19
A pplication of a dating methodology that integrates consideration of interactions between site formation and data retrieval has resulted in the detection of high incidences of multiple components and short reuse episodes on sites of both mobile groups and sedentary farmers. Mobile group reuse of sites previously interpreted as single- use locations is distinguishable, as is their ephemeral use of sites once occupied by more sta- tionary residents. These data reveal that mobile group occupations are quite often masked by our conceptual and methodological frameworks. Noticeably, application of the principles and approaches common to and suitable for sedentary systems can be inappropriate, and in fact, detri- mental in the archaeological study of mobile group sites. The customary practice of age- averaging mul- tiple chronometric dates obtained from a single site masks multi- componentcy and episodic reuse. Expectations for “contextual congruence” and the standard practice of eliminating apparent “anom- alous” results preclude the possibility of discover- ing the unexpected, identifying the unusual, and uncovering individual short- lived visitation events. Additionally, misreading of chronometric data may result from reliance on a single dating technique that may not be appropriate when addressing com- plex occupational sequences. When various chronometric techniques, includ- ing radiometric, are handled as dating methods— rather than simply as useful techniques— they can be considered and applied within the framework of informed theory (e.g., Dean 1978). Yet today, samples are often sought and results are applied and interpreted using accumulated knowledge and implicit models of place- use that are appropriate to sedentary rather than ephemeral occupations. This is especially problematic when attempting to identify and appropriately analyze the often- faint traces of mobile groups on landscapes and sites also CONTEXTUAL INCONGRUITIES, STATISTICAL OUTLIERS, AND ANOMALIES: TARGETING INCONSPICUOUS OCCUPATIONAL EVENTS Deni J. Seymour New methodologies are needed to address multiple componentcy and short reuse episodes that are characteristic of mobile group residential and logistical strategies. Chronometric results are often misinterpreted when evaluated within a frame- work suited to long- term sedentary occupations. The standard practices of age- averaging, eliminating apparent “anom- alous” results, and relying on high profile diagnostic tools and vessels and the most visible features— along with the expectation for “contextual congruence”—mask multi- componentcy and episodic reuse. High incidences of site reuse have been detected by considering alternate site development models and looking specifically for evidence of distinct shorter term occupations. Se necesita nuevas metodologías para hacer frente a sitios con componentes múltiples y episodios breves de reutilización que son característicos de las estrategias residenciales y logísticas de los grupos nómadas. De vez en cuando, resultados cronométri- cos son mal interpretados cuando se les evaluó con un marco adecuado para ocupaciones sedentarias a largo plazo. Las prác- ticas usuales de “la edad promedio,” la eliminación de aparente “resultados anómalas” y confiando en diagnósticos de alto perfil y los rasgos más visibles, junto con la expectación de “congruencias contextuales,” oculta componentes múltiples y episodios de reutilización. Alta incidencias de la reutilización de los sitios han sido detectadas por considerar modelos alter- nativos del desarrollo de los sitios, buscando específicamente evidencia para distintas ocupaciones de breve duración. Deni J. Seymour The Southwest Center, University of Arizona, 1052 North Highland Ave., University of Arizona, Tucson, Arizona 85721-0185 ([email protected]) American Antiquity 75(1), 2010, pp. 158–176 Copyright ©2010 by the Society for American Archaeology 158 AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 158

contextual incongruities, statistical outliers, and anomalies

  • Upload
    doxuyen

  • View
    216

  • Download
    0

Embed Size (px)

Citation preview

Page 1: contextual incongruities, statistical outliers, and anomalies

Application of a dating methodology thatintegrates consideration of interactionsbetween site formation and data retrieval

has resulted in the detection of high incidences ofmultiple components and short reuse episodes onsites of both mobile groups and sedentary farmers.Mobile group reuse of sites previously interpretedas single- use locations is distinguishable, as is theirephemeral use of sites once occupied by more sta-tionary residents. These data reveal that mobilegroup occupations are quite often masked by ourconceptual and methodological frameworks.Noticeably, application of the principles andapproaches common to and suitable for sedentarysystems can be inappropriate, and in fact, detri-mental in the archaeological study of mobile groupsites. The customary practice of age- averaging mul-tiple chronometric dates obtained from a single sitemasks multi- componentcy and episodic reuse.Expectations for “contextual congruence” and the

standard practice of eliminating apparent “anom-alous” results preclude the possibility of discover-ing the unexpected, identifying the unusual, anduncovering individual short- lived visitation events.Additionally, misreading of chronometric data mayresult from reliance on a single dating techniquethat may not be appropriate when addressing com-plex occupational sequences. When various chronometric techniques, includ-

ing radiometric, are handled as dating methods— rather than simply as useful techniques— they canbe considered and applied within the frameworkof informed theory (e.g., Dean 1978). Yet today,samples are often sought and results are applied andinterpreted using accumulated knowledge andimplicit models of place- use that are appropriateto sedentary rather than ephemeral occupations.This is especially problematic when attempting toidentify and appropriately analyze the often- fainttraces of mobile groups on landscapes and sites also

CONTEXTUAL INCONGRUITIES, STATISTICAL OUTLIERS, ANDANOMALIES: TARGETING INCONSPICUOUS OCCUPATIONAL

EVENTS

Deni J. Seymour

New methodologies are needed to address multiple componentcy and short reuse episodes that are characteristic of mobilegroup residential and logistical strategies. Chronometric results are often misinterpreted when evaluated within a frame-work suited to long- term sedentary occupations. The standard practices of age- averaging, eliminating apparent “anom-alous” results, and relying on high profile diagnostic tools and vessels and the most visible features— along with theexpectation for “contextual congruence”—mask multi- componentcy and episodic reuse. High incidences of site reuse havebeen detected by considering alternate site development models and looking specifically for evidence of distinct shorterterm occupations.

Se necesita nuevas metodologías para hacer frente a sitios con componentes múltiples y episodios breves de reutilización queson característicos de las estrategias residenciales y logísticas de los grupos nómadas. De vez en cuando, resultados cronométri-cos son mal interpretados cuando se les evaluó con un marco adecuado para ocupaciones sedentarias a largo plazo. Las prác-ticas usuales de “la edad promedio,” la eliminación de aparente “resultados anómalas” y confiando en diagnósticos de altoperfil y los rasgos más visibles, junto con la expectación de “congruencias contextuales,” oculta componentes múltiples yepisodios de reutilización. Alta incidencias de la reutilización de los sitios han sido detectadas por considerar modelos alter-nativos del desarrollo de los sitios, buscando específicamente evidencia para distintas ocupaciones de breve duración.

Deni J. Seymour � The Southwest Center, University of Arizona, 1052 North Highland Ave., University of Arizona,Tucson, Arizona 85721-0185 ([email protected])

American Antiquity 75(1), 2010, pp. 158–176Copyright ©2010 by the Society for American Archaeology

158

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 158

Page 2: contextual incongruities, statistical outliers, and anomalies

REPORTS 159

occupied by more sedentary groups. Focus has beenon the most robust component in stratified sitesrather than the many less conspicuous short- termoccupations that succeed it. Additionally, expecta-tions are that “ limited- use” sites will not presentevidence of periodic and sequential reuse and sointerpretations have tended to focus on a singleoccupational event rather than a diachronic suc-cession of events. These problems are compoundedby the fact that many chronometric techniques arenot sufficiently precise to parse multiple components— much less ferret out short- term occu-pations or to differentiate episodes of reuse— asoccur under conditions of high residential mobil-ity or reuse of logistical locations. A revised con-ceptual framework and minor adjustments in theapplication of available dating techniques result inaccess to new forms of knowledge. These issues are discussed using case studies

from the American Southwest (Figure 1) but are

applicable to problems throughout the world wheresimilar degrees of mobility prevail. This article is distilled from an on- going sys-

tematic effort to address the problem of detectingand dating the ephemeral traces of mobile occu-pations during the Terminal Prehistoric and Historicperiods (A.D. 1300–1800). The earlier portion ofthat time frame is often referred to as the Proto-historic. Remains of Protohistoric occupations areoften superimposed on, or intermixed with, themore obtrusive remains of earlier sedentary groups,and the methods discussed here are aimed at unrav-eling evidence of successive short- term uses onthese intensively occupied sites as well as on sitesthat may present little evidence of use overall.Expectations for differences in periodicity, dura-tion, and intensity of occupation have been derivedfrom the ethnographic and ethnohistoric recordsand from the development of theory aimed at ageneral archaeology for the study of mobile groups.

Figure 1. Map of Southern Southwest with sites and regions mentioned in text. (1) Jornada Mogollon Culture Area is asdefined by Lehmer (1948), although some consider its boundaries to continue east into the Trans-Pecos area (e.g., Corley1965). (2) Pintada Rockshelter. (3) FB 9423, North Otero Mesa Tipi Ring Site. (4) 41EP401 (FB 16715) in the FranklinMountains. (5) 1800s Apache Site in the Dragoon Mountains. (6) Llano Estacado. (7) Salinas Pueblo area. (8) Cerro RojoSite. (9) Sharples Site.

