But now comes in a most interesting and important point. If observations of the sun at solstice or equinox had been alone made use of, the true length of the year would have been determined in a few years. But the next scene in Egyptian history shows us that the true length of the year was not determined, but only an approximation to it. How was this? The astronomical answer is very simple. I have already referred to the common practice of all ancient peoples that we know of to make sacrifices at dawn, and have shown how, in order to do this, they took their time from a star rising before the sun. An observation of the so-called "heliacal rising" of a star--if the star were properly chosen--would give them the interval necessary for their preparations before the sun itself appeared; and, as the highest festival of all was that of New Year's Day, it was especially important that the work should be well done then. Now, if the stars had no precessional movement, the sun and stars, after each interval of a true year, would be in exactly the same position; but in consequence of the stars having the precessional movement to which I have before referred, the star so observed and the sun will _not_ be in exactly the same position after the interval of a true year. On this account, then, the difference of time between the heliacal risings will not represent the length of a true year. But, further, the heliacal rising of the star will not take place on the same day for the whole of Egypt, the difference between Thebes and Memphis, depending upon their latitudes, amounting to about four days; and, further still, the almost constant mists in the mornings in the Nile valley prevent accurate observations of the moment of rising. Still, as a matter of fact, the Egyptians defined their new year by the rising of a star, and the length of it by the interval separating two heliacal risings. Such a year could not be accurate; and again, as a matter of fact, their correction was not accurate, for the year was defined now as consisting of 365 days. It seems clear from this that the correction was made before the solar temples were in use. In any case the year of 360 days had naturally to give way, and it ultimately did so, in favour of one of 365. The precise date of the change is, as we have seen, not known.[71] The five days were added as epacts or epagomena; the original months were not altered, but a "little month" of five days was interpolated at the end of the year between Mesori of one year and Thoth of the next, as already stated.