You are on page 1of 828

Anatoly T.


or Science?
Antiquity and The Bible were crafted in X V-XVI century.
The Old Testament refers to mediaeval events.
Apocalypse was written after I4 8 6 a .d .

D r P ro f A natoly Fom enko and team dissect A lm a g e s t o f

an cien t Ptolem y com p iled allegedly in 150 a . d . and
considered to be the co rn er ston e o f classical history.
T h eir report states: A lm a g est was com piled X V I-X V II cy
w ith astro n o m ical data o f IX -X V I cy.

A llegedly a n cie n t E g y p tia n h o ro sco p e s p ain ted in

P haraohs to m bs o f the Valley o f Kings or cut in stone
in D end era and Esna for cen tu ries considered im p en e
trab le are decoded at last! All dates con tain ed therein
turn out definitely medieval and pertain to the X I cy a . d .
the earliest.

D iscov er hig h ly in tere stin g an gles, ch u n ky facts and

u p d ates to th e b io g ra p h ie s o f th e fa m o u s m ed iev al
astron o m ers Tycho B rahe and C op ern icus.

Reading this b o o k resem bles a test flight to th e distant

past landing a con clu sio n : the past is eventually b o th
d rastically closer and d ram atically different from one
taught in school. Fasten your seatbelts, please.

T he publishers will pay a 10,000 dollars USA in cash to

the first person who will not only declare but prove co n
sistently, with adequate m ethods and in sufficient detail
on the same or better academ ic level that the New C h ron
ology theory o f Full m em ber o f the Russian Academy o f
Science D r P ro f Anatoly T. Fomenko, Head o f the C hair o f
the D ifferential G eom etry o f M SU and his team is wrong.

This is History in the Making

ISBN 2-513^21-08-2
III llll 9 0 0 0 0

elam ere ^Publishing

9 7' 8 2 9 1 3 6 2 K 3 8 4 "
Entities should not be m ultiplied beyond necessity.
Friar W illiam O ckham
(allegedly 1285- 1349 )

Ptolemy is by no means the greatest astronom er o f the antiquity,

b u t ... the most successful con m an in the history o f science.
Robert R. Newton
American astrophysicist
( 1919 - 1991 )
ii I h is t o r y : f ic t io n o r sc ie n c e ?

A. T, Fom enko
Chronology 1
Introducing the problem. A criticism o f the Scaligerian chronology.
Dating methods as offered by mathematical statistics. Eclipses and zodiacs.

A. T. Fom enko
Chronology 2
The dynastic parallelism method. Rome. Troy. Greece. The Bible. Chronological shifts.

A. T. Fomenko, T. N. Fomenko, V. V. Kalashnikov, G. V. Nosovskiy

Chronology 3
Astronomical methods as applied to chronology. Ptolemys Almagest.
Tycho Brahe. Copernicus. The Egyptian zodiacs.

A. T. Fomenko, G. V. Nosovskiy
Chronology 4
Russia. Britain. Byzantium. Rome.

A. T. Fomenko, G. V. Nosovskiy
Chronology 5
Russia = Horde. Ottomans = Atamans. Europe. China. Japan. The Etruscans. Egypt. Scandinavia.

A. T. Fomenko, G. V. Nosovskiy
Chronology 6
The Horde-Ataman Empire. The Bible. The Reformation. America. Passover and the calendar.

A. T. Fomenko, G. V. Nosovskiy
Chronology 7
A reconstruction o f global history. The Khans o f Novgorod = The Habsburgs. Miscellaneous information.
The legacy o f the Great Empire in the history and culture o f Eurasia and America.

This seven volume edition is based on a number discovered and researched, as well as the new
o f our books that came out over the last couple of hypothetical reconstruction o f global history up
years and were concerned with the subject in ques until the XVIII century. Our previous books on the
tion. All this gigantic body o f material was revised subject o f chronology were created in the period of
and categorized; finally, its current form does not naissance and rather turbulent infancy o f the new
contain any o f the repetitions that are inevitable in paradigm, full o f complications and involved is
the publication o f separate books. All o f this re sues, which often resulted in the formulation of
sulted in the inclusion o f a great number o f addi multi-optional hypotheses. The present edition pi
tional material in the current edition - including oneers in formulating a consecutive unified con
previously unpublished data. The reader shall find cept o f the reconstruction o f ancient history - one
a systematic rendition o f detailed criticisms o f the that apparently is supported by a truly immense
consensual (Scaligerian) chronology, the descrip body o f evidence. Nevertheless, it is understandable
tions o f the methods offered by mathematical sta that its elements may occasionally be in need o f re
tistics and natural sciences that the authors have vision or elaboration.
Anatoly T. Fomenko, Tatiana N. Fomenko,
Vladimir V. Kalashnikov, Gleb V. Nosovskiy

or Science?

elamere Publishing

History: Fiction or Science?
Fomenko, Anatoly Timofeevich (b. 1945). Full Mem member o f computational mathematics and Cyber
ber (Academician) of the Russian Academy of Sci netics o f MSU. Worked as researcher in University of
ences, Full Member of the Russian Academy of Natu British Columbia, Canada; held courses o f topology
ral Sciences, Full Member o f the International Higher and applications in 1991.
Education Academy o f Sciences, Doctor of Physics
and Mathematics, Professor, Head of the Moscow Kalashnikov, Vladimir Viatcheslavovich (1942-2001).
State University Section of Mathematics of the De Doctor o f Physics and Mathematics, professor. Spe
partment o f Mathematics and Mechanics. Solved cialist in applied statistics theory. V. V. Kalashnikov
Plateaus Problem from the theory o f minimal spec was an authority on mass queuing models, reliability,
tral surfaces. Author o f the theory o f invariants and risk, imitation modelling o f stochastic systems. In
topological classification o f integrable Hamiltonian particular he developed criteria and estimates o f sta
dynamic systems. Laureate o f the 1996 National Pre bility, perturbation of various probabilistic processes.
mium o f the Russian Federation (in Mathematics) for Author of 160 scientific publications, including
a cycle o f works on the Hamiltonian dynamical sys 10 monographs. V. V. Kalashnikov was distinguished
tems and manifolds invariants theory. Author o f 200 with the USSR State Reward in 1986 for his works on
scientific publications, 28 monographs and textbooks modelling and simulation.
on mathematics, a specialist in geometry and topol
ogy, calculus of variations, symplectic topology, Ham Nosovskiy, Gleb Vladimirovich (b. 1958). Candidate
iltonian geometry and mechanics, computer geome of Physics and Mathematics (MSU, Moscow, 1988),
try. Author o f a number o f books on the development specialist in theory of probability, mathematical sta
of new empirico-statistical methods and their appli tistics, theory of probabilistic processes, theory o f op
cation to the analysis of historical chronicles as well as timization, stochastic differential equations, computer
the chronology of antiquity and the Middle Ages. modelling of stochastic processes, computer simula
tion. Worked as researcher of computer geometry in
Fom enko, Tatiana Nikolaevna (b. 1948). Mathe Moscow Space Research Institute, in Moscow Machine
matician, Candidate o f Physics and Mathematics, Tools and Instruments Institute, in Aizu University in
author o f books and scientific articles on algebraic Japan. Faculty member of the Department o f Mathe
topology and geometry, theory of algorithms, faculty matics and Mechanics MSU.

K indly order H istory: Fiction o f Science? V o lu m e 1 ( isb n 2-913621-07-4)

and Volum e 2 ( isb n 2-913621-06-6) with Am or

Published by Delamere Resources LLC

Professional Arts Building, Suite 1 ,2 0 6 11th Avenue S.E., Olympia WA 98501

ISBN 2-913621-08-2 | EAN 9782913621084

Anatoly T. Fomenko asserts the moral right to be identified as the author of this work.
Translated from Russian by Michael Jagger. Cover by Polina Zinoviev. Design & layout by Paul Bondarovski.
Project management by Franck Tamdhu.

Copyright 2007 Delamere Resources LLC.

All rights reserved. No part of this book may be reproduced, stored in a retrieval system, or transmitted, in any form
or by any means, without the prior permission of the publisher. Critics are welcome, of course,
to quote brief passages by way of criticism and review.

Also by Anatoly T. Fom enko

(List is non-exhaustive)

Differential Geometry and Topology

Plenum Publishing Corporation. 1987. USA, Consultants Bureau, New York and London.
Variational Principles in Topology. Multidimensional Minimal SurfaceTheory
Kluwer Academic Publishers, The Netherlands, 1990.
Topological variational problems. - Gordon and Breach, 1991.
Integrability and Nonintegrability in Geometry and Mechanics
Kluwer Academic Publishers, The Netherlands, 1988.
The Plateau Problem, vols.l, 2
Gordon and Breach, 1990. (Studies in the Development o f Modern Mathematics.)
Symplectic Geometry. Methods and Applications.
Gordon and Breach, 1988. Second edition 1995.
Minimal surfaces and Plateau problem. Together with Dao Chong Thi
USA, American Mathematical Society, 1991.
Integrable Systems on Lie Algebras and Symmetric Spaces. Together with V. V. Trofimov
Gordon and Breach, 1987.
Geometry o f Minimal Surfaces in Three-Dimensional Space. Together with A. A.Tuzhilin
USA, American Mathematical Society. In: Translation o f Mathematical Monographs, v o l.9 3 ,1991.
Topological Classification o f Integrable Systems. Advances in Soviet Mathematics, vol. 6
USA, American Mathematical Society, 1991.
Tensor and Vector Analysis: Geometry, Mechanics and Physics. - Taylor and Francis, 1988.
Algorithmic and Computer Methods fo r Three-Manifolds. Together with S.V. Matveev
Kluwer Academic Publishers, The Netherlands, 1997.
Topological Modeling fo r Visualization. Together with T. L. Kunii. - Springer-Verlag, 1997.
M odem Geometry. Methods and Applications. Together with B. A. Dubrovin, S. P. Novikov
Springer-Verlag, GTM 93, Part 1,1984; GTM 104, Part 2,1985. Part 3 ,1 9 9 0 , GTM 124.
The basic elements o f differential geometry and topology. Together with S. P. Novikov
Kluwer Acad. Publishers, The Netherlands, 1990.
Integrable Hamiltonian Systems: Geometry, Topology, Classification. Together with A. V. Bolsinov
Taylor and Francis, 2003.
Empirico-Statistical Analysis o f Narrative Material and its Applications to Historical Dating.
V ol.l: The Development o f the Statistical Tools. Vol.2: The Analysis o f Ancient and Medieval
Records. - Kluwer Academic Publishers. The Netherlands, 1994.
Geometrical and Statistical Methods o f Analysis o f Star Configurations. Dating Ptolemy's
Almagest. Together with V. V Kalashnikov., G. V. Nosovsky. - CRC-Press, USA, 1993.
New Methods o f Statistical Analysis o f Historical Texts. Applications to Chronology. Antiquity in
the Middle Ages. Greek and Bible History. Vols.l, 2, 3, - The Edwin Mellen Press. USA. Lewiston.
Queenston. Lampeter, 1999.
Mathematical Impressions. - American Mathematical Society, USA, 1990.

Overview o f the seven volumes............................................................................................................... ii

About the Authors..................................................................................................................................... iv
A bo by Analoly T, Fomenko.................................................................................................................... v
From the Publishers............................................................................................................................... xviii
Foreword by A. T. Fomenko................................................................................................................... 3


by A. T. Fomenko, V. V. Kalashnikov, G. V. Nosovskiy

1. A brief description of the Almagest................................................................................................ 7
2. A brief history of the Almagest....................................................................................................... 12
3. The principal star catalogues of the Middle Ages........................................................................ 13
4. The reason why the dating of the old star cataloguesis an important issu e........................... 14
5. Hipparchus.......................................................................................................................................... 15
6 . Ptolemy................................................................................................................................................ 15
7. Copernicus.......................................................................................................................................... 16
8. Tycho Brahe......................................................................................................................................... 23
9. Important research of the Almagest by the astronomer Robert Newton and
his book entitled The Crime of Claudius Ptolemy .................................................................. 28

Chapter 1 Some necessary information related to astronomy and history of astronomy

1. The ecliptic. The equator. Precession............................................................................................. 32
2. Equatorial and ecliptic coordinates................................................................................................ 34
3. The Methods of measuring equatorial and ecliptic coordinates.............................................. 35
4. The modern celestial sphere............................................................................................................. 37
5. Reverse calculation of objects positions on the celestial sphere.
The formulae of Newcomb-Kinoshita........................................................................................... 39
5.1. Necessary formulae................................................................................................................. 39
5.2. The algorithm for calculating past positions of stars....................................................... 40
6 . Astrometry. Ancient astronomical measurement instruments of the
XV-XVTI century................................................................................................................................ 42
7. Timekeeping and timekeeping devices in mediaevalastronomical observations................... 49

Chapter 2 A preliminary analysis of the Almagest star catalogue

1. The catalogue structure..................................................................................................................... 54
2. The analysis of the distribution of reliably and poorly identifiable stars
in the Almagest.................................................................................................................................... 56
3. Seven regions of the Almagest star atlas significantly differ from each
other by the number of reliably identifiable stars........................................................................ 68
4. Possible distortion of the star coordinates resulting from the atmospheric
refraction.............................................................................................................................................. 74
5. The analysis of the informata distribution across the Almagest catalogue............................. 75
6 . The analysis of the coordinate versions as specified in different manuscripts
of the Almagest catalogue. Comparison of the 26 primary manuscripts to
the canonical version of the catalogue........................................................................................... 77
7. Version density as the density of independent star observations.
Seven areas of the Almagest star atlas revisited with a new concurrence
with the previous results................................................................................................................... 80
8. In re the reliability of latitudinal and longitudinal measurements contained
in the Almagest.................................................................................................................................... 84
8.1. According to Robert Newton, the longitudes in the Almagest were
re-calculated by somebody; however, this suspicion does not arise
insofar as their latitudes are concerned.............................................................................. 84
8.2. Examples proving that the dating of the star catalogue by longitudinal
precession often leads to great errors. Mediaeval catalogues are subject
to becoming erroneously dated to an antediluvian epoch............................................. 86
9. The dubious nature of the traditional opinion that Ptolemys text implies
actual observations on his part, as well as his personal participation in the
stellar measurements and observations described in the Almagest.......................................... 89
10. What ecliptic point did Ptolemy use for longitudinal reference?........................................... 91
11. Peters sinusoid in Almagest latitudes.......................................................................................... 94

Chapter 3 Unsuccessful attempts of dating the Almagest. Reasons for failure.

Our new approach and a brief account of our results
1. The attempt to date the Almagest by a comparison to the calculated catalogues
reflecting the motion of the fastest stars........................................................................................... 95
1.1. The comparison of the Almagest catalogue to the calculated catalogues..................... 95
1.2. The attempt of dating the Almagest catalogue by proper movements of
individual stars........................................................................................................................ 96
1.3. Why the dating of the Almagest by individual star movements gives us
no reliable result...................................................................................................................... 99
2. An attempt of dating the Almagest catalogue by the aggregate of fast and
named stars as compared to the calculated catalogues............................................................... 101
2.1. The criteria one is to adhere to in ones choice of the stars for the purpose
of dating...................................................................................................................................... 101
2.2. The proximity interval system as applied to certain fast or named stars.................. 102
2.3. Dating the Almagest with the suggested method utilizing arc distances
of individual stars isan impossibility..................................................................................... 104
2.4. Dating the Almagest catalogue with the suggested method based on
latitudinal discrepancies of individual stars also proves impossible............................. 104
viii h is t o r y : fic t io n or sc ien c e? CHRON 3

3. The attempt to date the Almagest catalogue by the motion of individual

stars as compared to the objects in their immediate vicinity.................................................... 106
3.1. The varying geometry of stellar configurations as seen against the
background of immobile stars ........................................................................................... 106
3.2. The stars chosen for the experiment.................................................................................. 106
3.3. The behaviour of the individual discrepancies and the average
discrepancy.............................................................................................................................. 106
3.4. Negative experiment result................................................................................................... 107
4. The analysis of several erroneous works on the subject of dating the Almagest
by proper star m otions...................................................................................................................... 107
4.1. A lot of the errors are not produced by astronomical phenomena
and stem from the incorrect application of the methods offered by
mathematical statistics........................................................................................................... 107
4.2. The data in Y. N. Yefremovs works on the dating of the Almagest
were tailored to fit the desired result.................................................................................. 109
4.3. A vicious circle in the dating of the Almagest by the movement
of the star o 2 E ri...................................................................................................................... 109
4.4. Y. N. Yefremovs errors in the precision estimation of dating the Almagest
byArcturus............................................................................................................................. 110
4.5. Erroneous precision estimation of astronomical calculations:
another example..................................................................................................................... 113
4.6. The secondary analysis of the Almagest dating in the Samoobrazovaniye
(Autodidactics) magazine................................................................................................. 114
5. Conclusions and directions for further research. Our approach and a brief synopsis
of our main results............................................................................................................................. 115
5.1. The three problems one is confronted with: identifying the Almagest
stars, defining the nature of possible errors, and analysing the precision
of the catalogue....................................................................................................................... 115
5.2. The identification of the Almagest stars............................................................................ 116
5.3. Various types of errors in the catalogue............................................................................. 116
5.4 . The discovery of the systematic error in the Almagest catalogue.
Its compensation confirms the correctness of the declared catalogue precision........ 118
5.5. The compensation of the systematic error discovered in the catalogue gives
us an opportunity of dating the latter................................................................................ 120
5.6. The dating of the Almagest catalogue by the motion of its eight primary basis
stars after the rectification of the statistically discovered catalogue error.................... 120
5.7. The dating of the Almagest catalogue by the motions of its eight named basis
stars by an independent geometrical m ethod.................................................................... 121

Chapter 4 Who is who?

1. Preliminary observations.................................................................................................................. 123
2. Formal search of the fastest stars in the Almagest catalogue..................................................... 124
2.1. The star identification m ethod............................................................................................. 124
2.2. The result of identifying the modern stars as their counterparts from
the Almagest catalogue.......................................................................................................... 127
2.3. Corollaries................................................................................................................................ 129
3. The search of all the fast stars reliably identifiable in the Almagest catalogue...................... 130

Chapter 5 The analysis of the star catalogues' systematic errors

0. Basic conception................................................................................................................................. 133
o.i. A demonstrative analogy....................................................................................................... 133
o.2. The implementation of the method................................................................................... 134
0.3. The value of the systematic error cannot be used for the dating of the catalogue..... 135
1. Main definition................................................................................................................................... 136
2 . The parameterisation of group errors and systematic erro rs.................................................... 136
3. Calculating parameters y(f) and (p(f) with the method of minimal squares.......................... 139
4. variation of the parameters ystat(t) and VstaM over the course of tim e ................................ 141
5. The statistical properties of the estimates of ystat and (pstat....................................................... 144
6 . Corollaries............................................................................................................................................ 148

Chapter 6 Statistical and precision-related properties of the Almagest catalogue

1. Introductory remarks........................................................................................................................ 149
2 . Seven regions of the celestial sphere............................................................................................... 149
2.1. A characteristic of the seven areas that we have discovered in the
Almagest atlas.......................................................................................................................... 149
2.2. The disposition of the ecliptic poles for each of the seven regions
of the Almagest star atlas....................................................................................................... 150
2.3. The calculation of confidence intervals.............................................................................. 152
3. Our analysis of individual Almagest constellations..................................................................... 156
3.1. The compiler of the Almagest may have made a different error in case
of every minor constellation group..................................................................................... 156
3.2. The calculation of systematic errors for individual groups of constellations
in the Almagest........................................................................................................................ 156
3.3. Group errors for individual constellations from the well measured
celestial region of the Almagest are virtually identical to the systematic
error discovered as a characteristic of this area in general.............................................. 159
3.4. How the compensation of the systematic error that we have discovered
affects the precision characteristics of the environs of named stars............................. 160
3.5. The discovery of a single systematic error made by the compiler of
the Almagest for the region of ZodA and the majority of named stars........................ 161
4. Corollaries............................................................................................................................................ 162

Chapter 7 The dating of the Almagest star catalogue. Statistical and geometrical methods
1 . The catalogues informative kernel consists of the well-measured named stars................... 164
2. Preliminary considerations in re the dating of the Almagest catalogue by
the variations in the coordinates of named stars......................................................................... 165
3. The statistical dating procedure...................................................................................................... 171
3.1. The description of the dating procedure............................................................................ 171
3.2. The dependency of the minimax discrepancy A on f, y and (p for the Almagest........ 172
3.3. Results of dating the Almagest catalogue statistically...................................................... 172
3.4. The discussion of the result.................................................................................................. 174
4. Dating the Almagest catalogue by the expanded informative kernel....................................... 176
5. Dating the Almagest catalogue by a variety of 8-star configurations consisting
of bright stars...................................................................................................................................... 182
6 . The statistical procedure of dating the Almagest catalogue: stability analysis....................... 183
x h ist o r y : fic t io n or sc ien c e? CHRON 3

6.1. The necessity of using variable algorithm values.............................................................. 183

6 .2. Trust level variation............................................................................................................... 183
6.3. Reducing the contingent of the Almagest catalogue informative kernel..................... 184
6.4 . The exclusion of Arcturus does not affect the dating of the Almagest
catalogue substantially........................................................................................................... 185
7. The geometrical dating of the Almagest........................................................................................ 186
8. The stability of the geometrical dating method applied to the Almagest catalogue.
The influence of various astronomical instrument errors on the dating result..................... 188
8.1. Poorly-manufactured astronomical instruments may have impaired the
measurement precision.......................................................................................................... 188
8.2. Formulating the problem mathematically......................................................................... 189
8.3. The deformation of a sphere into an ellipsoid.................................................................. 190
8.4 . Measurement discrepancies in the ellipsoidal coordinate system .............................. 191
8.5. Estimating the distortion of angles measured by the marginally
ellipsoidal instrument .......................................................................................................... 191
8.6. Possible distortion estimation and the stability of the resultant dating....................... 192
8.7. Numerical value table for possible ellipsoidal distortions ............................................ 194
8.8. Conclusions............................................................................................................................. 194
9. Longitudinal behaviour of the named Almagest stars................................................................ 194
10. The behaviour of arc discrepancies in the configuration comprised of the
Almagest informative kernel.......................................................................................................... 196
11. Conclusions....................................................................................................................................... 198

Chapter 8 Tilt angle between the ecliptic and the equator in the Almagest
1. Ptolemys concept of the ecliptic tilt angle value and systematic error y ................................ 199
2. The Peters Zodiac and the sine curve of Peters............................................................................ 201

Chapter 9 The application of our method to the dating of other mediaeval catalogues
1. Introduction........................................................................................................................................ 209
2. Tycho Brahes catalogue.................................................................................................................... 209
2.1. A general characteristic of Tycho Brahes catalogue and the result
of our dating............................................................................................................................ 209
2.2. The analysis of Tycho Brahes latitudinal errors and the removal
of the rejects ............................................................. 210
2.3. The choice of the informative kernel for Tycho Brahes catalogue............................... 213
2.4 . The dating of Tycho Brahes observations......................................................................... 213
2.5. Conclusions.............................................................................................................................. 214
3. Ulugbeks catalogue............................................................................................................................ 217
3.1. A general characteristic of Ulugbeks catalogue and its dating result........................... 217
3.2. Systematic errors in Ulugbeks catalogue........................................................................... 218
3.3. The choice of the informative kernel and the A threshold. The dating
of Ulugbeks catalogue............................................................................................................ 218
3.4 . Conclusions............................................................................................................................. 219
4. The catalogue of Hevelius................................................................................................................. 220
4 .1. The dependency between the catalogues of Tycho Brahe and Hevelius....................... 220
4 .2. Conclusions............................................................................................................................. 221
5. The catalogue of A l-Sufi................................................................................................................... 221

Chapter 10 Additional considerations concerning the dating of the Almagest.

Stellar coverings and lunar eclipses
by A. T. Fomenko, G. V. Nosovskiy
1. Introduction........................................................................................................................................ 224
2 . Dating the planetary coverings of the stars. Calculations that involve
average elements................................................................................................................................. 225
3. The dating of the planetary star coverings described in the Almagest.
A more precise calculation............................................................................................................... 230
3.1. The adjusted algorithm.......................................................................................................... 230
3.2. The discussion of the mediaeval X-XI century solution................................................. 231
3.2.1. The T| of Virgo covered by Venus in 960 a.d .......................................................... 233
3.2.2. Mars covering the p of Scorpio in 959 a.d ............................................................. 235
3.2.3. Jupiter covering the 8 of Cancer in 994 a.d ............................................................ 236
3.2.4. Saturn approaching the y of Virgo in 1009 a.d ..................................................... 237
3.2.5. The chronology of the Almagest according to the X-XI century solution........ 238
3.3. Discussing the late mediaeval solution of the XV-XVT century..................................... 243
3.3.1. The T| of Virgo covered by Venus in 1496 a.d ........................................................ 243
3.3.2. Mars covering the p of Scorpio in 1497 a.d ........................................................... 244
3.3.3. Jupiter covering the 8 of Cancer in 1528 a.d .......................................................... 245
3.3.4. Saturn approaching the y of Virgo in 1539 a.d ..................................................... 245
3.3.5. Commentary to the late mediaeval solution......................................................... 245
4. The era of Nabonassar in accordance with the late mediaeval solution.................................. 246
5. The dating of the Almagests creation and how this book assumed its
present form. Ptolemy and Copernicus......................................................................................... 248
6 . The ancient Hipparchus as the apparent phantom reflection of Tycho Brahe,
the famous astronomer...................................................................................................................... 251
7. Ptolemys Almagest is most likely to have undergone its final edition already
after the death of Tycho Brahe, or the ancient Hipparchus.................................................... 254
8. According to Robert Newton, most of the lunar eclipses referred to in the
Almagest happen to be relatively recent forgeries........................................................................ 257

Chapter 11 Other problems and hypotheses arising from the dating of the Almagest catalogue
by A. T. Fomenkoy G. V. Nosovskiy
1. Certain auxiliary oddities of the Almagest.................................................................................... 259
1.1. What coordinates was the Almagest catalogue compiled in initially?........................... 259
1.2. The North Star as the first star of the Almagest catalogue.............................................. 260
1.3. Oddities inherent in the Latin (allegedly 1537) and Greek (allegedly 1538)
editions of the Almagest........................................................................................................ 262
1.4. The star charts of the Almagest............................................................................................ 266
2. The Almagest and Halleys discovery of proper star m otions................................................... 273
3. The identity of the ancient Emperor Pius, in whose reign many of Ptolemys
astronomical observations were performed. His geographical and chronological
localisation........................................................................................................................................... 275
4. Scaligerian datings of the manuscripts and the printed editions of the Almagest............... 276
4.1. Greek manuscripts of the Almagest..................................................................................... 278
4.2. Latin manuscripts of the Almagest..................................................................................... 279
4.3. Arabic manuscripts of the Almagest................................................................................... 279
h is t o r y : fic t io n or sc ien c e? CHRON 3

4.4. The first printed editions of the Almagest........................................................................ 280

4 .5. Questions concerning the Scaligerian datings of the Almagest manuscripts.............. 281
5. So what is the Almagest, anyway?................................................................................................... 283
6 . Oddities in the development of the astronomical science as portrayed in the
Scaligerian textbook ....................................................................................................................... 283
6.1. The efflorescence of the so-called ancient astronomy .................................................. 283
6.2. The beginning of the mysterious decline of the ancient astronomy in
Scaligerian history.................................................................................................................. 287
6 .3- The alleged millenarian return to infancy and the primitive character
of mediaeval astronomy........................................................................................................ 287
6.4. The astronomical boom of the Renaissance: original, not repetition........................... 289
6.4.1. The astronomical renaissance of the Arabs........................................................ 289
6.4 .2. The astronomical renaissance in Europe............................................................ 291
6.4.3. The boom of European astronomy in the XV-XVI century.............................. 293
6.5. Bottom-line chronological diagram which demonstrates oddities
inherent in the development of the astronomical science in the consensual
chronological paradigm of Scaliger and Petavius............................................................ 293
6 .6 . Corollaries............................................................................................................................... 302
7. Copernicus, Tycho Brahe and Kepler. The relation between Johannes Kepler
and the final version of the Copernican oeuvre........................................................................... 302
7.1. What we know about Copernicus and his astronomical endeavours.
Was the heliocentric cosmological system indeed discovered in the first
half of the XVI century and not any later?........................................................................ 302
7.2. Oddities in the Scaligerian story of how the book of Copernicus was
published................................................................................................................................. 306
7.3. Why it is believed that Tycho Brahe did not accept the theory of
Copernicus. In reality, the system invented by Tycho Brahe is identical
to the Copernican.................................................................................................................. 309
7.4. Is it true that the book of Copernicus, first published in the alleged
year 1543, has reached us in its initial shape and form ?.................................................. 313
7.5. Could Johannes Kepler be the editor or even co-author of the canonical
version of the Copernican oeuvre known to us today?.................................................. 315
7.6. The heliocentric cosmology and the Biblical stopped sun .......................................... 318
8. Anna Comnena considers Ptolemy her contemporary. In other words, Ptolemy
couldnt have lived earlier than the XII century a.d..................................................................... 319
9. Obvious dating o f the Ptolemaic epoch on Ptolemys portrait in the old German
Global Chronicles by Hartmann Schedel................................................................................... 320
10. The meaning of the word Pelusiensis (or Pheludiensis) in the full name
of Ptolemy.......................................................................................................................................... 321


by A. T. Fomenko, T. N. Fomenko, G. V. Nosovskiy

A foreword to Part 2 ................................................................................................................................ 327

From the preface to The New Chronology of Egypt. The Astronomical Datings
of Ancient Egyptian Monuments. Research of 2000-2002 by A. T. Fomenko and
G. V. Nosovskiy (Moscow, Veche, 20 0 2 )................................................................................................ 329
CO N TEN TS | x iii

Chapter 12 The Egyptian zodiacs

1. The Egyptian zodiacs and the likelihood of their reliable astronomical dating................. 334
2. The astronomical dating of Egyptian zodiacs and related difficulties. The reasons
why the Egyptologists eschew the astronomical dating of the zodiacs.................................... 350
3. Our new approach to the dating of Egyptian zodiacs................................................................ 354
4. The funereal character of zodiacs in Egypt.................................................................................. 355
5. Representations of the Egyptian zodiacs as used by the authors.............................................. 357
6 . Stylistic changes in the zodiacs from the Napoleonic Egyptian album ................................... 359
7. The names we use for the Egyptian zodiacs................................................................................. 364

Chapter 13 Former astronomical datings of the Egyptian zodiacs

1. The Round and the Long Zodiac of Dendera............................................................................... 366
2. The two zodiacs from E sn a .............................................................................................................. 372
3. Flinders Petries Athribis zodiacs..................................................................................................... 375
4. The Theban Zodiac of Brugsch....................................................................................................... 380
5. Astronomical dating in the works of the Egyptologists.............................................................. 388

Chapter 14 A new approach to the deciphermentof the Egyptian zodiacs

1. The shortcomings of the earlier decipherments of the Egyptian zodiacs............................... 390
2. A new approach to the interpretation of Egyptian zodiacs. Primary and
secondary horoscopes....................................................................................................................... 399
3. An Egyptian zodiac as a description of the entire calendar year that contains
the main horoscopes d ate................................................................................................................ 410
4. Unlike previous researchers, who stopped at a single interpretation version
they deemed best, we consider every possible decipherment option for the
Egyptian zodiacs................................................................................................................................. 410

Chapter 15 The symbolism of the Egyptian zodiacs. A new and more complete interpretation
1. Constellation symbols....................................................................................................................... 412
1.1. A ries........................................................................................................................................... 412
1.2. Taurus........................................................................................................................................ 413
1.3. G em ini...................................................................................................................................... 413
1.4. Cancer....................................................................................................................................... 416
1.5. L e o ............................................................................................................................................. 418
1.6 . Virgo.......................................................................................................................................... 421
1.7. Libra........................................................................................................................................... 422
1.8. Scorpio...................................................................................................................................... 423
1.9 . Sagittarius................................................................................................................................ 423
1.10. Capricorn................................................................................................................................ 425
1.11. Aquarius.................................................................................................................................. 426
1.12. Pisces........................................................................................................................................ 429
2. The ten-degree symbols and the resolution of the Egyptian zodiacs.................................... 431
2.1. The ten-degree marks in the Long Zodiac (D L )............................................................... 431
2.2. The division of the ecliptic into 36 parts and the exactness of
planetary representations in Egyptian zodiacs.................................................................. 434
3. Distinguishing between the male and the female figures in the Egyptian zodiacs................ 435
4. Planetary symbols of the primary horoscope............................................................................... 436
xiv | h is t o r y : fic t io n or sc ien c e? chron 3

4.1. The planetary r o d ................................................................................................................... 436

4.2. Saturn in the primary horoscope........................................................................................ 440
4.3. Seth, Anubis and Thoth as the symbols of Saturn and Mercury................................... 444
4.4. Confusion between Saturn and Mercury in astral symbolism...................................... 446
4.5. Our hypothesis in re the genesis of the old cult of Saturn.............................................. 447
4.6. Jupiter in the main horoscope............................................................................................. 448
4.7. Mars in the primary horoscope........................................................................................... 450
4.8. Venus in the primary horoscope......................................................................................... 452
4.9 . Mercury in the primary horoscope..................................................................................... 457
4.10. The attributes of Mercury in the Egyptian zodiacs........................................................ 458
4.11. Mercury drawn in two positions simultaneously............................................................ 459
4.12. Mercury as the symbol of the two-faced god Janus (Ivan)........................................ 460
4.13. The Sun in the primary horoscope.................................................................................... 462
4.14. The astronomical symbolism of the Egyptian eye symbol........................................ 464
4 .15. The Moon in the primary horoscope................................................................................ 466
5. Planetary symbols in secondary horoscopes................................................................................ 468
5.1. The first example: the planets from the secondary horoscope of
autumn equinox in zodiac D L .............................................................................................. 469
5.2. The second example: planets from the secondary horoscope of
winter solstice in the DR zodiac.......................................................................................... 471
5.3. Third example: planets from the secondary horoscope of summer
solstice in the AN zodiac....................................................................................................... 473
6 . Boats, snakes and other transposition symbols underneath the figures.................................. 474
7. Visibility indicators of the primary horoscopes planets............................................................ 476
8. Equinox and solstice symbols.......................................................................................................... 478
8.1. Autumn equinox symbols in V irgo..................................................................................... 478
8.2. Symbols of the winter solstice point in Sagittarius. The astronomical
hieroglyph of Sagittarius with a minimal horoscope..................................................... 480
8.3. Symbolism of the spring equinox point in Pisces............................................................ 482
8.4 . Symbols of the summer solstice point in Gemini. The astronomical
hieroglyph of Gemini with a minimal horoscope.......................................................... 484
9. Auxiliary astronomical symbols in Egyptian zodiacs.................................................................. 486
9.1. The Easter Full M o o n ............................................................................................................ 488
9.2. The solar bird in the Long Zodiac of Dendera (D L )....................................................... 492
9.3. The symbol of dusk and dawn............................................................................................. 493
9.4 . The decapitation scene next to Aquarius........................................................................... 493
9.5. The stabbed calf (bullfighting scene).................................................................................. 494
9.6. Wolf on a scythe in the zodiacs of Dendera...................................................................... 495
9.7. The conjunction of Mars and Saturn in the Long Zodiac of Dendera......................... 495
10. Legitimate and illegitimate zodiac decipherments.................................................................... 495
11. Observation point: Cairo or Luxor (Ibrim )?.............................................................................. 498
12. The beginning of a year in Egyptian zodiacs.............................................................................. 499

Chapter 16 Astronomical estimation of the dates ciphered in the Egyptian zodiacs: a methodology
1. Seven planets o f the antiquity. Zodiacs and horoscopes............................................................ 501
2. The possible presence of calculated horoscopes in Egyptian zodiacs...................................... 503
3. The motion o f planets along the Zodiac....................................................................................... 504

4. Dividing the zodiacal belt into constellations.............................................................................. 505

5. Astral calendar. How often do individual horoscopes recur?.................................................. 508
6 . The calculation of past planetary positions. The Horos software. Modern
planetary theory precision suffices for the dating of the Egyptian zodiacs............................ 510
7. The dating of an Egyptian zodiac with the aid of its primary and secondary
horoscopes regarded as a whole....................................................................................................... 511
7.1. First step. Defining the primary horoscopes planets. All possible options
are considered................................................................................................................................ 511
7.2. Second step. Calculating the dates for every interpretation option of the
primary horoscope................................................................................................................. 511
7.3. Third step. Solutions are tested to comply with such criteria as planetary
disposition, visibility indicators and secondary horoscopes. Rejection of
incomplete solutions............................................................................................................... 512
8. The coloured Egyptian zodiac...................................................................................................... 514
9. Unambiguous reconstruction of the dates transcribed in the Egyptian zodiacs.
Final (exhaustive) solutions............................................................................................................. 515
10. The constellation scale of a zodiac............................................................................................. 515
11. Points o f approximate planetary disposition in an Egyptian zodiac (best points)
and accounting for planetary order.............................................................................................. 516
12. Average distance between best points as the approximate quality criterion of
an astronomical solution................................................................................................................ 517
13. An example of the input data format used by the Horos program........................................ 518
14. Verification table for the astronomical solution........................................................................ 518

Chapter 17 Dates ciphered in the monumental temple zodiacs of Dendera and Esna
1. The zodiacs from Dendera and Esna as part of the grandiose royal necropolis
in the Royal Bight of the N ile....................................................................................................... 520
2. The Zodiacs of Dendera: history of discovery and research...................................................... 524
3. Deciphering the date of the Long Zodiac of Dendera (D L )...................................................... 525
3.1. The Long Zodiac of Dendera and the various representations thereof........................ 525
3.2. The Long Zodiac of Dendera in colour.............................................................................. 526
3.3. Constellation figures in the DL zodiac................................................................................ 526
3.4 . Planetary figures of the primary horoscope from the DL zodiac.................................. 526
3.5. Secondary horoscopes in the DL zodiac............................................................................. 531
3.5.2. Autumn equinox horoscope in the DL zodiac...................................................... 531
3.5.2. Winter solstice horoscope in the DL zodiac.......................................................... 533
3.5.3. Horoscope of spring equinox in the DL zodiac.................................................... 534
3.5.4 . The summer solstice horoscope in the DL zodiac................................................ 534
3.6. Validation and rejection of preliminary solutions............................................................ 536
3.7. The exhaustive solution of the Long Zodiac: 22-26 April 1168 A.d ................................ 537
3.8. Verification table for the exhaustive solution of the Long Zodiac................................. 538
4. The decipherment of the date from the Round Zodiac of Dendera (DR) ............................. 545
4 .1. Copies of the Round Zodiac from Dendera....................................................................... 546
4. 2. The coloured version of the Round Zodiac...................................................................... 547
4.3. Constellation figures in the DR zodiac............................................................................... 547
4 .4. Planetary symbols from the primary horoscope of the DR zodiac............................... 547
4.5. Secondary horoscopes in the DR zodiac............................................................................ 550
xvi | h is t o r y : fic t io n o r sc ien c e; chron 3

4.5.2. Autumn equinox horoscope in the DR zodiac...................................................... 550

4.5.2. Winter solstice horoscope in the DR zodiac.......................................................... 553
4.5.3. Spring equinox horoscope in the DR zodiac........................................................ 553
4.5.4 . Summer solstice horoscope in the DR zodiac...................................................... 553
4.6. The exhaustive solution for the Round Zodiac: morning
of 20 March 1185 a .d ............................................................................................................. 554
4 .7. Verification table for the exhaustive solution of the Round Zodiac.............................. 557
5. The decipherment of the dating contained in the zodiac from the Greater
Temple of Esna (E B ).......................................................................................................................... 562
5.1. Copies of the zodiac from the Greater Temple of Esna.................................................... 562
5.2. Coloured zodiac from the Greater Temple of Esna. Symbols of
constellations and planets in the primary horoscope...................................................... 563
5.3. The primary horoscope and the doubles of planets in the EB zodiac....................... 566
5.4. Visibility indicators in the EB zodiac.................................................................................. 567
5.5. Secondary horoscopes in the EB zodiac............................................................................. 567
5.6. The exhaustive solution of the EB zodiac: 31 March - 3 April 1394 a .d ...................... 569
5.7 . The verification table for the complete solution of the EB zodiac................................ 572
6 . The decipherment of the date from the zodiac found in the Lesser Temple
o f Esna (E M )....................................................................................................................................... 577
6 .1. Drawn copies of the zodiac from the Lesser Temple of Esna......................................... 578
6 .2. The coloured version of the zodiac.................................................................................... 578
6 .3. Constellation parentheses in the primary horoscopes planetary
row in the EM zodiac............................................................................................................ 581
6 .4. Planetary figures of the primary horoscope in the EM zodiac...................................... 583
6.5. Secondary horoscopes and auxiliary scenes in the EM zodiac....................................... 584
6.5.1. Autumn equinox horoscope in the EM zodiac..................................................... 584
6.5.2. Winter solstice horoscope in the EM zodiac......................................................... 585
6.5.3. The horoscope of vernal equinox and the additional scene
between Aquarius and Capricorn in the EM zodiac............................................ 586
6.5.4 . Summer solstice horoscope in the EM zodiac...................................................... 587
6.6. The exhaustive solution of the EM zodiac: 6-8 May 1404 a . d ....................................... 588
6.7. The verification table for the exhaustive solution of the EM zodiac............................. 589
7. The correlation between the solution dates and the New Chronology
as well as our reconstruction of history......................................................................................... 594

Chapter 18 Dates in zodiacs discovered inside Egyptian sepulchres

1. The Athribis zodiacs of Flinders Petrie (AV + A N ).................................................................. 601
1.1. The decipherment of the primary horoscope. Six options of planetary
identification............................................................................................................................ 601
1.2. Secondary horoscopes and additional scenes in the zodiacs from Athribis................. 606
1.3. Results of calculations including six options with rigid planetary ord er..................... 607
1.4 . Calculation results for six versions with random order of invisible planets................ 609
1.5. The exhaustive solution of the zodiacs from Athribis: 15-16 May 1230
for the Lower, and 9-10 February 1268 for the Upper Z odiac........................................ 613
1.6. A comparison of planetary positions in the solutions with those specified
in the zodiacs........................................................................................................................... 614
1.7 . Checking correspondence to the secondary horoscope of summer solstice................ 615

1.8. Verification by the scene of meeting in Leo .................................................................... 616

1.9 . The archaic June year as used in the zodiacs of Athribis................................................. 618
1.10. Final identification of the planetary bird s........................................................................ 619
1.11. Verification of the solutions stability................................................................................. 619
1.12. Corollaries............................................................................................................................... 619
2. The Theban Zodiac of Brugsch (B R )............................................................................................. 620
2.1. The demotic subscript horoscope in Brugschs zodiac.................................................... 621
2.2. The horoscope without rods from Brugschs zodiac.................................................... 621
2.3. The horoscope with boats in Brugschs zodiac.............................................................. 622
2.4. Corollaries................................................................................................................................ 625
2.5. The version with the baboon representing the moon in the
horoscope without rods ..................................................................................................... 625
3. The colour zodiac from Thebes (O U )......................................................................................... 628
3.1. Constellation figures............................................................................................................... 628
3.2. Planetary figures...................................................................................................................... 629
3.3. The primary horoscope and the extra conditions............................................................. 631
3.4. Preliminary solutions of the primary horoscope.............................................................. 631
3.5. Verification by compliance to additional criteria.............................................................. 633
3.6 . Corollary: the date transcribed in the OU zodiac is 5-8 September 1182 a.d ............. 634

Chapter 19 Dating results for Egyptian zodiacs

1. The general situation with the datings of the Egyptian zodiacs............................................... 638
2 . The stability of the datings that we came up w ith ...................................................................... 640
3. Unresolved issues in the decipherment of Egyptian zodiacs..................................................... 640
4. Astronomical dating of Sumerian tablets...................................................................................... 641
5. A list of 28 ancient zodiacs, discovered and dated by the authors recently............................ 643

1. Tables of fast and named stars of the Almagest that can be identified reliably..................... 647
2. The computer program of the geometrical method of dating of star
configurations by their proper movement taking into account the systematic
errors of the catalogue....................................................................................................................... 695
3. The description of the Horos program as used for the purposes of dating
the Egyptian zodiacs.......................................................................................................................... 706
4. Input data for the Horos program that yielded finite interpretation options........................ 709
5. Julian day numbers and the days of solstices and equinoxes as taken for
the beginnings of centuries in the p a st.......................................................................................... 723
6 . A list of solutions for the zodiacs from Athribis under less strict conditions........................ 724
7. A more precise drawn copy of the Greater Zodiac of Esna performed by
the authors of the present book in Egypt and based on the original...................................... 729
8. Our replies to the authors of certain erroneous works, who tried to refute
our astronomical datings.................................................................................................................. 737
9. WIKIPEDIA Damage Control on New Chronology (Fomenko) page.................................... 751

The complete bibliography to the seven volumes 755

x v iii | h is t o r y : f ic t io n o r s c ie n c e ?

From the Publishers

History: Fiction or Science? contains data, illustrations, charts
and formulae containing irrefutable evidence o f mathemati
cal, statistical and astronomical nature. You may as well skip
all o f it during your first reading. Feel free to use them in
your eventual discussions with the avid devotees o f classical
chronology. In fact, before reading this book, you have most
probably been one o f such devotees.

After reading History: Fiction or Science? you will develop a

more critical attitude to the dominating historical discourse
or even become its antagonist. You will be confronted with nat
ural disbelief when you share what youve learned with oth
ers. Now you are very well armed in face o f inevitable scepti
cism. This book contains enough solid evidence to silence any
historian by the sheer power of facts and argumentation.

History: Fiction or Science? is the most explosive tractate on

history ever written - however, every theory it contains, no
matter how unorthodox, is backed by solid scientific data.

The dominating historical discourse in its current state was es

sentially crafted in the XVI century from a rather contradic
tory jumble o f sources such as innumerable copies o f ancient
Latin and Greek manuscripts whose originals had vanished in
the Dark Ages and the allegedly irrefutable proof offered by late
mediaeval astronomers, resting upon the power o f ecclesial
authorities. Nearly all of its components are blatantly untrue!

For some o f us, it shall possibly be quite disturbing to see the

magnificent edifice o f classical history to turn into an omi
nous simulacrum brooding over the snake pit of mediaeval
politics. Twice so, in fact: the first seeing the legendary mil-
lenarian dust on the ancient marble turn into a mere layer of
dirt - one that meticulous unprejudiced research can even
tually remove. The second, and greater, attack o f unease comes
with the awareness o f just how many areas o f human knowl
edge still trust the elephants, turtles and whales o f the con
sensual chronology to support them. Nothing can remedy
that except for an individual chronological revolution hap
pening in the minds o f a large enough number o f people.
Anatoly T. Fomenko, Tatiana N. Fomenko,
Vladimir V. Kalashnikov, Gleb V. Nosovskiy

Chronology 3
Third volume o f History: Fiction or Science series

A stro n o m ica l m eth o d s in chronology

P t o lem y s A lm agest
T ycho B ra h e
C o per n ic u s
E g y p t ia n zodiacs

This book is dedicated to the new trend in science The second part o f the book deals with the new
associated with the development and use o f inde datings o f the Egyptian horoscopes. We are referring
pendent natural scientific methods for the dating of to the monumental bas-reliefs discovered in the tem
the ancient and mediaeval historical events. It is the ples o f the ancient Egypt, which depict zodiacal
follow-up to the first two books in the series, ChronI constellations and planets (horoscopes, in other
and C hron2 by Anatoly Fomenko. In the present words). They are all dated to deep antiquity today.
volume (C hron3) we date archaeological artefacts However, modern astronomy permits a different and
and historical texts by their astronomical content. more precise dating. It turns out that each and every
The problem o f independent dating as applied to ancient Egyptian horoscope that we found yields a
historical chronology has got a long history. The idea dating o f X II-X IX century a.d., no less. For instance,
o f applying the methods o f natural science for this the astronomical datings o f the ancient Egyptian
purpose is also far from novel. However, A. T. Fo horoscopes from the temples o f Dendera and Esna
menko, accompanied by a group o f mathematicians (Latopolis) unequivocally refer to the epoch o f the
and physicists from the Moscow State University, was XII-XV century. Apparently, some o f the Egyptian
the first to construct a systematic chronology from constructions that are dated to deep antiquity today
scratch using nothing but natural scientific methods were in fact built in the late Middle Ages.
completely unrelated to the Scaligerian chronologi The book also contains a number o f annexes.
cal scale. This was done in the early 1980s. In order Let us provide a brief synopsis o f the present vol
to distinguish between our chronology (constructed umes contents.
with the aid o f natural scientific methods and noth The first part o f the book deals with the famous
ing but) and the consensual chronology o f Scaliger problem o f solving the star catalogue from Ptolemys
and Petavius, we have called the former New Chron Almagest.
ology. The Introduction contains a concise overview of
The first part o f the present book is based on the the Almagests contents, as well as certain informa
work o f A. T. Fomenko, V. V. Kalashnikov and G. V. tion concerning the Almagest catalogue and a num
Nosovskiy entitled The Dating o f the Almagest Star ber o f other star catalogues. We explain why the prob
Catalogue, which came out in 1995 ( [METH3 ]: 1 and lem o f dating old star catalogues is o f interest to us,
[M ETH3]:2), and was subsequently revised in 2000 and cite information about mediaeval astronomers
([M ETH 3]:3). This book was revised yet again for associated with the creation o f star catalogues.
the present edition, and substantially so, with im Chapter 1 is a collection o f important facts related
portant new material added. to astronomy, astrometry, the history o f astronomi
4 I h ist o r y : fic t io n or sc ien c e? CHRON 3

cal instruments and the methods o f measuring star This gives us a new understanding o f the Almagest
coordinates. structure and allows us to develop a method o f dat
Chapter 2 contains a preliminary analysis o f the ing the catalogue.
Almagest star catalogue. We discuss a plethora of cor In Chapter 7 the Almagest star catalogue is dated
responding problems such as ambiguous star identi by two independent methods: statistical and geo
fication and certain anomalies pointed out by re metric. Both give us the same result - apparently,
searchers earlier, such as the Peters sinusoid. We also Ptolemys observations cannot predate 600 a.d. or
discuss the issue o f latitude and longitude precision postdate 1300 a.d., insofar as the Almagest star cat
in the Almagest catalogue. alogue is concerned (or its oldest part at the very
In Chapter 3 we analyse possible datings o f the least). Other parts o f the Almagest could be written
Almagest star catalogues based on standard methods much later, which must indeed be the case, as we
and ideas. We demonstrate that it is impossible to demonstrate in the chapters to follow.
date a catalogue by more or less standard and ele In Chapter 8 we explain the mysterious Peters si
mentary methods, pointing out the principal diffi nusoid and also analyse the value o f the angle be
culties that require a substantially more refined tween the equatorial and the ecliptic plane as cited in
method. We analyse a number o f known works for the Almagest.
this purpose, whose authors attempted to confirm In Chapter 9 we research and date other famous
the traditional dating o f the Almagest catalogue by old catalogues by Tycho Brahe, Ulugbek, Hevelius
proper star motions, exposing the reasons why they and Al-Sufi. These catalogues illustrate the method we
failed. suggest; the results are discussed.
At the end o f Chapter 3 we describe the concep Chapter 10 was written by A. T. Fomenko and G. V.
tion o f our star catalogue dating method. Nosovskiy. It considers the possibility o f dating the
In Chapter 4 we identify fast stars as the stars men Almagest by other astronomical observation data that
tioned in the Almagest catalogue. Obviously enough, it contains apart from the Almagest. The results are
such identification isnt always possible. Moreover, it in complete concurrence with our dating o f the Al
depends on the alleged dating o f Ptolemys observa magest star catalogue. We restore the Ptolemaic
tions in general. The same fast star whose position on chronology, or the chronological ideas adhered to
the celestial sphere changes over the years can be iden by Ptolemy himself or the XVI-XVII century editors
tified as several stars from the Almagest catalogue. o f his books. These ideas were subsequently forgot
This effect is important. A failure to comprehend it ten due to the erroneous conversion o f the Ptolemaic
has already led several authors (such as Y. N. Yefremov dates into their a.d . equivalents inherent in Scali-
and Y. A. Zavenyagin) to erroneous datings o f the Al gerian chronology.
magest catalogue. In Chapter 11, also written by A. T. Fomenko and
Chapter 5 contains mathematical results used in G. V. Nosovskiy, we discuss many other problems as
the statistical analysis o f star catalogues. We classify sociated with the dating o f the Almagest in general.
various catalogue discrepancies and discuss various The second part o f the book was written by A. T.
methods o f discovering the latter and compensating Fomenko, T. N. Fomenko and G. V. Nosovskiy; it de
the systematic compound. scribes the new method o f dating the Egyptian zodi
Chapter 6 contains the results o f our global sta acs. The method is used to date the ancient Egyptian
tistical calculations involving the entire Almagest star zodiacs from the temples o f Dendera and Esna, as
catalogue as well as its parts. The discovered statisti well as the horoscopes discovered inside Egyptian
cal characteristics o f different parts o f the Almagest tombs. All the dates turn out mediaeval and pertain
has made it feasible to find the well-measured and to the XII century a.d. the earliest.
poorly measured regions o f the celestial sphere. We
have discovered that the Almagest star atlas could be A. T. Fomenko
divided into uniformity regions whose stellar coor Moscow State University,
dinate precision differed drastically from each other. D epartm ent o f M athematics an d Mechanics

A. T. Fomenko, V. V. Kalashnikov, G. V. Nosovskiy

i. philosophy in theory and practice. Indeed, even

A BRIEF DESCRIPTION OF THE ALMAGEST notwithstanding earlier attempts to unite the two,
one could always see a great difference between them.
The Almagest is the famed mediaeval oeuvre that Firstly, although certain moral virtues might be pos
deals with astronomy, spherical geometry and calen sessed by a great multitude o f uneducated people, no
dar issues. It is believed to have been written by Clau study o f the ways of the Universe is possible without
dius Ptolemy, an astronomer, mathematician and ge prior education. Secondly, the former benefit the most
ographer from Alexandria. Historians date his lifetime due to incessant activity, whereas the latter relish in
to the II century a.d . We shall cite some brief infor the advancement of theoretical research. We therefore
mation about Ptolemy below. However, one must in deem it necessary to let our mental conceptions con
stantly point out that, according to certain specialists trol our actions most rigidly on the one hand, so as
in the history o f astronomy, Likewise his works, the to refer to a perfect and elegant ideal all the time,
personality o f Ptolemy was treated rather strangely by and, on the other, to direct most o f our energy to
history. His contemporaries have left no historical wards familiarising ourselves with a multitude o f ex
records o f either his life or his endeavours ... We quisite theories and learning many more things per
dont know so much as the approximate dates o f Ptol taining to the discipline commonly referred to as
emys birth and death or indeed any other details of mathematics in the narrow sense o f the word ... If we
his biography ([98], page 6 ). Figs. 0.1, 0.2, 0.3, 0.4, are to educe the primordial reason that has set the
0.5 and 0.6 reproduce ancient portraits o f Ptolemy. Universe in motion in the simplest form, it was the
According to Scaligerian chronology, the Almagest immanent and invisible God. The next section is the
was created in the reign o f the Roman emperor Anto ology ... The section that studies the material and the
ninus Pius, who reigned in 138-161 a.d . ever-changing qualitative aspects such as whiteness,
Let us instantly point out that the very literary warmth, sweetness, softness etc., is called physics ...
style o f the epoch, which is at times excessively Finally, the discipline concerned with the qualitative
grandiloquent and meandering, is more likely to hail motions and shapes... can be defined as mathemat
from the epoch o f the Renaissance than deep antiq ics ([704], pages 5-6).
uity, when paper and parchment (let alone books) The style is perfectly typical for late mediaeval sci
were luxuries. See for yourselves - the Almagest be entific (or, as they are also called, scholastic) works
gins like this. o f the XV-XVII century. One most vivid detail is the
O Sire, it appears to me that the true philoso reference to an invisible and immanent God by Ptol
phers made the most laudable distinction between em y - a characteristic element of the Christian dogma,
8 | h is t o r y : fic t io n or sc ien c e? CHRON 3 I PART 1


Fig. 0.1. Ancient drawing of Fig. 0.2. Ancient sculpture Fig. 0.3. Ancient depic Fig. 0.4. Ancient portrait of
Ptolemy dating from 1584. depicting Ptolemy from the tion o f Ptolemy from Ptolemy, where he looks
Ptolemy is holding a Jacobs rod. cathedral o f Ulm (around the Global Chronicle by like a typical mediaeval
Thevet. Les vrais protr. et vies 1469-1474). The statue was Hartmann Schedel. European. Taken from [98],
dhommes illustres... Paris, 1584. made by Jorg Sirlin the Senior. Augsburg, 1497. Taken page 7.
Taken from [704], page 431. Taken from [704], page 448. from [90], page 25.

quite alien to the polytheism o f the Olympians. And nowadays. According to the Scaligerite historians, it
yet Scaligerian chronology tries to convince us that was written almost two thousand years ago.
Christianity only became the official religion in the It has to be pointed out that certain researchers
IV century a.d ., and the ancient Greek Ptolemy consider the existing division o f the Almagest into
from the II century a.d. is clearly considered a pre- chapters to be more recent than the book itself, like
Christian author by the historical authorities. wise the names o f the chapters ([1358], pages 4-5).
We would like to introduce the reader to the However, this fact is o f no importance to us presently,
Almagests table o f contents, given that this funda since our only goal is to familiarise the readers with
mental scientific oeuvre is hardly a popular read the structure o f the Almagest.

7u k a ilo IOMCCS
toaom e

Fig. 0.5. Ancient portrait o f Ptolemy. Fig. 0.6. Ancient drawing o f Ptolemy on the Cosmosphere o f Vassily Kiprianov, 1707.
Wood engraving, XVI century. Taken Ptolemy is wearing something that resembles an Ottoman turban. Taken from [90],
from [ 1160], page 25. page 212.

T he Almagest: table of contents. 9. On individual issues related to the ascension

Volume 1.
10. On the angles constituted by the circumfer
1. Foreword.
ence that crosses the middles o f zodiacal
2. On the continuity o f narration.
constellations (ecliptic) and the meridian
3. On the spherical nature and motion of the
circle (meridian).
11. On the angles between the ecliptic and the
4. On the spherical nature o f the Earth in general.
5. The Earth as the centre o f the heavens.
12. On the angles and arcs formed by the same
6 . The Earth as a point as compared to the
circumference (the ecliptic) and the circum
ference that crosses the horizons poles.
7. On the immobility o f the Earth.
13. The values o f angles and arcs for different
8. On two different main kinds o f celestial
9. On individual concepts.
Volume 3.
10. On the sizes o f the chords.
1. On the duration o f a year.
11. Chord table.
2. Tables o f mean Solar motion.
12. On the arc between the two solstices.
3. On the hypotheses related to even circular
13. Preliminary data for spherical geometry
4. On the visible irregularity o f solar motion.
14. On the arcs between the equinoctial circle
5. On defining the irregularity quotients for
and the slanting circle (the equator and the
different position.
ecliptic, in other words).
6 . Solar anomaly table.
15. Declination table.
7. On the mean solar motion epoch.
16. On the sunrise phases in the straight sphere.
8. On the calculation o f the solar position.
Volume 2. 9. On the inequality o f daytime and nighttime.
1. On the general location o f the inhabited part
o f the Earth. Volume 4.
2. How to calculate the horizontal arc between 1. What observations the lunar theory must be
the equator and the ecliptic knowing the based on.
maximum daytime duration. 2. On lunar periods.
3. How to find the height o f the pole under 3. On individual values o f the Moons mean
similar assumptions and vice versa. motions.
4. How to calculate when and how often the Sun 4. Tables o f mean lunar motions.
happens to be right above ones head for 5. On the identical nature o f the events observed
different areas. under the simple hypothesis o f lunar motion,
5. How to calculate the gnomon proportions in either eccentric or epicyclical.
relation to the length o f the meridian shadow 6 . The definition o f the first (or simple) lunar
during solstices and equinoxes, knowing the inequation.
values mentioned above. 7. On the adjustment o f the Moons mean
6 . A list o f individual parallels special motions by longitude and anomaly.
characteristics. 8. On the epoch o f the Moons mean motions by
7. On simultaneous ascensions in the slanting longitude and anomaly.
spherical circle part that crosses the middles 9. On the adjustment o f the Moons mean posi
o f zodiacal constellations and the equinox tions and their epochs by latitude.
circle (the equator). 10. The table o f the first (or simple) lunar
8. 10 -degree arc ascension timetable. inequation.
10 h is t o r y : f ic t io n or sc ien c e? CHRON 3 I PART 1

11. On the fact that the discrepancy between the 6 . On the intervals between eclipse months.
lunar inequation value o f Hipparchus and 7. The construction o f eclipse table.
the one discovered by the authors results 8. Eclipse tables.
from calculations and not from a priori 9. Lunar eclipse calculations.
assumptions. 10. Solar eclipse calculations.
11. On the eclipse declination angles.
Volume 5. 12. Eclipse declination table.
1. On the construction o f the astrolabe. 13. Declination definition.
2. On the hypotheses o f the double lunar
inequation. Volume 7.
3. On the value o f the lunar inequation that 1. On the immobile stars, whose position in
depends on the Moons position in relation to relation to one another never changes.
the Sun. 2. On the retrograde motion o f the immobile
4. On the proportion value o f the lunar orbits star sphere alongside the ecliptic.
eccentricity. 3. On the circular nature o f the retrograde
5. On the declination o f the lunar epicycle. motion o f the immobile star sphere around
6 . How to calculate the position o f the Moon the ecliptic poles.
geometrically, relying on periodic movements. 4. On the methods o f compiling an immobile
7. Construction o f the full moon inequation star catalogue.
table. 5. Northern Hemisphere constellation catalogue.
8. The full moon inequation table.
9. On calculating the position o f the Moon in Volume 8.
general. 1. The Southern Hemisphere constellation
10. On the fact that the syzygy difference pro catalogue.
duced by the lunar eccentricity is marginal. 2. On the position o f the Milky Ways circum
11. On the lunar parallax. ference.
12. On the construction o f the parallax instru 3. On the construction o f the cosmosphere.
ment. 4. On the configuration characteristic for the
13. Estimating the lunar distances. immobile stars.
14. On the values o f visible diameters o f the 5. On simultaneous ascensions, culminations
Sun, the Moon and the shadow o f the Earth and descents o f immobile stars.
in syzygies. 6. On the first and last moments o f the
15. On the distance to the Sun and various im immobile stars visibility.
plications o f this calculation.
16. On the sizes o f the Sun, the Moon and the Volume 9.
Earth. 1. On the order o f the spheres o f the Sun, the
17. On individual values o f solar and lunar Moon and the five planets.
parallaxes. 2. On the aims o f our planetary hypotheses.
18. Parallax table. 3. On the five planets returning periodically.
19. Parallax definition. 4. Mean longitudinal motion table and the
anomaly o f the five planets.
Volume 6. 5. Primary postulations concerning the hypo
1. On the new moons and the full moons. theses o f five planets.
2. Compilation o f the mean syzygy table. 6 . On the character o f the hypotheses and the
3. New moon and full moon tables. respective discrepancies.
4. How to calculate the mean and the true syzygy. 7. Estimating Mercurys apogee position and its
5. On the limits o f solar and lunar eclipses. movements.

