Post by Admin on Sept 17, 2019 21:58:04 GMT
In order to discriminate among the European sources that were used for the tin ingots from Israel, the lead isotope data can be considered together with the artefact’s tin isotope composition. From the tin isotope systematics additional information regarding ore charges and metallurgical treatments can be inferred, particularly in combination with the trace elements.
Fig 14A and Tables 6 and S2 summarise the tin isotope compositions of the tin ingots. Overall, only positive δ124Sn values (relative to our in-house standard) are observed that spread over a large range between 0.05 ± 0.03 ‰ and 0.42 ± 0.01 ‰. Individual sites, however, show a much smaller variation, and differences between them become discernible. The ingots can be divided into three groups (Fig 14A): one with high δ124Sn values of greater than 0.19 ‰, one with medium δ124Sn values ranging from about 0.12 to 0.18 ‰, and a third one with low δ124Sn values of less than 0.12 ‰. The large variation of the tin isotope ratios implies that the different groups of ingots–even within the same archaeological context–were produced from different tin ore charges (for comparison of actual data with older analyses see S1 File). Compared with trace elements, no distinct relationships are recognisable overall, but individual sites or groups of ingots with smaller variations in the tin isotopes reveal relationships with trace elements such as iron, antimony, tungsten, lead, bismuth, and indium (Fig 15).
Fig 14. Tin isotope composition (δ124Sn) of the tin ingots examined in this study and comparison with tin ores.
(a) Isotope composition of the ingots without taking the pyrometallurgical fractionation into account. (a–f) Comparison of ingots with ores from the Erzgebirge province (b), the British Isles (c), the Iberian peninsula (d), Brittany, the French Massif Central, Egypt, Sardinia, Mount Bukulja and Monte Valerio (e) and central Asia (f). The horizontal bars represent the variation in the tin and are lowered by the value 0.1 ‰ as pyrometallurgical impact on the right hand-side (indicated by the arrow) to yield the estimated original isotope composition of the ingots (cf. [69]). The colours of the bars correspond to the colours of the symbols used for the tin ingots. The numbering for samples in (a) corresponds to the sequence in Tables 6 and S2. Legend applies to all diagrams (diagrams: D. Berger; data: G. Brügmann, to be published numerically in the PhD thesis of J. Marahrens).
The two ingots from Haifa have identical δ124Sn values within analytical uncertainties amounting to 0.15 ± 0.01 ‰ and 0.17 ± 0.02 ‰, respectively. Thus, these two ingots appear in fact to have belonged to a common find and archaeological context and were probably cast from the same metal batch as already suggested by their similar chemical and lead isotope compositions (Figs 12 and 15).
The isotopic composition of the fourteen ingots from Hishuley Carmel is on average identical with that of the Haifa samples (0.17 ± 0.08 ‰ vs. 0.16 ± 0.02 ‰). There is, however, a large variation of overall 0.15 ‰, and individual δ124Sn values range from 0.10 ± 0.03 ‰ to 0.25 ± 0.02 ‰ (Fig 14A and Tables 6 and S2).
Fig 15. Tin isotope composition (δ124Sn) of the tin ingots vs. the concentration of trace elements determined in this study.
The dotted lines and specified values represent the limit of detection of the Q-ICP-MS. Legend applies to all diagrams (diagrams: D. Berger; data: G. Brügmann, L. Lockhoff).
