Society for American Archaeology Ceramic Diversity in Chavín De Huantar, Peru Author(s): Isabelle C. Druc Source: Latin American Antiquity, Vol. 15, No. 3 (Sep., 2004), pp. 344-363 Published by: Society for American Archaeology Stable URL: http://www.jstor.org/stable/4141578 . Accessed: 18/10/2014 15:06 Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp . JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org. . Society for American Archaeology is collaborating with JSTOR to digitize, preserve and extend access to Latin American Antiquity. http://www.jstor.org This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions CERAMIC DIVERSITY IN CHAVIN DE HUANTAR, PERU Isabelle C. Druc Ceramic diversity in the ceremonial center of Chavin de Huantar in the Peruvian highlands is assessed by neutronactivation analysis, petrography,and stylistic analysis. Theseanalyses lead to a new interpretationof ceramicproductionin Chavin between ca. 850 and 200 B.C. Several compositional groups with very differentmineralogical paste types are identified. More than 30 percent of the ceramics are nonlocal. Most of these are bottles and fine wares, probably brought to the site as gifts or offerings. At the local level, ceramic production changed over time. A dramatic shift in resource procurement occurred at the end of the first occupation phase, and production became more diversified and intense as the site and its population grew larger Estudiosde activacidn neutr6nicay petrograffa,y andlisis estilistica muestranla diversidadcerdmicaque existe en el centro ceremonialde Chavinde en los Andes de Perd. Se proponennuevas interpretacionessobre la produccidncerdmica en Chavinentre 850 y 200Huantar, a. C. Variosgrupos de composicidnson identificados,con pastas mineraldgicasmuydistintas.La mayoria de las vasijas exogenas son botellas y cuencosfinos, probablementellevados al centro ceremonialcomo ofrendaso productosde intercambio.La activacidnneutrdnicarevela que, al nivel local, la produccidncerdmicacambia en el curso de existenciadel sitio. Se nota el uso de pasta volcdnicadurantelas dos primerasfases de ocupacidn (Urabarriuy Chakinani)y la produccidnde vasijas principalmenteutilitariascon poca variacidndeforma. En cambio, durantela tercerafase de ocupaci6n (Janabarriu)se usa un material intrusivocomo la granodioritaprocedentede la CordilleraBlanca. La produccidnes mds diversificadae intensa cuando el sitio conoce su desarrollo mdximo.El estudio de activacion neutrdnicamuestra tambien que las cerdmicascon un alto contenidodel elementocesium no son locales. La mayoriade estas cerdmicasson de estilo at'pico. Las cerdmicasno locales tienen composicionesquimicasdiversas lo que sugiere multiplesorigines. The ceremonial center ofChavindeHuantarone of the many galleries of the old temple. The is knownfor its fine architecture,extensive stone art, and beautiful ceramics. The Chavinoccupationspansnearly700 years,between 850 and 200 B.C. The site is located at the crossroads of selva (tropicalforest)-coastand northsouth highland routes (Figure 1), at 3,200 m elevationin theupperMosnaValley.TheU-shaped center has two main temples (the old temple, believedto be the firstone built,andthe new temple, an extensionor additionto the old structure), a small sunkencircularplaza, and a largerectanThesite gularplazaflankedby lateralconstructions. relates to the pan-Andean stylistic horizon of Chavinandis associatedwiththe spreadof Chavin ideology. Fringed creatures,supernaturalbeings withfeline, raptorialbird,andsnake-likeattributes arecarvedon stoneslabs,columns,andmonoliths. The Lanzon,a 5-m-highsculpture,is foundwithin ceremonialcenteris surrounded by a settlementthat reaches proto-urbansize by 250 B.C. Due to its location, religious importance,and the ceramic offeringsfoundin Chavin,this site has been interpretedas a pilgrimagecenter(Keatinge1981;Lumbreras1974). Its economicimportanceis reflected by the presence of nonlocal commodities like obsidian,ceramicsof foreign style, and products from the tropicalforestandPacificcoast. Spondylus shellsfromthe warmwatersof coastalEcuador are anotherexampleof long-distancetrade. Chavininfluencehas been felt in manyregions of Peru,in boththe highlandsandalong the coast. Many sites bear stylistic traits that link them to Chavinideology.Ceramics,textiles,bones,carved stones,andclay friezestestifyto this culturalaffiliation. The spreadof Chavinideology and style, however, is not clearly understood. Chavin de Isabelle C. Druc 0 Departmentof Anthropology,5240 Social Science Building, 1180 ObservatoryDrive, Universityof Wisconsin-Madison,Madison,WI 53706-1393 LatinAmericanAntiquity, 15(3), 2004, pp. 344-363 Copyright@2004 by the Society for AmericanArchaeology 344 This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions REPORTS Huantaris seen as representinga synthesisof earlier traditionsanda centerfromwhich ideas,if not objects, radiated.Indeed, Chavin-styleceramics are common at manyPeruviansites. It was, however,not certainif theirpresencewas the resultof imitation(diffusionof ideasor styles)or trade(diffusionof objects).This questionpromptedthe present INAA (instrumental neutron activation analysis) study, which completes earlier provenancestudieswithpetrographyandXRF(x-rayfluorescence) analyses of Chavin ceramics (Druc 1998a).A few compositionalstudieshadbeenpreviously undertakenbut not at a regionallevel and withprovenancein mind.Theseearlieranalysesare first reviewed to place the present study in perspective.Methodologyandresultsfollow. A Review of Chavin Ceramic Analyses Although stylistic analysis offers insight into the provenanceof ceramics,compositionalanalysisof the pastehelps to differentiatelocal productsfrom imports,thusprovidinga meansto investigateinterregional interactions.The initial compositional studiesof ceramicsfrom Chavinde Huantarwere undertakenin theearly1980s.Thefirststudy,a petrographicanalysisof nineceramicthinsectionsby RobertTracy,was publishedin RichardBurger's (1984) excavationreport.The samplescame from ceramicsfound outside the ceremonialcenter,in the surroundingChavin settlement.Tracyidentified several mineralogical compositions (with quartz,plagioclase,biotite,homblende,metamorphic, fine-grainedvolcanic, and plutonicigneous rockfragments).His observationscorrespondto my analyses. The second study, using Maissbauer spectroscopy (MS), instrumentalneutron activation analysis (INAA), and X-ray diffraction(XRD), was conductedby a Germanteam (Salazaret al. 1986) on 61 Chavinsherdsfrom an interiorcorridor of the old temple of Chavin,50 surfacefinds (Chavin),two ancientsoil samples,andone modem clay sample.The aim of the studywas to classify the samplesandlearnaboutfiringprocedure. The hierarchicalclusteringof the INAA results produced five groups with chemically distinct pastes.The largestgroupincluded57 Chavinand post-Chavinceramics,andthe recentclay sample. It was thereforeassumedthatthe wareshad been 345 locally produced.The MS analysisindicatedthat the ceramicswere firedin a reducingatmosphere followed by an oxidationphase.X-raydiffraction (XRD) identifiedsome of the mineralswithinthe ceramics: black talc and feldspar for the main group, and amphibole crystals for some of the other samples. These two studies showed that ceramics at Chavin de Huantarhad variedcompositionsand some were of nonlocal origin. Stylistic analysis had produceda similarconclusion(Burger1984; Lumbreras1977, 1993). Based on style, northern andcoastaloriginswereproposedfor thenonlocal wares. Anothercompositionalstudy(MS, INAA, thin section microscopy)was conductedby the same Germanteam(Lumbreraset al. 2003) on 70 sherds from the OfrendasGalleryof Chavinde Huantar. Thisgalleryis famousformorethan500 finebowls and bottles, believed to be offerings left in the gallery on multiple occasions (Lumbreras1977, 1993).The Ofrendasmaterialis, however,stylistically differentfromtheotherceramicsfromtheceremonialcenteror the surroundingsettlement.The hierarchicalclusteringof the INAA results conformed to stylistic groups within the Ofrendas material.Gebhardet al. (1996) suggesta nonlocal provenance,butno comparativematerialfromother siteswas analyzedto