Mafic outcrops of the southern Levant 

Contexts in source publication

Context 1

... order to extend the studies of Philip and Williams-Thorpe it was necessary to expand their data-set by collecting additional geological and archaeological samples, and analysing them for a wider range of trace elements, using ICP-MS. One of us (GPR) was able to field-collect additional geological and archaeological samples, whilst the database of geological samples was expanded through the collation of analyses reported by other 7,12,13 studies. Four geological and three artefactual samples reported by Philip and 8,9 Williams-Thorpe which had been analysed using XRF were reanalysed using ICP-MS, in order to assess the comparability of the data-sets, which were found to be in good 14 agreement. The geological samples collected for this study were taken from a number of outcrops along the Dead Sea Rift, from which there were either no or very few previously reported geochemical analyses. The artefactual samples were obtained from a set of sites currently under excavation, and for which reliable contextual data should be forthcoming; sites were selected to provide a broad geographical spread throughout the southern Levant. All these samples were analysed via the quadrupole ELAN 6000 ICP-MS at the Department of Geological Sciences, Durham. Samples were prepared and analysed using a routine technique developed at Durham for igneous rocks. Digestion, dilution and analytical protocols ensured that the data generated were of high quality, with reproducibility of key 15 elements and element ratios better than 5% at two standard deviations. One problem that was encountered when preparing the samples for analysis was that it was not generally possible to remove the weathered sections from the archaeological samples, given their generally small size. This introduced the prospect that it would not be possible to match artefacts to their original source, if post-manufacture weathering had altered their trace element concentrations. To assess this problem, a number of weathered sections from geological samples were analysed, as was one artefact which had a visible weathering rind. To further counteract any problems with weathering, we concentrated our attention on elements known to be relatively immobile during normal geological weathering conditions, 16 including the rare earth elements (REE). When the analyses of the unweathered samples are plotted with the analyses of their weathered counterparts (Figure 2) it can be seen that the amount of variation in the REE is negligible, being less than the variation due to analytical error. This was also the case for the high field strength elements (HFSE). Therefore only these ‘immobile’ elements were used to attempt to provenance the artefacts. This further demonstrates the usefulness of ICP-MS, which enables the low levels of REE and HFSE present in most mafic rocks to be measured at high precision. This is especially 17 the case as only a limited number of REE can be routinely determined by XRF. While elemental abundance plots, especially of REE, are useful to construct, they are subject to the affects of fractional crystallisation and the potential for crystal segregation on an individual outcrop scale. This means that elemental abundances may vary within a single outcrop, merely due to the effects of differences in the modal abundances of crystals (especially phenocrysts) within a given lava flow. Although fractional crystallisation effects are probably not a major problem in aphanitic basaltic lavas it can affect the elemental abundances, it is good policy to rule out any potential problems. This can be easily accomplished by using trace element ratios that are not normally disturbed during fractional crystallisation and crystal accumulation processes. Elemental ratios were therefore used instead of abundance plots (Figure 3). This plot utilises a number of HFSE 16 ratios and it can be seen that the artefacts form two main groups. This pattern is replicated when using REE element ratios (see Figures 4 and 5), with this plot usually 16 being used by geochemists to show the amount of REE fractionation. As part of the on- going research, these groups will also be quantified using statistical techniques, which can show the amount of variability within and between sources, but the use of two independent plots shows that these clusters are valid. When the Chalcolithic and Early Bronze artefacts are plotted separately (Figure 4) it can be seen that there is more clustering of the artefacts than in the Late Bronze and Iron Age artefacts (Figure 5) showing that a greater diversity of sources were being used to manufacture