Isotopic Ag–Cu–Pb record of silver circulation through 16th–18th century Spain

Abstract

Estimating global fluxes of precious metals is key to understanding early monetary systems. This work adds silver (Ag) to the metals (Pb and Cu) used so far to trace the provenance of coinage through variations in isotopic abundances. Silver, copper, and lead isotopes were measured in 91 coins from the East Mediterranean Antiquity and Roman world, medieval western Europe, 16th–18th century Spain, Mexico, and the Andes and show a great potential for provenance studies. Pre-1492 European silver can be distinguished from Mexican and Andean metal. European silver dominated Spanish coinage until Philip III, but had, 80 y later after the reign of Philip V, been flushed from the monetary mass and replaced by Mexican silver.

A particularly momentous time during the early history of modern European economy was the attempt by Hamilton (1) to demonstrate that the great Price Revolution (1520–1650) was largely fueled by the influx of American silver rather than by widespread coinage debasement and minting of the low-denomination copper “vellón”. The idea connecting silver influx to European inflation was actually proposed as far back as the16th century by the French philosopher Jean Bodin (2) and is commonplace in classical economics. Huge amounts of silver, ∼300 t annually (3–5), were mined in the Spanish Americas from the 16th to the 18th centuries. That much silver could not be absorbed locally by the American economy and therefore headed for the European market through major Spanish harbors (6), notably Seville (7), and to the Far East either directly through the Philippines or indirectly through Europe (8). The thesis that the Price Revolution in Spain was fueled by the influx of American silver has, however, become controversial in recent literature (9–11). More specifically, some authors emphasized that the arrival of American metals (ca. 1550 to ca. 1809) does not coincide with the period of inflation (ca. 1520 to ca. 1650) (9–11). Understanding silver monetary mass and circulation relies on three types of primary data: (i) the register of taxes collected when the silver bars received the royal stamp (the Quinto in Peru and the Diezmo in Mexico) (12, 13), (ii) the register of the European harbors used to import the silver shippings (1), and (iii) the compilation of contemporaneous gazettes (9). These data are imprecise or even incomplete, especially for trade registers between 1660 and 1809 (9), and do not take contraband and piracy silver into account (8, 14–17). In addition, any memory of the origin of the metal is lost by recoinage, whenever silver is exported or a new king comes to power, or upon debasement. Reliable tracers of the monetary mass and exchange that can see through the destructive alterations of coinage silver therefore are needed. Over the last 30 years, lead isotope compositions of metallic ores have been collected and gathered into large databases and broadly used as a tool for provenance studies of archaeological artifacts (18–20). The main factors of provenance analysis are (i) the contrast between ores produced by mantle-derived magmas with low ²⁰⁷Pb/²⁰⁴Pb, such as in Cyprus, southern Spain, and the Andes, and those produced in geologically ancient crust with high ²⁰⁷Pb/²⁰⁴Pb (such as magmatism from the Altiplano) (21) and (ii) the age of the crustal provinces from which the Pb ores were extracted. Unfortunately, the Pb isotope ratios of ores are strongly correlated with each other, which often makes provenance assignment insufficiently discriminating. More recently, the high precision of the multiple-collector–inductively coupled plasma mass spectrometry (MC-ICPMS) technique (22) allowed Cu isotopes to be added to the coinage tracers and a number of successful applications to the identification of the sources of metals used for coinage have been suggested (23). Although copper is primarily alloyed with coinage silver to improve metal hardness and resistance, it was also used for monetary debasement (17). Copper has two stable isotopes of mass 63 and 65, and, in contrast to the large variations in radiogenic Pb isotope abundances, which are due to the radioactive decay of U and Th, the abundance variability of Cu isotopes is due exclusively to the physico-chemical conditions of ore-forming processes (primary hydrothermal sulfides vs. low-temperature sulfides and hydrocarbonates) (23, 24) and remains within a few parts per 1,000. The other multi-isotopic coinage metal is silver (Au is monoisotopic), but beyond some preliminary data on silver ores (25–27) no archeological application has been attempted. Silver has two naturally occurring isotopes, ¹⁰⁷Ag (51.4%) and ¹⁰⁹Ag (48.6%). Evidence of 0.5‰ Ag isotopic variability among silver ores (25–27) provides a strong incentive to use the ¹⁰⁹Ag/¹⁰⁷Ag ratio as a provenance tracer despite the need for time-consuming high-precision isotopic analysis.