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 159

Page 3: contextual incongruities, statistical outliers, and anomalies

Distinctions between the Late and Terminal Pre-historic, Initial Contact, and Early Historic periodsillustrate recent advances in our understanding ofthe nature and timing of changes during this inter-val. Throughout the American Southwest, funda-mental transformations are recognized in local andregional systems in the 1200s (Late Prehistoric)when the predominant pattern of pit- house use isreplaced by compounds and pueblos (Cordell1984:328–337; Reid and Whittlesey 1997:100–101, 146, 190–191; Rouse 1962:27, 31). Thesedevelopments are concomitant with formation ofpopulation aggregations and intensification ofincreasingly complex agricultural systems that col-lapsed or reorganized around the A.D. 1400s andthen were replaced by Protohistoric groups andsystems (Cordell 1984:328–337, 360; Reid andWhittlesey 1997:107–110, 161–165, 198; Rouse1962:31, 40–48). This traditional view, however, may require revi-

sion. An ever enlarging set of chronometric datesfound in direct association with distinct materialculture sets is beginning to indicate that Protohis-toric groups with at least three general life wayswere present in the southern Southwest by or beforethe 1400s. These were the sedentary Sobaípuri- O’odham Cayetano complex, along with the earlyAthapaskan Cerro Rojo complex and a variety ofCanutillo complex mobile groups (see Seymour2002, 2004, 2008a, 2009a). Athapaskan presenceapparently preceded the other Protohistoric groupswith dates in the 1300s, if not the late 1200s (seeSeymour 2007a, 2007b, 2008a, 2008b, 2009b).This means that some of these historically knowngroups overlapped with prehistoric farmers andeither contributed to, witnessed, or experiencedtheir disruption.Like the traditional view that Protohistoric peo-

ples replaced already vanished sedentary groups,our views of these groups themselves also requirerevision. When mentioned in the 1500s by the firstEuropeans, many of these groups were very dif-ferent than those described in the 1600s, suggest-ing substantial changes had already occurred,making use of direct historical approaches anduncritical use of documentary sources imprudent(Seymour 2007a, 2008a, 2008b, 2009b). Thearchaeology of this Terminal Prehistoric and ear-liest Contact interval provides a view of the natureand timing of change that revises the later historic

record. Use of chronological terms such as “Ter-minal Prehistoric” and “Initial Contact” rather than“Late Prehistoric” and “Protohistoric” refocusesinvestigations on this critical period, to shorter inter-vals of time, and separates this sequence of events,processes, and the key players from the better- known prehistoric groups. An ever- increasing data-base of chronometric dates from an enlarging rosterof sites from the Terminal Prehistoric and InitialContact time frame provide a basis for under-standing a range of persistent research issues,including the presence and role of the less con-spicuous groups in the procession of local and large- scale events. New data combine with exist-ing and revised approaches of interpretation to forma new view of mobile groups who occupied por-tions of the Southwest as Europeans first encoun-tered the region.

Contextual Congruence: Diachronic Reoccupation vs. Anomalies

Dean (1978:251) discusses the principle of con-textual congruence, which posits that

dating anomalies may be detected on the basisof apparent inconsistencies between the datesand other archaeological data. [It is the archae-ologist’s] reservoir of accumulated knowledgethat enables him to place the site into some sortof temporal, spatial, and relational framework.This frame of reference permits the recogni-tion of dates that are inconsistent with theexpected temporal position of the site.

While seeking contextual congruence is commonpractice and fundamental to interpreting and eval-uating dates, expectations for contextual congru-ence create one avenue of interpretive misdirectionand can impede our understanding of mobile groupsites, making it difficult to recognize theseephemeral components. Preconception of a spe-cific occupation range creates a set of secondaryassumptions about the interrelation of features andartifacts in the archaeological context. These pre-conceptions may remain set, even in the face ofchronometric evidence to the contrary. Someresearchers have discarded dates that produced con-sistent results, even from two separate chronomet-ric techniques, in favor of aligning a preconceivedoccupational span with the presence of artifacts

160 amERican anTiquiTy [Vol. 75, no. 1, 2010

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 160

Page 4: contextual incongruities, statistical outliers, and anomalies

believed to be diagnostic (see Seymour2008c:Table 2). The existing conceptual model causes results to

be interpreted with respect to existing understand-ings of the data. It also assumes that when there aredivergent dates the prudent scholar will excludethese as anomalous because they do not conformto expectations derived from existing knowledge(e.g., Seymour 2008c). Based upon probabilities,statistical outliers can be reasonably excludedbecause they are not likely to relate to or inform onthe dominant or most easily recognized component.Yet, these statistical outliers may be the first hintof a secondary component or an unexpected event.A revised working model would encourage prac-titioners to also explore the possibility that the sup-posed anomalous dates provide evidence of anoccupational scenario that differs from expecta-tions. This involves entertaining alternative mod-els of site formation that allow for extremely subtleindications of multiple components associated withrepeated occupations, including later occupationsin the fill of prehistoric structures. This places the chronometric data requirement

higher up on the list of research issues to beaddressed during site investigations. It also neces-sitates a flexible research design that recognizes therole of dates in understanding other research issuesand that can accommodate a change in course wheninitial expectations are not met. Dates run concur-rent with fieldwork or between seasons allow timeto incorporate results that may take 6 to 12 monthsto receive, and enable the researcher to objectivelyrule out archaeological evidence of another com-ponent before rejecting the results because they aretruly anomalous. With access to results while field-work is underway, researchers can look specifi-cally for evidence of repeated use while still in thefield; that is, when appropriate methodologies canbe applied to the collection of these often unob-trusive data. This procedure has been instrumental in defin-

ing Terminal Prehistoric and Early Historic com-plexes throughout the southern Southwest. Thesecomplexes are known historically to have occurredbut until recently have defied archaeological recog-nition; in these cases in particular, the timely col-lection of chronometric dates during ongoingfieldwork has been crucial to the identification ofindividual and sequential mobile occupational

events. In many instances, the availability of dates,where previously there had been few or none, hasfundamentally changed understandings of the localor regional sequence.Concentration on the “richer” archaeological

record of sedentary peoples has also impeded animproved understanding of mobile groups. Whenlacking accurate accumulated knowledge of sitesand components left by mobile peoples, even themost parsimonious inferences can be biased towardthe better- known, more intensively studied groups,such as sedentary farmers. This is the risk of resort-ing to “common sense” in the face of anomaliesthat call for rigorous analysis. Schiffer (1987:309)has added to Dean’s (1978) earlier work in warn-ing against this intellectual trap. As Schiffer (1987) notes, “the traditional way

to deal with conflicting (i.e., variable) dates is toselect only those dates that are in agreement withone’s prior positions on chronological issues; thewidely practiced art of ‘accepting’ and ‘rejecting’dates” (1987:309). This practice is particularlyproblematic if the site lacks well- defined and con-tinuous strata— the connective tissue that allowsobservation of interruptions, intrusions, andsequential occupations. The often- subtle indices of multi- componentcy and reuse are most likely to bemissed when the researcher has pre- packaged a setof chronological assumptions concerning a site’soccupational history. Such bundled conceptualframeworks are difficult to unravel because they areso handy to tote around. But as Jeffrey Dean (per-sonal communication 2006) notes,

The purpose of identifying potential anom-alous dates (which are anomalous relative tosome defined target event), through the appli-cation of contextual congruence or any otherprinciple, is not to ignore or discard them.Rather, it is to highlight apparent irregularitiesthat have to be explained. Explanation shouldlead to the identification of erroneous dates(which do exist) or of otherwise unrecognizedhuman activities (multiple use episodes, etc.)at the site.

It is precisely the failure to “highlight apparent irreg-ularities that have to be explained” that has impededrecognition of the characteristic signatures of mobilegroup occupations, both on sites occupied only bymobile groups and those reoccupied by mobile

REPORTS 161

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 161

Page 5: contextual incongruities, statistical outliers, and anomalies

groups after sedentary occupations. Regrettably, multiple occupancy and multiple

componentcy are generally not expected on siteswith limited accumulations of cultural material,which sometimes also have barely perceivable fea-tures. Yet, mounting evidence from dozens ofrecently dated mobile group components is show-ing that even sites consisting of thin shallowdeposits, low accumulation indices, and vestiges ofinsubstantial features possess evidence of periodicreuse and may actually be persistent places.Methodological approaches appropriate to thetempo, scale, periodicity, and duration of mobilelandscape use are beginning to isolate mobile groupcomponents on sites that experienced multipleoccupancies. By taking into account gradations ofmobility and applying innovative approaches, it isnow possible to distinguish between (a) short- termepisodic use, (b) entirely transient one- time use, and(c) persistent episodic use over a short or longperiod. Greater progress has also been made inidentifying mobile group components that are posi-tioned over complex sedentary sites that were usedfor decades, if not centuries, even when these lay-ers of use are compacted into thin shallow strata orcollapsed onto a single surface as palimpsests. Con-textual incongruities, often in the form of anom-alous dates, are now seen as gateways to discoveryof mobile occupational events, revealing adiachronically dynamic cultural landscape. Con-textual incongruities that are portents of distinctoccupational events can, in due course, be con-firmed by other forms of data.

