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

Monday, 21 June 2021

More on the Stonehenge dolerites — multiple sources and no quarries

One of the unspotted dolerite samples from Stonehenge probably came from near here. This is Cerrigmarchogion.  The other dolerites — spotted and unspotted — came from many different places.

There’s a big new paper from Bevins, Pearce and Ixer. As with their other papers, the intention seems to be to find common sources for multiple samples from Stonehenge, and to identify the locations of “bluestone quarries” — but with every paper they publish they inadvertently demonstrate that almost all of the samples are unique and that they have come from different places.  This is exactly what we would expect of a collection of glacial erratics.  The idea of bluestone quarrying is dead, and it’s about time the geologists admitted it.

 Anyway, here are the details of the new publication.  It’s difficult to do a proper review right now because of the constraints of using an iPad in the wilds of Sweden, rather than my very versatile iMac at home. (Copying and pasting is difficult on an iPad.......). It’s a highly technical paper intended for specialists, and it continues the efforts by these three authors to study very intensively all the categories of rock found at Stonehenge and its environs, and to find provenances for all of the samples on the record. Unfortunately, there is no new fieldwork here — it appears that the sampled rock fragments and cores are the same ones that have popped up over and again in the literature from Bevins, Ixer and Pearce, mostly collected more than 30 years ago. Those samples have effectively been worked to death, and one has to wonder why there has been no new sampling around Mynydd Preseli which is more suited to the “bluestone debate” as it currently stands.

There are a couple of things that significantly detract from the value of this new study.

First, as mentioned above, the underlying assumption on the part of the authors that the bluestones at Stonehenge were quarried and transported by Neolithic tribesmen from the Preseli area to Stonehenge, rather than being collected up and used more or less where found. That forces them into other assumptions, including the assumption that two quarries have already been found, and the assumption that there are other quarries waiting to be found. This forces them, in their interpretations of the evidence, to assume a modest number of sources, whereas for any independent reader the evidence clearly points to multiple scattered sources, as one would expect of a collection of glacial erratics.

Second, the authors still refuse steadfastly to acknowledge that their ideas about quarrying and stone transport are hotly disputed in my book on the Stonehenge bluestones and in two peer-reviewed papers. I have said it before, and I will keep on saying it — this is reprehensible and is tantamount to scientific malpractice. I’m amazed that the referees and the editor responsible for the publication of this paper did not insist on proper careful citations of “inconvenient” papers and a consideration of pros and cons in the discussion and interpretation of the evidence.


The accurate identification of the sources of stones used in the construction of stone circles has the potential to play an important role in understanding the movements of people in ancient times, having a particular relevance and potential significance when long-distant transport has been involved. Tracing of sources to particular rock outcrops provides the opportunity for focussed archaeological excavations which might inform questions such as why particular stone sources were selected and exploited, as well as potentially revealing material evidence as to how the stones were extracted and subsequently transported from site. In the context of this paper, recent detailed provenancing of particular Stonehenge bluestones (see Ixer and Bevins, 2010, Ixer and Bevins, 2011, Bevins et al., 2011, Bevins et al., 2012, Bevins et al., 2014) has led to the discovery of two Neolithic quarry sites in the Mynydd Preseli area of north Pembrokeshire, at Craig Rhos-y-Felin (Parker Pearson et al., 2015b) and Carn Goedog (Parker Pearson et al., 2019). It has also recently been suggested that some of the bluestones may have been part of an earlier stone circle in the Mynydd Preseli area, at a location called Waun Mawn, which was partially dismantled, with some of the stones transported to Stonehenge (Parker Pearson et al., 2021).

So this article is hugely devalued by the ongoing adherence of the authors to a very silly ruling hypothesis which should have been abandoned years ago.  It’s also devalued by a lack of comparative sample analyses; I should like to have seen some analytical data from unspotted dolerites in other parts of the UK.    How similar, or how different, are they from the analyses presented in this paper?

