subsequent month of Elembiv implies arguably that Equos only had 28 days.^15
A month length of 28 days, unusual (though not impossible) in the context of
a lunar calendar, would suggest perhaps that the Coligny calendar was not
designed fromfirst principles, but rather on the basis of certain traditions that
had become established earlier through trial and error.^16
This model provides a reasonable solution to the calendar’s synchronism
with the lunar phases; but in relation to the solar year and seasons, thefive-
year cycle remains excessive, according to this model, byfive days. As a
remedy, the same scholars have conjectured further, more speculatively, that
thefive-year calendar was actually part of a longer, 30-year cycle. At the
beginning of this longer cycle (consisting of six smallerfive-year cycles), one
intercalary month (i.e. 30 days) would have been omitted, thus rectifying the
discrepancy rather neatly.^17
Evidence in support of this longer cycle—which is nowhere referred to in
the Coligny fragments—is weak and has been grossly overstated. In a brief
comment, Pliny (first centuryCE) states that the druids reckon the sixth day of
the moon as the beginning of months, of years, and of‘30-year eras’.^18 Pliny’s
notion that the month began on the sixth day of moon, i.e. at the moon’sfirst
quarter, is not intrinsically implausible (since thefirst quarter is easy to
observe) and not incompatible with the Coligny calendar, as the latter does
not specify the relationship between its months and the lunar phases.^19 But the
interpretation of his‘30-year eras’as calendar cycles is far from evident.
This phrase clearly has a chronological significance—it seems to represent a
way of counting years—but there is no reason to interpret it as the length of a
(^15) These festival days are named IVOS, and run from the end of the month to the beginning of
the next. At the beginning of Elembiv in year 2 there arefive IVOS days, whereas other months
begin with only three or four (see Olmsted 1992, tables 10–12). The unusually long run at the
beginning of Elembiv year 2 appears to be making up for a lost IVOS day at the end of Equos
(ibid. 92–3). However, IVOS days are perhaps not sufficiently attested in the extant fragments for
this inference to be conclusive.
(^16) If designed fromfirst principles, one would expect the length of the month of Equos to have
alternated through thefive-year cycle (e.g. with the sequence 29– 30 – 29 – 30 – 29), without resort
to a 28-day month. A 28-day month is unusual in lunar calendars: its putative occurrence in the
Greek Callippic cycle (on which see Ch. 1. 3) and in the lunar calendar of 1 Enoch has been
rejected as unlikely (Stern 2001: 6–7).
(^17) To be more precise, however, the 30-year cycle would still end with a discrepancy of over
one day from the solar year; but this could have gone unnoticed for several cycles.
(^18) Pliny,NaturalHistory16. 250:sexta lunaquae principia mensum annorumque his[the
druids]facit et saeculi post tricesimum annum, cited as evidence by Duval and Pinault (1986)
401, Olmsted (1992) 21, 61, 73, 132. This is the only literary source available on the Gallic
calendar; the implication, at the very least, is that the Gallic calendar was lunar. Onsaeculum, see
below. n. 20.
(^19) So Duval and Pinault (1986) 401, 415–17. Monard (1999) 83–104 argues that the Gallic
month began at the full moon, and dismisses Pliny’s report as erroneous. Olmsted (1992) 20–1,
61, 133 argues that Pliny’s comment refers to an earlier version of the Coligny calendar (see
further below, near n. 23).
Dissidence and Subversion 305