Such an observation is useful for more purposes than one. If the position of the star occulted is known--and it can be determined at leisure afterwards--we necessarily know where the limb of the moon was at the time of the observation. Then the time which the moon took to pass over the star enables us to calculate the diameter of our satellite; the different positions of the moon relative to the star, as seen from different observatories, enable us to calculate its distance. But if the disappearance takes place at the bright limb, the reappearance usually takes place at the dark, and _vice versâ_; and the two observations are not quite comparable. There is one occasion, however, when both observations are made under similar circumstances, namely, at the full. And if the moon happens also to be totally eclipsed, the occultations of quite faint stars can be successfully observed, much fainter than can ordinarily be seen close up to the moon. Total eclipses of the moon, therefore, have recently come to be looked upon as important events for the astronomer, and observatories the world over usually co-operate in watching them. October 4, 1884, was the first occasion when such an organised observation was made; there have been several since, and on these nights every available telescope and observer at Greenwich is called into action. It may be asked why these different modes of observing the moon are still kept up, year in and year out. 'Do we not know the moon's orbit sufficiently well, especially since the discovery of gravitation?' No; we do not. This simple and beautiful law--simple enough in itself, gives rise to the most amazing complexity of calculation. If the earth and moon were the only two bodies in the universe, the problem would be a simple one. But the earth, sun, and moon are members of a triple system, each of which is always acting on both of the others. More, the planets, too, have an appreciable influence, and the net result is a problem so intricate that our very greatest mathematicians have not thoroughly worked it out. Our calculations of the moon's motions need, therefore, to be continually compared with observation, need even to be continually corrected by it.