The strange case of the earliest silver extraction by European colonists in the New World A. M. Thibodeau*†, D. J. Killick‡, J. Ruiz*, J. T. Chesley*, K. Deagan§, J. M. Cruxent¶�, and W. Lyman** Departments of *Geosciences, ‡Anthropology, and **Materials Science and Engineering, University of Arizona, Tucson, AZ 85721; §Florida Museum of Natural History, University of Florida, Gainesville, FL 32611; and ¶Universidad Nacional y Experimental Francisco de Miranda, 4101 Coro, Venezuela Edited by David H. Thomas, American Museum of Natural History, New York, NY, and approved December 21, 2006 (received for review August 22, 2006) La Isabela, the first European town in the New World, was estab- lished in 1494 by the second expedition of Christopher Columbus but was abandoned by 1498. The main motive for settlement was to find and exploit deposits of precious metals. Archaeological evidence of silver extraction at La Isabela seemed to indicate that the expedition had located and tested deposits of silver-bearing lead ore in the Caribbean. Lead isotope analysis refutes this hypothesis but provides new evidence of the desperation of the inhabitants of La Isabela just before its abandonment. archaeological science � historical metallurgy � La Isabela � lead isotopes The first expedition of Cristóbal Colón (Christopher Colum-bus) to the Caribbean, from 1492 to 1493, returned to Spain with some gold taken from the Taı́no inhabitants of the island of Hispaniola (now divided between Haiti and the Dominican Republic). Columbus’s tales of the abundance of gold in the New World induced King Ferdinand and Queen Isabela of Spain to fund a second and much larger expedition (1). In early 1494, Columbus and �1,500 followers established the town of La Isabela on the northern coast of the present Dominican Republic (Fig. 1). His followers were eager to make their fortunes but were rapidly disillusioned by their failure to find any ores of precious metals (1, 2). La Isabela was beset by hunger, disease, hurricanes, mutiny, and conf licts with the Taı́no. In 1496 Columbus was recalled to Spain to account for his misadministration of the expedition. The few hundred remaining inhabitants of La Isa- bela struggled until early 1498, when they abandoned the town and marched across the island to found new settlements on the south coast. The archaeological site of La Isabela was intensively excavated in the late 1980s and early 1990s (1, 2). Among the finds were 58 triangular graphite-tempered assaying crucibles†† and 1 kg of liquid mercury, which was brought for the extraction of gold from powdered ore by amalgamation. The mercury was excavated from the ruins of the alhóndiga, a fortified structure erected for storage and protection of royal property (2). Within and imme- diately north of the alhóndiga, archaeologists also recovered �90 kg of galena (PbS), in blocks weighing up to 2 kg, and �200 kg of material classified in the field as metallurgical slag. Much of the ore and supposed slag was associated with a fired pit that appears to be the remnant of a small furnace that was located just north of the alhóndiga (2). Results and Discussion We examined polished sections of the galena and slag by using optical petrography and metallography, SEM, and electron microprobe analysis. Electron microprobe analysis of galena reveals highly variable silver content, ranging from �210 ppm to below the detection limit (�29 ppm) [see supporting information (SI) Table 2 and SI Methods]. The materials that are classified as slag can be divided into two groups. The first type, of which �15 kg was recovered, is crystalline and strongly magnetic. Optical and SEM analyses reveal abundant grains of metallic iron and crystals of a calcium–iron silicate in a calcium–iron–silica glass. This material is the result of a failed attempt to smelt iron with a calcareous f lux (coral): the temperatures achieved were not sufficient to allow the slag to f low freely, and thus it could not separate from the metal. The second and more abundant material (�185 kg) is a dense black glass. Optical and SEM analysis of multiple specimens shows that this glass is a lead silicate that contains tiny spheres of lead metal and chemically eroded remnants of quartz crystals that were shattered by thermal stress (see SI Fig. 4). Sulfur was not detected. Some samples of slag contain minute, complex inclusions of metallic lead, litharge (PbO), and metallic silver (Fig. 2). From these observations, we infer that some inhabitants of La Isabela were attempting to recover silver from galena by cupella- tion. The galena was reduced to metallic lead, a process that requires only that crushed ore be sprinkled on a bonfire of stacked wood (3). Sulfur is removed as sulfur dioxide fume, while any silver content