JournalofArchaeologicalScience40(2013)1122e1135 ContentslistsavailableatSciVerseScienceDirect Journal of Archaeological Science journal homepage: http://www.elsevier.com/locate/jas Empires and resources: Central Anatolian obsidian at Urkesh (Tell Mozan, Syria) during the Akkadian period Ellery Frahma,*, Joshua M. Feinbergb,c aDepartmentofArchaeology,TheUniversityofSheffield,NorthgateHouse,WestStreet,SheffieldS14ET,UnitedKingdom bInstituteforRockMagnetism,UniversityofMinnesota,310PillsburyDriveSE,Minneapolis,MN55455,UnitedStates cDepartmentofEarthSciences,UniversityofMinnesota,310PillsburyDriveSE,Minneapolis,MN55455,UnitedStates a r t i c l e i n f o a b s t r a c t Articlehistory: AlmostalloftheobsidianusedtocraftstonetoolsintheNearEastfromthePalaeolithiconwardorig- Received20May2012 inatedfromvolcanoesintwogeographicregions:CentralAnatoliaandEasternAnatolia.Fivedecadesof Receivedinrevisedform obsidiansourcinghasledtotheviewthatCentralAnatolianobsidianslargelyfollowedtheMediterra- 25July2012 neancoastandrarelyreachedfarthereastthantheMiddleEuphrates,whereasEasternAnatoliansources Accepted31July2012 almostexclusivelysuppliedsiteseastoftheEuphrates.ThispaperdiscussestheidentificationofCentral AnatolianobsidianartefactsattheBronze-AgesiteofTellMozan(Urkesh)innortheasternSyria.Mostof Keywords: theobsidiansatTellMozan(97%)camefromtheEasternAnatoliansources,asexpectedfromestablished NorthernMesopotamia distribution models. Artefactsof CentralAnatolian obsidian, however,wereexcavatedfromonewell- KhaburTriangle BronzeAge constrained context: the deposits on a palace courtyard that date to the height of the Akkadian Akkadianempire empire’s influence at this third-millennium Hurrian religious and political centre. In particular, the TellMozan obsidiancamefromtheKömürcüsourceofGöllüDag(cid:2).Potentialexplanationsforthisexoticobsidianare Urkesh discussed. This obsidianmighthave“piggybacked”on thedistribution ofCentralAnatolianmetalsor Exchangenetworks arrivedatthiscityasroyalgiftsorprestigeitems.OtherdiscussedmechanismsincludeAkkadian-linked Geochemicalobsidiansourcing changesineitherterritorialityinvolvingpastoralnomadsresponsibleforthearrivalofEasternAnatolian Magneticobsidiansourcing obsidians or identity construction of elites based on involvement in Central Anatolian economic and politicalnetworks. (cid:1)2012ElsevierLtd.Allrightsreserved. 1. Introduction 1964;Renfrewetal.,1966,1968).Theyshowed,bythestartofthe BronzeAge,theEasternAnatolianobsidian“interactionzone”was AcrossMesopotamiaandtheLevant,nearlyalloftheobsidian Mesopotamian,whereastheCentralAnatolianzonewasLevantine used to craft stone tools from the Palaeolithic onward originated and Anatolian. Later regional studies reinforce these trends (e.g., from volcanoes in Anatolia (i.e., modern Turkey; Fig. 1). The Cauvin and Chataigner,1998; Chataigner,1998, Chataigner et al., Anatolian obsidian sources occur in two geographic regions: the 1998).Simplyput,CentralAnatolianobsidiansareextremelyrare Central Anatolian sources (Acıgöl, Göllü Dag(cid:2), Nenezi Dag(cid:2), and inMesopotamia.Outofabout1000sourcedandpublishedMeso- HasanDag(cid:2))andEasternAnatoliansources(theBingölsources,Mus¸, potamian obsidian artefacts, the number with Central Anatolian NemrutDag(cid:2),MeydanDag(cid:2),andTendürekDag(cid:2),amongothers).Five origins can be counted on one hand, and most of them have no decadesofsourcingNearEastobsidianartefactshasledtotheview spatiotemporalcontext. that, outside of Anatolia, use of Central Anatolian obsidians is HerewediscussthediscoveryofCentralAnatolianobsidianat a principally Levantine phenomenon. These obsidians primarily theBronze-AgesiteofTellMozan(ancientUrkesh).Locatedinthe followedtheMediterraneancoastandrarelyreachedfarthereast northeastern cornerof Syria, this earlyHurrianpolitical and reli- thanMiddleEuphratessites,whereastheEasternAnatoliansources gious centre lies within the proposed “supply zone” for Eastern almostexclusivelysuppliedsiteseastoftheEuphrates.Thistrend Anatolianobsidians(Dixonetal.,1968).Thissitealsoliesnearthe wasfirstrecognisedbyRenfrewandcolleagues(CannandRenfrew, mouth of the Mardin Pass between the Anatolian highlands and Mesopotamianlowlands,permittingreadyaccesstoEasternAna- toliaanditsresources(Fig.2).IndeedmostoftheobsidiansatTell Mozan (97%) came fromthe Eastern Anatolian sources, as antici- * Correspondingauthor.Tel.:þ447402990202. E-mailaddresses:e.frahm@sheffield.ac.uk,[email protected](E.Frahm), pated from the established regional distribution patterns (Fig.1; [email protected](J.M.Feinberg). Renfrew and Dixon,1976; Chataigner et al.,1998). We identified, 0305-4403/$eseefrontmatter(cid:1)2012ElsevierLtd.Allrightsreserved. http://dx.doi.org/10.1016/j.jas.2012.07.019 1124 E.Frahm,J.M.Feinberg/JournalofArchaeologicalScience40(2013)1122e1135 areas” (137). These workshops, however, are all Palaeolithic debated.ASumeriantablet,unearthedatNippurincentralIraqby through Chalcolithic. Despite extensive surveys, Bronze-Age thejointexpeditionoftheAmericanSchoolsofOrientalResearch obsidian quarries remain unknown (Balkan-Atlı et al., 2011). Our andtheOrientalInstituteofChicago,famouslydescribesAkkadian findings,usingrockmagnetismasameansofhigh-resolutionintra- territory, attesting that the god Dagan gifted Sargon “the upper flow obsidian sourcing, suggest these Göllü Dag(cid:2) artefacts at Tell land, (namely) Mari, Yarmuti, (and) Ebla, up to the Cedar Forest MozancamefromatleasttwoquarryinglocationsoftheKömürcü (and)uptotheSilverMountain”(translatedbyKramer,1963;see obsidiansource. Fig.1forsitelocations).ThusitisgenerallyheldthattheAkkadian TheoccurrenceofthisexoticCentralAnatolianobsidianatTell empire extended from Southern Mesopotamia into Northern Mozan represents a change in Northern Mesopotamian regional Mesopotamia and westward to the Eastern Mediterranean coast dynamicsduringtheAkkadianperiod(circa2350e2150BCE).The (andperhapsCyprus). movementofGöllüDag(cid:2)obsidiantoUrkeshatthistimewasmost Thereare,however,largeuncertaintiesaboutAkkadiangeopo- likelyaccompaniedbyeortriggeredbyesomeotherphenomenon litical interests and the nature of their “control” across Northern involving the material (e.g., objects, technologies, people) and/or Mesopotamia. Adams (1966) proposes that Akkadian influence social(e.g.,ideas,identities)world.Themechanismsandrouteby across the north “fell decisively short of full imperial control” whichtheartefactswerebroughttoUrkesharepresentlyambig- during