Rock-on-rock wear under known normal loads was measured experimentally using 62 mm diameter rotating discs machined from a range of silicic rock types: one granite, two metamorphic rocks, and three sandstones of porosities varying between 7% and 28%. Wear debris was removed continuously from the contacting surfaces by water. We obtained a wear law of the form dT/dx = Aσn$phi$m, where σ is contact normal stress (MPa), dT/dx is a dimensionless wear rate (meter wear (T) per meter shear displacement (x)), $phi$ is porosity, and A, n, and m are empirical constants, 10-8.11, 8.33, and 4.50, respectively. Low-porosity rocks displayed surface polishing, and abraded particle size was substantially smaller than the starting grain size. Whole grains were plucked from high-porosity rocks, without any surface polishing, so that the wear product particle size remained the same as that of the starting material. The empirical wear law describes abrasion beneath a hard-bedded temperate glacier, provided we can estimate both the in situ normal stress between entrained clasts and the bed and the horizontal velocity of the clasts in the flowing ice. Normal stress concentration is expected to develop at clast/bed contacts from drag against ice flowing toward the bed, where pressure melting occurs. These experimental data are consistent with wear rates inferred from silt outflow in subglacial streams, provided a tenfold stress concentration can occur relative to the ice overburden pressure. The rate of production of fine-grained wear products under subglacial conditions is substantially less than by tectonic faulting.