All that having been said, it is of some interest in that it brings a new technique to the table — the analysis of rare earth elements.  It is suggested that one Stonehenge sample (SH45) probably comes from the Cerrigmarchogion area, but the other unspotted dolerite samples from the Stonehenge area are difficult to fix.  They are most likely to have come from eastern Preseli, around Carn Ddafad-las.  But the samples are different — they have not all come from the same place.  Sample SH44 is an anomaly — unlike anything else found at Stonehenge and different from all of the Preseli unspotted dolerites sampled.  

So once again, as with the studies of spotted dolerites, sandstones, rhyolites and ashes, the conclusion is that there are no preferred and identified sources which can be identified as quarries, but rather multiple scattered sources, not one of which has yet been “nailed down” by hard evidence.  I have made the point over and again over the last decade that there are around 30 different sources for the Stonehenge bluestones and the “bluestone debitage” — and every geological study published by Ixer, Bevins and their colleagues supports this contention.  They claim to have fixed some foliated rhyolites as having come from a quarry at Craig Rhosyfelin to “within a few square metres”, but the presented evidence does not support that claim.  They also claim that there was a spotted dolerite quarry at Carn Goedog, with “evidence” that is even more equivocal.  It is high time that they faced up to reality, as presented in their own papers.  


Revisiting the provenance of the Stonehenge bluestones: Refining the provenance of the Group 2 non-spotted dolerites using rare earth element geochemistry

Bevins, RE, Pearce, NJG and Ixer, RA

Jnl of Archaeological Science: Reports Vol 38, Aug 2021, No 103083.


The doleritic bluestones of Stonehenge, sourced from the Mynydd Preseli in west Wales, have been previously classified into three geochemical groups on the basis of compatible element geochemistry (Bevins et al., 2014). The majority of Group 1 (spotted) dolerites were identified as coming from the outcrop of Carn Goedog, Group 3 (spotted) dolerites were linked to the outcrops Carn Breseb, Carn Gyfrwy, outcrops in the vicinity of Carn Alw and an un-named outcrop west of Carn Ddafad-las and Group 2 (non-spotted) dolerites were identified as coming from either Cerrigmarchogion or Craig Talfynydd. A sub-set of the samples used by Bevins et al. (2014) have been re-analysed by solution nebulisation ICP-MS, including analyses of the rare earth elements (REE).

Analysis of the REE data reveals that Groups 1 and 3 dolerites from both Stonehenge and the Preseli have very similar REE patterns which strongly suggests that they are derived from a single intrusive body. Group 2 non-spotted dolerites are now divided, on the basis of their REE contents, into four Preseli and two Stonehenge sub-groups, (Groups 2i-2iv and Groups 2v-2vi, respectively) while Stonehenge orthostat sample SH44 plots apart from all other Stonehenge and Preseli samples in all discriminant diagrams used. The new data show that Preseli Group 2i dolerites have very distinct concave down “humped” patterns and bear no resemblance to any analysed Stonehenge dolerites. The source of Stonehenge Group 2v dolerites remains equivocal; they plot close to Preseli Group 2ii dolerites from Carn Ddafad-las and Garn Ddu Fach and have in common the presence of notable positive Eu anomalies, but they show minor differences, especially in relation to their Gdn/Ybn ratios. However, Stonehenge orthostat sample SH45 shows a near identical REE composition to Preseli Group 2iii dolerites from Cerrigmarchogion.

In terms of the interpretation of REE contents and chondrite-normalized patterns we found no differences whether using the ‘standard’ techniques used by geochemists, based on chondrite-normalized elemental ratios and values, or the quantitative approach using shape factors derived from polynomial curve fitting.

Saturday, 5 June 2021

Outwash gravel sheets in Central Pembrokeshire


Did a braided river like this once flow from the Trefgarn Gorge exit and then southwards across the site of Haverfordwest?

Extract from the geology map showing the main gravel occurrences in pink.  The present course of the Western Cleddau river can be picked out by the yellow band on the image, showing where the present flood plain is located.

I have long been intrigued by the extensive sheets of glaciofluvial gravel found in the Haverfordwest area, especially on the "plateau" incorporating Prendergast, on the NE edge of the town, and running up through the site of the Withybush Aerodrome (where of course they are massively disturbed) towards Rudbaxton and thence towards the southern exit of the Trefgarn Gorge.

