OVERDUE FINES ARE 25¢ PER DAY PER ITEM Return to book drop to remove this checkout from your record. NOV 1 3 2004 120304 I n’. F \ h 9‘ ELY” N _ 3*; , 3» MAR 9223399 g“; OCT 3 0 2004 oilln‘ Q. s a: .- 0 © 1979 MICHAEL ROBERT POLK ALLRIGHTS RESERVED BISON JUMP SITES IN THE NORTHWESTERN PLAINS OF NORTH AMERICA: A LOCATIONAL ANALYSIS By Michael Robert Polk A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF ARTS Department of Anthropology 1979 ABSTRACT BISON JUMP SITES IN THE NORTHWESTERN PLAINS OF NORTH AMERICA: A LOCATIONAL ANALYSIS BY Michael Robert Polk This study is a locational analysis of bison jump sites in the northwestern plains of North America. One hundred forty-six sites from Alberta, Montana and Wyoming were examined in an attempt to identify cultural preferences and environmental constraints which affected the site location decisions of prehistoric hunters. Bison jump site data and associated environmental information including various soil types, geology, topography, vegetation and water source associations were partitioned into a set of quantified variables and subjected to a series of statistical procedures. The most critical environmental and cultural variables identified for site location through various tests for degrees of significance were water source association and jump face direction. An interpretive framework provides evidence that jump face direction is strongly associated with prevailing wind direction. This knowledge may provide information relevant to seasonal site use, subsistence strategies and population movements. It is suggested that site proximity to permanent water sources reflects the use of Michael Robert Polk associated broken topography for bison jumping, or the water needs of human groups and/or bison herds. ACKNOWLEDGMENTS The research and writing of this thesis was carried out over a period of years from 1975 to 1979. It was originally conceived and initial background research begun while I was a graduate student at Idaho State University. Under the direction of the late Dr. Earl Swanson, a literature search was made and a research trip undertaken to the Universities of Wyoming, Montana and Calgary and to the Saskatchewan Natural History Museum to obtain site data. Additional site information was obtained from various sources during the period of analysis and write-up at Michigan State University. During my tenure at Michigan State, the College of Social Science and Department of Anthropology graciously provided computer time for this project. A large number of persons contributed to this effort and I acknowledge a debt of thanks to them all. The help provided by my committee members is particularly appreciated. Dr. William A. Lovis, Chairman of my committee, gave much needed advice on statistical problems and provided a critical, yet patient, sounding board for both good and bad ideas. Dr. Charles Cleland, as a member of my committee, gave very generously of his time to fully discuss with me his appraisal of each chapter. I also gratefully acknowledge the help of a large number of persons who made data available to me at several institutions. I would iii like to thank Mr. Gil Watson of the Saskatchewan Natural History Museum, Dr. Richard Forbis and Dr. Brian Reeves of the University of Calgary, and particularly Dr. Floyd Sharrock of the University of Montana for allowing me access to their data files. I also extend my thanks to Dr. George Arthur of the University of Regina for many fruitful discussions on the topic and for steering me towards so many research sources. The help and stimulation offered by fellow students and colleagues at Michigan State University in this effort was also important and valued. I am particularly grateful to Peggy Holman for many interesting and helpful discussions and for her moral support in com- pleting this project. I also acknowledge a debt to Shela McFarlin for throwing me a last minute lifeline. Finally, I would like to extend belated appreciation to my father and late mother for their love and support during my many years at college. My deepest appreciation goes to my wife Ann and daughter Jamie for their support and understanding during some of the most trying times. iv TABLE OF CONTENTS LIST OF TABLES O O O O O O O O O O O O O O O O O O O O 0 LIST OF FIGURES O O O O O O O O O O O O O O O O O O O 0 INTRODUCTION 0 O O I O O O O O O O O O O O O O O O O O O The Problem . . . . . . . . . . . . . . . . . . Hypotheses O O O O O O O O O O O I O O O O I 0 Chapter I. II. III. IV. ENVIRONMENT OF THE NORTHWESTERN PLAINS . . . . . . Study Area . . . . . . . . . . . . . . . . . . Paleoenvironment . . . . . . . . . . . . . . . Present Environment . . . . . . . . . . . . . . Physiographic Considerations . . . . . . . Climate . . . . . . . . . . . . . . . . . . Flora and Fauna . . . . . . . . . . . . . . PREVIOUS INVESTIGATIONS . . . . . . . . . . . . . Early Research . . . . . . . . . . . . . . . . Research Since 1960 . . . . . . . . . . . . . . Bison Jump Sites in Other Areas . . . . . . . . METHODOLOGY 0 O O O O O O I O O O I O O O O O O 0 Site Selection Criteria . . . . . . . . . . . . Additional Considerations . . . . . . . . . . . Cultural and Environmental Variables . . . . . LOCATIONAL ANALYSIS: SITES AND THEIR ENVIRONMENTAL SETTINGS O O O O C O O O O O O I O O O O O O O O 0 Nominal and Ordinal Scale Variables . . . . . . Frequency Distributions . . . . . . . . . . Frequency Cross-tabulation . . . . . . . . Interval Scale Variables . . . . . . . . . . . Nominal, Ordinal and Interval Scale Variables: Statistical Comparisons . . . . . . . . . . . Chapter Summary . . . . . . . . . . . . . . . . V Page vii OOH 11 ll 17 18 21 21 22 31 35 36 37 45 59 59 59 67 89 93 101 Chapter V. CONCLUSIONS 0 O O O I O O O O O O O Hypotheses . . . . . . . . . . . Significance for Future Research APPENDICES Appendix A. Identification of Variables . . . . B. List of Variable Codes . . . . . . C. Computer Printout of Site and Environmental REFERENCES CITED . . . . . . . . . . . . vi Page 104 110 111 113 115 123 131 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. LIST OF TABLES List of Sites Used in Analysis . . . . . . . . . . . List of Environmental Map Sources and Scales . . . . List of Counties and Number of Sites Investigated in Montana and Wyoming . . . . . . . . . . . . . . . Frequency Distribution of ACE Values . . . . . . . . Frequency Distribution of TYPEFALL Values . . . . . Frequency Distribution of JUMPFACE Values: 1 . . . . Frequency Distribution of JUMPHEIT Values . . . . . Frequency Distribution of JUMPFACE Values: 2 . . . . Frequency Distribution of WATERONE Values . . . . . Frequency Distribution of WATERDIS Values . . . . . Frequency Distribution of PERMWATR Values . . . . . Frequency Distribution of DISTPERM Values . . . . . Frequency Distribution of VEG Values . . . . . . . . Frequency Distribution of LANDFORM Values . . . . . Frequency Distribution of AREATOPO Values . . . . . Frequency Distribution of PRSRTOPO Values . . . . . Frequency Cross-Tabulation: Landform Class by Vegetation Type . . . . . . . . . . . . . . . . . . . . . . . Frequency Cross-Tabulation: Landform Class by Predominant 8011 Type (at 81128) 0 o o o o o o o o o o o o o 0 Frequency Cross-Tabulation: Landform Class by Predominant Soil Type (in standard area around site) . . . . . . . . vii Page 38 46 57 61 61 62 62 65 65 68 68 69 69 7O 70 71 74 75 76 Table 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. Frequency Cross-Tabulation: Landform Class by Predominant Surficial Geologic Feature (at site) . . . . . . . . . . Frequency Cross-Tabulation: Landform Class by Predominant Surficial Geologic Feature (in standard area around Site) 0 O O O O O O O O O O O C O O O O O O O O I O O 0 Frequency Cross-Tabulation: Landform Class by Nearest water source 0 O O O O O O O O O O I I O O O O O I O O 0 Frequency Cross-Tabulation: Landform Class by Nearest Permanent Water Source . . . . . . . . . . . . . . . . . Frequency Cross-Tabulation: Predominant Soil Type by Predominant Surficial Geologic Feature (at site) . . . . Frequency Cross—Tabulation: Vegetation Type by Predominant Soil Type (in standard area around site) . . . . . . . . Frequency Cross-Tabulation: Vegetation Type by Predominant 8011 Type (at Site) 0 O O O O O O O I O O I O O O O O 0 Frequency Cross-Tabulation: Vegetation Type by Predominant Surficial Geologic Feature (at site) . . . . . . . . . . Frequency Cross-Tabulation: Vegetation Type by Predominant Surficial Geologic Feature (in standard area around Site) 0 O O O I O O O O O O O O O O I O O I O O 0 O O O Varimax Rotated Matrix of Factor Loadings: All variables 0 O O I O O I O O O O O O O O O O O O O O O O SPSS PA/2 Analysis (R-Mode): Factors, Eigenvalues and Percent Variance . . . . . . . . . . . . . . . . . . . . Breakdown Analysis: Number of Washes by Predominant Surficial Geologic Feature (in standard area around Site) 0 O O O O O O O O O O O O O O O I O O O O O O O O Breakdown Analysis: Number of Washes by Predominant Surficial Geologic Feature (at site) . . . . . . . . . . Breakdown Analysis: Number of Streams by Predominant Surficial Geologic Feature (in standard area around Site) 0 O O I O O O C O O O O O O O I O O O O O I O O O Breakdown Analysis: Number of Streams by Predominant Surficial Geologic Feature (at site) . . . . . . . . . . viii Page 78 79 80 81 85 86 87 88 91 91 95 95 96 96 Table Page 35. Breakdown Analysis: Percentage Area Occupied by Predominant Soil Type By_Predominant Soil Type §y_Vegetation Type O O O O O I O O O O O O O I O O O O O O O O O O O O O 98 36. Breakdown Analysis: Percentage Area Occupied by Predominant Soil Type §y_Predominant Site Soil Type §y_Predominant Surficial Geologic Feature . I O O O O I I O O I O O O O O 98 37. Breakdown Analysis: Percentage Area Occupied by Dominant Surficial Feature §y_Predominant Soil Type By_ Predominant Site Soil Type . . . . . . . . . . . . . . . . 100 38. Breakdown Analysis: Percentage Area Occupied by Dominant Surficial Feature §y_Landform Type §y_Predominant Type Of BedrOCk O O O O O C O C O O O I O O O O O O O O O 100 ix LIST OF FIGURES Figure Page 1. Map of the Northwestern Plains . . . . . . . . . . . . . . . 8 2. Landform Map of Alberta and Saskatchewan . . . . . . . . . . l3 3. Landform Map of Montana and wyoming . . . . . . . . . . . . 15 INTRODUCTION Late prehistoric and protohistoric bison hunting techniques in the Northwestern Plains of North America have been the subject of increasing interest in anthropological studies. In part such interest stems from the extensive technological, social, and economic impact that a single resource, bison, had upon many Plains Indian groups in the area. For many historically known groups including the Crow, Blackfoot and Gros Ventre, the bison provided their most important food source. Archaeological evidence suggests such dependence was probably similarly shared by prehistoric groups in the Northwestern Plains for the last several thousand years. The Problem Traditionally, prehistoric bison procurement research has stressed descriptive studies or problem-oriented research structured by the culture-historical framework of the area. Most studies have been site specific or regionally specific even when attempting to identify and explain patterns of behavior covering a much broader area. Those which provided comparative sections covering a large area of the Northern Plains have done so either superficially or quite selectively. While a symposium and a synthesis of the literature of one procurement method--bison jumping-~appeared in the early l960s (Malouf and Conner 1962; Hurt 1963) they were largely pioneering studies identifying little 1 more than the parameters of the subject and some of the areas in need of research. But in pursuit of answers to a variety of problems related to kill sites, archaeologists have overlooked or ignored a major research topic: the spatial patterning of bison kill sites over a broad geographic region. It has been perhaps overlooked because it is con- sidered too obvious--the assumption is that conditions favorable for particular types of bison kills, particularly jumps, are self-evident. It is thought that the most important research task begins after sites are located. But not all "favorable" locations contain bison jumps while others, which seem less favorable sometimes do. This suggests that less obvious environmental and cultural variables or particular sets of variables may have contributed to a chosen location for a jump site. In the following study I intend to pursue the proposition that there was a particular identifiable set of variables which controlled the optimal locations available for one type of bison kill--bison jumps. For the purposes of this study a bison jump is defined as "the physical location of a buffalo jump including lines of rock piles, cliff and area where the animals landed and were slaughtered" (Conner 1962: 57). This study focuses upon several questions all of which relate to the spatial distribution of known bison jump sites in the North- western Plains. The most important focus will be determining why the jumps were located where they were. Important here is identification of the cultural preferences and environmental constraints which affected site location decisions of prehistoric hunters. This study's analysis will also provide enough detailed spatial information to answer several corollary questions: (1) does the locational data provide significant patterning and, if so, is the pattern predictable? and (2) do the environmental and cultural characteristics of bison jumps remain consistent throughout the Northwestern Plains? Answers to these questions should suggest powerful behavioral explanations for aspects of prehistoric communal bison hunting. At this point it should be stated that the severely limited number of dated jump sites restricts speculation about temporal patterns which may result from comparisons of the few dated sites. Hypotheses This investigation will follow the lead of several recent studies which utilized spatial modeling to establish the existence of patterned relationships between archaeological sites and the environment and between sites and other sites (Martin 1977; Green 1973; Gumerman 1971; Holman 1978; Roper 1975). These studies which stem from geographic locational analysis (Haggett 1965) have clearly shown the utility of isolating characteristics of the natural and cultural environment to establish their relationship with archaeological site distribution. My study, however, differs from previous research in that I propose to analyze the locational pattern exhibited by only one type of special use site--bison jumps. A corollary goal of the analysis is the formulation of a predic- tive testable model of jump sites which can be projected into as yet unsurveyed areas. That is, establishing the correlation between sites and environmental features and projecting this knowledge into similar areas and even into previously surveyed areas to discover overlooked sites (Green 1973: 279). Two propositions of locational theory are vital for analysis and model-building in this study: 1. Sites were located so as to minimize the effort expended in acquiring required quantities of critical resources. 