Archaeometry 51, 5 (2009) 763–773 doi: 10.1111/j.1475-4754.2008.00437.x A PROVENANCE STUDY OF THE GILGAMESH FRAGMENT FROM MEGIDDO Y. GOREN,1 H. MOMMSEN,2 I. FINKELSTEIN1 and N. NA’AMAN1 1 Department of Archaeology and Ancient Near Eastern Civilizations, Tel Aviv University, Israel 2 Helmholtz-Institut für Strahlen- und Kernphysik, Universität Bonn, Bonn, Germany A Late Bronze Age fragment of a clay cuneiform tablet with the Gilgamesh Epic was found in the 1950s on the surface at Megiddo. The presence of scribes in Megiddo is evident from the el-Amarna letters. This is the only first-class literary Mesopotamian text ever to be found in Canaan. The aim of the present study was to examine the origin of this tablet, by mineralogical and elemental methods. The petrographic and NAA results indicate that the tablet was not Mesopotamian, but was written in southern Israel. The implications of this result in view of the small corpus of scholarly cuneiform texts discovered in Egypt and the southern Levant in the second millennium BCE are discussed. INTRODUCTION The importance of Megiddo in the Late Bronze Age as one of the leading city-states in Canaan is indicated by a plethora of hieroglyphic and cuneiform historical sources (Ahituv 1984, 139–40; Moran 1992, 297–300, 363) and archaeological finds. The decisive battle of Thutmose III against a coalition of Canaanite city-states took place near Megiddo. A century later, the city played an important role in the events described in the Amarna letters. Megiddo is probably mentioned in a Hittite letter found at Bogazköy (Singer 1988). The series of Late Bronze palaces, which were excavated at Megiddo in the 1930s by the University of Chicago team, are among the most elaborate edifices unearthed in the territory of Late Bronze Canaan, and the Megiddo ivories (Loud 1939) testify to the opulence of the rulers who inhabited them. Egyptian finds and a Hittite seal found at the site (Singer 1988–9, 2002, respectively) demonstrate Megiddo’s international links, which stemmed from its strategic location on the international road that connected Egypt in the south with Anatolia and Mesopotamia in the north and north-east (Fig. 1). The fragment of a cuneiform clay tablet with the Gilgamesh Epic under discussion was found in the early 1950s by a shepherd from kibbutz Megiddo, near the dump-ramp created by the University of Chicago’s excavators on the northwestern sector of the mound (Fig. 2). The fragment was published by Goetze and Levy (1959), who dated it to ‘roughly the Amarna Age’ (ibid., 128). The date (Middle Babylonian period; Late Bronze Age) was accepted in the most recent editions of the tablet by George (2003, 339–47) and Horowitz and Oshima (2006, 102–5). The northwestern dump-ramp received the debris from the excavation of Area AA and hence it is reasonable to assume that the tablet originated from one of the Megiddo palaces that had been unearthed there. These include the palaces of Stratum X, probably dating to the *Received 23 October 2007; accepted 16 June 2008 © University of Oxford, 2008 764 Y. Goren et al. Figure 1 A map of west Asia and the eastern Mediterranean, with the location of Megiddo and other locations mentioned in the text. © University of Oxford, 2008, Archaeometry 51, 5 (2009) 763–773 A provenance study of the Gilgamesh fragment from Megiddo 765 Figure 2 The two sides of a fragment of the Gilgamesh Epic from Megiddo (photo: Y. Goren, courtesy of the Israel Museum in Jerusalem). Middle Bronze III; Stratum IX, dating to the Late Bronze I; Stratum VIII, probably dating to an early phase of the Late Bronze II (14th century bce?); and Stratum VII, dating to the Late Bronze II–III of the 13th–12th centuries bce (for the excavation, see Loud 1948, 15–33, figs 380–4; for the edifices of sub-strata VIIB and VIIA being in fact phases in a single palace, see Ussishkin 1995). The Megiddo palaces were excavated in a large area of roughly 50 ¥ 30 m, yet only the northern parts of the complexes were exposed; the published plans leave no doubt that walls belonging to the buildings continue to the south, into an