We report the results of Ar adsorption experiments on a terrestrial labradorite and lunar rock 15415 crushed in vacuo. The experiments were designed to test lunar atmosphere simulation models for the behavior of Ar on the lunar surface, as determined from the Apollo 17 mass spectrometer results. These models (Hodges, 1980, 1982) used a single adsorption potential to characterize the surfaces of lunar soil grains, with the result that high (6-7 kcal/mol) heats of adsorption were inferred Our experimental results show that very high adsorption potentials are indeed associated with fresh mineral surfaces, but that these energetic surfaces occupy only small fractions of the total surface area. Nonetheless, these small fractions of surface, if they can be maintained in the lunar regolith in steady-state condition, could be sufficient to account for the Apollo 17 mass spectrometer observations.