INAA of ochre artifacts from Jiskairumoko, Peru

481

This paper describes the analysis of 65 archaeological ochres from differentcontexts from the Terminal Archaic-Early Formative site of Jiskairumoko, Peru.Ochre is very COminon in many archaeological contexts worldwide, and also onthe site of Jiskairumoko. Instrumental neutron activation analysis (INAA) wasconducted to evaluate the heterogeneity of the elemental compositions of ochreartifacts used by residents of Jiskairumoko. An important correlation betweenthe chemical signature of artifacts and their place of manufacture is the basis formany artifact provenance or "sourcing" studies. Knowledge of the chemicalbasis and variation of sources and of artifacts allows distinction between ancientsources, artifact geochemistry and an understanding of ancient technologies andprovides further insight into ancient cultures and behavior.

Multi-elemental analysis has been applied to ochre and related materials farless often than it has to either ceramics or obsidian and other archaeologicalmaterials. Some exceptions include: the Natufian remains from EI-Wad Cave,Israel (1); modern ochre source sampling in Australia (2); Mousterian deposits atQafzeh Cave, Israel (3); and Paleo-Indian contexts at the Powars II site,Wyoming (4). While the earlier literature often quantifies elemental values fromochre analysis, the variation in the samples and sources is not often discussed orquantified. Geochemical variance has to be understood and quantified tointerpret the variation in a given source or site. In general, the variance in onlythe major elements such as Ba, Si, Mn and others was used to addressdifferences between sources or artifacts. Popelka-Filcoff (5), Erlandson (6), andMrzlack (7) have taken· initial steps to measure elemental variation, especiallytrace elemental variation in ochre. Efforts are taking place to expand thedatabase of chemical analysis, especially trace element data for ochres from theUnited States and. around the world to improve ochre and iron oxidecharacterization for archaeological questions. The ochre analysis fromJiskairumoko provides information on chemical variability in the ochre used atthis site and also contributes to a worldwide collection of analytical data onochre.

Jiskairumoko-a Late-Terminal Archaic Village

Jiskairumoko is located in the Lake Titicaca basin (Figure I), which is animportant region for understanding cultural change. Some of the earliestevidence for cultural complexity in the Andean highlands is found in thenorthern basin, at the Formative site of Pucara, and in the Tiwanaku Empire thatdeveloped in the southern basin during the Middle Horizon (8-10). Despiteprolonged archaeological research in the region, comparatively little attentionhas been paid to early prehistoric time periods. Excavations at Jiskairumoko

II

. I

483(Figure 2) represent the first systematic study ever conducted in the LakeTiticaca Basin of an open air Archaic Period residential site. The site is locatedin the Rio IIave drainage in thesouthwestem basin. Jiskairumoko was excavatedin natural layers employing a decapage approach (11-13), and horizontalexposures were carefully documented with digital photography and geographicinformation systems for site level photomapping.

Excavation results indicate that a shift from mobile to increasingly sedentarylife-ways occurred in the Rio IIave drainage about 3300 BC, when people appearto have begun living in pithouses(13). During this time, a low-level food !i

producing economy, c.r. (14), involving use of domesticated plants and animals,developed and lasted until about 1300 BC. X-ray fluorescence (n=63) (5) andportable X-ray fluorescence (PXRF) (n=550) (16) geochemical sourcing ofobsidian from the Rio Have, Rio Huenque, Rio Ramis and Rio Azangarodrainages demonstrates a highly redundant reliance on the Chivay source andimplies the existence of regular trade networks (17). Gold beads were discoveredin a burial that consisted of an adult and a child, and radiocarbon dated to about2100 BC. About 1300 BC, residential architecture at the site shifted tooccupation of above ground dwellings, which are associated with the appearanceof ceramics. Ochre was found within a wide variety of contexts spanning theapproximately two thousan~ years of the site's intermittent, but recurrent useduring the Archaic to Formative transition.

A comprehensive analysis of temporally diagnostic projectile points fromsurvey data indicates that use of obsidian increased significantly during the endof the Late Archaic and during the Terminal Archaic. This is the period of timeduring which came lids appear to have been domesticated in the region (18-27).Together these data imply that the use of regular camelid caravans thattransported goods over long distances developed on the altiplano around the endof the Late Archaic (28,29). This is the time period during which Jiskairumoko isfirst occupied. Jiskairumoko currently represents the earliest known example ofan early village in the Lake Titicaca Basin and provides data for understandingthe origins of plant cultivation and sedentary life. The deposit is a record of thishighly significant, yet poorly understood cultural period (30). Figure 2 is a mapof the site and outlines the blocks where samples of ochre were found.

Archaeological Contexts of Ochre Encountered atJiskairumoko

To interpret the significance of the presence of ochre, one must first attemptto discern if the use of the material was practical or symbolic. Practical uses ofochre include application as a preservative in curing hides, as an adhesive forhafting stone tools, and as medicine. Even practical objects can have symbolic

Figure 2. Jiskairumoko site map.

ldinlmum •••• "' ••••• lHI.OS(m)Maximum ~ 100 (m)

20I

"'"'-10I

SI

oIhh __ Site Topography; Contour Interval 0.05 (m)o Excavated Unlla (B = Block ~ T= Trench)o stone Walls O.S (m)grid lquares

485

I ~!

!lmeaning. Most mammalian vision is dichromatic and composed of two cone ,:pigments known as Sand M (3 I). The S cone pigment is maximally sensitive toshort and the M to medium visible wavelengths. '

Some primates, including humans, have a third cone pigment, known as Lthat is maximally sensitive to long wavelengths. This type of vision is calledtrichromacy. L cone pigments may have facilitated the detection of red fruitagainst a green foliage background (32-34), or the detection of new red foliage(35). Red is one of the three most fundamental color categories present in allknown languages (36), making it a human universal(37). Thus, importance of redappears to have a biological foundation, and ochre's red color makes it a likelysymbol in many cultural settings. Symbolically, ochre has many associatedconnotations including: menstrual blood (38), blood in hunting magic ritual (4),renewal (39), potency (40), and life stage transitions (41). Symbolic meanings ofthe color red are probably multivocal (42).

