Euctomen (432 B.C.)
Eudoxus (380 B.C.)
Callippus (340 B.C.)
Geminus (200 B.C.)
Hipparchus (130 B.C.)
accurate (134 B.C.)

Summer
91
90
91
92
92
92
92

Autumn
91
90
92
89
89
88
88

Winter
91
92
91
90
89
90
90

Spring
92
93
91
94
95
94
94

other than Hipparchus, it is not at all certain that any of these were based on
observation of equinoxes or solstices.

Ptolemy says that Hipparchus assumed season lengths 94 days for Spring and
92 days for Summer, but he does not say how Hipparchus got these values.
This tells us that the Sun
does not appear to move
around the ecliptic at a
uniform speed.

What are the major questions to be answered?

1. What is a year?
2. Is the length of the year constant?
3. What is the length of the year?
4. What are the lengths of the seasons?
5. How does the speed of the Sun vary throughout the year?
6. What kind of geometrical model would account for the phenomena
(observations)?
7. How does the length of the day vary?

6. What kind of geometrical model would account for the phenomena

(observations)?

www.scs.fsu.edu/~dduke/models

from the season lengths we know angles TZN and PZK, and using
simple geometry gives EZ = e = 2;30 and angle TEZ = 65;30.
VE

SS

WS

AE

How accurate is the model? Not bad in Hipparchus time.

0.5

135 B.C.

e = 2;30

0.4

0.3

error (degrees)

0.2
e = 2;10
0.1

average
e = 2;30
e = 2;10

average error
0

-0.1

-0.2

-0.3

-0.4
0

50

100

150

200
day of year

250

300

350

much less so in Ptolemys time (large average error).

2

A.D. 137
e = 2;30
1.5
e = 2;10

error (degrees)

average error
1

average
e = 2;30
e = 2;10
average
e = 2;30
e = 2;10

0.5

135 B.C.
0

-0.5
0

50

100

150

200
day of year

250

300

350

Some key terms:

mean motion refers to the average speed of some celestial body. They
knew that the speed could vary around the orbit, but they knew the
average speed was the distance around many orbits divided by the time
for many orbits. Mean motion is regular.
mean position refers to where the body would be if it always traveled
with mean speed. In reality the true position of the body would usually
be ahead of or behind the mean position.
motion in anomaly is the regular motion that actually causes the true
motion to differ from the mean motion, so the true motion appears to be
irregular. Thus irregular motions result from a compounding of regular
motions (mean and anomaly).

the equation is an angle that is the difference between the true position
and the mean position of a body. Thus
true = mean + equation

mean position
angle

true position
angle

What are the major questions to be answered?

1. What is a year?
2. Is the length of the year constant?
3. What is the length of the year?
4. What are the lengths of the seasons?
5. How does the speed of the Sun vary throughout the year?
6. What kind of geometrical model would account for the phenomena
(observations)?
7. How does the length of the day vary?

7. How does the length of the day vary?

the length of the day is determined mostly by how fast the Earth rotates about its
axis (or the celestial sphere, to the ancients).
However, because
(b) the Sun is moving on the oblique ecliptic, and
(c) the speed of the Sun varies on the ecliptic,
the actual length of time between successive noons varies slightly.
From day to day the variation is very small, but it does accumulate so that a day
in February can be about 15 minutes shorter than average, while a day in
November can be about 15 minutes longer than average, etc.

Ptolemy understood this very well, but does not tell us how he learned it.

Main features of Hipparchus solar model as reported by

Ptolemy:
there is only one variation: the speed around the ecliptic.
the eccentric and epicycle versions give equivalent explanations.
apogee is the direction of slowest motion on the ecliptic, perigee is the
direction of fastest motion.
the direction of the apogee is always 65 from the vernal equinox and the
eccentricity is always 2;30 (compared to 60). Both are determined from the
season lengths for Spring and Summer (94 and 92 days).
Ptolemy insists the model must predict that the time from slowest to mean
(average) speed is greater than the time from mean to fastest speed, for we
find that this accords with the phenomena [observations].
the Sun is always on the ecliptic, never north or south (no latitude).

The Background
A closer look at Ptolemys most carefully observed equinox and solstice dates.
(1) -146/9/27 midnight to 132/9/25 2 pm
(2) -146/9/27 midnight to 139/9/26 7 am
(3) -145/3/24 6 am
to 140/3/22 1 pm
(4) -431/6/27 6 am
to 140/6/25 2 am
suppose we compute the expected date of each event by multiplying the number
1 . We get
of intervening years by the assumed days per year, 365+ 14 300
(1)
(2)
(3)
(4)

132/9/25 1:46 pm
139/9/26 7:12 am
140/3/22 1:12 pm
140/6/25 2:19 pm

correct
(9/24 4 am)
(9/24 9 pm)
(3/21 4 pm)
(6/23 1 am)

Was it just too hard for Ptolemy to get right?

