You are on page 1of 21

Non-destructive provenancing of

bluestone axe-heads in Britain


Olwen Williams-Thorpe1, P.J. Potts1 & M.C. Jones2

The authors present a new procedure for discovering where stone artefacts come from without
having to cut a slice through them. The method is tested on axes of spotted dolerite bluestone from
Preseli in Wales, source of monoliths at Stonehenge.
Keywords: stone axe-heads, provenancing, magnetic susceptibility, X-ray fluorescence
See http://antiquity.ac.uk/ProjGall/williamsthorpe/ henceforward ‘web’ for the tables containing
the data which supports this paper.

Method
Introduction to non-destructive characterisation of axes
Scientific provenancing of stone axe-heads has become one of the key elements in assessing
trade and contact in prehistoric Britain (e.g. Bradley & Edmonds 1993). However, stone axe
studies using the traditional method of petrographic thin sectioning (mainly work by the
British Implement Petrology Committee; e.g. Clough & Cummins 1988) have suffered both
from the necessity of removing a slice or core from the implements, and also from a growing
recognition that petrography alone does not always provide an unambiguous answer to
questions of provenance (e.g. Fell & Davis 1988; Williams-Thorpe et al. 2003).
In order to address these problems, we have adapted and characterised at the Open University
several methods of non-destructive analysis for archaeological stone artefacts. The two
techniques that are most appropriate for stone axe studies are geochemical analysis using
portable X-ray fluorescence (PXRF; Williams-Thorpe et al. 1999); and magnetic susceptibility
measurements using a hand-held kappameter (Williams-Thorpe et al. 2003). The required
equipment is readily portable and therefore analyses can be made in situ at museums or in
the field. The facility with which the measurements can be made and, above all, the entirely
non-destructive nature of the work, has resulted in a considerable increase in the number of
artefacts that are made available to us for study (we estimate an increase of typically around
ten-fold over our previous studies that relied on destructive sampling of artefacts). The
geochemical and geophysical data that are obtained can be compared with analogous data on
potential outcrop sources, and considered in combination with petrographic information
derived from hand specimens, and from existing thin sections where these are available.
An important factor in our interpretation of magnetic susceptibility and PXRF data is the
use of a number of statistical techniques to support the comparison of artefact and source
data. In particular, an atypicality index has been developed to quantify the plausibility of an

1 Department of Earth Sciences, The Open University, Milton Keynes MK7 6AA, UK
2 Department of Statistics, The Open University, Milton Keynes MK7 6AA, UK

Received: 4 February 2003; Accepted: 9 September 2003

359
Olwen Williams-Thorpe, P.J. Potts & M.C. Jones

artefact originating at a particular source (Jones & Williams-Thorpe 2001). This index is
especially useful for our application, because it allows robust comparison of data that have
different analytical and sampling uncertainties (PXRF data may have larger uncertainties
than the source data with which they are compared). While statistics cannot always provide
unambiguous answers to questions of provenance, its use in this context introduces an
important element of objectivity in assessing, and comparing, artefact outcrop sources.
Notwithstanding the usefulness of these non-destructive methods, it is important to
recognise that there are limitations in their application. The surface characteristics of
unprepared rock or artefact surfaces – variations in relief and mineralogy, and weathering
layers – are of especial importance in PXRF, where most of the analytical signal originates
within the 1-2mm or so of material that is nearest to the analyser (magnetic susceptibility is
rather less affected by such factors, because the method assesses a greater depth of material,
around 20mm).
The presence of weathering layers may cause particular problems, because weathering
processes can alter the mineralogical and chemical composition of rocks from the original
(bulk rock) composition that is frequently required for provenancing purposes. We are
currently investigating ways of quantifying, and compensating for, the analytical effects of
weathering in certain rock types (Bernardini et al. in prep.).
The vexed question of whether petrographic or non-destructive (chemical and magnetic)
characterisation is more ‘successful’ for British stone axe provenancing has no easy answer.
Both methods rely heavily on the availability of a comparative database for potential sources,
so a comparison of robustness of technique is only meaningful for cases where such a database
exists for both petrography, and for chemical and magnetic characteristics. Williams-Thorpe
(in press, in Potts & West) commented on the degree of concordance between the two
approaches, and she broadly supported the view of Davis (1985) that the robustness of
petrographic assignments depends on the mineralogical distinctiveness of the rock type
considered. Yet that view appears to apply more truly to the numerically larger, and therefore
more commonly encountered, axe groupings: English Lake District axes (Implement Petrology
Committee (IPC) Group VI) and Whin Sill implements (Group XVIII), for example. As we
show below, the supposedly petrographically distinctive, but numerically much smaller, spotted
dolerite bluestone grouping (Group XIII) displays a startling lack of agreement between the
existing petrographic assignments and the new provenance information provided in this
paper.
The IPC petrographic database on stone axes, produced over a period of more than fifty
years by researchers, both professional and dedicated amateur, remains an invaluable resource
to anyone working in the field. However, the conclusions incorporate the particular range of
experience and knowledge of individuals, as well as the changing comparative petrographic
databases available to them. The numerical data that we use, on the other hand, are
reproducible, and therefore more easily scrutinised by other workers. And in the final analysis,
petrographic provenancing using thin sections still involves destructive sampling of
irreplaceable artefacts. Our non-destructive techniques do not.
Much of our work so far using PXRF and magnetic susceptibility has concentrated on
stone axes. However, the methods have wide applicability to other lithic artefacts, and we

360
Non-destructive provenancing of bluestone axe-heads in Britain

have successfully employed magnetic susceptibility and PXRF for the characterisation,
respectively, of granite, and of porphyry artefacts and quarry material (Williams-Thorpe &
Thorpe 1993; Williams-Thorpe et al. 2001). Using the methods in tandem (Williams-Thorpe
et al. 2003) or in combination with other non-destructive techniques (Williams-Thorpe et
al. 2000a), can further enhance their effectiveness in provenancing applications. And beyond
lithic studies lie further possibilities of employing PXRF, in particular, for the characterisation
of other classes of artefact including ceramics, glasses and metals.

Method

Figure 1. Map showing the find locations of reported Group XIII implements (as listed by Clough & Cummins
1988), and of five other archaeological pieces studied for this work (Car30, Car31, Car34, SilF and Wi389).
The list of axes examined is given in Table 1 (http://antiquity.ac.uk/ProjGall/williamsthorpe).

