American Philosophical Society

Saros Cycle Dates and Related Babylonian Astronomical Texts
Author(s): A. Aaboe, J. P. Britton, J. A. Henderson, O. Neugebauer and A. J. Sachs

Source: Transactions of the American Philosophical Society, New Series, Vol. 81, No. 6 (1991),
pp. 1-75
Published by: American Philosophical Society
Stable URL: http://www.jstor.org/stable/1006543 .
Accessed: 19/07/2013 15:22
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .
http://www.jstor.org/page/info/about/policies/terms.jsp
.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of
content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms
of scholarship. For more information about JSTOR, please contact support@jstor.org.
American Philosophical Society is collaborating with JSTOR to digitize, preserve and extend access to
Transactions of the American Philosophical Society.
http://www.jstor.org
This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

All use subject to JSTOR Terms and Conditions
TRANSACTIONS
ofthe
AmericanPhilosophicalSociety
Heldat Philadelphia
forPromoting
UsefulKnowledge
VOLUME 81, Part6, 1991
Saros Cycle Dates and Related

BabylonianAstronomicalTexts
A. Aaboe, J.P.Britton,J.A. Henderson,

0. Neugebauer, and A.J. Sachs
THE AMERICAN PHILOSOPHICAL SOCIETY
Independence Square, Philadelphia
1991

Copyright(? 1991 by The American Philosophical Society
Libraryof Congress Catalog

Card Number: 91-55342
InternationalStandard Book Number 0-87169-816-1
US ISSN 0065-9746

TABLE OF CONTENTS
I. Introduction ................................... 1
II. Saros Cycle Texts

TextA .............. ..................... 4
TextB .............. ..................... 10
TextC ................ .................... 12
TextD .............. ..................... 24
III. TextsContainingAstronomicalFunctions
TextE .............. ..................... 34
TextF .............. ..................... 63
TextG .............. ..................... 69
Bibliography .................. ................. 73

I. Introduction
T Nhe textspresentedhere are mostprobablyfromBabylon,

although theirexact provenance is unknown.1All concern
luni-solarphenomena withthe exceptionofa texton the last
visibilityofMercury,which is found on one side of a tabletwhose
other side deals with lunar eclipse magnitudes and longitudes.
The textsfall into two groups. One comprises what we have
called "Saros Cycle Texts,"which give the monthsof eclipse pos-
sibilitiesarrangedin consistentcyclesof 223 months(or 18 years).
Three ofthe fourtextsin thisgroup concernlunar eclipse possibil-
ities; the other treats solar eclipse possibilities analogously. In-
cluded in this group is B.M. 34597,known as the "Saros Canon,"
which we republish to correctseveral errorsin previous publica-
tions, and to clarifyits structure.
The second group oftextscontainsastronomicalfunctions.Two
(TextL and TextF) tabulatelunar longitudesat syzygiesin accord-
ance with a relativelycrude scheme, which approximatesuniform
motion and seems designed to facilitatecomputation. One of
these (TextL) also presentsa new functionwhich describes lunar
eclipse magnitudeswithconsiderableaccuracyand includes a cor-
rectionforzodiacal anomaly. A thirdtext(TextG) is a fragment
of a previouslypublished text,which we here call TextS. The frag-
ment enables us to restorethe functiondescribingeclipse magni-
tudes in Text S. This in turn contributesto our (still imperfect)
understandingof the analogous, but more sophisticatedfunction
in TextL. Finally,the last textin this group (TextM), occupies the
obverse of the tabletcontainingTextL and gives the longitude of
Mercuryat successive last visibilities(Q). The writingon it is at
rightangles to that of TextL on its reverse,and we treatit here
as though it were a separate text.
The periods covered by these textsare generallyearlier than
most ofthe dates associated withmathematicalastronomicaltexts,
I All the textsare in the BritishMuseum, and we publish them throughthe courtesy
of its trustees.For a discussion of the circumstancesof the acquisition of Rassam's "Baby-
lonian Collection" and the problems of establishingthe provenance of these texts,see
Reade [1986].
1

2 SAROS CYCLE DATES
although there is some overlap. The Saros Cycle texts include

dates fromthirteen18-yearcycles, which extend from -490 to
- 257 Of the textswith astronomicalfunctions,Text S concerns
solar eclipse possibilitiesfrom-474 to -456; TextM gives calcu-
lated positions ofMercuryat Q forthe period from- 423 to - 401;
TextL listslunar eclipse possibilitiesfrom- 416 to - 380; and Text
F gives approximatelongitudesof fullmoons from- 261 to - 256.
It should be unnecessary,but unfortunatelyis not, to remark
that textspresentingdates correctlyin several reigns beforethe
Seleucid Era cannot have been composed, at least not in theiren-
tirety,in advance of the events they describe. Indeed, we do not
know of a single astronomicalcuneiformtextin which a regnal
year exceeds the naturalreignof the king beforethe introduction
of a continuingyear count in the Seleucid Era. Thus we cannot,
alas, be more precise about when our textswere composed or
written.
The present paper began with the collaboration of the late
A. Sachs* and A. Aaboe around 1970on Saros textsin the British
Museum. The disjointedfragmentfromthe cornerofTextL raised
difficulties(not yet fullyresolved) that broughtthe enterpriseto
a prolongedhalt. Though questions remainunanswered, we pub-
lish the textsso that othersmay trytheirhand.
Our paper has been referredto in the literaturethrice:first,in
Aaboe [1972], n. 9; subsequently,in HAMA, p. 1106 as 'Aaboe-
Henderson-Neugebauer-Sachs[1975],"and lastlyin Britton[1989]
as 'Aaboe, et al. [1988]."
A. Aaboe's visitsto the BritishMuseum in the 1960s and early
1970s,duringwhich most of our textswere firsttranscribed,were
supported by grantsfromthe National Science Foundation and
the JohnSimon GuggenheimFoundation,which supportis grate-
fullyacknowledged.
* Abraham Sachs died 22 April 1983. Otto Neugebauer died 19 February1990,afterthe

present paper was submitted.

II. Saros Cycle Texts

TextA: B.M. 36910 (80-6-17,651)+ B.M. 36998 (80-6-17,742)
+ B.M. 37036 (80-6-17,780)
PLATE 1.

TEXT A
Contents:Regnal years and months of lunar eclipse possibilities

for(at least) year 31 of Darius I to year 30 of Artaxerses
II (-490 to -374) arrangedin 18-yeargroups.
PreviousPublications:Mentioned as Nos. *1422, *1423,and *1424
in LBAT.
Table 1; Translation:Table 2; Photograph:Plate 1
Transcription:
DescriptionofText:
TextA consists of threerejoined fragmentsand measures 31/2"
by 31/2".While no edges are preserved, traces of line 1 of the ob-
verse appear in columns IV' and V', showingthatitsverticalextent
is nearly complete. Horizontally,the text could have contained
one or more columns on eitherside.
The textgives columns of dates in the formof a regnal year
number and a month. No instance of a firstregnal year is pre-
served, so we do not know if the kings' names were given.2
Withineach column, successive dates are eithersix or fivemonths
apart,and each column begins aftera five-month interval.Succes-
sive dates in the same line differby 223 monthsbetween columns.
Each column has 38 dates which begin at line 1 ofthe obverseand
carryover the bottomedge to the reverse.There are 21 or 20 lines
on the obverse and 17 or 18 on the reverse.
Though TextA leads into,and partlyoverlaps,the Saros Canon
(TextC), it looks quite different: it is less carefullywritten;it has
no verticalrulingsseparatingthe columns of dates, nor horizontal
lines indicatingfive-month intervals;and "5 itu" (= fivemonths)is
not writtenaftera five-month interval.The dashed lines in Tables
1 and 2 thus have no counterpartsin the text.All in all TextA ap-
pears much less carefullyprepared than TextsB or C.
The characterof an intercalaryyear is indicated by "dir" if the
year contains a second Addaru (XI12), and by 2-kam (short for
"kin-2-kam")if it has a second Ululu (VI2) . This informationis
mostly writtenimmediatelybelow the year number, except in
Obv. 9',III' and when the intercalatedmonthitselfappears in the
text(Obv. 6',II' and Rev. 6',IV'). Month X112is twice written"dir"
and once "dir-se" (Rev. 6',IV'). In the lattercase it is the careful
2 Cf. TextC where they are given and TextD where they are omitted.
3 In translationssuch years are indicated by * and ** respectively.

6 SAROS CYCLE DATES
TABLE 1
ObYverse, I I I I I I
I') | I 1 1 ]3 51 I1[Sf
M) I dIju~ cbfL&p ri.Pif I afpin. rgaxm I ran? II
3') N1arl 14bar 11ba[r] 29 3uL 4
g6 . 15 s 23'
Io I1L6 i d6
C dI .46 ['IAri>.piat a-pi I irgan 3fati.
5') lc]r 5 bax 2. larc [olb&r
I I 5L 6 gU4 1r4 5t17i4
4')| j m 1ain de
3.kit U6 I Aa I
in i dr awv I
'r') 3se.:::l | r 13 iar 31[barb[j 5 bar I 7 25 Suf, I
g) ?_]: _~~16 k't1 &6-1 z[1 &i6__l jai1 a9pLA
9') I Ii Yzt'
' ?e I 3e. [ ie
t3[ L2JI[
to') I i1i.t i .t. kt.9t kV[i 3 [g kin
' t7- zt2gel[. e [ I
s1 33zi 1 15ti 1 33t2t Io kin I kin I I
13') I| dirz tz zI7zXtz .tire
Z I
HS') I I]'S ?T 1 6 Z 14 iz Zit 110 kIn n
5') I a.!1bk. ai57 IZZ .i 1 1z .d .il
- -
Il') 1 .
1]3g5arv gan I air ab ab ab [ I
1' ,,JI 18~ sa9 1 3& 5YJ[ I I I I
I[I 3a t 1 I I I I
20') 1 I 31 1 1 1
2se- - 1 Cd8ir
3 i1I I
I')9 I I r&,a1n
Y11d'i I I
I)I I Ij311bar 116/9 [ 1 1
3') 1 1 1 1 ]1z~~~~~~~~~c6
2.,4A4dL6 t[II
I I0 I Ji-.ir i0 1'ar I117 ar [16[4lIJ3i
I) lYkr,it 2.2kiit I dX I 6 1I 6
6') I 1jirAe~
5e r I3SLbarJ I11L1arj i
1') j 115kin 1z2kin I 1ktin Ii Ii
9 Iz 4ir 2 I te I LiV I I
I') 6 ab
iz 12Lkij I,
9
to' Uab ab7.Lrt,z 9[t
II') 11
]{ C IX [A4 2. 5j( I I I I
12') Idtrab j b l I I I I I
15') 11 I I I I I
4') I_ _ _ -_ - 11 t - - _I_ _- _?
- _I -

SAROS CYCLE TEXTS 7
TABLE 2
tA
Te',,x
Obv. Z Ur, u
31 1 13 1 Z IT
le Ti- 5* -Z ET 2Z N
Z, Vill ZK sK ZE
,3fr I /I I 12-Ilt I 23 fff I
ol I -W 9 I El I a I
K'Tz t5 1: IZ I 1 30 1 1 7* zr (o 2,' f
33 R 94- ,x0- I 01 I El1:1 Y-W FX I
K R.2-, 13 1 131 1 I 7* T-r Z5 F-I
ljq El: 6 U7- 7m Oji
to. 5`0 3Z0 9 EZ 01
XL I Ki I u I &U2, 1
36 5 1 15' 27 133 v- Id*a 1 ZE 127 a i
x ?1 'ET
-K 13Y -7 1 u 128 Kil
9 I
2i 297 -71 xu &a
I I
I-Z-E--i 2L LT 17 T-T 35*17 U f w
ju I *,.":T
y- x
Z -TY 36 V (3 LV iv
3
X
2-o, I ir LIE 137 -7
19'* Iq IJI
/3*Ev
T- CA
Re V.
I 5' r 2 11 2o 71 3ef /F 1- I /y 5
5= -I- - iK
-
&TI 2.1 I 3 IL I /S ff
(o -a iE V- VIL
Im XIT mzl Yo*-I 7 r iT
7* vl- E- ZZ 'O UT
I 1 u
/7
-.7--
lu M2. 1
z3 )a vr
yi1), U 1 1t
IFI 4IN 13)os. XI) Rz I I I
10,1 2* 2T -I- 19 7-L-I- -
x
(o hi 7 L-V- JE .7
K K Xi'
-7 1 2 (o EV 3 2 2b* --Io
X- I 11x AL xi-
1.2 27 N ig- 21 19

8 SAROS CYCLE DATES
alignmentofthe monthname which distinguishesitfromthe sev-

eral instancesof "dirse" (Obv. 9',III'; 11',IV'; 13',V';and 15',VI'and
Rev. 8',II') where "dir" is writtenunder the year and the meaning
is "XII2 year,month XII."
The textcontains three erasures, all of misplaced "dir"s. This,
togetherwiththe absence ofrulings,suggeststhatthe textwas not
a copy of a finishedtext,and thatthe scribe had some difficulty
in designating intercalaryyears. Traces of a "dir" in year 38 of
ArtaxerxesI, however,show that actual, ratherthan calculated,
intercalationsare recorded.4
4 During the reignof ArtaxerxesI intercalations

occur in the correctsequence required
by the nineteen-yearcycle, but only month XI12is used and V12 does not appear. Thus
years 19 and 38 have intercalaryXI12's,where we would otherwiseexpect VI2's. See PD3,
6-9.

TextB: B.M. 37044 (80-6-17,788)
PLATE 2.
Rew.
10

TEXT B

for(at least) year 10 to year 30 of ArtaxersesI (-454 to
-434).
PreviousPublication:Mentioned as No *1425in LBAT.
Table3; Photograph:
Transcription: Plate2
ofText:
Description
TextB is a small, well-writtenfragmentwith part of the upper
edge preserved. The surface of its reverse,where preserved, is
blank. A verticalline separates Columns I' and II'. TextB dupli-
cates the firstfouror fivelines of Columns III' and IV' of the ob-
verse of TextA, but with two differences.First,its columns are
separated by a verticalrulingand, second, it has the annotation
"5 itu" (= five months) in Col. I', line 1. Thus when complete,
TextB probablyappeared verymuch likethe Saros Canon (TextC).
TABLE 3
Obv&rs&
e'
l) Xudirapin apin
3) fzl bar Xq gf
4) 1 cLtt dir apkir
5) J3 ,bar,j
Z922
1TIL Y9 It
11

TextC ("Saros Canond'): B.M. 345975(Sp. 11,71)
Obv
N >'., ~~~~a PLATE

3
t ~~~~~~~7/ 9-S -ss<Z
'''_zg,'::2~~~~~~~~~~~~~~~~~~~~~~~"
0,0E::D
o .-A
5 TextC Not 34579 as in Neugebauer [1938], 248, 342].