-161/9/27
-158/9/27
-157/9/27
-146/9/27
-145/3/24
-145/9/27
-144/3/23
-143/3/23
-142/3/24
-142/9/26

6 pm
(9/27 2 am)
6 am
(9/26 8 pm)
noon
(9/27 2 am)
midnight (9/26 6 pm)
6 am
(3/24 3 pm)
6 am
(9/26 11pm)
noon
(3/23 9 pm)
6 pm
(3/24 2 am)
midnight (3/24 8 am)
6 pm
(9/26 5 pm)

-141/3/24 6 am
(3/24 2 pm)
-140/3/23 noon
(3/23 8 pm)
-134/3/24 midnight (3/24 7 am)
-133/3/24 6 am
(3/24 1 pm)
-132/3/23 noon
(3/23 6 pm)
-131/3/23 6 pm
(3/24 midnight
-130/3/24 midnight (3/24 6 am)
-129/3/24 6 am
(3/24 noon)
-128/3/23 noon
(3/23 6 pm)
-127/3/23 6 pm
(3/23 11pm)

so apparently Hipparchus was generally accurate to the nearest day.

Clearly Ptolemy computed those four dates. Why did he do that? There is no
objective evidence to help us, so we can only speculate. What we can say with
some certainty is that this was the rule, not the exception, for Ptolemy.

Ptolemy tells us that his solar model is the same as Hipparchus but gives us no
other background information. In Book 12 he does mention that Apollonius of
Perge (ca. 200 B.C.) had proved a rather complicated theorem involving the
epicycle model, so it seems likely that epicycles and eccentrics, and their
equivalence, had been studied for several centuries.
Also, as we shall see for the Moon and the planets, those models require both
moving apogee directions and (effectively) oscillating eccentricities. One might
think that for uniformity and unity Ptolemy would make the solar model more
like the models for the Moon and the planets, but he does not.

What Ptolemy does not tell us is that there was a lot of other activity developing
solar models both near his time and going back centuries.
Here are some examples:
(1) There are clear indications in the Almagest that Hipparchus himself used
solar models different from the standard one attributed to him by Theon and
Ptolemy. For example, the two pairs of eclipse longitude differences that
Hipparchus uses to find the unusual lunar eccentricities in Almagest 4.11 may
also be used to deduce the underlying solar models, and the resulting parameters
are equally unusual: e=7;48 and A=76;25 for Trio A, and e=3;11 and A=46;09
for Trio B. Although attempts have been made to understand the underlying
models, the analyses are neither conclusive nor satisfying. The solar parameters
are so bizarre that we might be tempted to speculate that Hipparchus is
somehow trying to use a lunar theory to learn something about the time variation
of solar theory (the trios date to about 380 and 200), and so it is perhaps
interesting that in both trio analyses the eclipses all occur near equinoxes and
solstices [more about this case in Lecture 3].

(2) Almagest 5.3 and 5.5 give three timed solar longitudes due to Hipparchus,
and these imply a solar model with parameters e = 2;16 and A = 69;05,
although it might be that the underlying model is actually based on season
lengths of 94 days and 92 days, for which the parameters are instead e = 2;19
and A = 67;08. Either way, the value of e is significantly improved over the
standard Hipparchan value 2;30.
(3) Theon of Smyrna mentions, quite matter-of-factly, a solar model with
periods of 365 days in longitude, 365 days in anomaly, and 365 18 days in
latitude. He also mentions that the Sun strays from the ecliptic by . Solar
latitude was mentioned as early as Eudoxus, and must have had some level of
use, since not only Theon but also Pliny mentions it, and Hipparchus felt
compelled to deny its existence. Ptolemy never mentions solar latitude.

(4) P. Oxy LXI.4163 is a fragment of a papyrus table from Oxyrhynchus

that gives a template for daily longitudes of the Sun to degrees and
minutes starting from the day of summer solstice. All indications are that
it is not based on the usual Hipparchan parameters.
(5) P. Oxy LIX.4162 is similar to P. Oxy LXI.4163 but appears to count
days starting when the Sun is at perigee and puts the cardinal points at 8
of the signs. In this case the indications are strong that the underlying
theory is kinematical, but even if it is, it seems not likely to be based on
the usual Hipparchan parameters.
(6) P. Oxy. LXI.4148 is a table of dates of summer solstices over a series
of years. The dates are in error by about five days in the years covered in
the fragment and are based on a year of length 365;15,22,46 days. There
are indications that the dates might have begun from a known
Hipparchan summer solstice measurement of 127 June 26 at sunrise.
For more information on the astronomical papyri of Oxyrhynchus see
http://www.chass.utoronto.ca/~ajones/oxy/

