the day and the year; the structure of calendars was thus determined by the
number of days in the months, and the number of months in the year. The
month was the fundamentalcomponent oflunar calendars, where it corre-
sponded, atleast in theory, to a fullcycle of the phases of the moon, normally
(but not always) beginning from the new moon; the typicallength of alunar
month was 29 or 30 days.Insolar calendars—of which, in Antiquity, the Julian
calendar was the only true exemplar—the fundamentalcomponent of the
calendar was the year, corresponding to a fullcycle of the sun or the seasons,
approximately365¼days. The division of the solar year into months is
artificial, as its twelve months are somewhatlonger thanlunar months (count-
ing 30 or 3 1 days) and out of step with the moon’s phases.Inlunar calendars,
conversely, the year is only an artificialsequence of twelve or thirteen months,
neither of which corresponds to a solar year or any other astronomicalreality.
Nearlyall lunar calendars in Antiquity counted twelve months in the year, but
occasionally added a thirteenth month (a procedure known as‘intercalation’)
so as to keep up, atleast approximately, with the sun and the seasons. These
calendars are sometimes called‘lunisolar’, a term, however, thatIshallgener-
ally not use.^7
Some ancient calendars also employed smaller time units, sometimes but
not always subdivisions of the month: for example, thehamuštumin early-
second-millennium Assyrian documents, which some have identified as a
seven-day week, but which alternatively may have meant a quarter of the
lunar month (thus variably seven or eight days); the decad or ten-day period in
Greek calendars representing a third of thelunar month (the third numbering
sometimes only nine days); the eight-daynundinaeor market week in the
Roman calendar; and the seven-day week in the Jewish calendar,later adopted
by Romans and Christians as a planetary, astrological scheme or as a
Jewish, biblicaltradition.^8 In this study, however,Iam more interested in
(^7) AlthoughIhave been criticized in the past for not using this term,Ishallrefrain from it
again in this work for severalreasons (see also Stern 200 1 : 1 – 2). The term‘lunisolar’is in my
view a misnomer, because it implies thatlunar and solar elements are equivalent in status;
whereas in most of these calendars thelunar element was dominant, as it defined thelength of
the months, whilst the solar year was only approximately tracked by a calendar year of twelve or
thirteen months. Furthermore,‘lunisolar’assumes that asolarcriterion governs the intercalation,
whereas in most ancient calendars it was on the basis of the seasons, the stars, or afixed
schematic cycle that the intercalation was governed. The term‘lunisolar’is also unnecessary,
because purelylunar calendars (i.e. without intercalation) are not attested in Antiquity before the
institution of theIslamic calendar in the seventh centuryCE(on the old Assyrian calendar, see
Ch. 2 n. 78); in the context of Antiquity, therefore, there is no need for a specialterminology to
distinguishlunar calendars with and without intercalation.
(^8) Hamuštumas a seven-day week: Veenhof ( 1995 – 6). The Greek decad:Ch. 1 nn. 1 3 and 67;
decads are also attested in ancient south Arabia (de Blois 1 998).Nundinae: Rüpke ( 1995 ). The
Jewish week:Ch. 4 n. 11 3; the seven-day week in Roman sources, Blackburn andHolford-
Strevens ( 1 999) 566 – 8 andCh. 6 n. 6 5.
6 Calendars in Antiquity