Some of the ideas discussed in this blog are published in my new book called "The Stonehenge Bluestones" -- available by post and through good bookshops everywhere. Bad bookshops might not have it....
To order, click

Friday 31 July 2020

Sarsen sources and instrumental /analytical uncertainty

In the midst of all the media celebrations about the solving of the great sarsen stone mystery, I'm quite intrigued by some of the comments from quite senior academics in Twitter conversations with David Nash. There seems to be more or less universal praise for the research, which suggests to me that nobody has actually read the article properly, and that academics have simply been reacting rather lazily to the press coverage and the items broadcast on the BBC, ITV,  and Sky.

I am really the only person concerned about the following?

Ref:  Nash et al, 2020:

1.  No attempt is made to explore the possibility that the Stonehenge sarsens were all picked up from the chalklands of Salisbury Plain.  (Instead, for some reason, areas like the North Downs, the East Anglian Heights, and Dorset were treated as higher priorities.)

2.  The conclusion that 50 of the 52 Stonehenge sarsens are from one common source is not adequately supported by Figure 2 of the paper, since only some of the 260 data points (5 readings were taken from each stone) are identified, and there are many "outliers" which are simply ignored.  I have been unable to access the full data sets and supplementary materials -- I don't know why. (I don't have a problem with most of the Stonehenge sarsens having a shared geochemistry -- that's what one would expect if the stones were picked up locally --  but it would be good to know just how wide the variability is, rather than just being told that 50 of the sarsens have come from a single source.)  Here is the link, if anybody else wants to try: content/full/6/30/eabc0133/DC1

3.  The "matching" of the Stonehenge sarsens to the sarsens of West Woods is dependent upon the examination of just ONE sarsen from West Woods and just ONE sarsen from Stonehenge.  This is completely unsatisfactory, as the authors of the paper must know.  And yet they have claimed that "most of the sarsens at Stonehenge have come from West Woods."

4.  I'm worried that the plotting of the trace element data for stone 58 and for the 20 sampled sarsen sites does not involve a direct "like for like" comparison.  This is because the core samples were done on rock taken from deep within the sarsen stone, while the other measurements were from non-invasive surface readings.  I would like to know how this might have affected the information plotted.

5.  The plot of trace element ratio data for West Woods is used by the authors of the paper as the "killer fact" that establishes beyond all reasonable doubt that the Stonehenge sarsens came from this site.  And yet, when we look at it carefully, its value is very limited indeed.


On this plot, the black line plots the ratios measured from the "missing but now found" core from stone 58.  The little cross bars show the range of uncertainty in the readings.  The pink band shows the range of instrumental / analytical uncertainty for the same trace elements at the selected site.  But note that the scale of this uncertainty here is the third highest of all the sarsen stone readings (only Monkton Down and Totterdown Wood have greater uncertainties).  This means that the validity of the main conclusion (namely that West Woods provides the best match) has to be questioned.  If the scale of the uncertainty had been just a little less, then the pink band would have become a thin pink line, and so many readings would have been adrift that West Woods would not have been a favoured candidate source at all.  Just look at the figure showing all 20 sites:

With less uncertainty (ie a thin pink line instead of a broad band) at West Woods, the best candidates for the "sarsen stone source area" would be the samples numbered 3, 4, 7, 11, 12, 13, 14, 17 and 20.  That is, nine other sites would have had equivalent or better fits...............

Sarsens in West Woods (Marlborough News)

I repeat -- the conclusion that the Stonehenge sarsens came from West Woods is not adequately supported by the data.


Note added 7th August 2020:  Since this post is still getting lots of hits (590 so far, and counting) and since most readers will not have the stamina to read more than 50 discussion contributions, this is the state of play:

Point 1.  This is still a matter of concern
Point 2.  Lead author David Nash is adamant that 50 of the Stonehenge sarsens are from a common source, and that the scatter seen on Fig 2 of the paper is down to natural variation within samples.    I'm still not convinced.  Further research will no doubt reveal the truth.
Point 3.  I'm happy to correct this.  In fact THREE stones from West Woods were sampled.  I was misled by an error in the paper (now acknowledged by David) which referred to "20 field samples" (ie one from each of 20 sampled field locations) when it should have referred to "60 field samples" (ie three sampled boulders in each location).  I should have checked this out in the Supplementary Data files, but initially had problems accessing them.  Anyway, apologies for that mistake.
Point 4.  Again, my mistake.  I was misled by some ambiguous wording in the paper to believe that ALL of the sampling was non-invasive.  In fact the non-invasive sampling was done just at Stonehenge (the stones are deemed too precious to harm in any way), while the stones from other sites were investigated via rock samples taken to the lab.  I should have read the paper more carefully.  There was a "like for like" comparison of the stone 58 (Stonehenge) core with the 60 or so samples taken from other locations.
Point 5.  David claims that the West Woods trace elements ratio graphic is as close to a perfect match with the data from core 58 as it is possible to get.  He also says that a perfect match would only be proposed if ALL of the element ratios overlap on the graphic.  To me, it looks as if one of them (Uranium?) doesn't overlap, but that may be down to my fuzzy computer screen.......  At any rate, I haven't changed my view that the best we can currently say is that "Of the 20 potential sarsen provenencing sites so far investigated, West Woods provides the best match, and on current evidence it is the most likely source for the bulk of the Stonehenge monoliths."  That is very different from saying "We now know where the Stonehenge sarsens came from."

Additional Note 10th August 2020.  David is very upset again, and claims that I am misrepresenting him.  In point 2 I used inverted commas around the words "scatter" and "natural variation" -- as I often do without any disrespect or cynicism.  If the word "cluster" is acceptable, I am not sure why there is a problem with the word "scatter."  Anyway, I have now removed the offending inverted commas, and hope that that will make him happy.  Time will tell whether the 50 Stonehenge sarsens are from exactly the same source, and whether there are other variables (as yet unidentified) that affect PXRF readings. (In my time, I have seen many wonderful new techniques being used.  Initially they are all believed to provide "correct" information -- until, later on, the correction factors kick in, and interpretations and conclusions have to be revised, sometimes very dramatically........)   Next, David is upset that I mention "some ambiguous wording" in my point 4.   This is getting ridiculous, and I'm getting fed up.  Almost all papers contain phraseology that could have been better, and this one by Nash et al  is no exception.   The last paragraph of the Introduction should have referred to "rock samples taken from a representative range of sarsen boulders" to make it clear that an invasive technique was involved.  And OK -- if David tells me that there is an overlap on the graphic for the U element ratio, I'll take his word for it.

It's a bit rich that David bangs on furiously about my "misrepresentations" of his scientific research, but cannot even bring himself to apologise for his own mistake in referring in the paper to 20 field samples of sarsen when the figure should have been 60.  Just a typo?  Hmmmm...

On everything else we will agree to differ, and if David wants to say anything else he can say it elsewhere.


chris johnson said...

Let me see if I get this right....
One stone probably from Stonehenge has been matched (probably) with one stone from West Woods?
And they call August the silly season for news...

BRIAN JOHN said...

That's about it, Chris, as far as I can see........ if it isn't silly pits at Durrington, it's silly sarsens at Stonehenge..... silliness prevails.

Tony Hinchliffe said...

Timothy Darvill, he of Bournemouth, has rushed out a Bournemouth University YouTube video which is mostly him, but also has a contribution from David Nash from Brighton University, who did the sarsen scientific research. It's not long enough to reveal very much.

Tony Hinchliffe said...

I think you need to put your 5 points of uncertainty you have made in this Post direct to David Nash at Brighton University, Brian.

David Nash said...

