sting of all the ancient constellations. If we could determine the
origin of these figures, their exact configuration as at first devised,
and the precise influences assigned to them in the old astrological
systems, we should have obtained important evidence as to the origin of
astronomy itself. Not indeed that the twelve signs of the zodiac were
formed at the beginning or even in the early infancy of astronomy. It
seems abundantly clear that the division of the zodiac (which includes
the moon's track as well as the sun's) had reference originally to the
moon's motions. She circuits the star-sphere in about twenty-seven days
and a third, while the lunation or interval from new moon to new moon
is, as we all know, about twenty-nine days and a half in length. It
would appear that the earliest astronomers, who were of course
astrologers also, of all nations--the Indian, Egyptian, Chinese,
Persian, and Chaldaean astronomers--adopted twenty-eight days (probably
as a rough mean between the two periods just named) for their chief
lunar period, and divided the moon's track round the ecliptic into
twenty-eight portions or mansions. How they managed about the fractions
of days outstanding--whether the common lunation was considered or the
moon's motion round the star-sphere--is not known. The very
circumstance, however, that they were for a long time content with their
twenty-eight lunar mansions shows that they did not seek great precision
at first. Doubtless they employed some rough system of 'leap-months' by
which, as occasion required, the progress of the month was reconciled
with the progress of the moon, just as by our leap-years the progress of
the year is reconciled with the progress of the sun or seasons.

The use of the twenty-eight-day period naturally suggested the division
of time into weeks of seven days each. The ordinary lunar month is
divided in a very obvious manner into four equal parts by the lunar
aspects. Every one can recognise roughly the time of full moon and the
times of half moon before and after full, while the time of new moon is
recognised from these two last epochs. Thus the four quarters of the
month, or roughly the four weeks of the month, would be the first
time-measure thought of;--after the day, which is the necessary
foundation of all time measures. The nearest approach which can be made
to a quarter-month in days is the week of seven days; and although some
little awkwardness arose from the fa