We know from the Almagest that Hipparchus knew the times of

equinoxes and summer solstices to an average accuracy of about day.
Since no ancient source explains how these times were determined, we
need to consider just how an ancient astronomer would measure the time
of an equinox or a solstice to that level of accuracy.
By definition,
an equinox occurs at the moment the Sun touches the equator, so its
declination = 0.
and a solstice occurs when the Sun touches either tropic circle
(Cancer to the north, Capricorn to the south), so its declination
= = 23;43D

It is clear from practical considerations that no one could have reliably

and routinely simply noted the moment when the Suns declination was
at a given value: 0 for an equinox or 23;43 for a solstice.
On the one hand, about half of the events will occur at night, when
the Sun is not visible.
On the other hand, even if the event happens in daylight, it is not
always the case that the Sun will be unobscured by clouds and in a
position in the sky favorable for measuring the declination
accurately.
In addition, for the solstices it is impossible to achieve day accuracy
with naked eye observations of any kind within a day or so of the event
since the declination of the Sun is changing extremely slowly near a
solstice.

It is most likely, then, that equinoxes and solstices were determined by

observing noon solar altitudes for a series of days before and after the events.
When the Sun is crossing the meridian at noon, it is relatively easy to measure
its altitude, and then knowing the geographical latitude, to compute the
declination. From the declination, it is easy to compute the Suns position on the
ecliptic (the longitude), and we know that Hipparchus knew how to do it.
But it is only at noon that such an easy determination is possible. It is then fairly
straightforward to estimate the time that the Suns declination reaches some
specific targeted value: 0 for an equinox, and maximum or minimum for a
solstice.
That series of daily altitude measurements were used to determine the time of
cardinal events can hardly be doubted, even though no surviving ancient source
has documented such an episode. Especially for the solstices, it is essentially the
only viable option for achieving day accuracy.

In fact, however, you dont really need equinoxes or solstices. Any trio of timed
longitudes would be adequate. Ptolemy provides two such analyses for the
Moon and one each for Mars, Jupiter and Saturn. Hipparchus, and perhaps his
predecessors and certainly his successors, knew the method, so it seems
inconceivable that it was not used multiple times to also determine solar model
parameters.
M1
M3

P
B

M2

Finally, Ptolemy insists the model must predict that the time from slowest to
mean (average) speed is greater than the time from mean to fastest speed, for
we find that this accords with the phenomena [observations].
First, the time differences Ptolemy refers to are undetectable using naked eye
observations, so this is something he is pushing not from empirical
observations, but from some unstated theoretical (or philosophical) bias.
Second, there is a perfectly good model for the solar motion that violates
Ptolemys rule: the concentric equant. Using the concentric equant one finds
that the time from least speed to mean speed is equal to the time from mean
speed to greatest speed.

In the concentric equant model the Earth is at the center E of the deferent, but
the center of uniform motion Z of the Sun S is displaced some distance e from
the center. Even though the Sun is now always at the same distance R = ES from
the Earth, the model still produces an apparent speed variation in the motion of
the Sun such that in one direction (the direction EZ) the Sun seems to be moving
slowest, and in the opposite direction it seems to be moving fastest.
The concentric equant model for the Sun is repeatedly attested in Indian texts,
all of which are generally supposed to be of Greco-Roman origin, and the
accurate value e = 2;10 is routinely used.

www.scs.fsu.edu/~dduke/models

For the equation of time, remember that it has two causes:

(a) the Sun is moving on the oblique ecliptic, and
(b) the speed of the Sun varies on the ecliptic,
so the actual length of time between successive noons varies slightly.
It turns out that Geminus, writing in about 50 B.C., mentions the equation of
time but for him only (a) is involved.
In the ancient Indian texts, on the other hand, the equation of time is attributed
only to (b). These texts are supposed to originate from Greco-Roman sources
from the time period between about 100 B.C. and A.D. 100, or post-Hipparchus
and pre-Ptolemy.

Early History of Season Lengths

based on a 365 day calendar (except for Hipparchus)
Democritus (460 B.C.)
Euctomen (432 B.C.)
Eudoxus (380 B.C.)
Callippus (340 B.C.)
Geminus (200 B.C.)
Hipparchus (130 B.C.)
accurate (134 B.C.)

Summer
91
90
91
92
92
92
92

Autumn
91
90
92
89
89
88
88

Winter
91
92
91
90
89
90
90

Spring
92
93
91
94
95
94
94

other than Hipparchus, it is not at all certain that any of these were based on
observation of equinoxes or solstices.

Ancient Astronomy
Lecture 3
February 14, 2007
Course website: www.scs.fsu.edu/~dduke/lectures

Lecture 3
Almagest Books 4 6
the Moon
the problem of parallax
the length of the various months
the first geometric model
the second geometric model
sizes and distances of the Sun and Moon
the background

the Moon and the Sun are both about the same size as viewed from Earth: they
both subtend about in the sky.
the distance to the Moon is not
negligible compared to the size of
the Earth.

best observations are times of lunar eclipses: at that time we can compute the
position of the Sun, and we then know that the Moon is exactly 180 away.

by the way, a solar eclipse is similar but a bit more complicated.

http://sunearth.gsfc.nasa.gov/eclipse/SEanimate/SE2001/SE2017Aug21T.GIF

drawn to scale:

What is a month?
There are several:
(a) return to the same star on the ecliptic (sidereal).

