With the two opening days of the calendar and the one division in the middle are often combined the three great sacrificial feasts, the autumn festival at the winter nights, the Yule festival at mid-winter, and the spring festival at the summer nights. It is true that the first of these festivals, which was celebrated at the beginning of a period of rest after the completion of the harvest and agricultural labour, denoted, as such festivals often do, the conclusion of the old year and the beginning of the new. That it was fixed for a definite day cannot be demonstrated any more than that the festival of victory in spring, celebrated before the Vikings went forth on their voyages, fell exactly on the summer night. On the contrary the time probably varied according to circumstances: the expression of Snorre lacks calendarial accuracy and remains indefinite:--“They should sacrifice against the winter to get a good year, and at mid-winter sacrifice for germination; the third sacrifice in summer, and this was a sacrifice of victory”[339]. In historical times the Yule festival is regulated by the Christian calendar; Snorre says that in heathen times it was celebrated at the _hökku_ night, but of this we have no certain knowledge. Things happened as in the Middle Ages and later: after a calendar has arisen the festivals are regulated by this, but they are not calendar-festivals, and in reconstructing the scheme of the calendar from the festivals very great caution must be exercised. Our conclusion is that the Germanic seasons, like the seasons in general, were not in themselves definitely limited divisions of time, and that alongside of the greater seasons smaller ones arose without there being any numerical determination of the relationship between the two. Seasons only become divisions consisting of a definite number of days when in the regulation of the calendar they are taken over as calendar divisions, as winter and summer were in Scandinavia. Where a calendar has arisen directly out of the seasons, the divisions, like the seasons, are of varying length[340]. This also shews that the Germanic seasons first attained a definite number of days through the calendar-regulation introduced from abroad. Further, when a calendar existed, the beginning of the seasons could be given with reference to this: the day varied according to circumstances, but the choice was limited in this manner, viz. that only a popular festival or saint’s day was appropriate as a distinguishing day. Here also, therefore, the calendar was the starting-point for the regulation of the seasons. A division of the year in the more accurate sense also first arose through the regulation of the calendar, since, owing to the method of calculation, the middle days of the half-year divisions became distinguishing days in the calendar. When the calendar came, the old festivals were also regulated by it.