While we thus recognise the possibility, at any rate, that multitudes of small lunar craters, say from a quarter of a mile to two miles in diameter, may have been formed by falling meteoric masses hundreds of millions of years ago, and may have remained unchanged even until now, we perceive that on the moon later processes must have formed many small craters, precisely as such small craters have been formed on our own earth. I consider, at the close of the essay above mentioned, the two stages of the moon's development which must have followed the period during which her surface was wholly or in great part plastic. First, there was the stage during which the crust contracted more rapidly than the nucleus, and was rent from time to time as though the nucleus were expanding within it. Secondly, there came the era when the nucleus, having retained a greater share of heat, began to cool, and therefore to contract more quickly than the crust, so that the crust became wrinkled or corrugated, as it followed up (so to speak) the retreating nucleus. It would be in the later part of this second great era that the moon (if ever) would have resembled the earth. The forms of volcanic activity still existing on the earth seem most probably referable to the gradual contraction of the nucleus, and the steady resulting contraction of the rocky crust. As Mallet and Dana have shown, the heat resulting from the contraction, or in reality from the slow downfall of the crust, is amply sufficient to account for the whole observed volcanian energy of the earth. It has indeed been objected, that if this theory (which is considered more fully in my "Pleasant Ways in Science") were correct, we ought to find volcanoes occurring indifferently, or at any rate volcanic phenomena of various kinds so occurring, in all parts of the earth's surface, and not prevalent in special regions and scarcely ever noticed elsewhere. But this objection is based on erroneous ideas as to the length of time necessary for the development of subterranean changes, and also as to the extent of regions which at present find in certain volcanic craters a sufficient outlet for their subterranean fires. It is natural that, if a region of wide extent has at any time been relieved at some point, that spot should long afterwards remain as an outlet, a sort of safety-valve, which, by yielding somewhat more quickly than any neighbouring part of the crust, would save the whole region from destructive earthquakes; and though in the course of time a crater which had acted such a part would cease to do so, yet the period required for such a change would be very long indeed compared with those periods by which men ordinarily measure time. Moreover, it by no means follows that every part of the earth's crust would even require an outlet for heat developed beneath it. Over wide tracts of the earth's surface the rate of contraction may be such, or