Introduction

Lead isotope databases such as OXALID (Stos-Gale and Gale 2009), IBERLID (de Madinabeitia et al. 2021), TerraLID (its prototypes were introduced as GlobaLID) (Klein et al. 2022), and the Lyon database (Milot et al. 2021a, 2021b; Vaxevanopoulos et al. 2022; Westner et al. 2023) hold almost 7000 Pb isotopic data entries on galena and other less common potential ores spread across the entire Western Europe, the circum-Mediterranean regions, and the Middle East. Although radioactive decay of uranium and thorium makes the variability of Pb isotope compositions very large, the sheer number of samples present a real technical challenge for assessing the provenance of silver artifacts. Mixed bullion adds complexity to these attempts; however, the problem of mixing was recently addressed with some arguable success by our group (Albarede et al. 2024b).

The challenge presented by silver isotopes is very different. In contrast to the abundances of Pb isotopes, which reflect the tectonic age of ore formation and the U/Pb and Th/Pb of its source(s) (Albarède et al. 2012), isotope fractionation of the two isotopes 107Ag and 109Ag in natural fluids and minerals is controlled by the energetics of Ag bonds in molecular species and the temperature of hydrothermal solutions (Fujii and Albarede 2018; Wang et al. 2022). Silver isotope fractionation becomes vanishingly small for solutions hotter than 400 °C and does not occur during the metallurgical process until the metal is almost completely evaporated (Berger et al. 2021). Silver isotopes should not, in general, be regarded as true provenance markers and are not expected to co-vary with Pb isotopes. The range of Ag isotope abundances measured in silver coinage of different ages spanning from the end of the Iron Age to modern times, in an ample field of geographical origins across several continents (Europe, Asia, Africa, Americas) is only a few parts per 10,000 (Fujii and Albarede 2018). In contrast, the range of silver isotope compositions displayed by the most common silver ores (galena (PbS), several types of Ag-sulfosalts, acanthite (Ag2S), and native silver from hydrothermal deposits (Arribas et al. 2020)) is an order of magnitude larger (Fig. 1). Such a contrast indicates that one very particular type of ore dominates the sources of the bullion used for silver coinage. Silver is not extracted from a multitude of different ores, but from galena and its low-temperature alteration products, such as sulfides and sulfates. A positive consequence of this observation is that, while Pb isotope compositions of galena and other silver-rich ores help identify potential ‘candidate’ bullion sources, silver isotopes help weed out a large fraction of the same candidates (Milot et al. 2021b, 2022; Vaxevanopoulos et al. 2022; Westner et al. 2023). When combined with archaeological evidence and ancient literary testimony, the pruning of potential silver sources and the identification of so-far unnoticed alternative ore districts using silver isotopes therefore may be of great historical value. Silver and lead isotopes are not provenance tracers in the same way, but they complement each other and, when used together, constitute a powerful provenance tool.

Fig. 1
figure 1

Histogram of Ag isotope compositions of coins from ancient Greece, Persia, Rome (Albarède et al. 2016; Desaulty et al. 2011; Milot et al. 2021b; Vaxevanopoulos et al. 2022), medieval Europe (Desaulty et al. 2011), Tudor England (Desaulty and Albarede 2013), and Spanish colonial Americas (Desaulty et al. 2011). The ε109Ag value of all but five of the 273 silver coins plot in the interval -1 to + 1 (-0.1 to + 0.1 permil). The spread of ε109Ag values in ores is much broader than in coins. Ores with ε109Ag values outside of the coin range can be excluded as sources of the silver used for minting

Full size image

In order to document the complementarity of Pb and Ag isotopes, let us summarize the current state of Ag isotope research in Archaeology. Over 95% of the silver coins and artifacts from ancient Greece, the Roman Republic and Early Empire, Medieval Europe, and Spanish Americas analyzed for silver isotopes have ε109Ag valuesFootnote 1 falling within the range of –1 to + 1 (Albarède et al. 2016; Desaulty et al. 2011; Desaulty and Albarede 2013; Eshel et al. 2022; Milot et al. 2021b, 2022; Vaxevanopoulos et al. 2022). These values will hereinafter be referred to as the ‘main range’ or group 1. Some rare Greek coins (Vaxevanopoulos et al. 2022) of smaller denominations, several Roman-period coins from the Balkans (Westner et al., submitted), and pieces of hacksilber (chopped bits of silver) from hoards in the Levant (Eshel et al. 2022) are isotopically lighter, hence falling outside the main range with ε109Ag values between -1.9 and -1.0. These are hereinafter referred to as group 2.

