A COMMON SOURCE FOR THE LATE BABYLONIAN CHRONICLES DEALING WITH THE EIGHTH AND SEVENTH CENTURIES.

• Author: GERBER, MANUEL
• Position: Critical Essay

The nature of the sources of the Babylonian chronicles remains a matter of controversy even after decades of occasional discussion. Here statistical evidence is presented that the sources of the chronicles dealing with political events up to the time recorded in Chron. 2 were astrological texts. The number of coincidences between exact dates in the examined chronicles and an adannu of 30 days after ominous planetary events is compared to the number that can be expected in a random sample of dates between 750-600. The null-hypothesis that the number of coincidences found in the chronicle sample is random could be rejected at a safe p [less than] 0.02. The long-sought "common source" of the late Babylonian chronicles dealing with the eighth and seventh centuries was therefore in all probability a corpus of astrological texts. Chronicles dealing with later times were compiled from texts of a different nature. This shift from one kind of source texts to another makes a diachronically differentiated approach in t he evaluation of the information contained in individual chronicles imperative. The design of the proposed test necessitates a sound basis for finding the possible Julian equivalents of Babylonian dates. A reanalysis of New Year's dates inferred from eclipse data in LBAT 1413-1417 and of New Year's dates between 626-539 is therefore presented in the first part of this paper. An average beginning of the year two weeks before equinox is demonstrated to be probable only until 730. From that time to the end of Nabopolassar's reign a constant ratio of [sim]7 intercalations per 19 years was maintained, which kept the average Babylonian New Year's date a few days before equinox. Only at the turn to the sixth century can a second shift to even later dates be demonstrated.

INTRODUCTION

THE BABYLONIAN CHRONICLES DEALING WITH the Neo-Babylonian period are late compilations, and it is commonly assumed that the sources used by the compilers were roughly contemporary with the events they describe. The nature of these sources, however, remains a matter of controversy. [1] Grayson (1975) hypothesized that the chronicles were compilations from a running account of Mesopotamian history, partly preserved in the astronomical diaries (Sachs and Hunger 1988). This view is still advocated (Hauser 1995), but is difficult to evaluate (Brinkman 1990). A connection with astrology has also long been suspected (most recently, Swerdlow 1998), and an attempt has been made to prove a partial dependence of the chronicles on Assyrian chronography (Weissert 1992). The view that the texts as a whole are historiographic in nature (van Seters 1983), implicit in many reconstructions of Neo-Babylonian history, was recently contested in the case of the chronicles concerned with the time before the fall of Assyria (Gerber 1998).

Despite this multitude of opinions, none of the proposed explanations has ever been rigorously put to the test. The present study attempts to fill this gap by testing the hypothesis that astrological compendia provided a common source for the chronicles dealing with the political history of the eighth and seventh centuries. The design of this test requires a sound basis for finding the possible Julian equivalents of Babylonian dates before Nabopolassar. Therefore, in the first section the evidence provided by known Babylonian New Year's dates is reassessed and an outline of intercalation practice and the resultant setting of the beginning of the Babylonian year during the eighth and seventh centuries is presented. Based on these results, the question of the chronicle sources is addressed in the second section.

1. INTERCALATION PRACTICE BEFORE NABOPOLASSAR

   Before Nabopolassar the Julian equivalents of a few New Year's Days can be inferred with varying accuracy from descriptions of dated astronomical events. Beyond that, the picture is vague (Hunger 1976-80: 298): According to the commonly held view, the aimed-for beginning of the Babylonian year in the eighth century fell about two weeks before vernal equinox (Huber 1982: 8-10; al-Rawi and George 1992: 61). There is no consensus about the situation in the seventh century. Kugler (1924) proposes an average date a few days before equinox, whereas Huber (1982) postulates a single shift of the average Babylonian New Year to about ten days after equinox during the reign of Nabopolassar. Assyrian evidence discussed by Parpola (1983: 381-83) points to an intermediary stage with dates around equinox already between 684-648, but Assyrian and Babylonian intercalation practice need not always be identical.

   Figure 1A shows a scatterplot of Babylonian New Year's dates from the eighth century to the Persian conquest. The range of the attested dates is roughly two months. However, at no period is the full range made use of, and there appears to be a general trend towards higher dates (indicated by a linear fit [solid line]). The main aim of the first part of this paper is to determine the range of possible New Year's dates at any given point in time between ca. 750 and the reign of Nabopolassar.

