formed a part of its mass, was travelling backward too, according to what we have said. Then under the friction of the tides raised on the planet by the Sun, the planet proceeded to turn over. It continued to do so until it spun direct. During this process there was no passage through zero of its moment of momentum considered with regard to itself, and therefore no difficulty on that score of supposing that it successively generated satellites at all degrees of inclination. That its children are of the nature of adopted waifs, Babinet's criterion (1861) would seem to imply. But it must be remembered that the Sun has been slowing up the planet's rotation now for aeons. As it turned over, its tidal bulges tended to carry over with it such satellites as it already had. This effect was much greater on the nearer ones, both because they were nearer and because they were much larger than the outer. So that the nearer kept with the planet, the others lagged proportionately behind. This suggests itself to account for the facts, but the subject involves so much that is uncertain that I submit the hypothesis with the distrust which Laplace has so eminently bespoken. I advance in its favor only the three striking facts: that a steady progression in their tilts of rotation is observable from Neptune to Jupiter and a substantially accordant one from Mars to Mercury; secondly, that the satellites turn their faces to their primaries, as likewise do Mer-