the artefacts in the later periods. Despite a 1,500 year gap between the end of the Early Bronze I and the start of the Late Bronze Age it is notable that most of the artefacts still form two distinct clusters, showing that the same two main sources were being exploited. Although these fractionation plots are capable of ruling out many of the outcrops as potential sources it was not possible to determine exactly which of a number of outcrops actually provided the material for the artefacts. In an attempt to constrain this further, the REE and HFSE of the potential source outcrop were normalised to the artefact sample and then plotted (see Figure 6 for an example of such a plot). These elements were selected as being the most immobile, even after weathering. A perfect match between an outcrop sample and an artefact would produce a completely flat line. This approach also enables the relative standard deviation between the composition of the artefact and that of the outcrop sample to be calculated. For many of the samples an RSD of less than 5% was obtained, which was taken as a positive indication that the outcrop was the source of the artefact in question. The RSD of the samples shown in Figure 6 are presented in Table 2. When this procedure was repeated for all of the artefacts it was possible to positively identify the source of most of the artefacts, with the two main clusters correlating with outcrops from the North Jordan Valley and, more surprisingly, from Mount Hermon (Figure 1). Philip and Williams-Thorpe had already identified the North Jordan Valley 9 outcrops as a probable source for outcrops, but Mount Hermon has not been previously identified as a potential manufacturing centre for mafic artefacts. Furthermore, the continued exploitation of these two sources for most of the analysed artefacts manufactured in all the periods studied is also surprising and suggests that there were good, technological reasons for choosing these particular outcrops. This therefore requires further work with a larger sample of artefacts and strengthens the argument that the mechanical properties of these mafic rocks need to be fully quantified before these 14 procurement systems can be fully understood. This study therefore reveals a more complex picture of the history of the region than was previously thought, and thereby demonstrates the usefulness of ICP-MS and the importance of provenance studies in archaeology. It is notable that none of the artefacts appear to come from the Jaulan outcrops, despite 18 their extensive nature, and the fact that there was a flourishing society in the region. This observation confirms what is already known about this society. One omission from the current study is any artefacts from the Jaulan itself, but it is hoped that samples will soon be analysed from this area. One problem with the current study is that there are still insufficient geological samples from a number of the outcrops, especially east and south of the Dead Sea (Figure 1). This has meant that it has not been possible to match all of the artefacts to outcrop samples with less than 5% relative standard deviation (i.e. at 95% level of confidence). Therefore, to extend this study, it will be first necessary to gather more geological samples. It will also be necessary to examine a greater number of artefacts, from a greater spatial and temporal range of sites. This will enable these provisional conclusions to be tested and will also provide a more complete picture of how past procurement systems may have operated in the southern Levant. However, these results have already revealed a complex picture of the procurement of mafic artefacts in the past, which has consequently changed our understanding of how these procurement systems operated. Nonetheless, some explanation of how these artefacts 14 were procured is still required. The limited number of sources indicates that the artefacts were very probably produced by specialists and were possibly used as some form of status symbol, especially as copies in other stones and pottery have also been found. A further problem with understanding the production of mafic artefacts is that currently no quarries or manufacturing sites are known. The archaeological examination of such sites is essential, as they provide important information about the organisation of the economy and 19 the way in which the society operated. The reason that these sites have not yet been located is in part due to the inability to pinpoint the source outcrops sufficiently to enable a focused search of a limited area. This may now well be possible and hopefully further work will enable the search to be focused even ...

