theory, however, there is no reason why the greatest depth should occur in one part rather than in another, while separate basins in the variously diverging arms of one lake seem most improbable. The lakes of Thun and Brienz form two basins of what was evidently once a single lake. The upper or Brienz basin is enormously deep, over two thousand feet, and the reason is obvious. The combined glaciers of the Lauterbrunnen and Grindelwald Valleys entered the main valley in a direction almost opposite to that of the Aare, piling up the ice against the great barrier of the Rieder Grat, so that it at length flowed downward with greatly increased grinding power; while lower down, toward Thun, the valley opens widely and would thus allow the ice to spread out with greatly diminished thickness. In our own country Loch Lomond and Ullswater have been found to consist of several distinct basins, and in none of our lakes have any indications of submerged river channels yet been found.
The third point of difference between lakes of erosion and those of submersion is the most important and the most distinctive, and furnishes, I think, what may be termed a diagnostic character of lakes of erosion. In most river valleys through a hilly or mountainous country outside of the glaciated districts, the tributary streams entering more or less at right angles to the main valley are seen to occupy small valleys of their own, which usually open out for a short distance at the same level before joining the main valley. Of course, there are also torrents which rush down steep mountain slopes directly to the main river, but even these have usually cut ravines more or less deeply into the rock. Now, if in such a valley we could mark out a contour line two hundred, three hundred, or five hundred feet above the level of the main stream, we should see that line continually turning up each side valley or ravine till it reached the given level at which to cross the tributary stream, and then turning back to the main valley. The contour line would thus form a series of notches or loops of greater or less depth at every tributary stream with its entering valley or deeply cut ravine, and if the main valley were filled with water this line would mark out the margin of the lake. As an illustration of this feature we may take the southwest coast of England, which has never been glaciated, but which has undergone a slight recent subsidence, as indicated by the submerged forests which occur at several places. The result of this submergence is that the lower parts of its larger river valleys have been converted into inland tidal lakes, such as Poole Harbor, Dartmouth Harbor, Kingsbridge River, Plymouth and Devonport Harbors, and Carrick Road above Falmouth. The Dart River is an excellent example of such a submerged valley, and its outline at high-water mark is shown at (3) on the accom-
