Episodes From the Early

History of Astronomy
Springer Science+Business Media, LLC
Asger Aaboe

Episodes From the Early

History of Astronomy

With 50 Figures

, Springer
Asger Aaboe
Emeritus Professor of Mathematics, History of Science, and Near Eastern
Languages and Civilizations
Yale University
New Haven, CT 06520
USA
asger.aaboe@yale.edu

Cover ilIustratwn: The frontispiece to Giambattista Riccioli, A lmagtstum Nouum

Bononia (Bologna), 1651. (Page 110)

Library of Congress Cataloging.in.Publication Data

Aaboe, Asger.
Episodes from the early history of astronomy I Asger Aaboe.
p . em.
Includes bibliographical references.
ISBN 978-0-387-95136-2 ISBN 978-1-4613-0109-7 (eBook)
DOI 10.1007/9781-4613-0199-7
1. Astronomy- History, 1. Title.
QBI5.AI5 2001
52O'.9- dc21 QO.()61919

Prin ted on acid·free paper.

10 2001 Springer Sciencc+Business Media New York

Orig inally publi shed by Springer-Verl ag New York, Inc . in 2001

All rights reserved. This work may not be translated or copied in whole or in part without the
written pennission of the publisher (Springer Science+Business Media, LLC). except for brief
excerpts in connection with reviews or seholarly analysis. Use in connection with any fonn of
infonnation storage and retrieval, electronic adaptation, computer software, or by similar or
dissimilar methodology now known or hereafter developed is forbidden. The use of genera l
descriptive names, trade names, trademarks, etc., in this publication, eve n if the fonner are not
especially identified, is not to be taken as a sign that such names, as understood by the Trade
Marks and Merchandise Marks Act, may accordingly be used freely by anyone.

Production managed by Allan Abrams; manufacturing supervised by J oe Quatela.

Typeset by Asco Typesetters, Hong Kong.

987654321
ISBN 978-0-3 87-95 136-2 SPIN 10780466
 For my grandsons,
Samuel and Tyler Alexander
Preface

More years ago than I can easily count, I published a small

book entitled Episodes from the Early History of Mathematics
(NML Vol. 13). Here I discussed in some detail a selection of
subjects from Babylonian and Greek mathematics that could be
fully mastered by someone with a background in high school
mathematics.
My own work, particularly since then, however, has largely
been concerned with ancient mathematical astronomy, especially
Babylonian arithmetical lunar and planetary theories. I had
the great good fortune in 1963 to get access to a large collection
of unstudied, relevant clay tablets in the British Museum, so
quite naturally I began thinking about writing an astronomi-
cal companion to the mathematical volume. I was well aware,
though, that it would be quite a different sort of enterprise.
When I wrote the former volume, I could take for granted that
my readers would have some familiarity with the elements of the
subject whose early history was my concern, for nearly everyone
has seen some basic arithmetic, algebra, and geometry.
It is, however, far otherwise for astronomy. If students take
any astronomy at all, they may learn of the evolution of stars,
and even of things that were certainly deemed unknowable in
my youth, such as what Mars's surface and the far side of the
moon look like in detail, but they remain in most cases woefully
ignorant of what you can expect to see when you look at the sky
with the naked eye, intelligently and with curiosity. I write this
from experience. I first became aware of the state of astronomi-
cal enlightenment when, many years ago , out of curiosity I took
a vote, yes or no, on a few questions from elementary spherical
astronomy in a mathematics class of students selected for their
excellence.

vii
Vlll Preface

I first asked if the sun rises and sets. After the students'
Copernican scruples were stilled, and they were sure I knew
they knew that it was really the earth's rotation that made it so
appear, they voted yes, the sun rises and sets. Next I asked the
same question about the moon and, after some mutterings about
phases, they voted yes again, as they did for the planets, for, as
they reasoned, "planet" means "wandering star." But for the
fixed stars their answer was a quick and unanimous no, for they
are fixed.* My final question was whether one can see the moon
in the daytime; again they voted no unanimously. This was dis-
turbing, for the moon obliged by being plainly visible, and in
broad daylight, through the lecture room's window, which I duly
pointed out to them to their astonishment.
It was thus clear that an introduction to naked-eye astronomy
would be necessary in a book on early astronomy; I have given
one here in Chapter O. I tried to keep it purely descriptive, un-
influenced by modern knowledge. A case in point is the retro-
gradation of planets. This phenomenon-that the planets, in
their slow eastward motion among the fixed stars, come to a
halt, reverse direction, and come to a halt again before con-
tinuing their eastward travel, is usually introduced in terms
of the Copernican, heliocentric system. The argument involves
the changing directions of lines of sight from a moving earth to
a moving planet, and it is difficult, as I know only too well, to
make students visualize, in this fashion, the phenomenon this
explanation is supposed to explain. This is not strange, for
planetary retrogression depends on the observer's being on the
earth and so is best and easiest accounted for in a geocentric
system.
In the astronomical introduction I simply describe the way the
sun, moon, stars, and planets appear to behave to anyone who
has the time, patience, will , and wit to observe and remember.
Later in the book I discuss the various mathematical models,
arithmetical and geometrical, that were devised in antiquity to
account for the observed behavior of these bodies, particularly
the planets.

* The correct answer is that some rise and set, while others never set,
and the rest never rise. The two exceptions are at the equator, where
all stars rise and set, and at the poles, where a star is either always
above or always below the horizon. Thus, the students would have
been nearly right if they had lived at either pole, but they did not.
 Preface ix

I need not, however, introduce the mathematical techniques

ancient astronomers used-Babylonian arithmetic and Greek
geometry and trigonometry-for I already treated them in my
previous little book.
In the following I concentrate on planetary theory and try to
avoid the moon as much as possible. Here I am reminded of a
story about Ernest Brown (1866-1938), professor of mathematics
at Yale University. Brown devoted his life to the study of the
moon's motion, and he published his lunar tables in three folio
volumes in 1919. Toward the end of his life he was inanely asked
what he could say about lunar theory. His answer was heartfelt,
" It is very difficult." And so it has always been, ever since its ele-
gant beginning in Mesopotamia some 2500 years ago, and I could
find no place for it within the limits I had set for this little book.
Nor shall I consider early attempts at accounting for the
planets' motion in latitude. These motions were referred to the
mean sun, so their descriptions remained unduly complicated
until Kepler saw that the planes of the planetary orbits pass
through the true sun (it is in this connection he wrote that
Copernicus was unaware of his own riches).
In Chapter 1 I introduce Babylonian arithmetical astronomy,
the earliest, and highly successful, attempt at giving a quantita-
tive account of a well-defined class of natural phenomena. I also
mention the preserved observational records and hint at how
the theories could have been derived from such material. [The
elegant English mathematician G. H. Hardy (1877-1947) was
particularly devoted to the theory of numbers-the domain of
mathematics dealing with the properties of whole numbers-
because of its purity in the sense that it found no application
outside mathematics. He would have been greatly amused to
learn that it was precisely number theory that, in the hands of
the Babylonian astronomers, became the first branch of mathe-
matics to be used to make a natural science exact.]
When I turn to Greek geometrical models for planetary be-
havior in Chapter 2, I can no longer afford to cite observations
or to deal with how the quantitative models' parameters were
derived from them. It is not that I deem these matters unim-
portant-quite the contrary-but such discussions would stray
too far from my main purpose, which is to describe the various
geometrical models for the planets and show how they work.
Further, I am particularly interested both in demonstrating
that epicyclic planetary models are not just ad-hoc devices for
x Preface

mimicking how a planet seems to behave, but are good descrip-

tions of how a planet in fact moves relative to the earth, and in
identifying their various components with their counterparts
in the solar system. Indeed, a well-read medieval astronomer
who considered a sun-centered system of planetary orbits could
immediately derive its dimensions in astronomical units (one
astronomical unit equals the mean distance from the earth to the
sun) from the parameters of Ptolemaic epicyclic models, as we
know Copernicus did (a note in his hand of this simple calcula-
tion is preserved in Uppsala University's library).
A crucial point in the demonstration is the transformation
from a heliocentric to a geocentric coordinate system. This
ought to be simple enough, but I have found it extraordinarily
difficult to make people visualize the same phenomenon in the
two systems. One sticking point is to make them realize that
even the nearest fixed star is so far away that the directions to it
from the sun and from the earth are the same (except for the
annual parallax, which is so small that it was not observed until
the 19th century).
The fixed-star sphere therefore looks the same whether viewed
from the earth or from the sun. For an observer on the earth, the
sun will seem to travel in a near-circular orbit, one revolution a
year relative to the background of the fixed stars. The planets,
in turn, will revolve in their orbits around the moving sun. This
arrangement, the "Tychonic system" as Tycho Brahe himself
modestly called it, is the one earthbound astronomers observe
in and the one the Naval Observatory uses for compiling the
Nautical Almanac (mariners, after all, are on the watery surface
of the earth and do not care about what things look like from
the sun).
The Copernican and Tychonic arrangements are geometri-
cally equivalent, for either implies the other and yields the same
directions and distances from one body to another. However, if
you insist that the origin of your coordinate system-the earth
for Ptolemy and Brahe, the mean sun for Copernicus (he has the
true sun itself travel in a small circle around the mean sun)-is
"at rest in the center of the universe," you do not, of course,
have dynamical equivalence between the two systems, and the
Copernican arrangement in Kepler's version, focused on the
true sun, became the basis of Newton's mechanical treatment of
the solar system.
In the Almagest Ptolemy does not give his planetary models
 Preface xi

absolute size: He measures a model's dimensions in units, each

of which is one-sixtieth of the deferent's radius, because he only
wants them to yield directions to the planets. So far they are
perfectly compatible with the Tychonic system and, in fact, if
we scaled the Ptolemaic models properly, they would also give
us the distances to the planets correctly.
However, in his Planetary Hypotheses Ptolemy constructs his
cosmological scheme, the Ptolemaic system, of snugly nested
spherical shells, all centered on the earth, each containing the
model of one planet, and with no wasted space. Here he commits
himself to the dimensions of the structure in terrestrial units,
and now agreement with the Tychonic system is no longer pos-
sible. I discuss these things in Chapter 3.
Though I originally intended to treat only ancient topics,
I could not help including in Chapter 2 a few remarks on later
modifications of Ptolemy's models at the hands of medieval
Islamic scholars. These revisions were mostly of three kinds: im-
provements of parameters; much-needed corrections of serious
flaws in his lunar theory; and attempts at replacing his philo-
sophically objectionable equant with combinations of philo-
sophically correct uniform circular motions that would work
almost as well .
I end Chapter 2 with a few remarks about Copernicus and
Brahe. Of Brahe's many achievements I only mention the
Tychonic system, for its arrangement is, as said, compatible with
the Almagest's planetary models.
In my comments on Copernicus's work, I concentrate on just
two aspects. First, I try to make clear precisely which problem
was resolved by the heliocentric hypothesis. The general lit-
erature is often vague on this point, suggesting, for example, a
desire for higher accuracy, or a dislike of epicycles, as a moti-
vation for the new system. Both suggestions are wrong. In fact,
the motivation lay in a desire to get rid of the awkward ques-
tions raised by the sun's curious role in the Ptolemaic planetary
models: For an inner planet the deferent's radius to the epi-
cycle's center always points toward the mean sun, while for the
outer planets the radii from the epicycles' centers to the planets
are all parallel and point in the same direction as that from
the earth to the mean sun. This strange role was difficult to
explain, but it becomes an immediate consequence of placing the
mean sun in the middle of the planets' paths, as we shall see.
(The Tychonic arrangement has the same virtue.)
xii Preface

Second, I discuss the fine-structure of Copernicus's planetary

models. His ideal is clearly to have each planet move uniformly
in a circular path-not, alas, around its center, but around an
equant point in analogy to Ptolemy's deferent. However, he has
committed himself to the exclusive use of uniform circular
motions and manages, by superposition of several of these, to
make his planets move uniformly around his ideal equant
points-exactly, not approximately-while their resulting paths
are nearly, but not quite, circular. I shall point out that these
arrangements are precisely what we have found in the works of
some of his Islamic predecessors.
In Chapter 3 I present, in some detail, Ptolemy's cosmology,
the first cosmological scheme to include quantitative models in
an integral way. It was long called the Ptolemaic system, even
though Ptolemy's authorship of it was established only a few
decades ago, when my then-colleague at Yale, Bernard Goldstein,
found Ptolemy's own description of it in an Arabic translation of
a lost part of the Greek original of his Planetary Hypotheses.
Though the Ptolemaic system is now most often talked about in
slighting terms, it is a logically pleasing structure that was , after
all, the basis for how educated people thought about the universe
for nearly a millennium and a half.
In the final chapter I show, from a more modern point of view ,
why Ptolemy's equant is so efficient. I take this opportunity to
sketch how Kepler proceeded in order to find the longitude of
a planet that moves according to his laws, and then I go on to
analyze how a planet seems to behave when observed from the
empty focus of its elliptical orbit, the one not occupied by the
sun. It turns out that its angular motion is uniform but for terms
involving second and higher powers of the ellipse's eccentricity,
so the empty focus plays the role of an equant point and, for
small eccentricities, ellipses are nearly circular. The combina-
tion of two eccentric circular motions with equants yields a
planetary model very close to Ptolemy's.
The result of this analysis is not to be found in the more recent
general literature, but it is far from new. In fact, in his Principia
Isaac Newton addressed the problem of a planet's angular
motion around the empty focus of its orbit and reached nearly
the same result as the one I derive in Chapter 4.
In the course of this derivation I trangressed the limits I orig-
inally imposed on myself, for I could not help including an inte-
gral or two . However, having set the rules myself, I felt free to
break them, and a few integrals never hurt anyone.
 Preface Xlll

At one time or another, I have dealt with most of the above,

very idiosyncratic selection of topics in courses and seminars
at Yale University. Additionally, in March of 1988, I was pleased
to be invited to give an Honors course at the University of
Pittsburgh; in my lectures there I also presented some of this
material. I am grateful to the students in both places who , by
their questions, made what I thought and wrote clearer.
Furthermore, I wish to thank my colleagues, Dr. John
Britton and Professors Bernard Goldstein and Alexander
Jones, for reading my manuscript and for helpful criticism and
suggestions.
Finally, I must acknowledge my indebtedness and gratitude
to Miss Izabela Zbikowska of the Polish Academy of Science
and Yale University. Without her constant help during the last
three years, this little book, begun so long ago, would still be
unfinished.

Acknowledgments
I have chosen special figures to illustrate the varied nature of
the sources for our knowledge of earlier astronomy: cuneiform
clay tablets and a Greek papyrus, fragmentary, but original; a
page of a Byzantine manuscript, many copies removed from the
original work; a page from the manuscript copy of De revolu-
tionibus ... in Copernicus's own hand, about as authentic as
evidence can get; and the frontispiece of an elegantly printed
astronomical treatise of the 17th century.
The photographs of the cuneiform texts (Figures 2 and 5 of
Chapter 1) are published through the courtesy of the Trustees
of the British Museum.
The photograph of the Greek papyrus (Figure 6 of Chapter 1)
is reproduced with the kind permission of the owner, Professor
Marvin L. Colker.
The Vatican Library provided me with a photograph of
Ptolemy's Kinglist from Vat. gr. 1291 and gave me permission
to publish it (Figure 14 of Chapter 2).
The photographic copy of the frontispiece of Riccioli's Alma-
gestum Novum ... of 1651 (Figure 20 of Chapter 2) was made for
me by members of the Institute for the History of the Exact
Sciences at Aarhus University, Denmark, the home away from
home for so many scholars, from a copy in their library.
I offer my thanks to these people and institutions.
XIV Preface

Advice to the Reader

I have arranged the chapters in approximate chronological
order, but each makes some sense even if read independently.
Some readers might profitably begin, after Chapter 0, with
Chapter 2 on Greek geometrical models, for they may find its
methods less strange than those of the Babylonian arithmetical
schemes.

New Haven, Connecticut Asger Aaboe

January, 2001
Contents

Preface Vll

0 What Every Young Person Ought to Know About

Naked-Eye Astronomy 1

1 Babylonian Arithmetical Astronomy 24

2 Greek Geometrical Planetary Models 66

3 Ptolemy's Cosmology 114

4 Kepler Motion Viewed from Either Focus 135

Selected Bibliography 171

xv
o
What Every Young Person
Ought to Know About Naked-Eye
Astronomy

In order to provide a starting point for an understanding of

ancient astronomical texts, I shall begin by presenting, in all
brevity, the basic elements of naked-eye astronomy. I shall, of
course, deal principally, but not entirely, with phenomena of
interest to ancient astronomers. Among these are many phe-
nomena, such as the first or last visibility of a planet or the
moon, that the modern astronomer shuns since they take place
near the horizon and further depend on imperfectly understood
criteria. Thus, these phenomena are not commonly discussed in
the modern astronomical literature and, more seriously, we lack
modern standards with which we may measure the quality of the
ancient results.
My presentation is entirely descriptive. I do not attempt to
explain why planets become retrograde, or that celestial bodies
really do not rise and set but that the earth's rotation makes it
appear that they do. Whoever does not feel comfortable talking
about the solar system without taking a detour via the sun is
welcome to do so.

The Celestial Sphere, Fixed Stars, Daily Rotation

We have no simple means of judging distances to celestial
bodies; we can only determine the directions toward them. To
describe what we thus observe, we introduce the celestial sphere
as a spherical surface with its center at the observer's eye and of
unit radius-which unit is irrelevant, but it is often thought
more comfortable to choose a very large one. A celestial object
is then mapped at, or identified with, the point on this spherical
surface at which the line of sight to the object pierces the celes-
A. Aaboe, Episodes From the Early History of Astronomy
© Springer-Verlag New York, Inc. 2001 1
2 O. What to Know about Naked-Eye Astronomy

tial sphere. The study of the behavior of this map of celestial

objects on the celestial sphere is called spherical astronomy.
If all celestial bodies visible to the naked eye are thus mapped
on the celestial sphere, it becomes apparent that the vast ma-
jority of them remain in fixed patterns with respect to each
other: They form recognizable constellations that, in turn, re-
main unchangeably distributed. These celestial bodies are called
the fixed stars. To the naked-eye observer there remain seven
exceptional objects: the Sun, the Moon, and the five bright
planets: Mercury, Venus, Mars, Jupiter, and Saturn; being ex-
ceptional, they are, of course, of particular interest.
Before describing the celestial sphere and its motion, I shall
briefly introduce, informally and without proofs, some basic ter-
minology and a few results from spherical geometry (i.e., the
geometry on the surface of a sphere).
A plane cuts a spherical surface, if at all, in a circle (or in one
point if the plane is tangent to the sphere). If the plane happens
to pass through the center of the sphere, the section is called a
great circle; otherwise the section is called a small circle. Great
circles are fundamental to spherical geometry; indeed, they play
much the same role that straight lines do in plane geometry.
Thus, through two points on a spherical surface that are not
diametrically opposite, passes one and only one great circle, and
the shortest distance between two such points measured on the
sphere is along the great circle joining them. By the "distance
between two points on the sphere," we mean the shortest dis-
tance, namely, the length of the shorter of the two great circle
arcs that join them; this distance is usually given not in linear
measure but in degrees (thus, the longer of the two great circle
arcs joining two points will be 360 minus the distance between
0

the points).
Associated with a great circle are two points called its poles ;
they are the end points of that particular diameter of the sphere
which is perpendicular to the plane of the great circle (i.e., they
are to the great circle what the North and South poles are to the
equator). Conversely, to two diametrically opposite points on the
sphere there corresponds one, and only one, great circle whose
poles they are. The distance from a pole to any point on the
corresponding great circle is 900 •
Two great circles always intersect in diametrically opposite
points, that is , they always bisect each other. The angle between
two great circles is the same as the distance (in degrees) be-
tween their poles.
 The Celestial Sphere, Fixed Stars, Daily Rotation 3

These few remarks may suffice for our present purposes, and
we can now return to the celestial sphere and its behavior.
The fixed stars permit us to get a hold on the celestial
sphere-they provide us with a coordinate system, if you will-
and enable us to perceive its motion. Ignoring for the moment
various very slow changes, we will observe that the celestial
sphere, with the fixed stars fixed upon it, revolves about a fixed
axis at a fixed rate of very nearly 366 1/4 (sic) revolutions per
year, relative to familiar fixed objects in our surroundings. This
axis, being a diameter of the celestial sphere, pierces it at two
diametrically opposite points called the celestial north and south
poles (the north pole is now very near the North Star, or Stella
Polaris). The great circle corresponding to these poles (i.e., the
great circle that slides in itself during the daily rotation, as this
motion is usually named) is called the celestial equator or, simply,
the equator.
A horizontal plane through the observer intersects the celes-
tial sphere in a great circle called the horizon, and a vertical line,
also through the observer, pierces the sphere at two points, zenith
above and nadir below, which are the poles (the term "pole" is
used here as in spherical geometry) belonging to the horizon.
Celestial objects below the horizon are invisible since the line of
sight to them would pass through the body of the earth: The
plane of the horizon is ideally tangent to the spherical earth.
The vertical plane through the observer and the north pole,
which because it is vertical also contains the zenith, intersects
the celestial sphere in a great circle called the meridian and
meets the horizon in two points: the north and south points.
The vertical plane through the observer perpendicular to the
plane of the meridian intersects the celestial sphere in a great
circle called the first vertical , which meets the horizon in the east
and west points. The equator passes through the east and west
points.
For a given place of observation, the directions to the north,
south, east, and west points as just defined, remain fixed in
relation to characteristic features of the visible neighboring
terrain. Furthermore, the elevation of the north pole does not
change.* These two facts together imply that the axis through
the north and south poles, about which the daily rotation takes

* Here there is no need for the cautionary remark about long-term

changes.
4 O. What to Know about Naked-Eye Astronomy

place, remains fixed for an observer at a given locality in rela-

tion to his or her terrestrial surroundings. The elevation of the
north pole above the horizon, in angular measure, is called the
terrestrial latitude of the place of observation and is usually
denoted by 4>. For Babylon we have 4> = 32f, very nearly.
In relation to a given horizon, the fixed stars are divided into
three classes: those that are always above the horizon; those
that are sometimes above and sometimes below the horizon; and
those that are always below the horizon.
The stars in the first category are called circumpolar. They are
all within a cap with the north pole as its center (for observers
in the northern hemisphere of the earth) and a radius equal to
the observer's terrestrial latitude 4>. Thus, the farther north you
live, the larger the region of the circumpolar stars.
A corresponding cap of equal size but centered on the south
pole contains the fixed stars that are never above the horizon.
All the fixed stars in the belt between these two circumpolar
caps will cross the horizon in the east and in the west and by the
daily rotation will be carried in a path partly above and partly
below the horizon. This daily path-a circle on the celestial
sphere-is traversed by the fixed star in slightly less than 24
hours. A star on the equator is as long in time above as below
the horizon, for its diurnal path is bisected by the horizon. Again
for an observer in the northern hemisphere, if a star is north of
the equator, it spends more time above the horizon than below-
the difference is larger the nearer the star is to the circumpolar
cap-and symmetrically for a star south of the equator.
Since a star is visible only at night and when above the hori-
zon, it follows that circumpolar stars are visible every night
of the year. Those circumpolar about the south pole are never
visible. A star in between is visible more nights the farther north
it is, for the interval of nighttime shifts throughout the year in
relation to a star's horizon crossings, as we shall see.
The sizes of the circumpolar caps and of the horizon-crossing
belt vary with terrestrial latitude, as mentioned. To illustrate
this, let us consider two extremal situations.
First, for an observer on the earth's North Pole, where the
terrestrial latitude 4> is 90°, zenith and north pole coincide. The
circumpolar caps meet at the horizon, and the horizon-crossing
belt vanishes. All visible stars are circumpolar, their diurnal
paths are parallel to the horizon, and no stars ever cross the
horizon. One has a chance of seeing only the stars on the north-
 Sun, Ecliptic, Seasons 5

ern celestial hemisphere, whereas the rest are never above the
horizon.
Second, for an observer on the terrestrial equator, it is the
circumpolar caps that vanish while the horizon-crossing belt
fills the entire celestial sphere. The north and south poles are in
the horizon, and all stars spend equal time above and below the
horizon. One has, theoretically, an equal chance of seeing all
stars, though in practice those stars near the poles never get
very far above the horizon.
The diurnal circle of a star in the horizon-crossing belt inter-
sects the meridian in the points of upper and lower culm ination
at which the star is at its greatest elevation above, and lowest
depression below, the horizon. For a circumpolar star these two
points mark its greatest and smallest distance from the horizon,
respectively.

Sun, Ecliptic, Seasons

For two reasons it is natural to begin an introduction to spheri-
cal astronomy, as I have, with the fixed stars. First, their be-
havior is simpler than that of the other celestial bodies: They
remain fixed in relation to each other, and all join in the uniform
diurnal rotation about a fixed axis. Second, they provide a con-
venient background against which the more complicated behav-
ior of sun, moon , and planets can be perceived and described. It
must be emphasized that once a celestial body has been placed
among the fixed stars it will, of course, partake of the same daily
rotation that they are subject to, in addition to any motion that
it will have relative to them.
The first example of this approach is the case of the sun. Let
us imagine, contrary to our everyday experience, that it were
possible to see the sun and fixed stars at the same time. (This is
now so for an observer outside the earth's atmosphere.) We
would then note that the sun, day by day, moves eastward very
slowly among the fixed stars, in the amount of about 1 per day .
0

It returns exactly to its original place after the lapse of one

year-that is, indeed, the definition of the year or, more pre-
cisely, the sidereal year-having traced out in that interval a
path among the stars which is a great circle. Year after year the
sun travels precisely the same great circle, which is called the
ecliptic.
6 o. What to Know about Naked-Eye Astronomy
In the time it takes the sun to complete one revolution in the
ecliptic, namely, in one year, the fixed stars revolve, as said,
very nearly 366 1/4 times relative to, say, the meridian. This is
the same as saying that a year has about 365 1/4 days, for the sun
has in that period revolved the same number of times as the
stars less the one revolution it itself has performed relative to
them, but in the opposite sense of the daily rotation.
The ecliptic plays a fundamental role as reference circle in
ancient astronomy. This is not surprising, for not only is it the
path of the sun, but the moon and the planets are always within
a belt extending at most 100 on either side of it.
The ecliptic has an inclination toward the equator of about
o
23 ~ • The two diametrically opposite points of intersection be-
tween ecliptic and equator are called the equinoxes. When in its
travel the sun happens to be in either one of these, its diurnal
path is the equator itself which is bisected by the horizon, so day
equals night in duration. When in its yearly motion the sun
crosses the equator from the south to the north, it is vernal or
spring equinox-this term is commonly applied both to the phe-
nomenon and to the point on the celestial sphere-and the other
crossing is called autumnal or fall equinox.
The point halfway between the equinoxes and at which the
sun is farthest north of the equator is the summer solstice, and
the diametrically opposite point is the winter solstice. When the
sun is at these points, the duration of daylight is longest and
shortest, respectively.
Let us , once more, consider the previous two extremal situa-
tions. First, for an observer on the North Pole of the earth,
where zenith and celestial north pole coincide, as do horizon and
celestial equator, the sun will appear in the horizon for the first
time at "vernal equinox" and will not set until "autumnal equi-
nox." From the vernal equinox the sun slowly gains elevation
above the horizon until it reaches an altitude of 23 -the r
'a mount of inclination of the ecliptic against the equator-at
summer solstice. In the course of 24 hours the sun will be seen
above every point of the horizon, and the shadow cast by a ver-
tical stick will revolve 360 From summer solstice the sun
0
•

slowly works its way back down to the horizon, and from fall
equinox to spring equinox it will be invisible.
On the earth's equator, where the celestial equator and the
first vertical coincide, day and night are always equal, and the
"equinoxes" are marked by the sun passing through zenith at
 Sun, Ecliptic, Seasons 7

noon. At the solstices the sun crosses the meridian farthest

north or south of zenith.
Two further special cases are of interest. One is to find the
zones on the earth where the sun becomes circumpolar just once
a year. For this to happen the circumpolar cap whose radius
is the terrestrial latitude ¢ must reach the sun when farthest
(i.e., 23f) from the equator. Thus, ¢ must be 90° - 23f = 66f,
At this northern latitude the sun will just reach, but not cross,
the horizon at midnight on summer solstice, and symmetrically
for the southern hemisphere. The two circles on earth of these
latitudes are called the polar circles.
The other special case is to determine where on the earth the
sun just reaches the zenith once a year at solstice. It is readily
r
seen that this happens at terrestrial latitude 23 north or south.
The corresponding two circles are called the tropics, the north-
ern of Cancer, the southern of Capricorn, for reasons that will
become clear later.
Finally, I shall mention a variant way of characterizing ter-
restrial latitude, namely, by giving the ratio of longest to short-
est daylight. This works, of course, only for places between, but
not on, the polar circles. Since, for reasons of symmetry, the
shortest day at a given locality equals the shortest night in
length, this ratio tells us how the sun's diurnal circle at summer
solstice is divided by the horizon; it is then a fairly simple matter
to determine, if one wishes, the elevation of the pole (i.e., the
place's terrestrial latitude) by means of spherical trigonometry.
On the equator of the earth this ratio is 1: 1. The farther north
a locality, namely, the higher the North Pole is elevated above
the horizon, the larger this ratio becomes until it loses definition
on the polar circle, where the longest day is 24 hours and the
shortest 0 hours.
We find this practice in ancient Greece, and a ratio of longest
to shortest daylight of 3 :2 for Babylon is, as we shall see, a fun -
damental parameter in Babylonian astronomy. I must empha-
size, however, that we have no evidence whatsoever that the
Babylonians were aware that this ratio changes as one travels
north or south.

Synodic Cycle of a Star Near the Ecliptic

As an introduction to phenomena of the kind dealt with in
Babylonian astronomy, let us consider a fixed star on or near the
8 O. What to Know about Naked-Eye Astronomy

ecliptic; we shall be concerned with when, where, and how long

it is visible from a place of observation of reasonable latitude.
Since the star is close to the sun's yearly path, at a particular
time of year the sun and star nearly coincide on the celestial
sphere or , as we say, are in conjunction. The star will be invisi-
ble, for whenever it is above the horizon, so is the sun. Star and
sun will rise simultaneously on that day.
When the star is about to rise the next morning, the sun will
have moved about 10 farther along the ecliptic so that the star
rises a little before the sun. The following morning the sun will
have moved yet another degree away from the star, so the inter-
val from starrise to sunrise has lengthened. Eventually a morn-
ing will come when the star rises so long before sunrise that the
sky is sufficiently dark for the star to be visible as it crosses the
horizon, if only for a short while until the dawn extinguishes it.
This is the phenomenon of first visibility (we usually denote it by
the letter I').
From the morning of first visibility, the star will rise earlier and
earlier, and we can follow it farther and farther from the eastern
horizon along its diurnal arc before dawn makes it vanish.
Half a year after conjunction the sun reaches a point on the
ecliptic 1800 from the star. Since ecliptic and horizon, being
great circles, always intersect in diametrically opposite points,
we now have the situation that when the star is in the horizon,
so is the sun. Thus, the star rises at sunset and sets at sunrise;
further, the star will be in upper culmination at midnight. We
say that now the star is in opposition to the sun, and we use the
letter e for this phenomenon. Weare able to see the star all
night long-less the intervals of dawn and dusk.
As the sun now progresses in its yearly motion, it approaches
the star from the other side, from the west. After each sunset,
when it gets dark enough to see it, the star will already be well
past the eastern horizon, and it will at its appearance get closer
and closer to the western horizon across which it sets.
There will now come an evening when the star appears only
just before it sets, while on the next evening the sun will have
gotten so close to it that the sky is not dark enough for the star
to be seen before it sets. This is the phenomenon of disappear-
ance of the fixed star, and we denote it by the letter D.
The star will now remain invisible while the sun catches up
with it. Star and sun will , once again, be in conjunction, and the
cycle is closed, having occupied just one year.
 Sun, Ecliptic, Seasons 9

What I have described here is called the fixed star's synodic

cycle, consisting of the synodic phenomena conjunction, first ap-
pearance, opposition, last appearance, and conjunction-once
again, all phenomena that place the star in special relation to
the sun.
The visibility phenomena, first and last appearance, or rand
fl, are determined by several factors, some of which are difficult
to control. First they depend on the brightness of the star: The
brighter the star is, the shorter the sun has to be removed from it
for it to be visible near the horizon. Second, they depend on the
inclination of the ecliptic against the horizon when the star is
rising or setting: The smaller the inclination, the farther the sun
must be from the star to ensure the darkness necessary for visi-
bility. Furthermore, these phenomena depend on the acuity of
the observer's eyesight and the quality of the atmosphere; in the
former there is clearly a personal variation, and the latter is
difficult, if not impossible, to ascertain for a place of observation
in antiquity.
Since the synodic period of a fixed star is one year, anyone of
the synodic phenomena can be used as a seasonal indicator. This
is, indeed, an ancient practice. It is well known that the first
appearance of Sothis, our Sirius, was used by the ancient Egyp-
tians as a herald of the rising of the Nile-two seasonal phe-
nomena that happened to coincide-and one finds various rustic
tasks tied to first or last appearances of certain fixed stars or
constellations in primitive societies as we learn, for instance,
from Hesiod's Works and Days .
Finally, a remark about the term "synodic"; synodos in Greek
means "getting together" or "meeting" and in astronomy is used
particularly for the coincidence in position of a celestial object
with the sun, that is, for conjunction. It is now used for any
phenomenon linking a star, a planet, or the moon to the sun in a
certain fashion.

Synodic Cycle of an Outer Planet

The planets are conspicuous among celestial bodies for at least
the following three reasons: They are all very bright; they do
not twinkle like the fixed stars; and they move relative to the
fixed stars though always staying close to the ecliptic. They are
divided into two classes consisting respectively of those that can
reach opposition to the sun and those that cannot.
10 O. What to Know about Naked-Eye Astronomy

The members of the first class-we call them outer planets-

are Mars, Jupiter, and Saturn, as far as naked-eye astronomy is
concerned. The second class consists of the two inner planets,
Mercury and Venus. Characteristic of them is the limit to how
far they can get from the Sun: Venus never gets farther from the
Sun than about 40°, and Mercury's limits are even narrower-
about 25°. We shall first be concerned with the outer planets.
The synodic behavior of Saturn has much in common with
that of a fixed star near the ecliptic. Like it, Saturn makes its
first and last appearance (also denoted rand [2), and near the
middle of its interval of visibility it is in opposition (8) to the
sun. The difference is that Saturn has a motion of its own: The
general trend of this motion is that from r to r (or from n to n),
that is, during one synodic period, Saturn moves about 12° along
the ecliptic in the same direction-eastward-as that of the
Sun's proper motion. This means that the synodic period of
Saturn is some 12 days longer than a year, for the Sun will
have to travel 12° in excess of one complete revolution in the
ecliptic to catch up with Saturn again, and the Sun travels
almost 1° per day.
Within each synodic period Saturn's motion is, however, more
complicated; after its first appearance I', Saturn first moves
directly, as we call eastward motion along the ecliptic, then
comes to a halt among the fixed stars-it is now at its first
stationary point <ll-then reverses its motion, or becomes retro-
grade, and then comes to a halt once again at its second station-
ary point 'lJ, where its motion finally becomes direct, once again.
Opposition happens in the middle of the retrograde arc. The
synodic phenomena of Saturn are then, in their proper order,
T: first visibility
<ll: first stationary point }
8: opposition retrogradation
'lJ: second stationary point
n: disappearance
An actual run of Saturn for some three synodic periods is
illustrated in Figure 1, where the dimensions perpendicular to
the ecliptic are exaggerated four times for the sake of clarity.
The dotted stretches from n to r are its arcs of invisibility in the
middle of which it is in conjunction with the sun.
The other outer planets, Jupiter and Mars, behave qualita-
tively like Saturn. The character and sequence of their synodic
 Sun, Ecliptic, Seasons 11

___,:...-...1.'
A3 50' 5,0' A.2
-1._ 4,0'
_ Al ,
30'
."-----_'---:-:--,- _
I_ dA2 i.. dAI------l1 ecliptic
I , I
I 1 I
r
r n!
0' I I
-o----~'l' !
<I>~ o-----~ r n:
Q <I>~ o----~'l' r n
0----<)-

e
FIGURE 1.

phenomena are quite the same, but their synodic arcs-the

progress from, say, r to f-are larger. Jupiter's synodic arc is
some thirty-odd degrees, and that of Mars is even in excess of
3600 • Thus, their synodic periods are longer than Saturn's, that
of Mars even longer than two years. I shall later return to these
matters and treat them from a more quantitative point of view.

Synodic Cycle of an Inner Planet

Since an inner planet is, as it were, tethered on a short leash to
the sun, it will, in the long run, travel among the fixed stars as
many times as the sun does or , on the average, once a year.
When an inner planet is, within its bounds, sufficiently west of
the sun, it will be visible in the eastern sky for an interval before
sunrise, and it is then called a morning star. When it is suffi-
ciently east of the sun, it will be seen in the western sky for a
while after sunset, and it is then called an evening star. Since
Mercury is tied to the Sun within rather narrow limits, it is vis-
ible, if at all, only for relatively short intervals before sunrise or
after sunset. The synodic phenomena of an inner planet are, in

l
order,

-'
~ . first appearance as an evening star;
greatest eastern elongation from the sun
1lJ: stationary point in the west
0: disappearance as an evening star;
inferior conjunction retrogradation
f: first appearance as a morning star
cI>: stationary point in the east;
greatest western elongation from the sun
2:: disappearance as a morning star;
superior conjunction
12 o. What to Know about Naked-Eye Astronomy
The phenomena that have not received Greek letters are not
considered in Babylonian astronomy. At inferior and superior
conjunction, an inner planet is, of course, invisible so it is not
odd that they are not among the Babylonian synodic phenomena
(the distinction between inferior and superior conjunction re-
flects our modern knowledge that the inner planets circle about
the sun). One may find it strange, however, that an inner planet
at greatest elongation from the sun-the situation when it is
visible the longest-is not given special consideration.