allowidentificationof origins. Stylistic affinitiespoint to the central and north coast, andto the northernhighlands(Burger1984; Lumbreras1993). Until 1994, no comprehensive study of a regional scale had been attemptedto understand how potterywas producedin the Chavinarea.Was it madeatdifferentworkshops?WereChavin-style ceramicscentrallyproducedandthendistributed? No ceramicworkshopsorkilnshavebeen foundin Chavinde Huantaror in the nearbyregion. A studyof ceramicproductionanddistribution in the Chavinsphereof influence(Figure 1) was conducted in 1993-1997 (Druc 1998a, 1998b), usingpetrographyandenergydispersiveX-rayfluorescenceanalysis(EDXRF).The studyincluded soil andmodemclay 284 ceramicsand69 modemrn samplesfromsix siteslocatedin five differenthighlandandcoastalregions(Chavinde Huantar,Huaricoto,Pallka,Anc6n,andtwo smallsitesin Nepefia). The analysis results showed that most of the Chavin-styleceramicsin the Chavinspherewere This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions 346 LATINAMERICAN ANTIQUITY Pacific [Vol.15, No. 3, 2004] Huaura OceanUma Garaga ' 0 100 km Uma Figure 1. Study region and sites (after Druc 1998a, Figure 4). locally made imitations.Furthermore,it proveda small-scaleexchangeof bottlesandfinebowls into the ceremonialcenterof Chavinde Huantar,(and not outwardfrom Chavin).This conformsto the idea of Chavinde Huantaras a pilgrimagecenter. The study also yielded new insights into ceramic production.Petrographicanalysisshowedthatthe wareswereproducedin severalworkshopsor production areas at each site, and that bottles were very often producedwith a fine paste. Thepatternof productionattheceremonialcenter of Chavinde Huantar,however,provedmore complexthanat anyothersite studied.The ceramics analyzedin the 1993-1997 study come from the ceremonialcenter(n = 86) and the associated settlement(n = 23). As seen in priorstudies, the ceramicsshow high compositionalheterogeneity. Six differentcompositionalgroupsandsubgroups were petrographicallyidentified,along with unattributedspecimens of atypicalcomposition.Two maingroupsareidentifiedas local. One is characterized by a volcanic-derivedpaste (pyroclastic sandwith quartz,plagioclase,biotite,hornblende, welded rhyolitictuff fragments),and the otherby an intrusive-derivedpaste (quartz, plagioclase, biotite, augite, granodioritefragments).A third composition groupof mixedintrusive-sedimentary was also foundto be local, on the basis of geological and abundancecriteria.Subgroupsaregranulometricor modal variants.Ceramicswith exotic pastesdisplaymetamorphiccomposition(quartzmuscovite-schist) or rare compositions (felsic, granitic)not correspondingto local geology (for detailssee Druc 1997, 1998a).The EDXRFanalysis suggestedthatat least 30 percentof the ceramics found in Chavinde Huantarwere of nonlocal production.This figureconformsto the findingof the INAA studyby Salazarandcolleagues(1986), where25 percentof thewareswerefoundto be nonlocal (20 out of 76 wares). This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions 347 REPORTS Table 1. Analyzed Sample of ChavinCeramicsby Types and Vessel Forms. Urabarriu Bowls Ollas Jars Bottles Total Chakinani 4 8 1 1 14 6 3 4 8 21 Althoughx-rayfluorescenceworkedwell atthe intersitelevel, it failed to distinguish individual groupsat the intrasitelevel andwouldnotpartition the corpus as accuratelyas petrographicclassification.Forthis reason,andto get a clearerpicture of ceramicproductionat Chavinde Huantar,it was decided to submit the Chavin corpus to another roundof analysis,this time using neutronactivation. In addition,the chemicalresultscould be triangulatedwith differentsets of data:petrographic, geological, stylistic, archaeological, and ethnographic.The resultsof this last studyarepresented here. The methodologywill be describedfirst.The resultsarethendiscussedin light of the chemical, mineral,and stylistic data.Finally,the discussion is enlargedto reacha higherinterpretative levelwith regardto ceramicproductionthroughtime in relation to the developmentof Chavinde Huantarand interregionalinteractions. Methodology Several questions defined the objectives of the INAA study. Its goal was identifying and interpretingintrasitechemicalgroupsandinvestigating compositionaldifferencesin relationto time and space. How many differentchemical groups are therein Chavinde Huantarceramics?Whatdo they reflect?Is therea change in ceramiccomposition throughtime? Is therea centralizationof production? The sample is comprisedof 81 specimens fromChavinde Huantar(56 sherds,and25 thathad alreadybeen used for EDXRF), and 20 pressed powderpellets from sherdsfrom the sites of Pallka, Huaricoto,Anc6n and Garagay(Figure1) for comparativepurposes.No bindingagentwas used for preparingthe pellets,whichwerekeptin a desiccatorafteruse. The sampleshaveknownmineral compositionfromthe previousanalyses. The Chavinsamplecomes fromthe threeoccupational phases and related styles identified by Janabarriu 15 5 3 5 28 Chavin-Style 1 1 0 3 5 Atypical 4 2 3 3 12 RichardBurger(1984, 1988) at Chavinde Huantar:Urabarriu(850 to 460 B.C), Chakinani(460 to 390 B.C.), and Janabarriu(390 to 200 B.C.). The differentwareformsin the samplearerepresentative of those found on the site: bowls, neckless ollas,jars,andbottles(Table1).Theceramicsfrom the ceremonialarea (n = 70) come from excavations by WendellBennett(1938, 1944), and bear identificationnumbers3700 and above (Table2). Theceramicsfromtheancientsettlementof Chavin de Huantaroutsidethe ceremonialcenter(n = 11, ID# B 1 and above) come from excavations by RichardBurger(1984). Table1 shows the number of samplesanalyzedby type andvessel form.The Chakinani,andJanabarriu Urabarriu, typesarespecific to the respectivetimeperiodsof the site occupation. The Chavin category groups ceramics of Chavin style, which could not be identified as Urabarriu,Chakinani,or Janabarriutype ware. Most of the analyzed ceramics from Chavin de Huantarare represented in Figure 2. They are grouped according to the chemical and mineralogicalgroupsin Table3. Vessel formsper group are listed in Table2. The neutronactivationanalysiswas conducted at the SmithsonianCenterfor MaterialResearch and Education(SCMRE)and at the nuclearreactorfacilityof theNationalInstituteof Standardsand Technology(NIST).Neutronactivationanalysisin general,andthe procedurefollowed at NIST,have been amply described elsewhere (Bishop and Crown 1994; Bishop et al. 1982; Glascock 1992; Perlmanand Asaro 1969). James Blackmanand RonaldBishop conductedthe elementalanalysis. The sampleswerepreparedfrom200 to 400 mg of pastedrilledfromthesherdcross-sectionsandfrom the pressedpowderpellets. Standardswere run at the same time with the powdersamplesenclosed in plasticvials.Of the 31 elementscommonlymeasuredwith INAA, the following 18 elementswere used for the statisticalanalysis:Na, K, Sc, Cr,Fe, Rb, Cs, Ba, La, Ce, Sm, Eu,Tb,Yb, Lu, Hf, Ta,Th. This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions 348 [Vol.15, No. 3, 2004] LATINAMERICAN ANTIQUITY Table 2. Vessel Formsper CompositionalGroup(See Table 3 for GroupCharacteristics). Group 1 3742a 3742b 3764e 3786e 3786h 3686i 3745c 3748c 3748d 3780b 3781c B 19 3781D B17 3757B 3787A 3787b Urabarriubowl Urabarriuolla Urabarriuolla Urabarriuolla Urabarriubowl Urabarriujar Chakinanijar Chakinanijar Chakinaniolla Janabarriuolla Janabarriubowl Chakinanibowl Atypical olla Chakinanibowl Chakinanibowl Urabarriuolla Janabarriubowl Group 2 B4 3781h 3781i 3781d 3780H 3783b B 13 3786b 3784c 3780c 3781j 3781k 3787c 3781e 3783H 3700 Janabarriubowl Janabarriuolla Janabarriuolla Janabarriubowl Janabarriubowl Janabarriuolla Chakinaniolla Urabarriuolla Janabarriubowl Janabarriubowl Janabarriujar Janabarriujar Janabarriubowl Janabarriubottle Janabarriubottle Atypical bowl Group 3 3780a 3783a 3786c P23 Urabarriuolla Janabarriujar Urabarriubowl Pallka bottle Group4 3748b 3764f 3774 3783d B21 3743a 3748e B14 3755a Urabarriujar Urabarriubottle Group7 Chavinbowl 3764b Janabarriubowl Janabarriubowl Chavin bottle Group8 Chavinbottle (spout fragment) 3775a Chakinanibottle (spout fragment) 3786f 3785 Atypical bowl Group 5 3762 3764a 3786g 3755b 3745a B20 3765a 37841 B8 3783c 3780e B21 