With the incentive that the input of American silver into the European monetary mass may be visible in the isotopic abundances of metals used for coinage, this work presents Pb, Cu, and Ag isotope data on silver and billon coins from Europe and the Spanish Americas. We first analyzed reference material from the Antique world (Greek, Hellenistic, Roman, Near Eastern) and medieval times, notably pre-Columbian Spain. We then analyzed the isotope compositions of Pb, Cu, and Ag in 16th–18th century American coinage from Mexico and South America and compared them to the isotope compositions of European Spanish coins of the same age. We discuss the problems associated with the allocation of the coinage metals to potential sources, notably isotopic modification during the metallurgical processes. We also discuss how these data can elucidate the history of the monetary mass and circulation in the world.

Results

The analytical techniques are described in SI Materials and Methods and the data are listed in Table S1. In the following, we first examine the isotopic results one element at a time and then describe how the data for different elements correlate with each other.

Lead.

Lead has four isotopes, the stable ²⁰⁴Pb and the radiogenic ²⁰⁶Pb, ²⁰⁷Pb, and ²⁰⁸Pb produced by radioactive decay of ²³⁸U, ²³⁵U, and ²³²Th, respectively. The choice of plots used to represent the data is particularly important: The traditional ²⁰⁷Pb/²⁰⁴Pb vs. ²⁰⁶Pb/²⁰⁴Pb, and ²⁰⁸Pb/²⁰⁴Pb vs. ²⁰⁶Pb/²⁰⁴Pb diagrams have a long history in geochemistry and are based on the well-understood control by the age of the ore and the U/Pb and Th/Pb ratios of its source (crust vs. mantle). In contrast, archeologists (28) favor different plots, notably ²⁰⁸Pb/²⁰⁶Pb vs. ²⁰⁷Pb/²⁰⁶Pb, which reduce the analytical noise by removing the correlations induced by relatively higher counting errors associated with the low abundances of ²⁰⁴Pb. However familiar these plots may be, those used for archeological purposes are more difficult to relate to the geological history of the ore source, a particularly important parameter because the Aegean, the Betic (southern Spain), and the American Cordilleras formed <120 Ma and are still geologically active, whereas most of Central Europe is underlain by Hercynian basement (250–400 Ma) (Fig. 1). The “model ages” T listed in Table S1 were calculated with the common Pb isotope composition and the ²³⁸U/²⁰⁴Pb of the crust (29, 30) using the formula given in SI Materials and Methods. As shown in ²⁰⁷Pb/²⁰⁴Pb–²⁰⁶Pb/²⁰⁴Pb and ²⁰⁸Pb/²⁰⁴Pb–²⁰⁶Pb/²⁰⁴Pb space (Fig. 2), ²⁰⁷Pb/²⁰⁴Pb and ²⁰⁸Pb/²⁰⁴Pb ratios are higher in Antique and Potosi coins than in Mexican and European medieval coins, which surprisingly overlap. The ²⁰⁶Pb/²⁰⁴Pb ratios of most Antique silver coins are derived from isotopically young provinces (<120 Ma) and fit sources within the Aegean, Asia Minor, and southeastern Spain (Betic Cordillera) (20, 28, 31, 32) (Figs. 2 and 3). The Basque–Cantabrian basin is also a suitable source but its Ag production was relatively minor (33). Exceptions are Gr2, a 500 BC Iona Miletos diobol; R118, a 118 BC Licinus Crassus denier from the Narbo mint in Gaul; and R121, a 121 BC denier of Caius Plutius from the Rome mint, which probably all derived their Pb from Hercynian (∼300 Ma) ores of the European basement. In contrast, Pb from all of the European medieval coins derives from the western European Hercynian basement. Spanish medieval coins form two distinct groups with coinage predating the Catholic Kings (1454–1474), having ²⁰⁶Pb/²⁰⁴Pb <18.5 and therefore bearing a European Hercynian signature contrasting with the largely >18.6 values of the Catholic Kings samples (1479–1504), which are more similar to the values found in the Betic Cordillera district. As expected, the ²⁰⁶Pb/²⁰⁴Pb ratios of coins from Spanish America are strongly imprinted by the rather recent magmatic activity of the Cordilleras (<130 Ma) and overlap with the Antique and precolonial European coins derived from the Aegean and Betic districts. ²⁰⁶Pb/²⁰⁴Pb in 16th–18th century Spanish coinage varies between precolonial and Mexican values. The Pb model ages of two 17th century coins, one French and one English, are conspicuously young (<70 Ma).