The Moon's Synodic Course

The moon moves directly, namely, eastward, among the fixed

stars at a rate of about 13° per day in an orbit that is inclined
approximately 5° toward the ecliptic; the moon will thus return
very nearly to the same position among the fixed stars after some
27 ~ days, an interval that is called one sidereal month.
To get some sense of the moon's swiftness, one may consider
that the moon's apparent diameter is slightly more than r, so
the moon moves a distance of its own width in less than one
hour. Thus, the moon's proper motion can be readily perceived
in quite a short time if one refers it to neighboring fixed stars.
When we considered the synodic cycle of a fixed star or of an
outer planet, the sun was the swifter body that ran away from,
and caught up with, the other. With the moon the roles are re-
versed. If sun and moon are in conjunction near a fixed star at
a certain moment, the moon will reach the star again after one
sidereal month, or some 27 ~ da ys, in the course of which the sun
has moved ahead only some 27°. Thus, it will take the moon a
little more than two extra days to reach the sun and, once more,
be in conjunction. The interval in time from conjunction to
conjunction is, then, somewhat over 29 days-on the average,
29.5309 days, to be more exact-and is called one synodic month
or one lunation. In addition to the year and to the day, this is a
most important time interval in ancient calendars, and funda-
mental in Babylonian astronomy.
When we now consider the sequence of synodic phenomena
within each synodic cycle, we should recall that the moon's
phase can be used as an additional indicator of its synodic state.
Furthermore, it should be remembered that the moon is the only
 Sun, Ecliptic, Seasons 13

celestial body that may be seen with ease simultaneously with

the sun, that is, in the daytime (one may actually at times follow
the progress of Venus across the day sky, but it is difficult, and
one has to know where to look) .
At conjunction the moon is invisible, unless it happens to
eclipse the sun; we shall return to the question of eclipses
later. As the moon now moves away from the sun at the rate of
some 12° per day-the difference between lunar and solar daily
motion-it increases its chances for being visible in two ways :
by getting farther from the sun, and by its lighted sickle growing
in width. At the first or second sunset-in exceptional circum-
stances, perhaps the third-after conjunction, the thin crescent
becomes visible in the west against the darkening evening sky
before the moon sets following the sun. This is the evening of
first visibility of the new moon and it marks the beginning of a
new month in the Babylonian calendar. Determining in advance
which evening the new moon will become visible is one of the
chief goals of Babylonian lunar theory and, as we shall see, is a
highly difficult task.
Some seven days after conjunction, the moon is 90° from the
sun, and half its visible surface is lighted. We say that the moon
is at first quarter. It will now rise near noon, when the sun cul-
minates; at sunset the moon at first quarter will be near culmi-
nation, and it sets near midnight.
Half a synodic month, or 14 to 15 days, after conjunction, the
moon reaches opposition to the sun. Its entire visible surface is
lighted and we have full moon . The full moon rises at sunset and
sets at sunrise. Eclipses of the moon occur at full moon. Both
conjunctions and oppositions of the moon are called by the
common term syzygy (strictly speaking, "syzygy" means pre-
cisely the same as conjunction, namely, the state of being yoked
together).
Some 21 or 22 days after conjunction, the moon is at its last
quarter, and a little before the next conjunction will be a morn-
ing when the waning sickle-the term "crescent" is still often
used, though a misnomer-is seen to rise just before sunrise
for the last time. This is the morning of last visibility of the
moon, and it is followed by an interval during which the moon
is too close to the sun, and its lighted part too slender, for
it to be visible. When conjunction is reached in the middle of
this interval of invisibility, the synodic cycle of the moon is
closed.
14 O. What to Know about Naked-Eye Astronomy

To recapitulate, the moon's synodic phenomena are, in order,

conjunction
first visibility
first quarter
full moon or opposition
last quarter
last visibility
conjunction
As said, the term synodic month is applied to the time interval
from one synodic phenomenon to the next synodic phenomenon
of the same kind. But there is a certain difference to which
I ought to draw attention. If the phenomenon is a syzygy-
conjunction or opposition-a specific synodic month will be
some fraction of a day in excess of 29 days; just how much will
depend on various factors, as we shall see. But when the phe-
nomenon is one that, like first visibility, can happen only at a
certain time of day, here sunset, any synodic month must be a
whole number of days-for first visibility either 29 or 30 days. In
either case the average length of the synodic month will, of
course, be the same, namely, 29.5309 days.

Lunar Orbit and Nodes

In the preceding discussion we ignored that the moon's orbit is
inclined to the ecliptic, for the inclination is so small that it
plays no significant role in a qualitative description of the
moon's synodic phenomena. However, where the moon is in its
inclined orbit can be decisive when one wishes to determine, for
example, precisely on which evening the moon will first become
visible, and for the prediction of eclipses it is absolutely essen-
tial to know how far away from the ecliptic the moon is.
The lunar orbit is, then, a great circle on the celestial sphere
which is inclined to the ecliptic at an angle of ca. 5°. The two
diametrically opposite points of intersection between the eclip-
tic and the moon's orbit are called the lunar nodes; the one
where the moon crosses the ecliptic from the south to the north
is called the ascending node, the other the descending node.
The first difficulty is that the nodes do not remain fixed among
the fixed stars, but have an appreciable retrograde motion of
not quite 2° per synodic month (this motion does not affect the
 Eclipses 15

orbit's inclination). Consequently, if the moon , a fixed star, and

the ascending node happen to coincide on the celestial sphere at
a certain moment, the moon will return to the ascending node
almost 3 hours before it reaches its nearest approach to the fixed
star, for the node has moved so as to meet it earlier. The period
of the moon 's return to a given node is called the draconitic (or
dracontic, or nodical) month and it is a trifle more than 27 ~ days
in length.
Some 7 days after the moon has been at the ascending node, it
will be as far north of the ecliptic as possible, namely 5°; after a
further 7 days or, more precisely, after in all half a draconitic
month, the moon will be at the descending node; after still
another 7 days the moon will be as far south of the ecliptic as it
can get; and after the lapse of a full draconitic month it will ,
once again, be at the ascending node.

Eclipses
A lunar eclipse happens when the earth intervenes between the
sun and the moon and deprives the moon of the sun's light or, in
other words, when the moon enters the earth's shadow. For this
to take place two conditions must surely be satisfied: First, sun
and moon must be seen in opposite directions by an observer on
the earth, that is, the moon must be in opposition, or be full ;
second, since the line from the observer to the sun is always in
the ecliptic, and since sun, earth (observer), and moon must be
in a straight line, the moon must be at, or near, a node.
A lunar eclipse is, of course, a real eclipse in the sense that
anyone who can but see the moon at the time-even an observer
on the moon itself-will notice that the moon is deprived of the
sun's light. For observers on the earth, this means simply that if
only the eclipse occurs during their night, they should be able to
see it.
A solar eclipse happens when the body of the moon intrudes
between the observer and the sun. The character of a solar
eclipse is thus quite different from that of a lunar eclipse. A
solar eclipse is a more subjective phenomenon; for an observer
somewhere in space it is signalled not, of course, by any change
in the aspect of the sun, but by the appearance on the lighted
surface of the earth of a black spot, the shadow cast by the
moon. The conditions that ensure that a solar eclipse will occur
16 O. What to Know about Naked-Eye Astronomy

at a given locality are, then, first the necessary one s that sun
and moon are in conjunction, that the moon is sufficiently close
to a node for its shadow to hit the earth, and that this happens
while it is day at the place of observation. In addition, one must
ascertain that the locality lies in the path of the moon's shadow.
To do this, one must know, at least, the shape of the earth and
the relative sizes and distances of sun, moon , and earth. If I may
introduce a historical remark, Ptolemy's Almagest (ca. A.D. 150)
is the earliest surviving work in whi ch t his problem is rea-
sonably dealt with and on the ba sis of which one may predict a
solar eclipse for a particular place with confidence. While the
Babylonian texts yield solid predictions of lunar eclipses , they
serve only to present necessary, but not sufficient, conditions for
solar eclipses or , if you wish, to issue warnings, but not predic-
tions, of solar eclipses.
The simple, necessary condition for solar as well as lunar
eclipses is then that the moon at syzygy be near a node. A rough
estimate of how often this happens is readily reached. Let us
assume that a syzygy of a certain kind, say, a conjunction of
sun and moon, takes place at a node, sa y, the as cending one. The
moon will reach that node again after the lapse of a dr aconitic
month, which, on the average, is 27.21 day s. However, the moon
will not catch up with the sun again until after a full synodic
month, which, in the mean, is 29.53 days. At the next conjunc-
tion the moon will then be 2.32 da ys' travel past the ascending
node. At the second conjunction, the moon will be twice that
distance from the ascending node, and so on. At the sixth con-
junction after the original one, the moon will be 6 x 2.32 = 13.92
da ys' travel from the ascending node; how ever, since h alf a dra-
conitic month, or 13.61 days, brings the moon from the ascending
to the descending node, the sixth conjunction will take place
just pa st the descending node. Thus, six synodic months bring
a conjunction-or an opposition-of sun and moon from on e
node to a little pa st the other, and eclipse possibilities will then
happen at intervals of mostly six, but occasionally of only five,
synodic months. An interval of onl y five months mus t intervene
at times since the six-month interval is clearly a little too long
and since, further, experience shows that t wo eclipses for a
given locality never take place only one month apart. (Modern
canons do show solar eclipses at an interval of onl y one month,
but in general on e of these will be visible only near the earth' s
North Pole and the other only near its South Pole.)
 Coordinate Systems and th e Zodiac 17

It is then proper to issue eclipse warnings at intervals of

mostly six , but occasionally of only five, months; what one can
guarantee is not, of course, that eclipses will happen at these
syzygies, but rather that eclipses will not occur at an y other.
This holds for both solar and lunar eclipses.

Coordinate Systems and the Zodiac

So far I have tried to keep the discussion of spherical astronomy
as qualitative as possible, but as the need arises to become more
precise and quantitative, we must introduce coordinates.
The basic device for fixing a position on a spherical surface
by a pair of numbers, or coordinates, is well known from geog-
raphy, where a location on the earth's surface is identified by
latitude and longitude. It is a great circle (see Figure 2) with a
fixed point F on it and its two poles, PI and P 2 • The coordinates
of a point 8 on the sphere's surface are now found in the fol-
lowing fashion. Through the poles PI and P 2 and th e point 8 is
drawn a great circle that intersects the basic great circle at the
point 8' (of the two possibilities, 8' is the one for which the arc
8'8 is less than 90°). The two coordinates (x,y ) of 8 are then
simply

F IGURE 2.
18 o. What to Know about Naked-Eye Astronomy

x=FS' ,
y=S'S,
where both arcs are measured in degrees. To avoid ambiguity in
the measure of the arc FS', the basic great circle is usually pro-
vided with a positive direction and FS' can then assume values
up to 360°. One may, as an alternative, choose to keep the values
of FS' below 180° and then indicate which way the arc is mea-
sured [this happens for terrestrial longitudes given in degrees
«180°) east or west of Greenwich]. For the other coordinate, the
arc S'S, already less than 90°, one must indicate whether S lies
toward PI or P z from S'. This may be done by a sign; say, posi-
tive toward PI and negative toward P z, or by words such as
north and south (as for terrestrial latitudes).
I shall mention here three coordinate systems on the celestial
sphere. In the first the basic great circle is the ecliptic, the fixed
point F is the vernal equinox (i.e., one of the points of inter-
section between the equator and the ecliptic), and the positive
direction on the ecliptic is that of direct motion (i.e., toward the
east). The coordinate measured on the ecliptic is called celestial
longitude, or simply longitude, when no confusion with terres-
trial longitude is likely. It is usually designated with the letter A.
The other coordinate, measured perpendicularly to the ecliptic,
is called celestial latitude or simply latitude, with a similar cau-
tion as for longitude. It is positive for points north of the eclip-
tic, and negative for the other hemisphere. It is usually denoted
by the letter (3.
This ecliptic coordinate system is without question the most
important in ancient astronomy; it is still used for problems
involving sun, moon, and planets, for they are all on or near the
ecliptic or, in other words, they have celestial latitudes that are
always small «(3 = +10 ° and _10 ° are the extremal values for
Venus, the planet showing the largest latitudes).
The longitude of the sun, whose latitude is always 0, increases
by 360° in the course of a year. Direct motion of a planet means
that its longitude increases; retrograde motion corresponds to
decreasing longitude. The synodic arc of a planet-s-zs.x, as we
have already called it-is the increment in the planet's longi-
tude from one synodic phenomenon to the next of the same kind.
While we count longitudes from 0° to 360 the classical
0
,

manner of giving longitudes is in terms of a zodiacal sign and

degrees within that sign. The ecliptic in this system is divided
 Coordinate Systems and the Zodiac 19

into 12 parts of exactly 30° each, beginning with the vernal

equinox, and they are named thus:
I Aries: 0° ~ A < 30° ~ Libra: 180° ~ A< 210°
ti Taurus: 30° ~ A < 60° m, Scorpio: 210° ~ A< 240°
I Gemini: 60° ~ A < 90° I Sagittarius: 240° ~ A< 270°
o Cancer: 90° ~ A < 120° 1; Capricorn: 270° ~ A< 300°
)l Leo: 120° ~ A < 150° = Aquarius: 300° ~ A< 330°
'I1J'Virgo: 150° ~ A < 180° x Pisces: 330° ~ A< 360°
and I have affixed to them their now-standard sigilla. Thus, to
give an example, Leo 27° is simply another way of denoting the
longitude 147°.
The Babylonian zodiac does not agree exactly with this de-
scription; at the period, from which most of our texts come, the
signs begin some 5° earlier than in the preceding list. We shall
return to the reason for this later.
The longitude of the sun at vernal equinox is then Aries 0° or
A = 0°; at summer solstice Cancer 0°, or A = 90°; at fall equinox
Libra 0°, or A = 180°; and at winter solstice Capricorn 0°, or
A = 270°.
I shall briefly mention two more coordinate systems. The first
has as its basic great circle the celestial equator whose poles are
the north and south poles, and the fixed point F is again the
vernal equinox. The first coordinate is called right ascension
and is measured from the vernal equinox toward the east so
that the sun's right ascension always increases. Right ascension
is usually denoted by the letter Q. The other coordinate, mea-
sured perpendicularly to the equator, is called declination and
is counted positive north and negative south of the equator.
Declination is usually denoted by the letter 8.
The sun's right ascension at the four cardinal positions just
mentioned is, respectively, 0°, 90°, 180°, and 270°; that is, its right
ascension here is the same as its longitude, which, of course, is
not so elsewhere. The declination of the sun at equinox is 8 = 0°,
at summer solstice 8 = 23 f, and at winter solstice 8 = - 23 f,
where the value 23 f is the inclination of ecliptic to equator.
In the equator and the ecliptic systems, the coordinates of
fixed stars will not change appreciably in the course of even
several years; that they change at all is largely due to the pre-
cession of the equinoxes, which we will describe later. The lon-
gitudes, latitudes, right ascensions, and declinations of the other
celestial bodies change at a moderate rate, most swiftly for the
20 O. What to Know about Naked-Eye Astronomy

moon, whose longitude grows by about 13° per day, while solar
longitude increases about 1° per day. The daily rotation has no
effect in these coordinate systems at all.
That is not so in the last system I shall introduce. Here the
horizon is the basic great circle whose poles are zenith and na-
dir, and the fixed point F is the south point. The first coordinate
is called azimuth and is usually counted from the south point
westward from 0° to 360°. It is denoted by Az. The other coordi-
nate, measured perpendicularly to the horizon, is called the
altitude and is denoted by h. It is counted positive for objects
above the horizon, and negative for those below.
The horizon system is a local system; it belongs to the observer
and his or her terrestrial surroundings. The altitude-what I
earlier called elevation-of the north pole is 1, the terrestrial
latitude of the observer, and the azimuth of an object not too
close to a pole changes about 360° in one day due to the daily
rotation.

Precession, Anomalistic Periods, and Various Other

Refinements
In this section I deal with various refinements-long-term
changes, and variation in velocities-which I have hitherto ig-
nored for the sake of simplicity of presentation.
The first of these is the precession of the equinoxes, or simply
the precession. Its effect is most easily described in the ecliptic
system: The precession causes the longitude of all fixed stars to
increase at a uniform, very slow rate of l Oin approximately 72 i
years, while their latitudes remain constant.
Of course, this phenomenon can be, and often is, described
in a different manner. Since longitudes are measured from the
vernal equinox, one can just as well say that the vernal equinox
moves in the retrograde direction-it precedes-among the
fixed stars along the ecliptic-a great circle that remains fixed
among the fixed stars since their latitudes do not change. Thus,
the equator, which serves to define the vernal equinox, moves
slowly among the fixed stars but in such a fashion that its incli-
nation to the ecliptic remains the same; the north pole travels
slowly on a small circle of radius 23f and the northern pole of
the ecliptic as its center.
 Precession, Anomalistic Periods, and Various Other Refinements 21

Right ascension and declination of a fixed star will therefore

both change, so it is more complicated to describe the effects of
the precession in the equatorial than in the ecliptic system.
One consequence of the precession of the equinoxes is that we
should distinguish between two kinds of year. One is the period
of the sun's return to a fixed star, and it is called the sidereal
year ; the other is the period of the sun's return in longitude (i.e.,
to the vernal equinox) and, since this governs the seasons, it is
called the tropical year. The latter is slightly shorter than the
former for the sun has to travel slightly farther to catch up with
a fixed star than with the vernal equinox, which moves slightly
to meet it during each of its revolutions. Values of the two are
1 sidereal year = 365.256 days,
1 tropical year = 365.242 days.
Similarly, we distinguish between a sidereal and a tropical
month, the periods ofthe moon's return to a fixed star and to the
same longitude, respectively. Their values are
1 sidereal month = 27.32166 days,
1 tropical month = 27.32158 days.
I must most emphatically warn that my inclusion of precession
in this presentation does not imply that it plays a role in Bab-
ylonian astronomical schemes. There is no evidence whatever
that the phenomenon of precession was known to Babylonian
astronomers; all indications are that their longitudes are side-
real longitudes-that their zodiac was fixed among the fixed
stars, that their year was the sidereal year, and that they were
unaware of any difference between this and the seasonal or
tropical year. It was Hipparchus (ca. 150 B.C.) who first drew the
distinction between the tropical and the sidereal year or who , in
other words, discovered the precession. I introduced the pre-
cession in this presentation to make clear the natural discrep-
ancy between Babylonian and modern longitudes, a discrepancy
that increases with time at the rate of 1 in 72 to 73 years.
0

Anomalistic Periods
In the previous discussion of the motions of sun and moon
among the fixed stars, I have been content to give crude esti-
mates of average daily progress. In a more precise account one
22 O. What to Know about Naked-Eye Astronomy

must recognize that the velocity-which in spherical astronomy

means, as it must, apparent angular velocity-of either body is
not constant, but changes periodically by sensible amounts.
The sun's daily progress varies between some 1.02° (or 1°1')
and 0.95° (or 57') in a regular manner; the period of this varia-
tion, that is, the time interval from high velocity to high veloc-
ity, or from low velocity to low velocity, is called one anom-
alistic year. Its length is
1 anomalistic year = 365.2596d
(where d = days) which is very close to the value of the sidereal
year. Within the historically relevant period we may ignore this
difference and shall, with the Babylonian astronomers, consider
solar velocity as a function of Babylonian, namely sidereal,
longitude.

r
For the moon the situation is quite different. Its daily progress
varies between some 15° and 11 (it makes no sense to present
sharper limits here), and the period from high velocity to high
velocity, called the anomalistic month, is
1 anomalistic month = 27.5545d .
This is substantially more than either the sidereal or the
tropical month-by some 6 hours-and the anomalistic month
was recognized as a separate and important parameter in Bab-
ylonian lunar theory, as we shall see. Abandoning for the mo-
ment the conventions of spherical astronomy, I may state that
the moon assumes its least velocity when at the point of its orbit
farthest from the earth, or at its apogee, and its greatest velocity
when nearest, or at its perigee; that the anomalistic month is
longer than the tropical month means, then, that the apogee and
the perigee of the moon both advance steadily in the ecliptic.

Synodic Arcs of the Planets

A quantitative investigation of the synodic arc .6.>' of a given
planet (i.e., the increment in the planet's longitude when it
moves from one synodic phenomenon to the next of the same
kind) will show that it is not constant but varies according to
where in the ecliptic the planet happens to be. To say this in
other words, .6.>' is a function of >., the longitude at which the
arc .6.>' begins. That this is correct when longitudes are counted
sidereally as in Babylonian astronomy can be established on the
 Pr eces sion, Anomali stic Peri ods, a nd Various Other Refinements 23

ba sis of modern planetary theory, but I shall not do so here,

though the necessary arguments are fairly simple.

Long- Term Changes

There are sever al long-term, or exceedingly slow , changes in
the various parameters I have mentioned, in the lengths of the
several kinds of month, in the length of the year, in the amount
of the precession, in the r elative positions of fixed stars, in the
places where synodic arcs achieve their maxima, and so on.
These minute variations, some of which are called secular since
they are only felt after the lapse of centuries, are of prime inter-
est to astronomers. I shall nonetheless pass them by, for in the
time interval of Babylonian astronomy, and within its margin of
precision, they play no significant role. The causes of some of
these very slow changes are still not quite under control; indeed,
some of their magnitudes are not yet unequivocally established.
Ancient observations, on ce properly understood and properly
treated, must play a centr al role in these matters, but our
understanding of ancient theoretical astronomy clearly does not
depend on such issu es at all.
1
Babylonian Arithmetical
Astronomy

In late May 1857 a committee, appointed by the Royal Asiatic

Society of Great Britain and Ireland, met in London to compare
four independent translations of an Assyrian text inscribed in
cuneiform characters in duplicate on two well-preserved clay
cylinders. Hormuzd Rassam had found them as foundation de-
posits in the ruins of ancient Ashur in 1853 when he was digging
in Mesopotamia on behalf of the British Museum. W. H. Fox
Talbot, the gentleman scientist, inventor of photography, and
linguist, had been given a copy of the text by H. C. Rawlinson,
the remarkable soldier, diplomat, and linguist, and sent his
sealed translation of it to the Society with the suggestion that
other scholars be invited to translate the same text so the results
could be compared to test the validity of the decipherment of
Assyrian for, as he writes, "Many people have hitherto refused
to believe in the truth of the system by which Dr. Hincks and
Sir H. Rawlinson have interpreted Assyrian writings, because it
contains many things entirely contrary to their preconceived
opinions."*
The Society chose the Reverend Dr E. Hincks, Ireland, Lieut.-
Col. Sir Henry C. Rawlinson, London, and the Orientalist Dr. J.
Oppert, who happened to be in London on leave from Paris. The
three sent in their translations of the text-the annals of the
Assyrian king Tiglath-Pileser I (l l th century B.C.) as it turned

*The committee's reports and the four translations are published in the
pamphlet: "Inscription of Tiglath Pileser I., King of Assyria, B.C. 1150,
as translated by Sir Henry Rawlinson, Fox Talbot, Esq. , Dr . Hincks,
and Dr . Oppert." Published by the Royal Asiatic Society. London,
1857. And again in : Journal of the Royal Asiatic Society, Vol. XVIII,
1861, pp. 150-219.

24
A. Aaboe, Episodes From the Early History of Astronomy
© Springer-Verlag New York, Inc. 2001
 1. Babylonian Arithmetical Astronomy 25

out to be-and the appointed committee opened and compared

them, including Fox Talbot's. In the words of a report dated May
29, 1857, "Having gone through this comparison, the Examiners
certify that the coincidences between the translations, both as to
the general sense and verbal rendering, were very remarkable."*
This dramatic event marks the end of the heroic epoch of the
decipherment of cuneiform script by the great amateurs, as they
had to be, and Assyriology had been established as a scholarly
discipline. Much hard and patient work with the cuneiform texts
lay ahead before Akkadian, as the Semitic language of the texts
is called, took its place beside Hebrew, Greek, Latin, and Egyp-
tian as one of the well-understood classical languages, but the
correct path had been found.
The texts are clay tablets, almost all unbaked, inscribed
with signs, each of which is a cluster of wedges impressed in
the wet clay with a sharpened stylus-"cuneiform" means
"wedge-shaped." The difficulties with the system of Hincks and
Rawlinson, as Fox Talbot called it, were two: They had been
forced to assume, first, that most signs could be read either as an
ideogram-an entire word-or as a syllable and, second, that in
neither case was there a one-to-one correspondence between
signs and readings. I shall give but one example. The sign <::::is
usually transcribed in the astronomical texts as ge6. As an ideo-
gram it may have one of the obviously related meanings: to be
black, black, and night. In Akkadian it would then be pro-
nounced saliimu (or declined form) , salmu, and musu, respec-
tively. However, as a syllable it is sixth in a list of 27 signs, each
of which can be pronounced gi or ge, and it may also have the
phonetic values mi, me, and sil . When one further realizes that
there are more than 500 distinct signs, one can readily perceive
the formidable obstacles both to the execution of the decipher-
ment and to its acceptance, but Hincks and Rawlinson were
right.
Three groups of texts will be of particular interest in what
follows . The first came to the British Museum in the early 1850s
from excavations in Kuyunjik-the ancient Ninive-in two lots.
The first was found by Layard in the southwest palace, and the
second by Rassam in the north palace, in the room, incidentally,
that was decorated with the famous reliefs of Assurbanipal's
lion hunt. Together they form the royal library-or Assurbani-
pal's library, as it is often called. It was a stroke of luck that this
large body of texts came to light this early, for it contains, often
26 1. Babylonian Arithmetical Astronomy

in standard Assyrian script, many copies of much older texts of

many varieties-e-Assurbanipal (668-627 B.C.) was an avid col-
lector. The bulk of the texts centered on the series Enilma Anu
Enlil of celestial omens comes from this library.
Second, in the last decades of the last century, texts began
arriving in the British Museum by the tens of thousands. They
were bought from dealers who, in turn, obtained them through
unscientific excavation. Among them is a group of some 1500
texts that come from what must have been an extensive astro-
nomical archive somewhere in the city of Babylon. Approxi-
mately 1200 of these are nonmathematical astronomical texts,
classified by A. Sachs as diaries, almanacs, and goal-year texts.
The rest are concerned with mathematical astronomy, and I
shall refer to them as being of the "A CT type" after the standard
abbreviation for O. Neugebauer's Astronomical Cuneiform
Texts, where most of them were published. I should add that
quite a few texts from the astronomical archive found their way
to other collections in Europe and America.
Last, there is a smaller group of texts, mostly of the A CT type,
that were found in Uruk, very likely within the domain of the
Res sanctuary. The largest single set of fragments from this site
is in the Archaeological Museum in Istanbul-Iraq was part
of the Turkish Empire before the first World War-but many
others, among them numerous potential joins with Istanbul
fragments, leaked from the excavations and ended up in many
other museums.
In respect of content we have, then, three principal classes of
astronomical texts. First there are those that in one way or an-
other are concerned with astronomical omens. At the core of
this class is the series of about 70 tablets now, as in antiquity,
known by its incipit Eniima Anu Enlil ["when (the gods) Anu
and Enlil"] , The contents of this collection of omens are pre-
sumably very old, but only a few Old-Babylonian (i.e., from the
early second millennium B.C.) fragments of some of the tablets
are known. The two-tablet series Mul Apin is more astronomical
in character and very likely younger than Eniima Anu Enlil.
Finally, there are the royal reports written by specialists in
astronomical omens from various cities of the kingdom to the
Assyrian king. Such a report may contain an observation and an
interpretation of its significance according to the canonical
texts, mostly Eniima Anu Enlil. These reports date from about
700 B.C.
 1. Babylonian Arithmetical Astronomy 27

The second class, that of the astronomical diaries and related

texts, contains or is based on a high proportion of observations.
They come from the astronomical archive in Babylon and span
in time the interval from about 750 B.C. to A.D. 75-the text from
this last year happens to be the latest datable text written in
cuneiform.
Third, the mathematical astronomical texts, those of the ACT
type, come from astronomical archives in Babylon and from
Uruk. They are, in respect of date, mainly, from the last four or
five centuries B.c.-the Uruk texts stop already around 150 B.C.
-so they represent one of the last, as well as one of the finest,
contributions of Mesopotamian culture.
In the following sketch of Babylonian astronomy, I shall con-
centrate on the third kind of texts, but the two groups of non-
mathematical texts cannot well be completely excluded from a
discussion of Babylonian astronomy, for they are connected both
to each other and to the ACT texts. Indeed, all three categories
of texts were kept in the archive in Babylon; the scribes who
wrote even the elaborate theoretically computed ephemerides of
ACT called themselves by the title tupsar Eniima Anu Enlil
["scribe of (the series) Eniima Anu Enlil"] ; the diaries can be
viewed as collections of raw material for omens, and they pro-
vided in the process the observational basis for constructing
the theories behind the ACT texts; and, finally, the ACT texts
predict precisely the core of the celestial phenomena recorded in
the diaries.
Last, a few remarks about the astronomical archive in
Babylon. Colophons, names of scribes, scribal families, runs of
the British Museum's accession numbers, and many other fea-
tures ensure its existence, but we do not know its precise loca-
tion within the vast ruins of Babylon, for all the texts from it
were excavated without records. The earliest diary is dated
by A. Sachs to -651, and he published six texts from A.D. 31 to
A.D . 75. These late texts throw a particularly interesting light
on the activities around the archive. Already in 275 B.C. the
government had moved from Babylon to Seleucia and Antiochus
ended Babylon's civil existence. The once great and glamorous
metropolis fell rapidly into decay, and some three centuries later
Strabo, who died circa A.D. 20, writes that the greater part of
Babylon is deserted and quotes the comic poet who said, "The
Great City is a great desert." Yet these six texts show that even
after Strabo's time there were still people living in the ruins of
28 1. Babylonian Arithmetical Astronomy

Babylon who not only knew, and taught others, the difficult art
of reading and writing technical Akkadian in cuneiform, but
who also had access to the astronomical archive and the desire
and competence to use and increase it. Here one may well recall
that Pliny (d. A.D. 79) in Nat. Hist . VI, xxx, 121-122, remarks
about Babylon, "The temple of Jupiter Belus still remains-
it was here the creator of the science of astronomy was-the
rest has reverted to desert." Indeed, many of the astronomical
texts from Babylon carry the invocation ina amat Bel u Beltia
lislim ["at the command of (the deities) Bel and Belti, may it
go well"].
The study of astronomical cuneiform texts began already in
the 1870s. The Orientalist P. Johann N. Strassmaier, S. J., was
then engaged in classifying and copying tablets in the British
Museum's collection, which was growing enormously in pre-
cisely those years. In the matter of texts that he deemed astro-
nomical he sought the advice of his fellow-Jesuit, the mathema-
tician and astronomer P. Joseph Epping.
Epping brilliantly and swiftly came to understand much of the
basic structure and terminology of Babylonian astronomy, and
he succeeded in correlating precisely Seleucid and Western
chronology. He summed up these early researches in a charming
little book, Astronomisches aus Babylon (1890), which still con-
veys a sense of the excitement of his pioneering efforts.
After Epping's death in 1894, his work was continued by
P. Franz Xavier Kugler, S. J., at first also in collaboration
with Strassmaier. Kugler addressed his task with rare vigor
and imagination, and he published his many results in volumes
entitled Die babylonische Mondrechnung (1900) and Sternkunde
und Sterndienst in Babel I (1907) and II (1909-1924) with three
supplements, the last of which Fr. Johann Schaumberger wrote
in 1935.
In his editions Kugler included astronomical texts of all kinds.
In the mid-1930s Otto Neugebauer selected those dealing with
mathematical astronomy to subject them to a systematic and
uniform treatment. Later he was given a copy of Strassmaier's
notebooks, which led to the identification of a large number of
new texts in the British Museum, and many texts from other
collections-those in Istanbul, Chicago, the Louvre, Yale, and
Berlin among them-came to his notice, so it was some 20 years
before he published his Astronomical Cuneiform Texts (A CT)
 General Background 29

(1955). It contains editions of all the texts of his chosen kind

known at that time, about 300 in number. Neugebauer included
for the sake of completeness some 50 previously published texts,
many of them augmented by joins.
Shortly before the appearance of A CT, A. Sachs got the op-
portunity to survey part of the collection of tablets in the British
Museum in order to identify astronomical texts. The result was
overwhelming. Before his search Sachs had been aware of 12
nonmathematical astronomical texts-astronomical diaries and
related texts-but now he had 1200, and they are still in the
process of publication. Further, Sachs was given a large number
of beautiful hand copies executed by the remarkable T. G.
Pinches. They had lain unnoticed in the British Museum for half
a century until Sachs published them with his comments and
datings as Late-Babylonian Astronomical and Related Texts
Copied by T.G. Pinches and J .N. Strassmaier (LBAT). Finally,
during his survey Sachs discovered some 200 texts concerning
mathematical astronomy. Many of these have been analyzed
and published in various journals, mostly by Sachs, Neugebauer,
and myself. Among the contributors to our further and deeper
understanding of already published texts, I must particularly
mention B. L. van der Waerden, A. Pannekoek, P . Huber,
J . Britton, and N. T. Hamilton.

General Background

Number System
Babylonian mathematics appears fullblown in Old-Babylonian
texts (from the early second millennium B.C.) and changes little,
if any, afterward. It deals principally with problems that we
today would classify as belonging to arithmetic, number theory,
and elementary algebra, although it also deals with certain
geometrical problems, some of them involving the use of the
Pythagorean theorem. It is based on a most efficient means of
writing numbers, a place-value system of base 60-thus we call
it sexagesimal-with many similarities to our decimal system.
The single digits are written as combinations of two signs: a
vertical wedge meaning 1, and a corner wedge meaning 10. The
numbers from 1 to 59 are written thus:
30 1. Babylonian Arithmetical Astronomy

r IT ffr v . .". .lJJ < -r . .« . .«<. .4t . .~. .4*

2 3 4 . . 7 . . 9 10 11 . . 2JJ . . 30. . 40. . 50 . . 59

(there is a late cursive form for 9 consisting of three diagonal

wedges).
The sexagesimal system uses these digits to express all whole
numbers, as well as certain fractions, by assigning importance to
a digit's place within a sequence of digits: Moving a digit one
place to the left increases its value 60 times, and moving it one
place to the right means to divide by 60, even beyond the units'
place. When we transcribe Babylonian numbers, we separate the
digits by commas so 1,25,30 can mean
1 . 602 + 25 . 60 + 30 = 5130.
However, the ancients did not indicate the units' place, nor did
they employ terminal zeros, so this sequence of digits could, in
principle, stand for 5130 multiplied by any power of 60, positive
or negative. In practice, there is rarely any doubt about the
absolute size of such a number, and when from context we have
determined the units' place, we separate the whole from the
fractional part with a semicolon (analogous to our decimal
point) so that
30 1
1,25;30 = 1· 60 + 25 + 60 = 85 ,
2
I should repeat for emphasis that the semicolon has no equiva-
lent in the texts but is solely introduced for the sake of conve-
nience in our translations.
In the late astronomical texts the separation sign ~ (which we
transcribe as ".") is used to indicate an empty sexagesimal place,
much like our zero, and also to help the reader distinguish, for
example, between 10,3 (written 10,.3) and 13.

Calendar
The Babylonian calendar is strictly lunar: A new month begins
on the evening when the crescent of the new moon becomes vis-
ible for the first time. A month contains either 29 days (hollow)
 General Background 31

or 30 days (full); on the average its length is the mean synodic

month, which is slightly more than 29! days. In the long run
there are then slightly more full than hollow months. I should
emphasize that the sequence of full and hollow months shows no
simple pattern. The ACT material contains several instances of
four consecutive full months; in one of the years in question we
also find three consecutive hollow months (ACT, p. 94). Indeed,
it is a very complicated matter to predict the visibility of the
new moon, and it is one of the great triumphs of Babylonian
lunar theory that it succeeded in doing just that so well.
A normal year consists of 12 lunar months, but since they
amount to only some 354 days, an extra month was occasionally
introduced in order to keep the year in step with the seasons.
From about the mid-fifth century B.C., these intercalations
followed a rigid scheme based on the so-called Metonic cycle,
which equates 19 years with 235 months. Every group of 19 years
contained seven years with intercalary months, six of them
inserted as a second month XII (we denote it Xlh) and one as a
second month VI (Vh), and the pattern is fixed modulo 19 years.
In our transcriptions we usually mark a 13-month year contain-
ing an XII 2 with a single asterisk, and with a double asterisk a
year with a V12 • The Babylonian year began near, and mostly
after, vernal equinox.
The Akkadian names of the months, and the late texts' ideo-
grams for them, are as follows:
I Nisannu bar VII Tasrftu dU6
II Aiaru gU4 VIII Arahsamna apm
III Simanu sig IX Kislimu gan
IV Dtizu su X Tebetu ab
V Abu IZI XI Sabatu ZIZ
VI Ulillu kin XII Adaru se
VI 2 kin-z-kam, kin a XII 2 dir-se, dir, a
In earlier times a date is given by regnal year of the reigning
king, month, and day, but in 311 B.C ., with the official reign of
Seleucus, once Alexander's general, began what became a con-
tinuing count of years. The Seleucid Era (S.E.), as it is called,
remained in use throughout the ACT material.
The continuing year-count of the Seleucid Era and the fixed
intercalation scheme are, of course, very useful when one wishes
to make astronomical predictions. However, I cannot tell how
32 1. Babylonian Arithmetical Astronomy

one in pre-Seleucid times would indicate a date far in the future

for I do not know of a single text that gives a regnal year number
in excess of the natural reign of the king.. Texts that date cor-
rectly in the reigns of several kings can obviously not have been
written in advance of the events they describe, at least not in
their entirety.

Units
In a certain sense the basic unit for measuring time in Bab-
ylonian astronomy is the synodic month, and this is intimately
connected with the calendar. For shorter time intervals the day
is employed, divided into 360 time-degrees (us).
In the planetary theory we find employed an artificial "day" ,
which is one-thirtieth of a synodic month; we call it lunar day
or tithi (T), for in modern times it was first encountered under
that name in Indian astronomy. The terminology of the available
texts draws no distinction between " day" and "tithi" ; both are
indicated by U4 or me.
Babylonian astronomy is principally concerned with the
moon, the sun, and the planets, all on or close to the ecliptic,
so it is not surprising that the most important celestial coor-
dinate is (celestial) longitude, >.. In the late texts we find the
ecliptic divided into 12 zodiacal signs, each of length precisely
30 degrees (us)-our modern degree is of Babylonian origin. The
standard sigilla and Latin names of the signs, and the corre-
sponding ideograms used in the texts are, in order,
'r Aries bun, lu ~ Libra rin
t; Taurus mul 'I11t Scorpio gir-tab, gir
n Gemini mas-mas, mas I Sagittarius pa
e» Cancer allax (KUSU) 1; Capricorn mas
~l Leo a := Aquarius gu
'I1)t Virgo absin, absin, (KI) x Pisces zib-me, zib
The zodiacal signs are, of course, derived from constellations
near the ecliptic. We are not sure when the important step of
replacing constellations by 30 segments was taken-it may
0

have been in the fifth century B.C. The normalized Babylonian

zodiac is sidereally fixed; thus, the systematic difference between
positions computed according to Babylonian and modern
theories changes with time at the rate of precession.
The degree is the standard Babylonian unit for expressing
 General Background 33

time as well as arcs. There are other measures of arc-cubit and

barleycorn, to name but two-but they will be introduced when
required.

Celestial Omen Texts

An omen consists of two parts: a protasis and an apodosis. Thus,
the first of in all 59 omens in the Venus Tablet of Ammisaduqa
(Tablet 63 of the series Eniima Anu Enlil) reads
In month XI, the 15th day, Venus disappeared in the west, it stayed
away 3 days, and in month XI, the 18th day, Venus became visible in
the east: springs will open, Adad will bring his rain, Ea his floods; king
will send messages of reconciliation to king.

Here the protasis is written in a past tense, the apodosis in a

future tense. It is reasonable to assume that the entire omen
would become a historical statement if the apodosis were
changed to the past tense. This assumption underlies the role
the Venus text has played in Old-Babylonian chronology. Kugler
realized that the tenth omen is incomplete and that instead of an
apodosis it contains what he brilliantly read as the name of the
eighth year of Ammisaduqa's reign, Year of the Golden Throne .
The omens in Sections I and III of Tablet 63 were consequently
taken to contain an observational record for the 21 years of
Ammisaduqa's reign and served to limit, on astronomical
grounds, the absolute date of the beginning of Hammurapi's
dynasty to a few possibilities in the early second millennium B.C.
The sequence of full and hollow months is highly irregular, and
so contains much information. By ingenious and sophisticated
use of this information, in addition to the Venus data, P. Huber*
has convincingly shown that only the "Long Chronology" with
Ammisaduqa 1 = -1701 = 1702 B.C. makes sense, so Hammurapi
began his reign in 1848 B.C., unless the Venus Tablet is alto-
gether rejected as chronological evidence.
Section II is a highly schematic interlude based on a crude
subdivision of Venus's mean synodic period. The date of this
scheme is moot. Section IV consists of a rearrangement of the
omens in Sections I and III in order of months.

