Glacier sliding has major environmental consequences, but friction caused by debris in the basal ice of glaciers is seldom considered in sliding models. To include such friction, divergent hypotheses for clast‐bed contact forces require testing. In experiments we rotate an ice ring (outside diameter = 0.9 m), with and without isolated till clasts, over a smooth rock bed. Ice is kept at its pressure‐melting temperature, and meltwater drains along a film at the bed to atmospheric pressure at its edges. The ice pressure or bed‐normal component of ice velocity is controlled, while bed shear stress is measured. Results with debris‐free ice indicate friction coefficients < 0.01. Shear stresses caused by clasts in ice are independent of ice pressure. This independence indicates that with increases in ice pressure the water pressure in cavities observed beneath clasts increases commensurately to allow drainage of cavities into the melt film, leaving clast‐bed contact forces unaffected. Shear stresses, instead, are proportional to bed‐normal ice velocity. Cavities and the absence of regelation ice indicate that, unlike model formulations, regelation past clasts does not control contact forces. Alternatively, heat from the bed melts ice above clasts, creating pressure gradients in adjacent meltwater films that cause contact forces to depend on bed‐normal ice velocity. This model can account for observations if rock friction predicated on Hertzian clast‐bed contacts is assumed. Including debris‐bed friction in glacier sliding models will require coupling the ice velocity field near the bed to contact forces rather than imposing a pressure‐based friction rule.