TEXT C

for(at least) year 4 of ArtaxerxesII to S.E. 406 (-400 to
-271) arrangedin columns of 18 years (223 months).
PreviousPublications:Publishedin hand copyby Strassmaier[1895]
and (in Pinches's copy) as No. 1428in LBAT;
and excerptedwithrevisionsin Aaboe [1972].
It has been discussed in, i.a., Epping and
Strassmaier [1893], Pannekoek [1917], and
Neugebauer [1938].
Translation:Table 4; Photograph:Plate 3
DescriptionofText:
TextC is a handsome fragment43/4"wide by 41/8" high. Vertical
rulingsseparatecolumns of dates, which include abbreviatedking
names afteryear 1 of each reign.7Horizontal alignmentis ob-
served throughout,so that dates in a given line increase by 223
monthsfromone column to the next.Where the intervalfromone
line to the nextincreasesby fivemonths,the entriesare separated
by a horizontalline, and the second entrycarriesthe annotation
"5-itu" as in TextB. These lines continue across the entiretext,
dividing each column into groups of seven or eight dates sepa-
rated by six-monthintervals.
As published by Strassmaierand in LBAT,TextC presentsele-
ments of seven columns of 38 lines each, beginningand ending
in the middle of a group of seven eclipse possibilities.In Strass-
maier's copy the columns on obverse and reverseappear in good
alignment,whereas Pinches's copy in LBAT shows columns which
do not connectcleanlyacross the tablet'sedges, but are somewhat
offset.In LBAT the identificationof "obverse" and "reverse"was
made to minimizethis shift.
Subsequently Aaboe [1972] proposed that "obverse" and "re-
verse" be interchanged,based on the curvatureofthe tablet.With
thisidentificationeach columnbegins just aftera five-month inter-
val, and the dates are consistentwith those of TextA and TextB
6 We use "S.E. N" to denote year N of the Seleucid Era. Month I of S.E. 1 began on
April 3, - 310.
7 For the Achaemenid king names see Sachs [1977]. The abbreviationsused in the text
are: u (Umasu) = ArtaxerxesIII; dr = Arses; da = Darius III; a = AlexanderIII (the Great);
pi = Philip III (Arrhidaeus); and an = Antigonus.The entryfor1 Seleucid Era is broken;
it probablyhad si.
13

14 SAROS CYCLE DATES
in the sense thatall dates in a given line are separated by a mul-

tiple of 223 months.In this arrangementthe textpreservestraces
ofeightratherthan seven columns,which ifcompletewould have
extendedfrom- 400 to - 257. On the rightside the narrownessof
the tabletmakes itlikelythatthe last preservedcolumn was in fact
the last column ofthe original.To the left,however,at least a third
and possiblyalmosthalfthe tabletappears missing.Thus the com-
plete text probably contained between 13 and 15 columns, ex-
tendingat least as farback as TextA (- 490) and possibly to - 526.
As in TextsA and B intercalaryyears with a X112are designated
by a "dir" beneath the year numberunless the intercalarymonth
itselfappears as an eclipse possibilityin that year. An exception
is 12 S.E. where the designation is omitted,although it is clear
fromthe months that the year contains a X112.Years with a V12
are designated by "kin-dir,"in contrastto TextA.
The text'suse of regnal years afterDarius III is as follows:
1 AlexanderIII (theGreat)follows5 DariusIII;
1 PhilipIV (Arrhidaeus)follows7 AlexanderIII;
1 Antigonusfollows6 PhilipIV; and
1 SeleucidEra follows6 Antigonus.
As discussed more fullybelow, this rational,but unconventional
practice differsfrom that described in PD3 and also fromthat
found in TextD. No colophon is preserved,but the textwas ob-
viously writtenafterthe adoption of the Seleucid Era.8
Critical
Apparatus:
For our identification of obverseand reverse,see above. All ref-
erences are to the translationgiven in Table 4.
Rev.33, Cols. V' and VI' and Rev.35, Col. VI': Pinches (LBAT) gives
"su" (month IV) for "du6" (month VII); this implies that he
copied what he saw and not what he thoughtshould be there.
Rev. 37 and 38, Col. I' : The text(and Pinches) has traces of "izi"
(monthV) and "ziz" (monthXI). Strassmaierrestoresa "dir"(in-
dicatingan intercalaryX112)in year 20, in agreementwith Sp. II
901 = B.M. 35328,9which forcesthe readings "su" (monthIV)
8 While 1 Seleucid Era begins in -310, Seleucus did not become king until 7 S.E.
(-304) (Sachs and Wiseman [19541,205), and the earliestattesteddate is 8 S.E. (PD3, 20).
As late as 10 S.E., however,we findthe date "year4" (of Seleucus) in a Diary for - 302/301
(Sachs-Hunger [1988],251). Thus the conventionof countingyears from1 S.E. regardless
of who was king, must have become general practicebetween -300 and -280 (31 S.E.),
when Antiochus I became sole king.
9 Published as No. 1394 in LBAT. Translatedand discussed by Kugler SSBI, 80-81. The
textis a Jupiterobservationtext,which is explicitlyas well as astronomicallydated. Obv.
26' has "dir-ge30" in a section beginning"year 20." This appears to be confirmedby sub-
sequent month names, although poorly preserved.

' I l~~ I ~~ I ~ ~Ii ~ ~~~~~~~~
In~~~~~~~~~~~ I I
N
F
o
on
eO
Ii I
I------ n - I l [
1t #cq
<~. i (A i (A Z411k2
I _1 I 1
ki12!Ii-
r>r,-Lo1-2 14,=I= W I
Kw *en c,, I 1 1 ?*I
1,~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~-
d I
I< ts % 14 > *- w 1s e > I bvls < s* e 1o '4 4 I
C- t| t| S
~
) : 60 |,1
1 k i t|Ix
&
1>p 1 61 SL
v
N :3 o "N 1 en \s
I I t ~~~~~~~~~~~~~~~~~~~~~~~I
~
I I I
L t ~~~~~~~~~-4 041 1
~~~~~~~j ~~~~~~~~~~~~~~~4
IA
1 | tG9Nld W 14k cq
hS
| bM C- >| c1r Vk)\i iW C taa11N)
1guo Ok~~~~~~~~~~~~t
Mpt>>g-IW$_:
$
ks
F
%e 0C
Ie-- 1S/ 0 II
eX1~~~~~(j 1,; X

and "ab" (month X). The text(with Pinches) implies that the
intercalationoccurred in year 21* which agrees with Text L
(below). This removesthe onlymeaningfulanomalyin the nine-
teen-yearintercalationcycleafter- 497,and moves the introduc-
tion of a consistentnineteen-yearintercalationscheme back to
the beginningof the fifthcenturyB.C.10
Commentary:
TextsA, B and C-despite differencesin orthography-all de-
rive froma single, consistentscheme. Each has columns of 38
dates (years and month names), which begin aftera five-month
intervaland run fromobverse to reverse.Furthermore,two dates
in a given line (countingfrom1 to 38) always differby a multiple
of 223 months,whetherin the same textor not. As a result the
five-monthintervalsalways occur afterthe same lines, dividing
each column into groups of 8-8-7-8-7 dates, where fivemonths
separate the groups, while consecutive dates within each group
are six months apart.
This arrangementofdates can be derivedfromthe assumptions
that: (1) sun, moon and node move uniformly;and (2) the sun re-
turns to its position relative to a node in 223 months.11The
second assumptioncorrespondsto an eclipse cycle,now generally
known as the "Saros," in which38 eclipse possibilitiesoccurin 223
months.12Our textsthus give the months of lunar eclipse possi-
bilitiesbased on this cycle.
By "eclipse possibility"we mean a syzygyat which the sun is
withinhalf a month'sprogressin elongation froma lunar node.
At such times solar eclipse possibilitiesoccur at conjunctionsand
lunar eclipse possibilities at oppositions. By this definition,as-
suming uniformmotion, therewill be exactlyone solar and one
lunar eclipse possibilityassociated with each passage of the sun
by a node.13 This agrees with the observational fact that for a
given location solar eclipses rarely,if ever,occur only one month
apart, and lunar eclipses never do.
10 The only divergencefromthe standard nineteen-yearintercalationscheme after
-497 is the previouslynoted (Note 3) replacementof VI2s with XII2s during the reign of
ArtaxerxesI. This obviously has no effecton the distributionof intercalaryyears.
11For this derivationfrom simple arithmeticalconsiderations see Aaboe [1972] and
Britton[1989].
12 See Neugebauer [1957], 141-143,and HAMA, 497 n.2 forthe historyof the modern
use of"Saros" forthe 223-montheclipse cycle,beginningwithHalley in 1691.In Babylonian
textsthiscyclewas called "18 years."We have used "Saros cycle"to mean 223 months,and
"Saros Cycle" to mean 223 monthswhich are also consistentwiththe arrangementin Texts
A-C (i.e., the firstmonthis a multipleof 223 months distantfromthose in line 1 of Texts
A-C).
13 For a full discussion of the theorypresented in SystemA, where the motion of the
sun and moon at syzygyis not uniform,see Aaboe and Henderson [1975].

SAROS CYCLE TEXTS 17
Solar ;
Months
EP DATE
L?unar I ALEXXI S = L-1
EP DATE 3 * XI it
1 2 IV 4 2 IV S=L
3 3 IV 6 3 IV .
=
S 4 IV 8 4 III S L-1
6 * X 9 * IX it
7 S III F 10 1I it
8 X 11 VIs l is
9 6 II 12 6 II S=L
VItd 10 VIg13p 7a
11 7 II 14 7 II it
12 * VIta 15 c ing la c
13 8 I 16 X112 S-=L-1
(SC 24)
FIGURE1
There is a simple relationshipbetween lunar and solar eclipse

possibilitieswhich, forthe Saros cycle, is shown in Figure 1. If
lunar eclipse possibilities are associated with a distributionof
dates into groups of 8-8-7-8-7 EP,14then solar eclipse possibili-
ties will be distributedinto groups of 7-8-7-8-8 EP, and the cycle
will begin 3 EP earlierthan the correspondinglunar cycle."5
For solar eclipse possibilities,therefore,the five-monthinter-
vals fallin the middle of the (six-month)groups forlunar eclipse
possibilities (and vice versa). Consequently, half of all solar
eclipse possibilitiesoccur in (i.e., at the end of) the same month
as the correspondinglunar eclipse possibility,while the otherhalf
occur in the precedingmonth.This fact,togetherwiththe location
of the five-monthintervalsallows us to establish with certainty
14 For convenience we use the abbreviation"EP" for"eclipse possibility"when refer-
ringto some numberofthemor to a specificone, and use the written-out expressionwhen

referringto the general phenomenon.
15 See Britton[1989], 21-24, forthe derivationof these relationships.

that TextsA, B, and C concern lunar, ratherthan solar, eclipse

possibilities.
It is naturalto ask how well this simple scheme agrees with the
actual recordof historicaleclipses. Table 5 shows the Julianyears
of all lunar eclipses visible in Babylon from the beginning of
Nabonassar's reignin - 746through- 238. The dates are arranged
to be consistentwith our textswhere theyoverlap,and the Cycles
are numbered so that Saros Cycle 1 (SC 1) is the firstcomplete
Cycle afterthe beginningof Nabonassar's reign. Horizontal lines
indicatethe boundaries between groups of eclipses separated by
a multipleofsix months,and thus correspondto five-month inter-
vals in our texts.
For 16 complete Saros Cycles, beginning with SC 13 in -526
and extendingthroughSC 27 (- 257), the scheme worksperfectly
in the sense that all lunar eclipses visible in Babylon occurred
in the given months.16In SC 12 and earlier Cycles, however,
eclipses at EP 16 occur one month earlier than in our scheme,
which shiftsthe boundary between Groups II and III one EP
earlier.Similar shiftsoccur in the boundaries between Groups I
and II and Groups IV and V in SC7 and between Groups III and
IV in SC 4. Thus beforeSC 5, whichbegan in - 670,onlythe boun-
dary of the Cycles themselves-i.e., the five-monthintervalbe-
tween Group V ofone Cycle and Group I ofthe next-is consistent
with our scheme. This boundary persistsfromsometime before
Nabonassar (- 746) through SC 27. In SC 28 the eclipse of
- 238:Oct 23 extends Group V, so that subsequent cyclesbegin 1
EP later.
The last preservedcolumn in the Saros Canon correspondsto
the last cycle (SC 27) which is fully consistentwith preceding
cycles.As we shall see below, a similarlist of solar eclipse possi-
bilitiesalso ends with SC 27, although in this case an additional
column was ruled offbut not filledin. Whetherthe scheme was
continuedpast the discontinuityat SC 28 remainsunknown. Nor
is it clear how farthe scheme in our textswas extended to earlier
periods,especiallybefore- 526 (SC 13) when the five-month inter-
vals between groups of actual eclipses are no longer consistent
with those in our texts. That some such scheme was used for
earlierdates is suggestedby the use of the term"5-itu"in several
early eclipse reports,since the termcan only referto the interval
16 Lunar eclipses did in factoccur in -274 and -256 (EP 1; SC 27 and SC 28), one
month afterthe indicated date, but neitherwas visible at Babylon. In the followingcycle
the correspondingeclipse (- 238:Oct 23) was visible at Babylon,thus violatingthe scheme.

[ f~~~ -tInIn ~ -a: ~~~~~~~~~~

ON a a In I ~ a0. 00a 0
A a-- Ir 0n a. a- r gI .Ia.2f- ' I . n I
a1 0a.
C14 - r-~ z ;-I a a a
N c7l 00
0aaaa a8 O0 0 0 00 0
tN aaaa~~~~~~~~~~~~~~~~~~~~~~~.
aaaa~~~~~~~~~~~~~~~~
a, a-r-- - a.A a ' r oIn In
In Ill In
a a
"I In
aa aa ao
N N C14C, C,4
aaa
C, N
0l
N~~~~~
-t 0,:~~~~~~~~~ ~ ~ ~ ~In In C,I ~ ~ ~-a~ ~ a~ ~ ~ ~a ~~~14 ~ 0
In
N C4
m ~In ~ C
~ ~a- ~a- ~ ~ ol a~*00 r-
CAa IN
a--.
a
'r 'IT Ina 'a. 0a
cn 0 r TIn.a ' aIna aan en I - aIn
0 IN aa- In-aaa -- aa a nI
~~~ 00 a~~~~~~~~~~~~0
r~- 1 ZR a.
~ ~ ~aaaInIl tC~ ~ ~ ~ ~ ~ ~ ~ ~ ~
aa aa a. a.a0. a- a a
1.0 a~~c r. -r-a-
aa o.a.aaa- I- c88 a In en 0 a-a-
0 0 0~~~ ~~~
~~~~0 1 01
1oo ~~~~aaa ~~~ aa a Ina In I
-- a ..a -
0 C- aaa- .o-aaaa-C.Oaaa.a00.aO- -aaa-0.r n

The termis found in the following

between eclipse possibilities.17
listof eclipse reports,where "EP-1" indicatesthatthe five-month
intervaloccurred1 EP earlierthan in our texts.In each case it cor-
rectlydenotes the boundary between groups as evidenced in
Table 5, whetheran eclipse was visible or not. This suggests the
existenceofsome scheme similarto thatofour texts,but reflecting
the actual distributionof eclipses in the earlierperiod.
LunarEclipseReportswith"5-itu"
Date SC:EP Visible? "5-itu" Reference
-746:Feb6 0:31 Yes EP-1 LBAT 1413
-685:Apr22 4: 8 Yes EP-1 LBAT 1416
-667:May 2 5: 8 No EP-1 LBAT 1416
-649:May 13 6: 8 Yes? EP-1 LBAT 1416
-631:Jun4 7: 8 Yes EP-1 LBAT 1416
-598:Feb 19 9: 1 Yes EP LBAT 1420*
-591:Apr2 9:16 Yes EP-1 LBAT 1420*
-588:Jul - 9:23 No EP-1 LBAT 1420*
-577:Jun- 10: 8 No EP-1 LBAT 1420*
-526:Apr - 13: 1 No EP B.M. 37276*
-422:Aug - 18:31 No EP-1 LBAT 1426
* = unpublished
The (unpublished) textB.M. 37276containsbriefobservational

reportsat consecutive eclipse possibilities,beginningwith EP 1;
SC 13 (3** Cambyses = -526). Only the top of the firstcolumn
is preserved,so it is impossible to tellthe structureand fullextent
ofthe text.Nevertheless,it begins withthe earliestSaros Cycle for
which the five-monthintervalsin our textsagree with the eclipse
record.
The earliestuse of the term"5-itu" to designate the beginning
of a new group of eclipse possibilitiesoccurs in the reportof the
eclipse of - 746:Feb 6. This eclipse was the firstin Nabonassar's
reign and is the earliest detailed eclipse reportwhich we have
fromBabylon.18
The,scheme underlyingour textsis consistentwith the histor-
17 While theoretically possible, eclipses separated by fivemonths are seldom, if ever,

observed and eclipses separated by eleven months are rare. In the 500 odd years covered
by Table 5 thereis only one instanceof two eclipses separatedby fivemonthswhich might
have been visible in Babylon (EP 16 & 17; SC 21). The firstof these had a magnitude of
only 0.1d, while the second was only marginallyvisible, if at all, beforesunrise.
18 The eclipse occurredin monthXII of the accession year of Nabonassar (i.e., at EP 31;
SC 0) and begins a series of consecutiveeclipse reportscoveringat least Group V of SC 0.

:, U
0l4-
I
''?
I , _> >.~->2t I~> xE55xx5z xxx E>>
> C1O tA4en N r-
_. I _ -, x 00 -I s ^ 0
o~ 2 > 2x2xE5 555Sx54- 0
-x>-xzx2x - W=5 0
5-x5x-5-x
o 0~~~~~~~~~~~~
n~~~~~~~~~~c co
~~~ ~ ~ ~ ~ ~ 4-4
0~~~~~~~~~~- 4
e
Q.) C 0l 0
0 1l> oo-~o~- 55 x0 eI T -
05 x x
cn V_ 'Q e
M4~4 4-t X t s - N
-
0) 4-4~~~~~~~~~~~~~~~~~~~~~~~
1 0 a
ral 00 .
00'
O
0 S ~~
>,
k 't
-> ~ *
o 0 00
0 4-- 0
- - 0-
F '0 Rj >0 N g 0 e e N e A t 0 t -_ e I
0
,3~~~~~~~ 222
-
_555 > > > >
>XX X >X= >XXX
EH |: > ,,5x5 xzxzx2x2

* x=:>>
xzxzxzxz
x2xlx5Rl
-
~ ~~ ~~~~~~~~~~~~~~1l x :2
0
*' > 5> x5> x2 x2x2
~ ~ ~~~
..L4"
.I
Ei > x>
5x5x5x>
- 0 - t o a C . en t--0
-~~~~0 0~~ 0 01 W) t0- - -0C,
w >^_>p 5.
,o..~~~~~~~~o a x<
>fi * x > >f
> > > >
v Ue Q - 4-4 - '0
'0~C1 C4 14 C4 C4 1
0. I4
.
U ~~ 4-en
0 - 0 '0
44O
4
4t '0
F
n'TW)' 5 t- 000'la, o _
= cs ~ Ar oFXoRNN 4 -4 'n A -- '0a '0 _NeaeO
-4 en " --'0 t 0

ical eclipse record forfifteenSaros Cycles beginningwith SC 13

and extendingthroughSC 27. Table 6 gives the dates of all lunar
eclipse possibilitiesforthisperiod, arrangedas in our texts.Rather
tidily,the table begins withan eclipse possibilityin monthI of 3**
Cambyses (-526) and ends with one in month X112of 54 S.E.
(-256). It is not impossible that the Saros Canon originallycov-
ered this same period.

