Reply to J.D. Muhly, "Early and the Taurus"

K. ASLIHAN YENER AND PAMELA B. VANDIVER

Abstract ical data that we have published in the past and in This response to J.D. Muhly's essay (supra pp. 239-53) this issue,2 but he has also not had access to our new focuses on a series of key issues that have arisen concern- findings from the 1992 season at G61tepe. In this reply ing the chronology, technology, and archaeological con- we present some new data and discuss the larger text in which prehistoric metallurgy developed. Additional radiocarbon dates and information on EBA question of tin metallurgy and metal exchange in ceramics from soundings in the Kestel mine are presented, Anatolia. In an appendix, Lynn Willies focuses on which are relevant to the dating of the operations. The local geology and the Kestel mine. tin-bronze industry at Tarsus and the question of "inten- Muhly has often argued for a scheme whereby tionality" in the manufacture of bronze alloys are further metallurgy was a "unique" discovery in only one area discussed. No data exist to support Muhly's contention followed by radial diffusion from the center of origin. that gold and iron were produced at G61ltepe and Kestel. It is stressed that although particles of cassiterite and tin Although he applauds the passing of traditional ideas metal are small, they are dense and characteristically col- such as stimulus-diffusion and "ex orient lux," he ored, and hence easily identified. Replication experiments nevertheless insists that the inspiration for metallur- in 1992 have suggested a method of producing tin metal gical development came from a single restricted area, compatible with the analyses of the crucibles and coatings. In an appendix, Lynn Willies discusses the geological such as Troy or Mesopotamia, and then spread out- nature of tin deposits in general and at Kestel in particular, ward through "indirect relations" between metal-pro- and considers Muhly's interpretation of the deposits in ducing regions.3 We expect that as intensive research the Eastern Desert.* in metal-rich highland regions of the world increases and their sophisticated indigenous metal technologies Professor Muhly is to be thanked for devoting se- are documented, Muhly's unilinear, monodimen- rious attention to the question of Early Bronze Age sional reconstruction will be replaced by a more com- tin, a topic that has so many significant implications. plex model. We believe that his commentary and the widespread interest shown by scholars, and by the lay scientific "INTENTIONALITY" AND DEFINITION OF BRONZE press, all reflect the importance of our findings to the field.' We welcome the opportunity to respond to his One of the most contentious issues in archaeomet- points, each of which touches on issues that we have allurgy has been the definition of "intentionality" in considered in depth and have discussed in print. the manufacture of bronze. Muhly dismisses the pres- Muhly seems to have overlooked some of the empir- ence of a local bronze industry in south-central Ana-

* Publication of this response was also made Proceedings possible inof the 24th International Archaeometry Symposium part by the AJA Matson Fund. (Washington, D.C. 1986) 309-20; K.A. Yener, H. Ozbal, E. See J.D. Muhly, "Early Bronze Age Tin and Kaptan, the Tau- A.N. Pehlivan, and M. Goodway, "Kestel: An Early rus," in this issue, supra pp. 239-53. Bronze Age Source of Tin Ore in the , Publications both lay and scholarly have presented ," our Science 244 (1989) 200-203; P.B. Vandiver, K.A. findings: see esp. P. Craddock, "A Short History Yener, of Fireset- and L. May, "Third Millennium B.C. Tin Processing ting," Endeavor n.s. 16 (1992) 145-50; C. Michel, Debris from"Dur- G61ltepe (Anatolia)," in P.B. Vandiver, J. Druzik, humid, son commerce et ses marchands," in D. and Charpin G.S. Wheeler and eds., Materials Issues in Art and Archaeology F. Joannes eds., Marchands, diplomats et empereurs. 3 (Pittsburgh, Etudes in press) 545-69; and E.V. Sayre, K.A. Yener, sur la civilisation misopotamienne offertes & Paul and Garelli E.C. (ParisJoel, reply to "Evaluating Lead Isotope Data: Fur- 1991) 253-73; T. Bass, "Land of Bronze," Discover ther Observations," 12:10 Archaeometry (in press). (1991) 62-66; L.L. Jobe, "Finding the Evidence," 3 J.D. Aramco Muhly, and Tin: The Distribution of Mineral World 43:1 (1992) 18-19; C. Holden, "Clues to Resources a Tin Age," and the Nature of the Metals Trade in the Bronze Science 256 (1992) 1136; and E. Pennisi, "Turkish Age Tin(Hamden, Mine Conn. 1973) 322-23; Muhly, "The Bronze Revises Bronze Age History," Science News 141 Age (1992) Setting," 309. in T.A. Wertime and J.D. Muhly eds., The 2 See bibliography supra p. 208 n. 6, p. 209 Coming n.12, ofand the Age of Iron (New Haven 1980) 28; and R. p. 215 n. 38. See also K.A. Yener and H. Ozbal, Maddin "The ed., The Beginning of the Use of Metals and Alloys Bolkardag District Survey of Silver and (Cambridge, Lead Metals Mass. 1988) 2. in Ancient Anatolia," in J.S. Olin and M.J. Blackman eds.,

