Studer's observation that glacier-erosion was distinguished by the production of convex forms, I proceeded to show that such forms naturally resulted from glacier working upon surfaces which had been antecedently broken up by diverse actions; and pointed out that when glacier-action was long continued, the obliteration of all angular surfaces, and of almost all curves, was inevitable. I concluded, therefore (and am prepared to accept all the responsibility which attaches to the conclusion), that the convexity of roches moutonnées was to be regarded as a proof that no great amount of glacier-erosion had occurred; that rock-surfaces with a small degree of convexity, which had obviously been glaciated, indicated a greater erosion; and that the degree of flatness bore a direct relation to the amount of power which had been employed. And further, that when unworn, angular rock-surfaces were found in the immediate vicinity of glaciated rocks, they were to be regarded as additional and confirmatory evidence that the depth of matter taken away by the glacier could not have been important, unless it could be shown that the angularity was due to subsequent operations.
Applying these conclusions to the case of the Valley of Aosta, we find—1. That as recent denudation has been unequal, throughout the valley, to obliterate polish and fine striations on the rocks, we are unable to believe that the vast number of angular surfaces which are found in contiguity to the abraded ones can possibly have been produced subsequently to the retreat of the glacier. 2. Their existence in connection with innumerable convex glaciated surfaces, throughout the valley, is irrefutable evidence that the valley was not excavated by glaciers. 3. The comparative scarcity of roches nivelées, combined with the other evidence, affords a strong presumption that the so-called excavation has not amounted, throughout the valley, to more than a very few feet of depth.
Hitherto, I have chiefly appealed to the bed (or floor) of the valley. Almost equally stubborn facts are obtainable from the slopes of its bounding mountains. If the valley had been excavated by glaciers, very emphatic traces would have been left behind every-
