Confessions of a serial polygamist: the reality of radiocarbon reproducibility in archaeological samples

Abstract

Since 1993 Historic England (and its predecessor English Heritage) has commissioned 9,074 radiocarbon measurements on archaeological samples. Over 80% of these have been interpreted within formal Bayesian statistical models. The multiple strands of reinforcing evidence incorporated in these models provide precise chronologies that make stringent demands on the accuracy of the radiocarbon results included in the analysis. Inter-laboratory replication is consequently a routine part of model construction and validation. We report an analysis of replicate measurements on 1,089 archaeological samples. It is clear that laboratory reproducibility accounts for only part of the observed variation. The type of material dated is also critical to the reproducibility of measurements, with some sample types proving particularly problematic.