Key contexts where ochre was encountered at Jiskairumokoare listed inTable I. Ochre was found inside a stone box at the base of a burial cairn at anearby Terminal Archaic (2578-2302 cal B.C.E.) site ofKaillachuro (43). Ochrewas not observed on either the working or hafting edges of any chipped stonetools recovered from Jiskairumoko or Kaillachuro, though ochre has beenobserved as a hafting adhesive on projectile points in the assemblage fromQilIqatani (J 8). None of the contexts listed in Table I suggest purely or evenlargely practical uses for ochre at Jiskairumoko.( 44)

Burned ochre fragments were found in a lens of ash on the floor of a Late "Archaic pithouse next to the structure's central hearth, which dated to 3385-3078 cal B.C.E. This thin, in situ, floor assemblage rests on the surface of thepithouse beneath an uninterrupted stratum of secondary rubbish fill that extendedacross the structure's extent. A key part of the pithouse's floor assemblage was aperpendicular debris arc that extended from the structure's central hearth. It wascomposed of ash, lithic debitage, burned camelid bone, fire altered rock, and alarge concentration of thermally altered ochre (44). Photomapping was appliedto piece plot the assemblage. The debris arc was ~xplored using multivariatestatistics, and defined using raster based unconstrained clustering.

These analyses revealed a perpendicular arc of debris expected for habitualhearth-side workspaces (45). The hearth-associated scatter found on the floor ofthis Late Archaic pithouse provides clear evidence that ochre was thermallyprocessed to change its color. Further it shows that thermal processing of ochreoccurred in a domestic context though it was primarily used in symbolic andritual contexts. At Jiskairumoko, preparation of ochre for symbolic and ritualpurposes appears to have been embedded in the, domestic setting. Theoccupation bears early evidence for prestige goods in burial contexts. Yet, ochrepreparation does not appear to have occurred in specialized architecture. Socialdifferentiation was taking place, but preparation of ritual materials took place indomestic contexts.

486

Table I. Contexts, Dates and Associations for Ochre at Jiskairumok

Ochre context 14C DateAssociationsB.C.E.

Drop zone from heat 3385-3078 Clay lined hearth inside atreatment in Pithouse 1 structure not far from deer

bones.1 palette and 1 abrader 2473-2119 On edge of and next to Pithouse

2072-1878

Ground dust at base of 1883-1680 Older adult female buried witha~Burial 3 lapstoneOchre stained manos No direct Unsexed adult buried withassociated with Burial 4 dates burned and unburned faunaland ochre stained lapstone remains from at least twofrom just above burial. individuals one adult and one

juvenileGround dust at base of No direct Individual buried with numerousBurial 5 dates red chert flakes placed at distal

end of intennent.

External ochre stain outside No direct Rock-soil feature and split rockRectangular Structures I dates altarand 2

Ochre stained animal bones No direct Rock-soil feature and split rock.dates altar

Ochre stained manos, No direct Rock pavement, perhaps relatedgroundstone fragments dates to cookingfound in rock pavement.

Ochre stained animal bones and a large ochre stain were found: in a rock-soil feature, near a split rock altar, associated with a concentration of obsidiandebitage, near early ceramic fragments, all located outside one of Jiskairumoko'slater above ground structures. This context for ochre decorated call1elid bonesimplies economic dimensions to ochre's symbolic meaning for Jiskairumoko'sEarly Fonnative occupants. Additionally, ochre stained manos and stone palletswere recovered from many burial and refuse contexts at Jiskairumoko. Togetherthese finds indicate that Jiskairumoko's residents used ochre as a pigment, andthat it often occurred in symbolically charged contexts like burial assemblages.

il

11'

487

Project Goals

This paper intends to accomplish several goals in the analysis of theJiskairumoko ochre. The chemical variability of ochre artifacts must bedetermined in order to evaluate its potential role in ancient trade and regionalinter-community interaction. Ideally, it would be valuable to know: if ochre wasa non~local resource, where it came from, if the source is local or not, andwhether multiple sources were used. Primary sources of ochre in the region havenot been located, making "sourcing" studies impossible; still, an exploratorystudy of ochre chemical variability is warranted for several reasons. There islittle knowledge about the chemical variability of ochre sources, reporting resultswill increase this understanding. Chemically variable groups could suggest useof either different sources or different portions of a variable source. Are therechemical differences in ochre remains recovered from different contexts ofJiskairumoko's multi-component occupation? Multiple structure types arelocated at the site and ochre is found in association with each kind. Are theseochres chemically similar or is there high variability? The site's earliestresidential context bears evidence of thermal processing of ochre. There is alsoochre in later secondary refuse heaps. Is there large variability in the ochresfound in the thermal processing context? Some have suggested that the color ofochre, (perhaps a result of thermal treatment) is independent of geographiclocation (46). Do the ochre samples found in this early residential context appearchemically similar to ochres recovered in secondary refuse contexts? Byunderstanding the chemical composition of ochre in different contexts,variability in use and procurement of the material can be explored.

Ochre Sampling

A total of 65 ochre samples recovered from a range of excavated contexts atJiskairumoko were analyzed by INAA (Table II). Ochre samples were recoveredas both singular lumps and powdered material. Only ochre masses were analyzedto prevent any possible contamination issues. Samples ranged in color fromyellow to dark brown, with varying shades of red and red-brown in between.These color variations may indicate differing iron concentrations or particlesizes, and may be due to inherent impurities in the mineral (47) or heattreatment. Table II lists the examined samples with Munsell values.

Samples were obtained from pithouse floor assemblages, the perpendiculardebris arc found next to the hearth of a Late Archaic pithouse, later expedienthearth features situated in secondary pithouse fill, secondary debris fill inpithouses, from along the edges of later rectangular structures, and from anexterior ritual area. Future studies will include a sample of archaeological ochrerecovered from each of the burial contexts where it was encountered.