TextD ("Solar Saros"): B.M. 36754 (80-6-17,488 + 564)
PLATE 4.
fev. .
24

TEXT D
Contents:Regnal years and months of solar eclipse possibilities

for,at least, year 11 ArtaxerxesIII to S.E. 53 (-347 to
-258), arrangedin eighteen-yeargroups.
PreviousPublication:mentioned as LBAT *1430.
Transcription:
Table 7; Translation:Table 8; Photograph:Plate 4
Colophon: ki-sa-risa' [. . .] which means "the knots(i.e., nodes) for
[solar eclipses?].'"19
DescriptionofText:
TextD is a single fragment51/4" wide by 31/4" high. Partsof the
top, rightand leftedges are preserved,althoughthe surfaceitself
is destroyednear the edges. The surface is divided into six col-
umns of 38 lines which continuefromobverse to reverseand are
marked by verticalrulings. The columns are of uneven widths,
rangingfrom3/4" (Cols. I and IV obv.) to 17/16" (Col. IV rev.), and
the text is generally less well finished than the Saros Canon.
Column VI is blank where preserved,and the verticalrulingsep-
arating Columns V and VI is not continued on the reverse. It
seems likelythat Column VI was leftblank throughout.
The formatofthe textis similarto thatof the Saros Canon (Text
C). The obversebegins at thebeginningofa group (i.e., aftera five-
month interval)and contains 20 lines, the reversecontaining18
lines. Horizontalalignmentis observed throughout,so thatdates
in a given line increase by 223 months fromone column to the
next. Five-monthintervalsare designated by "5 itu" and by hori-
zontal lines which extend across the tabletfromedge to edge, so
the distributionof eclipse possibilities is the same within each
Saros.
The text'suse of regnal years is as follows:it begins with years
10-21 ofArtaxerxesIII followedby years1-2 ofArses and years1-5
ofDarius III. BeginningwithColumn II we have years1-7 ofAlex-
ander III (the Great), followed by years 1-8 of Philip III (Arrhi-
daeus). So farno king's name has been preserved,but following
year 8 of Philip we have years3-6 of Antigonus,year-number"3"
being followedby a small "an"; then years 6-11 of Alexander IV,
but withouthis name; and finallyyear 7 of the Seleucid Era, the
19See Aaboe [1972], note 22.
25

TABLE 7
TextD)
Obv. I I JIlL
1) [iO be 5 i-t]l [5 2zL'5iEA]1[5 7-Lz 5 i [11te 5 ita, ' [35 se-5,& I E ,
[1:~~~~~~~~~~~1
]l[zr Ft
rSlFz 1 kin ]I [ki- It
[ ab~ ] [ ]z2h zz 4Z ]I[ g. leII[
12zu. 2 gu 6 iti IS iZr3i16[k. ]I[II
5) ab ab z7z Z,Z z [ 3]I
13gu. 3 gu 7 Wtt 3Z
). zOL [ Ii
a.b ab ab z{t ztz [ Ii
[I]
Ljsjig + LL _ _
' L_________
a 5 itu ynL5 itLu gan5itu gan5 Ltu ab 5 itu

10) 1 [5 9 5 gu4 9 sti 0 si (lo
[ aptr3
g apin gan
S5ig
3atf
gan
[16qt ] 6 guf 23 sig 4! sig
[ cpirl ap'Ln aptA angr
[I7bar_ 11 7 gLt 11gL4 24&q . 42sig
15)I kin5 ii duuSitu du6d5'5it.i>
it.
t.PStud La
[ Ge ] dir I se bar 25 bar '43[bar I
[1 kin ' [I kijLnj dz6 du6 [au6 I
ge gIt e I dir 2 bar [44bar I
t[I izi ] 2 kn J gJknj dll [ u44
20) [.t z ] [ ge I[ e IL[LdSrj[ 1[q
I bar I I (20
Rev. 1 11 I lV IVi
21) 1[20zi
i I [3 Lzt k~m ]I[ 4Jt361 IL
J' [q kin ]'[2z
'Lzz ]|
I_ ztz _ IL N __ ]I _~rd_]I[
__ 3[S __ _ Ii
[ 5 &J|[FsI 5 itit [o t 4z 5 ]' -zSi> 5,
7.,l '46 izi 5 ituI
t ab ] ab I ab I zLz I I zzz I [
?
25) lsig ] 5 t
's Ir,[u 2.t9gu I1471Fi-zilQ
I gan [ab ]rab I ab ] fzt]2
[2 sig [6 sig I I2rz' 30rW1i V[i15i
J gari 3[I]ari
7 13
ab ab ab
q z
[i sig 3 sig sig 51 tu.
30) [ganl
[an San ab ab 60
[2 bar51t 99 5 itt 14 9q 5 iti. 3Z f 5itu 50 St 5 'itu.
dA6 I apin apine. apin ;an
[3 bar 3 abar 159af 339i, 5/ 3t
[ d.63 do6 apin apin api1t
35) 4 barI ] bar 16bar 34l t* 52 u. S5
[ kin-2-k6] da6 dc6 apiyt apire
F te 5 bar 17bar 35 bar 53 qu,.f
ms) [5 kinLI kin-a-ka-1
| u6 ad6 | piI,
40)
-a-
A 1" 1
S E
IwQi. .. .. .. .. .. .. .. .. .. ..] 11 (40

TABLE8
Tex D
1 IO ffw. 5,E[ 5MO. 56 /7o 7F 5,no 3GSvZ .

(a ~~~~36Q:E
i1 z 37 iC
5 3-'V 3my ;7_0 I 39 1y
x~~~ 2o= ~~ 3g7V 2
f~~Zii v-!
L~7~7 39 27
si ?mto. 7R 5mo. () 5 3E S ?>i.
10. /5-y z g 2 o
1Eli'- v /0X 1
I?r1 7 ff (1ff 2 ff __1lF _ _
1-71
/ $o N3
j5.7e1 I
21. ZO 3 V r
27 1i
zg ( ly qg
W QO@sX qo
30. X _ _ _ _ _ _ _ _
LI sto. giT . (9f S>i. 3.2J S to. S
3131 IlW 33ff

2v~~~~~2
351 I LQEL I |171
/I | 4

"7" being followedby "se." The Seleucid Era is then used to the
end of the text.
Commentary:
Our text,which we call the "Solar Saros," gives the monthsof
solar eclipse possibilitiesfor5 (solar) Saros Cycles beginningwith
SC 23 and extendingthroughSC 27. That it concerns solar rather
than lunar eclipse possibilitiescan be seen fromTable 9, which
compares the dates fromcolumn VI' of the Saros Canon (SC 25)
with those fromcolumn III of the Solar Saros. The column from
the Solar Saros begins correctly3 EP earlier than that fromthe
Saros Canon. Indeed, the five-monthintervalsin the Solar Saros
are halfa group out ofphase withthose ofthe Saros Canon while
the relationshipbetweenthe monthsis just what we would expect
for solar eclipse possibilities(Figure 1, p. 17). Finally,in Table 9
we show dates of lunar and solar eclipses, taken fromOppolzer
[1887]. In every case but one, dates fromthe Solar Saros corre-
spond to dates of actual solar eclipses (not necessarilyvisible at
Babylon), just as dates of lunar eclipses correspondto dates from
the Saros Canon.
The one exceptionto the agreementbetween the Saros Canon
and the Solar Saros is thatthe lattercontainsgroups of 8-6-8-8-8
EP, whereas we would expectgroups of 7-8-7-8-8 EP. This anom-
aly also accounts forthe sole inconsistencybetween the dates in
our textand those of actual solar eclipses, whereinall ofthe dates
in line (EP) 15 are one monthtoo early.The simplestexplanation
is thatboth boundaries of Group II are (consistently)in errorby
1 EP-i.e., EP 8 should occur 1 monthearlierand EP 15 one month
later.This would restorethe expecteddistributionof7-8-7-8-8 EP
and leave all dates in our (emended) textin agreementwith dates
of actual solar eclipses -the firsteclipse possibilityin each group
correspondingin each instance to the earlier of the two "one-
month"eclipses listed by Oppolzer.
The purpose of textssuch as the Lunar or Solar Saros is not
clear to us. They are obviouslynot observationalrecords,since at
a givenlocationeclipses materializeat only a fractionofthe eclipse
possibilities. Nor are they forecastscontainingeclipse warnings
forthe futureforthe trivialreason thattextswhich give dates in
several reignscannot have been writtenin advance of the events
they describe, at least not in theirentirety.
One possibilityis that the textsserved as a guide to observa-
tional data on eclipses in the corpus of AstronomicalDiaries, al-
though the use of different dates than those used in the Diaries
to understandin thiscontext.Anotheris thatsuch texts
is difficult

TABLE 9
OppoLZet/uot,
cI. [Antig. 5& 53] -31i lf#-13
~~~o,I
No.. Date,
adD ______ g I - ~~~~~~
lo
3Sp~ ez X
~~~~~~~~~~3
0 11& ~ QiYY4
S E. I 7
5Vmo. (Alei . s)6 V 22 Y7
13 9' 309 Feb Y, *J -3o0Feb ZO 2/YS
2EC 7L A4ug15 21q
f38Q- 308Ja,#"2s-| x -3o0 Feb 9s3o I/
13 83 I/ aL1 3 __ S?_ | JccL S/Au 't 21 S7z
139Y -3o7 Jad 3 lLK 5 To. Dec Z9 21/53
/3's, Jua 9 -iy 9 la -3u7J^,25 21is6
Xg I Dbec18 26S767
5 if Smo. to ffi -306JuwI. 19 2i S
I 1_ I becg 8 2/67
1384 -3oSla,jQo 6 if f1 -3OSJVn 3 21sR
13 87 A/ovIZ 1 S?mMo. N/ov Z' 2(679
/398 -3oyafy1 g 7 if 5E. 7 I -30'YAprU3/11ay2Z 2l6o/l
1389 Oc31 | *i i | Oct7 1(7
1390 -3o3 A,pr2g | 812 -303 Apr 12 243
139 Oct21| 8 g OcJk6 2149
EiE5 o. 2
-3oZAP0Apr 2f(65
'
21
9 [93 1 Se1oZS 264
/312 -301 tlae 7 * ' I ttar 23
-3o 2 67
1393 Sep / to i' [to]: 1 768 j
(399 -3ooFeb2s : -3ooFeb 10 2164
13is, Auzo il: A g s 2170
I~~~ ~: ~ ~~~~~~~~~~~~~~
IiJa;,,qg 9 21 ?/
12 Emo. 12V" 26 2/7z
1394 -29gJa?& V (*' Z9
-218J./ tf8 2( 73
1397 J.0 i 13 ffE 13f Jt S .217y
1398 2ecAZs TX - 21 7Ja^ 8 21 ?S
1399 -297JJtu ('f If 1Smo . Ju" SAIuaL 121 7/7
/900 becJ9 | Nov zq 21 7q
I// -Z16dJ ( ff! 16 fa
/f -_296 .q 21 7q
I_* _ _ Ei /o g/7 /o
1 16 Ir s . 1l 3
zTrs/y t.3 2181
I l 71
1- A0ov 7 218Z
19o2 -29q9Apr18 17 I 17 1 -29y Kay 3 z2g3
/Yo3 Oct /_ IOt '27 .2 8y
IoYO -293Apr 7 | 18 II_ -1293L /pr.2 2ZgS
NYOS OCt L I 0Dat No
/Yo6 29211a.r24
S6o 81 lr Saros5 SolarEcI ipsc
Oppl2peZr;
served to identifyeclipse possibilitieswhich were similarly

affectedbylunaranomaly.Thisarisesfromthefactthat223 (syn-
odic) monthsverynearlyequal 239 anomalisticmonths,so that
each line in our texts-comprised of syzygiesseparatedby 223
months- designateseclipse possibilitieswithroughlythe same
lunaranomaly.Finally,textslikethesemayhave servedas either

TABLE 10. BabylonianYears Accordingto:

PTOLEMAIC
Julian SOLAR SAROS CANON
Year PD-3 SAROS CANON (Alexandria) NOTES
-334 1 DAR23 I DAR13 I DAR13 DAR13
-333 2 2 2 2
-332 3 3 3 3 D -322: 3 DARI3
-331 4 4 4 4
-330 5 5 5 1 ALEX3 D -330: VI - Alexander enters Babylon
-329 7 ALEX3 1 ALEX3 1 ALEX3 2 D -329; LBAT 1397 and BM 37043: 7 ALEX3
-328 8 2 2 3 D -328: 8 ALEX3
-327 9 3 3 4
-326 10 4 4 5
-325 11 5 5 6
-324 12 6 6 7 D -324: 12 ALEX3
-323 13 7 7 8
-322 14 1 PHIL3 1 PHIL3 1 PHIL3 D -322: 1 PHIL3 !; 11/29Alexander dies.
-321 2 PHIL3 2 2 2 D -321: 2 PHIL3
-320 3 3 3 3
-319 4 4 4 4
-318 5 5 5 5 DC: 5 PHIL3
-317 6 6 6 6
-316 7 7 1 ANTIG 7
-315 8 8 2 1 ALEX4 DC: 8 PHIL3
-314 2 ALEX4 3 ANTIG 3 2 ST: 3 ANTIG
-313 3 4 4 3
-312 4 5 5 4
-311 5 6 6 5
-310 1 S.E. 6 ALEX4 1 S.E. 6
-309 2 7 2 7 DC: 7 ALEX4
-308 3 8 * 3 8 * D -308: 8 ALEX4; Seleucus = General
-307 4 9 * 4 9* DC: 9 ALEX4
-306 5 10 * 5 10 *
-305 6 11 * 6 11 * SKL: Last year (?) of ALEX4
-304 7 7 S.E. 7 12 * SKL: 7 S.E. = 1st yr. Seleucus = King
-303 8 8 8 1 PTOL1 CT IV: 8 S.E. (earliest dated text)
-302 9 9 9 2 D -302: 9 S.E.
-301 10 10 10 3 D -302/301: 'Year 4' (of Seleucus)
-300 11 11 11 4
-299 12 12 12 5 D -299: 12 S.E.
Abbreviations: *.Posthumous
Ct IV: CuneiformTexts fromBabylonianTablets etc. in the BritishMuseum, vol. IV.
D -XXX: Diary forthe year -XXX; Sachs - Hunger[1988].