Fig. 1.

Map of Europe with the <90-Ma-old Alpine (in particular Aegean, Basque–Cantabrian, and Betic) domains indicated in gray overprinting the older western European Hercynian (250- to 400-Ma-old) basement.

Fig. 2.

Lead isotope compositions for Antique (n = 24), medieval (n = 23), Andean (n = 11), Mexican (n = 8), and 16th–18th century European (n = 25) coins. Analytical errors are smaller than the symbols.

Fig. 3.

(A) δ⁶⁵Cu (in parts per 1,000) vs. ε¹⁰⁹Ag (in parts per 10,000). (B) Pb model ages T (in millions of years) vs. ε¹⁰⁹Ag for Antique (n = 20), medieval (n = 23), Andean (n = 11), Mexican (n = 8), and 16th–18th century European (n = 24) coins. Sample GR2 is not reported on this graph due to its very different Cu isotope composition (δ⁶⁵Cu = −4.06‰, see text). Errors on δ⁶⁵Cu (2 SD) and ε¹⁰⁹Ag (2 SEM) are shown for each sample. Probability ellipses are represented for three groups of samples: Antique and medieval coins, Mexican coins, and Andean coins.

Copper.

Isotopic data are reported in δ notation, for which the isotope composition is cast as the deviation of a particular isotopic ratio with respect to the same ratio in a standard material [here National Institute of Standards and Technology (NIST) 976]: δ⁶⁵Cu = [((⁶⁵Cu/⁶³Cu)_sample/(⁶⁵Cu/⁶³Cu)_standard) – 1] × 1,000. The δ⁶⁵Cu unit is the part per 1,000 (per mil or ‰). Fig. 3A shows the Cu isotope compositions of analyzed samples. Except for sample (Gr2: δ⁶⁵Cu = −4.06‰), which was also anomalous for Pb isotopes, the δ⁶⁵Cu values of Antique coins range between −1.00 and +0.15‰, whereas the medieval, Mexican, Andean, and 16th–18th century European coins have δ⁶⁵Cu ranging from −0.5 to +1‰. δ⁶⁵Cu values for medieval samples fall between −0.5 and 0.5‰_. The range of δ⁶⁵Cu values for both Mexico (0.0 ± 0.2‰) and Potosi (+0.7 ± 0.2‰) is particularly narrow. The δ⁶⁵Cu of most European coins from the 16th–18th century, regardless of country, is similar to Mexican values (Fig. 3A).

Silver.

Isotopic data are reported in parts per 10,000, (ε), with ε¹⁰⁹Ag = [((¹⁰⁹Ag/¹⁰⁷Ag)_sample/(¹⁰⁹Ag/¹⁰⁷Ag)_standard) – 1] × 10,000 with respect to the standard SRM 978a.

The observed range of isotopic variations (from −1 to +2 ε-units) is ∼30 times the analytical uncertainty (∼0.1ε). The Ag isotope variability of Antique coins defines two groups (Fig. 3), the oldest having ε¹⁰⁹Ag ∼ 0.0 and the younger, mainly composed of Roman and Gallic coins, having ε¹⁰⁹Ag ∼ −0.5. The two groups are statistically different at the 99% confidence level. The ε¹⁰⁹Ag values of the medieval, European, and Spanish coins, with the exception of the Catholic Kings coins, overlap with those of the second Antique group (Fig. 3). As in the case for Pb, the Ag isotope composition of the Catholic Kings coins also is distinct from that of the rest of the medieval coins (ε¹⁰⁹Ag = from 0.0 to +1.5). The eight Mexican samples show very little ε¹⁰⁹Ag variability (0.7 ± 0.2). This range overlaps with the values of two Mexican native silver ores that we measured (ε¹⁰⁹Ag = 0.3 and 1.2), Pachuca and Guanajuato, two large mines actively exploited in colonial times. The ε¹⁰⁹Ag value of −5.3 reported by Hauri et al. (26) for the silver ore form Zacualpan (Mexico) is much more negative. In contrast, the Ag isotope compositions of Potosi coins are quite variable. In the Spanish samples from the 16th–18th century, the ε¹⁰⁹Ag varies from −1.0 to +1.0, a range that includes both the French (0.35) and English coins (0.32).