P3 P4 P14 P24 Atypical bottle Atypicaljar Atypicaljar Chakinanibottle Chakinanijar Chakinanibowl Chakinanibottle Janabarriubottle Janabarriubottle Janabarriubowl Janabarriubowl Janabarriubowl Pallka olla Pallka olla Pallkabottle Pallka bottle The elements were chosen accordingto smallest analyticalerror,expectedvalues in ceramics,and inspectionof the raw data.The alkali Cs and Ba were importantdiscriminatorsfor this specific set of samplesand were thus includedin the statistical analysis.Raw data were log transformedand datawere exploredvia multivariateanalyses. Differentclusteringmethodsas well as principal componentanalysis(PCA),using bothcovariance-variance and correlation matrices, were explored. Cluster analysis was conducted using log-transformeddata, mean Euclideandistances, andhierarchicalagglomerativecompleteandaverage link methods. Both linkage methods proved conclusivein groupingthesamples,yieldingstrong groupoverlap,which suggestsgenuinepatterning. Group 6 3784f B 15 3790B 3787d 3787B 3787a 3754B Atypical bottle Chakinanibottle Atypical bowl Urabarriubowl Atypical bowl Atypical bottle Chakinanibottle Janabarriujar Janabarriubowl Urabarriujar Urabarriumolcajete The resultsof the two clusteringmethodswill be discussed as they provide complementaryinformationon ceramicproductionin Chavinde Huantar. The average linkage allowed for more homogeneousclustersin termsof themineralcompositionof the samples.Clusteranalysis,however, imposes different patterning depending on the methodused and thereforerequiresexternalvalidation of the results (Aldenderfer1982; Baxter 1994;Shennan1988).Consequently,workingwith the raw listings was importantin evaluatingthe clusters,along with informationon style andmineralogy.In principalcomponentanalysis,the variance-covariancematrixwas preferredoverthe use of a correlationmatrix.The lattercentersthe distribution,an approachnot recommendedfor very This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions Table 3. Classificationof the Ceramic Samples Based on HierarchicalClusteringby the Average-LinkM Groups Group 1 N 101a(81) 18; 17.8%;(22.2%) Group l a Group2 3; 3%; (3.7%) 19; 18.8%;(23.4%) Group 3 Group4 Group5 4; 3.9%; (3.7%) 12; 11.9%(14.8%) 16; 15.8% (11.9%) Group6 Group7 7; 6.9%; (8.6%) 12; 11.9%;(3.7%) Group 8 3; 3%; (3.7%) Group9 + Outliers Descriptionb red slip ollas/bowls, combed ollas/bowl; 1 circle-dot bowl, 2 jars, 1 atypical olla; U, C, 3J 1 circle-dot bowl, 2 jars; J, C, atypical 5 ollas, 2 undec, 3 dec & 2 circle-dot bowls, 4 bottles, 2 jars, 1 atyp bowl; J, 3C, 2U dec. olla & jar, bowl, +Pallka bottle; 2U, 2J 6 bottles, 3 bowls, 2 circle-dot bowls, 1 atyp. Jar;U, C, J 4 Pallka +12 Chavin sherds, ollas, bottles, jars, circle-dot olla + bowl, atypical bottle + jar; C, J bottles, atyp. & dec. bowls and bottles; U, C, atypicals 3 Chavin, 3 Garagay,3 Huaricoto,3 Anc6n; ollas, bowls, bottles, jars; U, C, J olla, jar, circle-dot bowl; 2 U, J 7; 6.9%; (1.2%) Mineralogyc volcanic tuff, pyroclasticmaterial volcanics, pyroclastic material basic-interm.intrusive +2 sed/intr 2 volcanic, 2 granitic fine paste + sedimentary intrusive,sedimentary, 1 volcanic sedimentaryfine paste mixed qz-muscovite schist csed = sedimentery,intr = intrusive,metam = metamorphic,interm = intermediate,qz = quartz. This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions low Cs high Cs very hi Ce, Nd Yb, Lu Chak bowl 3708A, 2 Huaricoto, pellitic, + coast 2 Anc6n, 2 Garagay apercentagesare based on 101 samples (all sites included), and 81 samples for Chavin de Huantaronly (in parentheses) bU = Urabarriu,C = Chakinani,J = Janabarriu,dec = decorated,atyp = atypical. Chemis low Cr Ce, Hf high Cs 350 LATINAMERICAN ANTIQUITY heterogeneous samples, as is the case with the ceramiccompositionsfrom Chavinde Huantar. Results and Discussion Eightcompositionalgroupsof archaeologicalsignificancewere identifiedin the hierarchicalclassification of the 56 samples from Chavin de Huantarand 20 comparativesamples from four other sites. These chemical groups are presented in Table3, along with informationon mineralogy, form, style, chemistry,and possible source areas; the sherdsamplesfor each groupareillustratedin Figure2. The PCA scatterplot for ceramicsfromChavin de Huantarshowsfive groups(Figure3). The main chemical variablesresponsiblefor this classification are (by orderof importance):Cs, Th, La, Ce, Rb, andK, for PC1, andCR, LU,YB, SC for PC2. Thefirstprincipalcomponent(PC1)explains42.34 percentof the varianceandthe second(PC2)23.25 percent. The chemical groups show internal heterogeneity.Compositionalvariabilitywithinclusters is above 17 to 20 percentfor most of the elements, suggestingmultipleoriginsfor the samplesin the sameclusteror the existenceof severalworkshops using clay from differentlocalized sourcesin the same productionarea. Consequently,representation in multivariatespace, such as a PCA scatter plot, is not easily interpretedvisually.The groups are spreadout and give the impressionof being intermixedbecauseof the reductionin the number of dimensionsto two orthree,particularlyso when the samples from the comparative sites are included.Nevertheless,the chemicalgroupspoint to productionareas with internallysimilar geochemicalenvironment.Due to the degreeof compositionaldiversityin Chavinde Huantarceramics, the sampleclassificationby clusteranalysison logtransformeddatais moreinformativethanthePCA andprovidesthe basis for the discussion. Sample Classification Thehierarchicalclassificationreflectsthemainpetrographicgroups observed in mineral analysis: acidic volcanic, intermediateto basic intrusive, fine/sedimentary,and mica tempers.The clustering constructedwiththecompletelinkmethodfollows an acidic to basic trend, with the acidic [Vol.15, No. 3, 2004] volcanicsampleson one endof thedendrogramand the basic intrusivesampleson the other.Elemental variationsoccuraccordingly,allowingthe characterization of the mineral trends in terms of chemicaltendencies.Table4 gives the concentration rangesfor each compositionalgroupderived by this method. Values outside the range are in parentheses,while valuesforoutliersarenotgiven. The group numbers in parenthesesrefer to the groupsformedby averagelinkage listed in Table 3. Differencesbetween the acidic and basic poles are mostly seen amongtraceelements.The acidic pole is characterizedby low Sc, and high Rb, Cs, La, Ce, and Th (the reverseis true for the basic pole). Ironcontentshows differencesaccordingto the mineralogyor provenancewithinthe clusters, witha low rangein thevolcanicgroupanda higher one in the basic-intermediate intrusive group. Exceptions in chemical trends are observed for ceramics from the coast (Anc6n and Garagay), Huaricoto,a few Chavin samples, and the three mica-temperedceramics.Samplesfrom the comparativesites arefoundin compositionalgroups3, 5, 7 and 9 and as outliers. In the complete-linkdendrogram,some groups encompasssampleswith differenttempercomposition, althoughsimilaritiesin mineralcontentare observed.Chemicalvariationsthatexist in theclay may not be observedin petrographydueto the resolutionof the microscope.The clay matrix,which often representsmorethan50 percentof the paste of the Chavin ceramics (Druc 1997), is largely uncharacterized.Consequently,finer distinctions arereachedwithINAAthanwithpetrography when dealingwithfineorlow-temperedfragments.Samples with similartempermay thus displaychemical variationsrelatedto the use of a differentclay paste. This is the case for the two Garagayollas, G4 andG6, whichwerebothtemperedwithcrushed granodioritebuthaddifferencesin Cs, Ba, Tb, Ta, Th, andRb content.Theirmineralogicalcomposition is verydifferentfromtherestof theAnc6nand Garagayceramics (see Druc et al. 2001), hence theirseparateclustering. The dendrogrambuilt using the average-link method does not follow the same acidic to basic trendin the orderingof the clusters.Nevertheless, the constitutionof the groupsis similar.The average-linkmethodyields a morehomogeneousclusteringof the samplesrelativeto theirmineralogy. This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions REPORTS 351 Group 1 Volcanicpaste 42a 42b 64e 86e 48d 886i5c48c 81c 81D B19 87b 57B 80a 83a 