8. How planet Mercury gets the closest to the 4. The calculation o f Marss retrograde motion.
Earth twice in one move. 5. The calculation of Venuss retrograde motion.
9. On the size and proportions o f Mercurys 6 . The calculation o f Mercurys retrograde
anomalies. motion.
10. Mercurys periodic motion rectified. 7. Stationary point table construction.
11. On the epoch o f Mercurys periodic motion. 8. Stationary point tables. Amended anomaly
Volume 10. 9. Estimation o f the maximal possible distances
1. Estimating the apogee o f Venus. between Venus, Mercury and the Sun.
2. On the size o f the planets epicycle. 10. Tables o f maximal distances between the
3. On the relations between the eccentricities of planets and the true position o f the Sun.
planet Venus.
4. On the amendment o f the planets periodic Volume 13.
motions. 1. On the hypotheses that concern the latitudi
5. On the epoch o f the periodic motion o f Venus. nal motion o f the five planets.
6 . Preliminary data about other planets. 2. On the character o f motion in the alleged
7. Estimating the eccentricity and the apogee of inclinations and obliquities in accordance to
Mars. the hypotheses.
8. Estimating the epicycle o f Mars. 3. On the size o f the obliquities and inclinations.
9. Rectification o f the periodic motion o f Mars. 4. The construction o f tables for the individual
10. On the epoch o f the periodic motion of values o f longitudinal discrepancies.
Mars. 5. Table for latitudinal calculations.
6 . Latitudinal discrepancy calculations for the
Volume 11. five planets.
1. Estimating the eccentricity and the position of 7. First and last visibility moments for the five
Jupiters apogee. planets.
2. Estimating the epicycle of Jupiter. 8. How certain particular details o f Venus and
3. The amendment o f its periodic motion. Mars ascending and descending correspond to
4. On the epoch o f Jupiters periodic motion. consensual hypotheses.
5. Estimating the eccentricity and the position of 9. The method o f estimating the distance to the
Saturns apogee. Sun for individual cases o f heliacal ascensions
6 . Estimating the epicycle o f Saturn. and descents.
7. The amendment o f its periodic motion. 10. Tables o f heliacal ascensions and descents for
8. On the epoch of Saturns periodic motion. the five planets.
9. How the periodic motion can be used for a 11. Epilogue.
geometric calculation o f the true positions.
10. The construction o f the anomaly table. Therefore, the Almagest consists o f 13 volumes,
11. Tables for the estimation o f the longitudes o f which occupy 430 pages o f a broadsheet modern edi
the five planets. tion ([704]).
12. On calculating the longitudes o f the five This book is also concluded in the most remark
planets. able manner. The epilogue is as follows:
After we have made it all come to pass, o Sire,
Volume 12. and considered nearly everything that I believe nec
1. On the preliminary considerations concerning essary to be considered in such an oeuvre, inasmuch
retrograde motion. as the time that has passed appears to have helped
2. The calculation o f Saturns retrograde motion. with perfecting the precision o f our discoveries - by
3. The calculation o f Jupiters retrograde motion. no means having an idle boast as an ulterior motive,
12 | h is t o r y : fic t io n or sc ien c e? CHRON 3 I PART 1

but rather in order to be o f use to science; may our 4. The Earth is immobile (doesnt travel from
present work have an apropos and a fitting ending place to place).
([704], page 428). Many of these claims were educed from the Aris
As we can see, Ptolemys work is dedicated to a cer totelian philosophy according to Ptolemy himself.
tain Sire, or Czar. Historians appear to be greatly sur Furthermore, Volumes 1 and 2 are collections o f el
prised by this fact. Modern commentary is as follows: ements o f spherical astronomy - the spherical trian
This name [Sire = Czar - Auth.] was rather popu gle theorems, the method o f measuring the arcs (an
lar in Hellenistic Egypt in the epoch in question. We gles) by known chords etc. Volume 3 relates the the
have no other data about this person - we dont even ory of visible annual motion o f the Sun, discusses
know whether he was associated with astronomy in the dates of equinoxes, the length of a year etc. Volume
any way at all ([704], page 431). However, the very 4 considers the length of a synodal month, which is
fact that the Almagest was associated with the name the cycle of lunar phase repetition. It consists of circa
of a certain Czar can be proven by the following cir 29 days, 12 hours, 44 minutes and 2.8 seconds. The
cumstance. Apparently, Ptolemy was also ascribed same book relates the theory of lunar motion. Volume
royal ancestry in late antiquity and in the Middle 5 discusses the construction of certain observation in
Ages ([704], page 431). Also, the very name Ptolemy struments and continues the research of the theory
(or Ptolomy) is presumed to have been the dynasty of lunar motion. Volume 6 describes the theory of
name o f the Egyptian kings who reigned after solar and lunar eclipses.
Alexander the Great ([797], page 1076). The famous star catalogue that contains around
At any rate, according to Scaligerian chronology, 1020 stars is part of the seventh and the eighth vol
the Ptolemaic dynasty left the stage around 30 b .c. umes of the Almagest, which also discuss the prop
([797], page 1076) - more than a hundred years ear erties and characteristics o f immobile stars, the mo
lier than Ptolemy the astronomer, in other words. tions o f the stellar sphere etc.
Thus, the only thing that precludes us from identify The last five volumes of the Almagest contain a
ing the epoch o f the Ptolemaic rulers as the epoch of theory of planetary motion. Ptolemy mentions five
Ptolemy the astronomer is Scaligerian chronology. planets, namely, Saturn, Jupiter, Mars, Venus and Mer
Apparently, in the Middle Ages, when Scaligerian cury.
chronology had not yet existed, the Almagest was as
cribed to the Ptolemaic kings and none other - nam 2.
ing them as the organisers o f this grandiose endeav A BRIEF HISTORY OF THE ALMAGEST
our or the customers who had ordered this astro
nomical tractate. This is why the Almagest was As we have already pointed out, Scaligerian chron
canonised, becoming absolutely authoritative for a ology believes the Almagest to have been created in
long time to follow. It is easy enough to understand the reign of Emperor Antoninus Pius, in 138-161 a.d.
why the book begins and ends with a dedication to a Furthermore, it is presumed that the last observation
certain Czar, or Sire. It was the royal textbook on as included in the Almagest dates from 2 February 141
tronomy, in a way. We shall find out just when it was a.d . ([1358], page 1). The period o f Ptolemys ob
written in the present book. servations that the Almagest is based upon falls over
The first volume o f the Almagest voices a number 127-141 a.d .
o f general principles, in particular the following: The Greek name of the Almagest translates as Sys
1 . The sky is really a celestial sphere and rotates as tematic Tractate on Mathematics, emphasising the
such. fact that the Almagest represents the epochs sum
2. The Earth is a sphere located at the centre of the total of Greek mathematical astronomy. It isnt known
Universe (heavens). whether other astronomical textbooks comparable
3. The Earth can be considered a point in space as to the Almagest existed in the epoch of Ptolemy. Mod
compared to the distance to the sphere o f immobile ern scientists attempt to explain the unprecedented
stars. success o f the Almagest among the astronomers and

scientists in general by a chance loss o f the majority the Arabic made by Gerhard o f Cremona in Toledo
o f all the other astronomical works o f the epoch and finished by 1175 a.d . Greek manuscripts [of the
([1358]). The Almagest was the main textbook on Almagest - Auth.] started to reach the West in the XV
astronomy in the Middle Ages. If we are to believe the century; however, it was Gerards text that remained
Scaligerian chronology, it served in this quality for fif the basis o f books on astronomy for ages and gener
teen hundred years, no less, making a tremendous ations to come, up until the compilation o f a concise
impact on mediaeval astronomy in Islamic and Chris version o f the Almagest by Purbach and Regiomon
tian lands up until the XVII century a.d . The au tanus ... This was the first printed version of the Al
thority o f this book in the mediaeval scientific com magest (Venice, 1515). The sixteenth century wit
munity compares to nothing but Euclids Elements. nessed a wide propagation o f the Greek text (pub
As it is pointed out by Toomer, for instance ([1358], lished by Hervagius in Basel in 1538) and the waning
page 2 ), it is exceptionally hard to trace the history of of the Ptolemaic astronomical systems influence, not
the Almagest and its influence in the antiquity (be so much caused by the work o f Copernicus (which
tween the II century a.d . and the Middle Ages). One has been clearly influenced by the Almagest, be it the
usually judges the role of the Almagest as the standard form or the conceptions voiced therein) as by those
textbook for advanced students in the period o f the o f Tycho Brahe and Kepler ([1358], pages 2-3).
so-called decline o f the antiquity by the comments
o f Pappus and Theon o f Alexandria ([1358], page 2). 3.
The Scaligerian version of history tells us of a lugubri THE PRINCIPAL STAR CATALOGUES
ous and taciturn epoch that is presumed to have fol OF THE MIDDLE AGES
lowed - we shall discuss it in detail in Chapter 1 1 . For
the meantime, let us just point out the following char And so, the Almagest (its star catalogue in partic
acteristic of this fictitious Scaligerian stagnation age ular) ranks as the oldest more or less informative and
as given by a modern specialist in the history o f as detailed astronomical work that has reached our day
tronomy: After the astonishing efflorescence o f the and age. The approximate Scaligerian dating o f the Al
ancient culture on the European continent came a magest is the II century a.d. However, it is assumed
lengthy period o f stagnation and even regress in cer that Ptolemy used the star catalogue o f Hipparchus,
tain aspects - a 1000-year period commonly referred his predecessor who had lived in the II century b .c.
to as the Middle Ages... Not a single astronomical dis The catalogue in question has not survived in its orig
covery o f any significance was made in this millen inal form. Likewise other mediaeval catalogues, the
nium ([395], page 73). Almagest catalogue contains circa 1000 stars, whose
Furthermore, Scaligerian history is o f the opinion positions are indicated as their latitudes and longi
that in the V III-IX century the Almagest emerged tudes in ecliptic coordinates. It is presumed that no
from obscurity due to a growing popularity o f Greek other star catalogues but the one contained in the Al
science in the Islamic world and was translated into magest were known before the X century a.d.
Syrian; this was followed by several Arabic transla Finally, according to Scaligerian chronology, the
tions. At least five such translation versions are known first mediaeval star catalogue was compiled in the
to have existed by the middle o f the XII century a.d. X century a.d . in Baghdad by al-Sufi, an Arabic as
See more about this in Chapter 1 1 . Today it is be tronomer. His full name is Abd al-Rahman ben Omar
lieved that Ptolemys work, originally written in ben Mohammed ben Sala Abu al-Husain al-Sufi (903-
Greek, was still copied and even studied in the East, 986 a.d ., qv in [544], Volume 4, page 237). The cat
Byzantium in particular, but not the West. In the alogue o f al-Sufi has survived; a closer study reveals
Western Europe, all knowledge of this work remained it to be identical to the same old Almagest catalogue.
lost up until the early Middle Ages. Although several However, if the surviving copies and editions o f the
translations were made from Greek to Latin in the Almagest contain a star catalogue rendered to circa
Middle Ages, the primary source for the rediscovery 100 a.d. by precession as a rule (although there are
o f the Almagest in the West was a translation from exceptions), the catalogue of al-Sufi is the very same
14 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

catalogue rendered by precession to the X century attributed Ptolemy to the alleged II century a.d ., was
a.d . This fact is known quite well to astronomers - considered indisputable. However, in the X IX cen
see [1119], page 161, for instance. Let us also point tury a more meticulous analysis o f the stellar longi
out that rendering a catalogue to a random desired tudes contained in the Almagest revealed that pre
historical epoch was an easy enough task. A certain cession-wise these longitudes correspond to the epoch
constant would be added to the longitudes o f stars - o f the II century b .c. - the epoch o f Hipparchus, in
the same value for each and every star. This is a very other words. This is how A. Berry relates the situation:
simple arithmetical operation; actually, the Almagest The seventh and the eighth volumes [of the Almagest
describes it in great detail. - Auth.] contain a star catalogue and a description of
The next surviving catalogue in Scaliger-Petavius the precession. The catalogue, which includes 1028
chronology was compiled by Ulugbek in Samarqand stars (three o f them double) appears to be virtually
(1394-1449 a.d .). None o f the three is very precise, identical to that o f Hipparchus. It doesnt contain a
since they all indicate star coordinates using a scale single star that could be seen by Ptolemy in Alexandria
with a step of 10 arc minutes. Next, we have the famed and could not be seen by Hipparchus on the Rhodes.
catalogue o f Tycho Brahe (1546-1601), which is al Moreover, Ptolemy claims to have defined the value
ready substantially more precise. Brahes catalogue is of precession as 36" (and erroneously so) after a com
believed to be the greatest advance o f mediaeval in parison o f his observations to the data o f Hipparchus
struments and observation technology in general. and other astronomers. Hipparchus considers this
Post-Tychonian catalogues are abundant; however, value as the least possible result, whereas for Ptolemy
they are o f no interest to us presently. it is the final estimate. The positions o f stars in Ptol
emys catalogue correspond the most to their true
4. positions in the time o f Hipparchus, taking into ac
THE REASON W H Y THE DATING OF THE OLD count the alleged annual precession o f 36", and less
STAR CATALOGUES IS AN IMPORTANT ISSUE so - to their actual positions in Ptolemys epoch. It is
therefore very likely that the catalogue in question
Every new star catalogue is the result o f a great has got nothing in common with Ptolemys original
body o f work conducted by an observing as observations, being de facto the very same catalogue
tronomer; most likely, a whole group o f professional as that o f Hipparchus, with compensated precession
observers who needed all the professionalism, con only slightly altered by the observations o f Ptolemy
centration and meticulousness they could muster as and other astronomers ([65], pages 68-69).
well as the ability to use state-of-the-art measure The issue o f dating the catalogue becomes crucial
ment instruments o f their epoch to the maximum. in this case. Ever since the XV III century the as
Apart from that, a catalogue required a correspon tronomers and the specialists in history o f astronomy
ding astronomical theory, or cosmology. Thus, each have been analysing the Almagest catalogue and the
and every ancient catalogue was the epitome o f its Almagest in general, trying to sort out the data it
epochs astronomical thought. By analysing a cata contains, distinguish between the observations of Hip
logue we can find out a lot about the epochs qual parchus and Ptolemy etc. A great deal of literature has
ity o f measurements, the level o f astronomical been written about the dating of the observations that
knowledge etc. the Almagest catalogue is based on. We are by no
However, in order to comprehend the results o f a means attempting to analyse it in depth here and refer
given catalogues analysis, one must know the date o f the interested reader to [614], for instance, where one
its compilation. Any change o f date automatically can find a guide to the respective publications.
changes our estimates, our concept o f the catalogue We have another question to ask - is it possible to
etc. And it isnt always an easy task to calculate the date create a mathematical method that permits dating the
o f a given catalogues creation - this can be observed ancient star catalogue from within - in other words,
best in case o f the Almagest. Initially, in the XVIII by using nothing but the numeric information con
century, the veracity o f the Scaligerian version, which tained in the star coordinates that the compiler of the

catalogue included into his oeuvre? Our answer is in Hipparchus. The catalogue itself has not survived.
the positive. We have developed a method to serve However, it is believed that the ecliptic longitude and
this end, tested it on several veraciously dated cata latitude o f each star was indicated there, as well as the
logues, and then applied it to the Almagest. The reader magnitude. It is believed that Hipparchus localised the
shall find out about our results in the present book. stars using the same terms as the Almagest: the star
Let us now cite some brief biographical data con on the right shoulder o f Perseus, the star over the
cerning the astronomers whose activities are imme head o f Aquarius etc ([395], page 52).
diately associated with the problem as described above. One invariably ponders the extreme vagueness of
These data are published in Scaligerian textbooks. One this star localization method. Not only does it imply
must treat them critically, seeing as how the Scaligerian a canonical system o f drawing the constellations and
version o f history is based on an erroneous chronol indicating the stars they include - another stipulation
ogy (see ChronI and Chron2). We shall consider is that there are enough identical copies of a single star
other facts that confirm it in the present book. chart in existence. This is the only way to make the
verbal descriptions o f stars such as the above work
5. and help a researcher with the actual identification of
HIPPARCHUS stars. However, in this case the epoch o f the cata
logues propagation must postdate the invention of
Scaligerian history is o f the opinion that astron the printing press and the engraving technique, since
omy became a natural science owing to the works of no multiple identical copies o f a single work could be
Hipparchus, an astronomer from the ancient Greece manufactured earlier.
who lived around 185-125 b .c. He is also believed to Nearly the entire body o f information that we have
have been the first to discover the equinoctial pre on the ancient Greeks star science comes from the
cession, which shifts the equinox points across the two surviving works - Ptolemys Almagest and a
ecliptic in the reverse direction from which the lon work o f Hipparchus entitled A Commentary to
gitudes are counted in over the course of time. Ecliptic Aratus and Eudoxus, written around 135 b .c. ([614],
longitudes o f all stars grow as a result. Specialists in page 2 1 1 ). The issue o f stellar mobility - in other
the history o f astronomy tell us the following: Very words, whether or not individual stars move indi
little is known about the life o f Hipparchus. He was vidually in relation to the sphere o f immobile stars,
born in Nicaea (nowadays the city of Iznik in Turkey), was already discussed by Ptolemy, whose verdict was
lived in Alexandria for a while and worked on the negative (in particular, Ptolemy begins the VII volume
Isle o f Rhodes, where his astronomical observatory o f the Almagest with an analysis o f certain star con
was erected ([395], page 43). figurations cited by Hipparchus, a long time before
It is believed that the explosion o f a nova was the Ptolemys own epoch, claiming the configurations in
impetus which had made Hipparchus compile a cat question to be valid for his epoch as well ([704], page
alogue o f stars in the first place. Pliny the Elder (23- 210; also [614], page 212).
79 a.d .) is usually quoted in this respect - he reports Judging by this example and several others, Ptol
that Hipparchus discovered a new star as well as yet emy claims to have demonstrated the constancy o f rel
another star that came into being around that time. ative stellar positions ([614], page 213). Therefore,
According to other sources ([395], page 51), Hippar according to Scaligerian history, the proper star mo
chus noticed the explosion of a nova in 134 b .c. This tion issue first emerged in the II century a.d .
led Hipparchus to the idea that certain changes are
likely to take place in the stellar world - they are too 6.
slow to be discovered within the lifetime o f several PTOLEMY
generations. He decided to compile a 850-item star
catalogue in order to provide his distant descendants According to A. Berry, The last glorious name we
with such an opportunity ([395], page 51). encounter in the history o f Greek astronomy is that
Ptolemys Almagest tells us about the catalogue of o f Claudius Ptolemy. We know nothing about his life,
i6 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

apart from the fact that he lived in Alexandria around that refraction compensation remained a complex
120 a . d . His fame is largely based on the enormous task even in the times o f Tycho Brahe, or the second
astronomical tractate known as the Almagest - it is half o f the XVI century a . d ., will be related separately,
our primary source o f information on Greek astron in the Tycho Brahe section. One cant help suspect
omy, which can by all means be considered the de ing that the ancient Optics o f Ptolemy were writ
finitive encyclopaedia o f mediaeval astronomy. ten in this very epoch of the XVI-XVII century.
Several lesser astronomical tractates are ascribed As for the name of the Almagest, this is what we
to Ptolemy as well - some o f them are unlikely to be learn from A. Berry: The name o f the main manu
authentic, though. Also, Ptolemy was the author o f a script translates as The Great Work, although the
valuable work on geography, and, possibly, a tractate author himself refers to his book as The Mathemat
on optics as well. Among other things, the optics dis ical Work. The Arabic translators, whether out o f re
cipline includes the study o f light refraction in the at spect or accidentally, translated The Great Work as
mosphere of the Earth; it is explained in the book The Greatest Work, which is why the Arabs knew
that the light o f a star ... as it enters our atmosphere Ptolemys book as A1 Magisti, later known as Alma-
... and penetrates its lower and denser layers, must gestum in Latin, and, finally, into Almagest ([65],
eventually become curved or refracted. As a result, the page 64).
star will appear closer to the zenith as seen by the ob
server ... than it is in reality ([65], pages 64-65). 7.
It is however unclear whether or not the author of COPERNICUS
Optics could calculate refraction as a stellar lati
tude function. On the other hand, it is known that We shall select just a few necessary facts from the
Walther was the first to successfully attempt an in entire body o f available materials associated with Co
troduction o f atmosphere refraction compensation pernicus. Nicolaus Copernicus (1473-1543) is one of
... which Ptolemy could barely conceive o f ([65], the greatest astronomers o f the Middle Ages and the
page 87). However, the character in question lived in author of the heliocentric theory. His ancient portrait
the XV century a . d . - Bernhard Walther, 1430-1504 can be seen in fig. 0.7, and another one in fig. 0.8.
([65], page 85). Incidentally, his name was transcribed in a vari
So how does one date Ptolemys Optics? The fact ety o f ways - by Copernicus himself as well as his

Fig. 0.7. Ancient portrait of Copernicus Fig. 0.8. Ancient drawing o f Copernicus on the Cosmosphere o f Vassily Kiprianov.
(1478-1443). Taken from [1160], page 310. Taken from [90], page 212.

Fig. 0.9. The heliocentric system o f the world according to

Copernicus, as drawn in the atlas o f Andreas Cellarius Fig. 0.10. Fragment. A drawing o f Copernicus from a 1661
(Amsterdam, 1661). Taken from [1160], page 9. atlas. Taken from [1160], page 9.

contemporaries. He would occasionally write his Solar System, thus creating a heliocentric system of
name as Coppernic, reserving the Latin form o f the the Universe, qv in fig. 0.9. In the lower right corner
name, Coppernicus, for his scientific works. Much we see a portrait o f Copernicus (fig. 0.10).
less frequently he used the form Copernicus ([65], Copernicus reports having encountered a passage
page 90). By the way, could the name Copernic be a in one of Ciceros works, which had reflected the opin
derivative o f the Slavic word for competitor, which ion o f Hecataeus that the Earth revolves around its
is s o p e rn ik ? In the epoch that preceded the estab axis on a daily basis. These ideas were inherited from
lishment o f rigidified grammar rules the letter C the Pythagoreans. Philolaus claimed that the Earth
could stand for both S and K. moved around a central fire. It is perfectly clear that
The name Sopernik is in perfect concurrence his stance is already heliocentric in nature. Therefore,
with the scientific side o f the matter - namely, the the ancient Pythagoreans and Philolaus must have
prominent scientist can be regarded as a competitor been contemporaries o f Copernicus, or, alternatively,
o f his colleague Ptolemy and the author o f a new his immediate predecessors.
conception and theory. The very concept o f compe The idea that the Earth might not be the only cen
tition usually implies a certain chronological propin tre o f motion and that Venus and Mercury could also
quity, if not actual contemporaneity, o f the com revolve around the Sun is believed to be an ancient
petitors. Egyptian theory, which was also supported by Mar-
A. Berry reports: The crucial idea associated with cian Capella in the V century a . d . Nicolaus Cusanus,
the name o f Copernicus, owing to which De Revolu- a more modern authority (1401-1464) similarly in
tionibus is one o f the seminal works in astronomi clined to believe in telluric motion, either wasnt no
cal literature par none but the Almagest and Newtons ticed by Copernicus or deemed important enough
Principia, is that, according to Copernicus, the visi ... It is noteworthy that Copernicus remains taciturn
ble motions o f the celestial bodies are, for the greater about Aristarchus o f Samos, whose ideas o f telluric
part, different from their true motions, reflecting the motion were defined perfectly well [see Chapter 11
motions o f the observer carried away by the Earth for more details - Auth.]. It is possible that the re
([65], page 95). luctance o f Copernicus to refer to such an authority
Copernicus places the Sun at the centre o f the as Aristarchus can be explained by the fact that the
i8 h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

catalogue with 1024 stars in it. Specialists in the his

tory o f astronomy tell us that the catalogue is basi
cally identical to the catalogue of Ptolemy, the main
difference being that the former counts the latitudes
off the Gamma of Aries and not the vernal equinox
point ([395], page 109). Therefore, the initial refer
ence point did not necessarily coincide with the ver
nal equinox in the XVI century, whatever the reason.
The practice of choosing a different point as the be
ginning of the coordinate system may also have ex
isted before the XVI century - in the epoch of
Ptolemy, for instance. Berry also informs us of the fol
lowing: Whenever there were discrepancies between
the Greek and Latin version o f the Almagest, caused
by the inattentiveness of the scribes or the printers,
Ptolemy would accept either version without trying
to verify both by new observations ([65], page 103).
Our book pays a great deal o f attention to the pre
cision of the observations carried out by different as
tronomers. It would therefore be expedient to cite
some data concerning the degree of precision that
Copernicus had aspired to achieve. As A. Berry points
out, We have become so accustomed to associate the
renaissance of astronomy... with the growing metic
Fig. 0.11. Ancient engraving dating from 1635, found on the ulousness o f observation fact collection, believing
title page o f De Systemate Mundi by Galileo Galilei. We see Copernicus to be the primary figure of the Renais
the ancient Aristotle and Ptolemy, likewise the mediaeval sance, that it would make sense to emphasise the fact
Copernicus, who had lived in the XVI century, drawn as con that he was by no means a great observer. His in
temporaries. Ptolemy is wearing a turban on his head. This is
struments were of his own construction for the most
how the artist o f the early XVII century saw things; consen
sual Scaligerian chronology should naturally deem this quite
part, and greatly inferior to the instruments of Nassir-
odd. A publication o f Leiden, Bon. and Abr. Elsevier, 1635. Eddin and Ulugbek [the astronomers of the Muslim
Titular etching. Taken from [35], page 58, sheet XXXII. period who lived in 1201-1274 a.d. and 1394-1449
a.d ., respectively - Auth.]. Moreover, they were even
worse than the instruments that he could have or
later was accused of heresy for his scientific views dered from the craftsmen o f Nuremberg, had it been
([65], pages 95-96). his intention; the observations of Copernicus were
According to A. Berry, the plan o f De Revolu- few (27 are mentioned in his book, and we know of
tionibus is similar to that of the Almagest in gen a dozen or two more from other sources), and he was
eral ([65], page 97). O. Neugebauer is perfectly cor apparently unconcerned with attaining a particular
rect to remark as follows: There is no better way to degree of precision. The positions of stars that he had
convince oneself that the astronomical science o f the measured, which served him as the primary source of
Middle Ages concurs to that of the antiquity than to reference and were therefore of the greatest impor
perform a comparative study of the Almagest... and tance, allowed for discrepancies of 40' - greater than
De Revolutionibus by Copernicus. The two works the visible diameter of the Sun or the Moon. Hippar
are parallel - chapter by chapter, theorem by theorem chus would doubtlessly consider a discrepancy of this
and table by table ([571], page 197). sort a grave error ([65], page 93).
The book o f Copernicus is concluded by a star In fig. 0.11 we see an old engraving from the title

Fig. 0.12. The title page from the Celestial Atlas by Doppelmaier. The ancient Ptolemy and the mediaeval scientists o f the
XVI-XV1I century (Copernicus, Kepler and Brahe) are drawn as contemporaries, or at least as scientists o f the same epoch, con
versing between themselves. Taken from [926], page 73.

page o f The Two Primary World Systems, a book by mediaeval artwork o f this sort literally and to see pre
Galileo Galilei that came out in 1635 ([35], page 58, cisely what they show us. It is very likely that the con
sheet XXXII). The early XVII century artist portrays sensual metaphorical interpretation o f such artwork,
three scientists here - the ancient Ptolemy and which fuses the antiquity and the Middle Ages to
Aristotle next to the mediaeval Copernicus. They are gether, is a mere consequence o f Scaligerian chronol
depicted as contemporaries involved in a discussion ogy, which arbitrarily ascribes certain mediaeval con
o f scientific problems. Today we are told that all such temporaries to different epochs, severing all possible
mediaeval artwork (which is rather plentiful, as a connections between them. Ptolemy, for example, has
matter o f fact) happens to be o f a metaphorical na been cast into deep antiquity, whereas Copernicus
ture. Modern historians interpret the conversation more or less retained his own epoch - the XVI cen
between Copernicus and the ancient scientists as a tury.
symbol used by the mediaeval artist in order to em As a matter o f fact, Ptolemys headdress looks just
phasise the spiritual proximity between the great sci like a turban (see fig. 0.11). Could it be the result o f
entists o f the past and present. This is why the three his being an Ottoman scientist? Ptolemy also wears
are portrayed side by side, conversing at ease (fig. the turban-like headdress in yet another ancient por
0.12). This may indeed be the case. And yet everything trait - see figs. 0.13 and 0.14.
we have learnt about Scaligerian chronology (see In fig. 0.15 we see an old piece o f artwork dating
C hron I and C hron2) implies the potential viabil from 1666. It is evasively labelled allegorical - his
ity o f a different version - namely, that we are to take torians have no qualms about writing such things as
20 I h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

Fig. 0.13. An ancient drawing o f Copernicus next to a map of

the Old World. We see a headdress semblant to a turban on
Ptolemys head. A drawing from the 1507 world map by
Martin Waldseemller (Martin Waldseemllers Weltkarte Fig. 0.14. A close-in o f a fragment o f the previous drawing.
von 1507,Abb. S. 114/115). Taken from [1 0 0 9 ],page 12. Taken from [1009], page 12.

Fig. 0.15. An ancient drawing o f Claudius Ptolemy (standing

on the left), and three famous mediaeval cartographers:
Gerardus Mercator (sitting in the centre), Jodocus Hondius
and Willem Blaeu (sitting on the right). Title page from the
Concise Atlas by Johannes Jansson. Amsterdam, 1666.
An engraving by J. Weisheer made after the drawing of
S. Webbers. Chisel. Once again, historians suggest these
characters (Ptolemy and the three cartographers o f the XVI-
XVII century) to be separated from each other by some
1300-1400 years. We see two muses next to Ptolemy. Taken
from [90], page 6.

Fig. 0.16. A close-in o f a fragment of the above picture. We see

a pair o f mediaeval spectacles on the face of the ancient
Ptolemy. It is most likely that in the XVII century people still
remembered Ptolemy as a scientist from the epoch o f the XIV-
Apud Jo a n n * Ja u o in j H rede's. Anno M tjij i . : XVI century. Taken from [90], page 6.

Fig. 0.17. An ancient drawing o f Ptolemy

observing the stars. Etching on wood,
1517. We see Ptolemy wear a royal crown
- a mediaeval one, which is rather re Fig. 0.18. A close-in o f the
markable. We see such trefoil crowns in fragment with the medi
many mediaeval coats o f arms. Taken aeval royal crown on the
from: Gregor Reisch, Margarita philo- head o f the ancient
sophica ... Basel: Michael Furter, 1517. Ptolemy. Taken from
Fig. 0.16a. M onk with spectacles ([497:1], page 35). Taken from [1009], page 21. [1009], page 21.

the Allegory o f Cartography and the prominent car that spectacles appeared in the XIII century the ear
tographers: Claudius Ptolemy, Gerhard Mercator, liest ([497:1], pages 34-35). Around the middle o f the
Judocus Hondius and Willem Blau ([90], page 6). XIV century spectacles were already a very common
Ptolemy is on the left surrounded by two muses. object - a fresco o f 1352 depicts a bespectacled monk
However, the fact that the XVII century artist had no ([497:1], page 35). We reproduce this drawing in
doubts about portraying the ancient Ptolemy and fig. 16a.
three other cartographers o f the XVI-XVII century In fig. 0.17 we see an old portrait o f Ptolemy that
side by side may very well mean that he was perfecdy dates from 1517 ([1009], page 21). Ptolemy is wear
correct in his doing so. By the way, we see the an ing a trefoil royal crown on his head (fig. 0.18). These
cient Ptolemy wearing spectacles, a typically medi crowns are virtually identical to the kingly crowns
aeval object (fig. 0.16). This drawing also emphasises worn by the Evangelical Wise Men as portrayed on
a rather personal detail - Ptolemy appears to be ad the mediaeval sarcophagus o f the Three Wise Men,
justing the spectacles that have slid to the tip o f his for instance (it is located in the famous Cologne
nose. Ptolemy may have worn glasses in reality, and Cathedral in Germany - see Chron6, Chapter 3). We
this rather characteristic trait o f his may have been can also see three crowns o f the same trefoil design
remembered by the mediaeval artist and reproduced adorning the mediaeval coat o f arms o f Cologne (figs.
on the drawing. We feel obliged to remind the reader 0.19 and 0.20). Mediaeval crowns o f this shape are en-
22 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

Fig. 0.19. A fragment o f a mediaeval map depicting the

German city o f Cologne, dating from 1609. The engraving
was made by Abraham Hogenberg. We see three royal crowns
o f the same shape as the one worn by the ancient Ptolemy.
Taken from [1228].

Fig. 0.20. A close-in o f the fragment with the coat o f arms of

Cologne with the crowns. Taken from [1228].

/ \
*tt 1
ItoJ'.'li 9) 3

Fig. 0.21. An ancient French miniature of

the Rhemish Missal dating to 1285-1297
(Missel & lUsage de Saint-Nicaise de Reims.
The royal crowns we see here are o f the
same shape as the one worn by Ptolemy. Fig. 0.22. A close-in o f the fragment with the royal crowns. Taken from
Taken from [537], page 207. [537], page 207.

countered in a great deal o f mediaeval artwork por

traying royalties and dating from the XIV-XVI cen
tury (in Sweden, for instance).
We see trefoil royal crowns that are perfectly sim
ilar to the above in mediaeval French miniatures (such
as one may find in the famed Rhemish Missal created
between 1285 and 1297, for example). See [537], pages
194 and 207; also figs. 0.21 and 0.22.
Therefore, we see the ancient Ptolemy wearing
a famous mediaeval crown on his head. See more on
the history of the trefoil crown o f the Great = Mon
golian Empire in Chron7, Chapter 15:2.


Tycho Brahe (1546-1601) was one of the most

renowned astronomers of the Middle Ages, a profes
sional scientist who played a major part in the de
velopment of fundamental astronomical conceptions.
On 21 August 1560, in his second year at the Univer Fig. 0.23. An ancient portrait of Tycho Brahe. Taken from [1160],
sity of Copenhagen, there was a solar eclipse observed page 310.
as partial in Copenhagen. Tycho Brahe was aston
ished by the fact that this event had been predicted
earlier ([395], page 123). This event impelled Tycho November 1572 Tycho Brahe noticed a bright star in
Brahe to develop a deep interest in astronomy. the constellation of Cassiopeia, which hadnt been
An old portrait of Tycho Brahe can be seen in fig. there before. He instantly started the angular dis
0.23. In fig. 0.24 we see an old engraving that portrays tances between this new star and the main stars of
Tycho Brahe, his colleagues and his famous quad Cassiopeia as well as the North Star. Somewhat later,
rant. In fig. 0.25 we reproduce another version o f the Kepler wrote: Even if this star wasnt really an omen
very same engraving in order to draw the readers at o f any sort, it has heralded and made a great as
tention to the rather liberal manner in which the tronomer at the very least. The Tychonian super
copyists treated old artwork. The two versions strike nova was brighter than Venus, and could be seen for
one as identical at first sight; a more in-depth study 17 months with the naked eye, even in the daytime.
reveals substantial discrepancies. They lead to no con We are told that in 1576 King Frederick II of Den
fusion in this particular case, but the very fact that mark and Norway bestowed the Isle o f Hven near
mediaeval copyists did not deem it necessary to re Copenhagen upon Tycho Brahe. He also invested a
produce originals faithfully leads one to certain con large sum of money into the construction of the
clusions. Uraniborg observatory there - the name translates as
In 1569 Tycho Brahe was in Augsburg, the resi The Castle o f Urania. We shall discuss the possible
dence o f the craftsmen who manufactured instru true location of this observatory below, in Chapter 10.
ments o f sufficiently high precision for the observa It was most likely at a considerable distance from
tion o f celestial bodies. This is where Tychos quad Copenhagen. The observatory was equipped with
rant and sextant were made, followed by another precise angular instruments. Several years later, the
quadrant with a radius o f circa 6 metres. The full observatory o f Stjerneborg (Star Castle) was built.
height o f this instrument equalled 11 metres, and it All the measurement instruments were installed un
could count angles with the precision o f 10. On 11 derground so as to protect them from environmen-
24 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

Fig. 0.25. Another version (?) o f the old engraving presented

in the previous figure. Tycho Brahe and his quadrant. Mark
the fact that these two drawings differ from each other some
Fig. 0.24. An ancient drawing o f Tycho Brahe and his famous what; nevertheless, each o f them is declared to be original
quadrant. Taken from [1160], page 311. nowadays! Taken from [1029], page 24.

tal disturbances o f any kind. The Isle o f Hven be cision o f observations allowed for calculation preci
came a unique astronomical centre of global impor sion margin o f 10" or less (5" in case o f the wall
tance, and remained one for over 20 years. This is quadrant). The latter required 3 people for opera
where Tycho, accompanied by his apprentices, con tion - one to watch the celestial sphere and record the
ducted observations o f exceptional and unprece height o f the celestial object under study, another to
dented precision. Unique astronomical instruments write the data down in a journal, and yet another
were manufactured there as well ([395], page 126). person to record the time o f meridian crossing with
Diagrams and descriptions o f Tycho Brahes pri the aid of several chronometers, no less, installed right
mary instruments were published in his book enti there in the observatory (see figs. 0.24 and 0.25). In
tled The Mechanics o f Updated Astronomy (pub 1581 Tycho Brahe used a chronometer with an extra
lished in 1598). First and foremost, Tycho used quad hand for seconds, estimating their precision margin
rants with radiuses o f 42, 64 and 167 cm. The most as four seconds.
famous of all was the 194-centimetre quadrant, whose Another group o f instruments comprised the sex
arc o f cast brass was rigidly affixed to the eastern wall tants. Tycho Brahe oversaw and directed the manu
o f the observatory (precisely oriented at the North facture o f several armillary spheres. One must men
and the South). Special techniques o f raising the pre tion a large globe o f 149 centimetres in diameter,

Fig. 0.26. A diagram o f the Universe according to Tycho Fig. 0.27. A diagram o f the Universe according to Tycho
Brahe, taken from the atlas by Andreas Cellarius of Brahe, taken from the atlas by Andreas Cellarius of
Amsterdam and dating to 1661. Taken from [1058], page 20. Amsterdam and dating to 1661. Taken from [1058], page 20.
Left half o f the map. Right half o f the map.

whose surface was covered by thin sheets o f brass and onian cosmology taken from the atlas o f Andreas
depicted the Zodiacal belt, the equator and the posi Cellarius published in 1661 in Amsterdam ([1058],
tions o f 1000 stars; their coordinates were calculated page 20). We see Tycho Brahe in the lower right cor
over the many years o f Tychos observations. He was ner (fig. 0.28).
proud o f his creation, claiming No globe o f this size, This phase of success ended rather abruptly. Chris
manufactured with as much diligence and finesse, tian IV, the new King o f Denmark, expropriated
has ever been made anywhere in the world to the best Tycho Brahes estates, which had been providing him
o f my knowledge ... Alas, this true miracle o f sci with the funds necessary for maintaining the obser
ence and art was destroyed in a blaze in the second vatory in a constant state o f functionality. In 1597
half o f the XVIII century ([395], page 127). Tycho left Denmark and eventually setded down near
According to the evidence o f Tychos contempo Prague, founding a new observatory there. Johannes
raries, his work stamina was just as amazing as the Kepler began his career as Brahes apprentice (see fig.
meticulousness o f his scientific research. He checked 0.29). On 13 October 1601, Tycho Brahe fell ill and
and re-checked the results o f numerous observations died on 24 October 1601 at the age o f 55. The fa
personally, striving to bring them to perfection. In mous Uraniborg observatory was destroyed com
figs. 0.26 and 0.27 we reproduce the diagram o f Tych- pletely - there isnt a single trace o f it in existence
26 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

star catalogue. Only 777 stars had been observed and

measured properly, and so Tycho made haste to reg
ister all the rest o f the stars, wishing to add to the tra
ditional number ([65], page 126).
Let us linger on the precision o f Tycho Brahes ob
servations for a while. In the epoch o f Copernicus, a
single measurement step equalled 10' - just like it did
in the Ptolemaic epoch, since 10' also constitute the
value o f the Almagest precision margin. It is believed
that Tycho Brahe managed to make the measure
ments o f the equatorial star coordinates some 50
times more precise - namely, the average precision
margin o f the coordinates o f eight referential stars
measured by the wall quadrant equals 34.6" (33.2" in
case o f the astronomical sextant). This level of preci
sion is believed to be close to the theoretical possible
precision limit for any astronomical observations
conducted before the invention o f the telescope
([395], pages 128-129).
However, such great precision o f equatorial stellar
coordinate measurement was compromised by the
transition to the ecliptic coordinate system, which re
quires the knowledge o f the angle between the eclip
tic and the equator. Tycho Brahes calculations o f this
Fig. 0.28. A fragment o f the previous illustration depicting angle yielded the figure o f = 23 31' 5", which ex
Tycho Brahe. ceeds the true value by 2'. This can be explained by the
fact that Tycho corrected his star declination meas
urements taking refraction and solar parallax into ac
today. Alternatively, it could have been located in an count. Following Aristarchus of Samos, he accepted
altogether different place (see Chapter 10). the theory [? - Auth.] that the distance between the
In 1671 Picard went to Denmark in order to find Earth and the Sun was 19 times greater than that be
out about the fate o f Tycho Brahes observatory on tween the Earth and the Moon, which makes solar
the Isle o f Hven. Picard found a pit filled with rub parallax equal 1/19th of the lunar parallax, or 3'. Tycho
bish where the magnificent castle had formerly stood, wrote the following in this respect: the ancients ap
and was forced to conduct excavations in order to lo pear to have carried out the measurement in question
cate the foundation ([65], page 181). Thus, a great with enough attention to detail for us to adopt the end
deal o f information about the life and work o f Tycho value as sufficiently reliable. He made a mistake,
Brahe has been lost, notwithstanding the fact that he though . . . ([395], page 129).
didnt really live all that long ago. The was hardly any Thus, the precision margin o f the ecliptic stellar
one to use the large instruments o f Tycho after his coordinates in Tycho Brahes equals 2' or 3'. We shall
death - most o f them perished in the epoch of the Bo confirm this result independently, using our cata
hemian civil wars. Kepler managed to obtain a copy logue dating method; in particular, it allows us to es
o f Brahes observation records, but they were raw and timate the real precision o f star observations as con
unedited. Publications were few and far between ducted by the ancients.
([65], page 127). As we learn from A. Berry, obviously enough, the
It is believed that around 1597-1598 Tycho Brahe true precision o f Tychonian observations fluctuated
distributed some handwritten copies o f his 1000- significantly, depending on the character o f the ob

servation, the diligence o f the observer, and the pe

riod o f Tychos life when the observation was carried
out. The discrepancy between the coordinates o f the
nine stars that form the basis o f his star catalogue
and their equivalents yielded by the best modern ob
servations is smaller than f in most cases (equalling
2' in case o f just a single star). This error was caused
by refraction primarily - Tychos familiarity with the
latter phenomenon could not have been anything but
perfunctory. The positions o f other stars must have
been measured with less precision. Still, we shall
hardly be that much off the mark if we assume that
in most cases the precision margin o f Tychos obser
vations did not exceed 1' or 2'.
According to one o f the most frequently quoted
passages o f Keplers oeuvre, errata o f 8' were com
pletely out o f the question for Tychos planetary ob
servations ([65], page 128).
A. Pannekuk reports: Tycho estimated the direct
ascensions and declinations o f his referential stars,
totalling 21, with the greatest precision; the mean Fig. 0.29. An ancient portrait o f Johannes Kepler. Taken from
error value is less than 40" as compared to modern [926], page 69.
data ([643], page 229).
A. Berry suggests the following reasons why Tycho
Brahe was the first to attain a sufficiently high level to estimate the value o f this shift for different parts
o f observation precision: To a certain extent, such o f the celestial spheres. He came up with a rather
precision can be explained by the size and the excel rudimentary refraction table as a result, and made
lent construction o f his instruments - this is some regular refraction compensation an integral part of
thing that the Arabs and other observers had always all his further observations ([65], page 129).
sought to achieve. It goes without saying that Tycho Apart from that, Tycho Brahe accounted for the
used brilliant instruments - however, they became a parallax effect. He was among the first scientists to
great deal more efficient in his hands for two rea appreciate the full importance o f numerous repeti
sons, the first being his innovative use o f minor me tions o f the same kind o f observations under vary
chanical accessories, such as special dioptres or par ing conditions so as to make all the assorted random
ticular kinds o f horizontal gradation, and the second, errata introduced by individual observations neu
the fact that the motion range o f his instruments was tralise each other ([65], page 129).
very limited, which would substantially enhance their All the above facts demonstrate that Tycho Brahe
stability as compared to the devices that can be di was a perfectionist and a very meticulous astronomer
rected at any part o f the celestial sphere. o f great professionalism. This makes the following
Another great improvement was his systematic circumstance, pointed out by A. Berry, as well as many
compensation o f the inevitable mechanical imper other authors, seem very odd indeed: Unfortunately,
fections that affect even the best o f the instruments he did not measure the distance to the Sun, accept
as well as the more constant errata. For example, it ing the veracity o f the extremely rough estimate that
had been long known that the refraction o f the light had remained unaltered since the very epoch o f Aris
in the atmosphere makes the stars seem located tarchus, passing from one astronomer to another
somewhat higher than they really are. Tycho endeav ([65], page 130). From the consensual point o f view,
oured to carry out a series o f observations in order this institution o f astronomical heritage must have
28 I h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

been about two thousand years old in the epoch of P H A E N O M E N A .

mftdo obUquutdreulus cui adfcriptar font liter* a b c fc.p duod rat U & innft,
Tycho Brahe. If he did in fact consider this informa i t fubeo ainorbts par quosejrdem dtuiftoiies lirwarum dtcuntur, Quomodo ttutf
fubfigm spuottx progredUntur quarqut urn alia dixuaus fieriin muudo tx h o *
tion ancient, why didnt he verify it, being the bril fchenuoc non eft difficue djfcere.

liant professional that he was? It would be all the ^ D e rpbxrar conuerfione 8Caxe SCp olts Si circuits pr*ctpuif in raund*.
OW perios rnilm potuimus brcuiflim^mund per prarcipuas parte* aperubMU,
more natural given that he had made corrections jj^hinc um aeinceps,polos,& circulos deferibemus quaj n tc c fie e ft in ffh f'
rabarbarica propter fteiiarum loca difpouenda deferibere. Eft aurem iamtocomw
and new measurements to define nearly every astro fomundiimipfutn viMuerfumprxcipuasty eius partes eficfphgras;atq g lo b dupU
cemefte ration rotundttas motum,a!ium quo per plauiciem voluirur, confkien
nomical value that was of any importance at all ( [65], direftum fpanutn locum e loco mutns,quo quidem neqn mundus.ticque vila pars
cius mouetnr, alias eft quo in eodem quidem loco trancin', virtirur circa axemper
mud traic&um,cuiusextremitate*diaitttrpoli.Hoc modo veria fphera,paries qua?
page 129). fumvicinafpo!is,8,tardiuerunnir8Cbreujoreambisn: ipfi autcmpoli(qutLart-
ne i C icerone vertices dicuui)ornninotUMiroouemur,itc in caeli region fcprewri-
In fig. 0.30 we see a page from a 1537 edition of ooati vbl plauitrum eft 3 CCmfura,fydera wrdccircumferutr me occafu idiquo,
SCqiqdam ftellx ne locum quidem murare videuruJVpvoptcreaquod ibi plus eft#
the Almagest. Pari modo contra hunc per dire&am Itneam in iubterraneo C3?liloco,alter polot
eftnttnq nobis confpicuusEa autem lines per mundum dufta duobus tu loci* de
termination itta polos oftendehs>fuperiorem qui ab vi fa ar# icos dcimr,altima
fidterra contrarb fttuawar&icum,axis ntuttdt vocatur,qui defmirurmfidi dime*
9. liens circa quam voluitur.Huiusexti ema poiidtetttttovSt Latin vertices vefam in
t o feberoate quod fequitur,!inea a b, axis eft,a vcropoloip vnus,b aut alter.Cir
IMPORTANT RESEARCH OF THE ALMAGEST (ulorum vero qui in carlo um,ad ofteudendaphaomena,qudam deferibunf c-
MerKmemdi per fuoaxc.omms enim nta & quae naturaiirtr eft in caelo,8i quf
BY THE ASTRONOMER ROBERT NEWTON f lt cogitatione conctpttur ex curfusfui quafi veftigio, rirculutn deferibere inteRi-
gbuTtOtcunturcphi gracce "^ ^ '.latin c aequidiftantes>propterea quod vnaquarli-
AND HIS BOOK ENTITLED "THE CRIME OF Set eorum par1 8 polis SCab aequidiftSribus eodem interuailo abfim. Horum pri
mtiKvtrptcronf exordiu fumamuOarfticus eft circa polum mundi fepremreo
CLAUDIUS PTOLEMY" ttcmdbftusdnteruafto 24. propemodumparnumfemperfupcrterrammari*,
-* inrra cui* ambitum a-
ftracomprehenfa mlq
occidunt.Secundui eft
circulusmaiore inter-
We shall occasionally compare our results to the ualio circa polum ,di-
tfusxftiuus feu tropi
results of Robert Newtons fundamental scientific re cus cancrt,vel circulus
folftitialis. Cum cnim
search of Ptolemys Almagest ([614]). A portrait of fol quim plurimum ac
cefiit ad feptentrioucs,
Robert Newton can be seen in fig. 0.31. r% hunc deferibit die v i-
! delicti fblAlttal r.Xerti-
Robert Newton (1919-1991) was a prominent us eft arquinoftialu in
ter duosjpolospermc
American scientist. Let us cite some facts about him diumcarlu deferiptus,
lub quofol eurm,dum
from the official obituary of 5 June 1991 (died 2 June ttoftes dtebus facit pa*
1991 in Silver Spring, MD, USA). He was a scientist dit foils maximS egref
ftonem ad
deferibit iol die bruma
o f international renown due to his research con Ht& diftans ab xqua-
A uj torever
cerning the shape and the motion o f the Earth ... He
was a specialist in the theory o f ballistics, electronic Fig. 0.30. A page from a 1537 edition of the Almagest.
physics, celestial mechanics and satellite trajectory
calculation. His career started in APLs Space Depart
ment in 1957, where he was put in charge of the satel late 1970s he also became involved in the research of
lite motion research ... He is to be credited with his the ancient astronomical records o f solar and lunar
fundamental contribution to the m ajor improve eclipses ... This research gave him a reason to doubt
ments in navigation precision ... He was head of the the information contained in the main oeuvre o f the
space exploration programme and the developer of famous astronomer Claudius Ptolemy and to accuse
the satellite navigation labs analytical aspects ... He the latter of fraud in his book, The Crime of Claudius
was the chief architect of the Navys Transit Satellite Ptolemy ... Among other things, R. Newton was the
Navigation System, which was developed in the lab Professor of Physics at the Tulane University and the
oratory in the 1960s. This navigation system is still University o f Tennessee, having also worked for the
used by more than 50.000 private, commercial and Bell Telephone Laboratory... and developed the mis
military vessels and submarines ... His research of sile ballistics at the Allegany Ballistic Laboratory,
satellite motion made it feasible to calculate the shape Cumberland.
o f the Earth with greater precision, which has re We believe it to be perfectly appropriate to voice
sulted in more precise measurements ... R. Newton our attitude towards the famous book of Robert New-
was a member o f the Ad Hoc Committee on Space t o n - The Crime of Claudius Ptolemy ([614]), since
Development Director Board and became the leader it has become rather controversial among the mod
o f APLs Space Exploration Group in 1959 ... In the ern authors of works on the history of astronomy. I.

A. Klimishin, for instance, writes the following about contain any original astronom
Newtons book in [395]: What we encounter here is ical observation data at all - the
an intent to prove that nearly the whole bulk of Ptol Almagest data are the end
emys observations, which constitute the foundation product of somebodys conver
o f the Ptolemaic theory of solar, lunar and planetary sions and calculations aimed at
motion, happens to be a fraud ([395], page 56). I. A. making the initial observation
Klimishin doesnt counter Robert Newtons conclu data fit another historical
sions with any ostensible astronomical or statistical ar epoch. Moreover, a substantial
gumentation, opting to abandon the factual discus part o f the observations in
sion of the issue altogether and contenting himself cluded in the Almagest also re
with the following statement: And yet the main rea sult from later theoretical cal
Fig. 0.31. A portrait
son for Ptolemys universal fame was his theory of culations included in the o f Robert Newton,
planetary motion, which had made it feasible to cal Almagest ex post facto as the the American scien
culate the positions of planets dozens of years into the observations of the ancients. tist (1919-1991).
future, no less! ([395], page 56). However, the value 3) The Almagest could not
of the Ptolemaic model can by no means shed any have been compiled in 137 a . d .,
light on the Almagest star catalogues compilation his which is the epoch that the ancient Ptolemys life
tory or indeed reveal anything about the origins of the time dates to in the consensual history of today.
Almagest in general. Similar statements of disagree 4) Consequently, the Almagest was compiled in
ment with the conclusions made by Robert Newton some other epoch and requires a new dating. Robert
(containing no counter-argumentation of any sub Newton himself has made the assumption that the Al
stance) have been voiced by a number of other spe magest was in need of extra age, or a shift back
cialists in the history of astronomy, such as Gingerich wards in time that would place it in the epoch of Hip
([1153]). parchus - circa the II century b . c ., that is. However,
In reality, the book of Robert Newton encapsulates this does not alleviate any of the fundamental prob
his fundamental research of the Almagest with the aid lems discovered by Robert Newton.
o f mathematical, astronomical and statistical meth 5) Robert Newton had initially agreed with the
ods. It contains a vast body of statistical material and consensual hypothesis about the Almagest contain
several deep conclusions that sum up many years of ing Ptolemys own claim that all of his observations
Robert Newtons labour. These results elucidate the were carried out by none other but Ptolemy himself
nature o f difficulties associated with the interpreta - allegedly around the beginning of the reign of Anto
tion o f the astronomical data contained in the Alma ninus Pius, a Roman emperor. The Scaligerian dat
gest. It has to be emphasised that Robert Newton ing of his reign is 138-161 a . d . Robert Newton makes
hadnt a iota of doubt about the veracity of the Alma the instant self-implied conclusion that Ptolemy was
gests consensual dating (which falls over the period lying as a result. Actually, we shall deal with the issue
between the II century b . c . and the II century a . d . ) . o f just how strongly the information contained in
Robert Newton was no historian, and he had to rely the Almagest implies that Ptolemy carried out all of
on the Scaligerian version of history, using it as the his stellar observations by himself during the reign of
chronological framework for his own research. The Antoninus Pius.
main corollaries o f Robert Newton can be formu In other words, Robert Newton opines that Ptol
lated as follows: emy himself (or somebody else acting on his behalf)
1) The astronomical environment that corre was a fraud, seeing as how the Almagest data are pre
sponds to the beginning o f the a . d . era (as calculated sented as the results of actual astronomical observa
with the aid o f modern theory) is at odds with the tions when they really owe their existence to conver
observation m aterial included in Ptolemys sions and theoretical calculations.
Almagest. As a serious and renowned scientist faced by the
2) The surviving version of the Almagest does not necessity o f voicing a number o f straightforward ac
30 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

cusations against Ptolemy or his editors, Robert New However, by the point that Id written the first third
ton remained uncertain about the best form of his sci o f this book, I have discovered the evidence that
entific results publication. At the very least, this is proves the crime in question to be rooted much
the motivation he voiced in a private missive to A. T. deeper that I had expected initially. The present work
Fomenko, which had concerned with the history of is therefore a collection o f old and new evidence to
the creation and publication o f his book ([614]) in testify to the commission o f the crime in question
1977 (R. Newton and A. T. Fomenko exchanged a few ([614], page 10).
letters about the problems o f historical chronology in Robert Newton concludes his book as follows:
the 1980s). However, Robert Newton has finally con This is a final summary o f results. All of Ptolemys
sidered his discovery o f the situation with the Alma own observations that he uses in the Syntax [the Al
gest important enough to obey his duty o f a scientist magest - Auth.] have turned out fraudulent, inas
and even use his accusations as the headers o f some much as their veracity could be tested. Many o f the
o f his books paragraphs ([614]). Let us quote some observations ascribed to other astronomers are also
o f them, since they really do speak volumes. part o f Ptolemys fraud. There are theoretical errata
a5:4. The alleged observations o f the equinoxes galore in his work, and it also reveals a lack of com
and the solstices by Ptolemy. prehension on the part o f the author ... His models
5:5. The fabricated solstice o f 431 b .c. (the solstice for the Moon and Mercury contradict the most ele
o f Meton). mentary observations and must be considered a fail
5:6. Ptolemys observations aimed at the estima ure. The very existence o f the Syntax has resulted in
tion o f the ecliptic declination angle and the latitude the loss o f many authentic works written by the as
o f Alexandria. tronomers o f Greece - we have ended up with the
6 :6 . Four fabricated lunar eclipse triads. legacy o f a single solitary model, and we even lack so
6:7. Proof o f fraud. much as the certainty o f whether this contribution to
6 :8. The culprit. astronomical science can actually be attributed to
7:4. Fraudulent calculations and miscalculations. Ptolemy at all. I am referring to the equant model,
10:5. The falsification o f data. which was used for Venus and the external planets.
11:5. Falsified data concerning Venus. Ptolemy greatly diminishes its value by a somewhat
11:6. Falsified data concerning the external plan improper application o f the model in question. It is
ets ([614], pages 3-5). becoming perfectly clear that no statements made by
In the very first lines o f his foreword to [614], Ptolemy can be accepted at face value, unless they are
Robert Newton says the following. This book tells the confirmed by independent authors unaffected by
story o f a certain crime against science. I am neither Ptolemys influence. All the research based on the
referring to carefully planned criminal activity o f any Syntax must be started from scratch once again, be
sort, nor indeed to the kind of crime committed with it historical or astronomical.
the aid o f such devices as hidden microphones, mes I am yet unaware o f the other peoples possible
sages ciphered in microfilm, and so on. I am referring opinions; still, I can make but a single final judgement:
to a crime committed by a scientist against his learned the Syntax has turned out more detrimental to as
peers and apprentices and a betrayal o f professional tronomy than any other book ever written, and the
integrity and ethics - a crime that has forever de astronomical science would benefit greatly, had this
prived humanity o f certain fundamental informa book never existed.
tion pertaining to the most crucial fields o f astron Therefore, Ptolemy is by no means the greatest as
omy and history. tronomer o f the antiquity, but rather an even odder
I have demonstrated that the crime in question figure: he is the most successful con man in the his
was indeed committed in four o f my previously pub tory o f science ([614], pages 367-368).
lished works ... When I began my work on this book, A number o f other scientists are also rather scep
my objective had been to collect the materials scat tical about the part played by Ptolemy in the history
tered across several publications into a single b o o k ... o f science. In particular, A. Berry relates the follow

ing: There is a great deal of controversy in what con served in the epoch of 137 a.d . The research results
cerns the astronomers opinions o f Ptolemys merits. o f Robert Newton and Dennis Rowlins can be found
In the Middle Ages, his astronomical authority was in [1119] and [1120].
considered decisive ... Modern critics have discovered Furthermore, such works as [1119], [1120] and
the fact that Ptolemys works were largely based on [1182] address the issue o f the southernmost
those o f Hipparchus (actually, Ptolemy never made Almagest catalogue stars waning brightness. The mat
any secret o f it), and that the results o f his own ob ter is that the stars that arent elevated sufficiently
servations, if not de facto fraudulent, are largely sub high above the horizon lose a lot o f their luminosity,
standard at the very least ([65], page 72). due to the fact that the human line o f eyesight ap
Therefore, Robert Newton has proven the neces proximates the surface o f the Earth. As a result, the
sity o f re-dating the Almagest - astronomically as ray travels further in the atmosphere than in case of
well as mathematically. This leads us to the following the stars situated further away from the horizon. This
question - which epoch does the Almagest really per is why the southern stars appear dimmer to the ob
tain to? As we have mentioned earlier, Robert Newton server than they really are. Our analysis o f the south
himself suggests moving it backwards in time - to the ernmost Almagest stars luminosity has revealed that
epoch o f Hipparchus. Other points o f view are also the observations of these stars were carried out some
viable; we shall discuss them in detail below. At any where far in the south. In particular, these consider
rate, Robert Newton does not discuss the problem of ations rule out the very possibility that Ptolemy per
dating or even address it. Is it at all possible to find a formed his observations anywhere in the vicinity of
historical epoch that would fit the Almagest better the Isle o f Rhodes, which happens to be the consen
and effectively solve the problems discovered by Rob sual localization of his observation point ([1182]).
ert Newton, as well as the earlier researchers, no mat Alexandria in Egypt fits somewhat better - yet we
ter how distant from the Scaligerian dating of the Al shall find out that even Alexandria does not quite sat
magest? As we shall see further on, Robert Newtons isfy to the stipulations o f the Almagest data. The lu
suggestion to mitigate the controversy by means of minosity estimate o f the southernmost stars implies
shifting the Almagest backwards in time (into the an even more austral latitude.
epoch o f Hipparchus, that is) doesnt lead us any We must also note that the coordinates o f the stars
where. This is why we have to ask the obvious ques in question are measured exceptionally badly, with
tion o f whether there may be other possible shifts o f discrepancies o f several degrees, qv below. If the Al
the Almagest dating to consider - possibly, amount magest is indeed a product of the late Middle Ages, this
ing to longer periods than 200 or 300 years. This ques circumstance is easy enough to explain. Apparently, the
tion o f ours is perfectly justified from the mathe austral stars were added to Ptolemys catalogue as a re
matical and astronomical point o f view, and finding sult of observations carried out somewhere far in the
a correct answer is nothing short o f a duty from the South - possibly, India, and not Alexandria, or the
independent researchers point o f view. deck of a ship sailing the South Atlantic. The lumi
The publications o f R. Newton were followed by nosity o f the stars was measured correctly, though,
a work o f Dennis Rowlins ([1365]), wherein he uses unlike their coordinates. This may be explained by
an independent method to prove that the stellar lon the possible imperfections o f the southern observa
gitudes contained in Ptolemys catalogue have been tories, or a poor concurrence o f different observato
recalculated and altered by someone. In other words, ries data. Finally, if the southernmost stars were in
D. Rowlins claims that the stellar longitudes that we deed observed from some vessel, the low precision of
find in Ptolemys catalogue could not have been ob the end result is even less of a mystery.

Some necessary information

related to astronomy and history
of astronomy

1. the motion of the Earth in relation to the distant stars

THE ECLIPTIC. THE EQUATOR. and consider it the immobile centre o f the stellar
PRECESSION sphere. Our further references to celestial objects such
as the Sun, stars etc shall imply the identification of
Let us consider the motion o f the Earth along its said object with the point o f its projection over the
solar orbit. It is usually considered that it isnt the sphere of immobile stars.
Earth itself that rotates around the Sun, but rather the The ecliptic rotates with time, which is why it is
mass centre (gravity centre) o f the Earth-Moon sys known as the mobile ecliptic. In order to refer to the
tem, or the so-called barycentre. The barycentre is position o f the mobile ecliptic at a given point in
relatively close to the centre of the Earth as compared time, let us introduce the concept of instantaneous
to the distance between the Earth and the Sun. The ecliptic for a given year or epoch. The conception and
stipulations o f the present work allow us to consider the properties of instantaneous spin vector pertain to
the orbital motion of the barycentre around the Sun the discipline o f celestial mechanics. Fixed successive
identical to the orbital motion o f the Earth itself. instantaneous ecliptics for different epochs are some
Gravitational perturbations caused by planets cause times referred to as fixed ecliptics of said epochs. For
constant rotation o f the barycentre orbit plane. This instance, it is convenient to refer to the fixed ecliptic
rotation contains a certain primary sinusoidal com for 1 January 1900. The position o f the mobile eclip
pound with very high periodicity. It is complemented tic for any given point in time can be specified in re
by certain minor variable fluctuations, which we shall lation to a randomly chosen fixed ecliptic.
ignore. This rotating orbital plane of the Earth is called The Earth is considered a perfectly solid body in
the ecliptic plane. celestial mechanics. It is well known that a solid body
Sometimes the term ecliptic is used for referring possesses a so-called inertia ellipsoid, which is rigidly
to the circumference where the ecliptic plane crosses defined by its three semi-axes. The rotation of a solid
the imaginary sphere o f immobile stars. Let us as body is characterised by the value and the spatial at
sume that the centre of this sphere coincides with the titude o f spin vector CD. Vector CDis sometimes re
centre of the Earth that lies on the ecliptic plane. In ferred to as the instantaneous axis o f rotation. The
fig. 1.1. it is indicated as point O. We can disregard semi-axes o f the inertia ellipsoid are orthogonal, and

The North Pale and Around the Sun ([ 1295]) compiled by the em
inent American astronomer Simon Newcomb are
used in order to account for all the irregularities in
herent in the motion of the Earth.
The study o f cases (solid body configurations)
when the equations o f Euler-Poisson can be solved
with sufficient precision comprises an important area
o f modern theoretical mechanics, physics and geom
Let us consider vector co of instantaneous Earth ro
tation. It defines the axis o f rotation, or the instanta
neous rotation axis. The points where it crosses the
surface o f the Earth are known as instantaneous poles
o f the Earth, whereas those where it crosses the ce
lestial sphere, or the sphere o f immobile stars, are
Fig. 1.1. The sphere o f immobile stars. The ecliptic and known as celestial poles (North and South). Let us
equatorial coordinate systems. consider the plane orthogonal to the instantaneous
rotation axis o f the Earth that crosses the mass cen
tre o f the Earth. Its intersection with the surface of
can therefore be used as an orthogonal system o f co the Earth is known as the instantaneous equator of
ordinates. Thus, vector co can be defined by the pro Earth rotation, and the intersection with the celestial
jections o f x, y and z over the axes of inertia. The mo sphere is referred to as the true celestial equator, ce
ments of body inertia relative to these axes shall be lestial equator or equinoctial.
indicated as A, B and C, respectively. The rotation of Fig. 1.1 depicts the celestial sphere. Its centre is
a solid body is described in the dynamic equations of marked O. P stands for the North Pole of the ecliptic,
Euler-Poisson: and N - for the celestial pole. The ecliptic and the
equator have two intersection points, which are known
Ax + (C - B)yz = M a as the vernal and autumnal equinox points (indicated
By + ( A - C)xz = M b as Q and R in fig. 1 . 1 , respectively). The illustration
Cz + ( B - A ) x y = M c also demonstrates the alterations of the stars coordi-

In the right part o f the equations we have the pro

P recesm n
jections o f vector Ai, known as the external couple in and nutation
relation to the mass centre, over the same axes.
Moment Ai results from the effect o f solar and lunar
gravity on the ellipsoidal figure of the Earth. The Earth
is usually considered a two-axial ellipsoid rather than
triaxial - an ellipsoid o f revolution, in other words.
The position o f vector Ai in relation to the axes o f
inertia changes rapidly, and these changes are o f a
rather complex nature; however, the application of
modern theories of lunar and telluric motion makes
it feasible to calculate its evolution with sufficient
precision for any moment in time. This allows us to
solve the equation o f Euler-Poisson, or calculate the
The ecliptic
evolution o f vector co.
The Tables o f the Motion o f the Earth on its Axis Fig. 1.2. Precession and nutation.
34 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

nates in relation to the two coordinate systems of the the ecliptic as seen from the side of its north pole. The
celestial sphere - equatorial and ecliptic. precession vector is directed at the ecliptics South
Let us now consider a coordinate system that Pole.
would not rotate together with the Earth, but be based 2 . Constant velocity component 0 approximates
on the ecliptic instead. The new coordinate system 0.5" per year as o f today.
does not have to be orthogonal. The following axes 3. Constant velocity component cp is the average
are normally used for such coordinate systems: proper Earth motion velocity value with the perio
1 ) normal to the ecliptic plane; dicity of one day anticlockwise around axis AC (as
2 ) the intersection axis o f the ecliptic and equa seen from the North Pole o f the Earth).
torial planes, or the equinoctial axis; Let us note that axis OP, which is the normal to
3) inertia axis C. wards the ecliptic plane, belongs to the same plane as
The projections o f instantaneous angular velocity vector 0), which represents the instantaneous angle ve
vector (0 over these three axes are indicated as \j/, 0 and locity of the Earth, and axis OC, or the third axis of
(p. We have thus expanded the Earth rotation rate inertia. This plane rotates around axis OP due to pre
into three components. What is their geometrical cession.
meaning? The value o f \j/, is known as the Earth pre Nutational components inherent in velocities (\j/,
cession rate. This component defines the circular con 0 and cp) distort the above model - therefore, vector
ical motion of precession axis C, or the third axis of co doesnt follow an ideal conical trajectory, but a
inertia, around the normal OP, as shown in fig. 1 .2 . rather erratic wavy one instead, which approximates
Vector co = ON follows this conical rotation. Let us the shape of a cone. The trajectory of the vectors end
point out the close proximity of vectors co and OC. point is drawn as a wavy line in fig. 1 .2 .
For approximated calculations we can assume vector The two circumferences that pertain to the celes
COto coincide with axis OC. tial sphere (the equator and the ecliptic) intersect at
Owing to precession, the equinox axis, or the in the angle of e = +2327' in two points - Q and P, qv
tersection of the ecliptic and the equator, rotates within in fig. 1 . 1 . The Sun crosses the equator twice in these
the ecliptic plane. The rotation of 0 affects the incli points over the course o f its annual voyage along the
nation o f axis OC towards the ecliptic to a certain ex ecliptic. Point Q, which is where the Sun enters the
tent. Finally, the value of cpdefines the rate of the Earths Northern Hemisphere, is the point of the vernal equi
rotation around axis OC. In theoretical mechanics the nox. This is the point where the respective durations
value of cp is known as proper rotation rate. It is much o f daytime and night time equal one another every
higher than the angular velocities of \j/ and 0. From the where on the Earth. Point R corresponds to the au
point of view of theoretical mechanics, this circum tumnal equinox (see fig. 1 . 1 ).
stance reflects the fact that the stable rotation of a solid The mobile ecliptic is in constant rotation. There
body occurs around the axis that happens to be the fore, the vernal equinox point constantly shifts along
closest to the axis of maximal inertia moment, or the side the equator, simultaneously moving along the
shortest axis of the inertia ellipsoid. Let us remind the ecliptic as well. The velocity at which the equinox
reader that the Earth is somewhat flattened at the poles. point travels along the ecliptic is the actual longitu
Thus, CQ= \j/ + 0 + cp(+ standing for the summa dinal precession. The shift o f the equinox points pro
tion of vectors). Each velocity (\j/, 0 and cp) contains duces the equinox precession effect (see fig. 1 . 1 ).
a single constant (or nearly constant) component as
well as a great many small periodic ones, commonly 2.
referred to as nutations. If we overlook them, we shall EQUATORIAL AND ECLIPTIC COORDINATES
come up with the following model o f Earth rotation.
1. Constant velocity component \j/ is called longi In order to record the observations o f celestial
tudinal precession. It moves axis OC along the circular bodies, one needs a convenient coordinate system
cone with the approximate annual velocity of 50" (see that would allow one to fix the respective positions
fig. 1 .2 ). The equinoctial axis moves clockwise along of celestial bodies. There are several such coordinate

systems - first and foremost, the equatorial coordi torial coordinates caused by precession conform to
nates, which are defined as follows. much more complex formulae, which account for
In fig. 1.1 we see the North Pole indicated as N and the orthogonal turn o f the ecliptic that connects it to
the celestial equator, which contains arc QB. We can the equator.
estimate the plane of the celestial equator to coincide It is for this very reason that mediaeval astronom
with the plane of the Earth equator, given that the ers tried to compile their catalogues with the use of
centre o f the Earth corresponds to point O, which ecliptic coordinates, notwithstanding that equatorial
stands for the centre of the celestial sphere. Point Q coordinates are easier to calculate by observations,
is the vernal equinox point. Let point A represent a since such calculations do not stipulate to define the
random immobile star. Let us consider meridian NB, ecliptic plane. The position of the ecliptic depends on
which crosses the North Pole and star A. Point B is the motion of the Earth around the Sun and requires
the intersection of the meridian with the equatorial the use o f sophisticated methods for its calculation,
plane. Arc QB = a corresponds to the equatorial lon which, it turn, lead to additional systematic errata in
gitude of star A. This longitude is also known as di the coordinates of all stars. The discovery of the fact
rect ascension. The direction of the arc is opposite that the ecliptic fluctuates over the course o f time led
to the motion of Q, which is the vernal equinox point. to the use of equatorial star coordinates in catalogues
Therefore, direct ascensions of stars attain greater val instead of the ecliptic system. This system is still used
ues over the course o f time due to precession. - the advantage o f the ecliptic system is a thing o f
Meridian arc AB = 8 corresponds to the equato the past.
rial latitude of star A, which is also referred to as the
declination o f star A. If we are to disregard the fluc 3.
tuations of the ecliptic, the declinations of the stars THE METHODS OF MEASURING EQUATORIAL
located in the Northern Hemisphere diminish with AND ECLIPTIC COORDINATES
time due to the motion of vernal equinox point Q.
The declinations o f the stars in the Southern Hemi Let us briefly consider a number of actual meth
sphere slowly grow with time. ods used for the estimation o f equatorial and eclip
The daily motion of the Earth does not alter the tic coordinates. We shall relate a certain simple geo
declinations o f the stars. Direct ascensions change in metric idea that such measuring instruments as the
a uniform fashion and are affected by the Earths ro sextant, the quadrant and the transit circle employ in
tation velocity. their construction.
The ecliptic coordinate is also rather popular, and Let us assume that observer H is located in point
it was used very widely in the ancient star catalogues. cp on the surface of the Earth (see figs. 1.3 and 1.4).
Let us consider the celestial meridian that crosses It is rather easy to define line H N ' that is oriented at
the ecliptic pole P and star A (see fig. 1.1). It crosses the celestial North Pole and the parallel line ON. Next
the ecliptic plane in point D. Arc QD corresponds to we have to define the meridian that crosses point H
ecliptic longitude Zin fig. 1 . 1 , and arc AD represents and mount a vertical wall on Earth surface that shall
ecliptic latitude b. Precession makes arc QD grow by go along this meridian, qv in figs. 1.3 and 1.4. Marking
circa one degree every 70 years, which results in the the direction of the celestial pole on this wall as HN\
uniform growth o f the ecliptic longitudes. we can also indicate the equatorial like HK\ which is
If we disregard the fluctuations o f the ecliptic, we parallel to OK, by means of laying an angle -y from
can consider ecliptic latitudes b stable as a first ap direction H N 1. Right angle N H K 1can be divided into
proximation. This is the very thing that made eclip degrees, which gives us an astronomical instrument
tic coordinates so popular with the mediaeval as for angular measurements - a quarter of a divided cir
tronomers. The advantage of the ecliptic coordinates cle positioned vertically. Modern meridian instru
over the equatorial ones is that the value of b is con ments are based on this instrument as well - it can
stant, whereas the value o f Zgrows with the course of be used for measuring star declinations, or their
time as a result of precession. The alterations of equa equatorial latitudes, and also for marking the mo-
36 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1


