alden

Abstract

To examine regional patterns of ceramic production and distribution during the era of Inka domination in northern Chile, we determined theelemental compositions of 157 samples of archaeological ceramics and geological clays from the sites of Catarpe and Turi using instrumentalneutron activation analysis. We identified two major and three minor composition groups in the ceramics. The major groups, High Cr andLow Cr, are linked to clays from two broad geological contexts within the region, while the minor Low Na group is made up of ceramics importedfrom northwestern Argentina. The distribution of the composition groups indicates that, in the CatarpeeTuri region, patterns of ceramic produc-tion differed for different vessel types: jars were made from clay and temper acquired near the sites where the jars were used, while bowls weremade of material coming from more distant sources. The geographical distribution of the analyzed ceramics indicates that bowls were exchangedbetween Catarpe and Turi in a pattern more similar to tribute/extraction than to market exchange, with Catarpe being the dominant site. The com-positional analysis also demonstrates that Inka-style ceramics were being locally produced at sites in this region during the era of Inka domination. 2005 Elsevier Ltd. All rights reserved.

Keywords: Inca; Inka; Ceramics; Chile; Neutron activation

1. Archaeological background

In northern Chile, the Late Intermediate Period (SolorPhase, A.D. 900e1450) was an era of locally independent so-ciopolitical units (senoros) centered on oases scattered acrossthe arid Atacama Desert landscape [21]. According to ethno-historic sources, this region was incorporated into the Inka Em-pire during the reign of Topa Inka Yupanki. The advent of theInkas, whose presence defines the Late Horizon Tardo Phase(A.D. 1450e1536), was marked by the establishment of newsettlements incorporating structures built in Inka architecturalstyles, the construction of Inka style structures in previously

and ceramics, and the construction of a system of routes andpaths collectively called the Inka Road. These Inka featuresdid not, however, replace the existing Atacameno settlements,structures, ceramics, and trailsdthey only supplemented thecorpus of material culture already present in the region [1].

The two most important Inka administrative centers in thisregion were at Catarpe and Turi (Fig. 1). At Pukara de Turi,near the Rio Salado that drains into the upper Rio Loa, a smallsettlement was established around A.D. 900. The site, however,did not expand to its maximum size of about four hectares un-til around A.D. 1300, and the earliest evidence of an Inka pres-ence at Turi dates to around the end of the 14th century [2]. InIdentifying the sources of Inorthern Chile: results of a

John R. Alden a,*, Leah Ma University of Michigan Museum of Anthropology

b Oregon State University Radiationc Brazos Valley Museum of Natural History, Bryan, TX 77802

Received 28 June 2005; received in revised form 2

Journal of Archaeological Scienceoccupied sites, the introduction of new types of metal objects

* Corresponding author. Tel.: 1 734 663 8121.E-mail address: [email protected] (J.R. Alden).

0305-4403/$ - see front matter 2005 Elsevier Ltd. All rights reserved.doi:10.1016/j.jas.2005.09.015nka period ceramics fromneutron activation study

inc b, Thomas F. Lynch c

, 1215 Lutz Avenue, Ann Arbor, MI 48109, USACenter, Corvallis, OR 97331, USA

and Texas A&M University, College Station, TX 77843, USA

7 September 2005; accepted 29 September 2005

33 (2006) 575e594http://www.elsevier.com/locate/jascontrast, Catarpe Tambo, an Inka administrative center on theRio Grande de San Pedro about 10 km north of modern SanPedro de Atacama, appears to have been established by theInkas or their local surrogates when the region was added toInka territory somewhere around A.D. 1450 [19].

Turi is only about 80 km from Catarpe, and the Solor andTardo Phase ceramics from these sites are very similar. Ar-chaeologists use a single set of names to describe the potteryfrom the two sites, and in most cases it is not possible to dis-tinguish pottery from the Turi and Catarpe regions using visualexamination alone. However, because the two sites lie in dif-ferent geological regions and different drainage basins, itseemed likely that ceramics made near Catarpe and San Pedrode Atacama would have different elemental compositionsfrom similar-looking ceramics made near Turi.

neighboring sites in the San Pedro de Atacama oasis to seewhether the pottery from these settlements could be differen-tiated into distinctive compositional groups and identified withdifferent sources of clay and temper. If we were able to definesuch groups, we intended to use those data to determine pat-terns of ceramic production and exchange in this region duringthe period of Inka domination using compositional character-istics of the archaeological material rather than stylistic fea-tures of the ceramics or assumptions derived fromethnohistoric records pertaining to other areas of the Inka Em-

0 50kilometers

Rio

Sal

ado

Rio V

ilama

Vilama

Catarpe

Above 5000 meters

4000 to 5000 meters

3000 to 4000 meters

Below 3000 meters

Salar (Salt Flat) and Saline Lake

contour interval 500 meters

San Pedrode Atacama

Calama

Salar

de

Atacama

Archaeological Site

Modern Town

International Border

Rio

Gra

nde

Rio Loa

Fig. 1. Map of Calama region, showing locations discussed in text.Rio Salado

576 J.R. Alden et al. / Journal of ArchaeoWe initiated a neutron activation study of Tardo Period ce-ramics and geological clays from Turi, Catarpe, and severalToconceTuri

logical Science 33 (2006) 575e594pire. Such data would help us to elucidate directly the formand scale of economic reorganization that occurred in this

region following the advent of the Inkas, and ultimately allowus to compare the process of Inka colonization and exploita-tion in this region with the patterns observed in other areasthat were incorporated into the expanding Inka Empire.

2. Neutron activation analysis

Neutron activation analysis is a technique for determiningthe elemental composition of rock, clay, and other multi-ele-ment materials. A sample of material is bombarded with neu-trons from a nuclear reactor, transforming a small fraction ofthe atoms in the sample into radioactive isotopes. As these iso-topes decay they emit gamma rays with discrete energies thatare uniquely characteristic of the elements undergoing radio-active decay [13,14]. By measuring the energies and intensi-ties of the gamma radiation emitted by an irradiated sampleit is possible to determine the concentrations of 25e35 differ-ent elements in the sample being examined.

Over the past 30 years, archeologists have used instrumentalneutron activation analysis (INAA) to study ancient ceramicsand address questions concerning the organization of ceramicproduction and patterns of local, regional, and long-distanceexchange [5,7,11,15]. Based on differences in trace elementconcentrations, they have identified groups of vessels thatwere made using clay and/or temper from a single geologicalsource, and distinguished sherds of similar appearance thatwere produced from raw materials derived from different sour-ces. In some cases the actual sources of rawmaterials have beenidentified, but even in situations where the sources of the pot-terys component clay and temper remain unknown, composi-tional analyses have demonstrated that particular groups ofpottery were manufactured from material derived from differ-ent geological sources. Such information can be used to recon-struct patterns of ceramic production and exchange both withinlocal areas and between more distant centers of population.

3. Sample selection and analytical procedures

For this study, a total of 157 ceramic and clay samples weresubmitted for INAA at the University of Michigans Ford Nu-clear Reactor. The archaeological materials tested included 45samples from Catarpe Tambo, ten from Beter 3, five from So-lor 13, and five from Vilama 2dall Solor Phase and Tardo Pe-riod occupations in the San Pedro de Atacama oasisdas wellas 66 ceramics from Turi and three sherds collected from sitesalong the Inka Road. Twenty samples of clays from aroundSan Pedro, as well as two clays and one ethnographic ceramicsample from Toconce in the Turi region were tested in hopesof identifying sources for the material used to make the ar-chaeological ceramics.

Ceramic samples were chosen that (1) came from secure ar-chaeological contexts, (2) were identifiable by type and form,and (3) might be relevant to patterns of chronological, spatial,or typological variation. Samples were prepared by grindingaway all surfaces of the sherd using a Dremel high-speed drill

J.R. Alden et al. / Journal of Archaeoand a sintered carbide abrasive bit, removing any slip, paint, orsurface deposit that might contaminate the ceramic paste andtemper. A piece of the de-surfaced sherd one to two squarecm in size was broken off and soaked in distilled water for24 h to dissolve any soluble salts impregnating the ceramic.The water was changed and the sample soaked another day,then the process was repeated for a third day. The samplewas then air dried and ground by hand into powder using anagate mortar and pestle. The washed and powdered samplewas dried for 48 h in a desiccating oven.

Samples of geological clays were either washed or un-washed. The washed samples were mixed with distilled waterand allowed to settle for 24 h. The water was then decantedand the washing repeated. After the second washing thesample was air-dried and the clay fraction separated fromany non-target coarse material. Mixtures of clay and temperwere prepared from unwashed samples. As with the samplesfrom sherds, the samples of clay or clay and temper wereground into powder and dried.

Approximately 200 mg of powdered, desiccated materialfrom each sample was encapsulated in high-purity quartz glassand irradiated for 20 h in a core-face location in the reactor,subject to an average thermal neutron flux of 4.2 1012 ncm2 s1. Geological reference standards were irradiatedwith each batch of samples to calibrate the readings fromthe irradiated materials. Following irradiation, two separatecounts of gamma activity were acquired using a 30e40% ef-ficiency HPGe detector: a 5000-s count (live time) of eachsample after a 1-week decay period, and a 10,000-s count(live time) after a period of 5 weeks decay. The resulting gam-ma spectra permitted the quantification of 25 elements with in-termediate and long half-life isotopes (As, Ba, La, Lu, K, Na,Sm, U, Yb, Ce, Co, Cr, Cs, Eu, Fe, Hf, Nd, Rb, Sc, Sr, Ta, Tb,Th, Zn, Zr). Element concentrations were determined throughcomparison with three replicates of the standard reference ma-terial NIST1633A (coal fly ash). All data reductions werebased on current consensus element libraries utilized by theMissouri University Research Reactor for archaeological ma-terials [14]. Samples of NIST278 (obsidian rock), NIST688(basalt rock), and New Ohio Red Clay were included as checkstandards.

4. Statistical analyses: procedure

We used a series of statistical procedures to identify groupsof samples with similar elemental composition, where eachgroup is distinctly different from other such groups, andwith each group presumably representing a distinct source ofclay and temper. The ability to define such groups dependson the geochemical distinctiveness of the raw material sourcesbeing utilized [5], and it can be difficult to identify specificsources of clay or temper if those materials are derived fromextensive beds of geological material exhibiting a single com-positional signature. The analytical process followed nowstandard procedures for group identification and verification[14], including (1) preliminary identification of compositionalgroups, (2) group refinement to create statistically homoge-

577logical Science 33 (2006) 575e594neous core groups, and (3) classification of non-core membersinto their most likely compositional group.

