Download - LEAD ISOTOPE ANALYSES OF BRONZE AGE COPPER-BASE ARTEFACTS FROM AL-MIDAMMAN, YEMEN: TOWARDS THE IDENTIFICATION OF AN INDIGENOUS METAL PRODUCTION AND EXCHANGE SYSTEM IN THE SOUTHERN

Archaeometry

51

, 4 (2009) 576–597 doi: 10.1111/j.1475-4754.2008.00429.x

*Received 19 December 2007; accepted 28 May 2008© University of Oxford, 2008

0Blackwell Publishing LtdOxford, UKARCHArchaeometry0003-813X1475-4754© University of Oxford, 2008XXX

ORIGINAL ARTICLES

Lead isotope analysis of copper-base artefacts from al-MidammanL. Weeks

et al.

*Received 19 December 2007; accepted 28 May 2008

LEAD ISOTOPE ANALYSES OF BRONZE AGE COPPER-BASE ARTEFACTS FROM AL-MIDAMMAN,

YEMEN: TOWARDS THE IDENTIFICATION OF AN INDIGENOUS METAL PRODUCTION AND EXCHANGE

SYSTEM IN THE SOUTHERN RED SEA REGION*

L. WEEKS

Department of Archaeology, University of Nottingham, University Park, Nottingham NG7 2RD, UK

E. KEALL

Near Eastern and Asian Civilizations Department, Royal Ontario Museum, 100 Queen’s Park, Toronto, Ontario M5S 2C6, Canada

V. PASHLEY and J. EVANS

NERC Isotope Geosciences Laboratory, British Geological Survey, Keyworth, Nottingham NG12 5GG, UK

and S. STOCK

Conservation Department, Royal Ontario Museum, 100 Queen’s Park, Toronto, Ontario M5S 2C6, Canada

The results of the lead isotope analysis (LIA) of 15 copper-base artefacts from the BronzeAge site of al-Midamman, Yemen, are reported. The LIA data suggest the existence ofan indigenous Bronze Age metal production and exchange system centred on the southernRed Sea region, distinct from those in neighbouring regions of Arabia and the Levant.These preliminary results are highly significant for the archaeology of the region, suggestingthat local prehistoric copper extraction sites have thus far gone unrecorded, and highlightingthe need for systematic archaeometallurgical fieldwork programmes in the countriessurrounding the southern Red Sea.

KEYWORDS

: ARCHAEOMETALLURGY, LEAD ISOTOPE ANALYSIS, BRONZE AGE, RED SEA, ARABIA, YEMEN, COPPER, CORROSION

© University of Oxford, 2008

INTRODUCTION

A great deal of controversy has surrounded the technique of lead isotope analysis (LIA) in the25 years since it was first applied to study the absolute provenance of ancient copper-baseartefacts in the eastern Mediterranean (Gale and Stos-Gale 1982). The long and often acrimo-nious debate over the presentation and interpretation of lead isotope data (see Weeks 2003,Ch. 6) has in many respects overshadowed the contributions that LIA can make to our under-standing of ancient metal production and exchange systems. In the past decade, the standingof the technique has gradually improved (e.g., Pollard

et al

. 2006, 193–4) in tandem with abetter understanding of its correct application in an archaeological context.

Lead isotope analysis of copper-base artefacts from al-Midamman

577

© University of Oxford, 2008,

Archaeometry

51

, 4 (2009) 576–597

In this paper, we present the results of a small analytical programme of LIA, incorporatingthe analysis of 17 samples from 15 archaeological copper-base artefacts from the Bronze Age site ofal-Midamman, Yemen. When integrated with previous archaeometallurgical and geological studies, thenew data from al-Midamman are able to provide the first tentative evidence attesting to the existenceof an indigenous (Late) Bronze Age regional metal production and exchange system in the southernRed Sea region. Confirmation of this hypothesis and the development of a reliable and nuancedunderstanding of this putative exchange system will require a great deal of further archaeologicaland archaeometric research—a project for which we hope our study will provide some inspiration.

THE SITE OF AL-MIDAMMAN AND ITS DATE

Al-Midamman lies within 2 km of the Red Sea, on Yemen’s Tihamah coastal plain (Fig. 1).The site has been investigated by the Canadian Archaeological Mission since 1997, and consistsof a scattered settlement of ephemeral dwellings spread out over more than 4 km

2

(areas HWA,HWB, HWN and BNF), with intermittent adjacent clusters of upright megaliths reflectingeither alignments or individual commemorative monuments (Keall 1998, 2005; Pringle 1998;see also Khalidi 2006). There is some evidence of periodic human exploitation of the area

Figure 1 A map of the archaeological site of al-Midamman, Yemen, and its regional context.

578

L. Weeks

et al.


Archaeometry

51

, 4 (2009) 576–597

from the Neolithic through to the medieval Islamic period, but this paper focuses exclusivelyon material from the dominant and extensive ‘Bronze Age’ (following Maigret 1984, 426)occupation at the site.

Wind deflation of the exposed areas of the ephemeral Bronze Age occupation at al-Midammanhas left extensive surface scatters of household ceramics, obsidian microliths, grinding stones,stone beads and copper-base tools, interspersed between dunes. Excavations in the settlementareas have produced a similar range of inorganic remains to those found on the surface, inaddition to quantities of fish and some animal bones. However, the most archaeologically visibleBronze Age exploitation of al-Midamman consists of a cluster of standing stones in area HWBthat conceivably formed an alignment oriented towards the mid-winter sunset (Fig. 2). We maycall this here ‘Monumental Phase 1’. It was from this context that the site’s most significantcache of copper-alloy artefacts was unearthed (Giumlia-Mair

et al

. 2002, figs 2 and 3). Thereare other isolated stones toppled,

in situ

, in other parts of the site, as well as clear evidencethat many stones were once removed from their original contexts and re-used in the construc-tion of monumental buildings, best interpreted as temples (areas HWA, BNF). These activitiesrepresent ‘Monumental Phase 2’ at the site. Subterranean graves in area HWN were also builtof re-used stone (and hence also belong to Monumental Phase 2), with ceramic vessels interredas burial offerings. Although the veneration of upright stones morphed into the construction ofmonumental cultic buildings made from recycled megaliths, the utilitarian tools and vesselsfrom both monumental phases are indistinguishable.

The recovery of artefacts from across the site suggests (albeit inconclusively) that domesticsettlement was present during both monumental phases. Certainly, while the megalithic stand-ing stones of area HWB are tied indisputably to the metal cache featured here, other standingstones originally set up over infant burials in HWA may be from an earlier phase of the site’sexploitation. The areas surrounding the standing stones themselves lack the intense concentrationsof artefacts on the surface, but sufficient fragments of all the materials have been unearthed insuch contexts to assert that the cultic megaliths are to be associated with the same materialartefact culture as that present in the domestic settlement.

Figure 2 (a) Area HWB, overview of excavation trenches at the site of the standing stones; (b) the position of the cache artefacts originally beneath the base of a now-toppled upright.


579


Archaeometry

51

, 4 (2009) 576–597

The dating of the various phases of the use and occupation of the site is still somewhat imprecise,though a few fixed points for establishing chronological horizons are available. Specifically, acharred palm fruit from an ashy midden of a domestic settlement in area HWA produced a rawradiocarbon date of 2940 ± 60

bp

, with calibrated ranges from 1380 to 1340 (0.025) and 1320to 970 (0.929)

bc

(IntCal04 calibration curve, OxCal 3.10; Bronk Ramsey 1995, 2001). Thissingle radiocarbon determination is supported by assessment of the excavated and surfacepottery of the domestic settlement and the graves, which have rough parallels of form and detailin the Malayba/Sabir corpus datable from

c

. 1400 to 900

bc

(cf. Keall 1998, figs 9 and 10;Buffa and Vogt 2001, 437). Likewise, general architectural parallels for two of the rectilinearstone buildings from Monumetal Phase 2 (Building A, site HWA, and Building C, site BNF)can be found in Burned Building V at Sabir, dated to the late-second or early-first millennium

bc

(Vogt 1998). A third stone building (Building B, site HWA) has no clear ground planextant, but its distinctive wall decorations have precise parallels with the shallowly carveddesigns on the pillars of the so-called ‘Banat ‘Ad’ temples in Yemen’s desert interior, nowdated as far back as the tenth to ninth century, if not even earlier (Arbach and Audouin 2004,49–51).