361
Olwen Williams-Thorpe, P.J. Potts & M.C. Jones

Bluestone axes
We now apply our methods to the investigation of sources of British spotted dolerite ‘bluestone’
(IPC Group XIII) axes and related artefacts (‘related artefacts’ means axe-hammers, battle-
axes, maces and adzes; the term ‘axe’ is commonly used rather loosely to refer to all these
types, though ‘implement’ is, strictly, a more correct term).
Polished stone axes manufactured from the spotted dolerite which crops out in the Preseli
Mountains of south Wales are of special interest in studies of British prehistoric implements,
because this rock was also used for many of the bluestone monoliths of Stonehenge. The
distribution of spotted dolerite artefacts in Britain has long been used as supporting evidence
for the human transport hypothesis of the Stonehenge bluestones (Atkinson 1979; Green
1997), yet this element in the Stonehenge controversy has never been examined or tested.
The petrological identifications of many spotted dolerite axes were made during the first
half of the twentieth century (e.g. Stone & Wallis 1951) without the benefit of more recent
developments in provenancing techniques. Now that the ability of geochemical methods to
provide more specific data for source attributions has been demonstrated (e.g. Mandal 1997),
and those data can be derived non-destructively (e.g. Williams-Thorpe et al. 1999), it is
appropriate to re-examine all of the thirty or so reported spotted dolerite implements.
In this way we shall re-assess the distribution from source of this axe group using modern
methods, and also test whether the original classification of these axes is sufficiently robust to
support the Stonehenge human transport hypothesis.

Archaeological background of spotted dolerite axe Group XIII, and


the implements studied
Group XIII was defined by archaeologists, in consultation with petrologists, in 1951 (Stone
& Wallis 1951) to provide a framework for some dozen implements including axes, an axe-
hammer, and a mace that had all been identified as ‘preselite’, a term adopted to mean the
spotted facies of the Preseli dolerite intrusions. These, and subsequent, identifications were
made mainly on the basis of mineralogical features observed in petrological thin sections
(thus requiring a small sample to be sawn, or drilled, out of the implements), though occasional
identifications were made on the basis of hand specimen examination only (cf. Clough &
Cummins 1988: 157).
Preseli spotted dolerite has an unusual appearance in hand specimen, characterised by
whitish or pinkish metamorphic spots often up to 10 mm or so in diameter. In view of this,
it might be thought surprising that there are, within the published petrological reports of the
IPC, a number of revisions to the assignments of artefacts to Group XIII. For example, an
axe found in Crawley, Sussex (Sussex 6, IPC serial number 156) appears as ‘ophitic dolerite’
in 1951 (Stone & Wallis 1951: 142) and ‘dolerite’ in 1962 (Evens et al. 1962: 245), but by
1988 has been assigned to ‘Group XIII’ (Clough & Cummins 1988: 165). At least six further
instances of uncertain, or subsequently altered, published attributions to Group XIII may be
found, suggesting that ‘preselite’ is not always unmistakably distinctive in hand specimen
and/or in thin section. It is, indeed, the case that the supposedly diagnostic ‘spots’ may be
rather difficult to see on a slightly weathered artefact surface, or may be confused with white
crystals of different minerals. The Preseli metamorphic spots are also often much less obvious

362
Non-destructive provenancing of bluestone axe-heads in Britain

in thin section than in hand specimen (the spots are distinguished in microscope view only
as finer-grained areas comprising altered plagioclase feldspar and quartz inter-growth; Bevins
et al. 1989 and cf. Thorpe et al. 1991: 127).
Some of the difficulty in Group XIII assignments may, perhaps, be due to the introduction
in 1972 of ‘Group XXIIIb’, to include artefacts of non-spotted dolerite whose source was also
suggested to be in Preseli (Shotton 1972). Group XIII is now deemed to be ‘within the
petrographic and geographic range’ of the (broader) Group XXIII (Clough & Cummins
1988: 8-9). The fact that the characteristic metamorphic spots can, in some outcrops, be so
dispersed as to make possible the manufacture of an axe in which the spots are ‘missed’,
underlines the difficulties faced by IPC workers. It also strengthens the case for using
geochemical and other modern characterisation methods, which may provide provenance
interpretations which are less dependent on rock appearance and ‘spottedness’.
The effectiveness of geochemical characterisation for Group XIII has already been
demonstrated, since six reported Group XIII implements were analysed in previous studies
by Williams-Thorpe et al. (1999) and Jones and Williams-Thorpe (2001). The fact that four

Method
out of these six implements appeared not to be of Preseli spotted dolerite has, in large measure,
prompted the present study.
Cooney and Mandal (1995), working within the Irish Stone Axe Project, have suggested
on geochemical grounds that three further axes found on Lambay Island, near Dublin, may
have originated in Preseli (Cooney & Mandal op.cit.: 973, 978-9). An Irish link had earlier
been suggested by Alexander Keiller, who in 1936 identified two axes found in County
Antrim as ‘Presely rock’. However, these two axes were demoted to the rather less specific
types of ‘dolerite’ and ‘greenstone’, respectively, by Evens et al. (1962: 243).
Figure 1 shows the position of the Preseli area, and the find locations of the twenty-eight
axes and other artefacts assigned to ‘Group XIII’ and ‘Group XIII?’ by Clough and Cummins
(1988). We have now examined, and analysed, a total of twenty-four of these artefacts. The
remaining four are artefacts which we were unable to locate, being either in private collections,
lost, or stolen.
Although the core part of our project was to study those artefacts listed as Group XIII or
XIII? in Clough and Cummins (op. cit.), we also report here results for four additional
artefacts which had been linked to Group XIII in earlier IPC record cards or publications
(Car30, 31, 34 and Wi389), and an unworked flake that appeared in hand specimen to be
spotted dolerite (SilF). The find locations of these five pieces are also shown on Figure 1.
The studied artefacts include unperforated axes as well as shaft-hole implements (axe-
hammers, battle-axes, maces; implement types are noted on Figure 1). Archaeological contexts
are rarely recorded, the most secure being the Beaker and Bronze Age associations of Wiltshire
302 (Wilsford) and Radnorshire 4 (Clap yr Arian), respectively (Evens et al. 1962: 219;
Grimes 1935:278). The ten unperforated axes, or fragments thereof, suggest the inclusion
within Group XIII of earlier material, since such implements are generally assumed to be
Neolithic in age.
Figure 2 shows Radnorshire 4, the battle-axe from Clap yr Arian. The whitish metamorphic
spots are particularly clear on the left-hand side Figure 3 illustrates the manner of PXRF
measurement.

363
Olwen Williams-Thorpe, P.J. Potts & M.C. Jones

Characterisation of
the implements
Measurement procedures
The only preparation
carried out prior to analysis
was gentle wiping of the
implements with de-ionised
water, in order to remove
surface dirt and dust.
Figure 2. Photograph of the spotted dolerite battle-axe Radnorshire 4 from Clap yr Arian.
Magnetic susceptibility
measurements require the
stone axe or other artefact to
be held gently against the
active end of the instrument
(a KT-5 kappameter from
Exploranium G. S. Ltd.) for
about two seconds. A small
magnetic field is applied to
the measured object, from
which the instrument
computes and displays the
magnetic susceptibility, that
is the induced magnetism,
of the rock. Several
measurements (typically
between five and ten) are
made on each implement,
depending on its size and
shape. The minimum
sample size for effective
measurement is c. 50mm
across and 20mm thick, and
the maximum surface relief
or curvature must be less
than 10mm. Simple
arithmetic corrections are
routinely applied to Figure 3. The measurement procedure using the PXRF
magnetic susceptibility equipment. Photos from the authors, reproduced by permission
readings, in order to of Radnorshire Museum (Amgueddfa Sir Faesyfed).
compensate for non-standard sample size, and for irregular surfaces. The recommended
procedures for measurement and data correction of lithics are described in a series of papers
from the Open University, and procedures especially relevant to axes and similar artefacts are
in Williams-Thorpe et al. (2000b; 2003 and references therein), and Markham (2000).