American Journal of Archaeology 97 (1993) 255

This content downloaded from 160.111.254.17 on Mon, 24 Sep 2018 14:59:17 UTC All use subject to https://about.jstor.org/terms 256 K. ASLIHAN YENER AND PAMELA B. VANDIVER [AJA 97

tolia by questioning the tin-copper ratios lead isotopeanalyzed ratios. inCharacterization of ceramics and the Early Bronze assemblage from Tarsus. analysis Contraryof their clay sources, as well as obsidian sourc- to our view and those of his collaborators4 ing, have that illuminated 1% or complex networks and ex- more tin content in a bronze is significant, change patterns he has operating concurrently with metal concluded that there was no bronze industry trade. The exact at Tar-nature of the interactions, both of sus. With such a view of the as yet little-known local and interregional Cilicia, exploitation, should become he also discounts the contemporary and more nearby evident inAmuq the near future. bronzes. Even if one were conservatively to consider SOURCES OF TIN 2% tin content in a bronze as the "intentionality" limit, as Muhly suggests, then 12% of the objectsOver the fromyears Muhly has suggested a number of Tarsus should be considered intentional areas bronzes, as the "primaland source" of tin. In earlier papers one would have to explain the origin he of regarded the added the source of Near Eastern tin to be in tin. Furthermore, the remaining 12% central(of our and stated southeast Asia and Cornwall.5 Later he 24%) of objects that "unintentionally" favoredcontain Afghanistan 1-2% over the more remote southeast tin clearly indicate that the ores used Asia.6in Tarsus Also, areand despite evidence to the contrary, he rich in tin and, when smelted, yielded has discounted a so-called the Eastern Desert of Egypt, Erzge- "natural" tin-bronze. birge, Yugoslavia, as well as the other small pockets This evidence of a bronze industry at Tarsus is in the Black Sea sands, Cyprus, and the Troad, which corroborated by the presence of local tin-rich ores, have all yielded tin.7 which we have documented just to the north of Tarsus There have also been inconsistencies in his assess- in the Taurus Mountains. The 24% figure for tin- ment of the Taurus sources. At one time he regarded bronzes at Tarsus is not insignificant either for the them as important candidates for a tin source,8 reluc- amount of tin utilized as an alloying additive or as tantly accepting Taurus tin;9 later he totally rejected evidence for the use of local tin-rich ores. Notwith- their existence,10 then moved back to acceptance," standing Muhly's arguments in the preceding article, and finally, in this issue of AJA, he turns negative once tin-bronzes do occur in a variety of locations in Ana- again. Muhly has also interpreted the metallic nature tolia from the late Chalcolithic through the third mil- of the Taurus Kestel mine in several ways. Initially he lennium B.C. Their sources as well as the nature of congratulated us for having discovered the legendary their technology are being investigated. Our analytical "Silver Mountains" of the Akkadians;'2 later he ap- program in collaboration with the Turkish Geological plauded us for having discovered the oldest gold mine Survey carried out over a span of 10 years clearly of antiquity.'3 Now he suggests that we may have demonstrates that a number of tin sources in Turkey discovered an ancient iron mine.14 existed during the formative periods of bronze met- These are not matters for conjecture, but empirical allurgy. We are now researching the mosaic pattern data. The ore mined at Kestel was tin, as confirmed of interactions from the highlands to the urban cen- and published by a number of local and foreign spe- ters utilizing analytical methods such as elemental cialists.'5 Muhly's former coauthor and local expert, characterization of metals and the sourcing of ores by O. Oztunah, has now abandoned his former reserva-

4 E. Pernicka, F. Begemann, S. Schmitt-Strecker, 12 and Muhly A.P. et al. (supra n. 10) 212. Grimanis, "On the Composition and Provenance 13of J.D. Metal Muhly and E. Pernicka, "Early Trojan Metallurgy Artefacts from Poliochni on Lemnos," OJA 9 (1990) and Metals 272. Trade," in J. Herrmann ed., Heinrich Schlie- 5 Muhly 1973 (supra n. 3) 262-88; Muhly, "Tinmann. GrundlagenTrade und Ergebnisse moderner Archdologie 100 Routes of the Bronze Age," American Scientist Jahre 61 nach (1973) Schliemann's Tod (Berlin 1992) 315. 409-12. 14 Muhly (supra n. 1) 252. 6 J.D. Muhly, "Sources of Tin and the Beginnings of 15 L. Willies, "An Early Bronze Age Tin Mine in Anatolia, Bronze Metallurgy," AJA 89 (1985) 281; Muhly, review of Turkey," Bulletin of the Peak District Mines Historical Society R.D. Penhallurick, Tin in Antiquity, in Archeomaterials 2 11:2 (1990) 91-96; Willies, "Report on the 1991 Archaeo- (1987) 102. logical Survey of Kestel Tin Mine, Turkey," Bulletin of the 7 Muhly 1973 (supra n. 3) 102-103, 277-88, esp. 283. Peak District Mines Historical Society 11:5 (1992) 241-48; 8 Muhly 1973 (supra n. 3) 290-91. Willies, "Reply to Pernicka et al.: Comment on the Discussion 9 Muhly 1973 (supra n. 3) 102. of Ancient Tin Sources in Anatolia,"JMA 5 (1992) 99-103; 10 J.D. Muhly et al., "The Bronze Age Metallurgy of An- Willies, "Report to Sponsors: Early Bronze Age Kestel Tin atolia and the Question of Local Tin Sources," in E. Pernicka Mine in Turkey: 1992 Expedition," Bulletin of the Peak and G. Wagner eds., Archaeometry '90 (Basel 1991) 209-20. District Mines Historical Society (in press); A. Qevikbag and 11 J.D. Muhly, "Copper in Cyprus: The Earliest Phase," 0. Oztunali, "Uluklala-Camardi (Nigde) Havzasinin Maden in J.P. Mohen and C. Eluere eds., Dicouverte du mital (Paris Yataklari [Ore Deposits in the Uluklyla-Camardi (Nigde) 1991) 361. Basin]," Jeoloji Miihendisligi Dergisi 39 (1991) 22-40; A.

