Gault site: The 16 000 year old presence is looking better, but there are still problems

A new article last week claims that there is now definitive evidence of a 16 000 year old pre-Clovis human presence at the Gault site in Texas. The Gault site (Williams et al 2018) and the nearby Friedkin site (Waters et al 2011) have been some of the best documented claimed early sites in North America for several years now. The claim is that a group of hunter-gatherers lived in the area as long as 16 000 years ago and left traces of a lithic technology distinct from, but possibly ancestral to the later Clovis complex.

Pre-Clovis claim

The claim is important because, if well supported, it would be the earliest known material evidence of human presence in the Americas, and would pretty much require people to have come here through the Pacific coastal corridor (or some other unknown route) rather than the inland ice free corridor. As Potter et al (2018) reminded us just last week, the current state of the evidence on the earliest presence in the Americas requires us to be agnostic as to which corridor was used, or to accept both as routes for the earliest arrivers. If we accept that the earliest archaeological traces of humans in the Americas are younger than about 15 000 years, both corridors would have been available. If we push back human history in the Americas to 16 000 or more, then the inland route becomes much more difficult to accept.

Morrow et al (2012) published a detailed critique of the 2011 version of the Gault/Friedkin claim. Their most convincing points against the claim were that 1) the optically stimulated luminescence (OSL) dates reported by Waters et al seemed to be too old by about 1000 years for the Clovis layers of the site, and 2) that there was an increase in the proportion of very small lithics from the top to the bottom of the stratigraphy, indicating some natural process of sorting and down migration of the artefacts, rather than archaeological deposition at the geological age of the layers.

If the OSL dates for Clovis were too old by 1000 years, then perhaps the dates for the lower layers containing the pre-Clovis were also inaccurate, perhaps by even more. If there was a systematic relationship between proportion of small lithics and depth, perhaps the alleged pre-Clovis was in fact made of Clovis and later remains that had been driven down the layers by natural processes.

Last week, Williams et al published a new set of dates for their proposed pre-Clovis presence at Gault, right next to Friedkin. They clearly set out to address some of the critiques brought up by Morrow et al in 2012. They address some of the concerns, but there are still problems.

The Gault assemblage

There is definitely a cultural assemblage at Gault, and the lithics are definitely the result of human activity. But that hadn’t been questioned as far as I know. The question in this case is not whether the assemblage is the result of human activity, but rather whether it is really 16 000 years old.

Based on the material presented with the article, the assemblage does look different from the classic Clovis. Looking at the images supplied with the article, the layer that bears the pre-Clovis assemblage artifacts seems to be stratigraphically well separated from the Clovis assemblage above it. The artefacts recovered seem to show a good variety of shapes and sizes, casting some doubt on any sorting and down migration effects.

More importantly, the OSL dates reported for the Clovis layers in the site are consistent with Clovis in other places, and the underlying pre-Clovis assemblage dates are around 16 000 years old. So that would seem to cover both major objections raised by Morrow et al in 2012.

However, as is increasingly the case in articles published in high impact factor journals these days, the devil is in the supplementary materials and the reference section. Figure S4 shows two very interesting trends. First, there is clearly a drop in artefact frequency in the layers between the Clovis and the Gault. That’s good. But it also shows that, as was the case at Friedkin, the average weight of the identified artefacts in the lower layers is much less than in the Clovis and later layers. So either the people who produced the 16 000 year old pro-Clovis assemblage made lots of very small tools compared to their successors, or there is indeed some sorting and pooling of small flakes, tools, and fragments down at the bottom of the stratigraphy. Based on this article, I think this is still an open question.

The other main question is the dating. Yes, the Clovis and later dates are in the right ranges, which makes it more likely that the earlier dates are also good. However, the two phases of the OSL dating for each of the four samples were done at a single pair of labs. For something this important, I would like to see at least some replication. I would have split the four dates between at least two sets of labs, and if at all possible, I would have done more dates. Knowing that those dates were going to be the most closely examined evidence for their claim, it would have been prudent of Waters et al to beef up that part of their project.