area where excavations did not venture under late Iron II remains. Assuming that the building recently excavated at Hazor is a temple rather than a palace (Zuckerman 2007, 12–13; contra Ben-Tor, e.g., 1998), the Megiddo buildings are the largest and best preserved palaces ever uncovered in the territory of Late Bronze Canaan. The destruction layer of the palace of Stratum VII, which probably dates to the second half of the 12th century bce (Singer 1988–9; Ussishkin 1995; and see recent 14C results in Sharon et al. 2007 and their interpretation in Finkelstein and Piasetzky 2007), yielded a rich assemblage of finds, including the over 200 ivories (Loud 1939). Though the ivories were found in Stratum VII, they probably represent a collection, which spans a long time in the history of the Late Bronze palaces at Megiddo. Late Bronze Megiddo dominated the western part of the Jezreel Valley. Surveys conducted in this area revealed a dense network of rural sites—one of the most developed in Canaan—which constituted the agricultural hinterland of the city (for a summary, see Finkelstein et al. 2006). The presence of scribes in Late Bronze Megiddo, specifically in the 14th century bce, is evident from the six tablets sent from Megiddo which were found at el-Amarna (EA 242–6 and 365; see Moran 1992, 297–300, 363), tablets that were made at the site (Goren et al. 2004, 243–6). However, the literary quality of the Megiddo letters is not high and is less elaborate compared to those that were written in, for example, Gubla, Tyre or Jerusalem. The discovery of the Gilgamesh fragment—the only first-class literary Mesopotamian text that has ever been © University of Oxford, 2008, Archaeometry 51, 5 (2009) 763–773 766 Y. Goren et al. found in Canaan—at the site is surprising and indicates how little we know of the education of the local scribes in Canaan. The aim of the present study has been to examine the provenance of this unique literary tablet by petrographic and archaeometric methods, in order to supply additional information about its origin as well as its place in the small corpus of scholarly cuneiform texts discovered in Egypt and the southern Levant in the second millennium bce. METHOD The sampling and examination procedure followed the methodology that was developed by us for the examination of clay cuneiform tablets (Goren 2000; Goren et al. 2002, 2003a,b, 2004, 2007). First, the tablet was examined thoroughly under a stereomicroscope, in order to define the overall characteristics of its fabric. A shallow lamina, with dimensions about 10 ¥ 5 mm, was peeled off one of the broken facets of the tablet with the aid of a scalpel. Since, as it seems, the tablet was very lightly fired, cutting such a slice was easy and did not affect the museological value of the item. Another sample, this time as powder, was taken from the same broken facet after the removal of the outer surface with the aid of a corundum chisel. This material was used for the elemental analysis. The preparation of the sample for petrographic analysis was made in the following stages: the block sample was set in a small polyethylene mould (about 1 cm in diameter) and dried in an oven at 60°C for a few hours. Then, the cup with the sample was put into a desiccator under vacuum, where the sample was impregnated with Buehler Epo-Thin low-viscosity epoxy resin. After curing, the resulting pellet was used for the preparation of a standard thin section and subjected to routine petrologic examination under a Zeiss Axiolab-Pol polarizing microscope under ¥50–¥400 magnifications. Instrumental Neutron Activation Analysis (INAA) has been applied routinely at the laboratory in Bonn for many years to determine the place of production of ceramic objects, using its characteristic minor and trace elemental pattern (see, e.g., Mommsen 2004 and references therein). Up to 30 elemental concentration values can be measured using a small powder sample of about 80 mg, which is obtained by drilling at a fixed position or by moving a small drilling machine with a sapphire drill bit across a small area of a broken surface. The