The Canaanite Amphorae Project
This project arose from on-going research at two Egypt Exploration Society expeditions, Amarna and Memphis (Figure 1). At Amarna, a study by Margaret Serpico (Institute of Archaeology, University College London) of preserved visible organic residues in pottery vessels indicated a strong link between certain commodities and their transport in Canaanite Amphorae (see Organic Residues page). These storage jars were widely used in the eastern Mediterranean during the Late Bronze Age to hold a range of natural products such as oil, resin and honey. The designation of the amphorae as ‘Canaanite’ is based on their presumed manufacture in that region although centres of production are largely unknown. At Memphis, Janine Bourriau (McDonald Institute for Archaeological Research, University of Cambridge) undertook detailed microscopic study of the clays used to make the jars in order to form a classification system of the different clay fabrics. Confirmation that a number of different clay fabrics were used raised the possibility of identifying different geographical source locations and linking commodities to specific areas.
To explore this possibility, the Canaanite Amphorae Project was established in 1996. The project is co-directed by Serpico and Bourriau and has been undertaken in collaboration with a number of specialists. Further study of the residues has been undertaken by Dr Benjamin Stern and Dr Carl Heron (Department of Archaeological Science, University of Bradford) and examination of the clays using thin section petrography and chemical analysis has been undertaken by Dr Laurence Smith (McDonald Institute of Archaeological Research, University of Cambridge) and Dr Yuval Goren (Department of Archaeological Science, Tel Aviv University).
The aims of the project are:1) to establish a visual classification system for the fabrics used for the jars, thus enabling their identification to be secured wherever they occur;
2) to determine their regions of manufacture;
3) to examine the changes in vessel morphology for each fabric;
4) to identify the commodities they carried, whether in primary or secondary use;
5) to relate products to specific fabrics;
6) to reconstruct trade networks by linking areas of amphorae manufacture to industrial locations supplying the commodities carried in them;
7) to relate these data to the contexts in which the jars are found in Egypt .
Amarna and Memphis are especially suited to this study as quantities of sherds from these jars have been found at both (Figure 2). In addition, the sites complement each other chronologically and spatially: Memphis, capital of Egypt for most of the New Kingdom, represents an area of congested domestic housing with a long stratified sequence, while Amarna, briefly the capital under the pharaoh Akhenaten, offers evidence from a range of religious, royal and domestic contexts.
The first step was to try to establish a concordance between the fabrics found at Memphis and Amarna. While an advanced classification system had been developed for Canaanite Amphorae fabric at Memphis, the equivalent system at Amarna had been more general due to the overall volume of pottery from the site. This was achieved by studying clays at both sites with a binocular microscope. Table 1 shows the concordance for some of the more common fabrics and demonstrates that there is considerable overlap. A small number of fabrics have no definite at the opposite site, but most can be related to a similar clay at the other site, suggesting slight changes in source over time.
To ensure the accuracy of our groupings, thin sections were taken of all of these different fabrics and these were examined petrographically against an extensive comparative library of material from the eastern Mediterranean. This was followed by Neutron Activation Analysis (NAA). The results of this confirmed the concordance. Significantly, this method insures that the classification system can be related back to what an archaeologist actually sees sitting in the palm of his or her hand. It confirms that the distinctions we see with a 10x handlens are valid and offer crucial confirmation that we can transfer this classification system to different sites. Also, as suspected from the microscopic work, some of these fabrics seem to be related; likely to have come from the same general area or perhaps from varying deposits within a delineated area. From this, it has been possible to form the following groups:
|Group number||Amarna fabric||Memphis fabric|
|1||III.10, IV.5||P11, P30|
|2||IV.6, V.5||P31, P70|
|5||IV.1a, IV.12||P33, P51|
In some cases, it was possible to link these groups to the transport of specific commodities. This was achieved through a combined study of the preserved original hieratic inscriptions written on the exterior of the jars (Figure 3), which detail information about the contents, and by chemical analysis of the residues using gas chromatography/mass spectrometry (GC/MS) (see Organic Residues page).
While the inscriptions provide very useful administrative information on the nature and transport of commodities such as oil, resin and honey, they are not completely standardized: sometimes the year date of the ruler is given, sometimes not. Sometimes the name of the person transporting the product is given, sometimes not. The commodity is usually noted, but even then omissions do seem to occur. When the inscriptions were added is also unclear. They were probably written on the jars either at the origin of shipment, or more likely, en route or at the destination. Moreover, the numbers of jars thus inscribed are relatively small, indicating that not every jar in a shipment was marked.
