Using the Bayesian method to study the precision of dating by wiggle-matching.

Abstract

The "wiggle-matching" technique has been widely used for the absolute dating of a series of radiocarbon-dated samples connected in one floating chronology. This is done by calculations of SS statistics (the mean-square distance of (super 14) C ages of samples from the calibration curve) calculated for any assumed calendar age of the floating chronology. In the standard procedure the confidence intervals of true calendar age are derived from the width of the SS minimum, using the critical values of the chi-square distribution. This, however, seems oversimplified. Another approach is an extension of the Bayesian algorithm for calibration of single (super 14) C dates. Here, we describe in detail the Bayesian procedure and discuss its advantages compared to the SS minimization method. Our calculations show that for given errors of (super 14) C measurements, precision of dating the series is related to the shape of the SS curve around its minimum, rather than to the absolute value of SSmin. In some cases, dating precision may be improved more efficiently by extending the time span covered by the series rather than by improving the precision of the (super 14) C measurements. The application of the Bayesian method enabled us to delimit the age of the floating varve chronology from the sediments of Lake Gosciaz with distinctly better accuracy than was previously reported using the SS curve alone.