neutron irradiation of a whole set of such samples, including several samples of the Bonn pottery standard, takes place at the reactor of the GKSS research centre at Geesthacht, near Hamburg, at a thermal flux of 5 ¥ 1013 neutrons cm-2 s-1 for 90 min. Since the Bonn standard has been calibrated with the Berkeley/Jerusalem pottery standard (Perlman and Asaro 1969), the values can be compared directly. The whole procedure, giving more details, was described recently (Goren et al. 2007). A large databank of over 7000 samples from the eastern Mediterranean region is available for pattern comparison, mainly from Greece and the Aegean, but also from Cyprus (50 samples), Palestine (460 samples) and Egypt (282 samples, most of them imports), also comprising many reference pieces of known origin. The pattern comparison is done with the help of a set of computer programs developed in Bonn (Beier and Mommsen 1994; Mommsen et al. 2002, 615–17), which are able to filter out from a large databank all samples with similar or related composition. As a similarity measure, a squared modified Mahalanobis distance is calculated that takes into account both individual elemental measuring errors and various dilution effects, which result in constant shifts in elemental values. In addition, compared to other similarity measures, this distance has the advantage that it can be directly converted into a probability of chemical agreement. © University of Oxford, 2008, Archaeometry 51, 5 (2009) 763–773 A provenance study of the Gilgamesh fragment from Megiddo 767 COMPARATIVE MATERIALS The petrographic study of the Megiddo tablet relied on the collection of southern Levantine ceramics and clays in thin sections of the Laboratory for Comparative Microarchaeology of the Institute of Archaeology, Tel Aviv University, containing over 12 000 specimens. As the primary database of comparative materials for the petrography of Mesopotamian clay cuneiform tablets, we used the data obtained during the analysis of Amarna letters sent from Mitanni and Babylonia to Egypt (Goren et al. 2004). Other sources of information were more limited, since very little research has been done on the technology of pottery from this general area. Comparative data regarding the petrography of sites along the Euphrates was extracted from the thin-section collection of the ‘Gerald Avery Wainwright Archive of Mesopotamian Ceramics’ (Mynors 1986), now deposited in the Department of Scientific Research of the British Museum. Some data on the petrography of Euphrates sediments in pottery was also revealed from articles relating to ceramics from sites along the Euphrates drainage system. These include Tell Hadidi, Tell al-Sweyhat, Tell Banat and Tell Kebir on the upper Euphrates. More downstream, one finds Raqqa (Mason 1994; Mason and Cooper 1999), Terqa and Dilbat (Flint 1980), Dura Europos and Mari (Mason and Cooper 1999). Data on the Lower Euphrates come from Tell ed-Der, Nippur and Isin (Franken and Van As 1994). However, these studies do not supply any comparable petrographic or chemical data. Archaeometric research on the Mitannian letters from the Amarna archive began with the Laurence Berkeley laboratory team’s attempt to disclose the provenance of selected letters through NAA (Dobel et al. 1977). The Berkeley researchers were able to form a homogeneous elemental pattern in four (out of six measured) letters (EA 22, 24, 25 and 29). They compared this profile with pottery from Tell Fakhariyah, including Mitanni ware, and with a chemical database of selected ceramics from several sites that had been suggested as the location of Waššukanni, obtained in Edinburgh by Davidson and McKerrel (1976). The comparative material of this database also included data on clay samples that the latter collected from the Khabur headwaters, from Wadi Dara and Wadi Jaghjagh. Whereas the older Halafian pottery and the Mitanni pottery from Tell Fakhariyah measured in Berkeley generally agreed in composition with the pottery and most of the recent clay samples of the Edinburgh database, the pattern of the Mitanni letters from the Amarna