The δ124Sn values fall into all three isotope groups defined above, but with the exception of one sample they have high and medium high isotopic ratios greater than 0.12 ‰ (Fig 14A). All data taken together, distinct relationships with trace elements cannot be recognised (Fig 15). However, the groups of ingots with the heaviest and the intermediate isotope compositions tend to follow two parallel trends with negative slopes for manganese, iron, indium, tungsten, bismuth and lead, which are particularly well defined with the latter with R = –0.90 and –0.80, respectively (Fig 15). These trends could imply that with increasing tin isotope ratios the metal concentrations decrease. This can occur when tin metal experiences a kind of refining process during which the metal is melted under oxidising conditions [96; 98]. Depending on the applied temperature, tin isotope ratios will increase during repeated melting and casting processes because of the possible loss of tin vapours or due to dross formation. At the same time, tin metal will oust elements as dross (mostly oxides), which are not completely soluble within its crystal lattice (e.g. elements forming intermetallics such as Fe and Mn) or which are less noble. Regardless of whether such operations were actually carried out and how they worked in detail, at least three different ore charges must have been used to produce these ingots. Their origin could be different tin sources, but they could as well stem from different mines or locations of the same deposit. In this regard, the two Haifa ingots were probably made from tin of the same mine as the Hishuley Carmel ingots with the intermediate isotopic composition as suggested by similar tin isotope ratios and trace element patterns (Fig 15).
As shown by Galili and colleagues [26] the ingots of Hishuley Carmel have very different physical shapes varying from bar-like and round, hemispherical to bun-shaped (cf. Fig 4). The two bar-like and the two round shaped ingots each have very similar tin and lead isotope compositions (Fig 16). The two types of ingots have significantly different tin isotopic compositions where the bar-shaped ingots were produced from ores with heavy isotope composition, whereas round ingots derived from ores with the medium isotope composition. Although trace element concentration such as antimony, lead and indium differ among the two ingot types, this difference could be due to the variable depletion of these elements in the tin ingot depending on the temperature and redox conditions during the casting process as outlined above. However, hemispherically- or bun-shaped ingots have variable tin and lead isotopic compositions. They do not indicate a systemic relationship to any of the three groups defined by the tin isotope ratios (Fig 16) which implies that each ingot type was produced from a different charge of different tin ores. Hence, the physical shape of the ingots appears to contain no provenance information provided the recognised sample confusion does not carry weight here (all confused samples are accounted for).
Fig 16. Tin and lead isotope composition of the different shaped ingots from Hishuley Carmel and their relationship with the indium contents.
Note the strong correlation of the lead isotopes and indium (R = 0.74). Legend applies to both diagrams (diagram: D. Berger; data: G. Brügmann, B. Höppner, N. Lockhoff).
The conclusions from the tin isotopes for the Greek and the Israeli ingots cannot be transferred to the Uluburun ingots. The artefacts from the wrecked Turkish ship are (almost) completely corroded as well, yet what distinguishes them from the Cretan artefact is their burial context: all items corroded in seawater instead of terrestrial soil. This led to the formation of a different corrosion assemblage consisting predominantly of abhurite (Sn21O6(OH)14Cl16) along with romarchite and stannic oxide (Figs 7B and 10). Although a detailed explanation is lacking at present, most of the corroded artefacts from seawater contexts analysed so far show enrichment in heavy tin isotopes between Δ124Sn = 0.05 ‰ and 0.30 ‰ relative to the uncorroded metal [81]. Isotope discrimination in seawater thus seems to be more pronounced than in soil. Although we cannot verify such a fractionation for the Uluburun ingots, because we could not find and analyse a reasonable amount of preserved tin metal, their extremely heavy isotope compositions ranging from of 0.27 ± 0.02 ‰ to 0.42 ± 0.01 ‰ (Fig 14A, nos. 26–29) suggest that the isotope ratios were altered during the corrosion process and do not reflect the original composition of the tin. This conclusion is supported by the significantly different isotopic compositions of two samples from ingot KW 203 taken from its corroded core (FG-883210) and the surface (FG-883211) (Fig 14A, nos. 28–29). This difference might be explained by the formation of tin corrosion products with different oxidation states, as already suggested for the Mochlos tin (Fig 7B). Because of all these findings, it is not possible to source the tin of the actual Uluburun ingots with the aid of tin isotopes. Fresh material is needed in order to overcome this problem.
