The recent discovery of additional double-dated documents which do not
agree with the cycle of pap. Carlsberg 9 (even on Parker’s reading) further
disproves the claim that this cycle determined all lunar datings in the Ptole-
maic and Roman periods.^79
There is evidence, moreover, of other schematic lunar calendars in the
Ptolemaic period, which undermine the notion of a‘standard’25-year lunar
cycle. Pap. Rylands inv. 666, a fragmentary Greek papyrusfirmly dated to 180
BCE,^80 contains a lunar 25-year cycle that starts one year later than in pap.
Carlsberg 9 (its cycle starts in 181BCE, whereas the cycle of pap. Carlsberg 9
would have started in 182BCE). More importantly, it has different dates for
the lunar months^81 and a different sequence of 29- and 30-day months.^82 It is
possible that these cycles were designed in different periods (depending on
when exactly the cycle of pap. Carlsberg 9 is dated); cultural context may also
account for the differences between them, since pap. Rylands is distinctly
Greek not only in language, but also in other features that will be discussed
8 (two dates in 144BCE), 9 (212BCE), and 10 (237BCE), thus six cases out of seventeen. In each of
these cases, the cycle’s date is one day earlier.
(^79) Bennett (2003) 227–8, citing grMedinet Habu 43 and 44 (in both cases, the lunar date is
one day later than pap. Carlsberg 9), and n. 41 with references to further evidence; see also id.
(2008) 527–8, Lippert (2009).
(^80) Turner and Neugebauer (1949–50); Roberts and Turner (1952) iv. 56–62 (no. 589); see also
A. Jones (1997) 162, Depuydt (1998) 1294–5, and Lehoux (2007) 179–80, 474–7 (where it is
identified as P. Rylands 589). The dating to year 1 of Cleopatra and Ptolemy (Philometor)
appears in ll. 92–4, 108–111. The provenance is thought to be Philadelphia (Arsinoite nome):
Turner and Neugebauer (1949 81 – 50) 82, Roberts and Turner loc. cit.
According to pap. Rylands, the second lunar month of 181BCEbegan on 19 Phaophi,
whereas according to pap. Carlsberg 9, it would have been on the 20th. Note that 19 Phaophi
(= 24 Nov. 181BCE) would have been too early, as the old moon would have been still visible on
that morning. However, thefirst month of the pap. Rylands cycle begins on the right day, i.e. 20
Thoth (= 26 Oct. 181BCE), which corresponds to the day of invisibility of the old moon. How
exactly this cycle was designed remains, nevertheless, unclear. The phrasenoumeniakata selenen
(‘first day of month according to the moon’) frequently used in pap. Rylands (e.g. l. 95, cited
below, n. 100, and fr. 7, in Turner and Neugebauer 1949–50: 82) does not mean that the
beginning of the month was determined by lunar observation (as proposed by Depuydt 1998:
1295 n. 13). This phrase, in Greek sources, is used for a month that conforms to the moon and
has not been tampered with, but without implying anything about lunar observation (see Ch. 1.
4); in the context of pap. Rylands, it simply designates the beginning of alunarmonth as opposed
to that of an Egyptian civil month (Turner and Neugebauer 1949 82 – 50: 86).
According to pap. Rylands, the consecutive lunar months beginning in Phaophi and in
Hathyr are both of 30 days (so in 181BCEand—assuming the fragments have been correctly
collated—in 180BCE; the rest of the document is lost, although it does state explicitly, on ll. 97–8,
that the calendar is a 25-year cycle). This is precluded in pap. Carlsberg 9, where the total length
of these two months is always 59 days. Note also that the sequence of months in pap. Rylands
differs in other ways from Parker’sreconstructionof pap. Carlsberg: in the former, the months
from Choiak to Pachons are in a sequence of 29– 30 – 29 – 30 – 29 – 30, whereas in the latter they are
in the reverse.
152 Calendars in Antiquity