2. Sites were located with respect to critical on site resources. (Plog and Hill 1971: 12) The first proposition suggests that sites were located to gain access to a maximal number of needed resources with minimal effort. Since this analysis is concerned with a site type devoted exclusively to the exploitation of a single resource--bison, it might be assumed that those resources related to bison presence (soils, vegetation and possibly water) represented the most critical environmental variables in site location decisions. The fluctuating availability of bison in any one area and the fact that bison were driven from as far as 20 to 40 miles to a jump-off (MacGregor 1966; Schaeffer 1962; Forbis 1962: 63— 64), however, weighs against such a conclusion. The bison resource was an uncertain quantity and so probably represented a strong contributory factor rather than a dominant one. The second proposition refers to both potential on-site resources, such as water, and the immediate natural and cultural setting of the site. It is suggested that these variables were most critical in site location decisions because of the special topographic features required for jump sites and the relative permanence they offer for repetitive use. Thus, in light of the above propositions, it can be expected that this analysis will (1) show those variables most important to successful bison jumping (topography, geology and possibly water) to be the most critical to site location decisions and (2) show those variables most closely related to the bison resource (vegetation, soils) to be important, but dependent variables. Hypotheses: As a result of this study, it is expected that: Bison jump sites will show a definite predictable pattern most closely associated with topographic and geologic variables and possibly with water. While such a correlation could be antici- pated from cursory examination of jump sites, it is suggested that refined environmental parameters will be established which will provide vastly increased predictive potential. Critical environmental and cultural characteristics of bison jump sites will remain relatively consistent over the entire Northwestern Plains suggesting that, all things being equal, a most effective and productive method of killing bison was known over a large geographic area. CHAPTER I ENVIRONMENT OF THE NORTHWESTERN PLAINS Study Area The Great Plains of North America is a very cohesive environ- mental region. Physiographic, climatic and, to a large extent, biotic features distinguish it from adjacent regions including the Rocky Mountains to the west, the Parkland to the north, the Gulf lowlands to the south and, to a lesser extent, the tall grass prairies and wood- lands to the east. Scientists have, for purposes of discussion, found it valuable to divide the Plains into several sub-regions. This study will concen- trate on that portion of the Plains commonly identified as the North: western Plains. The Northwestern Plains (Fig. l) have themselves been defined several different ways in the past (Wedel 1961; Conner 1968; Frison 1978). For the purposes of this study a modified version of Stuart Conner's division was found most appropriate since it effectively incorporates the greatest concentration of bison jump sites and is not pretentious (Conner 1968: 13). Conner states that the boundaries are primarily " . . . a basis for discussion and not a definition" (1968: 13). In this scheme the Northwestern Plains constitute that portion of the short grass plains which lie between the Northern Rocky Mountains and the mixed grass prairie to the east. Politically, it includes Figure 1. Map of the Northwestern Plains. " . . . the southern one-half of Alberta and Saskatchewan, the eastern two-thirds of Montana, and the northern third of Wyoming" (1968: 13). Additionally, the extreme western portions of North and South Dakota and the eastern flanks of the Northern and Central Rocky Mountains are included (Hunt 1974). The latter regions encompass the large river valleys west of the Continental Divide in Montana and the large basins and valleys in north-central wyoming. These areas share many of the features of the Plains and, in fact, tend to blend imperceptibly into the Plains setting. Paleoenvironment Several models have been postulated which describe and explain Holocene environmental conditions in parts of western North America, two of which have gained particular attention. Antevs (1955) has postulated a model of gradual climatic change for the region which has traditionally provided the paleoenvironmental model for western North American archaeologists. The most significant part of this reconstruction is a gradual warming trend thought to have climaxed about 6000 years ago. Because Antevs' supportive data was primarily from the Great Basin, however, the degree to which the model is applicable to the Plains is largely unknown. Another more recent model of the Holocene climate postulates intermittent periods of environmental stability followed by rapid climatic shifts. Its principal advocate, Reid Bryson (Bryson and Wendland 1967; Bryson and others 1970) suggests that there is a corre— lation between particular climatic variables and biotic regions. When climatic variables shift, there appears to be a corresponding shift in 10 biotic regions. A cause and effect relationship has not been demon- strated, however, and, as Bryson emphasizes, the model may be pre- mature (1967: 296). The time span of major interest to this study is the last 2500 to 2000 years, since the vast majority of dated jumps fall within this time period. Bryson and Wendland (1967) propose that this period, called the Sub-Atlantic, was a time when biotic conditions were similar to the present, with only minor fluctuations primarily affecting the boundaries of biotic regions. Antevs' model postulates a gradual moderation of climatic conditions from the peak of the Altithermal 6000 years ago to today. Other environmental studies, drawing on pollen samples from southern Alberta (Wormington and Forbis 1975: 121) and northern Wyoming (Haynes 1965: 209), suggest that these portions of the Northwestern Plains have enjoyed a relatively stable climatic period for the last nine to ten thousand years. Thus, the consensus of the present evidence suggests the existence of a relatively stable environ- ment on the Northwestern Plains over at least the last 2500 years. In light of the above discussion, it is suggested that a detailed study of modern environmental variables at jump sites will generally reflect conditions prevalent at the sites within at least the last 2500 years. Some increase or decrease in erosional activity, due to unusually severe local climatic conditions or relatively recent grazing or other land-use practices, may have altered topographic variables and water availability at some locales. The assumption is, however, that general site settings have remained intact over time. 11 Present Environment Physiographic Considerations Contrary to the reports of many observers who characterized the Northwestern Plains as a level, windswept expanse, its topography is actually quite diverse. While a large part of the area consists of level plains, the isolated mountain ranges, high ridges, eroded badlands and river valleys serve to break up the level topography and provide a variety of micro-environments (Figs. 2 and 3). The Northwestern Plains lies within two major river drainages: the Saskatchewan and the Missouri. The Saskatchewan Drainage encompasses most of the Alberta and Saskatchewan Plains in Canada (Fig. 2). Lying at elevations between 2500 feet above sea level to 3500 feet a.s.l. and 1500 feet a.s.l. to 2000 feet a.s.l., respectively, these plains are characterized and modified by isolated mountain ranges, deeply incised river valleys (200 to 400 feet deep) and various glacial features such as drumlins, moraines, erratics and drainage channels (Hunt 1974: 328; Bostock 1976: 20). The Missouri Drainage is characterized by glacial features north of the Missouri River proper, and unglaciated terrain south of it. The unglaciated region consists of broadly terraced river valleys between which lie high, widely alluviated plains (Fenneman 1931: 63). The entire region, both north and south of the river, also con- tains a number of isolated mountain ranges, deeply incised river valleys, and eroded badlands. The mountains and river floodplains were often well-wooded in the past, providing diverse animal and plant resources for prehistoric inhabitants. 11 Present Environment Physiographic Considerations Contrary to the reports of many observers who characterized the Northwestern Plains as a level, windswept expanse, its topography is actually quite diverse. While a large part of the area consists of level plains, the isolated mountain ranges, high ridges, eroded badlands and river valleys serve to break up the level topography and provide a variety of micro-environments (Figs. 2 and 3). The Northwestern Plains lies within two major river drainages: the Saskatchewan and the Missouri. The Saskatchewan Drainage encompasses most of the Alberta and Saskatchewan Plains in Canada (Fig. 2). Lying at elevations between 2500 feet above sea level to 3500 feet a.s.l. and 1500 feet a.s.l. to 2000 feet a.s.l., respectively, these plains are characterized and modified by isolated mountain ranges, deeply incised river valleys (200 to 400 feet deep) and various glacial features such as drumlins, moraines, erratics and drainage channels (Hunt 1974: 328; Bostock 1976: 20). The Missouri Drainage is characterized by glacial features north of the Missouri River proper, and unglaciated terrain south of it. The unglaciated region consists of broadly terraced river valleys between which lie high, widely alluviated plains (Fenneman 1931: 63). The entire region, both north and south of the river, also con- tains a number of isolated mountain ranges, deeply incised river valleys, and eroded badlands. The mountains and river floodplains were often well-wooded in the past, providing diverse animal and plant resources for prehistoric inhabitants. 12 Figure 2. Landform Map of Alberta and Saskatchewan--after Bostock, 1970. Scale 1: 5,000,000 13 or: omo mmbfim omtz: @ om: om_P 2.4.5 8 E r\.. V G S A o L R H. H 5 Z_<._n_ m M. @ o «u .. «r E a 2.4.5 mmm gene camcmsm :anmnmmEmnwmmm annexe filo sans um>mam mnxaha oao «Icahn moo gunman woo Aqaoav amaze Hauflmwa Boo Noalomwa coo onuuomn moo oH-3owa soc N-aowa moo Hau>oma moo mash oofisoo mauufiwpm> cml>own goo umawwwwmwm .WEMVMJWDMWI Nummsmw. rad. .mflglwwww. pops—:2 mmmo. mHm>Qoamuc mom an em wmo A.c.:v Monaco maaoh onmwom ocw>oamuu New mm am fine Amaaac sous mcHUmez .A.n.:v umacoo mason oammwsm mcfi>mamuo Hem mm «N omo Amcnoav mo>wom mmlmmho Nmo Amqncav mo>mmm mlamno Hmc ooHIHmnm omo HIfimnm mqo HI>Opm wqo Hummam moo amelxmxm oqo Huxmxm mco nlmmxm «co Huumnm mqo Hucmwm Nqo anmwm qu Hunmam oqo «nomcm mmo mesa canon mafioESh Hiamwm wmo mauaawm amo zoommmwm alwawm emo cam comwm xmmuo wwmum Huamwm mmo mqsammm «mo mmnamwm mmo Nixmmm mmo ~nwmcm Hmo omlamom omo mwlcmwm omo malamom wmo ,mmwmwmmmwm, .lwammleWWI .Nwwsoo no umnaoz muww Honasz mmmo A.D.ucoov H mamc dash oamwmsm pmonzuo dash onwwzm waonox mmeah onwwsm mca>mamu0 maash onmmsm ocw>mamu0 '1‘ "'| I msmz muam m0m <0 «N «HM\«OH <0 «N H m: «N How an «N mooH 3a «N NOON m0 «N «MN m0 «N ovmommu NHOH <0 «N NooH <0 «N van MON 04 «N H« <0 «N mm <0 «N ma <0 «N NH <0 «N «OOH H0 «N mooH 90 «N mom m0 «N can new no «N cue: mam 00« mm «N mNNH an «N n«oH an «N moN mm «N «0N mm «N "| "x‘l- zucsoo no ponesz.0ufim a.s.uaoov H mamusaoo mswumm CHSuNS mHHmsuom mum mmuwm 03u mmozhk HoH >m «N HmH NooH Hm «N ONH HHHN mmH58 :mvnmo «Hm Hm «N «NH HHNmHV N6m HHHN :omHN Nuaanmm «HoH an «N NNH HmNoHv aom NHoH mm «N NNH mo« an «N «NH HHHM mwmummum NN an «N mNH HHHN “mumsouHsz mm as «N «NH mmm «a «N NNH sass oncwam «mo: auanamm Nmm «m «N NNH HoomHv Nanuu< "HNomHV zone maHoHcmz mesa mwuHm N:OH NHN\NHH «m «N HNH HmmmHv aoHHsz "HN«AH0 spam “HNmsHV esopm HeomHv acuamz "HoemH .NosHv usnuu< "HmoNHV sumxumum mesa onmcsm ocmuwHEN mom m0« m: «N «HH «o« m: «N mHH sash ammuo auuHN N«« m: «N «HH sass onHH=N smnwch NN\ON\HoN m: «N NHH «« m: «N NHH HoooHv pasuu< ease comuwumm «mm mo HmNoHV mH>ma AmomHV whom “A.v.:v mwsma A.w.c0 mHsoH AmcaHv ausm AmooHv comm: AoooHv uscuu< AwooH .NoaH .oNooHv umcaoo mwocmpmwo%i muHm muo> HHHM xmouu mmoo0 mam ouHm xmmu0 kwaHm mesa onwwsm zooucmH0 mesa somHm «mono mmHaomm mean onwwam :Hmmm owmm seas onccsm me NHN HHHM :omfim onumch HHHx xomuo noon \oesh onwmsm smousono HHHM swanmmo Ne mass onwwsm xwmuo vacono2 QESN xmmuo umasHH mam HHHM onmmsm mcHEBSU HHHx onmwsm Hmuoosw Nam dash socmm mesh onmmam sass onucsm cwomx mammlwme Nom MHm Nam «om HooH «OH NoH mmm 0mm Nmm ma ma HooH Nmm No« wo« Ho« M0 w« mm w« ow w« 00 m« Hm «N An «N an «N A3 «N A3 «N A3 «N A> «N A> «N neuoa GOuwH :oumH 3m «N 3m «N 3m «N 3m «N Hm «N kuclmw Ho HmnEDZ muwm H.«.u:oov H mHm :HumHHmo Ho mHHom ONH.m«N.«uH muumnH< No coHumumwo> may ooo.00N.th monopH< No :oHumumwm> use mmumum woufica macawshmuco0 coo.on.muH was No :oHumumwo> Hmusumz HmHusmuom mwmww mamazmmm mm4<0m Qz< mmomDCm m<2 Azm no HmHA N mam GOHumuowm> coHumumwo> xuowmumo 47 HoNoHV xooumom AmmaHv cmvH< HNmoHV ame< Hm«mH0 mumnuo «am HN«mH0 AH«mHV HNmoHv AmmmHv muocuo mumsuo muocuo muonuo can «so «so cam A«Nmav AHmmHV mumcuo wan Hmm3om uumwz uum>3 uumzz pom»: Hummz uumzz Ammmvaumxuo vow chose IIIIII" . mousow AmNaHV mumzuo cam meumd ooo.ooo.muH coo.OOmHH 000.comna owo.omHuH cmo.oaHnH owc.omHuH owo.omHnH owo.omHuH owo.omH"H owo.cmHnH ocm.mouH owm.«cmuH «am ooo.«NuH mHmom mvmcmo mo mcowwmm oHcamuonmkzm mmou< ucoomnw< cam manuaoz snowmoB mo xonom0 HmHomHo can madmquHm>£m mmou< unmounv< cam msmucoz cuoummm mo >w0Hom0 HmwomH0 mam >camuonmhsm muumnH< .ummcm Home «mm was No mHHom muumnH< .mummsm Human «an «nausea was No mHHom muumnH< .mummnm >pmmHm0 can uoomxomHm wnu mo wHHOm muumnH< .ummam um>HN NHH: was No mHHom muumnH< .smmsm mHHHm aaHmm was No mHHom mononH< .mumonm Hococwm can mmvfiunsuog can 00 mHHom muumnH< .ummnm «omHuz was 00 mHHom wcfiEo>3 .zucsoo cmvfiuosm mo mHHom mcmucoz .mou< hucsoo :oaum0 mo mHHOm mymz mum H.«.usoov N NHN£amuwoaoe mawom wHHom mHHom mHHom mHHom mHHow wawmumw 48 AmwomHV :mHHH HNcaHV anuH Ao«mH0 was: AH«mHV was: HHmmHv beam HHmmHV mmeaon HwomHV mumwumms HNoaHv ummeum AoooHv ummeum HomoHV meHmum HNmmHV ummeum HomoHv ummeum A«0¢Hv psoEEm: ' ll. ,' 'I'l..u.llt mousom c««.mmNnH uonoHOh .>onow0 o««.mmNnH owwHuncqu .honom0 oom.mouH xmmu0 somEHum mo >onom0 ocm.man xompo wwmum wo zonooc ocm.mouH amonm zmHHm> Medusa mo xonoo0 ooo.o«"H «HumnH< .xmmuo umsusHm Ho aonomu cco.omNuH mHHHm mmoua>0|umoEoHom .%onoo0 o««.mmNuH mononH< .mwkunnuoH .konowu o««.mmNuH .HoHuumumluooc 0mm .Nonoo0 o««.mmNnH muuoaH< .wooqomz uuom .>onomc o««.mmN H muumnH< .pm>Hm anm .Nonomo o««.mmNuH mononH< .uoxmmwom .xonom0 ooo.oco.muH Hmwowwuam HmHUHNunm muumnH< HmHUHuusm HmHUHNuam HmHUcham HmHUHNusm .< .m .3 .mmumum uanmuuuuom ocu :H Euom ecumusmlucmd mo mommmHo mHmom onom am: H.u.u=ouv N NHNwo~om0 xonom0 honow0 >onom0 >woaom0 HmHUHNusm HmHUHNNsm HmHocham HmHUHuuam HmHoHHusm HmHUHHusm Euowvcwq .Nmowmumo 49 AmmoHv vcwxufi3 vcm mmom AmmmHV muwcuo «cm m>0H HamcaHv anuH u'llll' IOJ‘ mousom oco.oomnH oco.oom"H o««.mmNnH t IO! II"- -onom mcmuco2 mo am2 owonom0 wCHEomz mo >moHom0 amnonH0 .monom0 I! it"- osmz am2 H.