the reign of Naram-Sin (159). Some researchers maintain uous,buttheirarrivalissynchronouswiththeknownheightofthe thattheAkkadiansexertedfullcontroloveragriculturalproduction Akkadianempire’sinfluenceatthecity.Avarietyofexplanations intheregion(WeissandCourty,1993).Othersproposetheirempire andcontextsareexploredhere.TheAkkadiansmayhaveeffected primarilyexertedinfluenceoverneighbouringpolities’distribution a period of exchange among Khabur-Triangle and Middle- routesforstrategicnaturalresources,particularlytogainaccessto Euphrates settlements, tapping into what S¸ahog(cid:2)lu (2005) terms such valuable highland materials as metals and timber (Nissen, theAnatolianTradeNetwork(ATN;Fig.1)andperhapsevendriven 1988; Michalowski,1993; Marcus,1998; Van De Mieroop, 2007). by access to Central Anatolian metals. Considering the similar Trade under the Akkadians is thought to have been principally occurrence of Göllü Dag(cid:2) obsidian in Bronze-Age Cretian palaces state-controlled,becomingaprivateendeavourduringthesubse- (Carter and Kilikoglou, 2007), the artefacts may have arrived at quent Old Assyrian Trade Colony period (OTAC, circa 1950e1750 Urkeshasprestigeitems,royalgifts,orsimilarly“activeagentsin BCE;Mallowan,1965;Veenhof,1997). the maintenance of complex social relations” (Gero, 1989: 103). ReadersinterestedinbroaderhistoricalcontextsofHurriansand This “exotic” obsidian may also reflect Akkadian-tied changes in Akkadians are directed to general books about this region and either identity construction of the Urkesh royals or territoriality period: Kurht (1995), Akkermans and Schwartz (2003), Snell involvingpastoralistsresponsibleforthearrivalofEasternAnato- (2005),andVanDeMieroop(2007).Additionally,aneditedbook lianobsidiansatTellMozan. on the socioeconomics of third-to second-millennium Northern MesopotamiaandAnatoliaisnewlypublished:Lanerietal.(2012). 2. Thepeople:HurriansandAkkadians 3. TheSite:TellMozanasUrkesh Knowledge regarding the Hurrians, who lived in Northern MesopotamiaduringatleastthethirdandsecondmillenniaBCE,is Tell Mozan sits at the north-middle of the Khabur Triangle in sofragmentarythatmanyauthorsusewordslike“mysterious”and northeastern Syria (Fig. 2), a transitional region between the “enigmatic” to discuss them. As explained by Wilhelm (1989), AnatolianmountainstothenorthandMesopotamianplainstothe Hurrianshad“oneofthemostimportantancientEasternciviliza- south.ThesettlementwasinhabitedsincetheHalafPeriod(sixth tions, and yet we have far less information, linguistic as well as millenniumBCE)andabandonedaround1300BCE,whenAssyrians historical and cultural, about them than we do about the Sume- assumed control across Northern Mesopotamia. Covering over rians, the Babylonians, the Assyrians, the Hittites, or the Canaan- 130ha(1.3km2)duringtheEarlyandMiddleBronzeAges,itwas ites” (v). Until recently, information about the Hurrians was among the largest cities in the region (Buccellati and Kelly- essentially limited to glyptic and epigraphic sources (e.g., name Buccellati,1988;Pustovoytovetal.,2011).Inaddition,itsitsnear lists, seal impressions, texts, and inscriptions) and subsequent theterminusoftheMardinPassintotheTaurusmountains’front linguisticanalyses.Basedonsuchevidence,theHurrianswerean range, the Tur Abdin. Akkermans and Schwartz (2003) propose ethnicminorityinNorthernMesopotamiansettlementsalongthe “Mozan’slocationatthenorthernedgeoftheKhaburplainsnear transitional zone between Anatolia and Southern Mesopotamia. theMardinsaddlemayindicatecontroloftheroutetothecopper ThusKurht(1995)reports,atthetime,“sourcesfortheHurriansare minesofeasternAnatolia”(285e286),inparticulardepositsnear exclusivelylinguistic: there are no artefacts or buildings that can Diyarbakır(Turkishfor“realmofcopper”). withanycertaintybedefinedas‘Hurrian’intype”(284).Thatyear, ThepoliticalimportanceofUrkeshismanifestedinthepalace Tell Mozanwas identified as the city of Urkesh, the political and complexonthetell’swesternside(AinFig.3).Thecomplexhastwo religiouscentreofthird-andsecond-millenniumHurrianculture, key components: a service wing and a formal-residential wing, previously known from texts preserved in Hittite archives bothwithcourtyards.Itwasconstructedduringthesite’sPhase2a (Buccellati and Kelly-Buccellati,1995). Forexample, a myth iden- andthereignofTupkish,aHurrianking,circaabout2260BCE.The tifiestheHurrianancestralgodKumarbiasthe“father”ofUrkesh, next two kings lived in this palace as well. The site’s religious whereheresides,anditsuggestskinshipbetweentheHurriansin significance is reflected in a temple (B), which sits atopa terrace Urkeshandpopulationsinthemountainousnorth(Hoffner,1990: thatrose30movertheplains,andamonumentalstonestaircase 46e47). Therefore, Tell Mozan is one of very few archaeological thatlinksthetempleterracetoaplaza(J). siteswheretheHurrianswereknowntohavebeentherulersand Despitesuchpoliticalandreligioussignificance,Urkeshdidnot amajorityofthepopulation. develop beyond a city-state (Buccellati, 2003). Instead, mountain TheAkkadianempire,centredinthecityofAkkadsomewhere villages in the region likely had Hurrian inhabitants who were inSouthernMesopotamia,reachedapoliticalapexfromthe24thto culturally linked to Urkesh. Such ties facilitated access to the 22ndcenturiesBCEafteraseriesofpurportedconquestsbySargon highland resources “even if the kings exerted no direct adminis- ofAkkadandhissuccessors.Somesuggestthisdynastyestablished trativeormilitarycontrolovertheruralhinterland”(Buccellatiand thefirstempireinhistory(Liverani,1993),butitsextentismuch Kelly-Buccellati,2001:27).Suchanorganisationofresourcecontrol E.Frahm,J.M.Feinberg/JournalofArchaeologicalScience40(2013)1122e1135 1125 Fig.3. ThelocationsofUnitsA7andA9relativetothepalaceexcavations(left)andAreasA,B,andJonthetell(right).CompiledandredrawnfromvariousUrkeshexpeditionmaps. mayhavebeenareasonthatUrkeshmightnothavefallenunder compared artefacts to geo-referenced geological specimens using Akkadianconquest(BuccellatiandKelly-Buccellati,2002).Instead, amultidisciplinaryapproachinvolvinggeochemistry(asameansto anAkkadianking,Naram-Sin(thegrandsonofSargon),mayhave identify “source” on the scale of individual obsidian-bearing lava established an alliance with Urkesh through the marriage of his flowsatavolcano)androckmagnetism(asameanstodistinguish daughterTar’am-AgadetoanunknownHurrianruler. “source”onthescaleofindividualquarryinglocationsataflow). 