Let a stream of pure water having a depth D, and with boulders on its bed, have a velocity V just sufficient to move them very slowly. Any larger boulders would not be moved. Any smaller boulders would move more quickly. Similarly, fine sand would be rolled more quickly than coarse sand. If the velocity of the stream increases, larger boulders would be moved. Streams are thus constantly sorting out the materials which they roll. If the bed is examined it will be found that large boulders exist only down to a certain point, smaller boulders, shingle, gravel, coarse sand and fine sand following in succession. If the water, instead of being pure, is supposed to contain silt, this may affect its velocity--it is not, however, known to do so--but, given a certain velocity, it is not likely that the rolling power of the stream is much affected by its containing silt. It is sometimes supposed that increased depth gives increased rolling power, because of the increased pressure, but this is not so. The increased pressure due to depth acts on both the upstream and downstream sides of a body. It is moved only by the pressure due to the velocity. When sand is rolled along the bed of a stream there is usually a succession of abrupt falls in the bed. After each fall there is a long gentle upward slope till the next fall is reached. The sand is rolled up the long slope and falls over the steep one. It soon becomes buried. The positions of the falls of course keep moving downstream. The height of a fall in a large channel is perhaps 6 inches or 1 foot, and the distance between the falls 20 or 30 feet. A fall does not usually extend straight across the bed but zigzags. It has sometimes been said that the inclination of the bed of a stream, when high, facilitates scour, the material rolling more easily down a steep inclined plane. The inclination is nearly always too small to have any appreciable direct effect. The inclination of the surface of the stream of course affects its velocity, and this is the chief factor in the case. A sudden rise in the bed of a stream does not necessarily cause rolled materials to accumulate there, except perhaps to the extent necessary to form a gentle slope. Frequently even this slope is not formed, especially if the rolled material is only sand. The eddies stir it up and it is carried on. The above remarks apply also to weirs or other local rises in the bed.