archives showed marked differences; for example, twice as much Ni (measured only in Berkeley) and only half as much Th and Hf. Only one clay sample (C61: Davidson and McKerrel 1976, 48, 56), taken about 27 miles north-east of Tell Halaf, has such low values of Th and Hf. The special clay bed used for forming the Mitanni letters may not have been sampled, since the only two samples from the Khabur valley (C64 and C65), collected 15 and 30 miles downstream of Tell Fakhariyeh, show the usual Th and Hf values for this whole area (Davidson and McKerrel 1976, 48, 56). Archaeometrically, the origin of the Mitanni letters is still an open issue. RESULTS The petrographic examination of the Megiddo tablet indicated the following features (Fig. 3): • Matrix: Calcareous, containing planktonic foraminifers, most likely of Palaeocene age, and their fragments (~2–3%). The matrix is light yellowish-tan in plain-polarized light (PPL), optically active with crystallitic b-fabric. Opaque minerals are infrequent, forming about 1% of the matrix and sizing up to 150 mm. Quartz silt is uncommon (below 1%) and is accompanied by very few accessory heavy minerals, including epidote and plagioclase. © University of Oxford, 2008, Archaeometry 51, 5 (2009) 763–773 768 Y. Goren et al. Figure 3 A microscopic view of the fragment of the Gilgamesh Epic from Megiddo in thin section, crossed polarizers. Field length 2.5 mm. Rounded sand-sized quartz grains within calcareous matrix with elongated sharp-edged voids, indicating the presence of chopped vegetal material. • Inclusions: Moderately to well-sorted sand (f:c ratio{0.062 mm} = 90:10) of the following: – Quartz: frequent to dominant (up to 550 mm), subangular to rounded, sometimes with undulose extinction and rarely with mineral inclusions. – Vegetal material: frequent (up to 500 mm lengthwise), usually identified by the resulting sharp-edged elongated void in the matrix, but often these voids still contain some badly preserved, unidentifiable tissue fragments or phytoliths. – Chalk: common, rounded (up to 450 mm) with foraminifers. – Chert: rare, subrounded (up to 450 mm) grains of replacement chert. The NAA examination of the Megiddo tablet (labelled Megi 2) rendered an elemental pattern given in Table 1. It was measured using the standard procedure described above. INTERPRETATION OF THE ANALYTICAL RESULTS According to its petrographic affinities, this tablet does not suit the composition of the Mesopotamian tablets that we have examined. The Euphrates River drains an area characterized by ophiolitic, basaltic, felsic volcanic and plutonic complexes that outcrop around its headwaters (Ponikarov 1966, Sheets I-37-II, I-37-III and I-37-IV). From the point where the two tributaries, the Balikh and the Habur, join the Euphrates and downstream, the petrofabrics display a gradual decrease in the appearance of ‘soft’ minerals that are derived from this lithology, namely amphiboles and serpentines, and gradual increase in ‘hard’ minerals (quartz, clinopyroxene and felsic volcanics). The common feature of all these components is that they are rounded and sorted by the effect of the river transportation, and that the clay is usually the river’s fine deposit. Further downstream, these trends are expected to increase, since there is no further source of supply of fresh igneous rocks or their derived minerals. Therefore, when one reaches Babylonia, it is expected that the unstable mafic minerals and their alteration products would © University of Oxford, 2008, Archaeometry 51, 5 (2009) 763–773 A provenance study of the Gilgamesh fragment from Megiddo 769 Table 1 Concentrations of elements measured by NAA of regional pottery group PalJ from Israel compared to sample Megi 2 corrected for a dilution (factor) of 18% with respect to this group: averages M and values C in mg g-1 (ppm), if not indicated otherwise, and spreads ␴ and measurement errors d, respectively, also in % PalJ, 23 samples, factor 1.00 As Ba Ca (%) Ce Co Cr Cs Eu Fe (%) Ga Hf K (%) La Lu Na (%) Nd Ni Rb Sb Sc Sm Ta Tb Th Ti (%) U W Yb Zn Zr Megi 2, one sample, factor 1.18 M1␴ % C1d % 4.48 1 0.72 555 1 181 6.30 1 1.63 65.2 1 3.50 17.4 1 1.17 100 1 3.92 1.32 1 0.24 1.35 1 0.041 3.84 1 0.16 14.8 1 2.41 11.1 1 1.62 1.13 1 0.22 29.0 1 0.81 0.44 1 0.023 0.61 1 0.20 26.7 1 0.97 62.9 1 23.2 41.7 1 4.56 0.48 1 0.089 12.7 1 0.55 5.14 1 0.13 1.15 1 0.062 0.77 1 0.035 7.45 1 0.36 0.65 1 0.076 1.80 1 0.34 1.34 1 0.24 2.97 1 0.12 75.8 1 25.6 379 1 81.0 16 33 26 5.4 6.7 3.9 18 3.0 4.1 16 15 20 2.8 5.1 32 3.6 37 11 18 4.3 2.6 5.4 4.6 4.9 12 19 18 4.1 34 21 19.3 1 0.12 1557 1 47.6 9.27 1 0.16 67.9 1 0.64 15.0 1 0.11 108 1 0.82 1.77 1 0.069 1.36 1 0.025 3.59 1 0.015 15.0 1 1.93 6.40 1 0.069 0.94 1 0.020 33.0 1 0.11 0.43 1 0.017 0.18 1 0.002 24.4 1 1.53 56.5 1 28.5 38.6 1 1.82 0.77 1 0.060 12.0 1 0.020 4.92 1 0.020 1.01 1 0.031 0.90 1 0.049 7.39 1 0.060 0.73 1 0.071 1.93 1 0.086 1.13 1 0.13 2.98 1 0.049 77.5 1 2.09 230 1 23.9 0.6 3.1 1.8 0.9 0.8 0.8 3.9 1.8 0.4 13 1.1 2.1 0.3 4.0 1.2 6.3 51 4.7 7.8 0.2 0.4 3.1 5.4 0.8 9.7 4.5 12 1.7 2.7 10 diminish considerably, and the relative proportion of stable minerals (first and foremost quartz) would increase respectively. These changes, which occur gradually along the drainage system, can be distinguished only over long distances (as we observed while examining the Gerald Avery Wainwright Archive of Mesopotamian Ceramics). Therefore, when examining the Babylonian tablets it is difficult to identify the exact production site, but the overall picture of lower Mesopotamian origin is quite clear. However, the petrographic characteristics of the Megiddo tablet are well known from the examination of the Canaanite letters from the Amarna archive. The matrix is obviously made of marl of the Palaeocene to basal Eocene age of the southern Levant, such as the Taqiye formation of Israel (for a discussion and references, see Goren et al. 2004, 257). Although © University of Oxford, 2008, Archaeometry 51, 5 (2009) 763–773 770 Y. Goren et al. Palaeocene marls are almost constant in their stratigraphic position, and even with regard to the details of their composition throughout the Levant, the combination of Palaeocene marl with quartzitic sand mixed with chalk and some chert is well recorded both from the local geology and the pottery assemblage of the ancient city of Gezer. Indeed, the letters of Milkilu and Ba‘lu-dānu, the rulers of Gezer in the Amarna Period, are petrographically similar to the tablet in question (Goren et al. 2004, 271–4). On the other hand, the local geology of Megiddo and the composition of the letters of Biridiya, the ruler of Megiddo, in the Amarna archive (Goren et al. 2004, 243–6) negate the possibility that the fragment of the Gilgamesh Epic under discussion was produced locally, at Megiddo. The NAA results generally support the petrographic results, although the location of the specific elemental pattern of the Megiddo tablet remains unknown; none of the over 7000 patterns of the databank has a statistically similar composition. Nevertheless, in the common picture of a multidimensional concentration space, where each sample corresponds to a point, the data point of sample Megi 2 is close to a larger group PalJ of samples originating from Israel (Ashdod, Ashkelon, Yavneh and Tell es-Safi/Gath) and has large distances to other samples from Anatolia, Cyprus and Egypt. It is also very different in composition from samples of the letters and clays from Mitanni. In Table 1, the local group of samples from Israel (PalJ) is compared to the pattern of Megi 2, which has been corrected for dilution by a best relative fit factor of 1.18 with respect to the PalJ group. Also, a comparison with various literature data of samples from Israel measured in Berkeley and Jerusalem gave the same result. Because of this general agreement with patterns from Israel, an origin there is highly probable; however, our database is not large enough to assign a specific site. DISCUSSION The petrography of the Amarna literary texts can enlighten that of the Megiddo fragment, the only available literary text discovered so far south of Ugarit. The petrographic and INAA results indicate that most of the scholarly tablets discovered