''^^ ;3TTRESS AND VV^ak^^^^ ^MONG KaYEN TA A^^ASAZI OF THE Thirteenth Century A.D. JONATHAN HA AC WINIFRED CREAMER 'ieldiana Anthropology, new series, no. 21 Publication 1450 August 31, 1993 FIELD MUSEUM OF NATURAL HISTORY. HICAGO UNOIS LIBRARY liRBANA-CHAMeAlGf STACKS CENTRAL CIRCULATION BOOKSTACKS The person charging this material is re- sponsible for its renewal or its return to the library from which it was borrowed on or before the Latest Date stamped below. You may be charged a minimum fee of $75.00 for each lost booic. Theft, mutllatien, and underlining of book* ore reaseni for dUclpllnory action and may ro«ult in dismitsol from tho Unlvorslty. TO RENEW CAU TELEPHONE CENTER, 333-8400 UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN OEC 1 8 1997 OCT 2 2 1998 AUG 2 8 1998 MAR 2 4 2000 APR272000 AUG 0 1 2005 When renewing by phone, write new due date below previous due date. L162 0^' STRESS AND WARFARE AMONG THE KAYENTA ANASAZI OF THE THIRTEENTH CENTURY A.D. Field Museum Press Anthropology Books Foote Canyon Pueblo, Eastern Arizona, by John B. Rinaldo The First Peary Collection of Polar Eskimo Material Culture, by James W. VanStone The Simms Collection of Southwestern Chippewa Material Culture, by James W. VanStone Indian Trade Ornaments in the Collections of Field Museum of Natural History, by James W. VanStone A Late Pre-Hispanic Ceramic Chronology for the Upper Moquegua Valley, Peru, by Charles Stanish Material Culture of the Blackfoot (Blood) Indians of Southern Alberta, by James W. VanStone 5 72. o S FA no . 5- Stress and Warfare Among the Kayenta Anasazi of the Thirteenth Century A.D. JONATHAN HAAS Curator Department of Anthropology Field Museum of Natural History Roosevelt Road at Lake Shore Drive Chicago, Illinois 60605-2496 WINIFRED CREAMER Assistant Professor of Anthropology Northern Illinois University DeKalb, Illinois 60115 Fieldiana Anthropology, new series, no. 21 Publication 1450 August 31, 1993 FIELD MUSEUM OF NATURAL HISTORY, CHICAGO © 1993 Field Museum of Natural History This book was set in Palatino and printed and bound by the Allen Press in the United States of America. Design by Pannell + Oliver. Edited by Dana Fos. Library of Congress Catalog Card Number: 93-70978 ISSN 0071-4739 Cover illustration: "Batwoman." Redrawn from the petroglyph by Becky Hammond. The location of the petroglyph on the back wall of Batwoman Cave is shown on page 102. CONTENTS Illustrations vii Acknowledgments xi 1 Tribes and Tribal Formation 1 2 Background and Methodology 11 3 Excavation in Long House Valley 37 4 Kayenta and Klethla Valleys 49 5 Tsegi Canyon Survey 89 6 Conclusions and Beyond 129 Literature Cited 139 Appendix A. Excavations at LHV72 and LHV73 Jonathan Haas and Winifred Creamer 149 Appendix B. Computer Methods: Geographic Information Systems Kenneth Kvamme 171 Appendix C. Chipped Stone Raw Materials from Long House, Klethla, and Kayenta Valleys Margerie Green 181 Appendix D. Temper Analysis of Ceramics William A. Lucius 201 ILLUSTRATIONS 2-1. Map of the Four Corners region 12 2-2. Aerial photograph of Long House Valley, facing southwest 13 2-3. Map of Long House Valley in the period a.d. 1000-1150 15 2-4. Map of Long House Valley in the period a.d. 1150-1250 17 2-5. Map of Long House Valley in the period a.d. 1250-1300 18 2-6. Looking northeast from Fireside toward Tower House 30 2-7. Line-of-sight connection between Fireside and Tower House 31 2-8. Aerial photograph of Long House Pueblo, facing northeast 32 2-9. Aerial photograph looking down on "pueblo of orientation" at Long House Pueblo 32 2-10. Aerial photograph of isolated mesa upon which Organ Rock Ruin is located 33 2-11. View facing south to back side of Tower House 34 3-1. Map of Tsegi Phase occupation (a.d. 1250-1300) showing location of sites worked on in the course of the project 38 3-2. Schematic map of LVH72, the Brown Star Site 40 3-3. General environmental location of LVH72 41 3-4. Projectile point found with Burial 2 at LVH72 41 3-5. Schematic map of LVH73, the Potential Site 42 3-6. General environmental location of LVH73 42 4-1. Total survey project area, with sector maps 51-54 4-2. Aerial photograph of Kayenta Valley, facing northeast 55 Vll viii Illustrations 4-3. Moqui Rock, facing west 56 4-4. Map of Moqui Overlook 57 4-5. Map of Kinpo 58 4-6. Map of Tachini Point 59 4-7. Map of Parrish Creek Site 60 4-8. Map of RB568 61 4-9. RB568, facing northeast 62 4-10. Aerial photograph, facing northwest, of mesa top upon which Table Top Ruin is located 64 4-11. Map of Table Top Ruin 65 4-12. Map of Rabbit Ears Pueblo 66 4-13. The monolithic "rabbit ears" around which Rabbit Ears Pueblo is clustered 67 4-14. The "flatirons" upon which Happy Valley Pueblo is located 68 4-15. Artist's reconstruction of the Happy Valley community 69 4-16. Aerial photograph of small mesa upon which Six Foot Ruin is located 70 4-17. Map of Six Foot Ruin 71 4-18. Aerial photograph, facing north, showing general location of Wildcat Canyon Ruin mesa within Tsegi Canyon 72 4-19. Aerial photograph of Klethla Valley, facing south 73 4-20. Aerial photograph, facing north, showing rocky outcrop upon which Hoodoo Heaven Ruin is located 74 4-21. Aerial photograph, facing west, of Alan's Fort Ruin showing location of the outcrop in the wash 75 4-22. Map of Valley View Ruin 76 4-23. Map of Thief Site 79 4-24. Aerial photograph of Thief Site 80 4-25. Map of Kin Klethla 82 4-26. Aerial photograph of Kin Klethla 83 4-27. Isolated skull excavated from Kin Klethla 83 4-28. Skull from Kin Klethla showing one of two cut marks 84 5-1. Aerial photograph of Tsegi Canyon system 90 5-2. Kiet Siel Pueblo 91 5-3. Map of Tsegi Canyon system 94 5-4. Turkey House, across the canyon from Kiet Siel 96 Illustrations ix 5-5. Twin Caves Pueblo 97 5-6. A field currently being farmed by Navajo residents in the canyon 100 5-7. Batwoman Cave, with good southern exposure 102 5-8. Unoccupied cave with relatively poor protection from the elements 104 5-9. Unoccupied cave with good protection from the elements 105 5-10. Cave showing aeolian deposition 106 5-11. Cave demonstrating serious erosion action 107 5-12. Cave with a stable roof 108 5-13. Cave with an unstable roof 109 5-14. Cave with a flat floor 110 5-15. Betatakin Pueblo, showing the extreme slope of the floor 111 5-16. Aerial photograph of a larger cave in the survey area 112 5-17. A smaller cave in the survey area 113 5-18. Access route into Twin Caves Pueblo 115 5-19. Access route into Cradle House 116 5-20. Access route into Lolomaki 117 5-21. Access route into Kiet Siel 119 5-22. Open access of large occupied cave in Bubbling Springs Canyon 119 5-23. Access into Twoozis Kin 126 B-1. Rasterization of contour data 173 B-2. A search along four lines crossing a cell of unknown elevation 173 B-3. Computer-produced shaded slope map of Klethla and Long House valleys 176 B-4. The 452 potential defensive hilltop locations identified by the computer 177 B-5. Line of sight 178 B-6. A digital elevation model segment showing elevations at and between two hypothetical defensive hilltops (positions 1 and 2) 178 ACKNOWLEDGMENTS Any multiseason research project such as the one reported here owes its success to support from many institutions and individuals. Our summers in the Four Corners were made memorable by many of the people we met. Those who are not named here are no less important to us, and we thank all those who gave us a hand, a piece of information, or an encouraging word during those halcyon days. The Harry Frank Guggenheim Foundation generously supported field- work in northeastern Arizona in 1984 and 1985. Then-Director of the Gug- genheim Foundation, Floyd Ratliff, and Executive Director Karen Colvard were enthusiastic in their encouragement and advice. The National Geo- graphic Society supported research on cliff dwellings in 1986 (grant no. 3377- 86). The Long House Archaeological Field School (L.H.A.F.S.), sponsored by the University of Denver, contributed to the 1984 and 1985 field seasons. The School of American Research provided logistical and financial support and laboratory space for the project from 1984 to 1989. To Doug Schwartz and Jane Kepp we extend our appreciation for long-term encouragement, advice, and logistical backing. This research could not have been carried out without the cooperation of the Navajo Nation and the Director of the Navajo Nation Archaeology De- partment, Anthony L. Klesert. The Bureau of Indian Affairs, Window Rock, also facilitated our research under permit ARPA-85-007. The project field camp was located at Navajo National Monument each year, and our stay was facilitated by the National Park Service staff at NNM, especially superintendent Steve Miller, rangers John Loliet and Norberto Ortega, and directors of maintenance Bob Patterson and John Laughter. We cannot begin to thank Bill and Ginny Cowles for their enthusiasm, fresh insights, and support. Robert and Margo Linton also provided support for the project and withstood an adventurous trip to the field site. One person XI xii Acknowledgments who put enthusiasm and effort into the project and who is not able to review the product is the late Paul Logsdon, aerial photographer extraordinare. Paul flew over the research area for us, defying frightening crosswinds, and making photographs of many of the focal sites mentioned in the text. We miss his courtly and gentle good nature, we miss his expertise, and we miss him very much. Legwork on the project was especially demanding as our research design designated tops of hills and mesas as survey areas. We appreciate the efforts of our field crews from each season, with special thanks to Marvin Jackson and Wilson Smith. Andy Christenson, Alan Dulaney, Steve Glass, Lori Rhodes, and Powys Gadd served as crew chiefs for different aspects of the survey portion of the project. Powys also visited often and gave us welcome cheer, boundless energy, and a fundamental commitment to the archaeological enterprise. We cannot overestimate the contribution made to the project by Jeff Dean and Lex Lindsay. They visited in the field, introduced us to Long House Valley and Tsegi Canyon, argued with us late into the night about Kayenta prehistory, and helped in many, many other ways. We could not have carried out this project without their assistance. Crew chiefs Lori Rhodes, Brooke Plastino, and Jay Reszka led the field crews working in Long House Valley. Laboratory work was facilitated by Mark Oldknow and Tony Thibodeau, and artifact photos were taken by Rod Hook and John Weinstein. Katrina Lasko and Zbigniew Jastrzebski drew the illustrations. Many people have read the manuscript and offered advice and counsel. Some of it we have taken, some we have not. These include Jerry Levy, J. Jefferson Reid, Jeff Dean, George Gumerman, Jane Kepp, and Jim VanStone. Many others have read parts of the manuscript in various stages of production and we thank all of them individually and collectively. I TRIBES AND TRIBAL FORMATION This is a monograph about the relationship between warfare and the for- mation of tribal-type polities among the prehistoric Anasazi of the south- western United States. It presents the results of archaeological research con- ducted in northeastern Arizona between 1983 and 1986. The project was initiated to begin to address the question of whether or not tribes may have existed in prehistoric, precontact situations. In 1975, Morton Fried argued that "tribes" as they were described in the ethnographic record were artifacts of colonialism. He then inferred that the entire tribal form of organization was an artifact of colonial contact and did not represent a native stage in a unilinear model of cultural evolution (see also Fried, 1983). Fried's examples, however, were all drawn from postcolonial situations and needed to be tested in a precolonial context — through the archaeological record. Although most tribal models were presented in the 1960s and Fried's critique was published in 1975, the issue of tribal formation and organization is of ongoing interest and importance in anthropology and archaeology. In Between Bands and States, Susan Gregg (1991) has recently argued that ar- chaeology must take an active role in addressing the question of tribes. "The key characteristics of tribes must be delineated; the role that tribal-type social organizations play in human social organization must be determined; and the conditions under which tribes have appeared must be identified" (Gregg, 1991:445). In reviewing the archaeological literature for evidence of tribal formation, there were a number of cases in different parts of the world. Voss (1980), for example, found evidence for the emergence of tribes in northwest Europe during the Neolithic (see also Gibson & Geselowitz, 1988), and Tuck (1971a, b) documented the evolution of an Onondaga tribe in the late prehistoric/ early historic period in New York (see also Englebrecht, 1974, 1978). In the southwestern United States, a survey conducted by Jeffrey Dean, Alexander 2 Stress and Warfare Among the Kayenta Anasazi J. Lindsay, Jr., and William Robinson (1978) in Long House Valley in north- eastern Arizona documented changes in settlement that corresponded closely with what would be expected in the development of a tribal polity (Creamer & Haas, 1985), Basically, Dean et al. (1978) found clusters of interacting villages coming together politically and economically in the context of what appeared to be conflict or war. To begin looking at how and why a tribal type of organization might evolve in a prehistoric context before any possibility of colonial pressures, additional archaeological survey and excavation were initiated in the area of Long House Valley. The project focused on Long House Valley itself and a surrounding area of about 25 km in radius in the heartland of the Kayenta branch of the Anasazi culture. The work was designed to elucidate the nature of prehistoric tribal organization, the factors affecting the process of tribal formation, and the possible role of warfare in tribal development. The specific focus of the research was on the period from a.d. 1250 to 1300, known as the Tsegi Phase of the Kayenta cultural sequence. This was a time of tremendous change throughout the entire Anasazi area, with new settlement patterns, new kinds of sites, new artifact types, and marked changes in social relations. Thus, the time and the place were set for an investigation of the formation of tribal polities in the prehistoric past. Tribal Organization The concept of "tribe" has long held a position of honor in anthropology. References to this or that "tribe" abound in the literature, and the label "tribe" has been assigned to one of the major stages in various schemes of cultural evolution. Generally, the term has been used to refer to bounded groups of culturally similar peoples. A tribe usually transcends the individual village by bringing multiple communities together in a larger regional polity. At an evolutionary level, this culture type is seen as lying somewhere be- tween the loosely knit "band" and the politically centralized "chiefdom" (Adams, 1975; Sahlins, 1968; Service, 1971; Steward, 1955; Gregg, 1991; cf. Barth, 1969; Friedman & Rowlands, 1978; Lewis, 1968; Johnson & Earle, 1987). While the concept has been widely used and accepted in the discipline, there has been considerable debate over the nature of the tribal polity and the question of how tribes come into existence. Discussions of how and why tribes develop rest to some extent on the concept of "tribe" itself. Just what is a tribe? Overview discussions of tribal organization in ethnography and cultural anthropology are graying today. As the field moves away from broad theories of evolutionary stages, there is a reluctance to try and build explicit models of specific "types" of societies or even to revise existing models. Yet the concept abounds in virtually every major journal of the discipline, and there seems to be some level of common understanding of what anthropologists mean when they talk about "tribes." Looking back in the literature, the most explicit discussions of the basic Tribes and Tribal Formation 3 concept of "tribe" appeared in the late 1960s and early 1970s. Recognizing that these discussions are inevitably dated in terms of current anthropological thought, they nevertheless give a general idea of what anthropologists refer to with the label "tribe." In one of the fullest presentations of a tribal model. Service (1971:101-108; see also Adams, 1975:225-228) sees the tribe as a network of social units or communities bound together by a series of pantribal sodalities. These so- dalities may be either kin-based (e.g., clans, kindreds, segmentary lineages) or non-kin-based (e.g., age-grade systems, warrior societies, religious asso- ciations). In all cases, however, their function is basically the same: They provide a means for culturally uniting individuals from different social or residence units. Residents of different communities, for example, who belong to the same clan are culturally joined together. Furthermore, the communities united by such clan ties, along with other sodality relationships shared by the respective members, would constitute the tribe in the Service model. This network of sodality relationships not only serves to unite the com- munities together as a tribe, but also operates to distinguish the member units of the tribe from other outside units. The formation of the tribe in this sense involves the formation of a cultural boundary, though the nature of this boundary is not clarified by Service. A second model, offered by Sahlins (1968, see also 1972), sees the tribe as a "pyramid of social groups" (1968:15) in which the parts are not necessarily subordinate to one another but play different roles in the social system. Thus, for example, at the basic level of the household, certain kinds of activities are carried out. At the next level of the clan or lineage, there is another set of activities, another set at the village level, and so on, with the "tribal" level being the most inclusive functional unit at the apex of the pyramid. In this model again there is the implication that since the tribe acts as a discrete functional unit at one level, it is a culturally bounded unit, just as in the Service and Adams model. However, Sahlins does state that at the tribal level, the unit may be tied into a "wider, intertribal field" (1968:15; see also Johnson, 1978, 1982). In