You are on page 1of 12

Received: 1 April 2016 Revised: 30 September 2016 Accepted: 27 November 2016

DOI: 10.1002/gea.21618

RESEARCH ARTICLE

Application of high-resolution Mobile Metal Ion (MMI) soil


geochemistry to archaeological investigations: An example
from a Roman metal working site, Somerset, United Kingdom

Graham C. Sylvester1 Alan W. Mann2 Andrew W. Rate3 Clare A. Wilson4

1 School of Geography and Environmental

Sciences, M087, The University of Western Abstract


Australia, Crawley, Western Australia, Australia An innovative application of Mobile Metal Ion (MMI) partial extraction soil geochemistry is used to
2 Independent Geochemical Consultant, South
identify below-surface archaeological features, using a previously incompletely surveyed Roman
Fremantle, Western Australia, Australia metal-working site at St. Algars Farm, Somerset, as a case study. Soil samples were taken and ana-
3 School of Geography and Environmental
lyzed for 53 elements by the MMI geochemical method. Lead, Tl, Ba, and Zn were found in very
Sciences, M087, The University of Western
Australia, Crawley, Western Australia, Australia
high concentrations and the sensitivity of the technique also enabled Ag, Au, and Sn to be mea-
4 Biological and Environmental Sciences, sured in anomalous concentrations. Elemental maps accurately outlined known metal working
University of Stirling, Stirling, United Kingdom areas. Principal component analysis and bivariate correlations identified two suites of associated
Correspondence elements: Pb, Ba, Tl, Ag, Au, Cu, Sb, the base and noble metal group (BNM), and Fe, Ti, Nb, Mn, Co,
Graham C. Sylvester, School of Geography and Cu, P, Li, Rb, Sc, Cs, K, Ga, P, Zr, Th, and Sn, the pegmatite (PEG) group. These were used to form
Environmental Science, M087, The University of
indices that delineate the metal working area and areas possibly related to the processing of peg-
Western Australia, 35 Stirling Highway, Crawley,
WA 6009, Australia. matite containing Sn. The high-sensitivity MMI data were compared with strong acid digest results
Email: sylvestergc@gmail.com from a limited number of the MMI samples; the MMI data showed better geochemical contrast
Scientific editing by Robert Speakman than the strong acid results. Multielement statistical similarity comparisons with off-site samples
suggest likely sources for the Pb and Sn used at the St. Algars site. The increased sensitivity of
MMI soil analysis combined with the multielement capacity allows a more detailed archaeological
interpretation.

KEYWORDS
archaeological prospection, magnetic gradiometry, metal extraction, MMI, partial extraction,
Roman, soil geochemistry

1 INTRODUCTION prospection (Aston, Martin, & Jackson, 1998; Bintliff, Davies, Gaffney,
Snodgrass, & Waters, 1992; Entwistle, Abrahams, & Dodgshon, 2000;
Geochemistry has had a small, but growing place in archaeological Schlezinger & Howes, 2000), interpretation of space use (Cook et al.,
investigations for almost a century. Soil chemistry was first applied 2010; Cook, Clarke, & Fulford, 2005; Cook, Kovacevich, Beach, &
to archaeological questions when it was discovered that phosphate Bishop, 2006), locating middens (Beck, 2007), inhumations and differ-
enrichment in Swedish soils was an indicator of prehistoric human entiated burial sites (Bethell & Carver, 1987; Sampiettro & Vattuone,
occupation and could be connected with prehistoric sites and deserted 2005), and identifying metal accumulation derived locally from ancient
medieval villages (Arrhenius, 1931). The advent of inductively coupled mining or mineral processing operations (Dunster & Dungworth,
plasma mass spectrometry (ICP-MS) presented the opportunity to 2012; Dunster, Dungworth, & Lowerre, 2012; Grattan, Gilbertson,
analyze quickly and at relatively low cost for a very much wider range & Kent, 2013; Maskall, Whitehead, Gee, & Thornton, 1996; Maskall,
of elements. This resulted in the analysis of soils associated with Whitehead, & Thornton, 1995; Mighall, Grattan, Lees, Timberlake, &
archaeological sites for an increased number of elements potentially Forsyth, 2002).
indicative of human activity (Middleton, 2004) Today, more than The formation of anthropogenic geochemical anomalies in soils is
20 elements have been found to be indicators of anthropogenic complex. Oonk, Slomp, and Huisman (2009) observed that retention
activity at specific sites over a wide range of archaeological contexts. and sequestration of elements in soils is rarely governed by a single
Applications include distinguishing past domestic and agricultural process. Adsorption, occlusion, ion exchange, isomorphic substitution,
activities (Davidson, Dercon, Stewart, & Watson, 2006), archaeological chelation, and precipitation reactions together with factors such as

Geoarchaeology. 2017;112. wileyonlinelibrary.com/journal/gea 


c 2017 Wiley Periodicals, Inc. 1
2 SYLVESTER ET AL.

