Bounding Chaco: Great House Architectural Variability across Time and Space Author(s): Ruth M. Van Dyke Source: , Vol. 69, No. 2, The Chaco World (Winter, 2003), pp. 117-139 Published by: Archaeological and Historical Society Stable URL: http://www.jstor.org/stable/30246714 . Accessed: 15/09/2011 15:16

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Arizona Archaeological and Historical Society is collaborating with JSTOR to digitize, preserve and extend access to Kiva.

http://www.jstor.org BOUNDINGCHACO

Great House ArchitecturalVariability Across Time and Space

RuthM. VanDyke

ABSTRACT Thebroad similarities that allow for the definition and recognition ofgreat house architectureat sites throughout the greater San JuanBasin strongly suggest the formis notthe product of independent invention. How and why were great hous- es builtin outliercommunities? What does the appearance of great house archi- tecturesuggest for relationships between outliers and Chaco Canyon, and for the socialstructure ofthe outlier communities themselves? As partof my dissertation research(Van Dyke 1998, 1999) I conducteda comparativearchitectural study designedto addressthese questions. In thispaper, I revisitthe original study, using188 casesfrom an expandeddatabase developed by participantsin the ChacoWorld meeting. Variables compared among the outlier cases include geo- graphiclocation, distance from Chaco Canyon,size, room count, kiva/room ratio,and presence/absenceof great , roads, earthworks, core-and-veneer masonry,and elevated kivas. Not surprisingly, Bonito style architectural attributes arefound to co-occurin theChacoan core area during the Late Pueblo II period. Landscapemodifications and multiple great kivas represent the Chacoan package laterin timeand furtherafield. Most interestingly, core-and-veneer masonry and kiva/roomratios are found to exhibitsignificant, patterned distributions inboth timeand space.An absenceof core-and-veneermasonry, together with a high kiva/roomratio, emerge here as a usefulway to set boundaries on thegeographic scopeof Chacoan influence.

RESUMEN Lassemejanzas amplias que nos permiten la definicidn yel reconocimientodela arqui- tecturade los "Great Houses" en todo el Cuencode San Juan sugieren convincemente que estaarquitectura nofuera una invenci6n independiente. C6mo y por que' se construtan los"Great Houses" en las comunidadesvecinas? iPor qud el aspectode la arquitectura de los "GreatHouses" presupone la relaci6n entre el Caio'nChaco y las comunidades

KIVA:The Journal ofSouthwestern Archaeology andHistory, Vol. 69, No. 2, (Winter2003), pp. 117-139. Copyright@ 2003 Arizona Historical and Archaeological Society. All rights reserved.

117 118 RuthM. VanDyke

vecinasytambidn explica la estructurasocial de estas cominidades? En partede mi tesis doctoral(Van Dyke 1998, 1999) lleve'a caboel estudioarquitect6nico comparativo que abord6estas preguntas. En esta investigaci6n vuelvo a analizar las mismas preguntas al usar188 casos sacados de la basede datos ampliada que ha sido desarrollada por los par- ticipantesenla reunionmundial Chaco. Los factores que se compararonenlos casos de las comunidadesvecinos son: posici6n geogrdfica, distancia del Caii6n Chaco, tamafio, nimerode cuartos,proporci6n de kiva/cuarto y presencia/ausencia de kivas grandes, caminos,trabajos de preparaci6ndel terreno,construcci6n con enchapado y madera, kivaselevadas. No fue una sorpresa ver que los atributos arquitect6nicos delestilo Boni- tocoinciden temporalmente enel drea principal de losChacos durantre elperiodo Ultimo PuebloII. Lasmodificaciones al paisaje y las kivas grandes multiples representan el esti- lode los Chacos mds tarde ymas extendido espacialmente. De mayor interds fue el hecho que la construccidnconenchapado y madera y la proporci6nde kiva/cuarto ponen de manifiestolas distrubucionesimportantes de patrones tanto en el espaciocomo en el tiempo.La ausenciade la construcci6ncon enchapado y madera y la proporci6nde kiva/cuartonossirven de limite geogrdfico dela influenciade los Chacos.

Chacoan greathouses are plannedbuildings of core-and-veneercon- structioncharacterized by distinct, banded masonry styles. Most con- tain multiplestories, large rooms with high ceilings, great kivas, and enclosed kivas.Earthworks and clearedor constructedroad segmentsare associatedwith manygreat houses. These architectural characteristics are sometimesreferred to as Bonitostyle architecture, following the termthat originatedwith Gladwin (1945). Communitiescontaining elements of Bonito style architecture are docu- mentedacross the San JuanBasin and adjacentareas. These sites are popularly called"outliers," a term that emphasizes their spatial relationship with respect to Chaco Canyon.Each outlier minimally contains a Bonito-stylegreat house. One or more greatkivas, road segments,earthworks, and a groupof surrounding smallsites may also be present.Despite the broad similarities that allow forthe definitionand recognitionof Bonito style architecture across a widearea, all out- lierswere not createdequal. Not onlydoes thecomposition of Bonito-style ele- mentsdiffer among communities,but outliergreat houses themselvesexhibit considerablevariability with respect to shape,size, and elevation(Figure 1). Neitherthe precise nature of the relationship between Chaco Canyonand theoutliers nor the function of Bonitostyle architecture within outlier commu- nitiesis well understood.A numberof modelshave been developedto explain therather dramatic spread of Bonitostyle architecture across an arid,agricultur- allymarginal landscape between ca. A.D. 890-1140. Earlyexplanations focused on thecanyon. More recent work has recognizedthat relationships between the BoundingChaco 119

LA16515

UpperKin Nizhoni

LA16508

Pierre's LowerKin Nizhoni LA16509

KinKlizhin Casamero

Guadalupe Andrews

Lowry KinBineola

Figure 1: Examplesof some Classic Bonito-phase great houses from across the Chaco world.See Figure1 in this volume's Preface for locations. 120 RuthM. VanDyke

Chaco Canyonand theoutliers must have been an importantpart of theraison d'etreof Bonito style architecture in bothareas. Thebroad similarities that allow for the definition and recognitionof great housearchitecture at sites throughout the greater San JuanBasin strongly suggest theform is notthe product of independentinvention. How and whywere great housesbuilt in outliercommunities? What does theappearance of greathouse architecturesuggest for relationships between outliers and Chaco Canyon,and forthe social structure of the outlier communities themselves? As partof my dis- sertationresearch (Van Dyke 1998, 1999) I conducteda comparativestudy designedto addressthese questions. My analysis was based on thepremise that outliergreat house similarityshould reflecta Chacoan sourcefor Bonito style architecture,and diversityshould reflectthe converse.Results were ambiguous, indicatingthat tremendous diversity exists within the confines of what is consid- eredChacoan, and pointingtowards a rangeof explanatorymodels rather than one singleexegesis. I concludedthat construction of Bonitostyle architecture is likelyrelated to communityritual and powerissues, and thata varietyof rela- tionshipsprobably existed between outlier communities and Chaco Canyon. Steinand Lekson's(1992) conceptof ritual landscape, and Renfrewand Cherry's (1986) peer-polityinteraction model bestdescribed how Bonitostyle architec- turemay have spread across a wide areawithout necessitating direct supervision fromChaco Canyon.

