This paper considers how to make the most out of the rather imprecise chronological knowledge that we often have about the past. We focus here on the relative dating of artefacts during archaeological fieldwork, with particular emphasis on new ways to express and analyse chronological uncertainty. A probabilistic method for assigning artefacts to particular chronological periods is advocated and implemented for a large pottery data set from an intensive survey of the Greek island of Antikythera. We also highlight several statistical methods for exploring how uncertainty is shared amongst different periods in this data set and how these observed associations can prompt more sensitive interpretations of landscape-scale patterns. The concluding discussion re-emphasizes why these issues are relevant to wider methodological debates in archaeological field practice.