But the main work of carving out the hills and mountains of the world is done by streams, rivers, and glaciers; and so we now pass on to consider how they perform their tasks. Water by itself, even when flowing fast, would be powerless to carve gorges and valleys in the solid rock; but the stones which torrents and streams carry along give them a marvellous grinding power, for with such material a stream continually wears away its rocky bed. Moreover, the stones themselves are all the while being rubbed down by each other, until finally they are ground down to fine sand and mud, which help in the work of erosion. Every mountain stream or torrent runs in a ravine or valley of some sort; and any traveller who will take the trouble to watch what goes on there may easily convince himself that the ravine, gorge, or valley has been carved out by the stream, aided by the atmospheric influences to which we have already alluded. But perhaps some may be inclined to look upon the ravine as a chasm produced by some violent disturbance from below, whereby the rocks were rent asunder, and that the stream somehow found its way into the rent. A little inquiry will dispel this idea. In the first place, such catastrophes are quite unknown at the present day; and as we have more than once pointed out, the geologist's method is to apply a knowledge of processes now in operation to the phenomena of the rocks, in order to read their history. Secondly, no conclusion can be accepted which is not supported strongly by evidence. If such a rending of the rocks had taken place, there would assuredly be some evidence of the fact. We should expect to find a great crack running all along the bed of the stream; but of this there is no sign. Go down in any weather when the stream is low, and look at the rocks over which it flows, and you will search in vain for such evidence. Instead of being broken, the rocks extend continually across. You would also expect to find the strata "dipping," or sloping away from the stream on each side, if they had been rent by such an upheaval; but here again we are met by a total want of evidence. Thirdly, a crack might be expected to run along more or less evenly in one direction. But look at the ravine, follow it up for some miles, and you will see that it winds along in a very devious course, not in a straight line.