BOUNDING CHACO

In thispaper, I revisitthe original study, using an expandeddatabase developed byparticipants in theChaco World meeting (Preface, this volume). As I see itcur- rently,there are two interrelated issues that can be addressedby comparing spe- cificarchitectural variables across time and space.One has to do withthe nature ofthe interaction between outliers and Chaco Canyon;the other has to do with definingthe limits of the interaction. In theoriginal study, I bounded my sample universein timeand space-I includedonly great houses datedto the Classic Bonitophase (ca. A.D. 1040-1100)from the San JuanBasin and immediately adjacentareas. The expandeddatabase offers an opportunityto look atvariabili- tyin greathouses across a widerexpanse of time and space.Before one can inves- tigatethe nature of interactionwith Chaco, it is necessaryto discernwhether, in fact,there was interactionat all. Whereare the boundaries of what we mayuse- fullyconsider to representthe Chaco world? Banded,core-and-veneer masonry was one ofthe initial, easily identifiable featuresused to characterizelarge Pueblo II structuresacross the San Juanbasin as "greathouses" (Marshallet al. 1979:15; Morris1919:105; Powerset al. 1983:16; Roberts1932:29; Roys 1936). The definitionof "greathouse" has since been expandedby some (e.g.,Lekson 1991) to includeany large structure locat- BoundingChaco 121 ed prominentlyin the midstof other,smaller structures, regardless of thepres- ence of shared,"internal" aspects of buildingconstruction such as core-and- veneermasonry. As a result,the expandeddatabase originally included large structuresas farafield from Chaco as Wupatki. Is it reasonableto thinkthat 11th century people living450 kmapart had meaningfuland regularsocial interaction,or at minimum,were members in a commonreligious sect? In theanalysis presented here, I notonly revisit the orig- inal studyto assesswhether my earlier findings hold witha largersample and to look fornew patterns, I also look fordifferences that likely represent boundaries to theChacoan experience. The analysisis based on theassumptions that some architecturalvariables do representparticipation in common learningframe- worksand socialactivities, and thatshared experiences are one wayto character- ize theChaco world.

The OriginalStudy The firstpart of the originalstudy involved a comparisonof generaloutlier attributesto assessthe nature of gross regional patterning, ifany. I soughtto deter- minewhether outlier community attributes such as greatkivas, roads, and berms exhibita patternedor regionaldistribution. Classic Bonito phase canyongreat housescover two-dimensional areas in excessof 2,000 m2. However,outlier great housefloor area was knownto be extremelyvariable. Using a databasecontaining 12 canyonand 39 outliergreat houses, Powers et al. (1983:313-315,344-345) definedthree great house size classesand usedthat framework to arguethat great housesrepresent a three-tiered, integrated settlement system. In theoriginal study, I investigatedwhether great house size is associatedwith distance from Chaco Canyon,regional outlier location, or the presence of great kivas, road segments, or earthworks.Great house size,great kiva size, and presence/absenceof roadsand earthworkswere compared against each otherand againstnorth/south location, region,and distancein km fromPueblo Bonitofor 62 cases (Van Dyke 1998). Comparisonsof dichotomous categorical variables were made using Fisher's exact test.Comparisons of polytomous categorical variables were made using Pearson's chi-squaretest. Comparisons of metricvariables were made usingthe Mann- WhitneyU-test and KruskalWallis one-way analysis of variance. P valuesof 0.05 or lesswere considered statistically significant. In the originalanalysis, I foundfew statistically significant associations. Great house size was not significantlyassociated with any othervariable, althoughmost of thesmall outlier great houses were located in thecentral San JuanBasin. Roads and earthworkswere found more commonly in thesouth, but thismay have been because,within the confinesof the sampledatabase, road and earthwork-relatedresearch was concentratedin thesouth (e.g., Fowler et al. 1987). The most importantfinding was a statisticallysignificant association 122 RuthM. VanDyke betweengreat kiva presence/absenceand distancefrom .Great kivaswere more likely to be absentin outlierswithin 40 or 50 km of Chaco Canyon.This pattern resonated with Breternitz et al's (1982; Doyelet al. 1984) idea of a Chaco Halo, wheresites located in the immediateenvirons of Chaco Canyonare consideredeconomically and demographicallysynonymous with sitesin thecanyon. I haveexplored this more thoroughly in a recentKiva article (VanDyke 2002). The secondpart of theoriginal study was designedto distinguishcanyon- directedfrom local greathouse constructionat outliers.The dataset included 61 ClassicBonito phase (A.D. 1040-1100)great house componentsfrom 55 outlier communities.The studyfocused on internal,or low visibilityaspects of great house construction,because these are most likely to containenculturative infor- mation(Carr 1995). The idea was thatlocal emulationshould be distinguish- able fromdirected, Chacoan construction in thatalthough the general form and appearanceof Bonitostyle architecture might be emulated,internal, precepts of the stylecould not. If outliergreat houses werebuilt by people fromChaco Canyon,then five internal variables were expected to co-occurand should be similarin theirdistribution across space. If outliergreat houses werebuilt by local communitiesto emulatethe structures of Chaco Canyonor oftheir neigh- bors,then the internal variables were expected to exhibitgreat disparity, with lit- tle co-occurrence.If communallabor practices led to Bonitostyle architectural informationbeing passed from community to community,then diversity among internalvariables was expectedto decreaseas distancefrom Chaco Canyon increases.Because most outlier great houses are unexcavated,internal variables weredrawn from surface data, although this situation is lessthan ideal. The fiveinternal variables used in the originalstudy included core-and- veneermasonry, banded facing,symmetry, elevated kivas, and kiva/roomratio. Forspecific information about how thesevariables were defined, and howprob- lemsof data comparabilityamong outliers were addressed, the interested reader is referredto VanDyke (1998, 1999). Analysesproceeded using the same statisti- cal measuresas describedabove. Core-and-veneermasonry was nearlyubiqui- tous in distribution,indicating that this constructiontechnique was widely shared.Banding and symmetryoccurred in approximatelyhalf the cases,but theydid not co-occurin a patternedrelationship, and no discerniblepatterning was foundin thelocational distribution of these variables or in theirassociation withother variables. Kiva/room ratio and elevatedkivas were not ubiquitous but exhibitedsome regionalpatterning. Where regional patterning exists, it suggests thatsome outliergroups shared information with each otherand withChaco, but othergroups did not. The resultsof the analysissuggested architectural information followed subregionalnetworks of interactionacross the Chaco world,but the specific natureof thesenetworks could not be furtherdefined. Great house architecture BoundingChaco 123 did not spreadin a linearor uniformmanner. I concludedthat the presence of greathouse architecturedoes not necessarilyentail direct contact with Chaco Canyon,and thatlarge-scale architectural similarities need notbe interpretedto indicateparticipation in one coherent,centralized, Chaco system.Rather, some outliercommunities may be local,in situdevelopments. A similarinterpretation was also reachedby Marshall et al. (1979:337) and Powerset al. (1983:341).

