Problematizing Bayesian approaches to prehistoric chronologies

In recent years Bayesian exploration of radiocarbon datasets has been employed widely in prehistoric archaeology. Pertinent especially to major biogeographic and behavioural changes such as human dispersals and extinctions, the spread of agriculture and culture change, the method can offer a powerful means to improve considerably the precision of prehistorians' investigation of some of the most major questions in human prehistory. As such its potential is profound - it has even been regarded as the third radiocarbon revolution - but its appropriateness is dependent on the assumptions that must be made of the samples selected for dating. How sound are these assumptions, and therefore how reliable are Bayesian analyses? Here, we introduce some aspects and assumptions that underline Bayesian modelling of radiocarbon measurements, and we problematize their application in Palaeolithic archaeology. We conclude that many existing models are faulty, and suggest some criteria for quality control in this field.