364
Non-destructive provenancing of bluestone axe-heads in Britain

PXRF measurements were carried out in the manner illustrated above in Figure 3, by
placing the implements in turn on the PXRF probe. In this method, a surface area of 25mm
diameter is exposed to X-rays from three radioisotope sources which are rotated in front of
the sample in sequence within the PXRF instrument. Fluorescent X-rays are detected by a
solid state (HgI2) detector. Element concentrations are computed by the instrument and
displayed immediately after measurement. Counting times for our work on spotted dolerites
were 340 s per measurement, and at least five measurements were made on most of the
implements. Quantitative data were obtained for five major and minor elements, and a
range of trace elements (Table 1, web). Recommended procedures, including simple safety
precautions, for lithic analysis are given in Williams-Thorpe et al. (1999) and Williams-
Thorpe in press, in Potts & West), while instrumental and analytical considerations are in
Potts et al. (1995; 1997a; 1997b). Like the magnetic susceptibility data, PXRF analytical
results require routine arithmetic corrections, to apply instrumental calibration factors and
to compensate for uneven sample surface relief. The derivation and application of these
corrections are detailed in the series of papers by Potts et al. referred to, but are also summarised,

Method
together with the measurement procedures, by Williams-Thorpe (in press op. cit.).
The weathering depths on the axes and other implements studied (cf. discussion above)
were observed to be variable, but mainly much less than 1 mm (weathered depths could be
assessed to some extent in hand specimen, and were confirmed for several Group XIII axes
by examination of the existing thin sections, which are most often taken from axe surfaces).
Since each PXRF measurement assesses a surface area of only about 25mm in diameter, we
were able to select the freshest-looking implement surfaces for analysis, using any recently
abraded or broken areas to aid in our assessment of weathering.

Results
The chemical analyses for the objects analysed for this paper (n=23) together with their
magnetic susceptibilities are given in Table 1, web. Chemical analyses for a further six Group
XIII artefacts were reported by Jones and Williams-Thorpe (2001), but since that paper did
not give magnetic susceptibility, these magnetic data are now presented in Table 1.

Characterisation of the Preseli source area


In order to assess the provenance of reported Group XIII implements, analogous data on the
source area in Preseli are required. Chemical data from a wide range of spotted dolerite
outcrops are already available in the published geological literature (summarised in Jones &
Williams-Thorpe 2001; see references therein for original data sources). However, we found
no existing magnetic susceptibility data for Preseli. As part of the present project, therefore,
a new susceptibility survey of the outcrops was carried out. Fourteen outcrops (twelve named
on O.S. maps and two small, unnamed outcrops) were measured, giving a total of 861
readings. The measurements were taken roughly on topographic grids established at outcrops
by pacing, and using as guides aerial photographs at a scale of 1:5000 (Millenium Maps™
from Getmapping™). This ensured as comprehensive coverage of outcrops as was possible,
bearing in mind that some parts are precipitous and not accessible.
The magnetic susceptibility measurements are shown in Figure 4.

365
Olwen Williams-Thorpe, P.J. Potts & M.C. Jones

The susceptibility of the


rocks is generally fairly low,
mainly in the range 0.40 –
0.90 x 10-3 SI, with a peak
at around 0.50 x 10-3 SI and
a small number of readings
between 1-2 x 10 -3 SI.
However, we recorded a
small number of readings
(n=8), in a restricted part of
the eastern end of the Preseli
outcrops, which were
anomalously high – up to
103 x 10-3 SI. These were
noted within two small areas
of rock (each a few metres
across), and possibly
represent mineralised parts
of the outcrops (a
Figure 4. Histogram showing magnetic susceptibility of Preseli spotted dolerite outcrops
geochemical drainage survey (n=853).
of Preseli suggested the
presence of some base metal mineralisation in this region; Cameron 1988). We cannot be
sure whether other such anomalous areas of rock might exist elsewhere in Preseli, but we
found none within the other 853 measurements. It seems reasonable, particularly as
mineralised rocks are unlikely candidates for implement manufacture, to regard the
distribution of readings shown on Figure 4 as representative of (un-mineralised) Preseli spotted
dolerite.

Comparison of implements with the Preseli source area


To establish the degree of similarity between the artefacts and the Preseli spotted dolerites,
we examine, firstly, the magnetic susceptibility information, and, secondly, the chemical
data, using in each case statistical atypicality calculations (based on the procedure described
by Jones & Williams-Thorpe 2001) to compare artefacts with source. Since the atypicality
probabilities underpin much of the interpretation made in this paper, the method of their
calculation is given in the Appendix.

Magnetic susceptibility data


Several of the artefacts have high magnetic susceptibilities: Wiltshire 10, for example, from
Windmill Hill, has a magnetic susceptibility of over 50 x 10-3 SI, and Pembrokeshire 13
(Nevern), Carmarthen 31 (Laugharne), and Carmarthen 34 (Marchgwyn) have susceptibilities
greater than 10 x 10-3 SI. These measurements are, therefore, immediate indications that
some artefacts are unlikely to be from Preseli.

366
Non-destructive provenancing of bluestone axe-heads in Britain

Atypicality probabilities and Mahalanobis-type distances (cf. Jones & Williams-Thorpe


op. cit.) for twenty-nine archaeological artefacts, compared with 853 Preseli measurements,
are given in Table 2 (web).
The lower the atypicality probability, the less typical the artefact is of the source (cf. Jones
& Williams-Thorpe 2001: 11). Atypicalities in Table 2 of > 0.1 give no reason to doubt that
the source is Preseli, but atypicalities < 0.01 indicate that a Preseli source is unlikely. An
atypicality of between 0.1 and 0.01 is more difficult to interpret – based on statistical criteria,
such measurements are expected to occur very occasionally within the source; in these cases,
other evidence (chemistry, petrography) may help to clarify the source.
The four artefacts which have the highest magnetic susceptibilities, together with Dumfries
8 (Watcarrick), have probabilities of much less than 0.01, and, therefore, are very unlikely to
belong to the Preseli distribution. Carmarthen 30 (Garn Wen, magnetic susceptibility 1.01
x 10-3 SI) has an atypicality probability of 0.09. For the remaining artefacts, the magnetic
susceptibility data are consistent with possible origin within Preseli, though at varying distances
from the centre of the source distribution (cf. Table 2, column 2, the Mahalanobis-type

Method
distances).