at Amarna were written in Egypt, including exercises of different kinds, literary and lexical texts, syllabaries and an amulet. Only three tablets of Mesopotamian origin—all of them literary—were discovered at Amarna. These are the myth of Adapa and the South Wind (EA 356) and two fragmentary texts whose genre is not clear (EA 342 and 344). No text of Anatolian or north Syrian origin has been traced. Unfortunately, the texts of the epic of šar taměāri (EA 359 and 375), which exhibit some linguistic peculiarities that can be attributed directly to the Akkadian of Bogazköy, have not been investigated thus far (Goren et al. 2004, 76–87). To explain these unexpected results, we suggested that texts of northern origin were brought to Egypt at the request of the royal court and copied there by local scribes. However, when the royal court left Amarna, the original texts were transferred to the new capital, and copies of inferior quality were left in their place. The preference of the Egyptian scribes for the master copies may thus explain the small number of scholarly tablets of northern origin discovered at Amarna (Goren et al. 2004, 87). A few scholarly tablets have been discovered in the land of Canaan; the origin of some of them has been scientifically established. The lexical tablets from Tel Aphek are made of local clay, and even the letter from Ugarit was copied at the site (Goren et al. 2007, 162–5). The prism fragment from Hazor was locally made and the origin of the inscribed liver model might also be local (Goren 2000, 34–7; Goren et al. 2004, 229). The Ashkelon lexical tablet (Huenergard and van Soldt 1999; Horowitz and Oshima 2006, 42–3), which was examined © University of Oxford, 2008, Archaeometry 51, 5 (2009) 763–773 A provenance study of the Gilgamesh fragment from Megiddo 771 petrographically by one of us (YG, as yet unpublished), was found to be made in the Lebanese littoral. The Beth-shemesh cuneiform alphabetical text (Sass 1991; Horowitz and Oshima 2006, 157–60) was also examined petrographically (by YG, as yet unpublished), and found to be made in the southern Coastal Plain of Israel. Summing up the results of our analysis, it seems that almost all of the local Canaanite scholarly texts were made inside the general boundaries of Canaan. However, it is apparent that within this broad area they were sometimes transported from one site to another. In this respect, the Gilgamesh fragment is not an exception. The results of the petrographic and NAA analyses of the Gilgamesh fragment indicate that the tablet was probably produced in Gezer. This fits the picture drawn from the analyses of the Egyptian literary texts and the local Canaanite scholarly texts, which indicate their local origin. It is clear that the Gilgamesh text arrived in Canaan at an earlier stage and that what we have is a copy of an original tablet. This raises the problem of the importance of the Gezer scribal school in the Late Bronze Age, as well as the issue of the route by which a literary text from Gezer reached the city-state of Megiddo. Vita (2000, 2005) has examined a group of south Palestinian letters sent from Gezer and other neighbouring cities. He concluded that the same scribe wrote about 24 letters, which he labelled the ‘Gezer-Corpus’. Among them are letters of Milkilu (EA 267–71), Ba‘lu-dānu (EA 293–4) and Yapahu (EA 297–300) of Gezer, and also Shuwardatu (EA 278–80), Shum-[..] (EA 272), Belit-nesheti (EA 273–4), Yahzib-Adda (EA 275–6), Tagi (EA 266) and Yahtiru (EA 296). According to the petrographic analysis, almost all the non-Gezerite letters in this list (EA 266, 275–6, 278–80 and 296) were not sent from Gezer. Assuming that Vita’s conclusions, which are based on the shape and ductus of the cuneiform signs, are valid, one can propose that the Gezer scribe travelled to neighbouring cities whose rulers were Gezer’s allies (e.g., Ginti-kirmil, Yahzib-Adda’s capital, Gath = Tell es-Safi and possibly Ashdod) and wrote letters on behalf of their rulers. Why Shuwardatu and Tagi would call on the service of a scribe from faraway Gezer to write certain letters for