UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Prehistoric cultural change at Kitselas Canyon Coupland, Gary 1985-07-31

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata


831-UBC_1986_A1 C68.pdf [ 23.92MB ]
JSON: 831-1.0058417.json
JSON-LD: 831-1.0058417-ld.json
RDF/XML (Pretty): 831-1.0058417-rdf.xml
RDF/JSON: 831-1.0058417-rdf.json
Turtle: 831-1.0058417-turtle.txt
N-Triples: 831-1.0058417-rdf-ntriples.txt
Original Record: 831-1.0058417-source.json
Full Text

Full Text

PREHISTORIC CULTURAL CHANGE AT KITSELAS CANYON by GARY GRAHAM COUPLAND B.A., The University of Toronto, 1976 M.A., Trent University, 1979 A DISSERTATION SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY i n THE FACULTY OF GRADUATE STUDIES (Department of Anthropology and Sociology) We accept this dissertation as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA December, 1985 (c) Gary Graham Coup land In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of fir^T***?*****),' frlXoL*^ The University of British Columbia 1956 Main Mall Vancouver, Canada V6T 1Y3 Date IA J/Ml/AEV tfU Abstract This study investigates the prehistoric transition from egalitarian to ranked social structure at Kitselas Canyon, Skeena River, British Columbia. It contributes to archaeological theory by developing and testing a model of the evolution of cultural complexity. A culture historical contribution is also made in the development of a prehistoric local sequence for Kitselas Canyon. The theoretical model argues that cultural change at Kitselas Canyon occurred as a deviation amplifying process. Beginning with an initial condition of environmentally limited access to the critical salmon resource, the cultural response was to impose further access restriction by placing control of the resource in the hands of fewer and fewer individuals. An important stage in this process is the formation of corporate groups, initially organized along egalitarian lines. It is argued that ranked corporate groups emerged from egalitarian corporate groups, not as a response to subsistence-related stress, but rather as a means of consolidating control over the most productive resource locations. This model is tested against archaeological data from Kitselas Canyon. The historic period is described to elucidate the ethnographic model, which represents the comparative base for the study of prehistoric social change. Data from the Paul Mason Site at Kitselas Canyon i f 1 is presented. Three cultural components at this site are identified through multivariate quantitative analysis. These results are then integrated with the existing sequence from the nearby Gitaus site to establish a five-phase prehistoric cultural sequence for Kitselas Canyon. This constitutes the temporal framework for the investigation of prehistoric social change. Social change is investigated in terms of a series of variables - residential permanency, storage, population aggregation, household variability, and status goods. Much of the investigation focusses on the Paul Mason Phase, dated at ca. 3000 B.P. This phase represents the development of salmon intensification, residential permanency, large-scale storage, and substantial population aggregation. However, there is no evidence of ascribed status differentiation, as seen historically among the Skeena River groups. This pattern is attributable to the formation of egalitarian corporate groups. This study demonstrates the importance of egalitarian corporate groups in the evolution of Northwest Coast society. Identification of such a group at Kitselas Canyon during the Paul Mason Phase supports the proposition that social ranking evolved as a deviation amplifying process with respect to resource access. i v Acknow1edgements As with any dissertation, there are many people to thank. I would like to start by expressing my gratitude to two people, without whose assistance and interest, this project would have fallen short of the mark. Professor R.G. Matson, chairman of my advisory committee, has guided my interest in prehistoric social organization. What words of wisdom exist in the following pages are attributable to his demands for thoughtfu1ness and thoroughness. It has been my pleasure to study under someone who I consider both teacher and friend. Dr. George MacDonald, Director of the National Museum of Man, and a member of my committee, first introduced me to the Skeena River and Kitselas Canyon. He is in many ways the "father of north coast archaeology". Without his support, both scholarly and financial, the Kitselas Project could never have been undertaken. Professor Donald Mitchell of the University of Victoria has helped me to understand some of the finer points of Northwest Coast ethnography, as well as providing valuable archaeological insights. Professor Richard Pearson has contributed greatly to my understanding of complex societies, and has pushed me to consider the reader of this dissertation. Professor David Pokotylo offered constructive theoretical and methodological criticisms, and did a thorough editorial job. To the above members of my committee, I am greatly indebted. V The National Museum of Man through the auspices of Professor MacDonald, and the British Columbia Heritage Trust provided financial support for the project. I am thankful for their assistance. The Kitselas Band administered the Heritage Trust funds. I would like to take this opportunity to acknowledge their friendship and support of the project. To Paul Mason, Ralph Wright, Roy Bolton and Wilfred Bennett, I owe a special debt of gratitude. The University of British Columbia provided me with financial support from 1981 to 1984 through the Charles and Alice Borden Fellowship in Archaeology. For this, I am most grateful. Moira Irvine, the most acknowledged person in B.C. archaeology, prepared the graphs, figures and photos with her usual speed and accuracy. Three seasons were spent in the field at Kitselas Canyon. For their work, often under adverse conditions, I thank David Archer, Al Mackie, Jenny Johnston, Jennifer Hayman, Colleen Rudy, Keith Johnston, Gwyn Langemann, Andrea Burbidge, Irene Hayman, Ruth Murdoch, Eric Hotz, Diva Vinceguerra, Debbie Thayer, Susan Reynolds, Glenn Bennett, Edison Bolton, Paul Mason Jr., Francis Wright, Kjerstin Smith, James Amos, Elaine Moore and Mason Hersey. Special thanks to Richard Mackie, Dave Walker and Jack Heppelwhite. For their help along the way, I thank Marty Magne, Deanna Ludowicz, Dana Lepofsky, Linda Roberts, Mike v i Broderick, Ann Underhill, Linda Burnard and Diana Alexander. I would also like to take this opportunity to thank my family, whose support has been of a more personal nature. To my parents, Donald and Elizabeth, and my sisters Kathy and Laurie, go my thanks for their unflagging support every step along the way. They were always there when I needed them. I am fortunate to be married to a woman, Carol, who has given me only encouragement and support in this endeavour. This dissertation could not have been completed without her. I dedicate this study to Carol and our children, Michael and Arran. vi i TABLE OF CONTENTS page ABSTRACT i ? ACKNOWLEDGEMENTS iv CHAPTER 1: INTRODUCTION 1 Orientation 1 Format of the dissertation 3 History of archaeological research on the north coast 6 Research objectives 22 CHAPTER 2: CORPORATE GROUPS AND SOCIAL RANKING: THE EVOLUTION OF CULTURAL COMPLEXITY 24 DefinitionsDeviation amplification 25 The deviation counteracting argument: a critical review 9 Social ranking as deviation amplification: the Northwest Coast 36 CHAPTER 3: KITSELAS CANYON: THE NATURAL ENVIRONMENT 65 General geographic setting 6Surficial geology 70 C1imate 73 Hydrology 6 Natural flora and fauna 77 Pa 1eoenvironment 99 Recent environment 102 Summary 1 1 0 vi 1 i CHAPTER 4: KITSELAS CANYON: THE CULTURAL SETTING 112 Introduct ion 1 12 Part 1 - Regional Ethnography 113 Cultural and linguistic affiliation 113 Settlement pattern and yearly round 114 Kitselas villages and territories 122 Subsistence technology and material culture 134 Demography 141 Social organization 144 Summary 149 Part 2 - Regional Culture History 150 Early evidence 15Prehistory of the lower Skeena 158 CHAPTER 5: THE PAUL MASON SITE 189 Introduction 18Site description 189 Sampling methodology 194 Stratigraphy at the Paul Mason Site 199 Dat i ng 210 Artifact descriptions 2 16 Summary 265 ix CHAPTER 6: THE PREHISTORIC SEQUENCE AT THE PAUL MASON SITE 267 Lithic analysis of Excavation H 274 Lithic analysis from the Floor 2 excavation 285 Lithic analysis of Floor 9 299 Lithic analysis of Excavation J 30Summary of the Paul Mason Site lithic analyses 317 CHAPTER 7: THE PREHISTORIC SEQUENCE AT KITSELAS CANYON 322 The Bornite Phase 32The Gitaus Phase 6 The Skeena Phase 328 The Paul Mason Phase 332 The Kleanza Phase 7 The Historic Period 340 Summary 341 CHAPTER 8: PREHISTORIC SOCIAL ORGANIZATION AT KITSELAS CANYON 342 Residential permanency 34Storage 350 Population aggregation and village formation 373 Household variability 388 Status goods 400 Summary: Prehistoric social organization at Kitselas Canyon 408 X CHAPTER 9: SUMMARY AND CONCLUSIONS 415 REFERENCES 436 APPENDIX I: EXCAVATIONS 46APPENDIX II: ARTIFACT TYPE DESCRIPTIONS 485 APPENDIX III: FAUNAL ANALYSIS (Linda Roberts) 545 xi LIST OF FIGURES F i gure Page 1.1 Archaeological sites and localities from the north coast area. 7 2.1 Schematic of change in resource diversity and niche width without and with intensification (after Christenson 1980). 41 2.2 Variable states related to the control of resource locations. 44 2.3 Deviation amplifying model of the evolution of status inequality: the Northwest Coast example. 63.1 Skeena drainage of northwestern British Columbia. 67 3.2 Kitselas Canyon. 9 3.3 Surficial geology of the Kitselas Canyon area. 71 3.4 Biogeoclimatic zones of the middle Skeena area. 79 4.1 Tsimshian linguistic divisions (from Duff 1965). 115 4.2 Schematic of Coast Tsimshian seasonal round. 117 4.3 Kitselas Canyon archaeological sites. 123 4.4 Gitlaxdzawk. 126 4.5 Gitsaex. 7 4.6 Kitselas territories of the Skeena Valley. 133 4.7 Prince Rupert Harbour Sites. 160 5.1 The Paul Mason Site, GdTc 16. 195.2 The Paul Mason Site: house floors and excavations. 197 5.3 Schematic matrix profile of the Paul Mason Site. 200 5.4 Schematic profile showing slope transformation during the occupation of the Paul Mason Site. 202 xi t 5.5 "'Cross-sect fonal schematic of house floor transformation. 204 5.6 Plan of floor 2 (depth = 70 cm). 205 5.7 Plan of floor 9 (depth = 70 cm). 206 5.8 Particle size of Paul Mason Site soils. 209 5.9 Radiocarbon dates from the Paul Mason Site plotted against depth of sample. 215 5.10 Implements with varying edge angles. 217 5.11 Implements showing different types of edge wear. 217 5.12 Implements with varying quality of edge retouch. 219 5.13 Modified cobble flake with platform cortex. 226 5.14 Utlized spall with smooth edge. 231 5.15 Saw and sawn fragment. 236 5.16 Implement with ground, bifacially bevelled edge. 241 5.17 "Pencil" showing blunted end. 245.18 Scraper re-sharpening flake. 255 5.19 Pieces Esquilleess (a) and bipolar core (b). 255 5.20 Microblades and microcore rejuvenation flakes. 255 5.21 Element spectre for Anahim I and Mt. Edziza III obsidian. 263 6.1 Unconstrained clustering of Excavation H stratigraphic units. 277 6.2 Constrained clustering of Excavation H stratigraphic units. 279 6.3 Artifact profiles for Excavation H clusters. 281 6.4 Multidimensional scaling of Excavation H stratigraphic units - dimensions 1 x 2. 283 6.5 Unconstrained clustering of floor 2 stratigraphic units with Excavation H clusters. 287 xl i i 6.6 Constrained clustering of floor 2 strat1 graphfc units. 289 6.7 Artifact profiles of floor 2 clusters and Excavation H clusters. 291 6.8 Multidimensional scaling of floor 2 and Excavation H clusters - dimensions 1 x 2. 296 6.9 Unconstrained clustering of floor 9 stratigraphic units with Excavation H clusters. 301 6.10 Constrained clustering of floor 9 stratigraphic units. 303 6.11 Artifact profiles for floor 9 and Excavation H clusters. 305 6.12 Multidimensional scaling of floor 9 strati graphic units - dimensions 1x2. 307 6.13 Unconstrained clustering of Excavation J with Excavation H clusters. 311 6.14 Constrained clustering of Excavation J stratI graphic units. 313 6.15 Artifact profiles of Excavation J and Excavation H clusters. 3 14 6.16 Multidimensional scaling of Excavation J stratigraphic units - dimensions 1x2. 3 16 6.17 Artifact profiles for the three components of the Paul Mason Site. 317 7.1 Artifact proportions for the five phases at Kitselas Canyon. 323 8.1 Cache pits on the east side of Kitselas Canyon. 353 8.2 "Box and dot" plot of between-f1oor spacing for the Paul Mason Site and Gitsaex. 382 8.3 "Box and dot" of house floor areas for Kitselas Canyon sites. 392 8.4 Gitsaex: house floor 2. 6 8.5 Gitsaex: house floor 6. 398 xiv 1.1 Profile of Trench A: south wall. 467 1.2 Profile of Trench B: south wall. 468 1.3 Profile of Excavation E units 1,4, Excavation B unit 2, Excavation D units 1, 4: north wall. 470 1.4 Profile of Trench C: south wall. 472 1.5 Profile of Trench G: east wall. 473 1.6 Profile of Excavation H: east wall. 477 1.7 Profile of Excavation J: east wall. 481 XV LIST OF TABLES Tab 1e Page 1.1 Summary of the North Coast Prehistory Project. 13 3.1 Climate of the lower and middle Skeena area. 75 3.2 Food plants available in the Kitselas Canyon vicinity. 81 3.3 Mammalian fauna of the middle Skeena area. 87 3.4 Avian fauna of the middle Skeena area. 89 3.5 Ichthyofauna1 species indigenous to the Skeena River. 92 3.6 Salmon escapement through Kitselas Canyon. 96 5.1 Radiocarbon dates from the Paul Mason Site. 214 5.2 Element peaks for five obsidian samples from the Paul Mason Site. 266.1 Artifact frequencies and proportions for Excavation H strat?graphic units. 275 6.2 Excavation H distance matrix. 2 79 6.3 Artifact frequencies and proportions for floor 2 stratigraphic units. 286 6.4 Floor 2 distance matrix. 289 6.5 Artifact frequencies and proportions for floor 9 stratigraphic units. 300 6.6 Floor 9 distance matrix. 3 03 6.7 Artifact frequencies and proportions for Excavation J stratigraphic units. 3 10 6.8 Excavation J distance matrix. 3 13 7.1 The prehistoric cultural sequence at Kitselas Canyon. 341 8.1 Cache pit measurements for Gitsaex. 360 8.2 House floor measurements and household estimates for Gitsaex. 364 XV 1 8.3 Cache pit measurements for the Paul Mason Site. 365 8.4 Housefloor measurements and household estimates for the Paul Mason Site. 367 8.5 Dwelling spacing for the Paul Mason Site and Gitsaex. 381 8.6 Artifact density for Component IV. 384 8.7 Housefloor measurements and household estimates for Gitlaxdzawk. 386 8.8 Summary statistics of dwelling size for Kitselas Canyon sites. 392 II. l Artifact type distribution at the Paul Mason Site. 540 III. l Fauna 1 samples and provenience. 547 1 Chapter 1 INTRODUCTION Or i entat i on This is a study of change in prehistoric social organization at Kitselas Canyon, a small canyon on the Skeena River, 145 km upriver from the coast at Prince Rupert. The primary objective is to develop a model of the evolution of cultural complexity - specifically, the transition from egalitarian to non-egalitarian or ranked society. A second objective is to develop a prehistoric sequence for the area, to be used as a framework for the investigation of cultural change. The evolutionary model is then evaluated against archaeological data from Kitselas Canyon and the lower Skeena River. In short, this is a study in "social" archaeology (Redman et a_L 1978:1). There is a famous cartoon in which two scientists are closely studying a complex formula on the chalkboard. The first scientist, who obviously developed the formula, is explaining it to the second scientist. There is a large blank space with numerous question marks at a critical point near the end of the formula. The second scientist is puzzled, and asks what happens here. The first scientist says, "A miracle occurs". For many years, archaeologists approached the origin and development of cultural complexity on the Northwest Coast in the manner of the above cartoon. Developing regional chronologies, based on sequences of phases. 2 archaeologtsts were aware that the social organization of prehistoric Northwest Coast cultures changed fundamentally from egalitarian to hierarchical. They could even suggest when this change occurred. However, analysis of the process of this change was considered beyond the bounds of archaeological research; simply, "a miracle occurred". Social anthropologists have long been fascinated by the rich and complex cultures of the Northwest Coast. A large body of ethnological research now exists, aimed at understanding the evolution and Integration of these cultures. Until recently, however, archaeologists contributed little of value to this research. Over the years, as the methods and techniques of archaeology were continually refined, the focus of archaeological investigations shifted from the materials of culture to the process of culture - from understanding prehistoric artifacts to understanding how prehistoric people 1i ved. This dissertation is concerned with the prehistoric people of Kitselas Canyon. It seeks to understand their relationships with each other, and how these relationships changed through time. In order to investigate this problem, data from two sites at Kitselas Canyon, Gitaus (GdTc 2) and the Paul Mason Site (GdTc 16), are analyzed. These sites are critical to an understanding of the evolution of cultural complexity at Kitselas Canyon. Together, their dated occupation was 3 from 5000 to ca. 2000 B.P. (see Allaire 1979; Chapter 5, •this volume). Most archaeologists agree that ranked societies evolved on the Northwest Coast during this time (see, for example, Ames 1981). It is recognized that cultural change at Kitselas Canyon did not occur in isolation. Data from two other important localities on the lower Skeena River - Prince Rupert Harbour and Hagwllget Canyon - are compared to the Kitselas Canyon material. This places Kitselas Canyon in the larger context of the lower Skeena River. I argue, contrary to an existing model, that Kitselas Canyon and the two localities mentioned above have existed as part of a cultural interaction system for the past 4300 years. A quantitative analysis of lithic artifacts from the Kitselas Canyon sites is employed to establish a cultural sequence for the area. This sequence is then used as a temporal framework for the investigation of social change. Specifically, we seek to establish stages of development in the transition from egalitarian to ranked society at Kitselas Canyon. A series of variables, sensitive to change in social organization, are selected and tested against the data (stone tools and other cultural features) to determine how the transition to cultural complexity occurred. 4 Format of the Dissertation The dissertation is organized into nine chapters. The present chapter gives an overview of the history of archaeological research on the north coast to elucidate the current state of knowledge of the prehistory of the area. Herein, the rationale for the present study is given, and the general problem orientation, including specific research objectives is presented. Chapter 2 presents the research design of the study. Previous arguments for the evolution of ranked society on the Northwest Coast are examined, and a new model of the evolution of cultural complexity is developed. In Chapter 3, an overview of the natural environment of the Kitselas Canyon area is presented, including the general geographic setting, geology and geomorphology, climate, modern flora and fauna, and pa1eoenvironmenta1 history. The cultural setting of Kitselas Canyon is reviewed in Chapter 4. This chapter is in two parts. Part 1 is an ethnographic overview of the Kitselas people, emphasizing aspects of settlement, subsistence economy, material culture and social organization. Archaeological data pertaining to the early historic period at two sites in Kitselas Canyon are used to supplement ethnographic information, and to provide a data base from which comparisons to the prehistoric record are made. Part 2 is a review of the regional culture history of the north coast, with special attention to 5 previous archaeological investigations at Prince Rupert Harbour, Kitselas Canyon and Hagwllget Canyon. Existing models of prehistoric cultural change are examined in this chapter. The present archaeological research conducted at Kitselas Canyon is discussed in Chapter 5. The Paul Mason Site is introduced, and the substantive results of the excavation and survey fieldwork are presented. Chapter 6 presents multivariate analyses of the Paul Mason Site lithic assemblages to identify discrete temporal components. Chapter 7 integrates the Paul Mason Site components with those from Gitaus, a prehistoric site at Kitselas Canyon previously analyzed by Allaire (1978, 1979), to establish a prehistoric cultural sequence for Kitselas Canyon. Chapter 8 examines the evidence for the evolution of cultural complexity at Kitselas Canyon. Expectations derived from the model developed in Chapter 2 are compared to the data from Kitselas Canyon to focus the investigation of social evolution. The concluding chapter discusses the efficacy of the model developed in Chapter 2 with respect to archaeological evidence from Kitselas Canyon. An interpretive statement concerning social evolution on the lower Skeena is presented, and comparisons are made to other areas of the Northwest Coast. 6 History of Archaeological Research on the North Coast The Northwest Coast so well studied by ethnographers and linguists, has been grievously neglected as a field for archaeological research (Drucker 1943:23). Despite Drucker's lament, further archaeological investigations in the area of northwestern British Columbia were not immediately forthcoming. In the southern part of the province, archaeologists led by C.E. Borden were quick to take up Drucker's challenge, and initiated research in the Gulf of Georgia and lower Fraser Valley regions. Another quarter of a century passed, however, before intensive archaeological investigations began to shed new light on the human prehistory of the north coast. In this review, the north coast is defined as the coast and adjacent inland areas from Milbanke Sound at the southern extent of Tsimshian territory to Yakutat Bay on the north (see Figure 1.1). It includes the north coast of British Columbia and the Alaska Panhandle. "Adjacent inland areas" refer to the lower drainages of the major rivers of the area; the Skeena, Nass and Stikine Rivers. Historically, this area is commensurate with the territories of the three northern ethnolinguistic groups: Haida, Tlingit and Tsimshian. The history of archaeological research on the north coast mirrors an archaeological construct that has served many researchers In the past. Following Willey and Sabloff (1974:21-87), there was a "Descriptive Period" and an "Historical Period". The Descriptive Period covers the 7 F i gure 1•1 Archaeological sites and localities from the north coast area. 8 first half of the twentieth century, and was characterized by speculations about recent population migrations and trait diffusions. The ensuing Historical Period, beginning in the mid-1960's, was marked by more Intensive research and the establishment of local cultural sequences, extending in some areas to the end of the Pleistocene. Evidence of an "Explanatory Period" in north coast archaeology Is limited (see Ames 1981). The north coast is the land that the "new archaeology" forgot. The Descriptive Period (1900-1965) Early Research in the Tsimshian Area The first archaeological research in the area was conducted by Harlan Smith during the first decade of this century, as part of the Jesup North Pacific Expedition under Franz Boas. Smith's work on the north coast was unfortunately more cursory than his investigations in southern British Columbia, but he was the first archaeologist to record shell midden sites in Prince Rupert Harbour and Metlakatla Pass (see Smith 1909). Later, Smith (1930) documented 18 sites for the National Museum of Canada along the Skeena River and its tributaries, as far upriver as Kitwanga. These included historic and protohistoric village sites at Kitselas Canyon (Gitlaxdzawk and Gitsaex), and at the mouth of the Kitsumkalum River. There were no excavations. Smith's primary effort was to record and preserve standing and fallen structural remains. 9 Inglis and MacDonald (1979:10) suggest that Smith's research was one of the first attempts at heritage conservation on the west coast. Smith was also concerned with obtaining archaeological evidence from upriver that might indicate that the Tsimshian had only recently migrated to the coast. According to MacDonald (1969a:242), Smith spent much time searching, in vain, for the interior village of Temlehem, the mythical homeland of the Tsimshian. In 1938, Philip Drucker (1943) conducted archaeological survey and test excavations along the northern Northwest Coast from Venn Passage, near Prince Rupert, south to Rivers Inlet, with the general objective of accumulating data that might lead to an understanding of the origin of the ethnographic cultural pattern. Three sites in the Prince Rupert Harbour area were tested and from these excavations, Drucker (1943:112-13) concluded that the area had been only recently settled. This interpretation is not surprising as Drucker, using the Direct Historic Approach, only tested sites with historic surface remains. Like Smith, Drucker subscribed to the theory of recent, interior origins of the Coast Tsimshian (Drucker 1943:34, 111). He stated that there appeared to be little difference between the archaeological and ethnographic cultures, and that the earliest components explored were essentially "coast cultures" similar to those of the historic period (Drucker 1943:112-13). Unfortunately, these "early" components 10 probably date to the late prehistoric period. It was Drucker's (1943:125) contention that the distinctiveness of the northern culture was attributable In large part to the exertion of Influences from the east and from the north. The decades of the 1940s and 1950s were marked by a paucity of archaeological research in the northwest. In 1954, C.E. Borden made a brief reconnaisance into the area, where he recorded and test excavated a possible prehistoric village site (GdTd 1) at the canyon on the Kitsumkalum River. In a recent re-visit to this site, J. MacDonald (1982) recorded a series of elliptical house depressions (not noted by Borden) on a terrace overlooking the river, and cache pits which Borden apparently mistook for house post depressions. James Baldwin, working with Borden in 1954, conducted excavations at the Co-op Site (GbTo 10) on Kaien Island in Prince Rupert Harbour. This work was not reported, however, until much later (Calvert 1968), when renewed archaeological investigations at Prince Rupert Harbour were underway. 11 Early Research in the Tlingit Area In 1949, Frederica de Laguna began archaeological and ethnological research at Yakutat Bay in southeastern Alaska (de Laguna, Riddel 1, McGeein, Lane and Freed 1964; de Laguna 1972). A student of Boas, de Laguna's approach was particularistic. The prehistoric Yakutat culture was considered in terms of its artifact traits, with the objective of determining the origin of these traits. Comparison of the Yakutat material to excavated assemblages from southwest Alaska (de Laguna 1934, 1956) led de Laguna et aj_. (1964:207-10) to argue that northern Northwest Coast culture was "built on a foundation" of long-standing A1eut-Pacific Eskimo contact. This Northwest Coast/Eskimo link was not only expounded by de Laguna. Borden (1962), for example, postulated that this Eskimo contact may have extended to the south coast of British Columbia during the Locarno Beach Phase. The early archaeological research of Smith, Drucker and de Laguna was heavily influenced by the anthropology of Franz Boas. Their approach was that of historical particularism applied to archaeology. Prehistoric cultures were seen as a series of diverse traits (artifact types), and the objective of archaeological research was to document the place of origin of these traits. In so doing, archaeologists could infer the directions of influence -in the form of trait diffusion and population migration - that shaped the prehistoric culture in question. 12 Unfortunately, there was little attempt to explain how this implied interaction led to the development of Northwest Coast culture. Trait diffusions and population migrations were simply seen as explanations in themselves. There was little or no effort to understand the process of cultural change. The Historical Period (1966-1981) Research in the Tsimshian Area The North Coast Prehistory Project, one of the largest archaeological investigations undertaken on the Northwest Coast, was initiated in 1966 by the National Museum of Man under the direction of G.F. MacDonald. With little previous research (Smith 1909; Drucker 1943) on which to base the study, the initial problem was essentially exploratory -to locate and test sites with sufficient time depth to detail the development of the elaborate cultural pattern known historically (MacDonald and Inglis 1981s37). The magnitude of the North Coast Prehistory Project makes a detailed review here difficult. Analysis of cultural material is ongoing, and much of the written material, particularly pertaining to the critical area around Prince Rupert Harbour exists only in manuscript draft form. The geographic scope of the project covered an area ranging from KJsgegas, at the junction of the Skeena and Babine Rivers, to Klusta on the northwest Queen Charlotte Islands. The history of work undertaken by the project has been Out 1ined by MacDona1d summarized below. 13 and Ing1 is (1981:37-43), and Table 1.1 Summary of the North Coast Prehistory Project Pr i nee Rupert Skeena R i ver Queen Charlottes Total recorded sites 200 25 40 265 excavated sites 1 1 4 3 18 radiocarbon dates 121 4 2 127 art i facts 18350 3450 100 21900 source: MacDonald and Inglis 1981:37-43. The investigations focused on Prince Rupert Harbour. Although the data have not yet been fully analyzed or published, MacDonald (1969a), In a preliminary report, was able to delineate a three-period chronolgy covering the last 5000 years. The time frame of this scheme was later altered, but not radically so, by MacDonald and Inglis (1981). The culture history at Prince Rupert Harbour, as presented by MacDonald, and the data on which it is based are reviewed in greater detail in Chapter 4. The antiquity and continuity of this occupation presented a strong argument against the "recent migration hypothesis" advanced by Smith 14 and Drucker. Upriver from Prince Rupert, the North Coast Prehistory Project conducted archaeological investigations at two locations along the Skeena River: Kitselas Canyon and Hagwilget Canyon. Investigations at Kitselas Canyon included excavations at Gltaus (GdTc 2) in 1968 (Allaire 1978; 1979) and at the historic site, Gitlaxdzawk (GdTc 1), in 1971 (Allaire and MacDonald 1971; Allaire, MacDonald and Inglis 1979). There is also a brief report of petroglyphs at Kitselas Canyon (Walker 1979). Ames (1971, 1973) has reported the results of a site survey in the middle Skeena Valley above Kitselas Canyon. Allaire's work at the deeply stratified Gitaus site resulted in the development of a three-component sequence, beginning ca. 4000 B.P. In 1970, Ames (1979a) excavated GhSv 2 at Hagwilget Canyon on the Bulkley River, near its confluence with the Skeena River. The site had been previously tested by MacDonald in 1966, and a carbon sample from this test produced an early age estimate of 3430 + 200 B.P. (GSC 746) (MacDonald 1969a:249). Results of the Gitaus and Hagwilget investigations are discussed in greater detail in Chapter 4. In 1979, MacDonald (1979) conducted excavations at the Kitwanga hi 11 fort, near the modern village of Kitwanga. This site included a small village or defensive refuge (five house floors) on top of a steep hill. This had apparently 15 served as a fort for the warrior chief Nekt, during the early historic period. This excavation, and the ethnohistorica1 research conducted at Gitlaxdzawk (Allaire and MacDonald 1971) were important because they focussed on the practice of aboriginal warfare and the use of defensive fortifications along the Skeena. Both sites were occupied during the early historic period when the fur trade was operative, and an active struggle for control of the Skeena River trade route was occurring (see Mitchell 1983a). Elsewhere along the Skeena, Acheson (1977) test excavated GgSw 5 at Gitseguk1a/Skeena Crossing. The importance of this investigation was the recovery of two chert microblades. At the time, these were the only microblades recovered from Skeena Valley. Unfortunately, the deposit was undated. In 1969 and 1970, Simonsen (1973) conducted archaeological investigation in the Hecate Strait-Mi1banke Sound area, as part of the North Coast Prehistory Project. Of the 108 sites recorded, 55 were habitations, and 40 were tidal stone fish traps. Excavations were conducted at one site in Higgins Passage, the Grant Anchorage Site (FcTe 4). Simonsen (1973:73-75) identified an early and a late component at this site. The late component has two radiocarbon dates of ca. 2100 B.P. Simonsen (1973:74) notes that the principal difference between the components was the increased emphasis placed on sea mammal hunting and 16 foreshore resources in the late component. Research in the Haida Area Another centre of archaeological investigation on the northern Northwest Coast is the Queen Charlotte Islands. Smith (1927) made a brief early reconnaisance to the islands as part of the Jesup North Pacific Expedition, describing a single earthwork site. Much later, Duff and Kew (1957) recorded a number of historic Haida sites during a brief survey. Initial archaeological investigations were conducted by G.F. MacDonald in 1966 and 1967, again as part of the North Coast Prehistory Project. Three Haida burial sites were excavated, including the Gust Island shelter (FhUb 1), Skungo Cave (GbUf 1), and a mass burial site at Tanu (FeTv 1) (MacDonald 1973). One non-burial site was excavated on the Queen Charlottes by the North Coast Prehistory Project, near the mouth of Honna River (FhUa 1). This was a shell midden site with basal dates of 3040 ± 100 B.P. (Gak 1870) and 3300 ± 100 B.P. (Gak 1871) (MacDonald 1969a:249). An assemblage of bone and ground stone tools was recovered which were similar in form to artifacts of the same time period at Prince Rupert Harbour. In 1969, Fladmark (1970) conducted archaeological investigations at Skoglund's Landing (FlUa 1). Excavation of this site recovered a lithic assemblage from a non-shell 17 context, dated to ca. 4000 B.P., constituting mainly retouched flakes and modified cobbles. Fladmark (1970:39) noted that this assemblage, lacking pecked and groundstone artifacts and chipped stone points, was unlike contemporaneous assemblages from Prince Rupert Harbour. On the basis of this dissimilarity, he suggested that early cultural developments on the Queen Charlottes occurred without mainland contacts until at least 4000 B.P. (Fladmark 1970:39). In 1970, Fladmark (1971, 1979a) excavated two Queen Charlotte Islands sites, Lawn Point (FiTx 3) and Kasta (FgTw 4). Two other sites, at Skidegate Landing (FhUa 7) and Honna River (FhUa 15) were surface collected. The importance of these excavations was the recovery of early Holocene assemblages (pre-dating 5000 B.P.) that included a microcore and blade industry. The possibility of earlier habitation (perhaps as early as 10,000 B.P.) of the Queen Charlotte Islands was examined by Hobler (1978) during a survey of Moresby Island conducted in 1974 and 1975. Hobler recorded 10 sites from the intertidal zone, which he interpreted as relating to a time of lowered sea levels at the end of the Pleistocene. In 1973, Severs (1974) excavated at Blue Jackets Creek (FlUa 4), near Skoglund's Landing. A lithic and bone tool assemblage was recovered in shell context, and radiocarbon dated between 4000 and 2000 B.P. 18 Research in the Tlingit-Tah1 tan Area In southeastern Alaska, at Groundhog Bay II, Ackerman (1968) and Ackerman, Hamilton and Stuckenrath (1979) have added to the record of early microblade components, with radiocarbon dates ranging from 9200 to 4200 B.P. The late component at this site has been assigned to the protohistoric period (Ackerman 1968:62) and appears to relate to the development of Tlingit culture in the area. In 1978, Davis (1979) excavated the Hidden Falls site in southeastern Alaska. He identified five components spanning the last 9000 years. The early assemblage included obsidian microblades. Lightfoot (1983) analyzed the Component 2 assemblage (4000 to 3000 B.P.) from this site. It was characterized primarily by groundstone and chipped stone tools. A similar assemblage was recorded by G. Clark (1979) at the nearby Coffman Cove Site. In 1969 and 1970, J.W. Smith surveyed and excavated in the Stikine-Tah1 tan region, in the area of Telegraph Creek, east of the Coast Range. Six sites were tested (Smith 1970), and five cultural components were established, based on the presence or absence of obsidian, microblades, and projectile points. The dates for the sequence, beginning at earlier than 9000 B.P. and ending at 500 B.P. were based on the obsidian hydration technique (Smith 1971:201). The main contribution of this research was the definition of the Ice Mountain Microblade Industry (Smith 1971, 1974), and the recognition of similarities in manufacturing 19 technique to west-central Alaska (D. Clark 1972) and Asia (Morlan 1967). Fladmark re-visited the Mt. Edziza area in 1982, where he surface collected and test excavated three localities. This research resulted in a full description of the lithic complex associated with microblades (not done by Smith), and a refinement of Smith's terminal dates for the sequence. Fladmark (1982b:347) placed the end of the microblade sequence at 4000 to 3000 B.P., substantially earlier than the 500 B.P. date given by Smith (1971:201). Summary In this review of the history of archaeological research on the north coast, a number of points are apparent. The first, and perhaps most important, is that archaeology in this part of the world has lagged behind archaeological advances in method and theory elsewhere in the Americas (see Willey and Sab 1 off 1974). The early research of H.I. Smith, Drucker, and De Laguna was primarily aimed at determining what, if any, archaeological record existed in the area, and where the origins of prehistoric Northwest Coast culture lay. Their work was exploratory. Smith made no attempt at reconstruction. Drucker's model of a north coast prehistoric culture with little time depth has long since been laid to rest. De Laguna's "normative" approach to Yakutat archaeology focused on artifact descriptions, and failed to explain satisfactorily how or 20 why classic Tlingit culture evolved In the area. Moreover, it Is now clear that the archaeology of the Deeriptive Period dealt only with the late manifestation of north coast prehistory (see MacDonald 1969a:242, 246). Following Smith, Drucker and de Laguna, most intensive archaeological research on the north coast has been conducted within the last twenty years. The primary effort of this research was to establish local chronological sequences, and it met with limited success. Chronological frameworks have been proposed for three areas along the Skeena River - Prince Rupert Harbour (MacDonald 1969a; MacDonald and Inglis 1976, 1981), Kitselas Canyon (Allaire 1979),and Hagwilget Canyon (Ames 1979a) - and a provisional sequence exists for the Queen Charlotte Islands (Fladmark 1975, 1979a). However, in each framework, the components are broadly defined temporally, and their relationship to aspects of cultural change, especially in the social realm, is not always clear. Other areas including the southeast Alaska coast, the mainland coast south of Prince Rupert, and the Stikine River-Mt Edziza area have been investigated and have yielded prehistoric cultural remains, but regional sequences for these areas have not been proposed. Still, many other areas including the Nass and Kitsumkalum drainages are virtually unknown archaeo1ogica11y. At this time, there is no single culture historical framework for the entire region, other than a basic early/late division. Indeed, Fladmark (1982a:101) has 21 recently noted that the cultural and environmental diversity of British Columbia may eventually defy simple generalizations on a macro-regional level. Systematic regional studies of settlement-subsistence patterns are almost wholly lacking for the area, although the extensive shell midden excavations at Prince Rupert Harbour have provided an enormous body of data for future study. The main problem in undertaking regional settlement studies has been archaeological visibility. Systematic surface survey in the dense, temperate, rain forest environment covering the region is extremely difficult. Moreover, ground moss or humus covers many small sites (lithic scatters, temporary camps), making recording difficult without shove 1-testing or random test-pit excavation. Both procedures are costly in terms of time and money, and not often productive in terms of finding sites. As a result, most archaeological research on the north coast has tended to concentrate on the excavation of deeply stratified and intensively occupied shell midden sites in fairly accessible locations. Archaeological research on the north coast has generally precluded consideration of regional assemblage variability, despite the fact that it is well known that seasonal movement inland was an important aspect of the aboriginal lifestyle. Most regional sequences developed to date fail to control for (or even consider) functional variability as a factor affecting artifact assemblage composition. Thus, when 22 differing artifact assemblages have been recovered from two or more nearby localities, they have generally been thought to reflect separate, localized populations, rather than discrete tool kits of a single population. Finally, with respect to changes in prehistoric social organization in northwestern British Columbia, some attempt has been made to document the development of classic Northwest Coast culture (MacDonald 1969a; MacDonald and Inglis 1976, 1981), but these efforts have primarily been aimed at describing different stages of development. There has been little attempt to formulate models to explain how or why these social changes occurred (but see Ames 1981). It is to the process of the evolution of cultural complexity that we must now turn our attention. Research Objectives Although gaps exist in all lines of archaeological research on the north coast, one of the least understood areas is prehistoric social organization. With the wealth of archaeological data from the lower Skeena River area - specifically Kitselas Canyon and Prince Rupert Harbour - we are now able to turn our attention to this important area of research. The primary objective of this study is to determine the nature and mechanism of change in prehistoric social organization at Kitselas Canyon. The critical focus is on the nature of social integration during early village 23 life at the canyon. In addition, another objective of this study is the development of a chronological sequence for Kitselas Canyon. This provides a temporal framework for the investigation of social change. 