The dating evidence for Monumental Phase 2 provides a broad

terminus ante quem

for theerection of the standing stones comprising Monumental Phase 1. However, evidence for theirabsolute age is minimal. The enigmatic ‘collonades’ of standing stones of Hajar al-Ghaimahin the Wadi al-Hamili district (some 100 km to the south of al-Midamman) have generallybeen attributed a Neolithic date of the fourth to third millennium

bc

(Bayle des Hermens1976; Vogt 1998, 124–5), but the site itself is devoid of pottery and its absolute age is equallydebatable. In this regard, the recovery at al-Midamman of a large neck fragment from an ovoidpottery jar immediately adjacent to the base of one of the Monumental Phase 1 megaliths ishighly significant. Not far away, and made in the same potting tradition, a cache of vessels thatmay be called libation pails (Keall 2004, fig. 14) attests to the close association of ceramicswith the standing stones. Moreover, two gold beads were also excavated at this site, similar tothose excavated outside Building B in area HWA. Such findings suggest that the use of thestanding stones of Monumental Phase 1 is not likely to have occurred too much earlier thanthe buildings and graves of Monumental Phase 2. This is commensurate with the assertion, aspresented above, that the material culture on the site remained consistent, despite the changesoccurring by way of monument veneration.

COPPER-BASE ARTEFACTS FROM AL-MIDAMMAN

Findspots for copper-base artefact fragments are moderately widespread across the ceramichorizons, but it is difficult from the current knowledge of the domestic settlement to determinewhich monumental phase the recovered artefacts belong to. In total, 101 copper-base artefactshave been recovered from the surface of the site and from excavations (Fig. 1). Four artefactsderived from surface collection across area HWB are analysed in the present study (Fig. 3),although such a limited number of samples does not necessarily represent a statistically validsampling of the overall assemblage of copper-base artefacts from surface/settlement contexts.

Much more significant for the present study, and for the site in general, was the discoveryin 1997 of a cache of copper-base artefacts set around a large core of obsidian found buriedbeneath one of the pillars of the major surviving standing stone alignment of MonumentalPhase 1 (Fig. 2), or excavated in the immediate vicinity of the monument. A total of 18 artefactsand fragments were recovered from this context (hereafter referred to as ‘the cache’), representing

580

L. Weeks

et al.


Archaeometry

51

, 4 (2009) 576–597

a range of seemingly utilitarian copper-base artefacts such as dagger and knife blades, points,spatulae, adzes and razors (Fig. 3). The 12 analysed samples from the cache and adjacentcontexts comprise the majority of the 17 samples for which data are presented in this study.

Given their archaeological context, it is clear that the artefacts from the cache must havebeen produced at the same time or earlier than the constructions of Monumental Phase 1.Some evidence for their general contemporaneity with Monumental Phase 1 is provided by thediscovery of typologically similar metal artefacts at Sabir in Yemen, which have been radiocarbondated to

c

. 1400 to 900

bc

(Giumlia-Mair

et al

. 2000, 42). In contrast, other typological comparisonswith Syrian artefacts dating from the Early to the Middle Bronze Age support a substantiallyearlier date, and may indicate that the deliberately buried cache of metal tools was producedbetween

c

. 2400 and 1800

bc

(Giumlia-Mair

et al

. 2002, 196, 206–7; Keall 2004, 53). However,the difficulties of the typological approach to the al-Midamman artefacts have been discussed(Giumlia-Mair

et al.

2002, 207), and additional efforts to broadly date the metal artefacts basedupon their composition are, as noted by Giumlia-Mair

et al

. (2000, 42), difficult to verify withouta detailed knowledge of the local developmental sequence of copper alloying in southern Arabia,a field of study that is in its infancy.

PREVIOUS ARCHAEOMETALLURGICAL ANALYSES

Artefacts from the metal cache under the standing stones in area HWB have previously beenstudied using a combination of compositional and metallographic analyses (Giumlia-Mair

et al.

2000, 2002; Shugar not dated a, not dated b). Initial compositional analyses were undertakenusing a combination of inductively coupled plasma atomic emission spectroscopy (ICP-AES),

Figure 3 The copper-base artefacts from al-Midamman analysed in this study (photo: Susan Stock and Heidi Sobol).


581


Archaeometry

51

, 4 (2009) 576–597

wavelength-dispersive spectrometry (WDS) on an electron microprobe, energy-dispersive spectro-metry (EDS) on a scanning electron microscope (SEM) and X-ray diffraction (XRD). The markedcompositional heterogeneity of these generally corroded samples makes the reconstruction ofthe original artefact composition, and thus the classification of alloy types, somewhat difficult.Nevertheless, these studies (Giumlia-Mair

et al

. 2002, table 1; Shugar not dated a, tables 1–3)indicate that the metal artefacts from the cache are copper-base alloys falling generally intothree main compositional groups: a group of ‘tin-bronze’ artefacts with

c

. 2.0 – 3.5% Sn (andsignificant impurities of As, Pb and Fe); a second group with

c

. 1.0 –2.5% As (and significantimpurities of Fe), hereafter referred to as ‘arsenical copper’; and a third group of relativelypure artefacts (with significant impurities of As, Fe and sometimes Ag), hereafter referred toas ‘copper’. Metallographic studies of uncorroded samples and of remnant microstructuralfeatures in corroded samples indicated that the artefacts were produced by casting followed bycold working and annealing to various degrees, and could have acted as functional tools/weapons(Giumlia-Mair

et al

. 2002, 202; Shugar not dated a, not dated b).

LEAD ISOTOPE ANALYSIS

Samples from the al-Midamman copper-base artefacts were taken at the Royal Ontario Museumusing a 1 mm drill bit and forwarded to the NERC Isotope Geosciences Laboratory (NIGL),Keyworth, UK, for analysis. The chosen samples included examples of all the known alloytypes from the site (i.e., copper, arsenical copper and tin-bronze) and come from a variety ofarchaeological contexts, including the cache under the standing stone monument and thesettlement area HWB. Most objects were highly corroded, and therefore samples varied in weight(averaging approximately 30 mg) and in texture (from a fine powder to small metallic chunks).Details of sample condition (i.e., ‘corroded’ or ‘metallic’) are given in Table 1.

Analytical techniques

Pb was extracted from each sample using the following procedure. A quantity of each samplewas placed into an individual, pre-cleaned, Savillex vial and dissolved in ~1 ml Teflon-distilled16 M HNO

3

by heating to 80°C overnight. The HNO

3

was evaporated to dryness and 1 ml of1 M HBr added to each vial. The samples were then left to stand (cold) overnight. Separationof the Pb from the dissolved fraction was achieved using ion exchange methods. Five drops ofcation exchange resin (AG1-X8) were added to pre-cleaned polypropylene columns, eachfitted with a 35

µ

m polyethylene frit. The resin was cleaned by eluting three times with onecolumn volume (CV) of Teflon-distilled 6 M HCl, followed by 1 CV Milli-Q water. The resinwas then pre-conditioned by addition of Teflon-distilled 1 M HBr. The sample was then addedto the column. Any Pb present in the sample forms stable bromide complexes with thepreconditioned column surface; other elements present in the sample matrix are eluted off thecolumn by washing with 1.5 CV of 1 M HBr. The isolated Pb fraction was then eluted offthe column by washing with 1 CV 6 M HCl. This fraction was collected into a pre-cleanedSavillex beaker and 1 ml Teflon-distilled 16 M HNO

3

was added to each individual sample.Each sample was then dried at 100°C overnight. Prior to analysis by multi-collector inductivelycoupled plasma mass spectrometry (MC-ICP-MS), the Pb fraction was taken into solution byaddition of 1 ml 2% Teflon-distilled HNO

3

.Pb isotope analysis of the samples took place over three analytical sessions. Initial analysis

employed a Nu Instruments Nu Plasma HR MC-ICP-MS. The two later sessions employed a

582

L. W

eeks

et al.

