9
177 Scientific analysis of lead-silver smelting slag from two sites in China Pengfei Xie and Thilo Rehren ABSTRACT This paper presents data on two sets of lead-silver slag samples from northern and south-west China. At present, very few slag studies on ancient and historic slags from China are available in English, and these two sites in particular are completely unstudied. The sites and slags were selected on the assumption that they represent lead/silver production. This paper shows the technological range of slags related to the production of these two met- als, and what technological information can be gained from their study. It is hoped that this paper will contribute to knowledge about ancient and historic smelting practices, by offering new and original data, and by comparing the findings from these sites to other published sites. Introduction Lead and silver are typically seen as very different metals. Lead is considered cheap and of little functional use apart from in architecture and as a metallurgical additive in cer- tain bronzes. Silver in contrast is one of the precious met- als, desired by mankind for its beauty, value and related functions such as jewellery and coinage. Despite this seem- ingly disparate nature, their ores often coexist in nature, and the smelting procedures of the two metals are closely related. Due to these natural affinities, they usually also appeared together in ancient Chinese texts, and even today their production can only be fully understood when stud- ied together. Silver does occur in nature as a pure or ‘native’ metal, and several rich silver minerals have been exploited in the past, mostly sulphides of silver together with variable quan- tities of other metals, such as copper, antimony or bismuth. However, the bulk of prehistoric and historic silver produc- tion is based on the smelting of silver-containing galena, lead sulphide. The silver concentrations in galena rarely exceed 0.5% by weight, and even at concentrations as low as 0.1 wt%, the value of the silver content is often higher than that of the lead metal. Smelting such mixed ore will not separate the two metals; instead, a lead-silver alloy, or bullion, will be produced. A second step is necessary to separate the two metals. During this cupellation the less noble metal (lead) is oxidised during re-melting in an open hearth, while the noble metal (silver) remains in its metal- lic state. Thus, at the end of the cupellation process all lead is present as lead oxide, or litharge, which can be mechani- cally separated from the silver metal. The litharge can then easily be re-smelted to lead metal, forming soft lead with less than c.100 ppm silver. Large quantities of lead metal are therefore produced (often more than 100 times the weight of the silver). Archaeological evidence of such smelting sites could include the slag from the first smelting, possi- bly litharge remains, and furnace fragments. The scientific study of these remains can offer much information about the technological processes, while evidence of cupellation also reveals the economic relationship between the two metals. The slag samples studied were collected by Professor Ko and Professor Li Yanxiang of the University of Science and Technology Beijing from two Chinese sites: Tang county and Shizhu county. The Tang county of Hebei province is located in northern China, 190 km south-west of Beijing. The Shizhu county of Chongqing municipality is situated in south-west China, with the Yangtze River winding across its western border. Historically, the two sites had abundant deposits of mineral ores of lead, silver, iron and gold. They both boast a long history of mining and smelting. Until now, no formal archaeological excavation has been carried out at these sites, and no scientific analytical study has been made of the slag and ceramic specimens from the two areas. Our research endeavoured to undertake scientific analysis of an initial selection of specimens from the two sites using optical microscopy, scanning electron micros- copy with energy-dispersive spectroscopy (SEM–EDS) and X-ray fluorescence (XRF). This paper presents the results of the slag analyses. Methodology In the sampling process, the specimens were separated into two groups: group T1 for site no. 1 (Tang county) and group S2 for site no. 2 (Shizhu county), and then labelled Offprint from J. Mei and Th. Rehren (eds), Metallurgy and Civilisation: Eurasia and Beyond Archetype, London 2009. ISBN 1234 5678 9 1011

Xie Rehren 2009 Lead Silver Slag China BUMA VI

Embed Size (px)

DESCRIPTION

Study of lead and silver smelting slag from historic China

Citation preview

Page 1: Xie Rehren 2009 Lead Silver Slag China BUMA VI

177

Scientificanalysisoflead-silversmeltingslagfromtwositesinChinaPengfei Xie and Thilo Rehren

abstract Thispaperpresentsdataontwosetsoflead-silverslagsamplesfromnorthernandsouth-westChina.Atpresent,veryfewslagstudiesonancientandhistoricslagsfromChinaareavailableinEnglish,andthesetwositesinparticulararecompletelyunstudied.Thesitesandslagswereselectedontheassumptionthattheyrepresentlead/silverproduction.Thispapershowsthetechnologicalrangeofslagsrelatedtotheproductionofthesetwomet-als,andwhattechnologicalinformationcanbegainedfromtheirstudy.Itishopedthatthispaperwillcontributetoknowledgeaboutancientandhistoricsmeltingpractices,byofferingnewandoriginaldata,andbycomparingthefindingsfromthesesitestootherpublishedsites.

