Kuzmin (Reference Kuzmin2019) uses the results of radiocarbon dating of Palaeolithic human remains from western Russia to raise questions about the reliability of dates based on ultrafiltration and single amino acids. He fails, however, to understand the chemistry underpinning and supporting the reliability of isolating hydroxyproline (Hyp) for AMS dating of bone. His suggested alternative approach, the dating of bulk collagen, can sometimes provide reliable ages, but at other times it does not. Simply because bulk methods work in one instance does not mean that they will work in all cases. This is because site-based contaminants can sometimes chemically cross-link with collagen molecules post-depositionally, and these bonds can be very difficult to break with routine bulk chemistry methods. These bonds can be broken, however, with acid hydrolysis, splitting the collagen molecules into single amino acids and then, by applying semi-preparative-HPLC (high-performance liquid chromatography) methods, separating the amino acids from one another and, therefore, from contaminating sources of carbon. This method is able to isolate contaminant-free single compounds for AMS dating. In the Oxford laboratory, we have focused on Hyp, which elutes between 15.5 and 17.5 minutes on our HPLC system, and provides 12.9 per cent of the total collagen amino acid carbon (Devièse et al. Reference Devièse, Comeskey, McCullagh, Ramsey and Higham2018a). We have shown that our method can separate Hyp without adding significant modern or background carbon from laboratory sources.
The Hyp approach is the most reliable method for dating bone because this amino acid is almost exclusively found in mammalian bone collagen, so it acts rather like a biomarker. Kuzmin (Reference Kuzmin2019) suggests that as Hyp is also found in vascular plants and soils, the extraction of this amino acid from a bone is no guarantee of exogenous carbon removal and therefore there could be Hyp contaminants in samples taken from Palaeolithic bone for dating. There is Hyp in plant glycoproteins, but the mass of these proteins in the cell walls and other components is so small that one would need grams to tens of grams of whole plant matter mixed with the animal collagen to deliver enough Hyp to contaminate it significantly. In addition, the cell wall and extensia components of plant matter are insoluble during the gelatinisation step of the collagen preparation process, and hence, even if somehow present with the bone, they would not be taken through to hydrolysis and AMS dating. In terms both of the limited quantity of these potential contaminants and their chemical exclusion during pre-treatment, therefore, any plant-derived Hyp, if it were present, would be insignificant. For these reasons, we can be confident that AMS dates on Hyp are highly reliable and are to be favoured (contra Kuzmin Reference Kuzmin2019). The reliability of the Hyp approach has now been demonstrated in a large number of cases at a variety of sites (e.g. Marom et al. Reference Marom2012, Reference Marom, McCullagh, Higham and Hedges2013; Nalawade-Chavan et al. Reference Nalawade-Chavan, McCullagh and Hedges2013; Devièse et al. Reference Devièse2017, Reference Devièse, Stafford, Waters, Wathen, Comeskey, Becerra-Valdivia and Higham2018b, Reference Devièse2019; Reynolds et al. Reference Reynolds, Dinnis, Bessudnov, Devièse and Higham2017; Sikora et al. Reference Sikora2017; Becerra-Valdivia et al. Reference Becerra-Valdivia, Waters, Stafford, Comeskey, Devièse and Higham2018; Bourrillon et al. Reference Bourrillon2018; Dinnis et al. Reference Dinnis, Bessudnov, Reynolds, Devièse, Pate, Sablin, Sinitsyn and Higham2018; McColl et al. Reference McColl2018; Moreno-Mayar et al. Reference Moreno-Mayar2018; Kosintsev et al. Reference Kosintsev2019).