4. Theory:Akkadianinfluence 5.1. Obsidianassemblageandsourcingsample The precise nature of Akkadian influence at Urkesh has been ObsidianaccountsforathirdoftheTellMozanlithicassemblage a topic of considerable debate. The discussion centres on seal whiletheothertwo-thirdsarecherts.Whereasobsidianmustbe impressionsthatread“[Of]Naram-Sin,thekingofAkkad,Tar’am- obtainedfromAnatolia,chertsareavailableasnodulesalongriver Agade,hisdaughter”(BuccellatiandKelly-Buccellati,2002:13).Her and stream beds throughout the region. Though Tell Mozan was sealings were excavated from Phase 2b strata of the palace continuously inhabited since the Halaf period (circa 6200e complex, meaning that she would have lived there soon after 5300 BCE), our work focused on the Early Bronze Age III (2700e Tupkish,whoseruledatestotheinitialportionofNaram-Sin’sreign 2200 BCE) to Late Bronze Age II (1400e1300 BCE). During this or a bit earlier. Hence strata before the palace date to the Early period,theassemblageisdominatedbytwotooltypes:(1)adhoc Akkadianperiod(circa2260BCE;Phase2a),TupkishandTar’am- flake tools and (2) blade tools, including prismatic blades, Agadedatetothemid-Akkadianperiod(circa2240BCE;Phase2b), segments, and blade-based tools such as geometric microliths, and the last king beforethe palace’s abandonment, Ishar-k(cid:3)anum, scrapers,andpoints.Chertandobsidiantoolswereretainedbeside datestotheLateAkkadian(circa2200BCE;Phase3). copperanditsalloys,whichwereusedtocraftobjectssuchasspear The implications of Tar’am-Agade at Urkesh are uncertain. pointsandpins.Fragmentsofanobsidianvesselandothercarved/ BuccellatiandKelly-Buccellati(2002)arguethatshewasmarriedto groundobjects,includingincisedbeads,datingtothisperiodhave thekingwhosucceededTupkish.Thisinterpretationsupportsthe alsobeendiscoveredatTellMozan. proposedalliancebetweenAkkadandUrkesh.Thealternative,that Afteranon-sitesurveyof820obsidianartefacts,asampleof97 Tar’am-Agade was there to either accompany or serve as an artefactswasapprovedforexportbytheSyrianDirectorateGeneral Akkadian administrator, cannot be discounted. Such an official at ofAntiquitiesandMuseums.Thus12%oftheobsidianassemblage UrkeshmightindicateAkkadiancontrolifnotconquest.Akkermans was chemically sourced. To gain export approval, artefacts were andSchwartz(2003)contendthatthe“discovery[ofhersealings]is non-diagnosticchipdebris(Figs.4and5).Thespatiotemporalspan of considerable historical importance, but the significance of of these artefacts reflects the site’s most recent excavations: 81 Tar’am-Agade’s presence is, as yet, elusive” (285). The situation artefactsfromAreaA(thepalacecomplex),13artefactsfromJ(the regardingTar’am-Agadereflectstheaforementioneduncertainties templeterraceandadjacentplaza),andthreeartefactsfromB(the aboutAkkadian“control”acrossNorthernMesopotamia.Itseems temple complex). The97 artefacts spanmorethanamillennium, mostlikely,however,iftheAkkadiansindeedsoughtcontrolover fromabout2400BCEtothesite’sabandonmentcirca1300BCE.The resources,changesatUrkeshmaybeexpectedwhentheirinfluence export agreement specified only non-destructive analyses of the wasgreatest,thatis,duringTar’am-Agade’spresence. artefacts. 5. Methodsandmaterials:artefactsandanalyses 5.2. Geochemicalanalysesforsourceidentification OuridentificationofKömürcüobsidianartefactsatTellMozan The artefacts’ chemical compositions were compared to over began with an on-site survey of the lithic assemblage, and we 900geo-referencedgeologicalobsidianspecimens,includingover 1126 E.Frahm,J.M.Feinberg/JournalofArchaeologicalScience40(2013)1122e1135 Fig.4. SourcedartefactsfromUnitA7:(left)obsidianfromGöllüDag(cid:2)(specificallyKömürcü)inCentralAnatoliaand(right)obsidiansfromEasternAnatoliansources. 450fromEasternAnatolia,280fromCentralAnatolia,and170from originated. Such flows, however, may cover a dozen or more the Caucasus. These artefacts and geological specimens were square kilometres, and layers of high-quality obsidian are buried geochemicallyanalysedusing twoanalytical techniques: electron across much of the flow (Fig. 7 in Frahm, 2012b). Obsidian is microprobe analysis (EMPA; using the procedures published in availablewherethislayerwasexposedbyerosion,tectonics,and/ Frahm, 2012a) and portable X-ray fluorescence (pXRF; Frahm, in or slope processes. The Kömürcü source, as noted in the Intro- press). EMPA has been previously used for obsidian sourcing by duction,consistsofobsidianexposureswherestreamsanderosion MerrickandBrown(1984)inEastAfrica,WeislerandClague(1998) have sliced into the Paleo-Kabak Tepe lava dome. Hence people in Hawaii, and Tykot(1995) in the Mediterranean and elsewhere would have collected high-quality obsidian from discrete loca- (TykotandChia,1997;Rosenetal.,2005).Obsidiansourcingusing tions, such as the aforementioned workshop sites near the pXRFhasbeenmainlytestedinEastAsia(PhillipsandSpeakman, obsidian exposures. These quarries, as we call them here, are 2009; Jia et al., 2010), the Americas (Craig et al., 2010; Nazaroff specificlocationswithinachemicallyhomogeneous(ornearlyso) et al., 2010; Millhauser et al., 2011), and Oceania (Burley et al., obsidianflow.Thusarchaeologicallyimportantactivities,including 2011;McCoyetal.,2011;Sheppardetal.,2011).Acombinationof the first stages of the reduction sequence, occurred at spatial theseanalyticaltechniquesandsubsequentdataanalysesallowed resolutionsthatconventionalgeochemicalsourcingstudiescannot us to detect subtle distinctions amongst geochemically similar typically resolve. Different groups and/or lithic traditions may sources(Frahm,2012b).Manypriorstudiesreportdifficultiesdis- also be associated with different quarries of a single flow (Davis tinguishingKömürcüfromotherlavaflowsontheeasternflanksof etal.,1992). GöllüDag(cid:2)(i.e.,EastKayırlı,EastBozköy)anduseacombined“East Different portions of an obsidian-bearing flow cooled at GöllüDag(cid:2)”sourceidentification(e.g.,Pernickaetal.,1997;Bellot- different rates and thus experienced different ranges of temper- Gurlet,1998;Chataigneretal.,1998;Poidevin,1998;Abbèsetal., atures, viscosity, and oxidation conditions. Within obsidian, 2001, 2003; Bressy et al., 2005; Carter and Shackley, 2007; microscopicmineralsaresuspendedintheglass,andtheirsizes, Poupeauetal.,2010). shapes, physical arrangements, and mineralogyare highlysensi- tive to the local cooling history. This, in turn, affects the rock 5.3. Magneticanalysesforquarrydistinction magneticpropertiesoftheobsidian.Magneticanalysisofobsidian