The new moon interval NA and the beginning of the Babylonian month (Q6562917)

The Babylonian calendar uses lunar months. The month began with sunset of the day on which the crescent moon was seen for the first time after the conjunction of Moon and Sun when it had been invisible. In principle, this event could simply be determined by observation. In the case of bad weather when the Moon could not be observed, the beginning of the month could have been delayed by several days. However, during the second half of the first millennium BC, the month length is always 29 or 30 days, as attested by the \textit{Astronomical diaries}. The author discusses the ways in which the month length is determined in these texts; obviously some calculation must have been involved.\N\NBabylonian methods for predicting the length of the month exist. They use the time interval from sunset to moonset on the evening of first visibility of the Moon, called NA in the texts.\N\NThe author has compiled a data base of all reports of month lengths plus NA intervals found in astronomical texts. These texts can be divided into observational and predictive (a more precise characteristic of the groups is given in the article). Both types contain indications of month length and of the interval NA. In the predictive texts, which were written before the events they describe, both month length and NA must have been calculated in advance. But in observational texts too, part of these data is marked as predicted. Including damaged reports, close to 2000 reports in all are considered.\N\NAfter a 29-day month, NA tends to be smaller than after a 30-day month. This can be explained by the fact that in the evening before the new moon the sky is for some time after sunset too bright for the Moon to be visible. Nevertheless, a NA interval can occur, but only on the following evening will it become apparent that NA is bigger than expected, and that there must have been a NA already on the preceding evening. So, in general, NA is smaller after 29-day months; but there is much overlap between the values of NA after a 29-day month and after a 30-day month. Therefore, NA is not sufficient to decide about the length of the month.\N\NRules for calculating NA are explicitly described in the texts. One such rule was first understood and formulated by \textit{L. Brack-Bernsen} [in: Under one sky. Astronomy and mathematics in the ancient Near East. Münster: Ugarit-Verlag. 5--19 (2002; Zbl 1076.01004)]. It uses lunar data from 18 and 18 1/2 years earlier. The author shows by his statistics that this rule produces the values of NA found in the predictive astronomical texts. In order to decide whether a month has 29 or 30 days, another rule detected by Brack-Bernsen can be applied. The author proposes a formulation slightly different from Brack-Bernsen's: ``If NA is smaller than NA of the previous month, then the previous month has 29 days; if NA is larger than NA of the previous month, then the previous month has 30 days''.\N\NHe then compares the data found in the observational texts with those of the predictive texts and can show that practically all month lengths and NA intervals in the astronomical texts were the result of calculation. Only borderline cases were decided by observation. Although the procedures used are attested only after the beginning of the Seleucid period, there are indications that they were already available from the 6th century BC onwards.