The use of summed radiocarbon probability distributions in archaeology: a review of methods

• Published in: Journal of archaeological science Vol. 39; no. 3; pp. 578 - 589
• Main Author: Williams, Alan N.
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01.03.2012; Elsevier
• Subjects: Australia and New-Guinea; Calibration curve effects; Dating; Excavation and methods; Laboratory methods; Methodology and general studies; Oceania; Prehistory and protohistory; Radiocarbon data; Sample size; Summed probability; Taphonomic bias; Australia; Oceania; Calibration curve effects; Summed probability; Sample size; Taphonomic bias; Radiocarbon data; Methodology; Sample; Archaeometry; Calibration; Statistics; Radiocarbon; Database; Dating; Taphonomy
• ISSN: 0305-4403; 1095-9238
• DOI: 10.1016/j.jas.2011.07.014

Using a database of Australian archaeological radiocarbon dates (n = 2996), this paper explores three key methodological issues associated with the use of summed probability plots of radiocarbon data: 1) the minimum sample size needed for a statistically reliable plot; 2) the effect of radiocarbon calibration on the structure of these plots; and 3) the application of a taphonomic correction to such time-series data. The results identify several protocols, recommended as best-practice when using summed probability plots: 1) a minimum sample size of 500 radiocarbon dates should be used, and the sample size and the mean of the standard deviations of the radiocarbon dates (ΔT) in the sample should both be reported; 2) a moving average trendline of 500–800 years should be used to offset the effects of the calibration process; and 3) Surovell et al. “Correcting temporal frequency distributions for taphonomic bias” [Journal of Archaeological Science 36 (2009) 1715–1724] is explored, with modifications and temporal limits (<25,000 cal years BP) proposed. Correction of time-series data using theoretical taphonomic correction curves is useful as a heuristic tool but can obscure real trends if applied routinely. Comparison between summed probability plots and other occupation data is presented and shows good correlation. However it is recommended that plots are supplemented by other archaeological indices and the cross-comparison of such multiple proxies will strengthen identification of occupation trends. ► Methodological issues with the use of sum probability plots are explored. ► The paper uses a dataset of 2996 14C dates from >800 sites across Australia. ► Analysis shows sample size, calibration and taphonomic loss all have an effect. ► Protocols for future production of sum probability plots are proposed. ► Comparison of plots and other archaeological indices show good correlation.