Regularity: the new moonAs mentioned above, Meritt believed against Pritchett that the festival calen-
dar at Athens (and elsewhere in Greece) was fundamentally regular.^44 This
was largely an assumption which, however reasonable, remained difficult to
prove. It meant that the attested cases of calendar irregularity (some of which
have been discussed above) had to be dismissed, rather unconvincingly, as
marginal or merely exceptional.^45 Positive evidence of regularity, as we shall
now see, is difficult tofind.
The assumption that the month was lunar is well attested already infifth-
centuryBCEsources such as Aristophanes,The Clouds 615 – 26 (see above), but
this does not meanipso factolunar regularity—as the same passage in Aris-
tophanes proves. Likewise, the names‘old and new’andnoumeniafor the last
andfirst day of the month respectively, which seem to imply a lunar calendar
regulated by the new moon,^46 do not necessarily reflect how the calendar was
reckoned in practice.^47 But another passage inThe Cloudsdoes seem to
calendar in Athens; but that such an attempt was ever made is completely unsubstantiated and
highly unlikely.
(^44) In some publications, Meritt (e.g. 1964) went as far as arguing that the Athenian calendar
conformed to the astronomical, Metonic calendar; this will be criticized below (}3). Meritt’s
position has ledWoodhead (1997), in his reconstruction of the Athenian epigraphic record, to
assumea prioria Metonic cycle, with a regular alternation of full and hollow months. However,
this assumption is unwarranted and indeed misleading, as—onWoodhead’s own admission—it
is regularly contradicted by the evidence. 45
Following this line of argument, Dunn (1998), followed in turn by Hannah (2005) 49, 51,
explains the attested cases of calendar disruption as being the result not of deliberate tampering
(Pritchett’s thesis), but rather of imprecisions inherent in the way the lunar calendar was
reckoned. He argues that discrepancies between the calendar and the true lunar month may
have resulted from delayed new moon sightings because of bad weather, or from inaccurate
predictions as to when the new moon wouldfirst be visible. But whilst this may well explain
minor discrepancies, e.g. why Athens was two days ahead of Sparta in 423BCEand two days
behind in 421 (ibid. 219; see above, n. 38), this does not explain how greater discrepancies could
occur. Dunn’s contention (1998: 214–17) that in a calendar based purely on new moon sighting,
the full moon can sometimes occur on the 9th of the month (which would explain Herodotus 6.
106) is plainly incorrect. Similarly implausible is his explanation (ibid. 220) of how e.g. a 5-day
discrepancy can accumulate between two lunar calendars (which would explain Aristoxenus,
Harmonica2. 37; see n. 37 above). In a number of cases, indeed, Dunn himself is forced to
concede that the discrepancy must have been due to deliberate tampering. Even if this tampering
was legitimate (e.g. for religious reasons, as Dunn attempts to explain away, e.g. 1998: 219; see
discussion below,}5), the fact remains that in these cases, the relationship between the calendar
and the lunar month was profoundly disrupted. 46
‘Old and new’(íÅ ŒÆd íÝÆ) designates the last day of the month, regardless of whether it is
hollow or full: see e.g. Pritchett and Neugebauer (1947) 23–31 (misinterpreted by Bowen and
Goldstein 1994: 697 n. 9) and n. 6 above. It probably means that the old moon is no longer
visible, but the new moon is not yet visible either (see further discussion in Hannah 2005: 43–4).
Noumeniameans‘new moon’, and can sometimes be used in this sense, i.e. the day when the new
moonfirst appears (e.g. Thucydides 2. 28. 1, on which see below,}4).
(^47) Indeed in the Roman period, the beginning of Julian calendar months was still called in
Greeknoumenia, even though the calendar was not lunar. In the Roman calendar, the terms
36 Calendars in Antiquity