A physically rational model for river width is critical to predict
macroscopic landscape evolution driven by fluvial sediment transport.
Growing evidence suggests that rivers widen until the stress exerted by
the fluid on the bed surface is close to the critical entrainment stress
of the bank material. In this study, we test the limits of this model as
a closure assumption in dynamically evolving river systems. We consider
a simple laboratory channel with a fixed water discharge, monodisperse
bed material, no sediment supply, and an initial relief that was
sufficiently large to guarantee a finite transport capacity. Over time,
the transport capacity approaches zero through changes in channel
morphology. Concurrent measurements of width and sediment load highlight
departures from theory that mirror empirical trends in bankfull alluvial
rivers. We suggest that lateral instability limits channel width at high
sediment loads.