INFORMATION TO USERS This manuscript has been reproduced from the microfilm master. UMI films the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer. Tbe quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins and improper alignment can adversely affect reproduction. In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion. Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand corner and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book. Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6" x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order. A Bell & Howell information Company 300 North Zeeb Road. Ann Arbor. Ml 48106-1346 USA 313.'761-4700 800/521-0600 ECOLOGICAL AND CONSUMER GROUP VARIATION IN EXPEDIENT CHIPPED STONE TECHNOLOGY OF THE PUEBLO PERIOD: AN EXPLORATORY STUDY IN THE SILVER CREEK DRAINAGE, ARIZONA by Eric James Kaldahl A Thesis Submitted to the Faculty of the DEPARTMENT OF ANTHROPOLOGY In Partial Fulfillment of the Requirements For the Degree of MASTER OF ARTS In the Graduate College THE UNIVERSITY OF ARIZONA 1995 UMI Number: 1362229 UMI Microform 1362229 Copyright 1995, by UMI Company. All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code. UMI 300 North Zeeb Road Ann Arbor, MI 48103 2 STATEMENT BY AUTHOR This thesis has been submitted in partial fulfillment of requirements for an advanced degree at The University of Arizona and is deposited in the University Library to be made available to borrowers under the rules of the Library. Brief quotations from this thesis are allowable without special permission, provided that accurate acknowledgment of source is made. Requests for permission for extended quotation from or reproduction of this manuscript in whole or in part may be granted by the head of the major department or the Dean of the Graduate College when in his or her judgement the proposed use of the material is in the interests of scholarship. In all other instances, however, permission must be obtained from the author. SIGNED:^- APPROVAL BY THESIS DIRECTOR This thesis has been approved on the date shown below: Cj. J j U f o Barbara J. Mills Assistant Professor of Anthropology /J/ay /9?s Date 3 ACKNOWLEDGEMENTS I would first like to extend my gratitude to Barbara Mills, for making the lithic collections of the Silver Creek Archaeological Research Project available to me for analysis, in addition to the Research Project's computing facilities. Without her patient advice, support, guidance, editorial and research recommendations, this research would not have been possible. I would also like to thank Steve Kuhn, whose chipped stone knowledge was invaluable in directing me to relevant sources of interest. His particular insights into quantitative methods helped me to arrive at more elegant and on-point statistical discussions than were initially formulated. I would like to thank Jeff Reid, whose knowledge of the Mogollon region helped locate relevant supporting literature. His thoughts on the Neolithic lithics of the Southwest served to encourage these investigations, and his advice in the midst of researching dilemmas proved both supportive and illuminating. I would also like to thank Deb Olszewski, who started me on the road to relevant chipped stone research topics at the beginning of my investigations. The expediency with which this research, manuscript, and my education have proceeded during the last two years would not have been possible without the support of the Jacob K. Javits Congressional Fellowship Program. Special thanks to Sarah Herr and Scott Van Keuren. Two people who kept eye out for source materials, kept an eye on me, shared ideas, offered advice, and supported ; - ^ all the ways of friendship. While the academic development of this research v uld not have proceeded without the help of my committee members, a quality life would not have been possible without Sarah and Scott's understanding, concern and kindness. As always, I thank my parents Ann and Phil Kaldahl. They are that foundation of support upon which rests the entire structure of my education, career and life. I can never thank them enough for always being there for me. 4 TABLE OF CONTENTS LIST OF TABLES 6 LIST OF FIGURES 9 ABSTRACT 11 CHAPTER I: EXPEDIENT TECHNOLOGY AND THE PUEBLO PERIOD 12 Mobility, Sedentism and Expedient Technology in the Southwest 13 The Influence of Modern Experimental Studies in Archaeological Interpretation 17 Impacting Variable,* of Interest on Chipped Stone Technology 19 Functional Form 22 Material Type 22 Technological Demands 22 Reduction Intensity 23 Conclusion 24 CHAPTER II: RESEARCH SETTING Ecological Context Geologic Context Social Context Sites Sampled in the Silver Creek Region Conclusion 25 25 31 40 41 46 CHAPTER III: RESEARCH METHODOLOGY AND DESIGN Analytic Factors, Levels, and Research Expectations Statistical Philosophy Attributes Recorded Material Types Technological and Functional Types Cortex Platforms Use Wear Completeness Metric Variables 47 47 52 52 52 53 53 54 54 54 57 5 TABLE OF CONTENTS-Continued CHAPTER IV: EXPEDIENT TECHNOLOGY IN VARIABLE ECOLOGICAL SETTINGS 58 Elevational Differences in Chipped Stone Assemblages 61 The Agave Factor and its Impact on Chipped Stone Assemblages .... 85 Conclusion 95 CHAPTER V: SOCIAL ORGANIZATIONAL EFFECTS ON CHIPPED STONE TECHNOLOGY Site Size Factor in Chipped Stone Assemblages A Consideration of Chert Conclusion 97 100 110 113 CHAPTER VI: SUMMARY DISCUSSION AND CONCLUSION The Proposition Technological Demands of Subsistence and Ecological Setting The Influence of Site Size on Chipped Stone Implications Building References for Further Investigation 114 114 115 116 116 119 APPENDIX A: SILVER CREEK ARCHAEOLOGICAL RESEARCH PROJECT LITHIC ANALYSIS KEY AND CODE BOOK 121 APPENDIX B: FREQUENCY TABLES 136 APPENDIX C: STATISTICAL NOTES Part A: Why these p-levels? Part B: Significance Testing of Spearman's Rho 159 159 160 REFERENCES 162 6 LIST OF TABLES TABLE 2.1: Correlation between Elevation and Precipitation 27 TABLE 2.2: Geologic Sequence of SCARP Area 32 TABLE 2.3: Surface Collected Sites and Characteristics of Interest 43 TABLE 3.1: Summary of Research Factors of Interest 49 TABLE 3.2: Frequency of Selected Material Types 53 TABLE 4.1: Elevation Factor: Sites in Elevation Levels 61 TABLE 4.2: Pearson chi-square test of homogeneity: Probability of Similar Debitage Category Frequencies Between Elevational Levels 62 TABLE 4.3: Pearson chi-square test of homogeneity: Probability of Similar Debitage Category Frequencies Between Sites and Within Elevational Levels 63 TABLE 4.4: Spearman's Rank Correlation Coefficients, Complete Flake Length by Elevation 71 TABLE 4.5: Spearman's Rank Correlation Coefficient, Thickness of Flakes with One Complete Dimension by Elevation 72 TABLE 4.6: Spearman's Rank Correlation Coefficients, Tested Cobble Weight by Elevation 78 TABLE 4.7: Spearman's Rank Correlation Coefficients, Core Weight by Elevation 78 TABLE 4.8: Pearson chi-square test of homogeneity: Probability of Similar Cortical Flake Type Frequencies between Elevational Levels 79 TABLE 4.9: Spearman's Rank Correlation Coefficients, Cortical Flake Length by Elevation (cortex > 10%) 80 TABLE 4.10: Spearman's Rank Correlation Coefficients, Thickness of Cortical Flakes with one Complete Dimension by Elevation (cortex > 10%) ... 80 7 LIST OF TABLES-Continued TABLE 4.11: Pearson chi-square test of homogeneity: Probability of Similar Frequencies of All Debitage Between Agave and Non-Agave Sites .... 86 TABLE 4.12: Pearson chi-square test of homogeneity: Probability of Similar Frequencies of All Basalt Debitage among Agave Sites and Non-Agave Sites 87 TABLE 4.13: Spearman's Rank Correlation Coefficients, Total Basalt Weight at Sites by Distance from Source 91 TABLE 4.14: Spearman's Rank Correlation Coefficients, Total Basalt Weight of Sites by Distance from Source, Separating Agave Factor 92 TABLE 4.15: Pearson chi-square test of homogeneity: Degree of Cortical Coverage on All Basalt Artifacts by Distance from Source Factor 94 TABLE 4.16: Spearman's Rank Correlation Coefficient, Cortical Coverage > 0% by Distance from Source 94 TABLE 5.1: Sites by Site Size Factor 98 TABLE 5.2: Pearson chi-square test of homogeneity: Probability of Similar Frequencies of All Debitage Types Between Site Size Levels 100 TABLE 5.3: Pearson chi-square test of homogeneity: Probability of Similar Frequencies of All Debitage Types Between Sites within Site Size Levels 101 TABLE 5.4: Spearman's Rank Correlation Coefficients, Complete Flake Length by Site Size Factor 105 TABLE 5.5: Spearman's Rank Correlation Coefficients, Thickness of Flakes with One Complete Dimension by Site Size Factor 106 TABLE 5.6: Spearman's Rank Correlation Coefficients, Core Weight by Site Size 109 8 LIST OF TASLES-Continued TABLE 5.7: Spearman's Rank Correlation Coefficients, Cortical Coverage (cortex > 0%) of Chert Flakes by Site Size 112 TABLE 6.1: Pearson Correlations, Coefficients of Determination, Probability Values for Median Length of Complete Flakes (90% Confidence Intervals) by Factors of Interest 118 9 LIST OF FIGURES FIGURE 1.1: Archaeological Project Areas 15 FIGURE 1.2: Impacting Variables of Interest on Chipped Stone Technology 21 FIGURE 2.1: Little Colorado River Drainage 26 FIGURE 2.2: Ecological Communities by Elevation 29 FIGURE 2.3: Surface Geology, Southern Silver Creek Drainage 34 FIGURE 2.4: Geologic Cross-Section of the Mogollon Rim, Fort Apache Region 35 FIGURE 2.5: Distribution of Surface Collected Sites 38 FIGURE 3.1: Frequency of Basalt Flakes by Distance from Source 51 FIGURE 4.1: Box Plot of Complete Diabase Flake Lengths, Elevation Factor 65 FIGURE 4.2: Box Plot of Complete Silicified Wood Flake Lengths, Elevation Factor 66 FIGURE 4.3: Box Plot of Complete Chert Flake Lengths, Elevation Factor 67 FIGURE 4.4: Box Plot of Complete Quartzite Flake Lengths, Elevation Factor 68 FIGURE 4.5: Correlations Between Median (90% C.I.) Complete Flake Lengths and Elevation 75 FIGURE 4.6: Frequency of Basalt Debitage, Agave vs. Non-Agave Sites .... 87 FIGURE 4.7: Agave Sites and Non-Agave Sites: Distance to Basalt Sources 90 10 LIST OF FIGURES-Conf/nued FIGURE 4.8: Frequencies of Basalt Core Weights by Distance to Source Factor 93 FIGURE 5.1: Box Plot of Complete S-Wood Flake Lengths, Site Size Factor 102 FIGURE 5.2: Box Plot of Complete Chert Flake Lengths, Site Size Factor 103 FIGURE 5.3: Box Plot of Complete Quartzite Flake Lengths, Site Size Factor 104 FIGURE 5.4: Correlations Between Median (90% C.I.) Complete Flake Lengths and Number of Rooms 108 ABSTRACT Lithic raw material variety and abundance reveals the technological utility of different source materials from 20 chipped stone surface collections in the Silver Creek area of east-central Arizona, from sites dating between the 9th and 14th centuries. A rich raw material environment obviates distance-from-source constraints, freeing debitage analysis from traditional spatial interpretations regarding the intensity of reduction. Rather the intensity of reduction and the frequency of distinct material types in each assemblage reflects the impact of social organization, community size, exchange and subsistence variation on the organization of chipped stone technology. 12 CHAPTER I: EXPEDffiNT TECHNOLOGY AND THE PUEBLO PERIOD The study that follows explores patterned variation in 20 chipped stone surface collections from Pueblo period sites in the Silver Creek drainage of eastcentral Arizona. While chipped stone assemblages are impacted by a variety of variables, of particular interest in this research is the co-variation between pre existing variables such as site size, ecological setting, raw material availability, and inter-site exchange networks with relevant variables in expedient lithic technology. Inter-assemblage variation of Pueblo period chipped stone technology has not enjoyed much attention until recent years; therefore, this study proceeds from an exploratory stance. The Silver Creek drainage is a lithic rich and diverse environment, with a variety of readily available raw materials. Consequently, variation in reduction intensity, raw material use, and length of manufactured cutting edge is assumed to reflect variables other than distance-from-source-bed constraints. This chapter considers expedient technology in the Southwest, the role of experimental chipped stone studies in current research, and a set of variables that impact chipped stone technology and can be explored with currently available assemblages. 13 Mobility, Sedentism, and Expedient Technology in the Southwest With greater sedentism in prehistoric Southwestern communities, chipped stone tool technology became increasingly expedient, characterized by less formal tools, a more generalized use of raw materials, and increased numbers of utilized flakes and other debitage (Leonard et al. 1989:106-108; Olszewski and Simmons 1982). Chipped stone tools in highly mobile societies are characteristically lightweight, multifunctional, and curated (Parry 1987:227-228; Parry and Kelly 1987:298-302). Every tool must ultimately perform a variety of tasks, and the tool's function alters throughout various stages of curation. In sedentary societies and lithic rich environments, the need for such tool kits declines. Expedient tools from all types of source material are quickly manufactured and simple in form. An expedient tool has a short use-life and is created to perform a limited set of tasks. Some archaeologists lament the declining information content of chipped stone tools in sedentary societies (Shott 1986:45-46), but perhaps it is the case that we are not properly utilizing the information available. Consider a lithic rich and diverse environment, where sedentary peoples can select from a wide range of readily available raw materials. In such a setting 1) the raw material selected for use reflects needs that only that material can fulfill, and 2) the blank form of an expedient tool is created to serve a limited set of tasks. The cutting edge of the simplest flake has a functional meaning other than tool curation and bifacial manufacturing steps. Expedient chipped stone 14 assemblages viewed across a landscape of shifting ecological and social environments should reflect adaptational variation. Some of the characteristics of expedient technology are discussed below as developed in previous research. Lithic analyses from large archaeological projects in the Southwest include collections that span the range from Archaic to Pueblo period sites. Broad trends in technological change have been observed throughout the region. In the Lower Chaco river region (Reher 1977), surface collections from the Coal Gassification Project (CGP) include sites from Archaic, Basketmaker, Anasazi, and Navajo periods. The lithic assemblages from Archaic to Anasazi indicate increased expediency of lithic technology, as evidenced by less formal tool production, higher tool/artifact ratios through time, due to more utilized flakes per core (Chapman 1977:446-451). An intensive lithic analysis of materials from the Navajo Mine Archaeological Program (Hogan et al. 1985:276-279), just north of the CGP study area (figure 1.1), demonstrated that longer flakes were indicative of Pueblo period sites, in contrast to Archaic sites which possessed shorter flakes posited to be the result of more bifacial tool manufacture. Further the shift to more expedient production is supported by higher frequencies of primary and secondary cortical flakes in Pueblo assemblages. 15 Archaeological Project Areas Utah Arizona New Mexico ^ Farmingt-on Gallup s 100 km Quemodo Project- Name Navajo Mines Coal GassiFication ProjectBlack Mesa Project TEP St..Johns Project. Silver Creek Project. Figure 1.1: Archaeological Project Areas 16 Archaeological investigations at Black Mesa (Leonard et al. 1989:107; Pariy 1987:215-220) observed many more expedient tools in Pueblo period sites than in the Basketmaker sites studied. The Tucson Electric Power St. Johns Project (Sullivan and Rozen 1985:764-769) concluded similar findings. Pueblo period lithic reduction strategies from the Lower Chaco River in New Mexico to the Upper Little Colorado and Salt river drainages in Arizona emphasize expedient flake tools, consuming a large number of cores relative to pre-Pueblo periods. Correlated with expedient technology is increasingly generalized raw material acquisition and more land extensive resource extraction. The Lower Chaco case (Chapman 1977:450) provides such evidence, with analyses indicating that a great variety of raw materials were selected. Clarifying this relationship, Hogan et al. (1985:256-258) demonstrate that the diversity of raw materials is greater during the Pueblo period. These researchers speculate that Archaic period lithic acquisition is embedded in the subsistence system, hence raw materials were gathered closest to occupation sites creating uneven distributions in these assemblages. The more sedentary Pueblo peoples extracted lithic raw material across a broad region in a variety of behavioral contexts, perhaps while visiting distant fields, traveling or extracting other resources. Studies conducted around Black Mesa (Leonard et al. 1989; Parry 1987:209-211) also demonstrate more generalized raw material acquisition, in contrast to Basketmaker assemblages which show a marked preference for raw 17 materials immediately adjacent to the sites. The Pueblo period assemblages have raw materials acquired from both local (0-15 km) and non-local (15-30 km) sources, with a preference for higher quality non-local silicates. Regardless of the fact that non-local materials were farther from sites, the tools manufactured were expedient, not curated. This further supports the Pueblo trend toward regional exploitation of lithic sources, acquired in large enough quantities to make a coreconsumptive expedient technology feasible. The Influence of Modern Experimental Studies in Archaeological Interpretation Modern flintknapping has shed great light on lithic technology's requirements, potentials and interpretations. However, most modern flintknappers strive to create aesthetically pleasing replicas of formal prehistoric tools (Crabtree 1982; Whittaker 1994). As mentioned above, Pueblo period chipped stone technology is characterized by increasing expediency (utilized flake tools) and a lack of formal stone tool technology (bifacially produced tools). Consequently some of the models and concerns presented in experimental studies and flintknapping experiments do not address the fascinating questions which can be put to sedentary Pueblo technology. Early attempts to analyze whole chipped stone archaeological assemblages (i.e., including debitage) merely attributed flakes to the detritus of bifacial manufacture (Collins 1975:16-26; Hill 1970:42; Martin et al. 1967:122-125). In recent experimental studies (Mauldin and Amick 1989; 18 Odell 1989b), the analysis of debitage as a simple reflection of bifacial manufacturing stages persists. Investigators strive to identify experimentally derived bifacial production "signatures" in the archaeological record (Patterson 1981, 1990). Although some investigations (Ahler 1989) have grown quite complex in testing percussion methods, stages of bifacial manufacture, etc., the emphasis remains on bifacial tool technology. Bifacial experimenters and archaeological modelers are certainly aware of the problems other technological methods represent Many factors will have to be considered before we can confidently apply the size distribution of waste flakes from biface reduction to archaeological problems. These potential problems may include the use of different manufacturing methods to achieve the same product, the combined deposition of waste flakes from different stoneworkinp technologies. inaccurate replication of the particular stoneworking tradition of interest, the effect of lithic raw materials, prehistoric selection of certain flake size categories, and modern relic hunting [Stahle and Dunn 1984:94; emphasis added]. When considering non-bifacial approaches in technology, lithic experimenters tend to dichotomize the results. Experimental and archaeological assemblages are characterized as bifacial reduction versus generalized "core reduction" (Ingbar et al. 1989) or amorphous flake producing industries (Burton 1980:138-139). Flake production is often viewed as an ill-defined, unpatterned process rather than the result of deliberate forethought and decisions by the knapper to create a tool with a task in mind, a view I do not share. 19 In comparing Southwestern sites of different time periods (Chapman 1977; Hogan et al. 1985; Leonard et al 1987; Pariy 1987; Sullivan and Rozen 1985), this biface-or-not dichotomy has been the central issue of concern, yet to be addressed is the variation in expedient technology per se. Debitage variation in experimentally derived bifacial production stage models has received much attention, whereas the existence of "other" non-bifacial technologies has been noted, but its sensitivity and meaningful variation remains unexplored. Such an exploration is the purpose of this study. Impacting Variables of Interest on Chipped Stone Technology Expedient tools are selected and discarded after a short use-life, little modification, and little curation (Parry 1987:227-228). The tool created is designed for a limited range of functions, hence in debitage analysis the choice of raw material and size of the artifact produced should reflect behavioral meaning quite distinct from the production sequence models that characterize a bifacial tool assemblage. For the archaeological researcher, expedient technology presents a different array of problems and concerns. The inter-site study of such assemblages demands the consideration of the many variables which impact a chipped stone technology. Recent thoughts on the variables contributing to Old World formal tool typologies (Kuhn 1992:126; Neeley and Barton 1994:287; Rolland and Dibble 20 1990:484-492) have sparked a general consideration of all the variables which impact chipped stone technology. Tools are shaped by the function they perform and the technological strategies of their makers. The physical properties of raw material, degree of reduction intensity, access to raw material, diversity of raw material, climate and ecological setting, consuming population structure, regional trading networks, subsistence strategy, historical development, and other resource extraction strategies combine to impact a chipped stone technology, whether expedient or formal. In an effort to simplify and consider these variables in a Southwestern context, I have chosen four larger variables of interest for this discussion (figure 1.2). It should be noted that post-depositional processes that have effected the assemblages of this study will not be discussed in this chapter, but will be addressed in chapter two along with the nature of the collections analyzed. Impacting Variables of Interest On Chipped Stone Technology Dtan»,te Some Material Type Functional Farm Sample Space Univetse DbtaQKeto Sonne Domeatlo NeocwklM ActMloltw Mrictfcb Reduction Intensity Figure 1.2: Impacting Variables of Interest on Chipped Stone Technology M 22 Functional Form. The tool forms manufactured by ancient knappers obviously shape the final disposition of a lithic collection. As previously discussed, bifacial and expedient technologies are tied to the degree of mobility. The shape and size of expedient tools serve a specific range of functions, and should be sensitive to different resource extraction and subsistence system patterns. Material Type. The physical characteristics of utilized raw materials clearly impact the chipped stone assemblage. Access to raw materials plays a large role in materia] type choice, further discussion of which will follow in the section on reduction intensity. Sedentary peoples of the Southwest employ an effective raw material acquisition system, as in the Black Mesa example (Leonard et al. 1989) where Pueblo people obtained higher quality cherts from non-local sources from off the Mesa proper. Whether this acquisition system operates by trade, casual acquisition during visits with other people, or is embedded within strategies of non-lithic resource collection, the system does work to procure desired materials in enough quantity to fuel a veiy core-consumptive expedient technology. Technological Demands. Stone tool technology is intimately tied to extracting resources from the environment, and consequently chipped stone is expected to be sensitive to ecological variation (Binford and Binford 1969:70-78; Patterson 1977:76-78). Biotic variation between sites presents different communities with different subsistence opportunities (Cartledge 1977:140-144). 