Kuzmin questions the notion that one should view older dates as being more reliable in Palaeolithic contexts after contextual and inbuilt age issues are taken into account, such as the assumption that mammoth bones, for instance, are not contemporaneous with human occupation. In fact, the opposite is true, due to the overwhelming effect of modern carbon contamination compared with older contamination. Modern sources of carbon have many more 14C atoms than older material. This has been demonstrated innumerable times and is once again explicable by basic chemistry. Contamination by old/depleted carbon will dilute the radiocarbon concentration, but large amounts are needed to influence ancient Palaeolithic samples because the radiocarbon concentration of the sample being dated is already depleted. Ten per cent contamination by old carbon, for instance, will yield a date that is 800 years too old, which is minimally older compared with the size of standard errors. Modern carbon, on the other hand, is significantly more influential; a one per cent modern contaminant at 40 000 years will result in a date that is 7200 years too young. This is why older dates are almost certainly more reliable and accurate than not and why radiocarbon specialists view them as such. A classic example of this—one of many—is the case of El Sidron, Cantabria (Higham Reference Higham2011). Here, the radiocarbon ages for a discrete group of Neanderthal remains that should all date to the same period in fact span from ~10 000 BP to 49 000 BP. The most reliable of these dates are the oldest ones because of the overwhelming influence of trace modern contaminants on the more recent dates.
Kuzmin (Reference Kuzmin2019) suggests that, when dating well-preserved collagen samples, routine preparation methods produce reliable dates, but when dating poorly preserved material, no fraction can produce reliable results. In illustrating examples of the latter case, Kuzmin mentions the sites of Okhladnikov and Anzick. In fact, these two sites comprise well-preserved collagen samples (Anzick samples are between 13.7 and 21 per cent collagen; the higher range is the equivalent of modern bone, and Okhladnikov is greater than five per cent, again well preserved according to Krause et al. (Reference Krause2007)). The reason for the divergent results is not to do with preservation, but rather because bones from the site were treated with conservatives and glues. Again, only Hyp approaches have solved this. At Okhladnikov (where dates spread over ~8000 years from 29–38 ka BP and where Kuzmin (Reference Kuzmin2019: 1063) says that “the reason for this deviation is unknown”), we have recently obtained new Hyp dates that show that the true age of these bones is beyond the radiocarbon age limit because of contamination that was not removed in the original attempts at dating. At Anzick, new Hyp dates on the Anzick 1 specimen place it chronologically, for the first time, along with the other dated artefacts (antler rods) from the burial assemblage (see Becerra-Valdivia et al. Reference Becerra-Valdivia, Waters, Stafford, Comeskey, Devièse and Higham2018). This result makes complete archaeological sense. Once again, the non-compound specific dates cannot be trusted due to the presence of contamination.
Kuzmin suggests that as bulk collagen radiocarbon dates can be demonstrated by numerous infinite ages, these types of dates must be reliable. He also suggests that as there are examples in which bulk collagen pre-treatment dates with and without ultrafiltration are identical, this negates the utility of using ultrafiltration methods. These interpretations ignore the fact that bone from different sites is subject to variation in post-depositional chemistry, soil pH, water and microbial influences—all variables that influence collagen contamination and diagenesis. Just because in one case bones of infinite age produce infinite dates does not mean that the method used to produce the results will work in all cases. Similarly, just because there is one example in which a single amino acid date produces the same result as a bulk collagen age does not mean that single compound dating is not needed in all cases. Rather, in these instances, it simply means it is probable that there is minimal to no contamination. Frustratingly, we cannot be sure that bulk collagen dates are reliable in every case because the analytical tools that we use to determine reliability (CN atomic ratios, %C, % collagen) are not sufficiently sensitive to demonstrate that all non-collagenous contaminants have been removed. In our experience, only compound-specific single amino acid dates are able to provide greater certainty because the measured sample is only Hyp, and therefore we can be confident that there is no museum-based or humic-derived contamination present. Cherry-picking one or two examples to support one's argument is not a scientifically rigorous approach. Kuzmin's (Reference Kuzmin2019) misleading and circular article serves only to confuse rather than add clarity. Single amino acid or compound-specific methods are the future for reliably dating the Palaeolithic, where the effects of modern contamination are so acute.
