and the solar year. Thus in 190BCEa solar eclipse was recorded on 11 Quintilis
(July) of the Roman calendar, when we know it occurred on (Julian) 14
March—a discrepancy of nearly four months, obviously caused by insufficient
intercalation. By 168BCE, however, a lunar eclipse was recorded on the night
before (Roman) 4 September, instead of (Julian) 21 June—a discrepancy of
only about two-and-a-half months, which suggests that by then the discrep-
ancy had been partly corrected through additional intercalation.^130 In Julius
Caesar’s time, the year had fallen short of some 90 days; the failure to
intercalate which caused this shortfall can be broadly attributed to the civil
wars and the political upheavals of the end of the Republican period.^131 For
this reason, 46BCE was extraordinarily extended into a 445-day year,^132
whereupon the Republican calendar was abandoned and the Julian calendar
was instituted.
The Republican calendar: originsThere are several indications that the Roman calendar, like most other calen-
dars in Antiquity, was originally lunar. Its month-lengths (in the Republican
period) of 28, 29, and 31 days and year-length of 355 days are loosely
compatible with a lunar calendar.^133 This may explain why Appian (second
centuryCE) and Dio (third centuryCE) state, somewhat oddly, that the months
of the Republican calendar were lunar,^134 although in reality, in the Republi-
can calendar, the intercalations of 22 or 23 days ruined their synchronism with
(^130) By‘Julian’I mean the backward (hence anachronistic) projection of the Julian calendar,
which is commonly used by modern historians and astronomers as a stable and well-known time
frame for ancient, pre-Julian times;‘Roman’refers to the Republican calendar that was then in use;
‘Quintilis’is the month later to be named Julius (July). The sources for these two events are Livy 37.
- 4, 44. 37. 8. See Bickerman (1968) 46, Samuel (1972) 163, Hannah (2005) 111–12, and in more
detail, Rüpke (1995) 290–2 (with further references) and Bennett (2005) 177–8. Bickerman
attributes the 190BCEdiscrepancy to a failure to intercalate during the Hannibalic wars, which
were concluded in 202 131 BCE, after which the discrepancy must have taken several decades to correct.
See references in previous n. Assuming with Brind’Amour (1983) 76–8 and Hannah loc.
cit. that the Roman calendar was aligned with the seasons around 70BCEand through much of
the subsequent decade (but see the reservations of Samuel 1972: 162–3 on the use of Plutarch,
Pompeius34 as evidence for 66BCE), four intercalations would have been omitted between the
mid-60s and 46BCE: for various reconstructions, see Beaujeu (1976), Brind’Amour (1993: 35–54,
123), Bennett (2004 132 b).
Suetonius,Julius Caesar40, but the sources do not all agree on thisfigure. For detailed
discussions see Ideler (1825–6) ii. 122–3, Beaujeu (1976), Rüpke (1995) 384–9.
(^133) Even though lunar month-lengths should not be less than 29 days or more than 30 days,
and an average lunar year of 355 days is slightly excessive: Michels (1967) 16–22, 145–72. The
lunar character of the 355-day year is unduly dismissed by Lehoux (2007: 49–50). See Ch. 3 n. 61.
(^134) Appian,CivilWar, 2. 154; Dio 43. 26; and for further sources see Samuel (1972) 159 n. 2,
168 n. 1, Brind’Amour (1983) 33–4. The same assumption seems to have been made by Plutarch
(Grafton and Swerdlow 1988: 18).
TheRise of the Fixed Calendars 207