23 Additionally other resources in different ecological settings (e.g., wood for fires, construction, wild plants) will impact lithic assemblages differently, resulting in a chipped stone technology which is adapted to a community's environment. This is especially the case with sedentaiy Pueblo peoples, whose expedient tools are assumed to have been manufactured and discarded close to the ecological setting in which they were used. Reduction Intensity. The degree of reduction in a chipped stone assemblage is dependent on the functional form of the tools created, the technological demands placed upon the tools, and access to raw materials. In the case of the Pueblo Southwest, expedient tools are characteristically less reduced than bifacial tools; however, all tools undergo some attrition during use which varies according to the tasks to which a particular tool is subjected. Access to raw materials is a frequently cited factor in understanding reduction intensity, as monitored through debitage and formal tool analysis. Distance from source presents particular constraints on technology, as transportation cost increases. In similar time periods a site in a lithic rich region may employ a flake technology while a site in a lithic poor region uses a bifacial technology (Burton 1980:138). Tools may be manufactured in their final forms close to lithic sources, while curated tools and blanks of known quality are transported to distant sites for use (Henry 1989; Ricklis and Cox 1993). Lithic scarcity can result in the recycling of 24 old resources and producing smaller, broken lithic fragments in assemblages (Olszewski 1993:190-192). The social organization and regional networks of the consuming population also plays a role in reduction intensity. Increasingly sedentary peoples (Shott 1986:45-46) acquire greater accumulations of material culture in general, and logistical trips from the center camp can effectively exploit regional resources (Leonard et al. 1989). Large populations can negatively impact access to resources as well, as localized resources become territorialized by social groups. Conclusion The variables which impact a chipped stone technology are obviously not discrete, but interactive. In an effort to isolate and focus upon the effects of any one variable, the others must be held constant or minimized in a fashion which is tailored to a specific archaeological data set and region. In the case of chipped stone technologies of the Pueblo Southwest, a case has been made for the sensitivity of such an expedient technology's debitage to the demands placed upon it in different prehistoric communities. Tool size and material choice in a lithic rich and diverse environment should reflect tool function. Each community's structure and ecological setting should shape its lithic assemblage, resulting in patterned lithic variation in similar contexts. In the next chapter, the research setting is explored. 25 CHAPTER II: RESEARCH SETTING To pursue patterned variation in chipped stone assemblages, regional ecological and geologic diversity needs to be assessed. Ecological variation will impact the resource extraction activities in which stone tools are employed, and geologic diversity obviously effects raw material availability. The social context of chipped-stone-using communities may also impact the amount of raw materials consumed and the movement of chipped stone materials across the landscape. To summarize the impact of ecological, geologic and social contexts, the Upper Little Colorado drainage in general and the Silver Creek drainage in particular will be described with regard to these variables. Site characteristics of interest will be summarized. Finally the nature of the chipped stone assemblages in the sample will be addressed, with concern for the processes which contributed to the formation of the collections. The strength of these collection for analyzing chipped stone behaviors of interest will be assessed. Ecological Context In the Upper Little Colorado drainage, increasing moisture generally correlates with increasing elevation, although this is somewhat of an oversimplification. Johnson (1976:26-29) analyzed precipitation records at 43 recording sites in the Little Colorado River basin (figure 2.1). 26 Little Colorado River Drainage ULnh Colorado Arlzona New Mexico Farmingtan Wine ow ^ac k / <\ y' iny 'ltil/ 90j7bl^j/ Gallup * ^^^inslgw--'' FlagstofT V Zuni y [N iS 100 km Q, . * bhaw Low \ 1 Quernado • Sprlragervl e Figure 2.1: Little Colorado River Drainage 27 Precipitation had been recorded at these sites for an average of 35 years, spanning elevations between 4165-8490 feet By re-analyzing Johnson's data, the correlations below are observed (table 2.1). Upon inspection, a positive correlation is apparent between elevation and precipitation, although the coefficient of determination values indicate that there are more factors involved in this relationship than elevation alone can account for. Table 2.1: Correlation between Elevation and Precipitation Type of Precipitation Correlation Coefficient (r) Coefficient of Determination (r2) Summer Rain 0.655 0.429 Winter Rain 0.515 0.265 Total Rain 0.597 0.357 Snow 0.824 0.680 Moisture-bearing weather systems in the winter can originate from the west or from the southwest of the region (Johnson 1976:33-36). Weather systems which approach from the southwest are much more variable annually, but large amounts of rain and snow are possible. These systems encounter the Mogollon escarpment, and the orographic effect results in much of the moisture being dropped in the immediate vicinity of the rim, with a rain shadow producing remarkable decreases in total moisture at comparatively short distances into the basin. For example, Show Low (6 km north of the Rim) receives 38.2 inches of 28 snowfall annually, while Snowflake (roughly 34 km north of the Rim) receives 16.9 inches (Johnson 1976:27). The Silver Creek sites included in this study are situated in close proximity to the rim, which allows them to take advantage of the moisture from these southerly storm systems. Temperature correlates quite closely with elevation: correlation coefficient 0.88 (Johnson 1976:31). While higher elevations enjoy greater moisture, they experience shorter frost-free seasons and lower year-round temperatures. For elevations above 6400 feet in the study area, the annual average monthly low temperature is at or below freezing (Johnson 1976:26-29). Ground water in the area is charged by winter precipitation, as most summer precipitation runs off rapidly (Johnson 1976:33). Ecological variation is closely tied to elevational differences and orographically produced rain shadows (figure 2.2). Lusher environments at higher elevations are characterized by ponderosa pine communities, and lower elevations by pinon-juniper communities (USDA Forest Service 1973:8-13; Plog 1981:25-29; Satterthwait 1976). It should be noted that the boundaries between ecological communities cannot be drawn with confidence across the region as local landfonns, upthrusts, and drainages in addition to elevation and distance from the Mogollon Rim result in very fine differences in ecological communities over restricted geographic areas. Ecological Communities by Elevation Elevation (ft.) 8000 7000 6500 — 6000 16-18 miles Vegetation Association Shortgrass Grassland Plant Community Dominant Vegetation Mammals Birds Reptiles S Amphibians Pifion-Junlper Woodland Pinon- Cottonwood Juniper Willow Ponderosa Pine Forest Ponderosa Pine PineBunchgrass Pine-Fir Mountain Meadow Fir-Aspen Ariz.Fescue Ponderosa Hairgrass Mt.Muhly Pine Sedges Ponderosa Douglas Fir Clover Gambel Oak Pine White Fir Douglas Fir Aspen White Fir Elk Muledeer Badger Bobcat Coyote Elk Muledeer Black Bear Red squirrel Elk Black bear Lion Red squirrel Goshawk Turkey Spotted Owl Robin Raven Blue gramma Galleta Sand Dropseed Pinon Juniper Sage Pronghorn Blaclctail Jackrabbit Cottontail Gray Fox Coyote Muledeer Raccoon Kit Fox Skunk Cotton Rock tail squirrel Meadowlark Night Hawk Horned Lark Lark Oriole Turkey Sparrow Red-tailed Goshawk RockWren hawk Chickadee Turkey W.Bluebird Sparrow Blacktailed rattlesnake Bullsnake Blacktailed Horned rattlesnake Lizard Kingsnake Cottonwood Ponderosa Willow Pine Walnut Gambel Oak Birch Alligator Gambeloak Juniper Gartersnake Elk Muledeer Porcupine Ht. Lion Horned lizard Kingsnake Pine-Fir Forest Elk Pocket gopher Deermouse Turkey Raven Spotted Owl Tiger BlkTailed salamander rattle snake Figure 2.2: Ecological Conmunities by Elevation to 30 The implications of elevational differences in moisture and temperature cannot be overlooked when considering farming communities of any time period. Corn demands a frost-free season of at least 110 days, and usually longer for year-to-year crop security (Hevly 1983:29). The early growth of a crop is also dependent on adequate moisture in the dry early spring of the study region; groundwater recharged through winter precipitation must be plentiful. Abruzzi's (1993) study of the Mormon colonization of the Little Colorado region neatly summarizes the farmer's conundrum: Generally, elevations above 6,000 feet do not provide sufficiently reliable growing seasons. In addition, climatic variability has yielded crop failures even at locations which display sufficient average growing seasons. Elevations below 6,000 feet, on the other hand, do not receive sufficient precipitation to produce healthy crops [Abruzzi 1993:84]. Paleoclimatic changes will vary the ecological community distribution diachronically, making the occupation of some regions differentially advantageous over time (Hevly 1983:40). From the technological perspective, it suffices to say that sites of different topographic contexts will be situated within different ecological communities. At higher elevations near the rim, the increased winter moisture will produce denser stands of trees and more ground cover than lower elevations. Moreover, at higher elevations mixed economies must rely upon "back-up" systems, either exchange networks, mobility or more wild resource exploitation, due to their tenuous growing season. 31 Geologic Context As discussed above, the geologic history of the area has contributed to current climatological variation along the rim. In a more direct way, this history determined the raw materials available to prehistoric peoples for construction, chipped stone and ground stone. Table 2.2 lists the geologic sequence for the Silver Creek study area. The attached map (figure 2.3) is taken from the Arizona Bureau of Mines Geologic Survey Map of Navajo and Apache Counties (Wilson et al. 1960). For the study area, this map is based upon walk-over surveys that have not been updated since the initial publication (Arizona Geologic Sociely Archivist, personal communication 1995). 32 Table 2.2: Geologic Sequence of SCARP Area Period QUATERNARY (QTs) Group Name None Material Types Formation Name None 1. Recent erosion deposition: sands, silts, gravels, conglomerates 2. Recent Volcanics: Basalt and other material (Qb) TERTIARY (Ts) None None Rim Gravel Deposition: Oligocene CRETACEOUS (Ks) None None Feldspathic sandstone, shale JURASSIC None None None for SCARP area TRIASSIC (TRm) None Moenkopi Sandstone, shallow marine and tidal flat deposits. PERMIAN Supai (PPs) Kaibab (Pk) Limestone, dolomite, sandstone, mudstone, gypsum, conglomerate, assorted colors of chert Toroweap Yellow and red sandstone, mudstone, sandy dolomite, gypsum Coconino (Pc) Sandstone Schnebly Hill Sandstone, mudstone, limestone, dolomite, evaporates Hermit Mudstone, siltstone, conglomerates, limestone Esplanade Sandstone Wescogame Red sandstone, siltstone Earp Limestone, sandstone, conglomerate Manakacha Sandstone PENNSYLVANIAN Supai 33 Table 2.2 (continued): Geologic Sequence of SCARP Area Period Group Name PENNSYLVANIAN Supai Formation Name Material Types Watahomigi Cherty micritic limestone, mudstone MISSISSIPPIAN None None None for SCARP area DEVONIAN None None None for SCARP area SILURIAN None None None in Arizona ORDOVIOAN None None None for SCARP area CAMBRIAN Tonto Tapeats Sandstone Sandstone Apache None Diabase sills and dikes run throughout the Apache group Troy Quartzite Gray quartzite with sandstone None Basalt layer, as yet unnamed Mescal Limestone Limestone with black and gray chert inclusions, sandstone, dolomite Dripping Spring Quartzite Light gray to purplish gray quartzites, jasper, sandstone Pioneer Shale Lower strata conglomerate with white vein quartz, quartzite of assorted colors, jasper, rhyolite and schist clasts. Upper strata is mudstone, claystone PRE-CAMBRIAN SunFace Geology: Southern Silver Creeh Drainage y® PPs Supol Group Pk Kalbob Limestone Ks Cretaceous Sandstone Pc Coconino Sandstone D~Rm| Moenkopl Formation [TsjRim Grovels •bj Quoternary Basalt QTs Quaternary Erosion 5 kllomeLers Figure 2.3: Surface Geology, Southern Silver Creek Drainage -p- 35 While using geologic reconstructions, the approach that follows is from the perspective of raw material utilization. The enormous lithic variability in the area is directly attributable to the deposition of the rim gravels, which represent a melange of materials from a number of geologic strata. In figure 2.4 (Peirce et al. 1979:4-11), a simplified geologic sequence is presented for the study area, cross-sectioning the Mogollon Rim looking toward the northwest. Geologic Cross-Section of the Mogollon Rim: Fort Apache Region * •avels Canyon Creek Gravel Supai Cambrian "--^Devonian Pre-Cambrian Pre-Cambrian ississippian Penn. * not to scale, looking northwest "igure 2.4: Geologic Cross-Section of the Mogollon Rim, Fort Apache Region The Precambrian through Pennsylvanian strata all dip northeast, resulting in older strata atop younger ones. At the time of the Laramide orogeny, the area south of today's Mogollon Rim was actually at elevations higher than the rim. This uplifting event resulted in enormous erosional activity, as boulder size fragments were carried from south to north in an ancient drainage system. Consequently Precambrian rocks were exposed at elevations higher than today's 36 Rim over an area covering at least 50 km. The rim gravel zone (somewhat of a misnomer as the "gravels" range from cobbles to boulders in size) contain remnants of an unknown number of Pre-Cambrian and later strata. During the Oligocene Epoch, further basin and range faulting and uplifting resulted in today's Mogollon Rim, which divides the Little Colorado River drainage trending northwest from the Salt and Gila river drainages to the southwest. The University of Arizona's fieldwork in the Silver Creek area is close to the rim gravel province, which is found at the highest elevations of the region today. Minimally this province includes deposits from the Pre-Cambrian Apache Group (Wrucke 1989), which encompasses several strata of quartzite, shale, limestone, and the remains of unnamed diabase sills and dikes. The next geologic group is the Supai, which spans both the Pennsylvanian and Permian periods. There has been some dispute about the continuity of strata from these periods in the area (Peirce 1989); for the moment, I follow the sequence proposed by Blakey and Knepp (1989:320). The older sandstone, limestone, and siltstone strata of the Pennsylvanian are somewhat ephemeral until the lower Permian. The sandstones of the Coconino formation clearly outcrop in the study area, and when combined with later sandstones of the Triassic (Blakey 1989:372-375) and Cretaceous periods (Nations 1989:442-444), provide the building material for the masonry structures of the Pueblo period. 37 The Kaibab limestone, the uppermost Permian formation (Blakey and Knepp 1989:336-337), includes numerous chert inclusions in a variety of colors and qualities. Recent erosion by active streams and washes has cut through the rim grave] province to expose strata as deep as the Coconino formation. During the Quaternary, volcanic eruptions to the east of Show Low resulted in basalt and vesicular basalt materials available on the surface (Lynch 1989). Figure 2.5 maps the sites of the study area atop the geologic provinces. The Tertiary rim gravel province is suspected to contain a very diverse array of lithic raw materials, including at least the Apache group's quartzites, diabase, limestone, chert/jasper, and shale. Also suspected to originate in this province are nodular silicified wood specimens. The rim gravels are within 4.5 km of all the sites in the study area. The exact surface outcrops and the dominant raw materials in any given section of the rim gravels is unknown, requiring extensive surface sampling across the region which has yet to be done. [n Distnibution oF SunFace Collected Sites /4;P;1^Pk • AR-05-,201 \ . AZ-PJLl v (260) 3 r / ^ t s 7 /f ^F^lne>dale ^ Ts ^9 ^ V? \ J^Jc7 ^ S ^ (/fVKjyl\ < W Ts \ zD) . Jo (?v j /Q \ 1 )Pk'AZ:P:12:196 //) MT^>^AZ:P:12:210 ^U^/j Qb \ AZ:P:12:242^ /iCAZ:P42:2ir 5*^6 Ob \ AZ:P:12:222«# •*^*£^1\ AZ:P:12227 >t5 Linden^H 1 <^-^> ! ' AZ:P:12:260 ^ A7-P-]?-?4RfP _ M V^^zei2:256^ T3 avo'^S •AZ:P12:65(_ _ 1 \ I \ #AZ:P:l^En^f4rL ^Wy^T ^K\ 2^ AZ:P:16:112 •AZ:P:1S:139 ShgfiT LSw /X/CO sf°^o\ /ViT , s AZ:P:16:153» f <^=H^Z:P:16/?6 ^ If5 ^C^yTY^ AZ:P:16:160^-Ts V / Is ( |Qb| Quoter-ner-y Basealt 1* | Pk Kolbab Limestone SurFQCB Collected Site Figure 2.5: Distribution of Surface Collected Sites Ts Rim Gravels 5_K^^ETG£S 39 A less random and perhaps more reliable location for chert is the Kaibab limestone. Once again most of the sites (18) are within 4 km of the formation. The remaining two, AZ:P:11:1[ASM] and AR-05-1201[USFS], are at distances of 6 km and 7.5 km respectively. As the above figures attest, a broad array of raw materials, from quartzites to chert, are at a short distance from the sites. Considering that higher quality silicates at Pueblo sites on Black Mesa were brought in from distances between 15 km and 30 km (Leonard et al. 1989:106108), the conclusion drawn is that the above material types are readily available in the Silver Creek area, and consequently distance-to-source constraints for raw material acquisition are obviated. This is not the case for basalt from east of Show Low. Nine sites are within 6 km of the basalt flows, nine are between 7 km and 10 km, and the remaining two are 27 km and 32 km from source. While vesicular basalt groundstone has been found in the area, basalt also appears in the chipped stone assemblages at all sites, making some basalt acquisition system common to the area. The role of active washes since the Oligocene should be noted. Active streams and washes have cut through the rim gravel province, bringing somewhat smaller pieces down to lower elevations. The importance of streams in exposing the Kaibab limestone and sandstone formations cannot be overestimated, as the 40 easiest quarrying of these materials would be in the drainages. No systematic study of outcrops and recent deposition in these washes has been made. Social Context The presence of exchange relationships and regional integration can create a context where raw materials or finished products can move around the landscape. The role of social and regional interactions in the Little Colorado drainage has been addressed by other researchers (Lightfoot 1981, 1984). As discussed above, most sites in the study area do not experience distance-to-source material constraints, except with regard to basalt. With that one exception, there would apparently be little need for networks to facilitate the movement of raw materials. However, the diachronic trend between formal and expedient lithic technology is a pan-North American phenomenon (Parry and Kelly 1987), and in spite of this switch, formal stone tools continue to persist side-by-side with expedient tools. Expedient technology (Koldehoff 1987:176-185) in the Mississippian context is a household produced technology, while many formal tools are manufactured at a restricted set of locales and exchanged across the landscape. With that example in mind, the Silver Creek area has several sites with great kivas, six of which are in our sample. Herr (1994:75-76) admonishes future researchers in the area to consider the role of great kivas and other integrative 41 features in settlement patterns and in the relationships between sedentism, integrative architecture, and economic organization. Variation in site size and the presence of integrative features should therefore be monitored with regard to basalt acquisition and differential consumption of all raw materials. Sites Sampled in the Silver Creek Region The Silver Creek area was the subject of several early survey and excavation expeditions at the beginning of this century (Fewkes 1904:136-167; Haury and Hargrave 1931; Hough 1901:297-302; Spier 1918:358-362, 1919). After a long hiatus in research, the area was again the focus of study for cultural resource management purposes over the last three decades. Research based on these collections has been the subject of socio-political models for the Little Colorado drainage (Lightfoot 1981, 1984). The arrival of the University of Arizona Field School in 1993, marked the advent of a new series of researchoriented surface collections and excavations in the Silver Creek area. During the 1993 and 1994 field seasons, the University of Arizona's Silver Creek Archaeological Research Project (SCARP) surface collected 20 sites. Located in Sitgreaves National Forest, the area has enjoyed numerous 100% cultural resource surveys. During the 1993 field season, SCARP relocated and collected a rarge of sites representing a broad array of site sizes and time periods, all of which minimally displayed masonry architecture and Pueblo period ceramics 42 (Mills et al. 1993:25-34). During the 1994 field season, the survey teams targeted sites with great kivas and sites of the Carrizo phase, in an effort to assess the comparability between previous archaeological surveyors with respect to site dating, location, and environment (Mills et al. 1994:35-43). Of particular interest was the re-location of Carrizo phase sites with agave, and the role that this potential cultivar played above the Mogollon Rim. A summary of site characteristics which are of interest to this research is presented in table 2.3 below. Those sites marked with an asterisk (*) were the subject of point-plotted collections. All other sites were surface collected using a strategy of one-by-one meter