DC: DiadochiChronicle;
Grayson
[1975].
SKL: Seleucid King List; Sachs and Wiseman [1954].
ST: Saros Tablet; BM 34576.
aids or exercisesin establishingfairlylong stretchesof local chro-

nology in antiquity,as theyhave in modern times.
Excursus-Post Achaemenid Dates:
BeforeAlexander,Babyloniandates reflecteda simple and con-
sistentpracticewith respectto regnalyears: namely thata regnal
year,once begun, continuedeven when a new kingacceded to the
thronein thatyear. Thus a king's accession year was counted as
his predecessor's last, so that his firstyear was his firstfullyear

as king. The practice avoided ambiguitywith respect to regnal

years.
Macedonian usage, in contrast,was thata king'sfirstyearbegan
withhis accession. Thus when a kingdied or was overthrown,an
eventcould be assigned to eitherthe last yearofthe departedking
or the firstyear of the new king.
Compounding the confusionin the period fromthe end of the
reignof Darius III (-330) untilca. S.E. 11 (-300) was the absence
of a single conventionon when reignsbegan, or even on who was
king. This is reflectedin Table 10, which describes threedifferent
methods of datingfromBabylon,togetherwith thatpreservedin
Ptolemy's"King List."
The firstcolumn of dates are those used in PD3, which agree
with those found in Diaries and the Diadochi Chronicle(Grayson
[1975])at least throughthe reignofPhilip III (Arrhidaeus). By this
conventionyear 7 of Alexander III (the Great) follows year 5 of
Darius 111,20 and the reign of Philip III extends throughyear 8
( - 315). For at least the firstsix yearsof the Seleucid Era, however,
the Diaries and the Diadochi Chronicledate in yearsof Alexander
IV, in contrastto the conventionused in PD3. Indeed, even as
late as S.E. 10 (-301) we find a Diary referenceto "Year 4" of
Seleucus, suggestingthat the inceptionof the reign of Seleucus
had not been firmlysettled.
In our textsthe dates fromthe Solar Saros are the least con-
sistent,the reigns of Antigonus,Alexander IV, and Seleucus be-
ginningin years3, 6, and 7 respectively.Such confusionseems un-
likely to have been wholly invented, suggesting that this
conventionreflectedsome contemporarypractice.In contrast,the
Saros Canon begins each reign including the Seleucid Era with
year 1. This rationalapproach seems likelyto be the farthestre-
moved fromcontemporarypractice,and may simplyreflecta later
attemptto establish a coherentchronology.
20 This implies that Alexander's reign began with his Macedonian accession in - 335
and thus avoids the problem thatby Macedonian conventionAlexander's accession (and
thus firstyear) in Babylon occurred in year 5 of Darius III. See PD3, 19-21.

III. TextsContaining AstronomicalFunctions

TextE ("TextM" and "TextL")
TextE1: B.M. 36651 (80-6-17,383)+ B.M. 36719(80-6-17,452)
+
B.M. 37032 (80-6-17,776)
+ B.M. 37053 (80-6-17,797)
34.<88E
T.8Es..LTEi
34

TEXT E
Contents:
Obverse ("TextM"): Regnalyearsand longitudesof69 consecutive
synodicphenomena forMercury(most likelyits last ap-
pearances as an eveningstar,Q), beginningin year 41 of
Artaxerxes I and continuingthroughyear2 ofArtaxerxes II
( - 423 to - 401).
Reverse("TextL"): Regnal years,months,longitudes,and eclipse
magnitudes of lunar eclipses for36 years, fromyear 7
of Darius II to year 24 of ArtaxerxesII ( -416 to - 380)).
TextM, Table 7; TextL, Tables 13, 14 and 15.
Transcriptions:
Photograph:Plates 5 and 6.
DescriptionofText:
An extraordinary featureof our textis that the writingon the
obverseis at rightangles to thaton the reverse.In the astronomical
cuneiformliteraturethis has been met twice before.21In all three
cases the contentsof obverse and reverseare diverse: in our text,
the obverse concerns Mercurywhile the reversehas to do with
eclipses.
There are severalindicationsthatthe textis a copy ofolder orig-
inals. In Reverse Ila, 28, we see what is most likely hi-pi,hi-pi
(broken,broken),verysmall and shallow,writtenthrougha very
fainthorizontalline. In Reverselb, 31, 32, and possiblyin Obverse
II, 13, we find"UR," the older formof Leo, while ReverseIlb, 31
and 33 have the later"A." In the materialrelatedto ACT thisusage
has been encounteredonly once before,viz. in B.M. 37024,a pro-
cedure textforMars, SystemA (Aaboe [1987],3), where both are
also attested.Further,the dates ofboththe Mercuryand the lunar-
eclipse data are in the neighborhoodof - 400. On the obverse,the
writingof9 in the older nine-wedgeformor the morerecentthree-
diagonal formis unstable. The reverse employs AB for ABSIN
(Virgo),a usage without precedent. Finally,the procedure text
ACT No. 816, which is intimatelyrelated to the Mercuryside of
this text, is clearly a copy, as is indicated by the presence of
[hi-p]ies-su ("recent break") in Section 5, and contains features
which seem to be non-standard.
21 TextJ(B.M. 36744)in Neugebauer and Sachs [1969] and TextsG and H in Aaboe and
Sachs [1966]. In addition, the so-called "Saros Tablet" (B.M. 34576) has what may be an
eclipse reportwrittenat rightangles to the main textat the bottomof the reverse(private
communication,C.B.F. Walker).
35

TextE2: B.M. 37162 (80-6-17,912)
TtEe24;^,,;A
36

TEXTS CONTAINING ASTRONOMICAL FUNCTIONS 37
All joinswereexecutedbyus; thoughtoo muchclayis missing

fora physicaljoin thereis no doubt thatthe fragmentB.M. 37162
(TextE2) derivesfromthe same tabletas TextE1. Since TextE2 pre-
serves the upper right-handcornerof the obverse (as well as of
the reverse)and since both obverse and reverseare preservedal-
most to theirbottomedges, we can estimatethe originalextentof
the textwith confidence.Its dimensions, when unbroken,were
51/2by 61/2inches.
On the obverse horizontaland verticalalignmentsare strictly
obeyed, even by sexagesimal digitsin theirproper places.
CriticalApparatus:
The digit9 is writtenin the older nine-wedge formin obverse
1,17,11,10,and 11,15while the cursivethree-wedgeformis used in
I,19 and 111,12.As noted, the 9 in 1,17is an errorfor8.
For Leo we findthe older UR in Reverselb, 31, 33, and possibly
in Obverse 11,13,while the laterA is used in reverseIlb, 31, 33.
Obv. 1,17: 13,19,2,48,45:the nine-wedge9 should be 8, an isolated
errorwithoutconsequence.
Obv. 11,14-18:24,51,48,48,45 in 11,15:8 should be 7, and this error
is repeated in the nextfour lines, thatis, as faras Column II
is preservedin this sexagesimalplace. If this is not merelya
copyist'serror,itwould show clearlythateach columnis com-
puted independentlyfromits initialvalue, forthe values in
the same lines of Column III are correct.Where Column II
is not preserved we have restoredthe correct7, though it is
likelythat the texthad 8.
Rev. Ia,14: 16,30: should be 15,30,an isolated errorwithoutcon-
sequence.
Rev.IIa,20: 15 Taurus (Taurusverydamaged) should be 24 Taurus.
Rev. Ia,20 and Ib,19: The traces of Month VIII (APIN) of year 16
Darius II should read Month VII (DU6). Alternatively year 16
ofDarius II would have to have a month X112, whereas TextA
explicitlygives V12 (Rev. VI', 5') as noted in PD3.
Rev. IIa,32,34 and IIb,31,33: Year 20 of ArtaxerxesII must here be
normal, while year 21 must have a month X112. As noted
above, this agrees with the evidence of the Saros Canon and
conflictswith B.M. 35328. PD3 makes year 20 an intercalary
year,based partlyon Strassmaier'smisreadingof the Saros
Canon.
Rev. Ib,33: 35 should be 36.

Commentary-Text M (Obverse):
The obverse of our textconcernsMercury,and we shall treatit
here as though it were a separate text.The fragmentary state of
the preserved surface and the unprecedented arrangementand
structureof the table would have been insurmountableobstacles
to our penetrationintothe contents,were it not forthe procedure
textACT No. 816.
In this procedure text are set forthseveral schemes of the
System A varietyconcerningMercury.The first,which we shall
call Model I, is given in Section 1; it serves to determinethe lon-
gitude of Mercuryat moments three synodic periods apart, or
year by year, as the text apparently has it, for three synodic
periods of Mercuryfall only littleshort of one year. The para-
metersof the generatingfunctionof this scheme are:
Aries 30? to Leo 30? : w1 = -16;52,30?
Leo 30? to Cancer 20;37,30? : w2 = - 16?
Cancer 20;37,30? to Aries 30? : W3 = - 10?
p - 6! = 20;42,51,52,30.
Z 19,12
Since the w's are negative,the zodiacal signs are here, as in the
text,to be taken in the sense opposite to theirusual order.
The scheme is based upon a distributionofintervalsin the three
zones of length
I1 = 0;0,52,44,3,45 12 = 0;0,50? 13 = 0;1,2,30?
and, as characteristic
of SystemA, one step corresponds to Z =
19,12 intervals,of whateverlength.22
Since
21.Z = I + 5,29,
21 steps, each correspondingto three synodic periods, lead to
a total lag of 5,29 intervals,which, in their respective zones
amount to
5,29*I1= 4;49,9,36,33,45?
5,29-12= 4;34,10?
5,29*13= 5;42,42,30 .
These are convenient checking parameters and, indeed, 21
triplesynodic periods are verynearly20 years. Section 4 of ACT
22 See Aaboe [1964], 219.

No. 816 is, as a matterof fact,concerned with preciselythese cor-

rectionsfor20 years,but gives them as
4;49,170 4;34,15? 5;32,51,300
though the zones are explicitlydescribed and identicalwiththose
of Model I. This is not the only instanceof internalinconsistency
in this text.
In Section 3, which is only partlypreserved,are foundelements
of a scheme, called SystemA3, forproducinglongitudes of char-
acteristicphenomena of Mercuryone synodic period apart. Be-
cause the preservedparametersare incomplete,we have no indi-
cation of the extentof the zones. As we shall see, however,the
parametersof this scheme-we call it Model I-must have been:
Leo 300 to Aries 300 : w1 = 1,50;56,150
Aries 300 to Cancer 200: w2 = 2,11;28,53,200
Cancer 200 to Leo 300 W3 = 1,45;11,6,40?
p = = 18,39
- 3;9,7,36,20,
This model is based on a distributionof 18,39 intervals of

length,in the respectivezones,
I1 = 0;18,45? 12 = 0;17,46,40? 13 = 0;22,13,200
to be taken 5,55 at a time.
Since forModel II
3Z= -54 (mod H),
the application of threeconsecutivesynodic arcs leads to a lag of
54 intervals.Indeed, the intervalsofModel II multipliedby 54 are
preciselythe w's ofModel I, but the period ofthis derivedmodel:
HI 18,39 = 6,13 = 20;43,20
H - 3*Z 54 18
is differentfromthat of Model I because of the displacement of
one endpoint of a zone fromCancer 200 of Model II to Cancer
20;37,300of Model I. Models I and II are thus not strictlycom-
patible. Neugebauer [ACT, 428] resolved this difficultyby as-
sumingthatCancer 20;37,300was probablyan errorforCancer 200
which, indeed, seemed a very plausible assumption until the
present textwas understood.
For quite surprisingly,the structureof TextM turns out to be
such that each column, by itself,is computed strictlyin accor-
dance with Model I, while the firstline obeys the rules of Model

II. Specifically,the entries of the firstline of the text-i.e., I(1),

11(1),111(1),where I(1) means Column I, line 1, etc.-give longi-
tudes of consecutive synodic phenomena of Mercurycomputed
accordingto Model II, while each of the columnar sequences
(i I(1), 1(2),1(3),1(4), .. .
(ii) II(1), 11(2),11(3),11(4),.. .
(iii) III(1), 111(2), 111(4),.. .
111(3),
denotes longitudes of synodic phenomena of Mercurythreesyn-
odic periods apart computed strictlyaccordingto Model I.
Thus it is clear that the 69 entries,read in the order
(iv) I(1), 11(1),111(1),1(2), 11(2),111(2),1(3), ... I
give longitudes of synodic phenomena of Mercuryone synodic
period apart, and thattheywould representthem systematically,
obeying a rathersimple period relation,if Model I and Model II
were trulycompatible.Indeed, the sequence (iv) above conforms
preciselyto the rules ofModel II untilwe reach the transitionfrom
111(3)to 1(4) which is the firstplace where the disparitybetween
the two models, viz., whether a discontinuityof the generating
functionsis at Cancer 20? or 20;37,30?,becomes relevant.Model
II, hithertohaving employed the discontinuityat Cancer 20? sev-
eral times,would yield Cancer 2;35,12,30?as a successor to 111(3)
while our texthas Cancer 2;25,50? as the entryin 1(4).
Similar discrepanciesoccur at 11(10)and 111(17).This last entry
yields explicit and complete evidence for the discontinuityat
Cancer 20;37,30?,while the situation at 11(10)is only partially,
though sufficiently, preserved. (See Table 11.)
It is clear,then, thatthe procedure textACT No. 816 must not
be tampered with despite internalinconsistencies,for it is pre-
ciselyfollowedin the presenttext,where relevant.We have failed
to discover any justification,astronomicalor otherwise,for em-
ploying incompatible schemes in a single text when consistent
and simple alternativeswere so close at hand.
It remains to explain the integerspreceding the longitudes in
TextM. At firsttheypresentedquite an obstaclefor,whereverpre-
served, theyare writtenso close to the followingnumberthatwe
initiallyread them as firstdigits. When this proved wrong we
thoughttheymightbe line numbers,since a continuousrun from
1 to 17 is preservedin Columns II and III. This assumption,how-
ever,was contradictedby the cornerpiece (E2), which shows that
one line precedes the line numbered 1 in Column II, as well as

TABLE 11
T@xt
M I B.M.
37053 Z
BiM.37162=Lb
Obv.'. 2. /o /YYo57sI'Ys /' / 12 r895 Y .
8, t,o YJL5 3(,9
g272S,, v2 i
'. 2 i/
'O__ 2 fo,S8,2sIg,9Y .,
?-ig )(
... 3o2' 3 7,2S;(8,VS
IV S 7 '12/g,8Y_
i ... /2,25;<o Y 7Jo, ?2s,/g9SK - 2i' 7 2S,IS9
ss
22,2657 \ C 2,, 7^298 'p .. /s,
..212tSO- __ t 3ZS2g~,/g,'/S .. Y2y2.,J2SIg,'/S
... //,2J2,Y8s X /3j Y37,2,2s,t8,,s n 0
... 3, 2,VY5 Q4
, /S52-1 o z3zo 6
S~~~~~~~~~~k
....
l7 ,2
J,22,Yo, (Y
2.Yi'S,Vy t
/ ?4gY9S tJY /s2G g3t S' /
./33 yS,2 , Io YS_
'z0 S-
, /3,1q,D,YS
..B.3,!,YgS
532,f8,s D ...~~~~~~13
Y .. yly v,Y8
>S/ y? Y2S3l7
3720 / S,33Yo
B.236657V ,
1t ..2SI A?Y
e1,3,O.
bythe numbrs 1a in Co lum, I, lns 2 an 1 233s,fo ther

is n2,Yo
very good reaSon
3i, dYulo to ue a b ng 20
these3numbers idct rega, yas 1fo te co
reigns aled us
to dae th ctents of T w hg pusi
bility.Bt/e en 750 and 03the are, l wo s3t, thaY
OrInvsIgaio f the only reann osiiiynml,ta
would
fittheregnalyearsf. Te e6
sli wu the first