Ag–Cu–Pb correlations.

Fig. 3A shows that each group of coins falls in a specific part of the ε¹⁰⁹Ag–δ⁶⁵Cu diagram with a few outliers. The field of Antique and medieval coins before the Catholic Kings plots toward the negative δ⁶⁵Cu and ε¹⁰⁹Ag quadrant. Seven Potosi coins stand out for their high δ⁶⁵Cu, whereas three of them have values close to 0.0‰: δ⁶⁵Cu and ε¹⁰⁹Ag are not correlated. Most European 16th–18th century and Mexican coins plot in a narrow field at δ⁶⁵Cu ∼ 0.0‰ and ε¹⁰⁹Ag ∼ 0.7. The Ag and Cu isotope compositions of Philip III Spanish coins are surprisingly similar to those of the pre-Catholic Kings medieval samples. The only significant correlation between Ag and Pb isotopes is observed for the Antiques coins with Pb model ages T < 200 Ma (r = −0.8) (Fig. 3B).

Discussion

First, we assess the extent to which isotope compositions can be used to identify the source of metal ores. We review the various processes that cause isotopic variability within a given ore deposit, including the mass-dependent thermodynamic fractionation of Ag and Cu isotopes, and the variability over regional distances, which is important for Ag as well as for Pb isotopes. Processing of coins may alter the original isotopic signatures, in particular isotopic fractionation during metallurgical processes, and willful additions or accidental contamination can overprint the original initial isotopic signature. Second, we use the Pb, Ag, and Cu tracers to infer the provenance of the metals contained within the samples analyzed in this study. Third, we examine how these provenance inferences relate to known historical events and what they teach us about silver circulation through the 16th–18th century Spanish economy.

Thermodynamic Isotope Variability Among Coexisting Ores.

The extent of Ag isotope fractionation among native metal and other Ag minerals (sulfides, sulfo-antimonides, and chlorides) from a single deposit or from nearby localities is not well studied, but some inferences can be made from available data. A single observation shows that the Ag isotope compositions of coexisting pyrite and native silver ore from Pribram (Czech Republic) are within error of each other (27). The conspicuous isotopic homogeneity of Mexican samples, however, makes a strong case against substantial thermodynamic fractionation among minerals. For example, the mining of both oxidized [colorados (red ores), inclusive of native silver, Ag chlorides and bromides, and Fe oxides] and sulfidic [negros (black ores), inclusive of pyrargyrite, pyrite, galena, and sphalerite] silver at the major camp of Zacatecas (34, 35) does not translate into measurable Ag isotopic variability among the coins struck in Mexico.

The range of Cu isotope variation among coexisting ores is clearly larger than that for Ag (23, 24). Both Cu isotopic homogeneity and values close to zero argue for the incorporation of metal from high-temperature ores into Mexican coins, possibly bornite or chalcopyrite. Likewise, the 10 samples from Potosi have similar δ⁶⁵Cu of +0.7 ± 0.2‰, whereas the 3 samples dated between 1620 and 1670 have values between −0.3 and +0.1‰. The narrow range of δ⁶⁵Cu within each group signals the use of high-temperature ores.

In contrast to the cases of Ag and Cu, lead isotope variability results from the radioactive decay of U and Th rather than thermodynamic fractionation. Consequently, Pb isotopes do not reflect on the predominant sulfide type, but mainly on the regional geologic context of the ore.

Origin of Regional Isotopic Variability.

Silver.