Post 1 of 2

Brian, it is very easy to be dismissive of research via a personal blog. I want to be as transparent as possible with our results and data, so this is a long response – I have split this across two comments due to character limits. If you don’t mind, I’d like to take each of the points of your blog in turn. I’ll start with point 2:

Point 2. Unfortunately, this point contains several errors and misunderstandings. First, Fig 2 – as explained in our paper – shows the results for 250 readings taken at Stonehenge, not ‘some of the 260 data points’. To explain the data collection process, PXRF takes a reading from across a small area of a sample. In a sarsen, if you hit a rare mineral other than quartz you get an anomalous reading. These 10 erroneous readings were omitted. If you include them it makes very little difference to the statistical results (trust me, we checked) and the main cluster still holds statistically.

Second, the cluster in our PCA results in Fig 2 has extremely high statistical confidence. This indicates that most of the stones at Stonehenge share a common chemistry. The shared chemistry suggests, as you acknowledge, that the sarsens mostly come from the same place. However, that same place does not need to be local. You are correct that there are outliers. However, these are not ‘simply ignored’ as you suggest. If you read our paper carefully (including the supplementary materials, where we present results stone by stone so you can see exactly how wide the chemical variability is – we are not hiding anything), you will see that we only attribute a stone as being different to the main cluster if all five readings fall outside the 95% confidence ellipse on the figure. If one or more reading falls within the ellipse, then the chemistry of that stone still shares overlapping chemical features with the cluster. It cannot, on this basis, be considered statistically ‘different’.

An important point that you do not acknowledge in your blog is that Stone 58, the source of the core used in ICP-MS analyses, falls bang in the centre of the cluster of PXRF results identified by PCA. This means that any further chemical readings from Stone 58 can be considered representative of the rest of the cluster. i.e. they are representative of the bulk of the sarsens. It would be logically inconsistent to say that one is happy that the sarsens come from the same place because they cluster and then not be content that Stone 58 is representative when it falls in the middle of that cluster. Either the sarsens are from the same area because they cluster AND Stone 58 is representative of this cluster, or they aren’t – it cannot be either/or.

You can view the supplementary materials here if you haven’t tracked them down via my tweet:

Point 3. This is incorrect. As explained in the paper, we analyse one stone from Stonehenge but take three replicate analyses from that stone. We also analyse three samples from West Woods, just the same as we did for all our other sites. Using the argument above, Stone 58 is representative of the bulk of the sarsens at Stonehenge; so if the analyses of the core samples from Stone 58 match the analyses from the three samples from West Woods, then we can reasonably say that the majority of sarsens at Stonehenge came from West Woods.

Point 4. This is incorrect. Our analytical method is explained in the paper. Both the trace element data for Stone 58 and the sampled sarsen sites were obtained using ICP-MS and, I should stress, in the same lab on the same instrument in the same batch of samples for avoidance of any error.

David Nash said...

Post 2 of 2

Point 5. You have interpreted Fig 3 incorrectly. All natural rocks contain a degree of chemical variability, so you would expect some variability in chemical readings for stones in one site. The pink band in the figure reflects that natural variability and includes – as we explain in the paper – instrumental error (which is perfectly normal in any analytical procedure). All that the differences in the width of the pink bands indicate is that there is a little more chemical variability across the sarsens we sampled at West Woods, Monkton Down and Totterdown Wood than the other sites. This is not surprising. The three samples from Stone 58, as natural stone, also contain inherent variability. This is smaller than the range of variability at each sampling area, but we show it using error bars nonetheless as this is good practice.

The sites you mention (3, 4, 7, 11, 12, 13, 14, 17 and 20) categorically do not provide an ‘equivalent or better fit’ than West Woods. At each of these sites, at least two chemical element ratios for the core from Stone 58 do not overlap with the respective pink data range. With our method, unless every single chemical ratio from the core analyses fell within the pink band representing a site then we would not call it a chemical match. West Woods is the only of our sample areas that provides that match. Even more interesting is that the other sampling areas in the Marlborough Downs do not get close to the same degree of match.

Finally, Point 1. This is the only point in your blog where your critique has some validity. However, there is a perfectly valid explanation. As part of our initial sampling strategy, we opted to sample only areas where there are reasonable sized clusters of large natural sarsens. Salisbury Plain does not meet that criterion. We also wanted to sample widely since – as explained in the paper – making the assumption that peoples who were prepared to transport bluestones from Wales would only use local sarsen sources was inconsistent. We do, however, intend to sample more locally in future studies as part of efforts to locate the source of the only two sarsens that fall fully outside our statistical cluster – stones 26 and 160.

I hope that this clarifies things for you. I won’t comment on the personal slights you make in your last post from 18:15 on 31 July 2020 as I would prefer to focus on the data. However, rest assured, I found them offensive.

BRIAN JOHN said...

Hi David

Thanks for these lengthy responses -- I was in the middle of an Email letter to you, asking for clarification on all these matters, so now that becomes unnecessary. I appreciate you going to all this trouble. And apologies re the tone of my earlier post -- I could have phrased that better, and have dumped it. Sorry if I have upset you there. Mind you, I still think there is too much pressure from university press offices to "sex up" papers and results -- and I noted that your statements in the paper are quite careful, but are transformed in your statements to the media into something much more "definitive". Classic "interpretative inflation" as discussed recently by Barclay and Btophy. And if I may say so, we still don't "know" that the Stonehenge sarsens came from West Woods -- the best we can say is that it is "likely" or "probable." I'll consider your other points in a couple of further comments.......

Unknown said...

Hello David: welcome! You say you were "making the assumption that peoples who were prepared to transport bluestones from Wales......". I would just like to make the important point that this is indeed an ASSUMPTION, nothing more. Glacial geomorphology is an exacting and dynamic science, and I'd like to invite you to delve deeply into Brian's Blog, because this is how I came to the conclusion that glacial geomorphology is worthy of deep consideration and reflection, as is the notion that glaciers played a part in fetching, by tremendous forces, a myriad collection of geological rocks of varying shapes and sizes of so - called "bluestones" to somewhere very generally towards the then future site of Stonehenge. This to me is just as amazing as any idea that human beings moved them all that way. Please consider - this is NOT, after all, primarily a Blog about Sarsen provenancing and transport.

BRIAN JOHN said...

Ah, the joys of the personal blog! Yes, it is easy to be dismissive of research in a blog -- but it's a handy format (or at least it should be!) in which people can express forthright opinions, react quickly to breaking news, argue honestly and (in most cases) politely, and test ideas. It's also rather a good format for informing and educating people via well illustrated and brief updates on research and developments in assorted specialist fields. I know that masses of students use this blog, because there is a lot on it now -- over 2,700 posts, 1.6 million hits, and goodness knows how many comments. The search facility on Blogger is good, and mine is heavily used. Not all blogs allow comments -- but I have always seen comments and the exchange of ideas as essential. We all get hot under the collar now and then, and my irritation about certain things sometimes leads to me expressing myself in terms that might have been more diplomatic -- but I take my fair share of stick from contributors too, and accept that as part of the deal! All in all, I see the blog as a cross between a pulpit, a schoolroom, a research seminar, and a staff common room....... devoted to freedom of expression, mostly on the record.

So David's comments are on the record, and I'm grateful that he has given the time to writing them and submitting them. Lots to respond to. I have a host of things to do this morning, but will come back to this later in the day.

Tony Hinchliffe said...

"UNKNOWN" on 1st August at 10.08 was me, David.

BRIAN JOHN said...
This comment has been removed by the author.
BRIAN JOHN said...

Sorry -- spelling listakes now corrected.....