with the goal of artefact sourcing is not new (e.g., McDougall, Conventionalgeochemicalsourcingtechniquesareoftenableto 1978; McDougall et al.,1983; Hammo,1984,1985; Schmidbauer identify the particular lava flow fromwhich an obsidian artefact et al., 1986; Urrutia Fucugauchi, 1999; Vásquez et al., 2001; Fig.5. SourcedartefactsfromUnitA9:(left)obsidianfromGöllüDag(cid:2)(specificallyKömürcü)inCentralAnatoliaand(right)obsidiansfromEasternAnatoliansources. E.Frahm,J.M.Feinberg/JournalofArchaeologicalScience40(2013)1122e1135 1127 any, spatial information). In contrast, hysteresis parameters e saturation magnetisation (Ms), saturation remanence (Mrs), coer- civity(Hc),andcoercivityofremanence(Hcr)eareaffectedmore directly by the local cooling history (and hence contain spatial informationas differentparts of a flowcooled at different rates). Thesemagneticparametersweredeterminedforall97TellMozan artefactsand704ofthegeologicalreferencespecimensviamajor hysteresis loops and back-field measurements. A paper on our techniques is currently in preparation (Feinberg and Frahm, in preparation), but here we present the first archaeologically significantresults. 6. Results:sourcedartefactsandtheircontexts Theresultsofourgeochemicalandmagneticanalysesrevealed the unexpected presence of Central Anatolian obsidian at Tell Fig.6. EMPAmeasurementsofAl2O3,TiO2,andZrdistinguishthreeartefacts(fullblack Mozan, potentially from two quarries at the Kömürcü source of circles)thatmatchthegeologicalspecimensfromGöllüDag(cid:2)(greyXs). Göllü Dag(cid:2); however, their stratigraphic contexts are equally important. Stewart et al., 2003; Weaver et al., 2005, 2009; Sternberg et al., 2010, 2011; Zanella et al., 2012); however, its application has 6.1. Geochemicalresults:KömürcüatGöllüDag(cid:2) been minor compared to conventional geochemical techniques. Many of these studies have had mixed success, usually due to AllTellMozanobsidianartefactswereexpectedtohaveEastern large variability in obsidian magnetic properties within a single Anatolianorigins.PreviousstudiesforBronze-AgeandChalcolithic flow. Because these studies attempted to distinguish amongst sitesintheKhaburTriangleattributedartefactsonlytotheEastern different sources (that is, the same goal for which geochemical Anatolian obsidian sources (Table 1). Only one artefact found on techniques have excelled for the last five decades), such a high the surface of Tell Halaf, lyingon thewesternmost fringes of the degree of intra-flow variation has been problematic (but not Khabur Triangle (Fig. 2), came from Göllü Dag(cid:2), although the necessarilydetrimental). precise source there is not known (Francaviglia and Palmieri, Ourapproachisnovelandusesthermal-dependenteandthus 1998). spatial-dependent e magnetic properties of obsidian to distin- ThreeofthesourcedartefactsfromTellMozan,however,came guish amongst different quarries (Feinberg et al., 2009; Johnson fromGöllüDag(cid:2),specificallytheKömürcüsourceonthevolcano’s et al., 2009; Hillis et al., 2010; Frahm, 2012c; Feinberg and easternslopes.Figs.6and7reveal,basedonbothEMPAandpXRF Frahm, in preparation.). Thus, some of the variations that hinder measurements, three obsidian artefacts are chemically distinct inter-flow differentiation are the very mechanisms that make from the others but match the Göllü Dag(cid:2) geological specimens. intra-flowdifferentiationpossible. Inparticular, magnetic param- Table2showsthat,whenthesameEMPAandpXRFdataareana- eters such as susceptibility (c) and natural remanent magnet- lysed in multi-dimensional space, Kömürcü obsidian is the best isation(NRM)areoftentoodependentonthemereabundanceof matchforallthreeartefacts.Asmentionedpreviously,priorstudies magnetic mineral grains to be useful (and hence contain little, if have had difficulties distinguishing Kömürcü obsidian from the surrounding sources. Our two independent analytical techniques bothidentifiedKömürcüobsidianasthebestmatch,corroborating theresult.Onekeytosuccesswasourreferencecollection,which includes82specimensfromGöllüDag(cid:2)and20fromthreeKömürcü locations. 6.2. Magneticresults:twoquarriesatKömürcü After Kömürcü was geochemically identified as their exact source, the three artefacts were compared magnetically to the geologicalspecimensfromthreeKömürcüoutcrops.Nineofthe20 specimens were suitable for measuring the aforementioned hysteresis parameters. Fig. 8 shows the artefacts and Kömürcü geologicalspecimensplottedusingthreeoftheseparameters.Note the geological specimens from three different Kömürcü locations fall into three discrete clusters. One of these locations (the black squares)was the famed Kaletepeworkshop area, whereobsidian wasworkedfromthePalaeolithictoChalcolithic(Balkan-Atlıetal., 1999,2011).ThemagneticresultssuggesttheTellMozanartefacts didnotoriginatefromanyofthesecollectionlocations,including theKaletepeworkshop,whichisconsistentwithalackofBronze Agematerialthere.Furthermore,theartefactsfallintotwoclusters, Fig.7. pXRFmeasurementsofRb,Zn,andZrdistinguishthreeartefacts(fullblack suggestingtheobsidiancamefromatleasttwodifferentquarries. circles)thatmatchthegeologicalspecimensfromGöllüDag(cid:2)(greyXs).ThepXRFdata WithoutadditionalsamplingattheKömürcüsource,wecannotbe werecalibratedusingasetof18Anatolianobsidianspecimensanalysedusingneutron activation analysis (NAA) and laboratory-based XRFat the Universityof Missouri’s more precise about the quarrying locations. At present, though, ResearchReactorCentre. we can propose that obsidian for the Tell Mozan artefacts came 1128 E.Frahm,J.M.Feinberg/JournalofArchaeologicalScience40(2013)1122e1135 Table1 TheresultsfrompublishedobsidiansourcingstudiesinNorthernMesopotamia(Fig.2).AsshowninFig.1,theMiddleEuphratessitesfallneartheedgeofreconstructedCentral Anatolianobsidiandistributionzones.IntheKhaburTriangle,justoneartefactfoundonthesurfaceofTellHalaf,onthewesternmostborderofthebasin,originatedinCentral Anatolia.CentralAnatoliansourceassignmentsarehighlightedinbold. Region(WesttoEast) Site Period Reference Artefacts’sourceassignments n Region Source MiddleEuphrates(NorthtoSouth) Dja’de Neolithic Pernickaetal.,1997 3 CentralAnatolia GöllüDag(cid:2) 2 EasternAnatolia BingölB 1 EasternAnatolia BingölA/NemrutDag(cid:2) TellHalula Neolithic Pernickaetal.,1997 7 EasternAnatolia BingölA/NemrutDag(cid:2) 6 CentralAnatolia GöllüDag(cid:2) 4 EasternAnatolia BingölB 2 undetermined 1 CentralAnatolia NeneziDag(cid:2) JerfelAhmar Neolithic Pernickaetal.,1997, 23 EasternAnatolia BingölB Abbèsetal.,2001,2003 22 CentralAnatolia GöllüDag(cid:2) CheikhHassan Neolithic Gratuzeetal.,1993, 17 CentralAnatolia GöllüDag(cid:2) Pernickaetal.,1997, 5 EasternAnatolia BingölB Abbèsetal.,2001,2003 1 EasternAnatolia BingölA/NemrutDag(cid:2) 1 Undetermined Mureybet Neolithic Gratuzeetal.,1993, 53 CentralAnatolia GöllüDag(cid:2) Pernickaetal.,1997, 5 EasternAnatolia BingölA/NemrutDag(cid:2) Abbèsetal.,2001,2003 2 EasternAnatolia BingölB AbuHureyra Neolithic McDanielsetal.,1980 47 EasternAnatolia BingölA/NemrutDag(cid:2) 28 EasternAnatolia BingölB 24 CentralAnatolia GöllüDag(cid:2) BalikhRiver TellAssouad Neolithic Gratuzeetal.,1993 3 EasternAnatolia BingölB 2 EasternAnatolia BingölA/NemrutDag(cid:2) KhaburTriangle(WesttoEast) TellHalaf surfacefinds FrancavigliaandPalmieri,1998 2 EasternAnatolia BingölA/NemrutDag(cid:2) 2 EasternAnatolia MeydanDag(cid:2) 1 EasternAnatolia BingölB? 