dissolves into the metallic lead. Cupellation is the second stage of the process, whereby metallic lead is melted under a cover of a charcoal fire and subjected to an oxidizing blast produced by bellows. This process oxidizes the molten lead to molten litharge but does not oxidize any noble metals present. To concentrate silver, the molten litharge must be removed as it forms by absorption into or reaction with the lining of the furnace base. Author contributions: D.J.K. and J.R. designed research; A.M.T., D.J.K., J.T.C., K.D., J.M.C., and W.L.; performed research; J.R. contributed new reagents/analytic tools; A.M.T. ana- lyzed data; A.M.T. and D.J.K. wrote the paper; and K.D. and J.M.C. excavated site and provided samples. The authors declare no conflict of interest. This article is a PNAS direct submission. †To whom correspondence should be addressed at: Department of Geosciences, University of Arizona, Gould–Simpson Building no. 77, 1040 East 4th Street, Tucson, AZ 85721. E-mail: amthibod@email.arizona.edu. �Deceased February 23, 2005. ††Martinón-Torres, M., Rehren. T., 34th International Symposium for Archaeometry, May 3–7, 2004, Zarogoza, Spain. This article contains supporting information online at www.pnas.org/cgi/content/full/ 0607297104/DC1. © 2007 by The National Academy of Sciences of the USA 85 o W 80 o W 75 o W 70 o W 65 o W 60 o W 16 o N 18 o N 20 o N 22 o N 24 o N Cuba Jamaica Puerto Rico La Isabela Caribbean Sea Atlantic Ocean Hispaniola V o y age to C uba and Jamaica, A p r il- S eptembe r , 149 4 Fig. 1. Location of La Isabela. Filled squares are approximate locations of Pb deposits in the Greater Antilles (5). Open square is approximate location of the Pueblo Viejo Au–Ag oxide deposit in the Dominican Republic (7). Arrows represent the principal exploration route traveled by the second expedition after La Isabela was established. www.pnas.org�cgi�doi�10.1073�pnas.0607297104 PNAS � February 27, 2007 � vol. 104 � no. 9 � 3663–3666 A N TH R O P O LO G Y http://www.pnas.org/cgi/content/full/0607297104/DC1 http://www.pnas.org/cgi/content/full/0607297104/DC1 http://www.pnas.org/cgi/content/full/0607297104/DC1 http://www.pnas.org/cgi/content/full/0607297104/DC1 http://www.pnas.org/cgi/content/full/0607297104/DC1 In medieval Europe, bone ash was commonly used for lining cupellation hearths (3), because it absorbs litharge but does not react with it. Using bone ash in the cupellation process makes it possible both to extract silver and to recover desilvered lead by crushing and resmelting the used hearth linings. At La Isabela, the cupellation hearth was evidently lined with quartz-rich sand, which reacted with the molten litharge. This process formed a lead silicate glass from which the lead could not be recovered. Additionally, the formation of the glass would have made the oxidation of lead less efficient and caused higher losses of metallic lead, and potentially silver, to the slag during the cupellation process (Fig. 2). The spatial association of cupellation with the royal storehouse supported the hypothesis that the second expedition might have found and assayed galena from Caribbean ore deposits. Although lead mineralization is scarce in the Greater Antilles, galena is present in small volcanogenic massive sulfide deposits in Cuba, Haiti, and Jamaica and in vein deposits in Puerto Rico (4, 5). All of these islands were visited during Columbus’s second voyage (Fig. 1). To determine the provenance of the galena, we measured the lead isotope ratios of 12 samples of galena and 2 samples of lead silicate glass. The results cluster tightly, and the majority of values overlap within the 2� analytical error (Table 1). The similarity of the isotopic ratios supports our interpretation that all of the galena found at La Isabela derives from the same source. The slight displacement of lead isotope ratios evident in the samples of lead silicate glass suggests some admixture with lead from sand. Beach sands in this region are rich in quartz and coral, and locally derived lead can substitute for calcium in the latter. The isotopic composition of lead in the ore and slag samples from La Isabela, however, does not fall within the known range of values obtained for galena-bearing deposits in the Caribbean. These samples include galena from the Greater Antilles and sulfide minerals from the Pueblo Viejo Au–Ag oxide deposit in the Dominican Republic (Fig. 3). Although the data are scarce, they are consistent with models of crustal evolution for the Caribbean region and with other lead isotope measurements of rock and ore from the Greater Antilles, the Lesser Antilles, and Central America (4 –10). We therefore conclude that the galena at La Isabela was not collected from a Caribbean source. Because historic documents do not