Back to David's points. First, point No 1. As the old Irishman said to the tourist who was asking the way: "if I was you, I would not start from here!" My grumble was that the researchers had completely failed to investigate the most obvious source for the bluestones -- Salisbury Plain. David says: " This is the only point in your blog where your critique has some validity." I think I might disagree with that, but we'll come to that shortly. On this specific point, David reveals a profound bias in the research. He says: "......we opted to sample only areas where there are reasonable sized clusters of large natural sarsens. Salisbury Plain does not meet that criterion." With all due respect, the reason why Salisbury Plain does not meet the desired criterion may just be because all of the usable sarsens have been collected and set into the Stonehenge monument -- and indeed, it is a perfectly valid thing to say that Stonehenge may bave been built where it is because that is where the stones were. Many people -- including Edward Pegler and David Field --have suggested that the Stonehenge sarsens may have been collected locally -- so it's not as if the idea is completely outrageous! David reveals the bias in all its beauty when he refers to the research assumption that"...... peoples who were prepared to transport bluestones from Wales would only use local sarsen sources was inconsistent." David's problem is that there is no evidence that the bluestones were transported from Wales by human beings. It has always been the classic Stonehenge myth, and should never have been used in framing a research project or in determining a research strategy.

BRIAN JOHN said...

Next, the other points, numbered 2 - 5. David points out that I am guilty of several misunderstandings, errors and "incorrect" statements. I am sorry about those, and apologise for my mistakes. In mitigation, I will repeat the point I made on the blog -- namely that I tried on many occasions to get at the "supplementary information" which explained the author's methods and provided extra data. but on each occasion I simply got the message "page not found." I have just tried again, and the data is still not accessible. Re the 250 readings and the Stonehenge information (in Fig 2) I don't think we are disagreeing that the Stonehenge sarsens have a lot in common -- ie they have a shared geochemistry --and that most of them come from one common area. The question is still "How wide is the variability, and what significance might that have for geographical distribution of spot sources?" As David has said elsewhere, there is quite a lot of natural variability in each sarsen stone. Never mind about the 95% confidence ellipse -- I am still to be convinced that the majority of the sarsens came from one location. How large? Measured in square metres, hectares, square kilometres? On point 3, I don't understand David's point. He says I am "incorrect" -- but he does not dispute that he and his colleague have based their conclusions on the matching of ONE STONE (three samples) from West Woods against ONE STONE (three samples) from Stonehenge. That latter stone was stone 58, chosen in order to be typical or representative, and because there was a handy core available. Point 4 -- OK, the same instrument and methods were used in the study for the West Woods stone and the Stonehenge stone, but I did wonder (and still wonder) whether a "fresh rock" sample should have been compared in this way with a "weathered surface" sample, since that might have given a "false positive" correlation. On point 5, David says I have interpreted Fig 3 incorrectly. Partly, that's because I changed my earlier text (which referred to the inherent variability of sarsen stone geochemistry) to talking about instrumental and analytical uncertainty, because that was the language used in the paper. At any rate, I'll accept David's point that there is a more accurate "fit" between Stonehenge and West Woods than there is between Stonehenge and any of the other 19 sites investigated. That still does not mean that West Woods is the source of the Stonehenge sarsens.

David Nash said...

Post 1 of 2

Thanks for your responses. I’m going to be relatively brief this time as you are still making incorrect statements and still being offensive, so I think it is time I disengaged. I’ll pick up on six points. You said:

1. Could it be that “…all of the usable sarsens have been collected and set into the Stonehenge monument -- and indeed, it is a perfectly valid thing to say that Stonehenge may bave been built where it is because that is where the stones were”.

This is an argument that has been rolled out repeatedly over the years, mostly by people who are not familiar with how sarsens/silcretes form. With all due respect to them, Edward Pegler and David Field are not geologists. I have published research on silcrete since 1994 so I know a little about this. The development of sarsens requires two things: 1. An area of host sands and 2. Silica carried within groundwater to cement these sands. Without both host sands and a silica supply, you don’t get sarsens. If you look at geological maps of the UK (e.g. the excellent BGS ‘Geology of Britain’ viewer at you will see that areas with sarsens today are close to or within areas underlain by sandy Paleogene sediments or their weathered remnants. Try the Marlborough area for example or the north of Brighton. To get a large area of really big sarsens, of the number and size present at Stonehenge, you’d need a really big area of sandy Paleogene sediment and these sediments would need to be really thick. Now have a look at the Stonehenge area on the BGS mapper. What do you notice? No Palaeogene sediments. No extensive areas of weathered remnants to suggest that Paleogene sediments were once there but have been eroded. On that basis, I would question whether there were ever extensive areas of big sarsens on Salisbury Plain. Put simply, no host sands = no sarsens. Even if there were sarsens, it would be incredibly convenient if every single one had been removed and used to build Stonehenge and only a few larger stones and scatters of smaller debris left behind.

2. “I tried on many occasions to get at the "supplementary information" which explained the author's methods and provided extra data. but on each occasion, I simply got the message ‘page not found.’ Brian, the link is right there on the paper home page – I also posted the link in my comments.

3. "How wide is the variability, and what significance might that have for geographical distribution of spot sources?" This is a pertinent question and one at this stage we cannot answer. We know that we have a direct match with West Woods, but we don’t know the exact location – that is the subject for future research.

4. “On point 3, I don't understand David's point. He says I am "incorrect" -- but he does not dispute that he and his colleague have based their conclusions on the matching of ONE STONE (three samples) from West Woods against ONE STONE (three samples) from Stonehenge. That latter stone was stone 58, chosen in order to be typical or representative, and because there was a handy core available.” Sorry Brian, but you have still got this wrong. Further, I did not say that. I will reiterate my previous comment. We DID NOT analyse one stone from West Woods – we analysed THREE STONES from across West Woods, just the same as we did for all our other sites. I know because I was there. I hope that this is now clear. To clarify further, we did not ‘choose’ stone 58 to be typical or representative as you suggest. Rather, we had access to a core from Stone 58 so we needed to check that it was representative of the sarsens at Stonehenge before we could draw wider inferences from its chemistry. It was.

David Nash said...

Post 2 of 2

5. “Point 4 -- OK, the same instrument and methods were used in the study for the West Woods stone and the Stonehenge stone, but I did wonder (and still wonder) whether a "fresh rock" sample should have been compared in this way with a "weathered surface" sample, since that might have given a "false positive" correlation.” I’m sorry Brian, but you are mistaken again here. For all our ICP-MS work we compared fresh stone from the core with fresh stone from the interior of sarsen blocks. This is explained on p.6 of the paper: “Any weathered outer surface material present on the samples was removed using a rock saw before dispatch to Spain”. Comparing fresh and weathered rock would, as you say, produce unrepresentative results.

6. “I'll accept David's point that there is a more accurate "fit" between Stonehenge and West Woods than there is between Stonehenge and any of the other 19 sites investigated. That still does not mean that West Woods is the source of the Stonehenge sarsens.” We’ll have to agree to differ Brian, as you clearly know a lot more about our results than I do. I’ll rehearse the argument one more time. The Stonehenge sarsens have a mostly coherent chemistry revealed by PXRF, which suggests they comes from one place (this, I think we agree on). Stone 58 sits right at the centre at the statistical cluster of these results, so its chemistry is representative of the majority of the sarsens at Stonehenge (this I also think we agree on). The chemistry of Stone 58, revealed by ICP-MS, shows a direct match with samples (plural) from West Woods (again, we agree here). Ergo, the stones from Stonehenge most likely originate from West Woods. This last point is the only one on which we appear to disagree. I still don’t understand why you don’t accept this. In archaeological provenancing studies, if the chemistry of ‘an artefact’ (in this case the core samples) matches the chemistry of a potential source area then a direct link can be made with that source area. We have a direct chemical match so we have a direct link. At this stage we don’t know which area of West Woods the sarsens came from, but – as with any piece of indictive research – that will need to be the subject of further investigation.