Natural silver isotope variability is very narrow (25–27) and a meaningful isotopic signal can be detected only with extremely precise measurements (±0.1ε). Except for a handful of data on native silver, the isotopic composition of silver ores is largely unknown. Native silver ores from Italy, Mexico, Canada, Russia, and Norway show ε¹⁰⁹Ag variations of ∼5 parts per 10,000 (25–27). A major finding of our work is that a substantial proportion of American Ag has isotope compositions that can be distinguished from those of the metal used in pre-1492 European mints. This is the case for the 8 Mexican coins analyzed here (ε¹⁰⁹Ag = 0.7 ± 0.2), which are distinct from the 24 Antique and 18 medieval European coins (ε¹⁰⁹Ag = −0.2 ± 0.6) (the Catholic Kings, 1479–1504, stand out as an exception) (Fig. 3). In contrast to the Viceroyalty of Peru, Mexican silver production was divided among several major camps (4). The metals analyzed here most probably come from different deposits, yet the Mexican Ag isotope signature is distinct. The spread of ε¹⁰⁹Ag in Potosi and Antique coinage reflects a different situation. The geological setting of the Cerro Rico mine at Potosi, which is hosted by a young (13.8 Ma) volcanic dome intrusive into a much older (Ordovician or ∼450 Ma) series of sediments, is complex (36). This complexity is apparent in the regional Pb model ages (Fig. S1). Different sources of silver may therefore be involved in the same ore deposit. In addition, different mines contribute to the same mint. In contrast to Mexico, where no particular silver mine dominated, Potosi was more productive than the other camps in Peru, which had difficulties with labor and capital (4). Ambiguities about the origin of silver coins arise from both the metal registration and minting. Registered Potosi production accounts for 80% of the silver minted over the history of colonial Peru, but other neighboring mines, such as Sicasica (1600), Tatasi (1612), and Padua (1652), also had their production registered in Potosi during the last half of the 17th century (4). The Potosi mint processed Potosi silver but also independently registered metal from nearby mines (Porco and Oruro) and occasionally dealt with silver from the major and more remote camp of Cerro de Pasco whenever the Lima mint closed down (Fig. S1). The spread of Ag isotope compositions of the Antique coins also calls for multiple sources of metal: ε¹⁰⁹Ag correlates with Pb model ages T (r = −0.8 if the three samples with T > 200 Ma are disregarded). The ore deposits therefore tap a very young geological source (T ∼ 0) of metal with ε¹⁰⁹Ag ∼ 0.0, mostly represented by the coins from the eastern Mediterranean Basin on the one hand and metal from the older basement (T > 200 Ma) mostly represented by Roman and Gaelic coins (ε¹⁰⁹Ag ∼ −0.5) on the other hand.

Copper.

As noted before, the δ⁶⁵Cu values separate most Potosi coins from the rest of the corpus. Because copper isotope compositions are controlled by ore genesis, it is unlikely that δ⁶⁵Cu is a true regional variable. Despite this, our observations indicate that the Potosi mint was using copper that came from an isotopically well-defined, but unidentified ore deposit. We therefore consider δ⁶⁵Cu of ∼ +0.7‰ as a geographically meaningful tracer of Potosi copper.

Lead.

The Pb model ages T of the samples (Table S1 and Fig. 3B), and to a large extent the ²⁰⁷Pb/²⁰⁶Pb and ²⁰⁸Pb/²⁰⁶Pb ratios, are reliable indicators of the geological age at which the crustal volume that gave rise to a particular ore body was isolated by geodynamic processes. For example, the ²⁰⁶Pb/²⁰⁴Pb ratios of the European Betic, Mexican, and Andean districts, which all derive from young provinces (<130 Ma), are similar and distinct from those derived from Hercynian northern Spain and western Europe, where crust mostly formed 250–450 Ma. Thus, it is difficult to use Pb model ages T (or, equivalently, Pb isotope ratios) to distinguish coins from southern Spain from those of the New World. Other ratios, such as ²⁰⁷Pb/²⁰⁴Pb, may fingerprint old Precambrian crustal segments, such as the Andean Altiplano (21), whereas ²⁰⁸Pb/²⁰⁴Pb variation relative to ²⁰⁶Pb/²⁰⁴Pb reflects the poorly understood variability of Th/U ratios among source rocks. One possible interpretation of the spread of Pb model ages T and ²⁰⁶Pb/²⁰⁴Pb ratios observed for Potosi coinage, and that contrasts with the homogeneity of Pb in Mexican coins, is the mixing of modern Pb from the Cerro Rico volcanics with Pb from the old sedimentary basement.

Effects of Extractive Metallurgy, Recycling, and Coinage.

