科研之友
微信
新浪微博
Facebook
分享链接
摘要
In interpreting radiocarbon dating results, it is important that archaeologists distinguish uncertainties derived from random errors and those from systematic errors, because the two must be dealt with in different ways. One of the problems that archaeologists face in practice, however, is that when receiving dating results from laboratories, they are rarely able to critically assess whether differences between multiple C-14 dates of materials are caused by random or systematic errors. In this study, blind tests were carried out to check four possible sources of errors in dating results: repeatability of results generated under identical field and laboratory conditions, differences in results generated from the same sample given to the same laboratory submitted at different times, interlaboratory differences of results generated from the same sample, and differences in the results generated between inner and outer rings of wood. Five charred wood samples, collected from the Namgye settlement and Hongreyonbong fortress, South Korea, were divided into 80 subsamples and submitted to five internationally recognized C-14 laboratories on a blind basis twice within a 2-month interval. The results are generally in good statistical accordance and present acceptable errors at an archaeological scale. However, one laboratory showed a statistically significant variance in ages between batches for all samples and sites. Calculation of the Bayesian partial posterior predictive p value and chi-squared tests rejected the null hypothesis that the errors randomly occurred, although the source of the error is not specifically known. Our experiment suggests that it is necessary for users of C-14 dating to establish an organized strategy for dating sites before submitting samples to laboratories in order to avoid possible systematic errors.
- 出版日期2016-3