waste and soil composition, soil pH, redox conditions, grain, and pore simple. The protocol is given in (Mann, 2010). When combined with
size can be responsible for the preservation of inorganic signals in ICP-MS (capable of returning low concentrations for over 50 ele-
archaeological soils. The retention and sequestration of trace elements ments), the MMI technique provides a potentially very powerful tool
in soils generally depend on their oxidation state, soil pH, the specific for archaeological investigation.
surface area, cation exchange capacity, and the presence of clay min- The recently completed European Agricultural Soils (GEMAS)
erals, Fe oxides, carbonates, phosphates, sulphides, and organic matter project (Reimann, Birke, Demetriades, Filzmoser, & OConnor,
(McBride, 1989; Oonk et al., 2009). 2014a,b) analyzed agricultural soils from a widely spaced sampling
The detection of an anthropogenic soil geochemical signature grid by XRF and ICP-MS after both SAD and MMI extraction. The
(anomaly) is based upon the contrast between the anthropogenic sig- MMI results identified anthropogenic contributions to soil chemical
nal (commonly weak) and the background. The anthropogenic con- composition (Sadeghi et al., 2015), some of which are potentially
tribution to the measured value may be relatively small when total of archaeological interest (Mann, Reimann, & de Caritat, 2014). To
or strong-acid digests are utilized because a substantial contribution evaluate the use of MMI soil geochemistry in the discovery, delin-
to the overall analytical signal will come from dissolved lithological eation, and definition of anthropogenic soil geochemical anomalies
material. This results in a lower peak/background ratio (geochemical associated with archaeological sites, this study was carried out on
contrast) than would be obtained by use of a weak acid or nonacid the well-documented Roman site at St. Algars Farm, Somerset, UK.
digest (Mann, 2010; Stanley & Noble, 2008). Even with the lowering Limited SAD investigations were included to determine whether the
of detection limits for elements, the problem of differentiating anoma- elemental species targeted by MMI provide more sensitive detection
lies from background remains, and it is only by reducing the relative of archaeological anomalies than total analyses.
background levels that the peak/background ratio can be effectively
enhanced. An acidic partial digest still dissolves part of the soil matrix,
whereas a nonacid partial extraction attempts to detach analytes from
2 MATERIALS AND METHODS
the soil matrix, with minimal dissolution of the matrix, thereby reduc-
ing background interference (Mann, 2010). Mann (2010) also found
2.1 Site description and sampling
that partial digestions and extractions appear to have the ability to
discriminate in favor of the active ionic (source-related) signature An MMI geochemical soil sampling program was undertaken on
of the sought after material and the elements indicative of its pres- and around the area of the Scheduled Monument centered at
ence, relative to their geochemical background, thereby improving 5110 31.2 N 218 39.0 W on St. Algars Farm, Selwood in eastern
both spatial and amplitude resolution of anomaly signals. Stanley and Somerset. St. Algars Farm is located adjacent to the Frome Road
Noble (2008) made similar findings in their evaluation of Navan base (B3092) about 2 km south of the village of West Woodland, some
metal deposits in Ireland. Here, it was found that weak nonacid partial 6 km south of the town of Frome and about 25 km south of the city
extractions (using solutions similar to those used in Mobile Metal Ion of Bath (Fig. 1). The farm lies on undulating land in the valley of the
[MMI]) were far more effective in the identification of soil geochemi- River Frome and is currently under permanent pasture and used for
cal anomalies derived from buried base metal mineralization than were cattle grazing. It has been subject to some plowing in the past few
total or strong-acid digests. The reason was that the weak extrac- decades. The site has a maximum elevation of approximately 108 m
tant dissolved only the material derived from the mineralization by above mean sea level. It slopes gently to the east and flattens toward
upward transport in solution and deposition on soil clays, iron oxides, an irrigation channel that extends from the nearby Frome River and the
etc. It dissolved negligible concentrations of elements from the soil farm pond, where the elevation is about 98 m above mean sea level.
minerals, which strong-acid digestion (SAD) does, and which increase The soil at the site is of heavy clay and has been mapped as Soilscape
background (and thereby its detectability) by lowering the peak/noise 18 of the Soilscape Soil Types map of England. These soils are
ratio. classified as slowly permeable, seasonally wet, loamy to clayey,
The MMI technique is based on a neutral alkaline solution con- commonly drainage-impeded, moderately fertile, low carbon grass-
taining both organic and inorganic ligands (Mann, 2010). The ligands land soils that drain into the local stream network (Cranfield
provide complexing ability, while the lack of an aggressive acidic (or University, 2015).
alkaline) component ensures that during extraction the soil matrix Archaeological investigation of the site (excavated trenches based
is not dissolved The analytical signal is most likely derived pre- upon geophysical results) recovered evidence of three periods of
dominantly from adsorbed and loosely attached ions on the exte- Romano-British occupation (Lambdin, 2011; Lambdin & Holley, 2011b,
rior of grains and grain boundaries. The amount of each element 2012a, 2012b). These are (1) an early villa with internal room divisions
extracted is much lower than obtained by total X-ray fluorescence and set within a square ditched area of first to second century CE age,
spectrometry (XRF) or by acid extraction (e.g., aqua regia), but the (2) a winged corridor villa of likely second to third century age, and
improved signal-to-noise ratio due to reduction in background has (3) a fourth century industrial site. Excavations have recovered
resulted in the technique being widely used in mineral exploration substantial amounts of common building materials including stone
(Mann et al., 1998); there is also an increased efficacy for resolu- rubble, roof and floor tiles, flue tiles, tesserae, fragments of painted
tion of anthropogenic anomalies (Mann et al., 2015) by discriminating wall plaster, and a cobbled surface that has been interpreted as
against matrix material. The sampling requirements for MMI are very a yard (Lambdin & Holley, 2011b, 2012a,b). Recovered artifacts
SYLVESTER ET AL. 3

F I G U R E 1 Map showing the location of St. Algars Farm, reference points and other places mentioned in the text. [Color figure can be viewed at
wileyonlinelibrary.com]

F I G U R E 2 St. Algars Farm magnetic gradiometry with interpreted archaeology (magnetic gradiometry after Lambdin, 2011). The Scheduled
Monument area is enclosed by the square box. [Color figure can be viewed at wileyonlinelibrary.com]

indicative of human occupation of the villa include substantial the site has uncovered evidence of considerable industrial activity, the
amounts of oyster shell, animal bone, glass gaming counters, local timing of commencement and end dates and whether different
and imported pottery, coins, and a copper brooch. A considerable activities were contiguous, sequential, or separated in time is
amount of glass and glass waste (pulled threads and trails), mis- not well defined. Other than a calibrated 14 C date of CE 131
shapen molten waste glass drops, and small broken chunks of glass 155 on cremation remains from the mausoleum, there is little
as well as crucible fragments indicative of glass working (probably in the way of secure dating evidence. Features of archaeologi-
fourth century) were recovered from the vicinity of the villa. No cal interest are shown on the magnetic gradiometry imagery in
conclusive evidence of a furnace on the site was derived from the Figure 2.
excavations. However, the discovery of cupellation products and The St. Algars soils sampled were gray in color, appeared uniform
waste, including lead and litharge are strongly indicative of Roman in grain size and texture, and showed little visual variation. The site
cupellation to produce silver from lead ore. Dipole magnetic anoma- had been plowed so there was little or no evidence of a humic mate-
lies, possibly indicative of hearth sites, are present in the geophysical rial in the near surface layer. Below the grass roots most of the soils
survey results and strong, coincident lead anomalies obtained from were visually unstructured and consist of a malleable, sticky clay-rich
portable XRF surface readings may be indicative of cupellation sites gray material. The area is underlain by Quaternary silt, sand, and gravel
(Dungworth, Comeau, & Lowerre, 2013). While excavation work on deposited from rivers as channel fill, terrace deposits, and flood plain
4 SYLVESTER ET AL.

