3. Results and Discussion

Provenance studies of silver artefacts based on lead isotope analyses rely on the measurements of each artefact's independent and stable isotope ratios 208Pb/206Pb, 207Pb/206Pb and 206Pb/204Pb, and their comparison to ores, metals and production materials characterised by the same measurements. Because the research question here concerns the identification of regions in the west that contributed mineral resources to pre-colonial silver-trade and production, we prioritize comparisons to data from galena and other lead ores. Using artefact data generated by thermal ionization mass spectrometry (TIMS) with a reported overall accuracy of better than ±0.1%, we detail consistencies between artefact and ore ratios where all three are identified at 0.2% or less; those established at 0.1% or less represent effectively identical data and appear in green (Table 2; and see data in Thompson and Skaggs 2013).

Based on their proximity to metalliferous regions, some eventual Phoenician colonies have been of particularly suspected significance to pre-colonial silver-trade. Our selection of comparative ore-data for this article is not comprehensive for the entire Mediterranean, but emphasises key zones in Sardinia that include the Sulcis-Iglesiente district in the south-west, where most of the island's lead-zinc-silver-bearing ores of the Caledonian (510-300 Ma) and primarily Hercynian (360-300 Ma) orogenies occur (Valera and Valera 2005), alongside the early Phoenician colonies of Sulcis, Bithia, Nora and Cagliari. Our comparisons of hacksilber artefacts to ores in the Iberian Peninsula are based more generally on the deposits for which isotopic data are available. The well-known colonial settlements eventually established in the southern Iberian Peninsula include, of course, Cádiz and also a concentration of what are usually characterised as agricultural sites along the Andalusian coast east of Gibraltar to Almería (Markoe 2000); less well known perhaps is the evidence cited for a 7th-century settlement as far north as Alácer do Sal on the Atlantic coast - about 250 miles north of Cádiz (Tavares 1993; Soares and Tavares da Silva 1980).

Bi-plot 1 (Figure 2) and Table 2 document the consistencies established between our artefact samples and lead ores in southern and south-western Sardinia, published galena from Huelva and the Ossa Morena zone in the south-western Iberian Peninsula, and the Los Pedroches-Alcudia Area in south-central Spain, where high Ag values have been related to Pb-Ag-Zn vein mineralisation (De Vos and Tarvainen 2006, 50). The 12 of our 25 artefact samples with lead isotope ratios fully consistent with Sardinian lead ores include two from 11th-century Tell Keisan, one from 11–10th-century Tel Dor, six from 10–9th-century Ein Hofez, and three samples from 10–8th-century Akko. At the same time, and with overlaps readily apparent, particularly in the 208Pb/206Pb ratios of Sardinian and Iberian ores ranging roughly between 2.10 and 2.13, ten samples from Ein Hofez are consistent with ores in the Iberian Peninsula.

Figure 2

Figure 2: Bi-plot 1 of Southern Phoenician hacksilber samples against lead ores from Sardinia and the Iberian Peninsula (24-31). See key for numbered citations.

At a most basic level, these results allow us to establish that lead isotope data from our hacksilber samples correlate with ancient documentary sources that indicate Phoenician pre-colonial silver trade with Sardinia and Spain. From here, however, geological trends in the western Mediterranean, our artefact samples and ancient documents tell us to take a wider view, one that incorporates evidence implicating the western Mediterranean at large as a region supplying silver and lead to the east prior to 800 BC.

Figure 3 Figure 4

Figure 3: Bi-plot 2 of Southern Phoenician hacksilber samples versus selected lead ores from southern France (32-34) showing overlaps with selected Sardinian deposits (Figure 2 sources). See key for numbered citations.
Figure 4: Bi-plot 3: ratios of selected lead ores in Sardinia, the Iberian Peninsula, Greece (Laurion), Central and East Anatolia and southeastern Turkey (Taurus) versus hacksilber data from Tel Miqne-Ekron (Table 1) and the southern Phoenician hoards (24-31, 35- 37, 46, 48 and 50). See key for numbered citations.