For Ag to be used as a provenance tool to its full extent, the geographical coverage of Ag isotopes in potential ores around the ancient Mediterranean world must be expanded. The purpose of the present study is to do exactly that: expand the Lyon Pb and Ag isotope database to Sardinia and, by geological extension, southern France because of their geological similarity and potential historical interests. Ancient Sardinia and Gaul are not widely perceived as major sources of silver, at least not on par with mining districts in the southern Aegean, Thrace, Macedonia, and Iberia. The galena and sulfosalt ores in the small districts of Northern Sardinia, though having been exploited in the past, are not rich enough in silver to have provided abundant bullion. Argentiera is the only silver-rich lead deposit (Orlandi and Gelosa 2007) characterized by Pb isotopic values that differ from those of the Iglesiente ores (Eshel et al. 2019).

The scarcity of Bronze Age silver artifacts found in Sardinia has been interpreted as indicating little mining activity of argentiferous ores in Sardinian chiefdoms during this period (Terpstra 2021; Valera et al. 2003). Nevertheless, recent Pb isotope studies reveal that Nuragic populations traded silver from the geologically older southern Sardinia with the Levant in the early Iron Age (Eshel et al. 2019, 2021; Gentelli et al. 2021) (see also Pearce’s (2018) review). At even later times, lead isotope data for Archaic coinage of Athens, Corinth, and Aegina are consistent with a small but distinctive contribution of South Sardinian-like silver (Albarede et al. 2024a) interpreted as some sort of ‘cash float’ of trade between Greece and southern Italy. By contrast, intensive exploitation of the deposits in the Iglesiente mining district by Carthage and the Romans after the First Punic War is attested to in the literature (Ingo et al. 1996; Valera et al. 2005).

Since Sardinia once belonged to the Hercynian margin of southern France and Catalonia until the Late Oligocene (~ 30 Ma) (Cherchi and Montadert 1982; Jolivet and Faccenna 2000; Puddu et al. 2021; Rehault et al. 1984; Romagny et al. 2020; Schettino and Turco 2011) and, together with Corsica, was embedded in an eastern extension of the Pyrenees (Casas 2010), we analyzed samples from southern France as well as samples from Sardinia. Extending our reconnaissance area in this way is further justified by lead isotope provenance studies pointing to ore deposits in southern France as potential bullion sources for hacksilber (Gentelli et al. 2021) and Archaic coinage (Stos-Gale and Davis 2020).

To summarize, the present work presents new high-precision Pb and Ag isotopic data for a whole region of the western Mediterranean which up until now was devoid of Ag isotope data. This new data set allows us to test whether Sardinia and southern Gaul are acceptable sources of bullion used to mint ancient silver coinage. Although access to abandoned mines and ore deposits in this region is limited and in most cases impossible, we nevertheless managed to obtain a total of 26 galena ores from southern Sardinia and southern France (13 samples from each region) from old mine heaps and collections. We then combined the newly measured values with literature data and reviewed the occurrences of potential circum-Mediterranean sources that may satisfy the 'Ag isotope condition' (a.k.a the ‘main range’ or group 1, but also group 2) mentioned above. Identifying plausible sources of bullion around the western Mediterranean using Ag isotopes (in combination with Pb isotopes as explained above) is the prime objective of the present work.

