Clearly we have here to do, not with a distinct type of cloud, but rather with the first step towards the formation of one, or the last stage in the life of one which is drying up. But sometimes the life of the cloud is so short that it never passes beyond this first stage; and it is by no means a universal rule for a growing sheet of cirrus to pass through this stage at all. It therefore represents a peculiar state of instability, and requires a name of its own. Sometimes patches of it will come and go in an apparently capricious manner for an hour or more before permanent condensation is effected or before the sky finally clears. But this is a rare event, since the slow change of conditions which has brought the stratum of air to the unstable condition is generally progressive, and instead of stopping at the critical point, goes beyond it, with the result that the condensation grows or the cloud disappears entirely. Change cirrus, or cirrus inconstans, would be an appropriate name for a kind of cloud which is so plainly indicative of instability. The critical condition referred to is, of course, that in which a particular stratum of air is just saturated, or is just on the point of forming visible cloud. If any cause is brought to bear on such a stratum which brings about even slight cooling, cloud must be produced; and, conversely, anything which results in the slightest heating must cause it to disappear. The shortness and haziness of the fibres, and the fact that they gather themselves into granules, shows that the cloud is formed in a stratum of air which is either still, or is moving as a whole, without any of those differential movements which seem to be necessary for the longer fibrous details. The causes which may bring about the local cooling and heating are easy to understand when we remember how the air will be affected by the uneven contours of the ground. As it passes over hill and valley the up-and-down movements of the lower layers, or even the disturbances caused by passing over a wood or clump of trees, all must be propagated upwards. Each disturbance must slowly spread laterally and diminish vertically, so that it will reach the cirrus zone as a broad and gentle dome-like oscillation. Suppose now a series of such slight upheavals to reach the critical level. The passage of the waves will mean alternate expansion and compression. Expansion means cooling, and therefore cloud-production; compression means heating, and therefore the destruction of cloud.