Anomalous Dates Case Study: Pintada Rock Shelter

Pintada Rock Shelter, located along Otero Mesa insouthern New Mexico, illustrates the problems thatoccur when one ignores episodic use sequencesand dismisses dates that deviate from a priori expec-tations. Investigating the Archaic occupation of thesite, MacNeish (1998, *2003a) dismissed eight lateradiocarbon and obsidian hydration dates that didnot fit his reconstruction of the occupational his-tory. The dates derived from a mixed layer thatoverlaid the Jornada Mogollon component in ZoneA in the upper 20 cm of sediments. Projectile pointsand bifaces diagnostic of the non- Athapaskanmobile Canutillo complex (Seymour 2002, 2009a)correspond well with some of the dates from this

zone, clearly indicating a Protohistoric mobileoccupation (Seymour and Church 2007). Yet,because the diagnostic attributes of this culturalphenomenon had yet to be defined, MacNeish(1998:165) assumed that the finely crafted toolsrepresented the Archaic occupants— who also pro-duced formal leaf- shaped bifaces that are onlysuperficially similar to the Protohistoric diagnos-tics (Seymour 2009a). He dismissed the late datesas “unfortunate” and “erroneous,” and in doing sooverlooked the Terminal Prehistoric component,collapsing the late- occurring artifacts into hisArchaic analysis (see Seymour and Church 2007).Had he attempted to find an explanation for theassociation of late dates with finely crafted tools,he might have been the first to determine that theyare clear indicators of a yet- to- be- defined Termi-nal Prehistoric (Protohistoric) Canutillo complexoccupation.

Sampling Bias of Sure Bets and Best Cases

At times, problems arise from the criteria used forselection of materials for dating rather than fromthe elimination of dates that have been run. Somesites produce so little datable material that allpromising samples are submitted. It is also com-mon, however, especially on multiple componentsites, to have a larger sampling universe with whichto work. In such cases sampling may be biasedwhen the researcher focuses on “strong patterns”(Plog 1984; Upham 1994), which “emphasizeobtrusive archaeological remains and conspicuouselements of the archaeological record, even if thereis also a low visibility record of human occupationin the region” (Upham 1994:118). These strongpatterns or best cases are contexts and chronomet-ric samples that are deemed most likely to providea usable or expected result and also to confirm andvalidate target events. For example, when follow-ing normal archaeological chronometric sample- selection criteria, mostly large thermal features withan abundance of clearly definable charcoal or thedarkest fill will be selected because these haveproven time and again to present definitive andexpected results. These features are favored becausethey are single- function (although often multiple- event), and are often predictably situated in partic-ular landscape settings and resource zones. Yetselection of these types of features can lead to a

162 amERican anTiquiTy [Vol. 75, no. 1, 2010

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 162

Page 6: contextual incongruities, statistical outliers, and anomalies

bias against datable features of other periods that(a) occur in different landscape settings, (b) wereused for different purposes, or (c) leave less obvi-ous signatures. When seeking evidence for Protohistoric mobile

group occupations, the “best cases” are defined asthe mirror opposite of the traditional model devel-oped for sedentary groups. Mobile group thermalfeatures are often more difficult to identify than thoseleft by sedentary occupations or repeated reoccupa-tions in the Archaic. This is due to the use of smallerpieces of fuel (brush, grass, twigs, etc.) that create a less- distinguishable deposit than wood charcoal(Seymour 1995, 2002, 2003a); the desirability ofsmaller fires so as not to attract attention (Seymour1995, 2002; Sweeney 1997:72); foodways involv-ing cooking techniques that do not demand sustained high- temperatures (Ayer 1965:14; Beckett and Cor-bett 1992:30; Treutlein 1965:53); preparation of sta-ples in roasting pits situated at some distance fromthe habitation setting (Opler 1996:357); and so on(Seymour 1995, 2002, 2005b, 2007c). Consequently,sample selection must involve thoughtful consider-ation of the ways in which materials, contexts, andtechniques were put to work by people practicingvery different lifeways.Because of their lower visibility, these types of

features are often not recognized— which is a chal-lenge that can be particularly problematic on mul-tiple component sites. Thermal features that aremore noticeable (because they contain large quan-tities of fire- cracked rock or dark fill from repeateduse) tend to be selected over the vaguest features.Yet, many times isolated deflated burned- rock scat-ters, vague charcoal- stained sediments, or charcoal- infused ash deposits reflect dates many centuriesapart from the more well- defined and obvious fea-tures of associated components, but their distinctnature becomes readily apparent only if they areselected for dating. The lower frequency andsmaller size of datable material in these featuresmeans that they are often bypassed for datingbecause they are either not noticed or, when facinga limited budget, they are not chosen because theyare more costly to run or are not as certain to returna date. A benefit of these less- obtrusive features isthat the use of brushy materials (short- lived species)means that they often do not tend to suffer from the“old wood” problem (Schiffer 1982, 1986; Sey-mour 2003a, 2005b).

“Sampling Bias” Case Study: The End of the Jornada Mogollon

The sampling bias toward large thermal featuresobtrusively colocated with the most densely dis-tributed artifact scatters can color the characteri-zation of entire occupational sequences. Theinadvertent bias away from sampling Terminal Pre-historic and Early Historic period features resultsfrom a lack of understanding mobility: degrees ofmobility; expected duration of stay; fuel choices;foodway customs; site reuse; and desire to main-tain low visibility in the face of enemy threat. Theseand other aspects of mobile patterns affect the waydates must be sought, selected, and interpreted. An example of the consequence of not consid-

ering the effects of sampling strategies is again pro-vided from the Jornada Mogollon area of southernNew Mexico and southwest Texas (e.g., Trans- Pecos). Miller (2001:106) notes a general down-ward trend in the number of datable contexts fromA.D. 1400 to 1500, a period that spans the Termi-nal Prehistoric and Early Historic periods in the Jor-nada Mogollon area. More recent work (Seymour2002, 2003a, 2004, 2007c; Seymour and Church2007) indicates continued occupation of the Jor-nada area at levels comparable to earlier periods.This occupation, however, manifests in differenttypes of features, contexts, and landscape settingsthan the earlier periods, and leaves a lighter foot-print. By looking mostly in the valleys (where mostcontract work has been conducted) the abundantevidence in the adjacent mountains and foothillshas been removed from the equation. This practicecontinues despite the fact that the ethnohistoric andethnographic records indicate these nonvalley loca-tions tended to be the focus of activity. Moreover,by ignoring intrusive features, incongruent dates,and out- of- place artifacts reuse of valley- based sitesis not being recognized.Expectations for use by a single culture group

for a specified period of time on Jornada Mogollonsites reduces the ability to recognize the unobtru-sive evidence of later use by visitors with a differ-ent footprint who might use the site and its featuresin a different way and who organize their space dif-ferently, including in a less- focused manner. Thesebiases can be counteracted by understanding thebehavioral context of features themselves and thevaried lifeway of the region’s more mobile residentsduring these later periods. The effect can also be neu-

REPORTS 163

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 163

Page 7: contextual incongruities, statistical outliers, and anomalies

tralized by recognizing the proclivity for these latefeatures on multicomponent sites to be passed overfor sampling and dating.

Redefining “Diagnostic”

A narrow definition of what constitutes a tempo-rally or culturally diagnostic artifact exacerbates thedifficulties created by dismissing contextual incon-gruities when assigning dates to components. Thisproblem is enmeshed in how diagnostics are clas-sified and the weight given them when assessingtemporal assignments. For example, it is commonknowledge that at a general and practical level,Southwestern diagnostics tend to include decoratedpottery and exclude plainwares, and include pro-jectile points while generally rejecting other toolforms owing to the belief that only a limited rangeof items have a distinctive form or style (e.g., Glad-win and Gladwin 1935; Hole and Heizer1977:173–182; Smith 1976:70–74). The problemof distinguishing a light mobile group footprintbegins with a reassessment of artifact attributes thatindicate mobile group presence. The bias toward pottery and projectile points, and

the strong emphasis on decorated pottery as anemblem of cultural identity, means that mobile sitesand components will be rendered invisible or willbe misinterpreted as limited- use settings created by pottery- producing sedentary groups. One reason forthis is that mobile groups rarely used or made pot-tery and when they did it tended to be plainware.Further, the pottery found on mobile group sites wasoften obtained from other groups, so, as I have pre-viously noted (Seymour 2002, 2003a, 2008d),assignment of cultural affiliation on the basis of afew sherds will often miss- identify the actual occu-pants. The assemblage for the component as a wholemust be examined, and an expectation for an amal-gamation of material from more than one groupincorporated into interpretative frameworks. It isnecessary to consider the succession of events affect-ing the distribution of incongruent pottery in mobilegroup settings. This paradoxically depends upon theability to recognize its incongruence, which in turnis dependent upon consideration of models appro-priate to mobile occupation and the ability to rec-ognize other indices of mobility and raiding.The abundance or distinctive nature of material

culture present is often overemphasized and is used

as confirmation of cultural and temporal affiliation(see Shepard 1956 for a discussion of reliability ofthe abundance criterion). This biases interpreta-tions toward assigning occupations to groups whoproduced and discarded more material culture— especially material culture that the archaeologistcan readily assign to a type (also see Upham1984:239–240). This is most pronounced when thematerial culture assemblage of the mobile compo-nent is impoverished, as is typically the case, butit is especially problematic on multicomponentsites. Low- density and low- diversity assemblagesmay result for a host of reasons (see Seymour2005b). Fewer items accumulate because of: (a)limitations on what (and how much) can be trans-ported by people on the move (Ball 1970; Opler1996:357, 371); (b) the prevalence of multifunc-tional and more durable tools that limit the num-ber of items potentially broken and left behind(Seymour 2002:323, 2005b); (c) the relatively highpercentage of perishable goods (Opler *1966); and(d) unique discard practices (i.e., the tendency for task- specific items to be discarded or stored at aseries of locations on an occupational circuit; Ball1970; Basehart 1960:104–105; Betzinez and Nye1959; Opler 1974:7; Seymour 2002, 2005b). Undersuch circumstances even small amounts of distinct,typologically secure material representing another better- known or more sedentary group tend toreceive greater emphasis when assessing a site’stemporal and cultural placement. Diagnostic material culture derived from other

groups may occur on mobile group sites for a num-ber of reasons, not just as a result of site reuse. Moreoften than not, material culture from a secondarygroup on a predominantly mobile group site is inci-dental, but because it is diagnostic it is frequentlyassigned greater weight than is warranted, leadingto a misinterpretation of the site’s occupational his-tory. Rather than representing an earlier or lateroccupation, this evidence of other groups may indi-cate: (a) information on the way the mobile groupinteracted with other groups (i.e., raiding or trad-ing; Goodwin 1929–1939; Opler 1996:398); (b)scavenging of prehistoric or contemporaneouscaches and vacant sites (Goodwin and Basso1971:231; Opler 1996:389); or (c) an incidentaloccurrence of an item at a location (such as a strayprojectile point or a pot break) as part of the gen-eral background scatter of artifacts across the land-

