Millar 1993: 407, 414–18), it is in this period that the change of calendar must
surely have taken place.^161
It is significant that in creating its new, 365-day calendar, the province of
Arabia chose not only to differ from the Julian sequence of months, but also to
refrain from exact synchronicity with the Alexandrian calendar. It is difficult
to argue, therefore, that the Alexandrian sequence of months was chosen
because of Arabia’s proximity to Egypt: if so, a conterminous calendar—
such as in Ascalon and Gaza—should surely have been preferred. In this
case, local particularism and a sense of autonomy seem to have been more
decisive factors.
The New Year date of 22 March was presumably not selected arbitrarily. It
may have been regarded as the date of the spring equinox, even though there is
no contemporary evidence to confirm this.^162 Alternatively, this date may
have been determined on the basis of the lunar dates of the old calendar. In
106 CE, when the calendar of the province of Arabia may have been instituted,
22 March occurred approximately at (or a little before)^163 the new moon of
what would have been of the month of Nisan/Xandikos. A closer match to a
lunar date can be achieved if we assume that the new calendar was instituted
later in that year (when, indeed, the province itself is more likely to have been
created). For example, the new moon was visible in the evening preceding 20
July 106CE, which corresponds exactly to 1 Loos in the new calendar of Arabia.
The new calendar could well have been instituted then, when the beginning of
the month of Loos of the old lunar calendar happened to be on 20 July. The
rest of the new calendar would have followed from this date, with a sequence
of months similar to the Alexandrian calendar and with a New Year on the
following 1 Xandikos, 22 March. The calendar of Arabia would thus have been
instituted almost immediately when the Roman province was created. This
would confirm that by 106CEthe conversion of lunar calendars intofixed,
Julian-type calendars had become accepted as an inherent aspect of Roman
provincial administration in the Near East.
The next stage in the spread of the Julian calendar, at least as far as the
evidence attests it, appears in Dura-Europos at the end of the second century,
when the city (with its surrounding region along the upper Euphrates) was
(^161) Grumel (1958) 173; see further Stern (2001) 39.
(^162) This explanation is assumed by Meimaris (1992) 37. The late-4th-c.Apostolic Constitu-
tions(5. 17. 1, 3) give the date of the equinox as 22 Dystros (i.e. March, in the Antiochene
calendar), a date which may go back to Anatolius’Easter cycle of the late 3rd c. (Grumel 1958:
31 – 2; Strobel 1984: 152; and see now Mc Carthy and Breen 2003), but which is not attested
earlier. The contemporary Roman tradition was to date all equinoxes and solstices on the 8th day
before the kalends (thus for the vernal equinox, 25 March): Pliny,NaturalHistory18. 221, 246,
256, 264, (311); Columella,On Agriculture, 9. 14. 11–12.
(^163) Conjunction occurred in 106CEon 23 March, which means that 22 March would have
been too early for a Babylonian-type calendar based onfirst appearance of the new moon.
292 Calendars in Antiquity