This content downloaded from 160.111.254.17 on Mon, 24 Sep 2018 14:59:17 UTC All use subject to https://about.jstor.org/terms 1993] REPLY TO J.D. MUHLY 257 tions and acknowledges 10% of the 250 excavatedthe in theimportant 1990 field season. tin min- eralization at Kestel.'6 Twenty-three Necip of these cruciblePehlivan fragments are typical of the Turkish Mining and Research of theInstitute larger sample and one was(MTA), selected because itthe geologist who actually discovered was abnormal. the There is noKestel gold or silver inmine, the slags panned out alluvial cassiterite from or deposited the on the crucibles' KuruCay surfaces, as analyzed stream by running immediately below the x-ray fluorescence, mine which hasentrance." a detection limit of hun- Contrary to Muhly's insistence that dredths of alluviala percent over an area cassiterite 1 cm2 by 200-300 does not exist at Kestel, hundreds [t in depth. A traceof of coppersand-sized was found only on the brilliant red cassiterite grains have one atypical been crucible, buteasily only present in separateda concentra- by man- ual means such as a pan tion of hundredths or a of vanninga percent. Oxidized tin prillsshovel, and very little water. Due to its are present high in the slags. specific These prills have the gravity,needle- particles of tin will sediment out like, or acicular,from morphology the of a moltenless metal, dense al- magnetite, hematite, and quartz. though This they have corrodedsimple or oxidized processto the oxide, was repli- cated a number of times at Celaller in the summer of cassiterite. No naturally occurring cassiterite has been 1992 by Bryan Earl of Cornwall (see below). As of this found in a glass with this acicular morphology any- writing, we have not been able to penetrate further where else. In addition, cassiterite is present as a fine into the strata in the stream to arrive at the original powder with a brownish to purplish and reddish tint concentrations of cassiterite available to the miners of on the inner surfaces of many of the crucible frag- antiquity, if those concentrations even still exist, be- ments. This cassiterite deposit is further evidence of cause they are buried in this mountainous area, ac- tin processing at an elevated temperature, because of cording to the Turkish Geological Survey, by perhaps the presence of tin and also for its morphology: the as much as 30-40 m of accumulated sediment. De- fine, nearly spherical particles in this powder are char- forestation in the region through the millennia has acteristic of particles produced by condensation from contributed to the erosion of the slopes, which has a vapor. We have identified some peaks of tin metal produced this overlying sediment. Coring to these by x-ray diffraction, but there are many other phases levels is thus an important, but costly and labor-inten- present, some of which overlap and so obscure the sive, future aim. We would not expect to see the peaks that identify tin metal. In addition, these other concentrations found in the third millennium B.C. on phases are better crystallized and present in far the surface anywhere today. This is also the case in greater amount, and thus detection is extremely dif- Cornwall, England. Earl, with Yener present, vanned ficult. The theoretical argument for the presence of the formerly tin-rich alluvial deposits in a Cornwall tin metal in the slag is that it is surrounded in glass stream and arrived at precisely the same results as at through which oxygen diffuses very slowly, at about KuruCay-up to 0.15% tin concentration in the 10-23 cm/cm/oC, and thus some of the original metal vanned sample. may remain. The amount of tin oxide that has been consistently THE EVIDENCE OF THE GOLTEPE CRUCIBLES analyzed in the slags by wavelength dispersive micro- At one time Muhly was an advocate of archaeomet- probe analysis is about 30 wt% average, although allurgy based on fieldwork and technical some analysis. areas contain over 90 wt%. The controversy Unfortunately, he has recently relied heavily here on is notthe- about tenths of a percent of tin, or about oretical approaches that do not spring from "infinitesimal" an em- amounts, but instead data are pre- pirical data base. Our data base consists of oversented 500in which the presence of tin as an element is excavated tin-containing crucible fragments foundfrom by the three different and well-established tech- processing and habitation site of Early Bronze niques Age in amounts far beyond the detection limits of Goltepe; we have intensively analyzed 24 of these,the techniques i.e., and in 24 individual samples. The

Qagatay and A.N. Pehlivan, "Celaller (Nigde-Camardi) kalay Kaptan, paper delivered at the 14th International Sympos- cevherleqmesinin mineralojisi [Mineralogy of the Celaller ium of Excavations, Surveys and Archaeometry, 25-29 May (Nigde-Camardi) Tin Mineralization]," Jeoloji Miihendisligi 1992, Ankara. Dergisi 32-33 (1988) 27-31; Craddock (supra n. 1) 145-50; 16 Qevikba? and Oztunall (supra n. 15) 22-40. E. Kaptan, "Tuirkiye Madencilik Tarihine ait Camardi-Ce- 17 Yener et al., "Kestel" (supra n. 2); Cagatay and Pehlivan laller K6yui Y6resindeki Buluntular," IV. Arkeometri Sonu- (supra n. 15) 27-31; and A.N. Pehlivan and T. Alpan, "Nigde (larz Toplantzsz (Ankara 1989) 1-16; Kaptan, "Tiirkiye Mad- Masifi Altin-kalay Cevherleqmesi ve Agir Mineral QaliSmalari encilik Tarihine ait Celaller (Nigde) Y6resindeki Sarituzla- On Raporu [Preliminary Report of the Gold-Tin Heavy Goltepe Buluntulari," V. Arkeometri Sonuclarz Toplantzsz (An- Mineral Mineralization of the Nigde Massif]," MTA Report kara 1990) 13-32; Kaptan, "Finds Relating to the History of (Ankara 1986). Metallurgy in Turkey," MTA Bulletin 111 (1990) 75-84; and