Introduction

Leadandsilveraretypicallyseenasverydifferentmetals.Leadisconsideredcheapandoflittlefunctionaluseapartfrominarchitectureandasametallurgicaladditiveincer-tainbronzes.Silverincontrastisoneofthepreciousmet-als,desiredbymankind for itsbeauty,valueand relatedfunctionssuchasjewelleryandcoinage.Despitethisseem-inglydisparatenature, their ores often coexist innature,andthesmeltingproceduresofthetwometalsarecloselyrelated.Due to these natural affinities, they usually alsoappearedtogetherinancientChinesetexts,andeventodaytheirproductioncanonlybefullyunderstoodwhenstud-iedtogether.

Silverdoesoccurinnatureasapureor‘native’metal,andseveralrichsilvermineralshavebeenexploitedinthepast,mostlysulphidesofsilvertogetherwithvariablequan-titiesofothermetals,suchascopper,antimonyorbismuth.However,thebulkofprehistoricandhistoricsilverproduc-tionisbasedonthesmeltingofsilver-containinggalena,lead sulphide.The silver concentrations in galena rarelyexceed0.5%byweight,andevenatconcentrationsaslowas0.1wt%,thevalueofthesilvercontentisoftenhigherthanthatoftheleadmetal.Smeltingsuchmixedorewillnotseparatethetwometals;instead,alead-silveralloy,orbullion,will be produced.A second step is necessary toseparate the twometals.During this cupellation the lessnoblemetal(lead)isoxidisedduringre-meltinginanopenhearth,whilethenoblemetal(silver)remainsinitsmetal-licstate.Thus,attheendofthecupellationprocessallleadispresentasleadoxide,orlitharge,whichcanbemechani-callyseparatedfromthesilvermetal.Thelithargecantheneasilybere-smeltedtoleadmetal,formingsoftleadwithlessthanc.100ppmsilver.Largequantitiesofleadmetalare

thereforeproduced(oftenmorethan100timestheweightof the silver).Archaeological evidence of such smeltingsitescouldincludetheslagfromthefirstsmelting,possi-blylithargeremains,andfurnacefragments.Thescientificstudyoftheseremainscanoffermuchinformationaboutthetechnologicalprocesses,whileevidenceofcupellationalsorevealstheeconomicrelationshipbetweenthetwometals.

TheslagsamplesstudiedwerecollectedbyProfessorKoandProfessorLiYanxiangoftheUniversityofScienceandTechnologyBeijingfromtwoChinesesites:TangcountyandShizhucounty.TheTangcountyofHebeiprovinceislocatedinnorthernChina,190kmsouth-westofBeijing.TheShizhucountyofChongqingmunicipalityissituatedinsouth-westChina,withtheYangtzeRiverwindingacrossitswesternborder.Historically,thetwositeshadabundantdepositsofmineraloresoflead,silver,ironandgold.Theybothboastalonghistoryofminingandsmelting.

Untilnow,noformalarchaeologicalexcavationhasbeencarriedoutatthesesites,andnoscientificanalyticalstudyhasbeenmadeoftheslagandceramicspecimensfromthetwoareas.Ourresearchendeavouredtoundertakescientificanalysisofaninitialselectionofspecimensfromthetwositesusingopticalmicroscopy,scanningelectronmicros-copywithenergy-dispersivespectroscopy(SEM–EDS)andX-rayfluorescence(XRF).Thispaperpresentstheresultsoftheslaganalyses.

Methodology

In the sampling process, the specimens were separatedintotwogroups:groupT1forsiteno.1(Tangcounty)andgroupS2forsiteno.2(Shizhucounty),andthenlabelled

Offprint from J. Mei and Th. Rehren (eds), Metallurgy and Civilisation: Eurasia and Beyond Archetype, London 2009. ISBN 1234 5678 9 1011

Page 2: Xie Rehren 2009 Lead Silver Slag China BUMA VI

PENGFEIXIEANDTHILOREHREN

178

T1A,T1B,T1C,T1D,T1E,T1FandS2A,S2B,S2C,S2D,S2Erespectively.Afterphotographicdocumentation,pol-ishedblockswereproduced for further analysis.SamplepreparationforopticalmicroscopeandSEM–EDSanaly-sesincludedcutting,mounting,grinding,polishingusingstandardmetallographicprocedures,andcarboncoatingtopreventachargebuildupfromtheelectronbeamduringSEM–EDS analysis. In contrast, sample preparation forXRFanalysis involvedcutting, crushing,milling topro-duceahomogenouspowder,whichwasthenmixedwithwaxandpressed intopellets.TheequipmentusedwasaLeica DMLM reflected light metallographic microscopeequippedwithadigitalcamera,aHitachiS-570SEMwithanOxfordINCAEDSsystemandaSpectroXLab2000X-rayfluorescencespectrometer.Thetechniqueswerecom-plementary, serving different purposes. SEM–EDSgivesfarbetterresultsformajorelementsinasinglephasethanXRFduetothegoodspatialresolution,whileXRFismoreusefulfordetectingminorandtraceelementsinthebulkmaterial(<0.1wt%).

