among which suitable and well-known days were chosen in the dating of the beginnings of the seasons also. For the general understanding it was necessary throughout to bring in popular saints’ days[332]. Tille attacks Weinhold very sharply but remains throughout under the influence of the method indicated by the latter: his work, however, has its good points, inasmuch as it refers to economic conditions, agriculture, the payments of rent, etc. The bipartite division, he asserts, is primitive Indo-European, the tripartite is of foreign (Egyptian) origin: both existed for a long time side by side. This fact is explained by an old sexpartite division of the year, since the six seasons could be run together either in twos or in threes. The beginnings of the half-years are given by natural phenomena, those of the three annual divisions are placed by Tille at March 13, July 10, and Nov. 11, old style: in the north on account of the climatic conditions they are pushed back a month. Hammarstedt[333] remarks very pertinently that the beginning of winter in November, in the north in October, belongs to the reckoning in half-years, and that hence arises the absurdity that Tille has to give Feb. 10 as the date for the beginning of spring in the north. But to assign Dec. 13 with Hammarstedt as the beginning of one of the three seasons agrees just as little with the natural seasons of the year. The principal error lies in the systematising, the seasons being regarded as periods of a definite number of days. This is not the case even to-day, and still less was it so, as we have seen, among primitive peoples. Still more clearly does the same error of method appear in Tille’s assumption of a sexpartite division of the year, or of sixty-day periods, as they are expressly termed. He refers to the six old Indian seasons, which are a comparatively late and artificial product called forth by the adoption of the names of the seasons in the reckoning by months[334], and to the pairs of months of the Syrian and Arabian calendar. He regards as 60-day divisions not only the smaller seasons mentioned above, p. 75, the duration of which was originally no less indefinite than it is to-day, but also the Germanic pairs of months, which owe their origin to an adaptation of the Roman months (for this see below, ch. XI). The 60-day periods are so far from being primitive that they first took their origin under the influence of the reckoning in months.