The Mill Bay area on the south shore of Milford Haven -- originally deemed to be the likely source area for the Altar Stone. Now back in the frame?
I have been taking some advice on that strange recent paper by Bevins, Ixer et al, and it looks as if my concerns are shared by others who know far more about sedimentary rocks than I do.
https://brian-mountainman.blogspot.com/2022/05/the-altar-stone-may-not-have-come-from.html
I said that I was not convinced by the evidence presented or by the interpretations of the authors -- and others are prepared to be rather more brutal in their assessment. This is a comment from an expert on pXRF work who prefers to remain anonymous:
"I have had a quick skim through the paper you attached and my general feeling is that the data are so compromised that I simply wouldn't have used them. Their understanding of XRF is sound, but the issues they have had with low totals and wet vs dry attenuation have forced them to be very selective in which data they accept and which they don't. In principle, it's certainly possible to acquire meaningful data for sandstones (the finer-grained the better), but the detector should be held orthogonal to a very flat dry surface (not laid on a tripod on an irregular wet surface) and the Ba, Sr, Ca etc could well be within clays and feldspars, rather than barytes and calcite. Such XRF is essentially a surface detection technique, so there is also the problem of weathering crusts to consider, not to mention burial within a chalk soil and whatever artefacts have been induced by archaeological processes and modern "witch" ceremonies!"
I also received a long analysis of the paper from Richard Thomas, who has himself been involved in Altar Stone work in the past. These are some of his points:
1. The authors make a series of assumptions used to bolster the claim that their methodology and results are valid. For example, although they do mention the "widely publicised concerns over the pXRF technique", they proceed to more or less ignore its documented shortcomings. What (very little) I know about pXRF is that it works best for essentially unweathered, compositionally homogeneous lithologies. It seems to me that there is no honest discussion here about whether pXRF is an appropriate technique for fingerprinting sandstones and determining their provenance. The authors use it and bingo, their results show it works. I would have liked to see them include some (non-ORS) sandstones as "controls" in their analyses. I'm very dubious about the utility of pXRF analysis for sandstone fingerprinting and provenance research. (I share Richard's concerns about the lack of analyses of controls for comparative purposes. Sadly, that is a shortcoming of many other papers by the same authors.).
2. Of course, the authors were 'forced' to use pXRF by English Heritage's stubborn refusal to allow a fresh sample (ideally a 1" core) to be taken from the actual Altar Stone -- from below the present ground surface (so it would be invisible to visitors) and then filled with concrete before reburial. Given the shenanigans that take place on every Summer Solstice at Stonehenge, which have included lighting fires on the Altar Stone, EH's attitude demonstrates a double standard to say the least.
3. Unless you are dealing with pure quartz arenites, the detailed composition of sandstones (at both large and small scales) is vertically and laterally variable. Source rock composition, the nature/energy levels/duration of the transporting mechanism(s) and subsequent diagenetic effects play key roles of course but, at a smaller scale, the interplay of stratification type (a product of bedform) and texture (i.e., grain size and sorting, combined with hydraulic equivalence) can generate significant local differences in sandstone mineralogy. For instance, heavy minerals can be concentrated at the toes of cross-bed foresets and micas (and other 'lights') in the troughs between ripple or dune crests. Even when conducting multiple analyses, if your analysis spot is only 8mm in diameter, such compositional heterogeneity has to be factored into pXRF sandstone data interpretation.
4. Bearing that in mind, I find it remarkable that this article contains NO descriptions of the grain sizes or other macroscopic characteristics of the samples, except (from Thomas, 1923) for the statement that the Altar Stone is a "grey-green micaceous sandstone". That in fact is my description; Thomas (1923) described its colour as "pale sage-green".
5. The conclusion that the Altar Stone and debris fragments are of essentially the same composition (p.6) is based on the values they obtained for "high atomic number elements" -- namely (table 4) Mn, Sr, Zr, Mo, Pb, Th and U, which presumably are in large part a reflection of the samples' heavy mineral assemblage compositions. Do the samples contain heavy mineral laminae? Lacking a macroscopic description of these sandstones we cannot tell figure 5 is of no help in this regard).