comparing the two models of tribe, it is apparent that they are not mutually exclusive. Instead, each places emphasis on different kinds of vari- ables in depicting the tribal form of organization. In the first, emphasis is placed on the network of social relationships, whereas in the second, it is on the hierarchy of functional roles. The Formation of Tribes There has been active recent debate over the formation of tribes, and the issues are more clear-cut (Whitehead, 1989). Generally, there are two different ideas about how and why tribes evolve: a "traditional" model that originated in the ethnographic evolutionary models of Service, Adams, and Sahlins, and an alternative model offered more recently in archaeology (Braun & 4 Stress and Warfare Among the Kayenta Anasazi Plog, 1980, 1982; Braun, 1984; Plog & Braun, 1983; Bender, 1985; Upham, n.d.; Voss, 1987). The traditional explanation assigns a primary causal role to warfare in the evolution of the tribe. Basically, it is argued that in the face of population growth related to the beginnings of plant domestication, groups found themselves in competition over land and other resources. In response to these conditions, groups that joined forces with one another would be better able to compete for scarce resources. Out of this competitive milieu, there emerged a new form of polity, a decentralized network of villages applied together in the face of external competition — the tribe. There is nothing in this demographic situation [population growth] alone which would necessarily lead to a tribal form of organization; probably the earliest growing societies simply divided and spread. With competition, however, the larger and better consolidated would prevail, other things being equal. Here again external offense-defense require- ments may well have been the selective factors, signifying as they do the importance of alliance and solidarity. (Service, 1971:101; see also Harner, 1970:70) The process of tribalization in this view clearly involves consolidation and coordination of different units, but the process also takes place within the context of competition and conflict or warfare. Again, to quote Service: The external policy of tribes is usually military only. Usually, too, the military posture is consistently held; that is, a state of war or near-war between neighboring tribes is nearly perpetual. Tribal warfare by its nature is inconclusive. Ambush and hit-and-run raids are the tactics rather than all-out campaigns, which cannot of course be economically sustained by a tribal economy and its weak organization. True conquest, furthermore, would be self-defeating, for the productivity of a defeated tribe would not be great enough to sustain the conquerors. Objectives seem to be booty ... or to drive the enemy out of a favored zone or prevent him from expanding. (1971:104; see also Adams, 1975:228; Car- neiro, 1970, 1978; Sahlins, 1968:17) Thus, while warfare is seen as central to the process of tribal formation, its rule is one of inducing consolidation through cooperation rather than con- quest. The unified tribe then emerges as a discrete political entity separated from its neighbors by hostile (or potentially hostile) relations. A second model that has recently been applied to explain the formation of tribes, argued most fully by Braun and Plog (1980, 1982; see also Bender, 1985), also places emphasis on cooperation. However, in this case, it is argued that the process of tribal consolidation will occur in response to any number of social and/or environmental "risks" (Braun & Plog, 1982:506-508). Con- ceivably warfare could be one of those "risks," but it is not seen as a primary Tribes and Tribal Formation 5 causal variable in tribal formation. Basically, Braun and Plog take issue with the notion that resource shortages and other environmental problems will necessarily lead to competition and conflict. They argue instead that, in the face of such problems, there is more likely to be an intensification of co- operative interaction and alliance between local communities. For example, other things being equal, an increase in population density or a reduction in the scale of residential movement would leave cores- idential units with increasingly smaller areas for direct exploitation. Under many environmental conditions, such a constriction of the area available for direct exploitation would increase the spatial and temporal variance in productive yields among neighboring coresidential units. Such increased local unpredictability in yields would present a potential increase in risk to each local community. If this increasing local unpre- dictability occurs within an already existing social network, we would expect to see an increase in the social connectedness within that network. That is, where lines of cooperation and communication already exist, increased local environmental unpredictability should lead to increased demand on these lines of integration. Except under the most extreme circumstances (e.g. see discussions in Dirks [1980] and in Winterhalder [1980]), sustained increases in such demands should lead to increased formalization of the existing lines of integration, as a consequence of the processes of selection discussed earlier. (Braun & Plog, 1982:508) Thus, for Braun and Plog, tribal polities form as a consequence of increased cooperation, integration, and communication between communities. This view differs from the traditional model in a number of ways. First, in downplaying the role of warfare, the interaction between tribal groups in Braun and Plog's model is significantly different from the competition and conflict envisioned in the more traditional model. Specifically, in the former model, not only is the expectation of intertribal conflict lacking, but there may well be an increase in certain kinds of positive interactions between tribal units (e.g., ritual exchange), as one tribal group establishes alliance relationships with other tribes (Braun & Plog, 1980). Furthermore, in the more traditional model, the formation of tribes in- volved not only increased integration (Service's pantribal sodalities) and coordination (Sahlin's structural hierarchy), but also formation of some kind of social boundary around the "tribe." This kind of boundary formation is a logical outgrowth of the emphasis on the importance of warfare in tribal formation. In a cooperative model, however, the process of boundary for- mation is quite distinct from the intensification of integration and coordi- nation (Braun & Plog, 1982:505; Bender, 1985:55). Accordingly, tribal polities can develop in the absence of the formation of concrete bounded tribes. This view requires a third model of what the "tribal" unit might look like. In a sense then, it might be possible to have a " tribal" form of organization without 6 Stress and Warfare Among the Kayenta Anasazi having a discrete social unit that might be readily identified as a bounded tribe. Tribe as an Evolutionary Stage There has been a recent argument in the archaeological literature against the analytical use of the concept of tribe. This argument harks back to the Boasian critique of the 19th century evolutionary models of Tylor and Mor- gan. Basically, it is argued that by focusing on certain "essential" elements in defining a tribal form of organization, the anthropologist winds up lump- ing wildly disparate cultures into one evolutionary type and avoids rich diversity for the sake of evolutionary uniformity (Leonard & Jones, 1987; Dunnell, 1980; Plog & Upham, 1983; Upham, n.d.; Braun, 1991). Leonard and Jones (1987), in particular, blanket all the evolutionary stage models with the same criticism: Invariably, applications of the model are conceived at a scale much too inclusive and indiscriminating for culture change to be monitored, and, in fact, the model serves to obscure change at the scale at which it primarily occurs (i.e., consider the material variation and culture change that goes unnoted with the punctuated stages of evolution and that the model demands). (1987:200) This type of generic argument against evolutionary stages such as "tribe" or "chiefdom" are really arguments against cultural typologies. By placing societies in broad stages or types, it is argued, the unilineal evolutionists miss the continuous range of variability found in the ethnographic and archae- ological record. Ultimately, however, this boils down to a classic debate between "lumpers" and "splitters" and constitutes a spurious argument against the evolutionary models and the different specific stages. While it is true that such models and stages gloss over a great deal of cultural diversity, they were never developed to account for diversity. They were developed to account for certain broad patterns of similarity that can be observed cross- culturally in the record of human societies. Any typology, be it for ceramics, house forms, or politics, is developed to answer certain questions about the phenomenon under study. Anthropology, fortunately, has moved beyond the point of trying to "discover" certain true, immutable types that are applicable in all circumstances and for all purposes. In the case of the different evolutionary models, the various stages or cultural types, such as tribe, have been generated to account for cross-cultural patterns in the evolution of specific aspects of cultural systems. Both Service (1971) and Adams (1975), for example, look at evolutionary patterns in terms of social integration and centralization of the different parts of the system, whereas Fried's model of political evolution looks at evolutionary patterns in terms of access to resources, status, and power. These different models Tribes and Tribal Formation 7 produce different explanatory arguments about the evolution of cultural systems and quite different evolutionary stages. They are not, however, contradictory to one another. They are simply focusing on different parts of the system. This is no different from having one ceramicist develop a typology of vessel form and function to learn about room use while another develops a typology of design elements to learn about social relationships. Neither typology is right or wrong and neither can be reasonably criticized for not encompassing the full range of ceramic variability. The typologies were developed with different questions in mind and serve different purposes for the analyst. In none of the evolutionary models is it argued that there is a lack of diversity within broad levels. Rather, it is argued that if one looks cross- culturally at certain aspects of the cultural system, there are patterns of similarity, and there is an evolutionary relationship among these patterns. In the case of the current project it is argued, following Service, Adams, and others, that "tribe" is a valid evolutionary type that can be effectively used to analyze well-integrated but decentralized regional systems of communi- ties. Such tribal systems represent an evolutionary development out of an antecedent pattern of unintegrated, independent band communities, and they precede the centralized regional systems of chiefdoms. Problem Domains In comparing the different models of tribes and tribal formation, a number of problem areas come into focus. First, in all of the models, the process of tribal formation is seen as involving increased interaction and solidification of bonds between separate social units. Beyond this level of agreement, questions remain about the nature and strength of those bonds. Are emergent tribal units held together by an essentially homogeneous web of sodality relationships? Or are individuals and communities involved in a hierarchy of relationships that vary in function and intensity as one goes from the household up to the tribal levels? Braun and Plog demonstrate that inter- community relationships intensify in the face of common risk, but they do not show how communities or social units actually interact in a tribal polity. It is quite likely, of course, that different cultures will have different patterns of interaction; however, without information on the nature of such inter- action in prestate aboriginal groups it will be impossible to generate anything more than a hypothetical model of tribal organization. A second problem domain concerns the causal variables contributing to the process of tribal formation. The ethnographers argue for warfare, while the archaeologists would broaden the field by subsuming warfare under the more general term of environmental "stress" or "risk." Braun and Plog convincingly argue that tribal polities can and do evolve in the absence of warfare; however, they do not address how warfare may play a role in the 8 Stress and Warfare Among the Kayenta Anasazi process of tribal formation. In fact^ a comparison of Braun and Plog's theo- retical model of tribal formation with the ethnographic evolutionary models leads to an alternative hypothesis regarding tribes and warfare. Namely, warfare acts to solidify existing unbounded tribal networks into concrete bounded tribes. In other words, warfare need not play a causal role in the intensification of social networks, but it may be a necessary variable in the formation of boundaries around those networks. Some empirical data to support this hypothesis may be drawn from the prehistoric Iroquois (Englebrecht, 1974; Tuck, 1971a,b; Whallon, 1968), but these data are suggestive at best, and the hypothesis is clearly in need of explicit testing with a more comprehensive data base (Englebrecht, 1978; Trigger, 1981). At the same time, it must be recognized that warfare itself may arise in response to or in combination with other forms of environmental "stress," such as demographic pressures or environmental degradation. Con- sequently, to fully understand how tribal polities form, a wide range of changes in the environment of the evolving society must be considered, even when warfare is present. A third problem area is related to the second, specifically, how does the process of boundary formation relate to the process of intensification or interaction? Braun and Plog point out that while often confused, these are not one and the same process. Consequently, if interaction increases with increased risk, then questions still remain about how and why boundaries may form around interaction networks. While the warfare hypothesis was mentioned above, it might also be hypothesized that intensification of in- tratribal links might progress to such a point that a de facto boundary was formed around the tribal unit. Such an event might be predicted when a group of communities was subjected to a qualitative and relatively permanent increase in the level of environmental stress. Beyond these basic questions of how and why tribal boundaries form, there are further questions related to the nature of such boundaries. Since no one argues that a tribal boundary eliminates all interaction between the tribe and outside groups, the question becomes one of how much interaction there is, and what form that interaction takes. If warfare is the primary vehicle of boundary formation, the intertribal interaction might be expected to be predominantly military. Input from outside groups would then be most likely to take the form of captives and material resources. On the other hand, if environmental stress is a primary vehicle for boundary formation, then intertribal interaction might be predicted to be cooperative. In such a case, input from the outside might take the form of limited trade or exchange of either scarce or luxury resources. Again, answers to questions about the formation and nature of tribal boundaries require collection of specific kinds of information from prehistoric societies. In examining these interrelated problem domains archaeologically, we do not expect to "prove" any of the individual models of tribal organization or formation. Rather, we hope to distinguish the components of the different Tribes and Tribal Formation 9 models having the greatest utility in explaining the process of tribal for- mation, and provide a more comprehensive empirical foundation for un- derstanding the tribal form of organization under pristine, prestate condi- tions. Summary During the 13th century, the Kayenta Anasazi in northeastern Arizona un- derwent a basic change in social organization. From nonhierarchical, dis- persed villages, they formed larger clusters of cooperating communities with some communal economic, ideological, and political activities. They looked like what we might expect for a prehistoric tribal type of organization. Archaeological evidence from the region may make it possible to elucidate the factors that contributed to the shift that occurred during the Tsegi Phase, A.D. 1250-1300. Central to the study of tribe formation is a consideration of the role of warfare as a catalyst to the other factors causing stress on the regional pop- ulation. Anthropological theories of societal change have long focused on the role of conflict in forging new configurations of settlement, economy, and ideology. In the case of the Kayenta Anasazi, integration of society appears to have persisted in the face of increasingly adverse environmental and economic conditions during the latter half of the 13th century, bringing the potential role of warfare into the arena of study. The emergence of tribal polities represents a critical stage in the evolution of cultural systems. It marks the time when village autonomy is transcended as people join in larger political networks. There is an increase in cultural diversification as tribal communities pull together internally and pull apart from other similarly organized communities. Varying relationships develop between the emerging tribal groups based on some combination of coop- eration, competition, and conflict. The cultural system overall becomes more complex, with tighter integration within the tribes and decreased integration between tribes. The study of the process of tribalization therefore offers potential insights into the beginnings of political aggregation, diversification, interaction, and integration. Below we explore the formation of prehistoric tribes in the American Southwest, in an effort to gain a greater understanding of some of the foundations of the evolution of political systems across cultures and time. An effort has been made to take in a large area and to use both traditional and nontraditional field methods. Though our net may be cast broadly, the data returned clearly show that regional studies such as this one yield the variety of information that is needed to address complex questions. 