the rate of erosion. There can be no question, however, that the nature of the valley rock is of profound importance, some weak rocks being eroded with relative rapidity by small glaciers, other rocks resisting the erosion of even large, powerful glaciers. Two causes, the size of the glacier and the nature of the enclosing rock, are, in all probability, of most importance in the modification of the height of valleys left hanging by more rapid erosion along the main trough.
An argument which has been advanced against the power of glaciers to erode, is the fact that rock islands sometimes rise from the floor of valleys through which powerful glaciers have passed. It has been claimed that such protuberances should have been erased if the glaciers were really eroding greatly. When the operation of glaciers as agents of erosion is truly understood, however, this argument seems to favor rather than to oppose glacial erosion. It is not to be supposed that glaciers would erode everywhere at the same rate. There is naturally a variation in the rate of erosion of a valley bottom dependent upon at least two important influences—nature of rock and rapidity of ice currents—both of which are liable to vary in any valley and thus necessarily give rise to irregularities in the ice-eroded valley bottom. Once an obstacle arose in the path of a powerfully moving glacier, it would have the tendency to split the ice current around itself, much as a sand bar spilts the current of a river. By interfering with the ice current in line with the obstacle, and by causing a concentration of movement on either side of it, the size of the obstacle would naturally increase. Rock knolls, islands and nunataks (Fig. 14) are such characteristic features in glacially eroded valleys that, when the full significance of glacial erosion is understood, I believe they will be found to constitute one of-the distinctive evidences of glacial erosion, to be-classed with hanging valleys, truncated spurs, steepened slopes and U-shaped profiles.
In discussions on glacial erosion much attention has been paid to rock basins,—basins with rock rims in the bottoms of glaciated, and oftentimes holding lakes. Such basins also occur on the fiord floors of the Inside Passage. Irregularities in erosion, due to differences in rock resistance and in ice currents, readily account for these. As Andrews has shown in his remarkable papers on glacial erosion in the New Zealand fiords, one important cause for such basins, and other forms of vigorous erosion, is the convergence of ice currents in a valley of smaller cross section, causing acceleration of motion. Rock basins must be added to the land forms resulting from and hence indicative of profound glacial erosion.
Another feature at first apparently opposing glacial erosion is that hanging valleys, truncated spurs, and steepened slopes are at times well developed on one side of a main trough and either absent or poorly developed on the other. This, however, seems a perfectly normal result