Content uploaded by Richard Bevins
Author content
All content in this area was uploaded by Richard Bevins on Oct 12, 2023
Content may be subject to copyright.
Journal of Archaeological Science: Reports 51 (2023) 104215
Available online 22 September 2023
2352-409X/Crown Copyright © 2023 Published by Elsevier Ltd. All rights reserved.
The Stonehenge Altar Stone was probably not sourced from the Old Red
Sandstone of the Anglo-Welsh Basin: Time to broaden our geographic and
stratigraphic horizons?
Richard E. Bevins
a
,
*
, Nick J.G. Pearce
a
, Rob A. Ixer
b
, Duncan Pirrie
c
, Sergio And`
o
d
,
Stephen Hillier
e
,
f
, Peter Turner
g
, Matthew Power
h
a
Department of Geography and Earth Sciences, Aberystwyth University, Aberystwyth SY23 3DB, UK
b
Institute of Archaeology, University College London, London WC1H 0PY, UK
c
Faculty of Computing, Engineering and Science, University of South Wales, Pontypridd CF37 4BD, UK
d
Department of Earth and Environmental Sciences, Universit`
a degli Studi di Milano-Bicocca, 20126 Milano, Italy
e
The James Hutton Institute, Craigiebuckler, Aberdeen AB15 8QH, UK
f
Department of Soil and Environment, Swedish University of Agricultural Sciences (SLU), SE-75007 Uppsala, Sweden
g
7 Carlton Croft, Streetly, West Midlands B74 3JT, UK
h
Vidence Inc., 4288 Lozells Avenue, Suite 213 – L, Burnaby, British Columbia, V5A 0C7, Canada
ARTICLE INFO
Keywords:
Neolithic
Stonehenge
Altar Stone
Sandstone analysis
Provenancing
ABSTRACT
Stone 80, the recumbent Altar Stone, is the largest of the Stonehenge foreign “bluestones”, mainly igneous rocks
forming the inner Stonehenge circle. The Altar Stone’s anomalous lithology, a sandstone of continental origin,
led to the previous suggestion of a provenance from the Old Red Sandstone (ORS) of west Wales, close to where
the majority of the bluestones have been sourced (viz. the Mynydd Preseli area in west Wales) some 225 km west
of Stonehenge. Building upon earlier investigations we have examined new samples from the Old Red Sandstone
(ORS) within the Anglo-Welsh Basin (covering south Wales, the Welsh Borderland, the West Midlands and
Somerset) using traditional optical petrography but additionally portable XRF, automated SEM-EDS and Raman
Spectroscopic techniques. One of the key characteristics of the Altar Stone is its unusually high Ba content (all
except one of 106 analyses have Ba >1025 ppm), reecting high modal baryte. Of the 58 ORS samples analysed
to date from the Anglo-Welsh Basin, only four show analyses where Ba exceeds 1000 ppm, similar to the lower
range of the Altar Stone composition. However, because of their contrasting mineralogies, combined with data
collected from new automated SEM-EDS and Raman Spectroscopic analyses these four samples must be dis-
counted as being from the source of the Altar Stone. It now seems ever more likely that the Altar Stone was not
derived from the ORS of the Anglo-Welsh Basin, and therefore it is time to broaden our horizons, both
geographically and stratigraphically into northern Britain and also to consider continental sandstones of a
younger age. There is no doubt that considering the Altar Stone as a ‘bluestone’ has inuenced thinking
regarding the long-held view to a source in Wales. We therefore propose that the Altar Stone should be ‘de-
classied’ as a bluestone, breaking a link to the essentially Mynydd Preseli-derived bluestones.
1. Introduction
Stonehenge is arguably the most iconic of Neolithic monuments in
the World. It stands on Salisbury Plain in Wiltshire and Parker Pearson
(2023, 161) considers that it was rst erected in the Late Neolithic
around 3000 BCE. The initial phase of construction was followed by four
further re-modelling phases, the last being in the Middle Bronze Age, ca.
1600 BCE. It was during the rst phase that according to Parker Pearson
(op. cit.) the bluestones were erected as a single ring of stones set in a
series of 56 pits known as the Aubrey Holes. Pitts (2022) called this ring
of stones ‘bluehenge’. The larger sarsen stones are thought to have been
brought to Stonehenge during construction Phase 2, at the end of the
Late Neolithic (ca. 2500 BCE). However, other authors have alternative
chronologies for Stonehenge; see for example Darvill (2022), who also
* Corresponding author.
E-mail address: rib24@aber.ac.uk (R.E. Bevins).
Contents lists available at ScienceDirect
Journal of Archaeological Science: Reports
journal homepage: www.elsevier.com/locate/jasrep
https://doi.org/10.1016/j.jasrep.2023.104215
Received 27 June 2023; Received in revised form 30 August 2023; Accepted 14 September 2023
Journal of Archaeological Science: Reports 51 (2023) 104215
2
questions whether the Aubrey Holes ever held bluestone monoliths.
The bluestones, predominantly of igneous origin, were termed the
‘Foreign Stones’ by early excavators at Stonehenge (for example Cun-
nington, 1884), being exotic to the Wiltshire landscape, in contrast to
the sarsen stones, which are identied as being of relatively local deri-
vation, Nash et al. (2020) recently stating that the principal source of the
Stonehenge sarsens was most likely West Woods, ca. 25 km north of
Stonehenge. The majority of the bluestones have been sourced to the
Mynydd Preseli area in west Wales (see Figure 1), ca. 225 km west of
Stonehenge, originally by Thomas (1923) and with more recent in-
vestigations by Thorpe et al., (1991), (Ixer and Bevins 2010; Ixer and
Bevins 2011), (Bevins et al. 2012; Bevins et al. 2014; Bevins et al. 2021)
and Pearce et al. (2022).
Monoliths used in the construction of stone circles are usually locally
derived. Linares-Catela et al. (2023) recently reported that stones used
in the El Pozuelo megalithic complex in Huelva, Spain were moved from
distances in the range of 50–350 m. One of the best documented ex-
amples from the Neolithic of Britain is the sourcing of stones used in the
Ring of Brodgar and the Stones of Stenness monuments on Orkney in
north Scotland which were quarried from sources around Staneyhill and
Vestra Fiold, no more than 5–10 km away (Downes et al., 2013; Richards
et al., 2013). It is the long-distance transport of the bluestones that
makes Stonehenge of particular interest; the bluestones in fact represent
one of the longest transport distances known from source to monument
construction site anywhere in the world (Parker Pearson et al., 2020).
Through the recent studies mentioned above there has been a
continued renement of the Preseli sources of some of the bluestone
lithologies, including Craig Rhos-y-Felin (the source of the main rhyo-
litic debitage at Stonehenge and possibly the buried stump of Stone
32d), Carn Goedog (the main source of the spotted dolerites) and Garn
Ddu Fach (the source of the non-spotted dolerite Stone 62). One blue-
stone, Stone 80, known as the Altar Stone, a grey-green (on fresh sur-
faces), micaceous sandstone, however, is anomalous in that it is not
derived from the Mynydd Preseli and surrounding area, and it is this
stone that is the subject of this paper.
2. Previous work and scope of this paper
As noted by Bevins et al. (2022a), one of the earliest references to the
Altar Stone was in a letter from Professor John Phillips of Oxford Uni-
versity to archaeologist Dr John Thurnam in 1858, suggesting that it
might have been sourced in the ‘…Devonian or gray Cambrian rocks.’,
possibly referring to the marine Devonian sequences in southwest En-
gland. Maskelyne (1878) mentioned this attribution but noted that his
assistant, Mr Thomas Davies, had informed him that such rocks could be
found in the Frome area, in the Mendips of Somerset. Thomas (1923)
considered that the Altar Stone is of Old Red Sandstone (ORS) age and
might have been derived from outcrops in west or south Wales lying to
the south or east of the Mynydd Preseli, either from beds of the Cosh-
eston Group (now called the Cosheston Subgroup) or the Senni Beds
(now called the Senni Formation).
The Altar Stone is the largest of all the bluestones, measuring 4.9 m
long by 1 m wide by 0.5 m thick with a slab-like form. Recent in-
vestigations by Ixer et al. (2019) and (Bevins et al. 2020; Bevins et al.
2022a; Bevins et al. 2023) have attempted to provenance the Altar Stone
by characterizing its chemistry and mineralogy using a range of
analytical techniques, in particular automated scanning electron mi-
croscopy (SEM-EDS), U-Pb zircon age determination and preliminary
portable X-ray uorescence (pXRF) analysis. Key characteristics include
abundant mica, heavy mineral laminae dening ripple bedforms, the
Fig. 1. Location map for samples analysed in this study, including the outcrop map of Old Red Sandstone sedimentary strata in the Anglo-Welsh Basin (bolder
colours) overlain on the background geological map of the area (faded colours). Contains British Geological Survey materials © UKRI [2023] from BGS GeoIndex
(onshore). Grid lines mark the British National Grid 100 km squares, designated by their 2-letter code (e.g. SN, see Supplementary Table 1). The location of
Stonehenge is shown at the bottom right of the map. The locations of the four high Ba ORS samples are indicated by their sample numbers.