However, the source of the tin in the remaining ingots can be discussed further as long as the change in the isotope ratios during the smelting process is taken into account. Smelting experiments with cassiterite recently carried out under prehistoric conditions demonstrated the loss of volatile tin species to induce a fractionation of tin isotopes. Relative to the ore, this caused heavier δ124Sn values in the tin metal by about 0.1 ‰ if 30% of the tin was recovered as tin metal [69]. We therefore have to consider this isotopic shift as ‘metallurgical impact’ when comparing metal artefacts with tin ores. Nevertheless, the experimental findings provide us with an important tool, namely that ores with heavier isotope compositions (in our case with higher δ124Sn ratios) than bronze or tin artefacts can be excluded as parental tin sources. The comparison of objects with each other is not significantly affected since isotope fractionation by smelting can be assumed to be more or less the same for all.
Fig 14B and 14F compare the ‘original’ isotope composition of the ingots (indicated as coloured bars)–i.e. the measured isotope composition minus the metallurgical impact of ca. 0.1 ‰ –with that of tin ores throughout Eurasia hitherto available from our database. Cassiterite from placers and lodes from the Erzgebirge region (Erzgebirge, Vogtland, Fichtelgebirge, Kaiserwald (Slavkovský les)), United Kingdom (Cornwall, Devon, Mourne Mountains), Brittany, the Massif Central, the Iberian peninsula (Portugal, Spain), Tuscany (Monte Valerio), Sardinia, Serbia (Mount Bukulja), Egypt, east Afghanistan (Hindu Kush), Uzbekistan (Lapas), Tadzhikistan (Mušiston, Pamir mountains), Kyrgyzstan and Kazhakstan are considered there. The figures clearly illustrate the principle difficulty in tin provenancing. Overall, the isotopic composition of cassiterite from the largest European tin provinces, namely Cornwall/Devon, the Saxon-Bohemian Erzgebirge and the Iberian peninsula, exhibits a wide variation and a significant overlap. The picture is somewhat differentiated when looking at single regions within these large tin provinces or even at single mines. Smaller variations and differing intervals can be recognised, which also applies to minor tin deposits such as those on Sardinia, at Monte Valerio, at Mount Bukulja or in the Mourne Mountains (North Ireland). The situation seems to be more promising with central Asian ores, but here our database is still lacking data from some major regions (e.g. Uzbekistan, Kazakhstan, Kyrgyzstan). For example, Afghan cassiterites from the Hindu Kush tend to be isotopically lighter compared for instance with Tadzhik tin ores from the Pamir and the data overlap very little (Fig 15F).
Given that, a positive assignment of a tin or bronze artefact to a specific tin mineralisation is not possible because of data overlap. The geographical location of the initial tin source can only be approximated by excluding those ore bodies whose isotope compositions differ from that of the artefacts. In the case of the Mochlos ingot this means that deposits having δ124Sn ratios of greater than 0 ‰, such as those from the French Massif Central and the smaller deposits of Monte Valerio, Sardinia, Mount Bukulja and the Mourne Mountains can be excluded if we consider a fractionation of Δ124Sn = 0.1% (for 30% tin recovery) during the smelting process (Fig 15E). Most of these deposits were made less likely because of their formation age (see above) or their ineptitude for Bronze Age tin exploitation [43; 45]. For the same reason, the other European mineralisations from the Erzgebirge, Cornwall/Devon, the Iberian peninsula and Brittany as well as those from Egypt would be excluded. Of the remaining possible sources, we do not have tin isotope data from Deh Hosein, but, as explained above, the polymetallic character of this deposit speaks against its use for the Mochlos tin. We therefore consider the central Asian tin deposits in the eastern part of Afghanistan and in Tadzhikistan as the most reasonable sources for the tin from Mochlos, not least because their tin isotope systematics (that we determined so far) agree well with the that of the ingot. In Tadzhikistan, cassiterite from Takfon (near the Mušiston deposit) and from Ghilnoye in the Pamir mountains shows the best match within our dataset (Fig 15F). One sample from Kyrgyzstan also shows a match, but we have just one data point that is not statistically significant. There are also more tin sources along the borderline of the Herat and Farah provinces in West Afghanistan and in the central Afghan Daykundi province [54; 56–57] from which we had no material to be analysed.