© U

niversity of Oxford, 2008,

Archaeom

etry

51

, 4 (2009) 576–597

Table 1

Lead isotope data for copper-base artefacts from al-Midamman, ordered by archaeological context and sample code

Archaeologicalcontext

Samplecode

Artefact Alloy type* Pb concentration

†

Condition

206

Pb/

204

Pb 2

σ

%

207

Pb/

206

Pb 2

σ

%

208

Pb/

206

Pb 2

σ

%

Cache zp97.110 Razor Tin-bronze (+Pb) ICP-AES: 2.44% Corroded 17.736 0.0084 0.87421 0.0046 2.1004 0.0120zp97.111 Spatula Tin-bronze ICP-AES: 0.42% Metallic 17.732 0.0102 0.87414 0.0052 2.1001 0.0129zp97.218 Razor Tin-bronze ICP-AES: 0.48% Corroded 17.735 0.0103 0.87425 0.0035 2.1006 0.0071zp97.224 Adze Copper ICP-AES: 0.26% Corroded 17.745 0.0102 0.87375 0.0035 2.1001 0.0069zp97.231a Point Tin-bronze ICP-AES: 0.37% Corroded 17.736 0.0114 0.87411 0.0039 2.1002 0.0078zp97.232 Dagger Tin-bronze ICP-AES: 0.15% Corroded 17.733 0.0104 0.87414 0.0036 2.1003 0.0072

zp97.236 Spatula Arsenical copperSEM/EDS: not detected

Metallic 17.589 0.0145 0.88348 0.0043 2.1163 0.0075EPMA/WDS: spots up to

c.

5.15%zp97.237 Point Copper ICP-AES: 0.08% Corroded 17.566 0.0111 0.88431 0.0039 2.1170 0.0076zp97.244 Dagger Copper ICP-AES: 0.03% Corroded 17.602 0.0145 0.88298 0.0025 2.1156 0.0061

Excavated nearcache

zp97.201a Dagger/knife

Arsenical copperSEM/EDS: not detectedEPMA/WDS: rare Pb inclusions

Metallic 17.744 0.0096 0.87364 0.0035 2.0995 0.0071zp97.201b Corroded 17.790 0.0085 0.87259 0.0046 2.1009 0.0119zp97.219 Adze Copper ICP-AES: 0.11% Metallic 17.759 0.0150 0.87326 0.0024 2.0997 0.0057

Domestic surface

zp97.67 Lump Copper (+S)SEM/EDS: not detectedEPMA/WDS: not detected

Corroded 18.458 0.0324 0.84395 0.0061 2.0413 0.0132

zp97.71 Razor Copper SEM/EDS: not detected Corroded 17.833 0.0100 0.87117 0.0046 2.1023 0.0120

zp97.156Dagger/knife

CopperSEM/EDS: 0.44%EPMA/WDS: 0.10%

Metallic 18.597 0.0110 0.83896 0.0039 2.0493 0.0075

zp97.216aPoint Arsenical copper ICP-AES: not detected

Corroded 18.356 0.0148 0.84933 0.0058 2.0775 0.0125zp97.216b Metallic 18.124 0.0110 0.85935 0.0036 2.0914 0.0076

* Alloy types are defined in the main text. Additional elements present in concentrations of greater than 1 wt% are listed in parentheses.

† ICP-AES: inductively coupled plasma atomic emission spectroscopy analysis of bulk sample composition (Giumlia-Mair

et al.

2002, Table 1). SEM/EDS: scanning electron microscopy

incorporating energy-dispersive spectrometry area composition scans (Shugar not dated a; Giumlia-Mair

et al.

2002, Table 1). EPMA/WDS: electron microprobe analysis incorporating wavelength-

dispersive spectrometry spot composition measurements (Shugar not dated a; Giumlia-Mair

et al.

2002).


583


Archaeometry

51

, 4 (2009) 576–597

VG (now Thermo) Axiom MC-ICP-MS. In both cases, the samples were introduced into theinstrument via an ESI 50

µ

l/min PFA micro-concentric nebulizer attached to a desolvating unit(Nu Instruments DSN-100 or Cetac Aridus, depending upon the mass spectrometer used).Prior to analysis, each sample was doped with a Tl solution, added to allow for the correctionof instrument-induced mass bias. For each sample, five ratios were measured simultaneously,three of which are presented in Table 1 (

206

Pb/

204

Pb,

207

Pb/

206

Pb and

208

Pb/

206

Pb). The precision and accuracy of the method was assessed through repeat analysis of a NBS

981 Pb standard solution (spiked with Tl). The average values obtained for each of the meas-ured NBS 981 ratios were then compared to the known values for this standard (Thirlwall2002). All sample data were subsequently normalized according to the relative daily deviationof the measured standard value from the true. Normalization to an international standard inthis way effectively cancels out the effects of slight daily variations in instrumental accuracy,and allows the direct comparison of the data obtained during different analytical sessions.The analytical errors reported for each of the sample ratios are also propagated relative to therespective reproducibility of this standard, to take into account the errors associated with thenormalization process.

Data

The normalized and error-propagated sample data are presented in Table 1. The Pb isotope datashow a wide isotopic spread, more than 5% on the

207

Pb/

206

Pb ratio. This would normally beregarded as a primary indicator of the use of metal from a number of discrete ore sources(Begemann

et al

. 1989, 273–5; Pernicka

et al

. 1993, 29; Stos-Gale and Gale 1994, 100),although individual ore sources displaying this much isotopic heterogeneity are known (e.g.,Faure 1977, Ch. 14).

The 17 analyses represent 15 artefacts in total, with two artefacts (zp97.201, zp97.216) eachhaving a pair of drilled samples analysed (referred to as ‘a’ and ‘b’). As discussed below, theresults from these data pairs are significant for understanding the possible effects of corrosionon isotopic composition. Subsequent sections focus upon isotopic variation by archaeologicalcontext and composition, and the implications of the lead isotope data for the absolute prove-nance of the al-Midamman artefacts.

DISCUSSION

Variation in isotopic ratios with corrosion (Figs 4 and 5)

The paired analyses of samples zp97.201 and zp97.216 consist of a metallic and a corrodedsample from each artefact, and in each case the isotopic data for the analytical pairs are divergentby more than the 2-sigma statistical error associated with individual measurements. Thedifferences between the paired analyses for zp97.201 are relatively small (the data points overlapwhen illustrated at the scale of Fig. 4) and do not affect the archaeological interpretation of thedata. In contrast, there is a larger divergence between the ratios measured in the two drilledsamples from zp97.216, equivalent to more than 1% on the

207

Pb/

206

Pb ratios.Although it is possible that the divergent data points for the paired samples reflect the internal

isotopic heterogeneity of these artefacts, previous Pb isotope analyses of archaeological arte-facts using laser ablation (e.g., Ponting

et al

. 2003) suggest that this is unlikely. Rather, thedivergent data pairs from zp97.201 and zp97.216 indicate that these samples were most likely

584

L. Weeks

et al.


Archaeometry

51

, 4 (2009) 576–597

contaminated with extraneous Pb from the burial environment during corrosion. The artefactsin question are particularly prone to such contamination, as their bulk Pb concentrations are verylow (Table 1), below the detection limit of ICP-AES analysis in one case. Other corroded sampleswith possibly low Pb concentrations include zp97.67 and zp97.71, although the SEM/EDS andEPMA/WDS analyses of these samples are less sensitive than ICP-AES and more prone to prob-lems associated with the insolubility and resulting segregation of Pb in copper. The pairedanalyses indicate that very low Pb corroded samples can be affected by contamination from theburial environment, although this effect can vary from minimal (as indicated by the data forzp97.201) to substantial. The near-complete corrosion of copper-base artefacts is common inArabian burial environments, and it is therefore useful to have an understanding of the potentialfor lead isotope ratios to be affected by post-depositional processes.

Figure 4 Lead isotope ratios of the al-Midamman artefacts, by alloy category. Data points with grey fill represent corroded samples.

Lead isotope analysis of copper-base artefacts from al-Midamman 585

© University of Oxford, 2008, Archaeometry 51, 4 (2009) 576–597

Most remaining samples from al-Midamman, both metallic and corroded, are considerablyhigher in Pb, with compositional analyses using ICP-AES and semi-quantitative SEM/EPMAindicating bulk Pb concentrations from c. 0.1–2.2% Pb (Table 1). Artefacts with such high Pbconcentrations are very unlikely to have been affected by contamination with extraneous Pbfrom the burial environment, to the extent that the latter will affect the isotope ratios.

