Stone 62 -- a nice academic spat about methodology

Unspotted dolerite stone 62 at Stonehenge (courtesy Simon Banton)

I came across this on the University of South Wales web site:

https://pure.southwales.ac.uk/ws/portalfiles/portal/39397422/1-s2.0-S2352409X26003093-main.pdf

https://doi.org/10.1016/j.jasrep.2026.105874

On following it up, I find that there are three papers and a rather intense and detailed academic spat -- all centred on the methodology for the spot provenancing of dolerites on Mynydd Preseli.  It makes a nice change to find an academic dispute that I am not actually involved in! Anyway, in the right corner, Pearce, Bevins, Ixer and Pirrie, and in the left corner Ciborowski and Nash. These two teams have argued rather violently once before, on the matter of the sarsens of West Woods and Stonehenge, but for now we will leave that to one side.

Here are the other two references:

https://doi.org/10.1016/j.jasrep.2025.105513 (the original article by C&N)

https://doi.org/10.1016/j.jasrep.2026.106012 (the C&N response to the Pierce et al response)

To summarise:

The dispute is very technical, but in essence it is about visual bivariate plots versus mathematical precision and standardization.  The clash exposes a fundamental rift in how these two teams of specialists believe archaeological stone sourcing should be conducted.

Pearce, Bevins, & Ixer have for many years had visual bivariate grouping graphs right at the centre of their provenancing work.  For example, in 2022 they used portable pXRF to analyze Stonehenge Stone 62. They plotted elemental data (like Zirconium, Titanium, Barium, and Rubidium) on a series of visual bivariate graphs.  They concluded that because the data points from Stone 62 visually overlapped with the "field" of a specific small Preseli outcrop called Garn Ddu Fach, it was the definitive source.
This conclusion was inevitably tied into the attempts by Parker Pearson and others to link Stonehenge to nearby ancient sites like Waun Mawn.

 Nash and Ciborowski stepped in to challenge this subjective (and perhaps biased) visual approach. They explicitly argued that relying on a few chosen elements in a bivariate space introduces possible error. They argued that you cannot just pick two elements at a time, draw a loop around a cluster on a graph, and call it an objective match. To remove human selection bias, they developed a highly standardized mathematical method (using geometric mean multi-element ratios).

 In a defense of their methodology, Pearce et al fired back, explicitly stating that “the simple graphical approach using bivariate plots produces the clearest assignment of artefacts to sources.” They argued that  complex mathematical equations actually overcomplicate the data and create false anomalies, whereas a clean X-Y plot provides the most practical "fingerprint". They even pointed out data-formatting flaws in Nash's critique, arguing that Nash's team used rounded data numbers that distorted the true placement on the graphs.

In their latest contribution Nash and Ciborowsky push back, repeat their concerns about the subjectivity involved in selected and possibly cherry-picked bivariate graphs, and argue for an acceptance of a standard mathematical procedure based on ratios of multiple elements against a single immobile element (Zr).

We have probably not heard the last of this............

I'm not going to take sides here, but I am going to say that I have expressed many concerns in the past about the possibility for corruption to creep into bivariate graphs when they are used for "definitive" spot provenancing.  The corruption might be very suble.  Ah, the lure of banner headlines in the media is never very far away.

But I am going to gently chide both teams involved in this dispute.  They argue about Stone 62 and about Carn Goedog, Carn Ddafad-las and Garn Ddu Fach -- and other upstanding tors -- as if they are the only locations that matter:  prominent features in the landscape which must have been (they assume) the focal points in the Neolithic landscape.  But it has never been demonstrated that upstanding tors have provided more monoliths for Stonehenge or anywhere else, in preference to less prominent intervening areas.  They have not, it appears, ever thought seriously about glacial entrainment mechanisms, and persist with the ancient belief that wherever the stones came from, they must have been fetched by Neolithic stone collectors from prominent sacred places.  Big mistake.

Second, both teams have neglected to  address the issue of sampling bias. Even though Richard Bevins and others may have taken hundreds of samples from the tors and other igneous outcrops on Mynydd Preseli, there are vast expanses of igneous terrain that have never been sampled at all.  The geochemical and petrologocal analyses that have been completed -- and reported in the literature -- may represent the places that happen to have been sampled, but they cannot be taken as truly representative of the igneous masses concerned.  We do not know everything about everywhere.  As I have said repeatedly with reference to Rhosyfelin and the claim by Ixer and Bevins that they have done accurate provenancing of Stonehenge fragments to "within a few square metres" they have no idea where their sampled rock layers may crop out elsewhere, because their sampling point density is nowhere near adequate.  The same is true of the sampling of dolerite on assorted tors in the uplands.

All of the authors involved in this current spat are explicitly claiming they understand the full range of variation of an entire geological formation based on chipping away at a small handful of prominent upland rocks. They treat the landscape as if it contains only a set number of possible "addresses," completely ignoring the millions of tons of un-sampled surface dolerite and low-lying, heather-hidden outcrops such as those of Waun Mawn.

Third, on a related matter, the huge geochemical and petrological variation that occurs within the Preseli sills has not been adequately recognized or factored in to the analytical work.  I pointed this out in my published paper on Carn Goedog:

 https://www.researchgate.net/publication/390565216_Carn_Goedog_on_Mynydd_Preseli_was_not_the_site_of_a_Bluestone_Megalith_Quarry 

... and have said the same thing repeatedly on my blog.  A spot sample from a particular location on Carn Goedog may tell us a lot about that precise sampling point, but it tells us just a part of the story relating to anywhere else. In the exchange of aggressive articles by out two teams as noted above, we find this classic statement from Pearce et al:

"The calculation approach fails to take into account even the most basic petrographic observations, which in the case of Stone 62 at Stonehenge shows that it is a non-spotted dolerite, hence cannot be sourced from a spotted dolerite source such as Carn Goedog." 

Pearce et al do not seem to know that the Carn Goedog tor contains both spotted and unspotted dolerite.  It seems to me that a little fieldwork might come in handy, now and then.

So Carn Goedog is by definition a spotted dolerite tor?  Think again, chaps, and 

preferably do some fieldwork........

Finally, this is a pretty damning admission in Ciborowski et al 2026: ".... incompatible element variation (at least as elucidated by pXRF) within individual bluestone monoliths may be similar in scale to the petrological variation between individual Preseli dolerite outcrops."  And yet they persist in arguing about individual Stonehenge monoliths like stone 62, and about individual tors as places where reliable and representative spot provenancing can be done!!

This is not the first time that attention has been drawn to the variations that exist within individual glacial erratics or Stonehenge monoliths.  The same argument is currently going on with regard to the physical and chemical characteristics of the Altar Stone...........

If we have a wide range of geochemical and petrological characteristics within individual tors of sills on Preseli, and a similar wide range even within individual monoliths like stone 62, accurate spot provenancing becomes virtually impossible.  So why do assorted highly qualified scientists continue to pretend that it is not only possible but desirable?

They do say that Stonehenge makes people slightly mad.   If that is so,  those (including me!) who spend their time agonising over Bluestone 62 (or any other) are probably completely bonkers!

As far as I am concerned, if it can be shown that Stone 62 (or any other) possibly comes from somewhere on Mynydd Preseli, that is quite enough to be going on with.