164 amERican anTiquiTy [Vol. 75, no. 1, 2010

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 164

Page 8: contextual incongruities, statistical outliers, and anomalies

scape (see Seymour 2002, 2005b for examples).The difficulty is in distinguishing the light imprintof mobile groups when more visible items of mate-rial culture seize the archaeologist’s attention.In such cases, it will be useful to extend our

modes of recognition, analysis, and inference build-ing to match the conditions presented by the archae-ological (and, in many cases, documentary) record.This must be accomplished on its own terms ratherthan expecting the record to conform to a set ofunrealistic or inappropriate models. The docu-mentary (ethnohistoric, ethnographic, and historic)record provides evidence of interethnic contact andunique forms of landscape use relative to earliergroups. When appropriately modeled, these guidescan help lay the groundwork for more accuratearchaeological expectations, especially when thesesources relate information that is in direct referenceto the groups and sites being studied. These can helpus incorporate high- profile diagnostics into ourinterpretive models of site use, including accuratelyassessing data as supplemental or incidental tooccupation by a mobile group. This allows us tosurmount the awkward reality that sites may con-tain both an amalgamation of material from morethan one group and present only a partial and scantset of materials from the target group.

Diagnostics Case Studies: Raiding Sites and Trading

Component invisibility in the face of highly rec-ognizable diagnostic artifacts is an especially pro-nounced problem for raiding and trading groupswho did not themselves create and discard largenumbers of artifacts. For example, when raidinggroups are in retreat they may take little of theirown material culture with them or they may inten-tionally clean up (Haley *1981:81–82; Opler1996:427–428; Seymour 2002:353), which wouldmake themselves less visible. What they do leaveoften includes materials obtained on the raid. Thismaterial, being diagnostic of another group, maybe useful for dating the site but frequently leads toinappropriate cultural assignments, such as whenthe presence of Hopi or Piro sherds on a mobilegroup site (Gilpin and Phillips 1998:25) are usedto assign cultural affiliation to those better- knowngroups. Yet, it is not uncommon to find discardedhorseshoes or non- native accoutrements alongApache raider’s escape routes, as is apparent, for

example, at 41EP401 (FB 16715) in the FranklinMountains near El Paso, Texas (Seymour 2002).Similarly, the most visible evidence on some lateApache sites often includes remnants of discardedcontainers (nails from wooden boxes; sherds frompottery vessels) that were emptied of their valuablecontents to lighten a load— as is evident at a smalllate (nineteenth- century) Chiricahua Apache habi-tation site I recorded in the Dragoon Mountainsnear Tucson, Arizona. Undoubtedly, similar factorsare applicable to the use and disposal of indigenouscontainers and materials obtained during raids. Thislikely accounts for the distribution of material cul-ture in contexts far beyond the reaches of the donorgroups, as seemingly occurs on some of the smallspecialized dunal sites I have recorded on the LlanoEstacado.While it is generally accepted that mobile groups

engaged in raiding and therefore the material cul-ture of other groups ended up in their sites, thisacceptance is generally not extended to ceramicvessels (nor is it generally considered in the pre- Contact period owing to the widespread assump-tion that raiding was a late- Apache adaptation).This is because a late nineteenth- century model ofraiding is being adopted that in many instanceswould preclude the transport of breakable and bulkyitems. Yet, the fast- and- furious hit- and- run tacticsof the late Apache adaptation are not characteris-tic of the earliest Historic period when substantiallydifferent raiding practices are documented in theethnohistoric record. In the 1530s Cabeza de Vacarecords that villagers fled to the hills leaving thevillage unattended or stayed and acceded to thetheft of material to avoid conflict (Hodge 1990;Seymour 2002). This allowed time for looters toremove and carry away select bulky and breakablematerials, such as vessels filled with grain thatcould be used for water- storage or cooking vesselsin their new contexts. Low densities of certain types of artifacts can

also be the result of knowledge of resources avail-able at a known destination. The Salinas Pueblosprovide a case in point. Documentary (ethnohistoricand historic) records indicate that mobile groupsvisited the pueblos to trade bison products for cornand processed goods (Baugh 1984, 1991; Kenner1969; Spielmann 1982; Vehik 2002). Rather thanbringing “coals to Newcastle,” Plains nomads trans-ported as many exchangeable goods as possible

REPORTS 165

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 165

Page 9: contextual incongruities, statistical outliers, and anomalies

and would have minimized extraneous utilitarianpottery or other equipment that would have beenavailable at the pueblos (Seymour 2006a, 2007d).Under such circumstances, intrusive (and poten-tially inferior) pottery might not be expected in thehost pueblo’s archaeological contexts. Moreover,since one of the primary purposes of such meet-ings was to exchange goods, material culture wouldhave moved between or across cultural contexts. Itis also likely that such meetings would haveincluded feasting rites outside the pueblo confines,where most of the interface seems to have takenplace (Seymour 2005a, 2006a, 2006b, 2007d,2008f), resulting in the deposition of Puebloan pot-tery in the mobile group locus.These processes are consistent with recent

research. Mobile group structure outlines have beenfound on the slopes near Salinas- area pueblos, sur-rounded by light scatters of Puebloan pottery andartifacts, with an occasional hint of Plains artifacts(such as nonlocal lithic materials and Plains- styletools; Seymour 2005a, 2006a, 2006b, 2007d,2008f). The Puebloan items result from centuriesof Puebloan discard and their limited use of suchlocations, but also from use of the host’s materialculture by visiting mobile groups. Similarly, mostof the identifiable (and valuable) Plains items (e.g.,nicely crafted bifacial knives) have been recoveredarchaeologically from purely Puebloan contexts,such as middens and house floors within or imme-diately adjacent to the pueblo itself. This “trans-ference” is a process that is expected among tradingpartners but has rarely been incorporated intoexpectations for distributions of material culture atmobile group encampments surrounding pueblos.Consequently, by privileging the most audible voice— that of decorated pottery and projectile points— over architecture, landscape use, andflaked stone, and by ignoring historically docu-mented processes, mobile groups have been ren-dered invisible.

Quantification and LowFrequency Occurrences

The problem being discussed is compounded bythe subtle forms of evidence that help archaeolo-gists distinguish between alternative interpretationsfor mobile group occupations. Interpretation of thissubtle and low- frequency evidence is often

obscured rather than illuminated by statisticalanalysis. Key evidence often comes in the form ofa small number of items, creating distributions thatsignificance tests would assign to chance. Archae-ologists are trained to seek commonalities ratherthan differences, and often discard outliers to“improve” their statistical results. Unfortunately, itis these outliers that may signal the presence ofmobile groups in mixed assemblages. The often- faint evidence of mobile groups overlying or cohab-iting a sedentary landscape lies outside the field ofvision, a viewshed that focuses on the dense clus-ter at the center of the graph or the linear arrange-ment of the better- understood and more obviousdistributions. This is exacerbated by high thresh-olds of adequacy— the higher the threshold, thehigher the probability that the patterns reflectbehavior (rather than poor collection methods, forexample), but also the greater the likelihood thatless robust signatures will be obscured. In fact, adeliberate consequence is that more unique mani-festations are relegated to irrelevance. Regrettably,an unintended consequence is that material evi-dence of most mobile groups falls at the outer edgesof the graph. The classic battleship curves and scat-ter plots of traditional analyses are useless whensearching for the presence of a people who onlybroke one pot during their entire occupation, espe-cially if that pot was not their own.The invisibility of so- called unique manifesta-

tions is exacerbated by the perceived need for quan-tification and the value of large numbers.Consequently, the denser and more complex thesedentary ceramic period occupation, the less likelythe mobile one will be seen. Relatively speaking,mobile occupations become more and more invis-ible in comparison to the increased robusticity ofthe sedentary dataset.

