Dating the middle to late woodland transition in the Illinois Valley: Radiocarbon and thermoluminescence dates from the Baehr-Gust site

Midcontinental Journal of Archaeology, Spring 2003 by Holt, Julie Zimmermann, Feathers, James K

In short, TL dating supports the second hypothesis: There were two components at the site, a Havana occupation followed by a White Hall occupation.

Discussion

To compare radiocarbon dates and TL dates, we calibrated the radiocarbon dates so that we would be comparing actual calendar years (see Methods above). Ignoring the margins of error, TL dates (as determined using the collected sediments for external dose rate) deviated from the calibrated radiocarbon dates by as much as 431 years (see Table 3). The TL dates from every sherd but one were younger than calibrated radiocarbon dates from the same feature.

Figure 1 shows the comparison between the TL dates and calibrated radiocarbon dates from the same features. The two data sets positively correlate, although the correlation is not significant at the .05 level (r = .4119; p

Why are the TL dates systematically younger than the calibrated radiocarbon dates? The radioactivity of the collected sediments was fairly consistent across the site, but how representative these are of the feature fill in which the sherds were found is not known. It is likely that the external dose rate is overestimated, because the features contained abundant organics, which attenuate the dose rate, and some ash, which generally has lower radioactivity. The ages were recalculated assuming 20 percent organics. We do not assume this is an accurate reflection of the organic content of these Woodland pits (compare Stein 1984), but values of that magnitude have been obtained from archaeological sites, for example, 16 percent from a Northwest shell midden (Holliday and Stein 1989) and 20-25 percent from a South African rockshelter (Feathers et al. 2001). Such an assumption makes the dates as much as 90 years older, reducing the discrepancy with the calibrated radiocarbon dates (Table 5).

We do not know the cause of the discrepancy, but a likely explanation is an overestimation of the unknown external dose rate. This assumption also allows assessment of the potential error stemming from the sherd pairs not having exactly the same provenience in the pits. The worst case is one where one of the sherds experienced a dose rate similar to that of the collected sediments and the other experienced the organics-attenuated dose rate. For Feature 91-XVIII, this worst case could make the date for one sherd 48 or - 169 B.C. and for the other A.D. 702 or - 115-still a very significant difference. For Feature 1(1987), this worst case would make the ages of the sherds statistically indistinguishable. Slightly different external doses may explain the discrepancy in age between those sherds.

The recalculated TL dates corroborate the relative ages of Feature 1(1987), Feature 4(1988), and Feature 89-XVIII. This is particularly important since Feature 4(1988) produced a surprisingly old radiocarbon date, one of the oldest at the site.

As discussed above, the TL dates clarify the age of Feature 91-XVIII, indicating that the calibrated radiocarbon date predates one of the ceramics. To reiterate, this feature contained several Havana-Hopewell sherds and an abundance of White Hall sherds. TL dates indicate that the Havana-Hopewell sherd dated from this feature was hundreds of years older than the White Hall sherd dated from this feature. It is clear based on the predominance of White Hall sherds within this feature and TL dating of these two sherds that this was a Late Woodland feature containing a few Middle Woodland sherds and probably some older charcoal as well.