Results and initial interpretation

The11analysedspecimensandtheresultswithsomeini-tialinterpretationsarepresentedbelow.Toavoidrepetitionin thepresentationofanalytical results, thedataarepre-sentedingroups.Onlyrepresentativesamplesareconsid-eredbelow.AnoverviewofhistoricalsourcesrelatingtosilversmeltinginChinaisgiveninXie(2005).FulldetailsoftheanalysesandresultsofouranalysesarepresentedinXie(2006).

Tang county

ThemacroscopicappearanceofsampleT1Aisdarkblackwith theflow textureon thesurface, suggesting itmightbe tapping slag (Fig. 1). This sample is representativeof themajorityoffindsstudiedhere, includingT1DandT1F,althoughwecannotjudgehowrepresentativeitisforthe entire assemblageon site.Themetallographic imageshows that sampleT1A isdominatedbyaglassymatrixwithnumeroustinybrightmetallicspots,andafewcrys-tals,bubblesandpores.Thelightergreycircularfeaturesintheslagglassarenoteworthy.SEM–EDSanalysisdidnotrevealanycompositionaldifferencebetweenthesurround-ingglassymatrixandthecircularfeatures;manyofthemcontainasmallinclusionorprillintheircentre.Weinter-pretthesecircularfeaturesasdevitrificationspheresaroundsome crystal nuclei, causedby a relatively slowcoolingof the slag (Fig. 2). Similar features havebeennoted inglassycrucibleslaginearlyIslamicUzbekistan(Rehrenet al.forthcoming).

ThechemicalcompositionoftheseslagsasdeterminedbySEM–EDSandXRFisgiveninTable1(mainoxides)andTable2(traceelements).Themaincomponentsoftheseslagsaresilica(55–65wt%),around15wt%ironoxide,5–10wt%lime,andloweramountsofalumina(6–8wt%)

andmagnesia(2–4wt%).SampleT1Dishigherinsilicathantheothersamples;thisisduetonumerousinclusionsofquartzparticlesintheslag.Theseinclusionsarepartlyabsorbedinthesurroundingglass,andhavenumeroussul-phideprillsintheirstructure,whichappeartoberesidualorematerial.SEM–EDSanalysesofthemeltphasebetweenthequartzinclusionsindicateasilicaconcentrationofc.58wt%, in linewith thebulkcompositionof theotherslagsamplesinthisgroup.Suchacompositionisinagreementwiththepredominantlyglassycharacteroftheslagasseenintheopticalmicroscope,andrelativelytypicalforahigh-temperature,stronglyreducingfurnaceregime.Duetothecomplexchemistry,itisdifficulttoestimatetheactualproc-esstemperature;testfiringswouldbenecessarytodeter-minetheliquidustemperature.

Onlya few inclusionswere found largeenough tobeanalysed.They included leadmetalprillswithup to2%silver,nearlypuresilverprills,andcomplexsulphidesrichinleadandiron,oftencontainingupto1%silver.

Figure 1SlagsampleT1A:typicalflowstructuresurface(centre).

Figure 2SlagsampleT1A:devitrificationspheres(lightgrey)inaglassyslagmatrix(widthofimagec.1mm).

Page 3: Xie Rehren 2009 Lead Silver Slag China BUMA VI

SCIENTIFICANALYSIS OFLEAD-SILVERSMELTINGSLAGFROMTWOSITES INCHINA

179

Forourinvestigationitisparticularlyimportanttonotethepresenceofbetween0.4and2wt%eachofzincoxideandleadoxide,indicatingthatthesearenon-ferrousslags.The very low level of copper (Table 2, 200–400 ppm)suggeststhatthisisnotcopperslag,butindeedleadslag.Furthermore, the presence of relatively high silver con-centrationsinbothmetalandsulphideprillssuggeststhattheproducedmetalwasasilver-richbullion,andthattheeconomicinterestinthissitewasprimarilysilverproduc-tion.This is furtherconfirmedby thepresenceofcupel-lation remains at this site (to be presented in a separatepublication).

