From this curious comparison, it becomes apparent that the variations of the difference in question depend upon the amount of variations of temperature which take place in the free air, and on the slowness with which such changes are communicated to the stagnant atmosphere of woods; in other words, as Herr Rivoli boldly formulates it, a forest is simply a bad conductor. But this is precisely the same conclusion as we have already arrived at with regard to individual trees; and in Herr Rivoli's table, what we see is just another case of what we saw in M. Becquerel's--the different progression of temperatures. It must be obvious, however, that the thermal condition of a single tree must be different in many ways from that of a combination of trees and more or less stagnant air, such as we call a forest. And accordingly we find, in the case of the latter, the following new feature: The mean yearly temperature of woods is lower than the mean yearly temperature of free air, while they are decidedly colder in summer, and very little, if at all, warmer in winter. Hence, on the whole, forests are colder than cleared lands. But this is just what might have been expected from the amount of evaporation, the continued descent of cold air, and its stagnation in the close and sunless crypt of a forest; and one can only wonder here, as elsewhere, that the resultant difference is so insignificant and doubtful. We come now to the third point in question, the thermal influence of woods upon the air above them. It will be remembered that we have seen reason to believe their effect to be similar to that of certain other surfaces, except in so far as it may be altered, in the case of the forest, by the greater extent of effective radiating area, and by the possibility of generating a descending cold current as well as an ascending hot one. M. Becquerel is (so far as I can learn) the only observer who has taken up the elucidation of this subject. He placed his thermometers at three points:[55] A and B were both about seventy feet above the surface of the ground; but A was at the summit of a chestnut tree, while B was in the free air, fifty feet away from the other. C was four or five feet above the ground, with a northern exposure; there was also a fourth station to the south, at the same level as this last, but its readings are very seldom referred to. After several years of observation, the mean temperature at A was found to be between one and two degrees higher than that at B. The order of progression of differences is as instructive here as in the two former investigations. The maximum difference in favour of station A occurred between three and five in the afternoon, later or sooner according as there had been more or less sunshine, and ranged sometimes as high as seven degrees. After this the difference kept declining until sunrise, when there was often a difference of a degree, or a degree and a half, upon the other side. On