«.uaoo0 N mHmuowwum0 50 Group One includes 12 nominal and ordinal scale variables. A portion of these served the purpose of record keeping and data standardization, while the remainder are cultural and site specific variables. All of the data for these variables were derived from site source materials, thus representing second-hand information provided by a large number of different individuals. Group Two includes 16 dicho- tomous and multistate nominal scale variables providing information on a number of environmental variables related to bison kill sites. These variables serve to eliminate extraneous information from the data and identify important variable trends. They also provide the primary data for case comparison and the identification of critical variables through cross-tabulation and statistical procedures. Group Three included eight interval scale environmental variables which provide the finest measure with which to compare cases and identify critical variables. Together with Group Two variables, they provide the infor- mation upon which the predictive statements rest. Group One variables were gathered almost entirely from site forms, unpublished manuscripts and published reports. In some instances, additional information was obtained from appropriate maps. These variables include STATE, denoting the state or province where the case was located, and COUNTY, indicating the county in the United States where the case was found. Cases from Alberta were excluded from this category and instead were given the designation 99. PRESEARCH denoted whether a site had been previously researched and TYPEPR indicated the type of previous research known. Previous research was not indicated unless the site was recorded as part of a systematic survey or other project for which a manuscript was written or for which literature 51 review was later given. SOURCE indicated where the information was derived for each case. LOCRELY denoted a measure of locational pre- ciseness for each case. This variable was used to test the relationship between progressively finer locational measurements and bison jump site settings. The next 6 Group One variables represent a range of identifiable cultural and site specific manifestations extant at bison jump sites. These variables were treated with caution since they are obtained from a wide variety of sources which contain some indirect measurements. Additionally, many of the site record sources did not identify all variables. Nevertheless, some trends were identified among these important aspects of jumps. The ACE acronym referred to the cultural period(s) of each case. While the age of only a minority of cases were known, those so identified were directly or indirectly measured by radiocarbon dating, geochronology or associated diagnostic artifacts. Thus, unlike other cultural variables, AGE represents a relatively reliable measurement. It was decided to group ages into the widely accepted culture-historical chronology suggested by Mulloy (1958) due to both the small number of dated cases and the broad time frames provided for many of the sites. Important modifications to this scheme based on more complete data have been recently suggested (Reeves 1970). For the purposes of this study the modifications would not be significant since they refer largely to the Middle Prehistoric Period, for which few cases are known. ASSOCAMP and DRIVLANE are presence-absence variables denoting the presence of a prehistoric campsite near a jump and drive lanes above the base of the fall. Campsites may or may not be directly associated with 52 the jumps they are near. In most cases an association is impossible to prove without excavation. Thus, the variable merely measures the degree of association between habitation sites and jump sites. Drive lanes provide a much clearer measure of direct association since the only Plains sites known to contain converging linear ground configurations are animal kills. For this reason, the existence of drive lanes alone, without an associated kill deposit, is sufficient to warrant a site's inclusion as a case. The remaining three Group One variables all concern the jumpoff location of sites. TYPEFALL identifies the type of fall represented. JUMPFACE denotes the direction in which the jump faced and hence the direction the bison were run in the final moments of the drive. JUMPHEIT refers to the height of the jumpoff. This variable was usually esti- mated so it was considered most appropriate to use grouped distances. The fall distance denoted the vertical distance for cliffs and cutbanks (as opposed to the combined vertical drop and steep incline following the drop, if any) and the incline distance for steep 310pes. This deci- sion was made to include the largest number of cases possible. Sources on most cases identifying cliffs and cutbanks provided a measurement of vertical drop, not the entire fall length. Group Two variables form the bulk of environmental identifiers, comprising 16 nominal scale variables. The variables represent five major environmental groupings including water resources, vegetation, soils, surficial land formations and bedrock formations, all gathered from a wide variety of environmental sources. Five of the variables concern water resources. Data for water variables was derived from 15 and 7% minute U.S. Geological Survey Quadrangle maps and topographic 53 maps of the Canadian Department of Energy, Mines and Resources (Scale l:50,000 and some l:125,000). WATERONE AND WATERTWO variables denote the nearest water source to each case and the next nearest water source to each case. PERMWATER concerns the nearest available permanent water source and WATERDIS AND DISTPERM refer to the distance to the nearest water and the distance to the nearest permanent water, respectively. It would have been preferable to provide exact distances from site to water sources, but this was precluded because only approximate locations were given for most sites. Thus, all distances to water were measured from the center of the most precisely given location for the site. For instance, for a site located within a k section, distances to water were measured from the center of the % section. The use of this proce- dure prompted grouping distance measurements into 400 meter and larger increments. The VEG variable denoted the natural vegetation type which occurs at each site. For cases within the United States Kuchler's Potential Natural Vegetation of the Coterminous United States (1964) was used while information on Canadian cases was derived from Lewis, Dowdling and Moss' The Vegetation of Alberta,pII (1928) and Moss' The Vegetation of Alberta, IV (1932). Four environmental variables concerned soil classification. Soil information was derived from reports and maps of the U.S. Soil Conservation Service, several Bulletins of the Montana Experiment Station, and soil survey Bulletins of the University of Alberta, College of Agriculture. SITESOL denotes the soil type predominant at the site while AREASOL indicates the soil type within the standard area around the site. Logistical problems posed by the wide variety of soil survey 54 reports used resulted in inclusion of only the most basic soil texture classes which appear in all reports. CORSESOL and STONISOL are two dependent variables denoting the relative coarseness of the predominant soil in each standard area and the relative stoniness of the soil in each standard area, respectively. CORSESOL and STONISOL were chosen for the purpose of exploring possible relationships with vegetation types. LANDFORM denotes the major geographic landform for each case as designated by Hammond (1964) in his Classes of Land-Surface Form in the Forty-eight States, U.S.A. (Fig. 2) and Bostock (1970) in Physiographic Regions of Canada (Fig. 3). This variable served to sort out major landform trends associated with bison jump sites. Following LANDFORM were three related topographic variables SITETOPO, PRSRTOPO and AREATOPO. SITETOPO refers to the surficial geologic features at each site, PRSRTOPO denotes the predominant surficial geologic feature at the site and AREATOPO identifies the predominant surficial geologic feature within the standard area. For sites in the United States William C. Alden's Physiography and Glacial Geology of Eastern Montana and Adjacent Areas (1932) and Physiography and Glacial Geolpgy of Western Montana and Adjacent Rggions (1953) were used, while in Alberta a series of surficial geologic maps produced by the Geological Survey of Canada and the Research Council of Alberta were used-7 The final two Group Two variables concern bedrock geology. SITEROCK denotes the predominant bedrock type occurring at each site and AREAROCK refers to the predominant bedrock type occurring within the standard area. The U.S. Geological Survey geology map of Montana (Ross and Witkind 1955) and of wyoming (Love, Weitz and Hose 1955) were used to plot cases within the United States. A series of Geological 55 Survey of Canada bedrock geology maps were used for cases in Alberta. Group Three data includes eight interval scale variables covering three major environmental areas. NUMWASH and NUMSTRM refer to counts of intermittent watercourses and perennial watercourses, respectively, over three kilometers in length within each standard area. These variables were established to test for a potential rela- tionship between stream density and bison jump location. The three kilometer length requirement for each watercourse was an attempt to reduce distortion in the data. In this case the distortion was fre- quently due to the existence of dozens of very short streamlets within a standard area. These streamlets drain minimal surface area and thus, if measured with longer watercourses, would distort the poten- tially greater effect longer streams have on topography and water resources within the standard area around each site. NUMSOL AND PERCTSOL denote aspects of soil type. The NUMSOL variable refers to a count of soil classes in the standard area around each case. This count was made to establish the density of different soil types within a standard area. PERCTSOL is the percentage of the standard area around a site occupied by the predominant soil type. This percentage, similar to that measured for surficial geologic features (see below), were obtained by overlaying gridded paper on appropriate maps and counting squares located within the desired soil class. Counts were made using 10 millimeter to the centimeter mesh graph paper. The remaining four interval scale variables concern some aspect of surficial topography. ELEV denotes the elevation of each site in 56 feet. The high and low measurements for each case were averaged to arrive at the given figure. NUMTOPO was a count made of the number of surficial geologic features within each standard area. AROCTOPO is the percentage of each standard area occupied by the dominant surficial geologic feature and STOCTOPO is the percentage of the standard area occupied by the surficial geologic feature predominant at the site. 57 TABLE 3 LIST OF COUNTIES AND NUMBER OF SITES INVESTIGATED IN MONTANA AND WYOMING County State Smithsonian Number Number of Sites Big Horn Montana 24 BH 9 Big Horn Wyoming 48 BH 0 Blaine Montana 24 BL 3 Broadwater MOntana 24 BW 0 Campbell Wyoming 48 CA 0 Carbon Montana 24 CB 1 Carter Montana 24 CT 2 Cascade Montana 24 CA 7 Chouteau Montana 24 CH 2 Converse Wyoming 48 CO 1 Crook Wyoming 48 CK l Custer Montana 24 CR 0 Daniels Montana 24 DN 0 Dawson Montana 24 DW 1 Fallon Montana 24 FA 0 Fergus Montana 24 FR 1 Fremont Wyoming 48 FR 0 Gallatin Montana 24 GA 4 Garfield Montana 24 GF 0 Glacier Montana 24 GL 2 Golden Valley Montana 24 CV 0 Hill Montana 24 HL 8 Hot Springs Wyoming 48 HO 0 Jefferson Montana 24 JP 3 Johnson Wyoming 48 JO 1 Judith Basin Montana 24 JT 0 Lewis and Clark Montana 24 LC 6 Liberty Montana 24 LT O McCone Montana 24 MC 0 Madison Mbntana 24 MA 5 Meagher Montana 24 ME 7 Musselshell Mbntana 24 ML 2* Natrona Wyoming 48 NA 0 Niobrara Wyoming 48 NO 0 Park Montana 24 PA 7 Park Wyoming 48 PA 0 Petroleum Montana 24 PT 0 Phillips Montana 24 PH 5 Pondera Montana 24 PN 0 Powder River Montana 24 PR 0 *These two sites are actually in Fergus County TABLE 3 (Cont'd.) 58 County State Smithsonian Number Number of Sites Prairie Montana 24 PE 0 Richland Montana 24 RL 1 Roosevelt Montana 24 RV l Rosebud Montana 24 RB 0 Sheridan Montana 24 SH 0 Sheridan Wyoming 48 SH 1 Stillwater Montana 24 ST 1 Sweetgrass Montana 24 SW 4 Teton Mbntana 24 TT 3 Toole Montana 24 TL 0 Treasure Montana 24 TE 0 Valley Mbntana 24 VL 2 Washakie Wyoming 48 WA 0 Weston Wyoming 48 WE 0 Wheatland Montana 24 WL 3 Wibaux Montana 24 WK 0 Yellowstone Montana 24 YL 0 Yellowstone National Park Montana 24 YE 0 Yellowstone National Park Wyoming 48 YE 0 CHAPTER IV LOCATIONAL ANALYSIS: SITES AND THEIR ENVIRONMENTAL SETTINGS The thirty environmental and cultural variables used in this study were subjected to a series of univariate and multivariate tests in order to determine their value as predictors of site location. Several SPSS subprograms (Nie and others 1975) were used to determine the distribution of individual variables and to explore the behavioral characteristics of sets of variables. Following is a discussion and interpretation of the results of those tests. Nominal and Ordinal Scale Variables Frequency Distributions The SPSS subprogram FREQUENCIES was generated to determine the degree of correspondence between case locations and various values of cultural and environmental variables. Absolute and adjusted frequency distributions were computed for all 146 cases on the nominal scale dichotomous and multistate variables and ordinal scale multistate variables. These variables included AGE, ASSOCAMP, DRIVLANE, TYPEFALL, JUMPFACE, JUMPHEIT, WATERONE, WATERTWO, VEG, SITESOL, AREASOL, CORSESOL, STONISOL, LANDFORM, SITETOPO, AREATOPO, SITEROCK, AREAROCK, PRSRTOPO, WATERDIS, PERMWATR AND DISTPERM (See Appendices A and B for a 59 60 description of these variables and their relevant values). Frequency distributions of most of these variables are displayed in Tables 4 through 15. Prominent frequencies and percentages are underlined. Inspection of the frequency distributions indicate that the variables exhibit a wide range of variability. Cases frequently asso- ciate with two or more different values in some variables and with only one in others. Among the Group One variables this pattern is clearly evident. Cases overwhelmingly associate with the Late Prehistoric Period value in the AGE variable and the bluff or cliff value in the TYPEFALL variable (Tables 4 and 5). On the other hand, cases frequently occur in association with several values of the JUMPHEIT and JUMPFACE variables (Tables 6 and 7). The strong association of cases with the Late Prehistoric Period appears to reflect the actual temporal distri- bution of sites, but their overwhelming association with bluffs and cliffs is partially the result of researcher error. Information per- taining to the type of fall for each site and all other descriptive information concerning sites were obtained from site records and other written sources. Written observations concerning the type of fall present were not standardized; rather, they varied with the knowledge and background of each observer. As a result, the terms "bluff" and "cliff" may refer to a wider variety of tOpographic situations than is actually the case. The data for other variables AGE, ASSOCAMP, DRIVLANE, JUMPHEIT and JUMPFACE was derived from the same sources as that for TYPEFALL. But these variables have considerably more credibility because they are not subject to interpretation, i.e., they are quantifiable. The variable ASSOCAMP, which measured the presence or absence of campsites associated with jump site locations, and the variable DRIVLANE, 61 TABLE 4 FREQUENCY DISTRIBUTION OF AGE VALUES Cultural Period of Site Frequencies Z Late Prehistoric 15_ 50.0 Late Prehistoric/Historic §_ 20.0 Late Middle Prehistoric 2 1.4 Late Middle Prehis./Late Prehistoric 2 1.4 Late Middle Prehis./Late Preshis./Historic 1 .7 Early Middle Prehistoric/Historic l .7 Early Middle Prehistoric/Late Middle Prehis. 