* Peter J . Huber, Astronomical Dating of Babylon I and Ur III. Malibu,

1982
34 1. Babylonian Arithmetical Astronomy

The Venus Tablet ends with a catch line-the first line of

Tablet 64-saying, "If Jupiter remains (in the sky) in the morn-
ing, enemy kings will become reconciled."
The catch line is, alas, much more typical of Enilma Anu Enlil
than Sections I and III, for the protases of the many thousands of
omens in the series are in general qualitative and contain little
significant or useful astronomical information.
According to Weidner, the 70 tablets of Enilma Anu Enlil can
be classified as follows: The first 23 concern the moon; the next
20 the sun; then follow a few tablets of meteorological omens;
and the last 20 deal with planets and fixed stars.
In the lunar protases, to take but one example, there is much
concern with the general appearance of the moon, for example,
on the important evening of its first visibility, whether it is light
or dark, which way its horns point, or whether it is surrounded
by a halo. Tablets 15 to 23 deal with similar aspects of lunar
eclipses. The apodoses here, as throughout the series, predict
events and conditions of concern to king or country: war and
peace, quality of the harvest, and bad weather, to name but a
few. This sort of astrology is called judicial in contrast to per-
sonal astrology with its horoscopes for individuals. So far we
have only encountered cuneiform evidence for personal astrol-
ogy in texts written well after the Assyrian period.
The imperfectly understood 70 tablets of Enilma Anu Enlil
are at the center of what at present seems a morass of related
texts: extracts, commentaries, and reports. Once all of this
material is brought under control, we shall very likely have a
firm grasp on the state of astrology in Mesopotamia, and the role
of " diviners," near the end of the Assyrian Empire, and we may
also be afforded some notion of the tradition of some of the texts.
Mul Apin (the Plow Star), the series of two, or maybe three,
tablets contains some omens, but it is principally a compendium
of astronomical knowledge. The treatment, theoretical if you
will, of various astronomical phenomena is always schematic in
the extreme. Thus, there is a section relating the first appear-
ances of certain fixed stars and constellations to an idealized
calendar in which a year consists of 12 months of 30 days each.
Likewise there is a part giving the length of daylight and night
for the 15th day (full moon) of each month in the same calendar.
This daylight scheme is based on a ratio of 2: 1 between longest
and shortest daylight, a ratio also found in Tablet XIV of Eniima
Anu Enlil and in a recently published text from the mid-seventh
century, though the latter text elsewhere gives the ratio as 3: 2.
 The Astronomical Diaries and Related Texts 35

In Hellenistic Greek astronomy the ratio of longest-to-shortest

daylight is a common way of denoting the terrestrial latitude of a
place. Here, following the later Babylonian texts, the canonical
ratio for Babylon (or rather for the entire Second Climate) is
3: 2, while 2 :1 corresponds to a parallel far north of Babylon's.
I must emphasize that the cuneiform texts give no evidence
whatever of an awareness that this ratio changes when you
travel north or south.
In this section of Mul Apin the length of daylight changes
from month to month by a constant amount from maximum to
minimum and back again. We see here an embryonic form of the
linear zigzag function that was to become one of the two major
devices of expressing periodic phenomena in the later texts of
the A CT type.
We cannot at present determine the dates of the various parts
of the series Mul Apin nor the tradition of the texts.

The Astronomical Diaries and Related Texts

To introduce the astronomical diaries I present part of a com-
posite diary translation from the edition by A. Sachs and H.
Hunger of all extant astronomical diaries. The tablet consists of
four rejoined fragments in the British Museum [Rm. 718 + Rm.
723 + BM 32840 (= 77 - 2 - 22,2) + BM 34130 (= Sp. 232)]; when
unbroken it gave reports for the last six months of year 60 of the
Seleucid Era (Oct. 252-Mar. 251 B.C.), but only those for months
VII, VIII, and XII are preserved. My excerpt covers month VII.
Time intervals are given in time degrees (1 = 4 min.), and the
0

frequently used cubit (kus) is here 20 , or maybe 2f, of arc.

No. -251
Obv.'
1 [Year 60,] king [Antioch]us. Month VII, the 1st (of which followed
the 30th of the preceding month), sunset to moonset: 13°, it was low.
Night of the 2nd , the moon stood 2~ cubits in front of Venus to the
west. Night of the 3rd, the moon was 1 ~ cubits in front of {) Ophiu-
chi, 2 cubits below Saturn.
2 [.... Night of the 4th (or: 5th), first part of the night, Venus was] 2
cubits [above] a Scorpii; last part of the night, Jupiter was 6 fingers
above p Leonis. Night of the 6th, beginning of the night, the moon
was [nn] cubits in front of f3 Capricorni. Night of the 8th, beginning
of the night, the moon was
36 1. Babylonian Arithmetical Astronomy

3 [(nn cubits) nn] fingers [... . 8 Capricorni.] The 11th, Mercury's first
appearance in the west, omitted. Night of the 12th, beginning of the
night, the moon was 1 cubit in front of [1]] Piscium. Night of the
13th , moonrise to sunset: 4° 10' . The 13th, moonset to sunrise: 3°
50' , measured?
4 [Night of the 14th, sunset to moonrise: 2° 20'; last part of the night,
the moon was] 1 cubit behind? 1] Tauri. The 14th, sunrise to moon-
set: 10°, measured. Night of the 15th, first part of the night, Venus
was 1 cubit above 'l3 Ophiuchi; last part of the night, the moon was
2r cubits behind 0 Tauri. Night of the 16th , last part of the night,
the moon was
5 [nn cubits behind ( Tauri. Night] of the 17th, last part of the night,
the moon was 2 cubits above I Geminorum. Night of the 18th, last
part of the night, the moon was I?! cubits in front of 01f3 Gem-
inorum. Night of the 20th , last part of the night, the moon was ....
in front of e Leonis. Night of the 21st, last part of the night, the
moon was
6 [nn cubits . . .. 0 Leonis. Night of the 22nd, last part of the night,]
the moon was 1 ~ cubits in front of'l3 Leonis. Night of the 23rd, last
part of the night, the moon was 1 cubit in front of f3 Virginis. Night
of the 25th , last part of the night, the moon was I ? cubit in front of 0
Virginis. Around the 26th, Mercury's last appearance in the west,
omitted.
7 [The 27th ?, Mars'? first appearance? in Libra"; moonrise to sunri]se:
[20°], measured. That month, the equivalent was : barley, 1 kur 1 pan
4 slit, at the end of the month, 1 kur [x pan] 3 slit 3 qa; new dates,
3 pan; mustard, 1 kur 2 pan 3 slit; cress, 2 pan 3 slit; sesame, 1 pan
5 slit;
8 [wool, nn minas] for 1 shekel of wrought silver. At that time,
Jupiter was in Leo; Venus was in Scorpius, at the end of the month
in Sagittarius; the 11th, Mercury's first appearance in the west,
omitted; the 26th , Mercury's last appearance in the west, omitted;
9 [Saturn was in Scorpius, at the end of the month in Sagittarius; the
2]7th?, Mars' first appearance in Libra. That month, the river level
rose 4 fingers, total: 17 was the na (gauge).

A standard diary is built up of six or seven units like the pre-

ceding excerpt, each of which spans one lunar month. The prin-
cipal concern is with the behavior of the moon . Six phenomena
and time intervals are recorded every month. First we find a
statement about the first visibility of the new crescent, for this
phenomenon marks the beginning of a new month. The diary
tells us whether the previous month was full or hollow and gives
the time interval from sunset to moonset (na) . Near the middle of
 The Astronomical Diaries and Related Texts 37

the month, when the moon is close to full, four time intervals are
recorded:
su: time from moonset to sunrise
na: time from sunrise to moonset
me: time from moonrise to sunset
ge6: time from sunset to moonrise
Near month's end the last visibility of the moon is noted as
well as the time from moonrise to sunrise (kur). Otherwise the
progress of the moon during the month is recorded, particularly
its passage above or below one of the 31 standard reference stars
given in Table 1. These "Normal Stars," as they are also called,
all lie less than 10° from the ecliptic. The curious gap of more
than 60° in the signs Aquarius and Pisces still wants expla-
nation, particularly since there are several fixed stars in this
region brighter than some of the normal stars included in the
list.
The diaries include two kinds of information about the five
planets. First, it is noted when a planet passes above or below
a Normal Star, and by what amount; "above" and "below"
probably refer to ecliptic coordinates. Second, the date of each
planetary phase is recorded together with the zodiacal sign in
which the planet is at that time or, for stations, with distance to
a Normal Star. The phases, or synodic phenomena, of an outer
planet are
I' : first appearance
<1> : first stationary point
8: "opposition"
\If: second stationary point
D: disappearance
and for an inner planet
r. first visibility }
<1>: stationary point in the east (morning star)
2:: disappearance
.:::.: first visibility }
\If: stationary point in the west (evening star)
D: disappearance
where the planet is retrograde from \If to <1>.
The study of planetary phases is fundamental to Babylonian
planetary theory, and it should be noted already here that the
38 1. Babylonian Arithmetical Astronomy

TABLE 1. The standard Babylonian reference stars as they appear in

text s after approximately -300
- 600 - 300 o
~ ~ ~

Babylonian name Star ,\ (3 ,\ (3 ,\ (3

mul kur sa dur nu-nu T) Pis c 350.7° + 5.2° 354.9° +5.2° 359.0° + 5.3°
mul igi sa sa g bun (3 Arie 357.9 +8.4 2.0 +8 .4 6.2 +8.4
mul tir sa sag bun Q Ari e 1.5 + 9.9 5.7 + 9.9 9.8 + 9.9
mul-mul T) Taur 23.9 +3.8 28.0 +3 .8 32.2 + 3.8
is da Q Taur 33.7 -5.7 37.8 - 5.6 42.0 -5.6
SUR gigir sa si (3 Taur 46.5 +5.2 50.6 +5 .2 54.8 +5.2
SUR gig ir sa U x ( Taur 48.7 -2.5 52.8 -2.5 57.0 -2.5
mul igi sase-pit mas-mas T) Gemi 57.4 -1.2 61.5 -1.2 65.7 -1.1
mul ar sa se-p it mas-mas Jl Gemi 59.2 -1.1 63.3 -1.1 67.5 -1.0
mas-mas sa sipa I Gemi 63.0 -7.1 67.1 -7.0 71.3 -7.0
mas-mas igi o Gem i 74.2 + 9.9 78.4 + 9.9 82.5 + 9.9
mas-mas dr (3 Gemi 77.5 +6.5 81.6 + 6.5 85.7 + 6.5
mul igi sa alla, sa u, (J Cane 89.7 -1.0 93.8 -1.0 98.0 -0.9
mul igi sa a lla, sa si I Cane 91.5 +3 .0 95.6 + 3.0 99.8 + 3.0
mul ar sa allax sa u, {j Canc 92.6 0.0 96.7 0.0 100.9 0.0
sag A c Leon 104.6 + 9.5 108.7 + 9.5 112.9 + 9.6
lugal Q Leon 113.9 +0.4 118.0 + 0.4 122.2 + 0.4
mul tur sa 4 kus ar lugal p Leon 120.3 0.0 124.4 0.0 128.6 0.1
GIS-KUN A (J Leon 127.3 +9 .7 131.4 +9 .7 135.6 +9 .7
gir dr sa A (3 Virg 140.5 +0 .6 144.7 +0.7 148.9 +0.7
dele sa igi absin I Virg 154.4 + 3.0 158.5 + 3.0 162.6 +3.0
sa, sa absin Q Virg 167.8 -1.9 171.9 -1.9 176.1 -1.9
rin sa u, Q Libr 189.0 +0. 7 193.2 + 0.6 197.3 + 0.6
rin sa si (3 Libr 193.3 +8.8 197.4 +8 .8 201.6 +8 .7
mu l murub sa sag gir -tab {j Scor 206.5 -1.7 210.6 -1.7 214.8 -1.7
mul e sa sag gir -tab (3 Scor 207.1 + 1.3 211.2 + 1.3 215.4 +1. 3
S14 Q Scor 213.7 -4.2 217.8 -4.3 222.0 - 4.3
mul kur sa kir, sil pa (J Ophi 225.3 -1.5 229.4 -1.5 233.6 -1.6
si mas (3 Capr 267.9 +4 .9 272.1 +4.9 276.2 +4 .8
mul igi sa suhur-ma s I Capr 285.6 -2.3 289.7 -2.3 293.9 -2.4
mul arsa suhur -rnas {j Capr 287.3 -2.1 291.5 -2.2 295.6 -2.2

computed ephemerides predict with considerable accuracy not

only dates but also planetary longitudes of the phases. Yet we
saw that the diaries mention nothing but the zodiacal sign of
a planet at phase, that is , its position is given only within an
interval of length 30°. It has been shown how theoretical
schemes of such excellence may be derived from a set of such
crude observations, for I do not doubt that the diaries provided
 The Astronomical Diaries and Related Texts 39

the observational basis for Babylonian mathematical astronomy.

The argument is, in principle, that the observations make up in
quantity for what they want in quality (Aaboe [1980]).
The diaries mention whenever an eclipse is seen at Babylon.
The text BM 45745 offers a spectacular example in its report on
the solar eclipse of April 15, 136 B.C., total at Babylon, very likely
the best account of a solar eclipse from antiquity:
[Year 175 (of the Seleucid Era), month XII2 •• •J Daytime of the 29th,
24° after sunrise, a solar eclipse beginning on the south west side [. .. J
Venus, Mercury and Normal Stars were visible; Jupiter and Mars,
which were in their period of invisibility, were visible in that eclipse
[. ..J (the shadow) moved from south west to north east. (Time interval
of) 35° from obscuration to clearing up (of the eclipse). In that eclipse
the north wind which [... J.
A group of texts, falling into several classes, derive from the
diaries. Thus, we find extracts of information, either about a
particular planet or, for the moon , about eclipses, usually for
an astronomically significant period. The "Goal-Year Texts,"
in Sachs's terminology, offer predictions for a certain year-
the goal year-about the moon and planets gleaned from older
diaries that antedate the goal year by appropriate periods. In the
standard order of the texts, these periods are for
Jupiter: 71 years and 83 years Saturn: 59 years
Venus: 8 years Mars: 79 years and 47 years
Mercury: 46 years Moon: 18 years
Ofthe two periods for Mars and for Jupiter, the first is used for
prediction of synodic phenomena and the other for the planet's
return to Normal Stars.
This method of predicting astronomical events is quite effi-
cient, which in part may explain that it survived side by side
with the more sophisticated theoretical schemes. As we shall
see, the recognition of the periodic character of many astro-
nomical phenomena and good values of the relevant periods lie
at the very base of both lunar and planetary theories ofthe ACT
texts.
The rest of the diaries' information, meteorological, economi-
cal, and historical, is not of direct concern to us here. I should
point out, however, that the diaries occupy a unique position
among documents of relevance to the study of ancient history.
The everpresence of the swiftly moving moon enables us to
40 1. Babylonian Arithmetical Astronomy

date the texts, if we can date them at all, to the very day, and in
sheer bulk, continuity, detail, and kind of information they are
unmatched.
We now recognize that several passages in Ptolemy's Alma-
gest refer to this group of texts. Two examples will suffice. First,
Ptolemy chooses the first year (747 B.C.) of the reign of Nabo-
nassar, king of Babylon, as epoch for his tables because "from
that time and on the old observations have been preserved, in
the large, until the present day" (AIm. III, 7). In the second pas-
sage he deplores that the older sustained sequences of planetary
observations are mostly concerned with stationary points and
heliacal risings and settings (AIm. IX, 2). We cannot tell in what
form Babylonian observations reached Ptolemy and Hipparchus
before him.

Babylonian Arithmetical Astronomy

When Neugebauer's Astronomical Cuneiform Texts (ACT) ap-
peared in 1955, it presented in transcription and translation, and
with penetrating commentary, all the then-known texts con-
cerned with mathematical astronomy, about 300 in number. All
derive from Uruk or Babylon and were likely written during the
last three or four centuries B.C. Roughly half of them are con-
cerned with the moon, and half with the five naked-eye planets.
We divide these texts according to content into ephemerides
and procedure texts. An ephemeris gives, among other informa-
tion, computed dates and longitudes for a sequence of interest-
ing phenomena of a planet or the moon, not necessarily-in fact,
rarely-day by day, as the name might suggest. The procedure
texts present rules for calculating ephemerides.
Ideally, a complete set of procedure texts and a set of initial
values should enable us to compute an ephemeris, but the road
to our present understanding of Babylonian theoretical astron-
omy has meandered from ephemerides to procedure texts and
back again. This is not only because the preserved procedure
texts are fragmentary, but also because they are written in
technically specific and very abbreviated terminology that in
many cases could be decoded only when the relevant rules had
been recovered from an analysis of ephemerides.
The texts are classified as belonging to System A or System B
according to the mathematical techniques they employ. Lunar
 Babylonian Arithmetical Astronomy 41

texts of System A come primarily, but not exclusively, from

Babylon, and those of System B mostly from Uruk. I begin the
following discussion by giving precise meaning to the distinction
between the two systems.
Babylonian mathematical astronomy has two features that
seem strange to modern eyes, and it may thus be in order to
mention them here. First, it is entirely arithmetical in character
or, in negative terms, there is no trace of geometrical models
like the ones we have been accustomed to since the time of
Eudoxos. Second, the cuneiform literature nowhere attempts to
justify the precepts of the procedure texts; thus it has rested
with modern scholars to uncover the underlying theoretical
structures.
The space allotted here does not permit me to present all of
Babylonian mathematical astronomy in the detail it deserves,
and I have chosen to emphasize the planetary theories at the
expense of lunar theory, contrary to their relative merit and
interest. Indeed, I prefer treating one part thoroughly to treating
the whole lightly, and planetary theory is by far the simpler
part. For lunar theory I merely aim at conveying a sense of its
complexity and sophistication, but not of its beauty, for it rests
in the subtle and elegant interplay of its technical details. Even
so, the planetary schemes serve to illustrate most of the charac-
teristic approaches and techniques of Babylonian mathematical
astronomy.

Planetary Theory
Babylonian and modern planetary theories differ in the very
formulation of their central problems. Since the rise of Helle-
nistic mathematical astronomy, we have wanted our theories to
enable us to answer the question: Given the time, where is the
planet? Thus, we consider time the independent variable and
seek means of deriving all other information from it.
The Babylonian approach is entirely different. First, all inter-
est, at least primarily, is focused upon the planet only when it is
at one of its characteristic synodic situations or phases: For an
outer planet they are first appearance (f), first stationary point
(<p), opposition (8), second stationary point (w), disappearance
(!1) (see Figure 1, which represents a run of Saturn with latitude
exaggerated four times).
The next bold simplification is that we disregard all but one of
42 1. Babylonian Arithmetical Astronomy

"3
•
60·
I
30·
I

,- <\"2 eclipt ic
I
I
r n'
~---:t::-=>'¥
a r n
-0----0-

FIGURE 1.

these, say the first stationary point <I> , and we now ask the
question: If we are given the longitude and the time at which a
certain planet happens to be at a first stationary point, where
and when will it next be at a first stationary point? What we, in
the Babylonian mode, consider and wish to reproduce is then, a
sequence of discrete points, in time and longitude.
The problem of finding a planet's longitude for a given date is
solved-if it is addressed at all-by interpolation between the
two nearest phases, as we shall see.

Planetary Theory: System A

In Table 2, columns I-IV, I have excerpted the first 25 of in all 56
lines of the text ACT no . 600, an ephemeris for Jupiter at first
stationary point, <I> , for the years S.E. 113 to 173 (-198 to -138).
According to its colophon, the text was written in Uruk S.E. 118,
VII 12 (-193, Oct. 5), thus at the beginning of the time interval of
its contents, so it is mostly a forecast.
In columns I and III we find year, month, and date. The year
number (S.E.) is written [e.g., 1-me 13 (1 hundred 13)], but all
other numbers are sexagesimals. The dates are in tithis and
fractions thereof; the fractions are solely of computational
interest. The use of the tithi (one-thirtieth of a synodic month)
frees us from concern about which months in the future will be
full or hollow; the date in tithis is, of course, always close to the
date in days. The years with a single asterisk (the text has " a" )
contain a month XII 2 , those with a double asterisk ("kin a") a
month V12 • The fixed 19-year pattern of intercalations of the late
period can readily be established from this short excerpt .
Column II contains the difference in tithis (~T) of the dates
minus 12 months. The convention of the text is that the differ-
 Babylonian Ari t hmetical Astro nomy 43

T ABL E 2
II III IV V VI VII VIIl
LIT mo. & date i. L1l Jul. date l date

1 113" 48; 5,10 I 28;41,40 21 I"> 8;6 11> - 198 Apr 22 I"> 3 11 4
114 48; 5,10 II 16;46,50 21 = 14;6 11> 36 - 197 May29 = 8 II 21
115- 48; 5,10 IV 4;52 )(20;611> 36 - 196 Jul 3 )( 14 IV 12
116 48; 5,10 IV 22;57,10 '1"26;6 36 -195 Aug 9 '1" 21 IV 29
5 117 48; 5,10 VI 11; 2,20 n 2;6 36 - 194 Sep 15 tj 26 VI 15
118- 45;54,10 VII 26;56,30 "" 5;55 33;49 - 193 Oct 19 n 29 VlI 28
119 42; 5,10 VIIl 9; 1,40 S!. 5;55 30 - 192 No v18 S!. 0 VIIl 12
120 42; 5,10 IX 21; 6, 50 III 5;55 30 - 191 Dec 20 III I IX 24
121- 42; 5,10 XI 3;12 "" 5;55 30 - 189 Jan 18
"" 1 XI 7
10 122 42; 5,10 XI 15;17,10 I1\. 5;55 30 -188 Feb 18 I1\. 2 XI 18
123- 43;16,10 XII 28;33,20 1 7;6 31;11 -187 Mar 21 1 4 XII2 4
125 48; 5,10 I 16;38,30 1">13;6 36 - 186 Apr 28 I"> 8 I 22
126- 48; 5,10 III 4;43,40 = 19;6 36 -185 Jun 3 = 13 III 10
127 48; 5,10 III 22;48,50 )(25;6 36 -184 Jul 9 )( 19 III 30
15 128 48; 5,10 V 10;54 tj 1;6 36 - 183 Aug 15 '1" 26 V 17
129- 48; 5,10 VI 28;59,10 n 7;6 36 -182 Sep 21 n 0 VII 3
130
131
45; 4,10
42; 5,10
VII 14; 3,20
VIII 26; 8,30
",,10;5
S!.10;5
32;59
30
- 181 Oct 24
"" 3 VlI 15
- 180 No v 23 S!. 5 VIII 28
132-- 42; 5,10 IX 8;13,40 1II 10;5 30 - 179 D ec 23 III 5 IX 11
20 133 42; 5,10 X 20;18,50 ""10;5 30 - 177 J an 23 "" 5 X 22
134- 42; 5,10 XlI 2;24 11\.10;5 30 - 176 Feb 22 I1\. 6 XlI 6
135 44; 6,10 XlI 16;30,10 11 2;6 32;1 - 175 Mar 26 1 9 XlI 21
137- 48; 5,10 II 4;35,20 1"> 18;6 36 - 174 Ma y 3 I"> 13 11 10
138 48; 5,10 II 22;40,30 =24;6 36 -1 73 IUD 8 = 18 II 28
25 139 48; 5,10 IV 10;45,40 )(30;6 36 - 172 Jul 14 )( 25 IV 18

ence between the date in line n and that in line n - 1 is 12

months plus the !IT listed in line n.
Column IV gives the longitude, A, of Jupite r at first stationary
point in terms of degrees of a zodiacal sign. In column V, whi ch
is not in the text, I have presented the differences, !lA, of these
longitudes. We note that columns II and V run parallel, so that
in each line
!IT - !lA = 12;5,10.
Thus, we need bring bu t one of these columns under contr ol to
uncover the structure of the te xt ; as we sh all see, there are good
reasons to beg in with !lA.
Column V, the total progress of Jupiter from one first station-
ary point to the ne xt, falls conspicuously in to constant stretches
44 1. Babylonian Arithmetical Astronomy

of either 36° or 30° separated by intermediate values. The key

to the structure of this column is the realization that the scheme
is tied to the ecliptic. As a procedure text would have it (e.g.,
No. 821):
From Gemini 25° to Scorpio 30° add 30°. Whatever exceeds Scorpio
30°, multiply it by 1;12 and add it to Scorpio 30°.
From Scorpio 30° to Gemini 25° add 36°. Whatever exceeds Gemini
25°, multiply it by 0;50 and add it to Gemini 25°.

Thus, the ecliptic is divided into two parts-the fast and the
slow arcs-inside which the phenomenon progresses in steps of
36° and 30°, respectively. If such a step crosses a boundary of the
arcs, the amount that reaches into the new zone is modified by
one of the factors 0;50 and 1;12. It is significant that these two
factors-5/6 and 6/5 in fractional form-are precisely the ratios
30 :36 and 36: 30. I shall give one example from the text:
In Column IV, line 5 we have n 2;6.
This is in the fast are, so we add 36
and obtain n 38;6
which exceeds n 25
by 13;6.
This multiplied by 0;50 is 10;55
which added to n 25 gives n 35;55
or Cancer 5;55, the longitude in line 6.
With these simple rules we may now continue the text as long
as we please: First we compute the column of longitudes and
form their differences, .6..\. From these we obtain .6.7 from
.6.7 = .6..\ + 12;5,10
and so the date columns.
In column VI I have translated the dates in columns I and III
into Julian dates; in column VII I have given the longitude of
Jupiter corresponding to these dates according to modern tables;
and in column VIII are the Babylonian dates of Jupiter's first
stationary point, again from modern tables.
When we compare these data it is well to keep in mind that
the vernal equinox, from which we count our modern longi-
tudes, played but a secondary role in Babylonian astronomy;
the Babylonian zodiac was sidereally fixed, and at the time of
the text Babylonian longitudes were about 5° larger than their
modern counterparts. Comparing columns IV and VII, the for-
 Babylonian Arithmetical Astronomy 45

mer rounded to nearest integer,

3° :::; A(text) - A(mod.) ~ 7°,
or, allowing for the systematic difference of 5°, that Babylonian
and modern longitudes never differ by more than 2°, and rarely
by that much. (If instead of the text's scheme we were to use
the mean value boA = 33;10°, the length of this interval would
increase from 4° to 11°.)
The Babylonian dates in columns I and III always precede the
correct dates in column VIII , but never by more than a week.
Here we may recall Ptolemy's complaint in Almagest IX 2
("On the difficulties of constructing a planetary theory") that it
is impossible to decide with any accuracy not where, but when, a
planet becomes stationary. The Babylonian tendency was, it
seems, to deem it stationary early.
The quality of the text remains the same to its end (and for a
very long while beyond). This seems curious when we consider
the crudity of the schemes, and hence wants explanation.

System A: Period Relations

An iterative scheme like the one for finding the longitudes in
our text, where anyone of the entries determines its successors,
is particularly vulnerable to corruption and decay due to com-
putational errors and to imperfections in the scheme itself.
The Babylonian astronomers guarded against mistakes in their
calculations by devising various checking rules that allowed
them to proceed in steps of many lines (an example in our text is
that an advance of 11 lines leads to an effective progress of 5° on
the fast arc, 4;10° on the slow).
As to the second point: To prevent the inaccuracies inherent
in all approximations to natural phenomena from accumulating
arbitrarily, the Babylonians made their schemes obey certain
relations that ensured precise return in longitude after astro-
nomically significant periods. Our text is built on the relation
that 391 synodic periods of Jupiter correspond to 36 revolutions
of Jupiter in the ecliptic and 391 + 36 = 427 years. Indeed, if we
begin with ~8;6 in line 1 of our text and calculate its 391 succes-
sors correctly, we end with ~8;6, and the longitudes have skip-
ped 36 times around the ecliptic.
To convince ourselves of this, we observe that the sequence
of longitudes in column IV can be considered as positions of a
46 1. Babylonian Arithmetical Astronomy

particle, one time unit apart, if it travels with a speed of 36° per
time unit on the fast arc and 30° on the slow. This interpretation,
though wrong and anachronistic, leads to the very rules for cal-
culating the longitudes and allows us to ask how long it takes
the particle to complete precisely one revolution in the ecliptic.
Since the fast arc measures 205°, and the slow 155°, this time
interval will be

p = 205 155 = 391 = 10.51 40.

36 + 30 36 "
Thus, one revolution lasts nearly 11 time units. Now, in the
terms of the text, one "time unit" is one " synodic phenomenon,"
and a fractional number of synodic phenomena makes little
sense. Thus, we put the previous relation in integral form :
391 synodic phenomena ~ 36 revolutions.
Since during one synodic period the Sun travels one complete
revolution in addition to , on the average, Jupiter's synodic arc,
the corresponding time interval is 391 + 36 = 427 years.

System A: General Theory

We have seen one example of what is, in fact, a general theory
underlying all planetary arithmetical models of System A. Such
a model enables us to calculate corresponding dates and longi-
tudes for consecutive planetary phases of the same kind, for in-
stance, first stationary points <P. The scheme for finding longi-
tudes is basic: It consists of a generating function and a rule for
deriving the synodic arc, ~)" from it. The generating function,
to use an anachronistic term, is a piecewise constant function of
longitude-a step function. The ecliptic is divided into arcs
a I , a2," " an , where ai stands for both an arc and its length; to
these are assigned the functional values WI, W2 , ... , to.; We have
that
al + a 2 + ... + a n = 360°.
Provided with an initial value ),I, we first seek the arc a i to
which ),1 belongs. If ),1 + ui, still lies in a i, then ~), is uu, and the
longitude of the next <P is ), 2 = ),1 + uu, If not, the interpolation
rule comes into play. It is, as given in the procedure texts, that
the part of to, reaching into ai+ l is modified in the ratio Wi+! : un,
and so on , if more zones are involved.
 Babylonian Arithmetical Astronomy 47

Thus, a sequence of longitudes may be computed from an ini-

tial value. The corresponding sequence of dates runs parallel to
it, as it were, for the synodic times are simply
b.t = b.A + C,
where C is an appropriate constant.
The principal reason such schemes work so well is that built
into them are certain period relations. If we form

and write P as an irreducible fraction

II
p= Z ' (II, Z) = 1,
we can show as before that the planet will return precisely to its
initial longitude Al after II applications of the synodic arc, and
not before, and that the phase has traveled Z revolutions in
the ecliptic in the process. The integers II and Z are in all cases
astronomically significant parameters. The corresponding total
time is Z years for Mercury, II + Z years for Venus, Jupiter, and
Saturn, and 2II + Z years for Mars.
The II computed positions arrange themselves very neatly in
the ecliptic: They form a set of points that are evenly distributed
within each zone ; indeed, within the arc D:i any two neighboring
possible positions are separated by the interval

as can be shown.
A sequence of positions of consecutive phases can be found
easily on the basis of such a distribution of II intervals in the
ecliptic: We let the phase start at anyone of the end points and
have it progress in steps consisting always of Z consecutive
small intervals, regardless of length.
In the scheme of our text we have II = 391 and Z = 36, so all
possible positions are either
30°
or 12 = 36 = 0;50° (slow arc)

apart (there may be , in fact there are in our text, intermediate

intervals around the boundaries of the zones). Thus, the fast arc
 A st ro no m y
A ri th m et ical
48 1. B ab yl on ia n
 Babylonian Arithmetical Astronomy 49

contains 205 intervals, and the slow 186; in all, there are 391
intervals on the ecliptic. Advancing one line in the longitude
column means advancing 36 consecutive intervals in the distri-
bution. Advancing 11 lines means an advance of
11 · 36 = 396 = 391 + 5 == 5 intervals (mod. 391).
The effective advance of 5 intervals after 11 lines means 5° on
the fast are, 4;10° on the slow, the checking rule I cited above.
There remains to consider the scheme for finding dates. The
rule
~t = ~A+ C

can be derived from the assumption that a certain phase always

takes place at a fixed elongation from the sun, but it may also be
based on experience. However, once established, whether on
theoretical or empirical grounds, there can be no doubt about
the value of the constant C: It must be the difference between
the mean synodic time, /st, and the mean synodic are, ~A . Thus,

FIGURE 2. Fig. · 2 shows the obverse of the cuneiform text B.M . 36300
from the British Museum's vast collection of Mesopotamian clay
tablets. Its columns, read from top to bottom and in order, constitute
one long sequence of longitudes, but not dates, of Saturn at consecu-
tive synodic phenomena I', <P, w, n, I', <P, w, n, etc. (i.e., first visibility,
first and second stationary points, disappearance), each computed with
a System A model for Saturn:
Leo 10° to Pisces 0°: w = 11;43,7,30°,
Pisces 0° to Leo 10°: W = 14;3,45°
and the period 256 phenomena correspond to 9 revolutions In the
ecliptic and (256 + 9 =) 265 years.
The reverse is a continuation of the obverse, and together they are
but four columns short of covering a complete period of 265 phenom-
ena of each kind, the longitudes distributing themselves very densely
in the ecliptic.
Such a text may have served as a template of sorts: If you are given
an initial longitude of a phenomenon, you are sure to find a close
approximation to it in the text, and you can simply read off the longi-
tudes of the subsequent phenomena.
This text and others like it are published in A. Aaboe and A Sachs,
" Some Dateless Computed Lists of Longitudes of Characteristic Plan-
etary Phenomena from the Late-Babylonian Period." Journal of Cune-
iform Studies, Vol. XX, pp. 1-33.
50 1. Bab ylonian Arithmetical Astronomy

T ABL E 3

IT Z 1-. w·1- I i Z"".

_1- ~

o A 1f",2.' 3,IfS 1
2-
300
log; 7,30
2/tS = 0; go
,
'YR= 0·730 .
M 12,2-3 t) 0 1 30 1/-, = 0 ; 30
2- 2€;2..0 17/3b= 0;2.8,20

Ai r tlY,33 1'1, e 1 1, 'fb Ilg = OJ 7,30

~ 2.
.3
,
.2. ,:l.f.lO
1,3'1; 13,20
1/ 6: 0 ;10
I/~ = 0; ,", /fa
'='
...... 2S; (3 s,» f 2, C/O 1'3 : 0; 2.0
2. " '16; lIo 2ffj : 0; 13,.20
.3 f 3& 'is" : OJ n;
AJ. 2 ZO,23 ' ,28 I 1,If7., 1/(,, '10 Sltg: 0; U" 'fo
2. l ,&j;2.D 11.3 :. 0 j zo
3 ',37 I/lf: OJ IS"
If 2.. q; 2.0 113 = 0; 20
52. 11,2y 3)37 1 f, V8; 30 Y'2. = 0; 30
2. 2,ojH,2.O ¥fj : OJ 33,.2.0
3 t, IfS; 30 '12" : OJ 30
If oZ , IS·37 30 S"/g : 0j 37,30
A Iff>
c! 2,13 2.,'h· 2 ;30
18 1
2. 30 12/3 : 1; '(0
3 !fo 2.y~ • z, 13,2.0
If 1) 0 3 '/3 = 3; ~o
S 1,30 S" : S"
b " 7; 30 3% • 3; I(S-

A 6,3/ 36 f 36 1 • f

1t A1 Ib, ('j 1,30

2.
f
30
3&
%:
~/s : Oj21f
a·So
2. 30 1/3 • 0;U)
A' b,3/ 3(; 1 30 0/" : O;So
2. 33; y!>- 'S/fb = 0; se, IS
3 3" 1 = 1
Lf 33·l(S I lfb = OJS{, I S-
A') ',31 36 1 30 ~ = o;So
J.. 33; IfS '¥ib = 0,.Sb,'s"
3 3b 1 = 1
'I
An) !j,S3 2) I
B ·YS"
30 f 'I, :
'SilE>= O,.S','S-
' : (" 'to
2. 33; 'i~ 1 '1y = ' ; (i)
3 3& W3 = li le
If 33· 'fS I Yy : 1· IS"
A 9 II; '13, 7,30 12.1"% : .s,
"Ii If) 1& I
2. IIf · 3 'is
Ij 7,30
I "lib ' I ; 33 '(S
 Babylonian Arithmetical Astronomy 51

-Z
TABLE 3 (continued)

lX,. VIS -~
~ ,
,?s £!. LX-J
i- .. Ii
3,IIf 0 2'fJIS J.2;2Z, e TJf /3 - ~27
2,'16 ~2.,€ X 27 - 7Jr 13
2 J1('1;30 S;2"} IZ;2.3 '72'1 - sc n, 30
3 IS'; 30 6 S''f X. '3;30--rw~"J
Z, I/fj 22,0 3; '3, 7J~~ '" J'2,1 - .lO It;,
:l,N ,3, 21f ~ (e;. - yo 30
)(0 -J"l-f
" -- x
(J I 'j, CJ
t.s» 5;30 3j '3, 7, 30 ~ ~ ~(,
2,ftf (0, ,J ~.<." 10
t, S6 "1,'10 )( (0 - dJJ b
{,30 S;;'1f 3; '3, 7, 2&;.. <[9 0 - ~30
' J3b ~ «e ::-0-:66
t ,2'} s: S'b Z 6 - y S
f,2S If, I~ y S - ll: 30
3,0 ',0 3; '7, 7,27,,, 90 - I' 30
1,0 I, ve ~ 0 - ::::z 30
1,0 :2,0 )(0 - Y 30
fO t .36 ~ 30 - It. 30
1) 0 2'1- 7· 231..0
) , O'Jr
f) 0 3b 6101
t,o 27 ~~
IJ0 '8 'It,J'''
IJo 1:<" z=
XV"
f 0 '6
3,.2S 3,2S to, 5"', "10 ~ 0 - X 1.S'
.2,35" 3, b :II' 2S - .", 30
3,2~ lj')~S Iq·S'<'J I{o ~ s- )r 1.7
z.ss 7, S'f .lr 2.7 - J"s
:1.,211 , ,
., -
2.,0 10·S,1I0 ~~-,.,,'3
S3 S"6; 31. ~ ~L
2,ts .2,tS Z .2.. - b" '7
Sl. SS,.Z8 ~ /7 - ~,

~,O 2J2. y 10jS', Yo ~ S; "Is - "1 5"; lIS'

S-6;1~- f) 0 ~ Sj'fS- ..B .l..
.2,IS' 2, ,S Z z - '1:517
'18; 'IS S'.l. 'd 17 - Q 5";'15
.2 0 I/{~ 10; 51, b,'lo !9 ., - "'l'}
'V€ 3~;J.'t "l ') - .t<" 27
.2,,U) 1)'IS J<"27-~17
S".l. Y';36 b' I) - ~ ~
3)20 2,33J .3' .2.8;2{,,1(0 6L (0 - .::::-30
:J.,Yo 1, If;.,),,/ )(. 0 - Ol. ro
52 1. Babylonian Arithmetical Astronomy

the constant is readily derived from II and Z. Here we need that

1 year = 12;22,8 synodic months = 12;22 ,8 ·30 tithis = 6,11;47" ,

a basic parameter of Babylonian theoretical astronomy (it is,
e.g., the period P of the longitude columns in lunar ephemerides
of System A). With these parameters we obtain

~>. = Z . 6,0
0

II
and, for Jupiter and Saturn,
-
~t = -rr'
II+Z
6,11;4 7"

(for Mars, II + Z should be replaced by 2II + Z).

