MOVING CLUSTERS. A third line of evidence, which also tells much the same story, may be briefly mentioned. The conspicuous groups of bright stars in the sky, such as the Great Bear, the Pleiades and Orion’s Belt, consist for the most part of exceptionally massive stars which move in regular orderly formation through a jumble of slighter stars, like a flight of swans through a confused crowd of rooks and starlings. Swans continually adjust their flight so as to preserve their formation. The stars cannot, so that their orderly formation must in time be broken by the gravitational pull of other stars. The lighter stars are naturally knocked out of formation first, while the most massive stars retain their formation longest. Observation suggests that this is what actually happens to a moving star-cluster; at any rate the stars which remain in formation generally have weights far above the average. And, as we can calculate the time necessary to knock out the lighter stars, we can at once deduce the ages of those which are left in. The result of the calculation confirms those already mentioned, so that we find that the three available astronomical clocks all tell much the same time. They agree in indicating an age of the order of five to ten millions of millions of years for the stars as a whole. Another line of investigation, to be mentioned later (p. 188) again points to a similar age. It is perhaps a little surprising that this age should prove to be so much longer than the age of the earth, although there is of course no positive reason why the earth should not have been born during the last few moments of the lives of the stars. It is perhaps also a little surprising that it should prove to be much longer than the age suggested, very vaguely it is true, by the cosmologies of de Sitter and Lemaître. If we accept the apparent velocities of recession of the most distant nebulae as real, we find that some thousands of millions of years of motion at their present speeds would just about account for their present distances from us, so that a few thousands of millions of years ago, the nebulae must have been far more huddled together than they now are. This is of course very different from saying that the time which has elapsed since the creation of the nebulae can only be a few thousands of millions of years, yet we might reasonably have expected _à priori_ that the two periods would be at least comparable. To state the difficulty in a slightly different form, a period of two thousand million years seems to have made a great deal of difference to the earth, and if the apparent speeds of nebular motion are real, it has made a great deal of difference to the general arrangement in space of the great nebulae, so that it is odd that it should make so little difference to the stars that we need to postulate an age a thousand times as great before we can explain their present condition.