Context 2

... order to extend the studies of Philip and Williams-Thorpe it was necessary to expand their data-set by collecting additional geological and archaeological samples, and analysing them for a wider range of trace elements, using ICP-MS. One of us (GPR) was able to field-collect additional geological and archaeological samples, whilst the database of geological samples was expanded through the collation of analyses reported by other 7,12,13 studies. Four geological and three artefactual samples reported by Philip and 8,9 Williams-Thorpe which had been analysed using XRF were reanalysed using ICP-MS, in order to assess the comparability of the data-sets, which were found to be in good 14 agreement. The geological samples collected for this study were taken from a number of outcrops along the Dead Sea Rift, from which there were either no or very few previously reported geochemical analyses. The artefactual samples were obtained from a set of sites currently under excavation, and for which reliable contextual data should be forthcoming; sites were selected to provide a broad geographical spread throughout the southern Levant. All these samples were analysed via the quadrupole ELAN 6000 ICP-MS at the Department of Geological Sciences, Durham. Samples were prepared and analysed using a routine technique developed at Durham for igneous rocks. Digestion, dilution and analytical protocols ensured that the data generated were of high quality, with reproducibility of key 15 elements and element ratios better than 5% at two standard deviations. One problem that was encountered when preparing the samples for analysis was that it was not generally possible to remove the weathered sections from the archaeological samples, given their generally small size. This introduced the prospect that it would not be possible to match artefacts to their original source, if post-manufacture weathering had altered their trace element concentrations. To assess this problem, a number of weathered sections from geological samples were analysed, as was one artefact which had a visible weathering rind. To further counteract any problems with weathering, we concentrated our attention on elements known to be relatively immobile during normal geological weathering conditions, 16 including the rare earth elements (REE). When the analyses of the unweathered samples are plotted with the analyses of their weathered counterparts (Figure 2) it can be seen that the amount of variation in the REE is negligible, being less than the variation due to analytical error. This was also the case for the high field strength elements (HFSE). Therefore only these ‘immobile’ elements were used to attempt to provenance the artefacts. This further demonstrates the usefulness of ICP-MS, which enables the low levels of REE and HFSE present in most mafic rocks to be measured at high precision. This is especially 17 the case as only a limited number of REE can be routinely determined by XRF. While elemental abundance plots, especially of REE, are useful to construct, they are subject to the affects of fractional crystallisation and the potential for crystal segregation on an individual outcrop scale. This means that elemental abundances may vary within a single outcrop, merely due to the effects of differences in the modal abundances of crystals (especially phenocrysts) within a given lava flow. Although fractional crystallisation effects are probably not a major problem in aphanitic basaltic lavas it can affect the elemental abundances, it is good policy to rule out any potential problems. This can be easily accomplished by using trace element ratios that are not normally disturbed during fractional crystallisation and crystal accumulation processes. Elemental ratios were therefore used instead of abundance plots (Figure 3). This plot utilises a number of HFSE 16 ratios and it can be seen that the artefacts form two main groups. This pattern is replicated when using REE element ratios (see Figures 4 and 5), with this plot usually 16 being used by geochemists to show the amount of REE fractionation. As part of the on- going research, these groups will also be quantified using statistical techniques, which can show the amount of variability within and between sources, but the use of two independent plots shows that these clusters are valid. When the Chalcolithic and Early Bronze artefacts are plotted separately (Figure 4) it can be seen that there is more clustering of the artefacts than in the Late Bronze and Iron Age artefacts (Figure 5) showing that a greater diversity of sources were being used to manufacture the artefacts in the later periods. Despite a 1,500 year gap between the end of the Early Bronze I and the start of the Late Bronze Age it is notable that most of the artefacts still form two distinct clusters, showing that the same two main sources were being exploited. Although these fractionation plots are capable of ruling out many of the outcrops as potential sources it was not possible to determine exactly which of a number of outcrops actually provided the material for the artefacts. In an attempt to constrain this further, the REE and HFSE of the potential source outcrop were normalised to the artefact sample and then plotted (see Figure 6 for an example of such a plot). These elements were selected as being the most immobile, even after weathering. A perfect match between an outcrop sample and an artefact would produce a completely flat line. This approach also enables the relative standard deviation between the composition of the artefact and that of the outcrop sample to be calculated. For many of the samples an RSD of less than 5% was obtained, which was taken as a positive indication that the outcrop was the source of the artefact in question. The RSD of the samples shown in Figure 6 are presented in Table 2. When this procedure was repeated for all of the artefacts it was possible to positively identify the source of most of the artefacts, with the two main clusters correlating with outcrops from the North Jordan Valley and, more surprisingly, from Mount Hermon (Figure 1). Philip and Williams-Thorpe had already identified the North Jordan Valley 9 outcrops as a probable source for outcrops, but Mount Hermon has not been previously identified as a potential manufacturing centre for mafic artefacts. Furthermore, the continued exploitation of these two sources for most of the analysed artefacts manufactured in all the periods studied is also surprising and suggests that there were good, technological reasons for choosing these particular outcrops. This therefore requires further work with a larger sample of artefacts and strengthens the argument that the mechanical properties of these mafic rocks need to be fully quantified before these 14 procurement systems can be fully understood. This study therefore reveals a more complex picture of the history of the region than was previously thought, and thereby demonstrates the usefulness of ICP-MS and the importance of provenance studies in archaeology. It is notable that none of the artefacts appear to come from the Jaulan outcrops, despite 18 their extensive nature, and the fact that there was a flourishing society in the region. This observation confirms what is already known about this society. One omission from the current study is any artefacts from the Jaulan itself, but it is hoped that samples will soon be analysed from this area. One problem with the current study is that there are still insufficient geological samples from a number of the outcrops, especially east and south of the Dead Sea (Figure 1). This has meant that it has not been possible to match all of the artefacts to outcrop samples with less than 5% relative standard deviation (i.e. at 95% level of confidence). Therefore, to extend this study, it will be first necessary to gather more geological samples. It will also be necessary to examine a greater number of artefacts, from a greater spatial and temporal range of sites. This will enable these provisional conclusions to be tested and will also provide a more complete picture of how past procurement systems may have operated in the southern Levant. However, these results have already revealed a complex picture of the procurement of mafic artefacts in the past, which has consequently changed our understanding of how these procurement systems operated. Nonetheless, some explanation of how these artefacts 14 were procured is still required. The limited number of sources indicates that the artefacts were very probably produced by specialists and were possibly used as some form of status symbol, especially as copies in other stones and pottery have also been found. A further problem with understanding the production of mafic artefacts is that currently no quarries or manufacturing sites are known. The archaeological examination of such sites is essential, as they provide important information about the organisation of the economy and 19 the way in which the society operated. The reason that these sites have not yet been located is in part due to the inability to pinpoint the source outcrops sufficiently to enable a focused search of a limited area. This may now well be possible and hopefully further work will enable the search to be focused even ...

