In the last chapter I referred to one of the difficulties of modern inquiries into the orientation of ancient temples, which arises from the fact that the sun has not always, at the solstices, risen or set at exactly the same points of the horizon. We now find ourselves face to face with the fact that the stars do not rise or set at the same points century after century. We saw that the change in the position of the sun on the horizon at the solstices is due to a very small change of obliquity of the ecliptic, so that in a matter of something like 6,000 years the position of the sun at sunrise and sunset on the horizon may be varied by, roughly speaking, 1 degree. But in the case of the stars the matter is very much more serious, because in the course of something like 13,000 years the rising-or setting-places of a star may vary by something like 47° along the horizon north or south. So that in the cases both of sun and stars there is no real fixity in the places of rising or setting, although of course those who made the first observations and built the first temples were not in a position to know this. The real cause of this precessional movement which causes the stars to change their places lies in the fact that the earth is not a sphere, its equatorial diameter being longer than its polar diameter, so that there is a mass of matter round the equator in excess of what we should get if the earth were spherical. Suppose that matter to be represented by a ring. The ring is differently presented to the sun, one part being nearer than the other, the nearer part being attracted more forcibly. If we take the point in the ring nearest the sun where there is the greatest attraction, and draw a line to the opposite point where the attraction is least, we can show that the case stands in this way: the sun's pull may be analysed into two forces, one of them represented by the line joining the centre of the sun and the centre of the ring, and another at right angles to it let fall from the point most strongly attracted on to the first line. The question is, what will that force at right angles do? The figure below represents a model illustrating the rotation of the earth on its axis, and the concurrent revolution of the sun round the earth once a year. To represent the downward force it is perfectly fair if I add a weight. The moment this is done the axis of the gyroscope representing the earth's axis, instead of retaining its direction to the same point as it did before, now describes a circle round the pole of the heavens. [Illustration: MODEL ILLUSTRATING THE PRECESSION OF THE EQUINOXES.]