   MATERIAL AND METHODS [2]

   A sample of 101 Babylonian New Year's dates between 748-539 was taken into consideration. The Julian equivalents of the 87 New Year's dates of the Neo-Babylonian empire (626-539) are known reliably with an error margin of one day (Parker and Dubberstein 1956; with a correction in Neugebauer and Sachs 1967: 189; Kennedy 1986: 222). As this error is not cumulative, it was considered negligible throughout this study. The time before Nabopolassar was represented by 14 New Year's dates inferred from lunar eclipse data in LBAT 1413-1417 (Sachs and Schaumberger 1955, Huber 1982). All dates are given as years B.C.

   New Year's Dates of the Empire Period 626-539. In a first step the complete set of New Year's dates of the Neo-Babylonian empire was examined. The data were checked for distributional irregularities by calculating the running mode [3] with a large window of 22 years. Discontinuities in the resulting values were assumed to indicate the presence of subpopulations, i.e., groups of New Year's dates separated by an uncompensated-for shift in the average beginning of the year. Furthermore, the distribution of dates was compared between individual reigns. Both procedures served only to determine the presence and approximate temporal position of such shifts. To determine appropriate boundaries between groups the data were examined for series of unusually numerous intercalations in the vicinity of the presumed breaks. Group boundaries were defined as coinciding with the beginnings of these series. An inverse squared distance smoothed curve [4] (8-year window) fitting the data was used to assess visually the effects o f similar intercalation patterns not coinciding with any of the presumed breaks.

   New Year's Dates of the Pre-Empire Period 748-626. The available sample of known New Year's dates before Nabopolassar is more problematic. It represents only 11% of the 122 years in question, and applying purely statistical methods one would risk obtaining biased results. Therefore, in order to augment the amount of information, a simple method was devised to calculate the number of inrercalations between known dates: An intercalation increases the New Year syzygy of the following year by [sim]18[degrees], while the syzygy following a year without intercalation is decreased by [sim]10[degrees]. Between 626-539, this corresponded to an increase of 17, 18, or 19 days per intercalation and a decrease of 10, 11, or 12 days per regular lunar year, respectively. For a given time-span, [delta]t and a given difference [delta]s between the corresponding New Year syzygies the number of intercalations can therefore be determined as follows:

   (1) i + j = \[delta]t\

   (2) ai - bj = \[delta]s\

   where i is the number of years with an intercalation, j the number of years without; a and b are field variates: a is the average increase [+ or -] SD (Standard Deviation) per year with intercalation, b the average decrease [+ or -] SD per year without intercalation. Joined and solved for i, (1) and (2) yield:

   (3) I = \[delta]s\ + b\[delta]t]\/a + b ; i [euro] N

   The extrapolated ratio of the number of intercalations per 19 years was used as an indicator of constancy. Where this ratio differed considerably from the ideal ratio of 7 intercalations per 19 years, it was compared to a LOWESS-smoothed [5] curve (t = 0.15) fitting the corresponding ratios during all periods equal to [delta]t in the empire-period sample. Smoothed values were chosen for comparison in order to compensate for irrelevant short-term fluctuations due to the presumed absence of a regular intercalation cycle. Where the deviations from 7/19 in the pre-empire sample were comparable to the values in the vicinity of group boundaries of the empire period, they were assumed to indicate a shift in the aimed-for beginning of the year and, thus, a group boundary.

   Determination of Relationships between Groups. In order to decide whether the two adjacent groups at the boundary between the samples should be merged into one, the procedure described above was extended to include dates before and after 626. Depending on the distribution of dates within each of the remaining groups, an appropriate test was chosen to determine if the groups represent statistically distinct populations of New Year's dates, i.e., whether or not the distributional differences between groups are random.

   Determination of the Range of Possible New Year's Dates at Any Point between 748-539. Assuming the range (excluding outlier dates) of New Year's dates in the largest group after 626 to be representative, ranges of the same width were then projected onto the earlier groups. Similarities of group characteristics, absence of contradicting evidence (i.e., dates beyond the projected range) and constancy of the aimed-for beginning of the year within groups were considered...