Feel free to comment but I will not be replying. I have tried to respond in a spirit of transparency and openness. However, you do not seem to be listening to my arguments and are still selectively misquoting information from our paper and my responses, both here and on social media. I particularly dislike the latter.

In your post on Saturday morning, you suggested that your blog is meant to be a place for “comments and the exchange of ideas”. Despite being warned to the contrary by colleagues, I decided to engage with that “exchange of ideas”. I now regret this. Blogs of this type can be incredibly useful, but to do this they need to focus on the evidence, reflect fairly on the science, and not stray into discussions of personality or engage in personal slurs. Respectfully, the tone you set through your posts, comments and the posts you approve as moderator will only limit engagement from the academic community – exactly the people who you should be attracting to discuss their results. Free speech is great but with it comes responsibility. You decide the areas to debate on your blog so you are responsible for ‘setting the bar’ for the discussion. The fact that you are frequently dismissive of research when it is clear from your posts that you don’t understand the science is very instructive. I wish you well for the future.

chris johnson said...

I would like to thank David Nash for his contributions which I have found very useful, so thanks again. My misunderstanding, based on various press reports, is that one stone had been sampled which is very probably from Stonehenge. It seems that most if not all sarsens at Stonehenge have been tested using the PXRF system and, according to the supplementary information which I did succeed to access, appear to have a clustered correlation which points to a single source. This is quite a remarkable step forward when it survives peer review by people who are a lot more familiar with PXRF than I am.
It is a shame that David will withdraw from the blog now. It is just getting interesting. I seem to remember several years ago that "Myris" was quite sceptical about X-ray analyses in general, but it seems science has moved forward or perhaps I did not understand Myris properly.

Tony Hinchliffe said...

Hear, hear,Chris. It is indeed a great shame that David has said he will withdraw from the Blog now, when his contribution and particular expertise were just getting very interesting. As a Person who used to work with the retrieval and dissemination of information for local government including environmental planning, as well as being a Geographer like David and Brian, I am very disappointed that this has occurred. Please reconsider.

BRIAN JOHN said...
This comment has been removed by the author.
BRIAN JOHN said...

(A couple of mistakes put right.....)

I cannot for the life of me see anything "offensive"in my last post, but if David sees it that way, we clearly have a problem, and if he sees fit to leave this discussion, fair enough. He's getting rather personal, with the encouragement of his colleagues -- two of whom are of course the very ones who have steadfastly refused (over a period of five years) to cite the two peer-reviewed papers written by Dyfed Elis-Gruffydd, John Downes and myself which argue that their so-called Neolithic bluestone quarries in N Pembs are not quarries at all. You do not just ignore relevant papers in academic debate just because they are inconvenient. Pots calllng kettles black. At least David has been good enough to engage in a discussion, so all due respect to him for that.

There are a number of points that I'll come back to, but I have been pondering over this "One stone or three?" issue with respect to West Woods.I have been rather confused by it, and it seems Chris might have been as well. David insists that it's there in black and white on P 6 of the article that THREE stones were analysed in West Woods. Well, yes and no. Quote: "Three subsamples of sarsen from the Phillips’ Core plus three samples from each of the 20 sarsen localities across southern Britain (Fig. 1) were processed and analyzed by ALS Minerals (Seville, Spain). Any weathered outer surface material present on the 20 field samples was removed using a rock saw before dispatch to Spain." It says "three samples", not "three stones". Then we have "20 field samples" -- that means one sample from each locality. If three separate stones had been sampled at each site, that would have been 60 samples, not 20. Then in his 21.16 post on 31 July, David says: "....we analyse one stone from Stonehenge but take three replicate analyses from that stone. We also analyse three samples from West Woods, just the same as we did for all our other sites." Again, that sounds like three samples from one stone, and not one sample from each of three stones. This is no doubt clarified in data file S1, but I am afraid it is still inaccessible..... I have been able to pick up the main supplementary data set as a PDF, but not file S1. Finally, in the media coverage, David says: "We sampled stones from 20 areas, including six in the Marlborough Downs, and analysed them using ICP-MS." ( In retrospect, David probably meant six areas, but I read it as six stones, one from each of six sites.

I respectfully suggest that what we have here is not an inadequate level of understanding on my part, but an inadequate explanation from the authors of the paper.

Tony Hinchliffe said...

As regards certain of the various names who are credited as co - authors with David Nash of the Sarsen Paper, I agree with what Brian is saying in his first paragraph of his 20.56 comment tonight. Kettles calling pots black indeed.

David should not be swayed by them, but should re - engage. This is an excellent blog.We welcome his contributions in the interests of Science.

David Nash said...

Much against my better judgement I had a look at your blog this morning and thought I should respond. A couple of direct replies to your post and then a more general observation. First, I was rather surprised that you jumped to the conclusion that I was referring to specific co-authors when I mentioned ‘colleagues’ in my comment. Sorry, but I haven’t spoken to either Tim or Mike about your blog – I was warned off by colleagues outside our team. Second, I had hoped that my explanation that we had sampled three stones at each of our 20 sites was clear. However, if we are down to the level of arguing over semantics then I am not that worried. Perhaps it’s time to move on. Third, apologies if you thought I was being personal.

Onto more general points. It is not easy to explain what I found offensive about your posts but – on reflection – I guess it comes down to the way that your comments (and those from others that you allow as moderator) question my integrity as a researcher. I’d include in that accusations of bias, implications that we are exaggerating our findings, the suggestion that we are being pressed by our institutions to over-egg our results, the notion that we are exaggerating our findings to boost press coverage, and that (because you can’t access it) we are somehow hiding our data. If you step back and look at your blog objectively, you follow the same pattern in some of your other posts. Further, you were pretty rude in your comment on 31 July that you have now deleted, and you are also rude about my co-authors. Some people might like that style of ‘debate’ but it isn’t my preference.

I’m not making any extra money out of this. If I wanted to make money, I wouldn’t be an academic. I am simply heading up a team of people who have – on the basis of some really detailed analyses and a lot of hard work – found something interesting about Stonehenge. We were sufficiently confident in our findings that we put the manuscript through tough peer review to get into one of the top science journals in the world. The media were understandably interested in our results. We have tried to communicate with them in as clear a way as possible. That isn’t easy when you are dealing with a press that wants simple messages about a complex study. There are some articles where we had more control, in that we were interviewed and allowed to correct copy in advance of publication. However, for the majority of others, we were very much in the hands of journalists and TV/radio editors.

I have enjoyed engaging with your blog because I want the wider public to understand our ‘science’. You have pointed out things that you didn’t understand about our study and – to the best of my ability – I have answered them. As I’ve said previously, we’re not hiding anything. Both Chris and Tony seem happy about this and I hope you are too. I am happy to discuss our findings further but – please – can we drop the combative tone now? Ask me straightforward questions and I’ll do my best to answer them. Bear in mind I’m a geoscientist not an archaeologist (despite what the media say), so I’m not fully rehearsed with all the archaeological nuances. Bear in mind also that – like the other authors whose work you dismiss – I’m a human being.

Tony Hinchliffe said...

David, as Bob Harris says on his introductory remarks to his weekly Country Show on Radio 2 on Thursdays, "Welcome along!!" I am very pleased.

BRIAN JOHN said...

