this process goes on with varying activity until the sheet of ice is so thick that it becomes itself a shelter to the water below, and protects it, to a certain degree, from the cold without.  Thus a given thickness of ice may cause a suspension of the freezing process, and the first ice-stratum may even be partially thawed before the cold is renewed with such intensity as to continue the thickening of the ice-sheet by the addition of fresh layers.  The strata or beds of ice increase gradually in this manner, their separation being rendered still more distinct by the accumulation of air-bubbles, which, during a hot and clear day, may rise from a muddy bottom in great numbers.  In consequence of these occasional collections of air-bubbles, the layers differ, not only in density and closeness, but also in color, the more compact strata being blue and transparent, while those containing a greater quantity of air-bubbles are opaque and whitish, like water beaten to froth.

A cake of pond-ice, such as is daily left in summer at our doors, if held against the light and turned in different directions, will exhibit all these phenomena very distinctly, and we may learn still more of its structure by watching its gradual melting.  The process of decomposition is as different in fresh-water ice and in land-or glacier-ice and that of their formation.  Pond-ice, in contact with warm air, melts uniformly over its whole surface, the mass being thus gradually reduces from the exterior till it vanishes completely.  If the process be slow, the temperature of the air-bubbles contained in it may be so raised as to form the vertical funnels or tubes alluded to above.  By the anastomosing of these funnels, the whole mass may be reduced to a collection of angular pyramids, more or less closely united by cross-beams of ice, and it finally falls to pieces when the spaces in the interior have become for numerous as to render it completely cavernous.  Such a breaking-up of ice is always caused by the enlargement of the open spaces produces by the elevated temperature of the air-bubbles, these spaces being necessarily more or less parallel with one another, and vertical in their position, owing to the natural tendency of the air-bubbles to work their way upward till they reach the surface, where they escape.  A sheet of ice, of this kind, floating upon water, dissolves in the same manner, melting wholly from the surface, if the process be sufficiently rapid, or falling to pieces, if the air-bubbles are gradually raised in their temperature sufficiently to render the whole mass cavernous and incoherent.  If we now compare these facts with what is known of the structure of land-ice, we shall see that the mode of formation in the two cases differs essentially.