2 BACKGROUND AND METHODOLOGY The southwestern United States offers an optimal laboratory for investigating the process of the formation of tribal polities in a prehistoric, precolonial context. There are several reasons for focusing on the Southwest for such an analysis. First, there is better chronological control for the prehistoric period in the Southwest than for virtually any other world area. Tight chronological control is critical to a study of tribal development because different phases of the process may occur within the span of a single generation. Second, at least for parts of the Southwest, there is an incomparable record of the past environment. Working with tree rings, pollen, and hydrology, scholars have reconstructed detailed records of past precipitation, erosion, groundwater fluctuation, and patterns of short- and long-term temporal and areal variation (Gumerman, 1988). This kind of detailed paleoenvironmental record is important for assessing the potential role of external or environmental stress in the process of tribal formation. Third, the large amount of previous archaeological work con- ducted in the Southwest provides a solid cultural foundation for the study of how tribes may form. The basic time-space frameworks have already been worked out and can be used as a starting point rather than an initial research objective. Finally, there are extant tribal groups in the Southwest, such as the Hopi and Zuni, that have deep historical roots extending back into prehistory. The Kayenta Anasazi With the Southwest as a broad laboratory, archaeological research was ini- tiated in one specific culture area, that of the Kayenta Anasazi, to begin addressing central questions about the nature of tribal organization and the causal forces behind the process of tribal formation (fig. 2-1). Again, there 11 12 Stress and Warfare Among the Kayenta Anasazi Map by Gigi Bayliss Figure 2-1. Map of the Four Corners region showing the general location of the research area. are several reasons for concentrating on the Kayenta Anasazi, who occupied what is today northeastern Arizona. First, a combination of tree-ring dates and well-analyzed pottery types provides for highly refined dating of Kayen- ta sites. This is true not only for excavated sites, but also for unexcavated sites discovered in the course of survey. Surface ceramics, in particular, allow the dating of Kayenta sites to within 50 years for most of the occupation sequence, and to within 25 years for the last century of occupation (Dean et al., 1978:29; Gumerman, 1988). The second reason for working in the Kayenta area is that numerous archaeological projects in the Kayenta area have provided a rich background for current problem-oriented research into tribal formation. Among the pro- jects in the Kayenta area are the survey and excavation of the Rainbow Bridge-Monument Valley Expedition (Beals et al., 1945; Hargrave, 1935; Christenson, 1983), the Glen Canyon Project (Lindsay, 1969; Lindsay et al., 1968), the Black Mesa Archaeological Project (Gumerman & Euler, 1976; Gumerman et al., 1972; Powell, 1983; Plog & Powell, 1984; Plog, 1986; Powell & Gumerman, 1987), and the Black Mesa Railroad Corridor Project (Swarthout et al., 1986). Together, these projects have fleshed out the broad outlines of Kayenta prehistory from the first millenium B.C. to regional abandonment in A.D. 1300 (Haas, 1989). Finally, because of the large amount of work done over the past two decades in the Kayenta region, the reconstructed paleoenvironmental record is among the best in the Southwest. Past environmental variability in the Kayenta Background and Methodology 1 3 Figure 2-2. Aerial photograph of Long House Valley, facing southwest, (Courtesy of the School of American Research, Santa Fe, New Mexico: 1993. Photograph by Paul Logsdon.) region has been assessed through a variety of paleoenvironmental studies, including analyses of tree rings, pollen, and hydrology. Intensive chrono- stratigraphic analysis of more than 50 independently dated stratigraphic _ sections in floodplain deposits provides detailed reconstructions of changes in alluvial groundwater levels and of alternating episodes of floodplain ag- gradation and dissection (Euler et al., 1979). Fluctuations in alluvial vs^ater tables coupled with aggradation or erosion would have had major conse- quences for the Anasazi inhabitants of the Kayenta region, who subsisted mainly by farming valley bottoms. Dendroclimatic analyses document an- nual variations in precipitation from A.D. 680 to 1970 (Dean & Robinson, 1977) that would have affected yearly crop yields. Access to these high- resolution paleoenvironmental data allows immediate investigation of the role of environmental stress in the development of political systems in the valley. Taking the region of the Kayenta Anasazi as a study area, research was designed and implemented in 1983-1986 to gain insights into the nature of tribal organization in a prehistoric setting and to address questions about the possible role of warfare and environmental stress in the formation of tribal-type polities. The focus of this research was on one small valley. Long House Valley, in the center of the Kayenta heartland. Long House Valley is a small wedge of land (75 sq km) nestled between Black Mesa on the south and the Shonto Plateau and Tsegi Canyon system on the north (fig. 2-2). The eastern edge of Long House Valley is delimited by Marsh Pass, which 14 Stress and Warfare Among the Kayenta Anasazi marks the mouth of the Tsegi Canyon system to the north and leads out into Kayenta Valley to the east. To the west, a string of hills separates Long House Valley from Klethla Valley. Long House Valley provided an ideal location to begin the investigation of the process of tribal formation. Long House Valley In 1978, Alexander J. Lindsay, Jeffrey Dean, and William Robinson completed a 100% survey of Long House Valley and recorded 551 prehistoric sites dating from ca. 2500 B.C. to a.d. 1300 (Lindsay & Dean, 1971; Dean et al., 1978). The evolving pattern of settlement that emerged from this survey looked on the surface like what might be expected to appear in a case of prehistoric tribal formation (Haas, 1980; Creamer & Haas, 1985). Although the history of occupation in the valley is complex and interesting, the time of immediate interest for the present project is the Pueblo III period, from A.D. 1100 to 1300, during which the archaeological and environmental data indicate the emergence of a tribal polity under conditions of stress. At the beginning of this period, between A.D. 1100 and 1150, the residents of Long House Valley were living in numerous small villages of two to ten rooms for one to four families. Some of these villages had kivas (ceremonial structures) while others did not. Sites from this period were distributed fairly evenly along the edges and across the floor of the valley (fig. 2-3). Although exact population estimates are not available, the relative population density in the valley was higher than in previous centuries, though not at a maxi- mum, which came in the 13th century (Dean et al., 1978; Effland, 1979). The environment in A.D. 1100 across all of northern Arizona was quite favorable for subsistence horticulture, with a relatively high water table, high annual precipitation, and near the top of a cycle of soil aggradation (Dean et al., 1985). With regard to social relationships, the Long House residents shared the same material cultural assemblage as people residing in other parts of the Kayenta region. There are indications that ceramics and the raw materials for making stone tools were being exchanged between valleys at this time (Green, 1983, 1986; Garrett, 1986). There is, however, no sign of any type of systematic interaction or coordination either between villages within the valley or between the valley residents and residents of neighboring valleys. Warfare and conflict are not manifested in any way in the settlement pattern or the burial population. Overall, the available data base from the Kayenta heartland at the beginning of the 12th century does not point to a supravillage tribal organization. Beginning in approximately A.D. 1150, the environment in northern Ar- izona began to change and the local population changed with it. There was a drop in the alluvial groundwater levels, a major short-term drought, and the start of a period of soil degradation caused by erosion (Dean et al., 1985). Background and Methodology 15 / LONG HOUSE VALLEY HABITATION SITES OCCUPIED BETWEEN AD 1000 AND 1150 1 Ml H 1 KM ■J 1 O HABITATION SITES — • UMIT OF SURVEY Figure 2-3. Map of Long House Valley in the period a.d. 1000- 1150 (adapted from Dean et al., 1978). In response to these environmental shifts, portions of the Kayenta region were abandoned completely. Black Mesa, immediately to the south and east of Long House Valley, was always marginal for horticulture, and with the drop in precipitation and the water table, the residents of this area were completely gone by a.d. 1150 (Powell, 1983). Between A.D. 1150 and 1250 there was a substantial shift of the population toward the northern half of the valley (fig. 2-4). This would have increased the population density locally, though the total population in the valley as a whole did not go up substantially during this period. As in the previous period, the villages remain small, and there are no overt indications of the emergence of a discrete tribal type of organization uniting the villages within 16 Stress and Warfare Among the Kayenta Anasazi the valley or bringing together the Long House residents with residents of neighboring valleys. At the same time, as inferred from the increased densities and clustering of villages in the northern half of the valley, there would have been increased interaction, coordination, and /or cooperation in the distribution and utili- zation of limited field systems. As with previous periods, there are no signs of either competition or conflict between the valley residents, nor signs of warfare with people outside the valley. However, the shift of the population out of the southern half of the valley resulted in the beginning of a de facto physical boundary between Long House residents and the people living in Klethla Valley to the south and east. The fact that Black Mesa to the south and west was also completely abandoned at this time further extended the physical boundary around the Long House population. In the latter half of the 13th century, there was continued degradation of the environment with increasing loss of arable land due to erosion and a dropping water table. The fall in the water table would have dried up many stable sources of drinking water as well, and while precipitation was variable annually, another major drought occurred in the 1260s (Dean, 1969). These environmental trends were accompanied by dramatic cultural changes among the Kayenta Anasazi of northeastern Arizona beginning in roughly a.d. 1250. Within Long House, population had increased, probably through immi- gration, to a maximum for the valley. The abandonment of the southern half of the valley was complete, and the population came to reside in five discrete clusters of villages (Effland, 1979; Dean et al., 1978) (fig. 2-5). Fur- thermore, for the first time, there appeared a clear differentiation between residential villages. Specifically, each of the clusters consisted of a single "focal" village of 75 to 400 rooms around which there were 2 to 13 smaller "satellite" villages of 2 to 25 rooms. Four of the five focal communities also had reservoirs for the collection and storage of water. Each of the focal villages, in addition to being much larger than the satellites, is distinguished by the presence of long, "spinal" roomblocks with cored, double-faced ma- sonry, and by open plaza areas (Dean et al., 1978:33). The focal villages also mark the first possible indication of conflict or warfare in the Kayenta region. All five focal sites were situated in defensible positions on a hilltop, knoll, or isolated cliff, overlooking the land around the satellite sites in the cluster. The focal communities were also located to afford a clear line of sight with each of the others and with all the primary access routes leading into the valley. The visual network of focal sites also indicates there was some degree of unification within the valley as a whole. These focal sites then provide the first indication of both warfare and possible tribal integration in Long House Valley. Taken as a whole, the existing survey and paleoenvironmental data for Long House Valley and the Kayenta region in general indicated at the be- ginning of the present project that this would be a good area to assess models Background and Methodology 17 LONG HOUSE VALLEY HABITATION SITES OCCUPIED BETWEEN AD 1150 AND 1250 1 Ml -I 1 KM O HABITATION SITES -• UMIT OF SURVEY Figure 2-4. Map of Long House Valley in the period a.d. 1150- 1250 (adapted from Dean et al., 1978). of tribal organization and tribal formation. There were signs of increased interaction in the clustering of villages and communal facilities; integration, as seen in the visually interlocking network of sites; a functional hierarchy, manifested in the focal and satellite villages; the environmental stress of drought and erosion; and the defensive posture of the focal sites, which pointed to a possible pattern of warfare. The initial survey data of Dean et al. (1978) from Long House Valley offered tantalizing but inconclusive insights into tribal organization and develop- ment in a prehistoric context. Substantial questions remained to be answered through examination of the sites and new, problem-oriented field research. Key problem areas centered on integration and conflict. On the integration 18 Stress and Warfare Among the Kayenta Anasazi / y^ — ^ v^ ho^^^T f ^^£ / \ I V. - "■ \ ''^M V %^ - / \- * ii / I ^ \ z o \ t^ 1 o 1 1 X 1 LONG HOUSE VALLEY 1 ^ HABITATION AND FOCAL SITES \ OCCUPIED BETWEEN \ AD 1250 AND 1300 1 A) j } 0 1 Ml / 0 1 KM ^r / ^ K X- / o HABITATION SITES -k FOCAL SITES ^0^ UMIT OF SURVEY Figure 2-5. Map of Long House Valley in the period a.d. 1250- 1300 (Tsegi Phase) (adapted from Dean et al., 1978). side, there were questions about the nature of political relationships within Long House Valley and between the valley and its neighbors. What was the nature of the relationship between the focal sites and the satellite sites? Were these two site types functionally different or did they differ only in size? Were the satellites homologous units, or were there structural differences between them? Were there specific sodalities holding together the different units within the Long House Valley system? On the conflict side, there were questions about the possible role of warfare in shaping the development of the Long House Valley polity. Was the environment the primary driving force behind the observed changes in the 13th century, or was warfare an intermediary causal factor? Was there a bounded Long House Valley polity Background and Methodology 19 or were they united with neighboring valleys in the Kayenta region? If there was conflict in the Kayenta region in the 13th century, was it intraregional or between the Kayenta and neighboring culture groups in the Mesa Verde or Hopi areas to the north and south? Answers to these questions about the nature and evolution of tribal or- ganization in the Kayenta heartland required the collection of new data from the field. Questions about integration required specific data from sites within the valley, to give insights into possible differences in site organization and function. Questions about conflict required information about a broader range of sites both within and outside the valley to give insights into the possible nature, extent, and importance of warfare in Long House Valley and surrounding areas. Methodology In addressing the issue of tribal formation in prehistory, it is necessary to operationalize a number of different concepts such that they can be recog- nized and assessed in the archaeological record. Specifically, the concepts of "interaction," "functional hierarchy," "integration," "boundaries," "environ- mental stress," and "warfare" must be operationalized. Interaction In the general models of tribal formation, there will be an increase in in- teraction, or connectedness, between the members of an emerging tribal unit. Measuring the degree of interaction can be done by looking at patterns of information exchange between individuals or groups and at manifestations of cooperative activities. Systematic analysis of information exchange in pre- historic societies is relatively new in archaeology and has focused primarily on an analysis of variability of style in artifactual production and decoration. It has been argued specifically that stylistic variability in ceramics and lithics serves to communicate messages about such things as group affiliation, own- ership, religious ideology, or maker of the artifact (Binford, 1972; Braun, 1977; Braun & Plot, 1982; Graves, 1981; Hodder, 1982; Plog, 1980, 1983; Wobst, 1977). The exchange of stylistically variable artifacts, therefore, between in- dividuals or social groups is a mechanism of communication or information exchange as much as it is an economic transaction between exchange partners. While the basic theory of information exchange, coupled with Braun and Plog's specific predictive model, offers one methodological avenue for using stylistic variability to measure social connectedness, a second avenue also merits investigation. In contrast to the information exchange model, an "in- teractional" or "sociological" model of style holds that stylistic similarities and differences are a reflection of the intensity of interaction between the markers of the artifacts (Graves, 1981; Voss, 1980; Washburn, 1978; see also Hill, 1970; Longacre, 1970; cf. Friedrich, 1970; Hodder, 1979). The more people 20 Stress and Warfare Among the Kayenta Anasazi interact or the greater the degree of social "connectedness," the greater the degree of similarity in stylistic elements. Following this interactional model, the process of tribalization should be characterized by an overall decrease in design