24 Chapter 2 CORPORATE GROUPS AND SOCIAL RANKING:  THE EVOLUTION OF CULTURAL COMPLEXITY Def i n i t i ons As the primary objective of this study is understanding the transition from egalitarian to ranked society, it is important at this point to formally define these terms. I use the definitions given by Fried (1967). An egalitarian society is one in which there are as many positions of prestige in any given age-sex grade as there are persons capable of filling them (Fried 1967:33). This definition allows for considerable status differentiation. The critical point is that in egalitarian societies, status is achieved, not ascribed. Leadership is transient. It rests on authority and competence, rather than power; it is situational, rather than personal (Fried 1967:83). ' A ranked society is one in which positions of valued status are somehow 1 imi ted so that not all those with sufficient talent to occupy such statuses actually achieve them (Fried 1967:109). In ranked societies, status differences are rigidly defined, usually according to a clearly distinguished principle of descent. Unlike egalitarian societies, status is ascribed, although there is generally some room for mobility. Leadership is regular and repetitive, and to the extent that a chief redistributes food and wealth, there is unequal access to resources. 25 Dev i at i on Amp 1 i f icat i on In order to conceptualize the evolution of cultural complexity at Kitselas Canyon, this chapter presents a model that explains how and why social hierarchies evolve. The model examines the relationship between resources (food and wealth) and human social organization. It is argued that access to resources will become increasingly restricted (to members of the population at large) as social organization becomes more complex. When the point is reached at which ascribed status inequalities exist, access to certain resources will not only be restricted to all non-group members of the population, but also to the majority of members within the community. In this sense, the evolution of status inequalities is a process of continually restricting access to resources to more and more people within the society. The process is morphogenetic rather than morphostatic; it amplifies deviation (Maruyama 1963). We are probably familiar with situations of deviation amplification in the form of "vicious circles" - those problems that only seem to compound themselves as events unfold. Maruyama (1963:164) defines deviation amplification as "...all processes of mutual causal relationships that amplify an insignificant or accidental initial kick, build up deviation and diverge from the initial condition". He offers the geological example of the weathering of rock. A crack in the rock collects water which freezes and expands the crack. More water collects, which further expands the 26 crack, and the process continues until it is possible for a tree to take root. The roots expand the crack even more. Thus, beginning with an initial condition (a crack in the rock) and an initial kick (collection of water), a series of deviation amplifying mutual causal processes are set in motion which have the effect of continually expanding the crack. Deviation amplifying systems function differently from deviation counteracting systems. In the former, change results from mutual positive feedbacks among the elements of the system, while in the latter, change results from mutual negative feedbacks. The application of the deviation amplifying model in archaeology has been discussed in general terms by Wood and Matson (1973), who include it as part of their "complex adaptive systems" model. The model was used in a specific archaeological case by Flannery (1968), who was concerned with the evolution of agriculture in Mesoamerica, Flannery argued that chance genetic changes in wild grasses, including teosinte, ultimately lead to maize intensification. The wild grass system was, according to Flannery, one of a number of wild procurement systems in highland Mesoamerica around 7000 B.P. Scheduled and seasonal use of each system served to maintain a high level of procurement efficiency, but prevented intensification of any one system. In this sense, seasonality and scheduling functioned as regulators in a deviation counteracting procurement system. This was the 27 initial condition. The Initial kick that changed the system to one of deviation amplification was the direct result of early "experiments" with the wild grasses to increase yield (Flannery 1968:81-82). Wild grasses had been a relatively minor procurement system, but the combination of genetic changes and human interaction caused it to increase in importance out of proportion to the other systems. At this point, deviation amplification took over. Cultivation lead to further favourable genetic crosses and back crosses which increased the yield even more. This promoted population increase and ever more intensive cultivation. The constraints of seasonality and scheduling were diminished under the demands of prolonged planting and harvesting seasons. Ultimately, the old wild procurement systems became secondary to the new system of domestication. On the northern Northwest Coast, the initial condition was similar to that proposed by Flannery for Mesoamerica. Around 3500 B.P., a number of resources were being used along the lower Skeena River, including land mammals (which may have been the single most important resource), sea mammals, intertidal species, a variety of plant foods, and fish including salmon. Further, it is likely that the constraints of seasonality and scheduling were operative. These resources were exploited when and where they were available. At times of the year when the availability of two or more of these resource systems overlapped, harvesting time was probably divided in varying proportions among each 28 of the systems, rather than concentrated on one system. This pattern of broad-spectrum or generalized procurement minimized the risk involved if one resource system failed in a given year. We may expect that social group size fluctuated in accordance with seasonal fluctuations in resource availability (see Smith 1981). It is unlikely that permanent social groups larger than the nuclear family would form. Fluidity of group membership was probably more important than group solidarity. Now let us consider one resource in particular -salmon. Like the wild grass system of Mesoamerica, salmon was probably not the critical resource for the early inhabitants of the lower Skeena area (ca. 5000 to 3500 B.P.) that it later became. Like wild grass, part of the reason for this had to do with scheduling constraints. The time for salmon harvesting is late summer which conflicts with the optimal time for land mammal hunting and plant gathering. Here, a critical difference exists. Wild grasses were readily available to the early Mesoamericans, but the potential of this system as a subsistence staple was low until genetic changes increased the yield. Salmon, on the other hand had the potential of being a subsistence staple on the Northwest Coast from at least 5000 B.P., and probably much earlier (see Chapter 3). The problem was not the lack of abundance of salmon for the early inhabitants of the lower Skeena, but the lack of availability or access to the resource. Salmon was both 29 abundant and predictable by 5000 B.P., but it was only seasonally available, and the access points or extractive locations were limited in number. Further, in order to make intensification worthwhile, salmon requires intensive processing and storage (see Schalk 1977; Matson 1983:134). In turn, this requires changes in technology and organization of labour. The basic argument here is that the minor importance of salmon to the early inhabitants of the lower Skeena had little to do with the productivity of the resource, but was rather a factor of the temporal and spatial restrictions of access, and the difficulty of processing and storing the resource for winter consumption. In this model, deviation amplification is seen in any ensuing process that has the effect of placing progressively tighter retrictions on access to the resource. The Deviation Counteracting Argument: A Critical Review In order to develop the model more fully, I will argue against an existing model of the evolution of cultural complexity among hunter-gatherers. Reference here is to the concept, popular among functiona1-eco1ogists, that non-egalitarian societies evolve as regulating or equilibrating systems. Essentially, this argument sees the evolution of cultural complexity as a deviation counteracting process. While the functiona1-eco1ogica1 argument has been 30 modified and refined over the years (see Orans 1975), many recent archaeological attempts to explain the evolution of status inequalities still rely on the deviation counteracting concept. For example, in his model of the evolution of cultural complexity In the Gulf of Georgia region, Bur ley (1979, 1980, 1983) postulates that a migration of salmon fishing specialists from the Hope/Yale locality at the end of the Baldwin Phase (ca. 2500 B.P.) was directly responsible for the rise of social ranking in the Marpole culture. Here, we see the deviation counteracting approach. Bur ley (1979:136) describes the demographic condition of the pre-Marpole culture (Locarno Beach culture) as being in "established density equilibrium", and as being a "closed system" (Bur ley 1979:140). Population influx from the lower Fraser Canyon upset this equilibrium, creating demographic stress. Social ranking, economically based on the newly introduced salmon specialization, was the mechanism adopted to restore the system to a state of equilibrium. At the risk of ovei—simplifying, Bur 1ey argues that a system, originally in a state of equilibrium, is subsequently forced out of equilibrium by some process (in this case, population influx). A new trait is then adopted (social ranking) which functions to restore equilibrium. It is implicit in Bur ley's argument that social ranking is need-serving and that ranking will not be adopted until the appropriate need arises. But how do we know that the proposed population influx from the canyon, if indeed it did occur, represented the 31 appropriate need? Further, is social ranking the only response to a "disequi1ibrating" process such as population influx? Is there no less costly response? In another recent archaeological example from the Northwest Coast, Ames (1979b, 1981, 1983) has considered ranking in terms of environmental monitoring and information flow within a proposed framework of stable and resilient systems. Ames (1983:183) states that social hierarchies will evolve "whenever circumstances constrain the capacity of the system to adapt internally", and "strategies which maintain internal values in the face of external changes will be positively selected. Ranking is such a strategy". Here again it is argued that the system in its present form cannot maintain internal stability because of external pressures. Restructuring is required and hierarchical restructuring is selected. Ames (1981:792) states that "ranking acts as a homeostatic mechanism maintaining the equilibria of systems". Again, at the risk of over-simplifying, ranking Is presented as a need-serving or deviation counteracting response. The efficiency of hierarchies at processing information in state level societies has been well demonstrated (e.g., Wright and Johnson 1975). But in these bureaucracies there is a real need for information processing because the hierarchical elite control vast territories and oversee the activities of large populations dispersed in many communities. To what extent are hierarchies more efficient than egalitarian 32 groups at processing Information, when group size rarely exceeds one hundred people, and territories are controlled by household groups? The basic premise of the deviation counteracting argument for social ranking is that hierarchies function to counteract subsistence related stress. In light of the above review, there is reason to question this premise. Did the ethnographic societies of the Northwest Coast counteract resource stress? One of the clearest explications of the deviation counteracting argument is that of Suttles (1960, 1962, 1968), later expanded by Vayda (1961), Piddocke (1965) and Weinburg (1965). The core of the argument is that the prestige system, which included the rank structure, was adaptive; it facilitated the adjustment of the population as a whole to its environment. The function of the system was to maintain high levels of food production and to equalize food consumption (Suttles 1960:304). Balancing resources was important because, according to Piddocke (1965:245), "scarcity of food was an ever-present threat... and without the distribution of food from wealthier local groups to poorer ones, the latter would often have died of hunger". The Sutt1es/Vayda/Piddocke model has been reviewed in detail on more than one occasion (see Drucker and Heizer 1967; Rosman and Rubel 1971; Ruyle 1973; Orans 1975), so my comments here will be brief. Suttles et a_L argued that while great surplus abundance may have characterized 33 Northwest Coast economy In general, this picture was less than true for individual local groups who occasionally faced food shortages due to fluctuations In resource productivity. Support for this claim 1s drawn from relevant ethnographic sources, particularly Boas (1921, 1935), in which accounts of starving people are given. Suttles (1960:298-301) argues that the principle means by which unbalanced food resources were equalized was through food-for-wea1th exchange, which preceeded pot latching. In this exchange, a man would bring food usually to an affinally related house. The latter was obliged to "thank" the donor with articles of wealth. As Suttles (1960:303) states, this type of exchange could lead to accumulation of wealth, since local inequities in resource productivity could be transformed into inequities in wealth. Less productive communities would be at a disadvantage, as they would soon be "unable to give back wealth in exchange for ...food" (Suttles 1960:303). Suttles argues that this potential inequity was regulated by the pot latch. At a pot latch, the host validated his claim to status by giving away wealth to guests. For Suttles, the function of the potlarch was the distribution of wealth, which "restored the purchasing power" of other communities. A critical question with respect to the food-for-wea1th exchange is whether the food-receiving household was actually in need of food. Piddocke, Vayda and weinburg all argue that the food-receiving households were being saved from 34 hunger, even possible starvation. In reality, this seems to have rarely been the case. Suttles (1960:298) states, " a man could at any time take food to a co-parent-in-law and expect to receive wealth In return" (emphasis mine). In other words, the exchange had little to do with starvation or temporary scarcity. In fact, one might expect the donor to take food to his affine during "good times", not "bad times", when the affine would likely have the most amount of wealth on hand. The food-receiving household typically consumed the food by holding a feast with other members of their community (Suttles 1960:298). The food received in the exchange was generally consumed in one sitting. This seems a very strange thing for "starving" people to do. If the food-receiving household was Indeed facing scarcity, why would they not initiate the exchange? Suttles (1960) is very explicit that this is a food-foi—wealth exchange, but it should be a wealth-foi—food exchange. I think it is unlikely that people who are hungry, and who also have wealth, would wait for food to be brought to them. Others (Drucker and Heizer 1967; Ruyle 1973; Orans 1975) have noted that individuals in wealthy, powerful households, capable of feasting or pot latching, would probably be affinally related to individuals in other wealthy, powerful households, given the preference for marriages between people of roughly equal status (Suttles 1960:297; Jorgensen 1980:164-68). Similarly, affinal 35 relationships would exist between poorer, less fortunate households. Thus, exchange and pot latching assured that resources circulated only among high status households, those very households that had the least to gain In terms of subsistence from the acquisition of food or wealth. Those households that could have benefitted in terms of subsistence from exchange or pot latching had no access to it. This much is admitted by Suttles (1973:622) who states, "the separate lower class settlements probably suffered most precisely because they had fewer extra-vi11 age affines", and we might add "poorer" extra-viI 1 age affines. Examples of shortages and starvation presented by Suttles (1968:59, 1973:622) and Piddocke (1965:247), which in itself seems to contradict the food-equa1izing hypothesis, may not be overstated, Drucker and Heizer to the contrary. The important point is that some households faced privation far more often than others, and these people, because of their lack of surplus food and wealth, would not have benefitted from the pot latch or food-for-wea1th exchange. Thus, we have the apparently contradictory situation in which pot latching and feasting are characterized by gormandizing and excess in the name of individual or family prestige, while households, and even entire communities of lower status often faced the prospect of food scarcity. Such disparity would not (could not) exist in an egalitarian society where access to resources is, by definition, free and equal. It is only in non-egalitarian societies that 36 surplus (of food and wealth) and starvation can co-exist. There is good reason to argue that the prestige system did not function to counteract or balance resource stress, but rather, amplified It, by concentrating food, wealth and prestige in the hands of the status elite. The purpose of the preceding discussion has been to demonstrate that there is room for re-interpretation of the functiona1-eco1ogica1 model of ranked societies as regulating or equilibrating systems. On the contrary, it may be argued that the pre-contact socioeconomic system of the Northwest Coast functioned in such a way that the rich got richer while the poor got poorer. Rather than re-distributing resources, the system concentrated resources (food and wealth) in the hands of powerful households. This cannot be seen as a homeostatic process. Social Ranking as Deviation Amplification:  the Northwest Coast The Initial Cond i 11on Let us now consider social ranking in terms of the deviation amplifying model. On the Northwest Coast, most archaeologists trace the origin of the development of "classic" Northwest Coast culture to about 5000 B.P. Fladmark (1982a:110) refers to the last 5000 years as the "Developmental Stage" in Northwest Coast prehistory. The initial economic condition at this time was one of resource 37 diversity. A variety of resources were exploited, including salmon, but as Mitchell (1971:57-58), Bur ley (1979:136), and Matson (1981:83) have argued for the south coast, there is little evidence for intensive salmon production until about 2500 B.P. This date also seems to hold for the north coast (see Chapter 4). The elements of salmon ecology (abundant, predictable, but localized) were critical to the ensuing developments of cultural complexity on the Northwest Coast. As Matson (1981, 1983) argued in his intensification model, salmon abundance made it possible to harvest a surplus, and reliability made control of the resource economical because of its consistent return. Since the resource is predictable, technology can be developed to utilize it fully (Matson 1983:138). It is also important that access to the resource is restricted, so that control of a particular access location would be worthwhile because it would ensure access year after year. Access to the river or to salt water did not necessarily ensure good access to the fishery. There were only certain discrete access points at which fish could be taken, given the available procurement technology. Moreover, al1 locations were not equally productive, as Suttles, Vayda and Piddocke have shown. Offshore reefs, estuarine zones, and river narrows or canyons were areas of high productivity. Other areas, such as heads of inlets or river sections lacking shallows or back eddies, would have had relatively low productivity. Other factors being 38 equal, salmon procurement locations near the beginning of the run are more desirable than locations near the end of the run for two reasons. First, salmon stop feeding entirely when they re-enter fresh water, and thus lose caloric value during upriver migration (Idler and Clemens 1959). Initially, this involves the loss of oil and fat content, which is not critical to the human consumer. In fact, this loss may be beneficial because preservation by smoking or drying is facilitated when the oil and fat are removed. Ultimately, however, the nutritional value of the meat is depleted as the salmon continue to expend energy in the upriver migration. Thus, locations near the beginning of the runs, in salt water or the lower river sections, would be more productive once the necessary technology for preservation was in place, since fewer fish would have to be taken to achieve a given caloric level (Kew 1976:8). Second, peak density occurs as the salmon enter fresh water, but this changes as breeding populations branch off into tributary spawning streams. The size of the run steadily decreases as the run moves upriver. Therefore, to exploit the maximum abundance of the runs, the most important procurement locations are those at the entrance or on the lower course of the main river in the system. The size of the runs also varies for different rivers. On smaller rivers, even ideal locations did not always ensure good catches. This variability among locations was important in terms of human demography and social structure. In an 39 example from the Interior, Sneed (1971) has shown that salmon abundance and human population size were strongly correlated. On the central coast, Donald and Mitchell (1975) demonstrated that salmon abundance was integrally related to population size and inter-group prestige. The Initial Kick Why become salmon specialists in the first place? Why would a population with a diversified economy change their strategy to resource specialization? Christenson (1980) argued that this change may occur as a result of changes in resource diversity and food niche width. Resource diversity refers to the number of different resources consumed, regardless of their proportional contribution. Niche width refers to the differing proportions or "eveness" in which food resources are consumed (Christenson 1980:34). According to Christenson (1980:37), low density populations will typically have low resource diversity and low niche width. Resource selection will be a function of two variables: cost of procurement and maximum yield. Low cost resources with high yield will be selected first, and higher cost or lower yield resources will be added on in an agg1omerative manner. The addition of high cost resources will usually be associated with some form of technological change. In addition, there might be some experimentation with improving the yield of certain resources, either through chance genetic changes (Flannery 1968:81-82), or by improving 40 growing conditions (e.g., adoption of irrigated rice paddies in Indonesia (Geertz 1963:29-31)). If intensification is not possible (or not successful) resource diversity, niche width and human population size will increase to a maximum point and then stop (see Figure 2.1a). If intensification is possible, resource diversity and niche width will increase to a certain point at which labour efficiency will begin to favour the intensification of one or more resources. At this point, niche width will decline as the non-intensified resources are exploited less often. Resource diversity will level off or perhaps even decline (see hatched line, Figure 2.1b) as high cost, low yield resources that conflict with the intensified resource are dropped altogether. Human population will continue to grow, however, as long as intensification improves resource yield. Christenson's model was aimed at explaining the evolution of agriculture in areas such as the Tehuacan Valley and the mid-western United States (see Christenson 1980:39-52), but it also has applicability to the intensification of salmon on the Northwest Coast. For example, in the Gulf of Georgia area, land mammal hunting predominated in the Old Cordilleran Culture until about 5000 years ago (see Matson 1981:80). Resource diversity and niche width were low at this time. In the ensuing St. Mungo Phase, resource diversity and niche width increased, with the most notable increase seen in the exploitation of shellfish and fish. Matson (1981:81) states that bay 41 4> > o o <n high population low populotion niche width intensification not possible high population HI > a> o o tn a> intensification begins low population niche width intensification possible Figure 2.1 Schematic of" change in resource diversity and niche width without and with intensification (after Christenson 1980). 42 mussel and salmon may have been the most important single resources at this time. Seal hunting also increased in importance during the St Mungo Phase, and it is likely that maximum resource diversity and niche width were reached by the St. Mungo/Locarno Beach interface (R.G. Matson, pers. comm., March 1985). The increase in shellfish exploitation, beginning about 4300 B.P. (Matson 1981:83), was particularly important. Shellfish are a low cost, low yield resource. They were not completely ignored during the Old Cordilleran Culture at Glenrose, but they were not intensively exploited (see Matson 1981:80). Recently, Croes and Hackenberger (1984:28) cogently argued that in pre-storing economies on the Northwest Coast, the limiting factor on population size was the winter resource base. Shellfish are the only coastal resource that is abundant in winter, and can be intensively exploited throughout the year. The increase in shellfish exploitation (ca. 4000 B.P.) seen at Glenrose and other sites is suggestive of an increase in coastal population at this time. Croes and Hackenberger (1984:37) even suggest that exponential population growth may have resulted in the over-exploitation of shellfish resources between 4000 and 3000 years ago. Even if the point of shellfish over-exploitation was not reached - and it seems unlikely that this happened all over the Northwest Coast - the intensive exploitation of a low yield resource such as shellfish as the primary means of supporting a large 43 population through winter would ultimately favour the adoption of a new economic strategy. It is at this point that intensification of a resource such as salmon may be adopted, including preservation and storage for winter consumpt i on. Intensification and Resource Control What are the correlates of developing resource specialization? One possibility is increasing control of specific resource extraction locations. This lies at the intersection of two variables: procurement technology and resource access (see Figure 2.2). If the procurement technology requires mobility and access to the resource is not restricted by the number of extraction locations available for use, then control of a specific location is unlikely (Figure 2.2, cell A). An example is seen in the aboriginal fishery of the middle Fraser drainage, including upper Fraser Canyon and the long series of rapids and canyons above it. The area, occupied by the Thompson, Li11ooet, and Shuswap, extends from Yale to Soda Creek (Kew 1976:12). Here, dip-netting was the main procurement technology. Dip-netting is a mobile technology that can be practised by a single individual. The necessary equipment is minimal and portable, and this allows the fisherman to move up and down the river from one fishing location to another. The best dip net locations were back eddies and areas of slack water along the river shoreline, where water turbidity was 44 Mobile TECHNOLOGY Fi xed A B resource control is not promoted C D resource control is promoted Unrestri cted Restricted Resource Access Figure 2.2 Variables states related to the control resource locations. 45 high, so that the salmon could not see the nets (Kew 1976:12). Dip nets were operated from natural stands such as rock ledges or boulders, or from wooden platforms. These finite locations along the middle Fraser exist in sufficient number so that dependable access to the resource was not a problem. Some of the more productive locations may have been "owned" by individuals (especially among the Lillooet - see Kennedy and Bouchard 1978:40), but it is not known to what extent this was practised prior to white contact. Not all middle Fraser groups recognized ownership, and it is apparent that many fishing locations in this region were open to common access. One could be assured of returning to the river year after year to good fishing locations, if not always the same ones. If access to the resource is restricted, in terms of a limited number of extraction locations, then control of specific locations will be important, and adoption of fixed procurement technologies will be favoured. Fixed technologies or permanent "facilities" (see Oswalt 1976:105-07) often require considerable labour efforts in construction and maintenance, but usually the high initial cost is offset by the yield. Accordingly, groups involved in the construction of fixed technologies will have a vested interest in their continual use. Cove (1982:4) argued that the use of fixed technologies such as weirs and traps by the Gitksan was an important factor in the establishment of lineage ownership of these locations. Ultimately, control 46 of fishing locations can form the basis of more complex notions of territoriality (see Dyson-Hudson and Smith 1978), with a spawning stream or river section as the territorial locus. Further, Cove (1982:4) argued that the combination of fixed technologies and restricted access led to the adoption of residential permanency by the Gitksan. The implication here is that sedentary settlement is one way of controlling access to important resource locations. Another likely correlate of increasing resource * specialization is change in domestic organization as a result of change in the type of production. Wilk and Rathje (1982:62Z-23) posit two basic types of production: linear and simultaneous. Linear production involves a series of activities that are scheduled to occur consecutively rather than at the same time. Among hunter-gatherers, linear production is typically associated with residential mobility. People move from one location to the next as resources in those locations become available. Although large households and base camps may occur on a temporary seasonal basis, linear production usually promotes small population aggregates for two reasons. First, resources tend to be dispersed rather than localized, which favours small dispersed social groups. Second, resources are often unreliable. Abundance may fluctuate from year to year, and small mobile groups are an adaptation to this uncertai nty. 47 Simultaneous production involves a number of people acting at the same time (Wilk and Rathje 1982:622). Large nucleated social groups are likely to develop when simultaneous tasks must be performed (see Pasternack, Ember and Ember 1976) if resources are localized and reliable. These conditions generally prevail for agriculturalists; rarely for huntei—gatherers. An exception is the Northwest Coast, where salmon meets the conditions of a localized and reliable resource. Here, salmon specialization is an example of simultaneous production. Coast Salish reef-netting is a case in point (Suttles 1951:161-62). Boat captains, fishermen and fish processors all came together for brief periods during the salmon runs to engage in intensive and complex simultaneous 1abour. On the northern Northwest Coast, task simultaneity was even more pronounced. The availability of salmon and other resources is attenuated here compared to the south coast because of environmental constraints (see Schalk 1977). Salmon runs in northern rivers are of shorter duration than the Fraser and Columbia River runs, and intensive simultaneous labour is required to extract the maximum amount of the resource. Places like Kitselas Canyon, the first canyon of the Skeena River, thus became probable localities for simultaneous production. Permanent households would be promoted here because they facilitated the need for complex divisions of labour 48 and more efficient completion of simultaneous tasks. In addition, Kitselas Canyon is only 2 km long. Once a prehistoric village was established, it would be unlikely to give up such a strategic location. Permanent occupation became a means of controlling the resource, and simultaneous production facilitated permanent residence. Returning to the deviation amplifying model, a critical level has now been reached. By engaging in simultaneous production and controlling resources at specific locations through permanent residence, social groups effectively impose greater access restrictions by limiting access to resources only to group members. In this sense, the cultural response to the initial condition of environmentally restricted access is to impose further access restrictions by bringing critical resource locations under the control of the resident group. In turn, this can create a positive feedback chain. As favourable locations come under control, it becomes important for other groups to establish control over remaining locations, further amplifying access restrictions. The Formation of Egalitarian Corporate Groups How may these resource control 1 ing groups be characterized? Are they in fact corporate groups? Recently, Hayden and Cannon (1982) considered the corporate group as an archaeological unit of analysis. They outlined four essential characteristics: 1) closed or restricted membership, 2) residential coherency, 3) internal 49 hierarchical structure, 4) economic pressure as the underlying cause of corporate group formation. It is important to note here that these criteria were aimed at identifying prehistoric corporate groups in the archaeological record. Not all are defining characteristics. For example, residential coherency can include substantial variability. Hayden and Cannon (1981:141-47) indicate that corporate group members need not 1ive in the same dwe11ing or even in the same community. They may be widely dispersed in a number of communities. With respect to hierarchical structure, Hayden and Cannon (1981:148) state, "there is always some sort of administrative or authoritative hierarchy which directs major decisions concerning the corporate group". Here, it is important to know how roles of authority are defined, and what decision-making powers a leader or leaders will have. In some societies, such as chiefdoms, corporate groups may exist in which the authoritative elite have absolute decision-making power over life, death and the redistribution of wealth. In other cases, corporate groups may exist in which ultimate decision-making powers are entrusted to the mutual agreement of all group members. Here, leaders may exist, but only in an advisory capacity or as managers of wealth and resources. An example of this type of leadership is the Melanesian "big man" (Sahlins 1963:289). Such a case is likely to develop when the corporate group is small and localized. This