alluvium. It is from the Quaternary flood plain alluvium that the local 2.3 Numerical and statistical analyses
soil at St. Algars Farm appears to have been developed. The basement
Basic statistical analyses on the edited dataset were performed using
geology consists of shallow marine calcareous sediments and lime-
Microsoft Excel.
R The distributions of each variable were checked for
stone of Jurassic age (British Geological Survey, 2015).
normality using the ShapiroWilk test implemented in R (R Core Team,
Sampling was undertaken using a predetermined 40 m 40 m grid.
2015); skewed variables were log10 -transformed prior to correlation
Sampling sites were selected to test the archaeology inferred from
and analyses. Numerous lead (Pb) values were above the upper detec-
geophysical survey and revealed by excavation. A total of 63 samples
tion limit (UDL; 20,000 ppb); however, based on the strong linear rela-
were collected over an area of 240 m 320 m. The predetermined
tionship between Pb and thallium (Tl) and because Tl commonly dis-
sample sites were located (Fig. 2) using a Garmin GPS 72 GPS unit
plays geochemical behavior similar to Pb (geochemical coherency), Pb
(accuracy to within 5 m). The actual sample sites were relocated with
values 20,000 ppb were replaced with calculated values from the
much greater accuracy (to a few cm) using a real time corrected Leica
regression of valid Pb concentrations against Tl (r2 = 0.75; P = 2 108 ;
GPS900 GPS unit. At each sampling location, the turf was lifted from a
log10 Pb = 4.75 + 1.354 log10 Tl). The extrapolated Pb values (Pbcalc )
20 20 cm2 area (the width of the spade used) and a pit was then dug
were used to overcome the distortion caused by the large number of
to a nominal depth of 15 cm. Duplicate soil samples (approximately 100
samples containing Pb at levels > UDL.
g) were then taken from the pit bottom, using a plastic spoon to prevent
Principal components analysis (PCA) was performed in R using cor-
sample contamination with metals. The sampling was thus designed to
relation matrix of scaled, centered log ratio transformed concentra-
capture soil from the 10- to 25-cm depth of the developed protocol and
tions, with the rare earth elements (REE; in this dataset, La, Ce, Pr, Nd,
was noted to be coincident with the plant root zone in accordance with
Sm, Eu, Gd, Tb, Dy, Er, and Yb) combined additively into a single vari-
previous observations (Mann, 2010). After sampling, the excavated soil
able because of their geochemical coherency and similarity in distri-
was returned to the hole and the turf sod replaced.
bution. The centered log ratio transformation removes any dependen-
cies between variables due to compositional closure (Reimann et al.,
2008) and also corrects skewed distributions. Elements containing too
2.2 Chemical analysis many below-detection concentrations (As, Cr, Sn) were not included
in the analysis. Contour plots and background maps were generated
The samples were sent to SGS laboratories in Perth, Western Australia
using Sigmaplot. Interpolation of the sampling grid to calculate con-
and were treated using the MMI-ME MMI analysis method in which 50
tours was achieved using anisotropic kriging based on linear-to-sill
g of sample is subject to leaching in a known volume of mixed lixiviants
semivariogram models.
solution (the solutions composition is commercial in-confidence but
Distribution maps of elements in surface soils were generated from
some useful background information is given above and in Mann et al.,
point data using simple kriging with exponential variogram models,
2012). One advantage of a relatively large sample charge and a par-
implemented using the R package geoR (Ribeiro & Diggle, 2015).
tial extraction is reduction in the nugget effect when dealing with
Variables to be predicted were log-transformed where necessary to
elements at low concentrations (Leduc & Itard, 2003). The resultant
remove skewness, and kriged predictions were made on a grid with
leachate is allowed to settle and is sampled after 24 hours. A subsam-
5 m spacings across the area sampled. Variogram models were fitted by
ple of the clear leachate is withdrawn and presented to an ICP-MS for
Cressies weighted least squares after fitting a first-order polynomial
elemental analysis. The ICP-MS instrumentation was a Perkin Elmer
trend surface, and with the maximum distance restricted to 40% of
Elan 9000 DRC II, with an argon plasma and a reaction cell with ammo-
the greatest intersample distance, as recommended by Reimann et al.
nia gas. The solution was analyzed for the following 53 elements: Ag,
(2008).
Al, As, Au, Ba, Bi, Ca, Cd, Ce, Co, Cr, Cs, Cu, Dy, Er, Eu, Fe, Ga, Gd,
Hg, K, La, Li, Mg, Mn, Mo, Nb, Nd, Ni, P, Pb, Pd, Pr, Pt, Rb, Sb, Sc, Se,
Sm, Sn, Sr, Ta, Tb, Te, Th, Ti, Tl, U, W, Y, Yb, Zn, and Zr. Quality control
and assessment (Caritat & Cooper, 2011) was based on several mea-
3 RESULTS
sures: (1) analysis in a single batch, (2) insertion of blind field duplicates,
(3) insertion of blind laboratory replicates, (4) insertion of laboratory
3.1 Elemental descriptive statistics
standards in the analytical stream at regular intervals, and (5) analysis
of blank MMI-extraction
R solution. Elemental values < lower detec- To facilitate investigation of the relationships between the elements
tion limit (LDL) were replaced in the final database with values set to of potential anthropogenic significance, basic statistical data were col-
0.5 LDL (Reimann, Filzmoser, Garrett, & Dutter, 2008). The elements lected and are shown in Table I. The local background at St. Algars
Bi, Hg, In, Mo, Pd, Pt, Ta, Te, and W were present in concentrations Farm is defined as being the mean of the first quartile MMI values for
< LDL in all samples and were removed from the database. Four ele- each element. Geochemical contrast for each element is defined as the
ments As, Cr, Nb, and Sn with >40% of values < LDL that would nor- ratio of the mean of the fourth quartile (4QM) to the mean of the first
mally be removed from the database (Reimann, Filzmoser, & Garrett, quartile (1QM).
2002) were retained as they provided valuable information relevant to Elements of interest include those which at St. Algars Farm show
the goals of the study. The element deletions resulted in a final 42 ele- strong (>10) and moderately strong (510) 4QM/1QM geochem-
ment suite for this study. ical contrast. Table I shows that Ag (19.0), Pbcalc (15.8), Ti (13.2), Au
SYLVESTER ET AL. 5

TA B L E I Statistical summary for St. Algars Farm MMI samples; elemental values are in ppb

Element Max. Median Max/1QM Mean SD 4QMean 1QMean 4QM/1QM


Ag 398 10 83.8 29.9 57.5 90.0 4.8 18.9
As 20 5 2.0 16.7 5.2 10.0 10.0 1.0
Au 2.2 0.2 44.0 0.4 0.4 0.6 0.1 12.4
Ba 3470 920 5.0 1090 506 1781 693 2.6
Ca 1,190,000 725,000 2.0 732,060 126 894,330 590,938 1.5
Cd 65 38 3.0 38.4 13.0 55 22 2.5
Ce 907 315 5.9 351 188 595 153 3.9
Co 439 38 27.3 66.6 78.1 164 16.1 10.2
Cu 2770 690 6.7 901 557 1584 412.5 3.8
Fe 202,000 39,000 8.6 46,490 31 82,690 23,560 3.5
Nb 6.4 0.7 25.6 1.4 1.3 2.6 0.3 10.2
Ni 2420 1110 3.5 1150 423 1773 700 2.5
P 1.3 0.6 2.8 0.7 0.2 1.0 0.5 2.1
Pb 20,000 20,000 3.4 15,878 6229 20,000 5829 3.4
Pbcalc 117,157 43,622 19.5 47,515 35,765 95,112 6013 15.8
Sb 12 2 12.8 3.3 2.5 7.9 0.9 8.4
Sn 17 0.5 34.0 1.1 2.2 2.9 0.5 5.8
Ti 1100 16 43.2 127 196 336 25.4 13.2
Tl 9.4 3.5 8.3 4.1 2.6 7.6 1.1 6.8
Zn 2910 590 14.6 676 484 1305 200 6.5