R.E. Bevins et al.
Journal of Archaeological Science: Reports 51 (2023) 104215
3
presence of early formed pore-lling baryte and kaolinite cement,
thought to be linked to burial diagenesis, a later calcite cement which
post-dates quartz overgrowths which occludes much of the available
porosity and a near absence of K feldspar. The presence of baryte is
reected in the high Ba contents as determined through the pXRF in-
vestigations presented in Bevins et al. (2022a, 2023), with the average
composition from all in situ analyses of the Altar Stone exceeding
2750 ppm.
Bevins et al. (2022a), on the basis of their preliminary pXRF analyses
and limited automated mineralogy data, were unable to offer any po-
tential source for the Altar Stone and remarked that they needed to ‘keep
an open mind over the potential source of the Altar Stone, especially as we are
not aware of any reports of baryte-bearing sandstones in the Old Red
Sandstone sequences of Wales and the Welsh Borderland’. This paper re-
ports on the ndings of further, continuing investigations of the Old Red
Sandstone (ORS) in Wales, the Welsh Borderland and the West Midlands
and Somerset in England (in the Anglo-Welsh Basin of Barclay et al.,
2015) based on an enlarged pXRF database, further automated SEM-EDS
analyses and initial ndings of Raman Spectroscopy analyses of an Altar
Stone fragment and an ORS sample from the West Midlands which bears
certain geochemical and mineralogical similarities to the Altar Stone.
The perspective we are forming is that it is perhaps time to broaden our
horizons, both geographically and stratigraphically, by looking else-
where other than the ORS Anglo-Welsh Basin (see Barclay et al., 2005;
Kendall, 2017 for the distribution of ORS strata in Britain) and perhaps
also to consider potential sources in younger sequences of Permo-
Triassic age.
3. Portable XRF analyses
In previous studies we have reported on portable XRF (pXRF) ana-
lyses of the Altar Stone (in situ analyses performed on two separate
visits), analyses of six small pieces of debitage (which were conrmed to
be fragments of the Altar Stone; Bevins et al., 2022a), and of sample
2010 K 240 from the collections of Salisbury Museum (sometimes
referred to as Wilts 277), which we conrmed as a piece collected from
the underside of the Altar Stone in 1844 (Bevins et al., 2023). During our
studies, and subsequently, we have analysed a total of 58 geographically
widespread samples of Old Red Sandstone from the Anglo-Welsh Basin
in an attempt to nd samples with a mineralogy and chemistry com-
parable with the Altar Stone (see Figure 1). The samples were drawn
largely from the set used by Hillier et al. (2006) in their X-ray diffraction
study of the ORS of the UK, supplemented by samples drawn from the
collections of the National Museum of Wales and a small number of
eld-collected samples. Sample site details are presented in Supple-
mentary Table 1.
3.1. Analytical methods
All analyses were performed using a Thermo Fisher Scientic™
Niton™ XL3t Goldd+handheld XRF analyser. The Niton pXRF uses a
2 W Ag anode X-ray tube, which can operate at between 6 and 50 kV and
0–200
μ
A, with operating conditions being varied during the “TestA-
llGeo” analysis method. The instrument can determine a range of ele-
ments in geological materials from Mg to U by use of different lters
which operate in sequence together to optimise sensitivity, although
light element analyses are less accurate without a He ush of the in-
strument and are sensitive to the presence of moisture in the sample. The
total analysis time was 100 s, divided between four operating modes
(Main range 30 s, Low range 30 s, High range 20 s, Light range 20 s)
using an 8 mm diameter analysis spot to give an analysed area
of ~ 50 mm
2
, with the spectra collected on a silicon drift detector, pro-
cessed and calibrated by the instrument’s manufacturer-installed cali-
bration. Here, across several separate periods of analyses, we performed
ve analyses of the weathered surfaces of each ORS sample and moni-
tored instrument calibration using a piece of the Big Obsidian Flow from
the Newbery Volcano in Oregon. All analyses are presented in the
Supplementary Table 2. Elsewhere we have discussed at length analyt-
ical methods and instrument accuracy, and these aspects of the method
are not revisited here (see Bevins et al., 2022a; Bevins et al., 2022b;
Pearce et al., 2022; Bevins et al., 2023).
3.2. Portable XRF comparisons
Fig. 2 shows a series of bivariate plots for the data for the ORS
samples from the Anglo-Welsh Basin compared with the analyses of the
Altar Stone (sensu lato, i.e., including the debitage fragments and 2010 K
240 shown to be derived from the Altar Stone). Here we concentrate on
those elements which are reported in most analyses, and which are
generally determined with good accuracy by pXRF (Bevins et al.,
2022b), including V, Rb, Sr, Zr, Nb, Ba and Th. Barium is signicant
because of the presence of abundant baryte (Ixer et al., 2019; Bevins
et al., 2020) together with calcite as a cement in the Altar Stone.
However, Ca concentrations, along with other light (low atomic num-
ber) elements, are affected by moisture in pXRF analysis and by surface
features/alteration, so are not considered here. In addition, Bevins et al.
(2022a) and Bevins et al. (2023) noted that Ca had been leached from
some of the in situ Altar Stone analyses.
As noted above, the Altar Stone contains high Ba, with all but 1 of the
106 analyses (plotted in red on Figs. 2 and 3) containing >1025 ppm,
the outlying analysis coming from Area C of the Altar Stone (Bevins
et al., 2022a) which was difcult to access, being partly under Stone 156
(a fallen lintel) and partly under Stone 55b (part of the Great Trilithon).
Barium concentrations are clearly far higher in the Altar Stone than the
majority of Anglo-Welsh Basin (AWB) ORS samples, with only four AWB
ORS samples having analyses which exceed 1000 ppm: these samples –
WM 6, 2009.46G.R.3a, LSF2-5504 and LORS 27 (locations shown on
Figure 1) – are plotted with black symbols in Figs. 2 and 3. The
remaining 54 AWB ORS samples are not individually identied in Figs. 2
and 3 and are plotted as green triangles. Strontium concentrations are
also generally higher in the Altar Stone (104 analyses >86 ppm) than
the AWB ORS samples, although a few AWB ORS analyses exceed
300 ppm Sr (not shown on Fig. 2 but see Supplementary Table 2). The
Ba-Sr distribution clearly separates the Altar Stone from the majority of
AWB ORS samples, with a strong positive correlation between Ba and Sr
in the Altar Stone (Sr =0.0092 Ba +91, r =0.71), whereas the rela-
tionship between all AWB ORS samples is poor (r =0.15). Of the four
high-Ba (>1000 ppm) ORS samples, WM 6 has a similar Ba-Sr compo-
sition to the Altar Stone, the two high Ba analyses in LSF2-5504 have
higher Sr than the Altar Stone, 2009.46G.R.3a has Sr <86 ppm in its
four high-Ba analyses, as does the sole high-Ba analysis from LORS 27,
although these samples both sit in the extended envelope of all Altar
Stone analyses. We take the strong Ba-Sr correlation to reect that Sr in
the Altar Stone substitutes for Ba in the baryte.
However, the Ba-Rb plot shows that 2009.46G.R.3a, LORS 27 and
LSF2-5504 all have Rb contents a factor of ~ 3 higher than in the Altar
Stone, ruling out a possible relationship. A similar relationship is seen
with K (not plotted). WM 6, however, has similar Rb concentrations to
the Altar Stone. Zirconium concentrations overlap for the Altar Stone
and AWB ORS samples; however, many AWB ORS samples have Zr
contents much lower than the Altar Stone, and a few Altar Stone ana-
lyses exceed the AWB ORS concentrations. In terms of Zr and Nb, there is
an overlap of WM 6 and 2009.46G.R.3a with the Altar Stone analyses,
but for the Altar Stone Nb =0.0126 Zr +7.35 (r =0.91), whilst for the
AWB ORS samples Nb =0.0194 Zr +5.2 (r =0.58), possibly indicative
of source regions with different Zr/Nb ratios.
For the ORS samples, V =3.56 Nb +54 (r =0.15), but for the Altar
Stone V =1.38 Nb +32.5 (r =0.34). Both V and Nb are likely to be
associated with Fe-Ti and Ti oxide (probably rutile) phases where Nb
will likely substitute for Ti, and V for Fe (GERM, 2021; Rollinson and
Pease, 2021), V particularly favouring inclusion within magnetite. Here,
WM 6 does not plot with the Altar Stone analyses, having higher V,
R.E. Bevins et al.