Context 3

... the southern Levant (Israel, Jordan and the Occupied Territories) mafic rocks were used in the manufacture of a wide variety of artefacts, including statues and bowls as well as more utilitarian artefacts such as grinders. Whereas igneous rocks are only located in certain parts of the southern Levant ( Figure 1, below), these artefacts are found on virtually every studied site in the region. Clearly, these artefacts must have been transported up to several hundred kilometres from their source outcrop. This is of great archaeological interest due to the potential constraints offered to questions relating to inter- 1 group contacts and how past societies operated and were organised. Artefacts were examined from the range of periods shown in Table 1. A large variety of goods were acquired over long distances, but are generally not amenable to provenancing. Textiles, spices and oils were widely distributed, but have usually perished. There have been a number of attempts to provenance metals, but these have met 1 with problems due to the potential for the mixing of sources. Artefacts made from rock have a far greater potential for elucidating long distance trade networks, as they are relatively common, virtually indestructible and do not generally undergo chemical or 2 physical changes during their manufacture, use or subsequent deposition. Given these advantages, there have been a number of attempts to provenance the mafic artefacts of the southern Levant. All provenance studies rely on matching unique features of the 2 composition of the artefact with the unique features of a single potential source outcrop. The provenance studies reviewed below attempted to use a variety of supposedly unique features of the artefacts to determine their source, with varying degrees of ...

