Indeed it is obvious that a glacier many hundred, or in some cases several thousand, feet in thickness, must exercise great pressure on the bed over which it travels. We see this from the striae and grooves on the solid rocks, and the fine mud which is carried down by glacial streams. The deposit of glacial rivers, the "loess" of the Rhine itself, is mainly the result of this ice-waste, and that is why it is so fine, so impalpable. That glaciers do deepen their beds seems therefore unquestionable. Moreover, though the depth of some of these lakes is great, the true slope is very slight. Tyndall and Ramsay do not deny that the original direction of valleys, and consequently of lakes, is due to cosmical causes and geological structure, while even those who have most strenuously opposed the theory which attributes lakes to glacial erosion do not altogether deny the action of glaciers. Favre himself admits that "it is impossible to deny that valleys, after their formation, have been swept out and perhaps enlarged by rivers and glaciers." Even Ruskin admits "that a glacier may be considered as a vast instrument of friction, a white sand-paper applied slowly but irresistibly to all the roughness of the hill which it covers." It is obvious that sand-paper applied "irresistibly" and long enough, must gradually wear away and lower the surface. I cannot therefore resist the conclusion that glaciers have taken an important part in the formation of lakes. The question has sometimes been discussed as if the point at issue were whether rivers or glaciers were the most effective as excavators. But this is not so. Those who believe that lakes are in many cases due to glaciers might yet admit that rivers have greater power of erosion. There is, however, an essential difference in the mode of action. Rivers tend to regularise their beds; they drain, rather than form lakes. Their tendency is to cut through any projections so that finally their course assumes some such curve as that below, from the source (_a_) to its entrance into the sea (_b_). [Illustration: Fig. 46.--Final Slope of a River.] Glaciers, however, have in addition a scooping power, so that if similarly _a d b_ in Fig. 47 represent the course of a glacier, starting at _a_ and gradually thinning out to _e_, it may scoop out the rock to a certain extent at _d_; in that case if it subsequently retires say to _c_, there would be a lake lying in the basin thus formed between _c_ and _e_. [Illustration: Fig. 47.] On the other hand I am not disposed to attribute the Swiss lakes altogether to the action of glaciers. In the first place it does not seem clear that they occupy true rock basins. On this point more evidence is required. That some lakes are due to unequal changes of level will hardly be denied. No one, for instance, as Bonney justly observes,[55] would attribute the Dead Sea to glacial erosion.