grid squares laid across artifact-dense regions of the sites. While most of these grid squares were 100% collected, a few of the high artifact density sites were the subject of 50% grid square collections. For the precise collection strategy at each site, see Mills et al. (1993, 1994). As a final note, AZ:P:12:227[ASM], marked with a plus sign (+), is notable for a lack of rooms and the presence of agricultural terracing. Table 2.3; Surface Collected Sites and Characteristics of Interest Site Number Elevation Site Integrative Agave Dating Distance Area n (feet) Size Features (AD) to Collected rm. # Basalt (m2) AZ:P:11:1ASM 6760 200 Plazas N 1275-1350 32.00 218 1516 AZ:P:12:65ASM 6460 10 none N 1100-1200 5.24 91 183 AZ:P:12:167ASM 6300 none 2 N 1000-1150 9.29 1092 702 AZ:P:12:196ASM 6470 5 none N 1000-1100 9.52 39 407 AZ:P:12:210ASM 6600 3 none N 800-1050 8.33 *3185 25 AZ:P:12:211ASM 6520 2 none N 800-1050 7.62 127 681 AZ:P:12:222ASM 6600 none 2 N 750-1000 8.93 105 42 AZ:P:12:227ASM 6640 +0 none Y 750-1000 483 264 8.57 AZ:P:12:242ASM 6560 none 3 Y 750-1100 9.52 110 436 AZ:P:12:248ASM 6675 none 15 N 750-1000 5.95 204 847 AZ:P:12:256ASM 6640 3 none N 750-1000 5.71 68 209 AZ:P:12:260ASM 6720 7 none N 850-1100 8.09 61 290 AZ:P:12:277ASM 6280 Kiva 14 N 900-1150 6.90 336 502 AZ:P:16:90ASM 6420 15 Kiva N 1100-1200 231 105 4.76 AZ:P:16:112ASM 6550 15 Kiva N 800-900 648 849 4.76 AZ:P:16:139ASM 6500 none 3 1050-1200 Y 5.00 80 195 AZ:P:16:153ASM 6560 Kiva 14 N 900-1100 3.57 1494 186 AZ:P:16:160ASM 6540 Kiva 10 N 850-950 85 86 4.29 AZ:P:16:176ASM 6530 5 Kiva 33 N 850-950 3.81 *3050 AR-05-1201USFS 6680 none 2 1100-1175 25 714 Y 27.00 * Site collected with point-proveniencing 44 The site characteristics listed in the table will be discussed in the analysis which follows, but a consideration of the formation processes that these collections passed through should first be addressed. The disposal of lithic debris in ethnoarchaeological contexts is quite consistent. After the initial reduction and shaping of tools, the large and sharp debris created is carefully removed from the living area of the site to a refuse pile. This is the case for both modern Ethiopian knappers (J. Gallagher 1977:410-411) and Mayan knappers (Clark 1991; Hayden and Cannon 1983; Hayden and Nelson 1981). If this pattern can be extrapolated to prehistory, most lithic debris and discarded tools should be placed in areas defined for the disposal of sharp and dangerous waste (Schiffer 1987:68-70). Subsequent trampling and down-slope movements further modify the assemblage into a secondary refuse context. While very small, hard-to remove lithic debris may not reach a refuse area, accumulations of concentrated lithic debris are expected to develop in sedentary communities. One experimental study (Gifford-Gonzalez et al. 1985) monitored the effect of trampling on obsidian debris in both sand and loam substrates. As the Silver Creek area soils are more similar to loam than sand, I focus upon the results from the trampling experiment performed on the loamy substrate. After 2 hours of continuous trampling, the lithic debris (n=641) of varying sizes migrated ve»y little vertically, with only 10 pieces passing below 2 cm in depth. Considering the fragility of the obsidian, veiy little edge damage resulted from the trampling, with only 14 pieces displaying breakage. Hence the effect of human or animal trampling on assemblages in such soil conditions would appear to be minimal; however, the effect of freezing and thawing in the soils of the region and its effect on lithic assemblages is unknown and would make an interesting future study. Artifact collectors have had a profound impact on the formal stone tools of the Little Colorado drainage, making any data comparisons of surface collected formal stone tools suspect (Francis 1978). Debitage from surface collections should be much less biased than formal tools when studying inter-site chipped stone variation. The analysis of debitage from these assemblages appears promising. If chipped stone refuse was deposited at a fixed location away from habitation, as suggested ethnoarchaeologically, a large amount of the systemic assemblage should have been placed in this context. Subsequent trampling by people or animals would not overly disperse the artifacts from their primary refuse position. The effects of deposition, slope, freezing and thawing is unfortunately unknown at this time. The strategy of placing collection units atop dense artifact areas (Mills et al. 1993:25-34; Mills et al. 1994:34-43) should have targeted the bulk of the secondary refuse that appeared on the surface. With an emphasis on debitage, rather than formal stone tools, recent collection by modern people should not unduly effect conclusions. 46 Conclusion After covering the ecological, geologic, and social context of the chipped stone assemblages included in this analysis, and after listing those site characteristics which will have some bearing upon the analysis of technology, the analysis methodology can now be discussed. 47 CHAPTER III: RESEARCH METHODOLOGY AND DESIGN Analytic Factors, Levels, and Research Expectations Large and complex data sets are difficult to work with, because their analysis involves the resolution of several problems at once. An effective strategy, of course, is not to attack them simultaneously, but to formulate a hierarchy of questions and work through them individually [Odell 1989b:176]. Before discussing an analytic methodology, the first question considered is the nature of the inquiiy. Exploring the variation in Pueblo period lithic technology is somewhat new, so a great deal of the Silver Creek analysis is exploratoiy in nature, assessing whatever recurring patterns in space, time and technology may present themselves. Previous large regional studies have approached the exploration of chipped stone assemblages by simply putting data into a multivariate statistical pattern and creating data clusters derived from attributes such as completeness (Sullivan and Rozen 1985) or edge measurements (Barton 1988:56-60). Explanation and interpretation follows from the identification of clusters. While this may provide a useful exploratory technique in the first stage, simply arguing the reality of clusters as behaviorally meaningful units post hoc leaves something to be desired. Proceeding from an a priori stance, the use of existing site characteristics can serve in the creation of research expectations. 48 As surveyed in chapter one, any chipped stone assemblage is the result of a variety of factors. When observing large multi-site assemblages, exploring differences in technology demands the minimization of some variables in an effort to explore patterns in a particular variable of interest. Borrowing terminology from experimental design (Kuehl 1994:6-7), a factorial treatment design can serve to clarify analytic considerations. In such a design (Kuehl 1994:6-14), the analyst proceeds by creating groups of maximum homogeneity within observational units, such that differences between the units can be highlighted. The identification of factors of interest is the first step, and many of the factors in this analysis follow from pre-existing site characteristics. The second step is the identification of meaningful levels within the factors which establishes comparative groups of interest. If this were an experiment in flintknapping, a factor would be raw material, while the "levels" of the factor would be the different raw materials themselves. Obviously the use of the word level in this instance does not necessarily imply a variation in magnitude, although that is one possibility. In the section below, I outline the factors and levels of interest in the analysis which follows (table 3.1). 49 Table 3.1: Summary of Research Factors of Interest Raw Material Level Factors of Interest Consumer Distance to Great Agave Group Basalt Kiva Room #'s (km) Basalt 0 3.57-5.57 Y Y Diabase 2- 5 6.57-9.57 N N Chert 6 - 11 27.0 Plaza S-Wood 12 - 26 32.0 Quartzite 200 Ecological Context (Altitude) 40 foot intervals (6280-6760) The raw material factor is the first concern, as the properties of these materials shape the knapping qualities of the stone as well as breakage, failure and the amount of debris produced during manufacture. The levels in this factor are the different raw materials themselves, and all subsequent data analysis proceeds by addressing each material class separately. The elevation factor serves as proxy for the ecological variation discussed in Chapter Two: Ecological Context. Ecological variation correlates with different subsistence opportunities, animal and plant communities. Further, elevational differences effect climatic patterning which effects needs for fuel, clothing, and shelter. The most difficult task involved is the division of an ecological continuum of change into discrete and useful units. To establish meaningful levels for this analysis, the sites were grouped into 10 elevational units of 40 feet 50 contour intervals. Ecological variation within each level is assumed to be minimal, while technological contrasts between them (after considering other impacting variables) are examined in light of different ecological contexts. The agave factor is also of interest to subsistence adaptation and its impact on technology. Agave above the Mogollon Rim exists outside of its natural range and is thought to have been deliberately transplanted by prehistoric peoples (Kunen 1994; Minnis and Plog 1976). Agave harvesting and processing is found in association with a particular lithic tool kit, as derived from ethnographic (M. Gallagher 1977:46-52) and archaeological evidence (Fish and Donaldson 1991:269; Fish et al. 1985; Van Buren et al. 1992). The two levels in this factor are lithic collections associated with agave sites and those which are not. Distance to source beds for all material types except basalt are minimal (see Chapter Two: Geologic Context). As basalt does occur at all sites and its reduction may be effected by distance-to-source constraints, distance from basalt source material is a factor of interest. To identify levels for this factor, a histogram of basalt flake frequency with respect to distance from source (km) was constructed (figure 3.1). At increments of 1 km, four modes are observed. For this factor four levels are established, creating one level near to source (3.57-5.57 km), a moderate distance from source (6.57-9.57 km), and far from source (27 km and 32 km). These levels will be considered as basalt is monitored across the landscape. 51 Etequaacy of Basalt Flaks by Distance from Source 0.1 10 T 3.67 10.57 ~r 17.67 T 24.57 31.67 Distance from Basalt Sources (km) ! 'igure 3.1: Frequency of Basalt Flakes by Distance from Source The factor of site size pertains to social organization and the size of consuming populations, which in turn impacts chipped stone assemblages. For the Silver Creek area, site size levels have been previously established based upon the frequency of particular site size occurrences (Mills and Newcomb 1994). The five levels in this factor are sites with no rooms, 2-5 rooms, 7-12 rooms, 14-26 rooms, and 200 rooms. The factor of integrative architecture should be considered, as such sites may indicate regional exchange networks that could influence lithic availability 52 and/or the transport of certain tool blanks across the landscape. The presence or absence of great kivas defines two levels, and one site containing plazas and kivas exists as its own level. These factors serve to shape expectations and frame questions as the lithic variation between sites is explored. The goal of analytic factors and their associated levels is to help organize observations and minimize differences within levels, so the contribution of particular factors to lithic variation can be highlighted. Statistical Philosophy This analysis proceeds as essentially exploratory; therefore, the statistical procedures will be primarily descriptive with a minimum of formal hypothesis testing (following Fish 1979:57-71). I avoid most multivariate techniques in favor of bivariate and exploratoiy statistics, as the more complex techniques do not appear warranted given the nature of the study (following Sackett 1989:58-61; Thomas 1978). Attributes Recorded Material Types. While the Silver Creek Archaeological Research Project (SCARP) records a wide array of material types, in the analysis that follows only five were focused upon, due to the well-established nature of their source 53 location, well-known flaking properties, and their unambiguous identification. Each of these substances has different chipping qualities and edge-wear characteristics. The easiest to knap are those with the sharpest and most easily damaged edges, these include silicified wood and chert. Harder igneous materials are represented by basalt and diabase, while the hardest and densest substance is quartzite (Whittaker 1994:66). Diabase, silicified wood, and quartzite originate in the rim gravel province, chert from either the rim gravels or the Kaibab limestone, and basalt from source beds east of Show Low. The frequencies of these materials in the collection are presented in table 3.2. Table 3.2: Frequency of Selected Material Types Basalt 357 Diabase 299 S-Wood 1740 Chert/Jasper 3479 Quartzite 844 Total 6719 Technological and Functional Types. Each artifact was coded with respect to its technological type, which included shatter, cores, flakes, tested cobbles, etc. Functional types included used cores, used flakes, scrapers, drills, projectile points, etc. Cortex. The amount of cortical material was coded by a series of categories: 0% cortical cover, 1-10%, 11-50%, 51-90%, and 91-100%. Using such variables, the earliest stages of lithic reduction can be monitored, and hence the 54 degree of reduction intensity (Crabtree 1982; Mauldin and Amick 1989; Sullivan and Rozen 1985). Platforms. Flaking platforms were recorded as platforms with cortex, without cortex and one scar, dihedral, or as having more than two flake scars. Use Wear. The presence or absence of use wear was recorded, but given the expedient nature of the technology veiy little wear could be expected. Completeness. The completeness of debitage was identified using Sullivan and Rozen's (1985:756-760; Rozen and Sullivan 1989b:180-181) debitage categories (complete flake, broken flake, flake fragment, debris, other). As their scheme has been applied to other Southwestern assemblages, using it facilitates comparability among analysts. Additionally, I feel that their debitage categories provide useful exploratory information for large data sets. These categories have been the focus of some debate in recent years (Amick and Mauldin 1989; Ensor and Roemer 1989; Prentiss and Romanski 1989; Rozen and Sullivan 1989a, 1989b; Shott 1994). The central issue of concern has been correlating certain reduction modes (e.g., core versus bifacial reduction) with debitage patterns. These arguments often emphasize experimental studies, which should be taken with some caution. Ensor and Romanski (1989) critique Sullivan and Rozen's interpretations concerning breakage frequency and flake distributions as characteristic of particular production modes, noting that they lack any basis in 55 experimental studies. Similar criticisms come from Amick and Mauldin (1989). Illustrations of the differences between experiment and reality are in order. Prentiss and Romanski (1989) conducted experimental bifacial and core reduction, and placed the resulting debris into a sandy substrate for subsequent trampling. Rather than allowing for natural conditions in the sandy soil, the researchers placed a plastic tarp 5 cm below the surface. At the end of the experiment, they noticed significant changes in the number of broken flakes. This result challenges Sullivan and Rozen's interpretations regarding reduction mode as monitored by completeness. The zone of turbation in this experiment was completely artificial, unlike the Gifford-Gonzalez et al. experiment (1985:814), where a sandy substrate was used without a plastic tarp. Only 5.4% of the artifacts (n=792) displayed edge damage in that study. Similarly Mauldin and Amick (1989) attempt to refute Sullivan and Rozen's flake-completeness to reduction-mode correlations by manufacturing three bifacially reduced cores and catching the debris produced on a plastic tarp for analysis. This experiment did not include subsequent resharpening of these tools, which would be expected in the curation life of a formal tool, nor any contrasting experiment with simple flake tools and their resulting debris. Many experiments do not address curatorial behavior and post-depositional processes. The proper role of completeness seems to lie in its most straight-forward predictions. Ingbar et al. (1989) found that by analyzing the percentages of 56 broken and complete flakes, bifacial reduction could be successfully distinguished from core reduction. Patterson (1981, 1990) finds that by monitoring the frequency of particular flake sizes, bifacial reduction can be identified experimentally and archaeologically. Mauldin (1993b:41-46) tentatively agrees that Sullivan and Rozen's debitage categories can distinguish the presence of bifacial core reduction from other reduction patterns. He uses their technique as the starting point in assessing assemblages along the Arizona Interconnection Project Transmission Line Corridor, and then proceeds to explore his data via other variables. This seems to be the best course to take, especially with large data sets. The relative percentages of complete flakes, broken flakes, and shatter can quickly explore the nature of assemblages with respect to manufacture and reduction intensity, and where patterns of interest are detected, further testing can proceed through other variables. In a slight modification of the Sullivan and Rozen (1985) debitage scheme, initial explorations of the SCARP database were conducted to examine the frequency of the four debitage categories along with the frequencies of tested cobbles and cores. By using those six mutually exclusive categories, all non-tool artifacts in the SCARP assemblages were exhaustively characterized. These combined variables served to quickly characterize the collections belonging to factors and associated levels of interest. Through the use of bivariate tables and 57 simple chi-square tests, initial similarities and differences were detected which directed further investigation through other variables. Metric variables. The length of flakes was recorded along the axis of applied force. Width was measured perpendicular to the wave of force at the widest point Thickness was measured perpendicular to the plane defined by the length and width measurements at the thickest point of the flake. All these measurements were recorded to the nearest 0.01 mm. For all other artifacts (e.g., cores, shatter, etc.), length was measured along the longest dimension, width was measured perpendicular to the long axis at the widest point, and thickness was measured perpendicular to the plane defined by the length and width measurements at the thickest point of the artifact. All measurements were recorded to the nearest 0.01 mm. All artifacts were weighed to the nearest 0.1 grams. (For the complete chipped stone code book used in the analysis, see Appendix A). 58 CHAPTER IV: EXPEDIENT TECHNOLOGY IN VARIABLE ECOLOGICAL SETTINGS In this chapter the following hypotheses are explored: 1) The general functional properties of expedient tools, as monitored by cutting edge, will vary with ecological setting, considered to be the result of different resource extraction demands at different elevations. 2) The cultivation of agave requires the use of hard, durable edged tools such that basalt will play an unique role in the technology of agave sites. As chipped stone technology is intimately tied to resource extraction, differing ecological contexts should result in technologically different assemblages (Binford and Binford 1969:84; Collins 1975:15-17). Ecologically patterned differences between Pueblo communities may occur in material type selection and the length of flakes manufactured, as these flakes are created to serve specific tasks in a specific environment. In the Silver Creek Archaeological Research Project (SCARP) study area, temperature and precipitation vary by elevation and the placement of topographic features. For the purposes of this study, the altitude of sites serves as proxy for changing ecologic settings. Elevational differences reflect a continuum of ecological change in plant and animal communities (Satterthwait 1976; USDA Forest Service 1973:7-10). As documented in modern times, these elevational differences present Upper Little 59 Colorado drainage residents with elevational-dependent subsistence/resource opportunities and problems (Abruzzi 1993:81-116). Pueblo peoples employed mixed economies of hunting, gathering, wild and domestic plant cultivation (Plog 1981:30-41). Given that this economic variety was situated in differing ecological communities, resource extraction strategies should vary with frost-free season, wild floral and faunal communities, amount of ground water, and choice of cultivars. Exclusive reliance upon intensive agriculture in early Pueblo periods (encompassing all sites in the sample except AZ:P:11:1[ASM]) has not been indicated for adjacent regions such as western New Mexico (Mauldin 1993a) and east-central Arizona (Reid 1989:71-80). Differential agricultural success is expected across elevational communities, noting that higher elevations will have greater moisture, but also a greater risk of killing frosts. Advantageous "mixes" of resource exploitation at higher elevation sites is likely to include more wild floral and faunal resources. One unique subsistence choice is the use of Agave parryi, which is found in association with archaeological sites above the Mogollon Rim (Minnis and Plog 1976). This is at the edge of the species' natural distribution (Gentry 1982:539) and is argued to be deliberately transplanted by site occupants. The precise role of agave in diet and subsistence technology above the rim is contested. Comparative data on agave exploitation comes from ethnographic information (M. Gallagher 1977:46-52), historical records (Dobyns 1988), and from arid basin archaeological research (Bohrer 1980:228-230; Fish and Donaldson 1991:269-272; Fish et al. 1985; Gasser and Kwiatkowski 1991; Hodgson et al. 1989; Van Buren et al. 1989). These researchers laud agave as providing a storable source of sugar and fiber, and a harvestable resource during early spring, a particularly harsh time in the subsistence year (Hodgson et al. 1989). Agave researchers have observed a particular prehistoric tool kit that consists of large quantities of heavy, durable stone tools made from igneous and quartzitic materials (Van Buren et al. 1989). Also uncovered in desert contexts are formal agave knives, none of which have been found in the SCARP collections. Agricultural features, such as rock walls where agave plants can flourish, have also been noted. In the sections that follow, chipped stone patterns which appear linked to elevational and hence ecological differences will be explored, as will the impact of agave on the SCARP area sample. 