InvestigationsUsing the ExpandedDatabase The expandeddatabase offers an opportunityto revisitmy original comparative studiesto see ifthe patterns I observed for 62 caseswill hold fora databasethree timesthat size, and whetherany new patternsmay be discerned.The expanded databasecontains 188 caseswith enough information to be usefulfor a similar analysis(see Table 1, Preface,this issue). The originalstudy cannot be replicated exactly,because not all thesame variables were collected, and some ofthem were assesseddifferently. The expandeddatabase also providesa broadertime hori- zon (A.D. 900-1300), affordingthe opportunityto move beyondthe Classic Bonitophase (A.D. 1040-1100)window of the originalstudy and assess how spatialpatterning at outlierschanges through time. Temporal, geographic, and architecturalvariables used in thecurrent analysis are discussed below.

TemporalVariation Databasecompilers assigned numeric date ranges to outliersbased on available information.In mostcases, date ranges were gleaned from surface ceramics; in the rarecases whereexcavations have been conductedor wherethere is standing architecture,tree-ring and radiocarbondates were included. For purposes of this study,it was importantto be able to sortoutliers by time period, both in orderto factorout noncontemporaneoussites from comparisons, and to trackchanges in outlierpatterning through time. To createcomparable date categories for the anal- ysis,I used theinformation in thedatabase to assigneach case to one or moreof threebroad temporalperiods: Early Pueblo II (A.D. 900-1040), Late Pueblo II (A.D. 1040-1150),and PuebloIII (A.D. 1150-1300).This resulted in fivepossible temporaldesignations (Table 1). Singlecomponent sites are best for comparison, butof 188 cases,77 aremulti-component. I did notwish to completelyeliminate the 77 multi-componentsites, as thisgroup contains many prominent outliers (e.g.,Kin Bineola), so I deviseda wayto incorporatethem into the analysis using sixsubjectively developed date groups (Table 2). GroupsA, D, and F are single- componentgroups. Groups B and E includerespectively all EarlyPueblo II, or PuebloIII, components,including multi-component background noise. Group C containsthe entire spectrum of LatePueblo II components;earlier and latersin- gle-componentsites are excluded, but multi-component sites are included. 124 RuthM. VanDyke

Table 1. Temporalassignments ofsites in the database (n = 188)

Date Code Periods DateRanges Breakdownby Region n 1 EarlyPueblo II A.D.900-1040 SanJuan Basin = 9; Arizona= 1; FourComers = 1 11 2 EarlyPueblo II A.D.900-1150 SanJuan Basin = 15;Southern = 18; -Late Pueblo II Arizona= 8; FourComers = 2 43 3 EarlyPueblo II A.D. 900-1300 SanJuan Basin = 1;Arizona = 1 2 -LatePueblo II -PuebloIII 4 LatePueblo II A.D. 1040-1150 SanJuan Basin = 22; Southern= 11; Arizona= 16; FourComers = 6 55 5 LatePueblo II A.D. 1040-1300 SanJuan Basin = 6; Southern= 6; -Pueblo III Arizona= 3; FourComers = 17 32 6 PuebloIII A.D. 1150-1300 SanJuan Basin = 7; Southern= 4; Arizona= 22; FourComers = 12 45

Table 2. Temporalgroups used in the analysis.

Date Total GroupDate Codes Included Breakdownby Region (n) A 1: EarlyPueblo II componentsonly SanJuan Basin = 9; Arizona= 1; FourComers = 1 11 B 1 & 2: includesall EarlyPueblo II SanJuan Basin = 24; Southern= 18; components,even when a Late Arizona= 9; FourComers = 3 54 PuebloII componentis also present C 2,3, 4, & 5: includesall LatePueblo II SanJuan Basin = 44; Southern= 35; components;excludes single-compon- Arizona = 28; FourComers = 25 132 entEarly Pueblo II andPueblo III sites D 4: LatePueblo II componentsonly SanJuan Basin = 22; Southern= 11; (smallersample than Group C, Arizona= 16;Four Comers = 6 55 butmore temporally pure) E 5 & 6: includesall PuebloIII compon-San JuanBasin = 13;Southern = 10; ents,even when a LatePueblo II Arizona= 25; FourComers = 29 77 componentis alsopresent F 6: PuebloIII componentsonly SanJuan Basin = 7; Southern= 4; Arizona= 22; FourComers = 12 45 BoundingChaco 125

GeographicVariation

The greaterSan JuanBasin may be dividedphysiographically into 20 or more subregions(e.g., Marshall et al. 1979:20-22), but thislevel of compartmental- izationyields outlier counts for each areathat are too smallfor meaningful com- parison.In myoriginal study, I combinedoutliers into three general regions by meansof two east-west boundaries drawn approximately atthe north edge of the Chaco Plateauand the southedge of the SouthChaco Slope. In the expanded database,outliers were assigned to one of 14 geographicallydefined regions (Fig- ure 1, Preface,this volume). Again, this level of subdivisionresults in outlier countsfor some areasthat were too small formeaningful comparison. To miti- gatethis problem, I collapsedthe expanded database into four broad geograph- ic categories-theSan JuanBasin, Southern, Arizona, and FourComers regions (Table3). Two additionalgeographic variables were also used. Cases werechar- acterizedas northor southbased on theirposition relative to an imaginaryeast- west line runningthrough Chaco Canyonand extendingin eitherdirection. Distanceis a metricvariable referring to kmbetween each outlier great house and PuebloBonito.

ArchitecturalVariables

The expandeddatabase contains presence/absence information for great kivas, roads,and earthworksfor each outliercase. It also containscounts for great kivas and roads,where present. This informationwas used to createtwo additional categoriesviable for statistical comparisons-multiple great kivas, and multiple roads-in whicheach case contained0, 1,or morethan 1 instanceof the feature. Thus,it was possibleto look forassociations not onlybetween the presence/ absenceof roads,great kivas, and othervariables, but also betweenthe presence ofmultiple roads, multiple great kivas, and othervariables. Greathouse size is assesseddifferently in the expandeddatabase than in myoriginal work. In theoriginal study, I assigned the metric variable area to each

Table 3. Fourcollapsed regions used in the analysis.