(12.4), Nb (10.2), and Co (10.2) show strong enrichment, while Sb (8.4), information. Strong relationships (Pearsons correlation coefficients >
Tl (6.8), Zn (6.5), and Sn (5.8) are moderately enriched. The enrichment 0.8) exist between Pb and Tl, Ti and P, Fe and Zr, Zr and La and between
of Pbcalc , Ag, and Au is indicative of the Roman lead smelting operations all REEs. Moderately strong (Pearsons correlation coefficients >0.7 <
at this site. 0.8) correlations were found between Pb and Sb, Tl and Sb, Tl and Ag,
Ag and Au, Ag and Cu, Nb and Fe, Ti and Nb, Fe and P, Mn and Co. These
correlations and a number of those which are interesting geochemi-
3.2 Single element spatial distributions
cally are indicative of two elemental suites of potential anthropogenic
To examine the patterns of areal distribution of the elements of inter- importance. The first is a noble and base metal (BNM) suite composed
est and to visualize the correspondence between elemental concentra- of the elements Pb, Tl, Ba, Ag, Au, Cu, and Sb, which is most likely indica-
tions and archaeological features, the raw data for each element were tive of the lead ore and possibly other material processed on site. The
plotted initially as Classed Post Maps (CPM) and then as contours of second is a suite composed of the elements Fe, Ti, Nb, Mn, Co, Cu, P, Li,
concentrations with an overlay of sampling locations as shown for Ag, Rb, Sc, Cs, K, Ga, P, Zr, Th, and Sn (PEG), which may be indicative of peg-
Au, Pb, Sn, and Tl in Figure 3. Contouring was chosen as the means matite lithology (London & Kontak, 2012) and could be reflective of the
of data presentation because it provides a clearer picture of elemen- treatment of Sn-bearing pegmatite on this site.
tal distribution. The interpolated contours conform to the known and
interpreted archaeology and were generated using the kriged CPM
data in Surfer
R
12 software.
3.4 Principal component analysis
The single element diagrams (Fig. 3) show that in the case of Pb
and Tl, large areas of anomalism exist at St. Algars Farm. For Ag Principal component analysis (PCA) confirmed the elemental asso-
and Au, the zone of anomalous soils is much smaller but covers the ciations deduced from the correlation analysis, and demonstrated
area of Roman lead processing defined by gradiometric magnetometry a credible relationship between the spatial location of samples and
(Lambdin, 2011) and excavation (Lambdin & Holley 2011b, 2012b). Tin their elemental suites (Fig. 4). While the complete set of principal
is most anomalous southeast of the lead processing area but elevated components explains all of the variance in the multivariate data, the
concentrations extend into that area. lower order components capture most of this variance and are most
useful for discriminating observations. Four components collectively
explaining nearly 80% of the multivariate variance were obtained (PC1
3.3 Correlations between elements
32.1%; PC2 23.1%; PC3 14.2%; PC4 8.4%). The PC1PC2 space pro-
A summary of the correlation analyses is shown in Table II. It includes vided clear discrimination of element suites and observations, and
the correlations for the 25 elements that provide the most useful higher order components did not separate observations adequately
6 SYLVESTER ET AL.

are associated in the NE quadrant and correspond to the Pegmatite


suite deduced by correlation analysis; the PEG group also includes
Mn, Co, Cu, Sn, Th, and Zr. These elements are often associated
with pegmatite lithology; ilmenite that contains Fe and Ti is also
commonly found in high concentrations in pegmatite. The observa-
tions from metal processing areas and having Sn > LDL fall between
element groupings corresponding to the BNM and pegmatite lithology
suites. This is consistent with our interpretations, since some of the
(now weathered to soil) ore material is suspected to be pegmatitic,
with subsequent extraction and processing of metals in the BNM
suite.

3.5 Spatial distribution of element indices


While individual elements in a suite can be used to define an area
of particular interest, examination of an elemental group (a compos-
ite index) can commonly provide better detail and definition of the
phenomenon giving rise to the anomalous suite (Mann, de Caritat,
& Prince, 2012; Smith, Campbell, & Litchfield, 1984). Our analysis of
the St. Algars Farm MMI datacorrelation analysis, PCA, and multi-
element plotsleads logically to the use of such indices. The index
points can be plotted and contoured to produce an Index plot. This has
been done for two suites, BNM and PEG for the area sampled at St.
Algars Farm (Fig. 5).
The Noble and Base Metal suite have been compiled into the BNM
index (Fig. 5a). It is a simple additive index constructed by taking the
F I G U R E 3 Spatial distributions of MMI Ag, Au, Pb, Sn, and Tl in soil at normalized to the mean values for the elements Ba, Ag, Au, Cu, Pbcalc ,
St. Algars Farm, using universal kriging (linear trend in log10 (MMI Zn)) Sn, and Tl. Index units are ppb. The contoured CPM values of the
with exponential variogram models with fitted nugget implemented in BNM index are shown superimposed on the gradiometric magnetic
geoR (Ribeiro & Diggle, 2015). Lower right panel shows estimated dis-
base image (Lambdin, 2011) to show the relationship with the inter-
tribution of kriging variance using the example of Tl, on a log10 (MMI Tl)
scale. [Color figure can be viewed at wileyonlinelibrary.com] preted archaeology in Figure 5a. The BNM index defines two quite dis-
tinct zones. The area between the 100 and 300 ppb contours includes
the settlement area and areas of known human activity. The 300 ppb
(see, e.g., PC3PC4 space in Fig. 4). PC1 was dominated by posi- contour encloses the known major areas of metal processing. The evi-
tive loadings for the refractory elements Ga, Sc, and Ti and negative dence suggests that roasting/smelting of primary ore was carried out
loadings for Ag, Au, Ba, and Ca. PC2 had the largest positive load- in and around Building 1, while silver cupellation of the smelted lead
ings for Ag, Cu, Sb, and Ti, with negative loadings for U, Y, and REE was undertaken in the winged villa area. The BNM index is a very valu-
(Fig. 4). able tool for defining the distribution of previous human activity on
Figure 4 shows a number of interesting features. Observation the site.
scores for soil sampled from the known metal processing areas The PEG + index is a multiplicative index constructed by multipli-
(,) coincide with the component loadings for the BNM element cation of the normalized to the mean values for the elements Nb, Zr,
suite. Similarly, observation scores for soil samples having MMI Sn Rb, Th, and Cs. The PEG + index (Fig. 5b) highlights not only a sam-
concentrations >LDL () coincide with the component loadings for ple (SAF15) at the southwestern edge of the sampled area, but sev-
the PEG suite. Local lithology (+) is characterized in PC1PC2 space eral samples on the eastern side, adjacent to the stream that drains
by an association of REE (including Y), refractory elements (Th, Zr), the St. Algars site. The PEG + index plot is very different from that of
elements that can be associated with carbonates (Ca, Mg, Sr, Cd, U, the BNMI plot and may be indicative of the presence of pegmatite or
Zn), and elements commonly cooccurring with Mn (Mn, Co, Ni). The Sn processing in the southern part of the area in the vicinity of sample
presence of Ca, Mg, and Sr in the local lithology suite and relatively SAF15.
high Ca values (Table II) confirm the underlying bedrock has a high
component of limestone. The elements Pb, Ba, Tl, Cu, Ag, and Au
3.6 The source of the lead processed at St. Algars
also show a high degree of similarity and are marked as the base and
Farm
noble metal (BNM) suite in the NW quadrant. A number of elements:
Li, K, Cs, Rb, with large ionic radius, or Ga, Sc, Ti, and Nb (and to a The Charterhouse lead mines, located about 40 km west northwest
lesser extent Fe and P) having high field strength (charge/radius ratio) of St. Algars Farm (see Fig. 1) have been considered (Dungworth,
SYLVESTER ET AL. 7

TA B L E I I St. Algars Farm MMI Pearson correlation coefficients (r) for elements of interest