Journal of Archaeological Science: Reports 51 (2023) 104215
4
which may separate it from the Altar Stone. This difference is consistent
with the high concentration of altered titaniferous magnetite (now
martite) and titaniferous hematite in WM 6 and their total absence in the
Altar Stone. The different V/Nb ratios between the ORS samples and the
Altar Stone may suggest a different mix of, or source for, the Fe-Ti ox-
ides, also consistent with the mineralogy of the opaque phases. Slightly
more overlap is shown for Nb and Ti (not plotted) which shows WM 6
with concentrations similar to the highest in the Altar Stone, but the
AWB ORS samples again have generally higher Ti.
Fig. 3 presents three triangular diagrams which conrm some of the
associations described above. The clearest distinction between the ORS
and the Altar Stone samples is seen in the alkali and alkaline earth
metals Ba, Sr and Rb, with only WM 6 showing any similarity to the Altar
Stone. The highly incompatible and immobile elements Zr, Nb, and Th,
which will reside in accessory phases in the sandstones, show a general
overlap, suggestive of generally similar sources and processes. The Ti-V-
Nb plot, however, suggests that the Altar Stone has lower Ti and V, and
higher Nb than the AWB ORS samples, possibly related to Fe-Ti and Ti
oxides, with WM 6 plotting in the middle of the ORS eld, and slightly
offset from the Altar Stone compositions.
From the above it is clear that the majority of the AWB ORS samples
have a very different chemical composition from the Altar Stone, with
only four ORS samples showing Ba >1000 ppm. Of these four high-Ba
samples, WM 6 is the only one which consistently plots close to, but
not always within, the eld of compositions of the Altar Stone. We have
investigated these four samples further using automated SEM-EDS and
Fig. 2. Bivariate plots of geochemical data for the Altar Stone and its related samples (red symbols), and undifferentiated Old Red Sandstone samples analysed in the
study plotted in green with those samples containing >1000 ppm Ba labelled separately (black symbols). On these diagrams, despite four samples containing more
than >1000 ppm Ba, only WM 6 plots consistently close to the eld of data from the Altar Stone. The plot of Nb - V excludes all analyses from WHB-3 (see Sup-
plementary Table 1) which has V between 240 and 410 ppm.
R.E. Bevins et al.
Journal of Archaeological Science: Reports 51 (2023) 104215
5
preliminary Raman Spectroscopy (for WM 6), as detailed below,
comparing with data from the Altar Stone and derived fragments, and
nally make comparisons with our ndings based on standard
petrography.
4. Automated SEM-EDS
Earlier work has utilised automated SEM-EDS analysis to quantify
the mineralogy in a textural context of the Altar Stone along with other
bluestone lithologies (Ixer et al., 2019; Ixer et al., 2022; Ixer et al., 2023;
Bevins et al., 2020; Bevins et al., 2021; Bevins et al., 2022a; Bevins et al.,
2023) and samples from one of the sarsen stones (Nash et al., 2021).
Based on the pXRF geochemical analyses, the four ORS samples with Ba
compositions comparable to the Altar Stone were selected for detailed
mineralogical analysis.
4.1. Analytical methods
Previous analyses utilised a QEMSCAN automated SEM-EDS plat-
form; however, in this study the analysis was carried out using an AMICS
system. Analysis was undertaken using a Hitachi SU3900 scanning
electron microscope tted with two large area (60 mm
2
) Bruker SDD
energy dispersive spectrometers and running the Bruker AMICS auto-
mated mineralogy package. Beam conditions were optimised for anal-
ysis and therefore an accelerating voltage of 20 kV coupled with a beam
current of approximately 10nA was used. All samples were measured
using the same analytical parameters and, to retain consistency with
previous (QEMSCAN) analyses, a mapping mode of analysis was used.
This analytical mode rst acquires a high-quality BSE image and then
systematically steps the electron beam across the sample at a preset
stepping interval of, in this case, 10 µm. An EDS X-ray spectrum is
collected at each point, compared with a spectral library of known
minerals and compositions, and a mineral assignment is made.
4.2. Comparison of results from the AMICS and QEMSCAN platforms
All automated SEM-EDS systems are based on the same technology,
but the data processing software differs. To allow direct comparison
between the previous analyses carried out using a QEMSCAN platform,
the new AMICS-based analyses replicated the previously reported min-
eral groupings. To test the comparison between the different analytical
platforms we repeated the analysis of two samples (2010 K 240 and WM
6) using AMICS, which had previously also been analysed using the
QEMSCAN platform. These replicate analyses used the same polished
thin sections, although the exact area of the sandstone measured in the
analyses will have differed slightly and hence some sample variance can
be expected. The replicate analyses for the two samples are provided in
Table 1 and are shown graphically in Fig. 4. Based on the data presented
in Table 1 and Fig. 4 there is a very strong correspondence between the
replicate analyses. Key, albeit small, analytical differences are that re-
ported muscovite abundance is higher within the QEMSCAN analyses
rather than the AMICS data (2.32 / 2.82 % versus 1.05 / 1.28 %
respectively); it is likely that some areas reported as muscovite in the
QEMSCAN analysis are reported to the illite mineral categories in the
AMICS dataset and vice versa. Dolomite is also apparently more abun-
dant in the QEMSCAN dataset when compared with the AMICS data; the
apparent increase in the calcite abundance in the AMICS data suggests
that dolomite is reporting to the calcite mineral grouping or vice versa.
All other mineral categories are within expected variance for replicate
analyses (see Pirrie et al., 2009). Neither dolomite nor muscovite
abundance is a critical characteristic discriminator in terms of com-
parison between known Altar Stone and questioned ORS samples. It is
reasonable, therefore, to make comparisons between the new AMICS-
generated results with those from previous QEMSCAN-generated results.
Fig. 3. Triangular variation diagrams for the alkali/alkali earth metals Ba-Sr-
Rb, the highly incompatible elements Nb-Zr-Th, and elements associated with
Fe-Ti oxides, viz. Ti-V-Nb.
R.E. Bevins et al.
Journal of Archaeological Science: Reports 51 (2023) 104215
6
4.3. New results
The mineralogical data for samples 2009.46G.R.3a, LORS 27, LSF2-
5504 and WM 6 are provided in Table 2 and are plotted in Fig. 5 against
previous analyses of samples proven as derived from the Altar Stone.
Based on the overall modal mineralogy there are signicant mineral-
ogical differences between the analyses of samples conrmed as derived
from the Altar Stone and the four ORS samples selected for analysis. The
mineralogical data support the pXRF data which indicate that if samples
2009.46G.R.3a, LORS 27 and LSF2-5504 are representative of the
location sampled, then these localities can be excluded as the source of
the Altar Stone. Geochemically sample WM 6 is the only one which
consistently plots close to the eld of compositions of the Altar Stone but
the automated mineralogy data show that it differs mineralogically
based on the abundance of detrital K feldspar, plagioclase and Fe oxides
and diagenetic calcite and baryte (which are key characteristics of the
Altar Stone mineralogy). Consequently, based on the automated
mineralogy dataset, if sample WM 6 is representative of the location
from which it was collected, then this area too can be excluded as the
source area for the Altar Stone.
5. Raman Spectroscopy
Raman Spectroscopy in provenance studies can be used both as a
mineralogical ngerprint and also to allow inferences to be made as to
the geological and geographical sources of sediment (Garzanti and
And`
o, 2007a). This is widely applied to both modern, unconsolidated
sandy/silty sediments (And`
o et al., 2011) as well as sedimentary rocks
(Garzanti and And`
o, 2007b). Here we have applied Raman Spectroscopy
to compare the Anglo-Welsh Basin ORS sample WM 6 (see above) to MS-
1, one of the debitage fragments derived from the Altar Stone (Ixer et al.,
2019; Bevins et al., 2020; Bevins et al., 2022a), to assess whether their
chemical similarity reported above is reected in their mineralogy and
how this mineralogy compares with that determined by automated SEM-
EDS.
The technique can be undertaken with mineral grain sizes down to a
few microns, is non-destructive, and can be applied to minute quantities
of material. As well as providing mineral abundance data, mineral
compositional information is also generated, and this method has ap-
plications in geoarchaeological studies (Zimmermann et al., 2016).
5.1. Sample preparation and analysis methods
Small amounts from the two samples analysed were rst dis-
aggregated and then the heavy minerals (HMs) were concentrated
following the protocol for gravimetric separation developed by And`
o
(2020) in order to calculate the heavy mineral abundance in the two
samples. A Renishaw inVia Reex® Raman Spectrograph at the Uni-
versit`
a di Milano-Bicocca was used with a 50x long working distance
(LWD) objective, coupled to a green laser (λ =532 nm). Raman spectra
were collected in the 150–1200 cm
−1
spectral range and in the high
frequency OH
–
region around 3100–4000 cm
−1
for hydrated minerals,
this combination allowing identication of individual mineral types as
well as mineral varieties. The masses available for the two samples
required different methods of preparation, as described below. Optical
analysis, in transmitted and reected light using a polarizing microscope
(Mange and Maurer, 1992), was combined with a single grain Raman
Spectroscopy approach (And`
o et al., 2011; And`
o and Garzanti, 2013) in
order to quantify mineral abundances in the heavy mineral separates.