Region(this analysis) Region(expanded database) n San JuanBasin CentralSan Juan Basin, Chuska Valley, Lobo Mesa/San Juan Basin 60 Southern Acoma,Lobo Mesa/Red Mesa Valley, Cibola/Zuni 39 Arizona BlackMesa, Defiance Plateau, Little , Rio Puerco, UpperChinle 51 FourCorners MiddleSan Juan, San Juan Foothills, Totah/Upper San JuanBasin 38 126 RuthM. Van Dyke

greathouse on the basis of two-dimensional roofed area in square meters, exclu- siveof plaza space (cf. Powers et al. 1983:313,Table 41). However, wetabulated roomcount rather than great house area for the expanded database because of disparitiesinthe methods used by the researchers tocollect areal information. So,room count isused here as a proxymeasure, albeit imperfect, forgreat house size.As an additional way to look at this variable, I used a stemand leaf plot to organizethe room count data into four broad categories. Small (1-11 rooms), medium(12-20 rooms), large (22-49 rooms), and very large (50 or more rooms)great houses are identified bythe categorical variable size group. Twointernal variables used in the original study-banding and symme- try-arenot employed here, because they do notappear in theexpanded database.The remaining three-core and veneer, kiva/room ratio, and elevated kivas,are included. It shouldbe notedthat most of the great houses in the expandeddatabase are unexcavated, which means that there is undoubtedly somedegree of error in the tallies of these variables. Evidence derived from sur- facesurvey isnot only incomplete, butis not always comparable, due to differ- encesamong investigators in terms of bothresearch agendas and field experience.I have not ground-verified everycase in the expanded database, but I haveattempted tominimize this problem by omitting cases where the evidence forthese variables seemed sketchy orambiguous. Inthe case of multicomponent greathouses, every effort was made to separate architectural building episodes. To theextent possible given database limitations, core-and-veneer masonry, kiva/roomratio, and elevated kivas were tabulated separately foreach temporal component,and these constituted discrete cases used in the analysis. As in the originalstudy, elevated kivas is taken here to include both second-story enclosed kivas,and tower kivas, which consists ofa seriesof up to four enclosed kivas ver- ticallystacked on top of one another (see Marshall etal. 1979:18 for a compari- son).This categorization ensures a large enough sample of elevated kivas for comparison,and it circumvents theconfusion surrounding theterm tower kiva, whichis sometimes used to describe second-story kivas (e.g., Irwin-Williams and Shelley1980; Lekson 1984:52).

Methods

As in theoriginal study, comparisons of dichotomouscategorical variables were made usingFisher's exact test. Comparisons of polytomous categorical variables weremade using Pearson's chi-square test. Comparisons of metric variables were made usingthe Mann-WhitneyU-test and KruskalWallis one-way analysis of variance.Correlations with distance were explored using regression analysis. P valuesof 0.05 or lessare considered statistically significant. BoundingChaco 127

Resultsand Interpretations Resultsof the comparisonsare presentedin Tables 4-8 and are discussedin detailbelow. In the interestof clarity,only significant results are listed.Table 4 presentsresults for the entire database; these comparisons, including sites from all temporalgroups, enabled the identificationof changesthrough time. Addi- tional comparisonsamong variables were conducted for each of the six date groupsto factorout the "noise"contributed by multiple temporal components. No significantassociations were found withinGroup A (n = 11), possibly becauseit is so small.Results for Groups B, C, D, and E arepresented in Tables 5, 6, 7, and 8, respectively.Significant results for Group F weretoo fewto warranta separatetable, but they are discussed in thetext.

GeneralPatterns ChangesThrough Time Althoughthe database is relativelycrude with respect to regionaland temporal definition,a cross-tabulationof temporaland locationalvariables supports basic,well-known shifts in regionalpatterning in the Chaco area betweenA.D. 900-1300. BetweenA.D. 900-1150,during the growth and peak of the Chaco phenomenon,most sites are locatedin the San JuanBasin and the Southern regions,with relatively few sites in otherareas. There is a tendencyfor Early Pueblo II sitesto appearin theRed Mesa Valley(see also Van Dyke2000). Dur- ingthe Late Pueblo II/PuebloIII period,there is a shiftin focusto theFour Cor- nersregion, and betweenA.D. 1150-1300,most sites are locatedin Arizona. Thesepatterns are supportedby significant associations between the categorical variabledate and north/south,region, and distance(Table 4). Date is significantlyassociated with rooms (Table 4), reflectingthe unsur- prisingpattern that sites in thedatabase in theFour Corners and Arizona regions datingfrom the Pueblo III periodtend to be largerthan sites in the San Juan Basincore dating from the Pueblo II period.Finally, date is significantlyassociat- ed withmultiple great kivas (Table 4), reflectingthe fact that there are no siteswith multiplegreat kivas in GroupA (refinedEarly Pueblo II), overhalf (52%) of all outlierswith multiple great kivas fall within Group B (EarlyPueblo II), and the largestgroup of sitesthat lack great kivas (34%) fallswithin Group D (refined LatePueblo II). Multiplegreat kivas are clearly an EarlyPueblo II phenomenon.

GreatHouse Size Greathouse size forthe sites in thedatabase, as measuredby room counts, varies between1 and 437 rooms.The categoricalvariable size groupconsists of 35 small,41 medium,38 large,and 41 verylarge sites. Size group is significantlyasso- ciatedwith region and distancefor the entire database (Table 4). Bothassociations P/A

ElevatedKiva -

P/A

Core/Veneer -

BermP/A periods

P/A time all Road in

Mult.Segment - 0.025

occupied

RoadP/A N/A - 0.033 were sites GreatP/A

188).Mult. Kiva - - 0.005 0.024 = (n Kiva multicomponent cases all GreatP/A N/A - - 0.007 0.008these and since

Rooms 0.004 0.011 - - 0.000 N/Ahere, variables all omitted for SizeGroup N/A - - - 0.008 N/A were 3 RegionN/A 0.000- - 0.014 - 0.007 0.008 - - comparisons Group of Date resultsDistance** N/AN/A 0.036- - - 0.004 0.000 - -

Bonito comprising Date*(n=186) 0.000 0.000 - 0.003 - 0.041 - - - - - Significant cases P/A Kiva Pueblo 4. to Kiva GreatP/A Road two the km Table N/SRegion SizeGroup Rooms GreatP/A Mult.Kiva Road P/A Mult.Segment BermP/A Core/Veneer P/AElevated P/A Kiva/Room Ratio* ** BoundingChaco 129 holdfor the Group C LatePueblo II subset(Table 6), andthere is a significant associationbetween region and roomsfor this group as well.The association betweenregion and rooms holds for the Group D refinedLate Pueblo II subset (Table7), which factors out some very large Pueblo III sites.In sum, most sites in theSan Juan Basin and Southern region are of the small, medium, and large vari- eties(1-49 rooms),whereas the very large sites (50 or morerooms), such as AztecEast, tend to be locatedin the .

Table 5. Significantresults for Group B (n= 54).

Distance Region Size Group Rooms Kiva/RoomRatio GreatKiva P/A - - - - 0.002 Mult.Great Kiva P/A - 0.026 - - 0.004 Core/VeneerP/A 0.010 - - - - ElevatedKiva P/A - - 0.007 0.000 -

Table 6. Significantresults for Group C (n= 132).

Distance Region Size Rooms Mult.Great Kiva/Room Group Kiva P/A Ratio Size Group 0.014 0.002 N/A Rooms - 0.000 N/A N/A GreatKiva P/A - - - 0.004 N/A 0.011 Mult.Great Kiva P/A 0.004 0.024 - - - Core/VeneerP/A 0.000 0.017 - - - Elevated Kiva P/A - - 0.009 0.000 0.049

Table 7. Significantresults for Group D (n= 55).

Region GreatKiva P/A ElevatedKiva P/A Size Group 0.003 - Rooms 0.021 0.023 0.042

Road SegmentP/A 0.003 - Multi.Road SegmentP/A 0.006 Core/VeneerP/A 0.049 - Kiva/RoomRatio - 0.032 130 RuthM. VanDyke

Table 8. Significantresults for Group E (n= 77).