1 alne
hav b , ines
ya 13 of E orhado
is no very good reason to use a numberingmodulo 20.
Our investigationofthe onlyremainingpossibility,namely,that
these numbers indicate regnal years from three consecutive
reigns, enabled us to date the contents of Text M with high plausi-
bility. Between -750 and 0 there are only two situations that
would fitthe regnal years for a Mercury phenomenon. The earlier
solution would have the firstline begin in year 13 of Esarhaddon
(-645), and here the phenomenon must be the firstvisibility of
Mercury as a morning star (r). The later possibility would assign
the firstline to year41 ofArtaxerxesI (- 423) and identifythe phe-
nomenon as the last visibilityof Mercuryas an evening star (9).

~~~~~~~~~~~~~~~~~~~~~~-
- - -}\4!-sQ$OeOi mu
fO
4 0XIN
t Q F
<a n WF t d q C s r t_ t A t p 99 WS xyE~~~~~~~~~~~~
m
oc, ~~~~~~~~~~-4
'3C>Fc2('mcjeXb
bc wkS W - t4VE tX @1 t- > }\1\1W 1t \\-}\ \|\ x
>~~~~~~~~~~~,
too -c o =n c-J
* 0_
-~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~V I
-~~~~~~~~~~~~~~~~~~~~~~~J z ()C
? a<a w G $>a- Q
> - 6 St -X
fi WO
H> Ee^ >m
oA~~~~~~~~~~~~~~-,O
St Q
%
>tR
4
m
N~~~~~~~~~~~~~~~~~~~~~~~~~~~~f-
% n <Z
Q< C-4 , c_ e
4 tS
a
; $ <2 t$tt*i E <? Q -t,t@ 2
mn t > 8 oos Z m> nvJ

=
w~~~C1
o>t>+>t
E bo c
W Qi i I I I I I\ i I Fl > > > I ? g B B Q $
S~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~t
W :t- - mn=%Z=
S~~~~~~~~~~~~~~~~~~~~c4 C -
w~~~~~~~~~~~~~I: r-;1-,1 kI~j \
@ L < a G 6 > W6 Dm
g W~~~~~~~~~~~~~C-
-~~~~~~~~~~~C-l
t o - m* a- WWF b o'i SW
A
2
4
u
>d
m C-

A comparison of the longitudes in the textwith modern com-

putations23shows that only the later possibilityneed be consid-
ered. Specifically,forthe earlierdating we find
-30 < Xtext- Xmod.< + 16
which is not acceptable, though a systematicdifferencebetween
Babylonian and modern longitudes of some 5?-10? is to be ex-
pected. Forthe laterdatingthe relevantdata are displayedin Table
12, and the differencesnow vary smoothly,and all lie in the
interval
+4 < Xtext- Xmod + 130
TextL (Reverse):
Text L contains two double-columns, each of which presents
data for38 lunar eclipse possibilitiesor one Saros. Foreach eclipse
possibilitywe are given the regnal year and month,longitude of
the correspondingfullmoon, and a numberclosely relatedto the
"magnitude" of that eclipse possibility.
Withineach Saros the data forsuccessive eclipse possibilitiesat
the same node are presentedin separatesub-columns.(We call the
double-columns I and II and theirparts Ia, Ib, IIa and IIb.) Col-
umns Ia and IIa concern eclipse possibilitiesat the descending
node, while Columns Ib and IIb concerneclipse possibilitiesat the
ascending node. Five-monthintervalsbetween successive eclipse
possibilitiesare marked in the textby a horizontalline thatruns
across both columns where preserved. For the sub-columns(i.e.,
eclipse possibilities at the same node) such a line indicates an
eleven-monthinterval.
The regnalyear numbersalone allow us to date the contentsof
the textto the 36 years fromyear 7 of Darius II (-416) to year 24
of ArtaxerxesII (-380). Thus Text L begins with EP 6 of Saros
Cycle 19 (see Table 6, page 21), in contrastto TextS,24which be-
gins withEP 1 of(solar) Saros Cycle 16. No kingname is preserved
in textL1, but textL2 (which was identifiedas part of TextL only
after we had dated L1) has a small ar-s'ufollowingyear 7 of
ArtaxerxesII, whose name also appears elsewhere as Arsu.25
23 The approximatedates of the Mercuryphenomena were computed fromSchoch's
tables in Langdon, Fotheringham,and Schoch [1928], 103-105and X, with the corrections

proposed by van der Waerden [1942].
24 Text S gives computed and observationaldata forthe 38 solar eclipse possibilities
comprising(solar) Saros Cycle 15 (- 474to - 456). It was originallypublished by Aaboe and
Sachs [1969], and is augmentedby TextG here. It is discussed in HAMA, 525-528; Moes-
gaard [1980], 78-79; and most recentlyBritton[1989].
25 See Sachs [1977].

This chronologicalsupport was, of course, welcome. However,

the cornerpiece L2 raised problems which we have failed to re-
solve to our satisfaction.To set these problemsin reliefwe present
firstour reconstructionof the textfromthe evidence provided in
L1, ignoringthe added evidence fromL2 exceptforthe suggested
position of the uppermost eleven-monthline. Table 13 gives our
transcriptionof L1, while in Table 14 we have separated the col-
umns more clearlyand reconstructedthe longitudes and magni-
tudes fromthe data preservedin L1.
Longitudes:
Withineach sub-columnthe longitudes are computed accord-
ing to the rules that to a twelve-monthintervalcorresponds an
effectivedecrease in longitude of 10;30? and to an eleven-month
intervala decrease of 10;30? and an additional full sign.
The solar travel according to this scheme correspondingto
these time intervalsis, therefore:
(in 12 months) d12X= 1 rotation- 10;30?; and
(in 11 months) d1jX = 1 rotation - 10;30? - 1 sign
Furthermore,since a Saros contains 14 twelve-monthintervals
and 5 of the otherkind, we get forthe totalsolar progressduring
one Saros:
(in 223 months) d223X= 14*d12X+ 5*d11X
= 19 rotations - 5,49;300
= 18 rotations + 10;300.
This leads to a solar progressin 19 years of
d235X = d12X + d223X
= 19 rotations + 0;O?,
which is a precise statementof the nineteen-yearcycle.
This relation,we are sure,is thebasis ofthe scheme,which rep-
resents a simple, but clever,device to avoid using mean motion
which, incidentally,would lead to non-terminatingsexagesimal
fractions.It can be summarized thus:
d223X = - d12X = 10;300 (mod. 3600) and (A)
d1iX= d12X- 1 sign (mod. 3600). (B)
One consequence of the scheme is thatforeach line the longi-
tudes in our text should increase by 10;300 fromColumn I to
Column II. This is trueforColumns lb and Ilb. In Columns Ia and
Ila, however,the differenceis 10;0?, which suggestsan errorin at
least one of these columns.

TABLE 13
TevtL 1 F
'30' ,,s- [BEJ
1.2, 3j,'+O It AiI
1. ]313 1930a,3.. 7 E
/F' [133,b
R 166,3caLta, 14,LZ0 /4/ /ab22G
1
Uf
5+ 3SJt9Stf La
'4. [IJy3 at~ 6 aIl 2s- 13 dt l5 a2ia Li
17. L 3 3,2-o ts 5tLZ9>3 ornxj31 33 L
I& ts apirvZ'?30 1vi,~
7,-2 BE 1' CbX %30 7s&
i9, {gt 9 (z
t6 ' 2-r 2BE
20 16 LO-PiLki Ilt 7rw"L 3,2 /So Wpi lS L7l1
2/. ~~~~~~~~17
bo-r, 7//L6' IGrj1Sgr
22 t L33o
33uLj.L ; 0F 371o
23 17 boar7,3o0gr 2]
2s. j18 ba,r 27 rLnt 33

ZS',. 2%,2O
246 2Q,3o3Th 10 SE
18 krm,
22. 5- Se, 14,3o al, 3,Yo
-
(t~~~~~t 2pb (o 0,
3o JZO t t. 1 3o,
z IY2
312 2,,;-/3o 9 Zia A z-

33, a,b 9,3oU(R 3S ZL.
ft s A 3KLY9o
33a. -t /<, 4
36. / 233SY 30o a o
25 7 f(9,30 cPO, Zo
So fartheserulesareidenticalwiththoseencountered in TextS
forcomputing thelongitudeofconjunctionat solareclipsepossi-
ofthetextswe first
bilities.To testtheconnectibility transform the

TABLE 14
Tex&L Ia lb lba
I a;, 7 2, 6 T-1 3 ]
: 8 I rf,30
mvI _ S ? W
L * t 7 W 2 30 I
: g M-8-
s. --~~~~~~~~~~~~1-3
Y- -- v r,
f 13a 7In[&r
0
9g 4-I 27,3ot 30 ,/30
Lo CI 6302
SpLZ7,3o,,~]
T%~r'73 loI
/0 X''0 Z_
I ?.
t Lz 2*2
25 *93],4' - - -K-- L-- -
[L{I X 26 300 ,zo? 13 j 1930X 7 ' 1 E/G 631 1E 30 t g

I
IS [7i]~zf/3 (4,3o&3 10,4 IZtiii L91 {?9 , /X - [ 1307/6 1I
[230
FYX 5 o 29 rr ZB - Zy Tr 3. J
IS?vii7 2G,3o t1df~

[7
2D
2, I .27 /B 3 3
-[7S8?;+
/173'
Vt
it3?
3,?
__6j__,, 2
ZY1IF,I Y __3 BE2
20 173317
L7 _0 3 2 0 .
30 L0ftgY 23
t Ei r~
15r:
0F130381<0x
21~~,30
2'to] 1x V~~4 ~ i ( e3022 J)( (0 > 8o22
f;~~~~~~~~~~~1
FJ93 3f]
*_ T )Y ZI B _ - I le; r'tf z Qot 29/3o]Q
it EF 23 Y' -Z-f i~ 7?aK 3 ( f
[3 " 0 t3 3. /o123 I ' /S Q 32?oJ
longitudesof conjunctionin TextS intolunarlongitudesat the

corresponding lunareclipsepossibilities
byaddingor subtracting
(dependingon theeclipsepossibility) thesolarprogressin halfa
monthto or fromthe longitudeof conjunction, plus 1800.If we
assumethatthesemi-monthly solarmotionis 14;400,26 and allow
foran advanceof10;3O?per Sarosas requiredbythescheme,we
can connedtthe longitudesin Columnslb and JIbofTextL pre-
ciselyto theircounterpartsin TextS. In contrast,thelongitudes
in ColumnsIa and IIa are 1;O?and O;3O?higherthanthevalues
whichagreewithTextS.
In TextS the lunarlongitudeat conjunctionincreasesin six
monthsby1750 fromascendingtodescendingnodeandby174;30?
fromdescendingto ascendingnode. In TextL thecorresponding
6 A moreaccuratevalueofthesemi-monthly solarmotionimplicit
in the12,223,or
235 monthrelationships in thelongitudeschemeis 14;33, ... Sincethevalue
reflected
14;40cresultsin longitudesat lunar
eclipse whichend in whole or half
possibilities
degrees,itsuse maysimplyreflect a desireto distinguish
easilylongitudesoflunarand
solareclipsepossibilities,
whileretaining simplefractions.

motions are 176? and 173;30? in Column I and 175;30? and 174?
in Column II. The lattervalues are consistentwith six months'
progress at 29;15? per month and six months at 29? per month.
These motions are also found in Text F (below), where several
values are connectiblewith Column II of TextL. Thus it seems
likelythatthe differencebetween TextS and Column II of TextL
was intended, and also that the longitudes in Column Ia of Text
L are in errorand 0;30? too high.
The relation
1 Saros = 223 months
18 revolutionsof the sun + 10;30?
implies a value of the year of
1 year = 2236360 months _ 12;22,7,51,. . . months,

18*360 + 10;30
which is fairlyclose to the standardvalue in SystemA of 12;22,8.
It is even closerto one ofthe yearlengthsimpliedby Lunar System
B, viz.
6,0 = 12;22,7,51,. . . months

1A
where (lA iS the mean value of Column A, the monthlysolar prog-

ress. To put the comparison anotherway, we have
(A = 29;6,19,200'm
whereas our presentscheme implies a mean monthlyprogressof
=18-6,0 + 10;30 = 29;6,19,22, . .
223 -96192, /r
We may here have a justificationof (LA.

TextL2:
There can be no doubt thatthe fragmentE2 (of which L2 is the
reverse)was once the upper right-handcornerofTextM, and thus
also of TextL. The ductus is much the same in E1 and E2, except
for the somewhat cramped writing on the reverse of E2; the
writingon the obverse is athwartthatof the reverse;and indeed,
the obverse of E2 fitspreciselyin our reconstructionof TextM as
indicated in Table 11, page 41. The reverseof E2, however,pre-
sents problemswhich we have been unable to reconcilewith our
reconstructionof TextL (Table 14).
In Table 15 we show a transcription(A) and translation(B) of
the cornerpieceL2, togetherwith what we would expectfromour

48 SARCS CYCLE DATES
reconstruction(C). In the last we have included restoredmagni-

tudes and have let 1', 2', . . . denote the line numbers of our re-
constructedtextas theyappear in Table 14, though it clearlydoes
not representthe top of the actual textcorrectly.
The top edge of the fragmentis preserved,leaving no room for
additionallines beforeline 1, which reads "(year) 7 ArtaxerxesII."
The difficultiesbegin immediatelywith line 2, where we find
tracesof monthVIII followedby 22,30 Taurus followedin turnby
a numberwhich begins with30. This is preciselythe entrywe find
in the reconstructedtext,but in line 3' instead of line 2. In line 3
(of the fragment)we find2,40,the magnitudewe would expectto
findat the end of line 4', Col. Ila. Thus farthe fragmentappears
simply to be missing the firsttwo lines of our reconstructed
text,although we would also expect a horizontalline denotinga
five-monthintervalbetween lines 2 and 3.
More severe difficulties occur in line 4 where we findtraces of
month followedby 28 Scorpio, followedin turnby an illegible
II
sign. The longitude, 28 Scorpio, is close to that found in line 2'
(Col. Ila) of the reconstructedtext(28,30 Scorpio). It is also con-
nectiblewiththe longitudesin TextS, unlike the otherlongitudes
in Column Ila.
TABLE 15
TextL A c
1. 17r-S 7 krtyerxetsl J.Zg3 '
2 r 22,3o I[ VI
vg 2
.2Z,30 3o3ox Z 7 X .Z,30 [333
3. 202o 3.) 7 91 zzA?o& 3L/,40
M. g
28gcr A7/ XI 28,s9 L21'&1
irIMl 23A3? 3 Z3FjfOJ

; s' g t7r L."
\1avnk.)
(b bla^i) ,R s, j 4,6' gm E/3
~~~~7,3a
7c zo3 J
?f,2o
7) ~~~ I3orC 2

Finally,in line 5 we findtraces of 3 Gemini followedby 23 and

what could be 40. This longitude is found in line 1' of our recon-
structedtext,where we would expect, however,24 forthe mag-
nitude. Followingline 5 the fragmentis blank forat least two and
possibly threelines.
Our fragmentthus gives jumbled data forthe firstfourdates in
Column II correspondingto lines 3', 4', 2', and 1' of our recon-
structedtext.Furthermore,the.blank lines in the fragmentshow
that the entrieswere not continued and imply that at least some
of the entriesin Table 14 were omitted.Finally,two data disagree
by small amounts with our reconstructedtext.
We have no satisfactoryexplanation for this anomalous frag-
ment,apartfromspeculatingthatall or mostofthe top ofthe orig-
inal was broken away somewhere above line 11'. If so, then the
corner fragmentcould simply reflectcalculations of some of the
missingentriesforthe top of Column II set down in wrong order.
EclipseMagnitudes:
Followingeach longitudein TextL is a numberwhich expresses
the distance of the moon at conjunctionfromthe inferioreclipse
limit-i.e., the extremenegativenodal elongationat which a lunar
eclipse is possible. The units of this function,which we shall call
I(L), are equal to 1/46thof the monthlyprogressin nodal elon-
gation (d1-q),and derive from a functionwhere d1l1 = 30;40?
making 1 unit of *(L) equal to 0;40? of nodal elongation.27For
convenience (and to be consistentwith the terminologyof ACT)
we shall use the term"eclipse magnitude" to describe 1(L) and
its values, although strictlyspeaking the functiondescribes nodal
elongations ratherthan magnitudes in the modern sense.
Underlying I(L) is a function, I(6), which reflectsuniform
motionand is based on an eclipse cyclecomprising46 EP over 270
monthsarrangedin six groups separated by five-monthintervals.
The period of this cycle,
270
P(6) = = 5;52,10,26,. . . (months/EP),
46
is poorer than that of the Saros,
223
P(5) = = 5;52,6,18,. . . (months/EP).
38
page 49
27 Britton[19891,14 ff.,discusses I(6), I(S), and

relatedfunctions.A demonstrationof
the correspondencebetween units of magnitudeand units of nodal elongationis given on
page 25.