While the following discussion of the al-Midamman artefacts emphasizes the data frommetallic samples, the data from corroded samples are also retained. This decision is justified notonly by consideration of the absolute Pb concentrations in the samples (and hence their lowlikelihood of contamination), but also by the broad similarities in the isotopic characteristicsof the corroded and uncorroded groups (Fig. 4). Not only do corroded and metallic examplesfrom the different alloy groups show comparable isotopic ratios, but also the general conclu-sions that can be drawn regarding the absolute provenance of the al-Midamman artefacts areexactly the same for the sub-set of 11 corroded samples as they are for the sub-set of sixmetallic samples. Finally, it is worth noting that the samples analysed in this study are the bestthat could be obtained without substantial destruction of such small and highly corroded arte-facts, numerous examples of which are of museum display quality.

Variation in isotopic ratios by alloy type (Figs 4 and 5)

Figure 4 shows the results of the LIA of the al-Midamman artefacts by their compositionalgroupings (copper, arsenical copper and tin-bronze, as discussed above). Artefacts of unalloyedcopper (n = 7) show a large diversity in their isotope ratios, falling into three rough ‘clusters’—onecluster of two artefacts (zp97.67, zp97.156) centred at 207Pb/206Pb ≈ 0.84; a second cluster ofthree artefacts (zp97.71, zp97.219, zp97.224) centred at 207Pb/206Pb ≈ 0.872; and a third clusterof two artefacts (zp97.237, zp97.244) centred at 207Pb/206Pb ≈ 0.883. The second and third ofthese clusters correspond to the main isotope groups observed in the artefacts from the stand-ing stones cache (see below). Similarly, artefacts of arsenical copper also show a wide range

Figure 5 Lead isotope ratios of a subset of the al-Midamman artefacts, by alloy category. Data points with grey fill represent corroded samples.

586 L. Weeks et al.


of isotope ratios (207Pb/206Pb ratios c. 0.859–0.883). Three of the analysed arsenical coppersamples (zp97.201a, zp97.201b, zp97.236) plot in the same general Pb isotope areas as unalloyedcopper and tin-bronze samples from the cache. Unlike the other alloy types, the tin-bronzes (n = 5)are extraordinarily homogeneous in their Pb isotope composition. Data points for the tin-bronzesare indistinguishable on Figure 4, and an expanded view of the data from the cache artefacts(Fig. 5) shows that all the tin-bronzes have isotopic ratios falling within analytical error of oneanother. Such isotopic homogeneity almost certainly indicates a shared source for the metal inthese artefacts, and might even be indicative of their origin in the product of one individualsmelting or alloying operation.

Overall, the Pb isotope data suggest that there is no strict isotopic differentiation betweendifferent alloy types. Interpreting such isotope patterns is difficult, due to lacunae in ourknowledge of the production and exchange systems underlying the al-Midamman copper-baseartefacts. We lack knowledge of how the alloys were produced, how much Pb was in individualalloying components and how much these concentrations varied, as well as how much isotopicvariation existed in the copper, tin or arsenic ores from possible sources. Such considerationsdemonstrate the interplay of metallurgical technology and ore source geology and isotopy,not to mention specific cultural and economic factors raised by the archaeological recordof al-Midamman, in the interpretation of archaeological LIA.

For example, if the al-Midamman alloys were produced by intentionally adding tin and/orarsenic (metal or ores) to raw metallic copper, then the alloying process would have led to amixing of Pb from the different constituents and thus to the production of an artefact with anisotopic signature intermediate between that of the alloying components. Within such a model,the linear array in the Pb isotope data for arsenical copper samples from al-Midamman mightbe regarded as a mixing line between a conformable copper deposit with Pb isotope characteristicsplacing it at or beyond one end of the linear distribution, and a conformable source of As orhigh-As copper with Pb isotope ratios at or beyond the other end of the distribution.

However, while such arguments are valid in theory, specific geological factors in SW Arabiaand general considerations of alloy production techniques provide an alternative perspective.First, many copper ore deposits of western Arabia (see below) are very old—Precambrian—and whole rock/ore samples show non-conformable Pb isotope signatures, reflecting in partthe effect of high U/Pb and Th/Pb ratios and in situ production of Pb since the time of initialore formation (Stacey et al. 1980; Ellam et al. 1990). The long linear arrays in the Pb isotopedata for specific alloys from al-Midamman could simply reflect local isotopic diversity ratherthan mixing lines between conformable isotopic end members. The highly variable isotopicratios of the ‘pure’ copper artefacts from al-Midamman are possibly an example of this effect.

The explanation of Pb isotope diversity within and between alloy groups is simplified by thesuggestion that the Pb isotope ratios of the tin-bronzes or arsenical copper artefacts are virtuallyidentical to the Pb isotope ratios of the copper they contain. For example, Pb levels in tin oresand metal are usually very low in comparison to Pb levels in copper and copper ores (Gale andStos-Gale 1982, 13; Muhly 1985, 80; Maddin 1989, 102; Pernicka et al. 1990; Pernicka 1995,106; Begemann et al. 2001, 66–8). Moreover, the al-Midamman arsenical copper most likelyrepresents the product of a mixed furnace charge of Cu-bearing and As-bearing ores (see Weeks2003, 113–21 for a general discussion of this issue), all the raw materials for which couldhave derived from the same mine source with consequently similar isotopic ratios (although,for complexities in this assumption, see Ixer 1999).

If such assumptions are valid for the al-Midamman copper-base artefacts, the isotopic clustersmay incorporate various alloy types because these different alloys were produced from the same



copper source. However, given our limited understanding of the manufacturing techniques of theal-Midamman artefacts and the general lack of information regarding primary metal extractionin Bronze Age southwestern Arabia, further laboratory and field work will be required beforereliable conclusions can be generated regarding the alloy/isotope patterns seen at al-Midamman.

Variation in isotopic ratios by archaeological context (Fig. 6)

There is a significant relationship between the archaeological context of artefacts at al-Midammanand their Pb isotopic composition. As can be seen in Figure 6, the nine artefacts analysed fromthe cache have isotopic ratios falling into two relatively tight clusters. Two additional samplesfrom artefacts found buried under other parts of the standing stones or immediately near them(zp97.219, zp97.201a) have similar Pb isotope ratios that fall into the more radiogenic (i.e.,lower 207Pb/206Pb and 208Pb/206Pb ratios) of the two cache isotope clusters.

In contrast, while one of the four surface artefacts (zp97.71) is isotopically similar to material fromthe cache, the remainder have significantly more radiogenic Pb isotope ratios. As there are multiplephases of human activity at the site, it may well be that the metal artefacts from surface contexts,distant from the standing stone monument itself, are chronologically distinct from the metal arte-facts buried as part of the cache. Certainly, the Pb isotope data do not contradict such a hypothesis.

The LIA results are a useful addition to the evidence for changing cultural expressions at al-Midamman over the period of occupation. The data indicate that, in addition to the changes inarchitectural styles and the expression of ‘monumentality’ seen at the site, changes were mostlikely taking place in the technological and/or economic spheres. That is, the sources of themetal artefacts found at the site appear to have changed substantially from the period of theproduction of the cache artefacts to the period of widespread domestic settlement acrossthe site in the 13th to ninth centuries bc. These changes may be conceived of as representinga re-orientation of exchange systems, and/or developments in the technology of metal miningand smelting in the region.

Figure 6 Lead isotope ratios of the al-Midamman artefacts, by archaeological context.

588 L. Weeks et al.


Lead isotope ratios and absolute provenance

In this section, a series of figures displays the results of the al-Midamman LIA in relation to(a) the archaeological contexts of the analysed artefacts, and (b) Pb isotope data for ore bodiesand artefact assemblages from southwestern Arabia and neighbouring regions. Contextualconsiderations have been somewhat simplified by dividing the al-Midamman artefacts into twogroups: those that come from the artefact cache and adjacent contexts under the standing stonemonument (labelled ‘Al-Midamman Stone Cache’) and those that were recovered from surfacecontexts away from the standing stones and adjacent to areas of settlement (labelled ‘Al-MidammanSettlement’). Only two measured Pb isotope ratios are illustrated in most of the following fig-ures, as this is sufficient to demonstrate isotopic divergence.

This component of our research remains preliminary, largely due to considerations of theextent and quality of background isotopic data from artefacts and ore bodies in south-west Asia.For some regions, such as Oman, Jordan and Israel, the relevant comparative data are fairly robustand unlikely to change dramatically with further archaeological fieldwork and laboratory analyses.Other regions, however, such as Saudi Arabia, Yemen and Egypt, have been the focus of muchless research. Pb isotope data for these regions tends to come overwhelmingly from geologicalstudies, and there is a real possibility that future programmes of archaeological research willradically alter our understanding of the Pb isotope profiles of ore sources and metallic productsfrom these regions.