1 .7 Early Middle Prehis./Late Mid. Pre./Late Pre. 2 1.4 SUB—TOTAL (valid cases) 30 100.0 MISSING CASES 116 TOTAL 146 100.0 TABLE 5 FREQUENCY DISTRIBUTION OF TYPEFALL VALUES Type of Fall At Site Frequencies Z bluff/talus slope l bluff/steep embankment/river cut bluff and steep embankment bluff and cutbank bluff or cliff talus slope/steep embankment talus slope steep embankment/river cut steep embankment river or creek bank other SUB-TOTAL (valid cases) 129 MISSING CASES l7 P‘H‘ \J #1 UJNJO\H‘P‘P‘N>UDNDQJU1 U'lNl-‘NH oowowox U1 H H O O N\ON O wwboooooo TOTAL 146 100.0 62 TABLE 6 FREQUENCY DISTRIBUTION OF JUMPFACE VALUES: 1 Direction Jumpoff Faces Frequencies Z Southwest 5 5.6 Southeast 3 3.3 Northwest 3 3.3 Northeast 11_ 12.2 West 5 5 6 South 1§_ 17.8 South-southeast 1 l 1 East 12_ 13.3 North 39_ 33.3 North-northwest 3 3.3 North-northeast 1 1.1 SUB-TOTAL (valid cases) 90 100.0 MISSING CASES 56 TOTAL 146 100.0 TABLE 7 FREQUENCY DISTRIBUTION OF JUMPHEIT"VALUES Height of Jumpofffl__u .1. -l-,rlr Frequencies Z over 50 meters 4 5.6 40 to 50 meters 2 2.8 30 to 40 meters 3 4.2 20 to 30 meters 9 12.5 10 to 20 meters 31_ 43.1 under 10 meters 22_ 31.9 SUB-TOTAL (valid cases) 72 100.0 74 MISSING CASES TOTAL 146 100.0 63 which measured the presence or absence of drive lanes, are potentially important for understanding the environmental and cultural context of jump site locations. Unfortunately, the site sample at hand does not provide a reliable measurement of the distribution of these features. While the present study established the existence of 34 campsites adjacent to jump sites and 46 sets of drive lanes, many more of these features may have gone unrecorded at known jump sites for a number of reasons. For instance, there may have been historic disturbance which obliterated the features or there may not have been thorough surveys made of the sites. It is also possible that the features never existed at many sites. The failure to accurately establish the presence or absence of these features at many sites, therefore, precludes formal use of existing information in the generation of a predictive site loca- tional model. Of the remaining Group One variables, AGE, JUMPFACE, and JUMP- HEIT appear to be the most reliable for purposes of this study. The JUMPFACE variable has the greatest potential for contributing to an understanding of site location. Cases frequently associate with several different values on this variable revealing a pattern which is not entirely compatible with existing literature on this aspect of jump sites. Limited observations have led to the conclusion that the vast majority of bison jump sites face either north or east; seldom west and almost never south (Malouf and Conner 1962: 44-45; Lewis n.d.). The distribution of facings derived from this study (Table 7), indicate that most jumps face north (33.3Z), and a lesser number face east or northeast (27.6Z). Contrary to previous statements, however, a large number of sites also face south (17.8%). It is impossible to know 64 to what extent this distribution is reflective of all jump sites in the Northwestern Plains since the cases evaluated in this study were non- randomly collected. Nevertheless, this distribution suggests that more variability may exist than formerly thought. In order to more clearly identify the diversity in the JUMPFACE variable, two additional frequency distributions were generated in which selected values of the PRSRTOPO and ELEV variables were controlled for. In the first analysis, site altitudes (ELEV) were divided into nine SOO-foot interval groups and frequency distributions of the JUMPFACE variable then generated for each group. The results were inconclusive due to the paucity of cases generated for each distribution. In con- trast, when controlling for the surficial geologic features ground moraine and unglaciated terrain (PRSRTOPO variable), enough cases were generated to reveal significant variability in the JUMPFACE distribution (Table 8). When controlling for unglaciated terrain, however, the results were mixed with a significant percentage of sites facing either south, north or northeast. An inspection of frequency distributions of the 16 Group 2 environmental variables revealed that, similar to Group 1 variables, some have greater potential for identifying site locations than others. For instance, for several reasons the variables WATERTWO, CORSESOL and STONISOL provided little information relevant to site placement. The WATERTWO variable, which identified the secondary water source of each case, was not accompanied by a distance measurement. In the absence of known distances to case locations the variable measures only the occurrence of secondary water sources within the study area, not their relation to the sample population. Additionally, the CORSESOL and 65 TABLE 8 FREQUENCY DISTRIBUTION OF JUMPFACE VALUES: 2 Direction Jumpoff Faces Frequencies* Z Frequencies** Z North 14 58.3 4 17.4 East 3 12.5 3 13.0 South 2 8.3 6 26.1 West 1 4.2 Northeast 3 12.5 5 2 .7 Northwest 1 .3 Southeast 2 .7 South-southeast 1 .3 Southwest 1 4.2 1 .3 SUB-TOTALS (valid cases) 24 100.0 23 10 .0 MISSING CASES 10 17 TOTALS 34 100.0 40 100.0 *of sites in ground moraine **of sites in unglaciated terrain TABLE 9 FREQUENCY DISTRIBUTION OF WATERONE VALUES Nearest Water Source Frequencies lake 5 4.1 river 14 11.6 river/spring/ lake 1 .8 wash 41_ 33.9 wash/ spring 2 1.7 wash/river 6 5.0 perennial stream 34_ 28.1 perennial stream/ lake 1 8 perennial stream/ river 2 1.7 perennial stream/ wash 14 11.6 perennial stream/ wash/ lake 1 .8 SUB-TOTAL (valid cases) 121 100.0 MISSING CASES 25 TOTAL 146 100.0 66 STONISOL variables, which measure the relative coarseness and stoniness of predominant soils in each site's standard area, provided minimal information due to problems which developed during collection of the data. After inspecting a number of site soils for these features, it became clear that different proportions of coarseness and stoniness frequently occurred in the same soil type within a given standard area. Several of the COARSESOL and STONISOL values were combined within their respective variables in an attempt to rectify this problem. Unfortu- nately, this procedure resulted in a diSplay of the variation, rather than the predominance, of soil coarseness and stoniness. Consequently, the two variables were eliminated from additional analysis. The SITETOPO variable, although it revealed considerable information relevant to site location, did not provide meaningful and comparative statistical results largely because it was not highly discriminating, i.e., it evaluated all surficial geologic features present at a site location rather than the predominant ones. Thus, SITETOPO was not readily comparable with other variables which identify predominant environmental features. The PRSRTOPO variable provided an appropriate evaluation of this aspect of the environment by identifying only predominant surficial geologic features at a site location. The remainder of the Group 2 variables consist of a series of concise assessments of vegetation, soils, landform, geology and water features and their association with case locations. Cases tend to associate with few values within each of these variables and in some, they largely associate with only one value. For example, cases measured on the SITESOL and AREASOL soil variables occur exceptionally often with loam and cases evaluated on the SITEROCK and AREAROCK variables 67 overwhelmingly associate with shale/sandstone. The strength of these values is undoubtedly a reflection of the Northwestern Plains environ- ment, i.e., loams are the predominant soil type and shale, sandstone, and related rocks are the most commonly occurring bedrock formation in this part of the Plains. The WATERDIS variable, which measures the distance from a site to the nearest water source, and the DISTPERM variable, which measures the distance to the nearest permanent water source, also strongly asso- ciate with only one value (Tables 10 and 12). In both instances, cases most frequently associate with a water source (WATERDIS) within 400 meters of a site or a permanent water source (DISTPERM) within 800 meters of a site. The balance of the Group 2 environmental variables tend to dis- play more variability than those discussed above, i.e., two or more values share a significant percentage of the total valid case population. Frequency distributions of these variables are displayed in Table 9 and Tables 11 through 16. Frequency Cross-tabulation Two-by-two contingency tables were generated by the SPSS sub- program CROSSTABS in order to determine if particular sets of variables co-occur with case locations. A large number of cross-tabulations were generated, but only a few revealed significant associations. All variable combinations which were analyzed will be discussed, but only those tables which suggest the existence of important associations will be illustrated. Prominent frequencies are underlined in the displayed tables. Chi-square significance tests were made for each of the 68 TABLE 10 FREQUENCY DISTRIBUTION OF WATERDIS VALUES Distance to Nearest Water Frequencies Z within 400 meters 101 84.2 within 800 meters 14 11.7 within 1600 meters 4 3.3 within 2400 meters 1 .8 SUB—TOTAL (valid cases) 120 100.0 MISSING CASES 26 TOTAL 146 100.0 TABLE 11 FREQUENCY DISTRIBUTION OF PERMWATR VALUES Nearest Permanent Water Frequencies Z lake 7 5.8 spring 5 4.2 river 34 28.3 perennial stream §§_ 56.7 perennial stream/lake 2 1 7 perennial stream/spring 2 1.7 perennial stream/river 2 1.7 SUB-TOTAL (valid cases) 120 100.0 MISSING CASES 26 TOTAL 146 100.0 Distance to Permanent Water FREQUENCY DISTRIBUTION OF DISTPERM VALUES 69 TABLE 12 Frequencies 3Q within 400 meters within 800 meters within 1600 within 2400 within 3200 within 5000 beyond 5000 meters meters meters meters meters NV DUO H P‘P‘UIH‘GD U1 \0 J.\ #000 SUB-TOTAL (valid cases) MISSING CASES 123 O moor-4001 100. TOTAL FREQUENCY DISTRIBUTION OF VEG VALUES Vegetation Type TABLE 13 146 Frequencies 100.0 3% K64/Southern Prairie Cordilleran Parkland Forest Northern Prairie Northern Prairie/Parkland Northern Prairie/Southern Prairie K98 K66 K64/K66 K63 K63/K64 K55 K16 Kl6/K64 K15 K12 K12/K55 K12/K15 TOTAL (valid cases) b 00 n: u: H‘O‘P‘H‘b‘h‘u1k) 1.1 Wi—‘ONNUJ-L‘N U) N \D I. N P‘uJP‘ n>a~a~ 1. b N o o o o o \I 1" \IN N Chl—‘i-‘NNH p.» b 0‘ O HNObe—‘Nb 100. TABLE 14 7O FREQUENCY DISTRIBUTION OF LANDFORM VALUES Landform Class Frequencies Z Mountains 2 1.4 Mountain Foothills 4 2.7 Alberta Plain 49_ 27.4 Alberta Plain/Mountain Foothills 5 3 4 D6 12 8.2 C6a 12 13.0 C6a/D6 l .7 C4b 7 5.0 B5d 9 6.2 BSb l .7 B4c 12 8.2 B4c/D6 l .7 84b 4 2.7 B3c 20 13.7 B3c/C6a 2 1 4 B3c/B4b 1 .7 B3b 5 3.4 B3b/D5 1 .7 TOTAL (valid cases) 146 100.0 TABLE 15 FREQUENCY DISTRIBUTION OF AREATOPO VALUES Surficial Geologic Features Frequencies Z unglaciated terrain 43. 33.3 dissected mountains ll 8 5 alluvium 1 .8 remnant stream terrace 5 3.9 glacial lake bed 14 10.9 glacial channels/ alluvium 2 1.6 ground moraine 59_ 38.8 moraine 1 8 moraine/ground moraine 2 1.6 SUB-TOTAL (valid cases) 129 100.0 MISSING CASES 17 TOTAL 146 100.0 71 TABLE 16 FREQUENCY DISTRIBUTION OF PRSRTOPO VALUES N Surficial Geologic Features Frequencies 1.x 0 C.) O 00 |. unglaciated terrain dissected mountains lO 7 7 bedrock 9 6.9 gravel l .8 alluvium 9 6.9 remnant stream terrace 8 6.2 glacial lake bed 10 7.7 glacial channels/alluvium 7 5.4 ground moraine 34_ 26.2 moraine 1 8 ground moraine/moraine 1 .8 SUB-TOTAL (valid cases) 130 100.0 MISSING CASES l6 TOTAL 146 100.0 72 contingency tables to determine the statistical significance of each cross-tabulation. Those measuring to at least the .05 level were deemed acceptable. Initially, the AGE and JUMPFACE variables were cross-tabulated with several environmental variables: AGE with LANDFORM, JUMPFACE, JUMPHEIT and PRSRTOPO and JUMPFACE with LANDFORM and JUMPHEIT. Signi- ficant associations were not demonstrated between any of these sets of variables. The negative associations which occurred with the ACE variable probably resulted from both the limited number of dated sites in the study population (30) and the lack of variability present in the sample. Half of the dated sites fall within the Late Prehistoric Period and more than 86 percent contain at least one Late Prehistoric Period component. Probable causes for the lack of association with the JUMPFACE variable are not clear. The cross-tabulation of the LANDFORM variable with a series of other environmental variables represented one of the most important segments of this analysis. The landform classes proposed by Hammond (1964) and Bostock (1970) divide the study area into a number of geo- graphically bounded areas with distinct topographic characteristics and altitudes (Figs. 2 and 3). Case locations within each of these regions share broad environmental characteristics. As a result, cross-tabulation of landform units with other environmental variables has considerable potential for identifying important differences between these cohesive groups of jump sites. Only sites located within Montana and wyoming were used in analysis of the LANDFORM variable because of the incompatibility of Canadian and United States landform units. Classes proposed by Hammond (1964) for the United States provide 73 more quantifiable and comparable physical descriptions of the separate units than those recently established for Canada (Bostock 1970). Thus, comparison between the two schemes was not possible without severely compromising the information content of the Hammond classes. Using the 95 case sample from Montana and Wyoming, LANDFORM was cross-tabulated with VEG, SITESOL, AREASOL, PRSRTOPO, AREATOPO, SITEROCK, AREAROCK, WATERONE and PERMWATR to determine if sets of these variables co-occur with case locations. Sites associated with the five LANDFORM values B5d, B4c, B3c, 04b and C6a occurred frequently and consistently in association with a limited number of values on each of the other nine values. The LANDFORM values fall into two major groups: (1) 85d, B4c, and BBC, tablelands with moderate to high relief and (2) C4b, open high hills, and D6 and C6a, high mountains and open high mountains with extreme local relief (Hammond 1964). In all contingency tables except one (AREASOL) chi-square tests confirmed significance to the .05 level. The cross-tabulation of landform classes with vegetation types revealed associations between tableland sites (B3c, B4c and 35d) and the short grass Plains vegetation (K64) and between open high mountain sites (C6a) and prairie grasses (K63) (Table 17). Contingency table analysis of the LANDFORM variable with SITESOL and with AREASOL revealed that loam soils most frequently occur in association with open high mountain and tableland sites (Tables 18 and 19). Specifically, loam soils occur with sites in the C6a, B4d and B3c landform classes on the AREASOL variable. The chi-square signifi- cance test of the AREASOL and LANDFORM contingency table, however, revealed that an unacceptably high level exists for this combination of variables. While this is true for the entire contingency table, the 74 TABLE 1 7 FREQUENCY CROSS-TABULATION: LANDFORM CLASS BY VEGETATION TYPE 2 9 1 7 9 1 3 1 I4 0 2 l S 1 6 A -9 ---- -i- --.--.-,- 31111-,- -31 m a b O 6 6 6 4 5 F D D C B D L D / / / / / A m a a b d b c c b c c c b b T 6 6 6 I... 5 5 l4 l4 l4 3 3 3 3 2 O L D C C C B B B B B B B B B B T 75 TABLE 18 FREQUENCY CROSS-TABLULATION: LANDFORM CLASS* BY PREDOMINANT SOIL TYPE* AmHo >mHo HHom cmvoum ESH>3HHm 1T- 10 40.39 H m LANDFORM D6 C6a C4b B5d 35b B4c B4c/D6 34b B3b 71 11 t 44 10 Tvll..1| TOTAL *at site 76 TABLE 19 FREQUENCY CROSS-TABULATION: LANDFORM CLASS* BY PREDOMINANT SOIL TYPE* Avcmm EmoH >mHo \EmoH uH«m\EmoH EmoH meo \EmoH Emoa EmoH >mHo \EmoH HHHm EmoH uHNm EmoH >mHo Esw>=HHm ILA. Hom=HHm HImacmnoamHome 14000000020000 7 H m «on SomeHmHomHm010000200300206 T W womuuwu MEmouumucmchu141000200000008 m #093003 1 0 0 0 O 0 0 0 0 0 0 0 0 0 1 M mchucsos S «muoommeEZOOOOOOOOOOOOm S A cfimuumu a cwumwomHucsl8_071017_0452131m m ouch—mm.» 0 m m m a b 0 6 6 6 A. 5 L F D D C B D m D / / / / / M a a b d b C C b C C C b b T 6 ,6 .6 4 5 s. A. 4 4 3 1. 3 1. 2 O L D C C C B B B B B B B B B B T *at site 79 TABLE 21 FREQUENCY CROSS-TABULATION: LANDFORM CLASS BY PREDOMINANT SURFICIAL GEOLOGIC FEATURE* \ E “U \ 0) m m m _- C: 1: OJ .0 0) +4 H 'c H I: m m m u m c H H u “C U) U) x m 0 O O to a) c: m .c: E E E m -H : u-H u m .4 o a m m 0 0-H u m a o -H 'U c 'U : Ea m m m u m m . - > a +4 :-H .4 -< .4 u m c c H o o a a m a m < n: fl‘H m a E u m m-H o H O H F a: c m -H o m m -H .H-H u c H o co LANDFORM": :Ju 'UE Hu 6!: 00m 6!: E one [—- D6 1 9 2 0 O O O 0 12 C6a 11_ 2 1 4 l 0 O O 19 C6a/D6 O O 1 O 0 O O O l C4b ' '7 O O O O O O O 7 B5d o o o o o _9_ o o 9 B5b 1 O O O O O O 0 1 B4c 9_ O O 2 O O O 1 12 B4c/D6 1 o o o 0 o o 1 0 1 B4b 4 O O O O 0 O O 4 BBC ' 3 o 1 3 1 11 o o 19 B3c/C6a 2 O O O O 0 O 0 2 B3c/B4b l O O O O O O O 1 33b 3 O O 2 O O O O 5 BZb/DS , 1 O O 0 O O O O 1 __- 1 1 TOTAL j43 11 5 11 2 20 l 1 84 *in standard area around site AHu\£mms mcHuaw\swms 12 17 12 16 78 12 18 80 TABLE 22 FREQUENCY CROSS-TABULATION: LANDFORM CLASS BY NEAREST WATER SOURCE sums ome \mcHuam\Ho>Hu Hm>Hu ome 7; 0 7; O mzcmme<3_ LANDFORM D6 C6a C6a/D6 C4b B5d B5b B4c B4c/D6 B4b B3c B3b szlns TOTAL 1.11.11 AHu\;mmB acfiuem\cmm3 12 17 12 16 78 12 18 80 TABLE 22 FREQUENCY CROSS-TABULATION: LANDFORM CLASS BY NEAREST WATER SOURCE sums ome \NCNHam\uw>Nh Hm>HH iii-111111111111 :1 i mzcmmh<3. j.|l.|0ll...l.l ll! D6 7 O 7; O O C6a C6a/D6 C4b B5d B5b B4c B4c/D6 B4b B3c B3b B2b/D5 33 TOTAL 1.11 '- 80 TABLE 22 FREQUENCY CROSS-TABULATION: LANDFORM CLASS BY NEAREST WATER SOURCE Afiu\£mw3 mcfiuam\:mm3 smma ome \wcHHam\uw>fiu um>Hp ome 7 O 7; O 12 18 33 mzcmm9<3. LANDFORM D6 C6a C6a/D6 C4b B5d B5b B4c B4c/D6 B4b 83c B3b szlns TOTAL 81 TABLE 23 FREQUENCY CROSS-TABULATION: LANDFORM CLASS BY NEAREST PERMANENT WATER SOURCE AHH\Emwuum Hmficcmuma maHuaw\Emwuum Hchcouoa ome\Emouum Hmficcmuoa Emmuum HmHssmuoa um>HH ucHHam mme 111-111-1111 ll-‘l MH<32mmm LANDFORM x D6 C6a .U: (lull-11-11-. 