However,unlike the Saros, it leads to units which convertconve-

nientlyinto degrees of nodal elongation,ii, by the relation
= O;40(4I - 18); 'I' =3 + 18. (C)
For thisreason (at least we can thinkofno other)'1(6), ratherthan

the correspondingfunctionbased on the Saros, was used as the
basis forboth 41(L)and 4I(S), a relatedmagnitudefunctionfound
in Column III of TextS.
The principalparametersof 4I(6) are:
Units Nodal
Parameter Description of I Elongation
d1'(6) Monthlychange 46 30;40?
d12I(6) 12-month change 12 8;0?
dilt(6) 11-month change -34 - 22;40?
d223I(6) 223-monthchange -2 - 1;20?
Eclipselimits 0 and 36 + 12;0?
In Text L we find that d124I'(L) is always 12 at one node and

10;40 at the otherforan entiregroup. These motionsswitchnodes
at betweenlines 25 and 26 which is the boundarybetween Groups
IV and V (Table 16). The preservedvalues of dlt(L) range from
-29;20 to -34;20. Thus 4I(L), though related,is clearlydifferent
from'I'(6).
Because its period is poor, I(6) requiressome adjustmentto be
used as the basis of a more accurate theory.Thus fittingit to a
Saros cycle (d223= 0) requires a cumulative adjustment of +2
units of magnitude to offsetthe factthat d2234I(6) = -2. This is
very nearly what we find in 4I(S), where the cumulativeadjust-
mentis +2;40 units and the fractionseems to have resultedfrom
a desire to preserveintegralvalues in a relatedfunction.
Ifwe examinethe changes in 'I'(L) fromone columnto the next-
i.e., in 223 months-we findthat when
d12I(L) = 12: d223I(L) = -1 (Group V),
but that when
d12I(L) = 10;40: d2234I(L) = -0;20 (Groups IV, V) or
-0;40 (Group I).
If we assume thatd223I(L) = - O;20wheneverd12I(L) = 10;40 ex-
cept in Group I (see Table 16), and thatd223I(L) = -1 whenever

d12I(L) = 12, the resultingsum of all such d223I(L) over 38 EP is

-26;40, so that the average value of
d223I(L) = - 26;40 ?38 = - 0;42,6, . . ., or
d223I(L) - d223*I(6) = + 1;17,53,
Since d223I(5) = 0 and d223I(6) = -2, this implies an underlying
eclipse period of
26;40 1223
P(L) = 26;0{P(6)
2.28 - P(5)} + P(5) = 223 +
/8691
= 5;52,7,45,
which correspondsto a monthlyprogressin nodal elongationof
dlq = 30;40,13,58,. .
As shown in the followingtable these parametersare virtually

identical with those found in System B (and indeed also with
modernvalues). Furthermore, ifwe subtractthe monthlyprogress
in nodal elongationfound above fromthe average monthlysolar
progress implicitin Text L (29;6,19,22, .0), we find for the
monthlymotion of the node
d1N = - 1;33,54,36, . .,
which also agrees much betterwithSystemB than withSystemA.
p dl- djN
SystemA 5;52,7,39,... 30;40,14,30? - 1;33,55,30?
System B 5;52,7,44, . . . 30;40,14,4, .? -1;33,54,44, . ..
Text L 5;52,7,45, . .. 30;40,13,58, .. - 1;33,54,36, . ..
Modern 5;52,7,45, .. . 30;40,14,1, . - 1;33,53,49, ...?
For *(L) to agree preciselywith (the period of) SystemB, the

sum of all 38 values of d223I(L) would have to equal -26; 18 in-
stead of - 26;40. While betteragreement(- 26;20) could have been
obtained by assigning - 0;40 as the saroslydifferenceto Group III
ratherthan Group I, P(L) is nearly as close to the period of Sys-
tem B as can be arrangedwith units of 0;20 and the requirement
thatall differenceswithina groupbe the same. In contrast,equiva-
lence with the period of SystemA would requirethatthe totalof
the saroslydifferences be - 24;48,or thatthe averagesaroslydiffer-
ence for10;40 velocitiesbe - 0;18, which is inconsistentwith any
combinationof - 0;20 and - 0;40. This suggeststhatthe intended
period of '(L) was the same as that of System B, and that the

saroslydifference- 0;40 in Group I was introducedto accomplish

this.
Using the twelve-monthand 223-monthdifferencesdescribed
above, we can reconstructall of 41(L)exceptforGroup II, Cols. lb
and Ilb. The resultis shown in Table 16, where (+) and (-) indi-
TABLE 16
Text -- - i (L) -------- Saros

L Grp la lb Ila Ilb d(223) Cycle
EP# # (-) (+) (-) (+) (II-I) EP#
I [24;40 24;0 [-0;40J 6
2 I [34 J [33 1 [-1;01 7
3 1 35;20 34;40 1-0;40J 8
4 [3;0 J 2;40 1-0;201 9
5 [ 1 [ 1 [-l;OJ 10
6 [13;40 13;20 1-0;20J 11
7 II [ 1 [ 1 [-1;0 12
8 [24;20 24;0 [-0;20J 13
9 1 1 [ 1 [-1;0J 14
10 [35;0 1 34;40 1-0;201 15
11 -5 [-6 1 1-1;0 1 16
12 3;40 [ 3;20 [-0;20 1 17
13 7 [6 J [-1;0J 18
14 III 14;20 [ 14;0 J 1-0;20 19
15 19 s18 1 1-1;0 1 20
16 25;0 [24;40 1-0;20J 21
17 31 [30 J [-1;0 1 22
18 -7;20 -7;40 -0;20 23
19 -2 -3 -1;0 24
20 3;20 3;0 -0;20 25
21 IV 10 9 -1;0 26
22 14;0 13;40 -0;20 27
23 22 21 -1;0 28
24 24;40 24;20 -0;20 29
25 34 33 -1;0 30
26 [-9 1 -10 [-1;0 1 31
27 4;0 3;40 -0;20 32
28 [3 1 2 [-1;0 33
29 V 14;40 14;20 -0;20 34
30 15 1 14 [-1;0 1 35
31 25;20 25;0 -0;20 36
32 [27 J 26 [-1;0 1 37
33 36;01 35;40 -0;20 38
34 [-3 1 -4 [-1;0 1 1
35 2;40 2;0 -0;40 2
36 I [9 1 8 [-1;0I 3
37 13;20 12;40 -0;40 4
38 21 1 20 I -1;0 1 5
Total = - 26;40

cate ascending and descending nodes. Values which have been re-
constructedassuming only that values of d124I(L)are constant
within a group are shown without [ ], while those derived from
the postulatedvalues of d2234I(L)are shown in [ ]. The groups are
numbered afterour Saros scheme for lunar eclipses described
above. Only the boundaries between Groups I/ILand Groups V/I
agree withthatscheme; the restare shiftedupwards by 1 EP, thus
distributingthe eclipse possibilitiesinto groups of 8-7-7-8-8 EP.
To reconstructthe remainingvalues of Group II + (Cols. lb and
Ilb) we need to understandthe structureof the discontinuitiesat
the boundaries between groups, or, more precisely,how doit(L)
varies. Table 17a shows the known values of dl1t(L), while 17b
shows the difference between these values and the corresponding
value foruniformmotion, doll(6) = -34.
We begin by notingthatover 223 monthsthe sum of the differ-
ences between the twelve-monthand eleven-monthchanges in
41(L) and the correspondingvalues forthe uniformmotionfunc-
tion, 41(6), must cumulativelyequal the differencebetween the
223-monthchanges in the two functions-i.e.,
Efdl2*(L) - d12'I(6)} + E{dllI(L) - d12I(6)} = d293I(L) - d293*(6),
where both sums are taken over 223 months.Substitutingthe ap-

propriatevalues of d+I(6) yields
E{dllI(L) + 34} = d223I(L) + 2 - E{d12'(L) - 12}, (D)
where
d2234I'(L)+ 2 = + 1, when d1tI(L) = 12 and
= +1;40 or +1;20, when d14f'(L)= 10;40.
In columns Ia and Ila ofTable 17 thereare, excludingthe bound-
aries between groups, eight intervalswhere d124I(L)= 10;40, or
where d124I'(L) - 12 = - 1;20. In these intervals'1(L) will fall8 x
(-1;20) = - 10;40 behind where it would be if d12t(L) = d12t(6)
= 12. To compensate for this slower progress, the boundary
jumps must reflecta cumulative correctionof +10;40 relativeto
dllt(6). Furthermore,to this must be added an amount corre-
sponding to d2234I(L) - d223?I(6), which here equals +1, since
d124(L) = 12 in Group I, Column la. Thus over 223 monthsthe
fiveboundaries must include a totalcorrectionof + 11;40 relative
to the eleven-monthchange of I(6), ifthe functionis to returnto
its initialvalue plus d223I(L) = -1. As shown in Table 17b, this
is exactlywhat we findin columns Ia and Ila.
At the ascending node, columns lb and Ilb, thereare six inter-
vals, excludingthe boundaries, where d12'I(L) = 10;40. Further-

TABLE 17A. dllI(L)

Boundary Ia (-) lb (+) Ia (-) Ilb (+)
Groups I/Il -31 - 30;20
Groups 11/111 -31;20 - 31;20
Groups III/IV - 32;20 -33 -32;20 -33
Groups IV/V - 33;40 -30 -34;20 - 29;20
Groups V/I -30 - 33;20 -30 - 33;40
Total - 158;20 - 158;20
TABLE 17B. dllI(L) + 34

Boundary Ia (-) lb (+) Ila (-) Ib (+)
Groups I/Il +3 [+1;40] +3;40 [+1;20]
Groups II/III +2;40 [+2 1 +2;40 [+2 1
Groups III/IV + 1;40 +1 + 1;40 +1
Groups IV/V +0;20 +4 -0;20 +4;40
Groups V/I +4 -0;40 +4 +0;20
Total + 11;40 [+ 9;20] + 11;40 [+ 9;20]
more, in Group I, Column lb d223I(L) = -O;40, which is + 1;20

greaterthanthe correspondingdifference of 1(6). Thus in columns
lb and Ilb the totalcorrectionat the boundaries relativeto d1lI(6)
is + 1;20 - (6 x - 1;20) = 9;20. This means that the combined cor-
rectionfromthe two unknown boundaries must equal +3;40 in
column lb and + 3;20 in column Ilb. Since d1lI(L) must be an in-
teger at the boundary between Groups II and III, there are four
pairs of correctionswhich mightserve. For column lb these are:
Groups I/II - 0;20 + 0;40 + 1;40 +2;40
Groups lI/Ill +4 +3 +2 +1.
If we now considerthe combined boundary correctionsat both
nodes, we findthatthese are most nearlyuniformand symmetri-
cal if we choose + 1;40 and + 2 fromthe above pairs. Table 18
shows the resultingcorrections.Any other choice yields a com-
bined correctionof at least + 5;40 at either the firstor second
boundaryofcolumnI, which is higherthan any attestedvalue and
not symmetrical.These thereforeseem the most probable correc-
tions, and we have used them to reconstructthe missing values
of '(L) in Group 11+, although tentatively, since we stilldo not
understandhow the individual boundary jumps were derived.
The values of I(L), as reconstructed,are presentedin column

TABLE 18. dolt(L) + 34

Boundary ColumnI ColumnII
GroupsI/II +4;40 +5;0
GroupsII/III +4;40 +4;40
Groups III/IV + 2;40 + 2;40
GroupsIV/V +4;20 +4;20
GroupsV/I +4;40 +4;20
Total +21;0 +21;0
(1) of Table 19 with the longitudesfromTextL in column (2). The

velocities, representedby the twelve-monthchanges in I, are
fixedrelativeto eclipse groups as follows: d12I(L) = 12 at the as-
cending node (Cols. lb and Ilb) in Groups II, III, and IV and at
the descending node (Cols. Ia and Ila) in Groups I, and V. Con-
verselyd12I(L) = 10;40 at the descending node of Groups II, III,
and IV and at the ascending node of Groups I and V.
For a given velocity(progress in nodal elongation) the longi-
tudes in Groups I and V fallwithinthose in Groups II-IV, so the
latter groups determinethe range of the two velocities. These
ranges are as follows:
d12I(L) = 12:
Col. I: 196;30? to 3;30? (Libra16;30? to Aries1;30?)
Col. II: 207;0? to 12;0? (Libra27? to Aries12?).
d12I(L) = 10;40:
Col. 1:28 23;0? to 188;0? (Aries23? to Libra8?)
Col. II: 33;0? to 198;0? (Taurus 30 to Libra18?).
Interzone midpoints:29
Col. I: 12;[O]? and 192;[O]? (Aries,Libra12?)
Col. II: 22;30? and 202;30? (Aries,Libra22;30?).
Ignoringthe small errorin Col. la, each zone of constantveloc-
ityextendsfor165?, and the two are separated by a gap of 15? at
each end.30The midpointsof these gaps are separated by 180?,
so there can be no doubt that the scheme assumes two equal
zones of constantvelocity,each coveringhalfthe zodiac, but fixed
relativeto the Saros Cycle ratherthan to the ecliptic,and thus ad-
28 As given in the text.Correctingthe longitudes in Col. Ia as discussed above would
reduce these values by 0;30?.

29 Assuming correctedvalues forthe longitudes in Column Ia.
30 This may explain the anomalous distributionof eclipse possibilitiesinto groups of
8-7-7-8-8 EP. The standard distributionwould add 1 EP to Groups II-IV in Text L and
therebycause the velocityzones to overlap.