Fig. 1.3. The principle o f stellar coordinate measurement. Fig. 1.4. Measuring the coordinates o f a star that passes a

ments when stars cross a given meridian, or the so- o f stars one needs a timekeeping device o f some sort
called vertical. in order to compensate the daily rotation o f the Earth
A series o f independent consecutive measurements and keep the orientation at the equinoctial point
makes it feasible to estimate the equatorial plane for constant.
the latitude o f observation with high enough preci The obvious complexity o f this task led to the fol
sion. At the same time, as it is obvious from the above lowing: for actual calculations o f ecliptic coordinates
elementary celestial mechanics, a measurement of astronomers would either use formulae o f the celes
longitudes requires a fixation o f moments when the tial spheres rotation or celestial globes with equato
stars cross the meridian. This requires either a suffi rial and ecliptic coordinate grids. The knowledge o f
ciently precise chronometer, or an auxiliary device equatorial coordinates would allow calculating their
providing for fast measurements o f longitudinal dis ecliptic equivalents. Naturally enough, there were in
tances between the star that interests us and a fixed evitable errata resulting from lack o f sufficient pre
meridian. At any rate, longitudinal measurements are cision in the estimation o f the comparative positions
a substantially more subtle operation. Therefore it is o f the ecliptic and the equator, as well as the attitude
to be expected that mediaeval astronomers meas o f the equinoctial axis.
urements o f direct ascensions are cruder than their This very concise discussion o f methods used for
declination measurements. the measurement o f ecliptic coordinates permits the
In order to measure the ecliptic coordinates o f estimation that the mediaeval astronomers are most
stars observer H must assess the celestial position o f likely to have used the following algorithm:
the ecliptic first. This operation is sophisticated 1 ) They would calculate the equatorial coordi
enough and stipulates a good understanding o f pri nates, the latitudinal measurements being more pre
mary elements o f solar and telluric motion. Ancient cise than the longitudinal.
methods o f measuring the declination angle between 2 ) Next they would estimate the position o f the
the ecliptic and the equator as well as the position o f ecliptic and the equinoctial axis in relation to the
the equinoctial axis with the aid o f the armillary equator.
sphere or the astrolabe are described in [614] and a 3) Finally they would convert the equatorial co
wealth o f other sources. It has to be noted that in ordinates into their ecliptic equivalents with the aid
order to measure the ecliptic coordinates o f a series o f special measurement instruments or trigonomet

ric formulae (or, alternatively, with the use o f a ce which possesses the magnitude o f 1 in the Almagest,
lestial globe with a double coordinate grid). has the magnitude o f 0.24 in The Bright Star
Moreover, since all the ancient measurement tools Catalogue, a modern source ([1197]), and Sirius, also
were inevitably installed upon the surface o f the a star of the first magnitude in the Almagest, possesses
Earth, the above algorithm is the only real method of the magnitude o f- 1.6 in the modern catalogue. Thus,
calculating the ecliptic stellar coordinates. Since a Sirius is brighter than Arcturus, although Ptolemy be
measuring instrument installed on the surface of the lieved them to be equally bright.
Earth takes part in daily rotation o f the Earth, the in The matter might be that in the antiquity the
strument in question is invariably tied to the equa brightness (or the magnitude) o f a star was estimated
torial coordinate system. by the observer in a very approximated fashion. Now
The application of our statistical methods to the adays stellar magnitude is estimated with the photo
data provided by the Almagest catalogue yielded a metric method. A comparison o f stellar magnitudes
confirmation o f the above algorithms usage, as we contained in the Almagest to their modern precise
shall demonstrate below. values as given in the work o f Peters and Knobel
([1339]) demonstrates that the discrepancy doesnt
4. usually exceed 1 or 2 measurement units.
THE MODERN CELESTIAL SPHERE In our calculations o f actual positions o f stars in
the past we were primarily referring to the bright star
In order to date an old star catalogue by the nu catalogue ( [ 1197] ), which contains the characteristics
meric values o f stellar coordinates contained therein, o f circa 9000 stars up to the eighth stellar magnitude.
we must be able to calculate the positions o f stars on Let us remind the reader that one can only see the
the celestial sphere for various points o f time in the stars whose magnitude is up to 6 or 7 with the naked
past. The information that we use for reference is the eye. According to Ptolemys claim, the Almagest star
existing description o f the celestial sphere in its mod catalogue contains all the stars from the visible part
ern state. The only data o f importance are the coor o f the sky up to the 6th magnitude.
dinates o f stars, as well as their magnitude and proper Ptolemy was exaggerating - there are more stars
motion rate. with magnitudes o f 6 and less in the visible part of
Jumping ahead, we can remark that the dating the sky than in the Almagest catalogue. This is one of
method that we suggest is only applicable if the re the reasons why the attempts to identify the Almagest
spective positions o f stars alter with the course o f stars with the stellar positions calculated in reverse
time. The rotation o f the entire celestial sphere re lead to ambiguities (see Chapter 2 for more details).
sulting from a transition to another coordinate sys On the other hand, it would be natural to assume
tem cannot be used for the purposes o f independent that all the stars that were actually observed by Ptol
dating. We shall discuss this in more detail below. emy or his predecessors still exist and can be found
Let us discuss the characteristics o f the stars that in the modern catalogue ([1197]).
we shall refer to in our research. J. Bayer, a prominent XVII century astronomer,
The magnitude o f a star in a modern catalogue is suggested a new system o f referring to stars in a con
the number that represents its brightness. The lower stellation. He suggested using letters o f the Greek al
the value, the brighter the star. There is an old tradi phabet instead o f a verbal description o f a given stars
tion o f indicating said values in star catalogues. The position in a constellation. The brightest star o f a
Almagest contains the magnitude values of all the stars constellation would be indicated by letter a , the sec
it lists. The brightest stars are indicated as the stars o f ond brightest one - by letter p, and so on. Later on,
the first magnitude, the less bright ones correspond Flamsteed (1646-1720) devised a special numeration
to the second magnitude and so on. Modern cata for stars in a constellation - more specifically, the
logues use the same scale for referring to the bright westernmost star o f a constellation was indexed as 1 ,
ness o f a given stars. However, stellar magnitudes can the next one to the east - as 2 , and so on. Flamsteeds
also be expressed as fractions. For example, Arcturus, numbers and Bayers letters are often used in combi
38 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

nation for referring to a star (32 a Leo and so on). The trajectories of stellar motion for the time in
Apart from that, some of the stars have individual tervals that interest us (2-3 thousand years) can be
names. Such named stars are comparatively rare - considered straight, which means that each of the
individual names were only assigned to stars that had stars coordinates on the celestial sphere change
special significance in ancient astronomy. For in evenly. This approximation is only valid for areas that
stance, 32 a Leo is called Regulus. lay at some distance from the pole, obviously enough.
We have used the following characteristics of stars The standard coordinate system for the celestial
from the modern catalogue ([1197]): sphere as given in the modern star catalogues is cus
1. Direct ascension o f a star for the epoch of 1900, tomarily based on the equatorial coordinates for the
which is transcribed as a 1900 below, expressed in epochs of 1900,1950 and 2000 a.d . We have chosen
hours, minutes and seconds. the system of equatorial coordinates for the beginning
2. The declination o f a star for the same epoch tran of 1900 a.d . Further calculations and coordinate sys
scribed as 1900 and measured in degrees, arc minutes tem conversions for a given epoch t were based on this
and seconds. system.
3. Stellar magnitude. First and foremost, in order to date the Almagest
4. Proper motion rate o f a given star. The proper catalogue we shall need the coordinates of stars with
motion rate is comprised of two elements, the first one high proper motion rates. Naturally, we shall only
being the star declination fluctuation rate and the sec consider the fast stars that are believed to be listed in
ond - the rate of its direct ascension alteration. How the Almagest.
ever, the coordinate grid of longitudes and latitudes on We have refrained from discussing the issue of
a sphere isnt uniform. The distances between adja whether or not the Almagest stars were identified cor
cent meridians diminish closer to the poles; therefore, rectly. We shall consider it in detail below. In order to
the stellar velocity component of direct ascension gives solve the identification problem we must know
one a wrong idea of the true, or visible velocity of a whether a given star had an individual name in the
star on the celestial sphere in the direction of the par ancient catalogues. The information about the me
allel. Therefore, some modern star catalogues give the diaeval names o f stars was taken from catalogues BS4
stellar velocity component of the direct ascension re ([1197]) and BS5 (online source).
duced to the equator. This means the value is multi In order to date the Almagest catalogue by proper
plied by the declination cosine, which makes it possi motion rates we shall require the following two lists
ble to interpret it as the local Euclidean length of the o f stars from the modern catalogues. We shall merely
stellar velocity vector projection over the equator (the describe them herein; the actual lists can be found in
parallel). This permits a comparison of the first stellar Annex 1 .
velocity components regardless o f their proximity to We shall refer to the first list as to the list o f fast
the pole. If the velocities arent reduced in this fashion, stars. In the first stage o f said lists compilation we
such comparisons require additional calculations. have selected all the stars whose speed by one o f the
Catalogues BS4 ([1197]) and BS5 (online source) coordinates at least is greater than 0. 1 " per year. This
that we have used, the velocities are reduced to the list was subsequently reduced to the stars that either
equator, which isnt the case with catalogues FK4 have Bayers Greek letter or Flamsteeds number in
([1144]) and FK5 (online source). Oddly enough, their name. Thus, we have rejected the stars that are
this fact isnt always mentioned in the descriptions of a priori useless for the dating for the Almagest. The
astronomical catalogues. The form o f the direct as matter is that nearly every star identified by the as
cension velocities has to be estimated from their ac tronomers as one o f the Almagest stars has an index
tual numeric values. in either Bayers or Flamsteeds system, or both; also,
The values of proper star motion rates are rather if a star from the Almagest is identified as one that
small. They dont normally exceed 1 " per year - the lacks such indices, this identification is always rather
fastest o f the stars visible to the naked eye, such as o 2 ambiguous ([ 1339]). The reason is clear enough. The
Eri, |i Cas, move at the rate o f 4" per year. catalogues of Bayer and Flamsteed were already com

piled in the epoch o f early telescopic observations, or culations quite a few times for different epochs. We
the X V II-X V III century. If a given star is omitted would first calculate the positions of stars on the ce
from those catalogues, it is either too dim or too dif lestial sphere for year tin coordinates a 1900 and 8 1900>
ficult to tell apart from the celestial objects in its im and then convert those into ecliptic coordinates lt and
mediate vicinity. bt for epoch t.
There may be other complications in the same Let us cite the necessary formulae that allow the
vein; therefore, one can hardly assume that a star of conversion of coordinates a 5 and 85 into coordinates
this sort can be veraciously identified as an Almagest ls0 and bs0 for any epochs s and s0. These formulae ac
star and that its position was measured with sufficient count for precession and proper star motion. Said for
precision by the ancient astronomers. mulae, as well as fig. 1.5, which illustrates them, were
The above selection gave us a list o f fast stars taken from [ 1222]. They are based on Newcombs the
visible with the naked eye, which can be found in ory as modified by Kinoshita. The actual coordinate
modern star catalogues and identified as Almagest conversion procedure is described in the next section
stars. Quite naturally, the veracity o f such identifica (5.2). In these formulae time moments s0 and s are
tions requires a separate research. We shall consider counted backwards from the epoch o f 2000 a.d. in
this problem below. Julian centuries, and 0 = s0 - s. See fig. 1.5.
Our list o f fast stars visible to the naked eye can
cp(5,s0) = 17452 '27,66* + 3289,80023% + 0,576264 "si?
be found in Table P l.l of Annex 1 . (1.5.1)
The second list o f stars is the list o f named stars. - (870,63478'' + 0,554988%)0 + 0,024578 " 0 ;
It is contained in Tables P1.2 and P1.3. In Table P 1.2 k(5,s0) = (47,0036"-0,06639% + 0,000569so)G (1-5.2)
the stars are arranged by names, and in Table P I.3 -
+ (-0,03320" + 0,000569% )9 2 + 0,000050'03;
by respective numbers from the Bright Star Catalogue
e 0(s,s0) = 2326'21,47"- 46,81559%
([1197]). This list contains all the stars which have in 2 3 (1.5.3)
- 0,000412% +0,00183%
dividual names according to BS4 ([1197]), or which
had such names in the past (Arcturus, Aldebaran, 6 iU i0) = o(s so) + (0,05130"-0,009203% )02 -0,007734"03;
Sirius etc). e(*Jo) = e 0(s, s0) + (-46,8156" - 0,00082% + 0,005489"sq)0
The lists of fast and named stars intersect - the + (-0,00041" + 0,005490 % )02 + 0,001830 "03;
same star can have a visible proper motion rate and
an individual name. Such stars are the most useful for iK ^o ) = (5038,7802" +0,49254% -0,000039"sq)0

the dating o f the Almagest. + (-1,05331" - 0,001513%)02 - 0,001530 "03;

X(5,50) = (10,5567"-1,88692%-0,000144 "5o)0
5. + (-2,38191"-0,001554% )02 -0,001661"03;
"REVERSE CALCULATION" OF OBJECTS' % i 0) = (5029,0946"+ 2,22280%+0,000264 "sq)0
POSITIONS ON THE CELESTIAL SPHERE. + (1,13157 " + 0,000212%)0 2 + 0,000102 "03. (L5'4)
Let us however note that the discrepancies between
5.1. Necessary formulae the corollaries made according to the actual theory of
Newcomb and its modification made by Kinoshita
Having the modern coordinates and proper m o ( [1222] ) that we have used are o f no consequence in
tion rates o f stars at our disposal, we can compile a sofar as our purposes are concerned. For any time
sufficiently precise star catalogue for any epoch in moment t of the historical interval under considera
the past. By sufficiently precise we mean the preci tion (between 600 b .c. and 1900 a.d.) the discrepan
sion that corresponds to modern astronomical the cies between the ecliptic coordinates of a star calcu
ories, which is quite sufficient for our purposes. Such lated according to Newcombs theory and those ob
precision can be considered absolute in comparison tained with the use o f its modified version ([1222])
to that of the old catalogues. are negligibly small in comparison to the errors of the
We had to perform retroactive star position cal Almagest. We have used [1222], since it gives the for
40 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

mulae for precession compensation in a format con ordinate system for 1900 a.d. Bear in mind that the
venient for computer calculations. position o f the stars on the celestial sphere changes
over the course of time in relation to any fixed sys
5.2. The algorithm for calculating past tem o f coordinates. The required calculations of the
positions of stars stars position are based on the known proper mo
tion rates va, v of the star by each of the coordinates
Let us provide a detailed description o f the algo a 1900, S 1900 (see Table 4.1, columns 5 and 6 ). We shall
rithm used for the calculation o f star catalogue K(t), come up with the following for non-reduced proper
which reflects the condition o f the celestial sphere motion rates:
for year t with sufficient precision, according to New
comb's theory. Here t is a randomly chosen epoch a(i) = a%00(i) = a-va- 1,
from the historical interval under consideration 8 0(t) = 8 % o (i) = 8 0 - v s -i.
(namely, one between 600 b .c. and 1900 a.d .). Epoch
t is calculated backwards into the past from the epoch Indeed, we can consider the proper motion rates
of 1900 in Julian years, in other words, t - 1 corre of each star by the coordinates a 1900>S 1900 to be con
sponds to the epoch of 1800, t = 10 - to the epoch of stant. The minus in the formulae cited above results
900 a.d ., t - 18 - to 100 a.d ., etc. The discrepancy of from the retroactive nature of calculations; the ve
several days that results from the differences between locity rate symbols va, v correspond to the normal
the Julian and the Gregorian calendar, and leads to flow o f time.
the situation where the epoch o f 100 a.d ., for in Before we can actually use this formula, we have
stance, fails to coincide with the epoch of 1 January to convert all the source values into a single meas
100 a.d . is o f no importance whatsoever. urement system. For instance, we can measure a(t)
The calculated star catalogues K(t) shall serve us and 8(f) in radians, and the velocities va, v - in (rad
for comparison with the old catalogue under study year) 10-2.
(such as the Almagest) with different values of t. Here Step 2. We have to shift from coordinates a 1900>
t shall stand for a random assumed dating of an old 8 j 9oo to coordinates Z1900, b 1900. We shall come up
catalogue. Thus, calculated catalogues K(t) must be with coordinates Z(t), b(t) of our star for the moment
transcribed in ecliptic coordinates for epoch t. As it t in spherical coordinates based on the ecliptic of the
has been pointed out, all known old catalogues are epoch of 1900 a.d. This is what we get:
compiled in ecliptic coordinates, be it Ptolemys Al
smb (0 - -sin a (t)cos 8 (Osine +sin8 (Ocose ,
magest or the catalogues of As-Sufi, Ulugbek, Coper o ,. ~o/ x o . . o
o sina (t)cos8 (t)cose +sin8 (Osine n r^
nicus, Tycho Brahe etc. tanl (t)= ----------------- 5--------- 5--------------- >
Let us assume that the modern equatorial coordi cosa V )co s8 (0
8 = 23027'8,26'.
nates of a star in a catalogue (such as [ 1197] ) are a 0=
a i9oo> $ = 8900. These coordinates reflect the position These formulae permit an unequivocal recon
of the star in question for 1900 a.d . in the spherical struction o f the values of (3(t) and a (t), since -9 0
coordinate system, whose equator corresponds to the < b(t) < 90 and \P(t) - a(t)\ x 90. The value o f e
Earths equator in 1900 a.d . The equator is defined by corresponds to the declination angle between the
the plane that is orthogonal to the axis of the Earths ecliptic of 1900 a.d . and the equator o f 1900 a.d . We
rotation. Let us remind the reader that this planes po refer the reader to the formula of 1.5.3, where one has
sition changes over the course of time. We have to cal to let 5 = -1 in order to make the transition between
culate the coordinates lp bv or the spherical coordinates 2000 a.d . and 1900 a.d .
whose equator coincides with the ecliptic, or the plane Step 3. We have to make a shift from coordinates
of the Earths rotation around the Sun for epoch t. /190o , b l900 to the auxiliary coordinates l 1 and b\ which

We should do the following for this purpose. are also tied to the ecliptic o f 1900. However, they
Step i . We have to calculate the stars coordinates have a different longitudinal reference point, which co
a(t), 8(t) for time moment t in the equatorial co incides with the intersection of the ecliptic of 1900 a.d.

Fig. 1.6. The sequence of steps that we use for reverse calculations of stellar positions and their past coordinates.

and that o f epoch f, or n i900 and U(t). This transition the choice o f the longitudinal reference point. In co
conforms to the following formulae: ordinates Z2, b2 this point corresponds to the inter
section of ecliptics ni900 and 11(f). The formulae of
l\t) = f ( t ) - cp, transition from Z1, b 1 to Z2, b2 correspond to the for
b\t) = b(t) mulae (1.5.5). Instead of 8 we have to take the angle
(p = 17357'38.436" + 870.0798"i + 0.024578"*2. (1.5.6) 81 between ecliptics n (t) and ni900:
Arc cpbetween the vernal equinox point of 1900 on e1= -4 7 .0 7 0 6 "*- 0.033769"*2 - 0.000050"*3.
the ecliptic n i90o and the intersection of n i900 and
n(t) conforms to the formula (1.5.1) if we re to assume This expression is derived from the formula (1.5.2)
that s0 = -1 and 0 = - t. Then the ecliptic n(s0) in where s = - 1 and 0 = -t.
fig. 1.5 shall correspond to the ecliptic n i900- Ecliptic Step 5. Finally, we have to make the transition
n(s) in fig. 1.5 shall represent the ecliptic of epoch t, from coordinates Z2, b2 to the ecliptic coordinates Zf,
which is of interest to us. Indeed, the time t is counted bt. This transition conforms to the following for
backwards from 1900 a.d. in centuries, whereas the re mulae:
mainder o f 0 = 5 - s0 is counted forwards from epoch
s0, also in centuries. Since we have agreed on s0 = - 1 , Zt= Z2 + cp + 'F, bt = b2,
which corresponds to 1900 a.d . (2000 - 100 = 1900),
we have to choose 0 = - t in order to make the epoch where cp is defined in (1.5.6) and is defined by for
s = 50 + 0 correspond to epoch t under consideration mula (1.5.4) with 50 = -1 and 0 = - t, therefore
in our formula (1.5.1).
Step 4. Next we have to make the transition from = -5026.872"t + 1.1314 "f2 + O.OOOlV.
coordinates Z1, b1to coordinates Z2, b2. These are spher
ical coordinates tied to the ecliptic n (t), whose only The sequence of steps 1-5 as described above is il
difference from the ecliptic coordinates Zt, bt is due to lustrated in fig. 1 .6 .
42 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

Let us conclude by pointing out that all the calcu The first instrument is the meridian circle, or the
lations necessary for the dating of a given star cata so-called transit circle as described by Ptolemy (see fig.
logue can be performed without accounting for the 1 .8). The instrument looked like a flat metal ring of a
Newcomb-Kinoshita theory. We shall consider this in random radius installed on a reliable support vertically
more detail below. The Newcomb-Kinoshita theory is in the plane o f the local meridian. The circle was
only used in order to obtain additional information graded (into 360 degrees, for example). Another ring
concerning the errata in the estimation o f the eclip o f a smaller diameter was placed inside the larger ring;
tic plane made by the author of the catalogue. The it could rotate freely, remaining in the same plane as
value of these discrepancies is the auxiliary factor that the larger ring (fig. 1 .8). There are two little metallic
confirms the correctness o f our corollaries. See plates with pointers attached to two opposing points
Chapters 6 and 7. on the inner ring (marked P in fig. 1 .8); the pointers
point at the grades found on the external ring. The de
6. vice is installed in the plane of the local meridian with
ASTROMETRY. ANCIENT ASTRONOMICAL the aid o f a level and the meridian line whose direc
MEASUREMENT INSTRUMENTS tion is defined by the shadow of a vertical pole at mid
O FTH EXV-XVII CENTURY day. Then the zero mark on the external ring of the
instrument is synchronised with the local zenith.
In Section 3 we have considered the general con The instrument described above can be used for
ception o f angular measuring devices used in as measuring the height of the Sun at given latitude. One
tronomy, which is important to us since it enables us must quickly turn the inner ring at midday until the
to estimate the position o f the equatorial line on the shadow o f one plate P covers the other plate P com
celestial sphere with sufficient precision. pletely. In this case, the position of the pointers on the
Let us assume that the observers line of eyesight plates shall tell us the height o f the Sun with the aid
is directed along half-line HK\ which moves along of the grade marks on the external ring. It has to be
the line o f the equinoctial in its daily rotation with pointed out that the instruments indications are to be
out any tergiversation. The attitude o f half-line HK' read after one fixes the plates in their proper posi
will naturally depend on the geographical latitude. tions. This tells one the height of the Sun already after
We can define the plane H IM , an orthogonal quad midday. Moreover, the meridian circle can measure the
rant parallel to the equatorial plane, which crosses the angle between the ecliptic and the equator.
celestial sphere precisely along the equinoctial, qv in The second instrument is the astrolabon as de
fig. 1.7. It is therefore possible to construct a station scribed by Ptolemy, which is more frequently referred
ary device in said point of telluric surface, oriented by to as astrolabe in our days. The latter term is me
the north-south meridian, which allows marking the diaeval in origin. According to the Scaligerian his
equator on the celestial sphere visually. This permits tory o f astronomy, the meaning o f the term astro
precise estimations o f equatorial stellar latitudes - labon has been changing over the course o f time.
during their crossing o f the quadrants vertical plane, We are told that in deep antiquity, or around the
for instance. As we have already pointed out, the meas very beginning o f the new era, the term astrolabon
urement of equatorial latitudes was hardly a compli was used for referring to the instrument that we shall
cated task for a professional astronomer o f the XIV- describe shortly. Ptolemy used one o f those. However,
XVI century. It required nothing but accuracy and in the Middle Ages the instrument in question was al
sufficient time for observations. In particular, it has to ready known as the armillary sphere, or armilla.
be expected that a careful observer could not make a Some modern astronomers are o f the opinion that
grave systematic error in the estimation of stellar dec Ptolemy describes the armillary sphere or the astro
linations for a given year. labon in his Almagest, and not the actual astrolabe
Now let us see how the simple general idea de (see [395], for instance). According to Robert Newton,
scribed above was implemented in real mediaeval in a renowned astronomer, it is likely that around the
struments. end o f the Middle Ages the term astrolabe referred

Fig. 1.8. The armillary circle.

to the device used for measuring the height o f a ce strument was manufactured when Ptolemys epoch
lestial body above the horizon. As for the device we was already considered ancient - it belonged to Tycho
describe herein [in accordance with Ptolemys indi Brahe, the famed XVI century astronomer ([1029],
cations - Auth.], by that time it was better known as page 13). The implication is that astronomical in
the armillary sphere, which is the distant ancestor of struments remained the same for fifteen hundred
the modern telescopes bearings ([614], page 151). years. As we can see, the instruments of the ancient
In order to avoid confusion with terms, we shall Ptolemy from the second century a.d. and the XVI
describe the two instruments separately - Ptolemys as- century scientist Tycho Brahe were almost identical,
trolabon and the astrolabe, or the mediaeval instru as though they were made in the same mediaeval
ment whose name is virtually identical to that o f Ptol workshop. An ancient drawing o f Tycho Brahes large
emys astrolabon. The primary elements of the astro- armillary sphere can be seen in fig. 1 . 12 .
labons (armillas) construction are shown in fig. 1.9. We must now describe the correct use o f this in
In fig. 1.10 we see the principal scheme o f the medi strument to the reader and also relate the astronom
aeval armillary sphere. Fig. 1.11 shows us the medi ical principles o f its construction. The main element
aeval armillary sphere - of Ptolemys type, according of the armillary sphere comprises two metallic rings,
to historians. Its diameter equals 1.17 metres. This in perpendicular to one another and rigidly joined to-

Fig. 1.9. A scheme of the astrolabon (armilla). Fig. 1.10. A scheme of the armillary sphere.
44 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

orthogonal to the plane of the ecliptic, crossing the

Earth surface along the meridian, qv in fig. 1.13.
Let us assume that the armilla is installed some
where along this meridian. The instrument can be lo
cated anywhere on the surface o f the Earth, but the
measurements must begin at midday, which is when
the instrument shall be on the meridian that is the
intersection o f said plane and the surface o f the
Earth. We assume the observer to know the direction
o f the Earth axis in this part o f the Earth; therefore,
the armillas NO axis shall be oriented in this direc
tion, parallel to axis N O ', qv in fig. 1 .1 3. Then, by ro
tating the first metallic ring around the armillas axis
NSywe shall install this ring in the plane of the merid
ian, which will happen when the shadow from the ex
ternal edge of the ring shall cover the inner part of
the ring exactly. Finally, having fixed the plane o f the
first ring, we must make the second ring orthogonal
to the first, so that its inner part would be covered
by the shadow cast by its outer part. Fig. 1.13 demon
strates that the second ring shall end up right in the
plane of the ecliptic as a result of these manipulations
(more precisely, it shall be parallel to the ecliptic
plane). As we have fixed both rings in the necessary
Fig. 1.11. The armillary sphere made in the XVI century; it position, the perpendicular P XP 2 to the second ring
used to belong to Tycho Brahe (1598). It is almost indistin
shall also be fixed, thus marking the pair o f polar
guishable from the instrument used by the ancient Ptolemy
in the II century A. D. These instruments are most likely to points Pi and P 2 on the first ring. We shall therefore
date to the same epoch - the XV-XVII century. Taken from be able to measure the angle P xON with sufficient
[1029], page 13. precision; it is obviously equal to the angle between
the ecliptic and the equator.
We have described the method that was allegedly
gether in points E x and E2. Let us henceforth refer to used by the ancient astronomers. Despite the geo
the rings as the first and the second (see fig. 1.9). metrical simplicity of the idea, one can clearly see the
The first ring rotates around the axis NS, which is numerous complications that introduce different er
parallel to the axis of telluric rotation. The centre of rata into the numeric value o f the measured angle. In
both rings is point O; P lP2 is the perpendicular to the particular, the observer must know the following pa
second rings plane. rameters:
Let us describe how one uses the armilla in order a) the direction of axis ON, which is parallel to the
to measure the angle between the ecliptic and the equa axis o f the Earth;
tor, for example. The most appropriate time for such b) the day of solstice;
measurements falls over the day of summer or winter c) the moment o f midday in this point o f Earth
solstice. The corresponding point on the orbit of the surface.
Earth is marked O' in fig. 1.13. It doesnt matter R. Newton made the following justified remark:
whether it corresponds to summer or winter solstice. The primary shortcoming of this instrument is that
Let us consider the plane that crosses the radial vector one has to be rather quick when one uses it, since the
CO', where C is the Sun, and the Earth axis is indicated rotation o f the Earth has a negative effect on the pre
as NO'. Since O' is the solstice point, this plane will be cision o f the device ([614], page 150). Indeed, in

fig. 1.13 we can see that the rotation of the Earth be

gins to turn the instrument around axis 0 'Nywhich
renders the above considerations invalid.
Strictly speaking, the points O (the centre o f the
armilla) and O' (the centre of the Earth), as seen in
fig. 1.13, are different points. The distance between
the two is equal to the radius of the Earth. However,
this discrepancy is negligibly small for the above cal
culations. Therefore, we can assume that O = O' in
sofar as these measurements are concerned, as shown
in fig. 1.13.
Let us come back to the measurements of the eclip
tic coordinates with the aid o f the armilla.
After the correct installation of the device as de
scribed above, it is tuned to the ecliptic coordinate
system for a short time, namely, the plane of the sec
ond ring E XE 2 is parallel to the ecliptic plane. Points
E 1 and E 2 shall correspond to the solstice points. Both
rings are presumed graded. Therefore, we can unam
biguously define points and R 2 on the second ring,
which shall correspond to the equinoxes. They divide
arcs E 1 and E2 in two halves. Points R l and R 2 are omit
ted from fig. 1.13 so as not to make the illustration too
cluttered. Thus, what we have on the second ring is a
scale with a fixed initial reference point ( R lf for in
stance, which is the vernal equinox point). We can
Fig. 1.12. The large armillary sphere of Tycho Brahe for
thus measure ecliptic longitudes and latitudes o f measuring the angular distances between luminaries (from
points on the celestial sphere, such as stars. M echanics R ejuvenated by A stronom y , a work of Tycho
However, let us reiterate that the daily rotation of Brahe. Windsbeck, 1598. Taken from [926], page 62.
the Earth quickly sets off the precision of the instru
ment. Therefore, one needs a precise chronometer in
order to compensate for the rotation of the Earth and The shadow from the pointer falls over the lower
tune the instrument. This is how the modern meas (northern) side of the meridian circle and can move
urement instruments are constructed - the rotation within the confines o f one quarter of the circumfer
of the Earth is compensated by the automatic track ence. Therefore, in order to measure the height of the
ing system. Sun it suffices to grade one quarter of the ring. The
In order to facilitate the measurements of celestial quadrant is therefore a plate o f some sort with a
objects ecliptic coordinates, a third ring is usually graded quarter of a circle installed in the plane o f the
added to the armillary sphere - a rotating one. The meridian. The height of the Sun above the horizon
axis of its rotation can, it turn, slide along the second at midday is indicated by the shadow o f the pointer
ring, which is positioned in the plane o f the ecliptic. that falls over the scale.
We shall omit these details, since they are of little im In fig. 1.15 we see the astronomical quadrant from
portance to us. a mediaeval book o f 1542 by Oronce Fine ([1029],
Let us now consider the third instrument, or the page 19).
quadrant (see fig. 1.14). This instrument is based on Fig. 1.16 shows us a small quadrant with a radius
the meridian circle and has a sharp pointer at its cen o f 39 centimetres, which belonged to Tycho Brahe
tre, which is perpendicular to the plane o f this circle. ([1029], page 26).
46 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

/ The axis
y > o f telluric

The m eridian

The half-rings
o f the armina
are shaded

^ The e q u in o x ^ in t St P2
\ ^ The Sun
\ /
\ /

Fig. 1.13. A scheme of utilising the armilla for the measurement of the angle between the equator and the ecliptic, for instance.

In fig. 1.17 we see Tycho Brahes sextant with a ra o f the circle. The instrument can be suspended ver
dius o f 1.55 metres, and in fig. 1.18 - another sextant tically; there is a special loop at the edge o f the plate
o f Tycho Brahe o f the same size ([1029], page 26). that serves this purpose. The plane o f the vertically
In fig. 1.19 we see the astronomer Hevelius por suspended circle could be directed at a celestial body,
trayed performing measurements with the aid o f the likewise the rotating mobile plank. This is how the
sextant ([1029], page 67). bodys height above the horizon was measured. Apart
The fourth instrument is the astrolabe (see fig. from that, after the measurement o f the height o f the
1 .20 ). The mediaeval astrolabe is a round metallic Sun at midday, one could also measure the observa
plate with a diameter o f some 50 centimetres, with a tion latitude. The precision o f such measurements
graded ring mounted rigidly on one o f its edges. At must have been rather low due to the primitive na
the centre o f the ring there is a mobile plank with vi ture o f the method used. It is believed that the in
sors mounted on an axis perpendicular to the centre strument in question could measure the observation

Fig. 1.15. An astronomical quadrant from a mediaeval book

Fig. 1.14. A scheme of the quadrant. by Finney. Taken from [1029], page 19.

Fig. 1.16. A small quadrant of Tycho Brahe (1598). Taken Fig. 1.17. The sextant of Tycho Brahe (1598). Taken from
from [1029], page 26. [1029], page 26.

Fig. 1.19. The astronomer Hevelius is using a large sextant

Fig. 1.18. Another sextant that belonged to Tycho Brahe for observations, assisted by his wife. Ancient engraving dat
(1598). Taken from [1029], page 26. ing to 1673. Taken from [1029], page 67.
48 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

In fig. 1.22 we reproduce an old picture of the fa

mous mediaeval astronomical instrument known as
the Turkish tool, or torquetum (turquetum).
Specialists in the history of science tell us the follow
ing: The torquetum (or turketum), whose name
can be translated as the Turkish tool, was character
istic for the mediaeval European astronomy, and em
bodies the intellectual heritage of Ptolemy as well as
the Islamic tradition... The torquetum was used for
measuring all three types of astronomical coordinates
and the conversions between different types of coor
dinates, which was stipulated by the Ptolemaic plan
etary theory ([1029], page 17). The instrument
shown in fig. 1.22 belonged to Petrus Apianus (1497-
1552). We are therefore told that the mediaeval Turks
revived the Ptolemaic theory o f measurements,
manufacturing the necessary tools after many years
of oblivion - namely, fifteen hundred years later than
point latitude with the precision of several arc min the ancient Ptolemy.
utes ([614]). As we are beginning to realise, the mediaeval O t
In fig. 1.21 we see an old astrolabe o f 1532 (Georg toman turketum was a contemporary o f the Ptole
Hartmann, Nuremberg). We see the front and the re maic devices. These instruments were made in the
verse of the astrolabe. XV-XVII century.

Fig. 1,21. The astrolabe o f Georg Hartmann from Nuremberg. We see both the front and the reverse sides of the instrument.
Taken from [1029], page 15.

a great extent. In particular, time was often consid

ered anthropomorphic before the invention o f the
clock - more specifically, its character and speed
would depend on the nature o f events. As we already
reported in Chron I, before the XIII-XIV century
timekeeping devices were a rarity and a luxury.
Sometimes even the scientists would lack them. The
Englishman Valcherius ... regretted the fact that the
precision o f his lunar eclipse observations o f 1091
was impaired by the absence o f a chronometer
([1461], page 68 ). Timekeeping devices o f low preci
sion were introduced in the Middle Ages: the usual
timekeeping devices in mediaeval Europe were sun
dials ... hourglasses and clepsydrae. However, sun
dials were only useful for sunny days, and clepsydrae
remained a rarity ([217], page 94).
In fig. 1.23 we see the astronomical rings o f the
XVII-XVIII century, which were used for telling the
time by the Sun in particular. The method o f their use
is shown in an old drawing that we reproduce in fig.
1.24. In fig. 1.25 one sees an old hourglass.
Mass production o f clepsydrae falls over the XIII-
XIV century. Clepsydrae were used by Tycho Brahe
(1546-1601). He used them in order to measure plan
Fig. 1.22. A mediaeval instrument known as turketum
(Turkish). Designed for estimating several types of celestial etary velocities ([954], page 36). In the Middle Ages
objects coordinates. It was also utilised in Ptolemaic plane the clepsydra was a popular device, its low precision
tary theory (Werner, 1533). Taken from [1029], page 18. notwithstanding. In order to make them more pre
cise, the constructors o f the clepsydrae had to take
into account the fact that the water doesnt leave the
7. vessel at a constant speed - the latter depends on the
TIMEKEEPING AND TIMEKEEPING DEVICES pressure, that is to say, the greater the level o f water
IN MEDIAEVAL ASTRONOMICAL in a vessel, the greater the pressure. The constructors
OBSERVATIONS o f the clepsydrae improved the construction some
what, making it more complex, so that the clock
As we have pointed out earlier, in order to conduct wouldnt slow down as the vessel on top emptied...
precise astronomical observations, the ancient as However, clepsydrae had the tolerance of around 10-
tronomers needed a chronometer with a minute hand 20 minutes per day, and even the best scientists o f
or some equivalent thereof. It would be expedient to the epoch couldnt think of a way to make them sub
recollect the history o f mediaeval timekeeping in this stantially more precise ([288], pages 32-33).
respect in order to compare the precision o f mediae At the end o f the IX century candles were used
val timekeeping devices to the relative precision of the widely for timekeeping purposes. For instance, Alfred,
coordinates included in mediaeval star catalogues, King of England, took candles of different length along
the Almagest catalogue in particular. on his voyages and ordered to light them one after
In general, it has to be mentioned that the very another ([217], page 94). This method o f timekeep
concept of time was rather idiosyncratic in the Middle ing was still used in the XIII-XIV century - in the
Ages. The analysis o f the ancient documents demon reign o f Charles V and other monarchs of the epoch.
strates that this concept differed from the modern to Timekeeping candles were known as the fire clock.
50 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

The hourglass was another popular timekeeping

device of the Middle Ages. The precision of the hour
glass depends on the stability o f the sand flow. In
order to make the hourglass more precise, one needs
to use sand o f as uniform a texture as possible, soft,
dry and forming no lumps inside the vessel. Mediae
val craftsmen o f the XIII achieved this by boiling the
mixture o f sand and marble dust with wine and
lemon juice, skimming it, then drying and repeating
the process nine times over. All o f these measures
notwithstanding, the hourglass remained a tim e
keeping instrument of low precision ([288], page 30).
In the XII century, the secular rulers o f Mons who
wanted to begin a process at a given time had to con
sult with the ecclesiastic authorities about the time of
day ([1037], pages 117-118).
Nowadays it is believed that the first mention o f a
mechanical chronometer dates from the end o f the
VI century a.d. ([797]). Then the devices disappear
for a long time to resurface already during the Ren
Fig. 1.23. An instrument of the XVII-XVIII century that was
aissance. According to the specialists in the history o f
used for solar timekeeping, among other things. Taken from sciences, the first mechanical clock was made by the
[1029], page 21. ingenuous and curious Italian craftsmen in the XIII
century ([954], page 38). The principle o f their con
struction is simple enough - a rope with a weight on
Many countries preserved this timekeeping method its end is woven onto a horizontal shaft. The weight
for a long time. aThe Japanese, for example, used time pulls the unwinding rope, which rotates the shaft. If
keeping devices consisting o f various incense sticks we are to attach a hand to the shaft, it will tell the time.
leaning one against another as recently as 200 years Despite the simplicity o f the principle, its practical re
ago. One could smell the hour by their aroma, as it alisation required a stable slow rate o f shaft rotation.
were. The Europeans used fire clocks as well - they This purpose was achieved by means o f using nu
were candles with special markings ([954], page 37). merous wheels, which transferred the rotation o f the
We can see that all these ancient timekeeping meth shaft to the hand, and clever regulators o f all kinds,
ods were used relatively recently; one must think, they installed to make the shaft rotation rate more or less
were invented not so very long ago. uniform. Mechanical clocks were constructions o f
Fire clocks were used in China for a long time formidable size. Enormous clockwork mechanisms
as well. Special kinds o f powdered wood were made were installed on the towers of cathedrals and palaces
into a paste, which would then be rolled into sticks ([954], page 38). A flywheel from Tycho Brahes clock
o f various shapes - spirals and so on. Occasionally, had 1200 notches and a diameter o f 91 centimetres
metal balls were tied to these sticks in certain places. ([288], page 35). The wheels of some clocks weighed
As the stick burned, they would fall into a vase and hundreds o f kilos. Due to the large weight o f their
make a sound. The precision o f fire clocks also left parts and substantial friction, wheel-based mechan
much to be desired - apart from the difficulty of mak ical clocks required lubrication and constant main
ing perfectly uniform sticks and candles, the speed of tenance. The daily tolerance rate o f such clocks
their combustion always depended on the atmos equalled several minutes ([288], page 35).
pheric conditions (wind, fresh air supply etc) ([288], It was only in the XV century that the spring re
pages 30-31). placed the shaft and rope in clockwork mechanisms.

The weight of clocks was reduced dramatically. Crafts

men o f the early XVI century mastered the con
struction o f mobile spring-based clocks that weighed
3 or 4 kilos. They were the rather heavy ancestor of GemmamPhryfium.
the modern mechanical watch ([954], page 39).
The invention o f the clock with a minute hand
must have been followed by the compilation o f a
more or less precise longitudinal star catalogue. What
is the significance of the minute hand? The matter is
that the celestial sphere and all the objects seen upon
it rotates at the speed o f one degree per 4 minutes;
therefore, a star passes 15 arc minutes per minute of
time. Star catalogues contain coordinates o f stars in
dicated with arc minutes - therefore, in order to make
the catalogue precision tolerance equal circa 15 arc
minutes, one needs to be able to track the time in
terval of one minute on a timekeeping device. The tol
erance o f circa 10 minutes (as in the Almagest, for in
stance) requires the ability o f measuring 40-second
intervals reliably. Higher precision o f a catalogue re
quires a higher precision o f timekeeping devices. O f MODIS OMNIBVS ORNATISSIMO
course, the observers could use their intuition for the Ac vert NobBi Domino IoanniKhmitter,
Screnifftm R e g i t i f Hungary Sccmwto
measurement o f short time intervals (one minute GemmaPhryfius S. 0#
and less), but this would introduce subjective errata Nter milita varia$ m im m tlw n genera,qu* diiieritfnmfo
ac admiratione dignis eflfinxit natura dor&us,vix imieniat
into the catalogue. vir oma-iiT. aUquocLquadminus fuo fungatur officio at<
humanmn genus Quod quoin a Deo Opt.Max. crcatum
Thus, the ancient astronomers who claimed their fit perfe&jUimum,rationeiMdiuinaanrmiparte prsedifoin,
catalogues to have a tolerance o f 10 ' needed to have <gua& eaqu* refte fitmr d%ercr,fdtarentr%& <raqu* prater officifi
iuntfogeret dmllarctur^nihminusagiCjno^uaf! qoadam attira!
a chronometer with a minute hand or some analogue Oh
thereof at their disposal. However, Ptolemy, who gives
Fig. 1.24. The astronomical rings of Gemma Frisius. A port
us a detailed description o f all the instruments re
able equatorial instrument that could be used at any latitude
quired for the measurements o f stellar coordinates ... for solar timekeeping, as well as many other approximated
(the armillary sphere etc) doesnt mention any chron astronomical observations (Apianus, 1539). Taken from
ometers and altogether refrains from the discussion [1029], page 21.
of the timekeeping problem and its direct relation to
the observations o f the celestial sphere, which is in a
constant motion. According to the history o f timekeeping, the hour
The hypothesis that chronometers with a minute hand was introduced into the mechanism o f a clep
hand could exist in the II century a.d . contradicts sydra in the XIII century a.d . ([544], Volume 4, page
Scaligerian information about the history of time 267) or even later. The timekeeping devices in ques
keeping devices, as we shall shortly see. tion had no pendulum, and were therefore o f low
Also, the above implies that if we really discover precision. It was only in the XTV century a.d . that dif
some sort o f catalogue whose precision tolerance ferent cities o f mediaeval Europe got tower clock
equals 10 arc minutes as declared by the author of the work mechanisms (Milan in 1306 and Padua in
Almagest, and this tolerance is verified by statistical 1344). It is reported that they were built by a certain
research, we shall have a good reason to assume that Dondi Horologiu. Clocks with springs moved by a
the compiler o f the catalogue was using a clock with weight were only brought into existence in the XV
a minute hand or some equivalent o f it. century. Walther was the first to use them for astro-
52 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

nomical observations, followed by many others up to

Tycho Brahe ([544], Volume 4, pages 267-268).
According to the history of sciences, various me
chanical clocks only had the hour hand initially. In the
middle o f the XVI century the minute hand was in
troduced, and the second hands invention took place
200 years later ([954], page 39). The invention of the
mechanical clocks minute hand is usually dated to
1550 a.d . ([288], page 36). It is believed that the first
chronometer was only constructed in the XVIII (1785,
by John Harrison). Harrison lived around 1683-1776
([ 1029], page 139). Harrisons chronometer is a com
plex enough instrument; it can be seen in fig. 1.26.
The modern mechanical clock, including the pen
dulum, was invented by Huygens in 1657 {[797]). In
1561 the Kassel observatory was built - a unique con
struction, since it was the first to embody the princi
ple o f rotating roof (a device used in most modern
observatories). After the death o f Regiomontan and
Walther, Landgrave Wilhelm IV o f Hessen-Kassel
(1532-1592), the creator o f said observatory, con
ducted extensive observations o f immobile stars (see
Chapter 11 below). In general, the primary purpose
of the Kassel observatory was the compilation of a star
catalogue ... The most remarkable innovation was the
clock used for timekeeping and measurements in
volving the motion o f the celestial sphere. The con
struction o f a clock whose precision was adequate
for this purpose owes its successful implementation
to the mechanical genius o f Biirgi [1522-1632 -
Auth.], and, in particular, to his discovery that the
clock can be regulated by the pendulum - apparently,
he hadnt made any attempts o f making this inven
tion public, and so the pendulum was reinvented be
fore it could be acknowledged by everyone [in re the
discovery o f Galileo and Huygens - Auth.]. By 1586,
the positions o f 121 stars were registered with the
greatest care, but the complete catalogue, which was
supposed to contain over 1000 stars, has never been
finished ([65], page 118).
The activity o f Tycho Brahe, who worked in the
same epoch, soon completely outshone the efforts of
the Kassel observatory. It is curious enough that the
scientists o f the Kassel observatory already used re
Fig. 1.26. The first chronometer created by John Harrison in fraction compensation to counteract the errata in
1735. The height of the instrument is 408 millimetres. Taken troduced by the refraction o f sunlight in the atmos
from [1029], page 140. phere ([65], page 118).

It was only in the time o f Huygens that the clock end o f his life, but we have no proof that he ever
became an integral part o f many astronomical in managed to make this idea come alive.
struments: One o f the inventions made by Huygens This invention has given us the opportunity to
completely revolutionized the art o f precise astro make precise observations, and, noting the gap be
nomical observation. Huygens attached the pendu tween two stars crossing the meridian, deduce their
lum to the clock that was set in motion by weights, angle distance to the west or the east, knowing the
in such a manner that the clock maintained the pen speed o f the celestial spheres motion.
dulum in motion, which, in turn, regulated the mo Picard was the first to appreciate the importance
tion o f the clockwork. o f this invention for astronomy, introducing correct
It is likely that Galileo planned to unite the pen timekeeping in the newly built Paris Observatory
dulum and the clockwork mechanism towards the ([65], page 177).

A preliminary analysis
of the Almagest star catalogue

1. division o f the ecliptic into twelve equal parts as stip

THE CATALOGUE STRUCTURE ulated by the epoch of observation. It has to be em
phasised that (strictly speaking) the even Zodiac is
The Almagest star catalogue comprises its seventh defined by the observable solar trajectory on the ce
and eighth books. We were using the canonical edi lestial sphere and not the zodiacal constellations per
tion o f the Almagest star catalogue for our research, se. The ecliptic arc covered by the Sun during the first
as published by Peters and Knobel ([1339]), as well month o f march (not the calendar march, but the
as the two complete editions o f the Almagest trans month that begins on the day o f vernal equinox) is
lated by R. Catesby Taliaferro ([1355]) and Toomer commonly referred to as Aries. The next equinoc
([ 1358]). The first Russian translation of the Almagest tial month o f April is when the Sun passes through
came out in 1998 ([704]). the constellation o f Taurus o f the even Zodiac. Next
Before we give our characteristic to the catalogue, come Gemini, Cancer, Leo, Virgo, Libra, Scorpio, Sagit
it would be expedient to remind the reader of a few tarius, Capricorn, Aquarius, and, finally, Pisces. This
concepts used in literature on the history of astronomy. is how the annual ecliptic circle ends. Thus, the even
The Almagest star catalogue was compiled in the Zodiac can be regarded as a simple way o f dividing the
ecliptic coordinate system. As we mentioned earlier, ecliptic into 12 equal 30-degree parts starting with the
in most o f its editions and copies stellar latitudes are vernal equinox point for a given epoch. Precession
rendered to the epoch o f circa 60 b .c. In other words, makes this initial point o f reference shift along the
the initial longitudinal reference point was recalcu ecliptic at the rate o f circa 1 degree per seventy years.
lated by someone to correspond to the position o f the These shifts are significant, but relatively small as com
sun in relation to the stars as they would appear to pared to the thirty-degree span of a whole sign. There
the observer from the middle o f the I century a.d . on fore, the even Zodiac, once chosen for its approximate
the day o f vernal equinox. correspondence to the constellations of the Zodiac, re
Stellar longitudes as indicated in the Almagest cat tains this correspondence to this day. In other words,
alogue relate to the so-called even Zodiac as counted if the Sun is in Aries (or March, according to the even
off the vernal equinox point of a given epoch. Let us Zodiac), it shall be near the zodiacal constellation of
explain that the even or monthly Zodiac is a mere Aries. The reverse is possible as well - namely, that the

Table 2 . 1 . Signs o f the even Zodiac corresponding to + 660' in the Almagest. The + or here refer to
30-degree arcs (or longitudinal intervals) as counted the respective location of the star in the Northern or
from spring equinox point o f the current epoch. Southern Hemisphere.
As we have pointed out already, zodiacal signs do
A bbreviated Longitudinal not correspond to zodiacal constellations, which is
Latin n a m e o f a sign Latin n a m e interval why the stars that pertain to a single zodiacal con
Aries Ari 0 -3 0 stellation can wind up in different zodiacal signs.
Taurus Tau 3 0 -6 0 The canonical version o f the Almagest catalogue
contained in the work o f Peters and Knobel ( [ 1339] )
Gemini Gem 6 0 -9 0
is presented as a table that consists of six columns.
Cancer Can 90 - 120
The first column contains the index number of a
Leo Leo 1 2 0 -1 5 0 given star in the Almagest. This numeration was de
Virgo Vir 1 5 0 -1 8 0 vised by the astronomer Bailey. Surviving manuscripts
Libra Lib 1 8 0 -2 1 0 o f the Almagest contain no numerical indexation.
Scorpius (Scorpio) Sco 2 1 0 -2 4 0 Bailey was a famous commentator and researcher of
Sagittarius Sag 240 - 270
the Almagest. According to Bailey, the sum total of
stars listed in the Almagest equals 1028. There are
Capricornus (Capricorn) Cap 270 - 300
minute discrepancies between the estimates o f dif
Aquarius Aqu 300 - 330
ferent researchers, one of the reasons for their very ex
Pisces Pis 330 - 360 istence being the fact that some stars were listed twice
in the Almagest (see [1339] for more details).
The stars are grouped by constellation in the Al
boundaries o f zodiacal constellations were once de magest; each of the constellations has a name. The
fined in such a manner as to correspond to the even Almagest lists 48 constellations all in all; we shall cite
Zodiac - the visible solar route, or the ecliptic, divided the actual list below. Some constellations have annexes
into twelve even parts. referred to as Informata - auxiliary stars that werent
In table 2.1 we cite the complete list o f signs (or included in the main list of stars comprising a given
arcs) that comprise the even Zodiac. All o f them are constellation. The Latin term informata translates as
counted off the variable vernal equinox point. shapeless or amorphous (informis, informitas
Stellar longitudes were transcribed with the aid of etc). In other words, the main list apparently contains
these arc signs (or month signs) in the Middle Ages. the stars that the ancient astronomer believed to form
For instance, 1520 in Taurus stood for 4520' as the skeleton of the constellation, whereas the stars
counted off the current vernal equinox point (or some listed as inform ata provide the backgrouond of
other point chosen as referential by the authors of a sorts. It is possible that the compiler of the catalogue
given catalogue for reasons o f their own). It has to be believed the stars of the informata category to be o f a
stated that the equinox point didnt always serve the lesser importance than the main stars. One must
referential purpose in the old catalogues. Let us con bear in mind that the ancient astronomy was closely
sider another example: w15o20 in Libra would mean linked to astrology, where the visual outline o f a con
22520' as counted off the point of reference. See table stellation was o f paramount importance. Some of the
2.1. This is how the longitudes are transcribed in the Almagest constellations have no inform ata whatso
Almagest catalogue. ever. A full list of constellations can be found below,
Ecliptic latitudes of stars in the Almagest are in in table 2 .2 .
dicated according to a simpler principle - namely, The second column of the table in [1339] contains
they are counted off the ecliptic that corresponds to a verbal description of the star in question and the part
the latitudinal zero degree and up to the ecliptic pole it plays in the general shape o f a given constellation.
corresponding to the 90th latitudinal degree. The Such descriptions are often rather vague. For instance,
Alpha o f Ursa Minor, for example, has the latitude of the Alpha o f Ursa Minor is referred to as the star on
56 I h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

the tip o f the tail in the Almagest. In the canonical the following famed researchers: Peters, Bailey, Schjel-
version of the Almagest ([1339]) verbal descriptions lerup, Pierce and Manitius. Different identifications
o f stars were taken from the Latin edition o f 1528 o f certain Almagest stars on the celestial sphere of
translated by Trebizond. They were verified by the our epoch suggested by said astronomers are also in
Greek edition. It is believed that the initial language dicated.
of the Almagest was Greek. See Chapter 11 for more We have partially processed this enormous body of
details concerning the history o f the Almagests man material. First o f all, it is very useful to indicate the lo
uscripts and first editions. cation o f the constellations mentioned in Ptolemys
The modem names of the stars can be found in the star catalogue geometrically. Let us use a modern map
third column o f the table in [1339]. Actually, this col that specifies constellation boundaries for this end. In
umn contains the names o f the Almagest stars iden fig. 2 . 1 . these boundaries are represented as uninter
tified on the star chart o f today. Said identifications rupted zigzagged lines. This is an approximated rep
are the result o f much labour performed by the sci resentation, of course, since the ancient constellations
entists whose research involved the Almagest. What had no rigidly defined borders. However, it suffices for
complicates such identifications is the rather whim a rough estimate, therefore we can assume that fig. 2.1
sical nature o f the verbal descriptions in question. gives us a correct qualitative representation o f how
Moreover, the very figures of constellations could vary the Almagest constellations are positioned on the ce
from one astronomical school to another over the lestial sphere.
course o f years. Therefore, the identification o f the Let us compare this illustration to the star chart
Almagest stars as some of the stars that we know today (with drawn constellations) from the first editions of
is anything but self-implied. Obviously enough, this the Almagest - in Greek and in Latin, dating from the
is the very first problem to solve before we can pro XVI century a.d. In fig. 2.2 we see a star chart o f the
ceed to analyse other characteristics of the catalogue. Northern hemisphere drawn by Albrecht Drer, and
An enormous body o f work was conducted by the in fig. 2.3 - the chart o f the Southern hemisphere by
XV II-XIX century astronomers in order to identify the same artist. Drer created these maps in 1515 (see
the Almagest stars. The final version can be found in [544],Volume 4 ,pages 204-205; also [9 0 ],pages 8-9).
[1339]. We shall forthwith refer to it as canonical. Drers map o f the Northern Hemisphere was in
The same source ([1339]) contains the table o f dis cluded in the 1527 edition o f the Almagest ([90],
crepancies between the opinions of different special page 8 ). Drers star chart o f the Southern Hemi
ists in re the identification o f a given star. This table sphere saw another edition in 1527, slightly altered
contains several such identifications of Almagest stars. (we reproduce it in fig. 2.4).
The fourth column contains the ecliptic longitude In figs. 2.5 and 2.6 we see two more star charts in
o f a star as related to the arc (or sign) o f the even Zo cluded in another edition o f the Almagest (dating
diac that the longitude value in question falls over. from 1551). It is very peculiar that although the an
The fifth column contains the stars ecliptic lati cient Ptolemy is supposed to have lived in the II cen
tude. tury a.d., some o f the constellation figures are dressed
The sixth column corresponds to the brightness in mediaeval attire ([543], pages 216-217).
(or size) o f the star. In figs. 2.7 and 2.8 we also reproduce the maps o f
the Northern and Southern hemisphere compiled in
2. accordance with the Almagest by the astronomer
OF RELIABLY AND POORLY IDENTIFIABLE Drers star chart does not contain any precise
STARS IN THE ALMAGEST borders o f Almagest constellations. The matter is that
Drer merely drew the symbolic figures o f zodiacal
The book ( [1339] ) contains a table entitled Iden constellations - Hercules, Pegasus etc. Nevertheless,
tification Discrepancies, which deals with the differ a comparison with the modern star chart demon
ent identifications o f certain Almagest stars made by strates that modern constellation borders are in good

Fig. 2.1. Modern boundaries of the constellations mentioned by Ptolemy in the Almagest.
58 I h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1


Fig. 2.2. Star chart o f the Northern Hemisphere drawn by Albrecht Drer in 1515. Taken from [544], Volume 4, page 204. See
also [90], page 8.

Fig. 2.3. Star chart of the Southern Hemisphere drawn by Albrecht Drer in 1515. Taken from [544], Volume 4, page 105.
6o | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

Fig. 2.4. Dlirers star chart of the Southern Hemisphere, published once again in 1527 - this time somewhat altered. According
to the commentators, the decorative framing was added subsequently, and includes a portrait of the painter ([90], page 9).
There was nothing of the kind in the map of 1515. Taken from [90], page 9.

correspondence with the figures from Drers star We shall end up with a curve whose motion from
charts in the Almagest. one constellation to another reflects the order of con
In fig. 2.9 we reproduce a page of the star catalogue stellations in Ptolemys list. It is remarkable that the
from an edition of the Almagest that dates from 1551. resulting curve attains the shape of a spiral that be
In fig. 2.10 one sees a page from the Greek version of gins with Ursa Minor and goes clockwise, up to the
the Almagest that was written in the IX century very end o f the Almagest list. This is precisely where
([1374], page 143). A page from another version of the celestial pole is, qv in fig. 2.1. Right next to it, in
the Almagest (in Latin, dating from the XIII-XTV cen Draco, we have the North Pole as well as the pole of
tury) is reproduced in fig. 2.11. In fig. 2.12 we see a the ecliptic. Let us follow the order of Ptolemys mo
page from George Trebisonds Latin version o f the tion across the celestial sphere as he lists the constel
Almagest (circa 1481 a.d. - see [1374]). It is most lations (see the curve in fig. 2.13).
likely that all these editions hail from the XVI-XVII The curve will obviously be divisible into several
century the earliest. We shall consider the issue of parts. First Ptolemy lists all the constellations num
their dating in the chapters that follow. Let us return bered 1-8, namely, the constellations of Ursa Minor,
to the Almagest star catalogue. Ursa Major, Draco, Cepheus, Bootes, Corona Borealis,
In fig. 2.1 the shaded circle represents the ecliptic. Hercules and Lyra. They are located in the area bor
The wide vertical stripe curved leftwards is the Milky dered by the zodiacal belt on the right and the Milky
Way. O f course, its borders are defined rather ap Way on the left.
proximately and demonstrate the distribution o f the Then the curve proceeds across the Milky Way.
densest parts o f the Milky Way. Inside the regions Ptolemy lists all the constellations included in the Milky
that correspond to constellations we have indicated Way or overlapping with the latter to a great enough
their names and numbers in accordance with the Al extent. Those are Cygnus, Cassiopeia, Perseus, Auriga,
magest. For example, Ursa Minor is the first constel Ophiuchus, Serpens and Sagitta (numbered 9-15).
lation listed in the Almagest, Ursa Major is the sec Next Ptolemy deals with the area that lays to the left
ond, Draco is the third etc. of the Milky Way (its left borderline is defined by the
The Almagest contains twelve named stars, or stars Zodiacal belt, qv in fig. 2.13). He consecutively lists the
that possess proper names. Verbal description of such constellations of Aquila, Delphinus, Equuleus, Pegasus,
stars always contains the formula vocatur (which Andromeda and Triangulum (numbered 16-21).
translates as named). Thus, vocatur Arcturus After that the curve moves on to the Zodiac and
stands for star named Arcturus. All these stars are goes around the centre of the star chart. Ptolemy pro
represented as large black dots in fig. 2 . 1 . They are as vides a list o f all twelve Zodiacal constellations,
follows: Arcturus, Previndemiatrix, Spica, Regulus, namely, Aries, Taurus, Gemini, Cancer, Leo, Virgo,
Acelli, Sirius, Procyon, Lyra = Vega, Cappella, Aquila, Libra, Scorpio, Sagittarius, Capricorn, Aquarius and
Canopus and Antares. We see that most of them hap Pisces (numbered 22-23).
pen to be located to the right o f the Milky Way, on Finally, Ptolemy leaves the Northern Hemisphere,
Milky Way or in its immediate vicinity. Canopus is de crosses the Zodiacal Belt and moves towards the
facto located outside o f the star chart, since the star Southern Hemisphere, listing the constellations in
in question lies very far in the South. the following order: Cetus, Orion, Eridanus, Lepus,
Let us enquire about the order of constellations in Canis Major, Canis Minor, Vela, Hydra, Crater, Cor-
Ptolemys list. This purpose stipulates the compilation vus, Centaurus, Lupus, Ara, Corona Australis and
of a new chart where every constellation is replaced Piscis Austrinus (numbered 34-48). This is where the
by the symbolic representation of its centre (a light Almagest star catalogue ends.
circle, qv in fig. 2.13). Obviously enough, constella Thus, Ptolemys order o f constellations is based
tion centres can only be defined approximately, but on a very obvious principle - the self-implied divi
no great precision is needed here, since we are only sion o f the star chart into several regions.
interested in a rough qualitative picture. Let us then We shall refrain from delving deep into the reasons
draw arrows to link adjacent constellations together. why the author o f the catalogue chose to list the con-
62 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

Fig. 2.5. Star chart o f the Northern Hemisphere from a 1551 edition o f the Almagest. Some o f the constellation figures are wear
ing mediaeval clothes, no less. Taken from [543], inset between pages 216 and 217.

Fig. 2.6. Star chart o f the Southern Hemisphere from a 1551 edition of the Almagest. The constellation o f Orion, for instance,
looks like a mediaeval knight. Taken from [543], inset between pages 216 and 217.
64 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

ttrn&rt G G m tn sd *'
vAfUnjptH *>&+*t^liwwWx ~
t,itnXwet'rfm Kj%*4rta*di>rt? ^
m& m i

Fig. 2.7. Star chart o f the Northern Hemisphere, compiled by the astronomer Bode in the XVIII century according to Ptolemys
Almagest. Published in C laudius P tolem aeu s B eob ach tu n g u n d B eschreibu n g d e r G estirne by J. E. Bode, 1795, page 238. Taken
from [544], Volume 4, inset between pages 184 and 185.

Fig. 2.8. Star chart of the Southern Hemisphere, compiled by the astronomer Bode in the XVIII century according to Ptolemys
Almagest. Published in C laudius P tolem aeus B eob achtu n g u n d B esch reib u n g d e r G estirne by J. E. Bode, 1795, page 238. Taken
from [544], Volume 4, inset between pages 184 and 185.
66 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

sphere to the left o f the Milky Way, which lays to the

iSo MAGNAE COMPOSITIO right of the Zodiac in fig. 2.14.
HU CU f t OliJMf I Miffl***1*
Alma**, Uwodwm. As we shall see below, such division o f the Alma
gest star atlas is anything but random and possesses
"cAdfeilxo&ictpyrocondU. several remarkable properties that permit a deeper
OFftfcfc* * & m t <flmrf*f*uftr*it*jA 'HS^or^r#? understanding of the statistical characteristics of the
* miuilpli<W*fo<6df4l*t . It *
Almagest star catalogue.
t | f ^ l <g||tiftW^rMebnf^ i A N !* *!*
occurs {Owi<&.*? Let us point out the specific and rather interest
wforifetr " |* n|boM<j
^ i f t i i i p 5 S 3 E H iaikm iorfir*
C^Mg<ftmic4totK>r^forftm "
^ j^ n
*rio|W If^R|
V ing manner of constellation listing characteristic for
Ik^feta^miorJCtievjUirur Jiju olbof-Uf*M the Almagest. For instance, the compiler of the cata
logue would be perfectly justified to list the events
moving in a spiral and shifting between parts A and
it tAaHwiitftgtiHgdu*r6(foHk*bomtij By making circular periodic movements around the
L~ i8 < ^ BB5ar
j S c q u r a t t ^ ^ I** pole. However, Ptolemy opts for a completely differ
IB cm lte itN w eiei8p icid m t6*|iii
T O w raaS ...._ ' t*~ ent approach. First he lists the constellations that lay
'-^ - m m ism i to the right of region Ai, then the constellations of that
\i 1m
4*|4 actual region, followed by the ones found on its left,
'S&& gg j ^ P i s ^ ^ the Zodiacal constellations, and, finally, the southern
I il, i i | i i t M i i i M i i m 'I stars. Ptolemy must have had some motives o f his
f ........... . - I" own that have led to this particular choice; the nature
P: i A i i ^ ^ of his motivation is however of little importance to
jPrc<dcra<ktftb m fp k o A ^
us. We are interested in the result - namely, the ac
* ** f o i p S ^ t a S U t ^ t ^ *m*.j* " * \A* a|
4 }oM
H tual method of listing stars as chosen above.
riwpt, i, JS w p ifM ^n w i_ (lofxdefum at 40! It is very important (and nowhere near obvious)
i j<>*mp^offK>da^ r , *M* 1

u f p U k b K I K^ufpohKm6xiuKYpo^if
jU*nA> t* liorcrirtdhnqu*mtmwiptkWo *** *
* that the division of the Almagest star atlas into regions
}nmUtr* tj Auih*^dbtfltt __ ' <, it* to is very closely linked to different precision charac
(OwdatoctpiiB LJSjMtel^asHTr, teristics of said regions.
Fig. 2.9. Fragment of the star catalogue from a 1551 edition As we have already pointed out, specialists adhere
of the Almagest. to different opinions in re the identification o f some
Almagest stars. The table reproduced in [1339] con
tains a list o f all discrepancies between the opinions
stellations in this particular way - let us simply point o f the five most prominent researchers and com
out the naturally occurring regions that the Almagest mentators of the Almagest. But what does the very fact
star atlas can be divided into (see fig. 2.14). of there being such discrepancies between the iden
Region M is the Milky Way, which divides the sky tifications o f different Almagest stars tell us?
into two parts. Then we have region A, which is the It tells us that the coordinates o f the star with sev
part o f the celestial sphere that lays to the right o f the eral different identifications were not measured with
Milky Way and goes up unto the very Zodiacal belt, sufficient precision by Ptolemy. Since the stars of the
comprising the right part of the latter. Region A con first and second magnitude constitute a minority, the
tains a part that consists of Zodiacal constellations ex rest can only be identified by the coordinates indi
clusively; we shall indicate it as Z od A cated in the Almagest. They need to be compared to
Next we have region B - the part o f the sky to the the coordinates of the modern stars in order to find
left o f the Milky Way that reaches up to the zodiacal a fitting equivalent on the celestial sphere. Obviously
belt and includes some of the latters left part - thus, enough, this method, which is often the only one
the part o f this region that consists o f Zodiacal con available for the identification of an unnamed and rel
stellations exclusively shall be labelled Zod . Finally, atively dim star, works well only in cases where Ptol
region D is the southernmost part o f the celestial emy had measured the coordinates of the star in ques-

aa-z.VLCi lt n n

Fig. 2.10. Greek version o f Ptolemys Almagest, allegedly Fig. 2.11. Latin version o f the Almagest, allegedly dating from
manufactured in the IX century. Taken from [1374], page 143. the X III-X IV century. Taken from [1374], page 146.

tion with sufficient precision. If there were serious sider some fixed constellation, the proportion o f du
errata in the process o f taking measurements, there biously identifiable stars that it contains shall demon
may be several identification options. strate how many stars in this constellation werent
The situation becomes particularly complex when measured with sufficient precision. The calculation of
the star under study is part o f an agglomeration of these proportions makes it possible to estimate just
stars whose brightness is more or less uniform. There how precisely Ptolemy measured the coordinates of
may be many different identifications o f a single Al the star in question.
magest star; the choice between them shall be hard Thus, we can select the percentage o f dubiously
to make. identifiable stars as the precision criterion o f Ptole
This is the reason for the controversial identifica mys observations for a given constellation. In other
tion o f certain Almagest stars. words, we need to calculate the value o f ( X /T ) x 100%
The final version o f identifications as cited in for every constellation, where T stands for the sum
the catalogue o f Peters and Knobel ([1339]) may have total o f stars and X - for the number o f dubiously
a greater or a lesser priority as compared to the opin
ions o f other researchers. We shall so far refrain from
discussing this issue in greater detail, since it is quite
beyond the scope o f our research. One finds the sci
entific accuracy o f Peters and Knobel most laudable
- they have diligently listed all the discrepancies be
tween different identifications in a single table. We
shall use this table in order to perform a few simple
yet extremely useful calculations. They give us the
opportunity to make important corollaries concern
ing the precision o f Ptolemys stellar coordinate meas
urements for different parts o f the celestial sphere.
The above permits the acceptance o f the hypoth
esis that if some Almagest star cannot be identified
unequivocally, its coordinates in the Almagest must Fig. 2.12. Another Latin version o f the Almagest, translated
contain errors. We can refer to such stars as dubiously into Latin by George Trebizond around 1481. Taken from
identifiable or poorly identifiable. Thus, if we con [1374], page 147.
68 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

identifiable stars contained by the constellation in

The end result shall accumulate a great deal of pre
liminary work conducted by the previous researchers
of the Almagest. There was a great deal of such re
search, therefore one has every reason to assume that
the average result of their activities may be considered
to represent a more or less veracious picture unaf
fected by the subjectivism o f certain specialists.
We have researched this issue and compiled our re
sults into table 2.2. This table contains eight columns.
In the first column one finds the number o f the
constellation as listed in the Almagest.
The second column contains a reference to the part
o f the celestial sphere where the Almagest constella
tion in question is located. Let us remind the reader
that there are seven such regions (we dubbed them
A, Zod A, B, Zod Bf C, D and M, qv in fig. 2.14).
The third column contains the name o f the con Fig. 2.13. An illustrative presentation o f the order in which
stellation (in Latin). Ptolemy lists the constellations in the Almagest. Constellation
The fourth column informs us o f the percentage of centres are marked by white points in our scheme.
poorly identifiable stars in the pure constellation
(sans informata).
In the fifth column the above percentage is calcu
lated for all the stars in a constellation, the informata
The sixth column contains the percentage of poorly
identifiable stars in the actual informata.
The seventh column contains the number o f stars
in a constellation.
The eighth column contains the number of stars in
the respective informata. Columns 5 and 6 are blank
in cases where there are no inform ata in a constella
tion, with zero in column 8. Table 2.2. lists all 48 con
stellations mentioned in the Almagest.


