fragment while it remained in the same position. The most striking evidence, in my opinion, that at times the whole mass of the glacier actually freezes, is drawn from the fact, already alluded to, that, while the surface of the glacier loses annually from nine to ten feet of its thickness by evaporation and melting, it swells, on the other hand, in the spring, to the amount of about five feet. Such a dilatation can hardly be the result of pressure and the packing of the snow and ice, since the difference in the bulk of the ice brought down, during one year, from a point above to that under observation, would not account for the swelling. It is more readily explained by the freezing of the water of infiltration during spring and early summer, when the infiltration is most copious and the winter cold has been accumulating for the longest time. This view of the case is sustained by Elie de Beaumont, who states his opinion upon this point as follows:-- "Pendant l'hiver, la temperature de la surface du glacier s'abaisse a un grand nombre de degres au-dessous de zero, et cette basse temperature penetre, quoique avec un affaiblissement graduel, dans l'interieur de la masse. Le glacier se fendille par l'effet de la contraction resultant de ce refroidissement. Les fentes restent d'abord vides, et concourent an refroidissement des glaciers en favorisant l'introduction de l'air froid exterieur; mais an printemps, lorsque les rayons du soleil echaffent la surface de la neige qui couvre le glacier, ils la remenent d'abord a zero, et ils produisent ensuite de l'eau a zero qui tombe dans le glacier refroidi et fendille. Cette eau s'y congele a l'instant, en laissant degager de la chaleur qui tend a ramener le glacier a zero; et la phenomene se continue jusqu'a ce que la masse entiere du glacier refroidi soit ramene a la temperature de zero."[H]