obviates the need for a complex system 50 of centralized information-processing, which some authors have suggested as a cause of social stratification (e.g., Wright and Johnson 1975). Indeed, most other definitions of corporate groups cited by Hayden and Cannon (1982:133-35) stress the importance of group solidarity and control or ownership of resources by the group rather than by a centralized authority figure. For example, Honigmann (1959:360) listed a number of criteria as definitive of corporate groups including enduring, selective, stable membership, conferring specific rights and duties, group ownership of wealth (emphasis mine), clear group identity and leadership. Weber (1947:45) defined a corporate group as "a social relationship which is either closed or limits the admission of outsiders by rules". Nadel (1951:160) stressed the sharing of rights and duties as definitive of corporate groups. Goodenough (1959:30-3 1) defined corporate groups as "groups that function as individuals in relation to property". Returning to Hayden and Cannon's definition, "some sort of hierarchy" is a vague concept. In a sense, all societies have some sort of hierarchy, even egalitarian societies. It is interesting that Hayden and Cannon (1982:142-43) suggest the Iroquois, who were stridently egalitarian (see Trigger 1969:85), as an example of a society with corporate group structure. In considering the above definitions, two characteristics stand out: 1) closed, stable membership, 51 2) mutual desire to achieve a common purpose. Residential coherency and centralized authority may be two ways of increasing the efficiency of the corporate group (i.e. achievement of a common purpose), but according to the above definitions, neither characteristic is definitive or even critical to the existence of a corporate group. It is apparent that corporate groups may be organized along egalitarian lines. To clarify this concept, we may draw a comparison to Matson's (1983) model of the evolution of cultural complexity on the Northwest Coast. The core of Matson's model, "is that ranked society, sedentariness, and large scale use of salmon resources should all be tied tightly together and that, initially, one should not occur without the other two". In the model developed in this study, it is argued that sedentism (related to resource control) and salmon intensification can occur in the absence of ranked society. In effect, social ranking is replaced in this triumvirate by egalitarian corporate groups. An abundant, reliable, localized resource (salmon) makes complex simultaneous production possible. This promotes sedentism for two reasons. First, complex simultaneous production requires a steady, dependable labour force. Secondly, permanent residence promotes ownership or control of critical resource locations (i.e., where groups may engage in complex simultaneous production). Under these conditions, egalitarian corporate groups will form if all 52 members of the resident group have equal rights of access to' the resource. As resource specialists, egalitarian corporate groups would have engaged co-operatively in complex simultaneous production and controlled specific resource locations in such a way that unrestricted access to the resource was available to group members (those who co-resided permanently), while access to non-members (non-residents) would be severely curtailed or even denied, as a result of the limited number of extractive locations, and the establishment of usufruct over each location by the existing corporate group. It does not seem unreasonable that egalitarian corporate groups could exercise control over critical resource locations because there is no reason to expect that outside groups would be any larger or better prepared to wrest control for themselves. If anything, it may be expected that outside groups would be smaller and more loosely organ i zed. This concept of group ownership of resources is not at all inconsistent with ethnographic evidence from the northern Northwest Coast. The Tsimshian, Haida and Tlingit had a we 11-deve1 oped concept of resource ownership, and it is generally argued that ownership was vested in the corporate lineage (Cove 1982:4 refers to this group as the "House"), with the lineage or "House chief" functioning as a "steward" (Drucker 1939:59) or "manager" (Richardson 53 1982:97-99) of resources. In Cove's (1982) analysts of Gitksan land ownership, it is clear that title to resources belonged to the House. The House chief gained (or lost) prestige on his ability to manage the resources of the House, and to represent the House at ceremonial gatherings. The concept of group ownership has been contested by some (e.g., Ruyle 1973) who argue that at the time of contact an incipient form of social stratification had developed on the Northwest Coast, with chiefs having clear title to resources. If this was the case, might we not expect the emergence of "village" or "regional" chiefs, commensurate with fully developed chiefdoms? If ownership was vested in the individual, rather than the House, what would prevent a chief from extending his power and authority beyond his own kin group? Yet, in a study of Hudson's Bay Company post journals from Fort Simpson, Mitchell (1983a:64) indicated that Tsimshian chiefs at Metlakatla held no real authority beyond their own group, "and even there his hold seems fragile". Hal pin (pers. comm.. May 1985) stated that although chiefs enjoyed certain rights and privileges that tended to set them apart from others in their House, they were not above being replaced by their followers if their work was deemed unsatisfactory. The ethnographic concept of group ownership of resources on the northern Northwest Coast may be a fundamental holdover from an earlier time, when corporate groups were organized along egalitarian lines. Egalitarian corporate groups: some ethnographic examples To further elucidate the concept of egalitarian corporate groups, some specific ethnographic examples are offered here. One example is the Hopi. Although transient leadership may exist within Hopi villages, they conform to Fried's definition of an egalitarian society. The Hopi household, the smallest unit of society, is an example of an egalitarian corporate group. Households may consist of numerous nuclear families organized in adjacent or quite separate dwelling units (Titiev 1944:51; Connelly 1979:545). Connelly (1979:549) identifies these as "management groups" that function as cooperative work units. This organization "provided a stabilizing effect, making possible production for survival" (Connelly 1979:549). On a larger scale, Hopi villages also function as egalitarian corporate groups. These are permanent, co-operative units for defense and production. "Each village is autonomous and has its own land" (Kennard 1979:554). The Huron provide another example of egalitarian corporate groups. There is some debate about whether the Huron were truly egalitarian at the time of European contact. Clan segments had "chiefs", but it is likely that these individuals were better defined as "big men". Civil chiefs were chosen from a number of qualified candidates (Heidenreich 1978:371), which suggests that their status was achieved rather than ascribed. There is no indication that Huron "chiefs" had unequal access to wealth or prestige 55 through redistributfve exchange. Hoarding or control of resources by an individual met with strong disapproval, and could lead to expulsion from the village (Trigger 1969:40-41). Undoubtedly, status differences existed within Huron society, but the above evidence is consistent with Fried's definition of an egalitarian society. Corporate identity existed among the Huron on many levels. The most basic corporate unit was the "fireside" group, consisting of two families that shared a hearth within a dwelling (Fenton 1978:303). This was the day-to-day producing unit. Huron longhouses were commensurate with the extended family. In terms of corporate identity, Heidenreich (1978:377) states: the longhouse represented the physical manifestation of their social and economic system. It was at this level... that the Huron values of family solidarity, economic co-operation, and rule by the mutual agreement of adults found their basic expression. On an even larger scale, the Huron village represented a corporate unit for defense, and was often strongly fortified. These two examples indicate the nature and extent to which corporate identity may be expressed in egalitarian soc i et i es. 56 The Transition from Egal1tar1an to Ranked Corporate Groups The next step In the model Is the restriction of access to resources to members within the group. This marks the transition from egalitarian to non-egalitar Ian or ranked corporate groups. Egalitarian corporate groups are unlikely to persist if two conditions exist: 1) variability exists in the amount that can be produced from one location or territory to another. 2) competition develops within the corporate group for household access to specific locations within the controlled territory. Egalitarian societies engage in "practical" production, intended to meet subsistence needs. Since there is no central hierarchy, there is no individual control of wealth, and no effort to produce a surplus beyond perceived subsistence needs ("social" production). In many cases, egalitarian social organization will be maintained because there is an upper limit to production. For example, on the Northwest Coast shellfish harvesting could not be the basis of surplus production because of the danger of over-exploitation and depletion of the resource (see Croes and Hackenberger 1984). The situation with salmon was clearly different. Given aboriginal technology, there was no upper limit to the amount of salmon that could be caught, provided one had access to a productive location. Moreover, a larger, hierarchically organized labour force could improve the efficiency of 57 complex simultaneous production. This might mean the incorporation of new complex techniques of extraction and processing, or the use of a variety of different techniques in different locations at the same time. It might also mean that some group members functioned primarily as subsistence producers, while others specialized in craft production, trade or warfare. The question still remains, however; why would egalitarian corporate groups re-organize hierarchically? The answer is not because hierarchies function to balance fluctuating resources. Rather, the answer has to do with a fundamental change in the transmission of rights to property, which results from intra-group competition among households for continued rights of access to prime extraction 1ocat i ons. As long as numerous extractive locations exist, the rights to use these locations are transmitted by village residence (Murdock 1949:80-81; Goody 1972:121). Inheritence is likely to be partible (divided among heirs) (Goody 1972), or new families may simply leave the village to establish use of new locations in other territories. This latter example of fissioning may be one way in which non-1 oca 1ized clans develop. As more and more extractive locations come under corporate control, the transmission of rights of access shifts .from, the village group to the 58 household group (Goody 1972; see also Wilk and Rathje 19"82:627-28) . For example, in Luzon, where land for swidden farming is plentiful, it is managed by village patri1ineages but rice paddies, which are at a premium, are owned and managed by households (Drucker 1977:9). This fundamental change in the transmission of rights to resources is stated by Wilk and Rathje (1982:627): "As property becomes more difficult to gain access to, the group that controls the resources becomes more and more strictly defined, eventually being reduced to the household itself". At this point the emergence of a leader or chief may become important. Questions regarding household membership can be settled by this individual, and more important inheritance of resources will tend to be reckoned through his line as a means of ensuring that ownership of resources continues to be vested in the same household from one generation to the next. In a sense, the chief functions as a hereditary "yardstick". Rather than partible inheritance, the corporate group is now practising impartibl inheritance (Goody 1972). This fundamental change in the transmission of rights to resources is not intended to balance or even-out a fluctuating resource base, but rather it is a strategy for ensuring control of critical resource locations by small, vested interest groups. For his part, the emerging chief is now in an advantageous position to encourage a change from "practical" 59 production to "social" production within his corporate group. Since there is disparity in productive capacity from one resource territory to another, some households will be able to out-produce others. This can result in an increase in the size of the most competitive households (see Donald and Mitchell 1975) or in the expansion of household control to include new territories, or both. The surplus production resulting from this expansion is controlled by the chief, and he uses it to finance his political activities. Earle (1977:217) argued that the complex redistributive system of the Hawaiian chiefdoms at the time of contact functioned in this capacity. He demonstrated that, contrary to Service's (1962) model, the small, localized socioeconomic units in Hawaii, the ahupua'a, were not resource specialists dependent upon regional redistribution, but were instead essentially self-sufficient. The redistributive system, based on social production, existed to channel the flow of wealth into the hands of a few elite individuals within society. I think that a similar situation existed prehistorica11y on the Northwest Coast. I suggest that as the forerunner to ranked society, the egalitarian corporate group formed the basic socioeconomic unit. These groups would have controlled critical resource locations such as Kitselas Canyon, where simultaneous production was possible. As more and more resource locations came under corporate control, it became increasingly important that access to 60 resources be more rigidly defined (as postulated by Goody 1972). A shift from partible to impartible inheritence would ensure household rights of access to specific locations from one generation to the next. For his part, the emerging leader could encourage a shift from practical to social production through expansion and hierarchical re-organization of corporate labour. The surplus could be transformed into wealth for feasting and pot latching. It is important to recall that among the northern coastal groups, prestige was conferred on the House, not the individual (see Cove 1982), so that in theory, all members of the corporate group would benefit, although in practice, it seems clear that the House chief was the greatest beneficiary. There is no shortage of evidence that Northwest Coast chiefs had differential access to wealth accrued through social production. For example. Boas (1921:1333-34) states that a Kwakiutl numaym chief could appropriate a certain portion of a hunter's catch. Meares (1790:258) states that among the Nootka, whaling and sea otter hunting were limited to chiefs (perhaps because only chiefs could organize the necessary labour force). Among the Coast Tsimshian, Mitchell (1983a:60) states that the Glspaxloats chief. Legale, gained control of Skeena River trade, during or shortly after the establishment of Fort Simpson. Over a 13 year period, after 1836, the Hudson's Bay Company journals record 32 trading excursions, with Legaic personally acting as trader on at 61 least 15 expeditions. The third and most critical level in the deviation amplifying model has now been reached. Not only does restricted access to resources exist on the intei—group level (level 2; see Fig 2.3), but there is also differential access within groups. House chiefs controlled direct access to wealth through control of trade and pot latching. In addition, chiefs took an active role in activities related to increasing household (i.e. labour) size. Slavery and the establishment of marriage alliances were means by which household size could be increased, and this in turn could increase the amount of surplus produced, and the potential for conversion to wealth. Warfare was another important means of direct acquisition of wealth from rival groups, and may also have been important in terms of competition between groups for control over critical resource locations (see Donald and Mitchell 1975:341-42; Ferguson 1983; Mitchell 1984). Thus, there were numerous ways in which wealth and prestige could be increased, and while low status or non-titled household members may have benefitted vicariously from their association with a powerful chief, it was clearly the chief who controlled access to resources within the household, and benefitted most from the conversion of surplus into wealth. 62 ABUNDANT PREDICTABLE RESOURCE ENVIRONMENTALLY LOCALIZED ACCESS First level of access restriction DEVELOPMENT OF NEW TECHNOLOGIES FOR RESOURCE PROCUREMENT DESIRE TO ENSURE ACCESS TO RESOURCE / SEDENTISM. RESOURCE t SPECIALIZATION PRESERVATION AND STORAGE SMALL-SCALE GROUP CO-OPERATION FOR PRODUCTION I MULTI-FAMILY CO-OPERATIVE HOUSING SMALL STABLE EGALITARIAN CORPORATE GROUPS. CONTROLLING RESOURCE ACCESS Second level of access restriction .EMERGENCE OF LEADER FROM WITHIN GROUP TO PROMOTE INCREASED PRODUCTION DEVELOPMENT OF INTERNAL GROUP HIERARCHY - LEADER OVERSEES PRODUCTION COMPETITION FOR MANPOWER TO INCREASE PRODUCTION / PRODUCTION OF SURPLUS"-\ CONVERSION OF SURPLUS TO WEALTH TRADE POTLATCH SLAVERY WARFARE LEADER CONTROLS SURPLUS AND WEALTH - PRODUCTION OF SURPLUS AND CONVERSION TO WEALTH FUNCTIONS TO «-FINANCE SOCIOPOLITICAL ACTIVITIES OF ELITE MARRIAGE ALLIANCES Third level of access restriction Figure 2.3 Deviation amplifying model of the evolution of status inequality: the Northwest Coast examp1e. 63 Summary The deviation amplifying model of the evolution of status inequality, based on increasingly restricted access to resources, is summarized in Figure 2.3. Initially, access to critical resources is restricted by spatial and temporal environmental constraints. The second level of access restriction is cultural - the formation of egalitarian corporate groups, who control productive resource locations, denying access to outsiders. The third level of access restriction is also cultural - the transformation of egalitarian corporate groups into ranked corporate groups. At this level, access to resources is not only denied to outsiders, but there is also unequal access within the group. This formed the basis of ascribed status inequality on the Northwest Coast. In conclusion, it is difficult to perceive of social ranking as part of a deviation counteracting system on the one hand (cf. Bur ley 1979; Ames 1981), while it creates differential access to resources on the other hand. In the model presented here, it Is argued that social ranking evolves as a process of amplifying, rather than counteracting access restrictions. In addition, I have argued that a critical sub-stage in the evolution of ascribed status inequality for some societies is the formation of corporate groups, initially organized along egalitarian lines. On the northern Northwest Coast, these were likely small, household and village groups, exhibiting some degree 64 of sedentism at or near a key resource location, specializing In the production of that resource, and capable of meeting subsistence requirements on a co-operative basis. This Is an essentially stable form of socioeconomic organization i n a number of soc i et1es (e.g., Iroquo is; Tr i gger 1969:85). On the northern Northwest Coast however, egalitarian corporate groups were eventually replaced by internally ranked corporate groups. The reason for this hierarchical re-structuring was to further consolidate control of territories and expand and re-organize the existing labour group to create surplus production, which in turn served to finance the sociopolitical activities of an emerging e 1 i te. This model of the development of cultural complexity on the Northwest Coast will now be tested against the archaeological data from Kitselas Canyon and the lower Skeena Ri ver. 65 Chapter 3 KITSELAS CANYON; THE NATURAL ENVIRONMENT In the preceding chapter, It was argued that the Initial condition for the formation of corporate groups was the presence of an abundant, reliable, localized resource. The chapter examines the natural environment of Kitselas Canyon. It is demonstrated that the canyon meets the initial condition. In addition, it is shown that that the unique location of Kitselas Canyon provides ready access to a wide variety of other resources in numerous biogeoclimatic zones. This further enhances Kitselas Canyon as a prime potential location for the formation of corporate groups. General Geographic Setting Kitselas Canyon is located on the Skeena River, 16 km east of Terrace, British Columbia, and approximately 150 km inland from the mouth of the Skeena. The canyon is situated at 540 36' north latitude, and 1280 25' west longitude. It is located within the Kitselas Indian Reserve Number 1. Kitselas Canyon lies within the Canadian Cordillera, one of the major physiographic divisions of Canada. The Canadian Cordillera of British Columbia has been divided by Holland (1964:27) into three major sub-divisions or "systems", including the Western, Interior and Eastern systems. These systems, oriented along a northwest to southeast alignment, conform to the Coast Mountain, 66 Intermontane and Rocky Mountain regions respectively. Kitselas Canyon is situated roughly on the boundary between the Coast Mountains and the Intermontane system (Figure 3.1). It is located in the extreme western flanks of the Hazel ton Mountains of the Interior system, and is bordered on the west by the Kitimat Ranges of the Coast Mounta i ns. The Kitimat Ranges (Holland 1964:4 1) comprise mainly granitic rocks and crystalline gneisses (Farley 1979:28), and extend from the Nass Range southward to the Bel la Coola River. Peaks range in elevation from 1200 to 3000 metres, and the mountains are crossed by major rivers in valleys that are less than 100 metres above sea level, creating maximum relief of over 2500 metres. The higher peaks and ridges are sharp-crested, and have cirque glaciers and permanent snowfields. Streams and rivers are deeply cut into canyon-like gorges (Duffel 1 and Souther 1964:7). The Hazel ton Mountains of the Interior System are quite distinct from the Coast Mountains. Comprising mainly folded and faulted sedimentary strata, the Hazel ton Mountains are much less steep than their coastal counterparts. Peaks are serrate to rounded, and the higher peaks have cirque g1ac i ers. Two interesting physiographic features, in addition to Kitselas Canyon, are located between the Kitimat Range and the Hazel ton Mountains. A surficial lava plain is located in the Nass Basin, south of the Nass River at Canyon 67 Figure 3.1 Skeena drainage of northwestern British Columbia. 68 Cfty. The source of the lava fs in the Tseax River in the Nass Ranges (Holland 1964:58). Perhaps the most spectacular physical feature of the area is the broad drainage valley or trough that extends from Kitimat Arm north across the Skeena River at Terrace, up the Kitsumkalum and Cedar River, ultimately meeting with the Nass River at Canyon City. The valley floor is unusually wide and flat for the area, ranging from 3 to 9 km in width. South of Terrace, the valley drains into the Skeena system via Lakelse Lake and Lakelse River. North of the Skeena, the valley roughly divides the Coastal Mountain System from the Intermontane System. South of the Skeena, this division runs for some distance along Zymoetz River, a tributary of the Skeena, just below Kitselas Canyon (Duffel 1 and Souther 1964:8). Kitselas Canyon cuts a deep gorge through the westernmost extent of the Hazel ton Mountains (Figure 3.2). The canyon is 1.7 km long, and is oriented approximately north/south. On the west side of the canyon, steep slopes rise abruptly from the river in most places. The highest peak on this side is Kitselas Mountain at 1488 metres. One exception to this steep relief is at mid-canyon, where a rocky promontory cuts into the river. The site of Gitlaxdzawk (the "Fortress") (GdTc 1) is located on this promontory. On the east side, the canyon walls rise vertically to a height of 20 to 30 metres, but then level on to a series of flat upper terraces. Behind these terraces, Bornite Mountain in the Bulk ley Range rises to Figure 3.2 Kitselas Canyon. 70 a height of 1751 metres. The Skeena Valley bottom elevation at Kitselas Canyon is 100 metres above sea level. A number of islands are exposed in the river, in the upper half of the canyon, depending on water level. These cut the river into numerous narrow, deep channels that are marked by rapids, boils and whirlpools. Shallow back eddies are located in numerous locations on both sides of the river. Immediately below Kitselas Canyon, Kleanza Creek empties into the Skeena River from the Bulk ley Range. The larger Zymoetz River, fed by McDonnel Lake also in the Bulkley Range, empties into Skeena River about 7 km below Kitselas Canyon. Above Kitselas Canyon, a number of small rivers and creeks cut the Hazel ton Mountains on either side of the Skeena. These include Shannon, Chimdemash, Hardscrabble. Sand, Legate, and Lorn Creeks. Surficial Geology The surficial geology of the Kitselas Canyon area is characterized primarily by g1aciof1uvia 1 deposits, laid down during glacial retreat up the Skeena Valley (Clague 1984). On the east side of the canyon, this deposit is a g1aciof1uvia1 veneer of sand and gravel ranging in depth from 0.5 to 1.0 metres. The west side of the canyon, within one hundred metres of the shoreline, is covered by a thicker g1aciof1uvia 1 blanket of 1.0 metre or more in depth (see Figure 3.3). This narrow strip widens perceptibly below 71 fluvial, deltaic, colluvial and organic sediments glaciofluviol - deltaic gravel and sand glaciomarine mud undifferentiated glacial sediments bedrock partially covered with till, colluvium Figure 3.3 Surficial geology of the Kitselas Canyon area, (source: Duffel 1 and Souther 1964; Clague 1984) 72 the canyon. As the steep slope rises from the west side of the canyon, the glaciofluvial material yields to exposed bedrock and a colluvial veneer. Bedrock exposures on both sides of the canyon are of metamorphosed sedimentary and volcanic rock, ranging in age from late Paleozoic to early Cretaceous times. Erosiona1 and depositional processes in Kitselas Canyon are generally not important due to the high, steep canyon walls that rise immediately from the river. However, at the upper end of the canyon, small river terraces are located on either side, and a narrow terrace exists just below the canyon on the east side. The site of Gitaus (GdTc 2) is located on the latter terrace. Erosion from river undercutting is extensive here, and it is likely that some of the site has been washed away. The terrain on the east side of the canyon is quite uneven, due mainly to the bedrock outcroppings, numerous deadfalls and small stagnant ponds in this coniferous rainforest environment. Large river terraces and floodplains exist just above Kitselas Canyon on the east side of Skeena River, and below the canyon at the mouth of Kleanza Creek. The latter is flanked by remnant deltas on either side. Colluvial sediments predominate on the steep west side of the canyon. An unusual feature is the presence of an active sand and gravel slide just below Kitselas Canyon on the east side. Gitaus is located in the area below the slide. It 73 is not known if the slide was active when the site was occupled. The local soil of Kitselas Canyon is derived from the glad of1uvia1 parent material and from colluvium. The soil type is mainly humo-ferric podosol, well-drained and very acidic (Valentine, Sprout, Baker and Lavkulich 1981). Leaching is extreme. This is to be expected, as the area is characterized by coarse-textured, acidic parent materials, high rainfall and dense coniferous forest. C1imate Kitselas Canyon is influenced by two major climatic divisions: coastal and continental (Inglis and MacDonald 1979:4). As one proceeds up the Skeena Valley from the canyon, the effect of the coastal climate, characterized by small seasonal variation in temperature, cold summers and mild wet winters, diminishes. It is steadily replaced by the continental climate with abrupt seasonal temperature changes, reduced precipitation and longer frost periods. Although the Coast Mountain range generally impedes the flow of Pacific maritime air into the interior, two gaps through the Coast Mountains, one by way of the Skeena Valley, and the other extending north from Kitimat to Terrace via the Kitimat-Kitsumka1 urn Valley, provide interior access for this moisture laden air. This influence is felt most noticeably at Terrace, but as distance from the coast increases, the maritime influence steadily weakens. At 74 Hazel ton, at the confluence of the Bulk ley and Skeena Rivers, the climate is significantly cooler and drier than at Terrace (British Columbia Lands Service 1969:63). At Terrace, during the period from April to September, precipitation averages 365.8 mm. During the period from October to March, precipitation, often in the form of snow, averages 947.4 mm. The wettest months of the year are October through December, and the driest months are May through July (Atmospheric Environment Service 1982: 234). This summer/winter variation in precipitation, with the winter average being significantly higher, is generally consistent with the coastal climatic regime. Terrace and Kitselas Canyon have moderately warm summers and mild winters. January is the coldest month of the year with a mean daily temperature of -5.90C. The warmest month is July with a mean daily temperature of 16.10C. Terrace experiences an average of 134 frost-free days per year (British Columbia Lands Service 1969:64). By contrast. Prince Rupert is wetter and temperature variation is less extreme; Hazel ton is drier and temperature variation is more extreme (see Table 3.1). This reflects the general upriver gradient of coastal to continental c1i mate. In addition to the coast/in land trend, altitudinal changes account for some climatic variation in the Kitselas Canyon area. Meteorological data on precipitation and temperature fluctuations do not exist to reflect these 75 Table 3.1 C1i mate of the 1ower Skeena area. JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC YEAR P R. 1 7 2. 9 3. 8 6. 1 8. 9 11. 3 13. 3 13. 5 11 7 8. 5 5. 0 2 6 7. 4 average mean daily temperature T -5. 9 -1. 4 1. 5 5. 7 9 9 13. 7 16. 1 15 8 11. 9 6. 4 0. 3 -3. 4 5. 9 in °C H -10. 7 -4. 8 -0. l" 5. 2 10. 2 13. 6* 15. 9~ 15. 5' 11. 2 5. 4 -1. 9 -7. 7' 4. 3 P. R. 184. 8 217 2 168. 9 175. 0 120 5* 110. 4' 112. 9~ 143'. 7' 231. 2 352. 4' 276 8" 241 1 2334. 9 total rainfall 211 943. in mm T 50. 1 56 4 41. 0 49. f 42 8* 42 5 56 6 63 6 98. 5 0 131. 8 90. 9 3 H 15 6 10 1 17. 2 16. 8' 32 5 56 0 55 8 53 4 56 5 74. 0 28. 7 13 4 430. 0 P R. 23 3 15 8 16. 2 5 1 0 1 0 0 0 0 0 0 0 0 0 2 3 8 19 6 84 1 total snowfall 105 403 in mm T 116 3 71 6 44 2 12 1 0 4 0 0 0 0 0 0 0 0 3 9 48 9 6 0 H 56 5 30 4 13 7 3 1 1 0 0 0 0 0 0 0 0 2 2 1 25 7 50 1 172. 8 total P R. 210 1 215 8 185 4 180 6 120 6 110 4 112 9 143 7 231 2 352 5 281 3 258 6 2403 1 precipitation in mm T 153 7 123 0 83 1 61 3 43 3 42 5 56 6 63 6 98 5 214 9 180 3 192 4 1313 2 ' H 76 5 36 2 30 2 20 2 33 7 56 0 55 8 53 4 56 7 77 9 57 0 71 0 624 6 number of days P R. 20 19 21 18 17 16 16 16 18 24 22 22 229 wi th T 18 17 15 14 13 12 12 13 16 22 19 21 192 precipitation H 16 11 10 10 11 13 14 15 16 18 16 17 167 number of frost P .R. 200 free days 134 (monthly figures T not available) H 69 P.R. = Prince Rupert, T. = Terrace, H. = Hazleton source: Atmospheric Environment Service 1982. 76 changes, but changes fn vegetation (see Table 3.2) from valley bottom coastal species that thrive on moderate temperatures to high elevation species adapted to alpine tundra conditions do indicate considerable variability in temperature. The important point here is that Kitselas Canyon is strategically located, so that one need travel only short distances from the canyon to experience a range of environmental conditions from coastal to interior, and from valley bottom to alpine tundra. Hydro1ogy The area is drained by the Skeena River, and south of Terrace by the Kitimat-Kitsumka1um Valley below Lakelse Lake (Clague 1984), via the Kitimat River into Kitimat Arm. The Skeena River accounts for most of the drainage in the area. The Skeena is one of the major rivers of Canada's west coast, draining an area of 42,217 km2 (Farley 1979:39). Many of the rivers and streams tributary to the Skeena drain large lakes. Babine Lake feeds the upper and lower Babine River; Morice Lake feeds the Bulkley system; McDonnel Lake feeds Zymoetz River; Kitsumkalum Lake feeds Kitsumkalum River; and Lakelse Lake feeds Lakelse River. All of these lakes are of economic importance in terms of sockeye salmon runs. The Skeena and its tributaries are permanent, year-round streams and rivers fed directly by lakes or mountain freshets from permanent snow fields and snowcaps. At Usk, on the Skeena 2 km upriver from Kitselas Canyon, the mean annual discharge is 900 m.3 per second, while in June at the height of the runoff, discharge can reach 3000 m3 per second (Farley 1979:39). The increased summer discharge coincides with the annual salmon runs, and was important to the aboriginal fishing technology. Higher summer water turbidity made techniques such as dip-netting and basket trapping more efficient (see Kew 1976:12). Natural Flora and Fauna This section outlines the animal and plant resources available in the Kitselas area. As suggested by its physiographic and c1imato1ogica1 position, Kitselas Canyon is in a unique location. The immediate canyon area is located within the Coastal Western Hemlock biogeoclimatic zone (Krajina 1973, 1976). This zone predominates on the north coast of British Columbia, and stretches up the Skeena Valley and adjacent Nass Valley to the middle regions of both rivers. The area surrounding Kitselas Canyon, however, is not entirely homogeneous. Major biogeoclimatic zones change with elevation, and more subtle zonal changes occur as distance from the coast increases. On the forested slopes flanking Kitselas Canyon, at elevations from 300 to 900 metres above sea level, the Subalplne Mountain Hemlock biogeoclimatic zone exists. At higher elevations, (1000 to 1700 metres ASL), the 78 Subalpine Engelmann Spruce-Suba1pine Fir biogeoclimatic zone occurs. Above the tree line, the Alpine Tundra zone exists. These zones all occur in patches on the middle and upper slopes flanking Kitselas Canyon. In the valley bottom. Coupe, Ray, Comeau, Ketcheson and Annas (1982:7) indicate a zonal change to a more interior orientation at a point about 30 km upriver from Kitselas Canyon. This marks the beginning of the Interior Cedar Hemlock biogeoclimatic zone, and coincides with a gradual widening of the valley bottom (see Figure 3.4). This diversity results in a wide variety of natural flora and fauna in the Kitselas Canyon area. It is convenient to consider three main altitudinal zones in the vicinity of Kitselas Canyon: alpine, middle s1opes/suba1pine, and valley bottom. Each offers different resources. The alpine zone was probably of least economic importance to the inhabitants of Kitselas Canyon. Mountain goat (Oreamnos americanus), marmot (Marmota monax) and cougar (Fe1 is conco1 or) are found seasonally at these high elevations. Mule deer (Odocoileus hemionus hemionus), kermody black bear (Ursus americanus kermodei), and grizzly bear (U. arctos horriblis) also range into the alpine zone during the summer and early fall months (see Table 3.3). J. MacDonald (1984: Table 1) lists a small mountain caribou (Rangifer tarandus montanus) population in the alpine zone adjacent to the Kitsumkalum Valley. The range of this population may also include the Kitselas area. Cowan and — —I Interior Cedar Hemlock Zone, Northern Transitional Subzone, Hozleton varianl Interior Cedar Hemlock Zone, Northern Transitional Subzone, Lower Nass Basin variant Subolpine Englemann Spruce - Suba I pine Fir Alpine Tundro Subolpine Mountain Hemlock Biogeoclimatic zones of the middle Skeena area, (source: Coupe et aj_. 1982; Krajina 1973, 1976) 80 Guiguet (1964:384) Indicate that mountain caribou inhabit 1:he eastern slopes of the Coast Range and the Bulkley Mounta i ns. The middle slope/subalplne zone Is characterized by mountain hemlock (Tsuga mertensiana) with lodgepole pine (Pinus contorta contorta), (P. c. 1 at i fo1i a) and amer i can green alder (Alnus viridus) also present. A number of important food plants occur in this zone (see Table 3.2), including one or more species of shrubs with edible fruits (Betulaceae, Capr i fo1i aceae, Empetraceae, Er i caceae, Grossu1ar i aceae, Roseaceae), horsetails (Equi staceae). ferns (Po1ypod i aceae), lilies (Lili aceae), orchids (Orch i daceae), and other herbaceous plants (Apiaceae, Araceae, Ara1iaceae, Cornaceae, Fabaceae, Onograceae, Orobanchaceae, and Urt i caceae). Of these, the edible berries of the various shrubs (especially Vacc i n i urn and Rubus) and other herbaceous plants were probably of greatest aboriginal economic i mportance. In addition, most fur-bearing animals of economic importance could be found in this zone through much of the year. The valley bottom was the richest and most varied ecological zone in terms of aboriginal economic importance. Major vegetation in this zone includes red cedar (Thuja  p1icata), yellow cedar (Chamaecyparis nootkatensis), western hemlock (Tsuga heterophy11 a), white spruce (Pi cea glauca), black spruce (P. mariana), and sitka spruce (P. sitchensi s). 