Our analysis o f table 2.2 implies the following:

Fig. 2.14. Approximated scheme o f the well-measured and
Corollary i . The seven regions that we mention
badly-measured celestial areas from the Almagest. One can
in section 2 contain the following Almagest constel plainly see that only some o f the areas are characterised by ac
lations: curate measurements and therefore stand out. The white area
- region A: constellations 1-8 and 24-29; was measured best in the Almagest.

Constel Percentage o f poorly-identifiable stars Number o f stars

Almagest Latin name of
lation In a pure In a constellation In the In a pure In the
celestial area the constellation
number constellation with informata informata constellation informata

1 A Ursa Minor 0 0 0 7 1
2 A Ursa Major 3.7 11.4 38 27 8
3 A Draco 0 - - 31 0
4 A Cepheus 0 7.7 50 11 2
5 A Bootes 27.3 26 0 22 2
6 A Corona Boreal. 0 - - 8 0
7 A Hercules 10.3 10 0 29 1
8 A Lyra 10 - 10 0
9 M Cygnus 0 0 0 17 2
10 M Cassiopeia 23 - - 13 0
11 M Perseus 3.8 6.9 33.3 26 3
12 A, M Auriga 21.4 - - 14 0
13 M Ophiuchus 25 20.7 0 24 5
14 M Serpens 0 - - 18 0
15 M Sagitta 0 - - 5 0
16 B Aquila 22.3 13.3 0 9 6
17 B Delphinus 20 - - 10 0
18 B Equuleus 100 - - 4 0
19 B Pegasus 10 - - 20 0
20 B Andromeda 13 - - 23 0
21 B Triangulum 0 - - 4 0
22 ZodB Aries 0 0 0 13 5
23 ZodB Taurus 21.2 25 36.4 33 11
24 ZodA Gemini 5.6 20 57 18 7
25 ZodA Cancer 0 23 75 9 4
26 ZodA Leo 11.1 17.1 37.5 27 8
27 ZodA Virgo 15.4 15.6 16.6 26 6
28 ZodA Libra 0 23.5 44.4 8 9
29 ZodA Scorpius 4.8 12.5 66.7 21 3
30 ZodB Sagittarius 12.9 - - 31 0
31 ZodB Capricornus 3.6 - - 28 0
32 ZodB Aquarius 26.1 24.4 0 42 3
33 ZodB Pisces 5.8 5.2 0 34 4
34 D Cetus 22.7 - - 22 0
35 D Orion 8.9 - - 38 0
36 D Eridanus 26.4 - - 34 0
37 D Lepus 0 - - 12 0
38 D Canis Major 5.6 41.3 100 18 11
39 C Canis Minor 0 - - 2 0
40 C Argo Navis 68.9 - - 45 0
41 C Hydra 16 22.2 100 25 2
42 C Crater 57.1 - - 7 0
43 C Corvus 0 - - 7 0
44 C Centaurus 81 - - 37 0
45 C Lupus 100 - - 19 0
46 C Ara 100 - - 7 0
47 D Corona Austr. 100 - - 13 0
48 D Pisces Austr. 8.3 38.9 100 12 6
Table 2.2. Percentage o f poorly identifiable stars in Almagest constellations.
70 I h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

- region B : constellations 16-23 and 30-33; This is the very reason why we introduced two
- region Zod A, which is part o f region A: constel separate columns in table 2.2 - one corresponds to
lations 24-29; the share o f poorly identifiable stars in the pure
- region Zod B> which is part o f region B: constel constellation, and the other - to the main stars o f the
lations 22 ,2 3 , 30-33; constellation with the inform ata added thereto. Our
- region D : constellations 34-38,47 and 48. analysis o f the fourth column demonstrates the pic
- region C: constellations 39-46; ture to be completely different here. Apart from the
- region M: constellations 9-15. pure constellations that were measured with rela
Corollary 2. The stars that constitute the infor - tively high accuracy, there are some whose stellar co
m ata in the Almagest were measured with compara ordinates are less accurate.
tively low precision, with the exception o f the follow For greater demonstrability, we have transcribed
ing: 1 star in Ursa Minor, 1 star in Bootes, 1 star in Her the numeric data from the fourth and the fifth col
cules, 2 stars in Cygnus, 5 stars in Ophiuchus, 6 stars umn in the following manner:
in Aquila, 5 stars in Aries, 3 stars in Aquarius and 4 Inside each o f the constellations reproduced as a
stars in Pisces, or 9 informata out o f the total of 22. certain area confined within a zigzagging border there
The remaining thirteen inform ata were measured are two numbers. The fractions nominator represents
very badly. Indeed, we find 38% o f poorly measured the share o f poorly measured stars in the current
stars in the inform ata o f Ursa Major, 50% in the in pure constellation, sans the informata. The fractions
form ata o f Cepheus, 33.3% in the inform ata o f Per denominator contains the percentage of poorly meas
seus, 36.4% in the informata o f Taurus, 57% in the ured stars together with the informata. There is no
informata of Gemini, 75% in the informata of Cancer, denominator if the constellation in question contains
37.5% in the informata of Leo, 16.6% in the informata no inform ata; however, the fraction line is nonethe
o f Virgo, 44.4% in the inform ata o f Libra, 66.7% in less present. The dotted line one sees in fig. 2.15 rep
the inform ata o f Scorpio, and 100% in the inform ata resents the Milky Way.
of Canis Major, Hydra and Piscis Austrinus. In order to facilitate the analysis o f the above pic
And so, there are lots o f poorly measured stars in ture, let us count the average share o f poorly identi
the inform ata o f the Almagest in general. It would be fiable stars separately (for each o f the seven regions
apropos to voice the hypothesis (one that doesnt af as described above). We shall add up the previously
fect our further research in any way at all, as a mat calculated rates for every constellation and divide the
ter of fact) that the stars collected in the informata did result by the number o f constellations in the region.
not constitute the primary constellation pattern, The result is represented in table 2.3.
which is why the measurement o f their coordinates Let us turn to fig. 2.16, where different regions are
was performed with less precision - especially if the represented by different kinds o f shading. They cor
star in question was a dim one. O f course, if a bright respond to varying levels o f observation quality.
star ended up among the inform ata, its coordinates White colour stands for values between 0% and 5%
could be measured with greater diligence. For in o f poorly measured stars. Dotted shading represents
stance, the famous Arcturus is part of the well-meas values falling between 6 % and 10%, slanted shading
ured informata o f Aquarius. However, table 2.2 shows - values between 21% and 30%, and, finally, black
us that in a typical situation the stars o f the informata field stands for values between 31% and 100% o f stars
are measured with less precision than the stars o f the whose coordinates lack precision.
pure constellation. Thus, the darker a given area, the worse the qual
It would therefore strike one as natural to separate ity o f its measurement in the Almagest. We instantly
the informata from the main stars of the constellation notice the fact that many austral constellations in
for the time being. Actually, this is how it is done in Area C, to the right o f the Milky Way, are measured
the Almagest - the inform ata stars are gathered in a very poorly indeed - we see a lot o f solid black shad
separate eponymous group. We shall consider the ing here, qv in fig. 2.16. On the other hand, the con
pure constellations alone. stellations in Area A are measured a great deal better,

Fig. 2.15. Inside each o f the constellations mentioned by Ptolemy and drawn as an area with zigzagged boundaries we specify
two numbers, the first one corresponding to the percentage o f poorly-measured stars in a constellation without informata, and
the lower - to the same in a constellation with the informata added.
72 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

precise definitions o f constellation borders, neigh

bouring constellations may become stretched in
such a way that they will fill the empty zones in fig.
2.16. We shall refrain from describing this procedure
in greater detail - there are few such blank spots, and
they hardly influence our results in any way at all.
For a more illustrative analysis of the above picture,
let us calculate the average percentage of poorly iden
tifiable stars in each of the above seven areas individ
ually by adding up the percentages calculated above
for each of the constellations and dividing the sub by
the total number of constellations for each area. The
result is represented in table 2 .3 .
Corollary 3. Region A is measured better than re
gions B, Cy D and M in the Almagest - namely, 6.3%
o f poorly identifiable stars in pure constellations
and 12 .6 % in constellations with added informata.
Corollary 4. Region B is measured worse than
region A in the Almagest, namely, we have 19.6% of
poorly identifiable stars in the pure constellations
Fig. 2.16. A demonstrable representation o f well-measured and and 19% in the constellations with the informata.
poorly-measured celestial areas from the Almagest. The darker Corollary 5. Region M, or the Milky Way, occu
the area, the less accurate the corresponding measurements. pies an intermediate position between regions A and B
- 10.5% of poorly identifiable stars in pure constel
lations and 10.3% in the constellations with informata.
there is a lot of white here. Area B , which lays to the Corollary 6. Regions C and D are measured the
left o f Area M, is measured worse than Area A, we see worst in the Almagest - namely, region D contains
a good deal o f double shading. Some of the areas in 27.4% o f poorly identifiable stars in pure constel
fig. 2.16 are marked with a question mark - they are lations and 36.9% in constellations with inform ata
the regions o f the modern celestial sphere that for added. For region C the percentage o f poorly identi
mally remain beyond the confines o f the Almagest fiable stars equals 52.9% in pure constellations and
constellations. Seeing as how the Almagest gives no 53.6% in constellations with informata.

Parts o f the celestial sphere A w/o B w/o

in the Almagest A B ZodA ZodB ZodA ZodB D C M

N um ber o f constellations 14 12 8 6 6 6 7 8 7

1-8, 16-23, 2 2 ,2 3 , 34-3 8 ,

Constellation num bers in the Almagest 1-8 16-21 24 -2 9 3 9-46 9-15
24 -2 9 30-33 30-33 4 7 ,4 8

Percentage o f poorly identifiable stars

6,3 19,6 6,4 27,6 6,2 11,6 27,4 52,9 10,5
in pure constellations (w/o informata)

Percentage o f poorly identifiable stars

12,6 19 8,1 26,5 18,6 11,9 36,9 53,6 10,3
in constellations w ith informata

Percentage o f reliably identifiable stars

93,7 80,4 93,6 72,4 93,8 88,4 72,6 47,1 89,5
in pure constellations

Table 2.3. Average percentage o f poorly identifiable stars as given for each o f the seven areas individually.

Corollary 7. Region Zod A is measured best in C:5p J

the Almagest - it is the part o f the Zodiac on the right
o f the Milky Way. It includes the constellations o f
Gemini, Cancer, Leo, Virgo and Scorpio. Here we have
The Southern j
Hemisphere ]
a mere 6 .2 % o f poorly identifiable stars in pure
Corollary 8. Region Zod B is measured much
worse than Zod A. Here we have 11.6% of poorly iden
tifiable stars in pure constellations. Region Zod B
comprises the constellations of Sagittarius, Capricorn,
Aquarius, Pisces, Aries and Taurus.
In order to get a better idea o f what the informa
tion in table 2.3 really stands for, we have drawn a di
agram, which is reproduced in fig. 2.14. Different
kinds o f shading correspond to different levels of
measurement precision, or the percentage o f dubi
ously identified stars. The white zone stands for areas A~zm
that contain 0% to 10% o f such stars, dotted shad
Fig. 2.17. Percentage o f dubiously identified stars in the pure
ing corresponds to levels o f 10 % - 20%, linear shad
constellations o f the Almagest, without accounting for the
ing - to those o f 20% -30% , and double shading rep stars listed in the informatae. It is quite obvious that the stars
resents zones o f the celestial sphere that contain 30% from group A were measured the best, and the percentage of
to 100% o f stars whose identity is ambiguous. dubious stars here is the lowest.
Another illustrative representation o f the above
information can be seen in fig. 2.17. The numbers o f
all 48 Almagest constellations are placed horizontally Corollary 9. The first primary statement. The
in such a way that they form groups, such as A, B, seven regions o f the Almagest star atlas that we have
Z od A, Z od B, A - Z od A (A without Zod A, that is), discovered differ by the precision of stellar coordinate
B - Zod B, C, D and M. The respective percentage o f measurements. Indeed, different kinds o f shading
dubiously identified stars in pure constellations is correspond to the seven celestial regions as described
aligned vertically. Each o f the constellation groups as above (A, B, C, D, M, Zod A and Z od B) in fig. 2.14.
listed above is represented by a certain horizontal seg Corollary 10. The second primary statement.
ment in fig. 2.17 - the average percentage value for 1) Further research o f star coordinates in the Al
the group under consideration. Fig. 2.17 makes it per magest has to be based on the stars from region A first
fectly obvious that the coordinates of stars in group and foremost, since it is the most accurately measured
A were measured with maximum precision (regions region with a minimum o f dubiously identified stars.
A, Zod A and A - Zod A). Corresponding values are 2) One mustnt base any corollaries on the study
the smallest. Group B is located much further up o f the stars from regions C and D. An exceptionally
in fig. 2.17, which stands for lower measurement pre large number o f poorly identifiable stars in this area
cision in this area. It is also apparent that the stars of tells us quite explicitly that the regions in question
the Southern Hemisphere were measured even worse. cannot be considered reliably measured. Refraction
The same information can be found in fig. 2.18, is one o f the reasons why the southern stars could not
which is based on the last line o f table 2.3, where the be measured with sufficient precision by the author
dubiously identified star percentage values in pure o f the Almagest - it is common knowledge that the
Almagest constellations are aligned vertically. This coordinates o f the stars located close to the horizon
graph is obviously implied by the graph in fig. 2.17 are affected by light refraction.
and represents the values o f the latter subtracted 3) We get the opportunity to differentiate the list
from 100%. o f 12 named stars by the level o f their reliability. The
74 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

stars measured with the greatest accuracy

correspond to region A and its immediate
vicinity. They are Regulus, Spica, Previnde-
miatrix, Procyon, Arcturus, Acelli, Anta-
res, Lyra (Vega), and Capella. The am
biguous stars are Sirius (region D), Aquila,
or Altair - region B, left border of the Milky
Way, and Canopus, which is altogether off
the chart. These stars ended up in the
poorly measured celestial regions.
Incidentally, the star Previndemiatrix
also has to be excluded from the list o f
good named stars for the following rea
son. Although this star can be identified
quite well (in particular, it is absent from
the list o f poorly identifiable stars, qv in
table 6 in [1339]), its coordinates as given Fig. 2.18. Percentage o f reliably identified stars in the pure constellations of
in [1339] are rather uncertain and not the Almagest.
substantiated with any references to the
original Almagest manuscripts. Peters re
ports the following about the coordinates o f the star The phenomenon o f refraction owes its existence
Previndemiatrix in the Almagest: Greek sources in to the properties o f the atmosphere that affect the
dicate 2010', and the Arabs - 1510' [a discrepancy measurements conducted from the surface o f the
o f five degrees, no less - Auth.]. Ulugbeks catalogue Earth; the latter is the case with all the ancient ob
contains the coordinates o f 1615'. Peters states 160', servations. From the mathematical point of view, the
following the catalogue o f Halma, likewise Bailey - atmosphere o f the Earth can be regarded as a set of
however, he points out that Halma gives no author concentric spherical air layers whose density is more
itative references. It is clear that Halmas 160' were or less uniform, changing from layer to layer.
taken from Halley, which is correct [?! - Auth.] but It is common knowledge that a ray o f sunshine is
not supported by any manuscripts ([1339], page subject to refraction as it moves between different
104). It is clear that a situation as ambiguous as this atmospheric layers o f different density (see fig. 2.19).
one requires the star Previndemiatrix to be excluded The ray becomes more vertical as a result, approxi
from further consideration. mating the normal, which is the perpendicular bor
Thus, eight out of twelve named stars o f the Alma der o f two layers.
gest end up in the reliably measured region o f the In fig. 2.20 we see a diagram o f the Earths at
celestial sphere: Regulus, Spica, Procyon, Arcturus, mosphere, presented as a set o f concentric layers
Acelli, Antares, Lyra (Vega), and Cappella. whose density diminishes as altitude grows. A ray of
light that comes from star A refracts as it moves from
4. one layer to another. As a result, it moves through
POSSIBLE DISTORTION OF THE STAR the atmosphere forming a certain curve that can be
COORDINATES RESULTING FROM calculated from the corresponding equation. This was
THE ATMOSPHERIC REFRACTION done in the theory o f atmospheric refraction. The re
sult is shown in fig. 2.20 - the observer located in
A researcher o f a star catalogue must always re point O on the surface o f the Earth perceives star B
member the physical phenomenon o f refraction, as part o f half-line OBywhile in reality the direction
whose influence can greatly distort the coordinates of is represented by half-line OA Therefore, refraction
the southern stars. lifts stars in a certain way.

Fig. 2.19. Refraction o f a ray o f light at the boundary between Fig. 2.20. Atmospheric refraction can distort the visible posi-
two different environments. tion o f a star on the celestial sphere.

The closer a star happens to be to the horizon, the reason why southern stars, which hang low above the
longer it will take a ray o f light to get through the at horizon, were measured rather badly in the Almagest
mosphere o f the Earth and the greater the elevation and the ancient catalogues in general.
o f the star. However, if the star is situated high We have already been confronted by this fact in
enough, the distortion o f its position shall be negli section 3, having witnessed the fact that the percent
gibly small. The theory o f refraction has an approx age o f poorly identifiable stars in regions C and D,
imated expression that characterises the refraction of which correspond to the southern part o f the celes
zenith distances - namely, stellar zenith distance or tial sphere, happens to be much higher than in regions
the angle between the direction o f zenith at the point A and B.
o f observation and the star direction, minus the value It would be apropos to remark that the phenom
approximately expressed in the following formula enon o f refraction was unknown to the ancient as
(for q < 70): tronomers, and even upon its discovery the precise
compensation o f refraction remained a formidable
B 273 problem - one that was only successfully solved in the
P tli
70 273*+ f* epoch o f Tycho Brahe. However, as it is mentioned
in [65] (page 129), Tycho Brahes compensation cal
stands for the zenith distance, B is the height of culations were rather far from perfection.
the barometers mercury column at the moment o f
observation rendered to 0 centigrade, and f is the 5.
air temperature in degrees (centigrade) at the obser THE ANALYSIS OF THE INFORMATA
vation location. The above formula demonstrates that DISTRIBUTION ACROSS THE ALMAGEST
the main variable component that affects refraction CATALOGUE
is tan. If the zenith distance is small (and the star is
high enough above the horizon), the value o f tan is Table 2.2 contains the information about the dis
small also, and the refraction is insignificant. tribution o f the inform ata across the Almagest con
As the stars get closer to the horizon, the value o f stellations. The table demonstrates that many con
component tan grows, and refraction distorts stel stellations possessed no inform ata at all - namely,
lar coordinates to a greater extent. This must be the only 22 Almagest constellations out of 48 possess in-
76 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

form ata. What is reflected in the presence or absence constellations, which constitute a separate group. We
of inform ata stars in a given constellation? There may are referring to constellation groups A, B, M etc.
be many opinions on this issue. The one we consider Therefore, for each o f the seven regions o f the star
to be the most plausible is as follows (it can be for chart discovered above we shall calculate a certain
mulated in brief as the following hypothesis): quantitative characteristic - the average share o f in
The inform ata were only indicated for the con form ata stars in a given group. The higher the share,
stellations that Ptolemy believed to be the most im the more stars ended up as informata.
portant. The result is represented graphically in fig. 2 .2 1 .
In other words, the very presence o f inform ata in We are following the same principle here as in
a constellation signifies that the astronomer was par fig. 2.17, namely, placing the numbers o f Almagest
ticularly interested in said constellation. constellations grouped by region (seven regions all in
It is possible that certain constellations were o f all, qv in fig. 2.17) on the horizontal axis. The aver
particular importance and therefore marked as such age share o f stars in the inform ata is indicated on the
on the celestial sphere. We do not ponder the reasons vertical axis. As a result, there is a horizontal segment
why there was an emphasis on these constellations - that corresponds to each area.
these reasons are o f no importance to us and may The information in fig. 2.21 has the following im
have been o f an astrological nature, for example. The portant implication.
stars o f such constellations would therefore be meas Corollary i . The distribution of ''informata den
ured several times for greater observation precision. sity in the Almagest star catalogue is in perfect con
Also, it might be that the observer, upon listing the currence with the distribution o f dubiously identified
stars that form the actual constellation figure, or the stars in the pure constellations o f the Almagest.
stars o f the pure constellation in our terminology, The same corollary can be reformulated as follows.
added some o f the background stars thereto - that The more attention was paid to one o f the constella
is to say, the stars that do not constitute the constel tion groups by the compiler o f the catalogue, the more
lations skeleton, but rather happen to be located in trustworthy the identity o f the stars in this group.
its immediate vicinity. This is how the inform ata may Indeed, as we can see in fig. 2.21, the highest den
have come into existence. sity of the informata can be observed in region Zod A.
As we already know, these stars (most probably Next we have region A. Furthermore, region A was
regarded as secondary) could be meas
ured worse on the whole than the stars % 1 I
o f the main constellation. 30 Z o iA -W 1 !
i 1 !
It would be interesting to observe the 1 1 |
distribution o f the inform ata across the The Northern Hemisphere end die zodiac j The Southern Hemisphere j
star chart o f the Almagest. | 1
r>:X-7< I
In order to provide a quantitative t. * . 1
20 I 1
characteristic o f this distribution, let us f l
do the following. We shall calculate she A : ';::'::- ;:' 1 !
1 1
share o f the inform ata stars for each of 1 i
the Almagest constellations - otherwise,
the value o f c = (a / b) x 100%, where a W a* !
stands for the number o f informata stars I*-;;* v.
and b for the full number o f stars in a
constellation with the inform ata added
Thus, if there are no inform ata stars Fig. 2.21. The distribution of inform ata density in the Almagest star cata
in a constellation, c - 0. Next let us cal logue. We can see that this density is in perfect concurrence with the distribu
culate the full share of inform ata in all tion of dubiously identified stars in the pure constellations of the Almagest.

clearly studied more attentively than region B. Region These variances were discovered in the 26 primary
M was the least accurately measured part o f the ancient manuscripts of the Almagest. Table IX in
Northern Hemisphere. Regions A and B were ob [ 1339] contains all such versions. The following man
served with greater diligence than region M. uscripts were used in its compilation (see Chapter 11
The least attention was paid to region C in the for an exhaustive list o f sources):
Southern Hemisphere. Although region D, also lo
cated in the Southern Hemisphere, enjoyed more at Greek manuscripts :
tention from the part o f the Almagests compiler 1) Paris 2389,
(poorly identifiable stars amounting to 10 .2% here), 2) Paris 2390,
this wasnt the case with region C (see fig. 2.17). Little 3) Paris 2391,
wonder - regions C and D comprise the southern part 4) Paris 2394,
o f the Almagest star atlas, which is characterised by 5) Venice 302,
lower observation precision on the whole than the 6 ) Venice 303,
stars o f the Northern Hemisphere and the Zodiacal 7) Venice 310,
constellations, as we have already mentioned repeat 8) Venice 311,
edly. Therefore, southern regions C and D must hence 9) Venice 312,
forth be considered separately and cannot be used in 10) Venice 313,
any conjectures due to low observation precision. 11) Vatican 1594,
Thus, figs. 2.17 and 2.21 lead us to an important 12) Vatican 1038,
conclusion. 13) Vat. Reg. 90,
Corollary 2. The above analysis confirms the 14) Laurentian 1,
previously discovered division o f the Almagest star 15) Laurentian 47,
atlas into seven regions of varying precision. Obser 16) Laurentian 48,
vation precision for each o f them is proportional to 17) Bodleian 3374,
the amount o f attention paid to this region. We are 18) Vienna 14.
primarily referring to the Northern Hemisphere and
the Zodiac. The higher the density o f the inform ata, Latin manuscripts:
the better the measurements o f the stars and the 19) Laurentian 6 ,
higher the percentage o f reliably identifiable stars. 20) Laurentian 45,
The lower the density of the informta, the smaller the 21) Vienna 24,
value corresponding to the percentage o f reliably 22 ) British Museum Sloane 2795.
identified and recognizable stars. Detailed numeric
data concerning individual Almagest constellations is Arabic manuscripts:
cited in table 2.4 o f Section 6 , and this is the source 23) British Museum 7475,
that the reader may refer to. The share o f inform ata 24) British Museum Reg. 16,
is indicated for each and every constellation. 25) Bodleian 369,
26) Laurentian 156.
THE ANALYSIS OF THE COORDINATE VERSIONS Table IX in [1339] contains 26 vertical columns
AS SPECIFIED IN DIFFERENT MANUSCRIPTS OF corresponding to the above manuscripts o f the Alma
THE ALMAGEST CATALOGUE. gest. Each row o f the table corresponds to some star
Comparison of the 26 primary manuscripts from the catalogue whose coordinates differed from the
to the canonical version of the catalogue canonical version. The table makes a very chaotic im
pression, since the versions are distributed randomly.
The work o f Peters and Knobel ([1339]) contains We must point out an important detail. Numbers
Table IX, where we see data that are at odds with the (or versions) found in a single line o f the table may
commonly used canonical version o f the catalogue. coincide with each other, which means that several
78 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

manuscripts contain the same version (of the stars o f these copies is presumed to be the end product of
longitude, for instance) that differs from the canoni a certain copy tree rooted in the lost original o f the
cal version. Almagest.
Let us consider an example, assuming that the At the same time, it is possible that the catalogue
longitude o f 1610' is mentioned four times in a sin wasnt merely copied, but rather complemented by
gle table row, whereas the longitude o f 1620' is in new observations conducted in the epoch o f the scribe.
dicated in seven table cells. If we are to assume fur New coordinates could be introduced into the cata
ther that there are no other longitude versions in logue as a result - the ones that the mediaeval re
said table row, there will be exactly two longitude searcher believed to be more precise than the originals.
values that differ from the canonical in all 26 above- It is therefore possible that the surviving versions of
mentioned manuscripts. We have simply considered the catalogue have reflected both kinds o f discrepan
the num ber o f versions here, regardlessly o f the cies - mechanical errata o f the scribes as well as the
number o f repetitions - a more in-depth study would results o f independent star observations and repeated
be very useful indeed. The total number o f different coordinate measurements. Which versions constitute
stellar longitude versions (with repetitions) appar the majority? Which o f the two versions that we for
ently equals 7 + 4 = 1 1 . mulate below happens to be closer to the truth?
Both numeric characteristics are important to us. 1) Contradictory versions we have at our disposal
The former is geometric and demonstrates the num today are nothing but errata introduced by the scribes.
ber o f different dots, or stars, which have to be drawn 2) Discrepancies between versions are primarily a
on the celestial sphere in order to account for all the result o f repeated independent measurements o f star
versions o f this stars coordinates contained in the coordinates conducted by a single observer (or group
manuscripts. The second characteristic corresponds o f observers) during a single epoch. The estimation
to manifestation frequency o f a given version. It is ob o f the epoch is a separate task.
vious that the more manuscripts insist on a single In other words, is it possible that the differing ver
version, the more reasons there are to try and find sions we have today arent necessarily copies o f the
out why this particular version happens to be so pop source catalogue - some are drafts, which were used
ular. for the compilation o f the catalogues final canonical
Table IX is very voluminous as per [1339], and so version. In order to find out which o f the two postu
there is hope o f finding certain tendencies that will lations is closer to the truth, we have processed table
be useful to our research. IX in [1339] and collected the results in table 2.4. Let
According to the Scaligerian viewpoint, the ver us comment on the principle o f our tables con
sions collected in Table IX ([1339]) result from scribes struction. It contains seven columns and 48 rows.
errata that have accumulated over the centuries as the The first column contains the constellation num
Almagest was copied many a time. The original o f the bers according to the list in the Almagest.
Almagest is presumed to have been lost a long time The second column contains the name o f the con
ago, and has only reached us as several mediaeval stellation (with the sum total o f stars in the constel
copies. Each o f the following copyists introduced new lation indicated in parentheses).
errata while copying the previous copy. As a result, we In the third column we have the number o f stars
have several versions of the catalogue today. O f course, in the informata o f the constellation in question (with
there could be errors made in the course o f copying, 0 used for constellations without the informata). The
since digits were transcribed as letters back then. Some percentage value of stars in a constellation comprised
letters can easily be confused for each other. This by the inform ata is indicated as well.
would lead to a certain distortion o f the original nu In the fourth column we see the full number o f
meric material. To sum up, we could say that Scaliger versions for longitudes and latitudes, as well as rep
ian history considers the differing manuscripts o f the etition frequency per single version (for the entire
Almagest and its catalogue to be nothing but me constellation with the inform ata included.
chanical copies introduced by different scribes. Each T h e fifth column corresponds to the full number of

Constel Name of constellations Amount of stars in an Number of options for latitudes and longitudes
lation and the amount informata and its in a constellation with informata
numbers of stars in pure percentage in comparison
Full number Average number
in the constellations with the constellation with
Almagest (without informatae) its informata included with miltiplicities w/o miltiplicities with miltiplicities w/o miltiplicities
1 Ursa Minor (7) 1 (12.5% ) 73 29 9.1 3.63
2 Ursa Major (27) 8 (22 .8%) 227 103 6.49 2.94
3 Draco (31) 0 150 89 4.84 2.87
4 Cepheus ( 11) 2 (15.4% ) 60 29 4.62 2.23
5 Bootes ( 22 ) 1 (4.3%) 132 55 5.74 2.39
6 Corona Boreal. ( 8) 0 25 17 3.13 2.13
7 Hercules (29) 1 (3.3% ) 202 79 6.73 2.63
8 Lyra ( 10) 0 49 22 4.9 2.2
9 Cygnus (17) 2 (10.5% ) 95 45 5 2.37
10 Cassiopeia (13) 0 60 28 4.62 2.15
11 Perseus (26) 3 (10.3% ) 87 49 3 1.69
12 Auriga (14) 0 68 35 4.86 2.5
13 Ophiuchus (24) 5 (17.2% ) 213 85 7.34 2.93
14 Serpens (18) 0 92 36 5.11 2
15 Sagitta (5) 0 43 12 8.6 2.4
16 Aquila (9) 6 (40.0% ) 49 36 3.27 2.4
17 Delphinus ( 10) 0 72 33 7.2 3.3
18 Equuleus (4) 0 6 5 1.5 1.25
19 Pegasus ( 20) 0 68 39 3.4 1.95
20 Andromeda (23) 0 78 39 3.39 1.7
21 Triangulum (4) 0 9 5 2.25 1.25
22 Aries (13) 5 (27.7% ) 83 41 4.61 2.28
23 Taurus (33) 11 (25.0% ) 259 110 5.89 2.5
24 Gemini (18) 7 (28.0% ) 192 60 7.67 2.39
25 Cancer (9) 4 (30.7% ) 107 44 8.23 3.38
26 Leo (27) 8 ( 22 .8%) 170 83 4.86 2.37
27 Virgo (26) 6 (18.7% ) 207 87 6.47 2.72
28 Libra ( 8) 9 (52.9% ) 85 39 5 2.3
29 Scorpius (2 1 ) 3 (12.5% ) 56 31 2.33 1.3
30 Sagittarius (31) 0 179 67 5.77 2.16
31 Capricornus (28) 0 217 85 7.75 3.04
32 Aquarius (42) 3 ( 6.6%) 207 109 4.6 2.42
33 Pisces (34) 4 (10.5% ) 246 96 6.47 2.53
34 Cetus ( 22 ) 0 130 54 5.91 2.45
35 Orion (38) 0 212 96 5.58 2.53
36 Eridanus (34) 0 210 81 6.18 2.38
37 Lepus ( 12 ) 0 71 36 5.92 3
38 Canis Major (18) 11 (37.9% ) 88 38 3.03 1.31
39 Canis Minor (2) 0 12 5 6 2.5
40 Argo Navis (45) 0 250 100 5.56 2.22
41 Hydra (25) 2 (7.4%) 209 73 7.74 2.7
42 Crater (7) 0 33 18 4.71 2.57
43 Corvus (7) 0 20 17 2.86 2.43
44 Centaurus (37) 0 179 70 4.84 1.89
45 Lupus (19) 0 133 57 7 3
46 Ara (7) 0 70 24 10 3.43
47 Corona Austr. (13) 0 85 31 6.54 2.38
48 Pisces Austr. 02) 6 (33.3% ) 72 36 4 2

Table 2.4. Number o f options for stellar coordinates in different constellations o f the Almagest.
8o I h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

versions for longitudes and latitudes without repeti 7.

tions given for the entire constellation, informata in VERSION DENSITY AS THE DENSITY
The sixth column is the average number o f differ Seven areas of the Almagest star atlas
ent longitudinal and latitudinal values with number revisited with a new concurrence with the
o f repetitions (per constellation, whole, informata in previous results
The seventh column is the average number o f dif In order to make conclusions from table 2.4 we
ferent versions (longitudes and latitudes) - taken shall perform an additional simple operation - namely,
without repetitions for the entire constellation, in calculating the average amount of stellar coordinate
form ata included. versions for all of the seven areas of varying precision
Let us comment the resulting table. The third col on the Almagest star chart as listed above. For this pur
umn serves as the basis o f fig. 2 .2 1 , which we discuss pose we shall divide the rows o f the last two columns
at length in Section 5. Values from this column cor o f table 2.4 into seven groups (A, B, M etc), and then
respond to informata density distribution in the Al average the values from a single group. The result is
magest star atlas. presented as table 2.5. The fourth row of the table
The principle behind the calculation o f values provides the basis for fig. 2.21 and shows the informata
from columns 4 and 5 is obvious enough. We counted percentage for every celestial region.
the full number o f variations for every star in a given The last two lines o f table 2.5 are the most im
constellation, with all the repetitions included. The re portant for table 2.5. The fifth line shows the version
sults for all stars in this constellation were subse density with multiplicities taken into account, whereas
quently added up. Let us emphasise that our current the sixth provides the same information without mul
objective is to study the distribution o f coordinate tiplicities, or repetitions. Let us turn to fig. 2.22 for a
variations across the entire catalogue. We see that the more demonstrative representation of these data. The
Almagest constellations are anything but uniform in horizontal line contains numbers o f the Almagest
this relation. Some constellations are poor in vari constellations grouped by the seven areas o f the star
ance. It has to be said that we did not consider lon chart, see fig. 2.17. In the vertical we see the average
gitudes and latitudes separately in this research, but amount o f versions for each o f these areas.
rather studied their sum characteristics for more con Tables 2.5 and fig. 2.22 lead us to the following
fident statistical corollaries. corollaries:

Parts of the Almagests A w /o B w /o

celestial sphere A B Z odA Z odB Z odA Z odB M D C

Number of constellations in an area 14 12 8 6 6 6 7 7 8

Compounds of an area (constellation 1 - 8, 16-23, 22,23, 34-38,

1-8 16-21 24-29 9-15 39-46
numbers according to the Almagest) 24-29 30-33 30-33 47,4 8

In form ata percentage in an area 16 9.2 7.3 6.7 31.8 11.6 5.4 10.2 0.9

Average number of versions for latitudes

5.72 4.68 5.69 3.5 5.76 5.85 5.5 5.31 6.09
and longitudes (with multiplicities)

Average number of versions for latitudes

2.53 2.23 2.63 1.96 2.41 2.49 2.29 2.29 2.59
and longitudes (without multiplicities)

Northern constellations and the zodiac Southern constellations

Table 2.5. Average number of versions for latitudes and longitudes in the Almagest constellations.

Corollary 2. Star coordinate density

(K.5J2 122& C:/M7 on the Almagest star atlas concurs per
ZottBi55 ; l fectly with the distribution o f the reli
~Zq$A : v; ;t i*
V.;* n ^ s o j ;1
ably identified stars in pure Almagest
* * rt
5:4,63- . * v -i
' f Average number * 'V -W ;l constellations as well as the inform ata
: of coordinate variants ; <y-V * .*.* ;.v i density distribution.
;f (with multiplicities} *. * ** > * i
.*\7.*;7.i v;-i We present the information which con
. -'I
r. - m ; I ;.v| cerns the distribution of said densities as
: :**:r.*.V* v*i ____ \*r*.**7*T^r
5- four tables - 2 .6 ,2 .7 , 2.8 and 2.9. Table 2.6
A-ZoA'2,63\' :* 4Zo&14$\ ** ***\ *,*i demonstrates the distribution o f safely
* V ' V .. I C 2.5S
:. d5 3 7 i identifiable stars in the pure constellations
\; : v* Average tim b er1
o f coordinate variants of the Almagest. The rows and the columns
(without multiplicities) J*//.; ,.;i
o f the table correspond to the following
The. .Northern Hemisphere and the zodiac* The Southern Hemisphere
. . . . . . . . . . regions that we discover on the Almagest
chart: A, B, A minus Zod A, B minus Zod
* .* * \ :h :
By Zod Ay Zod By My D and C. Three last
Fig. 2.22. Density distribution of stellar coordinate version numbers in the columns and rows of the table refer to the
Almagest catalogue. Densities are given with and without multiplicities. areas of the Southern hemisphere.
The cells of the table contain + and -
signs (or + = / - = , in some cases). Their
C orollary i . The version density graph with meaning is as follows. Let us consider the first row of
multiplicities concurs well to the one without them. the table, for instance, which corresponds to area A.
This implies that the logical patterns listed below The respective percentage is larger for area A than for
manifest in both graphs. Let us point out that the area B; therefore, we put a + on the crossing of the first
density graph without multiplicities has smaller am row and the second column. Furthermore, the per
plitude fluctuations as compared to the density graph centage is formally greater for area A than for A minus
that accounts for multiplicities. This is quite natural, Zod Ay but equal to the latter de facto; therefore, we
since when one includes them, the density fluctua put a + = sign into the respective cell; should this per
tions are observed more realistically; fig. 2.22 demon centage prove smaller, we use - ; if smaller but equal
strates precisely this. de facto, - = .

A w/o B w/o
A B ZodA ZodB ZodA ZodB M D C

A = + += + -= + + -1- +
B - = - + - - + +
A w/o ZodA -= + = + -= + + + +
B w/o ZodB - - - = - - - -= +
ZodA += + += + = + + + +
ZodB - + - + - = -= + +
M - + - + - += = + +

D - - - += - - - = +
C - - - - - - - =

Table 2.6. A comparison of the percentage of reliably identifiable stars in the pure constellations of the Almagest (without infor
mata) for different parts of the celestial sphere.
82 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

A w /o B w /o
A B Z odA Z odB Z odA Z odB M D C

A = + + + - + + + +
B - = + + - - + -= +
A w /o Z odA - - = + - - + - +
B w /o Z o d B - - - = - - += - +
Z odA + + + + = + + + +
Z odB - + + + - = + + +
M - - - -= - - = - +
D - += + + - - + = +
C - - - - - - - - =

Table 2.7. A comparison of informata density for various parts of the Almagest star atlas.

A w/o Bw/o
A B ZodA ZodB ZodA ZodB M D C

A = + += + -= -= + + -

B - = - + - - - - -

A w/o ZodA - + = + -= -= + + -

B w/o ZodB - - - = - - - - -

ZodA += + += + = -= + + -

ZodB += + += + += = + + -

M - + - + - - = + -

D - + - + - - - = -

C + + + + + + + + =

Table 2.8. A comparison of the relative stellar coordinate version numbers for various areas of the Almagest star atlas, with
multiplicities accounted for.

A w/o B w/o
A B ZodA ZodB ZodA ZodB M D C

A = + -= + + + + + -

B - = ~ + ~ - -= - -

A w/o ZodA += + = + += += + + +
B w/o ZodB - - ~ = ~ - - - -

ZodA - + -= + = -= + + -

ZodB - + -= + += = + + -

M - += ~ + - - = -

D - + ~ + - - s = -

C + + -= + + + + + =

Table 2.9. A comparison o f the relative stellar coordinate version numbers for various areas o f the Almagest star atlas, without

Fig. 2.23. A graph where we simultaneously see the following: 1) the distribution of whatever percentage the reliably identified stars
of the Almagest catalogue comprise; 2) the percentage of informatae in various areas of the Almagests celestial sphere, 3) average
number of stellar coordinate options in various manuscripts of the Almagest, with multiplicities, 4) average number of coordinate
options, without multiplicities. One can see that all four density graphs for the Northern Hemisphere correlate with each other well.

The implication is that when we look at table 2 .6 , distribution for the Almagest star atlas, and table 2.8
we can safely tell the comparative percentage o f reli gives us an opportunity to compare the version den
ably identifiable stars for every area pair. Table 2.6 is sity o f the Almagest stellar coordinates for different
a compact representation o f density distribution in celestial areas. The versions that constitute this table
all o f the star chart areas described above. were calculated with multiplicities, which means that
The next three tables are based on the same prin if the same version was encountered several times, the
ciple. Table 2.7 demonstrates the inform ata density entire amount was accounted for accordingly. If we
84 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

are to leave multiplicities out, or just count each ver coordinates o f those stars were indeed measured a
sion once, the result will be a comparative presenta great deal better on the average in Ptolemys epoch.
tion o f the relative coordinate version quantity for Thus, if we are to simplify the situation somewhat,
varying areas o f the Almagest star atlas, qv in table 2.9. one has reasons to presume that the 26 primary man
Tables 2.6-2.9 make it obvious that the distribu uscripts o f the Almagest are for the most part its
tion o f pluses and minuses is virtually equal, which drafts rather than mechanical copies. They were
implies a good correlation between the following four subsequently used for the creation o f the final canon
values: ical text. The Scaligerian version of these manuscripts
1 ) the percentage o f reliably identifiable stars in a origins does not concur with our conclusion. Indeed,
given area o f the Almagest star chart; why would mediaeval scribes copy the drafts to
2) inform ata density in the Almagest star chart gether with the final version for centuries o f end?
area in question; It would make a great deal more sense if we are to as
3) stellar coordinate version density with multi sume that both date to approximately the same epoch,
plicities; and the number o f copies was far from great. Let us
4) stellar coordinate version density without mul reiterate that observations of this manner shall not be
tiplicities. used in our research; they are but a number o f nat
In particular, the higher the informata density and urally arising questions which are to demonstrate
the coordinate version density in a given area, the several possible explanations o f the effect that we dis
more reliable the identification o f the stars located covered, nothing more.
therein. Finally, let us cite fig. 2.23 where we combine all
The implication is that we cannot interpret the o f the above density distribution graphs into one.
coordinate versions presented in the 26 manuscripts The dependency between various graphs is obvious.
o f the Almagest exclusively as scribe errors. Had this
been the case, this would lead us to the a priori false 8.
statement that the error rate growth for a given area IN RE THE RELIABILITY OF LATITUDINAL
results in better star identification. We must therefore AND LONGITUDINAL MEASUREMENTS
reject the hypothesis about this abundance o f ver CONTAINED IN THE ALMAGEST
sions being attributable to the inaccuracy o f the
scribes. In this case, the only reasonable explanation 8.1. According to Robert Newton,
of the effect discovered can be rendered as follows. the longitudes in the Almagest were
The multitude of different stellar coordinate ver re-calculated by somebody; however, this
sions in the Almagest manuscripts results from in suspicion does not arise insofar as their
dependent star observations performed several times latitudes are concerned
by an observer, or a group o f observers. Due to the
imprecision o f the instruments used for these obser Let us begin with the commentary in re the Alma
vations, the results would often differ from each other. gest measurement precision made by R. Newton, the
The more measurements o f a given stars coordinates astronomer. In general, we are o f the opinion that
were performed, the more versions would get into these observations o f his are applicable to a wider
manuscripts. Therefore, the areas o f the star chart spectrum o f issues. R. Newton actually gives us a very
with high coordinate version density are the ones forthright account o f a rather meandrous scenario
whose stars were observed several times with their around the readings and interpretations o f a great
coordinates measured anew; in other words, these number o f ancient astronomical documents. He is
areas enjoyed more o f the researchers attention than referring to the so-called principle o f error immor
the others. It is natural that the more attention a given talization, which can be formulated as follows. Let us
celestial region got, the more dependable the identi assume that the error of author A became published,
fications o f the stars it contains. As we shall demon and a later author B is referring to it in some man
strate in the subsequent chapters o f our book, the ner deeming the erroneous statement veracious. Thus

the error becomes immortalized in scientific litera numbers were transcribed as letters. This would fre
ture; erasing it from scientific literature becomes an quently cause confusion. For instance, according to the
impossibility. One can hardly be serious about there astronomers R. Newton ([614], page 215), Peters and
being no exceptions for this rule; however, there is a Knobel ([1339]), one could easily confuse the an
great number of examples that do follow this princi cient Greek digits for 1 and 4 due to the fact that the
ple - readers are likely to have quite a few such ex figure of 1 was transcribed as a , and one o f its widely-
amples o f their own ([614], page 165). used old forms was very similar to the letter 8, which
Something similar appears to be happening with stood for 4 - hence the confusion.
the Scaligerian interpretation o f the Almagest - its One has to make an important observation in this
dating in particular. The analysis o f the Scaligerian respect. Our research is based on the canonical ver
version, which dates it to the beginning o f the new sion o f the Almagest star catalogue translated in the
era requires a new study o f its content. This is a com work o f Peters and Knobel ([1339]). As R. Newton
plex scientific problem that requires a great deal of points out, a careful comparison o f various manu
labour. We accomplish a significant part o f this task script often reveals the errors made in the process o f
in our research, and the reader has the opportunity multiple copying and gives the researcher an oppor
to evaluate the complexity o f this task. The main dif tunity to correct them. Peters and Knobel studied the
ficulty is that one has to get to the very roots o f this Syntaxis [Almagest - Auth.] with the utmost at
or the other scientific statement or opinion. It ap tention; it is possible that their version o f this cata
pears that their overwhelming majority was initially logue is the most precise o f all ([614], page 216).
made with the a priori or taciturn presupposition We shall also be using the detailed analysis per
that the Almagest dates to an early a.d . century. Our formed by the astronomer Robert Newton in the large
excavations required the analysis of source material, special chapter IX o f his book ([614]) in order to
which requires a great deal o f work by itself. evaluate the reliability o f the longitudes and latitudes
Let us now get back to the issue o f the complex as given in the Almagest. We shall omit the details per
ity o f latitudinal and longitudinal measurements. In taining to the statistical analysis conducted by R. New
Chapter 1 we already explain that the very nature of ton and merely cite his results.
the ecliptic and equatorial coordinates allows to meas R. Newton wrote that the latitudes in the star cat
ure the latitudes more securely than the longitudes. alogue were most probably measured by a single ob
Also, the use o f an armilla, for instance, can gener server employing a single instrument for the pur
ate errors if the astronomer makes an incorrect eclip pose ([614], page 253). Further also: the latitudes
tic inclination choice. The matter is that the observer educed from the observations were put down in the
has to determine the angle between the ecliptic and the catalogue without alterations (it is however possible
equator and then fix it in order to use the instrument that there were errors in the transcription) ([614],
for the measurement o f stellar coordinates, for in page 249). According to R. Newton, the latitudes o f
stance, having adjusted it in accordance with the pre the Almagest star catalogue are a reliable enough body
viously found ecliptic inclination. In general, the o f material obtained as a result o f actual observations
armilla can be adjusted by any object whose latitude performed by either Ptolemy or one o f his predeces
and longitude are known. Ptolemy often used the sors (Hypparchus, for instance). This concurs per
Moon for this purpose. This makes it possible to cal fectly well with the inform ation cited above that
culate the coordinates o f any other object that might shows latitudinal measurements to be a lot simpler
interest us. However, in this case, as R. Newton is per as a procedure than the longitudinal, therefore, stel
fectly correct to remark, the imprcisions in the de lar latitude is a more reliably measurable coordinate.
termination o f the known objects coordinates auto The picture with the longitudes is drastically dif
matically lead to incorrect calculation of the second ob ferent. R. Newton claims that the longitudes werent
jects coordinates ([614], page 151). deduced from any observations whatsoever ... the
It also has to be borne in mind constantly that in longitudinal values are fabricated ([614], page 249).
case o f the Almagest we are dealing with copies where Further also: the multitude o f longitudes contained
86 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

in the star catalogue is highly unlikely to have been help meeting a certain end. According to N. A. Mo
determined from observations ([614], page 250). We rozov, for instance, this end could be formulated as
have already explained to the reader that the meas giving the catalogue an arbitrary amount o f extra
urements o f ecliptic longitudes prove to be a lot more age - in other words, we have a case o f falsification.
sophisticated and complex procedure than longitu However, we shall refrain from taking any sides a
dinal measurements. Furthermore, it is presumed priori and analyze longitudes and latitudes both to
that the longitudes in the Almagest catalogue were gether and separately.
rendered to 137 a.d. Such a rendition to an a priori Let us conclude with another summary made by
chosen date is quite simple; all it takes is adding some R. Newton: We get an altogether different picture
common constant to the ecliptic longitudes o f all the from the longitudes [as compared to the latitudes -
stars. This constant is proportional to precession and Auth.]. No colourable explanation can possibly be
depends on how much older the compiler o f the cat given to the fraction distribution in longitude, regard
alogue really wanted the longitudes to look. R. New less o f whether or not the observations were in fact
ton is o f the opinion that the original longitudes ob performed by a single person who had used a single
tained by the ancient observer experimentally were instrument for this purpose ([614], pages 146-247).
subsequently re-calculated anew by someone else.
This is his fundamental solution based on the analy 8.2. Examples proving that the dating of
sis o f how frequently degree fractions appear in the the star catalogue by longitudinal precession
catalogue: Longitudes were altered. Observation re often leads to great errors. Mediaeval
sults were made greater by several degrees and 40 catalogues are subject to becoming
minutes ([614], page 249). This operation (an addi erroneously dated to an antediluvian epoch
tion o f a whole number of degrees whose value could
be either positive o f negative, with a couple o f frac The Scaligerian version o f astronomy often uses
tions) could make the catalogue either gain or lose a the following apparently simple method for catalogue
considerable amount o f age at the will of its compiler dating. The ecliptic longitudes o f the old catalogues
or forger. Bear in mind that such an operation would stars are compared to the modern longitudes. The
be either altogether impossible with latitudes, or a resulting difference, which is roughly the same for all
great deal more complicated at the very least. How the stars, is then divided by the precession value,
ever, we cannot determine how many grades exactly which equals roughly 50 seconds per year or one de
were either added to the initial longitudes or sub gree in 70 years. This is how the historians determine
tracted therefrom if we are to base our research upon the residual between the dates o f the modern cata
nothing but the longitude analysis in the existing logue and those contained in the old one. In partic
copies o f the Almagest. R. Newton points out the very ular, this method allows to deduce the ecliptic co
same thing: The actual distribution of grade fractions ordinates from the 1538 edition o f the Almagest as
tells us nothing o f just how many grades were added equalling those which roughly correspond to some
to the initial longitude by Ptolemy ([614], page 251). early a.d . epoch.
Apart from the simple operation o f shifting all However, the method described above makes the
the longitudes by an unknown number o f grades taciturn implication that the compiler o f the old cat
mentioned above, R. Newton discovered traces o f alogue would count ecliptic longitudes from the ver
finer longitudinal recalculations ([614], pages 246- nal equinox point o f his era, or the epoch when the
247). Thus, someone had conducted an extensive star observations were conducted. Had this indeed al
body o f work in the field o f recalculating the ini ways been the case, the resulting residual accumulated
tially observed longitudes. Therefore, the modern by today could really be considered a result o f pre
list o f longitudes that we find in the Almagest does cession. Assuming this to be true, the method de
not represent the actual observational material, but scribed above would indeed give us the approximate
rather the likely result o f its having been processed date of the old catalogues creation. However, it is im
in a certain rather complex way which was meant to portant to emphasise that it wasnt in fact a charac-

Fig. 2.24. Star chart from a XVII century book by Stanislaw Lubienietski. One sees that the Gamma o f Aries was chosen as the initial
longitudinal reference point. This is where the equinoctial crosses the ecliptic. Taken from [543], inset between the pages 26 and 27.

teristic o f all the ancient authors to use the vernal constellation o f Argo Navis, closer to the right end of
equinox point o f their own epoch for the initial ref the map, and once again near the constellation o f
erence point. Ophiuchus near the left end o f the map - see fig. 2.24.
Let us linger on the above for a while. One should The ecliptic is represented by a thick horizontal line
nt get the impression that the astronomers o f as re with degree grades. One can see perfectly well that the
cent an epoch as the XVI-XVII century necessarily ecliptic and the equator cross right where the map
count the longitudes in the exact same manner as the boundary is located - at the y star of the Aries con
modern astronomers. We shall refer the reader to the stellation. There can be no doubt about this (see figs.
well-known C o m e to g ra p h y by the mediaeval author 2.25 and 2.26).
Stanislaw Lubienietski published in 1681: S. de Lubie Thus, all the stellar longitudes indicated by S. Lu
nietski, H is t o r ia u n iv e rs a lis o m n i u m C o m e t a r u m bienietski were smaller than the ones we find in the
([1257]). This book is a priori known to have been Greek longitudes from the 1538 Almagest by roughly
written in the XVII century. It lists many comets ob 7 degrees (see the respective comparative tables as well
served up until the year 1680. S. Lubienietski, its au as the actual charts in [544], Volume 4, pages 233-
thor, belonged to the XVII century school o f as 234, and also [543], inset between pages 26 and 27).
tronomers, preceding our time by a mere 300 years. Let us retort to the strange logic o f the Scaligerite
Lets take a closer look at how Lubienietski counts the historians which they advocate with such persistence
longitudes on his star charts. We discover that he uses and even obstinacy in their dating o f the Almagest by
the meridian crossing the y star from the Aries con the longitudes o f the Greek edition, thereby implying
stellation as the initial celestial meridian, qv in fig. Lubienietski to have counted the coordinates begin
2.24. The sine curve that stands for the equinoctial, ning with the vernal equinox point o f his epoch. In
or the celestial equator in this projection, is directly re that case his book will have to be dated to the V cen
ferred to as Aequator here, which is the legend that tury b . c ., since this is when the vernal equinox point
we see over the masts o f the Argonaut ship from the was really located near the first stars o f the Aries con
88 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

stellation, qv in Lubienietskis case, according to the o f Copernicus). The only difference is that the y of
most apropos comment made by N. A. Morozov [544], Aries occupies the longitude o f zero in the catalogue
Volume 4, page 33. However, Lubienietskis book was o f Copernicus ([1076]). The latter gives its coordi
written in the XVI century! nates as equalling 0 degrees 0 minutes o f longitude,
The ensuing absurd corollary is yet another proof and 7 degrees 20 minutes o f latitude (see [544], Vol
o f how careful one has to be in ones dealings with ume 4, pages 224 and 227). Thus, if we decided to
the dating method described above - which, as we date the catalogue o f Copernicus using the Scali
feel obliged to reiterate, has always been used by the gerian method described above, we would also date
Scaligerian historians in case o f the Greek edition of it to times immemorial, which would be perfectly er
the Almagest. roneous since it is presumed that Copernicus had
All o f the above implies lucidly that the astro lived in the XV-XVI century (1473-1543).
nomers o f the XV-XVII century a . d . hadnt yet come Thus, the precession o f the stellar ecliptic longi
to any unified agreement concerning the initial ref tudes cannot serve for any secure dating o f the cata
erence point for the longitude count. The unification logue whatsoever.
epoch would come after quite a while. Each as The varying initial reference points used for lon
tronomer would select his own point o f reference gitude count in the works o f the XVI-XVII century
guided by considerations o f his very own. authors as indicated above shouldnt surprise us at all.
Lubienietski, for one, used the first stars o f the Aries There were many different astronomical schools at the
constellation for this purpose. As for the Greek edi dawn o f this discipline, which would often compete
tion o f the Almagest, the star coordinates were with each other and adhere to different catalogue
counted from the meridian that crosses the ecliptic compilation rules etc. It is well possible that each
at the point whose longitudinal distance to the y of school remained loyal to a tradition o f its own which
Aries equals 640'. specified the rules for choosing the basis points, ref
Lubienietskis case is by no means unique. The erence points and so on. The considerations for such
star catalogue compiled by Copernicus provides for a choice may have been astronomical, religious, or of
a more impressive example. Copernicus also counts an altogether different nature.
the longitudes beginning with the y o f Aries, just like It was only when astronomy developed into a
Lubienietski (or, rather, the latter follows the tradition grown science when the necessity o f a unified system

Fig. 2.25. A fragment. Right side o f Lubienietskis chart, where Fig. 2.26. A fragment. Left side o f Lubienietskis chart, where
the equinoctial crosses the ecliptic near the Gamma o f Aries the equinoctial crosses the ecliptic near the Gamma o f Aries.
([1257]). Taken from [543], inset between the pages 26 and 27. Taken from [543], inset between the pages 26 and 27.

o f indications and concepts was realized that the as Apart from that, the Almagest is basically a textbook,
tronomical language became more uniform. In par or a guidebook for young astronomers and scientists
ticular, the vernal equinox point was agreed upon as in general that contains descriptions o f varying ob
the initial reference point (an invisible one, as a mat servation methods etc - a mediaeval astronomical
ter o f fact; furthermore, its celestial position changes encyclopaedia o f sorts. Here are a few examples to
with the passage o f time). This point cannot be af confirm this. We shall be using Toomers edition o f
fixed to some star located nearby. It is therefore hardly the Almagest ([1358]).
surprising that certain mediaeval astronomers would In his description of the transit circle in Chapter 1 ,
use an actual star for reference instead o f the equi Ptolemy tells us the following: We made a bronze
nox point - the y o f Aries, for instance. ring o f the fitting size [what size exactly? - Auth.] ...
When we study the Almagest star catalogue in our in order to use it as a transit circle, wherefore it was
book (the same is indeed true for other old star cat graded into 360 parts [degrees]; each of those were
alogues), we make sure our research is in no way de divided into as many parts as the instruments size
pendent on any presumptions that concern the par would allow [How many? - Auth.] ... We have fur
ticular longitudinal reference point used by the cat ther discovered an easier method for conducting such
alogue compiler. There are no such indications in the measurements, having forged a stone or wooden wall
actual star catalogues, after all. Our opponents might [?! - Auth.] to be used instead o f the rings ([1358],
counter that a direct reference to the choice o f the pages 61 and 62).
equinox point for the measurement o f longitudes can What we see here obviously differs from the de
be found elsewhere in the Almagest. scription o f an actual device used for measurements
However, if we are to be guided by such notions, by either Ptolemy alone, or himself and his team.
it shall imply the use o f some extraneous or foreign How else could one explains such ambiguity as fit
information which, as we must emphasize, is not con ting size, as many parts as the instruments size
tained in the star catalogue itself. However, our goal would allow, or stone or wooden wall? Really, was
is to date the catalogue by its own internal charac it stone or wooden?
teristics without citing any external sources. As for the Everything shall fall into place if we are to suppress
issue o f determining the dating o f the remaining texts the inner Scaligerite and realize that what we have in
together with its genesis is a problem o f its own, and front o f us isnt a report made by an observer, but
one that possesses no reliable single solution (see rather an encyclopaedic textbook that explains a po
[544] and [614]). tential student or scientist the construction o f vari
ous instruments; different methods o f conducting re
9. search etc.
THE DUBIOUS NATURE OF THE TRADITIONAL Consider the following passage from the Almagest,
OPINION THAT PTOLEMY'S TEXT IMPLIES for instance: Before [the reign of] Antoninus, when
ACTUAL "OBSERVATIONS" ON HIS PART, we conducted the most observations o f immobile
as well as his "personal participation" in the stars positions ([1358], page 328). Scaligerian as
stellar measurements and observations tronomy reads the implication o f Ptolemy claiming
described in the Almagest personal responsibility for the observations per
formed at the beginning o f the reign o f Antoninus
Ptolemys text can by no means imply the verac Pius into this phrase. The Scaligerian dating o f this
ity o f the consensual opinion, namely, that all the ob emperor is 138-161 a.d . However, Ptolemys phrase
servations and measurements that the Almagest con is rather vague and allows for different interpreta
tains were performed by the author in person. Its ac tions. Firstly, who are the we who conducted the ob
tual text allows for several interpretations. However, servations? Ptolemy himself or his predecessors from
what we are most likely to be seeing here represents the same scientific school? Furthermore, what exactly
the research result o f a great many astronomers and do the most observations refer to? The use o f we
not a single authors account of his own observations. etc has to be considered a distinctive of the Almagests
90 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

authors literary style rather than an indication o f his ner: a late mediaeval observer rendered the catalogue
actual participation in the research; it is also possible to the values corresponding with the reign o f Antoni
that the hoaxer editors o f the XVI-XVTI century were nus. By the way, the Almagest doesnt give us any
intending to create an impression that what the work datings for the reign o f Antoninus. As we already
in question had been written to relate the research of know, the simplest action which can be undertaken
a single person. in order to render a catalogue to any a priori known
For example, let us take account o f the words cho ancient epochs ecliptic coordinates is the subtraction
sen by Ptolemy as the introduction to the Almagest o f a suitable constant value from the original longi
star catalogue. It would be natural to expect the au- tudes. Furthermore, this explanation o f ours is ex
thor/observer who had conducted the research in plicitly confirmed by the text o f the Almagest! Ptol
question himself to provide detailed descriptions o f emy continues his thought right there: The latitu
how his research was conducted, which stars were dinal values always remain immutable; as for the
chosen for reference etc. Nothing o f the kind. Ptole longitudinal values [contained in the Almagest cat
mys text is very vague: alogue - Auth.], they allow for easy longitudinal cal
Again, the very same instrument [the astrolabon culations for other moments o f time as well, for
- Auth.] permits to observe as many stars as humanly which the distance between the current epoch and
possible, including those of the sixth magnitude. We the necessary moment in time needs to be recalcu
would always direct the first ring at the nearest bright lated assuming the alteration speed equal to 1 degree
star whose position in relation to the moon would al every 100 years. The resulting value would then have
ready be calculated by then ([1358], page 399). to be subtracted from that o f the current epoch in
This is followed by the description o f the method order to get a date in the past or added thereto for a
used for stellar coordinate calculations when the lon future date ([1358], page 340).
gitude is measured by relatively bright stars, and the Thus, Ptolemy gives a perfectly clear explanation
latitude in relevance to the astrolabons ecliptic ring. o f how one is to shift the star catalogue in time sub
This description is once again given in rather general tracting the constant, which would make it more
terms, followed by the remarkable phrase: ancient, or adding it for the opposite effect. Once
In order to represent the stars on a solid cosmos- again, this is very similar to a textbook that explains
phere in accordance with the method described above, the technique o f dating and re-dating star catalogues
we have arranged the stars into a table with four to students. This book may have also been a useful
columns ([1358], page 340). Further on we find ex source of all the necessary guidelines in the XVI-XVTI
planations o f the indications used in the table. The century a . d . , especially considering as how the con
table in question is the famous star catalogue. There struction of a cosmosphere as related in the Almagest
fore it turns out that Ptolemys catalogue was created does not require absolute longitudinal values -
with the main purpose o f using it for the creation of namely, they are counted from an arbitrarily chosen
a cosmosphere. immobile star. Ptolemy suggests to use Sirius for this
Once again, this resembles a textbook - in order purpose ([1358], page 405).
to make a globe, one has to do this and that. A pro- Apparently, the absolute values o f ecliptic stellar
pos, Ptolemy makes another reference to Emperor latitudes simply have never been used in Scaligerian
Antoninus in his description o f the table, or cata astronomy at all. Therefore, the longitudinal refer
logue: In the second column one finds the longitu ence point could be chosen more or less arbitrarily.
dinal value deduced from the research [conducted by Copernicus, for instance, having copied the Almagest
an anonymous scientist - Auth.] for the beginning of catalogue into Volume 6 o f his own Revolutionibus
Antoninus reign ([1358], page 340). O rbium C aelestiu m , with some circumstantiation,
Once again, one neednt interpret these words of counts latitudes off the y star o f the Aries constella
Ptolemys as evidence o f him having personally con tion, which was located at the distance o f 27 from
ducted observations in the epoch o f Antoninus. This the point o f vernal equinox in the epoch o f Coper
phrase can also be interpreted in the following man nicus.

