Assessing Systematic Bias In Museum Collections A Case Study Of Spindle Whorls

Archaeologists often overlook museum collections for research purposes due to concerns about the representativeness of the artifacts in the collection. Such concerns can be addressed by comparing non-scientifically produced collections to collections of known standing, such as those from more modern projects, using the same exploratory data analysis and basic statistical methods that archaeologists use to compare sets of artifacts in other situations. As a case study, this article compares spindle whorls recovered during the 2007 excavations at the Aztec site of Calixtlahuaca with two museum collections of whorls attributed to the site. A variety of metric and non-metric traits were recorded for each item in the three collections and were compared using Kolmogrov-Smirnov, Chi-square, and Fisher's exact tests. Both museum collections show a bias toward more elaborately decorated pieces, but with varying side effects on other attributes of analytical interest.