Contributions to Circumpolar Anthropology 7 National Museum of Natural History Smithsonian Institution 2008 enter ^^tudies Bryan C. Hood Towards an Archaeologt) of the Main K^gion^ Labrador BRYAN C. HOOD EDITED BY WILLIAM W. FITZHUGH Published by the Arctic Studies Center National Museum of Natural History Smithsonian Institution Washington, D.C. © 2008 Arctic Studies Center, National IVluseum of Natural History, Smithsonian Institution, Washington, D.C. All rights reserved Printed in the United States of America ISBN: 10: 0-9816142-0-5 ISBN 1 3: 978-0-9816142-0-5 Library of Congress Cataloging-in-Publication Data Hood, Bryan C, 1 955- Towards an archaeology of the Nain Region, Labrador / Bryan C. Hood ; edited by William W. Fitzhugh. p. cm. — (Contributions to circumpolar anthropology ; v. 7) Includes bibliographical references and index. ISBN-13: 978-0-9816142-0-5 (pbk. : alk. paper) ISBN-10: 0-9816142-0-5 (pbk. : alk. paper) 1. Indians of North America—Newfoundland and Labrador— Nain Region— Antiquities. 2. Eskimos—Newfoundland and Labrador— Nain Region—Antiquities. 3. Inuit— Newfoundland and Labrador— Nain Region— Antiquities. 4. Nain Region (N.L.)— Antiquities. 5. Excavations (Archaeology)—Newfoundland and Labrador— Nain Region. 6. Ethnoarchaeology— Newfoundland and Labrador—Nain Region. 7. Environmental archaeology—Newfoundland and Labrador— Nain Region. 8. Material culture—Newfoundland and Labrador—Nain Region— History. 9. Historic buildings— Newfoundland and Labrador—Nain Region. 10. Land settlement patterns—Newfoundland and Labrador—Nain Region— History. I. Fitzhugh, William W., 1 943- II. Arctic Studies Center (National IVluseum of Natural History) III. Title. E78.N72H66 2008 971 .8004'97-dc22 200801 2674 The paper used in this publication meets the minimum requirements of the American National Standard for Information Sciences—Permanence of Paper for Printed Library Materials, Z39.48-1 992. Technical Editor: Cara Seitchek Layout and Typography: Jody Billert / Design Literate, Inc. Production editor: Abigail McDermott Printed by United Book Press, Inc., Baltimore, MD This publication is Volume 7 in the Arctic Studies Center series. Contributions to Circumpolar Anthropology, produced by the Arctic Studies Center, National Museum of Natural History, Smithsonian Institution. THIS SERIES IS MADE POSSIBLE IN PART BY THE JAMES W. VANSTONE (1 925-2001) ENDOWMENT. ADDITIONAL SUPPORT FOR THIS VOLUME HAS BEEN PROVIDED BY THE INSTITUTE FOR ARCHAEOLOGY, UNIVERSITY OF TROMS0, NORWAY. Front Cover: Duncan Strong at Nukasusutok Island, August 1928. W. D. Strong collection © Smithsonian Institution, National Anthropological Archives. Negative No. 99-10592. Back Cover: Bryan Hood at Nukasusutok-5, Area 3, in 1980 (Photo: Morten Meldgaard). f Jan 7 "2009 j contents ^^^-^rari^.^ Pre fa ce i TABLE OF CONTENTS iv FOREWORD vi PREFACE vii ACKNOWLEDGEMENTS ix List of Figures xlli List of Tables chapter 1 1 INTRODUCTION The Goals of this Text A Brief History of Archaeological Research in the Main Region and Adjacent Areas Outline of Central and Northern Labrador Culture-History C h a p t e r 2 1 3 ENVIRONMENT OF THE NAIN REGION Overview of the Nain Region Bedrock Geology of the Nain Region Paleoenvironmental Change Recent Seasonal Resource Use in the Nain Region Nukasusutok Island Webb Bay/Port Manvers Run chapter 5 23 THEORY AND METHOD IN THE ANALYSIS OF SOCIAL SPACE Spatial Analysis in Archaeology The Maritime Archaic: Longhouse Models and Spatial Muddles The Functional and Social Dimensions of Maritime Archaic Dw/ellings Expanding the View: Other Archaeological Contexts Ethnography: Social Space and Meaning Methodological Strategy for the Spatial Analysis of Maritime Archaic Sites Paleoeskimos: Axial Features and Bilateral Organization Methodological Strategy for the Analysis of Paleoeskimo Spatial Patterns Notes on the Quantitative Methods. Conclusion: Behavioral Space and Signifying Practices cha pter 4- 53 NUKASUSUTOK ISLAND: OVERVIEW AND SURVEYED SITES Nukasusutok-1 (HcCh-4) Nukasusutok-2 (HcCh-5) Nukasusutok-4 (HcCh-6) Nukasusutok-6 (HcCh-8) Nukasusutok-9 (HcCh-1 1) Nukasusutok-1 (HcCh-12) Nukasusutok-1 1 (HcCh-1 3) Nukasusutok-1 3 (HcCh-1 5) Nukasusutok-14 (HcCh-16) Mount Pickle Harbour-1 (HcCh-17) cha pter ^ 61 NUKASUSUTOK-5; DESCRIPTIVE OVERVIEW The Site and its Environs Archaeological Activities and Field Methods Area 1 Area 2 Area 3 chapter 6 77 NUKASUSUTOK-5; AREA 2 SPATIAL ANALYSIS Limitations and Complications Preliminary Overview Area 2A Area 2B Area 2C C h a p t e r 7 139 NUKASUSUTOK-5: AREAS 1 AND 3 Area 1 Area 3 chapter S 157 NUKASUSUTOK-5: MATERIAL CULTURE Definitions Area 1 Area 2A Area 2B Area 2C Area 3 The Nukasusutok-5 Assemblage in a Comparative Perspective chapter 9 1 77 NUKASUSUTOK-5: INTER-AREA COMPARISONS AND CONCLUSIONS Inter-Area Comparisons Conclusions cha pter 1 1 89 NUKASUSUTOK-1 2: EARLY/MIDDLE DORSET AXIAL STRUCTURES Structure 1 Structure 2 Structure 3 Structure 4 Test Pits 1-3 Test Pits 4 and 5 Surface Collections Overall Assemblage Characteristics and Technological Organization Chronology Structural Comparisons Settlement Pattern and Seasonality chapter 11 213 POST-CONTACT INUIT SETTLEMENT ON NUKASUSUTOK ISLAND Post-Contact Inuit Society and Settlement Patterns in Labrador An Ethnohistoric Sketch of the Nukasusutok Settlement The Archaeology of Nukasusutok-7 (HcCi-8) The Archaeology of Nukasusutok-8 and W. D. Strong's Grave Collection Conclusion: Big-Men, Big-Women and Negotiating the 18^'^ Century European World System chapter 12 245 WEBB BAY/PORT MANVERS RUN: SMALL SCALE INVESTIGATIONS Port Manvers Run-1 (HfCj-5) Double Island Cove-1, L-1 (HeCj-1) Sunset Point (HeCj-8) Attu's Bight-1 (HeCk-4) Webb Point (HeCk-6) chapter 1 ^ 273 ATTU'S POINT: A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY Overview Excavation and Collection Methods L-1 Excavation L-5 Excavation L-7 Excavation L-9 Test Excavation Other Loci Inter-Locus Comparisons Conclusions C h a p t e r 1 -f 319 THE STRUCTURATION OF MARITIME ARCHAIC/PRE-DORSET SOCIAL BOUNDARIES IN LABRADOR Culture-Historical Background Working at Theoretical Boundaries: "Landscape " as a Platform for Dialogue? Partial Readings: Navigating the Boundaries Ecological Space Constructed Social and Ideological Landscapes Back to the Archaeological Context Paleoenviron mental Considerations A Note on Other Cultural Variables Navigation 1 : An Ecological Narrative Navigation 2: A Social and Ideological Landscape Narrative Conclusion Chapter 1^ 347 CONCLUDING REMARKS 351 REFERENCES CITED 363 INDEX foreword BY WILLIAM W. FITZHUGH In 1969 while in the second field season of my PhD work in Hamilton Inlet, Geoff Conrad, Peter Wells, and I flew 200 miles north to survey two locations at the for- est-tundra boundary in northern Quebec-Labrador. One of these locations was on the George River, where Dillon Wallace, Mina Hubbard, and W. B. Cabot had described Naskapi (today's Innu) camped in the early 1900s to intercept the annual caribou migration in August and September at Indian House Lake (Hutte de la Sauvage) near the Labrador border in northeastern Quebec. Here we found many historic Innu camps at the northern edge of the forest. This country was harsh and its scoured surface still bore the scars of continental ice sheets. For the Innu who lived here, starvation was always just a breath away, for if the 'deer' migration changed the Innu were often too weak to re-locate. After a few days we flew east to the comparative paradise of Village Bay, where Port Manvers Run joins the Labrador Sea at the southeast side of the Kiglapait Mountains. To the north lay the Qkak archipelago; to the south the island archipelago east of Nain, then the northernmost town in Labrador, settled by Moravian missionaries in 1771. As we waded ashore in the blis- teringly cold water, the contrast with the interior bar- rens was stark indeed. It wasn't that the land was so dif- ferent, for it had the same stunted spruce trees and tan- gled alder thickets, the same patches of wind-blown sand and winter-killed tree stumps. What was different was the wealth of resources provided by the ocean. The bountiful Labrador coast waters made the barrens seem like the most desolate place on earth. If you had a Iv choice - and in the competitive world of early Labrador history, one often did not - you would choose the coast, and if you know how to hunt sea mammals and navi- gate a treacherous coastline, it could sustain larger and more sedentary communities. Our brief visit to Village Bay and Thalia Point demonstrated great archaeological potential, with finds of Inuit camps dating to the past 5-600 years, the first Pre-Dorset and Dorset sites to be found in Labrador, and scattered traces of yet-to-be defined early Indian cultures. The region was well-described in detailed annual reports of the Moravians in the 18-1 9th C. and some ethnographic work had been done by geologist E.P. Wheeler, whose student, Stearns (Tony) Morse, and Peter Johnson had studied its geology and geomorphol- ogy. Wheeler and Morse were still working in the Nain area when we arrived, as was Terje Brantenberg, a social anthropologist. My brief excursion to Thalia Point convinced me that Nain, which lay just south of the coastal tree-line, was more central to the question of Indian-Eskimo boundary dynamics than Hamilton Inlet, where I had been working on this problem sincel968. Beginning in 1 973 I acquired a boat and started making excursions to Nain from Hamilton Inlet, and in 1975 1 began to focus exclusively on this area. In 1 977- 78 I undertook a large survey of the coast north of Nain, as far as Killinek and Port Burwell at the northern tip of Labrador. By this time Bryan Hood had become part of the field team, and when we returned to contin- ue work in Nain in the early 1 980s, Bryan began his own project at a large Maritime Archaic site at Nukasusutok 5. In later years he researched other sites at Nukasusutok and explored the inner fjords north of FOREWORD Main. Hood's work more than complements our earlier studies; he has advanced it in important ways, making the first detailed published test of the Maritime Archaic longhouse model, exploring Maritime Archaic- Paleoeskimo contacts, and expanding knowledge of Dorset settlement patterns. This volume is the first detailed description of Nain culture history to be pub- lished, spans a period of 6000 years, and includes a large number of sites from different environmental zones, from outer coast to interior bays and fjords. Hood's analytical study of settlement data using k- means and other methods is a model approach to the study of settlement information and the first such study to appear for any arctic or subarctic region. Description of Duncan Strong's finds from the early historic Inuit village at Nukasusutok, housed at the R.S. Peabody Foundation in Andover, Massachusetts, brings his study up toward the present day. While not a complete scenario of the early history of Nain it is a fine start at creating an accessible published literature on the archaeology of northern Labrador. I thank Bryan for his long and continuing dedica- tion to the peoples and cultures of Nain, and for his meticulous description and analysis of hard-won field data. It gives me great pleasure to publish this work through the Arctic Studies Center as the first of several future monographs on Smithsonian-related Labrador research. We dedicate this volume to the people of Nain in gratitude for the support and encouragement they have provided our collective efforts for many years. Heading home to Nam over the sea ice east of Strathcona Run, April 1996. (Photo. B. Hood) FOREWORD V reface The research described herein was undertaken in two phases separated by twelve years. The first phase tran- spired on Nul 2.5 Ca). In the Nain region, these outcrop primarily in a band between Okak and Webb Bay. Otherwise, most of the rocks that make up the Nain archipelago are part of the Middle Proterozoic Nain Plutonic Suite (1 .6-1 .0 Ca), which consists mostly of anorthosite, troctolite, diorite and granite (Ryan 1990; Ryan et al. 1995). This bedrock geology does not provide a great deal of raw material that would be sought after by stone knappers. Vein quartz is abundant throughout the region and was used frequently in the Early-Middle periods of the Maritime Archaic. A particularly large quartz vein occurs on one of the Red Islands, east of Paul Island. A prehistorically quarried quartz crystal outcrop, possi- bly used by Pre-Dorset, is located on "No Name Island" just east of Nukasusutok Island (Fitzhugh 1 981 :30). Aside from those mentioned above, virtually all the lithic materials required for prehistoric technology had to be acquired from outside the Nain region. Ramah chert, used extensively by the Maritime Archaic, Dorset and Pre-lnnu peoples, is derived from the Ramah Croup, which outcrops between Ramah Bay and Nachvak Fiord, ca. 300 km north of Nain (Cramly 1978; Lazenby 1980). Mugford chert, used extensively by Pre-Dorset, but also by Early/Middle Maritime Archaic and Intermediate Indian peoples, is found in the Mugford Croup, 1 50 km north of Nain between Okak and Napartok Bay (Cramly 1978; Lazenby 1980). Slates were used for ground tools by the Maritime Archaic, the Thule/lnuit and occasionally by Dorset peoples. Specific slate sources have not been identi- fied, but it is likely that many slates were derived from the Mugford Croup. Soapstone sources in central and northern Labrador have been documented from an archaeologi- cal perspective by Nagle (1 984) and from an economic geology perspective by Meyer and Montague (1993, 1994, 1995). No soapstone sources are currently known in the Nain area. The place name Soapstone Tickle, between the Red Islands, hints at a possible source, but the location is not known to local carvers. Otherwise, the nearest sources of soapstone are either 100 km north of Nain in the Okak region or 80 km to the south at Freestone Harbour near Utshimassits (Davis Inlet). Paleoenvironmental Change The post-glacial marine emergence chronology of the Nain area is based largely on radiocarbon dated archaeological sites on raised beaches (Fitzhugh 1977, Clark and Fitzhugh 1990, 1992). The oldest radiocar- bon date in the area is 7800±130 B.P. (reservoir cor- rected 7760±65 B.P.) on a geological sample of marine shells from 36 m on South Aulatsivik Island (Johnson ENVIRONMNENT 15 1 985:75; Clark and Fitzhugh 1990:301). The earliest archaeological radiocarbon date is 7065±70 B.P. from the Early Maritime Archaic Ballybrack-1 site, located at 41 m on South Aulatsivik Island (Fitzhugh 1978:85; Clark and Fitzhugh 1990:301). When these dates and elevations are linked with later dates from archaeolog- ical sites as well as a marine limit determination of 71 m, the estimated uplift curve suggests a date of ca. 8500 B.P. for the marine limit and the deglaciation of the outer coast of the Nain archipelago (Clark and Fitzhugh 1990:302). There have been suggestions that uplift rates dif- fer between the outer coast and the inner bays. According to Clark and Fitzhugh (1992:200) for Hamilton Inlet and Johnson (1 985:74) for the Torngat Mountain region, the inner bay marine limit is younger and lower. In the Nain area, Johnson (1969:1 10) iden- tified the marine limit at Webb Bay as lying at 41 m, substantially lower than the 71 m determined ca. 22 km to the east on South Aulatsivik Island. More recent investigations along Reid Brook between inner Voisey's Bay and inner Anaktalak Bay— south and 1 5 km west of Johnson's research area— place the marine limit at 93±2 m and provide a radiocarbon date of 7580±70 B.P. (7485±45 B.P., reservoir corrected) for marine shell at 50 m (Bell 1997). Johnson's marine limit must therefore be incorrect and the proposed differences in outer/inner coast uplift regimes are contradicted by the new data from Voisey's Bay. Figure 5 depicts the relative sea level curve pre- sented by Clark and Fitzhugh (1990) together with Bell's (1 997) suggested curve for Voisey's Bay. The lat- ter should be treated with caution since it is a simple exponential projection based on the local marine limit, the single radiocarbon date and undated marine ter- race elevations. Nonetheless, the Voisey's Bay curve suggests greater uplift in the inland area, which would be expected from an ice sheet that was thicker inland and thinning towards the coast. The archaeological 100 01 23456789 10 BP (1000) 5/ Relative sea level curves for Nain and inner Voisey's Bay (after Clark and Fitzhugh 1990 and Bell 1997). Sites marked: Nukasusutok-5, Area 2 (N5A2), Nukasusutok-5, Area 3 (N5A3). Attu's Bight (AB), Attu's Point (AP) and Nukasusutok 12 (N12). Study area at Webb Bay probably lies somewhere between these uplift profiles. Some of the archaeolog- ical sites discussed in the text are marked on the dia- gram; their positioning will be discussed later. During the 1970s, interest in the relationship between climate change and culture change (e.g., Fitzhugh 1972, 1977a) was pursued through the col- lection of pollen data from several locations in north- ern and central Labrador. As part of this program, one core was taken from Nain Pond (Short 1978), but it proved problematic because of reversed datings in the stratigraphic sequence. Similarities with cores taken in the Okak area, however, permitted reasonable extrap- olations. Cores were also taken in 1939 by Wenner (1947), but these are of limited utility since they pre- date the use of radiocarbon dating. Generally, the pollen data indicate the area was deglaciated by 10,500 B.P. and that by 9000 B.P. the ice front stood cf. Tanner's (1 944:245) determination of the marine limit at 83 m on Sandy Island, 20 km east of South Aulatsivik Island. 16 CHAPTER 2 on the interior plateau, at least 1 00 km from the coast. The post-glacial period was marked by tundra vegeta- tion until a transition to shrub tundra with a strong birch and alder component ca. 6700-6500 B.P.. After 4500 B.P., an open spruce woodland was present. Climatic cooling is indicated after 3500 B.P. (Short 1978:32). Initial efforts at dendrochronology in the Nain area began in the early 1970s, but the information only extended back to 1 769 (H.E. Wright, International Dendrochronologicai Data Base, National Oceanic and Atmospheric Administration [NOAA] 2003). Additional data are currently being collected, but the work is still at a preliminary stage. The other source of climate- related information is the oxygen isotope data from the Greenland ice cores. These data provide much higher chronological resolution than the pollen cores, but their high frequency variations can be difficult to interpret and attempts to correlate them with coarsely dated archaeological entities are problematic. In Chapter 14, ice core and pollen data are combined to discuss environmental changes relevant to under- standing shifts in Maritime Archaic settlement and the social boundary problems subsequent to Pre-Dorset colonization. Recent Seasonal Resource Use in the Nain Region Brice-Bennett (1977) outlines the seasonal varia- tions in Nain region fauna and the concomitant pat- terning of human activities during the 20^'^ century. The following discussion is based on her account as well as Taylor (1974), with supplemental information derived from local informants and personal observa- tion. The coastal resources are discussed first, fol- lowed by inner bay/interior species. In early spring Canada geese and various migrato- ry duck species return north and are hunted in early to mid-May. Later in the spring and the early summer, eggs are collected from duck and gull nesting areas. After break-up in mid-June, the migratory harp seal moves north from its whelpmg grounds off Newfoundland following schools of capelin. Large numbers can be netted or shot as they enter the bays traveling in herds. Although most of the harps summer off west Greenland they can occasionally be found in Labrador waters during the summer. The ringed seal is present year round in the Nain area, although ecologi- cal studies by Boles et al. (1980:54) suggest that dur- ing the summer some may migrate from the central coast to locations north of Okak. Harbor and grey seals haul out on seaward islands in the spring and break-up is said to be a good time for hunting bearded seals. During the open water season seals are most frequent- ly encountered in the outer island areas, but many move up the bays with high tide. Fishing was the major summer activity in the Nain area until the recent downturn in fish stocks. Arctic char begin their seaward runs from the rivers in early June and spend the summer among the islands until mid-August, when they return upriver to spawn. Salmon are fished in late July and August and cod in August and September. Minke and white whales are seen occasionally. Bowhead whales once frequented the Labrador coast, although they were rarely hunted by 18'^'^ century Inuit in the Nain area, possibly because poor bathymetric conditions reduced the like- lihood of their entering the bays (Taylor 1 988: 1 25). In the fall, Canada geese are pursued beginning the last week of August. Their main southward move- ment occurs in mid-September and few remain after October 25. Ducks and various sea birds are also exploited. In September, ringed, young bearded, grey and harbour seals move into the bays to feed. Brice- Bennett (1977:144) notes that the beginning of the harp seal southern migration varies from late September or October to late November or December. The later the commencement of the migration the more likely it is that the harps will stay seaward to avoid the already frozen coastal areas. Late and early 1 9'^'^ century Inuit settlements tended to be locat- ed in good locations for fall harp sealing (Taylor ENVIRONMNENT 17 1974:51) and numerous references in the Moravian Periodical Accounts make clear the importance of the fall seal hunt for storing up supplies for the winter. When ice formation begins in the bays, most of the seals that had previously been feeding there shift seaward. Those that are trapped in the bays after freeze-up are hunted basking on the ice, through their breathing holes or in patches of open water. During the winter, ringed seals are found in open water at the sina and at breathing holes. March is the worst month for sealing because pack ice jams against the landfast ice and eliminates the leads of open water. Come April, ringed seals may be encountered basking on the ice beside their breathing holes or at the sima. Walrus are rare today in the Nain region, but they formerly appeared at the sina in February and March. Polar bears move south with the pack ice, inhabiting the sina and coastal headlands from March until May. If the pack ice lingers until later in the summer, both polar bears and walrus may be found occasionally near the outer islands, although the bears sometimes wan- der as far up the bays as Tikkoatokak. Caribou is the main resource exploited by humans in the inner bays and on the interior plateau. Although they may be hunted at any time of the year, their hides are best for clothing in September and October. Brice- Bennett (1 977:1 58-1 59) notes the winter core areas for interior caribou herds, the nearest concentrations to Nain being in the Kiglapait Mountains, near Kingurutik Lake and between Tasisuak Lake and Anaktalik Brook. The geographical range of caribou varies over time. Prior to the mid-1960s, caribou primarily frequented the interior plateau, requiring lengthy hunting trips inland (Brice-Bennett 1 977:1 55). For example, in 1 780 Moravian missionary William Turner accompanied Inuit caribou hunters on two trips up the Fraser River valley and onto the plateau, up to 80 km inland from Nain (Taylor 1 969). At this time, prior to the Inuit accessing firearms, the Inuit brought kayaks inland on their late summer hunt (August-September) and travelled with dog sleds on a winter hunt (February). Meat might be cached for emergency use later in the winter. In more recent years, large caribou herds have moved out towards the coast, crossing over the sea ice to the outer islands. Most of these animals return inland before break-up, but some are trapped on the islands for the summer. Caribou are also encountered sporad- ically in the inner bays between spring and fall. Other animal species found in the interior or along the inner bays include: black bear, fox, wolf, arctic hare, wolverine, otter, mink, muskrat and lynx. Black bears, hare and fox can also be found on some of the larger coastal islands. During the winter, wolves may move over the sea ice to the islands in pursuit of cari- bou. The economically significant terrestrial birds are rock and spruce ptarmigan, which today are hunted in April/May when there is a heavy snow cover and it is still possible to travel over the sea ice. Nukasusutok Island Nukasusutok Island (NaKasetjutok in the current orthography) is located about 30 km southeast of Nain. According to Wheeler (1953:62-63), the island's Inuktitut name means "the place where the brothers quarreled." That name may be illuminated by ethnohis- toric information suggesting that during the 1 gth cen- tury brothers rarely shared households because of conflicts over authority within multi-family communal houses (Taylor 1 974:82). The island (Figures 3, 16) is approximately 10 km in length (east-west). It is indented on the north and the south by two bays; the southern bay is commonly known as Wyatt Harbour. According to geologist E. P. Wheeler II (cited in Morse 1 971 :94), Commander A.B. N. Wyatt of the British Navy survey ship Challenger (see Wyatt 1 934) attested to the fine anchoring qualities of these embayments by maintaining— doubtless with a degree of exaggeration— that the entire British Navy could be accommodated here. Less expansive, Morse (1971:95) noted that Wyatt Harbour is among the finest on the coast of Labrador. These bays effectively divide the island at its mid-point into two somewhat 18 CHAPTER 2 physiographically different portions. The western end of Nukasusutok is 3 km in width, contains the highest point on the island (298 m) and is generally rugged and mountainous. The eastern end is a thin northeast trending arm that ranges from 0.3 to 1 .3 km in width. This extremity is less rugged than the western tip of the island, has more flat low-lying areas, and eleva- tions do not exceed ca. 152 m. At both ends of the island most of the shoreline plunges precipitously down to the sea. Geologically, Nukasusutok Island is part of the Middle Proterozoic Nain Plutonic Suite and is underlain by gneiss, anorthosites, intrusive troctolites, diorites and monzonites (Davies 1974; Ryan 1990; Ryan et al. 1995). A combination of faulting, intrusions and weathering has produced a topography of rounded hills punctuated in places by parallel-walled "passes." The only lithic material on the island that was of pos- sible utility to prehistoric peoples is vein quartz, which outcrops in several places on the island. The vegetation of Nukasusutok is primarily that of a tundra community with plant species such as lichen, willow, dwarf birch, alder, Labrador tea, crowberry, blueberry, cranberry, cloudberry and various grasses. Although the exposed eastern arm of the island is strictly a tundra environment, certain well sheltered areas on the western end maintain substantial groves of black spruce. Spruce trees at the head of Wyatt Harbour have attained heights in the neighborhood of 3.0 to 4.5 m. The raison d'etre of any human occupation of Nukasusutok would be exploitation of the sea mam- mals, aquatic birds and fish found within the island's catchment area. Good spring sealing may be had to the southeast between Satok and Humby's Island. During the summer, seals are common in the "Bridges" pas- sage to the west, off the east side of Kikkertavak Island and among the small islands immediately to the east of Nukasusutok. Numerous aquatic birds may be had in the area during the open water season: black and eider ducks, teals, scoters, black guillemots, loons and gulls. The small islands to the east of Nukasusutok are a particularly good area for these waterfowl. Geese are seasonal visitors and may be encountered in the larg- er ponds on Nukasusutok. Ptarmigan also occur there. Arctic char, tom cod and occasionally salmon may be caught in the waters surrounding Nukasusutok. Minke whales sometimes move into Wyatt Harbour to feed. In contrast to the diverse and abundant marine fauna, Nukasusutok exhibits a paucity of terrestrial game. The arctic hare is a resident of the barren hills and several Inuit stone traps testify to the presence of fox. Caribou are known to frequent the island on rare occasions (such as 5-6 animals in 1992), crossing over the sea ice during the winter and becoming stranded after break-up. Several black bears were resident at the western end of the island in 1992. Webb Bay/Port Manvers Run Webb Bay is located 2 5 km north of the town of Nain (Figures 3 and 1 59). It is named after the settler fami- ly Webb, that maintains cabins at the head of the bay, but it is also known by the Inuktitut name Udjuktok, place of bearded seals (Wheeler 1953:94-95). Extending 1 5 km east-west by 3-7 km wide, Webb Bay is separated from the outer coast by South Aulatsivik Island, the largest island in the Nain archipelago. Access by boat is either from the south through the "First Rattle" at the tip of the Itilialuk Peninsula, or from the north via Port Manvers Run. The latter is a fjord-like passage that swings around the western side of South Aulatsivik and enters the sea at Thalia Point (Tikigdtsuk). The Run is significant as a transportation route because it constitutes an "inner passage" that can be used by boat travelers to avoid the exposed outer coast to the east of South Aulatsivik Island. During the winter, hunters can travel northwards to Okak by snowmobile or dogsled, either by going to the head of Webb Bay and following the Webb Brook Valley into Tasiuyak Bay, or by moving north up Port Manvers Run and then crossing the plateau behind the Kiglapait Mountains and thence down into Tasiuyak. ENVIRONMNENT 19 As far as game resources are concerned, Webb Bay is considered to be a "core area" for hunting ringed seals (Brice-Bennett 1 977:1 25), particularly in the spring. During the summer seals tend to be scarce, but ringed and harbor seals may be shot occasionally, par- ticularly off the point at the southern entrance to Port Manvers Run. The most important location for sealing along Port Manvers Run is the "Second Rattle," a nar- rowing of the Run which has such a strong tidal current that it remains an ice free polynya throughout the win- ter. This is also a "core area" for harbor seals (Brice- Bennett 1977:125). The narrow "tickle" between Igloo Island and the north shore of Webb Bay sometimes does not freeze over until February, providing extend- ed open water sealing possibilities, and it may be a good sealing location in spring and fall. The aforementioned winter travel routes can be used to access either the Kiglapait Mountain caribou herd to the north, or the Kinguritik herd to the west. Caribou may frequent the Webb Bay/Port Manvers area throughout the year, although they can be hard to find during the summer when they tend to be up the stream valleys and at relatively high elevations. In mid-April 1994 a small caribou herd was present in the interior valleys of South Aulatsivik Island. In late April tracks indicated a recent westward movement of this herd across Port Manvers Run and back towards the interior, presumably a shift towards spring calving areas. Black bears are also very common in the region, inhabiting forested areas near streams and wandering up to fairly high elevations on the mountains. Polar bears move through Port Manvers Run during the win- ter and may be found there occasionally in the sum- mer, especially in the northern portion from the Second Rattle to Thalia Point. Wolves and arctic hare also occur in the area, as well as porcupine and otter. Webb Bay Is a good location for wildfowl in the spring and fall, particularly geese and ducks (black and harlequin). Webb Brook at the head of the bay is an important salmon and char spawning stream (Brice- Bennett 1 977:133), but summer fishing can be variable In the bay. In 1992-94 the summer fishing was poor, while in early summer 2002 there was good char fish- ing near the southern entrance to Port Manvers Run. The bedrock at the western end of Webb Bay is Archean gneiss, while most of the remainder of the Webb Bay/Port Manvers Run area consists of anorthosites of the Middle Proterozolc Main Plutonic Suite. At the north end of the Run, extending north towards Okak, is the Kiglapait layered intrusion, a gab- broic body also dating to the Middle Proterozolc (Ryan 1 990; Ryan et al. 1 995). The only rock type of use to prehistoric peoples would have been vein quartz, which outcrops sporadically throughout the area. Scattered observations concerning the glacial geology of the Webb Bay/Port Manvers Run region were made by Vaino Tanner (1 944) and his associates, but the only detailed study is that of Peter Johnson (1969). Johnson studied the extensive glacio-fluvial and glaclo-marlne sand and gravel terraces on the north side of Webb Bay and Identified six phases in their formation. Early phases Involved the wasting of Ice into separate units centered on Webb Bay and Attu's Brook, then the development of a series of ice- dammed lakes and kame terraces at progressively lower elevations with runoff towards the east. A large marginal lake then formed from ice meltwater and runoff from the deglaciated highlands, contributing to subglaclal drainage down Port Manvers Run. According to Johnson, the Ice then withdrew sufficiently such that the eustatically rising sea cut a terrace at 41 m, which he considered to be the local marine limit. As noted previously, however, more recent data from Voisey's Bay (Bell 1997) suggest Johnson's 41 m marine limit Is incorrect. If we situate Webb Bay between the Voisey's Bay and Nain uplift curves shown in Figure 5, then a terrace cut at 40 m would date ca. 7000 B.P. 20 CHAPTER 2 Scenes from a bygone era: trapboats in Nam Harbour, 1978. (Photo: B. Hood) ENVIRONMNENT "Wet-Site excavation," Nukasusutok-5, Area 2C, 1980, with Doug Sutton and Morten Meldgaard (right). (Photo: B. Hood) CHAPTER 3 eorij and Method in the Analtjsis of Social Space One of the central themes of this text is the analysis of social space. In this chapter the theme is addressed at the scale of site structure and individual features. First up is a general discussion of the status of spatial analy- sis in archaeology, which focuses on some of the main trends relevant to the set of problems outlined here. Current interpretations of Maritime Archaic social rela- tions and dwelling space are then outlined and their explicit or implicit theoretical and methodological approaches are assessed. Given these considerations, a methodological strategy for the analysis of Maritime Archaic dwelling spaces is presented. A similar treat- ment of Paleoeskimo material follows. Spatial Analysis in Archaeology Spatial analysis of archaeological sites became an important theme in the late 1 960s as part of the New Archaeology's program of finding appropriate meth- ods for addressing conceptual problems. Much of the initial impetus was related to studies of functional vari- ation within and between sites, which involved identi- fying functionally distinct tool kits. Simple quantitative techniques based on grid counts and point data were borrowed from geography and ecology, such as near- est neighbor analysis and tool type correlation studies (e.g., Hodder and Orton 1976). It became evident, however, that archaeological data were not a direct reflection of behavior and that the attempt to identify tool kits through correlation studies was problematic (Speth and Johnson 1976). Recognizing the complex relationship between what Schiffer (1972) termed the archaeological and systemic contexts, attention turned to documenting the formation processes of the archae- ological record. Much of this involved ethnoarchaeo- logical studies of site structure (e.g., Yellen 1977; Binford 1 978, 1 983), which sought to identify "transla- tion rules" for inferring systemic context from archae- ological context. At the same time, a more varied range of quantitative methods was developed for rec- ognizing patterns in complex spatial data, such as k- means cluster analysis, unconstrained clustering, cor- respondence analysis etc. (e.g., Blankholm 1991; Carr 1984; Hietala 1984; Kintigh and Ammerman 1982). Subsequent work investigated site structure in terms of formation history, considering specific problems such as the effects of site maintenance behavior on patterning (Binford 1983, 1987; O'Connell 1987). As noted by Wandsnider (1 996), monitoring such process- es may require dropping traditional analytical entities such as artifact types in favor of novel units of analy- sis that are more relevant to formation processes, such as size-sorting indices employed as indicators of site maintenance behavior (Wandsnider 1996:353-359), or refitting studies used to reveal the micro-behaviorial level of site structure (Cziesia 1 989, 1 990). Whatever the methodological sophistication of spatial analysis, the central problem is the inferential process of connecting the observable archaeological context with the unobservable past cultural system. Ethnoarchaeological studies have been crucial for THEORY AND METHOD IN THE ANALYSIS OF SOCIAL SPACE 23 "Me.n'6" Out^side:- He/\r.th Model- 1 O 1 2 3-45 6/ Binford's external/internal hearth models. (6a) external hearth model (from Binford 1 983: 1 53). 6b/ internal hearth model. Palanga's house (from Binford 1983:177). Reproduced by permission. understanding the relationship between archaeological traces of behavior and the organizational variables generating those traces (Binford 1987). The most fre- quently cited ethnoarchaeological work on hunter- gatherer site structure is that of Binford (1978a, b, 1983, 1987). From the late 1970s through the 1980s Binford focused on human body mechanics as a univer- sal constraint that would produce consistent spatial patterns. The most well known example is the so- called "hearth-centered" model (Binford 1978b, 1983). According to Binford, the positioning of individuals around an outdoor hearth results in a characteristic pattern of a "drop zone" containing small items in the vicinity of the hearth and the seated individuals and a "toss zone" in front of the hearth and behind the hearth-users, where larger items are discarded (Figure 6a). Hearths inside dwelling structures do not exhibit this pattern since they are subjected to regular mainte- nance procedures of gathering debris and dumping it outside the dwelling (Figure 6b). The methods used for developing and presenting these models were largely qualitative and visual, with the goal of defining gener- al spatial patterns that could be regarded as behavioral configurations. Nonetheless, Binford (1981:286) emphasized that spatial data should be collected at as detailed a level as possible, since we do not yet know enough about spatial structures to justify coarser- grained recording. Binford used the hearth-centered model to provide an alternative interpretation of the French Paleolithic site Pincevent. According to Binford (1983:147-148), Leroi-Courhan and Brezillion (1972) imposed certain unevaluated assumptions when they concluded that the spatial patterning of debris associated with hearths indicated the presence of several tent structures. Instead, Binford (1 983:158-159) suggested the pat- terns derived from a rotation of seating positions around outdoor hearths. Debate continues as to how the spatial organization of Pincevent and other Upper Paleolithic sites should be interpreted (e.g., Audouze 1 987; Carr 1 991 ; Enloe et al. 1 994; Julien et al. 1 987; Koetje 1987, 1994; Simek 1984). Meanwhile, some researchers attempted to develop methodological tools for the analysis of hearth-centered distributions (Stapert 1989), while others discussed additional site formation considerations that complicate Binford's hearth-centered model (e.g., O'Connell 1987; Stevenson 1991). 24 CHAPTER 3 One of the problems with Binford's studies is that they mostly attribute spatial patterning in sites to "etic" behavioral factors such as body mechanics or practical "maintenance" strategies that determine disposal deci- sions. Although processualists eventually paid more attention to social factors such as the role of sharing, kinship and cooperative labor in the spacing of dw/ellings (e.g., Binford 1991; Cargett and Hayden 1 991 ; Whitelaw 1 991 , 1 994), there was no concern for the symbolic dimension of site structure. Symbolic meaning was taken up in Hodder's (1982, 1987) eth- noarchaeological research in east Africa, which explored the structuring effect of social relations, such as gender, and the symbolic aspects of discard related to concepts of dirt and purity (see D. Gordon [1980] for an earlier eastern subarctic example). In subse- quent post-processual studies, dwelling structures were seen as both the medium and outcome of social structuration processes, simultaneously channeling the physical movements of individuals and serving as symbolic metaphors encoding aspects of ideology (e.g., Hodder 1990:44-70; Parker-Pearson and Richards 1994; Yates 1989). Recent research takes a phenomenological rather than structuralist approach to constructed space, theorizing how "the body" sub- jectively experiences movement through dwellings and other aspects of the inhabited environment (Tilley 1994, 1999:40-49). The upshot of this discussion is that there are two main methodological approaches to spatial analysis (although these are not mutually exclusive). The first approach is model-centered (Carr 1991). This general- ly involves linking the observed archaeological context with an ethnoarchaeological model that provides insight into the unobserved past behavioral context. The model employed could be either an "etic" behav- ioral model or an "emic" socio-ideological model. The question of evaluating a model's correspondence with the spatial data is a central problem. There is a con- stant danger that model-centered approaches will assume what they set out to prove, and that fitting the model to the data will result in an uncritical accommo- dation argument (Binford 1981:284-285) in which the model becomes an unevaluated interpretative conven- tion or a rhetorical device used to construct a past-as- wished for. From the neo-positivist perspective, although data are recognized as constructed through a theoretical lens it is believed that independent middle- range methods can be applied to provide a measure of goodness of fit (Bmford 1981:21-30, 1982). The hermeneutic perspective, on the other hand, sees method as inseparable from theory, such that evaluat- ing goodness of fit is a negotiation process marked by fluidity and a lack of firm foundations for determining the plausibility of arguments. While data can in some sense "confront" theory, emphasis is placed on enhancing interpretive coherence and reducing incon- sistencies (Hodder 1999:39, 60-62). Nonetheless, there seems to be agreement that some kinds of methodological strategies are necessary to provide challenges to interpretative models and to prevent arguments from becoming self-serving tautologies. These strategies may involve contrasting arguments developed from different paradigmatic positions, applying different types of methods (quantitative and qualitative) with different assumptions to the same data set, and considering several different types of contextual information and different scales of analysis (e.g., Binford 1987:503; Blankholm 1991; Carr 1991; Hodder 1999:130-131). In this process, quantitative analysis can be used as a pattern recognition exercise aimed at exposing possible anomalies between models and data, a tool for resisting premature accommoda- tions. Such flexible methodological strategies might be considered part of a critical realist epistemology (Sayer 1 992, 2000). The other approach to spatial analysis can be termed empirical pattern recognition. Instead of tak- ing its point of departure in a conceptual model, this approach uses exploratory quantitative methods to identify statistically meaningful associations in spatial data. These patterns may then be interpreted post hoc THEORY AND METHOD IN THE ANALYSIS OF SOCIAL SPACE 25 wood cutting conical I I ri . scaffold slioreline tent 1 Recent Innu spring camp, Labrador (a) A duck wings hung on tree tent 2 square scaffold lil and hearth hearth Recent Innu summer tent B stove debris concentration 1 m scale (b) 7/ Recent Innu camp structure, (a) Innu spring camp, Labrador, (b) schematic diagram of a recent Innu summer tent fea- ture, Labrador. in relation to a model. The assumption is that statisti- cal techniques can provide results that are independ- ent of model premises, such that unanticipated pat- terns might be recognized (Carr 1 991 :226). While this property has some virtue, the important objection is that no statistical tool provides guidelines as to how the resulting patterns should be interpreted in terms of behavior. So, while a technique may provide a means of revealing patterning that is inherent in the data itself rather than an artifact of model conventions, the interpretations of the patterning are ultimately model dependent. There is also the danger that the elegance of a technique, such as the popular /c-means cluster analysis (Kintigh and Ammerman 1982; Simek 1984), may promote its use as a cookbook solution. This general discussion has implications for the methodology used in the spatial analysis of habitation areas undertaken in this text. The methodological strategy employed here involves the recursive use of models and empirical pattern recognition studies. Theory is used to frame models of Maritime Archaic and Paleoeskimo dwelling space, while the methods serve two roles: (1) evaluating the goodness of fit between models and data— their degree of accommo- dation or "coherence," and (2) as exploratory tools for partitioning complex data sets so as to expose unan- ticipated aspects of variation— anomalies between models and data that could require re-evaluation of the model assumptions. Methodological discussions for the Maritime Archaic and Paleoeskimo cases are pre- sented separately, since each cultural context presents its own specific problems of analysis. The Maritime Archaic: Longhouse Models and Spatial Muddles During the 1970s, excavations at Maritime Archaic habitation sites in central and northern Labrador failed to identify distinct traces of dwelling structures. Hearths were sometimes arranged in a linear fashion along raised beach terraces (e.g.. Rattlers Bight and Black Island Cove, Fitzhugh 1975, 1978a; Okak-2, Cox 1 977), but these and other features (Hood 1 981 ) were generally interpreted as the remains of individual fam- ily tent structures. Although the Maritime Archaic seemed "different" from the historically known Innu, the viewing lense for perceiving and interpreting these spatial patterns was traditional subarctic ethnography. This ethnographic model generated assumptions of nuclear family organizational units with modest tent structures as dwellings. For example, modern Innu spring-summer camps leave remains such as those sketched in Figure 7. Individual tents may be spaced at varying distances, dependent on both terrain and social relationships, and may be associated with exter- 26 CHAPTER 3 8/ Innu summer tent site, 1980. Small stakes mark stove location, door immediately to the right. Exterior midden between tent site and scaffold. Photo: W. Fitzhugh. nal features such as hearths, wooden scaffolds, skin- stretching frames, wood chopping areas and the sym- bolic placement of animal bones in adjacent trees. Dwelling remains may consist of small, single family tent floors with a central stove, along with trash distri- butions centered on the doorway and exterior area immediately outside the tent (Figures 7 and 8). Larger tents are also used, however. During the 1978 season of the Smithsonian Institution's Torngat Archaeological Project, probable Maritime Archaic rectangular structures were noted briefly at Aillik, near Makkovik on the central coast, while several linear structures up to 100 m long and associated with Late Maritime Archaic artifacts were identified at Nulliak Cove, north of Hebron Fjord (Fitzhugh 1980). The break-through came in 1980, when Smithsonian investigations at Aillik revealed the clearly defined remains of a "longhouse" embedded in a cobble beach. The feature measured in at 28 m long, 4 m wide, and was segmented into seven compart- ments with raised rock dividers. It was associated with several external conical cache pits (Figures 9 and 10). The longhouse was actually the latest in a sequence of structures that began on a higher and earlier raised beach. There appeared to be an "evolutionary" devel- opment over time from small, single-family pit-houses and rectangular structures towards progressively larg- er segmented rectangular structures, culminating in the longhouse. (Fitzhugh 2006). This remarkably clear longhouse and the sequence of earlier structures became a paradigm or interpretative model that could be used to make sense of the previously excavated hearth rows, the linear fea- tures at Nulliak Cove and other ambiguous spatial dis- tributions. The "longhouse model" revolutionized Maritime Archaic archaeology, revealing how the ethnographically derived expectations had hindered recognition of the pattern. The model made visible what was previously hidden and helped make sense out of spatial muddles. We cannot do without it. THEORY AND METHOD IN THE ANALYSIS OF SOCIAL SPACE 27 0°0q O (3 % ° °oo« Ooo" / CACHE PITS o \ TROUGH °° °Oo "o"" O ^? =00 o° =^ ^ o °° oOoO METERS ° °°l AILLIK-2 °°oo° 9/ Maritime Archaic longlnouse, Aillik-2 (after Fitzhugh 1984 33). Nonetheless, the model also poses certain dangers when applied as a handy template to data that are con- siderably more ambiguous than the "classic" feature at Aillik. At the descriptive or pattern recognition level, accurate delineation of dwelling size and configuration could be compromised by naive application of the longhouse template. There may be a temptation to interpret any linear distribution as a longhouse, partic- ularly distributions encountered on surveys and not subjected to substantive analysis. Such pattern recog- nition errors snowball on the inferential level when cor- relations of structure size and social composition are based on faulty description. Consequently, we have the ingredients for potentially deceptive accommodation arguments in which the longhouse model becomes a taken-for-granted interpretative convention. So, when dealing with ambiguous spatial data we have to bal- ance carefully between using the longhouse model as a necessary guide to pattern recognition and employ- ing methodological strategies to prevent unjustified reproduction of the model and to expose unanticipat- ed patterning. The Functional and Social Dimensions of Maritime Archaic Dwellings There appear to be at least three different forms of Maritime Archaic dwellings: (1) small pit-houses of ca. 3.0-3.5 m in diameter (either circular or oval double- room features, Fitzhugh 2006), (2) rectangular surface structures, and (3) "longhouses." The latter may be either surface structures or slightly excavated into cob- ble beaches. The current chronology, based on a com- bination of radiocarbon dating and elevation above sea-level, has pit-houses predating 6000 B.P.. Rectangular surface structures are also found prior to 6000 B.P.; these measure ca. 8-1 2 m in length and are partitioned into two to five segments by rock dividers (Fitzhugh 1984, 1985a, 1985b:62, 1986:56, 2006). Between 6000-5500 B.P., rectangular surface struc- tures at Aillik and in the Nain area range from 9-16 m long and contain two to four internal segments (Fitzhugh 1984:10, 13). A three-segment structure at Aillik West-1 is dated 5210±270 B.P. (Fitzhugh 1984:8). Between 5000-4200 B.P., a 30-40 m line of hearths at Okak-2 (Cox 1 977:1 84-1 85) has been inter- preted as a longhouse (Fitzhugh 1981:18), while at Black lsland-2, in Hamilton Inlet, 12 evenly spaced hearths suggest a 50 m long structure with 1 2-1 3 seg- ments (Fitzhugh 1975:122-125, 1981:17, 1984:13). Longhouse development reached its peak during the Rattlers Bight Phase (4200-3500 B.P.), with a 70 m lin- ear arrangement of hearths at Rattlers Bight in Hamilton Inlet, 27 surface longhouses ranging in length from 15-100 m at Nulliak Cove near Hebron, 28 CHAPTER 3 and the classic 28 m long seven segment semi-subter- ranean structure at Aillik (Fitzhugh 1981:18, 1984, 1986:56, 1 985a:88-89). The time of appearance of the first longhouses is unclear, but likely occurs between 5000 4500 B.P.. One possible conclusion to be drawn from this sequence is that longhouses developed by "accretion" (Fitzhugh, personal communication). Beginning with the basal units of single compartment THEORY AND IVIETHOD IN THE ANALYSIS OF SOCIAL SPACE 29 pit-houses and two segment rectangular dwellings, additional compartments were simply added on in a geometric progression, as clones of the basal units. The question, however, is what constitutes a "longhouse" ? What is the archaeological signature of such a feature and how does it differ from a "rectangu- lar structure" and a linear arrangement of completely independent tents or outdoor hearths? First of all, "rectangular structure" seems to designate surface features that exhibit tent-ring-like rock borders and internal stone segment dividers. The term "longhouse" is ambiguous since it denotes a feature that is also rectangular, but longer— with an unspecified minimum length threshold— and which may or may not display the aforementioned architectural features of rectangular structures. Additionally, the term long- house may be a misleading metaphor since it denotes certain ethnographically well-known dwellings (e.g., Iroquoian) that are not applicable to the Labrador case. Rather than being continuous architectural construc- tions. Maritime Archaic longhouses probably consisted of a linear series of linked individual tents. In the case of the 28 m long structure at Aillik there is clearly an architectural unity defined by its semi-subterranean excavation into a cobble beach, but this may simply have been the foundation for several closely packed tent dwellings. In cases that lack obvious architectural components, such as the surface features at Nulliak, the longhouse is an inference based on several lines of evidence, not all of which may be present or of equal quality in a given case: a linear series of regularly spaced hearths, a bounded linear distribution of lithic material (3.5-4.0 m wide), segment divider rocks and traces of wall border rocks. It is impossible to be more precise about the nature of the spatial patterning associated with rectan- gular structures and longhouses because no detailed distribution maps have been published. Partial plans of longhouses from Nulliak Cove and Rattlers Bight (Fitzhugh 1981: Figures 7 and 10, 1985a:96-97) indi- cate fairly regularly spaced (3-4 m) clusters of charcoal and fire-cracked rock (not always mutually associated) running down the center of the structures (Figure 1 1 a). In some cases, though, charcoal concentrations also occur close to the presumed walls. Tool and debitage distributions are described as being hearth-centered (Fitzhugh 1981:17-18, 1985a:98, 1985b:49), but the details of this spatial configuration are not specified. A schematic diagram of a longhouse from Nulliak Cove shows lithic distributions that are not always associat- ed with possible hearths (Figure lib; Fitzhugh 1985a:97). Some of the lithic concentrations are spa- tially independent, but most of them are joined along one side of the structure, suggesting an overlapping arc configuration or "ribbons" connecting separate concentrations. These examples should at least raise some questions as to how we might distinguish long- houses from lines of independent tents or outdoor hearths. In sum, our knowledge of Maritime Archaic dwelling structures is constrained by ill-defined terms and piecemeal information. The term "longhouse" orig- inated as an ad hoc category (Barsalou 1 983) formulat- ed to make sense of linear spatial distributions. Although useful for that initial purpose, the category needs to evolve further because considerable ambigu- ity results from its status as a conceptual combination in which archaeological attributes are mapped onto analog models from other contexts. It should also be pointed out that in addition to pit-houses, rectangular structures and longhouses, there are other types of Maritime Archaic features on the landscape that might be dwelling-related, such as isolated rock pavements (Fitzhugh 1978: Figure 14 illustrates one from Big Island, Sagiek Bay; the author has observed similar fea- tures on Uigortiek Island in the Nain region). Reducing ambiguity and error in our inferences will require a systematic study of Maritime Archaic features that doc- uments their range of variation and provides firmer cri- teria for distinguishing dwelling types with and with- out traces of architecture. 