Context 4

... first attempts to provenance mafic stone artefacts used petrographic analysis, but with 3 only limited success. Amiran and Porat examined a small number of artefacts and were able to rule out the southern Cisjordan outcrops, but were unable to determine whether the source was the Galilee, Jaulan (the Golan), or other Transjordanian outcrops (Cisjordan refers to land west of the Dead Sea Rift, while Transjordan refers to land to the east) 4 ( Figure 1). Hunt petrographically analysed six artefacts from Late Bronze Age Hazor, situated in the basalt fields in the north of Israel. All the artefacts were made from fine- grained basalt which only contained olivine phenocrysts, unlike the local outcrops which contain both pyroxene and olivine phenocrysts. However, although this study was able to show that the source of the artefacts was non-local, it was not able to pinpoint the actual location. These two studies both show that there are not enough variations in the mineralogy between different outcrops in the southern Levant to enable a single outcrop to be definitely identified as the source of the artefact. Furthermore, virtually all the artefacts are made from aphanitic (fine-grained) basalt, making the identification of sources impossible without geochemical analysis. Even if this were not the case, a further limitation is the large sample size required to cut a thin-section, limiting the number of artefacts that could be analysed using this technique. A different approach was therefore pursued by Weinstien-Evron et al., who used K-Ar dating to provenance Natufian (c10,000-8,000 BC) artefacts from three sites in the Galilee area. The artefact samples dated from about 6.5 to less than 0.25 Mya, whilst none of the outcrops closest to any of the sites had rocks with this range of ages. The two nearest areas where rocks with this range of ages are in close proximity is north of the Sea of Galilee and directly south of the Sea of Galilee, approximately 100 km from the furthest site, despite 5 the fact that basalt was available from sources closer to the sites. The main limitation of this study is that it is not possible to determine which of the two potential locations was actually used. It is not even possible to determine whether artefacts were manufactured from several different outcrops, with similar ages. This could be resolved by petrographic examination, but given the problems with different outcrops being mineralogically similar, there is no guarantee that this would be the case. A further problem in using K-Ar dating is that some of the argon may be released during weathering, thereby giving erroneous dates, which is a serious problem with K-Ar dates in the southern 6,7 Levant. Given the potential for differential weathering between the geological and archaeological samples it is therefore questionable whether artefacts can be provenanced using this method. This problem can be addressed using Ar-Ar dating, which enables 6 unreliable dates to be identified. However, whether K-Ar or Ar-Ar dating were used it would not be possible to demonstrate conclusively that the Jaulan is the source of the artefacts, as there are a number of different outcrops over a wide area with the same ages. This problem was noted by Amiran and Porat who commented that K-Ar ages (and, by implication, Ar-Ar ages) are “capable of indicating the general source region of a given basalt vessel, but not the exact locale within 3 that region.” The technique of argon dating is therefore only of limited use in provenance studies. Another approach has therefore been taken in two studies by Philip and Williams- 8,9 Thorpe , who used wavelength-dispersive X-ray fluorescence (WDXRF) to attempt to provenance mafic artefacts, using variations in their trace element concentrations to match artefacts to their source outcrop. For the first study they collected 16 Chalcolithic and EBI basalt bowls from four different sites in Jordan and 21 new samples from outcrops in Jordan. All the samples were analysed for trace and some major elements using XRF. The analyses of the geological samples were then used to define a number of different fields, grouping the samples by location. These fields were then used to provenance the archaeological samples. These indicated that the majority of artefacts from sites south-east of the Dead Sea probably originated from the Kerak flow, and also showed that the artefacts from the sites in the Wadi Faynan did not come from the nearby Dana flow. The results also showed that the artefacts from Sal in northern Transjordan probably originated near the site. However, the remaining artefacts could not be definitely assigned, due to a 8 lack of geological data and overlaps in the existing data. Philip and Williams-Thorpe therefore expanded their previous study by analysing another 35 archaeological samples and another 8 geological samples from the North Jordan Valley, again using WDXRF. They again used trace element ratios, especially Y/Zr, to provenance the archaeological samples. Using these criteria, they were able to demonstrate that sites in southern Transjordan used the most proximal sources for grinders, but another source (probably Kerak) for bowls. They were also able to show that sites in southern Cisjordan, and the Egyptian site of Maadi, did not obtain their basalt from the closest sources across the Dead Sea, as they had previously speculated. Instead, the basalt probably originated from the north of the region, thereby revealing a more complex picture than was originally 9 assumed. However, these studies were not able to conclusively source all of the artefacts analysed. As the authors conceded, even with the new outcrop samples there was still incomplete 9 outcrop data, meaning that the provenance of artefacts cannot be regarded as completely secure. Furthermore, the new geological data revealed that there was greater chemical variability in individual outcrops than was revealed by their first study. This both increased the overlaps between the outcrop fields and raises the possibility that additional outcrop samples could further increase the geochemical variability. In turn this could alter the most likely provenance of some of the artefacts. A further problem encountered was that they were not able to analyse 15 artefacts (30% of the artefacts collected), as they were too small for analysis by XRF. This therefore reveals a major limitation of the use of XRF in attempting to provenance the generally small samples usually available from artefacts. Similar work has also been undertaken in Syria by Lease et al., but using ICP-MS to 10,11 analyse trace elements. Using rare earth element (REE) plots and multi-element plots they were able to successfully source artefacts to a particular outcrop. They concluded that most of the artefacts from the two sites examined probably originated from the an outcrop approximately 30 km north-west of the sites, rather than from the outcrop situated less than 10,11 5 km north of the sites. These were more successful studies, due at least in part to the greater range of elements that could be analysed by ICP-MS and therefore illustrate the potential of using trace elements analysed by ICP-MS to successfully provenance artefacts. The use of ICP-MS therefore offers the advantage of analysing a greater number of trace elements at greater precision than XRF, using smaller samples. Despite this, there have only been a few archaeological applications of ICP-MS to analyse stone artefacts. This is probably due to a number of factors, including the fact that stone is harder to prepare for analysis than metal, as well as the generally low levels of collaboration between archaeologists and geologists. This situation is also not favoured by the general lack of archaeological interest in most stone ...