Within this mass, air is intercepted and held inclosed between the particles of ice. The process by which snow-flakes or snow-crystals are transformed into grains of ice, more or less compact, is easily understood. It is the result of a partial thawing, under a temperature maintained very nearly at thirty-two degrees, falling sometimes a little below, and then rising a little above the freezing-point, and thus producing constant alternations of freezing and thawing in the same mass of snow. This process amounts to a kind of kneading of the snow, and when combined with the cohesion among the particles more closely held together in one snow-flake, it produces granular ice. Of course, the change takes place gradually, and is unequal in its progress at different depths in the same bed of recently fallen snow. It depends greatly on the amount of moisture infiltrating the mass, whether derived from the melting of its own surface, or from the accumulation of dew or the falling of rain or mist upon it. The amount of water retained within the mass will also be greatly affected by the bottom on which it rests and by the state of the atmosphere. Under a certain temperature, the snow may only be glazed at the surface by the formation of a thin, icy crust, an outer membrane, as it were, protecting the mass below from a deeper transformation into ice; or it may be rapidly soaked throughout its whole bulk, the snow being thus changed into a kind of soft pulp, what we commonly call slosh, which, upon freezing, becomes at once compact ice; or, the water sinking rapidly, the lower layers only may be soaked, while the upper portion remains comparatively dry. But, under all these various circumstances, frost will transform the crystalline snow into more or less compact ice, the mass of which will be composed of an infinite number of aggregated snow-particles, very unequal in regularity of outline, and cemented by ice of another kind, derived from the freezing of the infiltrated moisture, the whole being interspersed with air. Let the temperature rise, and such a mass, rigid before, will resolve itself again into disconnected ice-particles, like grains more or less rounded. The process may be repeated till the whole mass is transformed into very compact, almost uniformly transparent and blue ice, broken only by the intervening air-bubbles. Such a mass of ice, when exposed to a temperature sufficiently high to dissolve it, does not melt from the surface and disappear by a gradual diminution of its bulk, like pond-ice, but crumbles into its original granular fragments, each one of which melts separately. This accounts for the sudden disappearance icebergs, which, instead of slowly dissolving into the ocean, are often seen to fall to pieces and vanish at once.