Preliminary group identification utilized a combination ofbivariate and multivariate statistical techniques to distinguishpossible groups within the data set and to identify outliers.Principal components analysis (PCA) was employed to sum-marize the primary dimensions of variability within the ele-mental concentration matrix, and samples that appeared tohold together across a series of bivariate plots and on PCscores were identified as potential compositional groups.

The second step was to test and refine the preliminarygroups, using multivariate statistical criteria to assess the ho-mogeneity of each group. For each case, the probability ofgroup membership was calculated from the Mahalanobis D2

statistic, a measure of the multivariate distance between thatcase and a group centroid relative to the dispersion of othergroup members. Group refinement and the calculation of theD2 statistic proceeds iteratively, and at each pass through thedata, cases with low probability of group membership andcases showing affiliations with more than one group are re-moved until one or several distinct core groups emerge.

The final step in forming groups involved assigning non-core samples to the composition groups they are most likelyto belong with. Typically, final classification utilizes multipleapproaches, including discriminant function (canonical vari-ates) analysis, the probabilities of group membership calculatedfrom the Mahalanobis D2 statistic, and the position of a caserelative to the 95% confidence interval ellipses for groupmembership defined on the principal components. In thisstudy, a case was classed as mixed if these criteria revealedprobabilities of greater than 5% for affiliation with more thanone group; a case was classed as an outlier (unclassed) if ithad low probabilities of membership in any group or if it felloutside the confidence intervals for group membership as plot-ted in canonical variate space.

5. Statistical analyses: results

5.1. Principal components analysis

The dimensional structure of the elemental concentrationswas first assessed and reduced using PCA, a mathematical pro-cedure that transforms a number of (possibly) correlated vari-ables into a smaller number of uncorrelated variables calledprincipal components. Because the goal of PCA is to reducethe number of dimensions in the data set, only the most signif-icant PCs are retained. Typically, this means keeping onlythose PCs whose eigenvalues are greater than 1, since compo-nents with eigenvalues of less than 1 account for less variancethan did the original variable [18].

In this data set, the PCAwas based on 21 elements that aremeasured more precisely by neutron activation (As, Ba, La,Lu, Na, Sm, U, Yb, Ce, Co, Cr, Cs, Eu, Fe, Hf, Nd, Rb, Sc,Ta, Tb, Th). After apparent outliers in univariate and bivariatespace were removed, we were left with a sample of 130 cases.Using the Kaiser criteria [18], we identified the first four prin-cipal components, jointly accounting for over 68% of the vari-

578 J.R. Alden et al. / Journal of Archaeoance in the compositional data, as the most significant (Table 1).The first component primarily represents variability among therare earth elements, while the second PC represents variation inthe first series transition metals, including iron, scandium, andchromium. The third component represents the incompatibleelements, including thorium, tantalum, and uranium, as wellas the alkali metals cesium and rubidium; the fourth representsfurther variation in the heavy rare-earth elements (lutetium andytterbium) and sodium.

5.2. Core group definition

Only samples from local style sherds were examined indefining reference composition groups for the Catarpe/Turi re-gion. Obvious imports and copies of imperial-style Inka ce-ramics were removed from the initial analyses, as weresamples of clay and temper and samples identified as outliersin the initial PCA analysis, leaving 109 samples to be used indefining core groups. Preliminary bivariate plots of elementsand principal components suggested two major compositiongroups, both consisting of local style sherds and thus pre-sumably of local manufacture. These groups are most clearlyseparated by differences in transition metal content, as illus-trated by the distribution of samples along the second principalcomponent (Fig. 2). (Composition groups are defined usingmultivariate statistics, and because plots like Fig. 2 are two-dimensional representations of a multidimensional data space,the 95% confidence ellipses sometimes appear to exclude sam-ples that are actually members of a particular group.) Since bi-variate plots show the clearest separation between the groups

Table 1

Principal components analysis of Catarpe/Turi ceramics

1 2 3 4

Principal components

Eigenvalue 6.01 4.51 2.26 1.61

Percent of variance explained 28.64 21.49 10.75 7.66

Cumulative variance explained 28.63 50.13 60.88 68.54

Total structure coefficients

Sm 0.9072 0.1209 0.1263 0.2011La 0.8220 0.0642 0.2943 0.0414Ce 0.8184 0.1259 0.3114 0.0721Eu 0.8034 0.4376 0.1454 0.0383Nd 0.7733 0.1389 0.0628 0.0903Yb 0.6564 0.0733 0.0999 0.5568Lu 0.5631 0.0884 0.0272 0.6735Tb 0.5104 0.1216 0.1994 0.0544Fe 0.0503 0.9503 0.0161 0.0419Cr 0.1315 0.9027 0.1360 0.1595Sc 0.2075 0.8944 0.0052 0.2191Co 0.2088 0.8048 0.1830 0.1105Hf 0.1423 0.6267 0.2405 0.3990Th 0.2337 0.0435 0.8680 0.0521Ta 0.2196 0.1372 0.7668 0.0675U 0.2297 0.2461 0.6555 0.3721Cs 0.0099 0.0411 0.6815 0.0037Rb 0.1791 0.4295 0.5522 0.4859As 0.2792 0.0956 0.4797 0.0292Na 0.1777 0.2280 0.3801 0.5823

Ba 0.1549 0.2065 0.3270 0.3509

logical Science 33 (2006) 575e594is evidenced by differences in chromium content, we designatethe two groups as High Cr and Low Cr. The High Cr

group (n 33) has higher values of transition metals overalland higher Cr:Th ratios, in contrast to the more abundantLow Cr group (n 52).

Subsequent verification and refinement of the two composi-tion groups utilized the Mahalanobis D2 statistic based onscores for the first four PCs. As noted by Glascock [14], cal-culation of Mahalanobis distances requires that group sizesexceed the number of elements; ideally, group size will be sev-eral times that of the number of elements. One way to assuremeeting this desired ratio is to base Mahalanobis distance cal-culations on PC scores, which encapsulate the most significantdimensions of variability within the data set. The probabilitiesof group membership calculated in this manner confirmed thepreliminary groups and identified samples with a strong prob-ability of group membership based on iterated jack-knifed dis-tance calculations.

In addition to the two main groups, several minor groupsshowing extreme values on one or more elements were identi-fied from the sample of local-style sherds (Fig. 3). These ex-treme-value groups may represent samples of local-stylepottery that are non-local in origin, or they may reflect prehis-toric exploitation of unusual clay sources or atypical concentra-tions of tempering agents. Minor groups include an Extreme Crgroup (n 3), with chromium concentrations of more than200 ppm, and a High Co group (n 6), of samples with elevatedconcentrations of cobalt. The small size of these groups pre-cluded rigorous testing of their internal homogeneity.

5.3. Linking composition groups to local clay sources

Fig. 2. Separation of local ceramic composition groups based on first two prin-

cipal components. Solid symbols show Core Members of each group, open

symbols show Non-Core Members; ellipses show 95% confidence intervals

for group membership.

J.R. Alden et al. / Journal of ArchaeoConfirmation that ceramics from both the High Cr and LowCr groups could have been locally manufactured was gainedby linking those two ceramic composition groups to samplesof local clays and mineral tempering agents (Table 2 andFig. 4). For each sample of clay, temper, or mixture of clayand temper, the probability of membership in either theHigh Cr or Low Cr group was assessed from the MahalanobisD2 statistic as calculated from scores for the first four PCs. Anumber of the clay samples show a significant probability ofmembership in either the High Cr or Low Cr group.

Three clay samples from two locations had a high probabil-ity of membership in the High Cr group: one washed and twounwashed samples from locations 3 and 4 in the San Pedro deAtacama region (Fig. 5). These samples were collected fromalluvial deposits in dried pools in the riverbed of the RioGrande de San Pedro, from directly below the site of Quitorand from beneath the bridge across the Rio Grande on theSan Pedro-Calama road, respectively. A single ethnographicceramic sample, made of clay from deposits near Volcan Tatioand tempering agents from near Toconce, also joined the HighCr group. Since a sample of clay from the Toconce-1 sourcewas not similar to either the High Cr or Low Cr groups, itwould appear that the tempering agents in the T-67 sampleare responsible for the High Cr compositional signature ofthis sample.

Clay samples from six locations joined with the Low Crgroup. These included samples taken from the bed of theRio Vilama, from alluvial deposits in a large arroyo debouch-ing into the Rio Grande just north of Catarpe, and from broaddeposits of Plio-Pleistocene material accumulated along theflanks of the chain of Andean volcanoes running along theChile-Bolivia border. In contrast, samples unaffiliated with ei-ther ceramic composition group are largely possible temperingsands or arbitrary mixtures of clays and temper. Most of thesetemper or clay-and-temper samples have substantially lower

Fig. 3. Distribution of minor composition groups, based on values of Cr and Co.

579logical Science 33 (2006) 575e594element concentrations than were evident in the ceramic orclay samples, suggesting that these samples contain a higher

port for the argument that the ceramics from Catarpe andTuri were locally produced from locally available clays andtempering agents.

a weight ratio basis. When Cogswell et al. fired oven-driedclay samples to 800e1000 C, they reported a weight lossof 4e5% [8], suggesting that a small but possibly significantincrease in element concentrations would occur after rawclay samples are fired. To simulate this effect, we increasedthe element concentrations observed in our dry clay samplesby 2% and 4% and recalculated PC scores and probabilitiesof group membership using the simulated concentrations.This simulation had only minor effects on the probabilitiesof group membership, although it did raise the probabilitiesthat Cl-02u was associated with the High Cr group and Cl-16 associated with the Low Cr group at above the 5% thresh-old (Appendix A). Although neither of those changes wouldalter the conclusions we came to using the raw clay elementconcentration data to calculate probabilities of group member-ship, they indicate that it would be preferable to use fired claysamples in any further INAA studies on material from thisregion.

5.4. Classification of non-core group samples

Ceramic samples initially excluded from the definition oflocal core groups (i.e., obvious imports and copies of imperial-style Inka ceramics) were tested for group membership inthe High Cr or Low Cr groups, again using the Mahalanobis2

Fig. 4. Distribution of clay samples relative to primary composition groups.fraction of quartz-rich sand than was typical for materials usedby prehistoric potters.