Quantification and Visibility Case Study: The Cerro Rojo Site

It is conceptually difficult to bypass the relativeabundance and familiarity of one component inorder to visualize the less- obtrusive ones. It is evenmore difficult to envision a component that has notbeen previously archaeologically defined (evenwhen it is dominant), when small amounts of thematerial culture of a known more obtrusive groupare present. The hilltop site of Cerro Rojo (FB9609; Seymour 2002, 2003a, 2004) provides an

166 amERican anTiquiTy [Vol. 75, no. 1, 2010

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 166

Page 10: contextual incongruities, statistical outliers, and anomalies

example. About 300 historic Apache (i.e., CerroRojo complex) and Canutillo complex features areintermixed with relatively limited Jornada Mogol-lon, Archaic, and historic mining components. Thedozen or so Jornada manifestations (fire pits, roast-ing pits, shelters, and shrines) are most visiblebecause they represent a known and more robustsignature and include associations of known diag-nostic artifacts, though they occur in relatively lowfrequencies. Owing to (a) the presence of this more recog-

nizable (though relatively inconsequential) Jornadacomponent, (b) the expectation that lesser knownmanifestations (such as Apache) are invisible, and(c) the application of inappropriate models of land-scape use and settlement types, previous recordersdid not recognize 95 percent of the spatial extentof the site or its features. Initially recorded as a 40m by 100 m fortified site of possible JornadaMogollon affiliation (Beckes et al. 1977), laterrecorders expanded the site to 115 m by 85 m,where 15 rock enclosures and several hundred arti-facts were acknowledged but it was determinedthat “there are no overt indications of a protohis-toric or early historic occupation at this site” (Baughand Sechrist 2000:184). Subsequent examination within a region- wide

context reveals that the Jornada components are rel-atively small and spatially restricted, representing limited- use localities within a larger settlement sys-tem. In contrast, the Terminal Prehistoric and His-toric mobile group components are widespread andrepresented by relative dense and abundant evi-dence. This mountaintop site contains thousands ofartifacts and 275 features dispersed in discrete lociacross about 130 acres (Seymour 2002, 2004,2008c, 2009c). Features include 212 structure out-lines that are similar in morphology to manythroughout the region, including those on a site thatwas photographed in 1886 during the famous Chir-icahua Apache Geronimo’s attempted surrender atCañon de los Embudos in Sonora, Mexico (Sey-mour 2009c); and to another, earlier location in theDragoon Mountains of Arizona that was where theApache leader Cochise met General Howard tonegotiate a treaty in 1872 (Seymour and Robert-son 2008). The association of at least eight uniqueplainware varieties and their dates with diagnosticstone tools confirms a picture of episodic settlementby distinct social groups falling within two (Cerro

Rojo and Canutillo) traditions. The Cerro Rojocomplex signals the presence of ancestral Apacheanraiders (Seymour 2002, 2003a, 2004, 2008c,2008d). An equally distinctive Canutillo complexfound at the Cerro Rojo Site may be assignable toany of a number of ethnohistorically known groupsfor the region, including the Jano, Jocome, Manso,Suma, or even some other, unknown group (Sey-mour 2002, 2003a, 2004, 2008c, 2008d). The prehistoric and historic components are also

distinguished by six radiometric and 11 lumines-cence dates, multiple petrographic analyses, exten-sive artifact attribute analyses, and test excavations(see Seymour 2004, 2008c). Dates on the later com-ponents range from the 1400s/1500s through theearly 1800s and are consistent with the area- specificand regionally applicable ethnohistoric record thatindicates people like these were present at first con-tact. In fact, the earliest applicable date (A.D. 1420± 80; Sample X1516) predates initial Europeancontact and indicates the occupants could havelooked out across the landscape to see the firstEuropeans crossing the Rio Grande on their waynorth into New Mexico. Documentary (ethnohistoric, ethnographic, and

historic) records report a mobile pattern of land-scape use. These sources also document the com-ing together of normally dispersed bands and localgroups into large aggregations in out- of- the- wayplaces, such as the deep and rugged Hueco Moun-tains where the Cerro Rojo Site is located (Seymour2004, 2008c). When the Cerro Rojo Site is inter-preted using regionally and temporally specificethnographic models and ethnohistoric data, itscharacter becomes clear. Cumulative, repeatedoccupation creates an archaeological pattern ofhouse clusters spread out across the mountain top.The placement of habitation features in discreteloci at Cerro Rojo suggests recurring though inter-mittent gatherings for specific occasions by a num-ber of individual social groups. These groupsmaintained spatial separation, presumably as areflection of geographically based social divisions.Defensive walls and ramparts augment the defen-siveness of the setting, tangibly conveying the per-ilous social world that was characteristic of thistime.Individual episodes of use at Cerro Rojo are

marked in part by discrete dates, including radio-metric results from charred material and from lumi-

REPORTS 167

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 167

Page 11: contextual incongruities, statistical outliers, and anomalies

nescence analyses run on pottery and burnedcaliche. The pottery selected for analysis met thefollowing criteria: (1) plain brownware attributedto Apachean groups; (2) plain brownware inferredto be associated with the river- side mission settle-ments (Valle Bajo); and (3) plain brownwares ofunknown origin but distinct from contemporane-ous types and from earlier types associated with theprehistoric Jornada Mogollon occupation. Classi-fications of Cerro Rojo’s plainware varieties arebased on petrographic analysis and visual exami-nation of the technological and compositionalattributes, and their placement in this late period isconsistent with their luminescence dates. Despite its association with Apache/Cerro Rojo

and Canutillo material culture complexes, it has notbeen assumed that all late- occurring pottery foundat the Cerro Rojo Site is directly attributable togroups who resided there. Some may have beenobtained via raiding, trading, and gifting fromgroups dispersed over a wide interaction sphere.These materials would have been brought to the siteduring occupation episodes spread out over somefive centuries. This inference is derived from dat-ing individual sherds and pot breaks from CerroRojo, some of which do not belong to either theCerro Rojo or the Canutillo complex, and whichcannot be assigned to any of the many pottery typesdescribed for the Jornada region (Seymour 2002,2003a, 2004, 2008c, 2008d). As our knowledge of the distribution of Termi-

nal Prehistoric and Historic plainwares in the Jor-nada region expands, the chronological and spatialanchoring of materials from Cerro Rojo shouldassist in developing a picture of the social and polit-ical affiliations that unfolded both before and afterEuropean contact. This valuable information wouldnot be available but for vigorous pursuit of chrono-logical information using the models suggestedhere.

Diachronic Occupation on a Single Site and Age Averaging

Age averaging relies on the reasoning that whenrecognized archaeological correlates on a site pointto a limited, short- term occupation, all dates shouldoverlap. Thus, averaging is a way to increase therobusticity of a date set. As Reed and Horn (1990)logically reason when averaging six radiocarbon

ages at the Navajo site of Kin ’ Atsá (LA 49498),“the tight clustering of the radiocarbon ages alsoargues for the accuracy of the age determinationsbecause the use of long- dead wood would result inmarked differences in ages between samples andgreater standard deviation” (1990:288).Confidence is heightened when multiple dates

produce similar results and are averaged becausethe “old wood problem” and the “ cross- sectioneffect” are less likely to be in play (Schiffer 1982,1986; Smiley and Ahlstrom 1998). Yet, efforts toguard against influence from these two widely rec-ognized problems may introduce other interpretivetroubles. Specifically, the practice of averagingresults obtained from multiple dates masks multi-componentcy and episodic reuse. This logical dis-juncture is most detrimental in mobile group studiesbecause averaging, or attempting to fit multipledates into a single synchronic event, means thatoccupations that take place over spans of two cen-turies or less are frequently obscured. For mobilegroups this is particularly problematic; non- occupational activities that take place during evenshorter intervals are completely masked. Schiffer (1987:309) finds statistical techniques

for isolating central tendencies unsatisfactorybecause this approach “is incapable of detectingbiases in a series of dates and because it treats alldates as being equally informative about past cul-tural events.” If one can demonstrate that all sam-ples are extracted from behaviorallycomplementary or equivalent contexts, such con-cerns may be obviated. Yet, frequently, this assump-tion is made prematurely and elements of multipleepisodes of use are collapsed into a single occupa-tional narrative. Often, the aggregate of a series of limited- use contexts and short- term activitiesresults in what appears to be a good example of a single- use event. A deeper understanding may beachieved when discordant dating results are allowedto cause interpretive pause as one considers a poten-tial lack of contextual equivalency. Entertaining alternative models of site forma-

tion has the benefit of highlighting subtle indica-tions of multicomponentcy and episodic reuse.Some sites produce non- overlapping dates widelyseparated in time. Often seen as irresolvable anom-alies, these sites may have been important locationsin specific adaptations occupied for short periodsof time in order to complete a single task. These

168 amERican anTiquiTy [Vol. 75, no. 1, 2010

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 168

Page 12: contextual incongruities, statistical outliers, and anomalies

sites— or the tasks carried out there— may not havebeen part of the normal seasonal round. This issueof occupational duration and episodic reuse forshort (or even variable- length) time spans is specif-ically relevant to whether an unexpected date isinterpreted as an anomaly or as site reuse. The dif-ference is discernable when dating results are indi-vidually evaluated rather than averaged.Recognition of a series of short- lived occupationsor uses can be facilitated by considering nontradi-tional forms of diagnostic artifacts. For example,when subject to multiple analytical techniques(INAA, petrographic thin sectioning, luminescencedating, and visual technological analyses), plain-wares can be differentiated from prehistoric pot-tery and then used for dating, even if not indicativeof specific or identifiable cultural affiliation (Sey-mour 2003a, 2008d). Understanding can be aidedby using parallel or converging lines of evidencethat increase confidence levels and take intoaccount the behavioral processes that influencesample discovery, selection, and interpretation.