Thus,
- - Z
C = ~t - ~>. = 6,11;4 + II . 11;4

= 12
m
+ (1 + ~) . 11;47"
(the rule's direct translation of degrees into tithis reflects that
the sun travels very nearly one degree per tithi). With II = 391
and Z = 36, we find
C = 12 m + 12;5,8,7, . .. 7",
which compares well with our text's use of 12;5,10 for the last
part.
In Table 3 I gathered the parameters of all planetary models
of System A and included those of the monthly motion of the
sun from the lunar theory. In addition, N. T. Hamilton recovered
parts of a System A scheme for Venus from ACT No. 1050, re-
cently published, but its parameters are not included in Table 3.

Planetary Theory: System B

Table.4 gives the first 22 of at least 62 lines of ACT No . 620,
an ephemeris for Jupiter at opposition for at least the years S.E.
127 to 194. It is arranged much like the previous text except
that here both difference columns are included, and we shall
first examine column IV, ~>..
 Bab ylonian Arithmetical Astronomy 53

TABLE 4
No. 620 I 11 III IV V
I. year LIT month & date LI). ).

0 2,7 49;42 V 27;36 37;37 )( 24;31

I 2,8 47;54 VII 15;30 35;49 'Y' 30;20
2 2,9- 46;6 IX 1;36 34;1 D 4;21
3 2,10 44;18 IX 15;54 32;13 lZll 6;34
4 2,11 42;30 X 28;24 30;25 Sl. 6;59
5 2,12-- 40;42 XI 9;6 28;37 I1J' 5;36
6 2,13 41;47,30 XII 20;53,30 29;42
"" 5;18
7 2,15 43;35,30 I 4;29 31;30 11\ 6;48
8 2,16 45;23,30 II 19;52,30 33;18 t 10;6
IV
"
9 2,17- 47;11,30 7;4 35;6 15;12
10 2,18 48;59,30 IV 26; 3,30 36;54 = 22;6
II 2,19 49;27 VI 15;30,30 37;22 )( 29;28
12 2,20- 47;39 VIII 3; 9,30 35;34 tl 5;2
13 2,21 45;51 VIII 19; 0,30 33;46 D 8;48
14 2,22- 44;3 X 3; 3,30 31;58 lZll 10;46
15 2,23 42;15 X 15;18,30 30;10 Sl. 10;56
16 2,24 40;27 XI 25;45,30 28;22 I1J' 9;18
I7 2,25- 42; 2,30 XII2 7;48 29;57 "" 9;15
18 2,27 43;50,30 I 21;38,30 31;45 11\ II
19 2,28- 45;38,30 111 7;17 33;33 t 14;33
20
21
2,29
2,30
47;26,30
49;14,30
III
V
24;43,30
13;58
35;21
37;9 "
=
19;54
27;3
22 2,31-- 49;12 VI2 3;10 37;7 'Y' 4;10

The entries in column IV decrease regularly by the amount

d = 1;48° per line until a minimum is passed between lines 5 and
6. From line 6 the values increase, again by 1;48°, until a maxi-
mum is passed between lines 10 and 11, at which point they
begin to decrease, and so on . If we plot ~>. as a function of line
number, we get a piecewise linear graph like the one in Figure 3.
One gets from an ascending branch to a descending branch of
such a zigzag function, as we call it, by following a simple re -
flection rule (often stated in procedure texts); if the application
of the line-by-line difference d (here 1;48°) leads to a value larger
than a certain fixed maximum M (here 38;2°), then the excess
over M is subtracted from M to yield the next value of the func-
tion, and symmetrically about the minimum m (here 28;15,30°).
The reflection in the maximum between lines 10 and 11 is then
54 1. Babylonian Arithmetical Astronomy

M-----------=>~:------------

m-"::'~---------------=~';--

line number

FIGURE 3.

executed thus:
Col. IV, 1. 10: 36;54°
+d: 1;48°
38;42°
-M: 38; 2°
0;40°
which subtracted from M gives: 37;22°,

the entry in line II.

The zigzag function, sometimes refined in various ingenious
ways, is one of the two basic modes of describing a simply peri-
odic component of a more complex astronomical phenomenon-
the other is the step function of System A. It is easy to compute,
and it has a simply controlled period:
= 2(M - m)
P d .

With the parameters of the present text, M = 38;2° and m =

28;15,30°, d = 1;48°, we obtain
391
P = 10;51,40 = 36 .

We recognize this as precisely the period of the System A

scheme for Jupiter. There it implied that IT = 391 applications of
the synodic arc lead to precisely Z = 36 revolutions in the
ecliptic. Here the period relation is that IT lines lead to precise
return in Do>. and embr ace Z " waves" of the zigzag function, but
 Babylonian Arithmetical Astronomy 55

it says nothing about the longitudes themselves. To ensure re-

turn in longitudes, we should require that the mean synodic arc
be
-bo,\ = z .II360° = 36391
. 360 °
= 33;8,44,48, . .. ,
and our zigzag function obliges by having as its mean value
J-l = ~(M + m) = 33;8,45°.
In column II of No. 620, boT is represented as a zigzag function
of parameters

J-l = ~(M + m) = 45;14 T

•

In this text we do not quite have that the difference between boT
and bo,\ is constant, but the amplitudes and differences, and so
the periods, of the two zigzag functions are identical. Thus, the
two functions are slightly out of phase, for no reason that I can
see other than numerical convenience. However, the difference
between the mean values of columns II and IV is 12;5,15, which
is close to the corresponding value 12;5,10 of the previous text.

Daily Planetary Positions

We have now seen how the date and longitude of the planetary
phases were computed. The day-by-day progress of a planet was
calculated by interpolation between neighboring phases. For
Jupiter (which here, as in other respects, is best documented
among the planets) we find several such interpolation schemes
of varying degree of complexity-the procedure text ACT No.
810 alone contains three variants.
Table 5 presents the most elaborate and sophisticated example
of such schemes. It is the result of P. Huber's ingenious recon-
struction and linking of two ACT texts, and I have excerpted the
most interesting parts. The first column gives year and date
(years in S.E., dates in days, not tithis), the last Jupiter's longi-
tudes, and the first and second columns the second and first dif-
ferences of the longitudes. The phases divide the text into
stretches of constant third difference, as indicated at the brack-
ets to the table's right.
It is worthy of note that in 164 B.C. (= -163), the year of the
text, Jupiter's retrograde arc was 9.8° (Tuckerman). This is a full
56 1. Babylonian Arithmetical Astronomy

TABLE 5
T AaJ\. AA A
f. 1'17 !E
z -
f 0
b
+12,\10
i- I)., 31, .If
21
2'1;'2,31, S1f
~ r
3 - 12. -+-1)., 31,' fJ. 21;1-o,I1)b
If - 18 tI2,3'l,2lf
,
"!; 37,'"
S. S - 2y +ll,J,,! '1,.S'o,3i
- 30 + 12,H,30 3, 'l.,30 /
7 - l' i-12)~9f ISiSY,2Y
If - 'I;.. +12,37,12,. 2831, , Jb
'1 - ¥8 -t/2.)',2/f 'fl , g
-Slf , , L:i3). =-0,.0,0/6
(0. (0 +1235'30 S3,1(3,30 0

1/ - 1,0 +12,3'1,30 I,. ','8

,

"r· 1'18 I 1
z
-II,'fg
-II,Sf
+ s7,Slf
+ If"
,c,; ',3{,
I,; 7,).J.,
~

3 -12- + 3'1 I,; 7,Sb

't -12, , + 21,Olf (b; 9, '7,S'f
~- -U,I2 + 'I, If2, Ib,' g,J..7, 36
~ -11.,2- - ).1,'f~ 1(,,. g, 5", ss:
12S", 7 -11,S2 - 33,3' 1'"·73.2., I'
g -II, 'fJ., - 'fS,I€
'l -/l)J.. - S',So I':
'" 11(, Si
i- r-o' g
1(.. If' 'I/.rt
10 -11,2J., -1,8/2.-
I '. I EJi.. + 0; 0, 0/10 0

m. .!l! If -2)2 -7,'<-, 9 /I i '(',/'1, '6 ./

s - 2,22.. -7,1'1,30 II; 3g, S'1,38
(, + ",3J.. -7,II,sr 1/;3/, If?, '10
7 T2,'f~ - ~ ~,1b
.
/I,·).~ 3~ 2'1

tl ).. : +0; 0/0,10 0

238. ~ 1 tIl, 'fl. - 3S; S1l 7;2], ~Jo .;I'

1.. TII,S2 - 2.'1, , 7;:U,'f3,J..'f
2'fo. 3 T/). J..
'i +1:1.; 1.2.
- 1:1., If
+ ).'1,1' ~:U,~3b
7;;U,3',).0
'P
S +12/ 00 , + 3',U,Jo 7j).).,3',~Jo
{" +12., I + 1'8,).3/30 7; ~1.o,2L.
,
Ii3,A : - 0; 0/ 0,5,30 0

37Y: +0,0,30 tl3, S"O, 130

, 1.B/Y, S'f, 23,30

ACTNos. (,S1j f. '55

 Lunar Theory 57

degree more than what we find in the text. The motivation of

this fine scheme cannot, then, be found among the concerns of
practical astronomy, for if the Babylonian astronomers could
tolerate an error of one degree in the length of Jupiter's retro-
grade arc, they surely had no need for so elaborate a scheme to
account satisfactorily for Jupiter's daily motion.

Lunar Theory
Babylonian lunar theories are classified as belonging to System
A or System B according to their manner of accounting for
monthly solar (and lunar) progress: System A employs a step
function oflongitude; System B a zigzag function of line number.
However, this is not the only, nor even the most significant, dif-
ference between the two systems.
The relations between Systems A and B for the moon are not
at all clear. The two systems were in use at the same time, in
both Babylon and Uruk, so one did not replace the other. An
analysis of System A reveals a tight and consistent inner struc-
ture, deducible from a rather small number of reasonable basic
assumptions-most of them are not mentioned in the procedure
text. In comparison, System B seems a collection of parts that
are at best but loosely connected and in one instance (column J)
of quite unwarranted complexity.

Lunar Theory: System A

In Table 6 I reproduced our reconstruction of a small fragment
of a lunar ephemeris of System A (ACT No. 20): The preserved
text is inside the dotted lines. I chose this example rather than a
better-preserved text to emphasize that all lunar ephemerides of
System A are strictly connectible (except for the last visibility
columns) and thus are slices, as it were, of one great continuing
ephemeris that we have been able to bring under control and
reconstruct, particularly with the aid of electronic computers.
This feature seems quite in harmony with the inner consistency
of System A, and we find nothing like it in the texts of System B.
Column T gives year (S.E.) and month, and everything in a
horizontal line concerns the situation near a syzygy in that
month-on the obverse the conjunction at the end of the month,
on the reverse the opposition at month's middle.
 J"q t !: 'li t ' I; OC '(,(. ' 6 - ,"t !:t ',: 'w I (~ ' f,(; ' t lit: ' !. 'n ' Sf. ' I I Ht: Ot t: ' I ;:: ' t + ut'lH 'c 10 0;::'{;9 'Rt ' t '0 'e: I 8" '(;
:;));oJ!P ()<I ' . 1 't W '7-t '6- 6 1 ·!. ' !I{; ul. 't t' ' ~ t ' t t 'il t ' St ' t !,' I 'Hi- ' I I ' IJ \/"' ·!.I ' I;; '(; + ~l1 'm; '(,' l it ';:C; ~H 1)(, ' r.1 ' f, ' !.I ' 8lJ ' I ' lI X
OJ Ll ' f (;~ 'lJ '\f -
'"I I tt ' I);' ' tt 'ut ' !" 't ; ;t !: I '~ I ;:: 'ec ' I I \If. ~ ;. 't t'i = (k": ' 6('; 'Ot '0 iii- - - -tt-~ ii i 01' 'Ht ' LS ' j: ' I '<:, II X
,t l \ 2IZ (;1
'!."9 't r.S '!,; -x-. 0;: ' &1 'j: ' !.I '! ',: ' f. !.q "!I;: 'vc '~ I 1Ir. 'IJc. 'm ' f - ur. ' !ol ' ~;. 't.: U' t t ~ lJ l U(: 't"9 ' Nt '& ' t; IX
lt l \"1_ __ .. ~t! <:<: 'ee'c 9 '1'1" 1+ 0; ' e l ';:t 't:;. 's; ' n: 'f. ~:I '!I;' ' !.I ' CI ;::1 "ot '!II 'H- r.!: 'c 't;: 'z ~ tt : 0 1 {)(, fr.S ' ~t 'M: ' 9 'z X
1l ',;r OC ud 1:& ' II 'e S ' t1; 'v-t- fi t ' !I III 'n 'liC '!" ' r. t l '~I;; 'm; 'CI !) ' ;: ' f 'n H ' K' '!.(: 'r. II tt : 0 1 (It '!l(: "tt '06 '6 1(, XI
g ...... - - I~ .. urde '(;, 'es ';. S ' H7 '6 + 0;: 'r.s 's ·.:t' i: !: I 'Hi: ' I t ' t l t ;: '!It ' 1'\ ' ' ~: - f t. ';,:t ' ,n : '(: R t t ~ fI( ut 'n '<:1 ' 6 I li A
-"~ np 6r.'4; t '(- 0 1 ' !.t" 'fj+ !.' I '! I"':' r.j: "n ' I ' CI ' I - u t 'lit "til} '(: J.., t t ~ I H 0.. ' f., ' !,'f, ',~ '1'( "
v 1"1 11" 1'11 ' f I II t ' i: tr. 11,\
-e U !~ tl f, it t 'r + t'C'l !, t ' t ' !.!.' lit ' ( 1'1: '~;: 's t ' lJl - -ii"7jt"7i"+ ·K- or.-';:~ To·t
_1.! ~~t ' t cz',n 'I: 01' II re sr (: 1.\
II I S(.' ' .:9 ' , m: 't l .(.' + !.'t 'l: ' !.!, 1 ot '~I t ' !. I ''If: ' !ol ' S f ' ;: ot ' ~ t: '! I ' ~ I ;: t ' ! I~" 'm: ' t + tt ' UI: ' f: := ~ t ~(: I Ot '!1 ' 11 '!.it ' CI ' z: ~\
us t '!iT't. O£ 'r r ' 1- lei
'"' ~' .. 'f: ' ..!.!,' I lit ' !It '.:.c.; 'n: ": 1 ' " of: ' ;: ~ ' t lo ' t l t": ' ''':t 'w ' '\1+ ,:' 't1: 'r. \..1 ut ' !,.: ~ fl or. 't l: ' £ 1 '6~' '01 ' (: ,\ 1
~! S c.(:'!.r. ';' Of. 'r t' t - ft 'r '!.I '"t: ' I I 'tif: ' f. (If: ';:;: ' j:t " :1 tt.: ' n: '!I t '... + (: I} ;:t: ' f: t Hf: ~~Il (1(. ' C( '8 '(: 1\I
t o:!
'!.!" Ili0;t " It
!ot \' ' !.!,'
It ' Ot ';. II.: 'er '~ - 1" 1
'"' I ~,!.' '~t ' t l '.. '!.' ' I' li t '(:(: ' II ' f: 1 ;: 1 '!It ' !ot '1:+ I;:; 'Cr. 'r. Ul m: ' j:;: ~ 8 ( 0 1' ' !I(: ' t (: ' L (,"~ 4t- II
>. JNt 6 'tt 'e f3i; ' tj - \"1 s l '01 ' !.t tlt: ' t~ ' wt 'm: 't o~: ';:;: ' '<l l ';: 1 1 't l ' l + 1)1
..:' '! I 't !.-' I ~ 4J(. fir. ' ElJ '~.:; ' I t ' (: 'r. I . Lt '(: I ' ,\ 3J
II
'd I~ :i .:J 'J :) II ,~ .I.
8o
'I
I=: J\!f{ ;)}'J !P SlJ \1: ' I 't ;, 11' 1 tlf: "t t ' !'o!t Itt..: ] I II' ';: !.'I ' !I" ' ,,, ' :t ' j' l I !<t ' I ';:t: 'r.- li t: '!.( :tj ' t ~~ _ tL'~!L;t~_ l k"': ' r.r.
'~!' 'OI 'I) I ' - ' IIX
o ut '1It ' ':'4,; 't:t tt , !J(,
~ )-J! P :> ~ !.t ';.c '(. (,1." ''1 's -I1-: '{i- I 'ot ' t: ~t '. t ';:r. ',;. '~ I l~' 't': ' I ;: - or. 'cj;; 'I ~' ' (; )( ot'ilt r.1 '~ ( 'r. II X
I I;:
13trJ ~~ 1.1~ I I 'f: ': ' ~ _ - -- - - ..!~ Ot' ~ ' (. ;.!.' \' ' t - ' t:t: '~'fJ '!.'t: ' ~ t 'm,,' ' j: n '~ t ' ~C '~ It ' t f ~1: 'SI 7"!.-":t-+ !It: ' ':'1 ' tt '(; = tt ~ !.:(, Ot ' !) ' Ilf 'ur. ' 7-1 '7, IX
< lit ' !Ie: ' t ' U: '!,'j: ' I: !.' t 's t ';:I: ' t ' t l I ":: 'tt: 't t 'r + j: !,! ' ,: ' !I(: '(. \.t tt~ m: ()(: It t "!:!; 'ot '6 'r. x
~~ ~:~ l~~:;~<~ ~ u~~' !~ :~~ ::~ ~~ :~t 'r.~ "" ' !It ' ':''.: ';:t ";:;: · I !.' ' I:; ~' t ' s t ';:f: ';:,: 'I: I ' !.!] Ii ' t ' .:.+ s ' t !: ' ~';: ' (: t tt :~ Sf; '!J ' (, XI
til 0 1: U le ~ 1,1 ~.. ....
:.t '!if' <. ~;: u !dc !.i ' yc, ' t '" ' f l.:' '!. + li t: ' f:!t ' polt: ' I;: 'I I "':'1 ' t l,.'t ' '' I'' ;' f: 'u t ';: 1 1':1 't<.i ' !os ' 1'+ t (: ';;t 'I:.: ';: lu tt ; ~~ ot '!l" ' t '6 't 'r. 111.\
~
,..., I ......u!d ,.. i)~ It ' t l . ~. tiV 'np / t:!I 't 111+ t t 't !.it ' 'oil' ';:1: '~' II ~ ~l~ '~!:. ~I~ ~j:...t Itt ' 01: 'fit ' (: ::::=: tJr ' S" u;:: ' ~ 'w 't <; ', '(: 11,\
Q) __...-iii: ' np I.l~ It ' .:, ' t ~u uPf (. " f. 't !.'f,,' ' j. l,,' ' ':'+ I IPi li t 'f-t 'I"': ';:t ! I ~ 'f lJt ' .... t '(':1: '! H ' II r s l ' I' I ' 1-:\' - ut ' Of. " l i C J.u tt ~lJ;: 1)l, ' r. I ' U: ' ~ '1 1.\
1)1:\ : 1 I U!'1. 1.1': J:t ',.t ·(. S(."- - - !; ! Ur. ';:t 't .:. 't + l'f'I <;t ' f: ' ':'': lit 'I I;; ' t t: ' .:,t '~ l l \ 1: 't CJt ' r.t: '! Il: 'r.I: '1 1 l '.; 'f:f: 'lj t ' !: - 1)f: inl ·!J(': ' r. ~t-'st ~ ir.-
-:.f- 01' '~ I t '!.c 't t '6~ ' I ,\
~l!:~\. '- I !7! .~l~ __':_':I(.. :.: S';: n~ (. 'CI't: !o J: + P'I t.:t ',: ' !.t u;: ' r.': ' s ' s;: 'Ii ' t n 't:£: ' !Ii : \ ' 1 ' j':1 I:r. '1)1 ' m: '~' - fll: 'oe 'r.r. 'r, ~ !.iI ' lt : 8(: (J<. ' r,£ 'ot: 'e: 'r. ,\I
~... .... ...t .nl· l.l ~ !,: '1'.: ',: :1!, .:t 't:t 'r- I:~' ' (:- 11~1 !.'t ' ,: '!.t,.' II;': '.:1 ';:;: 'ot ' lit 'I:t ' f: ~'t ' r.I: '!If; ' ,:S ' (: 1 !.-' t ' K' ' gt '!I of: ' t t 't f. ' c lJt 't t: :o O(: '~ '8(; '9 l'~ ' t; I II
h I.; ' na 1';:' ' I' I ' j: t:j: '!I - I II!1 !.'t ' t; ' ':' ~' lit ' ! , "J': ' 1':;,' 'm: '':' 1 'r. lJt ' t F: '!IC " it ' r.1 JI ~ ( '~J(. ~ (; Ot '!to ' t Z ' (, "8 'z II
< t tl " lr.q I:S '!It ' l lie ';; ' tj - Jl~ 1 c..' t' l: ' ':'c; '; t ' ~ I 'I;!: ' j: t t 'cc '!l C ' , (: 'I' I ~ !.'t 'sl ~ t nz; '!!It ' o r 'r. r . ! t '(,
I=: .It' q ;1 ~ ;:1 '(. 6 '; ;:: '6- ;q (If: '!i;::'S l; '..,1 i 01' 't: \ 't T l: 'l It '1: '!.' I J¥ ~ I " I ~ !l I ',\qa
oj I~"': ' r.r. \ ' ( itt : ' t l ';: I1X
k~{~~;f~;~~ ::i: ~ ! :~!:~
'2 II ' II
o 9 3 'I HV .L
>,
.0
oj
~
.....;

<Xl
lO
 Lunar Theory 59

Column <I> is a zigzag function whose period is the anomalistic

month; in fact, it is in phase with the apparent lunar velocity
(column F). It serves as a means of injecting lunar anomaly into
the theory. We have recently found that column <I> denotes the
length of 223 successive lunations, beginning at the syzygy for
which <I> is listed, minus 6585 days and assuming a constant
monthly solar progress of 30°, but it had long been known how <I>
served as the basis of finding the later column G. I shall not
explain here the very sophisticated methodology underlying the
relation of <I> to G (and to other similar functions).
In column B we find the longitude of the moon at syzygy or,
more correctly, monthly positions of the sun on the obverse, and
monthly positions of the sun increased by 180° on the reverse.
The sun moves according to the System A scheme whose pa-
rameters are given in Table 3 and whose period P = 12;22,8
months is the length of year that pervades all of Babylonian as-
tronomy (it serves as the value of the sidereal, tropical, and
anomalistic year; the Babylonian astronomers were unaware
that these three years were of different lengths). It should be
noted that lunar anomaly has so small an effect on the position of
syzygies that it may safely be ignored. Column C gives the length
of daylight, and column E is lunar latitude (in units, "barley-
corns," 72 of which equal one degree; the integral barleycorns
occupy the first two sexagesimal places). Lunar latitude turns
out to be a very simple function-a slightly modified zigzag
function-of the moon's elongation from the ascending node;
this nodal elongation is easily found from column B giving lunar
longitude combined with the underlying assumption that the
node retrogresses by the constant amount 1;33,55,30° per month,
and some initial position of the node. Eclipse warnings would be
issued whenever the new or full moon has smallest latitude at a
nodal crossing; this is done with an eclipse magnitude function
depending simply on column E, but which I have not bothered to
reconstruct.
Columns G and J together give the excess over 29 days of the
time from conjunction to conjunction, or opposition to opposi-
tion, where G depends on lunar, and J on solar anomaly. The
eclipse magnitude \l1 predicts lunar eclipses very well, but for
solar eclipses-treated precisely the same as lunar eclipses-it
can serve only as a possible maximum. Column C' gives a cor-
rection due to the variation in length of daylight, column K the
sum of G, J, and C', and column M lists the moment of syzygy (on
60 1. Babylonian Arithmetical Astronomy

FIGURE 4. Lunar system A

the obverse date and time degrees of conjunction before sunset of

the day), and the moments proceed from line to line by the
amount - K on the obverse, and K on the reverse. Finally, col-
umn P, preserved on obverse only, gives information about the
visibility of the new crescent; "1" means that the previous
month turned out to be full (30 days long), "30" means that it
was hollow (29 days long), and the following numbers give the
computed time from sunset to moonset.
In Figure 4 I show in schematic form the relations between the
columns of lunar System A.

Lunar Theory: System B

Figure 5 is a photograph of the reverse of ACT No. 122 (BM
34580 + ...), one of the best-preserved lunar ephemerides. It was
written in S.E. 209 IX 18, as the scribe tells us in the colophon,
and it treats new moons for the years S.E. 208, 209, and 210.
I cannot give a transcription of the text here, but I shall men-
tion a few of the characteristic features of lunar System B, to
which this text belongs. Monthly progress of the sun and moon
 II III v VI VII VIII IX x XII X III X IV XV XVI X VI!
f

o:
E;
-e
:;:.
~
o
....,
to
~
'<
0"
::l
~.
::l
~
(l>

S-
o
p...
'"
FIGURE5. Lunar ephemeris computed according to System B for consecutive new moons for the years S.E. 208, 209,
a>
and 210, and written in Uruk S.E. 209, month IX, day 18 (=103 BC, Dec. 22/3). [ACT No . 122 = BM 34580 + .. . .] .....
62 1. Babylonian Arithmetical Astronomy

is computed according to a zigzag function of parameters

M = 30;1,59°, d = 0;18,
m = 28;10,39,4°, J1 = 29;6,19,20,

and period P = 12;22,8,53,20 and with

360
- = 12;22,7,51, .. . ,
J1
where both last numbers are close to the canonical year length
of 12;22,8 months.
The remaining columns have functions analogous to those of
System A, but most of them are computed in isolation from the
rest. I shall only draw attention to column G, the eighth column
of a standard ephemeris. Here, as in System A, the time from one
conjunction to the next is 29d + G + J, where G's period is the
anomalistic month, and J's the year. Column G is a zigzag func-
tion of mean value
J1 = 3,11;0,50° = 0;31,50,8,20 day ,

to convert time degrees into days (1d = 6,0°). Since column J has
0° as its mean value, we then have the following Babylonian
value for the mean synodic month:
29 days + J1 = 29;31 ,50,8,20 days .
I draw attention to this parameter, for it is precisely the value
adopted by Hipparchus and, following him, by Ptolemy for the
length of the mean synodic month (Almagest IV, 2), as already
Kugler observed. This is but one example of the pervasive influ-
ence of the Babylonians on Hellenistic Greek astronomy, and
one of many supports for my conviction that all subsequent
activity in scientific astronomy-if not in the exact sciences-is
descended in direct line from the creation of the anonymous
Babylonian astronomers.

Survival of Babylonian Methods

Had I submitted my manuscript a decade earlier, I would have
ended the chapter on Babylonian astronomy here, but in the
course of that interval startling new evidence about the trans-
 Survival of Babylonian Methods 63

mISSIOn of Babylonian astronomical knowledge to Hellenistic

Greece, or rather Egypt, has come to light.
In the works of Hipparchus and Ptolemy we recognize, as said,
many parameters of Babylonian origin, and Ptolemy quotes
Babylonian observations. However, we find no evidence what-
soever that they knew anything about the Babylonian arithmet-
ical methods. I myself believed, but it was no more than a belief,
that the mere existence of Babylonian ephemerides had con-
vinced Hipparchus that a quantitative, mathematical descrip-
tion of certain astronomical phenomena, capable of yielding nu-
merical predictions that could be tested against observations,
was both possible and desirable. Consequently, I thought, he
adapted Greek qualitative, geometrical models to quantitative
ends, inventing trigonometry for the purpose. For this, of course,
he did not have to know how these ephemerides were computed,
and whether he, or Ptolemy, had such knowledge, but chose not
to write about it in their surviving works, remained an unan-
swerable question.
So it still does, but others in the Hellenistic world certainly
had control of Babylonian methods, as we have now learned. In
1988 Neugebauer published a small fragment of a Greek papy-
rus, a photograph of which had been sent him (see Figure 6). It
contains a column of numbers that, Heaven help us, forms 32
consecutive lines of a column G of Babylonian lunar theory,
System B, the very column that, as I said earlier, has a mean
value in which Kugler recognized Hippachus's and Ptolemy's
length of the mean synodic month. Such a column makes no
sense in isolation, so it is a safe inference that the techniques of
one of the systems of Babylonian lunar theory were known in
Roman Egypt in the first half of the first century A.D . (the papy-
rus is dated by the handwriting).
This was, indeed, an astonishing discovery, but it was only the
tip of the iceberg, for in the 1990s Alexander Jones, the brilliant
young Canadian scholar, edited and analyzed a great number of
astronomical papyri in a series of papers, culminating in his im-
pressive Astronomical Papyri from Oxyrhynchus in a massive
volume in 1999.
These papyri from Roman Egypt, dating from the first three or
four centuries A.D., contain many wonderful and unexpected
things. Of interest here is that we find ample evidence for Bab-
ylonian schemes, not only for the moon, but for the planets as
well, competently adapted to the local calendar, when required.
64 1. Babylonian Arithmetical Astronomy

1. 4,2[2, 34, 10
4, 0, 4,[10
3,3[7, 34, 10
3,1[5, 4, 10
5. 2,5[2, 34, 10
2,30, [4, 10
2, 7,[34,10
2, 0, [5
2,22,[35
10. 2,45, [5
3, 7,3[5
3,30, [5
3, 52,3[5
4, 15, [5

15 4, 21,[19, 10
3, 58,[49, 10
3, 36,1[9,10
3, 13,4[9, 10
2, 51,[19, 10
20. 2, 28,[49, 10
2, 6,[19, 10
2, 1,20
2, 23,5[0
2,46,[20
25. 3, 8,5[0
3, 31,2[0
3,53,5[0
4,16,20

4, 20, [4, 10
30. 3, 57,3[4, 10
3, 35, [4, 10
3, 12,3[4, 10

FIGURE 6. The papyrus (P . Colker) that offered our first evidence of

knowledge in Hellenism of the technical details of Babylonian arith-
metical astronomy. Its provenance is not known, but it is written in a
hand of the first half of the first century A.D.
It contains remnants of two columns, the first of which has so
far defied restoration. The second column, to the right of the vertical
ruling, presents 32 lines of a column G of Babylonian lunar System B,
as described at the end of this chapter. A restored transcription is
appended.
 Survival of Babylonian Methods 65

Thus, we have learned that Babylonian arithmetical methods

survived well into Roman times, very likely kept alive by the
needs of astrologers, but without leaving any trace in the as-
tronomy of the Ptolemaic tradition. Here we may recall the
opening lines of Ode I, xi, by Horace (65-8 B.C.), the lovely ode
that contains the admonishment carpe diem:
Tu ne quaesieris-scire nefas-quem mihi, quem tibi,
finem di dederint , Leuconoe, nee Babylonios
temptaris numeros 0 0 •

[Do not ask, Leuconoe (it is impossible (impious) to know), what end
the gods have ordained for me, for you, nor put the Babylonian num-
bers (tables) to the test oj 0 0
2
Greek Geometrical Planetary
Models

Greek Geometrical Models

Higher Greek mathematics is mostly concerned with geometry,
so it is not too surprising that the Greek detailed planetary
models were geometrical. The aim of such models was at first to
mimic the behavior of a planet, which, in the case of Saturn, is
indicated in Figure 1, with latitude exaggerated four times.
The figure does not show that the Sun plays a particular role
for the planet's behavior: When Saturn is at the middle of its
retrograde arc, it is also in opposition, 8, to the Sun, that is,
its longitude differs from the Sun's by 180 The same holds for
0
•

Jupiter and Mars, the other two outer planets. However, an

inner planet is at the middle of its retrograde arc when it is in
what we call "inferior conjunction" to the sun. I should here
remark that the ancients divided the planets into two classes
identical with our inner and outer planets, but according to
the criterion of whether they can reach opposition (our outer
planets) or not (our inner planets, which are, as it were, tethered
to the sun on a short leash). Please note that I am not attempting
to explain these things. I merely state them as facts discoverable
by fairly short runs of sustained observations, as are the follow-
ing crude period relations.
Let us now examine Saturn's behavior a bit more closely (see
again Figure 1). It is a matter of simple counting and record-
keeping that 29 loops lead to just about one revolution in the
ecliptic in a time interval of some 30 years.
For the other planets similar relations hold, but with different
numbers: For Jupiter we find that 11 loops bring the planet
nearly once around the ecliptic in the course of some 12 years.
For Mars the situation is a little more complicated. Here the

66
A. Aaboe, Episodes From the Early History of Astronomy
© Springer-Verlag New York, Inc. 2001
 Greek Geometrical Models 67

.\3 60" 30"

• I I

,. <1 .\2 eclipt ic

I

-o----:.t:::=>'i'
I
r 0 1

e r 0
0----0-

FIG URE 1.

shortest time interval that leads to a near return of the loops in

the ecliptic is 32 years. In the course of this time span, a synodic
phenomenon, say, first stationary point <1> , skips around the
ecliptic twice, very nearly, taking with it 15 loops. However,
Mars itself takes an extra turn around the ecliptic when it
travels from one <I> to the next, so the planet will have performed
15 (the number of loops) plus 2 (the number of the phenomena's
revolutions), or 17, revolutions in the ecliptic in 32 years.
For Venus it happens that a very simple period relation is
also very good , and it has been independently recognized in
several ancient cultures. It states that in eight years Venus goes
through its synodic cycle five times and travels very nearly eight
times around the ecliptic.
For Mercury, a planet we shall ignore here, the crudest esti-
mate of these parameters is that three synodic periods corre-
spond very roughly to one revolution in the ecliptic or one year.
To return to the problem of Greek geometrical planetary
models. It is perhaps natural to construct such a model of two
components: one that makes the planet oscillate back and forth
and another that drags the former along the ecliptic. At any
rate, that is what wa s done in the two types of models we shall
now consider (the latter comes in two versions).

The Homocentric Spheres of Eudoxos

The first model is the homocentric spheres of Eudoxos, the emi-
nent Greek mathematician and philosopher who lived during
the first half of the fourth century B.C. In mathematics he is par-
ticularly renowned for his theory of proportions and of exhaus-
tion (integration), both of which we find in Euclid 's Elements ; no
work of Eudoxos is preserved in its or iginal form . His major
68 2. Greek Geometrical Planetary Models

FIGURE 2.

contribution to astronomy was brilliantly pieced together by

Schiaparelli from secondary and tertiary sources, mostly Sim-
plicius's commentary to Aristotle's De caelo ("On the heavens").
Before we proceed I want to make it clear that " homocentr ic"
means "concentric" or that the spheres have a common center
that further is occupied by the observer.
The model's first component, the one that produces an oscil-
lation, consists of two concentric spheres, the inner being able
to rotate relative to the outer, and the outer being able to rotate
relative to a frame that for the moment we shall consider as fixed
(see the very schematic drawing in Figure 2). Weare now inter-
ested in the behavior of a point (the planet) on the "equator " of
the inner sphere, in fact, the very point that is also on the
" equator" of the outer sphere when the motion begins.
We now let the inner sphere rotate with a certain angular
velocity (say, in degrees or rotations per time unit: day, month,
or what have you) relative to the outer sphere, and the outer
with the same angular velocity in the opposite direction but now
relative to the frame.
If the two axes of rotations were not tilted against each other,
but coincided, the " planet " would not move at all relative to the
frame, for the rotation of the outer sphere would cancel that of
the inner. Now, if we tilt the two axes a little against each other,
it seems reasonable enough that the " planet" will wiggle about
its initial position a little bit and with a period equal to the
common period of the two rotations. But here the intuition of
most people (I am among them) gives out.
 Greek Geometrical Models 69

FIGURE 3.

The path of the "planet" for various tilts of the two axes was
analyzed by Schiaparelli by means of spherical trigonometry and
later by Neugebauer, who used only mathematical techniques
available to early Greek geometers. This path turns out to be a
curve on the spherical surface that looks much like a figure-
eight (see Figure 3)-its ancient name is a hippopede (a horse-
fetter, i.e., the rope loops used to hobble a horse's forelegs so it
cannot stray). In fact, the hippopede turns out to be the inter-
section of a sphere with a (small) right cylinder, tangent to the
sphere at one point as shown in Figure 3, which may have
been known in antiquity, for it is shown by ancient means by
O. Neugebauer in Exact Sciences in Antiquity. Here only one
sphere is drawn, but both axes and their " equators" are repre-
sented.
We now place this apparatus so that the vertical line of sym-
metry of the figure-eight is the ecliptic, and the whole thing is
given a forward motion in longitude (this may be achieved by yet
another sphere rotating, with the ecliptic as its "equator" rela-
tive to the fixed stars).
The third motion is direct or forward in longitude (toward the
left in Figure 4). The figure shows what the resulting path of the
"planet" would look like when observed from the common cen-
ter of the spheres for different forward motions. (I should warn
the reader that the horizontal scales are not to be taken as the
ecliptic-the ecliptic is a horizontal line parallel to the scales,
but bisecting the now-horizontal figure-eight.) At the top we
70 2. Greek Geom etrical Planetar y Models

C>C)
• i • i

---
200 180 160 140
ecliptic longitude
120 100 so

FIGURE
60

4.
---
40 20 o - 20 - 40 -60

have the hippopede itself, and as the forward motion increases,

we first see a curve that indeed looks somewhat like the appar-
ent path of a planet with its stationary points and retrograda-
tions. If, however, the forward motion is too swift, then the
"planet" can no longer become retrograde, but is merely slowed
down, as shown in the two lowest graphs of Figure 4.
When we toy like this with various velocities, we assume that
we can assign them at will. But when we deal with a specific
planet, that is not so. Indeed, the period of the motion in the
hippopede (i.e., of the rotations of the first two spheres) must
be the planet's synodic period, and the forward motion in longi-
tude must be the (mean) synodic arc per synodic period. Both
of these are determined within narrow bounds by even crude
period relations of the sort I just mentioned. Our only really free
choice in this model is, then, of the inclination of the two axes in
Figure 2.
Now , of the simple period relations I mentioned earlier, I must
emphasize the ones for our neighbors in the solar system, Venus
and Mars.
 Greek Geometrical Models 71

For Venus we have, as we saw,

5 synodic cycles ~ 8 years
~ 8 revolutions in the ecliptic of Venus,
and for Mars:
15 synodic cycles ~ 17 revolutions ~ 32 years.
In terms of Eudoxos's homocentric spheres, Venus must then
travel through its hippopede 5 times in 8 years, and the hippo-
pede must be carried around 8 times in the ecliptic in the same
8 years. Analogously, Mars must travel through its hippopede
15 times in 32 years, while the hippopede is carried 17 times
around in the ecliptic in the same 32 years.
It can be shown that in both cases the forward motion is so
swift that whatever the inclination of the axes of the two inner
spheres we must always end with a situation like that repre-
sented in the two lowest graphs in Figure 4: Venus and Mars
just cannot become retrograde, but are merely slowed down.
This result was already derived by Schiaparelli, but the proof,
though elementary, is rather messy, so I shall omit it and treat
more carefully the two models I shall consider next, where some
of the same issues arise.
Let me, however, already here reveal what it is I am after with
my arguments. We have become accustomed to consider the role
of a geometrical model of the motion of, say, a planet is that of
serving as a basis for computing the planet's position at a cer-
tain time in some relevant coordinate system. So it has, in fact,
been since, and including, the work of Ptolemy.
Now, deriving a planetary position (say, in ecliptic longitude
and latitude) from an Eudoxian model would surely involve
spherical trigonometry, a subject that very likely was not ad-
dressed seriously before Menelaos (first century A.D.).
For this reason alone-that the required mathematical tech-
niques for extracting numerical coordinates from the Eudoxian
models were not at hand until well after the models' invention-
we could conclude that the models' original purpose was quali-
tative rather than quantitative.
But even if that had not been so, the qualitative character of
these models would be amply established by their inability to
make Mars and Venus retrograde.
There is a hint, albeit very faint, that someone, probably
Eudoxos or his follower Calippus (Kalippos), was aware that the
72 2. Greek Geometrical Planetary Models

Martian model fails to produce retrogradations. According to

Simplicius, the period of Mars in its hippopede was taken to be
260 days, or precisely one-third of 780 days, which is a good
value of Mars's synodic period. Indeed, our crude relation
before, that 15 synodic periods correspond to 32 years (i.e.,
32 ·365.25 = 11,688 days) yields 11,688/15 = 779.2 days for this
parameter. Since Mars in the same 32 years revolves 17 times in
the ecliptic, the hippopede must be carried once around the
ecliptic in 11,688/17 = 687.5 days.
If we maintain this last carrying motion but adopt 260 days as
a synodic period-the period of travel through the hippopede-
the new planet can very well become retrograde, but we shall
now have three retrogradations when Mars has but one .
The motive for the adoption of the shorter period of the
motion in the hippopede may then have been a desire for
obtaining retrogradations, but at the cost of getting thrice as
many as one should. If this were so, someone must have been
aware of the deficiency of a Eudoxian model for Mars.