indicate that anyone from Columbus’s f leet reached the American mainland during the second expedition, we assume that the only other option is that the expedition brought galena from Europe. To evaluate this possibil- ity, we looked to ore deposits near Cádiz, the port from which Columbus’s second f leet sailed. The largest area of lead mineral- ization in Spain lies in the Los Pedroches–Alcudia Valley and the adjacent Linares–La Carolina ore fields (11). Collectively, these mining districts extend over an area of 10,000 km2 through the provinces of Ciudad Real, Badajoz, Córdoba, and Jaén and have been mined for thousands of years (11). Deposits from the Alcudia Valley are dominantly Pb–Zn polymetallic veins that contain variable but significant amounts of silver and copper (12). Galena is the dominant mineral within the Linares–La Carolina area (13). Santos Zalduegui and others (11) report lead isotope data for 125 samples of galena from the Alcudia Valley, Linares–La Carolina ore field, and the Los Pedroches batholith area. A comparison of 500 microns Litharge Metallic Lead Metallic Silver Shattered Quartz Crystal Fig. 2. Reflected light image of a lead silicate glass slag from La Isabela. Table 1. Pb isotope ratios measured on samples of galena and lead silicate slag from La Isabela Sample 206Pb/204Pb ratio 207Pb/204Pb ratio 208Pb/204Pb ratio 3a-G 18.224 15.614 38.369 3a-G (R) 18.221 15.610 38.354 6a-G 18.210 15.606 38.345 10a-G 18.211 15.608 38.349 13a-G 18.216 15.608 38.350 17a-G 18.214 15.608 38.349 17a-G (R) 18.212 15.607 38.346 23a-G 18.206 15.603 38.338 3b-G 18.196 15.602 38.328 6b-G 18.207 15.605 38.341 10b-G 18.210 15.606 38.345 13b-G 18.213 15.609 38.352 13b-G (R) 18.211 15.606 38.344 17b-G 18.218 15.609 38.347 23b-G 18.224 15.610 38.356 22-S 18.240 15.631 38.400 4-S 18.237 15.625 38.386 G, galena; S, lead silicate slag; R, repeat measurement. 15.60 15.62 15.64 15.66 15.68 15.70 18.1 18.3 18.5 18.7 18.9 19.1 38.20 38.30 38.40 38.50 38.60 38.70 38.80 38.90 Pb Slag Galena 206Pb/204Pb 2 0 8 P b/ 2 0 4 P b 2 0 7 P b /2 0 4 P b South Central Spain a b c d e Greater Antilles This Study: Fig. 3. Comparison of lead ore and lead silicate slag from La Isabela with galena and sulfide mineralization in the Greater Antilles (5, 7) and galena from the Alcudia Valley–Los Pedroches, and Linares–La Carolina mining dis- tricts in Spain (11). Individual groupings represent a, Cuba (n � 2); b, Haiti, Puerto Rico, and Jamaica (n � 4); c, Pueblo Viejo Au–Ag oxide deposit (n � 9); d, the Alcudia Valley, Los Pedroches, and Linares–La Carolina areas (n � 83); and e, Los Pedroches (n � 4). Error limits for 207Pb/204Pb measurements are shown; all other errors are smaller than symbols. Data from ref. 11 that are not shown lie outside range of the plot. For each area, n � the number of analyses represented by each shaded ore field. 3664 � www.pnas.org�cgi�doi�10.1073�pnas.0607297104 Thibodeau et al. these data with our measurements reveals that the galenas from La Isabela fall well within the range of values reported for the Linares–La Carolina fields and the Alcudia–Los Pedroches areas (Fig. 3). Other major sources of lead in Spain include the Betic Cordillera in the southeast (14), the western Pyrenees and Basque– Cantabrian basin in the north (15), the Rubiales ore deposit in the northwest (16), and the Catalonian Coastal Ranges in the northeast (17). However, the lead isotope ratios of these deposits (14 –17) do not match those of the La Isabela galenas. The lead isotope data alone cannot rule out the possibility that the galena came from some more distant part of Europe, the Mediterranean or North Africa. However, we believe that the combination of the specific historical circumstances, the geographical proximity of lead ore deposits to the port of departure of Columbus’s second f leet, the mineralogy of the deposits, and lead isotope data strongly support our conclusion that the galena recovered at La Isabela derives from the Alcudia Valley–Los Pedroches or Linares–La Carolina ore fields. Why would the second expedition have brought galena from Spain? The archaeological evidence from La Isabela (2) shows that metallic lead was used for shot and lead sheathing to protect the wooden hulls of ships. However, it is unlikely that the galena was brought for solely this purpose, because (i) galena would need to be smelted before use, and raw ore would have been an extremely inefficient use of the precious cargo space on the first transatlantic ships; and (ii) lead metal is among the supplies requested by Columbus as necessary for the colony after their arrival in 1493 (2). That almost all of the galena found at La Isabela was located in or near the alhóndiga supports the inference that it was royal property. Because the alhóndiga was used to