The BGS lead "isoscape map" -- a crude sketch map dressed up as a piece of high precision cartography?
Quite apart from the issue of proposing -- on the basis of extremely thin evidence -- that the famous cattle tooth found at Stonehenge came from Pembrokeshire, there is the question of the reliability of the lead "isoscape" map published by BGS, used by Evans et al, and assumed to be accurate. But how accurate is it?
It comes from an article by Jane Evans and others published in PLOS ONE journal in 2022. Here is a quote from the recent tooth article:
The variation in Pb isotopes across Britain is based on lead mineralization and the isotope composition is related to the timing of major tectonic events. This provides a broad-brush subdivision of Britain into three major Pb tectonic zones with the addition of Chalk/Limestone as a fourth zone (Evans et al., 2022b). The applicability of these rock and mineral zones to biological tracking is in its infancy and factors such as seawater/rainwater contributions are not yet assessed.Evans JA, Pashley V, Mee K, Wagner D, Parker Pearson M, Fremondeau D, et al. (2022)
Applying lead (Pb) isotopes to explore mobility in humans and animals.
PLoS ONE 17(10): e0274831. https://doi.org/10.1371/journal.pone.0274831
And how were the contour lines between areas of differing lead isotope signatures drawn? According to the relevant figure caption in the PLOS ONE article, contouring is based on Inverse Distance Weighting (IDW) with ten natural break intervals. What does this mean? Well, as I understand it each data point is used as the centre of a circle, and a determination is made regarding the circumferance within which values are probably similar. If there is a high density of sampling points with similar signatures, then the map can with a reasonable degree of confidence be given in the appropriate colour.
1. An IDW map is highly unreliable in areas with a low density of sample points. The method's core assumption—that local influence is the dominant factor—breaks down when there aren't enough local points to provide a meaningful average.
2. In areas with sparse data, such as South Wales, the prediction is essentially a guess, and you have no way of knowing how much you can trust it.
So there is an admission that the division of Britain into four "major Pb tectonic zones" is rather broad-brush and "in its infancy". In other words, the method is rather crude and untested. And yet it is being used by Evans et al (2025) in a piece of "high precision provenancing" to link one cattle tooth to Mynydd Preseli in Pembrokeshire..!!
So how was the lead isoscape map produced? The first thing that needs to be said is that it based on a remarkably small number of samples or data points (total 633?). It appears that there are none in Pembrokeshire, only three in South Wales and only three in the English Midlands.
Many of the data points and signature values come from an article by Blichert-Toft et al in 2016.......
We don't have any names for the four sites in South Wales, and neither do we have any of the "signature" values. I have tried to track them down, but we don't even know which article they came from (multiple citations are not very helpful when it comes to finding raw data). And some of the potential source articles are behind paywalls -- thus inaccessible.
The precise positioning of your contour can I suppose be inserted by your computer. But major errors can occur if a sampling point -- or many sampling points -- are located close to the edge of a significant geological outcrop such as that marking the edge of British Lower Palaeozoic rocks. In that case it will clearly be foolish to assume that the rocks within a circle centred on any particular data point will have the same signature. Some of them might, but most of them might not.........
Since there are no data points in Pembrokeshire, the colouring of the map must be based on the signature of rocks either in Ceredigion or Carmarthenshire. But in those counties the rocks are mostly sedimentaries and are very different indeed..........
On the lead isoscape map (at the head of this post) South Pembrokeshire is given a different value from North Pembrokeshire. The line between the two zones roughly coincides with the boundary between the Lower Palaeozoic and Upper Palaeozoic rocks in the county, and the line must have been inserted by hand, based on an assumption of two different signatures. That is not very scientific..........
All in all, I have a bad feeling about this map, and I have serious concerns about its scientific value. On checking up on the literature, I found the following points:
1. An IDW map is highly unreliable in areas with a low density of sample points. The method's core assumption—that local influence is the dominant factor—breaks down when there aren't enough local points to provide a meaningful average.
2. In areas with sparse data, such as South Wales, the prediction is essentially a guess, and you have no way of knowing how much you can trust it.
3. The IDW method cannot predict values higher or lower than the maximum and minimum measured values. How many of your sampling points were in "characteristic" locations? Many of them might have been in "exceptional" locations, for a variety of reasons. In sparsely sampled areas, this leads to an underestimation of high values and an overestimation of low values, as the interpolated value gets pulled toward the average of all available points.
4. Spatial autocorrelation is an issue. IDW assumes that nearby data points are more similar than distant ones. While this is useful for some isotope systems, it can be a problem for isotope systems that vary sharply or are heavily influenced by discrete geological features rather than gradual spatial changes, which can be the case for lead.
------------------------
To summarise: since the lead "isoscape" map used by Evans et al in the provenancing of a Stonehenge cattle tooth to Mynydd Preseli has no data points in Pembrokeshire, and uses highly questionable cartographic techniques, no precision is possible. So the provenancing is essentially worthless.
=====================
PS. In contrast to the very dodgy piece of mapping discussed above, here is an example of a good map. It is based on many thousands of data points, many of them described in the literature on multiple occasions over more than 200 years. This version is quite old, and there are frequent minor corrections, as there should always be in cartography.
I guess from a glance at the lead isotope map, the cow could just as easily have come from Southern Scotland. Don't I remember some evidence of pig (was it?) that was claimed to come from Orkney, or other similar northern places.
ReplyDeleteThe cow could just as well have come from a host of other places. SW Wales was only chosen by the authors because that is where the bluestones (or most of them) came from. i'm not sure that there have been any claims of pigs or other beasties comingv from Orkney to Stonehenge -- but a few years ago there was a huge row about the significance of the evidence of animal remains from Durrington Walls. That was to do with strontium isotope signatures. The spat got really nasty -- for daring to criticise the research, Scottish archaeologists Barclay and Brophy were viciously attacked by Madgwick and Co, in language that got dangerously close to defamation. I don't know all the details, but a court case was narrowly avoided.
ReplyDeletehttps://brian-mountainman.blogspot.com/2021/05/the-isotope-analysis-debate-gets-vicious.html
https://brian-mountainman.blogspot.com/2021/12/barclay-and-brophy-fight-back.html