them, while they used their own scribes for other messages, remains unclear. The palaeographic and petrographic analyses of the Amarna tablets indicate the centrality of the Gezer scribes in southern Canaan. The analysis of the Gilgamesh fragment brings a new element to the discussion. It indicates that the Gezer scribal school also studied the classical literary texts of the Mesopotamian scribal tradition and distributed them to other Canaanite centres. The question of when the original Gilgamesh text reached Gezer and by what mechanism remains unanswered. Another question concerns the circumstances by which the Gezer Gilgamesh tablet arrived at Megiddo. Did the Gezer scribe prepare a copy at the request of the Megiddo scribe and send it to his city? Was there a system of exchanging tablets between the main scribal schools at Canaan, so that students at one centre were able to study the tablets available at the other centres? Does the tablet reflect a period in which Gezer and Megiddo were allies, unlike the situation in the Amarna period, when there was enmity between the two centres? We need more literary tablets in order to answer these questions, and in the meantime we must live with questions that need to wait for reasonable answers in the future. ACKNOWLEDGEMENTS The petrographic study of the Gilgamesh fragment from Megiddo was carried out as part of a research project entitled ‘Provenance Study of the Corpus of Cuneiform Tablets from © University of Oxford, 2008, Archaeometry 51, 5 (2009) 763–773 772 Y. Goren et al. Eretz-Israel’. This research was supported by the Israel Science Foundation (grant no. 817/0225.0). The authors wish to thank O. Misch-Brandl, Curator of the Chalcolithic and Bronze Age Antiquities in the Israel Museum, and H. Katz, Head of the National Treasuries Division in the Israel Antiquities Authority, for their permission to analyse the tablet, now displayed in the Israel Museum in Jerusalem. We thank A. Middleton, from the Department of Scientific Research of the British Museum, who allowed us to use the thin sections belonging to the Gerald Avery Wainwright Archive of Mesopotamian Ceramics, deposited in the Department of Scientific Research of the British Museum by Syriol Mynors. The help of the irradiation service at the reactor of the GKSS research centre at Geesthacht is gratefully acknowledged. REFERENCES Ahituv, S., 1984, Canaanite toponyms in ancient Egyptian documents, Magnes, Jerusalem. Beier, Th., and Mommsen, H., 1994, Modified Mahalanobis filters for grouping pottery by chemical composition, Archaeometry, 36, 287–306. Ben-Tor, A., 1998, The fall of Canaanite Hazor—the ‘who’ and ‘when’ questions, in Mediterranean peoples in transition: 13th to early 10th centuries BCE (eds. S. Gitin, A. Mazar and E. Stern), 456–67, Israel Exploration Society, Jerusalem. Davidson, T. E., and McKerrel, H., 1976, Pottery analysis and Halaf period trade in the Khabur headwaters region, Iraq, 38, 45–56. Dobel, A., Asaro, F., and Michel, H. V., 1977, Neutron activation analysis and the location of Waššukanni, Orientalia, 46, 375–82. Finkelstein, I., and Piasetzky, E., 2007, Radiocarbon dating and Philistine chronology, with an addendum on el-Ahwat, Ägypten und Levante, 17, 74–82. Finkelstein, I., Halpern, B., Lehmann, G., and Niemann, H. M., 2006, The Megiddo Hinterland project, in Megiddo IV: the 1998–2002 seasons (eds. I. Finkelstein, D. Ussishkin and B. Halpern), B. 705–76, Institute of Archaeology, Tel Aviv University, Tel Aviv. Flint, A., 1980, A preliminary thin-section analysis of Terqa and Dilbat ceramics, Syro-Mesopotamian Studies, 3, 174–8. Franken, H. J., and Van As, A., 1994, Potters who used Euphrates clay, Mesopotamian History and Environment Occasional Publications, II, 507–18. George, A. R., 2003, The Babylonian Gilgamesh Epic: introduction, critical edition and cuneiform text, Oxford University Press, Oxford. Goetze, A., and Levy, S., 1959, Fragment of the Gilgamesh Epic from Megiddo, Atiqot, 2, 121–8. Goren, Y., 2000, Provenance study of the cuneiform texts from Hazor, Israel Exploration Journal, 50, 29–42. Goren, Y., Finkelstein, I., and