Another puzzling phenomenon, sometimes presented by cloud and fog, is that our instruments for detecting humidity show that the air within them is not always fully saturated. It seems probable that this is due to such cloud or fog having begun the process of drying up, or that in some way not fully understood the presence of the cloud particles after they have first come into existence may cause the withdrawal of some of the moisture from the intervening damp air. The surface of each minute droplet exerts a pressure on its interior similar to the pressure exerted by the film of a soap-bubble on the air within it, and it is conceivable that some of the uncondensed vapour from outside may diffuse through this enclosing surface film, and be retained there in consequence of the pressure. If this is so, and subsequent investigation can alone decide the matter, it will follow that when once cloud production has begun it will be continued until the air between the cloud particles is reduced so far below its saturation point that the tendency of the drops to evaporate, that is to say, for the imprisoned water to escape through the confining film, balances the retaining pressure. This consideration, however, is quite incompetent to affect the general explanation of cloud formation which has been given. Its result would be to carry condensation a little further than the exact saturation point, and to retard equally slightly the subsequent evaporation of the cloud particles. We have spoken of the typical cumulus as having a roughly pyramidal shape, and if the horizontal movement of the air is small, the loftiest point of the cloud will be situated approximately above the centre of its base. But if the horizontal movement increases in velocity, so that the top is in a more rapidly moving stratum than the base, it will lean forward in the direction of movement. This is a very common phenomenon, being generally shown by cumulus on a windy day. On much rarer occasions the converse occurs, and the top of the cloud lags behind the base, the explanation being a lessening of the velocity of the wind as the height above ground increases. But such conditions rarely occur, and when they do they are due to local eddies and affect only a limited area. Hence such clouds are isolated, and indicate a disturbed state of the air and uncertainty of weather. The clouds which lean forward are formed under conditions which are spread over wide districts, such as the rear of a large cyclone, and cumulus of that kind may follow one another across the sky for hours or even days as long as the wind persists.