, _ oO e O. ts 0O to to 0 en to 0 rs n O. ts O e
en FI~ n o.
0 F en o,
0 6 t 0 0 - F o e 0o t
en
0
0
to l tf te e~ e~ e ,
0 -
%0
I i ' l| Y 1 1 x I 1i
tw li N ci i |i |
0-,
0
Z~~TI _ > ?
_sO. - O.t .- 0 O, 0 O. - 0 . - O. - 0, - 0. -O
dod 0 ~~- 0o t It 0 N
0e 0. - . 0 t- O. 0 t- O 0 t- 01 t- 0 O
E- ~~ _o ? 0
' 0
o 0
+
o I -
,
0
N
-
4
o
%0
-
t_-
0
@
- 0
cSnN
t-
-
o 0 -
O
o
4
0
> D+ - st0 - CR 0 NI % I* O 0 0% - t 0% '1 4
-o i ? 0, O - t O. 0 o O, - 0 - 0, 0 O. 1 0 O <
c; 6; o g od od
oo _I 4 0 _i > 0
O-Z-
0-4 ci~~~
I -s 9 0, I r I o 0 ~ 0 t* 0 FI 0. F 0 N 0 'I 0 Z 0
es@,~~~~t
g obM$enN
<
~~~%O O 0 Z0 ' 0 t*- 0
| +11^ n 1_ - 1t > ^1> R g 21 ~~~m

>- n m>^1
I1
01- - 1- t- >c
t-
a N
a.-. 0 0 t- - 0 4 4 @0 0

vancing by 10;30? each Saros. This effectively (iferroneously)im-

plies that the variationin velocityis associated with eitherlunar
anomaly or nodal elongation, ratherthan with longitude, since
both ofthe formerreturnverynearlyto theirstartingpointsin one
Saros.
To see the fulleffectof the variationin velocitywe look at the
change in t(L) at successive eclipse possibilities,where it is large,
ratherthan at twelve and eleven monthintervalswhere it is not.
Column (3) ofTable 19 shows thischange,while column (4) shows
it net of the mean motion in nodal elongation, representedby
d+(6). Column (4) thus presents the inequalityin nodal elonga-
tion implicitin t(L), expressed in units of T.
Table 20 rearrangesthe data fromcolumn (4) ofTable 19 in order
of increasinglongitude. Data markedby an * occur at boundaries
which have beeen shifted1 EP fromthe patterncharacteristic of
T(6), and thus are likelyto be distorted.
The remainingdata are graphedin Figure2, wheretheyare com-
pared withthe theoreticalinequalityresultingfromthe sun's zodi-
acal anomaly,31expressed in units of T. For -400 this inequality
can amountto as much as + 4.07?,equivalentto + 6.1 unitsofmag-
nitude. The theoreticalinequality is zero at Babylonian longi-
tudes32of roughly 740 and 254? and reaches its minimum and
maximum at longitudes of 164? and 344? respectively.An addi-
tional inequality of +1.5 units arises from the lunar anomaly.
While individual values fall outside this band by up to 1.5 units
(10), the inequalityin t(L) clearlyparallels the zodiacal inequal-
ity very closely. In particular,the extremevalues derived from
the textoccur at very nearly the same longitudes as those of the
theoreticalinequality,while the zero values-although more dis-
persed-almost precisely bracket their counterpartsin the the-
oretical function. Finally, it is noteworthythat the maximum
inequality in t(L), ? 7;40 units = + 5.1, agrees very closely
with the theoreticalamplitude, +5.07? = +7.6 units,obtained by
combining the maximal inequalities due to zodiacal and lunar
anomalies.
'If(L)thus describes an inequalityin nodal elongationwhich is
very nearly in phase with the zodiacal inequality for the two
Cycles coveredby our text,but whose amplitudeappears to reflect
31 As remarkedby Aaboe and Henderson [1975],194,the nodal elongationof the moon
at syzygyis mainlydeterminedby the sun's position. Consequently,itsprincipalinequality

is due to the sun's zodiacal anomaly.
32 We have assumed the SystemA normin which the vernal point occurs at Aries 10?.
Thus by "Babylonian longitudes" we mean modern longitudes less 100.

TABLE 20
INEQUALITY IN * (L)
(units of I)
Col I ----- -----Col 11-----

dT(L) d+(L)
Bab. minus Bab. minus
Long. dI'(6) Long. d+(6)
1.5 4.7 1.5 4.0
23.0 3.3 12.0 5.3
33.5 2.0 33.0 2.7
42.0 1.7 43.5 1.3
44.0 0.7 52.5 2.0
52.5 3.3 54.0 0.0
54.5 -0.7 63.0 3.0
73.5 1.7 64.5 -1.3
84.0 0.3 84.0 1.3
95.0 0.0 94.5 0.0
106.5 -1.3 105.0 -0.7
116.0 -2.7 115.5 -2.0
124.5 -5.0 126.0 -3.3
135.0 -4.3 135.0 -5.7
145.5 -5.7 145.5 -5.0
156.0 -7.0 156.0 -6.3
156.5 * -6.0 * 166.5 -7.7
167.0 -3.3 166.5 * -6.7 *
177.5 -4.7 177.0 -4.0
188.0 -6.0 187.5 -5.3
196.5 * -9.0 * 198.0 -5.3
207.0 -3.3 207.0 * -9.0 *
217.5 -2.0 217.5 -2.7
228.0 -0.7 228.0 -1.3
228.5 -4.7 238.5 0.0
249.5 -3.0 238.5 -4.3
260.0 -1.7 259.5 -2.7
268.5 * -4.3 * 270.0 -1.3
270.5 -0.3 279.0 * -3.7 *
279.0 0.0 280.5 0.0
289.5 1.3 289.5 0.7
300.0 2.7 300.0 2.0
311.0 3.0 310.5 3.3
321.5 4.3 321.0 3.7
332.0 5.7 331.5 5.0
340.5 2.0 342.0 6.3
342.5 7.0 351.0 2.7
351.0 3.3 352.5 7.7
the combined effectsof the zodiacal and lunar anomalies. The re-
lationshipofthe inequalityto lunar anomaly is also evidentin the
basic structureof the textand especially in the constancyof the
twelve-and 223-monthchanges withineclipse groups. 1(L) thus
appears to antedate the clear separation of the two anomalies,

10.0
A Coll
0 Col 11 A
5.0 - Theoretical l
I 0.0 I A 13 1
+-2.5A AA
-5.0-A A0
-7.5--
- 10.0 I I I I I I I I I I I
0 30 60 90 120 150 180 210 240 270 300 330 360
Longitude (Babylonian)
FIGURE 2. Inequalityin If(L)
while representingan improvementover the more elementary

function,'(S).
Table 21 shows (1) 1(L) as reconstructed;(2) the nodal elonga-
tion (in degrees) at syzygy,computed from P.V. Neugebauer
[1929]; and (3) the resultingerrorin (correctionto) *(L). These
errorsare evenly distributedand relativelysmall, indicatingthat
1(L) is generallyquite good.
The averageerrorforeach group is shown in Table 22. In every
instancethe errorsat opposite nodes cancel, so the errorsforthe
entiregroup are small. Only Group III has an averageerrorgreater
than + 0.5 units; it is also the group with the distinctivecorrection
to d11o(6) shown in Table 18.
The mean errorforall 76 values of '(L) is negligibleand less
than the uncertaintyof modern calculations. The probable error
of a single value is + 93 units (0;37?). This shows that I(L) is ex-
tremelywell centeredand betterthan I(S), whose mean and prob-
able errorsare + 0.65 + 2.0 units (Britton[1988], 67). Finally,the
accuracy of '(L) is also remarkablyconsistent from group to
group, which suggests that empirical adjustments-ratherthan
purely theoretical considerations-may have influenced the
choice of values at the group boundaries.
'(L) differsfromlaterfunctionsin appearing to have been spe-
cificallyfittedto the period coveredby the text.We inferthisfrom
the general accuracyof the function,which depends on its close

ia - T, ' TTT?, ?0, '? , >0
0 0 ci 'i %0 Cs
-X fi \0 \ 0e t- e 0 t 0t% 0 t- en 0 00
i
%0 ~
l' 0% O0 N t i i0 ~- j~L%I
4 I
n_ 0 e0 oo~ _ 0 _ Or- _~O~_ c_ \
I ff 0 1 9' 1 9''TT0f 1 * I i ' i - ' N 1
oo~~~~~~~~~~
tv o, co o M M M M ,e tn o, F
6, _ o -- 0 e~ - N - oo o N- ' i_ -
~~~~~~~~~~~~~~~
%0F
n en
; Ue + - e o e t e~~~~~~~~~~~~~~~
I o -f I - -o I
ci O 4 4 tz 6f 6 ci
O t4 e 4 c
o -_ NI - t -o - F An
N o I en e * en en
_~~~~~~ -- ~~~~
z~~~~~~~~~~~~~~~~~t
_ _ _ _ I -
1. e oItk ct- o0 o~ N
e o tn D t- co o~ oo 00 N en Z t-c
e e - m
>- % S 2
1C h~~~~~~~~~~~~~ I
1-- - --- INcSNS^^n2rn

TABLE 22. Average Errorsin *(L) by Group (units of magnitude)

Group Group Group Group Group
Column I II III IV V Average
Ia (+) -1.0 +1.4 +0.2 -0.2 -0.6 -0.047
lb (-) +1.2 -1.4 -1.2 +1.1 +0.6 +0.137
Average I +0.1 +0.2 -0.6 +0.5 0.0 +0.045
Ila (+) -1.5 +0.7 +0.4 -0.5 -0.5 -0.321

Ilb (-) +1.6 -1.2 -1.5 +0.9 +0.7 +0.168
Average II +0.1 -0.1 -0.7 +0.2 +0.1 -0.074
Average (+) +1.4 -1.3 -1.4 +1.0 +0.7 +0.155

Average (-) -1.3 +1.1 +0.3 -0.3 -0.6 -0.183
Average ALL + 0.06 +0.04 -0.65 +0.33 +0.06 -0.014
Groups II, III, and IV - 0.095

Groups V and I + 0.064
Positive Errors 35
Negative Errors 41
Standard Deviation + 1.38 units
Probable Error + 0.93 units
correlationwith the zodiacal inequality,and which gets worse as

one moves forwardsor backwards in time. This apparent speci-
ficityis furtheremphasized by the factthat in the entireperiod
fromthe beginningof the reign of Nabonassar to the beginning
of the Seleucid Era, therewere only two tripletsof eclipses visible
at Babylon, which exhibitthe maximal inequalityin nodal elon-
gation due to the combined effectsof both the zodiacal and lunar
anomalies.
Bothsets ofeclipses occurin Column I ofour text.The firstcom-
prises EP's 7, 8, and 9 in Group II and begins with the eclipse of
-413: Sep 8. The second comprisesEP's 26, 27,and 28 in Group V
and begins with the eclipse of -403:Feb 23. The longitudes and
(modern) magnitudes forthese eclipses are:
Date Long. Mag. Date Long. Mag.

-413:Sep 8 339.890 14.7 -403:Feb 23 150.410 3.9
-412:Mar 4 159.510 10.5 - 403:Aug 18 319.930 1.0
- 412:Aug 28 329.030 13.1 - 402:Feb 13 139.590 19.1

In both cases the inequalityin elongationis +5.06?, equivalentto

+ 76 units of magnitude.In the firstcase, however,the inequality
acts to minimizethe variationin magnitude,while in the second
case the variationin magnitudeis maximized. The second group
of eclipses is also that forwhich we findthe maximal inequality
in I(L), namely ?7;0 units in Column I and +7;40 units in Col-
umn II.
In sum, we findin I(L) a functionwhich describesthe variation
in nodal elongationwith a relativelyhigh degree of accuracybut
in a manner which does not separate the independent compo-
nents due to the lunar and zodiacal anomalies. I(L) thus appears
to antedate both SystemA and SystemB, while possessing attri-
butes which appear relatedto each. In particular,the period rela-
tion which underlies I(L) appears identicalwiththatof SystemB,
as does the implicitmagnitudefora centraleclipse, I(L) = 18. In
contrastthe two zones ofconstanttwelve-monthprogressin nodal
elongation point towards System A's treatmentof the zodiacal
anomaly.Finally,the excellenceof I(L) as reflectedin the accuracy
ofits amplitude,phase (in relationto the zodiacal inequality),and
above all the mean positionofitsimplicitnode, pointsto a serious
and painstakingeffortin its constructionand appears to exclude
the possibilitythat it was merelya pedagogical excercise.

TextF: B.M. 36400 (80-6-17,176)
Contents:Full-moon longitudes monthlyfor S.E. 46-51 (-264 to

-258).
Table 23.
Transcription:
DescriptionofText:
B.M. 36400 listslunar longitudesof fullmoons monthlyforfive
years beginningwith S.E. 46,VI,and ending with S.E. 51,VI. Ob-
verse and reversecontain two columns each. Contraryto normal
practice,the order of the columns on the reverseis fromleftto
right.Yearsare separatedby horizontalrulingsexceptin obverseI;
the only preservedyear number is a "50" on the leftedge corre-
sponding to reverse 1,13, as indicated in the transcriptionin
Table 23.
In the last line the "9" is writtenwith nine wedges (although
as threediagonals everywhereelse), and the zodiacal sign Taurus
as "gU4" (an abbreviationof the earliernotation"gu4-an")in ob-
verse1,15and reverseL,8and 11,4,in contrastto the normalSeleucid
conventionof renderingTaurus as "muil-muil" or "muil."In astro-
nomical diaries, "gu4-an"virtuallydisappears as the designation
of the zodiacal sign by the beginningof the Seleucid Era. One of
the latestoccurrencesis fromS.E. 56 in a statementof Mercury
(A. Sachs), but it may well have been an anachronismby then;
the event had to be predictedfromthe correspondingphenome-
non that the scribe would have had to find in a diary dating 46
yearsbefore,in this way introducingthe likelihoodof contamina-
tion by the older terminology.
In the absence of an explicityear numberin the text,these in-
stancesmightwell have been takenas formalindicationsofa prob-
able pre-Seleucid date. On the lower edge are what appear to be
somethinglike the numbers24 and 2 writtenin a wettish,shallow
fashion, possibly even erased.
About half of the surfaceof the obverse is destroyed.The two
groups thatremainin the firstcolumn, amountingto seven lines
in all, presentproblemswhich we shall discuss later.Fromthe first
preserved line of obverse II to the very end of the text,however,
the longitudes follow a clear and consistentpattern.
Commentary:
The longitude scheme is very primitive:six months with a
monthlylunar progressof 29?, followedby a six-monthincrease
63

TABLE 23
Obv. I Jr
E ~~~~~I
LU&61 gqq
3) apin. I iaN-vWal
9a&r 30 0
s) L8b[... III
7~) barL
L9IJ4Lq. I3 |
bar
<9 I Q-
i
ii) [~~Lc I J~ 1,5[h4vn.
I9J+[
IIfuu1 lIdJu630[w.
13) ) ant I ll5rh X (9
I {Ee) 6 l1,1SrhtttU ab 9)15SA (13

S) apit10,30911+
ztz 1)30 absin,
[ga.m q tSa 6)30 Lrntj (15
Rev.sic
') baLr *,30 9i'r-tab Izt 21 slF(I
[31Ut 14)30
pa kt. 20,15 -
3) [si3 3,30 Mal cd6 19,30 hkLULT (3
j
S) L ZZ apu.
230 gu9
, 13,5
kiii, 'j5 hw,m't 9,5 u-
7) cLw6 30 fhlt] 81 17,15ailat

apin z9,15
1) I 16,30A (7
s) Ie
3an. 2g,3omas--mAg 15,30absit
ab 27, SZlax A4irt-4e11+30 rl/7 (9
I) zLz 27 A 'Abar 13,30 t/r-tab]
3 e 26 ab's 1 91k 12,30p[a ] (
13) bar 25 LrTLhJ sig 11,304as I
9U4 2* [tr-]tab I A, 10)309[ 1 (t3
15) [stgl23 pa. iZiL
I ,r7
[
q,45S
sior]
rY n tv
~~~~15t

of 29;15? per month,withoutregardto the region of the ecliptic.

These two parametersimply an increase in lunar longitude for
twelve months of
6.290 + 6*29;150 = 1740 + 175;300 = 5,49;300or
d12X= 6,0? - 10;300.
We have encounteredthis same relationshipin TextS and TextL,
and therecan be no doubt that the presentscheme is simply an
extensionto monthlymotion of the scheme found in those texts
fordepictinguniformmotion between eclipse possibilities.In so
doing the scheme in TextF takes no account of the effectivecor-
rectionin velocityintroducedeverytimethereis an eleven-month
interval.Instead, it simplyreflectsan averagemonthlyprogressin
lunar (and solar) longitude of 29;7,300,so that the year is
296O
= 12;1,37,51, . . . synodicmonths,
29;,3001m
which compares poorlywith 12;22,7,51,. . .-the value implicitin
textsS and L. Nevertheless,it is noteworthythatthe extremesix-
monthvelocitiesunder this crude scheme, 175;30? and 174?, are
preciselythose found in Column II of TextL forthe motionfrom
ascending to descending node and conversely.
Our textbegins with monthVI of 46 S.E., which is also a lunar
eclipse possibility(Table 6). No longitude is preserved, but the
next threelines imply a progressof 290 per month,which leads
to Aries30 as the firstentry.As noted above, thereare difficulties
with all the data preservedin Column I of the obverse. If we ex-
tend the scheme preserved in obverse II and the reverseback to
obverse I, we also arriveat Aries30 forthe firstline, but findthe
progressin lines 2-4 to be 29;150 per monthratherthan 290 as in
the text.In lines 14-17,on the otherhand, the progressesare con-
sistentwith the restof the text,but the longitudes are 10;300 less
than expected,suggestingthattheymayhave been originallycom-
puted formonthsone Saros earlier.33 Thus the errorsin obverse I
arise from(at least) two different and independent sources.
All the errors(or inconsistencies)occur priorto the five-month
intervalbetween eclipse possibilitiesin S.E. 47* XII and S.E. 48,
IV, while all preserveddata afterthatintervalare consistent.Also,
the reconstructedlongitudesforthe foureclipse possibilitiesprior
to that interval (46:VI, 46:XII, 47*:VI, and 47*:XII) are all con-
33 These fourlongitudes are also identicalwith those in obverse II, 9-12.

id ~ ~ ~ ~ ~ ~ - g(~ror0~\ 'I 0r,- 0 0~~
, > '
\9 'rx-- r .-\ - >
~~~~S~~~~~~~:l
suDd48 a- Xa
~ I~v ~ ~ ~ ~ ~ oo '
wss o ktl i aM Nbfix x '):
r.'n 000 00 I _ 00 n -o
n
H K, e
N
0 t> No ,a -
L liii b ; z'i r
A~~~~~~i q
C;
0~~~~~~ qNri w^ ,qN_______q
S~~ ~~~~ ft5 ?;z lI Ix O ~ ' I13 w- Ox IfI O-4 U oF I I4~