The macroscopic image of sample T1C differs fromthe previous group in that it shows twodifferent layers:aceramiclayeratthetopandaslaglayeratthebottom.Theslaglayerisdarkerthantheceramiclayer,whiletheceramic layer ismoreporous than the slag (Fig.3).ThemicrographofsampleT1Cshowsanumberofsmallcracks

intheglassymatrixthatwerebreakingupduringpolishing.Weinterpretthistoshowthattheglasshasaconsiderableamountofresidualtensioninitfromthecoolingphase.Inaddition,thissampleshowsdevitrificationspheressimilartotheprevioussample.Thebrightinclusionsintheslagareleadsulphideandleadandsilvermetalprills.Theyareallverylowincopperoriron.

Theceramicattachedtothisslagsamplehasbeenana-lysedbySEM–EDS;theaverageoffivedifferentareameas-urementswas60.2wt%silica,28.9wt%alumina,3.7wt%ironoxide,andaround2.5wt%eachoflimeandpotash,plus1.6wt%titania,andlessthan1wt%magnesia.Thisisahighlyrefractoryceramicbasedonakaoliniticclayandrichinquartzinclusions.Theshapeoftheceramicistubu-lar, althoughnotenoughwaspreserved todetermine thediameter.Thisisinterpretedasafragmentofatuyèrefilledwithsomeslagwhichwaseitherdeliberatelytappedthroughit,orwhichoverflowedintothetuyèrewhenthefurnaceoverflowed.Similarslag-filledtuyèresarewellknownfromAfricanironsmeltingsites,andseemtobearegularpartoftheprocessratherthananindicationofamalfunction.

Samples T1B and T1E have a very different macro-scopic appearance,microstructure andcomposition fromtheremainingsamplesofthissite(seeTables1and2).Theyaremoreporous,almost frothy (Figs4and5),andhavemanyinclusionsofahardwhitealloy.AccordingtoSEM–EDSanalysis, these inclusionsare iron-basedwithmorethan95wt%iron(Figs6and7);thebalanceismostlikely

Sample Na2O MgO Al2O3 SiO2 P2O5 SO3 K2O CaO TiO2 MnO FeO ZnO PbOT1A – 4.3 6.4 55.2 0.4 0.8 1.5 11.1 – 1.5 16.9 1.9 0.4T1C – 4.5 6.3 55.5 0.3 0.7 1.3 10.9 0.3 1.6 16.7 1.9 0.4T1D – 2.0 6.8 64.7 0.2 0.1 1.6 4.7 0.3 0.7 16.7 0.4 2.2T1F – 3.8 7.7 58.7 0.3 0.7 2.7 9.1 0.2 2.9 12.5 1.0 0.4T1B 1.5 2.0 14.1 61.3 – – 2.7 12.8 0.7 0.5 4.4 – –T1E 0.3 22.7 7.5 49.7 – – 1.7 14.9 0.2 0.8 2.2 – –(P)ED-XRFdata,quantifiedbytheTurboquantmethod.Datanormalisedto100wt%.

Table 1

Sample Pb Cu Sr Zn Ag Ba SbT1A 4200 200 230 – 15 1400 –T1C 4000 190 220 – 15 1150 –T1D 22000 360 200 3800 220 260 80T1F 4000 310 385 – 10 1850 –T1B 10 15 760 25 – 750 –T1E 5 15 350 15 – 280 –(P)ED-XRFdata,quantifiedbytheTurboquantmethod.

Table 2

Figure 3SlagsampleT1C:ceramic layer(top)surroundingaslagcore(darkergrey).

Figure 4IronslagsampleT1B:aporousmacrostructure.

Page 4: Xie Rehren 2009 Lead Silver Slag China BUMA VI

PENGFEIXIEANDTHILOREHREN

180

carbon.Chemicalanalysisshowsthatthisslaghasverylowlevelsofbasemetalsandalsoverylowironoxideconcen-trations,buthighlime,magnesiaandsilica.Thisisclearlynotabasemetalslag;thechemicalandstructuralcomposi-tionandthenumerousiron-richprillsaremorecharacter-isticofblastfurnaceironslag.Intriguingly,thechemistrydiffersconsiderablybetweenthetwosamples:T1Eisveryrichinmagnesia,withmorethan20wt%,whileT1Bhasonlyone-tenthofthis,butamuchhigheraluminalevel.T1Bisalmostentirelyglassy,whileT1Eisrichincrystalswithabout45wt%MgO,42wt%SiO2and11wt%FeO,butverylittlelimeandotheroxides.Thisindicatesanapproxi-mate formulaofMg2SiO4withsomesubstitutionof ironoxideformagnesia,i.e.aforsterite-richolivine.

Thus, among the few samples analysed is evidenceforbothlead-silversmeltingandforironsmeltingatthissite.Bothprocessesseemtobedrivenbyablastfurnacetechnology, but are clearly based on different ores. It isimpossibleatpresenttosaywhethertheyaretechnicallyorchronologicallyrelated,orwhethertheyarecompletelyindependentfromeachotherandonlyappeartogetherherebycoincidence.

