~ 3 ~
weeks of either of these lengths would be of no use to the moon observer. They would lead him wrong at the very outset. The only way of fitting a seven-day week into the lunar month of 29 to 30 days, or for the matter of that into the artificial month of 30 to 31 days, is to have one week at least of the four longer than seven days, or what is the same thing to intercalate one or more days at the end of the four weeks. Similarly an eight-day week can only be fitted into the month by shortening one at least of the four weeks. Of such a subdivisional or, as I will for convenience call it, a 'lunar' week, there are apparently traces in the primitive records of Babylonia and of ancient Persia[1]. But the 'lunar' week differs vitally from the continuous, and, while it is possible that the latter may have been developed from the former, I do not think the possibility rises above the region of speculation. The earliest forms of the continuous week of which we have any knowledge were justified by the nations which used them on grounds which have nothing to do with the moon or the month. There are, as I have said, three such forms, the Roman, the Jewish and the Planetary.
The Roman usage, by which the nundinae[2],