27d 07h 43m 12s

(b) return to the same declination (tropical).

27d 07h 43m 05s

(c) return to the same speed (anomalistic).

27d 13h 18m 33s

(d) return to the same latitude (draconitic).

27d 05h 05m 36s

(e) return to the same angle from the Sun (synodic).

29d 12h 44m 03s

The synodic month from one new moon or full moon to the next is the one
we use in daily conversation.

Sidereal Month (return to same longitude or fixed star)

Tropical Month (return to the same equinox or solstice)

Anomalistic Month (return to same speed, e.g. fastest or slowest)

Draconitic Month (return to the nodes)

Synodic Month (return to the Sun)

to get an eclipse we must have the Sun-Earth-lunar nodes lines up, and the
Moon fairly near a node (with about 7). On average we get about two eclipses
per year, somewhere.

Period Relations
Periodic (Saros)
6585d = 223m = 239a = 242d = 241t + 10 (about 18y)
Exeligmos (3x Saros)
19,756d = 669m = 717a = 726d = 723t + 32 (about 54y)
Hipparchus (Babylonian)
126,007d 1h = 4267m = 4573a = 726d = 4612t 7 (about 345y)
and
5458m = 5923d
Note that 126007d 1h / 4267m = 29d 12h 44m 02s (compared to 29d 12h 44m 03s)
All of these come from centuries of eclipse records in Babylon, starting around
750 B.C. if not earlier (remember that Alexander the Great conquered Babylon
in 323 B.C.)

Ptolemy and Hipparchus found that regarding just new moon and full moon,
when the Sun and Moon are in a line with the Earth, a simple model would
work.
http://www.scs.fsu.edu/~dduke/models.htm

However, in the more general case the simple model fails and Ptolemy uses a
more complicated model. http://www.scs.fsu.edu/~dduke/models.htm

Sizes and Distances of the Sun and Moon

Ptolemy gives an analysis which is extremely delicate to compute.

Ptolemy takes
= 0;31,20
= 2 3/5 = 1;21,20
L = 64;10
but for example / = 2 2/5 makes S < 0

to get S = 1210

The Background
Ptolemys usual fudging
Luni-Solar calendars
Babylonian models

Ptolemys fudging
for the simple model he produces two trios of lunar eclipses:
-720 Mar 19/20 7:30 pm
-719 Mar 8/9 11:10 pm
-719 Sep 1/2 8:30 pm

133 May 6/7 11:15 pm

134 Oct 20/21 11:00 pm
136 Mar 5/6 4:00 am

Analysis of both trios gives virtually identical results, and

changing any of the times by even a few minutes substantially
changes the results.
Later, in Almagest 4.11, he gives two more trios and once again
gets the very same answers. Such coincidences are very unlikely.

for the complicated model

(a) Ptolemy wants to know the maximum angle the true Moon can
differ from the average Moon. In the case of the simple model this
is 5. Ptolemy produces two observations which he analyzes to get
a maximum angle of 7;40 (in both cases). But he neglected
parallax, and if he had included it he would have gotten 7;31 and
7;49 for the two cases.
(b) Ptolemy needs to know the size of his new central epicycle, so he
produces two observation that both give him 10;19. In both cases
he miscomputes but still manages to get the same answer.
(c) Ptolemys complicated model makes the apparent size of the Moon
vary by almost a factor two. In reality it varies by about 15%
(maximum to minimum).

for the sizes and distances Ptolemy has to very carefully analyze
eclipses from 523 B.C. and 621 B.C. (why so ancient?). In the end he
finds
S 19
L
However, in about 240 B.C. Aristarchus, in a completely different kind
of analysis, also found S / L 19 . He assumed only that the angle MoonEarth-Sun was 87 at half-moon.

In between Ptolemy and Aristarchus, Hipparchus used slightly

different parameters to get a much different answer:
Ptolemy takes
= 0;15,40
= 2 3/5 = 0;40,40
L = 64;10

Hipparchus takes
= 0;16,37
= 2 1/2 = 0;41,33
L = 67;20

to get S = 1210

to get S = 490, or he
assumed S = 490 and
computed L = 67;20

the correct answers are about L = 60 and S = 23,000

Luni-Solar Calendars
The fact that the month is just a bit longer than 29 days caused a lot of bother
in establishing a workable calendar that keeps months properly aligned with the
year and its seasons.
Early try: Meton and Euctomen (about 430 B.C.): the Metonic calendar
19 years = 235 months = 6940 days
= 12 years of 12 months plus 7 years of 13 months
There are 125 full (30-day) months and 110 hollow (29-day) months
Resulting year is 365 5/19 days
Resulting month is 29 + 1/2 + 3/94 days

