The 19-year cycleThe institution of a 19-year cycle is generally ascribed to Darius’reign, around
the turn of thefifth century.^110 It is generally assumed that afixed 19-year
cycle was instituted as such from its inception, and that occasional deviations
from the cycle, during the following century, were merely errors or‘excep-
tions’. This common, scholarly perception must be revised, however, on two
counts. On the one hand, a 19-year cycle can be identified already from the
beginning of Darius’reign (in 522), and not at the turn of thefifth century. On
the other hand, it seems evident that the cycle remained subject to variation
until the Seleucid period (in the late fourth century). Deviations from a
standard 19-year cycle in the Achaemenid period should not be interpreted
as errors or exceptions, but rather as evidence that no standard cycle had yet
beenfixed.
Thefixed 19-year cycle that eventually emerged by the Seleucid period, as
attested at least in the Saros Canon texts, consists of a sequence of intercala-
tions at the following intervals:
Saros Canon 19-year cycle 3 – 3 – 2 – 3 – 3 – 2½–2½I shall refer to this as the‘Saros Canon cycle’(because it appears in the so-
called‘Saros Canon’texts); it is not to be confused with the‘Saros cycle’itself,
which is an 18-year cycle of lunar and/or solar eclipses (which these texts
present). The possible relationship between these two cycles, however, will be
considered below.
The rationale behind this 19-year sequence is not entirely clear. A more
homogeneous sequence could have been achieved by suppressing the short
interval of 2 years (in third position), and thus spacing out intercalations more
evenly. This would only have required a small modification to the second and
third intervals of the sequence, as follows:
Improved 19-year cycle 3 – 2½–2½– 3 – 3 – 2½–2½An even more homogeneous sequence could then have been achieved by
inverting the third and fourth intervals (see Table 2.3):
Optimal 19-year cycle 3 – 2½– 3 – 2½– 3 – 2½–2½Such improved cycles are not attested, however, in the Babylonian calendar.
Why the Saros Canon cycle was never modified as above is a question I shall
return to. It so happens that the latter,‘optimal’sequence occurred from 536/5
(^110) This common view supersedes an earlier theory that the cycle was instituted in the 4th c.;
for references, see above, n. 100. There is not much point in surveying the various starting-dates
that have been more specifically suggested (503/2, 498/7, 482/1, etc.).
The Babylonian Calendar 105