David -- let me just extend a hand of friendship here. I have NEVER questioned your integrity as a researcher, or your well-meaning intentions. I don't think anybody else has either. As you say, we are all human, and we all get obsessed with our pet theories and sometimes allow them to "guide" or direct our research to the extent that they become ruling hypotheses. They don't seem like ruling hypotheses to us, but they may well do to other people. Bias creeps in -- not by intention but mostly by accident. We are all liable, in sophisticated earth sciences field research, to believe that some things are TRUE when they are in fact just speculations or assumptions. All I ever try to do on my blog is scrutinize things as closely as possible, as I was taught to do as a student and as I tried to encourage my students to do as well. As Carl Sagan said in a post of mine a few days ago, we need to rediscover the art of "knowledgeably questioning" -- at a time when science is dummed down all the time by the media. So I subject things to scrutiny, and I go after inconsistencies and unadmitted assumptions where I see them, maybe too robustly for some tastes, but those who don't like my style can always spend time on other blogs instead!

So rest assured that I have great admiration for the technical expertise of you and your colleagues and for the vast energy and commitment you have brought to this sarsen project. It's just that I think your statements to the press (that we now know where the Stonehenge sarsens came from) are not adequately supported by the evidence! You clearly think differently. Fair enough.

By the way, I do not deliberately take a wrecking ball to every article I read -- if you look across my blog as a whole, I think you will find that the great majority of articles are cited by me with "neutral" summaries of what they are all about -- and in many cases I like them a lot, and encourage others to read them. I WANT people to read more specialist articles about Stonehenge and the Ice Age! My faithful blog followers will probably confirm that for you.

I am a bit upset that you think I'm being rude about some of your co-authors. Mike Parker Pearson and Tim Darvill have, over the last five years, consistently refused to cite in the specialist literature the two peer reviewed papers which I wrote with two colleagues, in spite of having many opportunities to do so. Fact. Don't ask me why -- ask them. Are the papers not worthy of citation? If they are lousy papers, why were they accepted for journal publication in the first place? "Disrespectful" is just one word we might use. Am I sore about it? Too right......

Anyway, gripe over.

One simple question: when you say "Any weathered outer surface material present on the 20 field samples was removed using a rock saw before dispatch to Spain" -- did you actually mean 60 field samples?

All the best

BRIAN JOHN said...

O happy day! I have managed at last to get at supplementary data file labelled S1, with all the data for the individual stones. Very interesting. Thank you. (I don't know what the problem was -- old computer with archaic software, probably.....)

chris johnson said...
This comment has been removed by a blog administrator.
BRIAN JOHN said...

Haha! Sure, I have an hypothesis, Chris. But I try not to be ruled by it. Whatever info comes along, fine. Truth will out. But for the moment I think it's reasonable, and supported by the info we have. I'm not aware of anything that shows it's unreliable -- and of course there is no killer fact on any of the stone transport hypotheses either.

I'm not aware of any major disputes over the X-ray techniques or data analyses -- this represents a fantastic advance, for all types of stone, but I suppose that as with all new techniques (as with amino acid dating and "aminostratigraphy") at first everything looks wonderful -- until, later on, quite dramatic reassessments have to be made, because not all possible variables are known and assessed. So the correction factors take time to emerge.

Myris has gone off to pastures new, by mutual agreement.

chris johnson said...

Should we agree Brian that you too have a ruling hypothesis? The stones were all lying about at the Stonehenge site. The green ones (now called blue) arrived on a glacier from Prescelli. The sarsens were just there, in a bed of stones now stripped clean and either poking above ground level in the circle or buried under the A303. But now I am guilty of letting my imagination run away again.

Might it just be possible that the creators of Stonehenge could have transported the sarsens from 20 miles away, and the much smaller and lighter greenstones over 150 or so miles? Finally we have some evidence for the sarsens at least, while we are still waiting for more bluestones to turn up on the glacial trail.
I wish Myris was still here to comment on the X-Rays. Somebody reminding me of him put me onto a very good book by Richard Bradley recently ...

David Nash said...

To answer your question Brian, yes we cleaned up all 60 samples before dispatch to Spain. Three rocks per site (a number we’ve found gives a good representation of silcrete chemistry from our work in Botswana). I hope you’re good with this now? Some of these samples could have been sitting around at the surface (or in the subsurface) for a lengthy period of time so it was important to remove any weathering patina. My previous work on sarsens (in Sussex) shows that quite a bit of iron (and by inference other soil minerals) can accumulate in cracks and void-spaces near the surface of a sarsen over time. However, once you get 3-5 cm through the outer surface, sarsen chemistry is remarkably consistent, both vertically and laterally, within individual boulders. This is what makes them useful rocks to study – individual boulders are internally consistent in terms of chemistry, so the main variation is from place to place (controlled by the chemistry of the host sands – again we know this from Botswana).

To reassure you about methods, we've been working on the use of ICP-MS data for silcrete provenancing since 2013. There are teams of geoscientists and archaeologists working on the use of PXRF for the same ends in Australia and we communicate regularly. We've been refining the elements we scrutinise over time. We started off using purely statistical methods for the final source matching, but I don't find that very satisfactory – I like to know that the chemistry of two samples absolutely matches rather than there being a high probability that it does. For this study we have developed a quantitative approach that provides – for the first time – the ability to source match with confidence.

I can’t comment on why individuals do/don’t cite specific papers. There are potentially a whole number of reasons, not simoply the personal ones you suggest. For our paper, we were limited by the journal to a maximum number of citations. In my other papers I don’t cite absolutely everything (otherwise my manuscripts would end up being 50:50 text to citations). I normally have a ‘mental stash’ of essential must-cite papers for each study and a second-tier of ‘cite if necessary’ papers. There are papers that I don’t cite because I think they are scientifically flawed but – if you talk to most academics – they rarely cite things because they don’t agree with the authors. I know from my own experience that some journals are ‘easier’ to publish in than others. Much of this is down to editors and the quality of reviewers they select. If an editor chooses an inappropriate reviewer, or one with less expertise, then weaker papers can always slip through.

We will indeed have to agree to disagree about the origins of the sarsens. Our data from Stone 58 shows a direct match with a specific area and we know that the chemistry of Stone 58 is representative of most of the stones at Stonehenge. Our referees were entirely happy with this. The silcrete geochemists (it's a niche field but there are a few of us out there) who have emailed me since Wednesday having looked at the supplementary data are happy with this. Any archaeologist doing provenancing work would be happy about this. It sounds as though other contributors are also happy with this. I’m not disputing that some of the sarsens at Stonehenge may be local to the monument, but I doubt it will be many for the reasons I outlined on 2 August. Put simply, there is no evidence for a really large area of really thick Palaeogene host sands having existed on Salisbury Plain. Without this sand body you would not get the number and size/thickness of really large sarsens needed to construct Stonehenge. I’d love to hear your views on this.

BRIAN JOHN said...

Thanks for clarifying that you should have said "60 field samples" in the text. Too late to change it now? Thanks also for the other useful info. Interesting. I've been giving some thought to the "sarsens on Salisbury Plain" issue, and will do another post on that rather than stretching out this conversation ad infinitum!!

BRIAN JOHN said...

Funny old world. I wondered why this post has been read 475 times, and all is revealed. My tweet of the other day has been retweeted by none other than Michael Sheen (did anybody else know that he is interested in Stonehenge?) -- and he has a following of many thousands. Anyway, I have done another tweet apologising for saying that only ONE stone was sampled at West Woods -- having been misled by that mistake in the paper. Of course I should have checked the data -- I did try, but could not get the link to work. Anyway, I have now corrected the earlier tweet and have said THREE stones were sampled at West Woods. So that puts the record straight.

