The hygienic dater

I've just been reading a useful paper by Andrew Millard, which reviews the chronometric dates of African and Near Eastern fossil hominids from the Middle and early Late Pleistocene. The overall theme is that we don't know the dates nearly as well as we would like -- or as well as many comparative analyses have assumed.

The highlight is the list of specimens with primary references to different date estimates. Anyone with a good training in paleoanthropology probably has a feel for which specimens have relatively good dates and which are real hands-up-in-the-air cases. Kabwe makes for a good example of the latter:

Kabwe (Broken Hill), Zambia. The remains of "Rhodesian Man," along with faunal remains, were discovered in 1921 by miners (Klein, 1973). The principal dating is based on Klein's (1973) assessment that the fauna is similar to that at Elandsfontein and broadly similar to those from Olduvai Gorge Upper Bed II through to Bed IV. There are no chronometric determinations. On the basis of the faunal correlation to Olduvai (Fig. 1), an age of younger than 1780 ka and, depending on the chronology for Olduvai, either older than 990 ka (on the long chronology) or, more likely, older than 490 ka (on the short chronology) may be assigned (see under Olduvai above). This is consistent with Elandsfontein being older than 330 ± 6 ka (Table 1).

Millard's discussion of "chronometric hygiene" takes up much of his discussion. This is nothing more than the simple idea that we should weed bad dates out of our analyses. For example, he singles out Florisbad as a specimen that has been handled poorly in the literature:

Use of the literature. In conducting this review of the chronometric evidence for African and Near Eastern hominids, the search for the detailed chronometric data was hampered by overreliance of many authors on the secondary literature. It is not uncommon to find a date cited from a publication, which upon checking simply cites another publication, which cites another, which cites the paper that first suggested the date. Frequently in such a chain of citations, the justification for the original date is lost, and in some cases, error limits disappear. For example, the ESR date of 259 ± 35 ka for the Florisbad hominid (Grün et al., 1996) can be applied to the Florisbad fauna, but somehow in the discussion of Stynder et al. (2001), this becomes simply "a maximum age of around 250 ka" (p. 372) for the Florisbad Faunal Span, and in McBrearty and Brooks (2000), it becomes a bald 260 ka age without any uncertainty for the Florisbad hominid itself. Sometimes, the primary proposal for a date is based solely on comparisons of morphology to the best-dated fossils at the time of publication, and for later papers to suggest evolutionary sequences based on this date is obviously problematic. Given the flux in dating methods, the fact that problems have often been identified some time after the introduction of these methods, and the changing understanding of the dates of faunal successions, every author should be beholden to check the basis of the dates cited and apply some basic chronometric hygiene (Millard 2008:19).

Of course, there is an irony here, since Millard's effort has generated a massive secondary source listing date estimates for all these hominids! I agree whole-heartedly with his sentiment, though -- everyone should do a better job of reading and citing papers.

But the effect of all this hygiene is to emphasize that most of the Middle Pleistocene remains a muddle, with very few well-resolved dates across the entire span. Millard describes faunal correlations as a relatively weak source of evidence in Africa. Above the time span effectively covered by ESR/TL, there is little to rely on.

References:

Millard AR. 2008. A critique of the chronometric evidence for hominid fossils: 1. Africa and the Near East 500-50 ka. J Hum Evol (in press) doi:10.1016/j.jhevol.2007.11.002