variability and a general pattern of stylistic homogenization in the tribal unit. While the information exchange and interactional models of stylistic vari- ability appear to offer opposing predictions for the patterns of stylistic vari- ability expected in the process of tribalization, in reality they may comple- ment one another. In particular, it is quite possible that some stylistic attributes may follow the expected patterns of the interactional model, and other kinds of attributes may follow the expected patterns of the information exchange model (Braun, 1983; Plog, 1983; Voss, 1980). Thus, in using stylistic variation as a means of examining interaction and social connectedness in emerging tribal societies, both possible patterns must be taken into consideration. Another means of examining changing patterns of interaction in the pro- cess of tribalization is through an analysis of evidence for cooperative activ- ities. The increased connectedness predicted for tribes should be characterized by greater degrees of cooperation between the member units of the tribal unit. Thus, there should be more joint activities carried out at all levels of the tribe, including households, kin units, communities, subtribal units, and the tribe itself. Internal cooperation within a tribe may also be manifested in the economic and social interdependence of the various tribal units. With the high level of integration predicted for tribal forms of organization, social and economic responsibilities may be consolidated and assigned or assumed by different parts of the tribal unit. Social consolidation for ceremonial or political activ- ities is to be expected as a means of explicitly uniting and integrating the member units of a tribe. Such social consolidation is exhibited by communal or shared ceremonial architecture and possibly other kinds of communal activity areas not directly related to the mode of production — for example, formal plazas or nonreligious communal buildings ("town halls"). Consolidation of economic responsibilities can be expected as an adaptive response in emergent tribes, given that tribes are argued to arise in response to severe environmental stress and /or intensified competition and conflict with neighboring groups. Under such conditions, limited pooling of re- sources and labor as well as small-scale specialization would enhance the economic viability and competitive position of the tribe as a whole. Pooling of resources and labor may be manifested archaeologically in communal storage facilities, small-scale communal labor projects such as field prepa- rations, water control devices, or defensive walls and fortifications. The eco- nomic specialization in tribes can be expected to be limited to the production of craft items such as ceramics and lithics. Specialization in construction activities, warfare, or the production of food would not be expected in these decentralized, nonhierarchical societies, where surplus production serves to ensure long-term survival rather than the support of discrete social strata of Background and Methodology 2 1 craftsmen, warriors, laborers, and the economic elite. This level of special- ization stands in contrast to the large-scale labor specialization and class formation characterizing more complex, politically centralized societies. Eco- nomic specialization in the prehistoric tribe would be marked by the con- centration of specific productive tools in the burials or residences of a small minority of the population. The large majority of the residences and burials, on the other hand, should not be marked by major economic differences or artifactual indications of specialization. In the Kayenta region, existing survey and excavation data indicated a number of patterns of interaction in the period from a.d. 1100 to 1300. In terms of ceramic designs, Kayenta pottery throughout this entire period was characterized by remarkable homogeneity. Pottery samples from one end of the Kayenta region to the other were virtually indistinguishable from one another in terms of patterns of design and in terms of the sequences of changes in designs over time. This homogeneity by itself is a strong indication of significant interaction between the component communities of the Kayen- ta region (Haas, 1989). No stylistic analysis has been done, however, to indicate the kinds of interaction patterns discussed by Braun and Plog. Within Long House Valley itself, the changing settlement pattern during this period is indicative of changing patterns of interaction. At the beginning of the period, the villages were spread evenly over the valley floor, with no direct evidence of interaction. By the Tsegi Phase, the villages are in distinct clusters, and within these clusters the bigger focal sites have plazas, large communal structures, and reservoirs. All of these characteristics fit the ex- pected pattern of increased economic and social interaction. Integration In the ethnographic models, integration is, in fact, a particular kind of in- teraction that is manifested primarily in the form of sodalities in tribal-type societies. The sodality relationships hold the parts of a tribe together and serve to distinguish it from other tribal units. Thus, there should be specific patterns manifested in the archaeological record of tribal societies. Although sodality relationships might be marked by stylistic variability in ceramics or lithics, burials and ceremonial architecture offer potentially more fruitful and less ambiguous avenues for detecting sodalities archaeologically. In re- gard to burials, it is possible that sodality membership may be marked by the inclusion of specific objects or types of burial furniture. Different age grades, warrior societies, religious societies, or clans, for example, may have material or symbolic correlates that could be buried with an individual at the time of death. At the same time, certain types of sodalities, particularly clans and religious societies, can be expected to have either sodality-specific architecture or distinctive architectural features that distinguish one sodality from another (see, e.g., Dozier, 1970; Eggan, 1950; Ellis, 1979; Hoebel, 1978; Ortiz, 1969). 22 Stress and Warfare Among the Kayenta Anasazi The predicted pantribal network of sodalities seen in the ethnographic tribal model should be manifested archaeologically in a number of different patterns. First, at the level of the community, there should be a number of different sodalities present in each independent community. For example, four or five different clans or age grades might be represented in each community of a tribal system. (Clans and age grades are used here only to illustrate the patterns of sodality relationships that may be expected. In actuality, both clans and age grades may be quite difficult or impossible to recognize archaeologically.) Second, a similar range of sodalities should be present at all of the communities. Thus, the same four or five clans or age grades should be found at all of the communities of the tribe. Third, there should be limited variability in sodalities at the tribal level as a whole. Continuing the example, the total number of clans or age grades found at the tribal level should not significantly exceed the total number of clans or age grades found at any individual community. Concomitantly, there should be an absence of markers of unique or community-specific sodalities (clans or age grades). Existing evidence of sodalities in the Kayenta region is at best indirect. In looking at the surface indications of other kivas in Long House Valley and at kivas excavated elsewhere in the Kayenta area, we find that the Tsegi Phase witnessed a great proliferation of kiva types (Dean, 1970; Lindsay, 1969; Lindsay et al., 1968). Within Long House Valley alone, there are ex- amples of round, oval, square, rectangular, D-shaped, and keyhole-shaped kivas. Although the function of these kivas is not known, in modern Western pueblos kivas are associated with kinship groups and /or religious societies (Eggan, 1950). If the Kayenta kivas were indeed associated with kinship or religious groups, and if different shapes represented different kinds of groups, then the proliferation of kiva types in the Tsegi Phase would signal a marked increase in the number of these groups cutting across Kayenta society. These are, of course, big "ifs"; however, the appearance of multiple kiva types in the Tsegi Phase is a fact, and one that needs explanation. Why would so many different types of kivas arise in the Kayenta area in the Tsegi Phase? One parsimonious answer is provided in Service's evolutionary model, where it is argued that tribal formation involves the development of a network of pantribal sodalities holding together the different communities and social units within the tribe. The development in the Tsegi Phase of different types of kivas within the same residential clusters is an expectable archaeological manifestation of such a pattern of sodality development. Along this same line, it is relevant to note that there is a much higher ratio of kivas to rooms in the small satellite sites than in the focal sites. In Long House Valley, for example, there are no indications of a single kiva among the 400 rooms of Long House Ruin, and signs of only two kivas each at Tower House, Fireside, Otherside, and Organ Rock Ruin, each of which has 75 to 200 rooms. In contrast, there are one or two kivas at almost every Background and Methodology 23 satellite site with only 3 to 25 rooms total. This kiva distribution indicates that certain kinds of ceremonial activities are more common at the satellite sites than at the focal sites. While this may not be directly related to sodalities, it is another indication of increased interaction within the clusters. Functional Hierarchy The functional hierarchy characteristic of tribal polities in the ethnographic record should have fairly clear-cut manifestations archaeologically. Each of the levels of the hierarchy has responsibility for the accomplishment of different kinds of tasks. An analysis of artifacts and architecture, therefore, should reveal specific activities that are carried out jointly by the member units of each hierarchical level. Consequently, there should be certain kinds of activities that are carried on predominantly at the household level, such as sleeping, eating, or cooking; activities that are carried on at the suprafamil- ial kin group level, such as storage, farming, or food processing; activities that are carried on at the community level, such as construction or archi- tectural maintenance; activities that are carried on at the multicommunity or subtribal level, such as water control or ceremonial behavior; and finally activities that are carried out at the tribal level, such as defense and offense. The precise nature of such a functional hierarchy should change from one tribe to another, as may the activities carried out at each level. However, if the model is an accurate one, there should be evidence of distinct organi- zational levels and of distinct activities carried out at each. Again, it should be stressed that evidence for a functional hierarchy may well be found accompanied by evidence for pantribal sodalities, as the two are not mutually exclusive patterns of organization. The most obvious manifestations of a functional hierarchy in the Kayenta region are to be seen in the emergence of the five distinct site clusters in Long House Valley and in the differentiation of the focal sites from the satellite sites. The clusters represent a new settlement unit above the level of the individual village. The two different site types reflect in turn a func- tional differentiation, with communal facilities and reservoirs concentrated in focal sites and religious structures more commonly associated with satellite sites. The survey data also indicated another possible functional level above the clusters, as manifested in the line-of-sight connections between the focal sites. To the extent that this visual "network" reflected coordinated com- munication and interaction between the clusters, it could be interpreted as another functional level within a valley- wide community hierarchy. Boundaries Boundaries around a tribal unit may be either physical or social, and they may exhibit greater or lesser degrees of permeability. A physical boundary 24 Stress and Warfare Among the Kayenta Anasazi may be marked by the geographical separation and isolation of the tribal unit. The limits of an island, for example, might demarcate the limits of a tribe, as might the sides of a valley or simply a broad band of unoccupied land. An artificial boundary might also be created by the tribal unit itself, in the construction of a defensive perimeter around the tribal residences and productive land. Such a perimeter might be marked by defensive walls, lookouts, and /or fortifications. Indications of physical boundaries around a tribal unit need to be con- firmed by evidence of corresponding social boundaries. It may be the case, for example, that residents of neighboring islands or valleys, while physically separate, are socially united into a larger unit. The formation of distinct social boundaries in the course of tribalization should be marked by a substantial decrease in interaction with external groups. As the member units of a tribe increase their internal connectedness, their ties with outside groups should decrease accordingly. Such a decrease in external interaction could be seen archaeologically in a substantial decrease in imported resources and trade goods. Immigration into the tribal area should also come to a halt as ties with the outside decline. (It should be noted, however, that tribal warfare is frequently marked by wife-stealing, and as a result, foreign women may be "imported" into a tribe that is otherwise socially bounded. Such an event should be manifested archaeologically by a limited intrusive assemblage of artifacts in a single household and in direct association with a much more comprehensive native or local assemblage.) Concomitant with the decline in ties with the outside, the tribal unit should become internally differentiated from other units around it. This process of self-identification of a tribe, often seen ethnographically in the emergence of a tribal name, may be seen archaeologically in tribal identification markers, such as specific stylistic attributes in ceramics, pictographs, burial practices, religious paraphernalia, or items of dress or jewelry. There should also be a general tendency toward material homogenization in a socially bounded tribal unit. With a decrease in interaction with the outside, there are fewer sources of introduced variation and innovation and a greater likelihood that the member units of a tribe will resemble each other more than they will any of their foreign neighbors. The available data from the Kayenta region were not particularly useful for making inferences about possible boundaries within or between the region. Broad survey data across the entire northern Southwest indicate that the Kayenta people were certainly interacting with their neighbors, as Kayen- ta pottery is found frequently as a "tradeware" in many outside areas. In- terestingly, relatively little foreign tradeware is found in the Kayenta region, though other kinds of materials are abundant, including obsidian, turquoise, and shell imported from the outside (Haas, 1989). Within the Kayenta region, no surveys have been conducted that would indicate whether or not there may have been intraregional boundaries. Background and Methodology 25 Environmental Stress Environmental stress on human populations may result from variability in climate, changes in hydrologic or erosional factors, or overexploitation of natural resources, including land. Looking at possible environmental stress in the archaeological record involves analysis of ecofacts — botanical, pollen, and dendrochronological samples — and pedological analysis of the soil. It is expected from Braun and Plog's model that the actual process of tribalization would be preceded or accompanied by environmental changes that intro- duced new sources of risk or stress into the local ecosystem. As noted in the brief summary of Long House Valley prehistory, there is clear evidence of environmental stress in the Kayenta region, beginning in about A.D. 1150. Various paleoenvironmental data indicate a drop in precip- itation, cyclical droughts, and increased erosion during this period. The direct impact of this stress will be discussed later in this monograph. ]Narfare Although this monograph is ultimately about the process of tribal formation, a seemingly disproportionate amount of attention is given to warfare. This is because for the Kayenta area, at any rate, warfare seems to be a primary causal variable leading to the development of tribal-type polities in the 12th and 13th centuries. Warfare, however, is only part of the process, and it must be placed in the context of a wider evolutionary trajectory in order for its role and significance to be understood. It is difficult to generate a common, cross-cultural definition of warfare that can be applied to all types of societies, and anthropologists have offered numerous definitions over the past century. Ferguson (1984), after comparing a number of different kinds of definitions, offered a broad definition to include a range of behaviors: "war can be described as follows: organized, purposeful group action, directed against another group that may or may not be organized for similar action, involving the actual or potential appli- cation of lethal force" (1984:5). In contrast to "advanced" forms of warfare found in chiefdoms, states, and other complex societies, tribal warfare is relatively simple. Staged, formal "battles" are less likely in tribal warfare, and there will not be large numbers of fatalities. Rather, tribal warfare may be expected to consist primarily of small-scale sporadic raiding with limited physical contact. The objectives of such raids would be wife-stealing, limited acquisition, and general destruction of an enemy's resources (Service, 1971). Clearly, raiding of this sort will not leave a dramatic mark on the archae- ological record. Nevertheless, there should be some signs that a prehistoric tribal group was either engaged in some form of conflict with foreign groups or was at least concerned about