The association of these clay samples with the primarycomposition groups of local style ceramics offers strong sup-

Table 2

Classification of clay samples to local ceramic composition groups

Sample Type Site Provenience

CL-03u Clay, unwashed Quitor Rio Grande

CL-03w Clay, washed Quitor Rio Grande

CL-04u Clay, unwashed San Pedro Rio Grande

T-67 Ethnographic ceramic Turi Toconce-1,

CL-01u Clay, unwashed Vilama Rio Vilama

CL-01w Clay, washed Vilama Rio Vilama

CL-02u Clay, unwashed Catarpe Arroyo nort

of site

CL-02w Clay, washed Catarpe Arroyo nort

of site

CL-05 Clay, unwashed Source 1 Source 1



CL-13 Clay and temper Source 3 Source 3


CL-04w Clay, washed San Pedro Rio Grande

CL-06 Temper Source 1 Source 1







T-68 Clay Turi Toconce-1,

T-69 Clay Turi Toconce-2

580 J.R. Alden et al. / Journal of ArchaeoSolid stars show samples identified with either the High Cr or Low Cr compo-

sition groups.One possible problem with this analysis is that our claysamples were not fired. Unfired clays contain small quantitiesof adsorbed water and organic material that will increase sam-ple weights and thus reduce element concentrations on

Group P (High Cr) P (Low Cr)

riverbed High Cr 0.467 0.000



near Tatio High Cr 0.190 0.020

riverbed Low Cr 0.000 0.159

riverbed Low Cr 0.000 0.175

h Low Cr 0.042 0.367

h Low Cr (mixed) 0.053 0.534

Low Cr 0.000 0.490

Low Cr 0.000 0.068

Low Cr 0.000 0.350

Low Cr 0.000 0.100

Low Cr 0.000 0.168

riverbed Unclassed 0.016 0.036

Unclassed 0.000 0.000







near Tatio Unclassed 0.001 0.000


logical Science 33 (2006) 575e594D statistic based on PC scores. This process identified a num-ber of samples (including some that had been identified on

stylistic criteria as presumed imports) that had a high probabil-ity of membership in either the High Cr or Low Cr group.These samples are designated as Non-Core in the Statuscolumn of Appendix B. This procedure also identified six sam-ples with compositions intermediate between the High Cr andLow Cr groups (labeled as mixed in the CompositionGroup column of Appendix B), and twelve samples withno significant probability (P < 0.05) of belonging to eitherof the two main Catarpe/Turi composition groups.

One additional cluster, a group of nine samples distin-guished by their low sodium content (Low Na), was isolatedfrom the set of samples that could not be classified as eitherHigh Cr or Low Cr (Fig. 6). These Low Na samples were

(CT-07, CT-18, and T-60) also had distinctly low iron to scan-dium (Fe:Sc) ratios, indicating that the Low Na group mightbe further divisible along that axis of variation.

The final classification of ceramic samples into composi-tional groups is presented in Appendix B, along with dataon the source and characteristics of each sample.

6. Discussion

6.1. Ceramic production, geology, and clay source areas

We know of no direct evidence of ceramic productiondkilns,pottery-making tools, wasters, or ceramic slagdfrom any sites in

0 10kilometers

N

Rio

San P

edro

Archaeological Site

Clay Sample Location

2500 m

2500 m

Contour Interval 100 meters

Vilama

Catarpe

Source 2Source 3

San Pedro

de Atacama

Quitor

3000 m26

00 m

Fig. 5. Site and clay sample locations in the area of San Pedro de Atacama.Sourc

Source 4

3000

m

Rio

Vilam

a

J.R. Alden et al. / Journal of Archaeoall from sherds that were classified as exotic based on theirpaste and surface treatment. Three members of this groupe 1

350 0 m

581logical Science 33 (2006) 575e594the Turi/San Pedro de Atacama region, and hence will not at-tempt to address questions about the organization of local or

regional ceramic production. However, because this region hasbeen relatively well surveyed, we assume that the majority ofthe pottery used at Catarpe and Turi was manufactured at or inthe immediate vicinity of the known settlements and not atsome undiscovered special production centers. It is also clearthat finished pottery may have been moved from site to sitewithin the region, through casual transport, market exchange,immigration, or state-controlled redistribution, and evidencefor such movement will be discussed later.

Ethnographic data from a worldwide sample of societiessuggest that most ancient potters would probably travel notmore than 1 km to obtain their clays and tempers while fewwould travel more than 7 km. [4]. In Toconce, however, a vil-lage about 20 km east of Turi, traditional potters make a roundtrip of 2 days on foot to reach one favorite clay source, andthey routinely gather tempering material from locations atleast 5 km from the town where they live [22]. Favored loca-tions for finding temper include sources of micaceous materiala moderate walk from Toconce, as well as a source at the footof Cerro Copacoya about a day and a half on foot from To-conce. Nevertheless, in societies without mechanized transportpotters surely preferred to exploit nearby sources of clayand temper [3], and we assume that the potters of Catarpeand Turi used local sources of clay and temper wheneverpossible.

Many of the clay samples tested had compositions consis-tent with membership in either the High Cr or Low Cr ceramiccomposition groups (Table 2 and Fig. 4). Clays associated withthe Low Cr group came from six sources in the vicinity ofmodern San Pedro (Fig. 5). The distribution of these sourcesrelative to the principal geological regimes in the Turi/San Pe-dro region (Fig. 7) suggests that Low Cr clays from Sources 1,2, 3, and the Rio Vilama derive from the Plio-Pleistocene

Fig. 6. Distribution of minor composition groups, based on values of Na and Cr.

582 J.R. Alden et al. / Journal of ArchaeoChaxas ignimbrite (a deposit of dacitic tuff, pumice, and smalllava bombs), while the Low Cr clays from Source 4 and thearroyo north of Catarpe derive from the Oligocene-MioceneSan Pedro Formation (sandstone, mudstone, and clay, withlimited intercalations of conglomerates and tuff) [20]. Together,these two formations are the primary sources for the alluvialdeposits associated with prehistoric occupations in the vicinityof San Pedro de Atacama. Clays from these deposits are com-positionally similar to our Low Cr group of sherds, and thegreat majority of the ceramics we sampled from this regionfell into the Low Cr composition group.

Turi is located in an area of alluvial gravel, sand, and clayderived from Pleistocene-Holocene volcanic deposits north-east of the site. Unfortunately, we were not able to collectclay samples from these deposits, and no amount of discussionof the regional geology can truly mitigate the lack of actualsamples. However, while the deposits around Turi and thosearound San Pedro differ in age and in details of petrologicalcharacter, they all derive from the extended episode of volca-nism responsible for creating the central spine of the Andeancordillera. Therefore, we suggest that the Low Cr clays used inthe production of Tardo Period ceramics at Turi, in a situationanalogous to what we were able to demonstrate at Catarpe/SanPedro de Atacama, originated from the erosion of volcanic de-posits, of different ages but broadly similar elemental compo-sition, coming from the range of high volcanoes to the east ofthe Turi/San Pedro region (Fig. 7). Such erosional deposits arereadily available in the immediate vicinity of Turi.

Clays associated with the High Cr composition group camefrom three locations: two places in the bed of the Rio Grandenear San Pedro de Atacama, and an ethnographic sherd, sam-ple T-67, made of clay from the Toconce-1 source and temperfrom an unspecified source of micaceous material near the vil-lage of Toconce. The Toconce-1 source (Fig. 7) is some 35 kmESE of the village of Toconce, and a sample of clay from thatsource (sample T-68) does not have a composition similar toeither the High Cr or Low Cr groups. We conclude that thetempering agents in sample T-67 are responsible for makingthe composition of this piece of modern pottery similar tothe composition of the High Cr sherds in our sample of prehis-toric pottery.

Our preliminary clay survey did not unambiguously identifythe geological sources of the High Cr composition group.However, the distribution of the three High Cr clay samplesand the concentration of High Cr ceramics at Turi (see Table 3)coincide in a suggestive way with the distribution of theCretaceous era Purilactis and Lomas Negras Formations(Figs. 5 and 7). Turi is less than 10 km north of a large expo-sure of Purilactis sedimentary rock that is being eroded by theRio Salado tributary of the Rio Loa, while the Rio Hojalar andRio Toconce, the northernmost tributaries of the Rio Salado/Loa, both collect outwash from the Lomas Negras Formationeast of Toconce [20]. Finally, along its upper reaches, the RioSalado that flows southward into the Rio Grande de San Pedroruns near a stretch of Purilactis Formation deposits severalkilometers in length. This tributary contributes a significantpart of the sediment carried by the Rio Grande, and it is pos-

logical Science 33 (2006) 575e594sible that the High Cr clays found in the bed of the Rio Grandederive from those Cretaceous era Purilactis deposits. However,

at least one hard rock mine is operating in intrusive graniticdeposits adjacent to the Rio Salado/Rio Grande, and thatmine is dumping tailings directly into the canyon of the river.Thus, it is possible that the elevated levels of chromium andother transition metal elements evident in the clay samplesfrom the Rio Grande alluvium are a result of modern miningactivity and not characteristic of the prehispanic riverine clays.At this time we consider this to be the most likely explanationfor the presence of High Cr clays in the Rio Grande alluvium,although further testing will be needed to prove or disprovethis hypothesis.

6.2. Distribution of composition groups by site

Because the ceramic samples tested here were not randomly

caution. Still, a cursory examination of the data in Table 3 re-veals several potentially meaningful differences between thematerial from Turi and that from Catarpe. Most notably, whilethe assemblage of ceramics from Turi contains approximatelyequal numbers of High Cr and Low Cr sherds, the Catarpesample has more than twice as many ceramics from the Low

0 50kilometers

Rio

Sal

ado

Rio V

ilama

Vilama

Catarpe

contour interval 500 meters

San Pedrode Atacama

Archaeological Site

Modern Town

International Border Granitic intrusions, Cretaceous

Purilactis and Lomas Negras Formations, CretaceousVolcanic andesites and scorias, Plio-Pleistocene & QuaternaryIgnimbrites, Plio-Pleistocene

Ignimbrites and redeposited volcanics, Miocene & OligoceneClay Sample Location

Ri o

Gra

nde

Fig. 7. Geology of the Turi/Catarpe area, Region II, Chile.

