Reply to Blockley, Donahue & Pollard

Article excerpt

We welcome Blockley et al.'s paper that raises the issue of calibration in the Late Glacial, although for reasons given below we believe their conclusions are premature. We are also gratified that they selected our paper from among several major books and papers on the period which have appeared since 1996 (Larsson 1996a; Otte 1997; Straus et al. 1996; Street & Terberger 1999; Tolan-Smith 1998). What this wider sample of the relevant Palaeolithic literature demonstrates -- about which Blockley et al. seem unaware -- is the degree of inconsistency in current approaches to calibration. Some of the papers in the edited volumes by Larsson (1996b) and Straus et al. (1996) calibrate, whilst others do not. Otte (1997) presents calibrated dates while Tolan-Smith (1998) prefers uncalibrated radiocarbon years for his study of the British data. This hesitancy is not confined to archaeologists, for the same is seen in geography. For example, at the English Heritage conference on Science in Archaeology, Edwards (1998) does not calibrate his pollen curves, although in a standard textbook such as Lowe & Walker (1997) calibration can be found. This is in a period where good calibration curves are available and the author is a former chair of the NERC Radiocarbon Accelerator Dating Service Advisory Panel. Consequently we agree with Street & Terberger (1999: 260), writing in ANTIQUITY, that in the Late Glacial:

Until there is consensus for a common calibration system, the presentation of the `raw dates' avoids confusion and even the uncalibrated radiocarbon dates provide convincing evidence for continuity or hiatus of settlement in different regions of Europe.

Let us be clear what Blockley et al. are advocating, namely the use of the 1993 and 1998 marine curves for calibrating radiocarbon determinations on terrestrial archaeological samples. It is important to realize just how problematic these curves currently are for uncritical archaeological use. The marine curves are based on paired U/Th and AMS [.sup.14]C determinations of corals. Although it is assumed that U/Th ages conform to a calendrical timescale, it is not proven. One only need look at the early history of the [.sup.14]C method to see how we once thought [.sup.14]C was an absolute method and it was only later realized that uncalibrated [.sup.14]C ages were not calendar years. There is also the question whether corals are closed-systems, for if loss or exchange of Th has occurred the data will be invalid. However, the over-riding problem for this period is the varying marine reservoir effect. Fluctuations in the amount of old carbon incorporated into the oceans during deglaciation means that the 400-year present-day offset cannot be assumed. Although we acknowledge that the marine curves are useful for indicating the long-term age offset between the [.sup.14]C timescale and calendrical time, the smoothed nature of the ocean-derived data and the variable marine reservoir effect means they should only be used with caution. It is because the calibration data are so heavily smoothed that Blockley et al. dispute our notion of a northward movement of peoples. We acknowledge that calibration should be a goal but to have simply adopted the marine data as it was available in 1997 seemed to us premature. The subsequent publication, in late 1998, of a new integrated decadel calibration curve (Stuiver et al. 1998) partly answers these concerns, but the precision is still poor in the Late Glacial once beyond the limit of the German Pine dataset (i.e. for ages [is greater than]11,871 calendar years BP).

In essence, Late Glacial Palaeolithic archaeology is undergoing the interest that surrounded the first calibration curve 25 years ago that gave us Wessex without Mycenae (Renfrew 1968). We need to ensure that, as calibration is extended, it is extended for a good archaeological reason. Clearly the issue of re-colonization and settlement pattern is fertile ground for wanting the precision which calibration offers. …