Epicyclic Models
The next model we shall consider-the epicyclic model- is
of unknown origin. We know from Ptolemy that Apollonios of
Perga, the great mathematician who wrote on conic sections
about 200 B.C. , proved an elegant theorem about stationary
points in epicyclic models, so the invention of this kind of model
antedates Apollonios if it is not, in fact, his own, which I doubt.
Consider now Figure 5. The plane of the paper is the plane of
the ecliptic viewed from the north. The observer sits at 0 on the
earth, and the vertical ray from 0 points to a certain fixed star.
The point C travels uniformly, relative to the direction from 0
to the star, and counterclockwise on a circle-the deferent-
with 0 as its center. The planet P in turn travels on a circle-
the epicycle-with C as its center and a radius CP less than that
of the deferent (OC) . The planet P's motion on this second circle
is also uniform, and we reckon it, as the ancients always did,
from the deferent's radius OC or its extension to A; A is called
the epicycle's apogee (the point farthest from the earth), and II
its perigee (the point closest to the earth).
We recognize here, once more, the two components of a plan-
etary model, one that makes the planet oscillate around a mean
position and the other that carries the first along the ecliptic.
 Greek Geometrical Models 73

FIGURE 5.

The former component is the epicycle; indeed, if we arrest C and

observe P's motion from 0, we shall see the epicycle edge on ,
and P will seem to oscillate in the ecliptic back and forth around
C. The period of this oscillation ought then to be the synodic
period-for Venus, 5 revolutions in 8 years, and so on.
The latter component consists of the motion of C, and with it
the entire epicycle, along the deferent. The rate of this motion
must be the mean synodic arc per synodic period-for Venus 8
revolutions in 8 years, for Saturn 1 revolution in some 30 years,
and so forth.
Before we proceed I should emphasize that if all we are inter-
ested in is the direction from the observer 0 to the planet P, the
absolute size of the model is of no concern to us. In fact, Ptolemy
shows in Almagest III, 3 that the direction OP remains the same
if you subject the model to a dilation or contraction (i.e., a simi-
larity transformation) with respect to O.
We shall then follow Ptolemy when we use OC, the radius of
the deferent, as our unit for measuring the dimensions of the
model-he, however, employs the sexagesimal system and calls
the radius of the deferent 60 parts.
Let us now return to Figure 5. ' As already said, we shall
normalize and set OC = 1, and the epicyclic radius CP = r < 1,
measured with OC as unit.
As I indicated in the figure, P can in principle move clockwise
or counterclockwise on the epicycle, relative to the deferent's
radius OC or its extension to the epicycle's apogee.
74 2. Greek Geometrical Planetary Models

If P moves clockwise-we shall call this sense of rotation

"wrong," for reasons to be set forth later-the planet ought to
become retrograde when it is near the apogee A of its epicycle.
If it moves counterclockwise-the "r ight " way, as we shall
see-it becomes retrograde when in the neighborhood of the
epicycle's perigee.
We have, in fact, evidence for the use of both senses of rota-
tion on the epicycle. Our main source for the wrong sense is a
long astrological Greek papyrus, Papyrus Michigan 149, written
in a hand most likely of the second century A.D., that is, about
the time of Ptolemy. However, that is not to say that the astro-
nomical parts of the papyrus reflect the high level of Ptolemy's
work. On the contrary, P. Mich. 149 preserves, encapsulated in
fossil form, as it were, purely astronomical parts that represent a
variety of older and more primitive astronomical schemes and
theories, even though these different schemes are not at all
compatible. This situation is quite common in astrological texts,
and that is why they are of great interest, not only to historians
of astrology, but to historians of astronomy as well.
In P. Mich. 149 we find a passage describing the behavior of
Venus during one synodic cycle in terms of the planet's distance
from the Earth. The description is a bit vague, but it leaves no
doubt whatsoever that the planet is assumed to move the wrong
way around on its epicycle. We would expect a corresponding
passage for outer planets, but here the papyrus is destroyed.
Even so, it is reasonable to assume that for outer planets, too,
the sense of rotation on the epicycle was the wrong way .
We find support for this assumption in the elder Pliny's ac-
count of astronomical matters in the second book of his Natural
History. Pliny's text shows many similarities to P. Mich. 149; but
in this instance it would by itself offer but weak evidence, for
Pliny's understanding of astronomy was faint, and his presenta-
tion consequently obscure.
Thus, there is incontrovertible evidence for models for Venus,
and likely for Mars as well, with the planet moving the wrong
way around on the epicycle, and we shall now proceed to show
that such models must have been entirely qualitative.

Wrong-Way Epicycle
Let us now consider an epicyclic model with the wrong sense of
rotation on the epicycle. We saw that in such a model the middle
 Greek Geometrical Models 75

FIGURE 6.

of the retrograde arc must be reached by the planet P when it is

at the apogee of the epicycle (see Figure 6).
Now let the center C of the epicycle move on the deferent so
that the radius from the observer 0 at the deferent's center to C
rotates counterclockwise about 0 with constant angular veloc-
ity J.L2 relative to OC or its extension. We follow Ptolemy and
normalize the model and call the deferent's radius 1 (Ptolemy
uses 60); measured with this unit of the epicycle's radius is r
«1). As we said before, the values of J.LI and J.L2 are determined
within rather narrow limits by simple period relations, and let
them be measured with the same unit, say, radians per time unit,
or rotations per time unit, or what you will-the choice of the
common unit is irrelevant.
Now let the planet P be at the apogee of its epicycle (see
Figure 6). The velocity of P now has two components: one in the
forward direction due to the motion of OCP in the amount of
OP · J.Ll = (1 + r) . J.Ll,
the other in the backward or retrograde direction due to the
motion of P on the epicycle in the amount of
76 2. Greek Geom etrical Planetary Models

CP . /l 2 = r . /l 2,

and the two components are along the tangent to the epicycle
atP.
Now , if the planet is to be retrograde, we must clearly have
that
r/l2 > (1 + r) . /ll (1)
or

/l2 > (~+ 1) ./ll ' (2)

Since r < 1 and positive, l/r> 1, so the inequality (2) implies

that

(3)

So if
(4)
an epicyclic model with the planet traveling the wrong way on
the epicycle cannot possibly yield retrogradation.
For Venus and Mars we even find that /ll > /l2, so the in-
equality (4) is amply satisfied, and thus these two planets will
not become retrograde no matter what value of r < lone adopts.

Epicycles and Movable Eccenters

Ptolemy wrote his great handbook on theoretical astronomy in
Alexandria circa A.D. 150. He named it The Mathematical Sys-
tematical Treatise, but we usually call it the Almagest. It is a
work in 13 books (or long chapters) of which the last 5 (Books
IX-XIII) are devoted to planetary theory.
Ptolemy's final planetary schemes are refined versions of epi-
cyclic models (with the right sense of rotation on the epicycle,
for which he argues in Alm. IX) and I shall describe them later.
Here I am merely about to argue that a simple epicyclic model is
a very reasonable means of accounting, in first approximation,
for a planet's behavior when viewed from the earth, and I shall
identify the elements of such models with their modern counter-
parts.
 Epicycles and Movable Eccenters 77

Figure 7(a) represents the standard epicyclic model of a planet

with the plane of the paper being that of the ecliptic viewed from
the north. The observer on the earth is at 0, the direction OA
points to some fixed star, the epicyclic center C revolves on a
circle of radius 1 (= 60 parts) so that the line OC rotates uni-
formly at the rate of J-ll degrees per day relative to the sidereally
fixed direction from 0 to A. Finally, the planet P moves on the
epicycle of center C and radius CP = r, so that CP rotates uni-
formly at the rate of J-l2 degrees per day relative to the extension
of OC beyond C. We should recall here that the planet is in the
middle of its retrograde arc when it is at the perigee (the point
nearest the earth) of its epicycle.
Ptolemy mentions in several places a model employing a
movable eccenter as a completely equivalent alternative to an
epicyclic model. His most detailed account of these matters is
found in the Almagest, Chapter 1 of Book XII, which he devotes
to the theory of retrograde motion. It is here that he gives
Apollonius of Perga credit for a beautiful, but far from trivial,
theorem on the determination of stationary points.
One can crudely and briefly say that a model with a movable
eccenter is an epicyclic model with a small deferent and a large
epicycle. I have drawn such a model in Figure 7(b)-in fact, it's
equivalent to the epicyclic model in Figure 7(a). The planet P in
Figure 7(b) moves on the large circle of center C' different from
the observer O-hence eccenter-while C' in turn moves on the
small circle of center O-hence movable eccenter.
In order to transform the epicyclic model in Figure 7(a) into
its equivalent in Figure 7(b), we must complete the parallelo-
gram spanned by OC and CP by adding C' . We observe that
OC' = CP= r,
so C' is always to be found on a circle of center 0 and radius r.
Likewise, P will always be on a circle of center C' and radius
OC= 1.
By comparing angles in Figures 7(a) and (b), we learn that C'
moves on the small circle-the concenter-so that the radius
OC' rotates counterclockwise with the angular velocity J-ll + J-l2
relative to the direction from 0 to the same fixed star we used in
the epicyclic model. Furthermore, P moves on the eccenter (the
large circle) so that the radius C' P rotates clockwise with the
angular velocity J-l2 relative to the extension of OC' beyond C' .
78 2. Greek Geometrical Planetary Models

FIGURE 7.
 Epi cycles and Mov abl e Eccenter s 79

If we start the motions in the two models properly, we shall

afterward always obtain the same direction and the same dis-
tance from the observer 0 on the earth to the planet P no matter
which version we us e.
I shall now show by three examples how period r elations of
the sort we have considered in varying contexts immediately
imply values of the angular velocities f.LI and f.L2 . My aim is not to
reproduce Ptolemy's derivation of his refined parameters, but
merely to illustrate their interdependence by consider ing very
simple, but reasonable, relations and, most importantly, to once
again emphasize the special role the sun plays in planetary
models.
Let us first consider the relations for Venus:
5 synodic cycles ~ 8 revolutions in the ecliptic ~ 8 years.
We seek to transform these data into other units, degrees and
days, and observe that on e synodic cycl e corresponds to one
revolution of the planet on its epicycle relative to the exte nsion
of OC-the planet becomes r etrograde in the neighborhood of
the epicycle 's perigee-and that one revolution of the synodic
pattern in the ecliptic means one revolution of the epicycle's
center C in the deferent relative to the direction from the ob-
server at 0 to a fixed star (we recall that the plane of the paper
for our models is that of the ecliptic). Additionally, on e r evolu-
tion in the respective circle corresponds to 360°, and for our
present purpose we set
1 year = 365 t days
(Ptolemy uses 365t - 1/300 da ys). For Venus we now have

_ 8· 360° _ . ol d
f.LI - 8 . 365 1 d - 0,59,8, .. .
4

[this is also the (mean) motion of the sun around the observer]
and
_ 5 .360° _ . ol d
f.L2 - 1d - 0,36,57, .. .
8 ·365-4
Analogously, for Mars we have
15 synodic cycles ~ 17 revolutions in the ecliptic ~ 32 years,
whence
80 2. Greek Geometrical Planetary Models

_ 17 · 360
0
_ • ol d
1t1 - 1d - 0,31,25 ...
32.365 4
and
_ 15 · 360
0
_ ol d
•
1t2 - 32 . 365 1 d - 0,27,43, . . .
4

Third, the simplest reasonable relations for Jupiter are

11 synodic cycles ~ 1 revolution in the ecliptic ~ 12 years,
and so
0
_ 360 _. ol d
1t1 - 1d - 0,4,55, ...
12 ·365-4
and
_ 11· 360
0
_ • 0 Id
1t2 - 1d - 0,54,12, . . .
12 ·365-4
For the sake of comparison, I listed in Table 1 Ptolemy's values
of 1t1 and 1t2 abbreviated to two sexagesimal digits. * I appended
the sums of 1t1 and 1t2 for the outer planets in the last column; all
three are the (mean) daily progress of the sun.
Thus, for an inner planet, the center of its epicycle [C in
Figure 7(a)] travels so that the line from the observer to it, OC,
rotates with the same angular velocity as the sun or, in other
words, the angle between OC and the direction from 0 to the
sun is constant. This constant is 0, for when the planet is in the
middle of its retrograde are, it is
(i) at the perigee of its epicycle and
(ii) in inferior conjunction with the sun.
In this configuration the observer, the planet, the center of its

*If I had given Ptolemy's values to three or more sexagesimal places,

I would have had to correct for precession, for his mean daily motion
in longitude of the epicycle's center (what we shall call m1 later) is
counted from the the vernal equinox-his longitudes are tropical, as
we say. However, our /11 describes mean motion relative to the defer-
ent's apogee, which is sidereally fixed and so subject to precession.
We should then subtract a little less than 6 in the third sexagesimal
place from m1 to get /11, for that is the amount of precession in degrees
per day.
 Comparison between Ancient and Modern Planetary Models 81

TABLE 1.

ff ~:. r: + f1"J.
SAtwJ\. tz
0/_
OJ Z, 0/ .•. OJ 57,7,•.• Oi 59/ 8, •.•
J7Afittr4 0; 1.t,59, •.• 0;5'1,9.... OjS'9,S, ...
MN-s d 0; 31,2', ... OJ 27, '(t, .•• 0; 5~J 8,•..
V(1W.$ 9 OJ 59,S, .•• Oj36, S~ ...
Mer-cura ~ 0; 59,5/ ... 3; b, l'f J . ..

SW1. 0 OJ S9, S, ...

epicycle, and the sun must therefore be in a straight line. In

a
particular, the directions from to C and from to the sun are a
equal in this situation, but then they must always remain so, for
the constant must be 0, as stated. Thus, for an inner planet the
a
line from to C always points to the sun.
We can argue in an analogous manner that for an outer planet
the epicyclic radius-CP in Figure 7(a)-is always parallel to
the direction from the observer a to the sun. First we remark
that the two directions differ by a constant; this is most easily
seen from the movable eccenter in Figure 7(b), where Ot)',
always parallel to CP, rotates with the angular velocity f.Ll + f.L2,
which for an outer planet equals the mean daily angular prog-
ress of the sun (see Table 1). Next we observe that this constant
must be 0, for when the outer planet is at the middle of its ret-
rogradation it is also
(i) at the perigee of its epicycle and
(ii) in opposition to the sun.
Thus, the role of the sun in an epicyclic planetary model is that
it guides the center of an inner planet's epicycle along the def-
erent, and an outer planet around its epicycle.

Comparison between Ancient and Modern Planetary

Models
For a proper appreciation of the Greek planetary models it is
important to recognize that they are more than a mere collec-
tion of ad-hoc devices that can reproduce the planets' apparent
82 2. Greek Geometrical Planetary Models

behavior. In fact, when appropriately scaled, they turn out to

be correct representations of the planets' motions relative to the
earth, in distance as well as in direction.
We shall continue for a while longer to ignore the deviations
from uniform mean behavior, as we have done so far in this
chapter. The modern mean model of the solar system then has
the planets move uniformly, with respect to the direction from
the sun to a fixed star, in circular orbits that have the sun as
their common center and that all lie in one plane, that of the
ecliptic. Mercury has the smallest orbit and then, in order of
increasing distance from the sun, are Venus, the Earth-now a
planet, though very special to us-Mars, Jupiter, and Saturn,
to consider only the planets known before the 18th century
(W. Herschel discovered Uranus in 1781).
Matters would then be very simple indeed if this very sim-
plified solar system were viewed from the sun. Fortunately, we
do not observe from the sun, but from the earth-a much more
comfortable place to live-so our task will be to transform this
heliocentric system so it displays how the planets appear to
behave to an observer on the earth.
Let us first consider the case of the sun. In Figure 8(a) I have
shown eight successive positions of the earth, E 1 , E 2 , . . . , E s, in
its simplified orbit, a circle with the sun 8 as its center. Let us
assume that the ray from 8 to E 1 points to a certain fixed star in
the ecliptic; the eight positions are drawn so that the angle sub-
tended at 8 between one of them and the next is always the
same, but then the time it takes the earth to pass from one posi-
tion to the next is always the same, too, for it moves uniformly in
its path relative to the direction from the sun to a fixed star. The
arrows from the various positions of the earth all point toward
the sun and are equal in length to the radius of the earth's
orbit-we call this mean distance of the earth from the sun the
astronomical unit.
We now translate this information into a coordinate system
where the earth is the origin-a geocentric system-to describe
how matters appear to us. In Figure 8(b) the earth is at E, and
we know that the sun is at the tip of the arrows beginning at
E 1 ,E2 , ... .E« in Figure 8(a). We draw them from E and obtain
the corresponding positions 8 1 ,82 , . . . , 8 s of the sun. These eight
positions lie on a circle of center E and with a radius equal to
one astronomical unit. They are equally spaced in time as in
direction, so the sun moves uniformly with respect to the direc-
 Compariso n between An cien t and Modern Planetary Models 83

tt)
£,

----0£7

----..0 53

5,

F IGURE 8.
84 2. Greek Geometrical Planetary Models

tion from E to say, 8 5 or any other direction that remains fixed

with respect to ES5 • Now SEl in Figure 8(a) points to a certain
fixed star, as we said, so ES 5 in Figure 8(b) points to the same
star. However, this direction from the earth to the star does
not change with the place of the earth in its orbit because the
astronomical unit is negligible in comparison with the distance
to even the nearest star. (Very strictly speaking, there is a
change-we call the effect of the earth's movement on the
direction from it to a star the "annual parallax of the star"-but
we are dealing with quantities of less than one second of arc
and so at the very limit of what is observable with even the most
refined modern techniques.) Thus, the sun will move uniformly
in its orbit in Figure 8(b) with respect to the direction from the
earth to a fixed star-I tried to illustrate these things in the
small diagrams in Figure 8.
We now turn to the problem of describing how the planets
behave relative to the earth, and we shall consider the two
slightly different cases presented by inner and outer planets.
Figure 9 deals with an inner planet. In Figure 9(a) we have the
planet P moving in its orbit around the sun S in the center, and
the earth E moves likewise, but in a larger orbit. We wish to
transform this situation into a model that displays directly the
two things we are interested in: the direction and length of EP.
We achieve this in Figure 9(b). The sun S moves, as we have
seen, in an earth-centered circle, uniformly with respect to the
direction from the earth E to a fixed star, and the planet P in
turn moves in its orbit, uniformly with respect to the same
direction and so also with respect to the extension of ES beyond
S. We recognize immediately that here we have a simple epi-
cyclic model; the radius of the deferent is one astronomical unit,
and the epicycle's radius is the planets' (mean) distance from
the sun.
Figure 10 deals analogously with the case of an outer planet-
Figures 10(a) and (b) present the heliocentric and geocentric
pictures, respectively-but here the geocentric version turns
out to be a model employing a movable eccenter. The radius of
the small circle-the concenter-is the astronomical unit and
that of the eccenter equals the planet's (mean) distance from the
sun. If we prefer the equivalent epicyclic model (look back to
Figure 7), as Ptolemy does, we have then that the radius of the
deferent is the planet's distance from the sun, while the epicyclic
radius is the astronomical unit.
Comparison between Ancient and Modern Plan etary Model s 85

b) *

FIGURE 9. Inner planet.

86 2. Greek Geometrical Planetary Models

a,)

FIGURE 10. Outer planet.

 Comparison between Ancient and Modern Planetary Models 87

In the Almagest, Ptolemy does not commit himself to an abso-

lute size of his models. He is, after all, only interested in the di-
rection from the earth to a planet, for that is all he can observe,
and if you enlarge or shrink a model, proportionally in all parts,
you will not change this direction, provided you keep the earth
fixed. Ptolemy does assign absolute sizes to his models, but in
another work, the Planetary Hypotheses , where he is concerned
with cosmology, as we shall see in the next chapter.
Here I shall scale Ptolemy's models according to the correct
geocentric planetary models, which we have just considered, in
order to show that Ptolemy's values of the epicyclic radii are
very good indeed. I wish to emphasize that in so doing, we do
not violate the Almagest in any way, but merely add one more
hypothesis.
This hypothesis is simply that the radius of the deferent for an
inner planet is the astronomical unit, while for an outer planet
the epicycle's radius equals the astronomical unit (see Figures 9
and 10 and recall that the concenter's radius is equal to that of
the epicycle).
As said, Ptolemy always calls the radius of the deferent 1, or
rather 60 ptolemaic units (p.u.) or parts. It is perfectly consistent
with the Almagest to have the ptolemaic unit change from planet
to planet, as we are about to . The epicyclic radius is r p.u. We
then get for an inner planet
60 p.u. = 1 a.u. (astronomical unit) ,
so

radius of epic ycle = r p.u. = :0 a.u.

For an outer planet we have

radius of epicycle = r p.u. = 1 a.u. ,
so

radius of deferent = 60 p.u. = 60 a.u.

r
Thus, if Ptolemy's values for r are good, then r/60 or 60/r
should be close to the planet's mean distance from the sun,
depending on whether it is an inner or outer planet.
In the first column of Table 2 I listed Ptolemy's values for r,
written sexagesimally, for the five planets. The other numbers
88 2. Greek Geometrical Planetary Models

TABLE 2.
r 60
r - -1r a
60

I-z 6;30 - 9.230 .. . 9.538 . ..

lj. 11;30 - 5.217 .. . 5.202 . ..
d 39;30 - 1.518 . . . 1.523 . . .
9 43;10 0.719 . .. - 0.723 . . .
t:) 22;30 0.375 . . . - 0.387 .. .
r: epicyclic radius in ptolemaic units
a: modern mean distance sun-planet in astronomical units

are written decimally, and the agreement between the values

derived from Ptolemy's radii and the modern mean distances
is considerable. We saw, furthermore, that Ptolemy's angular
velocities (mean motions) are excellent, so the simple models as
a whole are indeed very good representations of the planets'
mean behavior with respect to the earth in all matters but one:
namely planetary latitude.
How Ptolemy arrived at his values for radii and eccentricities
-we shall introduce them later-is too tangled a tale for inclu-
sion here. I shall only mention that the Almagest is in this
respect no more of an autobiographical account than are most
modern scientific treatises. He chooses to present a somewhat
neatened version of what surely must have been a very messy
process of trial and adjustment, as he himself says (Almagest
IX, 2).

Ptolemy's Refined Planetary Models

The planetary models we have considered so far are mean
models in the manner of Ptolemy. They represent the average
behavior of the planets very well, as we learned, and since the
underlying periods are very good , the errors inherent in all ap-
proximations do not accumulate arbitrarily, even in the course
of long time intervals. In the small, however, they do not serve
as well: They imply, for example, that all retrograde arcs are of
equal length, and that is not so.
The Babylonians were aware of this fact and constructed
their arithmetical models so they could account for it. The
 Ptolemy's Refined Planetary Models 89

Babylonians considered, at least primarily, a planet only when it

was at one of its characteristic phases (for an outer planet they
are first and last visibility, first and second stationary points,
and opposition) and addressed the question: If a certain phase of
a planet takes place at a given time and longitude, when and
where will this planet's next phase of the same kind occur?
Since a planetary phase happens when the planet has a certain
relationship to the sun (this is most easily seen in the case of
opposition), the Babylonians had only a single inequality or
anomaly (the technical terms for the deviation from mean be-
havior) to contend with, one depending on longitude, as we saw.
Ptolemy, however, wanted his planetary models to enable him
to answer the following question: Given the time, where is the
planet? (We have demanded the same of our planetary theories
ever since.) He had, therefore, to consider that the planet could
have any relationship to the sun, so he had one more anomaly
to account for than the Babylonians-one depending on the
planet's position with respect to the sun-and that is precisely
the task his epicycle enabled him to perform.
Ptolemy introduces the anomaly depending on longitude by
changing the workings of the deferent in two ways. We recall
that the center of the deferent played a triple role in a simple
planetary model:
(i) it was the place of the observer;
(ii) it was the point from which the center of the epicycle
always had the same distance; and
(iii) it was the point around which the motion of the epicycle's
center was unifrom.
Ptolemy now refines such a mean model by assigning each
of the three roles to a different point, thus creating what we call
his equant model for a planet (see Figure 11).
In Figure 11 we have, as before, a deferent of center C and
radius 1 (= 60 parts, or ptolemaic units) and an epicycle of
radius r ptolemaic units with its center C' on the deferent and
on which we find the planet P. However, we now observe from
the point 0; the motion of C' is uniform around the point Q with
respect to the direction to a fixed star. The points 0, C, and Q
are on a straight line, the apsidal line, pointing to some fixed
star; C is in the middle and
OC = CQ = e ptolemaic units (e < 60).
90 2. Greek Geometrical Planetary Models

* t

c
e
C
e
o

I
FIGURE 11.

Furthermore, the motion of P on the epicycle is uniform, not

with respect to the extension of OC' beyond C' as before, but
with respect to the extension of QC' beyond C' .
This is Ptolemy's simple and efficient equant model for a
planet's motion in longitude-his devices for finding a planet's
latitude are very complicated and will not concern us here.
In the last chapter of this book, "Kepler Motion Viewed from
Either Focus," I try to justify why this model is so efficient; here
I merely describe how simply one may derive a planet's position
from it . In this I shall not follow what Ptolemy expects a reader
and user of the Almagest to do, although he supplies all the
necessary tools, including trigonometric tables and examples of
their use. Rather, he saves the reader a lot of dreary trigono-
metric calculations by providing auxiliary tables so that the
 Ptolemy's Refined Planetary Models 91

crtto<r

FIGURE 12.

most one has to do is interpolate in them to find corrections that

are to be added to, or subtracted from, mean positions that, in
turn, can be taken directly from his mean motion tables.
What I outline ahead is akin to what Ptolemy had to do when
he set up his tables, and it involves solving a sequence of tri-
angles by means of plane trigonometry.
In an earlier book, Episodes from the Early History of Mathe-
matics, I presented Ptolemy's trigonometry in some detail. Here I
merely recall that plane trigonometry is the branch of elemen-
tary mathematics that enables you to calculate, as accurately as
you please, the measures of all the pieces of a triangle-sides as
well as angles-when you are given the measures of three inde-
pendent pieces: all three sides; two sides and an angle; or one
side and two angles (but not three angles, for if you know two ,
you also know the third, since their sum is 180°). To that end we
now employ a family of closely related trigonometric functions:
sine, cosine, tangent, and cotangent. For a given angle, they can
all be calculated with any desired precision.
Ptolemy does not use any of these but instead uses the closely
related chord function: The chord of an angle a-we write crd
a-is the length of the chord in a circle of radius 60 parts
subtended by a central angle equal to a . We have then (see
Figure 12)

crd a = 2 . 60 . sin ~ = 120 . sin ~ .

92 2. Greek Geometrical Planetary Models

In my earlier treatment of Ptolemy's trigonometry I showed

how he had calculated a table of chords of angles from 0° to 180°
in steps of half a degree and how he used this table to solve all
cases of triangles with three given independent pieces. We shall
now see how this suffices to make his equant model for a planet
yield an answer to the following question: Given the time, what
is the planet's longitude?
We take for granted that the model is completely determined;
that means, first, that all its dimensions are known. We have
[see Figure 13(a)] the observer at 0, the center of the deferent at
C, the equant point Q on the extension of OC so that
OC= CQ=e;
the deferent's apogee is at A, its perigee at II. The center of the
epicycle is C' , its radius is r, and the planet itself is at P. The
radius of the deferent is set equal to 60 parts (60P ) , and e and r
are measured in such parts.
The angles that grow uniformly with time are marked 0: and 13
in Figure 13(a); in order to find them, we must know their rates
of growth as well as their initial values. Ptolemy did not, in fact,
find 0: directly, but as the difference between>' (C'), the uni-
formly growing mean longitude of the epicyclic center C', and >.
(A), the longitude of the apogee A, which Ptolemy, correctly
assumed to be sidereally fixed, and so subject to precession. I
show, in Figure 13(a), the direction ~ (C') as coming from Q, the
center of uniform motion.
We must then know ml and m2, the rates of growth of ~ (C')
and 13, respectively, as well as the values ~o (C') and 130 of these
two quantities at epoch (year 1 of Nabonassar, day 1 of month
Thoth, noon), when the apogee's longitude was >'0 (A). The apo-
gee, we recall, is subject to precession, and Ptolemy's value of its
rate is 1° per century. We observe that 0: also grows uniformly
and at the rate of ml diminished by the rate of precession (we
called the resultant angular velocity J.Ll above), but Ptolemy
chooses not to make use of this.
We should futher note that for Venus
ml = 0;59,8,17,13,12,31old,
Ptolemy's value for the daily mean solar progress in (tropical)
longitude, and for the outer planets
ml + m2 = 0;59,8,17,13,12,31old ,
 Ptolemy's Refined Planetary Models 93

•
A:

A\c') '-,_

A(C)·--raT~T:-:-+--L-l

~;::U:---.!...-l 0-
e
C
e
o

1T'

FIGURE 13. A & B

as a consequence of the special role the Sun plays in the models:

For Venus, QC' points toward the mean sun, whereas for an
outer planet the direction C' P is the direction to the mean sun.
We are now prepared to find >.(P), the planet's longitude at a
given moment T. Our first step is to calculate t, the time interval
from epoch (when t = 0) to T-for Ptolemy epoch is the begin-
94 2. Greek Geometrical Planetary Models

ning of Nabonassar's reign in -747 (746 B.C.). In the following

Excursus on Calendars and Chronology I describe how the char-
acter of the calendars complicates the problem of finding t, how
Ptolemy's Kinglist enables us to solve it, and the structure of the
Kinglist itself.
Next we obtain

A(A) = Ao(A) increased by 10 per century since epoch,

and

X(C') = Xo(C') + mI ' t,

so we get

a = X(C') - A(A),

and furthermore,

/3 = /30 + m2 . t.
These things having been given and found, we can now
dispatch our problem by solving three triangles in quick
succession.
First consider Figure 13(b). In 6.QCC' we have

LQ = 1800 - a,
CQ = eP ,
CC' = 60P ,

so we can calculate the angles u and v (and the side QC') .

Next, in 6.COC' we now know that

LOCC' = 1800 - v,
OC = eP ,
CC' = 60P ,
so we now can find the angles x and y and the side OC' . We set

1= u+x.

Third, we turn to Figure 13(a) and consider 6.0C'P, in which

we now have
 Excursus on Calendars and Chronology 95

C'P = r" ,
OC'-just found ,
LOC'P = 180 0
- (,8 + , ),
and so we can derive 8 (and OP).
Finally, we now have for the planet's longitude
>.(P ) = >'(A) +Y + 8
= >'(C') + 8,
and we are done.

Excursus on Calendars and Chronology

In everyday usage, "time" can mean either of two things: a
certain instant-"a point in time"-or "duration" or "time
interval." A common, and important, problem in astronomy is to
find the time interval between two points in time. This ought to
be easy enough, but often it is not-just try to calculate how
many days you have lived so far . The problem is, of course,
caused by the structure of our calendar with its leap years and
months of different lengths-28, 29, 30, or 31 days.
The Mesopotamian calendar was worse in this re spect; as we
saw , a date was given in the form
year n of King N , month x, day y ,
and a moment was indicated by a number of time degrees before
or after sunset or sunrise. The calendar was strictly lunar-a
new month began at sunset on the evening when the new moon
could first be seen, so a month was either hollow or full, that is,
it was either 29 or 30 days long. Furthermore, a calendar year
consisted of either 12 or 13 such months, and these years of dif-
ferent lengths were not distributed according to a fixed pattern
(the " Metonic" cycle of 19 years) until the last 400 to 500 years
B.C.
In order to find the precise number of days between two Mes-
opotamian dates, one would then have to know the character of
all the intervening years and months. It is clear that such infor-
mation was to be had, for in the Almagest we find Babylonian
observations presented in the splendidly uniform Egyptian cal-
endar, about which more below.
96 2. Greek Geometrical Planetary Models

Ptolemy says that Hipparchus "arranged" the Babylonian

observations, and "arranging" may well mean that he saw to it
that the Babylonian dates were translated into their Egyptian
equivalents. The Egyptian calendar was very simple, and therein
lies its virtue. The Egyptian year was invariably 365 days long,
and it was divided into 12 months of 30 days each, and 5 extra
("epagomenal") days-no exceptions, no intercalations. One
may object that this year is too short, and that the calendar is
therefore not "right," but there is, of course, no such thing as
one, or even a, "right" calendar. The most important things
about a calendar are that everyone within the culture agrees on
it and, next, that it more or less does what you want it to do. We
think it important to keep the year in step with the seasons,
within a day or so, and so we have to contend with years with
different numbers of days, for the tropical year is, alas, not a
whole number of days long.
The Egyptian choice of a constant year of 365 days is clearly a
gesture in the direction of the tropical year, but it is about a
quarter-day too short. In the course of a life-span of 80 years-
rarely attained in antiquity-the date of, say, the vernal equinox
would then creep some 20 days forward toward higher dates
through the month or months, but this was a small price to pay
for the rigid regularity of the calendar, at least from an astron-
orner 's point of view .
The year in which an event took place was, as said, given as a
regnal year of the reigning king of Babylon, and so it was in
Hellenistic Egypt. To make use of this information, you must
obviously know when the king reigned, and you can learn that
from Ptolemy's Kinglist, which he appended to the Almagest. In
it he presents the sequence of successive kings, first of Babylon
and then of Egypt, and the lengths of their reigns in Egyptian
years, as well as a running total, beginning with Nabonassar,
king of Babylon (-746 = 747 n.c.). (See Figure 14.)

After Ptolemy; Copernicus, Brahe

Ptolemy's Almagest, composed in the second century of our era,
has reached us in the original Greek in Byzantine manuscripts,
transferred to parchment from the less durable papyrus. The
printed editions of the Greek text-the best and most recent was
delivered by the Danish classical scholar J . L. Heiberg from 1898
 After Ptolemy; Copernicus, Brahe 97

to 1903-as well as the English translation by G. J . Toomer

(1984) are based on such manuscripts from the ninth and tenth
centuries.
The Almagest remained the central astronomical treatise for
nearly a millennium and a half whereever serious positional
astronomy was cultivated, also outside the Greek-speaking
domains, first in the Islamic world and later in medieval and
Renaissance Europe. That is not to say that it was slavishly
accepted and passed on . Indeed, we find both improved parame-
ters and alternative geometrical models suggested by later as-
tronomers, but the Almagest remained the touchstone against
which they tested such models.
Indeed, around A.D . 1000 Islamic astronomers began subjecting
the Almagest to constructive criticism and emendation, on both
astronomical and philosophical grounds-it had been translated
several times into Syriac and Arabic in the late eighth and
ninth centuries. It was a Latin translation of one of these Arabic
versions, made in Toledo in 1175 by Gerard of Cremona, that in
effect introduced the Almagest to the West; earlier Latin trans-
lations from the Greek had seen scant use . What Ptolemy called
The Mathematical Systematic Treatise (Syntaxis) now got the
Latin name Almagesti or Almagestum, a transliteration of an
Arabic term that, in turn, probably was a transliteration of the
Greek megiste (syntaxis), meaning "the greatest (treatise),"
though this appropriate title is found in none of the surviving
Greek manuscripts.
I give but a few examples of such emendations and criticisms;
simplest among them is the introduction of improved values of
parameters, particularly those that measure extremely slow
rates of change, for to determine such slow changes well, one
must observe and keep records over long time intervals. Indeed,
Ptolemy was well aware of this and invited his successors to
improve what he himself had had to be satisfied with. Notable
among such slow changes is the " precession of the equinoxes,"
the extremely slow increase in (tropical) longitude of the fixed
stars. Hipparchus, who discovered the phenomenon, had esti-
mated it to be at least 10 per century, and Ptolemy adopted that
value, a nice round number, easy to remember, but too low-the
correct rate is l Oin some 71.6 years. Islamic astronomers pro-
posed values closer to the modern one.
A related problem concerns the apogee of the solar model.
Hipparchus described the behavior of the sun by a simple eccen-
98 2. Greek Geometrical Planetary Models

/ ;:::::=-

f WII M I T U M V .tl
.\ IIITOlllO:T1t Tllti
I , 1 ~ ," r1t C " t,
J "~ lt,WM ll II> I.. ,
l.r'-t- -
I
I tc H·
\h l " ~~. ! ~ It . II 07,
I fVW !loii t :;:- -- . 1(1 It
I ,~i ""I "" "'J _ 11 .1 1..
. _"I t~ H
" I "'f"ll" 1 II}., tM1'
It 1,-,-Mt..IH'1·wt
" "' WITH I> 1<+ t 7.

m
(W TII " . , IMI
tHICl-IOtWlt t I 11+
,-.!.~ fO m:rt~ _ In. I~l>,
M r T" ,
I, to "11 '
A. T ~
1'~ .· -I07
1 IA 'r 'l

__ '~~lliJ
f~. y~
" _.YKl
__ I ''t 'H
l ie" v~
, ll r 'nr
..--
I ~ t. II

j ~ !IA
A. ....
H' ,1>1\
r .h,

FIGURE 14. Ptolemy's Kinglist. Figure 14 shows Ptolemy's Kinglist as it

appears on folio 16, verso, of the Byzantine manuscript Vaticanus
graecus 1291, written in the first half of the ninth century. The manu-
script contains otherwise a version of Ptolemy's Handy Tables, his
elaboration and extension of the Almagest's tables, occasionally with
altered parameters.
The first set of three columns of the Kinglist is headed "Years of
the Kings." The first 20 lines concern 18 kings of Babylon in order of
their consecutive reigns, beginning with Nabonassar and ending with
Nabonaides, and two interregna. The next 10 lines have the subhead-
ing "Persian Kings" and list the rulers of the Persian dynasty, begin-
ning with Cyrus and ending with Dareius III; in the last line we find
Alexander "the Founder."
The second set of columns is headed " [Years] of those after
 After Ptolemy; Copernicus, Brahe 99

tric model; it has the sun moving in a circle, uniformly around

the center, which is outside the earth. He assumes that the
apogee-the point where the sun is farthest from the earth-is
tropically fixed, that is, its longitude measured from the vernal
equinox remains constant, and Ptolemy adopted Hipparchus's

Alexander the Founder" and begins with Alexander's two successors

as king of Babylon, Philip (Arrhidaeus) and Alexander (the younger).
Then the list changes to the Macedonian rulers of Egypt, all but the
last called Ptolemy, and here they are given only their cognomina,
Lagos, Philadelphos, Euergetes I, Philopater, etc. Ptolemy I, once
Alexander's Companion in Arms, and founder of the dynasty, is here
identified as Lagos, his father's name, and not as Soter, his more usual
cognomen. The last of the line is Cleopatra and then, as noted in the
margin, the list gives the names of the Roman "kings," Augustus,
Tiberius, Gaius, and so forth.
The second column of each set records the length of each reign,
in Egyptian years of 365 days , and the third keeps a running total,
beginning with Nabonassar's reign. In this version of the Kinglist, the
total ends at 424, including Alexander's 8 years, and begins afresh with
Philip's reign-not only was an Era Nabonassar in use, but also an
Era Philip. [Year 1 of Era Nabonassar, month Thoth of the Egyptian
calendar, day 1 corresponds to -746 (= 747 B.C.), February 26.]
Ptolemy's Kinglist or Canon was long the principal key to ancient
chronology, and the Babylonian part has on the whole been confirmed
by the cuneiform evidence as it became available in the course of the
last century and a half.
Those who would enjoy reading what was written some 1200 years
ago might need to know the numerical values assigned to the Greek
letters. They are given in Table 3, where the classical Greek alphabet
has been augmented by three obsolete letters: digamma (the old waw),
also called stigma, for 6; qoppa for 90; and sampi for 900.
The first two lines say
NABONAZAPOY I~ I~
NA~POY B IS
which means
of Nabonassar 14 14
of Nadros 2 16.
For a recent reconstruction of the Kinglist, see G. J . Toomer's
Ptolemy's Almagest, p. 11.
100 2. Greek Geometrical Planetary Models

TABLE 3.