Anonymous said...

David Nash writes,

"Put simply, there is no evidence for a really large area of really thick Palaeogene host sands having existed on Salisbury Plain. Without this sand body you would not get the number and size/thickness of really large sarsens needed to construct Stonehenge. I’d love to hear your views on this."

The "local sourcing" of sarsens does not require these came from the geology of Salisbury Plain! These could have been brought and dumped at Salisbury Plain by natural agency (either glacial or periglacial or meltwater streams on ice surfaces). The reason why also smaller sarsens at Salisbury Plain also need to be considered.

BRIAN JOHN said...

Quite so. Instead of simply assuming human agency in the movement of the stones, i would have liked some consideration of the "natural" options -- as discussed by Peter Worsley and various others over the years. I hope that that will come, since there will have to be further research on Salisbury Plain before this matter is resolved,.

Tony Hinchliffe said...

By gum! It's all been happening on here in literally 12 HOURS, since I last looked (and briefly commented) at this Post. And, apparently, Michael Sheen and his fans have been scrutinising this sarsen Stonehenge story too..........great stuff.

As 'Anonymous' says at 22.00, there certainly are smaller sarsens on Salisbury Plain, e.g. next to Robin Hood's Ball [cursus monument]. Surely the Heel Stone too?

There is a chasm of a difference, by the way, between David Nash and The Contributor Formerly Known As Myris etc etc, who was in contrast positively cryptic. What a refreshing change!

Tony Hinchliffe said...

"Sarsen, formerly, a sandy crust produced under running water, began to break up and drift downslope during successive periglacial periods coming to rest in coombes and valleys.....On Salisbury Plain small boulders of sarsen lie in some abundance between Imber and Chitterne while a number of larger - sized stones can be found on the higher ground around Larkhill." FROM "The Making of Prehistoric Wiltshire", D Field & D McOmish, 2017, page 35 in Chapter 3, 'The Making of Wiltshire Landscape'.

Anonymous said...

Simply put, this study by Nash et al does not go far enough. As good as it may be going as far as it does! It does not settle the question of agency of transport.

Knowing more about the match or not of Stonehenge sarsens to sarsens in Salisbury Plain can help. No Nash study can be complete without this knowledge.

And if it should be that the Stonehenge sarsens match the rubbish sarsens lying around Salisbury Plain, what "human agency" and intend brought them there?

But Nature is less selective and indiscriminate!

BRIAN JOHN said...

Thanks for assorted contributions, folks! But please read my guidance notes on the landing page -- I don't like Anonymous posts, although I do sometimes let them through! On my computer one can choose to post anonymously or by using one's name. If a comment is worth making, it is worth owning too -- and I see no reason why people should not actually use their own names if they have something to say. Thank you!

Tony Hinchliffe said...

My last comment, at 00.12 this morning, continues with this, regarding sarsen boulders etc on the ground, visible to prehistoric peoples:-

"Undoubtedly,they formed a significant component of the Mesolithic landscape, difficult to negotiate in places, and they may even have been perceived to possess supernatural qualities."

The authors reference an earlier book, authored by themselves AND Graham Brown, who is the archaeologist for the Army Training Areas on Salisbury Plain. That book is an Oxbow publication and is about the Avebury landscape.

Moreover, the current book's two authors acknowledge the help of, and providing of additional comments by, the Director of Wiltshire Museum, the Wiltshire County Archaeologist, and also renowned archaeologist Jim Leary.


David Nash said...

Part 1

Thank you for your apology on Twitter Brian. Much appreciated.

Before this thread fizzles out, I’d like to know where you now stand exactly. Looking back at your original post, I think we have discussed – through a sometimes-difficult sequence of exchanges – all of your five concerns. I think we now agree that most of the sarsens at Stonehenge share a common chemistry, as shown by PXRF. We also agree that this implies they came from a common source area. We agree that Stone 58, which falls at the centre of the statistical cluster of PXRF data, is representative of the chemistry of the sarsens at Stonehenge. We also agree that the chemistry of Stone 58, as determined by ICP-MS, is a direct match with the chemistry of multiple stones at West Woods. We also agree that West Woods is the only permissible match from our natural sarsen sites. Is that all fine?

If so, the only element of our study that you appear to now be challenging is the inference made from our results that most of the sarsens came from West Woods. Reading your various comments, this appears to be because we have not (yet) established exact how large an area is covered by the ‘West Woods Geochemical Signature’, so we do not know if it extends south to include Salisbury Plain. Is this fair?

If so, I can offer you some pointers. Have a look again at our results in Figure 3 and think about the following questions:

1. Could the West Woods signature extend to the north? Answer: No. The chemistry of sarsens to the north of the Kennet (Clatford, Piggledene, sites on Fyfield Down) are different. This difference occurs over a very short distance, so we are dealing with a relatively sharp boundary.

2. Could the West Woods signature extend to the west? Answer: No. The catchment from which the Lockeridge Dene sarsens were eroded sits immediately to the west of West Woods and the sarsens have a different signature to West Woods. This difference occurs over a very short distance, so we are dealing with a relatively sharp boundary.

Now, as I suggested before, if you haven’t done so already, have a look at the BGS ‘Geology of Britain viewer’ at Click on ‘Go to Location’ in the top right-hand corner and type in West Woods. You’ll see that West Woods sits at the western end of a very large area of east-west aligned Palaeogene sediments extending out of the London Basin. These include the Lambeth Group and London Clay (which includes some sandy units) plus areas of ‘Head’ and clay-with-flints. It is the presence of this large area of thick east-west aligned Palaeogene sediments that provided the host sands for the large numbers of very large sarsens on the Marlborough Downs. As I’ve said before, we know from the silcrete literature that ‘no host sands = no sarsens’.

David Nash said...

Part 2

Now consider the following questions:

3. Could the West Woods signature extend to the east? Answer: potentially for 4-5 km within the east-west aligned area of Palaeogene sediments until you hit the Lambeth Group. To my knowledge, in situ sarsens have not been identified within the Lambeth Group (indeed, in situ sarsens have only been identified in two places in Britain).

4. Could the West Woods signature extend to the south? Answer: given (a) the geological structure of the region and (b) the complete lack of evidence for former extensive Palaeogene sediments on Salisbury Plain, probably no. It just doesn’t make geological sense. Sure, there are patches of clay-with-flints on Salisbury Plain that could reflect former areas of Palaeogene sediment. However, these are mapped by BGS as geologically much younger than the sarsens. They are also very small.

As I’ve said before, you would need a very big area of very thick sands to generate sarsens of the size and thickness of those at Stonehenge. Sarsens are a form of groundwater silcrete. We know from studies in France and Australia that these types of silcrete develop as undulating sheets or lenses of silica-cemented sand close to former water tables. Individual groundwater silcrete sheets or lenses vary considerably in thickness across their area. You would need an original groundwater silcrete layer that reached at least 1.5 m thick in places to generate sarsens of the size seen at Stonehenge. If there was once a silcrete sheet (or sheets) on Salisbury Plain that reached that thickness, then there would almost certainly be a large number of silcrete boulders that were not large or thick enough to make the grade (have a look at the Marlborough Downs as an analogue – few huge sarsens, plenty of smaller ones). Where have all those boulders gone? As Tony points out, there are some sarsen boulders and localised areas with sarsen debris on Salisbury Plain, but these boulders/debris are all way smaller than Stonehenge. Why, if there was once a large area of sarsens, aren’t farm buildings and walls full of the stuff as they are in other part of southern Britain where sarsens occur? Do the few sarsen boulders on Salisbury Plain represent a smoking gun? On the basis of the preceding argument, I very much doubt it.