Geological setting

The geology of southwestern Sardinia is largely dominated by sedimentary Cambro-Ordovician rocks (Fig. 2). The geological units in the Iglesiente-Sulcis mining district consist of low-grade metamorphic rocks, belonging to the so-called “External zones” of the Variscan orogen (Franceschelli et al. 2005). The Lower Cambrian succession is subdivided into the basal Nebida Group and the overlying Gonnesa Group, which consists of carbonate rocks hosting Zn-Pb mineralization considered as sedimentary-exhalative (Sedex) and large, carbonate-hosted, massive sulfide Zn/Pb (Irish-type) ore deposits (Bechstädt and Boni 1994; Boni et al. 1996; Santoro et al. 2023). The Cambrian sediments were deformed by a first tectonic phase in the upper Ordovician (the Sardic phase). At the end of the major Variscan orogeny, the Lower Paleozoic basement was intruded by post-collisional granites, which caused the formation of skarn-type deposits. Variscan tectonics and magmatism were followed by a long continental period, associated with erosion and deep karstification of the Cambrian carbonates, which periodically underwent hydrothermal dolomitization episodes. From the Permian onwards, southern Sardinia experienced several hydrothermal phases comparable to those having occurred in other European terranes. The associated ores consist of low-temperature veins and paleokarst breccia fillings in the Cambrian carbonates, which contain mainly Ag-rich galena and barite mineralizations (Santoro et al. 2023).

Fig. 2
figure 2

Geological map of southwestern Sardinia with the locations of the samples analyzed in the present work (red circles) (Boni, unpublished work)

Full size image

The Montagne Noire is the southern extension of the French Massif Central (Fig. 3). It consists of three main structural domains: (1) a metamorphic axial zone made up of complex domes of gneiss and migmatites surrounded by micaschists; (2) a northern flank composed of imbricated tectonic nappes of Cambrian to Silurian rocks; and (3) a southern flank made up of large nappes involving Cambrian to Carboniferous strata (Álvaro et al. 2008). In the Montagne Noire, carbonate rocks were deposited during part of the Cambrian, as in the Iberian Peninsula and Sardinia. At the southern edge of the Massif Central, in a region known as the Cévennes, several nappes are correlated with those of the northern Montagne Noire. In Les Malines mining district the two southernmost units are exposed, represented by Les Malines ‘autochtonous’ and overlying Saint Bresson units. The Cambrian stratigraphic record of the above units is an analogy to the northern Montagne Noire, in the broad terms of paleogeographic and paleotectonic evolution. The stratigraphic position of the northern Montagne Noire Sedex-type mineralization corresponds roughly to that of the mineralization at Sanguinède and Montdardier in the Cévennes (Orgeval et al. 2000). An exception is represented by the classic Les Malines mine ores, which are of Triassic age (Orgeval et al. 2000). In contrast to the Sedex-type occurrences of the northern Montagne Noire, the southern Montagne Noire deposits, set in a shallow carbonate platform, are considered as Mississippi Valley-type (Lescuyer and Giot 1987; Marignac and Cuney 1999) and can be compared with part of the Pb ores of the Iglesiente in Sardinia. In between the Cévennes and the Montagne Noire there is a small area consisting of Late Proterozoic and Cambrian terranes intruded by the Mendic granite of terminal Ediacaran age (Leveque 1990).

Fig. 3
figure 3

Simplified geological map of southern France with the locations of the samples analyzed in the present work (modified after Asch (2005))

Full size image

Several reviews and monographs on the Variscan basement of the Pyrenees have been published over the last three decades (Casas et al. 2019; Denèle et al. 2014; Guitard et al. 1995; Laumonier et al. 2008). The axial zone of the Pyrenees, which hosts part of the samples analyzed in this study, has been involved in two pre-Alpine events: the Cadomian event (= pan-African) across the Proterozoic-Cambrian boundary, and the Carboniferous Variscan event (= Hercynian). Discontinuities contemporaneous with the Ordovician Sardic phase are ubiquitous in this area.