Shizhu county

Theslagsfromsiteno.2aremostlytapslagswithclearflowstructures(Fig.8).Therearesomegreenandbrowncor-rosionfeaturesvisible.Noceramicfragmentsareattachedtotheseslags.

ThemicrographofsampleS2Ashowsclearevidenceoftapping.Figure9illustratesaquenchingzonenearthesur-facewheretheslagquicklysolidifiedasaglass,whileinthecentre,wherethecoolingratewouldhavebeenslower,it has formed crystals. The micrograph of sample S2B (Fig.10)showsamagnetiteskinseparatingtwosubsequenttappinglayers.Suchamagnetiteskinformsonlywhenslagflowsintoairatambienttemperature,i.e.outsidethefur-

(a)

Figure 5IronslagsampleT1E:analmostfrothymacrostructure.

Figure 6Micrographof iron slagT1E: calciumsilicate inclusions(grey)inaglassymatrixwithnumerousmetallicprills(bright)andabundantporosity(black)(widthofimagec.2mm).

Figure 7IronslagsampleT1E:close-upofametalprillshowinganiron-ironcarbonalloystructure(widthofimagec.0.2mm).

Figure 8SlagsampleS2B:cleartappingstructure.

Page 5: Xie Rehren 2009 Lead Silver Slag China BUMA VI

SCIENTIFICANALYSIS OFLEAD-SILVERSMELTINGSLAGFROMTWOSITES INCHINA

181

Figure 9MicrographofslagsampleS2D:theglassyquenchedouterregion(bottom)andthemorecrystallineinnerpart(widthofimagec.1mm).

Figure 10MicrographofslagsampleS2B:amagnetiteskin(lightgrey,centre)separatingtwosubsequentslagflows(widthofimagec.1mm).

Figure 11(a)MicrographofslagsampleS2B:typicalstructurerichinsulphideinclusions(bright),andvarioussilicatecrystals(differentgreyshades)inaglassymatrix.(b)MicrographofslagsampleS2A:severaldifferentsilicatephasesinaglassymatrixtogetherwithnumeroussulphideinclusions(bright)(widthofimagec.0.2mm).

Figure 12(a)MicrographofslagsampleS2E:thesilicateslagphase(left,dark)andthelargemetallicpart(right,bright)(widthofimagec.1mm).(b)detailofthemetallicpartinthesamesampleshowingacopper-richmatrixandvariouscopper-arsenicphases(differentgreyshadesandcrystalshapes)(widthofimagec.0.1mm).

(a) (b)

(a) (b)

Page 6: Xie Rehren 2009 Lead Silver Slag China BUMA VI

PENGFEIXIEANDTHILOREHREN

182

nace.Incontactwiththeoxygenoftheairsomeoftheironintheslagnearthesurfaceissufficientlyoxidisedtoformmagnetitecrystals.Thisskin is thencoveredby thenextflowofslagandisthereforepreservedinthecontactzonebetweenthetwoflows.Often,thesetwoflowsexhibitdif-ferentcrystalsizes,duetodifferentcoolingspeeds.

The samples have different amounts of sulphide andmetalinclusions(Fig.11).Theyalsohavevarioussilicatecrystals,mostlypyroxeneandbarium-richfeldspars,andaremoreporousthanthepreviousgroupofleadslags.Theirchemicalcompositiondifferssignificantly;theyarericherinironoxide(35–40wt%)andhavemuchlesslime(lessthan5wt%).Inaddition,threehavehighlevelsofleadoxide (2wt%PbO)andzincoxide(3wt%ZnO),whilethefourthhas only 0.3wt%PbO and 0.9wt%ZnO.High bariumoxidelevelsaremostcertainlyduetothepresenceofbaryte(BaSO4)asaganguemineral,atypicalcomponentofmanyleadores.Allsamplescontainedmetalandsulphideprills;inS2A theseweremostly ironarsenide (70–90wt%Fe,10–30wt%As)andlowamountsofcopperandantimony,whilethesulphideswerepredominantlyleadsulphide.InsampleS2Bweonlyfoundleadsulphideinclusions,whileinsampleS2Dasinglegoldprillwasfound(91wt%Au, 8wt%Feand1wt%Zn),andnumerouslead-ironsulphideprills.Overall,itisreasonabletoassumethatthethreesam-plesS2A,BandDareallleadsmeltingslags,althoughthepresenceofdifferentmetallicphasesindicatesthattheorewasrelativelycomplex,andthatothermetalscouldhavebeenproducedaswell.ThereislessindicationhereforthepresenceofsilverintheslagscomparedtotheTangcountysite;nofragmentsoflithargeorothercupellationremainswerefoundduringfieldwork.