All of the sources for the Haifa tin could have been used for the production of the ingots from Hishuley Carmel and Kfar Samir. If one assumes a common origin for the ingots in one cargo because they were found very close together at one site, this would exclude tin deposits that do not completely cover the range of isotope ratios observed (Land’s End granite in Cornwall, the Castelo Branco region in Portugal or Cerro de San Cristóbal, Logrosán in Spain [140–141]; Fig 15 and Table 7). However, it is possible that tin was collected from various mining areas in large centres or trading ports such as Ugarit, Byblos, Tyre, Cyprus, Cornwall etc. [18], and thus individual ships transported cargoes containing tin ingots of mixed provenance. The different groups and shapes of ingots observed at Hishuley Carmel and Kfar Samir might reflect this situation. In this case, the different isotopic groups have to be discussed separately and the interpretation is further complicated when the tin isotope composition is considered alone. Yet, by including the trace element patterns of the Mediterranean tin ingots, the potential sources can be confined further. Because the elemental composition is quite similar to those of the Salcombe ingots (Fig 8), and the latter were certainly made from Cornish or Devonian tin ores [94], a British provenance of the tin from Israel is currently the most reasonable. The comparably high indium concentration in the ingots that is a typical feature of Cornish cassiterites might be the most helpful indication. On the other hand, the very low indium contents of the Uluburun ingots as well as their differing elemental pattern in general indicate a source of ore other than southwest England. It would be tempting to infer geologically very old tin deposits (such as Egypt or India) from the existing lead isotope data (literature data and own data; Fig 12A, grey and yellow dots), but because of the corroded nature of the objects, neither the chemical data nor the lead and tin isotopic compositions provide any reliable information on their provenance. For the Israeli ingots the situation is much better, and all parameters taken together indicate that the ingots (except for the isotopically lowest group of Kfar Samir south) were produced from tin ores of the same deposit, albeit different ore charges or ores from different mines were certainly used. Given this and the proximity of the sites, it even seems possible that the suspected ships belonged to a fleet that sank for example during a single storm event.
Fig 14A and Tables 6 and S2 summarise the tin isotope compositions of the tin ingots. Overall, only positive δ124Sn values (relative to our in-house standard) are observed that spread over a large range between 0.05 ± 0.03 ‰ and 0.42 ± 0.01 ‰. Individual sites, however, show a much smaller variation, and differences between them become discernible. The ingots can be divided into three groups (Fig 14A): one with high δ124Sn values of greater than 0.19 ‰, one with medium δ124Sn values ranging from about 0.12 to 0.18 ‰, and a third one with low δ124Sn values of less than 0.12 ‰. The large variation of the tin isotope ratios implies that the different groups of ingots–even within the same archaeological context–were produced from different tin ore charges (for comparison of actual data with older analyses see S1 File). Compared with trace elements, no distinct relationships are recognisable overall, but individual sites or groups of ingots with smaller variations in the tin isotopes reveal relationships with trace elements such as iron, antimony, tungsten, lead, bismuth, and indium (Fig 15).
Fig 14. Tin isotope composition (δ124Sn) of the tin ingots examined in this study and comparison with tin ores.
(a) Isotope composition of the ingots without taking the pyrometallurgical fractionation into account. (a–f) Comparison of ingots with ores from the Erzgebirge province (b), the British Isles (c), the Iberian peninsula (d), Brittany, the French Massif Central, Egypt, Sardinia, Mount Bukulja and Monte Valerio (e) and central Asia (f). The horizontal bars represent the variation in the tin and are lowered by the value 0.1 ‰ as pyrometallurgical impact on the right hand-side (indicated by the arrow) to yield the estimated original isotope composition of the ingots (cf. [69]). The colours of the bars correspond to the colours of the symbols used for the tin ingots. The numbering for samples in (a) corresponds to the sequence in Tables 6 and S2. Legend applies to all diagrams (diagrams: D. Berger; data: G. Brügmann, to be published numerically in the PhD thesis of J. Marahrens).