Sampling and analytical techniques

Ore sampling has become a modern challenge. For safety and liability reasons, most major abandoned mine shafts in several European countries have been blasted, flooded, and sealed to prevent the intrusion of unprotected collectors. An even more pervasive limitation is the continually shrinking space dedicated to public rock collections stored and curated by universities and other academic institutions. The field of ore geology and its perception by the public has considerably changed over the last few decades. As large numbers of researchers with skills in Ore Geology have retired, the space that was dedicated to their ore and rock collections has been reassigned to newly develo** fields and the collections themselves scattered, discarded, or moved to premises inappropriate for conservation. We therefore were limited to rare, unfortunately often poorly documented, but always properly curated samples that had been preserved in collections accessible at the University of Naples and the Museum of the Ecole des Mines de Paris. Additionally, several samples were personally collected by one of us (M.B.) in SW Sardinia.

Most of the samples analyzed in the present study consist of galena, with the exception of the bournonite (PbCuSbS3) sample 63,600 from the Cambrian-late Proterozoic terranes of Brusque and the cerussite sample S-38 from Les Malines (Cévennes). The names and origins of all the samples are listed in Table 1.

Table 1 Ore localities, Pb and Ag isotope data, and major and trace element data of the analyzed samples
Full size table

In southern France, we sampled three distinct units of Variscan crust, which have a very similar geological history: the Montagne Noire, the Cévennes, and the axial zone of the Pyrenees. In the Peyrebrune district (samples 59,051, 59,040, and Tarn), west of the Montagne Noire, predominantly Zn-(± Pb)-F veins occur emplaced in early Paleozoic schists (Munoz 1997), which were already exploited by the Gauls. To the north-east of the Montagne Noire (Brusque, sample 63,600), strata-bound Pb–Zn ores are found in Cambrian carbonates (Guérangé-Lozes et al. 1982). In the Cévennes, the samples from Les Malines and Durfort (samples 63,025, S-38, and Durfort) belong to multiple ores deposited during successive phases in Paleozoic and Triassic carbonates (Le Guen et al. 1991). The axial zone of the Pyrenees contains sedimentary-exhalative Pb–Zn deposits formed during the Paleozoic (Aulus-Argentière, sample 65,644 and 65,649; Aulus-Lauqueille, sample 65,726; Seix, sample 65,647; Abères, sample 65,826). Vein-type mineralizations within Late-Silurian and Devonian calcschist and limestone apparently produced significant amounts of silver (Munoz et al. 2016), although the details of how much bullion was extracted from these remote and high-altitude mines at any given time remain essentially unknown. Fragments of amphoras found in gullies indicate that these mines were known since at least Roman times. An inventory of Pb-Ag ores is available for the zone of interest, known as Couserans (the Ariège department), in Dubois (1997) and an updated discussion on Pyrenean ore mineralogy can be found in Cugerone (2018).

The techniques used to measure major and trace element concentrations and purify silver and lead for high-precision isotope analysis by MC-ICP-MS are fully described in a number of recent publications by our group (Milot et al. 2021a, 2021b; Vaxevanopoulos et al. 2022; Westner et al. 2023). For two galena samples,’Tarn’ and ‘Durfort’, there was not enough Ag for high-precision isotopic analysis. We suspect that, for these two samples, the low yields of the Ag separation resulted from unusually large abundances of organic material in the ores interfering with the column chromatography.

Results

Binary plots of Pb isotopic ratios are often misleading because they account poorly for the 3-dimensional character of the data (Albarède et al. 2020): inclusion of a given data point in the projections of a 3-dimensional field in two 2-dimensional plots is not sufficient to demonstrate that the data point in question is part of the original 3-dimensional field. Nevertheless, to comply with the usual practice of Archaeometry, Pb isotopic ratios have been plotted in Fig. 4 in the conventional manner. To avoid overcrowding the figures, only ore samples for which both Ag and Pb isotope compositions are currently available are plotted, whether this be the conventional binary Pb isotope plots or the maps of Figs. 5,6,7, the latter of which emphasize regional clusters (Milot et al. 2022; Vaxevanopoulos et al. 2022; Westner et al. 2023). These maps therefore include both the new samples of this study and samples from the literature with both Ag and Pb isotope data. Note that, on the maps, overlap** points have been made visible by a small random ‘jittering’ around the true location. The precise locations of the samples are given in the original publications reporting the Pb isotope compositions.