potential conflict. The latter may be objectified in the construction of defensive features, such as walls or moats, or the 26 Stress and Warfare Among the Kayenta Anasazi deliberate selection of defensible site locations. Deliberate selection of such locations may be inferred when ready access to resources, water, or arable land is sacrificed in exchange for elevation, difficult access, unrestricted or strategic vistas, or physical protection from attack. "Defensible" localities are those that can be protected from entrance by outsiders. These may include geographical features of the landscape that limit access, such as hilltops, mesa tops with steep escarpments and few access routes, steep-sided ridges, in- accessible caves, narrow spits of land surrounded by water or a steep drop- off, or any location that is amenable to controlled entrance. "Defensive" sites may also be in "defensible" locations, but in addition such sites will have specific cultural features, such as perimeter walls or artificially restricted access routes, explicitly designed to defend the site from attack. Constructed defenses may be applied to any site, whether or not it is in a naturally defensible position. These include walls, palisades, deadfalls, moats or ditches, retractable ladders, windowless or slit-window structures, and generally large-scale construction. Indirect defenses include interior wells, reservoirs, and the stockpiling of food and resources. Line-of-sight visibility between neighboring villages provides another kind of defense mechanism in that it provides for communication between allies in case of attack. Thus, if one community comes under assault by an enemy group, if they have line-of-sight visibility with neighboring communities, they can signal for support and reinforcements. It is likely that under conditions of endemic warfare, a number of defense mechanisms will be employed together, with a combination of natural and constructed defenses. Specifically, during such periods, we would expect sites consistently to be constructed in defensible locations that could be protected from raids. There should also be open lines of visible communication between close neighbors, and within a larger system of communities some villages should command views of the main geographical access routes into the area. Turning from site-specific kinds of archaeological evidence for warfare, it should also be possible to distinguish, at least on a general level, the scope of the conflict (in terms of who was fighting whom) and the scale of the conflict. If the conflict is relatively small scale — between individual villages — we should find individually protected defensive sites, each attempting to protect a water source, plots of farmland, and stored foodstuffs. Under such conditions, the settlement pattern should be relatively dispersed, with some buffer zone, or no-man's-land, separating the field systems of the individual villages. If the conflict has a wider scope, with enemies located at some distance from one another, then the defenses might not be focused on the individual villages but, rather, on defensible "central places," where common goods could be stored and protected. Such central places might also serve as de- fensive redoubts to which the population could retreat in the face of a threatened attack. Under such conditions of more distant enemies, the buffer zone would be much wider and would surround clusters of friendly or allied Background and Methodology 11 villages. Alliances themselves point to large-scale warfare (see Ferguson, 1984), as villages unite defensively in the face of a commonly perceived outside enemy. Such alliances might be detected archaeologically in the form of joint regional defensive systems, with sites strategically placed to oversee primary access routes into the area. There may also be increased exchange of resources between allied villages, as a means of cementing the alliance relationships, and decreased trade between the alliance villages and outside, potential enemy villages. Alliance might also be marked stylistically by distinctive design motifs in the ceramics and other art forms. In addition to the indirect indicators of conflict seen in the defensive measures taken by villagers, conflict between neighboring settlements or regions should also have a direct manifestation in the archaeological remains. Arson or the deliberate burning of structures would be another indication of raiding and conflict. Although it may be difficult to detect deliberate burning in the archaeological record, tribal warfare should result in increased burning of sites, in general, and differential burning of storage facilities, since one of the goals of raiding is destruction of the enemy's resource base (see Zier, 1976). Attacks aimed at destroying the goods of others would be reflected in a pattern of burned storage rooms as opposed to living rooms (where fires commonly occurred as accidents). Some evidence of combat would be un- avoidable, such as a significant number of broken forearms ("parry fractures") among the population, skull fractures, marks of scalping, and the taking of trophy heads. The frequency of deaths of young adult males, between the ages of roughly 18 and 35, should be higher in the presence of tribal warfare; however, warriors are also often accorded special burial and thus may actually appear in lower frequencies in the regular community burial population. Warriors may be given separate cemeteries (e.g., Arlington Cemetery), or they may be accorded special consecration distinct from the rest of the population (e.g., the Plains practice of placing the warrior dead in trees). Warriors might also be buried with an increased frequency of weaponry. Finally, the initiation of warfare should result in an increase in weaponry generally. While weapons used for hunting in tribal societies might just as well be used against one's enemy, there should be an overall increase in the frequency of weapons in times of war. Within the Kayenta area specifically, it is necessary to distinguish the possible archaeological manifestations of warfare from material consequences of the population responding to the stress and pressures imposed by the degrading environment in the 12th and 13th centuries. Thus, for example, increased and protected storage facilities cannot be seen as evidence of war- fare, since the people would have tried to store more food in the good years to offset the scarcity of bad years. They also would have made a greater effort to protect those food resources from damage from the environment and natural predators. At the same time, while increased storage and protection of food resources can be attributed to environmental pressures, deliberate 28 Stress and Warfare Among the Kayenta Anasazi placement of the storage facilities in militarily defensible positions cannot be attributed to those same pressures. Likewise, it can be expected that in times of heavy erosion of arable land, as was the situation in the 12th and 13th centuries, the people would have moved their villages away from even the smallest plots of arable soil. Again, however, when the people explicitly selected defensible locations for their villages, rather than simply moving to any of the myriad nonarable locations nearby, we must appeal to warfare, and not just the environment, for an adequate explanation. Warfare in the Southwest Before discussing the research into tribal formation in Long House Valley, it is helpful to review the evidence for conflict and war in the broader arena of the Anasazi region as a whole. Warfare is a frequently cited but seldom studied phenomenon in the prehistory of the American Southwest (see Wilcox & Haas, 1991, for an overview of evidence for warfare throughout the Southwest). Archaeologists have turned to warfare as one possible ex- planation of the abandonment of large portions of the northern Southwest at the end of the 13th century (Cordell, 1984), and recently warfare has been mentioned in the context of sporadic "defensive" sites that have been noted in parts of Utah, Colorado, New Mexico, and Arizona between a.d. 1200 and 1300 (Dean et al., 1985; Gumerman, 1984; Mackey & Green, 1979; Lightfoot, 1984; Martin & Plog, 1973; see also Danson, 1957; Farmer, 1957; Woodbury, 1959). Despite frequent references to warfare and defense in the literature, little research has been directed at systematically examining the origins and nature of prehistoric warfare in the archaeological record of the Southwest. Gumerman (1984:107) summarizes the general state of knowledge about warfare in the Southwest in addressing possible causes of the abandonment of the Colorado Plateau: Intuitively, these "defensive" sites, along with some burned sites and very rare evidence of cannibalism, suggest that warfare was an important factor in the population movement [abandonment]. The warfare hy- pothesis, however, has not been tested rigorously. The Anasazi always utilized cliff shelters and seldom is there evidence in burial populations of violent, man-inflicted death. . . . There is also the possibility of inter- necine warfare. . . . However, little archaeological evidence exists for internecine warfare. Proponents of this hypothesis rightly ask what kind of evidence would one expect; the hypothesis is not easily testable in archaeological context because a different, potentially hostile culture would not be identifiable in the archaeological record. In the Kayenta area in particular, most scholars in the past have recognized defensive attributes at some sites and at least consider the possibility that there may have been warfare in the region in the 13th century (see Dean, 1969, 1986; Lindsay, 1969; Ambler, 1985; Ambler & Sutton, 1989; Ryan, 1977). Background and Methodology 29 But while there is an awareness that the Kayenta region along with the rest of the northern Southwest may have witnessed some conflict in the 13th century, the nature and intensity of the conflict are not well documented and are poorly understood. There is also no recognition of the potential importance of warfare in population movements, settlement patterns, and eventual abandonment of the entire Anasazi region. Jeffrey Dean, for example, one of the leading scholars of the Kayenta Anasazi, has noted the defensive possibilities of many of the cliff dwellings in Tsegi Canyon: "Some attributes of cliff dwellings suggest that they may have served defensive purposes, either as fortified villages or as refuges in times of strife" (1986:14). Yet he also dismisses warfare as a significant variable in any of the major events of the 13th century in the Kayenta region. Spe- cifically, Dean has argued that warfare did not play a causal role in the aggregation of the population and the appearance of much larger pueblos in the 13th century; in the large-scale movement of villages from the canyon floors up into the protected cliff dwellings in the canyons; and in the aban- donment of the area in approximately a.d. 1300 (Dean, 1969, 1986; Dean et al., 1978). He notes, "evidence from the Tsegi permits rejection of disease, the so-called Great Drought of 1276 to 1299, and conflict with enemy groups as primary factors in the final Kayenta Anasazi withdrawal from the area" (1969:195). Instead, Dean suggests, "[t]he most important consideration here is the interaction of the culture of the people actually living in the canyon with specific environmental factors" (1969:195). The dismissal of warfare and the appeal to strictly environmental variables are logical outgrowths of the history of archaeological research in the South- west. Since the 1940s, when tree-ring dating was developed for the region, archaeologists have relied on the climatological and chronological data that tree-ring analyses produce in interpreting survey and excavation data (Dean, 1969; Dean & Robinson, 1977). For the Kayenta /Tsegi Canyon area, for ex- ample. Dean (1967, 1969) conducted exhaustive tree-ring dating of cliff dwell- ings and sites in Long House Valley. Studies indicate that severe arroyo cutting and drought occurred during the 13th century, a pattern that has been confirmed elsewhere in the northern Southwest at that time (Dean et al., 1985). With such a firm and conclusive data base on the past environment of the ancient Anasazi, it has not been unreasonable to correlate environ- mental changes with observable cultural changes and to posit a causal re- lationship between them. However, the data obtained from the present research project demonstrate that warfare was much more prevalent than had been previously recognized and point to a more complex causal rela- tionship between the environment and the observed cultural changes. Defense in Long House Valley In the course of this project, the five Long House Valley focal sites recorded in the survey of Dean et al. were Long House Ruin, Organ Rock Ruin, Tower 30 Stress and Warfare Among the Kayenta Anasazi Figure 2-6. Looking northeast from Fireside toward Tower House. Arrow points to a notch in the talus slope between the two sites. House, Fireside Ruin, and Otherside Ruin. Names of ruins used in this report are those either given to the sites in the course of this project or assigned by previous investigators in the area. In some cases, published site names — for example. Long House or Organ Rock — have been used elsewhere to designate other sites. Rather than give such sites new names, however, we have kept the names that have already been established in the literature for the Kayenta sites. In assigning new names for sites as part of this project, we have tried to avoid the use of names already in use for sites in other areas. We have also tried to use names that may be descriptive of a distinctive site feature but that are not loaded with explicit or implicit connotations. In many cases we have followed in the time-honored Southwest tradition of mild humor, as in site names such as "Heltagito" (Breternitz, 1959). Aside from their distinctive internal features, these focal sites represent the clearest sign of defense in the valley. All five sites are situated on de- fensible hilltops or prominences, and each is located so as to be able to see all of the other four. In one case, where a talus slope of Black Mesa obstructed the natural view between two focal sites, a deep notch was cut in the slope to create a line of visibility between them (figs. 2-6, 2-7). Together, the five focal sites hold commanding views of all access routes into the valley: Organ Rock Ruin overlooks Marsh Pass, and Fireside Ruin and Otherside Ruin both stand watch over the south flank of the valley leading out to Klethla Valley. Background and Methodology 31 Figure 2-7. Line-of-sight connection between Fireside and Tow- er House (not to scale). Long House Ruin (LHV108), the largest focal site, includes over 400 rooms spread over one-half hectare of rocky slope above the valley floor (fig. 2-8). Surface remains at Long House reveal that access into the village could be gained only by way of two access corridors on either side of the site, and both of these were blocked prehistorically by cross walls. Rooms were ar- ranged in tiers around a two-story-high "spinal" roomblock, the "pueblo of orientation" (see Lindsay, 1969), which traverses the crest of the hill at the northern edge of the site. There is a dam and reservoir on the east edge of the site that catches runoff from the sandstone above. Due to the porosity of the local sandstone, it is unlikely that this reservoir could have held water permanently. The nearest source of permanent water was probably a spring in the bedrock behind the site. The pueblo of orientation at Long House is approximately 30 m in length, and it had few doors and no interior cross walls (fig. 2-9). While its function is unclear at this time, communal storage of food resources is the most likely 32 Stress and Warfare Among the Kayenta Anasazi Figure 2-8. Aerial photograph of Long House Pueblo, facing northeast. (Cour- tesy of the School of American Research, Santa Fe, New Mexico: 1993. Pho- tograph by Paul Logsdon.) Figure 2-9. Aerial photograph looking down on "pueblo of orientation" at Long House Pueblo. (Courtesy of the School of American Research, Santa Fe, New Mexico: 1993. Photograph by Paul Logsdon.) Background and Methodology 33 Figure 2-10. Aerial photograph of the isolated mesa upon which Organ Rock Ruin is located. This view shows the east cliff of the mesa, which was the source of the "Organ Rock" name given to the site by the Rainbow Bridge-Monument Valley expedition. (Cour- tesy of the School of American Research, Santa Fe, New Mexico: 1993. Photograph by Paul Logsdon.) explanation for such a large and open structure. The size of the building itself also indicates communal construction, and it could have served for storage at least for focal site residents if not for inhabitants of the entire Long House site cluster, which includes five additional satellite sites. Another focal site. Organ Rock Ruin (LHV404) overlooking Marsh Pass, 34 Stress and Warfare Among the Kayenta Anasazi Figure 2-11. View facing south toward the back side of Tower House. The pueblo of orientation is on top of the rock outcrop. is perched in a spectacular position on top of an isolated 200-m-high sand- stone cliff (fig. 2-10). The only means of access to this site (and its satellites) is by way of a steep talus slope followed by a vertical 15-m climb up steps and hand- and toeholds through a narrow crack in the cliff edge. There is no water on top of the cliff (though site residents did build at least one reservoir to catch runoff from the bedrock) and negligible arable land. The nearest source of water is Laguna Creek, which is 200 m down and several hundred meters to the east. There is arable land surrounding the base of the cliff and more on the alluvial terraces of the canyon bottom. Roomblocks were constructed in several locations across the top of the cliff. There is an L-shaped block of rooms by the reservoir composed of about 75 rooms, and at least three other clusters of rooms were recorded, including a pueblo of orientation with double-faced walls, trash areas, and a group of mealing bins. Entry to the site was restricted to a single narrow and steep passageway, making the Organ Rock Ruin extremely inaccessible and highly defensible even in the absence of any recorded constructed defenses. Tower House (LHV14) is located on a knoll with