1 CentralAnatolia GöllüDag(cid:2) 1 undetermined TellKashkashok LateNeolithic Gratuzeetal.,1993 4 EasternAnatolia BingölA/NemrutDag(cid:2) 4 EasternAnatolia BingölB TellGudeda EarlyBronzeAge Chabotetal.,2001 4 EasternAnatolia BingölA/NemrutDag(cid:2) TellMullaMatar BronzeAge Pernickaetal.,1997 1 EasternAnatolia BingölA/NemrutDag(cid:2) Tell‘Atij EarlyBronzeAge Chabotetal.,2001 4 EasternAnatolia BingölA/NemrutDag(cid:2) 2 undetermined ChagarBazar Chalcolithic CannandRenfrew,1964 1 EasternAnatolia BingölA/NemrutDag(cid:2) LateNeolithic CannandRenfrew,1964 1 EasternAnatolia MeydanDag(cid:2)? TellBrak unknown ForsterandGrave,2012 4 EasternAnatolia BingölA/NemrutDag(cid:2) LateChalcolithic Khalidietal.,2009 4 EasternAnatolia BingölA/NemrutDag(cid:2) 3 EasternAnatolia BingölB 1 EasternAnatolia MeydanDag(cid:2) Surfacefinds FrancavigliaandPalmieri,1998 4 EasternAnatolia BingölA/NemrutDag(cid:2) 1 EasternAnatolia BingölB? TellBarri Surfacefinds FrancavigliaandPalmieri,1998 18 EasternAnatolia BingölA/NemrutDag(cid:2) 2 EasternAnatolia MeydanDag(cid:2) 2 EasternAnatolia BingölB? TellHamoukar Surfacefinds HallandShackley,1994 9 EasternAnatolia BingölA/NemrutDag(cid:2) 1 undetermined LateChalcolithic Khalidietal.,2009 27 EasternAnatolia BingölA/NemrutDag(cid:2) 2 EasternAnatolia BingölB 2 undetermined 1 EasternAnatolia MeydanDag(cid:2) Surfacefinds FrancavigliaandPalmieri,1998 16 EasternAnatolia BingölA/NemrutDag(cid:2) HirbetTueris Surfacefinds HallandShackley,1994 11 EasternAnatolia BingölA/NemrutDag(cid:2) from two different quarries, neither of which was the Kaletepe somewhatconjecturalinthattheircontinuitycannotyetbeshown workshop. in a single exposed profile. Table 3 summarises the stratigraphic, chronological, and historical data for A7 and A9 as well as the 6.3. Stratigraphiccontext:servicecourtyardofthepalace quantitiesandtypesofobsidianartefactsintheirstrata.Notethat, forbothunits,almostone-thirdoftheexcavatedobsidianartefacts Theartefacts’contextsareassignificantastheirorigins.Twoof weresourced.Mostimportantare thecontextsfortheGöllüDag(cid:2) theartefactswerefoundinUnitA9andoneinUnitA7(Fig.3),both obsidian.InA9,forexample,twoobsidianartefactswererecovered ofwhichincludethepalace’sservicecourtyardanditsfoundations from Feature 156, an accumulation on a pebble surface likely aswellasstratafromlaterphases(inparticularPhase4,thefirst deposited while craft activities were conducted in the courtyard, post-palacesettlementperiodcirca2100e2000BCE,andPhase5, andbothartefactsareassignedtoGöllüDag(cid:2).Thestrataaboveand thesecondsuchperiodcirca2000e1800BCE).UnitA2separatesA7 belowcontainedonlyEasternAnatolianobsidians.Thesameistrue andA9,andithasbeenexcavatedonlythroughthePhases4and5 in A7: Göllü Dag(cid:2) at the service courtyard entrance with Eastern strata. Hence, the stratigraphic links between A7 and A9 remain Anatolian obsidians above and below. Additionally, the magnetic E.Frahm,J.M.Feinberg/JournalofArchaeologicalScience40(2013)1122e1135 1129 Table2 Threeartefacts’sourceassignmentsbasedonpXRFandEMPAdata.GöllüDag(cid:2)isthebestmatchforthesourcecomplex,andamongtheGöllüDag(cid:2)sources,theKömürcüsource isthebestmatch.Bestmatchesaredeterminedbythehighestproportionofnearestneighbours(%NN)basedonmulti-dimensionalEuclideandistancematrices. Artefact pXRFsourceassignments EMPAsourceassignments Sourcecomplex Source/outcrop Sourcecomplex Source/outcrop Bestmatch %NN Bestmatch %NN Bestmatch %NN Bestmatch %NN A7q892f261k12 GoöllüDag(cid:2) 100% Kömürcü 32% GoöllüDag(cid:2) 96% Kömürcü 95% A9q463f156k3piece1 GoöllüDag(cid:2) 100% Kömürcü 38% GoöllüDag(cid:2) 93% Kömürcü 80% A9q463f156k3piece2 GoöllüDag(cid:2) 100% Kömürcü 30% GoöllüDag(cid:2) 43% Kömürcü 31% data suggest (1) the two A9 artefacts originated from different tradepersebutinsteadthetradeofanothermaterialorobject,the quarries and (2) the one A7 artefact might have come from the movementofpeople,oranotherphenomenonaltogether,including samequarryasoneA9artefact. territoriality,diplomacy,oreliteidentity. TheseresultsarestratigraphicallyshowninFig.9,highlighting use of Göllü Dag(cid:2) obsidian soon after the service courtyard’s 7.1. ArrivalviatheMiddleEuphrates? constructiondespiteuseofEasternAnatolianobsidiansbeforeand after.Therefore,GöllüDag(cid:2)obsidian,potentiallyfromtwodifferent quarriesatKömürcü,coincideswiththepresenceofTar’am-Agade. Obsidian from Göllü Dag(cid:2) certainly did not follow a 600-km straight line to Urkesh. Instead, a likely path seems a west-east Given the limitations of negative evidence, we cannot strongly assert that use of Göllü Dag(cid:2) obsidian was exclusive during this route from the Middle Euphrates, where settlements frequently used Göllü Dag(cid:2) obsidian, at least during the Neolithic (Table 1, period.Asofyet,however,noothersourcedartefactscanbetimed so preciselywithTar’am-Agade,so mechanismsforadecrease in Figs.1 and 2). No post-Neolithic sourcing studies are published for Middle Euphrates sites, so concurrent data are unavailable. thearrival ofEasternAnatolianobsidiansaregivensomeconsid- Such a route is consistent with the Göllü Dag(cid:2) artefact from the eration.Itisequallylikely,ifnotmoreso,thatobsidianinthepalace surface of Tell Halaf, on the far western edge of the Khabur service courtyard arrived at Urkesh through mechanisms distinct Triangle.ItisalsocongruouswiththeAkkadiansaccessing,what fromobsidianinothercontexts.Severalpossibleexplanationsare S¸ahog(cid:2)lu (2005) terms the Anatolian Trade Network (hereafter