The size and shape of equilibrium channels in uniform, noncohesive sands, 0.67 mm and 2.0 mm in diameter, were studied experimentally in a laboratory flume 52 feet long in which discharge, slope, sediment load, and bed and bank material could be varied independently. For each run a straight trapezoidal channel was molded in the sand and the flume set at a predetermined slope. Introduction of the discharge was accompanied by widening and aggradation until a stable channel was established. By definition a stable equilibrium existed when channel width, water surface slope, and rate of transport became constant. The duration of individual runs ranged from 2 to 52 hours depending upon the time required for establishing equilibrium.

Stability of the banks determined channel shape. In the 2.0 mm sand at a given slope and discharge, only one depth was stable. At this depth the flow was just competent to move particles along the bed of the channel. An increase in discharge produced a wider channel of the same depth and thus transport per unit width remained at a minimum. Channels in the 0.67 mm sand were somewhat more stable and permitted a 1.5 fold increase in depth above that required to start movement of the bed material. An increased transport was associated with the increase in depth. The rate of transport is adequately described in terms of the total shear or in terms of the difference between the total shear and the critical shear required to begin movement.

In these experiments the finer, or 0.67 mm, sand, began to move along the bed of the channel at a constant shear stress. Incipient movement of the coarser, or 2.0 mm, sand, varied with the shear stress as well as the mean velocity. At the initiation of movement a lower shear was associated with a higher velocity and vice versa.

Anabranches of braided rivers and some natural river channels formed in relatively noncohesive materials resemble the essential characteristics of the flume channels. For a given slope and size of bed material the discharge per unit width in the laboratory channels was similar to that computed for anabranches and river channels measured in the field. Unlike most natural channels, despite impressive bank erosion, the channels in the laboratory only meandered at supercritical flows associated with very steep slopes. These conditions involving shallow depths, high velocity, and steep slopes are uncommon in most natural rivers.