One has to point out that the work o f Copernicus, stice or an equinox point, qv above (VII:4, [1358],
as history o f astronomy is telling us, wasnt appar page 340).
ently appreciated until a century after his death, in Stellar longitudes in the Almagest are indeed in
Keplers epoch, or the XVII century ([614], page 328). dicated separately for every arc sign o f the uniform
See Chapter 10 for more details. One can therefore zodiac and counted from the beginning o f the re
ask the legitimate question o f the exact date when the spective arc sign. In other words, the stellar longi
book attributed to Copernicus nowadays was writ tudes that we encounter in the Almagest should not
ten or edited. Could it have been the early XVII cen be considered absolute and are counted off a single
tury and not the XVI - Keplers epoch, in other chosen point on the ecliptic. Instead o f this, the rel
words? ative longitudes contained by every respective arc sign
o f the uniform Zodiac are given, totalling to 12 . It is
10. also pointed out that one o f the quadrants is oriented
FOR LONGITUDINAL REFERENCE? Therefore, the calculation o f some absolute lon
gitudinal value requires the addition o f a certain in
As we already know, the choice o f the initial lon teger number o f degrees divisible by 30, or the size of
gitude count reference point influences the longitu a certain arc sign o f the even Zodiac. The absolute
dinal precession dating o f the catalogue to a sub ecliptic longitudes o f the catalogue can only be de
stantial extent. Let us conduct a more in-depth study duced after this procedure, which is hardly all that
o f the question which point o f the ecliptic was used complex in principle.
by Ptolemy for longitudinal calculations in his cata Let us illustrate by the following example. The
logue. It is traditionally assumed that he had used North Stars longitude in the Almagest is given as
the vernal equinox point for this purpose, likewise Gem 010. In order to calculate the absolute longi
many late mediaeval astronomers. tude value, we have to add an integer number o f de
It turns out that the initial reference point issue as grees to 0 10 ' that equals 60, as contemporary tra
rendered by Ptolemy is far from simple, and cannot dition suggests. This is the number o f degrees be
be resolved without controversy if we are to use noth lieved to correspond to the beginning of the Gem arc
ing but the text o f the Almagest for that end. Let us sign o f the even Zodiac. We shall thus get the value
turn to the Almagest and provide the relevant quo o f 60 10'. If we are to consider it to be the ecliptic lon
tations. gitude o f the North Star as compared to the vernal
Ptolemy writes that we shall be using the names equinox point, it shall correspond to the position the
o f the Zodiac signs in order to refer to the corre latter had occupied in the beginning o f the new era.
spondent twelve parts o f the tilted circle which shall One observes a perfectly similar situation with the
begin in the equinox and solstice points. The first remaining longitudes o f the thousands o f stars con
twelfth part that begins at the vernal equinox point tained in the Almagest catalogue. The simplicity of the
and whose direction is counter to that o f the Universe abovementioned calculations notwithstanding, one
shall be known as Aries, the next as Taurus . . . (11:7 has to point out that this is our first opportunity to
- [704], page 45). The signs in question are merely misinterpret the source data offered by the Almagest,
the arcs of the even Zodiac - not stellar longitudes. namely, the fact that the integer degree values corre
Furthermore, when Ptolemy tells us o f the longitudes, sponding to zodiacal signs depend on the choice o f
he describes the second (longitudinal) column o f his the first arc sign o f the even Zodiac, whose begin
star catalogue as follows: In the second column we ning coincides with the initial reference point - ver
find their [referring to the stars - Auth.] longitudi nal equinox, or, possibly, some other point on the
nal positions educed from observations conducted ecliptic. The alteration o f the first Zodiac sign shall
in the beginning of Antoninus reign. These positions apparently alter the absolute degree values added.
are located inside the Zodiac signs; the beginning of The vagueness o f Ptolemys phrase leaves plenty of
each Zodiacal quadrant is determined by either a sol space for interpretation.
92 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

As we shall find out, Ptolemys description o f the who advocated using Sirius for reference, could well
cosmosphere does not use the vernal equinox point have chosen another very bright star - Arcturus. This
for initial reference. He writes that as it makes no is a star o f great luminosity; most importantly, its
sense to mark the solstice and equinox points on the longitude is expressed as an integer in the catalogue
globes Zodiac (since stars maintain no constant dis - namely, 27 o f Virgo. Why would one perform a
tance to these points), we should select a number of thousand operations with fractions when it would be
fixed immutable reference points among the immo a lot simpler and less time-consuming to perform the
bile stars. The brightest o f those is the star in the very same operations with degrees expressed as an
mouth o f Canis Major [Sirius, that is! - Auth.] ... integer?
then for each or the remaining immobile stars in the One can make the natural presumption that a cer
catalogue [apart from Sirius - Auth.] we must mark tain constant value was either added to, or subtracted
its location [longitude - Auth.] rotating the gradu from, the initial longitudes o f the Almagest, which
ated ring around the ecliptic pole - the point that we made the longitude o f Sirius a fractional value in
must mark on this rings ecliptic is to be at the exact stead o f an integer. Therefore, this value had to com
same distance from the reference point that we dis prise a certain amount o f degrees and 40 minutes,
covered (Sirius) as lays between the star in question since the longitude o f Sirius in the modern version
and Sirius in the catalogue ([1358], page 405). o f the Almagest catalogue equals 17 40.
Thus, Ptolemy gives us a direct reference to Sirius This is where we unexpectedly run into a good
as to a convenient absolute beginning for the eclip concurrence with the result o f R. Newton ([614]).
tic longitude count. This is completely at odds with He proves that the longitudes contained in the cata
the consensual version which tells us that Ptolemy logue were recalculated by someone, with an indefi
would definitely use the vernal equinox point for ref nite amount o f degrees and 40 minutes added to the
erence. original longitudinal values, and bases his conclusion
Furthermore, since the Almagest is an astronom on altogether different considerations - those o f a
ical encyclopaedia of sorts, it may have been compiled statistical nature. We deem such a good concurrence
from the works o f various astronomers from differ between two varying observations to be anything but
ent schools in its present form. Therefore, different random.
measurements principles may have been used for dif One has to make the following general observa
ferent parts o f the Almagest - in particular, it is pos tion, which bears no formal relation to astronomy, but
sible that the longitudinal reference point in the might yet prove useful for our understanding o f the
Almagest catalogue varies as taken for its different role and the place of the Almagest. Modern literature
parts. on the history o f astronomy gives one the impression
All o f this indicates that the attempts to date Ptol that the Almagest chapters dealing with stars are a
emys catalogue by longitudinal precession may lead commentary o f sorts, or an annex to the central doc
to gravest errors, which is exactly what we see in some ument, which is the star catalogue. However, we are
modern works on the history o f astronomy, qv below. o f a different opinion. The primary content o f these
Other contentious issues arise as well. The quo chapters is Ptolemys guidelines for the construction
tation mentioned above demonstrates that the cre o f the cosmosphere whereupon one was to point out
ation o f a cosmosphere requires circa 1000 astro the locations o f the stars. The actual construction
nomical operations - namely, the subtraction o f the process, the paint one needs to use for the purpose
longitude of Sirius from the longitudes o f a thousand etc are described with great detail; the catalogue it
other catalogue stars. However, the longitude o f self is but a reference table for the construction of
Sirius is expressed as a fraction in the Almagest cat the cosmosphere.
alogue, namely, 1740' o f Gemini. It is perfectly clear It is quite possible that such cosmospheres were
that the operation o f subtracting this number from used for astrological or mystical purposes in the
other longitudes a thousand times shall consume a Middle Ages. The most curious fact is that the history
great deal o f labour. On the other hand, Ptolemy, of astronomy has many references to the construction

o f such cosmospheres - however, this celestial globe with greater precision. Not only the longitudes were
construction epoch isnt even close to the beginning made more realistic, but the latitudes as well - those
o f the new era, it pertains to the Middle Ages. In par corrections may have varied from star to star. As a re
ticular, the first news o f such globes that we have date sult, the astronomers o f every new generation would
from the epoch o f Tycho Brahe, who constructed a compile a maximally accurate new catalogue for
cosmosphere himself ([395], page 127); this was con themselves (inasmuch as their instruments would
sidered an important task. We are told that the large allow, o f course). This very method was used for sci
brass-plated cosmosphere, 149 centimetres in diam entific applications, such as navigation, as opposed to
eter, deserves to be mentioned separately. Its surface obsolete near-forgotten catalogues which contained
bore the representations o f the Zodiacal belt, the many errors due to the imprecision o f the primitive
equinoctial, and the positions o f 1000 stars whose early instruments.
coordinates had been determined over the years o f If anyone in the XVI-XVII sought to fabricate and
Tychos observations. Tycho proudly confessed: I be introduce a falsified ancient history, the approach
lieve that no other cosmosphere of this size, built with may have been radically different. Some recently-
such accuracy and precision, has ever been made any compiled star catalogue would be taken, and his lon
where in the world. He also claimed that multitudes gitudes shifted into the past, or the necessary his
would come to Denmark specifically in order to ad torical epoch - the early a . d . period, for instance. The
mire the cosmosphere. Alas, this true wonder o f sci operation was simple and did not consume much of
ence and art perished during a blaze in the second half the hoaxers time. After that they would loudly claim
o f the XVIII century ([395], page 127). having discovered an extremely ancient star cata
Thus, the respective Almagest chapters fit into the logue. Let us reiterate that the simplest and fastest fal
epoch o f the XVI-XVII century perfectly well. sification method would employ a shift o f all stellar
Furthermore, experts in history o f astronomy sug longitudes by a single constant value. Apparently, this
gest that even if the longitudes o f the Almagest were is how the personal observations o f Ptolemy from
recalculated, it was for a more recent epoch and never the II century a . d . came into existence, as well as
backwards. We are being convinced that the recal many other observations conducted by early me
culation o f old stellar longitudes for the current diaeval astronomers. The hoaxers couldnt just open
epoch was a common enough practice amongst me a modern catalogue, since they would be immedi
diaeval astronomers. References are also made to the ately caught, and preferred to use some catalogue dat
early mediaeval catalogues predating Brahe. Me ing to 100-200 years backwards, well-forgotten and
diaeval astronomers are supposed to have been too out o f print already.
lazy to conduct new research. They would rather
grab an ancient catalogue dating from times im
memorial, alter all o f its values by the factor o f a sin
gle constant and come up with modern star coor
dinates as a result, subsequently using this ancient
but so conveniently updatable catalogue in their
own research.
One has to admit that this hypothesis looks rather
strange. It is unlikely that each new generation o f as
tronomers would contend itself with a mere fabri
cation o f the kind o f catalogue they needed via a
shift o f longitudes contained in some old and rather
obsolete, catalogue. Every new epoch creates new and
more advanced astronomical instruments. Therefore,
it is most likely that the astronomers o f every subse Fig. 2.27. The sinusoid o f Peters in the latitudes o f the Almagest
quent epoch would measure stellar coordinates again, star catalogue.
94 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

as a result will represent latitudinal discrepancy as a

longitudinal function. The sine curve of Peters can be
seen in fig. 2.27. It is very much like a sine curve with
the amplitude o f circa 20. One could choose a sinu
soidal curve considered best in its class for the ap
proximation o f the curve in fig. 2.27. The resulting
sine curve was named after Peters.
The appearance o f Peters sinusoid is very hard to
explain within the framework o f the modern ideas of
the Almagest. At any rate, we have found no reason
able explanation o f this distinctly periodical phe
nomenon in any kind o f literature.
Fig. 2.28. The somewhat odd graph o f average longitudinal
One has to point out that [1339] contains no de
discrepancy as a function o f ecliptic longitude in the Almagest tails related to the calculation o f this curve by Peters.
catalogue. In particular, we learn nothing o f the actual Zodiacal
stars he used for calculations. Therefore, in order to
confirm the actual existence o f the effect and study
11. it we had to recalculate the curve in question for all
PETERS' SINUSOID IN ALMAGEST the Zodiacal stars with the aid o f a computer. Our re
LATITUDES sults, as well as their implications and related com
mentary can be found in the chapters to follow. Let
Let us now consider the latitudes o f the Almagest us however jump ahead for a moment and divulge to
star catalogue. This is where we immediately discover the reader that we find a perfect explanation for this
a most peculiar effect that defies explanation in the strange sine.
paradigm o f earlier Almagest studies. We shall be re NB. Apart from the latitudes, Peters also studied
ferring to this effect as to the Peters sine curve. The the longitudes o f the Almagest catalogue ([1339]).
matter at hand is as follows: Peters analyses the aver He counted the average latitudinal discrepancy for
age error distribution in the Almagest as a longitu 10 -degree sectors and came up with the graph that
dinal function. For this purpose he calculates the po we see in fig. 2.28. The curve represents the behav
sitions of the modern skys Zodiacal stars for 100 a . d . , iour o f the average longitudinal discrepancy as a func
or the alleged epoch o f the Almagest creation. Then tion o f ecliptic longitude. It is remarkable that the
Peters calculates the latitudinal discrepancy o f A, = graph is drastically different from the one with the Al
- b{. Thus, B{ is the latitudinal value o f star i from the magest latitudes. The longitudinal graph is by no
Almagest, and b{ - the meaning o f its latitude for 100 means sinusoidal; its amplitude is smaller; besides,
a . d . as per Peters. Therefore, the A, value demon and it has two rather distinct local maxima. It is pos
strates Ptolemys error in the determination o f star sible that this oddly irregular nature o f the longitu
Vs latitude, made under the assumption that the Al dinal curve is a result o f the mysterious ecliptic lon
magest was created around 100 a . d . Peters proceeds gitude recalculation as discovered by R. Newton in
with the division o f the ecliptic into 10 degree inter [614] (see section 8). As it has been pointed out, the
vals and then calculates the average latitudinal dis longitudes o f the Almagest catalogue are by no means
crepancy value for all the Almagest stars that wind up a reliable source o f information; therefore, we have
in this interval, which naturally varies from one in no reasons to study the resulting graph more atten
terval to another. tively. Such analysis would only make sense if the lon
A special graph has been built as a result, one that gitudinal recalculation mechanisms, which must have
demonstrates how the average latitudinal discrepancy been used by later astronomers (possibly o f the XVI-
manifests along the ecliptic. Points o f the ecliptic can XVII century), could be reconstructed, which we be
be characterized by ecliptic longitude; the graph built lieve to be a very difficult task at this point.

Unsuccessful attempts of dating

the Almagest. Reasons for failure.
Our new approach and a brief
account of our results

i. We shall refrain from going into detail about the

THE ATTEM PT TO DATE THE ALMAGEST quality criteria o f such correspondence and merely
BY A COMPARISON TO THE CALCULATED define the meaning o f comparing the Almagest to
CATALOGUES REFLECTING THE MOTION catalogue K{t) with a given t value. What this im
OF THE FASTEST STARS plies is selecting the same coordinates from cata
logue K(t) and the Almagest. The comparison in
1.1. The comparison of the Almagest catalogue question makes year t serve for the alleged dating o f
to the calculated catalogues the observations that the Almagest catalogue is based
upon. Therefore, in order to compare the coordi
In Chapter 1 we refer to the algorithm o f recal nates o f the stars in the Almagest with their coordi
culating the modern positions o f celestial objects nates in the calculated catalogue, one has to set the
backwards into the past. Thus, what we have at our Almagest ecliptic into the same plane as the ecliptic
disposal presently is the Almagest catalogue com o f the calculated catalogue K(t).
piled in ecliptic coordinates in some unknown epoch However, such a superimposition shall allow for
and the set {K(t)} o f the calculated star catalogues. nothing but latitudinal comparison, whereas we also
They reflect the real situation on the celestial sphere need to compare stellar longitudes. In other words,
that we computed for a given time moment t. Let us we shall have to impose the Almagest star atlas over
try and determine the desired value o f the date or the real one for epoch f, supposing t to be the real time
the epoch when the Almagest catalogue was com when the Almagest author performed his observa
piled. We shall begin with the following idea which tions. This requires marking the vernal equinox point
appears quite simple and try to compare the posi for epoch t on the Almagest ecliptic. This point is to
tions o f individual stars in the Almagest to their po be selected in such a way that the average longitude
sitions in the calculated catalogues K(t); after that we error for the Zodiacal stars o f the Almagest would
shall try to select such a value t* for the evaluation equal zero. Bear in mind that we are using the table
o f the date tA that it would make the Almagest data o f traditional identifications o f the Almagest stars
correspond to those contained in the catalogue K (t*) with the modern star chart as given in [1339] for our
in the best way possible. comparison with the longitude o f the relevant stars
96 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

from the catalogue K(t). It isnt that formidable a task

to select such an equinox point. As it is known (qv in
[1040] and [1339]) that t = 18.4, or corresponds to
the Aries arc sign on the Almagest ecliptic for 60 a . d . ,
shifting with the speed of roughly 49.8" for each year
t - the precession speed, that is.
We cannot quite evade errors in our choice o f the
vernal equinox point on the Almagest ecliptic with the
method indicated above, which is optimal statisti
cally. Its complete evasion would be achieved if we
merely compared stellar latitudes without taking the
longitudes into account whatsoever. This is what we
shall do below, in Chapters 3-5. We shall analyze the
latitudes and the longitudes separately. The consid
erations given in the current section are o f a prelim
inary character.
Fig. 3.1. The m otion o f the real Arcturus as compared to its
position specified in the Almagest. This graph doesn't
1.2. The attempt of dating the Almagest
account for the systematic error made by Ptolemy or
catalogue by proper movements compensate it.
of individual stars

Let us choose nine o f the fastest stars for com

parison, indicated in the Almagest according to
[1339]. These are the stars, whose proper movement
speed exceeds 1 " per year. Their list is as follows:

a Cent (969) - 4.08" per year,

o 2 Eri (779) - 3 .6 8 " per year,
a Boo (110) = Arcturus - 2.28" per year,
T Cet (732) - 1.92" per year,
a CMa (818) = Sirius - 1.33" per year, Fig. 3.2. The m otion o f the real Sirius as compared to its
y Ser (265) - 1.32" per year position specified in the Almagest. This graph doesn't
account for the systematic error made by Ptolemy or
i Per (196) - 1.27" per year,
compensate it.
a CMi (848) = Procyon - 1.25" per year,
r\ Cas (180) - 1.22" per year.

All these stars are contained in the Almagest, ac

cording to traditional identifications ([1339]). The
numbers given to them by Bailey in the serial nu
meration o f the Almagest are in parentheses. Let us
represent each o f these Almagest stars as a circle with
out any shading, see figs. 3.1-3.8. We decided to omit
a Centauri, since the coordinates o f this star which
lays far to the south are given in the Almagest with
Fig. 3.3. The m otion o f the real Procyon as compared to its
the gigantic 8-degree error. In fig. 3.4, apart from the position specified in the Almagest. This graph doesnt
Almagest star 779, one can also see the neighbouring account for the systematic error made by Ptolemy or
stars 778 and 780 and the trajectories o f real stars compensate it.

Fig. 3.4. The motion o f the real stars o2 Eri and Eri as compared to the Almagest data. This graph doesnt account for the
systematic error made by Ptolemy or compensate it. The numbers o f the stars are given in accordance to a modern catalogue
98 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

Fig. 3.5. The motion o f the real star x\Cas as compared to its Fig. 3.8. The motion o f the real star y Ser as compared to its
position specified in the Almagest. This graph doesnt account position specified in the Almagest. This graph doesnt account
for the systematic error made by Ptolemy or compensate it. for the systematic error made by Ptolemy or compensate it. Star
numbers are given according to a modern catalogue ([ 1197]).

numbered 1332, 1362 and 1363 from the catalogue

([1197]). Thus, we have eight stars left.
Let us now regard the small neighbouring areas of
each o f these eight stars in Ptolemys star atlas. We
shall be using these star coordinates as given in the
Almagest. Each o f these areas contains one o f the
eight fast stars listed above. Furthermore, we share the
opinion [1339] that Ptolemy did in fact observe all
o f these eight stars, and that they are really present in
Fig. 3.6. The motion o f the real star i Per as compared to its
his catalogue.
position specified in the Almagest. This graph doesnt account Now let us superimpose the star atlas compiled
for the systematic error made by Ptolemy or compensate it. from the calculated catalogue K (t) which reflects the
state o f the real celestial sphere for epoch t, over
Ptolemys star atlas compiled from the Almagest; we
shall be using the method described above, and per
form this procedure for every t moment. We shall
now draw our eight fast stars among the stars o f the
The method o f imposing the calculated atlas K(t)
over Ptolemys atlas depends on the choice of epoch t.
Moreover, each o f the eight fast stars changes its po
sition in relation to the other stars from the calculated
catalogue K(t) with an alteration of t. Thus, the way
Fig. 3.7. The motion o f the real star T Cet as compared to its
these stars shall be represented on Ptolemys atlas
position specified in the Almagest. This graph doesnt account shall also depend on the time t. We will come up with
for the systematic error made by Ptolemy or compensate it. eight new trajectories on Ptolemys atlas correspon-

ding to the shift o f our eight fast stars after the alter Table 3.1. Approximate datings o f the Almagest
ation o f t. These trajectories can be seen in figs. 3.1- catalogue by the proper movements o f eight fastest
3.8. Let us emphasize that we are not yet taking into stars observable with the naked eye.
account the systematic error in stellar locations that
we discovered the Almagests compiler to have made. Dating closest Minimal
We shall relate the story of this error in detail below. to the star distance to the
What are the t moments that we are considering observation time star o f the
Star name in the Almagest Almagest
now when the real fast stars are the closest to how they
were represented on Ptolemys atlas? Arcturus = a Boo 900 A .D . 40'
Generally speaking, these moments vary from star Sirius = a CMa 400 A .D . 10'
to star. For the eight stars listed above we shall mark Procyon = a CMi 1000 A .D . 20'
them as tly t2, . . t8. If it turns out that all the values
o2 Eri 50 A .D . 5'
o f tj ( 1 < i < 8 ), or a considerable part o f them at the
T| Cas 1100 B .C . 40'
very least, turn out to be close to each other as well
as some averaged value o f f*, it shall be strong argu i Per 9700 a . d . 70'
mentation in favour o f the theory that the true time xCet 220 a . d . 15'
o f the Almagests authors observations is close to t*. ySer 700 A .D . 80'
However, this doesnt appear to happen. Indeed,
the values tt are chaotically scattered across the time
interval - 7 0 < t < 30), or 1000 b . c . - 9000 a . d .! The epoch o f the ancient Hipparchus, that is. Let us
range is just too great. Let us compile the results into draw the precision circle around the point that rep
table 3 .1 to make them more illustrative. The fact that resents a fast star in the Almagest whose radius will
the individual datings tt are spread across this great equal the average error rate for the constellation that
a range is hardly surprising. The matter is that each contains the star in question, qv in figs. 3.4-3. 8. The
of the eight stars under comparison is represented in projection o f this circle over the trajectory o f the cal
the Almagest with a certain error which is rather se culated star that reflects the movement o f a real fast
rious. star across the celestial sphere shall give us an
The idea o f the possible rate o f this error for an idea o f the possible error rate pertinent to the indi
individual star can be obtained from the average arc vidual dating t{ by the star in question as compared
declination in the constellation that the star in ques to the real date o f the catalogues compilation. Let us
tion is part of. Under the arc declination we under also point out that the individual star measurement
stand the gap between the stars position in the errors that we know nothing about can differ from
Almagest and its true calculated position. Strictly the average error rate drastically. The radius o f the
speaking, the indicated average error depends on the precision circle for Arcturus, Procyon, Sirius and
alleged dating o f the Almagest - due to the proper other named stars was chosen as equalling 10 ', or
movements o f stars, for instance. However, the stars the Almagest catalogue scale grading value. See
on the celestial sphere are almost immobile for the figs. 3.1-3.3.
most part. It appears that the rate of this average error
is only marginally dependent on the epoch that the 1.3. W hy the dating of the Almagest
stellar coordinates are calculated for. The precision by individual star movements gives us
level that is o f interest to us allows to disregard this no reliable result
In order to calculate the average error rate, we The question that inevitably arises in this regard
have used the comparison table that contains the star is whether the results achieved with the use o f one or
positions in the Almagest together with their real several o f the eight stars listed above can be trusted
positions for 130 b . c . that we encounter in the work more? In that case, this is the star which we must use
o f Peters and Knobel ([1339]) - calculated for the for the purpose of evaluating and dating Ptolemys re
10 0 h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

search, rejecting the datings based on all the other calculation must have been very high indeed (see
stars as not reliable enough. It is natural to use the [968], for instance). At the same time, there is noth
stars whose coordinates are the most correct in the ing very noticeable about the star o 2 Eri. It cannot be
Almagest. But how does one choose them? distinguished from the stars surrounding it, them
In some works it was suggested to evaluate the being just as dim.
precision o f Ptolemys measurements for each o f the Furthermore, the star traditionally associated with
stars in question basing our judgement on the cal o 2 Eri is merely described as an average star in the
culated arc discrepancy for a given star - using the Almagest. Therefore, we would be justified to ask an
last column o f the cited table, in other words. The im other perplexed question after taking a look at fig. 3.4.
plication would be that the coordinates o f the star o 2 Why would the Almagest star #779 possibly be iden
Eri were measured by Ptolemy with the precision tified as o 2 Eri? It is perfectly clear that this is a con
rate o f 5', for instance, and those o f Arcturus - with clusion one can only arrive at in case when the co
the precision rate o f 40'. This is exactly what the au ordinates o f the real star o 2 Eri and the star #779
thors o f [273] Y. N. Yefremov and Y. D. Pavlovskaya from the Almagest correlate with each other opti
had done. They had tried to date the Almagest by mally - better than those o f o 2 Eri and the star #778,
proper movements and worked with the same list of for instance. However, due to the significant proper
9 stars in particular. This approach would yield a motion velocity o f o 2 Eri this clearly implies that its
dating which would be close to the Scaligerian - 50 identification as any star o f the Almagest is greatly
b . c . , qv in table 3.1. The evaluation o f the possibil dependent on the time we date the Almagest to.
ity that this dating is erroneous is a separate issue For instance, if we knew that the Almagest was
which we shall consider below. To jump ahead very written in 1000 b . c . , we could identify o 2 Eri with the
briefly, we shall merely state that the possible error Almagest star #778, and then successfully date the
rate o f Yefremov and Pavlovskayas method was es Almagest to the very same year 1000 b . c . judging by
timated perfectly unrealistically in [273]. the minimal possible distance between o 2 Eri and
This approach instantly leads us to the following the star #778, which would serve as sound proof
set of questions. The first one concerns the rather ab o f our a priori dating.
surd situation in which all three stars of the first mag A propos, this identification makes the concur
nitude out o f nine, namely, Arcturus, Sirius and Pro- rence between the coordinates o f o 2 Eri and the Al
cyon (and ones that have names o f their own in the magest even better than the traditional version, as
catalogue at that) were measured by Ptolemy very one can plainly see in fig. 3.4. If we assume that the
roughly, with error rates approximating an entire de Almagest was written in 1500 a . d . , or the XVI cen
gree. Yet the dim and poorly-visible star o 2 Eri was for tury, for instance, we might identify the star o 2 Eri
some reason measured with the utmost precision, the as the Almagest star #780 and date it to the late
discrepancy equalling a mere 5'! Let us explain that Middle Ages, or even a future epoch, qv in fig. 3.4.
the magnitude of this star according to modern meas It is clear that ruminations o f this sort lead to a
urements equals a mere 4.5, which means it is very vicious circle. The dating o f the observations based
dim. on proper star motion requires a reliable identifica
All o f this is most bizarre indeed. Such bright and tion o f said star as one contained in the Almagest, all
famous stars as Arcturus, Procyon, Regulus and Spica o f this independently from its presumed dating.
must have served Ptolemy in his research as control However, even if we are to disregard o 2 Eri, we still
points, or, at the very least, their coordinates were cannot use the remaining eight fast stars for a secure
measured with the utmost care and precision. Their dating, even now. The dating dispersion is too great
exceptional importance to ancient astronomy is re for all the different stars. Even the datings made by
flected in the very fact that they have own names in the stars o f the first magnitude out o f the eight stars
the Almagest. There are even special sections o f the under study (Arcturus, Procyon and Sirius) are scat
Almagest concerned with the measurements o f some tered over the 600-year interval between 400 a . d .
o f them. Therefore the precision o f their coordinate and 1000 a . d . , qv in table 3.1.

Furthermore, one neednt forget that the datings 2.

deduced in such a manner (900 a . d . for Arcturus) AN ATTEM PT OF DATING THE ALMAGEST
only represent the moments when the real positions CATALOGUE BY THE AGGREGATE OF FAST
o f the stars are the closest to those given in the AND NAMED STARS AS COMPARED TO
One also needs to specify the time intervals sur
rounding these datings for which the deviation val 2.1. The criteria one is to adhere to in one's
ues would fall into a range conforming to precision choice of the stars for the purpose of dating
The gravity o f the situation is all the greater that In section 1 we demonstrate that the comparison
if we are to use average values for the evaluation of o f the Almagest with the calculated catalogues K(t)
just how precisely this star or the other was measured by the eight of the fastest stars doesnt allow us to in
in the Almagest, we shall be making a certain error a dicate a t* value that makes the Almagest correlate
priori, knowing nothing of the individual errors made with the catalogue K (t*) in the best possible manner.
in the measurement o f the stars in question by For each star the value o f f* = t* is unique and dif
Ptolemy. fers from the values o f other stars significantly. The
scatter range for different stars equals several millen
Let us formulate the corollaries: nia. Therefore, the approach as described above is too
1. Before one can use the coordinates of a separate rough, and gives us no substantial result.
star as given in the Almagest for the purposes o f dat However, it might turn out that once we make the
ing, one needs to make sure that identifying the star sample include a lot more stars than eight, we shall
in question as a star observed upon the modern ce come up with such a set of individual datings {t*}
lestial sphere does not depend on a presumed dating whose larger part will fall into a rather short time in
o f the Almagest, which would lead us to a vicious cir terval. At the end o f the day, even an interval o f circa
cle once again. 500 years would suffice; in this case we would be given
2. Even for the fastest o f stars, the shifts made due some sort o f opportunity to obtain the information
to proper motions are small enough inasmuch as the concerning the real date o f Ptolemys research (tA).
span o f the historical period is concerned (see figs. Apart from that, making the sample more inclusive
3.1-3.8). Therefore, a dating would require a selection might enable us the use o f mathematical statistics
o f stars whose positions in the Almagest would be methods for the estimation o f the tA value.
measured with enough precision. A star than only What other stars should one include in the sam
shifts by 2" in a year will shift by a mere 3.3' over the ple? It is clear that only the fast and relatively well-
period o f a century. measured stars fit the purposes o f dating. These two
Therefore, if we want to use an individual fast star criteria - proper motion velocity and the record pre
for the dating o f the Almagest with the precision cision in the Almagest, complement each other in gen
range o f circa 300 years, we must be certain that the eral, since the faster the star, the greater the error we
precision o f this stars position as given in the Al can make for its coordinate in the Almagest without
magest does not exceed the discrepancy rate of 10 . affecting the dating by the star in question.
According to the estimations o f researchers, the real These considerations lead us to the choice o f the
precision o f the Almagest is a lot lower in general following stars for the comparison o f the Almagest
([1339]). with the calculated catalogues K(t).
The stars whose coordinate precision discrepancy 1 ) The stars which move fast enough. Let us choose
rate exceeds 20' are all but void o f utility for us. The 0.5" as the annual speed threshold pertaining to a
dating interval is 1200 years minimum if we are to use single equatorial coordinate at least a 190o and 8 190o
them for dating purposes. for the epoch o f 1900 a . d . , qv in table 1.1).
This issue is considered in more detail below (see 2) Famous or named stars, or the stars which have
Chapters 5 and 6 ). old names of their own (see table P I .2 in Annex 1 ).
102 h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

Naturally, named stars may have received their

names already after the creation o f the Almagest,
which appears to be true for many stars. However,
firstly, the stars names are unlikely to have been for
gotten with age, although they may indeed have al
tered. In other words, named stars o f Ptolemys epoch
remain such until the present day. Secondly, the fact
that a given star received a name o f its own tells us
that it had been charged with a particular significance
in old astronomy. It would therefore be self-implied
that Ptolemy had paid more attention to named stars
than to others, which would be manifest in their more
precise measurement especially.
Fig. 3.9. The m otion o f a real star near the position specified
Let us choose the interval o f 0 < t < 30 as the a pri for it in the Almagest.
ori time interval for our research (1100 b . c . to 1900
a . d . , that is). Bear in mind that the letter t refers to

the time counted backwards from 1900 a . d . in cen correspond to [1339]. The distances on the celestial
turies. sphere would be measured on the geodetic arc that
connects the stars. Bear in mind that geodesic lines
2.2. The "proximity interval" system as applied on a sphere, or the line o f the shortest local lengths,
to certain fast or named stars are the arcs o f large circumferences or flat cross-sec
tions that go through the centre o f the sphere. Such
Let us merge the lists o f fast and named stars from distances on spheres are called arc distances; we shall
tables P l.l and P 1.2 (from Annex 1 ) in order to study simply refer to them as distances.
them together. We shall choose those stars from the Let the distance between the stars be minimal for
multitude that one finds in the Almagest according the moment t* = We have dubbed moment t* the
to [1339]. The resulting list consists o f circa 80 stars. individual dating by a given star in section 1 . When
Let us calculate the trajectory o f every star from this t deviates from the t* value into either direction, the
list in the Almagest coordinate grid as we have done distance between the real calculated star and its rep
in section 1 for the eight fastest stars. resentation in the Almagest begins to grow.
Be sure to mark that for this purpose we have Let us conside the dating interval [t\, t*2] = [tn ,
fixed a certain t value as the presumed dating and cal ti2] where the distance in question does not exceed 30'
culated the location o f each star for the epoch t in correspond to every star with the number i from the
the ecliptic coordinates o f the epoch. This position list. This interval can actually be empty, which shall
can be represented as a point on Ptolemys star atlas be the case if the distance between the calculated star
- that is, an atlas built from the Almagest catalogue and the respective star from the Almagest exceeds 30'
under the assumption that it was compiled in epoch for moment t. The centre o f the interval shall be de
t. Changing the value o f the alleged dating t within fined by value t*. See fig. 3.9.
the range o f the historical interval under study, we The 30' limit for the arc distance between the
are making the star, or point, move along Ptolemys Almagest star and the corresponding calculated star
atlas across the stars o f the Almagest. As time t alters, was chosen with the goal o f having most o f the
the calculated star i moves across the stars o f the Al Almagest stars stay within it. Indeed, if we are to con
magest (proper star motion as well as the slight shifts sider the average square error rate in the arc distance
o f the ecliptic that take place with the course of time). for the Almagest stars to exceed 40' (which concurs
The distance between the calculated point or star with the research conducted in [1339] and [614]),
and the Almagest star that this star becomes identi more than half the stars in the Almagest must be rep
fied as also changes in its turn. The identifications resented with the precision rate of circa 30'. We are

basing this on the hypotheses o f normal error distri 2) Failing to cross the a priori interval 0 < t < 30
bution and o f error independence as taken for indi and located beyond the area o f the graph.
vidual stars. Due to the approximate nature o f our 3) Covering the a priori interval completely.
narrative, possible discrepancies that these presump In the latter case, the coordinates o f the star must
tions might lead to do not affect our corollaries. have been measured with enough precision for the
The set o f the intervals that we calculate in this 30' interval; however, one cannot date the observa
manner, or the proximity intervals, can be seen in tions in the interval between 1100 b . c . and 1900 a . d .
fig. 3.10. What we see here is the time axis beginning by the positions o f such stars since their movement
with t - 0, or 1900 a . d . , and ending with t = 30, or is too slow.
1100 b . c . Each interval has a centre defined by the op Let us give Baileys numbers o f the Almagest stars
timal dating t{ for a given star. We also mark the points for which the 30-minute proximity intervals cover
for which the distance between the Almagest star, the entire interval 0 < t < 30 given a priory (see [ 1339]
or the position given in the Almagest, and the calcu and [ 1024] ). These are the stars with numbers 35,36,
lated star, equals 10 ' and 20' (see fig. 3.9). Lines rep 1 6 3 ,1 9 7 ,2 2 2 ,3 1 6 ,3 1 8 , 375 and 768.
resenting distances under 10' are heavier as seen in fig. Only partial intervals are given for many stars.
3.10. The ends o f the intervals are marked with point This happens when part o f the interval is located out
ers where they stay within the graph. side the a priori interval o f 0 < t < 30 and thus fails
Many o f the stars in our list o f fast and named to be represented in fig. 3.10.
stars do not have a corresponding interval in fig. 3.10. Next to each interval one sees the number o f the
This should imply the interval in question to be: corresponding Almagest star in Baileys numeration.
1) Altogether nonexistent (in cases when the dis The name of the modern star identified as the cur
tance between the Almagest star and the calculated rent Almagest star, as well as its own special name, in
star remains greater than 30' in all cases). case o f its existence, is given next to the equal sign.

25 20 15 10 5 1
600 B.C 100 B C 400A D . 900A.D. 1400A.D.
- - --I---------------------:------------------------
8 1 8 * Sirius 848=Procyon
^ ..
t 110=Arcturus
457=y Lao ^
" 1....1 ' 1 |
^ 488=D enebola
4 6 9 = Rgulas

2 3 4 = a Oph
7 3 6 = Bellatrix
553=a$cor ............
3 f5 = a A n d g^.
4 4 0 = y Gem ^ 349=yAndr
7B 1* I 0 n ______ _ _ _______I ________________________ . ____________________________^
28 8 =A quiia 7 3 3 = $ Cet
^ ___
2 4 = a UMa

1 4 9 = Lyra = Vega
185= $ Cas

Fig. 3.10. Intervals o f maximum proximity between visibly mobile fast or named stars with their corresponding positions as
specified in the Almagest.
10 4 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

In fig. 3.12 we reproduce a similar graph for latitudes;

the moment t = 18 is represented with a dotted line
and stands for the Scaligerian dating o f the Almagest
(around 100 a . d . ) .

2.3. Dating the Almagest with the suggested

method utilizing arc distances of individual
stars is an impossibility

Fig. 3.10 tells us very explicitly that time values t

which would belong to all the maximal proximity
intervals simultaneously do not exist. Let us raise the
precision threshold starting with the 30' value as cho
sen above, in order to obtain the desired values o f t.
The intervals as seen in fig. 3.10 shall grow respec
tively, with pointers indicating the direction o f Fig. 3.11. Latitudinal discrepancy for the real calculated star
and its position as specified in the Almagest.
growth. At some moment, all the intervals shall begin
to intersect. Let us see what value o f t and precision
threshold value it should take for this intersection to stable in the following sense. An alteration in the
occur the first time. It turns out that it takes place with compound o f the stars under study (which is obvi
t 12, or around 700 a . d . , with the precision thresh ously chosen rather arbitrarily) can shift the dating
old o f about 60', or one degree. If we keep raising the moment rather significantly.
precision threshold, the intersection interval will grow It is clear that such a situation makes all claims o f
in both directions from the point t - 12 . a reliable deduction o f the Almagest catalogue com
However, we cannot regard point t = 12, or 700 pilation date quite void.
a . d . , as a reliable enough estimate o f the date when

the author o f the Almagest catalogue carried out his 2.4. Dating the Almagest catalogue with the
observations since the intersection o f all maximal suggested method based on latitudinal
proximity intervals in fig. 3.10 only takes place at discrepancies of individual stars also proves
the precision threshold o f 1 degree, which implies impossible
the existence o f very poorly-measured Almagest stars
in this set. The error in the estimate o f their position Let us consider another method o f calculating
contained in the Almagest equals one degree at the maximal proximity intervals for the Almagest stars
very least. from our list o f fast and named stars. This method is
Furthermore, if we are to estimate the precision of similar to the one described above, the difference
stellar coordinates from below with the aid o f the se being that this time the distance between the Almagest
lective average square arc error in the optimal point star and the corresponding calculated star is com
t - 12 , we shall have to raise the acceptable error rate posed o f the latitudinal discrepancy and not arc seg
value (or the precision threshold) excessively (over 2 ments. By latitudinal discrepancy we mean the pro
degrees). However, such a value o f the precision jection length o f the interval that connects these two
threshold shall make the acceptable maximal prox stars over the Almagest coordinate grid meridian (see
imity interval intersection cover the entire period fig. 3.11). The choice o f a latitudinal discrepancy (as
between 500 b . c . and the present (see fig. 3.10). Such opposed to longitudinal, for instance) was made out
a corollary is o f zero scientific interest, since it is per of the following considerations: firstly, it is well known
fectly understandable that the Almagest was created that the Almagest star latitudes are more precise than
somewhere in this great time period. the longitudes (qv in [1339], for instance, as well as
Moreover, the very dating o f 700 a . d . is rather un- Chapter 2 o f the present book). Secondly, the latitu

dinal discrepancy does not depend on how we posi tervals in fig. 3.12 than in fig. 3.10, which represent a
tion the Almagest in relation to the calculated cata greater amount o f stars.
logue K(t) in terms of longitudes, qv in Chapter 1. Maximal proximity intervals for all the stars in fig.
Thus, we shall manage to evade making additional er 3.12 apart from two stars in Centaur (935 = 2g Cent
rors which may result from such juxtaposition as well and 940 = 50 Cent) also begin to intersect at the level
as the possible arbitrary choice o f the initial longitu o f t - 12, or approximately 700 a . d . , latitudinal pre
dinal reference point (see Chapter 1 ). cision threshold equalling 40'. This is somewhat bet
In fig. 3.12 we see the resulting maximal proxim ter than the 60' value that we got in the previous case,
ity interval set for the case when the latitudinal dis but still nowhere near precise enough. We are brought
crepancy represents the distance. Once again, the to the dating o f roughly 700 a . d . once again, but, as
proximity intervals which cover the entire interval of in the above case, we cannot consider this result re
0 < t < 30, or 1100 b . c . to 1900 a . d . , are absent from liable due to the considerations related above; there
the graph. The Almagest numbers o f the stars whose fore, this method o f dating the catalogue gives us no
30-minute latitudinal proximity intervals cover the in tangible results.
terval 0 < t < 30 completely are as follows: 1, 35, 36, In general, regardless o f the fact that the transition
78, 111, 149, 163, 189, 222, 234, 287, 288, 315, 316, from the arc discrepancy to the latitudinal discrep
318, 349, 375, 3 93,410,411, 424 ,4 6 7 ,4 6 9 , 510, 713, ancy helps us rectify the errors o f the Almagest to
733, 760,761, 768,812 and 818. some extent and therefore allows for more precise sta
A comparison o f fig. 3.12 and fig. 3.10 demon tistical corollaries, the resulting intervals o f possible
strates that the longitudes of the Almagest stars under datings remain too great. They cover the entire period
study are indeed a lot more precise than their posi o f 4 < t < 20, or 100 b . c . - 1500 a . d . Such intervals
tions on the celestial sphere defined by both latitude give us no useful information in re the date o f
and longitude. This is exactly why one sees more in Ptolemys observations.

25 20 W 15 W 5 i
BOO B.C. 100 B.C.
t m A D - 900A D { ,4 0 0 A D - 848=Procyon
9 05= a Hydra
..r..................................... * 110=Arcturus
288 = Aquiia

Ihi. 41 .. KI I I m
h Hnrr 488-D ensbofa
;"Ni.1 ......... ^ /SP- ]Cas
i fct IOC o f*
. ' 247= 36A Oph ............................... i . ................................. ik. n e tt ....

..................... ...... ....... ........... ........ If OWjr

-------------------------------------------------- ~ w 5 5 7 = e Scor
i < 7 7 0 . n2 C ri
m 1 * 78fl=38o' Eri
783=238 Eri
( 6 = W /2 53 Pup (
9 *0= 6 Cent
$Z9=gLup * ---------------------

103 = e Boo
< -------------------------------

Fig. 3.12. Intervals o f maximum latitudinal proximity between the visibly mobile real fast stars and named stars and the
corresponding Almagest stars.
io6 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

3. 3.2. The stars chosen for the experiment

CATALOGUE BY THE MOTION OF We shall keep comparing the positions o f all in
INDIVIDUAL STARS AS COMPARED TO THE dividual fast-moving stars on the real star chart with
OBJECTS IN THEIR IMMEDIATE VICINITY their positions as specified in the Almagest. However,
now we shall be comparing the positions o f the stars
3.1. The varying geometry of stellar on the real chart and in the Almagest to a certain set
configurations as seen against the background o f referential stars pointed out on the real star chart
of "immobile stars" as well as the Almagest. For this set we have chosen
either named stars (Aldebaran, Scheat etc), or those
In sections 1 and 2 we tried to date the catalogue which definitely stand out in brightness amongst the
with rough methods based on various stellar config stars that surround them. We excluded the stars whose
urations altering over the course o f time due to the coordinates might have been affected by refraction
proper movements o f individual stars that comprise from the list o f referential stars. 45 stars altogether
them. We have considered each star in the configu were chosen, among them such visibly mobile ones
ration individually, comparing its calculated position as Arcturus, Sirius, Procyon, Capella, Aquila = Altair,
to the one given in the Almagest. In order to compare Denebola, Caph and Regulus. Thus, the position of
all these positions we had to use the Newcomb the a mobile star on the real celestial sphere is determined
ory that describes the movement o f the ecliptic co in reference to a basis that is mobile as well. The re
ordinate system used in the Almagest across the sulting picture alters depending on the alleged dat
sphere o f immobile stars over the course o f time. ing and is compared to the respective picture as re
Let us see what results we can obtain from the flected in the Almagest.
method o f dating the Almagest that will not use the Let us take the average configuration discrepancy
Newcomb theory. The idea behind a method of this o f stellar arc distances as the deviation measure:
sort is simple. One doesnt compare the positions o f
individual stars on the real theoretically calculated Ai (0 = T 7 X | Preal (S i j (S i j ) |
star chart to their positions in the Almagest, but rather A M
the geometry o f stellar configurations (which change N stands for the quantity o f referential stars,
due to the proper movements o f stars) to the config preai(Si, Oj} t) is the arc distance between the star S{
urations from the Almagest catalogue. The only thing and the referential star Oj on the real celestial sphere
required from us for such a comparison is the knowl o f epoch t. Furthermore, pA/m(Sl-, 0 ; ) is the arc dis
edge o f velocity values o f the individual stars proper tance between the star S{ and the Almagest star Oj.
motion - not the Newcomb theory. The time moment t{ when the value o f A (f) reaches
Although the errors resulting from the Newcomb its minimum shall be referred to as the individual
theory are rather small (several orders smaller than dating by the star in question. If the individual dat
the Almagest catalogue grade value), the study of con ing values t{ for all the fast stars o f the Almagest cat
figurations is a lot simpler this way from the calcu alogue or at least their majority fall into a short
lus point o f view. enough time interval, said interval should either in
Proper movements o f stars are nowadays meas clude the real date o f Ptolemys observations tA or be
ured with great precision with the aid of telescopic ob located in its immediate vicinity. However, the real
servations ( [ 1144] and [1197]). The values o f proper status quo appears to be altogether different.
star movements and the table that identifies the
Almagest stars as their counterparts on the modern 3.3. The behaviour of the individual
star charts comprise the only data that we are to use discrepancies and the average discrepancy
here. The identification table was borrowed from
[1339]; we have omitted the ambiguous cases indi We have studied the behaviour o f the A, (t) dis
cated therein. crepancies for eight rather fast stars contained in the

that the reasons for such a great scatter range o f in

dividual datings arent related to the conversion
method as applied to the coordinates o f the celestial
sphere, but rather relate to the low precision o f co
ordinates offered by the dated catalogue, the possible
heterogeneity o f the catalogue etc. The latter might
be caused by different positions o f the instrumental
ecliptic during measurements performed in differ
ent observatories, which produce different system
atic errors for various groups o f stars.
In section 5 o f the present chapter we shall ana
lyze the coordinates o f the Almagest stars as well as
the general structure o f the Almagest catalogue in
order to discover all the factors that might be caus
ing this.

Fig. 3.13. Individual discrepancies for mobile stars and the
average discrepancy in eight configurations. It is obvious that 4.1. A lot of the errors are not produced by
one can make no definite conclusions.
astronomical phenomena and stem from the
incorrect application of the methods offered
Almagest catalogue, namely, Capella (Baileys num by mathematical statistics
ber = 222), Arcturus (110), Aquila = Altair (288),
Denebola (488), Regulus (469), Sirius (818), Procyon Let us analyze different authors attempts to date
(848) and C aph(189). the Almagest by proper star movements.
We have deliberately chosen the most famous The articles o f the astronomers Y. N. Yefremov and
and the brightest o f the Almagests fast stars and omit Y. D. Pavlovskaya ([273] and [274]) were published
ted the dim ones. As we point out above, the coordi in reference to our publications; they represent an at
nates o f dimmer stars may be represented in the tempt to confirm the Scaligerian dating o f the
Almagest very imprecisely. Therefore, their inclusion Almagest star catalogue by proper star motion. The
into the sample can make the scatter range o f indi corollary formulated in [273] is as follows. The
vidual datings a lot wider. Almagest catalogue can be dated to an early a . d .
Fig. 3.13 demonstrates the graphs o f individual epoch by proper star motion with the precision
discrepancies for the indicated fast stars Af (f) as t threshold o f 100 years. The authors go as far as nam
functions as well as the average graph for all these ing the date o f 13 a . d . 100 years.
stars. Unfortunately, this graph turns out almost uni In [274], which is a more in-depth publication, the
form over the entire time interval o f 1100 b . c . - 1 9 0 0 authors formulate their corollary with more caution:
a . d . (see fig. 3.13). The Almagest star catalogue has thus already been
observed in the antiquity; most probably, by Hip
3.4. Negative experiment result parchus. It is however possible that the brighter stars
were observed by Ptolemy himself. Some sort o f ar
Our refusal to use the Newcomb theory did not gumentation to support this can be found in the fact
lead to the concentration o f different datings by in that the epochs that we got for Arcturus and Sirius,
dividual stars on the time axis. The implication is the two stars o f the first magnitude present in our
io 8 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

sample, are 2-4 centuries more recent than those for the Almagest as well as on the real celestial sphere
the rest o f the stars ([274], pages 189-190). and the alleged dating o f the observations conducted
However, the actual contents o f [273] and [274] by the author o f the Almagest catalogue. We are told
imply no such corollary. Let us briefly follow the rea that the epoch T0 can be estimated with enough con
soning patterns o f Y. N. Yefremov and Y. D. Pavlov fidence, the minimum o f the function Ar 2 (t) being
skaya using their more extensive publication ([274]), drastic and deep ([274], page 183). However, the il
although everything we say shall also refer to their ear lustration that the authors o f [274] are referring to
lier work ([273]). Let us point out that Y. N. Yefremov (page 185, ill. 3) implies that the alteration o f the al
hasnt made any scientific publications on this sub leged dating by 1000 years makes the value o f the
ject ever since the respective publications o f said square average discrepancy VA r 2( t ) alter by a mere
works ([273] and [274]) in 1987 and 1989. However, maximum o f 13 for all configurations except for a
quite a few o f his popular articles have appeared in single group, that of o 2 Eri. See more about this group
newspapers and literary magazines. Still it has to be below.
said that both his publications ([273] and [274]) con Let us see how significant the 13 deviation from
tain errors which were pointed out to their author in the square average discrepancy really is for the situ
our book [METH3]:2, pages 99-103. It would make ation regarded by Y. N. Yefremov and Y. D. Pavlov
sense for Mr. Yefremov to correct these errors prior skaya. The Almagest scale grade value equals 10',
to advertising the results o f his research in popular whereas the real precision threshold o f the stars in the
press. Moreover, we are o f the opinion that these er Almagest estimated as the square average arc dis
rors cannot be corrected - in particular, due to the crepancy equals roughly 30' (see [1339] and [614]).
erroneous dating offered by Y. N. Yefremov, qv below. If we are to base our estimations on the proper move
The dating of star catalogues with the method de ments o f the stars under study, it will imply that the
scribed in [273] and [274] is based on the compari precision estimate according to the method offered
son of stellar configurations that alter over the course in [274], which is based on the minimal square aver
o f time with the respective configurations as given in age configuration discrepancy, must allow for the
the Almagest. It turns out that the main part in the value o f this discrepancy to fluctuate within a much
change o f an individual configuration is played by a greater range than 13' - circa 20'-30'. This leads to the
single star contained therein, the fastest one (the dating intervals o f 2-3 millennia. In other words, the
group o f Arcturus, the group o f T Cet etc). We shall possible discrepancy rate for the dates cited in [274]
be using the same terminology. equals 1000-1500 years. See more details concerning
The dating o f a catalogue by an individual con the precision o f the method related in [273] and [274]
figuration is supposed to be such a dating for which below. However, dating the observations performed
the set o f pairwise distances between the stars o f this by the Almagest compiler with such low precision
changing configuration is the closest to the set of such doesnt allow for making a distinction between
distances as given in the Almagest. Proximity is de Ptolemys epoch and our age, let alone the Scaligerian
fined in the square average sense. datings o f the respective lifetimes o f Hipparchus
What one gets as a result is naturally a certain ap (II century b . c . ) and Ptolemy (II century a . d . ) . Such
proximation of the date when Ptolemy or some other a result is o f zero scientific value. It is obvious that the
observer who had compiled the Almagest catalogue Almagest was created during the last two millennia
were making observations - not the actual date. What at any rate.
are the possible discrepancy rates o f such approxi Therefore, this error, as well as the ensuing mis
mation, one wonders? There is no factual reply to takes made by the authors in question, is o f a math
this question given anywhere in [274]. ematical nature and not astronomical. The methods
The discussion of the issue of discrepancy rates for o f mathematical statistics are either misused or alto
the resultant datings is left out in favour o f a refer gether neglected. The claims made by Y. N. Yefremov
ence to the dependency graph o f the square average in re the alleged high precision o f his methods dont
discrepancy between the sets o f pairwise distances in hold up to the simplest criticisms. It is most peculiar

that Y. N. Yefremov keeps insisting on the veracity of method such as this one is mere tailoring o f re
his erroneous results in the field o f Almagest-dating search results in such a way that they would corre
publicly after all these years, the situation being as spond to the desired values known a priori.
described above. This concerns his numerous public All o f this makes the results claimed in [273] and
speeches and popular magazine and newspaper pub [274] wholly insubstantial. These results are erro
lications oriented at the general public. neous, and therefore cannot confirm the Scaligerian
datings o f the old star catalogues.
4.2. The data in Y. N. Yefremov's works
on the dating of the Almagest were tailored 4.3. A vicious circle in the dating
to fit the desired result of the Almagest by the movement
of the star o2Eri
Y. N. Yefremov and Y. D. Pavlovskaya claim in [274]
that the star catalogue dating method that they offer Let us analyze the dating o f the Almagest by the
was tested on three veraciously dated catalogues - group o f o 2 Eri as offered in the works o f Y. N. Yefre
namely, the catalogues o f Ulugbek, Tycho Brahe and mov ([273] and [274]) in more detail, since it is this
Hevelius, and that the application o f the method in dating that Y. N. Yefremov bases his conclusions upon
question to all three catalogue gave an incredibly pre de facto.
cise result. The dates when the catalogues o f Tycho We have already referred to the star o 2 Eri above,
Brahe were compiled were restored with the preci in section 1 . Bear in mind that its identification as one
sion threshold o f 30-40 years, and Ulugbeks cata o f the Almagest stars is largely dependent on the al
logue, the least precise o f the three, was dated with leged dating o f the catalogue. In other words, the an
the mind-boggling precision o f 3 years! swer to the question o f who is who in the Almagest,
However, one cannot overlook the alarming cir or, in other words, whether the star o 2 from the con
cumstance that each of these datings was calculated stellation o f Eridanus is represented in the Almagest
by its own stellar configuration - namely, the datings at all, and if so, under which name, varies to a great
for the catalogues o f Tycho Brahe and Hevelius were extent as the a priori known dating o f the catalogue
obtained from the Arcturus groups, and the dating of changes.
Ulugbeks catalogue comes from the data obtained Let us remind the reader that the star o 2 Eri moves
from the group o f T Cet. Other stellar configurations fast enough, which changes its celestial position. In
for each of the three catalogues in question arent con the course o f its movement it becomes consecutively
sidered at all. Why would that be? We shall promptly identified as different stars o f the Almagest - namely,
answer this question. the three o f them that one finds on the historical in
Furthermore, the main result of Y. N. Yefremov terval o f the last 2,500 years. Baileys numeration of
and Y. D. Pavlovskaya concerning the dating o f the these Almagest stars is as follows: 778, 779 and 780.
Almagest is also de facto obtained from a single soli The star #779 is traditionally identified as o 2 Eri (qv
tary configuration - group o 2 Eri, although they make in [1339] ) due to the mere fact that in the beginning
formal references to having studied 13 configura o f the new era the star o 2 Eri had occupied a position
tions. The analysis o f the datings that they came up close to that o f the star 779 on th Almagest star atlas.
with for all three catalogues demonstrates that in each However, what we face here is clearly an implica
case the choice of the actual stellar configuration used tion o f the Almagests being roughly dated to the be
for the dating o f the catalogue was conditioned by the ginning o f the new era. If we are to make no pre
Scaligerian dating o f said catalogues creation, whose sumptions in re the dating o f the Almagest, we in
veracity the authors o f [273] and [274] were trying stantly find other candidates which we could identify
to prove. In other words, Y. N. Yefremov and Y. D. Pav as the moving star o 2 Eri. For instance, on the inter
lovskaya chose such stellar configurations for each val o f 900-1900 a.d ., the star which corresponds to
catalogue in [274] that would concur best with the the real position o f o 2 Eri is #780. On the other hand,
Scaligerian dating o f the catalogues compilation. A the star #779 from the Almagest does not remain
no | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

unidentified in this case either, since it can be suc Simple considerations allow for an easy estimation
cessfully identified as the star 98 Heis (see [1339], o f the real precision that the methods leading prin
page 117). Furthermore, this is the exact identifica ciple is based upon (as related in [273]). Indeed, the
tion o f this star which was made by the astronomer Almagest position o f a given moving star is deter
Pierce, qv in [1339]. mined in relation to certain stars in its vicinity ([273])
We must emphasize that the star o 2 from the con - the Arcturus group in case o f Arcturus. The Arc
stellation o f Eridanus is rather dim, likewise the ones turus group contains 11 stars. The position o f Arc
that surround it. Their magnitudes range from 4.2 to turus in relation to this group is used for the estima
6.3. Therefore, the only way o f identifying them as tion o f its position on the star chart theoretically cal
Almagest stars is coordinate comparison. The bright culated backwards for the epoch t. These positions are
ness o f these stars is roughly the same, and Ptolemys then compared to each other.
verbal descriptions of the stars in this part of Eridanus All the stars o f the Almagest are measured with er
are laconic and extremely vague. Therefore, a reliable rors o f some sort. This definitely applies to the
identification o f these stars by any other properties group stars - in particular, all the stars from the
but their coordinates is impossible. The proof o f o 2 group o f Arcturus. Let us however make the tempo
Eri being veraciously identified as a star from the rary presumption that the measurements o f the stars
Almagest catalogue as cited in [274] is based on late in the vicinity o f Arcturus were carried out with ideal
identifications o f the Almagest stars, or, alternatively, precision. Even in this case the error rate in the Alma
upon dating the catalogue to II century a . d . in actu gest location o f Arcturus cannot be less than 10 ' by
ality. The use o f such proof for independent dat any coordinate, since this is the grade value o f the
ing obviously leads us to a vicious circle. Almagest star catalogues coordinate scale. In reality,
Therefore, what we see in the works o f Y. N. Yefre this rate has to be raised due to the imprecise coor
mov and his co-authors ([273] and [274]) is in fact dinates o f the stars in a given group.
the assumption that the Almagest was compiled in This leads to the arc distance error o f circa 14 for
the early days o f the new era used as the basis for the [273]. If the possible error rate for each o f the coor
corollary that the Almagest dates to 13 a . d . 100 years. dinates equals 10', it shall equal 14' for the hypotenuse
This is the very vicious circle that were talking about. according to the Pythagorean theorem. Proper move
ment speed for Arcturus is roughly 2" per year. There
4.4. Y. N. Yefremov's errors in the precision fore, Arcturus covers the distance o f 14' in about 420
estimation of dating the Almagest by Arcturus years. This is but a rough estimation o f the methods
Let us now turn to Arcturus - the second and last In reality, the actual precision o f the position o f
star discussed in the work o f Y. N. Yefremov and Y. D. Arcturus in the Almagest may be given with an error
Pavlovskaya ([273]). The Almagest identification of rate that substantially exceeds 14', and the dim stars
the Arcturus is unambiguous. The first proper mo in its vicinity could be measured with even less pre
tion dating o f the Almagest that we encounter in cision. What we are referring to here is naturally the
[273] is 250 a . d . Then the authors adjust this dat arc distance error. As we shall see below, the latitude
ing and end up with the dating of 310 a . d . 360 years o f Arcturus was measured with sufficient precision
calculated by one of the configurations. We shall deal in the Almagest - however, this does not apply to its
with this adjustment below. longitude (see Robert Newtons research in [614],
The dubiety of the results published in [273] and for instance). Moreover, one has no reasons to as
[274] was also commented upon by other authors. sume that Ptolemy measured any o f the dim group
M. Y. Shevchenko, for one, makes the justified re stars precisely. Therefore, the real precision o f the
mark in re [273] that the catalogue dates to the I cen method related in [273] is a lot worse than 420
tury b . c .; however, the precision and hence the ve years. Therefore the interval of possible datings of the
racity o f this result leaves much to be desired so far Almagest obtained with this method is a priori
on page 184 o f [968]. known to be greater than 200 b . c . - 700 a . d .

Let us now comment upon the random error the dispersion estimation o f the local error in the
modelling method as offered in [273] and [274] for Almagest catalogue. Y. N. Yefremov tells us rather
the precision estimation o f the resultant dating. For plainly that grouping the same n quantity o f stars in
instance, this method brought Y. N. Yefremov to different ways, we shall obtain a number o f estima
the conclusion that his dating o f the Almagest to tions Ex,p. They arent independent; therefore, instead
roughly 300 a . d . had the precision o f 300-400 years o f averaging them we shall choose the maximal value
(see [273], page 311, and [274], page 181). which shall be considered the estimation o f the local
The method o f minimal squares is used for the coordinate determination error in the Almagest cat
purposes o f dating in [273] and [274]. The elemen alogue ([273], page 311). One wonders just why.
tary calculations cited above demonstrate the preci Firstly, the local error o f the Almagest has to be esti
sion o f this method to be estimated in accordance to mated separately, which is necessary for the under
the individual error rate pertinent to the Almagest po standing o f just what minimal level variation we must
sition o f the star under study divided by the speed o f allow for in order to reliably cover the real dating of
its proper movement. the catalogues compilation. When Y. N. Yefremov
Y. N. Yefremov uses the method o f random mod takes the actual minimal value for dispersion esti
ulation o f the Almagest errors in order to raise the mation, he basically fails to allow for the variation o f
precision o f his method. The precision o f the mod this minimum altogether.
elling method that he suggests (multiple perturba Secondly, the sample volume used for the averag
tions o f the Almagest star coordinates resulting from ing of the value in question is too small (circa 5-6 in
the application o f some random value comparable dependent observations) and doesnt permit to con
to the catalogue precision) isnt estimated anywhere sider Y. N. Yefremovs estimation precise enough.
in his works. Nevertheless, this method will only work Local error needs to be estimated from a much greater
if the results o f these random perturbations shall quantity of stars.
make the Almagest stellar coordinates approximate Furthermore, Y. N. Yefremov models random per
the real ones with distinctive probability. However, turbations o f Ptolemys coordinates using his esti
due to the effect o f the individual error mentioned mated local error rate as basis. He writes that the
above, the probability o f such coincidence with the knowledge o f the error rate Ex, pfor each group makes
area o f real coordinates shall most probably be very it feasible to conduct a numerical experiment in order
low. At any rate, this probability has to be estimated; to study how the estimation o f T0 is affected by ran
there isnt so much as a hint of such estimation any dom coordinate errors. Let us model the corrections
where in [274]. In general, the methods offered by the o f stellar coordinates from the Almagest catalogue,
authors o f [273] and [274] dont hold water from the considering these corrections to be distributed nor
point o f view o f mathematical statistics. mally with the average o f zero and the square average
The dating modelling method as offered by Y. N. error x,p for each group and calculate the respective
Yefremov can be formulated in the following manner. value o f T0. Having repeated the procedure 100 times,
One is to consider a certain vicinity of a fast star - Arc- we can build a distribution graph for the resultant es
turus, for instance. Then one is to use the method of timations o f T0 ([273], page 312). Y. N. Yefremov
minimal squares in order to determine the date which proceeds to tell us that the common interval for all
gives us a minimal square average discrepancy o f the the groups with the square average errors for the
mutual distance set o f the Almagest stars from the set epochs o f T 0 taken into account is the I century b . c .
of the same values in the real stellar configuration ([273], page 313). Y. N. Yefremov also makes the fol
that alters over the course o f time. This dating is used lowing flabbergasting statement: the probability rate
for the estimation o f the real date when the catalogue o f T q s random value exceeds 900 equals 0.2, and that
was compiled, which is unknown. Y. N. Yefremov for a group with maximal dispersion. Therefore, the
marks said dating as T0. Almagest catalogue is most unlikely to be a mediae
Furthermore, the resultant minimum o f square val forgery ([274], pages 188-189). Thus, Y. N. Yefre
average discrepancy is for some reason declared to be mov apparently assumes that the average date must be
112 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

close enough to his randomly modelled date T 0, es This is untrue. The abovementioned simple esti
timating this proximity whilst taking the square av mation demonstrates the real date to be far enough
erage errors as calculated above into account ([273], from the centre o f such modelled distribution (for
page 313). instance, this range exceeds 420 years for Arcturus,
This approach is utterly delusional. It is obvious qv above). At the same time, the scatter range o f
that what Y. N. Yefremov determines to be the aver modelled dates around a shifted date might not be
age modelled date T 0 is merely his initial estimation all that great. The matter is that Y. N. Yefremov takes
o f T0 with some random perturbation added thereto an unreasonably low value of the square average error
by the author himself. As for the distribution of mod obtained from parabolic minimum for this model
elled dates, what he comes up with is a random dis ling, making no specific estimations o f this error for
persion with the centre equalling T0 for a given group. some reason.
Y. N. Yefremov is of the opinion that the real date Apart from that, it is easy enough to estimate that
must be close to the centre o f this dispersion, or, in even if one is to model the correction for the coor
other words, that the random perturbations that he dinates o f a single star, the probability o f returning
introduced have a certain real probability o f covering to its true position is very small in general. This is con
the real positions o f Ptolemys stars. In other words, firmed by the following simple calculation. Let us as
he hopes that his modelling will randomly cancel out sume that Ptolemys individual error for a given star
Ptolemys errors, estimating their probability to be equals 45 arc minutes. Such errors are typical for the
real. This is the exact meaning o f the passage quoted Almagest - a great number o f stars it contains were
above where Y. N. Yefremov tells us that a post-900 a . d . measured a lot worse ([1339]). Let us re-emphasize
dating can only be achieved in the course o f this mod that we are referring to the arc error. Latitudinal er
elling with the minute probability rate of 0.2. He is rors are a lot smaller, as we shall demonstrate below.
o f the opinion that this makes a mediaeval dating of If we apply the above calculations to Arcturus, for
the Almagest highly improbable. instance, the implication is that in order to model an
However, one has to bear in mind that his initial actual dating that would differ from the original by
dating Toy which the modelled datings are grouped 400 years maximum, one has to hit the 14-minute
around differs from the real date by a certain value. range around the stars real location (provided that
The value o f this shift, as we have demonstrated the group stars have already fallen into necessary po
above by simple calculations, can be great enough. sitions and do not affect the dating too greatly). The
In case o f Arcturus its lowest possible value is 420 maximum probability of the value falling into this 14-
years, qv above. Said shift is defined by Ptolemys in minute range from a position shifted by 45' can be
dividual error in the estimation of a given stars co estimated as the probability o f its falling into the
ordinates, as well as individual errors for the stars of shaded sector on fig. 3.13a.
the chosen group. Also, our calculations demonstrate If we are to consider the probability of a perturbed
that the value in question is largely dependent on point being located in the 60' radius o f point A to
the group choice. Therefore, some individual error equal 1 , we end up with the probability o f 0.1 for its
is already inherent in the value T0, possibly a serious location in the shaded sector. Thus, even in this ideal
one. When Y. N. Yefremov models his additional er case the probability rate o f obtaining the necessary
rors for group stars, he already distributes them dating randomly - not even the correct dating, but
around a certain dating which might be shifted side rather one that wont differ from it by more than 400
ways to a substantial degree. However, in his refer years, equals 0.1. Still, Y. N. Yefremov is o f the opin
ence to the graphs o f modelled distributions, Y. N. ion that the probability threshold o f 0.2 already suf
Yefremov appears to assume that the real dating must fices for rejecting the post-900 a . d . datings as im
be located near the centre o f these distributions in probable.
every case - at least, within a certain confidence in The authors o f [274] claim that the results o f cal
terval with the probability ratio o f 0. 8, since he con culations performed by other fast stars (which arent
siders the probability o f 0.2 to be too low. cited in their work for some reason) confirm the con-

elusions made in the research o f Arcturus and o 2 2. If we are to strip the works in question ([273]
Eridani. However, this statement does not correspond and [274]) from all such circular considerations,
to reality. the discrepancy we end up with does not contradict
Let us provide a single vivid example. Among the our dating, qv below.
fast stars which were processed by the authors o f [273] 3. The positions o f Y. N. Yefremov and Y. D. Pav
and [274] we find Procyon, a star which was famous lovskaya that concern the precision estimates o f their
in mediaeval astronomy. Our research (qv in sec method (and the correction modelling o f the Alma
tion 1, for instance) demonstrates that Y. N. Yefre gest) as seen in [273] and [274] are mathematically
movs method must have led to the dating of roughly illiterate and void o f meaning in our opinion.
the X century a . d . by Procyon, which would blatantly 4. The authors o f [273] and [274] failed to con
contradict his conclusions. For a mysterious reason, sider Procyon, which gives a blatantly non-Scaligerian
[273] tells us absolutely nothing about the results for dating, for some unknown reason.
Procyon. The work o f Y. N. Yefremov and Y. D. Pavlovskaya
Finally, the method related in [273] and [274] is ([273]) was published in the Doklady Akademii
largely dependent on the group contingent choice for Nauk SSSR in 1987. We pointed out the errors con
the fast star under study. We have checked how the tained in [273] and [274] in our articles [350] and
result of the dating by the Arcturus group changes de [355], which were published in the Doklady Aka
pending on the choice o f various stars for this group. demii Nauk SSSR in 1989 and 1990, respectively.
It turns out that when we change the contingent o f Apart from that, we have personally addressed Y. N.
the group, the Arcturus dating may vary from 0 a . d . Yefremov with a criticism o f his errors at the semi
to 1000 a . d . - that is, the results can fluctuate with nar hosted by the Institute of Natural Scientific and
the amplitude of up to a thousand years. This very cir Technical History in 1989. Y. N. Yefremov did noth
cumstance completely invalidates the method offered ing to rectify the errors in question - moreover, he
by Y. N. Yefremov. evades all attempts o f their discussion.