365 5/9 is longer than 365 by 1/76 day. Hence Callippus (about 330 B.C.)
suggested a new calendar with four successive 19-year Metonic cycles but
leaving out 1 day from one of the cycles:
76 years = 940 months = 27,759 days
= 4 x 235 months = 4 x 6,940 days 1 day
Resulting year is 365 1/4 days
Resulting month is 29 + 1/2 + 29/940 days
The fraction 29/940 is about 1/32.4 whereas a slightly more accurate value is
1/33, and this was known to Geminus and hence would have been widely
known.
There may have been even older and simpler calendars. Geminus describes ones
with 8 years = 99 months and 16 years = 198 months and 160 years = 1979
months. In all of these either the month or year length is not good enough.

The Antikythera Mechanism

the pin-and-slot mechanism to simulate Hipparchus model

Babylonian Astronomy
During the late 1800s some 50,000 or so clay tablets were sent to the British
Museum from Babylon and Uruk.

About 250 of the tablets related to astronomy were studied by two Jesuit priests,
Fathers Epping and Strassmaier in the late 1800s and followed by Father
Kugler in the early 1900s.

The work of the three Fathers revealed a previously unsuspected history of very
involved mathematical astronomy developed in Babylon starting about 450 B.C.
Before their work science in Babylon was generally associated with ideas like
magic, mysticism, and astrology. These people were often referred to as the
Chaldeans.
Whereas the Greek models were designed to give the position of the Sun or
Moon at any moment in time, the Babylonians were interested in predicting the
times and position of sequences of quasi-periodic events new moon, full
moon, etc.

The Babylonians used a purely lunar calendar. The lunar month begins on the
evening when the lunar crescent is first visible shortly after sunset.

Such a definition has a number of intrinsic difficulties, and Babylonian lunar

theory was developed to deal with these complications.
How many days are in a lunar month? Each such month is either 29 or 30
days, but we need to know which in advance.
This clearly involves both the varying speed of the Moon and the varying speed
of the Sun. Remember that the Moon covers about 13 per day and the Sun
about 1 per day, but these are averages. So we must account for the departure
from average throughout each month.
There are seasonal changes due to the angle between the ecliptic and the horizon
and also changes due to the varying latitude of the Moon.

setting

western
horizon

Spring

Fall

setting

latitude
5

western
horizon
Spring

setting
latitude
5
western
horizon
Spring

The astronomical diaries were kept for many centuries and are night-by-night
accounts of where the various celestial objects were to be found:

The result was a list of eclipses covering about six centuries, which Hipparchus
apparently had access to.
In addition, the Babylonians kept extensive records of several centuries of
observations of the times between rising/setting of the Moon and the Sun.

The lunar theories

Each tablet is a set of columns of
numbers

Almost all of these changes vary fairly smoothly somewhat like sine and cosine.
The Babylonian astronomers invented schemes for approximating this kind of
variation.

Nothing survives to tell us how these schemes were created. What do survive are
a small number of procedure texts which give the rules the scribes need to
compute each column.

Ancient Astronomy
Lecture 4
Course website: www.scs.fsu.edu/~dduke/lectures

Lecture 4
Almagest Books 78
the stars
precession
the constellations
rising and setting and the calendar
the background

Ptolemy is now ready to discuss the stars. Recall that

first, he measures the Sun w.r.t. the equinoxes and
solstices
then he measures the Moon w.r.t. the Sun
now he will measure the stars w.r.t. the Moon
next he will measure the planets w.r.t. the stars
But first he must deal with a small complication:
the stars move!

The goal is to measure the position of stars on the celestial

sphere. Lets see what is involved.

Throughout each night the stars rise in the east and set in the west.

since the latitude of Alexandria is about 31 (similar to Tallahassee) the

celestial equator is about 59 above the southern horizon.

the declination coordinate is the distance of each star from the celestial
equator. It is easiest to measure when the star crosses the southern
meridian.

stars rotate along circles of constant declination parallel to the celestial

equator

from north pole to south pole run parallel lines of constant right
ascension, always perpendicular to the lines of declination.

so one convenient set of coordinates is (right ascension, declination).

Another set is defined by the ecliptic, which is oblique to the equator:

This is the coordinate

system used by Ptolemy

Ptolemy says he used an armillary sphere to measure the position of a star.

The problem is: these coordinate systems do not stay fixed w.r.t. each other.

The real reason is that the Earth is like a spinning top, hence the name
precession.

However, to the ancient observers precession shows up as the equinoxes and

solstices, and the north celestial pole, moving with respect to the stars. But the
speed is very slow: they estimated at least 1 per century (in fact, about 1 in 72
yrs).