81 Table 3.2 Food Plants Available in the Kitselas Canyon Vicinity Spec i es Edible berries beaked hazelnut (Corylus cornuta) coast red elderberry (Sambucus pubens) common snowberry (Symphoricarpos albus) high-bush cranberry (Viburnum edule) soapberry (Shepherdia canadensis) crowberry (Enpetrum nigrum) k i nn i k i nn i ck (Arctostaphylos uva-ursi) creeping snowberry (Gaul then" a hispidula) sa 1 a 1 (Gaultheria shalIon) Labrador tea (Ledum groen1 andicum) Alaskan blueberry (Vaccinium alaskaense) dwarf huckleberry (Vaccinium caespitosum) blue huckleberry (Vaccinium membranaceum) Hab i tat Zone valley bottoms, lower slopes (2)* riparian, streamside (3) thickets, open slopes (3) swamps, stream banks, (2) al1uvial sites dry open woods (2) various (3) dry, open woods (2) bogs, sphagnum hummocks (2) streamside, thickets (1) streamside, swamps (3) coastal forests, humus (1) mountain slopes (3) mountain slopes (3) oval-leaf huckleberry streamside to (Vacc i n i urn ova 1i fo1i urn) mountain slopes (3) 82 Table 3.2 (continued) Spec i es wild cranberry (Vacc i n i um oxycoccos) Habi tat sphagnum bogs Zone (3) red blueberry (Vaccinium parvifolium) bog blueberry (Vaccinium uliginosum) mountain cranberry (Vaccinium vitis-idaea) stink currant (Ribes bracteosum) prickly gooseberry (Ribes lacustre) trailing black currant (Ribes 1 axiforum) northern gooseberry (Ribes oxyacanthoides) servi ceberry (Amelanchier a 1nifo1 ia) western crabapple (Mai us fusca) bitter cherry (Prunus emarginata) choke cherry (Prunus virqiniana) nootka rose (Rosa nutkana) red raspberry (Rubus idaeus) th i mb1eberry (Rubus parvi florus) valley bottom forests sphagnum bogs dry bogs, thickets rIparIan riparian to sub-alpine riparian to sub-alpine dry, open valley bottom open mountain slopes swamps, bogs to mounta in si opes riparian to wet slopes dry, exposed sites thickets to open woods riparian to open woods open woods (3) (3) (3) ( 1 ) (3) (3) (3) (2) ( 1 ) ( 1 ) (2) (3) (2) (3) 83 Table 3.2 (continued) Spec i es sa1monberry (Rubus spectabilis) c1oudberry (Rubus chamaemorus) strawberry bramble (Rubus pedatus) red blackberry (Rubus pubescens) Canadian bunchberry (Cornus canadensis) blue leaf strawberry (Fragaria virqiniana) Habitat riparian, swamps sphagnum bogs forested slopes riparian, moist woods moist humic soil, riparian to mid-slopes open exposed sites Zone ( 1 ) (2) (3) (3) (3) (3) Non—berries cascade mountain ash (Sorbus scopulina) sitka mountain ash (Sorbus sitchensis) horseta i1 (Equi setum arvense) lady fern (Athyrium fi1ix-femina) spiny wood fern (Dryopteris assimilis) male fern (Dryopteris filix-rnas) sword fern (Polystichum muni turn) bracken fern (Pteridium aquilinum) mid-slopes to sub-alpine open mid-slopes to sub-alpine open woods moist woods moist woods moist woods, talus slopes wet., shaded woods dry, open sites (3) (3) (3) (3) (3) (3) (3) (3) 84 Table 3.2 (continued) Spec i es Hab i tat Zone nodding onion open, exposed sites (3) (A 1 1 i urn cernum) rice root moist, open, mountain slopes (3) (Fritillaria camschatcensis) Solomon's seal riparian, moist woods (3) (Smilacina stellata) fairy slipper moist, shaded woods (3) (Calypso bulbosa) wild saraspari1 la moist, shaded woods (2) (Aralia nudicaulis) fireweed open exposed sites (3) (Epilobium angustifo1iurn) cinquefoi1 swamps, bogs (3) (Potentilla palustris) stinging nettle moist, shaded valley bottom (3) (Urtica d i o i ca) * (1) coastal (2) interior (3) coast and interior Sources: Turner (1978); Coupe et a 1 . (1982) 85 In addition, broad leaf deciduous species, including black cottonwood (Populus balsimfera), common paper birch (Betu1 a  papyrifera), bog birch (B. glandulosa), and small trees including douglas maple (Acer glabrum), trembling aspen (Populus tremuloides) and red alder (Alnus rubra) all occur in this zone, especially in swampy, marshy or streamside locations. All of the food plant families listed for the middle s1opes/suba1pine zone (see Table 3.2) occur in the valley bottom in greater or less abundance, although there are often differences in the time of ripening between these two zones. Two families not present on the middle slopes, Nymphaeceae and Typhaceae occur in swampy or marshy 1ocat i ons. Most mammalian species of the lower Skeena occur in the valley bottom zone (except the alpine species). Duffel 1 and Souther (1964:3) report that coast deer (Odocoi1eus  hemionus columbianus) are found occasionally in the lower Skeena Valley, although Cowan and Guiguet (1965:369) indicate that mule deer and sitka deer (0. h. sitchensis) may also be found here. Sitka deer are generally regarded as a coastal species, whereas mule deer are more common in the interior. In fact, Kitselas Canyon is situated roughly on the distributional boundary between these latter two sub-species, so it is likely that both are indigenous to the area (see Cowan and Guiguet 1965:371). 86 Table 3.3 lists all mammalian species whose present range includes the Kitselas Canyon area. Avian species found in the Kitselas area are listed in Table 3.4. As stated previously, Kitselas Canyon roughly straddles the coastal-interior boundary. The importance of this in terms of the variety of edible species - both plant and animal - cannot be underestimated. The ameliorated coastal climate of Kitselas Canyon means that a variety of coastal food plants, such as sa1 a 1, salmonberry, crabapple and Alaskan blueberry are common in the Kitselas Canyon valley bottom area (Coupe et a_l_ 1982:32, 37, 45, 49). A short distance upriver, the effect of the wet coastal climate is attenuated. Drier, more open conditions exist; food plants that are rare or absent on the coast are more abundant here. These include kinnikinnick, high-bush cranberry, serviceberry and choke cherry (Coupe et aj_ 1982:25, 29, 45, 46). In addition, mule deer, mountain caribou and mountain goat, al1 found at or near Kitselas Canyon, are rare or absent on the coast. The single most important resource to the aboriginal economy was fish. Nolan (1977:401-11) has listed 293 i chthyofauna1 species indigenous to the north coast area. Most of these spend their entire lives in salt water. 87 Table 3.3 Mamma 1 i an Fauna of" the M i dd 1 e Skeena Area I nsect i vores Cinerus shrew (Sorex cf nereis) 1* wandering shrew (Sorex vagrans longicauda) 3 navigator shrew (Sorex paiustris navigator) 1 little brown myotis (Myotis lucifugus alascensis) 3 yuma myotis (Myotis yumanens i s saturatus) 1 Lagomorphs varying hare (Lepus americanus) 2 Rodents marmot (Marmota monax petrensis) 2 northwestern chipmunk (Eutamias amoenus) 2 red squirrel (Tamiasciurus hudsonicus picatus) 1 flying squirrel (Glaucomys sabrinus zaphaeus) 1 american beaver (Castor canadensis belugae) 1 (C# c. sagittatus) 2 white-footed mouse (Peromyscus maniculatus) 3 northern bog-lemming (Synaptomys boreal is wrangeli) I (S. b. da 1 i ) 2 boreal redback vole (Clethrionomys gapperi caurinus) 1 (C. g. saturatus) 2 long-tailed vole (Microtus longicaudus littoral is) 3 muskrat (Ondrata zibethica osoyoosensis) 3 88 Table 3.3 (continued) Spec i es Zone meadow jumping mouse (Zapus hudsonius hudsonius) 3 western jumping mouse (Zapus princeps saltator) 3 american porcupine (Erithizon dorsatum) 3 Carn i vores coyote (Can is latrans inoculatus) 3 wolf (Canis lupus fuscus) 1 (C. 1. columbianus) 3 red fox (Vulpes fulva abietorum) 3 black bear (Ursus amer i canus kermodei) 3 grizzly bear (Ursus arctos horriblis)marten (Martes americana caurina) 3 fisher (Martes pennant i Columbiana)short-tailed weasel (Mustela erminea richardsoni) 3 mink (Mustela vison) 3 wolverine (Gulo 1uscus luscus) 3 striped skunk (Meph i st i s meph i st i s) 3 Canadian river otter (Lutra canadensis) 3 cougar (Felis concolor missou1 ensis)Canada lynx (Lynx canadensis) 2 P i nn i peds hair seal (Phoca vitulina richardi) 1 89 Table 3.3 (continued) Spec i es Zone Art i odacty1s mule deer (Odocoileus hemionus hemionus) 2 (0. h. sitkensis) 1 moose (Alces a Ices) 3 mountain caribou (Ranqifer tarandus montanus) 2 mountain qoat (Oreamnos amer i canus americanus) 3 * (1) coasta1 (2) interior (3) coast and interior source: Cowan and Guiguet (1965) Table 3.4 Avian Fauna of the Middle Skeena Area Waterfow1 common loon (Gavia immer) red-throated loon (Gavia stellata) red-necked grebe (Podiceps gricenqa) horned grebe (Podiceps auritus) great blue heron (Ardia herodius) 90 Table 3.4 (continued) Canada goose (Branta canadensis) mallard duck (Anas piatyrhynchos) pintail duck (Anas acuta) green-winged teal (Anas carolinensis) blue-winged teal (Anas discors) American widgeon (.t^areca americana) shove 1er (Spatula clypeata) common goldeneye (Bucephela clangula) Barrow's goldeneye (Bucephela islandica) harlequin duck (Hi str ioni cus h i str ion i cus) surf scoter (Me 1 an i tta persp i c i11ata) hooded merganser (Lophodytes cucullatus) common merganser (Mergus merganser) sora (Porzana Carolina) American coot (Fulica americana) black oystercatcher (Haematopus bachmani) Upland Game Birds blue grouse (Dendragapus obscurus) Franklin grouse (Canachites franklini) ruffed grouse (Bonassa umbellus) spruce grouse (Canachites canadensis) sharp-tailed grouse (Ped i ocetes phas i ane11 us) rock ptarmigan (Laqopus lupestris) willow ptarmigan (Laqopus laqopus) 91 Table 3.4 (continued) white-tailed ptarmigan (Lagopus leucurus) band-tailed pidgeon (Columba fasciata) Birds of Prey goshawk (Accipter gentilis) sharp-shinned hawk (Accipter striatus) red-tailed hawk (Buteo .jama i cens i s) bald eagle (Ha 1iaetus 1eucocepha1 us) marsh hawk (Circus cyaneus) osprey (Pandio haliaetus) pigeon hawk (FaICQ columbarius) sparrow hawk (FaICQ sparverius) sources: Godfrey (1966); Guiguet (1961, 1968). Only 31 species spend part or a 1 1 of their lives in the fresh water of the Skeena or Nass systems. These include 14 anadromous species, which spend most of their adult lives in the sea or estuaries, but enter fresh water to spawn in streams, rivers or lakes; 5 adfluvial species, which spend most of their adult lives in large river or lakes, 92 ascending smaller streams to spawn; and 12 f1uvia1-1acustrine species which spend their entire lives in fresh water, but are not adfluvial. The species included in these three general categories are listed in Table 3.5. Most, but not all 3 1 species may be found in Skeena River at Kitselas Canyon. Eulachon (Thaieichthys pacificus) run in Skeena River, but not in great abundance, and spawn below Kitselas Canyon. The major eulachon run occurs in the Nass River, and a smaller run occurs in the Kitimat River. Two other anadromous species, river, lamprey (Lampetra  ayres i) and green sturgeon (Acipenser medirostris), are both very rare in Skeena River, and it is not known if they occur as far Upriver as Kitselas Canyon (Carl Clemens and Lindsey 1967:30, 33; Hart 1973:23, 82; Scott and Crossman 1973:56, 91). Table 3.5 Ichthyofauna1 Species Indigenous to the Skeena River Anadromous river lamprey (Lampetra ayres i)* green sturgeon (Acipenser medirostris)* white sturgeon (Acipenser transmontanus) longfin smelt (Spirinchus tha1eichthys) coastal cutthroat trout (Salmo clarki clarki) steel head trout (Sa1 mo gai rdner i) (marine) 93 Table 3.5 (continued) Dolly Varden trout (Sa1ve1i nus ma 1 ma) (marine) pink salmon (Oncorhynchus gorbuscha) chum salmon (Onchorhynchus keta)* coho salmon (Oncorhynchus kisutch) sockeye salmon (Oncorhynchus nerka) chinook salmon (Oncorhynchus tshawytscha) Adfluvial longnose sucker (Catostomus catostomus) white sucker (Catostomus commersoni) largescale sucker (Catostomus macrocheilus) rainbow trout (Sa1 mo ga i rdner i) (freshwater) Dolly Varden (Salvelinus malma) (freshwater) F 1 uvia 1-Lacustrine western brook lamprey (Lampetra r i chardson i) longnose dace (Rhinichthys cataractae) redside shiner (Richardsonius balteatus) threespine stickleback (Gasterosteus aculeatus) coastrange sculpin (Cottus aleuticus) prickly sculpin (Cottus asper) peamouth (Mylocheilus caurinus) lake whitefish (Coregonus clupeaformis) mountain whitefish (Prosopium williamsoni) northern squawfish (Ptychochei1 us oregonensis) lake trout (Sa1ve1i nus namaycush) 94 Table 3.5 (continued) burbot (Lota lota) * rare in Kitselas Canyon sources: Rost1und (1952); Carl, Clemens and Lindsey (1967); Nolan (1977). Thus, of the 293 species of fish occurring in or near north coastal waters, less than 107, (N=28) can be found in the Skeena system in the vicinity of Kitselas Canyon. In addition, seven of these species achieve a maximum size of less than 20 cm. With the possible exception of longfin smelt (Spirinchus tha1eichthys), it is unlikely that any of these were of aboriginal economic importance. Although ichthyofauna1 diversity is minimal at Kitselas Canyon (relative to the coast), this does not mean that ichthyfaunal biomass is minimal. Quite the opposite is the case, primarily due to the presence of one genus of fish - Qnchorhynchus (salmon). All five species of North Pacific salmon that migrate into North American river systems occur at Kitselas Canyon. Four of these - sockeye (Oncorhynchus nerka), pink (0. gorbuscha), coho (0. k i sutch), and chinook (0. tschawytscha) - occur in large numbers in Kitselas Canyon. In fact, high proportions of the Skeena 95 runs of each of these species pass through Kitselas Canyon on their way to upriver spawning beds (see Table 3.6). The fifth species, chum (0. keta). occurs in Kitselas Canyon, but in unknown abundance. Chum generally run a short distance in fresh water before spawning, and according to Aro and Shepard (1967:286), the Skeena is only a minor producer of chum. Sockeye and coho run long distances upriver before spawning. Sockeye only spawn in streams that are fed and flushed by lakes (Forester 1968). The major sockeye runs are in the upper and lower Babine Rivers and Fulton River (Milne 1955) (see Table 3.6). The major coho run is in Lakelse River, just below Kitselas Canyon (see Table 3.6). Chinook or pink may run long or short distances, depending on the breeding population. The major pink run is in Kispiox River in odd years, and in Lakelse River in even years. The major chinook runs are in Bear River, lower Babine River, Kispiox River and Morice River. Aro and Shepard (1967:296-99) indicate that approximately 83.4% of a 1 1 spawning sockeye, coho, pink and chinook pass through Kitselas Canyon annually. 96 Table 3.6 Salmon Escapement Through Kitselas Canyon Escapement Spawn Spawn ing Locat i on (X 1000 fish) Start Peak End Sockeye Upper Babine River 1 16.0 Sept. Oct. Oct. Lower Babine River 84. 0 Sept. Oct. Oct. Bear Lake 5-10 Sept. Oct. Oct. Fal1s Creek 2-5 Sept. Sept. Oct. Pinkut Creek 29. 0 Aug. Sept. Sept. Four Mile Creek 2.3 Aug. Aug. Sept. Fulton River 86. 0 Sept. Oct. Oct. Grizzly Creek 9.9 Aug. Sept. Sept. Pierre Creek 23. 7 Jul y Sept. Sept. Tachek Creek 2.2 Aug. Sept. Sept. Twin Creek 7.9 Ju 1 y Aug. Sept. Willi ams Creek * 5-10 Aug. Aug. Sept. Ink (odd and even years) Odd Even Lower Babine River 28. 1 15.7 Aug. Sept. Sept. Kispiox River 475.5 59.0 Jul y Sept. Sept. Kitwanga River 158.8 71.3 Aug. Sept. Sept. Lakelse River * 206.3 273.5 Aug. Sept. Oct. Skeena River 165. 0 25.5 Aug. Sept. Sept. 97 Table 3.6 (continued) Spawn i ng Locat i on Escapement (X 1000 fish) Start Spawn Peak End Chum Skeena River unknown Coho Lower Babine River 2-5 Sept. Oct. Nov. Upper Babine River 2-5 Sept. Oct. Nov. Upper Bulkley River 2-5 Oct. Nov. Dec. Lower Bulkley River 1-2 Oct. Nov. Dec. Gosne11 Creek 2-5 Sept. Sept. Nov. Kispiox River 5-10 Sept. Oct. Nov. Lakelse River * 20-50 Sept. Oct. Nov. MacDone11 Lake 1-2 Sept. Oct. Nov. Morice River 5-10 Aug. Sept. Nov. Ch i nook Lower Babine River 5-10 Sept . Sept . Sept. Bear River 10-20 Sept . Sept . Oct. Upper Bulkley River 0.5-1 Aug . Sept . Sept. Cedar River 0.5-1 Aug . Sept . Sept. Kispiox River 5-10 Aug . Sept . Sept. Morice River 5-10 Aug . Sept . Oct. * depart Skeena River just below Kitselas Canyon source: Aro and Shepard (1967) 98 Chinook may be present year round in Kitselas Canyon, but the major run, along with the sockeye run, begins in June. Both species occur in the canyon in abundance as late as August or September, as successive breeding populations run through the canyon. Pink and coho begin to run through the canyon in early July. The runs of pink are short; spawning usually occurs in September. The coho run lasts longer, with spawns occurring into November. Other ichthyofauna1 species occurring in Kitselas Canyon include the anadromous and adfluvial varieties of trout: cutthroat (Salmo clarki), steel head (S. gai rdneri), and Dolly Varden (Salvelinus malma). These species do not approach the abundance or predictability of salmon, but were undoubtedly important nonetheless in the aboriginal economy. This is especially true of steel head and cutthroat which spawn in late winter and early spring, a time when other resources are depleted. The adfluvial species are similar to the anadromous species in that spawning is a cyclical event that occurs at the same time each year, and takes place in shallow creeks and streams. These fish are most predictable and accessible at this time. The disadvantage of adfluvial and fluvial lacustrine fish is their reduced abundance relative to anadromous species. Without access to the productive marine biocycle, the amount of exploitable ichthyobiomass in spawning adfluvial and fluvial lacustrine fish is much less than among anadromous fish. Fluvial lacustrine species have the additional disadvantage of being unpredictable and generally inaccessible. They do not spawn in large numbers or at predictable times, and spend most of their time in deep water (large rivers or lakes). Pa 1eoenv i ronment The pa 1eoenvironmenta1 conditions of Kitselas Canyon, and surrounding area are not well known at this time. Two studies of late Quaternary environmental change have been conducted in the Terrace area (Duffel 1 and Souther 1964; Clague 1984). These studies focussed primarily on the history of late Pleistocene glaciation and de-g1ac?ation in the Kitimat-Kitsumka1um Valley, and in the Skeena Valley below Terrace. The data from these studies can be extrapolated to provide a general picture of the late Quaternary environment of the larger area of the lower Skeena including Kitselas Canyon. At the climax of the late Wisconsinan Fraser glaciation (ca. 14,500 B.P.), a massive ice sheet covered much of British Columbia. Along the north coast area, three major ice lobes projected down Nass Valley, Skeena Valley and the Kitimat-Kitsumka1um Valley. They coalesced over eastern Hecate Strait (Clague 1984:41), and a complex of confluent ice tongues enveloped all valleys and lowlands in the area. Clague (1984:6-7) suggests that during the climax of the Fraser glaciation, peaks and ridges above 1500 to 2000 metres 100 fn the Coast and Hazel ton Mountain ranges probably projected above the surface of the ice to form nunataks. De-glaciation began about 13,000 years ago, and occurred by a combination of downwasting and complex frontal retreat (Clague 1984:2). The outer coast became ice-free first, in response to destabi1ization of the western periphery of the Cordilleran ice sheet by eustatically rising seas. By 12,500 to 12,000 B.P., the western margin of the Cordilleran ice sheet had disintegrated in Hecate Strait leaving discrete lobes at the present day mouths of the Nass and Skeena Rivers, and at the base of Douglas Channel. At this time, glaciers were retreating rapidly by calving, and were soon confined to fjords and mountain valleys. Clague (1984: Figure 9) indicates that by 11,000 B.P. the Skeena Lobe had retreated about 70 km up Skeena Valley from the present river mouth. At the same time, the Kitimat Lobe had almost completely retreated from Douglas Channel, remaining only in Kitimat Arm. By 10,500 to 10,000 B.P., these two piedmont lobes had split in Skeena Valley at Terrace. From here, the Skeena glacier continued to retreat westward up Skeena Valley, while the Kitimat-Kitsumka1 urn glacier retreated northward up Kitsumkalum Valley. Kitselas Canyon was probably ice-free by 10,000 B.P. or shortly thereafter. During the period of deg1aciation, Skeena Valley and other valleys and lowlands in the area were isostatica11y depressed. As the glaciers retreated the sea closed in. 101 often in contact with the ice front. At about 10,000 B.P., Skeena Valley existed as an arm of the sea to a point well above Kitselas Canyon. Marine shell collected from mud near Terrace yielded a radiocarbon age of 10,200 + 100 B.P. (GSC 2306) (Clague 1984:15). Similar marine flooding, in contact with the retreating glacier, also occurred in the Kitimat-Kitsumka1um Valley. It halted where Kitsumkalum Valley narrows north of Pine Lake, and constructed a large delta sandur complex graded to a shoreline about 200 metres above present sea level (Clague 1984:20). Clague's (1984) study, which is the first detailed analysis of the de-g1aciation of this area of the north coast of British Columbia, focuses on these deltas, and on the large ice-frontal delta sandur complexes in particular. There are two in the Kitimat Valley, and both mark periods of stagnation or stabilization during the ice retreat. One is located 5 km south of Lakelse Lake; the other is 17 km north at Terrace airport. The Lakelse sandur is the largest, stretching completely across the valley, and covering an area of 60 km2. Both sandurs graded to shorelines about 200 metres above the present. Despite the presence of these stillstands, Clague (1984:18) argues that deglaciation from Kitimat to Tsrrace took no more than 1000 years. Radiocarbon dates on glaciomarine sediments in this part of the valley range from 9900 to 10,800 B.P., and do not become systematically younger towards the north (i.e., the direction of retreat). This suggests that the 102 outwash plain probably formed in a relatively short time period, and that around these still stands glacial retreat was very rapid. Marine transgressions culminated in the Skeena Valley (including Kitselas Canyon) and Kitimat-Kitsumka1um Valley about 10,500 B.P. (Clague 1984:24; Clague, Harper, Hebda and Howes 1982), when the sea was 200 metres higher relative to the land than at present. Isostatic rebound at the close of the Pleistocene and during the early Holocene lead to rapid emergence of Skeena Valley and Kitimat-Kitsumka1um Valley. Clague et aj_ (1982) indicate that shorelines had fallen to about 120 metres above present by 10,000 B.P., and to +35 metres by about 9300 B.P. Glacial retreat was diachronous, so the timing of emergence probably differed locally. Clague (1984:26) states that the sea in the Skeena and Kitimat-Kitsumka1um Valley areas had fallen to its present level relative to the land by 8000 B.P. He indicates no major sea level fluctuations after this time. The Recent Environment This section examines the potential for human occupation of Kitselas Canyon and the lower Skeena during the late Pleistocene and Holocene. The seminal research on late Pleistocene and Holocene environments on the north coast of British Columbia was conducted by Heusser (1960, 1965) as part of a larger study of North Pacific North America. The north coast area studied was Prince Rupert, where cores 103 from three peat bogs were taken. Heusser's (1960:181-89) four phase climatic scheme, covering the period from the late Pleistocene to the present remains the yardstick for measuring late Quaternary climatic change, although It is thought by some to be overly simplistic (e.g., Bryson, Barrel's and Wend land 1970) for an area stretching from California to western Alaska. However, in the absence of any more detailed post-Pleistocene environmental recontruction for the area pertinent to Kitselas Canyon, Heusser's generalized scheme is adopted here with some modification (cf. Miller and Anderson 1974). Heusser's reconstruction is as follows: 1) late glacial, 2) early Postglacial, 3) hypsitherma1, 4) late Postglacial. Each phase is characterized by climatic change from its predecessor, and by distinctive floral communities. The late glacial phase in northern British Columbia, beginning about 11,000 years ago was significantly colder and drier than the modern climate. At Prince Rupert, willow, alder and lodgepole pine dominated. Sedges and the Po1ypod i ceae were present as ground cover, but ferns were present only in small amounts. Alpine tundra conditions were present at considerably lower elevations than today. The Kitselas Canyon area may have been inhabitable for man at this time, but it would have been a harsh environment, with river valleys clogged by ice, and minimal ground cover to support animal species. The early Postglacial phase in northern British Columbia 104 began about 10,000 years ago, and lasted until about 8000 B.P. Heusser (1960:183) indicates that a cool, moist climate continued to prevail on the north coast, although somewhat warmer than that which dominated the final interval of the late glacial phase. At some time during the early Postglacial, climatic conditions were similar to those that exist today. Lodgepole pine, alder and sedges and ferns were the principal members of the vegetation community. Sitka spruce and western and mountain hemlocks became established at this time, but only in small, numbers (Heusser 1960:183). In the area of Kitselas Canyon, east of the Coast Mountains, it is not clear whether wet climatic conditions existed during the early Postglacial, as they did on the coast. Miller and Anderson (1974) have compared pollen cores from the Taku district in the Alaska panhandle (after Heusser 1952, 1960) with samples from At 1in in northwestern British Columbia, east of the Coast Mountains. This contrast is analogous to that between Prince Rupert Harbour and Kitselas Canyon. During the early Postglacial, the lodgepole pine maximum occurred at Taku about 9000 B.P. (Miller and Anderson 1974:46), indicating cool, wet conditions, as suggested by Heusser (1960:183). However, inland at Atlin, Miller and Anderson (1974:46) state that a spruce woodland environment, indicative of a cool dry climate, existed at this time. They relate this to an east-west shifting across 105 the Coast Mountain ranges of the Arctic Front - the line of demarcation between high pressure anti-eye 1 onic continental weather, and low pressure, cyclonic maritime conditions - throughout the Holocene. When this front was located west of the Coast Mountains, the associated storm path would produce cool, wet conditions on the coast, and cool, dry conditions inland. When the front was situated east of the Coast Mountains, relatively warmer and drier conditions would have existed on the coast, while warm, wet conditions would have existed inland (Killer and Anderson 1974:54). The implications of this inverse pattern for Kitselas Canyon are not clear. If the Taku/Atlin climatic pattern held for Prince Rupert/Kitse1 as Canyon, then a cool, dry early Postglacial climate may have existed at Kitselas Canyon. However, two additional factors exist to further confuse the picture. First, Taku/Atlin are much farther north than the lower Skeena. Therefore, latitudinal change as a factor affecting climatic variability cannot be ignored. Secondly, the Skeena Valley allows maritime air to penetrate to Kitselas Canyon. No similar condition exists at At 1 in, where today semi-arid conditions prevail. Given this evidence, climatic conditions at Kitselas Canyon during the early Postglacial may have been cooler and somewhat drier than on the coast, as is the case today. The question of sea level fluctuations at the end of the early Postglacial remains contentious. Heusser 106 (1960:194) and Fladmark (1975:151-52) both argue that as late as 7500 B.P., sea level in the Prince Rupert area may have been 40 to 50 metres higher than at present. Fladmark (1975:152) indicates that sea level at Prince Rupert did not stabilize to present conditions until shortly before 5000 B.P. This model of continued sea level fluctuations into the middle Holcene was important for Fladmark because it served as the basis for his "salmon climax productivity" model (Fladmark 1975:293-96). Briefly, this model states that sea level fluctuations along the Northwest Coast to 5000 B.P. resulted in reduced salmon productivity relative to the late Holocene, as river and lake beds (salmon spawning locations) would have been unstable during the period of sea level fluctuation. Fladmark (1975:293-96) argued that it was only with sea level stabilization and the resulting "climax productivity" of salmon, ca. 5000 B.P., that the ethnographic Northwest Coast cultural pattern began to deve1 op. There are two problems with this model. CIague's Lakelse Lake cores indicate that marine shorelines in the Kitimat-Kitsumka1 urn Valley dropped 200 metres in the 2000 year interval of the early Postglacial, and that shortly after 8000 B.P., the sea level in the Prince Rupert-Kitimat area was at its present level (Clague 1984:26). Mathews' (1979) reconstruction of early Holocene sea levels is consistent with Clague's model. Mathews (1979:150) argued that sea levels, river and lake beds had stabilized by no 107 later than 8000 B.P. Thus, we may envision salmon climax productivity at or shortly after this time. Even if climax productivity did not occur until 5000 B.P., it is not clear that this was important. The size of the climax runs were not seriously affected by intensive fishing, even during the ethnographic period (see Kew 1976:2). It is entirely possible that "sub-climax" runs were still large enough to make salmon specialization profitable for the human procurer. The relevant point here is that salmon probably existed in abundance on the Northwest Coast long before the advent of intensive procurement. Sea level stabilization and the attainment of salmon climax productivity probably had little to do with the evolution of the ethnographic north coast cultural pattern. The Hypsithermal interval (Heusser 1960:184-86) succeeded the early Postglacial on the north coast of British Columbia at about 8000 B.P. and lasted until roughly 3500 B.P. This interval was marked by successiona1 sub-divisions of forest cover. The first succession, lasting until about 6000 B.P., was marked by sitka spruce, alder and mountain hemlock, and was associated with ground cover dominated by sedge and fern. The later succession was marked by increased importance of western hemlock, which eventually predominated over sitka spruce. Temperature and humidity fluctuations occurred throughout this 5000 year interval. Heusser (1960:185) 108 states that at Prince Rupert, successions of peat types (two horizons of ligneous peat separated and underlain by sedge peat) indicate two discrete episodes of humidity increase, followed by decrease. Miller and Anderson (1974:49-50) indicate that at Atlin, warm, wet conditions existed throughout the Hypsitherma1. This is consistent with their shifting Arctic Front model. Taku was experiencing warm, dry conditions at this time. Again, this has ramifications for Kitselas Canyon. While relatively warm, dry conditions (although fluctuating) existed at Prince Rupert during the Hypsitherma1, it is possible that increased humidity and storminess prevailed at Kitselas Canyon. This implies that more extensive, luxuriant vegetation existed in the area throughout the Hypsithermal than in preceding and succeeding intervals. It is unlikely that an open, semi-arid environment ever existed at Kitselas Canyon during the Holocene. If Miller and Anderson's model has even remote applicability to Kitselas Canyon, it is entirely possible that the forest cover of the late Hypsithermal was even more extensive than it is today. The final climatic interval in Heusser's scheme is the late Postglacial. On the north coast, in the Prince Rupert area, this began about 3500 B.P., and is marked by cooler temperatures and generally higher precipitation (although again fluctuating) than at present. Heusser (1960:187) noted that these conditions were particularly 109 conducive to glacier development. Western hemlock remained the dominant species on the coast, but sitka spruce appears to have declined, being replaced by mountain hemlock and lodgepole pine. On the ground, heath and Sphagnum became dominant, succeeding Lysichiturn. At Atlin, Miller and Anderson (1974:47-48) divided the period covered by the late Postglacial into three intervals. Their pollen cores indicate that the earliest interval (to 2500 B.P.) was marked by continuing wetness and decreasing temperatures. This is indicated by the appearance of alpine fir in the Atlin spruce forest. The middle interval to 750 B.P. was characterized by cooler and somewhat drier conditions than the preceding interval. The most recent interval embraces the "Little Ice Age" on the immediate coast (Taku), but Miller and Anderson (1974:47) state that at Atlin conditions were actually slightly warmer and drier than in the preceding phase. At Kitselas Canyon, cool, wet conditions are likely to have existed during the late Postglacial because of the ameliorating effects of the maritime air mass, via the Skeena Valley access route. Late Postglacial climatic conditions at Kitselas Canyon were probably cooler than during the preceding Hypsitherma1, and similar in general to climatic conditions that exist in the area today. 110 Summary Kitselas Canyon was probably habitable by 8000 B.P. at the latest. At this time, postglacial temperatures were becoming warmer, and precipitation may have been increasing moderately. Rapid forest succession may be envisioned. In the ensuing 5000 years, warmer conditions may have existed (relative to the present). At no time however, should we envision truly xerothermic conditions at Kitselas Canyon. The ameliorating effects of the off-shore maritime air mass would have maintained moderate to high moisture levels throughout the Hypsitherma1, and inland, storm tracks from the Arctic Front may have actually produced wetter conditions during the thermal maximum than at present. The last 3000 years have been characterized by cool and moderately moist conditions, similar to the present. Little vegetation change from the preceding phase is envisioned, except for the forest floor replacement of Lys i ch i turn by heath and Sphagnum. Coniferous forest conditions similar to those of the present may have developed at Kitselas Canyon shortly after 8000 B.P., and were almost certainly in existance by 5000 B.P. Sea levels were stable, or nearly so, by 8000 B.P., with only minor oscillations after this time. River downcutting, as a response to these oscillations, would have been minimal. Thus, we may envision salmon runs approaching present size by shortly after 8000 B.P. Throughout the Holocene, minor climatic osci11 at ions may have resulted in the altitudinal raising or lowering of Ill certain bfotfc communfties, but fn general, major environmental change fn the last 8000 years fs unlikely. The Initial condition of the model presented in Chapter 2 was one fn which access to resources was restricted by environmental conditions. Corporate groups are most likely to form at primary locations of resource access. The natural environment of Kitselas Canyon satisfies the fnitfal condition. An abundant, reliable, localized resource -salmon - could be procured at Kitselas Canyon. In fact, Kitselas Canyon is a critical procurement location on the Skeena River. Further, a variety of other resources were available in the area because of the unique location of the canyon. This contributed to the importance of the canyon as a primary location for resource control. 112 Chapter 4 KITSELAS CANYON; CULTURAL SETTING Introduction This chapter includes detailed ethnographic (Part 1) and prehistoric (Part 2) information on the Kitselas and lower Skeena areas. The ethnographic summary presents a model of Kitselas life at the time of European contact. This represents the culmination of evolutionary trends in prehistoric settlement, subsistence and social organization. It is demonstrated that the Kitselas were permanent residents of the canyon, who controlled a defined territory along the Skeena River. Their subsistence was oriented strongly toward salmon. The Kitselas were organized in ranked corporate groups, based on ascribed status differences. This model constitutes a base of comparison for the prehistoric evidence from Kitselas Canyon. The prehistoric summary begins with an examination of the evidence for early occupation of the north coast. The remainder of this section focusses on the lower Skeena localities of Prince Rupert Harbour, Kitselas Canyon and Hagwilget Canyon. The archaeology of these three areas is presented in detail; there is, at present, no comprehensive discussion of lower Skeena archaeology in the literature. These data are used in subsequent chapters to investigate change in prehistoric social organization. Existing models of prehistoric cultural change from Prince Rupert Harbour (MacDonald 1969a) and Kitselas Canyon (Allaire 113 1979) are critically examined. Part 1 REGIONAL ETHNOGRAPHY Cultural and Linguistic Affiliation Kitselas Canyon is the traditional homeland of the Gitdsilasshoo people (Emmons 1912:468) or Kitse1 as. "people of the canyon". Although visited on many occasions during the late 19th and early 20th centuries, the Kitselas were never described in a detailed ethnography. Brief descriptions exist (Schoolcraft 1860; Dawson 1881; Dorsey 1897; Emmons 1912; Shotridge 1918), but these are generally incomplete and in some cases unreliable. For example, Schoolcraft (1860) published a Kitselas census, but he never actually visited Kitselas Canyon. Much of the information in these reports relates to a period when the Kitselas villages were abandoned or in a state of decline. Allaire, MacDonald and Inglis (1979), using early photographs and the fieldnotes of H.I. Smith, and Man"us Barbeau and William Beynon, have meticulously compiled an ethnohi storic reconstruction of the Kitselas, focussing mainly on Gitlaxdzawk (the "Fortress"), one of the historic Kitselas villages. This report is of great value, not only because it describes each house structure at the site, but it also identifies the owner of each house, and where possible, provides the owner's phratric and clan affiliation. Linguistically, the Kitselas speak a Tsimshian dialect more closely related to Coast Tsimshian than to Gitksan 114 (Allaire et a 1. 1979:58). Duff (1965:18-19) refers to the Kitselas and the Kitsumkalum, who occupied the Kitsumkalum Valley north of Terrace, as the Canyon Tsimshian. and he too includes them linguistically with the Tsimshian tribes of the lower Skeena (see Figure 4.1). Allaire et a 1. (1979:58) suggest that this coastal rather than interior affiliation of the Kitselas was due to their closer proximity to the summer villages of the Coast Tsimshian on the Skeena below the canyon, than to the more distant Gitksan villages of the middle and upper Skeena. Emmons (1912:468) states that the Kitselas were intermediary between the Coast Tsimshian and the Gitksan, but it is not clear whether Emmons is speaking of Kitselas regional group affiliation here, or their role in coast-interior trade relations. Settlement Pattern and Yearly Round The General Coast Tsimshian Model Despite their ties to the coast, the Kitselas differed from most other Coast Tsimshian groups in some important respects. One difference was in settlement pattern. Prior to 1834, when the Hudson's Bay Company trading post was established at Port Simpson, the Coast Tsimshian local groups or tribes wintered in villages on the coast. Each tribe - Ginakangeek, Gitando, Gispakloats, Gilutsau, Gitlan, Gitwilgiots, Gitsees, Ginadoiks, Gitzaklahth (Garfield 1939:175-76) - had its own winter village location in the Figure 4.1 Tsimshian linguistic divisions (from Duff 1965). 116 Metlakatla Pass area of Prince Rupert Harbour (Boas 1916:394; Garfield 1939:177; Drucker 1950:160, 1965:115). These were permanent villages in the sense that they were maintained year after year, and the dwellings were owned by "House" groups. They were only occupied seasonally, however, from roughly November to March. Winter was primarily a time for social and ceremonial activities among the Tsimshian. Subsistence pursuits were minimal, but not altogether absent. There are references to land and sea mammal hunting (Boas 1916:339; Garfield 1951:15-16; Drucker 1965:119), and Inglis (1977:2) states that shellfish collecting and bird hunting were also practised. The annual round of the Coast Tsimshian groups of Metlakatla can be seen in terms of three major seasonal movements (see Fig 4.2). From their winter villages, most local groups moved in March to the Nass estuary for the eulachon fishery. Each winter village group had its own camping location on the lower Nass River (Boas 1916:44; Garfield 1939:199, 270, 1950:13, 15), and some household groups maintained permanent dwellings at these locations (Garfield 1939:277). Eulachon was a critical element in the Tsimshian diet because the oil or grease rendered from the fish was an important source of carbohydrates (Kuhnlein et al. 1982:155; Stewart 1975). Grease was highly desired by most northern groups, and the Coast Tsimshian and Niska in particular actively traded it. Garfield (1951:16) states that these groups 117 128° w 0 50 km 1 fcj it -I A J / late winter - early spring: Noss River eulachon fishery. Some task groups (broken arrows) engage in sea mammal hunting, herring and halibut fishing en route to and from Nass River. 