Region Rooms GreatKiva P/A Core/VeneerP/A ElevatedKiva - 0.029 0.019 -

Kiva/RoomRatio 0.014 N/A - 0.011

GreatKivas Greatkivas are found at 108 outliersin theexpanded database. Of these, 25 cases containmore than one greatkiva. Great kivas are reportedas absentfrom the remaining80 outliers.The categoricalvariable great kiva presence/absence is sig- nificantlyassociated with rooms, and withelevated kivas, for the entiredatabase (Table4; n = 188; p = 0.004). The associationwith rooms is maintainedfor Late PuebloII GroupsC-E (Tables6-8). Duringthe Late Pueblo II period,great kivas are muchmore likely to be presentat siteswith high room counts. The associa- tionwith elevated kivas in theoverall database probably reflects the fact that ele- vated kivas also are more common at pueblos with high room counts,as discussedfurther on. The categoricalvariable multiple great kivas is significantlyassociated with regionfor the entire database (Table 4). Caseswith 0 greatkivas are more likely in theSan JuanBasin-great kivas are lacking for 52% ofthe outliers in thisregion acrossall timeperiods. Cases with a singlegreat kiva are distributed fairly evenly acrossthe four regions. Nearly half (48%) of all caseswith more than one great kivaare found in theSouthern region. This significant association is maintained forthe subsetof EarlyPueblo II periodsites (n = 54, p = 0.026) and forthe GroupC subsetof LatePueblo II periodsites (Table 6; n = 132,p = 0.048). As I foundin theoriginal study (Van-Dyke 1998, 2002), greatkivas are less likelyto be presentat outliersin thecentral San JuanBasin. Furthermore, most sites with multiplegreat kivas are found in theSouthern region. Multiplegreat kivas also aresignificantly associated with rooms and elevated kivasfor the entire database (Table 4) and forthe Group D refinedLate Pueblo II subset(Table 6). Thisagain reflects the fact that great kivas are more common at largersites, and this patternis particularlystrong during the Classic Bonito phase,or LatePueblo II period.

Roads Roadsegments are recorded as presentin 91 casesand absentin 38 cases;there is no informationavailable for the remaining59 cases.Where roads are present, multiplesegments are recordedfor 36 cases. In the originalstudy, roads were significantlymore likelyto be presentin the south (Van Dyke 1998:218). Althoughno significantassociations were found between roads and north/south in the expandeddatabase, there are significantassociations with distance and BoundingChaco 131 region.Presence/absence of roads is significantlyassociated with regionfor the GroupD refinedLate Pueblo II subset,reflecting the fact that, during the Chaco heyday,an outlierwithout a road segmentcould be locatedmost anywhere, but an outlierwith a road segmentwas mostlikely located in the San JuanBasin (Table 6). Multipleroad segments are significantlyassociated with region and dis- tancefor the entiredatabase (Table 4). Theseresults hold forthe GroupC Late Pueblo II subset(Table 6), and theassociation with region holds forthe Group D refinedLate Pueblo II subset(Table 7). In theFour Comers and in Arizona,far fromChaco Canyon,outliers most likely have either no roadsor multipleroads, whereasin theSan JuanBasin and theSouthern region, nearer to Chaco Canyon, outliersmost likely have either no roadsor one roadsegment.

Earthworks Earthworks,orberms, are recorded as presentin 34 casesand absentin 3 casesin theexpanded database; there is no informationavailable for the remaining 151 cases.No significantassociations were found between the presence/absenceof earthworksand anyother variable for any time period. The lackof associations undoubtedlyreflects, in part,the uneven distribution of informationabout this variable.Of the 34 cases whereearthworks are present,27 are locatedin the southernhalf of the Chaco world.I noteda similarpattern in theoriginal study (Van Dyke 1998:218-219),where 9 out of 10 outliersrecorded as havingearth- workswere located in thesouth. As withroads, this pattern may be a functionof the factthat much of the researchthat has recognizedberms and earthworks (Fowleret al. 1987; Marshalland Sofaer1988) has been concentratedin the southernpart of theSan JuanBasin. The case of Bluff,in southeastUtah, is an interestingexception (Jalbert and Cameron2000). The functionsof bermsand earthworksin Chaco Canyonand at outliers are poorlyunderstood (Wills 2001). The Bluffberm is one of the fewoutlier earthworksto have been tested;in a recentreport, Cameron (2002) provides details and synthesizesinformation about outlierberms. Stein and Lekson (1992) havesuggested that earthworks are part of a rituallandscape, symbolical- lydefining space in frontof the great house. Following an observationmade by JohnRoney (personal communication 2000), I have suggestedthat these fea- tures,which contain a confusingarray of trash, sterile earth, and construction debris,may representfake middens built to bolsterlegitimacy through con- structedappeals to thepast (Van Dyke 2003).

InternalVariables

Threeof the fiveinternal variables used in the originalstudy-core-and-veneer masonry,elevated kivas, and kiva/roomratio-were revisitedhere. In the expandeddatabase, all threeexhibit interesting patterning. 132 RuthM. VanDyke

Coreand VeneerMasonry Becausecore-and-veneer masonry is an internalvariable not immediately visible to outsideobservers, it is a good indicatorof participationin commonlearning frameworks(sensu Lechtman 1977; Lemonnier1986; Dobres2000). In myorig- inal study,core-and-veneer masonry was significantlyassociated with the depen- dentvariable distance, suggesting that in the relativelyrare (4 out of 64) cases whencore-and-veneer masonry was lackingat a LatePueblo II greathouse, it was faraway from Chaco Canyon.This relationship holds for the current study, and the expandedgeographic and temporalinformation allows us to make even moresense of it. Core-and-veneermasonry is recordedas presentat 93 casesand absentat 45 cases in the expandeddatabase. The variableis significantlyassociated with regionand distancefor the entire database (Table 4). Bothrelationships hold for the LatePueblo II GroupC (Table 6), and theassociation with region holds for therefined Late Pueblo II GroupD (Table7). Thesefindings indicate a patterned distributionof core-and-veneersites that holds through the Late Pueblo II peri- od, but disappearsfrom the data setswith Pueblo III components.Not surpris- ingly,core-and-veneer masonry is morecommon in the Chaco corearea. Most (76-82%) greathouses in theSan JuanBasin and theSouthern region contain core-and-veneermasonry through the Late Pueblo II period.However, in Ari- zona and theFour Comers region, only about half (45-55%) ofthe Late Pueblo II greathouses containcore-and-veneer masonry. When broken down intothe Chaco Worlddatabase's smaller regions (Figure 1, Preface, this volume), this pat- ternholds. Approximately half the great houses in themiddle San Juan,San Juan foothills,Totah/Upper San Juan,Rio Puerco,Defiance Plateau, and UpperChin- le areascontain core-and-veneer masonry. The masonryis entirelylacking in the BlackMesa and LittleColorado areas. The factthat the masonry is presentabout halfthe time in subregionsdistant from Chaco suggeststhat its use is notmerely a functionof the presence of tabular and blockyCliff House sandstone,or some othersuitable building material. Core-and-veneermasonry is significantlyassociated with multiple road seg- ments(p = 0.030, n = 32) and withkiva/room ratio (p = 0.018,n = 44) forGroup F, suggestingthat all threevariables were a recognizablepart of a packageof architectiralattributes that continued through the end of Pueblo III period,in the post-Chaco world. In my conclusions below, I argue that the presence/absenceof core-and-veneermasonry can and should be employedto set spatial,as well as temporal,boundaries around the Classic Bonito-phase Chaco world.