Ptolemy tells us that Hipparchus was quite careful about this, e.g. he wondered
if just the stars near the ecliptic were involved, or all the stars? To test this he
left a list of star alignments good in his time, and invited future observers to
check them. Ptolemy did, and added some new ones:

To determine the speed of precession several kinds of observations

were used. First, eclipses when the Moon was near a star:

The changes in
declination for
stars near an
equinox:

Lunar occultations of stars a few hundred years apart in time:

Ptolemy recognizes 48 constellations: 21 north of the zodiac, 12 in

the zodiac, and 15 south of the zodiac. He gives the ecliptic
longitude and latitude of 1,028 stars (including 3 duplicates shared
by two constellations).
1
2
3
4
5
6
7
8

Ursa Minor (Little Dipper)

Description
The star on the end of the tai.
The one next to it on the tail
The one next to that, before the place where the tail joins
[the body]
The southernmost of the stars in the advance side of the
rectangle
The northernmost of [those in] the same side
The southern star in the rear.side
The northern one in the same side
The star lying on a straight line with the stars in the rear side
[of the rectangle] and south of them

longitude
d m
UMi 1 1
60 10
UMi 1 2
62 30
UMi 1 3

70 10

+
+

latitude mag
d m
V
66 0
3
70 0
4

name
1Alp UMi
23Del UMi

74 20

22Eps UMi

4 16Zet UMi
4 21Eta UMi
2
7Bet UMi
2 13Gam UMi

UMi 1 4
UMi 1 5
UMi 1 6
UMi 1 7

89
93
107
116

40
40
30
10

+
+
+
+

75
77
72
74

40
40
50
50

UMi 1 8 i

103

71 10

UMi

For most people in antiquity the real interest in the stars resulted from their
relation to the annual calendar. Ptolemy and many people before him published
something like this (e.g. 14 hrs is the longest day at latitude 31):
Epiphi (the 11th month).
1. Summer solstice. 13 hours: the middle one of the belt of Orion rises.
14 hours: the one on the trailing shoulder of Orion rises.
2. 15 hours: the bright one of Perseus rises in the evening.
5. 14 hours: the one common to Eridanus and the foot of Orion rises. 15
hours: the one on the leading shoulder of Orion rises.
6. 13 hours: the one on the head of the leading twin [of Gemini] rises.
14 hours: the middle one of the belt of Orion rises, and the last one of
Eridanus rises, and the one on
the head of the leading twin [of Gemini] rises.
7. 14 hours: the bright one of Corona Borealis sets in the morning.
8. 15 hours: the one on the head of the leading twin [of Gemini] rises. 15
hours: the one common to Pegasus and Andromeda rises in the
evening.
9. 15 hours: the one on the head of the leading twin [of Gemini] rises.

latitude 31
14 hrs
longest day

latitude 41
15 hrs
longest day

a few relics survive:

The Background
Constellations were known to the Babylonians. Many but not all
are related to the Greek versions.
The constellations were well-organized in Greece no later than 380
B.C. and probably considerably earlier. We know this from the
famous poem of Aratus written about 270 B.C. that was derived
from two works by Eudoxus.

In about 130 B.C. Hipparchus wrote a Commentary to Aratus and

Eudoxus that, for the most part,
(a) severely criticized most earlier astronomers for not being
accurate enough, and
(b) gave Hipparchus own version of the rising and setting of
constellations that established a new level of accuracy and
precision.

For example, Hipparchus wrote:

1.1.5 Since I observe that even on the most important points Aratus conflicts both with
phenomena and with the things that really happen, but that in practically all details not only
other commentators but even Attalus agree with him, I thought it good for the sake of your
learning and the common benefit of others to make an accounting of the things that seem to
me to be erroneous. I undertook to do this not because I chose to enhance my image by
refuting others. That is hollow and altogether mean; indeed, I think, on the contrary, that we
must give gratitude to all who engage in taking upon themselves rigorous tasks for the
common benefit. However, I undertook this so that neither you nor others who seek wisdom
might stray from scientific knowledge concerning phenomena in the universe. Many have
suffered this; and it is easy to understand why. For the charm of poems acquires for their
statements a certain reliability, and almost everybody who has commented upon this poet
submits to his statements.
1.1.8 Eudoxus treated the same material concerning phenomena as did Aratus, but with greater
understanding. Naturally, then, the poetry is also regarded as trustworthy because so many
great mathematical astronomers concur. And yet, it is not appropriate that one assail Aratus,
even if he happens to err in certain points. For he wrote the Phaenomena closely following
Eudoxus material, but without observing for himself and without promising to report the
opinion of mathematical astronomers in matters concerning the heavens; this is where Aratus
makes mistakes in his Phaenomena.

Hipparchus own version of the rising of the Crab:

3.3.1b
When the Crab is rising, together with it rises the zodiac from 23 of the Twins
until 18 of the Crab. On the meridian is the portion from 5 of the Fishes until 1m of
the Ram. And the first star to rise is the one in the tip of the northern Claw; the last is that
in the tip of the southern Claw.
Of others on the meridian, the first is the bright star in Andromedas head; the last is the
leading star of the three in the Rams head, and the bright, unassigned star lying toward
the south along the middle of the Sea-monsters body, and the southern of the following
stars in the quadrilateral of the Sea-monster, and Andromedas left foot which is a little
short of the meridian.
The Crab rises in 1 hours.