The two ingots from Haifa have identical δ124Sn values within analytical uncertainties amounting to 0.15 ± 0.01 ‰ and 0.17 ± 0.02 ‰, respectively. Thus, these two ingots appear in fact to have belonged to a common find and archaeological context and were probably cast from the same metal batch as already suggested by their similar chemical and lead isotope compositions (Figs 12 and 15).
The isotopic composition of the fourteen ingots from Hishuley Carmel is on average identical with that of the Haifa samples (0.17 ± 0.08 ‰ vs. 0.16 ± 0.02 ‰). There is, however, a large variation of overall 0.15 ‰, and individual δ124Sn values range from 0.10 ± 0.03 ‰ to 0.25 ± 0.02 ‰ (Fig 14A and Tables 6 and S2).
Fig 15. Tin isotope composition (δ124Sn) of the tin ingots vs. the concentration of trace elements determined in this study.
The dotted lines and specified values represent the limit of detection of the Q-ICP-MS. Legend applies to all diagrams (diagrams: D. Berger; data: G. Brügmann, L. Lockhoff).
The δ124Sn values fall into all three isotope groups defined above, but with the exception of one sample they have high and medium high isotopic ratios greater than 0.12 ‰ (Fig 14A). All data taken together, distinct relationships with trace elements cannot be recognised (Fig 15). However, the groups of ingots with the heaviest and the intermediate isotope compositions tend to follow two parallel trends with negative slopes for manganese, iron, indium, tungsten, bismuth and lead, which are particularly well defined with the latter with R = –0.90 and –0.80, respectively (Fig 15). These trends could imply that with increasing tin isotope ratios the metal concentrations decrease. This can occur when tin metal experiences a kind of refining process during which the metal is melted under oxidising conditions [96; 98]. Depending on the applied temperature, tin isotope ratios will increase during repeated melting and casting processes because of the possible loss of tin vapours or due to dross formation. At the same time, tin metal will oust elements as dross (mostly oxides), which are not completely soluble within its crystal lattice (e.g. elements forming intermetallics such as Fe and Mn) or which are less noble. Regardless of whether such operations were actually carried out and how they worked in detail, at least three different ore charges must have been used to produce these ingots. Their origin could be different tin sources, but they could as well stem from different mines or locations of the same deposit. In this regard, the two Haifa ingots were probably made from tin of the same mine as the Hishuley Carmel ingots with the intermediate isotopic composition as suggested by similar tin isotope ratios and trace element patterns (Fig 15).
As shown by Galili and colleagues [26] the ingots of Hishuley Carmel have very different physical shapes varying from bar-like and round, hemispherical to bun-shaped (cf. Fig 4). The two bar-like and the two round shaped ingots each have very similar tin and lead isotope compositions (Fig 16). The two types of ingots have significantly different tin isotopic compositions where the bar-shaped ingots were produced from ores with heavy isotope composition, whereas round ingots derived from ores with the medium isotope composition. Although trace element concentration such as antimony, lead and indium differ among the two ingot types, this difference could be due to the variable depletion of these elements in the tin ingot depending on the temperature and redox conditions during the casting process as outlined above. However, hemispherically- or bun-shaped ingots have variable tin and lead isotopic compositions. They do not indicate a systemic relationship to any of the three groups defined by the tin isotope ratios (Fig 16) which implies that each ingot type was produced from a different charge of different tin ores. Hence, the physical shape of the ingots appears to contain no provenance information provided the recognised sample confusion does not carry weight here (all confused samples are accounted for).
Fig 16. Tin and lead isotope composition of the different shaped ingots from Hishuley Carmel and their relationship with the indium contents.
Note the strong correlation of the lead isotopes and indium (R = 0.74). Legend applies to both diagrams (diagram: D. Berger; data: G. Brügmann, B. Höppner, N. Lockhoff).