ElevatedKivas In theoriginal sample, elevated kivas were heavily concentrated in thesouth and were significantlyassociated with road segmentsas well as greathouse size. BoundingChaco 133

Thesefindings are borneout here.Elevated kivas are recordedas presentat 42 and absentat 76 outliersin theexpanded database. For the database as a whole, thepresence/absence ofelevated kivas is significantlyassociated with six variables: greatkivas, multiple great kivas, road segments, multiple road segments, sizegroup, and rooms(Table 4). The associationwith rooms holds forGroups B, C, D, and E (Tables 5-8), and the associationwith size groupholds for GroupsB and D (Tables5 and 7). Elevatedkivas, when present, occur more often at largersites; none arepresent in thegroup of "small"great houses (less than11 rooms). This mayreflect the factthat larger sites are morelikely to incorporateseveral con- structionepisodes across time, in whichlater, "higher" kivas were built on top of earlierarchitecture and deposits. In thedatabase as a whole,most sites that lack elevated kivas also lackroad segments.The presence/absenceof elevatedkivas also is significantlyassociated withgreat kivas (Table 8) and multiplegreat kivas (Table 6). Mostsites that lack elevatedkivas also lackgreat kivas. These associationsindicate that these three variables-roads,great kivas, and elevatedkivas-occur more often at thesame sites,and less oftenin isolation.The patternsmay also reflecta tendencyfor all threevariables to appearat largersites, although neither great kivas nor road seg- mentsare, independently, significantly associated with size group.

Kiva/RoomRatio In the originalstudy, kiva/room ratios were significantly different in the north and south;great houses in thesouth (n = 46) had a mediankiva/room ratio of 1:9,whereas great houses in thenorth (n = 11) had a mediankiva/room ratio of 1:20.The 1:20 kiva/roomratio figure for northern great houses is similarto a fig- ure obtainedby Lipe (1989:56, Table 1), who estimateda kiva/roomratio of 1:15.2 fornine Chacoan great houses in theMesa Verdearea, although contem- poraneous,"local" pueblos exhibited kiva/room ratios of 1:6.5. Steward(1937), notinga kiva/roomratio of 1:5-1:6 as characteristicofPueblo II habitationsites, interpretedthis to mean thateach familygroup constructed its own kivafor its own ceremonies,and a latershift to kiva/roomratios of 1:15-1:25 reflecteda changein theorganization of ritual.Lipe and Hegmon(1989) furthercontend- ed thatkivas at thispoint acquireda new functionas settingsfor integrative social activities.Whatever the specificfunctions of kivas in greathouses- whetherceremonial, domestic, or both-the differencesin kiva/roomratio pat- ternsobserved in theoriginal study suggested that northern Chacoan great house kivaswere used differentlythan kivas in otherareas, or in non-Chacoannorthern sites. In the currentstudy, for n = 122 cases, kiva/roomratios range from extremesof 1:3 to 1:175,with a medianof 1:10.Significant associations are pre- sentwith kiva/room ratio through all dategroups, but interestingly, there is a shift in resultsbetween date groups with only Pueblo II components(B-D), and date 134 RuthM. VanDyke groupsthat include Pueblo III components(E and F). Kiva/roomratio is signifi- cantlyassociated with the presence/absence ofgreat kivas and multiplegreat kivas for the entiredatabase, and theserelationships hold forthe Pueblo II date groups (Tables4-8). Greathouse size is undoubtedlya factorhere-one or moregreat kivasare moreoften present at largersites, and kiva/roomratios are not inde- pendent of room count. P values below 0.05 can be obtained between kiva/roomratio, room count,and sizegroup for every sample set, but I do not considerthese results to be meaningful,because the variables are not indepen- dent-both kiva/roomratio and sizegroup were derived, in part,from room count. It would be usefulto explorethis pattern further with an independentmeasure ofsite size, such as area. Some of the mostinteresting kiva/room ratio results have to do withdif- ferencesbetween the Pueblo II (GroupsA-D) and Pueblo III (GroupsE and F) sites.The mediankiva/room ratios for both groups are similar (1:10 and 1:12.5, respectively),but the smallest kiva/room ratio for sites with only Pueblo II com- ponentsis 1:60, whereasthe smallestfor sites with Pueblo III componentsis 1:175. This patternconfirms and underscoresthe findingsfrom the original study-thespecial role of kivas in lateChacoan great houses seems to be intensi- fied,as thereare fewer of them, and morerooms. ForGroup E, kiva/roomratio is significantlyassociated with region (Table 8). This reflectsthe presenceof 9 large,late siteswith few kivas in the Southern region,particularly the Acoma and theCibola/Zuni areas (Figure 1, Preface,this volume).However, many of these sites, such as BossonWash, Cerro Prieto, and GonzalesWell, are listedin the expandeddatabase as containing50 or more roomson thebasis of rathertenuous information summarized in Fowleret al. (1987). For both GroupsE and F, kiva/roomratio is significantlyassociated with core-and-veneermasonry (see Table8 forGroup E; forGroup F p = 0.018,n = 44). LatePueblo III sitesthat lack core-and-veneer masonry have a mediankiva/room ratioof 1:8 (GroupE) or 1:7 (GroupF), whereasLate Pueblo II sitesthat exhibit core-and-veneermasonry have a mediankiva/room ratio of 1:14 (GroupE) or 1:13.5 (GroupF). Thisfinal pattern, I arguebelow, has importantimplications forour ability to distinguishtruly Chacoan great houses from very large pueblos.