Chemical data
The PXRF chemical analyses of the artefacts (Table 1) were compared with the Preseli database
of forty-five rocks previously used by Jones and Williams-Thorpe (2001: 4-5).
In the first instance, it is useful to examine bi-variate graphs using key discriminants plotted
relative to Zr, an element which is generally immobile in rocks and which is frequently used
to illustrate geochemical discrimination (cf. e.g. Rollinson 1993). Figures 5 and 6 include
data for all the archaeological pieces analysed for this work, and also for those six previously
analysed (Williams-Thorpe et al. 1999; Jones & Williams-Thorpe 2001).
Thus, from Figures 5 and 6, it is apparent that several of the artefacts lie outside the
analytical field of Preseli rocks. Artefacts with concentrations of Zr that are higher than those
of Preseli (in which Zr reaches a maximum value of ca. 100 ppm) are of particularly doubtful
Preseli origin. These include Pembrokeshire 13, Carmarthenshire 31, Wiltshire 10,
Carmarthenshire 34, Essex 23 and Suffolk 94. Examination of the other elements plotted in
Figure 5 (Nb, Y, Ba and Sr) shows further outlying and marginal samples, including Anglesey
10, Merioneth 8, Dumfriesshire 8, Cheshire 11, Shropshire axe-hammer 61, Carmarthenshire
30, Sussex 6, Sussex 178 and Cardiganshire 25.
The elements plotted in Figure 5 include those likely to be most representative of whole
rock composition in PXRF of dolerite artefacts (cf. Williams-Thorpe et al. 1999: 224). Y was
shown by Williams-Thorpe et al. (op. cit.) to be subject to only minor depletion in slightly
weathered artefact surfaces, but this depletion might, nevertheless, help to explain the apparent
offset between Preseli rocks and artefacts (Figure 5(b)) which, in other graphs of Figure 5
and 6, are within Preseli limits. Nb concentrations in many of the artefacts are around detection
limits (6 ppm; Potts et al. 1995), but the graph (Figure 5(a)) usefully serves to demonstrate
chemical differences between some higher Nb artefacts, and Preseli.

367
Olwen Williams-Thorpe, P.J. Potts & M.C. Jones

An atypicality index was used to


compare the chemistry of the
artefacts with that of Preseli.
While still following in general the
procedure described in Jones and
Williams-Thorpe (2001), we have
now extended the exercise to
include a larger number of
elements in addition to the four
(Zr, Nb, Y, Sr) which, for a variety
of analytical and geochemical
reasons, were identified by Jones
and Williams-Thorpe (op. cit.) as
most useful for their case. Thus,
in Table 3, atypicality probabilities
for each artefact in relation to
Preseli are given for four, six, and
seven elements in turn. The
atypicalities should be interpreted
in the same way as for the
magnetic susceptibility data, that
is, a very low atypicality
probability casts doubt on a Preseli
source.
Suffolk 94, Wiltshire 10,
Carmarthenshire 30, 31 and 34,
are immediately identifiable as
extremely unlikely to belong to the
Preseli distribution (Table 3,
column 2), with atypicalities near
zero (0.0005 or < 0.0005). The
inclusion of Fe and Rb in the
atypicality calculations (Table 3,
column 3) adds Cardiganshire 25,
Cheshire 11, Dumfries 8, Surrey
34 and Sussex 178 to the list of
very unlikely Preseli artefacts
(again, with atypicality
probabilities of 0.0005 or <
0.0005). And when Ba is also
included (Table 3, column 4),
Shropshire axe-hammer 61 is also Figure 5 a&b. Graphs showing axes and other archaeological pieces studied
far from the Preseli distribution. (PXRF data; error bars are typical sd of the mean), together with rocks of the
Preseli spotted dolerite area (data from the literature, see text). Named samples
are discussed in the text. County abbreviations are given in Table 1.

368
Non-destructive provenancing of bluestone axe-heads in Britain

Sussex 6 and SilF are doubtful (a)


Preseli members, with lowest
atypicalities of 0.0073 and 0.0018
respectively. The result for SilF is
curious, because this piece (an
unworked flake from Silbury Hill)
appears very similar in chemistry
to Wiltshire 108 and 109 (cf.
Table 1), which are ‘allowed’ by
atypicality as Preseli. The low
(depleted, or perhaps poorly
determined because of sampling
uncertainty) Y concentration in
SilF (12 ppm; n=2), together with
the negative Rb (- 4 ppm) may be

Method
having an undue effect on the
atypicality probability in this case.
The remaining artefacts analysed
for this work – Car23, Dev1,
Do41, Ha41, Ra4, and Wi 108,
109, 118, 302 and 389 – all have
compositions consistent with an (b)
origin within the Preseli dolerites.
Results for the six artefacts
previously analysed (and presented
here with atypicalities for a larger
number of chemical elements) are
mainly in line with our previous
conclusions: Essex 23 and
Pembrokeshire 13 are unlikely to
be from Preseli; Anglesey 10 and
Essex 29 have chemistry that
could originate in Preseli; while
Merioneth 8 differs from from
Preseli mainly in its anomalously
high Sr concentration, but is very
difficult to interpret because of its
single PXRF measurement. Only
for Monmouthshire 8 do our new
atypicalities make a qualitative
difference; this implement now
appears less typical of Preseli. Figure 6 a&b. Graphs showing axes and other archaeological pieces studied
(PXRF data; error bars are typical sd of the mean), together with rocks of the
Preseli spotted dolerite area (data from the literature, see text). Named samples
are discussed in the text. County abbreviations are given in Table 1.

369
Olwen Williams-Thorpe, P.J. Potts & M.C. Jones

Table 3. Atypicality probabilities for elemental data for axes and other archaeological pieces relating
to Group XIII, in comparison with outcrops of the Preseli spotted dolerite area
IPC/Other reference Using Zr Y Sr Using Zr Y Sr Nb Rb Using Zr Y Sr Nb
and Nb and Fe Rb Fe and Ba
Analysed for this paper
Ca25 0.0156 0.0001 0.000
Car23 0.1890 0.3813 0.5013
Ch11 0.0343 0.0005 0.0010
Dev1 0.1649 0.0939 0.1139
Do41 0.3670 0.3052 0.1284
Dmf8 0.0663 0.0000 0.0001
Ha41 0.1584 0.3056 0.4170
Ra4 0.6704 0.1842 0.1771
Shah61 0.0351 0.0024 0.0000
S94 0.0000 0.0000 0.0000
Su34 0.4488 0.0001 0.0003
Sx6 0.4049 0.0375 0.0073
Sx178 0.0016 0.0000 0.0000
Wi 10 0.0000 0.0000 0.0000
Wi108 0.2256 0.2778 0.2326
Wi109 0.3541 0.1616 0.1463
Wi118 0.2887 0.4888 0.3609
Wi1302 0.7355 0.8089 0.7134
Car30 0.0005 0.0000 0.0000
Car31 0.0000 0.0000 0.0000
Car34 0.0000 0.0000 0.0000
Wi389 0.0901 0.1738 0.2609
SilF 0.0082 0.0018 0.0039
Analysed by Jones and Williams-Thorpe (2001)
An10* 0.0385 0.0370 0.0547
E23 0.0831 0.0003 0.0007
E29 0.4095 0.1146 0.1462
M8 0.0053 0.0045 0.0065
Me8* ** 0.0000 0.0002 0.0005
P13 0.0000 0.0000 0.0000
Notes: county abbreviations as in Table 1; * based on one PXRF measurement per axe; ** ignoring anomalously high
Sr, atypicalities are 0.0117, 0.0219 and 0.0311 respectively (for 3, 5 and 6 elements); standard errors used in
atypicality calculations are derived from repeat axe measurements (cf. Table 1) – this accounts for small differences
between the atypicalities given here for axes previously studied by Jones and Williams-Thorpe (2001), and data in
Jones and Williams-Thorpe (op cit.) table 5; Fe values used are as oxides (total iron as Fe2O3).

