But where direct observations are still so scanty, and the interpretations of the facts so conflicting, it is the part of wisdom to be circumspect in forming opinions. This much, however, I believe to be already settled: that any theory which ascribes the very complicated phenomena of the glacier to one cause must be defective and one-sided. It seems to me most probable, that, while pressure has the larger share in producing the onward movement of the glacier, as well as in the transformation of the snow into ice, a careful analysis of all the facts will show that this pressure is owing partly to the weight of the mass itself, partly to the pushing on of the accumulated snow from behind, partly to its sliding along the surface upon which it rests, partly to the weight of water pervading the whole, partly to the softening of the rigid ice by the infiltration of water, and partly, also, to the dilatation of the mass, requiting from the freezing of this water. These causes, of course, modify the ice itself, while they contribute to the motion. Further investigations are required to ascertain in what proportion these different influences contribute to the general result, and at what time and under what circumstances they modify most directly the motion of the glacier. That a glacier cannot be altogether compared to a river, although there is an unmistakable analogy between the flow of the one and the onward movement of the other, seems to me plain,--since the river, by the combination of its tributaries, goes on increasing in bulk in consequence of the incompressibility of water, while a glacier gradually thins out in consequence of the packing of its mass, however large and numerous may be its accessions. The analogy fails also in one important point, that of the acceleration of speed with the steepness of the slope. The motion of the glacier bears no such direct relation to the inclination of its bed. And though in a glacier, as in a river, the axis of swiftest motion is thrown alternately on one or the other side of the valley, according to its shape and slope, the very nature of ice makes it impossible that eddies should be formed in the glacier, and the impressive feature of whirlpools is altogether wanting in them. What have been called glacier-cascades bear only a remote resemblance to river-cascades, as in the former the surface only is thrown into confusion by breaking, without affecting the primitive structure;[I] and I reiterate my formerly expressed opinion that even the stratification of the upper regions is still recognizable at the lower end of the glacier of the Rhone.