depressions of greater height, we have relatively many depressions of less height. Furthermore, there are a great many more of both projections and depressions than there were of the former variety, and they are both of much less height or depth. Evidently, therefore, we have here, in part at least, a different class of phenomena from what we have previously considered. Now we perceive at once that two factors enter here which did not enter in the case of the short and sharp irregularities. The long and low depressions occur, as we shall recall, almost always over the dark areas, while the short and sharp ones do not. In the next place, the average height or depth of the long and low irregularities is much nearer the value of the irradiation constant, that is, the amount by which a bright object seems bigger on account of its brightness; which would cause the dark areas to seem depressed. From these facts we infer that most of the depressions of this class are due to the character, not to the contour, of the surface where they occur; partly to the direct effect of lack of irradiation, partly to sombreness of the surface, which would cause the light to fade from them at a greater relative distance from the terminator. On eliminating these depressions, therefore, we find ourselves left with very few depressions as against nearly 200 projections. The excess in number of the