5.1.1. MS-1 – Altar Stone debitage fragment
Only 0.1973 g of sample MS-1 (>2
μ
m) was available for study, being
the residue following X-ray diffraction (XRD) analysis of a detached
fragment from the sample. This is close to the limit of the gravimetric
protocol for heavy mineral separation using only a few milligrams
available in forensic investigations. After wet sieving at 500
μ
m using a
steel sieve, the sample was dried and weighed, with the >500
μ
m
fraction giving 0.0231 g and the >2–500
μ
m fraction giving 0.1453 g
(74 % of the sample). The sieved yield represented a weight loss
of ~ 15 %, considered acceptable when working with samples of only
fractions of a gramme. The 2–500
μ
m fraction was separated into heavy
and light mineral fractions in a centrifuge using the non-toxic heavy
liquid sodium polytungstate (SPT) at a density of 2.90 g/cm
3
. The light
fraction weighed 0.1334 g and the total heavy mineral grains were 0.002
g, representing only 1.4 % by weight of the sample, 2 mg being the
minimum amount of material for preparing a grain mount.
For MS-1 all the separated heavy mineral grains were mounted on a
slide with Canada Balsam (n =1.54) for optical identication of the
mineralogy, to document the texture of single grains and to perform
analyses by Raman Spectroscopy. Considering the wide grain size range
this HM sample comprises (2–500
μ
m), it is essential to apply a point
counting method to determine the mineral frequency which can then be
transformed into abundance percentages by accounting for the grain
volumes. Observations in transmitted and reected light enable the
entire mineral population to be described, here giving 206 transparent
HM grains and 325 opaque, turbid, phyllosilicate, carbonate and “light
minerals”. Raman Spectroscopy is then nally applied to the mineral
separate to differentiate magmatic (schorl) versus metamorphic (dra-
vite) tourmalines, garnet types, apatite, and carbonates. In MS-1, HM
grains are well sorted, mostly sub-rounded, with very rare opaque
minerals, no Fe-hydroxides, and Ti-oxides comprising 12 % of semi-
opaque heavy grains (Fig. 6). In the transparent HM suite, apatite
(29 %) is dominant, occurring with zircon (9 %), garnet (8 %), rutile
(6 %), tourmaline (4 %), red spinel (2 %), with trace quantities of blue-
green amphibole, anatase and epidote. Baryte is very common (40 %),
occurring with “light minerals” (27 %), chlorite (15 %), authigenic Ti-
oxides (12 %) (occurring as granular rutile and anatase), undifferenti-
ated carbonates (5 %, not calcite), and biotite (2 %). Apatite shows a
well-rounded to sub-rounded shape, whilst baryte is more angular with
corroded rims, and chlorite is larger in size and rounded to sub-rounded.
Table 1
Comparison of replicate analyses using both a QEMSCAN and an AMICS auto-
mated SEM-EDS mineralogy platform.
2010 K 240a 2010 K
240b
WM 6a WM 6b
Analytical System QEMSCAN AMICS QEMSCAN AMICS
Quartz 51.03 52.69 53.83 54.73
K Feldspar 0.34 0.25 1.27 1.23
Plagioclase 12.15 11.59 10.45 10.71
Muscovite 2.32 1.05 2.82 1.28
Biotite 0.48 0.48 0.83 0.85
Kaolinite 3.37 3.18 2.63 3.14
Chlorite 4.05 3.59 6.51 4.37
Illite & Illite-
smectite
4.97 5.14 6.42 6.51
Fe-Illite & Illite-
smectite
0.89 1.59 4.51 4.97
Calcite 18.80 19.44 9.10 10.29
Dolomite 0.59 0.01 0.20 0.01
Ferroan Dolomite 0.04 0.00 0.03 0.00
Fe Oxides 0.00 0.00 0.32 0.44
Chromite 0.01 0.02 0.02 0.02
Pyrite 0.00 0.00 0.00 0.00
Baryte 0.13 0.11 0.27 0.24
Anhydrite 0.00 0.00 0.00 0.00
Halite 0.00 0.00 0.00 0.00
Rutile & Ti Silicates 0.37 0.31 0.27 0.23
Ilmenite 0.04 0.08 0.19 0.19
Apatite 0.24 0.20 0.16 0.16
Garnet 0.05 0.09 0.13 0.48
Tourmaline 0.06 0.03 0.02 0.03
Zircon 0.05 0.05 0.03 0.05
Undifferentiated 0.00 0.09 0.00 0.07
R.E. Bevins et al.
Journal of Archaeological Science: Reports 51 (2023) 104215
7
5.1.2. WM 6: Anglo-Welsh Basin ORS sample from the West Midlands
A larger sample of WM 6 was available for analysis, which had been
previously disaggregated for XRD analysis. The quantity of powder was
sufcient to apply a standard preparation protocol, and a representative
aliquot was obtained by splitting it using the method of Parfenoff et al.
(1970) to give an initial dried fraction of 2.0515 g.
After wet sieving the sample at 500
μ
m with a steel sieve,
the >500
μ
m fraction was 0.0047 g, the 5–500
μ
m fraction was 1.8553 g
(90 % of the sediment), with 0.1406 g <5
μ
m. The 5–500
μ
m fraction
was centrifuged in SPT, giving a light fraction of 1.7694 g and the heavy
grains of 0.0393 g (2.1 % by weight). A representative aliquot of HMs
was obtained using a micro-rife box and prepared as a grain mount
with Canada Balsam. Once again, considering the wide grain size range
(5–500
μ
m), a heavy-mineral point counting method was applied. Op-
tical inspection identied 200 transparent HMs together with 372 opa-
que, turbid grains, phyllosilicates, carbonates and “light minerals”.
Heavy mineral grains in WM 6 are poorly sorted, most of the grains are
angular, and both opaque minerals (16 %) and semi-opaque Fe-hy-
droxides are common (15 %). In the transparent HM suite, garnets
(28 %) are dominant, occurring with apatite (13 %), zircon (11 %), rutile
and spinel (3 %), tourmaline (2 %), and trace amounts of epidote and
anatase (1 %). Other common minerals in WM 6 include Ti-oxides
(13 %), the platy minerals chlorite (9 %) and biotite (5 %, mostly
deeply weathered), and nally “light minerals” (5 %) and minor car-
bonate (1 %, not calcite).
5.1.3. Comparison between WM6 and Altar Stone samples
The HM compositions of MS-1 (Altar Stone) and WM 6 (ORS) are
compiled in Table 3. Raman analyses of detrital HM suites show mark-
edly different HM abundances for the two analysed samples, indicating
they are different from each other and from different sources. Specif-
ically, the Altar Stone contains well sorted HMs, which are mostly sub-
Fig. 4. Comparison of the reported modal mineralogy for samples (A) 2010 K 240 and (B) WM 6 based on replicate analyses using both QEMSCAN (analysis a) and
AMICS (analysis b) automated SEM-EDS platforms.
R.E. Bevins et al.
Journal of Archaeological Science: Reports 51 (2023) 104215
8
rounded, with very rare opaque minerals and no Fe-hydroxides, as well
as containing 29 % apatite, 8 % garnet (mainly solid solutions of
almandine and spessartine) and 4 % tourmaline (both schorl and dra-
vite). In contrast WM 6 contains HMs which are poorly sorted and
angular, with only 13 % apatite, and both opaque minerals (16 %) and
semi-opaque Fe-hydroxides being common (15 %), whilst garnet is
abundant (28 %) but apatite (13 %) and tourmaline (1.5 %) less so
(dravite being absent). These differences clearly indicate that WM 6 and
MS-1 were not sourced from the same lithologies and hence have
different sources.
6. Comparisons between results of standard petrographic
examinations of WM 6 and the Altar Stone and results from
automated SEM-EDS and Raman Spectroscopy
Standard petrographic examination generally concurs with the
automated SEM-EDS and Raman Spectroscopy results but does provide
additional information. There are some differences of note between the
standard petrography results and the analytical investigations of sample
WM 6 and samples proven to have been derived from the Altar Stone. In
thin section, whilst the detrital grains in WM 6 are dominated by
monocrystalline straight extinction quartz, along with plagioclase and K
feldspar, there are also abundant mudstone clasts (Fig. 7), the latter not
seen in the analytical investigations. These mudstone clasts are typically
compacted and commonly form a pseudo-matrix around the quartz
grains. These grains could potentially either be mudstone lithic sedi-
mentary clasts, or alternatively, mudstone intraclasts. Kaolinite, chlo-
rite, illite and Fe-illite are reported as present in both the known Altar
Stone samples, in this case 2010 K 240, in which the total clay content is
13.5 % (AMICS data), and also in WM 6, but the total clay content is
signicantly higher (19 %; again AMICS data) in the latter section. In
addition, sample WM 6 is notably coarser grained than 2010 K 240, as
illustrated in Fig. 7.