steep-sided arroyos on either side (fig. 2-4). Pecked steps and grooves in the stone face of the knoll provided access to the rear of the site, though there is relatively open access to the front of the site today. It is likely, however, that the front three sides of the village were blocked prehistorically by the construction of walls. On the highest part of the knoll stands the pueblo of orientation with wall remnants over 2 m tall, and, as at Long House, this room is long and unbroken Background and Methodology 35 by cross walls. It is on the northern side of the site, where it was built on the edge of a sheer-sided rock outcrop. The entire knoll upon which the village was built is bedrock, and several room floors and kivas were cut into the rock itself. With room walls blocking the more easily accessible south side and with a sharp cliff edge on the north side. Tower House, like Long House and Organ Rock, was in a defensible position. As mentioned, the line of sight from Tower House to Fireside was artifi- cially created prehistorically by means of a notch cut into the obstructing talus slope. Through this "V" it was possible to see back and forth between the highest points of these two focal sites. The overt effort to establish visibility indicates that visual contact or communication by signaling be- tween the focal sites in the valley was of great importance to the residents and points to cooperation and alliance among the people of the valley. Fireside (LHV137) is located on the summit of a steep-sided coUuvial ridge extending perpendicularly away from the escarpment of Black Mesa. The site consists of a long pueblo of orientation along the north edge of the village, with several additional roomblocks enclosing small plazas (and pos- sibly kivas). This is the only focal site where a reservoir has not been located, and yet there is no natural source of water within 2-3 km of the village. The elevated position of the hilltop makes this site defensible from the valley floor below, and there is a large roomblock obstructing access from the ridge extending off Black Mesa to the east. The site is also strategically located at the southern end of the occupied portion of the valley (during the Tsegi Phase), with an open view for several kilometers to the south toward the hills separating Long House Valley from Klethla Valley. Directly across the valley floor from Fireside is Otherside (LHV159), a large L-shaped pueblo on a hilltop on the northwest flank of the valley. Otherside also has a pueblo of orientation on its north side, with a large central plaza, a number of roomblocks, and a heavy scatter of trash downhill from the pueblo of orientation. The site is bordered by arroyos on both the north and south and by a steep drop-off on the east side. The reservoir is in the arroyo to the north. Though there are no specific defensive features, the hilltop position of the village places it in a defensible position, and there is good visibility to the other Long House Valley focal sites from this location. As with Fireside, Otherside stands watch over the southern access route into Long House Valley from the direction of Klethla. Aside from their individual characteristics, the five focal sites together point to a commonly perceived concern with defense by the residents of Long House Valley. The sites' inhabitants could all communicate back and forth with one another, and they jointly held commanding views of all primary access routes into the valley. The sites are also all placed in topo- graphically defensible positions, and defensive features as well as visibility appear to have taken precedence over access to both water and arable land. The residents of Long House Valley do seem to have been concerned about raiding or other forms of attack from outside the valley. But on the basis of 36 Stress and Warfare Among the Kayenta Anasazi the sites recorded by Dean and his colleagues, several questions remained unanswered. It was unclear who the people of Long House Valley were afraid of, and whether this was a very localized and idiosyncratic incidence of people defending themselves against real or perceived enemies or part of a much larger pattern of regional conflict and war. 3 EXCAVATION IN LONG HOUSE VALLEY To begin answering questions about the nature of political interaction in Long House Valley during the Tsegi Phase (a.d. 1250-1300), excavations were initiated at a number of sites. Two satellite sites were targeted for excavation, LHV72 and LHV73, both in the cluster associated with the Tower House focal site (fig. 3-1). The purpose of these excavations was twofold: to determine whether or not these two sites were functionally similar to one another in terms of the range of activities carried out by the occupants, and to determine possible patterns of interaction between the two adjacent com- munities. Complementing the intensive excavations at these two sites, testing was carried out at two focal sites. Long House and Tower House, and at two additional satellite sites associated with the Long House cluster. These test excavations were designed to provide information about similarities and differences between focal sites and satellite sites, and between satellite sites from different clusters. Finally, the excavations within the valley provide a comparative base for assessing similarities and differences between the Long House Valley occupation and the occupation of adjacent valleys. This inter- valley comparison allows us to make concrete inferences about cultural boundaries around the Long House Valley polity and the nature of interaction between Long House Valley and other areas. The two sites chosen for excavation, LHV72, named the Brown Star Site (NA10,829), and LHV73, named the Potential Site (NA10,830), both satellite communities of the Tower House cluster (Effland, 1979), are located at the far western edge of the cluster. They are both small, with four rooms and a kiva at LHV72 and eight rooms and two kivas at LHV73. They are located only 50 m from each other, and on the basis of surface ceramics, later confirmed by dendrochronological dates, were both occupied during the early part of the Tsegi Phase (ca. A.D. 1243-1280) (see tables 3-1, 3-2). The two sites are then representative of one component in the Long House Valley socio- 37 38 Stress and Warfare Among the Kayenta Anasazi Organ Rock Ruin HABITATION AND FOCAL SITES OCCUPIED BETWEEN AD 1250 AND 1300 O HABITATION SITES "^ FOCAL SITES LIMIT OF SURVEY Figure 3-1. Map of the Tsegi Phase occupation (a.d. 1250-1300) showing the location of sites worked on in the course of the project. political system, the satellite site, and provide a good sample for investigating patterns of interaction and variability within the valley. Excavations were conducted to reveal similarities and differences between the two contem- poraneous sites in terms of ceramics, kiva characteristics, storage patterns, and the range of activities carried out at each. Details on the excavations at LHV72 and LHV73 are contained in Appendix A. Excavation in Long House Valley 39 TABLE 3-1. Tree Ring Dates from LHV72 Provenience Date Structure 1, Layer 2, ash lens 1228fp-1272vv Structure 1, SWV^, Layer 2 1193fp-1243v Structure 1, SWy2, Layer 2 1193fp-1243B Structure 1, SWV2, Layer 2 1204fp-1243vv Structure 3, NEV2, Level V 1193fp-1256vv Structure 3, SVJ'A, Layer 4 1178fp-1243rB Structure 3, fill of mortar bowl in floor 1008fp-1136vv Fill of mortar bowl to south of Structure 2 1 1 92f p- 1 237 vv Summary of LHV72 LHV72 (figs. 3-2, 3-3) w^as a small habitation site occupied during the Tsegi Phase in Long House Valley. Based on tree-ring dates, the site was probably constructed in the 1240s and occupied at least into the 1270s. Activities carried out at the site included food preparation, indicated by the mealing complex, and ceremonial activities in the kiva. Periodic remodeling and reflooring efforts were carried out, resulting in the conversion of a room with a jacal wall into an all-masonry room. No evidence of burning was uncovered during the investigation of do- mestic rooms. The burned areas found in rooms were evidence of fires built within the room, not of conflagration. Though few in number, the burials TABLE 3-2. Tree Ring Dates from LHV73 Provenience Date Structure 7, Structure 7, Structure 7, Structure 7, Structure 7, Structure 9, Structure 9, Structure 9, Structure 9, Plaza, Wi/i, Plaza, WV2, Plaza, WV2, upper hearth upper hearth upper hearth upper hearth upper hearth NWi^, Level VI hearth hearth SEV2, Layer 2 Level IV Level IV Level IV lllOfp- 1177fp- 1095fp- 1070fp- 1125fp- 1136fp- 1213fp- 1187fp- 1199fp- 832fp- 1129fp- 802fp- 1208VV 1253VV 1173VV 1160VV 1193VV 1245VV 1271VV 1257VV 1248w 921vv 1206VV 948vv 40 Stress and Warfare Among the Kayenta Anasazi Excavation in Long House Valley 41 Figure 3-3. The general environmental location of LHV72. recovered from the site point to differential treatment of male and female individuals. An older adult female was interred in a burial pit w^ith numerous grave objects, while a male was interred with cranium missing, in a contorted position, and with a projectile point in the vicinity of the pelvis. The projectile (fig. 3-4) point is not characteristic of the ^ ° ■ Figure 3-4. Projectile point found with Burial 2 at LHV72. Summary of LHV73 On the surface, LHV73 (figs. 3-5, 3-6) looked like an example of a small Tsegi Phase pueblo, C-shaped, with eight rooms that were divided into two living units of four rooms each. Excavation of the site showed that this interpretation was substantially incorrect in the physical layout of the site. The C-shaped portion of the site includes seven rooms, one of which. Structure (S.) 9/10, is double the size of rooms at this site and elsewhere in the valley. An eighth room set apart from the roomblock, which was thought to be a rectangular kiva, con- tained the remains of a mealing bin com- 42 Stress and Warfare Among the Kayenta Anasazi 3000"^ O Figure 3-5. Schematic map of LHV73, the Potential Site. Figure 3-6. The general environmental location of LHV73. Excavation in Long House Valley 43 TABLE 3-3. Comparison of Ceramics from LHV72 and LHV73 Plainware Whiteware Orangeware Site Obs. Exp. Obs. Exp. Obs. Exp. Total LHV72 LHV73 Totals 1,457 1,380 6,054 6,131 7,511 353 396 1,802 1,759 2,155 580 614 2,765 2,731 3,345 13,011 Chi-square = 13.29. df = 2. P = 0.005-0.001. Obs. = observed, Exp. = expected. plex and was probably used as a work area for grinding corn. An oval kiva turned out to be two oval kivas, one partially superimposed over the other. However, the function of the rooms at the site does correspond to use by two clusters of people, either two families or an extended family. There are two living rooms, S. 7 and S. 9/10, and the other rooms, S. 3, S. 4, S. 5, S. 6, and S. 8, appear to have been storerooms, as they contained no interior features. The grinding complex in S. 1 could have been used cooperatively by all residents, as could the kiva. Intersite Comparisons One of the clearest contrasts between LHV72 and LHV73 is in the kivas at each site. The kiva at LHV72 is rectangular, while that at LHV73 is oval. The two sites are in close proximity to each other and might be assumed to have similar structures, but this is not the case. It is difficult to assess the full significance of the kiva shapes, but it can be inferred that there were some substantial differences in the ceremonial affiliations of the people in the neighboring communities. At the same time, the pattern of room construction and the general composition of the artifactual assemblages are quite similar. Specifically, the ceramic and nonceramic artifacts found at each site are typologically indistinguishable from each other. However, there are some significant differences in the relative frequencies of ceramic types at the two sites (table 3-3). LHV72 has more than expected plainwares and less than expected whitewares and orangewares, while LHV73 has the opposite pat- TABLE 3-4. Comparison of Ceramics from LHV72 and LHV73 by Percentage Site Plainware Whiteware Orangeware LHV72 LHV73 61% 57% 15% 17% 24% 26% 44 Stress and Warfare Among the Kayenta Anasazi TABLE 3-5. Ceramics at LHV96, LHV98, Long House, and Tower House Long Tower Ceramic LHV96 LHV98 House House Category (3 Pits) (2 Pits) (7 Pits) (5 Pits) Plainware Kiet Siel Gray 470 80 1,348 422 Moencopi Corrugated 75 131 321 429 Tusayan Corrugated 26 40 59 48 Unknown corrugated 0 5 8 1 Total 571 256 1,736 900 Whiteware Kayenta B/W 7 20 50 27 Tusayan B/W 45 55 200 99 Flagstaff B/W 0 2 2 2 Sosi B/W 7 7 25 4 Dogoszhi B/W 5 5 15 10 Black Mesa B/W 3 1 13 2 Kana'a B/W 1 1 1 2 Unidentified B/W 65 55 125 109 Undecorated whiteware 84 116 202 159 Total 217 262 633 414 Orangeware Kiet Siel Polychrome 7 3 33 8 Kayenta Polychrome 0 5 9 2 Tusayan Polychrome A 120 130 300 194 Tusayan Polychrome B 15 27 49 27 Unknown polychrome 5 6 9 6 Kiet Siel B/R 2 0 16 1 Tusayan B/R 36 26 136 56 Unidentified B/R 6 5 31 22 Tsegi B/O 3 3 21 7 Tsegi Orange 175 170 410 162 Orange with red slip 40 3 134 68 Total 409 406 1,148 553 Totals 1,197 924 3,517 1,867 tern. At the same time, it should be pointed out that when calculated as percentages of the whole assemblage, the distribution of the three wares at the sites is very similar (table 3-4). With a few exceptions, there are similar nonceramic artifactual assemblages at the two sites. The exceptional artifacts found at LHV72 include a stone "fire dog" and two small stone cylinders with no immediately apparent function. At LHV73, exceptional artifacts include two large floor polishers. Excavation in Long House Valley 45 TABLE 3-6. Comparison of Ceramics from Sites in Long House Valley Plainware Whiteware Orangeware Site Obs. Exp. % Obs. Exp. % Obs. Exp. % Total LHV72 LHV73 LHV96 LHV98 1,457 1,278 6,054 5,681 571 640 256 494 61 59 48 28 353 429 1,802 1,906 217 215 262 166 15 17 18 28 580 683 24 2,765 3,034 25 409 342 34 406 264 44 Long House 1,736 1,881 49 Tower House 900 999 48 633 631 18 1,148 1,005 33 Totals 10,974 414 335 22 3,681 553 533 30 5,861 20,516 Chi-square = 435.95. df = 10. P = <0.001. an axe, two abraders, a maul, two loomblocks, three sharpening stones for bone tools, a bone needle, and several miscellaneous ground stone objects of unknown function. The greater diversity found at LHV73 may well be due to the greater size of the collection, rather than to a greater diversity of activities at the site. However, the presence of two loomblocks and a bone needle at LHV73 does indicate cloth-working activities at this site that are not present at LHV72. Test Excavations at Other Sites In addition to the intensive excavations at LHV72 and LHV73, four other sites in Long House Valley were tested to provide comparative ceramic samples from focal sites and satellite sites associated with other clusters. Because LHV72 and LHV73 are part of the Tower House cluster. Tower House was one of the focal sites selected for sampling. Then, to balance the data from the Tower House cluster, additional testing was conducted in the Long House cluster. Long House Ruin itself was tested along with LHV96 and LHV98, two small adjacent satellites at the west side of the cluster. At each site, two to seven 1- x 1-m test pits were excavated in 20-cm artificial levels, and all material was screened through V4-in. mesh screen. The distri- bution of the ceramics is given in Table 3-5. A comparison of the distribution of the major ware categories at the four sites tested and the two sites excavated indicates that there were significant differences in almost all cases between the observed and expected frequencies (table 3-6). However, when relative frequencies of the wares are considered, some general patterns appear. There is confirmation that when compared to other sites in the valley, LHV72 and LHV73 share a similar pattern of the distribution of the different wares. Similarly, LHV96, Long House, and Tower House share a similar distribution pattern. In contrast, LHV98, with a much lower relative frequency of plainwares and higher frequencies of whitewares 46 Stress and Warfare Among the Kayenta Anasazi TABLE 3-7. Tsegi and Pre-Tsegi Phase Black-on-White (3/ W) Ceramics in Long House Valley LHV72 LHV73 LHV96 LHV98 Long House Tower House B/W Era No. % No. % No. % No. % No. % No. % Tsegi Phase B/W Pre-Tsegi Phase B/W 78 67 39 33 675 88 92 12 52 77 16 23 75 82 16 18 250 82 56 18 126 86 20 14 and orangewares, stands out as unique in the sample of sites tested in the valley. There is clearly no pattern of similarity demarcating the satellite sites in general, nor a pattern distinguishing the focal sites from the satellite sites. Based on the limited testing at LHV96 and LHV98, Long House and Tower House, we are unable to offer an explanation for the observed patterns of ceramic assemblages in these sites. Beyond the general patterns seen in the distribution of the various wares in the different sites, there is another pattern that should be noted in terms of the relative frequencies of specific whiteware types at the different sites. Whitewares in the Kayenta region are often used as the primary technique for dating sites on the basis of surface remains. Because each of the different types has been independently dated through dendrochronology, they can be used to assign specific absolute dates when found on the surface of a site. The excavated data from Long House Valley, however, require some mod- ification in the use of whitewares for dating. Two whiteware types in particular, Kayenta Black-on-white and Tusayan Black-on-white, are commonly used to distinguish sites dating to the Tsegi Phase, A.D. 1250-1300. Other Black-on-white types, such as Flagstaff, Sosi, or Dogoszhi, are used in dating earlier sites. However, the data from Long House Valley indicate that the earlier types are consistently found on sites that are known to date exclusively to the Tsegi Phase. The relative frequencies of Tsegi Phase types and pre-Tsegi Phase types (collectively) are given in Table 3-7. LHV72, for which there are good tree-ring dates of a.d. 1246-1272, actually has the highest percentage of pre-Tsegi Black-on-white pottery with 33.3%, while the other five sites have percentages ranging from 12.0 to 23.5. The implication of these data is that all of these sites, including the two focal sites of Long House and Tower House, were constructed and occupied exclusively in the Tsegi Phase. Thus, even though earlier ceramics are found with some frequency on these sites, they do not necessarily indicate earlier occupations. Conclusions Taken as a whole, the excavation data from Long House Valley provide only limited insights into the possible formation of tribal-type polities in the Excavation in Long House Valley 47 Kayenta heartland during the Tsegi Phase. Although some functional dif- ferences between focal sites and satellite sites could be determined from surface survey, further differences were not manifested in the artifactual assemblages obtained from very limited testing. 