6. Are the authors correct in concluding that the rock fragmentss they analysed came from the Altar Stone? I'm sceptical and, at best, think more substantive evidence is required. However, even if they are right, I believe that any definitive study of the Altar Stone's provenance must be based on samples taken from the Altar Stone itself, not small, proxy specimens. In my opinion, any such provenance study has to begin by thoroughly documenting the Altar Stone's macroscopic characteristics followed by detailed thin section analysis. As I've noted before, thin sections can be a wonderful source of information about a sandstone's original compositional and textural characteristics and its subsequent diagenetic history. The authors here are enamoured with high-tech analytical techniques but are in danger of neglecting the fundamental building blocks of any sandstone provenance work. (NB. Way back in 1973 (!) I noted that the visible portion of the Altar Stone was a fine-grained, muscovite-rich, small-scale trough cross-laminated, grey-green sandstone that appears similar to a number of such sandstones found within the Lower ORS sequence of South Wales. Its largest surface is bedding parallel and stratigraphically, the block appears to be lying right way up.)
7. There is still a discussion about the petrographic description of the Altar Stone given by H.H. Thomas in his classic 1923 paper; was it really based upon a thin section taken from the Altar Stone itself? Not everybody accepts that, but I prefer to give my namesake the benefit of the doubt.
8. Since they found barite in an Altar Stone proxy sample (FN 196) using SEM-EDS (see Bevins et al., 2020; and figure 5 of this new article), the authors seem to have become obsessed with its importance in unravelling the provenance of the Altar Stone. Perhaps they're right, but I'm a bit dubious about that since, with a maximum modal content of 0.8% BaSO4, it is still just a trace mineral. I think other, more obvious characteristics will prove to be of greater value. For their pXRF data presented in table 5, the authors assume all Ba to occur as barite but (as they grudgingly acknowledge in the table caption) it seems far more likely to me that the Ba is actually present within K-feldspars, micas and clay minerals (e.g., illite).
9. To be frank, I found the "Altar Stone and ORS comparisons" section to be laughable. I don't think anyone with a reasonable level of knowledge of the ORS sections of South Wales and the Welsh Borderlands and familiarity with the Altar Stone, would expect it to be derived from any formation other than the Senni Formation or its lithostratigraphic equivalents. Provenance studies should be about comparing like with like. It would be shocking if the Altar Stone (proxies) and samples from the other ORS formations listed in table 1 did not exhibit major compositional differences. Based upon just 3 samples from the Mill Bay Formation, Bevins et al. (2020) ruled out the Cosheston Subgroup as a source for the Altar Stone. Without further study however, I believe this conclusion to be premature since the Burton Cliff and Llanstadwell Formations both contain some thick, grey-green sandstone units. Again, I may be proven wrong, but I fully expect the source of the Altar Stone to turn out to be a Lower ORS Senni Formation-related sandstone unit from South Wales.
====================
So there we have it. It does not look as if the specialists in pXRF work or the specialists in ORS / Devonian rocks are going to be very impressed by this paper, and it probably should not have been published. We still cannot take it that the fragments analysed by the authors actually did come from the Altar Stone; and we cannot take the main conclusions of this paper as reliable.
The origin of the Altar Stone is as mysterious as ever -- notwithstanding the contents of this new paper, it still appears probable that it came from somewhere in the ORS Welsh sequence, with the balance of possibilities tending towards the Senni Formation. But who knows? Maybe it did, after all, come from the shores of Milford Haven? Now that would be fun, wouldn't it?
The message from all of this? The provenancing of the Stonehenge bluestones is not very easy at the best of times, and a lot of the work over the past 20 years by Bevins, Ixer and others (which should have been strictly geological) has been spoiled by contacts with the archaeologists that have been far too close for comfort. There has been an ongoing quest for spectacular results and banner headlines linked to quarries and "lost circles". Over and again, the quality of the science has been compromised. As I have said repeatedly on this blog, there have been too many assumptions and distortions, and too many ruling hypotheses........