Altar Stone sample 2010 K 240 and ORS sample WM 6 have sharply
contrasting opaque mineralogies, most clearly evidenced by the occur-
rence of hematite. It is notably present in WM 6 (0.44 %; AMICS data) as
martite replacing primary magnetite but also within ilmenohematite
and as titaniferous hematite-rutile intergrowths; much occurs as ne-
grained intergrowths with TiO
2
minerals replacing original iron tita-
nium oxides including ilmenite. Hematite pigment (<1 µm size crystals)
is also widespread within the matrix and occurs along the cleavage
planes of altered biotite. This contrasts with the almost total absence of
hematite in the Altar Stone (AMICS analyses show 0.00% Fe oxides),
Table 2
Modal mineralogy of selected ORS samples with elevated Ba geochemical sig-
natures based on AMICS analysis.
WM 6 LORS 27 LSF2.5504 2009.46G.T.3
Quartz 54.73 50.76 48.70 57.35
K Feldspar 1.23 3.44 0.06 0.02
Plagioclase 10.71 0.06 14.10 16.15
Muscovite 1.28 0.73 2.36 2.63
Biotite 0.85 0.25 1.41 2.79
Kaolinite 3.14 0.20 0.06 0.06
Chlorite 4.37 1.30 4.24 6.08
Illite & Illite-smectite 6.51 3.01 11.58 8.29
Fe-Illite & Illite-smectite 4.97 0.98 3.60 4.12
Calcite 10.29 38.73 13.23 1.23
Dolomite 0.01 0.35 0.00 0.00
Ferroan Dolomite 0.00 0.03 0.00 0.00
Fe Oxides 0.44 0.03 0.00 0.06
Chromite 0.02 0.00 0.01 0.02
Pyrite 0.00 0.00 0.00 0.00
Baryte 0.24 0.00 0.11 0.10
Anhydrite 0.00 0.00 0.00 0.00
Halite 0.00 0.00 0.00 0.00
Rutile & Ti Silicates 0.23 0.05 0.34 0.48
Ilmenite 0.19 0.01 0.01 0.00
Apatite 0.16 0.03 0.01 0.15
Garnet 0.48 0.03 0.06 0.13
Tourmaline 0.03 0.01 0.02 0.03
Zircon 0.05 0.00 0.04 0.06
Undifferentiated 0.07 0.00 0.05 0.25
Fig. 5. Modal mineralogy of samples conrmed as derived from the Altar Stone compared with the analysed ORS samples.
R.E. Bevins et al.
Journal of Archaeological Science: Reports 51 (2023) 104215
9
Fig. 6. Heavy mineral grain mounts (mounting medium n =1.54) displaying different heavy-mineral suites. MS-1 is dominated by transparent heavy minerals, with
common sub-rounded grains compared with WM 6 which shows abundant, larger, angular semi-opaque Fe-oxides and hydroxides with a different suite of HMs
dominated by corroded and etched garnets. Abbreviations (alphabetic order) follow Kretz (1983): Ant - Anatase; Ap - Apatite; Brt - Baryte; Car - Carbonate (un-
specied); Chl - Chlorite; Fe-Hydrox - Red and yellow-orange Fe-hydroxides; Grt - Garnet; Op - Opaque (unspecied); Rt - Rutile; Ti-Ox - White Ti-oxides; Zrn - Zircon.
Scale bar =100
μ
m.
Table 3
Frequency of heavy mineral grains in samples MS-1 and WM 6 from Raman Spectroscopy analysis. Numerical grain frequency counted by optical microscopy in
transmitted and reected light using a point counting method (PCm). Percentages of grain size fractions and of total heavy grains (%HM/Tot) included here. Per-
centage grain frequency recalculated from numerical count data. Zircon-Tourmaline-Rutile (ZTR) index after Hubert (1962). Percentage of total heavy grains (%HM/
Tot) and percentage of transparent heavy minerals in weight (tHM %weight) given, with varieties of tourmaline (schorl or dravite) indicated.
Numerical grain frequency Percentage Grain Frequency
Location Stonehenge West Midlands Location Stonehenge West Midlands
Sample Altar Stone ORS Sample Altar Stone ORS
Number MS-1 WM 6 Number MS-1 WM 6
Class analysed 2–500 um 5–500 um HM tot% 1.4 2.1
Counting method PCm PCm tHM %weight 0.5 0.7
Operator Sergio Ando Sergio Ando zircon 8.7 11.0
zircon 18 22 tourmaline 3.9 1.5
dravite 2 0 rutile 6.3 3.0
schorl 6 3 Ti Oxides 1.0 1.0
rutile 13 6 apatite 29.1 13.0
anatase 2 2 others 39.8 39.0
apatite 60 26 epidote 0.5 1.0
baryte 82 78 garnet 7.8 27.5
blue-green hornblende 2 0 amphibole 1.0 0.0
spinel 4 6 spinel 1.9 3.0
epidote 1 1 Total 100 % 100 %
clinozoisite 0 1 ZTR 19 16
garnet 16 55 % transparent HM 39 % 35 %
Total transparent 206 200 % opaque HM 0 % 16 %
opaques 2 94 % Fe Ox 0 % 15 %
Fe Ox-Hydrox 0 87 % Ti Ox 12 % 13 %
Ti Ox 65 77 % chlorite 15 % 9 %
chlorite 79 52 % biotite 2 % 5 %
biotite 10 28 % carbonates 5 % 1 %
carbonates 28 6 % “light minerals” 27 % 5 %
“light minerals” 141 28 Total 100 % 100 %
Total Opaque 325 372
Total (all) 531 572 zircon 8.7 11.0
<5 um (g) 0.000 0.141 dravite 1.0 0.0
5–500 um (g) 0.145 1.855 schorl 2.9 1.5
>500 um (g) 0.023 0.005 rutile 6.3 3.0
% ne tail cut (g) 0 % 7 % anatase 1.0 1.0
% class (g) 74 % 90 % apatite 29.1 13.0
% coarse tail cut (g) 12 % 0 % baryte 39.8 39.0
TOT excluded (g) 12 % 7 % blue-green hornblende 1.0 0.0
Total sieved (g) 0.1973 2.0515 spinel 1.9 3.0
Total used (g) 0.145 1.855 epidote 0.5 0.5
Light fraction (g) 0.133 1.769 clinozoisite 0.0 0.5
Dense fraction (g) 0.002 0.039 garnet 7.8 27.5
%HM/Tot 1.4 2.1 Total Transparent 100 % 100 %
R.E. Bevins et al.
Journal of Archaeological Science: Reports 51 (2023) 104215
10
where primary iron titanium oxides only comprise ne-grained sec-
ondary TiO
2
minerals (their original secondary ne-grained hematite
being lost). This is reected at the hand specimen/outcrop scale, WM 6
being a red sandstone whilst the Altar Stone is grey-green.
7. Discussion - where next?
Based on the pXRF sample screening and more detailed integrated
geochemistry and mineralogy all of the examined locations from the Old
Red Sandstone in the Anglo-Welsh Basin can be excluded as the source of
the Altar Stone. No other ORS locations with comparable Ba concen-
trations as observed in the Altar Stone are known in outcrops of the
Anglo-Welsh Basin, thus suggesting that we should perhaps exclude the
Anglo-Welsh Basin from further investigations, leading us to consider
broadening our horizons, both geographically and stratigraphically.
Key sedimentological characteristics of the Altar Stone are that it is a
very ne- to ne-grained, well sorted sandstone. Unidirectional ripple
cross lamination is present and is dened by the presence of subtle heavy
mineral laminae. At the thin section scale there is no apparent bio-
turbation and no fossils are recognised within the samples examined.
The sandstone has undergone moderate compaction, with the calcite
cement inferred to relate to burial diagenesis. However, there is no
apparent tectonic fabric (e.g. cleavage) implying that it has not under-
gone signicant deformation. In addition, there is also a lack of stylolites
or chemical dissolution surfaces which might result from such
deformation.
The overall dimensions of the Altar Stone at Stonehenge and its ge-
ometry (measuring 4.9 m long by 1 m wide by 0.5 m thick) suggest that
the original bed thickness must be >50 cm, with widely spaced (~5 m)
vertical joint sets; the tabular nature suggests that the original bed ge-
ometry has a tabular rather than strongly channelised or lenticular form.