Age Averaging Case Study: The Sharples Site

A prime example of the potential pitfalls of age- averaging, as well as the ways in which use of twochronometric techniques in an integrated method-ology address this repeated- use issue, is providedby work on the Sharples Site (AZ DD:8:44, ASM).This site, situated on a terrace that overlooks theSanta Cruz River near Tubac, Arizona, wasmapped, surface collected, and the mobile groupcomponent selectively excavated by the author anda volunteer crew. The mobile group occupation isrepresented by a veneer of material culture thatpartially overlies a near- surface radiocarbon- datedprehistoric (A.D. 1100s) sedentary occupation andunderlies a later historic Apache occupation. Analy-sis of multiple chronometric samples using twodifferent techniques indicates an even more com-plex site history relating to Canutillo complexmobile groups in the Terminal Prehistoric period(Seymour 2009a). Where radiocarbon isolated onlyone, luminescence dating of surface and near- surface sherds provided evidence of two to threedistinct occupations. In addition, radiocarbon dat-ing could not differentiate the fifteenth- centuryoccupation suggested by the luminescence results. The string of dates suggests multiple episodes

of use and reuse. This scenario is supported by

ethnographic descriptions and archaeologicallydevised models of mobile group landscape use andbehavior, and is consistent with other on- site data(Seymour 2009a). The replication of stationary fea-ture types and the relatively high density and diver-sity of artifacts (that is, relative to most regionalmobile group sites) at the Sharples Site provide evi-dence of what was probably a multiple series ofencampments by small residential groups. In anyone area artifacts were relatively sparse but cumu-latively, over the entire spatial distribution of thecomponent, diversity was relatively high with areplication of tool kit elements and maintenancedebris. Pairs of rock hut outlines from mobilegroups were scattered around the site withoutregard to one another or to any formalized plan.Specialized work stations— individually consist-ing of groundstone, anvil stones, hide workingstones, and chipping stations— showed evidence oflight use and duplication of purpose, presumablyindicating replacement during each visit to the siteor perhaps communally focused use areas duringa somewhat larger gathering.Two AMS radiocarbon dates were run on small

flecks of charred brushy material, one from a struc-ture, the other from an extramural thermal feature.Both individually and together they provide a gen-eral idea that this component was occupied duringthe inclusive dates of A.D. 1430 to 1630 (Table 1).These two internally consistent radiocarbon resultssuggest that they are dating events that occurredwithin relatively the same analytical period. Theindividual 2-sigma standard deviations of the radio-carbon determinations are quite broad, and whenthe dates are considered together (overlapping thestandard deviations to identify dates that are not sta-tistically different), the statistical interval isextended even more (Figure 2a). Consequently,these results are not especially useful in differen-tiating between distinct occupations or for sug-gesting which part of this calibrated date intervalis most likely to indicate component use.In comparison, five luminescence dates were

run on plainware sherds (see Table 1). When con-sidered together— overlapping the standard devia-tions to identify dates that are not statistically different— they seem to narrow the occupation evenfurther, from the mid-1400s to mid-1500s (Figure2b). Yet, when a single or continuous longer- termuse is not assumed and the sample results are exam-

REPORTS 169

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 169

Page 13: contextual incongruities, statistical outliers, and anomalies

170 amERican anTiquiTy [Vol. 75, no. 1, 2010

Table 1. Radiocarbon and Luminescence Dates Obtained from AZ DD:8:44 (ASM).

Measured Conventional Sample No. Material/ Calibrated Calibrated Radiocarbon Radiocarbon C14, OSL Context Luminescence Date (2�) Date (1�) Age Age

Beta-190621 Charred brushy material, A.D. 1450–1650 A.D. 1470–1530 340 ± 40 350 ± 40Feature 33 (500–300 cal B.P.) (480–420 cal B.P.)

and A.D. 1550–1630(400–320 cal B.P.)

Beta-191895 Charred brushy material, A.D. 1430–1520 A.D. 1440–1490 200 ± 40 410 ± 40Feature 21 (520–430 cal B.P.) (510–460 cal B.P.)

and 1580–1630(380–320 cal B.P.)

OSL: X2549 Plainware sherd, flat on A.D. 1450 ± 40; structure floor, Feature 3 A.D. 1416–1496

OSL:X2068 Rio Rico Plain, Tubac A.D. 1524 ± 40; variety, Sample 7 A.D. 1484–1564

OSL:X2069 Whetstone Plain, A.D. 1524 ± 60; Amado variety, Sample 8 A.D. 1464–1584

OSL:X2070 Baca Float variety, A.D. 1604 ± 40; Sample 9 A.D. 1564–1644

OSL:X2071a Arivaca variety, A.D. 1574 ± 40; Sample 10 A.D. 1534–1614

Figure 2. Averaging effects of chronometric results. (a) Radiometric results averaged to produce an A.D. 1450–1650 con-fidence interval. (b) Luminescence results averaged to produce an A.D. 1484–1564 overlap period.

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 170

Page 14: contextual incongruities, statistical outliers, and anomalies

ined individually— as potentially dating distinctbehavioral events— a very different pictureemerges. They seem to indicate two distinct peri-ods of mobile group use (Figure 3). This is con-sistent with mobile group landscape use anditinerancy (Seymour 2002, 2009d). The earliestoccupation is indicated by one of many sherdsfound lying flat on a structure floor (OSL: X2549).The date for this sample (A.D. 1416–1496) is sur-prisingly consistent with the calibrated 1-sigmaresult for one of the radiocarbon samples (A.D.1440–1490; Beta-191895; Table 1; Figure 5). Twoadditional sherd samples date to the mid-1500sthrough early 1600s, seemingly representing a sec-ond period of use that does not overlap with the sherd- dated structure (Samples 9 [OSL:X2070]and 10 [OSL:X2071a]; see Figure 3). Another two sherd samples fall within these

potential periods (Samples 7 [OSL:X2068] and 8[OSL:X2069]; see Table 1; Figure 4). If averagedwith the previous two samples, as often occurs intraditional analyses, the dates for these final twosherds would fill a gap between the two periodsdescribed above, indicating a continuous occupa-tion between the mid-1500s and the mid-1600s. Onthe other hand, if these latter two overlapping sam-

ples are combined and averaged with the one fromthe structure floor they suggest that the site wasoccupied in the late 1400s, with a second period ofuse sometime between the mid-1500s to mid-1600s. It is also possible that these sherds datingto this intermediate period represent a third occu-pation. In the absence of averaging, the lumines-cence technique is just precise enough to suggestup to three occupations, perhaps hinting at anepisodic series of an even greater number of occu-pations, as expected for highly mobile hunter- fisher- gatherers. At a minimum, luminescencedating of surface and near- surface sherds providesevidence of two discrete occupations, while theradiocarbon technique isolated only one and couldnot specifically differentiate the fifteenth- centuryoccupation. The sherds used to date this occupation provide

evidence of a general regional presence by Termi-nal Prehistoric and Contact period groups. Thesegroups were distinct from their prehistoric prede-cessors and their ceramics are unlike prehistoricpottery known for the area. For the purposes of dat-ing site use, it is less important to know who themakers of the pottery were than to understand howthe pottery found its way into the targeted mobile

REPORTS 171

Figure 3. Unaveraged dates considered in a mobile framework. Luminescence results considered individually show atleast two distinct occupations, as expected for mobile groups.

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 171

Page 15: contextual incongruities, statistical outliers, and anomalies

group contexts (e.g., cultural affiliation and chrono-logical placement are separate analytical endeav-ors, which regrettably are often collapsed). Thesample of dated pottery includes two sherds thatare clearly early O’odham and three previouslyundescribed Terminal Prehistoric plainwares thatcould have originated with any of a number ofgroups. The early O’odham pottery (Samples 8[OSL:X2069] and 10 [OSL:X2071a]) likely orig-inated in a nearby O’odham settlement and weretransported to this location by mobile groups whoobtained the vessels through raiding or by trade,two well documented forms of interaction in thisarea (Seymour 2005b, 2007b, 2009a). The histor-ically referenced O’odham site of Tubacawas sit-uated nearby (Dobyns 1995), and an excavatedcontemporaneous (A.D. 1424–1524) Sobaípuri- O’odham settlement is known from further southon the river, thus placing the O’odham in proxim-ity to these mobile groups who resided at theSharples Site. On the other hand, the earliest occu-

pation is indicated by a coarse- tempered, rough- surface sherd, one of many found lying flat on astructure floor (OSL: X2549). The sherd corre-sponds to no described type. Two additional, andtechnologically distinct, sherd samples (Samples 9[OSL:X2070] and 10 [OSL:X2071a]; see Figure3) represent the second period of inferred use.Another two sherd samples may fill the gapbetween these potential use episodes. Though cur-rently untyped, these sherds seem to be derivedfrom the Sonoran brownware tradition (Samples 7[OSL:X2068] and 8 [OSL:X2069]; see Table 1;Figure 4), as do most indigenous pottery types inthe southern portion of the American Southwest.

Site reuse by mobile groups who move acrossexpansive areas is expected. They are drawn to loca-tions that naturally possess certain important char-acteristics, and are therefore likely to select manyof the same locations repeatedly (Seymour 2009c,2009d). Resource procurement is typically viewedas an extractive process in which materials are drawn

172 amERican anTiquiTy [Vol. 75, no. 1, 2010

Figure 4. Dates interpreted in a high mobility framework reveal a possible third occupation. A period of overlap mayindicate a third possible short-term occupation, consistent with other evidence on the site, or a longer term occupationthat encompasses many decades.