CORROLARIESI 4.5. Erroneous precision estimation of

1. The result of dating the Almagest by proper star astronomical calculations: another example
motions as claimed by Y. N. Yefremov and Y. D. Pav
lovskaya in [273] and [274] is based on thin air. Fur Let us consider another publication that deals with
thermore, some o f the considerations one encounters the issue of Almagest dating ( [ 179] ). Its authors, Y. S.
in said works contain a vicious circle. Goloubtsova and Y. A. Zavenyagin, refer to Galley re
porting that over the time that passed between Pto
lemy and Galley (up to 1690, which is when Flam
steeds star catalogue was created), Arcturus shifted
in the Virgo direction by 1.1 degrees. Having com
pared this to the annual shift value for Arcturus
(2.285"), Goloubtsova and Zavenyagin perform the
following simple calculation, writing that if we are
to divide 1.1 degrees by 2.285 angular seconds per
year, we end up with 1733 years. Finally, once we sub
tract 1733 from 1690 (or the year when Flamsteeds
catalogue was compiled), we shall come to the con
clusion that the Almagest catalogue was compiled in
43 b . c . The discrepancy error rate for the coordinates
o f neighbouring stars is a lot smaller than the error
o f the actual coordinate, since the subtraction re
Fig. 3.13a. moves the systematic error. Therefore, the average
ii4 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

error rate in the positions o f bright stars in relation error in the bright stars positions in the Almagest
to their neighbours in the Almagest does not exceed not exceeding 0.1 degrees, or 6 ', isnt based on any
0.1 degrees [? - Auth.]. The implication is that the thing whatsoever. Why 6 ' and not 2 or 15? Having
possible dating error rate does not exceed 150 years said everything about the precision estimation prob
([179], page 75). lem o f the Almagest stellar coordinates, we deem a
Thus, if the authors o f [273] date the catalogue to deeper study o f this issue superfluous.
250 a . d . by Arcturus (and even to 310 a . d . after mak The authors o f [179] did not limit their research
ing their adjustment, estimate precision equalling to the study o f Arcturus and its behaviour. They also
360 years in this case), the authors o f [179] per attempted to date the catalogue by another fast and
form a single solitary arithmetical calculation and well-known star - Procyon. Let us quote: We get a
date the Almagest to 43 a . d . , also by Arcturus, with similar result once we date the Almagest by the proper
the much greater precision rate o f 150 years. movement o f Procyon, namely, that the Almagest cat
However, the text from [179] as quoted above is alogue was compiled in 330 b . c . , with the possible
oriented at the reader who will not bother checking error rate o f 300 years... The Procyon dating serves
the real stellar configuration on the celestial sphere. as a perfectly independent corroboration o f the Arc
The calculations o f the authors o f [179] are based on turus dating, both o f which take us to the last cen
the taciturn implication that the own movement vec turies before the new era ([179], pages 75-76).
tor o f the modern Arcturus is directed exactly at its However, just as they had done in case o f Arcturus,
Almagest location. Had this indeed been the case, the authors did not take the direction o f Procyons
their calculations would have some sort of reasoning movement into account for some reason. Let us see
to back them up. However, this doesnt appear to be what dating we shall get if we are to use their
the case. In fig. 3.1 one sees the real movement di method for our own accurate calculations which
rection o f Arcturus in relation to its position as spec take real stellar positions into account. It turns out
ified in the Almagest. One can plainly see that Arctu that the real trajectory of Procyons movement is such
rus moves visibly sideways from its Almagest posi that a rough Procyon dating is the X century a . d . , no
tion. Therefore, it isnt the value o f 1.1 degrees that less (see section 1). It goes without saying that the
has to be divided by 2 , the way it is done by the au issue o f this datings precision remains standing.
thors o f [179] for some reason, but one that is a great
deal smaller, and shall yield the dating o f approxi 4.6. The "secondary analysis" of the Almagest
mately 900 a . d . , albeit with a significant possible error dating in the "Samoobrazovaniye"
rate due to the rough nature o f the method itself. See (Autodidactics") magazine
our considerations in re the precision o f this method
above. In the first 1999 issue o f the Muscovite magazine
Thus, dating the Almagest to 43 a . d . with the pos Samoobrazovaniye ([263]) we find a publication
sible discrepancy rate o f 150 years, as Y. S. Goloub- by A. S. Doubrovskiy, N. N. Nepeyvoda and Y. A.
tsova and Y. A. Zavenyagin claim to have done, is Chikanov entitled On the Chronology o f Ptolemys
completely out o f the question. Almagest. A secondary mathematical and method
Let us also point out that the very concept be ological analysis which deals with our dating o f the
hind [179], which implies the random errors in the Almagest by proper star movements in particular.
Almagest to be a result o f proper star movement, is Unfortunately, the authors o f [263] failed to fa
perfectly erroneous. Its absurdity is all the more ob miliarize themselves with the necessary astronomical
vious if we are to consider the examples o f slowly issues and thus made the false conclusion that the
moving stars which are almost immobile. The divi dating o f the Almagest by proper star movements is
sion o f a non-zero error o f the Almagest in the posi unreliable in general, as the speeds o f proper star
tion o f a star might yield any infinitely ancient ob movements are known rather badly, which is pre
servation dating. sumably reflected in great controversy one finds in as
The claim made by the authors o f [179] in re the tronomical literature.

Further in [263] we encounter a comparative table Almagest has to be preceded by an in-depth study of
o f proper movements as taken from the Astronomi- certain rather complex issues from the part o f the re
cheskiy Yezhegodnik (The Astronomical Yearly) searcher. It actually requires a great deal o f time and
and the catalogue [1197]. For instance, the reader is effort, even from a specialist.
invited to compare the values contained in both cat
alogues (-0.1098; -0.2001) and (-1.155; -1.998) re 5.
spectively. These are the proper movement speeds of CONCLUSIONS AND DIRECTIONS FOR
The authors o f [263] tell us exactly the following A BRIEF SYNOPSIS OF OUR MAIN RESULTS
in this respect: As for the analysis o f the fast star
motion, we must point out that the data concerning 5.1. The three problems one is confronted with:
the stellar speed taken by Fomenkos group from the identifying the Almagest stars, defining the
catalogue... [followed by a reference to the bright nature of possible errors, and analysing
star catalogue ([1197]) - Auth.] differ considerably the precision of the catalogue
from those contained in the Astronomicheskiy
Yezhegodnik ([263], page 23). Sections 1-3 contain accounts o f several attempts
Having cited this remarkable table on page 24 of to date the Almagest on the basis o f the numerical
[263], its authors come to the following conclusion: material contained in Ptolemys star catalogue. All o f
As one sees from the table, estimating the age o f the these attempts has proven futile. We have discussed
catalogue by proper star movements is a more than them in such great detail for two reasons - firstly, the
dubious activity which doesnt stand up to criticism. reader can get a better idea o f what the complexities
However, the speed vector compounds which are o f the self-sufficient dating o f the star catalogue re
compared in this table werent just given in different ally are - the dating that would be based on nothing
coordinate systems, but also in different measurement but the catalogues numerical material, that is. Sec
units! This is easy to observe from the above exam ondly, we wanted to provide some basis for raising the
ple - were dealing with the equatorial coordinate sys issues that we shall relate in more detail further on.
tem for the epoch o f 2000 a . d . in one case and the The main corollary that we come to at the pres
equatorial coordinate system for the epoch o f 1900 ent stage is as follows. The dating o f the Almagest re
a . d . in the other. These coordinate systems differ from quires a meticulous preliminary analysis o f the cata
each other. The above example demonstrates the scale logue. This analysis must relate to the following issues.
discrepancy. According to the Pythagorean theorem, 1. Identifying the Almagest stars as the ones ob
the given vector speed components o f Arcturus suf served on the contemporary celestial sphere. In sec
fice for the calculation o f said vectors length which tion 1 we demonstrate that this problem doesnt al
shall already be independent from the coordinate sys ways have an unambiguous solution; furthermore,
tem. However, in the first case it is ten times smaller the solution in question might depend on the alleged
than in the second, which stems from the fact that dif dating o f the catalogue. Therefore, before we can pro
ferent catalogues use different proper movement ceed with dating, we have to find and reject all cases
scales. In one case the measurement unit used equals of dubious identification o f the Almagest stars as their
l / 1000th o f a second per year, and in the other it is 1 modern counterparts.
second per century. The units differ by a factor o f ten. 2 . The nature o f possible errors contained in the
One needs no commentary here. It is obvious that Almagest catalogue. The error rates in stellar coordi
before suggesting that the reader should compare any nates characteristic for the Almagest lead one to the
values o f any kind, said values need to be given in the conclusion that the dating o f the catalogue cannot be
same scale. estimated with more precision on the historical in
We shall refrain from discussing the authors own terval as based on proper star movements. However,
attempts o f dating the Almagest ([263]), merely stat this statement becomes generally false if we manage
ing that we are o f the opinion that the dating o f the to discover the systematic compound in the errors o f
n 6 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

the Almagest star positions. In this case we may get As it has already been mentioned, [1339] contains
an opportunity to compensate it, thus raising the pre the identifications of the Almagest stars as their mod
cision o f the catalogue, which, in turn, may allow us ern counterparts. Nevertheless, we have conducted
to date the latter regardless o f the error in question. the identification process from scratch in order to se
3. The precision of the Almagest catalogue attainedlect the stars to be analyzed, see Chapter 4. The iden
with different stellar subsets. The goal o f this analy tifications contained in [1339] were thus confirmed
sis is the choice o f the star group from the Almagest for the most part.
whose coordinates must have been measured by Ptol However, we have discovered several modern stars
emy with some guaranteed precision level 8 . Once that can be identified as different Almagest stars for
we manage to locate such a group, it shall define the different epochs t. Such are o 2 Eri and JI Cas, for in
set of possible Almagest datings, namely, making fea stance. These stars were identified in [ 1339] under the
sible the datings that will allow the guaranteed pre assumption that Ptolemys observations were con
cision level 8 to be attained for the stars o f this group. ducted around the beginning o f the new era. Basing
If the resultant dating interval proves to be a great deal the dating o f the Almagest catalogue on the analysis
shorter than the a priori known historical interval, we of such stars makes no sense, for we shall simply end
shall obtain purposeful information about the date up with a vicious circle. All such stars were excluded
when the Almagest star catalogue was compiled. This from further consideration.
concept shall be used below (see Chapters 5-7). Let us also point out that the identifications and
Let us briefly discuss each o f the three issues as coordinates o f the stars o 2 Eri and JI Cas are consid
listed above. Their more detailed rendition can be ered doubtful.
found in the chapters to follow.
5.3. Various types of errors in the catalogue
5.2. The identification of the Almagest stars
We have demonstrated above that a simple com
There is a rather large amount o f handwritten parison o f the calculated stellar coordinates to those
copies as well as several mediaeval printed versions contained in the Almagest catalogue doesnt permit
of the Almagest where the ecliptic coordinates o f in to estimate the dating o f the latter. This is explained
dividual stars differ from one another. Most o f these by the huge discrepancy rates inherent in the Almagest
copies and editions (although not all) were brought catalogue for the most part. Therefore, we can only
to roughly 60 a . d . by precession. The implication is succeed if we analyze the Almagest errors o f differ
that if one were to compare the stellar longitudes ent nature meticulously.
from a given copy o f the Almagest with the precisely We shall divide the errors into three types: group
calculated stellar longitudes for 60 a . d . , the average errors, random errors and rejects.
discrepancy rate shall equal zero. Such a comparison Under group errors we shall understand various
is only possible due to the fact that identifying most data distortions resulting from observations or re
o f the Almagest stars with those on the modern ce calculations and leading to the shift o f a star group
lestial sphere leaves no room for doubt. on the celestial group as a whole.
The source text that we used was the Almagest Random errors are o f an individual character and
catalogue containing over a thousand stars in the owe their existence to imprecise observations rang
exact same form as it is given in the fundamental ing within the grade value o f the measurement in
work o f K. Peters and E. Knobel ([ 1339]). Several co strument for the most part. A distinctive trait of such
ordinate variants from [1339] were also included in errors is that they shift each star on the celestial sphere
the list o f stars under analysis. In the preliminary by a random value which has a zero average.
stage we neither doubted the veracity of stellar coor Rejects are a product of circumstances which were
dinates from the Almagest, nor the fact that they were either unforeseen by the compiler or unknown to
given in ecliptic coordinates rendered to 60 a . d . due him: copy errors, refraction etc. They also affect the
to precession. coordinates o f individual stars, and their values are

usually much greater than the measurement instru searchers - see [1339], [614] and [544], for instance.
ment scale precision. Rejects are a rather scarce type We shall merely mention possible reasons for the ex
o f error. istence o f such errors here.
The most important task is to define and com Error x might result from the fact that the observer
pensate the group errors. Suitable methods are dis or a later compiler o f the catalogue had for some rea
cussed in Chapter 5 where, apart from providing the son adjusted the catalogue to make it fit a dating
formulae necessary for their calculation, we also that would differ from that o f the real observation. It
demonstrate how to determine the precision o f the is possible that this operation used to serve some
resulting values. methodological end - for instance, making the cata
The estimation o f different types o f errors in the logue conform to some round or important date. It
Almagest stellar coordinates is dealt with in Chapter 6 . could also have been used for a deliberate distortion
We find out that the coordinates o f stars as given in o f the real observation date ([614]), or, alternatively,
the Almagest do indeed contain significant group er it may result from changes in the initial longitudinal
rors manifest as the shifts o f the respective stellar con reference point. We have already demonstrated that
figurations on the celestial sphere as a whole. ancient astronomers could count longitude from var
The values of group errors may in fact differ for ious points on the ecliptic. A change of the initial ref
various stellar groups - constellations, for instance, erence point would naturally lead to some constant
hence their name. However, we shall witness that in being added to all ecliptic longitudes and hence the
sofar as large enough celestial areas are concerned, alteration o f the catalogues dating, if it were to be
group errors o f the Almagest and other old star cat dated by longitudinal precession.
alogues coincide for various constellations and equal It is understandable that the latitude o f a star is in
the single error for the entire area. We shall refer to dependent from error x. This makes latitudinal co
such an error as the systematic error of a given cata ordinates more reliable, which is the very reason why
logue for a given celestial area. we shall be considering longitudes and latitudes sep
Each of the shifts defining a group error can be de arately. The consideration o f latitudinal discrepan
scribed by three parameters. We shall choose the fol cies requires just two parameters to define a group
lowing base errors as such, qv in fig. 1.1, Chapter 1. error - p and y, for instance.
Error x in the location o f the vernal equinox point What is there to say about the values o f p and y?
Q(tA) made by the observer in the observation year Equatorial latitudes o f stars are easy enough to de
tA in the ecliptic direction. In other words, x is the pro termine from actual observation with enough sim
jection o f the Almagest catalogue vernal equinox plicity and precision ([75]). Therefore, one should
point shift sideways from its real position over the expect error p to be small enough for the moment o f
ecliptic. observation, provided the observer was accurate
Error p in the location o f point Q(tA) in the di enough. Error y is o f a principally different charac
rection o f the meridian, or the projection o f the error ter. The determination o f the ecliptic position is
vector over the ecliptic meridian. achieved as a result o f rather complex observations
Error y in the angle e between the ecliptic and the and calculations, qv in Chapter 1 . Therefore, the value
equator. The change o f a stars ecliptic coordinates by o f error y might be significantly greater than that of
the ground observer needs to be preceded by the es error p.
timation o f the angle e between the ecliptic and the The works [544] and [1339] contain indications
equator, regardless o f the measurement method. If at the fact that the systematic error y is indeed in
the observer made the error y in the estimation of said herent in the Almagest. Moreover, some o f the Alma
angle, the ecliptic o f the catalogue shall be shifted in gests researchers estimated the value o f this error as
relation to the position o f the real ecliptic in the ob roughly 20'. Our calculations confirm this, qv in
servation year by the value o f y. Chapter 6 .
The possibility that group errors may be inherent We shall occasionally use parameters 9 and y in
in the Almagest has been discussed by many re stead o f p and y since they are more convenient from
ii8 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

Fig. 3.14. Specifying the parameters of the systematic error in the ecliptic coordinates of the stars with the aid of the parameters
y and (p or y and (3. In the present example x = 0.

the point o f view o f calculation. Their meaning is anisms o f error genesis. Calculations allow the dis
clarified in fig. 3.14. Inasmuch as the latitudinal dis covery of errors themselves but tell us nothing o f how
crepancies are concerned, the group error is rendered and why they were made - possible reasons may dif
to a mere misplacement o f the ecliptic plane, which fer from the abovementioned.
we shall be referring to as the catalogue ecliptic.
One can define the mutual disposition o f the cata 5.4. The discovery of the systematic error
logue ecliptic and real ecliptic plane for catalogue in the Almagest catalogue. Its compensation
compilation epoch tA if one is to fix angle cp between confirms the correctness of the declared
the equinox axis QR for epoch tA and the plane rota catalogue precision
tion axis CD, as well as fixing the plane angle y be
tween the two ecliptic planes - the true and the false. The real moment tA o f the catalogues compilation
We shall hereinafter define the parameters o f group remains unknown to us. Therefore we should calcu
errors with the values o f cp and y for the most part. late the values o f parameters y(t) and cp(t). The cal
Generally speaking, the compiler o f the catalogue culation method is a combination o f the minimal
may have made different group errors in his study o f square method and the spherical regression problem.
different celestial areas. Possible reasons include in Its precision properties are discussed in Chapter 5.
strument readjustment, the choice o f a different ob The results o f our calculations can be represented
servation point etc. as graphs ystat(t) and cpstat(t)y qv in fig. 3.15. These
In Chapter 2 we discover seven parts o f the Alma graphs were built after the processing o f the Almagest
gest star catalogues which are naturally distinctive as stellar coordinates for large celestial areas. The stat
seen on the celestial sphere, and differ by their relia index indicates that the corresponding values were
bility characteristics in the Almagest, see fig. 2.14. In educed by methods o f statistics. They are actually es
Chapter 6 we shall see that the same celestial areas in timates o f discrepancy parameters inherent in the
the Almagest also differ in group error values and positions o f the Almagest stars, and demonstrate said
precision characteristics. discrepancy to be uniform for several large areas of
To sum up, one can say that the reasons for the ex the celestial sphere. The estimations were made under
istence o f group errors and other discrepancies as the assumption that the catalogue was compiled in
listed above only serve to explain the possible mech epoch t, and are thus t functions. We shall be using

the term systematic errors for the error in question The next issue that we are confronted with is the
as well as its compounds, parameters y(t) and cp(t). nature o f the discovered parameters ystat and (pstflt. Is
W hat is the relation between these errors and it true that the calculated values o f ystat and cpsfflf are
group errors? If the large celestial area under study close enough to real group errors for the entire cat
consists o f several constellations, systematic errors alogue, or at least the stars from area A?
discovered with the aid o f statistical methods shall It is quite possible that the compiler o f the cata
represent averaged group error values for different logue made individual group errors for each con
constellations. It is only in case when all group errors stellation; in this case, the values that we have calcu
equal each other that they coincide with the respec lated shall de facto represent a sum o f various aver
tive systematic error. aged group errors, the result o f such averaging being
This is the only case where we shall not differen non-zero due to the relatively small number o f con
tiate between the definitions o f group error and stellations in general.
systematic error. In order to answer this question, we have consid
We have built confidence intervals Iy and 1^ o f ac ered all the Zodiacal constellations and the neigh
ceptable y and cp values around each value o f ystat(i) bourhoods o f most named stars. Calculations have
and cpstati*) Let us clarify that ystat and cpsfflf are but shown that the value of y ? fA as calculated for area
punctual statistical estimations o f unknown param
eters; the latter define the systematic error made by
the compiler of the catalogue, and the values of such
estimations are by no means equal to the values o f ac
tual unknown parameters. Once we build the confi
dence intervals around the calculated punctual esti
mates ystat and (pstaP we can claim the true parame
ter values to fall into these intervals with a given
degree o f certainty.
The method o f building confidence intervals,
which is widely used in statistical problems, is related
in Chapter 5. Actual results pertaining to the Almagest
are cited in Chapter 6 .
We have conducted an analysis o f errors for all
seven celestial areas o f the Almagest as discovered
above, having determined their respective systematic
error values as well as the values o f the remaining
square average latitudinal discrepancies resulting from
the compensation of the discovered conditional sys
tematic errors. What we discovered as a result was that
areas A and Zod A are the most precisely-measured of
all, qv in Chapter 6 and table 2.3. A propos, these are
the areas where most of the named Almagest stars are
located. Another discovery was that after the com
pensation of systematic error, more than half the stars
from area A ended up with the latitudinal discrepancy
o f 10 maximum (see Chapter 6 ). The percentage of
such well-measured stars is even greater for area Zod
A - 63.7%. Thus, the declared 10' precision rate of
the catalogue was confirmed for the latitudes o f the Fig. 3.15. The behaviour of parameters y stat fr) and cp stat fr)
majority o f stars from a rather large celestial area. in time.
120 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

Zod A applies to all the constellations from area A at those o f the stars which have names o f their own in
least. In other terms, y ft? A should be regarded as the his catalogue. There arent too many such stars - only
systematic compound that affects all the stars from twelve. They really comprise a very convenient basis
the well-measured celestial area A which also con in the visible part o f the sky. Their complete list is as
tains most of the named stars. However, we can make follows: Arcturus, Regulus, Spica, Previndemiatrix,
no such claim for the value o f c p I t is curious that Capella, Lyra = Vega, Procyon, Sirius, Antares, Aquila
this conclusion about the nature o f compounds ystat = Altair, Aselli and Canopus; twelve stars altogether.
and (pstai can serve as argumentation in favour o f the All o f these stars are bright and clearly visible
theory that the coordinate measurements for the against their background. What is especially impor
Almagest catalogue were conducted with the use of tant for the purposes o f dating, some o f them have a
the armillary sphere. See Chapter 6 for more details. rather high proper movement speed - for instance,
Arcturus, Procyon and Sirius. Some o f the others also
5.5. The compensation of the systematic shift across the celestial sphere rather visibly, namely,
error discovered in the catalogue gives us Regulus, Capella, Antares and Aquila = Altair.
an opportunity of dating the latter However, we had to exclude two o f the twelve stars
from consideration instantly - namely, Canopus and
The compensation o f the discovered systematic Previndemiatrix, the reason being that Ptolemys co
error allowed us to reduce the latitudinal discrepancy ordinates o f Canopus were greatly affected by refrac
for area Zod A of the Almagest from 17.7' to 12 .8'. This tion, and they can be regarded as a reject from the
resulted in the possibility o f dating the catalogue. statistical point o f view; as for Previndemiatrix, Ptol
We have already pointed out that the declared 10- emys initial coordinates o f this star were lost, and
minute precision rate o f the Almagest is indeed at simply remain unknown to us today, qv in Chapter 2 .
tained for most o f the stars in the catalogue. The Two more stars (Sirius and Aquila, or Altair) were
question that one comes up with here is whether there rejected due to the fact that the systematic error is
are any stars at all for which the declared Almagest different in their case, as our analysis shows, and the
precision rate will be guaranteed? value o f said error cannot be determined for these
It is known that the observer always uses the sys two stars. Therefore, the dating o f the Almagest cat
tem o f referential points, or stars, on the celestial alogue was made on the basis o f the remaining 8
sphere in stellar coordinate measurements, qv in [968], named stars. Their list is as follows:
for instance. This measurement method is natural and Arcturus, 16, a Boo, Baileys Almagest number 1 10;
has been used by all mediaeval astronomers. Tycho Regulus, 32, a Leo, number 469;
Brahe, for one, used 21 referential stars for his meas Spica, 67, a Vir, number 510;
urements ( [ 1049] ). The modern system o f referential Capella, 13, a Aur, number 222 ;
points consists o f several thousand stars which are Lyra = Vega, 3, a Lyr, number 149;
collected in the so-called fundamental catalogues (see Procyon, 10, a CMi, number 848;
catalogue FK4, for instance - [1144]). The Almagest Antares, 2 1 , a Sco, number 553;
contains indications that Regulus and Spica must be Aselli, 43, y Cnc, number 452.
among these referential stars. Special sections o f the
Almagest are dedicated to the measurement o f their 5.6. The dating of the Almagest catalogue
coordinates. by the motion of its eight primary basis stars
Let us formulate the following axiom. If the de after the rectification of the statistically
clared precision o f the catalogue is confirmed, it discovered catalogue error
should be guaranteed for the majority o f the refer
ential stars from the catalogue in question. The proposed hypothesis leads us to the implica
What are the stars that should have necessarily been tion that for the desired catalogue compilation epoch
included in the number o f the Almagests referential tAy all o f the eight named basis stars o f the Almagest
stars? First and foremost, Ptolemy must have used must have a maximal latitudinal discrepancy o f 10 '.

On the other hand, we know that the catalogues

systematic discrepancy compound y must fall into
the confidence interval Iy built around the statistical
estimation ystat(tA) for epoch tA. We thus come to a
natural dating method.
Let us consider the confidence interval Iy around
ystat(t) with the value of t and the level o f confidence
being fixed and select a certain subset St from values
that fall into it, which will compensate the given sys
tematic error compound y and make the latitudinal
discrepancies for all o f the eight named basis stars
less than 10', or the grade value o f the Almagest cat
alogue coordinate scale, with y in St, qv in fig. 3.16.
In general, set St can be empty. Let us find all the
Fig. 3.16. Dating the Almagest catalogue with the statistical
values o f the presumed datings t for which the sets
St are not empty. These very values shall comprise the
possible dating interval, since for all o f the presumed
datings t from this interval the latitudes o f all eight
named stars are measured with the precision rate
o f 10 .
We shall refer to the described dating procedure
as statistical, since it is based on the values of ystat(t)
discovered with statistical methods. A more explicit
description o f this procedure can be found below, in
Chapter 7, alongside a detailed discussion o f the
achieved dating results.
It turns out that the dating interval begins in 600
a . d . and ends in 1300 b . c . Although its length equals

700 years due to the low precision o f the Almagest,

this interval is located at a considerable distance from Fig. 3.17. Dating the Almagest catalogue with the geometrical
the Scaligerian dating o f the Almagests creation. method.

5.7. The dating of the Almagest catalogue

by the motions of its eight named basis stars interval can become if the confidence areas expand
by an independent geometrical method indefinitely.
We shall give a geometrical answer to this ques
The confidence intervals used for the statistical tion. Let us once again select a fixed time moment t
procedure contained a certain subjectively chosen pa as a candidate for the desired dating moment. After
rameter, namely, the level o f confidence, which rep that we shall define the set Dt o f such y values that a
resents the minute probability which we can disregard turn o f the real ecliptic by this angle for epoch t shall
in statistical corollaries. Therefore one can actually make the latitudinal discrepancy o f all the 8 named
discuss the issue o f the dating interval being de stars conform to the 10 -minute threshold with a cer
pendent on the chosen level o f confidence. Our corol tain value o f parameter (p, qv in fig. 3.17. It is obvi
lary that the group error for the 8 named stars equals ous that Dt contains subset St whatever the value of
the systematic error for area Zod A is also o f a statis t might be. Therefore, we shall discover all the possi
tical nature and may therefore prove incorrect. Hence ble values o f t for which the latitudes o f all 8 named
the question o f just how much greater the discovered stars shall not differ from the respective stellar lati-
122 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

tain stability unaffected by the variation of the initial

ClAVDII hypotheses, the declared precision o f the catalogue,

the reduction or expansion o f the dating contingent
P T O L E M A E I PE- o f the referential stars, and also the non-linear meas
lufienfis Alexandria omnia quae urement instrument distortions.
The viability of our method has also been tested
extant opera,praeter G eographiam ,q uam
nondifamiliformanuperrimeardidimus: fuminacu n S c diligcmiacaftigata on the star catalogues compiled artificially as a result
abErafmoOfiialdoScHitfchenfuchfio, &ahcodcmIlagoicainA! * o f modelling random errors in stellar coordinate ob
magdlumprxfarionr,&fideUfsimisin prioreslibros
annotationibuf illudrata.quemadmo* servations. The observation dates defined in mod
dum(t q u e n i paginatalo* elling concur with the results of dating by our method
go indio*.
in every case.
Apart from that, the dating method that we offer
was successfully tested on several well-known old cat
alogues. We have used it for dating the catalogues o f
Ulugbek, Al-Sufi, Tycho Brahe and Hevelius. In every
case the traditionally known datings o f the old star
catalogues under study were confirmed with our
methods, the Almagest catalogue being the sole ex
ception. This is apparently an indication that the tra
ditional dating o f Ptolemys lifetime contains a gi
B A S 1 L E
gantic error o f several centuries or even over a mil
lennium. See Chapter 9 for more details.
Anno, f$fl
Our main corollary is as follows. The star cata
logue o f the Almagest was created in the interval be
Fig. 3.18. The title page from a 1551 edition of the Almagest. tween 600 a . d . and 1300 a . d . The Scaligerian dating
The handwritten dating Anno 1551 is most noteworthy o f the Almagest catalogue (II century a . d . ) is ipso
indeed; the book is likely to have been dated retrospectively, facto proven gravely erroneous.
in the XVII-XVIII century.
We shall conclude this chapter with citing the front
page o f a 1551 edition o f the Almagest (see fig. 3.18).
tudes as given in the Almagest by more than 10' after It is most curious that the publication date is written
a certain rotation o f the ecliptic. by hand, in the exact same place o f the books front
A most important fact is that the resultant maxi page where one expects to find a printed date. It is
mal possible geometrical dating interval coincides possible that this date was inscribed on the book as
with the interval discovered by statistical methods. See late as the XVII or even the XVIII century, possibly
Chapter 7 for more details. with the goal o f making the book seem published in
Another fact that we shall demonstrate in Chapter the XVI century, its real publication date being much
7 is that the proposed dating method possesses a cer more recent.

Who is who?

1. twelve have proper names in the Almagest. They are

PRELIMINARY OBSERVATIONS simply described as star at the middle o f the neck,
star at the tip o f the tail, star at the end of the front
As we have seen, the dating of the Almagest by leg, the brighter of the two stars on the left knee
proper star movements might turn out erroneous if etc. Such descriptions are more often than not com
there was an error in identifying the fast stars used for pletely insufficient for a reliable identification of one
dating as their Almagest equivalents. The problem of Almagest star or another as its modern counterpart.
identifying the Almagest stars, or, more precisely, the Numerous researchers o f the Almagest have al
Ptolemaic descriptions of stars, as real, or modern ready performed an identification of the stars con
stars - the ones that we can observe today, that is, tained therein as the modern stars by comparing the
often turns out extremely complex. In some cases, Almagest star coordinates to those of the modern
there is no unambiguous solution at all. Obviously stars. The results of this identification can be found
enough, we havent been the first ones to address the in the work of K. Peters and E. Knobel, for instance
problem of identifying the stars in the Almagest cat ([ 1339]). They cite a table where each Almagest star
alogue. This problem has been known to researchers corresponds to a modern star. [1339] also contains the
for quite a while. However, it is of extraordinary im table of discrepancies between the identifications sug
portance to us, since no dating o f the Almagest star gested by different researchers. However, it has to be
catalogue by proper star motion rates is possible be emphasized that all prior identifications were made
fore the problem in question is solved. by astronomers who trusted the Scaligerian hypoth
Let us remind the reader that the Almagest cata esis, which notably affected the identification result
logue contains 1025 stars. However, only twelve of in many cases.
them have names of their own in the Almagest cata Indeed, if the position o f a dim and otherwise un
logue, which use the formula vocafwr (named). remarkable star with a high proper motion velocity
Those are Arcturus, Aquila (Altair), Antares, Previn- has altered notably over the period of time between
demiatrix, Acelli, Procyon, Regulus, Spica, Vega = the beginning of the new era and our days, it will
Lyra, Cappella, Canopus and Sirius (the latter is re identify as different Almagest stars in different epochs.
ferred to as The Hound). No other stars but these It is pointless to date the catalogue by such stars, since
12 4 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

the epoch o f the catalogues compilation will be cho contained in Table 4.1. This table contains the equa
sen depending on the chosen identification. Multiple torial coordinates o f stars for the epoch o f 1900 a . d .
possible identifications will lead us to multiple dat (for the time moment o f t = 0 in our system, and the
ings o f the catalogues compilation. proper motion components o f star velocities ren
Apart from that, in this situation it is altogether dered to the equator for the epoch o f 1900 a . d . The
impossible to be certain that the fast star in ques first column o f Table 4 .1 contains the index o f the star
tion is in fact included in the Almagest. Most o f the according to Bayer and Flamsteed. Some o f the data
stars are dim and their order of magnitude is between contained in Table 4.1 were taken from the previous
4 and 6 . Many of these dim stars werent included in edition o f the catalogue ([1197]). The discrepancies
the Almagest catalogue for the simple reason that there between the numeric values contained in both edi
are more such stars than the catalogue contains, and tions are minute and negligible in our case.
so there are cases when a single Almagest star can be According to the data contained in this table, the
identified as several stars visible with the naked eye. formulae o f transforming the equatorial coordinates
All such cases need to be taken into account so as not into their ecliptic equivalents with proper star motion
to base the dating method on ambiguous scenarios. velocities taken into consideration (see Chapter 1 )
However, in general, we did not doubt the fact were used in order to determine the ecliptic coordi
that the star identifications o f Peters and Knobel were nates Li(t) and B ^t) o f star i on the celestial sphere
made diligently and in good faith ( [ 1339] ). Our com (1 < i < 78) for epoch t.
putations have proved this viewpoint correct. Possible We built an estimated 8-area for each o f the above
errors result from nothing but the implied incorrect 78 fast stars - in other words, a circle whose radius
dating of the Almagest star catalogue - the Scaligerian equals around the calculated position o f the star on
early a . d . dating. In order to rule out the effects o f the the celestial sphere for each assumed dating t between
Scaligerian dating, we have performed the Almagest 1100 a . d . and 1900 a . d . (0 < t < 30), see fig. 4.1. After
identification o f fast stars anew. that, we calculated the arc distance (A, i, t) between
star A from the Almagest catalogue and the estimated
2. position o f fast modern star i, with estimated coor
FORMAL SEARCH OF THE FASTEST STARS dinates equalling (Lf(f), B ^t)) in epoch t for each of
IN THE ALMAGEST CATALOGUE the assumed dates (t).
If (A, i, t) < e, modern star i is likely to identify
2.1. The star identification method as star A from the Almagest catalogue in the moment
o f t. Otherwise, no such identification is likely. Thus,
We are only concerned with the issue o f identify the identification (or capture) only took place when
ing the notably mobile stars in the Almagest cata area around the star i from the modern catalogue
logue, which may be o f use for dating purposes. The
faster the star, the more precisely we can date the cat
alogue by its position - but only given that the star
in question is reliably and unambiguously identified
in the catalogue that we attempt to date. In the first
stage we have chosen but 78 o f the fastest stars from
the bright star catalogue ( [ 1197] ) in order to identify
them formally as Almagest stars. Double stars are
counted as a single star. The stars that we have cho
sen have a minimal proper movement velocity o f 0.5"
per year by at least one of the coordinates in the equa
torial system o f the epoch o f 1900 a . d . It has to be
said that the majority o f these stars are rather dim. Fig. 4.1. The circular area around a modern star that moves
A list o f the fastest stars visible to the naked eye is across the celestial sphere together with the star.

Table 4.1 A list o f the fastest stars in the catalogue ([1197]). We have chosen all the stars whose speed equals
0.5 sec/year minimum by at least one o f the equatorial coordinates ( a and 8) fo r the epoch o f 1900.

Modern name Star V V5 Magnitude

o f the star number a !900 $1900 measurement measurement o f the star
(where in the catalogue unit unit in the catalogue
applicable) [1197] m 0 t 0.001 "/year 0.0017year [1197]
h s

6 00 01 08 49 38 560 -37 5,77

11 P Cas 21 00 03 50 58 36 527 -178 2,42
77 00 14 52 65 28 1708 1163 4,34
98 00 20 30 77 49 2223 326 2,90
159 00 32 12 25 19 1383 -8 5,71
173 00 35 31 24 21 640 -329 6,24
176 00 35 44 60 01 886 451 5,79
24 T| Cas 219 00 43 03 57 17 1101 -523 3,64
222 00 43 08 4 46 752 -1142 5,82
JJ. Cas 321 01 01 37 54 26 3430 -1575 5,26
52 T Cet 509 01 39 25 16 28 -1718 860 3,65
637 02 06 19 51 19 2108 651 6,28
660 02 10 57 33 46 1155 -240 5,07
753 02 30 36 6 25 1807 1459 5,92
1 8 1 Per 937 03 01 51 49 14 1267 -81 4,17
1006 03 15 36 62 57 1332 659 5,48
1008 03 15 56 43 27 3056 744 4,30
1010 03 16 02 62 53 1328 655 5,16
23 8 Eri 1136 03 38 27 10 06 -92 744 3,72
40 o2 Eri 1325 04 10 40 -7 49 -2225 -3418 4,48
1614 04 55 51 -5 52 557 -1089 6,50
15 A. Aur 1729 05 12 06 40 01 528 -659 4,85
2083 05 51 44 50 24 74 568 5,00
2102 05 53 20 63 07 135 540 4,53
9 a CMa 2491 06 40 45 16 35 -545 -1211 1,60
10 a CMi 2943 07 34 04 5 29 -706 -1030 0,48
78 P Gem 2990 07 39 12 28 16 -623 -52 1,21
2998 07 39 51 44 55 -72 -563 5,22
3018 07 41 51 39 59 -293 1663 5,39
3384 08 28 57 31 11 -1119 757 6,36
3951 09 55 15 32 25 -522 -436 5,60
4098 10 21 54 49 19 81 -892 6,50
53 % UMa 4375 11 12 51 32 06 -431 -593 4,41
83 Leo 4414 11 21 42 3 33 -723 177 6,50
4486 11 33 29 45 40 -594 18 6,39
4523 11 41 45 39 57 -1538 393 5,04
126 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

M odern nam e Star VB ^8 M agnitude

o f the star num ber a !900 $1900 measurement m easurement o f the star
(where in the catalogue unit unit in the catalogue
applicable) [11 9 7 ] h m s O r
0.001 "/year 0.001 "/year [1197]

4540 11 45 29 2 20 742 -277 3,80

5 P Vir 4550 11 47 13 38 26 3994 -5800 6,46
4657 12 10 02 -9 44 31 -1024 6,12
4710 12 17 51 67 05 -748 243 6,38
43 p Com 4983 13 07 12 28 23 -799 +876 4,32
5019 13 13 10 17 45 -1075 -1076 4,80
5072 13 23 32 14 19 -237 -583 5,16
5183 13 42 00 6 51 -513 -114 6,32
5189 13 43 10 35 12 -522 -178 6,47
5209 13 45 50 23 53 -575 -310 6,48
5568 14 51 37 20 58 1041 -1745 5,76
v2 Lup 5699 15 15 03 47 57 -1621 -275 5,71
41 ySer 5933 15 21 50 15 59 307 -1292 3,86
15 p CorB 5968 15 57 13 33 36 -200 -774 5,43
6014 16 04 16 6 40 235 -744 6,02
6060 16 10 11 -8 06 227 -508 5,56
26 e Sco 6241 16 43 41 34 07 -613 -256 2,36
36 Oph 6401/2 17 09 12 26 27 -464 -1146 5,33; 5,29
6416 17 11 28 46 32 975 213 5,58
6426 17 12 09 34 53 1167 -176 5,89
6458 17 16 55 32 36 126 -1047 5,36
6518 17 25 18 67 23 -529 0 6,31
6573 17 33 57 61 57 253 -513 5,31
46 jllHere 6623 17 42 33 27 47 -313 -748 3,48
6752 18 00 24 2 31 256 -1097 4,07
58 T| Ser 6869 18 16 08 -2 52 -554 -697 3,26
44 %Dra 6927 18 22 52 72 41 521 -356 3,57
7373 19 20 12 11 44 722 640 5,16
7644 19 55 32 67 35 845 -680 6,07
7703 20 04 38 36 21 449 -1568 5,32
7722 20 09 03 27 20 1244 -178 5,73
7875 20 31 46 50 53 309 -569 5,12
3 T| Cep 7957 20 43 15 61 27 91 822 3,43
61 Cyg 8085/6 21 02 25 38 15 4135 3250 5,21; 6,03
8148 21 13 59 26 46 -539 -352 6,56
8387 21 55 43 57 12 3940 -2555 4,59
8697 22 47 20 9 18 522 49 5,16
8832 23 08 28 56 37 2073 299 5,56

contained star A from the Almagest catalogue on Each of the table's rows corresponds to a pair of
some a priori dating interval [t* , i*] (a fragment of identified stars - the fast modern star whose num
the historical interval 0 < i < 30). Obviously, differ ber is taken from the catalogue ([1197]), and the Al
ent Almagest catalogue stars could wind up in the magest star which we shall mark as A. If the fast
same area 8 o f the modern star z, simultaneously as modern star isnt identified as the Almagest star A
well as with different t values. In some cases, the re whatever the value of 8 - that is to say, if the Almagest
gion around a fast star didn't contain any Almagest star A isnt captured by the 8 circumference of the fast
stars, regardless o f the rvalue under consideration. modern star in question, we put a dash into the re
The above identification method is, of course, rather spective position in the table. For instance, star 1325
rough. In particular, it makes sense to choose values from [1197] cannot be identified as Bailey's star #780
of the capture radius that happen to be several times from the Almagest anywhere on the historical inter
greater than the error margin value o f the catalogue val 0 < t < 30 with 8 = 0.5.
under study. It turned out that the actual identifica If a star numbered z is just identified with a single
tion hardly depended on the radius values (e) at all, star A from the Almagest catalogue, what we indicate
owing to the fact that the stars of the Almagest are dis in the respective row is Bailey's number of star A, as
tributed across the celestial sphere rather sparsely. well as the time intervals for which the identification
takes place with different values of 8. Star whose i
2.2. The result of identifying the "modern" value equals 21 (11 P Cas, that is) can thus be iden
stars as their counterparts from the Almagest tified as the star A = 189 with 20 < t < 30, if 8 = 0.5
catalogue and on the entire interval o f 0 < t < 30 if 8 > I o.
Should star i have several identification options,
When we were giving a general description of the all o f them are indicated in the corresponding row,
Almagest catalogue, we already mentioned that the and the time interval that we regard is the one for
catalogue precision level as declared by the compiler which the Almagest catalogue star under study is
equals 10 ' (with latitude and longitude considered closer to star z than other stars that it may be identi
individually). Hence, the arc distance measurement fied as. The star with i = 1325, for instance, or 40 o 2
precision as declared in the Almagest roughly equals Eri, can be identified as different Almagest stars on
14', which is V~2~times lower than the individual different time intervals (numbers 778, 779 and 780
measurement precision for each coordinate. However, in Bailey's numeration). The column that corresponds
this declared precision happens to represent a record to the value 8 = 1.5 tells us that while 0 < t < 10, star
value o f sorts, that is, such precision can only be at i = 1325 is the closest to Almagest star A = 780 (in
tained for well-measured stars - such as the named Bailey's numeration). Nevertheless, let us note that if
basis ones. Real precision might well prove to be sev t = 10, the distance between the stars i = 1325 and A =
eral times lower. 779 is also less than 1.5.
We shall consider the precision issues in more de The reason for identifying the modern stat zas the
tail below (Chapters 5 and 6 ). For the meantime, we Almagest star A for the moment t is as follows. If one
can safely leave the topic alone and choose such a is to assume that the Almagest catalogue was compiled
value for the capture radius 8 that will be several times in year tythe most fitting candidate for playing the
greater than 14'. This is exactly what was done, part of A-numbered star from the catalogue is the z-
namely, we chose the values for 8 to equal (V2)0, 1 , numbered star from the modern catalogue ([1197]).
(IV2)0, 2. Table 4.2 contains the fast star identifica Table 4.2 demonstrates that the choice o f the 8
tion results for the abovementioned time interval of value hardly affects the identification results at all.
0 < t < 30 - between 1100 b .c . and 1900 a .d ., that is. This choice is arbitrary in many respects, and is only
The only fast stars that we find in this table are the dictated by the following informal considerations.
ones whose environs capture at least one star from Firstly, the radius of 8 must be comparable to the ac
the Almagest catalogue with a minimum o f one t for tual catalogue precision level. Secondly, it has to be
the indicated values o f 8. sufficiently big for the identified pair list to contain
128 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

Table 4.2. Time intervals o f possible identifications o f the fastest stars as their Almagest counterparts fo r varying
inclusion range values o fe . Alleged dating param eter t has values that fluctuate between 0 and 30, which correspond
to the changing alleged Almagest catalogue creation dating interval beginning with 1900 a . d . and stretching back
wards in time with a step o f 100 years. The value o f t = 0 corresponds to 1900 a . d . ; t = 30 corresponds to 1100 b . c .

Time intervals o f fast star identification fo r varying inclusion range values

N um ber o f the star N um ber o f the star ofe.. We indicate intervals applicable to the alleged dating param eter t,
in the star in the Almagest which fluctuates between 0 and 30
catalogue [1 1 9 7 ] star catalogue
= 0.5 = 1.0 = 1.5 = 2.0
21 189 [20.30] [0.30] [0.30] [0.30]
219 180 - [0.30] [0.30] [0.30]
321 185 - [6.27] [0.30] [0.30]
509 723 [4.30] [0.30] [0.30] [0.30]
660 360 [8.30] [8.30] [8.30] [8.30]
-II- 361 [0.7] [0.7] [0.7] [0.7]
753 716 - [10.30] [2.30] [0.30]
937 196 [27.30] [0.30] [0.30] [0.30]
1136 783 [0.13] [0.30] [0.30] [0.30]
1325 778 [29.30] [29.30] [29.30] [29.30]
-II- 779 [19.25] [14.28] [12.28] [12.28]
-II- 780 - [0.8] [0. 11 ] [0. 11 ]
1614 775 - - [0.30] [0.30]
1943 848 [0.17] [0.30] [0.30] [0.30]
2491 818 [8.30] [0.30] [0.30] [0.30]
2990 425 [0.30] [0.30] [0.30] [0.30]
2998 882 [0.30] [0.30] [0.30] [0.30]
4375 32 [0.3] [0.30] [0.30] [0.30]
4414 486 [0.30] [0.30] [0.30] [0.30]
4540 501 - [14.30] [0.30] [0.30]
4657 732 - - [0.30] [0.30]
5019 527 [8.30] [0.30] [0.30] [0.30]
5188 935 - - [0.30] [0.30]
5288 940 - [0.21 ] [0.30] [0.30]
5340 110 [5.13] [0.25] [0.30] [0.30]
5460 969 - - - [0.30]
5699 979 [0.25] [0.30] [0.30] [0.30]
5933 265 - [8.30] [0.30] [0.30]
6241 557 [0.30] [0.30] [0.30] [0.30]
6401 247 [17.30] [0.30] [0.30] [0.30]
6623 125 - [0.30] [0.30] [0.30]
6752 261 - [4.30] [0.30] [0.30]
6869 279 [0.28] [0.30] [0.30] [0.30]
7957 79 - [0.22] [0.30] [0.30]
8085 169 - - [22.30] [20.30]
8697 327 - - [0.7] [0.7]
- // - 328 [28.30] [8.30] [8.30] [8.30]

something in the first place; end result should not be measurement precision, the intersection point being
affected by the possible aberrations contained in the the approximate date o f the catalogues creation.
catalogue. Thirdly, the value o f should not be ex However, table 4.2 demonstrates that the value o f
cessive to keep the identification result definite. that we get in such a manner is too great. It will take
In particular, table 4.2 shows us that 36 out o f the several millennia for this distance to be covered -
78 stars under study could be identified. These iden even by the fastest o f stars. However, in this case the
tifications do not contradict the ones indicated in date in question will be determined very unreliably,
[1339]. Moreover, the overwhelming majority of them with a possible millenarian aberration. In particular,
coincides with the previously known identifications. a dating like this shall be largely dependent on the stel
The visible exception is the star whose i number equals lar contingent under study. Adding or subtracting a
1325, or o 2 Eri. The work of Peters and Knobel points single star, for instance, can significantly affect the
out the dubiety o f this stars identification. Our re dating. This is exactly why we describe the stage o f
search demonstrates that it can be identified as dif classifying stars by the precision o f their measure
ferent stars of the Almagest on different time intervals. ment separately in Chapter 3 for - it is a necessary
Bearing in mind its rather low luminosity, the identi procedure required for a reliable dating.
fication o f the Almagest stars A = 778,779 and 780 as
real celestial objects is highly dubious. Therefore we 2.3. Corollaries
have to exclude these three stars from further consid
eration, which we have already done. Corollary 1. Most of the stars in the Almagest
Table 4.2 contains an example o f the opposite as catalogue were identified correctly by the researchers
well. For instance, the Almagest catalogue star A = 169 that preceded us.
in Baileys numeration became identified as two mod Corollary 2. Out o f the 78 fastest stars borrowed
ern stars simultaneously (#8085 and #8086 in the from a modern bright star catalogue ([1197]) and
modern catalogue - [1197]). visible to the naked eye, 36 stars can be reliably iden
The results presented in table 4.2 tell us that new tified as Almagest stars (see table 4.2).
identifications o f stars are an exception and not the Corollary 3. Only the following fast stars from
rule. This is explained by the low mobility of the over table 4.2 are identified ambiguously with = 1.5.
whelming majority o f the stars as well as the fact that a) Star o 2 from the constellation o f Eridanus =
the stars from the Almagest catalogue are at a signif 40 o 2 Eri, numbered 1325 in [ 1197] can be identified
icant distance from each other on the celestial sphere. as the following Almagest stars (in Baileys numera
The stars that we shall base our research upon were tion, for different alleged epochs).
not re-identified; we shall therefore use their corre Almagest star 778 for the interval o f 1100 b .c. -
sponding numbers in Baileys numeration without 800 b .c.;
quoting the numbers o f [1197]. The star will be Almagest star 779 for the interval o f 700 b .c. -
named should such a necessity arise. 800 a.d .;
The table that we cite might lead one to the ques Almagest star 780 for the interval between 900
tion o f whether one can use the resultant time inter a.d . and the present epoch.
vals for fast star identification in the Almagest in order b) Star 660 from [1197] can be identified as the
to date the latter. It appears that no reliable dating can following Almagest stars:
be calculated in this manner. The reasons are dis Almagest star 360 for the interval o f 1800-1900
cussed above in great detail (see Chapter 3). a.d .;
We feel we should sum up with the general ob Almagest star 361 before 1800 a.d .
servation that if one were to exclude the ambiguously c) Star 8697 from [1197] can be identified as two
identified stars from the list and make equal some Almagest stars in different epochs:
minimal value which would make all the identifica Almagest star 327 for the interval o f 1200 a.d . -
tion intervals intersect with each other, this value 1900 a.d .;
could serve the ends o f evaluating the real fast star Almagest star 328 before 1200 a.d .
130 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

3. mind the reader that the coordinate system choice of

THE SEARCH OF ALL THE FAST STARS one epoch or another by no means implies that the
RELIABLY IDENTIFIABLE IN THE ALMAGEST star positions were calculated for the same epoch.
CATALOGUE These phenomena are not related in any way at all.

In the previous section we were looking for pos Step 2. Selecting the stars that have either
sible identifications o f fast stars seen with the naked Bayers or Flamsteed s indications.
eye as the Almagest stars. This would allow us to in Further one, we have picked out just those stars
stantly reject the stars which are a priori useless for a whose indication either included a Bayers letter or
proper movement dating o f the Almagest due to the a Flamsteeds number, or both. We have already
fact that the possible identification o f these stars as mentioned our motivation for doing this above. The
their counterparts from the Almagest is largely de reason is that the systems o f Bayer and Flamsteed are
pendent on the alleged dating. the XVII-XVIII century heirs o f Ptolemys stellar po
Let us now ask an altogether different question - sition description method which would describe the
which ones o f the relatively fast modern stars can be stars relative position in a given constellation ver
identified in the Almagest catalogue with absolute bally. It would be natural to assume that when these
precision? The search o f these stars is the necessary astronomers introduced a new system o f indicating
preliminary work that has to be done before we can stellar positions, they studied the Almagest very
date the catalogue by proper star movements. This pedantically, ascribing their new indication to a star
formulation o f the problem differs from the one of whose identification would leave no place for doubt.
fered in the previous section. Before we have used a Had we kept back the stars which neither have Bayers
rough formal method for the rejection o f the stars letter nor Flamsteeds number in their name, it would
which obviously cannot be identified as the Almagest mean that were keeping back the stars that Bayer and
stars reliably. As a result, many o f the poorly quali Flamsteed were doubtful about. And what we seek to
fied stars were not excluded from our research. How evade first and foremost is the effect of the suspicious
ever, we shall be needing a meticulously verified list stars that can lead us to erroneous datings based
o f fast stars which can be reliably identified in the upon false identifications.
Almagest. This task requires some additional work Why have we chosen Bayer and Flamsteed in par
from our part, and well get right to it. ticular - from the great multitude o f later as
In order to solve the problem, we have taken the tronomers o f the XV II-XX century who studied the
modern electronic version o f the catalogue BS5 which Almagest? This was primarily caused by the fact that
contains all the stars visible to a naked eye - about they were the ones to introduce the new indications
nine thousand of them altogether. Catalogue BS5 is a o f stars which reflected the old tradition that they
more precise version o f the bright star catalogue BS4 were based upon. The generations o f astronomers
([1197]). We have checked the electronic version of that followed them were already using the new indi
BS5 for misprints having compared it to the printed cation for their studies, and the old tradition had
edition o f BS4 ([1197]). All the misprints were cor soon been forgotten as obsolete. Metaphorically
rected. speaking, the astronomy teacher of Bayer could point
out the stars on the sky (and then the respective places
Step i . Selecting the stars for speed . in the Almagest describing said stars) with his finger,
We have picked out all the stars from the cata quoting their names as given by Ptolemy - the star
logue BS5 whose annual proper movement speed on Virgos shoulder, the star on the hoof o f Pegasus
equals 0.1 sec (by one o f the coordinates in the equa etc. The following generations of young astronomers
torial system for the epoch of 1900). These speeds were would already learn the names o f these stars as the
taken from the printed catalogue BS4 ([1197]), since Delta o f Virgo, the Epsilon o f Pegasus and so on.
in the catalogue BS5 the speeds are given in equato The Almagest catalogue terminology became com
rial coordinates for the epoch o f 1900 a . d . Let us re pletely obsolete.

Step 3. T he selection of stars which have worst-measured in the Almagest. The implication is
OLD NAMES OF THEIR OWN. that even if the position o f a star is measured well
The catalogue BS4 ([ 1197]) contains the complete enough but falls into one of these areas, the error in
list o f Star names found in old and more recent texts its coordinates can substantially affect the proper
on pages 461-468. The texts in question date back to movement dating, making it extremely imprecise.
the antiquity and the Middle Ages. We cite this en Having performed the selection described above,
tire list in tables P1.2(a) and P1.2(b) in Annex 1 . We we ended up with a total o f 76 stars.
have picked out those o f the stars we ended up with
in the previous stage which can be found in this list Step 5. Selecting the stars by the local star
o f old stars possessing names o f their own. CHART IMAGE.
The reasons for such a selection are as follows. We In the final stage we have chosen only those stars
want to exclude all possible errors in our identifica which can be unequivocally located on the sky by
tion o f the stars which shall be used for the dating o f Ptolemys coordinates, even if one is to allow for the
the Almagest. It is obvious that if a star has a medi gigantic errors o f 2-3 degrees. We have meticulously
aeval name of its own, it makes its identification more verified the correctness o f luminosity as stated in the
reliable. Named stars have clearly been o f special in Almagest, as well as the veracity o f Ptolemys de
terest for the old astronomers, hence the very fact of scription. If any discrepancies were found, the star
their having names. Since old astronomy was based would be rejected at once.
on the Almagest to a great extent, one is to expect that As a result, the only stars that we decided to keep
these stars could be identified in the Almagest more in our list were the ones which can be isolated among
reliably than others. the stars o f comparable luminosity and also corre
spond to the coordinates o f a single star in the Alma
Step 4. T he selection of stars that fall into gest that cannot be identified as any other star even
THE WELL-MEASURED CELESTIAL AREAS OF THE if we are to allow for an aberration o f several degrees.
Almagest. We have used the star atlas ( [293] ), as well as the sim
We proceeded to exclude the stars which wound up ple and convenient software package called Turbo-Sky
in celestial areas C and D o f the Almagest catalogue. which can display a detailed map o f any given celes
We shall explain the reason for that in Chapter 6 . These tial area accounting for stellar luminosity. This pro
are the areas for which we can neither calculate nor gram also includes a telescope feature giving a 25x
compensate the systematic error o f the Almagest com zoom.
piler. Apart from that, our analysis of Ptolemys meas During this last selection stage 8 stars o f 76 were
urement precision for different areas o f the sky (see rejected, which leaves us with 68 stars. The rejected
Chapter 2 ) demonstrates areas C and D to be the 8 stars are listed in table 4.3.

Table 4.3. Eight stars rejected in the fin al stage o f filtration o f the 76-star list.

1 2 3 4 5 6 7 8
BS5 N am e ? M BS5 vai900 Vl900 Bailey's num ber m a

921 25p Per 3.39 +0.130 - 0.102 204 4

2484 31 Gem 3.36 -0.115 -0.194 441 4
4057 4 1 Leo 2.61 +0.307 -0.151 467 2
6913 22X Sgr 2.81 -0.043 -0.185 573 3
8610 6 3 k Aqr 5.03 -0.070 -0.114 651 4
321 30[X Cas D 5.17 +3.423 -1.575 185 4
343 336 Cas D 4.33 +0.229 -0.017 185? 5
7348 a Sgr D 3.97 +0.030 - 0.121 593 2-3
132 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

The first column of table 4.3 contains the stars fiable stars according to [1478]. The stars we are re
number according to the bright star catalogue BS5. ferring to are marked D (for dubiously-identifiable).
The second column contains the name o f the star. In All three stars were discarded in the final filtering
the third column we find the letter D which stands o f our list.
for disagreement (referring to different researcher To size up, we could say that we got a list o f stars
versions) which we borrowed from the electronic ver which can be identified as their Almagest counterparts
sion o f the Almagest. The corresponding explanatory reliably and whose proper movement is visible from
materials tell us that the discrepancies between the celestial areas A, Zod A, , Zod B and M. The list con
opinions o f various astronomers are quoted accord tains a total o f 68 stars; it can be seen in table 4.4
ing to [1478]. The book also accounts for the dis from Annex 1 at the end o f the book.
crepancies pointed out by Peters and Knobel ([ 1339]). Let us emphasize that the resultant list contains the
The fourth column contains Baileys numeration, or complete kernel o f the eight named Almagest stars
the Almagest number given to the suggested doppel- which we already mentioned above. These eight stars
ganger o f the star in question. The eighth column are collected in the very beginning o f the list and
contains the luminosity value according to Ptolemy. marked with block letters. This is the primary list we
We must emphasize that the previous list o f shall use in our final dating o f the Almagest catalogue
76 stars contained a total o f three dubiously-identi- by proper star movements.

The analysis of the star catalogues'

systematic errors


0.1. A demonstrative analogy

The necessity o f analyzing the errors contained in

star catalogues was already explained above. First and
foremost, we are referring to the Almagest; however,
the method in question shall also be applied to other
catalogues - real ones as well as artificially generated
ones. In the present chapter we shall demonstrate
how to discover and compensate the systematic error.
The idea behind the method is simple and quite nat Fig. 5.0. A target with traces o f bullet shots.
ural. Moreover, it has been used in mathematical sta
tistics for quite a while now. In order to explain the
basic concept, let us consider the following example. rifle - however, we can estimate the displacement
Let us assume that we are regarding the results o f a value. A sensible way o f doing this would require us
shooting competition as shown on the picture. to determine the geometrical centre o f all the results
The dots represent bullet holes. How great is the and draw a vector from the bulls eye to the calculated
hit accuracy? The answer is obvious - not that great centre (vector S on the scheme). How do we formally
at all. However, we can see that the actual grouping calculate vector S? The procedure is a simple one. We
o f shots is good enough. This leads us to the as have to take vectors xt which correspond to the re
sumption that the rifleman is in fact a good one; as sult o f the shooting and to average them by the total
for the fact, that the bullets hit a spot which lies side amount o f shots N:
ways from the bulls eye, it can be explained by a de N
fect in his rifle-sight. Obviously, we can say nothing
about the nature o f said defect without seeing the
134 I h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

We must also point out that vector S can be cal

culated alternatively from the problem o f square
average discrepancy minimization - we have to find iV i=l
vector S which provides for the minimum o f the with the systematic latitudinal error equalling
2 > , - s ) 2. /=1
i=l These errors, as we already mentioned, may result
Here we estimate that (x{ - S)2 = (xn - SY)2 + (xi2 from the incorrect estimation o f the ecliptic plane as
- S2)2j where xfl, xi2 and Sv S2 are the respective co well as a number o f other reasons which remain un
ordinates o f vectors x{ and S. known to us. We shall not be able to say anything in
The accuracy o f the actual rifleman can then be re the exact nature o f these circumstances - however,
characterized by the result scatter range around the we shall put forth a number o f hypotheses in this re
discovered centre; this accuracy is thus a lot greater spect. All o f this notwithstanding, we can, and will,
than the accuracy o f hitting the bulls eye. The cal compensate the error that they caused. It requires
culation o f vector S represents the actual systematic nothing but the alteration of the catalogue coordinate
error compensation procedure for this example system similarly to how it was done in the rifle ex
(whose value equals S, respectively). ample - one that would make the resultant average
Formally, if we are to use a different coordinate sys longitudinal and latitudinal errors equal zero.
tem moving its initial point sideways from the bulls
eye by vector S, the shooting results as given in the 0.2. The implementation of the method
new coordinate system shall only contain random
compounds (resulting from shaking hands etc), with In this section we shall demonstrate the practical
no regular compound. application o f the general concept related above.
Let us now return to the star catalogue and as First o f all, let us emphasize that we shall only
sume that we need to check whether there may be a compensate the latitudinal error. The reasons were all
systematic error in some part o f the catalogue and to named above - basically, it allows to minimize the
determine its value should such an error indeed exist. error in calculations, which is vital, considering the
Let us assume that we arent confronted by the prob low precision o f the old catalogues.
lem o f dating so far - that is to say, we know the date Thus, what we have at our disposal is the cata
when catalogue tA was compiled for certain (A is for logue from which we have selected a large group o f
Almagest, o f course - still, all the above considerations stars whose total number equals N, with the coordi
are valid for other catalogues as well). We would then nates (Z bi)f=1. Their doubles from the modern cat
have to compare the real coordinates o f the stars for alogue are already known to us from the previously
the moment tA (known from precise modern cata conducted identification procedure. Let us use the
logues) to the coordinate values taken from the cat indications (Lf(t), Bi(t))f=J for referring to the coor
alogue under study which pertain to the part thereof dinates o f said doubles calculated for moment t. Let
that is used in our research. This comparison requires us now assume that we want to examine the possible
the calculation o f the average discrepancy rate for the systematic error value under the assumption that the
coordinates under comparison, just like we did in the catalogue compilation date is tA.
example with rifle shot accuracy. Let us define
Let the total o f stars from the chosen area equal
N. We shall use the indications Z, and I , for the actual If = 1 ,.^ ) ,^ = ^ )
ecliptic longitude o f star i in the catalogue under
study and its exact longitudinal value, respectively. and introduce the latitudinal discrepancy
In this case, the average (systematic) longitude error
shall equal ABf = - Z?,.

Our goal is to minimize the value of (or the trusted interval for the real value o f parame
ter y), and
C2= X(A5,^)2-----
/=1 i<p (e) = [(pSMt- y e > ( p stat + y e ]
by changing the coordinate system, or simply draw
ing a new coordinate grid that differs from the one which is the trusted interval for the real value o f the
used in the catalogue. parameter <p. It can be demonstrated (qv below) that
The change o f the coordinate grid can be para- the values o f x and y can be calculated by the for
meterised by two values if we are to consider the mulae
problem o f minimizing the expression mentioned *2e> y . ~ *2e>

above: y and <p. They can be seen in fig. 5 .1 below. Let

us explain what they stand for. Here y is the angle be where q represents (1 - ) - the fractile o f the
tween the real ecliptic and the ecliptic o f the cata
logue, whereas <p represents the angle between the standard normal distribution as calculated from the
equinox line and the line o f intersection between the tables.
real ecliptic and the catalogue ecliptic. Thus, if we are to define a certain confidence level
Thus, having solved the problem o f minimizing 1 -e, we can guarantee that the real value of y falls into
the abovementioned expression, we can calculate the the interval Iy(e), and the value o f (p falls into the in
values o f ystat and (pstat which can parameterise the terval /^(e) with a probability o f no less than 1 -e.
coordinate system alteration and give us the initial
minimum. Their explicit form can be seen below, in 0.3. The value of the systematic error cannot
formulae 5.5.2 and 5.5.3. be used for the dating of the catalogue
The value o f Gmin is a residual square average lat
itudinal error that we end up after the compensation Let us now provide a somewhat different inter
o f the systematic error. The explicit form of the resid pretation of the calculated values o f ystat and <p5tof. The
ual dispersion formula o min can be seen below, after use of stellar coordinates (it suffices to consider noth
formula 5.5.10. It results from using ystat and <p5taf as ing but the latitudes, as a matter of fact) permits an
the parameters for the square average aberration ex easy calculation o f the ecliptic poles PA (for the cat
pression. The derivation o f these formulae can be alogue under study) and P (t) for the calculation cat
seen below. alogue o f the moment f, qv in the diagram.
However, we cannot presume to have found the It is obvious that the arc distance between PA and
systematic error (or, rather, the parameters ystat and P (t) equals ystat precisely, and that the compensation
(p$tat that characterize it) with absolute precision. The
matter is that individual measurement errors (which
are o f a random nature) also affect the values o f ystat
and (pitof. Therefore, we can only claim that the real val
ues o f the systematic error are close to ystat and <p5tof.
In order to make our statement more precise, let us
introduce the concept of a trusted interval. Let 1-e
stand for a certain level of trust. If e = 0.1, for instance,
the level of trust shall equal 0.9. The level of trust rep
resents the probability that guarantees the precision
o f our results; the trusted interval is the interval that
includes the unknown real value o f the parameter with
a minimal probability of 1 -e. Let us define

fy ( ) V i Sta t x e> Y stat Xel Fig. 5.0a. The two poles - on the ecliptic and in the catalogue.
136 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

o f the systematic error requires nothing but the su 1.

perposition o f these two poles. Let us now consider MAIN DEFINITION
the changes in the general picture that take place over
the course o f time. Since P(t) shifts within the limits From this chapter and on we shall assume to be
o f one degree, we can use a flat diagram and assume dealing with a catalogue whose every star has a single
that the motion o f P(t) is uniform, qv in the diagram. double among the stars o f the modern catalogue. Ac
Velocity v o f this uniform motion is easy enough cordingly, we shall be using index i in order to iden
to calculate if we know the values o f ystat for two dif tify the stars, as well as /, and b{ for the ecliptic longi
ferent points. We can then calculate the moment t* tude and latitude o f star i in the Almagest, respec
when the position o f the real pole is the closest to that tively. L,(t) and Bf(t) shall be used for referring to the
o f the catalogue pole. Prima facie we might assume real longitude and latitude o f star i in epoch t. Bear in
that this moment can be declared as the dating mo mind that time t is calculated backwards from 1900
ment deduced from processing the coordinates o f a a . d . and measured in centuries - that is to say, t =

great many stars. However, we have already demon 3.15 shall correspond to the year 1900 - 3.15 x 100 =
strated the fallacy of such logic; therefore, it has to be 1585 a . d . , for instance, and t = 22.0 shall correspond
said that one cannot date the catalogue to the moment to the year 1900 - 22 x 100 = 300 b . c .
of f*. Indeed, if the possible systematic error in Ptol Let tA equal the unknown time of the Almagest cat
emys estimation o f the ecliptic can equal the value of alogue compilation. The real longitude and latitude
8, all the moments in time that correspond to the pas o f star i for the year when the catalogue was compiled
sage o f the pole P(t) through a circle with the radius shall be indicated as I f and B f - that is, I f = Lt(tA)> B f=
equalling 8 whose centre lays in the point PA should Bi(tA). Let ABi(t) = Bi(t) - bt stand for the difference
be regarded as possible candidates for the moment of between the real latitude o f star i for moment t and
dating. However, we do not know the value o f 8. We its latitude as given in the Almagest. The value o f
can naturally estimate it, but only given that we know ABi(t) shall be referred to as the latitudinal discrep
the dating o f the catalogue. A different presumed dat ancy for moment t. This value shall stand for the error
ing shall yield a different estimation value. Therefore, in the estimation o f the latitude o f the Almagest star
this value already contains the presumed dating. i under the condition that it was compiled in epoch
Thus, depending on Ptolemys systematic error, t. It is natural that the real error in the estimation of
or the error in the determination o f the ecliptic, the the latitude is represented by AB,(tA) = AB f.
moment f* can either precede the real date o f the cat As we already pointed out in Chapter 3, we only
alogues compilation or postdate it. In the former have to analyze the latitudinal errors in the case of the
case, the catalogue (or, rather, the part o f it for which Almagest. The reasons for this were explained in de
we are trying to estimate the value o f ystat), gains extra tail above.
age, beginning to resemble a catalogue compiled in
the year t*. In the latter case (when t* postdates the 2.
real compilation dating) the catalogue becomes more THE PARAMETERISATION OF GROUP ERRORS
recent. Below we shall see that both these possibili AND SYSTEMATIC ERRORS
ties are implemented in the Almagest. However, the
terms extra age and more recent refer to a cata Let us consider a certain group o f stars such as a
logue where the systematic errors were not compen constellation or several constellations. We shall define
sated. What we end up with after the compensation the group error in the latitudinal coordinates o f these
is a refined catalogue which only contains random stars as the error in the estimation o f stellar latitudes
errors whose square average value can be estimated for the group in question resulting from the motion
to equal c min, although no individual value can be o f the stellar configuration under study across the ce
determined. lestial sphere as a whole. Therefore (we shall put a spe
Let us now consider the practical use o f the gen cial emphasis on this circumstance due to its exten
eral idea as specified above in more detail. sive use below), any subset o f this configuration also

shifts across the celestial sphere as a whole with the sin P = sin y sin (p (5.2.1)
same angle as the entire configuration. Such shifts
have three degrees o f freedom - that is, they can be The third degree o f freedom is defined by the ro
described by the specification o f three parameters tation o f the sphere around the axis P a? a> qv in
which we shall shortly define. fig. 5.1. However, this rotation only affects stellar lon
In fig. 5.1 one sees a diagram of the above. The po gitudes, leaving their latitudes intact. Therefore, we
sition of the real ecliptic for the time moment tA is shall not be considering this degree o f freedom. Let
represented on the celestial sphere whose centre is in us point out that instead o f the parameters specified
point O. The respective points o f the vernal and au we could choose any other set o f basis parameters
tumnal equinoxes are marked Q and R on the eclip that define the rotation o f the sphere. This obviously
tic. Point P represents North Pole o f ecliptics. Point cannot affect the further conceptual development of
E represents the position o f a given star. As we have our method.
already mentioned, all the group errors for a fixed Let us now study the distortion o f the real coor
stellar group in the ecliptic latitude made by the com dinates o f star i as affected by the systematic error of
piler o f the catalogue can be considered to stem from this kind. The real latitude B f and the latitude o f this
the miscalculation o f the ecliptic pole without ex star I f are equal to the lengths o f arcs EE' and QE'
ception, or the result o f the fact that the compiler counted clockwise as seen from pole P, respectively.
used the wrong point for the pole - PA instead o f P. The respective distorted latitude and longitude b{ and
This point corresponds to the perturbed ecliptic equal the lengths o f arc EEA and QAEA. Bear in mind
which is referred to as the catalogue ecliptic in fig. 5.1. that the latitudes of stars whose real longitudes are
Its position can be determined in a unique way after greater than the latitude o f point D and smaller than
we determine the following two parameters - firstly, that o f point C are reduced, whereas other latitudes
angle y between the lines OP and OPA, or the very increase, qv in fig. 5.1. This corollary does not apply
same plane angle between the planes o f the real eclip to all stars, strictly speaking. It is false for the stars lo
tic and the catalogue ecliptic. Secondly, we must cal cated at the angle distance o f y or less from the poles
culate angle (p between the equinox line RQ and line P and P '. However, since the value of y is anything but
CD that results from the intersection of the real eclip great, there are very few stars which can be found in
tic plane with that o f the catalogue ecliptic. This pa-
rameterisation is convenient for analytic purposes.
However, we shall also be using value (3 alongside (p,
which can be interpreted as follows (see fig. 5.1). The
shift o f the ecliptic can be decomposed into the com
position o f two rotations - one around the equinox
axis RQ equalling angle y, and the other around the
axis that also lies within the plane o f the ecliptic and
is perpendicular to axis RQ and equals angle (3. Thus,
P stands for the length o f arc QAQ which pertains to
the large circumference that goes through pole PA
and point Q. The astronomical meaning of the point
Qa is clear enough. It is the vernal equinox point on
the ecliptic o f the catalogue. It is obvious that angles
y and cp unambiguously define the angles y and (3;
the reverse is also true. The desired relation can be de
termined from the consideration o f a spherical right-
angled triangle CQAQ. The angle at the vertex QA is
a right one, the angle at the vertex C equals y, and the
length o f arc CQ equals p. The result is as follows:
138 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

and cp* which lead us to the solution o f the problem

in question.

a 2(Y*> (p*) min, (5.2.3)


o 2 (y, <p) =(Bf - b f - y sin(L?+ <p))2.