AD' 1 I " 10 P P '00

e ~ z K"2c !: if 200
r ~ 3 I\~~ T t' ~OO
~ ~ 4 M ~4C Y V 400
E( 5 N '\I So 4' 'J' 500
F ~ 6 ::: ; 60 X 'X, be"
Z ; 7 o 0 70 'l' V 700
H 1\, s n1Y'"80 .n. ~ 8oc
e f 9 a 4 90 ~ " 900
'Z5' , 11) er".. 0

model without change. Several early Islamic astronomers real-

ized that the solar apogee is certainly subject to precession, just
like the planetary apogees, and some, at least as early as al-
Biruni (ca. A.D . 1025), even detected a further proper motion.
Two further examples will suffice, but they concern more sub-
stantial and serious criticism of Ptolemy's work. In one of them
the criticism is justified, in the other not. The first concerns his
lunar model, which, alas, is too complicated for inclusion here.
I shall merely say that in order to account for the fact that
lunar anomaly (the moon 's deviation from uniform behavior) is
larger at quadrature than at syzygy, Ptolemy introduces a crank
mechanism in his model that pulls the epicycle nearer at quad-
rature to make it appear larger-indeed, nearly twice as large.
Ptolemy conveniently does not mention that the moon itself
consequently ought to look twice as large at quadrature than at
syzygy (when it is new or full), and anyone can see it does not.
This serious flaw was removed by the replacement of the crank
linkage by other geometrical devices, one of which-the Tus1
couple-I mention later.
The other criticized deficiency was considered much more
fundamental: When he introduced the equant model, Ptolemy
committed an act of philosophical heresy. Aristotle had argued
that the only self-sustaining natural motion was uniform circu-
lar motion, and his argument had been so effective that natural
philosophers, including Copernicus, remained persuaded for
nearly two millennia. Ptolemy himself pays his respects to the
principle of uniform circular motion in the preface to the Alma-
gest, but he blithely violates it when he constructs satisfactory
models for the moon and planets. To be sure, the motion on the
equant model's deferent is both circular and uniform, but it is
 After Ptolemy; Copernicus, Brahe 101

FIGURE 15.

not uniform around the circle's center, and so not of constant

speed, and that would not do for a proper Aristotelian.
Medieval Islamic scholars expended much effort and ingenu-
ity on the construction of models in which combinations of truly
uniform circular motions very nearly did the same job as the the
equant model. Indeed, there was never any question about the
equant model's efficiency, and the test for a philosophically cor-
rect model remained that it produce nearly the same positions as
an equant model.
The earlier center for astronomical activities of this kind was
at Maragha in northwest Iran with Nasir aI-DIn al -Tusl (1201-
1274) and his school. In the 14th century the brilliant Ibn al-
Shatir of Damascus wrote a treatise proposing models that were
composed entirely of uniform motions and that worked as well
as Ptolemy's.
One important component of these structures is what we now
call a "TusI couple" after its inventor in Maragha. This lovely
and simple geometrical device produces, surprisingly, a har-
monic oscillation in a straight line as a result of only two com-
bined uniform circular motions and can be presented in two
equivalent forms.
In Figure 15 let A, 0, and B be points on a fixed straight line
such that
AO = OB = 2a.
102 2. Greek Geometrical Planetary Models

Consider now the circle with center a and radius a and another
circle with its center C on the circumference of the first circle
and the same radius a. Let X be one intersection between the
second circle and the line AB (the other is 0) and draw the di-
ameter OCD in the second circle. Now tlOCX is isosceles (with
two sides equal to a), so the angles at a and X are equal, say u,
and they are so marked. Consequently, we always have
L.XCD = 2u ,
for it is exterior to the third angle in the triangle and thus equal
to the sum of the first two (this is the same as saying that the
sum of the angles in a triangle is 180°).
We now use this bit of geometry when we devise a kinematic
model: We consider the first circle as a deferent and the second
circle as an equally large epicycle. We let the deferent's radius
OC rotate counterclockwise around 0, uniformly with respect to
the fixed direction OB, namely, we let u grow uniformly; addi-
tionally, we let the epicyclic radius CX rotate in the opposite
direction-clockwise-uniformly with respect to the extension
CD of OC, but with twice the angular velocity of u. If we start
the motion with u = O-so X is at B-X will move from B to A
and back again, and the oscillation will be harmonic, for we
have
OX = 2a · cos u
(with sign).
This is a 'fils! couple, and it produces a harmonic oscillation
in a straight line in a way that could not offend even the sternest
Aristotelian eye, namely, as the result of combining two uniform
circular motions, in opposite directions and the second twice as
fast as the first.
We can also interpret Figure 15 in a different and " mechani-
cal" way: We draw the circle with a as a center and OB = 2a as
a radius, and we note that
arc BD = arc XD = 2a . u
if u is measured in radians. Thus, a point on the circumference
of a circle will describe a rectilinear, harmonic oscillation when
the circle rolls without slipping, and with constant speed, inside
another circle that is twice as large.
If we place such a 'fils! couple in the plane of the ecliptic with
the diameter AB perpendicular to the line of sight, we make the
 After Ptolemy; Copernicus, Brahe 103

point X oscillate harmonically and we have thus gained the fol-

lowing two advantages: First, the motion is generated entirely
by means of uniform circular motions; second, the point X
remains at practically the same distance from us .
Similarly, if we see to it that the couple's diameter AB always
points to the observer, or some other fixed point, we can make
the distance from the fixed point to the point X vary periodically
without interfering with the direction to it .
Thus, the TusI couple enables us to vary either the direction or
the distance to an object, whereas with the old device of an epi-
cycle it is either both or nothing, and this makes it a very useful
and versatile component of philosophically correct models.
An important example of what can be achieved with this
device is a result by Nasir aI-DIn al-Ttlsi and two of his fol-
lowers. Using a combination of only uniform circular motions,
they succeeded in making the center of the epicycle move uni-
formly around the old Ptolemaic equant point-exactly, not
approximately-while the new orbit of the epicyclic center stays
very close to the old deferent.
I shall be more specific. In Figure 16 I show a Ptolemaic
deferent of center C and radius p. The observer is at 0, and
the equant point is Q. The eccentricity-much exaggerated-is e
(here I use e to indicate the length OC, not a ratio, so e = ep,
where E is what we usually mean by eccentricity) and A is the
apogee, IT the perigee. The epicycle's center C' moves uniformly
around Q (i.e., the angle a grows uniformly with time) and not
around the center C; this is what was considered so philosophi-
cally offensive. Furthermore, I display, following E. S. Kennedy,
three equivalent attempts at doing nearly as well as Ptolemy,
but with nothing but uniform circular motions.
The first is by Nasir aI-DIn al-Tusi himself. He extends the line
QC' to D so that QD = p, the deferent's radius. He then places a
Tfisi-couple with its center at D and principal diameter along
QD and its extension, and with the radius of the small circle
equal to eJ2. If the angles are as shown in the figure , C"-the
equivalent of X in the preceding discussion of the device-is his
epicyclic center. Nasir aI-DIn uses the version of his couple with
a circle rolling inside another that is twice as large. I show the
arrangement in Figure 17. Since C" lies on the line QD, it moves
uniformly around Q when a grows uniformly with time, pre-
cisely like Ptolemy's C'.
We should further note that the radius of the small circle is
104 2. Greek Geometrical Planetary Models

,, A
"

o
QD·MF'· O~aCC'&f
~·£ ·f

1T
FIGURE 16.

chosen so that C" coincides with the apogee A and the perigee II
when 0: equals 00 and 1800 • This ensures that C" will not stray
far from the old deferent, particularly when e is small.
In Figure 16 I also indicate two almost identical other ways of
reaching the point C" via uniform circular motions (for the sake
of clarity, I omitted the circles themselves and drew only the
relevant radii). The first variant is due to Qutb aI-DIn al-Shlrazi
(1236-1311), who manages with only two circles. The first has its
center at M, the midpoint of CQ, and radius p; and the second
has its center at F, where MF is parallel and equal to QD, and
radius FC" = e/2. The motions are as indicated in Figure 16.
 Of Copernicus and Brahe 105

FIGURE 17.

The second version is from the work of Ibn al-Shatir of

°
Damascus (1304-1375/6). His model is quite like Qutb al-Din's,
but instead of going from to F via M, Ibn al-Shatir goes via G,
where G is the fourth vertex in the parallelogram spanned by 0 ,
M, and F . Qutb aI-DIn needed only two circular motions, the
first of which had the eccentricity 3/2 e, while Ibn al-Shatir
needed three, but they are without eccentricity.
Ibn al-Shatir's model is found in his treatise, nobly entitled
The Final Quest Concerning the Rectification of Principles, in
which he also makes ample further use of al-Tusf's device. In
the 1950s E. S. Kennedy rediscovered this treatise, and we were
astounded to learn that component for component, link for link,
Ibn al-Shatir's models correspond to those of Copernicus, per-
haps with the order of the links changed.

Of Copernicus and Brahe

I shall now conclude this chapter on Greek geometrical models
with a brief sketch of parts of the work of Copernicus and Brahe.
106 2. Greek Geometrical Planetary Models

(I \.
\ \
\ \
-.
\,
\

FIGURE 18, This plate shows the Copernican arrangement of the uni-
verse as Copernicus drew it himself (not to scale). Outermost we see
the immobile sphere of the fixed stars, inside which are the planetary
orbits (or spheres), among them the Earth's, and the Sun in the middle.
The orbits' legends include approximate periods: Saturn 30 years,
Jupiter 12 years, Mars 2 years, the Earth with the Moon 1 year, Venus
9 months, and Mercury 80 days.
The figure is part of a manuscript copy of De reuolutionibus .. . in
Copernicus's own hand. He gave it to his student Rheticus. After
 Of Copernicus and Brahe 107

This is quite proper, for though they in a sense marked the be-
ginning of modern astronomy-Copernicus with his cosmology,
Brahe with his vast corpus of the best observations yet-their
theoretical work was very much in the ancient tradition.
But first I cannot help remarking that if one is accustomed to
working on ancient astronomy, a field where one must stretch
to the utmost every fragment of primary evidence, be it of clay
tablet or papyrus, and where treatises must be laboriously
reconstructed from copies of copies, many times, and centuries,
removed, one becomes entirely overawed by the wealth of source
material preserved from the Renaissance. We have correspond-
ences, journals, all sorts of records, contemporary printed edi-
tions (Brahe had his own printing press and papermill on
Hveen), and even a manuscript copy of De reuolutionibus .. . (see
Figure 18) written very beautifully by Copernicus himself in red
and black ink (it is now in Krakow and in pristine condition, for
it is not the copy used by the printer).
Thus, we have quantities of information about Copernicus and
Brahe, and accounts of their lives can be found in many places,
and in varying degrees of detail, so I provide just the briefest
biographical sketches of them.
Nicolaus Copernicus was born in 1473 in Torun (Thorn),
Poland, in his father 's handsome house that still stands inside
the medieval city walls. He died in 1543 in Frauenburg (Fram-
bork) two months after the publication of his principal work,
De reuolutionibus orbium coelestium ("On the revolutions of the
celestial spheres") . He attended the University of Krakow and
afterward studied medicine and canon law (and astronomy on
the side) in Italy. He spent most of his life as canon in the
Chapter of Warmia, where his uncle and protector, Lucas
Watzenrode, was bishop from 1489 to 1512.
De reuolutionibus .. . is in its structure, compass, and mathe-

changing ownership many times, it is now in the Jagiellonian Library

in Krakow.
The whole manuscript has been published in facsimile twice, first by
Verlag R. Oldenbourg, Berlin-Munich, in 1944 (this figure is copied
from this edition) and again by the Polish Academy of Sciences in
Warsaw in 1975.
108 2. Greek Geometrical Planetary Models

matical and astronomical methods very like the Almagest except

in two respects: It includes cosmology, and it adheres strictly to
the principle of uniform circular motion. Copernicus's cosmol-
ogy, based on the heliocentric hypothesis, is, of course, what he
is famous for . I shall later compare it to Tycho Brahe's geocen-
tric version in the context of the astronomy of the time.
As an illustration of the second point of difference, I can
mention that Copernicus used two of the three models in Figure
16, with obvious modifications. First, in the Commentariolus, a
short, early presentation of his principal ideas that was circu-
lated in manuscript, we find Ibn al-Shatir's model from Figure
16, but with the sun at 0 and the planet at C". In De revolu-
tionibus .. . itself, perhaps to save one circle, Copernicus re-
placed it with Qutb al-Din's obvious alternative, as I show in
Figure 19. Here S is the sun, P' the planet according to both of
Copernicus's models, and P where the planet would be accord-
ing to a Ptolemaic equant model.
The degree of agreement between Copernicus's models and
those of Ibn al-Shatir is in general such that independent dis-
covery is quite out of the question, but we do not know where,
and in what form, Copernicus learned of the earlier work. It has
often been claimed that Copernicus wanted to get rid of the
equant point. This is wrong. In fact, he strove to preserve it, but
in a philosophically correct manner, and he succeeded in doing
so by following the work of the Maragha School, as we have
seen.
Tyge (in Latin, Tycho) Brahe (1546-1601) was born into a
noble Danish family-a Rosenkrans and a Guldensteren (Brahe's
spellings) are among the ancestors whose coats of arms he dis-
plays around his portrait-at his family's seat Knudstrup in
Skane, since 1658 a Swedish province. He spent most of his
earlier adulthood traveling in Europe, getting acquainted with
leading astronomers and instrument makers concerned with
improving instruments and observational techniques. He made
observations that gained him such renown that when he was
30 years old, his king, Frederik II, gave him free use for life of
the island Hveen-also Swedish since 1658-and several bene-
fices whose proceeds he used to build, maintain, and adorn his
observatory Uranienborg and associated workshops. Until he
quarelled with the government of the young Christian IV and
went into exile in 1597, he and his staff made and recorded a
body of observations that became the basis of modern astronomy.
 Of Copernicus and Brahe 109

1T
FIGURE 19.

With his best instruments he reached an accuracy of less than

one minute of are, by far the best until that time, and probably
near the limit of what is achievable without the telescope. Fur-
ther, and most importantly, he departed radically from the prac-
tice of his predecessors in his observational program. Hitherto
one had mostly observed, say, planets when they were in partie-
ulaly interesting situations, well suited to yield parameters of
preconceived models. However, Brahe and his staff ran sus-
tained sequences of indiscriminate observations without regard
to hypotheses or models. When Brahe died in exile in Prague in
1601, his observations remained at the disposal of Johannes
110 2. Greek Geometrical Planetary Models
 Of Copernicus and Brahe 111

Kepler, who in their wealth could find whatever he needed for

the construction of his new theories and models.
Brahe proposed an alternative cosmological scheme to that of
Copernicus-he modestly called it the Tychonic system, and so
do we-and I shall present and discuss the two systems together,
in qualitative terms.
The name of Copernicus immediately calls to mind his he -
liocentric cosmological scheme: The (mean) sun is at rest in
the "center" and the planets travel in orbits around it, among
them the earth, spinning on its axis. His reason for proposing
a theory that seems to be at such odds with everyday experience,
and common sense, lies in the peculiar role the sun played
in Ptolemy's planetary models: For an inner planet, the radius
of the deferent points to the mean sun, whereas for an outer
planet, the epicyclic radius to the planet is parallel to the di-
rection from the earth to the mean sun. Indeed, these seeming
peculiarities are immediate consequences of the heliocentric
hypothesis. I must emphasize that once one has suggested a he-
liocentric solar system, one can immediately find its dimensions
in astronomical units from the Almagest's parameters: An inner
planet's distance from the sun is given by r160, and an outer
planet's by 601r, where r is the planet's epicyclic radius mea-
sured in parts (Ptolemaic units), one part being one-sixtieth of
the deferent's radius, as we saw earlier (page 88).
The Tychonic system is the geocentric equivalent of Coperni-
cus's: The earth is at rest, the sun revolves around it once a

FIGURE 20. The frontispiece to Giambattista Riccioli, Almagestum

Novum .. . , Bononia (Bologna), 1651. Riccioli (1598-1671) was a Jesuit
scholar and so preferred the Tychonic system. He found it quite com-
patible with the recent discoveries in observational astronomy, and
there are many indications of his awareness of them among the signi-
ficant elements in the allegorical picture, including a telescope,
Mercury and Venus showing crescents, Saturn with its rings (though
they look more like ears, as they did at the first sightings), Jupiter with
its moons, and the rough surface of the Moon.
Urania weighs the Tychonic against the Copernican system and
finds the latter wanting. The Ptolemaic system lies on the ground
under the Tychonic. A reclining Ptolemy says "Erigor dum Corrigor,"
which may be translated as "I am exonerated even as I am corrected"
-Riccioli knew, of course, that the Almagest is compatible with the
Tychonic system.
112 2. Greek Geometrical Planetary Model s

year, and the five planets travel around the sun. The Tychonic
planetary models are then very like the properly scaled Ptol-
emaic models, in movable eccenter version for the outer planets,
that we discussed previously.
It cannot be overstressed that the Copernican and Tychonic
arrangements yield identical results as far as directions and
distances within the solar system are concerned, and it takes
evidence of a different sort to distinguish between them. For
example, the telescopic discovery of the phases of Venus is often
quoted as proof of the Copernican hypothesis. It is no such
thing: It merely shows that Venus goes around the Sun, not that
the Earth does, and it is in agreement with the Tychonic system
as well.
Furthermore, as soon as we try to calculate planetary posi-
tions, we are, in a sense, working in a system of the Tychonic
type, for we are concerned with how things look from the earth,
not from the sun. The Nautical Almanac is, as it were, produced
on a Tychonic hypothesis.
One had, then, to seek beyond the solar system for observa-
tional confirmation of the Copernican system, and it wa s slow in
coming. It had been clear ever since antiquity that a heliocentric
system would imply an apparent yearly change in stellar posi-
tions as we on the earth travel in a vast orbit around the sun,
and no such yearly parallax was observed until Bessel did so in
the 1830s.
The problem of a heliocentric hypothesis combined with a
failure to observe an annual parallax is nicely put by Archi-
medes (d. 212 B.C.) in the preface to his Sandreckon er (Heath's
translation):
But Aristarchus of Samos (early third century B.C.) brought out a
book consisting of some hypotheses, in which the premises lead to the
result that the universe is many times greater than that now so called.
His hypotheses are that the fixed stars and the sun remain unmoved,
that the earth revolves about the sun in the circumference of a circle,
the sun lying in the middle of the orbit, and that the sphe re of the fixed
stars, situat ed about the same centre as the sun, is so gre at that the
circle in which he supposes the earth to revolve bears su ch a propor-
tion to the distance to the fixed stars as the centre of the sphe re bears
to its surface.
Brahe had tried in vain to observe an annual parallax and
argued that either there was not any, in which case Copernicus
was wrong, and his own system was to be preferred, or it was
 Of Copernicus and Brahe 113

less than what he could observe. He knew that his accuracy of

observation was 1', which then had to be an upper limit on the
annual parallax. This, in turn, placed a lower limit on the dis-
tance to the fixed stars. Now, he believed to have measured the
angular diameters of stars to be between 1/3' and 2', and a star of
diameter, say, L' would be as large as the orbit of the earth,
which he found absurd. (Note that a star's annual parallax can
also be defined as the angle subtended by the earth's orbit at the
eye of an observer on the star.) By the way, Brahe's estimate of
the apparent size of a fixed star was based on an optical effect
(diffraction) that makes a point source of light seem larger the
brighter it is; telescopes showed that no matter how large the
magnification, fixed stars still appear as points.
After the introduction of telescopes, astronomers eagerly
sought the annual parallax of stars but did so in vain until F. W.
Bessel in the years 1837-1840 found the annual parallax of 61
Cygni to be 0.35".
As a matter of fact, another, but unforeseen, consequence of
the earth's motion had been observed by Bradley about a cen-
tury earlier. It turned out to be what we call the " annual aber-
ration, " the apparent deflection of the line of sight to a star in
the direction of the earth's velocity, which, to be sure, is small in
comparison to the velocity of light, but not negligible. The effect
is rather like what you observe when you drive through rain
falling vertically down: The rain seems to come slanting down
toward your windshield, falling from a point ahead of the one
directly overhead. The aberration is about 20.5" for a star near
the pole of the ecliptic. It should be noted that the annual par-
allax is smaller the farther the fixed star is from us, whereas the
aberration is independent of distance.
By the time of its strict observational confirmation, the Co-
pernican system had, of course, long been generally accepted,
due particularly to the work of Kepler and Newton. Keplerian
motion is the subject of the last chapter of this book, but New-
ton's dynamics and laws of motion are beyond our scope.
3
Ptolemy's Cosmology

Textual Problems
The Ptolemaic system has long been in common use as the name
for the cosmological scheme that was eventually replaced by the
Copernican. It consists of seven nested spherical shells, one for
the Sun, one for the Moon, and one for each of the five planets,
all surrounded by the sphere of the fixed stars. All of this re-
volves once a day about a central, stationary Earth, while Sun,
Moon, and planets move appropriately and much more slowly,
each within its own sphere. This scheme is most often repre-
sented graphically by its intersection with the plane of the
ecliptic (the Sun's orbit)-it then becomes a system of nested
annuli-for the planets and the Moon are always close to this
plane.
In modern times the Ptolemaic system is usually encountered
only when it is being compared-not to its advantage-to the
Copernican scheme by authors and teachers who , more often
than not, are wholly ignorant of its finer structure, dimensions,
and underlying principles. For pedagogical reasons the virtues
of Copernicus's system and the flaws of Ptolemy's are equally
emphasized, and one is left with the impression that anyone but
an utter idiot ought to have embraced Copernicanism immedi-
ately after having been exposed to it and discarded with scorn
the older views.
I find it a pity that a vague and skewed image is thus created
of the cosmological system that prevailed for nearly a millen-
nium and a half in the West, and for longer in the Near East, a
system of pleasing harmony and inner consistency, and with all
its parts neatly measured in terrestrial measure. To give but one
example of the confident and matter-of-fact manner in which

114
A. Aaboe, Episodes From the Early History of Astronomy
© Springer-Verlag New York, Inc. 2001
 Textual Problems 115

medieval authors approach their universe, I can point to the

place in Dante's Conuiuio where he says that "Mercury is the
smallest star in the heavens, for the quantity of its diameter
is not more than 232 miles according to what Alfraganius
(al-Farghani) posits, who says it to be 1/28 of Earth's diameter
which is 6500 miles."*
In the following I set forth the basic assumptions, explicit as
well as implicit, on which the Ptolemaic system rests, and sketch
how its details are derived from them, but I begin by tracing the
difficult path that led to our present control of the evidence for
the system's genesis.
It seems a tautology that the Ptolemaic system is due to the
astronomer Ptolemy, or Claudios Ptolemaios, of Alexandria,
whose principal work on astronomy, the Mathematical Collection
or the Almagest, as it is more commonly called, was written
about A.D. 150. Mathematicians, and even more so historians of
mathematics, know all too well, however, that the person whose
name is affixed to a theorem is not necessarily the one who first
stated and proved it, and it seemed as if the Ptolemaic system
was a case in point. Indeed, the few who had read Ptolemy were
well aware that this system is nowhere described in any of his
works that have survived in the original Greek, and some of
these few-I for one, alas-were willing to argue, quite persua-
sively, we thought, that the system was not his.
Our earliest textual evidence for it was found in the Hypo-
typosis ("Outline") of Astronomical Hypotheses ("Models") by
Proclus (A.D. 412-485). Proclus was the next-to-last director of
Plato's old Academy in Athens and was passionately devoted to
the old pagan Greek culture that was being threatened on all
sides by the new Christian barbarism, as he thought it. You get
the feeling when you read his works that he saw himself as one
of the last outposts of the culture he loved so well, and he did
what he could to preserve it for posterity by writing highly
competent summaries of most branches of Greek learning. His
works-the survivors of which, by the way, are still not com-
pletely edited-were indeed, he would have been gratified to
know, much studied and admired during the early Renaissance.
The text of his Hypo typos is was fortunately well published with
a German translation by Karl Manitius in 1909, and it remains

*1 owe this reference to Noel Swerdlow.

116 3. Ptolemy's Cosmology

the best elementary introduction to classical astronomy in the

Ptolemaic manner. In most of this work Proclus summarizes
the contents of the Almagest, but he also devotes a chapter to a
description of the cosmological system of nested spheres, which
has no counterpart in the Almagest. It is curious, and seemed
significant, that Proclus does not mention Ptolemy anywhere in
this part, while elsewhere he is lavish with both credit and
praise.
It is quite natural that a cosmological scheme is nowhere to be
found in the Almagest, for Ptolemy's concern here is with posi-
tional astronomy and eclipse theory. In other words, the princi-
pal aim of the Almagest is to enable you to answer the question:
Given your location on the earth, and given the time, in pre-
cisely which direction should you look in order to see a given
celestial body? For only one of the bodies-the moon-is its
distance from the earth of practical interest in this connection,
and tables of the moon's daily parallax are duly given (they
matter most for the prediction of the visibility and magnitude of
solar eclipses at a given place).
The obvious place for Ptolemy to present his cosmological
system, we thought, was in his Planetary Hypotheses (the Greek
word "hypotheses" is here best translated by "models") and it
was not there. Our best edition was that delivered in 1907 by the
Danish classical philologist J. L. Heiberg (to whom we owe so
many definitive editions of Greek mathematical and astronomi-
cal texts) in a volume presenting Ptolemy's minor astronomical
works. He gave all that survives in Greek (18 printed pages),
which he took to be Book I. There are, however, two manu-
scripts of an Arabic translation of the Hypotheses , one in Leiden
and the other in the British Museum, and both contain Book II
as well as Book I. The Arabic manuscript in the British Museum
was written in 1242, while none of Heiberg's Greek manuscripts
antedates the 14th century. The brilliant young German Orien-
talist Ludwig Nix translated the Arabic text into German, but
he died before putting the finishing touches on his work. Hei-
berg decided to print on facing pages Nix's translation via the
Arabic of his Greek text as far as it went. Heegaard and Buhl
(a mathematician and an orientalist) prepared Nix's translation
of Book II for publication-they had the use of the Leiden
manuscript-and Heiberg printed it after the Greek and German
of the first part. The work as it then stood dealt with the con-
struction of planetary models for motion in longitude and lati-
 Principles, Assumptions, and Construction of the Ptolemaic System 117

tude, largely based on the Almagest's parameters, but nowhere

did it give instructions for fitting the models for the different
planets together into a cosmological whole.
This is where matters rested until 1964 when the German
scholar Willy Hartner published the results of his close study of
Proclus's Hypotyposes and commentary to Plato's Timaeus and
of various Arabic treatises on cosmology. He came to the con-
clusion that the Ptolemaic system must have been set forth by
Ptolemy himself in the Planetary Hypotheses, very likely at the
end of Book II.
A few years later Noel Swerdlow was beginning to work here
at Yale on a doctoral thesis on medieval cosmology, and very
naturally we were discussing Hartner's article. This prompted
my then-colleague, Bernard Goldstein, who directed Mr. Swer-
dlow's thesis, to wonder if the missing part might not be found in
a Hebrew version of the Planetary Hypotheses even though it
was absent, so it seemed, in the Arabic. He consulted Steinsch-
neider's work on medieval translations into Hebrew and learned
that he had, in fact, in his desk drawer a microfilm that among
other things contained the wanted work. To his great and good
surprise he found that where the Greek (and German) of Book I
gave out, the Hebrew continued with Ptolemy's own account of
the system of nested spheres. Goldstein obtained microfilms of
the two Arabic manuscripts and found, again to his surprise,
that they, too, had the same long passage in the same place-
one cannot understand how it came to be omitted in Heiberg's
edition. Goldstein published the complete Arabic text, with
commentary and a translation of the newly found second part of
Book I, in 1967. At long last it had been firmly established that
the Ptolemaic system was indeed due to Ptolemy.
I have gone into these matters at such length to convey a
sense of the vagaries, and the pleasures, of scholarship, and of
the importance of consulting the texts themselves.

Principles, Assumptions, and Construction of the

Ptolemaic System
I now finally turn to a statement and discussion of the principles
and assumptions that underlie Ptolemy's cosmological scheme,
and then I shall sketch how he derives it from them.
118 3. Ptolemy's Cosmology

First Ptolemy assumes that the fixed-star heaven is spherical

and moves like a sphere around a spherical earth, very small in
comparison and at rest in the center. He argues for all these
things in the Almagest, Book I, and they are of course not new
with him. Here I might insert a remark on the recognition of the
earth's sphericity. In the mid-third century B.C. Eratosthenes
had given a good estimate of the earth's circumference, and at
least from that time on no person educated in the classical tra-
dition, be it in antiquity, in the Islamic countries, or in the Latin
West, had doubted that the earth is round. The popular notion
that Columbus proved it is utterly false ; he merely advocated a
value of its size-whether he believed it or not-that was much
too small and used it to argue for a shorter western route to
China. As everyone knows, he never got there.
Ptolemy does not argue for the next fundamental principle,
but merely takes it for granted. It amounts to this: Each of the
planets, including Sun and Moon, has its own particular range
of possible distances from the Earth; in other words, once a
planet has been at a certain distance, no other planet can ever
attain that same distance, except perhaps at the extremes of its
range. It follows that each planet can call an earth-centered
spherical shell its own private domain if the outer and inner
radii are the greatest and smallest distances, respectively, from
the Earth to the planet, or at least part of such a shell, truncated
symmetrically to the ecliptic to allow the planet to move in
latitude.
Further, Ptolemy says [Planetary Hypotheses (PH) I, ii, 4] that
it is most plausible that the spherical shells are snugly nested,
"for ," he continues, " it is not conceivable that there be in nature
a vacuum, or any meaningless and useless thing. " Thus, the
greatest distance of one planet from the Earth equals the small-
est distance of the next.
Ptolemy now accepts that we meet the planets, as we proceed
outward from the Earth, in the following sequence: Moon,
Mercury, Venus, Sun, Mars, Jupiter, and Saturn, before coming
to the fixed stars. Ptolemy mentions this order in Almagest IX, 1,
but only in passing. This seems quite natural when one con-
siders that his goal with the Almagest's planetary models is to
predict directions to the planets, and their distances are so great
in comparison to the size of the Earth that the place of observa-
tion has no effect on the direction of the lines of sight to them.
Indeed, if we could measure a difference in direction to a given
 Principles, Assumptions, and Construction of the Ptolemaic System 119

planet when viewed from two places on the Earth, we would

have a means of finding how far away it is. This is what Ptolemy
puts in other words when he says in this passage that the only
sure way of finding planetary distances is by determining plane-
tary parallaxes, and none is sensible to the naked eye (I shall
return to the problem of finding a parallax). He ascribes the
order he accepts to the "older astronomers" and adds that there
has been considerable discussion of the matter. Thus, some more
recent astronomers have, he says, placed all the planets beyond
the sun because no one has ever seen a planet cross the disc of
the sun. Ptolemy counters this argument, both in the Almagest
(IX, 1) and the Planetary Hypotheses (I, ii, 2), by saying that such
transits would be extremely rare and that they would likely be
invisible to the naked eye since a solar eclipse of magnitude
equal to a planetary diameter cannot be seen. Thus, he finds it
plausible that the sun separates the three planets that can reach
opposition (our outer planets) from the two that cannot (our
inner planets). Otherwise the order reflects that a slower planet
ought to be more distant, and all agree, he says, that the moon is
nearer to us, and the fixed stars farther away, than any planet.
The next assumption is completely new with Ptolemy. In the
Almagest (III, 3) he proves that the size of his epicyclic models
does not matter at all, as far as predicting directions to the
celestial bodies is concerned, just so long as all proportions and
mean motions are correct. In modern times there has been much
discussion of whether in the Almagest Ptolemy views his models
as descriptions of how the planets actually move in space or
merely as devices to serve as the basis for calculating planetary
positions. That even the well-informed can take part in this dis-
cussion shows, of course, that the question is irrelevant to the
. Almagest's aims. In the Planetary Hypotheses, however, there is
no doubt on this point, for Ptolemy now assumes that his
detailed quantitative models, correctly scaled, really represent
the way the planets move around us. (This assumption happens
to be very nearly right, as we have seen; it is in his choice of
absolute sizes that Ptolemy errs.) This assumption enables
Ptolemy to find the ratio of inner to outer radius of each spheri-
cal shell. To see how this is done, let us consider his detailed
model for a planet (Figure 1) other than Mercury.
The entire model is at first thought of as lying in the plane of
the ecliptic through the Earth E-Iater it is tilted in various
ways to represent the planet's movement in latitude. The planet
120 3. Ptolemy's Cosmology

FIGURE 1.

P is carried on the circumference of the epicycle of radius rand

center C', which, in turn, moves on a larger circle, the deferent,
of center C and radius R. The distance from the Earth E to the
center C of the deferent is called the eccentricity, e. The line EC,
the apsidal line, remains fixed with respect to the fixed stars.
Though it is irrelevant for our present purpose, I should mention
that the epicycle's center C' moves uniformly, not around the
deferent's center, but as seen from the equant point Q placed on
the apsidal line so that CQ = EC = e, and that the planet P
moves uniformly, not with respect to CC' , but to QC'; thus, it is
the two angles marked with arrows in Figure 1 that grow uni-
formly with time. When Ptolemy introduces the equant point, he
blithely violates the philosophical doctrine of uniform circular
motion as the only permitted component, but it is this bold step
that enables him to achieve such excellence so economically.
Such a model is used for Venus as well as for the outer planets.
The only difference, except in the necessary parameters, is that
in Venus's model the radius of the deferent or rather, more pre-
cisely, the line QC' from the equant point to the epicycle's cen-
ter, points toward the mean sun, while for any outer planet the
epicycle's radius C' P to the planet is parallel to that direction.
 Principles, Assumptions, and Construction of the Ptolemaic System 121

Ptolemy's model for Mercury is, of necessity, more complicated,

and I shall not discuss it here.
Ptolemy measures the dimensions of each model with a unit
that is one-sixtieth of the radius of the deferent. The actual size
of this unit is irrelevant when we want to find the direction of
the line of sight from the Earth to the planet (EP) as Ptolemy
has shown. In the Almagest he makes no attempt at finding it for
any planet, but it is otherwise in the Planetary Hypotheses . With
the new assumption that a planet really moves according to such
a model of correct size, he first determines the ratio of the
planet's least-to-greatest distance from the Earth. Indeed, if the
planet's spherical shell has inner and outer radii in that same
ratio, then it can just accommodate an epicyclic model of the
right proportions.
The planet is farthest from the Earth when the center of the
epicycle is at the apogee of the deferent and when the planet is
at the apogee of the epicycle ("apogee" means the point most
distant from the Earth). We then obtain that the greatest dis-
tance D of the planet is [see Figure 2(a)]
D = R + r + e = 60 + r + e Ptol. units.

FIGURE 2.
122 3. Ptolemy's Cosmology

TABLE 1

a:D fi' D'

Moon. <t 3+:65 1 /2
~rcu.r.!1 t; 34:88 2"- /5
5 /32.
VtK.\4S 9 16: '04 #

Sun. 0 23 :2.5 32 # L/ 35
Ma.rs cI 1: 7 35 # [,..245
J~4 23: 37 2.45# ,/39~
SRlUY't\. 11 5:7 394"- 55J

Likewise, we get for closest approach d of the planet [see Figure

2(b)]

d =R- r - e = 60 - r - e Pto!' units.

Thus, the ratio D :d is easily found. Let us consider the model for
Venus. Here we have, as everywhere, R = 60 and, according to
Almagest X, 1-5, r = 43;10 and e = 1;15 (here I follow the now-
standard way of writing sexagesimal fractions: 43;10 means 43
10/60, etc.). We then get
D = 60 + 43;10 + 1;15 = 104;25
and
d = 60 - 43;10 - 1;15 = 15;35,
so
d : D = 15;35 : 104;25.
In the Planetary Hypotheses Ptolemy rounds this ratio to 16 : 104.
He proceeds likewise for the other planets and obtains the ratios
of least-to-greatest distance given in the first column of Table l.
I included the corresponding ratios from the Almagest for the
Sun and Moon. Ptolemy takes over Hipparchus's solar model
without change; it is a simple eccentric circle with an eccen-
tricity of 2 ~ parts in 60. His complicated lunar model does indeed
give rise to the outrageous variation in lunar distance in a ratio
of nearly two to one . I cannot go into detail with these matters
here but shall merely hint that this awkwardness is the result
of Ptolemy's successful attempt at accounting for the lunar in-
 Principles, Assumptions, and Construction of the Ptolemaic System 123

equality now called the "evection," which manifests itself as an

enlargement of the moon's possible deviation from mean behav-
ior at first and last quarters. Ptolemy achieves this by pulling
the moon's epicycle closer to the earth at quadrature to make it
seem larger, but at the price of a much-too-large variation in
lunar distance and, one should expect, in the moon's apparent
size, but he is silent on this embarrassing point in the Almagest.
We are now in a position to construct a scale model of the
universe. If we call the inner radius of the lunar sphere d' = 1,
then the outer radius becomes D' = 65/34, or to the nearest
integer, 2. The principle of nesting spheres demands that this D'
also be d' for Mercury. Applying the ratio of D: d = 88 :34, we
obtain D' = 5, again to the nearest integer, which, in turn,
must be d' for Venus, and thus we can proceed all the way
out through the spheres and obtain the values for d' and D' in
Table 1.
If just one of these dimensions could be given in some terres-
trial unit of length, the whole universe could then be measured
with that unit, and Ptolemy's next step is to introduce the
Earth's radius as a yardstick, but, as we shall see, he gets more
than he really needs and wants. In the Planetary Hypotheses I, ii,
3, he refers to the result obtained in Almagest V, 15, where he is
concerned with finding the distances to the Moon and Sun in
terrestrial radii (t.r.), which he needs for forecasting solar
eclipses. His method certainly goes back to Hipparchus, and I
present it here in quite ahistorical and much simplified form, in
respect of both geometrical argument and parameters, in order
to display its essential core.
The argument proceeds from the following two pieces of data:
(i) The apparent diameters of the Sun and Moon are both equal
to half a degree, very nearly, and
(ii) the apparent diameter of the Earth's shadow cone, cut at the
distance of the Moon, is 2 ~ times as large as the apparent
diameter of the Moon. I shall return to how such estimates
could be obtained.
Figure 3 represents-with no pretense whatsoever at being
drawn to scale-the Sun and the Earth, with the Earth's shadow
cone extending toward the right, and with part of the lunar orbit
drawn in . The angles ¢ and 'ljJ are the apparent radii of the Sun
and the shadow cone's circular cross-section at the distance of
the Moon, respectively, and our data tell us that they are
124 3. Ptolemy's Cosmology

/
/lww-orbit

FIGURE 3.