Despite these problems, the consistent occurrence of the names of certain commodities on jars in certain clay fabrics does indicate that there are clear links between products and specific geographic areas.
The residue analyses included examination of both visible and adsorbed deposits (Figure 4). At Amarna, visible residues were still preserved in many cases, at Memphis, no visible residues remained. Solvent extractions were carried out on small samples where there were no discernable contents and the extractions were then analysed. While this proved very useful for the residues from Amarna, no residues could be detected in the Memphis samples. This was probably due to the overall poorer preservation of organics at the latter site.
Summary of the Fabric Groups and ContentsBased on the combined results of the residue and fabric analyses, the Groups can be described as follows:
Fabric Group 1
The clay fabrics in this group (Figures 5 and 6) are dominated by fine chalk, limestone and coastal quartz sand inclusions, which immediately indicated a source somewhere along the coast of Egypt north to the present Israel/Lebanon border (Figure 7). Also found in this group were small amounts of basalt and chert. This was a fortuitous combination as it occurs only in the seaward portion of the Jezreel Valley, where riverine deposits of basalt drain into the area of coastal sand. The region in question is delineated here on the map. It includes two well-known sites, Tell Abu Hawam, and Tell Nami, located just to the south, which have both been cited as commercially important ports during the Late Bronze Age. Examination of a range of vessels from these two sites indicated that they are visually identical to Amarna III.10/Memphis P30.
Several dozen Canaanite amphorae sherds in this Group at Amarna still had visible resinous residues (Figure 8) and solvent extraction was also carried out on sherds where no visible residues remained. In addition, a small number of sherds from earlier excavations at Amarna which had hieratic inscriptions denoting the contents as sntr, usually translated as incense, were discovered in museum collections in England. A small number of sherds were found which mentioned honey rather than incense (Figure 9). This is not completely surprising as documentary evidence, notably the Annals of Thutmosis III, indicate that resin, honey and oil could all be imported from one region.
Analyses of samples from current and earlier excavations indicated that the residues were consistently pistacia resin (Pistacia spp.). These trees and shrubs (Figure 10) are virtually non-existent in Egypt but widely distributed in the eastern Mediterranean. It should be mentioned, however, that other volatile components, no longer detectable, may have been present in antiquity, but nonetheless it is clear that sntr consisted predominantly, if not exclusively, of pistacia resin. As confirmation of this identification, pistacia resin was also identified in samples taken from shallow bowls where it had clearly been burned as incense (link to Organic Residues page).
No residues consistent with the identification of honey were detected. This may be because of the comparatively less robust nature of these residues which are more prone to environmental and insect attack. The comparative scarcity of inscribed sherds mentioning honey compared to those mentioning sntr, and the fact that honey could be produced within Egypt, may indicate that the primary commodity transported to Amarna in amphorae in Group 1 fabrics was incense, with honey imported to a lesser extent.
Therefore, Group 1 has a relatively restricted geographic origin and is clearly linked to the transport of pistacia resin, widely used as incense, and to a lesser extent, to the shipment of honey.
Fabric Group 2
Clays in Group 2 (Figure 12) are much less homogenous than those in Group 1 although Group 2 does have some similar components. As in Group 1, coastal quartz sand, calcareous inclusions and a small quantity of volcanic rock fragments are present. This would appear to indicate a partial geographical overlap with Group 1. However, Group 2 is characterized by a lower percentage of basaltic fragments and sometimes a highly silty matrix derived from red soil (Hamra). Such a composition would suggests that Group 2 would extend slightly further south than Group 1 (Figure 13). Initially, it was thought that Group 2 distribution would extend some distance north, but more recent detailed study of the proportions of quartz sand indicates that Group 2 is unlikely to reach further north than Akko.
Hieratic inscriptions again link this group to the transport of sntr with honey more rarely mentioned. As for Group 1, it was possible to analyze resinous residues from a number of vessels from previous and current excavations at Amarna including some inscribed for the transport of sntr to confirm this. More recently, this has been broadened to include a number of sherds with only adsorbed residues. As was the case for pistacia resin found in Group 1 jars, there was no evidence of contamination or reuse in the samples tested here. Moreover, the analytical results revealed no consistent compositional similarities or differences between the pistacia resin from the Group 1 and Group 2 jars. Thus, it is not possible to determine whether the sources for the pistacia resin were the same or varied. None of the results indicated the presence of honey.
