My own experience of these associated air and water cells is derived almost exclusively from lake-ice, in which I have often observed them in considerable numbers. In examining whether the liquid contents had ever been frozen or not, I was guided by the following considerations. If the air be that originally entangled in the solid, it will have the ordinary atmospheric density at least; but if it be due to the melting of the walls of the cell, then the water so formed being only eight-ninths of that of the ice which produced it, _the air of the bubble must be rarefied_. I suppose I have made a hundred different experiments upon these bubbles to determine whether the air was rarefied or not, and in every case found it so. Ice containing the bubbles was immersed in warm water, and always, when the rigid envelope surrounding a bubble was melted away, the air suddenly collapsed to a fraction of its original dimensions. I think I may safely affirm that, in some cases, the collapse reduced the bubbles to the thousandth part of their original volume. From these experiments I should undoubtedly infer, that in lake-ice at least, the liquid of the cells is produced by the melting of the ice surrounding the bubbles of air. But I have not subjected the bubbles of glacier-ice to the same searching examination. I have tried whether the insertion of a pin would produce the collapse of the bubbles, but it did not appear to do so. I also made a few experiments at Rosenlaui, with warm water, but the result was not satisfactory. That ice melts internally at the surfaces of the bubbles is, I think, rendered certain by my experiments, but whether the water-cells of glacier-ice are entirely due to such melting, subsequent observers will no doubt determine. [Sidenote: "LIQUID LIBERTY."] I have found these composite bubbles at all parts of glaciers; in the ice of the moraines, over which a protective covering had been thrown; in the ice of sand-cones, after the removal of the superincumbent debris; also in ice taken from the roofs of caverns formed in the glacier, and which the direct sunlight could hardly by any possibility attain. That ice should liquefy at the surface of a cavity is, I think, in conformity with all we know concerning the physical nature of heat. Regarding it as a motion of the particles, it is easy to see that this motion is less restrained at the surface of a cavity than in the solid itself, where the oscillation of each atom is controlled by the particles which surround it; hence _liquid liberty_, if I may use the term, is first attained at the surface. Indeed I have proved by experiment that ice may be melted internally by heat which has been conducted through its external portions without melting them. These facts are the exact complements of those of "regelation;" for here, two moist surfaces of ice being brought into close contact, their liquid liberty is destroyed and the surfaces freeze together. THE MOULINS. (25.)