Fig. 4
figure 4

Conventional Pb isotope plots for Sardinia and southern France. Iberian samples and ‘others’ from Milot et al. (2021b, 2022). Southern Greece and the Balkans, including Laurion, Macedonia, and Thrace, from Vaxevanopoulos et al. (2022). Balkans (including Thrace and Macedonia) from Westner et al. (2023). ‘Others’ stand for localities that do not belong to these regions

Full size image
Fig. 5
figure 5

Map of the 204Pb-normalized ratios for samples for which Ag isotopes are also known (from this work and the literature: Milot et al. 2021b, 2022; Vaxevanopoulos et al. 2022; Westner et al. 2023). Note the strong clustering of some silver ore provinces

Full size image
Fig. 6
figure 6

Map of the 206Pb-normalized ratios for samples for which Ag isotopes are also known

Full size image
Fig. 7
figure 7

Map of the Pb model age Tm and μ (238U/204Pb) and κ (232Th/238U) values for samples for which Ag isotopes are also known

Full size image

The Pb isotope compositions of the galena samples analyzed here are consistent with geological and archaeological literature data. This is the case for both Sardinia (Boni and Koeppel 1985; Boni et al. 1996; Orgeval et al. 2000), the Pyrenees (Marcoux and Moelo 1991; Marcoux 1987; Munoz et al. 2016), and the southern and southeastern part of the Massif Central of France (Cévennes) (Baron et al. 2006; Brevart et al. 1982; Charef 1986; Le Guen et al. 1991, 1992; Ploquin et al. 2010).

In general, the Pb isotope data are well regionalized (Figs. 5,6,7):

  • 206Pb/204Pb and Pb model ages Tm show a sharp contrast between four groups: (i) Sardinia; (ii) Sierra Morena and Central Europe; (iii) northeastern Spain and southern France; and (iv) the coastal Betics, Serbia, North Macedonia, and Greece.

  • As expected, 207Pb/204Pb and µ values show some similarity between southern Sardinia and southern France. 208Pb/204Pb and κ values also show similarities.

  • Low values of 208Pb/204Pb and high values of 208Pb/206Pb from southern Sardinia (Iglesiente) and Central Europe contrast with samples from the coastal Betics in southern Spain, Serbia, North Macedonia, and Greece. The Sierra Morena in south-central Spain and southern France are intermediate between the two groups.

Overall, Late Proterozoic and Early Paleozoic model ages are found in southern Sardinia and occasionally in the Sierra Morena and southern France (Fig. 7). Paleozoic ages are more common in south-central Spain, northern Spain, southern France, and central Europe, while Late Mesozoic and Cenozoic ages prevail in the coastal Betics, the Balkans, and Greece.

The ore samples analyzed for silver isotopes (Fig. 8) have been subdivided into three groups corresponding to the thresholds previously observed in silver coinage. The main range coinage group (group 1) has ε109Ag values ranging from -1 to + 1 parts per 10,000 with respect to the NIST 978A reference material and is consistent with 95% of the values observed in Bronze to Iron Age hacksilber hoards from the Levant, in coins from ancient Greece and Rome, in medieval European coins, and in coinage from Spanish Americas (16-18th C). A second, isotopically light group (group 2) is defined by its ε109Ag values extending from -1 to -2 parts per 10,000, values that are found in some hacksilber hoards (Eshel et al. 2022) from the Levant and in a small number of archaic Greek coins (Vaxevanopoulos et al. 2022). The ε109Ag values of the third group (group 3), also referred to as ‘external’, fall outside the -2 to + 1 interval and do not correspond to any silver coinage or artifacts known so far. No samples from Sardinia fall in the main group, while five of them fall in the isotopically light group, and the rest fall in the external group (Table 1).

Fig. 8
figure 8

Map of the ε109Ag values of the galena samples analyzed in this work (Sardinia and southern France)

Full size image

Ores from regions covered by the broad denominations of the Balkans, Macedonia, Thrace, and Greece make up most of the likely sources of silver for the main coinage group 1. Some ore deposits from the Sierra Morena, the eastern Central Pyrenees, the southern Massif Central (notably the mine of Peyrebrune), and central Europe also represent acceptable bullion sources for this group. Group 2 is represented by Sardinia, the coastal Betics, and ore deposits scattered over the Balkans, Thrace, Iberia, and southern France. In the context of currently available data of ancient artifacts, the external group 3 does not carry any particular meaning.