~Table II. Sampled Ochre Characteristics: Block, Provenance, Level, Context and Munsell Color 0000ID Block Provenance Context Munsell Color305 3 X35a/7NIIIa Pithouse fill 2.5 YR 3/4 Dark red-

brown306 3 Y36c17NIIIa Pithouse fill 2.5 YR 5/6 Red307 3 Y34a/oaNII Matrix lOR 3/6 Dark red308 . 3 W34b/5NII Pithouse Fill 2.5 YR 4/8 Red309 3 Y35d7Niiia Pithouse External Stain 2.5 YR 6/6 Light red310 3 Y34d1oaNII Matrix 2.5 YR 2.5/4 Dark red-broWn311 4 GG24c/oa/IV Outside Rectangular Structure 1 2.5 YR 3/3 Dark red-brown312 4 FF21 clF 14NII Hardpan-Occupation Interface 2.5 YR 2.5/3 Dark red-brown313 4 KK24b/oa/IV Outside Ritual Area 2.5 YR 3/3 Darkred-brown314 4 JJ21cIFINII Hardpan-Occupation Interface 2.5 YR 2.5/4 Dark red-brown315 4 II21dIFINII Hardpan-Occupation Interface 2.5YR 3/4 Dark red-brown316 6 KK-25- Outside Ritual Area 2.5 YR 3/6 Dark redCIF3/III

317 6 LL-3IBIF6/III Rectangular Structure 2 Edge 10 YR 4/8318 6 MM32A/Oa/II Outside Rectangular Structure 2

I

319 6 LL28d/HFI/III Rectangular Structure 2 Edge 2.5 YR 2.5/3 Dark red-brown

320 6 LL28c/ob/IV Rectangular Structure 2 Edge 2.5 YR 3/4 Dark red-brown

321 6 LL27c/ob/IV Rectangular Structure 2 Edge 2.5 YR 2.5/3 Dark red-brown

322 6 LL25d/ob/IV Outside Ritual Area 5 R 3/2 Dark brown323 8 P22a1 1O/IV Pithouse floor 10 R 3/3 Dusky red324 8 Q23c/l0/IV Pithouse Occupation Surface 2.5 YR 3/4 Dark red-

brown325 8 P23a1IO/IV Pithouse Occupation Surface 10 R 3/2 Dusky red326 9 KK21D/Fl/IV Late Archaic Pithouse Fill lOR 5/2 Weak red327 9 x26c/F8/x Late Archaic Pithouse Fill 2.5 YR 3/3 Dark red-

brown328 9 X26B/F8/X Late Archaic Pithouse FIll 7.5 YR 3/2 Dark brown329 9 X28d/F8/X Late Archaic Pithouse Lower 7.5 YR 3/2 Dark brown

Fill330 9 Y28a1F8/x Late Archaic Pithouse Fill 10 R 3/2 Dusky red331 9 Y28a1F6/IX Late Archaic Pithouse Fill 10 R 3/2 Dusky red·332 9 Y28b/F6/IX Late Archaic Pithouse Fill 7.5 YR 3/1 Very dark

gray333 9 Y28c/F6/IX Late Archaic Pithouse Fill 7.5 YR 3/2 Dark brown334 9 Y28d/F6/IX Late Archaic Pithouse Fill 7.5 R 3/4 Dark brown335 9 Y28d/F6/IX Late Archaic Pithouse Fill 7.5 R 4/1 Dark gray336 9 Y27a1FI/ii Plowzone 7.5 R 4/3 Brown

Continued on next page. 01:0.co\C

"'"Table II. Continued.\0QlD Block Provenance Context Munsell Color337 9 X26c/F3/X Late Archaic Pithouse Fill 2.5 YR 3/3 Dark red-

brown338 9 AA/28c1F2/IX Hearth in Upper Late Archaic 10 R 3/2 Dusky redFill

339 9 X28c/FI 2/IX Late Archaic Pithouse Fill 10 R 3/3 Dusky red340 9 Y28d1F8/X Late Archaic Pithouse Fill 10 R 3/2 Dusky red .341 9 Z28b/F3/IX Late Archaic Pithouse Fill 10 R 3/2 Dusky red342 9 Y27/F7/vii Late Archaic Pithouse Fill 10 R 3/2 Dusky red343 9 X26d1F9/vii Late Archaic Pithouse Fill 7.5 R 3/6 Dark brown344 9 Y27b/F7/vii Late Archaic Pithouse Fill 7.5 R 3/4 Dark brown345 9 Y27a/F7/vii Late Archaic Pithouse Fill 7.5 R 3/2 Dark brown346 9 BB25dIF l/ii Plowzone 2.5 YR 3/3 Dark red-brown347 9 A25c/F Vii Plowzone 5 R 3/4 Dark brown348 9 BB25b/F l/ii Plowzone 10 R 3/4 Dusky red349 9 Y25b/F Vii Plowzone 7.5 R 3/2 Moderate redbrown350 9 AA25b/Fl/ii· Plowzone 7.5 R 3/2 Moderate redbrown351 9 BB24d1Fl/ii Plowzone 2.5 YRJ/3 Dark red-brown352 9 AA28a/F l/ii Plowzone 7.5 R 3/2 Moderate redbrown_353 9 X26c/F8/viii Late Archaic Pithouse Fill ~atekr¢d;"

brown354 9 Z27b/F7/IX Late Archaic Pithouse Fill lOR 3/4 Dusky red355 9 Y27c/F8/viii Late Archaic Pithouse Fill Near 7.5 R 3/4 Dark browna Small Hearth

