A method of analyzing the extent to which two-dimensionally located points are randomly distributed; a measure of the relationship between a cluster of points in a pattern based on the expected value and the observed value. The statistic equals observed value divided by expected value. This method of analyzing the degree of dispersion in a distribution pattern was first developed by plant ecologists studying the concentration of certain species. A nearest-neighbor index (usually denoted by the symbol R), is calculated from the ratio of the average observed distance from each point in the pattern to its nearest neighbor, to the average distance expected if the pattern were randomly distributed, which depends solely on the density of the pattern being studied. The index R varies from 0.00 for a totally clustered pattern through 1.00 for a random distribution to a maximum of 2.15 for a completely regularly spaced pattern. The index is influenced by the size of the study area chosen; it is therefore essential to select a relevant framework for the distribution being studied. With any boundary, however, it is possible for the index to be distorted by the 'boundary effect' to give a figure closer to the maximum than would be justified; this arises because the nearest neighbors of points near to the boundary may in fact lie beyond the boundary and hence not be properly counted, thus increasing the figure for the observed mean distance. It is also essential that the points in the pattern being analyzed are of the same date and similar function, and that the pattern should be complete. The index R describes only a part of the total pattern and can serve as a useful basis for asking more detailed questions about the factors that underlie the observed pattern. The technique has been useful to archaeologists studying the distribution of sites over a landscape and their relation to each other.