30 CHAPTER 3 / // Maritime Archaic ionglnouses from Nullial< Cove-1 . (a) Structure 15, (b) Structure 1 1 (after Fitziiugin 1985a 97; repro- duced by permission, Arctic Anthropology, University of Wisconsin Press). If we restrict ourselves to the established dwelling features, the question is: to what extent are these dif- ferent types or sizes of structures part of an evolution- ary development over time versus functional or sea- sonal variants? In the pre-6000 B.P. period, the coexis- tence of pit-houses and rectangular surface structures suggests they may be seasonal variants, with pit-hous- es used in colder periods (Fitzhugh 1 985a:89). If that was the case, however, the outer coast locations of the pit-houses do not correspond well with the postulated interior-maritime settlement pattern, in which fall-win- ter settlement is posited for the inner bays or near interior (Fitzhugh 1978:83-84). However, pit-houses might have been used on the coast in the colder spring or fall periods while exploiting migrating harp seal herds, and rectangular structures during the summer. Alternatively, the lack of pit-houses after 6000 B.P. might be explained by a shift of winter settlement from the outer coast to the near-interior (Fitzhugh 1985a:88). Questions could also be raised concerning the contrast between the virtually semi-subterranean longhouse feature at Aillik and the surface longhouses at Nulliak Cove. Is this a seasonal difference, or is the Aillik feature merely a situational variant used on rocky cobble beaches? We have very little information concerning how Maritime Archaic dwellings were constructed. Few details have been published, aside from hints of post- holes at the Aillik longhouse (Fitzhugh 1984:33). The assumption is that the rectangular surface structures were tent dwellings and that the longhouses represent a series of individual tent structures placed adjacent to each other. The pit-houses and longhouses excavated into cobble beaches may also have been roofed in this manner; there is no evidence that they were covered with sod. Estimates of the duration of dwelling occupations is problematic in the absence of preserved bone mate- rial. Fitzhugh (1985a:98) suggests the Nulliak Cove longhouses were used for short periods of days or weeks on the basis of low rates of debris accumulation inside the structures and the absence of middens. This short occupation is related to Nulliak Cove's role as a repeatedly-used staging area for trips to the Ramah Bay chert sources, part of an "expeditionary" model of chert procurement involving pronounced long-distance seasonal mobility (Fitzhugh 1985b:50). On the other hand, longhouses used as seasonal base camps within a local settlement system may have been occupied longer, resulting in greater accumulation as seen in the "sheet midden" deposits at Rattlers Bight (Fitzhugh THEORY AND METHOD IN THE ANALYSIS OF SOCIAL SPACE 31 1981:18, 1985b;49). Consequently, we might expect seasonal, functional, and geographical differences in the nature of longhouses and their duration of occupa- tion. In the absence of bone material these differences might be reflected in the types of features and lithic assemblages present, the ratio of primary versus sec- ondary refuse (Schiffer 1987:58-72) and the degree to which peripheral areas of the sites exhibit build-ups of external activity areas (Binford 1983:190). Of course, these variables are complicated by other factors such as group size and composition and culturally codified disposal practices. The social significance of longhouses has been discussed from somewhat different directions. Although reliance on subarctic Innu ethnography as an analogy source led to difficulties in recognizing the longhouse pattern in the first place, Fitzhugh (1981, 1984, 1985a) turns to this analogy again when inter- preting social space in longhouses, but it should be emphasized that he has not presented a full and explicit account of his position. He characterizes Maritime Archaic society in line with ethnographic interpretations of subarctic Algonkian society as egali- tarian, with emphasis on individual families and the personal autonomy of individuals (Fitzhugh 1 985a:l 04-1 05). Nonetheless, when seen in terms of the temporal development from small pit-houses to longhouses, there appears to have been an increase in co-residential group size over time, from single family units (or small groups of two or three families, if clus- ters of two or three pit-houses indicate contemporary use), towards large multi-family co-residential long- houses consisting of 50-100 individuals (Fitzhugh 1 984:20). Additionally, variation in mortuary treatment may indicate some degree of status distinction in Maritime Archaic society (Fitzhugh 1 978:85, 1 981 :32). A consequence of this social model is that when analyzing architecturally distinct Maritime Archaic dwelling features, Fitzhugh emphasizes the signifi- cance of the individual segments or compartments of the structures, which he considers to be individual family living floors. In the case of the more ambigu- ous linear distributions of hearths and iithics, he views hearth-centered distributions as indicative of individual family living floors (Fitzhugh 1984:18, 1985a:98). The individual family floor areas can be thought of as structural "modules" (Binford 1 983:1 47). A pit- house then consists of a single such module while a longhouse is a linear accretion of multiple modules. In earlier published papers, Fitzhugh (1984:8-9, 1985a:96) argued that the regu- lar spacing of hearths and debris distributions relat- ed to longhouse segments is strongly repetitive, indi- cating the dwelling compartments were used simulta- neously. More recently, however, Fitzhugh (2006: 12-13) suggests that perhaps not all longhouse modules were contemporary. The central segments of longhouses tend to have greater accumulations of debris than the peripheral segments, suggestive of longer occupation. Dwellings may have been organ- ized around a stable central unit, with additional seg- ments being added or subtracted on an "accordion" basis in relation to the arrival and departure of fami- lies. In this reading, the overall configuration of a longhouse is largely determined by repetitive "microscale" patterns at the module level. Modular organization plays a conceptual and methodological role in how spatial patterning in dwellings is linked to possible social distinctions in Maritime Archaic societies. Fitzhugh (1985a:98) views Maritime Archaic social organization as "gener- ally egalitarian" because the contents of longhouse segments are repetitive in terms of household activi- ties and material culture. Nonetheless, there are indi- cations that some artifact production and discard, particularly that involving Ramah chert, slate and soapstone, may be concentrated in particular parts of a longhouse. The possibility that these segments were communal work areas is dismissed, since the content and configuration of these areas is similar to the other segments. In any event, the implication is that relative social equality/distinction might be 32 CHAPTER 3 investigated through the degree of module redundan- cy or variability (Fitzhugh 1 984: 1 1 ). Modular organization is one analytical framework we can use in analyzing Maritime Archaic dwellings but, as indicated, it is also linked to ethnographically derived premises about individual family-focused social organization as well as the assumption that the overall configuration of a longhouse results from repeated replication of microscale processes at the modular level. As an alternative, the continuous linear pattern may be generated by macroscale processes which unite several modular units into a spatial logic partly determined by the meaning of the dwelling as a whole. This unifying spatial logic may be social and ideological rather than simply a functional conse- quence of dwelling geometry— the cloning of individ- ual family modules. Structural linearity may be an organizational principle in which the practical ordering of domestic space in the course of daily life is a vehi- cle for actively structuring social relations, as well as being a symbolic representation of those relations. The unifying linear principle could indicate an emphasis on connected space, a merging of private and public space that expresses a corporate social logic rather than a logic of familial segmentation. Elsewhere, the author suggested that; The spatial logic of the longhouse, consisting of an extremely contrived linear ordering of individual social units, conveyed a tension within Maritime Archaic society between the autonomy of individual household units and the collective organization of emergent cor- porate groups. Maritime Archaic seasonal mobility might have entailed tendencies towards group fission, flexible autonomous social units and temporary settlements. Household autonomy was spatially encoded in the segmentation of longhouses into multi- ple compartments representing individual household floors. These individual social modules were incorporated into a larger col- lective unit by the overall structure of the dwelling. The longhouse structure physically created a corporate unit and gave that unit a temporary (seasonal) material existence through the organization of domestic space. Furthermore, the longhouse helped establish an ideological fiction of collectivity in a social world which otherwise tended towards sea- sonal fragmentation. I suggest that this pat- tern signifies the emergence of situational rather than permanent corporate principles and expresses flux in Maritime Archaic com- plexity processes (Hood 1995:95). The methodological consequence of such a social model is that spatial analysis must be multiscalar. At the microscale level the focus is on patterning within individual modules. At the mesoscale level the patterns that cross-cut or unite modules are considered. At the macroscale, patterns at the level of the dwelling as a whole are investigated. One of the main practical problems in developing accurate descriptions of Maritime Archaic dwelling structures is delimiting them in size. This is particular- ly the case for ambiguous linear distributions without clear architectural demarcation. Longhouse interpreta- tion may be complicated by overlapping between non- contemporary structures and the resulting palimpsest accumulations. Without systematic radiocarbon dating of hearths, combined with refitting studies, we have no firm grounds for assuming that longhouse segments constitute a contemporaneous unity. This methodolog- ical problem has broader interpretative significance in that lacking good control over segment contemporane- ity we have no grounds for accurate dwelling size esti- mates, thus no grounds for inferring the demographic and social composition of a settlement or for postulat- ing symbolic order and meaning. The foregoing discussion focused on identifiable dwelling structures, but it is obviously not the case THEORY AND METHOD IN THE ANALYSIS OF SOCIAL SPACE 33 that all patterns at Maritime Archaic sites are the con- sequence of activities conducted within the confines of dwellings. To avoid unwarranted accommodation argu- ments we need alternative models for identifying dis- tribution patterns unrelated to dwelling structures. Binford's (1 983:1 56-1 59;165-1 70) well-known distinc- tion between patterns connected with indoor and out- door hearths and his discussion of "extensive" activity areas is one point of departure. Expanding the View: Other Archaeological Contexts Longhouses are an unusual dwelling form among ethnographically known hunter-gatherers. North American subarctic ethnography contains examples of large communal structures (see below), but none of these had linear dimensions approaching those of Late Maritime Archaic dwellings and there is little documen- tation of their spatial patterning. Otherwise, longhous- es are better represented in the ethnography of seden- tary agricultural societies such as the Iroquoians. Since the latter analogy is problematic for northern hunter- gatherers, it may be useful to take a comparative archaeological view and consider the spatial patterning associated with longhouse-like features in the eastern subarctic, arctic and elsewhere. What spatial patterns result when people are enclosed in linear structures? This will hardly be comprehensive, but will attempt to develop a sense of the variability in spatial patterning. Beginning close to Labrador, an interesting exam- ple is Seguin's (1995) work on sites dating 700-1000 B.P. in the James Bay region of northern Quebec. Seguin applied Binford's hearth-centered model and Cree ethnography to the interpretation of hearth-cen- tered distributions lacking evidence of structural remains. Archaeologically, she distinguished between external hearths, small structures with single central hearths and longhouses with two or multiple hearths. Her brief consideration of the ethnographic material suggested there are regional variations among Cree groups in the positioning of activities and persons vis a vis hearths. In the Mistassini area, men's and women's activities are conducted on the same side of the hearth, while in the Chisasibi area they are con- ducted on opposite sides. The longhouse structure posited for site GdFc-2 (Figure 12a; Seguin 1995:44) consisted of four hearths placed in a ca. 9 m long line, spaced at ca. 1 m intervals; lithic concentrations were found on opposite sides of two hearths and on one side of the other two. Seguin interpreted this distribu- tion as a ca. 1 3 m long, 5.5 m wide longhouse, with the lithic concentrations representing activities con- ducted between the hearths and the dwelling walls. In contrast, Loring (1 992:245, 250-260) described a probable longhouse (shaputuan) dated 1800-1500 B.P. at Daniel Rattle-1, Area IV, near Davis Inlet on the Labrador coast. There was no trace of walls; the fea- ture consisted of a 4.5 m long, 1 .0 m wide, raised lin- ear hearth with two smaller cobble hearths slightly dis- tant at both ends. The central hearth was tightly packed with calcined bone fragments, bone ash, ther- mally-altered Ramah chert flakes and heat-shattered biface fragments, and fire-cracked rocks. Lithic tools and flakes occurred in a major concentration extend- ing 3.5 m outwards on one of the long sides of the hearth feature, with a two diffuse clusters on the oppo- site side and minor concentrations associated with the small end-hearths. Returning to Quebec, at site CbFd-9 (Figure 12b) Seguin (1995:43) identified two features (B and E) as external hearths, although the excavated area was too small to identify possible drop/toss zones. Six other hearths (A, C, D, F-H) were placed in a 30 m long line, spaced at intervals of 4.4 to 6.6 m. Lithic remains were concentrated within 1 .0 to 1.5 m of the hearth centers, in five cases on opposite sides of the hearths, in one case as two clusters on the same side. Each hearth is interpreted as the focus of a separate tent structure, but little justification is given for this conclu- sion. Hearths C, D and F are closely positioned and are aligned on the same axis, hearths C and H lie slightly apart from the first group and are aligned on a slight- 34 CHAPTER 3 1 2/ Pre-contact Recent Indian dwelling features, (a) longhouse, CdFc-2, Riviere Laforge, Quebec (after Seguin 1995:44). (b) individual tent structures, CbFd-D, Riviere Laforge, Quebec (after Seguin 1995:43). (c) tent dwelling, Kamarsuk Area II, Labrador (after Loring 1992:266, 270). Figures a and b reproduced by permission, I'Association des archeologues du Quebec. ly different axis, and hearth A is out of alignment with the C-D-F group. It is unclear why the C-D-F and C-H groups could not be considered longhouses; the only difference appears to be that their lithic concentrations are oriented towards what would be the middle of a longhouse rather than lying between the hearths and the walls. This would seem to be a weak criterion given the ethnographic observations on variability in the placement of activities around hearths. Indeed, with Fitzhugh's longhouse model in mind, one might even consider the possibility of a larger structure, given a refit between hearth areas A and F. Nonetheless, Seguin's material is very relevant to the Maritime Archaic longhouse discussion, since it provides some concrete archaeological examples of alternatives in spatial patterning and it attempts to apply an ethnoar- chaeologically-derived behavioral model. Seguin's individual structure interpretation can be compared with Loring's (1992:266-276) discussion of a Late Prehistoric dwelling from Kamarsuk Area II, north of Davis Inlet, Labrador (Figure 1 2c). The feature was a slightly semi-subterranean oval tent structure, 4 by 5 m in size, with a raised central cobble hearth. The hearth contained calcined bone fragments and thermally altered Ramah chert. There was a pro- nounced concentration of debitage on one side of the hearth, but relatively little on the opposite side, sug- gesting the latter was used as a sleeping area. A sec- ond concentration of Ramah chert debitage at one end of the structure and extending outside suggested a doorway midden. This spatial pattern is broadly simi- lar to that identified by Seguin and bears a strong resemblance to Binford's "archetype" model for hearth- centered patterns expected in enclosed and main- tained spaces (i.e., Palanga's House, Figure 6b). A distinctly different pattern was found at the late Point Revenge Complex Winter Cove-4 site (ca. 500 B.P.) in Hamilton Inlet, Labrador (Fitzhugh 1 978b). This feature was a 4 by 8 m oval tent ring with a single cen- tral internal hearth and nine external hearths. Some of the external hearths were almost superimposed by the tent ring such that, in contrast to Fitzhugh's interpre- tation, they may represent earlier activity. Very little lithic material was found inside the tent ring; only a small concentration adjacent to the well-constructed hearth and along the southeastern wall of the feature. This seems consistent with Binford's inside-hearth model. Most of the lithics were concentrated around the external hearths, which were less substantial cob- blestone clusters. The tools and flakes mostly lay adja- cent to the hearths (akin to drop zones), but in one case there was a wide scatter of debitage extending out at least 3 m from a hearth, evidently a "non-main- tained" zone. Fitzhugh (1 978b:l 58-1 59) suggested the THEORY AND METHOD IN THE ANALYSIS OF SOCIAL SPACE 35 structure represents a warm weather occupation by a multi-family group. Returning to the spatial structure of longhouses, another archaeological case of such features can be found in Late Dorset Paleoeskimo society (ca. 500- 1 000/1 200 AD). These features range from 1 to 45 m in length, 5-7 m wide, and sometimes incorporate stone walls up to 1 meter in height. The central corri- dors may lack distinctive hearths, although there may be traces of a central "passage" with rock construc- tions and burned fat. The walls sometimes contain small niches with traces of burned fat and possible domestic activities. In some cases, lines of external hearths and caches run parallel to the structures (Appelt and Cullov 1998:148-149, 1999;24-35; Damkjar 2000; McChee 1971; Plumet 1985; Schledermann 1 990:202-251 ). Investigators agree that these structures represent the communal ritual activi- ties of multi-family groups, probably conducted during the summer, but they disagree on whether the struc- tures were roofed tent dwellings or whether the walls were merely symbolic. Plumet (1985:228-229, 366- 367, 371) argues that the spatial logic was of individ- ual family floor spaces along the walls, with a linear central communal area that was a symbolic spatial transformation of the axial structure principle used in smaller Dorset sod-house dwellings. Perhaps the clearest example of longhouse spatial patterning is found in a 15.5 m long feature at the David site in northwest Greenland (Appelt and Cullov 1999:36-40; Figure 13a), where a series of separate lithic concentrations were closely linked to a central passage feature. Some of the concentrations lay adja- cent to stone-slab pits of unknown function. Four of these clusters were regularly spaced along the east side of the passage (ca. 7 m between center points); the clusters were all somewhat oval and of broadly similar dimensions (4-6 m long, 3-4 m wide). On the west side of the passage were two linear distributions of similar dimensions (5-6 m long, 2 m wide) and a small flake concentration at one end of the passage. There were also four small clusters of tools lying out- side the main concentrations. Thus, there were slight differences in depositional patterns on either side of the passage such that the configuration was somewhat asymmetric. The eastern wall contained traces of at least six regularly spaced (1.5 m) slabs with burned fat suggestive of hearth areas. It may be significant that the largest debris concentrations occurred on this side of the structure. Schledermann's (1 990:209-21 1 , 239, 245) spatial distribution data from Ellesmere Island exhibit gener- ally similar tendencies. At the Longhouse site, partial excavation of a 45 m long feature suggested a pattern of paired lithic/bone clusters on opposite sides of the central area. The clusters were of similar size (mostly 2.0 by 1 .0-1 .5 m) and fairly evenly spaced along each side (mostly 2.0-2.5 m between their center points). Although there were sporadic traces of charcoal and burned bone there were no distinct hearths. The pat- tern in the 14.5 m long structure at the Cove site sug- gests seven clusters of variable dimensions on each side of the passage. The clusters along the southern half of the structure seem semi-regularly spaced (1 .5- 2.7 m between center points) and have higher find densities than those on the northern side of the struc- ture. Although those on the northern side are also reg- ularly spaced (1.5 m between center points), they are arranged in two groups separated by 4.5 m of low den- sity: four at the western end of the structure, two at the eastern end. Again, there is a hint of slight asym- metry in cluster placement and deposition, as at the David site feature. A slightly different pattern was found in Late Dorset House B at the UNC-1 1 site in the Ungava region (Plumet 1985:105-134, 209-231, hors texte 9 and 1 0). The feature was ca. 36 m long and 5 m wide (Figure 13b). The central axis of the structure had a slightly raised ridge containing stone-lined boxes and pits with traces of burned fat and red ocher. In the wall area there were several niches consisting of flat slabs and burned fat that were spaced at 2-3 m intervals. A 36 CHAPTER 3 schematic summary indicates the lithic distribution was primarily concentrated in the southern end of the structure. Debris occurred along the walls as near-con- tinuous distributions, with hints of separate but over- lapping clusters spaced about two meters apart, evi- dently associated with the wall niches. Seven smaller clusters (mostly 1.5-2.0 m in diameter) occurred clos- er to the center, adjacent to the box and pit features. The spatial patterning in these structures is rea- sonably consistent, with small, relatively discrete, sometimes regularly spaced, debris clusters along both sides of the central axis of the features. These clusters are associated with rock constructions and traces of burning. Possible hearth niches in the walls at the David site and UNC-1 1 seem to be associated with larger and denser debris accumulations. Hence, although we lack evidence for "classic " hearth-centered distributions there are strong indications that the dis- tributions are feature-related, probably to cooking/heating/ illumination areas with oil lamps. A slight exception to the pattern of relatively discrete debris clusters is UNC-1 1, where there was greater accumulation of material adjacent to the walls and overlap between niche-related distributions. It should be kept in mind, however, that although these feature- centered distributions seem similar to those associat- ed with domestic hearths, the spatial organization of Late Dorset longhouses was likely structured by spe- cial ritual practices. A final archaeological example comes from farther afield: the well-known, although not well understood, "longhouse" structures from the Russian Upper Paleolithic site Kostenki (Crigor'ev 1967; Klein 1969:114-117, 170-177). The three structures described by Klein had similar features: lines of 9-10 hearths, mostly spaced ca. 1.5-2.0 m apart. Distributions of the associated artifactual materials suggested one dwelling 35 m in length by 1 0-1 5 m in width, a second 35 m in length by 5 m in width, and a third 23 m in length by 5 m in width. Cache pits with mammoth bone and smaller circular structures that might have been house-pits were also present. These features, and the complex sites of which they are a part, present similar analytical problems, and then some, to the Maritime Archaic longhouses that lack clear architectural features. Crigor'ev (1993:58-59) notes that delineation of the dwelling boundaries is arbitrary line-drawing through density distributions, but the lithic distributions from Kostenki I, Horizon 1 (Crigor'ev 1 967:345; Klem 1 969:1 1 7) at least provide a sense of the spatial structure at the site (Figure 1 3c). Taking into account that some excavation units were destroyed, the most dense flake concentrations were located primarily within a 1 -3 m distance of the central hearths, although towards the southern end of the pro- posed dwelling the debitage extended up to 4 meters distant. The concentrations were all on the eastern side of the dwelling, with the exception of two small clusters on western side of the hearth line. The south- ern portion of the dwelling had a near continuous dis- tribution of flakes paralleling four of the hearths and there may have been a separate concentration in the middle near two hearths, as well as a separate concen- tration paralleling the three northernmost hearths. If, for the sake of argument, we accept Crigor'ev's dwelling boundary, it would appear that the longhouse dwelling was surrounded by storage pits and there might have been some external lithic reduction or dumping of waste from inside the dwelling. It is interesting to compare the spatial patterns of the archaeological longhouses described above. There are obvious parallels between Seguin's northern Quebec longhouse, the Late Dorset structures and the Kostenki situation— clusters of debris on either side of central hearths (or central "features" in the case of Late Dorset)— although the Kostenki pattern is one of less discrete clusters with much more overlap between what may or may not have been individual hearth-cen- tered distributions. The more continuous distributions at Kostenki are similar to the wall-related concentra- tions in the Ungava Late Dorset longhouse. Each of these cases varies somewhat in the position of relative- THEORY AND METHOD IN THE ANALYSIS OF SOCIAL SPACE 37 I'- David Site Thule District ^ stone wall hearth j I lithic concentration ^2 tool outlier — 2m scale (a) UNG-11 Ungava Bay J house wall hearth Q rock feature 1 lithic concentration — 2m scale (b) 0' o (!)(£] (c) Kostenki I Russia inferred wall ^ hearth [~~] lithic concentration 2m scale /5/ /.afe Dorset and Russian Upper Paleolithic longhouses. (a) Late Dorset longhouse, David site, Thule district, Greenland (after Appelt et al. 1 999 36-40). (b) Late Dorset longhouse, UNC I 1 , Ungava, Quebec (after Plumet 1985: hors texte 9, 10); reproduced by permission of author), (c) Upper Paleolithic longhouse, Kostenki I, Russia (after Crigor'iev 1967:345: repro- duced by permission. Current Anthropology, University of Chicago Press). ly empty or maintained space: in the Quebec Late Pre- Contact case it is near the walls, in the Late Dorset longhouses it may occur in the center, near the walls, or at one end of the structures, while at Kostenki it consists of most of one entire side of the structure. In sum, these archaeological examples exhibit similar micro-scale patterns involving feature-centered distri- butions that probably result from commonalities in body mechanics. But they also suggest that feature- centered distributions were organized into macro-level configurations through different kinds of social prac- tices or cultural logics. In the case of Maritime Archaic longhouses we might anticipate that micro-level fea- ture-centered distributions will be similar to those in the behavioral models, but that their integration into the dwelling as a whole will be culturally idiosyncratic. We should now briefly consider some ethnographic examples that illustrate the mutual structuring of spa- tial practice and cultural logic. Ethnography: Social Space and Meaning The preceding discussion of archaeological spatial patterning identified some micro-level behavioral con- sistencies, but the interpretive scope of these observa- tions is limited: "We are left with a bunch of pieces, but cannot explain why they are assembled the way they are" (Whitelaw 1994:219). The integration of micro-patterns into macro-configurations is structured by differing social and cultural logics that were hinted at but not illuminated in the preceding discussion. The behavioral dimensions of dwelling space will now be related to social and ideological aspects of spatial organization using ethnographic analogies, beginning with social organization linkages then shifting towards the symbolic. The strongly behavioral ethnoarchaeological stud- ies of spatial patterns have been supplemented by greater attention to the role of social organization in community patterning. To a great extent this has focused on the degree to which dwelling proximity in settlement layouts correlates with social relationships such as kinship, sharing and labor cooperation (e.g., Binford 1991; Gargett and Hayden 1991; Whitelaw 1991; Yellen 1977). Whitelaw's (1991) cross-cultural study of hunter-gatherer community space concluded that there are clear linkages between spatial and social proximity, and that increases in social scale result in more organization or patterning of community struc- ture. In situations involving cooperation in labour or resource sharing, close spacing of habitations facili- tates interaction, but also the monitoring of behavior and possessions (Whitelaw 1991:168), which plays into the regulation of cooperation and sharing. Whitelaw (1 991 : 1 65) notes that linear spatial arrange- 38 CHAPTER 3 merits are primarily found in arctic and subarctic regions; strangely, he relates this to dispersal of social units along shorelines and claims that "...because lin- ear arrangements allow only one or two neighbors, they do not particularly emphasize the coherence of a group." Such a conclusion is thin, to say the least, and it seems absurd to suggest that linear arrangements such as Maritime Archaic longhouses contribute little to group coherence. Although cross-cultural general- izations can point to general patterns to be investigat- ed, they are not very context sensitive and in a case such as hunter-gatherer longhouses, which are very rare in the ethnographic record, they completely fail us. It is necessary, therefore, to explore specific con- textual studies in order to arrive at a more dynamic view of spatial patterning. Archaeological treatment of the meaningful dimensions of space has built on the seminal work of Bourdieu (1979) and Levi-Strauss (1963) on dwelling organization, combined with Ciddens' (1984) struc- turation theory, the work of diverse human geogra- phers and influences from phenomenology. At the risk of oversimplifying, the common thread is that space is organized in accordance with symbolic principles, but spatial meanings are invoked, reproduced or changed through human practice (Parker Pearson and Richards 1994:5). Meaningful space is constructed through social action and, in turn, that construction acts back upon social action: meaning-laden space is both a product and precedent of agency. This holds for the small scale of dwelling sites as well as for large scale landscapes (Tilley 1994). The following is a nice sum- mary statement of this view, as applied to the organi- zation of housing space: The house and the body are intimately linked. The house is an extension of the person; like an extra skin, carapace or second layer of clothes, it serves as much to reveal and dis- play as it does to hide and protect. House, body and mind are in continuous interaction, the physical structure, furnishing, social con- ventions and mental images of the house at once enabling, moulding, informing and con- straining the activities and ideas which unfold within its bounds. A ready-made environment is fashioned by a previous generation and lived in long before it becomes an object of thought, the house is a prime agent of social- ization (Carsten and Hugh-Jones 1995:2). A focus on the interplay of mind, architecture and action leads to consideration of the metaphorical role played by housing as a material representation of social structure and ideology, either as a mirror or transforma- tion. Seen from the structuralist tradition, components of houses can represent social divisions and hierar- chies, while the dwelling as a whole can give the appear- ance of unity to opposing principles (Carsten and Hugh- Jones 1995:8, 12) or encode cosmological metaphors. As emphasized above, however, these are not static abstract structures but codes and materials that are produced and reproduced through daily practice. An example of this perspective is Yates' (1989) structuralist analysis of social space in Sami tents. Gender was structured by divisions into male (back) and female (front) space, while the generations (par- ents and young children, versus older children) were distinguished laterally. In both cases the central hearth served as the fulcrum of the spatial differentiation. These socio-spatial divisions were paralleled by con- ceptual divisions based on Sami cosmology, such that a series of binary oppositions were played out within the horizontal dwelling space: male/female, back/front, north/south, winter/summer, sacred/pro- fane, clean/unclean, death/life, hunting/milking, hunt- ing blood/menstrual blood. The hearth was the crucial center that mediated these oppositions. Furthermore, the vertical space of the tent was divided into levels associated with divisions of the heavens, with different spiritual personages associated with each level. Consequently, the tent was a locus in which gender THEORY AND IV1ETH0D IN THE ANALYSIS OF SOCIAL SPACE 39 and other structures were produced and reproduced in daily practices. Siberian Nenets tent interiors and camp lay-outs are also structured by gender divisions and rules for movement (Golovnev and Osherenko 1999:32-39; Haakanson 2000:82-89). Tent doors normally face east. Within the tent the central hearth is the anchor point for an invisible symbolic siyangi line that extends westwards across the floor and continues outside the tent in the opposite direction from the entrance. Movement of all people within the tent is restricted to the front and sides of the hearth because it is not per- mitted to cross the siyangi line behind the hearth. Outside the tent, adult women are not permitted to cross the siyangi line when they are within sight of the camp. This means that women cannot move behind the tents and must instead depart and return from the front entrances by following the same path. Women's work areas are therefore in front of the tents. Although the movement of adult men is restricted within the tent it is not restricted outside; the men's work area is behind the tent where the siyangi line projects west- wards. These symbolic orientation and movement reg- ulation principles have consequences for the lay-out of multi-tent camps (Figure 14). The tents must be set up in a line along a north-south axis in order to maintain the principle that women cannot cross the multiple siyangi lines that extend from the western (back) sides of the dwellings (Haakanson 2000:219). The spacing between the tents is related to the nature and close- ness of personal relationships (Haakanson 2000:221). It is interesting to compare this rule-structured lineari- ty of individual dwellings with the linear organization of Maritime Archaic longhouses. Tanner (1979:73-107) discusses the symbolic dimensions of space among the Mistassini Cree. For much of the year the Cree use single-family tent dwellings. When they are used at winter aggregation sites, Tanner (1979:83) notes they are arranged in a line. He suggests that a location on one end of the line will be chosen by the hunting group leader. Single-fam- ily dwellings display a consistent pattern of internal organization. A central hearth is placed back from the doorway and a living area is positioned at the back of the dwelling behind the hearth and opposite the door. Each individual has their own place at the rear of the dwelling, while the front of the dwelling is communal space. The tent space is also divided by gender, with one side associated with women, the other with men. Within the gender divisions there are further spatial positionings in relation to marital status and age. Household items and foods are also ordered in relation to the gender division. The meat of large animals (moose, caribou, bear) is stored on the male side, while that of smaller animals (beaver, etc.) is stored on the women's side and each gender attends to the prop- er ritual treatment and disposal of the bones. This basic pattern recurs when the Cree move to winter communal dwellings, which are shared by sev- eral families. Divisions of male and female space are maintained, with the proviso that adjacent families cannot place members of the same gender adjacent to each other. The hunting party leader selects a position near the doorway. Tanner (1979:86-87) notes that "...the move to a communal dwelling brings about a minimal alteration in the use of domestic space by the individual commensal group, and the relative autono- my of each group is preserved. " This is consistent with Fitzhugh's (1985a) interpretations of Maritime Archaic longhouses. It should be noted, however, that unlike other frequently cited ethnographic cases (e.g., Bourdieu 1970; Carsten and Hugh-Jones 1995; Yates 1989), the organization of Cree interior space is not structured by geographic or cosmological symbolism (Tanner 1 979:87). The eastern subarctic ethnographic record con- tains sporadic references to large dwellings that could be called "longhouses." Among the Labrador Innu and the Quebec Cree a large multi-hearth tent (Innu: shaputuan) was associated with the makushan cari- bou feast (Henriksen 1 973:35-39; Tanner 1979:162- 1 69, 1 80). Among the Ojibway, bark longhouses were 40 CHAPTER 3 women's path East corral women's path siyangi line siyangi line siyangi line siyangi line siyangi line firewood • hearth women's work area I I sled men's work area 14/ Spatial organization of a Nenets summer camp (after i-laal (r\ Q\ \j .L. \\J . \ } 1 7 1^1 f,7 A U A\ 1 7 =^1 1 Q Q 71 Cray-banded 3 (0.1) 0.1 (0.0) 0.1 (0.0) Red-brown 0.9 (0.0) 0.5 (0.0) 2.6 (0.1) Indet. 19 (0.4) 2 (0.0) tion in the middle is somewhat higher), but the middle has an extremely low overall flake total, entirely lacks patinated Ramah chert and Mugford chert, and has only a single slate flake. Thus, the low intensity of lith- ic reduction in the middle subarea stands as a marked contrast to the two lobes and is also consistent with its low frequency of retouched implements (n =14, see Table 7), though not with the relatively high propor- tion of quartz cores (see below). The "other" subarea has quartz/Ramah chert proportions similar to the south lobe. If we consider slate color variants by frequency, both lobes are dominated by gray slate, while green slate is the second most frequent variant. Nonetheless, a chi-square test (x^ = 80.92) indicates there are sig- nificant differences in the spatial distributions of the slate variants, the expected frequencies pointing to a slight over-representation of gray slate in the north lobe and slight a over-representation of green slate in the south lobe and "other" subarea. The frequency of green slate in the "other" subarea— most being con- centrated east of the south lobe— could imply a behav- ioral link between the south lobe and the "external" area to the east. If the comparison is conducted by weight, however, there is parity in gray and green slate between the two lobes, but a repeated over-propor- tional amount of green slate in the "other" subarea (Table 6). Thus, the similarities in slate type propor- tions between the south lobe and the "other" subarea parallel the similarities in quartz/Ramah chert propor- tions between the same subareas. Variation in flake raw materials can also be evalu- ated independent of the model units. Although it can- not be assumed that the distributions of tools and deb- itage are conditioned by the same depositional vari- ables, a tool-cluster based quadrat grouping of flakes might reveal additional dimensions of spatial variation. The quadrats were grouped according to the 9-cluster solution, but the 1 m^ units did not coincide perfectly with the clusters. The resulting table (not shown) was not particularly enlightening, since the frequency vari- ations were fairly consistent with the model-based analysis. A better way of handling the debitage distri- bution independent of model assumptions was exploratory correspondence analysis of raw material frequencies by quadrats. The correspondence plot in 90 CHAPTER 6 iTable 7: Area 2A Tool Classes by Model Units. PT Bl PR FP SC UF SP CA SB ST BC CO TM OTH TOTAL N 2.5 6 3 7 6 25 2 2 2 8 1 3 1 3 1 1 91.5 MID 0.5 2 1 4 1 4.5 1 1 4 S 8 6 2 2 1 1 41 2 1 4 35 32 1 1 .5* 1 2 1 58.5 OTH 2 2 1 1 4 27 6 7 12 8 5 2 77 TOTAL 1 3 16 6 10 22 97 4 3 12 51 57 37 7 6 341 % 3.8 4.7 1.8 2.9 6.5 28.4 1 .2 0.9 3.5 1 5.0 16.7 10.9 2.1 1.8 * a fragment conjoining with another in a different unit was assigned a value of .5 PT: points (bifacial), BL bifaces, PR: preforms (biface), FP: flake points, SC: scrapers (endscrapers and flake knives), UF: utilized flakes, SP: slate points, CA: celt/adze, SB: slate blanks (unmodified plates and blanks), ST: slate tool fragments (ground slate flakes and unidentified tool fragments), BC: bipolar cores, CO: block cores (quartz), TM: tool manufacturing (hammerstones, schist tablets, whetstones), and OTH: other (stemmed flakes, linear flakes, graver). Figure 45 indicates that the first axis (horizontal) accounts for 75.2% of the variation, seriating the quadrats relative to their proportions of Ramah chert and quartz. The units on the left side of the axis are dominated by Ramah chert while the units on the right side are dominated by quartz. The second axis (verti- cal) accounts for 18.2% of the variation, scaling the quadrats relative to slate proportions. Units placed towards the top of the axis have unusually high fre- quencies of slate. Symbol coding of quadrats by the model-based units indicates a clear pattern. Most of the north lobe quadrats and all the middle quadrats are pulled right by high quartz frequencies, while most of the south lobe quadrats are pulled left by abundant Ramah chert (although several are pulled by quartz). "Other" quadrats are distributed in both the quartz and Ramah chert groups, although three quadrats are out- liers pulled by high frequencies of slate. Ten of the 1 3 "other" quadrats that group with the south lobe are actually adjacent to that lobe, while 10 of the 15 "other" quadrats that group with the north lobe lie adjacent to that lobe. Mugford chert has very little influence on the plot. Thus, the correspondence analy- sis indicates that the model-based units are to a great extent replicated by model-independent analysis. Tool Distribution The spatial distribution of tool classes may indicate something about functional or depositional variation within Area 2A. The analyses are conducted using 14 largely functional classes, some of which (e.g., bifacial points) could be further subdivided into formal/func- tional variants (see Chapter 8 for a description of the tool material). Table 7 outlines the tool class frequen- cies in relation to the model-based units introduced in the debitage analysis. Variation in the table can be scaled by correspondence analysis and visualized in a plot (Figure 46). The first axis of the plot accounts for 44.9% of the variation and primarily contrasts the south lobe on the right side with the north lobe slight- ly towards the left. The south lobe is positioned with bipolar cores, bifacial points, scrapers, utilized flakes and slate tool fragments. The north lobe is only distin- guished by flake points. Common to both lobes are block cores, bifaces and preforms. The middle subarea is pulled to the extreme left of the plot, probably NUKASUSUTOK-5: SPATIAL ANALYSIS 91 because of a modest frequency of block cores combined with an absence of bipolar cores. Along the second axis, account- ing for 34.9% of the variation, the "other" subarea is distin- guished towards the bottom of the plot by slate blanks and tool manufacturing items. Although the low tool frequencies limit the robustness of the analysis, it does point to contrasts between the subareas. The lobes are not replicate behav- ioral modules, but they do share broad similarities. A more detailed assess- ment of tool class variation can be had by combining the quan- titative data with the spatial dis- tribution diagrams. While some of the tool classes can reason- ably be evaluated in terms of their point distributions, oth- ers— specifically, utilized flakes, block cores and bipolar cores— lack systematic point proveniencing and are best evaluated by quadrats. Both types of information are provid- ed here. There are some inter- esting patterns in the distribu- tions of block and bipolar cores (Figures 47 and 48). The largely unworked quartz block cores are frequent in the middle area between the lobes, suggesting they were cached or dumped there. Half of the cores in this area are located in quadrats with few quartz flakes, while the 45/ Area 2A: correspondence analysis plot of flake raw materials by quadrat, coded by analytical model units. 