356 9 BB28a/F7/v Hearth in Upper Late Archaic 7.5 R 3/6 Dark brownFill

357 9 Z28c/F4/IX Hearth in Upper Late Archaic 7.5 R 3/2 Dark brownFill

358 9 AA26c/Fl/ii Plowzone 7.5 YR4/4 Dark brown359 9 AA24c/F2/iii Plowzone 10 R 4/2 Dusky red361 9 X24d1oa/v Late Archaic Pithouse Fill 10 R 3/2 Dusky red362 9 AA27b/F8N Late Archaic Pithouse Fill 10 R 3/2 Dusky red363 9 Y25d/F9/viii Outside Stain (Shallow) 7.5 R 2.5/4 Dark brown364 9 Z27 c/F2/IX . Hearth in Upper Late Archaic 10 R 3/4 Dusky redFill

365 9 BB27d1oa/vii Matrix 10 R 3/3 Dusky red366 9 X26B/F9/xiv Hearth in Base of Late Archaic 7.5 R 2.5/4 Dark brownPithouse

367 9 Y27bFll/XIII Hearth in Base of Late Archaic 10 R 3/4 Dusky redPithouse

368 9 Y27d/FII/XII Hearth in Base of Late Archaic 10 R 3/3 Dusky redPithouse

369 9 Y27d1FIIlXII Hearth in Base of Late Archaic lOR 3/4 Dusky redPithouse

370 9 AA24c/F2/iii Plowzone 7.5 YR 4/4 Dark brown.j:o\C-

492

Experimental

The INAA analysis followed standard analysis parameters used ,inArchaeometry Lab at the University of Missouri Rese'Reactor (48). Each sample was broken into smaller pieces with a rock ha'The crumbled sample was then pulverized into a powder using a Brazilian a 'mortar and pestle. The powdered ochre material was dried at 100 ·C ovembefore preparation for INAA. For both the short and long irradiatiapproximately 60 mg of sample was used. In the short irradiation, the sampwere irradiated in 1.2 mL high-density polyethylene vials for five seconds atthermal tlux of approximately 8.0 x 1013 neutrons cm-2 S-I. After a decay of'minutes, the samples were counted for 720 seconds on a high resolution HPQdetector. For the mid and long irradiations, the samples were sealed in higpurity quartz vials and irradiated for 24 hours at a thermal neutrontlux dapproximately 5.2 x 1013 neutrons cm-2

S-I. The sample was allowed to decay forseven days, and the "mid" count data was acquired for 2,000 seconds onautomated sample changers. The samples were allowed to decay an additionalthree weeks, and the "long" count data were acquired for 10,000 seconds. Thecomparator standards used in the INAA measurements were NIST SRM 1633a(Fly Ash) and SRM 688 (Basalt), and the quality control standards were NISTSRM 278 (Obsidian Rock) and Ohio Red Clay.( 48) The elemental data ispresented in Table III.

Mathematical and Statistical Treatment of Data

The raw concentration data were subjected to several mathematical andstatistical transformations. The ratio of the element of interest to Fe helps offsetinherent variation in Fe across the data set. A loglo transform is a standardstatistical conversion for elemental data. This transformation reduces the"weighting effect" from very small to very large concentrations in the data (48).More details on the calculations and reasoning behind these transforms can befound elsewhere (5).

Although INAA can routinely measure approximately 30 elements by themethods described, several elements were below the detection limits, or wereotherwise unreliable elements for ochre. For this study, 16 elements were used:As, Ce, Co, Cr, Dy, Eu, La, Mn, Nd, Sb, Sc, Sm, Sr U, V, Yb and Zn. Theseelements are similar to other ochre studies as those related to the "Fe-oxidesignature" and not the surrounding minerals (5).

'.J.

'III

493Statistical Operations

An initial study of the data was undertaken with a cluster analysis, to outlinepossible clusters and groups within the elemental data. this was performed withthe Clustan software (ClustanGraphics, Edinburgh, Scotland). The clusteranalysis included the use of a hierarchical tree diagram to display the results.Distances were calculated using a squared Euclidean distance. The linkagebetween groups was calculated by the increased sum of squares. The results ofthe cluster analysis outlined five distinct groups, henceforth referred to asGroups I, 2a, 2b, 3a, and 3b.

A principal components (PC) analysis was also performed to evaluate theelements in the data set that contributed to the variance. PC plots graphicallyindicate a linear combination of original variables, oriented in the direction ofgreatest variance. PC space also displays the elements with the greatest variationby graphically displaying them with the longest vectors.

In addition, a canonical discriminant analysis (CDA) was performed on thegroups defined by the cluster analysis. This statistical procedure was performedto evaluate group differences as defined by the cluster analysis. CDA analysisassumes that the groups are different and calculates the largest differencebetween the groups (48).

Results of Sample Variance

The concentrations of Fe ranged from as low as 4 weight percent to as highas 67 weight percent, with the majority of the samples in range from 35-50 %Fe. The mathematical transform of the elemental ratio to Fe helped to normalizethe data for this range of Fe concentration in the samples. After the mathematicaltransform and PC analysis, several observations were made. Principalcomponents I thru 8 described 95% of the total variance for the dataset. Figure 3is a plot Qf PC2 vs PC4 as an example of one of several possible permutations.On this plot, the longest vectors represent the most variance in this PC space.Elements with the longest vectors in the various permutations of PC bivariateplots in total included: Co, Mn, Zn, Eu, Sm, Ce and La. Bivariate plots of theseelements were explored to ascertain which pairs of elements could be used tovisually describe group associations.

The cluster analysis identified five discrete groups based on the trace elementchemistry of the selected ochre samples. These five groups were analyzed severalways to discern possible relationships between the variation in the elementalcomparison of the artifacts and possible archaeological significance.

Bivariate plots based on archaeological context were examined to investigatepotential spatio-temporal contextually specific groupings (Figures ~).

