calendar scheme or as a cycle.^20 Even if it does mean a 30-year cycle, there is
no reason to assume that Pliny is referring to the same calendar as that of
Coligny. The theory of the 30-year cycle is thus not grounded on any satisfac-
tory evidence; although it provides a good synchronism between the Coligny
five-year calendar, the solar year, and the lunar month, it remains no more
than a conjecture.
More recently, Olmsted (1992) has argued that the Colignyfive-year calen-
dar was part of a longer 25-year cycle. This is based on his analysis of the
three-letter signs (TII, ITI, IIT) that appear recurrently in the calendar, and
which he interprets, together with other recurring signs (e.g. N), as indicators
of solar phenomena such as the solstices. This interpretation is only fully
compatible with a 25-year cycle in which the month of Equos counted 28 days
in year 2, but 29 days in year 4 of thefive-year cycles.^21 These month lengths
would have led to a six-day discrepancy from the solar year afterfive years,
which would have been rectified by the omission of one 30-day intercalary
month every 25 years. The three-letter signs, in this interpretation, would
indicate the solstices and other solar dates in three successive 25-year cycles.^22
Olmsted (1992) 131 argues that originally a 30-year cycle was followed (and is
still represented in the Coligny calendar by the N solar notation), but was later
replaced by a 25-year cycle to enable, through the system of the three-letter
signs, the long-term prediction of solstices with much greater accuracy.^23
But although the 25-year cycle is well synchronized with the solar year (as
well, in fact, as the 30-year cycle), it is hopelessly out of step with the lunar
month. Indeed, the additional day in Equos year 4 (with a 29-day month)
causes a one-day discrepancy from the moon everyfive years, which for a
lunar calendar is very significant: it means that the calendar could not have
retained its synchronism with the moon, and hence could not have remained
lunar, for much longer than ten orfifteen years. This failure to conform to the
moon runs counter to the general character of the Coligny calendar, of which
the structure of 29- and 30-day months is very clearly designed to be lunar.
Olmsted recognizes this problem but argues that the addition of one day to the
(^20) Pliny’s word for‘era’,saeculum, could mean something like‘generation’: see Duval and
Pinault (1986) 401, Monard (1999) 98–9, and onsaeculumin general,Weinstock (1971) 191–7,
Feeney (2007) 145 21 – 8.
As explained above (near n. 15), there are indications that Equos counted 28 days in year 2,
but not necessarily in year 4. 22
In each successive cycle the solar events occur one day later, to account for a small, one-day
discrepancy between the 25-year cycle and the sun (cf above, n. 17).
(^23) Olmsted’s model is endorsed by Pinault (1996), who, however, criticizes some philological
aspects of his theory. Olmsted (1992) 89–90, 100 identifies the starting-point of the calendar as
the winter solstice, unlike McCluskey (1998) 59–76, who, apparently unaware of his work, argues
in favour of half-way points between the solstices and equinoxes, on the basis of Celtic survivals
in early medieval Christian calendars (which Olmsted in turn explains differently, in a manner
consistent with his theory: 1992: 20, 131–3).
306 Calendars in Antiquity