Evaluating measurement uncertainty in amino acid racemization analysis: Towards a new chronology

Unlike other Quaternary dating methods, amino acid racemization (AAR) geochronology has the potential to provide age estimates that span the entire Quaternary period, a crucial period for understanding past climate change and human evolution. It has become a critical technique for Quaternary Science and uses the time/temperature dependent kinetics of protein decomposition to provide relative age estimates of fossil samples. The accuracy of age estimates relies heavily on the accuracy of analytical data and accurate determinations of uncertainty estimates.
This thesis takes internationally established principles of measurement uncertainty determination and applies them to AAR. Analytical uncertainty is considered in the context of intra- and inter-laboratory measurement results. A retrospective evaluation of intra-laboratory precision using ANOVA is given, and results from an inter-laboratory proficiency study, evaluated as estimates of bias, are summarised (paper submitted). The final sections look at uncertainty from existing archaeological site data, including sampling effects. A model is proposed that utilises decomposition correlations between amino acids to provide a priori uncertainty estimates. These are then used to update observed site data using a Bayesian approach to derive posterior uncertainty estimates and D/L values. A further model is tentatively presented which could potentially be used to derive quantitative age estimates once uncertainty within the kinetic and temperature models have been characterised and accounted for.