This content downloaded from 160.111.254.17 on Mon, 24 Sep 2018 14:59:17 UTC All use subject to https://about.jstor.org/terms 258 K. ASLIHAN YENER AND PAMELA B. VANDIVER [AJA 97 other elements that are present as major particles constituents are grouped together. In fact, we can identify are iron, calcium, silicon, and aluminum. it as Although a strong red color when only 10% of the submi- their absolute amounts vary, their proportions cron hematite are particles are present and mixed with relatively fixed. This indicates either a 90%consistent, quartz rawand calcite, and the density difference of material composition involving careful hematiteselection is much of lessore than that of cassiterite. The from the mine or, alternatively, control evidence in issample overwhelmingly positive in its confirma- preparation and processing for the smelting tion that tin opera- processing occurred at G61ltepe. tion. Such consistent ratios argue for a consistent and REPLICATION EXPERIMENTS intentional workshop practice. The variation in the amount of iron oxide may be related to Excavations its function at G61tepe have so far provided an as a flux, an agent that aids melting orextensive lowers corpusthe of artifacts related to tin production. temperature at which melting occurs. Artifacts Iron oxide from is the Kestel slope and the mine galleries found in lesser quantities in examples in corroborate which greater the presence of a tin production industry segregation has occurred, suggesting aat more G61tepe, success- with the possibility of functional and ful smelt, and in other instances, too much diachronic iron oxidevariation. The evidence for tin smelting is present and segregation is less pronounced, relies on 1)indi- the enormous quantities of crucible frag- cating a smelt that has not been as efficient. ments and the analyses of their tin-rich slag, 2) the The presence and size of the particles composition of purple toof the local ore body, 3) the presence of reddish-brown cassiterite in the sediments workshop in the mine floors with activity areas related to tin pro- and the size of the tin prills in the slags cessing,are not relevantand 4) the presence of bar-ingot molds. We issues. One of the purple particles of pure have cassiterite constructed a model for tin production at G61- is pictured in figure 20 of our article tepe(supra based p. 234). on data generated from the 1990-1992 A spoonful of sediment from the mine field brought seasons, back the geological survey, replication exper- from the first field season contained several iments, different- and laboratory analyses of production debris. colored particles that were identified Ourvisually. research The goal is to narrow the gap between theory reddish-purple ones can be easily separated and data as ausing con- fundamental premises of physical and centrate from the black (iron oxides) chemicaland white science, ones which influence the structure of our (quartz and calcite) by merely swishing investigation. them around Each phase of the production process in a glass with water in it. They can also from be separatedmining to finished product is in the process of by winnowing because their densities are being so different.identified, its elements defined, and archaeo- Once one learns that purple is the magic logical, color metallurgical, asso- and mineralogical implications ciated with cassiterite, then it is relatively investigated easy to iden- to characterize the nature of the tech- tify where it is concentrated during nology. excavation. Since very little information for crucible Likewise, the crucible fragments look likesmelting vegetal of fi-tin existed prior to our investigation,'8 ber-tempered, coarse-ware cookpots, butour withreplication a cer- experiments in the summer of 1992 tain brownish residue and black globules helped or dripsdefine on hypothetical manufacturing stages us- the surfaces. Recognition of their special ing contents local raw also materials and identified the expected requires careful observation. Furthermore, archaeological in grind- data associated with each stage. ing the slags that can be detached from The the experiments surfaces were conducted in collaboration of the crucibles, careful observation and with separation Bryan Earl, an eminent mining engineer from according to color and density is a key Cornwall to recognizing who specializes in ancient tin technologies. the tin. Concentrating fine particles ofThese an especiallyexperiments have demonstrated that smelting dense material present as about 30 wt%required of a multistepglass process to produce tin metal with requires no great skill or insight. For instance, refining aaccomplished bottle by washing, density separation, of red iron oxide, or hematite, more and commonlygrinding with lithic tools and subsequent remelt- known as rust, consists of particles that ing. average Corroborating 0.1 evidence was provided by tin ox- p, or 0.0001 mm, yet we have no trouble ide identifiedseeing the by scanning electron microscope on the red color and identifying it as red ocher surfaces when of the several grinding tools from Early Bronze

18 See attempts by B. Earl, "Tin Preparation the and West Smelt- of England: A Study of an Old Art," Journal of the ing," in J. Day and R.F. Tylecote eds., The Industrial Historical Revo- Metallurgy Society 19 (1985) 153-61; and R.D. lution in Metals (London 1991) 47-83; Earl, "Melting Penhallurick, Tin in Tin in Antiquity (London 1986).