Ba Cd Cs Er Eu La Li Nb Rb Sb Sn Tl Zr Ag Au Ce Co Cu Fe P Mn Ni Ti Pb Zn
Ba 1.00
Cd -0.08 1.00
Cs -0.06 -0.21 1.00
Er -0.41 0.38 0.17 1.00
Eu -0.28 0.34 0.10 0.94 1.00
La 0.08 0.07 0.25 0.71 0.82 1.00
Li 0.24 -0.02 0.23 0.18 0.29 0.45 1.00
Nb 0.15 0.01 0.47 0.25 0.24 0.50 0.59 1.00
Rb -0.25 -0.11 0.69 0.32 0.18 0.16 0.15 0.45 1.00
Sb 0.43 -0.14 0.02 -0.20 -0.14 0.20 0.14 0.28 0.01 1.00
Sn 0.18 -0.30 0.38 0.06 0.03 0.29 0.33 0.48 0.38 0.26 1.00
Tl 0.69 -0.13 0.02 -0.19 -0.07 0.38 0.27 0.32 -0.24 0.73 0.24 1.00
Zr -0.16 -0.05 0.45 0.67 0.70 0.80 0.46 0.56 0.51 0.15 0.39 0.13 1.00
Ag 0.75 -0.27 -0.15 -0.59 -0.47 -0.08 0.16 0.08 -0.42 0.43 0.10 0.75 -0.35 1.00
Au 0.60 -0.10 -0.18 -0.38 -0.27 0.01 0.32 0.14 -0.37 0.35 0.01 0.56 -0.19 0.70 1.00
Ce -0.08 0.48 0.16 0.78 0.86 0.83 0.37 0.42 0.09 0.03 0.04 0.18 0.70 -0.26 -0.05 1.00
Cr -0.48 -0.01 0.08 0.44 0.40 0.14 -0.10 -0.10 0.39 -0.28 -0.04 -0.58 0.42 -0.59 -0.47 0.21 0.43
Cu 0.58 -0.47 0.04 -0.46 -0.35 0.06 0.40 0.26 -0.15 0.46 0.28 0.62 -0.06 0.74 0.67 -0.22 -0.17 1.00
Fe 0.03 -0.05 0.40 0.52 0.56 0.79 0.52 0.71 0.44 0.30 0.52 0.27 0.81 -0.12 -0.05 0.55 0.27 0.22 1.00
P 0.22 -0.13 0.55 0.15 0.11 0.38 0.48 0.74 0.62 0.38 0.62 0.28 0.56 0.09 0.14 0.20 -0.05 0.39 0.73 1.00
Mn -0.57 0.14 0.13 0.39 0.29 0.05 0.06 0.01 0.43 -0.30 0.10 -0.55 0.33 -0.71 -0.44 0.18 0.74 -0.33 0.20 0.10 1.00
Ni 0.04 0.23 -0.29 0.32 0.46 0.46 0.31 0.05 -0.49 -0.13 -0.18 0.25 0.25 0.15 0.29 0.56 0.20 0.09 0.14 -0.18 -0.08 1.00
Ti 0.03 -0.34 0.65 0.26 0.21 0.43 0.50 0.72 0.73 0.20 0.65 0.10 0.70 -0.12 -0.06 0.21 0.24 0.24 0.75 0.86 0.25 -0.21 1.00
Pb 0.52 0.12 -0.11 -0.06 0.03 0.28 0.10 0.12 -0.32 0.70 0.05 0.85 -0.01 0.52 0.43 0.24 -0.32 0.40 0.11 0.11 -0.43 0.18 -0.13 1.00
Zn -0.29 0.83 -0.04 0.56 0.47 0.20 0.11 0.17 0.15 -0.18 -0.09 -0.28 0.21 -0.50 -0.24 0.49 0.05 -0.43 0.24 0.11 0.43 0.19 -0.05 -0.08 1.00

Ba Cd Cs Er Eu La Li Nb Rb Sb Sn Tl Zr Ag Au Ce Co Cu Fe P Mn Ni Ti Pb Zn

Very strong 0.67 Interesting geochemically


Strong
Moderately Strong

Comeau, & Lowerre, 2013; Dunster & Dungworth, 2012) a possible The very strong multielement correlation between the St. Algars
source of the lead, which was processed at the site. To investigate Farm soils (especially those in the area associated with the transport
this possibility, five soil samples were collected from around the and processing of the lead ore) and those from Charterhouse provide
Charterhouse Lead mine and subjected to multielement MMI analysis. strong support for the hypothesis that the Charterhouse lead mines
The results were compared with those obtained from the St. Algars were the source of the lead processed at St. Algars Farm.
site using the Degree of Geochemical Similarity (DOGS) statistical
method described in Mann, de Caritat, and Sylvester (2016). The
3.7 Possible Sources for Tin at St. Algars Farm
log-transformed mean elemental concentration for each element of
the Charterhouse soils was used as the comparator against which all The closest possible source of Sn ore available in the Roman period
of the log-transformed St. Algars Farm samples were evaluated. The would have been from Dartmoor in Devon, a distance of about 130
DOGS correlation coefficients reveal a pattern very similar to that km to the southwest, where tin was mined by the Romans at Erme Val-
shown by the plot of the BNM index for the St. Algars Farm MMI ley (Thorndycraft Pirrie, & Brown, 2004) which is located to the south
samples and is strongly indicative that the ore materials processed at of Dartmoor. It is also known that tin from Cornwall was used in the
St. Algars Farm were sourced from the Charterhouse lead mines. Cor- manufacture of Roman pewter as remnants of pewter molds have been
relation coefficient values range from 0.798 in the area of geochemical found at the tin mining site of Leswyn St. Just (Henig, 2002).
background to 0.898 near the northwest trackway, along which it is In order to establish a possible provenance for the Sn observed in,
believed the ore was transported to the metal processing site. Values for example, sample SAF15 (and four others with similar multi-element
of greater than 0.880 were obtained for samples (13) from northwest profiles, as indicated by DOGS r correlation coefficient values of >
trackway area, the metal processing area and areas adjacent to it. A 0.95) at St. Algars Farm, a small number of additional MMI samples
contoured plot of the DOGS correlation coefficient values is shown were taken at likely sites for Sn in Devon and Cornwall. Table III con-

in Figure 6. This plot shows a strong similarity to that of the BNMI tains their locations, and in addition the DOGS correlation factor r

plot in Figure 5a. It is noteworthy that some of the samples with the (Mann et al., 2016) for these samples versus the SAF15 sample.

highest values are those which are enriched in Ba, a trace element Sample DC05 from the banks of the River Plym, immediately south
present as the accessory mineral barite in the Charterhouse ore of Dartmoor has the highest correlation with SAF15 (r = 0.820), but
and not commonly present in this type of Pb deposit in the United samples DC04 from the banks of the River Erme, and the GEMAS
Kingdom. sample #3449 from a field near Penzance in Cornwall also have high
8 SYLVESTER ET AL.

F I G U R E 5 Plots (a) base and noble metal index (BNM) (b) pegmatite
+ index (PEG) at St. Algars Farm. The scheduled area that consists of
the winged villa, cupellation zone, and burnt area is shown as a square,
while Building 1 lies to the north of it in the center of the sampled area.
The locations of these archaeological features are shown in Figure 2.
Contours are superimposed on a gray magnetic gradiometry base