86c P Pallka Group 3 Volcanic+ Intrusive Group 2 Intrusive S81h 817 81d (*Sedimentary) 80 B4 080c* 81j 84c3 87c 81k 83 81e U0J 3700 3700 0 4 cm 8 Figure 2. The ceramic sample from Chavin de Huantar grouped according to chemistry and mineralogy. Identifying numbers for samples collected by Bennett (1938, 1944) in this figure correspond to sample numbers listed in Tables 2 and 5, but lack the initial "37"; e.g., 42a corresponds to 3742a in Tables 2 and 5. This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions 352 LATIN AMERICAN ANTIQUITY [Vol. 15, No. 3, 2004] Group4 Finepaste+ Sedimentary 48b 64f 74 '83d B21 43a 48e B14 55a Group 5 Chavin S64(*Volcanic 55b + ceramics 65a 45aB20 +P3,P4, P14,P24 GO 841 Pallka InstrusiveandSedimentary paste) B8 83c B1 80e *Group6 Sedimentary(*Finepaste) 87B 64b 87a* 54B Group7 Chavin Huaricoto Ancon Garagay 75a Group 8 Micapaste Micapaste 86f 85 4cm 0 Outlier Figure 2. continued. This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions 4 cm 353 REPORTS 44 20 + -1- 0) A ++ - O. + 0 + ?A - * _+ + o * AFine * paste * Sedimentary + Intrusive Atypicalpaste S, 0a Volcanic A* Mica + -2 -3 -3 -2 -1 0 PrincipalComponent1 1 2 Figure 3. PCA scatterplot of ceramics from Chavin de Huantar on the first two principal components. Covariance matrix, log transformed data. Cs as Site Discriminant In orderto allow a closer examinationof the data andto see whichelementscharacterizeeachgroup, the chemical raw data were listed following the orderof the hierarchicalclustering.One element in particular,cesium (Cs), provedto be important in theclassificationof thesamples.Cesiumappears to be a site discriminator,with high Cs content (above 11 ppm)suggestinga nonlocalprovenance fortheChavinsherds.However,thiscriterionis not thesole indicatorof foreignprovenanceandit must be appliedin conjunctionwith otherinformation like the style, mineralogy, and archaeological provenanceof the sherds. Cesium is likely to be found in late-stage-formationcrystalsin acidicrocks,in leucite,feldspar, and similarminerals(BruceVelde,personalcommunication2001). Itis also foundas tracesin black mica(biotite)(FoucaultandRaoult1995),in white mica (muscovite) in pegmatite,but not in metamorphicrocks (BruceVelde, personalcommunication2001). Thispointis importantto understand the relativelylow Cs contentin the ceramicstemperedwith quartz-muscoviteschist (mica temper, Table 3). These wares are nonlocal, as no metamorphicrocksareknownin the regionof Chavin. High cesium contentmay also point to the use of a micaceousclay to producethe ceramics,due to the preferentialabsorptionof Cs associatedwith micaceousclaysovergeologicaltime(RonBishop, personalcommunication1998). As a corollary,ceramicsfound in Chavinwith Cs contentbelow 11 ppm areprobablylocal. This is observedfor all theceramicsin thebasicto intermediateintrusivegroupandthesedimentary-intrusive group,and for most ceramicsin the volcanic group.Theirmineralogyalso correspondsto local resourcesaroundChavinde Huantar. This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions 354 LATINAMERICAN ANTIQUITY E The chemical tendencyshown in Table 3 and theassociatedmineralogysuggestthatCs is a componentof acidic volcanicrocks, while it is of low abundancein ceramicswith intermediateto basic intrusivepaste. Its presenceis linked to the geological environmentof the sourceareafor ceramic production.The high Cs content of the coastal ceramicsfromAnc6n andGaragay,andHuaricoto in the Callej6nde Huaylas,is probablyrelatedto the volcanicsof thoseregions.On the contrary,the Mosna-Chavinregionis characterized by intrusivegranodioritic and sedimentary geology, which explains the lower Cs content in local Chavin ceramics. k00 ,.0 ~~oo ' 00N 00-r/t N0 0 CH Cli I CM SIn Nm Sz IIt [Vol.15, No. 3, 2004] c~1\A Non-Local Ceramicsat Chavinde Huantar Cl C\ 0) ?M CC/) c 000-u C 00 I CO 0: 0I 0 u .0`~ 00 0 I I 9~ C \ ~ N 0) E Nt O cO 0) H 00 0) rn 0)).0 k0) 06 C,4C Sc,0CcO~ O Oo~ 0,0 I I'-I C C vi v E 0 -]~ CC o 00 OO 0 0) I0. ~I~~ I ; \d -:l o6 00 o Cl ed CD U ~a > 0~0 Z 0 .0 ~c#C. O. ~0~eC.)0 -- 0 3 0)0) 0+ 0. " CO ) +0 0) II C) 0 0~ >5 COCO O0 C0.SQ .00. U3 0 U Q)20 0)0 ~ Based on cesium content,mineralogy,local geology, and style, severalceramicsfound in Chavin de Huantarwereidentifiedas nonlocal.Inthecluster analysisof the INAA results,they standout as outliers or they group separatelyfrom the main Chavinceramics(Groups1 and2). Mostof thestylistically atypicalceramicsare found in Groups7, 6, and 5. High Cs contentis found in abouthalf of the stylistically atypical wares from Chavin (7/12), theirnonlocalinterpretation. therebystrengthening These include four jars (3742c, 3748b, 3764a, 3786g) and three bowls (3708A, Mosna bowl 3755a, oversized bowl 3764b). Chakinanibowl 3708A has very high Cs (56.90 ppm) and low Na and Fe values. Its paste is not tempered and extremely fine, unlike any other paste found in Chavin.An earliermicroprobeanalysisof the clay matrixof bowl 3708A showedSi/Al ratiosof 4.20 to 4.91, well above local Chavin ceramics with ratios from 1.98 to 2.00 in the illite-montmorillonite range(Druc 1997; Newman 1987). ThehighCs contentof severaladditionalsherds suggests they are not local, althoughthey are of Chavinstyle. This is the case for two jars (3744, 3745a), four circle-and-dotbowls (B21, 3775a, 3787b, 3784d), three plain bowls (B1, 3783c, 3783d), olla 3741d, andbottle3755c. Sherd3774, withhighCs content,comes froma concaveundecoratedbowl found in a cell in the ceremonialarea (Bennett1938)andis probablylateChavinin date. mineralcomposition Ithas a granitic/metamorphic with very coarse,alteredrock fragments,dissimilarto theusualbasicto intermediateintrusivecom- This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions REPORTS positionof the local Group2. Its contextsuggests thatit is an offering(Bennett1938). The other atypical bowls and bottles (3762, 3787B, 3790, 3784f, 3787a) have normalCs content but they cluster separately from the local ceramics. They have sedimentary or intrusive pastes.Theirstyle,whichis clearlynotChavin,and chemistrysuggestnonlocalproduction. Atypical olla 3781D clusters with outliers 3787A and B 19 within the volcanic local group. Althoughof uncommonshape,this warecould be from the Mosnaregion.Similarly,threevolcanictemperedceramicsform a subgroupjoining local Group1, butthey areprobablyof differentorigin. These are Janabarriu circle-stampedbowl 3787b, Chakinanijar 3744, and atypical red-slippedjar 3742c. Bowl 3787b andjar 3744 lack the typical welded tuff fragmentsof the local group,but display the same monomineralicgrains.They have a high Cs content,suggestinga nonlocalorigin.Furthermore,jar 3744 has a slip containing quartz unlike any other local slip, while the use of volcanic temperin bowl 3787b is uncommonduring Janabarriu times. Threeceramicsfrom Chavinand Pallkabottle P23 (Group3) clusterseparatelyfromthe intrusive local Group 2. These are Urabarriubowl 3786c withintrusivepaste,Urabarriu decoratedolla 3780a with volcanic temper,and volcanic-temperedjar 3783a with graphiteslip over a texturedbody.The latteris probablya Chakinaniware. The Cs contentof these ceramicsis not significantlyhigh, but the mineralogydiffersfrom the local intrusiveor volcanic groups. The varied mineralogy of this group suggests at least two distinct production areas.The two volcanic-temperedwares lack the welded tuff fragmentsof the local groupandhave pyroxenecrystalsin theclay matrix.Theycouldbe regional productionsfrom the White Cordillera. BottleP23 foundat Pallkahas a moreacidic-intrusive pastethanthe local Chavingranodioritepaste and could not have been made in Chavin. Twoothervolcanic-tempered waresdo notclusterwiththelocalvolcanicgroup.Theirhighcesium contentsuggestsa nonlocalorigin.This is the case for Janabarriu oversizedbowl 3764b and atypical jar 3786g. They arefoundin the chemicalclusters 7 and 5, suggesting an origin outside the Mosna Valley. The origin of bowl 3764b is discussed below. 