The conclusions from the tin isotopes for the Greek and the Israeli ingots cannot be transferred to the Uluburun ingots. The artefacts from the wrecked Turkish ship are (almost) completely corroded as well, yet what distinguishes them from the Cretan artefact is their burial context: all items corroded in seawater instead of terrestrial soil. This led to the formation of a different corrosion assemblage consisting predominantly of abhurite (Sn21O6(OH)14Cl16) along with romarchite and stannic oxide (Figs 7B and 10). Although a detailed explanation is lacking at present, most of the corroded artefacts from seawater contexts analysed so far show enrichment in heavy tin isotopes between Δ124Sn = 0.05 ‰ and 0.30 ‰ relative to the uncorroded metal [81]. Isotope discrimination in seawater thus seems to be more pronounced than in soil. Although we cannot verify such a fractionation for the Uluburun ingots, because we could not find and analyse a reasonable amount of preserved tin metal, their extremely heavy isotope compositions ranging from of 0.27 ± 0.02 ‰ to 0.42 ± 0.01 ‰ (Fig 14A, nos. 26–29) suggest that the isotope ratios were altered during the corrosion process and do not reflect the original composition of the tin. This conclusion is supported by the significantly different isotopic compositions of two samples from ingot KW 203 taken from its corroded core (FG-883210) and the surface (FG-883211) (Fig 14A, nos. 28–29). This difference might be explained by the formation of tin corrosion products with different oxidation states, as already suggested for the Mochlos tin (Fig 7B). Because of all these findings, it is not possible to source the tin of the actual Uluburun ingots with the aid of tin isotopes. Fresh material is needed in order to overcome this problem.
However, the source of the tin in the remaining ingots can be discussed further as long as the change in the isotope ratios during the smelting process is taken into account. Smelting experiments with cassiterite recently carried out under prehistoric conditions demonstrated the loss of volatile tin species to induce a fractionation of tin isotopes. Relative to the ore, this caused heavier δ124Sn values in the tin metal by about 0.1 ‰ if 30% of the tin was recovered as tin metal [69]. We therefore have to consider this isotopic shift as ‘metallurgical impact’ when comparing metal artefacts with tin ores. Nevertheless, the experimental findings provide us with an important tool, namely that ores with heavier isotope compositions (in our case with higher δ124Sn ratios) than bronze or tin artefacts can be excluded as parental tin sources. The comparison of objects with each other is not significantly affected since isotope fractionation by smelting can be assumed to be more or less the same for all.
Fig 14B and 14F compare the ‘original’ isotope composition of the ingots (indicated as coloured bars)–i.e. the measured isotope composition minus the metallurgical impact of ca. 0.1 ‰ –with that of tin ores throughout Eurasia hitherto available from our database. Cassiterite from placers and lodes from the Erzgebirge region (Erzgebirge, Vogtland, Fichtelgebirge, Kaiserwald (Slavkovský les)), United Kingdom (Cornwall, Devon, Mourne Mountains), Brittany, the Massif Central, the Iberian peninsula (Portugal, Spain), Tuscany (Monte Valerio), Sardinia, Serbia (Mount Bukulja), Egypt, east Afghanistan (Hindu Kush), Uzbekistan (Lapas), Tadzhikistan (Mušiston, Pamir mountains), Kyrgyzstan and Kazhakstan are considered there. The figures clearly illustrate the principle difficulty in tin provenancing. Overall, the isotopic composition of cassiterite from the largest European tin provinces, namely Cornwall/Devon, the Saxon-Bohemian Erzgebirge and the Iberian peninsula, exhibits a wide variation and a significant overlap. The picture is somewhat differentiated when looking at single regions within these large tin provinces or even at single mines. Smaller variations and differing intervals can be recognised, which also applies to minor tin deposits such as those on Sardinia, at Monte Valerio, at Mount Bukulja or in the Mourne Mountains (North Ireland). The situation seems to be more promising with central Asian ores, but here our database is still lacking data from some major regions (e.g. Uzbekistan, Kazakhstan, Kyrgyzstan). For example, Afghan cassiterites from the Hindu Kush tend to be isotopically lighter compared for instance with Tadzhik tin ores from the Pamir and the data overlap very little (Fig 15F).