There is also that the Clovis dates were actually done separately from the pre-Clovis dates, which is unfortunate, but at least they were all done by the same labs (Rodrigues et al 2016). So we have a pair of labs getting good Clovis dates, and then the same pair getting pre-Clovis dates from the same site later. Better than getting good Clovis dates from one pair of labs, and then getting pre-Clovis dates from another. But still not ideal.

Overall, the dates are good, but not exactly great. Table S1 shows a range of 16.7 ± 1.1 to 21.7 ± 1.4 ky, which is quite a bit of variation. The sample highest in the stratigraphy is dated at 18.7 ± 1.2, older than the two middle ones, which at least are tightly clustered both in depth and in OSL age. The older one, which is mercifully at the bottom, is very old indeed, and it isn’t clear from the text whether it is closely associated with any of the archaeological remains.

Then there is the nagging voice in the back of the mind that says OSL has made a lot of progress as a dating technique in the past couple of decades, but it is still a bit mysterious. A recent survey of advances in OSL, written to reassure archaeologists about its accuracy and reliability, especially in the context of early American sites, does nothing of the kind (Forman and Waters 2016).

For example, from a layman’s reading of both Forman and Waters (2016) and Williams et al (2018), Uranium disequilibrium seems to be quite important in OSL dating, but for some reason “disequilibrium measurements were not carried out for the Gault samples”.  But not to worry, because “several inferences can be made about the potential impact that this would have on the ages we report” (Williams et al 2018). Again, not reassuring, and definitely not something you want to read in a paper that wants to change our understanding of the peopling of the Americas.

Almost there

Gault is getting close to being an honest to goodness well attested pre-15k presence in the Americas. The dating is more solid than it was on the basis of the Friedkin dates alone, but there are still some problems. Mainly, we need a few more dates done by several labs, to show replicability. The question of down migration of artefacts is still open. However, if there is down migration and sorting, it is not simple and straightforward, since there is a significant and obvious gap between the Clovis layers and the Gault layer. I look forward to more work from these teams.

References

Forman SL and Waters MR 2016. Optically stimulated luminescence dating and the peopling of the Americas, PaleoAmerica 2https://www.tandfonline.com/doi/full/10.1080/20555563.2015.1136722

Morrow JE, Fiedel SJ, DL Johnson, M Kornfeld, M Rutledge, WR Wood 2012. Pre-Clovis in Texas? A critical assessment of the “Buttermilk Creek Complex”, Journal of Archaeological Science 12: 3677-3682. https://www.sciencedirect.com/science/article/pii/S030544031200218X?via%3Dihub

Potter BA, JF Baichtal, AB Beaudoin, L Fehren-Schmitz, CV Haynes, VT Holliday, CE Holmes, JW Ives, RL Kelly, B Llamas, RS Malhi, DS Miller, D Reich, JD Reuther, S Schiffels, TA Surovell 2018. Current Evidence allows multiple models for the peopling of the Americas, Science Advances 4  http://advances.sciencemag.org/content/4/8/eaat5473/tab-pdf

Rogrigues K, WJ Rink, MB Collins, TJ Williams, A Keen-Zebert, GI Lopez 2016. OSL ages of the Clovis, Late Paleoindian, and Archaic components at Area 15 of the Gault site, Central Texas, USA, Journal of Archaeological Science: Reports 7:94-103. https://www.sciencedirect.com/science/article/pii/S2352409X16300827

Waters MR, SL Forman, TA Jennings, LC Nordt, SG Driese, JM Feinburg, JL Keene, J Halligan, A Lindquist, J Pierson, CT Hallmark, MB Collins, JE Wiederhold 2011. The Buttermilk Creek complex and the origins of Clovis at the Debra L Friedkin site, Texas, Science 331: 1599-1603.

Williams TJ, MB Collins, K Rodrigues, WJ Rink, N Velchoff, A Keen-Zebert, A Gilmer, CD Frederick, SJ Ayala, ER Prewitt 2018. Evidence of an early projectile point technology in North America at the Gault Site, Texas, USA, Science Advances 11http://advances.sciencemag.org/content/4/7/eaar5954