2.0- 1.5- c\ SP^ 1.0- Bl 0.5- Axis 1 (44.9%) ^'a • . ^ N Pj^^ -2.0 -1.5 -1.0 -0.5 .ST SC ^BC 1.0 1 1 1 CO A ^""^ ^UF A OTH -1.0- OTH • ^ SB 1 ^ - Tool class ' • Model unit -2.0- ^TM -2.5- Axis 2 (34.9%) 46/ Area 2A: correspondence analysis plot of tool classes by analytical model units. 92 CHAPTER 6 47/ Area 2: distribution of Mock cores by quadrat. 48/ Area 2 distribution of bipolar cores by quadrat. Other half are positioned within or adjacent to a clus- ter of quartz debitage at the northeastern corner of the south lobe. Consequently, the placement of block cores is partly independent of quartz reduction loca- tions. There is also a slight concentration of block cores at the northern end of the north lobe. Here, how- ever, the cores are clearly associated with quartz flake concentrations. In contrast, bipolar cores, the interme- diary and final product of quartz reduction, are mostly found within the south lobe, where they are associated with the greatest densities of quartz flakes but are largely independent of the distribution of block cores. At the north lobe there are small clusters of bipolar cores at each end of the eastern side; the cluster near the northern end is associated with a concentration of quartz flakes, but the southern cluster is not. Both NUKASUSUTOK 5: SPATIAL ANALYSIS 93 49/ Area 2 distribution of utilized flal) J (. 1 .a; DidCK cneri iL \iL.iLj 11 if, Q\ 18 K ^1 jIdLc lOldl \ A ^\ 1 ^7 1 ^ 1 J V 1 O • Gr66ri Oo 1 1 c aD Gra,y 11 ? ?L. CJ 78 D Q /H r^c/^\A/nKcu Ui uwri 11 7 e.D Vjidy DdnUcU 11 :3 inuei. 11 11 7 Schist 1 (1.1) 4 (5.2) 5 (1.5) Other 1 (0.6) 1 (1.3) 2 (0.6) *.5 denotes a conjoining fragment Table 9: Area 2A Tool Raw Material Weights by Model Units. In Grams. North Lobe Middle Lobe South Lobe Other N (%) N (%) N (%) N (%) Quartz 1751 (68.8) 1161 (94.7) 1759 (76.4) 1475.6 (58.2) Ramah chert 284.2 (11.2) 64.4 (5.3) 302.3 (1 3.1) 165 (6.5) Black chert 12 (0.5) 32 (1.4) 1 5 (0.6) Gray Mugford 25.3 (1.0) 16 (0.7) 4 (0.2) Slate—total 294.7 (11.6) 0.3 (0.0) 160.6 (7.0) 1 55.1 (6.1) Green 111.5 0.3 98.7 144.1 Gray 16 19.8 9.9 Red-brown 167 3 1.1 Gray-banded 0.2 6.1 Gray/schist 33 Schist 177 (7.0) 33 (1.4) 554.5 (21.9) Other 0.3 (0.0) 167 (6.6) Distribution of Red Ocher and Bone Fragments Figure 61 depicts the distribution of red ocher stains and bone fragments. It is possible that more red ocher stains were present but went undocu- mented given the difficulty of identifying them in the very dark brown podzolized sand or because some were erased by deflation. Taking the data at hand, red ocher stains are found almost exclusively within the lobes, where they occur within or closely adjacent to hearths. Stains are associated with both NUKASUSUTOK-5: SPATIAL ANALYSIS 101 Table 10: Area 2A Tool Classes; Quadrats Grouped by K-Means Cluster. PT Bl PR FP sc UF SP CA SB ST BC CO TM OTH TOTAL Cluster 2 7 6 2 2 1 1 32 2 1 4 1 9.5* 25 3 4 2 120.5 Cluster 4 0.5 1 1 1 2 8 2 1 2 3 3 24.5 Cluster 7 3 2 1 2 2 22 4 7 1 3 1 1 2 1 70 Cluster 8 1 4 2 5 4 20 1 6 6 8 1 1 59 Cluster 9 1.5 3 3 1 5 2 2 16.5 10 1 2 2 67 TOTAL 13 16 6 10 22 97 4 3 12 51 57 37 7 6 341 * .5 denotes a conjoining fragment PT: points (bifacial), Bl: bifaces, PR: preforms (biface), FP: flake points, SC: scrapers (endscrapers and flake knives), UF: utilized flakes, SP: slate points, CA: celt/adze, SB: slate blanks (unmodified plates and blanks), ST: slate tool fragments (ground slate flakes and unidentified tool fragments), BC: bipolar cores, CO: block cores (quartz), TM: tool manufacturing (hammerstones, schist tablets, whetstones), and OTH: other (stemmed flakes, linear flakes, graver). Table 1 1: Area 2A Tool Raw Material Frequencies by K-Means Cluster. Cluster 2 Cluster 4 Cluster 7 Cluster 8 Cluster 4 N (%) N (%) N (%) N (%) N (%) Quartz 22 (18.3) 4 (16.3) 21 (30.0) 13 (22.0) 20 (29.9) Ramah chert 55 (45.6) 13.5 (55.1) 35 (50.0) 33 (55.9) 21.5 (32.1) Black chert 10 (8.3) 1 (4.1) 4 (6.8) 3 (4.5) Cray Mugford 3 (2.5) 1 (4.1) 1 (1.4) 1 (1.7) 2 (3.0) Slate— total 26.5 (22.0) 5 (20.4) 11 (15.7) 7 (11.9) 19.5 (29.1) Green 1 1 3 4 3 9 Cray 12.5 5 1 8.5 Red-brown 1 2 2 1 Gray-banded 2 1 1 Gray/schist Indet. 1 1 Schist 3 (2.5) 2 (2.9) 1 (1.7) Other 1 (0.8) 1 (1.5) TOTAL 120.5 24.5 70 59 67 " .5 denotes a conjoining fragment surface and pit hearths, although they seem most lobe and two small stains just south of the south intensive with pit features. Deviating from this pat- lobe. No uncrushed ocher fragments were observed tern is a patch 1 m north of hearth 8 at the north at Area 2A. 102 CHAPTER 6 The paucity of bone is a consequence of both taphonomic factors and recovery techniques. The extremely acidic soil (pH 4.0-5.0) destroyed all but a few fragments of highly calcined bone. On the other hand, it is possible that more fragments could have been recovered with fine-screening (only 1/4 inch mesh was used). All bone fragments at Area 2A are associated with the two lobes. At the north lobe, a sin- gle fragment is found in the pit of hearth 8, another single fragment lies 1 m to the southeast and another single piece lies between hearths 1 and 6. At the south lobe, six fragments are associated with hearth 5. None of these bone fragments could be identified to taxon. Feature-Centered Distributions, Size-Sorting, and Cultural Formation Processes Since the behavioral modules suggested in the Area 2A material might imply dwelling structures we should consider more explicitly the ethnoarchaeologically derived feature-centered models. As discussed in Chapter 3, we might expect to find different types of distribution patterns associated with exterior and inte- NUKASUSUTOK-5: SPATIAL ANALYSIS 103 53K84y 50x84y 47x83y CI 53x76y 50x7 1y Ground slate flakes and tool fragments 44>i79y 1 o^ 46x63y 43x75y 1 2 I I I o © • <• 41x67y Ground slate flakes and tool fragments (n) edge-struck flakes n 11 41x61y i 57/ Area 2 distribution of ground slate flakes and ground tool fragments. rior hearths. In general, exterior hearths may have more extensive debris distributions with drop- and toss-zones characterized by size-sorting of cultural material: smaller items in the drop-zones adjacent to hearths, larger items in the toss-zones further away— a roughly concentric zonation. Interior hearths may exhibit an adjacent concentration of cultural material but may lack clear drop- and toss-zones because of periodic maintenance of dwelling interiors designed to prevent the accumulation of material in the sleeping areas near the dwelling walls. Maintenance of dwelling interiors may result in the development of exterior sec- ondary refuse deposits such as "door dumps" (Binford 1983:177). Thus, instead of concentric size-sorting there is a distinction between small-sized items near interior hearths and larger items in exterior dumps. Evaluation of the total debitage distribution as well as distributions of individual raw materials does not indicate the concentric and gradual "distance decay" patterning expected at exterior hearths. Flakes are obviously clustered adjacent to hearths, but they are generally concentrated on one side of 104 CHAPTER 6 53x84y 50x84y 53x76y 50x7 1y 47x83y Miscellaneous tools FSC= flake scraper/knife GSC= graver/scraper GST= grindstone HAM= hammerstone LFK= linear flake SFK= stemmed flake SHF= scfiist fragment SHT= schist tablet WHS= whetstone 45x63y SH£ Miscellaneous tools FSC= flake scraper/knife GSC= graver/scraper GST= grindstone HAiVI= hammerstone LFK= linear flake SFK= stemmed flake SHF= schist fragment SHT= schist tablet WHS= whetstone 41x61y 58/ Area 2: distribution of miscellaneous tools. the hearths rather than forming a ring around them. The overall distributions are bounded in linear "lobe" patterns which, to some extent, may be the result of overlapping hearth-centered distributions. The outer edges of the lobes (particularly the southern one) are fairly sharp, suggestive of a boundary effect. At the same time, the inner edges of the lobes indicate cen- tral areas with minimal debitage accumulation, either a consequence of differential activity placement or maintenance of the central areas. The tool distribu- tion patterns are also closely linked to hearths, although the concentration of block cores in the "middle" and the area east of the lobes deviates from the overall pattern. If the overall configuration of the debitage and tool distributions seems more in line with an interior hearth model, what about the size-sorting criterion? One problem in assessing size-sorting is that lithics may be less sensitive to the discard and maintenance practices noted ethnoarchaeologically for bones and implements of organic materials, given much less vari- ation in size dimensions. Nonetheless, since lithics are NUKASUSUTOK-5: SPATIAL ANALYSIS 105 the only available data class their potential for reveal- ing relevant patterning should be explored. A distinc- tion between large/small tools and tool fragments was made with reference to a histogram of all tool maxi- mum lengths for Area 2. Given a skewed distribution with over 60% of the tools less than 40 mm long, small was defined as <39.9 mm, large as >40.0 mm. Given the relatively small number of tools and their variable numbers across the units, it did not seem advisable to calculate a size-sorting index. Instead, a subjective evaluation was made in which units with either high absolute numbers of large tools or high proportions of large to small tools were identified and marked on Figure 62. At the north lobe, units with large tools lie between hearths 7 and 8, adjacent to hearths 2 and 6, and associated with possible hearth 1 9. The latter con- centration is strongly affected by the presence of sev- eral quartz block cores. There are no units with con- centrations of large tools outside the open northern side of the lobe, as might be expected from a "door dump." At the south lobe, a unit with large tools is directly associated with hearths 3a and 3b while anoth- er lies in the relatively debitage-free "corridor" between hearths 3a and 4. Large tool units also occur at both the eastern and western edges of the middle area. Most of the large implements on the eastern side are unused quartz block cores. As discussed previously, these may represent a raw material cache rather than a discard dump. In sum, there is little evidence for the systematic removal of larger tools from within the modules to secondary deposits elsewhere. Debitage might also have been subject to mainte- nance related size-sorting, especially quartz, the expe- dient reduction of which produced a wide array of large chunks. Quartz chunks might be more annoying on a dwelling floor than small Ramah chert retouch flakes and the larger quartz fragments would be easi- er to collect and discard outside a dwelling. However, size-sorted patterns of debitage might also reflect the spatial staging of lithic reduction sequences. Initial 106 CHAPTER 6 Tool class • Cluster 1.5- CO^ 1.0- Axis 1 (38.4%) 0.5- 9 7' -1.0. .OTH^ 1 ^PR -3.0 -2.0 1.0 *H 2.0 1 3.0 11 1 ST^ BC^ -0.5- -1.0- -1.5- 'SB . 2 PT 4 ^SP Axis 2 (33.5%) 60/ Area 2A: correspondence analysis plot of tool classes byk-means cluster, based on the overall Area 2 nine-cluster solution. ing the associated tool context. Reduction related size-sorting might be indicated by associa- tions between block cores and larger debitage fragments and bipolar cores and smaller deb- itage fragments, while mainte- nance related size-sorting might be signaled by the co- occurrence of both core types and larger debitage sizes. If all reduction stages occurred at the same locale the entire range of core types and debitage sizes would be present. Investigation of quartz size-sorting was undertaken by reduction of quartz block cores resulting in the pro- duction of large fragments might occur in one area, then some of those fragments could be removed to another location for further bipolar reduction, produc- ing a concentration of small fragments. Nonetheless, it might be possible to distinguish between maintenance related and reduction related size-sorting by consider- calculating a size-sorting index (SSI; Wandsnider 1996, see Chapter 3) for each of the 1 m^ quadrats at Area 2A. The SSI summarizes the representation of small and large items in a quadrat, with high negative values indicating a dominance of large items, high positive values a dominance of small items, and values near indicating a mixture of sizes. The large and small flake NUKASUSUTOK-5: SPATIAL ANALYSIS 107 o H11 OH17 Units with concentrations of large tools H9 H)2 H15 o 0.11 0.01 -i^ 0.85 (jO.S? 0.03 0.11 0,08 -0.81 o.-ts -0.01 ') 0.36 -0.36 Size-sorting index (SSI) predominance of large quartz flakes predominance of small quartz flakes u -0.02 11 62/ Area 2: size distribution of tools and quartz flakes. categories were constructed with reference to a line graph of quartz flake sizes for Area 2 as a whole. Around 85% of the flakes were <35 mm in size and the graph flattened out markedly above 35 mm, so "large" was defined as >35 mm. Figure 62 displays the size- sorting indices; units dominated by small or large flakes are highlighted. At the north lobe there is a unit with predominantly small flakes that lies between hearths 2, 7 and 8. On its northern flank are three units with a predominance of large flakes; two of these also have significant numbers of large tools, as does one unit to the southwest. Hearth 6 lies in one of the aforementioned units dominated by large flakes, but adjacent units to the north and east are dominated by small flakes. Hearth 1 is situated in a unit with prima- rily large flakes and adjacent to a unit in the middle subarea with mainly large tools (quartz cores). One unit north of the north lobe (49x79y) has a preponder- ance of large flakes, but its location in the area of probable overlap between Areas 2a and 2b renders it problematic. At the south lobe the unit containing hearth 3a is dominated by large quartz flakes while an 108 CHAPTER 6 1 1 6 2 1 1 1 Total biface-thinning flakes 1 2 11 3 6 1 4 5 1 3 2 4 3 2 2 7 1 2 2 2 1 :c ) 4 2 7 7 2 1 2 -( 8 4 8 21 2 7 \ 1 o 14 1 5 5 9 19 1 ^ 4 20 ^ 4 3 3 7 9 24 16 6 @ 3 1 63 22 12 5 5 2 7 5 0" 7 30 a:: 1 2 3 9 10 1 1 1 2 5 6 6 67 65 25 C 9 L ( ) 3 2 1^ 2 . 1 14 2 i 6 11 24 9 c 2 2 1 2 1 2 2 ( . 1 2 1 2 1 Large (>30mm) biface-thinning flakes 1 1 t 1 c 1 1 :> 1 - o T 2 2 1 3 / (J 2 4 1 10 7 3 1 1 1 _ 1 2 10 2 4 5 3 C ) ( ) _^1 4 2 2 4 4 1 1 ( 1 1 63/ /Area 2: distribution of Ramah chert biface-thinning flal30 mm), an unusually high proportion. Figure 97 shows the distri- bution of BTFs by quadrat; three were associated with Feature 1, three with Feature 2 and 1 1 with Feature 3. Eight of the ten large BTFs were associated with Feature 3. Dating Table 37 summarizes the radiocarbon dates for Area 3. Only one of the four dates is reliable. The date of 5090±95 B.P. was run on alder/birch and willow char- coal procured from the subsurface of Feature 1 , thus it is contextually more secure than the overlying charcoal deposit from which the 2845±70 B.P. date was derived. The latter sample and all the other later date samples contained coniferous material (Dosia Laeyendecker, personal communication). Given that pollen (Short NUKASUSUTOK 5: AREAS 1 AND 3 149 L-1 quartz flakes 22x121y 19x121y 23x1 15y L-2 quartz flakes 22x121y 19xl21y 18x1 15y 23x1 15y 3 119 94 84 4 10 18 105 282 20 5 69 8 245 48 2 8 15 74 233 62 9 9 31 5 11 Quartz flakes 86/ Area 3 distribution of quartz flal Ground slate flake Scfiist tablet • Utilized flake 90/ Area 3: tool distribution and k-means cluster analysis. oval deposits of red ocher (Features 1 and 3) lie oppo- site each other along an axis that is almost north- south. The other two features, which either lack red ocher (Feature 4) or contain only a modest amount (Feature 2), lie opposite each other along an east-west 23x1 15y 22x121y 19x121y Block and D=1 18x115/ disc cores D= disc core 92/ Area 3 distribution of block cores. axis. This may be merely coincidental and we have no means of ascertaining contemporaneity between the features other than a couple of refits possibly connect- ing Features 1 and 2. Nonetheless, the nearly equidis- tant spacing of the features relative to each other sug- gests deliberate placement. The only patterning in lith- ic materials consonant with this spatial arrangement is that the major red ocher features (1 and 3) tend to be more strongly associated with Ramah chert and slate. Understanding these features and their contextual relationships will require comparative data from other Maritime Archaic sites. 154 CHAPTER 7 23x1 15y 22x121y 19x121y 1 1 1 1 ar cores 1 18x115y 95/ ,4refl 3; distribution of bipolar cores. 18x1 15y 22x121y 19x121y ST=.5 SB=.5 ST=1 C=1 SB=5 — ; ST=.5 tools 18x1 15y ST= slate tool (fragment) SB = slate blank C = celt 94/ Area 3: distribution of slate tools. 23x1 15y 22x121y 19x121y Ground slate flakes 18x1 15y 22x121y 19x121y FK=1 Bl=1 Miscellaneous tools FP=1 FP=2 ES=1 FP=1 ST=1 ST=1 81=1 18x1 15y FP = flake point ES=endscraper ST = schist tablet Bl=biface FK= flake knife 96/ Area 3: distribution of other tool classes. 23x1 15y 22x121y 19x121y Blface-thinning flakes total (large) i(i; 2(1) 10(7) 1(1) 18x1 15y 97/ Area 3: distribution of Ramah chert biface-thinning flakes. 95/ Area 3: distribution of ground slate flakes. NUKASUSUTOK-5: AREAS 1 AND 3 155 O f Z 3 5 98/ Nukasusutok-5: stemmed bifacial points, a: Area 1 surface: b: Area 2A: c: Area 2C1 . CHAPTER 8 Nukasusuto Material Culture The preceding chapters outlined the spatial patterning at Nukasusutok-5 and made general references to the material culture insofar as it was relevant to the spatial patterning. The function of this chapter is to present a descriptive overview of the assemblages from each subarea. Each assemblage is broken down by tradition- al implement classes and the formal variation in these classes is summarized. A systematic consideration of technological organization at the site is presented in Chapter 9. DEFINITIONS For the most part, the categories of lithic implements used here follow the system developed by the Smithsonian Institution during the 1970s. There are deviations, of course, and some categories need more precise definition. Consequently, a series of clarifica- tions is presented prior to describing the assemblages. Bifacial Points and Flake Points A bifacial point is retouched over most of its dorsal and ventral surfaces; it may be made on a flake or from a bifacial preform. A flake point is made on a flake and has retouch limited to its margins, either bifacial, uni- facial or on alternating edges. Bifaces and Biface Preforms Biface denotes implements with bifacial retouch, either a generic form that does not suggest a particular end product or an implement fragment that is not further NUKASUSUTOK-5: MATERIAL CULTURE identifiable. Biface preform denotes bifacially retouched implements that are assumed to represent the penultimate reduction stage for a specific imple- ment class. Endscrapers, Flake Knives and Utilized Flakes Endscrapers are small implements, generally trapezoid or triangular in form, with clearly defined abrupt angle retouch at one end and varying degrees of retouch on their lateral edges. Flake knives are flakes with fairly distinct, continuous and moderate-to-abrupt retouch along a lateral edge; this form might also be termed a backed knife or flake. The term does not exclude a scraper function. Utilized flakes bear traces of inten- tional retouch or use wear that is less distinct, gener- ally discontinuous and usually, though not always, shallow angled. Bipolar Cores This term replaces the problematic piece esquillee. The latter seems to have been introduced into Labrador archaeology by McChee and Tuck (1975); a contemporary paper (Fitzhugh 1975:122) refers to them as "quartz nuclei or wedges." There has been a long debate in the lithic technology literature regard- ing bipolar cores and piece esquillee; since these implements constitute such a large part of Early/Middle Maritime Archaic assemblages the main points of the discussion will be summarized here. The heart of the debate is whether piece esquillee are 1 57 b c 99/ Nukasusutok-5 bifacial points and bifaces. a: stemmed point, Area 2A: b: stemmed point, Area 28; c: side-notched biface, Area 2A. d bipoint preform. Area 2C2, e bipoint. Area 2A. f biface, Area 2B. tools or exhausted bipolar cores. MacDonald's (1968:88) discussion of Paleoindian lithics regards them as tools for wedging or slotting bone, antler or wood. LeBlanc (1992) provides evidence suggesting this may be the case in some contexts. But most com- mentators have either regarded piece esquillee as bipolar cores or have tried to clarify the distinctions between the two (e.g., Flenniken 1 981 ; Hayden 1 980; Kuijt et al. 1 995; Lothrop and Cramly 1982; Shott 1989). When cryptocrystalline materials are involved the bipolar technique is seen as a method for reduc- ing small nodules that cannot be handled by direct percussion. In the case of vein quartz, however, the non-isotropic structure and resultant blocky fracture renders attempts at controlled flaking by direct per- cussion difficult. Consequently, the use of vein quartz is frequently associated with bipolar reduction tech- niques aimed at producing large quantities of deb- a ^ c 6 1 00/ Nul c u 8 - 10 15 Bivariate metric analysis provides more interesting results. Figure 107 is a scatter diagram of Axis 1 length against thickness, coded for dif- ferent raw material types. There is a slight tendency for three groups in the plot: (1) speci- mens <20 mm long and <8 mm thick, (2) those from 21-34 mm in length and between 6-14 mm thick, and (3) those >30 mm long and >12 mm thick. Ramah chert specimens cluster in the shorter/thinner area of the plot while quartz is spread out more broadly, although with a clear break between "medium" and "large" groups. The Mugford and black chert specimens lie in the mid- dle. The contrasts between quartz and Ramah chert likely indicate differences in the sizes of the raw mate- rials when bipolar reduction commenced: quartz reduction began with block cores and large flakes while Ramah chert bipolar cores are probably the result of recycling broken tool fragments. By way of comparison, Flenniken (1981:48) divides his Northwest Coast quartz bipolar cores into three size categories, although he provides no his- togram or scatter plot data on the shape of the distri- butions. Lothrop and Cramly's (1982:16) mostly chert pieces esquillees exhibit a unimodal size distribution suggestive of a continuous reduction process. Although the non-unimodal size distribution at Area 2A may simply be an artifact of sample size or reflect stages of core exhaustion (MacDonald 1968:86), it could also be related to purposeful selection of core sizes for the production of different types of flakes. For example, Flenniken (1981:43-46) identifies what he terms "microliths" as deliberate products of the bipolar reduction of quartz. This category may be A Ramah Quartz • Chert 20 25 30 Length (mm) 35 40 45 107/ Area 2A bipolar core size distriiyution. more a reflection of the investigator's expectations and sorting procedures than of prehistoric intentional- ity, but it is clear that careful bipolar reduction can produce at least some fairly regular columnar or linear flakes, although they will likely constitute a small pro- portion of the total debitage. The quartz debitage from Area 2A (and the other subareas) contains little that resembles deliberately produced linear or columnar flakes. Consequently, it is more likely that discontinu- ities in the bipolar core size distribution are the result of selectivity in the size of shatter chosen for further bipolar reduction. It is difficult, however, to see the smallest bipolar core size group as the result of reduction aimed at pro- ducing flakes since flakes produced from these cores would be extremely small and cores under 20-25 mm are difficult to hold between the fingers without risk- ing injury during bipolar percussion. If any of the bipo- lar cores were used for splitting/wedging functions it was perhaps these. Block Cores Area 2A produced a total of 37 quartz block cores, which can be divided into five sub-groups. The first NUKASUSUTOK-5; MATERIAL CULTURE 165 f108/ Nukasusutok-5 slate and schist tools, a: ground slate point, Area 2A. b flake adze. Area 2A, c: reworked flake adze. Area 2A, d slate blank. Area 2A. e celt/adze blank. Area 2A, f: schist whetstone, Area 2A. group consists of 1 1 tabular or sub-tabular cores, unmodified except for one or two flakes removed (e.g., Figure 1 10:b). Summary statistics are as follows: Length: r = 58.3-1 2 1 .4 mm, x=82.1, sd = 20.1 Width: r = 41 .4-70.3 mm, x = 56.6, sd = 1 1 .6 Thickness: r = 1 5.7-46.7mm, x=31.7, sd = 9.06 The second group consists of five quartz beach cobbles marked by various degrees of water-rolling. Four are battered and two of these were split in half. The remaining specimen has only a single flake removed. Their size range is slightly smaller than for tabular cores. Summary statistics are as follows; Length: r = 29.8-63.5 mm, x=52.2, sd = 1 3.4 Width: r = 25.1-52.6 mm, x=43.1, sd = 12.1 109/ Nukasusutok-5: ground slate tools, a: adze, Area 2A; b-c: celts, Area 2C. Thickness: r = 22.7-35.2 mm, x=30.3, sd= 5.8 The third group consists of 1 3 cores with traces of bipolar percussion. Some of these could be called bipolar cores, but they are distinguished from the lat- ter because of their size and their place at the initial phase of the reduction process. Figure 110:d illus- trates an example with crushing on its long sides and on a flat platform at one end; the opposite end has large step fractures. The core is broken across its width, but there was clearly an attempt to split it lengthwise using bipolar percussion. Summary statis- tics are as follows: Length: r= 53. 1 -1 22.8 mm, x= 72.4, sd=18.9 Width: r= 35.2-70.9 mm, x=51.5, sd=10.2 Thickness: r= 25.5-43.7 mm, x= 33.0, sd= 6.1 166 CHAPTER 8 / 1 0/ Nukasusutok-5: quartz biface and cores, a: biface, Area 2A: b: tabular block core, Area 2A: c: core/hammer- stone, Area 3; d: block core with bipolar percussion, Area 2A. 1 1 1/ Nukasusutok-5: slate and schist implements, a-c: ground slate point tips. Area 2A, d: ground slate tool frag- ment (point stem''). Area 2A; e schist tablet. Area 3: f: ground slate tool fragment. Area 2A: g schist tablet. Area 3. The fourth group consists of two specimens, one discoid the other lenticular in form. The former has three or four flakes removed while it is uncertain whether the latter has been culturally modified. Finally, the fifth group consists of 5 cores with irregular shapes and the following summary statistics: Length: r = 50.2-82.2 mm, x = 66.2, sd = 13.2 Width: r = 40.2-65.9 mm, x=52.1, sd = ll.l Thickness: r = 30.4-48.6 mm, x=36.6, sd = 6.9 Ground Slate Points Portions of four ground slate points were found at Area 2A. Three of the four are distal fragments while one is most of a medial portion. Two of the distal fragments have broad flat blades (Figure 1 1 1 :a-b); one is made of green slate the other of beige slate. The third distal fragment is narrower with a lenticular cross-section (Figure 1 1 1 :c) and made of gray slate. The medial specimen is made of green slate and consists of the lower portion of the point's blade and the upper part of its stem (Figure 1 08:a). Its blade element is very thin and flat (W= 22.6 mm, TH= 3.5); the stem is formed by a slightly rounded constriction. Ground Slate Adzes Three examples, all made of green slate, suggest two types of slate adzes. The first (Figure 109;a) is a large triangular form with bifacially retouched margins and a plano-convex cross-section (L= 105.6 mm, W= 53.0, TH= 1 3.3). Very fine polishing at the bit end extends 40-70 mm back from the working edge. The bit has an edge angle of 60 and exhibits very little evidence of wear. The poll end of the tool has a small surface NUKASUSUTOK-5. MATERIAL CULTURE 167 ground to an angle of 60°; this may be a trace of a sec- ond reworked bit or a hafting modification. The second example (Figure 1 08:b) is a small trapezoid adze made on a flake with a remnant striking platform at its poll end (L = 64.5 mm, W = 33.9, TH = 6.5). The implement has light dorsal retouch on its lateral margins, a bit angle of 35 and polishing limited to within 1 2-1 8 mm of the working edge. The third specimen (Figure 1 08:c) is probably a reworked flake adze similar to the latter type. It consists of a triangular flake with fine abrupt retouch on the distal end, slight retouch on its lateral edges and polishing remnants on one distal corner and one lateral edge. Its maximum width (33 mm) is very similar to that of the previously described flake adze. Slate Blanks Three items are classified as slate blanks. One is a tri- angular celt or adze preform of green slate (Figure 108:e), bifacially flaked over its entire surface, with preliminary grinding of a few facets near the bit end (L = 62.9 mm, W = 42.7, TH = 1 7.0). The second is a celt or adze preform of gray slate made on a thin slate plate (Figure 1 12:b). The implement is unifacial- ly retouched on its lateral margins and widest end; two step fractures at the wide end are ground slight- ly (L = 82.4 mm, W = 49.0, TH = 1 0.7). The third item is a bifacially flaked fragment of green slate (Figure 1 12:d). Slate Plates Of the nine slate plates or fragments thereof, six are made of green slate, two of red-brown slate and one of beige slate. All but one are thin plates of slate with varying amounts of edge retouch (e.g., Figure 1 1 2;a, e, f). The remaining example is a thicker ovoid piece with bifacial retouch on one edge and extensive step frac- turing on the opposite edge (Figure 1 12:c). Ground Slate Flakes and Tool Fragments Of the 49 items categorized as ground slate flakes, 24 (49.0 %) are made of gray slate, 1 6 (32.7 96) of green slate, four (8.2 %) of gray-banded slate, two (4.1 %) of red-brown slate, two (4.1 %) of indeterminate color slate and one (2.0 %) of greenstone. Seven of these specimens are edge-struck flakes from tools; four have medial ridges while three have convex surfaces. Two items are classed as ground slate tool fragments. One is a piece of gray slate with one edge ground perpen- dicularly and two other edges bifacially flaked (Figure 1 1 1 :f), apparently a reworked implement fragment. The other is a piece of green slate with rounded later- al edges suggestive of a ground slate point stem (Figure 1 1 1 :d), Hammerstones A single hammerstone was recovered, consisting of a small water-rolled cobble with battering concentrated at one point. Whetstones The single whetstone (Figure 108:f) is a thin bar of dense gray schist (L = 87.2 mm, W = 30.7, TH = 5.2). The working edge runs the length of one side of the tool, with wear on both faces but mostly on one. The opposite lateral edge is ground perpendicularly. Several linear incisions made by a pointed tool are vis- ible on one face of the implement as well as on the perpendicular non-working edge. These do not seem to constitute decoration. Schist Tablets and Ground Schist Fragments Three schist tablets and two fragments of ground schist were recovered. The three tablets are all made of chlorite schist. Two (Figure 1 1 3:a, b) are rectangu- lar with somewhat battered lateral edges suggestive of deliberate shaping (L = 117.4 mm, 100.1 mm; W = 67.1, 56.4; TH = 31.8, 21.2). Both have one smoothed surface, possibly from grinding. The third tablet (Figure 113:d) is a knife-shaped bar with no obvious modification (L = 165.0 mm, W = 44.6, TH = 1 7.2), so it is uncertain whether the implement is a cultural affordance. 168 CHAPTER 8 Graver (?) An unusual tool made of black chert may be a graver (Figure 103:p). It resembles an endscraper with over- size spurs extending from each distal corner (L= 37.4 mm, W= 18.4, TH= 4.9). The left distal spur is shaped like a screwdriver tip while the right spur tapers to a perforator-like point. The concave distal edge exhibits scraper-like wear traces and parts of both spurs also exhibit use-wear. Another possibility is that the speci- men is an effigy. Stemmed Flakes Three Ramah chert implements are classified as stemmed flakes because they exhibit more systematic retouch than utilized flakes and a tendency towards stem definition, but they do not resemble flake points. One is a roughly triangular flake (L= 55.7 mm, W= 39.2, TH= 4.3) with a short stem formed by alternating unifacial retouch on a slightly indented left shoulder and more sharply defined right shoulder (Figure 104:a). The lateral margins of the "blade" have fine bifacial retouch limited to the margins. The second (Figure 104:b) is a proximal fragment with a stem formed by a single dorsal flake removal on each side; the lateral edge margins above the stem have fine bifa- cial retouch. The third specimen (Figure 104:c) is a tri- angular flake with dorsal retouch distally and on one lateral edge (L= 36.0 mm, W= 41 .2, TH= 3.5). Linear Flakes Two items are classified as linear flakes. One is made of Ramah chert (Figure 103:r) and has a short blade- like form with two medial "arrises," dorsal polishing on its distal end and dorsal utilization wear on its lateral margin (L= 38.7 mm, W= 14.0, TH= 3.4). This speci- men may be a product of columnar fracturing during bipolar reduction. The other example is a medial frag- ment of gray Mugford chert (Figure 1 03:q); one lateral edge is retouched ventrally along its entire length, the other is retouched dorsally along half its length (L= 28.4 mm, W= 14.6, TH= 2.3). A third linear flake frag- ment of Ramah chert is illustrated (Figure 103:s), but it was dropped from the analysis because of uncertain provenience. AREA 2B Bifacial Points The single bifacial point from Area 2B is a proximal fragment made of quartz (Figure 99:b). The specimen is crudely flaked bifacially and has a contracting stem. One lateral edge has an angled rather than indented shoulder while the other bears only a slight indenta- tion indicative of a tentative shoulder (W= 26.4 mm, TH= 1 3.5). Flake Points The single Ramah chert flake point is an uncertain specimen bearing limited unifacial retouch, some of which seems designed to taper the flake near its tip and even out one lateral margin. Bifaces The only biface (Figure 99:f) is the distal or proximal fragment of a thin, very finely flaked, asymmetric biface of Ramah chert (W = 45.3 mm, TH = 7.6). The quality of flaking suggests the implement is a finished product ("bifacial knife"), but it might be a preform for a lanceolate bipoint such as those illustrated in Figure 99:d, e. Comparison of the biface with the latter two points suggests a reduction process in which the sym- metric lateral edge was "keeled" against the hand while the asymmetric edge was reduced towards the bipoint form. This technique might have minimized the risk of breakage when producing the narrow lanceolate forms. Endscrapers The single endscraper (Figure 103:h), made of gray Mugford chert, is trapezoid with a slightly concave dis- tal edge and a unifacially retouched lateral edge (L= 15.1 mm, W= 21.1, TH=9.9, edge angle= 60 ). NUKASUSUTOK-5: MATERIAL CULTURE 169 / 1 2/ Nukasusutok-5 slate plates and blanks, a, c, e-f: plates, Area 2A: b, d: blanks, Area 2A. Flake Knives Both examples are fragmentary and made of Ramah chert. Utilized Flakes Of the 21 utilized flakes, 17 (71.4 %) are made of Ramah chert, four (1 9.0 %) of black chert, one (4.8 %) of slate and one (4.8 %) of quartz. Bipolar Cores Of the six bipolar cores, five are made of quartz, one of Ramah chert. Most of these are fragmentary, so no summary metrics can be presented. Block Cores Two quartz block cores were recovered. One is a sub- tabular specimen, unused except for the possible removal of two flakes, the other is unused and prism- / / 3/ Nukasusutok-5: schist tablets, a-b, d: Area 2A; c: Area 2C. shaped (L= 78.2, 1 1 5.8 mm, W= 56.0, 90.6, TH= 39.0, 70.3). Ground Slate Flakes and Tool Fragments Of the five ground slate flakes, four are made of green slate and one of gray slate. A sixth flake, made of green slate, is an edge fragment from a celt or adze. Grindstones (?) Two conjoining fragments of a sandstone plate were found at Area 2B. It is uncertain whether this was a cul- tural affordance, but it could have served as a grind- stone for slate tools. Bone Pendants A small fragment of a bone pendant was recovered from the fine-sieved charcoal sample taken at hearth 1 1 , dated 6040±90 B.P.. Ca. 9 mm in length and 2.4 170 CHAPTER 8 / 1 4/ Nukasusutok-5: fragment of bone pendant, Area 2B. (Length 9 mm). mm in thickness, the artifact has a gouged hole, flat- tened and rounded edges, one incised line and two small notches on one face, and a single incised line on the other face (Figure 1 1 4). AREA 2C Bifacial Points The single bifacial point is from Area 2C1 (Figure 98:c). Made from Ramah chert, it is nearly complete, finely flaked, exhibits a contracting stem, rounded shoulders and has a biconvex cross-section (W= 28.2 mm, TH= 8.5). It is virtually identical to a specimen from Area 2A (Figure 98:b). Biface Preforms The two biface preforms are made of Ramah chert and are complete. One could perhaps be classed as an / / 5/ Nukasusutok-5: Area 3 flaked implements, a-c: flake points: d: endscraper: e: biface preform; f: disc core; g: biface fragment; h: flake knife. unfinished lanceolate bipoint (Figure 99:d), but here it is treated as a preform since its asymmetric shape results from retention of a small portion of unfinished edge (L= 91 .1 mm, W= 28.6, TH= 1 1 .8; biconvex cross- section). As noted above, it is possible that Figure 99:f- d-e illustrates a reduction sequence. The other exam- ple is made of thermally altered Ramah chert and exhibits a lanceolate form with a striking platform at its base (Figure 102:c); it was broken about one third of the way down from its distal end (L= 48.2 mm, W= 19.1, TH= 7.0; biconvex cross-section). The break occurred during the final tip-to-base retouch sequence since the tip is well retouched but the medial and prox- imal portions are still roughly flaked. Bifaces All five biface fragments are made of Ramah chert. Two large fragments (Figure 101:c, d) are similar in NUKASUSUTOK-5: MATERIAL CULTURE 171 form: asymmetric with slight stemming or notching and plano-convex cross-sections resulting from exten- sive retouch on their dorsal surfaces but only margin- al retouch on their ventral sides (W = 41.1, 45.5 mm, TH = 10.7, 13.3). Figure 101;d might be a stemmed point preform, but both could conceivably be early stages in the production of bipoints using the asym- metric reduction technique described previously. A third specimen is the proximal fragment of a large biface made on a flake (Figure 1 01 :b), with a remnant striking platform on its proximal end and retouch limited mostly to its lateral edges (W = > 60 mm, TH =10.3). Another implement is the distal end of a large thick biface, possibly a quarry blank (Figure 102:i). The last example is a lateral edge fragment of thermally altered Ramah chert. Endscrapers One of the two specimens is made of black chert, the other of Ramah chert. The black chert implement (Figure 103:e) is trapezoid in form, has double concav- ities on its working edge and abrupt dorsal retouch on its lateral margins (L = 22.4 mm, W = 23.8, TH = 9,0). The Ramah chert specimen (Figure 103:m) is problem- atic since Maritime Archaic endscrapers are almost never made from this material. The implement has a stem created by unifacial retouch along its lateral mar- gins (dorsally on one edge, ventrally on the other). The distal "ears" produced by the stemming have dorsal retouch and the distal edge itself is broken, with two notch-like flakes removed and the subsequent imposi- tion of slight edge retouch (L = 36.7 mm, W = 31.5, TH = 7.7). Polishing is present on the distal corners. Utilized Flakes Of the 31 utilized flakes, 26 (83.9%) are made of Ramah chert, four (1 2.9%) of quartz and one (3.2%) of black chert. Bipolar Cores Only two bipolar cores were found at Area 2C; one is made of quartz, the other of Ramah chert. Block Cores Of the 1 9 block cores, 1 8 are made of quartz, one of gray quartzite. Of the quartz cores, 1 1 were tabular in shape. Of these, eight were unused except for minor flaking, four may have been fractured off larger blocks and two have traces of bipolar percussion on their long ends. Four have slightly rounded cortex suggestive of water-rolling. Five cores were prism shaped. Three of the latter are virtually unused while 2 exhibit traces of bipolar percussion. Two cores are simply fragments of blocks. The gray quartzite core is a fragment of a water-rolled cobble split by bipolar percussion. A few flakes removed on one edge might indicate use as an expedient scraper. Summary statistics are as follows: Length: r = 48.7-97.7 mm, x = 74.9, sd = 15.8 Width: r = 31.9-88.9 mm, x = 56.5, sd = 15.5 Thickness: r = 19.1-70.4 mm, x = 33.5, sd = 12.5 Disc Cores The single example of a disc core is made of Ramah chert (Figure 1 05:j). It is a small bifacial disc with sharp edges and no evidence for crushing that might indicate bipolar reduction (L = 40.0 mm, W = 39.6, TH = 27.5). Ground Slate Celts Two celts were found at Area 2C, one complete and the other near-complete; both are made of green slate. One is a large sub-triangular specimen (Figure 109:b) with bifacial flaking along its lateral edges and on its poll end (L = 148.3 mm, W = 60.2, TH = 19.7). Polishing is limited to an area 38-48 mm back from the bit edge. The bit has little damage besides some crushing and small flakes removed from near the mid- dle of the working edge. Near the poll end some of the flake facets are polished, presumably hafting-related. Red ocher stains are visible on the celt surface. The second example (Figure 109:c) is a small, presumably sub-triangular, celt with bifacial retouch on both later- al edges, one of which has a thin perpendicularly ground facet (L = >82.0 mm, W = 39.2, TH = 1 2.4). One CHAPTER 8 face of the tool is mostly polished and flat; the other face is completely polished, but the edge portion has canted grinding that produces a ridge running parallel to the lateral edge. Ground Slate Celt/Adze One artifact of greenstone is the proximal (poll) frag- ment of a celt or adze (Figure 108:d). It is bifacially flaked and the lateral edges have slight perpendicular grinding. The poll edge is slightly ground on one side and one medial flake facet is slightly ground. Slate Plates The lone slate plate, of indeterminate slate color (due to weathering), has fine flaking along one edge and an area of possible edge grinding. Ground Slate Flakes and Tool Fragments There are 22 ground slate flakes, of which eight (36.4%) are made of green slate, six (27.3%) of beige slate, three (1 3.6%) of gray slate, three (1 3.6%) of red- brown slate and two (9.1%) of gray-banded slate. One flake is an edge-struck fragment from a tool. Hammerstones The single hammerstone is a small egg-shaped cobble with slight crushing at one end. Schist Tablets Two schist tablets were found. One is a large rectan- gular tablet of chlorite schist (Figure 113:c) with battered edges and one smoothed surface sugges- tive of use in grinding (L = 171.0 mm, W = 67.7, TH = 27.1). The other is a thin triangular tablet of phyllite schist; it is uncertain whether this was a cultural affordance. Schist Fragments Six fragments of schist were collected, of which two display grinding indicative of cultural modification. Stemmed Flakes One Ramah chert implement is classified as a stemmed flake (Figure 104:d). Made on a long flake, it has light dorsal retouch on one lateral edge while a slight stem shoulder is formed by ventral retouch (L = 82.9 mm, W = 36.8, TH = 5.3). Perhaps it could also be classified as a flake knife. Red Ocher Nodules The vicinity of hearths 1 4 and 1 5 produced eight small red ocher nodules, weighing a total of 52 g. Three tiny nodules near hearth 12 weighed 0.4 g. AREA 3 Flake Points All three flakes points are made of Ramah chert. One is almost complete with a contracting stem (Figure 115:a); the margins of its lateral edges and stem are retouched unifacially on alternate sides and the blade element is slightly serrated (L = 43.1 mm, W = 22.0, TH = 2.2). Another specimen is a proximal fragment with a stem element formed by a retouched shoulder on one lateral edge and a shallow notch on the oppo- site edge (Figure 1 1 5:b). A striking platform is retained at the base of the stem and the blade margins are retouched on alternate sides (W= 21.4 mm, TH= 2.9). The third example (Figure 1 1 5:c) is missing part of its base and has slightly serrated bifacial retouch limited to its margins (L= 28.4 mm, W= 18.7, TH= 3.1). Biface Preforms The single example is made of Ramah chert and is a proximal fragment with a convex base (Figure 1 1 5:e). Bifaces All three biface fragments are made of Ramah chert. Two are edge fragments with deep flake scars, reminis- cent of quarry blanks (Figure 1 02:h, 1 1 5:g), while the third is the lateral edge of a smaller biface. NUKASUSUTOK-5: MATERIAL CULTURE 173 Endscrapers The single endscraper from Area 3 (Figure 115:d) is made of an unusual extremely fine-grained beige chert, presumably from Cape Mugford. It has a straight distal edge (50' angle) and distinct polishing on its lat- eral edges (L = 19.0 mm, W = 19.1, TH = 2.8). Flake Knives The single flake knife (Figure 1 1 5:h) is made of Ramah chert and has two distinct retouch areas, both dorsal; the lateral edge angle is 60 while the proximal edge is 20* (L = 54.8 mm, W = 35.1 , TH = 9.2). Bipolar Cores A total of five bipolar cores was found at Area 3; two are made of quartz, three of Ramah chert. Block Cores Of the 20 quartz cores, seven are tabular in form, six are made from rounded water-rolled cobbles, three are squarish or prism-shaped, two conical, two semi-dis- coidal and one irregular. Three tabular specimens are unused or exhibit minimal modification; two were probably split with bipolar percussion and two others were fractured. Two of the tabular cores are slightly water-rolled. Of the rounded cobbles, three bear traces of bipolar percussion while a fourth is heavily battered, suggesting use as a hammerstone or resistance to repeated efforts at bipolar reduction (Figure 110:c). One squarish and one conical core show traces of bipo- lar percussion. Summary statistics are as follows: Length: r = 38.2-93.9 mm, = 70.8, sd = 12.4 Width: r = 36.9-86.1 mm, = 56.5, sd = 12.2 Thickness: r = 24.0-72.2 mm, = 40.9, sd = 11.2 Disc Cores The single disc core (Figure 115:f) is a small Ramah chert specimen with bifacial flaking and severe step fractures on both faces (L = 44.7 mm, W = 39.3, TH = 18.8). / 16/ Nul26.2 mm, TH = >1 2.0) suggest it is not a 174 CHAPTER 8 projectile point but possibly the shaft of a more robust tool such as a small gouge. ^ Slate Blanks The single slate blank is a conjoined specimen made on a plate of gray slate (Figure 1 1 6:d). It is sub-rectan- gular in form and bifacially retouched along three edges; the fourth edge bears a remnant of the original flat plate edge (L= 99.6 mm, W= 52.2, TH= 8.0). Ground Slate Flakes and Tool Fragments Of the 1 3 ground slate flakes, seven (53.8%) are made of beige slate, four (30.8%) of gray slate and one each (7.7%) of green and red-brown slate. One flake is a tool fragment with a ground, heavily battered bevel (Figure 1 16:c). Schist Tablets and Fragments Two schist tablets were recovered. One is a small knife- shaped bar of chlorite schist (Figure 1 1 1 :e) that appears worn, although not necessarily by cultural modification (L = 81.9 mm, W= 28.6, TH = 6.7). The other is an unmodified plate of phyllite schist (Figure 1 1 1 :g); its association with Feature 3 suggests it was a cultural affordance (L = 1 1 7.3 mm, W = 1 01 .9, TH = 1 0.2). Red Ocher Nodules Tiny fragments of red ocher were associated with two features. Nearby Feature 2 were nine fragments weigh- ing a total of 3.2 g. Adjacent to Feature 3 were three fragments, weighing 14.2 g. THE NUKASUSUTOK-5 ASSEMBLAGE IN A COMPARATIVE PERSPECTIVE As discussed more fully in Chapter 9, significant prob- lems are encountered when comparing Maritime Archaic assemblages. Raised beach sites that have been reoccupied repeatedly, like Nukasusutok-5, have horizontal stratigraphy and potential palimpsest deposits. Consequently, surface collections, test pits and limited excavations that do not define behavioral- ly relevant units (such as dwelling modules) may pro- duce an unrepresentative material of limited value for inter-assemblage comparisons. Nonetheless, some general observations should be made regarding the position of the Nukasusutok-5 assemblage relative to other material from northern and central Labrador. Since the radiocarbon dates from Nukasusutok-5 are spread between 6000-5000 B.P. it is hardly surpris- ing that the material shares similarities with both the Naksak Complex of northern Labrador (7000-6000 B.P.) and the Sandy Cove Complex (5200-4500 B.P.) of the central coast. Nukasusutok-5 and Naksak sites share lanceolate bipoint projectiles, side-notched bifaces, convex-based bifaces, endscrapers, small celts and a high frequency of bipolar cores (cf., Fitzhugh 1978:Figures 6-9). They differ, however, in that Nukasusutok-5 lacks nipple-based points but has a dis- tinct inventory of flake points. Nukasusutok and the Sandy Cove Complex have in common tapered stem points, convex-based bifaces and copious use of quartz reduced by bipolar techniques. They differ in that Nukasusutok has more extensive use of Ramah chert while Sandy Cove lacks endscrapers. The general stylistic sequence of Maritime Archaic projectile points goes from early triangular and nipple- based types (7500-6000 B.P.) to more sharply shoul- dered, tapered stem forms (6000-3500 B.P.). Only two projectile points from Nukasusutok-5 are reasonably associated with the 6000 B.P. component. The speci- men from near hearth 1 1 at Area 2B, dated 6040±90, probably has the best context; it is quartz and has a tapered stem angled with the blade rather than clear ^In contrast, an Early Maritime Archaic gouge from Koliktalik-1 in the Main area has a relatively cylindrical cross-section (Fitzhugh 1978: Figure 6j) and Late Maritime Archaic gouges have triangular or rectangular cross-sections (Fitzhugh 1978: Figure 5m, Tuck 1976; Plate 33). NUKASUSUTOK-5: MATERIAL CULTURE shoulders (Figure 99:b). A larger but similar quartz point (Figure 99:a) was associated with the south lobe at Area 2A, which has one radiocarbon date of 5670±1 75 B.P.. A point from Area 2C1 was found ca. 2 m from hearth 1 0, dated 6050±80; it is made of Ramah chert and has a tapered stem with rounded shoulders (Figure 98:c). The latter point is virtually identical to one from the southern lobe at Area 2A (Figure 98:b), dated ca. 5600 B.P.. The several stem fragments from the site (Figure 100) provide no information on shoul- der form. The rounded shoulder points from Nukasusutok-5 (Figure 98:b, c) are very similar to a point from Karl Oom lsland-3 in the Nain area, dated 6080±380 B.P. and 6120±120 B.P. (Fitzhugh 1 985:83, 1990:301). Considered as a whole, the Nukasusutok points and stem fragments bear a general resemblance to those from the Cull Arm-1 site in the Nain area, dated 5605±160 B.P. and 3285±80 B.P. (probably multi-component; Fitzhugh 1978:66, 78), as well as slightly later sites in other regions: Okak-2 in Okak Bay (4765±85 B.P., 4905+80 B.P.; Fitzhugh 1978:66, 78) and Sandy Cove-2 at Hamilton Inlet (no radiocarbon dates; Fitzhugh 1 972:92, 288). In conclusion, elements of the lithic assemblages from the various subareas at Nukasusutok-5 resemble those from Naksak Complex sites dated ca. 6000 B.P., although there are also similarities with later sites in other regions. Initially, the latter tendencies led Fitzhugh to disregard the ca. 5500 B.P. radiocarbon dates and place Nukasusutok-5 in the 5000-4500 B.P. range, based on "...typological convergence toward Rattlers Bight point styles and greater use of Ramah chert at the expense of quartz and Sandy Cove similar- ities..." (Fitzhugh 1978:77). This placement has been repeated more recently (Clark and Fitzhugh 1992:196). It should now be clear from the larger excavated assemblage and additional radiocarbon dates that there is no basis for such a dating other than the problematic 4645+65 B.P. assay from Test Pit C at the western end of the site. The excavated assem- blage from Nukasusutok-5 can therefore constitute a baseline for defining a late Naksak phase in the 6000- 5000 B.P. time range. -r 1,0 FP SP SB -- 0.5 -1.5 Axis 1 (68.3%) h- OTH -1.0^ PI ^-0.5 • SC UF, A Tool class • Area 2B PR Bl 0.5 i^H- CO 1.0 1.5 ST CA 2C A ---0.5 TM -1.0 Axis 2 (22.0%) / / 7/ Correspondence analysis plot of tool classes and Nukasusutok-5 subareas. 