=Melting and freezing.=—It has already been shown (p. 279) that the initial or fundamental cause of glacial motion must be operative at the heads of glaciers where the temperature is lowest and the material most loosely granular. In this condition, there is reason to believe that motion takes place between the grains, rather than by their distortion through the displacement of their laminæ. The fact that the granular structure is not destroyed, as it would be by the indefinite sliding of the crystal plates over each other, sustains this view. The inference is that the gliding planes play a notable rôle in glacial movement only in the basal parts of the lower ends of glaciers, where the greatest thrusts are developed, and where the granules have become largest and most completely interlocked. At the heads of glaciers, where motion is initiated, there may be great downward pressure, but not vigorous thrusts from behind, and probably only moderate thrusts developed within the body itself. There seems therefore no escape from the conclusion that the primal cause of glacial motion is one which may operate even under the relatively low temperatures, the relatively dry conditions, and the relatively granular textures which affect the heads of glaciers. These considerations lead to the view that movement takes place by the minute individual movements of the grains upon one another. While they are in the spheroidal form, as in the névé, this would not seem to be at all difficult. They may rotate and slide over each other as the weight of the snow increases; but as they become interlocked by growth, both rotation and sliding must apparently encounter more resistance. The amount of rotary motion required of an individual granule is, however, surprisingly small, and the meltings and refreezings incident to shifting pressures and tensions, and to the growth of the granules, seem adequate to meet the requirements. In order to account for a movement of three feet per day in a glacier six miles long, the mean motion of the average granule relative to its neighbor would be, roundly, ¹⁄₁₀₀₀₀ of its own diameter per day, or one diameter in 10,000 days; in other words, it would _change its relations to its neighbors_ to the extent of its diameter in about thirty years. A change of so great slowness under the conditions of granular alteration can scarcely be thought incredible, or even improbable, in spite of the interlocking which the granules may develop. The movement is supposed to be permitted chiefly by the temporary passage of minute portions of the granules into the fluid form at the points of greatest compression, the transfer of the moisture to adjoining points, and its resolidification. The points of greatest compression are obviously just those whose yielding most promotes motion, and a successive yielding of the points that come in succession to oppose motion most (and thus to receive the greatest stresses) permits continuous motion.