Different processes were involved in the extractive metallurgy of silver:

• i) Smelting consists of heating the metal ore, possibly after a first stage of roasting, with a reducing agent, commonly charcoal. During the medieval period, some metallic Pb was added at the beginning of the smelting process to facilitate the recovery of Ag directly from galena (37).

• ii) After the 1550s in Mexico and the 1570s in Peru, the kind of amalgamation known as the patio process of silver extraction allowed silver recovery from very low-grade ore. The ore was first finely crushed. Added to this ore were large quantities of common table salt (NaCl), vitriol (CuSO₄ and FeSO₄), known as magistral, and mercury, with a typical ratio of lost mercury to silver produced of 1.5 (38). The resulting amalgam was boiled off and both silver and mercury were retrieved. Mercury used in Spanish America came from three sources: the Huancavelica mines (1,500 km north of Potosi) (Fig. S1), which provided mercury to the viceroyalty of Peru; the Almaden mines in southern Spain, which supplied mainly Mexico and less frequently Potosi; and the Idria mines in modern Slovenia, which were tapped occasionally to make up for any shortfalls from the two principal sources (12, 39).

• iii) Cupellation is a purification stage that separates metals easily oxidized, typically Pb and Cu, from Ag, which remains metallic. It often involves litharge (PbO) addition. This stage was used after smelting and for recycling preexisting metals (coins and silverware) with large Cu contents.

For the metallurgical process to induce isotope fractionation, it must cause partial vaporization of the metals or involve a solid or a liquid phase that coexists with the metal, notably silicate-rich slag or Pb-rich oxides. Because little isotope fractionation of stable metal isotopes is expected at the high temperatures of metallurgy, the yield has to be poor for isotope fractionation to be observed. Baron et al. (37) concluded from smelting experiments that the Pb isotope signature of ores is preserved during metallurgy. Likewise, Cu extraction and refining processes do not alter the copper isotope signature of copper ores (40). For Ag, no effort was spared to keep the yield as high as possible, and hence any potential isotope fractionation of Ag was minimized. The homogeneity of Ag isotopic compositions in Mexican coins and their similarity with those of local ore suggest that fractionation related to metallurgy is weak. The patio process spread in Mexico starting in the 1550s while smelting and silver recovery by lead cupellation still persisted. Smelting was largely used in the late 17th century due to a shortage of mercury in Mexico (4) and was in general preferred for the treatment of Ag-rich ores, which was the case of Pb ores (34). Poor miners and Indian laborers, who received part of their wages in ore (39), also favored it. Until the 18th century, it has been estimated that roughly half of the silver produced in Mexico came from the smelting process (4). Extractive metallurgy, whether smelting or amalgamation, is therefore not a significant source of Ag isotope variability. This assessment may not be valid for Peru, where it is known that yields were poor (4).

Use of additives during metallurgical processes or recoinage (37, 41) is expected to distort fingerprinting if the additives and the ore have a different origin. Likewise, litharge (PbO) addition during postsmelting purification by cupellation may overwhelm the original Pb isotope signature. Such problems are particularly serious if mining, ore treatment, and metal purification take place at different locations. In the patio process, the many additives are a concern, notably mercury. The Pb content of mercury and cinnabar is not recorded, but given the solubility of lead, both in mercury and in silver (∼1%) (42, 43), amalgamation may have altered the Pb isotope composition with respect to the original ore. In addition, lead contamination by cupellation is expected during the refinement frequently required before recoinage (44). Accession of new monarchs, design of new coins, silver import, and coinage debasement were all opportunities to introduce foreign Pb into the metal. New World silver reminted in Spain is expected to involve European lead and to obscure the American Pb isotope signature of the metal. Likewise, Andean and Mexican silver went through multiple recoinage in the Americas (1728, 1772, and 1786) (17), but local reprocessing is less likely to corrupt the isotopic signature.