Clearly, unidirectional ripple cross lamination can develop in a very
Fig. 7. Petrographic comparison of Sample WM 6 (A, B, E, F) with sample 2010 K 240 (C, D, G, H), which is proven to be derived from the Altar Stone. Images A, C, E,
G cross polarised light; B, D, F, H plane polarised light. Note the abundance of mudstone clasts in WM 6 when compared with 2010 K 240.
R.E. Bevins et al.
Journal of Archaeological Science: Reports 51 (2023) 104215
11
wide range of depositional settings, although the presence of the heavy
mineral laminae would be more consistent with a uvial depositional
system. The absence of trace or body fossils may also indicate that a non-
marine depositional setting is perhaps more likely. Overall, the sediment
characteristics and lack of evidence for tectonism or signicant meta-
morphism would suggest that the source of the Altar Stone was a post
Caledonian, non-marine Devonian or post-Devonian sandstone unaf-
fected by tectonism.
The detrital mineralogy of the Altar Stone is dominated by sub-
angular to sub-rounded monocrystalline quartz grains showing
straight extinction although rare, larger grains are strained. Plagioclase
is much more abundant than K feldspar. Lithic grains (rock fragments)
are the same size as the quartz and feldspar grains; most are internally
ne-grained and include siliceous “cherts”, polycrystalline metamorphic
quartz, phyllite and ne-grained sandstone, along with rare, ne-
grained graphic granite and quartz-chlorite intergrowths. Detrital
muscovite dominates over biotite. Heavy minerals identied optically
and through automated mineralogy and Raman Spectroscopy include Fe
oxides, chromite, rutile and Ti oxides, rare ilmenite, apatite, garnet
(mostly almandine and spessartine), tourmaline (mostly schorl) and
Fig. 8. Outline geological maps contain British Geological Survey materials © UKRI [2023] from BGS Make-a-Map (https://www2.bgs.ac.uk/discoveringgeology/
geology-of-britain/make-a-map/map.html). Stream sediment geochemistry data from the G-Base project, contains British Geological Survey materials © UKRI [2023]
from https://www.bgs.ac.uk/datasets/g-base-for-the-uk/.
R.E. Bevins et al.
Journal of Archaeological Science: Reports 51 (2023) 104215
12
zircon.
Geochemically, the Altar Stone is characterized principally by its
high Ba content, related to the presence of a baryte cement, with an
average Ba abundance of around 2800 ppm. This contrasts markedly
with the average of all of the ORS samples analysed from the Anglo-
Welsh Basin, with Ba averaging 437 ±293 ppm (range
128–2665 ppm). Baryte is a highly insoluble mineral, resistant to
chemical weathering, and for this reason Ba concentrations in stream
sediments often closely reect the underlying geology (British Geolog-
ical Survey, 2000; Everett et al., 2019). In the stream sediments lying on
the ORS in Wales, Ba concentrations are typically between the 5th-75th
percentile of the national range (Everett et al., 2019), being between 247
and 695 ppm, which compares well with the pXRF analyses (average
437 ±293 ppm, see Fig. 8). It is to be expected therefore that the strata
which sourced the Altar Stone would have elevated Ba. Fig. 8 shows
simplied geological maps for the UK which indicate the outcrop of
Devonian and Permo-Triassic rocks, these being predominantly non-
marine strata which are geologically consistent with the lithology of
the Altar Stone. These maps are superimposed on geochemical maps
highlighting areas where Ba in stream sediments exceeds 942 ppm
(>90th percentile), which may provide clues as to the source of the Altar
Stone. Also shown in Fig. 8 are sections of the detailed Ba stream sedi-
ments maps, which at a larger scale show some of the high-Ba regions
more clearly. The lack of high Ba over Devonian sequences in Wales is
clear, although there is high Ba on the north Somerset coast, but these
are deformed Middle Devonian marine sequences and thus could not be
the source of the Altar Stone. There is elevated Ba in Northern Ireland
(no Ba stream sediment data exists for the Republic of Ireland) over
dominantly conglomeratic Middle Devonian sediments. In NE Wales,
Cheshire and the Welsh Borderland, high Ba coincides with Permo-
Triassic sequences, as does local Pb-Zn-baryte mineralisation. Perhaps
of signicance is the fact that there is just a single signicant Neolithic
monument in the NE Wales/Cheshire/Merseyside area, that being the
remains of a megalithic tomb comprising six red sandstone orthostats
with carved motifs known as the Calderstones in Allerton near Liverpool
(Forde-Johnston, 1958). Interestingly, in this area during the Bronze
Age, Alderley Edge was the site of one of the earliest known copper
mines in Britain in baryte-bearing, sometimes pebbly, Triassic sedi-
ments. High Ba is recorded between Leeds and York, while on the
eastern edge of the Triassic sequences adjacent to the Pennine Fault
there is high Ba, the latter associated with mineralisation in the Pennine
ore eld (Colvine, 1995). Here, however, no signicant Neolithic
monuments occur in these areas, although further north in the Vale of
Eden area a string of monuments occurs, including the stone circle
known as Long Meg and her Daughters (Sharpe, 2022). Long Meg itself is
a 3.8 m monolith consisting of red sandstone, generally thought to have
been cut from cliffs of the Permian Penrith Sandstone Formation
bordering the River Eden close by.
Further north, in Scotland, non-marine Old Red Sandstones are more
abundant than Permo-Triassic sequences. On the east and west coasts of
the Isle of Arran, stream sediment Ba exceeds 942 ppm, associated with
Permian or ORS sequences (British Geological Survey, 1993). On the
west coast of Arran there are remains of a number of Neolithic stone
circles at Machrie Moor. Circles 2 and 3 are of pebbly red sandstone,
probably Permian in age and thought to be from nearby Auchagallon
(see Richards, 2013). Another high-Ba occurrence is above similar
Permian lithologies east of Prestwick, and there are a small number of
high-Ba locations above ORS strata in the Midland Valley (south of the
Highland Boundary Fault), but once again these are not associated with
known Neolithic contexts. What is interesting to note is that Hillier et al.
(2006) reported the presence of a dioctahedral interlayered chlorite-
smectite (tosudite) in sandstones of the ORS Strathmore Group in the
Midland Valley Basin. Tosudite also occurs in the Altar Stone sandstone
(authors’ unpublished data). North of the Highland Boundary Fault, ORS
strata occur in the Orcadian Basin, around the Moray Firth, through
Sutherland, and on Orkney and Shetland. Across these sandstone areas
the Ba stream sediment concentrations are all greater than the 50th
percentile (i.e. >533 ppm, yellow on Fig. 8) which contrasts with the
ORS strata in the Anglo-Welsh Basin, where the majority of Ba stream
sediment concentrations are below the 50th percentile range (i.e. below
533 ppm, green on Fig. 8) (Everett et al., 2019). There are high-Ba
stream sediment concentrations (>942 ppm) in several locations in
Caithness, in southwest Mainland Orkney and across Shetland. Orkney
contains among the nest Neolithic settlements and monuments in the
UK, including the Ring of Brodgar and the Stones of Stenness on Orkney,
both constructed using ORS age sandstone identied as being quarried
at Vestra Fiold and Staneyhill within a few kilometres of the stone circles
(Richards, 2013).
On Orkney, the basement granites underlying (unconformably) the
Old Red Sandstone in the southwest of Mainland have elevated Ba
(Lundmark et al., 2019 and see Fig. 8), although the Ba anomaly in this
area seems to relate to copper-uranium-rare earth element (plus baryte)
mineralization linked to an exhumed oil reservoir (Heptinstall et al.,
2023). In Shetland, high-Ba levels coincide with metamorphic and
plutonic rocks underlying the Devonian Walls and Sandness formations
(Melvin, 1985), which include, on Papa Stour, Shetland, some Middle
Devonian volcanic sequences. Basalts in these sequences have vesicles
inlled with baryte (up to 10 cm across as agate-baryte amygdales), a
testament to Ba-mobility in the basin during diagenesis of the Middle
ORS (Mykura and Newsier, 1976). A baryte cement in the ORS of these
areas may be expected and might account for the elevated stream
sediment Ba.
Thomas (1923) divided the stones used in the construction of
Stonehenge into two groups, namely the sarsens (of relatively local
origin) and the ‘foreign stones’ or the ‘blue stones’ and he included the
Altar Stone in his ‘blue stone’ group, hence assuming a common ‘cul-
tural’ origin. In doing so he sought a provenance for the Altar Stone in
west or south Wales, suggesting possible sources in the Cosheston Group
around Milford Haven or possibly from somewhere in the outcrop of the
Senni Beds in ‘Glamorganshire’ (now parts of the counties of Carmar-
thenshire, Powys and Monmouthshire). It is clear from his paper that he
thought all of the bluestones were from a single source area in west or
south Wales brought together within a unied effort.
This Wales source for the Altar Stone has remained unchallenged for
almost a century. The Altar Stone has frequently been referred to as an
anomalous bluestone, both in its lithology and in its size and weight. It is
also not known when it arrived at Stonehenge (M. Pitts, pers. comm.).