discussedbelow. ATN;circa2500e2100BCE;Fig.1).GöllüDag(cid:2)liesnearthenexus of the ATN, which reached the Middle Euphrates. It has been 7. Interpretationanddiscussion suggested that Tell es-Sweyhat, located in Middle Euphrates (near Jerf el Ahmar, Fig. 2), was the location of an Akkadian Determining the nature of Akkadian influence at Urkesh is trading enclave, providing support for such a route. Most aheavyburdentoplaceonthreeexoticobsidianartefacts.Wecan, recently, though, the excavators have interpreted the area in though, consider a number of hypotheses regarding mechanisms questionasaproductioncentreandcaravanseraionawest-east fortheartefacts’arrivalinthepalacecomplex.Westressthat,when trade route (Zettler,1997; Holland, 2002), remaining consistent discussing “obsidian trade” in third-millennium Northern Meso- withourhypothesisedpath. potamia,suchexchangeshouldnotnecessarilybeconceptualised BytakingadvantageoftheATN,Naram-Sin,underwhoserule asobsidian-driven.Instead,itshouldbeunderstoodmorebroadly the Akkadian empire reached its greatest extent, could have assomephenomenoninwhichobsidianplayedeitheraprimaryor gained access to various Central Anatolian resources (including secondary role. Obsidian distribution might not reflect its own GöllüDag(cid:2)obsidian),evenwithoutextendingAkkadianinfluence far beyond the Middle Euphrates. When the ATN (2500e 2100 BCE) and later OATC system (1950e1750 BCE) are consid- ered together, there was an established trade network between CentralAnatoliaandNorthernMesopotamiaformuchofthespan between 2500 and 1750 BCE.In the latter period, exchangewas intensive and managed in “modern” ways (Veenhof,1997). Yet, despitetheseknownlong-distanceroutes,GöllüDag(cid:2)obsidianis only known at Urkesh concurrently with Naram-Sin’s reign and Tar’am-Agadeinthepalace.TheAkkadians,directlyorindirectly, may have been an impetus for exchange amongst Khabur- Triangle (e.g., Urkesh) and Middle-Euphrates (e.g., Tell es- Sweyhat)settlements. ConsiderthattheGöllüDag(cid:2)obsidianartefactswereunearthed intheservicecourtyardofthepalacecomplex.Urkeshlikelyhad,at leastinpart,apalaceeconomy.Thecomplex,particularlyitsservice courtyard,mayhaveservedasmorethanjusttheroyalresidence and centre of government. It also possibly served as a craft productioncentre,amarketplacethatadministereddistributionof goods,andacaravanseraithatfacilitatedlong-distanceexchange. Fig.8. Aplotofthreemagnetichysteresisparametersesaturationremanence(Mrs), Depositionofobsidianartefactsmighthaveoccurredthroughany coercivity(Hc),andcoercivityofremanence(Hcr)erevealsthatthegeologicalspeci- of theseactivities. Althoughthe mechanismcannotbeidentified, mensfromthreedifferentKömürcülocations(solidcircles,squares,andtriangle)fall eachscenarioimpliessomedegreeofpalacecontroloroversight. intodiscreteclusters.ThethreeTellMozanartefacts(opencircles)fallintotwodistinct Thisraises thedistinct possibilitythatobsidianarrived atUrkesh clusters,suggestingthattheycamefromatleasttwoquarriesdifferentfromthethree geologicalcollectionlocations. embeddedwithinsomeotherpalaceactivity. 1130 E.Frahm,J.M.Feinberg/JournalofArchaeologicalScience40(2013)1122e1135 Table3 Summarisedstratigraphy,chronology,andobsidianartefacts(includingartefacttypesandproportionsourced)forunitsA7andA9.ThefeatureswithCentralAnatolian obsidiansarehighlightedinbold. Unit Feature Phase Approxdate Period Relevant Obsidian Basicartefacttypes Sourcedartefacts Feature/stratumcomments (MiddleChron) individuals n n % A7 f56 Phase4/5 2100-1800BCE UrIII/Old 2 Flake/debitage(1), 1 50% Post-palacesettlementperiod; Babylonian blade(1) housesandburialsrevealan f63 2 Flake/debitage(1), 1 50% intensivesettlementover blade(1) formerpalace f69 2 Flake/debitage(1), 1 50% scraper?(1) f121 Phase3? 2200-2100BCE LateAkkadian 6 Flake/debitage(5), 5 83% Accumulationsincourtyardarea blade(1) whileitwasanopenspacesoon f148 2 Flake/debitage(1), 1 50% afterthepalacewasabandoned blade(1) f261 Phase2b 2240BCE mid-Akkadian Tar’am-Agade 2 Flake/debitage(1), 1 50% Accumulationatentranceto blade(1) servicecourtyard f465 Phase2a 2260BCE EarlyAkkadian Tupkish 4 Flake/debitage(3), 3 75% Foundationsofthepalacecomplex blade(1) anditsservicecourtyard;both f480 1 Flake/debitage(1) 1 100% datetothepalace’sconstruction 21 A7totals-sourced 14 67% strata 48 A7totals-allstrata 14 29% A9 f126 Phase4? 2100-2000BCE UrIII 8 Flake/debitage(7), 3 38% Scatteredpost-palaceoccupation; core?(1) sparsehouses f98 Phase3 2200-2100BCE LateAkkadian 10 Flake/debitage(9), 6 60% Accumulation,openspaceafter scraper(1) palaceabandoned f156 Phase2b 2240BCE mid-Akkadian Tar’am-Agade 2 Flake/debitage(2) 2 100% Accumulationonthecourtyard’s pebblesurface f247 Phase2a 2260BCE EarlyAkkadian Tupkish 5 Flake/debitage(3), 3 60% Foundationsofservicecourtyard; blade(2) f247sitsdirectlyatopf260;both f260 2 Flake/debitage(2) 2 100% datetothepalace’sconstruction 27 A9totalsesourced 16 59% strata 56 A9totalseallstrata 16 29% 7.2. AegeanParallels:“piggybacking”onmetals? Dag(cid:2) volcano havedeposits rich in silverand copper (Yeneret al., 1989; Yener and Vandiver, 1993). Hence Carter and Kilikoglou AsimilarsituationoccursinBronze-AgeCrete,whereGöllüDag(cid:2) (2007) suggest that Göllü Dag(cid:2) obsidian arrived on the island obsidians have been rarely found alongside artefacts from the through“piggybacking”onthemetalstradeviaanetworksuchas abundant Aegean obsidian sources (Melos and Giali). Carter and theATNorOATC. Kilikoglou (2007) sourced 60 Middle Bronze Age artefacts (circa This proposal is supported by lead isotope analyses of Cretan 20th century BCE) fromMalia in central Creteand identified five copper artefacts, which have signatures consistent with Central artefacts of Göllü Dag(cid:2) obsidian. They propose GöllüDag(cid:2) obsidian AnatolianoresbeginningintheMiddleBronzeAge(SolesandStos- reachedCreteduetoexploitationofa“crucialmetalliferouszone” Gale,2004;PoursatandLoubet,2005).Thus,asmetalmovedfrom surroundingthevolcano(YenerandVandiver,1993:238;seealso CentralAnatoliatoCrete,sotoomayhaveobsidian.Furthermore, Wiener,1991:327e328).Indeed,theTaurusmountainsnearGöllü they state obsidian distribution may have been embedded in Fig.9. AsimplifiedstratigraphyofthesourcedobsidianartefactsinUnitsA7andA9.OnlyGöllüDag(cid:2)obsidianwasidentifiedsoonafterthepalaceservicecourtyard’sconstruction despiteuseofEasternAnatolianobsidiansbeforeandafter. E.Frahm,J.M.Feinberg/JournalofArchaeologicalScience40(2013)1122e1135 