The purpose of copper alloying is either improved coin hardness and resistance (∼5–10%) or debasement (17). In general, Pb and Cu ores form in different environments (13). Pb contents in chalcopyrite are rather limited and, for fine silver coins, Pb contamination associated with Cu addition should be less important than during purification. The Ordinance of Medina de Campo (1497) defines the weight of the Real (3.434 g) and stipulates that it contains 3.195 g of silver (93% Ag and 7% Cu) (45). The use of Cu isotopes to trace Andean and Mexican silver in Spanish colonial coins is relevant only for exports consisting of silver coins. Otherwise the copper signature is inherited from wherever copper was added, possibly in Europe. According to Garner (14), if there is little doubt that silver was exported in different forms, coins were the favorite one. Although up to a few percent metallic Cu can be alloyed with Ag (46), silver of lower fineness is prone to unmixing and, during oxidative high-temperature reprocessing, to oxidation (47): Removal of Cu by cupellation is then necessary before realloying. Coinage recycling preexisting coins can therefore be suspected to contain copper freshly added at the site where recycling took place.

Provenance Assessment.

Antique and pre-1492 European coins.

Most Antique coins are characterized by young Pb model ages T and low (<0.1) ε¹⁰⁹Ag values. The very young Pb model ages T of the silver coins from the eastern Mediterranean area (with ε¹⁰⁹Ag ∼ 0.0, see above) are consistent with the prevalence of either local metal sources or sources in the Betic district in Southern Spain (33). The intermediate (∼100 Ma) model Pb ages T of the Roman and Gaelic coins are geologically unusual and suggest that mixed sources of Pb were used in silver metallurgy. In contrast, most European medieval and a handful of Roman coins are characterized by Pb model ages T older than 200 Ma and negative ε¹⁰⁹Ag isotope compositions. Clearly, Antique silver either is a minor component of the medieval silver monetary mass or has been largely reprocessed using old Hercynian Pb. A puzzling case is that of the coins of the Catholic Kings (1479–1504), in which the young Pb model ages T (28–120 Ma) reflect the prevalence of a Betic metal (20, 31, 32). This period corresponds to the capture of the kingdom of Granada and of its rich Ag mines by the Spanish kings. The scatter of ε¹⁰⁹Ag for the Catholic Kings coins is uncharacteristic. The only coin of the Catholic Kings (ES32) with a Hercynian signature was struck in a northern mint at Burgos, whereas the other coins were struck in Seville or Granada in southernmost Spain.

Mexico.

The ε¹⁰⁹Ag of Mexican coins is very similar to native silver from Pachuca and Guanajuato, two large mines exploited by the Spaniards, analyzed in this study (+0.3 and +1.1, Table S1), but differs from the value of −5.3 ± 0.5 found by Hauri et al. (26) for Zacualpan. This difference suggests that Zacualpan silver was not used for coins analyzed so far. The Pb isotope compositions are consistent with what is known of the Mexican sulfide ores in the region of colonial Spanish mining in the center and the north of the country (48, 49).

Andes.

The only available ε¹⁰⁹Ag value from a local mine (Porco, Table S1) falls within the range of Potosi coins. Likewise, the isotope composition of Pb present in the coins struck in Potosi is consistent with literature values for the neighboring Ag ore deposits of Cerro Rico, Oruro, and Porco (ref. 50 and references therein) (Fig. S2). In contrast, the Pb isotope composition of Cerro de Pasco next to Lima is clearly different (51, 52). The spread of Pb isotope compositions requires the contribution of a rather old end member, which can be European Pb introduced by amalgamation, local Pb from the Paleozoic basement, or possibly an unknown source (Fig. S2). The two samples PotoD and PotoF with the oldest Pb model ages T (329 and 252 Ma, respectively) and also distinctly high ε¹⁰⁹Ag (+1.6 and +3.3) suggest that they represent a local variation rather than contamination or isotope fractionation during the amalgamation process. For the rest of the Potosi coins, compelling evidence for what created the Ag isotope variability is missing. Whether a low yield of the extractive metallurgy or regional isotopic variations, the Ag isotopic signature is heterogeneous and not particularly distinctive.

European 16th–18th century coins.