Table 3

Distribution of ceramic composition groups at sampled sites

Site High

Cr

Low

Cr

Extreme

Cr

High

Co

Low

Na

Mixed Uncl. Total

Catarpe 11 23 2 0 7 1 1 45

Turi 25 21 0 6 2 2 10 66

Beter-3 1 8 0 0 0 1 0 10

Vilama 0 5 0 0 0 0 0 5

Solor-13 1 1 1 0 0 1 1 5Rio Salado

TuriRi

o Lo

a

J.R. Alden et al. / Journal of Archaeoselected, any apparent patterning must be interpreted withToconce

583logical Science 33 (2006) 575e594Road Survey 2 0 0 0 0 1 0 3

Cr group than from the High Cr group. When the Catarpesherds are combined with the ceramics from Beter-3, Vilama,and Solor-13 (three other sites in the environs of modern SanPedro de Atacama), the combined sample has almost threeLow Cr sherds for every High Cr sherd.

Despite the scarcity of High Cr sherds at the sites aroundmodern San Pedro, all three of the Extreme Cr samplescome from sites in the San Pedro region. Conversely, all sixof the High Co sherds and ten of the twelve unclassified sherdscome from Turi. It is also worth noting that, as shown in Ap-pendix B, five of the six High Co sherds have significant prob-abilities of membership in the High Cr composition group.The limited distribution of the minor ceramic compositiongroups in particular regions implies that the clay and temperused to manufacture those vessels came from geological de-posits in the neighborhood of the sites where each of thesegroups is found. (Although most of the Low Na sherds comefrom Catarpe, that distribution reflects a sampling biasdtheLow Na sherds are primarily Yavi-style ceramics importedfrom the eastern side of the Andes, and more Yavi-style sherdswere selected for testing from Catarpe than from Turi. A sim-ilar sampling bias is evident in the Unclassified group, whichincludes four mica-tempered (pasta con mica) sherds fromTuri. A small number of mica-tempered sherds were excavatedat Catarpe, but none of those sherds were selected for neutronactivation.)

The predominance of Low Cr sherds in Catarpe and the SanPedro area sites and the presence of Low Cr clay sourcesthroughout that area indicate that the primary clay sourcesused by prehistoric potters in the environs of San Pedro deAtacama were Low Cr geological deposits from north andeast of the modern town, or alluvial clays derived from thosegeological deposits. If the High Cr ceramics found in the SanPedro area sites were locally manufactured rather than im-ported, then the prehistoric potters in that region were appar-ently using alluvial clays from the bed of the Rio Grande tomanufacture those ceramics. The prehistoric potters of theSan Pedro region, however, certainly used High Cr claysmuch less commonly than clays from Low Cr sources. Thisis true even for Catarpe, where alluvial deposits from theRio Grande, which is within 100 m of the site, would havebeen an extremely convenient source for clay and temper.

The pottery used by the inhabitants of Turi was made usingclays of a wider variety of compositions: High Cr, Low Cr,High Co, and several unclassified clays. This pattern may re-flect sampling strategies, since the Turi samples were selectedto examine the entire range of paste types found in the ce-ramics from the site. More probably, however, it is a resultof Turis geological situation, in an area where clays derivedfrom a variety of bedrock regimes are readily available.

6.3. Distribution of composition groups within ceramictypes and vessel forms

A significant proportion of the sherds analyzed in this

584 J.R. Alden et al. / Journal of Archaestudy, particularly among the samples from Turi, are not rimsherds, so vessel forms for those samples are not directlydeterminable. However, each of the Tardio phase ceramicgroups is by definition closely associated with a small set ofvessel forms [23]. As a result, ceramic group categories canbe used as a surrogate indicator of vessel form. The primaryform associated with each ceramic group is listed in Table 4,and rim profiles of sampled sherds are shown in Figs. 8 and 9.

As Table 4 shows, no sherds classified as Ceramic Groups1, 2, 7, or 40 fall into the High Cr compositional groupdtheyare either Low Cr (21 examples) or from one of the outliercompositional groups (eight examples). These four groups (re-spectively, Rojo Alisado, Rojo Burdo, Turi Gris Alisado, andRojo Alisado doble cuerpo jars) are associated with medium,large, and very large jars or occasionally, deep basins, andthey are relatively poorly made, with rough to lightlysmoothed exterior surfaces and relatively coarse paste. Be-cause such vessels are rough and cumbersome, it is expectedthat they were seldom transported far from the places wherethey were made.

The pattern of elemental compositions found in Group 38(red slipped and smoothed exterior; red smoothed interior)vessels is like that of the large jars of Groups 1, 2, 7, and40. Except for one example in the High Cr group, all sherdsfrom Group 38 are Low Cr, from one of the minor groups,or unassigned. Groups 30 and 51, which are ceramic typesfound only at Turi and are also primarily jars [23], show thesame compositional pattern.

In summation, of the medium to large sized jars in the Turi/San Pedro de Atacama region (typological Groups 1, 2, 7, 30,38, 40, and 51), 36 of 52 were made from Low Cr clays andtempering agents either demonstrably (in the San Pedroarea) or presumably (for Turi) available in the immediateneighborhood of the sites where the vessels were used. Fifteenof the sherds from these typological groups fell into the minor,mixed, or unclassified composition groups. Only one sherdfrom the typological groups typical of jars fell into the HighCr composition group.

Table 4

Distribution of composition groups by local ware types

Ceramic type

and primary

vessel form

Ceramic composition group

High

Cr

Low

Cr

Extreme

Cr

High

Co

Mixed Unclassed Total

Gr. 1dJars 0 3 0 3 1 2 9

Gr. 1?dJars 0 1 0 0 0 0 1Gr. 2dJars 0 11 2 0 0 0 13

Gr. 7dJars 0 2 0 0 0 0 2

Gr. 9dBowls 12 6 0 0 1 1 20Gr. 9AdBowls 2 4 0 0 0 0 6

Gr. 9BdBowls 1 2 0 0 1 0 4

Gr. 30dJars 0 1 0 2 0 0 3

Gr. 32dBowls 11 4 0 0 1 1 17Gr. 36?dBowls 1 0 0 0 0 0 1

Gr. 37dBowls 7 1 0 0 0 0 8

Gr. 38dJars 1 11 1 1 0 2 16

Gr. 40dJars 0 4 0 0 0 0 4

ological Science 33 (2006) 575e594Gr. 51dJars 0 3 0 0 0 1 4

Not defined 1 1 0 0 0 0 2

The compositional group distribution observed for the typo-logical groups dominated by bowlsdGroups 9, 32, 36, and37dis quite different from the distribution observed for groupswhere the predominant forms are jars. The bowls in these fourtypological groups are very similar in size and shape (Fig. 9),and because they may have been functionally equivalent, anypatterning evident in the distribution of bowls belonging tothe various composition groups is particularly interesting. (Itis possible, however, that the groove around the rim of thegrooved rim bowls served to facilitate tying a cover over thetop of this particular type of bowl. If this was the case, thenthe grooved rim bowls might have been more suited for trans-porting dry materials than other bowl types.)

flattened rims, are ubiquitous in Solor and Tardo Phase occu-pations in the Calama region. Ayquina vessels come in twocolor variants (9A and 9B, reddish brown and grayish brown),and Group 9 is the only typological group (15 High Cr; 12Low Cr) with roughly equal numbers of sherds in the HighCr and Low Cr composition groups.

Group 32 (Dupont) ceramics are also ubiquitous in Solorand Tardo Phase occupations in the Calama area. Dupont ce-ramics may be somewhat earlier than Ayquina, but they appearin the same range of forms and show the same geographicaldistribution. The only notable difference between the twotypes is their color: Dupont vessels are dark gray to black,while Ayquina ceramics range from reddish to grayish brown.

CT-19Gr. 38, Low Cr

CT-20Gr. 38, Low Cr

CT-36Gr. 31?, Low Cr

centimeters 100

CT-21Gr. 1, Unclassed

CT-39Gr. 38, Low Cr

CT-22Gr. 1, Low Cr

CT-16Gr. 31B, Low Na

CT-23Gr. 36, Low Cr

CT-24Gr. 1?, Low Cr

CT-25Gr. ?, Low Cr

CT-42Gr. 38, Low Cr

CT-43Gr. 38, Low Cr

T-35Gr. 1, High CoT-34

Gr. 1, Low Cr

T-25Pasta con mica

Unclassed

Other Jar and Vessel Forms

AribaloidJars

Fig. 8. Rim profiles of analyzed ceramicsdJars and other forms.CGr. 2CT-14

Gr. 31B, Low Na

CT-57Gr. 2, Low Cr

CT-58Gr. 2, Low C

CT-59Gr. 2, Low Cr

CT-56Gr. 2, Low Cr

Flaring

J.R. Alden et al. / Journal of ArchaeoGroup 9 (Ayquina) vessels, which are primarily medium-sized bowls with simple round rims (SRR bowls) or interiorCT-53Gr. 2, Extreme Cr

CT-54Gr. 2, Extreme Cr

T-55, Low Cr

r

T-36Gr. 1, High Co

T-42Gr. 30, High Co

Rim Jars

585logical Science 33 (2006) 575e594Because such variation in color may reflect nothing more thanunintentional variations in firing conditions, archaeologists

working in the region have hesitated to assign too much signif-icance to the distinction between the two ceramic types. How-ever, the sample of sherds tested here indicates that Dupontvesselsd11 High Cr: 4 Low Crdare noticeably more likelyto have been made from High Cr clays than are Ayquina ves-sels, hinting that the distinction between Ayquina and Dupontmay be more meaningful than archaeologists have heretoforebeen able to demonstrate.

Seven out of eight of the Group 37 ceramics (bowls withred slipped and polished interior, smoothed light brown exte-rior) fall into the High Cr composition group. Group 37 pot-tery is readily distinguished from other Tardo Phase pottery,and because at least half of the sherds that were tested from

grooved-rim bowls, were being made at a single locus of pro-duction in the Turi/San Pedro region.