2 early summer: Skeena River fishing, hunting and gathering. Small orrows indicate migrotion to tribal territories. 3 fall: return to winter village'ot Prince Rupert Harbour. Figure 4.2 Schemtic of Coast Tsimshian seasonal round. 118 were the primary producers. The Niska traded inland via the "grease trails" with the Gitksan and Carrier. The Coast Tsimshian traded with the people of Kitkatla and with the Haida. The many groups that congregated at the Nass undoubtedly resulted in some of the largest, pre-contact aboriginal aggregations on the north coast. While most Coast Tsimshian journeyed to the Nass in the spring, some task-specific groups formed to pursue two other resources that were also available at this time; herring and halibut. On the north coast, herring spawn in early spring in great numbers in shallow water close to shore (Carl 1964:22; Hart 1973:97), and halibut, after spawning in deep water in late winter, move into shallow water offshore in the spring (Nolan 1977:321). If the eulachon fishery ended early, some groups might proceed to the offshore islands where they engaged briefly in sea mammal hunting and halibut fishing (Boas 1916:399; Drucker 1965:117-18). This practice was usually ad hoc. and did not occur every year. The salmon runs began in May or June, and this signalled the second major phase of the Coast Tsimshian annual pattern of movement. The winter village groups journeyed up the Skeena River to fish for salmon and collect berries. There is some disagreement over the summer residence pattern. Boas (1916:399) indicates that the winter village groups simply relocated themselves to summer villages on the Skeena. However, Garfield (1939:277, 1951:16) states that the winter 119 village groups did not stay together, and even the large household groups fragmented into smaller family groups that became the primary residential and productive units. It is likely that both authors are correct. The Coast Tsimshian tribal groups had traditional, contiguous territories along the Skeena (Inglis and MacDonald 1979:3). Each tribe derived its name from these territories. For example, Ginadoiks means "people of the swift current" (Garfield 1939:176). These territories were sub-divided into resource areas that were owned by the various House groups. Inglis and MacDonald (1979:3) indicate that each tribe had a major village location in its traditional summer territory, usually at the confluence of the Skeena and a major tributary (see also Allaire 1984:88). Most of the time, however, was spent in small camps of simply constructed dwellings, smokehouses and drying racks, distributed throughout the territory. Each House owned multiple camps, and from each camp, one or two families engaged in subsistence activities. Men fished and hunted in late summer and early fall. Women processed fish, and also collected berries (Inglis and MacDonald 1979:7). A great deal of mobility could be expected of these small family units at this time, as they moved from one resource location to another. In late October or early November, when the peak of the salmon runs had passed, the third major seasonal movement occurred. The Coast Tsimshian returned to their winter villages at Metlakatla. 120 The Kitselas Model This brief description of general Coast Tsimshian settlement and annual round differs markedly from that of the Kitselas. The Kitselas wintered in the canyon (Emmons 1912:468). They also spent their summers there, collecting food resources, mainly salmon and berries. This resulted in a less mobile lifestyle than that of the Coast Tsimshian, in that there was no residential shift from summer fishing to winter village location. Allaire et a 1. (1979:70) state, "Despite their limited territories the Kitselas could depend on sufficient quantities of salmon going through the canyon each summer to live in their village throughout most of the year". The Kitselas followed "a way of life very similar to that on the coast except that the Kitselas were very much more sedentary" (Allaire et a 1 . 1979:71). There is some disagreement concerning Kitselas mobility in the spring. Allaire et a 1 . (1979:71) state that the Kitselas joined the rest of the Tsimshian at the eulachon fishery on the Nass River, "going there on foot by the trail through the Kitsumka1um Valley". McNeary (1976:117) on the other hand, states, "the Canyon Tsimshian did not depend on the Niska for eulachon oil, as they had an alternative source in the Kitimat people of Douglas Channel". It is not clear whether the Kitselas actually fished for eulachon (as implied by Allaire et a 1.), or traded for grease (as implied by McNeary). In the latter case, it is unlikely that a major spring residential move occurred. 121 More likely, only a brief expedition involving a few Kitselas traders would have been mounted. If the Kitselas did fish for eulachon, ft is unlikely that all Kitselas people moved to the Nass. The journey from Kitselas Canyon to the Nass, whether by land or by water, is much longer and more arduous than from Prince Rupert Harbour to the Nass. Likely, many people stayed home. In any event, the Kitselas were much less mobile than the Coast Tsimshian of Prince Rupert Harbour, spending most, if not all, of the year at Kitselas Canyon. The sedentary lifestyle of the Kitselas was based on an extreme dependence on salmon - perhaps more so than the other Coast Tsimshian groups. In part, this dependence was facilitated by the Kitselas' control of the canyon, a critical fishing location on the lower Skeena. Although other resources, especially land mammals and berries, were available in the vicinity of the canyon (see Tables 3.2, 3.3), the Kitselas had little or no direct access to the marine and foreshore resources that were used by other Coast Tsimshian groups. Allaire et a 1. (1979:136-37) suggest that the Kitselas may have traded for these resources. This is supported by the presence of sea mammal bones and marine shell, recovered at Gitlaxdzawk. In return, the Kitselas may have traded mountain goat wool or soapberries, both of which are more abundant on the east side of the Coast Mountains than on the west side. Herring, cod, halibut, shellfish and sea mammals were all 122 important resources to the Coast Tsimshian, but not to the Kitselas. As Emmons (1912:468) states, to compensate for the lack of these resources the Kitselas "have always looked to the river with its wealth of salmon for their chief support". Kitselas Villages and Territories At the time of European contact, the Kitselas occupied two villages in the canyon. One, Gi11axdzawk (the "Fortress"), "people of the ravine" (Allaire et al. 1970:65) or "people of the place where they steal canoe bottom boards" (Emmons 1912:469) was situated on a rocky promontory near the upper end of the canyon on the west side (see Figure 4.3, 4.4). At present, this promontory is connected to the river shore, adjacent to the CNR line. Prior to railway construction, the promontory may have been separated from the shore by a narrow channel (Allaire et a 1. 1979:72). The other village was Gitsaex, "people at the edge of the lake", located on the east side of the river at the north end of the canyon (see Figure 4.3, 4.4). Allaire et a 1. (1979:66) state, "These two sites constituted a single community, and were the only two occupied in historic times". 123 Figure 4.3 Kitselas Canyon archaeological sites 124 Origins The Kitselas, like many Coast Tsimshian tribes, claim Temlehem, a mythical village on the upper Skeena, as their place of origin (Robinson and Wright 1962). The myth states that after a famine or great flood at Temlehem some families migrated downriver, and eventually settled on the west side of the river just below Kitselas Canyon at Tsunyow. "landing place" (Emmons 1912:469). Sometime later, these original settlers were joined by new people, and according to Allaire et a 1. (1979:65), this resulted in a move across the river to Gitaus, "people of the sand bar" (see Figure 4.3). According to the myth, the Kitselas abandoned Gitaus when beavers mined the soil from under the village. This is interesting, given that the archaeological excavation of Gitaus in 1968 was prompted by the erosion of the site as a result of extensive river undercutting. In addition, there is an unstable slide area directly behind the site. These factors may have been important reasons for not maintaining a permanent village here. Allaire et a 1. (1979:65) state that after leaving Gitaus, the Kitselas established their village at Gitlaxdzawk. Gitsaex was established soon after, when overcrowding at Gitlaxdzawk caused some people to move across the river. The timing of the events described above is difficult to establish because they are based on oral history. When Emmons passed through the canyon in 1909, Gitlaxdzawk and 125 Gitsaex were abandoned. Of Gitsaex, Emmons (1912:470) states: The decayed remains of other carvings and house-timbers half-buried in the moss and overgrown with brush confirm the statement of the natives that this was the largest and most important of the villages hereabouts. In 1879, George Dawson visited Kitselas Canyon, and described Gitaus and Gitlaxdzawk as follows: Near the lower end of the canyon, on the right bank are perched a few Indian houses, with some rudely executed carved posts... At the upper end of the canyon, on the opposite or left bank, is a second and larger village with eight or ten houses, a few in good order, but most in various stages of dilapitation [sic] (1881:13) . Dawson apparently did not see Gitsaex, and the structures at Gitaus were probably associated with the white settlement that developed during the 1870's gold rush. G i 11axdzawk Gitlaxdzawk was first investigated in 1927 by H.I. Smith as part of a joint National Museums of Canada -Canadian National Railway reconstruction project. The site was not investigated from an archaeological perspective, however, until 1971 when mapping and test excavations were conducted by Allaire and MacDonald (1971). The village area covered about 100 metres (north/south) by 50 metres. The implication that the site was fortified is reflected today by its location. It is separated from the river bank by a dry gulley, which may have been a river channel prior to railway construction. The site is Figure 4.4 Gitlaxdzawk. 127 surrounded by steep slopes and its access by river or by land is difficult at all times. The village was located on top of this promontory which consisted of three terraces. Allaire and MacDonald (1971:49) described the residential pattern, at Gitlaxdzawk as unusual for a Tsimshian village. Instead of the usual row of houses facing the water, the houses were placed irregularly and oriented in various directions (see Figure 4.4) as a result of the unusual site topography. The remains of ten house floors were cleared and mapped. The general construction of the houses is described by Allaire and MacDonald (1971:50): The type of the houses corresponds to that used by the Tsimshian for the same period on the coast. They are thirty feet square with two heavy roof beams (nearly two feet in diameter) placed halfway between the lateral walls and the centre of the house. The adzed planks which supported them were of the same width and about eight inches thick and nine to eleven feet high ...the upper plank of the lateral wall was joined to the short corner posts by either mortice and tenon or lap joints. The gable planks were also fitted into the corner posts by the same techniques. In one photograph the gable planks of one house are shown with a series of notches to hold smaller rafters. Similar notches held the transverse rafters on the upper plank of the lateral walls. The covering of the walls was probably made of vertical split planks fitted into a grooved sill at both ends of the house. The covering of the lateral wa1 Is must have consisted of horizontally laid planks. Three.houses at Gitlaxdzawk were unusual (Allaire and MacDonald 1971:50) (see Figure 4.4). House 2 was a traditional structure that was covered with modern sawn planks. House 3 was large and had a raised platform at the rear, and House 7 was also very large, had a dug out 128 floor, and was built partly on piles over the cliff above the river (see Table 8.7 for house floor dimensions). The dwelling structures at Gitlaxdzawk are examined in greater detail later in this study. Gitsaex Gitsaex has not been described in detail in the literature. The site was surveyed and mapped in 1981 (MacDonald and Coupland 1982), and the following description is based on that investigation. The site is situated on a flat bench or terrace, approximately five metres above the high water line of the Skeena River. It is bounded on the north side by a shallow bay and swampy marsh, and on the southwest side by two islands at the head of the canyon. The east and southeast portions of the site are bounded by slopes that rise steeply away from the site. Surface features at Gitsaex include seventeen house floors (see Figure 4.5). These appear as flat, rectangular levelled areas, in most cases bounded by low earth ridges that were formed as a result of the levelling of the floors. Post depressions helped to define the perimeters of the floors. In a few cases, pieces of the wall footings remained in place, precisely defining the location of the structural wall. The orientation of the floors is consistent with the ethnographic village pattern. There are two rows of floors, arranged parallel to each other and facing the river. 129 Figure 4.5 Gitsaex 130 There are eight floors fn the front (west) and nine floors in the back (east) row. In addition, a cultural feature including five circular surface depressions was identified between floors 2 and 3. These were evidently post holes of free-standing columns. The entire eastern portion of the site includes numerous and varied cultural features. There is a small area (12 metres by 12 metres) located east of floors 14 and 15 that is marked by a lithic scatter (mainly fire-cracked rock). There is also an earth ramp which may have functioned as a formal entrance to the site from the east. It is located east of floor 12. Six circular cache pits are located in this area. These may have been used for food storage. Three large, flat, unmarked stone slabs, likely grave markers, were recorded, and there is an historic Christian grave stone located east of floor 17. No standing architecture remains at Gitsaex, although this condition has prevailed for only a short time. Emmons (1912:470) described standing house frames in 1909. The most significant features identified from archaeological reconnaisance at the site were the seventeen levelled floors, thought to have been formerly associated with plank house superstructures. The average size of these floors was about 10 metres by 10 metres (individual floors are examined in greater detail later in this study) which is consistent with house floor size at Gitlaxdzawk. Post depressions along the edges of most floors at the front 131 and back walls suggest the use of off-centred main beams as described by Allaire and MacDonald (1971) for Gitlaxdzawk and by Boas (1916:46-48) for the Coast Tsimshian village of Kitkatla. Pieces of wall sills at Gitsaex in association with floors 3 and 16 correspond to plank house construction techniques noted at Gitlaxdzawk (Allaire and MacDonald 1971:50). In addition to the earth ramp, there are two likely access points to Gitsaex from the river. These are small beach areas (see Figure 4.5) where canoes could be landed at the north and south ends of a large back eddy that fronts the site. The north beach is directly in front of floor 2 and the adjacent small open area with the five post depressions. The south beach fronts floor 6. Gitsaex was a large village occupied during the early historic and perhaps late prehistoric periods. The house floors, arranged in two rows, are spaced at roughly equal distance (see Chapter 8), and do not overlap. It is reasonable to infer from this that most if not all floors were used simultaneously during the peak occupation of the village. This was probably in the late 18th or early 19th century. Our Kitselas informant stated that after Gitsaex was abandoned as a village, it continued to function as a burial site. This is substantiated by the presence of a late 19th century Christian tombstone on site. In addition, a small cemetery with seven graves is located on the upper terrace east of Gitsaex. 132 Terr i tor i es There is some confusion concerning the territories held by the Kitselas (Figure 4.6). Allaire et a 1. (1979:70-71) simply state that Kitselas territory extended "some thirty miles along the Skeena", and was bounded above the canyon by Gitksan territory and below by the Coast Tsimshian. Inglis and MacDonald (1979:3) indicate that Gitlan territory included the area around the modern town of Terrace and the lower Zymoetz (Copper) River, and Gispakloats territory extended beyond this, virtually to the foot of the canyon. Inglis and MacDonald's map, based on the fieldnotes of Wallace and Beynon, does not indicate the extent of Kitselas territory, but it is implied that the Kitselas did not control the river below the canyon, or that their claim to this territory is disputed. Emmons (1912:468) states that the Kitselas "held the canyon, claiming the river from Lome Creek above to the little canyon below". Lome Creek, a tributary of the Skeena is 40 km upriver from Kitselas Canyon. Little Canyon is at Terrace, 16 km below Kitselas Canyon. This conflicts with Wallace and Beynon's notes, especially concerning Gitlan and Gispakloats territories. The river between Terrace and Kitselas Canyon today is regularly fished by members of the Kitselas Band, but the situation may have been different in the early historic or prehistoric periods. Boas (1916:381) refers to a Gispakloats group who camped one summer "below the canyon at G'at-aus (Sandy Camp)", 133 Figure 4.6 Kitselas territories of the Skeena Valley. 134 where they dried salmon and other provisions. It is not stated whether the Gispakloats actually owned the camp, or simply received permission from the Kitselas to use it one summer. The upriver extent of Kitselas territory to Lome Creek is not disputed. This area is important because it is substantially drier, less rugged and more open in forest coverage than the area around Kitselas Canyon. As stated in Chapter 3, this facilitated access to a variety of resources, both floral and faunal, that are either non-existent or not readily available near the coast. Subsistence Technology and Material Culture The subsistence technology and material culture of the Kitselas are poorly documented. The following review is based on information culled from ethnographies of the Tsimshian and Niska, and is thought to be generally applicable to the Kitselas. Fishing Technology The Kitselas were river fishermen, and as such were likely to have used all river fishing techniques known to the Tsimshian. Prior to European contact, their principal food procurement technology included dip nets (Boas 1916:400; Drucker 1950:169, 239; Nolan 1977:135-36; Allaire etaK. 1979:71), traps (Boas 1916:50; Drucker 1950:166-67; Nolan 1977:140-42; McNeary 1976:93-95;) and weirs (Drucker 135 1950:166-67; Nolan 1977:140-41; McNeary 1976:93). According to Drucker (1950:166-67) and Nolan (1977:135-36), the dip net was the only netting technology used aboriginally by the Tsimshian. The Tsimshian dip net consisted of a hoop, manufactured by bending and lashing together two arms of a natural wooden fork. The net was attached to the hoop and the hoop was then attached to the end of a long pole (Boas 1916:400). The net itself was probably made of stinging nettle fibres or sinew (McNeary 1976:74). Duff (1952:63) states that dip-netting is most effective for sockeye, which swim close to the surface. Allaire et a 1. (1979:71) suggest that dip-netting was the Kitselas' principal fishing technique. However, our informant stated that the use of dip nets was secondary to wooden traps or baskets. The Kitselas trap was similar to one described by Drucker (1950:166-67). Essentially, it was a large cylindrical basket, weighted in the bottom with rocks so that it was submerged below the surface. The trap could be used with or without a funnel entry. The funnel opened downstream, and fish caught therein would be forced into the trap. Most fishermen operated numerous traps of this type simultaneously in different locations. Another trap that may have been used by the Kitselas -is the type described by McNeary (1976:93-95) for the Niska: A fence was built part way out into the river on a frame of stakes driven into the river bed. This fence channelled the fish into the trap. Apparently the trap forced the salmon to ascend to the surface of the water where they met a barrier that was so constructed 136 that as the Fish tried to jump they were knocked backwards into a tray or bin above the water. Morice (1893:89) described a similar trap used by the Carrier where rivers were deep and swift. It is not clear to what extent the Kitselas used weirs. Nolan (1977:140-42) indicates that weirs were a very important Tsimshian fishing technique, used in shallow spawning streams. Although these would have been ineffective in the canyon where the river is deep, Kitselas territory included a number of shallow spawning streams, including Kleanza Creek at the foot of the canyon, and Chimdemash, Legate, Oliver and Lome Creeks, all above the canyon. They may also have fished Zymoetz River. Weirs may have been constructed on any of these tributaries. The fish were probably speared as they congregated below the weir. Nolan (1977:133) describes single and double-headed salmon harpoons, and two-pronged leisters used by the Coast Tsimshian. The Niska used a leister with two detachable toggle heads, lashed together with root and barbed with bone (McNeary 1976:95). Dip-netting is a mobile technology. It allows the fisherman to move up and down the river and use more than one dip-net location. Traps and weirs are fixed technologies, associated with permanent fishing sites. The most productive of these locations often had processing stations located nearby (McNeary 1976:95), so that the fish could not only be harvested in large numbers, but they could 137 also be efficiently processed for storage. The Tsimshian preserved fish by means of smoking. The lower Skeena climate is generally too wet to allow the wind drying techniques used in other places such as the Fraser Canyon. Today, most smoking is done in smokehouses. Formerly, low open fires and frame racks were used. Rotten cottonwood (McNeary 1976:97) or alder (Drucker 1965:118) was used for smoking salmon. Fish were generally processed by women. The fish was cleaned and the head and tail were removed. The body of the fish was then spread out and the flesh scored. The fish was smoked for a few hours until firm, and then the flesh cut into strips to be fully smoked (McNeary 1976:97). Fish were smoked for winter consumption. Coho salmon could sometimes be caught into December (McNeary 1976:96). For the remainder of the winter the Tsimshian, and particularly the Kitselas, subsisted mainly on smoked salmon and dried berries. Storage The Coast Tsimshian of Metlakatla stored their food inside their winter dwellings, in smokehouses, or occasionally in cabins constructed for storage (Garfield 1951:11). Within dwellings, food was stored in boxes which were sometimes kept in underground pits beneath the sleeping benches (Garfield 1951:10), or stacked high against the house walls to form partitions for individual family space 138 (Drucker 1965:120). Garfield (1951:10-11) also states that "dried foods were stored on shelves suspended from the beams and rafters or hung in bundles from the beams". There is no mention of food storage in pits outside the winter dwellings. This is perhaps the most basic of storage techniques, but at Prince Rupert Harbour winter precipitation is heavy and usually in the form of rain, so it is likely that food stored in outdoor pits would quickly spoil. Among the Niska, however, McNeary (1976:90) states that "fish were stored in pits near each house". This was in addition to the use of boxes (McNeary 1976:80). MacDonald (1979:69, Figure 29) indicates that the Gitksan also made extensive use of food storage pits. The hillfort at Kitwanga has a number of food storage pits at the base of the hill. Some of these are over 200 metres from the residential area of the site. Although it is not documented, the Kitselas also stored food in pits. MacDonald (pers comm, July 1981) stated that food pits were lined with birch bark or skunk cabbage leaves. The pits were often located outside the immediate area of the village, perhaps as a deterrent to raiding. In part, the use of food storage pits by the "upriver" Tsimshian was facilitated by the fact that winter temperatures are usually at or below freezing, and precipitation is in the form of snow. Coho salmon and steel head trout can both be taken late in the year, and so are particularly suited to this type of storage. 139 Hunting and Trapping Allaire et al. (1979:71) state that the Kitselas hunted and trapped in the winter. Hunting and trapping was important for the Gitksan (Cove 1982:4-5) and for the upriver Niska (McNeary 1976:101). It is likely that the same was true for the Kitselas, at least relative to other Coast Tsimshian groups who had access to marine and foreshore resources. The area around Lorne Creek, at the upriver end of Kitselas territory, was probably one important hunting area. The sub-alpine and alpine zones were also important for hunting marmot and mountain goat. McNeary (1976:101) states that land mammals were more frequently trapped than hunted by the Niska. The Niska used two types of traps; the deadfall and the snare. Both varied greatly in size, and were suitable for catching animals ranging in size from rabbits to bears (McNeary 1976:103). Hunting equipment included the spear and the bow and arrow. McNeary (1976:102-03) states that mountain goats, highly prized for their meat and wool, were hunted with spears by the Niska. The hunting season began in September (mountain goat) and generally continued through the winter, depending on snow conditions and the availability of the resource. Very little hunting and trapping were done during the summer when fishing was of primary importance. 140 Plant Gathering Like hunting, plant foods supplemented the essentially fish oriented diet of the Kitselas. Spring through fall was the gathering season. Among the Niska, this was primarily women's work; men fished and hunted. In the spring, hemlock cambium was edible when the sap was running. Most edible greens were available from April through June. These included stonecrop, cow parsnip, lady fern and Kamchatka lily. The Niska roasted fern roots in earth ovens, and the Kitselas may have done the same, although no earth ovens were recorded during the survey of Kitselas Canyon. Berries, including wild crabapple, were collected in late summer and early fall. These were probably more important to the Kitselas than greens because they could be preserved for winter consumption. Blueberries, huckleberries, soapberries and salal were all available in Kitselas territory. The Niska crushed these berries and spread them over skunk cabbage leaves under which a low fire burned. When the mash dried, it was rolled up and stored for later consumption (McNeary 1976:109). Soapberries do not grow in the coastal environment, but they do grow in the upriver end of Kitselas territory, near Lome Creek. They are considered "chief's food" among the Niska, who traded for them with the Gitksan (McNeary 1976:109). Plants were not only important for food. Stinging nettle was spun into twine for nets and snares, and birch 141 bark and skunk cabbage leaves were important for wrapping food. Birch bark was also used to make containers, and spruceroot was woven into baskets (McNeary 1976:75-76). Alder and cottonwood were used as fuel. Finally, a discussion of the plant use habits of any Northwest Coast group would be incomplete without mention of the importance of cedar. Tributes to cedar have been made on numerous occasions elsewhere (H. Stewart 1984; Hebda and Mathewes 1984), so I will be brief here. The Kitselas used cedar to construct their dwellings, manufacture canoes, and make containers (boxes and baskets), rope, clothing, mats, personal adornment and monumental art. Boas' (1916:46) general ode is appropriate as well to the Kitselas: "It may be said that the salmon and cedar are the foundations of Northwest Coast culture". Demography Population estimates for the two historic Kitselas villages have been reviewed by Allaire et a 1. (1979:67-68) and Allaire (1978:28-29). There are four estimates pertaining to the late 19th century, before the complete abandonment of the viilages. These range from 186 (Kane 1859: Appendix I) to 1000 (Emmons (1912:468). Allaire et a 1. (1979:68) favour the lower of these two estimates, arguing that at a time when European diseases were ravaging aboriginal populations in North America, an estimate of 1000 people at Kitselas Canyon seems inordinately high. 142 In fact, all four estimates are neither useful nor reliable. Kane's and Schoolcraft's estimates were the only ones based on census data (taken in 1845). Kane reported 72 men, 49 women, 29 boys, 31 girls and 5 slaves. Apparently, there was some discrepancy over these numbers because Schoolcraft used the same data, and his figures were 82 men, 47 women, 29 boys, 31 girls and 5 slaves. Adding to the confusion is the fact that the census was probably not taken at Kitselas Canyon. Allaire et al. (1979:66) state that the first white men on the Skeena did not arrive at Kitselas Canyon before 1859. The census was probably taken from Fort Simpson on the basis of native estimates. The inaccuracy of this data is further seen in the ratio of adult males to females. It approaches 2:1. This discrepency is not accounted for In any of the estimates, and I think it contributes to the unreliability of the data. In addition, the number of dwellings in the census was estimated at twenty. Allaire et a 1. (1979:67) state that this gives an average of 100 people in ten houses per village (Gitlaxdzawk and Gitsaex), or ten people per dwelling. This is extremely low by most Northwest Coast standards, where household size is usually estimated at 20 to 25 people (e.g., Donald and Mitchell 1975:333; McNeary 1976:128). Fort Simpson was established in 1834, and it is reasonable to assume that there was a steady population decline from this time, as more and more families moved to the coast. Therefore, the 1845 census, which indicates many unoccupied 143 dwellings and others occupied by only small numbers of people, probably reflects a depleted population at Kitselas Canyon. In 1879, Dawson (1881:20) estimated the Kitselas population at 300 in eight or nine houses. He cautioned that these figures are "uncertain owing to the number of people away at the coast and elsewhere, yet calling this place their home". Thus, the number could have been substantially higher than 300, and in fact probably was, given that Dawson did not see Gitsaex when he passed through the canyon. Dawson's figure was used by Allaire and MacDonald (1971:48) in their population estimate of Gitlaxdzawk. This estimate gives about 30 to 35 people per dwelling (not counting absentees), which is higher than the usual estimate of 20 to 25 people. It is unfortunate that no population estimate exists for Kitselas Canyon prior to 1834. We are left to speculate about what the size of this population may have been. If Dawson's estimate is brought in line with the ethnographic model (20 to 25 people per dwelling), the Gitlaxdzawk population may have been 200 to 250 people. But given the statement by Allaire et_ a 1 . (1979:65) about overcrowding at the site, it is possible that Dawson's estimate is close to being accurate. A conservative estimate would be 250 people. Gitsaex was a much bigger village in terms of number of house floors (17 compared to 10 at Gitlaxdzawk). Conservatively, the maximum Gitsaex population may have 144 been 50% larger than at Gitlaxdzawk. This gives an estimate for Gitsaex of 375 people, making a total of 625 people for both villages. Social Organization The social organization of the Kitselas is consistent with that of other Coast Tsimshian groups. The Coast Tsimshian are divided into four units, called phratries (Garfield 1939:173). They are Laxg i bu (wolf), Laxski k (eagle), Ganhada (raven), and G i spawedwada .(blackfish or killer whale). Phratries are sub-divided into clans, non-1 oca 1ized matrilineal descent groups that share properties and privileges (Garfield 1951:20). The localized segment of a clan - the people who functioned as a group - was a House or lineage (Garfield 1951:22). Theoretically, each phratry could be represented in each tribe or winter village group by a lineage. For example, Allaire ejt a 1 . (1979:68, 111) state that at Gitlaxdzawk there were Eagle, Blackfish and Raven Houses. There were no Wolves at Gitlaxdzawk, but there may have been a Wolf House at Gitsaex (Allaire et al. 1979:68). The lineage or House was the autonomous socioeconomic unit of Tsimshian life. Villages were local units consisting of various families, some related, some not (Barbeau 1917:403). They functioned as political units (with contiguous territories, a name etc. - Mitchell 1983b) although not with the same level of integration as Houses. 145 The House was a corporate unit that held title, in the sense of proprietory rights, to lands and resources (Garfield 1951:14). The importance of House, power and property has been analyzed by Cove (1982). The analysis pertains to the Gitksan, but it has applicability to the Kitselas and Coast Tsimshian as well. Briefly, a House held title to a territory because it had merged its essence with that piece of land. A House's essence was its stock of supernatural powers acquired by ancestors of the House from supernatural beings (naxnox) (Cove 1982:7). Supernatural beings were identified with specific locations (spanaxnox). A person entering that space could acquire powers from the being, and in so doing could become a "real person" (semooget). Acquired powers became part of the House and were given life by it (Cove 1982:7). The locale (spanaxnox), and its resources, also became part of the House. The complex of crests, names, myths and songs stood in relationship to a House, its powers and territories. In one way or another, they expressed the powers of the House. For example, crests stand for the form taken by a supernatural being when it was encountered (Hal pin 1973:107). Typically depicted on poles, house posts and dancing blankets, crests publicly stated a House's stock of powers (Cove 1982:9). Carved poles were often complete statements of acquired powers. Placing a pole in the ground was a means of putting a House's powers into it, acting 146 as a "deed" to a territory (Duff 1959:12; Cove 1982:9). Names also express power. For example, "real" names, a political extension of chieftainship, denote the ancestor who acquired powers from a supernatural being. These names were given to acknowledge that encounter and the subsequent transformation to real status. Myths and songs were historical statements about how, when and where powers were received. Through oral tradition, a House preserved and legitimized its powers and titles to territories (Cove 1982:11). In a sense, the property of a House was not the territories that it used, but rather the complex of crests, names and oral traditions that expressed the exclusive rights of access (title) of a House to the territories. These expressions or evidence of title were formally presented at potlatches or feasts. Feasts provided a public forum for the host House to validate newly acquired or existing powers, their origins and associated rights in property (Cove 1982:12). Guests at feasts were witnesses to the host's claims. The giving of gifts to guests was in part payment for their witness, but also an endorsement of the successor to title and a demonstration of his or her worth i ness. The Kitselas conformed to Fried's (1967:109) definition of a ranked society. Within each House, the members were ranked according to primogeniture. Status differences were ascribed. Garfield (1939:177) states that this resulted 147 in a virtual caste system among the Tsimshian. At the top were chiefs (semooget) and close members of their lineage who collectively constituted Tsimshian nobility. These people held ranking names and usually lived in the same dwelling. The name of the dwelling and the name of the chief were usually the same. The wealth of the House, in the form of slaves, accumulated food stores and manufactured goods, was controlled by the chief. Chiefs held high ranking names, enjoyed prominent seats at pot latches, and had the right to distribute House-owned wealth. Chiefs made decisions about resource use, but their decisions were not binding; a chief could not deny another House member access to subsistence resources (Drucker 1939:58). Next in rank to chiefs and nobles were common people (lekayiget). Garfield (1939:178) states that these people formed the bulk of Tsimshian population. They included people who assumed hereditary personal names, but not "real" names. While they were House members, and related to the chief, they were not in the direct ancestral line of the chief. They typically lived in separate dwellings, near that of the chief, and the senior or most important individual in the dwelling assumed the name of the dwelling. For example, at Gitlaxdzawk, the Eagle chief was Gitxan. but other Eagle dwellings included N i eskadek, N i eswi txo, Iyoos, and Tacolepl i p (Allaire et a 1 . 