Error variation in OSL palaeodose estimates from single aliquots of quartz: a factorial experiment

We use a factorial experiment to study systematic and random differences between measured OSL palaeodoses for a variety of quartz samples. These include samples that have absorbed either a large or small natural or laboratory-induced radiation dose, either with or without prior heating or bleaching. The systematic factors studied are the size of the test dose, the preheat temperature and the number of quartz grains in each multi-grain aliquot. Palaeodoses were estimated using a single-aliquot regenerative-dose protocol. The main parameter of interest is the amount of random variation, over and above that due to photon counting statistics, to be expected between estimates from aliquots that have received the same radiation dose. This over-dispersion is generally larger for natural samples than for artificially bleached ones, and it varies from about 1% in the most favourable cases to about 18% for small aliquots of a sample that had received a natural dose of about 46 Gy. The latter is comparable to the over-dispersion reported for single grains of natural quartz that are thought to have been well-bleached at the time of deposition. The factorial experiment also revealed a number of systematic effects. In particular, measured palaeodoses using a preheat temperature of 260degreesC were systematically lower than those using 180degreesC, by up to about 5% in some cases. (C) 2004 Elsevier Ltd. All rights reserved.