1.1 7; R: 7; O O 44 23 C4b B5d B5b B4c B4c/D6 B4b B3c B3b B2b/D5 TOTAL 82 eliminated as a category (in PERMWATR variable). Tableland sites with considerable or high relief (B4c and B5d) most frequently occur near rivers in the PERMWATR variable, but when washes are included as a cate- gory (in the WATERONE variable) rivers drop from importance and are replaced by washes. Tableland sites with moderate to considerable relief (B4c and BBC) and sites in open high mountains (C6a) also most frequently occur with washes. Additional cross-tabulations were made between landform classes and the SITETOPO, AREATOPO and AREASOL variables selecting for environ- mentally distinct regions of the study area which might serve to differ- entiate sets of cases. One set of tables were generated for cases within unglaciated areas, a second for those in glaciated regions and yet another for all cases lying above 4500 feet in altitude. In all analyses, the cells within the tables produced too few cases that were too widely dispersed to identify significant trends. The remainder of cross tabulations encompassed the entire sample population and focused upon case associations with soils and vegetation. In one analysis the site soil type (SITESOL) was compared with predo- minant site surficial features (PRSRTOPO) (Table 24). Sites in ungla- ciated terrain most frequently occur with loam soils while sites in ground moraine areas (glaciated) were found to associate with loams and, to a lesser extent, with eroded soils. A cross-tabulation between AREASOL and AREATOPO revealed case associations similar to that between SITESOL and PRSRTOPO, except that eroded soils did not even occur as a AREASOL category. The comparison between AREASOL and AREATOPO variables resulted in an unacceptably high chi-square level of probability. This was probably due to the large number of empty contingency table cells 83 TABLE 24 FREQUENCY CROSS-TABULATION: PREDOMINANT SOIL TYPE* BY PREDOMINANT SURFICIAL GEOLOGIC FEATURE* '0 \ OJ r—4 \ E .o m m m m c a c m m s -H -H 'o u .x m m m o u m .c H H u 'U 0‘) m v-4 0 O O c~ m m a E E E E o. -H : u-H .x s u o .H .4 s o o c- 0-H o m o -H -H a o m moa 'o a 'o : 6a m m m u o o > m m -H -H > c -H c-H .4 m' —-u m c H > a E H u u a a m a m -< u: u>u m s 'c m -H u 6 6-4 0 u o u E4 SITESOL E 586°” :1- 7‘. 882:. "3.7.2.1; 21° 8 alluvium 5 0 0 0 O 0 0 1 2 1 O 9 eroded soil 1 O O 4 0 3 O 0 O 9_ 0 0 l6 clay 1 O 0 O l 0 2 0 O 0 O 4 clay loam . 3 0 0 0 1 1 O O 3 O 0 8 i clay loam/ ' 1 0 0 0 O 0 0 O 0 O 0 1 clay silt loam O O O 0 0 O O 1 0 0 0 1 loam 21_ 4 2 l 2 2 5 1 16_ O 1 55 loam/ 2 O O 0 O 0 O 0 0 O O 2 silt loam sand loam l l O O 0 0 1 O 0 O 0 3 sand 0 O O O 1 O O 0 O 0 O 1 TOTAL 34 5 6 1 8 3 8 3 3o 1 1 100 *at site 84 and the table's large size (8 by 9 cells), both of which tended to mask the paired value associations. A cross-tabulation of the VEG AND AREASOL variables revealed that loam soil is associated with sites located in several different vegetation zones (Table 25). The most frequent occurrence is with sites in the K64/Southern Prairie vegetation type. Loam occurs with less intensity with sites in Northern Prairie and K63 vegetation types. In contrast to this pattern, analysis of the VEG and SITETOPO variables suggests that site soil type tends to vary in association with vegetation types (Table 26). Loam frequently occurs with sites in the K64/Southern Prairie, Northern Prairie and K63 vegetation types while eroded and alluvial soils often occur with sites in the K64/Southern Prairie type. Eroded soils are also often associated with sites in Northern Prairie vegetation. The greater variability exhibited in soil and vegetation types at site locales (SITESOL) than in their surrounding area (AREASOL) may, in part, be due to the different localized tepographic settings. Sites are often located on high, steep cutbanks above riversand washes which frequently produce eroded soils and, depending upon conditions, alluvial soils. These conditions are usually localized, however, and so tend to occur most often at site locations rather than in the surrounding area. Surficial geologic features of sites and surrounding vicinity (PRSRTOPO and AREATOPO) also appear to vary in association with vege- tation type (Tables 27 and 28). Sites in unglaciated terrain frequently occur with the K63 vegetation type while those in both unglaciated terrain and ground moraine areas are often associated with the K64/ Southern Prairie type. Sites located in areas where ground moraine, 85 TABLE 25 FREQUENCY CROSS-TABULATION: VEGETATION TYPE BY PREDOMINANT SOIL TYPE* \ E m 0 H H . H \ E . o E E SE E 44E :1: . E m m m w m m H m o H. z o o o o o H o H O, H '0 H H HH H (pH to, > w \ \ >~. d <. D U >. >. u u >» E E >~ E >- c n.1, H o to m H H“! :6 mm was I: H oz, H L- H H -H HH 0 OH 0H m 0 VEG <12, m m U U m mu H Ho HU ID [-4 K64/Southern l 0 0 3 0 32_ 0 0 1 42 Prairie Cordilleran O 0 0 0 O 0 2 0 0 0 2 Forest Parkland O 1 O 2 0 O 2 0 0 0 5 N. Prairie 0 0 4 0 0 2 l3 0 0 0 19 N. Pr./Pk1d. 0 O O 2 0 O 1 0 0 0 3 N. Pr./S. Pr. 0 0 0 O 0 O l O 0 O 1 K98 0 O 0 0 0 0 0 0 0 1 1 K66 0 0 0 0 0 l 1 0 O 0 2 K64/K66 O O O O 0 O l 0 0 0 1 K63 o o o 4 1 0 p o 1 2 16 K63/K64 O O 0 0 0 0 2 O 0 0 2 K55 0 0 0 0 0 0 1 1 O O 2 K16 O 0 .0 1 O O 2 0 O 0 3 K12 1 0 0 1 0 0 1 1 0 0 4 TOTAL 2 l 4 l3 1 3 72 2 1 4 103 *in standard area around site 86 TABLE 26 FREQUENCY CROSS-TABULATION: VEGETATION TYPE BY PREDOMINANT SOIL TYPE* 1 :3 \ E ' s 3. 5 5 53- 5 '3 .4 a o o o 0 v4 0 H “U H H H H :0 > m \« >. .4 m s 1: >- >» >->. u 5 E14 '6 w: < e- H- o m m m 6 -4 m m—a : c E4 H H L: H H H H H O OH to m 0 VEG co m w o o o u m —- Ham m m _j: K64/Southern 7 6 2 4 O O _23 0 1 0 43 Prairie i : Cordilleran { 0 0 0 0 0 O 2 O 0 O 2 Forest ’ Parkland O 2 O 1 0 O 2 0 0 0 5 No. Prairie 0 2_ 4 2 0 0 19_ 0 0 l 26 N. Pr./Pkld. 0 0 0 l O 1 l O 0 0 3 1 N. Pr./S. Pr. i O O 0 O 0 O 1 0 0 0 1 K98 i 0 0 O 0 O 0 0 O 1 O 1 1 K66 1 1 O O 0 O O 2 0 0 O 3 1 K64/K66 % 0 O 0 0 O O 1 0 0 O 1 1 - K63 * 1 o 0 1 o 1 _1__o_ 2 o o 15 l 1 K63/K64 I 0 O 0 O 0 2 0 0 0 0 2 I K55 I O 0 0 1 1 1 O O 0 O 3 1 K16 E O 0 O O O O 3 0 0 0 3 I K16/K64 1 O 0 0 0 0 O 0 0 1 O 1 I K12 I 1 O 0 O 0 0 5 0 1 0 7 '1 T TOTAL {10 17 6 10 l 5 60 2 4 1 116 *at site 87 TABLE 27 FREQUENCY CROSS-TABULATION: VEGETATION TYPE BY PREDOMINANT SURFICIAL GEOLOGIC FEATURE* H=HHm \Hocnmco HmHowHa «mg mme HmHume NUNHHNU 5mm..." u m USN-Hemp EaH>=HHm Hw>muw xoouuon mchucaoE wouommmHv :Hmuumu woumHomecn «I. O 34 10 10 4O CACHmmmm K64/Southern Prairie Parkland Prairie No. N. Pr./Park1d. N. Pr./S. Pr. K98 K66 K64/K66 K63 K63/K64 K55 K16 K16/K64 K15 K12 K12/K55 K12/K15 TOTAL *at site FREQUENCY CROSS-TABULATION: TABLE 27 87 VEGETATION TYPE BY PREDOMINANT SURFICIAL GEOLOGIC FEATURE* 'U \ (D H \ E .o o m m m a c a Q) m c: H H 'o u .x m m m m u m .c u H u '0 U) a) H U o O o m m c E E E E FL H C U -r-l ,2 :3 u m H H .‘3 Q) (I) c c-H- o m u .4 -H a u m m-H x: c: 'o c B m m m u o m > m m -H -H > a -H ;:,4 :2 m H-:- m a u > a g u o o a s m s m . u: 601- m a 'o m -H u m 6.4 o p o u En m c w H o m H r4 m w -H v4.4 H o H O VEG IL : L! 'o E .o a: m H u on cmcu a: E «>2 64 K64/Southern 13_ 0 0 0 0 5 3 2 21_ 1 0 46 Prairie 1 Parkland 0 0 2 0 O 0 0 O 0 O 0 2 No. Prairie 0 O 6 1 2_ 0 2 0 2_ O 0 27 N. Pr./Parkld. I O 0 0 0 0 0 0 0 2 0 0 2 1 u. pr,/s, pr, 0 O O 0 0 O O 0 1 O O 1 K98 § 0 O O O O 0 1 0 0 0 0 1 K66 4 0 0 0 0 O O 0 1 0 0 5 K64/K66 0 0 0 O 0 0 1 O 0 0 O 1 K63 ll_ 2 0 O 0 2 1 3 0 0 1 20 K63/K64 1 0 O O 0 O 1 O 0 O O 2 K55 2 O 0 O 0 O 1 1 0 0 0 4 K16 3 0 0 0 0 0 O O 0 0 0 3 K16/K64 2 O 0 0 0 O 0 O 0 O O 2 K15 0 2 0 O 0 0 0 0 0 O 0 2 K12 4 4 0 O 0 l 0 l 0 0 O 10 K12/K55 1 0 0 0 O O O O 0 0 O 1 K12/K15 0 2 l O 0 O 0 O O O O 3 TOTAL 40 10 9 1 9 8 10 7 34 1 1 121 *at site 88 TABLE 28 FREQUENCY CROSS-TABULATION: VEGETATION TYPE BY PREDOMINANT SURFICIAL GEOLOGIC FEATURE* '6 ~« (00) H \ m .n a» o m o a c c m m c 44 -H 'o H .x m to m m H m .c :4 u c- ‘6’ 3‘25 3" ”e?- 8 D. "'4‘: U-v-O D U v-1 0-1:! Q) Q) c: o-H o w -H a E m m-H w: 'o c a H m m w u > m m -H -H > a ::v4 *4 .4 -< -H H m a s E m o o a :3 a m m < m asp m 3 H4 u a 6'4 0 o H H H a: co HOH muH HH L- H o c VEG < DU '65“! Hm 0(- mm OD 602 E E-‘ K64/Southern 14_ 0 0 l l ‘23 0 1 44 Prairie Parkland 0 0 0 0 0 O 2 0 0 2 N. Prairie 0 O 0 0 3 O 21_ 1 O 25 N. Pr./Pk1d. 0 O 0 0 O 0 2 0 O 2 N. Pr./S.Pr. 0 O 1 O O 0 O 0 O 1 K98 0 O O 0 1 O 0 0 O 1 K66 4 0 0 O O 0 2 0 O 6 K64/K66 O 0 0 O l 1 O O O 2 K63 12_ 2 O 2 2 O 0 1 0 l9 K63/K64 1 0 0 0 l 0 0 0 O 2 K55 2 0 O O 2 O 0 O 0 4 K16 3 O 0 O 0 0 0 0 O 3 K16/K64 2 0 0 O O O 0 0 O 2 K15 O 2 0 0 0 0 0 O O 2 K12 4 4 0 2 O 0 O O O 10 K12/K55 1 0 O O O 0 O 0 O 1 K12/K15 O 3 0 O O 0 0 O O 3 TOTAL 43 11 l 5 14 2 50 2 l 129 *in standard area around site 89 bedrock or alluvium is predominant are often associated with Northern Prairie vegetation. This association only extends to the area sur- rounding the site (AREATOPO), however, for cases in ground moraine areas. Bedrock and alluvium features are usually too limited in extent to predominate within the broad contiguous standard area surrounding a site. Interval Scale Variables The eight interval scale environmental variables investigated in this study which reflect aspects of soil, water, topography and alti- tude associated with case locations, were all subjected to multivariate analysis. In part, this method of analysis was chosen to supplement information on associations previously discovered during the frequency and cross-tabulation tests of Group 1 and 2 variables and site locations. More importantly, however, it will provide a more encompassing view of the patterns of relationships exhibited by variables than was possible in the more cumbersome nominal scale analyses. The analytic results are enhanced by the much greater information potential possessed by interval scale variables. In addition to the multivariate analysis, a frequency distribution test was performed on the altitude variable ELEV. The multivariate analysis procedure, SPSS subprogram FACTOR, was performed on all of the interval scale variables in order to gain maximum associational information. Prior to this analysis, however, a series of univariate statistics were produced of each variable by SPSS sub-program CONDESCRIPTIVE in order to establish the degree to which the data approximate a normal distribution. Near normal data distribu- tions are necessary to insure the reliability of the multivariate 9O analyses. An examination of measures of kirtosis and skewness of each variable revealed that none even approached three standard deviation units, thus indicating a near normal distribution. Inspection of correlation coefficients of the eight variables produced by SPSS sub-program PEARSON CORR revealed one set of multi- collinear variables and several that were moderately intercorrelated. All of these coefficients were significant to the .001 level. Only three, however, equaled or exceeded the Pearson's r level of .74. The R—Mode factor analysis generated included all cases and all eight variables. The SPSS PA/Z method of factoring (Principal Factoring with iteration) was used in which communality estimates are automatically given and iteration employed to improve the estimates. Three factors were revealed accounting for 100 percent of the total variance present in the three eigenvalues (Table 30). Varimax rotation of the factor matrix was made following extraction of appropriate eigenvalues resulting in significant factor loadings on several variables. The rotated factor matrix is displayed in Table 29. Significant loadings are underlined. Factor 1 is characterized by three variables: positively by AROCTOPO and STOCTOPO and negatively by NUMTOPO. These variables exhibit two strong relationships: (1) an association between the pro- portion of surficial geologic features at a site location and within a standard area; and (2) an inverse relationship between the percentage of area occupied by a predominant surficial feature at or around a site and the number of types of features present. These associations suggest that when a single predominant surficial feature occupies a large proportion of a site or area surrounding a site, multiple surficial 91 TABLE 29 VARIMAX ROTATED MATRIX OF FACTOR LOADINGS: ALL VARIABLES Variable l 2 3 NUMWASH -.07 -.08 .g_9_7_ NUMSTRM -.02 .51: -.23 NUMSOL -.19 .29 .05 PERCTSOL .22 ‘ - 2_3§_ .14 ELEV .13 ._3_§_ .20 NUMTOPO - _8_2_ .07 .01 AROCTOPO ._9_6_ - . 16 - . 08 STOCTOPO .88 -.06 .05 TABLE 30 SPSS PA/2 ANALYSIS (RrMODE): FACTORS, EIGENVALUES AND PERCENT VARIANCE Factor Eigenvalue Percent Variance l 2.66 53.5 2 1.35 27.3 3 .95 19.2 100 = Cumulative Percentage 92 features will not tend to occur. Conversely, when the percentage of the predominant feature is low, there will tend to be a diversity of surficial features. Factor 2 is characterized by four variables: positively by NUMSTRM, NUMSOL and ELEV and negatively by PERCTSOL. Two relationships characterize this factor: (1) a positive association between altitude, stream density and the diversity of soil types within a site's standard area; and (2) an inverse relationship between, on the one hand, alti— tude, perennial stream density and diversity of soil types, and, on the other hand, the percentage of area occupied by a predominate soil type within a site's standard area. These associations suggest that when sites are high in altitude they tend to have a high density of perennial streams and a diversity of soil types in the site's standard area. The Opposite would also be true. Also, when sites are high in altitude and there is a density of perennial streams and diversity of soil types in a site's standard area, the percentage of the predominant soil type will tend to be low. Conversely, when sites are low in altitude and there are low numbers of streams and few soil types, the predominant soil will occupy a relatively large percentage of the standard area. Factor 3 is characterized by only one variable: NUMWASH. It appears to have a tightly clustered, independent nature because it loads extremely high on the factor and does not display significant relationships with other variables. Results of the statistical test performed on the altitude variable ELEV were inconclusive. A frequency distribution was generated by SPSS sub-program FREQUENCIES revealing 121 valid cases having an elevational range of 2020 feet to 8400 feet above sea level. The 93 sample represents an almost continuous distribution so that potential site grouping are not discernible within particular elevational ranges. This suggests that elevation may not have been an important selective factor in site location. Nominal, Ordinal and Interval Scale Variables: Statistical Comparisons The univariate and multivariate analyses described in the previous section revealed several significant associations between environmental features and site locations. These analyses, however, did not provide the means to compare the nominal and ordinal scale variables with interval scale variables. To explore the information potential of such comparisons it was decided to implement SPSS sub- program BREAKDOWN. This program sub-divided the means, standard deviations, and variances of selected interval scale variables among value categories of one or two nominal or ordinal scale variables in order to establish the degree of correspondence between them. The measures of central tendency and dispersion of the NUMWASH, NUMSTRM, PERCTSOL, AROCTOPO and STOCTOPO variables were examined among values of eleven nominal and ordinal scale variables. The most significant values of the analyses discussed are displayed in Tables 30 through 38. A one-way analysis of variance was produced for each procedure to test whether the means of the population sub-samples were significantly different from one another. This test is intended for paired variables, so that only the first two were analyzed even when additional variables were included in the same analysis. All analyses were measured at the .05 level of significance. 