of 12 months; if on the third
evening, that the year ought to be a full one of 13 months.

This was precisely the method followed by the Akkadians some 4000 years
ago. For Prof. Sayce and Mr. Bosanquet translate an old tablet in
Akkadian as follows:--

"When on the first day of the month _Nisan_ the star of stars
(or _Dilgan_) and the moon are parallel, that year is normal.
When on the third day of the month _Nisan_ the star of stars
and the moon are parallel, that year is full."[315:1]

The "star of stars" of this inscription is no doubt the bright star
Capella, and the year thus determined by the setting together of the
moon and Capella would begin on the average with the spring equinox
about 2000 B.C.

When Capella thus marked the first month of the year, the "twin stars,"
Castor and Pollux, marked the second month of the year in just the same
way. A reminiscence of this circumstance is found in the signs for the
first two months; that for the first month being a crescent moon "lying
on its back;" that for the second month a pair of stars.

The significance of the crescent being shown as lying on its back is
seen at once when it is remembered that the new moon is differently
inclined to the horizon according to the time of the year when it is
seen. It is most nearly upright at the time of the autumn equinox; it is
most nearly horizontal, "lying on its back," at the spring equinox. It
is clear from this symbol, therefore, that the Babylonians began their
year in the spring.

[Illustration: POSITION OF THE NEW MOON AT THE EQUINOXES.]

This method, by which the new moon was used as a kind of pointer for
determining the return of the sun to the neighbourhood of a particular
star at the end of a solar year, is quite unlike anything that
commentators on the astronomical methods of the ancients have supposed
them to have used. But we know from the ancient inscription already
quoted that it was actually used; it was eminently simple; it was bound
to have suggested itself wherever a luni-solar year, starting from the
observed new moon, was used. Further, it required no instruments or
star-maps; it did not even require a knowledge of the constellations;
only of one or two conspicuous stars. Though rough, it was perfectly
efficient, and would give the mean length of the year with all the
accuracy that was then required.

[Illustration: BOUNDARY-STONE IN THE LOUVRE; APPROXIMATE DATE, B.C.
1200.

(