In summary, Group 2 is, like Group 1, associated with the transport of sntr and honey. These two commodities are not mentioned in hieratic inscriptions on any of the other fabric groups found at Amarna, nor are there visible resinous deposits on amphorae in any of the other groups. Therefore, Groups 1 and 2 confirm that the resin industry was centred south of Lebanon and the famous port of Byblos .
Fabric Group 3
This group has a very distinct appearance even to the eye (Figure 14) and is characterized by basalt (occasionally reaching coarse grain size), limestone, chalk and chert inclusions. The presence of microfossils (foraminifera) and rare coralline algae fossils indicate a source near earlier coastal deposits. The absence of coastal sand with the combination of wadi sand and basalt temper points to an inland rather than coastal source. Similar components occur in the Galillee region of northern modern Israel, but the more likely sources appear to be found in the region of the Akkar Plain in southern Syria and around the coast of northern Lebanon (Figure 16).
Unfortunately, this fabric Group is comparatively rare at both Memphis and Amarna. Only one sherd with an hieratic inscription has been found, mentioning meat. Residue analyses were limited due to the scarcity of sherds in this fabric and as yet there is no secure link between this fabric group and a specific commodity.
Fabric Group 4
Group 4 is characterized by multi-coloured inclusions of limestone, serpentine, altered basalt and dolerite (Figure 18). Less common are schist and radiolarian chert inclusions, but planktonic and benthic foraminifera are also present. These components are common to both western Cyprus in the Troodos Mountains, and northwest Syria, extending into adjacent areas of Turkey (Figure 19). Distinguishing between Cypriot and Turkish sources is problematic but it is currently thought that the presence of radiolarian chert in this group is indicative of a Syrian rather than Cypriote origin. Notably, the ancient site of Ras Shamra and its port at Minet el-Beidha lie within the likely source area in western Syria.
Previous excavations at Amarna have produced a number of Canaanite amphorae sherds in this fabric group with inscriptions naming the contents as oil (Figure 20). Although there are no visible oily residues in the jars, GC/MS analysis of solvent extracted samples did confirm the presence of lipid matter, possibly a vegetable oil, in sherds in this fabric group. Analysis of the 2mm sections taken through the sherd wall and analysed separately revealed the highest concentrations of residue on the inner wall, falling rapidly through the wall of the vessel as would be expected. The subsequent slight rise on the exterior wall may represent some contamination, but is significantly lower than the concentrations on the inner wall so there is little doubt that the contents were lipid and not contamination.
Identification of the botanical source is problematic as vegetable oils contain largely the same fatty acids, varying only in proportion. Over time, these fatty acids degrade making identification unreliable. Both olive (Figure 21) and sesame have been suggested as the botanical identity of oil, but this issue is still unresolved. Archaeobotanical remains of sesame in the Near East are scant until later in the Iron Age, although some evidence is emerging for an earlier date. In contrast, olives were certainly common in the Near East and there is evidence of oil pressing at the Syrian site of Ras Shamra in the Late Bronze Age. In addition, Ras Shamra’s nearby coastal port site of Minet el-Beidha clearly was a centre for the transport of Canaanite amphorae as a storeroom filled with rows of over 80 Canaanite amphorae was found at the site. Sadly, most of these are now missing or have been destroyed over the years, but it was possible to examine the one single remaining example in the Louvre Museum with a hand lens and confirm that it is Group 4.
Thus, Group 4 is clearly linked to a trade in oil, most likely although not certainly from the area of northwest Syria/southeast Turkey .
Fabric Group 5
This fabric group can be distinguished by the dominant presence of microfossils deriving from shallow water algae and planktonic foraminifera (Figure 23). Differences in colour of the microfossils are visible with a handlens, varying from white to cream coloured to dark grey, but whether these variations are significant has been difficult to establish. The inclusions also point to beach sand found largely north of modern Israel, into Lebanon or Syria. Chert and rare serpentinised inclusions are present, but the high content of carbonates rather than quartz is characteristic the central area of the eastern Mediterranean coast (Figure 24). Thus, this group can be assigned generally to the Lebanese coastal plain, although the closest parallels for some of the examples are located along the north Lebanese coast, above Byblos.
Like fabric Group 4, study of inscribed sherds in this group from previous excavations at Amarna indicates that amphorae in clays of this group held oil (Figure 25). GC/MS analysis of adsorbed residues solvent extracted from sherds in this group again detected lipid material consistent with an oil. Comparative examination of analyses of contents from Groups 4 and 5 revealed no instructive differences, which was as expected given the compositional similarity of oils.