FIGURE 4.

(i) and (ii) 1/1 = 2 ~ . </> = r.

The angles marked 7ro and 7r. are what we call the horizontal
parallaxes of the Sun and Moon, and we get immediately that

(</> and 1/1 and the big angle at the Earth's center add up to 180°,
but so do 7ro , 7r., and the same big angle). Thus, if we can deter-
mine one of the parallaxes, this relation gives us the other.
To see what is involved in finding the parallax of a celestial
object, consider Figure 4. Here C is the center of the earth, 0 is
an observer on the earth's surface whose horizon is indicated by
the dotted line, S is the object whose altitude as seen by 0 is the
angle h, and the angle 7r is what we call its daily parallax. It is ,
then, the difference between the directions of the lines of sight to
S from 0 and C, respectively, and if we call the terrestrial radius
OC 1, and the distance from C to S d, we obtain immediately
 Principles, Assumptions, and Construction of the Ptolemaic System 125

sin 7r sin(90 + h) cos h

= --'---::---'--
1 d d
The parallax vanishes if h = 90°, that is, if S is in the observer's
zenith, and assumes its largest value when h = 0, that is, when S
is in the observer's horizon. This last value of 7r we call the hor-
izontal parallax of S-we encountered it for the sun and moon
earlier-and if we measure it in radians, we can replace sin 7r by
7r for small values of 7r and get

where d is measured in earth radii.

Translating the statement about solar and lunar horizontal
parallaxes into radians, we get

or, in terms of solar and lunar distances,

1 1 1
-+-=-- ,
do d.. 65.5
where do and d.. are measured in earth radii.
At the end of Almagest V, 11, Ptolemy says that Hipparchus
found the distance to the moon by assuming the distance to the
sun: At one time he took the sun to have a barely observable
parallax, and at another he supposed that it had none at all.
(Hipparchus's works on these, as on most matters, are lost.)
This used to seem a curious procedure, for of the two distances
that to the sun is by far the more difficult to measure (in fact, the
astronomical unit, as we call it, was not well determined until
the end of the 19th century). However, as has now become clear,
particularly through the work of Swerdlow and of Toomer,
Hipparchus saw that the contribution of lunar parallax to the
total of 7/8° (= 1/65.5 rad.) in the relation above was predomi-
nant, for he had come to believe, as was true, that solar parallax
was not directly observable with his means. Thus, the relation
could serve to give him a good first estimate of the distance to
the moon. More precisely, if he assigned to the solar parallax
the least observable amount (very likely 7 minutes of are, as
Swerdlow has argued) and then let it vanish, he would obtain
a lower and an upper bound for lunar parallax, and hence an
126 3. Ptolemy's Cosmology

upper and a lower bound for lunar distance. Since the two limits
would not be very far apart, this approach would give him a
good idea of the order of magnitude of the moon's distance from
the earth. To illustrate this argument with our crude values
above, we would get as the upper limit of lunar parallax, namely
with vanishing solar parallax,
1f " -- 7° -
8 -
1
65.5
rad . , or d" = 65.5 t.r.
as the lower limit of lunar distance. This is of the right order of
magnitude, though too large by about 10 percent; a solar paral-
lax of 7' would increase d" to 75.5 terrestrial radii. The modern
values of lunar and solar mean horizontal parallax are 57' and
8.8", respectively, to which correspond lunar and solar mean
distances of a little over 60 terrestrial radii and about 23,500
terrestrial radii, so the interval from 65.5 to 75.5 terrestrial radii
for the lunar distances fails to capture the correct value. This is,
of course, due to the crudeness of the two pieces of initial data-
the values I assigned to the apparent diameters of the sun and
the shadow-and not to any flaw in the method.
As I said, my derivation of the relation between solar and
lunar distances from the two pieces of data is very much
simplified-this simplification goes back at least to J . L. E.
Dreyer. Ptolemy performs his calculations using trigonometry,
and so in all probability did Hipparchus before him, and both of
them use data different from mine. It seems that Hipparchus
found the apparent diameter of the sun (and moon) to be 1/650 of
a complete circle (= 0;31,14°) by means of his diopter, and he
determined the shadow's diameter to be 2! times that of the
moon, very likely by measuring with a water clock the time
the moon needed to traverse the shadow during a nearly central
lunar eclipse.
To digress for a moment, Hipparchus was certainly not the
first to find the apparent solar diameter. In his sole surviving
work, On Sizes and Distances of Sun and Moon, Aristarchus of
Samos (third century B.C.) presents the puzzling value of 2° (a
fifteenth of a sign, as he puts it), while Archimedes (287-212 B.C.)
ascribes to him the discovery that it is about 1/720 part of the
circle of the zodiac (i.e., 1/2°). Archimedes tells of this in the
preface to his book The Sand-Reckoner and proceeds to describe
his own determination of that parameter, the earliest explicit
account of a scientific measurement that I know of. Archimedes
says that he pointed a firmly fixed ruler with a small cylinder
 Principles, Assumptions, and Construction of the Ptolemaic System 127

standing on it toward the rising sun-one can bear looking at

it then-and holding his eye at the ruler's end he moved the
cylinder first so it just covered the solar disc and then so the sun
could just be seen on either side of it, and marked the two posi-
tions of the cylinder. From these two measurements, and even
taking into account that the eye does not see from a point but
from an area (the pupil) whose size he also determined, Archi-
medes calculated that the apparent solar diameter is more than
1/200 of and less than 1/164 of a right angle, limits that still
hold. It is noteworthy that the result of the earliest measurement
is given as an interval, and not as a single number, but it should
not be surprising, for we are dealing with Archimedes.
For his derivation of the two diameters, Ptolemy scorns the
use of diopter and water clock. Instead he analyzes two lunar
eclipse reports from Babylon in the light of his refined lunar
theory. During these eclipses one-fourth and one-half of the
lunar diameter, respectively, reached into the shadow, and from
this he calculates that the lunar diameter must be 0;31,20° and
that the shadow's diameter is 2 ~ times that of the moon. With his
version of the above argument, the connection between solar
and lunar distance is established.
Already in Almagest V, 13, Ptolemy has, however, found the
distance to the moon. He does it by determining lunar parallax
as the difference between a computed and an observed lunar
altitude; this is proper, for his lunar theory represents the moon
as seen from the center of the earth. From the parallax he can
find the distance from earth to moon in terrestrial radii, as we
saw earlier, for one particular configuration of the lunar model,
and so he can assign all parts of the model terrestrial scale. I
shall omit his observation and calculations and merely mention
that he finds the greatest distance D to the moon to be 64;10
t.r.-he abbreviates it to 64 in the Planetary Hypotheses. It is
with this value that he enters the relation between solar and
lunar distances-he maintains that lunar and solar diameters
appear equal when the moon is at its farthest from the earth and
the sun at its mean distance-and he finds the mean distance to
the sun to be 1210 terrestrial radii.
As I said, all we need to measure the universe in terrestrial
units is to have just one of the distances in Table 1 so deter-
mined, but here Ptolemy presents us with two , which may prove
embarrassing: a greatest distance of the Moon of 64 t.r. and a
mean solar distance of 1210 t.r., which imply smallest and
128 3. Ptolemy's Cosmology

TABLE 2. Distances in terrestrial radii (t.r .)

a:D ti 1)
et 34:65 33 -' 64
~ 34:85 6+ ' ;' 166
9 10f 166 ' 1079
j 6:
0 2;:2,5 1160 (12.10),1260
d' 1: 7 1260- 8820
-'

It 23: 37 8810" )4 189

11 5:7 14189' 19865

greatest solar distances of 1160 t.r. and 1260 t.r., respectively.

Yet we can but try. Proceeding as before, but beginning with
D = 64 t.r. for the Moon, we find immediately d = 64 t .r. for
Mercury (see Table 2). We apply the ratio of D: d and find
D = 166 t.r. for Mercury, which becomes d for Venus. With
some trepidation we multiply d = 166 t.r. by the ratio 104: 16 and
obtain D = 1079, which miraculously is very close to d = 1160
for the Sun. Ptolemy now says (P.H. I, ii, 3)
Since the least distance of the Sun is 1,160 earth radii, as we men-
tioned, there is a discrepancy between the two distances which we
cannot account for; but we were led inescapably to the distances
which we set down . So much for the two (planetary) spheres which lie
closer to the earth than the others. The remaining spheres cannot lie
between the spheres of the Moon and the Sun, for even the sphere of
Mars, which is the nearest to the earth of the remaining spheres, and
whose ratio of greatest to least is about 7 : 1, cannot be accommodated
between the greatest distance of Venus and the least distance of the
Sun. On the other hand it so happens that when we increase the dis-
tance to the Moon, we are forced to decrease the distance to the Sun,
and vice versa. Thus, if we increase the distance to the Moon slightly,
the distance to the Sun will be somewhat diminished and it will then
correspond to the greatest distance of Venus. [Goldstein's translation]
Ptolemy is right, for the relation
1 1
-d + -d = constant
o ~
 Principles, Assumptions, and Construction of the Ptolemaic System 129

FIGURE 5. The Ptolemaic system (P.H. I) drawn approximately to scale

with the sub-Martian orbs enlarged tenfold. The earth's diameter is
1/33 of the sublunar orb's.

implies, of course, that if one of the distances is increased, the

other must decrease. However, he does nothing to close the
slight gap between the spheres of Venus and the Sun but con-
tinues building the system outward from the solar sphere, taking
D for the Sun to be d for Mars, and so on . Outside Saturn's orb
we have the fixed stars at a distance of very nearly 20,000 earth
radii, and the system is complete.
In Figure 5 I have drawn the spheres in proper proportion; the
130 3. Ptolemy's Cosmology

--- c > >

FIGURE 6. E: earth; C: center of deferent; C': center of epicycle; P:

planet (Mars); d: smallest distance; and D: largest distance.

dotted area is the gap between Venus's and the Sun's orbs, and
for the sake of clarity I have enlarged the spheres inside Mars's
10 times and moved them upward, out of the way. You are asked
to envisage this part compressed to one-tenth of its size and
placed into the shaded hole in Mars's sphere.
The cosmological system is now finished and measured in
earthly measure-we shall presently transform the dimensions
into everyday units. We have seen that it arises as a combina-
tion of old and new components, and chief among the new are
those of a quantitative character: The spheres are assigned defi-
nite sizes and are constructed so that they can accommodate the
best quantitative astronomical models of the day. No wonder,
then, that the system gained such widespread and lasting popu-
larity. I drew Figure 6 to show how a planetary model fits into
the planet's spherical shell. The figure represents the intersec-
tion between the sphere and the ecliptic. I chose Mars because
 Principles, Assumptions, and Construction of the Ptolemaic System 131

its eccentricity is so large that I could afford to draw the model

in correct proportion in all its parts. In principle, Mars can be
anywhere in the white eccentric annulus, but it can never
invade the two shaded lunules-one must remember that C, the
center of the deferent, is fixed, whereas C' and P can be any-
where on their respective circles.
Ptolemy now says (P.R. I, ii, 4) that the circumference of
the earth is 18 myriad (180,000) stades, which agrees with his
Geography , where he maintains that one degree measured along
a meridian (i.e., going due north or south) is 500 stades. From
this it follows that one terrestrial radius is 2;52 myriad stades
(he actually gives it, Egyptian fashion, as 2 + 1/2 + 1/3 + 1/30
myriad stades). The stade is a Greek unit of length always, as far
as we know, equal to 600 feet, but the length of the Greek foot
seems to have varied with time and place. Thus, it is impossible
to say how good his estimate of the earth's size is, but it is cer-
tainly of the right order of magnitude, though very likely quite a
bit too small (one degree of latitude is about 364,500 English and
American feet, while 500 stades is 300,000 Greek feet). When he
adopts this value, he follows Marinos (died circa 110-120),
whose work on geography is lost, though a good part of its con-
tents are known, for Ptolemy refers to it extensively. The esti-
mate of 180,000 stades for the earth's circumference, adopted
by Marinos, seems to go back at least to Eratosthenes (third
century B.C.), according to Cleomedes (fourth century A.D.), but
so does another estimate that was very popular before Marinos
and Ptolemy and variously given as 250,000 or 252,000 stades
(the latter yields very neatly 700 stades per degree of latitude).
However, Eratosthenes was certainly not the first to attempt to
measure the earth, for his older contemporary and aquaintance
Archimedes (ca. 287-212 B.C.) tells that "some" had tried to
prove that its circumference is 300,000 stades.
To reach the larger of his two estimates, Eratosthenes pro-
ceeded from the following pieces of data:
(i) Alexandria and Syene (modern Aswan) lie on the same
meridian;
(ii) at summer solstice the sun culminates in zenith at Syene
(its rays reach the bottom of a well, it is said), but 1/50 of a
full circle south of zenith at Alexandria;
(iii) the distance between the two places is 5000 stades.
The circumference of the earth is then simply 50 x 5000 stades,
132 3. Ptolemy's Cosmology

or 250,000 stades. The smaller estimate seems to arise in the

same fashion, but a locality directly north of Alexandria across
the sea plays the role of Syene.
In his Geography Ptolemy shows how one may derive the
length of a degree from the latitudes and the difference in longi-
tude of two places of known distance, measured in stades, by
means of spherical trigonometry. As he was well aware, there
are two difficulties with this procedure. It is fairly easy to estab-
lish the latitude of a locality, but it is otherwise when one wants
to find the difference in longitude between two places, for it
involves comparing their local times. In antiquity it was impos-
sible to carry the time of one location to another-it remained
so, within the margin of what is useful, until the end of the 18th
century-so one had to rely on lunar eclipses to serve as uni-
versal time signals for the comparison. The other difficulty is, of
course, with measuring the distance between places that are far
apart. In ancient itineraries such distances are most often given
by the number of days it took to travel from one to the other,
a number much affected by irrelevant circumstances like the
nature of intervening terrain.
These two difficulties are also inherent in the older method,
for to ascertain that two distant places lie on the same meridian
one must show that they have the same longitude (the longitude
of Syene is, in fact, 3° more easterly than that of Alexandria),
and it is just as hard to measure a distance in the direction north
or south as in any other direction. The only advantage of the
older method is that it does not depend on spherical trigonome-
try, which is just as well, for this subject had not yet been
invented at the time of Eratosthenes.
Much has been written in modern times about the quality of
the various measurements of the earth and, indeed, if you pick
the length of a stade carefully you can show that Eratosthenes's
larger value is wonderfully accurate. However, I hope that my
short discussion has shown that such efforts are historically
irrelevant, for quite apart from the uncertainty about the length
of the Greek foot or stade, all underlying data are nothing like
modern measurements, but rather rough estimates given in nice
round numbers.
To return to the principal matter, we can now simply multiply
all the distances in Table 2 by 2;52 to convert them from terres-
trial radii into myriad stades, and the universe is measured in
terms of a unit that all have experienced by walking over it on
 Principles , Assumptions , and Construction of the Pt olemaic System 133

T ABLE 3.
Mean Apparent True diameter Volume
distance in diameter comp ared to compare d to
Earth radii compare d the Earth's the Earth 's
to th e Sun's
Moon 48 113 l +~
1
40
1 1 1
Mercury 115 15 27 19,683
1 1
Venus 622l 10 l+~ 44
Sun 1,210 1 512 166!
Mars 5,040 1 117 112
20
Jupiter 11,504 12
1
4l + to 82l+~
1
Saturn 17,026 18 4l + ~ 79l
1. Magn. Stars 20,000 1
20 4l + ~ 94~ +~

their own two legs. Ptolemy performs these multiplications in

P.H. I, ii, 4.
In the final section Ptolemy finds the sizes of the planets, the
Sun, and the Moon. Here he has to call on a tradition of esti-
mates of the apparent angular diameters of planets and fixed
stars, a matter not dis cussed in the Almagest. The values he
chooses to use are given in the se cond column of Table 3, and
they represent the apparent diameters of planets and fixed stars
as fractions of the Sun's apparent diameter. He ascribes some of
them to Hipparchus and sa ys that he found the others himself,
but he does not sa y how. However, they can surely be nothing
but rough estimates, depending more on the brightness of a
planet than on the size of its apparent disc , for it was not until
after the introduction of the telescope that one could see that
planets present a disc and that fixed stars, no matter how bright,
remain mere points, no matter how great the magnification.
Ptolemy assumes that these estimates obtain when the celes-
tial bodies are at mean distance, so he next takes the average of
d and D in Table 2 and finds the values listed in the first column
of Table 3. The too-large value of the apparent lunar diameter
does indeed follow from his much-too-small value of the mean
distance of the Moon.
Ptolemy now refers us to the Almagest (V, 16), wh ere he has
shown that the solar radius is 5! terrestrial radii and, indeed, an
angle of 0;31,20 0-the
Sun's apparent diameter-subtends a
chord of very nearly 11 t .r. in a circle of radius 1210 t.r., the
134 3. Ptolemy's Cosmology

Sun's mean distance. If we now multiply the fractions in the

second column of Table 3 by 5!, we would find the true planetary
diameters in terrestrial diameters provided that the planets
were all placed at the same distance as the Sun, which, of
course, they are not. We must then bring each planet to its
proper distance or, more precisely, divide by 1210 and multiply
by the appropriate mean distance listed in the first column of
Table 3 in order to obtain each planet's true diameter in terres-
trial diameter. In Mercury's case we must then take the 1/15
from the second column, multiply it by 5!, divide by 1210, and
multiply by 115 from the first column. Ptolemy rounds the result
to 1/27, which is entered in the third column. [The value 1/28
quoted by Dante from al-Farghani (see p. 115) is, in fact, closer,
for our calculation yields 0.0348, in decimals, while 1/27 = 0.0370
and 1/28 = 0.0357.] Thus, Ptolemy arrives at the true diameters
of the planets compared to the Earth's in Table 3-all manu-
scripts agree on the value 1/4 + 1/20 for Venus, but this is an
obvious error for the correct 1/4 + 1/30.
Ptolemy now tacitly assumes that all celestial bodies are
spherical like the Earth, and in the last column of Table 3 their
volumes are compared to the Earth's. The numbers here are
simply the cubes of the numbers in the previous column.
4
Kepler Motion Viewed from
Either Focus

In the first chapter we analyzed a Babylonian planetary text and

saw that the Late-Babylonian astronomers had shaped arith-
metic into a powerful tool for addressing astronomical problems.
In contrast, the Greek planetary models, qualitative as well as
quantitative, employed geometrical models with moving parts
and a knowledge of how to combine velocities. We call such
models cinematical, for cinematics is the branch of mathematics
concerned with motion but without regard to masses and forces.
Finally, dynamics-the study of the behavior of a system of
masses in terms of the forces that act upon them-was created
in its useful form by Isaac Newton (1642-1727). In his Philoso-
phiae naturalis principia mathematica (Mathematical Princ iples
of Natural Philosophy), usually called the Principia (1687), he
formulated the basic rules of dynamics and successfully applied
them to a variety of physical problems, among them that posed
by the solar system.
The last major chapter in the story of the cinematical
approach to planetary astronomy was written by Johannes
Kepler (1571-1630), though he himself would very likely have
objected vehemently to this characterization of his work, as we
shall see. By degrees he formulated the laws that still bear his
name, among them his discovery that each planet, including
earth, travels in a planar, elliptic orbit with the sun in one focus
and such that the area swept out by the line joining sun and
planet grows uniformly with time. More specifically, as Figure 1
illustrates, he found that the planet P travels in an ellipse whose
major axis remains very nearly fixed relative to the fixed stars
and with the sun S at one focus. The far endpoint of the major
axis from S is called the aphelion A ("aphelion" means the point
farthest from the sun), and the planet moves so that the shaded
A. Aaboe, Episodes From the Early History of Astronomy
© Springer-Verlag New York, Inc. 2001 135
136 4. Kepler Motion Viewed from Either Focus

FIGURE 1.

area ASP is c - T, where c is an appropriate constant and T the

time elapsed since the planet was last at aphelion.
In the present chapter, I shall try to make clear what is
involved in finding a planetary position when planet and earth
are known to obey Kepler's rules; then I shall briefly outline
how Kepler himself solved this problem; finally, I shall show
that Ptolemy's equant model owes its success to a certain prop-
erty of Keplerian motion-that if you observe a planet, not from
the sun, but from the other and empty focus of its orbit (a much
more comfortable place to be), then the planet will seem to
travel very nearly uniformly around you.

Properties of Ellipses
Before I turn to these things, it might be well if I briefly review
some of the results and techniques we shall need in the fol-
lowing, particularly such that concern ellipses. In current texts
on analytic geometry the ellipse is commonly introduced as the
locus of points satisfying a certain geometrical condition. Let
there be given two points, F I and F 2 , and a distance 2a > F IF2 ;
then the ellipse is the locus of the points P the sum of whose
distances from F 1 and F2 is always 2a, namely,
F1P + F2P = 2a.
If we now introduce the eccentricity e as the ratio between
F 1F2 and 2a, and a coordinate system as in Figure 2, then we can
derive the equation of the ellipse with the standard techniques
of analytic geometry-we use the distance formula-and after a
 Properties of Ellipses 137

"

FIGURE 2.

reduction we get
x2 y2
-+
a 2 (1 - e2)a 2 -1
- ,

or, setting (1 - e 2 ) • a 2 = b2 ,
x2 y2
a 2 + b2 = l.

The points A, B , C, D in Figure 2 are called the vertices of the

ellipse, F 1 and F 2 are its foci , 0 is its center, the distance
OA = a its semi-major axis, and the distance OB = aJl=e2 = b
its semi-minor axis.
When one reads older texts it is useful to know that latus rec-
tum is the term used for the chord through a focus and perpen-
dicular to the major axis. For its length we get
2b 2
I =- = 2a(1 - e2 ) .
a
If we solve the equation of the ellipse for y, we obtain
b
y = ±-Ja
a
2 - x2 ,

where the signs refer to upper and lower halves of the curve,
respectively. If PI of coordinates (x,Yl ) is a point of the ellipse's
circumscribed circle, which has center 0 and radius a, we find
138 4. Kepler Motion Viewed from Either Focus

FIGURE 3.

or agam
YI = ±va2 - x 2 .
If P and PI are two points, the former on the ellipse and the
latter on the circle, with the same abscissa and on the same side
of the major axis (see Figure 3), we get for their ordinates, from
the preceding equations
b
Y = -YI·
a
We may then think of the ellipse as the curve we obtain when we
compress the circumscribed circle in a constant ratio (b: a) in a
direction perpendicular to the major axis. Thus, if you think of
the areas above the major axis and below the upper halves of
ellipse and circle as made up of a great many, very narrow ver-
tical strips, and likewise for the lower halves, it should become
clear that the areas of ellipse and circle have the same ratio as
the ordinates. The circle's area is 1ra2 , so the area enclosed by
the ellipse is
b
- . 1ra2 = nab ,
a
 Properties of Ellipses 139

FIGURE 4.

FIGURE 5.

One more property of the ellipse will be of use in the follow-

ing . If P is any point on the ellipse, then the focal rays to P , F1P
and F 2P, will form equal angles u with the tangent to the ellipse
at P; see Figure 4. This implies that the ellipse will reflect rays
emitted from one focus so that they pass through the other-
hence the name "focus" for the points F 1 and F 2 •
Finally, the equation of an ellipse in polar coordinates takes
on a particularly pleasing form if the pole is placed at a focus .
With the pole at F 1 and the polar axis along the major axis as in
Figure 5, the equation of the ellipse is
l
2 a(l - e2 )
r= =
1 - e cos () 1 - e cos ()
,
140 4_Kepler Motion Viewed from Either Focus

rs r(8")

o o
FIGURE 6.

and with the pole at F 2 and the polar axis as before, it is

I
2 a(l - e2 )
p=
1 + e cos 'P 1 + e cos 'P
These formulas can be derived, or verified, with the standard
techniques of analytic geometry.
Two formulas that have to do with curves in a polar coordi-
nate system will be employed at the end of this chapter, and I
may as well include them in this swift review. If we are given a
curve whose equation in polar coordinates is r = r(O), then the
area enclosed by the curve and the rays 0 = 01 and 0 = O2 [see
Figure 6(a)] is

Furthermore, if P is a point on the curve, and u the angle from

the polar ray through P to the forward half of the tangent at P
[see Figure 6(b)], we have
dO
tan u = r --
dr
or, where appropriate, for example, when drf di) = 0,
1 dr
cot u = ~ - dO-
 Kepler's Equation 141

All these things were known in antiquity except those that

have to do directly with rectangular or polar coordinate sys-
tems. In fact, in matters concerning the geometry of the ellipse,
Kepler refers to Archimedes (ca. 287-212 B.C.) and Apollonius
(ca. 200 B.C.) as his authorities. The ellipse is one member of a
family of curves that the Greeks called the conic sections, for
these curves were first thought of as generated when you cut a
conical surface with a plane.
Before Apollonius the surface was always assumed to be that
of a right circular cone, and the cutting plane had to be perpen-
dicular to a generator-various attempts, none quite convinc-
ing, have been made at explaining this curious convention. With
this old definition it is then the angle at the vertex of the cone
that determines the kind of conic we get: When this angle is
acute, we obtain an ellipse; when right, a parabola; and when
obtuse, a hyperbola. The earlier Greeks consequently called the
three kinds by the names "acute-angled," "right-angled," or
"obtuse-angled" conic sections, and Archimedes still used this
terminology in the books where he talked of such curves.
Apollonius's great work on conic sections was in eight books
of which the first four survive in the Greek, the next three only
in Arabic, while the last is lost. Here he began with a new defi-
nition of conic section and introduced the names for them that
we still use. For him the conical surface could be that of any
circular cone, right or skew, and he considered any plane cut in
such a surface as one of his conic sections. He showed, inciden-
tally, that, despite the more relaxed definition, his family of
curves is identical to that obtained under the older convention.
I should emphasize that for Kepler the works of these and
other ancient authors were not objects of historical study, but
treasured sources of technique and knowledge.

Kepler's Equation
We are now prepared to address the problem that has been cen-
tral in planetary theory since the work of Ptolemy: Given the
time, find the direction from us to the planet in question. We saw
how Ptolemy solved it, but now we assume that the planets and
the earth obey Kepler's rules, which, of course, they do, except
for minor perturbations caused by mutual attraction between
the planets.
142 4. Kepler Motion Viewed from Either Focus

FIGURE 7.

The first step toward that goal consists in finding the position
of a planet in its orbit at a given time. Figure 7 represents the
planet P on its ellipse with the sun S at one focus and the apsi-
dalline SA always pointing to very nearly the same place among
the fixed stars. Before we begin our calculation, we must be pro-
vided with information about the orbit: the direction of its apsi-
dalline and its size and shape or , in other words, its semi-major
axis a and its eccentricity e. I ought to include among these ele-
ments two more indicating how much, and around which line
through S, the plane of the orbit is tilted against the plane of the
ecliptic, which is that of the earth's orbit. However, for the five
classical planets the angles of inclination are rather small, so for
the sake of simplicity we may assume that our planet's orbit lies
in the ecliptic without much consequence for the planet's longi-
tude. Further we must know the planet's period p, usually given
in days, that is, the constant number of days it takes the planet
to travel once around its orbit (e.g., from the aphelion A and
back again) as well as time t(A) when it last was at A [all possi-
ble values of t(A) are separated by whole multiples of pl.
I cannot afford here to describe how these elements were first
derived from observations and later refined by degrees, for this
story is as complicated as it is worth telling. For our present
purpose it suffices to take them as given.
If we are then presented with an arbitrary time t and wish to
find where in its orbit the planet is at that moment, we must first
find how long a time interval T has elapsed since the planet last
was at aphelion and obtain
T =t- t(A) .
 Kepler's Equation 143

The planet moves so that the shaded area in Figure 7 is pro-

portional to T, that is,
area ASP = c · T,
where c is a constant we can find from the provided elements.
Indeed, when T equals p , the period of the planet's motion, then
the radius vector SP will have swept out the entire area of the
ellipse since the planet last was at aphelion, and we get
area of ellipse = nab = c . p,
which yields
1I"ab
C=- ,
P
and so
1I"ab
areaASP=-· T .
p
We now have the area expressed in terms of time and orbital
elements, and we will endeavour to find the planet's position
from this or, more precisely, we will seek the radius vector
r = SP and the heliocentric elongation () from the aphelion A
(see Figure 7).
To that end we introduce an intermediary quantity E, already
called the eccentric anomaly by Kepler, in the following fashion
(see Figure 8). First we draw the orbit's circumscribed circle, of
center 0 and radius a. From the planet P we drop the perpen-
dicular PPo to the apsidalline and extend it the other way until
it intersects the circumscribed circle at Pl. It is the angle AOPI
that we call the eccentric anomaly E, and we shall now express
the uniformly growing area in terms of E and so establish a
connection between the time T and E.
Here we notice first that the area bounded by the apsidalline,
the circumscribed circle, and the line SPI consist of two simple
parts:
area ASPI = sector AOPI + triangle OSPI .
If we measure E in radians, we have

sector AOPI = !a 2E

(note that the area of the sector is proportional to E and that to

E = 211" corresponds the area of the whole circle, 1I"a 2 ). Triangle
144 4. Kepler Mot ion Viewed from Either Focus

FIGURE 8.

aSPI has as its base as = ea and as its altitude PoP!> which is

POP I = aPI . sin E = a sin E ,
so we have

Thus, we get
area ASPI = !a 2E + !ea 2 sin E = !a 2 (E + e sin E ).
We recall that we obtain the ellipse if we compress its cir-
cumscribed circle in the constant ratio b: a in the direction per-
pendicular to its major axis, and note that if we reduce the
ordinates of the points on the line segment SP1 in the same
ratio, we obtain the ordinates of the points of SP (we think of
the major axis as the abscissa axis). It should then be clear that
the areas ASP and ASPI also have the ratio b: a, and we have

area ASP = ~ . area ASPI = ~ab(E + e sin E ).

a 2
Combining this with our earlier result that
nab
areaASP=-· T ,
p
 Kepler's Equation 145

we obtain

E + e sin E = 271" ' T .

p

It is customary to denote the right-hand side by M . We note

that M is a quantity that grows uniformly with time; when T
goes from 0 to p, M increases from 0 to 271", as do E and (), and M
can therefore be thought of as the measure in radians of an
angle whose right side is, say, SA and whose left side performs
one revolution uniformly while the planet travels from aphelion
and back again. We call Mthe mean anomaly. With this notation
we have

I E + e sin E ~ M. I
I should remark that M is not on Figure 8. When M is 0 or 71",
M and E are equal; otherwise, M > E for the upper half of the
orbit in Figure 8 where e sin E is positive, and M < E for the
lower half. If the eccentricity e is zero, so the orbit is a circle, we
always have E = M and so () = M . Uniform circular motion is
thus a possible, though unlikely, Keplerian motion.
The boxed formula is Kepler's equation, and it allows us to
determine E when the time, and so M, is given. Finding E from
this equation is not a trivial matter, and I shall postpone a dis-
cussion of how it can be attempted, but let us assume for the
moment that it has been done so that we now know E.
Once we have the eccentric anomaly E, our difficulties are
over, for we can then easily find the position of the planet P in
its orbit or, more precisely, its distance r from the sun and its
heliocentric elongation from the aphelion, () (see again Figure 8).
We have
POPl = a sin E,
so

PoP = ~a . POPl = b sin E = ~ . a sin E.

Furthermore,
GPo = a cos E ,
146 4. Kepler Motion Viewed from Either Focus

so
SPo = ea + a cos E = a(e + cos E).
Since t:lSPoP is a right triangle, we have
r 2 = Sp 2 = SpJ + P Op2 = a2(e + cos E)2 + (1 - e2)a2 sin 2 E ,
from which we get, after a slight reduction,
r 2 = a 2(1 + 2e cos E + e2 cos 2 E) = a 2(1 + e cos E)2,
so
r=a(l+ecosE).
As to (), we have from t:lSPoP
() _ SPo _ a(e + cos E) _ e + cos E
cos - SP - r - 1 + e cos E '
from which () is readily found.
Thus, a knowledge of E allows us to determine rand () through
the simple formulas

e + cos E
r = a(l + e cos E) and () = arccos E.
1 + e cos

The second formula implies the prettier formula

() ~ E
tan '2= V~·tan 2'
For the sake of completeness, I repeat here the formula from
page 139:
a(l- e2 )
r--'----
- 1 - e cos ()'

Resolution of Kepler's Equation

We do not meet Kepler's area law as we know it until his Epit-
ome of Copernican Astronomy, published in four parts from 1617
to 1622, but Kepler's equation appears already in Astronomia
Nova of 1609. Here he talks not about the area of a part of the
elliptic orbit, but of the area inside the circumscribed circle-
 Resolution of Kepler's Equation 147

the eccentric circle, as he calls it-which we call area ASPl in

Figure 8. He shows that it provides a measure of "the sum of all
the radii" from the sun to the corresponding part of the ellipse.
Indeed, the left side of our form of Kepler's equation
E+esinE = M ,
when multiplied by ~a2 is the area ASPl , as we saw in the
course of its derivation. Here is what Kepler has to say about his
equation (Astronomia Nova, Chapter 60):
If the mean anomaly (our M) is given, then there exists no geomet-
rical method for deriving either the corrected anomaly (our B) or the
eccentric anomaly (our E). For the mean anomaly is composed of two
areas, one sector (area AOPl in Figure 8) and one triangle (OSPl ) .
While the former is measured by the arc on the eccentric circle (the
orbit's circumscribed circle), one obtains the latter by multiplying the
sine of this arc with the area of the greatest triangle (!ea 2 ) and cutting
off the last places. There are infinitely many ratios between an arc and
the corresponding sine. If then the sum of the two be given, it is
therefore impossible to say how large the arc may be, and how large its
sine, that correspond to this sum; i.e, unless we first find out how large
the area is that belongs to a given are, i.e. unless we construct tables
of (E + e sin E) and use them backwards.
That is my own way of thinking. The more wanting in geometrical
beauty it may appear to be, the more strongly do I urge the geometers
to solve for me the following problem:
Given the area of a part of a semicircle, and given a point of the
diameter, to find an are, and an angle at the point, so that the sides
of the angle and the arc enclose the given area. Or: to cut the area of
a semicircle in a given ratio from an arbitrary given point on the
diameter.
It is enough for me to believe that a solution a priori is impossible
because the arc and the sine are heterogeneous in nature. But whoever
proves me wrong and shows the way (to a solution) will seem to me as
great as Apollonius.
Terseness was not a characteristic of Kepler's prose style, but
he is right when he says that a solution is impossible if by that
he means that E cannot be explicitly written in terms of the
kind offunctions of e and M that he knew, for example, trigono-
metric functions and their inverses. The equation is what we
call transcendental, for one can show that its solution cannot be
expressed as a finite combination of elementary functions of e
and M. [I here insert that by an "elementary function" we still
mean one that L. Euler (1707-83) dealt with in his Introductio in
148 4. Kepler Motion Viewed from Either Focus

Analysin Infinitorum (1748), the ancestor of all textbooks on cal-

culus; they are essentially exponential and trigonometric func-
tions, their inverses, and algebraic functions.] That a solution
exists is obvious, for the function
y = x+ e sin x, o::; e < 1,
has a positive derivative (1 + e cos x ~ 1 - e > 0), so it is strictly
increasing and therefore has an inverse.
Kepler suggests that though it is hard to find E when M is
given, it is easy to find M when we know E (and e). Thus, one can
readily make a table of M = E + e sin E for a dense set of values
of E and use it backward. As we shall see, this is in effect what
he does . The disadvantage of this method is that we must make a
separate table for each value of e, that is, for each planet.
A considerable literature grew up on the solution of Kepler's
equation-already Newton gave much attention to this problem
-but if all you want is a specific, numerical solution, it can be
obtained swiftly and painlessly, particularly if you have at hand
a little calculator with trigonometric functions built in. The
following iteration method works well.
First a remark on units: We have so far assumed that E
and M in Kepler's equation were measured in radians, but in
astronomy one prefers to work with degrees. We therefore mul-
tiply Kepler's equation by 180/7r and get
180 180. 180
-·E+-·esmE=-·M
7r 7r 7r
or, if we now let E stand for the measure of that angle m
degrees, and likewise for M ,

E + 180 . e sin E = M.
7r
It might have been more elegant and, more importantly,
clearer if I had introduced a particular typeface for the measure
of the angles E and M when expressed in degrees rather than in
radians, but I adhere to astronomical custom. However, I shall
set
180
-·e= c
7r
so that Kepler's equation in the form
 Resolution of Kepler's Equation 149

TABLE 1
e 0.0933 0.2056

M 600
600
E1 55.370491 49.798210
E2 55.601327 51.002710
E3 55.589123 50.844860
E4 55.589767 50.865317
E5 55.589733 50.862662
E6 55.589735 50.863007
E7 55.589735 50.862962
Es 50.862968
Eg 50.862967
ElO 50.862967

E+csinE = M
reminds us that E and M are measured in degrees (sin E is, of
course, the same no matter what unit E is measured in).
We now rewrite the equation as
E = M - e sin E,
and from a given value of M we compute a sequence of what
turns out to be ever-better approximations of E, beginning with
Eo=M:
E l = M - e sin M ,
E2 = M - e sin E l ,
E3 = M - e sin E2 ,

If, or rather when, an E n+l turns out to be equal to En (to the

desired number of decimal places) we have a solution of the
equation. One can prove in general that this process con-
verges for 0 ::; e < 1, but this is not relevant for each specific in-
stance, for the method is self-checking. In Table 1 I listed two
sequences of En's, one for Mars (e = 0.0933) and one for Mercury
(e = 0.2056), to show how quickly it works (the fractional parts
of E are written decimally).
To show how Kepler himself attacked his problem in practice,
I reproduced a piece of the Rudolphine Tables (1627) in Table 2.
It consists of the top and bottom of the first of three pages de-
voted to the anomalies of Mars (Kepler's page is too long to be
 150 4. Kepler Motion Viewed from Either Focus