Eastern Arabia (Figs 7 and 8) There is archaeologically well-documented, if sporadic,large-scale primary production of copper in Oman and the United Arab Emirates (UAE) fromat least the second half of the third millennium bc through to the Islamic period (Hauptmann

Figure 7 Lead isotope ratios of the al-Midamman artefacts, by archaeological context, in comparison to copper ores from Oman and Bronze Age copper ingots from Oman and the Persian Gulf region (data from Chen and Pallister 1981; Calvez and Lescuyer 1991; Stos-Gale et al. 1997; Prange 2001; Weeks and Collerson 2003). Absolute 2-sigma errors are smaller than the symbol size.



1985; Prange 2001; Weeks 2003, Ch. 2). This area might therefore be considered as one of theregional possibilities to have supplied metal to al-Midamman. Isotopic data from copper orebodies in Oman are illustrated in Figure 7, along with the isotopic data from al-Midamman.Only one of the al-Midamman samples shows isotopic ratios on the edges of the known Omanicopper ore Pb isotope compositions: the arsenical copper artefact zp97.216b. However, BronzeAge copper ingots from Oman (as well as the UAE and Bahrain) have somewhat unusual Pbisotope signatures that do not consistently overlap with known Omani copper ore sources(Prange 2001; Weeks and Collerson 2003, 2005). Figure 7 shows that some of the Omaniingots are broadly isotopically similar to the al-Midamman surface artefact zp97.156.

Furthermore, there is not a complete overlap between the isotopic composition of copper-baseartefacts found in prehistoric Oman and the UAE and the local ore sources, so a comparisonbetween the al-Midamman artefacts and contemporary finished artefacts from southeasternArabia is necessary. As illustrated in Figure 8, the al-Midamman artefacts are completely iso-topically distinct from contemporary (i.e., Umm an-Nar Period, Wadi Suq Period, Late BronzeAge and Iron Age) southeastern Arabian finished artefacts, despite the similar range of alloysbeing utilized in the two regions (Weeks 1997, 1999, 2003; Prange 2001; Weeks and Collerson2003). The overriding impression gained from Figures 7 and 8 is of a general Pb isotopic dis-similarity between the al-Midamman artefacts and the copper-base metal circulating in BronzeAge/Iron Age eastern Arabia (i.e., within the Persian Gulf exchange system).

Northwestern Arabia (Fig. 9) A very long history of copper production has been documentedin the Jordan rift valley of northwestern Arabia, from the fourth millennium bc onwards. Themost famous sites are Feinan (Jordan) and Timna (Israel), although production outside of thesespecific locations also took place (Rothenberg 1972; Hauptmann 2000; Weisgerber 2006).Figure 9 indicates that there is very little isotopic overlap between copper-base artefacts fromal-Midamman and ores and raw copper from Timna, Feinan or neighbouring deposits. With the

Figure 8 Lead isotope ratios of the al-Midamman artefacts, by archaeological context, in comparison to Bronze Age and Iron Age copper-base artefacts from Oman and the UAE (data from Weeks 1999; Prange 2001; Weeks and Collerson 2003). Absolute 2-sigma errors are smaller than the symbol size.

590 L. Weeks et al.


exception of one sample (zp97.216b), the al-Midamman samples are depleted in their thorogenicPb isotope component in comparison to the northwestern Arabian production sites. The differencebetween the Pb isotope composition of the al-Midamman artefacts and Middle Bronze Age artefactsfrom the Levant is further emphasized by the inclusion in Figure 9 of LIA data for copper-basealloys from the Middle Bronze Age site of Ein Zig in the Negev (Segal et al. 1999, fig. 6).

The results of this comparison suggest that Levantine copper sources did not supply themetal used to manufacture the al-Midamman artefacts. This conclusion is important, given thetypological comparisons that have been noted between the al-Midamman cache and LevantineEarly/Middle Bronze Age metal artefacts (Giumlia-Mair et al. 2002, 196).

Ores from the Arabian Shield (Fig. 10) The geological unit known as the Arabian–NubianShield characterizes western Saudi Arabia and parts of coastal Yemen, as well as adjacent areasof northeastern Africa across the Red Sea, from eastern Egypt to Ethiopia (Stoeser and Frost2006). As illustrated in Figure 10, there is great similarity between the Pb isotope compositionof artefacts from al-Midamman and samples derived from base metal deposits in western SaudiArabia (collected largely in the course of geological research – see Stacey et al. 1980; Bokhariand Kramers 1982; Ellam et al. 1990). In particular, the two major isotopic clusters of artefactsfrom the cache find close Pb isotope parallels in the analysed galenas from Saudi Arabian basemetal vein and massive sulphide deposits. Moreover, the outlying al-Midamman copper artefactswith low 207Pb/206Pb ratios (zp97.156 and zp97.67) can be paralleled in a general way with some ofthe more radiogenic whole-rock and whole-ore samples from Saudi Arabian base metal deposits.

All of the artefacts from al-Midamman have isotopic matches or near-matches with basemetal deposits from the Arabian Shield, with the exception of the surface sample zp97.216b. Asdemonstrated above, this sample shows tenuous isotopic parallels with southeastern Arabianand/or northwestern Arabian (Levantine) copper sources.

Figure 9 Lead isotope ratios of the al-Midamman artefacts, by archaeological context, in comparison to copper ores and raw copper from Feinan (Jordan), Timna (Israel), and nearby sites (data from Gale et al. 1991; Hauptmann et al. 1992). Absolute 2-sigma errors are smaller than the symbol size.



Northeastern Africa (Fig. 11) The isotopic similarity between Saudi Arabian ores and theal-Midamman artefacts raises the possibility of a local (i.e., south-west Arabian) origin for themetal used at al-Midamman, particularly as such thorogenically depleted isotopic signaturesare relatively rare in south-west Asia and Europe. However, as noted above, the geology ofnortheastern Africa (the Nubian Shield) is very similar to that of western Saudi Arabia andYemen (the Arabian Shield), and there are numerous base metal deposits in northeasternAfrica, some of which have been exploited since early times (Shortland 2006). As Pb isotopesare primarily determined by the age and geology of the ore body, the similar underlying geology

Figure 10 Lead isotope ratios of the al-Midamman artefacts, by archaeological context, in comparison to base metal vein and massive sulphide deposits in Saudi Arabia (whole-rock data from Ellam et al. 1990; whole-ore data from Doe and Rohrbough 1977; Stacey et al. 1980; Bokhari and Kramers 1982; galena data from Stacey et al. 1980). Absolute 2-sigma errors are smaller than the symbol size.

592 L. Weeks et al.


of the ore bodies in northeastern Africa and western Arabia means that they can be difficult todistinguish isotopically.

The Pb isotope data for ores from Egypt (predominantly Pb ores) collected as the result ofgeological prospection and research are shown in Figure 11. While most of the Egyptian Pbores have 207Pb/206Pb ratios of less than 0.84, there are a few examples that have similar Pbisotope ratios to artefacts from al-Midamman, including those from the cache. Although it ispreferable to isotopically compare copper artefacts and copper ores (rather than Pb ores as usedhere), it is known that copper deposits do occur in the Eastern Desert of Egypt in a geological settingsimilar to those of the Pb ore deposits (Garenne-Marot 1984; Hussein and El Sharkawi 1990),and that some show traces of ‘ancient’ exploitation (Garenne-Marot 1984, 99). This raises thepossibility that copper mined in the Eastern Desert of Egypt may have had Pb isotope characteristicssimilar to the Pb ores from the region. Moreover, the trade of Egyptian copper-based artefactsto the land of Punt in the second millennium bc (Dixon 2004) may provide further supportfor the possibility of Egyptian metal reaching the southern Red Sea region at that time.