This content downloaded from 160.111.254.17 on Mon, 24 Sep 2018 14:59:17 UTC All use subject to https://about.jstor.org/terms 1993] REPLY TO J.D. MUHLY 259

Fig. 1. Bryan Earl smelting tin in a crucible, July 1992, Celaller, Turkey. (Photo K.A. Yener)

Age contexts.'9 The smelting process was simple and ticism about the "useful" concentrations of tin in the straightforward and did not require technical sophis- Taurus and the technological skill required of ancient tication. No fixed installations were needed. A tem- workers. perature range of ca. 850-1000' C is sufficient and The second part of our hypothesis, that smelted tin was easily attained with a blowpipe. was subsequently refined in "melting" crucibles and The first step in the experiment entailed the iden- cast into standardized ingots of tin metal, is still being tification and selection of the ore charge to be used investigated. Production at G61tepe apparently was in the smelt. In the 1990 season at Gl1tepe, several aimed at the production of ingots: multifaceted caches of pulverized ore were discovered that had a molds with impressions for bar-shaped ingots were fine powdery consistency, and contained 0.3-1.8 wt% excavated in pit-house workshops. Although tin- tin with a distinct reddish-purplish color. Vessels con- bronze pins, awls, rings, and other fragments contain- taining this ore had been excavated in sealed deposits ing 4.75-12.3% tin were found in several of the pit- on the floors of Early Bronze II/III pit-houses. This houses, it is more likely that the alloying was carried "low grade" 1% cassiterite ore mixture was then en- out elsewhere. The presence of non-local goods such riched to approximately 20% by vanning (panning as imported pottery and the lead isotope correlations with a shovel) with a cup of water. The enriched ore of Taurus ores with artifacts made of copper-based was next placed in a "homemade" crucible, using local or silver metal from distant sites confirm exchange clay and chaff temper (as described above), and cov- with neighboring regions. Given the numbers of ered by successive layers of charcoal. Smelting of this grinding stones and crucibles and the amount of other ore produced slag and tin prills that emerged inside industrial debris, however, in addition to the context the crucible after 20 minutes of blowing through a of production in pit-house workshops, the quantity of blowpipe (fig. 1). This experiment will be published tin being produced at G61tepe and Kestel was signif- elsewhere in more detail and should ease initial skep- icant. Such production reflects a strong demand for

19 R. Hard, personal communication, August 1992.

This content downloaded from 160.111.254.17 on Mon, 24 Sep 2018 14:59:17 UTC All use subject to https://about.jstor.org/terms I I I I I I I I I workings unor 20 40 50 70 80 90 100 110 , 2 16 - ' shaft 50o .17 .inclined

14 C13 >

-40 40- ~I1 entry rsmall window Surface - mined.:: Area rlp /

00 l \2 3aI V

t"j -20 VI'II . , - Line survey only > - I 92M- Mortuary C tI1 -po First C~a haber Va \0- ;> > Entrance I >I

-10 %1 a

<

tI1;>

Z

S srveystation Large Chamber

T5 archeological trench6 V slope down all614

threork g, 0 10 2O m wat

0o-30 10 20 30 40 so eo 70 o gow

Fig. 2. Map of the Kestel mine, 1991. (Lynn Willies)

This content downloaded from 160.111.254.17 on Mon, 24 Sep 2018 14:59:17 UTC All use subject to https://about.jstor.org/terms 1993] REPLY TO J.D. MUHLY 261 tin metal. The extent, The discoverydirection, of a large diabase and mortar orregularity anvil of this interaction are of particular with two circular hollows interest, on one surface provided especially with respect to the role informationtin played about the specific toolsin of it,extraction and and we hope to trace such interactions beneficiation using in this chamber.20 trace Other surface elementfinds in and Moss- bauer studies of tin. chamber VI included a lamp, a diabase pestle, and third-millennium B.C. ceramics. The upper 40 cm of DATING OF MINING ACTIVITY AND PRESENT the sounding yielded a mixed deposit of medieval and RESEARCH STRATEGY Early Bronze Age sherds and diabase tools. The EBA The initial dating of the operations at sherds the Kestelare predominantly dark-burnished and un- mine relied heavily on radiocarbon dates burnished and stylistic varieties, red-burnished, and micaceous- studies of ceramics found in five soundings finished from wares. 1987 Interestingly, a pink ware similar to to 1988. New soundings in 1990-1992 by one the dated collab- to Karum IV levels21 at Kfiltepe suggests orating mining team led by Lynn Willies tantalizing expanded connections to very late third-millennium our knowledge of the extent of the mine central (fig. Anatolian 2). This sites. Below 40 cm, the pottery be- material is now being prepared for publication, came more buthomogeneous a with a dark, highly polished short summary is presented here. The discoveryEBA ware predominating. of Some cruder examples wares from different periods on the surface such as a ofhole-mouth the jar and several straw-tempered galleries and divergent extraction techniques types also suggest emerged in the lowest strata, suggesting the that ore may have been removed over several existence periods of a Late Chalcolithic phase in this mine as and that a chronological sequence of mining well. It ismight also possible that the Chalcolithic pottery be reconstructible. slid into this gallery from open-pit mining operations In order to test the assumption that cassiterite was situated 50 m upslope. Open-cast mining may have the targeted mineral in the Bronze Age, four small preceded the shaft and gallery systems and the Chal- (ca. 1 x 2 m) soundings (S1-S4) were initiated inside colithic pottery may have slipped in through the ver- galleries II, III, VI, and VII. These soundings were tical shaft that was dug after the extraction pit not dug stratigraphically, but soil and charcoal sam- operations. A more precise Late Chalcolithic date ples were collected at arbitrary 10 cm levels for geo- for this pottery must await study of comparable se- chemical, mineralogical, and radiocarbon analyses. quences22 elsewhere in the Qamardi and Nigde area The assumption was that the detritus of mining activ- when they are excavated. A new radiocarbon series ity would yield important information about dating from soundings in 1991 and 1992 should expand our the mine and about the original ore body composition. knowledge of the range of dates for the various min- Sounding S2, measuring 1.0 x 1.5 m, was placed in ing operations. chamber VI at the confluence of five upsloping gal- A layer of rubble intermixed with EBA sherds was leries some of which measured a scant 60 cm in di- reached at -60 cm. This basal unit layer of collapse ameter. These galleries had circular cross-sections and from -60 to -93 cm had a massive character and differed morphologically from the larger entrance suggested spoil from mining activity. Only samples chambers I and II. Careful study of the stratified below the -40 cm level were utilized for radiocarbon assemblage indicated that the debris of mining dates dating and mineral identification. The radiocarbon to the Bronze Age. Also supporting this view were results on samples of charcoal yielded dates of 2070 - the massive layers of rubble with third-millennium 80 B.C. (at -68 cm, calibrated 2 sigmas, Struiver and ceramics, which were overlaid by thin horizons dating Pearson curves, 2874-2350 B.C.), 2030 - 100 B.C. to the Byzantine period but without any associated (-68 cm, calibrated 2 sigmas = 2870-2200 B.C.), and mining debris. Thus, that there are Byzantine strata 1880 - 65 B.C. (-93 cm, calibrated 2469-2133 B.C.), in the mine dated by ceramics and radiocarbon (char- and suggested that strata between -68 and -93 cm coal samples from sounding S2 at -30 cm = A.D. should be dated to the third millennium B.C. Another 380 ? 60, calibrated A.D. 347-609, 2 sigmas) does radiocarbon date comparable to these dates comes not necessarily indicate that mining took place during from sounding S1 at -60 cm, 1945 ? 70 B.C. (cali- these later periods. brated 2576-2147 B.C.).