4 KAYENTA AND KLETHLA VALLEYS The settlement pattern data from Long House Valley, described in Chapter 2, provided the first indication of the emergence of a pattern of warfare in the 13th century in northeastern Arizona. The aggregation of the population into much larger villages and tight settlement clusters and the movement up onto hilltops and prominences indicated an overt concern for attack from the outside. To obtain a more comprehensive picture of the nature and development of conflict in the Kayenta heartland, additional research was initiated in neighboring areas. Since Long House Valley is in a unique position in the middle of a natural east-west corridor across northern Ari- zona, the defensive posture taken by the valley's residents may have been a highly localized response to their vulnerable geographic position. At the same time, we know that across the northern Southwest, the period between A.D. 1150 and 1300 was a time of decreased precipitation, increased erosion of arable land, a lowered water table, and cyclical droughts (Euler et al., 1979; Dean et al., 1985). If such environmental conditions did result in intergroup conflict, then Long House Valley would have been part of a much larger regional pattern of warfare, as the residents of the northern Southwest re- sponded to the common environmental pressures. Finally, since the villages and field systems in the Valley were much too close together for there to have been internal fighting and raiding among the Long House Valley res- idents, who were their enemies? Were they worried about an attack from their immediate neighbors, or was the threat coming from much farther away, outside the Kayenta region? To begin addressing these issues, new archaeological investigations outside Long House Valley to the east in Kayenta Valley and to the west in Klethla Valley were undertaken in 1984 and 1985 (fig. 4-1). Additional survey was conducted in 1986 in the Tsegi Canyon system to the north (Chapter 5). To the south. Black Mesa was abandoned by A.D. 1150 (Powell, 1983) and thus 49 50 Stress and Warfare Among the Kayenta Anasazi does not directly enter the picture of events occurring in the 1150-1300 period (though it is highly likely that some residents of Black Mesa would have moved into Long House Valley and neighboring valleys in the first half of the 12th century). The fieldwork in the open valleys on either side of Long House Valley was specifically designed to clarify the apparent pattern of Kayenta warfare in three ways: 1. Was the defensive pattern seen in Long House Valley unique, or was there instead a more widespread, regional pattern of defensive sites and warfare? 2. Were the Long House Valley residents concerned about attack from their immediate neighbors or about more distant enemies? 3. Were the Long House Valley residents a separate political entity with their own bounded alliance network, or were they part of a larger multivalley "tribal" alliance? The Kayenta and Klethla Valley surveys were explicitly designed to address these questions. If the Klethla and Kayenta valleys were found to lack the pattern of focal sites in defensible locations, then the Long House Valley defensive system could be interpreted as a unique response to the central location of Long House Valley in a trans-Plateau pathway. On the other hand, if a similar pattern was found in the neighboring valleys, then concern over defense occurred over a much wider area and warfare can be inferred to have been occurring at a regional rather than local level. Likewise, if the line-of-sight visual network in Long House Valley was found to be linked to similar networks in the neighboring valleys, then it could be argued that the Long House Valley residents were part of a larger alliance system extending beyond the geographic boundaries of the one small valley. A link between the valleys would also indicate that the conflict was probably not occurring between neighboring communities; rather, the people of adjacent valleys joined in communication would have been concerned with common enemies in out- side areas. Conversely, warfare between valleys — internecine Kayenta war- fare— should have been marked by a clear break in the defensive networks of neighboring valleys. Geographic Information Systems and Archaeological Survey To maximize the effectiveness of fieldwork, in terms of both cost and time, we turned to computer technology to select a sample of each valley for survey. Using a Micro-Terrain Information System, a geographic information system developed to aid archaeologists in locating specific environmental features (Kvamme, 1983), we were able to produce a computer-generated map of the research area that pinpointed potentially defensible locations and Kayenta and Klethla Valleys 51 Long House Valley Upper Klethla Valley Figure 4-1. Outline of total survey project area. indicated intervisibility between such locations (see Appendix B for a detailed discussion of the program developed). The topographic characteristics of the known defensible hilltop sites in Long House Valley were used as a baseline, and all hilltops and elevated areas in the Klethla and Kayenta Valley survey areas sharing those characteristics of relief and limited access were compiled on a set of maps. The computer was also used to generate a cross-list of intervisibility between each of these defensible locations. (In laymen's terms, the computer put itself, figuratively, on top of each hill and then listed what other hills could be seen.) This intervisibility list included the possible points of visibility between each of the neighboring valleys, so we could determine possible patterns of intervisibility between Long House, Kayenta, and Klethla. The hilltops and elevated areas, designated high probability target areas (HPTAs), were transferred onto U.S. Geological Survey topographic maps for use in the field. A total of 627 HPTAs in the Klethla and Kayenta valleys were marked by the computer. Of these, 233 were ultimately surveyed. Included within this sample were all HPTAs that allowed visual contact between different valleys. In addition to the HPTAs, a sample of nondefen- sible areas — essentially the open flats on the valley floors — were also targeted for survey. Each of the sample areas, designated low probability target areas (LPT As), was 1 sq km. Eleven LPT As, five in Klethla Valley and six in Kayenta Valley, were surveyed. These LPTAs were surveyed as a cross-check and control on the HPTAs and to locate possible "satellite"-type communities. 52 Stress and Warfare Among the Kayenta Anasazi Figure 4-la. Southwestern sector of survey project area showing the high and low probability target areas surveyed and the sites discovered and re- corded. * FOCAL SITES • OTHER TSEGI SITES A ALL OTHER SITES U UPPER KLETHLA VALLEY L LAGUNA CREEK DRAINAGE LV LONG HOUSE VALLEY HL HIGH PROBABILITY TARGET AREA (LAGUNA CREEK DRAINAGE) LL LOW PROBABILITY TARGET AREA (LAGUNA CREEK DRAINAGE) HU HIGH PROBABILITY TARGET AREA (UPPER KLETHLA VALLEY) LU LOW PROBABILITY TARGET AREA (UPPER KLETHLA VALLEY) Kayenta and Klethla Valleys 53 so I KM Figure 4-lb. Central sector of survey project area. While the overall number of LPT As is low, the actual acreage covered by survey compared to that of HPTAs is not dissimilar (figs. 4-la-c). Kayenta Valley Kayenta Valley rises from about 1750 m in elevation at the town of Kayenta to 1850 m at the base of Skeleton Mesa (fig. 4-2). The research area investigated in 1985 extended from Marsh Pass at the southwest corner of the valley to the town of Kayenta, a distance of about 20 km. Using the computer tech- niques discussed above, an area of over 300 sq km was targeted for sample 54 Stress and Warfare Among the Kayenta Anasazi :^ /)r'''l il'., aL-TO *L.9 %1 -OQ.,/- ,.,^..s ^^ ^ Figure 4-lc. Northeastern sector of survey project area. surveys in 1985. Survey teams investigated the escarpments of Skeleton, Tyende, and Black mesas that border the valley on the west, north, and south, respectively. Hilltops and prominent points across the valley floor were also selected by the computer program for survey. Overall vegetation and climate are similar in this area to Klethla and Long House valleys. Goals for the Kayenta Valley were the same as those for the Klethla survey: to see whether or not the settlement pattern change noted for the Tsegi Phase in Long House Valley extended east through Kayenta Valley out toward Monument Valley. If Kayenta Valley was found to have a similar kind of defensive settlement pattern, it would be important to know if Long House Valley focal sites could be seen from focal sites in Kayenta Valley and if a visibility network existed among focal sites in Kayenta Valley. A total of 149 HPT As were surveyed, as were 6 LPT As, and 175 sites were recorded. Ninety-seven of these dated on the basis of surface ceramics to the Kayenta and Klethla Valleys 55 Figure 4-2. Aerial photograph of Kayenta Valley, facing northeast. (Cour- tesy of the School of American Research, Santa Fe, New Mexico: 1993. Photograph by Paul Logsdon.) periods prior to a.d. 1100, and 110 sites, over half those recorded, dated to between a.d. 1100 and 1300. The other sites ranged from Archaic through Pueblo II. Tsegi Phase sites included 40 sites or site components and ranged in size from trash scatters to 150 or more rooms. Ten focal sites were located. These are all on hilltops or other prominent isolated features and possess a range of defensible and defensive features. In general, the characteristics of these ten sites are similar to the focal sites of Long House Valley, some having open plaza arrangements, pueblos of orientation with cored, double-faced masonry, with reservoirs, walled en- trances, and perimeter walls. All of the focal sites in Kayenta Valley were also located in defensible locations. All were located on HPT As, on geograph- ical features identified by the computer program as analogous to the focal sites in Long House Valley. Of the other 31 smaller, satellite-type sites dating to the Tsegi Phase, 8 were in defensible locations, and 23 were in nonde- fensible locations, usually close to water sources or on cultivable land. Sixteen of those nondefensible sites were within 300 m of a focal site identified by the survey, forming an overall settlement pattern similar in all respects to that found in Long House Valley. Two of the focal sites recorded in 1985 were located on isolated hills out on the valley floor. The site of Moqui Rock (LCD- 169) is on an isolated sandstone knob south of Tyende Mesa and west of the town of Kayenta at the confluence of Laguna and Parrish creeks (fig. 4-3). Known locally since 56 Stress and Warfare Among the Kayenta Anasazi Figure 4-3. Moqui Rock, facing west. Ruin covers virtually all the upper sur- faces of the rock outcrop. early this century, the site has been largely destroyed by visitors. What remains is the heavily rubble-strew^n remains of a very large pueblo with reservoirs, a rock-cut kiva, and a number of room outlines and viga holes carved into the rock of the hill itself. The settlement was perched on the top of the rock outcrop and extended partway down the sides. A double-faced stone wall, part of the pueblo of orientation, was located high on the south side of the site. Access to the site was by way of a narrow crack in the rock marked with hand- and toeholds on the north side of the site. It is possible today to climb directly into the south side of the site, but, because of large- scale deterioration of the architectural remains, it is impossible to determine whether this side of the pueblo was walled or open aboriginally. The highly restricted route on the north side, however, would indicate that access into the pueblo was probably limited in all directions. The concentration of struc- tures well above the surrounding farmland and above the lowest slopes of the knob suggests that the site was defensive. There are large areas of clear bedrock down below the outcrop that certainly would have been habitable prehistorically, and an occupation here would not have used up any valuable arable land. If the inhabitants merely wished to be off land suitable for cultivation, we might expect the bulk of rooms to have been built on the low stone slopes of Moqui Rock rather than on the upper portions of the steep-sided outcrop. From Moqui Rock, visibility is excellent to most of the other focal sites. Of the other focal sites in the valley, only Six Foot Ruin is Kayenta and Klethla Valleys 57 LCD 93 MOQUI OVERLOOK ^ITE LOCATION ••;.♦.' ARCHITECTURE -* PRINCIPAL ACCESS ROUTE \'\- •:\ TRASH SCArrER/MIOOEN I I BEDROCK \miM RUBBLE (SCHEMATIC-NOT TO SCALE) Figure 4-4. Map of Moqui Overlook. not visible from Moqui Rock. (As noted below. Six Foot Ruin is located in the mouth of Tsegi Canyon and is not part of Kayenta Valley proper.) Moqui Overlook (LCD-93) is the other site located on the valley floor (fig. 4-4). This site is a completely walled 40-50-room pueblo on a low hill across Laguna Creek from Moqui Rock. The two sites are separated by about 1 km. The pueblo itself is U-shaped, with a perimeter wall around the exterior enclosing the rooms and a small plaza area. A kiva depression is located just outside the wall. The main block of the pueblo has a double-faced wall along the north side. From this location there is clear visibility to six of the other Kayenta Valley focal sites. This site differs from the others in that the location is exposed and the low sides of the hill are not particularly difficult to climb; however, it offers good visibility of wide stretches of arable land below, and it was culturally protected by the construction of a perimeter wall. Thus, Moqui Overlook lacked some of the naturally defensible attributes of the other sites, but the residents took overt steps to make the village deliberately defensive. Four focal sites are located at the base of the escarpments that border Kayenta Valley on three sides. Kinpo (LCD-50) includes 100-125 rooms with double walls along the roomblocks at the west edge of the site (fig. 4-5). The site was constructed on a sandstone outcrop just above the valley bottom- lands. A wash leading into Laguna Creek runs along two sides of the site. 58 Stress and Warfare Among the Kayenta Anasazi LCD 50 KINPO J*,'. . ARCHITECTURE ■*■ PRINCIPAL ACCESS ROUTE TRASH SCATTER/MIDDEN BEDROCK WmM RUBBLE (SCHEMATIC-NOT TO SCALE) Figure 4-5. Map of Kinpo. and the creek itself is within 200 m. Access to the site is from the northeast, as the steep sides of the outcrop and the walls of the creek limit access on the east and south. The west side of the site was walled, and the highest portion of the site may have had a perimeter wall as well. No reservoir was located in the vicinity of this site, though bedrock pools are common in this Kayenta and Klethla Valleys 59 Figure 4-6. Map of Tachini Point. part of the valley and could have been employed as water catchment basins. Visibility from this site was clear to five of the other Kayenta Valley sites. Tachini Point (LCD-174; also known as Gnat Hill) is a site similar to Moqui Overlook. Located on a spur extending out from Black Mesa (fig. 4-6), it is 60 Stress and Warfare Among the Kayenta Anasazi 'i '.'.-! ARCHITECTURE ^. PRINCIPAL ACCESS ROUTE V^:}^::i^\ TRASH SCATTER/MIDDEN tggggg} RUBBLE (SCHEMATIC-NOT TO SCALE) Figure 4-7. Map of the Parrish Creek Site. composed of roomblocks in a U-shape around a small plaza area and may originally have been rectangular, completely enclosing the plaza. The exterior of the site appears to have been walled, with an entrance on the south or southwest side. From Tachini Point, visibility extends to most of the other focal sites and, significantly, back to the Organ Rock Ruin in Long House Valley. This is particularly important because Tachini Point links Long House Valley with Kayenta Valley through visibility between these two sites. In addition to its pivotal position in intersite visibility, Tachini Point's defensive features are primarily its location on a steep-sided, narrow ridge and the perimeter wall. Unfortunately, construction of a modern highway across the southern side of the ridge upon which the site sits prevents us from ascer- taining whether or not there was a wall limiting access to the site area. Byron Cummings is reported to have excavated at Tachini Point (Jeffrey Dean, pers. comm.), but there is no record of these excavations (Turner, 1962). The Parrish Creek Site (LCD- 166) is located on a sandstone outcrop along the southwest edge of Tyende Mesa (fig. 4-7). There are 60 to 100 rooms at this location, with the pueblo of orientation at the site's highest point. Most of the rooms are arranged in tiers across a sloping sand pocket within this irregularly shaped rock formation. The site appears to have been facing Kayenta and Klethla Valleys 61 Figure 4-8. Map of RB568. Parrish Creek, presently 200 m to the south across low dunes. Hand- and toeholds ascend a tall sandstone "chimney" at one side of the site. This chimney is a prominent feature of the landscape and can be easily seen from the surrounding areas. Another access route climbs from the creek's flood- plain at the east edge of the site to the roomblocks. This route is walled at its lowest point. Much of the site would have been hidden from view within the sand pocket; however, a small knob covered with rubble forms the highest part of the site and would have been easily visible from a distance. 