store metallurgically important materials (such as mer- cury), we believe it is most probable that the galena was brought to La Isabela as a reagent for assaying ores for precious metals. In late medieval Europe, galena was used as a reagent in fire assays by smelting precise amounts of galena with other pow- dered ores together in a crucible so that the resulting lead would collect any noble metals present (M. Martinón-Torres, personal communication). After cupellation of the lead metal, the gold and silver content of the assayed ore could be determined by subtracting the quantities assumed or known to be present in the reagent galena (M. Martinón-Torres, personal communication). The low average silver content of the galena supports the interpretation that it was brought as an assaying reagent. The archaeological distributions of galena, slag, and cupelling residues were heavily concentrated just inside and outside of the north end of the alhóndiga. Piles of slag and galena were mapped on either side of two stone pillars that the excavators interpret as foundations for an arched portal through the north wall of the structure, which gave access to the furnace area, the beach and the shipyard (2). The scatter of slag and ore through the entranceway and into the structure suggests that the cupellation of the galena and the smelting of iron were among the last activities to have taken place at the alhóndiga, when there was no longer any concern about maintaining access to the beach or about dumping industrial waste within the royal storehouse. In early 1496, the Spaniards had located gold deposits in southern Hispaniola and subsequently established the port of Santo Do- mingo on the southern coast of the island. Throughout the latter half of 1496, Spanish settlement was gradually transferred to Santo Domingo, which significantly reduced the population of La Isabela (1). Despite the relocation, La Isabela remained occupied through 1497, although food rationing, hunger, and sickness heavily plagued the shrinking settlement (1). In May of 1497, La Isabela was sacked, and the alhóndiga was looted by a band of discontented rebels led by Francisco Roldán (1). Although a few settlers likely remained until late 1497, the site was effectively abandoned by 1498 (1). Given the archaeological context of the ore and slag, the cupelling of these ores likely postdates the breakdown of royal authority that led to the looting of the alhóndiga. We suggest that toward the end of the site’s occupation, some desperate survivors sought to salvage what they could from La Isabela before abandoning it. With the help of at least one person with basic knowledge of assaying procedures, they set out to cupel the galena stored in the alhóndiga. In so doing, they modified the standard medieval cupelling process by substituting beach sand for the usual bone ash hearth. Whether this unusual process ref lects a shortage of bone ash or a lack of metallurgical knowledge we cannot say. However, the failed attempt to smelt iron, which appears to be contemporary with the processing of galena, supports the latter inference. Thus, we believe that the processing of lead and iron ore at La Isabela was not the work of skilled assayers working for the Crown, but instead the result of the imperfect knowledge held by those at La Isabela who remained during the final stages of its occupation after the relocation of mining activities to other parts of Hispaniola. That �88 kg of galena was left unprocessed suggests that those cupelling the ore were disappointed with the return and abandoned their efforts. In conclusion, what seemed at first to be evidence of the earliest European mining and processing of precious metals in the New World appears, on further inspection, to be poignant testimony to the disillusion- ment and desperation of settlers who had embarked on the second expedition in the hope of making their fortunes. Methods Twelve samples of galena and two samples of lead silicate glass were analyzed for this study. All samples were chipped off larger pieces of material and accurately weighed. Six of the galena samples were taken from pieces that had previously been thin-sectioned and examined by using SEM. Each sample was stored in a clean dry Falcon tube (BD Biosciences, San Jose, CA). Samples were then transferred into clean 24-ml Tef lon containers. Ten milliliters of 8 M HNO3 was added to each galena sample, and 2 ml of HF was added to each sample of lead silicate glass. All samples were capped and heated on a hot plate at 125°C. To ensure total dissolution, samples were periodically placed in an ultrasonic bath. An additional sample that was composed solely of 10 ml of 8 M HNO3 was processed alongside the Pb samples to measure the total process blank. After