Altar Stone paper
"The Stonehenge Altar Stone was probably not sourced from the Old Red Sandstone of the Anglo-Welsh Basin: Time to broaden our geographic and stratigraphic horizons?” by Richard E. Bevins, Nick J.G. Pearce, Rob A. Ixer, Duncan Pirrie, Sergio Ando, Stephen Hillier, Peter Turner, Matthew Power.
Journal of Archaeological Science: Reports 51 (2023) 104215
Comments from Dr Richard Thomas
In 1923, H.H. Thomas (p. 245) wrote: “From general considerations, however, the type of heavy [mineral] residue and the lithology of the rock as a whole are sufficient to make the identification of the altar-stone with the Old Red Sandstone of South Wales almost a matter of certainty.” It’s been 50 years since I knelt by the Altar Stone to take some photos and was greatly amused to hear two ladies behind me say: “what’s ee doin’?” … “ee’s takin’ a picture of a ten-pence piece.” My immediate impression at that time was to wholeheartedly agree with Thomas’s assessment of the Altar Stone’s probable provenance, and nothing in this latest publication leads me to change my opinion.
1. The presentation, superficially at least, appears to be authoritative. The element concentrations are what they are. Using ternary plots of 3 elements or element ratios is a well-established and effective procedure for discriminating between, or correlating, igneous units such as lavas and volcanic ashes. The problem I have with them using such an approach for sandstones is that there are many variables that contribute to the initial composition of a fluvial sand body when deposited, and many more that affect its diagenetic history and hence, its ultimate composition (e.g., mineral dissolution and replacement, etc.). It would be interesting to compare and contrast other detailed studies attempting to correlate sandstones on the basis of their geochemical composition.
2. Another issue I have is that the heavier elements they use in their diagrams (e.g., Nb, Th and Zr) are concentrated within the sandstones' heavy mineral fraction (such as in zircons, tourmalines and rutiles = ZTR). The more mature your sandstone is (i.e., well sorted and composed of more resistant minerals), the more likely it is that the heavy mineral fraction will be ZTR dominated. As a result, you have to compare apples with apples -- i.e., sandstones with similar grain sizes and levels of textural and mineralogical maturity. Have the authors done this? I don't know because I haven't seen their sample site table, but I suspect not.
3. As far as I know, there has been no recent sampling of the Altar Stone itself – apparently, English Heritage will not permit it. In my view, unless such sampling occurs (e.g., a 25cm x 2.5 cm diameter core drilled from the underside of the Altar Stone), the provenance of the Altar Stone can never be conclusively proven. This lack of unequivocal Altar Stone samples is one reason for Bevins et al.’s heavy reliance upon high-tech analytical methods.
4. Unfortunately, I don’t have access to their Supplementary Table 1 which lists the site details for the 58 samples they analysed from the ORS of the Anglo-Welsh Basin, nor the 2006 paper by S. Hillier et al. whose sample set they draw upon extensively.
5. Bevins et al. frequently state they have “proven” (or “confirmed”) that various debitage samples were derived from the Altar Stone. Key amongst such samples is ‘Wilts 277’ (a.k.a. 2010 K 240) which they state is a thin section collected from the underside of the Altar Stone in 1844. They base this “proof” on comparisons between their portable XRF analyses from the weathered surface of the Altar Stone with SEM-EDS data from Wilts 277. They place great emphasis on the ‘high’ Ba content of the Altar Stone (to which we’ll return).
6. Rob Ixer has told me that he and Richard B do not think H.H. Thomas examined thin sections taken from the Altar Stone. I do not know the reason they believe this but, from my reading of the latter’s paper, I disagree.
7. Thomas (1923, p. 244) states that “the Altar Stone is exceedingly rich in garnet of small dimensions”. In addition, the upper surface of the Altar Stone, which is small-scale trough cross-laminated, is highly micaceous (with muscovite the dominant species). In 2010, I examined thin section Wilts 277 and told Rob and Richard that I thought it couldn’t be a specimen of the Altar Stone because it did not contain sufficient muscovite or garnet.
8. In my opinion, since sandstone composition (especially in fluvial sandstone bodies) varies both laterally and vertically, and with distance from source, due to textural differences and the influence of hydraulic equivalence, the use of whole-rock geochemical results (and primarily, in this case, the concentrations of one element) is not a viable provenance matching method for most sandstones.
9. The use of SEM-EDS analysis to generate modal compositional data for sandstones is also problematic because a “stepping interval” of 10 microns (as used by Bevins et al.) means that you sample individual grains within rock fragments and inclusions within mineral grains, etc.
10. Subjectively, I feel that Bevins et al. are tackling the analysis of sandstones almost as if they were igneous rocks (e.g., page 4: [Zr, Nb and Th] “will reside in accessory phases in the sandstones” – presumably they mean within heavy minerals.
11. Muscovite abundance is one of the Altar Stone’s definitive features, and yet Bevins et al. (page 5) state that it is not “a critical characteristic discriminator in terms of comparison between known Altar Stone and questioned ORS samples.”
12. We need more information about the Anglo-Welsh ORS – after all, most ORS sandstones are red – and the authors have not stressed the importance of the (primary) grey-green colour of the Altar Stone from a provenance perspective.
13. For example, they have invested a great deal of effort in comparing Wilts 277 with their WM 6 sample – because of the latter’s ‘elevated’ Ba content. Surely, in sandstone provenance studies the idea is to compare like with like, whereas WM 6 is red and coarser grained than Wilts 277 and so it’s hardly a surprise that they are not a match.
14. Based on their previous work, Bevins et al. have started their Altar Stone provenance study at the microscopic level. For example, they base their heavy mineral analysis of sample MS-1 on 0.1973g of material – which seems unlikely to constitute a statistically representative sample.
15. I believe such a study should instead begin with the ‘big picture’ (basin-wide outcrop) level before homing in on candidate sandstones’ microscopic characteristics. In terms of the latter, thin sections offer a wealth of provenance-matching information that Bevins et al. ignore or barely mention.
16. I haven’t kept up with the literature, but I don’t think there have been any modern, detailed studies of the petrography and diagenetic histories of grey-green L. ORS sandstones across the A-W Basin, have there?
17. From personal observations I know that grey-green sandstones within the Senni (Beds) Fm. vary considerably from west to east in terms of their detrital compositions and dominant cements (cf. Heol Senni, Brecon Beacons and Primrose Hill, etc. quarries).
18. While Bevins et al.’s discovery of relatively ‘high’ baryte concentrations in the Altar Stone is certainly interesting, it should be regarded as the ‘icing on the cake’ (to confirm or dismiss potential provenance matches for the Altar Stone) rather than the primary investigative tool.
19. Bevins et al. also make a big deal about the relative lack of high Ba concentrations in stream sediments on the A-W ORS. I don’t find this to be a concern. What was the spacing between samples? Look at the relative sizes of the outcrop areas of the ‘Red Marls’ and Brownstones versus that of the Senni (Beds) Fm.
20. I think Bevins et al. are barking up the wrong tree(s) but support them in their desire to solve the mystery.