Na’aman, N., 2002, The seat of three disputed Canaanite rulers according to petrographic investigation of the Amarna tablets, Tel-Aviv, 29, 221–37. Goren, Y., Finkelstein, I., and Na’aman, N., 2003a, The expansion of the kingdom of Amurru according to the petrographic investigations of the Amarna tablets, Bulletin of the American Schools of Oriental Research, 329, 1–11. Goren, Y., Finkelstein, I., and Na’aman, N., 2004, Inscribed in clay: provenance study of the Amarna tablets and other Near Eastern texts, Institute of Archaeology, Tel Aviv University, Tel Aviv. Goren, Y., Bunimovitz, S., Finkelstein, I., and Na’aman, N., 2003b, The location of Alashiya: new evidence from petrographic investigation of Alashiyan tablets from el-Amarna and Ugarit, American Journal of Archaeology, 107, 233–55. Goren, Y., Na’aman, N., Mommsen, H., and Finkelstein, I., 2007, Provenance study and re-evaluation of the cuneiform documents from the Egyptian residency at Tel Aphek, Ägypten und Levante, 16, 161–71. Horowitz, W., and Oshima, T., 2006, Cuneiform in Canaan: cuneiform sources from the land of Israel in ancient times, Israel Exploration Society, Jerusalem. Huehnergard, J., and van Soldt, W., 1999, A cuneiform lexical text from Ashkelon with a Canaanite column, Israel Exploration Journal, 49, 184–92. Loud, G., 1939, The Megiddo ivories, The University of Chicago Press, Chicago. Loud, G., 1948, Megiddo II: Seasons of 1935–1939, The University of Chicago Press, Chicago. Mason, R. B., 1994, Islamic glazed pottery 700–1250, Unpublished Ph.D. dissertation, University of Oxford. Mason, R. B., and Cooper, L., 1999, Petrographic analysis of Bronze Age pottery from Tell Hadidi, Syria, Levant, 31, 135–47. © University of Oxford, 2008, Archaeometry 51, 5 (2009) 763–773 A provenance study of the Gilgamesh fragment from Megiddo 773 Mommsen, H., 2004, Short note: Provenancing of pottery—need for the integrated approach, Archaeometry, 46, 267– 71. Mommsen, H., Beier, T., and Hein, A., 2002, A complete chemical grouping of the Berkeley neutron activation analysis data on Mycenaean pottery, Journal of Archaeological Science, 29, 613–37. Moran, W., 1992, The Amarna letters, Johns Hopkins University, Baltimore, MD. Mynors, H. S., 1986, Mesopotamian ceramics of the third millennium BC with analysis of pottery from Abu Salabikh, Kish and Ur, Unpublished Ph.D. dissertation, University of Southampton, Southampton. Perlman, I., and Asaro, F., 1969, Pottery analysis by neutron activation, Archaeometry, 11, 21–52. Ponikarov, V. P. (ed.), 1966, The geological map of Syria, 1:200 000 (20 sheets with geological cross-sections and explanatory notes), Technoexport, Moscow. Sass, B., 1991, Studia alphabetica. On the origin and early history of the Northwestern Semitic, South Semitic and Greek alphabets (OBO 102), Freiburg. Sharon, I., Gilboa, A., Jull, T. A. J., and Boaretto, E., 2007, Report on the first stage of the Iron Age Dating Project in Israel: supporting a low chronology, Radiocarbon, 49, 1–46. Singer, I., 1988, Megiddo mentioned in a letter from Bogazkoy, in Documentum Asiae Minoris Antiquae: Festschrift für Heinrich Otten zum 75. Geburtstag (eds. E. Neu and C. Ruster), 327–32, Harrassowitz, Wiesbaden. Singer, I., 1988–9, The political status of Megiddo VIIA, Tel Aviv, 15–16, 101–12. Singer, I., 2002, A Hittite seal from Megiddo, in Across the Anatolian Plateau: readings in the archaeology of ancient Turkey (AASOR 57) (ed. D. C. Hopkins), 145–7, American Schools of Oriental Research, Boston. Ussishkin, D., 1995, The destruction of Megiddo at the end of the Late Bronze Age and its historical significance, Tel Aviv, 22, 240–67. Vita, J.-P., 2000, The Gezer-Corpus von El-Amarna: Umfang und Schreiber, Zeitschrift für Assyriologie, 90, 70–7. Vita, J.-P., 2005, Der biblische ortsname Zaphon und die Amarnabriefe EA 273–274, Ugarit-Forschungen, 37, 673–7. Zuckerman, S., 2007, Anatomy of a destruction: crisis architecture, termination rituals and the fall of Canaanite Hazor, Journal of Mediterranean Archaeology, 20, 1–31. © University of Oxford, 2008, Archaeometry 51, 5 (2009) 763–773