Summary of the new evidence for Group XIII membership and its
relationship to the existing petrographic evidence
In Table 4, we summarise our new information on the twenty-eight objects listed as Group
XIII by Clough and Cummins (1988). We comment on the evidence for membership of
Group XIII – Preseli spotted dolerite – and also of Group XXIIIb – the broader dolerite IPC
grouping that includes unspotted Preseli dolerites.
Eight objects conform very well to Preseli petrology, and are confirmed by our studies as
Group XIII spotted dolerites. Me8, more marginal in chemistry, may also be retained in the

370
Non-destructive provenancing of bluestone axe-heads in Britain

Group. In addition, two objects, An10 and in particular Wi302, are candidates for Group
XXIIIb. E29 and Sx6 are less likely members of Group XXIIIb in terms of their magnetic
susceptibility and chemistry respectively, but we cannot discount the possibility. All the other
reported Group XIII objects that we examined differ significantly from Preseli, and may,
therefore, need re-assessment in terms of axe group classification.
Table 4 also reports the results for the additional five objects that we studied (that is, pieces
not listed as Group XIII by Clough & Cummins op. cit.). Wi389 is Preseli spotted dolerite,
and the balance of evidence may be in favour of a Preseli origin for SilF.
Two artefacts listed in Table 4 present particular problems in interpretation. M8, spotted and
with magnetic susceptibility firmly within Preseli, was allowed by Jones and Williams-Thorpe
(2001) as Preseli in origin though with rather marginal chemical composition. However, it is
excluded from Preseli using the more rigorous error assessment adopted for the present work.
Shorter PXRF counting times (and, therefore, larger uncertainties) of the early measurements
of which M8 formed a part may contribute to the apparent ambiguity of these data.
Sx178 is more difficult to account for. It also appears spotted, and has Preseli type magnetic

Method
susceptibility, but a chemical composition atypical of Preseli. The PXRF analysis is based on
ten measurements at the longer count times that we now adopt, on surfaces that appear
reasonably smooth and planar, and not heavily weathered. Unusually low X-ray backscatter
peaks (cf. Potts et al.. 1997a) observed for this piece are likely to reflect the relatively high Fe
concentration (nearly 14 per cent Fe2O3; Table 1). The data from this artefact could, therefore,
suggest an outcrop of spotted dolerite with a composition that is not represented among our
Preseli source material.
The petrographic evidence was reviewed from IPC record cards, and from our re-examination
of twenty existing Group XIII thin sections. Several of those implements which differ from
Preseli magnetically or chemically also have some differentiating features in thin section. But
these are mainly slight, such as some finer-grained texture than is typical of Preseli in Sussex 6.
And, since the Preseli dolerites encompass a range of textures and mineralogy (cf. Bevins et al.
1989), it is sometimes only in the light of additional new data that provenance can be confirmed.
Dumfries 8, on the other hand, (listed only ‘Group XIII?’ in Clough & Cummins 1988) is
rather more exceptional in petrography, being an altered (uralitic) doleritic rock.
It seems likely that the presence of an ophitic or sub-ophitic doleritic petrography may
have sometimes been taken as a determining characteristic in assigning implements to Group
XIII. The presence or absence of the diagnostic spots cannot always be assessed by thin
section examination alone, since the spots are sometimes so dispersed in the rock that, as H.
H. Thomas noted, ‘ … there is every possibility of their being unrepresented in a micro-section’
(Thomas 1923: 243).
Ixer (1994; cf. Ixer et al. in press) has made the case for use of opaque mineralogy studies
in lithic provenancing, and the potential of this method for Group XIII axes is enhanced by
preliminary studies that suggested it might be possible to differentiate Preseli outcrops on
the basis of their opaque mineral assemblages (Ixer in Thorpe et al. 1991; Ixer 1997). Following
up this approach is hampered at present by the severe limitations in using the existing axe
thin sections (designed for transmitted light petrography) for opaque mineralogy, and also
by the need for more detailed source sampling in order to take account of the sensitivity of
the method to minor lithological variations.

371
Olwen Williams-Thorpe, P.J. Potts & M.C. Jones

Table 4. Summary of new information on Group XIII membership


Is the object Could the object
Preseli spotted be Preseli non-
IPC/Other dolerite? spotted dolerite? Other
reference Object type Find site* i.e. Group XIII i.e. Group XXIIIb comments
Objects assigned to Group XIII in Clough and Cummins (1988)
An10 axe-hammer LLANFAETHLU NO possibly unspotted**
Ca25 axe-hammer PENLAN NO no unspotted
Car23 battle-axe TRELECH A’R BETTWS YES n/a spotted
Ch11 axe BARROW NO no unspotted
Dev1 mace High Peak, SIDMOUTH YES n/a spotted
Do41 small axe, MAIDEN CASTLE YES n/a spotted
reworked
Dmf8 axe Blackburn Bridge, NO no unspotted
WATCARRICK
E23 axe Near LAMARSH Hall NO no unspotted
E29 axe-hammer THAXTED NO possibly possible though
rather marginal
Group XXIIIb
(chemistry lowest
atyp. 0.11; mag
sus atyp. 0.18);
unspotted
Ha41 axe Bankes Heath, YES n/a spotted
BOURNEMOUTH
Ha47 axe WORTHY DOWN, not examined
Worthy
M8 small axe St Brides, NETHERWENT ? ? chemically at
edges of Preseli
distribution
(Fig. 3, Table 3);
spotted
Me8 axe-hammer Near ARTHOG, Group XIII? n/a chemistry
Llangelynnin, marginal Preseli;
Bwlch Gwyn Farm spotted
P13 mace Cilgwyn Mawr, NEVERN NO no unspotted
P40 axe-hammer LLANFYRNACH not examined
or battle-axe
Ra4 battle-axe CLAP YR ARIAN, YES n/a spotted
Llansantffraed-Cwmdeuddwr
Shah61 axe-hammer Chirbury, WOTHERTON NO no unspotted
S94 axe LAKENHEATH NO no unspotted
Su34 axe HERSHAM NO no distinction from
Preseli rests on Fe
Rb Ba; unspotted
Sx6 axe CRAWLEY NO possibly chemistry
marginal Preseli;
unspotted
Sx35 battle-axe TWINEHAM not examined
Sx178 shaft-hole adze NEWHAVEN ? ? chemistry is
or mace-head atypical of Preseli;
spotted
Wi10 miscellaneous WINDMILL HILL NO no unspotted
fragment
Wi108 miscellaneous STONEHENGE, YES n/a spotted
artefact fragment South Barrow, context unknown

372
Non-destructive provenancing of bluestone axe-heads in Britain

Table 4. Summary of new information on Group XIII membership (continued)