"'"1,0

"'"Table III. Elemental Data from the Jiskairumoko OchresID As Ce Co Cr Dy La Mn Nd Sb Sc Sm Sr U V Yb Zn

305 449.2 166.0 57.7 12.5 1.1 169.5 1726.5 53.3 15.5 9.1 4.0 1778.4 5.4 311.9 0.6 193.5306 392.3 128.8 5.7 34.0 2.3 130.5 176.3 40.3 27.4 12.9 4.2 893.5 6.6 240.9 1.7 121.3307 191.3 273.6 5.2 9.7 60.7 298.9 200.3 56.1 8.0 8.5 14.7 3530.9 18.2 262.2 31.4 60.0308 32.8 52.0 1.1 12.8 1.4 35.2 581.2 14.2 1.1 18.3 2.1 1021.6 2.6 321.1 1.7 84.1309 9.2 93.0 6.7 63.8 2.7 53.3 226.1 35.4 0.9 12.6 5.1 652.5 5.1 115.9 1.7 116.0310 734.7 34.5 9.1 31.7 3.4 33.0 538.9 18.9 3.8 4.2 3.3 0.0 2.7 390.9 1.7 313.0311 81.3 161.6 1.8 3.0 1.1 221.4 58.8 26.2 4.6 0.9 2.6 842.5 5.9 60.8 0.3 11.3312 297.1 171.4 14.3 8.9 1.6 127.5 1391.3 34.8 9.0 3.4 5.3 509.7 9.7 327.5 0.7 81.0313 62.8 54.3 5.6 9.7 0.9 49.5 396.8 0.0 3.2 3.7 2.5 704.9 9.1 69.3 0.4 42.3314 163.6 113.9 5.9 6.4 0.9 103.2 919.7 47.3 10.4 1.2 3.9 0.0 17.0 126.1 0.0 29.0315 92.3 5.2 0.6 5.5 0.2 5.5 36.4 0.0 3.5 1.2 0.1 148.6 0.0 74.9 0.0 0.0316 292.4 172;9 14.0 7.9 1.6 137.8 590.2 22.1 12.6 3.5 3.6 1277.8 6.5 337.6 0.7 62.6317 106.1 225.8 1.2 164.6 1.2 285.1 65.0 32.4 5.7 6.2 3.8 3699.9 1.3 178.4 0.0 17.4318 63.6 11.7 0.2 2.3 0.0 19.2 17.2 10.9 5.8 0.3 0.4 776.1 1.2 12.8 0.0319 126.5 523.5 0.9 7.4 0.6 773.8 56.5 158.0 1.2 1.1 2.8 3641.9 42.3 0.7320 280.8 567.5 14.6 8.2 2.4 508.8 843.8 96.9 14.6 1.8 9.4321 68.8 30.9 1.5 2.5 0.0 35.8 56.2 0.0 4.4 0.8

322 69.2 172.7 566.9 5.0 0.0 135.3 4773.6 28.3 4.0 1.4 4.3 893.5 9.0 97.3 0.2 51.7323 79.4 67.3 20.1 7.0 0.4 69.6 1491.6 16.6 4.6 4.2 1.5 1587.3 1.9 94.2 0.3 18.3324 223.3 202.3 7.8 9.1 1.3 215.0 366.8 21.3 8.8 5.1 4.5 1589.8 14.3 263.6 0.9 57.9325 200.5 201.5 11.6 8.6 1.1 173.6 838.5 41.6 10.6 2.1 4.6 994.2 14.9 154.2 0.7 27.1326 108.8 77.6 1.7 53.2 0.0 117.7 71.5 67.5 2.2 3.6 1.3 1186.2 3.1 141.5 0.0 17.5327 132.3 274.5 7.5 12.0 2.9 232.7 245.4 53.3 5.9 8.2 8.3 1245.1 7.5 166.1 0.9 39.5328 94.9 190.3 4.3 11.0 2.4 154.9 331.2 52.5 4.4 4.3 6.1 962.5 5.8 121.0 0.7 26.0329 40.3 70.0 14.2 21.4 0.7 50.4 1212.8 21.0 1.5 8.8 2.5 1010.2 3.6 148.4 0.8 82.8330 100.2 150.2 1.4 10.3 0.6 168.1 165.1 25.5 3.0 5.2 1.7 2680.3 3.7 185.1 0.0 0.0331 192.6 326.9 186.4 3.5 0.0 290.4 12144.8 55.5 15.4 1.4 8.8 0.0 18.1 190.4 0.0 22.6332 102.2 111.2 4.1 11.3 0.8 100.6 266.9 0.0 2.5 8.3 2.3 1755.4 4.8 191.8 0.4 0.0333 41.9 76.5 1.1 20.9 1.2 68.9 151.4 14.1 2.1 14.9 2.1 1515.6 4.2 162.0 0.7 16.2334 93.1 210.1 4.0 7.1 2.1 172.6 187.3 53.6 4.9 6.6 6:8 1869.4 10.6 161.4 0.8 27.6335 92.7 234.8 1.4 8.4 3.6 169.0 93.5 65.1 3.9 4.1 9.7 348.2 16.6 127.4 1.2 8.9336 226.9 153.0 4.8 16.5 0.7 157.3 817.2 35.9 13.4 5.5 3.4 152.1 6.4 279.0 0.6 24.5337 171.6 402.2 12.1 16.0 2.8 414.5 344.8 69.2 8.0 6.2 7.4 574.3 9.3 228.2 1.0 132.8338 108.0 133.7 2.7 11.7 0.6 146.2 178.9 27.9 4.1 6.0 2.4 2416.5 3.6 124.7 0.3 18.0339 137.3 230.7 3.2 15.7 1.1 235.8 202.7 39.8 4.8 3.5 2.9 2444.2 3.9 138.2 0.2 9.0340 69.3 93.3 6.8 10.8 1.2 93.2 512.0 21.4 2.7 6.2 2.2 1808.9 4.3 275.0 0.4 60.7

