Information about dates from early in the Nabonassar Era has come down to us mostly from Greek sources. No tablet inscribed with the name Nabu-nasir or his immediate successors has been absolutely dated. For direct evidence of the period historians turned to Babylonian records of lunar eclipses.
The moon shifts rapidly across a background field of stars, about ½ degree an hour. Hence errors in calculation of the moon's position come to light expeditiously, and they are readily noticeable to astronomers observing the sky and to historians checking ancient records.
The moon's advantageous features are counterbalanced by the complexity of its orbit. Astronomers since Babylonian times have struggled to discover and describe the cyclical patterns traced by the moon. Thousands of observations of the moon in cuneiform texts attest to the close attention paid to lunar comings and goings.
Eclipses are a boon to astronomers. The lineup of sun, moon and earth provides a reference that astronomers can use to test and refine their calculations.
Lunar eclipses offer additional benefits to historians. The eclipses are visible over the whole night-side of the globe and ancient peoples diligently logged them. A record of an eclipse can be used to peg a relative chronology to an absolute date.
Avid sky-watchers like the Babylonians had no difficulty working out the rudimentary pattern of lunar eclipses. They occur every 5th or 6th month, discounting about 1/3 of eclipses that occur entirely in daylight hours.
A more elusive pattern operates on a lengthy cycle of 223 months. This period of about 18 years is known as a Saros. (The term is loosely applied to a cycle of eclipses separated by 18 years, perhaps better labeled a Saros cycle.) The eclipses of a Saros cycle share a striking feature - they look the same. Hence the timing and prognostic characteristics of an impending eclipse can be determined by checking earlier eclipses of its Saros cycle.
A tablet found in the 19th century provided the first hint that Babylonian astronomers were familiar with the Saros. Now known as the Saros Tablet, it lists the years 7 Nabonidus, 8 Cyrus, and so on into Seleucid times. Next to each entry is written the number 18, and the entries are in fact 18 years apart.
Initial attempts to fit a Saros cycle to the list of years were inconclusive, but an intriguing discovery emerged when the list was extrapolated back in time. Backtracking 11 intervals of 18 years from the initial entry (-548) brings the date to Nabonassar 1 (-548-11x18=-746).
A Saros period is 18 years plus about 10 days. Hence the year of an eclipse in a Saros cycle can generally be obtained by adding 18 to the year of the preceding eclipse. On the rare occasion that an eclipse occurs within 10 days of the end of a year, 19 must be added to arrive at the year of the next eclipse.
Now to the denouement: The last entry of the Saros Tablet is a year that draws to a close with an eclipse. Historians propose that this eclipse marks the finale of an 18-year sequence, and the next entry would require a 19-year jump. In effect, after mulling over the problem for a century, historians settled on a particular Saros cycle and declaredThe Mystery of the Saros Tablet solved.
In addition to tracking a Saros cycle from one eclipse to the next, the Babylonians worked out the internal structure of the period. The 18 years interval, or more precisely 223 months, allows for exactly 38 eclipses possibilities, 33 spaced 6 months apart and 5 separated by 5 months.
The Babylonians organized this scheme in tables of Saros cycles that begin as early as 747 BC. They continued mapping the Saros scheme until the 1st century AD.
The primary texts buttressing the Grand Saros (No. 2, 3 & 4 in Lunar and Planetary Texts) list 52 entries that match modern computations and 4 that may be wrongly filed.
Checking all the entries in the eclipse texts, no further mismatches crop up in the other 93 records from the period of the Grand Saros (-746 to -314).
A Saros cycle generally peters out after a few centuries and new cycles come into being. Moreover, not all eclipses of a Saros cycle are visible. The changing pattern of eclipses led Babylonian astronomers to organize their observation in a multiplicity of Saros cycle configurations.
Still, the Grand Saros Canon is noteworthy. Except for two low magnitude lunar eclipses that Babylonian astronomers may have missed, the Grand Canon correctly predicts every eclipse possibility over 400 years, from -746 to -314.
A document akin to the Grand Saros may have been source of Ptolemy's eclipse data from Babylon.
The impressive scope of the Grand Saros Canon and of the Saros scheme in general does not insure the validity of individual entries. The four earliest dates are computed from the text of tablet shard LBAT 1413. The preserved text is insufficient to determine a unique set of dates for these eclipses. At least one historian hesitated to ratify Feb 6, 747 BC as the earliest date recorded in a Saros.
The oldest preserved text with the characteristic 8-7-8-7-8 scheme of the Saros may come from the 6th century. The tablet (LBAT *1420) registers eclipse possibilities from -603 to a final entry in -575, and it likely was inscribed soon thereafter.
If 747 BC marks the beginning of regular eclipse recordings, then apparently within two centuries of astronomical observations the Babylonians had identified and systematized the Saros.