Is the object Could the object
Preseli spotted be Preseli non-
IPC/Other dolerite? spotted dolerite? Other
reference Object type Find site* i.e. Group XIII i.e. Group XXIIIb comments
Wi109 miscellaneous STONEHENGE, YES n/a spotted
artefact fragment Aubrey Hole 24, topsoil/turf
Wi118 axe STOCKTON Earthworks YES n/a spotted
Wi133 axe-hammer Ebbesbourne Wake, not examined
FIFIELD (Fyfield) Down
Wi302 battle-axe WILSFORD S. Barrow 54 NO yes Chemistry and
mag sus typical of
Preseli; unspotted
Other objects with petrographic or other similarities to Group XIII
Car30 large axe GARN WEN Farm, NO no unspotted
Llanglydwen, Cilmaenllwyd
Car31 small axe The Pill, LAUGHARNE NO no unspotted
Car34 axe-hammer MARCHGWYN, NO no unspotted

Method
Cilmaenllwyd
Wi389 probably AVEBURY AREA, not YES n/a spotted
unworked piece far from West Kennet long barrow
SilF*** unworked flake Very probably from Group XIII? n/a very low Y and
STONEHENGE Rb affects PXRF
(n=2) atypicalities;
spotted
Notes: An10, E23, M8, Me8 and P13 discussed by Jones and Williams-Thorpe (2001); * shortened name, as used in
text and tables, is in upper case; ** ‘unspotted’ or ‘spotted’ refers throughout to the metamorphic spots characteristic
of Preseli; *** SilF is part of a collection made by R. Faulkner in c. 1849 of material from various Wiltshire sites
including Stonehenge and Silbury Hill; investigation by R. Cleal (pers. comm.) has indicated that the piece is almost
certainly from Stonehenge, not, as was thought at the time of analysis and naming of the sample, from Silbury Hill.

Discussion and conclusions


Comments on techniques
The non-destructive PXRF technique has been successful in producing data of analytical
quality (accuracy and precision) appropriate for comparison with the published source
analytical database (which has been produced using well-established laboratory XRF
techniques on crushed rock samples). The magnetic susceptibility, a relatively recent addition
to our work on stone axes, has also provided some extremely useful information, especially in
helping to identify some of the most clearly anomalous (in terms of Group XIII membership)
implements. Interpreted using atypicality calculations, the new data provide indications that
alter some previous (petrographic) assignments.
In the case of Group XIII, the chemical and magnetic data have identified a larger variety
of petrological types and, therefore, outcrop sources, within the studied assemblage than was
previously apparent from the petrographic work alone. The petrographic descriptions are
very rarely in actual contradiction to our new work; rather, the features observable in some
thin sections are insufficiently diagnostic to exclude non-Preseli dolerites.

373
Olwen Williams-Thorpe, P.J. Potts & M.C. Jones

Comments on archaeological implications


Figure 7 illustrates the new information on the distribution of Preseli spotted and unspotted
dolerite implements in the UK.
The number of spotted dolerite implements confirmed in this work as originating from
Preseli is fewer than half of that previously reported as Group XIII. It is notable that many of
the implements that we would now exclude (from Group XIII) are those at the edges of the
geographic distribution of artefacts, namely those found in East Anglia, the Welsh borders
and Scotland. The revised distribution is much reduced, being limited to Wales and part of

Figure 7. Map showing the new information on the distribution of Preseli spotted and unspotted dolerite
implements (all the objects shown were previously assigned to Group XIII). A question mark indicates uncertainty
in our provenance conclusion (cf. Table 4).

374
Non-destructive provenancing of bluestone axe-heads in Britain

southern England. Future analysis of the many dolerite axes that have been examined in thin
section, but which proved to have insufficiently diagnostic mineralogical features for
assignment to a specific implement Group (cf. Clough & Cummins 1988, passim) may, of
course, add to the list of Preseli products. Such analysis is probably more likely to increase
numbers of the less distinctive unspotted Preseli products.
The evidence regarding spotted dolerite axes from Preseli in Ireland is contradictory. Three
axes identified by Cooney and Mandal as probably originating in south-west Wales (1995:
979) appear petrographically similar to Group XIII in transmitted light thin sections, yet
have differing opaque mineral assemblages (Cooney & Mandal op. cit.: 978). Chemically,
they are characterised by high Zr and Y concentrations (137 – 180 ppm, and 34 – 47 ppm
respectively; data from Mandal 1996: 192), and, so, are unlike the Preseli dolerite intrusions
discussed above. Overall, the evidence for Group XIII axes in Ireland is unconvincing at
present.
The confirmed Preseli dolerite artefacts include battle-axes and a mace, and several
unperforated axes (cf. Table 4). The dating evidence remains essentially unchanged from

Method
that previously reported in the literature for Group XIII; associations with round barrows
and with Beaker finds still apply (Radnorshire 4, and the unspotted Preseli dolerite battle-
axe Wiltshire 302). Unperforated axes are regarded as Neolithic (e.g. Bradley & Edmonds
1993; Pitts 1996), though secure associations of spotted dolerite implements with pre-Beaker
contexts are lacking.
Preseli dolerite was used for three highly polished and carefully finished battle-axes
(Carmarthenshire 23, Radnor 4, and Wiltshire 302), all very similar in style, and suggestive
of a large amount of care and time expended in their manufacture. The first two listed were
found in Wales (and have very obvious spots), and the third in England (and there may be
another example among the implements that we were not able to examine). These battle-
axes are, however, unusual within the Preseli implements. Notwithstanding the surface
weathering that can sometimes obscure original polish, most confirmed Group XIII objects
are unexceptional in style and appearance (cf. Lynch 1975, 124). It is also notable that the
style of the three battle-axes is not exclusive to Preseli rock; similarly shaped implements
were made from other dolerites, for example Whin Sill (IPC Group XVIII; cf. Williams-
Thorpe et al. 2003).
Although artefacts of spotted dolerite appear to be less widespread and less numerous than
was previously thought, some have been found beyond the generally accepted limits of
glaciation in Britain (cf. Figure 7), implying human dispersal. The available data present a
picture of a small and morphologically disparate group of implements distributed (perhaps
traded) within a limited area mainly within Wessex, possibly both before and during the first
use of bluestones at Stonehenge (which was c. 2600-2500 BC: Cleal et al. 1995). Two of the
confirmed Group XIII pieces are reported as ‘miscellaneous artefacts’ (neither appears to be
an axe) and were found at Stonehenge; they (and also, perhaps, SilF) may be part of the large
assemblage of bluestone fragments found in many contexts at and near Stonehenge, rather
than part of bluestone axe trade. The identification of an unworked fragment of spotted
dolerite bluestone from the vicinity of Avebury/West Kennet long barrow (Wiltshire 389)
adds to the occurrences of bluestone apparently associated with a variety of prehistoric
monuments in Wiltshire (cf. Thorpe et al. 1991, 108-113).