20 Kaptan 1989 (supra n. 15) fig. 3. gonic texts that specify a tin-mountain called Zar-sur, and a 21 A. Ozten, private communication: Karum IV, III = ca. mention of a "mine" that may be tin. 2100-2000 B.C., Middle Chronology. It is interesting to note 22 G.D. Summers, "Chalcolithic Pottery from Kabakulak Muhly's discussion (1973 [supra n. 3] 288-89) of pre-Sar- (Nigde) Collected by Ian Todd," AnatSt 41 (1991) 125-32.

This content downloaded from 160.111.254.17 on Mon, 24 Sep 2018 14:59:17 UTC All use subject to https://about.jstor.org/terms 262 K. ASLIHAN YENER AND PAMELA B. VANDIVER [AJA 97

Sounding S2 has identified several features The soundings of this in the mine have provided impor- chamber. First, cassiterite was being mined tant ininformation the third about the tools employed in the millennium B.C. The technique entailed technology firesetting, of Bronze Age mining and ore pulveri- then battering the ore with heavy hammerstones.zation as well as an indication that some amount of Mortars, pestles, and bucking stones (stones habitation with or shelter one probably was provided by the flat surface and a hollow in the middle) mine. indicate The Kestel that excavations have helped define an some ore beneficiation was also taking interactiveplace inside system the between the mine and Giltepe, with mine. It is apparent from the stone tool its specializedtypes found sectors devoted to smelting tin. Tin inside that hammerstones for battering smelting and entailed pulver- an initial ore enrichment at Kestel, izing the ore were being utilized. But, followed surprisingly, by grinding and washing phases at G61ltepe. bucking stones indicate that grinding Firedalso intook crucibles place. of various sizes, the resulting tin The debris may have been used as backfill metal in was the fabricated mine. into ingots by pouring the molten The presence of beneficiation in the minemetal into suggests molds. The data from this project have led that some fine-scale processing was essential to the irrefutable for conclusionex- that tin was mined at traction, and may indicate that veins Kestelwere and small processed or at G61tepe in the third millen- that testing of the particular area of niumthe B.C.deposit We can bynow question where the products empirical means was necessary and waswent andmore what effi-the impact of this strategic industry ciently carried out within the mine. It maywas on also the bronzebe that producers. In the developing field some tools were stored or discarded within the mine. of archaeometallurgy, Muhly's careful examination of Therefore, the ore was battered, pecked, and en- published material in the past has been an inspiration riched inside the mine. Secondly, the presence of for researchers, including us. But such evaluations pottery with open forms, the domestic faunal remains, require continual testing and support from empirical and a hearth suggest a certain minimal amount of data. Old ideas die hard, especially when not sup- domestic activity also took place inside the mine. ported by empirical data. Larger ceramic forms, presumably to contain water CONSERVATION ANALYTICAL LABORATORY for beneficiation and drinking or foodstuffs, were SMITHSONIAN INSTITUTION perhaps used for short-term storage. WASHINGTON, D.C. 20560