Collectively, the evidence from the study of Group 4 and Group 5 amphorae and contents indicates an oil industry focused largely in modern Lebanon and Syria.
Fabric Group 6
Identification of the geographic source of fabric Group 6 has been problematic. Many of the inclusions are similar to those in Group 4, but generally occur in finer sizes in this group (Figure 27). This range of inclusions points to a location in Cilicia, northwest Syria or along the southern coast of Cyprus (Figure 28). However, in comparison to Group 4, radiolarian chert is virtually absent. This absence may point to a source along the southern coast of Cyprus but more research is needed as visual study of pottery from sites in Cyprus with a hand lens has not as yet revealed any good matches within the local assemblages.
Unfortunately, no inscribed sherds have yet been identified. Similarly, visible residues are not present on sherds in this group. While negative evidence, this does suggest that the contents were not resin as resin is generally very well preserved at Amarna.
Morphology, Potmarks and Painted Decoration of Amarna Canaanite AmphoraeIn the past, it was suggested that the shape of Canaanite amphorae changed during the Amarna period. Following this theory, prior to the Amarna period, the shape of the amphorae was more bulbous, with rounded shoulders and a tapering, but still rounded base (Figure 29). At some point, during the Amarna period, the shoulders became more carinated and the base more conical (Figure 30). However, the results of the project point to a more complicated evolution. While the examples at Amarna can show some variation, there is clearly no distinct chronological point of change. Given the number of fabric groups and their distribution, it might be expected that the shape of amphorae in some fabrics would change at different rates. However, this would be very difficult to trace within the Amarna timeframe. What does seem to happen is that amphorae in some fabric groups tend to show more variation than others. Notably, there are more examples of rounded shapes in Group 2 than the other fabric groups, but carinated shapes also occur.
Other distinctions also occur, but these are not hard and fast rules. All that can be said is that, at Amarna, certain features, a thicker base, a more extended rim, are more likely to be found in certain fabric groups rather than others. Thus, while a thick base is a characteristic occasionally found in Groups 1 and 4, thick bases do occur in other fabrics and the absence of such a base does not preclude an identification of a jar as Group 1 or 4.
Similarly, some fabric groups at Amarna, notably Group 5, are more likely to have potmarks than others. These can be pre- or post-firing marks, usually on the upper handles and taking forms such as circular depressions ranging in number from one to three (Figure 31), or horizontal cut marks, again one to three, across the upper handle.
A few sherds have traces of painted marks or decoration (Figure 32), all post-firing and all comparatively uncommon. A few rare examples from previous excavations (in particular from Petrie’s excavations) at Amarna have painted floral collars on the upper shoulders. There is again no consistency to their occurrence and the only tendency that can be observed is for broad red painted curved linear marks on the bodies of some Group 1 jars.
At this stage, all of these issues, morphology, potmarks and decoration, are under assessment and a more detailed consideration will be forthcoming. However, it is unlikely that the results will be any more definitive than the observations here, with it only possible to recognize a tendency for a certain feature or trait without a clear cut link to a fabric group or a time span.
Distribution of Canaanite Amphorae at AmarnaCanaanite amphorae occur in nearly all of the excavated areas at Amarna. The comparative frequency of their occurrence is difficult to trace at this time as the full pottery assemblages have only been studied in a few areas. Notably, a high concentration was found on a water route to the Workmen’s Village, suggesting that they had been re-used for water transport. As a rule, however, there has been little physical evidence of re-use. Where resin is preserved on the surface of Group 1 and 2 amphorae, there is rarely any evidence of other products. Most often, the surface just seems to have a thin coating of mud, probably due to the sticky nature of the resin coating on the walls. No other substances have been identified in the residue analysis and there are no signs of extraneous debris embedded in the residues of these jars.
Interestingly, several hundred diagnostic sherds from Canaanite amphorae have been found at the Workmen’s Village so, despite the potential for re-use, the inhabitants clearly did have access to imported products.
In addition, the ability to link certain fabric groups to a geographical location and to specific commodities raises the possibility of tracing the distribution of amphorae and natural products across the site. The full implications of this are under review. At this time, it can be noted that Group 1 amphorae, associated with the transport of pistacia resin used as incense (see Organic Residue page), are far more common in religious contexts such as the Small Aten Temple, rather than settlement areas, such as the Workmen’s Village. That is not to say that incense is rarely found at the Workmen’s Village – it is, but that Group 1 amphorae are less common, suggesting that private individuals of the status associated with the Workmen’s Village were less likely to be able to access whole amphorae of it.