1 76 CHAPTER 8 NuLasusutok-^: Inter-Area Co arisons elusions The previous chapters have described and interpret- ed each of the subareas at Nukasusutok-5. While each subarea stands on its own as an analytical unit, they will now be compared to consider the implica- tions of intra-site variation. Evaluation of intra-site variation is based on comparisons of tool assem- blages, lithic raw materials, technological organiza- tion and behavioral inferences drawn from spatial patterning. Area 1 is omitted, since the scale of investigation was too limited to draw conclusions. Areas 2A and 2B contained spatial configurations that were not completely exposed by excavation, but their assemblages will be considered nonetheless. The chapter ends with some observations on the sig- nificance of Nukasusutok-5 for understanding pat- terns in Early/Middle Maritime Archaic assemblage structure. INTER-AREA COMPARISONS Tool Assemblages The first characteristic to be compared is variation in tool assemblages. Table 38 outlines the tool class percentages at each subarea, showing obvious differ- ences in the proportions of utilized flakes, bipolar cores, block cores and slate tool fragments. Figure 1 1 7 displays a correspondence analysis plot based on the frequency data. The first dimension accounts for 68.3% of the variation and distinguishes Areas 2C and 3 on the right, pulled by block cores, from Area 2A on the left, pulled by bipolar cores. The second dimension, which accounts for 22.0% of the variation, contrasts Area 2B on the bottom, pulled by its high percentage of utilized flakes, with assemblages in which block cores are frequent and flake points have a modest significance: Area 3 in particular and Area 2A to a lesser extent. Thus, the subareas are distin- guished mainly in terms of lithic reduction and expe- dient tool variables. The more form-invested tool classes do not occur in high enough frequencies to draw reliable inferences about intra-site patterning. Tool Raw Materials Table 39 provides a percentage comparison of tool raw material frequencies between the subareas. Quartz tools occur in similar percentages across the Area 2 subareas while Area 3 has a slightly higher proportion. Ramah chert is also fairly even across the Area 2 subareas, but is slightly less frequent at Area 3. Mugford and black cherts occur in relatively low percentages over most of Area 2, with the excep- tion of Area 2B, which has a high proportion of black chert. Area 3 is notable for containing only a single tool of Mugford chert. Slate varies slightly, with sim- ilar proportions in Areas 2A and 2C, a higher propor- tion in Area 3 and lower percentage in Area 28. Schist is most frequent at Area 2C. Thus, the differ- ences between the subareas are fairly minor. NUKASUSUTOK-5: INTER-AREA COMPARISONS AND CONCLUSIONS Table 38. Inter-Area Percentage Comparison of Tool Classes. PT Bl PR FP SC UF SP CA SB ST BC CO TM OTH TOTAL 2A 3.8 4.7 1.8 2.9 6.5 28.4 1.2 0.9 3.5 1 5.0 16.7 10.9 2.1 1.8 341 2B 2.4 2.4 2.4 7.2 50.0 14.3 14.3 4.8 2.4 42 2C 1.0 5.2 2.1 2.1 32.0 3.1 1.0 22.7 2.1 20.6 7.2 1.0 97 3 4.5 1.5 4.5 3.0 19.4 1.5 1 .5 22.4 7.5 31.4 3.0 67 PT: points (bifacial), Bl: bifaces, PR: preforms (biface), FP: flake points, SC: scrapers (endscrapers and flake knives), UF: utilized flakes, SP: slate points, CA: celt/adze, SB: slate blanks (unmodified plates and blanks), ST: slate tool fragments (ground slate flakes and unidentified tool fragments), BC: bipolar cores, CO: block cores (quartz), TM: tool manufacturing (hammerstones, schist tablets, whetstones), and OTH: other (stemmed flakes, linear flakes, graver). Table 39. Inter-Area Percentage Comparison of Tool Raw Materials. Area 2A Area 2B Area 2C Area 3 Quartz 23.5 21.4 23.5 32.8 Ramah chert 46.3 47.6 38.8 35.8 Mugford chert 2.3 2.4 1 .5 Black chert 5.3 9.5 2 Slate 20.5 16.7 24.5 26.9 Schist 1.5 8.2 3.0 Other 0.6 2.4 3.1 Total 341 42 98 67 Flake Raw Materials The next comparison is flake raw material percent- ages, as summarized in Table 40 and visualized in a stacked column graph (Figure 118). There are mod- erate differences in the proportions of quartz and Ramah chert between the Area 2 subareas, but quartz almost totally dominates at Area 3, associat- ed with a high frequency of block cores. Areas 2B and 2C have more patinated (thermally altered) Ramah chert. Area 28 has a high percentage of black chert, while Area 3 has little slate and entirely lacks Mugford and black chert. Technological Organization Organizational studies of lithic technology in "Americanist" archaeology have treated lithic pro- curement and use as embedded within a regional settlement system (e.g., Andrefsky 1 998:1 89ff.; Bamforth 1991; Binford 1979; Nelson 1991; Odell 1996; Parry and Kelly 1987). Although a regional scale of analysis is impossible here, given the absence of detailed information on other Maritime Archaic lithic assemblages, the discussion is con- ducted with this broad scale in mind. Another perspective connects social agency theory with 178 CHAPTER 9 Table 40. Inter-Area Percentage Comparison of Flake Raw Materials. Area 2A Micd £.0 Area 3 Qudrtz 1 d P Oj.d rvdiilall Li Ici I fin 1 fi7 7 rdLlllaLcU rvdillctll 1 7 ^ Q ^ fi n n Mugford chert 0.7 1.4 0.2 Black chert 0.7 8.8 1.3 Slate 5.4 3.9 2.9 1 .5 Total 8248 2330 1809 1828 Nukasusutok-5 Flake Raw Materials 100 % 90% 80 % 70 % 60 % 50 % 40 % 30 % 20 % 10% 0% Slate Black chert Mugford chert n Paiinafed Ramah chert S Ramah chert Quartz Area 2A Area 2B Area 2C Subarea Area 3 / 1 8/ Stacked column graph of Nukasusutok-S flake raw materials by subarea Leroi-Courhan's concept of chame operatoire to focus on technology as sequences of choices struc- tured by cultural schemas (e.g., Chilton 1999; Dobres 2000; Gamble 1999; Karlin and Julien 1994; Perles 1992; Schlanger 1994; Stark 1998). This approach is used implicitly here at the site-specific level, with an assumption of linkages to the regional scale through the settlement system. Chapters 6-7 deployed certain organizational concepts as part of the descriptive framework for the spatial analysis, so the subareas will now be compared more explicitly from this perspective. Each lithic raw material is treated separately since they are used in different kinds of reduction systems. Quartz Quartz is ubiquitous in Early/Middle Maritime Archaic assemblages in central and northern Labrador, but largely goes out of use after 4500 B.P.. It is widely dispersed as veins in the Precambrian bedrock, including two outcrops identi- fied on Nukasusutok Island, one on a hill north of Nukasusutok- 5. Archaeologically, quartz occurs primarily in the form of unmodified or minimal- ly retouched flakes, block cores and bipolar cores, but it was also used occasionally for crude projectile points. The quartz reduction system began with the collection of small quartz blocks, either tabular pieces derived from the bedrock or water-rolled pebbles from beach deposits; both are in evidence at Nukasusutok- 5. The spatial analysis of Area 2A suggested that un- or minimally modified blocks were cached in the mid- dle area between the two behavioral modules, while most reduction occurred within the modules. Reduction was accomplished by both freehand and NUKASUSUTOK-5: INTER-AREA COMPARISONS AND CONCLUSIONS Table 41. Variation in Elements of Quartz Technology at Nukasusutok-5. No. Tools Block Cores Bipolar cores Flakes Area 2A 80 37 34 2588 Area 28 9 2 5 332 Area 2C 22 1 7 724 Area 3 24 1 8 2 1569 bipolar percussion, but the lacl< of partially retouched cores intermediate between the minimally modified blocks and bipolar cores indicates that reduction pro- ceeded rapidly from a raw block to an exhausted bipo- lar core. The desired product of the reduction process was flakes, since very few visibly retouched imple- ments are found. A wide variety of these flakes could have been used expediently for simple cutting/scrap- ing functions or specific size/shape variants might have been selected for hafting, along the lines of Flenniken's (1981:60-96) discussion of quartz "microliths" hafted in wooden handles being used for processing fish. Flakes might also have been selected for further bipolar reduction to produce even smaller fragments. At any stage in the reduction process bipo- lar cores could have been recycled for use as scrapers, wedges, etc. At Nukasusutok-5, quartz reduction characteris- tics vary slightly between the sub-areas. Table 41 com- pares aspects of quartz technology at each subarea, with the exception of the small assemblage from Area 1 . The subareas at Area 2A are combined, so some of the spatial variation is masked. For all subareas, block and bipolar cores constitute ca. 80% or more of the quartz items classed as tools. Thus, as already noted above for Area 2A, quartz was rarely used to produce recognizable retouched implements; the primary goal was flake production, although the cores could have been put to secondary uses. Areas 2A and 2B have rel- atively high proportions of bipolar cores relative to block cores and total tools, while Area 3 has few bipo- lar cores and Area 2C none. While these observations may be affected by incomplete excavation at Areas 2B and 2C, there is at least a hint that bipolar reduction was less significant at Areas 2C and 3. If bipolar reduction was less prominent at Areas 2C and 3, these subareas might also exhibit quartz flake size distributions with a better representation of larger sizes than the other subareas since bipolar reduction could lead to more thorough reduction of quartz than the freehand percussion technique. The line graphs of quartz flake size distributions (Figure 1 1 9) do not provide much support for this. Area 3 has the lowest percentage of material in the <10 to <20 mm categories (i.e., fewer small flakes) and a slightly higher percentage in the 35-40 mm category. Area 2C has the highest percentage of material in the 1 0-1 5mm category (small flakes) and a slightly elevated percent- age in the 25-30mm category. Area 2A, which has large numbers of bipolar cores, has the lowest percent- age of flakes in the smallest <1 0mm category, but the highest in the 15-20 mm and 20-25 mm categories. Thus, by themselves, the quartz flake size distribu- tions are not very clear indicators of differences in reduction patterns. More detailed analysis of quartz fracture patterns (cf., Callahan et al. 1992; Knutsson 1998) is necessary to fill out this picture. Ramah Chert Ramah chert was derived from its bedrock sources 300 km to the north in the Ramah Bay region (Gramly 1 978; Lazenby 1980). The material could be quarried direct- ly from the bedrock or easily picked up as large blocks or plates in stream beds or from loose talus slopes. These blocks were processed at or near the sources into biface blanks of varying sizes, which were then 180 CHAPTER 9 transported south. The blanks were modified as needed by bifacial reduction into the required tool form. Consequently, Ramah chert debitage in the Nain region con- sists mostly of biface-thinning and other secondary retouch flakes, with little trace of pri- mary reduction flakes bearing cortex. "Stage 1" quarry blanks seem to be rare in the Nain area, but smaller bifacial point pre- forms are frequent. Ramah chert was used primarily for projectile points throughout the time span of the Maritime Archaic, although flakes from biface reduction were selected for use as expedient tools (uti- lized flakes, flake knives, flake points). In contrast to Dorset, the Maritime Archaic almost never used Ramah chert for endscrapers. In sum, this is a classic example of a curated biface technology (cf., Kelly 1988; Parry and Kelly 1987). Table 42 compares ele- ments of Ramah chert technolo- gy between the subareas. Area 3 stands out for its general paucity of Ramah chert, yet it also has: a) a relatively high pro- portion of tools to flakes, and b) a high proportion of large biface-thinning flakes (BTFs), perhaps linked to the presence of possible quarry blank frag- ments among the bifaces. These variations may be minor. Quartz flake size distribution 35,0 -1 <10 <20 <30 <40 <50 <60 <70 <80 <90 <100 5 mm classes / / 9/ Quartz flake size distribution at Nukasusutok-5. Ramah chert flake size distribution 45,0 T <10 <20 <30 <40 <50 <60 <70 <80 5 mm classes i 20/ Ramah chert flake size distribution at Nukasusutok-5. Fine-grained chert flake size distribution <10 <15 <20 <25 <30 <35 <40 <45 <50 5 mm classes 121/ Fine-grained chert flake size distribution at Nukasusutok-5. NUKASUSUTOK 5 INTER-AREA COMPARISONS AND CONCLUSIONS 181 Table 42. Variation in Elements of Ramah Chert Technology at Nukasusutok-5. Tools Bifacial Points Bifaces Preforms Retouched Flakes* Utilized flakes Bipolar Cores BTFs (N/large) Flakes Area 2A 1 58 12 1 5 6 1 7 88 19 637/88 4957 Area 2B 20 1 2 1 5 1 149/8 1669 Area 2C 41 1 5 2 1 25 2 114/14 1006 Area 3 23 1 2 4 13 3 17/10 238 'includes flake points, flake knives and stemmed flakes Table 43. Variation in Elements of Fine-Crained Chert Technology at Nukasusutok-5. Tools Endscrapers Util./ret. flakes Bipolar Cores Flakes Area 2A 26 16 5 4 122 Area 2B 5 1 4 237 Area 2C 2 1 1 19 Area 3 1 1 but they parallel other behavioral differences linked to the presence of the red ocher features. Otherwise, the large number of BTFs at Area 2A is clearly associated with the substantial numbers of bifacial points, biface fragments and preforms. The bifacial tool component varies from highs of 2 1 % and 1 9.5% at Areas 2A and 2C, respectively, to lows of 1 3% and 5% at Areas 3 and 2B, respectively. The proportion of expedient tools (retouched and utilized flakes) is high in all the assem- blages, ranging from 63% to 85% of all Ramah chert implements. It is notable that Ramah chert occasional- ly was subjected to bipolar reduction, but this was a secondary technique implemented on broken tools or flakes produced by direct percussion. It is unclear whether the purpose was production of small flakes to maximize the use-life of the material or a specific func- tion such as wedging/splitting. Figure 120 charts the size distribution of Ramah chert flakes at the four subareas. The distributions have generally similar profiles except for the smallest <10 mm flake size category. For the latter. Areas 3 and 2A have relatively low percentages (17.2% and 24.4%, respectively) while Areas 2C and 2B have high- er percentages (39.2% and 36.6%, respectively). All areas were screened with the same mesh size, but the differences could indicate some kind of collection bias since the low-percentage areas were both excavated in 1979 and the high percentage areas in 1992-93. Reviewing the results for quartz (Figure 1 1 9), the Area 2A excavation from 1 979 also exhibits a slightly lower percentage in the smallest size class than the other subareas, which might support the case for collection bias. Looking more closely at the line graph, Areas 2B and 2C have almost identical profiles from the 1 5-20 mm class onwards. Area 3, however, has a consistent- ly higher representation of the larger flake sizes from 1 5-20 mm onwards, while Area 2A is intermediate. If we hypothetically assign a slightly larger number of flakes to the smallest size class to compensate for col- lection bias it is likely that Area 3 would still have a 182 CHAPTER 9 relatively low percentage in the smallest size class and the overall profile would remain similar. Although the difference may, therefore, be "real," the small sample size from Area 3 (238 flakes) should also be taken into account. Fine-Grained Chert The term fine-grained chert refers to both Mugford cherts of varied colour (gray, green, beige, patinated) as well as black chert, some of which was derived from the Mugford sources (Cramly 1978; Lazenby 1980), some of which might have been from sources in Sagiek Bay (Chapter 2). These cherts occur as thin tablets that are transportable without the systematic preliminary reduction used for Ramah chert. Nonetheless, not a single core of the material occurs at Nukasusutok-S. Fine-grained cherts were used almost exclusively for the production of small endscrapers, although one biface fragment is present in the Nukasusutok-5 col- lection. A few utilized flakes and bipolar cores are also present. As noted in the comparison of flake and tool raw materials (Tables 39, 40), fine-grained cherts occur in rather low percentages. The low frequency of these items is underlined in Table 43, which combines the Mugford and black cherts. Only Area 2A has a sub- stantial number of tools, while Area 2B has twice as many flakes but substantially fewer tools. This might be interpreted as a high rate of tool discard at Area 2A, contrasted with more production or retouch at Area 28 combined with less discard, or curation of finished products to other localities. The almost complete absence of fine-grained cherts at Area 3 reinforces the impression that this subarea differs from the others in function, meaning and chronology. It should be emphasized, however, that low overall frequency of fine-grained chert need not signify minor cultural or organizational significance. Considerable effort was expended to acquire a material that was used mostly for a single purpose: endscrapers. Figure 121 shows the size distribution for fine- grained chert flakes at Area 2. Again, the portion exca- vated in 1979 (Area 2A) has a lower percentage of material in the smallest size class. Areas 2A and 28 parallel each other closely, while Area 2A exhibits a slight peak in the 20-25 mm size class. Overall, there is a strong skewing towards small flakes of 10-15 mm or less, which implies emphasis on secondary and ter- tiary retouch rather than any form of core reduction. In sum, the flake size distribution, the small size of the finished tools (endscrapers) and the lack of cores sug- gest that fine-grained cherts were curated as finished tools or flake preforms. Slate The most likely source for the slates is the Cape Mugford region. Several colour and texture variants were used. In contrast to the fine-grained cherts, it appears that raw plates of slate were often transport- ed "as is" from the sources then reduced as needed. This is suggested by the presence of several minimal- ly worked plates in the Nukasusutok material. These raw materials were then roughed into blanks and pre- forms by direct percussion. The latter were then ground into finished tools, possibly with schist tablets. Table 44 outlines elements of the slate technolo- gy system at Nukasusutok-5. Area 2A was the obvious center for slate reduction. A considerable number of raw slate plates and tool blanks were present, either cached as raw materials or discarded during the man- ufacturing process. The frequency of ground slate flakes, in some cases clearly edge-rejuvenation flakes, attests to the reworking of finished or partly finished tools. Attempts to refit these flakes to the tools found in the subarea were unsuccessful, suggesting the "cores ' were removed from the site. Combined with the low frequency of finished tools, this could imply main- tenance of curated implements. Areas 2C and 3 exhib- ited only modest indications of tool production and maintenance while Area 28 had the fewest indicators of slate reduction. The flake size distribution for slate is portrayed in Figure 122. The line graph profiles are extremely variable, most likely an expression of ran- NUKASUSUTOK 5: INTER-AREA COMPARISONS AND CONCLUSIONS 183 Slate flake size distribution <30 <40 5 mm classes 122/ Slate flake size distribution at Nukasusutok-5. dom variations in small samples rather than meaning- ful patterning. the social composition of the groups. But the differences could also be the result of "fall- out" from activities positioned at different points on the pro- curement and reduction cycle; that is, a consequence of raw material availability and contin- gent organizational constraints rather than a direct reflection of site-function requirements or social variables. Sorting out these alternatives requires situ- ating Nukasusutok-5 within the regional and supra-regional contexts of Maritime Archaic settlement and procurement systems. SUMMARY The Maritime Archaic inhabitants of Nukasusutok-5 organized their lithic technology into two primary sys- tems: a) a highly expedient component using an abun- dant local raw material (quartz), and b) a highly curat- ed component involving non-local materials. The latter consisted of different reduction systems for each of the materials involved. The Ramah chert system involved the transport of large bifaces that were: a) reduced along a core-tool trajectory to produce "high performance" form-invested tools (points, bifaces), while b) debitage from core reduction was employed for expedient tools (utilized flakes etc.). Fine-grained cherts were transported as small tablets, flakes or fin- ished tools, with a strong focus on a specific imple- ment— endscrapers— and little expedient flake use. Slate was transported as raw plates, blanks and fin- ished tools; production and maintenance involved a time-consuming grinding process after initial reduc- tion. The subareas at Nukasusutok-5 differ slightly in the emphasis placed on these organizational systems. These differences may signal choices made in response to site-function considerations or variation in CONCLUSIONS Nukasusutok-5 is one of the few Early/Middle Maritime Archaic sites that has been excavated sufficiently to provide a reasonably representative sample for under- standing spatial patterning and assemblage composi- tion. As such, it is difficult to place the site in a broad- er systemic context because there are insufficient com- parable assemblages to work with. By way of conclu- sion, a number of points that could be pursued in future research are discussed. In the last major synthesis of northern Maritime Archaic culture-history, Fitzhugh (1978:72) grouped most of the Early/Middle Maritime Archaic sites in the Nain region into the "Naksak Complex," although he recognized that this entity was problematic: Many of these sites are deflated or partly eroded, with artifacts found on the surface and in buried deposits. Local topographic conditions rendered some sites useful for long periods, while others were occupied briefly. Until these factors can be analyzed it seems better to acknowledge the probable 184 CHAPTER 9 Table 44. Variation in Elements of Slate Technology at Nukasusutok-5. Tools'^ Ground flakes/fragments Plates/blanks Schist Flakes Area 2A 7 51 1 1 5 443 Area 2B 6 92 Area 2C 3 24 1 6 52 Area 3 2 1 3 1 3 20 Points, celts/adzes "lumped" nature of this complex (Fitzhugh 1978:72). Fitzhugh goes on to refer to "considerable typo- logical variation" at some sites, particularly those situ- ated in topographical locations that could be used over long periods of time, and he mentions sites of differ- ent character and dating that may form the basis for defining additional archaeological complexes, such as the Nukasusutok and Gull Arm Croups (Fitzhugh 1 978:72-77). This sense of barely ordered variability is understandable since the data base was derived most- ly from surface surveys and small test excavations, and radiocarbon dates were scarce. But ordering that vari- ability will be difficult without a better understanding of individual site structure. Nukasustok-5 illustrates some of these problems. It is located in a protected harbour on a prominent tombolo beach that was a magnet for settlement from 6000 B.P. to 4500 B.P. and perhaps later. Area 2 revealed three or four occupation components repre- senting at least two chronological phases (ca. 6000 B.P. and 5500 B.P.). Components of these two phases overlapped slightly at the junction of Areas 2A and 2B. Within Area 2A two behavioral modules that might rep- resent dwelling features were identified. The modules were similar in spatial organization but varied some- what in tool and flake raw material contents. Area 2C might have contained a similar dwelling module. Area 2B exhibited a different behavioral pattern consisting of a central pit-hearth complex ringed by deposits of fire-cracked rock. This pattern was interpreted as either an outdoor feature with a concentric drop-toss zone or a dwelling module with fire-cracked rock dis- carded outside along a tent wall. The Area 2 subareas displayed slight differences in their frequencies of tool types and lithic raw materials and in their technologi- cal organization, but the chronological distinctions suggested by the radiocarbon dates were not paral- lelled by clear differences in tool typology. The proba- bly later occupation at Area 3 (5000 B.P.) was not so much typologically different as it was behaviorally dis- tinct from the Area 2 occupations. At the intra-site level of analysis, the Nukasusutok-5 material underlines: a) the importance of excavating and analysing entire behavioral units so as to provide a meaningful baseline for comparison, and b) the need to acquire several such samples from large, complex and chronologically diverse sites in order to understand both their chronological position and their role in the regional settlement system. At the inter-site level, the variability uncovered in the Nukasusutok-5 material raises problems as to what is an adequate unit of inter-assemblage comparison. The subareas cannot simply be lumped together and com- pared with some other composite of assemblages or with more discrete assemblages. To be meaningful, comparisons and culture-historical integration must be based on behaviorally distinct assemblages, such as the possible dwelling modules at Area 2A. Another implication of the present analysis is that comparisons based on traditional tool typology NUKASUSUTOK-5: INTER-AREA COMPARISONS AND CONCLUSIONS 185 may be of limited value for understanding settlement systems and social process. It may be more useful to investigate other dimensions of variation such as pat- terning in different lithic reduction systems, as dis- cussed in this chapter. The intra-site analysis of Nukasusutok-5 suggests that assemblages could be scaled in relation to which components of lithic reduction systems are present or absent and in rela- tion to the articulation between expedient and more "organized" components of the technologies. Unfortunately, such analyses cannot be undertaken until such a time as a sufficient number of appropri- ate assemblages become available. The analysis of Nukasusutok-5 also contributes to the empirical and methodological aspects of Maritime Archaic dwelling structure studies. At Area 2A it was possible to identify two behavioral modules that might be dwelling remains. Two alternative interpreta- tions were suggested: (1) each module represents a small tent structure of 4 by 3 m in size; the spatial organization of activities at each structure was highly symmetrical and the dwellings had a common orienta- tion to a large central rock, and (2) each module was a component of a larger rectangular segmented struc- ture of 4 by 6-1- m in size; although the spatial organ- ization of the modules was highly symmetrical the modules were integrated into the rectangular struc- tures in slightly different ways. Both interpretations were problematic, however, since the multiple hearth features could indicate re-occupation events rather than discrete dwellings. Consequently, although the "longhouse model" provided a framework for identify- ing spatial patterning, the meaning of that patterning remained ambiguous. More detailed investigations of Early/Middle Maritime Archaic sites will be necessary to clarify how the Area 2A patterns fit into the range of variation in Maritime Archaic site structure. The methodological strategy outlined in Chapter 3 involved playing off a model-based analysis versus model-independent pattern recognition techniques. The model-based analysis focused on hearth-centered distributions that could be viewed within the segmen- tal or modular framework of the longhouse model. Qualitative visual analysis of flake distributions proved to be the crucial element in identifying behav- ioral modules. Other lines of evidence were then explored within this postulated modular framework, largely through the use of simple cross-tabulation techniques. K-means cluster analysis was used as a model-independent exploratory technique to identify tool groupings by physical proximity rather than by sorting them into preconceived behavioral modules. This proved useful for mapping hearth-centered distri- butions at the micro-level and for pointing to possible variations in the lithic materials associated with each feature, but the sample sizes in each cluster were too small to have much statistical significance. The clus- ter analysis could not provide a basis for combining hearth-centered distributions into larger behavioral units; ultimately, this had to be accomplished by qual- itative means and within model assumptions. A differ- ent model-independent technique, correspondence analysis, was used to explore patterning in flake raw materials at Area 2A. The results were consistent with the model-based analysis. Overall, the analysis showed that relatively sim- ple qualitative and quantitative methods are capable of detecting meaningful patterning at Maritime Archaic sites lacking in traces of architecture that would facilitate the definition of analytical units. But this patterning would have been difficult to identify if the flake distributions had not been plotted sys- tematically in the field. Consequently, future analy- ses of Maritime Archaic (and other) sites should point plot debitage distributions so as not to miss patterns that would be indistinct in quadrat sum- maries. Spatial data should be as high resolution as possible, otherwise it may be difficult to identify behavioral modules such as those that were crucial to the present analysis. 186 CHAPTER 9 Edward Flowers and Sam Saimat looking for basking seals on the sea ice near Dog Island, April 1994. NUKASUSUTOK-5: INTER-AREA COMPARISONS AND CONCLUSIONS 187 1 23/ Nukasusutok-l 2: overview towards the southwest. CHAPTER 10 Nukasusutolc-lZ: Earlt)/ Middle Dorset Axial Structures 6 Nukasusutok-1 2 (HcCh-14) was discovered during a 1979 survey when a well preserved axial structure (Structure-1 ) was observed almost completely exposed in a blow-out. The few tools noted on the surface indi- cated a Dorset affiliation, but the feature bore a strik- ing resemblance to Pre-Dorset axial structures and it seemed to differ from Middle and Late Dorset tent ring structures recorded in northern Labrador by the Torngat Archaeological Project (e.g., R. Jordan 1980:611). An Early Dorset affiliation was suspected, so the site was investigated further in 1980. It was revisited briefly in 1992. The site is situated at the head of a shallow bight on the south side of Nukasusutok Island, west of Wyatt Harbour (Figure 1 6). It lies at an elevation of 5.5-7.5 m asl. at the western end of a narrow east-west trending pass between steep rocky hills. At the time of occupa- tion, higher sea level would have resulted in a well-pro- tected cove extending considerably further inland than at present, providing an excellent location for boat landings. Lithic material occurs sporadically over an area of roughly 1500 m^, at least half of which is wholly or partially deflated, the rest is covered with lichen and crowberry. Two axial features (Structures 1 and 2) were clearly visible near the middle of the beach, while traces of a third (Structure 3) lay towards the southern side of the pass, and a fourth (Structure 4) was identified in a test pit on the northern edge of the pass beside a rock-fall (Figures 123 and 124). Three test pits (TP 1-3) in a small area on the south side of the pass revealed a rich tool deposit, but no trace of structural remains. It is possible that addition- al structures are present at the site, deeply embedded in the gravel. Most of the site is underlain by aeolian sands and gravel, but two large clay boils are located on a low ridge in the center of the main deflated area near Structure 2. The three test pits (TP 1-3) on the south side of the pass exposed a thick basal clay mixed with small cobbles. The 1980 fieldwork consisted of the complete excavation of two axial features (Structures 1 and 2) and the excavation of seven test pits, two of which revealed the presence of additional structures. The two excavated structures will be presented first, followed by the test units. STRUCTURE 1 Structure 1 was the lowest feature identified at the site, located at ca. 6.5 m asl. in the middle of the beach pass. When discovered in 1 979 the feature was almost completely exposed in a sand-gravel blow-out. A total of 23.25 m^ was excavated to expose entirely the ^Nukasusutok-1 2 was described previously in Hood (1 986). This chapter is based on the earlier report, but the material has been reworked, supplemented with tool descriptions and reinterpreted. It should be regarded as the "authoritative" account. NUKASUSUTOK-1 2 EARLY/MIDDLE DORSET AXIAL STRUCTURES 189 oQo-^ . rock fall OVO^O 1 1 24/ Nukasusutokl 2 sitemap. Structural components and recover any buried lithics. The structure consisted of a ca. 6 m long axial feature aligned NNW-SSE, running obliquely to the beach front (Figures 1 25 and 1 26). Excavation at the southern end of the structure required the removal of 25 cm of lam- inated aeolian sand and thin humus horizons to expose a 2 cm thick black humus zone that appeared to correspond with the floor level of the feature. This humus level was limited to the outer periphery of the excavation, mostly outside what were interpreted as tent anchor rocks. Consequently, the structure took the appearance of an axial feature bisecting a 3.5-4.0 m wide sub-rectangular patch of gravel. Although it is possible this gravel patch is a partial consequence of deflation, its orientation relative to the feature sug- gests a thin sod layer may have been removed to pro- vide a gravel floor for the dwelling or that it was tram- pled away during the occupation. The axial feature was 5.0 m long and 0.75-l.Om wide. It was constructed of parallel rows of thick flat- topped rocks embedded in the gravel end-to-end lengthwise, but was disturbed at its northern end. Several rocks on the eastern edge of the feature were slightly displaced to the west and on the western edge some of the axial rocks were missing. At both ends of the feature were high large rocks, morphologically unlike those composing the feature itself. These end- rocks lay perpendicular to the feature's long axis and formed the feature terminus, although they were placed slightly back from the feature and did not form a continuous alignment with it. There were no trans- verse partitions or paving stones inside the axial bor- der rocks, but at each end of the feature were the remains of hearths. Just south of the feature's center lay two thin flat slabs, one of schist. Both slabs were notched on one longitudinal edge and modified to be flat on the other edge. These were presumably soap- stone vessel support rocks. To the north of the fea- ture's center were a broken, blubber-stained, fire- cracked rock and a thin slate slab with a battered and notched edge, probably the remains of additional ves- sel support rocks. The only tools associated with the structure were the proximal portions of two Ramah chert microblades (W= 1 2.4, 1 2.4 mm, TH= 4.5, 2.8). One of these was surface collected from near the terminus rocks at the north end of the feature, the other was excavated from within the black humus zone at the southern boundary 190 CHAPTER 10 1x6y 5x6y 1x1y o (7 5x0y Gravel floor edge Burned fat-stained rock Schist slab F Flat rock N Notched rock • Microblade of the excavation. The latter specimen may not be associat- ed with Structure 1 since sever- al rocks at the humus zone level of this end of the excavation may be part of another buried feature. No flakes were found in Structure 1 . 1 25/ Nukasusutok-l 2: Structure 1. 1 26/ Nukasusutok-1 2: Structure 1, view towards the south. NUKASUSUTOK 1 2 EARLY/MIDDLE DORSET AXIAL STRUCTURES 191 ]27/ Nukasusbttok-l 2 Structure 2. STRUCTURE 2 Structure 2 (Figure 127 and 128) was located near the cen- ter of the beach, about 28 m east of Structure 1 and adjacent to the large clay boils. Its eleva- tion was no more than a meter above Structure 1 and it was ori- ented at an oblique angle to the beach front, running NE-SW. The structure consisted of an axial feature 4.0 m long and ca. 0.7 m wide, constructed of thick flat slabs embedded in the sur- face gravel end-to-end length- wise. The feature boundary rocks were displaced at the cen- ter of the alignment for a distance of 1.5 m and the scatter of rocks at both ends of the feature indicated a greater degree of disturbance than at Structure 1 . The southern end of the feature was bounded by two high large terminus rocks, structurally continuous with the feature, but the northern terminus was indistinct. The axial feature contained at least two and pos- sibly three hearth areas, one at each end of the align- ment and perhaps another in the middle. At the south- ern end of the feature two thin, flat, notched and blub- ber-encrusted schist slabs lay fallen between the fea- ture boundary rocks near a concentration of crumbled schist fragments. A third notched and burned fat- encrusted schist slab was wedged vertically against a western boundary rock. These notched slabs were probably used as soapstone vessel supports and frag- ments of a near-complete soapstone vessel were found nearby. Several small horizontally placed flat slabs inside the feature may be the remnants of a floor pave- ment. At the northern end of the alignment were a pair of fallen notched and burned fat- encrusted schist slabs that lay near a concentration of crumbled schist fragments and a soapstone vessel sherd. Several hori- zontal flat slabs, one of which was levelled with a shim 10x10y 12K8y ^ Axial passage border rock Burned fat-stained rock CD Schist slab F Flat rock N Notched rock rock, may be the remains of an internal floor pave- ment. In the disturbed center of the feature were three blubber-encrusted rocks, a stray notched schist slab and a soapstone vessel sherd. These may be the remains of a third hearth, but considering the degree of disturbance they could have been thrown there from the other hearth areas. The overall dimensions of the structure were diffi- cult to determine given the absence of obvious perime- ter anchor rocks and the lack of axial terminus rocks at the north end of the passage. But judging from the artifact distribution and the position of possible perimeter rocks the structure may have been about five meters long and four meters wide. Spatial Patterning Unlike lithic sterile Structure 1, Structure 2 contained 74 tools. The original spatial analysis of the distribu- tion patterns (Hood 1986) was complicated by artifact provenience problems related to missing records. During re-analysis only 53 of the 74 tools could be point-plotted with certainty; most of the missing items are microblades. Additionally, some of the point-plot- ted microblades could not be related back to specific 192 CHAPTER 10 1 28/ Nukasusutok-1 2: Structure 2, view towards the southwest. catalogued implements, which made it difficult to dis- tinguish reliably between utilized and unutilized microblades. Although the missing information could skew the analysis somewhat, it is necessary to work with what is at hand. A nephrite flake and a slate flake 7x1 Oy J L. lOxlOy 1 29/ Nukasusutok-1 2: Structure 2 tool distribution and k-means cluster analysis were also included in the analysis, bringing the num- bers to 76 total, and 55 plotted items. The point-plot of tools (Figure 129) indicates that most were concentrated within a 1-1.5 m dis- tance of the axial feature. Following the bilateral organization model, a simple cross-tabulation analysis was presented in the original site description (Hood 1986:64), dividing the provenienced tools into spatial units associ- ated with the feature or lying to the east or west of it. A revised version of this table is presented in Table 45. Most of the tools are located close to the feature or on its eastern flank. There is little clear pat- terning in tool class distribu- tions besides the limitation of Biface A Endblade Endblade preform • Microblade * Nephrite flake B Side-nolched biface Slate bitace • Slate flal>e ^ Soapstone vessel sfierd f& Tip-fluting spall + Utilized nake NUKASUSUTOK-1 2: EARLY/MIDDLE DORSET AXIAL STRUCTURES 193 Table 45. Nukasusutok-1 2: Structure 2 Tool Classes in Relation to the Bilateral Axial Feature Model. iviia tdSl west Prov. N (%) No Prov. TOTAL N (%) Triangular endblades z ")Z 4 (7 3^ 4 Endblade preforms 3 (5.5) 3 (3.9) Side-notched bifaces 11 1 (1.8) 1 (1.3) Slate bifaces 11 1 (1.8) 1 (1.3) Ramah chert bifaces 1 1 (1.8) /I 5 (6.6) Tip-fluting spalls 11 -)c 3 (5.5) D 6 (7.9) M icroblades 1 3 1 7 D 36 (65.5) 1 D1 D 49 (64.5) utiiizeo TiaKes 11 1 (1.8) 1 2 (2.6) Slate flakes 1 1 (1.8) 1 (1.3) Nephrite flakes 1 1 (1.8) 1 (1.3) Soapstone vessels 3 3 (5.5) 3 (3.9) TOTAL 22 24 9 55 21 76 Table 46. Nukasusutok-1 2: Structure 2 Tool Classes by Front/Back Divisions. Front Back Middle TOTAL N (%) Triangular endblades 4 4 (7.3) Endblade preforms 3 3 (5.5) Side-notched bifaces 1 1 (1.8) Slate bifaces 1 1 (1.8) Ramah chert bifaces 1 1 (1.8) Tip-fluting spalls 1 2 3 (5.5) Microblades 20 16 36 (65.5) Utilized flakes 1 1 (1.8) Slate flakes 1 1 (1.8) Nephrite flakes 1 1 (1.8) Soapstone vessels 2 1 3 (5.5) TOTAL 31 23 1 55 endblade preforms to the eastern flank and soap- stone vessels to the axial feature. A division between front (north), back (south) and middle subareas in Table 46 indicates that the only clear difference is the presence of all endblades and endblade preforms in the front subarea. In neither case are there indica- tions of gender-based activity variation. /(-means cluster analysis was used for a model independent analysis of the point plotted tools; a five cluster solution seemed most reasonable (Figure 129). Two of the clusters (1 and 3) have centroids closely associated with the hearths at opposite ends of the axial feature in what might be termed hearth- tending positions. Cluster 2 lies slightly back from 194 CHAPTER 1 Table 47. Nukasusutok-1 2: Structure 2 Tool Classes by K-Means Cluster. C-1 C-2 C-3 C-4 C 5 TOTAL Endblades 3 1 4 Endblade preforms 2 1 3 Side-notched bifaces 1 1 Ramah chert bifaces 1 1 Slate Bifaces 1 1 Tip-fluting spalls 1 2 3 Microblades 7 12 4 8 5 36 Utilized flakes 1 1 Slate flakes 1 1 Nephrite flakes 1 1 Soapstone vessels 1 1 1 3 TOTAL 14 1 5 8 10 8 55 7x10y lOxlOy 5x9y Sx4y RCW8_ S=1 GM= 2 RC= 3 RC= 18 CR= 1 _ o RC=4 Q= 1 N= 1 ^0 12x8y CR Crystal quartz GM Green Mugford chert N Nephrite O Quartz RC Ramah chert S Slate 9x4y 1 30/ Nukasusutok-1 2: Structure 2 flake distribution. the northernmost hearth, but seems positioned in relation to that feature. Clusters 3 and 4 lie in similar positions on opposite sides of the middle of the axial feature where there was uncertain evidence for a third hearth. Each of these clusters is located equidis- tant from, and thus equally accessible to, the hearths at the ends of the axial align- ment. From a different per- spective, clusters 3 and 4 could be arm's length discard areas associated with the hearth- tending positions at each end of the axial alignment. Overall, there is a hint of behavioral symmetry at the two hearths. The tool associations for each cluster are compared in Table 47. Given the low fre- quencies in each cell the distri- bution is difficult to evaluate, but there is a slight over-repre- sentation of microblades in cluster 2, adjacent to the north- ern hearth. Inspection of individual tool type distribu- tions provides little to supplement the foregoing observations other than revealing a tendency for end- blades and endblade preforms to be located towards the south and tip-fluting spalls towards the north. In other words, deposition of the preforms and finished NUKASUSUTOK-1 2 EARLY/IVIIDDLE DORSET AXIAL STRUCTURES 195 tools is not associated witln deposition of the by-prod- ucts of tine manufacturing process. Tlie refits of a tip- fluting spall and a soapstone vessel are spatially prox- imate, providing little information on material move- ments at Structure 2. In contrast to the relatively large tool assemblage, only 55 flakes were found, almost exclusively small secondary retouch flakes of Ramah chert. Their distri- bution by 1 m^ units is displayed in Figure 1 30 (some flakes are missing because of uncertain provenience). The majority of flakes are distributed outside the axial feature, but within a 1 m distance from it. Their con- centrations correspond well with the clusters identified in the tool analysis. Tool and flake raw materials are outlined by fre- quency in Table 48 and by weight in Table 49. By fre- quency, Ramah chert dominates both tools and flakes. The weight figures for tools are distorted by soap- stone, but if the latter Is excluded then Ramah chert constitutes 86% of the flaked tool materials. The weight figures for flakes are misleading; what seems to be a dominance of quartz is mostly the result of a sin- gle quartz nodule that is heavier than the 48 small Ramah chert flakes. The single flake of white chert is fossiliferous with a carbonate cortex and is not from the Newfoundland west coast. It appears to be of northern Quebec or Eastern Arctic origin. Table 48. Nukasusutok-1 2: Structure 2 Tool and Flake Raw Material Frequencies. Tools N (%) Flakes N (%) Ramah chert 65 (87.8) 48 (87.3) Gray Mugford chert 1 (1.8) Crystal quartz 5 (6.8) 1 (1.8) Quartz 2 (3.6) Slate 1 (1.4) 1 (1.8) White chert 1 (1.8) Nephrite 1 (1.8) Soapstone 3 (4.1) Table 49. Nukasusutok-1 2: Structure 2 Tool and Flake Raw Material Weights. In Crams. Tools g(%) Flakes g (%> TOTAL g (%) Ramah chert 107.9 (23.2) 9.9 (34.1) 1 17.8 (23.8) Gray Mugford chert 0.1 (0.3) 0.1 (0.0) Crystal quartz 2.9 (0.6) 0.03 (0.1) 2.93 (0.6) Quartz 14.0 (48.2) 14.0 (2.8) Slate 14.6 (3.1) 0.5 (1.7) 15.1 (3.1) White chert n.a. Nephrite 4.5 (1 5.5) 4.5 (0.9) Soapstone 340.0 (73.1) 340.0 (68.8) Tool Assemblage Endblades All but one of the four Ramah chert end- blades are fragments. The single complete specimen (Figure 131:c) is small (L=19.6 mm, W= 14.2, TH= 2.9) with a very slight basal concavity and tiny tip-fluting spalls (L= 2.5, 3.5 mm) on the dorsal surface. A proxi- mal fragment (Figure 1 31 :d) is only slightly larger, with a straight base (W = 1 7.1 mm, TH = 3.1 ). The other two specimens are distal fragments of larger endblades (L = >28, >35 mm). One (Figure 1 31 :e) has two tip-fluting spalls (L = 14.0, 16.0 mm) on its dorsal surface while the other broke during removal of the first spall. Endblade Preforms The three preforms are made of Ramah chert. One is complete, triangular and straight-based (Figure 131:a), with fine retouch on its ventral side and large flake scars on its dorsal side (L = 35.5 mm, W = 24.9, TH = 5.5). The blunt tip was retouched ventrally to form a platform from which two tip-fluting spalls were removed dorsally (L = 1 6.8, 1 9.0 mm). The other two preforms are distal fragments. One (Figure 1 31 :b) has a distal platform from which a single flute was removed down the center of the dor- sal side (L= 26.8 mm), probably a production failure, while the other (Figure 1 31 :f) has a convex distal end with faint retouch for platform preparation. Side-notched bifaces: The single example (Figure 131:h) is the proximal fragment of a Ramah chert biface with broad side-notches positioned towards the 196 CHAPTER 10 1 31/ Nukasusutok-1 2: implements from Structure 2. a, b, f: endblade preforms: c-e: endblades, g: biface blank; h: side-notched biface: i: ground slate biface: j-o: micro- blades. medial portion of the tool such that the implement resembles a "box-based" biface. The basal edge is bifa- cially flaked and convex in form. There is slight grind- ing on the base as well as in one of the notches (W= 24.7 mm, TH= 5.7, notch HT= 12.0, 14.6, notch L= 7.1 , 7.5, notch W= 2.2, 2.0). Slate Bifaces: The distal fragment of a black slate biface is either the tip of a lance or a large knife (Figure 1 31 :i). Each face of the tool has a flat medial surface, but only one face is polished. The lateral edges are bevelled for a width of 6-9 mm on each face, but there is also fine secondary bevelling ca. 0.7 mm wide extending along the immediate working edge. Bifaces: All five of the bifaces are made of Ramah chert. One (Figure 131:g) is a complete asymmetric blank (L= 82.0 mm, W= 42.6, TH= 16.4). Two are dis- tal fragments, of which one convex-tipped implement might be an endblade preform while the other might be a tip-fluted endblade fragment. The remaining two specimens are lateral and proximal fragments. Tip-fluting spalls: All five tip-fluting spalls are made of Ramah chert and all are primary removal spalls. Three of them could be considered relatively large (>25 mm in length). Metrics for the three com- plete specimens: L= 16. 