F I G U R E 4 Biplots for principal components analysis (PCA) of MMI


compositional data for soil samples from St. Algars Farm, Somerset,
UK. Plot (a) shows observations and component loadings in PC1-PC2
space; plot (b) shows the equivalent for PC3-PC4 space
F I G U R E 6 Contoured plot of DOGS correlation coefficients for log-
transformed St. Algars Farm MMI multielement soil data compared
correlation (r values = 0.784). All of the samples in Table III had sig- with the log-transformed multielement means for the Charter house
nificant concentrations of MMI Sn (the highest Sn = 160 ppb being in soil
the River Plym DC05 sample), and other elements diagnostic of peg-
matites or s-type granites, for example, Cs, Nb, and Ti. These elements
are included in the pegmatite index, and of course in the calculation of SAD with concentrated aqua regia. Comparison of SAD results with
DOGS r values. the MMI results has been undertaken for a range of elements to eval-
uate the relative effectiveness of MMI to provide meaningful results
in the documentation of the archaeological features of this site. The
3.8 Comparison of St. Algars Farm soil MMI and
geochemical contrast was determined for a range of elements of inter-
strong acid digest results
est for each technique. The contrast chosen was maximum/minimum
To evaluate the comparative efficiency of MMI extraction with SAD, rather than 4QM/1QM because only 10 samples were analyzed by
10 of the MMI samples selected to represent the range of geochem- both techniques and quartile statistics were considered not to be
ical environments (background to anomalous) were also subjected to meaningful. Comparative figures are shown in Table IV.
SYLVESTER ET AL. 9

TA B L E I I I DOGS r values for MMI samples from Devon and Corn- TA B L E V Comparison of MMI SAF and MMI GEMAS 1QM values
wall versus sample SAF15
St. Algars (SAF) GEMAS SE England SAF/GEMAS
Sample Location County Dogs r-Value Element 1Q Mean 1Q Mean 1QM/1QM
SAF15 St. Algars farm Somerset 1 Ag 4.75 2.64 1.80
GEMAS #3449 Penzance Cornwall 0.784 As 10 8.23 1.22
DC04 R. Erme Dartmoor Devon 0.784 Au 0.05 0.5 0.10
DC05 R. Plym Dartmoor Devon 0.820 Ba 693 237 2.92
DC06 Two Bridges Dartmoor Devon 0.241 Ca 590,938 552,140 1.07
DC07 Dartmeet Devon 0.600 Ce 153 12 12.75
DC08 R. Dart Dartmoor Devon 0.726 Cu 413 750 0.55
BO09 St Neot R. Bodmin Cornwall 0.629 Fe 23,560 16,000 1.47
CO10 R. Fal Grampound Cornwall 0.427 Nb 0.25 0.25 1.00
CO11 Wheal Jane Truro Cornwall 0.378 P 460 1729 0.27
Pb 5829 150 38.86
TA B L E I V Mean geochemical contrast (maximum/minimum) for Sb 0.94 2.59 0.36
selected elements for MMI and SAD analyses from 10 selected soil Sn 0.5 0.26 1.92
samples from St. Algars Farm
Ti 25.44 12 2.12
Element MMI Contrast SAD Contrast MMI/SAD Contrast
Tl 1.13 0.25 4.52
Ag 15.6 14.18 1.1
Zn 200 461 0.43
Zn 2.56 1.63 1.6
Elemental values are in ppb. Students t tests conducted on the 1QM values
Cu 6.85 3.42 2.0 from both the SAF and GEMAS elemental suites demonstrated in all cases
Ba 2.26 1.6 1.4 that the differences were significant at the 99.9% level.

Tl 1.6 1.14 1.4


Sn 51.25 30.94 1.7
interest are shown in comparison with the GEMAS 1QM values in
Sb 173 57.67 3.0
Table V.
Ce 1.43 1.08 1.3 Comparison of the 1QM MMI data for St. Algars farm and GEMAS
Th 1.43 1.07 1.3 shows that a large number of elements at SAF are anomalous com-
Zr 1.29 1.03 1.3 pared to GEMAS SE England, even at background concentrations. Pb
Cs 2.05 1.02 2.0 (38.9) and Ce (12.8) are strongly enriched at St. Algars Farm; Tl
Rb 2.15 1.13 1.9 (4.52) is moderately enriched whilst Au (0.1), P (0.27), Sb (0.36),
Nb 1.64 1.1 1.5 and Zn (0.43) are depleted. It is probable that the relative enrichment
Li 1.88 1 1.9 of Pb and Tl represented by the 1QM is indicative of the widespread
background pollution to soils on the St. Algars Farm site from the
Roman lead processing activities. That pollution probably resulted
For all 13 elements the MMI extraction provided a greater geo- from the roasting and smelting of Pb ore, which produced consid-
chemical contrast than did the SAD, a direct consequence of the lower erable lead vapors and litharge (PbO), much of which was dumped
background concentrations achieved by MMI because it does not sig- or stored at various places, in or adjacent to the metal processing
nificantly dissolve soil matrix materials. areas on the site (Dunster & Dungworth, 2012). The fact that the
1QM concentrations of other elements associated with the Pb pro-
cessing (Au, Ag, Cu, Sb, and to a lesser extent Ba) are not very dif-
ferent from the GEMAS concentrations is indicative that these met-
4 DISCUSSION
als were not widespread across the site (thereby creating an ele-
vated SAF 1QM had they been so), but are substantially contained
4.1 Geochemical context of St. Algars Farm
within the metal processing areas and associated areas on human
To place background concentrations from St. Algars Farm into a activity.
more regional context, the background values (1QM) were compared
with the MMI 1QM values obtained from agricultural soil samples
4.2 The effectiveness of MMI in defining
taken in southeast England as part of the Geochemical Mapping of
multielement anthropogenic soil geochemical
Agricultural Soils (GEMAS) regional soil sampling program (Reimann
anomalies
et al., 2014a, 2014b). Fifty-four samples taken approximately 50 km
apart comprise the GEMAS southeast England database used for The distribution plot of the BNM suite elements (Fig. 2) clearly delin-
this comparison. The St. Algars Farm 1QM values for elements of eates an area of anomalously high elemental concentrations over the
10 SYLVESTER ET AL.