Discussion and Conclusions

The picturethat emerged from the original study of n = 62 greathouses was of set of outlierscharacterized by diversityrather than homogeneity,requiring a rangeof interpretations.The resultsof the expandedoutlier database analysis clarifythose findings.Many of the patternsobserved in the originalstudy are strengthenedand upheld,and new progresshas been made towardsseparating theChacoan from the merely Chacoesque (sensu Hurst 2000). BoundingChaco 135

Withrespect to patterningof general variables such as size,great kivas, and landscapefeatures, I reconfirmed a number of earlierobservations. Sites of any size arelocated in theSan JuanBasin, but very large sites tend to be late,and they tendto be locatedfurther away from Chaco. The Southernregion not only has a concentrationof EarlyPueblo II greathouses but also a concentrationof very large,Late Pueblo II greathouses with multiple great kivas. Most sites that lack greatkivas are located close to Chaco Canyon,and thesegenerally date from the LatePueblo II period(see also Van Dyke2002). Mostsites with one road seg- mentare in theSan JuanBasin. There are strong associations among great kivas, elevatedkivas, and highroom counts, especially during the Late Pueblo II peri- od. In sum,the suiteof Bonitostyle characteristics hangs together, particularly duringthe Classic Bonito phase or latePueblo II period. Movingout fromthe Chaco core area,great houses eitherlack roads,or theyare associatedwith multiple road segments.Earthworks and multipleroad segmentsare commonly found at LatePueblo II and PuebloIII sitesin theFour Cornersand in Arizona.The furtherremoved from Chaco Canyonin bothspace and time,the more important landscape modifications such as roadsand earth- worksappear to become.Formal landscape modifications may be moreimpor- tantin thepost-Chaco world (Fowler and Stein1992). Alternatively,the patterns maymerely reflect the fact that road and earthworkresearch is notgeographical- lyconsistent across the sample. Internalvariables are expectedto help distinguishChacoan greathouses fromvery large sites. Three internal variables used in the originalstudy were comparedfor the expanded database. Elevated kivas do notappear to be particu- larlysensitive indicators with respect to thisissue. Elevatedkivas occur more oftenat larger,Late Pueblo II greathouses-they appear to be partof the package describedabove, that includes road segmentsand greatkivas. All threefeatures occurmore commonly together than they do in isolation,and all occurmore commonlyat largesites. Core-and-veneermasonry and kiva/roomratios, however, both appearto be usefulindicators of Chacoan-ness. In theexpanded database, core-and-veneer masonryexhibits significant, patterned distributions in bothtime and space.It is concentratedin the San JuanBasin and the Southernregion during the Late PuebloII period.However, it is presentin onlyabout half the contemporaneous greathouses in Arizonaand theFour Corners. A closerlook at geographyreveals thatthe masonry is absentin theBlack Mesa and LittleColorado areas. I do not believethis is merelya functionof the availability of suitable building materials; thisassertion could be confirmedwith geologic studies. Justas we use theabsence of core-and-veneer masonry to setan initialtem- poralboundary on the ClassicBonito phase (Hawley1934, 1938), we can use the absenceof core-and-veneermasonry for setting spatial boundaries on the Chaco world.Obviously these boundaries are not rigid, but because the mason- 136 RuthM. VanDyke ry techniqueshould be passed throughcommon learningframeworks and sharedlabor pools, I would argueone is morelikely to see non-Chacoansites withcore-and-veneer, than Chacoan sites that lack the masonry style. The impor- tanceof core-and-veneeras a Chacoan markerincreases during the post-Chaco, PuebloIII period,when it is significantlyassociated with multiple road segments and verylow kiva/roomratios. Addingkiva/room ratio to themix clarifies the picture still further. The typ- ical Late Pueblo II Chacoan greathouse has a kiva/roomratio of about 1:15. Movinginto the 1100sand later,some greathouse kiva/roomratios become quiteextreme; for example, Salmon only has one kivafor 175 rooms.These low ratiosreflect the functions of both kivasand roomsin Chacoan greathouses- whateverwas goingon at thesestructures (see Durand,this volume), one appar- entlyneeded fewkivas and lots of rooms.This patternis in starkcontrast to those sites with kiva/roomratios of 1:8, or even higher.Remember, Lipe's (1989:56, Table 1) non-Chacoan,domestic Mesa Verdeansites had kiva/room ratiosof 1:6.5. Clearly,kivas in thesedomestic sites have different, perhaps less integrativefunctions. Following Lipe, I would arguethat sites with very high kiva/roomratios, perhaps less than 1:10, are less likely to be Chacoan. Thingsget really interesting when the pattern of sites with high kiva/room ratiosis juxtaposedwith the pattern of sites that lack core-and-veneer masonry. I would arguethat these are the sites least likely to be Chacoan.Given the statisti- callysignificant associations between these two variablesdiscovered here, the argumentshould hold particularlywell forPueblo III sites.Of 100 sitesin the database where both variablesare tabulated,12 sites lack core-and-veneer masonryand havekiva/room ratios higher than 1:10. As one mightexpect, these aremainly late sites in theFour Comers or in Arizona.Whatever was goingon in the BlackMesa and LittleColorado areas,I would argue,it is not something directlyrelated to Chaco. Not everybig siteis bestdescribed or understoodas a "Chacooutlier."

CONCLUSION

Thisstudy, then, is an encouragingstep towards putting some spatial boundaries aroundthe Chaco world.These boundaries are permeable,not rigid,of course, and theychange over time. Bonito style architectural attributes occur together in theChacoan core area during the Late Pueblo II period.Landscape modifications and multiplegreat kivas represent the Chacoan package later in timeand further afield.The absenceof core-and-veneermasonry together with high kiva/room ratiosseem a usefulway to definethe geographic scope of Chacoan influence. By the Pueblo III period,the largesites tabulated here are eitherhuge, core-and- veneerpueblos containing multiple road segments,a greatkiva or two,and one or two (probablyelevated) kivas...or they are huge pueblos withlots of little BoundingChaco 137 kivas,lacking core-and-veneer masonry. The firstgroup of sites,I argue,can be considereddirectly connected in someway back to ClassicBonito phase Chaco. The secondare something different. I thinkwhat we areseeing across time and spaceis thespread of an iconic style,affiliation with the Chacoan package-and not everybodywas a member. Massive,visually impressive architectural features obviously contained symbolic meaningsfor prehistoricbuilders and viewers-some outliercommunities sharedcommon meanings with Chaco and each other,and some did not.Are the less Chacoan-lookingsites emulators, or are theyoutside the networkalto- gether?We do not presentlyhave enoughof a handle on internalvariables to determinethe nature of the relationship these "outsiders" had withChaco. How- ever,with better data on internalvariables such as bandingand kivafloor fea- tures,and withbetter temporal resolution, we should be able to improveour understandingof interactions inside the Chaco world.

ACKNOWLEDGMENTS

This studyemerged from the stimulatingand thought-provokingChaco World summitorganized by JohnKantner, Nancy Mahoney,and KeithKintigh in September1999 at ASU. My thanks to them,and to TimPauketat and Dennis Gilpinfor encouraging me to revisitmy dissertation project using the expanded database.Gwinn Vivian, Barbara Mills, and ChuckAdams provided invaluable guidanceand supportfor the original project. John Kantner, Keith Kintigh, Peter McKenna,Ron Towner, and Tom Windes kindly offered critical commentary at variousstages of manuscriptcompletion, and thearticle is, I hope,stronger as a result.Any errors are, of course, my own responsibility.