It has been possible for some time to document geological trends that broadly distinguish the isotopic characterisation of some key western Mediterranean ores from eastern ones (Knapp and Cherry 1994, 117); this is of critical significance to questions of pre-colonial silver-trade. Bi-plot 3 (Figure 4) shows that the 208Pb/206Pb ratios of galena from the relatively young deposits in Laurion (Greece) and the Taurus range in south-eastern Turkey (less than 100 Ma) plot far enough below the older deposits in Sardinia, Spain and southern France (greater than 100 Ma) to allow generalised distinctions between these western and eastern ores that become apparent around the x/y coordinates 0.84-0.85 and 2.08-2.09. Our research questions prompt the identification of exceptions to this trend within the Mediterranean basin. A few samples from Seeliger's central and eastern Anatolia group (Figure 4), for example, plot around 0.84-0.85 and 2.08-2.09. At the same time, samples from the lead ores of the Betic Cordillera, Almería and Murcia in Spain fall below 0.84/2.09, while Tunisian galena plots above 2.05 and as high as 2.09 (Skaggs et al. 2012), and belongs to the same geological category as our selected eastern Mediterranean ores. That is, because the ratios of 208Pb/206Pb are influenced more by the amount of parent uranium and thorium than the overall age of the deposit, we identify the relevant trends using calculated model age curves based on comparisons of non-radiogenic isotope 204Pb and radiogenic 207Pb and 206Pb (Figure 5). The data points from eastern Mediterranean and Tunisian ores, which plot farther above and to the right of 15.65 and 18.6, are characteristic of relatively young emplacements and derivation from shallower crustal materials, or even subducted sediments. The western ore ratios 206Pb/204Pb, which fall below 18.6 and closer to the upper crustal and the orogenic mountain formation curves (Zartman and Doe 1981), suggest that they are the result of older emplacements and that their derivation is from deeper crustal material. In short, these data points identify trends that exclude the silver-bearing ores of Laurion in Greece and the Taurus mountains as candidate sources for our 25 hacksilber samples from the southern Phoenician hoards.

Figure 5: The comparison of calculated growth curves and selected ores in the Mediterranean: AC = the average crustal curve (38), WM = the western Mediterranean (35); UC = the upper crustal curve and OR = and the deep crustal curve for orogenic mountain formation (20) versus in the western Mediterranean (32, 24, 29, 31, 19), Turkey (36, 39, 40), and the Aegean (37, 41, 42). Error = 0.1% amu. See key for numbered citations. [View static image]

We have limited our discussion so far about the loci of pre-colonial silver trade mainly to Sardinia and Iberia, but Classical authors, such as the 1st-century Greek historian Diodorus Siculus, also cited the early importance to the Phoenicians of silver supplies in Sicily and North Africa. It is also clear that the Greeks were targeting the metalliferous regions of the west for colonisation before 600. To the settlements already mentioned, we can add sites like Phoenician Lixus in Atlantic Morocco, near mineral deposits that include galena of the Upper-Moulouya District, and the Greek port of Marseille, at the foot of France's Pb-Ag-Zn rich southern Massif Central. The latter was geographically (and, in terms of early metals trade, potentially) connected via the Rhône-Saône valleys, Burgundy, the Seine and the English Channel to supplies of tin in Cornwall. Biplot 2 (Figure 3) presents the developing picture from lead isotope data that identifies consistencies between 15 of our 25 selected samples of southern Phoenician hacksilber and previously published galena from Les Malines, Massif Central South and Mont Lozère (variously), in southern France. Our sample 2HFZ006 is also consistent with galena from Mibladén in Morocco (Table 2). These basic numerical consistencies (that is, the data without interpretations) between individual samples of hacksilber and multiple ores are the result of overlaps in the isotopic field of major deposits of galena in the west, the number of which are expected to keep growing. Our understanding of relevant ores in Algeria, for example, needs to be expanded from the study already well begun by Belkacem Touahri (1991). Meanwhile, the initial lead isotope data from galena there already suggest a relationship to the older, deeper crustal sources of the western Mediterranean, and the elemental analyses that detected levels of silver above 300ppm in galena indicate the potential of some deposits for silver production.