61 Elevational DifTerences in Chipped Stone Assemblages Table 4.1: Elevation Factor: Sites in Elevation Levels Elevation Level Elevation Interval (feet above sea level) Site Numbers A 6281-6320 AZ:P:12:167[ASM] AZ:P:12:277[ASM] no assemblages for these intervals B 6401-6440 AZ:P:16:90[ASM] C 6441-6480 AZ:P:12:65[ASM] AZ:P:12:196[ASM] D 6481-6520 AZ:P:12:211[ASM] AZ:P:16139[ASM] E 6521-6560 AZ:P:12:242[ASM] AZ:P:16:112[ASM] AZ:P:16:153[ASM] AZ:P:16:160[ASM] AZ:P:16:176[ASM] F 6561-6600 AZ:P:12:210[ASM] AZ:P:12:222[ASM] G 6601-6640 AZ:P:12:227[ASM] AZ:P:12:256[ASM] H 6641-6680 AZ:P:12:248[ASM] AR-05-1201[USFS] I 6681-6720 AZ:P:12:260[ASM] J 6721-6780 AZ:P:11:1[ASM] 62 Exploratory investigations proceeded by using the debitage categories of Sullivan and Rozen (1985; Rozen and Sullivan 1989b). Within each material type, the frequency of debitage categories was monitored across several factors of interest. In the case of the elevation factor (table 4.1), differences in debitage category frequency were assessed with a chi-square test of population homogeneity, revealing statistically significant differences (a=0.05) between elevation levels (table 4.2) for all materials classes except basalt. Table 4.2: Pearson chi-square test of homogeneity: Probability of Similar Debitage Category Frequencies Between Elevational Levels Material Type Probability Value Sample Size Basalt * 0.170 357 Diabase * 0.003 297 Silicified Wood 0.000 1736 Chert 0.000 3457 Quartzite 0.000 843 * Cell counts low (more than 1/5 of the cells contain counts < 5), results should be taken with caution ** frequencies of artifacts reported in Appendix B, Tables B.1-B.5 Further testing demonstrates that within elevation level (i.e., within those levels encompassing multiple sites), only a few demonstrate statistically significant differences. Debitage categories appear to be internally homogenous for the majority of levels tested (table 4.3). 63 Table 4.3: Pearson chi-square test of homogeneity: Probability of Similar Debitage Category Frequencies Between Sites and within Elevational Levels Diabase 0.714* S-Wood 0.184* B (1 site) auto C (2 sites) 0.588* 0.249* 0.263* 0.532 0.080* D (2 sites) 0.715* 0.046* 0.048* 0.102 0.197* E (5 sites) 0.014* 0.460* 0.000* 0.469* F (2 sites) auto 0.073* 0.050* 0.290* G (2 sites) 0.004* 0.427* 0.018* 0.007 0.005* H (2 sites) 0.509* 0.416* 0.308 0.137 0.686 N/A 0.017 Quartzite A (2 sites) auto 0.838 Chert 00* Basalt © Elevation Levels (# of sites in level) auto auto 0.472* auto I (1 site) auto auto auto auto auto J (1 sites) auto auto auto auto auto # of internally homogenous elevation levels out of nonautocorrelated levels (a=0.05) 4 of 6 5 of 6 4 of 7 5 of 7 6 of 7 * Cell counts low (more than 1/5 of the cells contain counts < 5), results should be taken with caution "auto": with only one site in test, results are auto-correlated "N/A": no artifacts observed ** Frequencies of artifacts reported in Appendix B, Table B.6-B.10 Concerns are justifiably raised when examining a table of frequencies such as the ones presented above. By simple laws of probability, multiple significance 64 tests reported upon en masse will demonstrate higher error rates. This concern is usually met by raising the level of a, to reduce overall experiment-wise error rates (Kuehl 1994:101-102). In this instance, I have not chosen to do this for reasons which are discussed at greater length in Appendix C, Part A: Why these p-Levels? The function of the Pearson chi-square is simply to identify potentially interesting similarities and differences in a factor of interest, in this instance elevation. This initial exploration suggests potential similarities in chipped stone by elevation level. A test for homogeneity can only indicate the simplest similar versus different dichotomy, but these first observations encourage further testing through other variables by elevational differences. The similarity of basalt debitage categories across elevation levels indicates that this factor is not a powerful tool for exploring that material type's variation. In a flake-tool technology, material type and flake cutting edge, as monitored by complete flake length, carries the functional "signal" of the artifact (see Chapter 1). Length of complete flakes across elevation levels is next explored. Figures 4.1-4.4 are box plots of complete flake lengths by elevation levels. The median values are recorded to the side of each box plot. None of the distributions are normally distributed, as determined by the inspection of histograms from each site and each elevation level (not presented here). This fact makes mean flake length an unreliable statistic for summarizing central tendencies. 65 Box Plot of Complete Diabase Flake Lengths Elevation Factor Medians 17.0 40.5 d) 3) c o» cd > CD 0) H *34.5 G 41.6 26.0 D 35.0 LU - 10.8 € B N/A 29.6 J 0 20 40 60 80 100 Length (mm) Figure 4.1: Box Plot Elevation Factor of Complete Diabase Flake Lengths, 66 Box Plot of Complete S-Wood Flake Lengths Elevation Factor Medians 24 30 * H a5 Q 8 25 * G 29 / m 32 <3 E 25 UJ D 26 21 B 23 A" 21 0 ± ± 10 20 J 30 40 50 60 70 Length (mm) Figure 4.2: Box Plot of Complete Lengths, Elevation Factor Silicified Wood Flake 67 Box Plot of Complete Chert Flake Lengths Elevation Factor Medians 21 31 H CD ** 25 G •i r F -CD 29 3? c o -*—» CO > CD LU 27 * E 25 D 25 C 16 B 19 O A" 20 I 0 10 20 30 40 50 60 70 80 Length (mm) Figure 4.3: Box Plot Elevation Factor of Complete Chert Flake Lengths, 68 Box Plot of Complete Quartzite Flake Lengths Elevation Factor Medians 29 41 © © H 29 ~ G 40 c o cO > 6 29 D LU - h; 28 C -a>/ to B A" 14 25 21 - I o 50 100 150 Length (mm) Figure 4.4: Box Plot of Complete Quartzite Flake Lengths, Elevation Factor 69 These box plots depict a progression of median flake lengths from shorter to longer as elevation increases. A 1 cm shift in the median is observed from the lowest elevation level (level A) to the second highest level (level I) for diabase, silicified wood and chert. In the case of quartzite, the median shift is 2 cm. Two exceptions are noticed. The highest elevational level (J) consists of 1 site AZ:P:11:1[ASM], encompassing some 200 rooms. This site size is completely out of proportion to the rest of the assemblage, all other sites being 15 rooms or less. In chapter 5, the variation in flake length as it relates to site size will be explored in more detail. While anticipating later discussion, a negative correlation exists between site size and the length of complete flakes, which in the case of AZ:P:11:1[ASM] would suggest that population size is a confounding variable. Elevation level H also consistently demonstrates lower median lengths compared to immediately adjacent lower elevations, but remains higher than the median lengths of the lowest elevation levels. These patterns are argued to be behaviorally meaningful based on several points. First the patterns appear in all materia] classes except basalt, which implies that the change in length is not a fluke or a particular characteristic of a given material type. Further the box plots reveal distributional shifts in length values at higher elevation levels (i.e., the whole distribution of flake length shifts to larger values with elevation, such that changes in median values are not merely an artifact of influential outliers). A direct test of these relationships can be carried out with the nonparametric Spearman's correlation coefficient, which ranks the values of several variables and analyzes the relationship. Significance testing of Spearman's correlation coefficients requires the use of several tables (Sokal and Rohlf 1981:584-585, 607), and is not performed by most statistical packages (see Appendix C, Part B: Significance Testing of Spearman's rho). The correlation's significance is directly tied to the size of the sample being tested. Consequently very low correlation coefficients are needed to satisfy significance for large samples and conversely. The increasing length of cutting edge in different ecological settings as monitored by complete flake length and site elevation is tested below (table 4.4). 71 Table 4.4: Spearman's Rank Correlation Coefficients Complete Flake Length by Elevation Material Type Sample Size Correlation Coefficient Significance Test: Accept or Reject (a=0.05)* Basalt 82 -0.161 accept Diabase 91 -0.010 accept Silicified Wood 453 +0.099 reject Chert 736 +0.043 accept Quartzite 231 +0.216 reject *H0: No relationship between variables Monitoring only complete flakes ignores a large portion of the assemblage, namely broken flakes. Complete flake length is a poorly preserved metric variable, as natural use will result in the breakage of flakes. As metric variables correlate quite highly with one another, the use of a conserved dimension such as thickness or width can serve as proxy to flake length. For the SCARP assemblages, the most highly correlated variable to complete flake length is thickness, as monitored through Pearson's correlation coefficient. In the case of complete silicified wood flakes, the correlation is 0.551. Similar correlations are found in the other material classes. I take these results with some concern, as this correlation indicates that roughly 30% of the variation in complete flake length can be accounted for with variation in thickness. In table 4.5, the database 72 is expanded by including flakes with complete length, complete width or both, and the relationship between thickness and elevation of sites assessed. Table 4.5: Spearman's Rank Correlation Coefficients Thickness of Flakes with One Complete Dimension by Elevation Material Type Sample Size Correlation Coefficient Significance Test: Accept or Reject (a=0.05)* Basalt 119 -0.061 accept Diabase 129 +0.042 accept Silicified Wood 649 +0.084 reject Chert 1155 +0.056 reject Quartzite 342 +0.186 reject *Ho: No relationship between variables The results indicated by the earlier test of homogeneity are mirrored in significant complete flake length differences. Basalt and diabase flake lengths and thicknesses are not significantly correlated with site elevation. Silicified wood and quartzite are significantly positively correlated with elevation for complete flake length, and along with chert are significantly correlated for thickness. Finally the attribute of interest is cutting edge, to which a length measurement of the longest axis serves only as proxy. Length is a one dimensional measure of a two dimensional edge. Rarely is cutting edge calculated due to the time, effort, and measurement ambiguity it represents, in spite of the fact that cutting edge in an 73 expedient technology is one of the most critical variables to monitor. Sheets and Muto (1972:633) estimated total cutting edge in a blade production experiment by doubling the length measure. Using median flake lengths only, doubling the length measure and comparing the lowest sites (level A) to the highest comparable sites (level I, excluding AZ:P:11:1[ASM]) yields a 30% increase in the cutting edge of silicified wood, 35% in chert, and 49% in quartzite. To develop this relationship directly and without the use of the Spearman's rankings, the logical tool is the parametric Pearson correlation coefficient, regressing cutting edge on site elevation. Unfortunately, a linear model would not be appropriate as applied to the data in its current form. The predicted values (i.e., flake lengths) need to be normally distributed for an assessment of the correlation to be valid. Further, many of the smaller complete flakes were probably not used as tools, and hence their cutting edge potential is of no particular functional interest. Deciding which lengths would make usable tools would require very arbitrary decisions with questionable results. To address the problems of non-normality and usable flake lengths, the central tendency of each site's complete flake length distribution was calculated for each material class with medians, and 90% confidence intervals were constructed about these median values. This yielded small normal distributions for each site, as confirmed by tests of normality. Using these trimmed distributions, correlations between elevation and the central tendency of flake lengths can be assessed. Figure 74 4.5 plots complete median flake length distributions against elevation, as well as reporting Pearson's correlation coefficient (r), the coefficient of determination (r2), the f-ratio probability value to assess the significance of correlation (p), and sample size (n). AZ:P:11:1[ASM], due to its non-comparable site size, was not included. No patterning was observed in the analysis of the residuals. 75 Correlations Between Median (90% C.I.) Complete Flake Lengths and Elevation SiabM* Fltkii 100 6200 6300 6400 6600 6600 6700 6800 Klevatlon (feet) x-0.177 1^*0.031 p-C.1*7 n»57 60 Chart Flake* eo 60 Sllioifled wood Flakes 16200 8300 6400 6600 6600 6700 6800 SlevaEion (feet) r-0.442 r2-0.195 p-0.000 n-13* 00 Ouartxite Flake* 60 0200 6300 6400 6000 6600 6700 6800 Elevation (£eet) z-0.400 r2-0.lt p-0.000 n-lC7 6200 6900 6400 6600 6600 6700 6800 Slewation (feet) L-0.44C r2-0.19S p-0.000 n-9» Figure 4.5: Correlations between median (90% C.I.) Complete Flake Lengths and Elevation 76 Statistically significant relationships (a=0.05) exist between elevation and flake length for silicified wood, chert, and quartzite flakes. While a positive correlation is apparent for diabase, the result is not significant. For the others, 19.5% of the variation in silicified wood flake length can be accounted for with elevational differences, 16% for chert, and 20% for quartzite. While obviously elevation is not the whole story of increasing central tendencies in these distributions, it does account for a portion of it. The hypothesis entertained to explain this pattern is that ecological communities at higher elevations are lusher, wetter, and colder. The uncertainty of agriculture at higher elevation forces higher altitude communities to rely on other strategies, which likely include increased wild resource processing in plants and animals relative to lower altitude peoples. Additionally, construction material and wood fuel at higher elevations must be extracted from stands of ponderosa pine, oak, and other hardwoods, trees that are thicker and more difficult to cut through than the pinon and juniper species of lower elevations. Hide processing of more wild game and processing larger arboreal resources would be facilitated by longer-edged scraping tools and heavier cutting implements. The technological demands placed upon tools of different elevations vary in such a way that longer edges are more efficacious in higher elevation communities. In the absence of excavated floral and fannal materials to document differing resource exploitation at sites of different elevations, this hypothesis must remain speculative. Other hypotheses have been 77 entertained, but on the whole have been found unsatisfactory given the data available. One alternate hypothesis is that longer flakes correlate with higher elevations because larger cores are found at higher elevations. The higher elevation sites in the study area are closer to the rim gravel province, which contains quartzite, silicified wood, diabase and chert nodules (compare table 4.1 with figure 2.5). The nodules in this province are quite large and were laid down during the original formation of the gravels (Peirce et al. 1979). Communities at lower elevations would have to transport nodules from the rim gravel province or acquire nodules moved by washes, resulting in the size sorting of smaller nodules at lower elevations. This effect could result in the flake length correlations found above, assessing the explanatory power of this supposition requires testing via other variables. The most direct test is to assess the size of tested cobbles, to see if smaller size cobbles are more prevalent at lower elevations. The Spearman's correlation coefficient was used to compare site elevation with tested cobble weights in each material class (table 4.6). 78 Table 4.6: Spearman's Rank Correlation Coefficients Tested Cobble Weight By Elevation Materia] Type Correlation Coefficient Sample Size Significance Test: Accept or Reject (a=0.05)* Diabase 5 +0.224 accept Silicified Wood 28 +0.092 accept Chert 117 +0.066 accept Quartzite 31 +0.030 accept •"Hq: No relationship between variables No material class demonstrates statistically significant (a=0.05) correlations between tested cobbles weights and elevation. The size of raw material nodules can also be estimated by the size of cores, as measured by their weight. Table 4.7 assesses core weight differences between elevation levels. Table 4.7: Spearman's Rank Correlation Coefficients, Core Weight by Elevation Material Type Sample Size Correlation Coefficient Significance Test: Accept or Reject (a=0.05)* Diabase 28 +0.151 accept Silicified Wood 216 -0.038 accept Chert 304 +0.042 accept Quartzite 104 +0.005 accept •H,,: No relationship between variables Once again, no statistically significant correlations exist between site elevation and core weight. As a final attempt to support this alternative hypothesis, cortical flakes can be examined. Higher amounts of cortical coverage are expected at higher elevations as a function of site-proximity to rim gravel sources. First the frequencies of flakes in all cortical classes were tested with a chi-square test of homogeneity (table 4.8). Table 4.8: Pearson chi-square test of homogeneity: Probability of Similar Cortical Flake Type Frequencies Between Elevational Levels Material Type Sample Size Probability Value Diabase 143 0.362 Silicified Wood 724 0.315 Chert 1363 0.005 Quartzite 373 0.210 * Warning more than 1/5 of the fitted cells are sparse (frequency < 5) ** frequency of artifacts reported in Appendix B, Table B.11-B.14) Only chert indicates any significant difference in cortical flake type frequencies at different elevation levels. By selecting complete flakes with a high degree of cortex (cortical coverage > 10%), the length of such flakes may serve to indicate the size of the original nodule, with longer flakes expected to occur at 80 higher elevations (tables 4.9-4.10). As before the length of complete flakes and the thickness of flakes with at least one complete dimension are correlated with elevation. Table 4.9: Spearman's Rank Correlation Coefficients Cortical Flake Length By Elevation (cortex > 10%) Material Type Sample Size Correlation Coefficient Significance Test: Accept or Reject (a=0.05)* Diabase 29 -0.226 accept Silicified Wood 95 -0.036 accept Chert 207 +0.043 accept Quartzite 120 +0.161 reject *Ho: No relationship between variables Table 4.10: Spearman's Rank Correlation Coefficients Thickness of Cortical Flakes with one Complete Dimension by Elevation (cortex > 10%) Material Type Sample Size Correlation Coefficient Significance Test: Accept or Reject (a=0.05)* Diabase 31 -0.245 reject Silicified Wood 123 +0.034 accept Chert 287 +0.021 accept Quartzite 170 +0.119 accept *H0: No relationship between variables 81 The only significant positive correlation was observed in quartzite for complete flake length, but not in thickness. The establishment of positive correlations between cortical flake length and elevation remains ambiguous. Every effort to support the hypothesis that bigger nodules were being used at higher elevations due to proximity to the rim gravels has yielded unsatisfactory results. Consequently source material size differences as a factor influencing longer flake production at higher elevations does not appear to be valid. Rather the initial hypothesis, concerning technological demands brought about by differing ecological settings, remains a more satisfying explanation for cutting edge increases with site elevation for silicified wood and quartzite, and to a lesser extent chert. The favored hypothesis requires further testing through other material classes, particularly floral and faunal information that can only be gained through excavation. In the absence of that, some extant ethnographic and archaeological studies provide some useful references in evaluating the supposition. The use of stone tools to scrape cow hides is documented among three ethnic groups in Ethiopia (J. Gallagher 1977). Of interest to this discussion is the use of high quality silicates (in this case obsidian) to make a scraping tool, which was quickly utilized and discarded. Informants would use four scrapers per hide, resharpening their tools after an average of 100 strokes. Ragged edges which can 82 damage the hides develop very quickly. Consequently the tools have very short use lives and develop little or no use wear. In the Southwest expedient tool case, Chapman's (1977:433-435, 450-451) evaluation of raw material selection and debitage edge-wear analysis indicate that Pueblo peoples of the Lower Chaco preferentially used silicified wood debitage for scraping activities, probably due to the fracturing and edge-properties of that substance. Following from the ethnographic example, expedient flake tools of high quality silicate substances would be discarded fairly quickly during hide processing. For the SCARP area, if higher elevation communities process more hides due to the greater consumption of animals and perhaps the need for more hides in colder contexts, then longer high quality silicate flakes of silicified wood and chert, which would cover more surface area more effectively, may reasonably result. In a more general perspective, economies engaged in greater wild animal resource procurement tend to have longer tools, which seems to be a macro-level trend observed as wild resource economies in North American prehistory shift to increasingly sedentary agricultural economies. In more agriculturally dependent economies, the utilization of smaller local game results in smaller formal and expedient stone tools. This pattern appears in both Southwestern and nonSouthwestern assemblages. For instance increasing use of small-fauna in late prehistoric agricultural economies of northeastern Ohio corresponds with a shift 83 to smaller tools (Brose 1978:90-97), with excavated faunal assemblages documenting the shift in game exploitation. Finally, Hayden (1978:188-193) points out that wood exploitation is the largest consumer of lithic resources, based on his observations of Western Desert aborigines. While sedentary Pueblo peoples obviously are not living an Australian lifestyle, the consumption of wood for fuel, the construction of structures, and the manufacture of wooden tools requires large amounts of chipped stone. Different ecological communities will obviously present different wood procurement problems (Plog 1981:38-41). The fuel packet of a ponderosa pine forest is somewhat difficult to consume, while pinon-juniper forest species are much more readily processed. The requirements of processing higher elevation plant species would more effectively be met with large, heavier stone implements (such as large quartzite pieces), relative to lower elevation processing requirements. In summary, ecological variation presents technologies with different challenges, to which a chipped stone tool assemblage is somewhat sensitive. The logical companion to chipped stone tool studies is in-depth analyses of all resources taken from the environment. In the absence of this information, patterned variation in technology across differing ecological communities can be argued to result from differing exploitation needs, not random human knapping behavior or distance-to-source constraints in the SCARP context. Flake length seems to be a sensitive variable, which is tied to behaviorally meaningful relationships between cutting edge length and resources exploited. While most subsistence information from these sites requires excavation, the presence of agave fields on some sites presents a factor which can guide chipped stone assemblage research. 