Other possible patterns are currently under review.
Canaanite amphorae and eastern Mediterranean tradeIt has long been recognized that Canaanite amphorae were widely distributed across the eastern Mediterranean and therefore are important indicators of trade. The ability to relate fabric groups to a source location and to specific commodities can, for the first time, enable us to see patterns in the distribution of jars in certain clay fabric groups and in the distribution of natural products. In order to explore the spread of these jars in antiquity, a number of sites in Cyprus, Turkey and Crete have now been visited and the information gathered is under assessment.
AcknowledgmentsThe project has received funding from the Natural Environment Research Council (NERC), the Gerald Averay Wainwright Committee (Oxford), the Egypt Exploration Society, the McDonald Institute for Archaeological Research (Cambridge), the British Academy and the Society of Antiquaries. For permission to study and/or sample sherds, we would very much like to thank the Supreme Council for Antiquities in Egypt, the Department of Ancient Egypt and Sudan at the British Museum (London), the Petrie Museum of Egyptian Archaeology (University College London), the Department of Antiquities at the Ashmolean Museum (University of Oxford), the Department of Near Eastern Antiquities at the Louvre (Paris), the Israel Antiquities Authority, Michal Artzy (Director, Tell Nami, Israel), Sophocles Hadjisavvas (Director of Antiquities, Cyprus), Ian Todd and Alison South (Directors, Kalavassos, Cyprus) and Cemal Pulak (Director, Uluburun, Institute of Nautical Archaeology).
Bibliography arising from the project
Smith, L., Bourriau, J., Goren, Y., Hughes, M., and Serpico, M. 2004. The Provenance of Canaanite amphorae found at Memphis and Amarna in the New Kingdom: results 2000-2002, in J. Bourriau and J. Phillips (eds), Invention and Innovation: The social context of technological change 2: Egypt and the Near East 1650-1150 BC. Oxbow Books, Oxford, 55-77.
Serpico, M. 2003. Quantifying Resin Trade in the Eastern Mediterranean during the Late Bronze Age, in R. Laffineur (ed.), Metron. Measuring the Aegean Bronze Age. Proceedings of the 9 th International Aegean Conference, Yale University 18-21 April, 2002. Annales d’archéologie égéenne de l’Université de Liège, 224-230.
Serpico, M., Bourriau, J., Smith, L., Goren, Y., Stern, B. and Heron, C. 2003. Commodities and Containers: A Project to Study Canaanite Amphorae Imported into Egypt during the New Kingdom, in M. Bietak (ed.), The Synchronisation of Civilisations in the Eastern Mediterranean in the Second Millennium BC II. Proceedings of the SCIEM2000 Euro-Conference, Haindorf, May 2001. Osterreichischen Akademie der Wissenschaften, Wien, 365-375.
Stern, B., Heron, C., Corr, L., Serpico, M., and Bourriau J. 2003. Compositional Variations in Aged and Heated Pistacia resin found in Late Bronze Age Canaanite Amphorae and Bowls from Amarna, Egypt, Archaeometry, 457-469.
Bourriau, J., Smith, L., and Serpico, M., 2001. Chapter 7: The provenance of Canaanite amphorae found at Memphis and Amarna in the New Kingdom, in A. Shortland (ed.), The social context of technological change: Egypt and the Near East 1650-1150 BC. Oxbow Books, Oxford, 113-46.
Serpico, Margaret and White, Raymond 2000. The Botanical Identity and Transport of Incense during the Egyptian New Kingdom, Antiquity 74 (286), 884-97.
Stern, B., Heron, C. P., Serpico, M. and Bourriau, J. 2000. A comparison of methods for establishing fatty acid concentration gradients across potsherds: a case study using Late Bronze Age Canaanite amphorae, Archaeometry 42 (2), 399-414.
Smith, L., Bourriau, J., and Serpico, M., 2000. The provenance of Late Bronze Age transport amphorae found in Egypt, Internet Archaeologist 9.
Serpico, Margaret. 1996. Mediterranean Resin Trade in New Kingdom Egypt, Unpublished Ph.D thesis, University College London.
Organic residue analysis at the University of Bradford http://www.brad.ac.uk/staff/bstern/molecular/Canaanite.html