However, in viewof the complex compositionof theleadslagsinthisgroupitisnotsurprisingtonotethattheothertwosamplesfromthissite,S2CandE,areconsider-ablyricherincopper(3–6wt%)anddifferinotherdetails,too.Theiroverallchemicalcharacteristicsarequitesimi-lar though,with a high barium content and very similarironoxideandlimelevels.SampleS2Ccontainednumer-ousmetalandsulphideprills:someprillshadanaveragecompositionof90wt%Pb,7wt%Cuandafewpercentarsenic;otherswere70wt%Cuand30wt%As;yetoth-erswerenearlypureironmetalwithalowcoppercontent.Thesulphideinclusionswerebothlead-andcopper-rich.SampleS2Ehadahugemetallicprilltrappedinit;SEM–EDSanalysisgaveanaveragecompositionofabout70wt%Cu,15–20wt%As,5–8wt%Fe,andminorquantitiesoflead,sulphur,antimonyandzinc.Occasionally,individualsmallprillsintheslagS2Ewereidentifiedasbeinglead-oriron-dominated,butalwayswithahighcoppercontent.Overall,weinterpretthesesamplesascopperslag,possi-

blyexploitingthesameoredepositbutobtainingdifferentmineralsfromit.

Thus,wesee twodifferent,butclosely relatedmetal-lurgical processes here: one smelting leadmetal and theotherarsenicalcopper.ThehighbariumcontentofsamplesS2CandS2Ematchesthecopperslagcompositioninlate

Sample MgO Al2O3 SiO2 P2O5 SO3 K2O CaO TiO2 FeO CuO ZnO BaO PbOS2A 1.1 6.8 34.4 0.2 3.5 1.2 3.8 0.8 33.3 – 3.1 9.8 1.5S2B 1.1 6.7 33.9 0.2 3.0 1.1 4.1 0.8 34.3 – 3.3 9.5 1.8S2D 1.7 5.5 32.8 0.1 2.0 0.8 4.5 0.5 40.4 – 3.4 6.0 2.0S2C 0.8 3.0 21.5 0.1 9.2 0.3 1.5 1.0 40.8 6.4 0.9 12.3 0.3S2E – 2.2 23.0 – 5.7 – 1.5 – 42.7 2.9 0.9 20.6 –(P)ED-XRFdata,quantifiedbytheTurboquantmethod.Datanormalisedto100wt%.

Table 3

Sample Cu Sb Sr AsS2A 1100 75 1300 3200S2B 950 30 2000 1100S2D 830 25 1100 1350S2C 70000 – 1150 3700(P)ED-XRFdata,quantifiedbytheTurboquantmethod.

Table 4

prehistoricXinjing(Mei2000:52).Theyalsoshowhighcopperandarsenicconcentrations,andmanyinclusionsofcopper-arsenicalloy(Fig.12).Itislikelythattheseslagsoriginatefromthesmeltingofacomplexore,andthatbothleadandcopperwereproduced.Atthisstage,itappearsasifthereweretwophysicallyseparateoperations,andnotajoint lead-coppersmeltingprocess.However,moreworkandexcavationwouldbenecessarytoclarifythistechnicalaspect.Furthermore, it isuncertainwhether thesmeltingalsoincludedtheproductionofsilverorwhetherthiswaspredominantlyabasemetalmetallurgy.

From the above analyses, we summarise that samplegroupS2consistsofthreelead(silver?)slagsamples,andtwoarsenicalcopper(lead?)slagsamples.Thethreeleadslagsamples(S2A,S2BandS2D)aremostlyglassyandhaveonlysmallmetalorsulphideinclusions,whilethetwocopperslags(S2CandS2E)appear lessglassyandhavemanymoremetalandsulphideinclusions.Sofar,nofirmevidenceforsilverproductionhasbeenfoundamongtheslagsfromthissite.

Conclusions

Through scientific analysisof11 slag samples from twosites,wefindthatsevenareleadandpossiblylead/silversmeltingslags,twoareironslags,andtheremainingtwoarecopperslagsfromamixed(arsenic)copper-leadore.

