Comparing archaeological settlement systems with rank-size graphs: a measure of shape and statistical confidence

Abstract

Rank-size analysis of settlement systems in archaeology has focused strongly on departures from log-normality, but it is not clear that a log-normal pattern should ordinarily be expected. Direct comparison of observed rank-size curves to each other is of greater utility in identifying chronological change and inter-regional variation in settlement dynamics as reflected in rank-size patterns. A coefficient of shape is proposed to facilitate comparison and to serve as a basis for bootstrap assessment of statistical confidence in any differences that are observed. The approach is illustrated with analysis of archaeological settlement data from China, Iran, Peru, and Mexico. These analyses provide insight into how many settlements must be observed to achieve high confidence and into the impact of curve shape on confidence.