Continued on next page.""\0Ul

359 132.8 265.4 3.5 8.6 2.3 267.2 218.3 69.5 6.0 5.0 7.8 1008.6 8.6 208.0 0.9 0.0361 107.6 101.3 20.5 43.1 1.9 72.3 1917.7 18.1 4.6 5.2 4.1 525.5 6.8 143.3 1.0 58.8362 181.5 167.2 9.5 28.2 ' 2.0 177.7 351.1 37.3 9.3 6.3 4.2 1276.2 6.7 130.4 1.2 52.3363 110.6 264.7 101.3 8.9 1.8 257.6 855.4 65.6 5.4 3.6 7.1 2103.2 6.1 132.7 0.6 11.9364 81.6 166.5 5.3 10.5 2.7 142.2 306.4 50.7 3.4 5.1 7.2 813.2 3.6 105.9 1.2 40.5365 402.5 1707.2 3.6 12.1 16.3 1858.3 262.3 313.6 22.5 l.l 50.4 339.6 93.5 650.1 8.9 0.0366 61.2 86.4 3.0 25.9 0.8 88.9 233.6 30.7 3.0 4.3 3.2 1286.7 3.4 110.2 0.4 20.8367 138.3 159.1 14.0 23.0 2.2 120.4 527.1 27.8 6.2 9.0 5.0 337.6 8.1 158.8 l.l 138.4368 143.5 187.0 3.0 6.2 1.8 165.8 182.3 49.1 6.5 3.2 6.6 343.5 8.6 123.3 0.5 15.4369 139.4 174.9 8.7 14.9 1.3 136.5 1219.1 37.8 6.3 6.5 4.9 826.4 6.6 183.4 0.6 85.2370 172.9 46.9 1.4 6.8 0.5 67.8 104.7 7.9 7.8 1.7 1.5 785.0 4.0 64.7 0.4. 0.0371 69.2 14.5 1.5 7.6 0.3 5.5 139.9 11.3 144.3 1.5 1.7 757.4 1.8 0.0 0.4 25.2372 80.3 57.3 11.7 30.9 2.4 37.1 764.8 31.0 4.9 11.4 4.1 840.3 3.4 110.8 1.5 80.2373 72.6 0.0 1441.4 9.4 0.3 0.7 291.6 0.0 I.I 2.4 0.3 1111.5 0.0 159.0 0.4 33.2

0.0

o

-51>

o

-1.0 -0.5

o Hordpono H<>.rth" Late Arch.ic Plthouse fill+ Matrix

o Outsidev Plthouse

)( Plowmne• ReClOngula, Structure

o

"

-1.5-2.0

I09,o[Co/Fe]

-2.5

+ V

-3.0

+

o

-3.5

..··0.2;) ·..·0.20 ....O.l~) ·O.lO "0.05 "'0.00 0.05 O.~O 0.15 0.20

.

~~--'----r---....,..----r-----'r----T;;""---r--_....,.. --..I

('oj

NI

PC4

"'.I

N-oj'

I -4.5 -4.0

Figure 3. Principal components plot (PC 2 vs. PC 4) demonstrating elementsthat drive variance in Jiskairumoko ochre

498

'"0

;5

(''';

NC)

UC-

o s.c'

'"~;

"0

Figure 4. Bivariate plot ofloglo [Sm/Fe] vs log 10 [Co/Fe). Samples are plottedby groups as determined by context.

il"!'I

II'" i·I

(j)+p.j.

I

II .00 A KAA 00,.;

<0."~

I At;

~ 9..0::2 A 0~ ..•.<>"j 00 ".b (,Cl I

..Q x tf(,CO A"j. A

I <>

'" 0.,£. AI x

.[]

v

o Hardpan

[] HearthA L.t. An:h.icPlthou,. fm+ M.trlx

o Outsidev Plthou,eK Plowzonc• Re<:tdnglll,,, Structure

499

-0.2

K

-1.0 -0.6

if 0

-1.4-1.8-2.2-2.6

<D..•.I -3.0

10glO[Zn/Fe]

Figure 5. Bivariate plot ofloglo (Eu/Fe] vs 10gJO{Zn/Fe]. Samples are plottedby groups as determined by context ..

o Group 1A Group 2..1" Group 2b"j' Group 3.

<> Group 3b'0..j1·3,C··2.() "';.8 ....:.4 •... , .":. ·O.G Cl.:1 0.&

log lO[Mn/Fe]

,

ItIi

Figure 6. Bivariate plot of loglo {Eu/Fe] vs 10gJO{Mn/Fe]. Samples are plottedby groups as determined by a cluster analysis. Group 1 does not have enough

samples to form an ellipse.

500

")

N

/~ ,

0)\ 0 0, I

Na 0u

I

NI

"lI -4 '··3 -2 -1

CD 1o 2

<:> Group 1A Group 2a

o Group 2b+ Group 3a<' Group 3b

Figure 7. Canonical discriminant analysis plot (CD2 vs CDI) Samples areplotted by groups as determined by a cluster analysis.

Artifacts were divided into several groups based on archaeological contextand plotted according to 10glO[SmlFe] vs 10glO[Co/Fe] (Figure 4), 10glO[Eu/Fe] vs10glO[ZnlFe] (Figure 5). The third elemental bivariate plot was groupedaccording to cluster analysis and plotted by 10glO[Eu/Fe) vs 10glO[MnlFe] (Figure6). Figure 6 indicates visual grouping of the samples in an elemental plot, withsome samples associated in a large group, but other clusters are both visually andstatistically distinct.

Canonical discriminant analysis (Figure 7) indicates that, by using CD 1 andCD2, groups defined by the hierarchical cluster analysis are well defined andhave analytical merit. Group 2a, 2b, and 3a appear to be associated in onecentral aggregation, while Group 3b is somewhat separated from this centralconcentration. Group 1 emerges as a distinct cluster that is very different fromthe others ..

The groups found in this project display variation in their chemicalcomposition. Because sources were not analyzed in this study, the variation inthe sources cannot be quantified. However, the variation in the groups impliesdifferent procurement practices or variation within the procurement site.Depending on the source and the geochemistry of the region, differing elements

501

and element groups can be used to characterize the sources and artifacts. Sub-source variation in the major source may also be present(5). It can be seen fromthese data that the trace element analysis of ochre is helpful for understanding itsvariability. The following section will discuss archaeological interpretation ofthe variation in ochre composition at the site of liskairumoko.