375
Olwen Williams-Thorpe, P.J. Potts & M.C. Jones

Studies of lithic resource procurement strategies in British prehistory are bedevilled by the
difficulty of distinguishing between primary sources (material at outcrops), and secondary
sources (rock removed from outcrops by, for example, glaciation, and perhaps available locally
to the users). The case of spotted dolerite implements is further complicated by the possibility
that they could have been made from the same assemblage of rocks or erratics that formed
the Stonehenge monoliths, whether these were obtained in Preseli or were available nearer to
Salisbury Plain. The working and polishing of the implements has removed any original
surface that may have held clues to the type of source rock acquisition.
Spotted dolerite axes have been regarded as illustrating a special regard for Preseli rock,
helping to justify the concept of human selection and transport of spotted dolerite for
Stonehenge monoliths. Our new information on Group XIII axes and other implements has
reduced the number that can be reliably assigned to this group, and the extent of their
geographical distribution. Notwithstanding evidence for other prehistoric, Neolithic, activity
in the Preseli area (Darvill & Wainwright 2001), it appears that spotted dolerite was used for
a very small proportion of British stone implements, mainly of unremarkable styles.

Overall conclusion
Non-destructive analysis has shown that spotted dolerite was much less widely used for
implement manufacture in Britain than has previously been thought, thus weakening the
argument that stone axe production and trade constituted a prompting influence for
Stonehenge monolith selection and transport.

Appendix: atypicality calculations


Let d be the number of variables measured on source material and artefacts; d = 1 for magnetic
susceptibility and d = 4, 6 or 7 for PXRF measurements. Suppose there are n separate
measurements on the source material and k such measurements on an artefact. Let x- and S be
the vector of sample means and the sample variance-covariance matrix, respectively, of the
source measurements (the ith diagonal element of S is the sample variance of the ith variable
and the (i, j)th element of S, i ≠ j is the sample covariance between the ith and jth variables). Let
y- be the vector of sample means based on artefact measurements. Then, following Jones and
Williams-Thorpe’s (2001) modification of standard atypicality-type calculations (e.g. Krzanowski
1988), the (modified) Mahalanobis distance of the artefact from the source material is
n(n − d )
m0 = ( y − x )T (S + P / k )−1 ( y − x )
(
d n2 − 1)
Here, superscripts T and –1 stand for vector transpose and matrix inverse, respectively. We
take P to be a diagonal matrix, the non-zero elements of which are the sample variances of
the artefact measurements (in so doing, we depart slightly from Jones and Williams-Thorpe
(2001), where we used analytical errors provided by the measuring instrument). Finally, to
calibrate m0, the atypicality value is defined to be the probability of
P( M0 ≥ m0 )
getting as or more extreme a value as m0 assuming multivariate normality. The approximate
distribution of M0 which we use is the F distribution on d and n – d degrees of freedom.
376
Non-destructive provenancing of bluestone axe-heads in Britain

Acknowledgements
This work forms part of an on-going project at the Open University, in co-operation with the IPC (now Implement
Petrology Group or IPG) and others, aimed at refining the distributions of a number of stone axe groups, using
non-destructive geochemical and geophysical methods of characterisation. The work on Group XIII axes which
is reported in this paper was wholly funded by a grant from the Leverhulme Trust (F/00269/C) which we gratefully
acknowledge. The following museums and individuals are thanked for their help in permitting and facilitating
analysis of archaeological objects, and for loans of thin sections: Mike Stokes, Shrewsbury Museums Service;
Moya O’Mullane, Grosvenor Museum, Chester; Ros Cleal, Alexander Keiller Museum, Avebury; Martin Wright,
Salisbury Museum; Paul Robinson, Devizes Museum; John Allan, Exeter Museum; Peter Woodward, Dorchester
Museum; Peter Berridge, Colchester Museum; Elizabeth Walker, Richard Bevins, Jana Horak and Heather Jackson,
National Museum and Galleries of Wales Cardiff; Siobhan Ratford, Dumfries Museum; David Clark, Tullie
House Museum, Carlisle; John Faithfull, Hunterian Museum, Glasgow; Emma O’Connor, Lewes Museum;
Michael Rowe, Elmbridge Museum; Gillian Varndell, British Museum; Christopher Chippindale and Sharon
Webb, Cambridge University Museum; Michael Freeman, Ceredigion Museum, Aberystwyth; Gavin Evans,
Carmarthen Museum; Rachel Scholl, Radnorshire Museum; David Smith, Natural History Museum; Stephen
Minnitt, Somerset County Museums Service, Taunton; Rob Ixer, University of Birmingham. Special thanks are
due to Bill Pratt, Nur Bahal and Jovanna Scorsone at the Royal Ontario Museum for their help with the loan of

Method
Sx 178. For permissions and advice regarding work in Preseli, we are grateful to Sid Howells and the Countryside
Council for Wales; and J. E. George of George, Davies and Evans (Solicitors), and Lady Marcher and the Barony
of Kemes. Ros Cleal kindly allowed us to report analyses of SilF and also provided very helpful information on
the likely find location of this piece. We thank Heather Pegg and Will Adams of Radnorshire Museum for
allowing us to use photographs of a battle-axe in their collections (Figure 2). Ian Rigby helped in the examination
of petrological thin sections, and the diagrams were produced by John Taylor.

References CLOUGH, T.H.MCK. & W.A. CUMMINS (ed.). 1988.


Stone Axe Studies Volume 2. Council for British
ATKINSON, R.J.C. 1979. Stonehenge: archaeology and Archaeology Research Report No. 67. London:
interpretation. Harmondsworth: Penguin (3rd Council for British Archaeology.
edition).
COONEY, G. & S. MANDAL. 1995. Getting to the core of
BERNARDINI, F., P.J. POTTS, M.C. JONES, P.C. WEBB & the problem: petrological results from the Irish
O. WILLIAMS-THORPE. In preparation. Effect of Stone Axe Project, Antiquity 69: 969-980.
weathering on in situ measurements of geological
samples by portable X-ray fluorescence analysis: DARVILL, T. & G. WAINWRIGHT. SPACES – exploring
dolerite and rhyolite outcrops in the Preseli Neolithic landscapes in the Strumble-Preseli area of
Mountains, South Wales. southwest Wales, Antiquity 76: 623-624.
BEVINS, R. E., G. J. LEES, & R. A. ROACH. 1989. DAVIS, R.V. 1985. Implement petrology: the state of the
Ordovician intrusions of the Strumble Head- art – some problems and possibilities, in P. Phillips
Mynydd Preseli region, Wales: lateral extensions of (ed.), The archaeologist and the laboratory. Council
the Fishguard Volcanic Complex, Journal of the for British Archaeology Research report No. 58.
Geological Society of London 146: 113-123. London: Council for British Archaeology.
BRADLEY, R. & M. EDMONDS. 1993. Interpreting the axe EVENS, E.D., L.V. GRINSELL, S. PIGGOTT & F.S. WALLIS.
trade. Cambridge: Cambridge University Press. 1962. Fourth report of the Sub-Committee of the
South-Western Group of Museums and Art
CAMERON, D.G. 1988. A geochemical drainage survey Galleries on the Petrological Identification of Stone
of the Preseli Hills, south-west Dyfed, Wales, Axes, Proceedings of the Prehistoric Society 28: 209-
Transactions of the Institute of Mining and 266.
Metallurgy, Section B: Applied Earth Science 97:
B42. FELL, C.I. & R.V. DAVIS. 1988. The petrological
identification of stone implements from Cumbria.
CLEAL, M.J., K.E. WALKER, & R. MONTAGUE. 1995. In T.H.MCK. Clough & W.A. Cummins 1988:
Stonehenge in its landscape. English Heritage 71-77.
Archaeological Report 10. London: English
Heritage. GREEN, C. P. 1997. Stonehenge: geology and prehistory,
Proceedings of the Geologists’ Association 108: 1-10.