355 The final clearly nonlocal group is the mica group(Group8, Table3). Thewhite mica flakesin the clay matrixcome from crushedquartz-muscovite schistfragmentsaddedto theclayby thepotter.Thismineralogicalcompositionis foreignto the Mosna Valley geology. A high Cs content was expectedforthisgroup,butit is relativelylow (6.38, 7, and 11 ppm) due to the metamorphicorigin of the temper.In this group,the elementalcompositionfortheJanabarriu bowl 3775a circle-impressed differs from that measuredfor the two Urabarriu wares,suggestinga differentorigin. Table 5 summarizes these results. Nonlocal ceramicsat Chavinde Huantarvaryin form,paste composition,and date, as will be detailedbelow. The provenancediversity for volcanic-tempered wares bears particularsignificancewhen considering the production scenario throughout the Chavinoccupationof thesite.Compositionaldiversity suggests multiplenonlocal origins. Nonlocal waresreach30 percent,andupto 40 percent,if one considersthewaresincludedin thechemicalgroups 3 through8. However,the local productionof several ceramicsin the fine wareGroup4 and mixed compositionGroup5 cannotbe ruledout. Local Productionat Chavinde Huantarand Local Geology Local geology and the abundancecriteria(Harbottle 1982; Shepard 1968) determinethe local characterof a group.Groups1 and2 (Table3) consist of ceramicswithmineralcompositionsmatching the local geology. They are termed "local groups"andrepresent22.2 and23.4 percentof the Chavinsample.Partof Group4, characterizedby a sedimentarypaste, is certainlyalso local. The threemainpetrographic groupsidentifiedin Chavin ceramics(volcanic,intrusive,andsedimentary)are compositiontypes foundwithina 20 km radiusof the site (Figure4). The local geology consists of Quaternaryalluvial and glacial deposits at the bottomof the valley, sedimentary rocks of Late Jurassic age (ChicamaFormation:dark shale and dark limestone, sandstoneand tuff), and LowerCretaceous sedimentaryrocks (Oy6n Formationand Goyallarisquisgagroup:shale, slate, coal seams, limestone quartzite,and arenite)on the valley slopes (Cobbinget al. 1996:73-74; Turneret al. 1999). Sediments with mineralsand rock fragmentsof This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions 356 LATINAMERICAN ANTIQUITY [Vol.15, No. 3, 2004] Table 5. Non-local Ceramicsat Chavin de Huantar. Atypical wares with high Cs content - N 6/81 (7.4%) 3708A outlier pellitic untemperedpaste; decoratedlight gray bowl, Chak period 3742c Gr 1 C1, red slip jar, prob.Urab period 3748b B 1, mix sed, volc, intr.;Urab red slippedjar Gr 4 3755a F mix sed, few intr,no hornblende;Mosna orange slip bowl with red paint Gr 4 3764b Gr 7 Cl many micas in the paste; Jan oversized red slipped bowl Gr 5 3786g C1 pyroclastics,very coarse grains;incised jar Chavin style non-local wares with high Cs - N 8/81 (9.9%) 3741d Gr 7b B2; black olla 3744 Gr 1 C3 coarse volc sand, no tuff, crushedquartzslip; oxidized Chakjar 3755c Gr 7 F fine-sandgrains of basic intrusiveorigin in mica paste with mixed lithics (intr,sed, volc); Chak bottle 3774 Gr 4 D; reducedChavinbowl 3783d Gr 4 B1; Jan red slip bowl 3784d Gr 4 B 1; Jan circle-and-dotblack bowl 3787b Gr 1 C3 Jan circle-and-dotblack bowl B21 Gr 4 B1 grains perpendicularto walls; Jan circle-and-dotbrownbowl Atypical wares with normal Cs content (cluster separatelyfrom the "local"wares) N 5/81 (6.1%) 3762 Gr 5 graniticmaterial,not from White Cordillera;bichromebottle chamber,zoned red with graphite,probablyJan 3784f Gr 6 Bl; incised bottle chamber,dull beige paste and very fine walls 3787a Gr 6 F; fragmentfrom squarered bottle 3787B A2; black bowl with red paint within incision Gr 6 3790 Bi ; incised bowl Gr 6 Chavin style wares, normal Cs content (cluster separatelyfrom the "local"wares) N 4/81 (4.9%) 3765a Gr 5 granitictemper,medium to coarse sand grains;Chavinbottle 3775a Gr 8 E; Jan circle-and-dotbowl, brown micaceous slip 3785 Gr 8 E; Urab?Body fragmentwith interiornet impressions 3786f E; Urab red slip jar Gr 8 Legend:Al, A2 Materialof basic to intermediateintrusiverock in the paste. B 1, B2, B3 Materialof sedimentaryorigin in the paste. Cl, C2, C3 Materialof volcanic origin (pyroclastics)with embayed quartz,plagioclase, green hornblende,brownbiotite, and welded tuff fragments(in C1). D Acid intrusiverock fragmentsand very alteredminerals,very coarse grains. E Paste temperedwith quartz-muscoviteschist fragments. F Fine paste with few inclusions of mixed origin and fine to medium sand grains. Urab = Urabarriu,Chak = Chakinani,Jan = Janabarriu,sed = sedimentary,intr = intrusive;volc = volcanic. mixed sedimentarycompositionare found at the valley bottom. The White Cordillerais composed of a granodioritic-tonaliticbatholith,intrudedinto volcanic and sedimentarysequences(Cobbinget al. 1981; EgelerandDe Booy 1956). Greenhornblendeis a chief componentof the WhiteCordilleragranodiorite(RobertStrusievicz,personalcommunication 1999) and is found in quantityin the thin sections fromtheChavinintrusivegroup.Glacierscoverthe peaks of the White Cordilleraabove 5,000 m elevation(BodenlosandEricksen1955;INGEMMET 1995;Turneret al. 1999). Sourcesof magmaticmaterialareless extensive thanintrusiverocksaroundChavin.Volcaniccompositions in the form of ignimbriticmaterialand tuff fragmentsoutcropto the south in the Mosna headwater(Cobbinget al. 1996).Daciticextrusives with welded tuffs outcrop10 km south of Chavin at PampaJunin(EgelerandDe Booy 1956), a tributaryof the Rio Mosna, 1 km above the junction with QuebradaTambillo on the road to CatacChavin. Rhyolitic tuff is also reportedas stocks northeastof Chavin,above San Marcos near the Antaminamine (Bodenlos and Ericksen 1955). Greenhornblendeis presentwithinthevolcanictuff This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions REPORTS 357 ,IWO Sa ..0 aro Z. ?Huat .14r~ .. . +rChavin . ,t* .,~)?, ?r?~2 T-11M, *- Iti~?)?)trrC,~ ~c-Z.1l 11-1 ut r?(?~~?~r),?trrrrlb r??r?????~rr -r-?r~?? ?rrtz lp VIP= Sedimentary Rocks and Deposits LowerCretaceous, GoyllarisquizgaGroup(shale, siltstone, Ssandstone, slate, quartzite,limestone) i . . LowerCretaceous, Oyon Formation(shale, sandstone, limestone, tuff) LateJurassic, ChicamaFormation(darkshale and limestone, sandstone, tuff) Quaternary,morainicdeposits, sand, gravel, silt, till Quaternaryfluvioglacial,morainicand aluvialdeposits Igneous Rocks Intrusiveneogene granodiorite A: Leucogranodiorite; B: Tonalite-granodiorite Neogene riodacite Paleogene, CalipuyGroup(volcanics) Figure 4. Geology of the Chavin de Huantar area (After Turner et al. 1999, Figure 2; Druc 1998, Figure 5; Cobbing et al. 1996, Map 20i). 1. Pampa Junin volcanics (dacitic tuff); 2. Quilloc volcanics (Rhyolite, dacite); 3. Atamina-Contonga volcanics (rhyolite); 4. Cerro Torregaga (riodacite). This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions 358 LATINAMERICAN ANTIQUITY fragmentsand in the paste of the volcanic-temperedsherds.This is peculiarbecausehornblende is normallybrown in volcanics. Ignimbritictuffs from the Black Cordillerahave no hornblende (RobertStrusievicz,personalcommunication1998 and 1999), which rules out a provenancewest of the Rio Santafor the tuff-temperedsherdsfound in Chavin. Othersourcesof volcanictuff are closer to the site. The LateJurassicChicamaFormationunderlyingtheMosnaValleyis composedof volcanictuff along with darkshale, darklimestone, and sandstone (Figure 4; Turner et al. 1999:48-49). Althoughthis formationcould have been a source for the Chavinpotters,its mixed sedimentaryand tuff compositiondoes not match the paste of the volcanicgroup.However,bouldersof volcanictuff are reportedaroundthe site and along the river banks(Turneret al. 1999:55),andtheymaybe part of a rock fall or landslidefrom the uppervalley slope to the east. Unfortunately,therehas been no petrographicanalysisof these boulders.However, it is likely thatthe local pottersused these soft tuff rocks as temperingmaterial. Theuse of volcanictuffwas notrestrictedto potteryproduction.Twodifferentkindsof porousvolcanic tuff were used for carving some of the columnsandtenonnedheadsof Chavin(Turneret al. 1999:55).It is likely thatthe potterswere aware of this stonecarving"industry"andmay havebenefitedfromit by usingtuffdebris.This impliesthat theywouldnothaveneededto go too farto get their temperingmaterial,if stone carving was a local activity,as suggestedby theamountof stonesculpturesin Chavin. Theremaywell