areas in which excavation has confirmed that lead processing was car- Cr, As, Sc, Al, Ga, Rb, K, Li, Zr, Th, P, Cr) presented by sample SAF15,
ried out and in the broader zone of human settlement (Lambdin & Hol- located near the southern margin of the investigated area.
ley, 2011b, 2012). The wide range of elements, combined with the abil- A further indication of the effectiveness of multielement MMI soil
ity to detect anomalies at low concentrations and the high geochemical geochemistry was provided by the comparative study of the soils at St.
contrast for many elements demonstrate that MMI clearly defines the Algars Farm with those in the vicinity of the Charterhouse lead mine
lead processing areas. The BNM index plot graphically augments the and known tin mining sites in Devon and Cornwall. The results strongly
distribution plots and provides greater definition of the elements that support the contention that the Pb processed at St. Algars Farm was
characterize the processing and also the spatial limits of those areas. sourced from the Charterhouse mines, and that Sn bearing material
The results also indicate that precious metals may have been extracted from Devon and/or Cornwall has been incorporated into metallurgical
in areas other than just the predominant Pb processing site. procedures at the St. Algars Farm site.
There is evidence from both the pegmatite (PEG) elemental suite
and the comparison with s-type granite samples from Devon and
Cornwall (Table III) that processing of tin-bearing pegmatite has 5 CONCLUSIONS
taken place on the St. Algars site. Tin, which has been found at
concentrations (from strong-acid digest) of up to 164 ppm at St. Soil analysis using MMI clearly defines and characterizes multielemen-
Algars Farm (G. Sylvester unpublished data), is anomalous in the tal suites (Pb, Tl, Au, Ag, Sb, Cu, Ba, Nb, and Sn), which delineate the
MMI analyses on this site in several samples (including the type documented metal processing activities on this site. MMI also outlined
samples of the two defined elemental suites) and its presence at new geochemical anomalies (of the same elements) within the area of
these concentrations is unexpected, as the underlying geology pre- metal processing and human settlement, which may indicate significant
cludes it being of in situ origin. It is almost certain that the tin used buried archaeological features and which require follow-up investiga-
in the pewter manufacture at nearby (25 km) Camerton was Cor- tion. In addition, other elemental suites (Nb, Ti, Sn, Fe, Cr, As, Sc, Al, Ga,
nish. Pewter from Camerton and other local sites in Somerset (e.g., Rb, K, Li, Zr, Th, P, Cr) have defined geochemical targets which, although
Lansdown, 5 km north of Bath) was used extensively in Bath (about not located in areas of former human settlement (as defined by the
25 km away) for the production of vessels and Bath Curse Tablets magnetic gradiometry), may be indicative of tin ore processing on the
(Flint and Gordon 1999) and it is possible that the tin found at St. site and require follow-up work.
Algars Farm was used to produce pewter, as fragments of pewter MMI soil geochemistry has been shown here to be an effective
were recovered during field walking during the 1970s (Lambdin & archaeological prospection and documentation tool. Sampling can be
Holley, 2011b). readily undertaken by field personnel and the analyses are relatively
The MMI multielement concentration plots for the metal process- inexpensive given that high-quality data for up to 53 elements can be
ing area and pegmatite lithology samples have shown that each group obtained at low ppb concentration detection limits when linked with
displays enrichment in a characteristic elemental suite. However, there ICP-MS. Discrimination against dissolution of the soil matrix by the
are some elements that are found to be present in elevated concen- MMI extraction solution has allowed anthropogenic derived sources of
tration in both the metal processing and pegmatite lithology groups. anomalism to be enhanced. These features make MMI an ideal tool for
These elements include Sn, as noted above, Nb, Ti, and to a lesser archaeological application as it allows the identification and use of a
extent Cs, Ga, Li, and Tl. The overlap of these elements between groups wider range of elements that may have anthropogenic significance.
may mean that there was also an overlap of Pb and Sn processing activ- It is concluded that MMI soil geochemistry is a valuable tool for
ities on the site (as indicated by the overlapping distribution of Sn locating and defining areas of potential archaeological significance on
between the BNM and PG elemental suites in Fig. 4). It could also be this site and is made significantly more powerful when integrated with
showing that pollution from these individual activities was widespread the results of geophysical investigations.
on the site as has been stated for the Pb processing by Dunster and
Dungworth (2012).
ACKNOWLEDGMENTS
One of the results of this work has been the delineation of MMI
geochemical anomalies, which are located in areas that appear to be The contribution from SGS minerals in providing analysis results for

underlain by potentially interesting archaeology and which have not, the soil samples is gratefully acknowledged. The archaeological advice

as yet, been investigated. These targets include a very strong multi- and assistance with the sampling provided by Dr. David Dungworth is

element anomaly of (Pb, Au, Ag, Pb, Tl, Sb, Cu, Ba, Nb, Sn) over what also gratefully acknowledged. Statistical advice was supplied courtesy

appears, from the magnetic gradiometry, to be a building (Building 1), of Martin Firth, University of Western Australia. The authors would

located about 100 m north of the winged villa within the metal pro- also like to thank the anonymous reviewers for their valuable com-

cessing area defined by the MMI geochemistry. There is a similar, but ments and suggestions that have considerably improved this paper.

slightly weaker anomaly of the same elements (Area 1) located adja-


cent to the Building 1 (Fig. 2) anomaly and a number of spot high
REFERENCES
anomalies on, or adjacent to, the interpreted northwest trackway in
Arrhenius, O. (1931). Die Bodenanalsye in dienstder Archaeologie.
the zone of human settlement. These anomalies all warrant follow-up Zeitschrift fur pflanzenernarhung, Dungung und Bodenkunde B10
investigation as does the strong multielement anomaly (Nb, Ti, Sn, Fe, Jahrgang, 427439.
SYLVESTER ET AL. 11