For sherds with rims, which allow vessel form to be unam-biguously defined, the distribution of bowl and jar forms withincomposition groups and between sites clarifies the patterns ofdistribution indicated by the data on ceramic typologicalgroups. At Catarpe bowls with all three types of rim form oc-cur in both High Cr and Low Cr variants (Table 5). At Turi,however, only High Cr bowls are found. At Beter, Vilama,and Solordthe sites in the environs of San Pedrodthere isa clear predominance of Low Cr composition bowls, whileonly High Cr bowls occur in the small number of bowl rimssampled from sites located during the road survey. All the un-

CT-26Gr. 9B, Mixed

centimeters100

CT-11Gr. 31, Low Na

CT-32Gr. 32, High Cr

CT-27Gr. 36?, High Cr

CT-15Gr. 31A?, Low Na


CT-33Gr. 9B, Low Cr



CT-40Gr. 37, Low Cr

CT-34Gr. 9B, Low Cr


CT-31Gr. 9A, Low Cr

T-08Gr. 9, High Cr

T-17Gr. 37, High Cr

T-31Gr. 32, High Cr

T-45Gr. 9, High Cr

T-05Gr. 9, High Cr

T-10Gr. 9, High Cr

T-39Gr. 36, Low Cr

T-09Gr. 9, High Cr

T-07Gr. 9, High Cr

T-03Gr. 9, High CrT-02

Gr. 9, High Cr

T-49Gr. 38, Low Cr

InteriorFlattened Lip

Bowls

Grooved Rim Bowls

Other Bowl Forms

Fig. 9. Rim profiles of analyzed ceramicsdBowls.CT-44Gr. 32, High Cr

CT-45Gr. 32, High Cr CT-46

Gr. 32, Low Cr

CT-50Gr. 9A, Low Cr

CT-51Gr. ?, High Cr

CT-52Gr. 9A, Low Cr

T-46Gr. 32, High CrT-18

Gr. 37, High Cr

T-30Gr. 37, High Cr

Simple Round

586 J.R. Alden et al. / Journal of Archaeothis group have the very distinctive grooved rims, it seems pos-sible that Group 37 ceramics, and particularly the Group 37CT-48Gr. 9A, Low Cr CT-49

Gr. 9A, High Cr CT-47Gr. 32, High Cr

RS-02Gr. 1, Mixed

RS-01Gr. 9B, High Cr

RS-03Gr. 9A, High Cr

T-20Gr. 36, Mixed

T-48Gr. 32, Unclassed

Rim Bowls

logical Science 33 (2006) 575e594ambiguous jar forms from both Catarpe and Turi fall withinthe Low Cr composition group.

6.4. Reconstructing economic patterns from the ceramiccomposition data

Hayashida has observed that no single model of potteryproduction and distribution applies to all regions of theInka Empire [16]. The discussion that follows attempts to elu-cidate several features of the ceramic economy of one partic-ular Inka provincial domaindthe Catarpe/Turi regiondandthereby add new detail to the picture of how the Inka admin-istered and exploited different regions in their vast empire.

None of the samples chosen for neutron activation were ran-domly selected. However, the set of bowl rims that were ana-lyzed do represent a blind sample of the bowls found at eachof the sites. That is, there is no bias toward any particular com-position group inherent in the procedures used to pick sherdsfor analysis. Thus, the chi2 statistic can appropriately be usedto test the statistical significance of the distribution of bowlsin the two main composition groups between various sites.

Table 6 shows the distribution of High Cr and Low Crbowls in the Catarpe/Turi region. If bowl composition typesare independent of the sites where the bowls were found,then the distribution observed here differs from the expecteddistribution at the 0.005 level of significance. In short, if thesample of bowls selected for analysis is indeed unbiased, itis extremely unlikely that the observed distributiond100%High Cr bowls at Turi, 61% High Cr bowls at Catarpe, and20% High Cr bowls at sites in the San Pedro de Atacamaoasisdis accidental. We therefore reject the null hypothesisthat bowl composition types and geographical locations varyindependently.

Table 5

Composition groups of principal vessel forms, by site

Site Source

group

Grooved

rim

bowls

Int. flat

lip bowls

SRR

bowls

Total

bowl

rims

Jars

Aribaloid Other

Catarpe High Cr 4 2 5 11

Low Cr 1 2 4 7 6 6

Turi High Cr 2 7 4 13

Low Cr 1

Beter High Cr 1

Low Cr 4

Solor High Cr 1

Low Cr

Vilama High Cr

Low Cr 4

Road survey High Cr 2 2

Low Cr

Table 6

Contingency table, distribution of common composition groups in bowls

High Cr Bowls Low Cr Bowls Total

Catarpe 11 7 18

Turi 13 0 13

Beter/Solor/Vilama 2 8 10

J.R. Alden et al. / Journal of Archaeo26 15 41

Chi-squared 15.67; d.f. 2; P < 0.005.The observed distribution of High Cr bowls and Low Crjars at Turi indicates that the potters at that site habitually uti-lized two separate sources of raw materialda High Cr sourceto make their bowls, and a Low Cr source to make their jars.We have argued, using geological data, that the Low Cr mate-rial was collected from the area immediately around the site,and that the High Cr material came from Cretaceous era de-posits found 8e10 km south of Turi, where they are beingeroded by the Rio Salado. It is also significant that the compo-sitional data from neutron activation analysis supports the dis-tinction observed visually by Varela et al., [23] between theaplastic inclusions typical of the Turi Standard 1 pastes foundin Groups 1, 2, 38, and 51 (the Turi jars) and the inclusionstypical of the Turi Standard 2 pastes associated with the Group9, 32, 36, and 37 Turi bowls. (A similar relationship betweenceramic wares and tempering agents is evident in Inka-era pot-tery from the Cusichaca Valley, near Cuzco [17].) Standard 1temper is angular and larger than the Standard 2 temper typicalof bowls, which has smaller and more rounded particles ina wider variety of colors. Because it seems likely that therounded particles in Standard 2 temper came from river-washed sands, the observations made visually by Varelaet al., support the suggestion that the potters of Turi collectedHigh Cr material from the bed of the Rio Salado, and that theyused this material exclusively for manufacturing bowls.

The pattern of vessel composition is less clear at Catarpeand in the neighboring sites in the oasis of San Pedro de Ata-cama. At Catarpe, all the jars are made from Low Cr material,but bowls are found with both High Cr and Low Cr composi-tions in roughly equal proportions. In the neighboring sitesbowls of both compositions are also present, although LowCr bowls are noticeably more common than High Cr bowls.

This pattern could be explained in two ways. Tardo periodpotters at Catarpe and in the San Pedro region might havemade bowls from both High Cr and Low Cr raw materials,or they might have only used Low Cr clays for making pottery,and imported vessels of the High Cr compositional group. Butthe only source for High Cr clays identified in the San Pedroregion is the alluvium of the Rio Grande, a river that flowswithin 100 m of Catarpe. Since the Rio Grande alluviumwas the most easily available source of clay for the pottersof Catarpe, we expect that they would have used that clayfor making large jars and basins as well as for making bowls.Because no jars from Catarpe fell into the High Cr composi-tion group, we believe that during prehispanic times the RioGrande alluvium was not a source of High Cr clay or temper.We instead favor the second possibility, that potters at Catarpeand in the San Pedro region made all their pottery using LowCr material and that the High Cr composition bowls found atCatarpe and in the San Pedro region were acquired throughsome form of exchange with Turi. Petrographic analyses ofthe bowl sherds, it should be noted, might allow us to test thesealternate hypotheses.

If this reconstruction of ceramic production and distributionis valid, then the exchange between Turi and Catarpe was no-

587logical Science 33 (2006) 575e594tably one-sideddHigh Cr bowls are coming to Catarpe fromTuri, but Low Cr bowls are not going to Turi from Catarpe.

Furthermore, sites in the San Pedro oasis are getting a notice-ably lower proportion of bowls from Turi than Catarpe is get-ting. This pattern is suggestive of a regional ceramic economybased on state-controlled extraction rather than state-managedproduction, and state-controlled redistribution rather than dis-tribution through casual transport or market exchange. This re-construction of Inka period economic organization in theCatarpeeTuri region accords closely with the suggestionmade by DAltroy et al. [12] that the Inkas may have directlyrequisitioned ceramics in some regions, particularly duringthe early stages of imperial occupation, acquiring ceramicsfrom existing pottery production specialists rather than impos-ing direct control over local ceramic production or resettlingforeign potters (mitmaqkuna) into the region. The observeddistribution of High Cr and Low Cr bowls also implies thatin the Inka administrative system, Catarpe was the regionallydominant center during the era of Inka control in northernChile, with Turi being a subservient site.

6.5. Compositions of non-local style ceramics

Small numbers of exotic ceramics (sherds with pastes, sur-face treatments, or painted decoration that appear non-local instyle) are found at both Turi and Catarpe. A series of thesesherds were sampled to determine their compositions. If thosecompositions were similar to the local style ceramics it wouldindicate that the sampled sherds were locally produced copiesof foreign-style ceramics, while if the compositions were dif-ferent it would indicate that they had been produced elsewhereand imported into the site where they were found.

Five types of pottery were considered as potentially exotic.Group Pasta con mica sherds, sampled only at Turi, exhibita distinctive paste containing large quantities of micaceoustemper. Group 36 or 36-Inka sherds are red slipped and bur-nished on both faces, and the sherds labeled as 36-Inka aresimilar to the shallow plates with modeled duck or animalheads and tails on their rims that are a common componentof the imperial Inka ceramic assemblage [6,10]. Group 31 in-cludes three varieties of Altiplano pottery: fine red paste(Group 31), Hedionda black on tan (Group 31A), and Yavi(Group 31B) [23]. Twenty-four sherds from these groupswere sampled to determine their elemental compositions.

As Table 7 shows, four of the five Pasta con mica sherdsfrom Turi do not fit into any of the composition groups definedin this study. Although the presence of mica may be sufficientto alter the compositional profile of these sherds, the principalcomponents procedure is robust enough that we suspect themicaceous pottery from Turi was made using clay or temperfrom a geologically distinct source with a distinctive composi-tion. Further testing will be needed to define a compositionalgroup for this class of ceramics.