However, there are a number of reasons to question the possibility that the al-Midammanartefacts were made of metal obtained from Egypt, particularly Lower Egypt. In the NewKingdom (1550 to 1069 bc), both texts and tomb paintings attest to the exploitation in LowerEgypt of ‘Asiatic’ copper and copper from Alashiya (Cyprus) (Muhly 1973; Garenne-Marot1984, 103). In the Ramesside period (13th/12th centuries bc) there was, in addition, intensiveEgyptian exploitation of the Timna mines (Weisgerber 2006, 13), copper from which does notcorrespond isotopically with the al-Midamman samples (as demonstrated above). In addition,Pb isotope data discussed by Stos-Gale et al. (1995) and Shortland (2006) undermine thehypothesis of a Lower Egyptian provenance for the metals from al-Midamman. Lead isotopeanalyses of copper-base artefacts (chiefly tin-bronzes) from the XVIIIth Dynasty site ofAmarna (Stos-Gale et al. 1995, table 2) do not overlap with the data from al-Midamman.Moreover, the Pb isotope data for lead ores and lead and copper artefacts presented by Short-land (2006, fig. 1, table 2) suggest that the extensive base-metal ore deposits in the Eastern

Figure 11 Lead isotope ratios of the al-Midamman artefacts, by archaeological context, in comparison to Pb ores, Pb artefacts and copper-base artefacts from ancient Egypt (data from Doe and Rohrbough 1977; Stos-Gale and Gale 1981; Stos-Gale et al. 1995, table 2; Shortland 2006, table 1). Absolute 2-sigma errors are smaller than the symbol size.



Desert of Egypt were not used to produce even metallic lead (which appears, like copper, tohave been imported from the Mediterranean region; Shortland 2006, 666). Finally, of the copperdeposits in the Eastern Desert discussed by Garenne-Marot, it appears that one, Umm Samiuki(Garenne-Marot 1984, fig. 2), has isotopic characteristics substantially different from those ofthe al-Midamman artefacts (Stos-Gale et al. 1995, 132). Of course, there are also the largecopper deposits of the Sinai Peninsula to consider. These sources were definitely exploited forturquoise in the New Kingdom, but debate continues regarding the dating of the evidence forcopper smelting in the region (Garenne-Marot 1984, 97–9). Unfortunately, Pb isotope data areas yet unavailable from these potential sources.

If sources on the west side of the Red Sea were important to the metallurgy of Bronze AgeYemen, then regions further to the south, in modern day Sudan or Ethiopia, may have beenmore important than those in the north. For example, copper ores, copper smelting furnacesand moulds datable to the Old Kingdom (Dynasties IV–V) have been recorded near Buhen innorthern Sudan (Garenne-Marot 1984). Furthermore, as the Nubian component of the Arabian–Nubian Shield continues well into the area of modern-day Ethiopia (Stacey and Hedge 1984,fig. 1), base-metal deposits geologically similar to those in western Arabia might well haveexisted that far south in the western Red Sea coastal region.

In this context, it is worth noting that the strong connections in later Bronze Age materialculture between the adjacent African and Arabian coasts of the southern Red Sea have ledsome scholars to suggest that communities in these regions were part of the same culturalgroup (Fattovich 1978; Munro-Hay 1993; Keall 2004). In fact, the general ‘connectedness’ ofcommunities on the African and Arabian shores of the southern Red Sea, from early prehistoryto the present day, is an issue that emerged clearly in the proceedings of the first of the recentRed Sea Project conferences (Lunde and Porter 2004; see also Edens and Wilkinson 1998,102–5). After the initial compositional analyses of the artefacts from the al-Midamman cache,Giumlia-Mair et al. (2002, 207) tentatively suggested that:

The different metals used on the western and the eastern coast of the Arabian Peninsulacould reflect the existence of two different spheres of influence: one encompassing bothcoasts of the Red Sea and the second the countries around the Persian Gulf.

In accordance with this hypothesis, the isotopic data in the present study highlight the possibilityof production of the al-Midamman artefacts within southwestern Arabia or adjacent regions ofAfrica, and the clear isotopic divergence between southwestern and southeastern Arabian oresand artefacts further supports the existence of a distinct regional metal production and exchangesystem centred upon the southern coasts and hinterlands of the Red Sea. In fact, Edens andWilkinson (1998, 105) have noted that there is a relative abundance of metal artefacts at laterBronze Age coastal sites in southwestern Arabia in comparison to contemporary inland sites,and they suggest that this may reflect the greater participation of coastal groups and individuals inthe maritime exchange relations linking the region with northeastern Africa.

SUMMARY AND CONCLUSIONS

The Pb isotope analyses show a number of interesting patterns when information on archaeo-logical context and alloy composition is considered. First, the al-Midamman artefacts that canbe securely associated with the cache under the standing stone monument (Monumental Phase 1)have relatively high 207Pb/206Pb ratios in excess of 0.87, whereas surface finds from area HWB(Monumental Phase 2 or later) tend towards more radiogenic 207Pb/206Pb ratios of 0.839–0.871.

594 L. Weeks et al.


Thus, the Pb isotope analyses do not suggest a strong connection between surface materials and thedeposits of the standing stone monument. A second significant observation is that there are nostrict differences in Pb isotope composition between the various copper-base alloy types usedat al-Midamman. Clusters of artefacts characterized by similar Pb isotope ratios can incorporatealloys of copper, arsenical copper and tin-bronze.

The LIA has proven most important in the initial consideration of the absolute provenanceof the metal used to produce the al-Midamman artefacts. A number of the negative provenanceconclusions are clear, in particular the demonstration that the neighbouring mining regions ofsoutheastern Arabia and the Levant did not supply the metal (or artefacts) used at al-Midamman.One fundamental isotopic characteristic distinguishes the al-Midamman artefacts from metalsand ores used in neighbouring regions of Arabia and the Middle East, i.e., a relative depletionof their thorogenic Pb component.

This discovery raises the possibility that an indigenous copper production and exchangesystem exploiting local copper sources existed in the southern Red Sea region by the BronzeAge, a hypothesis further supported by the isotopic similarities between al-Midamman artefactsand base metal deposits from the Arabian–Nubian Shield in southwestern Arabia and northeasternAfrica. The possibility of ‘local’ primary copper extraction in South Arabia is supported bymore than just the occurrence of copper-base artefacts at al-Midamman and the results of theirPb isotope analysis. The al-Midamman assemblage represents only a small part of the growingbody of copper-base artefacts from Yemen dating to as early as the fourth millennium bc (e.g.,Crassard and Hitgen 2007; Steimer-Herbert et al. 2007). Moreover, despite little archaeologicalresearch on this issue, there is some evidence for the pre-modern exploitation of copper depositsin parts of western Saudi Arabia and Yemen to add to the solid evidence for the local smeltingof copper and tin ores in the first millennium bc at Hajar Ar-Rayhani in Yemen (Fleming andPigott 1987). One report, for example, describes the copper deposit at Kutam, in SW SaudiArabia near the Yemeni border, as marked at the surface by ancient trenches and pits in a zone500 m long and 100 m wide. Slag piles, estimated to contain about 50 000 t of material, lie onthe southwest flank of the ridge and in the adjacent valley (Saudi Geological Survey 2007).Although most documented copper mining and smelting sites in Saudi Arabia appear to beIslamic in date (de Jesus et al. 1982; Peli 2006), there is possible evidence for their earlierexploitation (Hester et al. 1984). The existence of a substantial prehistoric indigenous metalextraction and exchange system in southern Arabia would perhaps come as no surprise tothose familiar with the numerous copper-base artefacts characteristic of the region in theclassic South Arabian period of the mid-first millennium bc (e.g., Glanzman 2002; cf.Keall 2004, 53).

When considering this possibility, however, a number of factors outlined above suggest thatour positive provenance assignations from the al-Midamman LIA are tentative at best. Thissituation is typical of archaeological LIA, reflecting both logical limitations of the techniqueand the imperfect background Pb isotope database. In the case of al-Midamman, the geologicalsimilarities between western Arabia and northeastern Africa, including the Sinai Peninsula,mean that the ultimate delineation of the provenance of the artefacts will be difficult, requiringnot only more Pb isotope analyses of local ores and artefacts but also field research to documentloci of metal extraction in these regions. Although our preliminary analyses support thepossibility of an indigenous metal extraction and circulation system in the southern Red Searegion, the potential of LIA for archaeological provenance studies will only be fully realizedwithin a coherent research framework combining intensive archaeometallurgical fieldwork andlaboratory analyses. For ancient South Arabia, such work remains a project for the future.