havebeenmultipleceramicproductionlocationsorworkshopswithintheresource areasdescribedabove,which in turnmay account for compositionalvariationsrelatedto spatialdistributions.Ignimbriticor pyroclasticdepositsrepresent a good example of such compositional variability,as their mineralassemblagesvary by depthof depositflow anddistancefromthe source. This can account for the chemical heterogeneity observedwithina single cluster.However,to confirmthe locationof the procurementareas,intensive samplingandpetrographic analysisof sandand should be conducted to identifypetroclay samples facies (Miksa2000). [Vol.15, No. 3, 2004] CeramicProductionthroughSpace and Timein Chavin Elevenceramicsamplesfromtheancientsettlement to the northand from aroundthe ceremonialcenter were included in the analysis. They are dispersed through the hierarchical classification, togetherwiththose fromthe ceremonialarea,suggestingthatthechemicalcompositionof thesherds is not linked to spatialdistributionwithin the site areaas a whole.Inotherwords,no particularworkshop was producingexclusivelyfor the residential or ceremonialareaorusing a differentsourcelocation. Variationsarenoted,however,with regardto pastecompositionlinkedto a particularoccupation phase. Ceramic Production during the Urabarriu Phase. Volcanicmaterialis prevalentin Urabarriu and Chakinaniceramics(Group1, Table3). Only a few ceramicswereproducedwithintrusiveorsedsherdstemimentarymaterial.The few Janabarriu with volcanic material are pered probablynonlocal. The diversityof pastes within the volcanic group and the domesticcharacterof the wares(ollas and jars) suggest a differentproductionscenariothan for later periods.The chemical diversityand the domestic characterof the volcanic wares (ollas, bowls, a fewjars)duringtheUrabarriuphasepoint to the existenceof multiplesmall-scaleproducers aroundandaboveChavin,whilefinebowlsandbottles were producedin or broughtinto the valley. Volcanicmaterialresiststhermalstresswell (Rice 1987; Shepard 1968), which suits the domestic characterof the Urabarriuwares. These wares reflect a demand for simpler,less-diverse forms thanthose of laterperiods.This patternof multiple local productionlocationsalso is observedfor other sites and is associated with compositional diversitywithin the coarse ware inventory,while fine bowls and bottleshave a more homogeneous compositional profile (Chapdelaineet al. 1995; Druc 1998a).Burgernotesthe coarseexecutionof domesticwaresfrom the highlandsites surrounding Chavinde Huantar,whichcontrastswitha finer qualityof productionforthelocal waresin Chavin, suggesting the existence of differentworkshops supplying the hamlets and the center (Burger 1984:185).No pasteanalysisof ceramicsfromthe highlandsiteswas conducted,butthepresentanalysis and ethnographicevidence suggest the pres- This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions REPORTS ence of severalpottingcommunitiesthatmayhave producedfortheChavinpopulationorforintra-village consumption.As statedearlier,ceramicproductionduringtheUrabarriu phasemighthavebeen linkedto or benefitedfromvolcanicstone carving for the temple,indicatingthatat least some of the workshopswere locatedclose to the center. CeramicProductionduringthe Chakinaniand JanabarriuPhases. The use of volcanicresources declined,and half of the ceramicswere tempered with sedimentarymaterialor a mixedtemper(sedimentary,volcanic, and intrusivefragmentsin the paste)duringthe Chakinaniphase.Intrusivematerial from the White Cordillerawas mostly used during Janabarriutimes to produce all types of wares.Only a few ceramicshave a sedimentaryor volcanicpaste.This suggests a switch in resource areasandworkshoplocations,andtheexploitation of intrusivematerialsfrom resourceareaswest of Chavin,along the flanksof the White Cordillera. Workshopsat the time of the maximumextentof Chavinde Huantar(390-200 B.C.) were probably locatedto the west, in agreementwith the distribution of modernpottingvillages along both sides of the WhiteCordillera(Druc 1996, 2001). No temporalvariationis observedforthemixedcomposition and fine-pasteceramics (Group4). This was also seen in an earlierstudy based on a largersamplethan the one used for INAA (Druc 1998a). These ceramicswere probablyproduced in smallquantitythroughoutthe occupationof the ceremonialcenteror were broughtfrom production centersoutsidethe MosnaValley.The present chemical analysis suggests that some of these ceramicswere not producedlocally. 359 Paste and CeramicStyle jars, 1 olla) also display a varietyof pastes.They are temperedwith sedimentary(2 jars, 1 bottle, 1 bowl), welded tuff (3 jars and 1 olla), or intrusive materials(1 bottle and 1 bowl), and two are finepastebowls.Theyappearin differentchemicalclusters,suggestingdifferentorigins.Thestyleandhigh Cs contentin halfof the atypicalwaressuggestthat they were broughtto the site. Also suggested as nonlocalare some 18 Chavin-stylesherds,which clusterwith ceramicsfromsites otherthanChavin or are outliers.These are bottles, plain and decoratedbowls,two circle-dotbowls, a quartz-slipped jar, anda blackolla. They varyin mineralcomposition (volcanic, intrusive,sedimentary,and pellitic) and often presenta high Cs content. It is interestingto note the rarityof ollas and absence of coarse ware in the nonlocal ceramics. Instead,the majorityof the nonlocalceramicsare nondomesticwares,liquidor transportcontainers (jars/bottles)andbowls, withfine walls,oftendecorated,and with good surfacefinish. Their ceremonial function or offering status is not demonstrablebutcan be presumed.The compositionaldifferencesand styles suggestthe existence of differentcontextsof warecirculation.Distributionnetworksforceramicsapparentlydo not apply here.Neitherthe productionnorthe distributionof ceramicsappearsto have been controlledor organized.Rather,ceramicdistributionapparentlywas the resultof occasionalevents.Because Chavinde Huantarwas an importantreligiousandeconomic center,local and nonlocalproductsmusthave circulatedatparticularoccasionsas gifts orofferings, or havebeen acquiredduringfairs.It is also possible that some of the nonlocal wares could have beenbroughtto Chavinas containersandwerenot the exchangeproductper se. Anotherresearchvenueis thecorrelation of styleand pastecomposition.No particular pasteformulawas reservedforthemanufacture of a specificwarestyle. The circle-and-dotdecoratedwarescommonto the Janabarriuphase occur in differentclusters, displaying differentchemical and mineralcompositions,bothlocal andnonlocal,rulingoutcentralized production.Thestampedwareswereproducedwith differentmaterials: volcanic,intrusive,mixedlithologies, sedimentary,and even metamorphic.Of the eight stampedceramicstested,five arenonlocal. The 12 stylisticallyatypicalwares in the sample fromChavinde Huantar(3 bottles,3 bowls, 5 InterregionalRelationships RelationshipswithPallkain the CasmaValleyand with the Callej6nde Huaylasare suggestedby the analysis.SeveralceramicsfromChavinoccuralong withceramicsfromPallkain compositionalGroup 5. The Pallka sherdshave an intrusivetemperof graniticcompositiondifferentfromthe granodiorite/tonalitetype found in Chavin.It is difficultto ascertainthe origin of this mixed Pallka-Chavin group.Theatypicalstyle of severalceramicsin this groupandtheirsubclusteringwithinGroup5 suggest a nonlocalorigin for bottle 3762, jars 3764a This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions 360 LATINAMERICAN ANTIQUITY and3786g, andbowl B20. Bottle3762 is decorated with graphiteoverred,a style calledWacheqsaby Lumbreras(1977, 1993) and supposedlyfromthe Cupisnique region on the north coast or correspondinghighlands.However,the use of graphite decorationis alsoknownfromtheKotosh-Huanuco region,southeastof Chavin(Burger1984;Lumbreras 1993). Bottle 3765a was originally attributedto the coast by discriminantanalysisusing x-rayfluorescence data (Druc 1998a). It is now groupedwith Pallkasherds,regardlessof the clusteringmethod, even whensamplesfromthecoastalsites of Anc6n andGaragayareincluded.The CasmaValleyis the probableprovenancesite of this bottle.This placementis morein agreementwithbottlestyle at both sites and, above all, with petrographicanalysis (unavailablefor