85 The Agave Factor and its Impact on Chipped Stone Assemblages Four sites in the study area have associated plots of Agave parryi (AZ:P:12:227[ASM]; AZ:P:12:242[ASM]; AZ:P:16:139[ASM]; AR-05-1201[USFS]. As the plant is occurring outside of its natural range, propagates best by cuttings, and is almost always found in the vicinity of human habitation, agave appears to have been deliberately brought into the area for cultivation by prehistoric people (Kunen 1994; Minnis and Plog 1976). Tool assemblages from these sites have not been well-studied for the region. In Hohokam research (Van Buren and Skibo 1992) and ethnographic Apache examples (M. Gallagher 1977), agave processing tools are typically large, heavy implements made from hard durable materials such as igneous substances. Bearing that in mind, the investigation of basalt for the area provides some interesting patterning between agave and non-agave sites. As the agave sites targeted by SCARP fall in the Carrizo phase with small site sizes (Mills et al. 1994:35), the larger and later site AZ:P:11:1[ASM] will not be included in the statistical comparisons that follow. Preliminary investigations pursued contrasts in debitage frequency for all debitage types (i.e., tested cobbles, cores, and Sullivan and Rozen's  debitage categories). Contrasts between agave sites and non-agave sites in a test for homogeneity yielded the following results (table 4.11): 86 Table 4.11: Pearson chi-square test of homogeneity: Probability of Similar Frequencies of All Debitage Between Agave and Non-Agave Sites Material Type Sample Size Chi-square probability Basalt* 330 0.041 Diabase* 262 0.227 Silicified Wood 1394 0.065 Chert 2753 0.089 Quartzite 780 0.331 * Cell counts low (more than 1/5 of the cells contain counts < 5), results should be taken with caution ** frequencies of artifacts reported in Appendix B, Tables B.15-B.19 At the a=0.05 significance level, only basalt reflects differences between agave and non-agave sites. Examining the frequencies of basalt debitage between agave and non-agave sites, homogenous results are observed, indicating internal similarities in basalt debitage frequencies between these sites (table 4.12). 87 Table 4.12: Pearson chi-square test of homogeneity: Probability of Similar Frequencies of All Basalt Debitage among Agave Sites and among Non-Agave Sites Sample Size Level Chi-square probability among Agave sites * 54 0.194 among Non-Agave sites * 276 0.139 * Cell counts low (more than 1/5 of the cells contain counts < 5), results should be taken with caution ** artifact frequencies recorded in Appendix B, Table B.20-B.21 The nature of the dichotomy between agave and non-agave site assemblages is next pursued. Histograms were constructed, contrasting the frequencies of basalt debitage types at agave and non-agave sites (figure 4.6). Frequency of Basalt Debitage Agave vs. Non-Agave Sites as i g Agave Sites atM02 - 0.1 - 1 » A 5 « D»bit»9» Type 25 20 16 10 6 OA 06 • 1 (U • fa 03 02 • Q1 - Bon-Agave Sites 1 2 3 4 s « Dafcltag* Typ« Eayi 1-Tasted Cobbles 2Kk>xes 3-Complat« Flakes 4-Broken Flake 5-Flake Fragment 6-Shatter 7igure 4.6: Frequency of Basalt Debitage, Agave vs. Non-Agave Sites 88 Inspecting the histograms reveals differences in the frequency of complete basalt flakes, compared with other debitage types between agave and non-agave sites, although these higher frequencies of complete flakes are not matched by significantly longer cutting edges at agave sites. A Kruskal-Wallis test of variance between agave and non-agave site complete basalt flake lengths (n=75) yielded a probability value of 0.336 (exclusive of AZ:P:11:1[ASM]). A similar test performed on thickness of basalt flakes with at least one complete dimension (n=107) yielded a probability value of 0.796. The frequency of complete basalt flakes at sites with agave is on average 10% higher than those sites lacking agave. With multi-use technology, it is not reasonable to conclude that basalt was only used for agave processing and harvesting tasks. However given archaeological examples of agave tools (Van Buren et al. 1992:92), heavy, durable-edged implements are found preferentially in agave fields. If this is the case above the rim, basalt could have been used preferentially in agave related tasks. Three hypotheses for the debitage frequency pattern can be entertained. First, the chipped stone assemblages collected by SCARP (Mills et al. 1993:28-29; Mills et al. 1994:38-43) were situated in the vicinity of room blocks or terracing features (e.g. AZ:P:12:227[ASM]), not in the agave fields themselves and hence the entire chipped stone profile may not be represented. The preferential processing of agave with basalt tools could result in the discard of broken items in the agave field/processing areas away from 89 the habitation zone, deflating the amount of broken flakes in the samples collected. A simple test of this would involve excavation and collection in the fields themselves, which will have to await future research. The second explanation assumes that the collection contains a representative sample of debitage, with people at agave sites using reduction strategies that maximize complete flakes or who are preferentially bringing whole flake blanks into agave sites. This is similarly testable by collection outside of the habitation areas to confirm the presence of representative samples on-site, ascertaining that the disproportionate frequency of complete flakes is indicative of agave sites in general. Finally, when chipped stone resources are scarce, one strategy for conservation is reducing lithic material to usable blanks such as complete flakes near source beds (Henry 1989; Ricklis and Cox 1993), and then moving the blanks away from the source via exchange or the movement of people. Basalt represents the only material type in the study area that has a distance-to-source constraint. A distance-to-source factor may be driving observed differences in the collections. Testing this hypothesis can be carried out through other variables using the current collection. As discussed in Chapter 3, one factor of interest is the distance of sites from basalt source beds. Four levels can be identified through the use of histograms (see Chapter 3, figure 3.2) and are portrayed in figure 4.7. Agave Sites and Non Agave Sites: Distance io Basalt. Sources N AR-05-1201 i—ii—ii—1<=^ ' AZ:P:11:1 \ V °^Q0 •b| Quaternary Basalt Site Number- Agave Site AZ:P:12:277 AZ:P:12:196 AZ:P:12:210 AZ:P:12:242 • _/^VAZ:P:12:211 AZ:P:12:222<_ AZ:P:12227f> Linden AZ:P:12:260•. • AZ:P12:6 AZ:P;12:248 A AZ:P:16:9 AZ£;12:256 ^ •AZ:P:16:139 AZ:P:16:112 JT AZ:P:16:153A AZ:P:1 AZ:P:16:160 A • Level A: 3.57~5.57 km Level B: 6.57~9.57 km ^ Level C: 27 km ^ Level •: 32 km 5 kilometers Figure 4.7: Agave Sites and Non-Agave Sites: Distance to Basalt Sources io o 91 If a distance-to-source constraint is operative, certain expectations can be tested. The total weight of basalt in site collections will drop with increasing distance. In table 4.13, a Spearman's correlation is calculated to compare the weight of basalt in collections with distance from source. When comparing all sites, a significant negative correlation is observed. Table 4.13: Spearman's Rank Correlation Coefficients Total Basalt Weight at Sites by Distance from Source Distance Range of Sites (kilometers) Sample Size Correlation Coefficient Significance Test: Accept or Reject (a=0.05)* 3.57-32.0 (All Levels) 357 -0.166 reject 3.57-9.57 (Levels A&B) 297 -0.052 accept 6.57-27.0 (Levels B&C) 280 -0.092 accept 27.0-32.0 (Levels C&D) 60 -0.435 reject *H0: No relationship between variables It is also instructive to note correlations between adjacent distance-to-source levels. The negative correlation between distance-from source levels is not significant for sites between 3.57 and 9.57 kilometers, which might be expected; however that theie is not a significant negative correlation between sites in the 6.57-9.57 kilometer range and the one site 27 km from source is most unexpected. Distance has nearly 92 tripled, yet only a weak negative correlation with reduced basalt weight is indicated. A very strong decrease in total basalt weight is then observed when comparing the 27 km site to the 32 km site. The 27 km site in question is an agave site AR-051201[USFSj. This observation prompts an assessment of total basalt weight at agave sites and non-agave sites as correlated with distance to source (table 4.14). Table 4.14: Spearman's Rank Correlation Coefficients Total Basalt Weight of Sites by Distance from Source Separating Agave Factor Distance Range of Sites (kilometers) Sample Size Correlation Coefficient Significance Test: Accept or Reject (a=0.05)* All Sites 357 -0.166 reject Non-Agave Sites 303 -0.200 reject Agave Sites 54 -0.174 accept *H0: No relationship between variables The weight of basalt at agave sites does not demonstrate significant negative correlation with distance to source. This fact stands in sharp contrast to the weight and frequency of basalt cores as they are compared across distance-to-source levels (figure 4.8). 93 Frequencies of Basalt Core Weights by Distance to 400 r Source Factor 300 3200 & •H<D S100 o o o CO ° 0 7igure 3.57-5.57 6.57-9.57 27.0 32.0 Distance to Source Levels (km) 4.8: Frequencies of Basalt Core Weights by Distance to Source Factor With decreasing core weight not matched by decreasing total weight at agave sites, it could be the case that basalt flake blanks are being transported across the landscape more frequently than cores. As a means of exploring this, a Pearson chisquare test was performed comparing classes of cortical flakes with distance from source levels. No significant differences were detectable (table 4.15). 94 Table 4.15: Pearson's Chi-Square Test of Homogeneity Degree of Cortical Coverage on All Basalt Artifacts by Distance from Source Factor Percent Cortical Coverage Level 0% 11-90% 1-10% Total 3.57-5.57 101 73 12 186 6.57-9.57 63 43 5 111 22.0 15 16 2 33 32.0 21 5 1 27 Pearson Chi-Square p-value = 0.285 Cortical classes are rank-order variables and can therefore be correlated to other variables using Spearman's correlation coefficient (Kitchens 1987:107-109). When cortical coverage is compared with site distance-from-source, no significant correlation results (table 4.16). Table 4.16: Spearman's Rank Correlation Coefficients Cortical Coverage > 0% by Distance from Source Basalt Artifact Sample Sample Size Correlation Coefficient Significance Test: Accept or Reject (a=0.05)* All Artifacts 157 0.106 accept Flakes 50 -0.163 accept *H0: No relationship between variables 95 Given all these observations, we might conclude that flake blanks are moving across the landscape more than basalt cores, and that agave sites, which demonstrate high frequencies of complete basalt flakes relative to other kinds of sites, are preferential consumers of basalt flakes. This is further supported by the fact that the total weight of basalt at agave sites is not significantly negatively correlated with distance from source beds. Given this patterning in chipped stone, a large sample of surface material in conjunction with other agave site correlates such as site aspect and site size (Kunen 1994), could generate predictions of agave cultivation even when the plants are no longer growing at a site. The difficulty lies in the fact that basalt is a rare material class, and large chipped stone samples would be required for the prediction. By analyzing frequencies of complete basalt flakes and assemblage basalt weight in light of regional correlations with distance from source, potential agave sites may be identifiable. Conclusion In this chapter, elevation and the presence of agave seem to correlate with particular patterns in the chipped stone tool assemblages. In particular as elevation increases, length of complete flakes also increases. The favored interpretation for thi^; co-variation lies in the efficacy of longer cutting edges for communities dependent upon greater relative frequencies of wild resource exploitation, and for 96 communities exploiting different arboreal resources. It is argued that a chipped stone technology is sensitive to the economic basis of the community, and that even in mixed economies of a relatively restricted geographic region, chipped stone can be sensitive to differing resource exploitation strategies. Basalt seems to pattern very little with respect to elevation, but agave sites in the assemblage have higher frequencies of complete basalt flakes relative to nonagave sites (30% compared to 20%), and unexpectedly high amounts of cortical material and total weight given distance-to-source variables. Small sample sizes make conclusions difficult, but it is suggested that basalt may play a more pronounced role in the chipped stone technology of agave sites. Not all patterned variation has been successfully explained by ecological and resource exploitation factors. In the next chapter, the impacts of consuming population size, regional networks, and bifacial manufacture are explored. 97 CHAPTER V: SOCIAL ORGANIZATIONAL EFFECTS ON CHIPPED STONE TECHNOLOGY This chapter will test the hypothesis that chert is particularly influenced by variation in consuming population size due to its role in both bifacial and expedient production modes. Chipped stone assemblages are impacted by the social organization of the community creating and using the technology. The size of the consuming population obviously effects the amount of raw material acquired, and the number of people available for raw material acquisition. Social organization and exchange relationships may effect the movement of rare materials from source beds to consumers across a region. Sites which are prominent in the exchange system will have larger amounts of rare resources. In the Silver Creek surface collected sites, the consuming populations reside in pueblos ranging from 2 to 15 rooms (table 5.1), in addition to one much later pueblo with 200 rooms. These sites can be divided into site size classes, based on previous research assessing commonly occurring site sizes across the study region (Mills and Newcomb 1994). 98 Table 5.1: Sites by Site Size Factor Site Number Site Size Level Number of Rooms A 0 AZ:P:12:227[ASM] B 2- 5 AZ:P:12:167[ASM] AZ:P:12:196[ASM] AZ:P:12:210[ASM] AZ:P:12:211[ASM] AZ:P:12:222[ASM] AZ:P:12:242[ASM] AZ:P:12:256[ASM] AZ:P:16:139[ASM] AZ:P:16:176[ASM]* AR-05-1201[USFS] C 7- 10 AZ:P:12:65[ASM] AZ:P:12:260[ASM] AZ:P:16:160[ASM]* D 14 - 15 AZ:P:12:248[ASM] AZ:P:12:277[ASM]* AZ:P:16:90[ASM]* AZ:P:16:112[ASM]* AZ:P:16:153[ASM]* E 200 AZ:P:11:1[ASM] * Designates a Great Kiva site 99 At the regional level, six of the sites have large integrative structures in association with roomblocks. All of these structures are defined as circular great kivas, with the exception of AZ:P:16:176[ASM], which has a rectangular great kiva. AZ:P:11:1[ASM] has several integrative architectural features, including a plaza and possible kivas. The Pueblo period is characterized by the movement of peoples and exchanges of materials (Reid 1989). In a region of abundant lithic resources, exchange of raw materials should be unnecessary; however, basalt is a location-restricted material and some movement along exchange lines may be indicated. This would especially be assumed for the larger sites with great kivas, which may serve as foci of a regional community (Herr 1994; Reid 1989:74-75). In later Mississippian expedient-tool contexts, Koldehoff (1987:176-185) argues that formal tools may be produced by few people and exchanged to others, whereas expedient flake technology is the result of household production. Unfortunately the data for this sort of analysis in the SCARP collections is lacking, minimally requiring the presence of a representative formal tool sample, spatial and technological data. As discussed previously, formal tools are infrequent in the collections due to casual collecting (Francis 1978). While analyzing "expedient" chipped stone assemblages, the continued manufacture of bifacial tools should not be forgotten (Parry and Kelly 1987). The effects of the site size factor on the chipped stone assemblages is now addressed. 100 Site Size Factor in Chipped Stone Assemblages As in Chapter 4, preliminary investigations into the effect of site size on chipped stone variation were made by chi-square tests of homogeneity, performed on all debitage types by site size level (table 5.2). Table 5.2: Pearson chi-square test of homogeneity: Probability of Similar Frequencies of all Debitage Types Between Site Size Levels Material Type Probability Value Sample Size Basalt * 0.307 357 Diabase * 0.312 297 Silicified Wood 0.010 1736 Chert 0.000 3460 Quartzite * 0.000 842 * Cell counts low (more than 1/5 of the cells contain counts < 5), results should be taken with caution ** frequencies of artifacts are reported in Appendix B, Tables B.22-B.26 These first results indicate significant differences (a=0.05) between debitage frequencies for silicified wood, chert, and quartzite. Using this information, differences between sites within site size levels are examined for these three material classes (table 5.3). 101 Table 5.3: Pearson chi-square test of homogeneity: Probability of Similar Frequencies of All Debitage Types Between Sites within Site Size Levels Site Size Levels (# of sites in level) Silicified Wood Chert Quartzite A (1 site) auto auto auto B (10 sites) 0.002* 0.000* 0.052* C (3 sites) 0.004* 0.891* 0.192* D (5 sites) 0.000* 0.000* 0.000* E (1 site) auto auto auto * Cell counts low (more than 1/5 of the cells contain counts < 5), results should be taken with caution "auto": with only one site in test, results are auto-correlated ** frequencies of artifacts are reported in Appendix B, Tables B.27-B.29 Internally similar debitage frequencies are only found in chert level C, quartzite levels B and C. Internal consistency within levels is spotty rather than regular. A more exploratory examination of the data is indicated. The most salient functional indicator in a utilized flake tool kit is the length of complete flakes as a monitor of cutting edge. Box plots of complete flake lengths were constructed for each site class (figure 5.1-5.3) for silicified wood, chert and quartzite. Due to skewed distributions, the central tendency is assessed by medians recorded to the side of each box plot. 102 Box Plot of Complete S-Wood Flake Lengths Site Size Factor Medians 24 e * 23 £ CD M _ C 26 CD CD» CD . 4** 24 29 0 10 20 30 40 50 60 70 Length (mm) Figure 5.1: Box Plot of Complete Lengths, Site Size Factor Silicified Wood Flake 103 Box Plot of Complete Chert Flakes Lengths Site Size Factor Medians 21 © 21 CD 3? <D N 30 _ C 0D <D 25 CO 30 0 10 20 30 40 50 60 70 80 Length (mm) Figure 5.2: Box Plot of Complete Chert Flake Lengths, Site Size Factor 104 Box Plot of Complete Quartzite Flake Lengths Site Size Factor Medians e 29 28 0) <D M h- 29 CO <D CD 25 b 40 0 50 100 150 Length (mm) Figure 5.3: Box Plot of Complete Quartzite Flake Lengths, Site Size Factor 105 As site size gets larger for silicified wood and chert, there is a shift toward smaller medians. There are no dramatic shifts in total length distributions across site size classes. Spearman's correlation coefficients between flake metric variables and the number of rooms at each sites were calculated (tables 5.4-5.5). Site size and complete flake length have significant negative correlations for diabase, basalt and chert. Significant negative correlations are observed for quartzite and chert only when comparing flake thickness with site size. Hence, the only agreement in correlation significance between the two tests was for chert. Table 5.4: Spearman's Rank Correlation Coefficients Complete Flake Length by Site Size Factor Material Type Sample Size Correlation Coefficient Significance Test: Accept or Reject (a=0.05)* Basalt 82 -0.244 reject Diabase 91 -0.180 reject Silicified Wood 453 -0.072 accept Chert 736 -0.120 reject Quartzite 231 +0.031 accept *H0: No relationship between variables 106 Table 5.5: Spearman's Rank Correlation Coefficients Thickness of Flakes with one Complete Dimension by Site Size Factor Material Type Sample Size Correlation Coefficient Significance Test: Accept or Reject (a=0.05)* Basalt 119 -0.084 accept Diabase 129 -0.140 accept Silicified Wood 649 -0.078 reject Chert 1155 -0.082 reject Quartzite 342 +0.087 accept *H0: No relationship between variables The above relationships require clarification. As the exploratory chisquare tests in table 5.2 and 5.3 indicated dissimilarities in silicified wood, quartzite and chert for the site size factor, the cutting edge of these material classes are pursued via regression. Relating the influence of two metric variables such as number of rooms and flake lengths is ideally approached with a linear model. As in Chapter 4, the shift in central tendency is the issue of concern, with changing cutting edge length being of greatest functional interest. Filtering small "unusable" flakes without wild qualitative judgements is difficult. To avoid such judgements, each site's complete flake length distributions were trimmed by constructing 90% confidence intervals about the medians. These small distributions satisfy normality requirements for regression analysis hypothesis 107 testing, as confirmed by Lilliefors tests of normality. The non-comparability of AZ:P:11:1[ASM] once again presents a problem. To pass a predictive regression line from 15 rooms to 200 rooms is simply not appropriate, as the model would go through a region which completely lacks data. Consequently for the purposes of relating site size to median complete flake lengths, AZ:P:11:1[ASM] is not included in the relationship. Perhaps with the addition of more assemblages in the intervening site size classes, such a comparison will become possible. In figure 5.4, a regression line is constructed by plotting complete median flake lengths against the number of rooms, also reported are Pearson's correlation coefficients (r), the coefficients of determination (r2), the f-test probability ratio to gauge the significance of the correlations, and sample sizes (n). No patterning was revealed in the analysis of the residuals. 108 Correlations Between Median (90% C.I.) Complete Flake Lengths and Nuaiber of Rooms 60 Silicified Wood Flakes 00 r Chert Flakes a so 0 6 10 16 20 Nvmbei of Rooms r--o.l«2 r2-0.033 P-0.032 n-13« 0 6 10 16 20 Number of Roams X--0.300 r2-0.090 p-0.000 11-167 Quaitzite Flakes 0 6 10 15 20 Number of Roams r--0.l28 r2-o.016 p-0.210 n-98 Figure 5.4: Correlations Between Median (90% C.I.) Complete Flake Lengths and Number of Rooms 109 Only silicified wood and chert have significant (a=0.05) correlations. 