The Spanish vellón coin ES4 of Charles V (1516–1555), the half-real ES40 of Philipp II (1556–1598), and four coins of Philipp III (1598–1621) (ES19, ES21, ES48, and ES45) all have δ⁶⁵Cu at ∼0.0‰. With the exception of the intermediate ES40, this group of coins also has negative ε¹⁰⁹Ag. These characteristics do not fit American metal supply and can be explained by the alloying of pre-1492 silver with Pb of local origin. Both Charles V and Philipp II coins have Hercynian Pb model ages T (287 and 247 Ma, respectively). The Pb model ages T of the Philipp III coins are consistent with the locality of the mint: ES19 (39 Ma) and ES21 (36 Ma) were struck in Seville, whereas ES45 (268 Ma) was struck in Valladolid. The Toledo sample (ES48, 147 Ma) is intermediate. In contrast, from Philipp V (1700–1746) onward, silver isotopes (ε¹⁰⁹Ag ∼ +0.7) unambiguously demonstrate the prevalence of Mexican silver in the Spanish monetary mass of the 18th century. The observation that few Spanish colonial coins have Pb model ages T as young as those of Potosi and Mexico does not imply that Spanish silver was not imported from the Americas. Rather, it emphasizes widespread recoinage and refining, which mostly used local European lead. The δ⁶⁵Cu and ε¹⁰⁹Ag of the 17th century French and English coins and their young model ages point to a strong contribution of Mexican metals (Fig. 3).

Silver Circulation.

The Spanish quest for silver in the New World started as early as 1498 (53), but it was only in the early to mid-16th century that the major mines were opened (4). European Spanish coins from the 16th and early 17th centuries (Charles V to Philip III) show no input of American metal, which suggests that coins and cobs struck in Mexico and Peru were not realloyed into Spanish coinage. Of the silver mined in the Americas, 20% stayed on the continent (12). Another 10% were used to buy Asian silk, porcelain, and spices (14), and yet another 15% fell in the hands of pirates or was smuggled (15), leaving ∼200 tons to reach Seville every year of the late 1500s and early 1600s (4, 5, 9). Philip II defaulted on the Spanish debt in 1557, 1560, 1575, and 1596 (54, 55) and reopened the Rio Tinto and other Spanish silver mines (56), which confirms that American silver did not stay in Spain very long. As stated by Braudel (ref. 57, p. 205) “every consignment of American silver was quickly dispersed in all directions, almost like an explosion.” Silver was mostly exported from Spain to repay the huge loan obtained from the German bankers to secure Charles V's election and to repay Genoese bankers for other major loans. It was also lost to subsidize the wars in The Netherlands; to buy grains, cloth, and paper that Spain did not produce itself in sufficient quantities; or simply because petty money was driving out good silver (16). However, the absence of the New World treasures from Spain is not sufficient to reject their role in the Price Revolution. According to Flynn (10), the influx of American silver may have had a global influence on international markets and may explain Spain's inflation as a reflection of the overall European Price Revolution. Our isotopic data on the coins of Philip V (1700–1746) indicate that Mexican silver found its way to the Spanish silver monetary mass only in the aftermath of the Utrecht treaty (1713), which marks a major break in Spanish involvement in foreign wars, which in turn coincides with the onset of a major phase of silver mining expansion in America (4, 5, 58). These data also show that the isotopic signature of European silver until Philip III (1598–1621) is not detectable in Philip V coins. If the corpus analyzed in the present work is a representative sample of the contemporaneous metal, the implication is that by the time of Philip V, Mexican silver had actually replaced the silver monetary mass circulating under Philip III, which was partly exported and partly diluted.

Conclusions

This work demonstrates that silver isotopes can be used successfully to trace the origin of coinage. The usefulness of Pb isotope compositions as tracers can be strengthened by using model ages T, which represent a geologic characteristic of ore deposits, but is made ambiguous by silver purification and reprocessing, which often involve local sources of Pb distinct from the ore sources. The combined use of Ag, Cu, and Pb can distinguish pre-1492 European silver from Mexican and Andean metal sources. European silver dominated Spanish coinage until Philip III but, 80 years later under the reign of Philip V, had been flushed from the monetary mass and replaced by Mexican silver.

Materials and Methods

The corpus is composed of 94 samples described in Table S1. After cleaning, a piece of coin was cut off with pliers and dissolved in nitric acid. Ag was precipitated by addition of ascorbic acid. Cu was separated in HCl medium and Pb in HBr medium on anion exchange columns and then analyzed by MC-ICPMS (22), using Pd (Ag), Zn (Cu), and Tl (Pb) to correct for instrumental mass bias. Details relative to sample preparation and MC-ICPMS measurement protocols are provided in SI Materials and Methods.