94 The initial analysis was made on the NUMWASH variable, which measures the number of intermittent watercourses within a site's standard area. It was broken into sub-populations by the VEG, AREATOPO and PRSRTOPO variables. The comparison with the AREATOPO and PRSRTOPO variables are of particular interest (Tables 31 and 32). Both showed a strong association between unglaciated terrain sites and a relatively large number of washes and between ground moraine sites and a small number of washes. Several alternative explanations can be suggested for these associations. One is that site location was signi- ficantly influenced by different stream densities in the two areas. It is most likely, however, that because Pleistocene ice advances obli- terated or altered former stream drainage patterns in ground morainic areas, there has not been sufficient time for stream patterns to fully mature. The result has been a low density of intermittent streams. The analysis of the NUMWASH and VEG variables revealed a pattern of stream density associations similar to that discussed above. This pair of variables is deleted from discussion, however, because the standard deviations of the values were very large and the sample sizes quite disparate. The result was a highly skewed association pattern. The NUMSTRM variable was statistically broken down by PRSRTOPO and AREATOPO revealing very similar patterns of association (Tables 33 and 34). Unglaciated terrain sites tend to occur in areas with slightly higher stream densities than do ground moraine sites. The standard deviations for these values are so large, however, that the mean differences must be considered negligible. Mixed results were yielded when the PERCTSOL variable, measuring the percentage of area occupied by a predominant soil type in a site's 95 TABLE 31 BREAKDOWN ANALYSIS: NUMBER OF WASHES* BY PREDOMINANT SURFICIAL GEOLOGIC FEATURE* AREATOPO Value Mean Standard Dev. Cases For entire population 6.27 4.39 98 l. unglaciated terrain 8.79 3.53 33 7. ground moraine 3.87 3.47 37 *in standard area around site TABLE 32 BREAKDOWN ANALYSIS: NUMBER OF WASHES* BY PREDOMINANT SURFICIAL GEOLOGIC FEATURE** PRSRTOPO Value Mean Standard Dev. Cases For entire population 6.24 4.41 97 l. unglaciated terrain 8.48 2.90 29 9. glaciated terrain 4.24 3.32 25 *in standard area around site/ **at site 96 TABLE 33 BREAKDOWN ANALYSIS: NUMBER OF STREAMS* BY PREDOMINANT SURFICIAL GEOLOGIC FEATURE* AREATOPO Value Mean Standard Dev. Cases For entire population 3.33 2.43 99 l. unglaciated terrain 3.46 1.89 33 7. ground moraine 2.71 2.63 38 *In standard area around site TABLE 34 BREAKDOWN ANALYSIS: NUMBER OF STREAMS* BY PREDOMINANT SURFICIAL GEOLOGIC FEATURE** PRSRTOPO Value Mean Standard Dev. Cases For entire population 3.37 2.42 98 1. unglaciated terrain 3.41 1.78 29 9. glaciated terrain 2.35 2.65 26 *in standard area around site/**at site 97 standard area, was statistically broken down by AREASOL and VEG and by SITESOL and PRSRTOPO. The first analysis, a three-way breakdown between PERCTSOL and AREASOL and VEG, suggests that there are only slight differences between the areal soil percentage means of sites located in K64/Southern Prairie vegetation and those located in the Northern Prairie type (Table 35). The 6 percent difference in means may be due to the non-random nature of the data base or possibly to the better adaptation of K64/Southern Prairie vegetation to loam soils. The analysis of the PERCTSOL variable with SITESOL and PRSRTOPO reveals stronger associations than those described above. Within the loam value, sites associated with ground moraine and unglaciated terrain display a 10 percent mean difference in the predominant soil type per- centage figure (Table 36). The difference may stem, in part, from the non-random data base, small sample size, overlap of the standard deviation figures, or from a combination of these factors. It is also possible, however, that it stems from the surficial terrain differences between the two areas. Stream dissected areas (unglaciated terrain) exhibit a greater diversity of microenvironments than do the topo- graphically more uniform ground moraine areas. As a result, it is expected that soils would also exhibit greater diversity. Thus, soil types would tend to occupy smaller contiguous parcels than those found in ground moraine areas. Alternatively, there may have been a loca- tional preference for establishing ground moraine sites in largely uniform soils and locating sites in unglaciated terrain in areas with some soil type diversity for as yet unknown reasons. In any case, the mean percentage difference between these values is not excessive. Comparisons of sites associated with eroded soils and those associated 98 TABLE 35 BREAKDOWN ANALYSIS: PERCENTAGE AREA OCCUPIED BY PREDOMINANT SOIL TYPE* BZ_ PREDOMINANT SOIL TYPE* BY VEGETATION TYPE AREASOL Value VEG Value Mean Stan. Dev. Cases For entire population . . . . . . 57.54 18.78 84 7. loam . . . . . . . . . . . . . . . 63.23 16.51 66 l. K64/Southern Prarie 66.71 16.05 35 5. Northern Prarie 60.46 18.17 13 *in standard area around site TABLE 36 BREAKDOWN ANALYSIS: PERCENTAGE AREA OCCUPIED BY PREDOMINANT SOIL TYPE* BY PREDOMINANT SITE SOIL TYPE §Y_PREDOMINANT SURFICIAL GEOL. FEATURE** SITESOL Value VEG Value Mean Stan. Dev. Cases For entire population. . . . . . 60.0 18.97 69 2. eroded soil . . . . . . . . . . 56.58 14.90 12 9. ground moraine 59.44 14.43 9 7. loam . . . . . . . . . . . . 67.66 16.97 38 1. unglaciated terrain 64.67 12.83 9 9. ground moraine 74.67 18.57 15 _. *in standard area around site/ ** at site 99 with loam soils reveals an even greater disparity in the predominant soil percentage. The mean standard deviations are considerable, but it is likely that much of the disparity in means stems from the contrasting t0pographic settings of the two soil types. Eroded soils tend to occur along river and stream courses while loams are most often found in broad, contiguous zones between watercourses and in undissected plains. AROCTOPO wa broken down by AREATOPO and SITESOL and resulted in a disparity in the mean AROCTOPO percentage of eroded and loam soils similar to that above (Table 37). The high standard deviations show considerable dispersal of the percentages which, to some degree, negate the mean differences. Nevertheless, the difference may also be attri- butable to the contrasting t0pographic settings in which the soils occur. The analysis of both AROCTOPO and STOCTOPO by the LANDFORM variables resulted in several associations which, in mean percentage difference, far exceed any of those discussed above. The associated values in the STOCTOPO analysis, however, have exceptionally high standard deviations which indicate that the value means are not necessarily the focus of percentage distribution. Rather, they repre- sent the mid-point in a series of widely dispersed STOCTOPO percentages. As a result, the value associations are minimally significant. The breakdown of the AROCTOPO variable by the LANDFORM and AREAROCK variables resulted in much more significant percentages (Table 38). The two AREAROCK sandstone/shale values moderately deviate from the mean, but their respective ranges only overlap by 4 percent. This distributional pattern strongly suggests that sites associated with sandstone/shale bedrock in the Alberta Plain tend to occur in more surficially diver- sified settings than do tableland sites of moderate relief (B3c) which PERCENTAGE AREA OCCUPIED BY DOMINANT SURFICIAL FEATURE* BY_ 100 TABLE 37 BREAKDOWN ANALYSIS: PREDOMINANT SOIL TYPE* §Z_PREDOMINANT SITE SOIL TYPE AREATOPO Value SITESOL Value Mean Stan. Dev. Cases For entire population . . 77.14 19.31 98 1. unglaciated terrain . 80.86 18.46 35 7. loam . 81.50 18.46 22 7. ground moraine . . 75.76 19.44 42 2. eroded soil . . 68.86 17.03 14 7. loam . 81.94 18.46 18 *in standard area around site TABLE 38 BREAKDOWN ANALYSIS: PERCENTAGE AREA OCCUPIED BY DOMINANT SURFICIAL FEATURE* BY_ LANDFORM TYPE* §Y_PREDOMINANT TYPE OF BEDROCK* LANDFORM Value AREAROCK Value Mean Stan. Dev. Cases For entire population . . . . . . . . 76.70 19.96 125 17. Alberta Plain . . . . . 63.86 12.14 28 6. shale/sandstone . . . . 65.40 11.69 _25 5. D6 . 83.58 21.59 12 6. C63 . . . . 71.47 22.34 19 11. B4c . 73.75 15.98 12 14. B3c . . 84.58 21.36 19 6. shale/sandstone . . . 89.06 15.93 16 *in standard area around site 101 occur with sandstone/shale bedrock. These sites tend to occur in more surficially uniform settings. While the cause of the difference is not clear, it cannot be attributed to dissimilar landforms since both regions have comparable topography. It is possible, however, that the difference stems from dissimilar map scales. The surficial geologic maps used for Alberta were to a scale of at least l:253,440 while those in Montana and wyoming were all at a scale of 1:500,000. Thus, the larger scale maps for areas in Alberta would tend to more finely parti- tion surficial features than would the Montana maps which are tiwce that size. The result might be a higher number of surficial geologic features and a correspondingly lower percentage of area occupied by the predominant feature within the standard area surrounding each jump site in Alberta, the exact pattern displayed by the BREAKDOWN analysis of AROCTOPO BY LANDFORM BY AREAROCK. Chapter Summary The results of the univariate and multivariate analyses provide evidence of both regional and sub-regional site locational patterning in the study area which is summarized in the following paragraphs. The analyses suggest that, regardless of sub-regional assignation, jump sites tend to date to the Late Prehistoric Period and occur with cliffs or steep slopes measuring up to 20 meters high. Sites also tend to be associated with loam soils and sandstone and shale bedrock, though bedrock formations are not always surficial. Most importantly, however, are indications that jump sites: I. tend to face either in a north, northeast, east or south direction; 102 2. usually lie within 400 meters of a water source, most often a wash or perennial stream; and 3. often lie within 400 meters of and most frequently within 800 meters of a permanent water source, almost invariably a perennial stream or river. The analyses also suggest that there exists sub-regional variation in site location. While differences focus upon several environmental features, the most prominent and important are the ground moraine and unglaciated terrain. In large part, surficial geologic assignation depends upon whether the site is to the north of the Missouri River (entirely glaciated) or south of it (largely unglaciated). Sites in glaciated terrain tend to be association with a more diversified vegetation pattern than those in ground moraine areas. Those in both types of terrain strongly associate with short grass vegetation, but only sites in unglaciated terrain occur frequently with the denser grasses of the mountain foothills. Ground moraine sites also tend to associate with higher densities of intermittent streams (washes) and a higher percentage of loam soils than do sites in unglaciated terrain. On the other hand, there is a negligible difference between the two in perennial stream density. Most importantly in this site variation, however, are suggestions that: 1. while sites in both glaciated and unglaciated terrain are strongly associated with perennial streams and washes, those in ground moraine areas (primarily in Alberta) also frequently occur with rivers; and 2. that jump sites in ground moraine most often face north while those in unglaciated terrain frequently face either north, northeast or south. Site locational variation is also apparent within the sub- regional landform units of Montana and Wyoming. Case locations, regardless of their location in those two states, are strongly associated 103 with loam soils, but bedrock type varies with topographic setting. Plains (tableland) sites in areas of moderate and high relief most frequently occur with sandstone and shale bedrock formations (though formations may not be surficial), while Plains sites with considerable local relief and those in open high mountains are most often associated with a variety of sedimentary rocks. Water features, however, appear to have perhaps the most striking amount of site diversity. The nearest water source to Plains and open high mountain sites tends to be washes, whereas sites in high mountains more frequently occur near perennial streams. When permanent sources of water are considered separately, there is a tendency for sites in both open high mountains, high mountains, and open high hills to occur near perennial streams, and Plains (tableland) sites near rivers. CHAPTER V CONCLUSIONS The analysis of prehistoric bison jump site locations in relation to cultural and environmental variables performed in this study has revealed several important site location patterns. These results provide evidence for immediate interpretation and speculation and possess con- siderable predictive potential and research value for future investi- gations. In this final chapter an interpretive framework will be sug- gested for the critical variable relationships identified in Chapter IV, followed by a discussion and revision of the general hypotheses outlined in the Introduction in relation to the results of the analysis. Finally, there will be a discussion of the significance of the study in relation to its practical applicability for future research. The 146 prehistoric jump sites considered in this study were evaluated on variables reflecting various cultural and environmental features. The cultural variables focused upon the age and selected topographic feature options available at site locations while the environmental variables evaluated specific aspects of: water asso- ciation, geologic feature association, vegetation association, soil association, and topography. Of these seven major feature asso- ciations, two reveal close associations with sites, suggesting strong site distributional patterns and providing predictability in site 104 105 location. Given the results of the statistical analyses, supplemented by archaeologically and ethnographically available evidence, the most critically important variables to occur in association with site loca- tions are (1) the direction in which jump sites face, and (2) water features. The strong patterning observed in the JUMPFACE variable strongly suggests a single cause, while several factors probably con- tribute to the close association between sites and water features. Following is a discussion of each of these associations. Several alternative causes for the association of water features with jump sites are suggested by the fact that the nearest water source to most sites is either a wash, perennial stream or river (Tables 9 and 11). One explanation suggests that the sample case population is skewed toward areas associated with watercourses, especially permanent ones. While all of the areas surveyed by field workers who recorded the bison jump sites used in this study are not known, it is likely that at least some of the surveys were entirely carried out along watercourses with no attempt to cover adjacent regions where there may have been additional sites. If many of the surveys were carried out in this way it could partially explain why most jump sites appear to occur along watercourses. Nevertheless, site proximity to water sources may represent a strong site locational pattern despite some distortion in the data base. Assuming that this is an accurate tendency, the case associations can be explained in other ways. For instance, it is possible that the dissected terrain often created by permanent (and ephemeral) watercourses provided the required topographic setting to effectively operate bison drives. Steep slopes or sheer cliffs were needed for jump locations and the 106 partial impediment which watercourses often created to bison movements may have helped funnel them to particular locations. It can also be argued that jump sites are frequently in close proximity to permanent water (within 800 meters) because of a human need for water itself. Camps were often established at jump sites where peOple remained for weeks at a time subsisting on