Appendix: Early Bronze Age Tin Working at Kestel LYNN WILLIES

In the controversy over Kestel, my role With has beena small team of experienced workers, I have that of a bit-part player, contributing from now thebeen side- to Kestel for three successive summers, lines in my own particular specialization surveying of ancient and sampling the mine, with minor archae- mining.23 I confine myself here to discussion ological of excavation. the Interim reports have been pub- mining aspects of Muhly's intriguing paper lished in for this the first two visits,25 the second too late for issue of AJA.24 The controversy over EBA Muhly tin working to have seen prior to his latest commentary. I at Kestel over the past few years has had have the tobenefit rely on coworkers with specialist skills for of leading to a very close examination of potteryalmost every examination (Sylvestre Dupre and Behin Ak- statement written on it, but has also led soy), to a varietylocal geology and geochemistry (Necip Pehlivan of claims, some rather frivolous, by several and Ergun writers. Kaptan of the MTA), and chemical analysis Since Muhly's case fundamentally rests (Hadion whether Ozbal of Bogazigi University). Significantly, tin was indeed mined at Kestel in the Early none Bronze of these specialists appear to share Muhly's Age, and since he has, by selective quotation, doubts inferred about Kestel. As yet, only limited accounts of that I am doubtful this was the case, theI hope finds this from the mine are published, while the contribution will clarify matters. results of radiocarbon dating are also awaited. Never-

23 Acknowledgments are due to the Historical at Kestel. Metallurgy Society and the Peak District Mines Historical 24Society, See also Williesand 1992, "Reply" (supra n. 15) 99-103. above all the British Academy, which have provided 25 Willies 1990much (supra n. 15) and Willies 1992, "Report" of the still very limited funding for the field (supra investigations n. 15).

This content downloaded from 160.111.254.17 on Mon, 24 Sep 2018 14:59:17 UTC All use subject to https://about.jstor.org/terms 1993] REPLY TO J.D. MUHLY 263 theless, it is possible practicability to give of processing here a low-grade an deposit, outline and of mining at Kestel with reasonable confidence. subsequently smelting a concentrate of it. The 1992 season at Kestel established the earliest It would not have been fundamentally difficult to mining so far found. An inclined shaft was reopened locate the primary tin deposit, once its importance in a mined joint to give access to an otherwise virtually had been realized (very much easier, for instance, than inaccessible area of workings. This fortuitously went locating silver in low-grade argento-jarosites, as was through some 5 m of material tipped into the joint done in the Late Bronze Age at Rio Tinto in Spain). from above, i.e., after the original working, containing The exploration method would involve the same sim- a great deal of pottery tentatively dated to ca. 3000 ple technique used for gold in streams, and the sub- B.C. The shaft gave us access to a long mined chamber sequent search for "shode ore" up hillslopes. Al- reused as a mortuary chamber, which seems from the though laborious, the methods used to mine the ore pottery scatter to have been its purpose throughout were also simple, using firesetting technology well the EBA. It was originally sealed off from other mine developed by EBA times, though Kestel is an early workings, but was subsequently broken into, possibly example. Processing of the ore was again laborious during the Byzantine period. The mortuary chamber but simple, very similar to the production of flour, by had been worked by firesetting, which, with stone first bruising the ore, which is very friable after fire- tools, was the mode of working for most of the mine. setting, followed by a fine-grinding process using a Remains of crucibles found in the shaft infilling may quern. Once ground down to a fine powder it could indicate that contemporaneous on-site smelting oc- be concentrated using water either by some refine- curred also at Kestel, but firm conclusions must await ment of panning, or by pouring thin slurries down further examination. sloping surfaces or tables. As has been shown by Bryan Although there are a few remains of later periods, Earl, who will issue his own report, it is possible to the rest of the accessible mine workings seem gener- van tin from the mine, and to concentrate it on the ally to have been used throughout the EBA, then vanning shovel. Crude but effective hand-processing largely abandoned. The age of the large area of open- of finely ground, low-grade tin ores containing much working at the surface has not yet been ascertained, hematite was a normal feature of Cornish hand meth- but it cuts through the earlier underground mine ods in past centuries. working. A radiocarbon date of a sample from a fire It may not always have been necessary to concen- used to heat the rock should provide more definite trate the ore beyond a hand-picking stage: samples of evidence, but much more work is required. ore found at G61ltepe with up to 1.5% tin may have The survey, since the first report seen by Muhly, been just as they were when derived from the mine. has shown the mine and the surface-working area to A small sample was smelted by Earl, who is an ac- be substantially larger than originally thought. There knowledged specialist on tin. There is no reason to is no reason to downgrade earlier estimates of a pos- doubt that some low-grade ore could have been sible output of a hundred or few hundred tons of tin smelted directly, if wastefully in terms of fuel, or it metal, at 0.1% to 1% tin grades within the whole rock could relatively easily have been concentrated, with removed. some loss of tin. Samples collected in the second season (1991), de- To conclude the case for tin processing at the Kestel signed to test various possibilities, gave very variable mine: even after intensive exploitation, the mine bears results, up to 2315 ppm Sn, which complement sam- considerable tin, which was largely worked in EBA ples earlier assayed by Professor Ozbal, and of course times, using appropriate methods. There is an enor- the much higher grade ore found in small pots at mous stone tool assemblage suitable for crushing and G1ltepe. Further sampling took place in 1992. There liberating the tin from its matrix. Simple methods is no doubt that the mine was capable of producing were available, used for gold and other minerals, tin, though it would also have been able to produce suitable for concentrating the ore. Tin is relatively vast quantities of iron oxides, and we may even find easy to smelt, and there is an abundance of evidence at some future time that some gold was produced. for the contemporaneous use of crucibles on a nearby That all this work was to produce tin is not, in my site (and on-site too), some of which have been shown words, cited by Muhly, self-evident (supra p. 251). It to have powdered tin oxide lining their inner surfaces, rarely is! This conclusion can only come from the and which have a slag containing prills of tin now related evidence, in this instance the metallurgical mostly oxidized to cassiterite, which could have re- evidence of tin smelting found at Galtepe by Yener sulted from the unwanted hematite being fluxed with and her coworkers, supported by evidence of the locally available quartz.