If anything, I would expect the few sarsens you find on Salisbury Plain to have a chemistry more closely related to the Palaeogene sediments further south in the Hampshire Basin. Here, you are dealing with a very different set of Palaeogene sediments with very different mineralogies, so I would expect a very different signature.

In conclusion, I would argue that West Woods has a distinctive chemical signature. This may extend a little to the east from our sampling area, but it doesn’t make geological sense for it to extend northwards, westwards or southwards. The good news about all of this is that it is eminently testable. If anyone has a spare £5k we could easily characterise another 6-7 sites and analyse the various Palaeogene sediments and come up with further answers. I would test probably 3 more sites in West Woods and some sites on Salisbury Plain, but focussing on areas with sarsen debris rather than single boulders or monuments that may/may not involve human agency.

Right, I’ve spent enough time here and need to get on with my day job. Thanks all.

chris johnson said...

Very interesting David, and persuasive. As a complete PXRF layman I must admit that I look at several of the sample results your team has published from different locations and they look more or less the same. I also recall that sarsen is 99% silica and therefore identifying variations are relatively small and may deviate strongly. Not understanding the technology sufficiently I cannot judge whether 3 samples per site is sufficient to identify a common theme - presumably it is as otherwise your expert judgement would not be so confident.
I understand from people who have visited West Woods that the sarsens look plausible and large enough to have supplied Stonehenge. One person I trust greatly does believe in your science and that you have solved the puzzle. It is also on Tim Daw's putative route which is satisfying too. The sheer size of the sarsens at Stonehenge has persuaded several experts in the last century to look north for a source, the sarsens found on the plain being smaller reportedly. Why not West Woods...
I confess I was biased initially by at least one of the names lending their authority to your study being an academic of dubious quality - preferring a good narrative to solid evidence. I think I know what I am talking about here having spent years in the Prescelli hills looking for and at the supposed "evidence".
Thanks again for sharing your expertise on this blog. It is much appreciated.

BRIAN JOHN said...

Chris -- I like your honesty! Yes -- we have seen it over and again in recent years -- good basic research from the geologists and other specialists on the provenancing of both the igneous bluestones and the sedimentary bluestones, then the Altar Stone, and now the sarsens -- but as soon as the presentation of evidence and its interpretation is over and done with, along comes the discussion and conclusions, chock-full of speculative archaeological narratives which should not really have been present at all. One's instinct is to groan and say "OMG! Here we go again......" Sound research, time and again devalued by the assumptions imposed by senior archaeologists who just happen to be joint authors, presumably for "sexing up" purposes and to attract the attention of the media. (Maybe to attract research grants as well?) Nobody will admit to that happening, because of corporate responsibility for everything contained in an article............

Barclay and Brophy were, I think, spot on with their analysis of Interpretative inflation (or maybe "inflammation"):

1. presentation of data with relatively restrained preliminary interpretation in the first part of the original academic paper;

2. less tentatively, in the later part of the paper (and in the Abstract) more far- reaching interpretation, with less support offered;

3. even more ambitious claims in media releases prepared by the universities, incorporating direct quotations from the authors;

4. creation of attention-grabbing headlines and soundbites in the media by journalists working from the press releases, further amplified through interviews with the lead authors, and affected by the media outlet’s own political angle.

If we don't, who else is going to say these things?

For me, problems of credibility are compounded when some of these unnecessary assumptions spill over from the back ends of these learned papers into the front ends or introductions. The alarm bells of "assumptive research" start ringing, and I find myself -- for better or worse -- instinctively mistrusting the research methods and the results of the authors. I am sure I am not alone in this.

So my message to earth scientists writing papers on Stonehenge-related topics is this -- please stick to the science. And be very careful who you get into bed with! In the meantime, I urge everybody to read this if you have not done so already:

Gordon J. Barclay & Kenneth Brophy (2020): ‘A veritable chauvinism of prehistory’: nationalist prehistories and the ‘British’ late Neolithic mythos, Archaeological Journal,
DOI: 10.1080/00665983.2020.1769399

BRIAN JOHN said...

I must go through David's points carefully, and work out where we now agree and disagree! Will try to come back to it tomorrow.

Anyway, we are apparently agreed that when the authors of the paper said "Any weathered outer surface material present on the 20 field samples was removed using a rock saw before dispatch to Spain" they were in error, and caused considerable confusion, since they should have referred to "60 field samples". So that's all right then.

Kosta Dean said...

We can know where the sarsens came from and still not know how they got there.

BRIAN JOHN said...

Yes, that's one of my points. Far too much emphasis on wording like "brought from" and "sourced in" West Woods when more neutral wording would have been much more reliable and appropriate for a serious scientific paper.

Kosta Dean said...


If the 'rubbish' sarsens of Salisbury Plain were to match the Stonehenge sarsens, what would that tell us?

BRIAN JOHN said...

Interesting question. It could tell us several things. If the match were to be closer than the match between the 3 West Woods sarsens and stone 58, that would mean that Salisbury Plain was the most likely source of the Stonehenge sarsen monoliths. I'll use the same term as David Nash and his colleagues -- "most likely" -- because you can never be certain in provenancing studies because you cannot be 100% sure of where certain rocks outcrop, unless you have an incredibly detailed knowledge of local geology. (Look what happened at Rhosyfelin and Carn Goedog, where the "spot provenancing" claims of the geologists are undermined because they simply do not know where all the outcrops are.......) The most obvious interpretation would be that the suitable stones (mostly "rectilinear", as David points out) were collected, and the "rubbish stones" left behind. Or they were missed at the time because they were lost in the undergrowth or not exposed at the surface. If the match were to be the same as that between the West Woods stones and stone 58, again we would assume that Salisbury Plain was the main collecting area, simply because in practical terms that would have been easier. Others might argue that even if stones which were similar geochemically were available on the Plain, they might have collected them from West Woods anyway simply because they were more frequent there, or had more suitable in shapes........... but that would mean accepting the assertion that the "stone collectors" had more or less unlimited manpower resources and great technical skill. "No problem," they say. "If they could collect 82 bluestones from Pembrokeshire, this was technically a doddle....." And so we get into circular reasoning territory. Round and round we go.....

BRIAN JOHN said...

Thanks to David for his concluding (?) points. Yes, in reality there is probably not much on which we differ dramatically. On the matter of the "direct match" and the "common chemistry" of the Stonehenge sarsens and the 3 West Woods stones, which was the original cause of my concern, I still have concerns about how close the match really is -- it is not perfect, and still does not justify the statement "We now KNOW where the Stonehenge sarsens came from." I will go with your rather more circumspect wording -- saying that West Woods is the "most likely" of the sites investigated. We will have to set the technical arguments to one side until the paper has been digested and assessed by other researchers who are familiar with the techniques used here, the data analysis methods, and the graphic representations of the results. That may, I suppose, take some little time.......

Re the extent of the "sarsen outcrops", yes, I agree that the sarsens to the north look different geochemically, which might lead one to suggest that the further south you go, the more like the Stonehenge sarsens the sarsens in the ground become. (This might be a very important point.) So yes, the West Woods sarsens become the best available match. I have looked at the BGS viewer maps, and they are very interesting. Thanks too for the useful info on the characteristics and extent of the Palaeogene beds. Yes, I accept the point that sarsens with the same or very similar chemical signatures could extend to the east of West Woods.

I take on board David's points about huge rectilinear sarsens like those at Stonehenge probably not occurring naturally on Salisbury Plain, but the jury os still out, as he admits himself. There are abundant records of sarsens (including big ones) on the Plain, and we cannot assume that every detail of local geology is already known and represented on the BGS maps. (I know of several places in Pembrokeshire where rocks outcrop where they are not supposed to be present......) So I will maintain my working hypothesis that Stonehenge was probably built of sarsens picked up from the neighbourhood -- and as David says, this issue will not be resolved until further work is done.