377
Olwen Williams-Thorpe, P.J. Potts & M.C. Jones

GRIMES, W.F. 1935. Notes on recent finds of perforated POTTS, P.J., P.C. WEBB & O. WILLIAMS-THORPE.
axe-hammers in Wales, Archaeologia Cambrensis 90: 1997(a). Investigation of a correction procedure
267-278. for surface irregularity effects based on scatter peak
intensities in the field analysis of geological and
IXER, R.A. 1991. Appendix 1. Opaque petrography of
archaeological rock samples by portable X-ray
Stonehenge bluestone and Carnmenyn dolerites, in
fluorescence spectrometry, Journal of Analytical
R. S. Thorpe et al. 1991: 150-152.
Atomic Spectroscopy 12: 769-776.
–1994. Does ore petrography have a practical role in
–1997(b). The bulk analysis of silicate rocks by
the finger-printing of rocks? In N. Ashton & A.
portable X-ray fluorescence: effects of sample
David (ed.), Stories in Stone: 10-23. Lithics Studies
mineralogy in relation to the size of the excited
Society Occasional papers 4, London.
volume, Geostandards Newsletter 21: 29-41.
–1997. Detailed provenancing of the Stonehenge
ROLLINSON, H. 1993. Using geochemical data. Harlow:
dolerites using reflected light petrography, in A.
Longman Scientific and Technical.
Sinclair, E. Slater & J. Gowlett (ed.), Archaeological
Sciences 1995: 11-17. Oxford: Oxbow Monograph SHOTTON F.W. 1972. The large stone axes ascribed to
64. North-West Pembrokeshire, in F. Lynch & C.
Burgess (ed.), Prehistoric man in Wales and the West:
IXER, R.A., O. WILLIAMS-THORPE, R.E. BEVINS & A.D
85-91. Bath: Adamas and Dart.
CHAMBERS. In press. A comparison between ‘total
petrography’ and geochemistry using a portable STONE, J.F.S. & F.S. WALLIS. 1951. Third Report of the
XRF as provenancing tools for some Midland axes, Sub-Committee of the South-Western Group of
in E.A. Walker, F. Wenban-Smith, & F. Healy Museums and Art Galleries on the Petrological
(ed.), Proceedings of the ‘Lithic Studies in the Year Identification of Stone Axes, Proceedings of the
2000’ Conference. Oxford: Lithic Studies Society, Prehistoric Society 17: 95-158.
Oxbow Books, Occasional Paper No. 8. (Pre-print
THOMAS, H.H. 1923. The sources of the stones of
available at: http://www.rosiehardman.com/
Stonehenge, The Antiquaries Journal 3: 239-260.
goodprovenance/onlinepubs.html).
THORPE, R. S., O. WILLIAMS-THORPE, D. G. JENKINS &
JONES, M. C. & O. WILLIAMS-THORPE. 2001. An
J. S. WATSON, with contributions by R.A. IXER &
illustration of the use of an atypicality index in
R.G. THOMAS. 1991. The geological sources and
provenancing British stone axes, Archaeometry 43:
transport of the bluestones of Stonehenge,
1-18.
Wiltshire, UK, Proceedings of the Prehistoric Society
LYNCH, F. 1975. The impact of landscape on prehistoric 57: 103-157.
man, in J.G. Evans, S. Limbrey & H. Cleere (ed.),
WILLIAMS-THORPE, O. In press. The application of
The effect of man on the landscape: the Highland
portable X-ray fluorescence analysis to
Zone: 124 -127. London: Council for British
archaeological lithic studies, in P.J. Potts & M.
Archaeology Research Report 11.
West (ed.), Portable X-ray fluorescence spectrometry:
MANDAL, S. 1996. The Petrology of the Irish Stone Axe, capabilities for in situ analysis. London: Royal
Unpublished Ph.D. Thesis, University of Dublin. Society of Chemistry.
–1997. Striking the balance: the roles of petrography WILLIAMS-THORPE, O. & R.S. THORPE. 1993. Magnetic
and geochemistry in stone axe studies in Ireland, susceptibility used in non-destructive provenancing
Archaeometry 39: 289-308. of Roman granite columns. Archaeometry 35: 185-
195.
MARKHAM, M. 2000. Provenance studies of British
greenstone implements using non-destructive WILLIAMS-THORPE, O., P.J. POTTS & P.C. WEBB. 1999.
analytical methods. Unpublished Ph.D. Thesis, Field-portable non-destructive analysis of lithic
Open University, UK. archaeological samples by X-ray fluorescence
instrumentation using a mercury iodide detector:
PITTS, M. 1996. The stone axe in Neolithic Britain,
comparison with wavelength-dispersive XRF and a
Proceedings of the Prehistoric Society 62: 311-371.
case study in British stone axe provenancing,
POTTS, P.J., P.C. WEBB, O.WILLIAMS-THORPE & R. Journal of Archaeological Science 26: 215-237.
KILWORTH. 1995. Analysis of silicate rocks using
field-portable X-ray fluorescence instrumentation
incorporating a mercury (II) iodide detector: a
preliminary assessment of analytical performance,
Analyst 120: 1273-1278.

378
Non-destructive provenancing of bluestone axe-heads in Britain

WILLIAMS-THORPE, O., M.C. JONES, P.J. POTTS & WILLIAMS-THORPE, O., P.C. WEBB, M.C. JONES &
I. J. RIGBY. (based in part on field observations by I.J. RIGBY. 2000b. Magnetic susceptibility thickness
D. Peacock, J. Phillips & N. Bradford). 2001. corrections for small artefacts and comments on
Geology, mineralogy and characterisation studies the effects of ‘background’ materials, Archaeometry
of Imperial Porphyry, in V.A. Maxfield & D.P.S. 42: 101-108.
Peacock (ed.), The Roman Imperial Quarries, Survey
WILLIAMS-THORPE, O., M.C. JONES & P.C. WEBB.
and Excavation at Mons Porphyrites 1994-1998,
2003. Non-destructive geochemical and magnetic
Volume 1: Topography and Quarries. Cairo: Egypt
characterisation of Group XVIII dolerite stone axes
Exploration Society: 305-318.
and shaft-hole implements from England, Journal
WILLIAMS-THORPE, O., P.C. WEBB & R.S. THORPE. of Archaeological Science 30: 1237-1267.
2000a. Non-destructive portable gamma ray
spectrometry used in provenancing Roman
granitioid columns from Leptis Magna, North
Africa. Archaeometry 42: 77–99.

Method

379

You might also like