This content downloaded from 160.111.254.17 on Mon, 24 Sep 2018 14:59:17 UTC All use subject to https://about.jstor.org/terms 264 K.A. YENER AND P.B. VANDIVER, REPLY TO J.D. MUHLY

The case against is not without point, but in Do the we wait a millennium for the deposit to build up theoretical manner posed, it is as readily answered. again? It Perhaps the deposits or the cracks in the base seems to be appropriate to use an old miner's ofsaying the wadi are deep-and might repay open-working here about the location of ore, "where it is, there by pick it and shovel. Assume "Muhly and Partners" is"-in other words, it does not necessarily relate had topicked up all the cassiterite they could find-he human expectation or convenience. It does not says occur a few kilograms, so let us call it 10 kg for conve- at Kestel in the large pieces found elsewhere by nience, Muhly in the top 10 cm of the deposit. No plan is and others (not today anyway), but he fails to provided, show but let us assume an average width of 10 alternative sites with such large pieces that also m, have and a sampling length of 200 m-a pleasant walk evidence of ancient mining. Indeed the difficulty in the of hot conditions.Thus to extract his next 10 kg finding ancient tin ore sites is the underlying will thread entail the removal of some 500 tons of host ma- of this controversy, and obviously is a good reason terial, for and a grade of some 2% ore, or 1.6% metal in working low grades of ore in discovered deposits. the whole In rock. Still pretty good, but we can expect my experience of ancient mines, low-grade ore the was promoters of such a prospectus to have chosen frequently worked-compare for instance the the very best section they could, and this "proven" grade extensive workings at Wadi Amram26 near Timna- of ore might not continue far: all other grades would for low-grade copper ore, a metal that is infinitely at best be unproven. Very soon of course, even such more plentiful than tin, and found in much highereasily obtained ore as this would run out, and the concentrations elsewhere. Muhly raises the question company will need to consider mining the veins. Here of lack of evidence for the treatment of the "(hypo- the grade is likely to be lower, since it has not been thetical) rich ores that were all mined out during processed the by weathering, but the reserves are likely Early Bronze Age" (supra p. 247). This in the contextto be very much greater. Demand for the metal is of his pebble-sized samples from renewable deposits greater than ever. (?) is his rather than my hypothesis, but I would remark So far, of course, there has been no mention of that processing of high grades is not normally transportconsid- facilities, timber supply, food-growing po- ered a major problem if low ones can be tackled. tential, population, water supply-but in the middle Finally, I need to warn the readers of this journalof the Eastern Desert? Examined in these terms, per- not to part with good money to practical prospector haps Kestel appears a more realistic prospect. We Jim Muhly without a full professional examination should ofcertainly be suspicious of the viability of any what might be called the prospectus for Egyptian easily Tin visible, previously unworked, deposit in the Properties, Inc. His prospectus bears all the hallmarks ancient civilized world. of selling dubious mine-stock! Consider. Exotic, A far- great deal of Prof. Muhly's commentary is a away place (just dots on a small-scale map, nouseful site and perceptive summary of the situation, but plan); actual massive veins of cassiterite mineraliza- despite the impressive backing of the references he tion; area not previously exploited; and necessary cites, his whole argument concerning Kestel (and only to walk up the bed of a dry wadi to collect muchseveral else) relies on publications of interim results- kilograms of cassiterite. Nothing could be simpler: intended a to allow others to share and cooperate in self-renewing source of alluvial cassiterite, easily the workre- in progress. That there was EBA working coverable raw material that assays at 80% tin. of tin Plus at Kestel is now beyond reasonable doubt. In photographs of a few formless pebbles we are the asked remaining part of the "heroic age of archaeomet- to believe are pure cassiterite (I do really believe allurgy," they Timna is taking some three decades to bring are), and associated with a description of a well-known to published fruition. In more words from old miners, cassiterite deposit where the ore came up to the "When tree you find tail, hang on until you find t'cow." At roots (though not on the same lode-not even onKestel the and G61ltepe the tail has been identified; defin- same continent!). All published in a respected journaling the whole cow will take some time yet. whose readers are known to be interested in tin min- ing, but who do not have professional expertise PEAK in DISTRICT MINING MUSEUM mineral exploitation. A classic trap. MATLOCK BATH Let me add a touch of cynicism. What happens DERBYSHIRE DE4 3PS when the first few inches of wadi have been worked? ENGLAND

26 L. Willies, "Ancient Mining at Wadi Amram, (1991) Israel," 109-38. Bulletin of the Peak District Mines Historical Society 11:3

This content downloaded from 160.111.254.17 on Mon, 24 Sep 2018 14:59:17 UTC All use subject to https://about.jstor.org/terms