ACKNOWLEDGEMENTS

The authors are indebted to a number of people who provided helpful commentary on theanalyses and our interpretation, including Dr Aaron Shugar (Buffalo State University), ProfessorJulian Henderson (University of Nottingham), Dr Peter Magee (Bryn Mawr College),Dr Ronald Farquhar, Mr Ross Thomas (University of Southampton), and two anonymousreferees. Any remaining errors or omissions are solely the responsibility of the authors.Analytical costs were supported by grants from the University of Nottingham and theRoyal Ontario Museum.

REFERENCES

Arbach, M., and Audouin, R., 2004, Archaeological discoveries in the Jawf. Franco-Yemen rescue operation of thesite of as-Sawda’ (ancient Nashshan), Centre francais d’archeologie et de sciences sociales, Sanaa.

Bayle des Hermens, R. de, 1976, Premiere mission de recherches prehistoriques en Republique arabe du Yemen,L’Anthropologie, 80, 5 –37.

Begemann, F., Schmitt-Strecker, S., and Pernicka, E., 1989, Isotopic composition of Pb in early metal artefacts:results, possibilities, limitations, in Old World archaeometallurgy (eds. A. Hauptmann, E. Pernicka and G. A.Wagner), 269 –78. Der Anschnitt Beiheft 7, Bochum.

Begemann, F., Schmitt-Strecker, S., Pernicka, E., and Lo Schiavo, F., 2001, Chemical composition and lead isotopyof copper and bronze from Nuragic Sardinia, European Journal of Archaeology, 4, 43 – 85.

Bokhari, F. Y., and Kramers, J. D., 1982, Lead isotope data from massive sulfide deposits in the Saudi Arabian Shield,Economic Geology, 77, 1766 – 9.

Bronk Ramsey, C., 1995, Radiocarbon calibration and analysis of stratigraphy: the OxCal program, Radiocarbon, 37,425 – 30.

Bronk Ramsey, C., 2001, Development of the radiocarbon program OxCal, Radiocarbon, 43, 355 – 63.Buffa, V., and Vogt, B., 2001, Sabir – cultural identity between Saba and Africa, in Migration und Kulturtransfer:

Der Wandel vorder- und zentralasiatischer Kulturen im Umbruch vom 2. Zum 1. Vorchristlichen Jahrtausend (eds.R. Eichmann and H. Parzinger), Kolloquien zur Vor- und Fruhgeschichte 6, 437–50, (Akten des InternationalenKolloquiums, Berlin November 1999), Habelt, Bonn.

Calvez, J. Y., and Lescuyer, J. L., 1991, Lead isotope geochemistry of various sulphide deposits from the Oman mountains,in Ophiolite genesis and the evolution of the oceanic lithosphere (eds. Tj. Peters, A. Nicolas and R. G. Coleman), 385–97,Ministry of Petroleum and Minerals, Sultanate of Oman.

Chen, J. H., and Pallister, J. S., 1981, Lead isotopic studies of the Samail Ophiolite, Oman, Journal of GeophysicalResearch, 86(B4), 2699–708.

Crassard, R., and Hitgen, H., 2007, From Uafir to Balhaf – Preventive archaeological survey and rescue excavationsalong the Yemen LNG pipeline route, Proceedings of the Seminar for Arabian Studies, 37, 43–59.

de Jesus, P., Al-Sugiran, S., Rihani, B., Kesnawi, A., Toplyn, M., and Incagnoli, J., 1982, Preliminary report of theancient mining survey 1981 (1401), Atlal, 6, 63–79.

Dixon, D. M., 2004, Pharaonic Egypt and the Red Sea arms trade, in Trade and travel in the Red Sea region: proceed-ings of the Red Sea Project 1 (eds. P. Lunde and A. Porter), 33–42. BAR International Series 1269, Society forArabian Studies Monographs No. 2, Archaeopress, Oxford.

Doe, B. R., and Rohrbough, R., 1977, Lead isotope data bank: 3458 samples and analyses cited. United StatesDepartment of the Interior Geological Survey Open Report 79-661.

Edens, C., and Wilkinson, T. J., 1998, Southwest Arabia during the Holocene: recent archaeological developments,Journal of World Prehistory, 12, 55–119.

Ellam, R. M., Hawkesworth, C. J., and McDermott, F., 1990, Pb isotope data from late Proterozoic subduction-relatedrocks: implications for crust-mantle evolution, Chemical Geology, 83, 165–81.

Fattovich, R., 1978, L’Etiopia ed i regni sudarabici, in Archeologia: culture e civiltà del passato nel mondo europeoed extraeuropeo (ed. L. Fasani), 350–9, Mondadori, Milan.

Faure, G., 1977, Principles of isotope geology, John Wiley and Sons, New York.Fleming, S. J., and Pigott, V. C., 1987, Archaeometallurgy, in Site reconnaissance in the Yemen Arab Republic, 1984:

The stratigraphic probe at Hajar Ar-Rayhani (eds. W. D. Glanzman and A. O. Ghaleb), 171–81, The Wadi Al-Jubah Archaeological Project Volume 3, American Foundation for the Study of Man, Washington, DC.

596 L. Weeks et al.


Gale, N. H., and Stos-Gale, Z. A., 1982, Bronze Age copper sources in the Mediterranean: a new approach, Science,216(4541), 11–19.

Gale, N. H., Bachmann, H. G., Rothenberg, B., Stos-Gale, Z. A., and Tylecote, R. F., 1991, The adventitious produc-tion of iron in the smelting of copper, in The ancient metallurgy of copper: archaeology–experiment–theory (ed.B. Rothenberg), 182–91, Metal in History: 3, Institute for Archaeo-Metallurgical Studies, London.

Garenne-Marot, L., 1984, Le cuivre en Égypte Pharaonique: Sources et métallurgie, Paléorient, 10, 97–126.Giumlia-Mair, A., Keall, E. J., Shugar, A. N., and Stock, S., 2002, Investigation of a copper-based hoard from the

Megalithic site of al-Midamman, Yemen: an interdisciplinary approach, Journal of Archaeological Science, 29,195–209.

Giumlia-Mair, A., Keall, E., Stock, S., and Shugar, A. N., 2000, Copper-based implements of a newly identifiedculture in Yemen, Journal of Cultural Heritage, 1, 37– 43.

Glanzman, W. D., 2002, Arts, crafts and industries, in Queen of Sheba. Treasures from ancient Yemen (ed. St. J. Simpson),110–16, British Museum, London.

Hauptmann, A., 1985, 5000 Jahre Kupfer in Oman, Band 1: Die Entwicklung der Kupfermetallurgie vom 3.Jahrtausend bis zur Neuzeit, Der Anschnitt Beiheft 4, German Mining Museum, Bochum.

Hauptmann, A., 2000, Zur frühen Metallurgie des Kupfers in Fenan/Jordanien, Der Anschnitt Beiheft 11, GermanMining Museum, Bochum.

Hauptmann, A., Begemann, F., Heitkemper, E., Pernicka, E., and Schmitt-Strecker, S., 1992, Early copper producedat Feinan, Wadi Arabah, Jordan: the composition of ores and copper, Archaeomaterials, 6, 1–33.

Hester, J., Hamilton, R., Rahbini, A., Eskoubi, K. M., and Khan, M., 1983, Preliminary report on the third phase ofancient mining survey, southwestern province, 1403 A.H./1983, Atlal, 8, 115–141.

Hussein, A. A. A., and El Sharkawi, M. A., 1990, Mineral deposits, in The geology of Egypt (ed. R. Said), 511–66.A. A. Balkema, Rotterdam.

Ixer, R. A., 1999, The role of ore geology and ores in the archaeological provenancing of metals, in Metals in anti-quity (eds. S. M. M. Young, A. M. Pollard, P. Budd, and R. A. Ixer), 43–52, BAR International Series 792,Archaeopress, Oxford.

Keall, E. J., 1998, Encountering megaliths on the Tihamah coastal plain of Yemen, Proceedings of the Seminar forArabian Studies, 28, 139–47.

Keall, E. J., 2004, Possible connections in antiquity between the Red Sea coast of Yemen and the Horn of Africa, inTrade and travel in the Red Sea Region: proceedings of the Red Sea project 1 (eds. P. Lunde and A. Porter), 43–56, BAR International Series 1269, Society for Arabian Studies Monographs No. 2, Archaeopress, Oxford.

Keall, E. J., 2005, Placing al-Midamman in time. The work of the Canadian Archaeological Mission on the TihamaCoast, from the Neolithic to the Bronze Age, Archaologische Berichte aus dem Yemen, 10, 87–100.