9% of the variation in median complete chert flake lengths can be accounted for with room size, and only 3.3% for silicified wood. The results indicate that elevation, an ecological indicator, has a more powerful effect on median flake length variation for silicified wood and quartzite, while social group variables do not. Further consideration of the role of chert and the effect of consuming population size is warranted. In summary debitage frequency differences and complete flake length variation has not produced significant patterning with respect to population sizes except in the case of chert. As an alternate measure of lithic consumption differences, core weights are correlated with site size using Spearman's correlation coefficients (table 5.6). Table 5.6: Spearman's Rank Correlation Coefficients Core Weight By Site Size Material Type Sample Size Correlation Coefficient Significance Test: Accept or Reject (a=0.05)* Basalt 28 +0.156 accept Diabase 28 -0.048 accept Silicified Wood 216 +0.023 accept Chert 304 -0.019 accept Quartzite 104 -0.184 reject *H„: No relationship between variables 110 Significant negative correlations only exist for quartzite, which demonstrated no significant correlation between flake length and site size. Of all the material types, chert merits further consideration as it pertains to consuming population size. A Consideration of Chert Chert is the highest quality silicate locally available in the study area, and the production mode of chert tools appears different from other material classes. As observed in tables 5.4 and 5.5, the metric dimensions of chert flakes are significantly negatively correlated with site size level, the only material class with an unambiguous result. The inspection of complete chert flake length by site size level (figure 5.2), coupled with the relationship observed in regression (figure 5.4), demonstrates a tendency for smaller flakes to be produced at larger sites. 9% of the variation observed in median chert flake length can be accounted for with site size variation. The natural corollary of smaller flakes is significantly smaller cores at larger sites, which has been demonstrated not to be the case (table 5.5). These facts are best explained by the use of chert in bifacial tool manufacture. Chert, a highly isotropic and easily worked material, (Cotterel and Kamminga 1987; Luedtke 1992:63-98; Whittaker 1994:70-74), may be used preierentially in bifacial tool manufacture. Such a conclusion has been drawn by researchers of the Chevelon Drainage (immediately west of the SCARP study Ill area), where chert was demonstrated to be the preferred material for bifacial tool manufacture among Pueblo period communities (Reid 1982:162-164). Of the 8273 artifacts analyzed from SCARP surface collections, only 26 tools were recovered, of which 16 (62%) were made of chert. When including all formal tool categories in conjunction with flakes showing clear utilization, the sample expands to 149 artifacts, of which 97 (65%) are chert. AZ:P:11:1[ASM] has been the focus of excavation. Nineteen bifacially produced tools were recovered from the 1994 excavation, 18 of which are chert (Mills et al. 1994:27-30). Parry and Kelly (1987:295-297) remark that expedient tool technologies do not replace formal tool technologies in North America, but rather the two manufacturing techniques continue to co-exist in late prehistory. In the SCARP area, bifacial technology seems to be preferentially produced with chert as opposed to other material classes. Some of the hallmarks of bifacial reduction include the increasing relative frequencies of very small lithic debris associated with final manufacturing steps, re-sharpening, and maintenance activities (Patterson 1981, 1990; Patterson and Sollberger 1978; Stahle and Dunn 1984). Larger consuming populations appear to preferentially manufacture more bifacial tools, generating more small-size chert debris compared to smaller-sized communities. This is supported by cortical flake evidence, which indicates significant negative correlation between cortical coverage on flakes and sites size (table 5.7). 112 Table 5.7: Spearman's Rank Correlation Coefficients Cortical Coverage (cortex > 0%) of Chert Flakes by Site Size Sample Size 486 Correlation Coefficient -0.176 Significance Test: Accept or Reject (a=0.05)* reject *Ho: No relationship between variables Chert has significant negative correlations between flake size and site size and between cortical coverage and site size. This might be explained as more intensive chert reduction at larger sites, but this would fail to account for the insignificant correlation between weight of cores and site size (from table 5.5). Rather than more reduction intensity at larger sites, these observations point to the preferential role of chert in bifacial tool production, which tends to produce smaller flakes in bifacial manufacture and curation, but need not necessarily produce more reduced cores. The concentration of more people at larger sites would create a larger bifacial tool consumer group, resulting in the co-variation of site size with chert assemblage variation. This is not to say that chert is only used in bifacial manufacture, but that chert appears to be used in both expedient and bifacial tool production modes, and appears to be a preferred raw material in bifacial manufacture. This is consistent with previous discussion, where expedient flake tools are argued to be particularly sensitive to the context in which they are situated. Hence, silicified wood and 113 quartzite are sensitive to the technological demands placed upon them in differing ecological circumstances, while chert operating in both production modes demonstrates significant (a=0.05) correlations to both elevation and site size variation. Conclusion The size of the consuming population effects the patterning of chert, accounting for variation which is unexplained by the ecological proxy of site elevation. This result is indicative of chert's preferential role in bifacial tool manufacture, whose debitage is effected by large numbers of bifacial tool producers and consumers. The role of regional social organization in the movement of lithic materials cannot be satisfactorily addressed in the current data base. In the following chapter, the variables that impact a chipped stone assemblage are reviewed in light of the Silver Creek chipped stone assemblages, and the ability of debitage analysis to reveal the effects of factors of interest. 114 CHAPTER VI: SUMMARY DISCUSSION AND CONCLUSION The Proposition: After the analysis of 20 chipped stone assemblages from Pueblo period sites in the Silver Creek area, the sensitivity of expedient Southwestern lithic technology in sedentary communities can be assessed. Any chipped stone technology is influenced by a variety of factors (Rolland and Dibble 1990:484492). The chipped stone technologies of the Silver Creek area are in a context of changing social and ecological communities. Due to chipped stone abundance and diversity, the reduction of most materials at these sites need not be influenced by distance-to-source constraints. A technology situated in a lithic poor environment may only reflect the adequate conservation of a scarce resource. In a lithic rich environment, chipped stone analyses may reflect other variables, including number of consumers (a social dimension) and the technological challenges presented by environment and subsistence strategies (an ecological dimension). The expedient/utilized flake technology of puebloan communities, rather than carrying little information, should be particularly sensitive to technological demands. A expedient tool is created to serve a specific function and then discarded (Parry 1987), creating large assemblages of tools designed for a limited 115 functional life. The tasks in which these tools performed should pattern by the context of the community and its resource extraction behaviors. Technological Demands of Subsistence and Ecological Setting Elevational difference in the Silver Creek area reflect patterned differences in natural biotic communities, creating variable moisture, temperature and wild resource extraction zones. The positive correlation of increasing flake length observed in chert, silicified wood and quartzite with elevation has been argued to represent a technological reflection of different ecological resources and challenges presented to these communities. With higher elevation and shorter frost-free seasons, more wild plant and animal processing is expected to "back-up" fragile agricultural systems. Wood fuel resources in higher elevation contexts differ from lower elevation resources by size and processing demands. Wild game and hide processing requires efficient cutting and scraping technology. Given these points, longer cutting edges and heavier implements are more efficacious in higher elevation contexts. Agave cultivation presents its own unique need for durable-edged heavier tools. Higher amounts of basalt in agave sites at unexpectedly long distances from basalt source beds (e.g., AR-05-1201[USFS]), coupled with unexpectedly high amounts of cortical material on flakes, might indicate that basalt served a special role in agave processing. The high relative frequencies of complete basalt 116 flakes at agave sites, when contrasted with non-agave sites, similarly points to the special involvement of basalt in agave processing. The creation of larger databases of agave site chipped stone and subsequent acquisition of chipped stone found in agave fields should help to develop an understanding of agave "tool kits" above the Mogollon Rim. The Influence of Site Size on Chipped Stone The consuming population size, a variable not correlated with ecological community, seems to have a role. A general trend of smaller flakes, especially for chert at larger sites in the study area, has been indicated. Chert flake length varies with site size to a degree not observed in other material classes. The special sensitivity of chert assemblages to site size seems tied to the preferential use of chert in bifacial technology. While all bifacial tool users generate large amounts of very small flakes during consumption, production, and re-sharpening, the existence of larger consuming populations seems to increase the frequency of smaller chert flakes in assemblages. Implications I have assumed that the production of an expedient flake technology is not stochastic, rather manufacturers create chipped stone tools which are suited to the tasks at hand, and these tasks are shaped by ecological circumstance. 117 1) Expedient flake tool technology appears to be sensitive to ecological setting, due to its short functional life, the tool serves in a specific resource extraction context generating patterned variation where resource extraction zones differ. 2) When expedient and bifacial technologies exist side-by-side, some material classes are chosen preferentially to serve in each technology. For the Silver Creek area, chert seems to be preferentially selected for bifacial production, although its sensitivity to elevation should not be dismissed. This is most clearly illustrated in table 6.1, where the correlations between factors of interest in this study and the cutting edge of complete flakes (monitored in the central tendencies of complete flake length distributions: 90% confidence intervals about the medians) are summarized. In particular, chert flake length demonstrates a higher coefficient of determination (r2) for the site size factor than do the other material classes. 118 Table 6.1: Pearson Correlations, Coefficients of Determination, Probability Values for Median Length of Complete Flakes (90% Confidence Intervals) by Factors of Interest Material Type Statistic Elevation Site Size Distance to Basalt Source % r2 sum of significant results -0.014 0.0% 0.917 0.0% Basalt n=55 r= % r2= P= 0.067 0.4% 0.628 -0.190 3.6% 0.165 Diabase n=57 r= % T2= P= 0.177 3.1% 0.187 -0.080 0.6% 0.553 Silicified Wood n=138 r= % r1P= 0.442 19.5% 0.000 -0.182 3.3% 0.032 N/A Chert n=167 r= % TZ= P= 0.400 16.0% 0.000 -0.300 9.0% 0.000 N/A Quartzite n=98 r= % T1= P= 0.446 19.9% 0.000 0.128 1.6% 0.210 N/A 0.0% N/A 22.8% 25.0% 19.9% 3) In a lithic rich and diverse environment, particular resource extraction behaviors may preferentially consume certain material types. Agave sites and the consumption of basalt being a case in point. 4) A priori site information can serve as factors of interests, guiding the analysis of variation, such that the effects of a given factor can be quantitatively assessed in the conclusion of research. 119 Building References for Further Investigation Chipped stone assemblages, even highly expedient tools of sedentary puebloan peoples, contain patterned variation which reflects differing social and ecological contexts of technology. This study proceeded from an exploratory stance, and preliminary results generate prospects for further inquiry. 1) Wild resource exploitation strategies generate different tool demands. Socalled "mixed" economies need not remain black boxes, quantifiable only with human osteological remains. Mauldin (1993a) has alleged on ground stone evidence that truly intensive agriculture is not found until late prehistory for puebloan communities of western New Mexico. Ezzo's (1992, 1994) osteological analysis of Grasshopper Pueblo people has demonstrated that different people within the same community have different diets. Brose's (1978) work in the Mississippian of Ohio demonstrate chipped stone variation between habitation sites and specialized sites synchronically and diachronically. Chipped stone technology can trace, in a quantifiable fashion, different exploitation emphases within mixed economies, even in narrow geographic and temporal stretches, which can test or further support conclusions drawn from such studies listed above. Hypothesis-generating potential exists in the analysis of expedient technological variation to understand exploitation strategy differences between and perhaps within communities. 2) Sedentary puebloan community chipped stone variation need not be a noninformation bearing artifact class. While it is true that this study did not account for 120 all variation in cutting edge, 20-25% co-variation between factors of interest and observable chipped stone attributes contribute to further understandings of the Silver Creek area. Excavations at sites of varying altitudes are expected to generate differences in floral and faunal assemblages consistent with the technology which exploited them. SCARP's ongoing research should be able to assess if higher altitude communities relied more upon wild plant and animal processing, and assess the fuel consumption needs and difficulties for communities of different ecological contexts. Debitage analyses are one avenue toward exploring these differing strategies and generating hypotheses for future testing. APPENDIX A Silver Creek Archaeological Research Project Lithic Analysis Key LITHIC MATERIAL I.D. 00 01 indeterminate indeterminate, nonlocal 10 11 12 13 14 igneous, nfs basalt, andesite (dark, £ine- medium grain) diabase granite obsidian 20 21 22 23 24 sedimentary, nfs sandstone shale, claystone conglomerate limestone 30 31 32 33 34 35 silicates, nfs quartz (crystal) agate, chalcedony silicified sandstone/siltstone silicified wood quartz 40 41 chert jasper 51 52 metamorphic, nfs quartzite, metaquartzite 90 91 92 93 minerals, nfs- comment turquoise hematite ochre 122 CORTEX Estimate cortex on all surfaces of the item. Cortex is any stone surface recognizable as "natural" or non-produced. Estimate cortex for only the portion of the flaked lithic that is present, do not extrapolate. 0 none 1 1-10% 2 3 4 10-50% 50-90% 90-100% ITEM CONDITION 0 1 2 3 4 no complete dimensions complete (weight complete, all margins intact) nearly complete (length and width measurements complete; weight is not complete) complete length (width and weight not complete) complete width (length and weight not complete) FLAKE PORTION This code list is prioritized, so if the flake has more than one of the flake portions present, choose the portion nearest the top of the list. 0 1 2 3 4 5 not applicable (ex: shatter) profile (proximal- distal portions available) proximal distal medial unknown 123 DEBITAGE CATEGORY (SULLIVAN AND ROZEN 1985; ROZEN AND SULLIVAN 1989b) This code provides a different way of recording the completeness of flakes, which is similar to analyses in other Southwestern lithic studies. 0 1 2 3 4 other (anything utilized, core, tested cobble) complete flake (flake platform and all margins in tact) broken flake (flake with a platform but lacks complete margins) flake fragment (a clear flake without platform) debris (angular debris, shatter) PROJECTILE POINT PORTION This code list is prioritized, so if the projectile point has more than one of the projectile point portions present, chose the portion nearest the top of the list. 0 1 2 3 4 5 not applicable profile base/ complete base/ incomplete tip midsection PLATFORM TYPE 0 1 2 3 4 5 not applicable cortical platform, no scars noncortical platform, no scars dihedral platform, 2 scars faceted platform, >2 scars indeterminate USE WEAR 0 1 absent present 124 FLAKED LITHIC TECHNOLOGICAL TYPE 00 01 02 03 04 05 06 07 08 09 10 15 angular debris, shatter core flake biface thinning flake notched flake unidirectional edge uniface bidirectional edge biface tested cobble manuport other (comment) "Edge" minimally intrusive production flakes; "face"- production flakes intrusive to, or nearly to, the mid-line of the item's face(s) FLAKED LITHIC FUNCTIONAL TYPE 00 01 02 04 05 06 07 08 09 10 15 not applicable used core used flake spokeshave chopper, flaked axe scraper graver, perforator drill knife projectile point other (comment) WEIGHT To the nearest .1 gram LENGTH To the nearest .01 mm 125 WIDTH To the nearest .01 mm THICKNESS To the nearest .01 mm COMMENTS The comments field must be used if an item has been coded as other in the TECHNOLOGICAL TYPE or FUNCTIONAL TYPE fields. Additional information that may be desirable includes whether there is evidence of heat treatment (particularly on cherts), whether the raw material of the item is of nonlocal origin, if this can be determined or the geologic formation from which the item is derived (applicable for cherts and sandstone particularly) again, if this is possible to determine. Information such as "preform" or "reused" is not obvious from the coding sheet, but may provide useful information in later analyses is worth noting, as are any unusual characteristics such as serrated or asymmetrical projectile points. INITIALS Initials of analyst 126 Silver Creek Archaeological Research Project Lithic Analysis Code Book Provenience Information From bag Raw materials See rock and mineral guide Cortex Values of this variable are simple grouped estimates of the proportion of the study item's surface which is composed or cortex (natural, unproduced surface). Cortex here includes all surfaces recognizable as not being the product of human action. Item condition This variable gives a measure as to how complete the item is, and how useful each of its measurements will be in future analyses. Flake Portion The terms 'top' and 'bottom' to 'towards the proximal end' and 'towards the distal end', respectively. 'Horizontal' means parallel to flake width and 'vertical' means parallel to flake length. Not applicable— the study item is not a flake (example: core, shatter) Profile- study item has proximal, medial and distal portions present- no horizonal fracture. Proximal- platform, and bulb of percussion are present. Neither the platform nor the bulb of percussion need to be complete. Snap fracture is present on bottom of flake portion. 127 Distal- this is the end opposite the bulb of percussion— the natural termination of the flake. The distal end should show a feather fracture, a step or hinge fracture, but not a snap fracture. Snap fracture is present on top end of flake portion. Medial- both the proximal and distal ends of the flake are missing, only the flake midsection remains. Two snap fractures are present. Unknown- the item in question is a flake, but it is too damaged to orient. Projectile Point Portion The terms 'top' and 'bottom' to 'towards the tip' and 'towards the base', respectively. 'Horizontal' means parallel to point width and 'vertical' means parallel to point length. Not applicable— the item is not a projectile point Profile— the study item has tip, complete base and midsection present— no horizonal fracture. Base/ Complete- this end of the projectile point shows evidence of hafting by exhibiting side or corner notching, basal thinning, or grinding, or any combination of these characteristics. Enough of the base must be present to distinguish whether notching symmetrical or asymmetrical for this code to be used. Projectile point bases are often used in typology. Horizontal snap fracture present at top of item. Base/ Incomplete- this end of the projectile point shows evidence of hafting by exhibiting side or corner notching, basal thinning, or grinding, or any combination of these characteristics. Not enough of the base is present to distinguish whether notching is symmetrical or asymmetrical. At least one notch must be complete to assign the item this variate. Horizontal snap fracture present at top of item. Tip- the point of the projectile point- where the margins converge. If the actual tip is missing, but would be within approx. 1 millimeters, encode the item as a tip. Horizontal snap fracture present at bottom of item. 128 Midsection- neither tip nor base are present, but the item has manufacturing scars, possibly indications of notches in combination with converging edges indicating it was probably, at least morphologically, a projectile point. Two horizontal snap fractures present, one at top, one at bottom of item. Platform type Not applicable— proximal portion of flake not present Cortical— platforms with any amount of cortex Plain- platforms with no apparent cortex, consisting of a single surface unbroken by flake scars Dihedral— platforms with no apparent cortex, two flake scars Faceted- platforms with no apparent cortex, more than two flake scars Indeterminate- proximal portion of flake with major portion of platform missing Usewear Usewear is present if there is evidence of microflaking interpreted as the result of use (not retouch), edge rounding/ dulling (especially useful for coarser stones), edge frosting (the frosty white character of edge rounding on obsidian), striations, polish (reduced irregularity of surface topography, with flake scars evened out; increased light reflectivity), etc. Technological Type This is the first of two typological evaluations given to each item that is modified by either use-wear or production flaking. This typology focuses on the production technology of tools in the assemblage(s). Variable values are arranged, very generally, from lower levels of production input to higher levels. It is possible that a tool will fit into one or more of these categories— for example many pueblo projectile points will have bidirectional edges with one surface more fully flaked (unifacial). Since these types are rank ordered chose the code for the "higher" level of production input (the code with the higher number). If the item is too technologically unusual to treat in this manner place the it in the other category with an explanation. 129 Some important definitions include: Dorsal-- the "back" of the flake that, if still attached to the core, would be the exterior surface of the core. It is this portion of the flake that exhibits cortex (for the first flake taken off a core) and/or scarring from previous steps in the reduction process. Ventral- this portion of the flake would, if attached to the core, face the interior of the nodule. It is this portion of the flake that exhibits the "flake characteristics" such as bulb of percussion, eraillure scar, force waves, and possible lines radiating from the platform. "Edge"-- an edge is formed on a relatively flat item by short flakes that extend from the margin no more than 1/4 of the distance across the item's dorsal and/ or ventral surface. "Face"-- a face is produced by relatively longer flakes which extend at least 1/3 to 1/2 the distance from the margin across the dorsal and/ or ventral surface of the flake, so that the entire surface, or most of the surface, is formed by production flake scars. Angular debris, shatter— this category should be applied to items which are obviously the result of lithic reduction but which do not exhibit any flake or tool morphology Core-- in a rigid technical sense any rock from which a single flake has been removed is a core, including most of the following tool categories. Such a definition, however, is not very helpful in constructing archaeologically useful typologies. Here, a core is an item from which at least two flakes have been removed, and on which no apparent attempt has been made to produce a relatively low-angle (< approx. 