62 Stress and Warfare Among the Kayenta Anasazi Figure 4-9. RB568, facing northeast. The entire outcrop is covered with remains of the Tsegi Phase ruin. Several room floors w^ere cut into the rock of the knob and a room was constructed on its highest point. Concealment could not have been of primary- concern to the site's occupants; however, a wall of large sandstone blocks restricted access to the site from the west, and another wall limited access from the east. Rock pools above the site on sandstone outcrops of Tyende Mesa may have been used for water catchment, although no clear remains of reservoir construction were recorded. It is likely that Parrish Creek would have provided a stable source of water for the site residents. Satellite sites of Parrish Creek are numerous and close by. There are three sites within 200 m of the north, south, and east edges of the site, each located on level sandy areas that would have been suitable for cultivation. Visibility from this site is limited to sightings up and down Parrish Creek, to Moqui Rock and to RB568. The site of RB568 was first located and partially excavated by the Rainbow Bridge-Monument Valley Expedition (RBMV) between 1935 and 1938 (fig. 4-8) (Beals et al., 1945). The site had over 100 rooms arranged over a level area southwest of Parrish Creek. The roomblocks were constructed out from a sandstone knob, which was covered with architecture when the site was occupied. Floors were pecked into the surface of the knob from bottom to top, as was at least one kiva (fig. 4-9). Burials held the primary interest of the RBMV excavators, and 50 inter- ments were recovered. Recent analysis of this material revealed patterns of Kayenta and Klethla Valleys 63 particular interest to the present project. There was a disproportionately low number of adult males among the burial population from RB568. There were half as many males (four) as females (nine) in the group 35-50+ years, and there were no males identified in the 18-35 age group (seven females were identified) (Crotty, 1983:28). This suggests that males were given alternative burial treatment, a pattern that would be consistent for a culture at war in which warriors would be buried apart from the rest of the population. Focusing as they did on burials, the RBMV investigators did not note the defensive features present at RB568. George Brainerd, the principal excavator, noted: Site RB 568 was selected [for excavation] for a number of reasons. It appeared to be a large and rich site, and its extensive architectural remains were in the open, a fact that distinguished it from the much better-known cave ruins. (Beals et al., 1945:72) Some note was made of the extent of the rooms on the knob at the northeast corner of the site, and though no direct mention of defense was made, the characteristics are consistent with defensibility: . . . the bluff to the east showed unusual features characteristic of at least three sites of this immediate area and time period. There was evidence of ten niches cut into the steep sandstone slope of the bluff. . . . Footholds led up to the ends of some of them. . . . Two of them had groups of "loom holes" at either end. It is suggested that these niches are the remains of rooms perched on the cliffside, supported in front by a stiltlike wall. . . . Such rooms occur elsewhere in cave dwellings dated shortly after this site. . . . There were also several surface rooms on the blurf, including two on the summit, which rises fifty feet above the main ground level of the village and one hundred feet above the floor of the neighboring canyon. These two rooms may well have served as a lookout. It is hard to guess why the inhabitants of the village built rooms in such inconvenient and windy locations. (Beals et al., 1945:82) Overall, it appears that RB568 was a focal site related to others in the vicinity. Pueblo of orientation double-faced masonry was part of a roomblock con- structed toward the west side of the site, possibly RBMV's Group III. Though no evidence of a reservoir was located, reservoirs have been reported (Dean, pers. comm.), and the site is very near Parrish Creek, which flows most of the year. The site's location is secluded, like the Parrish Creek site. Although blown sand prevented a determination of whether or not there may have been a defensive wall protecting all or part of the site, the steep sides of this natural outcrop would have made it virtually impregnable to attack from below. From RB568 there is a clear line of sight to the nearest focal sites. 64 Stress and Warfare Among the Kayenta Anasazi '^^§h^'^ Figure 4-10. Aerial photograph, facing northwest, of the mesa top upon which Table Top Ruin is located. (Courtesy of the School of American Research, Santa Fe, New Mexico: 1993. Photograph by Paul Logsdon.) Parrish Creek and Moqui Rock, and to four of the other seven Kayenta Valley focal sites. Table Top Ruin (LCD-98) is located on an isolated narrow mesa top adjacent to the escarpment of Black Mesa (fig. 4-10). The site is on what is probably the most defensible location anywhere in the valley. The extremely steep sides of the mesa rise 165 m off the valley floor. Once these sides are scaled, a visitor is faced with a 4-10-m-high cliff wall around the entire circumference of the mesa top. It took the project survey crew 2 days to find a way to negotiate this cliff face. They eventually did find a single route of access into the site, through a narrow crack in the face of the cliff beneath the mesa top. The site itself consists of 50 to 100 rooms in roomblocks clustered toward the center of the mesa top (fig. 4-11). A series of interconnected and dammed rock pools formed a reservoir with a controllable outlet over the side of the cliff. The pueblo of orientation is constructed in the center of the site, near a circular opening in the mesa top that was artificially enlarged and may have served as a kiva. The nearest land suitable for cultivation is at the base of the mesa, and the nearest source of permanent or stable water is several kilometers away. Visibility from the site is excellent, and six of the other Kayenta Valley focal sites can be seen. Two other focal sites within the Kayenta Valley proper. Rabbit Ears (LCD- 9) and Happy Valley (LCD-52), were discovered by the survey crews. Rabbit Kayenta and Klethla Valleys 65 3 LU o Q a. Q (0 03 S m ST. 3 o UJ i?! t: LU _i if 5 CO C3 I o o z I o oc n CI a. cr UJ < Q. 1- m 3 o H 3 66 Stress and Warfare Among the Kayenta Anasazi Figure 4-12. Map of Rabbit Ears Pueblo. Kayenta and Klethla Valleys 67 Figure 4-13. The monolithic "rabbit ears" around which Rabbit Ears Pueblo is clustered. The large upright stones are a distinctive landmark on the horizon and are visible from a great distance to the east and west. Ears, at the north edge of the research area, overlooks Monument Valley to the north and east and Tyende Mesa to the south (fig. 4-12). This site consists of two clusters of rooms, one set around the base of two 10-m-high sandstone monoliths that give the site its name, and the other out on the flat area of the mesa top. The mesa is easily defensible, in that the top is surrounded by sheer cliff walls. Access is limited to three narrow and very steep cracks, which today can be scaled only with the use of log ladders. The top of the mesa is relatively extensive, approximately 300 x 500 m, most of which is exposed bedrock. Along the south edge, three rock pools have been converted into reservoirs by the construction of stone walls across the low end. Sections of wall have been constructed along the edge of the mesa near the access routes, as well. This site can be recognized easily from a distance by the V-shaped "rabbit ears" monoliths (fig. 4-13). Visibility from the site into Monument Valley is excellent, and it is possible that Rabbit Ears served as a gateway or point of communication with groups to the north and east in Monument Valley, an area not yet well studied (see Neely & Olson, 1977). Rabbit Ears also can be seen from Moqui Rock, Moqui Overlook, and Table Top, and it is likely to have been the northernmost outpost of the Kayenta Valley system of focal sites. Happy Valley is located along the escarpment of Skeleton Mesa, 265 m above the floor of Kayenta Valley (fig. 4-14). Like Rabbit Ears, the site can 68 Stress and Warfare Among the Kayenta Anasazi Figure 4-14. The "flatirons" upon which Happy Valley Pueblo is located. Site is on the right-hand outcrop. (Courtesy of the School of American Research, Santa Fe, New Mexico: 1993. Photograph by Paul Logsdon.) be identified from a great distance by the unusual and distinctive rock formation on which it is located. Access to the site is strictly limited to one major route, which is walled in at least two places, crosses a steep bedrock slope via pecked hand- and toeholds, and skirts the edge of a 100-m cliff. Several smaller Tsegi Phase satellite sites are located along the slope leading to the main focal site area. Happy Valley, the largest site of this cluster, is located at the very highest point of the hilltop (fig. 4-15). Sheer cliffs drop 20-40 m around three sides of the site, broken only by narrow cracks. Three of these cracks provided the alternate access routes to the upper ruin. In each case, these cracks were nearly vertical and could be climbed only through the use of cut hand- and toeholds in the rock. A wall, 1-2 m in height at the top of each crack, further restricted access into the site. Rather than serving as primary access routes into the site — all three exit to the west, where there are only barren hills and the escarpment of Skeleton Mesa — these seem more to have been auxiliary exit routes away from the main site area. Roomblocks, some with at least two stories of rooms, were arranged in terraces in sandy pockets at the peak. The nearest cultivable land would have been at the base of the hill, 265 m below. Rock pools and a reservoir several hundred meters down the slope of the hill provided the nearest Kayenta and Klethla Valleys 69 L^^^ Figure 4-15. Artist's reconstruction of the Happy Valley community located around the rock outcrop. source of water. Visibility from this site extends into Monument Valley and to all but two of the other Kayenta Valley focal sites. The remaining focal site. Six Foot Ruin (LCD- 130), is not located in Kayenta Valley proper but in the mouth of Tsegi Canyon, where the canyon opens into Marsh Pass (fig. 4-16). The site is on a small mesa that projects from the escarpment of Skeleton Mesa, on the north side of Laguna Creek, directly across from Organ Rock Ruin. Access to the site requires climbing a long sloping hill, traversing around the base of a cliff on the edge of Skeleton Mesa, crossing a narrow saddle, and climbing through a narrow crack in the rock cliff. As with Rabbit Ears and Table Top, the mesa top of Six Foot Ruin is set off by a 4-10-m-high cliff around the circumference. The 60 rooms and kiva cover much of the top of the mesa, with some rooms built right to the edge of the cliff and the others dotting the rest of the mesa (fig. 4-17). It could not be determined whether or not the entire cliff edge was surrounded by a defensive wall, but there remain today sections of a wall built along at least parts of the cliff edge. Curiously, portions of this wall are constructed in areas where it would have been impossible to scale the cliff face without technical climbing gear. Preservation of structures at Six Foot was the best of any site visited, with sections of the pueblo of orientation standing 2 m high (thus the site name). A possible reservoir was located on the mesa top as well. The view from this site extends to the Organ Rock Ruin focal site 70 Stress and Warfare Among the Kayenta Anasazi Figure 4-16. Aerial photograph of the small mesa upon which Six Foot Ruin is located (foreground). Note its proximity to Organ Rock across the canyon bottom. (Courtesy of the School of American Research, Santa Fe, New Mexico: 1993. Photograph by Paul Logsdon.) in Long House Valley and over the south end of Tsegi and Wildcat canyons. Although none of the other focal sites in Kayenta Valley are visible from Six Foot, it is tied into the Long House /Kayenta system by its visual link with Organ Rock Ruin. Like Rabbit Ears, Six Foot may have served as a connector or point of communication by signaling with groups that lived in the canyons, tying them into a communication system that extended from Long House Valley and Kayenta Valley into the canyons and toward Mon- ument Valley. Subsequent to the Kayenta Valley survey, field crews discovered yet an- other major focal site even farther up into Tsegi Canyon beyond Six Foot (fig. 4-18). This site. Wildcat Canyon Ruin (LCD-176), directly links the out- side valley systems with the complex of cliff dwellings in the canyons. Wildcat Canyon Ruin has direct visual links with Six Foot and Organ Rock Ruin and, on the other side, looks directly down on Swallow's Nest Cave, the first of the cliff dwellings in the canyon. Wildcat Canyon Ruin is typically situated on top of a small mesa, rising 200 m up from the canyon bottom. Access again is by way of a narrow crack in the upper sheer cliff wall. Upon gaining the top of the mesa, one must walk across a narrow saddle to get to the site area. Prehistorically, two separate walls were constructed across this saddle to restrict access. The site itself consists of over 200 rooms, including a pueblo of orientation, spread over the north end of the mesa. There is no Kayenta and Klethla Valleys 71 Figure 4-17. Schematic map of Six Foot Ruin. 72 Stress and Warfare Among the Kayenta Anasazi Figure 4-18. Aerial photograph, facing north, showing the general location of the Wildcat Canyon Ruin mesa within Tsegi Canyon. (Courtesy of the School of American Research, Santa Fe, New Mexico: 1993. Photograph by Paul Logs- don.) reservoir associated with the site, though it is likely that Laguna Creek, immediately below the site, would have served as a permanent source of water. It should be pointed out that with 200 rooms at Wildcat Canyon Ruin, 60 rooms at Six Foot Ruin, and 200 to 300 rooms at Organ Rock Ruin and its satellites, there would have been 400 to 600 people living within about a 2-km radius of the mouth of Tsegi Canyon. Given that there were only 500 to 600 occupied rooms in the canyons proper during the Tsegi Phase, the new survey data now indicate that there were almost as many people living right at the mouth of the Tsegi Canyon system as lived in all of the site canyons put together. Klethla Valley Like Long House Valley to the east, the floor of Klethla Valley lies at an elevation of approximately 2061 m. From the low hills separating Long House Valley and Klethla Valley, the 1984 research area extended 22 km to the west, where Klethla Valley and the Shonto Plateau begin to merge. Within the research area, Klethla Valley is clearly defined by the steep escarpment of Black Mesa on the south and by the hills, mesas, and sandstone outcrops of Shonto Plateau on the north (fig. 4-19). Annual precipitation averages about 280 mm, and the predominant vegetation is pinyon-juniper forest on the Kayenta and Klethla Valleys 73 Figure 4-19. Aerial photograph of Klethla Valley, facing south. (Courtesy of the School of American Research, Santa Fe, New Mexico: 1993. Photograph by Paul Logsdon.) valley slopes and a combination of sagebrush and rabbitbrush on the valley floor (Dean et al., 1978:27-28). The survey was conducted to see if there was a network of focal sites in Klethla similar to that found in Long House Valley, and to see if the Long House Valley focal sites were visually linked with contemporaneous sites in Klethla. The survey was concentrated in the upper (eastern) end of Klethla where it abuts the lower reaches of Long House Valley. The research area investigated in 1984 extended over 200 sq km. A total of 84 HPT As were surveyed along with 5 LPT As of 1 sq km. A total of 290 sites were recorded (fig. 4-1). Of these, 82 dated prior to a.d. 1100, 32 dated to the period a.d. 1100-1300, and 1 proto-Hopi site was located. More than one-quarter of the sites recorded dated to the 200-year period being investigated; the rest spanned over 3,000 years. Twenty-three sites dating to the Tsegi Phase were recorded, and the pattern of site types and locations mirrors in some ways that found in Long House Valley. Five of the sites were much larger than the others and had similar characteristics to the Long House Valley focal sites, with the open plazas, and a "pueblo of orientation" having cored, double-faced masonry. Eleven of the Tsegi Phase sites, including two focal sites, were in nondefensible locations. All of these sites were located near dunes suitable for dry farming, and one, UKV-72, was associated with several man-made catchment basins and check dams. The remaining 12 sites from this period were all located in defensible positions. By and large, the sites in Klethla Valley conform to the 74 Stress and Warfare Among the Kayenta Anasazi Figure 4-20. Aerial photograph, facing north, of the rocky outcrop upon which Hoodoo Heaven Ruin is located. Architectural rubble covers the end of the outcrop in the foreground of the photograph. (Courtesy of the School of Amer- ican Research, Santa Fe, New Mexico: 1993. Photograph by Paul Logsdon.) pattern of defensively located focal sites surrounded by much smaller satellite villages. Twelve Tsegi Phase sites are in defensible locations on hilltops, cliffs, or other geographic promontories that restrict access. Some of these sites also possess overtly defensive features such as walls across access routes, access limited to a single walled corridor, and hand-and-toe trails leading to the tops of nearby high rock outcrops suitable for lookout or signaling stations. These sites range in size from one or two roomblocks of four rooms each to focal sites of 100 or more rooms. They are often also associated with reservoirs, or bedrock pools to catch rain and runoff. Four of the 12 defensible sites are actually satellites of the large focal site. Hoodoo Heaven (UKV-92), which is located on the irregular sloping sides of a large solid sandstone outcrop rising 25 m above the valley floor (fig. 4-20). Double-coursed masonry was employed in the construction of at least one structure, though this may have been a kiva rather than the pueblo of orientation found in other focal sites. There is only one access route up onto the outcrop, and there is an architectural wall built across this single route. The surrounding land is suitable for cultivation, and the areas that would have been fields can all be observed from the uppermost section of the site. A number of catchment basins occur in the bedrock around the site, and in addition, at the highest part of the site there is a reservoir in a natural declivity Kayenta and Klethla Valleys 75