In fact this is pretty accurate in 130 B.C.:

It turns out that using all the data that Hipparchus gives we can
(a) conclude that he had an extensive catalog of star coordinates, and
(b) figure out the errors on many of his star positions.
The correlation of the Commentary and Almagest errors should be small
if the catalogs are independent, but large if the catalogs share a common
heritage.
It is clear that Ptolemy copied his star coordinates from Hipparchus.
10

805
892

805

Almagest Errors (degrees)

892
918
918

995

992

-5
996

918
918

996

-1 0

995

-1 5
-1 5

-1 0

-5

C o m m e n ta ry E rro rs (d e g re e s )

10

What Ptolemy almost certainly did was take Hipparchus

coordinates, probably in equatorial right ascension and declination,
convert them to ecliptical longitude and latitude, and then add 2
to the longitudes to account for 265 years of precession at 1 per
100 years.

This in spite of his explicit claim:

Hipparchus does have some strange things, though:

1.4.1 Eudoxus is in ignorance concerning the North Pole, when he says
this:
There is a certain star which remains ever in the same spot; and this
star is the Pole of the world.
Upon the pole lies not even one star; rather it is an empty place beside
which lie three stars. With these the point on the Pole forms nearly a
square, according to Pytheas of Massilia.

It is difficult to know what Eudoxus or Hipparchus is referring to,

though:

There were probably a number of other star catalogs around. Star

globes seem to have been popular, and two very nice originals
have survived: the Farnese Atlas and the Mainz Globe.

Ancient Astronomy
Lectures 5-6
Course website: www.scs.fsu.edu/~dduke/lectures

Lectures 5-6
Almagest Books 913
geocentric vs. heliocentric point of view
the wandering stars, or planets
the two anomalies
the eccentric plus epicycle and its problems
the equant
latitude
distances
the background

In reality the Earth and all the other planets revolve around the Sun.
Nevertheless, we can imagine a reference frame in which the Earth is
at rest, and ask what would a correct theory look like in that reference
frame?

Answer: it would look very much like the theory created by the Greek
astronomers.

And note: modern astronomers compute first the planets orbiting the
Sun, and then have to figure out the position of the planet relative to
the Earth.

Problems for heliocentric theory:

Earth in motion??? can't feel it

no parallax seen in stars

geocentric = ego-centric = more

"natural"

Keplers Three Laws of planetary motion:

1. orbits are ellipses, Sun at focus

3.

P2 =1
a3

for each planet

2. equal area in equal time

Instead of the Earth circling the Sun, we would have the Sun
circling the Earth.

http://www.csit.fsu.edu/~dduke/venhelio.html

For the inner planets, not only

does the planet revolve on an
epicycle, but the center of the
epicycle is always lined up with the Sun.

For the outer planets, the radius of the epicycle is always

parallel to the direction of the Sun from the earth.

http://www.csit.fsu.edu/~dduke/juphelio.html

All of the planets have, from time to time, a retrograde

motion, i.e. the slow motion from west to east stops,
then reverses into an easy to west motion, then stops
again and resumes a west to east motion.
http://www.astronomynotes.com/nakedeye/animations/retrograde-anim.htm

In reality this happens because planets closer to the Sun

move faster than planets farther from the Sun.
In a geocentric theory, this happens because of the
counter-clockwise motion on the epicycle.
http://www.astro.utoronto.ca/~zhu/ast210/both.html
http://www.scs.fsu.edu/~dduke/models.htm
And note the close relation to the Sun. In a heliocentric
picture it is clear that in retrograde the SunEarthplanet are
in a line. In a geocentric picture it is not required, but the
Greeks knew that had to assume it to be true.

Counting the number of retrograde episodes and planetary

orbits over many years gives the period relations:
planet
Saturn
Jupiter
Mars
Venus
Mercury

years
59
71
79
8
46

orbits
2
6
42
13
191

retrogrades
57
65
37
5
145

Note that the solar year is somehow involved for every

planet! Such relations are completely ad hoc in a geocentric
view but exactly as expected in a heliocentric view.

In the Almagest Ptolemy says little about the distances to the

planets:

For all the models Ptolemy assumes a deferent circle of radius R =

60 and an epicycle of radius r < 60. Comparing the heliocentric
distances and the Almagest geocentric distances gives
modern
planet
Mercury
Venus
Mars
Jupiter
Saturn

a
0.3871
0.7233
1.5237
5.2028
9.5388

r
23;14
43;24
39;22
11;32
6;17

Almagest
r
22;30
43;10
39;30
11;30
6;30

As far as we know, nobody after Aristarchus (ca. 230 B.C.) and

before Copernicus (A.D. 1540) was willing to make the leap to the
heliocentric picture.