REFERENCES Bretemitz,Cory D., DavidE. Doyel,and Michael P. Marshall 1982 Bissa'ani: A LateBonito Phase Community on Escavada Wash, Northwest New . NavajoNation Papers in Anthropology14. NavajoNation Cultural Resource Man- agementProgram, Window Rock. Cameron,Catherine M. 2002 SacredEarthen Architecture inthe Northern Southwest: The Bluff Great House Berm. AmericanAntiquity 67(4):677-695. Carr,Christopher 1995 A UnifiedMiddle-Range Theory of Design. In Style,Society, and Person: Archae- ologicaland EthnologicalPerspectives, edited by C. Carrand J.Neitzel, pp. 171-258. PlenumPress, New York. Dobres,Marcia-Anne 2000 Technologyand Social Agency: Outlining a Practice Framework forArchaeology. Blackwell, Oxford. Doyel,David E., Cory D. Breternitz,and Michael P. Marshall 1984 ChacoanCommunity Structure: Bis sa'ani and the Chaco Halo. In RecentResearch on ChacoPrehistory, edited by W. J. Judge and J. D. Schelberg,pp. 37-54. Reportsof the ChacoCenter 8. Divisionof Cultural Research, , Albuquerque. 138 RuthM. VanDyke

Fowler,Andrew P., and John R. Stein 1992 TheAnasazi Great House in Space, Time, and Paradigm. In Anasazi Regional Organiza- tionand the Chaco System, edited by D. E. Doyel,pp. 101-122.Maxwell Museum of AnthropologyAnthropological Papers 5. Albuquerque. Fowler,Andrew P., John R. Stein, and RogerAnyon 1987 AnArchaeological Reconnaissance ofWest-Central : The Anasazi Monuments Project.Office of Cultural Affairs, Historic Preservation Division, Albuquerque. Gladwin,Harold S. 1945 TheChaco Branch: Excavations atWhite Mound and in theRed Mesa Valley. Medal- lionPapers 33. GilaPueblo, Globe. Hawley,Florence M. 1934 TheSignificance ofthe Dated Prehistory ofChetro Ketl, Chaco Canyon, New Mexico. Monographsofthe School of American Research 2, SantaFe. 1938 TheFamily Tree of Chaco Canyon Masonry. American Antiquity 3(3):247-255. Hurst,Winston B. 2000 Chaco Outlieror BackwoodsPretender? A Provincial Great House at Edgeof the CedarsRuin, . In GreatHouse Communities across the Chacoan Landscape, edited byJ. Kantner and N. Mahoney,pp. 63-78.Anthropological Papers of the University ofArizona 65. The University ofArizona Press, Tucson. Irwin-Williams,Cynthia, and Phillip H. Shelley(editors) 1980 Investigationsat the Salmon Site: The Structure ofChacoan Society in the Northern South- west.Eastern New Mexico University Printing Services, Portales. Jalbert,Joseph Peter, and Catherine M. Cameron 2000 Chacoanand LocalInfluences in ThreeGreat House Communities in theNorthern San JuanRegion. In GreatHouse Communities across the Chacoan Landscape, edited by J.Kantner and N. Mahoney,pp. 79-90.Anthropological Papers of the University of Arizona65. The University ofArizona Press, Tucson. Lechtman,Heather 1977 Stylein Technology-SomeEarly Thoughts. In MaterialCulture, Styles, Organization, and Dynamicsof Technology, edited by H. Lechtmanand R. Merrill,pp. 3-20. West Publishing,New York. Lekson,Stephen H. 1984 GreatPueblo Architecture ofChaco Canyon, New Mexico. University of New Mexico Press,Albuquerque. 1991 SettlementPatterns and the Chaco Region. In Chacoand : Prehistoric Region- al Systemsinthe American Southwest, edited by P. L. Crownand W. J. Judge, pp. 31-55. Schoolof American Research Press, Santa Fe. Lemonnier,Pierre 1986 TheStudy of Material Culture Today: Toward an AnthropologyofTechnical Systems. JournalofAnthropological Archaeology 5:147-186. Lipe,William D. 1989 SocialScale of Mesa Verde Anasazi Kivas. In TheArchitecture ofSocial Integration inPre- historicPueblos, edited by W. D. Lipeand M. Hegmon,pp. 53-71.Occasional Papers ofthe Crow Canyon Archaeological Center 1. Cortez,Colorado. Lipe,William D., andMichelle Hegmon (editors) 1989 TheArchitecture ofSocial Integration in Prehistoric Pueblos. Occasional Papers of the CrowCanyon Archaeological Center 1, Cortez, Colorado. Marshall,Michael P., and Anna. Sofaer 1988 TheSolstice Project: Archaeological Investigations in the Chacoan Province, New Mexico. Ms.on file,Laboratory ofAnthropology, Santa Fe, New Mexico. Marshall,Michael P., John R. Stein, Richard W. Loose, and Judith E. Novotny 1979 AnasaziCommunities ofthe San Juan Basin. Public Service Company of New Mexico, Albuquerque. Morris,Earl H. 1919 TheAztec Ruin. Anthropological Papers of the American Museum of Natural History 26, Part1, New York. BoundingChaco 139

Powers,Robert P., William B. Gillespie,and Stephen H. Lekson 1983 TheOutlier Survey: A Regional View of Settlement in the San Juan Basin. Reports of the ChacoCenter 3. U.S.Department ofthe Interior, National Park Service, Albuquerque. Renfrew,Colin, and John F. Cherry (editors) 1986 PeerPolity Interaction and Socio-politicalChange. Cambridge University Press, Cam- bridge. Roberts,Frank H. H.,Jr. 1932 TheVillage of the Great Kivas on theZuni Reservation, New Mexico. Bureau of Ameri- canEthnology Bulletin 111. Washington, D.C. Roys,Lawrence 1936 Masonryof Lowry Ruin and of the Southwest. In LowryRuin in Southwestern Colorado, byP. S. Martin,pp. 115-142.Field Museum of Natural History Anthropology Series 23 (1). Chicago. Stein,John R., and Stephen H. Lekson 1992 AnasaziRitual Landscapes. In AnasaziRegional Organization and theChaco System, editedby D. E. Doyel,pp. 87-100.Maxwell Museum of Anthropology Anthropolog- icalPapers 5. UniversityofNew Mexico, Albuquerque. Steward,Julian 1937 EcologicalAspects of Southwestern Society. Anthropos 32:87-104. VanDyke, Ruth M. 1998 TheChaco Connection: Bonito Style Architecture in Outlier Communities. Ph.D. disserta- tion,The University ofArizona, Tucson. UMI, Ann Arbor. 1999 The ChacoConnection: Evaluating Bonito Style Architecture in Outlier Communi- ties.Journal ofAnthropological Archaeology 18(4):471-506. 2000 ChacoanRitual Landscapes: The View from the Red Mesa Valley. In GreatHouse Com- munitiesacross the Chacoan Landscape, edited by J.Kantner and N. Mahoney,pp. 91-100.Anthropological Papers of the University ofArizona 65. The Universityof ArizonaPress, Tucson. 2002 The ChacoanGreat Kiva in OutlierCommunities: Investigating Integrative Spaces acrossthe San Juan Basin. Kiva 67(3):231-248. 2003 Memoryand the Construction ofChacoan Society. In Archaeologies ofMemory, edited byR. M. Van Dyke and S. E. Alcock,pp. 180-200.Blackwell Publishers, Oxford, and Malden,Mass. Wills,W. H. 2001 Ritualand MoundFormation During the Bonito Phase in ChacoCanyon. American Antiquity66(3):433-451.

RuthM. VanDyke Departmentof Anthropology ColoradoCollege ColoradoSprings, CO 80903 (719)389-6362 [email protected]