Parker Pearson (2023) considered that the bluestones (56 in number)
were erected in the set of stone holes known as the Aubrey Holes during
the Stage 1 construction phase (c. 2950 BCE). Because of its size the Altar
Stone would have looked at odds amongst a ring of smaller bluestones so
a possibility, to explain its anomalous characteristics, is that it arrived at
a different time and from a different source area to the bluestones. ‘De-
classifying’ the Altar Stone as a bluestone frees up thinking regarding a
potential source for the stone and has led us to consider that it is an
appropriate time to broaden our horizons, both geographically and
stratigraphically in our search for the source of the Altar Stone. This is
the next phase of our investigation, in which we will seek to try to match
the distinctive lithology, mineralogy and geochemistry of the Altar
Stone to Old Red Sandstone sequences across the other regions of Britain
described above, considering also younger strata of Permian and Triassic
age. We will be in part guided by tools such as the barium stream
sediment distribution maps, but we will also endeavour to gain an un-
derstanding of the source provenance of the Altar Stone component
minerals by extending the use of Raman Spectroscopy and also by age
determinations of those minerals. We intend also to collaborate with
archaeologists to explore the proposed long-distance links between
Stonehenge and other regions of Britain, such as evidence that cattle and
pigs feasted on at Durrington Walls were brought from western and
northern areas, including Scotland (Madgwick et al., 2019, but see Evans
et al., 2022). This long-distance connection occurred during the Stage 2
construction phase (c. 2500 BCE) so maybe the Altar Stone arrived
R.E. Bevins et al.
Journal of Archaeological Science: Reports 51 (2023) 104215
13
during this period, well after the bluestones were erected? The timing of
these links needs to be further explored in order to try to discover when
the Altar Stone arrived at Stonehenge. These considerations will inform
the next phase of our investigations.
8. Conclusions
For the last 100 years the Stonehenge Altar Stone has been consid-
ered to have been derived from the Old Red Sandstone sequences of
south Wales, in the Anglo-Welsh Basin, although no specic source
location has been identied. Our extensive sampling, petrographic ex-
aminations, portable XRF analyses, automated SEM-EDS investigations
and very preliminary Raman Spectroscopy have similarly failed to
provenance the stone. Indeed, only four samples from our dataset have
Ba levels comparable to those in the Altar Stone and more detailed in-
vestigations of those four samples discounts each sample and its location
as being linked to the source of the stone. We have concluded that the
Altar Stone appears not, in fact, to come from the ORS of the Anglo-
Welsh Basin and further, we propose that the Altar Stone should no
longer be included in the “bluestone” grouping of rocks essentially
sourced from the Mynydd Preseli. Accordingly, in our on-going pursuit
of the provenance of the Altar Stone we consider it time to broaden
horizons, both geographically and stratigraphically, to include parts of
Britain with evidence of Neolithic peoples and their monuments.
Attention will now turn to the ORS of the Midland Valley and Orcadian
Basins in Scotland as well as Permian-Triassic of northern England to
ascertain whether any of these sandstones have a mineralogy and
geochemistry which match the Stonehenge Altar Stone.
Declaration of Competing Interest
The authors declare that they have no known competing nancial
interests or personal relationships that could have appeared to inuence
the work reported in this paper.
Data availability
Data submitted as Supplementary tables
Acknowledgements
Salisbury Museum and Amgueddfa Cymru – National Museum Wales
are thanked for the loan of samples used in this study. We also thank
Keith Ray and Jon Morris for assistance with the collection of eld
samples. The senior author acknowledges nancial support for this work
from the Leverhulme Trust through award of an Emeritus Fellowship.
Finally, very helpful comments from two reviewers led to improvements
in this contribution.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.jasrep.2023.104215.
References
And`
o, S., 2020. Gravimetric separation of heavy minerals in sediments and rocks.
Minerals 10, 273. https://doi.org/10.3390/min10030273.
And`
o, S., Garzanti, E., 2013. Raman spectroscopy in heavy-mineral studies. In: Scott, R.
A., Smyth, H.R., Morton, A.C., Richardson, N. (Eds.), Sediment Provenance Studies
in Hydrocarbon Exploration and Production. Geological Society, London, Special
Publications, pp. 395–412. https://doi.org/10.1144/SP386.2.
And`
o, S., Vignola, P., Garzanti, E., 2011. Raman counting: A new method to determine
provenance of silt. Rendiconti Lincei 22, 327–347. https://doi.org/10.1007/s12210-
011-0142-4.
Barclay, W.J., Browne, M.A.E., McMillan, A.A., Pickett, E.A., Stone, P., Wilby, P.R. (Eds.),
2005. The Old Red Sandstone of Great Britain. Geological Conservation Review
Series, No. 31. Joint Nature Conservation Committee, Peterborough, p. 393.
Barclay, W.J., Davies, J.R., Hillier, R.D., Waters, R.A., 2015. Lithostratigraphy of the Old
Red Sandstone successions of the Anglo-Welsh Basin. British Geological Survey
Research Report RR/14/02, 96 pp.
Bevins, R.E., Ixer, R.A., Webb, P.C., Watson, J.S., 2012. Provenancing the rhyolitic and
dacitic components of the Stonehenge Landscape bluestone lithology: New
petrographical and geochemical evidence. J. Archaeol. Sci. 39, 1005–1019. https://
doi.org/10.1016/j.jas.2011.11.020.
Bevins, R.E., Ixer, R.A., Pearce, N.J.G., 2014. Carn Goedog is the likely major source of
Stonehenge doleritic bluestones: Evidence based on compatible element
discrimination and principal component analysis. J. Archaeol. Sci. 42, 179–193.
https://doi.org/10.1016/j.jas.2013.11.009.
Bevins, R.E., Pirrie, D., Ixer, R.A., O’Brien, H., Pearson, M.P., Power, M.R., Shail, R.K.,
2020. Constraining the provenance of the Stonehenge ‘Altar Stone’: Evidence from
automated mineralogy and U-Pb zircon age dating. J. Archaeol. Sci. 120 https://doi.
org/10.1016/j.jas.2020.105188.
Bevins, R.E., Ixer, R.A., Pirrie, D., Power, M.R., Cotterell, T., Tindle, A.G., 2021.
Alteration fabrics and mineralogy as provenance indicators; The Stonehenge
bluestone dolerites and their enigmatic spots. J. Archaeol. Sci. Rep. 36 https://doi.
org/10.1016/j.jasrep.2021.102826.
Bevins, R.E., Pearce, N.J.G., Ixer, R.A., Hillier, S., Pirrie, D., Turner, P., 2022a. Linking
derived debitage to the Stonehenge Altar Stone using portable X-ray uorescence
analysis. Mineral. Mag. 86, 688–700. https://doi.org/10.11180/mgm.2022.22.
Bevins, R.E., Pearce, N.J.G., Parker-Pearson, M., Ixer, R.A., 2022b. Identication of the
source of dolerites used at the Waun Mawn stone circle in the Mynydd Preseli, west
Wales and implications for the proposed link with Stonehenge. J. Archaeol. Sci. Rep.
45 https://doi.org/10.1016/j.jasrep.2022.103556.
Bevins, R.E., Pearce, N.J.G., Pirrie, D., Ixer, R.A., Hillier, S., Turner, P., Power, M.R.,
2023. Assessing the authenticity of a sample taken from the Altar Stone at
Stonehenge in 1844 using portable XRF and automated SEM-EDS. J. Archaeol. Sci.
Rep. 49 https://doi.org/10.1016/j.jasrep.2023.103973.
British Geological Survey, 1993. Regional geochemistry of southern Scotland and part of
northern England. British Geological Survey, Keyworth, Nottingham.
British Geological Survey, 2000. Regional Geochemistry of Wales and part of West
Central England: Sediment and Soil. British Geological Survey, Keyworth,
Nottingham.
Colvine, R., 1995. Regional Geochemistry of the Lake District and adjacent areas.
Geochemistry Group of the British Geological Survey, Keyworth, Nottingham
(British Geological Survey), 1993. viii+98 pp, 46 coloured maps+1: 250 000
geological map.
Cunnington, W., 1884. Stonehenge notes: The fragments. Wiltshire Archaeological and
Natural History Magazine 21, 141–149.
Darvill, T., 2022. Mythical rings? Waun Mawn and Stonehenge Stage 1. Antiquity 390
10.15184/aqy.2022.82.
Downes, J., Richards, C., Brown, J., Creswell, A.J., Ellen, R., Davies, A.D., Hall, A.,
McCulloch, R., Sanderson, D.C.W., Simson, I.A., 2013. Investigating the Great Ring
of Brodgar. In: Richards, C. (Ed.), Building the Great Stone Circles of the North.
Oxbow Books, pp. 90–118.
