Radiocarbon analysis of the early archaeological site of Nauwalabila I, Arnhem Land, Australia: implications for sample suitability and stratigraphic integrity

This study presents the results of an extensive radiocarbon dating program at the Nauwalabila I site in northern Australia. The results show that the radiocarbon chronology at Nauwalabila is reliable to similar to 130 cm depth, but below this depth coarse charcoal has been variably altered during a period in the early Holocene when an ephemeral groundwater table reached close to the ground surface of the time. Below similar to 150 cm none of the radiocarbon ages can be considered to indicate reliably the age of deposition of the sediments. Luminescence dates near the surface and at I 10 cm are concordant with the radiocarbon chronology in the upper part of the sequence, and hence the aberrant radiocarbon results below similar to 150 cm do not constitute a reason to doubt the accuracy of the luminescence chronology deeper in the stratigraphy. A conservative estimate of the age of the sequence, based on extrapolation of results from that portion of the sequence where the radiocarbon chronology is considered to be reliable, is consistent with the chronology proposed previously from luminescence dating. Both chronologies therefore suggest occupation of the site before 50,000 years. Based on sediment characteristics and the distribution of quartz, chert, quartzite and 'quartz crystal' artefacts, there is no evidence that there has been significant vertical displacement of artefacts relative to the surrounding sand matrix. Both chemical alteration and physical translocation of charcoal contributed to the aberrant ages at depth in the deposit. The results point to the need for careful assessment of the suitability of charcoal for radiocarbon dating prior to analysis and to the dangers of relying on a small number of radiocarbon dates in the development robust site chronologies. Strategies for screening samples for suitability include (i) microscopic examination, (ii) not analysing samples unless they survive the full ABOX pretreatment, (iii) not analysing samples unless the material is significantly larger than the sediment matrix, (iv) using CHN analysis on both untreated and pretreated material to check for organic contamination and (v) using stepped combustion to check for concordancy in the ages of carbon released at successively higher temperatures. (C) 2002 Elsevier Science Ltd. All rights reserved.