Modelling chronologically ordered radiocarbon dates in R

Abstract

Studies with multiple radiocarbon dates often contain useful information on the relative locations of the dated levels. Such information can be used to obtain robust, integrated site chronologies, with at times more precise ages than those of the individual dates, where outliers can be identified and downweighted, and where the ages of any undated levels can also be estimated. Examples include trees with radiocarbon dates separated by exactly known amounts of yearly tree-rings, or sedimentary sites where ages further down the stratigraphy can be assumed to be older than ages further up. Here we present coffee, an R package for Bayesian models that apply chronological ordering for fossils and environmental events. Coffee runs natively within the popular and versatile R environment, with no need for importing or exporting data or code from other programs, and works with plain-text input files that are relatively easy to read and write. It thus provides a new, transparent and adaptable educational and research platform designed to make chronology building more accessible.