the killed animals. A dependable local water source would have been necessary to sustain such a group. Thus, the availability of water may have been an important consideration in the decision of where to construct a jump. Jumps may also have been constructed near permanent water sources to take advantage of bison movements to such sources or at least to take advantage of the bison's protracted presence within a limited area surrounding a permanent water source. In contrast to the numerous causation hypotheses proposed for the above association, considerably fewer potential explanations appear available concerning the direction in which jumps face. Statistical analyses of this variable indicate that jumps tend to face north, northeast, east or south and that there is a decided preference for north facing sites in glaciated locations. One explanation for this might be that the site sample is skewed, caused by over-selection of sites with one or another facing direction. Because the sample was non-randomly collected the validity of this argument is, of course, unknown. If one assumes, however, that the sample accurately reflects the total site population in the region, at least two alternative explanations of the situation are possible. First, it is possible that the directions in which jumps face are a natural result of the particular orientation of the surficial 107 topography, i.e., the direction in which watercourses flow and cliff rock breaks off. This was probably a limiting factor in the directional choices available, although because of the meandering nature of many watercourses in the Northwestern Plains, depending upon one's location, it is possible to find sufficiently lethal cliff drOps and steep slopes facing in any or all direction. Thus to fully explain these patterned facings yet another explanation must be sought. One which may, in large part, help explain this patterning are seasonal prevailing wind patterns. A slight digression is necessary to fully explain this hypothesis which emanates in modified form from ideas suggested earlier by H. P. Lewis (n.d.) and several members of the 1961 Symposium on Buffalo Jumps (Malouf and Conner 1962: 44-45). It is well-known that bison possess an exceptionally well- developed sense of smell (Hornaday 1889: 418; McHugh 1972: 149; Garretson 1938: 46). As such, prehistoric hunters of bison would have found it necessary to remain upwind of the animals most of the time to avoid detection during drives. This could be accomplished by moving the bison wi£h_the wind so that during a drive the animals would not sense hunters ahead and to the sides of them (possibly hiding behind rock cairns); rather, they would tend to flee from a few hunters or "runners" they sensed moving up behind them gently urging them on. If a jump was repeatedly used, this technique might also keep the animals from balking at the scent from rotting carcasses of other bison driven over the same jump one or more seasons before. The importance of wind direction in driving bison has been historically documented for the Northern Plains Indians. Several descriptions of Assiniboine drives stress the point that a favorable 108 wind was required before the drive could begin (Henry and Thompson 1965: 519; De Smet 1905: 1029-1030). When blowing from this direction runners could slowly start the bison moving by setting small dung or grass fires which would drift toward the animals and cause them to slowly move away from the smoke and toward the enclosure (Henry and Thompson 1965: 519; Weekes 1948: 16). The smoke may have also helped mask the human scent of the runners which could be even more frightening to the bison than smoke. The same technique was described by McDougall (1896: 278) for the Cree Indians in Saskatchewan. The need for favorable wind direction to successfully drive bison was also documented among the Blackfoot (Henry and Thompson 1965: 577). In this case, the Indians made several attempts to drive the animals toward the pound, but were not successful until the wind shifted to a more favorable direction. Depending upon the location in the Northwestern Plains, pre- vailing winds may or may not tend to shift with the change of seasons. For instance, in Sheridan, wyoming prevailing winds are consistently from the northwest year around whereas in Great Falls, Montana they are consistently from the southwest and in Helena from the west year around (Cordell 1971; Lowers 1960). In contrast, in Havre, Billings and Miles City, Montana prevailing winds tend to shift from season to season and even month to month. In light of these examples of Plains wind patterns, it would have been desirable to change the direction of the bison drives in areas where seasonal wind patterns shifted, if bison were jumped during more than one season of the year. The site data displayed in Table 6 suggests that such seasonal adjustments may have been made in certain areas. Such changes might take the form 109 of several different sites in one area with different facing directions. At the same time, if jumps were only used during one season or in an area where the prevailing winds were constant year around, there would be little reason to have sites with different directional facings. If the data displayed in Table 6 represent accurate population tenden- cies, it suggests a powerful predictive method for determining site seasonality. As an example of this, it could be predicted that jump sites located in the Miles City area having north to northwest facings would probably represent either summer or late fall kills since prevailing wind patterns at those times are from the southeast and south-southeast. Similarly, those sites having south to southeast facings would probably represent winter or late summer kills since at these times the prevailing wind direction is from the northwest. Ideally, after compiling wind direction information for all sites, the information could be plotted on maps with seasonal use areas delimited. If this variable represents a true tendency in seasonal site use, such a map could have considerable research value, not only in its own right, but also by providing an independent test of other seasonal data. It is recognized that this hypothesis is highly speculative and that jump facing and wind direction were only some of the many considerations in choosing a site location and season for jumping bison. Nevertheless, it suggests a potentially useful method for establishing site seasonality and differences in seasonal population movements in the Northwestern Plains. Only one substantivi objection to the idea can be immediately discerned: the fact that some drive lanes are circuitous and thus would not always follow prevailing wind directions. The effects of 110 local topographic conditions, which may change wind patterns, could account for some directional differences in drive lanes. Regardless, however, the site data at hand reveal relatively few sites at which drive lane patterns are described at all and even fewer with circuitous or angled turns. Unless there exists substantial archaeological and environment data to prove this idea untenable, only additional seasonality determinations from faunal material in jump site deposits will suffice to verify its potential worth as a predictor of site seasonality. Hypotheses The results of this analysis made only partially confirm (or partially reject) the two hypotheses proposed in the introductory statement of this thesis. As such, they require modification to accurately reflect the probable site locational information discovered during the variable analysis. Following is a restatement of those hypotheses incorporating the new information. As a result of detailed statistical analysis of bison jump sites in the Northwestern Plains it has been demonstrated, and it is expected that additional research will confirm, that: 1. Bison jump site locations show a definite, predictable pattern most closely and importantly associated with water features. Landform units and surficial geologic features provide modi- fying, but secondarily important, influences upon water feature association. 2. Critical environmental and cultural characteristics of bison jump sites do not tend to remain consistent over the entire Northwestern Plains. a) Types of water features nearest jump sites vary from area to area (although close proximity to water sources is constant throughout the region). 111 b) Jump site facing directions tend to vary from area to area. In light of these modifications it is clear that there are fewer identifiable environmental features critical for jump site location than originally proposed. As a result, future investigators should be able to more closely examine the few critical variables identified in this study and explore additional features not yet considered. Significance for Future Research The mass of detail produced by a study of this magnitude often makes it difficult to step back and properly assess its practical significance. Nevertheless, several aspects can be singled out which have value for future research efforts. First, and most importantly, this study provides a testable predictive model of bison jump locations in the Northwestern Plains. While only a few environmental variables were identified as most crucial for site location, the combination of variables covering soil type, vegetation type, surficial geologic feature type, bedrock type, and water type can effectively delimit general conditions under which sites are most likely to be found. Through field tests the model can be refined to make it a most valuable predictive tool. In the process, it can also provide an important source from which to interpret site distribution in terms of subsistence strategies, site seasonality, seasonal population movements and the place of communal bison jump sites within the larger socio-economic systems of prehistoric cultures. The study also indicates considerable deficiencies in the jump site data base throughout the Plains. Additional site information 112 and especially improved quality of the information are necessary to allow development of a more solid framework upon which to formulate more refined testable hypotheses and to speculate about socio-economic patterning. There is a particular need for more temporal information on jump sites. Currently, a very small number of sites are dated providing only a very general time frame for the use of this bison procurement method in the Northwestern Plains. Changes which occurred in the form or function of this method over time are documented at only a few individual sites and are so little known regionally that even speculation is difficult. Finally, this study points to the need for additional locational analyses of sites in the Northwestern Plains. Regional efforts such as Loendorf's analysis of campsite selection patterns in the Prior Mountains of Montana (1970) are an invaluable means of establishing site to environment and site to site relationships as well as identifying associations with which to speculate about the subsistence strategies and other socio-economic aspects of resident prehistoric and proto- historic cultures. APPENDICES APPENDIX A IDENTIFICATION OF VARIABLES Acronym ID STATE COUNTY PRESEARCH TYPEPR SOURCE LOCRELY AGE ASSOCAMP DRIVLANE TYPEFALL JUMPFACE JUMPHEIT WATERONE WATERTWO VEG SITESOL AREASOL CORSESOL STONISOL LANDFORM SITETOPO AREATOPO SITEROCK AREAROCK PRSRTOPO WATERDIS PERMWATR DISTPERM APPENDIX A IDENTIFICATION OF VARIABLES Definition case number state or province county in U.S.A. site is in previous research known type of previous research source of information about site locational reliability of site cultural period of site associated campsite associated drive lanes type of fall at site direction jumpoff faces height of jumpoff nearest water source to site secondary water source to site type of vegetation at site soil type predominant at site soil type predominant within standard area relative coarseness of predominant soil in standard area relative stoniness of predominant soil in standard area major landform designation for site surficial geologic formation at site surficial geologic formation within standard area predominant bedrock type at site predominant bedrock type within standard area predominant surficial geologic feature at site distance to nearest water source nearest permanent water source distance to nearest permanent water source 113 Acronym NUMWASH NUMSTRM NUMSOL PERCTSOL ELEV NUMTOPO AROCTOPO STOCTOPO 114 Definition count of intermittent watercourses over three kilometers in length within standard area count of perennial watercourses over three kilometers in length within standard area count of soil classes in standard area percentage of standard area occupied by predominant soil type elevation of site in feet count of surficial geologic features in standard area percentage of standard area occupied by dominant surficial feature percentage of standard area occupied by surficial geologic feature predominant at site APPENDIX B LIST OF VARIABLE CODES APPENDIX B LIST OF VARIABLE CODES STATE 1. Wyoming 2. Montana 3. Alberta COUNTY 2. Big Horn, Montana 4. Blaine 7. Carbon 8. Carter 9. Cascade 10. Chouteau 11. Converse 12. Crook 15. Dawson 17. Fergus 19. Gallatin 21. Glacier 23. Hill 25. Jefferson 26. Johnson 28. Lewis and Clark 31. Madison 32. Meagher 36. Park, Montana 39. Phillips 43. Richland 44. Roosevelt 47. Sheridan, Wyoming 48. Stillwater 49. Sweetgrass 50. Teton 53. Valley 56. Wheatland 99. Alberta PRESEARCH 1. known 2. unknown 115 116 TYPEPR 1. survey 2. test excavation . excavation . documentary research none 1 and 2 1 and 3 1 and 4 2, 3 and 4 all HP MNQVO‘U‘L‘M 0... SOURCE Site Form Published Report Unpublished Report 1 and 2 2 and 3 l and 3 All \JO‘U‘IbU-JNH O LOCRELY . Within 8 secion Precisely located Within % of k section Located within k section Located within section UIJ-‘WNH O AGE Late Prehistoric Late Prehistoric and Historic Late Middle Prehistoric Late Middle and Late Prehistoric Late Middle, Late and Historic Early Middle and Historic Early Middle and Late Middle Early Middle, Late Middle and Late Prehistoric CDNO‘UIJ-‘WNH O ASSOCAMP 1. Known 0. Unknown DRIVLANE 1. Known 0. Unknown TYPEFALL 1. Other 2. River or Creek Bank 3. Steep Embankment 117 TYPEFALL continued 40 5. 6. Steep Embankment and River Cutbank Talus Slope Talus and Steep Embankment 7. Bluff or Cliff 8. Bluff and Cutbank 9. Bluff and Steep Embankment 10. Bluff, Steep Embankment and Cutback ll. Bluff and Talus JUMPFACE 1. SW 2. SE 3. NW 4. NE 5. W 6. S 7. S, SE 8. E 9. N 10. N, NW 11. N, NE JUMPHEIT 1. Over 50 meters 2. 40 to 50 meters 3. 30 to 40 meters 4. 20 to 30 meters 5. 10 to 20 meters 6. Under 10 meters WATERONE 1. Lake 2. River 3. River, Spring, Lake 4. Wash 5. Wash, Spring 6. Wash, River 7. Perennial Stream 8. Perennial Stream, Lake 9. Perennial Stream, River 10. Perennial Stream, Wash 11. Perennial Stream, Wash, Lake WATERTWO l. Marsh 2. Lake 3. Spring 4. River 5. Wash 6. Wash, Spring 118 WATERTWO continued 7. Wash, River 8. Wash, River, Lake 9. Perennial Stream 10. Perennial Stream, Lake 11. Perennial Stream, Spring 12. Perennial Stream, Wash VEG 1. K64/Southern Prairie = grama-needlegrass-wheatgrass 2. Cordilleran Forest 3. Parkland 5. Northern Prairie 6. N. Prairie, Parkland 7. N. Prairie, S. Prairie 8. K98 - Northern Floodplain Forest 9. K66 = wheatgrass-needlegrass 10. K64, K66 - grama-needlegrass-wheatgrass/wheatgrass-need1egrass 11. K63 - Foothills Prairie 12. K63, 64 8 Foothills Prairie/grama-needlegrass-wheatgrass 13. K55 8 Sagebrush Steppe l4. K16 = Eastern Ponderosa Forest 15. K16, K64 - Eastern Ponderosa Forest/grama—needlegrass-wheatgrass l6. K15 - Western Spruce-Fir forest l7. K12 - Douglas Fir Forest 18. K12, K55 8 Douglas Fir Forest/Sagebrush Steppe 19. K12, K15 - Douglas Fir Forest/Western Spruce - Fir Forest SITESOL l. Alluvial 2. Eroded 3. Clay 4. Clay Loam 5. Clay Loam, Clay 6. Silt Loam 7. Loam 8. Loam, Silt Loam 9. Sandy Loam 10. Sand AREASOL l. Alluvial 2. Eroded 3. Clay 4. Clay Loam 5. Silt Loam 6. Silt Loam, Clay Loam 7. Loam 8. Loam, Clay Loam 9. Loam, Silt Loam, Clay Loam 10. Sandy Loam 119 CORSESOL 1. Gravelly 2. Coarse, Medium, Fine 3. Coarse, Medium, Fine, Stony 4. Coarse, Medium, Fine, Cobbly 5. Coarse, Medium, Fine, Cobbly, Stony 6. Coarse, Medium, Fine, Gravelly 7. Coarse, Medium, Fine, Gravelly, Stony STONISOL 1. Rockland 2. Stony 3. Stony, Rockland 4. Not Stony 5. Not Stony, Rockland 6. Not Stony, Stony 7. Not Stony, Stony, Very Stony LANDFORM l. Mountains (Alberta) 2. Mountain Foothills (Alberta) 3. Alberta Plain (Alberta) 4. Alberta Plain, Mountain Foothills (Alberta) 5. D6 - High Mountains (3000-5000 feet local relief, less than 20% of area gently sloping) 6. C6a - Open High Mountains (3000-5000 feet local relief, 20-50% of area gently sloping, more than 75% of gentle slope is in lowland) 7. C6a, D6 - Open High Mountains/High Mountains 8. C4b - Open High Hills (500-1000 feet local relief, 20-50% of area gently sloping, 50 to 75% of gentle slope is in lowland) 9. 