Any year that is regulated by the Pleiades, or by any other group of stars, must, as we have seen before, be what is called a sidereal, and not a solar year. Now a year in uncivilised countries can only mean a succession of seasons, as is illustrated by the use of the expression "a person of so many summers." It is difficult of course to say when any particular season begins by noticing its characteristics as to weather; even the most regular phenomena are not certain enough for that; we cannot say that when the days and nights become exactly equal any marked change takes place in the temperature or humidity of the atmosphere, or in any other easily-noticed phenomena. The day therefore on which spring commences is arbitrary, except that, inasmuch as spring depends on the position of the sun, its commencement, ought to be regulated by that luminary, and not by some star-group which has no influence in the matter. Nevertheless the position of such a group is much more easily observed, and in early ages could almost alone be observed; and so long as the midnight culmination of the Pleiades--judged of, it must be noticed, by their appearance _on the horizon_ at sunset--fairly coincided with that state of weather which might be reckoned the commencement of spring conditions, no error would be detected, because the change in their position is so slow. The solar spring is probably a later discovery, which now, from its greater reasonableness and constancy, has superseded the old one. But since the time of the sun's crossing the equator is the natural commencement of spring, whether discovered or not, it is plain that no group of stars could be taken as a guide instead, if their indication did not approximately coincide with this. If then we can determine the exact date at which the Pleiades indicated by their midnight culmination the sun's passage across the equator, we can be sure that the spring could only have been regulated by this during, say, a thousand years at most, on either side of this date. It is very certain that if the method of reckoning spring by the stars had been invented at a more remote date, some other set of stars would have been chosen instead. Now when was this date? It is a matter admitting of certain calculation, depending only on numbers derived from observation in our own days and records of the past few centuries, and the answer is that this date is about 2170 B.C.