There was an outcrop of these gravels on the bank that dropped away beneath Cherry Grove, where I lived as a child, and I was intrigued by them even at that tender age! 

But why does this "gravel sheet" not run along the present route of the Western Cleddau river and to the west of the A40 road?  Was there a braided outwash river to the east of the road?  That would have been a strange situation, since much of the land to the east of the road is above the 50m contour whereas much of the land to the west of the road is below 50m.  Or might the river have migrated westwards and removed vast quantities of glaciofluvial material in the process?

Various attempts have been made to recognize river terraces in the Haverfordwest area, but they have never been very successful, partly because of the extent of the built-up area and partly because terrace remnants are small and difficult to identify.  Traditionally, the gravels in the Haverfordwest area have been interpreted either as pre-glacial river gravels laid down at a time of higher sea-level, or as glaciofluvial gravels associated with one of the earlier glaciations (Anglian?).  This would make sense, if the Gwaun-Jordanston channels are assumed to have been formed by vast torrents of meltwater flowing under the ice and flowing first westwards and then southwards, through Trefgarn Gorge and towards Milford Haven.  This too makes sense, with an ice surface gradient sloping down from NW towards SE.   If the gravels are stained and even rotten -- and hence very old -- this would support the Anglian glacial hypothesis........ and the assumption that Devensian or LGM ice could not have affected central Pembrokeshire.

On the other hand, I have been thinking for some time that Devensian ice did not simply skid to a halt along the coast of St Bride's Bay, but pushed well inland.  To his credit, Prof Dai Bowen was the only person to have suggested this in his Devensian / Weichselian ice limits maps:

Recently some big exposures have been opened up in association with the building of a new Haverfordwest High School on the site of the old Sir Thomas Picton School.  Near Prendergast Cemetery and the County Archives building there is a huge mound of excavated gravels, incorporating large boulders which suggest to me either a very powerful meltwater torrent or the close proximity of glacier ice.  Sadly, I have not been able to examine any in situ gravels which might indicate the direction of meltwater flow.........

Another braided river plain (sandur), showing the complexity of anastomosing channels and suggesting the frequent lateral displacement of the dominant water routes.  We should not strictly refer to this as a "flood plain" since the "flooding" is going on for much of the time during the melting season, here, there and almost everywhere........  Multi-channel rivers like this are a nightmare to cross on foot with a heavy pack!

Kaldalon braided river and sandur in NW Iceland.  We crossed this one many times in 1960, usually in the middle of the "night" when the water level was low.

When I examined the apparently fresh glaciofluvial gravels at Picton Point, I became convinced that they were Devensian in age, and that they were laid down nor far from an ice edge:

It would be logical for the gravel sheet north of Haverfordwest to be the same age, and to have formed in similar circumstances.

If we look at the topographic map of north Pembrokeshire we can see where the main drainage routes are.  The Gwaun-Jordanston meltwater channel system has flummoxed researchers for well over a hundred years, but most people nowadays accept that they are very old, having been cut (and then modified) by sublgacial meltwater during several glacial episodes.  But they must have been used by huge torrents of meltwater at the end of the Devensian (LGM) glaciation as well.  If Lake Teifi and the other features in the Teifi Valley were created in the WAXING phase of the LGM ice advance, that makes it quite likely that meltwater escaped westwards through Cwm Gwaun and the other big rock-cut channels prior to the LGM and after it as well.  All very confusing.  That means a complex history of meltwater flow and landform development, between 30,000 and 20,000 years ago.

Main meltwater routes used during the Late Devensian.  It is assumed that meltwater flowed westwards and then southwards through Trefgarn Gorge when the ice was far advanced, and then escaped northwards into Cardigan Bay once the ice edge had retreated to the north of the present coastline.

So where should we draw the Devensian ice edge at the time of the LGM?  I am inclined to think we are talking of retreat phases here, rather than a terminal of "end moraine" position.  Watch this space.......

Rock avalanche

A fabulous image.  A rock avalanche onto the surface of Scud Glacier, British Columbia, June 2020.