amyntoros wrote:
I know now why I thought that the whole of the Greek world considered nightfall to be the beginning of the day. It's in Robert Garland's article,
Countdown to the Beginning of Timekeeping, History Today April 1999:
It was the Romans who chose midnight as the transition between one day and the next, though they had no way of calculating when that moment occurred. In the Greek World, by contrast, the calendar day began at moonrise.
Moonrise???? 
Surely he means
sunset. Moonrise can happen at any time of the day or night depending on the phase of the moon. I think he is confusing days and months. To the extent that Greek months were lunar, the evidence is that they were based on the first moonrise of the new crescent moon. The Babylonians did it the same way, while the Egyptians based their lunar (temple) month on the invisibility of the old crescent moon at sunrise.
Greek calendrics is a nightmare. Each Greek city or region had its own sequence of months, and its own convention for marking the start of the year. Years were counted eponymously (e.g. by the name of the Athenian archon), and since we don't have a list of eponyms it is often impossible to assign a fully dated inscription to a particular year, or at least a matter of endless controversy. While months were usually lunar in principle, you could, at least in some calendars, insert intercalary days and remove days whenever there was a need, as Alexander did at Tyre. Athenian months seem to have been particularly elastic. We have an inscription dated to the 8th intercalary repetition of a certain day in Elaphobolion. Also, it was necessary to insert intercalary months from time to time in order to keep the year aligned to the sun, but again the Athenian evidence is that intercalation could be done after any month. A good recent account that is not highly technnical is Robert Hannah, "Greek and Roman Calendars: Construction of Time in the Ancient World" (Duckworth, 2005).
There are a few explicit ancient statements about the start of the day, e.g. Pliny, NH 2.79 "The days have been computed by different people in different ways. The Babylonians reckoned from one sunrise to the next; the Athenians from one sunset to the next; the Umbrians from noon to noon; the multitude, universally, from light to darkness; the Roman priests and those who presided over the civil day, also the Egyptians, and Hipparchus, from midnight to midnight." Otherwise, the ancient evidence mostly comes from astronomical/astrological sources, like the Almagest. The trouble with these is that they are very specialised texts, and they require careful interpretation. E.g. Pliny's statement that the "Egyptians" based their day on midnight is only true if you consider the "Egyptians" to be Alexandrian astronomers, but Egyptian evidence otherwise indicates a day starting at dawn or sunrise. Despite Pliny (who ought to have known) Egyptian horoscopes with Roman dates require the Roman day to begin at dusk. Likewise, the Babylonian evidence from the astronomical diaries clearly shows a sunset-based day, but Pliny says they began their day at sunrise.
This problem of the start of the day in ancient times was studied by Bilfinger in the 19th century. The Ephemerides are one of a very small set of non-astronomical documents that give a sequence of day and night events organised by date, and they show that day came before night on any given date. For Bilfinger this was conclusive evidence that the Macedonian day began at dawn. On the other hand, the "Chaldean" observations in the Almagest, which have Macedonian dates, have months which begin on the same (Julian) day as first crescent visbility, which seems to prove that the Macedonian day in Babylon began at sunset (but, this could be adaptation to the Babylonian day, there are only three samples, and they are technical texts).
The Egyptian evidence comes from double-dated papyri, and is quite noisy, but does seem to favour a month based on first visibility. Although the data is clearly not good enough to prove this one way or the other, Samuel leapt to the conclusion that the Macedonian day in Egypt must also therefore begin in the evening (instead of, say, the following dawn). However, since this is not native Egyptian practice, it could only be explained as perpetuation of original Macedonian practice. Hence his need to explain away the Ephemerides as non-Macedonian practice.
amyntoros wrote:I'd happily accept that the Macedonians might have had a different system but it would have been an exception in the Greek world, therefore I'm most curious to know how this can be interpreted from the sources. Let's say a Greek or Macedonian was writing about the 26th of Dios, for example, and the Macedonian day DID start at moonrise. If he wrote "during the night of the 26th" or "on the night of the 26th" wouldn't that mean anytime from evening of the 25th through to dawn on the 26th by our calculations - because when night fell on the 26th that would have been the beginning of the 27th?
Yes, but it only causes confusion if you are trying to map it to a day that starts at a different time, something you only do if you were trying to be extremely precise about it, which most ancient authors were not. However, this is a big deal for modern scholars. Students of the Macedonian calendar in Egypt frequently resort to the supposed phase mismatch between Egyptian and Macedonian days as a way of correlating double-dated documents to their preferred calendrical reconstruction. My own opinion is that they are mostly trying to make the evidence more precise that it warrants.
Again the Ephemerides are an example of this. They date Alexander's death to 28 Daisios, but Aristoboulos dates it to 30 Daisios. Now, the last day of the month was usually called the 30th regardless of whether it had 29 days or 30. Knowing that the Ephemerides used dawn-based days, you can reconcile these dates within the same definition of the calendar month by supposing that Aristoboulos used dusk-based days, so that his month started, in effect, 12 hours before that of the Ephemerides, but both were based on the same observation of the first crescent moon. That is in fact how the two dates used to be reconciled, until recently. But this explanation only works if Alexander died at night.
amyntoros wrote:So, how would a writer in Roman times have interpreted such dates, if he was, let's say, reading from the Ephemerides or an older Greek work that quoted the Ephemerides? Plutarch lived in Greece under Roman rule, so wouldn't the calendar and dating system he was accustomed to have been a Roman one? Even if not - although I tend to think it must have been so - Aelian wrote in Rome so he surely considered the day to begin at midnight. So what did they think when they read records taken from the Ephemerides? Aelian says that
"On the fifth of the month of Dius he was drinking with Eumaeus, they say; then on the sixth he slept because of the amount he had drunk." If we were interpreting a Greek calendar system which said Alexander was drinking on the fifth of the month, then by
our reckoning (because he most likely would have begun drinking in the evening) it would have been the evening of the fourth. And the "next day", which he slept through, was actually the day of the fifth? Aaack! Am I making any sense or am I just making everything more confusing?
No, its confusing all right. Plutarch was a Greek, from Boeotia, and living in Delphi. He seems to have largely treated Greek and (pre-Julian) Roman dates as though they were strictly lunar,whether they were or not. We have an explicit statement from him (in Roman Questions 24) that he regarded the Kalends, Nones and Ides of pre-Julian Roman dates as though they actually represented the new moon, first quarter and full moon; he regarded it as approximation that was "close enough". Whenever he makes a statement about the lunar phase of a pre-Julian Roman date, it matches this principle.
The same thing happens with Greek dates. Take the example of Gaugamela, which Plutarch dates to 26 Boedromion. In the late fourth century the Athenian calendar was particularly elastic. There is no way that an Athenian at Gaugamela could possibly have known the actual date in Athens, since it depended on local and arbitrary decisions made by the archon, possibly day-to-day and certainly not much more than a few days in advance. Even a few weeks or months later, by the time the news reached Athens, it would very likely have taken a complicated retrocalculation to figure it out. However, in the second and first centuries BC the Athenian calendar became more closely regulated by the moon -- we have dates which are "according to the archon" (variable) or "according to the moon (or the god)" (aligned). What Plutarch (or his source) could have known is that Gaugamela happened 11 days after a lunar eclipse, and, quite possibly, that there was a lunar eclipse in Boedromion of that year. For Plutarch lunar eclipse = full moon = 15th day of the month, hence the date of Gaugamela was 15+11 = 26 Boedromion.
Chris