Ifwecomparethecompositionofthelead-silverslagsfromthetwosites(seeTables1and3),wefindthatincon-sidering themajorelements,groupS2slagshavehigherconcentrationsofFeO,BaO,ZnO,SO3,TiO2andAs2O3thangroupT1,whilegroupT1slagshavehigherlevelsofSiO2,MgO,CaOandMnO.Amongthetraceelements,groupS2

Page 7: Xie Rehren 2009 Lead Silver Slag China BUMA VI

SCIENTIFICANALYSIS OFLEAD-SILVERSMELTINGSLAGFROMTWOSITES INCHINA

183

sampleshavemuchhigherconcentrationsofcopper,arsenicandstrontium,whilegroupT1samplesshowhigherlevelsofsilver.Thesedifferenceswereclearlycausedbydifferentoredepositsandspecificallyadoptedsmeltingprocesses,aswouldbeexpectedfromtheirdifferentgeologicalorigin.

FromtheTangcountysitethereisevidencefortheuseoftuyèrestoblowairintothefurnaces.Theyaremadefromahigh-qualityrefractoryclayandwouldhavelastedalongtimeevenathightemperatures.Itisassumedthattheyfilledwithslagwhentheblowingstoppedattheendoftheproc-essandtheslaglevelhadincreasedinthefurnace.Morefieldworkisnecessarytoinvestigatethisaspectfurther,bylookingatfragmentsoftuyèreswithoutslagfillingandpos-sibleotherfurnacefragmentsthatwouldprovideimportantinformationaboutthefurnacetype.Furtherarchaeologicalworkisalsonecessarytodatethemetallurgicalactivitiesatbothsites.

LeadsmeltinginChinabeganveryearlyandthemetalwasmass-producedintheChineseBronzeAgetomeettheenormousdemandforlead-containingbronzeobjects(Sunand Li 2003).Archaeological excavations have demon-stratedthatasearlyastheShangdynasty(16th–11thcen-turyBC),theChinesepeoplewereabletomakealmostpureleadmetal(Li1984).However,itisnoteworthythatalltheslagsstudiedherearefromatypetypicalofadvancedblastfurnacetechnology,asisknownfromancienttextandillus-trations(seealsoMeiandRehren2005).

EarlyleadslagsareknownfromnumeroussitesinancientEurope;fromtheBronzeAgethroughtheRomanperiodandtheMiddleAgesintotheearlymodernperiod.Thispapercannotgiveacomprehensiveoverviewoftheirchemistryandmineralogybutsufficeittosaythatleadslagsarefarmore variable in their appearance and composition thantypicalcopperorironslags,reflectingthewiderchemicalvariabilityofleadoresandtheassociatedgangueminerals.Theclosestparallelsintermsofbulkcomposition,glassyappearanceandstronglyreducingconditionsarefromvari-ousmedieval to earlymodern sites in theSchwarzwald,south-west Germany (Goldenberg 1994), thought to bebasedonawater-poweredblastfurnacetechnology.

Ofparticularinterestistheevidenceformultiplemetal-lurgicaloperationsatbothsites.InTangcounty,thelead-silver slag ismixedwith typical iron blast furnace slag;themacroscopicandmicroscopicappearanceofbothslagsissosimilarthatitcanbeassumedthatverysimilarfur-naceswereemployedintheirproduction,eventhoughtheir

chemicalcompositionissignificantlydifferent,asbefitsthesmeltingoftwosuchdifferentmetals.IntheShizhucountysite,tworelatedmetalswereproduced:leadandcopper.Wecanspeculatethattheycamefromthesameoredeposit,butmayhavebeensmeltedseparately.Clearly,thesesiteshaveahugepotentialforfurtherarchaeologicalandarchaeomet-allurgicalwork.

Acknowledgements

WewouldliketoexpressourgratitudetothelateProfessorP.UckooftheUCLInstituteofArchaeology,whoasExecutiveDirectorofthe InternationalCentre forChineseHeritageandArchaeology fa-cilitatedPengfeiXie’sstudyatUCL;theInstituteofHistoricalMet-allurgyandMaterialsofUSTB,andinparticularProfessorKoandProfessor Li Yanxiang, for permission to study the material; and DrMarcosMartinón-Torres,SimonGroomandKevinReevesoftheUCLInstituteofArchaeologyfortheirgeneroushelpinvariousas-pectsofthisresearch.

References

Goldenberg, G. (1994)Archäometallurgische Untersuchungen zur Entwicklung des Metallhüttenwesens im Südschwarzwald (doc-toralthesis,UniversityofFreiburg,Germany).

Li,Minsheng(1984)‘Xianqinyongqiandelishigaikuang’,Wenwu 10:85–8[inChinese].

Mei,Jianjun(2000)Copper and Bronze Metallurgy in Late Prehistoric Xinjiang: Its Cultural Context and Relationship with Neighboring Regions.BARInternationalSeries865.Oxford:Archaeopress.

Mei,JianjunandRehren,Th.(2005)‘CoppersmeltingfromXinjiang,northwestChina.PartI:Kangcunvillage,Kuchecounty,c.18thcenturyAD’,Historical Metallurgy39:96–105.