Nine out of the ten Group 31A, Group 36, and Group 36-Inka sherds have compositions that fall into the High Cr, LowCr, or Mixed groups. These sherds thus appear to be locallymade copies of foreign-style ceramics. Aribaloid jars, which

588 J.R. Alden et al. / Journal of Archaeowere not considered as potentially exotic because their pastesand surface treatments were visually indistinguishable fromthe pastes and surface treatments of other Tardo pottery, arealso a vessel form that is part of the standard Inka assemblage.The six aribaloid jar rims from Catarpe fell into the Low Crcomposition group (Table 5), indicating that they were also lo-cally made. It should also be noted that the scarcity of Inka stylepottery in the Catarpe/Turi region contrasts markedly with thesituation in the Upper Mantaro Valley of Peru, where Inka stylepottery, including Inka polychrome, is very common [10].

Costin, working with ceramics from the Upper MantaroValley of Peru, concluded that the advent of Inka rule in thatregion resulted in no quantitative change in the organizationof local production of ceramics, but adds that the Inka didestablish a new production system to procure politically andsocially symbolic vessels of the Inka style [9]. Because theforms and fabrics of the Inka period Tardio ceramics are virtu-ally identical to the forms and fabrics characteristic of thepre-Inka Solor Phase material, we believe that when theInka incorporated the Catarpe/Turi region into their expandingempire, they did nothing to alter the technology or organiza-tion of local ceramic production. They did, however, obtaina small number of locally made copies of Inka style vessels.

In contrast to the locally produced Inka style pottery, all theGroup 31 (Altiplano) and 31B (Yavi) ceramics that were testedfell into the Low Na composition groupda composition notfound in any of the core group of local style sherds fromTuri and Catarpe. Thus, Group 31 and 31B ceramics appearto be genuine exoticsdproduced in some other region and im-ported into the Turi/San Pedro area through some form oflong-distance exchange. The source area for Yavi pottery isin the Jujuy region of northwestern Argentina, and the similarcompositions of the Yavi and Altiplano sherds indicate that theGroup 31 sherds tested in our study also came from the east ornortheast of the Turi/San Pedro region.

7. Conclusions

Most of the Tardo style ceramics tested in this study fellinto two major composition groups, divided most clearly onthe basis of their chromium content. The High Cr group ap-

Table 7

True exotics versus local copies

Ceramic

type

Ceramic composition group

High

Cr

Low

Cr

Extreme

Cr

High

Co

Low

Na

Mixed Unclassed Total

Gr. Pasta

con mica

1 0 0 0 0 0 4 5

Gr. 31A,

Hedionda

2 1 0 0 1 1 5

Gr. 31? 1 1

Gr. 31,

Altiplano

0 0 0 0 2 0 0 2

Gr. 31B, Yavi 0 0 0 0 6 0 0 6

Gr. 36 0 2 0 0 0 1 0 3

Gr. 36dInka 1 1 0 0 0 0 0 2

logical Science 33 (2006) 575e594pears to be associated with clays and/or tempering agents de-riving from Cretaceous era deposits of the Purilactis or Lomas

of cobalt (the High Co group). Geological sources have notbeen identified for either of these compositional groups.

Yavi-style pottery, distinguished by a notably low sodiumconcentration (the Low Na group), was imported into Turiand Catarpe from the Jujuy region of northern Argentina. Itseems likely that the Altiplano style sherds in the Low Nagroup were also produced somewhere east of the crest of theAndes. In contrast, Inka style vessel forms found at Turi andCatarpe, particularly aribaloid jars and shallow plates, fallinto the High Cr and Low Cr composition groups, and we con-clude that they were locally produced.

Medium to large-sized jars in typological Groups 1, 2, 7,and 40 all fall into the Low Cr or one of the minor local com-position groups; of 29 samples in these typological groups,none had a High Cr composition. A similar pattern is evidentin the Group 38 sherds, where only one of the 16 sherds testedwas in the High Cr group. We argue that the very large, large,and medium sized jars typical of these typological groupswere made in or near to the settlements where they werefound, using sources of clay and temper available in the imme-diate neighborhood of the site.

Medium sized bowls, in typological Groups 9, 32, and 37,were made with clay and temper from both High Cr and LowCr sources. Because no Low Cr bowls in these typologicalgroups were found at Turi, we conclude that the potters atTuri deliberately utilized a special source of High Cr clayand temper to manufacture their bowls. This source, we sug-gest, was somewhere downstream of the area where the Rio

Appendix A. Simulation of firing on clay sample element co

INAID P (High Cr) P (Low Cr)

CL-01u 0.000 0.159

CL-01u 0.000 0.154

CL-01u 0.000 0.144

CL-01w 0.000 0.175

CL-01w 0.000 0.171

CL-01w 0.000 0.162

CL-02u 0.042 0.367

CL-02u 0.064 0.343

CL-02u 0.095 0.309

CL-02w 0.053 0.533bution rather than distribution through casual transport ormarket exchange. In addition, the observed distribution suggeststhat Catarpewas the regionally dominant center during the era ofInka occupation in northern Chile, and that, in the Inka provin-cial administrative system, Turi was subservient to Catarpe.

Acknowledgments

The Beter 3, Solor 13, and Vilama 2 samples were selectedby Agustin Llagostera; samples from Catarpe Tambo and theInka Road sites were selected by John Alden and Tom Lynch;and samples from Turi, which were collected during a projectfunded by Fondecyt (#1011006), were selected by Carlos Al-dunate, Victoria Castro, Mauricio Uribe, and Varinia Varela.Geological clay samples were collected by Alden (the San Pe-dro de Atacama region) and Varela (the samples from To-conce). The authors would like to thank these scholars forproviding materials for our study. We would also like to thankthe staff of the Museo LePaige in San Pedro de Atacama fortheir assistance with our ceramic analyses and for their unstint-ing hospitality, and we are grateful to Chiles Consejo de Mon-umentos Nacionales for giving permission to export samplesof archaeological material for neutron activation analysis.Frances Hayashida and Jeff Parsons offered many useful sug-gestions about the content and organization of this paper, andtwo anonymous reviewers offered criticisms and observationsthat helped us improve our presentation in material ways; wethank them all for their contributions.

ncentration data

Treatment Location Group

Clay 0% Below Vilama Low Cr

Clay 2%

Clay 4%

Clay 0% Below Vilama Low Cr

Clay 2%

Clay 4%

Clay 0% Next to Catarpe Low Cr

Clay 2%

Clay 4%

Clay 0% Next to Catarpe Low CrNegras formations, while the Low Cr group appears to be as-sociated with deposits deriving from Plio-Pleistocene and Ho-locene era volcanism. Because the High Cr and Low Crgeological source formations are both widely distributedacross the region and potential source deposits have notbeen extensively sampled, it is not yet possible to equate eitherof the two main composition groups with a single geologicallocus or specific archaeological production locations.

Prehispanic potters in the area around San Pedro de Ata-cama occasionally used a source of raw material that yieldedpottery with notably high concentrations of chromium (the Ex-treme Cr group). Similarly, potters working in Turi occasion-ally used a source of clay or temper with high concentrations

Salado cuts through the Purilactis Formation south of Turi.In contrast, Group 9, 32, and 37 bowls at Catarpe exhibitboth High Cr and Low Cr compositions.

Potters at Catarpe may have had access to both High Cr andLow Cr clay sources, and we cannot dismiss the possibility thatboth High Cr and Low Cr bowls could have been made atCatarpe. We think it more likely, however, that potters workingat Catarpe and in the area of San Pedro de Atacama were onlyusing Low Cr raw materials and that the High Cr bowls foundin these sites were imported from Turi. If this is the case, theexchange of vessels between Turi and Catarpe was distinctlyone-sided; a pattern suggesting state-controlled extraction ratherthan state-managed production, and state-controlled redistri-

589J.R. Alden et al. / Journal of Archaeological Science 33 (2006) 575e594(continued on next page)

Appendix A (continued)

INAID P (High Cr) P (Low Cr) Treatment Location Group

CL-02w 0.075 0.586 Clay 2%

CL-02w 0.102 0.461 Clay 4%

CL-03u 0.467 0.000 Clay 0% Below Quitor High Cr

CL-03u 0.468 0.000 Clay 2%

CL-03u 0.443 0.000 Clay 4%

CL-03w 0.339 0.000 Clay 0% Below Quitor High Cr

CL-03w 0.301 0.000 Clay 2%

CL-03w 0.250 0.000 Clay 4%

CL-04u 0.055 0.000 Clay 0% SPeCalama bridge High CrCL-04u 0.071 0.000 Clay 2%

CL-04u 0.087 0.000 Clay 4%

CL-04w 0.016 0.036 Clay 0% SPeCalama bridge Unclassed

CL-04w 0.026 0.033 Clay 2%

CL-04w 0.041 0.029 Clay 4%

CL-05 0.000 0.490 Clay 0% Source 93-1 Low Cr

CL-05 0.000 0.405 Clay 2%

CL-05 0.000 0.519 Clay 4%

CL-06 0.000 0.000 Clay 0% Source 93-1 Unclassed

CL-06 0.000 0.000 Clay 2%

CL-06 0.000 0.000 Clay 4%


CL-07 0.000 0.001 Clay 2%

CL-07 0.000 0.001 Clay 4%


CL-08 0.000 0.078 Clay 2%

CL-08 0.000 0.088 Clay 4%


CL-09 0.000 0.000 Clay 2%

CL-09 0.000 0.000 Clay 4%


CL-10 0.000 0.016 Clay 2%

CL-10 0.000 0.023 Clay 4%


CL-11 0.001 0.423 Clay 2%

CL-11 0.001 0.494 Clay 4%


CL-12 0.000 0.003 Clay 2%

CL-12 0.000 0.003 Clay 4%


CL-13 0.000 0.129 Clay 2%

CL-13 0.000 0.161 Clay 4%


CL-14 0.000 0.221 Clay 2%

CL-14 0.000 0.281 Clay 4%


CL-15 0.000 0.024 Clay 2%

CL-15 0.000 0.033 Clay 4%

590 J.R. Alden et al. / Journal of Archaeological Science 33 (2006) 575e594CL-16 0.000 0.032 Clay 0% Source 93-4 Unclassed

Appendix A (continued)

INAID P (High Cr) P (Low Cr) Treatment Location Group

CL-16 0.000 0.046 Clay 2%

CL-16 0.000 0.063 Clay 4%

T-67 0.190 0.020 Clay 0% Toconce Low Cr

T-67 0.155 0.010 Clay 2%

T-67 0.117 0.005 Clay 4%

T-68 0.001 0.000 Clay 0% Toconce Unclassed

T-68 0.001 0.000 Clay 2%

T-68 0.001 0.000 Clay 4%

T-69 0.000 0.000 Clay 0% Toconce Unclassed

T-69 0.000 0.000 Clay 2%

T-69 0.000 0.000 Clay 4%

0% raw, dried; not fired (actually irradiated and analyzed); 2% assume 2% weight loss with firing; 4% assume 4% weight loss with firing (cf. Cogswellet al., 1996). Note on procedures: Water in dry clay can dilute element concentrations. To simulate element concentrations in fired clays, values were enriched

by multiplying by 2e4% prior to calculating PC scores. Simulation suggests that firing clays would have a minimal effect on results.