TABLE 2

6~ Ta!J,,1ar1l1lJ 1(lItiolphi
Tabula JEquationum MAR TIS.
Anomalia Interee- Anomwa Inrerco- Inrervallii
Eccenrri. Iumnium, Anomw Inrervallu Eccenrri, Iumnium, Anomwa C•• Ul....
c•••,_,. c...Ur cozqllUa. ":••Ul·· c.",.,••,.. c•• lAt.
Mil, ..rtr po). .",1.... r;,"_ 1II1,.,u1h.J' •..,11,...
cozquara. rit.lfltI
-
, Par. ,,- , ,. - ,
° " Gr.
0. 0. °
I 66 ...6f 3° t " 119 60 Gr. " 164-57 1
0. 0. °
I .11'0
1°9 6•
166+61 °
----
1.39·14
3I
0·51· 9
III I °
17.16 .37 4;1818
1644+7
12

-- 0 . I ·H 0·50. 3
1 111)0
0 .54.+ 1 1°9 60
166+56 °
a .+4. 1 O·f r.1 3
32 J 16"10
18 ·1l·P +;174'
16+~19
II

0.11. 7 0.;0. 3 1.49· U 10J/17 '.4 1 .·P 0.; 1.1 8 19·17.19 49 6 61

I II
3 1'110 166446 H '1+9 0 164-18 7
0.,6 .+0 0·50. ~ 1·44· l 1°91° 1·13.) I 0.; 1.11 30.n~ ~ .I+
...0.1: .13 I SI • 0 166+~1
I
34- 113 1 0 100\005 1
3 I. 8·11 ';'yjOI
Il

1°9 4 1 a , j $.1 ° o.f 1.19

---- 0 ·50. 3
-"---"
H8,:1:1
1
.... "
I)
;- 1'090 1664- 11 35 1111° 16~911
0.17.4 6 0·50. 4 ...·n ·1 ; 1°930
---1
) . 0·5 1'~4 p . H·I 45>4 16
16 376 9 I'
6 1807 o. 166 3 8 8 36
,. 7 . 11 1+5>7°
P·f9· 14- 49)19
----0.ll.·8 o·fO. f
7 11 04 0
----
;.18· 7 5 0 9 16
~ -- 3
166360 37
0 .;1 ·39
1475>° 16 1 6 1 3
14
0 .) 1.10 0 .50. 6
8 I¥OlO
6.n ·+9 1 0899
166 31 8
, H
3.11 .4 {. o·P·4-f
1+600
3 ~·54--;0 49'4°
16 3+ 7+ '4
0.44 . 1• 0·50. 7 7· 17 ·P 5 0 17 9 ,.1,".10 o·f 1·51 34-·5°·19 491+9
- .- ,
I
-. ~: ,
I I
I • "070
I

I ......
I • 1°',} '"
I'_' ~ 0.;0 ·34 19· 1o.: U 0 19 6\ 4-6 . 59 . 9 477 67
__
1 I--+'-~ 0·n ·17
116 1° 1610 19
19
11 J69~0 16S4.J7 I f1
1·19 ·' ~ 0·5 O.}.8 :0. f·U 1°)41 4010·19 0·fP4 +7·55-lB 47 641
-----
1~ 16170 16 534-3 9 H Iii' a 160844 I'
1. +., I 0.5° .4-1. z r , 0 .13 101S1 4.14.• 1 o·B ·P ...8.fl.11 47317
----
14- '''7 60 16P",; II ;+ I I .10 16064-6
15>
11 .55. 17 1° 1• 6
---
1~
:1. Q.] 0 0·5°,-44
1664C 16fl4-~
10
4·17 ·4 ° 0·51 ·19
Sf 1°9+°
+9·+8.+8 474°4
16044()
.0
101 64 4. 10 .1 3 0·53·4-7
--. - ----
1. t ..... ;, ~ 0 ·5°48 11·5°·33 fO·4-S·3" 47 '7;1
10 10
::6 16 p o 16 50 36 f6 · 107°C 16014 4
>.I,.}4 0 ·5°·P 13 ·H·4-1 1 0100 +··4. I 0·5 JoSS fl.·p·16 411 ~ 1
10 10
17 '6)90 16+9 16 57 1 041 " 160°39
z , 1 ... 31 0.;0·56 140....°.5 1 100H +.17 . 6 0 ·5+ ~ 51 ·~9· 6 47° 14
II 10
18 161 10 16 40 8 11 58 10100 15983°
1.19.19 0.51. ° 1;·36. 5 4996"4 +.30. Ii 0.540.I I SH6. ° + 6 894
16.,0 164-694-
Jl - --- - 994 0 159 ( 1)
II
19 59
I ., .... 3 0.5 J. + 16.31.10 +919 1 4·) ,. 1 0 .54-. 19 54-.3 1.58 4 67 6 ;
11 ZI
3° 11960 16+571 60 9 69 0 159+°9
1.'9.1 4 0·51. " 17.16·V 4\111~ •• IS·SO 0·54·17 55 .30. ° 4 6 63 0)
-
 Resolution of Kepler's Equation 151

T ABLE 3

En IIn L\M
L\On I X 105 rn X 105
n
L\On On
e sin En L\M (sexag.) lIn rnl X 105
n
180
€ = e . - , e = 0.09265, a = 1.52350 a .u.;
7r
E: eccentric anomaly;
M : mean anomaly, from aphelion; M; = En +
e sin En;
0: heliocentric longitude, from aphelion;
r: distance from sun in astron units;
L\Mn = M n - M n- I ; L\On = On - On-I·

adapted to the present format and remain legible). There is

an entirely analogous set of three pages for each of the other
planets and for the sun-we shall presently see why this last set
is for the sun and not the earth.
The page is divided into two columns, each of which consists
of 31 rows, except for headings. Such a horizontal row is built
up of four little boxes, and I presented the structure of such a
row in schematic form in Table 3. The upper number in the first
or leftmost box represents the eccentric anomaly E in degrees,
and E is the independent variable of the tables in the sense that
Kepler chose values for E first and then derived the rest from
them. There is a row corresponding to every whole number of
degrees of E from 0 to 1800 • The index n refers now to the
0

number ofthe row of boxes; it has nothing to do with the indexes

in Table l.
The third box gives in degrees, minutes, and seconds what
Kepler calls the corrected anomaly, our 0, calculated from his
version of our formula
o e + cos E
cos = 1 + e cos E '

where his value of the eccentricity for Mars is e = 0.09265.

The upper number in the fourth box represents the distance
from sun to Mars, our r, measured in astronomical units and
calculated from the equivalent of our formula
r=a(l+ecosE) ,
152 4. Kepler Motion Viewed from Either Focus

with the mean distance a = 1.52350 a.u. Kepler gives these

numbers to six places, but without decimal points, so that they
in fact are r- 105 .
As already remarked, it may be difficult to solve Kepler's
equation
E+csinE=M
for E when the time, and so M, is given, but it is easy to find M
when E is given, and that is, in fact, what Kepler does , keeping
E as his independent variable for a short while longer. Indeed,
the lower number in the first box is simply e sin E, expressed in
degrees, minutes, and seconds and computed from the value of E
just above it (Kepler calls the quantity e sin E "the physical part
of the equation" or correction). The sum of the two numbers in
the first box is, then, the value of M to which correspond the
values of () and r in the same row . In other words, Kepler has
very simply succeeded in turning the table of equally spaced
values of E and the values of () and r derived from them into a
table correlating unequally, but still closely, spaced values of M
with their corresponding values of () and r.
These planetary tables are preceded by a set of mean motion
tables that enable you to find the mean anomaly M correspond-
ing to any given time t for each planet. I shall describe the sim-
ple structure of these tables later. If the value of M you obtain in
this way happens to be one of the 181 sums of the two numbers in
the first boxes, you are provided directly with the corresponding
() and r . If not-and this is, of course, much more likely-you
must find () and r by linear interpolation, and some of the re-
maining numbers in the boxes are there to help in this process.

Example
Given M = 60°, find () and r from Table 2.
The nearest values of M in Table 2, and their corresponding
values of () and r, are
M = 55° + 4;20,53° = 59;20,53° ~ () = 50;45,30° ~ r = 1.60446
and
M = 56° + 4;24,2° = 60;24,2° ~ () = 51;42,16 ~ r = 1.60244.
The lower number in the second box belonging to E = 56° is
the ratio of the difference between the two ()-values and the two
M-values, expressed sexagesimally, so we learn that the incre-
 Resolution of Kepler's Equation 153

ment in () corresponding to an increment in M of l Oin this in-

terval is 0;53,55°. Our value M = 60° is 0;24,2° smaller than
60;24,2°, so the associated value of () is found thus:
from Table 2 : () = 51;42,16
0;53,55·0;24,2 = 0;21,36
() = 51;20,40°.
so for M = 60° we get by subtraction
Likewise we find, but now by direct interpolation without the
benefit of a provided interpolation coefficient, that to M = 60°
corresponds r = 1.60320 a.u.
To test the reliability of Kepler's tables, I computed () and r
directly from M = 60°, but using his values of eccentricity and
mean distance, e = 0.09265 and a = 1.52350, for Mars. Solving
Kepler's equation, I found the eccentric anomaly E = 55.6189322°
and hence
e+ cos E
() = arccos E = 51.342568° = 51;20,33,14°
1 + e cos
compared to his () = 51;20,40°, and
r = a(l + e cos E ) = 1.60321 a.u.
compared to his r = 1.60320 a.u.

The rest of the numbers in Kepler's tables are, as indicated

in his headings and in Table 3, logarithms of the interpolation
coefficients and of r provided to be used in further calculations
in combination with his various logarithmic tables in the Ru-
dolphine Tables. Kepler's logarithms were close to, but not
identical with, our natural logarithms, but I cannot go into these
matters here.

Remark
Those already familiar with planetary astronomy will have no-
ticed that the formulas I derived and used look slightly different
from the ones they learned from modern textbooks. The reason
is that for obvious reasons I adhered to the older convention,
introduced by Kepler and followed by Newton and many others,
of referring the motions to the aphelion instead of the perihe-
lion, as is the current practice.
154 4. Kepler Motion Viewed from Either Focus

Indeed, if M, E, and 0 are counted from the perihelion (the

point on the orbit nearest the sun), Kepler's equation will now
be
E - e sin E = M,
and we further have
cos E - e
cos O= - - - -
1- e cos E
and
r = a(l - e cos E) ,
as is readily checked.

Heliocentric and Geocentric Longitudes

Since the work of Ptolemy, celestial longitudes have been
counted along the ecliptic from the point occupied by the sun at
spring or vernal equinox. This point marks the beginning of the
zodiacal sign Aries on the celestial sphere and is consequently
marked 'rO°. It is, then, the direction from the earth to the sun at
vernal equinox that provides the basis from which geocentric
longitudes are counted.
We have hitherto indicated the direction from the sun to a
planet relative to the planet's apsidal line, but no two of these
lines coincide. When we now introduce heliocentric longitude to
indicate directions from the sun in the plane of the ecliptic (that
of the earth's orbit), we reckon it from the same basic direction
as geocentric longitude (note that two parallel directed lines
have either the same or opposite directions). This convention
assures that the geocentric longitude and the heliocentric lon-
gitude of a fixed star are equal, for the star is, to all practical
purposes, infinitely far away so that the lines of sight to it from
earth and sun are parallel.
Figure 9 represents the earth's orbit with the sun S in one
focus-here, as elsewhere, I have greatly exaggerated the ec-
centricity and have let the apsidal line point in a direction that
has no particular relation to the correct one. E; and E a are the
positions of the earth at vernal equinox and autumnal equinox,
respectively, and E v , S, and E a lie on a straight line as indicated.
The direction to '1'0 is, as said, the direction from E; to S, so the
0
 Heliocentric and Geocentric Longitudes 155

E, q - - - - - o - - - - - - - - p

FIGURE 9.

.- _.- .- - ._.- .- .- ._.- 'r 0'

FIGURE 10.

basic direction, the zero direction, or what you will , from which
we measure heliocentric longitudes is that from 8 toward E a , the
place of the earth at autumnal equinox.
In Figure 10 l(E) is the heliocentric longitude of the earth and
-\(8) the geocentric longitude of the sun. As the figure shows,
they differ by 180 This simple relation is crucial for what fol-
0
•

lows. Both -\(8) and l(E) increase by 360 in one year in the
0

counterclockwise sense if the ecliptic-the plane of the paper-

is viewed from the north.
Figure 11 indicates what we need to do when we wish to find
the heliocentric longitude l(P) of a planet P. We must be pro-
vided with the heliocentric longitude l(A) of the aphelion A and
increase it by the corrected anomaly o. Kepler gives us the
means of finding l(A) in the mean motion tables that precede
156 4. Kepler Motion Viewed from Either Focus

FIGURE 11.

the anomaly tables for each planet. We obtain 8 either via the
eccentric anomaly E from Kepler's equation or by interpolation
in his tables and have
l(P) = l(A) + 8.
To enter Kepler's mean motion tables we must be provided
with a value of the time, t, given in terms of year, month, day,
and hour. He reckons his years according to the Julian Christian
calendar in which every fourth year is a leap year so the average
length of the year becomes precisely 365 1/4 days. He gives
chronological tables for converting dates in other calendars-
Jewish, Islamic, and Gregorian-into Julian dates.
The Gregorian calendar, the one we now follow, was in-
troduced by Pope Gregor XIII in 1582 by a papal bull that de-
creed that in that year October 4 should be followed by October
15. Its aim was to bring vernal equinox back to March 21 and
keep it always near this date on which it had fallen at the time of
the Council of Nicea in 325 when rules for calculating Easter
had been laid down (Easter Sunday should be the first Sunday
after the full moon that happens on, or next after, the day of
vernal equinox; to calculate its date in advance is a problem of
some nicety). The Julian year of 365.25 days is slightly longer
than the tropical (seasonal) year of 365.2422 days, so vernal
equinox had slipped back to March 11; the new calendar pre-
vents such slippage by omitting leap days in years ending in two
zeros, except when the year number is divisible by 400 (e.g., 1600
and 2000). The new calendar was adopted immediately in most
Catholic countries, but the Protestant northern parts of Europe
 Heliocentric and Geocentric Longitudes 157

did not change over until much later, and in different years in
different countries. England held to the Julian calendar until
1753, and by then the difference between dates in Old and New
Style, as they were called, had grown to 11 days. Thus George
Washington was actually born on February 11, 1732, Old Style,
but his birthday is now celebrated on February 22, the equiva-
lent date in New Style.
Kepler has his year begin at noon on January 1. It may sound
trivial that January 1 is the first day of a new year, but we find,
in fact, a bewildering array of conventions of day and month of
New Year's Day, sometimes even different for different purposes
at one place and one time (we still have our fiscal years, academic
years, and tax years beginning on different dates). March 1 was
widely used for that purpose, and in such a calendar September,
October, November, and December are, indeed, the seventh,
eighth, ninth, and tenth months as their names indicate, and the
leap days at the end of February are added at the very end of the
year. In England the new year began on March 25 (Lady Day,
Annunciation Day) until 1751.
When Kepler begins his astronomical " day" at noon rather
than at midnight, he follows a tradition that goes back to
Ptolemy. The advantage of this convention is that the observa-
tions made during a given night all carry the same date, and it
was followed in most astronomical works until the end of 1924,
from which time astronomers had agreed to switch to the civil
calendar's midnight epoch.
Kepler reckons hours for the meridian of the small island
Hveen where Tycho Brahe had observed (ca. 12 r east of Green-
wich) , so his noon means noon at Uraniburg, Brahe's observa-
tory on Hveen. The Rudolphine Tables contain a list of terres-
trial longitudes-counted from this meridian-(and latitudes) of
a great number of important cities, as well as a map of the world
with coordinates drawn in , which enables you to convert your
local time to Hveen time.
The two parameters given in the mean motion tables for each
planet are the heliocentric longitude of the aphelion, l(A) , and
the mean heliocentric longitude of the planet, Z(P ), which in our
notation is
Z(P) = l(A ) + M ,
that is, the longitude of the aphelion increased by the mean
anomaly. [Note that Z(P) would be the heliocentric longitude of
158 4. Kepler Motion Viewed from Either Focus

the planet if the eccentricity e were zero so the orbit would be

circular and the planet's motion uniform with () = E = M.l
Kepler does not include tables for the earth among his plane-
tary tables; instead we find tables enabling us to find the geo-
centric longitude of the sun, 'x(S). In the mean motion tables we
find
'x (Ag ) = I(A ) + 180 0

and
X(S) = Z(E) + 1800 = I(A) + M + 1800 = 'x(Ag ) + M ,
where I as before stands for heliocentric longitude, ,X for geo-
centric longitude, Sand E for sun and earth, A for the earth's
aphelion, and Ag for the sun's apogee, the point farthest from
the earth on the sun's orbit.
For each planet, and for the sun, Kepler gives the values of
these two parameters for the beginning of every century for an
enormous time span. Further, he lists how much these parame-
ters grow in any (whole) number of years up to 100, in the inter-
vals from the beginning of the year to the end of each of the 12
months, and in any number of days up to 31. When year and date
are given, by simple addition we can then find Z(P ) I(A ), X(S),
and 'x(Ag ).
If the planetary orbit's major axis were truly fixed in relation
to the fixed stars, then I(A ) would grow precisely at the rate of
the precession (Kepler's value is very nearly 51" per year).
However, he assigns to each aphelion a very slow proper motion
so that I(A) grows at slightly different rates for different planets.
For a planet we now find
M = Z(P) - I(A) ,
and we enter the anomaly tables with this value of M to find ()
and r as described above. We immediately get
I(P) = I(A) + (),
but we save the value of r for later use.
For the sun we find likewise
M = X(S) - 'x (Ag )
and enter the solar anomaly tables with M to find () and r. These
tables are identical to the ones we would get for the earth if it
were treated like any other planet. Again we have
 Heliocentric and Geocentric Longitudes 159

._._.... 'Y' 0#
.- .- .-. eye 0#

FIGURE 12.

A(S) = A(Ag) + a,
and r is reserved for use in our next and final step, that of finding
the geocentric longitude of the planet.
Figure 12 represents the earth E and an outer planet P at a
particular moment in their orbits surrounding the sun S. Our
aim is to find A(P), the planet's geocentric longitude. We have
just seen how we may find l(P) and A(S) as well as the distances
SP and ES (they are the values of r) . We now consider triangle
ESP. We know two of its sides and can find angle PSE, for (see
Figure 12)
t:".PSE = l(E) - l(P) = 180 + A(S) -
0
l(P)
[due to space constraints, l(E) is not marked on the figure] .
Thus, we know an angle and the two enclosing sides of the tri-
angle, and we can find its remaining pieces by the standard
methods of trigonometry. We particularly want angle SEP, for
then we have the desired longitude
A(P) = A(S) + t:".SEP,
and, at long last we are finished.
So you see that in the end we still have to solve the same old
triangle earth-sun-planet that one has had to deal with since at
least the time of Ptolemy.
160 4. Kepler Motion Viewed from Either Focus

Kepler Motion Viewed from the Empty Focus

We shall now address the problem of how a planet appears to
move when observed from its orbit's empty focus -the one not
occupied by the sun. Our aim will be, first, to show that the
planet moves very nearly uniformly around this focus and,
second, to obtain an estimate of its motion's deviation from uni-
formity. There are, as we shall see , good historical reasons for
doing these things.
First let us see , however, in a purely pragmatical way if it is at
all worthwhile to tackle the problem. In Figure 13 and Table 4 I
compared three models for the planet Mars, whose eccentricity e
I have taken to be 0.0933, and whose mean distance from the sun
I have used as a unit of length.
Model I is the Kepler ellipse in which the planet P moves
about the sun S according to the area law . In Table 4 the col-
umns headed I list the values of rand () (in degrees and decimal
fractions thereof) corresponding to the values of the mean
anomaly M given in the leftmost column and produced by Model
1. I found the values of rand () by solving Kepler's equation and
otherwise following the procedures set forth above.
In Model II I retained the correct orbit, that is, an ellipse of
semi-major axis 1 and eccentricity e = 0.0933 with the sun in one
focus, but here I let the planet move uniformly about the other
focus F with the correct period. I did the latter by letting the
angle at F equal the mean anomaly M. In the columns headed II
I entered the values of rand () corresponding to the values of M
in the leftmost column, and in the columns I - II are the differ-
ences between the correct values of rand () and those derived

I II 1II
FIGURE 13.
 Kepler Motion Viewed from the Empty Focus 161

TABLE 4

r 8-
M I II I-II zrr I-III I II I-II 1lI I-1Il
o· 1.0'733 /.0133 0 1.0"03 0 o· o· o· 0° o·
IS 1.0' 06 /.0'07 -0.0001 1.0'101 -0.• 0 0 ' I~S!" 12$.· o.orP 12.. 'It 0.0'1°
Jo 1.08:1.7 1.08:J..'l - 0.00 0 I 1.0'1,37 -0.00 10 2S,IS 25'.0" o.o~ 25';0'- 0 .0'
'IS' 1.0''''1 1.070/ -0 . • 0 02 1.07:2.0 -0.002.1 38. 03 .37.'l~ 0 .11 37.8'( 0 ·17
60 l.oS;).7 I.0 S.2.'} -0.0002 I.0 S'S8 -0.0031 SI .2"1 5'1.17 0 .12, S'/'2.D o.o"}
7~ 1.031"1 1.03.2.1 -o.ooo:t 1.035''} -0.00'10 65: 03 ''!·''IY 0 .0") 'If.''I8 O.oS
'0 I.oon l.oog7 o. l .o/~o -0.00'0 7'}.37 7'-3'1 0.03 7~.38 -0.01
loS q'1glf2 O.C/BY;.. 0 0 .~g8Y -O.oo'f:l. (N./fl 'jl{.'i~- -0.01( "jLf.'f} -0.08
1:1.0 0.'1603 0.'''02- 0.ooo t 0.'}~37 -0.003/f 1I0.2~ /10,31 -0.0'1 I/O.3S' -0.13
lils 0.'}3Sg 0·'31>7 0.0001 0.''JII1f -0 .0023 1;u,.7"1 121.."2. -0.13 126.''1 -0./5
Iso 0.'}.2./7 0.'1.216 0.0001 O·1:0.g -0.00 II NY.o? 1't'f,li - 0./2. flI'po -0./3
16S' O.'1I Ob O."IloS 0.0001 0.'11 a'} -0.0003 16 1.8? 1"/.1'- -0 .07 1'1. '16 -0.07
/So 0.1067 0."0"7 a /).'}o (, 7 0 180 180 o 180 0

from Model II. How the latter is done I leave as an exercise for
the reader. (Hint: Set FP = p and
1- e2
p- .
-1 +e cos M '
thus, two sides and an angle are known in triangle SEP, and r
and () can be found.)
In Model III I replaced the elliptical orbit by its circumscribed
circle (of radius 1). I placed the sun S at the distance e from the
center and made the planet P move uniformly about the "equant
point" Q, which lies symmetrically to S with respect to the cen-
ter, so SQ = 2e. The angle at Q is then the mean anomaly M. The
values of rand () derived from this model and corresponding to
the values of M in the leftmost column are listed in the two col-
umns III, and the differences between the correct values and
these are in the columns I - III. Once again, I leave their deri-
vation as an exercise for the reader. (Hint: Introduce the circle's
center C; then CP = 1. Triangle CQP then has three known
pieces, so the rest of them can be found. Finally, solve triangle
SCP.)
The numbers in the difference columns speak for themselves:
They are very small, indeed, even the ones in columns I - III.
Thus, Models II and III are good approximations to the Kepler-
162 4. Kepler Motion Viewed from Either Focus

FIGURE 14.

ian motion of Model I, and the excellence of Model III is perhaps

particularly surprising, for here we changed not only the man-
ner of motion, but the orbit itself. Here we must remember,
however, that an ellipse with small eccentricity deviates only
very little from its circumscribed circle. With Mars's eccentric-
ity of 0.0933 and semi-major axis 1, we find for the semiminor
aXIS

VI - e2 = 0.9956
(for e = 1/3, we get 0.943, and 0.866 for e = 1/2).
Thus, in this case at least, it seems close to the truth that the
planet moves uniformly about the empty focus , and we shall now
investigate just how close (or far) it remains so in general.
Let the ellipse in Figure 14 be the orbit of the planet P; the
sun S is in one focus, H is the empty focus, and A the aphelion.
Let the eccentricity be e and set, as above, the semimajor axis
equal to 1. We have then the polar equations of the orbit
1- e2 1 - e2
r= and p = (i)
1 - e cos e 1 + e cos ¢
where r+ p = 2, so
dr = -dp. (ii)
If u and v are the angles from the rays from Hand S, respec-
tively, to the forward half of the tangent at P (see Figure 14), we
 Kepler Motion Viewed from the Empty Focus 163

have (see page 140)

dO dif>
tan u = r dr and tan u = p dp 0
(iii)

Because the ellipse reflects a ray from one focus through the
other (see page 139), we have that u + u = 1r (we measure in
radians, since we use calculus), so
tanu- tanu (iv)
and hence

which, with (ii), yields

r dO = p dif>o * (v)
The planet obeys Kepler's law, and the area that grows uni-
formly with time is

area ASP = J.L T = J9 !r 2 ae, (vi)

2
0

where T is the time elapsed since the planet last passed the
aphelion A. We wish to establish a direct link between T and if>,
and to that end we transform the integral in (vi) using r = 2 - P
and relation (v):
1
1
<1>
area ASP = -(2 - p) p dif>o (vii)
o 2
0

We now express p in terms of if> according to (i) and obtain

area ASP = ! (1 _ e2 ) • J 2 + 2e cos if> -

<I> 1: e dif>
2

2 0 (1 + e cos if»
= ! (1 -
2
e 2) . J 1 +(1 2e+ ecoscosif>if+»
0
<I> 2
e2
dd: (viii)

*1 am indebted to Professor D. To Whiteside of Cambridge University

for having shown me this Newtonian shortcut. Newton had results
about the near-uniform growth of ¢ much like the one below. (See his
Principia, Book I, Prop. XXXI, Scholium, and D. T. Whiteside (ed.),
The Mathematical Papers of Isaac Newton, Vol. VI, Cambridge, 1974,
p.172.)
164 4. Kepler Motion Viewed from Either Focus

Finally, we introduce the mean anomaly M, recalling that

area ASP = ~ VI - e2 . M
2
(when M = 27r, area ASP = ttab = 7rVl - e 2 ) , which, with (viii),
gives

M = ~f ¢ 1 + 2e cos ¢ +2e2 d¢ . (ix)

o (1 + e cos ¢)
This integral is not at all pleasant, but if we develop the coef-
ficient and the integrand in infinite series according to increas-
ing powers of e, and then integrate term by term, we get

M =
,I..
lfI - 4:1 e2 sin
. 2
2¢ - 3e3 sm ¢ + ...,
. 3
(x)

which, when inverted in the rather brutal manner of the 17th

and 18th centuries, yields

¢ = M + ~ e2 sin 2M + ~ e3 sin 3 M + ... , (xi)

so ¢ is equal to M except for terms that contain e raised to the

second and higher powers. Thus, the planet moves practically
uniformly about the empty focus when the eccentricity is small.
In sum, if we replace
(i) a planet's elliptical orbit by its circumscribed circle, and
(ii) its Keplerian motion by uniform rotation about the empty
focus, we shall have achieved an excellent approximation to
its behavior, for in either case we have committed errors
that involve only second and higher powers of the eccen-
tricitye.
In case (i) we recall that an ellipse's minor semi-axis is

b= aVl -e a(1 _e:),

2 ;::;::

where a is its major semi-axis, and in case (ii) we invoke equation

(xi). In other words, the kind of motion that Ptolemy employed
on the deferent of his equant model is a very good approximation
to a Keplerian motion of a planet around the sun with the sun at
the observer's place 0 , the eccentricity of the circle the same as
the ellipse's, and the empty focus as the equant point Q.
 Kepler Motion Viewed from the Empty Focus 165

So we have seen that Ptolemy was on the right track when

he abandoned uniform circular motion, philosophically correct
though it then was and long remained, and replaced it with a cir-
cular motion that is uniform about a point other than the center,
but we have not yet justified his planetary models as a whole.
We saw earlier that for an inner planet, the deferent repre-
sented the sun's orbit around the earth (the equivalent of the
earth's orbit around the sun), while the epicycle corresponded to
the planet's orbit around the sun. For an outer planet it was the
other way around: The deferent was the planet's orbit around
the sun, while the epicycle was the sun's orbit around the earth.
With the advantage of hindsight we can see that Ptolemy's
planetary models should have been composed of two equant
models, approximating the model composed of two Keplerian
ellipses, for determining a planet's geocentric position shown in
Figure 12.
Since that is so, one may well ask how Ptolemy was led to
propose a deferent that acts like a very good approximation to
a Keplerian ellipse, but an epicycle with a purely concentric
motion that is nearly uniform (it is precisely uniform relative to a
point on it opposite the equant point) and not like the Keplerian
motion.
Before trying to hint at an answer, I should say that the ques-
tion is partly wrong. Indeed, Ptolemy cannot make an equant
model work for Mercury but has to introduce a quite complex
crank mechanism to carry the epicycle around. I have avoided
mentioning Ptolemy's model for Mercury so far-and I shall not
discuss it in detail now-for its complexity places it outside the
scope of this little book. The trouble is caused by Mercury's
great eccentricity (slightly over 1/5).
The case of Venus is readily disposed of. Its orbit has so small
an eccentricity (less than 1/100) that it is already very nearly
circular and its motion very nearly uniform. An equant model
with the sun's geocentric orbit as a deferent, and an epicycle
with Venus's mean distance from the Sun (in a.u.) as a radius, is,
then, a very good approximation to the behavior of Venus as
seen from the earth.
While Venus seems to be made to fit an equant model, the
case of the outer planets is not so simple. I cannot here justify
Ptolemy's models for them in detail or try to assess how accu-
rately they yield a planetary position. I shall rather suggest a
point of view from which his models look quite plausible.
166 4. Kepler Motion Viewed from Either Focus

t
A

FIGURE 15.

To that end we consider, once again, the model representing

the geocentric motion of an outer planet correctly: A deferent
that is the equivalent of the planet's orbit around the sun, and
an epicycle that corresponds to the sun's apparent orbit around
the earth. When we approximate these two components with
eccentric circles with equant points, we get an arrangement like
the one illustrated in Figure 15.
In Figure 15, 0 is the observer on the earth, C the center of
the deferent whose radius CO' = a astronomical units is the
major semi-axis of the planet's elliptical orbit of eccentricity e',
and the distance DC is e'a a.u. The point 0' moves on the defer-
ent uniformly, with angular velocity f.Ll, about Q, which lies on
 Kepler Motion Viewed from the Empty Focus 167

the extension of the line OC, with OC = CQ. However, A' is no

longer the epicycle's center C' , for we now have an eccentric
epicycle representing the sun's elliptical orbit around the earth.
The apsidalline II'A', on which we have the equidistant points
O'C'Q', is of fixed direction, so it moves parallel to itself. The
radius of the " epicycle" is 1 a.u., and we have
O'C' = C'Q' = e a.u.,
where e is the eccentricity of the sun's (earth's) elliptical orbit.
The planet P moves on the "epicycle" uniformly, relative to
C'A', around Q'. We note that LQO'A' must grow at the uni-
form rate J-ll so that the apsidal line II' A' always will have the
same direction, and the angular velocity of P relative to C'A'
is not J-l2 of Ptolemy's old model, but J-ll + J-l2.
It should further be noted that we have scaled the model
correctly, in astronomical units, so that it yields not only a fine
approximation to the direction from the observer to the planet,
but also to its distance OP.
We shall now see that by a slight rearrangement of parts in
Figure 15, the planet can be recognized as moving on an epicycle
whose center travels on an eccentric deferent. To that end we
complete the parallelogram CO' C' and call the fourth vertex C".
Two of the parallelogram's sides are of (constant) length a, the
other two are oflength e, and these last two sides, CC" and O'C',
have fixed direction. The point C" is then a fixed point, and
it can plainly be considered the center of a circle of radius
C" C' = a on which the center C' of the epicycle travels.
So an outer planet's geocentric behavior can be very well ap-
proximated by an epicyclic model with an eccentric deferent.
The eccentricity of the deferent is the vector sum of the eccen-
tricities of the earth's and the planet's orbits (the eccentricities
are here considered not as ratios, but as distances measured
in a.u.).
This is as close as we can get to a Ptolemaic model for an
outer planet with this sort of approximation. Two things are
missing for perfect agreement. First, we ought to have found
that the epicycle's center C' moves uniformly around the ideal
Ptolemaic equant point E that lies symmetrically with a with
respect to the eccentric circle's center C" (see Figure 15). In
fact, C' moves uniformly about the midpoint E' of the line join-
ing Q, the old deferent's equant point, and the ideal equant point
E, as can readily be seen. Note that
168 4. Kepler Motion Viewed from Either Focus

TABLE 5
e a e' ae'
m a.u. in a.u.
Mars 0.017 1.52 0.093 0.14
Jupiter 0.017 5.20 0.048 0.25
Saturn 0.017 9.54 0.056 0.53
e: earth's eccentricity;
e': planet's eccentricity.

QE" = E'E = e,
the earth's eccentricity.
Second, the planet does not move uniformly with angular
velocity P,l + P,2 relative to a fixed direction, but about the point
Q' that is removed from it by the distance e a.u., the earth's
eccentricity.
So we have found that a planetary model composed of two
equant models, approximating the correct Keplerian ellipses,
and Ptolemy's planetary model agree in all but one respect: The
centers of uniform motion do not coincide but are displaced-
deferent's equant from deferent's equant, and "epicycle's"
equant from epicycle's center, respectively-and by the amount
e a.u. in both cases, where e as before is the earth's eccentricity.
How much that is compared to other dimensions of correctly
scaled models is shown in Table 5.
In the Almagest Ptolemy deals with the planets in the order we
recognize from his cosmology: Mercury, Venus, Mars, Jupiter,
Saturn, and the equant model first appears in the Almagest's
Book X, where he is concerned with Venus. He first finds its
epicycle's radius to be 43;10P , if the deferent's radius is 60P , and
that the deferent's eccentricity is 1;15P = 1/48, all this by con-
sidering Venus's greatest elongation from the sun (we recall that
the direction from us to the sun is, roughly, the direction to the
epicycle's center); the argument is hinted at in Figure 15. He
then determines that the point around which the epicycle's
center moves uniformly-the equant point-is a further 1;15P
removed from us beyond the deferent's center, so our distance to
the equant point is 2;30P = 1/24. We have then an equant model
where the deferent's center bisects the line from observer to
equant point.
 Kepler Motion Viewed from the Empty Focus 169

Ptolemy now takes for granted that this sort of equant model
works for an outer planet as well, with the obvious modification
that now it is the direction from the epicycle's center to the
planet that is the same as the direction from us to the sun, and
he devotes each of the next three books of the Almagest to find-
ing the parameters of such a model that best suit one of the
outer planets.
The Almagest is surely not an autobiographical work, and
we find many instances of Ptolemy's changing the sequence of
events leading to a certain result for pedagogical reasons or
reasons of clarity. Much can be said for Mars, and not Venus,
being the planet for which he first discovered the equant model,
for (always excepting Mercury) Mars is the planet that, because
of its great eccentricity, most clearly shows deviations from
mean behavior-after all, it was by an analysis of Mars that
Kepler found his laws. Yet it is quite possible that here the
Almagest's order of presentation with Venus's as the proto-
typical model is also the historically correct order of discovery,
for , as we saw, Venus seems ideally suited to suggest a deferent
with an equant point.
Venus's deferent should be the equivalent of the sun's orbit
around the earth, so we should consider Ptolemy's solar model.
He takes it over unchanged from Hipparchus, and it is the one
example in the Almagest of the simplest efficient approximation
to Keplerian motion: the pure eccenter, a uniform circular
motion viewed from a point other than the center. It works
about as well as an equant model if its eccentricity is chosen to
be twice that of the Keplerian ellipse-this may be intuitively
clear from the above discussions-but only as far as directions
are concerned. For distances it does not work well at all: Thus,
in the apsides it would place the sun at the distances 1 ± 2e,
respectively, where e is the sun's (earth's) eccentricity, and they
ought to be 1 ± e.
Indeed, Ptolemy's (and Hipparchus's) value for the solar
eccentricity is 1/24, and for the eccentricity of Venus's deferent
he has 1/48, precisely half, as one would expect if the solar model
was to yield correct distances as it must if it is to serve as a
deferent in a model that accounts for-the variation in Venus's
greatest elongation.
What Ptolemy thought about the similarity of his solar model
to Venus's deferent is a very good question (which means that I
cannot answer it) . However, in the Almagest they are different
170 4. Kepler Motion Viewed from Either Focus

in this respect: The sun's apogee is assumed to be tropically

fixed, that is, to have constant longitude, while the deferent's is
sidereally fixed so that its longitude grows at the rate of the
precession of the equinoxes. Incidentally, the assumption that
the solar apogee has constant longitude is one of the flaws in
the Almagest that early Islamic astronomers corrected, as I
mentioned earlier.
Selected Bibliography

ACT: O. Neugebauer, Astronomical Cuneiform Texts. London 1956.

Reissued by Springer-Verlag. The standard edition of all cuneiform
texts dealing with mathematical astronomy known at the date of
publication.
HAMA: O. Neugebauer, A History of Ancient Math ematical Astron-
omy. 3 vols . Springer-Verlag, 1975. The standard referen ce work.
Asger Aaboe, Episodes from the Early History of Mathemat ics (New
Mathematical Library, Vol. 13). New York, 1963. Kept in print by
The Mathematical Association of America.
Asg er Aaboe, " Observation and Th eory in Babylonian Astronomy."
Centaurus (1980), Vol. 24, pp. 14- 35.
Bernard R. Goldstein, "The Arabic Version of Ptolemy's Planetary
Hypotheses." Trans . Am. Philos. Soc. N.S. 57,4 (1967).
Hermann Hunger and David Pingree, Astral Sciences in Mesopotamia .
Brill, Leiden, 1999. An expert and detailed survey of Mesopotamian
astrology and astronomy with a full bibliography.
Alex ander Jones, "Astronomical Papyri from Oxyrrhynchus." Mem. of
Am. Philos . Soc. (1999), Vol. 233. An epoch-making publication that
changed our ideas about Hellenistic astronomy.
E. S. Kennedy, Colleagues and Former Students, Studies in the Islamic
Exact Sciences. Beirut, 1983. The startling first papers on the
Maragha School and its followers are among the included essays.
O. Neugebauer, The Exact Sciences in Antiquity. 2nd ed. , Providence,
1957. (Dover, 1969). The classical monograph on ancient mathematics
and astronomy.
Olaf Pedersen, A Survey of the Almagest. Odense Univ. Press, 1974.
A very useful, clear, and detailed summary and analysis.
N. M. Swerdlow and O. Neugebauer, Mathematical A stronomy in
Copernicus's De revolutionibus. 2 vols. Springer-Verlag, 1984.
A detailed account of Copernicus's complicated astronomical
models.

171
172 Selected Bibliography

G. J . Toomer, Ptolemy's Almagest. Springer-Verlag, 1984. The defini-

tive English version with commentary.
Christopher Walker (ed.), Astronomy before the Telescope. British
Museum Press, London, 1996. An excellent collection of essays
written by experts for a general audience.