Discussion

As expected from the very different physical processes controlling Ag and Pb isotope variations, the correlation between ε109Ag and the different Pb isotope ratios is weak and statistically non-significant at the 95% confidence level (correlation coefficient  r ~ ± 0.4). The existence of two regions with contrasting Pb isotope compositions (the Aegean vs the western Mediterranean, Figs. 5,6,7) nevertheless explains to some extent why r is not truly zero.

For Sardinia, the probability that the present samples belong to the main group 1 and therefore were actually used for coinage is statistically low. In contrast, values from the isotopically light group are clearly identified. The small number of samples, of course, places some limits on a broad generalization. Archaeological evidence nevertheless suggests that Sardinian metallurgy took off in the last quarter of the fifth millennium BCE or later (De Caro et al. 2013; Pearce 2018). Evidence for mining of argentiferous galena has been recorded from the Iglesiente province, notably at the Monteponi and Montevecchio mines (Ingo et al. 1996; Valera et al. 2003, 2005). Eshel et al. (2019, 2022) observed that the Pb isotope ratios of ores from San Giovanni are consistent with the silver artifacts they analyzed. The chronology of the mining works is still incomplete. Craddock (1995) and Pearce (2018) suggested that, in early prehistoric times, silver may have been preferentially extracted from supergene acanthite (Ag2S), cerargyrite (a.k.a., chloragyrite AgCl), and argentiferous cerussite, which have since been mined away, rather than from hypogene galena and associated sulfosalts. Although, to some extent, these results may be a consequence of limited ore sampling, it seems unlikely that these rare minerals ever represented a major source of silver in Sardinia with respect to Ag-hosting galena. In addition, supergene Ag minerals display large 109Ag/107Ag fractionation with respect to hypogene galena (Arribas et al. 2020). A provisional conclusion hence is that at least part of the silver mined in Sardinia belongs to the isotopically light group (-2 ≤  ε109Ag ≤ -1) and should be recognized as such in silver artifacts and coins. However, further Ag isotope studies are warranted on Ag-rich ores (> 1000 ppm) before the significance of Sardinia on the silver circuits will be firmly established.

The situation for silver mining is different during the Iron Age. Isotopically light silver similar to that characterizing group 2 has been found in hacksilber hoards from the Levant by Eshel et al. (2022). Evidence from the combined Pb and Ag isotope signatures is more difficult to interpret. Artifacts with a Sardinian Pb isotope signature do not have the Ag isotope characteristics of the present Sardinian ores. Using new software, which relies on Pb isotopes and takes mass-dependent fractionation into account to locate the provenance of samples (Albarede et al. 2024a), we confirm Eshel et al.’s (2022) general findings for some hacksilber hoards that the lead isotope compositions of the Tel Dor, ʽAkko, and Meggido hoards (all from Early Iron Age I), and possibly those of Tel Keisan (Late Iron Age) (Eshel et al. 2018), point to a possible Pb source from Sardinia. The Sardinian Pb isotope imprint on Iron Age artifacts may therefore be occasionally strong. The ε109Ag values reported by Eshel et al. (2022) for these hoards are, however, typical of the main group 1. The samples reported on by these authors from the Shiloh hoard do belong to the isotopically light Ag group, but their high 206Pb/204Pb values are not compatible with a Sardinian source. The search therefore must continue in order to identify silver ores from Sardinia that belong to the main group 1. Additional silver isotope data on potential silver ores are definitely a high priority for future research.

