In many cases, the apparent change in the period of the seasons is a purely local phenomenon, which is probably compensated by a higher temperature in other months, without any real disturbance of the average thermometrical equilibrium. We may easily suppose that there are analogous partial deviations from the general law of precipitation; and, without insisting that the removal of the forest has diminished the sum total of snow and rain, we may well admit that it has lessened the quantity which annually falls within particular limits. Various theoretical considerations make this probable, the most obvious argument, perhaps, being that drawn from the generally admitted fact, that the summer and even the mean temperature of the forest is below that of the open country in the same latitude. If the air in a wood is cooler than that around it, it must reduce the temperature of the atmospheric stratum immediately above it, and, of course, whenever a saturated current sweeps over it, it must produce precipitation which would fall upon or near it. But the subject is so exceedingly complex and difficult, that it is safer to regard it as a historical problem, or at least as what lawyers call a mixed question of law and fact, than to attempt to decide it upon _a priori_ grounds. Unfortunately the evidence is conflicting in tendency, and sometimes equivocal in interpretation, but I believe that a majority of the foresters and physicists who have studied the question are of opinion that in many, if not in all cases, the destruction of the woods has been followed by a diminution in the annual quantity of rain and dew. Indeed, it has long been a popularly settled belief that vegetation and the condensation and fall of atmospheric moisture are reciprocally necessary to each other, and even the poets sing of Afric's barren sand, Where nought can grow, because it raineth not, And where no rain can fall to bless the land, Because nought grows there.[176] Before stating the evidence on the general question and citing the judgments of the learned upon it, however, it is well to remark that the comparative variety or frequency of inundations in earlier and later centuries is not necessarily, in most cases not probably, entitled to any weight whatever, as a proof that more or less rain fell formerly than now; because the accumulation of water in the channel of a river depends far less upon the quantity of precipitation in its valley, than upon the rapidity with which it is conducted, on or under the surface of the ground, to the central artery that drains the basin. But this point will be more fully discussed in a subsequent chapter. There is another important observation which may properly be introduced here. It is not universally, or even generally true, that the atmosphere returns its humidity to the local source from which it receives it. The air is constantly in motion,