60 degrees) edge or margin (thus turning it into a core tool). Flake— a flake is a piece of stone removed from a core by fracture, exhibiting a positive fracture cone or bulb, one of several characteristic fracture surface morphologies, and/or a localized point or area of force application. If flakes are further reduced by production flaking they become other technological types. Bifacial thinning flake— this value indicates a particular kind of flake removed in the process of biface reduction, which is distinguished here because it represents an increased degree of technological concern and fracture control. Such flakes, 130 because they are removed from items which are already shaped bifaces, have several distinguishing characteristics: (1) they are quite thin in relation to their width, (2) they are frequently rather wide in relation to their length, (3) they usually have small, narrow platform areas which are actually removed segments of the biface margin and (4) their dorsal surfaces display the scars which were previously the biface surface. Notched flake- a flake which is unmodified except for a notch in one edge, formed by a single application of force- sometimes used as spokeshaves. Unidirectional edge- these are items on which an edge has been produced by repeated force application (retouch) in only one direction across that edge, removing short flakes from only one surface of the item. Edge flake scars are located on only the ventral or the dorsal flake surface. Uniface— this technological type includes items on which one surface has been predominantly produced by flaking, and is therefore largely covered by flake scars, most of which extend at least 1/3 to 1/2 the distance across the surface. "Face" type flakes are located only the ventral or dorsal flake surface. Bidirectional edpe— these items display an edge produced the repeated removal of short flakes from both surfaces in both directions across the edge. Edge flake scars are located on both the ventral and dorsal flake surfaces. Biface- bifaces have both surfaces covered largely by production scars. "Face" type flakes are located on both the ventral and dorsal flake surfaces. Tested Cobble-a nodule that has been split open and is covered with 50% or more cortex with no evidence of flake removal Manuport—a unmodified stone (over 90% cortex) which has been obviously moved to the site by human agency Other- other technological types, either common or uncommon in the relevant literature may be included along with an explanation in COMMENTS (one uncommon, but not unheard of, example might be utilized shatter) Functional Type The second typological evaluation given to each item is a general functional class assignment. Two criteria are used in this evaluation; actual indications of use, 131 and produced morphological suitability for certain categories of use. Values are arranged generally form least to greatest probable production input. Not applicable- this value indicated the study item is/ was apparently not used, and has not been modified into some shape suitable for use; i.e., it is apparently a non-functional item (it is a by product of the manufacture of some other functional item). This value includes cores from which flakes or blades were produced but that have not apparently been used. Used core- items designated as used cores virtually always demonstrate use as a vertical fore applier, or hammerstone, after they have had two or more significant (or shape-altering) flakes removed. Whether the flakes were removed for the purpose of shaping the core/ hammerstone or for some other purpose is immaterial in this decision. Used flake— this value is assigned to all flakes that are not further modified by production after their removal from a core, and that demonstrate some type of obvious use wear. This involves the difficult analytic decision as to whether certain small-scale morphologic characteristics— especially microflake scars— are the product of use or are simply incidental/ accidental. The guidelines given earlier in the use-wear variable/ value description should be used to help provide analytical consistency, but some element of observer preference and variation will likely remain. There will also likely be preference given to certain raw material types that break easily, or show small scars more clearly. Interpretation of this value should therefore be tempered to recognize this inherent variation. Spokeshave— these are usually formed on notched flakes where the notch shows evidence of scraping or polish. Chopper, flaked axe- this value is intended to group heavy vertical force appliers, either hafted or unhafted, on which an edge or point has been produced by flaking. These are normally rather large items, often formed on cobbles or thick slabs by the removal of a few large flakes. Any hafting elements such as grooves or notches are discounted for this classification, though they should be briefly described as comments, and attention is focused on the bit, or "business end" of the item. Hammerstones, mauls, and ground bit axes are included elsewhere with nonflaked lithic ground stone tools. Scraper— this functional category groups items on which a certain kind of edge has been produced, and that may, or may not, demonstrate certain kinds of use wear. Contiguous or overlapping flake scars produce a regular, high included 132 angle (> approx. 50 degrees) edge that may be straight, concave, or convex in plan view. If microflaking is present it is essentially restricted to one of the edgeforming surfaces, normally that which is produced by flaking. Edge rounding may or may not be present, but any striations are perpendicular to the produced edge, and also restricted to a single surface. For scrapers, edge strength is more important than edge sharpness. Graver, perforator- these are functionally rather specialized items, though they may be technologically quite simple and generalized. A small, short projection has been produced by unidirectional or bidirectional flaking, often on a small, thin item. These items are suitable for incising or perforating materials considerably softer than stone such as hide, wood,or bone. Drill— this tool type is also characterized by a produced projection on a relatively small item, but the projection is long and narrow, and suitable for drilling holes in somewhat harder materials such as bone or even softer stone. The items may or may not demonstrate some kind of hafting mechanism, though they are often produced as small bifacial items with notches, perhaps from worn or broken projectile points. If actual use wear is present it should have resulted from rotary motion: tip and/ or edge rounding, microflaking, and/ or striations perpendicular to the tip edges. Knife- this tool type is functionally quite generalized, and groups all items on which a low included angle edge (< approx. 40 degrees) has been produced by unidirectional, bidirectional, or possibly bifacial flaking. Here, edge sharpness is more important than edge strength. Items with such edges are suitable for cutting a variety of softer materials such as fiber, meat, skins, or even softer wood. Slight edge rounding and/ or polish may be found on both edge- forming surfaces, and any striations are parallel or slightly oblique to the edge. It is quite likely that "knives" and "saws" might both be included in this tool category. Knives need not display overall bilateral symmetry, and production input may be rather minimal although some retouch is required to distinguish them from used flakes. Overall, this category can be technologically quite variable. Projectile point— projectile points are normally rather small, thin items on which a distinct point and some hafting mechanism (generally) have been produced at opposite ends. They usually display general bilateral symmetry, and technological production input may vary from relatively low to very high. Points may have been usea for either the atl-atl or the bow-and-arrow propulsion mechanism, although the latter are more likely on pueblo period sites. Generally these two types of points are distinguished by size, with the atl-atl points being larger. If a large 133 probable point is found on a pueblo period site it is worth considering the possibility of it functioning as a knife. There is clearly a great deal of functional overlap between knives and projectile points in that a great may projectile points, even quite small ones, are totally suitable for light cutting tasks and may display use wear indications to this effect. Such items should be classed as projectile points rather than knives in order to recognize their more specialized functional and/ or technological nature. Other- other functional types may be indicated here, along with an explanation in COMMENTS. A possible inclusion in this category is manufacturing blanks such as projectile point preforms. Weight Weight, in grams Length This measurement is taken with calipers, in millimeters, along the axis of applied force- from where the platform is/ was to where the flake/ "tool" ends/ ended. Take the measurement at the point of longest length. If the item is not a flake (shatter, tool) or is too damaged to orient use the maximum dimension of the item as length. This category, as with all measurement categories, should never be blank. Width Take maximum width measurement with calipers, in millimeters, perpendicular to the axis of applied force-- the distance between the margins of the flake/ "tool". If the axis of applied force was not determined in the measurement of length, take width measurement perpendicular to the axis measured in the length category. Thickness Take the maximum thickness measurement with calipers, in millimeters. The thickness measurement is the distance between dorsal and ventral surfaces of the flake/ "tool". If the axis of applied force was not determined in the measurement of length, take width measurement perpendicular to the axis measured in the length category and the axis measured in the width category. 134 Comments The comments field must be used if an item has been coded as other in the TECHNOLOGICAL TYPE or FUNCTIONAL TYPE fields. Additional information that may be desirable includes whether there is evidence of heat treatment (particularly on cherts), whether the raw material of the item is of nonlocal origin, if this can be determined or the geologic formation from which the item is derived (applicable for cherts and sandstone particularly) again, if this is possible to determine. Information such as "preform" or "reused" is not obvious from the coding sheet, but may provide useful information in later analyses is worth noting, as are any unusual characteristics such as serrated or asymmetrical projectile points. 135 LITHIC SYSTEM DERIVED FROM Camilli, Eileen n.d. "General and Lithic Codes" (code sheet) (this provided some general ideas about projectile point portion, and flake portion although both categories were modified) Phagan, Carl J. 1985 Bandelier Archeological Survey Lithic Analysis System, on file at National Park Service Southwest Regional Office, Santa Fe. (raw materials, cortex, usewear definition, technological type and functional type adapted from this system. Some portions of this manual are used verbatim or nearly verbatim) Rozen, Kenneth C. 1984 Flaked Stone in Hohokam Habitation Sites in the Northern Santa Rita Mountains Archeological Series No. 147, Vol. 2, Part 1, ed. by Alan Ferg, Kenneth C. Rozen, William L. Deaver, Martyn D. Tagg, David A. Phillips Jr., and David A. Gregoiy. Cultural Resource Management Division Arizona State Museum, Tucson, pp. 421- 604. (platform type adapted from this system, and personal communication with the author) Rozen, K.C., and A.P. Sullivan 1989 The Nature of Lithic Reduction and Lithic Analysis: Stage Typologies Revisited. American Antiquity 54:179-184. Sullivan, Alan P., Ill, and Kenneth C. Rozen 1985 Debitage Analysis and Archaeological Interpretation. American Antiquity 50:755-779. Specific material types were derived from a review of contract literature in the area. 136 APPENDIX B: FREQUENCY TABLES Table B.l Basalt Debitage Category Frequencies by Elevation Factor Complete Flake Broken Flake Flake Fragment Shatter Total A 9 3 10 32 54 B 1 0 0 2 3 C 5 1 1 4 11 D 5 2 1 13 21 E 16 5 7 31 59 F 0 0 1 0 1 G 4 0 0 14 18 H 34 16 11 94 155 I 1 2 0 5 8 J 7 4 1 15 27 82 33 32 210 357 Level TOTAL WARNING: MORE THAN ONE-FIFTH (FREQUENCY < 5) OF FITTED CELLS ARE SPARSE 137 Table B.2 Diabase Debitage Category Frequencies by Elevation Factor Level Complete Flake Broken Flake Flake Fragment Shatter Total A 13 2 3 5 23 B 1 0 0 1 2 C 3 5 1 7 16 D 16 4 6 17 43 E 19 9 4 20 52 G 13 1 0 14 28 H 10 3 5 53 71 I 4 0 1 2 7 J 13 5 3 14 35 Total 92 29 23 153 297 WARNING: MORE THAN ONE-FIFTH OF FITTED CELLS ARE SPARSE (FREQUENCY < 5) 138 Table B.3 Siliclfied Wood Debitage Category Frequencies by Elevation Factor Level Complete Flake Broken Flake Flake Fragment Shatter Total A 64 14 19 83 180 B 15 0 0 9 24 C 26 8 12 61 107 D 64 14 14 126 218 E 112 32 37 253 434 F 2 2 1 10 15 G 34 5 9 75 123 H 38 18 15 142 213 I 17 14 8 41 80 J 83 32 17 210 342 455 139 132 1010 1736 Total 139 Table B.4 Chert Debitage Category Frequencies by Elevation Factor Level Complete Flake Broken Flake Flake Fragment Shatter Total A 138 44 58 371 611 B 14 8 1 24 47 C 45 17 33 214 307 D 68 23 33 172 296 E 107 54 48 355 564 F 9 2 3 19 33 G 33 11 10 96 150 H 112 72 40 402 626 I 17 14 14 71 116 J 197 94 52 364 707 Total 740 339 290 2088 3457 140 Table B.5 Quartzite Debitage Category Frequencies by Elevation Factor Level Complete Flake Broken Flake Flake Fragment Shatter Total A 33 3 7 34 77 B 5 1 0 3 9 C 7 3 2 23 35 D 35 11 8 39 93 E 48 35 12 137 232 F 0 0 2 5 7 G 16 2 4 36 58 H 61 22 15 139 237 I 7 1 1 24 33 J 20 12 3 27 62 232 90 54 467 843 Total WARNING: MORE THAN ONE-FIFTH OF FITTED CELLS ARE SPARSE (FREQUENCY < 5) 141 Table B.6 Basalt Debitage Category Frequencies within Elevational Levels Shatter Flake Frag. E Level Site Number Complete Flake Broken Flake A 12:167 4 2 7 26 39 12:277 5 1 3 6 15 B 16:90 1 0 0 2 3 C 12:65 0 0 0 1 1 12:196 5 1 1 3 10 12:211 5 2 1 11 19 16:139 0 0 0 2 2 12:242 5 0 4 6 15 16:112 5 5 2 19 31 16:153 5 0 0 1 6 16:160 1 0 1 5 7 F 12:210 0 0 3 0 3 G 12:227 3 0 0 1 4 12:256 1 0 0 13 14 12:248 25 14 10 73 122 5-1201 9 2 1 21 33 I 12:260 1 2 0 5 8 J 11:1 7 4 1 15 27 D E H 142 Table B.7 Diabase Debitage Category Frequencies within Elevational Levels Level Site Number Complete Flake Broken Flake A 12:167 11 2 2 22 37 12:277 2 0 1 3 6 B 16:90 1 0 0 1 0 C 12:65 0 1 1 2 4 12:196 3 4 0 5 12 12:211 16 2 5 15 38 16:139 0 2 1 2 5 12:242 4 4 1 6 15 16:112 10 3 0 10 23 16:153 2 0 1 2 5 16:160 3 2 2 2 9 12:227 10 1 0 8 19 12:256 3 0 0 6 9 12:248 7 3 5 44 59 5-1201 3 0 0 9 12 I 12:260 4 0 1 2 7 J 11:1 13 5 3 14 35 D E G H Flake Frag. Shatter L Z 143 Table B.8 Silicified Wood Debitage Category Frequencies within Elevational Levels Flake Frag. Shatter £ Level Site Number Complete Flake Broken Flake A 12:167 33 83 12 46 99 12:277 31 6 7 37 81 B 16:90 15 0 0 9 24 C 12:65 8 2 7 28 45 12:196 18 6 5 33 62 12:211 54 9 11 83 157 16:139 10 5 3 43 61 12:242 36 5 12 55 108 16:112 42 20 15 156 233 16:153 26 4 8 32 70 16:160 0 0 1 6 7 16:176 8 3 1 4 16 12:210 1 0 0 0 1 12:222 1 2 1 10 14 12:227 27 5 8 43 83 12:256 7 0 1 32 40 12:248 13 5 6 66 90 5-1201 25 13 9 76 123 I 12:260 17 14 8 41 80 J 11:1 83 32 17 210 342 D E F G H 144 Table B.9 Chert Debitage Category Frequencies within Elevational Levels Shatter Flake Frag. 2 Level Site Number Complete Flake Broken Flake A 12:167 70 16 28 218 332 12:277 68 28 30 153 279 B 16:90 14 8 1 24 47 C 12:65 12 7 7 55 81 12:196 33 10 24 159 226 12:211 58 21 23 131 233 16:139 10 2 10 41 63 12:242 31 9 6 112 160 16:112 51 28 26 177 282 16:153 17 10 10 32 69 16:160 6 5 4 27 42 16:176 2 2 0 7 11 12:210 7 0 0 10 17 12:222 2 2 3 9 16 12:227 22 7 6 34 69 12:256 11 4 4 32 81 12:248 44 32 21 205 302 5-1201 68 40 19 197 324 I 12:260 17 14 14 71 116 J 11:1 197 94 52 364 707 D E F G H 145 Table B.10 Quartzite Debitage Category Frequencies within Elevational Levels Shatter Flake Frag. E Level Site Number Complete Flake Broken Flake A 12:167 18 3 4 18 43 12:277 15 0 3 16 34 B 16:90 5 1 0 3 9 C 12:65 1 0 0 12 13 12:196 6 3 2 11 22 12:211 34 11 8 34 87 16:139 1 0 0 5 6 12:242 12 12 7 34 65 16:112 28 16 3 87 134 16:153 4 3 1 7 15 16:160 3 3 1 8 15 16:176 1 1 0 1 3 12:210 0 0 0 2 2 12:222 0 0 2 3 5 12:227 11 1 0 8 20 12:256 5 1 4 28 38 12:248 50 16 11 104 181 5-1201 11 6 4 35 56 I 12:260 7 1 1 24 33 J 11:1 20 12 3 27 62 D E F G H 146 Table B.11 Diabase Cortical Flake Frequencies by Elevation Factor Percent Cortical Coverage Level 11-50% 1-10% 0% 51-90% Total A 13 1 4 0 18 B 1 0 0 0 1 C 7 1 0 1 9 D 16 3 5 3 27 E 23 2 6 1 32 G 4 4 6 0 14 H 14 1 1 1 17 I 3 1 1 0 5 J 158 2 3 0 20 Total 96 15 26 6 143 WARNING: MORE THAN ONE-FIFTH (FREQUENCY < 5) OF FITTED CELLS ARE SPARSE 147 Table B.12 Silicifed Wood Cortical Flake Frequencies by Elevation Factor Percent Cortical Coverage Level 11-50% 1-10% 0% 51-90% Total A 75 8 11 3 97 B 11 2 1 1 15 C 40 1 5 0 46 D 59 9 21 4 93 E 133 17 26 4 180 F 2 1 1 1 5 G 35 4 9 0 48 H 51 5 12 2 70 I 30 3 5 1 39 J 97 14 20 0 131 533 64 111 16 724 Total WARNING: MORE THAN ONE-FIFTH (FREQUENCY < 5) OF FITTED CELLS ARE SPARSE 148 Table B.13 Chert Cortical Flake Frequencies by Elevation Factor Percent Cortical Coverage Level 0% 11-50% 1-10% 51-90% Total A 75 8 11 3 97 B 11 2 1 1 15 C 40 1 5 0 46 D 59 9 21 4 93 E 133 17 26 4 180 F 2 1 1 1 5 G 35 4 9 0 48 H 51 5 12 2 70 I 30 3 5 1 39 J 97 14 20 0 131 533 64 111 16 724 Total WARNING: MORE THAN ONE-FIFTH (FREQUENCY < 5) OF FITTED CELLS ARE SPARSE 149 Table B.14 Quartzite Cortical Flake Frecpaencies by Elevation Factor Percent Cortical Coverage Level 11-50% 1-10% 0% 51-90% Total A 17 11 9 5 42 B 1 2 2 1 6 C 5 3 4 0 12 D 16 8 26 4 54 E 35 14 34 12 95 F 2 0 0 0 2 G 7 3 7 5 22 H 26 22 43 6 97 I 1 1 5 1 8 J 16 2 13 4 35 126 66 143 38 373 Total WARNING: MORE THAN ONE-FIFTH (FREQUENCY < 5) OF FITTED CELLS ARE SPARSE 150 Table B.15 Basalt Debitage Type Frequencies by Agave Factor Level Tested Core Complete Broken Flake Shatter Flake Frag. Flake Cobble £ Agave 3 4 17 2 5 Non-agave 2 23 58 26 26 141 276 Total 5 27 75 28 31 164 330 WARNING: MORE THAN ONE-FIFTH (FREQUENCY < 5) OF FITTED CELLS 23 54 ARE SPARSE Table B.16 Diabase Debitage Type Frequencies by Agave Factor Level Tested Core Complete Broken Flake Shatter Cobble Flake Frag. Flake 2 Agave 1 8 17 7 2 Non-agave 2 18 62 17 17 95 211 Total 3 26 79 24 19 111 262 WARNING: MORE THAN ONE-FIFTH (FREQUENCY < 5) OF FITTED CELLS 16 ARE 51 SPARSE Table B.17 Silicified Wood Debitage Type Frequencies by Agave Factor Level Agave Tested Core Complete Broken Flake Shatter Cobble Flake Flake Frag. E 7 63 98 28 32 147 Non-agave 14 111 273 79 83 459 1019 Total 21 174 371 107 115 606 1394 375 151 Table B.18 Chert Debitage Type Frequencies by Agave Factor Level Tested Core Complete Broken Flake Shatter Flake Frag. Flake Cobble 2 Agave 22 73 131 57 43 Non-agave 72 189 410 184 194 1087 2136 Total 94 262 541 241 237 1378 2753 291 617 Table B.19 Quartzite Debitage Type Frequencies by Agave Factor Level Agave Tested Core Complete Broken Flake Shatter Cobble Flake Flake Frag. Z 8 22 35 19 10 53 147 Non-agave 21 74 176 59 39 264 633 Total 29 96 211 78 49 317 780 Table B.20 Basalt Debitage Type Frequencies between Agave Sites Site Number Tested Core Complete Broken Flake Shatter Cobble Flake Flake Frag. Z 12:227 0 0 3 0 0 1 4 12:242 0 1 5 0 4 5 15 16:139 0 1 0 0 0 1 2 5-1201 3 2 9 2 1 16 33 Total 3 4 17 2 5 23 54 WARNING: MORE THAN ONE-FIFTH OF FITTED CELLS ARE SPARSE (FREQUENCY < 5) 152 Table B.21 Basalt Debitage Type Frequencies between Non-agave Site Number SjLtes Tested Core Complete Broken Flake Shatter Flake Flake Frag. Cobble E 12:65 0 0 0 0 0 1 1 12:167 1 2 4 2 7 23 39 12:196 0 1 5 1 1 2 10 12:210 0 0 0 0 1 0 12:211 1 3 5 2 1 7 19 12:248 0 10 25 13 10 64 122 12:256 0 0 1 0 0 13 14 12:260 0 1 1 2 0 4 8 12:277 0 1 5 1 3 5 15 16:90 0 0 1 0 0 2 3 16:112 0 4 5 5 2 15 31 16:153 0 0 5 0 0 1 6 16:160 0 1 1 0 1 4 7 Total 2 23 58 26 26 141 276 WARNING: MORE THAN ONE-FIFTH OF FITTED CELLS ARE SPARSE (FREQUENCY < 5) 153 Table B.22 Basalt Debitage Type Frequencies by Site Size Factor Level (rooms) Tested Core Complete Broken Flake Shatter Flake Frag. Flake Cobble Z 0 0 0 3 0 0 1 4 2-5 5 10 29 7 15 67 133 7-10 0 2 2 2 1 9 16 14-15 0 15 41 19 15 87 177 200 0 1 7 4 1 14 27 Total 5 28 82 32 32 178 357 WARNING: MORE THAN ONE-FIFTH OF FITTED CELLS ARE SPARSE (FREQUENCY < 5) Table B.23 Diabase Debitage Type Frequencies by Site SjLze Factor Level (rooms) Tested Core Complete Broken Flake Shatter Cobble Flake Flake Frag. E 0 0 1 10 1 0 7 19 2-5 2 15 40 14 9 48 128 7-10 0 1 7 3 4 5 20 14-15 1 9 22 6 6 51 95 200 0 2 12 5 3 13 35 Total 3 28 91 29 22 22 124 WARNING: MORE THAN ONE-FIFTH OF FITTED CELLS ARE SPARSE (FREQUENCY < 5) 154 Table B.24 Silicified Wood Debitage Type Frequencies by Site Size Factor Level (rooms) Tested Core Complete Broken Flake Shatter Flake Frag. Flake Cobble 2 0 4 12 27 5 8 27 83 2-5 12 89 193 51 55 282 682 7-10 0 19 25 16 16 56 132 14-15 5 54 126 35 36 241 497 200 4 42 83 32 16 165 342 Total 25 216 454 139 131 771 1736 Table B.25 Chert Debitage Type Frequencies by Site Size Factor Level (rooms) 0 Tested Core Complete Broken Flake Shatter Cobble Flake Flake Frag. 2 3 6 22 7 6 67 135 292 105 118 7-10 8 30 34 25 25 118 240 14-15 16 89 193 104 88 488 978 200 8 42 195 94 52 316 707 102 304 736 335 289 2-5 Total 24 70 748 1465 1694 3460 155 Table B.26 Quartzite Debitage Type Frequencies by Site Size Factor Level (rooms) Tested Core Complete Broken Flake Shatter Flake Flake Frag. Cobble Z 1 2 11 1 0 5 20 2-5 15 38 87 37 30 118 325 7-10 6 17 11 4 1 22 61 14-15 7 39 102 36 18 172 374 200 1 8 20 12 3 18 62 0 30 104 Total 90 231 52 335 842 WARNING: MORE THAN ONE-FIFTH OF FITTED CELLS ARE SPARSE (FREQUENCY < 5) Table B.27 Silic ified Wood Debitage Type Frequencies by Site Size Factor Level Site Tested Core Complete Broken Flake Shatter 2 rooms Number Cobble Flake Flake Frag. 0 12:227 4 12 27 5 8 83 27 2-5 12:167 5 8 33 8 12 33 99 12:196 0 5 18 6 5 28 62 12:210 0 0 0 1 0 0 1 12:211 4 13 9 54 11 67 158 12:222 0 2 1 2 1 8 14 12:242 1 9 36 5 12 45 108 12:256 0 8 0 7 1 24 40 16:139 1 5 14 10 3 28 61 16:176 0 2 8 3 1 2 16 5-1201 1 28 25 13 9 47 123 7-10 12:65 0 8 2 2 7 26 45 12:260 0 17 17 14 8 80 24 16:160 0 0 0 0 1 6 7 14-15 12:248 0 5 13 6 14 52 90 12:277 0 6 6 31 7 31 81 16:90 0 0 4 15 0 5 24 16:112 5 20 15 23 41 128 232 16:153 0 26 4 8 25 7 70 200 11:1 83 32 16 165 4 42 342 Total 139 25 454 131 216 771 1736 Table B.28 Chert Debitage Type Frequencies by Site Size Factor Level Site Tested Core Complete Broken Flake Shatter E rooms Number Cobble Flake Flake Frag. 0 12:227 3 8 22 7 6 24 70 2-5 12:167 26 19 70 16 28 173 332 12:196 17 21 33 10 23 122 226 12:210 0 2 7 0 0 8 17 12:211 3 58 22 21 23 108 235 12:222 0 0 2 2 3 9 16 12:242 2 14 31 9 8 96 160 12:256 5 1 11 4 56 4 81 16:139 6 10 7 2 10 28 63 16:176 1 1 2 2 0 5 11 5-1201 44 68 11 39 19 143 324 7-10 12:65 11 6 1 12 7 44 81 12:260 6 13 16 14 54 117 14 16:160 6 6 1 5 20 4 42 14-15 12:248 18 2 44 31 187 303 21 12:277 31 68 4 28 30 118 279 16:90 6 2 14 8 16 1 47 16:112 33 3 50 27 26 141 280 16:153 5 10 1 17 10 26 69 200 89 193 104 11:1 16 88 488 978 335 Total 102 304 736 289 1694 3460 Table B.29 Quartzite Debitage Type Frequencies by Site Size Factor Level Site Tested Core Complete Broken Flake Shatter rooms Number Cobble Flake Flake Frag. 0 12:227 1 2 11 1 0 5 2-5 12:167 2 7 17 3 4 8 12:196 2 1 6 3 2 8 12:210 0 1 0 0 0 1 12:211 2 2 34 11 8 30 0 12:222 0 0 0 2 3 12:242 4 9 12 12 7 21 12:256 2 7 5 4 19 1 16:139 1 2 0 0 1 2 16:176 0 0 0 1 1 1 5-1201 2 9 11 6 3 25 7-10 12:65 2 5 0 0 5 1 12:260 4 10 0 7 1 11 16:160 0 2 3 3 6 1 14-15 12:248 4 5 50 16 11 96 12:277 0 5 15 0 3 11 16:90 0 0 0 5 3 1 16:112 3 27 3 28 16 57 16:153 0 2 3 1 5 4 200 11:1 1 8 3 20 18 12 Total 30 104 90 52 231 335 S 20 41 22 2 87 5 65 38 6 3 56 13 33 15 182 34 9 134 15 62 842 159 APPENDIX C: STATISTICAL NOTES Part A: Why these p-levels? The ability of a statistical test to significantly identify differences is called the power of the test (1-B). Power is inversely related to the confidence level a, which is the probability of avoiding Type I errors (i.e., rejecting a null hypothesis in favor of an alternate hypothesis, when in reality the null hypothesis is the true one). Increasing level-a diminishes the power of the test while diminishing the chance for error. Increasing the power of the test admits the possibility of committing more errors while discriminating more statistically significant differences (Kuehl 1994:84-106). When a group of statistical tests are performed, the probability that one or more of the tests will result in a Type I error increases. Often, when several tests are reported simultaneously, a higher a-level is set for each test in an effort to avoid this cumulative error effect. While the "matrix format" is often the context of setting a higher a-value, it bears some reflection. In the course of this manuscript, numerous tests of significance were performed. While not all of them were neatly summarized in a single matrix of results, the same laws of probability and additive error apply. The more tests performed in any single stuay on any set of data, the greater the probability of at least several erroneous results occurring. 