Prof Christopher Ramsey

Abstract:

One of the key factors in understanding past climate change is the development of robust age models that have sufficient chronological precision for comparing different palaeoclimate archives, while retaining accuracy. Recent developments in Bayesian age modelling are applied here to the Swiss varve lake, Soppensee. We develop revised age models for this sequence using the available stratigraphic information to constrain the calibrated radiocarbon ages. We begin by using stratigraphical order as the only constraint and then sequentially increase the information incorporated into the model, using relative varve age, sample depth, and varying depositional models. Within this Bayesian framework, we develop internally robust models that significantly improve the dating precision and are applicable to both varved and non-varved sequences. We then compare the model output to wider chronological information, such as the age of the Laacher See Tephra in order to test the model accuracy. The results of this exercise suggest that these methods can be used to build very reliable improved age models in a variety of records. © 2007 Elsevier Ltd. All rights reserved.

Abstract:

Boaretto et al. (2005) published 68 radiocarbon dates relating to 30 samples from 10 Iron Age sites in Israel as part of their Early Iron Age Dating Project. Though the main goal of their paper was an interlaboratory comparison, they also presented results of Bayesian models, calculating the transition from Iron Age I to Iron Age II in Israel to be about 900 BCE instead of the conventional date of about 1000 BCE. Since this date has great importance for all of Eastern Mediterranean archaeology, in this paper we examine the results in light of the dates published in the above-mentioned article. Our paper was revised in light of new data and interpretations published by Sharon et al. (2007). Following a survey of the contexts and specific results at each site, we present several Bayesian models. Model C2 suggests the date range of 961-942 BCE (68% probability) for the transition from Iron Age I to Iron Age II, while Model C3 indicates a somewhat later date of 948-919 BCE (compare the date 992-961 BCE calculated at Tel Rehov for the same transition). In our Model D, we calculated this transition date at Megiddo as taking place between 967-943 BCE. Finally, we calculated the range of dates of major destruction levels marking the end of the Iron Age I, with the following results: Megiddo VIA: 1010-943 BCE; Yoqneam XVII: 1045-997 BCE; Tell Qasile X: 1039-979 BCE; Tel Hadar: 1043-979 BCE (all in the 68.2% probability range). Figure 4 indicates that the transition between Iron I and II probably occurred between these above-mentioned destruction events and the dates achieved in our Models C2 or C3, namely during the first half of the 10th century BCE. This study emphasizes the sensitivity of Bayesian models to outliers, and for reducing or adding dates from the models. This sensitivity should be taken into account when using Bayesian models for interpreting radiometric dates in relation to subtle chronological questions in historical periods.

Abstract:

The performance of the CO2-accepting SO-110 ion source at Oxford Radiocarbon Accelerator Unit has been investigated in detail. The purpose was to clarify the possibilities of accelerator mass spectrometry (AMS) radiocarbon measurements with real-time separation, e.g. GC AMS or HPLC AMS. The construction of a gas test injector based on the continuous-flow technique made it possible to characterize the response of the ion source to continuous and pulsed input of CO2 gas. The source exhibited remarkably good linearity over a wide range of CO2-pulse sizes and fast rise time, but the peak shape varied and memory effects were significant. Appropriate tuning of the gas source proved to be critical.