Likewise, the ε109Ag values of five isolated coins from Cyzicus (Mysia), Corinth, Abdera (Thrace), Chersonesus (Thrace), and Caria vary between -1.04 and -1.92 (Vaxevanopoulos et al. 2022), while their Pb isotope compositions and tightly grouped Pb model ages (24–52 Ma) suggest Aegean sources. Some silver sources exist in Troad, Mysia, and Caria (Yigit 2009, 2012), but only few Pb isotope data are available (Wagner et al. 1985). In either case, the dilemma is between isotopically light silver not being a reliable provenance tool and lead used for cupellation potentially coming from a source very different from the mine where the silver originated. If, as argued above, galena should be abundant wherever silver is exploited, the former conclusion takes precedence. Of course, other factors, such as the market price for lead, may enter into the equation as foreign lead transported by merchant ships might have been more competitive than locally extracted metal. Lead ingots found in shipwrecks and particularly abound in Roman times (Domergue and Rico 2003; Tisseyre et al. 2008; Trincherini et al. 2001, 2009) attest to intense long-distance trade of this metal. Whether lead of distant origins prevailed on Pb markets remains a topic for further investigations, at least for the East Mediterranean world.

Gentelli et al. (2021) suggested that some Iron Age hacksilber hoards from the Levant (Beth Shean, Tel Keisan, and Tel Dor) comprise pieces for which a southern Gaul origin is plausible, which the new provenancing software developed by Albarede et al. (2024a) confirms. The present work identified ores with ε109Ag values similar to the main coinage group 1 in southern Gaul, east-central Pyrenees, southern Massif Central, and the Cévennes. An isolated value from Bourg d’Oisans in the French Alps (CRPG 1046, (Milot et al. 2021b)) is also part of this group. This is consistent with archaeological finds of Ag mining activity from the Roman period in southern Gaul (Abraham 2000; Baron et al. 2006; Bonsangue 2011; Domergue and Leroy 2000; Feugère and Py 2007; Ploquin et al. 2010). Dubois (1997) argues that argentiferous galena was exploited from the Albères (maritime east Pyrenees) and Esplas-de-Sérous lodes (east-central Pyrenees) since Roman or even pre-Roman times, and mentions amphora fragments in gullies from the Couserans. Writing at the turn of the first millennium before the Roman conquest, the Greek historian and geographer Strabo mentions (3.2.8; 4.1.12; 4.2.2) silver mining in the lands of the Ruteni (southern Massif Central) and the Gabales (northern Cévennes) (see Hirt’s (2020) review). In their review of Iron Age and Roman metallurgy, Domergue et al. (2006), possibly influenced by Diodorus Siculus (Libr. Hist. 5.27),Footnote 2 argue from observations of ancient mining works that Gaul was, overall, not a major silver-producing region, i.e., not on par with the Aegean and Iberia, neither under the Gauls nor under the Romans. However, both in the field and in ancient literature, statistics are inadequate. Therefore, excluding Gaul a priori from silver provenance studies may introduce significant bias.

As a closing remark, silver production by a particular mining district is difficult to evaluate, first because reliable numbers are missing and, second, because some mining works may not have been preserved. In addition, many potentially important samples have become notoriously difficult to obtain. Thanks to the proximity of active silver mines, mints of Athens and Thasos in Greek times stroke local production (Albarede et al. 2024a). In contrast, the Lugdunum (Lyon) workshops from the Late Roman Republic and Early Empire are located far from any significant source of bullion, but are nevertheless known to have processed large silver issues (Sutherland 2018). Production estimates may also be amplified for political reasons at the time, notably as communication warfare. Moreover, mints concentrated artisanal expertise, skillful engravers, and well-trained slaves, and as such may have attracted non-domestic bullion.

Conclusions

New Ag and Pb isotope compositions on galena samples from southern Sardinia suggest that silver extracted from local mines is different from the most common bullion used to mint silver coinage. The connection between hacksilber hoards from the Levant and ores from Sardinia is confirmed. At this stage, however, the overall number of samples analyzed is still too small to assess whether Sardinian ores were a major contributor to Greek and Roman silver bullion.

Mines from southern Gaul produced silver with a range of Ag isotope compositions that render these ores consistent with those of the main silver pool (group 1). Hence, not including this area among the significant silver sources in antiquity reflects biases of ancient literature, which does not mention southern Gaul, and the lack of archaeological evidence for major mining works. As far as Pb and Ag isotopes are concerned, southern Gaul cannot be excluded as a potential source of silver bullion.