1979:68). Commoners formed the bulk of the labour force in 148 Tsimshian society. According to Garfield (1939:178), their lot in life was to "help their chief" because generally they were unable to acquire property to initiate a pot latch for themselves. At the lower end of the social scale were slaves and their families. If there was a class distinction in Tsimshian society, it was between freemen (chiefs, nobles, commoners) and slaves. Slaves, typically captives of raids or wars, held no rights in property and had no status. They were often given as pot latch gifts. Early research on Northwest Coast slavery held that slaves were not functioning members of society (Drucker 1939; Codere 1950). Recent research, however, has stressed the economic importance of slaves in everyday life (Donald 1983:114; Ruyle 1973:610-11). Slaves provided cheap labour, and a House that owned slaves could increase production at low cost. Mitchell (1984:45) states that slaving was a "business", and predatory warfare for capturing slaves was a "business venture". Slaves were viewed as wealth, and through the pot latch, this wealth could be transformed into prestige (Mitchell 1984:45). It is conceivable that much of the warfare and raiding that was endemic historically on the Northwest Coast was aimed at acquiring slaves as a means of increasing social standing and productive effi ci ency. 149 Summary This ethnographic summary of the Kitselas has Introduced three salient points. First, the Kitselas were permanent, yeai—round occupants of the canyon. Second, their subsistence was based on the intensive production of salmon. Third, the Kitselas were part of a ranked society, as defined by Fried (1967). Status differences were ascribed from birth, and there were only a limited number of positions of high status (semooget) within society. Although chiefs did not technically own resources - these were controlled by the corporate lineages - it is clear that chiefs had unequal access to resources (food, wealth, and prestige) through their central position in redistributive exchange. In subsequent chapters, the prehistoric evidence from Kitselas Canyon will be compared to this model of the Kitselas to determine the nature of prehistoric social i ntegrat i on. 150 Part 2 REGIONAL CULTURE HISTORY Early Evidence This section investigates the existing culture history of the lower and middle Skeena Valley, and reviews models of prehistoric cultural change that have been proposed for the area. The earliest evidence of human settlement in this area is ca. 5000 B.P. Radiocarbon dates of this age have been obtained from cultural components in two localities; Kitselas Canyon (Paul Mason Site, GdTc 16) (Coupland 1984:61), and Prince Rupert Harbour (Dodge Island, GbTo 18; Kitandach, GbTo 34; Ridley Island GbTn 19) (MacDonald and Ing1i s 1981:43). At Hagwilget Canyon (GhSv 2), Ames (1979a:202) has inferred an initial date of occupation of 4500 to 5000 B.P., based on the radiocarbon date, 3430 ± 200 B.P. (GSC 746), obtained from a test excavation by MacDonald (1969a:249). This was not a basal date, but Ames has suggested an initial occupation 1000 to 1500 years earlier, based on rate of sedimentation. It is significant that there is no dated occupation in excess of 5000 years in the region. Elsewhere in northwestern British Columbia (and Alaska and Yukon), there is evidence of human occupation 3000 to 5000 years earlier than in the Skeena Valley. Fladmark (1982a:109) suggested that the absence of early Holocene prehistoric components at the mouth of the Skeena "probably reflects lower sea levels... and the 151 submergence of most inner coastal sites". However, this does not explain the absence of pre-5000 B.P. components at Kitselas Canyon or at Hagwilget Canyon. Moreover, it is not clear that sea levels were lower in Prince Rupert Harbour during the early Holocene. Elsewhere, Fladmark (1975:152, 1983:72) suggested that prior to 5000 B.P. north coast sea levels were actually higher than at present, and in Prince Rupert Harbour, sea levels may have stood 40 metres higher than at present about 7500 B.P. (Fladmark 1975:151). Following this argument, if early Holocene sites exist in the area, they would be located above current sea levels. It will be recalled from Chapter 3 that Clague's (1984:3) research in the Kitimat/Prinee Rupert Harbour area, the most recent study to date, indicates only minor sea level fluctuations during the last 8000 years. There has been no shortage of archaeological investigation in the Prince Rupert Harbour region during the last 20 years. Yet after recording nearly 200 sites, eleven of which were excavated, and obtaining 121 radiocarbon dates from the area (see MacDonald and Inglis 1981:41), the earliest evidence for human settlement of the area is 5000 B.P. Early settlement elsewhere in northwestern British Columbia is nonetheless important, and is reviewed here as a prelude to the discussion of the post-5000 B.P. settlement of the lower Skeena area. 152 The Late Pleistocene: the Question of Origins There Is disagreement between the earliest evidence for human occupation of the northern Northwest Coast, and the earliest possiblity for man in the region. The earliest evidence, based on radiocarbon dated cultural components, is between 9000 and 8500 B.P. The earliest possibility, according to some (e.g. Fladmark 1975, 1979b) is at least 2000 to 3000 years earlier, and assumes the existence of pre-Clovis man in North America. Charles Borden (1968a, 1969, 1975) argued that the antecedents of northern Northwest Coast culture were to be found in the late Pleistocene assemblages of south central Alaska and southwest Yukon. Borden thought that the early assemblage at the Healy Lake Site (see Cook 1969), stratigraphica11y earlier than a radiocarbon date of 11,090 i 170 B.P., was similar to later assemblages recovered on the northern Northwest Coast. This early Alaskan assemblage was characterized by well-established microcore-and-b1ade and burin industries. Borden (1975:12) included the Healy Lake assemblage and virtually all other late Pleistocene assemblages from the Yukon River drainage in his "Early Boreal Tradition". While recognizing local and regional specializations over this broad area, Borden (1975:12) generalized that: these northern late Pleistocene - early Holocene groups possessed a flexible culture, including subsistence techniques capable of being adapted to the efficient exploitation of a wide range of ecological niches, from steppe and tundra to lake and riverine 153 environments. Further... when opportunity or need arose some of their interior utilization techniques were readily adaptable to the exploitation of coastal resources". The presence of early microblade-bearing assemblages in central Alaska and southwest Yukon prompted Borden (1969:5) to argue that coastal settlement was from the interior via major river valleys. This model implies initial settlement of the northern coast no earlier than 10,000 B.P. because river valleys would not have been ice-free prior to this time. Current dates from archaeological sites on the north coast support this argument (see Chapter 1, and this chapter). An alternative model to Borden's, presented by Fladmark (1975, 1979b), posits the coast as a migration corridor at a time of lowered sea levels during the late Pleistocene. Fladmark argues that settlement of northwestern British Columbia and southeastern Alaska was from the coast rather than the interior. This model, based entirely on pa 1eoenvironmenta1 evidence, hinges on the presence of a chain of ice-free refugia, a narrow migration corridor between the ice-front and salt water. To date, Fladmark's research has been aimed at demonstrating that suitable conditions existed for Pleistocene human migration along the coast. This has become a point of contention in recent years. In addition to the fact that there is no archaeological evidence to support the model, Mathews (1979:150) argues that at the time of the glacial climax (17000 to 13000 B.P.), the ice front was in contact with 154 the open ocean over much of the distance between Cape Flattery and southeast Alaska. Mathews (1979:150) argues that the benthic fauna of the shore zone was eradicated at this time, and the food sources of inland refugia were meagre. There would also have been the difficulty of "leap-frogging" over broad stretches of barren ice from one refugia to the next. At present, if the "coastal route" model is to be accepted, two lines of evidence must be forthcoming. First, there must be some consensus among Quaternary researchers that the environment of the Northwest Coast was habitable during the late Pleistocene. Secondly, the model requires clear archaeological evidence in the form of radiometrica1 Iy dated components that pre-date the period of river valley deglaciation that opened migration corridors from the northern interior to the coast. The Ear1y Ho1ocene: Sett1ement and Subs i stence A variety of terms has been applied to the early Holocene culture of the northern Northwest Coast, including the "Early Boreal Tradition", the "Microblade Tradition" (Carlson 1979), and the "Early Coast Microblade Complex" (Fladmark 1975). Despite the variation in names, there is general agreement that the early occupants of the north coast looked to the sea for subsistence. Carlson (1979:224) states, "the culture of the early Microblade Tradition people was likely specialized in the areas of fishing and sea 155 hunting". In reviewing the early Holocene culture from the Queen Charlotte Islands, Borden (1975:23) states, "the small size as well as the location of the sites suggests an early maritime estuarine adaptation by small transient groups". This interpretation is in need of review. It is based on meagre evidence at best, and is directly at odds with the fauna 1 data, where it exists. The components in question are from a handful of sites on the Queen Charlotte Islands, the southeast coast of Alaska, and the central coast of British Columbia. The assemblages are typically small, and are characterized by microblades, a variety of unifacially retouched flakes, cobble cores and tools, rare abraders and rare notched or ground sinker stones. Some assemblages have poorly formed leaf-shaped bifaces and unifaces, but Fladmark (1975:238-41; 1982a:109) states that these are rare and do not occur in all assemblages. Organic artifacts (bone, antler, wood) were almost certainly part of this early coastal complex, but they were not preserved (Fladmark 1982a:109). Also important is the absence of shell in these early components. On the Queen Charlotte Islands, early lithic assemblages were recovered from three sites, each located on raised beach or estuarine deposits: Kasta and Lawn Point (Fladmark 1971), and Skoglund's Landing (Fladmark (1970). Kasta and Lawn Point form the basis of the Moresby Tradition of the Queen Charlotte Islands sequence (Fladmark 1975, 1979a). 156 Skoglund's Landing, which lacks radiocarbon dates and microblades, may also be a component of this tradition. Farther north, early Holocene lithic assemblages, technologically comparable to the Moresby Tradition, have been recovered at Groundhog Bay 2 (Ackerman 1968; Ackerman, Hamilton and Stuckenrath 1979), and at the nearby Hidden Falls Site (Davis 1979). These latter two coastal sites, with radiocarbon dates of ca. 8500 B.P., have the oldest dated microb1ade-bearing assemblages on the north coast, and indicate that the Early Boreal Tradition had reached the coast by the early Holocene. Moving south to the central coast, the early assemblages from Namu (Carlson 1978, 1979; Hester and Nelson 1978) are included by Fladmark (1982a:108) in the early coastal complex. The relevant assemblages here are from stratigraphic units I la and lib, for which radiocarbon dates range from 9700 to 5100 B.P. The presence of a faunal assemblage from zone lib provided the first direct evidence of subsistence for an early coastal component. The assemblage includes elements of land mammals (deer, dog) sea mammals (seal, sea lion, sea otter, dolphin) and fish (salmon). Although analysis of this data was based only on raw counts of bones, Conover (1978:96) indicates that the coastal forest was the habitat most heavily exploited at this time. In general, the evidence cited above provides no clear support for a marine specialist economy on the north coast 157 during the early Holocene. Borden (1975:24) admits that "evidence for fishing is weak" at Kasta and Lawn Point, and "direct evidence for the taking of sea mammals has not yet been found". Borden's interpretation of the Moresby Tradition, as a marine-estuarine adaptation, is based solely on the location of the Kasta and Lawn Point sites (on raised beaches). No bone harpoons or fish hooks were recovered. This may be a result of poor organic preservation, but this does not explain the paucity of abraders, which should be numerous if bone fishing and sea hunting tools were manufactured. It is difficult to generalize about the economy of a cultural tradition on the basis of two small site components. A marine-estuarine focus may have existed at Kasta and Lawn Point (although this not clear), but other, potentially larger Moresby Tradition components may be located away from raised beaches, and may have little to do with a marine-estuarine adaptation. The early Holocene "sea hunting" hypothesis is biased because only coastal sites have been investigated. Moreover, at one coastal site - Namu — the fauna1 evidence suggests that land mammal hunting was an important aspect of the subsistence pattern until at least 5000 B.P. (Conover 1978:96). Following Borden's model of the Early Boreal Tradition, the early emigrants to the north coast (ca. 10,000 B.P.) had a generalized economy that included hunting and fishing. 158 Upon reaching the coast, adaptation to the new environment was made, specifically to marine resources. However, the land mammal hunting base continued to be important. It wi11 be seen shortly that land mammal hunting continued to be an important aspect of the early economy (5000-3500 B.P.) at Prince Rupert Harbour. Prehistory of the Lower Skeena The preceeding section briefly outlined the late Pleistocene and early Holocene cultural sequence in northwestern British Columbia and adjacent areas, leading up to the earliest known settlement of the lower Skeena Valley/Prince Rupert Harbour area. This sets the framework for investigating cultural developments in the lower Skeena area after 5000 B.P. This section critically examines the archaeology that has been conducted in the latter region (since the mid-1960's) , and reviews the relevant existing data on which our study of social change is based. Three localities are considered here: Prince Rupert Harbour, Kitselas Canyon and Hagwilget Canyon. Prince Rupert Harbour The earliest and most intensive archaeological investigations along the lower Skeena were conducted at Prince Rupert Harbour. These included excavations in 1967 and 1968 at Garden Island (GbTo 23), Dodge Island (GbTo 18), Parizeau Point (GbTo 30), the Boardwalk Site (GbTo 159 31), Grassy Bay (GbTn 1), and Lucy Island (GbTp 1) (see Figure 4.7). The first archaeological excavation at Prince Rupert Harbour was actually conducted in 1954 by James Baldwin at the Co-op Site (GbTo 10), but was not reported until much later (Calvert 1968). MacDonald's (1969a) original synthesis of culture history at Prince Rupert Harbour was a three-horizon scheme, in which he designated lower (4500 to 2500 B.P.), middle (2500 to 1500 B.P.), and upper (1500 to historic period) horizons. Subsequent investigations led to an alteration of the sequence. The term "period" was substituted for "horizon", and the dates were extended (MacDonald and Inglis 1981). The revised sequence is: Period III (5000 to 3500 B.P.), Period II (3500 to 1500 B.P.), and Period I (1500 to historic period). The revision was based in part on post-1969 excavations at Boardwalk, Parizeau Point, Lachane (GbTo 33) (Inglis 1974:27-31), Kitandach (GbTo 34) (Inglis 1972:101-05), the Baldwin Site (GbTo 36) (Inglis 1974:27-31), and K'nu (GcTo 1) (Inglis 1972:101-05). The earliest radiocarbon dates from Dodge Island, Kitandach and Lachane approach 5000 B.P., and the evidence for rapid shell midden accumulation, which marks the beginning of Period II, was dated about 3500 B.P. The basic theme of MacDonald's reconstruction - one of basic cultural continuity with a net accretion of cultural elements through time - remained unchanged. MacDonald (1969a:242-43) stated that all artifacts relating to economic 160 Figure 4.7 Prince Rupert Harbour sites. 161 and manufacturing activities show continuity from the earliest to latest periods. The differences among the periods are essentially based on changes in the frequency of artifacts, rather than their presence or absence. For example, the lithic industries, including chipped stone, ground slate and pecked stone, all peak in frequency during early Period II (ca. 3500 to 3000 B.P.) (MacDonald and Inglis 1981:45), but are present in other periods. Basic tool forms are said to have overwhelming continuity throughout the sequence (MacDonald 1969b:259). This is also true of the cobble and flake tools (choppers, cores, cortex spalls) which are present in abundance in each of the three periods. Generally, the assemblage became more diverse through time, with new tool forms (adzes, net sinkers, socketed bone points, bone scrapers, bark shredders and mauls) gradually added. The only significant deletion was bilaterally barbed bone harpoons in Period I. A weakness of MacDonald's model is that, to date, quantitative supporting evidence is generally lacking. Ames' (1976) study of the bone tool assemblage from the Garden Island Site, occupied during the middle and late periods, offers partial support. A factor analysis of the bone tools from the site indicated little temporal variability in bone tool form or function (Ames 1976:63-92). Despite the lack of quantitative analysis, MacDonald (1969a) effectively dispelled the old argument for a late Tsimshian migration to the coast. Many of the lithic and 162 shell tools, thought by Smith and Drucker to be representative of developed Coast Tsimshian culture, are shown by MacDonald to be present from at least the middle period of occupation. These include ground slate daggers, stone clubs, ground slate pencils, mussel shell knives and zoomorph i c art. The addition of new artifact forms in the late period is argued to represent ongoing cultural change (MacDonald 1969b:259), not population displacement. These additions (splitting adzes, mauls, bark shredders, stone bowls, shell rattles, bone combs and pins) were clearly not the result of downriver migration, as Smith and Drucker believed, because none of these artifacts have been recorded in excavations upriver at Gitaus (A11 aire 1978), the Paul Mason Site (Coupland 1984) or Hagwilget Canyon (Ames 1979a). Indeed, MacDonald (1969b:259) states that on the north coast there was never any great influx of new traits that might be suggestive of discontinuity. In MacDonald's view the only important addition to the late period was the classic art style, and this has been shown to have earlier antecedents (MacDonald 1983:101). To date, most of the analysis of the Prince Rupert Harbour material has centred on the artifacts. Studies related to prehistoric settlement and subsistence have, unfortunately, been more cursory. The most complete reported fauna 1 analysis comes from the first site excavated at Prince Rupert Harbour, the Co-op 163 Site (Calvert 1968). Recently, this site was re-designated as part of the Lachane Site (MacDonald and Inglis 1981:43). Calvert identified three stratigraphic zones at the site. The lower zone (I) comprised dark greasy soil and a small amount of mussel shell. This is consistent with the matrix of Period III components as described by MacDonald and Inglis (1981:42-45) from other Prince Rupert Harbour sites (Boardwalk, Dodge Island, Lachane). A radiocarbon date of 3040 + 100 (Gak 1477) from Zone I is somewhat late for an early period date (but not overly so), and as it was the only sample dated from the site, it should not be given undue importance. Zone II comprised ash and humus, and was culturally sterile. This may represent abandonment of the site during Period II, because Zone III, a 1.5 metre deep shell deposit, includes splitting adzes and hafted mauls, which are diagnostic artifacts of the late period. Although the fauna 1 assemblage from Co-op was small, and collected by uncertain excavation techniques, two points are important. One was the obvious increase in the use of shell after Zone I. The other point is the apparent emphasis placed on land mammal hunting in the occupation represented by Zone I (as compared to Zone III). Whereas sitka deer was present in abundance in both zones, elements of mountain goat, wapiti, and smaller land mammals were restricted to Zone I. Stewart (1974:25) noted a similar predominance of land mammal elements from the lower levels (Period III) of the Boardwalk Site. 164 Sea mammal remains, recovered from both zones at Co-op, attest to a long-standing maritime adaptation. Fish bones, including salmon, were recovered, but not in great quantity. This low recovery is probably due to the excavation of one of the test pits without the use of screens. Calvert (1968:96) concluded that in addition to sea mammal hunting, fishing, and some shellfish (blue mussel) collecting, land mammal hunting was an important economic pursuit to the early occupants at Co-op. Given a similar pattern at the Boardwalk Site (Stewart 1974), it is unlikely that the predominance of land mammal elements at Co-op was merely a factor of site function. A reasonable inference is that the Period III inhabitants of Prince Rupert Harbour had a generalized economy, in which both terrestrial and maritime fauna were of major importance, and shellfish was of minor importance. This represents a continuation of the early Holocene subsistence pattern of the north coast. The question of seasonality at Prince Rupert Harbour is more difficult to address. Wapiti elements in Period III components suggest winter occupation, which is consistent with the late prehistoric and ethnographic settlement pattern. The paucity of salmon elements may be due in large part to excavation technique. Alternatively, it is possible that salmon fishing was conducted upriver in summer and fall (as done ethnographica11y), which again implies winter use of Prince Rupert Harbour. 165 During the early period, only blue mussel (Myti1 us  ca1i fornianus) was present in large amounts in the middens. At the onset of Period II, there was "rapid midden build-up" (MacDonald and Inglis 1981:45). In addition to blue mussel, cockle (C1 i nnocard i urn nutta1 i), butter clam (Saxidomus  g i gantus), horse clam (Tresus capax), and sea urchin (Strongy1ocentrotus drobach i ens i s) were all present in the middens in large amounts. This represents new and intensified use of foreshore habitats. An important question is whether the increasing importance of shellfish was correlated with a decrease in the use of other resources. In other words, did this change in subsistence pattern merely represent a shift in preference, or an increase in resource diversity? In her preliminary investigation, Stewart (1974:25) stated that sea mammal elements increased in frequency in the upper levels of the Boardwalk Site. Calvert's analysis revealed a similar trend at the Co-op Site. Thus, more intensive use of foreshore resources was associated with increased use of maritime resources at the onset of Period II. Land mammal bones no longer predominate in the shell deposit, as they did in the pre-shell zone, but land mammal elements continue to be present, albeit less frequent, in Period II components (Stewart 1974:25; Calvert 1968:96). Changes in the artifact assemblage support the inference of continued land mammal hunting, at least during the early part of Period II. Chipped stone tools, including 166 square-based lanceolate points and leaf-shaped bifaces, "peak in frequency in this period" (MacDonald and Inglis 1981:45). Ground slate points and "pencils" (see Appendix II) also occur in abundance. These tool forms are generally associated with land mammal hunting, and their increase in frequency during Period II suggests that land mammal hunting continued to be important at this time. The increase in shellfish procurement is particularly interesting. The addition of clam and cockle to blue mussel indicates exploitation of intei—tidal resource habitats that were not previously important. MacDonald and Inglis (1981:45) have inferred a substantial population increase from this rapid midden accumulation. To the extent that increasing shellfish (and sea mammal) procurement added to, rather than replaced land mammal hunting, I believe there is justification in MacDonald's and Inglis' inference. It is not the increase in midden size per se, but rather the apparent increase in resource diversity, reflected in the middens that substantiates an inference of population growth during Period II. The conditions of increasing shellfish exploitation and human population growth at Prince Rupert Harbour are consistent with the conditions described for the southern Northwest Coast (Croes and Hackenberger 1984) at roughly the same time, 4000 to 3000 B.P. It will be recalled from Chapter 2 that Croes and Hackenberger argued that, in the absence of a storing economy, the limitation to population 167 growth in this area was the winter resource base. Shellfish was the most suitable resource for intensive exploitation throughout the winter. The implication here is that the occupants of Prince Rupert Harbour had reached a point of maximum resource diversity and niche width (cf. Christenson 1980) by ca. 3500 B.P. Turning to change in prehistoric social organization, the earliest evidence for the transition from egalitarian to ranked society occurred at ca. 2500 B.P. Primarily, this is seen in the burial evidence recovered from Prince Rupert Harbour (Cybulski 1972, 1973, 1974, 1975, 1978, 1979). Over 200 human burials were recovered during the course of excavations at Prince Rupert Harbour, and most of these date to the latter half of Period II (ca. 2500 B.P.) (MacDonald 1983:101-02). Cybulski (1978) reported 12 modified human skull fragments, four altered long bones, and nine complete or partial skeletons with skull and/or long bone modification from this time range. In all cases, the modification was post-mortem, and Cybulski (1978:27) speculated that the preponderence of skull modification may have had ritual significance. Underlying this was the belief that the dead held great "power", an indication that only those few individuals in society who deigned to harness that power (chiefs, shamans?) would tamper with the remains of the dead. Another possibility is that the modified human long bones reflect cannibalism, a practice recorded 168 ethnographical1y for the Tsimshian (Garfield 1939:313-16). Although these suggestions are tempting, they are also highly speculative. Less speculative is the evidence for inter-group or intra-group hostilities beginning about 2500 B.P. Cybulski (1979) and MacDonald and Inglis (1976:77) report mass interments with osteological evidence of healed forearm "parry" fractures, which may have been incurred in attempts to block striking blows, and evidence of blows to the head causing severe skull fractures. MacDonald (1983:101) states that bone clubs appear initially in the Prince Rupert Harbour middens at 2500 B.P., and a basalt dagger, which may date to this time was also recovered from a burial context. The earliest evidence for the use of elaborate stone clubs, which Duff (1963) called the "Skeena River club style", is 2000 B.P. MacDonald (1983:102) anticipates that further research may push this date back to 2500 B.P., based on a 2500 year old stone club from Blue Jackets Creek on the Queen Charlotte Islands (Severs 1974). The presence of "war" clubs at Prince Rupert Harbour during later Period II, in addition to the osteological evidence, has lead MacDonald and Inglis (1976:76) to argue that inter-group hostilities were present at this time, in which raids for slaves or other property may have been conducted. Recently, Ferguson (1983) argued that warfare was an important factor in the complex system of redistributive exchange of the late prehistoric and historic Northwest Coast, and Donald (1983) and Mitchell (1984) have 169 shown that slave labour was an important factor in intensified production and the improvement of social standing (see also Ruyle (1973)). There is also evidence of long-distance trade for status goods during later Period II, seen in the presence of amber beads, copper ornaments and dentalium. These materials were probably items of personal adornment that symbolized the accumulation of wealth by some members of society. MacDonald and Inglis (1981:52) state specifically that ornamental items of copper, amber, dentalium and sea otter teeth, which appear initially about 2500 B.P. (MacDonald 1983:101), have been recovered from some Period II burials. They argue this to be a reflection of status differentiation in the community, although it is not clear whether the "status" burials are independent of age and sex. Both lines of evidence - warfare and possible slavery, and differential status reflected in burials - are consistent with the postulate of ascribed status differences by 2500 B.P. A third, possible line of evidence relates to changes in dwelling size. MacDonald and Inglis (1981:52) argued that there was an increase in dwelling size, associated with population growth, at Prince Rupert Harbour during Period II. They do not state specifically when this increase occurred, but in light of the above evidence for ascribed status differences, 2500 B.P. would seem to be a likely date. Social ranking and large, corporate households were 170 integrally related during the ethnographic period. There are, however, two problems with the "dwelling floor" argument. First, no complete dwelling floors are reported from Period III components, so it is not clear how MacDonald and Inglis base their claim for increased dwelling size during Period II. Secondly, it is equally unclear whether the authors are referring to actual Period II dwelling floors or generalized living floors. Reference is made to large hearth features, floors of beach sand and gravel, post moulds and concentrations of fire-cracked rock (MacDonald and Inglis 1981:52), all of which may be related to habitation, but do not necessarily define the extent of the dwellings. There is no reference to surface outlines of dwellings, from which size could be accurately measured. Inglis (1972:102-03) reported housefloors, recognizable from surface contours at Kitandach and K'nu, but these are historic. No housefloors from Period II context have been described from Lachane or the Baldwin Site (Inglis 1974:27, 30). At Boardwalk, MacDonald and Inglis (1976:17) state that Period II housefloors were identified by black, humus-rich soil and gravel with little or no shell. This may simply represent activity areas in generalized living floors. Features typical of housefloors, including post moulds, slab-lined hearths and caches of boiling stones, indicate that housefloors were probably present within these living floor areas, but not necessarily that each living floor 171 represented a discrete housefloor. In the absence of housefloor depressions or a clear line of post moulds indicating a structure wall, it is often extremely difficult to delineate the boundaries of dwellings within generalized living floor horizons. Indeed, MacDonald (1969b:259) has cautioned that settlement pattern studies on the Northwest Coast have often been neglected in the past because "house structures and settlement information is extremely difficult to obtain from shell middens". While it is tempting to accept MacDonald and Inglis' argument for larger dwellings during Period II, evidence of this increase awaits further research. If ranked corporate groups evolved during the latter half of Period II at Prince Rupert Harbour, as suggested by the above evidence, what developments in social organization occurred during the first half of Period II (3500 to 2500 B.P.)? There is evidence to indicate the initiation of artistic expression at Prince Rupert Harbour at this time. This is seen in the presence of decorative personal effects including incised canine tooth pendants, inverted human figure bone pendants, and siltstone concretions modified with incised ribs, joint marks, vertebral columns, and other skeletal features (MacDonald 1983:101). MacDonald emphasizes the importance of the anthropomorphic representation of this art, because it is the basis of later developments of the classic Northwest 172 Coast style. He states that as early as 3500 B.P., the inhabitants of the harbour area were "approaching a threshold where graphic symbols of corporate identity became meaningful" (MacDonald 1983:101; emphasis mine). MacDonald argues not only that a system of graphic symbolism expressed in art and requiring shared cognitive modes and beliefs was developing, but that a cohesive group structure was also developing, whereby these symbols derived meaning. Yet, MacDonald does not suggest evidence of ascribed status differences at Prince Rupert Harbour until at least 2500 B.P. In other words, corporate identity may have preceded social ranking during early Period II. As we shall see, it was during this period that egalitarian corporate groups evolved at Kitselas Canyon. With the onset of Period I, MacDonald and Inglis (1981:52) state that the Northwest Coast pattern was "in full stride". It is reasonable to refer to this period as "Developed Coast Tsimshian". Analogous terminology is now used by some for the Gulf of Georgia, where the late prehistoric period, which has equivalent dates to Period I on the north coast, has been referred to as Developed Coast Salish (e.g. Ham 1982; Fladmark 1982a:115). Late prehistoric material remains from both areas are consistent with the regional ethnographic material cultures, and it is possible to associate the late prehistoric and ethnographic inhabitants of the respective areas. Whereas the Marpole culture may have been the "climax" phase in Gulf of Georgia prehistory, in terms of the elaboration and diversity of the assemblage. Period I, the Developed Coast Tsimshian phase, marked the cultural climax on the north coast. The pattern of gradually elaborating the assemblage, by adding new tool types and maintaining old ones, continued in this period. MacDonald (1969a; MacDonald and Inglis 1981) has shown that this pattern is consistent, at least in the Prince Rupert Harbour area, for the entire 5000 year prehistory of the region. MacDonald and Inglis (1981:52) state that zoomorphic art flourished in this period, and the style is clearly linked with that of the ethnographic northern Northwest Coast. Zoomorphic labrets, mauls and adzes all apear about 1000 B.P. (MacDonald 1983:102). Zoomorphic designs on bone combs and pins also appear at this time, but MacDonald (1983:103) suggests that these may have greater antiquity, given the presence of similar items from Queen Charlotte Islands components dated between 2500 and 2000 B.P. (Severs pers. comm. to MacDonald 1977). In fact, MacDonald (1983:103) concedes that many of the Period I elements from Prince Rupert Harbour with zoomorphic designs may have antecedents in late Period II components. 