Appendix B. Ceramic sample data and compositional group assignments

Sample no. Site Provenience Ceramic group Ceramic form Composition group Status P (High Cr) P (Low Cr)

B-06 Beter 3 Nivel 17 Gr. 32, Dupont Bowl rim High Cr Core 0.558 0.017

CT-27 Catarpe CT 818:2 Gr. 36? Grooved rim bowl High Cr Core 0.743 0.004

CT-28 Catarpe CT 844:5 Gr. 37 Grooved rim bowl High Cr Core 0.880 0.003



CT-32 Catarpe CT 815:2 Gr. 32, Dupont Interior flattened

lip bowl

High Cr Core 0.550 0.007

CT-35 Catarpe CT 864:1 Gr. 32, Dupont Interior flattened

lip bowl


CT-44 Catarpe CT 812:5 Gr. 32, Dupont Small SRR bowl High Cr Core 0.233 0.000

CT-45 Catarpe CT 818:5 Gr. 32, Dupont Med SRR bowl High Cr Core 0.255 0.001

CT-47 Catarpe CT 883:1 Gr. 32, Dupont Med SRR bowl High Cr Core 0.660 0.001

CT-49 Catarpe CT 849:6 Gr. 9A, Ayquina Med SRR bowl High Cr Core 0.467 0.012

RS-01 Road Survey PN 67:1 Gr. 9B, Ayquina Med SRR bowl High Cr Core 0.313 0.000

RS-03 Road Survey SP 13:1 Gr. 9A, Ayquina Med SRR bowl High Cr Core 0.535 0.001

T-01 Turi R141 c2 Gr. 9, Ayquina Body High Cr Core 0.592 0.002

T-02 Turi R141 c2 Gr. 9, Ayquina Interior flattened

lip bowl


T-03 Turi R141 c2 Gr. 9, Ayquina Interior flattened

lip bowl


T-04 Turi R87 c2a p. ocupacional Gr. 9, Ayquina Body High Cr Core 0.346 0.000

T-05 Turi R87 c2b p. ocupacional Gr. 9, Ayquina Thick SRR bowl High Cr Core 0.713 0.000

T-07 Turi R183 c. superficial Gr. 9, Ayquina Grooved rim bowl High Cr Core 0.829 0.000

T-08 Turi R393 c. superficial Gr. 9, Ayquina Interior flattened

lip bowl



lip bowl


T-10 Turi R87 c1, superficial Gr. 9, Ayquina Interior flattened

lip bowl


T-17 Turi R183 c. superficial Gr. 37 Grooved rim bowl High Cr Core 0.615 0.000

T-18 Turi R141 c1 basural Gr. 37 SRR bowl High Cr Core 0.602 0.000

T-19 Turi R141 c1 basural Gr. 37 Body High Cr Core 0.471 0.009

T-30 Turi R393 c. superficial Gr. 37 Small SRR bowl High Cr Core 0.396 0.000

T-31 Turi R393 c. superficial Gr. 32, Dupont Interior flattened

lip bowl


T-32 Turi R393 c. superficial Gr. 32, Dupont Body High Cr Core 0.681 0.000

T-44 Turi R523 c. superficial Gr. 32, Dupont Body High Cr Core 0.325 0.000


lip bowl


T-46 Turi R87 c3a p. ocupacional Gr. 32, Dupont SRR bowl High Cr Core 0.482 0.000

591J.R. Alden et al. / Journal of Archaeological Science 33 (2006) 575e594T-47 Turi R87 c3a p. ocupacional Gr. 32, Dupont Body High Cr Core 0.499 0.000

(continued on next page)

Appendix B (continued)


T-50 Turi R87 c2a p. ocupacional Gr. 38 Body High Cr Core 0.457 0.000

CT-51 Catarpe CT 879:4 not defined Small SRR bowl High Cr Non-Core 0.506 0.001

S-04 Solor 13 Cuadricula 2 Gr. 9, Ayquina Interior flattened

lip bowl

High Cr Non-Core 0.522 0.001

T-06 Turi R87 c2b p. ocupacional Gr. 9, Ayquina Body High Cr Non-Core 0.094 0.001

T-58 Turi R141 c1 basural Exotic, Hedionda Body High Cr Non-Core 0.240 0.001

T-59 Turi R141 c1 basural Exotic, Hedionda Body High Cr Non-Core 0.342 0.009

T-62 Turi R273 c1b basural Gr. 36, Inka Escudilla (?) High Cr Non-Core 0.466 0.000

T-65 Turi R273 c2a basural Gr. Pasta

con mica

Body High Cr Non-Core 0.295 0.000

B-02 Beter 3 Nivel 12 Gr. 9, Ayquina Bowl rim Low Cr Core 0.027 0.363

B-03 Beter 3 Nivel 17 Gr. 9, Ayquina Bowl rim Low Cr Core 0.000 0.943

B-04 Beter 3 Nivel 3 Gr. 32, Dupont Bowl rim Low Cr Core 0.001 0.486

B-07 Beter 3 Nivel 10 Gr. 40 Doble cuerpo jar Low Cr Core 0.003 0.359




CT-19 Catarpe CT 742:8 Gr. 38 Aribaloid jar Low Cr Core 0.000 0.437



CT-24 Catarpe CT 876:8 Gr. 1? Aribaloid jar Low Cr Core 0.000 0.106

CT-31 Catarpe CT 898:2 Gr. 9A, Ayquina Grooved rim bowl Low Cr Core 0.001 0.519

CT-33 Catarpe CT 849:3 Gr. 9B, Ayquina Interior flattened

lip bowl

Low Cr Core 0.000 0.541

CT-34 Catarpe CT 861:2 Gr. 9B, Ayquina Interior flattened

lip bowl

Low Cr Core 0.000 0.083

CT-36 Catarpe CT 750:8 Gr. 31? Curving flared

rim jar

Low Cr Core 0.001 0.122

CT-39 Catarpe CT 847:11 Gr. 38 Sinuous-neck jar Low Cr Core 0.010 0.581

CT-40 Catarpe CT 850:1 Gr. 37 Large other bowl Low Cr Core 0.012 0.459

CT-42 Catarpe CT 895:1 Gr. 38 Doble cuerpo jar? Low Cr Core 0.000 0.674

CT-43 Catarpe CT surface:1 Gr. 38 Tall flaring

rim cup

Low Cr Core 0.009 0.432

CT-46 Catarpe CT 878:9 Gr. 32, Dupont Small SRR bowl Low Cr Core 0.002 0.693

CT-48 Catarpe CT 765:4 Gr. 9A, Ayquina Small SRR bowl Low Cr Core 0.006 0.506

CT-50 Catarpe CT 876:3 Gr. 9A, Ayquina Med SRR bowl Low Cr Core 0.001 0.448

CT-52 Catarpe CT 885:5 Gr. 9A, Ayquina Small SRR bowl Low Cr Core 0.005 0.529

CT-56 Catarpe CT 862:4 Gr. 2 Med flaring

rim jar

Low Cr Core 0.000 0.487


rim jar

Low Cr Core 0.000 0.415


rim jar

Low Cr Core 0.000 0.325


rim jar

Low Cr Core 0.000 0.617

S-03 Solor 13 Cuadricula 2 Nivel 12 Gr. 9, Ayquina Body Low Cr Core 0.002 0.478

T-11 Turi R183 c. superficial Gr. 51 Body Low Cr Core 0.000 0.521

T-15 Turi R141 c1 basural Gr. 38 Body Low Cr Core 0.000 0.510




T-23 Turi R141 c1 basural Gr. 2 Plastered jar body Low Cr Core 0.000 0.405



T-34 Turi R393 c. superficial Gr. 1 Large club

rim bowl/basin

Low Cr Core 0.000 0.979



T-49 Turi R87 c2a p. ocupacional Gr. 38 Ledge rim bowl Low Cr Core 0.000 0.064

T-52 Turi R500 c2 p. ocupacional Gr. 1 Body Low Cr Core 0.000 0.878

T-53 Turi R87 c2a p. ocupacional Gr. 2 Body Low Cr Core 0.000 0.471

T-54 Turi R87 c2a p. ocupacional Gr. 2 Body Low Cr Core 0.000 0.542

T-55 Turi R332 p. de chullpa Gr. 51 Body Low Cr Core 0.003 0.514

592 J.R. Alden et al. / Journal of Archaeological Science 33 (2006) 575e594T-56 Turi R332 p. de chullpa Gr. 51 Body Low Cr Core 0.001 0.878

Appendix B (continued)


T-63 Turi R273 c1b basural Gr. 38 Body Low Cr Core 0.001 0.512

V-01 Vilama N12/W16 surface Gr. 9, Ayquina Bowl rim Low Cr Core 0.001 0.456

V-02 Vilama N4/E0 surface Gr. 9, Ayquina Bowl rim Low Cr Core 0.001 0.596

V-03 Vilama Cuadricula 2, Nivel 2 Gr. 32, Dupont Bowl rim Low Cr Core 0.000 0.156

V-04 Vilama N4/W24 surface Gr. 32, Dupont Bowl rim Low Cr Core 0.000 0.415

V-06 Vilama Recinto 20 Gr. 38 Body Low Cr Core 0.000 0.462

B-01 Beter 3 Nivel 3 Gr. 9, Ayquina Bowl rim Low Cr Non-Core 0.001 0.345

CT-23 Catarpe CT 870:1 Gr. 36 Aribaloid jar Low Cr Non-Core 0.000 0.249

CT-25 Catarpe CT 880:3 not defined Aribaloid jar Low Cr Non-Core 0.000 0.989


rim jar

Low Cr Non-Core 0.012 0.121

T-26 Turi R393 c. superficial Gr. 7 Body Low Cr Non-Core 0.000 0.237

T-39 Turi R273 c. superficial Gr. 36 Everted rim bowl Low Cr Non-Core 0.016 0.310