Like the Sun and Moon, the speed of the planets also varies
smoothly as they circle the zodiac, so the planetary orbits each
have an apogee and a perigee.

A new idea, the equant, solves the problem.

The equant is very similar to Keplers ellipse, and accounts

very well for Keplers 1st and 2nd Laws.
www.scs.fsu.edu/~dduke/kepler.html

http://people.scs.fsu.edu/~dduke/kepler3.html
Combining the periods and distances gives Keplers 3rd Law:
Mercury
Venus
Earth
Mars
Jupiter
Saturn

a
0.38
0.72
1.00
1.52
5.22
9.23

a3
Period
0.05
0.24
0.37
0.62
1.00
1.00
3.50
1.88
142.02 11.83
786.53 29.50

P2
P2/a3
0.06 1.10
0.38 1.02
1.00 1.00
3.54 1.01
140.03 0.99
870.25 1.11

So the Almagest models are indeed very much like the real
planetary orbits when viewed from Earth.

For some reason Ptolemy

makes the model for
Mercury more complicated.

http://people.scs.fsu.edu/~dduke/mercury.html

Like the Moon, the planet orbits are tilted relative to the Suns
orbit.

outer

inner

Outer planet http://people.scs.fsu.edu/~dduke/latitude.html

Inner planet http://people.scs.fsu.edu/~dduke/latitude2.html
Note that these make good sense in a heliocentric view.

In the Planetary Hypotheses Ptolemy writes:

http://people.scs.fsu.edu/~dduke/ptolemy.html
Using his nesting assumption Ptolemy gets:

Early Greek Planetary Theories

The Keskintos Inscription (found on Rhodes about 1890) and
probably carved about 100 B.C.

These are period relations as before, but much longer:

planet
Saturn
Jupiter
Mars
Venus
Mercury

years
29140
29140
29140
?
?

orbits
992
2450
15492
?
?

retrogrades
28148
26690
13648
?
?

The texts of ancient Indian astronomy give us a sort of

wormhole through space-time back into an otherwise
inaccessible era of Greco-Roman developments in astronomy.

Indian Planetary Theories

Conventional wisdom:
The orbits of the planets are concentric with the center of the earth. The single
inequalities recognized in the cases of the two luminaries are explained by mandaepicycle (corresponding functionally to the Ptolemaic eccentricity of the Sun and lunar
epicycle, respectively), the two inequalities recognized in the case of the five starplanets by a manda-epicycle (corresponding to the Ptolemaic eccentricity) and a
sighra-epicycle (corresponding to the Ptolemaic epicycle). The further refinements
of the Ptolemaic models are unknown to the Indian astronomers.

The Indian theories have even longer period relations:

planet
Saturn
Jupiter
Mars
Venus
Mercury

years
4,320,000
4,320,000
4,320,000
4,320,000
4,320,000

orbits
146,564
364,224
2,296,824
4,320,000
4,320,000

retrogrades
4,173,436
3,955,776
2,023,176
2,702,388
13,617,020

In fact, the numbers the Indians text quote for Venus and Mercury are the
number of heliocentric revolutions for each planet in 4,320,000 years:
Venus:
Mercury

7,022,388 = 4,320,000 + 2,702,388

17,937,020 = 4,320,000 + 13,617,020

eccentric (manda)

sin q( ) = e sin

epicycles (sighra) tan p( ) =

r sin
1 + r cos

1 = + 2 q( )

Aryabhatas text says:

(1)

= A

(2)

= S 1

2 = 1 + 1 2 p( )

half the mandaphala obtained from

the apsis is minus and plus to the
mean planet. Half from the
sigraphala is minus and plus to the
manda planets. From the apsis they
become sphutamadhya [truemean]. From the sigraphala they
become sphuta [true].

(3)

= 2 A 3 = + q( )

(4)

= S 3

= 3 + p( )

Jupiter
Almagest

Sunrise

-1

-2

500

502

504

506

508

510

512

Most of the difference is due to poor orbit parameters in the

Sunrise model.

What happens if we use identical orbit elements in both models?

Jupiter

Equant

Eccentric

Equant

Mars

Sunrise

-0.5

Eccentric

Sunrise

15
10
5
0

-1.5

-5
-10

-2.5
500

502

504

506

508

510

512

-15
500

502

504

506

508

Therefore, it is clear that the Almagest equant and the Indian

models share the same mathematical basis.

510

512

Arabic astronomers were very unhappy with the equant since it

violates Aristotles principle of uniform motion in a circle. By
about A.D. 1250 they had developed several alternatives that are as
good as the equant and use only uniform motion.

http://people.scs.fsu.edu/~dduke/arabmars.html
The same issues bothered Copernicus (ca. 1520-1540) and he
used the same models, although we do not know how he
learned about them.