Journal of Archaeological Science
Richard E. Bevins, Rob A. Ixer, Nick J.G. Pearce
Journal of Archaeological Science
Available online 19 November 2013
In Press, Accepted Manuscript
Some thoughts on the new paperThis seems to me to be an excellent paper, involving meticulous research and careful presentation of the results. It is highly sophisticated, and because most of it is really only going to be comprehensible to a small group of igneous petrologists / geochemistry specialists it may not have a very wide readership! It will be much cited, but seldom read, as is the way with detailed research........... So let's try to summarise what the main findings are, given the highly simplified nature of the authors' press releases and the banal coverage by the media.
Below I reproduce a few of the key statements from the text. But the following points are important:
1. There is no new fieldwork here. The authors have not been doing any new sampling at Stonehenge, and nor have they been furiously collecting new samples on Mynydd Preseli.
2. The authors have re-analysed samples already collected over past decades by Richard Bevins and others, and by Olwen Williams-Thorpe and her colleagues prior to the publication of this big report: Thorpe, R.S., Williams-Thorpe, O., Jenkins, D.G., Watson, J.S., Ixer, R.A., Thomas, R.G.,
1991. The geological sources and transport of the bluestones of Stonehenge, Wiltshire, UK. Proc. Prehist. Soc. 57, 103e157.
3. A total of 68 samples have been re-analysed using very sophisticated techniques: 22 samples from Stonehenge (from monoliths and debitage) and 46 samples of Preseli dolerite -- spotted and unspotted. It's worth noting that only 3 of these samples have come from Carn Goedog (which is a very big hillside tor stretching across several hundred metres of terrain.)
4. It is concluded that 11 (and possibly 12) of the 22 Stonehenge dolerite samples have probably come from the Carn Goedog outcrops. These are called Group 1 samples.
5. Three of the Stonehenge samples have possibly come from the Cerrigmarchogion area, on the crest of the Preseli ridge and some way to the south-west of Carn Goedog. These are called Group 2 samples.
6. Five of the Stonehenge samples and the Boles Barrow sample have come from outcrops in the Carn Gyfrwy-Carn Breseb-Carn Alw area (including the un-named outcrop west of Carn Ddafad-las). These are called Group 3 samples.
7. One sample from Stonehenge (numbered SH44) has not been assigned to any of the three groups mentioned above.
8. There is a possibility that Group 1 and Group 3 dolerites might be closely related -- and may ALL have come from the Carn Goedog outcrop. By the same token, some of them or most of them might have come from some other as yet unsampled part of the same intrusion, which stretches cross-country for about 3 km before being truncated by faults at both ends.
9. Carn Meini -- for long assumed to be the location of THE bluestone quarry -- does not figure at all in the new provenancing work, and we can see from the detailed mineralogy plots that the Carn Meini samples (around 18 of them?) seem to fall outside the ranges of the three Groups of Stonehenge samples analysed by the authors.
10. The dolerite sills in this part of eastern Preseli are highly variable or "geochemically heterogenous." Also, they have still not been adequately mapped, and whilst there seems to have been quite intensive sampling of the Carn Meini outcrops in the past, there is still a shortage of samples from the Carn Goedog outcrops. Expect more surprises when more fieldwork is undertaken!
11. So has the Carn Meini Quarry now been replaced in the minds of archaeologists by the Carn Goedog Quarry? If it has, they should think again. There is a lot of variation in the Stonehenge samples, so even the eleven Group 1 samples have NOT all come from the same precise location. Overall, we are looking at Carn Goedog rocks that have possibly come from an area several hundred metres across -- and quite possibly some of them have come from outcrops or source areas as yet unsampled.
12. The 22 Stonehenge samples analysed in this paper have probably come from 22 different locations. Some of the samples show strong similarities, but each one is unique. What the geologists have done here is to have narrowed down the provenancing of the sampled Stonehenge dolerites to three or four areas within this landscape of dolerite sills in eastern Preseli -- but there is no way we can talk about point locations or quarries.
13. When this is all put together with the work that the same authors, and others, have done on the rhyolites and sandstones found in the Stonehenge bluestone and debitage sample assemblage, it still seems to me that we are looking at an assemblage of glacial erratics that have come from many different places -- but preferentially from eastern Preseli -- on the route of the Anglian Irish Sea Glacier.
14. Remember that there are 43 known bluestones at Stonehenge. There is also a vast area of debitage at Stonehenge that has not been excavated or sampled. The authors of this paper look forward to the day when EH will allow ALL of the bluestones to be sampled and analyzed -- and will allow the "unexplored" parts of the Stonehenge regolith also to be sampled. Expect further surprises -- but maybe not in our lifetimes.........
