We have reduced published petrographic data on 90 surface soils from Apollo sample return sites in terms of olivine, pyroxene, plagioclase, and opaques (mostly ilmenite) for a best estimate. To do so, we have recast the best estimates of modal mineralogy of mare and highland lithic fragments to add to the monomineralic population of soils. We have assumed that agglutinates and regolith breccias mimic those of local soils. We have thus arrived at a working summary of mineral distribution in areas traversed by Apollo astronauts. In addition, we have proportioned these averages to geologic units in the premission maps to estimate the average mineral composition of areas up to 30 km¿30 km around the landing sites. The calibration and testing of remote sensing signals for planetary exploration depend largely on the ground truth provided by lunar samples. Chemical compositions and the modal distribution of petrologic constituents of most lunar soils are known. These can be grouped spatially to represent areas that coincide with pixels of remote sensing observations and measurements. Signals from certain petrographic constituents such as lithic fragments, breccias, and agglutinates, however, are understandably ambiguous. Petrologic inference from remote sensing signals would be facilitated if petrographic data from actual samples were to be reduced and expressed as sums of single phases occurring in each pixel of observation. ¿ 2000 American Geophysical Union