this sampleat the time of the XRF analysis).PallkabottleP23 withfineintrusivepaste clustersin Group3 withChavinoutliers,separately from the otherPallkasherds.Its origin, however, is not Chavinde Huantar,due to its mineralcomposition thatdoes not matchlocal Chavinmineralogy. Along withthe sherdsfromtheceremonialcenter of Huaricotoin the Callej6nde Huaylasis the oversizedbowl 3764b.Itis a largered-slippedbowl materialand temperedwithpyroclastic-ignimbritic weldedtufffragmentsat a time (Janabarriu phase) when thistype of temperwas no longerused in the region of Chavinde Huantar.The presenceof an andesite fragmentin the paste suggests that the materialcame froma volcanicrockof the Tertiary Calipuy Formation.This rock type is not found nearChavin.Themineralcompositionof thisbowl is consistentwith sourcesof volcanictuffandignimbritecommonin theCallej6nde Huaylas(Robert Strusievicz,personalcommunication1999; Cobbing et al. 1996). The Callej6n is easily reached througha mountainpass just above Chavin.The provenanceof this bowl, however,is not the small ceremonialcenterof Huaricoto,judging from the lack of mineralsimilaritywith the ceramicsfrom thatsite. SeveralotherceramicsfromChavinclusterseparately from the Chavin local groups (#3744, 3742c, 3787b, 3755a, and B21) or from samples fromthe coast. Manyof these come fromatypical ceramics,whose style ormineralcompositionsuggests a nonlocal provenance,as yet unidentified. [Vol.15, No. 3, 2004] These sherds have been discussed in relationto theirhighCs content.TheMosnaRed-over-Orange oripaintedsherdin Group4 is also of unattributed gin. Its style suggestsa northernprovenance,from the regionof Pacopampa(Lumbreras1993). Last, the finesand-pasteChakinanibottle3755c andsedimentarypaste olla 3741d are linked to samples fromthe coastalsites of Anc6nandGaragay.Their mineralogy,however,is differentfrom the coastal paste found in Anc6n and Garagay(Druc et al. 2001) and rules out a centralcoast provenance. Conclusions Priorchemical analyses hinted at the wide compositionaldiversityof the ceramicsin Chavin de Huantar.FurtherpetrographicanalysisandINAA studiesallow this diversityto be linkedto production scenarioschangingover time, and to a considerable amount of nonlocal wares. Several observationson ceramicproductioncan be made at the local and interregionallevels. Accordingto chemical and mineralanalyses,severalcentersof productionwere supplyingthe ancientsettlement and ceremonialcenter of Chavin de Huantar.At least threelargeproductionareascan be proposed: two in thelocal highlandssouthandwest of Chavin and one in the valley bottom. Severalworkshops were probablyactivein each area,as suggestedby the internal heterogeneity of the compositional groups.Also, manyproductioncentersmust have been in use at differenttime periods.The use of material,oftenwithwelded ignimbritic-pyroclastic tuff fragments,is observedduringthe Urabarriu phase, the first occupation phase at Chavin de Huantar.The workshopsusing this materialwere probablylocatedupto 12kmfromtheresourcearea (estimateddistance based on ethnographicdata, Druc 1996), south and southeastof Chavin, and northeastof the site abovethe village of San Marcos (Figure5). Some of the workshopsalso could havebeen locatednearthe site, using tuffboulders scatteredalongtheMosnariverbanks,ortuffdebris fromvolcanicstone-carvingactivity.Laterin time, duringthe maximumextensionof the site, the tendency was to use intrusiveraw materials,and the workshopsmusthavebeen locatedwest of the settlement,usingmaterialsfromtheslopeof theWhite Cordillera. Thethirdproductionareais characterized by the This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions 361 REPORTS ~ ai .. -CI*'Mallas <• - ,C r ile ......... .i • . a .................. ........ ............. ...:.. havi R..-:i .... S om. // : :..... ..... ...... .. n.W p N "San arcos Bla c .. . . .r ...... a l l i ~i~ lji l j i :ii: i :iiii ; } Ii•ii iii:ii ............... ii~iiiiii!;:ii l?•:I!•: lii: ..... . a;:i B lanca .. .....j .......::A: .......ii..:i. ..!! !. 3:40 : 2ooo. . . . . . .l . de!.!..2..!. !i . . . . .'•~~~~l:: . .. . . . .. . .. . .. . H ohvn : : l..... allejonde . .... . ....J . 1 3000I 4000 m 30004000.......50.0. . Pocenters s ible of I6 Hu ..... ..... ...... . . . - I! ceramic I :- iI: 1Ij 1 4 : 7 production 5000 - l 6768 las~ . .......... tio ofeiprodu ll--:il~l~ 5 ~jii~ km:ll ...... I :I I I 1Ir I 1I i i iI m . i iI1 Ii I: :i I I;U Ii i iI i i Figure 5. The Chavin region and possible ceramic production areas suggested by the analysis. use of sedimentarymaterialandthe productionof fine paste ceramics. However, productionmust have been low andthe workshopsfew andlocated nearthe site of Chavin,in the valley floor or just aboveit. The workshopsin this areawere engaged in small-scaleproductionof bottlesanddecorated bowls throughoutthe existence of the site. The workshopslocatedhigherup in the valley on the slope of the Cordillerararelyproducedbottles. DuringUrabarriutimes, when volcanicmaterial was predominantlyused, the most common wares were ollas and bowls. As the site grew and became moreimportant,local ceramicproduction becamemorediversifiedandintrusivematerialwas used to producea widerrangeof ceramics,including bottlesandjars. The shift in procurementarea andworkshoplocationscannotbe explainedbased on pasteanalysisalone.Resourceexhaustion,landslidesthatburiedsources,ortechnologicalchanges are possibilities.However,among more probable factors were the economic expansionof the site, population growth, higher demand for ceramic products, and the establishmentof new potters, who perhapswere from the Callej6nde Huaylas and more familiarwith the intrusivematerials. This production pattern corresponds to the observeddevelopmentof Chavinde Huantar.During the Urabarriuphase,the settlementwas small This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions 362 [Vol.15, No. 3, 2004] LATINAMERICAN ANTIQUITY and located immediatelynorthof the ceremonial area.Chavingrew duringthe Chakinaniphase to an estimated 15 ha and to 42 ha duringthe Janabarriuphase,with the settlementextendingsouth of the ceremonialcenter(Burger1984, 1992).The ceremonialcenteralonecoveredsome 1.5ha in the early periodsto 5 ha at its maximumextent.Severalvillages abovethe valleyfloorandin the highlands surroundedChavin.The highlandvillages appearto have been occupied mainly duringthe Urabarriu phase,ata timewhenvolcanicresources were used for ceramicproduction.Archaeological evidence shows a dispersedpopulationduringthe Urabarriuand Chakinaniphases,and a moreconcentratedand dense occupationduringJanabarriu times, reflectedin the ceramicproductionpattern. During the Urabarriuphase, the diversityof volcanicpastessuggestsdispersedceramicproviders, probably responding to low demand. A more diverseandintenseproductionis suggestedduring the Janabarriuphase, when the valley was more heavily populated.The absenceof atypicalwares in thevolcanicandintrusivegroupsis in accordance with the hypothesisof local highlandproduction for utilitarianware. The region providingthe Chavin de Huantar populationwithutilitarianwaremusthaveextended well beyondthe site. Untilthe early 1970s,potters fromYacyaandMallas(Figure5), 30 km northin the highlands above the lower Mosna Valley, walkedto Chavinto bartertheirpots (Druc 1996; RichardBurger,personalcommunication1998). These conclusionssupportBurger'shypothesisof the Chavin Temple drawing services and goods from communitiesup to the lower Mosna Valley and southern Callej6n de Huaylas (Burger 1984:249).In additionto the religiousimportance of the ceremonialcenter,Chavinmusthavebeen a centralplace for fairs and exchange of local and interregionalproducts,including ceramics. Economic and culturalinteractionsexplain the wide diversity of pastes and variety of origins of the ceramicsin Chavin. At the level of interregionalinteractions,the importanceof Chavinde Huantaris demonstrated by thenumberof atypicalandnonlocalwaresfrom the Urabarriuphaseon. None of the sites analyzed previously (Pallka, Huaricoto,Anc6n, Garagay [Druc 1998a;Druc et al. 2001]) show this diversity of pasteandnumbersof atypicalandnonlocal wares,attestingto the interregionalscope