T-61 Turi R273 c2a basural Gr. 36, Inka Body Low Cr Non-Core 0.000 0.526


rim jar

Extreme Cr Assigned 0.000 0.000


rim jar

Extreme Cr Assigned 0.000 0.000

S-05 Solor 13 Cuadricula 2 Gr. 38 Aribaloid jar (?) Extreme Cr Assigned 0.000 0.000

T-29 Turi R393 c. superficial Gr. 38 Body High Co Assigned 0.179 0.000

T-33 Turi R393 c. superficial Gr. 1 Body High Co Assigned 0.059 0.001

T-35 Turi R141 c1 basural Gr. 1 Large club

rim bowl/basin

High Co Assigned 0.018 0.019

T-36 Turi R141 c1 basural Gr. 1 Large flaring

rim jar

High Co Assigned 0.380 0.000

T-40 Turi R141 c2 basural Gr. 30 Body High Co Assigned 0.151 0.000

T-42 Turi R141 c2 basural Gr. 30 Flaring rim jar High Co Assigned 0.319 0.003

CT-07 Catarpe CT 751 Exotic, Yavi Body Low Na Assigned 0.000 0.001


CT-11 Catarpe CT 818:14 Exotic, Altiplano Sm bowl w/ red painted lip Low Na Assigned 0.006 0.001

CT-14 Catarpe CT 841:3 Exotic, Yavi Black-on-red flaring

rim jar

Low Na Assigned 0.026 0.000

CT-15 Catarpe CT 849:2 Exotic, Altiplano Bent-wall bowl Low Na Assigned 0.001 0.000

CT-16 Catarpe CT 850:2 Exotic, Yavi Black-on-red painted Low Na Assigned 0.005 0.001


T-43 Turi R56 c. superficial Exotic, Yavi Body Low Na Assigned 0.000 0.000

T-60 Turi R87 c. superficial Exotic, Hedionda Body Low Na Non-Core 0.000 0.270

B-05 Beter 3 Nivel 12 Gr. 32, Dupont Bowl rim Mixed Assigned 0.050 0.236

CT-26 Catarpe CT 762:1 Gr. 9B, Ayquina Grooved rim bowl Mixed Assigned 0.447 0.058

RS-02 Road Survey PN 155:1 Gr. 1 Med SRR bowl Mixed Assigned 0.114 0.437

S-01 Solor 13 Cuadricula 2 Nivel 1 Gr. 9, Ayquina Body Mixed Assigned 0.175 0.245

T-20 Turi R183 c. superficial Gr. 36 SRR bowl w/ nubbin Mixed Assigned 0.140 0.062

T-37 Turi R141 c1 basural Exotic, Hedionda Body Mixed Assigned 0.061 0.092

CT-21 Catarpe CT 774:3 Gr. 1 Aribaloid jar Unclassed Assigned 0.000 0.045

S-02 Solor 13 Cuadricula 2 Nivel 2 Gr. 9, Ayquina Body Unclassed Assigned 0.000 0.019

T-12 Turi R183 c. superficial Gr. Pasta

con mica

Body Unclassed Assigned 0.001 0.001

T-13 Turi R183 c. superficial Gr. 38 Body Unclassed Assigned 0.004 0.008

T-14 Turi R183 c. superficial Gr. 38 Body Unclassed Assigned 0.000 0.022

T-24 Turi R183 c. superficial Gr. Pasta

con mica


T-25 Turi R141 c1 basural Gr. Pasta

con mica

Flaring rim jar Unclassed Assigned 0.000 0.000

T-48 Turi R87 c3a p. ocupacional Gr. 32, Dupont SRR bowl Unclassed Assigned 0.008 0.000

T-51 Turi R500 c2 p. ocupacional Gr. 1 Body Unclassed Assigned 0.000 0.013

T-57 Turi R26 ex. de chullpa Exotic, Hedionda Body Unclassed Assigned 0.004 0.000

T-64 Turi R273 c1b basural Gr. Pasta

con mica


T-66 Turi R273 c1b basural Gr. 51 Body Unclassed Assigned 0.039 0.000

593J.R. Alden et al. / Journal of Archaeological Science 33 (2006) 575e594

References

[1] C. Aldunate, The Inka in Tarapaca and Atacama, in: C. Aldunate,

L.E. Cornejo (Eds.), In the Footsteps of the Inka in Chile, Museo Chileno

de Arte Precolombino, Santiago, 2001, pp. 18e33.[2] C. Aldunate, Tawantinsuyu dominion over Turi, in: C. Aldunate,

L.E. Cornejo (Eds.), In the Footsteps of the Inka in Chile, Museo Chileno

de Arte Precolombino, Santiago, 2001, pp. 39e43.[3] D. Arnold, Ceramic Theory and Cultural Process, Cambridge University

Press, Cambridge, 1985.

[4] D. Arnold, Andean ceramic technology: an ethnoarchaeological perspec-

tive, in: I. Shimada (Ed.), Andean Ceramics: Technology, Organization,

and Approaches, MASCA Research Papers in Science and Archaeology,

Supplement to Vol. 15, University of Pennsylvania Museum of Archae-

ology and Anthropology, Philadelphia, 1998, pp. 353e367.

[5] R. Bishop, M.J. Blackman, Instrumental neutron activation analysis of

archaeological ceramics: scale and interpretation, Accounts of Chemical

Research 35 (2002) 603e610.

[6] T.L. Bray, Inca pottery as culinary equipment: food, feasting, and gender

in imperial state design, Latin American Antiquity 14 (2003) 1e22.[7] T.L. Bray, L.D. Minc, M.C. Ceruti, J.A. Chavez, R. Perea, J. Reinhard, A

compositional analysis of pottery vessels associated with the Inca ritual

of Capacocha, Journal of Anthropological Archaeology 24 (2005) 82e100.

[8] J.W. Cogswell, H. Neff, M.D. Glascock, The effect of firing temperature

on the elemental characterization of pottery, Journal of Archaeological

Science 23 (1996) 283e287.

Technology, Organization, and Approaches, MASCA Research Papers

in Science and Archaeology, Supplement to Vol. 15, University of Penn-

sylvania Museum of Archaeology and Anthropology, Philadelphia, 1998,

pp. 283e312.[13] M.D. Glascock, Tables for Neutron Activation Analysis, University of

Missouri, Columbia, 1991.

[14] M.D. Glascock, Characterization of archaeological ceramics at MURR

by neutron activation analysis and multivariate statistics, in: H. Neff

(Ed.), Chemical Characterization of Ceramic Pastes in Archaeology,

Monographs in World Archaeology No. 7, Prehistory Press, Madison,

WI, 1992, pp. 11e26.

[15] M.D. Glascock, H. Neff, Neutron activation analysis and provenance re-

search in archaeology, Measurement Science and Technology 14 (2003)

1516e1526.

[16] F. Hayashida, New insights into Inka pottery production, in: I. Shimada

(Ed.), Andean Ceramics: Technology, Organization, and Approaches,

MASCA Research Papers in Science and Archaeology, Supplement to

Vol. 15, University of Pennsylvania Museum of Archaeology and An-

thropology, Philadelphia, 1998, pp. 313e335.[17] R. Ixer, S. Lunt, Petrography of certain pre-Spanish pottery of Peru, in:

A. Middleton, I. Freestone (Eds.), Recent Developments in Ceramic

Petrology, 81, British Museum Occasional Papers, 1991, pp. 137e164.

[18] H.F. Kaiser, The application of electronic computers to factor analysis,

Educational and Psychological Measurement 20 (1960) 141e151.

[19] T.F. Lynch, L. Nunez, Nuevas evidencias inkas entre Kollahuasi y Ro

Fro (I y II Regiones de Chile), Estudios Atacamenos 11 (1994) 145e

164.

594 J.R. Alden et al. / Journal of Archaeological Science 33 (2006) 575e594[9] C.L. Costin, Production and exchange of ceramics, in: T.N. DAltroy,

C.A. Hastorf (Eds.), Empire and Domestic Economy, Kluwer Academic/

Plenum, New York, 2001, pp. 203e242.

[10] T.N. DAltroy, State goods in the domestic economy: the Inka ceramic

assemblage, in: T.N. DAltroy, C.A. Hastorf (Eds.), Empire and Domes-

tic Economy, Kluwer Academic/Plenum, New York, 2001, pp. 243e264.

[11] T.N. DAltroy, R.L. Bishop, The provincial organization of Inka ceramic

production, American Antiquity 55 (1) (1990) 120e138.

[12] T.N. DAltroy, A.M. Lorandi, V. Williams, Ceramic production and use

in the Inka political economy, in: I. Shimada (Ed.), Andean Ceramics:[20] N. Marinovic, A. Lahsen, Hoja Calama, Carta Geologica de Chile No.

58, Servicio Nacional de Geologa y Minera, Santiago, 1984.

[21] L. Nunez, Cultura y conflicto en los oasis de San Pedro de Atacama, Ed-

itorial Universitaria, Santiago, 1992.

[22] V. Varela, Ensenanzas de alfareros Toconcenos: tradicion y tecnologa en

la ceramica, Chungara 34 (2) (2002) 225e252.

[23] V. Varela, M. Uribe, L. Adan, La ceramica arqueologica del sitio "Pu-

kara" de Turi: 02-TU-001, Actas del XII Congreso Nacional de Arqueo-

loga Chilena, Boletn No. 4 del Museo Regional de la Araucania, Tomo

II, Temuco, Chile, 1993, pp. 107e121.

Identifying the sources of Inka period ceramics from northern Chile: results of a neutron activation studyArchaeological backgroundNeutron activation analysisSample selection and analytical proceduresStatistical analyses: procedureStatistical analyses: resultsPrincipal components analysisCore group definitionLinking composition groups to local clay sourcesClassification of non-core group samples

DiscussionCeramic production, geology, and clay source areasDistribution of composition groups by siteDistribution of composition groups within ceramic types and vessel formsReconstructing economic patterns from the ceramic composition dataCompositions of non-local style ceramics

ConclusionsAcknowledgmentsSimulation of firing on clay sample element concentration dataCeramic sample data and compositional group assignmentsReferences

alden

Documents