1131 diplomatic contacts between Crete’s first “Minoan palaces” and silver)implytheAkkadiansmayhaveaccessedthesemetalsfrom Central Anatolian kingdoms.In other words, theobsidian arrived Central Anatolia via a supra-regional exchange network. Would, onCrete,theyhypothesise,intheformof“royalgifts”orprestige though,CentralAnatolian metals have supplemented or replaced objects.GöllüDag(cid:2) obsidianhasbeenchemicallyidentifiedatone Diyarbakırresources?Thesoleevidencewehaveatpresent(albeit other Proto-palatial Cretan site: Knossos, specifically within the only negative evidence with all the associated caveats) is that Central Palace Sanctuary (Wiener,1991; Panagiotaki,1998,1999). Eastern Anatolian obsidians are not yet identified in the same Theyalsonote that GöllüDag(cid:2) lies near Kültepe, a majorregional stratum as the Göllü Dag(cid:2) obsidian. Thus it is worth at least trading centre that served as a key node of the OATC network consideringscenarioswherebyEasternAnatolianobsidiansmight (Veenhof,1972;SherrattandSherratt,1991;Dercksen,1996,2005), havedecreasedat,orevendisappearedfrom,Urkesh(althoughthe andastorehouseintheKülteperoyaltemplecomplexheld3000kg burdenonourartefactsmaynowbetoogreat).Todosoinvolves ofobsidiansourced,atleastinpart,toGöllüDag(cid:2)(KatsujiKobayashi discussing probable mechanisms by which Eastern Anatolian inCarterandKilikoglou,2007;Özguç,1996). obsidiansarrived. TheGöllüDag(cid:2)obsidianatUrkeshmayrepresentaverysimilar phenomenon.Theartefacts may notreflect awest-east“obsidian 7.4. EasternAnatolianobsidiansandtranshumance trade”somuchasametalstradeinwhichGöllüDag(cid:2)obsidianwas an embedded element either deliberately or incidentally. The Eastern Anatolian obsidians were likely brought to Urkesh, at obsidianstillcouldhavebeenanimportantcomponentofexchange leastinpart,bytranshumantnomads.Therolesoftranshumancein althoughnotwhatcompelledtheAkkadianempiretoparticipatein obsidian distribution, particularly in Eastern Anatolia, have been CentralAnatoliannetworks.Theimpetusmayhavebeenmetals. long discussed (e.g., Hole, 1968; Wright, 1969; Crawford, 1978; Muchhasbeenmadeofthereferenceto“theSilverMountain” Williams-Thorpe,1995;Cauvin,1996;Chataigner,1998,Chataigner regarding Akkadian territory in the aforementioned tablet from et al., 1998). Archaeological evidence supports this subsistence Nippur;however,itdoeshintataninterestinthismetal.Infact, modelintheregionformillennia(UrandHammer,2009)andeven Mesopotamian silver artefacts may support this interpretation. suggestsawidespreadreturntonomadisminthethirdmillennium AlmostallsilverknownintheNearEastisderivedfromsilver-rich BCE (Sallaberger, 2007). In addition, we have third- and second- galena,aleadsulphideoreusedinantiquity(Moorey,1985:107e millennium accounts of nomadic and semi-nomadic peoples in 126, Ebeling and Meissner, 1997: 130). Therefore, all silver arte- NorthernMesopotamia(e.g.,Edzard,1976;Postgate,1976).Conse- facts contain trace quantities of lead, and a number of third- quently, regarding Hurrian populations living in the mountains millennium silver Mesopotamian artefacts have lead isotope through the Mardin Pass, Wilhelm (1989) asserts that they most signatures consistent with Central Anatolian ores (Yener et al., likely practiced “varied forms of ‘mountain nomadism’” supple- 1991:561).Itispossible,underNaram-Sin,theAkkadianstapped mentedbytradewithvillagersandcity-dwellers(16). intothe ATN to access silver and other metals from ore deposits While acknowledging differences in politics and climate within the vicinity of Göllü Dag(cid:2), and obsidian distribution was betweentheBronzeAgeandtoday,wecanconsiderethnohistorical insteadasecondaryphenomenon. accountsfromthisregion.Cribb(1991)documentsinteractionsof Because the Göllü Dag(cid:2) obsidian was discovered in the palace Alikan nomads, who summered at Nemrut Dag(cid:2) volcano, with complex,itispossiblethattheartefactsdidnotarriveatUrkeshvia villagerstothesouth: simple “down-the-line” exchange but instead as gifts or other Duringthewintermonths,thetribeisdispersedinsmallunitsof prestige items intended for royalty. Later second-millennium twotofivetentspitchedwithinorontheoutskirtsofvillagesto Akkadian-language tablets also refer to obsidian (ṣurru) having thesouthoftheTaurusMountains.Althoughthevillagershave economic, apotropaic, and symbolic value and being suitable as no tribal or kinship connection with the Alikan As¸iret, the akinglygift(Gelbetal.,1962:257e258).Analogoustothesituation wintering nomads become a temporary part of the village at Malia on Crete, obsidian might have been a component of community, drawing on its services and land resources, and “international relations” in the Bronze Age and served as a gift coming under the authority and protection of the [governor]. between elites to accompany metal shipments and/ordiplomatic (198) communiqués.Thispossibilityraisesaseriesofquestions.Towhom at Urkesh would the gifts be sent? An Akkadian administrator in The Alikan were integrated into Tur Abdin villages and their appreciation for trade conducted for the good of the empire? A economiesduringwinter,andnaturalresourcesfromtheirsummer Hurriankingresistingagainstit?Asaroyalgift,obsidianbecomes pasturesmayhavebeenusedforbarter.Crawford(1978)pointsout, lessabouttradeandmoreaboutdiplomacy. basedonhistoricalrecords,migrationsofnomadicgroupsreadily produce a criss-crossing de facto exchange network. Surveys of 7.3. Ametalsparadox? nomadicpastoralists,conductedduringthe1930sFrenchoccupa- tionofSyria,mentionKurdishpastoralistswhoseasonallymigrated Considering the circulation of metal and other materials with between Lake Van and Jebel Sinjar (Fig. 10; Haut-Commissariat, themovementofobsidianintheNearEastisnotanewsuggestion 1930). Other groups migrated between Jebel Sinjar and central (e.g., Wright, 1969), and given the proximity of the Diyarbakır Iraq (British Naval Intelligence, 1944). Yet other nomads moved copperminestotheMardinPassandTellMozan,thepotentialfor east-west and intersected these groups, enabling an intersecting interwovendistributionhasbeenconsideredbefore.Inthecaseof exchange networkamongstthem.Furthermore,theytradedwith Akkadians potentiallyaccessing Central Anatolian metals such as villagers along their routes. Likelymechanisms forexchange also silver,lead,andcopper(andthusGöllüDag(cid:2)obsidian),thereisan includereligiousfestivalsandgatheringssuchastheannualtrading