Archaeological Interpretation·

When considering archaeological interpretation of the results from chemicalcharacterization, we return to the five questions described in the project goals.

1) What is the overall'variability of ochre? Variability is present, but it isnot large. Given the paucity of ochre chemical characterization stUdies,evaluating the degree of variability is problematic. Hierarchical clusteringsuggested the presence of five groups, which were corroborated by canonicaldiscriminant analysis. Groups I and 3b appear to be the most dissimilar from themajority of samples. Groups 2a, 2b, and 3a may comprise a single heterogeneouscluster.

2) Are there chemical differences in ochre remains recovered from differentcontexts? Variability is present, but does not appear to be context-specific. Themajority of samples are scattered throughout, as seen in Figures 4 and 5. Thesamples that comprise this aggregation come from all of the depositionalcontexts that were sampled from liskairumoko. Ochre sampled from differentarchaeological·contexts at liskairumoko is not chemically distinct.

3) Are ochre artifacts from different structures chemically similar ordifferent? Ochre artifacts from different structures are chemically similar.Samples 368 and 320 are very close together in CD space as well as theloglO[Sm/Fe] vs log 10[Co/Fe] and 10glO[Eu/Fe] vs 10glO[Zn!Fe] bivariatediagrams (Figure 4 and 5). Sample 368 comes from a sealed context at the baseof the Late Archaic pithouse while sample 320 was recovered from the edge of alarge rectangular structure. Both samples came from clear and unambiguouscontexts yet are very similar chemically, indicating that either the same source orthe same portion of a heterogeneous source was used throughout the span ofliskairumoko's occupation.

4) Is there large variability in the ochre from the thermal processing area?Figures 4-6 and the CD plot (Figure 7) indicate that there is considerablechemical variability in the samples taken from the hearth in the base of the LateArchaic pithouse and from the hearth-associated debris arc. This variabilitysuggests that ochre in this activity area was obtained either by ]) various trips todifferent parts of a single source or 2) various exchanges for ochre from differentsources. It is not likely that the thermal treatment of the ochre significantlyaltered the trace chemical signature of the ochre. Chemical variability of ochre

502

artifacts in this context suggests. either different portions of a single sOUthat multiple sources are represented. This information serves as an addiline of evidence indicating that the debris arc next to the hearth was noutcome of a single behavioral episode. Instead, it represents habitual heart .activity ...

5) Is the ochre from this context similar to ochre found in secondary refUElemental and CD space plots all indicate chemical similarities between o¢artifacts recovered from the central hearth and debris arc encountered at the bof the Late Archaic pithouse. The occurrence of chemically similar ocartifacts in both contexts provides additional evidence that secondary redeposits were fonried by cleaning hearths and removing debris from insCstructures. __

Several other observations can be made from Figures 6 and 7. Group"l'exhibits the greatest distance in CD space. This group consists of samples frp!Ji:the plow zone, deep within the fill of Pithouse 3, and a stain outside Pithouse3"Group 1 may represent a source, or portion of a source, that was not generanypreferred by occupants of Jiskairumoko. None of the samples forming Group 1were recovered from early occupational contexts at Jiskairumoko, though noneof them are unambiguously later either. Samples comprising Group 3b are-allfrom plow zone contexts except for samples 315 and 342. These latter two arenot from early contexts. Thus, Groups I and 3b, which are the most chemicallydistinct, do not appear to include any samples recovered from early contexts. Forexample, none of the ochre samples from Late Archaic pithouse contexts aremembers of either Group I or 3b. These two groups may represent two portionsof a single heterogeneous source or two additional sources that were notpreferred by residents of Jiskairumoko.

Conclusions

While the artifacts were not compared to the ochre sources from the region,some tentative conclusions about the variability of ochre and its use atJiskairumoko can be drawn. Statistically distinct groups were found in ochrefrom Jiskairumoko. Major elements, such as Fe and Mn, are as important astrace elements, such as the rare earth elements, in studying ochre variability.Viewed in a number of different dimensions and transformations, the majority ofochre samples appear to form a single rather heterogeneous congregation thatcomprises all of the depositional contexts that were sampled. Ochre use withinany given depositional context is composed of members of more than onestatistically defined group. These statistical Clusters also consist of samples frommultiple depositional contexts. Without locating and geochemicallycharacterizing a range of ochre sources it is impossible to determine if thesereflect multiple sources or a single heterogeneous source. Later in time, either

503two additional portions of a source or two additional sources may have comeinto use. Additional work locating and characterizing Andean ochre sources isneeded. However, it appears that ochre followed a procurement trajectorydifferent from obsidian, which was obtained from a single non-local source inthe Colca Canyon near Arequipa( 49).

Acknowledgments

It

For the analytical work in this study, the authors wish to acknowledge theArchaeometry Laboratory Staff for preparation of samples, as well as theNational Science Foundation Graduate Fellowship (Rachel Popelka-Filcoft) andNational Science Foundation grant #0504015 (Principal Investigator: Michael D.Glascock) for funding. Fieldwork was supported by: National ScienceFoundation grant #9816313 for fieldwork in 1998,and a supplement forfieldwork in 2002; National Science Foundation Equipment grant #9978006 in .1999, and John H. Heinz III Charitable Trust in 1997 and 2002. MarkAldenderfer was Principal Investigator on all grants for fieldwork.

References

1. Weinstein-Evron, M.; Ilani, S. J. Archaeol. Sci. 1994,21,46]-467.2. David, B.; Clayton, E.; Watchman, A. L. Austral. Archaeol. 1993,36, 56-

57.