15. Many thanks to Rob Ixer for sharing this paper and making it possible for me to conduct this review of the main findings. He and his fellow authors will will no doubt correct me if I have got anything wrong!!
Extracts:From p 2:
Bevins et al. (1989) provided an account of the dolerites exposed
in the area between Fishguard and the eastern Mynydd Preseli,
identifying them on field, petrographical and geochemical evidence
as representing a suite of intrusive doleritic sills which were
emplaced at a high crustal level and which are the lateral equivalents
of basaltic lavas (and associated sub-volcanic doleritic sills)
comprising the basic member of the Ordovician age Fishguard
Volcanic Group, the major expression of which is exposed further to
the west (Bevins, 1982). Bevins et al. (1989) suggested that the
basaltic magmas were erupted in a submarine environment in a
graben or half-graben structure centred in the Fishguard to
Strumble Head area, with the magmas being channelled up the
bounding faults. The Preseli district was peripheral to this graben
structure and accordingly the magmas were emplaced as high-level
sills in the adjacent sedimentary sequence.
From p 3:
In this paper we turn our attention to a re-examination of the
geochemistry of the various spotted and non-spotted dolerites from
the Stonehenge Landscape, including samples from the monoliths
themselves and debitage from various other locations including the
Heelstone Ditch, the Avenue west ditch, Aubrey holes 10 and 22, and
hole Y6. An analysis of a dolerite sample from Boles Barrow is also
included. We review these data, published previously by Thorpe
et al. (1991) but with the addition of one new analysis for monolith
34, and compare them with the eastern Mynydd Preseli dataset
of Bevins et al. (1989) and Thorpe et al. (1991) along with some
previously unpublished analyses by Thorpe and colleagues. In this
re-examination we take a different approach to interpreting the data
from that adopted by Thorpe et al. (1991) and we reach alternative
conclusions. The dataset utilised in this paper comprises analyses of
22 Stonehenge monolith and debitage samples and 46 Preseli
From p 4:
Thorpe et al. (1991) concluded that they could
determine three different sources, arguing that:
most of the Stonehenge dolerite monoliths (their SH33 Group)
and debitage fragments were derived from a single geographic
source in the eastern part of Mynydd Preseli, either the Carn
Meini-Carn Gyfrwy, or the Cerrigmarchogion or Carn Goedog
outcrops; SH42 is identical (for the immobile elements) to Carn Breseb;
and SH44, SH45 and BB1 are close in composition to dolerite from
However, Ixer (1996, 1997), mainly on the basis of reflected light
microscopy, reached some slightly alternative conclusions for some
of the monoliths, namely that: SH33, 37, 43, 49, 61, 65 and 67 were sourced from ‘Carn
Goedog Carn Meini’; SH44 and SH45 were sourced from Carn Ddafad-las; and
SH42 was sourced from Carn Breseb (Ixer, 1996), ‘although
petrography suggests Carn Bica or Carn Ddafad-las.’ according to
Finally, Ixer (1997) could provide no satisfactory petrographical
match for sample SH62.
Later, Williams-Thorpe et al. (2006) reported on new PXRF data
for outcrops in the eastern Mynydd Preseli and concluded that Carn
Meini might not in fact be the source of any spotted dolerite artefacts
(and indeed therefore also any of the Stonehenge monoliths)
and noted that the original WDXRF data of Thorpe et al. (1991)
contained ‘clues’ which suggest in fact that the Carn Goedog
outcrop contains dolerites with a composition closer to the main
Stonehenge (SH33) group identified by Thorpe et al. (1991).
What is clear, nevertheless, is that on the basis of hand specimen
and transmitted light microscopy by Thomas (1923), whole rock
geochemistry by Thorpe et al. (1991), and reflected light microscopy
by Ixer (1996, 1997), and new PXRF and magnetic susceptibility
data (Williams-Thorpe et al., 2006) the Stonehenge doleritic
bluestones can be provenanced to the eastern Mynydd Preseli.