Rehren,Th.,Groom,S.andAnarbaev,A.(forthcoming)‘SteelmakingatAkhsiket:ananalysisofUzbekcruciblesteelslags’.

Sun, Shuyun and Li, Yanxiang (2003) Zhongguo Gudai Yejin-jishu Zhuanlun.Beijing:ChineseScienceandCulturePress[inChinese].

Xie,Pengfei(2005)Documents of Ancient Chinese Silver Smelting Technology (unpublishedmaster’s thesis, Institute ofHistoricalMetallurgyandMaterials,UniversityofScienceandTechnologyBeijing).

Xie, Pengfei (2006)Scientific Analysis of Slag Samples from Two Sites in China (unpublishedMSc dissertationUCL Institute ofArchaeology,London).

Page 8: Xie Rehren 2009 Lead Silver Slag China BUMA VI

ForewordbyWeidongLuo,Chancellor,UniversityofScienceandTechnologyBeijing viiForewordbyRobertMaddin,ChairmanoftheBUMAStandingCommittee ixPreface xiAcknowledgements xiiiListofcontributors xviiIntroduction xxi

Early metallurgy across Eurasia

AncientmetallurgyintheEurasiansteppesandChina:problemsofinteractions 3EvgenijChernykh

EarlymetallurgyinChina:somechallengingissuesincurrentstudies 9JianjunMei

MetaltradeinBronzeAgeCentralEurasia 17LiangrenZhang

Documentaryandarchaeologicalevidenceforanantiquecopper-nickelalloy(baitong) 26productioninsouthernChinaanditsexportationtoIndiaFrançoisWidemann

MetaltradebetweenEuropeandAsiainclassicalantiquity 35AlessandraGiumlia-Mair,MichelJeandinandKen’ichiOta

TheblackbronzesofAsia 44PaulCraddock,MaickelvanBellegem,PhilipFletcher,RichardBlurtonandSusanLaNiece

Bronze casting technologies in ancient China

OriginsandevolutionofthecastingtechnologyofAnyangbronzeritualvessels:anexploratorysurvey 55YuLiu

ThreeWesternZhoubronzefoundrysitesintheZhouyuanarea,Shaanxiprovince,China 62WenliZhou,JianliChen,XingshanLei,TianjinXu,JianrongChongandZhankuiWang

Newresearchonlost-waxcastinginancientChina 73WeirongZhou,YaweiDong,QuanwenWanandChangsuiWang

IncipientmetallurgyinYunnan:newdataforolddebates 79TzehueyChiou-Peng

Contents

Page 9: Xie Rehren 2009 Lead Silver Slag China BUMA VI

METALLURGYANDCIVILISATION:EURASIAANDBEYOND

vi

AstudyofthesurfacecraftofweaponsfromtheBa-ShuregionofancientChina 85ZhihuiYaoandShuyunSun

ProductionofsignatureartifactsforthenomadmarketinthestateofQinduring 90thelateWarringStatesperiodinChina(4th–3rdcenturyBCE)KatherynM.Linduff

Ancient iron and steel technologies in Asia

Anearlyiron-usingcentreintheancientJinstateregion(8th–3rdcenturyBC) 99RubinHanandHongmeiDuan

FromwesternAsiatotheTianshanMountains:ontheearlyironartefactsfoundinXinjiang 107WuGuo

SouthIndianIronAgeironandhighcarbonsteel:withreferencetoKadebakeleand 116comparativeinsightsfromMel-siruvalurSharadaSrinivasan,CarlaM.Sinopoli,KathleenD.Morrison,RangaiahGopalandSrinivasaRanganathan

SurvivaloftraditionalIndianironworking 122VibhaTripathiandPrabhakarUpadhyay

Finestructures:mechanicalpropertiesandoriginofironofanancientsteelsword 129excavatedfromanoldmoundinJapan MasahiroKitada

Specialisationiniron-andsteel-makingintheearlyMiddleEastandCentralAsia: 134myths,assumptionsandareassessmentofearlymanuscriptevidence BrianGilmourAncient metallurgical and manufacturing processes

Theearlyhistoryoflost-waxcasting 147ChristopherJ.Davey

Anaturaldraughtfurnaceforbronzecasting 155BastianAsmus

Theliquationprocessutilisedinsilverproductionfromcopperore: 163thetransfertoanddevelopmentinJapanEijiIzawa

AtechnicalstudyofsilversamplesfromXi’an,Shaanxiprovince,China, 170datingfromtheWarringStatesperiodtotheTangdynastyJunchangYang,PaulJett,LynnBrostoffandMichelleTaube

Scientific analysis of lead-silver smelting slag from two sites in China 177PengfeiXieandThiloRehren