Aston, M. A., Martin, M. H., & Jackson, A. W. (1998). The use of heavy metal Lambdin, C., & Holley, R. (2011b). St. Algars Project Group, St. Algars
soil analysis for archaeological surveying. Chemosphere, 37(3), 465477. Villa Excavation Summary, Bath and Camerton Archaeological Society
Beck, M. (2007). Midden formation and intrasite chemical patterning in (Unpublished report).
Kalinga, Philippines. Geoarchaeology, 22(4), 453475. Lambdin, C., & Holley, R. (2012a). St. Algars Project Group, project
Bethell, P. H., & Carver, M. O. H. (1987). Detection and enhancement of design. St. Algars Roman Villa Evaluation Trenches, Bath and Camerton
decayed inhumations at Sutton Hoo. In A. Boddington, N. Garland, & R. Archaeological Society (Unpublished report).
Janaway (Eds.), Death decay and reconstruction (pp. 1021). Manchester: Lambdin, C., & Holley, R. (2012b). St. Algars Project Group. Archaeological
Manchester University Press. Field Evaluation, Bath and Camerton Archaeological Society (Unpub-
Bintliff, J. L., Davies, B. E., Gaffney, C. F., Snodgrass, A., & Waters, A. (1992). lished report).
Trace element accumulations on and around ancient settlements in Leduc, C., & Itard, Y. (2003). Low sampling density exploration geochemistry
Greece. Geoprospection in Archaeological Landscapes, P. Spoerry (ed.), for gold in arid and tropical climates: Comparison between conventional
Oxbow Books, Oxford. geochemistry and BLEG. Geochemistry: Exploration, Environment, Analy-
British Geological Survey. (2015). Geology of Britain viewer. Retrieved from sis, 3, 121131.
mapapps.bgs.ac.uk/geologyofbritain/home.html London, D., & Kontak, D. J. (2012). Granitic pegmatites: Scientific wonders
Caritat, P. de., & Cooper, M. (2011). National Geochemical Survey of Aus- and economic bonanzas. Elements, 8, 257261.
tralia: The geochemical atlas of Australia, 2011/20, 557. Geoscience Mann, A. W., Birrell, R. D., Mann, A. T., Humphreys, B., & Perdrix, J. L. (1998).
Australia Record, Geoscience Australia. Application of the mobile metal ion technique to routine geochemical
Cook, D. E., Kovacevich, B., Beach, T., & Bishop, R. (2006). Deciphering the exploration. Journal of Geochemical Exploration, 61(13), 87102.
inorganic chemical record of ancient human activity using ICP-MS: A Mann, A., Reimann, C., Caritat, P. de., & Turner, N. (2014). Mobile Metal Ion
reconnaissance study of late Classic soil floors at Cancun, Guatemala. Analysis of European Agricultural Soil. Chemistry of Europes agricul-
Journal of Archaeological Science, 33(5), 628640. tural soil. Geologisches Jahrbuch B,. B. Reimann C., M., Demetriades, A.,
Cook, S. R., Banerjea, R. Y., Marshall, L.-J., Fulford, M., Clarke, A., & van Filmoser, P. & OConnor, P. 103, 203-231.
Zwieten, C. (2010). Concentrations of copper, zinc and lead as indica- Mann, A., Reimann, C., Caritat, P. de., Turner, N., Burke, M. 2015. Mobile
tors of hearth usage at the Roman town of Calleva Atrebatum (Silch- Metal Ion analysis of European agricultural soils: bioavailability, weath-
ester, Hampshire, UK). Journal of Archaeological Science, 37(4), 871 ering, geogenic patterns and anthropogenic anomalies. Geochemistry:
879. Exploration, Environment, Analysis, 15, pp 99112.
Cook, S. R., Clarke, A. S., & Fulford, M. G. (2005). Soil geochemistry and Mann, A. W. (2010). Strong versus weak digestions: Ligand-based soil
detection of early Roman precious metal and copper alloy working at extraction geochemistry. Geochemistry: Exploration, Environment, Analy-
the Roman town of Calleva Atrebatum (Silchester, Hampshire, UK). sis, 10(1), 17-26.
Journal of Archaeological Science, 32(5), 805812. Mann, A. W., de Caritat, P., & Prince, P. (2012). Bioavailability of nutrients in
Cranfield University. (2015). LandIS Soil Portal. Retrieved from Australia from mobile metal ion analysis of catchment outlet sediment
www.landis.org.uk samples: Continental-scale patterns and processes. Geochemistry: Explo-
Davidson, D. A., Dercon, G., Stewart, M., & Watson, F. (2006). The legacy of ration, Environment, Analysis, 20, 116.
past urban waste disposal on local soils. Journal of Archaeological Science, Mann, A. W., de Caritat, P., & Sylvester, G. C. (2016). Degree of geochemical
33, 778783. similarity (DOGS): A simple statistical method to quantify and map affin-
Dungworth, D., Comeau, B., & Lowerre, A. (2013). St Algars, Selwood, ity between samples from multi-element geochemical data sets. Aus-
SomersetGeochemical survey. Technology Report. Portsmouth UK: tralian Journal of Earth Sciences, 63(1), 111122.
English Hertiage, 37 p. Maskall, J., Whitehead, K., Gee, C., & Thornton, I. (1996). Long-term migra-
Dunster, J., & Dungworth, D. (2012). St. Algars Farm, Frome, Somerset tion of metals at historical smelting sites. Applied Geochemistry, 11(12),
The analysis of lead: Working wasteTechnology report. Research 4351.
report. Portsmouth: English Heritage. Maskall, J., Whitehead, K., & Thornton, I. (1995). Heavy metal migration in
Dunster, J., Dungworth, D., & Lowerre, A. (2012). Characterising metal- soils and rocks at historical smelting sites. Environmental Geochemistry
working through a geochemical survey of plough soil Research Report and Health, 17, 127138.
Series, 30-2012. Portsmouth: English Heritage, 44 p. McBride, M. B. (1989). Reactions controlling heavy metal solubility in soils.
Entwistle, J. A., Abrahams, P. W., & Dodgshon, R. A. (2000). The geoarchae- In B. A. Stewart (Ed.), Advances in soils science (Vol. 10, pp. 156). New
ological significance and spatial variability of a range of physical and York: Springer-Verlag.
chemical soil properties from a former habitation site, Isle of Skye. Jour- Middleton, W. D. (2004). Identifying chemical activity residues on prehis-
nal of Archaeological Science, 27, 287-303. toric house floors: A methodology and rationale for multi-elemental
Flint, V. & Gordon, R. (1999). Witchcraft and magic in Europe, Ancient characterization of a mild acid extract of anthropogenic sediment.
Greece and Rome. Witchcraft and Magic in Europe. B. R. L. M. Gijswijt- Archaeometry, 46(1), 4765.
Hofstra, R. Porter. London, The Athlone Press. 2. Mighall, T. M., Grattan, J. P., Lees, J. A., Timberlake, S., & Forsyth, S. (2002).
Grattan, J. P., Gilbertson, D. D., & Kent, M. (2013). Sedimentary metal- An atmospheric pollution history for leadzinc mining from the Ystwyth
pollution signatures adjacent to the ancient centre of copper metallurgy Valley, Dyfed, mid-Wales, UK as recorded by an upland blanket peat.
at Khirbet Faynan in the desert of southern Jordan. Journal of Archaeo- Geochemistry: Exploration, Environment, Analysis, 2, 175184.
logical Science, 40(11), 38343853. Oonk, S., Slomp, C. P., & Huisman, D. J. (2009). Geochemistry as an aid in
Henig, M. (2002). The art of Roman Britain. Routeledge. archaeological prospection and site interpretation: Current issues and
research directions. Archaeological Prospection, 16(1), 3551.
Lambdin, C. (2011). St. Algars Project Group, Geophysical Survey. St.
Algars Roman Villa, Bath and Camerton Archaeological Society R Core Team (2015). A language and environment for statistical computing.
(Unpublished report). Vienna, Austria: R Foundation for Statistical Computing.
12 SYLVESTER ET AL.

Reimann, C., Birke, M., Demetriades, A., Filzmoser, P., & OConnor, P. Schlezinger, D. R., & Howes, B. L. (2000). Organic phosphorus and elemental
(2014a). Chemistry of Europes agricultural soilsPart B: General back- ratios as indicators of prehistoric human occupation. Journal of Archaeo-
ground information and further analysis of the GEMAS data set. Han- logical Science, 27, 479492.
nover: Schweizerbarth. Smith, R. E., Campbell, N. A., & Litchfield, R. (1984). Multivariate statistical
Reimann, C., Birke, M., Demetriades, A., Filzmoser, P., & OConnor, P. techniques applied to Pisolitic Laterite at Golden Grove, Western Aus-
(2014b). Chemistry of Europes agricultural soilsPart A: Methodology tralia. Journal of Geochemical Exploration, 22(1), 193216.
and interpretation of the GEMAS data set. Hannover: Schweizerbarth. Stanley, C. R., & Noble, R. R. P. (2008). Quantitative assessment of the accu-
Reimann, C., Filzmoser, P., & Garrett, R. G. (2002). Factor analysis applied to racy and precision of exploration techniques using minimum probability
regional geochemical data: Problems and possibilities. Applied Geochem- methods. Geochemistry: Exploration, Environment, Analysis, 8, 115127.
istry, 17, 185-206. Thorndycraft, V. R., Pirrie, D., & Brown, A. (2004). Alluvial records of
Reimann, C., Filzmoser, P., Garrett, R. G., & Dutter, R. (2008). Statistical data medieval and prehistoric tin mining on Dartmoor, southwest England.
analysis explainedApplied environmental statistics with R. Chichester: Geoarchaeology, 19(3), 219-236.
John Wiley & Sons
Ribeiro, P. J., Jr., & Diggle, P. J. (2015). geoR: Analysis of geostatistical data.
R package version 1.7-5.1. How to cite this article: Sylvester GC, Mann AW, Rate AW,
Sadeghi, M., Albanese, S., Morris, G., Ladenberger, A., Anderson, M., Can- Wilson CA. Application of high resolution Mobile Metal Ion
natelli, C., Lima, A., & DeVivo, B. (2015) REE concentrations in agricul- (MMI) soil geochemistry to archaeological investigations: An
tural soil in Sweden and Italy: Comparison of weak MMI extraction with
example from a Roman metal working site, Somerset, United
near total extraction data. Applied Geochemistry, 63, PP 2236.
Kingdom. Geoarchaeology. 2017;00:112. https://doi.org/
Sampiettro, M. M., & Vattuone, M. A. (2005). Reconstruction of activity
areas at a formative household in Northwest Argentina. Geoarchaeology,
10.1002/gea.21618
20(4), 337354.