of interactionswith the ceremonialcenter.Most nonlocal wares are bottles and fine bowls thatwere probably not destinedfor domesticuse in view of their style, decoration,and rarity.One bottle may be fromthe ceremonialcenterof Pallkain the coastal valleyof Casma,whilea largebowlprobablycomes from the Callej6nde Huaylas,on the otherside of theWhiteCordillera.Abouta thirdof the nonlocal waresarenon-Chavinin style andmay come from cultural areas outside Chavin influence whose inhabitantswere awareof the religious centerof Chavinde Huantar. This INAA study yields a clearer picture of ceramicproductionand interactionsin Chavinde Huantar.However,none of the conclusions presentedherecould havebeen reachedsolely on the basis of neutronactivation.Petrographicandgeological datawereveryimportantin interpretingthe chemicalresults,along with stylistic information. By combining these different sets of data more informationcan be extractedfrompaste analysis. Acknowledgments.I would like to express my thanksto Ron Bishop and Jim Blackman, who conducted the chemical analysis at the Smithsonianlaboratory,which is maintained at the National Instituteof Standardsand Technology,and to Ron Bishop for his advice on statisticaltreatmentof the data. This paperalso benefitedfrom the comments of clay mineralogist Bruce Velde. The interpretationof the data and conclusions are, however, my sole responsibility. I thank the Social Sciences and Humanities Research Council of Canada for a postdoctoral grant (fellowship NB 756-970120), the Smithsonian Center for Material Research and Education for allowing me to conduct this study, and the Museum of NaturalHistory in New York for lending many ceramics from the Bennett collection. References Cited Aldenderfer,MarkS. 1982 Methodsof ClusterValidationforArchaeology.World Archaeology14: 61-72. Baxter,Mike J. 1994 ExploratoryMultivariateAnalysis in Archaeology. EdinburghUniversityPress,Edinburgh. Bennett,WendellC. 1938 1938 Expedition,Callej6nde Huaylas,Peru.Unpublished notes.AmericanMuseumof NaturalHistory,New York. 1944 TheNorthernHighlandsof Peru:Excavationsin the Callejdnde Huaylasand at Chavinde Huantar.AnthropologicalPapersof theAmericanMuseumof NaturalHistoryVol. 39, Pt. 1. AmericanMuseumof NaturalHistory, New York. Bishop, RonaldL., and PatriciaL Crown 1994 ExperimentalProceduresandParametersUsed in the This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions REPORTS NeutronActivationAnalysisof theGilaPolyInstrumental chromeCeramics.InCeramicsandIdeology.SaladoPolychromePottery,editedby PatriciaL. Crown,pp.227-231. Universityof New Mexico Press,Albuquerque. Bishop, Ronald.L., RobertL. Rands,andGeorgeR. Holley 1982 CeramicCompositionalAnalysis in Archaeological Perspective.AdvancesinArchaeologicalMethodand Theory 5:275-330. Bodenlos,AlfredJ., and GeorgeE. Ericksen 1955 Lead-ZincDeposits of CordilleraBlanca and Northern CordilleraHuayhuash,Peru. United States Government PrintingOffice,Washington,D.C. Burger,Richard.L. 1984 The PrehistoricOccupationof Chavinde Huantar, Peru.Universityof CaliforniaPublicationsin Anthropology 14. Universityof CaliforniaPress,Berkeley. 1988 Unity andHeterogeneityWithintheChavinHorizon. In PeruvianPrehistory,edited by RichardW. Keatinge, pp. 99-144. CambridgeUniversityPress,Cambridge. 1992 Chavin and the Origins of Andean Civilization. Thamesand Hudson,London. Chapdelaine,Claude,SantiagoUceda, and Greg Kennedy 1995 Activaci6nneutronicaen el estudiode la producci6n local de la cerimica ritualen el sitio Moche, Peru.Bulletinde l'InstitutFrangaisd'EtudesAndines24:183-212. Cobbing,John,W. S. Pitcher,J. J. Wilson,J. W. Baldock,W. P. Taylor,W. McCourt,andN. J. Snelling 1981 The Geology of the WesternCordilleraof Northern Peru. Overseas Memoir 5. Instituteof Geological Sciences, London. Cobbing,John,Agapito SanchezE, WilliamMartinezV., and HectorZ6arate O. 1996 Geologiade los cuadrangulosde Huaraz,Recuay,La Union, Chiquiany Yanahuanca.Boletin no 76. Instituto Geol6gico Mineroy Metalurgico,Lima. Druc,IsabelleC. 1996 De la ethnograffahacia la arqueologia:Aportes de entrevistascon ceramistasde Ancash (Perni)parala caracterizaci6nde la cerimicaprehispinica.Bulletinde l'Institutfrangaisd'e'tudesandines25(1):17-41. 1997 Caractdrisationet analyse de provenancede la productioncdramiquedans laire d'influenceChavin(Andes centre-nord):Un aspect des relations interrigionalesac I'Horizonancien. Dissertationthesis. Departementd'anthropologie,Universit6de Montreal,Montr6al. 1998a CeramicProductionand Distributionin the Chavin Sphereof influence.BritishArchaeologicalReports,InternationalSeries 731. HadrianBooks, Oxford. 1998b Producci6ny distribuci6nde la cerimica en Chavin de Huaintar,Apendice I. In Cronologiay economia de Chavin de Hudntar, edited by Richard L. Burger, pp. 303-314. PontificiaUniversidadCat61lica del Perni,Lima. 2001 Soil SourcesforCeramicProductionin theAndes.In Archaeologyand Clays, edited by Isabelle C. Druc, pp. 95-105. British ArchaeologicalReportsS942. Hadrian Books, Oxford. Druc, Isabelle C., RichardL. Burger,Regina Zamojska,and PierreMagny. 2001 Anc6n andGaragayCeramicProductionat the Time of Chavinde Huantar.JournalofArchaeologicalScience 28:29-43. Egeler,CorneliusG., andThijs De Booy 1956 Geology and Petrology of Part of the Southern CordilleraBlanca,Peru.GeologySerie 19:1-86. Foucault,Alain, andJean-FrangoisRaoult 1995 Dictionnairede gdologie. Masson,Paris. Glascock,MichaelD. 363 1992 Characterizationof Archaeological Ceramics at MURRby NeutronActivationAnalysis andMultivariate Statistics.InChemicalCharacterization of CeramicPastes in Archaeology,editedby HectorNeff, pp.11-26. Monographs in WorldArchaeology No 7. PrehistoricPress, Madison,Wisconsin. Harbottle,Garman in Archaeology.In Con1982 ChemicalCharacterization textsforPrehistoricExchange,editedby J. E. Ericsonand TimothyK, Earle,pp. 13-51. AcademicPress,New York. INGEMMET(InstitutoGeol6gico Mineroy Metaldrgico) 1995 Mapa geoldgico del cuadrangulo de Huari, 19-i. INGEMMET,Lima. Keatinge,RichardW. 1981 The Natureand Role of Religious Diffusion in the EarlyStagesof StateFormation:An ExamplefromPeruvian Prehistory.In TheTransitionto Statehoodin theNew World,editedby G.D. JonesandR. R. Kautz,pp 172-187. CambridgeUniversityPress,Cambridge, Lumbreras,Luis G. 1974 ThePeople and Cultureof AncientPeru. Smithsonian InstitutionPress,Washington,D.C. 1977 Excavacionesen el temploantiguode Chavin(sector R): Informede la sextacampafia.NawpaPacha 15: 1-42. 1993 Chavinde Huantar Excavacionesen la Galeria de las Ofrendas.DeutschenArchaeologischeInstituts,Band 51. Philippvon Zabern,Mainz. Lumbreras,Luis G., R. Gehard,W. Hiiusler,FedericoKauffman-Doig,J, Riedner,G. Sieben, and UrsulaWagner 2003 MossbauerStudyof CeramicFindsfromthe Galeria de las Ofrendas,Chavinde Huantar.HyperlineInteractions 150:51-72. Miksa,ElizabethJ. 2000 Correspondenceand DiscriminantAnalyses of Sand and Sand TemperCompositions,Tonto Basin, Arizona. 42:273-299. Archaeometr, Newman,A.C. D. (editor) 1987 Chemistryof Clays and ClayMinerals.Mineralogical Society,Monograph6. LongmanScientific& Technical, New York. Perlman,I., andFrankAsaro 1969 PotteryAnalysisby NeutronActivation.Archaeometry 11:21-52. Rice, PrudenceM. A Sourcebook.Universityof Chicago 1987 PotteryAnalysis: Press,Chicago. Salazar,R., UrselWagner,FritzE. Wagner,W. Korschinsky,M. Zahn,Josef Riederer,andFedericoKauffmann-Doig 1986 Missbauer Spectroscopy and Neutron Activation Analysisof RecentFindsfromChavin.In Proceedingsof the 1984 International Symposiumon Archaeometry, edited by J. S. Olin and M. J. Blackmann,pp. 143-151. SmithsonianInstitutionPress,WashingtonD.C. Shennan,Stephen 1988 Quantifyingarchaeology.EdinburghUniversityPress, GreatBritain. Shepard.AnnaO. 1968 Ceramicsfor theArchaeologist.CarnegieInstitution of Washington,Washington,D.C. Turner,RobertJ. W., RosemaryJ., Knight,andJohnRick 1999 GeologicalLandscapeof the Pre-IncaArcheological Site at Chavinde Huantar,Peru.CurrentResearch1999D, Geological Surveyof Canada:47-56. ReceivedMarch29, 2002; AcceptedOctober21, 2002; RevisedAugust13, 2003. This content downloaded from 129.81.226.78 on Sat, 18 Oct 2014 15:06:01 PM All use subject to JSTOR Terms and Conditions