smaller may lose while the larger gain. [Illustration: +Figs.+ 286–91.—Snowflakes. (Photographed by W. A. Bentley.)] Another factor that enters into the process is that of pressure and tension. The granules are compressed at the points of contact and put under tension at points not in contact, and the pressure and tension are, on the average, likely to be relatively greatest for the smallest granules. Tension increases the tendency to evaporation and adds its effects to curvature, and the capillary spaces adjoining the points of contact probably favor condensation. Ice expands in crystallizing and pressure reduces the melting-point, while tension raises it. The effect of this is slight (p. 276), and it probably plays little part in glacial action, but it is to be correlated with the much more important fact that _compression produces heat_ which may raise the temperature of the ice to the melting-point, while tension may reduce the temperature to or below freezing. There is therefore a tendency for the ice to melt at the points of contact and compression, and for the water so produced to refreeze at adjacent points where the surface is under tension. This process becomes effective beneath a considerable body of snow, and here the granules gradually lose the spheroidal form assumed in the early stages of granulation and become irregular polyhedrons interlocked into a more or less solid mass. A third factor is also to be recognized, though its effectiveness is unknown. Under severe wind pressure, air penetrates porous bodies with appreciable facility. The “breathing” of soils and the curious phenomena of “blowing-wells” and “blowing-caves” teach us of the effective penetration and extrusion of the air under variations of barometric pressure. In the snow-fields, and in the more granular portions of glaciers near their heads, the porosity is doubtless sufficient to allow of the appreciable penetration of the atmosphere. During a part of the time, the probable effect is the condensation within the ice of moisture from the air, and during another part, evaporation from the ice. These alternating processes are attended by oscillations of temperature. While the balance between loss and gain of substance may be immaterial, the oscillating nature of the process and the fluctuations of temperature are probably favorable to granular change.