Khalidi, L., 2006, Settlement, culture-contact and interaction along the Red Sea coastal plain, Yemen. The Tihamahcultural landscape in the late prehistoric period 3000–900 BC, Ph.D. dissertation, University of Cambridge.

Lunde, P., and Porter, A., 2004, Trade and travel in the Red Sea region: Proceedings of the Red Sea Project 1, BARInternational Series 1269, Society for Arabian Studies Monographs No. 2, Archaeopress, Oxford.

Maddin, R., 1989, The copper ingots and tin ingots from the Ka6 shipwreck, in Old World archaeometallurgy (eds.A. Hauptmann, E. Pernicka and G. A. Wagner), 99–105, Der Anschnitt Beiheft 7, Bochum.

Maigret, A. de, 1984, Archaeological activities in the Yemen Arab Republic, 1984, East and West, 34, 423–54.Muhly, J. D., 1973, Copper and tin, Transactions, The Connecticut Academy of Arts and Sciences, 43, 155–535.Muhly, J. D., 1985, Lead isotope analysis and the problem of lead in copper, Report of the Department of Antiquities,

Cyprus 1985, 78–82.Munro-Hay, S., 1993, State development and urbanism in northern Ethiopia, in The archaeology of Africa: Food,

metals and towns (eds. T. Shaw, P. Sinclair, B. Andah and A. Okpoko), 609–21, One World Archaeology 20,Routledge, London.

Peli, A., 2006, Les mines de la péninsule Arabique d’après les auteurs arabes (VII e–XII e siècles), ChroniquesYéménites, 13. Available at website http://cy.revues.org/index.html (accessed 31 July 2008).

Pernicka, E., 1995, Gewinnung und Verbreitung der Metalle in prähistorische Zeit, Jahrbuch des Römisch-GermanischenZentralmuseums, Mainz, 37, 21–134.

Pernicka, E., Begemann, F., Schmitt-Strecker, S., and Grimianis, A. P., 1990, On the composition and provenance ofmetal objects from Poliochni on Lemnos, Oxford Journal of Archaeology, 9, 263–98.

Pernicka, E., Begemann, F., Schmitt-Strecker, S., and Wagner, G. A., 1993, Eneolithic and Early Bronze Age copperartefacts from the Balkans and their relation to Serbian copper ores, Praehistorische Zeitschrift, 68, 1–54.

Pollard, A. M., Batt, C. M., Stern, B., and Young, S. M. M., 2006, Analytical chemistry in archaeology, CambridgeUniversity Press, Cambridge.

http://cy.revues.org/index.html



Ponting, M., Evans, J. A., and Pashley, V., 2003, Fingerprinting of Roman mints using laser-ablation MC-ICP-MSlead isotope analysis, Archaeometry, 45, 591–7.

Prange, M. K., 2001, 5000 Jahre Kupfer in Oman. Band II. Vergleichende Untersuchungen zur Charakterisierung desomanischen Kupfers mittels chemischer und isotopischer Analysenmethoden, Metalla (Bochum), 8, 1–126.

Pringle, H., 1998, Yemen’s Stonehenge suggests Bronze Age Red Sea culture, Science, 279, 1452–3.Rothenberg, B., 1972, Timna: valley of the Biblical copper mines, Thames & Hudson, London.Saudi Geological Survey, 2007, Metallic Resources > Copper, available at website http://www.sgs.org.sa/index.cfm?

sec=74&sub=195&sub2=264&page= (accessed 18 December 2007).Segal, I., Halicz, L., and Cohen, R., 1999, A study of ingots and metallurgical remains from ‘Ein Ziq and Be’er

Resisim, in Metals in antiquity (eds. S. M. M. Young, A. M. Pollard, P. Budd and R. A. Ixer), 179–86, BARInternational Series 792, Archaeopress, Oxford.

Shortland, A. J., 2006, Application of lead isotope analysis to a wide range of Late Bronze Age Egyptian materials,Archaeometry, 48, 657–69.

Shugar, A. N., not dated a, Metallurgical investigation of copper artefacts from al-Midamman, Yemen, unpublishedreport submitted to Dr E. J. Keall, Royal Ontario Museum, Canada.

Shugar, A. N., not dated b, Metallurgical investigation of a bronze dagger from Yemen, unpublished report submittedto Ms Susan Stock, Royal Ontario Museum, Canada.

Stacey, J. S., and Hedge, C. E., 1984, Geochronological and isotopic evidence for early Proterozoic crust in the easternArabian Shield, Geology, 12, 310–13.

Stacey, J. S., Doe, B. R., Roberts, R. J., Delevaux, M. H., and Gramlich, J. W., 1980, A lead isotope study of miner-alization in the Saudi Arabian Shield, Contributions to Mineralogy and Petrology, 74, 175–88.

Steimer-Herbert, T., As-Saqqaf, A.-R., Lavigne, O., Sagory, T., Saliege, J.-F., Machkour, M., and Guy, H., 2007,Rites and funerary practices at Rawk during the fourth millennium bc, Proceedings of the Seminar for ArabianStudies, 37, 281–94.

Stoeser, D. B., and Frost, C. D., 2006, Nd, Pb, Sr, and O isotopic characterization of Saudi Arabian Shield terranes,Chemical Geology, 226, 163–88.

Stos-Gale, Z. A., and Gale, N. H., 1981, Sources of galena, lead and silver in Predynastic Egypt, Reviewd’Archéometrie, 5 (Actes du XXème Symposium International d’Archéometrie), 285–95.

Stos-Gale, Z. A., and Gale, N. H., 1994, Metals, in Provenience studies and Bronze Age Cyprus: production, exchangeand politico-economic change (eds. A. B. Knapp and J. F. Cherry), 92–121, Monographs in World ArchaeologyNo. 21, Prehistory Press, Madison, Wisconsin.

Stos-Gale, Z. A., Gale, N. H., and Houghton, J., 1995, The origin of copper metal excavated in El Amarna, in Egypt,the Aegean and the Levant: interconnections in the second millennium BC (eds. W. V. Davies and L. Schofield),127–35, British Museum Press, London.

Stos-Gale, Z. A., Maliotis, G., Gale, N. H., and Annetts, N., 1997, Lead isotope characteristics of the Cyprus copperore deposits applied to provenance studies of copper oxhide ingots, Archaeometry, 39(1), 83–123.

Thirlwall, M. F., 2002, Multicollector ICP-MS analysis of Pb isotopes using a (207)Pb-(204)Pb double spike demon-strates up to 400 ppm/amu systematic errors in Tl-normalization, Chemical Geology, 184(3–4), 255–79.

Vogt, B., 1998, Frühe Kulturen an der Kuste des Roten Meeres und des Golf von Aden, in Jemen, Kunst und Archäologieim Land der Königen von Saba (ed. W. Spiepel), 123–7, Kunsthistorisches Museum, Wien.

Weeks, L. R., 1997, Prehistoric metallurgy at Tell Abraq, United Arab Emirates, Arabian Archaeology and Epigraphy,8, 11–85.

Weeks, L. R., 1999, Lead isotope analyses from Tell Abraq, United Arab Emirates: new data regarding the ‘tin problem’in western Asia, Antiquity, 73/279, 49–64.

Weeks, L. R., 2003, Early metallurgy of the Persian Gulf: technology, trade, and the Bronze Age world, AmericanSchool of Prehistoric Research and Brill Academic Publishers, Boston.

Weeks, L. R., and Collerson, K. D., 2003, Lead isotope data from the Gulf, in Early metallurgy of the Persian Gulf:technology, trade, and the Bronze Age world (ed. L. R. Weeks), Ch. 7, 145–63, American School of PrehistoricResearch and Brill Academic Publishers, Boston.

Weeks, L. R., and Collerson, K. D., 2005, Archaeometallurgical studies, in The Early Dilmun settlement at Saar (eds.R. Killick and J. Moon), Ch. 9, 309–24, London–Bahrain Archaeological Expedition, Institute of Archaeology,UCL, London.

Weisgerber, G., 2006, The mineral wealth of ancient Arabia and its use I: Copper mining and smelting at Feinan andTimna – comparison and evaluation of techniques, production, and strategies, Arabian Archaeology and Epigraphy,17, 1–30.

http://www.sgs.org.sa/index.cfm?sec=74&sub=195&sub2=264&page=

http://www.sgs.org.sa/index.cfm?sec=74&sub=195&sub2=264&page=