4, 27.1, 27.5 mm, W= 15. 6, 1 1 .2, 9.3, TH = 1 .3, 3.1 , 1 .8. Microblades: The raw materials of the 49 microb- lades (MNI = 24, based on proximal ends) are over- whelmingly Ramah chert (n = 44, 89.8%), with five examples made of crystal quartz (1 0.2%). Retouched or utilized microblades constitute 40.4% of the total. Of these, only three items exhibit traces of hafting: one crystal quartz specimen has bilateral notching (Figure 128:o), while another has a constricted stem, and one Ramah chert implement has a bilateral constriction. Two microblades may have been the result of core- trimming. Using only proximal portions and specimens from which reliable measurements could be taken (n=l 5), the average width of the Ramah chert microb- lades was 1 2.3 mm. Platform angles were measured to the nearest 5", resulting in a range from 75-90 (n=23), with a distribution across the four angle classes as fol- lows: 75 = 4 (1 7.4%), 80 = 1 2 (52.2%), 85 = 1 (4.3%), 90" = 5 (26.1%). Of the 22 proximal specimens upon which platform preparation technique could be identi- fied, 17 (77.3%) had plain platforms, four (18.2%) retouched platforms, and one (4.5%) had a crushed platform. The size of microblade fragments may be the result of random breakage patterns, but in some cases microblades might have been snapped deliber- ately for use in particular tasks. For the Ramah chert specimens, length measurements of proximal frag- ments (n=22) cluster into two groups, ca. 1116 mm and 21-31 mm, while the medial fragments (n=l 3) exhibit a marked clustering between 14-20 mm. The distal fragments are relatively few (n=7) and exhibit a wide range of values. These groupings may simply result from the mechanics of tool breakage, but they NUKASUSUTOK 12 EARLY/MIDDLE DORSET AXIAL STRUCTURES 197 seem consistent enough that deliberate action should also be considered. Utilized flakes: The two examples are both made of Ramah chert. Soapstone vessels: Three vessels were identified from a total of 5 sherds. One vessel is '/< complete (Figure 1 32), shallow (depth= 31 .5 mm), with one long side slightly curved and the other markedly bowed out so as to produce an asymmetric form with a length over 103 mm and a width ca. 66 mm. Wall thickness varies from 2.7 mm at the rim to 7.1 mm near the base. The rim is flattened with slightly rounded edges. The interior bottom exhibits striations and gouge marks indicative of an attempt to thin the vessel, a process that produced a small hole in one corner of the vessel bottom. The vessel weighs 71 g. A second ves- sel (Figure 133:a) is represented by a wall fragment with a small rim portion (TH= 3.1-5.7 mm). The third specimen is a wall sherd (TH= 3.3-5.7 mm) from a fair- ly deep vessel (>70 mm) featuring a flattened rim with round edges and an encrustation of burned fat on both the mner and outer surfaces (Figure 1 33:b). Discussion Structure-2 is a double or perhaps triple-hearth axial feature. The distribution of lithic material and rocks indicate the feature was associated with a rela- tively large tent (ca. 5m long by 5-5 m wide). The lith- ic materials were concentrated within 1 .0-1 .5 m of the axial feature, suggesting sleeping areas further out towards the tent walls. Cluster analysis of the tools suggests a degree of behavioral symmetry in the use of space at each end of the axial feature, with hearth- related activities or deposition, although most of the lithics occur on the west side of the feature. A slight front/back distinction noted in the clustering of all endblades and endblade preforms at the front (north end) of the axial feature is the only hint of spatial dif- ferentiation by gender, if one invokes the stereotype of male association with harpoon technology. This weak inference need not contradict the possibility that the 1 32/ Nukasusutok-1 2: soapstone vessel from Structure 2. 133/ Nukasusutok-1 2: soapstone vessel fragments from Structure 2. behavioral symmetry between the hearths, in combina- tion with the assumed size of the dwelling, might indi- cate the presence of two domestic units within the structure. Seen in terms of what did and did not "fall out" of the technological organization system, the tool and debitage samples suggest a relatively short duration of occupation and a limited range of activities. Most prominent among what fell out of the system were: (1 ) a substantial rate of microblade production and (2) a modest deposition of endblades, endblade preforms and tip-fluting spalls. The former may be related to various expedient tasks while the latter indicates 198 CHAPTER 1 a b c 1 34/ Nukasusutok-1 2 implements from Structures 3 and 4. a: side-notched biface, Structure 3, b: endblade, Structure 3; c: burin-like-tool preform, Structure 3; d, e: side-notched bifaces, Structure 4, f: endblade preform, Structure 4; g: microblade, Structure 4. preparation for and/or implementation of sea mammal hunting activities. What did not fall out of the system were flakes and domestic processing tools. The small quantity of Ramah chert debitage and its near-exclu- sive secondary retouch flake character suggests limit- ed modification of curated implements such as the endblade preforms and the biface blank. The scarcity or absence of domestic processing tools such as bifa- cial knives, scrapers and burin-like tools indicates that either activities associated with these implements were not undertaken or that such items were less likely to enter into the archaeological record at this locality. STRUCTURE 3 A 1 m^ test pit 25 m south of Structure 1 (Figure 1 24) revealed the probable remains of a third axial feature consisting of cobbles and flat slabs. The test pit and later surface collection in the area produced nine tools: a tip-fluted endblade, a tip-fluting spall, a side-notched biface, a burin-like tool preform and four microblades. A total of 17 flakes was collected, of which 15 were Ramah chert, one crystal quartz and one schist. The Ramah chert endblade (Figure 1 34:b) is com- plete, with a fairly straight base and some medial basal thinning (L= 42.7 mm, W= 1 8.2, TH= 5.6). The tip-flut- ing was placed dorsally and the scars are relatively short (L= 13.0, 16.5 mm). The single side-notched biface of Ramah chert (Figure 134:a) is also complete and displays a slightly asymmetric blade, a biconvex cross-section, a straight bifacially thinned base and slight edge grinding above one of the notches (L= 57.6 mm, W= 30.5, TH= 5.8, notch HT= 9.2, 9.0, notch L= 5.4, 6.6, notch W= 2.2, 2.1). The burin-like tool preform (Figure 1 34:c) is com- plete and made on a long flake of nephrite. One later- al edge has a bifacially flaked wide notch or stem while the opposite edge bears a slight trace of a notch formed by dorsal retouch. The implement has a slight basal concavity and it is unground (L= 48.5 mm, W= 1 9.8, TH= 5.8, notch L= 1 3.4, notch W= 1 .9). All four microblades are made of Ramah chert; two are retouched distal fragments while the other two are uncertain proximal fragments. STRUCTURE 4 Ten meters north of Structure 1 , adjacent to a rock-fall beside the north hill-face, a 1.5 by 1.0 m test pit revealed part of a fourth axial feature (Figure 124). Several thick flat-topped rocks and a flat slab were associated with burned sand, charcoal, red ocher patches and a considerable quantity of Ramah chert debitage, some of which was thermally altered. Tools recovered from the test unit Include two side-notched biface bases, one primary tip-fluting spall, one tip-flut- ed endblade preform, three microblades and two uti- lized flakes. The debitage collection consists of 253 unaltered Ramah chert flakes (78.3 g by weight), 77 thermally altered Ramah chert flakes (29.5 g), one quartz flake (0.6 g), and one schist flake (0.4 g). NUKASUSUTOK-l 2: EARLY/MIDDLE DORSET AXIAL STRUCTURES 199 Table 50. Nukasusutok-1 2: Test Pits 1-3 Tool Class Frequencies. TPl TP2 TP3 Surface TOTAL Triangular endblades 1 z 3 (4.3) Side-notched bifaces -> 1 4 (5.8) Bifaces 1 1 2 (2.9) Burin-like tools 1 1 (1 .4) Tip-fluting spalls 1 z oO 1 12(1 7.4) Microblades, unutilized I 7 zU 29 (42.0) Microblades, utilized i. 1 -7 / 1 (1 4.5) Microblade cores 11 1 (1 .4) Utilized flakes 1 2 3 (4.3) Ground schist flakes 2 2 (2.9) Ground nephrite flakes 2 2 (2.9) TOTAL 7 14 46 2 69 Both the side-notched biface bases are made of Ramah chert. One (Figure 1 34;d) has a single pair of wide notches and a slightly concave base (W= 30.4 mm, TH= 8.2, basal depth= 3.0, notch HT= 6.0, notch L= 13.1, notch W= 3.0). The other (Figure 1 34:e) has two pairs of notches and a straight base (edge A notch HT= 6.0, 1 4.6 mm, edge B notch HT= 5.9, 13.1, edge A notch L= 2.9, 4.2, edge B notch L= 3.6, edge A notch W= 1 .6, 1 .8, edge B notch W= 1 .6) The endblade preform (Figure 1 34:f) is made on a flake of thermally altered Ramah chert and appears to be a production failure (L= 55.3 mm, W= 22.9, TH= 5.8). It retains a striking platform proximally, has bifacial retouch limited to its margins and a single ventral tip- flute that encompasses the entire tip of the implement and terminates in a step fracture. The margins of the flut- ing scar were retouched slightly subsequent to spalling. All three microblades are made of Ramah chert and are medial fragments. One is retouched. A large specimen is illustrated in Figure 1 34:g. The two uti- lized flakes are both made of Ramah chert. Two radiocarbon dates were run from a sample of conifer charcoal associated with the structural rocks. The first assay produced a date of 930±64 B.P. (Sl- 5536). Since this result indicated a Late Dorset dating, which seemed inconsistent with the tool collection, the sample was resubmitted and subjected to nitra- tion pre-treatment to remove uncharred cellulose. The second date was 1110±80 B.P. (SI-5822), marginally older than the first and equally inconsistent with the tool collection. TEST PITS 1 3 Seven meters south of Structure 3 was a vegetated area that was probed with three 1 m^ test pits (Figure 1 24). Test Pits 2 and 3 adjoined each other while Test Pit 1 lay 1 m to the northwest. Test Pits 2 and 3 pro- duced a surprisingly dense concentration of Ramah chert debitage and tools in a matrix of brown sand and cobbles superimposed on a soft clay deposit lying 1 5 cm below ground surface. No features could be dis- cerned in the morass of cobbles and clay. The area might have contained a midden or structure that was badly disturbed by severe clay movement. Table 50 provides an overview of tool frequencies, distributed across the three test pits. The tool collection from these test units is marked by a large number of microblades (56.5%) and 200 CHAPTER 10 contains a microblade core fragment. Altliough trian- gular endblades are only modestly represented, a fair- ly large number of tip-fluting spalls are present, including eight primary removal spalls. The latter might indicate a focus on endblade production from curated preforms. Tool and flake raw materials are summarized by frequency in Table 51 and by weight in Table 52. Both tool and flake materials are dominat- ed by Ramah chert. Table 51. Nukasusutok-1 2: Test Pits 13 Tool and Flake Raw Materials by Frequency. Tools N (%) Flakes N (%) Ramah chert 62 (89.9) 21 1 (87.2) Crystal quartz 1 (1 .4) 1 (0.4) Cray Mugford chert 1 (0.4) Creen chert 1 (1 .4) Nephrite 3 (4.3) 2 (0.8) Schist 2 (2.9) 27(11.2) Table 52. Nukasusutok 1 2: Test Pits 13 Tool and Flake Raw Materials by Weight. In Grams. Tools g (%) Flakes g ( %) TOTAL g(%) Ramah chert 128.5(91.3) 107.3(76.0) 235.8 (83.7) Cp/stal quartz 0.6 (0.4) 0.5 (0.4) 1.1 (0.4) Cray Mugford chert 1.7(1.2) 1.7 (0.6) Green chert 2.8 (2.0) 2.8 (1.0) Nephrite 6.8(4.8) 11.7(8.3) 18.5 (6.6) Schist 2.0(1.4) 19.9(14.1) 21.9 (7.8) Tool Assemblage Endblades: All three triangular endblades are made of Ramah chert. A complete tip-fluted specimen (Figure 136:a) has a slightly concave base and appears to be fluted on its dorsal side (L= 21.1 mm, W= 14.6, TH= 3.2, basal depth= 1.0, tip-fluting L= 10.6, 14.0). A proximal fragment (Figure 136:b) is bifacially flaked with a slightly asymmetrical concave base; it is impos- sible to determine if it was tip-fluted (W= 13.1 mm, 3 5 135/ Nukasusutok-1 2 implements from test pits, surface and Structure 1 . a: biface, TP- 1 . b: endblade, TP- 1 , c: endblade preform, TP-4, d-e: utilized flakes, TP-4, f: ground slate point/knife base, surface; g: microblade, surface; h) microblade. Structure I . TH= 2.8, basal depth= 1.0). Another proximal frag- ment (Figure 1 35:b) is bifacially flaked (completely on the dorsal side, marginally on the ventral side) and has a straight to slightly convex base with slight basal thin- ning (W= 22.9 mm, TH= 3.5). Tip-fluting spalls; All of the 12 tip-fluting spalls are made of Ramah chert. The seven whole or near complete specimens range in length from 12.3-36.4 mm and two size groups are represented: 12.3-13.6 mm and 21.5-36.4 mm. Four of the five fragmentary specimens clearly fall into the larger size class. These large and small classes may indicate the tip-fluting of different sizes of preforms or endblades (cf., Plumet and Lebel 1 998). None of the intact finished endblades NUKASUSUTOK 1 2 EARLY/MIDDLE DORSET AXIAL STRUCTURES 201 a b c d 1 36/ Nukasusutok-1 2: implements from test pits, a, b: endblades, TP-3; c-e: side-notched bifaces, TP-3; f: burin-like-tool, TP-3: g: ground schist, TP-3; h: microblade core fragment, TP-2; i-l: microblades, all TP-3. from Nukasusutok-1 2 exhibit tip-flutes with a length corresponding to the larger group, although one pre- form from Structure 2 does. Eight of the spalls are pri- mary, four are secondary. Side-notched bifaces All four bifaces are proximal fragments and all exhibit a single set of notches. Three of the four are made from Ramah chert while the fourth is made from a green chert (either Mugford or Newfoundland). One of the Ramah chert implements (Figure 1 36:c) is from a fairly large biface and displays wide side-notches and a slightly concave bifacially thinned base (W = 33.4 mm, TH = 6.9, notch HT = 5.3, 9.5, notch L = 13.0, 12.6, notch W = 3.0, 3.0, basal depth = 2.0). A second Ramah chert example (Figure 136:e) lacks one basal corner and has a biconvex cross-section and a slightly concave base with basal thinning (W = 20.3 mm, TH = 4.4, notch HT = 7.3, notch L = 5.4, notch W = 1.2). The third Ramah chert specimen was missing when the collection was re- analysed in the early 1990s but it is illustrated else- where (Hood 1 986:Figure 5:1). The implement is small- er and narrower than the other bifaces and exhibits predominantly unifacial dorsal retouch and a straight base (base W = 1 3.8 mm, TH = 2.8, notch HT = 3.2, 3.4). It roughly resembles a Croswater "box-based" endblade, although it has a bifacially thinned base in contrast to the abrupt unifacially retouched bases of the latter. The green chert specimen (Figure 1 36:d) has a straight bifacially thinned base (base W = 30.0 mm, TH = 6.2, notch HT = 5.4, 5.1). Bifaces Both examples are made of Ramah chert. One is the distal portion of a large implement over 78 mm long and at least 37 mm wide, presumably a large knife or lance point (Figure 1 35:a). The other is a small distal fragment. Burin-like tools: The single specimen of nephrite is complete but for a portion of its proximal end (Figure 1 36:f). Polishing traces on the proximal edges suggest a bilateral slightly constricted stem with a unilateral notch just below the bevelled portion of one lateral edge. The distal end is straight and symmetri- cally bevelled to a cutting edge. The distal lateral edges are ground flat and the dorsal and ventral sur- faces are flat and fully ground and polished (L = > 29.0 mm, W max = 16.3, W distal = 12.7, TH = 4.3, notch HT = > 16.0). Microblades: All but one of the 39 microblades are made of Ramah chert; the single exception is crys- tal quartz (a few examples are illustrated in Figure 1 36:i-l). All specimens are fragmentary: 1 8 proximal, 1 5 medial and six distal. Traces of utilization or retouch are found on 10 (25.6%). An MNI count of 17 can be formulated for the non-utilized specimens and an MNI of two for the utilized implements. One of the 202 CHAPTER 1 utilized specimens has proximal retouch (unilateral constriction) that may indicate hafting. In another case, a medial fragment has bifacial retouch on one of its snap surfaces that could indicate preliminary retouch for a projectile base. Additionally, a distal fragment is completely retouched on its ventral sur- face, which may also indicate an intended use as a projectile. The average width of 1 5 proximal frag- ments is 12.5 mm. Platform preparation is primarily plain (n= 6) or retouched (n= 5), while one example each occurs for grinding, crushing and indeterminate. Platform angles were distributed as follows; 60° = 1 , 65° = 3, 70' = 1 , 75° = 2, 80° = 5, 85 = 1 and 90° =1 . A major propor- tion of the proximal fragments (1 of 1 7) have lengths in the range of ca. 13-28 mm; the medial fragment lengths exhibit greater spread, but eight of 1 7 range between ca. 1 2-25 mm. This is broadly consistent with the pattern observed at Structure 2. Microblade cores: The single Ramah chert microblade core (Figure 136:h) is a core-face frag- ment that bears negative scars of five long (ca. 75 mm) microblades. Three of these microblades were struck from a core platform that was broken off when the core face was detached. This detachment occurred when the core was turned over and struck on a flat basal surface used as a platform for the attempted removal of microblades from the opposite direction. Secondary retouch on the edges of the core fragment suggest it may have been re-used as an expedient scraper. Utilized flakes: All three specimens are made on large (52-62 mm diameter) flakes of Ramah chert. In one case virtually all edges of the flake exhibit fine utilization retouch combined with blunting of exposed facets; the implement may be an expedient scraper. Ground schist flakes: The two specimens are both small (ca. 17 and 28 mm long); one is illustrated in Figure 1 36:g. Schist was likely used to grind nephrite burin-like tools and perhaps slate implements. Ground nephrite flakes: Two of the three flakes are small (L= 1 5.0, 24.6 mm) with limited areas of dis- tinct grinding. A third specimen is a larger linear flake (L= 72.9 mm) with faint traces of polishing on its facets but no obvious grinding. Discussion These three test pits provide little indication as to the function of this part of the site. Any features once present were obliterated by fluctuating movements of the underlying clay deposit. Nonetheless, the large quantity of tools and flakes recovered from these units suggests the presence of a dwelling structure or mid- den. The most striking aspect of the assemblage is the high frequency of microblades and tip-fluting spalls. The former suggest a high rate of expedient tool pro- duction while the latter suggest an emphasis on end- blade production— a "gearing-up" for hunting. The lack of scrapers is notable. TEST PITS 4 AND 5 Two other 1 m^ test pits were excavated within in situ deposits 1 m east of Structure 1 . Test Pit 5 was ster- ile but the central portion of Test Pit 4 contained part of a shallow depression, measuring 45 by 60 cm, which contained red ocher stains and charcoal. A small notched schist slab with adhering burned fat was recovered from the deposit. A second small charcoal patch lay beside the north wall of the test unit. Charcoal from the central depression was submitted for radiocarbon dating. A first run produced a date of 1 000±75 B.P. (SI-5535), which seemed much too late in relation to the tool material from the site. The sample was re-run following nitration pre-treatment, resulting in a date of 1 660±90 B.P. (SI-5821). Test Pit 4 produced 10 tools, all of Ramah chert; an endblade preform, a tip-fluting spall, six microb- lades, and two utilized flakes. The endblade preform (Figure 135:c) is made on a crudely retouched Ramah chert flake and is only retouched marginally— bifacial- ly at the tip, mostly dorsally on the lateral edges— NUKASUSUTOK-1 2: EARLY/MIDDLE DORSET AXIAL STRUCTURES 203 while the base is thinned dorsally from a snap edge (L= 36.4 mm, W= 19.3, TH= 4.0). The tip-fluting spall is a large medial fragment (L >30.0 mm) of a primary spall. The six microblades are comprised of four medial and two proximal fragments. Two medial fragments are completely retouched on their ventral surfaces, possi- bly for use as projectiles. One of the utilized flakes has unifacial dorsal retouch along all its margins and a slight stem at one end (Figure 135:e), while the other has unifacial dorsal retouch along two lateral edges (Figure 135:d). A total of 50 flakes was collected, of which 47 (94.0%) were of Ramah chert, one (2.0%) of crystal quartz, one (2.0%) of black chert, and one (2.0%) of nephrite. SURFACE COLLECTIONS Two implements not directly associated with struc- tures were surface collected. A complete retouched microblade (L= 47.9 mm, W= 13.1, TH= 3.7) lay 3 m west of the 1 xl y marker at Structure 1 (Figure 1 35:g). The proximal portion of a ground slate point or knife (Figure 1 35:f) was found 6 m south of the 5x5y mark- er at Structure 2. The tool is made of green slate and has a constricted stem formed by a pair of wide side- notches. The base is straight with narrow, symmetri- cally ground bevelling. Above each notch is a slight trace of edge bevelling with surfaces running in slight- ly different directions. The medial faces of the imple- ment are largely unground (TH= 4.9 mm, notch HT= 4.5, 3.5, notch L= 14.2, 1 5.5, notch W= 3.5, 2.5). OVERALL ASSEMBLAGE CHARACTERISTICS AND TECHNOLOGICAL ORGANIZATION Consideration of technological organization at Nukasusutok-1 2 is complicated by rather divergent tool and debitage sample sizes and differences in the representativeness of the samples. As far as tools are concerned, only Structure 2 and Test Pits 1-3 have sufficient numbers to work with. Although Structure 2 has very little debitage, the com- plete excavation of the feature constitutes a fairly rep- resentative sample. The material from Test Pits 1-3 has a reasonable number of tools and flakes, but is a lim- ited area sample from an uncertain context. The test unit at Structure 4 produced a few tools and a modest debitage collection, but the sample cannot be consid- ered representative. Given these limitations the discus- sion can only hope to point out some tentative dimen- sions of technological organization at the site. The most striking characteristic of the Nukasusutok-1 2 assemblage is the absence of pro- cessing tools common in other Dorset sites: endscrap- ers and sidescrapers. This absence implies that either the activities in which these tools were used (skin scraping and wood working) were not conducted at the site or that the implements were highly curated as part of a hafted tool kit. Instead, microblades are the pre- dominant implement type, suggesting a high frequen- cy of production for use in expedient tasks. The pro- portion of microblades with traces of utilization varies from 25% at Test Pits 1-3 to 40% at Structure 2. Structure 2 had a modest number of endblades, end- blade preforms and tip-fluting spalls, while Test Pits 1- 3 had a substantial number of tip-fluting spalls, includ- ing several primary removal spalls (but no endblade preforms). In both cases a focus on endblade produc- tion is suggested. As far as the debitage is concerned, Structure 2 was notable for a very small amount of flakes (and thus a high toolTlake ratio), while Structure 4 and Test Pits 1-3 had a relatively larger quantity of flakes suggestive of more emphasis on tool production and maintenance. Considered in isolation these observations are perhaps not that significant, but their relevance is clearer when placed in a comparative context with other Dorset lithic assemblages. The only available material for such a comparison is Nagle's (1 984, 1 986) analysis of Dorset Ramah chert use. Nagle concluded that distance-decay processes associated with "down- the-line" exchange from the Ramah chert sources had a significant effect on the nature of Dorset lithic 204 CHAPTER 10 assemblages, particularly with regard to tool size and frequency. Preforms for bifaces and endblades are both smaller and less frequent with distance from the source areas and debitage also decreases in size and frequency. But frequency and size may also respond to other variables. Microblade production can generate either a few or many implements in a single production sequence, so the frequency of microblades may be more closely linked to the situational factors behind individual production events than to the raw material supply. The same could be said for the frequency of tip-fluting spalls. Size differences in endblades may have more to do with intended functions— such as seal versus walrus hunting— than with raw material avail- ability (Nagle 1986:98). One of the questions Nagle (1984:295, 315-316, 434) addressed using Nain Middle Dorset material was whether winter semi-subterranean houses were "gear- ing-up" sites where tools were produced for use during other periods of the annual round while sites from other seasons were marked by strategies of tool main- tenance and conservation (i.e., curation behavior). He concluded that the evidence supported his general line of reasoning, but that inter-assemblage differences were not substantial: While sites occupied in the late fall and winter may have been the focus of the majority of flaked stone manufacturing activities in preparation for the ensuing year, it appears that for the most part Dorset knappers were routinely making many new tools throughout their entire seasonal round. Yearly lithic needs were evidently planned carefully, and sup- plies budgeted reasonably well (Nagle 1 984:434). How does the material from Nukasusutok- 1 2 fit into this picture? Table 53 compares the tool assem- blages from Structure 2 and Test Pits 1-3 at Nukasusutok-1 2 with five of the Nain region Dorset assemblages analysed by Nagle (1 984:Appendix A). Koliktalik-1 (highlighted in the table) consists of two semi-subterranean winter houses (Cox 1978:107, 110-111; Fitzhugh 1976:138-140; Nagle 1984:217- 228; Spiess 1978), St. John's lsland-3, L-4, is an outer island mid-winter hunting camp containing a tent ring with an axial pavement (Nagle 1984:228-230; Spiess 1 978:55), Black Island ! A has a possible tent struc- ture (Nagle 1984:230-232), and Dog Bight L-3 lacked a discernable structure but is assumed to be a tent site (Fitzhugh 1976:138; Nagle 1984:209-210). Dog Bight L-3 is Early Dorset, while the rest are Middle Dorset. The proportions of tool types at Nukasusutok- 12 are most similar to Dog Bight L-3, with high per- centages of microblades and tip-fluting spalls, except that endscrapers and burin-like tools are better repre- sented at Dog Bight. The other two tent-dwelling sites have very different profiles. St John's lsland-3 has high proportions of endblades, endscrapers and burin-like tools, but relatively few microblades. This emphasis on sea mammal hunting and processing activities is consistent with the location of the site near the ice- edge and traces of walrus hunting in the small faunal assemblage (Spiess 1978:55). Black Island-IA is dom- inated by tip-fluting spalls, suggesting final-stage "gearing-up" for hunting. The variation in these tent dwelling assemblages probably reflects differences in the situational variables affecting lithic use in each case. In contrast, the winter house assemblages from Koliktalik are heavily loaded with processing/mainte- nance tools such as bifaces, endscrapers and burin- like tools. Given the frequency of endblade preform production at Koliktalik and their scarcity at all the tent dwelling sites, it seems likely there was a spatial staging in production, final retouch and use in a chaine operatoire marked by a high degree of cura- tion. But it should be kept in mind that there are sig- nificant formational differences between the tent and house assemblages, since the latter are accumulations resulting from multiple occupation phases ("coarse- NUKASUSUTOK 1 2 EARLY/MIDDLE DORSET AXIAL STRUCTURES 205 Table 53. Comparison of Main Region Dorset Lithic Assemblages. Semi-Subterranean Houses Highlighted in Gray. Nukasusutok-1 2 Nukasusutok-1 2 Dog Bight St. John's lsl.-3 Black IsL-la Koliktalik-1 Koiiktalik-1 S-2 TP 1-3 L-3 L-4 H-1 H-2 Bifaces 6 (8.7) D \y.o) 11 '?\ 1 U VO. 334 \L£..Z>) 1 7 tr ri 7\1 / J Biface Preforms 1 (1.4) 1 \j \\J. 1 ) Endblades 4 (5.8) 11 ?\ ^.Hj \ 1 D.^J or HA Q^ Endblade Preforms 3 (4.3) -7A {A Q\ Z 1 (3./) Tip-Fluting Spalls 6 (8.7) 1 2 (1 9.7) 45 (1 3.0) 15 (12.7) 33 (55.9) 252 (16.8) 43 (7.5) Endscrapers 4 (1.2) 1 1 (9.3) 3 (5.1) 66 (4.4) 46 (8.1) Microblades 49 (71 .0) 39 (63.9) 266 (76.7) 41 (34.7) 1 1 (18.6) 419 (28.0) 165 (28.9) Burin-Like Tools 1 (1.6) 8 (2.3) 9 (7.6) 98 (6.5) 35 (6.1) TOTAL 69 61 347 118 59 1498 570 grained" assemblages in Binford's [1980:17] sense) and the former are probably the result of more dis- crete events ("fine-grained" assemblages). Organiza- tional dynamics are difficult to infer from coarse- grained assemblages. Another dimension of variability is debitage characteristics. Table 54 outlines the Ramah chert toohflake ratio for the Nain area sites (derived from Nagle 1984; Appendix A). Although the Nukasusutok test pits cannot be considered a repre- sentative sample, their ratio is consistent with the other Dorset sites, whether tent sites or the semi- subterranean houses at Koliktalik. Structure 2, on the other hand, stands out for its unusually high ratio of tools to flakes. Compared with the other localities. Structure 2 reflects an extremely low rate of lithic reduction combined with a high rate of tool discard. This ratio suggests little emphasis on tool replenishment or "gearing up" strategies, presum- ably related either to an anticipated ease of tool replacement or the absence of domestic processing and maintenance activities which might place high- er demands on raw material use. The organization of lithic reduction may also be reflected in flake size distributions and the frequency of biface-thinning flakes. Although the Nukasusutok- 12 assemblages are not ideal for size analysis given small samples (Structure 2) and samples of uncertain representativeness (test pits), size distribution data for Structure 2 (completely excavated). Structure 4 (test-pitted) and Test Pits 1-3 are presented in Figure 137. Table 55 displays the frequency of biface-thin- ning flakes. Structure 2, which had only 48 Ramah chert flakes, is distinguished by a high proportion of flakes in the <1 5 mm and <1 mm categories and low frequencies in the <20 mm and larger categories, sug- gesting emphasis on late stage retouch rather than multi-stage reduction. The low frequency of biface- thinning flakes suggests limited reduction of pre- forms. Structure 4 and Test Pits 1-3 have slightly high- er percentages of flakes in the <20 to <30 mm range and both have relatively large numbers of biface-thin- ning flakes, indicating greater reduction of preforms than at Structure 2. Although the sampling problems need to be kept in mind, these data suggest spatial differences in reduction patterns at Nukasusutok-1 2. 206 CHAPTER 1 Nukasusutok-12 Ramah Chert Flake Size Distribution -S2 \ -S4 -TP1-3 <5 <10 <15 <20 <25 <30 5 mm classes <35 <40 <45 13// Nukasusutok-1 2: Ramah chert flake size distribution. Table 54. Ramah Chert ToohFlake Ratios for Dorset Sites in the Nain Area Tools Flakes ToohFlake Ratio Nukasusutok-1 2 S-2 63 38 1.66 Nukasusutok-12 TPl-3 62 210 0.3 Dog Bight L-3 339 1021 0.33 St. John's lsland-3 L-4 127 669 0.19 Black Island 1 A 51 479 0.1 1 Koliktalik-1 H-1 2536 16281 0.16 Koliktalik-1 H-2 1038 7712 0.1 3 Table 55. Distribution of Biface-Thinning Flakes at Nukasusutok-12. Biface-Thinning Flakes Structure 2 4 Structure 4 30 Test Pits 1-3 53 The Ramah chert debitage size distribution has greater value when compared with other Dorset sites in the Nain region. Comparison necessitated re-sorting the Nukasusutok-12 debitage in accordance with the size category system used by Nagle (1984:261-262): category 4 (>% < '/j inch), category 3 (>'/: inch, < 1 inch), category 2 (>1 inch, < 2 inches). Nagle eliminated a fifth category (<'/i inch) as unreliable, so the same was done for Nukasusu- tok-12. Four of the previously discussed Dorset components were used in the comparison; Koliktalik-1 houses 1 and 2, St. John's lsland-3, L-4, and Dog Bight L3 (data from Nagle 1 984: Appendix B). In the comparison by frequency (Figure 1 38), Structure 2 at Nukasusutok-12 has a profile similar to all the other Dorset sites, being domi- nated by category 4 debitage (>% inch <'/j inch). In contrast, Structure 4 and Test Pits 1 -3 are distinctly different from all the other components in their high proportions of larger category 3 flakes (>'/j inch, < 1 inch). When the proportions from the three Nukasusutok-12 components are averaged, the site has a deviant profile with a high proportion of class 3 flakes. In the comparison by weight (Figure 1 39), Structure 2's profile of balanced classes 4 and 3 most closely resem- bles Koliktalik House 2, while Structure 4 and Test Pits 1-3's profiles with high class 3 proportions are most similar to St. John's Island. The averaged Nukasusutok profile is also very close to St. John's Island. Although the intra-site variation at Nukasusutok-12 should be kept in mind, it is interesting that its greatest similari- ties are with St. John's Island, since both sites are sea- sonal hunting camps associated with axial structures. NUKASUSUTOK-12: EARtY/MIDDLE DORSET AXIAL STRUCTURES 207 Ramah chert size distribution by frequency, Nagie categories and Nain Region Sites 100 90 - 80 70 60 % 50 40 30 20 10 tL I Cat4 Cats m Cat2 S2 S4 TP1-3 KKH1 KKH2 DBL3 Component/Site StJI N12av 1 38/ Nain region Dorset site Ramah chert size distributions by frequency, using Nagle's (1984) categories. The preceding discussion positions Nukasusutok- 12 relative to other Nain area Dorset assemblages, drawing out some general similarities and differ- ences, but it is confined by the vagaries of the Nukasusutok sample and by the limited comparative base. Nonetheless, the Nukasusutok- 1 2 assemblage might be characterized as representing a hunting camp marked by a high production of expedient microblades, the final stage reduction of harpoon endblades, as indicated by tip-fluting spalls and end- blade preforms, and the reduction of biface pre- forms, as implied by the biface-thinning flakes. The size/weight distribution of the Ramah chert debitage suggests that Nukasusutok-1 2 reduction patterns are most similar to another outer island hunting camp, St. John's lsland-3, L-4, although the tool assemblages of the two sites differ in that Nukasusutok-1 2 has a higher rate of microblade production but a lower rate of endblade discard. If, as Nagle (1 986:98) observed, variation in microb- lade frequency is conditioned by highly contingent individual production decisions, then it may be that the contrasts in endblade discard are the only significant differences between the sites. To conclude, partly follow- ing Nagle's (1984, 1986) obser- vations one might say the nature of any individual Dorset assemblage is a product of the articulation of three dimensions: (1) general constraints imposed by the structure of the Ramah chert delivery system on the form and quantity of raw materials circulating with- in different regions, (2) production decisions linked to specific situations, and (3) social and symbolic con- text. At this point we have a strong set of inferences concerning the first point, a good start on but still inadequate understanding of the second, and rather little penetration of the third. CHRONOLOGY Table 56 outlines the radiocarbon dates from Structure 4 and Test Pit 4. As noted previously, the first run Table 56. Radiocarbon Dates from Nukasusutok-1 2. Location First Run Second Run Lab Nos. 1 6 Calibration of Second Run (nitration pre-treatment) (OxCal 3.10) Structure 4 930±64 B.P. 1 1 10+80 B.P. SI-5536, 5822 AD 780-1020 Test Pit 4 1000±75 B.P. 1660±90 B.P. SI-5535, 5821 AD 250-540 (both samples of conifer charcoal) 208 CHAPTER 10 70 60 50 40 30 20 10 S2 dates did not correspond with what was expected on the basis of tool typology (2200-2000 B.P.), so the samples were run a second time with nitration pre- treatment to remove uncharred cellulose. Although this result- ed in a significantly earlier assay for the sample from Test Pit 4, the result for Structure 4 was little different. Regrettably, neither of these samples was associated with clearly diagnos- tic tools. The material from Structure 4 included two side- notched bifaces (one with a sin- gle pair of notches, the other multiple notched), three microblades, a tip-fluted point preform, a tip-fluting spall, and a utilized flake. Typologically, the bifaces seem inconsistent with the 1 1 10±80 B.P. Late Dorset dating. Test Pit 4 contained an endblade preform, a tip-fluting spall, a possible scraper fragment, six microblades (two of which were completely retouched on their ventral surfaces), a uti- lized flake and a schist fragment with adhering burned fat. The best that can be said is that this material is not inconsistent with a 1 660±90 B.P. Middle Dorset dating. Given the ambiguity of the radiocarbon dating the remainder of the chronological discussion must be based on typological comparisons. Since very little Labrador Dorset material has been reported in full, the discussion will be somewhat impressionistic. Typologically, Nukasusutok-1 2 appears to be inter- mediate between Early Dorset material such as Dog Bight L-3 from the Nain region, dated 2455±75 B.P. and 2400±70 B.P. (Cox 1978:99, illustrated in Fitzhugh 1976, Figure 12, m-s), and lluvektalik-1 from Okak Bay, dated 2845±60 B.P. on charcoal mixed with burned fat (Cox 1977:152-1 64, illustrated in Cox 1978:108), and Middle Dorset material from Koliktalik-1 in the Nain area (Cox 1 978:99; illustrated in Fitzhugh 1 976, Figure Ramah chert size distribution by weight, Nagle categories and Nain region sites £1 eacat4 Eacats S4 TP1-3 KKH1 KKH2 DBL3 Component/Site StJ I N12av 1 39/ Nain region Dorset site Ramah chert size distributions by weight, using Nagle's (1984) categories. 1 2 a-l, and Cox 1 978: 1 09). The Nukasustok-1 2 materi- al bears some resemblance to the few tools illustrated from Avayalik-7 in northernmost Labrador, originally believed to date to the Early/Middle Dorset boundary, ca. 2200-1800 B.P. (R. Jordan 1980:609-610), but which produced a later than expected date of 1 395±70 B.P. (Morlan 2001-2005). Several Nukasustok-1 2 assemblage attributes reflect this Early/Middle Dorset character. First, there is a high frequency of microblades, which is regarded as typical of Early Dorset sites (Cox 1978:107). In a correspondence analysis (not shown here) of Labrador Dorset assemblages based on data from Nagle (1984; Appendix A), Nukasusutok-1 2 was grouped together with all the Early Dorset sites on the basis of their shared high frequency of microblades. As observed above, however, microblade production was probably situationally conditioned such that one should be wary of using it as a chrono-typological indicator. Second, the triangular endblades are straight-based or only slightly concave and four of them (from Structure 2, Structure 3 and Test Pits 1 -3) appear to be tip-fluted on their dorsal surfaces, as are two endblade preforms (Structure 2). Dorsal tip-fluting is thought to be an NUKASUSUTOK-1 2: EARLY/MIDDLE DORSET AXIAL STRUCTURES 209 early Dorset trait. There is no trace of unifacial end- blades or basal thinning on the dorsal surface of end- blades as found in IVliddle Dorset (Cox 1978:107). Third, a high side-notched or quasi "box-based" biface (Figure 131:h) from Structure 2 and a small unifacial notched biface resembling Croswater style (Hood 1 986:Figure 5:1) from Test Pits 1 -3 have an Early Dorset stamp, although the latter may simply be an expedient tool (cf., Nagle 1 984:347-348). Fourth, a burin-like tool preform from Structure 3 that was destined to be rela- tively long and have broad side-notching (Figure 1 34:c) seems more similar to the Early Dorset forms from lluvektalik-1 (Cox 1978:108) than the narrow-notched Middle Dorset forms from Koliktalik-1 (Cox 1 978:1 09). A fully ground tabular nephrite burin-like tool with a "squared" working end from Test Pit 3, however, exhibits the remnants of a proximal stem and traces of a small notch on a lateral margin, similar to Middle Dorset forms (Figure 1 36:0- The single and multiple side-notched bifaces from Nukasusutok-1 2 could be found in either Early or Middle Dorset contexts. To conclude, on typological grounds we might propose that Structure 2, Test Pits 1-3 and probably Structure 3 are Early/Middle Dorset, perhaps dating 2200-2000 B.P.. On the basis of the radiocarbon dates, unsupported by typological data. Structure 4 and Test Pit 4 may be Middle Dorset or later. STRUCTURAL COMPARISONS A major difficulty in providing a broader view of the Nukasusutok-1 2 structural remains is the lack of pub- lished accounts of Dorset structures in Labrador or else- where. Three semi-subterranean houses have been given some preliminary documentation: one of the two Middle Dorset houses from Koliktalik-1 in the Nain region (Fitzhugh 1976:130-131; Cox 1978:107-111), Iglusuaktalialuk lsland-4 West, a Middle Dorset house in the Okak region (Cox 1 977:1 37-1 38), and Okak-3, a Late Dorset house in the Okak region (Cox 1977:195-196, 1 978:1 10-111). Koliktalik-1 House 1 lacked an axial fea- ture, but there were a series of central pits associated with a partial pavement. The other two houses had well- defined stone axial features. Dorset surface structures are described as consist- ing of rectangular tent rings with central cobble or slab pavements lacking clear borders, or more ephemeral rock and pavement features (Cox 1978:11). The only published example is a tent ring from Avayalik-2 in northernmost Labrador, typologically dated to the Early/Middle Dorset "transition" (2200-1800 B.P.). This was a rectangular structure with a central boulder align- ment but without box hearths, slab pavements or verti- cal boundary rocks (R.Jordan 1980:612, Figure 4). One of the two radiocarbon dates from the structure is con- sistent with the typological dating (2000±75 B.P., 1 345±70 B.P.;). In the Nain area, tent ring features of broadly similar character are known from an Early Dorset component at the Dog Island West Spur site and the Middle Dorset St. John's lsland-3, L-4 site. The Nukasusutok-1 2 structures are clearly of a dif- ferent character than the previously mentioned surface structures in that they consist of long axial features with hearth areas and they lack distinct tent rings. These con- trasts may indicate functional or seasonal differences. The lack of ring rocks may indicate the tents were anchored in snow and perhaps used in the early spring or late fall. As discussed in the spatial analysis of Structure 2, the presence of hearth-related pairs of tool clusters at each end of the mid-passage hints at behav- ioral symmetry, possibly indicative of two domestic units within a large dwelling. This inference can be compared with R.Jordan's (1980:612) preliminary conclusion that the tent ring at Avayalik-2 was used by two family units. Structures 1 and 2 at Nukasusutok-1 2 bear a strong resemblance to each other. Although slightly different in size, each had a well defined axial feature, clear evidence for a hearth at each end of the passage, possible traces of a third central hearth, notched schist slabs and traces of large terminus rocks at one or both ends of the pas- sage. These structural and organizational similarities might indicate proximity in time, but the almost total lack of lithics at Structure 1 hinders chronological inferences. 