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 172

Page 16: contextual incongruities, statistical outliers, and anomalies

away from a site. A formulation more consistentwith the known behavior of mobile human groupswould highlight the key fact that alterations, causedby use or occupation, often enhance a place’s valuefor future uses. Various kinds of alterations may beexpected based on a comparison to a region- widesample of mobile group sites investigated by theauthor and others that show: (a) reworking and reuse

of discarded tools and materials, (b) shelters that arebuilt within prehistoric structures, and (c) thermalfeatures that have been reused (Seymour 2002,2003a, 2003b, 2005b; also see Gregory 1981).These expectations also derive from analysis of theethnographic and ethnohistoric records that docu-ment mobile groups reusing prehistoric locationsand collecting prehistoric tools for reworking (Flint

REPORTS 173

Figure 5. Correspondence between dates derived from two different dating techniques reveals a distinct and the earliestmobile occupation. When a single or continuous longer-term use is not assumed and the sample results are examinedindividually—as potentially dating distinct behavioral events—several potential short-term occupations are distinguish-able. The result for earliest luminescence sample (flat-lying sherd on structure floor; A.D. 1416–1496) is surprisingly con-sistent with the calibrated 1-sigma result for one of the radiocarbon samples (A.D. 1440–1490; Beta-191895).

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 173

Page 17: contextual incongruities, statistical outliers, and anomalies

and Flint 2005:417; Goodwin and Basso 1971:231;Russell 1975:111). These expected alterationsinclude those made as the same group repeatedlyreturns to a place, or modifications resulting froma totally different set of occupations that make theplace useful in ways the earlier occupants could nothave imagined. The location may gain value onceused because artifacts are left in place, outcrops ofraw lithic material are exposed, walls have alreadybeen constructed, surfaces are cleared of cobbles,or disturbance contributes to favorable circum-stances for edible species to grow. At the SharplesSite the previous prehistoric occupation clearlyenhanced the location for future use; the walls ofprehistoric structures and a compound were incor-porated into the hut walls of later mobile groups,depressions cleared of rocks were reused for vari-ous daily activities, and prehistoric stone founda-tions and discarded tools and debris became rawmaterials for tools.

Concluding Remarks

Given the high incidence of multiple reoccupationsand of later occupation by unrelated groups, itseems essential that alternative hypotheses andmethodologies be explored and divergent datessought. The amount of site reuse detected by recentstudies suggests that it is time to reformulateresearch approaches and consider that multipledates are likely to be useful and meaningful ratherthan problematic. Recently obtained data indicatethat a single- use scenario is quite rare indeed, evenon sites with limited assemblages and flimsy struc-tures. However, unless a researcher looks veryspecifically for evidence of repeated use while stillin the field, such evidence will likely go unnoticed;the signs tend to be faint and sometimes confus-ing, and data are easily interpreted in alternate ways.I advocate consideration of an observationalmethodology that, if applied effectively, will ensurethat the correct observations will be made on thesite and that analysis of the data after the fact willreveal a detailed occupational history.No current dating technique alone is going to

isolate multiple episodes of reoccupation if they alltook place within a few decades. Other lines of evi-dence need to be used, along with careful applica-tion of chronometric data, in order to make theseoccupational distinctions. This requires consider-

ing other indices of cultural and temporal affilia-tion beyond the old standbys of decorated potteryand projectile points and it requires a reevaluationof plainware variability and other indices of diag-nosticity. It means taking into account gradationsof mobility, and considering various forms of lim-ited reuse. It involves the even more challengingtask of identifying mobile group components thatare positioned over complex sedentary sites, andbeing able to isolate them even in cases where dat-able materials are not present. Such considerationsalso raise the ethical issues as to whether impactsto the later and lighter components are appropri-ately mitigated as researchers delve into deeperstratigraphy, and of whether the multiethnic tenureof places is being recognized.

Acknowledgments. My deepest appreciation is extended toJeffrey Dean, Mark Harlan, Darden Hood, Michael Schiffer,and H. Wolcott Toll for taking the time to comment on draftsof this paper and to Mark Harlan for preparing the Spanishabstract. Ed Rhodes, Jean Luc Schwenninger, JamesFeathers, William Deaver, and David Hill provided veryhelpful comments on samples, sampling approaches, or onportions of this paper. Some of the data presented in thispaper result from research I conducted while at LoneMountain Archaeological Services under contract with FortBliss. Appreciation is extended to the Sharples family forallowing access to their land to investigate their importantsite, to Bill Cox for showing the site to me, and to the manyvolunteers and past employees who assisted during surveyand with excavations.

References Cited

Ayer, Mrs. Edward E. (translator)1965 The Memorial of Fray Alonso de Benavides. Horn andWallace, Albuquerque.

Ball, Eve1970 In the Days of Victorio: Recollections of a WarmSprings Apache. University of Arizona Press, Tucson.

Basehart, Harry W.1960 Mescalero Apache Subsistence Patterns and Socio- Political Organization: Sections I and II. A report of the Mescalero- Chiricahua Land Claims Project. University ofNew Mexico, Albuquerque.

Baugh, Timothy G.1984 Southern Plains Societies and Eastern Frontier PuebloExchange During the Protohistoric Period. In CollectedPapers in Honor of Harry L. Hadlock, edited by Nancy L.Fox, pp. 156–167. Papers of the Archaeological Societyof New Mexico No. 9, Albuquerque Archaeological Soci-ety Press, Albuquerque.

1991 Ecology and Exchange: The Dynamics of Plains- Pueblo Interaction. In Farmers, Hunters, and Colonists:Interaction between the Southwest and Southern Plains,edited by K. A. Spielmann, pp. 107–127. University of Ari-zona Press, Tucson.

174 amERican anTiquiTy [Vol. 75, no. 1, 2010

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 174

Page 18: contextual incongruities, statistical outliers, and anomalies

Baugh, Timothy G., and Mark Sechrist 2001 Protohistoric Apachean Adaptations within the Basinand Range Province of South- Central New Mexico andWest Texas: A Perspective from the Fort Bliss Reserva-tion. On file with the Conservation Division, Directorateof Environment, Fort Bliss, Texas. TRC Mariah Associ-ates, Inc. Report, El Paso.

Beckes, Michael R., Alan R. Dulaney, John D. Pigott, RobertF. Scott IV, and Nancy Kays Smiley

1977 A Cultural Resource Inventory and Assessment ofMcGregor Guided Missile Range, Otero County, New Mex-ico.Research Report No. 65, Part II. Texas ArcheologicalSurvey. The University of Texas, Austin.

Beckett, Patrick H., and Terry L. Corbett1992 The Manso Indians. COAS Publishing and Research,Las Cruces, New Mexico.

Betzinez, Jason, and Wilbur Sturtevant Nye1959 I Fought With Geronimo. University of NebraskaPress, Lincoln.

Cordell, Linda S.1984 Prehistory of the Southwest. Academic Press, New York.

Corley, John A. 1965 Proposed Eastern Extension of the Jornada Branch ofthe Jornada Mogollon. In Transactions of the First Arche-ological Symposium for Southeastern New Mexico andWestern Texas. Lea County Archeology Society Bulletin 1,pp. 30–36. Hobbs, New Mexico.

Dean, Jeffrey S.1978 Independent dating in archaeological analysis. InAdvances in Archaeological Method and Theory, Volume1, edited by Michael B. Schiffer, pp. 257–314. AcademicPress, New York.

Dobyns, Henry F.1995 Tubac through Four Centuries: A Historical Resumeand Analysis. Prepared for the Arizona State Parks Board15 March 1959, Reformatted by Tubac Presidio State His-torical Park, August 1995 and revised. Electronic docu-ment, http://parentseyes.arizona.edu/tubac/index.html,accessed September 8, 2009.

Flint, Richard, and Shirley Cushing Flint2005 Documents of the Coronado Expedition, 1539-1541:They Were Not Familiar with His Majesty nor Did TheyWish to Be His Subjects. Southern Methodist UniversityPress, Dallas.

Gilpin, D., and D. A. Phillips, Jr.1998 The Prehistoric to Historic Transition Period in Ari-zona, Circa A.D. 1519 to 1692. Prepared by SWCA forthe Arizona State Historic Preservation Office, Phoenix.

Gladwin, Winifred, and Harold S. Gladwin1935 The Eastern Range of the Red- on- Buff Culture.Medallion Papers No. XVI. Gila Pueblo, Globe.

Goodwin, Grenville1929–1939 Pottery. Manuscript 17, Goodwin Papers, 1929-1939. Arizona State Museum Archives, University of Ari-zona, Tucson.

Goodwin, Grenville, and Keith H. Basso1971 Western Apache Raiding and Warfare. The Univer-sity of Arizona Press, Tucson.

Gregory, David A.1981 Western Apache Archaeology: Problems andApproaches. In The Protohistoric Period in the NorthAmerican Southwest. AD 1450–1700, pp. 257–274.Anthropological Research Papers No. 24. Arizona StateUniversity, Tempe.

Haley, James L.1981 Apaches: A History and Culture Portrait. Universityof Oklahoma Press, Norman.

Hodge, Frederick Webb (editor)1990 The Narrative of Alvar Nunez Cabeza de Vaca. InSpanish Explorers in the Southern United States 1528-1543, pp. 1–123. The Texas State Historical Association,Austin.