35d - Tablelands, high relief (1000-3000 feet local relief, 50—80% of area gently sloping, more than 75% of gentle slope is an upland) 10. B5b - Plains with low mountains (1000-3000 feet local relief, 50-80% of area gently sloping, 50 to 75% of gentle slope is in lowland) ll. B4c - Tablelands, considerable relief (500-1000 feet local relief, 50-80% of area gently sloping, 50 to 75% of gentle slope is on upland) 12. B4c, D6 13. B4b - Open High Hills (500-1000 feet local relief, 20-50% of area gently sloping, 50 to 75% of gentle slope is in lowland) 14. 33c - Tablelands, moderate relief (300—500 feet local relief, 50-80% of area gently sloping, 50 to 75% of gentle slope is an upland) 15. BBC, C6a 16. B3c, B4b 17. 33b 18. 33b, D5 120 SITETOPO 1. Unglaciated 2. Dissected mountains 3. Bedrock 4. Alluvium 5. Alluvium, Bedrock 6. Remnant stream terrace bench and alluvial fan 7. Glacial lake bed 8. Glacial lake bed, Alluvium 9. Glacial and outwash channels 10. Glacial and outwash channels, Remnant stream terrace bench ll. Glacial channels, Glacial Lake Bed 12. Glacial channels, Lake bed, Alluvium 13. Ground Moraine 14. Ground Moraine, Bedrock 15. Ground Moraine, Stream Gravel 16. Ground Moraine, Alluvium 17. Ground Moraine, Alluvium, Bedrock 18. Ground Moraine, Glacial and outwash channels 19. Ground Moraine, Channels, Alluvium 20. Unglaciated, Remnant stream terrace bench and alluvial fan 21. Unglaciated, Glacial lake bed 22. Unglaciated, Glacial lake bed, Remnant stream terrace bench and alluvial fan 23. Unglaciated, Glacial channels, Alluvium 24. Moraine 25. Moraine, Ground Moraine AREATOPO l. Unglaciated 2. Dissected mountains 3. Alluvium 4. Remnant stream terrace bench and alluvial fan 5. Glacial Lake bed 6. Glacial channels, Alluvium 7. Ground Moraine 8. Moraine 9. Moraine, Ground Moraine SITEROCK 1. Alluvium 2. Stream terrace bench and alluvial fan 3. Quartzite 4. Other sedimentary rock 5. Glacial deposits 6. Sandstone, Shale 7. Shale, Alluvium 8. Shale, Sandstone, Conglomerate 10. Limestone, Quartzite ll. Limestone, Quartzite, Shale, Sandstone, Other sedimentary rock 12. Limestone, Shale, Sandstone 13. Limestone, Shale, Sandstone, Conglomerate 121 SITEROCK continued 14. Extrusive fragmental rock, Other igneous rock 15. Andesite 16. Andesite, Alluvium AREAROCK 1. Alluvium 3. Other sedimentary rock 4. Other sedimentary rock, Terrace deposits 5. Glacial deposits 6. Shale, Sandstone 7. Shale, Other sedimentary rock 8. Shale, Sandstone, Conglomerate 10. Limestone, Quartzite ll. Limestone, Quartzite, Shale, Sandstone 12. Limestone, Shale, Sandstone 13. Limestone, Shale, Sandstone, Conglomerate 14. Other igneous rock 15. Other igneous, Limestone, Shale, Sandstone l6. Extrusive fragmental rock, Other igneous rock 17. Andesite 18. Extrusive fragmental rock, Other igneous rock, Limestone, Shale, Sandstone l9. Extrusive fragmental rock, Slate PRSRTOPO Unglaciated Dissected mountains Bedrock Gravel Alluvium Remnant stream terrace bench and alluvial fan Glacial lake bed Glacial channels, Alluvium Ground Moraine Moraine Ground Moraine, Moraine l—‘OCmNOUIbUNl-d O h‘h‘ WATERDIS 1. Within 2400 meters of the site 2. Within 1600 meters 3. Within 800 meters 4. Within 400 meters . Perennial stream . Perennial stream, Lake 122 PERMWATR continued Perennial stream, Spring Perennial stream, River 6. 7. DISTPERM 1. Beyond 2. Within 3. Within 4. Within 5. Within 6. Within 7. Within 5000 meters of the site 5000 meters 3200 meters 2400 meters 1600 meters 800 meters 400 meters APPENDIX C COMPUTER PRINTOUT OF SITE AND ENVIRONMENTAL DATA C APPENDIX COMPUTFR pRINTOUT OF SITE AND ENVIRONMENTAL DATA 1 L 7 ITEROCKQ R9740 dosages! 0U OIDO 020 I210 OO—N 0N u¢m¢zm I Q UHnU0 030 I so O-ON ohm IO. zmmquosoo DU) 005- 0 OF- Odi—MHOII U OH-(VJQUJD oat...“ .042 ONInnDr-J O ’0Q4macffil mus; o I mam baseman: I ((331.00. 00‘ 0- O o I o on" «00091085! 0 mn€¢HDH> n 0| .0 I ow l>-O‘.JU.\DN.J namoowcu om omz 00-4 0 vacuums): oh- OUUO-JUNH H04»: OO\I O—ILLWFOKQW H on. 010th \vch) 04...: o O «.43: 40:21.: (v.1ICMnrc VIA ONV) cum ITETOPO T0 0 DUFthlh-NLHQO- HUN 'Zflfl-UnUH-U XDULDOIWIIILJWC HCQUIPHMUDQF‘C LLVSN>W~DDQJOQC< DATA LIST 123 MISSING VALUES 124 Q Q B Q h Q F N ¢ ¢ Q h h F F F Q h h h F Q m 70 Q C '0 n '0 t H H C é C C ¢ 0 O C C’ t C ¢ O O 0 ¢ ¢ ¢ H N '0 t C’ C O O C C C' C C 0‘ 0* 0‘ 0‘ 0‘ 0‘ a“) In F O 0“ 10 1D In 0* '0 IO 0‘ 0‘ m 0‘ Q Q Q Q Q Q Q N Q Q Q Q Q Q Q Q Q Q Q Q Q P! Q Q Q Q Q Q Q N Q Q Q Q Q Q Q Q Q Q Q Q Q H b N h h h n N ID ID h B h h F h h N F O F O '0 a m n F) 0‘ N 0 h '0 to Q t In '0 F- ") O O C C H H H H H H H H H H H H H H n n n '0 If) n N l0 IO N N n :0 :0 IO IO '0 n '0 n n N N N N O c N H N H o O o O o o o o N o N N N N N D (u N N N N O O O o O O O O N C) N h F N h 5 Q Q n c o o c: c: o o O h o h F N N N O h 0 '0 V ¢ D O t J N O N D O N O N H H H H H H r~ m m m Q Q m m In :0 m m on m m m o o c o o o o o o o O o o c a o o o c o 0 4’ 0" 0" 0‘ C m 0‘ H N N H 0" 0‘ N 0' 0‘ 0‘ H 0‘ 7100 GCNthtNtho-ih hhhehhhh OQOHOO‘DanaeQ90CHOOOOHDwaon-OQQooaxman-«rohoo Q Q It‘s Q c m an Q Q N h H Q 00 N no ‘0 4 tr 0* o c: o a) Q Q 0 0‘ o O 0*- co m a) an c a- mHa‘wcchNc‘c‘a‘mhtooHHhNh r‘umwhmr‘éu”mocthcDWFQFQFGNFFQNNnbthQNFthh DQOWOND‘ OCQC QOCOG‘60H?HGOQOWO¢O¢HWGVOOQD¢QYOOQ soot! ohoNHnoNHQONOUDCHmr-‘LnocaomomcmoNasync—tho: to H r) H N Q 0 H o a N N c: N N h H :0 €- 0‘ c OHONCNOHono—ocwonOHrwfithcoc occéenomNa‘ 4r comma to If) If) '0 F) c n F) I". 6' n '0 n H M n F) F) P) F‘. F. I”. M u: 8 1r. 8' I") 6 Us 1:) c In '0 9') F: H r’? .1 '0 F: 3 TD IO Q If! :0 P“! O F ¢ Q (a an a. N (a no I; a: n a; Pr. HfiHu‘.HnHmH¢Hr-cmHHHNHmh-coHcch-cHmnmficnxdmphfimfih G. 6.. CL 1" N 0.} (C a: a: K o c; o c: o o r~ ~2 c: o N c: :3 N D o c: H m o o n c n In a t Q tr '0 #HC :2) at 00 CD a: co COHN a; N (I. Q 03 a co m at a) Q 00 e a o o o o H c: c: o H o o H c o a H o o n r". n :0 N n H m ax c3 '0 c m m Q ¢ :0 or. a. h a. 0* 0‘ 0‘ m O‘- HO‘ 0‘ 0‘ 6‘ 0* 0‘ 0' O‘- U‘ 0‘ 0‘ 0‘ o. 0‘ C‘ 0‘ 6‘ 0 O‘ 0‘ O‘~ 0* O‘- O‘ 0‘ m 0‘ 0‘~ 0‘ 0‘ 0‘ 0‘ 0‘ UN 0‘ U" 0‘ 0‘ ON 0': h c If 0 In 0 a: c: c: C Q on N H H n c: 0‘ o O‘ '0 If) r) n n n n n n '0 n n n :0 '0 n to If) n 0‘ n :0 O‘- HNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHN HfiCCC‘uN'JnUC UTWHDGFFNNU‘O‘OOHHNNW'OQ‘CWID\D‘DFFNNU‘U‘OOHH HHHHHHHHHHHHHHHHHHHHNNNN 125 r~ Q r~ o v~ o r~ r~ m o o o h h Q r~ h o o h h c N 7~ o n o n H c a c: o c c n t c o a t c t N c o c o c t N a o o c c C c t o o c t m o a m a o o :3 ex :0 c 0% o N o o 0 tr :0 o 0‘ Q Q Q Q 0 D Q Q Q Q Q Q Q o a a o o o c: Q Q Q Q 9 a Q Q Q Q Q Q Q Q a a a o a o a h c c h o a o o h h 0‘ h a Q a o o h h o h c o a Q o c: o o n c 1.0 Q o F a 9 ea c n o n H H H H H H H n c t n c n c c n n n n N n n n n n n n n o N N N Q N c n c N N N H o N Q c o v~ c Q a N N N N N N N N N N N N N N N N o '0 N IO 0 b c h h h n n h b h at h h c h o h o n H h F Q N h h to n h h h h h r~ h c h h F m N u": u" Q m In H In 10 m m m ID '0 :0 Q 0') H m In H In a a c a c: o a a o o o o a o a o o o o c: a HO‘UJDC‘OtUOINODOO‘OU‘N 0‘ 2 0 0 l FND‘OCVDNHHOOONFNDCOOFF H ONNCOQNHCQDQWNOQ'FNQCDONOHOQOOWQ¢NOCDQONOGOQ NQCNCDCWQLTNHIDGFNQQQHQN 0‘ C‘UJGEID QI-‘CDQDGN 0‘ O ‘5 OCH OGDQQO QO‘QH C‘QG H g 0‘ Q Q n m m G a N 0‘ '0 '9 Q O O C Q Q V) CC ‘0 NWV‘OLNNNFFQNNFCFGU‘U NWHQHONCFFNiFCFNOGDFFENn (I a: C5 N U, ‘C CD QHQH C 61000.”: motOQOQOQDCOMOnDCOCONONOQONOQONONONON HNou' ahomraoomou afiomoau:oomoaomoNooomQa‘HmHNcn Q N '0 0‘ m 0* h N H m (h 0‘ c to Q m h 0‘ 0- «:r h onowooomoQomacoomooaHmhmoa‘ocOHOQOQOONONO each m n c O‘- IO U“ n n '0 0* 0‘ 0‘ c n n c c 0‘ 0‘ n N If) :n c d N €' € 0 Q Q G c '0 H N t m «I G' F". n N F H d' {'3 N Q Q N a: O H (I m f’) N H co Q :n b HQHQHcHQHnHQHU; Hc-Hq-HhHmnFHanHm-HIOHFQCHGHcHn OWQHQHHHHHOQODOOOOQOOQ wnmccmcchcnmmnmncpmcm Q Q o c H c H H H H H o H o c: o c: o c: H o c: o F F) ¢ c: m 0- c N N 0‘ O‘~ U~ 4‘ O‘ t N m m 0‘ C 0‘ a" 0‘ c 0‘ 0‘ wow» 0 m o-occmmaxcr mom m c- 0‘00‘00‘ INT 0‘ m m m 0‘ m 0‘ OK a~ 0‘ 0‘ Ch 0' m m as c~ a: 0‘ 0‘ O c: N N C: N 0‘ F‘: c 0‘ 0* 0‘ a o 6‘ H N H a~ 87- N m m F) If) n n no-n F) noncnonon men :0 IO nonmn [00‘ HNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHN NNnncc meQCQF-O‘QDHHNNnnc{QJWQQFFQQmO‘ OOHHNNVON‘: NNNNNNNNNNNNNNnnnnnmnmnnnnnnmnnnnncc'c tecto- 126 In 9 F Q F F F F F F F F F F In In In F O Q o n a c IO 9 Q Q I0 t c c C It «0' t c e «r c Q Q 0 o c t O Q Q c Q C Q C' c c c n C c I0 0 Q 0‘ IO IO II”.- 9 IO IO 0 o N H H H H H H H N N H H o c Q o o N N Q Q 0 I0 '0 In to Q Q Q If) H Q Q H H H H H H H H H O Q Q o o N N Q Q 0 IO IO IO :0 Q Q Q n N Q Q H H H H H H H H F F F F o F F :3 Q N H H H H H H H N N H H C c c F O I!) Q Q Q N H H H H H H H N H H H H H H H H IO If) I0 I!) I0 If) If) H H Q Q Q Q Q Q Q Q m In F F H H d' o Q c Q o ¢ H H c c c t C c C 4- c O c: Q (‘0 O o N o N N H H N N N N N N N N N 6' O D F O N F F F F c c c t c Q c F F Q Q o c N o N N N F F F 0‘ F F d‘ F In F F F F D c: m m m an n n n N N F H H H H n t H F F m m H H H H H H H H H o O o a Q Q Q o o c O a Q Q Q a Q Q Q o o a- c: .13 an o c: 0 ID a .n m L” I N In a a m I“ Q 3') Q u H 3 O F c o c: o 0‘ o F F C) O F c c ¢ 0 c O a H H H H mmomcmnoacncHnHQchQQNQNOQmNoomoQQooQooch FHHNQHHQQQQQQQQQQOQOQ 1.". o a Q QQQ Q QQQQQ N c c Q @HC‘HHHU‘HU‘HOHQH H I." 0‘ Q F 1 m N Q Q Q N N Q N o a o c: o c Q IfFNancLGO-QchaanHQFEFQFQFmFQFQFQF:HQHQFQQO :6 H Q Q H H HHHHH H H OI!) onononowoccc OQHQHNHIOGIOHIOHIOOHOHHHOHHHOHOH uooa~ou)HNQo\ count ooHcoooocoooooooooHcoucoomom u‘: c~ N Q 0‘ a- a\ o .n In F c N O IO Q Q m o o 0‘ GQOO‘OFGWOC‘ omomQchonOmomQQomooochocococxom N o~ IO N O 0‘ m c c c If) n IO If: e c c c- Q 0- 0‘ Q n n a c c 1.”) c c- c o c N N n n u": 1.0 a :r H N a) Q If) N F F If) If; x I Q Q Q Q Q Q Q Q (I Q HmHmHQHnHr3H¢HmN¢N¢QQQQQQQQQQHQHQNQQQFQHQHQ on. Q Q Q Q Q Q Q Q Q Q c: CC CC I..: o o c: c: O O Q Q H H H H H o c. '0 J“. H o o u: N I '0 8 c 3 33 (2 N m N! ~17 m N O.) Q: N CD GD 00 Q Q Q Q’. Q Q Q Q Q Q Q Q Q Q Q Q L4 0 C) O O D O H H H H H H H C) 0 H H H O O H O‘- L‘ O" 0‘ 0‘ 0‘ IO n Q m In m m H H N Q m 0‘ O‘ 0‘ aura O‘O‘O-NO‘O‘G‘OO 0‘ N N N N N N N N N F QO‘QO‘ 0 a~ O‘ 0‘ O- (I- U‘ 0‘ a“. c 0 IO 0‘ ON 0‘ 0‘ O :3 0‘ O c: N o In F Q m 0‘ 0‘ 0‘ "3 "0"”; nonwhmnom WHO] NHNHNHNH‘V N N N N N05619- Ho;HNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHN o cmmQQFFNQQOOQHHNNQQFFQQO-O‘COHHNNIOIOG¢ICIDQQFF ¢c¢¢ Q are Cctc QmmmmnmmmmIDIDIDIDIDQQQQQQQQQQQQQQQQ 127 F F 0’) F Q n n F F Q Q n c F Q Q Q F n F F n N n c N N n c c Q t m C Q Q Q n N n n :0 ¢ 4r n c n N «t c C O ¢ ¢ 6 Q Q Q '0 C '0 Q 6' F F F F Q F F 0‘ H H H H Q H Q F O H 0‘ m U“ H Q H n c I!) Q Q Q Q Q Q Q n F m If) If: to Q Q Q H H Q H t c Q Q F Q Q Q Q Q C In H H Q 0' Q Q Q H In In H In In .0 In F F H H H Q H an m Q H F F F F F H F N F F n H H H H N H Q H c H Q n n N N H H N H H H c c H F H t H c c c Q Q Q Q Q F N Q 0 Q m H H H H H H H H H H H H H Q c Q Q Q G Q c «C' 0' Q O O Q t c Q Q Q Q 8 F c u 7') F F N F N N N F O N a N c F F N N N F Q F F F c F F F Q F Q Q Q U) Q F F F H H H F J F F F n F c H F H Q Q Q Q Q H F F H F Q 6; H H H H N H H 0‘ H Q H F H H H H H H H H H H H H H H H o Q O O O O O D O D O O O O O D O O O O O V) J N m 6' J? a“. m 0‘ D ID N 0‘ c d o O u‘. .31 F m u" H H N m c F ¢ O t F F O 3 0' 0 F C) O N ID 0 N N QFQanQnQHmQQnQQQHQQQa‘momeQQFQecQQFQQQQQQ 0* Q I“ Q H Q Q Q N O F Q L“) Q N c Q Q Q Q Q Q QHQ Q 0- Q Q QHQ QHQ c In U‘HO- Q QHQ QHO‘HQH F Q n M Q Q IO Q H Q 0‘ UK In Q '0 Q C: F Q Q Q F3“ QQFmFroFQQQQQQQQQNFFQHQNQDFHJIHQQQFQFQF O H H H H H H HH H H H QNQHQNQNHIOQNQNQHQcQHQNQNHmQHQQQNHHQQHHQHQH QQQQQQ .‘JOODHCHJQOOOCCo‘csHlnHOHQv-OOOD‘HO‘OCJHJ‘;CJU)OJ)QO a CC. Q Q In N Q H c 0- N N Q at 0‘ IJ‘ H Q Q U) m QmQFQc Qa QQQQQQQmochQQONvacQc‘QmQHQmQFmQQQ IO M n n r. n a- N K) 0‘. '0 tr.- 3 m 0‘ 0‘. a- c N N N a O H N 1’) c an m n «r: F: n m H ¢ N) H m c d‘ c '0 Q G H O Q Q c N C: Q Q Q Q Q Q a. u: Q N Q HchquHFHFancHnHFHQHQc mmCVHQHonan'OWKJU‘WFF‘F Q Q Q a. Q Q Q C) Q N C C.) c: Q o H H N N o H H H n H Q Q «n n U1 4' a Q Q Q Q Q to Q Q Q '0 c 0.. c :0 Q Q Q Q Q Q D C c: C) O H O o O O O H H H H U H H H H H F‘: '0 N n H H n N H m m H 0‘ m 0‘ 0* H H H H H m a Q o o w Q Q Q mmFON wwm mmme H n n m H H H H '0 H H H H N N G! N N Q C) Q If: D H H F H 0‘ 0‘ N O" 7‘ 0‘ O“ 0 n m r! C N NHN N NHOJHNHN NHNO‘NO‘NHNO‘N NU‘NO NHNHN N N HNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHN QQ00‘QQHHNNIOIOG'cIDLDQQFFQQU‘O‘QQHHNNIOF‘JG‘CIDIDQQFFQQ QQQQFFFFFFFFFFFFFFFFFFFFQQQQQQQQQQQQQQQQQQ 128 F r~ r~ r~ r~ a o F F F F c F In 4: v~ v~ v~ F r~ «3 '0 '0 7’) F) If) D c ID '0 '0 ¢ 00 n O Q n C C C Q’ 6' G’ C t C t O 4‘ C Q C’ C C C’ C O t C C d‘ d’ '0 0‘ 0‘ 0‘ 0‘ O 0 0 m H H H H z o F 0 Q N N \D 0 ~13 so \D \D so '0 '0 n n '0 '0 Q In G n Q n n n : H \D 0 ~D ‘0 0 t t C N t O O In N t H H c H m \D H H H F F F F F C m H H H H H 0‘ H In H ‘0 N O H d‘ n n r: n r. .0 o m H H H H In a F as H N o w u: H H H H N N H H 0‘ 0‘ 0" 0‘ 0‘ ~13 0 m H H H H H H ~D m H) ID ID 0 m H H H H H H Q C in O SD 0 D O t 0 Q 0 N O O O D O O \D O (”u N N N N O CD :2 N F N N F F O O O O O F O F vs F a c o F F 4' t F F o o o a o F o F N N H F O c O H F t C F d’ O o D O o F O H H H H H H H H H H H H H H F) F n H H H F H H H H H H H H o a a c: o a a c: o c c: a o a o c a o o o o h m L. ) m u ; o 0‘ u ; In In m .n 0* 0‘ c a m m c 0* 0 N O N N \D O t H 43 VD F C \D t C \D D F F O F H H H H cacaococoomnomeOchmancHmcHwamFooccmmcm OOQOOCNNFFFFF¢1HHONN¢ gfimfimHo—‘ofia: U3 c- o O n O'- 0 O 0‘ ‘0 l0 OHO c: O H o o o a o n 0 o N a F m H o N In H o H D ID FOFOFaFoh-OFanocFFMFFFFFFFHmochFFFoanFFc H H H H H H H H oHHHHHHHHHHQ’HmHmONonoNHNod'ONHNHch HHHIDHG'HIO ooooomothuocooc:coco—neocooooouomomomooone“: m w \0 no H q: m ‘0 O \D F m H n O. F N H c o H cocoacomo~aowomo~ooa~OFomOHo—oooomomomomomowoc N N N N N 4r c c '0 n M c c C Q G' a) ID m u) 6' a c a c c n N n In H N n 4: ID H n In a- H n m u; a; CD Q 4: no no Q 0' F 0‘ 0‘ N N C or. CD a a.- F') a. ICF”FIGstIOFnFHwHQHchsomHP‘anNmanmHaH¢nmnuncm a. a) q; a; m a: m c; u: H H H H H H c: O C: s: H O F? O H H H C: H o H '0 d' 3 ¢ F2- 0;; C d; c 0' 10 m c :0 Q a CD 0.: a. F) as C m CD a: a: m or. m :0 H H H H H H o o O o H C) H O H H H o H o H H H H H H 0‘ N N 10 N '0 c N F) N H \O r, F) m m m n n to If: no ID ID no m as «.- as c H H H H H N u; N N N N N N N N N N N N N N '0 n '0 n '0 '0 In C F) to C 0‘ N H G C- c CI; ID F V.) N n c P? F F? N N N N N NO‘-NNNNN N N N N N N NHNHN N N OJ HNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHNHN o‘mOOHHNNnnccmmmehfiwmmooHHnnttmmmeFmemmoo mmommmmmmwoc»mmma‘mmmmmmocoooocooooooooooo—nm HHHHHHHHHHHHHHHHHHHH 129 c F F F t so 0 c F VD c Q F a o ~D \D H 6 ~13 o C 8 ¢ O '0 C O O O Q C" '0 H O a t H t O H O t t c ¢ ¢ 1' O O t n :0 V0 0 D o t 0' '0 O n o N H N H H H H H N n 0 N N 0* 0‘ O O‘- U‘~ a Q 0* so CI: m H so F \o \D \D n N c: 0 Q o \D ~D 0 0 H \D H H H H H H H H t C O t t m N N c t c d‘ 0 '0 s0 0 G \O H 0 H H H H H H N H N H H H H H N N t N N F F F F F F F F N H N H H O H H N N O N N '0 IO '0 n I“) F) U” n N N H H H H H H H m 0 6 0 Q VD In In n n F m In O C t t t c t t H H H H H H H H H o O O 0 \D so so 0 O D D O O o \D so so 0 c c a O O G F I": If) F F O O O O O F n F) F n N N o o o o F F F F F o o c o a F F F F o H F c: F F F F F F F o o o c: o F F F F F H H F 0‘ F F n H H t C o m H o 0" H H H H H 0‘ H 0‘ H H H H H H H H H H H H H o c: o o o o c o o o o a o o a o I: c: o o o c». .n 5". tn :0 m o o c c 6* N In a c: N N m to :0 o H c F o o t o o o F F c l0 H o O c c Q 4‘ H o H H H cocoaocewo‘wcccooocococmowcmFoU‘onmccaccomoo c: c o a un F o o c F