apparentparadox.TherewasnotonlycopperoreintheTurAbdin bazaarsdescribedbyCrawford(1978). highlands near Diyarbakır but also silver (and lead). In fact, one Hurrianmythrelatesthetaleofayounggod,Silver,wholivesinthe 7.5. States,nomads,andterritoriality mountainous hinterland and travels to Urkesh in search of his father(Hoffner,1990:46e47).Copper,silver,andleadapparently Thesemechanismsofobsidiandistributioncouldhavechanged were readily available via the Mardin Pass; however, Göllü Dag(cid:2) if territorialityalso changed under Akkadian influence. To use an obsidianartefactsatUrkesh(andsomeleadisotopesignaturesin ethnohistorical analogy, consider the aforementioned Kurdish 1132 E.Frahm,J.M.Feinberg/JournalofArchaeologicalScience40(2013)1122e1135 ties,ratherthanstateadministrativecontrols,“wouldhavemadeit difficultforanoutsider,suchasNaram-Sin,toreplacewithhisown thecontroloftheUrkesh[kings],andthusanalliancewouldhave beenawiserpoliticalchoice”(BuccellatiandKelly-Buccellati,2000: 155). Perhaps, with the mere presence of Tar’am-Agade in the palace,Hurriansin thehighlandsindeedresistedAkkadianinflu- ence and endeavoured to limit access to resources within their purview, such as the Diyarbakır copper mines. In turn, Urkesh might have been forced to seek metal and other mountain resources elsewhere, increasing or initiating trade with Middle Euphratessettlements. 7.6. Exchangeandeliteidentity Anotherexplanationisthat,ashasbeenshownintheAegean (Dayetal.,1998;Whitelawetal.,1997),eliteidentitycan,atleastin part, be formed and sustained by involvement in exclusive economic and political networks, resulting in shared material cultureamongsttheparticipants.ItispossiblethattheAkkadians simply focused on trade routes across the Mesopotamian plains, towards the Eastern Mediterranean, resulting in greater trade on a west-east route along the Tur Abdin and a linkage to the ATN. Could an Akkadian administrator or Hurrian king have bolstered theirstatusatUrkeshbyengagingwithCentralAnatoliankingdoms viatradeanddisplayingprestigeobjectslikeobsidianvessels?Such vessels have been found at Bronze-Age Anatolian and Meso- potamiansites(e.g.,Tobler,1950;Childe,1954;Woolley,1955),and fragments were even discovered in the temple complex (B) at Fig.10. AnnualmigrationroutesofseveralnomadicgroupsmappedbyBes¸ikçi(1969), Urkesh. In a Bronze-Age palace complex at Acemhöyük, near the formingadefactoexchangenetwork.Recordsfromthe1930srevealthatpastoralists CentralAnatolianobsidiansources,excavatorsdiscoveredaseries summerednearLakeVaninTurkeyandwinterednearJebelSinjarontheIraq-Syria ofobsidianvessels,includingabroadbowl,aflaskrepairedusing border. Three decades later, their wintercamps lay just within the Turkey border duetostate-levelchangesinterritoriality.RedrawnandeditedfromBes¸ikçi(1969). gold, a tall decorated vessel, and a fluted flask with two animal- head handles (Özten,1988). At Tepe Gawra in Iraq, avessel from a fourth-millennium royal tombwas sourced toAcıgöl in Central pastoralistswhomigratedbetweenLakeVan(inTurkey)andJebel Anatolia (Renfrew et al., 1968; cf. Blackman, 1984 in Iran). As Sinjar (on the Syria-Iraq border) in the 1930s. Bes¸ikçi (1969) a “flashy” prestige object, obsidian becomes less about trade and documentedtheirmigrationsthreedecadeslater(Fig.10),andno moreaboutstatusandidentity. longerdidthetribeswinterinSyriaandIraq.Instead,thewinter Such a strategy could even have been used by the Urkesh campslayjustinsidetheTurkeyborderduetoachangeinhowthe governor, whether a Hurrian king or Akkadian administrator, to states delineated and defended their territories. Thus, the spatial forgeandsustainanidentityindependentoftheTurAbdinhigh- organisation of the groups changed, and their resources were no lands.Aspreviouslydiscussed,Urkeshinhabitantsare thoughtto longer traded with villagers across borders. Similarly, nomadic have maintained cultural ties to Hurrian groups in the north groups who decades ago traded Dead Sea salt with villagers in (Buccellati and Kelly-Buccellati, 2001). Various scholars have Turkey no longer do so (Ashkenazi, 1938; Maxwell-Hyslop in proposed that the mountains of southeastern Anatolia were Crawford,1978). aHurrian“homeland”(Speiser,1953;Wilhelm,1989;Stein,1997; Territorialityneednotinvolveaggressiontodefendgeograph- Steinkeller, 1998; Akkermans and Schwartz, 2003; cf. Benedict, icalperimeters.Itmayalsobeexpressedthroughsocialboundaries. 1960; Kurht,1995; von Dassow, 2008). Given the likelihood that When territories are difficult to defend physically, groups can theTurAbdinmountainswereinhabitedbyHurrians,anapproach maintain boundaries socially through signals (e.g., elaborate toeliteidentityconstructionbasedupontradeandcontactswith greetings, distinctive material culture), kinship, exchange and CentralAnatoliankingdomsseemsmorelikely fromanAkkadian reciprocity, language, and similar means (Dillian, 2002: 102). governor seeking to distance their administration from the high- Hence, while it is possible that Akkadian territoriality may have landsandapossibleHurrianhomeland.Thisisapotentialresolu- affected nomadic migrations, it is equally possible highland pop- tiontothe“paradox”ofengagingintheCentralAnatolianmetals ulations, perhaps Hurrians groups, changed their territoriality in trade while copper, silver, and lead metals from Diyarbakır were response to Akkadian encroachment. In such a case, territoriality availableviatheMardinPass. may be a form of resistance against Akkadian influence (i.e., the Akkadian empire installing a queen or an administrator in the 8. Conclusions palace of a Hurrian political and religious centre). As mentioned earlier, Hurrian control over northern mountain resources may As expected from established regional patterns, 94 sourced havebeenbasedonculturalties.Consideringreasonsforapossible obsidianartefactsatTellMozancamefromtheEasternAnatolian Akkadianallianceratherthanconquest,Buccellati(2008)proposes sources.Threeartefacts,however,unearthedinthepalaceservice that Urkesh controlled a northern hinterland through Hurrian courtyard,datingtotheheightofAkkadianinfluenceatUrkesh(i.e., cultural ties and “Naram-Sin realized that, while he could easily when Tar’am-Agade resided in the royal palace), are “exotic” haveconqueredthecity,hecouldnothaveovercomeamountain obsidianfromtheKömürcüsourceofGöllüDag(cid:2)inCentralAnatolia. insurgency”(3).Resourceandterritorialcontrolbasedoncultural Theprecisemechanismsandroutebywhichtheseartefactswere