complete dissolution, all solutions were diluted to a Pb concentration of �200 ppb based on sample weight. Sr and Pb were separated on the same Sr–Spec resin (Eichrom, Darien, IL) columns by using protocol developed at the University of Arizona. Data for a mixed National Bureau of Standards (NBS)- 981:Tl yields data with reproducibility similar to that described in ref. 18. Runs of United States Geological Survey basalt standard (BCR-2) are in excellent agreement with BCR-2 values measured by Woodhead and Hergt (19), using double-spiking thermal ionization mass spectrometry methods. Lead isotope analysis was conducted on a GV Instruments (Hudson, NH) multicollector inductively coupled plasma mass spectrometer (MC-ICP-MS) at the University of Arizona. Sam- ples were introduced into the instrument by free aspiration with a low-f low concentric nebulizer into a water-cooled chamber. A blank, consisting of 2% HNO3, was run before each sample. Before analysis, all samples were spiked with a Tl solution to achieve a Pb/Tl ratio of �10. Throughout the experiment, the standard National Bureau of Standards (NBS)-981 was run to monitor the stability of the instrument. All results were Hg-corrected and empirically normalized to Tl by using the exponential law correction according to ref. 18. To correct for machine and interlaboratory bias, all results were normalized to values reported by Galer and Abouchami (20) for the National Bureau of Standards (NBS)-981 standard (206Pb/204Pb � 16.9405, 207Pb/204Pb � 15.4963, and 208Pb/204Pb � 36.7219). Inter- nal error ref lects the reproducibility of the measurements on Thibodeau et al. PNAS � February 27, 2007 � vol. 104 � no. 9 � 3665 A N TH R O P O LO G Y individual samples, whereas external errors are derived from long- term reproducibility of NBS-981 Pb standard and result in part from the mass bias effects within the instrument. In all cases, external error exceeds the internal errors and is reported here. External errors associated with each Pb isotopic ratio are as follows: 206Pb/ 204Pb � 0.028%, 207Pb/204Pb � 0.028%, and 208Pb/204Pb � 0.031%. Mark Baker provided a great deal of assistance with sample prepa- ration and analysis. Marcos Martinón-Torres provided valuable in- formation on assaying practices in medieval Europe. This work was supported by a graduate fellowship from the National Science Foun- dation Integrative Graduate Education and Research Traineeship program in the Archaeological Sciences (to A.M.T.). The Dirección Nacional de Parques de la República Dominicana provided funds for excavation of La Isabela and permitted scientific analysis of the samples described here. Excavations were also supported by National Endowment for the Humanities Grant RK20135-94 and the National Geographic Society. The development of analytical methodology and isotopic analyses were supported by the Keck Foundation and National Science Foundation Grants EAR-9976676, EAR-0125773, and EAR-0409417. 1. Deagan K, Cruxent JM (2002) Columbus’s Outpost Among the Tainos: Spain and America at La Isabela, 1493–1498 (Yale Univ Press, New Haven, CT). 2. Deagan K, Cruxent JM (2002) Archaeology at La Isabela, America’s First European Town (Yale Univ Press, New Haven, CT). 3. Craddock P (1995) Early Metal Mining and Production (Smithsonian Inst, New York). 4. Cumming GL, Kesler SE (1976) Earth Planet Sci Lett 31:262–268. 5. Cumming GL, Kesler SE, Krstic D (1981) Earth Planet Sci Lett 56:199 – 209. 6. Cumming GL, Kesler SE (1987) Chem Geol 65:15–23. 7. Cumming GL, Kesler SE, Krstic D (1982) Econ Geol 77:1939 –1942. 8. Lebrón MC, Pefit MR (1994) Tectonophysics 229:69 –100. 9. O’Blein S, Guillot S, Lapierre H, Mercier de Lépinay B, Lardeaux JM, Millan Trujillo G, Campos M, Garcia A (2003) J Geol 111:89 –101. 10. Davidson JP (1987) Geochim Cosmochim Acta 51:2185–2198. 11. Santos Zalduegui JF, Garcia de Madinabeitia S, Ibarguchi JI, Palero F (2004) Archaeometry 46:625– 634. 12. Palero FJ, Both RA, Arribas A, Boyce AJ, Mangas J, Martin-Izard A (2003) Econ Geol 98:577– 605. 13. Lunar R, Moreno T, Lombardero M, Regueiro M, López-Vera F, Martı́nez del Olmo W, Mallo Garcı́a JM, Saenz de Santa Maria JA, Garcı́a-Palomero F, Higueras P, et al. (2002) in The Geology of Spain, eds Gibbons W, Moreno T (Geological Society, London), pp 473–510. 14. Arribas A, Tosdal RM (1994) Econ Geol 89:1074 –1093. 15. Velasco F, Pesquera A, Herrero JM (1996) Miner Deposita 31:84 –92. 16. Tornos F, Arias D (1993) Eur J Mineral 5:763–773. 17. Canals A, Cardellach E (1997) Miner Deposita 32:243–249. 18. Rehkämper M, Mezger K (2000) J Anal At Spectrom 15:1451–1460. 19. Woodhead JD, Hergt M (2000) Geostand Newsl 24:33–38. 20. Galer SJG, Abouchami W (1998) Mineral Mag 62A:491– 492. 3666 � www.pnas.org�cgi�doi�10.1073�pnas.0607297104 Thibodeau et al.