- Untersaal (Raetia), Monday 23 May,luna5 (231) (CILiii. 11943).
- Vasio (Gallia Narbonensis), Monday 19 October 470,luna 1751 (CIL
xii. 1497, Diehl 1927). - Tiaret (Mauretania Caesariensis), Tuesday (12)^52 August 480,luna 21
(CILviii. 21550, Diehl 4385).
The use oflunadatings in inscriptions from outside Italy may reflect, in some
cases, local calendrical practices: for example, no. 19 may be related to the
Gallic lunar calendar that has been discussed above, and no. 15 is most
probably related to the Jewish lunar calendar (more on this below). However,
it would be wrong to generalize from these cases thatlunadates were always
reflections of local provincial calendars. At least two of the inscriptions from
the Danube provinces (nos. 16 and 18) were set up by army veterans; in these
cases, it was through the medium of the Roman army, rather than because of
local calendrical practices, that the practice oflunadating spread. There is no
doubt that thelunaformula was a specifically Latin-language tradition, origi-
nally from Italy (as the early inscriptions from Ferentium and Pompeii testify)
and even in the late Empire still largely confined to Italy (or more specifically,
Rome). Its appearance in other provinces of the Roman Empire was only the
result of the general diffusion of Latin culture in theWest.
The ability to record any day of the lunar month, evident in these inscrip-
tions, suggests that lunar days were regularly and continuously counted.
Material evidence of lunar counting in Roman society can be recovered
from numerous fragments of Latinparapegmatathat tracked, with numbered
peg holes, the days of the lunar month.^53 For example, theparapegmafrom
Trajan’s baths in Rome includes a row of the seven planets (depicted as
personifiedfigures), a zodiac wheel, and a sequence of numbers from I to
XXX, all with peg holes. Its purpose was clearly to indicate, by moving a peg
along the holes, the current day of the planetary week, the current zodiac sign,
and the current day of the lunar month—all which the Julian calendar or
traditionalfastidid not represent. The lunar character of the third sequence is
evident from the fact that it has only 30 days (whereas some Julian months
(^51) This lunar date is erroneous; see discussion below, n. 82.
(^52) The inscription readsidusAugustae, i.e. 13 August, which is inconsistent with the other
calendrical data. Diehl suggests as a correction eitherWednesday orluna20; the former is a
possibility (the latter is not), but my suggested correction (pridie idusAugustae, i.e. 12 August)
fits better the lunar date, as it brings it in line with the conjunction (on which see discussion
below).
(^53) As Lehoux (2007, esp. 14, 28, 142–3) has shown, Greek and Latinparapegmatadiffer in this
respect, as the former are, in content and function, mainly astro-meteorological (see Ch. 1. 3),
whereas in the Latin traditionparapegmataare more calendrical and designed to track the days
of different (and incompatible) cycles such as lunar, hebdomadal, and nundinal (see also
Degrassi 1963: 299–313, Salzman 1990: 8–10).
316 Calendars in Antiquity