cloudy days the difference tended to a minimum. During a rainy month of April, for example, the difference in favour of station A was less than half a degree; the first fifteen days of May following, however, were sunny, and the difference rose to more than a degree and a half.[56] It will be observed that I have omitted up to the present point all mention of station C. I do so because M. Becquerel's language leaves it doubtful whether the observations made at this station are logically comparable with those made at the other two. If the end in view were to compare the progression of temperatures above the earth, above a tree, and in free air, removed from all such radiative and absorptive influences, it is plain that all three should have been equally exposed to the sun or kept equally in shadow. As the observations were made, they give us no notion of the relative action of earth-surface and forest-surface upon the temperature of the contiguous atmosphere; and this, as it seems to me, was just the _crux_ of the problem. So far, however, as they go, they seem to justify the view that all these actions are the same in kind, however they may differ in degree. We find the forest heating the air during the day, and heating it more or less according as there has been more or less sunshine for it to absorb, and we find it also chilling it during the night; both of which are actions common to any radiating surface, and would be produced, if with differences of amount and time, by any other such surface raised to the mean level of the exposed foliage. To recapitulate: 1st. We find that single trees appear to act simply as bad conductors. 2nd. We find that woods, regarded as solids, are, on the whole, slightly lower in temperature than the free air which they have displaced, and that they tend slowly to adapt themselves to the various thermal changes that take place without them. 3rd. We find forests regarded as surfaces acting like any other part of the earth's surface, probably with more or less difference in amount and progression, which we still lack the information necessary to estimate.