Abstract

Shell midden sites are a common feature in the archaeological record of coastal areas and analysis of shell has played an important role in archaeological problems such as the evolution of subsistence systems. However, spurious conclusions may be reached when shell sampling and quantification give inadequate consideration to the structure of shell deposition. In particular, there is a tendency for differences in species representation between stratigraphically arranged samples to be conceptualized as trends of change through time, without regard for alternative explanations of variability such as sampling effects and horizontal inhomogeneity, or postdepostional effects. At the Duwamish site, 4SKi23, Seattle, Washington, shell lenses were collected as discrete features to provide analytic control in interpreting the variability of shell samples from larger depositional units. The results show that individuel shell lenses are environment-specific collections, and that two contemporaneous samples may differ as greatly as two stratigraphically distinct samples. This has implications for the type of sampling necessary to achieve a representative species composition for a given component.