Fig. 5.2. The dependency between the systematic latitudinal
discrepancy and the longitude. Had there been no other errors in the catalogue
except for the systematic ones, the relation 5.2.3 would
transform into the equation o 2(y*> cp*) = 0. However,
such a small area. Virtually none o f those are con the presence of random errors in stellar coordinates
tained in the Almagest catalogue. As we shall see, the makes the minimum o f 5.2.3 differ from zero.
value of y equals circa 20'. In our situation, the catalogue compilation mo
Bearing in mind the value o f y being minute, one ment tA remains unknown; therefore, we must cal
can suggest the following approximated formula for culate the systematic errors for all possible values of
the latitudinal discrepancy: t from the interval 0 < t < 25 under study, namely, the
position o f the real ecliptic and the equinox axis are
A Bf = y sin(L? + cp). (5.2.2) calculated for every value o f t. Then, just as we see it
in fig. 5.1, the parameters y = y[t), cp = cp(t) and |J =
In other words, the systematic error in stellar lat (3(f) are introduced; they define the relative positions
itude estimation can be represented with the sine o f the catalogue ecliptic and the ecliptic for epoch t.
curve we see in fig. 5.2. It is very much like the curve The values o f y(t) and cp(t) are found as the solution
discovered earlier by Peters and Knobel ( [1339] ) when o f the problem
they were processing the data from the Almagest cat
alogue. The error rate o f formula 5.5.2 does not ex o^YW> <PW> *) - > min, (5.2.4)
ceed T for the stars whose \bA\< 80 and is therefore
o f no importance to us, so we shall consider the for where
mula 5.2.1 absolutely precise. For the sake o f propri
ety we shall exclude the stars whose absolute latitu a 2(y> <P> f) = Z (A B ;(t)-y sm (L ,(t) + <p))2. (5.2.5)
dinal values exceed 80 degrees from further consid
eration. We shall refer to the systematic error Once again, had this case been ideal (with no
hereinafter, since the methods described are only valid other discrepancies but the systematic error inher
under the assumption that we are considering a large ent in the catalogue), the relation 5.2.4 could be tran
group o f stars. The verification o f whether or not the scribed as the following equation (disregarding the
discovered discrepancy coincides with group errors minute effects o f proper star movement): c 2(y(t)y
for individual constellations is a problem in itself. Its cp(f), t) = 0 .
application to the Almagest is considered below, in As for the proper movement effects, let us remind
Chapter 6. the reader that the quantity o f visibly mobile stars
Assuming that the time tA o f the catalogues com on the celestial sphere is very small as compared to
pilation is known, we can calculate the parameters y the entire number of the Almagest stars. The solution
and cp which define the systematic error as follows: o f this last equation would exist for all the values o f
1) We shall calculate the real latitudes B f and lon t; however, these equations would not enable us to cal
gitudes L f for all the stars from the group under con culate the date o f tA. It is all the more impossible to
sideration (corresponding to the moment tA). calculate it from the relation 5.2.4 which acts as a
2) Then we must find the values o f parameters y* substitute for the equation in question when we con

sider a real catalogue containing random errors. We Let us point out that all such values can be calcu
can merely calculate the systematic error as a func lated for any time moment t, depending on the val
tion of the alleged dating t. This error is naturally de ues o f the modern stellar coordinates as well as the
pendent on the presumed dating due to the fluctua star coordinates in the Almagest catalogue.
tion of the ecliptic over the course o f time. It is pre Obviously, the minimization problem 5.2.4 is
cisely why we arent referring to the dating o f the equivalent to the minimization problem
catalogue, but rather the deduction o f its systematic
error as a function o f the alleged dating t. o (Y> <P> *) min> (5.3.1)
The real catalogue contains random errors apart
from the indicated systematic errors. Therefore, the in the sense that the parameters y(t) and cp(f) defined
discrepancies are random, and their values by the relation 5.3.1 coincide with the parameters de
are scattered around the sine curve o f their average fined by the solution o f the problem 5.2.4.
value as seen in fig. 5.2. Assuming that other errors As we already pointed out, solving problem 5.3.1
o f the catalogue than the systematic ones are of a ran only makes sense for large stellar groups, and since we
dom nature, the problem o f calculating y(t) and cp(f) shall study the statistical properties o f such a solution
is one o f regression parameter determination. below, we shall hereafter use ystat(t) and (p5tef(t) in order
to refer to values which satisfy to relation 5.3.1.
3. The value of
,(f), qw(t), t) (5.3.2)
is rather transparent from the point of view of physics.
Let us find the solution for the minimization prob It is the square average latitudinal discrepancy as ap
lem 5.2.4 and 5.2.5 expressed as y(t) and cp(f). Below, plied to the group o f stars under study for moment
in actual examples, this problem will be considered t resulting from the compensation o f the discovered
for groups containing different quantities o f stars. systematic error in ystat(t) and cp5taf(f). As we shall see
We shall therefore be using the following standard below, the value o f Gmin(t) is hardly dependent on
ized values for our calculations for which N will de time at all due to the extremely low proper movement
fine the quantity o f stars in the group under study. velocity o f most stars. Thus, we shall also use the in
dication Gmin. Bear in mind that the square average
latitudinal discrepancy prior to the compensation of
oo(Y>f>0=-^r<*2(y,<M), this error would equal the following value for m o
ment t
1 N
s b ( t ) = ' A B i ( t ) sin1,(0, I i N
" /=i Oinit = * 0 ( 0 ,0 , t) = (0) 2 , (5.3.3)
1 N
c b (0 = -rrZ Aj5/ (0 L, (0,
N =1 Thus, the difference Aa (t) = o init(t) - o min(t) esti
mates the effect o f compensating the systematic error
7smM, <Psml(t).
Further on when we shall define the values of y ^ (t)
and cp5iaf(f) from the relation 5.3.1, we shall presume
^ ( O ^ Z c o s 2^ ) , the time moment t to be fixed. We shall therefore omit
" M argument t from our calculations, that is, we shall use
L, instead o f L,(t), sb instead o f sb(t) etc.
(O ^ E s in M O c o s L O O . In order to find the minimum in the relation 5.3.1,
** /=! we shall take the partial derivatives o f functions aj(y,
14 0 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

cp, t) by y and cp and render them to zero. Bearing the order partial derivatives o f the function o 2(y> cp, t)
formula sin(Lt + cp) = sinLtcoscp + cosLtsincp in mind, with respect to y and cp:
we shall end up with the following equations:

d2cr2(Y,<p,t )
sb coscp + cb sincp = a ll( t ) -
= y [s2 cos2cp + 2d coscp sincp + c2 sin2cp], (5.3.4) y=rjtoi(0<p=q>rt,/(0

d2a 2(y,<p, 0
- c b coscp + sb sincp = 1 2 (0 =
= Y [ - d cos2cp + (s2 - c2) coscp sincp + d sin2cp]. (5.3.5)
2cr2(y,(p, t)
If we divide the equation 5.3.4 by 5.3.5, we shall get O22W :
5<p2 Y=Y sm A > ) ^ = < P s m t (t )
sb + cb tan cp s 2 + 2 d tan cp + c 2 ta n 2 cp

~ cb + sb ta n (P - d + ( s 2 - c 2) tancp + d ta n 2 cp Keeping in mind the equations 5.3.4-5.3.7, we can

easily determine the following expressions for these
Once we render both parts o f this equation to a partial derivatives:
common denominator, we shall come to the follow
an = 2 ( s 2 c o s 2 (pstat + 2 d c o s q > stat sincpJto, + c 2 s m 2 (?stat) =
ing equation concerning tancp:
= (2 / YStat)(sb cos ^stat + cb sin9 stat)>
a12 = 2 (cb cos (pstat - sb sin cp^)>
(l+ ta n 2cp)(cbs2- s bd) + (l+ ta n 2cp)tancp(cbd - s bc2) = 0.
a 22 = 2st a t ( s 2 sin2 y s ta t ~ 2 d ^ stat COS(pJto, +C2 COS2(p ^ ).

This makes it easy to calculate the tangent o f the (5.3.8)

optimal value o f cpstflt: In order to estimate the errors in calculating the
square average error rate o(y, cp, t) considering the
aberration o f the values y and cp from the calculated
sb ^ ~ cbs 2
t a n <Pstat (5.3.6) optimal values cpstat and ystat>let us use the following
c b d ~ Sbc 2 decomposition of the function (f{y , cp, t) for the vicin
ity o f point (y(t),q>(t)):
The equation 5.3.6 permits a unique determina
tion o f cpstflt; after that, the optimal value o f ystat can <j 2(y , 90 * + a u (0 ( y - y s t a t i * ) ) 2 +
be deducted from 5.3.4, for instance: +2fl12(0(Y - Ystat (0)(<P - V stat (0 ) + ^22(0(9 " stat (0 ) 2
y stat =
In the last formula we disregard the terms o f mag
sb cos 9 stat + cb sin cpstat nitude order three and higher as related to the dif
ferences Y - Ystat(t) and cp - cpsfflf(t).
s 2 cos 2 (P stal + 2d cos <pito, sin <psto, + c2 sin2 <pstat
Formula (5.3.9) allows for an elementary estima
tion o f the sensitivity o f the square average error o(y>
4 cbd 2 - 2 sbcbd + slc\ + c j s j cp, t) to the variation o f parameters y and cp. For this
S C purpose it suffices to determine the values a n , a l2
(5.3.7) and a22 pertinent to the right part o f 5.3.9. After the
estimation of ysm(t) and cpsiai(f), they can be easily cal
Formulae 5.3.6 and 5.3.7 make it feasible to find culated by the formula 5.3.8.
the desired solution o f the problem of calculating the Formula 5.3.9 demonstrates that the level curves
estimations for cpstflt and ystat by the method o f min o f square average errors manifest as ellipses on the
imal squares. plane (y> cp), qv in fig. 5.3. The centre o f the ellipsis is
It would be expedient to conduct a sensitivity in point (Ystat, cpstflt) for which the value o f the square
analysis for this problem. Let us regard the second- average error equals Gmin. The direction o f the ellip-

computer for a quantitative calculation o f temporal

dependency estimation for ystat(t) and ([>5tat(f), qv in
Chapter 6. We shall merely analyse the qualitative be
haviour o f these functions herein.
Let us once again consider the celestial sphere, as
suming all o f the stars thereupon to be immobile for
the sake o f simplicity, thus returning to Ptolemys
conceptions, albeit merely for the sake o f simplifying
the argumentation and the calculations. We are well
entitled to it since the percentage of the stars with no
ticeable proper movement velocity (ones that move
by several arc minutes over the 2500-year time inter
Fig. 5.3. Level curves o f square average error a (y, <p, t) where
val under study) is comparatively low. Such stars
t is a fixed value.
hardly affect the calculation o f parameters ystat(t) and
(pstat(t) that we are concerned with presently.
tic axes and the relation between them are determined In fig.5.4 one sees the celestial sphere as well as the
by the standard analytical geometry formulae through real ecliptic for catalogue compilation epoch tA. It
the values a n , a n and a22, namely, the tilt angle a of would be expedient to compare figs. 5.1 and 5.4. In
one o f the ellipse axes is determined by the following the epoch tA that remains unknown to us the eclip
relation: tic pole P{tA) was occupying a certain position on the
celestial sphere. The compiler o f the catalogue was
tan 2 a = -------- . naturally not ideally precise in his indication o f the
% + a 22 ecliptic on the celestial sphere. Therefore the pole PA
o f his catalogue ecliptic assumed a position differ
The second axis is perpendicular to the first. The ing from that o f P(tA).
lengths o f the axes relate to each other as Xl/X2, where Let us draw the arc o f a large circle that shall con
Xxand X2 are the roots o f the quadratic equation nect the pole P(tA) with the respective vernal and au
tumnal equinox points Q and R. In addition, we shall

A2 - X (a n + a22) + (on a 22 - afe) = 0.


Above we have made the assumption that the mo

ment t is fixed. We shall now consider how the pas
sage o f time affects the behaviour o f the calculated
values ystat and <psttt.
This behaviour can be determined from the for
mulae cited in the previous section. These formulae
contain the values L{(t) and B^t) which define the
temporal dependency o f ystat and (p5tot. The changes
o f the longitudes (Lf(f)) and the latitudes over
the course o f time have been studied well enough, qv
in Chapter 1. The respective calculations were o f a Fig. 5.4. Geometrical definition o f the angles <p and y on the
complex enough nature and required the use o f a celestial sphere.
142 | h i s t o r y : f i c t i o n o r s c i e n c e ? CHRON 3 | PART 1

their shifts are a priori known to fall into the range

o f one degree. We shall thus make a two-dimensional
copy of the polar part of fig. 5.4 - see fig. 5.5.
As it is obvious from fig. 5.5, the real ecliptic pole
shifts over the course o f time due to the fluctuation
o f the ecliptic. For the interval under study the value
o f this shift equals a mere 25' on the average, and so
we can draw it as a straight line (see the dotted line
in fig. 5.5). The motion o f the ecliptic pole along this
line can be considered uniform with a high enough
degree o f precision. And so, the distance between the
poles P(t) and P(fA), for example, shall equal v(tA - 1)
Fig. 5.5. Calculating the qualitative temporal dependency of
JstaM and cpJtet(iA). where v is the velocity o f the ecliptic poles motion.
This velocity approximates 0.0 T per year. As we have
mentioned above, in the observation epoch tA the
draw the arc o f the large circle D(tA)D'(tA) that shall catalogue compiler had made a mistake in his esti
pass through P(tA) and cross the recendy-built arc Q mate o f the ecliptic plane which resulted in the shift
P(tA)R at a right angle in point P(tA). If we knew the o f the catalogue ecliptic pole into point PAwhich dif
date tA, then the method o f minimal squares as de fers from P(tA). Should this result in the perpendicular
scribed in section 3, would give us the opportunity between PA and the ecliptic pole motion trajectory
to find the parameters o f y and cp that define the mu cross it in point t * > tA as is the case in fig. 5.5, this
tual disposition o f the ecliptic for epoch tA and the error o f the compiler shall obviously add extra age to
catalogue ecliptic. Fig. 5.4 demonstrates that these the catalogue ecliptic, namely, make it correspond
very angles also define the mutual disposition o f the best to the ecliptic o f the year t*. The opposite hap
poles P(tA) and PA on the celestial sphere - namely, pens if this perpendicular crosses the trajectory in
the value o f y equals the length o f the arc P(tA)PA in point t* < tA- the authors mistake would thus make
arc values, and angle cpequals the angle PAP(tA)D (tA). the catalogue more recent. In order to give the reader
As we point out in Chapter 1, the celestial position an impression o f the real value correlations, we shall
o f the ecliptic alters in the course o f time. This is the indicate that for the Almagest the distance between
manifestation o f the ecliptic fluctuation effect. There ecliptic poles P(0) for 1900 a . d . and P(19) for the
fore, the ecliptic pole for moment t that differs from early a . d . period roughly equals 20 - the value ap
tA shall be located in point P(t) which will also differ proximates that o f the error ystat(tA).
from P(tA). Earlier we mentioned that the value ystat(tA)
The ecliptic pole trajectory on the celestial sphere equalled the length o f segment P(tA)PA, whereas
is indicated with a dotted line in fig. 5.4, one that cpstatic) was equal to angle PAP(tA)D'(tA). In a simi
crosses the points P(t) and P(tA). Thus, in order to lar manner, ystat(t) = P (t)P A. Here the horizontal line
combine the ecliptic o f epoch t with the catalogue on top refers to the length o f the segment. However,
ecliptic, one has to superpose poles PA and P(t) over angle PAP(t)D \t) does not equal cpstat(t)ysince by mo
one another. The length o f arc P(t)PA equals the value ment t the vernal equinox axis would shift by the
o f ystai(f), and the location of the ecliptic rotation axis value G)(fA- 1).
that provides for such a superposition can be para- Here costands for the annual angle velocity o f pre
meterised by angle PAP(t)D\t) where arc D(t)D '(t) is cession that roughly equals 50", qv in Chapter 1. This
parallel to arc D(tA)D'(tA). shift corresponds to the value o f angle D\t)P(t)S(t)
In order to understand the quantitative behaviour in fig. 5.5. Thus, cpstat(t) is equal to the angle
o f the functions ystat(t) and cpstat(t) better, let us use a PAP(t)S(t)ywhere angle D'(t)P(t)S(t) = co(tA- 1).
two-dimensional drawing depicting just the ecliptic In order to evade such bulky indications, let us
pole shifts. This is permissible, since the values o f assume that

X (t ) = P(t)P(tA), y = P (t A)P(t*), Z PAP(t*), 7 stat(^)

M/(0 = ZPAP(t)iy(t), 8 = ZD{tA)PitA)Pit).

The value o f ystat(tA) can be referred to as the eclip

tic estimation error; it has the order o f 20 in the
Almagest. Angle 8 does not depend on t and equals
the angle between the motion direction o f the eclip
tic pole and line D(tA)D'(tA) as estimated above. It is
obvious that

Z~ Kstatuta) sin(8
y = Yst*t(tA) cos(8 - (fw(fj).

Since x(t) = v(tA - 1), from fig. 5.5 we get

y tat(0 = J(v(tA- 0 + y ) 2 + z2 =

= ^ y 2tat(iA) + 2M tA - t ) + V2(tA - t ) 2. Fig. 5.6. Approximate view of the functions ystat{t) and

Quite obviously, the minimal value o f this func for the catalogue compilers accuracy, that is, the val
tion is reached with t = t*. If we are studying a case ues o f ystat(tA) and <pstatic)- Formulae 5.4.1 and 5.4.2
o f 11 - tA| \tA t |, the function o f y$tat^) behaves also define the nature o f the dependency Pstat(t)yqv
almost as if it were linear: in formula 5.2.1.
Let us discuss the geometrical meaning o f these
Ystatic * Ystatic) + w o s (5 %,at((A))((A ~ ') calculations. We shall consider the Ptolemian coordi
nates of a certain star groups considering the obser
The function of (pstat(t) is also easy enough to find: vations to have been carried out in the time moment
t. We must then compensate the systematic error
<Pstat ( 0 = 0 + to( t A - t ) - arctan ystat(t), q> 5 or rotate the entire group by angle
vy + v f o - / ) / ystat(t) around the axis which is on the distance <pstat(t)
(5.4.2) from the equinox axis. We shall assume that we have
been perfectly precise in our estimation o f the sys
Once again, i f \ t - t A\ \tA - f*|, one can use lin- tematic error. Then the catalogue ecliptic pole PA shall
ear approximation: become superimposed over the real pole P(t). Obvi
ously, such a superimposition will not make the lati
tudinal discrepancies o f the stars equal zero, since the
<Prta<(0 = <p,to< ( ^ ) + [ t o + vsul(S -0 -
L y stat ) J catalogue also contains random errors. However, these
errors do not affect the position o f the ecliptic pole,
Naturally, the formulae that we end up with can having a null average value - or, rather, they affect it
only give us a general idea of the character o f such to a very small extent which is inversely proportional
functions as ystat(t) and <pstat(i). In fig. 5.6 we can see to the quantity o f the star group under study.
an approximated representation o f these functions From fig. 5.5 we see that the shift o f pole PA to
that we get from the formulae 5.4.1 and 5.4.2. It is ob wards the point P(t) can be decomposed into a com
vious that their actual form depends on the error rate position o f two shifts - PA to P(tA) and P(tA) to P(t)
14 4 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

in a single possible manner. The parameters ystat(tA)

and 9 stat(tA) which define the first shift refer to the ob
servers error, namely, the error made by the cata
logue compiler in the estimation o f the ecliptic plane.
The second shift is defined by the centenarian fluc
tuation o f the ecliptic plane which can be calculated
by Newcombes theory.
All o f the above also implies the following corol
lary. Let us mark the latitudinal discrepancy o f star i
calculated for the presumed observation moment t as
AB,(i), and the same discrepancy for moment t after
the compensation o f the systematic error as AB?(t) = are located in points t{ and whose radiuses equal
- Ystatit) sin(Li(t) + 9 stat(t)). Then the values of Ystat(ti)> i = 1.2. Fig. 5.7 demonstrates how one calcu
ABf(t) shall be independent from t and equal the ran lates the values o f ystat(t) and 9 stat(t) that correspond
dom errors made by Ptolemy in the estimation of the to arbitrary t values. It just has to be noted that the
latitudes. The situation changes when mobile stars line S'S that angle 9 stat(t) is counted from crosses the
enter the stellar group under study. For them the trajectory o f the ecliptic pole motion at angle 8 (f).
value o f ABf(f) shall depend on the time t. The de This angle can also be calculated with the aid of New
pendency character is defined by the values of indi combs theory. The astronomical meaning o f the
vidual random errors as well as the direction o f straight line S'S is obvious enough - it is a straight
proper motion velocities o f all stars as viewed at once. ened out part o f a large circumference pertaining to
In particular, for the unknown epoch tA the value of the celestial sphere that crosses the ecliptic pole P(t)
AB i(tA) shall equal the random latitudinal error for o f epoch t and is perpendicular (at P(t)) to another
star i. It would be natural to expect that if this star large circumference which also crosses P(t) and the
moves fast enough, and happens to be well-measured equinox point o f epoch f.
at the same time, the value of AB f(t) should reach its In a similar way, the calculation of the parameters
minimum somewhere around the point tA. The size ysm(t) and 9 5tot(f) for all the values of t shall require the
of this minimum range depends on the value and the knowledge of two values only - 9 5tot(fi) and 9 statih)-
velocity o f a given stars proper motion and equals We shall however work with angle y. It is a pithy
hundreds o f years even for the fastest o f stars - value, being the error in the estimation o f the tilt
Arcturus, for instance. angle between the equatorial and the ecliptic plane.
The above consideration and fig. 5.5 have a rather Let us point out that this angle can be fixed with the
important implication that in order to determine the use o f the armillary sphere, for instance. Therefore,
pole PA o f the catalogue ecliptic we only need to know the error y inherent in the value o f this angle may be
the two values o f ystat which will correspond to two an instrumental error o f the armillary sphere, qv in
respective time moment values - 11 and t2. Chapter 1 . Thus, error y arises in the course o f as
Indeed, Newcombs theory makes it relatively easy tronomical observation naturally. Apart from that,
to determine the ecliptic pole motion speed v, qv in the choice o f y for the representation o f a parameter
Chapter 1. Let us fix two arbitrarily chosen time m o shall further receive statistical validation.
ments and t2 (see fig. 5.7). We shall use the formula
5.3.7 to calculate the values o f y a n d ys(ai(f2)- 5.
Let us now draw the line o f the ecliptic poles motion THE STATISTICAL PROPERTIES
through time, marking the points t1 and t2 thereupon. OF THE ESTIMATES OF y * * AND 9^
The scale we have to choose must make the distance
between the two points equal v|t2 - t \ . The position We shall now consider the problem of calculating
of the ecliptic pole PA is determined as the intersec the parameters y and 9 which define the systematic
tion point o f the two circumferences whose centres error o f the catalogue as a problem o f statistics. Let

us assume the following for this purpose: the cata Let us perform a substitution for sb and cb in the
logue compiler introduced the systematic error at formulae related above, using the difference
time moment said error is defined by parameters yA sin(Li(tA) + (pA) - ^ instead o f Ab{ and apply said
yA and (pA. Apart from that, let us assume the latitude substitution to formulae 5.3.6 and 5.3.7. We shall
o f each measured star to have been affected by the come up with the following expressions for the val-
random perturbation with a zero average as a re ues <p5M( and ystat.
sult o f the observation error, or = 0. It is pre
sumed that random errors which correspond to lN
(s2 cos Lf(tA)-d sin Z, (tA))
different stars are independent and distributed uni tancpA + ^=1________________
Y^(^2~'52c2) COS(Py4
formly. Let o 2 = E ^2 stand for the dispersion o f the tantpJ(a,= (5.5.2)
random value this dispersion remains unknown 1+ -
y A( d 2-STC2) cosy A
to us in general.
The latitude o f star i shall assume the following
form in these presumptions: i N
(sin L t (tA) + tan (pA cos Z, (tA))
y stat=yA ---------------------------------------- . (5.5.3)
bi = B,(tA) - yA sin (L ;(fJ + <pA) + ^ (5.5.1) 2
(s2 + 2</tancpA + c tan 2 cpA)coscpA

From the statistical point o f view, what we have in Let us introduce the value
front o f us is a sample that consists o f N realizations
of random values { b ^ o f the 5.5.1 variety. This sam R = (yA(<P - S2C2) cos ( p j- 1.
ple has to be used for the statistical calculation o f y
and cp parameters of yA and cpA, as well as the calcu In this case 5.5.2 can be transcribed as
lation of the a value which is equal to the square av
erage equation error. We shall localize the problem R N
ton<*>A + T7 2 X ( s 2 cos L i ( t A ) ~ d sinL i ( * A ))
immediately and study the estimations o f cp = (pstat tonq>stat = ------ -M-----------------------------------. (5.5.4)
and y = ystat calculated with the minimal square 2
1 + t j Y & t (c sin Li (tA) - d cos Lt (tA))
method. These estimations have the form o f 5.3.6 N /=l
and 5.3.7. Most o f our attention shall be turned to
wards the estimation o f the yA value for reasons ex The condition = 0 tells us that the received es
plained at the end o f Section 4. timation o f parameter ystat is not shifted, that is:
Formula 5.5.1 looks traditional for regression
analysis. Indeed, this equation claims observation &fstat = yA- (5.5.5)
error Ab{ = B,(tA) - b{ to be a random value with the
average yA sin(Lf(tA) + cpA) depending on unknown The dispersion for the estimation of ystat expressed
parameters yA and cpA, and the dispersion a 2. One has through Dy looks like this:
to estimate the values o f unknown parameters using
the minimal square method and determine the sta Dy = ----------------------------------------------------------=------ * ( 5 *5 *6 )
tistical qualities o f the estimations received. Under 2
N(cos <pAs + 2i/coscpA sincpA + c2sin cpA)

such conditions, the curve T(x) = yA sin(x + cpA) is

usually referred to as the line o f regression. If observation errors are distributed normally,
Let us define the values o f y and cp using the rela the same applies to the value ystat, and the first two
tions expressed in 5.3.6 and 5.3.7. Discrepancies Ab{ moments (5.5.5 and 5.5.6) define its entire distribu
are random by presumption. Therefore, the estimates tion. This fact shall give us an opportunity to build
o f <pstat and ystat that we get from these formulae are the trust interval for the value o f yA.
random values as well. Let us study their statistical The estimation analysis o f cpsfflt is a bit more
qualities and consider their relation to the unknown complex. Let us used the equation rendered from
true values o f cpA and yA. formula 5.5.4:
14 6 I h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

tanq>Jto, - tancpA = f(x)=(x/V5(tU), (5.5.9)

R ^
T7 Y A (( s2 + d tan cp^) cos L (tA) - (d + c tan cpA) sin L (tA))

o (x) = (2ji)_1/2 jcxp(-u / )du.
A /=!___________________________________ -00
1+ 2
(c sinL i itA) ~ d cos L i (tA))
The values o f ystat and (psfflf as calculated above are
the so-called punctual estimations o f the unknown
parameters yA and <pA. Since we have found the dis
as well as the fact that for large values o f N the sec tribution functions for these estimations, one can
ond item in the denominator o f the right part of 5.5.7 study the issue o f possible errors inherent therein.
is a small value. This value is indeed o f a random na Let us answer this question in standard terms used for
ture, with a null average and the dispersion of trust intervals based on formulae 5.5.5, 5.5.6, 5.5.8
and 5.5.9.
In mathematical statistics the problem o f confi
22 2
Nyl(s c - d )cos yA 2 dence interval calculation is dependent on the fol
lowing situation that we shall illustrate with the ex
If are distributed normally, the same applies to ample of estimating the value o f yA. This value is a de
the value under study. It has the following implica terministic error o f a very certain nature made by the
tion for the Almagest: even for N = 30 the probabil compiler o f the catalogue. As a result o f the statisti
ity PNthat the denominator o f the right part o f 5.5.7 cal estimation o f yA - with the aid o f the minimal
shall be negative does not exceed 5 x 10'3. This prob square method in our case - we end up with the ran
ability diminishes drastically with the growth o f N: dom value ystat. One wonders about the boundaries
P50 < 2.5 X KT4, P80 < 4 x 10-6, P 100 < 3 x 10-7, P200 < o f the unknown value yA if we already managed to de
8 X lO-13, P300 < 2 .5 X 1 0 -8. termine ystat.
Formula 5.5.7 implies that, in general, E tan <psfat ^ In order to keep these boundaries from becoming
tan <pA. However, we can easily obtain distribution trivial, we have to define the acceptable error rate
function F(x) o f the random value tan(p5iflf - tan<pA probability - that is, the probability o f specifying such
from this formula which we need for the estimation boundaries that shall not contain the true value o f yA.
of the trust interval for <pA. Indeed, if we are to dis Let us use for referring to the acceptable error rate
regard the rather improbable case o f the denomina probability. Confidence level shall equal 1 - in such
tor in 5.5.7 becoming negative, we can educe the ex a case. The random value o f y5tat is distributed nor
pression for F(x) from this formula: mally, with parameters defined by formulae 5.5.5 and
5.5.6. Therefore, for x > 0 we shall have
F(x) = P(tan <psfflf - tan <pA < x) = F(r\x - x),
P (hr 5/a/-Y^I<^)=
where random value r\x has the form of Let us define the value o f (/2) - the ffactiles o f
normal xE distribution from the equation:
Tlx = 77 Z / ((j2 + d(tan(pA + x) cos L (tA) ) - ( d + c2(tan <pA + x) sin L (tA))).
" i
O(j)^xe) - O{ - J d ^ xz) = 1- e,
Therefore, if values ^ are distributed normally
with the dispersion equalling a 2, value r\x shall have or, alternatively, another equation that gives the same
Gaussian distribution with a null average and the dis result 0(-^Z)~xe) = e/2.
persion of Then the interval

R2a 2 fy() ~' (Ystat~~ ^e) Ystat * tb) (5.5.10)

DOlx) = ~ n r ~ ( C2S2 - d 2)(s2 + 2d(x + tany A) + c2(x + tan <pA) 2).
shall represent the confidence interval for yA with
(5.5.8) confidence level o f 1 - . This follows from P(|ysfaf -
Thus, YA| > X e )= e .

Liber l .
h*c fiiam artgulusqui a pr'mdpio Saaitta guio e te sequaliitftt&rfliquuf t rd
.4iii!((tf .V iird i qua o L c aequiii, <9 erst dcfnouftrand,
si tmiminan ptiticmiti ftf qm
a pritwipui Aqyart) rkififtm ffidti<mifti M
tu d iM tf&ta'gradtitmtell r / . j ; f iiu if
twofufrair* kim nobit qua pn utimmite
quaataacmifi miiuitthiffctiinullum ar*
culicsariiofiibui dUkrtiu cti. bed qufium
i^uium & (m rfcniif icgcmi fif fluaulord
fctiptuw iii iigfiurd^utiicidfr diflch

qfDicertomqucKJ pumflenim diametra/

liter oppofttorum orientlt! angufui unt
m aim reidet alt anqtilo sltfius dudbut
fC(f1is3tqualH<l,nam(ifirc.iIufn hortZv fi
fis A b g defcriperimus oblnjuum ctsans
circuiiim A E G F m A & G pun&is fcipfo
inrerferlrcitiirr$ Itimil f ad & f a e du
iHisrcdtsarqualcsfunt fed f i d tpfi fgd
tXmgutiiafywithtswidfodfmktU arquaMscft. Vtrtjngiuif unui FGO&C b a
fttoertt$ ri*#* jWt. duoircAosfaciune.
t*p. X. A
liificcpsaurcm demonirabmus

D quomedo ki data nobis deelnac

none,angulos ettam ,quo obii'
quusorcuius ad honzontema
*,inue*Tiius,faciliorenamq; uiailU rdi
quiscapinmir,quodigtfur qtn ad meridi a-
mm tutin ,t;dem tilts iunr qui adfcdiorbti
Mmzuo?tmltimt,peifpiaHim cft.Seduc
idfdtui cfiamofbe rapurur, prmm <k/
mof^randum cft.Puiufla oNfep rircuii
qua ab ccdcm qukioAiali punCto qu*ti f s e ( ri ta Tehtbunt, quonmttwn
fr dfftan uiogioa qpdad im m horis nguli qui adatodon borizowem nptct
tfmeofiflltuuiitur^aquaki Mount untur,qui|ab rodem cquioodnt A no
qfSitenimmeridianus circulus a b g d , xqualMCfdifanf.xqualadcRiftmi um,
PMfiiaodHdis circuit imkirculus a i & ( W b iw q u * * q t ^ o a b f o < k m f o l
lim z o fi iy ir^ c irc ttb a 1 0 f fidcfcri ttM pndoM m alm lO fldaH dbin
bantitrduarobtfqtiicirculiportiont! f i t P * >*"!* ocddcnuli., duobus fa u l
& c ktrt f c ptmAa. Aiimmm
aequtnot ij pun&um cle fupponanrur, &T
f i BT c l arcus equates, <coiH|ybsei}2 h
a ir 8 f 0 te w|uaJf$fff<,q*ioc iiidcapcft
fftm u m in te Citrla^f*figunrqaa
tefuix,<pam pFFea qu* dcmoiiltrata
funt trialatcra unfits, tnbuslattribusaltc/
Hs (tngulaiitgulis pqualafunt J l
Pr^ffica l tiorzordspoitlo E l * /
qu*J f m , feniltcr E f affcnfs IC
irfctnfusjqutangulusquoque 8 1 f an-

Fig. 5.8. A page from a 1551 edition o f the Almagest.

148 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

When we try to calculate the value o f xe, we must for instance, which would render Newcombs calcu
particularly lean upon the value o f Dr which depends lations unnecessary. Then instead of the curves ystat(i)
on the unknown parameters a 2 and (pA. As it is usu and (pstef(t) we shall have constant values correspon
ally done in mathematical statistics, we shall replace ding to observation errors - however, the coordinate
a 2 in the formula for Dy by the residual dispersion system shall pertain to the epoch o f 1900 a . d . Then
we would draw confidence intervals around these
l N constant values whose length will not depend on t. We
CTo(Y stat ^ stat tA Y stat sin(Z#,-{t A) + cpstat))
iy i=1 shall end up with the same interval o f possible cata
logue datings as we did in our estimation o f errors y
defined by formula 5.5.3, and (pA by (p5fflf. Catalogue and (p for the presumed dating epoch t if we carry out
compilation moment tA also remains unknown to us; the statistical dating procedure described below. The
thus, all the calculations as listed above have to be only information we shall lose after this shall be the
carried out for all the time moments t in order to es estimated real values o f ystat and (psfflt.
timate the systematic error ystat(t), (p5fflt(f), assuming
the catalogue to have been compiled in the random 6.
fixed epoch t. COROLLARIES
In a similar way we can educe the confidence in
terval for (pA with the confidence level of 1 - e. This Corollary i . The group error o f a stellar config
interval ^ ( e ) shall look like this: uration results in said configuration shifting across the
celestial sphere as a whole. This shift can be parame
f y y )
*Pstat ~ ~ 2
1 + tan (p - yE tan y stat
^ stai + i 2
1 + tan q>stat + y E tan(pJto/;
r terized by two parameters, namely, y and (p (or y and
P), if we are to consider latitudinal discrepancies ex
ye being the solution o f the equation F(y) - F (-y e) = Corollary 2. The latitudinal discrepancies in
1 - e, where distribution function F is defined by the herent in the catalogue can be reduced as a result o f
equality 5.5.9, that is, e/2 - ffactile of the correspon compensating the group errors.
ding normal distribution. Corollary 3. If group errors coincide for a large
Note: the above estimations o f the true error rates part o f the catalogue, this common error is called sys
for y and (p in the catalogue as the presumed dating tematic and can be discovered by statistical methods.
functions are not only important for our being able Under the condition that the catalogue compila
to compensate them, but also for the indirect verifi tion epoch equals f, the values o f parameters (p(t) and
cation o f just how correct the suggested approach y(t) can easily be assessed with the minimal square
happens to be. For instance, if we came up with such method. The corresponding estimations o f ystat(t) and
a value o f ystat that would be several times greater (pstat(t) have the respective forms o f 5.3.6 and 5.3.7.
than the catalogue precision rate, it would indicate at Corollary 4. It suffices to know the values o f
the existence o f substantial effects that we did not ystat(h) and ystat(t2) for two different moments in time
take into account. for the reconstruction of functions ystat(t) and (pstat(f).
However, inasmuch as the dating itself is con Corollary 5. Confidence intervals J 9(e) and JY(e)
cerned, the actual value of ystat takes no part in the cor for the real values o f parameters (p(t) and y(t) were
responding procedure. All we need to know is the calculated under the assumption o f random errors
length of the respective trust interval. Therefore, one being distributed normally. See the respective for
could simplify the calculations to a great extent in the mulae 5.5.11 and 5.5.10.
following manner. One would have to calculate ystat Let us conclude by reproducing a page from a 1551
and (p5fflf for any fixed moment in time t0: 1900 a . d ., edition o f the Almagest in fig. 5.8.

Statistical and precision-related

properties of the Almagest catalogue

1. that the discovered values o f y$tat and <pstot do in fact

INTRODUCTORY REMARKS possess the nature o f systematic errata in substantial
parts o f the Almagest catalogue, and are by no means
In the preceding chapters we have estimated that a mere result o f numerous group errors superim
one o f the primary problems with the dating o f the posed over each other and differing from one small
Almagest by proper star motions is the problem of group o f stars to another.
real precision o f the Almagest catalogue star latitudes As a result, we shall calculate the area o f the ce
for different celestial regions. Therefore, one needs to lestial sphere that was measured well enough by
conduct a meticulous analysis o f star coordinate er Ptolemy. In fact, it turned out rather significant. Our
rata in the catalogue in general and different parts of dating o f the Almagest shall be based on star coordi
the latter. A preliminary and rather rough analysis has nates from this very area - one where Ptolemys cal
already been conducted (see Chapters 2 and 4). culations were the most precise.
The primary instrument o f this chapter shall in
clude the methods o f systematic star coordinate er 2.
rata calculation as described in Chapter 5. First of all, SEVEN REGIONS OF THE CELESTIAL SPHERE
we shall demonstrate that seven regions o f the Alma
gest star atlas as described above do actually differ 2.1. A characteristic of the seven areas that we
from each other by the system error rate as well as have discovered in the Almagest atlas
random measurement errata. We shall find errors in
ecliptic pole estimation for each o f these areas, as well In Chapter 2 we have described seven areas that the
as the values o f residual square average star coordi celestial sphere can be divided into; they are also very
nate errata. Moreover, we shall build confidence in manifest in the Almagest catalogue, qv in fig. 6.1.
tervals o f systematic error parameters ystat and <pstat for In this chapter, we analyse Ptolemys coordinates of
each o f the areas. 864 stars in total. These 864 stars were what we ren
Next we shall analyse certain comparatively small dered the 100 stars o f the Almagest to after a filtration
celestial areas - constellations and environs o f indi o f the following sort. Firstly, the so-called informata
vidual stars. The goal o f this analysis is to make sure stars were removed due to reasons considered in Chap-
150 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

is very clearly defined by the named stars of the Alma

gest. One gets the impression that Ptolemy ascribed a
special significance to celestial area A. This is also con
firmed by our preliminary analysis in Chapter 2. As
we shall see below, area A turns out the most impor
tant for our dating research. It also has to be pointed
out that the area in question contains the celestial pole
(marked N) and the ecliptic pole (marked P).
Named stars that surround area A must have
served Ptolemy as a basis o f some sort when he was
performing his observations. He referred to them as
he moved further towards the centre of area A, meas
Fig. 6.1. Seven areas that we discovered in the star chart uring the coordinates o f all the other stars. Measure
according to the Almagest. Named stars are represented by ment errata accumulated as he moved from one star
black dots. to another. One should therefore expect the stars from
region A that lay outside the Zodiac to be measured
worse in general than zodiacal stars. Half of the Alma
ter 2 - they arent included in the canonical constel gests named stars (6 out o f 12) are either part o f the
lations. Secondly, we have also filtered out the re Zodiac, or located in its immediate vicinity. The Zo
jects and the ambiguously identified stars. Table 6.1 diac includes Regulus, Spica, Antares, Previndemiatrix
contains precise indications concerning the Almagest and Aselli. Procyon is right next to the Zodiac.
stars that a given region includes, and the residual
amount of stars after the filtration for each area. We 2.2. The disposition of the ecliptic poles
have used Baileys numeration in this table, or star for each of the seven regions of the Almagest
numbers from the Almagest catalogue. star atlas
Let us consider fig. 6.1, which represents the divi
sion o f the celestial sphere into the abovementioned Let us first locate the disposition o f the ecliptic
regions. All 12 named Almagest stars are marked as poles for each o f the seven celestial regions o f the Al
black dots. It is easy to see that the outline o f area A magest. In Chapter 5 we demonstrate that the posi
tion o f the ecliptic pole in relation to the catalogue
stars is set by parameters ystat and <pstoi. These param
Region Bailey's numeration for the region before eters are estimated from the catalogue by the appli
o f the sky and after the filtration o f the catalogue
cation o f the minimal square method in accordance
in the Almagest before after
with the formulae (5.3.6 and 5.3.7).
A 1-158 and 424-569 249 Let us calculate the values o f parameters ystat and
B 286-423 and 570-711 262 <pstot for each o f the seven celestial regions separately.
C 847-977 116 Afterwards we shall mark each corresponding posi
tion o f the ecliptic pole in fig. 6.2. In the same illus
D 712-846 and 998-1028 143
tration we shall also define the motion o f the real
M 159-285 94
ecliptic pole P(t) that corresponds to the variations
ZodA 424-569 124 o f the alleged dating.
Zod B 362-423 and 570-711 168 In fig. 6.2 we have used the following segment as
an example: it connects the ecliptic pole for celestial
Table 6.1. The distribution o f the Almagest stars across the
area B with the real ecliptic pole for epoch t - 10
celestial areas with the specification o f just how many stars
remained in each o f said areas after the filtration o f the cata
marked P (10). The length o f this segment equals
logue. We were using Baileys enumeration, or the numbers yftat(lO). The angle between this segment and the
o f the stars as specified in the catalogue o f the Almagest. line that stands for arc D (10) D '(10), whose defini-

tion was cited in relation to fig. 5.4 and 5.5; its value
is equal to <pftat(10). Obviously enough, any other
epoch can be taken as t, ditto area B, and respective
values o f ystat and (pstat can be deduced with the aid
of fig. 6.2.
Table 6.2 contains the values o f ystflt(18) and
cpstat( 18) that we have calculated for each o f the seven
celestial regions. These positions provide an unam
biguous definition o f the observer ecliptic pole for
each o f the areas. However, we may have just as eas
ily taken any pair o f y5tat and <pstflt values for a random
t. We refer you further to section 5.4. Apart from that,
table 6.2 contains the values of o init(\S) and the resid
ual o min square average latitudinal discrepancies re
sulting from the compensation of the systematic error Fig. 6.2. The respective disposition of the mobile ecliptic pole
(see formulae 5.3.2 and 5.3.3). In section 5.4 we P{t) and the ecliptic poles as estimated for each o f the seven
demonstrate that o min does not depend on time mo parts comprising the Almagest catalogue.
ment t under consideration, if we disregard the in
substantial influence o f the proper star motion.
Therefore, o min is defined by the ecliptic pole exclu cision. The value o f Pinit( 18) corresponds to the per
sively, which can be estimated statistically for this centage o f the stars whose latitudinal discrepancy
group o f Almagest stars. doesnt exceed 10' for the dating o f 100 a . d . (t = 18),
As for proper star motion, it has to be pointed out 10' being the Almagest catalogue minimal scale step.
that it hardly affects either the estimated systematic The value o f Pmin corresponds to the share o f the stars
error ystat(t), (pstat(t) or the residual square average dis whose latitudinal discrepancy doesnt exceed 10' after
crepancy o f star coordinates in the Almagest cata the compensation o f the systematic error. This value
logue. Therefore, we can omit all references to the ef is hardly affected by the dating o f the observations for
fect o f proper motion, although it was obviously al large quantities o f stars as considered presently.
ways taken into account in our calculations. The disposition o f the statistically definable Alma
We have chosen the value o f t = 18 for table 6.2 just gest poles shown in fig. 6.2 as related to the trajectory
because this time moment corresponds to the Sca- o f the true poles motion tells us that in every celes
ligerian dating o f the Almagest. tial area except C the systematic error o f the Almagest
Further on, Table 6.2 contains the following sta catalogue makes the catalogue more ancient even
tistic characteristic o f Almagest stellar coordinate pre- as compared to the epoch o f Hipparchus. Let us re-

Areas of the Almagest sky

Characteristics A B C D M ZodA ZodB
J s m ,( 1 8 ) 18.5 13.6 9.7 26.6 19.4 16.4 20.0
9sfaf(18) 34.0 -34.5 -122.5 -52.7 -50.5 -21.7 -23.5
20.5 21.8 23.4 27.3 23.0 17.7 24.0
m in 16.5 19.2 22.5 24.4 20.5 12.8 19.3
Pim,(18),in% 36.5 35.5 33.6 28.7 37.2 30.6 30.9
Pm m > in % 50.6 43.5 43.1 35.7 45.7 63.7 44.0

Table 6.2. Calculated values o f error parameters ysifli(18) and 9 ^ ( 1 8 ) as specified in the Almagest for different celestial regions.
152 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

mind the reader that the system error minimum in 2.3. The calculation of confidence intervals
celestial region C falls over t ~ 10, or the year -9 0 0
(900 A.D.). Still, as we have mentioned above, the dis In the previous section we calculated discrete sta
position o f the pole o f Ptolemys Ecliptic isnt in tistical estimates ystat and cp5faf for the unknown pa
any way related to the date o f the catalogues compi rameters of the Almagest catalogues systematic error
lation. This disposition simply tells us the character (y and (p). We have already reminded the reader the
and the value o f the systematic error made by Ptolemy definition o f confidence intervals in section 5.5. Let
in the measurements of star coordinates as conducted us make the visual representation o f the result as fol
for one celestial region or another. lows. First we shall build dependence graphs for t and
Another implication made by fig. 6.2 is that the the estimates o f ystat(t) and cpstat(t)ywhere 1 < t < 25.
statistically estimated pole positions for regions A, Then we shall draw stripes on the resulting graphs,
ZodA and ZodB are rather close to each other - in whose vertical sections shall be the confidence inter
other words, Ptolemy appears to have made the same vals Iy(e) and ^ ( e ) with confidence level e = 0.1.
systematic error for each o f these celestial regions. Confidence intervals shall be calculated in accordance
We shall come back to this fact below, in our analy with the formulae 5.5.10 and 5.5.11.
sis o f individual Almagest constellations. Further The result o f these calculations can be seen in figs.
more, the ecliptic pole defined by region B o f the Al 6.3-6.9. More data on the borders o f different confi
magest catalogue is also located next to the pole for dence levels e and the two values o f the alleged Alma
groups A, ZodA and ZodB, as we see from fig. 6.2. The gest catalogue dating (t = 7, or 1200 a.d ., and t = 18,
position o f the pole for area M lays further away, and or 100 a.d .) can be found in table 6.3. This table con
that o f area D - even further off. Apparently, the sys tains the values o f half-widths o f confidence intervals
tematic error o f the Almagests areas M and D has a Iy(e). Let us remind the reader that the centre o f the
different value than that o f area ZodA. Area C looks confidence interval for y and each fixed value o f t is
like an obvious reject in fig. 6.2. the non-shifted estimate o f ystat(t), qv in section 5.5.

1200 A .D . 100 A . D .
Area \, > 0.1 0.05 0.01 0.005 0.1 0.05 0.01 0.005
xl 2.6 3.1 4.1 4.5 2.7 3.2 4.2 4.6
Xl 11.7 14.0 18.3 20.0 16.6 19.8 25.9 28.4
Xl 2.7 3.2 4.2 4.6 2.6 3.1 4.0 4.4
xl 14.7 17.4 22.8 25.0 22.1 26.2 34.4 37.6
xl 4.6 5.5 7.2 7.9 5.1 6.0 7.9 8.7
xl 91.1 108.2 141.9 155.2 60.7 72.2 94.7 103.5
x\ 6.3 7.4 9.8 10.7 7.2 8.6 11.3 12.3
xl 28.3 33.6 44.1 48.2 37.8 44.9 58.9 64.4
xl 5.4 6.4 8.5 9.2 6.5 7.7 10.1 11.0
xl 28.2 33.5 43.9 48.0 42.4 50.3 66.0 72.2
Xl 2.5 2.9 3.9 4.2 2.5 3.0 4.0 4.3
Z odA
xl 11.4 13.6 17.8 19.5 18.1 21.5 28.2 30.8
xy 3.5 4.2 5.5 6.0 3.4 4.1 5.4 5.9
Z odB
*6 14.3 17.0 22.3 24.4 19.8 23.5 30.8 33.7

Table 6.3. Semi-width values x\ o f confidence interval /y(e) and x l o f confidence interval /<p(e) for different confidence levels
o f E and two presumed datings o f the Almagest catalogue - 1200 A.D. (f = 7) and 100 A.D. (f = 18).

The confidence interval /^(e) for cpis, generally speak

ing, asymmetrical in relation to cpstoi(t), since this es
timate might be shifted. However, the abovementioned
asymmetry is insignificant enough, and one may con
sider <pstat(t) the approximate centre of the confidence
interval. x\ stands for the semi-width of interval JY(e),
and - for the semi-width of interval /^(e).
The figures one finds in tables 6.2 and 6.3 imply
the following. Almagest area ZodA is the most accu
rately measured celestial region. This is obvious from
the fact that the compensation o f the discovered sys
tematic error for this group o f stars allows reducing
the square average error to 12.8f. Also, it turned out
that 64% o f the stars ended up with a latitudinal dis
crepancy o f less than 10'.
The second most precise group o f stars pertains
to the Almagest area A, where the square average lat
itudinal discrepancy became reduced to 16.5' after
the compensation o f the systematic error. The share
o f stars whose latitudinal discrepancy is under 10'
has grown to over 50% in this area.
Confidence intervals Iy(e) and /^(e) for celestial Fig. 6.3. The behaviour o f systematic errors Jstat(t)ycpsiai(f)
areas ZodA and A turned out to be o f similar sizes, qv and pstat(t) for celestial region A in the Almagest.
in table 6.3, although the precision o f measurements
is higher in area ZodA. This is explained by the het
erogeneous quantities of stars for these parts. The less
stars, the greater the size of the confidence interval; the
latter is reduced by higher measurement precision.
The data from Table 6.2 confirm Ptolemys claimed
precision o f 10f, insofar as stellar latitudes are con
cerned, at least.
The next best measured groups o f Almagest stars
are concentrated in areas B and ZodB. Their precision
characteristics are rather close to each other. The
residual square average error is approximately equal
to 19'. Stars with a latitudinal discrepancy o f under
10' constitute 44% o f these groups. The positions of
the ecliptic pole calculated by these Almagest sky parts
seem close to the pole positions o f areas A and Z od A
at a cursory glance; however, they end up in respec
tive confidence intervals only with sufficiently small
values o f 8 ~ 0.01, which means that the systematic
errata o f celestial areas B and ZodB may differ from
those of A and ZodA. Moreover, the stars in areas A
and ZodA were measured with substantially greater
precision than those in areas B and ZodB. Below we Fig. 6.4. The behaviour o f systematic errors ystat(t), cpstoiW
shall cite more evidence that testifies to this. and Pstof(t) for celestial region B in the Almagest.
15 4 I h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

Fig. 6.5. The behaviour o f systematic errors ystat(t), cpstat(t) Fig. 6.7. The behaviour o f systematic errors ystat(t), (pstat(t)
and Pstat(t) for celestial region C in the Almagest. and pstat(t) for celestial region M in the Almagest.

Fig. 6.6. The behaviour o f systematic errors ystat(t), (pstat(f) Fig. 6.8. The behaviour o f systematic errors ystat(t), (pstat(t)
and Pstot(i) for celestial region D in the Almagest. and Pstoi(i) for celestial region ZodA in the Almagest.

Celestial region in the Almagest atlas

ZodA A

H 1.11 0.82

12 0.042 -0.03
22 0.073 0.13
^min 12.8 16.5'
= 0.1 1.3 1.2
= 0.05 1.8' 1.7
= 0.01 3.0 1.8
e = 0.005 3.5 3.3

14.1 17.7

= 0.05 14.6' 18.2
^ max
= 0.01 15.8 19.3
= 0.005 16.3 19.8

Table 6.4. The values o f a ( l l ) , a{ 12) and a(22) as calculated

Fig. 6.9. The behaviour o f systematic errors ystat(t), (pstef(f) for the Almagest, assuming the date o f its compilation to be
and Pstat(t) for celestial region ZodB in the Almagest. close to 100 A.D. (t = 18).

The stars in areas C, D and M were measured systematic errors pertinent to celestial regions A,
worse than those in areas A and B. Moreover, the val ZodA, B and ZodB.
ues o f ystat and <p5tof estimates only end up inside con The analysis o f tables 6.2 and 6.3 has already made
fidence intervals o f areas A, ZodA, B and ZodB when us enquire about the values o f the square average error
the values o f are very small indeed, which means that one should consider great and small. Let us refer
that we must allow for the existence o f such system to the sensitivity analysis as described in Chapter 5.
atic errata in areas C, D and M that differ from the The solution scheme can be seen in fig. 6.10.
Let us draw the ellipsoidal level curves o f function
a 2 (y, (p, t) on coordinate plane (y, (p) according to for
mula 5.3.9. We shall draw the rectangle R(e) on the
same plane, with coordinate projections Iy(e) and
J9(e). In fig. 6.10 it is the shaded rectangle. In this case,
the probability that the true value of system error (y, (p)
lays inside this rectangle is 1 - 2e or greater. Let us find
chaxiz) = max G2 (Y> <P>t)ywhere the maximum is taken
for each o f the pairs (y, (p) e R(e). The resulting value
of GmflJC(e) defines the permissible square average dis
crepancy with a confidence level o f 1 - 2e, whereas the
difference o f - O min defines the permissible ex
pansion o f the square average discrepancy due to the
Fig. 6.10. Estimating the allowable variations o f the square lack of sufficient precision in the estimation of pa
average latitudinal discrepancy values. rameters y and <p by the values o f ystat and (p5toi.
156 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

Table 6.4 contains the values o f a n , a 12, a 22 for ce ori exclude the possibility that the compiler made a sep
lestial areas A and ZodA for the time moment of t = 18; arate group error in case of every small star group such
they define the level curves of the square average error. as an individual constellation. In this case, parameters
These level curves are calculated with the aid o f for yxat and cp5tof are but average meanings of the true group
mula 5.3, which stipulates the measurement of yin arc errata, and will be o f little use to us for this reason.
minutes and cp in degrees. The table also contains the We have to note that the sizes o f confidence in
values o f Aa = c max(e) - c min calculated for the ex tervals for the values o f cpstatfound in Section 2 are
treme values o f e = 0.1 and = 0.005. It has to be said rather substantial. This may be explained by the low
that the resulting values appear to change little over sensitivity o f latitudinal discrepancies to the turn
time. These figures demonstrate the obvious preci angle cpas well as the non-systematic nature of the
sion division between areas A and ZodA on the one cpsfaferror. In other words, it is possible that parame
hand, and B and ZodB on the other. Indeed, even with ters ystat and cpstafhave a different nature, namely, ystat
the confidence level o f 1 - 2e = 0.99, the square aver is the result o f an observers error that affects all stars
age error value o f the confidence area constructed for (an error in the estimation o f the ecliptics position),
the region ZodA remains less than the minimal error whereas cpsfafis the averaged value o f numerous indi
value o f celestial regions B and ZodB. vidual errors. Such a difference in the behaviour of
A similar statement shall also be true for celestial the parameters is easy to explain if we consider the
region A. Although <3A max o f region A is greater than primary astronomical instrument o f the Almagest
<5Bmax, this is only true for < 0.01. Other values make epoch, for instance - the armillary sphere (see
error levels o f celestial regions A and B substantially Chapter 1). The angle between the equatorial and
different, or separated by a statistical criterion. It must ecliptic plane is fixed once and forever in the very
be added that the stars in the ZodA group are just as construction o f this instrument. If there was an de
different from their counterparts from group A pre fect in the latter, it would affect the coordinates of each
cision-wise, since for all values considered the value and every star measured with the aid o f this armil
of c max found for ZodA is less than c min calculated for lary sphere. The error in the estimated value o f angle
region A. cpis o f an altogether different nature. It is individual
Furthermore, table 6.3 demonstrates that the pa for each star and changes as the observer measures the
rameter cpsfaf cannot be calculated with sufficient pre coordinates o f several consecutive stars.
cision, especially for the poor quality regions C, D One must therefore find the group errors charac
and M. This is confirmed by the sizes o f confidence teristic for individual Almagest constellations and
intervals /<,,(). For example, the full range o f this in compare them to the systematic error o f ZodA, the
terval exceeds 180 degrees in case o f area C. best measured group o f Almagest stars.

3. 3.2. The calculation of systematic errors

OUR ANALYSIS OF INDIVIDUAL ALMAGEST for individual groups of constellations
CONSTELLATIONS in the Almagest

3.1. The compiler of the Almagest may have The present section analyses a total o f 21 small
made a different error in case of every minor groups o f Almagest stars. Their list can be found in
constellation group Table 6.5, whose structure is completely identical to
that of Table 6.1. Our only additional indication con
Further analysis is necessary due to the following cerns the principle o f selecting limited stellar config
problem. Parameters y ^, and cpstot, which define the urations. All o f the above are zodiacal constellations
systematic error, have been found for some large group from the Almagest, likewise the environs o f named
of stars. They correspond to the turn of the ecliptic that stars, with the exception of Canopus and Previndemia-
minimises the square average discrepancy for the stars trix (made for abovementioned reasons), as well as Pro-
contained in this group. However, one must not a pri cyon, due to the paucity o f stars in its environment.

Almagest star group Bailey's star numbers for the group Number o f stars in a group
i. Zodiacal constellations
Aries 362-371, 373,374 12
Taurus 380-388,390, 391,393-410 29
Gemini 424-440 17
Cancer 449-454 6
Leo 462-481,483-488 26
Virgo 497-516, 518-520 23
Libra 529-534 6
Scorpio 546-565 20
570-573, 575-583, 585, 586, 590,
Sagittarius 22
591, 593, 594, 596-598
Capricorn 601-608,610-627 26
Aquarius 629-650,652-656,658-660,662-668 37
Pisces 674-695, 697, 699-701, 704-706 29
2. Environs of named Almagest stars
Antares 546-569 24
Cappella 220-233 14
Aquila 286-300 15
Vega = Lyra 149-158 10
Arcturus 88-96,98,100-110 21
Sirius 812,818-835, 837-846 29
Spica 497-503, 505-515, 518-526 27
Regulus 462-481,483-488,491-493 29

Table 6.5. Stellar compound o f 21 Almagest star groups; for each o f the latter, the values o f systematic (group) errors were
calculated. These groups include all the zodiacal constellations o f the Almagest, as well as the neighbourhood o f 12 named
Almagest stars, with the exception o f Canopus and Previndemiatrix. The table contains Baileys enumeration, or star num
bers as given in the Almagest catalogue.

The location of group errors for individual Alma However, the value o f <5G min defines the lower
gest constellations is associated with the following boundary o f possible square average errata for group
problems. Let us consider a certain star group G and G. This minimal value o f possible error results from
find the corresponding values of y Gat and (pGat by ap turning the coordinate system by angles y G stat and
plying the method o f minimal squares. This will also <pGat. Obviously enough, the values o f y Gat and (pGat
define the minimal possible residual square average can greatly differ from those of ystat and (pstof, which
discrepancy o ^ in, as well as the share of stars whose were calculated for a larger number o f stars that had
residual latitudinal discrepancy is less than 10'. This will included group G.
also define P ^ in in relation to the time moment t = 18. The identity criterion o f group error for group G
However, due to the small sizes o f certain star groups, and the systematic error calculated for a large num
the statistical discrepancy of estimates y Gat and <pGat is ber o f stars could be expressed as the approximated
too great to serve as a basis for justified corollaries. equation o Gin ~ o G where o f is the residual square
158 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

average discrepancy for group G after the coordinate happens to prove our claim independently. It also
system is rotated by angles ystat and (pstat. Indeed, the needs to be pointed out that both proportions are
above approximated equation means that ystat and temporally independent, if we are to disregard the
cpsfflf are almost optimal values. In order to support proper star motion. Therefore, their practical verifi
this criterion, let us define the auxiliary values of cation can only be conducted for a single moment in
and P?, which stand for the share o f stars from group time - any one such moment, that is.
G whose latitudinal discrepancy does not exceed 10 We have calculated the values o f g g and P^ for dif
after the respective rotations o f (yGtat and (pGat) and ferent Almagest groups G and the time moment of t =
(Ystat and <pstat). Should we also observe a case of 18. Let us reiterate that these values equal to the re
P p w e can conclude that group G does indeed pos spective square average latitudinal discrepancy and
sess the same systematic error value as the stars o f a the share o f stars whose latitudinal discrepancy value
greater group. We must note that the latter approxi does not exceed 10', given that the pole o f the eclip
mate proportion is not implied by the former, but tic coincides with the pole defined for the most accu-

_G _G pr 9tmt.
Star group Indication of G Ginit min o? PG
x mm P?
l. Zodiacal constellations
Aries Z1 19.7 17.2 18.9 45.5 45.5 72.7
Taurus Z2 23.2 18.1 20.6 27.6 41.4 41.4
Gemini Z3 17.8 10.5 11.0 29.4 82.4 58.8
Cancer Z4 13.8 4.3 5.2 33.3 100.0 100.0
Leo Z5 20.2 11.1 11.2 19.2 65.4 65.4
Virgo Z6 18.4 13.6 14.4 39.1 56.5 47.8
Libra Z7 8.4 6.1 9.3 83.3 83.3 83.3
Scorpio Z8 18.8 13.7 15.1 30.0 65.0 55.0
Sagittarius Z9 16.4 14.3 15.8 30.4 60.9 60.9
Capricorn Z10 16.2 10.6 11.3 42.3 65.4 57.7
Aquarius Z ll 28.6 17.3 19.2 18.4 44.7 44.7
Pisces Z12 22.5 21.5 21.7 51.7 41.4 34.5
2. Environs of named Almagest stars
Antares SI 17.7 12.6 13.8 33.3 70.8 58.3
Acelli S2 15.7 11.0 12.1 33.3 58.3 66.7
Cappella S3 34.6 30.3 34.0 35.7 14.3 64.3
Aquila S4 24.0 23.7 26.7 40.0 33.3 13.3
Vega = Lyra S5 20.0 14.1 17.1 50.0 60.0 30.0
Arcturus S6 24.2 17.2 20.0 19.0 38.1 28.5
Sirius S7 15.2 11.9 25.9 47.4 52.6 15.8
Spica S8 17.9 14.1 14.5 44.4 48.1 48.1
Regulus S9 25.2 21.0 21.1 17.2 58.6 58.6

Table 6.6. Calculation results for the 21 Almagest star groups. Here af nit, 0 G
min > correspond to square average latitudinal
discrepancies in group G - the initial and the remaining, as well the one that we come up with after compensating the sys
tematic error in G as estimated for ZodA. We also cite the stellar percentage values o f Pfm-f , P ^ , P? with a m inimal latitu
dinal discrepancy o f 10'.

rately measured group o f stars in area ZodA. In other markable quality in the Almagest. The square aver
words, the condition is that the group errors must age error and the percentage of stars with the max
equal the values o f y ^ f 4 and cp^f4. imal latitudinal discrepancy o f 10' calculated under
The square average latitudinal discrepancy and the the assumption that the group error is equal to ( y ^ f 4,
percentage o f stars whose latitudinal discrepancy ( p a r e only marginally different from the values
value doesnt exceed 10' (in group G, without the o f Gmin and Vmin calculated for the optimal ecliptic
compensation o f the systematic error) were tran pole position in the constellation under study. The
scribed for t = 18 as Gfnit and P fnit, respectively. greatest discrepancy between the two was noted in the
If the value o f a ? exceeds the minimal possible most orderly constellation o f Libra, where no value
value o f a in, but very slightly so, we are entitled to Of a ini o min or Oj exceeds 10', and P,1( = Pmin = Pj =
assume that the group error value o f star group G 83,3%. Such is the percentage o f stars whose latitu
equals the systematic error value o f celestial region dinal discrepancy value is less than 10'. The equation
ZodA. The difference between the values o f P? and Pinit - Pmi = P x is easy to explain - the constellation
P ^ is yet another proximity criterion for group error in question all but lays on the equinoctial axis, thus
and systematic error. Let us remind the reader that the remaining quite unaffected by the turn.
values GG minand ggare temporally independent for the However, this corollary may also be true for the
immobile stars and only marginally depend on time constellations from celestial region ZodB, although
in case o f their mobile counterparts. A similar state with more details to take into account. However, the
ment shall be true for the stars that end up within the veracity or inveracity o f this corollary is o f no im
10' interval o f the latitudinal discrepancy. portance to us presently, since celestial region ZodB
Table 6.6 contains the numeric data that we have doesnt contain any named Almagest stars.
calculated. A more visual representation thereof can We must nevertheless point out a single curious
be found in figs. 6.11 and 6.12. Fig. 6.11 contains the fact that was found out in relation to the constella
information about the values of G ^in and o f, as well tion o f Aries. Although the value o f became lower
as VGand P ^ in, for all the zodiacal constellations of the in comparison to Ginit after the compensation o f the
Almagest (indicated Z l , . . . , Z12). Fig. 6.12 contains systematic error discovered earlier (one must also
respective results for the environs o f the named note that the difference between Gminand Ginit is rather
Almagest stars; they are marked S I , . . . , S9. One must small), but Px Pim>= Pmi - in other words, the shift
say that the environs o f the named Zodiacal stars in o f the ecliptic pole into the position calculated for
the Almagest do not fully correspond with the re ZodA made it possible to raise the share o f well-meas
spective Zodiac constellation. These environs are con ured Almagest stars in the constellation o f Aries to
stituted by a group o f stars from this constellation, 72.7%.
which have received a name in Bayers system. These The general conclusion resulting from our the
stars are usually the brightest and the most reliably consideration o f all zodiacal constellations is as fol
identifiable stars of the Almagest, which makes them lows. If the proportion Gmin Ginit is true for the op
more solid corollary basis. timal value o f Gmini the conjecture that the group
error equals the systematic error for celestial region
3.3. Group errors for individual constellations ZodA and the ensuing compensation o f this error
from the well measured celestial region of the lead us to the proportion o f Gy Ginit; other valid
Almagest are virtually identical to the proportions include Px Vinit and Pmin P **-This
systematic error discovered as a is true o f the following Almagest constellations:
characteristic of this area in general Gemini, Cancer, Leo, Virgo, Scorpio, Capricorn and
The key implication o f the cited graphs and o f If the value o f o min is close to o ta- o min < o , < c ,m(
Table 6.6 is that the zodiacal constellations from ce as a rule, and the effect of placing the ecliptic pole into
lestial region ZodA (namely, Gemini, Cancer, Leo, the position that corresponds to area ZodA is hardly
Virgo, Libra and Scorpio) possess the following re manifest at all. This is true o f the Aries constellation
i6o | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

Fig. 6.11. The dependencies o f o min, a v Ginit, Pmirt, Vinit Fig. 6.12. The dependencies o f Gmin, Gv <5iniv Pmin, Pp ? init
for the zodiacal constellations. for the areas around named stars.