Quote: "If anyone has a spare £5k we could easily characterise another 6-7 sites and analyse the various Palaeogene sediments and come up with further answers. I would test probably 3 more sites in West Woods and some sites on Salisbury Plain, but focussing on areas with sarsen debris rather than single boulders or monuments that may/may not involve human agency." Three cheers for that. Any wealthy benefactors out there? Michael Sheen maybe?

Not much more to be done for the time being. But it would have saved us all a lot of fuss and bother if the Salisbury Plain sarsens had been looked at from the beginning, rather than as an afterthought.

BRIAN JOHN said...

I'm inclined to terminate this discussion here, since we are now going round in circles. If there are any other points, they can maybe be linked to another post which I'm currently working on. Thank you all for a very stimulating chat!

David Nash said...

Swift response 1:


I’m glad you’re persuaded by my arguments and that that ‘one person you trust greatly’ is also persuaded. A number of geochemists that I trust greatly who have looked at our data are also persuaded. I might add, they also like our new method for quantitative ‘wiggle matching’.

Your observation about some of the analyses looking similar is understandable given the scale of the graphs in Figure 3 (you try fitting 20 graphs with identical axes on a single figure). Visually some appear to match. However, if you look at the data you can see that only site 6 (West Woods) directly matches for all chemical element ratios. I explain this to Brian in my comment at 21:16 on 31 July (headed ‘Post 2 of 2’). We would only claim a direct match if ALL element ratios overlapped – you can’t be a little bit out on a couple of elements and claim a match.

Your point about sarsens being 99% silica is absolutely correct (Stone 58 is 99.7% silica). That’s why we used ICP-MS (not PXRF) to do our detailed matching of Stone 58 to potential source areas. ICP-MS allows you to test for a much wider range of elements with greater precision for many elements. PXRF was only used to test for in-site variability at Stonehenge.

Our use of 3 samples is based on experience of working on silcrete geochemistry in the UK and Botswana. Stewart Ullyott and I have made forensic analyses of sarsen boulders from Sussex (also developed in Lambeth Group sediments and with a similar petrography to those at West Woods). This involved analysis of multiple samples along transects vertically through and laterally across a cut stone. The chemistry is remarkably consistent throughout so it is safe to assume that a sample from a stone is representative. From work in Botswana we’ve found that once you get above 3 samples from a silcrete area the mean chemistry varies very little, so 3 is a safe number of samples to encompass in-site variability.

David Nash said...

Swift response 2:


You’ve made your point about interpretative inflation (I think) three times now. I get it. You’ve also spotted a typo. Well done. However, as you say, if you had looked at the data spreadsheet from day 1 (or indeed day 3 when I included the link in my first comment) you would have known this wasn’t a problem.

But enough tetchiness. Your point about the chemistry of stones on the Salisbury Plain is an interesting one. However, if we test some sarsens from Salisbury Plain and find they match as well as those at West Woods (the match can’t be ‘closer’ – we are dealing with absolutes here – they are either permissible matches or they aren’t) then all we know with confidence is that the West Woods geochemical signature extends south onto Salisbury Plain.

We are dealing with hypotheticals here, but I would not then (as a geoscientist) automatically take the ‘least effort’ approach often assumed in archaeological theory. From my research in Botswana I know that these models don’t always work – we’ve found that Middle Stone Age folk transported silcrete 300 km to make stone tools when there is equally good silcrete 70 km away. Rather, I would look at other properties of the stones in the area covered by the ‘expanded West Woods geochemical signature’ and compare them against equivalent properties for the Stonehenge core sample to see which matches best.

In parallel with this study, we are working on a second paper purely on the Stonehenge core samples. You’ll like this one because there is less archaeological interpretation. We have run a huge variety of state-of-the-art analyses on the core samples precisely so that scientists in the future can use the data for sourcing. Seriously, this 67-mm-long sample of core is going to be the most-analysed piece of stone other than moon rock. We have looked in detail at the petrography (thin-sections, SEM, cathodoluminescence), mineralogy (QEMSCAN) and geochemistry (XRF, micro-XRF, ICP-MS, ICP-AES, stable isotopes). We have only reported on geochemical data so far but will be submitting the other results to an open access journal later this year. We will also be archiving the full data set for free public access (this is something I’m passionate about).

An obvious question is ‘why haven’t you run these analyses on the West Woods samples’? A short answer is because they are bloody expensive – much of the analyses represent ‘good will’ on behalf of the labs involved because they recognise the importance of the core as an artefact. However, if we get more funding, we will do this, and expand our sampling around West Woods and onto Salisbury Plain. However, whatever more we find out, there is no disputing that we have a direct chemical match between Stonehenge and West Woods. That cannot be taken away.

David Nash said...

Thanks Brian - you make some valid points here. I don't agree with your assertion that the match between West Woods and Stonehenge is 'not perfect'. All the chemical ratios match within the limits of instrumental error and you will never get better than that. However, I now have around 5,000 words towards a paper on 'Was there ever a large silcrete sheet on Salisbury Plain' - so thank you everyone. After a rocky start (excuse the pun), I've enjoyed the debate and am glad of the opportunity to explain our science. You see, with a bit dialogue and communication, we can resolve initial differences. Definitely time to put this to bed now.

BRIAN JOHN said...

I thought we were finished, but my tolerance knows no bounds! Now we'll leave this in the anticipation of future work on chemical matching. End of thread.

David Nash said...

Brian, I thought we had ‘closure’ on this and yet you add an update to your blog post that shows that you still don’t fully understand some of the points you are making. I am absolutely not leaving this as the last word – and every time you post something that is misleading or incorrect about our work, I will correct you. I am sorry – but I will not have you misrepresenting scientific research when you clearly do not understand it.

Your point 1: We will have to agree to differ on this.

Your point 2: I hate to do this as I will come across as an arrogant academic, but I have been working and publishing on silcrete/sarsen petrography and geochemistry since 1994. It is one of my main research specialisms. I have co-authored the most highly cited review on silcrete (Nash & Ullyott, 2007). I KNOW that if you analyse sarsen with a PXRF you will get some scatter on the results. This is due to natural variation. There is no need for the inverted commas. You are wrong. Please correct your wording here.

Your point 3: Thank you.

Your point 4: Could you point me to the “ambiguous wording” please? It couldn’t be any clearer in the paper. Opening line of the Results section on p.2: “Nondestructive chemical analyses of all 52 sarsens present at Stonehenge were undertaken using PXRF”. Regarding the analysis of sarsen samples from across southern Britain (bottom p.4): “Each of these samples was analyzed by ICP-MS/-AES using the same analytical protocol as applied to the Phillips’ Core samples from Stonehenge”. Please correct your wording here as your comment is misleading.

Your point 5: First three sentences. I hate to have to do this again, but you clearly do not understand the analytical procedures involved in collecting and interpreting geochemical data. I explained your point repeatedly in my responses to your comments and I will do it again. Look at the data as it will answer your concerns about Figure 3. In our method, we ONLY attribute a direct match between Core 58 and a potential source area if ALL element ratios overlap on the graphic. For West Woods, and nowehere else, ALL element ratios (even U) overlap on the graphic. You are wrong. Please correct your wording here.

Your point 5: Last two sentences. We will have to agree to differ on this.

BRIAN JOHN said...

I have made corrections on the post, and my patience has run out. No more comments will be allowed on this thread. Time to move on.