174 Kitselas Canyon The excavations at Gitaus, at the lower end of Kitselas Canyon revealed a cultural sequence reflecting an estimated 2500 years of occupation, beginning at roughly 4000 B.P. (Allaire 1979:46-48). In Allaire's (1978) analysis of the cultural material (mainly stone tools) recovered from the site, three cultural components are identified. The earliest component, contained within the lower strata of the site (zone VI), is identified by Allaire (1979:46) as a component of the type of cultural pattern present on the coast at the time, defined as Period III at Prince Rupert Harbour. Radiocarbon dating of the initial occupation at Gitaus (3680 + 130 B.P., GSC 1157; 3760 ± 140 B.P., GSC 1113; 4100 + 310 B.P., GSC 1141) (A1 1 aire 1978:286) supports a late Period III temporal association. This is a small assemblage of 46 retouched stone tools and 153 cobble spalls (Allaire 1979:29). Cobble and flake tools predominate in this assemblage (53.4%), with groundstone implements (25.7%) (mainly abraders), and chipped stone tools (19.0%) comprising the remainder of the assemblage (Allaire 1979:34). The second or middle component at Gitaus is interpreted by Allaire (1979:47) as an indigenous complex, comprising the middle stratigraphic zones (V, Ill-lower, Ill-upper, II+III). Allaire (1979:47) calls this the "Skeena Complex". It is characterized by an increase in proportion and variety of unifacially and bifacially chipped stone tools. The Skeena Complex includes 401 retouched stone 175 tools and 584 acute angle cortex spalls. Of the retouched tools, chipped stone comprises 40.27, of the zone V assemblage, increasing to 59.0% in zone Ill-upper. The diagnostic element of the Skeena Complex is the elongate lanceolate point form, including variants with convex, straight and concave bases, and occasionally with tapering or stemmed bases. Many of the chipped stone implements of the Skeena Complex were manufactured from a distinctive green chert (57.8%) (Allaire 1978:287). Groundstone tools (abraders) decline drastically in zone V to only 3.0% of the assemblage, increasing to 15.2% in zone Ill-upper. Cobble tools actually increase in zone V to 57.3% of the assemblage, but then decrease to 25.7% in zone Ill-upper. It should be noted that zone V includes gravel fill and probably represents a disturbed context (Allaire 1979:33). Finally, the third and most recent component identified at Gitaus by Allaire comprises the upper stratigraphic zones (I, II). This is termed the "Kleanza Complex" (Allaire 1979:48), and it is characterized by an increase in the proportion of cobble tools (55.07.) and groundstone tools (25.37,). New elements include ground slate points with hexagonal cross section, labrets, slate mirrors, and a variety of shaped abraders, including ovoid, trapezoid and rectangular forms. Abraders (shaped and unshaped) now comprise 21.5% of the total assemblage, which Allaire (1979:45) suggests as an Indication of a wel1-developed bone industry not preserved at the site. Chipped stone 176 tools decrease in the Kleanza Complex to 19.7%, but lanceolate point forms are still present. The dating of the Gitaus components is problematic. Although the initial occupation of the site is firmly established at around 4000 B.P., no radiocarbon dates were obtained for the subsequent duration of the sequence. Allaire (1979:46) argued that the Skeena Complex could be assigned a date of roughly 3400 B.P., based on a date of 3430 + 200 B.P. (GSC 746) from Zone A of the Hagwilget Canyon site, GhSv 2 (MacDonald 1969a:249). This zone produced a chipped stone assemblage, including lanceolate points, similar to that of the Skeena Complex at Gitaus. In addition, the Skeena Complex is stratigraphica11y superior to the dates of 4000 B.P. from Zone VI. Allaire's dating of the Kleanza Complex (2500 to 1500 B.P.) is also based on certain trait similarities, shared in this case with the late time range of the middle period at Prince Rupert Harbour. Late period elements such as splitting adzes and hand mauls were not recovered. Recently, a charcoal sample from lower Zone I at Gitaus was submitted for radiocarbon dating. This yielded a date of 2390 ± 85 B.P. (NMC 1318), which supports Allaire's proposed initial date for the Kleanza Complex. In his reconstruction of the prehistoric cultural developments at Gitaus, Allaire (1979:48-50) argued that a migration of people, presumably from farther upriver, is the best explanation to account for the introduction 177 of the Skeena Complex. The Kleanza Complex was thought to be the result of coastal acculturation at the end of the Skeena Complex (Allaire 1979:49). The interior migration argument for the Skeena Complex is based on the numerous similarities in artifact types between the Skeena Complex at Gitaus and the early materials recovered from GhSv 2 (Allaire 1979:46). These similarities include para 1 1e1-f1aked lanceolate points, and a chipped stone tool kit including various bifaces (daggers, knives, picks), scrapers (keeled, thumbnail, flat, fan-shaped), perforators, gravers and notched and denticulated flakes. Allaire (1979:48) argued that the early occupants of Hagwilget Canyon and Kitselas Canyon were, in fact, the same population, who migrated from the former locality to the latter around 3600 to 3400 B.P. Allaire (1979:48) and Ames (1979a:208-l0, 1979b:234-36) both regard the people of the Skeena Complex as an "interior" population, ethnically and culturally distinct from their coastal neighbours at Prince Rupert Harbour. Their contention is that a diversified, well-made chipped stone assemblage, which characterizes the Skeena Complex, is more typical of the interior than the coast. Some similarities, however, do exist in stone tool forms between the Skeena Complex and the Period II assemblages at Prince Rupert Harbour. Of particular importance here is MacDonald and Inglis' (1981:45) statement that chipped stone tools "peaked in frequency" at Prince 178 Rupert Harbour during Period II. This includes long, square-based lanceolate points and bifaces, leaf-shaped bifaces and scrapers, all similar in form and identical in manufacturing technique (Ames 1979a:208) to those of the Skeena Complex. In addition, it now appears that there was contemporaneity between the Skeena Complex and early Period II at Prince Rupert Harbour, given MacDonald's re-assignment of the beginning of the middle period to 3500 B.P. To be sure, the coastal and upriver tool assemblages at this time are not identical. But does this mean that they necessarily represent two distinct populations? This important question has direct relevence to the model of cultural change proposed in this study. I have argued (Chapter 2) that long-term economic changes led to the permanent settlement of critical resource localities such as Kitselas Canyon, and ultimately to the formation of egalitarian corporate groups. This model assumes evolutionary change within the same cultural group - not change as a result of population displacement. Let us examine the evidence from Kitselas Canyon and Prince Rupert Harbour. The assemblage differences exist mainly in terms of the frequency of occurence of certain artifact types (especially chipped stone tool forms) rather than presence or absence. For example, projectile points and bifaces (19.07.) and formed uni faces (12.0%) occur more frequently in the Skeena Complex at Gitaus than at 179 Prince Rupert Harbour during Period II. MacDonald and Inglis (1981:46) list lanceolate points, 1eaf-shaped bifaces and formed unifaces as present (but rare) at this time. They also list flaking detritus and cores, which suggests that these tools were manufactured at Prince Rupert Harbour (i.e., not acquired through interior trade). Moreover, both localities share high proportions of cobble tools (25% at Gitaus, Skeena Complex) and cortical spall tools. Thus, the differences between early Period II assemblages from Prince Rupert Harbour and the Skeena Complex at Gitaus appear more quantitative than qualitative, and the presence of para 11e1-sided lanceolate points in both localities seems especially important. Lanceolate points have also been recorded from the Queen Charlotte Islands (Fladmark 1970) and southeastern Alaska (Keithan 1962). Further, even the quantitative differences between the assemblages from the two localities may be distorted because of differential factors of preservation (shell versus non-shell contexts). Bone tools occur frequently at Prince Rupert Harbour (see Ames 1976). No bone tools were recovered from the Skeena Complex at Gitaus, but abraders (5.07.) are present in the Skeena Complex (especially Zone III), which suggests that bone tools were manufactured at this time, but not preserved. There may be another reason for the paucity of chipped stone tools at Prince Rupert Harbour, and in most prehistoric coastal assemblages. Coastal archaeology in British Columbia 180 has concentrated almost exclusively on deep shell midden sites. Survey is often done from boats, and typically only the coastline is investigated. This represents an extreme sampling bias. If shell midden components primarily reflect shellfish harvesting, then the absence of chipped stone tools from these contexts does not seem surprising. Prehistoric land mammal hunting has been we 1 I-documented for Prince Rupert Harbour, Namu and other coastal localities. Despite this fact, the hinterland around these areas has been virtually ignored by archaeological investigations. I suspect that the reasons for this are the often impenetrable bush and the fact that sites away from the shoreline are difficult to recognize. It may be that entirely different tool kits would be found in these contexts, perhaps including high proportions of chipped stone too1s. Turning to the interior of British Columbia, it is unclear which prehistoric components pre-dating 3000 B.P. Allaire and Ames are referring to as possible places of origin for the Skeena Complex. There are very few sites with dates coeval with the Skeena Complex. Those that do exist - Tezli (Donahue 1975, 1977), Punchaw Lake (Fladmark 1976) - have uncertain relations to the Skeena Complex. For example, Fladmark (1976:28) recovered six square-based lanceolate points from Punchaw Lake, but these are described as small and poorly made, which is quite different from the elongate, well-made Skeena Complex points. Moreover, 181 the dating of the Punchaw Lake Site is very uncertain; it is possible that any or all of these points are no more than 500 years old. More problematic is the apparent lack of consensus over what constitutes a middle prehistoric (ca. 3500 B.P.) interior tool assemblage. In his proposed prehistoric chronology of the north-central interior of British Columbia, Helmer (1977:95, 97) argued that the prehistoric period from 5000 to 2000 B.P. is characterized by fish-tail and leaf-shaped points, corner-notched points, and microblades. Only leaf-shaped points occur in abundance in the Skeena Complex at Gitaus (there is one fish-tail point); corner-notched points and microblades are not present. Farther north in the southern Yukon and the Fisherman Lake area of the Northwest Territories, Taye Lake Phase sites (MacNeish 1964; Workman 197B) and Julian Complex sites (Fedirchuk 1970; Millar 1981; Morrison 1984) are the closest analogues to the Skeena Complex. Lanceolate points with square bases have been recovered from sites in both areas, but again these points are different in morphology from the Skeena Complex points. They tend to be quite wide in relation to their length and not truly para 11e1-sided (see Morrison 1984:45 for lanceolate points from the Julian Complex). Moreover, these northern assemblages also include notched points and burins, which are not found in the Skeena Comp1 ex. 182 Thus, while the Skeena Complex does share some elements with interior assemblages (emphasis on stone chipping, lanceolate points), there are many "interior" elements not present in the Skeena Complex (notched points, burins, microb1ades), and there are many "non-interior" elements included in the Skeena Complex (abraders, ground slate, r i bbed stones). Allaire (1979:47) concluded that the Skeena Complex was part of the Northern Archaic Tradition. This may ultimately prove correct, but the differences noted above are important. Perhaps more important is the most recent assessment of the validity of the Northern Archaic: ... traits often appear to be distributed in time and space essentially independent of one another. Patterns and associations can shift disconcertingly, so that maximal cultural syntheses often seem more arbitrary than elsewhere" (Morrison 1984:54). It is difficult to accept the Skeena Complex as an aspect of the Northern Archaic Tradition when disagreement exists as to the nature or even the existence of the latter. An alternative to the "population displacement" model proposed by Allaire and Ames is that the Skeena Complex represents a seasonal inland orientation toward riverine and forest exploitation within a larger annual settlement/subsistence system. At other times of the year, the same population may have located near the mouth of the Skeena River (Prince Rupert Harbour) to exploit marine, terrestrial and foreshore resources, and perhaps to take advantage of the milder winter climate. This should not 183 seem unusual. It was a pattern of settlement mobility practised by the historic Coast Tsimshian. This alternative model is developed further in Chapter 7. Hagw i1 get Canyon Hagwilget Canyon is located at the confluence of the Skeena and Bulkley Rivers near Hazel ton B.C., roughly 100 km upriver from Kitselas Canyon. The three cultural zones at GhSv 2 at Hagwilget Canyon cover a minimal time range from 3400 B.P. to the historic period. The earliest zone, A, is characterized by a predominance of chipped stone tools, including lanceolate points, leaf-shaped points and bifaces, preforms and formed unifaces. Chipped stone tools comprise roughly 50% of the Zone A assemblage, modified cobbles 23%, and groundstone implements including ground slate and abraders 16%. Allaire (1979:46) includes the assemblage from this zone in the Skeena Complex, and Ames (1979a:208) generally agrees, noting numerous similarities. Cultural zone B probably reflects a period of decreased activity at GhSv 2. Two cores, an abrader, some rolled birch bark and scattered fire-cracked rock were recovered (Ames 1979a:207). Although radiocarbon dates are lacking for this zone, it appears to represent the middle and late prehistoric periods (roughly 3000 B.P. to the historic period). Ames (1979a:210) suggests sporadic use of the 184 site as a fishing camp during this time. Zone C, although undated, includes clay pipe fragments, gunflints and glass, and likely represents an historic Carrier occupation. Aboriginal implements include bone tools (18.77. of the assemblage) ground slate (12.57.), and cobble tools (6.37.). Chipped stone tools are not represented (Ames 1979a:212). The single radiocarbon date (3430 + 200 B.P., GSC 746) was assigned by Ames (1979a:208) to Zone A. This seems reasonable, based on the depth of the sample (1.7 metres). There is no apparent justification, however, for Ames' (1979a:208) contention that this is a terminal date for Zone A. There was no attempt to correlate the physical stratigraphy from MacDonald's test unit to Ames' excavations, and it is not stated whether MacDonald's sample was recovered at a depth that coincided with a change in tool assemblage or stratigraphy that might reflect the end of Zone A. At best, it is only possible to assign the date to somewhere within Zone A. It is further argued by Ames (1979a:208) that this date should be considered in conjunction with the three dates from Zone VI at Gitaus, even though those dates are 300 to 500 years earlier than the Hagwilget date. This manipulation of radiocarbon dates appears to be aimed at an attempt to establish contemporaneity between the end of Zone VI (beginning of the Skeena Complex) at Gitaus and the end of Zone A at Hagwilget Canyon. In turn, this could be used to support a model of population migration from 185 Hagwilget Canyon to Kitselas Canyon about 3400 B.P. If we accept the existing radiocarbon dates for Gitaus and GhSv 2 at face value, it is equally plausible to argue that the Skeena Complex existed contemporaneously at Kitselas Canyon and Hagwilget Canyon, some time during the fourth mi 11 enium B.P. Rather than inferring population displacement on the basis of similar tool assemblages, it may be argued that these assemblages simply represent similar seasonal adaptations to similar environments. Based on existing data, the Skeena Complex represents a riverine-forest adaptation, emphasizing land mammal hunting and fishing, centred on the Skeena River between Kitselas Canyon and Hagwilget Canyon. Both components (Gitaus, GhSv 2) represent minimally summer-fall base camp occupations. This interpretation of seasonality is consistent with the apparent emphasis on land mammal hunting and fishing, which are essentially summer-fall activities. At Gitaus, 58 fauna 1 elements were assigned to the Skeena Complex (Allaire 1978:297). Of these, 41 were land mammal elements, including marmot, fisher, beaver, caribou and black bear), and 16 were fish. At GhSv 2, zone A, fauna\ material included an unspecified amount of mammalian and fish remains. This included a concentration of mammal and fish bones in a storage pit at the bottom of zone A (Ames 1979a:203). Of particular Interest was the identification of sub-adult fisher (Martes pennanti) elements at Gitaus, a 186 strong indicator of late summer occupation (see H. Savage in Allaire 1978:157-58). Likewise, the presence of salmon bones in both components (Allaire 1978:297; Ames 1979a:206) is a further seasonal indicator of summer occupation. The interpretation of both components as base camps is based on the large and diversified tool assemblages recovered from each, and the presence of multiple hearth features in each component (Allaire 1979:37; Ames 1979a:213). Allaire (1979:37) states that numerous small post moulds in Zone Ill-lower at Gitaus indicate the use of drying racks. If so, this would be a further indication of site use during summer months. Alternatively, these post moulds may have been associated with crude summer dwellings. Either interpretation generally supports the summer base camp model. The possibility of winter occupation of these sites at this time is less likely, but cannot be completely ruled out. There is no evidence of winter house pits at either site, or of large post moulds that might be associated with winter dwelling construction. This may reflect sampling error, although large post moulds were recorded in the upper zones at both sites. The Skeena Complex at Gitaus also lacks fire-cracked rock (which was abundant in the Kleanza Complex) and pit features. Again, the absence of these features does not deny the possibility of winter habitation. 187 Summary In this review of north coast prehistory, it has been argued that the early inhabitants of the area had a generalized economy, adapted to land and sea mammal hunting, and fishing. This generalized focus was maintained with the earliest settlement of Prince Rupert Harbour at 5000 B.P. (Period III), as seen at the Boardwalk and Co-op sites. By 3500 B.P. (Period II), shellfish procurement had become an important subsistence strategy, and sea mammal hunting was becoming increasingly important. In addition, the inhabitants of the coast were probably moving upriver seasonally, to hunt and fish for salmon. It is likely that maximum resource diversity and niche width were achieved on the lower Skeena at this time. By 2500 B.P. (middle of Period II), there is evidence from the Boardwalk Site burials for social ranking at Prince Rupert Harbour. At Kitselas Canyon, Allaire and Ames argued that the Skeena Complex at Gitaus represented population displacement from the interior. Allaire (1979) further argued that the Kleanza Complex could be seen in terms of the "coastal acculturation" of the people of the Skeena Complex. However, there is reason to doubt the "population displacement" argument for the Skeena Complex. Further, given the uncertain dating of the Kitselas Canyon and Hagwilget Canyon components, we cannot be sure that the Kleanza Complex and the Skeena Complex are contiguous. 188 This casts doubt on the "acculturation" argument. The following chapters (5 and 6) present and analyze new archaeological data from Kitselas Canyon. These data fill in the gaps in the existing chronological sequence, and provide^ the basis for a revised sequence. This sets the temporal framework for the investigation of change in soc i a 1 organ i zat i on. 189 Chapter 5 THE PAUL MASON SITE Introduct i on This chapter presents archaeological data from the Paul Mason Site at Kitselas Canyon. The surface contours and sub-surface matrix of the site are described. Radiocarbon age estimates are given, and the lithic artifacts from the site are classified and described. These data are used to expand the existing cultural sequence (see Chapter 7). They also provide the data base for the investigation of prehistoric social change (see Chapter 8). Site Description The Paul Mason Site was named in honour of Paul Mason, a village elder of the Kitselas Band. There is no Tsimshian name for the site. Its existence was unknown to the members of the band until it was recorded during the course of archaeological survey in 1981 (Figure 5.1). The site is situated in a heavily forested area on the upper terrace level on the east side of the canyon, 138 metres above sea level, and roughly 25 metres above the July (high) water level of the Skeena River. It is located at mid-canyon, and is sheltered from the main channel of the river by a small island directly in front of the site. At low water, this channel becomes dry, creating a long narrow back eddy. In late spring and early summer. Figure 5.1 The Paul Mason Site, GdTc 16. 191 water flowing on the near side of the island is notably Vess turbulent than in the main channel of the river. The site measures roughly 175 metres long by 105 metres wide, oriented on a northwest by southeast axis. The topography is rugged and uneven. Below the site, the canyon wall rises almost vertically from the river. A steep slope extends back from the canyon wall for a distance of 60 metres, with a vertical rise of 10 metres. At the base of the slope, near the north end, there is a deep gully or swale. This may be an ancient stream bed that was active at a time when the river was at this level. The slope ultimately yields to a relatively flat narrow terrace that extends the length of the site, and is 50 metres in width. Following CIague's (1984) model of glacial recession in the area, this terrace may have formed as a peri glacial beach prior to 8000 BP. The site is rimmed by another steep slope extending back from the terrace, and characterized by formidable outcroppings of bedrock. This slope extends in a curving arc from the north end of the site, along the eastern edge, and finally intersects the canyon wall at the southern end of the site. The modern vegetation on the site is mixed. The slope rising from the canyon wall is characterized by mature growth of alder, hemlock, cottonwood, spruce and white birch. Generally, surface visibility is very good, although the upper terrace at the rear of the site includes a thick undergrowth of thimb1eberry, ferns, devil's club, thistle 192 and stinging nettles. A sparse growth of young alder, hemlock and white birch is also found here. The vegetation on this part of the site suggests disturbance of recent or i g i n. Cultural surface features at the Paul Mason Site include a series of prepared house floors, indicative of a small village occupation. These features are recognizable by their very distinct surface contours. In effect, the floors were "dug in" to the upper slope at the edge of the terrace to prepare a series of flat level surfaces. Earth-moving was deepest at the upslope end of the floors, while no excavation is apparent at the downs lope end. The fill from these excavations was piled around the perimeter of the floors. Ten prepared floor features were identified (see Figure 5.1). Their orientation is generally consistent with the ethnographic village pattern for the Tsimshian area, and much of the Northwest Coast. Floors are oriented side by side, facing the river, each floor being separated from the next by a ridge of earth fill. The floors are arranged in two rows, one behind the other. The "back" row (floors highest on the slope and farthest from the river) includes six floors, and the "front" row (closest to the river) has four floors. Other surface features at the Paul Mason Site include a small fresh water spring, lined at the surface with flat stone slabs. It measures about 1 x 1 metre, and is located on the slope, south of the prepared floor area (see Figure 193 5.1). A small area of the slope, about 5x5 metres, has been levelled around the spring. There are two other small levelled areas on the slope at the south end of the site, but it is unclear whether these were cultural or natural formations. Each measures about 5x5 metres. They may have been used as activity areas. On the upper terrace level, two prepared floor features were identified. One (floor 11) is located about 20 metres northeast of the back row of floors (see Figure 5.1). This floor is larger than any other floor at the site. It is rectangular in shape and defined by a low earth ridge around the entire perimeter. The interior has been dug out to a depth of about 0.5 metres. The floor is oriented northwest/southeast and measures 12 x 10 metres. The second floor (floor 12) is located at the extreme south end of the site, on the upper terrace level at the crest of the slope (see Figure 5.1). This floor is also defined by a low earth ridge around the entire perimeter. It is rectangular in shape, oriented northeast/southwest (facing the river), and the interior has been excavated to a depth of approximately 0.6 metres. This floor measures 8x5 metres. The location of this floor affords an excellent view downriver from the site. Floors 11 and 12 are similar in that the earth ridges defining their perimeters are squared in relation to the floor - that is, little slumping or erosion of these ridges has occurred. By contrast, the earth ridges surrounding 194 the floors on the lower slope are rounded as a result of erosion. The difference in extent of erosion suggests that the floors on the upper terrace are of more recent origin than the floors on the slope. Samp1i ng Methodo1ogy Binford (1964) and Struever (1971) were among the first archaeologists to emphasize the importance of devising a sampling strategy as a coherent and integrated part of an overall research design. The basic problem is to derive a sample that best reflects the population at large. They urged a strategy based on the principles of probability or random sampling as a means of satisfying the criteria of data adequacy and representativeness. This was primarily aimed at archaeological investigations of regional scope. Sampling a deeply stratified site, however, often poses particular problems (see Brown 1975:155). One problem pertains to "site coverage". It is often logistically impossible to draw representative samples from the entire area of the site. Flannery (1976:68) cautions that it may even be unwise to try. To do so, usually involves the excavation of small randomly placed units ("telephone booths"), and the data derived from these deep dark holes is usually uninformative, and can be misleading because of the lack of stratigraphic control. Another problem is defining the extent of the population to be sampled. This can prove difficult if the population 195 in question is a buried cultural horizon of unknown extent. How does one ensure that a representative sample is drawn? Typically, the population is arbitrarily defined because the archaeologist does not know the extent of the buried hor i zon. If the site to be tested is deeply stratified, and only a few excavation blocks can be dug, it may be preferable to adopt a judgemental sampling strategy. Asch (1975:191) states that a non-probabilistic design is often useful "to investigate spatial structure" and "to select more representative, if biased, samples when only a few sampling units can be excavated". The Paul Mason Site provided a good case for judgemental sampling. The primary aim of the research was to obtain archaeological data pertaining to prehistoric social organization. The presence of a series of prepared floor features, whose boundaries were well-defined, afforded the opportunity to collect archaeological data relating to the size and construction of dwellings, household size, and domestic organization. Random sampling from a population of only 12 prepared floors would not likely increase the representativeness of the sample. Moreover, selective sampling would ensure the excavation of large, well-defined floors that did not have obstructions to excavation (such as large trees growing i n the f1oor). To determine the depth and spatial extent of the 196 cultural deposit, soil cores were taken from around the site and five 1 x 1 metre test units were excavated. Three of these units were excavated within prepared floor areas on the slope (test units 1, 2, 3), one in a non-floor area of the slope (test unit 4), and one on the upper terrace (test unit 5) (see Figure 5.2). These tests revealed a deeply stratified cultural deposit on the slope, greater than 1.5 metres. By contrast, the cultural deposit on the upper terrace was thin and discontinuous. It was decided that excavation should be concentrated on the slope. This decision was guided by two factors. First, the surface outlines of the ten house floor depressions on the slope indicated that this was the main residential area of the site. Secondly, the deepest cultural deposit was located here, which implies more intensive human activity, and perhaps a greater time range of occupation than the upper terrace. Thus excavation of the slope would yield data related to village life and domestic organization, and in addition, data related to earlier occupations of the site. This would allow for investigation of the evolution of village life at Kitselas Canyon in diachronic perspect i ve. Two prepared floors, 2 and 9 (see Figure 5.2) were selected for excavation. Both floors were excavated by means of a trench oriented along the long axis of the floor. In the excavation of floor 2, it was possible to extend units off the north side of the trench to reveal most of 197 Figure 5.2 The Paul Mason Site: house floors and excavat i ons. 198 one side of the floor. In the excavation of floor 9, a crossing trench was excavated at roughly the mid-point of the floor. These were the major prepared floor excavations. The purpose of these excavations was two-fold. First, they were intended to provided data relating to domestic organization, including household size and internal household divisions. Secondly, they were intended to provide data related to dwelling construction technique. Floor 12, the small floor on the upper terrace was also tested by means of a 3 x 1 metre transect. A problem with the excavation of prepared floors, especially those that have undergone extensive earth-moving, is that the underlying cultural deposit has often been disturbed (see Wilmeth 1977; Fladmark 1982a:123). To control for disturbance, and to develop a reliable, complete cultural sequence for the site, it was necessary to conduct excavations away from areas of possible disturbance. Two areas (H and J), adjacent to the prepared floors were selected for excavation (Figure 5.2). Excavation H was located immediately north of the prepared floors. Excavation J was just west or "in front of" the prepared floors. These areas were selected for excavation because their proximity to the prepared floors indicated that they were probable midden areas, associated with the occupation of the floors. These excavations were expected to yield a relatively high frequency of artifacts from undisturbed cultural deposits. The non-floor excavations were important for two 199 reasons. First, they would yield an undisturbed stratigraphic record to control for disturbance that had taken place in the prepared floor areas. Second, they would provide data concerning spatial variability at the s ite. Excavations were undertaken in block or trench areas, utilizing 1 x 1 metre units. All excavation was done by trowel, and all excavated material was screened through 6 mm mesh. Excavation followed natural soil layers whenever possible, and 10 cm levels within layers. A total of 64 square metres of the site was excavated. Stratigraphy at the Paul Mason Site The range in depth of exacavations at the Paul Mason Site was from 0.60 to 2.13 metres. In the four main excavations, the cultural deposit was overlaid by the surface littermat. At the base of the 1ittermat was a continuous discrete layer of charcoal and partially burned wood, 3 to 8 cm thick. A similar layer was encountered in most of the shovel tests conducted during the survey of the east side of the canyon. It probably represents a forest fire of the late 19th or early 20th century, as burning seems to have been intensive in some areas, and moderate to absent in other areas. Underlying the littermat was a thin layer of humo-ferric podosol, representing post-occupational soil development at the site (see Figure 5.3). This horizon averaged 10 200 60 120-180 -1 c m I i 11 e r m a t charcoal, partially burned wood Po s t - o c c u po t i o n a I Horizon humo - ferric podosol Cultural Deposit, Upper Strati graphic Zone dark brown silty sand 76 charcoal 8 carbon ash burned soil fire cracked rock unmodified rock beach sond gravel Lower St ra t i graphic Zone red brown silty-sand 79 charcoal, carbon, ash, burned soil I unmodified rock 20 cobbles 8 boulders Percentages based on volume, derived from column samples from Excavation H and Floor 2. Proportions under 5% ore estimates. Figure 5.3 Schematic matrix profile of the Paul Mason Site. 201 cm thick, and was encountered In each of the four main excavation areas. lt was a light yellow-brown humus, typical of the non-cultural matrix In and around Kitselas Canyon (Valentine, Sprout, Baker and Lavkulich 1981:104). The cultural deposit underlying the post-occupational horizon included two physical stratigraphic zones. The upper zone was a silty-sand matrix, dark brown in colour. It included intrusions of pure coarse sand, gravel, charcoal and ash. Fire-cracked rock was also abundant throughout the upper layers of this zone. The dark brown soil zone constituted most of the cultural deposit at the site, ranging in depth from 0.50 to 1.75 metres. It included numerous lenses and discontinuous layers. Most of the cultural material was recovered from this zone. In the two house floor excavations this zone was "wedge-1ike" in shape. It was thickest near the west or downs lope end of the floors, and became progressively thinner toward the rear or upslope end (see Figure 5.4). This was most apparent in Trench A of floor 2 (see Appendix I; Figure 1.1). The wedge shape is a result of the levelling of the slope to prepare the floors. The most substantial amount of earth would have been removed from the rear area of the floor, with little or no earth removed from the front of the floor. The fill resulting from this earth-moving was then deposited around the perimeter of the floor forming ridges on either side. The cultural material, originally deposited in the lower levels of the dark brown soil zone. 202 A - hill-slope prior to human occupation and deposition of cultural material C - hill-slope modified by house floor preparation Figure 5.4 Schematic profile showing slope transformation during the occupation of the Paul Mason Site. 203 may have been disturbed as a result of floor preparation, disturbance was probably greatest toward the rear of the f1oor. The profile of Trench B of floor 2 (Appendix I; Figure 1.2) shows that in cross-section, the floor was probably formed in two levels. A lower level extended along the middle of the floor, and a raised bench extended along either side. This perspective is also shown in Figure 5.5, a cross-sectional schematic of house floor transformation during the construction, occupation and abandonment phases. At the time of construction, an earth ridge was formed as the central floor area was levelled. During occupation, debris probably accumulated on the ridge in the narrow space between dwelling walls. After abandonment, and the collapse of the dwellings, this debris would have slumped down into the central house floor areas. Figures 5.6 and 5.7 are plan drawings taken at 70 cm below surface of the excavated areas of floors 2 and 9, respectively. These give some indication of the internal organization of the dwellings. Both dwellings had two hearths, one located in the middle of the floor; the other near the front or downs lope end of the floor. This suggests that the dwellings were occupied by more than one family. The positioning of one hearth near the front of the dwelling may be related to a desire for heat in winter. Lithic concentrations were located in the central areas of each floor, identifying these areas as the main activity 204 dwelling floor at time of construction I dwelling floor prior to abandonment / dwelling floor after abandonment post mold accumulated occupational debris Figure 5.5 Cross-sectional schematic of house floor transformat i on. 205 Figure 5.6 Plan of floor 2 (depth = 70 cm). 206 Figure 5.7 Plan of floor 9 (depth = 70 cm). 207 loci. Figure 5.6 shows that the lithic concentration in froor 2 stopped abruptly at the base of the raised lateral bench on the north side. This suggests use of the benches as sleeping areas. Post moulds were located along the mid-line of both floors. This probably reflects the use of a central ridge pole in the construction of the dwellings. In addition, two large post moulds were identified near the crest of the ridge, flanking floor 2. These were probably external wall posts, although it is not known if they were used in conjunction with split planks. The lower zone of the cultural deposit included a thin series of banded soil layers, oriented along a southwest slope, roughly following the general surface slope of this portion of the site. These layers were clearly different in colour and texture from the overlying dark brown zone. The colour range was from yellow-brown to red-brown, and the soil tended to be moister and stickier than the overlyi zone. This zone ranged in thickness from 0.10 to 0.50 metres, and was encountered in the floor 2 excavation and Excavation H. Underlying the cultural deposit in most units was a culturally sterile deposit of poorly sorted glacial till. This was a mixture of gravel-sized rocks and large boulders most of which had been water-rounded. In some areas, most notably Excavation H, glacial till was not present. The cultural deposit rested directly on bedrock. 208 Detailed stratigraphic descriptions of the four main excavation areas, floor 12, and the test units are presented i n Append i x I. Soil samples were taken at 10 cm intervals, beginning at a depth of 0.30 metres below surface, and continuing to 1.70 metres below surface. Granulometric analysis of these samples indicates that the predominant soil grains in the cultural deposit are in the coarse to fine sand range (particle size 0.125 to 0.5 mm) (Figure 5.8). According to Shack ley's (1975:53) four principle varieties of sand grains, the soils from the Paul Mason Site are "worn, rounded and glossy grains...this type of surface texture indicated the action of running water, and the characteristics are found on sand grains of marine or fluviatile deposits". Soil pH levels reveal strongly acidic soil, ranging from 4.6 to 4.9. 209 GR AVE LLY-SILTY-SAND I 63 cm below surface grain size > 8.0 = 2.6 % grain size < .002 = 2.0 % CLAYEY-GRAVELLY - SILT Y - SANO 147 cm below surface grain size > 8.0 = 3.5% groin size < .002 = 4.4 % a> a. GRAVELLY - SILTY - SAND S 119 cm below surfoce grain size > 8.0 - 6.9 % grain size < .002 = 2.1 % SILTY - SANDY - GRAVEL D - 170 cm below surfoce grain size >8.0 = 16.8 % grain size < .002 = 3. I % |- 99.99 - 99.95 - 99.9 cloy size modified Wentworth scole Figure 5.8 Particle size of Paul Mason Site soils. 210 Pat i ng There are 19 radiocarbon dates for the site (Table 5.1, Figure 5.9), ranging from 635 + 100 B.P. to 5050 ± 140 B.P. A few of these estimates are problematic in terms of their stratigraphic position. This is not suprising, considering the amount of earth-moving that occurred during the occupation of the site. A few other dates seem quite recent, but again, there are plausible explanations for these. In general, however, these estimates present a coherent picture of the sequence of occupation of the site. The earliest, 5050 + i4n B.P. (SFU 259), was taken from the middle of the lower stratigraphic zone (red-brown zone) of Excavation H. A date of 4655 + 130 (NMC 1293) from Excavation H marks the transition between the lower (red-brown) and upper (dark brown) zones. The date 4745 ± 195 (NMC 1292), is from the bottom of the dark brown zone in Trench G, unit 1. These estimates suggest that the time of transition from the red-brown zone to the dark brown zone was probably about 4700 B.P. No radiocarbon sample was obtained from the lower stratigraphic zone in the excavation of floor 2 or Excavation J. Of the remaining 16 dates, 13 fall into two major clusters. The earliest of these includes eight estimates ranging from 3780 + 120 to 4395 ± 130 B.P. Using the single standard deviation values, these dates overlap between 4030 and 4185 B.P., and all eight are within a 365 year span 211 (3900 to 4265 B.P.). Four of these samples were recovered from the excavation of floor 2; three from the excavation of floor 9; and one from Excavation H. Their stratigraph?c positions are generally in good agreement. All but one were recovered from the middle or lower portions of the dark brown soil zone, with depths ranging from 0.9 to 1.4 metres below surface. The exception is the date 4395 + 130 (NMC 1290). That sample was taken from the upper dark brown soil zone (Trench A, unit 5) at a depth of 0.6 metres below surface. The date seems too early for this depth. The sample comprised scattered pieces of charcoal from across the unit. Considering the amount of earth-moving that took place at the rear of the floor, it is possible that some, if not all, of this sample had been re-deposited during floor preparation. Six of the above eight samples were recovered from depths of 0.9 to 1.4 metres below surface. This includes a date of 4280 ± 95; WSU 2923) taken from the hearth feature at the base of the upper dark brown layer in Excavation H, the ostensibly undisturbed cultural deposit. It will be recalled from Chapter 4 that the dates from the early component at Gitaus also clustered around 4000 B.P. This suggests a temporal overlap in the occupation of these sites. The second cluster of radiocarbon estimates from the Paul Mason Site includes five dates ranging from 2750 + 90 to 3230 + 160 B.P. The samples were recovered from four different areas of the site; two from Excavation J, and 212 one each from Excavation H, floor 9 and floor 5 (test unit 3)"". Each of these samples was recovered from the upper portion of the dark brown soil zone. Their range in depth below surface was from 0.45 to 0.9 metres. Using the single standard deviation values, these dates are within 230 years of each other, and three of the dates - 3080 + 85 (WSU 292 1); 3130 ± 100 (SFU 132); 3230 ± 160 (SFU 133) - overlap between 3070 and 3165 B.P. These dates appear to reflect a discrete occupation of the site. They are separated from the previous group of eight dates by 500 to 1000 years. The fact that two of the samples were recovered from prepared floor features, and al1 five were recovered from the upper portion of the dark brown soil zone suggests that these dates are associated with the occupation of the floors. The three most recent dates were all taken from the top of the dark brown zone. The first two, 635 + 100 (NMC 1291) and 890 ± 160 (SFU 135), were both taken from depths of 0.2 metres in floor 9 and floor 5, respectively. These dates seem inordinately late compared to the other dates for the site. Moreover, in the lithic assemblage from the Paul Mason Site (see following section), there are no artifact forms to indicate an intensive late prehistoric occupation of the site. One possible explanation for these dates is that the samples may have been contaminated by the surface layer of charcoal and partially burned wood. Both samples were recovered from just below this layer. 213 It is unlikely that the 1330 ± 90 (SFU 260) date was affected by contamination from surface charcoal. It was taken from the top of a cultural hearth feature, 0.3 metres below surface, sufficiently deep to reduce the possibility of contamination. One possibility, which may be applicable to each of the three late dates, is that they represent periodic short-term re-use of the site. This possibility seems most likely for the 1330 ± 90 date. Because Kitselas Canyon is an important fishing location, in an area where fishing was critical to the late prehistoric aboriginal economy, it is not unlikely that the canyon was used intensively during the last 2000 years (although archae1o1ogica1 evidence of this has yet to be recovered). Thus, periodic re-use of a site that offered existing prepared floor features is a d i st i net possi b i1 ity. In summary, radiocarbon estimates from the Paul Mason Site indicate that the most intensive settlement of the site was from roughly 5000 to 2800 years ago, with numerous dates clustering at 4200 to 4000 B.P., and again at 3200 to 2800 B.P. Periodic short-term re-use of the site may have ensued throughout the late prehistoric period. 