In the light of these findings, the eventual concentrations of early Phoenician and Greek colonies on the coasts of the metalliferous regions in the west appear as reflections of metals-based connectivity begun earlier than 800 BC, whatever the primary economic orientation of a given colony may have been. But here we must ask, 'how much earlier?' Metals-based connectivity has already been attested (although not always clearly dated and provenanced) by finds in Sardinia, Sicily, Corsica, southern coastal France and (possibly) Ibiza of 'oxhide' ingots and their fragments that typify copper trade originating in Cyprus during its Late Bronze Age (c. 1550-1200 BC) (Lo Schiavo et al. 2009, 497). Likewise, lead artefacts with western Mediterranean ore-signatures were identified in Middle to Late Cypriot contexts almost 20 years ago (Knapp and Cherry 1994), and, in a recent re-evaluation of older data, Stos-Gale and Gale (2010) published lead objects found in early and late 2nd millennium Cyprus that have signatures consistent with the lead, copper and silver deposits of Massif Central, and Cambrian Sardinian lead ore deposits, respectively. The research question that asks whether Phoenicians were actively engaged in trans-Mediterranean metals trade between 1200 and 800 BC identifies an archaeological need for these re-evaluations; such finds permit, inter alia, the argument that all of the silver artefacts with western Mediterranean ore-signatures now identified in Phoenician contexts between 1200 and 800 could have been made from recycled lead or silver derived and/or imported to the east before 1200. But this is not the only complication.

There are also serious limitations and complexities inherent to the archaeological record of the Mediterranean basin between 1200 and 800; those, along with the documented overlaps in the isotopic ratios of major deposits of lead ores in the west, advocate the transparent reporting of multiple possibilities of ore origin, when they occur. That is, earlier provenance studies have focused on identifying a singular 'most likely' or 'real' provenance for each metallic artefact, and often report these origins in ways that obscure or ignore defendable alternatives; the lack of diagnostic criteria that might allow us to distinguish artefacts made with silver derived from ores in Sardinia and southern France, for example, tells us, instead, to develop ways to work effectively in view of multiple possibilities of provenance, and to disclose those possibilities. In any case, archaeologists must also grapple with a lack of remains from silver production and trade in the west securely datable between 1200 and 800, which might point to a lack of production and trade, or just as well to insufficient or inconclusive research and the ephemeral nature of transient trade that will have left few traditionally recognised, securely datable markers of contact.

This article identifies a need for research involving the continued characterisation of argentiferous and lead ores in the west (for example, we have too little visibility into the character of Sicilian, North African and Egyptian deposits), as well as native or aurian silver. This includes a need for improved documentation of the relative suitability in antiquity for silver production of potentially relevant ores. Such advances, however, will not resolve many of the uncertainties that characterise the pre-colonial period so that we can count on an Occam's Razor approach (at least one that depends on knowledge of the history of mining in each potentially relevant region) to identify the 'real' source of the metal used to make each artefact (cf. Stos-Gale and Gale 2010, 400). Even if it becomes possible to identify evidence of extraction or production datable to the pre-colonial period, later periods of exploitation will have, in a necessarily uncertain number of cases, obscured or obliterated other earlier remains, as well as particular mineralisations exploited in antiquity. For now, the lead isotope data presented here are, instead, better able to identify regions where further investigations of ore characterisations and ancient exploitation may prove productive.

The necessity of taking into account such complex realities is perhaps most clear in the Iberian peninsula's Pyritic Belt, where intensive exploitation in antiquity is attested by an estimated six million tons of ancient slag at Rio Tinto, and the extraction of silver from jarosites with low lead concentrations required additions of lead or litharge. That is, (potentially imported) lead used to extract silver sourced from Rio Tinto will have given its isotopic signature to silver produced with it. At least by the 7th century, shipwrecks in the western Mediterranean attest to the maritime transport of sometimes large quantities of galena or its derivatives, including lead ingots and litharge (Muhly 1998 with important discussion; Renzi et al. 2009).