160 As with any study, the researcher needs to consider their goals when making a-level choices. I have not chosen to set a study-wise error rate for the entirety of this manuscript for several reasons, the most important of which is the exploratory nature of this study and its goal of producing fruitful avenues of future inquiry. To increase the a-level of the whole study would decrease the power of the tests performed. While increasing a may diminish error rates, the ability to discriminate differences in this study would decrease. I would argue that more stringent a-level requirements should be in force when testing theories concerning human behavior and material culture which enjoy a histoiy of research and are of particularly intense interest. To sacrifice power in the preliminary stages of assemblage investigation may protect us from error, while blinding us to interesting lines of inquiry. Part B: Significance Testing of Spearman's rho The Spearman's correlation coefficient ranks all the values of the variables compared and calculates a coefficient from -1 to +1, which characterize the positive or negative correlation of the variables (Sokal and Rohlf 1981:584-586, 607). Significance testing in this study sought to reject the null hypothesis that Spearman's rho=0 (i.e., that no correlation between variables existed). Tables have been written for such significance testing, but exact values were calculated in 161 this study by using a t-statistic for n < 80 and a z-statistic for n > 80. These were calculated as follows: First the Spearman's correlation coefficient r was calculated by the statistical package SYSTAT. For n < 80, ts=iV(n-2)/(l-r2) The calculated t, was then compared against a cumulative t-distribution with n-2 degrees of freedom. If the calculated value exceeded t(n.j, „ =0.05)> the hypothesis of no correlation was rejected. For n > 80, t^zVn-3, where z is the z-transformation of the correlation coefficient r, available in statistical texts and tables. The probability of the critical value t, was then assessed using a cumulative z-distribution. Cumulative probabilities greater than 0.95 resulted in the rejection of the null hypothesis of no correlation. 162 REFERENCES Abruzzi, William S. 1993 Dam That River! Ecology and Mormon Settlement in the Little Colorado River Basin. University Press of America, New York. Abler, Stanley A. 1989 Mass Analysis of Flaking Debris: Studying the Forest Rather than the Tree. In Alternative Approaches to Lithic Analysis, edited by D. Hemy and G. Odell, pp. 85-118. Archaeological Papers of the American Anthropological Association no. 1, Washington, D.C. Amick, D.S., and R.P. Mauldin 1989 Comments on Sullivan and Rozen's "Debitage Analysis and Archaeological Interpretation." American Antiquity 54:166-168. Barton, C. Michael 1988 Lithic Variability and Middle Paleolithic Behavior: New Evidence from the Iberian Peninsula. British Archaeological Reports International Series 408, Oxford. Binford, Lewis R., and Sally R. Binford 1969 Stone Tools and Human Behavior. Scientific American 220(4):70-84. Blakey, Ronald C. 1989 Triassic and Jurassic Geology of the Southern Colorado Plateau. In Geologic Evolution of Arizona, edited by J.P. Jenney and S.J. Reynolds, pp. 369-396. Arizona Geologic Society Digest 17, Tucson. Blakey, Ronald C., and Rex Knepp 1989 Pennsylvanian and Permian Geology of Arizona. In Geologic Evolution of Arizona, edited by J.P. Jenney and S.J. Reynolds, pp. 313-347. Arizona Geological Society Digest 17, Tucson. Bohrer, Vorsila L. 1991 Recently Recognized Cultivated and Encouraged Plants among the Hohokam. The Kiva 56:227-235. 163 Brose, David S. 1978 A Model of Changing Subsistence Technology in The Late Woodland of Northeastern Ohio. In Lithics and Subsistence: The Analysis of Stone Tool Use in Prehistoric Economies, edited by D.D. Davis, pp. 87-115. Publications in Anthropology no. 20, Vanderbilt University, Nashville. Burton, John 1980 Making Sense of Waste Flakes: New Methods for Investigating the Technology and Economics Behind Chipped Stone Assemblages. Journal of Archaeological Science 7:131-148. Cartledge, Thomas R. 1977 Human Ecology and the Changing Patterns of Co-Residence in the Vosberg Locality, Tonto National Forest Central Arizona. Cultural Resources Report no. 17. USDA Forest Service, Southwestern Region, Albuquerque. Chapman, Richard C. 1977 Analysis of the Lithic Assemblages. In Settlement and Subsistence Along the Lower Chaco River: The CGP Survey, edited by C.A. Reher, pp. 371-452. University of New Mexico Press, Albuquerque. Clark, John E. 1991 Flintknapping and Debitage Disposal among the Lacandon Maya of Chiapas, Mexico. In The Ethnoarchaeology of Refuse Disposal, edited by E. Staski and L.D. Sutro, pp. 63-78. Arizona State University Anthropological Research Papers no. 42, Tempe. Collins, Michael B. 1975 Lithic Technology as a Means of Processual Inference. In Lithic Technology: Making and Using Stone Tools, edited by E. Swanson, pp. 15-34. Mouton Publishers, Paris. Cotterell, B., and J. Kamminga 1987 The Formation of Flakes. American Antiquity 52: 675-708. Crabtree, Don E. 1982 An Introduction to Flintworking, second edition. Occasional Papers of the Idaho Museum of Natural Histoiy no. 28, Pocatello, Idaho. 164 Dobyns, Henry F. 1988 Piman Indian Historic Agave Cultivation. Desert Plants 9(2):49-53. Ensor, H.B., and E. Roemer 1989 Comments on Sullivan and Rozen's Debitage Analysis and Archaeological Interpreation. American Antiquity 54:175-178. Ezzo, Joseph A. 1992 Dietary Change and Variability Grasshopper Pueblo, Arizona. Journal of Anthropological Field Archaeology 11:219-289. 1994 Paleonutrition at Grasshopper Pueblo, Arizona. In Paleonutrition: The Diet and Health of Early Americans, edited by Kristin D. Sobolik, pp. 263-277. Center For Archaeological Investigations, Occasional Paper No, 22, Southern Illinois University. Fewkes, Jesse 1904 Two Summer's Work in Pueblo Ruins: 22nd Annual Report of the Bureau of American Ethnology. Government Printing Office, Washinton, D.C. Fish, Paul 1979 Statistical Methods in Description and Taxonomy. In The Interpretive Potential of Mousterian Debitage, pp. 57-71. Arizona State University Anthropological Research Papers no. 16, Tempe. Fish, Suzanne K., and Marcia Donaldson 1991 Production and Consumption in the Archaeological Record: A Hohokam Example. The Kiva 56:255-275. Fish, Suzanne K., Paul R. Fish, Charles Miksicek, and John Madsen 1985 Prehistoric Agave Cultivation in Southern Arizona. Desert Plants 7(2):107-112. Francis, Julie E. 1978 The Effect of Casual Surface Collection on Variation in Chipped Stone Artifacts. In An Analytic Approach to Cultural Resource Management: The Little Colorado Planning Unit, edited by Fred Plog, pp. 114-132. Arizona State University Anthropological Research Papers No. 13, Tempe and USDA Forest Service Cultural Resources Report No. 19, Albuquerque. 165 Gallagher, James P. 1977 Contemporary Stone Tools in Ethiopia: Implications for Archaeology. Journal of Field Archaeology 4:407-414. Gallagher, Marsha 1977 Contemporary Ethnobotany among the Apache of the Clarksdale, Arizona Area: Coconino and Prescott National Forests. Archaeological Report no. 14, USDA Forest Service, Southwestern Region, Albuquerque. Gasser, Robert E., and Scott M. Kwiatkowski 1991 Regional Signatures of Hohokam Plant Use. The Kiva 56(3):207226. Gifford-Gonzalez, Diane P., David B. Damrosch, Debra R. Damrosch, John Piyor, and Robert L. Thunen 1985 The Third Dimension in Site Structure: An Experiment in Trampling and Vertical Dispersal. American Antiquity 50:803-818. Gently, Howard S. 1982 Agaves of Continental North America. University of Arizona Press, Tucson. Haury, Emil W., and Lyndon L. Hargrave 1931 Recently Dated Pueblo Ruins in Arizona. Smithsonian Miscellaneous Collections 82(11). Smithsonian Institution, Washington, D.C. Hayden, Brian 1978 Snarks in Archaeology; or Inter-assemblage Variability in Lithics (a view from the anipodes). In Lithics and Subsistence: The Analysis of Stone Tool Use in Prehistoric Economies, edited by D.D. Davis, pp. 179-198. Publications in Anthropology no. 20, Vanderbilt University, Nashville. Hayden, Brian, and Aubrey Cannon 1983 Where the Garbage Goes: Refuse Disposal in the Maya Highlands. Journal of Anthropological Archaeology 2:117-163. Hayden, Brian, and Margaret Nelson 1981 The Use of Chipped Lithic Material in the Contemporary Maya Highlands. American Antiquity 46:885-898. 166 Henry, Donald O. 1989 Correlations between Reduction Strategies and Settlement Patterns. In Alternative Approaches to Lithic Analysis, edited by D.O. Henry and G.H. Odell, pp. 137-155. Archaeological Papers of the American Anthropological Association no. 1, Washington, D.C. Herr, Sarah A. 1994 Great Kivas as Integrative Architecture in the Silver Creek Community, Arizona. Unpublished Master's thesis, Department of Anthropology, University of Arizona, Tucson. Hevly, Richard H. 1983 High-Altitude Cultural Resources in Arizona. In High Altitude Adaptations in the Southwest, edited by Joseph C. Winter, pp. 22-40. Cultural Resources Management Report no. 2, USDA Forest Service, Southwestern Region, Albuquerque Hill, James N. 1970 Broken K Pueblo: Prehistoric Social Organization in the American Southwest. Anthropological Papers of the University of Arizona no. 18, Tucson. Hodgson, Wendy, Gary Nabhan, and Liz Ecker 1989 Prehistoric Fields in Central Arizona: Conserving Rediscovered Agave Cultivars. Agave 3(3):9-ll. Hogan, Patrick, Janette M. Elyea, and Peter N. Eschman 1985 Intensive Lithic Analysis. In Economy and Interaction Along the Lower Chaco River, edited by Patrick Hogan and Joseph C. Winter, pp. 275-285. Office of Contract Archeology and The Maxwell Museum of Anthropology, Univerisity of New Mexico, Albuquerque. Hough, Walter 1903 Archaeological Field Work in Northern Arizona. The Museum-Gates Expedition of 1901. Report of the United States National Museum. Ingbar, E.E., M.L. Larson, and B.A. Bradley 1989 A Nontypological Approach to Debitage Analysis. In Experiments in Lithic Technology, edited by D.S. Amick and R.P. Mauldin, pp. 117135. British Archaeological Reports International Series no. 528, Oxford. 167 Johnson, Daniel M. 1976 Precipitation and Stream Flow in the Little Colorado River Basin, Unpublished Master's thesis, Department of Geography, Arizona State University, Tempe. Kitchens, Lariy J. 1987 Exploring Statistics: A Modern Introduction. West Publishing Company, Los Angeles. Koldehoff, Brad 1987 The Cahokia Flake Tool Industry: Socioeconomic Implications for Late Prehistoiy in the Central Mississippi Valley. In The Organization of Core Technology, edited by Jay K. Johnson and Carol A. Morrow, pp. 151-185. Westview Press, Boulder. Kuehl, Robert O. 1994 Statistical Principles of Research Design and Analysis. Duxbury Press, Belmont, California. Kuhn, Steven L. 1992 Blank Form and Reduction as Determinants of Mousterian Scraper Morphology. American Antiquity 57(1):115-128. Kunen, Julie 1994 Agave Cultivation and Seasonal Site Use in the Silver Creek Drainage. Paper presented at the 8th Mogollon Conference, El Paso. Leonard, Robert D., F.E. Smiley, and Catherine M. Cameron 1989 Changing Strategies of Anasazi Lithic Procurement on Black Mesa, Arizona. In Quantifying Diversity in Archaeology, edited by Robert D. Leonard and George T. Jones, pp. 100-108. Cambridge University Press, New York. Lightfoot, Kent G. 1981 Prehistoric Political Development in the Little Colorado Region, eastcentral Arizona. Ph.D. dissertation, Department of Anthropology, Arizona State Univeristy. 1984 Prehistoric Political Dynamics: A Case Study from the American Southwest. Northern Dlinois University Press, DeKalb. 168 Luedtke, Barbara E. 1992 An Archaeologist's Guide to Chert and Flint. Archaeological Research Tools 7, Institute of Archaeology, University of California, Los Angeles. i Lynch, Daniel J. 1989 Neogene Volcanism in Arizona: The Recognizable Volcanoes. In Geologic Evolution of Arizona, edited J.P. Jenney and S.J. Reynolds, pp. 681-700. Martin, Paul S., William A. Longacre, and James N. Hill 1967 Chapters in the Prehistory of Eastern Arizona, III. Fieldiana: Anthropology 57. Mauldin, R. P. 1993a The Relationship Between Ground Stone and Agricultural Intensification in Western New Mexico. The Kiva 58:317-330. 1993b Laboratory Analysis: Lithics. In The Chihuahua Lake Basin to the Chihuahua Desert: Archaeological Studies along the Arizona Interconnection Project Transmission Line Corridor; Data Recovery Report, edited by J. Simon Bruder, Section 5 pp. 29-103. Zuni Archaeology Program Report No. 347, Zuni Archaeology Program, Zuni Pueblo. Mauldin, R.P., and D.S. Amick 1989 Investigating Patterning from Experimental Bifacial Core Reduction. In Experiments in Lithic Technology, edited by D.S. Amick and R.P. Mauldin, pp. 67-88. British Archaeological Reports International Series no. 528, Oxford. Mills, Barbara J., Trixi Bubemyre, Doug Gann, Sarah Herr, Chuck Riggs, and Ruth Van Dyke 1993 Report of the 1993 Activities of the Univerisity of Arizona Archaeological Field School, Silver Creek Archaeological Research Project, Sitgreaves National Forest. Department of Anthropology, Univeristy of Arizona. Mills, Barbara J., Sarah Herr, Eric Kaldahl, Scott Van Keuren, and JoAnne Newcomb 1994 Reports of the 1994 Activities of the University of Arizona Archaeological Field School, Silver Creek Archaeological Research 169 Project, Sitgreaves National Forest. Department of Anthropology, University of Arizona. Mills, Barbara J., and JoAnne Newcomb 1994 Prehistoric Population Dynamics in the Silver Creek Area, East-Central Arizona. Paper presented at the 8th Mogollon Conference, El Paso. Minnis, Paul E., and Stephen E. Plog 1976 A Study of the Site Specific Distribution of Agave parryi of East Central Arizona. The Kiva 41:299-308. Nations, J. Dale 1989 Cretaceous History of Northeastern and East-Central Arizona. In Geologic Evolution of Arizona, edited by J.P. Jenney and S.J. Reynolds, pp. 435-446. Arizona Geologic Society Digest 17, Tucson. Neeley, Michael P., and C. Michael Barton 1994 An New Approach to Interpreting Late Pleistocene Microlith Industries in Southwest Asia. Antiquity 68:275-288. Odell, George H. 1989a Experiments in Lithic Reduction. In Experiments in Lithir Technology, edited by D.S. Amick and R.P. Mauldin, pp. 163-197. British Archaeological Reports Intentional Series no. 528, Oxford. 1989b Fitting Analytical Techniques to Prehistoric Problems with Lithic Data. In Alternative Approaches to Lithic Analysis, edited by D.O. Henry and G.H. Odell, pp. 157-182. Archaeological Papers of the American Anthropological Association no. 1, Washington, D.C. Olszewski, Deborah I. 1993 The Late Baradostian Occupation at Warwasi Rockshelter, Iran. In The Paleolithic Prehistory of the Zagros-Taurus, edited by D. Olszewski and H. Dibble, pp. 187-205. University Museum Press, University of Pennsylvania, Philadelphia. Olszewski, Deborah I., and Alan H. Simmons 1982 Tools for Thought: Some Comments on the Analysis of Puebloan Chipped Stone Assemblages. The Kiva 48:109-115. 170 Parry, William J. 1987 Technological Change: Temporal and Functional Variability in Chipped Stone Debitage. In Prehistoric Stone Technology on Northern Black Mesa, Arizona, edited by W. Parry and A. Christenson, pp. 199-256. Center for Archaeological Investigations, Occasional Paper no. 12, Southern Illinois University, Carbondale. Parry, William J., and Robert L. Kelly 1987 Expedient Core Technology and Sedentism. In The Organization of Core Technology, edtied by Jay K. Johnson and Carol A. Morrow, pp. 285-304, Westview Press, Boulder. Patterson, Leland W. 1981 The Importance of Flake Size Distribution. Contract Abstracts and CRM Archaeology 3:70-72. 1990 Characteristics of Bifacial-Reduction Flake-Size Distribution. American Antiquity 55:550-558. Patterson, L.W., and J.B. Sollberger 1978 Replication and Classification of Small Size Lithic Debitage. Plains Anthropologist 23:103-112. Patterson, Patience E. 1977 A Lithic Reduction Sequence: A Test Case in the North Fork Reservoir Area, Williamson County, Texas. Bulletin of the Texas Archaeological Society 48:53-82. Peirce, H. Wesley 1989 Correlation Problems of the Pennsylvanina-Permian Strata of the Colorado Plateau of Arizona. In Geologic Evolution of Arizona, edited by J.P. Jenney and S.J. Reynolds, pp. 349-368. Arizona Geologic Society Digest 17, Tucson Peirce, H.W., P.E. Damon, and M. Shafiqullah 1979 An Oligocene (?) Colorado Plateau Edge in Arizona. Tectonophysics 61:1-24. Plog, Fred 1981 Cultural Resources Overview: Little Colorado Area, Arizona. Bureau of Land Management, Phoenix. 1f 171 Prentiss, W.C., and E.J. Romanski 1989 Experimental Evaluation of Sullivan and Rozen's Debitage Typology. In Experiments in Lithic Technology, edited by D.S. Amick and R.P. Mauldin, pp. 89-99. British Archaeological Reports International Series no. 528, Oxford. Reher, Charles A., editor 1977 Settlement and Subsistence Along the Lower Chaco River: The CGP Survey. University of New Mexico Press, Albuquerque. Reid, J. Jefferson, editor 1982 Cholla Project Archaeology, Volume 2, The Chevelon Region. Cultural Resource Management Division, Arizona State Museum, Archaeological Series No. 161, University of Arizona, Tucson. 1989 A Grasshopper Perspective on the Mogollon of the Arizona Mountains. In Dynamics of Southwestern Prehistory, edited by Linda S. Cordell and George J. Gumerman, pp. 65-94. Smithsonian Institution Press, Washington, D.C. Ricklis, Robert A., and Kim A. Cox 1993 Examining Lithic Technological Organization as a Dynamic Cultural Subsystem: The Advantages of an Explicitly Spatial Approach. American Antiquity 58:444-461. Rolland, Nicolas, and Harold L. Dibble 1990 A New Synthesis of Middle Paleolithic Variability. American Antiquity 55:480-499. Rozen, K.C., and A.P. Sullivan 1989a Measurement, Method, and Meaning in Lithic Analysis: Problems with Amick and Mauldin's Middle-Range Approach. American Antiquity 54:169-175. 1989b The Nature of Lithic Reduction and Lithic Analysis: Stage Typologies Revisited. American Antiquity 54:179-184. Sackett, J. 1989 Statistics, Attributes and the Dynamics of Burin Typology. In Alternative Approaches to Lithic Analysis, edited by D. Henry and G. Odell, pp. 51-82. Archaeological Papers of the American Anthropological Association no. 1, Washington, D.C. 172 Satterthwait, Leonn 1976 The Chevelon Creek Area: Environmental Characteristics. In Chevelon Archaeological Research Project, edited by F.T. Plog, J.N. Hill, and D.W. Read, pp. 6-17. Department of Anthropology, University of California, Los Angeles. Schiffer, Michael B. 1987 Formation Processes of the Archaeological Record. University of New Mexico Press, Albuquerque. Sheets, Payson D., and Guy R. Muto 1972 Pressure Blades and Total Cutting Edge: An Experiement in Lithic Technology. Science 175:632-634. Shott, Michael J. 1986 Technological Organization and Settlement Mobility: An Ethnographic Examination. Journal of Anthropological Research 42:15-51. 1994 Size and Form in the Analysis of Flake Debris: Review and Recent Approaches. Journal of Archaeological Method and Theory 1: 69-110. Sokal, Robert R., and F. James Rohlf 1981 Biometry: The Principles and Practice of Statistics in Biological Research. W.H. Freeman and Company, New York. Spier, Leslie 1918 Notes on Some Little Colorado Ruins. Anthropological Papers of the American Museum of Natural History 18(4), New York. 1919 Ruins in the White Mountains, Arizona. Anthropological Papers of the American Museum of Natural History 18(5), New York. Stahle, David W., and James E. Dunn 1982 An Analysis and Application of the Size Distribution of Waste Flakes from the Manufacture of Bifacial Stone Tools. World Archaeology 14:84-97. Sullivan, Alan P., Ill, and Kenneth C. Rozen 1985 Debitage Analysis and Archaeological Interpretation. American Antiquity 50(4):755-779. 173 Thomas, David Hurst 1978 The Awful Truth about Statistics in Archaeology. American Antiquity 43:231-244. U.S. Department of Agriculture, Forest Service 1973 Environmental Statement: Mogollon Rim Area Land Use Plan. U.S. Department of Agriculture, Coconino, Sitgreaves, and Tonto National Forest, Phoenix. Van Buren, Mary, James M. Skibo, and Alan P. Sullivan III 1992 The Archaeology of an Agave Roasting Location. In The Maratia Community in the Hohokam World, edited by S.K. Fish, P.R. Fish, and J.H. Madsen, pp. 88-96. Anthropolgocial Papers no. 56, University of Arizona, Tucson. Wilson, Eldred D., Richard T. Moore, and Robert T. O'Haire 1960 Geologic Map of Navajo and Apache Counties, Arizona. Arizona Bureau of Mines, Tucson. Whittaker, John C. 1994 Flintknapping: Making and Understanding Stone Tools. University of Texas Press, Austin. Wrucke, Chester T. 1989 The Middle Proterozoic Apache Group, Troy Quartzite, and Associated Diabase of Arizona. In Geologic Evolution of Arizona, edited by J.P. Jenney and S.J. Reynolds, pp. 239-258. Arizona Geological Society Digest 17, Tucson.
* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project