214 Table 5.1 Radiocarbon dates from the Paul Mason Site Laboratory No. NMC 1291 SFU 135 SFU 260 WSU 2922 WSU 2920 WSU 2921 SFU 132 SFU 134 SFU 133 SFU 255 SFU 256 SFU 257 SFU 258 WSU 2923 SFU 261 NMC 1290 NMC 1293 NMC 1292 SFU 259 Date (uncorrected) 635 ± 100 890 ± 160 1330 + 90 2750 ± 90 2840 ± 85 3080 ± 85 3130 ± 100 3230 ± 160 3780 ± 120 4060 ± 120 4 130 + 90 4250 ± 100 4270 ± 200 4280 ± 95 4350 ± 320 4395 ± 130 4655 ± 130 4745 ± 195 5050 ± 140 Un i t Layer Depth (cm) C3 C 21 TU3 - 26 C2 E 30 J1 E 70 Jl D 44 H2 D 57 C4 E 78 TU3 - 65 C4 F E2 H Al H C8 G C8 F H2 D A7 H A5 E H3 D 1 1 0 95 140 1 1 4 96 1 1 0 98 58 103 Gl H/J 102 H3 G 101 215 -i 1 1 r -i 1 1 r SFU256 ~i r IO(H SFU 133 SFU 25? WSU 2923 NMC 1293 NMC 1292 SFU26I -t-SFU258 SFU 259 SFU 132 6 50t WSU 2922 SFU 134 WSU 2921 NMC 1290 50 SFU 135 —1 L 500 1000 2500 J I I L. 3000 3500 I L 4000 C14 years BP Figure 5.9 Radiocarbon dates from the Paul Mason Site plotted against depth of sample. 216 Artifact Descriptions A total of 1194 artifacts was recovered during excavation at the Paul Mason Site. All are stone tools. Grinding and chipping were the principal means of stone tool manufacture. Extensive pecking was not In evidence.. Chipping was most frequent on large, unformed cobb1es and flakes. Formed, chipped bifaces and unlfaces were also manufactured, but less frequently than the cobble-based implements. Classification of stone tools was based on physical attributes, including size, shape, location and quality of retouch, edge angle, presence or absence of wear, and raw mater ia1. With respect to size, some artifacts (cobble core tool8» utilized flakes, spalls) were separated into large or small types on the basis of weight, length or a combinat1 on of the two (see Plate IV a-c). Edge angle also refers to a bipartite quantitative division. Retouched, utilized edges were classified as o "steep" if they exceeded 45 (angles measured at 2.mm from the edge; see Pokotylo 1978:215) (Figure 5.10). "Acute" o edge angles were less than 45 (Figure 5.10). Implements o with edge angles less than 45 were probably used as knives o or cutting tools, whereas edges greater than 45 .were best suited to skinning or hide-scraping, plant fibre shredding, and heavy wood or bone working (Semenov 1964; Wilmsen 1970; Gould 1971). 217 r- 3 a Figure 5.10 Implements with varying edge angles. a- implement with steep edge angle b- implement with acute edge angle Figure 5.11 Implements showing different types of edge wear. a- battered undercut edge; b,c- polished smooth edge. 218 Two types of wear were distinguished: microf1aking, in which edges were often undercut; and abrasive dulling, in which edges were smoothed and polished, often with parallel striat ions (Figure 5.11). Acute, polished edges with parallel striations are inferred as cutting edges. Steep edges with microflaking may reflect scraping, chopping, shredding or rasping activities. Distinction was made with respect to quality of retouch. "Regular" retouch refers to numerous, small, evenly-spaced flake scars on the edge or face of the impl.ement. "Irregular" retouch refers to a few large, unevenly-spaced flake scars, usually on the edge of the implement (Figure 5.12). Classification was made independent of the component in which the artifacts were found, although this undoubtedly had an effect on the distribution of certain artifact types. Three lithic categories - cobble and flake, ground stone, and chipped stone - form the major classificatory divisions. These are sub-divided into 19 artifact classes The artifact classes are divided into a number of individual types (see Appendix II for artifact type descriptions). Cobble and flake tools are, by far, the most numerous in the assemblage; they are described first, followed by groundstone tools, then chipped stone or flaked tools. Comparison to other established classifications, especially Mitchell (1971), Matson (1976), Allaire (1978) and Lightfoot (1983), is made where appropriate. 219 F i gure 5.12 Implements with varying quality of edge retouch a- Implement with irregular edge retouch, b- Implement with regular edge retouch. 220 Cobble and Flake Tools Cobble implements and large. Irregular flake tools based on cobble cores consltute the largest; portion of lithic artifacts recovered from the Paul Mason Site (60.9% of all artifacts; N=718). These are simply manufactured implements, based on locally available lithic material (primari1y: beach cobbles). Classification of these implements on morphological characteristics was often difficult. They generally show evidence of only minimal reduction. For example, core tools were formed, in some cases, by the removal of only a few flakes. In other cases, flake tools were used with little or no edge retouch (although wear is present). The question of tool function is also difficult to address, except in a general sense. Undoubtedly some implements served more than one purpose. Although it is difficult to perceive such implements as "personal gear",, subject to curat ion, (Blnford 1979, 1980), this does not mean that some cobble tools could not have been extensively used. For example, Ackerman (1968:68) suggested a variety of uses for cobble Implements from Groundhog Bay 2, including tree bark removal and processing, chopping down saplings, and gIrdlIng.larger trees. One of the best functional treatments of cobble and flake tools from a Northwest Coast site Is by Mitchell (1971). He considered basic tool attributes such as size and overall shape, edge angle, location and pattern of wear. 221 and raw material to determine probable function. The ;kfnds of uses suggested by Mitchell (1971:102-08) for the various cobble and flake tools from Montague Harbour Include chopping, pulping and rasping for core tools, and scraping or cutting for flake tools. Core tools were generally argued to be woodworking tools (see also Borden 1968; Ackerman 1968). Ackerman (1968:68) has argued that choppers from level II at Groundhog Bay 2 were later functionally replaced by ground stone adzes. Although chopping tools were prolific at the Paul Mason Site, no ground adzes or celts were recovered. 222 I-A Cobble Tools (N=360) These implements, sub-divided into six types, are core tools, based on beach cobbles and pebbles. Most have received minimal reduction, and retain a large amount of original cortex. The larger of these implements - cobble core tools, choppers, split cobble tools (Plate I, II, III) (see Appendix II) - were probably used as heavy woodworking tools (Borden 1968; Ackerman 1968). Most implements (N=330; 91.77,) have steep edge angles. Edge wear is invariably in the form of microf1aking, with substantial undercutting suggesting heavy use. These tools were based on locally available raw material - primarily argil lite, sedimentary rock and basalt. Although reduction was unsystematic in terms of core preparation, there was a marked tendency for the sequential removal of flakes, beginning at one end or side of the cobble, using the cortex edge as a ready-made striking platform (Figure 5.13). Cobble tools occur in great frequency at most north coast sites. The cobble tools from Gitaus (see Allaire 1978:135, 45, 294) are similar to those from the Paul Mason Site in reduction technique and raw material. Calvert (1968:39) reports only three "large unifacially flaked pebble tools" from the Co-op Site. Despite this paucity, Calvert (1968:39) states, "pebble tools are found in great numbers on the Prince Rupert beaches and have been excavated at 223 a-d cobble core tools; Plate I e pointed cobble implement 224 Plate II a-d cobble choppers Plate III a-c split cobble tools; d-f acute angle cobble tools. 226 gure 5.13 Modified cobble flake with platform cortex. 227 all levels of Tsimshian sites". At Hagwilget Canyon, (GhSv ZY zone A, cobble tools were the most frequent tool type, comprising 23.0% of the assemblage. Lightfoot (1983:110-18) describes flaked cobbles from Component II at the Hidden Falls Site. Component II has radiocarbon dates in the 3200 B.P. range, which are consistent with the late cluster of dates at the Paul Mason Site. The cobble tools from Component II do not, however, compare favourably with the Paul Mason Site material. Angular fragments of rhyolite, andesite and tuff were used extensively at Hidden Falls; argil lite and sedimentary rock were rarely used. Cobble tools have considerable spatial and temporal range on the north coast. They occur less frequently in the Interior. Their high frequency at the Paul Mason Site in all levels (30.2% of the overall lithic assemblage) suggests a coastal affiliation. I-B Utilized Flakes (N=215) This class includes flakes with evidence of wear, either in the form of abrasive dulling or micro-flaking (Plate IV). Although the distinction is often subtle and subjective, edge modification on these tools is argued to be the result of use, rather than retouch. The implements are sub-divided into six types (see Appendix II) on the basis of extent and location of cortex, flake size and edge angle. Included among these types are large and small utilized Plate IV b large utilized spalls; c small spalls; d-e acute ilized flakes; f-g steep angle utilized flakes 229 cortex spalls (Plate IV a-cj Figure 5.14). These Implements have acute edge angles, and often have polished edges with parallel strlatlons. Some Northwest Coast archaeologists (Hanson 1972; Allaire 1979) have argued that cortex spalls were used In the initial stages of fish processing. Like cobble tools, the distribution of cortex spalls is widespread spatially and temporally on the Northwest Coast. At Gitaus and Prince Rupert Harbour, they occur in great frequency in all levels (Allaire 1978; MacDonald and Inglis 1981:46). They are reported from south coast sites, including Montague Harbour I (Mitchell 1971:102) and Glenrose (Matson 1976:133). They are also reported from numerous interior sites In British Columbia. The vast majority of the utilized flakes, including cortex spalls, from the Paul Mason Site (178; 82.97.) have acute edge angles, which suggests that they were used as cutting tools. A wide variety of raw materials were used, but sedimentary rock (N=59) and argi11ite (N=86) were the most common. I-C Retouched Cobble Flakes (N=52) These large flake implements, based on river cobbles, are sub-divided Into three types. Two pertain to retouched spalls (Plate V a-d). Following Matson (1976:135), these implements are sub-divided according to edge angle - steep or acute. It Is likely that steep angle retouch cortex 230 Plate V a-b steep angle retouch spalls; c-d acute angle retouch spalls; e-h retouched cobble flakes; f-j cortex-based cores 231 Figure 5.14 Utilized spall with smooth edge. 232 spalls were used as scraping tools. Acute angle retouched spalls were probably used as cutting implements. Allaire (1978:147) defines a class of retouched cortex spalls from Gitaus - ecailles enti^eres^ retouche corticale - but they are not sub-divided by edge angle. The third type is large retouched flakes from unprepared cobble cores (Plate V e; Figure 5.13). These are consistent with ecailles de galets or grande ecailles de galets from Gitaus (Allaire 1978:145-51), and with "large utilized flakes" from Glenrose (Matson 1976:133). These implements are generally larger and more simply manufactured than the retouched flakes of the chipped stone category. The raw materials are sedimentary rock (N=24), argil lite (N=21), basalt (N=4), slate (N=2), and sandstone (N=l). I-D Cobble Cores (N=42) The implements in this class are cores based on river cobbles. There is no indication that these implements were used as tools, although that possibility cannot be totally discounted. The implements are sub-divided into three types based on extent of reduction and orientation of the core for flake removal (see Appendix II). Two types are consistent with the "unformed cores" and "cortex-backed cores" (Plate V i-j) defined by Matson (1976:131) at Glenrose. The third type includes minimally reduced cobbles (see Matson 1976:142). At Gitaus, Allaire (1978:151) 233 identified only one type of cobble core - nuc1e\ i nformes. Raw materials include sedimentary rock (N=23), argil lite (N=16), and basalt (N=3). I-E Hammerstones (N=59) These are unformed cobbles, modified by battering or crushing, usually at one end. In four cases, a large flake or fragment was removed, as a result of hammering. The size of these implements ranged along a continuous gradient; there were no obvious break points for distinguishing types on the basis of size. Locally available raw materials were used, including sedimentary rock (N=43), argil lite (N=14), and igneous rock (N=2). I I Groundstone Tools Stone grinding was essentially a coastal industry, and one of substantial antiquity. In the Gulf of Alaska, Clark (1968) has documented the existence of a ground slate industry as early as 6000 B.P. Stone grinding was not well developed in the interior. Groundstone implements constitute 15.5% (N=180) of the total lithic assemblage. At the Paul Mason Site, there is evidence of a complete stone grinding industry, including all stages of manufacture. Stone saws were recovered, as well as sawn and snapped fragments of metamorphic and microcrysta11ine rock. Shaped and unshaped abraders were recovered throughout 234 the deposit. Formed groundstone implements, including projectile points, drills, "pencils" and single-edged knives, were also recovered, although these were rare. The presence of numerous abraders and rare groundstone implements indicates that something in addition to stone was ground at the site. The most probable candidates are bone, antler and wood, all organic materials not likely to survive in the highly acidic, non-shell matrix. The general groundstone classes are based on probable function (saws, abraders, etc.), raw material (slate, green schist, sedimentary, etc), and general morphological characteristics (form, location and extent of grinding). I I—A Saws and Sawn Fragments (N=14) This class includes all implements related to stone sawing, one of the initial steps in the manufacture of groundstone tools (Plate VI a-d). Stone sawing has been described by Clark (1979:170) in the context of the Ocean Bay sites where there is evidence of stone sawing and ground slate manufacture at 6000 B.P. Clark states that thin, often unretouched boulder flakes were used as saws. They were recognized by smooth, longitudinally striated edges (Figure 5.15). Tablets were typically sawn one third to one half way through, "and then snapped through the remaining septum" (Clark 1979:170). At Gitaus, Allaire (1978:130, 132) identified two stone saws and ten sawn fragments. Stone saws from the Paul Mason 235 Plate VI a-b stone saws; c-d sawn fragments; e-f finished ground slate; h-1 thin ground slate fragments 236 F i gure 5.15 Saw and sawn fragme 237 Site were manufactured from sedimentary rock, argl11Ite and green schist. 11-B Ornamenta1/DecoratIve Objects (N=14) A variety of ornamental and decorative Items were recovered from the Paul Mason Site (see Plate VII). These Include five polished, spherical stones (Plate VII a-e), a slate disc bead (Plate VII f), polished fragments of mica and serpentine (Plate VII g-h), and a bird effigy (Plate VII j). One of the Incised tablets Is a ribbed stone (Plate VII k). In the Tsimshian area, ribbed stones were manufactured as early as 3500 B.P., and became common between 3000 and 2500 B.P. In addition to formed ornaments and items of personal adornment, 19 small, unmodified fragments of ochre (red and yellow) and 9 small nodules of graphite were recovered. These materials were probably cut and ground, and used as pigments. II-C Pointed Groundstone Tools (N=22) Ground, pointed tools from the Paul Mason Site are sub-divided into five types based primarily on form and raw material. One type - "ground slate points" - includes implements that were well-made (Plate VIII a-g). In general,' however, other tools were poorly manufactured. Grinding was minimal and the implements appear unfinished. Often, poor quality raw material such as green schist or sedimentary 238 Plate VI I a-b sperical polished hematite; c-e spherical stones; f slate disc bead; g mica fragment; h highly polished fragment; i incised tablet; j bird effigy; k ribbed Plate VIII a-g ground slate points; h-i ground and chipped bifaces; j groundstone drill; k-m pointed ground schist implements 240 rock was used, instead of high quality slate. Groundstone points are common from most north coast sites. The finished or nearly finished slate points from the Paul Mason Site (N=7) closely resemble those from Prince Rupert Harbour (see MacDonald and Inglis 1981:46; Calvert 1968:36). They are typically thin and hexagonal in cross-section; the edges are bifacially bevelled (Figure 5.16). In contrast, ground slate points from the Hidden Falls Site (Lightfoot 1983:70) appear thicker and more lanceolate than the Paul Mason Site points. Many have stemmed or tapering bases, attributes not in evidence at the Paul Mason Site. The ground schist points from the Paul Mason Site (Plate VIII k-m) (N=10) are interesting for two reasons. Allaire (1978:122) reports two schist points from Gitaus, but otherwise schist is rarely used in the manufacture of ground stone points on the Northwest Coast. It is possible that schist was more readily available than slate at Kitselas Canyon, or that schist was simply preferred over slate by the inhabitants of the canyon because it was easier to grind. Secondly, the schist points at the Paul Mason Site have squared lateral edges, rather than sharp, bevelled edges. They do not resemble groundstone points from other Northwest Coast assemblages. 241 Figure 5.17 "Pencil" showing blunted end. 242 11—D Blunted Groundstone Implements (N=20) The implements in this class are long and flat or rod-like. They have been ground to form a rounded or blunted tip at one end. The grinding is continuous on to the proximal ends of the implements, so it is apparent that the blunt tips were formed in manufacture, not derived through use. These implements are subdivided into three types. Two types include long, thick, poorly manufactured implements (Plate IX g-q). The third type includes facetted "pencils" (Plate IX a-f; Figure 5.17), which are common in Period II components at Prince Rupert Harbour (MacDonald and Inglis 1981:45; Calvert 1968:36). These are finished rod-1ike implements. Drucker (1943:57), who first defined pencils, indicated that those from Prince Rupert Harbour had either pointed or blunt tips. All pencils from the Paul Mason Site (N=5) have blunt tips. They may have been used as stone awls. Slate (N=7) and green schist (N=6) were the most commonly used raw materials. Sedimentary rock (N=5) and argil lite (N=2) are also present. I I—E Miscellaneous Groundstone (N=30) A number of small fragments and broken or unfinished groundstone implements were recovered. They are grouped into seven types based on raw material, and on existing morphological characteristics, such as edge angle. Five probable fragments of thin slate cutting implements were Plate IX a-f pencils; g-q other blunt groundstone implements 244 recovered (Plate VI h-1). In other cases, miscellaneous grinding and/or wear was present, but functional classifications could not be made. II-F Finished Ground Slate (N=3) These are tabular, rectangular pieces of well-made ground slate Implements. Unlike the Implements In the previous class these are large fragments (almost complete) that give some indication of the general shape of the i mp1ement. One implement Is trapezoidal, thin, and extensively ground on both faces (Plate VI e). One face is slightly concave in cross-section; the other face Is slightly convex. Both lateral edges and one end are ground flat (the other end is broken). The implement may be a shaped slate abrader. Tabular, shaped slate abraders are reported from Component II at Hidden Falls (Lightfoot 1983:52). The other two fragments are smaller and less distinctive of overall shape (Plate VI f-g). Both resemble the implement described above in that they have evidence of extensive bifacial grinding, and two edges intersecting at right angles that are also ground flat. 245 11—G Unshaped Abraders (N=43) Flat abrasive stones are divided into two classes: unshaped and shaped. Unshaped abraders show no evidence of forming or shaping at the edge. The only shaping of these implements resulted from use, in the form of abrasion on one or two surfaces. They are subdivided into eight types according to size, shape and location of abrasion (Plate X g-h). As stated earlier, at the Paul Mason Site, abraders were probably used to manufacture groundstone implements such as points and pencils, and also to manufacture implements of bone, wood and antler. In three cases, pigment stains were visible, indicating that abraders were also used to grind or crush these materials. II-H Shaped Abraders (N=34) Shaped abraders have abrasion on one or both faces, and are distinguished from unshaped abraders in that the edges are formed, usually by squaring the edge to the abrading surface, and grinding the edge flat. They are divided here into four types (Plate X a-f). Shaped and unshaped abraders have wide spatial and temporal distributions on the Northwest Coast. They are present throughout the Prince Rupert Harbour sequence, apparently with little change in frequency through time (see MacDonald and Inglis 1981:46). At Gitaus, abraders occur in highest proportions in the early and late complexes, 246 Plate X a-c trapezoidal abraders; d-e edge f re-worked uni facial abrader; g-h rounded abraders; unshaped abraders 247 but they are also present In the Skeena Complex. Shaped abraders are most abundant fn the Kleanza Complex (see Allafre 1979:45, 1978:300), with ovofd, rectangular and trapezoidal forms present. Farther south, Mitchell (1971:167) describes shaped and Irregular abraders from Montague Harbour, where they.. occur throughout the sequence. 11 — I Grinding Stones (N=5) These Implements have been used extensively as grfndfng tools; each has a concavity on one face. The surfaces of the abraders described above are flat to slightly convex. Two implements fn thfs class are bar-shaped, and have a deep linear concavity extending the length of one face (Plate XI a-c). These implements may have been whetstones, used to sharpen the edges of bone or stone cutting Implements. The other three implements are flat discoid river cobbles of coarse crystalline material (Plate XI d-f). Each has a unifacial, shallow circular depression, with evidence of grinding and pecking. These may be milling stones. Matson (1976:155) reports a probable milling stone from the St Mungo Component of the Glenrose Site, and Mitchell (1971:125) describes two "round abrasive stones" with shallow depressions from Montague Harbour I that are similar to the Paul Mason Site implements. Farther north, Allaire (1978:131) describes four possible milling stones (fragments ^a depressions) from the 248 Plate XI grinding stones 249 Kleanza Complex at Gitaus. Raw materials include sedimentary rock (N=l), argil lite (N=2), and sandstone (N=2). I I I Chipped Stone Tools Chipped stone tools from prepared cores were present at the Paul Mason Site, but like the groundstone tools, the technology was poorly developed. Chipped stone tools constitute 23.6% (N=281) of the total lithic assemblage. Some features of the chipped stone tools that distinguish them from cobble and flake tools include the fact that chipped stone tools tend to be small and we 11-formed (with the exception of retouched flakes). Many implements were manufactured from obsidian. Other lithic materials used rarely in the cobble and flake category include chert, chalcedony and vitreous basalt. Argil lite and other sedimentary rock, which were used frequently in the manufacture of cobble and flake tools were also used in the manufacture of chipped stone tools. The chipped stone category is divided into five classes: formed bifaces, formed unifaces, retouched flakes, chipped cores, and microblades. The lack of development of good quality stone chipping at the Paul Mason Site is fairly typical of most recorded north coast sites. MacDonald (1969) and MacDonald and Inglis (1981) indicate that scrapers, chipped bifaces and chipped projectile points were present in the Prince Rupert Harbour 250 components, but occurred Infrequently, especially during the early and late prehistoric periods. The same Is generally true of the Queen Charlotte Island sites (see Fladmark 1970, 1971; Severs 1974), and sites In the Tlingit area (de Laguna et a 1. 1964; Ackerman 1968, Lightfoot 1983). At many of these sites, stone chipping is limited to flakes, cores and angular fragments, with no formed implements present. In most cases, bone implements greatly outnumber chipped stone implements, and it seems likely that pointed bone tools or groundstone functionally replaced pointed chipped stone tools in many activities. As previously noted, there is strong circumstantial evidence in the presence of the varied and numerous abraders to indicate the presence of a developed bone tool industry at the Paul Mason Site. This may partially explain the paucity of formed bifaces and the virtual absence of chipped stone projectile points. III-A Formed Bifaces (N=33) These implements are sub-divided into six types, based on general form, extent of reduction and edge angle. They all show some evidence of being "formed" by bifacial flaking around the periphery of the implement. Only one projectile point is included here. This is a small, probable dart point, manufactured from obsidian (Plate XII e). It was recovered from the lower dark brown soi1 zone. The remainder of the implements show great variability 251 Plate XI I i chipped bifaces; j biface edge fragment; k-m unilateral backed") bifaces 252 In form and qualIty of manufacture. Many are Irregular and thick, with a circular to oval outline. Of the regular or well-made Implements, four are complete, leaf-shaped bifaces, manufactured from argi1 lite and basalt (Plate XII b-c, f). They have regular bifacial flaking around the entire periphery. Leaf-shaped bifaces are common from Interior sites, less so from north coast sites. The regular, leaf-shaped bifaces from the Paul Mason Site are generally consistent in form, size and manufacturing technique with those from Prince Rupert Harbour (see MacDonald and Inglis 1981:48) and from Gitaus (Allaire 1979: Plate 4a, Plate 5a). They are distinct, however, from the lanceolate points from these two localities. The lanceolate points from Gitaus and Prince Rupert Harbour are long, narrow and parallei-sided, often with square bases. In contrast, leaf-shaped implements from these two areas, and from the Paul Mason Site, are not para 11e1-s1ded, and usually have round bases. A variety of raw materials were used in biface manufacture. The most common were obsidian (N=8), argil lite (N=9), sedimentary rock (N=5) and basalt (N=5). III-B Formed Unifaces (N=13) Formed unifaces Include implements with retouch covering most or alI of one face, reflecting an attempt to shape the implement. They are divided here into three types including two types of steep angled formed unifaces (Plate 253 XIII a-c, f) , and perforating implements (Plate XIII d-e). Steep angled formed unifaces were initially defined by Sanger (1970:76), and later, by Pokotylo (1978:220). The retouched margins of these implements have edge angles o greater than 45 , implying a scraping function. Three scraper re-sharpening flakes (see Frison 1968; Shafer 1971) were recovered (Plate XIII f; Figure 5.18). These are the dulled, retouched edges of steep angled formed unifaces. Each was made from obsidian, which suggests that formed obsidian implements were maintained and curated more than implements based on other types of raw materials. Steep-angled formed unifaces of obsidian and chert occur frequently in all levels at Prince Rupert Harbour and Gitaus. I I I—C Retouched Flakes (N=77) A variety of types of retouched flakes was recovered from the Paul Mason Site (Plate XIII g-y). This class is distinguished from "modified cobble flakes" primarily on the basis of size (smaller), absence of original cortex, and quality of retouch (better). They show no evidence of forming. Retouched flakes are sub-divided into seven types based on edge angle, bifacial or unifacial retouch, and quality and nature of retouch. This breakdown follows Matson (1976). There was a greater preference for steep unifacial retouch (N=49) (Plate XIII h-p) than for acute angle retouch 254 Plate XIII a-c formed unifaces; d-e perforators; f scraper re-sharpening flake; g denticulated retouched flake; h-m regular steep retouched flake; n-p irregular steep retouched flake; q-s regular acute retouched flaked; t-u irregular acute retouched flake; v-w regular bifacial retouched flake; x-y irregular bifacial retouched flake Figure 5.20 Microblades and microcore rejuvenation flakes. 256 (N=28) ((Plate XIII q-u), which suggests that the majority of these implements were used as scraping tools. Bifacial retouch was rare (N=6) (Plate XIII v-y). This is consistent with Gitaus, where Allaire (1978:114-16) reports 121 retouched flakes, of which only six were bifacially retouched. Three denticulated retouched flakes were recovered (Plate XIII g). Denticulated retouched flakes are not reported from other lower Skeena sites. Farther south, Matson (1976:122) reports denticulated flakes from the Glenrose Site. A wide variety of raw materials was used. The most common are obsidian (N=28), chert (N=18), argil lite (N=10), and basalt (N=7). Only four of the chert flakes are green chert. This is in contrast to Gitaus where 57.87, of chipped stone tools (N=41) in the early Skeena Complex were manufactured from green chert (see Allaire 1978:287). That proportion decreased to 14.5% (N=7) in the Kleanza Complex. The high proportion of green chert in the Skeena Complex is associated with the increased manufacture of chipped stone tools in this component. It occurs rarely at the Paul Mason Site, where stone chipping was poorly developed. It also occurs infrequently in the early component at Gitaus (14.37,; N=3), where again cobble tools and groundstone tools predomi nate. 257 III—D Chipped Cores and Pieces Esquillees (N=41) Two stone flaking techniques are represented at the Paul Mason Site - direct, freehand percussion, and bipolar flaking. Direct, freehand percussion was used for most tools in the chipped stone category, and all tools in the cobble and flake category. In this class, chipped cores were all reduced by freehand percussion (Plate XIV j-1). This technique was used on all raw materials represented at the site. This is consistent with most north coast sites (see Lightfoot 1983:98). Bipolar flaking is represented in this artifact class in two types: bipolar cores (Plate XIV m-n) and pfeces esquillees (Plate XIV a-i). In this technique, the implement is held against an anvil, and then struck with a hammerstone. Force is applied simultaneously to both ends. This technique tends to shatter the implement, and is less controlled than freehand percussion. The distinction between bipolar cores and pieces esqui1 lees made here fol lows that made recently by Hayden (1980:2-3), who states that pieces esquillees were tools usually produced on thin flakes (see Figure 5.19). This is contrary to MacDonald (1968), who introduced the term to North America. MacDonald (1968:85-89) argued that pieces esquille'es from Debert were core tools, often thick in cross-section, with evidence of bipolar flaking. Both authors agree, however, that pieces esquille'es are tools. Hayden (1980:3) also states that pieces esquille'es lack * g h a-i pieces esquillfles; cores Plate XIV j-1 chipped cores; m-n bipolar 259 primary flake removals (for micro-tool use), lack flake scars that extend the length of the tool, and often show evidence of the ventral flake scar of the original flake. In contrast, bipolar cores are true cores, sometimes thick, with no evidence of a ventral scar. They show evidence of primary flake removals extending the length of the core. The bipolar technique is evident on a variety of raw materials at the Paul Mason Site, including obsidian (N=5), argil lite (N=7), sedimentary rock (N = 4), quartz crystal (N=3), basalt (N=2), and chert (N=l). This is in contrast to Hidden Falls II, where Lightfoot (1983:98) reports that the bipolar technique was used almost exclusively on cha1cedony. Ill—E Microcores and Microblades (N=118) Although no complete microcore was recovered from the Paul Mason Site, two definite fragments or rejuvenation flakes were recovered. One is made of obsidian (Plate XV a'; Figure 5.20d). It is a mid-section fragment of the fluted surface of the core. There are five parallel blade scars present, each 7 to 8 mm wide. The other is chert (Plate XV b'; Figure 5.20c). It is a relatively thin flake from the fluted surface of the core. The flake appears to have been removed by a force applied to the base of the core, perhaps in an effort to rectify the platform striking angle. There are four parallel blade scars present on the flake. 260 hi g h i r ( i I • ( l m n I I I IM I >3 Plate XV a-z microblades; a'-b' rejuvenation flakes 261 Unfortunately, ft fs Impossible to reconstruct the shape of the cores from the existing fragments. Matson (1976:126) states that the cylindrical mlcrocore form is common along the coast, at least in the Gulf of Georgia region (see also Mitchell 1968), whereas the keeled form is more common in the interior. A total of 116 microblades was recovered. These are blade implements, with length greater than or equal to two times the width (Plate XV a-z; Figure 5.20). They are either triangular or trapezoidal in cross-section. Edge retouch was present on only four microblades, but 57 had evidence of use, in the form of microflaklng or polish. Where present, retouch and use wear were restricted to the lateral edges of the implement. This may be related to side hafting or direct use on some material. In either case, a cutting function is indicated. Two blades show evidence of nicking and tiny flake scars along one edge, while the other edge Is rounded or polished, perhaps as a result of rubbing or abrasion in a haft. There is no evidence of end wear or retouch that might result from use as points. Sixteen complete microblades were recovered. Undoubtedly many mlcroblades were broken accidentally, before, during or after use. However, in light of the preponderance of lateral edge wear, it Is also likely that many blades were purposely broken to create relatively short straight sections, that could be readily hafted. 262 Obsidian was used overwhelmingly in the manufacture of microblades (N=106; 91.4%). Chert (N=8) and quartz crystal (N=2) were also used. As stated previously, there is no local obsidian source in the Kitselas Canyon area. To determine the source of the obsidian, two microblades and three unretouched obsidian flakes were submitted for X-ray flourescence analysis by D. Godfrey-Smith of Simon Fraser University (Godfrey-Smith 1984). The results (Figure 5.2 1; Table 5.2) show that the microblades were made of obsidian from the Anahirn area in the south-central interior of British Columbia. The other three flakes of obsidian are from the Mt. Edziza flow. Microblades have been recovered from a number of prehistoric components in the southern and central interior. The best known of these are the components from the Lochnore-Nesikep locality in the Southern Plateau (Sanger 1970). The dating of these components has come under considerable recent scrutiny (e.g. Stryd 1973). According to Fladmark's (1982a:128) most recent synopsis, microblades in the southern and central Interior are not known to be present in quantity before ca. 6600 B.P., although a 7500 B.P. date exists for one microblade from the Drynoch Slide Site (Sanger 1968). On the central coast at Namu, obsidian microblades are also present at ca. 6000 B.P., and one dates to ca. 7800 B.P. (Carlson 1979:220). The obsidian source for most of these microblades is Anahirn Peak (Carlson 1979:220). 263 scatter peaks Figure 5.21 Element spectre for Anahirn I and Mt Edziza III obsidian (source: Godfrey-Smith 1984). 264 Table 5.2 Element Peaks for Five Obsidian Samples from the Paul Mason Site Obsidian Sample Element 2872* 7161 6941 8279 9493 K 0.77 6.84 5.72 0.99 0.85 Ca - - - - -Ti 0.22 + 0.29 0.26 0.21 Mn 0.06 2.30 2.59 0.17 0.03 Fe 33.93 108.92 102.51 33.61 34.13 Zn 1.62 15.60 16.18 1.74 1.77 Ga - - - - -Pb/As 0.33 7.32 4.73 0.59 0.59 Rb 9.45 179.80 183.70 9.30 9.07 Sr + 4.05 3.46 + + Y 10.16 105.10 105.90 10.00 10.16 Zr 100.00 100.00 100.00 100.00 100.00 Nb 13.79 135.90 132.10 14.10 13.88 * artifact catalogue number - 7161 and 6941 are microblades + element below detection limits - element present in sample but cannot be reliably extracted source: D. Godfrey-Smith 1984 265 Sumrnary This chapter has presented archaeo1gica1 data from the Paul Mason Site, a prehistoric site in Kitselas Canyon to augment the existing data from Gitaus. The Paul Mason Site is deeply stratified, with two discrete soil zones. The surface contours reflect the placement of 10 dwelling floors on the slope, overlooking the canyon wall. This suggests a small village occupation during the later use of the site. A suite of 19 radiocarbon estimates from the site cluster at three distinct time periods: 5000 to 4500 B.P., associated with the lower soil zone; 4300 to 3800 B.P., associated with the bottom of the upper soil zone; and 3200 to 2700 B.P. associated with the upper deposits including the dwelling floors. The poor conditions for organic preservation at the site precluded recovery of non-1ithic artifacts, although it is likely that a bone tool industry existed, based on the presence of numerous abraders. Lithic artifacts were divided into 19 classes. These are further sub-divided into 74 artifact types (see Appendix II). Cobble and flake tools, grouped into four classes, predominated at the site, accounting for 60.97, of all artifacts. These implements were typically large, and expediently manufactured from locally available raw material. The groundstone and chipped stone categories are not well represented. Groundstone tools, grouped into nine classes, constitute 15.57, of the 266 assemblage. Shaped and unshaped abraders were most prevalent in this category, comprising 42.8% (N=77) of all groundstone tools. Hexagonal ground slate points and "pencils", similar to forms from Prince Rupert Harbour, were also recovered. The chipped stone category (23.6% of the assemblage) included bifaces, but lacked projectile points. Steep-edge scrapers were also rare, which suggests that land mammal hunting was not of critical importance to the occupants of the site. The presence of a microblade technology is interesting. This represents the first rad i ometr i ca1 1y dated context for microblades in the Skeena area (see Chapter 6). In the next chapter, the lithic artifacts from the Paul Mason Site are analyzed in terms of their stratigraphic position to identify discrete cultural components. 267 Chapter 6 THE PREHISTORIC SEQUENCE AT THE PAUL MASON SITE This chapter presents the analysis of the lithic data from the Paul Mason Site. The objective Is to delineate discrete cultural compon