Sometimes, when a number of cold summers have succeeded each other, preventing the glacier from melting in proportion to its advance, the accumulation of materials at its terminus becomes very considerable; and when, in consequence of a succession of warm summers, it gradually melts and retreats from the line it has been occupying, a large semicircular wall is left, spanning the valley from side to side, through which the stream issuing from the glacier may be seen cutting its way. It is important to notice that such terminal moraines may actually span the whole width of a valley, from side to side, and be interrupted only where watercourses of sufficient power break through them. To suppose that such transverse walls of loose materials could be thrown across a valley by a river were to suppose that it could build dams across its bed while it is flowing. Such transverse or crescent-shaped moraines are everywhere the work of glaciers. All these moraines are the land-marks, so to speak, by which we trace the height and extent, as well as the progress and retreat, of glaciers in former times. Suppose, for instance, that a glacier were to disappear entirely. For ages it has been a gigantic ice-raft, receiving all sorts of materials on its surface as it travelled onward, and bearing them along with it; while the hard particles of rock set in its lower surface have been polishing and fashioning the whole surface over which it extended. As it now melts, it drops its various burdens on the ground; boulders are the mile-stones marking the different stages of its journey, the terminal and lateral moraines are the framework which it erected around itself as it moved forward, and which define its boundaries centuries after it has vanished, while the scratches and furrows it has left on the surface below show the direction of its motion. All the materials which reach the bottom of the glacier, and are moving under its weight, so far as they are not firmly set in the ice must be pressed against one another, as well as against the rocky bottom, and will be rounded off, polished, and scratched, like the rock itself over which they pass. The pebbles or stones set fast in the ice will be thus polished and scratched, however, only over the surface exposed; but, as they may sometimes move in their socket, like a loosely mounted stone, the different surfaces may in turn undergo this process, and in the end all the loose materials under a glacier become more or less polished, scratched, and grooved. These marks exhibit also the peculiarity so characteristic of the grooves and scratches on the bed and walls of the valley: they are rectilinear, trending in the direction in which the superincumbent mass advances, though, of course, owing to the changes in the position of the pebbles or boulders, they may cross each other in every direction on their surface.