210 CHAPTER 1 SETTLEMENT PATTERN AND SEASONALITY Drawing together the discussion so far, the Nukasusutok-1 2 lithic assemblage gives the impres- sion of a focus on sea mammal hunting with minimal evidence for a wider range of domestic activities. In the case of Structure 2, the high toohflake ratio indicates a high rate of tool discard with minimal tool production and maintenance, suggesting a social unit arrived with a limited tool kit prepared at another locality, perhaps a winter house, and functioned with low replacement requirements. The latter may not hold for the other features at the site, the sampling of which revealed larger quantities of debitage. The excavated axial fea- tures point to the use of large tents (perhaps family pairs?) and the presence of additional structures indi- cate the site was used repeatedly by small social units, probably around the Early to Middle Dorset "transi- tion," but possibly later as well. The potential seasonal use of the site can be con- sidered with reference to century Labrador Inuit settlement. According to Taylor (1974:51), Contact Period house sites were typically located with good access to fall harp seal hunting. There is an 18*^^ cgp. tury communal house complex just west of Nukasusutok-1 2 (Nukasusutok-8, Chapter 1 1 ) and east of Wyatt Harbour the narrow passage between a rocky islet and Nukasusutok Island is said to be a good place for intercepting harp seals. The area between Nukasusutok and Nochalik Island to the southwest has reasonably good open water sealing during the sum- mer. Mid-winter ice-edge sealing is not likely from the site since the usual position of the sina is off Humby's Island, over 20 km to the east (Figure 1 40). The area to the southeast between Satok Island and Humby's Island can be good for hunting seals basking on the ice in the spring. This area often has thin ice (visible as a darker shading in Figure 4), so open water areas may develop early in the spring. Given these considera- tions, the optimal periods for site occupation are dur- ing the harp seal migrations in the fall or the spring, basking seal hunting in the spring, or open water seal- ing during the summer. If we consider the Nain region Dorset site distribu- tion as a whole (Figure 140), it is clear that Dorset activity was concentrated in a 30 km wide band in the mid-outer island zone. Only a handful of sites are situ- ated at inner bay locations. Four of the five Middle Dorset semi-subterranean house sites^ are located within 7-8 km of the sina (as judged from Figure 4); Koliktalik-1 (Fitzhugh 1976:131, 140; Nagle 1984:217- 228; Spiess 1 978), Ford Harbour-4 (Fitzhugh 1 981 :37), jonathon lsland-3 (site files. Cultural Heritage Division) and Drawbucket Tickle-2 (site files. Cultural Heritage Division). Spiess' (1978) interpretation of the faunal remains from the houses at Koliktalik-1 suggests the site was occupied October to December to exploit the southward harp seal migration and from March to June for hunting basking seals and the northward migrating harps. Cox and Spiess (1 980:660-661 ) hypothesize the site was abandoned January through March and postu- late movement to camps on more seaward islands near the ice-edge due to a lack of breathing hole sealing techniques. An axial pavement structure at St. John's lsland-3, L-4, is proposed as such a camp, oriented towards open water seal and walrus hunting. A possible variation on this pattern may be inferred from the fifth semi-subterranean house site located 8 km east of Nukasusutok-1 2: No-Name Island- 2 (Sutton et al. 1981). This locality has two houses, with House 1 radiocarbon dated at 1285±75 B.P., 1260±60 B.P. and 915±100 B.P. and House 2 dated 1450±65 B.P. (Fitzhugh, personal communication); in other words late Middle Dorset to Late Dorset. ''Middle Dorset sod houses have also been registered at Skull Island (Fitzhugh 1981:37), but subsequent investigation led Susan Kaplan to suggest they may actually be Thule houses on top of a Dorset component (site files. Cultural Heritage Division). This locality is located ca. 2-3 km from the sina. NUKASUSUTOK 1 2 EARLY/MIDDLE DORSET AXIAL STRUCTURES 140/ Nain region Dorset site distribution. Original map © 2004 produced under licence from Her Majesty the Queen in Right of Canada, with permission of Natural Resources Canada. Excavations in a poorly preserved midden revealed the presence of walrus, but mid-winter hunting seems unlikely since the site is located 10-12 km distant from the sina. Given the tendency for thin ice and early open water in the area to the south between Satok and Humby's Island, it is possible walrus were hunted there in the first open leads of the late winter/early spring. They could also have been taken later in the open water period when the eastern end of the small island would serve as an excellent haul-out locale. With these winter sites as anchor points we could postulate that Dorset settlement patterns involved relatively short distance "lateral" (east-west) shifts on the scale of 1 5-30 km between fall-winter- spring semi-subterranean house sites located near the outer coastal fringe and spring-summer-fall tent dwellings at outer and mid-island locations, with similar or longer distance movements also possible on a north-south axis. Periodic shifts between alter- native winter house locations may have occurred pri- marily as "longitudinal" (north-south) movements. However, if the Nukasusutok-1 2 occupation, or at least part of it, pre-dated the documented Middle Dorset winter houses it could very well have been part of a different settlement pattern configuration than the one outlined here. 212 CHAPTER 10 Post-Contact Inuit Settlement on N u leasusutok Isia As noted in Chapter 4, there are several localities with traces of Post-Contact Period inuit activity on Nukasusutok Island. These were not central to the fieldwork conducted in 1 979-80 or 1 992-93, although the midden at the 18'^*^ century communal house site Nukasusutok-8 (HcCh-10) was tested. Subsequent to this testing, the Robert S. Peabody Museum at Phillips Academy, Andover, Massachusetts, was visited in order to inspect collections made on Nukasusutok Island in 1928 by William Duncan Strong, anthropolo- gist with the Rawson-MacMillan Subarctic expedition. Strong's field notes and photographs were consulted at the National Anthropological Archives, Smithsonian Institution, Washington D.C. Strong excavated some of the house structures at Nukasusutok-8 and collected skeletal material and arti- facts from graves at various places on the island. He never published this or other Inuit material he collect- ed between Nain and Hopedale, although the skeletal material was used in a physical anthropological study by T. D. Stewart (1939). The artifact material was first housed at the Field Museum in Chicago, but was even- tually transferred through an exchange agreement to the Peabody Museum, Andover. The value of the collec- tion is limited because Strong's cursory field notes lack critical provenience details, but the material is suffi- ciently interesting to justify an overview here, although the short time available for documentation did not permit a full analysis. Strong's material is supple- mented by a small test-pit collection derived from Nukasusutok-8 in 1992 and a brief survey of Nukasusutok-7 in 2004. Offsetting the problematic archaeological data is a relatively rich ethnohistoric record for the Nukasusutok-8 settlement. J. Garth Taylor's research on the Moravian missionary diaries uncovered refer- ences to the settlement between 1772-1794 and the Moravian Periodical Accounts (abbreviated as P. A.) contain passing references to later settlement on Nukasusutok Island. Some of the documentary infor- mation pertains to well-known Inuit individuals, so the usually anonymous archaeological context can be con- nected with the agency of real people and the social organization of the community can be specified. These ethnohistoric data are marshaled to construct a com- posite settlement history from the late 18^^^ to early 1 9*^*^ centuries, which is followed by a consideration of the archaeological data. But a brief sketch of post-con- tact Inuit society and settlement patterns is first pre- sented to set the scene. POST-CONTACT INUIT SOCIETY AND SETTLEMENT PATTERNS IN LABRADOR The Thule ancestors of the Labrador Inuit colonized northern Labrador by AD 1250-1 300 (Fitzhugh 1 994). At the lglosiatik-1 site in the Nain area, Thule-style sod houses with traces of metal apparently date to the early 16^^ century (Fitzhugh 1994:258; Kaplan 1983:216, 455-462). During the 16^^ century Inuit POST-CONTACT INUIT SETTLEMENT ON NUKASUSUTOK ISLAND 213 expanded southwards to Hamilton Inlet, the Strait of Belie Isle and the Quebec North Shore. Documentary information and archaeological data suggest the Inuit presence in southern Labrador was a year-round occu- pation that lasted until the mid-1 700s, when the increasing scale of European activity may have led to a contraction of Inuit occupancy towards Hamilton Inlet. During the 1 gth and ]7^^ centuries the southern Inuit were in contact with a variety of Europeans who con- ducted whaling and fishing in the area: Basques, French, English, Portugese, Spanish and Dutch (Martijn and Clermont 1980: Stopp 2002:75-76). Inuit both traded for and plundered European goods (Stopp 2002:83; Trudel 1980). Although some European wares made their way northwards along the Labrador coast through Inuit social networks, the degree of cul- ture change in the north was limited. Change accelerated during the 18'-'^ century. In 1 702 the French Sieur de Courtemanche was granted a concession in southern and central Labrador and he established a base of operations at Bradore on the Quebec North Shore. In 1 735 the Quebec trader Marsal received a concession for sealing and trading at Cape Charles and in 1737 Louis Fornel was one of three partners who acquired a similar concession at Chateau Bay. In 1 743 Fornel established a post for trading with the Innu at Northwest River in Hamilton Inlet; after his death in 1 745 this post and another at Rigolet were run by his wife until ca. 1 755. French rela- tions with the Inuit in southern Labrador were general- ly marked by considerable conflict (Kennedy 1995:20- 23; Stopp 2002:82-83, personal communication; Zimmerly 1 975:36-40). Britain acquired control of the Labrador coast in 1 763 after the end of the Seven Years War. British mer- chants overtook operations at Chateau Bay and a defensive blockhouse called Fort York was construct- ed there in 1 766. In 1 770 George Cartwright overtook the trading post at Cape Charles and founded addition- al posts, including his northernmost establishment at Sandwich Bay in 1 775. Cartwright ceased operations in 1 786. Other English merchants competed within the same area and by 1 784 Quebec traders were estab- lished in Hamilton Inlet (Kennedy 1995:25-41; Zimmerly 1975:42-44). During the 1760s and early 1 770s large numbers of Inuit from the central coast- perhaps mostly from Hamilton Inlet, but also parties from Hopedale and Nain— made trips to the English trading centers at Cape Charles and Chateau Bay. George Cartwright's diary indicates that large groups of Inuit did not appear south of Hamilton Inlet after 1 773 (Stopp, personal communication), but major Inuit traders were still making trips to Chateau Bay until ca. 1 790 (see below). The establishment of Moravian mis- sions in central and northern Labrador after 1 771 pro- vided northern Inuit with closer trading options, which became more advantageous in the late 1780s and 1 790s when Moravian trade policies were liberalized. During the course of the 1 7^^ and 1 8'^'^ centuries, Inuit tools traditionally made from stone were replaced by metals and a wide array of European goods was incorporated into Inuit life. Dwelling forms changed from the earlier small single-family sod houses to large multi-family communal houses (Kaplan 1983, 1985). The first Moravian missionary settlement was estab- lished at Main in 1771, followed by Okak (1776), Hopedale (1 782) and Hebron in 1 830 (and several oth- ers later in the 1 9^*^ century). Eventually, the previous- ly dispersed Inuit settlement became concentrated at the mission stations, where systematic Christianization and incorporation into European material culture systems occurred. But this process did not transpire rapidly. Moravian letters from the late 18^*^ century repeatedly express frustration at the dif- ficulties of acquiring and retaining converts and bemoan the negative influences of "heathens." A poignant note is struck by a comment in the Hopedale diary for 1 795 that "...their hearts be as hard as rocks and as cold as the ice of their country" (P.A. 1: 1795:351 ). Christianity was not consolidated until 1804-1805 when a "revival" spread through the mis- sion settlement areas (Hiller 1971:86). Nonetheless, 214 CHAPTER 1 1 "heathen" Inuit still dominated northernmost Labrador for many years thereafter (Loring 1998). Communal Houses and "Big-Men" Traders Much of the research on Post-Contact Inuit settlement has focused on the so-called "communal house phase" of the 18'^^ century. These semi-subterranean winter houses were rectangular, 7-16 m long and 6 m wide, with long entrance passages and generally three sleep- ing platforms (Kaplan 1983:238). Taylor's (1974:71) ethnohistorical data indicate the dwellings were occu- pied by an average of 20 people. The household was generally composed of several closely related families, often fathers and their married sons or brothers (Taylor 1 974:74-75) and polygyny was frequent (Taylor 1974:67). Winter settlements consisted of 1-8 houses (average= 2.4; calculated from Taylor 1974:71). Information on the social relationships between house- holds is more limited than for within households, but Taylor (1974:77-78) suggested kinship sometimes played a role, either links between brothers or uxorilo- cal ties. Certain males had leadership positions within the households by virtue of their kinship positions (father/son or father/son-in-law), but leadership authority at the settlement level was less marked and was tied to personal qualities, shamanist power and success in trading for European goods. There were few mechanisms for solving disputes above the household level and authority conflicts between brothers were common (Taylor 1974:80-84). The latter resonates in the translation of Nukasusutok (NaKasetjutok) as "the place where the brothers quarreled" (Wheeler 1 953:62) and in a story recounted by Manase Fox (1979) con- cerning conflict between brothers, one a leader at Nukasusutok. The shift from single-family to multi-family com- munal dwellings has been discussed from different perspectives. Schledermann (1976) offered an environ- mental determinist proposal: climatic cooling and the concomitant decline of whale hunting decreased resource availability, resulting in Inuit co-residential arrangements to facilitate sharing and fuel conserva- tion. An unpublished paper by Taylor (1 976) attributed the development to social aggregation around individ- uals prominent in trading baleen for European goods, a position adopted by R. Jordan (1978) and Kaplan (1983:351-359; also Jordan and Kaplan 1980) and clearly analogous to Sahlins' (1963) "big-man" model and the north Alaskan umealik, a man of wealth and boat owner (Spencer 1 959). Kaplan went further, sug- gesting these high profile individuals and their eco- nomic and political activities indicate "...that a form of hierarchical organization was in operation in 18'^'^ cen- tury Neo-Eskimo society" (Kaplan 1 983:352). Inuit mid- dlemen from the central coast acquired baleen, whale- bone, seal oil and furs from groups in the north and then traveled to southern Labrador trading centers such as Cape Charles and Chateau Bay. There they acquired European goods that later were circulated northwards, either down-the-line along existing social networks or by the traders transporting them north- wards themselves (Kaplan 1983:351-353). A well- known trader from the Main region, Tuglavina, facilitat- ed this activity by acquiring a two-masted sloop (P. A. 1 : 1 794:251). Kaplan (1 983:355-359) saw these processes in terms of the emergence of new "entrepreneurial" social roles, increased competition for resources and family rivalries. Multi-family households headed by boat own- ers constituted expanded production units and created social dependencies. Household heads could use these production units to generate the resources needed for participation in regional trade, which in turn provided the economic tools and symbols of status needed to retain followers and recruit multiple wives, who sup- plemented the household's labour power, served as prestige symbols and extended kinship-based eco- nomic networks. Shamanist power could also be used to reinforce social influence. The Jordan-Kaplan model was challenged by Richling (1993), who drew quite different conclusions from the connection between trade goods and multi- POST-CONTACT INUIT SETTLEMENT ON NUKASUSUTOK ISLAND 215 family households. Richling saw the 18'^'' century not as a period of radical shift towards hierarchy, but as a time when Inuit adjusted to changes by intensifying traditional egalitarian social practices. He questioned R. Jordan's (1978:184) assumption that trade goods were regarded as private property and thus not subject to sharing beyond the household. Instead, Richling (1993:74) suggested intensified communalism through sharing was a traditional Inuit means for deal- ing with economic scarcity and that access to scarce European goods would also have been handled through sharing relationships. Although Richling was probably right in questioning the relevance of private property concepts, his invocation of a blanket commu- nal social ethic as sufficient explanation for the devel- opment of multi-family houses leaves much of the dynamics of 18^'^ century Inuit society unaccounted for. It seems more productive to theorize household heads as active agents whose strategies drew on tradi- tional internal social resources such as personal char- acteristics, hunting skill, generosity and shamanist power, and who had varying degrees of ability to exploit the new external resources that became avail- able through engagement with the expanding periph- ery of the European world system. This leaves open the possibility for varied sets of local outcomes, some marked by a degree of social inequality, others marked by equality maintenance mechanisms. In order to put human faces on the abstract cate- gory of Inuit agents, a short historical narrative derived from Moravian missionary sources is presented in the next section. The late 18'^'^ century Moravian accounts provide an interesting glimpse of the prevailing social dynamics, albeit from a Moravian point of view. This story of two well-known Inuit highlights some of the processes outlined above. An Ethnohistoric Vignette: Tuglavina and Mikak One individual whose life intersected with the Nukasusutok settlement for a time was the "big-man" trader Tuglavina, frequently referred to in the late 1 8'^'' century Moravian accounts. His sometime wife Mikak also has a high profile in the historical records, although there is no indication that she resided on Nukasusutok with Tuglavina. The following is a brief biographical synopsis, culled from J. Garth Taylor's publications and the Moravian Periodical Accounts. It provides a sketch of important aspects of Inuit social and economic life during the late 18*^^ century, but it should be emphasized that neither individual can be taken as representative of a "typical" Inuk of the time. Mikak was the daughter of the Inuit leader Nerkingoak. After an initial 1765 encounter with Moravian missionaries on an exploratory trip, Mikak was captured in 1 767 near Cape Charles in southern Labrador. In 1 768 she was taken to England where she resided until her return to Labrador in 1769. She moved within London society and helped promote the Moravian cause; the mission eventually received a land grant in 1769. When the Moravian missionaries returned in 1 770 on another exploratory trip to find a location to establish a mission they met Mikak and her new husband Tuglavina near Byron Bay (north of Hamilton Inlet). Mikak clothed herself in a gold- trimmed gown given to her by the Dowager Princess of Wales, and a king's medal. The couple accompanied the Moravians northwards to Nain, Tuglavina serving as guide (Taylor 1979:740, 1984:18-19; Whiteley 1 979:536-537). One might speculate as to whether Tuglavina's forming a partnership with a woman who had such close ties with the English was a deliberate strategy on his part to facilitate access to European goods and acquire status. The Moravians regarded Tuglavina as "...a man of vast authority among his countrymen" (P. A. 3: 1799:435) "...and his word was absolute law" (P.A. 1: 1 794:251 ). His prominence was based partly on hunt- ing success, physical strength and cleverness, but also on the widespread belief that he possessed exception- al powers of sorcery that were given to him by Torngak (a spirit). If, after consulting with Torngak, he stated 216 CHAPTER 1 1 that someone should be killed, the declaration was implemented. Tugiavina himself allegedly killed sever- al people and his shamanist exhortations resulted in many additional killings by others. During the early years of the mission Tugiavina was a thorn in the side of the Moravians, "seducing" their converts (P. A. 1 : 1794:251). He and Mikak did not live at the mission (Whiteley 1 979:537); as noted below, we know that he was resident at the Nukasusutok settlement in 1 776- 77. Despite his activities, Tugiavina maintained a "sub- missive" tone in the vicinity of the missionaries (P.A. 1 1 794:25 1 ) and assisted them in various ways. In 1 775 he used a boat constructed for him by the Moravians to transport some missionaries scouting for a new mis- sion site to the Hopedale area. In 1 780 a missionary accompanied him into the interior on a caribou hunt- ing trip (Taylor 1969, 1979:740, 1984:21). Mikak and Tugiavina had a difficult relationship. Prior to Mikak's return from England Tugiavina had been married to Mikak's younger sister. Tugiavina took possession of the sister again in 1772, which alienated the group she had been with and displeased Mikak. Mikak was also displeased when Tugiavina tem- porarily exchanged her for another man's spouse. A feud was brewing with another group in the Nain region because of complaints that Mikak's family did not share the goods they had accumulated. In 1 774-75 Tugiavina and Mikak wintered in the interior, subsist- ing off stored caribou meat. In the fall of 1 775 the couple went to northern Labrador on a trading expedi- tion, collecting baleen at Nachvak, among other places. In the spring of 1 776 Tugiavina abandoned Mikak in favour of her two sisters. Mikak then lived with another man who lacked resources and social sta- tus (Taylor 1984:21-22). By 1782 Tugiavina had four wives (Taylor 1979:740, 1984:23). In 1 782 Tugiavina went to Chateau Bay in south- ern Labrador where he traded at European establish- ments. Mikak and her new husband followed. Tuglavina's southern trading continued until ca. 1 790. At the outset he acquired a two masted sloop in the south (P. A. 1: 1794:251), which was instrumental for his ability to gather baleen from Inuit in northern Labrador. The baleen was exchanged for European goods at Chateau Bay and the goods were then trans- ported to the north. Tugiavina also acquired firearms, which the Moravians had been unwilling to supply, and he solicited Inuit to accompany him on his trading trips, which did not please the missionaries (Taylor 1 979:740). On his return to Nain in 1 784 he appeared at the mission wearing a European officer's outfit con- sisting of a coat, wig, laced hat and sword (P.A. 1: 1794:251). In 1783 Tugiavina was baptized in the south and given the Christian name William, but he returned to Nain in 1784 "as sinful as ever," although he assisted the mission in various ways (P. A. 3: 1 799:435-437). Mikak lived mostly in the Hamilton Inlet area until 1 795 when she returned to the mission at Nain, where she died about the age of 55 (Whiteley 1 979:536-537). Tuglavina's life as a trader ended by 1 790 "...when his sloop became unseaworthy and he had only a single dog to pull his sled" (Taylor 1979:740). In 1790 he joined the mission settlement in Nain, now with only one wife. He was not accepted as a full member of the Nain congregation until 1 793, since he had difficulty conforming to Moravian expectations. In 1 793 the missionaries wrote that Tugiavina and Sikfigak's fami- lies departed the mission for other places "...where they spent the winter in many wicked ways. ..Tugiavina is often very uneasy in his conscience and promises to return, but cannot put his good intentions into execu- tion" (P.A. 1: 1793:21 5). But by 1795-96 Tugiavina began to "testify" to other Inuit and remained a convert despite "harassment" from "heathen" Inuit (P.A. 1: 1 795:354, 2:60). He died of pleurisy in 1 798 at ca. 60 years old, a few days after returning from a caribou hunt (P.A. 3: 1 799:328, 435). 18^*^ Century Subsistence-Settlement Systems The 18'^'^ century Inuit communal house economy must also be situated in terms of subsistence-settle- POST-CONTACT INUIT SETTLEMENT ON NUKASUSUTOK ISLAND 217 ment systems. The following is a short summary of central points. Semi-subterranean sod house settlements were occupied during the winter, beginning in mid-October when Inuit moved into their houses prior to the fall hunt for southward migrating harp seals (Taylor 1974:51). According to Taylor (1974:51), all the late 18'^'^ century winter house settlements in the Nain- Okak regions were located in areas with access to migrating harp seals. Harps and other seals were hunt- ed by kayak until the formation of sea-ice. Fall sealing was highly productive such that seals were first cached near the hunting sites then transported by sled to the settlements later in the winter (Taylor 1974:35, 51 ). Bowhead whales were hunted in November by umiak crews of 1 1 -1 5 men (Taylor 1 988: 123, 125). Moravian records from 1771-1784 indicate the most productive whaling areas (in descending order) were Hebron, Nachvak, Sagiek, Okak, and Hopedale, with a total of 63 whales reported taken during this time span. No whales were reported as captured in the Nain area, probably because of the unfavorable bathymetric con- ditions of the island-studded archipelago (Taylor 1988:125), but stranded drift whales were exploited when available (Taylor 1988:127). After freeze-up in mid-December, Inuit activities were centered on breathing hole sealing through new thin ice, especially along the sJna or at tidal "rattles." This was often undertaken by male hunting parties based in temporary snow-house camps. But January through April was a relatively low productivity period for sealing. Caribou meat taken on fall hunts in the interior might be retrieved from distant inland caches (Taylor 1 969:1 57, 1974:51-55). March and April were marked by sealing and walrus hunting at the sina and rock cod could be jigged through the sea-ice and char through the ice on inland lakes. The sod house settlements were abandoned in late April when households established tent camps on the outer islands for spring sealing. Both basking seal (ottuk) techniques and open water kayak hunting were employed, with bearded and harbor seals as primary prey. In June the northward migration of the harp seals would be exploited by open water hunting, occasional- ly walrus and beluga could be had, and sea birds were hunted and their eggs collected (Taylor 1974:55-56). After the ice break-up in mid-June people moved to the inner bays where large regional gatherings occurred at aggregation sites. In the Nain area this was at Amitok (Pardy Island), where 200-300 people participated. Kayak-based sealing was conducted in the bays and char and salmon were fished on their seaward runs (Taylor 1974:18, 56). In late August and until October, some families went inland to hunt caribou for their skins and to cache meat for the winter. Those remaining on the coast engaged in sealing, fished char and salmon on their landward migrations, caching them for future use, and fished cod (Taylor 1974:57-58). On the Cusp of Change: Into the Early 19*^1 Century During the late 1 8'^'^ and early 1 9*^*^ centuries a series of changes occurred as central coast Inuit gradually abandoned their communal houses and resettled at the Moravian mission centers. Although the missions at Nain, Okak and Hopedale were established between 1 771 and 1 782, it was not until the early 1 9th centu- ry that the majority of Inuit resided in these settle- ments. In 1 793 the Moravians reported four houses with a total of 60 Inuit at Nain (P.A.I :1 793:21 5). In 1 795 the Nain mission reported a congregation of 28 plus 40 others who were permitted to live at the set- tlement (P. A. 1 :1 795:354). In 1 806, after the "revival" episode, there were 42 baptized and 22 unbaptized Inuit (P.A. 4:1 806:1 25). By 1810 there were 91 in the congregation (P.A. 5:1810:55) and in 1819 there were 146 congregants and a total of 1 75 inhabitants in the Nain settlement (P. A. 7: 1 81 9: 1 66). The slow recruit- ment rate well into the early 19'^'^ century suggests that Inuit assessed the costs and benefits of affiliating themselves with the missions and initially found few 218 CHAPTER 1 1 advantages in doing so. Yet by the second decade of the 19*^^ century the dynamics of autonomy/depend- ence relations had shifted to the extent that autonomy was no longer a viable strategy on the central coast. The primary attraction of settling at a mission sta- tion was close and reliable access to European goods since, as noted above, the Moravians' frustration with a low conversion rate indicates that religious enthusi- asm was not a central factor (Kaplan 1983:365). Initially, the Moravians tried to deal with this problem through a strategy of handling economic transactions and religious conversion as separate spheres. Goods and services had to be paid for and food was not dis- pensed freely except in times of dire need. Since the mission store was accessible to both converts and "heathen," conversion had no direct economic benefits and residence at mission settlements might incur sub- sistence disadvantages because of the non-optimal placement of the missions relative to game resources (Hiller 1971:84-85). Thus, until the early 1800s Inuit retained their autonomy in the hinterlands of the mis- sion settlements, leaders attempted to lure converts away from the missions and traders continued their travels to southern Labrador. In the late 1780s and 1790s the Moravian strate- gy of separating the spiritual realm from utilitarian rewards was undermined by organizational changes involving closer links between mission policy and eco- nomic practices, such as providing firearms to com- pete with southern traders and allowing credit (Hiller 1971:94-95). These practices increased the potential benefits of mission affiliation and subverted the ability of independent Inuit traders to sustain their activities and retain followers. Increased economic engagement with the Moravians impacted social relations and mate- rial culture. The availability of guns for caribou hunting eliminated the use of drive systems and large group cooperation, while sealing and fishing with nets pro- vided larger immediate returns with minimal coopera- tion. In the 19'^'^ century, fox trapping drew Inuit into credit/debt relations and greater dependence on European goods since trapping did not generate direct subsistence rewards (Kaplan 1983:361-363). Those Inuit who settled at the missions not only underwent religious conversion but also significant social transformations. The missionaries abhorred the multi-family houses and their associated "heathen" practices so mission Inuit were required to live in sin- gle-family dwellings. The Moravians were concerned about the potential bad influence of "heathens" outside the settlements, so converts were encouraged to hunt and fish close to the missions in order to minimize contacts with "heathens" and to permit Moravian mon- itoring of the converts' behavior. This tethering was problematic for a traditional economy requiring mobil- ity and opportunism, especially at Nain, where the mis- sion was established at an inner bay location many kilometers distant from the best winter hunting areas on the outer islands. The Moravians therefore had to accept a degree of residential flexibility and seasonal mission occupation on the part of their converts (Hiller 1971:89-9;, Kaplan 1983:365). An important shift in subsistence practices encouraged by the Moravians was an intensification of cod fishing in the fall to lay in a stored food supply to offset frequent winter short- ages. But as late as 1814 the Moravians reported diffi- culties in establishing a storage-oriented cod fishery because Inuit were satisified with 1 00-200 fish and did not like that type of work (P. A. 6:1 81 4:54). An important aspect of mission-related social transformation concerns women's roles. The status of women in Inuit society in general has been the subject of considerable discussion (e.g., Briggs 1974; Ciffen 1930; Cuemple 1986). A strong division of labor by gender is evident in the ethnography, but the function- al complementarity of roles is generally stressed. At the same time, however, accounts of violence towards women abound, wife-stealing is a recurrent theme and the symmetry or asymmetry of spousal exchange has been debated. The ethnohistorical material from Labrador is no different. Taylor (1974:80) refers to accounts of runaway wives and spousal violence, and POST-CONTACT INUIT SETTLEMENT ON NUKASUSUTOK ISLAND 219 the politics of polygyny (Taylor 1 974:67-70) were open to abuse as well. From one perspective, multiple wives provided the male household leader with more labor power and higher prestige, while from another per- spective co-wives provided each other with labor assis- tance (Cabak 1 991 :42). The distribution of these costs and benefits is hard to determine from the published sources and there is no information on the ranking of co-wives within households. Women could also be shamans (Taylor 1974:87, 1989) and thus possess spiritual power. Taylor (1974:87) refers to a joint per- formance of seal hunting prediction by Sattugana (woman) and Millik (her husband) at Nukasusutok in 1774, commenting: "...it is interesting to consider the indirect control Millik would have over the economic activities in his camp through his wife's reputation as a shaman." But Sattugana seems to have had direct influence over her husband as well, forbidding him to undertake actions on advice from spirits (Taylor 1 974:88). The history of Mikak and Tuglavina is a com- plex mixture of a strong woman as social agent, con- flict and spousal abandonment. Whatever the conflicting evidence, there are strong grounds for assuming that in many cases an Inuit woman's status in 1 8^^ century Labrador was not enviable. Perhaps this is an important reason why women were the core converts when the Moravian mis- sions began to attract residents towards the end of the 18^^ century. In 1799 the Nain mission observed that "...in general, the men do not seem so determined in their resolution to abide faithful as the women" (P. A. 2:1799:328) and it appears that women were impor- tant catalysts in the religious revival of 1804-1805 (Brice-Bennett [1981], cited in Cabak 1991:58). Cabak (1991:62-67) suggests that Inuit women affiliated themselves with the missions out of an interest in Christianity as a source of hope and security, a desire to change the unhappy or abusive circumstances of their lives, to take advantage of educational opportuni- ties and childcare possibilities within a larger commu- nity, to extract themselves from the marginality of wid- owhood and old age, and to acquire European goods. As far as the latter is concerned, archaeological mate- rials from 19'^'^ century middens at the Nain mission settlement suggest that Inuit women were active in integrating European goods into household contexts and that the use of these goods may have contributed to women's prestige (Cabak 1991:181; Cabak and Loring 2000). AN ETH NOH ISTORIC SKETCH OF THE NUKASUSUTOK SETTLEMENT References to Nukasusutok in the ethnohistoric sources have been patched together to form a sketchy settlement history. The synopsis is derived primarily from J. Garth Taylor's and Helga Taylor's published research, which draws on the mostly German language Moravian mission diaries and letters. Also consulted were the Moravian Periodical Accounts (P. A.), which are annually published English summaries of mission diary entries and letters. Most of these accounts probably refer to Nukasusutok-8, but it is possible that some of the early 19'^'^ century references to Nukasusutok might refer to a dwelling at Nukasusutok-7. The demographic data pertaining to the Nukasusutok-8 settlement are outlined in Table 57. In the following account the settlement history is broken down according to the years for which relevant infor- mation is available. References to other localities are included when the information is relevant to under- standing the Nukasusutok settlement. Table 57. Demographic Data for the Nukasusutok Settlement from Moravian Sources (Taylor 1974:16). No. of Houses Population 1 773-74 23 6 1 774-75 (3) (60) 1 776-77 4 62 1 779-80 2 33 1 781-82 2 50 Parentheses indicate estimates 220 CHAPTER 1 1 1772-73 The brothers Millik and Pattiguk, who would winter the next year on Nukasusutok, each had a winter house on Niatak Island (7 km northwest of the Nukasusutok set- tlement). Each household contained 20 people (Taylor 1974:77). Millik's wife Sattugana was an eminent shaman and performed a ceremony during the winter (Taylor 1974:87, 1985:123, 1989:300). 1 773-74 The Nukasusutok settlement consisted of two houses containing a total population of 36 people (Taylor 1974:16, 71). Millik (father-in-law) and Okarloak II (son-in-law) shared a household (Taylor 1974:75). Millik's wife Sattugana gave a shamanist performance concerning the weather and seal hunting possibilities (Taylor 1974:87, 1989:301). Taylor (1974:87) sug- gests that Sattugana's reputation as a shaman may have contributed indirectly to Millik's control over community economic activities. Sikkuliak (father) and Kigluana (son), who would later reside at Nukasusutok, shared a house at Satosoak Island (30 km west of Nukasusutok). 1774-75 Millik ordered his sled and 1 2 dogs to be prepared for the missionaries when they left the Nukasusutok set- tlement. He sent his eldest son Aumarak and Akbik with them. Taylor (1974:80) cites this incident as one of the few documented examples of authority asser- tion by a household head. 1776-77 In December 1 776 a stranded whale was towed to the Nukasusutok settlement just before freeze-up (Taylor 1974:76-77, 1988:128, 1990:52). The baleen was described as "marketable," the meat as "spoiled, " but the skin was still edible (Taylor 1988:128, 1990:59). Given this bonanza of whale skin, people from all over the Nain region converged on Nukasusutok to feast and celebrate. The community consisted of four sod houses with a total of 62 people and eight snow hous- es that probably contained at least 38 visitors, for a total of 1 00-1 50 people (Taylor 1 974:1 6, 1 990:60). In January 1 777 the Inuit built a kagglk (festival house) of snow, 5.3 m high and 8.3 m in diameter, with an entrance passage. The kaggik activities mostly involved men, and included the nullutak game (in which a bone with holes is suspended from the ceiling and participants attempt to skewer the holes with sticks) and boxing (Taylor 1990:53-54). Taylor (1 974:78) was able to use the Moravian census data to reconstruct the kinship relations among the sod house inhabitants (Figure 141). Household 1 : Sikkuliak was the head. The unit con- sisted of 10 people, including Sikkuliak's two wives, six children, and the married son Kigluana (Taylor 1974:75). Household 2 Millik was the head. The unit consist- ed of 1 6 people, including Millik's three wives. Millik's second wife was Sikkuliak's daughter, therefore Millik was Sikkuliak's son-in-law. This provides evidence for an uxorilocal relationship. The household also includ- ed five of Millik's children, two of whom were married with their own families. Household 3: Pattiguk was the head. The unit con- sisted of six people, including Pattiguk's three wives and two children. Note that the brothers Pattiguk and Millik had lived in the same settlement at Niatak in 1 772-73. Household 4: Tuglavina was the head. The unit consisted of 1 1 people, including Tuglavina's three wives (not Mikak, who had been abandoned the previ- ous spring) and Tuglavina's brother, with his two wives and three offspring. Taylor (1974:81) assumes the 1777 Beck census ordered the houses in terms of importance, implying that Sikkuliak (household 1) was most prominent. Taylor goes on to suggest that Sikkuliak's pre-emi- nence was due to his kinship tie as Tuglavina's broth- er (possibly elder), as well as being Millik's father-in- law. Additionally, we know that the brothers Sikkuliak POST-CONTACT INUIT SETTLEMENT ON NUKASUSUTOK ISLAND 221 AtOtO 1=o AST) =0=0 Household 4 (Tuglavina) o=A A A 6 6 6 Household 1 (Sikkuliak) "Z)f1tOtO Household 2 (Millik) 1 = = = 6^ Household 3 (Pattiguk) 141/ Kinship relations of households at the Nukasusutok settlement, winter I 776-1 777. After Taylor (1 974:78): reproduced by permission of the Canadian Museum of Civilization. and Tuglavina quarreled periodically (Taylor 1974:82) and that they probably never shared a household, although in this case they were co-resident at the set- tlement. Parenthetically, this fraternal relationship is interesting in light of the etymology of Nukasusutok given by Wheeler (1953:62-63): "the place where the brothers quarreled." 1777-78 Sikkuliak and his son-in-law Millik began to build a house together at Okak (Taylor 1 974:74). 1 779-80 This winter the Nukasusutok settlement consisted of two houses and a total of 33 people (Taylor 1974:16). Pattiguk resided here and his nephew Kapik left the mission settlement at Nain to live in Pattiguk's house- hold (Taylor 1974:74). Sikkuliak and his son Kigluana shared the other household (Taylor 1974:75). 1780-81 Sikkuliak and his son Kigluana shared a household at Kheovik, at Voisey's Bay (Taylor 1 974:75). 1781-82 Two households with a total of 50 people resided at the Nukasusutok settlement (Taylor 1974:16, 71). A stranded 1 8-20 foot minke whale was towed to the set- tlement (Taylor 1974:32, 1988:128). Three families moved away from Nukasusutok: Millik, his married son Aumarak and Naksuk (Taylor 1974:74). Sikkuliak and Ketornek shared a household at Nain (Sikkuliak was the uncle of Ketornek's wife: Taylor 1 974:74). A spring camp was established seaward from Nukasusutok. On May 26 it consisted of three tents (22.5 people esti- mated) and on May 27 four tents (30 people estimat- ed: Taylor 1974:18). 1783 The Nukasusutok Inuit found "...a hole in the ice with two sled-loads of sea birds," probably dovekies (Taylor 1974:54). In other words, there was an open water area that attracted a large number of birds, which were hunted rather thoroughly. 1784 Millik and his eldest son Aumarak were killed by Tuglavina and others while trading at Cape Charles in southern Labrador (Taylor 1974:92). 1794 Apkajunna, his family and two others visited Okak in February from Nukasusutok. He was instrumental in organizing the construction of a festival house at 222 CHAPTER 1 1 Okak. A 1 783 Moravian account states that Apkajunna was originally from Sagiek and a 1 784 account men- tions him as a whale harpooner at Okak. According to accounts from 1 787 and 1 794, he later became a mid- dleman in the trade with Europeans, acquiring a wood- en boat and wintering in the Hamilton Inlet region (Taylor 1990:62). 1800 Only a few "heathen" families remained in the Nain area. It appeared that the "heathen" were: "...more than ever convinced of the necessity for conversion" (P.A. 2: 1800:472). In December the angakok (shaman) Sigsikak went from Nukasusutok to Nain with his "whole numerous family" (P.A. 3: 1801:12). Two Englishmen arrived unexpectedly and came to the Nukasusutok settlement to catch seals in nets and win- ter there, but the Inuit "...didn't seek the company of the strangers." (P.A. 2: 1800:472) 1805-6 A group of "heathen" were reported at Nukasusutok. They maintained relations with traders and settlers in Hamilton Inlet (P.A. 3: 1805:489). At Hopedale in 1806, Siksigak was baptized and re-named Mark, which the Moravians regarded as a "striking conver- sion" (P.A. 4: 1806:87). 1809 The missionaries in Nain reported: "We have heard with much concern, that a man, who had lived two winters on our land, and even last winter gave good hopes that he would be converted, has been seduced by the hea- thens at Nokkasusuktok to change his mind, and no more intend[s] to live on our land" (P.A. 4: 1809:453). 181 1 During the winter Inuit from Nukasusutok came fre- quently to the mission at Nain to trade. Their "princi- ple leader" moved to Nain in February, seemingly ready for conversion (P.A. 5:1811 :1 30). Discussion One of the major points emerging from this short set- tlement history is that Nukasusutok-8 represents repeated occupations by many different household units. These households circulated among several set- tlement locations over the years, some close by in the Nain area, others further afield in Okak, Sagiek and Hamilton Inlet. Nonetheless, there appears to have been a degree of consistency in the social composition of the settlement during the 1 year period of 1 772-82, with the families of Sikkuliak, Millik and Pattiguk being mentioned frequently. This is probably a consequence of the fraternal relationship between Millik and Pattiguk and Millik's marriage to one of Sikkuliak's daughters (see Figure 1 38). But although some settlements may have been associated with specific families or kin units for a time, residential mobility was high and structured by both kinship and opportunistic factors. Another point is that despite considerable contact with the Moravians in the early years after the estab- lishment of the Nain mission in 1771, the inhabitants of the Nukasusutok settlement remained unconverted. By the early 1800s Nukasusutok was regarded by the missionaries as an outpost of recalcitrant "heathens." Not only did they still resist conversion, but they encouraged converts residing at the Nain mission to abandon their new beliefs. Furthermore, the Nukasusutok Inuit maintained contacts with traders and European settlers in Hamilton Inlet, an undesirable link in Moravian eyes, since it undercut Moravian influ- ence and control. This, along with the residence of sev- eral prominent traders and shamans at the settlement, suggests that Nukasusutok was a prominent site of resistance to European domination. There are two obvious archaeological conse- quences of this settlement history. First, the sod hous- es at Nukasusutok-8 were probably cleaned out or ren- ovated periodically, thus it is unlikely that the house and midden contents can be related to any of the doc- umented individuals and the house floor contents will mostly represent the last occupation phase. Second, POST-CONTACT INUIT SETTLEMENT ON NUKASUSUTOK ISLAND 223 we might expect that a settlement composed of Inuit resisting control by the IVloravians might contain a combination of traditional Inuit and European goods, but unusual quantities of European goods might be present, particularly if prominent traders such as Tuglavina and the nascent middleman Apkajunna were resident there for extended periods. Another possibil- ity might be the presence of artifacts related to shamanist practices, since Sattugana, Tuglavina and Sigsikak all were believed to possess special powers. THE ARCHAEOLOGY OF NUKASUSUTOK-7 (HcCI-8) Nukasusutok-7 is a Contact to post-Contact Period site located on Pumialuk Point on the southwestern portion of Nukasusutok Island (Figure 16). The area was tra- versed by a Smithsonian associate in 1 973 but was not revisited until contract and provincial government archaeologists passed by on a mineral exploration- related survey in 1 997 (Penney 1 997; K. Reynolds, per- sonal communication). The author conducted a brief reconnaissance in 2004. The site designation lumps together five separate localities that are distributed along a 700 m stretch of the point. Each locality is described separately, proceeding westwards. L-1 On an extensive flat area near the base of the point were a large oval tent ring (5 by 5 m) and a small stone grave or fox trap. L-2 A large stone grave was located in a narrow beach pass. Constructed of boulders, the grave was 3.5 by 1 .5 m in size and was oriented east-west. A gift cache (1 by 1 m) directly adjoined its western end. This fea- ture might be Strong's Grave 3 (see below), which had a gift cache and skeletal remains oriented with the head towards the east, but his description is so sketchy it is impossible to be sure. Strong mentioned that the chamber of Grave 3 contained broken brass and iron objects and the cache had broken metal objects. L-3 Two oval tent rings (both 4 m in diameter) and one square tent ring (4 by 4 m). L-4 Four square tent rings (two 5 by 5 m, one 5 by 4 m, one 4 by 4 m); one contained a probable internal hearth. A clay pipestem was noted on the surface of the beach crest. At the western side of the beach pass was a wood cache positioned under a bedrock overhang. L 5 A broad beach pass contained four features. On the western side of the pass was a semi-subterranean house. Ten meters east of the house was a small rec- tangular stone "frame" and 20 m east of this were two more such frames. Semi-Subterranean House This feature was square, 5 by 4.5 m in size. Its well- preserved back and side walls (30-50 cm wide) were built up with sod to a height of 25 cm above the exter- nal ground surface and 50 cm above the internal floor level. The front of the house lacked a distinct above- surface wall; it was simply dug down ca. 20 cm from the turf surface. There was no clear entrance passage, only a shallow depression 1.5 m wide by 2.5 m long that extended from the center of the front wall. The entrance faced south. There was a single rear sleeping platform that encompassed the entire rear wall and was 1.3 m wide. Three small shovel tests were excavated in an attempt to find diagnostic materials to date the struc- ture. Test Pit 1 was placed 1 .4 m from the front wall of the house, just east of the entrance depression. It revealed a 20 cm thick and extremely compact peat layer on top of beach gravel. About 10 cm below the peat surface was a charcoal layer that contained two 224 CHAPTER 1 1 well preserved seal bones. Test Pit 2 was excavated in the center of the house floor, 1 m from the front wall. Here there was a 10 cm deep peat layer on top of beach gravel. Towards the bottom of the peat were two small iron fragments, a poorly preserved seal bone and a piece of wood embedded vertically into the floor. Little charcoal was encountered. Test Pit 3 was placed 2 m from the front wall, on the western side of the entrance depression. It contained 25 cm of peat on top of beach gravel; no cultural material was observed. This semi-subterranean structure is notable for its "light" construction, exhibiting a form more similar to a transitional season qarmat than a winter house. The relatively small size of the dwelling, plus its lack of a well-defined entrance passage, is suggestive of a 19'^^ century structure. Unfortunately, no datable finds were recovered. The paucity of cultural material suggests the dwelling was occupied for only a short period of time. Stone "frames" The three stone "frames" were rectangular alignments of small rocks, similar in size (1 .3 by 0.80 m) and ori- ented approximately north-south. They may mark early Christian graves. THE ARCHAEOLOGY OF NUKASUSUTOK-8 AND W. D. STRONG'S GRAVE COLLECTION The sod-house village Nukasusutok-8 (HcCh-10) is located towards the southwest portion of the island at the base of a steep spruce-clad valley (Figures 16, 1 42, 143). W. D. Strong was informed about the Nukasusutok house structures by an Inuk from Main. In 1928 Strong and an expedition companion spent the period of August 1-7 on the island investigating Nukasusutok-8 and several Inuit graves. This occurred at the tail-end of the Rawson-MacMillan expedition and it is evident from Strong's diary entries that he was fed-up and looking forward to heading home. Kaplan's (1 983:469) visit in 1 980 identified five sod houses, but Strong's field notes indicate seven dwellings, two of which may be tentative. Since the two "missing" struc- tures could be obscured beneath today's heavy vegeta- tion cover. Strong's sketch map is used as the basis for Figure 142 and his numbering sequence for the houses^ is retained. Also present at the site are a tent ring and the remains of a kayak stand or cache. A grave lies near the shore ca. 200 m west of the site. Besides Kaplan's brief visit in 1980, more recent inves- tigations of the site include the author's 1 992 test-pit- ting of the midden in front of House 5 and a recon- naissance by J. Garth Taylor in 1966 (Taylor 1966). Strong excavated much of House 1 , part of House 5, and tested Houses 4, 6 and 7. He found relatively little artifact material within the houses. Much of his interest seems to have been directed towards graves located in several places on the island, some of which contained considerable quantities of artifacts. Unfortunately, the Peabody Museum collection is sim- ply registered as from "Nukasujuktok Island," with no indication as to which parts of the material derive from the house excavations versus the graves, or to which graves the material belongs. Strong's fieldnotes and photographs can be used to associate a very few artifacts with specific graves, but over 90% of the material must be regarded as of uncertain prove- nience. Nonetheless, given his comments on the paucity of artifacts within the excavated houses it is likely that much of the museum collection is derived from the graves. Strong's fieldnotes suggest some items excavated by him are either not present in the Peabody Museum material or were never collected and brought south, possibly because they were regarded as too "modern" (e.g., European ceramics and an iron shovel). The seven sod houses at Nukasusutok-8 are ^Strong's and Kaplan's (1983:470) house numbering is identical for Houses 1 and 2, but Kaplan's Houses 3, 4, and 5 correspond to Strong's Houses 5, 6, and 7, respectively. POST-CONTACT INUIT SETTLEMENT ON NUKASUSUTOK ISLAND 225 sea 1 42/ Nukasusutok-8 site map. Based on W. D. Strong's 1928 field notes, with modifications. Used by permission, National Anthropological Archives, Smithsonian Institution. arranged in two different groups (Figure 142). Houses 1 , 2 and 3 lie furthest from the modern shore- line and are built into the side of a steep slope, while Houses 4, 5, 6 and 7 are excavated into a relatively flat beach surface closer to the contemporary shore- line. Two house types are present. Houses 1, 2 and 5 are large rectangular communal houses with long straight entrance passages; this form was common in the 1 8^^ century. They measure about 8 m along their rear walls and 7 m along their side walls. Houses 1 and 5 have prominent midden deposits. Houses 6 and 7 are smaller. House 6 measures 5 m along its side, while House 7 measures 6 m along its rear wall and 5 m along its side walls, both have short entrances rather than long entrance passages. Kaplan (1 983:469) dates these structures to the 1 9^'^ century and Strong's field notes confirm this in referring to them as "regular board houses," in which were found an iron shovel and "pink plates." No size data are available for the currently "invisible" Houses 3 and 4. Strong's sketch map and his field note comments that iron and porcelain from House 4 indicate it is the same age as House 5 suggest that the former may have had communal house dimensions. This would increase the total of communal houses to at least four, which would be consistent with Moravian obser- vations of four occupied sod houses in the winter of 1 776-77. House 3 was not tested by Strong so there are no chronological indicators. House 1 Strong began the excavation of House 1 by digging a trench up the long entrance passage towards the front of the house. He refers to a thin and scattered midden deposit full of iron, brick, china, porcelain and cham- ber pot handles. These European objects were domi- 226 CHAPTER 1 1 1 43/ Nukasusutok-8: overview towards the southwest, August 1928. W. D. Strong Collection, © Smithsonian Institution, National Anthropological Archives, Negative No. 99-10590. 