Given that, a positive assignment of a tin or bronze artefact to a specific tin mineralisation is not possible because of data overlap. The geographical location of the initial tin source can only be approximated by excluding those ore bodies whose isotope compositions differ from that of the artefacts. In the case of the Mochlos ingot this means that deposits having δ124Sn ratios of greater than 0 ‰, such as those from the French Massif Central and the smaller deposits of Monte Valerio, Sardinia, Mount Bukulja and the Mourne Mountains can be excluded if we consider a fractionation of Δ124Sn = 0.1% (for 30% tin recovery) during the smelting process (Fig 15E). Most of these deposits were made less likely because of their formation age (see above) or their ineptitude for Bronze Age tin exploitation [43; 45]. For the same reason, the other European mineralisations from the Erzgebirge, Cornwall/Devon, the Iberian peninsula and Brittany as well as those from Egypt would be excluded. Of the remaining possible sources, we do not have tin isotope data from Deh Hosein, but, as explained above, the polymetallic character of this deposit speaks against its use for the Mochlos tin. We therefore consider the central Asian tin deposits in the eastern part of Afghanistan and in Tadzhikistan as the most reasonable sources for the tin from Mochlos, not least because their tin isotope systematics (that we determined so far) agree well with the that of the ingot. In Tadzhikistan, cassiterite from Takfon (near the Mušiston deposit) and from Ghilnoye in the Pamir mountains shows the best match within our dataset (Fig 15F). One sample from Kyrgyzstan also shows a match, but we have just one data point that is not statistically significant. There are also more tin sources along the borderline of the Herat and Farah provinces in West Afghanistan and in the central Afghan Daykundi province [54; 56–57] from which we had no material to be analysed.
All of the sources for the Haifa tin could have been used for the production of the ingots from Hishuley Carmel and Kfar Samir. If one assumes a common origin for the ingots in one cargo because they were found very close together at one site, this would exclude tin deposits that do not completely cover the range of isotope ratios observed (Land’s End granite in Cornwall, the Castelo Branco region in Portugal or Cerro de San Cristóbal, Logrosán in Spain [140–141]; Fig 15 and Table 7). However, it is possible that tin was collected from various mining areas in large centres or trading ports such as Ugarit, Byblos, Tyre, Cyprus, Cornwall etc. [18], and thus individual ships transported cargoes containing tin ingots of mixed provenance. The different groups and shapes of ingots observed at Hishuley Carmel and Kfar Samir might reflect this situation. In this case, the different isotopic groups have to be discussed separately and the interpretation is further complicated when the tin isotope composition is considered alone. Yet, by including the trace element patterns of the Mediterranean tin ingots, the potential sources can be confined further. Because the elemental composition is quite similar to those of the Salcombe ingots (Fig 8), and the latter were certainly made from Cornish or Devonian tin ores [94], a British provenance of the tin from Israel is currently the most reasonable. The comparably high indium concentration in the ingots that is a typical feature of Cornish cassiterites might be the most helpful indication. On the other hand, the very low indium contents of the Uluburun ingots as well as their differing elemental pattern in general indicate a source of ore other than southwest England. It would be tempting to infer geologically very old tin deposits (such as Egypt or India) from the existing lead isotope data (literature data and own data; Fig 12A, grey and yellow dots), but because of the corroded nature of the objects, neither the chemical data nor the lead and tin isotopic compositions provide any reliable information on their provenance. For the Israeli ingots the situation is much better, and all parameters taken together indicate that the ingots (except for the isotopically lowest group of Kfar Samir south) were produced from tin ores of the same deposit, albeit different ore charges or ores from different mines were certainly used. Given this and the proximity of the sites, it even seems possible that the suspected ships belonged to a fleet that sank for example during a single storm event.