0-te ?1 t S^>
- tZ te1 ss |t X s^> tt
La hEl& t}
l l l l~~~~~~~~~b
l l
~~~~
S, bo $~I ;<~~~~~
IS Sx Sl tI c I F <1FE t?1d
I! S)t )N tIC;;)
Ie, iC 81 S ?0 1 t
Ae l?t n?0 % ???? (
>m i 1? qfi 1q $fi sfi1 t1t% F w 4> fi_ ?t
{_
N~~~~~r N N
;~~~~~~4 SOe >oo0WOWWmutt e O r-

N sv Bs > s ,b,>% @N * *s 4 ** s*Zt
^*sG
%
G
Ms
6*
6s
^

nectiblewith Column II of TextL, whereas those afterthe interval

are not.34
In Table 24 are listed, forreasons of comparison, first,the pre-
served longitudes from the text; next, the longitudes B2 con-
formingto the consistentcorpus of SystemA texts;and last, the
longitudes of the fullmoon computed by modern methods,with
modern parameters,and obeyingthe modern conventionof mea-
suring longitudes fromthe vernal equinox.
There is excellentagreementbetween B2 and the textnear the
years' ends, and the maximal deviation of nearly 50-the text's
longitudes are almost everywherelargerthan B2-is then a mea-
sure of the quality of the primitivescheme.
Though it has nothingto do with our textwe have included in
the last column the differences between B2 and the modern longi-
tudes; they all lie in the interval5;30? + 10. This is remarkable,
for B2 takes into account only solar anomaly, while the modern
values depend on lunar anomaly as well.35
34 This followsfromthe factthatin the underlyingscheme the omittedmonthin a five-

monthintervalhas an implicitlunar velocityof 300, whereas in TextF the omittedmonth
has a velocity of 290 or 29;15? (here 290). Thus every five-monthintervalthe scheme
reflectedin TextF advances 10 or 0;45? relativeto the underlyingsheme.
35 See Aaboe and Henderson [1975], 194f.

TextG: B.M. 36580 80-6-17,590)
Obv
3,6
3 6 73 7
PLATE 7.
67 3
Rev 40
XSI.

TEXT G
Contents:Partsof Columns II and III of TextS36concerningsolar

eclipse possibilitiesfrom11 Xerxes,VIII to 8 ArtaxerxesI,
IV (-474 to -456)
Table25; Photograph:
Transcription: Plate7
ofText:
Description
TextG comprises part of the upper leftcorner(obverse) of the
tabletcalled TextC [B.M. 36737 (80-6-17,470)]in Aaboe and Sachs
[1969], to which it is now physicallyjoined. All edges are badly
rubbed, as are the edges of Text C, so the surfaces do not join
closely.However,thereis no doubt at all about the join as may be
seen in Plate 7.
Partofthe upper edge is preserved.37The obversecontainsCol-
umns II and III of Text S forthe firstfive eclipse possibilitiesof
(solar) Saros Cycle 16. Column II containsvalues of 4b which are
consistentwith the reconstructionin Aaboe and Sachs [1969], 17.
Beneath the firstfour4 values are numberswhich we stilldo not
understand.These are followedalternatelyby the terms"me" and
"zalag" which are otherwiseunattestedin TextS,38but which sug-
gest intervals,probablymeasured in time degrees.
Justenough ofColumn III is preservedin the obverseto confirm
the structureofthe functionI(S) forthe firstgroup ofeclipse pos-
sibilitiesand to show that the values agree with the traces pre-
served in TextB (Aaboe and Sachs [1969], 12).
The reverseis verypoorlypreserved. Only threevalues of Col-
umn III can be clearlyread along with the ending of one 4 value
in Column II and the term"sa zalag" under what would be Col-
umn IV, which agrees with the (partiallypreserved) heading for
the same column in TextB. Of greatestimportance,however,are
the preservedvalues of Column III, which permitthe secure re-
constructionofthe functionI(S) forthe sole group of eclipse pos-
sibilitieswhere its structurewas uncertain.
The edge of the textcontainstracesof the followingfournum-
bers writtenacross the edge fromobverse to reverse.
36 See note 24 above.

37 We are indebted to ChristopherWalkerof the BritishMuseum forhis carefulcolla-
tion of this fragment,in which he furnishedseveral of the readings presented here.
38 Elsewhere in TextsS the termssiI and gin alternatelyfollow similarnumbers.
69

TABLE 25
.
TexdtGaL?I
Xer,1 21 620 rS'
(33 Xaclae
1 ~~2, /I 37,57?4 ,40

__ ~~~)17
23 1 2 52 2 3 l
Vq2IZI Sa 7j
2z,4jsz21~,Z6q3,
g~ ~~
:z2JC q2 32
_ __ _lo_ _
2,zo3j
E[21
C9Ld-)

Column II obv (III rev): 2,1]3,20

if : ]5,40
Column III obv (IV rev): 3,22
30
The firstof these is probably2,13,20used simply as the name of
Column 4 or as a truncatedvalue of 4.39 The rest is obscure.
Commentary:
Fora detailed discussion ofTextS, includinga descriptionofthe
consolidated textwhich incorporatesthe evidence of TextG here,
see Britton[1989].
39 Cf. Aaboe [1968], 8 and ACT I, 212 forthis use of 2,13,20.

BIBLIOGRAPHY
Aaboe [1964]: A. Aaboe, "On Period Relations in Babylonian Astron-

omy."Centaurus10 (1964): 213-231.
[1968]: A. Aaboe, "Some Lunar AuxiliaryTables and Related Texts
fromthe Late Babylonian Period." Danske Vidensk.Selskab,Mat.-fys.
Medd.: 36, 12.
[1969]: A. Aaboe, "A Computed List of New Moons for319 B.C. to
316 B.C. fromBabylon: B.M. 40094." DanskeVidensk.Selskab,Mat.-fys.
Medd.: 37, 3.
[1971]: A. Aaboe, "Lunar and Solar Velocitiesand the Length of
Lunation Intervalsin BabylonianAstronomy."DanskeVidensk.Selskab,
Mat.-fys.Medd.: 38, 6.
[1972]:A. Aaboe, "Remarkson the TheoreticalTreatmentofEclipses
in Antiquity."Journal fortheHistoryofAstronomy 3 (1972): 105-118.
[1984]: "Mesopotamian Mathematics,Astrology,and Astronomy,"
to appear in Cambridge AncientHistory,Volume III 2: 276-292.
[1987]: A. Aaboe, 'A Late-BabylonianProcedureTextforMars, and
Some Remarkson RetrogradeArcs,"FromDeferent toEquant:A Volume
ofStudiesin theHistoryofSciencein theAncientand MedievalNear East
in HonorofE.S. Kennedy,ed. by D.A. King and G. Saliba (Annalsofthe
New YorkAcademyofSciences,vol. 500, New York,1987), 1-13.
Aaboe and Hamilton [1979]: A. Aaboe and N.T. Hamilton, "Contribu-
tions to the Study of Babylonian Lunar Theory." Danske Vidensk.
Selskab,Mat.-fys.Medd.: 40, 6.
Aaboe and Henderson [1975]:A. Aaboe and J.A.Henderson, "The Baby-
lonian TheoryofLunar Latitudeand Eclipses Accordingto SystemA."
ArchivesInternationalesd'Histoiredes Sciences25 (1975): 181-222.
Aaboe and Sachs [1966]: A. Aaboe and A.J.Sachs, "Some Dateless Com-
puted Lists of Longitudes of CharacteristicPlanetary Phenomena
from the Late-Babylonian Period." Journalof CuneiformStudies 20
(1966): 1-33.
[1969]: A. Aaboe and A.J.Sachs, "Two Lunar Textsof the Achaeme-
nid Period fromBabylon." Centaurus14 (1969): 1-22.
ACT: 0. Neugebauer, Astronomical CuneiformTexts,3 vols., (London,
1955), Lund Humphries. Reprinted(1983), Springer-Verlag.
Alm.: Ptolemy's Almagest,trans. and annot. by G.J.Toomer,(New York,
Berlin,Heidelberg, Tokyo,1984), Springer-Verlag.
Bernsen [1980]: L. Bernsen, "Some Investigationson the Ephemerides
of the BabylonianMoon Texts,SystemA." Centaurus24 (1980): 36-50.
Bowen and Goldstein [1989]: A.C. Bowen and B.R. Goldstein, "Meton
of Athens and Astronomyin the Late FifthCenturyB.C.," A Scientific
73

74 BIBLIOGRAPHY
Humanist:Studiesin HonorofAbrahamSachs, ed. by E. Leichty,M.deJ.

Ellis, P. Gerardi (Philadelphia, 1989), Univ. Museum, 39-81.
Britton[1987]:J.P.Britton,"The Structureand Parametersof Column 4,"
FromAncientOmensto Statistical Mechanics:Essayson theExactSciences
presentedto AsgerAaboe, ed. by J.L. Berggrenand B.R. Goldstein,
(Copenhagen, 1987), Univ. Library.
[1989]: "An Early Function for Eclipse Magnitudes in Babylonian
Astronomy."Centaurus32 (1989): 1-52.
Epping and Strassmaier [1893]: J. Epping and J.N. Strassmaier,"Der
Saros-Canon der Babylonier."ZA 8 (1893): 149-178.
Goldstine [1973]:H.H. Goldstine. "New and Full Moons from1001B.C.
to A.D. 1651." Memoirsof theAmericanPhilosophicalSociety94 (1973;
reprinted,1991).
Grayson [1975]: K. Grayson,Assyrianand BabylonianChronicles(Locust
Valley,NY, 1975), J.J.Augustin.
HAMA: 0. Neugebauer, A Historyof AncientMathematicalAstronomy
(New York-Heidelberg-Berlin, 1975), Springer-Verlag.
Huber [1973]:P.J.Huber,"BabylonianEclipse Observations:750 B.C. to 0,"
unpublished paper.
Kugler [SSB]: F.X. Kugler,Sternkunde in Babel,1 (1907); II
und Sterndienst
(1909-24), Munster,Aschendorff.
JCS: Journal ofCuneiform Studies.
Langdon, Fotheringham,and Schoch [1928]: S. Langdon, J.K. Fother-
ingham, and C. Schoch, The VenusTabletsof Ammizaduga(London,
1928), OxfordUniv. Press.
LBAT: T.G. Pinches, J.N. Strassmaier,A.J. Sachs, Late BabylonianAstro-
nomicaland RelatedTexts(Providence,1955), Brown Univ. Press.
Moesgaard [1980]: K.P. Moesgaard, "The Full Moon Serpent. A Founda-
tion Stone of AncientAstronomy."Centaurus24 (1980): 51-96.
Neugebauer [1938]: 0. Neugebauer,"Untersuchungenzur antikenAstro-
nomie III.,"QS B 4 (1938): 248-252; and (V): 407-411.
[1957]: 0. Neugebauer,TheExactSciencesinAntiquity, 2nd ed. (Provi-
dence, 1957), Brown Univ. Press, Reprinted(New York,1969), Dover.
[1989]: 0. Neugebauer, "A Babylonian Lunar Ephemeris from
Roman Egypt."A Scientific Humanist:Studiesin HonorofAbrahamSachs,
ed. by E. Leichty,M.deJ.Ellis, P. Gerardi (Philadelphia, 1989), Univ.
Museum, 301-304.
Neugebauer and Sachs [1967]: 0. Neugebauer and A.J. Sachs, "Some
AtypicalAstronomicalCuneiformTextsI." JCS 21 (1967): 183-218.
[1969]: 0. Neugebauer and A.J.Sachs, "Some AtypicalAstronom-
ical CuneiformTextsII." JCS 22 (1969): 92-113.
Oppolzer [1887/1962]:T. v. Oppolzer, Canon of Eclipses,trans. by 0.
Gingerich(New York,1962), Dover.
Pannekoek [1917]:A. Pannekoek,"The Originof the Saros." Koninkligke
Akad. van Wetenschappente Amsterdam.Proceedings 20 (1917): 943 ff.
PD3: R.A. Parkerand W.H. Dubberstein,BabylonianChronology (Provi-
dence, 1956), Brown Univ. Press.

BIBLIOGRAPHY 75
PV Neugebauer [1929]: P.V. Neugebauer, Astronomische Chronologie,2

vols., (Berlin-Leipzig,1929), De Gruyter.
[1934]: P.V. Neugebauer, "Spezieller Kanon der Mondfinsternisse
fur Vorderasienund Aegyptenvon 3450 bis 1 v. Chr." Astronomische
Abhandlungen (1934): 9-2.
QS: Quellenu. Studienz. Geschichte d. Mathematik,Astronomie, u. Physik.
Reade [1986]: J.E. Reade, "Rassam's Babylonian Collection: the Excava-
tions and the Archives."Introductionto E. Leichty,Catalogueof the
BabylonianTabletsin theBritishMuseum,VolumeVI: Tablets fromSipparI
(London, 1986), BritishMuseum Publications.
Sachs [1974]: A.J. Sachs, "Babylonian Observational Astronomy."Phil.
Trans.Roy.Soc. LondonA. 276 (1974): 43-50.
[1977]: A.J.Sachs, "AchaemenidRoyalNames in BabylonianAstro-
nomical Texts'"AmericanJournal ofAncientHistory2 (1977): 129-147.
Sachs and Hunger [1988]: A.J.Sachs and H. Hunger,Astronomical Diaries
andRelatedTexts fromBabylon(Vienna, 1988),Osterreich.Akad. d. Wiss.
Sachs and Wiseman [1954]: A.J.Sachs and D.J.Wiseman, "A Babylonian
King List of the HellenisticPeriod." Iraq 16 (1954): 202-212.
Stephenson and Clark [1978]: ER. Stephenson and D.H. Clark,Applica-
tionsofEarlyAstronomical Records,Monographs on AstronomicalSub-
jects: 4 (New York,1978), OxfordUniv. Press.
Strassmaier [1895]: J.N. Strassmaier,"Der Saros-Canon Sp II, 71." ZA
10 (1895): 64-69.
van der Waerden [1942]: B.L. van der Waerden, "Die Berechnung der
ersten and letzen Sichtbarkeitvon Mond und Planeten und die
Venustafelndes Ammisaduga." SachsischeAcad. d. Wiss. zu Leipzig,
Berichted. Math.-phys.Ke. 94 (1942): 23-56.
[1966]:B.L. van der Waerden,Die Anfdnge derAstronomie: Erwachende
Wissenschaft II, (Groningen, 1966), Noordhoff.
[1968]: B.L. van der Waerden,"The Date of Inventionof Babylonian
PlanetaryTheory."ArchiveforHistoryoftheExactSciences5 (1968): 70-78.
ZA: Zeitschrift furAssyriologie und verwandteGebiete.


American Philosophical Society

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

American Philosophical Society

Uploaded by

Copyright:

Available Formats

Saros Cycle Dates and Related Babylonian Astronomical Texts

Author(s): A. Aaboe, J. P. Britton, J. A. Henderson, O. Neugebauer and A. J. Sachs

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

VOLUME 81, Part6, 1991

Saros Cycle Dates and Related

A. Aaboe, J.P.Britton,J.A. Henderson,

THE AMERICAN PHILOSOPHICAL SOCIETY

Independence Square, Philadelphia

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

Libraryof Congress Catalog

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

II. Saros Cycle Texts

Bibliography .................. ................. 73

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

T Nhe textspresentedhere are mostprobablyfromBabylon,

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

although there is some overlap. The Saros Cycle texts include

* Abraham Sachs died 22 April 1983. Otto Neugebauer died 19 February1990,afterthe

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

Contents:Regnal years and months of lunar eclipse possibilities

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

alignmentofthe monthname which distinguishesitfromthe sev-

4 During the reignof ArtaxerxesI intercalations

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

Contents:Regnal years and months of lunar eclipse possibilities

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

N >'., ~~~~a PLATE

5 TextC Not 34579 as in Neugebauer [1938], 248, 342].

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

Contents:Regnal years and months of lunar eclipse possibilities

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

in the sense thatall dates in a given line are separated by a mul-

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

There is a simple relationshipbetween lunar and solar eclipse

14 For convenience we use the abbreviation"EP" for"eclipse possibility"when refer-

ringto some numberofthemor to a specificone, and use the written-out expressionwhen

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

that TextsA, B, and C concern lunar, ratherthan solar, eclipse

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

[ f~~~ -tInIn ~ -a: ~~~~~~~~~~

A a-- Ir 0n a. a- r gI .Ia.2f- ' I . n I

aa o.a.aaa- I- c88 a In en 0 a-a-

0 C- aaa- .o-aaaa-C.Oaaa.a00.aO- -aaa-0.r n

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

The termis found in the following

The (unpublished) textB.M. 37276containsbriefobservational

17 While theoretically possible, eclipses separated by fivemonths are seldom, if ever,

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

> C1O tA4en N r-

EH |: > ,,5x5 xzxzx2x2

-~~~~0 0~~ 0 01 W) t0- - -0C,

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

ical eclipse record forfifteenSaros Cycles beginningwith SC 13

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

This content downloaded from 128.197.27.9 on Fri, 19 Jul 2013 15:22:07 PM

Contents:Regnal years and months of solar eclipse possibilities

-~~0 0 0 01 W) t0- - -0C,