144/ Nukasusutok-8: House 5 overview towards the southwest, August 1928. W. D Strong collection, © Smithsonian Institution, National Anthropological Archives, Negative No. 99-10591. nant to a depth of two or three feet (60-90 cm), at which point a frozen layer containing bone fragments was encountered. Strong's rough sl N A /-y' 0/ (7 O 8x1 2y F Flat rock Outer pit edge I I Inner pit Burin spall * Endblade + Microblade • Utilized flake 11x12y lOxlOy llxlOy / 74/ Double Island Cove- 1 , Structure 1 : tool distribution. 8x1 5y 9x1 5y > N H /-y' D 40.0 mm, W= 1 1 .0, TH= 2.7), with fine use retouch along part of one edge. ATTU'S BIGHT-1 (HeCk-4) Attu's Bight-1 is a large Rattler's Bight Phase (4200- 3500 B.P.) Maritime Archaic site, located on the north side of Webb Bay, just west of the entrance to Port Manvers Run (Figure 163). It was discovered and sur- face collected by the Smithsonian Institution in 1985 (Fitzhugh 1 986:57) and revisited by the author in 1 992 and 1 993. The site consists of cultural remains extend- ing for about 350 m along a set of sand and gravel ter- races 19-21 m as I. (Figure 1 80, 1 81 ). Smithsonian field notes identify three different beach levels (A, B, C); that were suggested to be indicative of chronological differences within the site. However, the vertical dis- tance between these terraces is minimal (0.5-1.0 m) and the slope is very gradual, so the levels may have little relevance for relative dating the site contents. Indeed, according to the general uplift regime for the Main inner bay area (Figure 5) the 19 m terrace emerged 6000-5500 B.P., well prior to the Rattlers Bight phase. Since the terrace front is an erosion face that plunges steeply from 1 9 m to the modern shore- line, during the Rattlers Bight Phase there were proba- bly no lower beaches suitable for habitation. In any event, the large horizontal extent of the 19-21 m sur- face was ideal for longhouse placement and it provid- ed a broad field of view over Webb Bay. Most of the site is completely deflated by wind erosion. The little surface vegetation present consists of moss-lichen and isolated stands of spruce trees. Thicker spruce growth is found immediately north of the site, downslope to the south, and to the west. Flakes and implements of Ramah chert and other materials lie exposed on the sand/gravel surface, dis- tributed in linear patterns parallel to the beach levels. The 1985 Smithsonian survey identified nine separate loci and the 1992-93 investigations added three more. These 1 2 loci (Figure 181) are described briefly below, based on Smithsonian field notes and later observa- tions. Test excavations conducted at L-3/6 in 1 992-93 are described thereafter. Overview of Localities L-l: At the extreme eastern end of the terrace was a Ramah chert scatter. The 1992 surface collection included three biface fragments of thermally altered Ramah chert, of which two are probably point pre- forms (Figure 1 90;j, k), while the third is an edge frag- ment from a larger bifacial implement (Figure 190:i). Also recovered was a large green slate flake (Figure 191:a) with a slightly polished medial surface and scraper-like retouch along one edge (L= 79.6 mm, W= 64.4, TH= 7.7). L-2 Parallel to the southeastern terrace front was a scatter of Ramah chert flakes, 60 m long, 6 m wide. Just north of this was a packed cobble feature, sub-rec- tangular in shape and 2.5 by 1.0 m in size. The Smithsonian surface collection consisted of two Ramah chert micropoints, a ground slate preform and two uti- lized flakes (one of Ramah chert the other of "chert"). L-3 and L-6: Located on the middle beach level near the eastern end of the terrace, these loci contained the only in situ portion of the site. The Smithsonian regis- tered two separate loci here, but they probably consti- tute a single occupation area, of which perhaps two- thirds is deflated. L-3/6 was marked by a scatter of Ramah chert flakes 40 m long and at least 5-6 m wide, the easternmost portion of which extended into the in situ deposit. At L-3 the Smithsonian collected a Ramah chert stemmed point, three flake points (two of Ramah chert, one of unspecified material), four Ramah chert biface fragments and a slate celt fragment, while at L- 6 they collected two Ramah chert biface fragments and a slate celt fragment. The 1 992-93 test excavations are described below. WEBB BAY-PORT MANVERS RUN: SMALL SCALE INVESTIGATIONS 261 182/ Attu's Bight L-3/6: view towards the east. L-4 Positioned on the highest beach level, L-4 consist- ed of a small scatter of quartz flakes, from which the Smithsonian collected two quartz biface fragments and a Ramah chert flake knife. In 1992 a Ramah chert biface preform (Figure 190;c) was collected 7 m north of L-4. The implement is fully flaked dorsally but only partially flaked ventrally, and it bears a striking plat- form distally (L= 43.5 mm, W= 21.9, TH= 9.2). L-5 Situated on the upper portion of the middle beach level and adjacent to L-3/6, L-5 was a lithic scatter con- taining flakes of Ramah chert, Mugford chert and slate. The Smithsonian collected two stemmed points, a sin- gle-shouldered biface and a biface fragment, all of Ramah chert. L-7: Located on the central portion of the highest beach level, L-7 was a 120 m long scatter of Ramah chert flakes. The Smithsonian surface collected four flake points, three of Ramah chert, one of quartz. In 1992 a complete Ramah chert stemmed point was recovered (Figure 190:a). The implement is "typical" of the Rattlers Bight phase, with bifacial flaking, a bicon- vex cross-section, a slightly tapering stem and a slight- ly convex base (L= 47.0 mm, W= 21.2, TH= 6.2, stem L= 16.4, stem W[shoulder/base] = 14.1/1 1.3). L-8: Near the center of the site, slightly back from the eroded terrace front, was an oval ring of cobbles, 4 m in diameter. The feature was associated with large chunks of Ramah chert, banded slate and quartz flakes. The Smithsonian surface collected a stemmed point preform, two biface preforms and a sidescraper, all of Ramah chert. L-9: Near the western end of the site, on the highest beach level, was a 50 m linear scatter of Ramah chert flakes. Smithsonian field notes mention the presence of a stemmed point resembling Sandy Cove Complex form. L-]0: Lying between L-7 and L-9, but slightly lower, was a 1 m diameter cluster of about 10 head-sized rocks associated with smaller cobbles. A couple of Ramah chert flakes lay on top of the feature, which may be a burial cairn. West of the feature was a 1 by 1 m scatter of Ramah chert flakes. L-11: This was the westernmost cultural feature observed at the site, lying in an isolated area 40 m south of some sand dunes. It consisted of a cluster of about 1 head-sized or smaller rocks, 1 m in diameter, without associated lithic material. The feature may be a burial cairn. L-12: Located on the highest beach level and towards the northeast corner of the site, L-1 2 was a small scatter of Ramah chert flakes. A Ramah chert flake point was surface collected (Figure 190:b); it has bifacial retouch limited to the edge margins, a slight shoulder on one lateral edge, and slight basal thinning (L= 23.9 mm, W= 11.6, TH= 3.7, shoulder HT= 9.3). 262 CHAPTER 12 46x62y 49x52y deflated § in situ 46x47y */ in situ 49x47y Feature 2 53x50y / Edge of deflatii Charcoal ^ Bone fragments 1 m 183/ Attu's Bight L-3/6: excavation with flake distribution. Test Excavations at L-3/6 In 1992 and 1993 small scale test excavations were undertaken in the />? s/Tw deposits at L-3/6 (Figure 182) to acquire material for radiocarbon dating and to determine if there was much likelihood of identifying traces of dwelling structures or other features. In 1 992 a small grid system was set up on the western edge of the in situ deposit, including part of the deflated area. A total of 1 7.8 m^ was investigated, of which only 3 m^ was excavated in situ, the rest was simply a trow- el-through of the exposed sand and gravel. Tools and flakes were point-plotted, while flakes were collected in 1 m^ units. A profile from the in situ deposit indi- cated possible traces of two occupation layers, so in 1993 two 1 m^ test pits were excavated further east- ward in the m situ deposit to provide additional strati- graphic information (Figure 183). The sediments in the vicinity of L-3/5 consisted of coarse sand and pea-size gravel. The depth of defla- tion varied from 10-40 cm. A stratigraphic profile (50x51 y/50x50y) taken from the edge of the in situ deposit exhibited some interesting details (Figures 1 84 and 1 85). A total of 1 1 separate stratigraphic units were identified, including at least four buried humus layers that indicate a repetitive cycle of temporarily stable vegetation surfaces followed by eolian deposi- tion. The main cultural level was unit 9, 35 cm below the surface, a sand layer that exhibited a shallow basin-like feature excavated slightly into the underly- ing sand/gravel. Unit 9 contained flakes, tools, red ocher and oxidized sand, and charcoal flecks. Immediately above unit 9 was unit 8, a humus layer containing charcoal flecks. The other stratigraphic units were culturally sterile, except for unit 2, a faint buried humus layer 1 5 cm below the surface that con- tained two Ramah chert flakes. In other parts of the in situ deposit Ramah chert flakes were exposed in a shallow deflated area 1 0-1 2 cm below the surface, sug- gesting there may have been an occupation at roughly that depth. But given the instability of sand surfaces. WEBB BAY-PORT MANVERS RUN: SMALL SCALE INVESTIGATIONS 263 -'A- /84/4ffM'5 6;g/7f L-3/6 excavation at the edge of the in situ deposit. redeposition by deflation and eolian processes cannot be ruled out. In 1 993, test pit 50x50y was excavated adjacent to the 1 992 profile. At the 1 cm level, seven Ramah chert flakes were encountered, as well as one flake of 50x5 1y 1 . Gray sand 2. Buried humus, Ramah chert flake 3, 5, 7. Yellow-brown sand 4, 6. Buried humus 8. Buried humus with charcoal flecks 9, Red-yellow sand, charcoal, Ramah chert flakes 9a, Basin feature, oxidized sand, minimal charcoal 9b Charcoal beneath cultural layer 10, Yellow-brown sand/gravel 11. Gravel lens 50x50y 185/ Attu's Bight L-3/6: profile. greenstone and a split cobble with a faint trace of red ocher on its surface. An extremely faint spot of red ocher was observed in the sand. Given the paucity of material it is difficult to conclude there was a dis- tinct occupation floor at this level, but the split cobble and red ocher might counter the redeposition argument, since the former would be resistant to wind action and the latter would have been erased by it. No flakes were observed at a corresponding level in test pit 53x50y, further to the east. In both 1993 test pits, however, a distinct occupation floor was found corresponding to the lower unit 9 level in the 1992 test, although this floor lay somewhat deeper (45 cm) than in the 1 992 units (35 cm). Two features were identified in the basal unit 9 deposits, both of which appeared to be small hearths (Figure 1 83). Feature 1 was located near the edge of the deflated area. It was a 60-70 cm in diameter oval patch of red-yellow oxidized sand mixed with char- coal and faint red ocher stains, accompanied by a 55 by 35 cm oval concentration of 38 small burned bone fragments. None of the bone fragments were identifi- able, although one might be a rib fragment. There were no distinct hearth border or other "functional" rocks associated with the feature. The slightly basin- shaped profile of the feature suggests a fire was sim- ply kindled in the upper portion of the ground sur- face. A sample of charcoal from this concentration was radiocarbon dated to 4080±100 B.P. (Beta- 571 26). Within and adjacent to the feature were small Ramah chert retouch flakes and a few tools: a slate celt fragment, a utilized flake of Ramah chert, a ground slate flake and a small mica fragment. Feature 2 was found at the bottom of test pit 53x50y, 264 CHAPTER 1 2 3.5 m east of Feature 1. It consisted of a 60 cm in diameter oval cluster of fist-sized cobbles associated with a considerable amount of charcoal and a few Ramah chert flakes. Two small red ocher stains occurred nearby. Charcoal from this feature was radiocarbon dated to 4080±90 B.P. (Beta-7 1 477). Figure 186 depicts the spatial distribution of tools while the tool classes are listed by frequency in Table 68. The implements lay in two elongated east/west distributions separated by 1.0-1.5 m, one extending out from the Feature 1 hearth, the other along the southern edge of the excavation. These distributions parallel that of the flakes (see below). The small collection is dominated by utilized flakes of Ramah chert. Also collected were two small mica fragments and 1 1 schist fragments. Table 68. Attu's Bight L-3/6 Tool Class Frequencies. N (%) Stemmed points 3 (1 5.0) Bifaces 2 (10.0) Flake points 1 (5.0) Stemmed flakes 1 (5.0) Utilized flakes 10 (50.0) Celts 1 (5.0) Ground slate flakes 1 (5.0) Split cobbles 1 (5.0) TOTAL 20 The flake distribution is shown in Figure 187. The plotted distribution of flakes in the deflated areas is probably misleading since eolian processes have doubtless smeared the spatial distributions and the material could represent a palimpsest of differ- ent occupation levels. Nonetheless, the frequencies of Ramah chert flakes indicate a distribution about 4-5 m wide (north/south) with two dense concentra- tions: one associated with the Feature 1 hearth area near the center and another ca. 1.0-1.5 m to the south. Thermally altered patinated Ramah chert is found in most of the excavation units, but is most frequent in the hearth area. Table 69 indicates the flake raw material frequencies are almost totally dominated by Ramah chert, typical of Rattlers Bight Phase localities. A fairly high percentage of the Ramah chert is patinated. Figure 188 portrays the size distribution of Ramah chert flakes. A total of 78.4% of the flakes measure < 20 mm and 61.7% < 1 5 mm, suggestive of an emphasis on secondary retouch. Additionally, a total of 1 34 biface-thinning flakes (BTFs) was identified (Figure 189). BTFs were distributed over most of the excavated area, although units with the largest quantities tended to be those with the most flakes overall. Additionally, 42 (3 1 .3%) of the BTFs were >30 mm in size, sugges- tive of the reduction of large quarry blanks or pre- forms. This percentage is much higher than those recorded for the Middle Maritime Archaic site Nukasusutok 5 (Area 2A = 1 3.5%, 2B = 5.4% and 2C = 12.3%). The large BTFs were also distributed fairly evenly across the units. Table 69. Attu's Bight L-3/6 Flake Raw Material Frequencies. Excavation Surface TOTAL N (%) Ramah chert 674 31 1 985 (88.2) Patinated Ramah 76 48 1 24 (11.1) Cray slate 2 1 3 (0.3) Creenstone 2 1 3 (0.3) Cray Mugford chert 1 1 (0.1) Quartz 1 1 (0.1) TOTAL 754 363 1117 The total weights of the lithic raw materials (tools and flakes) are given in Table 70. Ramah chert is clearly dominant; together, patinated and unpati- nated Ramah account for 89.2% of the material. Patinated Ramah chert constitutes a larger proportion of the Ramah chert flakes by weight (21 .7%) than by frequency (1 0.1%). WEBB BAY-PORT MANVERS RUN: SMALL SCALE INVESTIGATIONS 265 46x47y 49x47y 186/ Attu's Bight L3/6: tool distribution. 46x52y 46x47y 49x52y RC=3 RC= 32 RC= 20 PAT=4 RC=4 PAT= 1 GR= 1 GM= 1 RC= 28 PAT=9 RC= 55 PAT= 14 RC= 16 PAT=2 SL= 1 Q= 1 RC=8 RC= 21 RC= 43 PAT= 6 RC= 257 PAT= 8 RC= 28 PAT= 7 missing data RC= 139 PAT= 1 Deflated portion RC=84 PAT=4 / RC= 172 PAT= 65 SL=2 RC= 45 PAT=2 GR= 1 RC= 20 50x50y 53x50y L-1 (10 cm) RC=7 GR= 1 GM= Gray Mugford chert GR= Greenstone PAT= Patinated Ramah chert Q= Quartz RC= Ramah chert SL= Slate 49x47y 187/ Attu's Bight L-3/6: flake distribution. 266 CHAPTER 1 2 L-3/6 Ramah Chert Flake Size Distribution 35 30 25 20 I 15 10 5 <5 <15 <20 <25 <30 <35 <40 <45 <50 <55 5 mm classes 1 88/ Attu's Bight L-3/6: Ramah chert flake size distribution. Table 70. Attu's Bight L-3/6 Total Lithic Raw Material Weights (Flakes and Tools). In Grams. Flakes g Tools g TOTAL g (%) Ramah chert 434.0 66.2 500.2 (71.6) Patinated Ramah 120.0 3.0 123.0 (17.6) Gray slate 0.8 0.8 (0.1) Green slate 3.0 3.0 (0.4) Red slate 68.0 68.0 (9.7) Greenstone 2.4 2.4 (0.3) Gray Mugford chert 0.5 0.5 (0.1) Quartz 0.5 0.5 (0.1) TOTAL 629.2 143.6 624.0 Tool Assemblage Stemmed points: The three stemmed points are all made of Ramah chert. Two are stem fragments of sim- ilar dimensions that retain striking platforms on their bases (Figure 190:e, f). The third is a marginally retouched flake fragment with a striking platform or break at the base of a slight stem (Figure 190:1); it is unclear if this is a finished product or a preform. Bifaces: Two Ramah chert biface fragments were recov- ered, one a medial portion the other a distal element bearing an impact fracture (Figure 1 90:d, h). Both are prob- ably from stemmed points. Flake points: The single exam- ple, made of Ramah chert, is a medial fragment with dorsal marginal retouch on opposite lateral edges (Figure 190:g). Stemmed flakes: This Ramah chert implement (Figure 190:m) exhibits limited marginal retouch on opposite edges that creates a stem-like constriction; the stem base was thinned to remove the striking platform. It could be a point pre- form or simply a utilized flake. Utilized flakes: All of the 1 utilized flakes are made of Ramah chert. One fragment has fine bifacial marginal retouch on opposite edges, possibly indicative of intended use as a flake point. Two implements are made on large biface-thinning flakes, probably derived from quarry blanks. Celts: The implement identified as a celt is a medial fragment made on a plate of red slate (Figure 191:b). The only grinding occurs on two opposite edges that are ground perpendicular to the plate surface (W= 70.0 mm, TH= 14.7). Ground slate flakes: The single example is made of green slate. Split cobbles: A water-rolled flat cobble is split trans- versely and bears a faint trace of red ocher on its edge. It may have functioned as a hammerstone, red ocher crusher, or both. WEBB BAY-PORT MANVERS RUN: SMALL SCALE INVESTIGATIONS 267 46x52y 46x47y 49x52y 1/1 6/2 9/3 3 4/1 5/3 6/3 1/1 7/5 3/2 17/4 10/2 missing data 21/6 deflated portion / 11/1 12/6 11 7/2 50x50y 53x50y Ramah chert biface-thinning flakes total/large 49x47y 1 89/ Attu's Bight L-3/6 distribution of Ramah chert biface-thmning flakes. Discussion The excavation of L-3/6 suggests that two components may be present. The earlier component is the floor buried 35-45 cm below the modern surface. It contains two hearth features with virtually identical radiocarbon dates (4080±100 B.P. and 4080±90 B.P.). The charcoal comprising these samples was not identified to wood species, so it is possible that if spruce trees were used the dates could be somewhat older than the occupa- tion. The later component is a trace layer 1 cm below the modern surface, inferred from a few in situ flakes, a split cobble, a small red ochre stain and deflated Ramah chert flakes lying near the same level. There are no clear chronological indicators for the later compo- nent, but enough time elapsed for at least two humus layers to have developed and been covered by drifting sand. This two component interpretation falls if the upper layer is composed of redeposited material. Smithsonian field notes suggested a Sandy Cove-like point from L-9 could indicate occupations pre-dating 4500 B.P.. However, nothing else in the artifactual material suggests anything other than a Rattler's Bight Phase dating and the slight differences in beach ter- race elevation do not provide much basis for relative dating within the site. Consequently, the vast majority of the material probably pertains to repeated occupa- tion during the Rattler's Bight Phase. Linear scatters of Ramah chert such as those found at Attu's Bight have been interpreted as traces of Maritime Archaic longhouse structures (Fitzhugh 1981, 1984). While this may have been the case here, we should not assume that flake scatters ranging from 40-1 20 m long necessarily represent single continuous structures. Besides the possibility of non-contempora- neous overlap, the stratigraphic situation at L-3/6 rais- es the possibility that some of the flake scatters in deflated parts of the site could be palimpsests of more than one occupation. Two aspects of Attu's Bight seem significant in comparison with some other Rattler's Bight Phase 268 CHAPTER 1 2 190/ Attu's Bight Tools, a stemmed point, L-7: b: flal19-21 m). Five small localities (L-1 to L-5) were identified on the primary beach surface, two more (L- 6 and L-7) were recorded slightly down-slope of the northwestern corner of the beach, while one (L-8) was observed down-slope to the northeast. L-1 was located near the rock outcrop at the northern end of the main beach terrace and consist- ed of a few Ramah chert flakes and tools. A stemmed point and a flake point of Ramah chert were surface- collected. The stemmed point is similar to Rattlers Bight styles (Figure 1 79:m). The dorsal side of the implement is completely flaked but the ventral side is retouched only on the lateral margins. A remnant striking platform runs oblique to the stem base and although the shoulders and upper stem are retouched the basal portion is unmodified, presum- ably unfinished (L = 47.9 mm, W = 26.4, TH = 6.1). The flake point (Figure 1 79:n) is formed by unifacial dorsal retouch limited to the edge margins (L = 33.1 mm, W = 13.8, TH = 2.7, stem L = 10.6, stem W [shoulder] = 9.2, stem W (base) = 7.7). L-2 was situated near the southeastern corner of the terrace and consisted of a small rock mound, 1 m in diameter, with no visible lithic association. It resembled the cobble features at Attu's Bight (L-1 and L-1 1) and may be a burial cairn. L-3, L-4 and L-5 270 CHAPTER 1 2 were all located towards the southern end of the ter- race and all consisted of small scatters of Ramah chert flakes. L-6 was a gently sloping beach surface situated 40-50 m northwest of, and several meters below, the rock outcrop at the northwest end of the beach. L-5 contained a blow-out with exposed flakes of Ramah and Mugford chert and a Ramah chert biface stem, possibly Maritime Archaic (Figure 179:k). Small cobbles and charcoal flecks were also visible on the surface. A 50 by 50 cm test pit adja- cent to the blow-out produced some charcoal, but no associated cultural material. Systematic shovel tests were conducted up and down the L-6 beach, but they did not reveal additional traces of occupation. L-7 was a lithic scatter in a deflated sandy cleft beside the bedrock knoll at the north end of the main terrace and up-slope from L-6. Flake material exposed on the surface consisted of Mugford chert, crystal quartz and slate. A surface collection pro- duced a few Pre-Dorset tools and a biface of either Pre-Dorset or Maritime Archaic origin. The Pre- Dorset implements include a medial fragment of a Pre-Dorset crystal quartz biface with fine parallel pressure flaking (Figure 1 79:j), two small microb- lade fragments of dark gray Mugford chert, a com- plete burin spall of gray Mugford chert and a uti- lized flake of patinated Mugford chert. The medial portion of a burned Ramah chert biface (Figure 1 79:1) might be Maritime Archaic, but a Ramah chert biface of similar width (ca. 40 mm) was found in a Pre-Dorset context at Attu's Point (Figure 201 ;a. Chapter 1 3). A blow-out 1 5 m north of L-7 contained a dark gray Mugford chert microblade fragment. A test pit excavated beside the blow-out did not reveal additional cultural material. Finally, L-8 was located downslope from the northeast corner of the main beach terrace. Yellowish-green slate flakes similar to those found in Pre-Dorset sites in the area were observed on the surface. Discussion Webb Point provides evidence for limited Maritime Archaic and Pre-Dorset activity. The most definite traces of Maritime Archaic presence were found on the prominent terrace, where evidence for Pre- Dorset was lacking. The stemmed point and flake point found at L-1 might indicate a Rattlers Bight phase or slightly earlier occupation. Pre-Dorset appears restricted to the lower beach levels, although possible Maritime Archaic implements were found at L-6 and L-7. The varying beach eleva- tions may indicate chronological differences in occupation period, but the material is so scanty that no reliable conclusions can be drawn. The minimal occupation of the main beach terrace was surprising given its topographic prominence and its physical similarities with the beach terrace at Attu's Bight. Indeed, the limited use of Webb Point contrasts strongly with the amount of Maritime Archaic and Pre-Dorset activity on the north shore of Webb Bay. The next chapter (13) describes Attu's Point, a major Pre-Dorset locality on the north shore of Webb Bay. Following that, Chapter 14 integrates the material presented in Chapters 1 2 and 1 3 and welds it into a more comprehensive discussion of Maritime Archaic/Pre-Dorset social relations in northern Labrador. WEBB BAY-PORT MANVERS RUN: SMALL SCALE INVESTIGATIONS 271 Northern entrance to Port Manvers Run, Thalia Point on right. July 1997. (Photo B. Hood) 272 CHAPTER 1 3 Attu's Point: A Pre-Dorset Centra! Place'' on Webb Baij In 1993 surveys were undertaken west of the entrance to Port Manvers Run along the north shore of Webb Bay. A Pre-Dorset locality dubbed Attu's Point (HeCk-5) was discovered near a rocky point overlooking the narrow tickle between Igloo Island and the mainland (Figure 163). In the middle of a raised beach was an extremely well-preserved axial structure containing a box-hearth filled with thermoliths.^ The initial survey identified at least a dozen loci of Pre-Dorset activity, of which five had traces of dwelling structures, as well as a locus with Maritime Archaic material. Since Attu's Point lay only 1.8 km west of the Late Maritime Archaic occupation at Attu's Bight, and since Attu's Point is the only major Pre-Dorset inner bay locality in the Main region, the site was deemed high priority and was investigated further in 1994. The 1994 work identified several additional loci, for a total of 21 find spots. These substantial traces of occupation suggest that Attu's Point was a "central place" in the Nain region Pre-Dorset settlement system: a repeatedly used seasonal site that contrasts with the scattered and ephemeral small localities that otherwise characterize Pre-Dorset landscape use. OVERVIEW The Attu's Point site is spread out along two west-fac- ing raised beach terraces ranging between 10-12 m asl. (Figures 1 92 and 1 93). The beach deposits consist of medium-sized gravel and they are bordered on the southeast and west by eroded outcrops of anorthosite bedrock. Some portions of the beaches are deflated while other parts are covered with a thin layer of crow- berry or moss. Open spruce forest begins immediately east of the site. The 21 loci were arranged in three linear distribu- tions on different beach levels. L-1 to L-5, L-20 and L- 21 lay near the edge of the 1 m beach terrace, L-6 to L-8 and L-1 7 were in a line slightly higher up and L-9 to L-1 6, L-1 8 and L-1 9 lay on the uppermost terrace between 11-12 m. At least one structure from each of these three levels was excavated or tested (L-1 , L-5, L- 7 and L-9) and surface collections were made at some of the other loci. All but one locus exhibited evidence of Pre-Dorset material; L-1 2 produced traces of Late Maritime Archaic. Each of the excavated features will be described in turn, along with a spatial analysis and descriptive sum- mary of the tools. The surface collected loci will then be described briefly. A summary of the significant spa- tial patterns and a general culture-historical conclusion is presented at the end of the chapter. ^Thermolith is a generic descriptor for rocks heated within a box-hearth for uncertain intended function (heating, cooking or other practices); it denotes both fire-cracked and non-thermally altered specimens. ATTU'S POINT: A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY 273 1 92/ Attu's Point overview towards the northeast. EXCAVATION AND COLLECTION METHODS Excavation of the structures involved removal of the thin crowberry vegetation mat and troweling through the upper few centimeters of the beach gravel. Given time constraints, the size of the excavation areas at each structure was relatively small. Although most of the structurally relevant space was excavated at L-1 , L- 5 and L-7, a better picture would be available if a larg- er area around the periphery of the features had been exposed. All tools were point-plotted. Flakes were point-plotted by raw material types, then bagged by 1 m^ units. The gravel was not screened, so small implements such as burin spalls and retouch flakes may be somewhat under-represented. L-1 Excavation L-l was located on the lowest beach ridge, 10 m asl. (Figure 193). Because the 1993 survey produced an interesting surface collection and suggested the presence of structural remains, a 14 m^ area was excavated in 1 994, revealing a cluster of rocks asso- ciated with hearth remnants (Figures 194 and 195). The presumed structural feature consisted of a north-south oriented concentration of rocks covering an area of 3.5 by 2.0 m. A hearth area composed of three flat slabs and a charcoal concentration was identifiable at the northern end of the concentration. A small charcoal sample from the hearth was radio- carbon dated to 3750±60 B.P. (Beta-7761 1 ; ^ ^c/l ratio -26.1). A line of three rocks extending north- west from the hearth to a small boulder gave the impression of perimeter delineation, but it was oth- erwise difficult to discern an internal order in the associated scatter of rocks. The feature may have been disturbed, but it is possible that it was orient- ed differently from our expectations, or that it is not of the anticipated axial form. 274 CHAPTER 1 3 Tool Distribution Table 71. Attu's Point L-1: A total of 49 tools was recovered. Figure 196 shows Tool Classes by K-Means Cluster. the tool distribution and the results of a /c-means clus- ter analysis. A five cluster solution seemed best, defin- C-1 C-2 C 3 C 4 C 5 TOTAL N (%) ing a hearth-centered group (cluster 5), another im- Endblades 4 4 (8.2) mediately west of the hearth (cluster 3), a small cluster Endblade preforms 1 1 2 4 (8.2) near the western boundary of the excavation (cluster Bifaces 1 1 2 (4.1) 1), and two diffuse clusters on the southern side of the Burin spalls 1 1 5 4 11 (22.5) excavation (clusters 2 and 4). Table 71 compares the Microblades 3 2 4 1 3 1 3 (26.4) tool contents of these clusters. Although it is difficult Scrapers 1 1 2 (4.1) to draw inferences from such low frequencies there is Scraper/burins 1 1 (2.0) a contrast between the prevalence of burin spalls, Utilized flakes 2 1 6 1 10 (20.4) microblades and utilized flakes in cluster 3 adjacent to Adzes/celts 1 1 2 (4.1) the hearth, and the concentration of endblades in clus- TOTAL 13 5 17 5 9 49 ter 1 , 2 m west of the hearth. 1 93/ Attu's Point site map. ATTU'S POINT; A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY In the absence of a clearly defined axial feature, the spa- tial divisions used for the bilat- eral model-based tool distribu- tion analysis are association with the central rock feature, or placement on its east or west flanks (Table 72). The majority of tools (55.1%) lie in the vicinity of the rock feature. Although the frequencies are rather low, there are slight dif- ferences in where the tool types were discarded. Most of the endblades and preforms occur on the flanks of the rock feature, including a cluster of five implements deposited together on the west side (pos- sibly a cache), while the major- ity of the burin spalls, microb- lades and utilized flakes are closely associated with the fea- ture. Table 73 breaks down the tool distribution along a north- south division (front/back) running along the 6y grid line. In this case, nearly 80% of the tools lie on the northern (sea- ward) end of the rock feature, with endblades, burin spalls, microblades and utilized flakes being particularly well represented. 4x8v 6x8y N F Flat rock Q Charcoal 3x4y 6x4y /94/Afru's Point L-1 : rock feature and total flake distribution. 195/ Attu's Point L-1 : overview towards the east. 276 CHAPTER 1 3 3x4y 6x4y 1 96/ Attu's Point L-1 : tool distribution with k-means dusters. Table 72. Attu's Point L-1: Table 73. Attu's Point L-1: Tool Classes by Tool Classes by Bilateral Position. North-South (Front/Back) Position. Central (Axial) East West TOTAL N (%) North (Front) South (Back) TOTAL N (%) Endblades 4 4 (8.2) Endblades 4 4 (8.2) Endblade prefo rms 1 2 1 4 (8.2) Endblade preforms 2 2 4 (8.2) Bifaces 1 2 2 (4.1) Bifaces 1 1 2 (4.1) Burin spalls 9 1 1 1 1 (22.5) Burin spalls 10 1 1 1 (22.4) Microblades 7 3 3 1 3 (26.4) Microblades 10 3 1 3 (26.5) Scrapers 1 1 2 (4.1) Scrapers 1 1 2 (4.1) Scraper/burins 1 1 (2.0) Scraper/burins 1 1 (2.0) Utilized flakes 7 1 2 10 (20.4) Utilized flakes 9 1 10 (20.4) Celts/adzes 1 1 2 (4.1) Celts/adzes 1 1 2 (4.1) TOTAL 27 9 13 49 TOTAL 39 10 49 ATTU'S POINT: A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY 277 Tool raw material distribution patterns can be evaluated visually in Figure 197. Mugford chert tools tend to lie near the central rock feature, although they are also found on its flanks. Crystal quartz tools are clustered near the hearth (similar to the flake distribu- tion, see below). The other materials are too few to indicate a pattern, but the two implements of Ramah chert are located some distance away from the central rock feature. Quantitatively, the distributions can first be evaluated by their location relative to east-west bilaterality (Table 74). The only clear pattern is the exclusive association of crystal quartz with the rock feature. When broken down by the north-south (front/back) division (Table 75) there is little clear pat- terning other than the exclusive association of crystal quartz and gray-speckled chert with the northern (hearth-related) division. Tool raw materials can also be seen in relation to the /(-means clusters (Table 76), but the numbers are too low for the comparison to be worth much. Table 74. Attu's Point L-1: Tool Raw Material Frequencies by Bilateral Position. Central (Axial) East West TOTAL N (%) Dark gray Mugford chert 1 2 2 4 18(36.7) Light gray Mugford chert 6 4 6 16(32.7) Cray-speckled chert 2 1 3 (6.1) Black chert 1 1 (2.0) Ramah chert 1 1 2 (4.1) Crystal quartz 6 6(12.3) Gray silicified slate 1 1 (2.0) Slate 1 1 2 (4.1) TOTAL 27 9 13 49 278 CHAPTER 1 3 4x8y 6x8y 3x4y 3x4y o oo. 0^ Mugford chert flakes (all variants) 4x8y 6x4y 6x8y 3x4y 6 ^ oo 6x4y 3x4y Black chert flakes 1 98/ Attu's Point L-1 flake raw material distributions. 3x4y r^/—? I ! Patinated chert flakes 4x8y Crystal quartz flakes 4x8y 6x8y O oo 'O u Ramah chert flakes ATTU'S POINT: A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY 279 Table 75. Attu's Point L-1: Tool Raw Materials by North-South (Front-Back) Position. North South TOTAL N (%) Dark gray Mugford chert 1 5 3 18 (36.7) Light gray Mugford chert 12 4 16 (32.7) Cray-speckled chert 3 3 (6.1) Black chert 1 1 (2.0) Ramah chert 1 1 2 (4.1) Crystal quartz 6 6 (12.3) Cray silicified slate 1 1 (2.0) Slate 1 1 2 (4.1) TOTAL 39 10 49 Flake Distribution The total flake distribution (Figure 194) parallels the tool groupings identified by the cluster analysis (Figure 196). There are concentrations on either side of the hearth (similar to tool clusters 3 and 5), a tight concen- tration west of the hearth (cluster 1 ) and a diffuse con- centration in the southeast corner of the excavation (cluster 2). Figure 198 presents point-plots for each flake raw material type while Figure 199 displays quadrat counts. The Mugford and patinated chert distri- butions are very similar, which is not surprising since the latter is probably a weathered version of Mugford chert. They are spread in a roughly linear east-west dis- tribution centered on the hearth area at the northern end of the rock feature. The marked concentration of Mugford chert on the western side of the excavation lies only a few centimeters distant from the tight tool concentration of cluster 1, which included a cache of four endblades and an endblade preform (all of Mugford chert), two microblades, a scraper and two uti- lized flakes. In the absence of refitting data, one might speculate as to the relation between the Mugford flake concentration and production of the endblades. The distributions of slate and Mugford chert are almost mutually exclusive. Slate clusters in the south- east corner of the excavation (with over 70% in a single unit), although a few flakes are present near the cen- ter. A slate adze bit was found 1 m west of the slate concentration. Mugford cherts have a more diffuse dis- tribution, but are mostly associated with the hearth area and the northwestern portion of the excavation. This placement broadly parallels the tool distribution. Crystal quartz occurs in a small cluster immediately west of the hearth, virtually in the center of the crystal quartz tool distribution, which was also hearth related (Figure 197). Black chert has a diffuse distribution, but occurs almost exclusively west of the rock feature; the single tool of this material lay west of the rock feature (Figure 197). The two Ramah chert flakes are found towards the north end of the rock feature. Table 77 outlines the distribution of flake raw materials according to their east-west bilateral posi- tion, tabulated by Im^ quadrats (unit 6x5y missing). Overall, the flakes are dominated by Mugford chert and slate; 61 .1% of the debitage was deposited on the west side of the structure, consistent with, though not as marked, as the tool distribution. A chi-square test (x^ = 85.80) indicates a statistically significant differ- ence between the two areas at the .05 level of signifi- cance. This difference is based on a marked tendency for light gray Mugford chert and crystal quartz to be found on the west side, while slate is associated prima- rily with the east side. Variation can also be seen in the north-south (front/back) direction, with a not entirely artificial division imposed along the 6y line. Table 78 indicates that 64.6% of the debitage is found on the north side of the rock feature. A chi-square test ("2 = 103.59) indicates a statistically significant difference between the two areas. The pattern is linked to the high frequency of light gray Mugford chert (and to a much lesser extent, patinated chert and crystal quartz) on the north side versus a preponderance of slate on the south. Integration The distribution of Mugford chert and crystal quartz tools and flakes is largely coterminus, while the slate, black chert and Ramah chert implements are found 280 CHAPTER 1 3 11 3 Black chert flakes 4 2 2 5 2 5 14 16 1 5 10 33 Light gray Mugford chert flakes 41 4 6 3 2 1 2 2 14 Crystal quartz flakes 6 2 6 6 1 4 7 Patinated chert flakes 1 1 3 Dark gray Mugford chert flakes Ramah chert flakes Slate flakes Gray speckled Mugford chert flakes 1 99/ Attu's Point L-1 : flake raw materials by quadrat. outside the distributions of their debitage. An attempt to refit burin spalls to the single scraper/burin was unsuccessful. While the spalls mostly cluster towards the north end of the rock fea- ture, generally associated with the hearth, the buri- nated implement lies on the western periphery of the excavation. Table 79 tabulates tool and flake raw materials by weight for L-1 as a whole. Slate is the dominant material (54.2%), with gray Mugford cherts ranked second (22.6%). Several observations can be made if Table 79 is used in conjunction with Tables 74 and 77 (tool and flake frequencies). The high proportion of dark gray Mugford chert among the tools by both weight and frequency is not parallelled by a high pro- portion of such flakes; in fact, light Mugford chert dominates the debitage. This suggests most of the tools made of dark gray Mugford were manufactured elsewhere (either at Attu's Point or another site) and curated into the locus. The opposite pattern is seen with patinated chert; no tools were present but a modest amount of debitage was, indicating the ATTU'S POINT A PRE DORSET "CENTRAL PLACE" ON WEBB BAY 281 Table 76. Attu's Point L-1: Tool Raw Material Frequencies by K-Means Cluster. V. 1 C-2 Dark gray Mugford chert 4 2 6 1 5 18 Light gray Mugford chert 6 2 5 2 1 16 Cray-speckled chert 1 2 3 Black chert 1 1 Ramah chert 1 1 2 Crystal quartz 3 3 6 Cray silicified slate 1 1 Slate 1 1 2 TOTAL 13 5 17 5 9 49 Table 77. Attu's Point L-1 Flake Raw Material Frequencies by East-West Bilateral Position. East West TOTAL N (%) Dark gray Mugford chert 7 (5.9) 4 (3.6) 1 1 (4.8) Light gray Mugford chert 33 (27.9) 66 (59.5) 99 (43.2) Cray-speckled chert 1 (0.8) 4 (3.6) 5 (2.2) Patinated chert 14 (11 .9) 18 (16.2) 32 (13.9) Black chert 5 (4.3) 5 (2.2) Ramah chert 2 (1.8) 2 (0.9) Crystal quartz 4 (3.4) 14 (12.6) 18 (6.8) Slate 54 (45.8) 3 (2.7) 57 (25.6) TOTAL 94 140 229 removal of tools from the locus. Ramah chert is bet- ter represented by tools than by flakes. Having presented the lithic distribution patterns along several dimensions, the overall spatial configura- tion at L-1 can be evaluated. The initial assumption was that the central rock feature could indicate a dis- turbed axial structure running roughly north-south, oblique to the beach front. Inspection of the individual distributions in relation to the feature suggests an alternative interpretation. If the more numerous item classes are considered (e.g., microblades, utilized flakes, Mugford chert tools, crystal quartz tools and flakes), there is a strong tendency for them to display a roughly northeast-west linear distribution ranging from the hearth to the edge of the excavation on the 7x line. The clustering of slate flakes and some microblades in the southeast corner of the excavation is the primary deviation from this pattern. In light of these tendencies one can postulate an axial feature running parallel with the beach front and consisting of a hearth at its northeastern end and a small, roughly rectangular feature 1 by 2 m in size that extends west- wards. The presence of a similar configuration at L-5 (see below) might support this interpretation. It is also instructive to compare the small concentration of Mugford chert flakes in the northwest corner of the L- 282 CHAPTER 1 3 Table 78. Attu's Point L-1: Flake Raw Material Frequencies by North-South (Front-Back) Position North (Front) South (Back) TOTAL N (%) Dark gray Mugford chert b (4.U) 5 (6.2) 1 1 (4.8) Light gray Mugford chert OA tec O \o4 (5b. o) 15 (18.5) 99 (43.2) Cray-speckled chert C to A\5 (3.4) 5 (2.2) Patinated chert -) C / 1 "7 C \(1 /.d) 6 (7.4) 32 (1 3.9) Black chert i U.O) 2 (2.5) 5 (2.2) Ramah chert 2 (1.3) 2 (0.9) Crystal quartz 16 (10.8) 2 (2.5) 18 (6.8) Slate 6 (4.1) 51 (62.9) 57 (25.6) TOTAL 148 81 229 Table 79. Attu's Point L-1: Total Lithic Raw Material Weights (Tools and Flakes). In Grams. Tools g (%) Flakes g (%) TOTAL g (%) Dark gray Mugford chert 32.5 (10.8) 16.1 (9.0) 48.6 (10.1) Light gray Mugford chert 21.3 (7.1) 38.6 (21.6) 59.9 (12.5) Cray-speckled chert 19.1 (6.4) 4.9 (2.7) 24.0 (5.0) Patinated chert 13.2 (7.4) 13.2 (2.8) Black chert 1.0 (0.3) 2.1 (1.2) 3.1 (0.6) Ramah chert 41.7 (1 3.9) 0.8 (0.4) 42.5 (8.9) Crystal quartz 1 1 .7 (3.9) 16.6 (9.3) 28.3 (5.9) Slate 173.0 (57.6) 86.8 (48.5) 259.8 (54.2) 1 excavation— which would lie at the western end of a beach-parallel axial feature— with the concentration of Mugford chert flakes at the foot of the axial structure at L-7 (see below). In any event, it appears that the L- 1 hearth is located at the terminal end of a feature rather than in the middle, as expected from a classic axial structure. Another possibility is that the scatter of rocks in the southwestern portion of the excavation may represent a second overlapping feature. L-1 Tool Assemblage Endblades: Three of the four endblades are complete, one is a distal fragment (Figure 200:a-d). Three are made of dark gray Mugford chert, one of light gray Mugford chert. The complete specimens are triangular and two have straight bases and biconvex cross-sec- tions (L= 25.1, 16.1, 21.4 mm, W= 9.5, 7.7, 9.2, TH= 2.2, 1.5, 1.9). One implement has a plano-convex cross-section and a concave base (depth= 1 .3 mm); its dorsal face is completely retouched, but ventrally only the edge margins are retouched (Figure 200:d). The lateral edges of the latter specimen are distinctly ser- rated; the other three endblades also have slightly ser- rated edges. Endblade preforms: Of the four endblade pre- forms (Figure 200:e-h), one is complete, two are prox- imal fragments and one a distal fragment. Two are made of light gray Mugford chert, one of dark gray ATTU'S POINT A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY 283 f g h 200/ Attu's Point L-1 tools, a-d endblades; e-h: endblade preforms. Mugford chert and one of crystal quartz. Two of the three specimens with a proximal portion exhibit straight bases while the third is straight-based with an eared basal corner (Figure 200:f). The complete imple- ment of crystal quartz (Figure 200:e) has near-final retouch completed from its tip to its medial region and an unfinished basal portion, indicating a distal to prox- imal flaking sequence (L= 23.8 mm, W= 1 6.1 , TH= 8.6). The two proximal fragments differ somewhat in width (W= 1 3.4, 1 9.9 mm, TH= 2.8, 3.4). One of these frag- ments (Figure 200:g) is flaked mostly on its dorsal face, upon which two long parallel basal-thinning flakes were removed medially. Bifaces: Two biface fragments were recovered, one made of Ramah chert, one of crystal quartz. The Ramah chert specimen (Figure 201 :a) is the proximal end of a biface that originally must have been ca. 10 cm long (W= 39.3 mm, TH= 6.9). There is a remnant striking platform near the base, which seems less fin- ished than the rest of the implement. The crystal quartz specimen is a small lateral edge fragment. Combination scraper/burins: The single speci- men of a combination scraper/burin (Figure 201 :c) is made on a blade-like flake of dark gray Mugford chert (L= 70.5 mm, W= 20.6, TH= 8.1). It consists of a con- d f ^!!!B5il^S!Siiiai^!? CM. O I £ J •? 5 201 / Attu's Point L-1 tools, a biface, b sidescraper; c: burm/sidescraper , d-e microblades. 