For this article, we establish consistencies between our artefact samples and ores only to the extent needed to implicate the silver-rich regions of the west at large as suppliers to eastern economies before colonisation. For non-specialists we note that, once well established, the consistencies between the lead isotope ratios of artefacts and ores do not change, but the number of identified consistencies can and are expected to keep growing. For this reason, it is not our goal here to pinpoint singular origins for each sampled artefact, but rather to consider questions that rely on the existing ability to identify evidence consistent with derivation in the west. At the same time, our findings generate a need (or at least a wish) for sufficiently strong diagnostic criteria capable of identifying or favouring a particular origin when an artefact's lead isotope ratios are consistent with multiple deposits. Trace elemental analyses may yet prove helpful.

With reference to our findings in 2003, and based on a study of early incuse coins of western Greece, Giuseppe Giovannelli et al. (2005, with references) discussed the potential capacity of relatively high gold contents in silver to indicate derivation from oxidized ores. The basic idea is that silver from lead sulphide (galena) would not normally contain more than 0.1% of gold, while the gold/copper ratios of weathered ore zones may at times be higher than 0.25%; relatively high gold and bismuth levels may indicate the derivation of silver from a polymetallic ore or an oxide, rather than galena. This means some of our 19 samples with relatively high levels of gold (Table 2) could have lead isotope signatures that reflect a source of (possibly imported) lead or litharge used to purify silver from a polymetallic ore or an oxide through cupellation, rather than primary ore, and that at least some of these artefacts may have been produced from the gold- and silver-bearing jarosites in south-west Spain's Pyritic Belt, or from cerussites, which have been suspected as important sources of silver in antiquity, particularly in Egypt (Ogden 2000, 170). Data from our samples of hacksilber found at the Philistine site of Tel Miqne-Ekron in Cisjordan, also help identify the need to consider minerals and ores other than galena as potential sources of silver. Bi-plot 3 (Figure 4) includes 36 samples out of 50 taken from the 6 hoards found at Ekron that plot with the silver-bearing galena of Laurion in Greece. At the same time, they are inconsistent with the older deposits of the western Mediterranean and Energy Dispersive X-ray Florescence (ED XRF) detected levels of gold above 0.25% in the majority of these 36 samples (Table 1, see Appendix). Their isotopic signatures and levels of gold suggest derivation from cerussites at Laurion, but we will need to report more detailed analyses.

Analyses by ED XRF are a qualitative guide, typically used to determine which elements are present, and are not a substitute for more rigorously accurate quantitative measurements obtainable by other methods. As a case in point, and as we were beginning to build the next stages of research, we entrusted some of our samples and research questions to S. Shalev of the Weizmann Institute of Science, where levels of gold were detected as high as 11% and 13% in samples for which Oxford's Isotrace Laboratory reported levels between 2-4%. With support from the National Endowment of the Humanities, and in cooperation with the Alexandria Archive Institute and the California Digital Library, it has become possible to construct a digital platform for the presentation of more detailed elemental analyses of each piece of hacksilber in our sample-set. These will appear on-line at the Hacksilber Project on Open Context, alongside an indefinitely expanding account of the consistencies established between the lead isotope ratios of our hacksilber samples and the growing body of comparative data from geological sources and production remains. For now, we observe that the ED XRF analyses from Oxford's Isotrace Laboratory (some of which were cross-checked for accuracy by M. Cowell at the British Museum Research Laboratory) identified levels of gold above 0.1% in 19 of the 25 southern Phoenician hacksilber samples, including sample 2HFZ008 with 20.8% Au and 3HFZ001 with 35% Au. On present evidence, the least likely of our samples to have been derived from polymetallic ores or oxides include DOR004, KSN001, HFZ003, HFZ005, 2HFZ002; these are isotopically consistent with Sardinian and Iberian ores and have levels of gold reported below 0.1% (Table 2).


Internet Archaeology is an open access journal. Except where otherwise noted, content from this work may be used under the terms of the Creative Commons Attribution 3.0 (CC BY) Unported licence, which permits unrestricted use, distribution, and reproduction in any medium, provided that attribution to the author(s), the title of the work, the Internet Archaeology journal and the relevant URL/DOI are given.

University of York legal statements

File last updated: Wed Oct 16 2013