3. Hovers, E.; Ilani, S.; Bar-Yosef, 0.; Vandermeersch, B. Curro Anthro. 2003,44,491-522.

4. Stafford, M. D.; Frison, G. C.; Stanford, D.; Zeimans, G. Geoarchaeol.2003,18,71-90.

5. Popelka-Filcoff, R. S.; Robertson, J. D.; Glascock, M. D.; Descantes, C. J.Radioanal. Nucl. Chem.; in press.

6. Erlandson, 1. M.; Robertson, J. D.; Descantes, C. Am. Antiq. 1999,64,517-526.

7. Mrzlack, H.M.A. thesis, University of Colorado-Denver, Denver, CO 2003.8. Moseley, M. The Incas and Their Ancestors. Thames & Hudson: New York,

NY, 2001.9. Stanish, C. Annu. Rev. Anthro. 2001, 30, 41-64.10. Stanish, C. In Ancient Titicaca: The Evolution of Complex Society in

Southern Peru and Northern Bolivia; University of California Press:Berkeley, CA, 2003; P 354.

II. Lavallee, D.; Julien, M.; Wheeler, J.; Karlin, C. In Telarmachay: Chasseurset Pasteurs Prehistoriques des Andes; Institut Fran~ais D'Etudes Andines:Paris, 1985; Vol. I.

504

12. Aldenderfer, M. S. In Montane Foragers Asana and the South-C'Andean Archaic; University of Iowa Press: Iowa City, lA, 1998; P 32

13. Craig, N. Ph.D. thesis, University of California at Santa Barbara;'Barbara, CA, 2005.

14. Smith, B. D. J Archaeol. Res. 2001,9, 1-43.15. Shackley, S. M.; Eklund, E.; Ogasawara, C. Report. Source Provenanc

Obsidian Artifacts Jiskairumoko (189), Peru; University of CalifoBerkeley Archaeological XRF Laboratory: Department of AnthropolBerkeley, CA May 2004; p 10.

16. Speakman, R. 1.; Popelka, R. S.; Glascock, M. D.; Robertson, J. D. RepAnalysis of Obsidian Artifacts from Southern Peru Using a Field-PorttibX-Ray Fluorescence Spectrometer; University of Missouri Research React'Center: Columbia, MO, 2005.

17. Craig, N.; Aldenderfer, M. In Advances in Titicaca Basin ArchaeologyIKlarich, E. A.; Stanish, C., Eds.; Cotsen Institute of Archaeology: Lds~Angeles, CA; in press. '

18. Aldenderfer, M. S. Qi//qatani and the Evolution of Pastoral Societies in theTiticaca Basin; in press.

19. Baid, C. A.; Wheeler, J. Mountain Research and Development 1993, 13,145-156.

20. Dransart,P. World Archaeol. 1991,22,304-319.21. Dransart, P. Earth, Water, Fleece, and Fabric: An Ethnography and

Archaeology of Andean Came/id Herding. Routledge: New York, NY 2002.22. Hesse, B. J. Ethnobiology 1982, 2,1-15.23. Kuznar, L. A. Awatimarka: The Ethnology of an Andean Herding

Community. Harcourt Brace: Orlando, FL 1995.24. Nunez, L. Boletin de Antropologia Americana 1981,2,87-120.25. Nunez, L. Chungara 1982, 9, 80-122.26. Rick, J. Prehistoric Hunters of the High Andes. Studies in Archaeology.

Academic Press: New York, NY, 1980.27. Zlatar, V. Chungara 1983,10,21-28.28. Dillehay, T. D.; Nunez, L. In Recent Studies in Pre-Columbian

Archaeology; Saunders, N. J.; Montmollin, O. D., Eds.; BAR InternationalSeries: Oxford, 1988; Vol. 421, pp 603-634.

29. Browman, D. Am. Scientist 1981, 69,408-419.30. Quilter, J. J. World Prehistory 1991, 5, 387-415.31. Rowe, M. H. News in Physiol. Sci. 2002, 17, 93-98.32. Mollon, J. D. J. Experimental Biology 1989,146,21-38.33. Sumner, P.; Mollon, 1. D. J. Experimental Biology 2000,203, 1963-1986.34. Sumner, P.; Mollon, J. D. J. Experimental Biology 2000,203, 1987-2000.35. Dominy, N. J.; Lucas, P. W. Nature 2000,4/0,363-366.

\1

U,I

d

Ig

y.

505

36. Berlin, 8.; Kay, P. Basic Color Terms; University of California Press: !

Berkeley, CA, 1969.

37. Brown, D. E.' Human Universals; McGraw-Hili, Inc.: New York, NY, 1991.38. Knight, C.; Power, C.; Watts, I. Cambridge Archaeol. J. 1995,5,75-114.39. Wreschner, E. E. Curro Anthro. 1980,21,631-644.40. Bolton, R. Curro Anthro. 1980,21,633-635.41. Stevenson,P. Curro Anthro. 1980,21,633-635.42. Turner, V. W. The Forest of Symbols: Aspects of Ndembu Ritual; Cornell

University Press: Ithaca, NY, 1967.43. Aldenderfer, M. S. In Kaillachuro: A Formative Mortuary Complex from

the Southwestern Lake Titicaca Basin; Institute of Andean Studies,Berkeley, January, 1998; Berkeley, CA, 1998.

44. Craig, N.; Moyes, H.; Aldenderfer, M. J. Archaeol. Sci.; in press.45. Binford, L. In Pursuit of the Past: Decoding the Archaeological Record;

Thames and Hudson: New York, NY, 1983; pp 1-256.46. Elias, M.; Chartier, c.; Prevot, G.; Garay, H.; Vignaud, C. Mater. Sci. Eng.

B 2006, 127, 70-80.47. Tankersley, K. B.; Tankersley, K.G.; Shaffer, N. R.; Hess, M. D.; Benz, J.

S.; Turner, F. R.; Stafford, M. D.; Zeimans, G.; Frison, G. C. PlainsAnthropologist 1995, 40,185-194.

48. Glascock, M. In Chemical Characterization of Ceramic Pastes inArchaeology; Neff, H., Ed.; Prehistory Press: Madison, WI, 1992; pp 11-26.

49. Burger, R. L.; Asaro, F.; Salas, G.; Stross, F. Andean Past 1998, 5, 202-223.