(as we have pointed out, the percentage o f well-meas 3.4. How the compensation of the systematic
ured stars grew dramatically in case o f Aries), as well error that we have discovered affects the
as Taurus, Libra, Sagittarius and Pisces. precision characteristics of the environs
Out o f the constellations pointed out above, good of named stars
precision characteristics o f Libra from celestial area
ZodA remain virtually unchanged after the shift of the The situation with the environs o f named stars in
ecliptic pole from the optimal position to the posi the Almagest is more diverse. First o f all, let us point
tion that corresponds to ZodA. Precision character out the environs of Aquila and Sirius. In both cases,
istics o f Aries become even better after this opera the compensation o f the discovered systematic error,
tion, and those o f all the other constellations remain characteristic for celestial region ZodA, leads to the fol
average. lowing. Firstly, we observe a growth o f the square av
Taurus is a typical example, with Ginit = 23.2, Gmin erage latitudinal discrepancy, which is rather sub
= 18.1', a , = 20.6, P,mt = 27.60/g and ? min = P, = stantial in case o f Sirius - from 15.2' to 25.9'. Sec
41.4%. The constellation o f Pisces differs from all the ondly, the share o f well measured stars shrinks (from
other Almagest constellations, with Vmin < Vinit and 40% to 13.3% for Aquila, and from 47.4% to 15.8%
Pi < P,m> given that o init = o mi = Cj. for Sirius). The obvious conclusion to make is that the

group error o f the compiler made during the meas

urements o f the environs o f Aquila and Sirius is sub
stantially different from the systematic error o f celes
tial region ZodA. Unfortunately, it is impossible to
calculate these errors veraciously. Therefore, Sirius
and Aquila were excluded from further consideration.
The environs o f other named stars have basically
the same properties as the zodiacal constellations -
namely, stars from the environs o f Antares, Acelli,
Arcturus, Spica and Regulus greatly reduce the square
average error, bringing it close to the minimal possi
ble values after the compensation o f the group error,
which equals the systematic error for region ZodA.
The percentage of stars whose latitudinal discrepancy
value is smaller than 10' (Px) shall dramatically grow
as compared to the initial value o f Pimt. The environs
o f Cappella have the same property as the constella
tion o f Aries - namely, the square average latitudinal
discrepancy of this area doesn't change much after the
shift o f the ecliptic pole from the initial position to
the optimal position and then also into the position
calculated for celestial region ZodA. However, in the
last o f said positions the share o f stares that fit into
the ten-minute latitudinal discrepancy value grew
drastically in the vicinity o f Cappella, reaching 64.3%.
For comparison, let us point out that in the initial po
sition this share equalled 35.7%, and as little as 14.3%
in the optimal position dictated by the square aver Fig. 6.13. The dependencies o f o min, o 2, o imt, Pmin, P2, Pimt
age latitudinal discrepancy. On the contrary, the stars for the zodiacal constellations.
neighbouring with Vega demonstrated a substantial
reduction o f the square average latitudinal discrep
ancy. However, when we shifted the ecliptic pole into ture o f ystat)ythe issue o f whether or not the error of
the position characteristic for celestial region ZodA, cp5tat might be systematic as well remains open. Let us
the number o f stars with the latitudinal discrepancy solve it in the following manner. Let us consider some
value of 10 minutes and less was reduced substantially. individual Almagest constellation. We shall not go
Therefore, the nature o f group errors in the environs beyond the zodiacal constellations - the six named
o f Vega and Cappella remains unclear. Little wonder stars pertain to the Zodiac, at any rate. Let us calcu
- one might as well recollect that these stars lay at late the characteristics o f o 2 and P2 for these constel
quite some distance from the celestial region of ZodA. lations, which can be done as follows. The first char
acteristic is the residual square average discrepancy,
3.5. The discovery of a single systematic error and the second - the share o f stars in a constellation
made by the compiler of the Almagest for the whose latitudinal discrepancy does not exceed 10'.
region of ZodA and the majority of named stars Both characteristics result from the statistical error
calculated for region ZodA, and cp(2), calculated
Although we have discovered a certain proximity as a necessary pre-requisite for the minimization o f
between the characteristics o f o l and P x, respectively the g 2 error. In other words, this is what we come up
to Gmin and Pmin (which testifies to the systematic na- with for constellation G:
16 2 I h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

of = of (t) = min oG(f) = min oG( (p, t), o f (p ^ f1 can be an averaged result o f many individ
<P (p ual measurement errors, and there is no reason to
<p<2) = arg min o G( y ^ f , <p, i). consider it a systematic error. Moreover, the value of
<pstflt is calculated rather roughly, which makes it rather
uninformative in this respect.
Let us compile table 6.7, which is similar to
table 6.6. Moreover, some data recur for better de- 4.
monstrability. In table 6.7 the values o f and are COROLLARIES
replaced by o 2 and P2. Let us also draw these data as
fig. 6.13, which is similar to fig. 6.11. Both the table Corollary i . It has been proven statistically that
and the drawing make it obvious that the compen the ecliptic poles o f stars from celestial regions A and
sation o f systematic error y ^ f* in zodiacal constella ZodA are very close to one another, which makes the
tions from celestial area ZodA and the variation of the values o f the systematic error made by the compiler
(p value may give us minimal possible values of o 2, of the Almagest in these parts o f the sky the same.
which are very close to Gmin or even equal to o min. Corollary 2. The statistical analysis that we have
Likewise, the value o f P2 will be close to Vmin or equal conducted gives one no reason to believe that the
thereto. Remarkably enough, the same is true for the systematic error values o f the Almagest catalogue for
constellations from celestial region ZodB. celestial regions C, D, Ai, B and ZodB have anything
All o f the above proves it beyond any doubt that in common with such values characteristic for areas
the value o f y ^ f 1 that we have discovered is indeed A and ZodA. Systematic errors o f areas C, D, and M
the systematic error made by the compiler o f the are very likely to differ from their counterparts in
Almagest catalogue as he measured the stars from ce areas A and ZodA. We can say nothing o f any sub
lestial region ZodA, as well as named stars, with the stance about the errors that characterise celestial re
exception o f Sirius, Aquila and Canopus. The value gions B and ZodB in the Almagest, since the numer-

Star group Indication of G P ?. PG pg

Gmin r m it x min

Zodiacal constellations
Aries Z1 19.7 17.2 17.2 45.5 45.5 45.5
Taurus Z2 23.2 18.1 20.2 27.6 41.4 41.4
Gemini Z3 17.8 10.5 10.6 29.4 82.4 82.4
Cancer Z4 13.8 4.3 4.5 33.3 100.0 100.0
Leo Z5 20.2 11.1 11.1 19.2 65.4 65.4
Virgo Z6 18.4 13.6 14.4 39.1 56.5 52.2
Libra Z7 8.4 6.1 6.1 83.3 83.3 83.3
Scorpio Z8 18.8 13.7 13.7 30.0 65.0 70.0
Sagittarius Z9 16.4 14.3 14.4 30.4 60.9 56.5
Capricorn Z10 16.2 10.6 10.6 42.3 65.4 65.4
Aquarius Z ll 28.6 17.3 18.7 18.4 44.7 47.4
Pisces Z12 22.5 21.5 21.7 51.7 41.4 37.9

Table 6.7. Calculation result for the zodiacal constellations o f the Almagest. Here af nit, a ifl , a f represent the square average
latitudinal discrepancies in group G - the initial and the remaining, as well the one that we come up with after compensating
the systematic error in G as estimated for ZodA with the optimal choice o f parameter (p. We also cite the stellar percentage
values o f Pfmt, P ^in , P f , as calculated after similar compensation, with a minimal latitudinal discrepancy o f 10'.

ical material that we have at our disposal doesnt per tions o f Gemini, Cancer, Leo, Virgo, Libra, Scorpio,
mit anything in the way o f an unambiguous statis Sagittarius and Capricorn, as well as the environs o f
tical conclusion. Antares, Acelli, Arcturus, Spica and Regulus, hap
Corollary 3. The precision o f star coordinate pens to be close to the systematic error o f y sffA,
measurements is much higher for A and ZodA than which is characteristic for ZodAythe part o f the sky
it is in case o f any other celestial region. measured best in the Almagest, and might even co
Corollary 4 . The residual square average latitu incide therewith.
dinal discrepancy for celestial region ZodA equals Corollary 7. Nothing definite can be said about
12.8' in the Almagest. About 2/3 o f all stars from this the values o f group errors made by the compiler o f
part o f the sky have the latitudinal discrepancy o f less the Almagest in cases o f Aries and Taurus. They may
than 10 ', which make them fit the declared 10 ' pre coincide with the errata discovered for ZodA or be dif
cision margin of the Almagest catalogue. Correspond ferent from their values. The errata in the environs o f
ing values for celestial region A equal 16.5' and 1/2. the named stars Cappella and Vega cannot be calcu
Corollary 5. A study o f the Zodiacal constella lated, either.
tions and the environs o f named stars in the Almagest Corollary 8. Group errors in the environs o f
makes it possible to conclude that parameter y, which Sirius and Aquila differ from the error that is char
stands for the error in the angle o f the ecliptic, is a acteristic for celestial region Z odA . However, we
systematic error. As for parameter <p, it may well be a havent managed to calculate the values o f these er
squared value o f group or individual errors. rors. The group error made for the constellation o f
Corollary 6 . Group error y for the constella Pisces is also likely to differ from yf^f\

The dating of the Almagest star

catalogue. Statistical and geometrical

1. measurements o f most stars coordinates have always

THE CATALOGUE'S INFORMATIVE KERNEL been based on the so-called reference stars, whose
CONSISTS OF THE WELL-MEASURED number is rather small as compared to the total num
NAMED STARS ber o f the stars in the catalogue.
Let us begin by reiterating a number o f consider
The analysis of the Almagest star catalogue related ations voiced in the preceding chapters, which will
in Chapters 2-6 had the objective o f reducing latitu serve as a foundation o f our dating method.
dinal discrepancies in star coordinates by compen Unfortunately, we do not know which reference
sating the systematic error as discovered in the cata star set was used by the author o f the Almagest. All
logue. we do know is that it must have included Regulus
As a result, we have proven that the Almagest com and Spica, since the measurement o f their coordi
pilers claim about the precision margin o f his meas nates is discussed in separate dedicated sections o f
urements being less than 10' is justified - insofar as the Almagest. However, it would make sense to as
the latitudes o f most stars from celestial area A are sume that the compiler o f the catalogue was at his
concerned, at least. We believe this circumstance to most accurate when he measured the coordinates o f
be of paramount importance. named stars. As it was stated above, there are twelve
However, we can only date the Almagest catalogue such stars in the Almagest: Arcturus, Regulus, Spica,
by considering fast and a priori precisely measurable Previndemiatrix, Cappella, Lyra = Vega, Procyon,
stars. In other words, dating purposes require indi Sirius, Antares, Aquila = Altair, Acelli and Canopus.
vidual error estimates. Our statistical characteristics The identity o f Ptolemys reference stars (as used
can tell us nothing about the precision o f actual star for planetary coordinate measurements) is an issue
coordinate measurements or the stars measured with that was studied in [ 1120]. The stars in question turn
the greatest precision. out to be as follows (Ptolemy actually mentions them
The choice of such stars can only be defined by rea as ecliptic reference stars): Aldebaran = a Tau, Regu
sonable considerations based on known practical lus, Spica and Antares. Three o f them have proper
methods of measuring stellar coordinates as used by names in the Almagest - namely, Regulus, Spica and
the ancients (see Chapter 1). It is known that the Antares. Apparently, Ptolemy also had to add Aldeb-

aran to their number for the purpose o f planetary It would make sense to put forth the following hy
observations. Incidentally, all four stars are included pothesis. If the precision rate claimed by the compiler
in our table 4.3. o f the catalogue was actually true, it is guaranteed to
The twelve named stars of the Almagest are bright, manifest as such in the case o f the catalogue's in
clearly visible against their background and provid formative kernel after the compensation of the group
ing a convenient basic set o f reference points on the error.
celestial sphere. The most important circumstance is This is the very hypothesis that out method o f
that a sufficiently large part o f these stars are charac dating star catalogues relies upon.
terised by substantial proper motion rates, especially However, the fact that the informative kernel o f a
Arcturus, Procyon and Sirius. star catalogue has the ability to assist us with the dat
Seven o f the named Almagest stars are located in ing o f the latter is far from obvious. In general, the
celestial area Zod A or its immediate vicinity. They are fact that we did manage to reconstruct the true val
as follows: Arcturus, Spica, Procyon, Acelli, Previnde- ues o f random errors inherent in the Almagest cata
miatrix, Regulus and Antares. Nine of the named stars logue by group error compensation does not imply
surround area A - the above set needs to be comple that the individual errors in the coordinates of the cat
mented by Lyra = Vega and Cappella. Thus, even if alogue's kernel stars are the same. It doesn't seem too
these 12 stars weren't used for reference, their coor likely that a discrepancy of this sort actually exists -
dinates are still most likely to have been measured the central star o f a group appears to have the same
with sufficient precision. sort of error in its coordinates as its closest neigh
However, despite the probable high precision of bours. However, strictly speaking, the hypothetical
their coordinates as measured in the Almagest, the existence of such a discrepancy has to be taken into
stars comprised in this group are by no means o f account nonetheless. Apart from that, one mustn't
equal importance. Our analysis has revealed the fol rule out the possibility that the coordinates o f a star
lowing: included in the catalogue's informative kernel were
1) Canopus is located far in the south, and meas measured with an error margin o f more than 10'.
urement precision is greatly affected by refraction in All o f the above tells us that if we do manage to
such cases. Therefore, all efforts o f the Almagest's find a moment in time conforming to the require
compilers notwithstanding, the coordinates o f this ments o f our hypothesis, we shall once again prove
star as given in the catalogue are a priori known to the correctness o f our initial statistical conjectures.
be more than one degree off the mark.
2) The coordinates o f Previndemiatrix as meas 2.
ured by the compiler o f the Almagest remain un PRELIMINARY CONSIDERATIONS IN RE
known to us - we are only familiar with results of later THE DATING OF THE ALMAGEST CATALOGUE
3) Group errors in the environs o f Sirius and OF NAMED STARS
Aquila fail to concur with the errata inherent in the
coordinates o f all the other stars, as we have discov In section 1 we singled out a group of stats that we
ered in Chapter 6. We are incapable o f calculating the have called the Almagest's informative kernel. We shall
rates of these errors, and, consequently, compensation consider its behaviour in detail below. What we shall
is a non-option in their case. analyse herein is the behaviour o f all 12 named
Thus, we end up with 8 named stars that we can Almagest stars at once. This preliminary study demon
use for the purpose o f dating. The stars that surround strates perfectly well how much greater the precision
them have a single group error in their coordinates rate o f the Almagest catalogue becomes after the com
- at least, the y component o f this error is the same pensation o f its systematic error. It also provides ad
in each and every case. We shall be referring to these ditional explanation to the fact that three named stars
stars as to the informative kernel of the Almagest cat out o f twelve (Canopus, Sirius and Aquila = Altair)
alogue. break the homogeneity of the entire sample. We learn
i 66 I h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

The name o f a star and Years

the respective Bailey's number 1800 A .D . 1400 A .D . 900 A .D . 400 A .D . 100 A .D . 200 b . c .
Arcturus (110) 37.8 21.2 0.9 19.3 31.4 43.3
Sirius (818) 23.6 18.3 11.7 5.1 1.2 2.6
Aquila = Altair (288) 8.6 9.4 10.5 11.8 12.6 13.4
Previndemiatrix (509) 13 14.3 15.8 17.1 17.8 18.4
Antares (553) 32.6 29.5 25.5 21.6 19.3 17
Aselli (452) 30.5 28.5 25.9 23.2 21.5 19.8
Procyon (848) 11.2 16 21.9 27.6 31.1 34.4
Regulus (469) 17.5 16.6 15.4 14 13 12.1
Spica (510) 2.4 0.7 1.3 3.1 4.2 5.2
Lyra = Vega (149) 15.4 14.2 12.5 10.8 9.8 8.7
Capella (222) 21.9 21.7 21.3 21 20.8 20.6
C anopus(892) 51 54.2 58.2 62.3 64.8 67.3

Table 7.1. Latitudinal discrepancies o f the 12 named Almagest stars and their dependency on the presumed dating. The systematic
error discovered in the Almagest catalogue isnt compensated herein.

these stars to be rejects in relation to all the other 30' or 40'. It is peculiar that the brightest and most vis
named stars. Below in our study of all 12 named stars ible star o f the Northern hemisphere would be ob
as a whole we shall be using the coordinates o f served by either Ptolemy or Hipparchus this much
Previndemiatrix from [1339] which were apparently worse than all the other stars. Furthermore, the text
calculated by Halley. We shall use A B ft, y, cp) for re o f the Almagest implies that the coordinates o f Reg-
ferring to the difference between the latitude of star i ulus were measured several times during the compi
from the informative kernel o f the Almagest after the lation o f the catalogue, and that the star in question
compensation o f the systematic error (y, cp) and the is known to have been one o f the referential points for
true latitude as calculated for epoch t. the measurement of all the other stars in the cata
C o n s i d e r a t i o n i . Let us observe the correlation logue. It would be natural to expect that Ptolemy had
between the latitudinal precision o f the named stars been exceptionally careful in his measurement of this
coordinates in the Almagest with the grade value of star; therefore, its latitudinal discrepancy shall be less
the catalogue equalling 10', assuming that the latter than 10'. Let us point out that for another bright star
contains no global systematic errors. Table 7.1 contains on the ecliptic - namely, Spica, whose coordinates had
the absolute latitudinal discrepancy values o f all 12 also been measured by Ptolemy during the initial stage
named Almagest stars depending on the alleged dat to be used for reference later (see Chapter VII.2 o f the
ing t. In the first column we see the given stars Alma Almagest, or [1358]), has a latitudinal discrepancy of
gest number (in Baileys numeration). The rates o f 5?- less than half the catalogue grade value.
latitudinal discrepancies are given in arc minutes. Let us now consider the systematic error that we
Table 7.1 demonstrates that for 7 out of 12 named discovered in the Almagest (see Chapter 6). As the y
Almagest stars the latitudinal discrepancy exceeds the compound o f this error only varies slightly over the
limit of 10'. The columns that correspond to 100 a . d . , interval between the beginning of the new era and the
which is the Scaligerian dating o f the Almagest (Ptol middle ages, and the variations o f the cp value also
emys epoch) or 200 b . c . (the epoch of Hipparchus) hardly affect the picture, we shall use the values y0 =
draw our attention primarily because o f the outra 21?, cp0 = 0. The value y0 = 21? is the average value of
geous error in the coordinates o f Arcturus - around y(t) for t from the a priori known interval.

The name o f a star and Years

the respective Bailey's number 1800 A .D . 1400 A .D . 900 A .D . 400 A .D . 100 A .D . 200 b . c .
Arcturus (110) 29.9 15.5 2.3 20 30.5 41
Sirius (818) 44.2 39.2 32.7 25.9 21.8 17.5
Aquila = Altair (288) 27 28.7 30.7 32.5 33.5 34.4
Previndemiatrix (509) 15.6 14.9 13.8 12.6 11.8 11
Antares (553) 13.3 11 8.5 6.2 4.9 3.7
Aselli (452) 13.2 10.2 6.5 2.9 0.9 1.1
Procyon (848) 8.1 4 1.2 6.7 10.1 13.5
Regulus (469) 6.1 3.5 0.4 2.7 5.1 6.2
Spica (510) 5.1 4.9 4.4 3.7 3.3 2.7
Lyra = Vega (149) 5.1 6.7 8.5 10 10.8 11.5
Capella (222) 1.3 1.5 2.1 2.9 3.5 4.2
C anopus(892) 71.5 75 79.2 83.1 85.4 87.6

Table 7.2. Latitudinal discrepancies of the 12 named Almagest stars and their dependency on the presumed dating as given after
the compensation of the systematic error in the Almagest stellar coordinates specified by parameters y0 = 21' and % = 0.

We shall proceed with building the table num to study other values of parameters y and cp. The pres
bered 7.2, which is similar to 7.1, the sole difference ent chapter contains extensive calculations and more
being that the systematic error defined by parameters explicit statements below.
y0 = 2 T and (p0 = 0 in all the stellar coordinates is now C onsideration 2. The following line o f argu
taken into account and compensated in the calcula mentation might provide additional information per
tion o f latitudinal discrepancies. tinent to the dating of the Almagest catalogue. Let us
A comparison o f the two tables demonstrates the consider the latitudinal discrepancies A y , (p) o f
precision characteristics o f the named Almagest star a certain Almagest star set , 1 < i < n as a whole for
coordinates to have improved drastically for all pos each moment t and all the values o f y and cp.We shall
sible datings after the compensation o f the systematic use them for building empirical function graphs of
error. The latitudes o f Regulus and Spica prove to be latitudinal error distribution for star set E: Ft y 9 (x) =
measured with the precision rate of up to 5' for every (1 / n) # { i : Y><p)|^ * }, where n represents the
alleged dating between the beginning o f the new era quantity of stars in set E. A comparison o f these dis
and the end o f the Middle Ages. This correlates well tribution functions for different values o f parame
with the fact that these two stars enjoy a great deal o f ters t, y and cpcan allow us to try finding such a com
attention in the text o f the Almagest - qv in the book bination o f these values that will minimize the lati
itself, Chapter VII.2 ([1358]). Moreover, if we are to tudinal errors o f set E stochastically. The error
place the dating on the interval o f 6 < t < 10, or 900- difference rate for different values o f t, y and cpshall
1300 a.d ., the latitudinal discrepancy does not ex be their average difference value. We can obviously
ceed 10', or the catalogue scale grade value, for 8 come to no quantitative conclusions so far since we
named stars out o f 12 - the ones located in celestial only have 12 observations at our disposal, and we
area A which we discovered in Chapter 6 as we were shall thus be merely referring to the qualitative pic
analyzing the entire stellar aggregate o f the Almagest ture as a first approximation.
catalogue. The error difference rate in question can be rep
It goes without saying that the above considera resented as the area contained between the distribu
tions need to be more explicit. In particular, we have tion graphs of the functions Fn>Y1>91(x) and Y2)<p2(x)
i6 8 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

proves to be the optimal dating o f the catalogue (ap

proximately 1550 a . d . ) . Indeed, this is the very dat
ing for which the errors in the 13 named catalogue
stars shall be minimal in the abovementioned sense.
The date 1550 a . d . is really close to the known epoch
when Tycho Brahes catalogue was compiled, namely,
the second half o f the XVI century.
Let us provide a list o f these 13 stars from Tycho
Brahes catalogue. First and foremost, they are Regu-
lus, Spica, Arcturus, Procyon, Sirius, Lyra = Vega, Ca-
pella, Aquila and Antares, which are also included in
the list o f the named stars from the Almagest. Apart
from that, there are four more stars: Caph = P Cas, De-
nebola = P Leo, Pollux = P Gem and Scheat = P Peg.
We shall now consider the empirical distribution
functions Ft y ^ for star set E that consists of 12 named
Almagest stars (see section 1). In fig. 7.3 one sees the
graphs o f these functions for t = 5, or 1400 a . d .,
t = 10, or 900 a .d ., t = 18, or 100 a .d ., and t = 20, or
100 b . c . with varying values o f y. The value o f cp is
Fig. 7.1. Empirical functions o f error distribution in stellar considered to equal zero everywhere, since the gen
latitudes. eral picture is hardly affected by cp variations. The
values t = 10, or 900 a .d ., and y= 2T are optimal -
that is to say, they generate the least serious errors.
as drawn on a single draft. Both areas contained be The resulting graphical representations o f the
tween the graphs have to be taken with either a plus functions Ft y 9 for the Almagest isnt very sensitive
or a minus depending on which function we find to to changes in the contingent o f the named stars. Let
the right and to the left o f the area in question (see us cite the empirical distribution functions for all 13
fig. 7.1). The distribution function Ffo, m poM that is stars which were used in the Tycho Brahe example,
located to the left o f all the other functions Ft y ^ on having taken the coordinates from the Almagest this
the average corresponds to minimal latitudinal error time, qv in fig. 7.4. The values o f t = 10 and y= 21'
rates for set E. It would be natural to consider the remain optimal for this star list as well. In fig. 7.4 one
dating t0and the systematic error value (y0, cp0) as aP- can clearly see the difference between the values o f
proximations to the real observation date and the real y= 21' and y= 0 already pointed out above - namely,
systematic error as made by the observer. that all the graphs corresponding to y= 21' taken as
Let us illustrate the above with the example of an a whole are located to the left o f the graphs built for
other famous star catalogue dating to the second half y= 0 in general, indicating the lower error rate o f the
o f the XVI century and compiled by Tycho Brahe. former as opposed to the latter. In other words, the
The informative kernel that we shall be using is com value o f y= 2 1' is better than y= 0 for all the t dates
prised o f 13 named stars from Tycho Brahes cata from the a priori chosen interval.
logue. We have calculated the empirical distribution Consideration 3. Let us conclude with discussing
functions Ft y <pfor y = cp = 0 and three different val the issue o f just how possible it is to expand the list
ues o f 1 1 = 3 (1600 a . d . ) , t = 3.5 (1550 a . d . ) and t = 4 of the named Almagest stars used as a basis for proper
(1500 a . d . ) . The result can be seen in fig. 7.2. This il movement dating. Yet the coordinate precision o f this
lustration demonstrates quite well that without con expanded list (latitudinal at least) may by no means
sidering the possibility o f a systematic error inherent deteriorate. The first impression one gets is that the
in Tycho Brahes catalogue (y = cp = 0) epoch t = 3.5 most natural way to extend the list would be includ-

ing all the stars which have names of

their own nowadays into it (see table
P I.2 in Annex 1). Most stars received
names in the Middle Ages, but this
practice continued into the XVTI-XIX
century. It is possible that many o f
them were particularly significant for
the Almagest catalogue compiler. We
shall proceed to select just those stars
from table P I.2 (from Annex 1) whose
names are capitalized in the exact
same manner as [1197] does it; such
are the most famous o f the named
stars. Their number is 37, qv listed in
table 7.3.
However, it turns out that such an
expansion o f the Almagest's inform
ative kernel drastically reduces the Fig. 7.2. Empirical distribution functions for Tycho Brahes catalogue; the optimal
sample's coordinate precision, and we value of t0 = 3.5.
are particularly concerned about the
latitudes being affected. Let us con
sider the expanded kernel that con
tains 37 Almagest stars as listed in
table 7.3. Fig 7.5 demonstrates how
the mean-square discrepancy behaves
for these 37 stars depending on the al
leged dating o f the Almagest. Having
calculated this discrepancy, we would
allow for the variation o f the system
atic error's calculated rate to fluctu
ate within 5 with the step value of
1 minute for parameter y and within
30' with the step value of 1 minute for
parameter p. The resultant graph
demonstrates that although the min
imum is reached around 400 a . d . , it Fig. 7.3. Empirical distribution functions Ff y ^ for the 12 named Almagest stars.
is very inexplicit. The minimal mean- The value of cp equals zero in every case.
square value roughly equals 18 min
utes. If we are to allow for a variation
o f this value within a two-minute range, or a mere Furthermore, this vague picture is what we get in our
10%, we shall end up with a dating interval o f 1800 analysis of the latitudes, which are more precise in the
years, no less - between 600 b . c . and 1200 a . d . It is Almagest catalogue, as we know. The longitudinal
perfectly obvious that this result is o f no interest to picture is even vaguer.
us, the reason being that the average precision o f In figs. 7.6 and 7.7 one sees the dependency graphs
Ptolemy's calculations is too low for the 37-star list for the quantity o f stars in the extended kernel whose
under consideration. It is clearly insufficient for the calculated latitudinal error does not exceed 10 and 20
dating o f the catalogue by proper star movements. minutes respectively and the presumed dating o f the
170 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

No by Stellar Stellar Modern name o f the star and its ancient proper
BS4 Bailey's magnitude V&0900) V5(1900) magnitude name as specified in caps in the Bright Stars
and number according to [1197] [1197] according to Catalogue ([1197])y which indicates that the star
BS5 BS5 the Almagest in question was known very well in the past
5340 110 -0.04 -1.098 -1.999 1 16Alp Boo (ARCTURUS)
1708 222 0.08 +0.080 -0.423 1 13Alp Aur (CAPELLA)
3982 469 1.35 -0.249 +0.003 1 32Alp Leo (REGUL)
2943 848 0.38 -0.706 -1.029 1 lOAlp CMi (PROCYON)
5056 510 0.98 -0.043 -0.033 1 67Alp Vir (SPICA)
6134 553 0.96 -0.007 -0.023 2 21Alp Sco (ANTARES)
7001 149 0.03 +0.200 +0.285 1 3Alp Lyr (LYRA=VEGA)
3449 452 4.66 -0.103 -0.043 4-3 43Gam Cnc (ASELLI)
15 315 2.06 +0.137 -0.158 2-3 21 Alp And (ALPHERATZ)
21 189 2.27 +0.526 -0.177 3 llB et Cas (CAPH)
188 733 2.04 +0.232 +0.036 3 16Bet Cet (DENEDKAITOS=DIPHDA)
337 346 2.06 +0.179 -0.109 3 43Bet And (MIRACH)
617 375 2.00 +0.190 -0.144 3-2 13Alp Ari (HAMAL)
1231 781 2.95 +0.057 -0.110 3 34Gam Eri (ZAURAK)
1457 393 0.85 +0.065 -0.189 1 87Alp Tau (ALDEBARAN)
1791 400 1.65 +0.025 -0.175 3 112Bet Tau (ELNATH)
2491 818 -1.46 -0.545 -1.211 1 9Alp CMa (SIRIUS)
2890 424 1.58 -0.170 -0.102 2 66Alp Gem (CASTOR)
2990 425 1.14 -0.627 -0.051 2 78Bet Gem (POLLUX)
4057 467 2.61 +0.307 -0.151 2 41Gaml Leo (ALGIEBA)
4301 24 1.79 -0.118 -0.071 2 50Alp UMa (DUBHE)
4534 488 2.14 -0.497 -0.119 1-2 94Bet Leo (DENEBOLA)
4660 26 3.31 +0.102 +0.004 3 69Del UMa (NEGREZ)
4905 33 1.77 +0.109 -0.010 2 77Eps UMa (ALIOTH)
4914 36 5.60 -0.238 +0.057 3 12Alpl CVn (COR CAROLI)
5054 34 2.27 +0.119 -0.025 2 79Zet UMa (MIZAR)
5191 35 1.86 -0.124 -0.014 2 85Eta UMa (ALKAID)
5267 970 0.61 -0.020 -0.023 2 Bet Cen (AGENA)
5793 111 2.23 +0.120 -0.091 2-1 5Alp CrB (ALPHEKKA)
5854 271 2.65 +0.136 +0.044 3 24Alp Ser (UNUKALHAI)
6556 234 2.08 +0.117 -0.227 3-2 55Alp Oph (RASALHAGUE)
6879 572 1.85 -0.032 -0.125 3 20Eps Sgr (KAUS AUSTRALIS)
7557 288 0.77 +0.537 +0.387 2-1 53Alp Aql (ALTAIR)
7602 287 3.71 +0.048 -0.482 3 60Bet Aql (ALSHAIM)
8162 78 2.44 +0.150 +0.052 3 5Alp Cep (ALDERAMIN)
8728 1011 1.16 +0.336 -0.161 1 24Alp PsA (FOMALHAUT)
8775 317 2.42 +0.188 +0.142 2-3 53Bet Peg (SCHEAT)

Table 7.3. A list of fast stars possessing old names of their own according to BS4 ([1197]), all transcribed in capital letters as the
most famous stars in the Middle Ages. All the celestial areas are represented here. The list is preceded by the 8 stars from the in
formative kernel of the Almagest, some of which dont rank among the fast stars.

Almagest. The error was calculated

after the compensation o f the sys
tematic error y = 20. We observe fluc
tuations within a more or less con
stant value range for the entire time
interval under study. A 10-minute lat
itudinal range covers 3-13 stars in
various years, whereas about 11-16
stars wind up within the 20-minute
range. These graphs give us no reli
able inform ation concerning the
most probable dating o f the cata
In fig. 7.8 we cite the mean-aver
age discrepancy dependency graph
similar to the graph in fig. 7.5. How
ever, the only stars we took into con
sideration this time were the ones that Fig. 7.4. Empirical distribution functions for the 13 bright named Almagest stars
got a latitudinal discrepancy o f less with t - 1 ,5,10,15 and 20. Continuous lines: 7 = 21 '; dotted lines: 7 = 0.
than 30 minutes for a given dating.
One sees the graph to consist o f gen
tly sloping parabola segments whose minimums fall 3.
on different years on the time axis. Thus, various THE STATISTICAL DATING PROCEDURE
parts o f the 37-star list contain the valleys o f respec
tive parabolas scattered all across the historical in 3.1.The description of the dating
terval. procedure
The discovered instability o f the valleys tells us
that this dating method is a very imprecise one due The hypothesis about the named stars of the Alma
to the fact that the valleys o f many parabolas are sit gest measured in correspondence with the aberra
uated at a considerable distance from the catalogue tion rate o f 10 minutes allows us to give a rather ap
compilation date. Therefore, a variation of the stel proximate real dating o f the Almagest in section 2. We
lar contingent shall distribute these valleys chaoti proved that the configuration o f the Almagest cata
cally over the entire length o f the historical interval. logue informative kernel varies over the course o f
In general, the graph in fig. 7.8 has its extremely time at a high enough speed for us to determine the
poorly-manifest valley fall on the period o f 700-1600 catalogue compilation date. Therefore one finds it
a.d ., which is o f zero use for a reliable dating. makes sense to set the problem of estimating the pos
We have also considered other possibilities o f ex sible dating interval.
panding the Almagests informative kernel - for in The following procedure that we shall refer to as
stance, using stellar luminosity as a criterium. Nearly statistical appears to be o f the most natural and ob
all o f them led to a drastic decrease in stellar coor vious character; it is based on the hypothesis that the
dinate precision and what can be de facto regarded named Almagest stars were measured with a declared
as eliminating o f dependency between the dating o f 10-minute latitudinal precision. Furthermore, we
the observations and the extended list characteristics. shall base our research on the statistic characteristics
However, it turns out that the informative error does o f group errors as rendered in Chapter 6. The statis
in fact allow a natural expansion without a drastic tical dating procedure is as follows:
precision decrease. This issue is considered in detail A) Let us specify the confidence level 1 - 8 .
below. B) Now we shall consider time moment t and trust
172 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

interval Iy(e) for the compound y5f^ ( t ) o f the group (p) > 20' is true. On every drawing the parameters
error in area Zod A. Now to estimate the value are marked by a large dot.
The diagrams demonstrate that the spot with
A(f) = min A(f, y, q>), (7.3.1) double shading that corresponds to the maximal lat
itudinal discrepancy of 10' for the eight named Alma
where the minimum is taken for all y in JY(e) with gest stars only exists for time moments falling into the
varying (p values, while the value o f range o f 6 < t < 13, or the interval between 600 and
1300 A . D .
A(f, y, (p) = max |AB,(t, y, <p)| The area with normal shading that corresponds to
the maximal latitudinal discrepancy o f 15' only exists
defines the maximal discrepancy for all the stars from for 4 < t < 16. Maximal sizes of these areas are reached
the informative kernel as calculated for the presumed at 7 < t < 12. For t > 18 the acceptable interval alter
dating t. Parameters (y, (p) define a certain turn o f a ation area defined by correspondent confidence in
celestial sphere - quite arbitrarily so, as a matter of tervals contains no points where A(t, y, (p) < 20'. In
fact, qv in fig. 3.14. particular, this is true for the Scaligerian dating of the
C) If the educed value o f A(t) does not exceed the epochs when Ptolemy and Hipparchus lived.
declared catalogue precision rate o f 10, time m o Furthermore, when we attempt to date the Almagest
ment t should be regarded as the possible catalogue catalogue to 100 a . d . or an earlier epoch, the latitu
compilation date. Otherwise the catalogue cannot be dinal discrepancy minimax A(t) turns out to be two
dated to epoch t. times greater than the declared 10-minute precision
Quite obviously, the result o f applying this dating o f the Almagest catalogue. For datings preceding 100
procedure depends on the subjective choice o f trust a . d . the value o f A(t) exceeds even the mean-average

level 1 - e. Therefore its stability shall have to be tested residual error for the stars from areas A, Zod A, B and
against the variations o f 8, which is carried out below. Zod By being close to the square average residual Al
magest error for celestial area Ai, or rather dim stars
3.2. The dependency of the minimax o f the Milky Way (where the observations o f such
discrepancy A on t, y and cpfor the Almagest stars were complicated by the abundant stellar back
ground which would impair their precision making
We shall draw a graph for 8 of the named Almagest its rate unacceptably low for the bright named stars).
stars comprising the informative kernel to represent One therefore has to reject the dating of the Almagest
the dependency o f the minimax latitudinal discrep to the epoch of roughly 100 a . d . or earlier as contra
ancy A(f, y, (p) on all three variables. This dependency dicting the Almagest catalogue.
is shown as a sequence o f diagrams in figs. 7.9 and Thus, figs. 7.9 and 7.10 demonstrate that the area
7.10. Every diagram here corresponds to some fixed permitted by the values of y and (p fundamentally gives
moment t. The diagrams are given for t = 1 , . . . , 18. us no opportunity o f making the latitudinal discrep
For other t values the respective diagrams prove void, ancy o f all 8 stars comprising the Almagests inform
as is the case with t - 1. Let us remind the reader that ative kernel less than 10' for epochs preceding 600
t = 1 corresponds to 1800 a . d . , and t = 18 - to the be a.d. If we are to raise the error rate threshold to 15',
ginning of the new era. The horizontal axes o f the di the earliest possible dating o f the Almagest is 300 a.d.
agrams bear the values of y, and the vertical - the
values o f (p. 3.3. Results of dating the Almagest catalogue
Double shading marks the areas for which A(t, y, statistically
q>) < 10.
Shaded areas correspond to 10' < A(t, y, (p) < 15. Let us assign variation area St(a ) o f parameter y
The area filled with dots corresponds to 15' < A(f, in the following manner:
y, (p) < 20'.
For the rest o f the drawings, the expression A(f, y, St( a ) = { y : min A(t, y, <p) < a }

Plot o f N Jn _ep s vs t

8" ai
e 1 1*
H t - E M~a

4200 -800 -400 0 400 800 1200 1600 2000 -1200 -800 -400 0 400 800 1200 1600 2000
t t

Fig. 7.5. The square average discrepancy for the 37 Almagest Fig. 7.7. Vertical axis: number of Almagest stars from the list of
stars listed in Table 7.3 as the presumed dating function. The 37 (qv in Table 7.3) whose latitudinal discrepancy doesnt
systematic error y of the Almagest catalogue was compensated exceed 20'. Horizontal axis: presumed dating of the catalogue.
in the calculation of the discrepancy. Apart from that, the de
sired square average discrepancy was minimised in accordance
with the variations of y = ystat 5'; p = 0 30'. Plot o f sigma vs t

4100 -600 400 400 900 40G 1900 Fig. 7.8. The square average deviation for the 37 Almagest stars
t listed in Table 7.3, whose latitudinal discrepancy doesnt exceed
30 minutes for the presumed dating in question. The graph is
Fig. 7.6. Vertical axis: the number of Almagest stars from the built as a function of the presumed Almagest dating. In the
list of 37 (qv in Table 7.3) whose latitudinal discrepancy search of the discrepancy, the catalogues systematic error y was
doesnt exceed 10 minutes. Horizontal axis: presumed dating compensated. Apart from that, the square average discrepancy
of the Almagest catalogue. was minimised by the variations of y = ystat 5'; (5 = 0 30'.

Set St(a ) may yet turn out empty. Let us consider be referred to as the possible dating interval o f the
the intersection o f set St(a ) and the confidence in Almagest catalogue.
terval Iy(e) built around the value o f yft{^ ( t ) . If this The result o f calculating St(a ) for the Almagest is
intersection isnt empty, we can declare moment t to represented graphically in fig. 7.11. The dots fill the
be the possible epoch o f the Almagest catalogues union o f sets St{a ) for a = 1 0 . The surrounding out
compilation in accordance with the statistical dating line corresponds to the value a = 1 5 . We shall find a
procedure. All o f such moments t taken as a whole can use for it later.
17 4 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

22 24 26 ZB 3 0 3 * 34 f* 2t 23 25 27 29 $t 33 29 22 24 29 23 39 32 rf

Fig. 7.9. One sees the dependency A(f, y, q>) for the time values t beginning with 1, or 1800 a . d . , and ending with t = 18, or
100 b . c . The area with double shading corresponds to A < 10'. The area with single shading corresponds to 10' < A < 15'. The
area filled with dots corresponds to 15' < A < 20'. The large dot corresponds to parameter pairs of y ^ ^ M ,

The graph of the function y u s e d herein was 3.4. The discussion of the result
calculated in Chapter 6 (see fig. 6.8). The values o f
trust intervals /y(e) that correspond to different val The length o f the possible catalogue dating inter
ues o f can be found in table 6.3. Fig. 7.11 implies val we ended up with equals 700 years: 1300 - 600 =
that the possible dating interval is the same for = 0.1, 700.
= 0.05, e = 0.01 and = 0.005 - namely, 6 < t < 13. The interval is a rather large one for a number of
If we are to translate the resultant dating result reasons. We already named the first one - the low
into regular years, we shall see that the possible dat precision o f the Almagest catalogue, even if we are to
ing interval in the Almagest catalogue begins in 600 accept Ptolemys declared precision o f 10'.
a . d . and ends in 1300 a . d . Such low precision makes it impossible to date the

Fig. 7.10. The previous figure continued.

catalogue to a narrower time interval since even the This approach obviously leads to the broadening
fastest o f the named stars under study (Arcturus) al o f the Almagest catalogue dating interval. Indeed, if
ters its latitude by a mere 10' every 260 years. we could consider cp to be a group error like y, we
The value is great, and it is greater still for other would select parameter cp from the confidence strip.
kernel stars. This would raise the value o f min A(t, y, cp) and thus
The second reason stems from the fact that we narrow the possible dating interval.
have only used the trust intervals o f the group errors However, as it has been pointed out above, we do
y compound, having minimized the value A(f, y, cp) not have enough reasons to consider cp a group error
by various possible values o f cp, qv in formulae 7.3.1 in stellar groups from the Almagest that we have
and 7.3.2. studied.
176 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

the exact method used by Ptolemy in order to meas

ure stellar latitude. It is known quite well in history of
astronomy that such measurements were conducted
with bright basis stars used as a framework of sorts
which the desired stellar positions would be educed
from in all the measurements to follow. The coordi
nates of these stars would be measured with the ut
most precision and used later on. Ptolemy does not
specify the exact stars that he used for basis; as we can
see from the text of the Almagest, such basis stars have
at least been Regulus, Spica, Antares and possibly Al-
debaran (see page 247 of [1120], for instance). Three
o f them - namely, Regulus, Spica and Antares - have
names o f their own in the Almagest that employ the
formula vocatur . . . (named . . . ), qv above. We
formulated the idea that the named stars o f the Alma
gest received names because they served as the basis
for Ptolemys observations in the first place. This idea
is confirmed by the fact that, as we have proved, the
named stars o f the Almagest really possess the Ptole
maic reference precision o f 10' (insofar as the lati
tudes are concerned, at least) in areas A, Zod A, B and
Zod B . This isnt true for the longitudes, but we already
mentioned that it is a great deal more difficult to ob
serve the longitudes than the latitudes. Apart from
that, longitudinal precision was most probably lost
Fig. 7.11. Result of the statistical dating procedure as applied when the Almagest catalogue had been re-calculated
to the Almagest catalogue and using its eight named stars. in order to correspond to other epochs. Therefore the
latitudes cannot serve as a criterion o f Ptolemys real
precision. It is only the latitudes that one can rely
4. upon for this purpose.
DATING THE ALMAGEST CATALOGUE BY We could prove none of the above for other celes
THE EXPANDED INFORMATIVE KERNEL tial areas, since the systematic error rates could not be
established reliably. Therefore we shall refrain from
The issue o f expanding the informative kernel of going beyond celestial areas A, Zod A, B and Zod B in
the Almagest has been discussed above at the end of our search for possible informative kernel extensions.
section 7.2. It was discovered that if we expand the Let us ask about what other stars except for the
kernel choosing bright and fast stars for this purpose basis ones - the top ranking stars, that is, would also
without following any system, we cannot get an in be measured very well by Ptolemy? Quite naturally, the
formative dating. We already understand that this is ones located in the immediate vicinity o f the basis
explained by the low average precision o f Ptolemys stars - the primary reason being that Ptolemys coor
measurements, and this concerns even the bright dinates are most likely to have followed links of sorts,
stars. The question o f what principle one could use when the coordinates o f the stars close to the basis
in order to expand the 8-star informative kernel o f the ones would be measured first, and he would proceed
Almagest without the loss o f latitudinal precision re further taking the previously-calculated coordinates
mains open. into account, step by step. Nowadays we understand
We managed to solve this problem. Let us ponder that this measurement method inevitably leads to ran

dom error dispersion growth, which means greater

coordinate measurement errors. The further a star is
from the referential kernel, the worse it shall be meas
ured on the average.
It would thus make sense to attempt an extension
of the informative kernel, adding the stars ranking
second thereto, which are bright enough, well-iden
tified and located in close proximity to the basis stars.
One would then have to proceed with the third rank
o f stars which are further away, the fourth rank
which is even further and so on. If we notice this
process to be accompanied by a slow decrease in av
-1200 -S00 -400 0 400 800 1200 1600 2000
erage latitude precision remaining virtually the same t
for the basis stars and the ones closest to them, we shall
Fig. 7.12. Square average latitudinal discrepancy graph after the
ipso facto confirm our presumption that the top
compensation of the systematic error for the eight first lever
ranking stars were really included into the basis ref
stars. These eight stars comprise the informative kernel of the
erential framework. We shall also get the opportunity Almagest catalogue. According to our calculations, these very
to extend the dating kernel o f the catalogue as well stars served as reference points in Ptolemys observations. The
as checking (and, possibly, correcting) our dating. square average discrepancy was minimised in accordance with
This idea was implemented in the following man the variations of parameter y for the interval of ystat 5', and the
variations of parameter |3 for the interval of 0 20'. The graph
ner. First o f all we would have to use nothing but the
reaches its minimum in 900-1000 a . d . , at the level of 5-6 arc
stars which have perfectly sound and dependable minutes. The discrepancy equals 12' for the Ptolemaic epoch of
Almagest identifications as well as observable proper the II century a . d . , which exceeds the minimum by a factor of
movement. They are listed in table 4.3. There are 68 two. The discrepancy for the epoch of Hipparchus (the II cen
such stars altogether. Bear in mind that the 8-star in tury b . c . ) approximately equals 14'.
formative kernel is included in this list in its entirety.
Eight information kernel stars were taken to rep
resent the top level. We have calculated the latitu
dinal mean-square aberration for all o f them after
the compensation o f the systematic error. Systematic
error y was calculated in Chapter 6. We allowed for a
fluctuation o f this errors value within the range of
5 ' with a 1-minute step. Parameter (3 would define
the excesses within the limits o f 20' with the same
step value. The mean-average discrepancy for each
presumed dating o f the catalogue would be selected
as the minimal value achieved by said variations of
parameters yand (3. The result is presented as the de -1200 -800 -400 0 400 800 1200 1600 2000
pendency graph o f the square average discrepancy of t
the presumed Almagest catalogue dating. The graph Fig. 7.13. Square average latitudinal discrepancy graph after the
built for eight o f the informative kernel stars, or top compensation of the systematic error for the nine second level
level stars, can be seen in fig. 7.12. stars located at the maximal distance of 5 degrees for the base
The graphs minimum is reached around 900-1000 ones. The square average discrepancy was minimised in accor
dance with the variations of parameter y for the interval of ystat
a . d . at the level o f 5-6 arc minutes. This means that
5', and the variations of parameter (3 for the interval of 0 20'.
the guaranteed latitudinal measurement precision for The graph reaches its minimum in 1000-1100 a . d . , at the level
Ptolemy equalled 10-15\ Indeed, all the stars o f the of 9-10 arc minutes. The square average discrepancy equals 15' at
informative kernel are measured with the precision least for the epoch of II century a . d . and the ones preceding it.
178 | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 | PART 1

-1200 -800 -400 0 400 800 1200 1600 2000 -1200 -800 -400 0 400 800 1200 1600 2000
t t
Fig. 7.14. Square average latitudinal discrepancy graph after Fig. 7.15. Square average latitudinal discrepancy graph after
the compensation of the systematic error for the twelve third the compensation of the systematic error for the fifteen
level stars located at the maximal distance of 1 0 degrees for fourth level stars located at the maximal distance of 15 de
the base ones. The square average discrepancy was minimised grees for the base ones. The square average discrepancy was
in accordance with the variations of parameter y for the inter minimised in accordance with the variations of parameter y
val of yat 5', and the variations of parameter p for the in for the interval of ystat 5\ and the variations of parameter p
terval of 0 20'. The graph reaches its minimum in 900 a . d . , for the interval of 0 20'. The graph reaches its minimum in
at the level of 1T. The discrepancy equals 14' and more for 800-900 a . d . , at the level of 10-1T. The discrepancy equals
the epoch of 1 0 0 a . d . and the ones preceding it. 12' for the epoch of 100 a . d .

o f 10' or better, as we have already observed. This is ence and thus were measured several times with the
in perfect concurrence with the scale grade value cho utmost precision. The rest of them must have been
sen by Ptolemy - 1 0 '. measured following a link from a referential star.
As for the epoch o f the II century a . d . , the dis Nevertheless, the graph we encounter in fig. 7.13
crepancy here reaches 12'. This is two times the per is still informative enough. The discrepancy graphs
missible minimal value, which makes the early a . d . minimum is reached around 1 0 0 0 - 1 1 0 0 a . d . at the
epoch completely unacceptable for the Almagest cat level o f 9-10 arc minutes. The square average dis
alogue, let alone the epoch o f Hipparchus that is crepancy is substantially greater for the epoch o f the
supposed to have preceded it, for the discrepancy II century a . d . as well as the ones preceding it. It
equals circa 14' for the II century b . c . equals 15' for 100 a . d . , which is substantially greater
All the stars from table 4.3 were taken as the sec than 150% o f the minimal value.
ond level stars which are at no further distance from The third level stars are all the stars from table
the closest informative kernel star than 5 degrees. 4.3 that are located at the maximal distance o f 10 de
There proved to be 9 such stars including the infor grees from the informative kernel. We discovered
mative kernel. It turned out that we needed to add star there to be 12 such stars including the informative
478 Cnc (#3461 in catalogues BS4 and BS5). The re kernel. Apart from 478 Cnc, the informative kernel
sultant square average discrepancy graph can be seen was expanded to include 14o Leo (#3852), 8t| Boo
in fig. 7.13. It is plainly visible that the picture drasti (#5235) and 26e Sco (#6241).
cally changes once we add a single star to the eight that The discrepancy graph is demonstrated in fig. 7.14.
comprise the informative kernel - and it is just one, It hardly differs from what we had in the previous step
which is close to them, well-visible to the naked eye, at all. This is well understood. We are still very close to
and isolated to boot. The reason is most likely to be the informative kernel, which still comprises 3/4 of the
that the named stars were used by Ptolemy for refer total amount o f stars in the sample. The graphs min

imum is reached in 900 a . d . or at the level of IT . The

discrepancy for the epoch o f 100 a . d . and earlier the
discrepancy equals 14 or more. Judging by fig. 7.14,
the most possible dating of the Almagest catalogue is
the interval between the alleged years 400 and 1400 a . d .
We have taken all the fourth level stars from table
4 .3 - the ones located at the maximum distance of
1 5 degrees from the informative kernel. There are

1 5 such stars, new additions being 7 8 ( 3 Gem ( 2 9 9 0 ) ,

7 9 Vir ( # 5 1 0 7 ) and 2 4 | X Leo ( # 3 9 0 5 ) . The discrep

ancy graph can be seen in fig. 7 . 1 5 . The graphs min

imum is reached around 8 0 0 - 9 0 0 a . d . at the level o f
1200 400 406 0 400 800 1200 1600 2000
10- IT . The discrepancy equals 12 for the epoch o f
100 a . d . Thus, the value o f the minimal square aver
age discrepancy hardly alters at all. Apparently, for Fig. 7.16. Square average latitudinal discrepancy graph after
distances under 15 Ptolemys tools would still allow the compensation o f the systematic error for the twenty-two
to measure stellar coordinates against the actual basis fifth level stars located at the maximal distance o f 20 de
grees for the base ones. The square average discrepancy was
stars, and not following links.
minimised in accordance with the variations o f parameter y
Finally, for fifth level stars we took the ones in
for the interval o f ystat 5', and the variations o f parameter (3
cluded in Table 4.3, located at the maximal distance for the interval o f 0 20'. The graph reaches its minimum in
o f 20 degrees from the informative kernel. There are 400-800 a . d . , at the level o f 22-23'. This is the level that we
22 such stars including the informative kernel - the find to be characteristic for the Almagest catalogue in gen
newcomers are 112(3 Tau (#1791), 60i Gem (#2821), eral. In other words, the proximity of the base stars ceases
to be effective at the distance o f some 15-20 degrees. The
685 Leo (#4357), 29yBoo (#5435), 3(3 CrB (#5747)
graph became almost even due to the significantly lowered
and 5 a CrB (#5793). precision o f calculations at such distance from the base stars.
The discrepancy graph is shown in fig. 7.16. The The discrepancy equals 23' for the beginning o f the new era,
graphs minimum is reached around 400-800 a . d . at 24' for the epoch o f the V century b . c . etc.
the level o f 22-23'. This is the mean-square error
level which is characteristic for the Almagest cata
logue in general, which is to say that the effect o f the alleged datings might make one star look measured
basis star proximity ceases to manifest at distances o f well and another poorly, and vice versa.
15-20. The graph became almost even due to a vis We shall continue with compensating the system
ible decrease in measurement precision at such a dis atic error discovered in the Almagest catalogue and
tance from the basis stars. The discrepancy equals 23' make y as well as (3 fluctuate within the same range
for the beginning o f the new era, 24' for the epoch of as above.
the V century b . c . , and so on. The amount of stars that we find in the sample
The last step demonstrates a drastic drop in meas after such a selection shall be represented on the same
urement precision. The square average error rate grew drawing as the discrepancy. The resulting picture can
by a factor of two. Therefore, before we move on in be seen in fig. 7.17. One sees that the minimal square
our extension o f the catalogues informative kernel, let average discrepancy drops to 9' once again for 800-900
us agree to count the square average discrepancy using b . c . , whereas the Scaligerian epoch o f Ptolemy and

only those stars for reference who get a maximal lat Hipparchus, or 400 b . c . - 100 a . d ., makes the discrep
itudinal error o f 30 minutes for the assumed dating ancy values maximal, reaching up to 12'. Let us point
o f the Almagest catalogue. This shall allow us to ex out that the resultant discrepancy values o f 9' for the
clude the star which Ptolemy measured the worst from presumed dating period of 800-900 a . d . correlate very
the very beginning. The choice of such stars naturally well with the discrepancy limit of 30' as specified be
depends on the alleged dating of the catalogue. Certain forehand. The matter is that the normally-distributed
i8o | h is t o r y : f ic t io n o r s c ie n c e ? CHRON 3 I PART 1

sigma N

18 45
16 40
14 35
12 |30
10 I 25
5 20
6 15
4 10
2 5
4200 800 400 400 800 1200 1600 2000
4100 *600 -100 400 900 1400 1900
Variables Fig. 7.17. Square average latitudinal Variables Fig. 7.19. A similar square average
sigma discrepancy graph for the collected sigma latitudinal discrepancy graph for the
NJuleps stars from table 4.3 located within Njn eps group o f stars from table 4.3 located
20 degrees from the stars o f the within 30 degrees from the stars o f
catalogues informative kernel. the catalogues informative kernel.
eps * 30" One can also see the graph for the ep$8830* We also presented a graph for the
number o f stars in this group. The number o f stars in the group.
<1 20 degrees stars whose latitudinal discrepancy d * 30 degrees
exceeded 30 minutes for the pre
sumed dating in question were
excluded from the sample. The
systematic error o f the catalogue
was compensated.

-1200 -800 -400 0 400 800 1200 1600 2000 -1100 -600 -100 400 900 1400 1900
t t

Variables Fig. 7.18. A similar square average Variables Fig. 7.20. A similar square average
sigma latitudinal discrepancy graph for the sigma latitudinal discrepancy graph for the
Njmjsps group o f stars from table 4.3 located NJnjeps group o f stars from table 4.3 located
within 25 degrees from the stars o f within 35 degrees from the stars of
the catalogues informative kernel. the catalogues informative kernel.
eps 30* We also presented a graph for the We also presented a graph for the
eps 30*
number o f stars in the group. number o f stars in the group.
d* 25 degrees d * 35 degrees


-110 0 -600 -100 400 900 1400 1900

t t
Variables Fig. 7.21. A similar square average Variables Fig. 7.22. A similar square average
sigma latitudinal discrepancy graph for the sigma latitudinal discrepancy graph for the
~~~~ NJnjjps group o f stars from table 4.3 located group o f stars from table 4.3 located
within 40 degrees from the stars o f within 45 degrees from the stars of
the catalogues informative kernel. the catalogues informative kernel.
ep s*30' eps~30*
We also presented a graph for the We also presented a graph for the
40 degrees number o f stars in the group. d 45 degrees number o f stars in the group.

random value with the square average discrepancy o f reached after 800 a .d . This sample contains 30 stars.
circa 9'-10' is likely to remain within the limits o f 30' The amount o f stars in the sample varies between 20
or 3a, the probability rate being close to 1. and 31 stars for different presumed datings. Around
Let us now expand the maximal distance between the beginning o f the new era the discrepancy rate is
the stars and the catalogues informative kernel from roughly equivalent to 13', which is close to the maxi
20 to 25. We shall still only regard the stars whose mal value for the graph in question.
latitudinal error does not exceed 30' for the presumed In figs. 7.20,7.21 and 7.22 one finds similar graphs
dating in question. See the resulting graphs in fig. 7.18 for the stars whose distance from the Almagest cata
representing the discrepancy as well as the amount of logue kernel does not exceed 35, 40 and 45, re
stars included in the sample for each presumed dat spectively. The sample consists o f roughly 40 stars.
ing. The square average discrepancy minimum is The latitudinal square average discrepancy minimum
reached on the interval between 800 and 1000 a .d ., becomes less manifest and drifts towards the future.
equaling circa 9.5'. The maximal discrepancy rate is The graph in general begins to look more and more
roughly equivalent to 12.5' and is reached around 400 horizontal.
b .c . The Scaligerian epoch o f Ptolemy and Hippar C orollary. Thus, the Almagest catalogue can be
chus, or the beginning o f the new era, has a discrep dated by the proper movement o f a configuration of
ancy rate approximating the maximum - about 12'. roughly 20 stars. The most possible dating interval
The amount of stars in the sample varies from 21 to falls on the same epoch as above, namely, 600-1200
24. There are 23 stars in the sample corresponding to a .d . We also discover that one has to use reliably iden
the minimal square average discrepancy. tifiable stars which arent located at too great a distance
We shall proceed to raise the acceptable distance be from the informative kernel (20-25 maximum). If we
tween the stars and the kernel from 25 to 30, keep are to exclude the stars who get a maximal 30-minute
ing all other parameters just the same as they were. The latitudinal discrepancy for alleged dating t from the
result can be seen in fig. 7.19. Once again, the mini sample, we shall end up with about 20 stars. This pro
mal possible latitudinal discrepancy can only be vides for a graph with a well-manifest minimum as
182 I h i s t o r y : f i c t i o n o r s c i e n c e ? CHRON 3 I PART 1

Plot of sigma VS t <20 Stars: 8+12 selected from ZodA) sumed dating can be seen in fig. 7.23. The poorly-
28 manifest minimum corresponds to the level o f 23'. It
26 is reached on the interval between 400 and 800 a.d .
24 : H*'*']
22 A mere T above the minimum, and we shall cover the
20 ; entire interval o f 400 b .c. and 1500 a.d. Therefore,
g 16 this list doesnt permit any reliable datings due to the
& 14 low average precision o f the stellar latitudes that it
*55 12
10 contains. Even the eight informative kernel stars can
8 not improve the average latitudinal precision o f this
4 list owing to the fact that most o f the visibly mobile
2 stars from area Zod A are rather dim, and were there
fore measured rather badly by Ptolemy on the aver
-1200 -800 -400 0 400 800 1200 1600 2000
t age. Bear in mind that the average precision o f his lat
itudinal measurements equals 12'-13' for the entire
Fig. 7.23. Square average latitudinal discrepancy graph for Z od A area, which is a lot better than the 23' that we
20 stars: 12 stars from table 4.3 located in celestial area
get for the 20 stars in question.
Zod A yexcluding the informative kernel stars, and 8 stars o f
the informative kernel. As one sees from the graph, the lati
We have thus managed to expand the informative
tudinal precision for this list is substantially lower than that kernel o f the Almagest without any substantial pre
for the area Zod A on the average. cision losses to 15 reliably and unambiguously iden
tifiable Almagest stars that are also visibly mobile, by
which we mean that their minimal annual proper
seen in fig. 7.18. The latitudinal discrepancy mini movement speed equals 0.1" by one of the coordinates
mum o f 9' is reached on the interval o f 800-1000 a.d. at least. The choice o f the celestial coordinate system
The interval o f 600-1200 a.d . corresponds to a dis is o f little importance here, and so we are using the
crepancy rate very close to the minimal, one o f 9'- 1900 a.d . equatorial coordinates for the sake o f con
9.5. The epoch o f 400 b .c. - 100 a.d . corresponds to venience, since they are used in the modern star cat
the maximal discrepancy rate o f 11.5-12'. alogues that we have used. Let us now cite the final
Let us emphasize that the minimal discrepancy of list o f the 15 stars that enable a proper movement
circa 10 can only be reached for a group o f several dating o f the Almagest. The BS4 number o f the star
dozen stars on the condition o f their proximity to is specified in parentheses ([1197]).
the informative kernel o f the Almagest. All the other
methods o f selecting the stars from the combined 1) 16a B oo (5340); 2) 13a Aur (1708); 3) 32a Leo (3982);
areas A, Z od A, B, Z od B and M - by luminosity, 4) 10a C M i (2943); 5) 67a V ir (5056); 6) 21a Sco (6134);
fame etc leave us with the discrepancy minimum of 7) 3a Lyr (7001); 8) 4 3 y C n c (3449); 9) 78p G em (2990);
roughly 20, which is typical for the Almagest in gen 10) 475 C nc (3461); 11) 14o Leo (3852); 12) 24p Leo (3905);
eral. Remaining within a single well-measured area 13) 79^ Vir (5107); 14) 8q Boo (5235); 15) 26e Sco (6241).
(Z odA ) is also a non-option. For example, let us re
gard all the visibly mobile stars from this area as a 5.
whole, that is, all the stars from table 4.3 that pertain DATING THE ALMAGEST CATALOGUE
to celestial area Z od A. There are 12 such stars if we BY A VARIETY OF 8-STAR CONFIGURATIONS
dont consider the informative kernel; adding the CONSISTING OF BRIGHT STARS
8 stars that comprise the latter to this amount shall
give us a total o f 20 stars. Unfortunately, the latitude The idea behind this calculation as well as the cal
precision for this list is rather low - a great deal lower culation itself are credited to Professor Dennis Duke
than that o f area Z odA in general. The corresponding from the State University o f Florida, an eminent spe
square average latitudinal discrepancy graph for these cialist in data analysis. He suggested to study all pos
20 stars as a function of the Almagest catalogues pre sible configurations of eight named Almagest stars.

Professor Duke chose a set of 72 stars whose Almagest proved asymmetric in relation to the distribution
magnitude is less than 3 (bear in mind that the lower centre (qv in fig. 7.24), the average estimation for this
the value, the brighter the star) for this purpose. Then distribution turns out to be shifted sideways. If we are
he selected all the 8-star combinations from this num to take this effect into consideration, the more accu
ber whose maximal latitudinal error in the Almagest rate estimate o f the mean-square aberration shall
catalogue does not exceed 10' for a certain non-zero yield an even smaller value.
time interval (tv t2) that covers the entire period be Moreover, the centre o f the selective distribution
tween 400 b .c . and 1600 a.d . The total amounted to is located near the year 800. Had the sample elements
736 eight-star combinations out o f 500.000 possibil been independent, one could come to the conclusion
ities. Each one o f these combinations specifies a dat that the real dating o f the Almagest catalogue com
ing interval (tly t2) of its own. Professor Duke stud pilation can be located within
ied the set o f such dating interval centres, or the set 800 (3 X 400) / V 736,
o f values {tl + 12) / 2. It turns out that if one is to build or 800 45 years. However, one cannot consider the
a frequency distribution histogram o f these centres sample elements to be independent since the real pre
on the time axis, one sees a manifest maximum on cision o f the 800 a.d . dating for the Almagest is a
the interval o f 600-900 a.d ., qv in fig. 7.24. Therefore, great deal lower than 4 5 years. Nevertheless, the
the epoch o f the V II-X century a.d . is the most likely early a.d . period dating or an even earlier one can be
date when the Almagest catalogue was compiled. regarded as highly improbable in this situation, and
The approach suggested by Professor Duke has all but out o f the question.
the advantage that poorly-measured or excessively
slow stellar configurations are automatically excluded 6.
from the sample due to the fact that their dating in THE STATISTICAL PROCEDURE OF DATING
tervals are either void for the 10-minute latitudinal THE ALMAGEST CATALOGUE:
threshold, or great enough to go well beyond the his STABILITY ANALYSIS
torical interval o f 400 b .c. - 1500 a.d . as chosen by
Professor Duke a priori. It turns out that after such 6.1. The necessity of using variable
a rigid selection one is still left with a great many algorithm values
configurations, namely, 736 o f them, each one con
taining eight stars. If we are to chose the dating in The implementation o f the dating procedure as
terval centre of some such configuration as a dating described above involved a rather arbitrary choice o f
with a latitudinal level o f 10, we shall end up with the certain values defining the algorithm, whereas other
Almagest catalogue dating that shall contain some values result from statistical conclusion. One there
random error, or a perturbed catalogue compilation fore has to check the behaviour of the resultant dat
dating. Once we build a distribution graph o f these ing interval in case o f said values being subject to al
perturbed datings, we shall be able to date the Alma teration.
gest catalogue with a great deal more precision than
in case o f using a single configuration. 6.2. Trust level variation
The natural assumption is that the true dating of
the catalogue equals the average value o f the ran The value o f 8 that determines the trust level was
domly perturbed datings. This average can be esti chosen rather arbitrarily. Bear in mind that in statis
mated by the empirical distribution that we have at tical problems it represents the acceptable error prob
our disposal. Considering the true perturbation dis ability rate, that is, 8 = 0.1 stands for the error prob
tribution to be close to normal, it is easy to estimate ability rate o f 0.1. The smaller the value o f 8, the
its dispersion. The selective mean-square distribu greater the trust interval. The dependency o f the trust
tion aberration as seen in fig. 7.24 roughly equals 350 interval size