Evans, J.A., Pashley, V., Mee, K., Wagner, D., Parker Pearson, M., Fremondeau, D.,
Albarella, U., Madgwick, R., 2022. Applying lead (Pb) isotopes to explore mobility in
humans and animals. PLoS One 17 (10), e0274831. https://doi.org/10.1371/
journal.pone.0274831.
Everett, P.A., Lister, T.R., Fordyce, F.M., Ferreira, A.M.P.J., Donald, A.W., Gowing, C.J.
B., Lawley, R.S., 2019. Stream sediment geochemical atlas of the United Kingdom,
OR/18/048. British Geological Survey 94.
Forde-Johnston, J.L., 1958. Megalithic art in the north-west of Britain: The calderstones.
Liverpool. Proc. Prehist. Soc. 23, 20–39.
Garzanti, E., And`
o, S., 2007a. Heavy-mineral concentration in modern sands:
Implications for provenance interpretation. In: Mange, M.A. & Wright, D.T. (Eds.),
Heavy Minerals in Use. Developments in Sedimentology Series, vol. 58, pp. 517–545.
Garzanti, E., And`
o , S., 2007b. Plate tectonics and heavy-mineral suites of modern sands.
In: Mange, M.A. & Wright, D.T. (Eds.), Heavy Minerals in Use. Elsevier, Amsterdam,
Developments in Sedimentology Series, vol. 58, pp. 741–763.
GERM, 2021. Geochemical Earth Reference Model (GERM) Partition Coefcient (Kd)
Database, https://www.earthref.org/KDD/.
Heptinstall, E.A., Parnell, J., Armstrong, J.G.T., Schito, A., Akinsanpe, T.O., 2023.
Copper, uranium and REE mineralisation in an exhumed oil reservoir, southwest
Orkney. Scotland. Geosciences 13, 151.
Hillier, S., Wilson, M.J., Merriman, R.J., 2006. Clay mineralogy of the Old Red Sandstone
and Devonian sedimentary rocks of Wales. Scotland and England. Clay Miner. 41,
433–471.
Hubert, J.F., 1962. A zircon-tourmaline-rutile maturity index and independence of
composition of heavy mineral assemblages with gross composition and texture of
sandstone. J. Sediment. Petrol. 32, 440–450.
Ixer, R.A., Bevins, R.E., 2010. The petrography, afnity and provenance of lithics from
the Cursus Field, Stonehenge. Wiltshire Archaeological & Natural History Magazine
103, 1–15.
Ixer, R.A., Bevins, R.E., 2011. Craig Rhos-y-Felin, Pont Saeson is the dominant source of
the Stonehenge rhyolitic debitage. Archaeology in Wales 50, 21–31.
Ixer, R., Bevins, R., Turner, P., Power, M., Pirrie, D., 2019. Alternative Altar Stones?
Carbonate-cemented micaceous sandstones from the Stonehenge Landscape.
Wiltshire Archaeological and Nat. History Magazine 112, 1–13.
Ixer, R.A., Bevins, R.E., Pirrie, D., Power, M., Webb, P., 2022. Stonehenge Dacite Group
D: Fact or ction? Wiltshire Archaeological & Nat. History Magazine 115, 117–132.
Ixer, R.A., Bevins, R.E., Pirrie, D., Power, M., 2023. Treasures in the attic. Testing
Cunnington’s assertion that Stone 32c is the ‘type’ sample for Andesite Group A.
Wiltshire Archaeology and Nat. History Magazine 116, 1–15.
R.E. Bevins et al.
Journal of Archaeological Science: Reports 51 (2023) 104215
14
Kendall, R.S., 2017. The Old Red Sandstone of Britain and Ireland — A review. Proc.
Geol. Assoc. 128 (3), 409–421. https://doi.org/10.1016/j.pgeola.2017.05.002.
Kretz, R., 1983. Symbols for rock forming minerals. Am. Mineral. 68, 277–279.
Linares-Catela, J.A., Romero, T.D., Molina, C.M., Puro, L.M.C., 2023. Choosing the site,
getting the stones, building the dolmens: Local sourcing of andesites at the El
Pozuelo megalithic complex (Huelva, Spain). Archaeol. Anthropol. Sci. 15, 101.
https://doi.org/10.1007/s12520-023-01799-0.
Lundmark, A.M., Augland, L.E., Bjerga, A.D., 2019. Timing of strain partitioning and
magmatism in the Scottish Scandian collision, evidence from the high Ba–Sr Orkney
granite complex. Scott. J. Geol. 55, 21–34. https://doi.org/10.1144/sjg2018-001.
Mange, A., Maurer, H.F.W., 1992. Heavy minerals in colour. Chapman and Hall, London,
p. 147.
Maskelyne, N.S., 1878. Stonehenge: The petrology of its stones. Wiltshire Archaeological
and Nat. History Magazine 17, 147–160.
Melvin, J., 1985. Walls Formation, western Shetland: Distal alluvial plain deposits within
a tectonically active Devonian basin. Scott. J. Geol. 21, 23–40. https://doi.org/
10.1144/sjg21010023.
Mykura, W., Newsier, J., 1976. The geology of western Shetland (Explanation of one-
inch geological sheet western Shetland; comprising Sheet 127 and parts of 125, 126
and 128). Keyworth, Nottingham.
Nash, D.J., Ciborowski, T.J.R., Ullyot, J.S., Parker Pearson, M., Darvill, T., Greaney, S.,
Maniatis, G., Whitaker, K.A., 2020. Origins of the sarsen megaliths at Stonehenge.
Sci. Adv. 6 https://doi.org/10.1126/sciadv.abc0133.
Parfenoff, A., Pomerol, C., Tourenq, J., 1970. Minerals in grains: methods of study and
determination. Masson et Cie (Editeurs). 120, Boulevard Saint-Germain, Paris (6e),
550p.
Parker Pearson, M., 2023. Stonehenge: A brief history. Bloomsbury Academic, p. 191 pp.
Parker Pearson, M., Bevins, R.E., Ixer, R.A., Pollard, J., Richards, C., Welham, K., 2020.
Long distance landscapes: from quarries to monument at Stonehenge. In:
Boaventura, R., Mataloto, R., Pereira (Eds.), Megaliths and Geology. Archaeopress,
Oxford, pp. 183–200.
Pearce, N.J.G., Bevins, R.E., Ixer, R.A., 2022. Portable XRF investigation of Stonehenge
bluestone 62 and potential source outcrops in the Mynydd Preseli, west Wales.
J. Archaeol. Sci. Rep. 44 https://doi.org/10.1016/j.jasrep.2022.103525.
Pirrie, D., Power, M.R., Rollinson, G.K., Wiltshire, P.E.J., Newberry, J., Campbell, H.E.,
2009. Automated SEM-EDS (QEMSCAN) mineral analysis in forensic soil
investigations; testing instrumental variability. In: Ritz, K., Dawson, L., Miller, D.
(Eds.), Criminal and Environmental Soil Forensics. Springer, pp. 411–430.
Pitts, M.W., 2022. How to build Stonehenge. Thames & Hudson Ltd, London, p. 240.
Richards, C., 2013. Building the Great Stone Circles of the North. Oxbow Books, Building
the Great Stone Circles of the North, p. 322.
Richards, C., Brown, J., Jones, S., Hall, A., Muir, T., 2013. Monumental risk: megalithic
quarrying at Staneyhill and Vestra Fiold, Mainland, Orkney. In: Richards, C. (Ed.),
Building the Great Stone Circles of the North. Oxbow Books, pp. 119–148.
Rollinson, H.R., Pease, V., 2021. Using geochemical data: To understand geological
processes, 2nd edn. Cambridge University Press, Cambridge, p. 346.
Sharpe, K.E., 2022. East of Eden: monumental rock art in Cumbria, England. In: Mazel,
A., Nash, G. (Eds). Signalling and performance: Ancient rock art in Britain and
Ireland. Archaeopress Archaeology, pp. 96–121.
Thomas, H.H., 1923. The source of the stones of Stonehenge. Antiqu. J. 3, 239–260.
https://doi.org/10.1017/S0003581500005096.
Thorpe, R.S., Williams-Thorpe, O., Jenkins, D.G., Watson, J.S. with contributions by Ixer,
R.A., Thomas, R.G., 1991. The geological sources and transport of the bluestones of
Stonehenge, Wiltshire, UK. Proceedings of the Prehistoric Society, 57, 103–157. Doi:
10.1017/S0079497X00004527.
Zimmermann, U., Kristoffersen, E.S., Fredriksen, P.D., Bertolino, S.A.R., And`
o, S.,
Bersani, D., 2016. Provenance and composition of unusually chrome and nickel-rich
bucket-shaped pottery from Rogaland (southwestern Norway). Sed. Geol. 336,
183–196. https://doi.org/10.1016/j.sedgeo.2015.09.001.
R.E. Bevins et al.