What remains to be resolved however is which are the exact source
outcrops and then definitively matching any Stonehenge material
From p 12:
On the basis of the above analysis we consider that eleven (and
possibly twelve) samples, representing over half of the Stonehenge
dolerite monolith and debitage samples analysed to date (but
excluding the Boles Barrow sample), can be sourced on the evidence
of compatible element chemistry to the castellated outcrop
of spotted dolerite called Carn Goedog. A further three samples are
possibly sourced to the Cerrigmarchogion-Mynydd-bach area,
more specifically to outcrops at the western end of that line of E-W
trending outcrops, or to Craig Talfynydd. This leaves five Stonehenge
samples and the Boles Barrow sample as coming from outcrops
in the Carn Gyfrwy-Carn Breseb-Carn Alw area (including the
un-named outcrop w of Carn Ddafad-las), with SH44 remaining
unassigned in this study. This result is somewhat surprising as Carn
Meini was considered by Thomas (1923) as the most likely source of
most of the Stonehenge doleritic bluestones (see Darvill et al.,
2008), although he did identify Cerrigmarchogion as another
possible provenance for Stonehenge doleritic bluestones. Our proposal
for Carn Goedog as the principal source ( 55%) of the analysed
Stonehenge doleritic bluestones gives credence to the
suggestion by Williams-Thorpe et al. (2006) that the original
Thorpe et al. (1991) data contained clues indicating that this
outcrop might be the source for many of the Stonehenge bluestones
rather than Carn Meini.
From p 12:
In summary, we believe we can explain the origin of the three
Stonehenge dolerite groups by petrogenetic modelling involving
crystal fractionation and/or crystal accumulation, processes which
cannot be detected using the incompatible element approach
adopted by Thorpe et al. (1991).
From p 14:
9. Summary and next steps
On the basis of bivariate plots involving the compatible elements
MgO, Ni, Cr and Fe2O3 this study has identified three
geochemical groupings amongst the Stonehenge monolith and
debitage samples. Differences between these three groups can be
explained in terms of high crustal level igneous processes, namely
crystal fractionation and/or crystal accumulation, affecting a series
of melts derived from a common parental source and can be
determined at the outcrop level in the eastern Mynydd Preseli. Use
of these compatible elements contrasts with the use of incompatible
elements as applied by Thorpe et al. (1991) which are unable to
detect such geochemical differences related to high crustal level
igneous processes in the samples in question.
The bivariate plots suggest that the three Stonehenge dolerite
geochemical groups are derived from three principal geographical
sources, namely Carn Goedog (Group 1), Cerrigmarchogion or
possibly Craig Talfynydd (Group 2) and a set of outcrops including
Carn Breseb, Carn Gyfrwy, those in the vicinity of Carn Alw and the
un-named outcrop immediately west of Carn Ddafad-las (Group 3)
(but note that sample SH44 remains unassigned in this study).
Whilst this agrees in part with the suggestions of Thorpe et al.
(1991) on the basis of whole rock geochemistry and Ixer (1996,
1997) on the evidence of reflected light microscopy the principal
conclusion presented here is that at least 55% of the Stonehenge
monoliths and fragments analysed to date can be sourced to Carn
Goedog. The PCA plots support the association of the Group 1
Stonehenge dolerites with Carn Goedog but also suggest that Group
3 dolerites might come from Carn Goedog, rather than the outcrops
listed above. What seems likely, and is hinted at in some of the
bivariate plots (e.g. Figs.10 and 11), is that the Groups 1 and Group 3
magmas were derived from the same magma batch and that the
difference between Group 1 and Group 3 dolerites relates to
varying mineral concentrations (resulting from crystal fractionation
and/or crystal accumulation) in the samples and that the
compatible element plots highlight this feature. Group 3 samples
might be from an as yet un-sampled part of the Carn Goedog
outcrop, bearing in mind that Jones et al. (2005), on the basis of
PXRF investigations, identified that a number of the eastern Preseli
outcrops were geochemically heterogeneous. Further sampling of
the Carn Goedog intrusion would serve to clarify if this is the case or
Overall, the PCA plots, based on a broader range of elements
(some of which are not compatible in the mafic minerals in the
crystallizing assemblage in the dolerites), serve to support the
conclusions reached on the basis of the compatible bivariate
elemental plots presented here.
Clearly there remain some uncertainties over the exact source of
a small number of Stonehenge dolerite samples and we believe that
generation of a new geochemical dataset, based on ICP-MS techniques,
offers the best way forward to resolve the sources of these
samples. A dataset based on ICP-MS techniques would allow not
only for the generation of analyses for some elements with a precision
and accuracy greater than those derived by WDXRF techniques
but would also allow for additional elements to be added to
the dataset, including for example Sc, Ta, Th, U and the rare earth
elements. These elements would serve to provide further information
about high crustal level igneous processes, such as the role
of clinopyroxene in the crystal fractionation process (by using
Rayleigh fractionation modelling techniques) and also any additional
complicating processes such as crustal contamination of the
melts during magma ascent (by assimilation fractional crystallization
modelling). Some of these elements, not available in the current
dataset, would serve to complement the role that the elements
MgO, Ni, Cr and Fe2O3 have provided in the current study. However,
what is also a pre-requisite is the opportunity to re-analyse well
contexted Stonehenge dolerites, most especially the orthostats.