Hole, Frank, and Robert F. Heizer1977 Prehistoric Archaeology: A Brief Introduction. Holt,Rinehart, and Winston, New York.

Kenner, Charles L.1969 A History of New Mexico- Plains Indian Relations.University of Oklahoma Press, Norman.

Lehmer, Donald J. *1948 The Jornada Branch of the Mogollon. Social Science

Bulletin No. 17. The University of Arizona, Tucson.MacNeish, Richard S.1998 Excavation of Pintada Rockshelter on McGregor Fir-ing Range in New Mexico. Publications in AnthropologyNo. 12. El Paso Centennial Museum, the University ofTexas at El Paso, El Paso.

2003 Pendejo Cave.University of New Mexico Press, Albu-querque.

Miller, Myles R. 2001 Post- Pueblo, Protohistoric, and Early Mission PeriodArcheology in Western Trans- Pecos Texas and South Cen-tral New Mexico, A.D. 1450–1680. Bulletin of the TexasArcheological Society 72:105–164.

Opler, Morris E.1974 Lipan and Mescalero Apache in Texas. Apache Indi-ans X. Garland Publishing, New York.

1996 [1941] An Apache Life- Way: The Economic, Social,& Religious Institutions of the Chiricahua Indians. Uni-versity of Nebraska Press, Lincoln.

Plog, Fred 1984 Exchange, Tribes, and Alliances: The northern South-west. American Archaeology 4:217–223.

Reed, Alan D., and Jonathon C. Horn1990 Early Navajo Occupation of the American Southwest:Reexamination of the Dinetah Phase. Kiva55(4):283–300.

Reid, Jefferson, and Stephanie Whittlesey1997 The Archaeology of Ancient Arizona. The Universityof Arizona Press.

Rouse, Irving1962 Southwest Archaeology Today. In An Introduction tothe Study of Southwest Archaeology, edited by Alfred Vin-cent Kidder, pp. 1–53. Yale University Press, New Haven.

Russell, Frank 1975 The Pima Indians. The University of Arizona Press,Tucson Arizona.

Schiffer, Michael B.1982 Hohokam Chronology: An Essay on History andMethod. In Hohokam and Patayan: Prehistory of South-western Arizona, edited by Randall H. McGuire andMichael B. Schiffer, pp. 299–344. Academic Press, NewYork.

1986 Radiocarbon Dates and the “Old Wood” Problem:The Case of the Hohokam Chronology. Journal of Archae-ological Science 13:13–30.

1987 Formation Processes of the Archaeological Record.University of New Mexico Press, Albuquerque.

Seymour, Deni J.1995 Behavioral Approach to the Recognition of Incon-spicuous Apachean Sites. Paper presented at the 68th

Annual Pecos Conference, Gila Wilderness, New Mexico. 2002 Conquest and Concealment: After the El Paso Phaseon Fort Bliss. On file with the Conservation Division,Directorate of Environment, Fort Bliss, Texas. Lone Moun-tain Report 525/528, Albuquerque and El Paso.

REPORTS 175

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 175

Page 19: contextual incongruities, statistical outliers, and anomalies

176 amERican anTiquiTy [Vol. 75, no. 1, 2010

2003a Protohistoric and Early Historic Temporal Resolu-tion. Manuscript on file at Conservation Division, Direc-torate of Environment, Fort Bliss, Texas. Lone MountainReport 560-003. Also on file at the Arizona State Museum,Tucson.

2003b The Cerro Rojo Complex: A Unique IndigenousAssemblage in the El Paso Area and Its Implications ForThe Early Apache. Proceedings of the XII Jornada Mogol-lon Conference in 2001. Geo- Marine, El Paso.

2004 A Ranchería in the Gran Apachería: Evidence of Inter-cultural Interaction at the Cerro Rojo Site. Plains Anthro-pologist 49(190):153–192.

2005a Mobile Visitors to the Eastern Frontier Pueblos. Man-uscript on file with the National Park Service, Mountainair,New Mexico.

2005b The Implications of Mobility, Reoccupation, andLow Visibility Phenomena for Chronometric Dating. Man-uscript on file at the Arizona State Museum Library, Tuc-son.

2006a Athabascan and non- Athabascan Mobile Traders atthe Salinas Pueblos, in the Galisteo Basin, at Pecos Pueblo,and near Paa- ko. Paper presented at the 79thAnnual Meet-ing of the Pecos Conference, Navajo Dam, New Mexico.

2006b Stranger Sojourners: Mobile Group Visiting Behav-ior. Manuscript on file with the National Park Service,Mountainair.

2007a An Archaeological Perspective on the Hohokam- Pima Continuum. Old Pueblo Archaeology Bulletin No.51(December 2007):1–7.

2007b A Syndetic Approach to Identification of the His-toric Mission Site of San Cayetano del Tumacácori. Inter-national Journal of Historical Archaeology11(3):269–296.

2007c Misconceptions about Luminescence Dating: Appli-cations for the Protohistoric. Paper presented at the 15th

Biennial Meeting of the Jornada Mogollon Conference,El Paso, Texas.

2007d With a Grain of Salt: Mobile Visitation and Residenceat the Salinas and Galisteo Area Pueblos. Paper presentedat the 80thAnnual Meeting of the Pecos Conference, PecosPueblo, New Mexico.

2008a Despoblado or Athapaskan Heartland: A Method-ological Perspective on Ancestral Apache Landscape Usein the Safford Area. In Crossroads of the Southwest: Cul-ture, Ethnicity, and Migration in Arizona’s Safford Basin,edited by David E. Purcell, pp. 121–162. Cambridge Schol-ars Press, New York.

2008b Delicate Diplomacy on a Restless Frontier: Seventeenth- Century Sobaípuri Social and EconomicRelations in Northwestern New Spain, Part II. New Mex-ico Historical Review 83(2):171–199.

2008c Surfing Behind The Wave: A Counterpoint Discus-sion Relating To “A Ranchería In the Gran Apachería.”Plains Anthropologist 53(206):241–262.

2008d Apache Plain and Other Plainwares on Apache Sitesin the Southern Southwest. In Serendipity: Papers in Honorof Frances Joan Mathien, edited by R. N. Wiseman, T. C.O’Laughlin, C. T. Snow, and C. Travis, pp. 163–186.Papers of the Archaeological Society of New Mexico No.34. Archaeological Society of New Mexico, Albuquerque.

2008f Pliant Communities: Seasonal Mobile Group Visita-tion at the Eastern Frontier. Paper presented at the 41st

Annual Meeting of the Society for Historical Archaeol-ogy, Albuquerque.

2009a The Canutillo Complex: Evidence of ProtohistoricMobile Occupants in the Southern Southwest. Kiva74(4):421–446.

2009b Evaluating Eyewitness Accounts of Native Peoplesalong the Coronado Trail from the International Border toCibola. New Mexico Historical Review 84(3):399–435.

2009c Nineteenth- Century Apache Wickiups: HistoricallyDocumented Models for Archaeological Signatures of theDwellings of Mobile People. Antiquity 83(319):157–164.

2009d Distinctive Places, Suitable Spaces: Conceptualiz-ing Mobile Group Occupational Duration and LandscapeUse. International Journal of Historical Archaeology13(3): 255–281.

Seymour, Deni J., and Tim Church2007 Apache, Spanish and Protohistoric Archaeology ofFort Bliss. Historic and Natural Resources Report 03-06,Cultural Resources Program, Directorate of Environment,Fort Bliss, Texas.

Seymour, Deni J., and George Robertston2008 A Pledge of Peace: Evidence of the Cochise- HowardTreaty Campsite. Historical Archaeology 42(4):154–179.

Shepard, Anna O.1956 Ceramics for the Archaeologist. Carnegie Institutionof Washington, Publication 609, Washington, D.C.

Smiley, Francis E., and Richard V. N. Ahlstrom1998 Archaeological Chronometry: Radiocarbon and Tree- Ring Models and Applications from Black Mesa, Arizona.Occasional Paper No. 16, Center for Archaeological Inves-tigations, Southern Illinois University, Carbondale.

Smith, Jason W.1976 Foundations of Archaeology. Glencoe Press, BeverlyHills.

Spielmann, Katherine Ann1982 Inter- Societal Food Acquisition Among EgalitarianSocieties: An Ecological Study of Plains/Pueblo Interac-tion in the American Southwest, Volume I. Ph.D. disserta-tion, University of Michigan. University Microfilms, AnnArbor.

Sweeney, Edwin R.1997 Making Peace with Cochise: The 1872 journal of Cap-tain Joseph Alton Sladen. University of Nebraska Press,Lincoln.

Treutlein, T. E. (Translator)1965 Missionary in Sonora, The Travel reports of JosephOch, S.J. California Historical Society, San Francisco.

Upham, Steadman1984 Adaptive Diversity and Southwestern Abandonment.Journal of Anthropological Research 40(2):235–256.

1994 Nomads of the Desert West: A Shifting Continuum inPrehistory. Journal of World Prehistory 8(2):113–167.

Vehik, Susan C.2002 Conflict, Trade, and Political Development on theSouthern Plains. American Antiquity 67(1):37–64.

Submitted June 19, 2009; Revised September 6, 2009;Accepted September 15, 2009

AQ75(1) Seymour_Layout 1 1/4/10 3:28 PM Page 176