Three factors contribute to differences between archaeological samples: sampling error and functional and temporal differences between the parent populations from which the samples are drawn. Many archaeologists have explained such differences as due only to temporal variations between the parent populations, thereby ignoring sampling error and relegating functional specificity to a minor role.
A statistical analysis of potsherd frequencies in samples made by the Chicago Natural History Museum from an eastern Arizona pueblo, the Carter Ranch site, was undertaken using UNIVAC. By taking into account sampling error, differences between parent populations appeared that could most logically be attributed to functional differences. From the calculations, four possible functional constellations of pottery types were discovered, four types of rooms were isolated, a probable ceremonial complex of five pottery types was established, and differential areal disposal of sherds was found. It was impossible to demonstrate temporal differences between provenience units on the basis of sherd frequencies. Any such differences played a minor part in contributing to sample variation as compared to functional differences between provenience units.