202/ Attu's Point L-1 tools, a: adze; b: celt/adze preform or proximal fragment. cave scraper running distally-medially on one lateral edge and a burination plane along the other lateral edge, with removals from a platform on the slightly convex, unifacially retouched distal end. The scraper 284 CHAPTER 1 3 4x7y 7x7y 2x6y ) 0' o o o ,^0 o — 1 — 2x3y F Flat rock ® Charcoal 203/ Attu's Point L-5: features and total flake distribution. edge angle is 50 and there are three burin spall removals. Slightly below the mid-portion of each later- al edge there is a faint shoulder, which creates a slight stem effect proximally. A striking platform is present proximally and both the platform and bulb of percus- sion are highly polished, presumably by hafting wear. There is no facial grinding. The middle of the scraper 204/ Attu's Point L-5: possible axial structure, view towards the southwest. edge and the distal end of the tool exhibit wear polish; the first burination scar has pol- ished edges and small fractures while the third burination scar has clear striations on the adja- cent ventral surface, perpendi- cular to the scar edge. Scrapers: Two implements were collected. A complete sidescraper (Figure 201 :b) is made of dark gray Mugford chert and has a slightly concave working edge (L= 43.8 mm, W= 23.9, TH= 8.2). The tool is dor- sally retouched on all lateral edges while on the ventral side the proximal end has been flaked to thin out the bulb of percussion, although the striking platform remains at the base. The concave working edge and distal end are distinctly polished by wear. Seen alongside the combi- nation scraper/burin (Figure 201 ;c), it is likely that the sidescraper is a reworked and perhaps exhausted ver- sion of the latter, since both have very similar basal forms and dimensions. Rather than representing dif- ferent "types," they may indicate start and end points in a reduc- tion sequence. The second implement is an edge fragment made of black chert. Burin spalls: Of the 1 1 burin spalls (MNI=10), seven were complete, three were proximal fragments and one a distal fragment. All were made of gray Mugford chert vari- ants. None bore traces of grinding and all were second- ary spalls. Their metrics can be summarized as follows: ATTU'S POINT A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY 285 xL = 1 5.9, r = 9.2-21 .8, xW = 3.3, r= 2.2-4.3, xTH = 1 .7, r = 1 .1 -2.4. Microblades: Of the 1 3 microblades, three are complete, six are medial fragments, three are proximal fragments and one is a distal portion. Based on proxi- mal elements, MNI=1 0. Nine are made of gray Mugford cherts and four of crystal quartz. Two specimens (Figure 201:d,f) are large enough to be considered "macroblades" (L = 61.0, 56.7 mm, W = 17.7, 20.9, TH = 5.6, 4.8). Metrics for the others are as follows: xW = 9.4 mm, r = 6.1-15.0, xTH = 2.1, r = 1.4-2.9. Platform angles: 80* = 3, 75 = 1, 70' = 1. Platform preparation: retouched= 2, retouched/ground = 1, plain = 2. Eight specimens exhibit use-wear or light retouch. Adzes/Celts: The two specimens are fragmentary and both are made of green slate. The first is the dis- tal portion of an adze (Figure 202:a); one lateral edge is bifacially retouched while the other is retouched abruptly from the dorsal (top) side of the implement, resulting in an asymmetrical cross-section. A consider- able amount of cortex is still present on the dorsal sur- face. The dorsal edge of the bit is only polished 1-3 mm back from the edge, while the ventral polishing is mostly limited to an area within 7.5 mm of the bit edge, although a few facets further back are also pol- ished. Use-wear consists of a slight rounding of the bit edge, although some rather large step fractures on one ventral corner may indicate heavier work pressure or retouch. L = >64 mm, W = 41.8, TH = 16.2, Edge Angle = 70'. The second specimen (Figure 202: b) is the prox- imal end of an adze or celt. It is made on a rectangu- lar slate tablet, the lateral edges of which are retouched with small flakes removed at 90' from both the dorsal and ventral surfaces of the tablet such that the squared edge and rectangular cross-section is maintained. The flake facets on what appears to be 286 CHAPTER 1 3 Table 80. Attu's Point L-5: Tool Classes by K Means Cluster C-1 N (%) C-2 N (%) C-3 N (%) C-4 N (%) TOTAL N (%) Endblades 1 (6.7) 3 (8.4) 4 (5.6) Endblade preforms 1 (2.8) 1 (1 .4) Stemmed points 1 (14.3) 1 (1 .4) Biface fragments 1 (7.7) 3 (8.4) 4 (5.6) Burins 1 (6.7) 2 (5.6) 3 (4.2) Burin spalls 2 (1 3.3) 2 (1 5.4) 9 (25.0) 2 (28.5) 1 5 (21.1) Microblades 4 (26.6) 3 (23.1) 12 (33.3) 1 (14.3) 20 (28.1 ) Microblade cores 1 (6.7) 1 (2.8) 2 (2.8) Scrapers 2 (1 5.4) 1 (14.3) 3 (4.2) Scraper/burins 1 (14.3) 1 (1.4) Utilized flakes 6 (40.0) 4 (30.7) 5 (13.9) 1 (14.3) 16 (22.4) Celts/adzes 1 (7.7) 1 (2.8) 2 (2.8) TOTAL 15 13 36 7 71 the dorsal surface are polished slightly and one later- al edge is slightly rounded. L = > 62 mm, W = 33.7, TH = 16.7. Utilized Flakes: Eight of the ten utilized flakes are made of gray Mugford chert, one of black chert and one of gray-speckled chert. L-5 EXCAVATION L-5 was located on the lowest beach ridge at 1 m asl. (Figure 193). Initially, L-5 appeared to contain the remains of a well-preserved axial structure, so an exca- vation of 21.5 m^ was opened up. Unfortunately, the structure turned out to be less distinct than anticipat- ed (Figures 203 and 204). Towards the southern end of the excavation was a hearth composed of a few flat slabs, a small scatter of charcoal and a number of small rocks that may have been thermoliths. What appeared to be the remnants of an axial structure extended out from the hearth, suggesting a feature 2 m long, 0.75-1.0 m wide, oriented N-S, parallel with the shoreline, but with the hearth placed at one end of the axial structure rather than in the middle (cf., L-1). A small charcoal sample procured from the hearth area was radiocarbon dated to 3790±70 B.P. (Beta-77612; ]3q/'\2q rgfiQ -24.7). To the north of the axial struc- ture was a cluster of rocks associated with a charcoal concentration and a substantial quantity of tools and debitage. This area may represent an outdoor dump or activity area at the entrance to the axial structure or it could be the remains of a second, badly disturbed axial structure with a NW-SE orientation. Finally, in the southwest corner of the excavation and just outside the excavation limits was another cluster of rocks, including a few flat slabs. Although there were no associated lithics, the rock cluster might indicate the presence of another feature. Tool Distribution A total of 82 tools was recovered; some were scattered around the postulated axial feature, but most (68.7%) ^ Items excluded from the analysis due to lack of point provenience are displayed by quadrats in Figure 206. ATTU'S POINT: A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY Table 81. Attu's Point L-5: Tool Classes by General Spatial Position. AXiai In (/o) iNOrin In (^; 1 U 1 AL N Endblades AH (b. ! } 4 (4.y) Endblade preforms 1 ( 1 .3) 11 ( 1 -*^) Endblades/points TL (3.U) c. KlA) Stemmed points 11 (1.3) 11 Biface fragments 3 //I c ^ (4.b) i (3.7) Burins (4.5) 3 (3.7) Burin spalls d. \\ l.^) 1 c (ZZ.O) 1 / (ZU. /) Microblades 3 ( 1 O.O) 1 D 1 o1 y / 3 3 3 \(^3.2) Microblade cores ~%L. (3.U) z /3 /l A Scrapers ( 1 0.0) 3 (3./) Scraper/burins 1 (6.2) 1 (1.2) Utilized flakes 6 (37.5) 18 (27.3) 24 (29.3) Celts/adzes 1 (6.2) 1 (1.5) 2 (2.4) TOTAL 16 66 82 Table 82. Attu's Point L-5: Tool Raw Material Frequencies by General Spatial Position. Axial N (%) North N (%)TOTAL N (%) Dark gray Mugford chert 6 (37.6) 30 (45.5) 36 (43.9) Light gray Mugford chert 3 (4.5) 3 (3.7) Light gray silicified slate 2 (12.5) 6 (9.2) 8 (9.8) Patinated chert 2 (3.0) 2 (2.4) Gray-speckled chert 1 (6.2) 5 (7.6) 6 (7.3) Crystal Quartz 2 (12.5) 8 (12.1) 10 (12.2) Ramah chert 3 (18.8) 8 (12.1) 1 1 (13.4) Black chert 1 (6.2) 3 (4.5) 4 (4.9) Slate 1 (6.2) 1 (1.5) 2 (2.4) TOTAL 16 66 82 were located to the north in the possible dump/activi- ty area. A bilateral analysis could not be undertaken given the paucity of tools associated with the axial fea- ture. Instead, k-means cluster analysis was used to define groupings based on proximity rather than model-derived spatial categories. The analysis used a total of 71 tools for which precise point provenience was available; a four cluster solution seemed most appropriate (Figure 205). One of these clusters (cluster 2) is centered on the hearth area of the axial structure while two others (clusters 1 and 3) partition the dump/activity area into two groups. Cluster 4 is a dif- 288 CHAPTER 1 3 Table 83. Attu's Point L-5: Tool Raw Material Frequencies by K-Means Cluster. C-1 N (%) C-2 N (%) C-3 N (%) C-4 N (%) TOTAL N (%) Dark gray Mugford chert 6 (40.0) 3 (23.1) 24 (67.7) 2 (28.6) 35 (49.3) Light gray Mugford chert 1 (6.7) 2 (5.5) 3 (4.2) Light gray siliclfied slate 2 (13.3) 2 (1 5.4) 4 (5.6) Patinated chert 1 (2.8) 1 (14.3) 2 (2.8) Cray-speckled 3 (20.0) 1 (7.7) 1 (2.8) 1 (14.3) 6 (8.5) Crystal quartz 2 (13.3) 2 (1 5.4) 3 (42.8) 7 (9.9) Ramah chert 1 (6.7) 3 (23.1) 5 (13.9) 9 (12.7) Black chert 1 (7.7) 2 (5.5) 3 (4.2) Slate 1 (7.7) 1 (2.8) 2 (2.8) TOTAL 1 5 13 36 7 71 fuse distribution of questionable significance immedi- ately northwest of the axial feature. Table 80 displays the variation in tool classes between the clusters. Cluster 3, immediately north of the presumed axial fea- ture, has by far the largest proportion of the tool col- lection (N = 37, 51 .4%). It also has a tendency towards more endblades, perhaps burins and burin spalls, as well as microblades and microblade cores. Cluster 2 UF=4 UF= 1 UF= 1 UF= 1 EB= 1 UF= 1 Bl= 1 BS= 1 Utilized flakes (UP) Endblades (EB) Bifaces (Bl) Burin spalls (BS) 206/ Attu's Point L-S no provenience tools by quadrat. ATTU'S POINT: A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY (associated with the hearth) and cluster 1 exhibit some- what higher proportions of utilized flakes. A broader spatial comparison can be made between the frequency of tool types associated with the presumed axial feature and the concentration to the north of the feature using a dividing line running along 4.30x. Table 81 shows clear differences between the two subareas. Endblades, points, bifaces, burins and microblade cores are exclusively associat- ed with the northern subarea, while burin spalls and microblades are predominantly to the north. Scrapers and the combination scraper/burin are exclusive to the southern area associated with the axial feature. The relationship is statistically significant at the .05 level (/^= 22.49). The first step in evaluating variation in tool raw materials is to consider their distribution relative to the axial feature/northern subarea division used in the tool type analysis (Table 82). There is no significant difference between the two areas (-/^= 3.26). The dis- tribution of tool raw materials can also be compared across the cluster groups (Table 83). It is difficult to evaluate the low frequencies, but higher proportions of dark gray chert occur in clusters 1 and 3 and crystal quartz is absent from cluster 1 . Point distributions of tool raw materials are shown in Figure 207. 289 4x7y 7x7y 2x6y * A ^ ^ A 2x3y o B Black chert Crystal quartz A Dark gray Mugford chert + Gray speckled Mugford chert • Light gray Mugford chert Light gray silicified slate Patinated chert * Ramah chert Slate 207/ Attu's Point L-5 tool raw material distribution. Table 84. Attu's Point L-5: Flake Raw Material Frequencies by Analytical Quadrant. NW N (%) NE N (%) SW N (%) SE N (%) TOTAL N (%) Dark gray Mugford 9 (10.7) 21 (12.0) 4 (13.8) 3 (10.7) 37 (1 1.7) Light gray Mugford 1 (1.2) 5 (2.9) 2 (6.9) 2 (7.1) 10 (3.2) Gray-speckled chert 10 (5.7) 1 (3.6) 1 1 (3.5) Black chert 2 (1.1) 1 (3.4) 1 (3.6) 4 (1.3) Ramah chert 5 (6.0) 30 (17.1) 2 (6.9) 7 (25.0) 44 (13.9) Patinated chert 1 (1.2) 12 (6.9) 1 (3.4) 14 (4.4) Biostrome chert^ ^ 1 (1.2) 1 (0.3) Crystal quartz 1 (3.4) 3 (10.7) 4 (1.3) Slate 17 (20.2) 78 (44.6) 6 (20.7) 8 (28.6) 109 (34.5) Clastic slate^ ^ 43 (51.2) 1 7 (9.7) 8 (27.6) 2 (7.1) 70 (22.2) Light gray silicified slate 7 (8.3) 4 (13.8) 1 (3.6) 12 (3.8) TOTAL 84 1 75 29 28 316 ^ 1 Biostrome chert denotes possible stromatolite structure. ^ ^ Clastic slate denotes a material containing small rounded clast inclusions. 290 CHAPTER 1 3 oh © o O Mugford chert flakes (all variants) if o Black chert flakes 7x7y o oo'- o G7 o o' Go p. Q o o o -JO 30 • Patinated chert flakes Crystal quartz flakes o o .O O <3 o Ramah chert flakes o 2 Slate flakes 208/ Attu's Point L-5: flake raw material distributions. ATTU'S POINT: A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY 291 Table 85. Attu's Point L-5: Total Lithic Raw Material Weights (Tools and Flakes). In Grams. TOOLS g (%) FLAKES g (%) TOTAL g (%) Dark gray Mugford 69.8 (26.8) 1 9.9 (7.8) 89.7 (1 7.4) Light gray Mugford 8.0 (3.1) 8.5 (3.3) 16.5 (3.2) Cray-speckled chert 2.3 (0.9) 5.5 (2.2) 7.8 (1.5) Black chert 5.5 (2.1) 1.5 (0.6) 7.0 (1.4) Ramah chert 1 3.1 (5.0) 1 7.7 (7.0) 30.8 (6.0) Patlnated chert 0.9 (0.3) 8.5 (3.3) 9.4 (1.8) Blostrome chert 0.4 (0.2) 0.4 (0.1) Crystal quartz 7.2 (2.8) 1.1 (0.4) 8.3 (1.6) Slate 1 10.0 (42.2) 1 1 2.5 (44.2) 222.5 (43.2) Clastic slate 72.5 (28.5) 72.5 (14.1) Light gray silicified slate 43.9 (16.8) 6.2 (2.4) 50.1 (9.7) No refits of broken tools were identified and attempts to refit burin spalls with the three burins were unsuccessful. The majority of burins and burin spalls lie clustered together on the north side of the excavation (Figure 205). Two of the three burins are made of dark gray Mugford chert, as are 1 3 of the 1 6 spalls, while one burin is made of patinated chert, as is one spall. A single spall of gray-speckled chert lacks a burin of the same material. Flake Distribution The total flake distribution is shown in Figure 203. The vast majority of flakes (81%) are concentrated on the northern side of the excavation. This concentration actually consists of two smaller elongated clusters, one associated with a small charcoal concentration, the other running parallel ca. 50 cm to the east. The axial feature has relatively few flakes, but these seem to be concentrated in two clusters; one on the southeast edge of the hearth, the other at the northwest corner of the axial alignment in the vicinity of three flat slabs. In sum, the overall debitage distribution tends to par- allel the /(-means clusters of tools. The individual flake raw material distributions are shown as point plots and quadrat counts in Figures 208 and 209, respectively. Ceneric Mugford chert, patinated chert and Ramah chert tend to be found on the east side of the northern concentration, while generic slate is found on both sides. Black chert occurs mostly in two small concen- trations, one at the northwest corner of the excavation, the other at the northwest corner of the axial feature. There are few crystal quartz flakes, but two are locat- ed close to the hearth associated with the presumed axial feature (cf., the hearth-related crystal quartz dis- tribution at L-1). Ramah chert occurs in two clusters, one on the east side of the northern concentration, the other adjacent to the axial feature hearth. Given the strong spatial clustering, a quantitative analysis of the debitage was conducted using four quadrants, distinguished along an east-west axis by the 5y line and along the north-south axis by the 4x line (Table 84). These quadrants separate the two clusters of the northern concentration and distinguish the latter from the axial feature, and they are general- ly consistent with the results of the /c-means analysis of the tool assemblage. There are several clear ten- dencies. Most of the Ramah chert is found on the east side of the excavation, particularly in the northeast quadrant. Slate is mostly distributed on the north side, also primarily in the northeast quadrant. Gray-speck- 292 CHAPTER 1 3 2 1 7 14 1 11 1 B= 1 1 2 3 1 C= 1 6 Black chert Crystal quartz (C) Blostrome chert (B) Dark gray Mugford chert 1 4 2 5 5 1 5 7 1 1 1 1 1 Light gray Mugford chert Gray speckled Mugford chert Patinated chert 1 1 29 1 5 14 64 1 7 6 11 2 1 5 1 10 2 2 4 5 26 1 1 5 9 3 7 Ramah chert Slate Clastic slate 1 2 5 1 3 Silicified slate 209/ Attu's Point L-5: flake raw material distributions by quadrat. led Mugford chert and patinated chert are exclusively and primarily (respectively) associated with the north- east quadrant. Another noticeable pattern is the ten- dency for "plain" slates to cluster in the northeast quadrant, while "clastic" slate is mostly found in the northwest quadrant. The latter pattern suggests in situ distributions resulting from two separate knap- ping episodes using slightly different raw materials rather than dumping behavior resulting from floor ATTU'S POINT: A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY 293 a b c d e 2 1 0/ Attu's Point L-5 tools, a-d endblades. e: stemmed point, f: endblade preform, g-i burms, j, k endscrapers. I: burm/endscraper. cleaning, which would probably mix the two slate types more thoroughly. Integration Comparison of the tool and flake raw material plots (Figures 207 and 208) indicates the distribution of Mugford chert flakes and tools largely coincides, except for a small concentration of tools in the extreme northwest corner of the excavation. Ramah chert tools and debitage tend to be spatially associat- ed and are fairly tightly clustered around the hearth area and the northeast portion of the external dump/activity area. Both black chert and crystal quartz have a somewhat different distribution pattern than the other materials, tending to be found towards the more westerly portion of the excavation on the periph- ery of the other materials' main distributions. If we compare the frequencies of raw materials present in the tool assemblage to those among the flakes (Tables 82 and 84) there are some marked dif- ferences. The high percentage of dark gray Mugford chert tools contrasts with a relatively low percentage of flakes, suggesting that some of the tools entered the locus in finished form. The frequency of crystal quartz debitage also seems low relative to the quantity of tools, suggesting the same pattern. The frequency of slate debitage is high compared to the few tools, which might indicate that some of the finished products or preforms were removed from the locus. This is under- scored by the presence of clastic slate debitage, but the absence of tools of this material. Table 85 tabulates tool and flake raw materials by weight for L-5 as a whole. All materials occur in low weights, but slates dominate, followed by dark gray Mugford chert. The most striking variations are that dark gray Mugford chert and light gray silicified slate are more significant as tools than as flakes; the former is consistent with the frequency data. The initial assessment of L-5 suggested the local- ity consisted of an axial feature with a hearth at one end and an activity area to the north, or that perhaps the latter was a disturbed structure. Support for the disturbed structure alternative might be seen in the clustering of the flake distributions, one concentration related to a charcoal deposit and rocks, the other con- centration lying 50 cm to the east (Figure 203), hinting at the bilaterality sometimes associated with axial fea- tures. On the other hand, a similar pattern could also be associated with an isolated hearth without an accompanying mid-passage structure. But there may be a third alternative. We might deconstruct the axial structure interpretation by pointing out that the fea- ture consists of a hearth on the south side connected by a single line of rocks to a cluster of rocks and flat slabs on the north side. These two components could be separate features, with the small charcoal patch at the northern cluster of flat slabs representing a second disturbed hearth. Seen in conjunction with the north- ern activity area, we might then have a linear series of three hearths spaced at 1.5 m intervals. Such a linear 294 CHAPTER 1 3 feature, possibly consisting of four hearths, was exca- vated at Port IVlanvers Run-1 (Chapter 12). L-5 Tool Assemblage Endblades: Of the four endblades from L-5 (Figure 210:a-d), one is nearly complete and the others are proximal, medial and distal fragments. One is made of dark gray Mugford chert, one of light gray Mugford chert, one of light gray silicified slate and one of black chert. The nearly complete example (Figure 210:a) is triangular with a biconvex cross-section and a straight base that is thinned on both sides by two parallel flakes placed medially. It is finely flaked with distinctly serrated lateral edges and there is a small area of medial grinding on one face (L= >17.7 mm, W= 10.2, TH= 1.4). The proximal example (Figure 210:b) has a slightly concave base (depth= 0.6 mm) with two paral- lel basal-thinning flakes on one side and one central basal-thinning flake on the other side. A striking plat- form remains on one lateral edge. The implement may be unfinished (W= 14.6 mm, TH= 2.7). The other two examples are small fragments deemed to be end- blades by virtue of their size and flaking patterns. Pomts/endblades: Two distal fragments are diffi- cult to classify as points or endblades. One is made of Ramah chert (Figure 210:d) and has distinct edge ser- ration and a plano-convex cross-section resulting from near-complete dorsal flaking combined with ventral retouch limited to the edge margins. The other is made of dark gray Mugford chert and has fine edge serra- tion, slight medial grinding, a biconvex cross-section and a slight hint of a shoulder (Figure 210:c). Combined with its slightly greater thickness, these attributes suggest this specimen may represent a point rather than an endblade. Endblade preforms: The single example is a dis- tal fragment made of gray-speckled Mugford chert (Figure 210 0. It bears minimal lateral retouch limited to within 2.5 mm of the edge. Points: The single definite specimen (Figure 210;e) is the proximal/lateral portion of a crystal 21 1/ Attu's Point L-5 tools, a, b: microblade cores; c-j: microblades: /c adze, I: celt/adze preform. quartz point with a contracting stem and a biconvex cross-section (TH= 5.7 mm, stem L= 1 1 .4). Bifaces: Three small biface fragments were found, all lateral edge fragments, two are made of crystal quartz, one of Ramah chert. The latter has a very slight hint of a shoulder, suggesting it was a lanceolate biface. Burins: Two of the three burins are fashioned from flakes, one is made on a microblade. Both of the flake burins are made of gray Mugford chert while the microblade burin is made of patinated chert. None of the implements are ground. One of the flake burins is relatively large (Figure 210:g) with fine retouch on both lateral edges, dorsal on the left, ventral on the ATTU'S POINT: A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY 295 right (L = 37.6 mm, W = 24.1, TH = 4.8). There is a faint shoulder on each side and a slight notch-like indentation on the right side, where the retouch is bifa- cial. The proximal end is thinned bifacially but two long parallel flakes were used to remove the bulb of percussion on the ventral side (similar to the technique used on some endblades). The right distal corner is bifacially retouched, providing a platform for the oblique removal of four burin spalls. The other flake burin (Figure 210:h) exhibits minimal unifacial retouch on its lateral edges (L= 25.4 mm, W=14.9, TH=6.7). A single spall removal runs obliquely from a slightly retouched area on the distal end to a terminus marked by a steeply retouched notch (65 ) on the opposite lat- eral edge. The notch might be either a remnant con- cave scraper edge or a termination control feature. The retouched edge is wear-polished. The third burin (Figure 21 0;i) is a very small imple- ment made on the proximal fragment of a microblade (L= 1 2.6 mm, W= 1 1 .8, TH= 2.8). The left distal corner has a small notch that may have been made for a con- trolled snap of the microblade, but the notch also served as a platform for the oblique removal of four tiny burin spalls. No other retouch is present. Combination scraper/bunns The single imple- ment (Figure 210:1) is made on a lozenge-shaped flake of crystal quartz (L= 35.6, W= 22.4, TH= 6.5). The medial-distal portion of one lateral edge has abrupt scraper-like retouch on the dorsal side while on the other lateral edge the medial-proximal portion is retouched similarly on the dorsal side. Burinations are present at both ends of the implement; one end has two spall removals— the first an angle burination, the second oblique across the tip of the implement— while the other end has a single angle burination. Burin spalls: Of the 17 burin spalls (MNI = 13), eight are complete, five are proximal fragments, three are distal fragments and one is a medial portion. Eight are made from dark gray Mugford chert, five from gray Mugford chert, one from light gray Mugford chert, two from light gray silicified slate and one from Ramah chert (the latter is an uncertain spall). Two spalls are primary, the rest are secondary. None are ground. Metrics are as follows; x L = 16.0 mm, r = 9.6-20.5, X W = 3.8, r = 2.4-5.2, x TH = 1 .9, r = 1 . 1 -3.2. Scrapers: Three scrapers were recovered. One is the distal fragment of an endscraper (Figure 210:k) made on a blade of light gray silicified slate, with a rel- atively narrow convex "nose" (W = 1 4.9 mm, TH = 3.8, edge angle = 40°). Another is an end/sldescraper (Figure 2 1 0;j) made on an irregular linear flake of dark gray Mugford chert. The partially retouched distal end is convex and half of one lateral margin is also retouched (L = 33.7 mm, W = 13.5, TH = 9.1, edge angle = 65 ). The third specimen is a possible medial fragment of black chert that has shallow dorsal retouch on opposite edges. Microblades: Of the 20 microblades (MNI=12), fourteen are made of Mugford chert, four of crystal quartz and two of Ramah chert (Figure 211: c-j). Only one is complete; 1 1 are proximal fragments, six are medial fragments and two are distal fragments. Metrics are as follows: xW = 8.08 mm, r = 5.1-14.6, X TH = 2.1, r = 1.1-4.4. Platform angles: 85' = 1, 80' = 7, 75 = 1, 70 = 3. Platform preparation: plain = 1, ground = 1, plain/ground = 1, retouch = 2, retouch/ground = 3. Four specimens have traces of deliberate retouch while eight exhibit minor notching and use-wear. Microblade cores: The first of the two cores (Figure 21 1 :b) is made on a sub-rectangular to trian- gular block of dark gray Mugford chert (L =56.4 mm, W =33.4, TH = 20.59). A triangular platform is posi- tioned at one of the long ends of the chert block. The platform surface is fully retouched by shallow flakes and has a platform angle of 80°. Flake scars from six blade removals run the length of the front and sides of the core. The blades ranged in size from 37-55 mm long and 5.5-10 mm in width. The back of the block is not retouched. The second example (Figure 21 1 :a) is a tabular-shaped core of light gray silicified slate (L = 51.6 mm, W = 32.7, TH = 14.3). The plat- 296 CHAPTER 1 3 form is located on one of the long ends of the core. The edges of the core are retouched slightly, including some thinning on the long end opposite the platform. The platform is long, narrow and flat, without preparation; a well defined point of impact inden- tation is found on the platform edge. The platform angle is 80". One microblade ca. 29.5 mm long and 10.1 mm wide was removed from the narrow longtitudinal edge of the core. Celts/adzes: Both speci- mens are made of green slate. A definite adze (Figure 211:k) is made on a slate tablet and is broken longtitudinally (L = >56.7 mm, W = >27.2, TH = 1 2.7). The implement has a rectangular cross-section with lateral edges flaked perpendicular to the tool faces. On the dorsal side of the tool polish is mostly limited to within 4.5 mm of the tool edge, although it extends up to 1 6 mm back. The ventral side is almost completely polished while the lateral edges are unpolished. There is little indication of use wear on the bit apart from three small fractures on the ventral side and a slight rounding of the edge. The second implement is a complete unground preform for a celt or adze (Figure 21 1 :l). Its overall form is "pick-like," with a square cross-section resulting 21 2/ Attu's Point L-7: axial structure, view towards the northeast. 21 3/ Attu's Point L-7: box-hearth with thermoliths. View towards the southeast. 2 1 4/ Attu's Point L-7 box-hearth, thermoliths removed. View towards the northeast. ATTU'S POINT; A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY 297 Table 86. Attu's Point L-7: Tool Classes by K-Means Cluster. Cluster 1 Cluster 2 Other TOTAL N (%) Endblades 4.5* 1.5 6 (25.0) Endblade preforms 1 1 (4.2) Biface fragments 2 2 (8.3) Burins 2 2 (8.3) Burin spalls 6 6 (25.0) Microblades 3 1 4 (1 6.7) Utilized flakes 2 1 3 (12.5) TOTAL 20.5 2.5 1 24 half numbers denote refits across units m 3x7y 3x4y Gray Mugford chert concentration F Flat rock 21 5/ Attu's Point L-7: axial structure and total flake distribution. from perpendicular retouch on all sides (L = 1 1 5.4 mm, W = 24.9, TH = 24.3). Utilized flakes: The 24 utilized flakes are made of a variety of materials: seven of dark gray Mugford chert, five of light gray Mugford chert, one of light gray 7x7y silicified slate, one of crystal quartz, six of Ramah chert and one of slate. The relatively high frequency of Ramah chert spec- imens is noteworthy. L-7 EXCAVATION L-7 was located in the middle of the site, just below the edge of the second beach ridge at 1 1 m asl (Figure 193). The locality consisted of a small but perfect- ly preserved axial feature, around which a total of 1 2 m^ was excavated (Figures 212 to 215). A debitage concentration at the foot of the feature extend- ed slightly outside the limits of the excavation, so it is possible that an external midden area remains uninvestigated. The axial feature was just under 3 m long, 1 m wide and was oriented northwest- southeast, perpendicular to the shoreline (Figures 212 and 21 5). A beautifully preserved box-hearth lay at the center of the feature (Figures 2 1 3 and 2 1 4). It was com- prised of four thick inclined rock slabs that created a 7x4y 298 CHAPTER 1 3 Table 87. Attu's Point L-7: Tool Classes by Structural Position. East Axial West Ax/E Ax/W TOTAL N (%) Endblades 1.5 3.5 1 6 (25.0) Endblade preforms 1 1 (4.2) Biface fragments 2 2 (8.3) Burins 1 1 2 (8.3) Burin spalls 1 5 6 (25.0) IVlicroblades 1 1 2 4 (16.7) Utilized flakes 2 1 3 (12.5) TOTAL 2.5 10.5 4 1 6 24 compartment of 60 by 40 cm. The compartment was filled with tightly packed fist-sized thermoliths (some fire-cracked) that overlay three small horizontal flat stone slabs at the base of the feature. The gravel at the bottom of the hearth contained a few flecks of solid charcoal as well as a faint stain of powdered charcoal. Two small clusters of probable thermoliths were found outside the hearth, one immediately to the east of the hearth within the axial feature, another outside the fea- ture 1 m north of the center of the hearth. There was also a 1 .7 m long scatter of small rocks 1 .3 m west of the axial feature that likely represents thermoliths 21 6/ Attu's Point L-7 tool distribution and k-means clusters. ArrU'S POINT: A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY 299 Table 88. Attu's Point L-7: Tool Raw Material Frequencies by Structural Position. East Axial West Ax/E Ax/W TOTAL N (%) Dark gray Mugford chert 2.5 3.5 2 6 14 (58.3) Light gray Mugford chert 3 3 (12.5) Cray-speckled chert 1 1 (4.2) Crystal quartz 1 1 1 3 (12.5) Ramah chert 3 3 (12.5) TOTAL 2.5 10.5 4 1 6 24 cached or discarded against the wall of a tent. There was no trace of tent anchor rocks. A tiny charcoal sample collected from beneath the flat rocks at the bottom of the hearth was submitted to Isotrace Laboratory, University of Toronto, for AMS dating. The result was very surprising: 50±50 B.P. (TO- 4793). Of all the charcoal samples from Attu's Point, this one seemed to have the strongest contextual asso- ciation, virtually sealed underneath an undisturbed hearth feature. The anomalous result might be attrib- utable to recent burning of roots that had penetrated beneath the hearth or post-excavation contamination. Tool Distribution The L-7 tool distribution is plotted in relation to a /(-means cluster analysis in Figure 216. Cluster 1 is 3x7y 7x7y 3x4y 7x4y Circle indicates location of 6 dark gray Mugford chert items lacking point provenience Crystal quartz A Dark gray Mugford chert + Gray speckled Mugford chert • Light gray Mugford chert * Ramah chert 21 7/ Attu's Point L'7: tool raw material distribution. 300 CHAPTER 1 3 3x7y 3x4y S Concentration Gray Mugford chert flakes 7x7y 3x7y 7x4y 3x4y Black chert flakes 3x7y 3x4y 7x7y 3x7y 7x4y 3x4y Crystal quartz flakes Ramah chert flakes 3x7y 3x4y 7x7y 7x4y Slate flakes 2 1 8/ Attu's Point L-7: flake raw material distributions. ATTU'S POINT; A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY Table 89. Attu's Point L-7: Flake Raw Material Frequencies by Structural Position. East N (%) West N (%) TOTAL N (%) Dark gray Mugford chert 18 (12.8) 1 57 (54.5) 1 75 (40.8) Gray-speckled chert 1 9 (13.5) 54 (18.8) 73 (17.0) Patinated chert 14 (4.9) 14 (3.3) Black chert 3 (2.1) 1 (0.3) 4 (0.9) Ramah chert 4 (2.8) 19 (6.6) 23 (5.4) Crystal quartz 6 (4.3) 1 (0.3) 7 (1.6) Slate 91 (64.5) 42 (14.6) 1 33 (31.0) TOTAL 141 288 429 positioned on the northwestern arm of the structure while Cluster 2 flanks the feature. Six of the seven quadrat provenienced items (five burin spalls and a burin) can be associated with Cluster 1, while one item (utilized flake) lay somewhere between the two clusters. A refitted endblade connects the two clus- ters. All the implements are found on the northern (seaward) side of the excavation and Table 86 indi- cates almost all the tools belonged to Cluster 1 at the northwest corner of the axial feature. Although the distinct structure could facilitate a model-based bilateral analysis, the paucity of tools limits the value of such an approach. Nonetheless, Table 87 outlines the distribution of the tool classes partitioned by spatial position relative to the feature. Little can be concluded except that the majority of tools were associated with the feature and that biface frag- ments and burin spalls tend to be associated with the feature. Spatial distributions of the tool raw materials are shown in Figure 217 and summarized in Table 88. Given the low frequencies little can be said except that light gray Mugford chert and Ramah chert tend to be more closely associated with the axial feature than are the other raw materials. The total flake distribution (Figure 215) indicates deposition restricted to the north side of the structure, the same pattern as the tool distribution. Looking at the distribution of individual flake raw materials (Figure 218), a major concentration of Mugford chert Table 90. Attu's Point L-7: Total Lithic Raw Material Weights (Tools and Flakes). In Grams. Tools g Flakes g TOTAL g (%) Dark gray Mugford chert 5.9 52.1 58.0 (25.0) Gray-speckled chert 0.4 10.7 11.1 (4.8) Patinated chert 7 7.0 (3.0) Black chert 2.6 2.6 (1.1) Ramah chert 4.3 2.6 6.9 (3.0) Crystal quartz 3.4 3.3 6.7 (2.9) Slate 139.3 139.3 (60.1) 302 CHAPTER 1 3 3 1 1 1 15 147 10 Dark Gray Mugford chert flakes B= 1 B= 1 p= 1 B= 1 p= 11 B= 1 p= 1 Patinated Mugford chert (P) Black chert (B) 16 4 52 2 Gray speckled Mugford chert R= 1 C= 1 R= 1 C=2 R= 1 C= 3 R= 19 C=2 C= 1 Ramah chert flakes (R) Crystal quartz flakes (C) 1 1 60 4 1 8 16 31 9 Slate flakes 219/ Attu's Point L-7: flake raw material distributions by quadrat. flakes (generic) is found at the northern terminus of the axial feature. This concentration is strongly sug- gestive of a restricted l14.1 mm, TH = 2.3). The sixth specimen is a very small complete endblade made of Ramah chert (Figure 220:a). It has a biconvex cross-section, a slightly concave base (depth= 0.5 mm) and faint edge serration (L= 16.3 mm, W= 8.5, TH=1.7). There is also slight medial grinding on both faces. Endblade preforms: The single endblade pre- form (Figure 220: f) is a proximal fragment made of gray Mugford chert. It has a straight base and rough bifacial flaking (W= 16.8 mm, TH= 2.9). Points/endblades: The two fragments are so small it is impossible to determine which tool class is represented. One of them is a distal element made of crystal quartz (Figure 220:g) with a biconvex cross-section and slight edge serration. The other is a medial fragment made of light gray Mugford chert with a biconvex cross-section. Burins: The two burins are both made of gray Mugford chert. One is a small specimen that initially was classed as an endblade fragment (Figure 220: h). The implement is made on a very thin flake with a remnant basal striking platform. It is completely bifacially flaked and a burination platform was pre- pared using a small bifacial distal notch. Two spalls were removed obliquely from the notch; the first sheared off most of one lateral edge of the flake while the second is very short and runs parallel to the first. Much of the medial portion of one face is ground while the other face has slight grinding dis- tally (L = 20.6 mm, W = 9.3, TH = 1.6). The second example is a distal fragment, also made on a small thin flake (Figure 220:i). Fine retouch towards the distal end of one lateral edge was used as a platform to remove a transverse spall. This spall surface was then employed as a platform to remove two small spalls from the other lateral edge of the flake. Near the distal end on both faces there are numerous stri- ations that run 70-80 to the burin spalls; this appears to be incipient grinding. 5x7y 5x5y - 5x3y I I Hearth border rock F Flat rock Core fragment + Microblade • Utilized flake A Black chert flake * Slate flake 6x3y 221/ Attu's Point L-9: axial structure. Burin Spalls: Of the six burin spalls, three are complete, two are proximal fragments and one is a medial fragment. All are made of gray Mugford chert. One IS a primary spall, the rest secondary. The pri- mary spall is long (24.7 mm) while the two complete secondary spalls are half its length (1 1 .7, 1 3.6 mm). Additional metrics: xW = 3.5 mm, r = 2.0-4.4, xTH = 2.1 , r = 1 .2-3.3. Microblades: All four of the microblades (MNI = 2) are medial fragments. Three are made of gray Mugford chert, one of crystal quartz (Figure 220: j, illustrates a Mugford chert example). Reliable widths vary between 6.5-8.6 mm. One has retouch on both lateral edges while two others exhibit slight use-wear on their margins. Utilized Flakes: Two of the three utilized flakes are made of Ramah chert, one of crystal quartz. ATTU'S POINT: A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY 305 222/ Attu's Point L-9 tools, a. core fragment; b: macrob- lade. c. utilized flake, d macroblade. L-9 TEST EXCAVATION L-9 was located on the uppermost beach ridge at 1 2 m asl. (Figure 193). At first glance it appeared to consist of a well-preserved axial structure, so a 1 m by 4 m excavation unit was placed along the length of the fea- ture (Figure 221 ). After exposure, a hearth was clearly identifiable but other structural elements were less dis- tinct. The axial feature was 3.5 m long, 0.80-1.0 m wide and was oriented east-west, perpendicular to the beach front. The slightly disturbed hearth was com- posed of three thick border rocks and a fourth large flat rock that surrounded a pair of small horizontal flat slabs, creating a compartment of 60 by 40 cm. Some of the fist-sized stones within the hearth area were likely thermoliths and small fire-cracked rocks were noted on the eastern edge of the hearth. Small flecks of charcoal were found in the gravel on the outer perimeter of the hearth and beneath some of the hearth and central axial feature rocks. Just outside the excavation at the western end of the feature was a small rock cluster that might be part of the L-9 feature, but it could also represent a separate structure. A 223/ Attu's Point surface collected tools, a, b: bifaces (L-4); c large Ramah chert flake (L-4), d: adze (L-3), e-f: burin, endblade preform (L-15). small charcoal sample procured from the gravel imme- diately adjacent to the hearth rocks was submitted to the Isotrace Laboratory, University of Toronto, for AMS dating. The result was an unacceptably recent date of 690 ± 60 B.P. (TO-4794). The sample was not from a sealed context and could conceivably be related to a post-occupation vegetation burn. Very little lithic material was found in the L-9 exca- vation: two "macroblades," one core fragment, a biface fragment and two utilized flakes. Both macroblades (Figure 222:b, d) are made of gray Mugford chert; one is complete (L = 56.8 mm, W = 18.9, TH = 8.9), the other a proximal fragment. Both have a platform angle of 80'; one has a unprepared plain platform the other platform is retouched. Both have slight use-wear on their edges. The core fragment (Figure 222:a) is made 306 CHAPTER 1 3 a224/ Attu's Point surface collected tools, a-b Maritime Archaic biface, Maritime Archaic stemmed point (L-l 2): c: lanceolate biface (L-6): d: lanceolate biface (east of L5). of gray Mugford chert. The biface fragment is a tiny, finely flaked lateral edge element of black chert. One of the utilized flakes is made of gray Mugford chert, the other of crystal quartz (Figure 222:c). Eight flakes were collected, five of which are made of black chert, two of slate and one of dark gray Mugford chert. OTHER LOCI Throughout the course of the work at Attu's Point the site surface was scanned for traces of additional struc- tures and artifact concentrations. Surface collections were made at some of these localities. Each is described briefly below (see Figure 193 for locations). L-2: Located on the first beach terrace at 1 m asl., there was a possible axial structure, 3.5 m long, oriented northwest-southeast. Visible on the surface was a microblade and flakes of Mugford chert, Ramah chert and slate. L-3: Situated on the first beach terrace at 1 m asl., a few rocks and flat slabs were observed, but no clear structure. A flake scatter measuring 4.5 m north- south and 2.5 m east-west contained Mugford chert and abundant slate. A complete adze of green slate (Figure 223: d) was surface collected. The implement is sub-rectangular in form, with abrupt perpendicular flaking on one lateral edge and more irregular retouch on the other lateral edge (L = 78.6 mm, W = 33.2, TH = 2 1 .6). The dorsal side of the adze bit was formed by flaking out a concavity, then polishing no more than 4.5 mm back from the edge (edge angle = 75 ). On the ventral side the bit is mostly polished within 6.0 mm of the edge, although it extends back to 21.5 mm. Edge wear consists of a few small fractures and a slight rounding of the edge. L-4: This locality was situated on the edge of the lowest beach terrace, just under 1 m asl., so it could be the youngest locus at the site. Various rocks were present, but no clear structure. A lithic scatter measur- ing 4.5 m north-south and 2.5 m east-west contained Mugford chert, slate and large Ramah chert flakes. A Ramah chert biface fragment, a biface preform of black chert and a large Ramah chert flake were surface col- lected. The Ramah chert biface (Figure 223:b) is bro- ken in the middle and is missing part of its mid-sec- tion. It is unfinished, with a slightly convex base bear- ing traces of grinding and there are three slight notch- es near the proximal end of one lateral edge (L = ca. 76.5 mm, W = 32.1 , TH = 1 3.7). The black chert biface preform (Figure 223:a) is leaf-shaped: one face of the distal end is fairly well flaked but the rest of the imple- ment is only retouched marginally (L = 49.8 mm, W = 26.0, TH = 7.7). The large Ramah chert flake (Figure 223:c) is the largest piece of that material observed at the site (L = 82.9 mm, W = 32.1 , TH = 1 3.7). East of L-5: About 4.75 m east of the L-5 exca- vation the mid-section of a lanceolate biface made of Ramah chert (Figure 224:d) was surface collected. It is completely retouched bifacially, with a biconvex cross-section. Weak shoulders occur just below the mid-section on both lateral edges, but there is more distinct stemming near the base (L = > 49 mm, W = 23.9, TH = 8.1). ATTU'S POINT; A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY 307 225/ Attu's Point surface collected tools, a stemmed point (L-14): b-c: biface tip, bunn (L~l 5), d bunn (L-16), e-f: biface base, sidescraper (L-14). L-6 Positioned on the upper portion of the first beach level at 10.5-1 1 m asl., this locality consisted of an indistinct rock structure, flakes of slate, crystal quartz and black chert, as well as a lanceolate biface fragment and a core tablet. The biface fragment (Figure 224:c) is a proximal specimen made of gray Mugford chert, with a biconvex cross-section and a slight contraction towards the base (W= 21 .4 mm, TH= 5.6). The core tablet is made of black chert; three lin- ear flakes removed from one end of a flat-flaked plat- form suggest it was a test piece for a microblade core (L = 76.4 mm, W = 62.3, TH = 22.6). L-8: Located on the upper portion of the first beach level at 1 1 m asl., an axial structure may be present in a moss-covered area 3.8 by 3.0 m in size. Flakes of Mugford and Ramah chert were noted, along with a large microblade of Mugford chert. L-IO: This locality was situated in the middle of the second beach terrace, just under 12 m asl. There appeared to be a small axial structure, 2.4 m long, 70 cm wide, oriented northwest-southeast. A bifacial implement of Ramah chert was noted nearby. L-] I: Near the middle of the second beach ter- race, at just under 12 m asl., was an indistinct rock structure associated with flakes of Ramah chert, slate and crystal quartz, as well as a large black chert microblade. L-12: This find spot was situated at the extreme eastern edge of the site, on the uppermost beach level at 1 2 m asl. It consisted of a Ramah chert stemmed point of Late Maritime Archaic Rattler's Bight form and a Ramah chert biface fragment that might also be of Maritime Archaic origin. A 1 m^ test pit was excavated in a mossy area 2 m southeast of these surface finds, since a few rocks suggested the possibility of a fea- ture, but no cultural material was encountered. The stemmed point is complete except for its tip (Figure 224:b). It has well-defined shoulders and a striking platform at the base of its stem. The dorsal side of the stem is completely retouched, but the ventral side is only partially retouched. Only the margins of the blade element are retouched (L = >40 mm, W = 20.2, TH = 4.9, Stem L = 12.5 Stem W[shoulder]= 11.8, Stem W[base] = 9.5). The Ramah chert biface is a distal frag- ment (Figure 224:a). Its dorsal face is almost complete- ly flaked while the ventral retouch is limited to the mar- gins ( W = >45 mm, TH = 6.3). L-13 This small locality was situated on the uppermost terrace of the beach, near the southwestern edge of the site at just under 1 2 m asl. It consisted of a dense concentration of green slate flakes in front of a boulder, suggestive of a reduction locale for adze/axe production. L-14 & 15: These two localities seemed continu- ous with each other and were located on the upper- most beach terrace at the western extremity of the site, about 1 1 m asl. L-14 contained possible structur- al rocks and a surface collection procured there con- sisted of a stemmed point, a biface base, and a large sidescraper. Slate flakes were also observed. The stemmed point (Figure 225:a) is a distal fragment 308 CHAPTER 1 3 made of Ramah chert with a biconvex cross-section and slightly serrated lateral edges (W= 1 1 .2 mm, TH= 2.5). The biface base (Figure 225:e), also made of Ramah chert, exhibits a faint trace of a shoulder 1 7.5 mm above the base; it is probably part of a shoul- dered lanceolate implement (cf., Figure 224:c, d). The complete sidescraper (Figure 225:0 is made on a blade-like flake of dark gray Mugford chert (L = 57.0 mm, W =21 .6, TH =6.3). The working edge is slightly concave, with a 65" angle. Both lateral edges are retouched and ground slightly, while the ventral side of the proximal end exhibits some thinning retouch. Use-wear in the form of edge rounding and a few small ventral fractures is present on the screwdriver- like distal tip and the distal portion of the working edge. The implement is almost identical to the combi- nation burin/scraper from L-1 (Figure 201 :c), minus the burination. At L-1 5, 2 m northwest of L-1 4, flakes of Mugford chert and slate were observed and an endblade pre- form, biface tip and two burins were surface collected. The endblade preform (Figure 223: f) is triangular, although missing its tip, and made from crystal quartz. It has a slightly concave base, which has been thinned bifacially, and a roughly biconvex cross-section (W = 16.2 mm, TH = 4.0). The biface tip (Figure 225:b) of patinated chert has a biconvex cross-section. Both burins are made of gray Mugford chert and are unground. The first specimen (Figure 225:c) is made on a large flake with a striking platform at the base (L = 33.4 mm, W =21 .6, TH =6.2). On its dorsal side the proximal end is retouched with long flakes extending over the entire surface, while on the ventral side retouch is limited to the edge margins. One lateral edge has a slight shoulder while the other exhibits a small notch below the termination of a burin scar. The distal end platform has unifacial dorsal retouch, from which eight burin spalls have been removed obliquely. The second burin (Figure 223:e) is also made on a flake with a striking platform at its base (L =28.5 mm, W =1 7.5, TH = 4.3). The proximal portion of one later- al edge has a faint shoulder created by fine ventral retouch and slight grinding. The other lateral edge exhibits ventral retouch that thinned the bulb of per- cussion. The distal end is "beaked" with three burina- tion planes that involved 10 visible spall removals. L16: This small locality was located downslope from L-1 4/1 5 at 1 1 m asl. It consisted of a lithic scat- ter of Mugford chert flakes, a burin and an endblade; only the burin was collected. The burin (Figure 225:d) is made on a flake of gray Mugford chert with a striking platform at the base (L = 28.9 mm, W = 20.9, TH = 6.2). Its lateral edges are slightly retouched dorsally, producing faint constric- tions on each edge. The distal end has oblique dorsal retouch, including a small notch. Two burin spalls were removed, slightly obliquely. There is no facial grinding. L- 1 7. This locality lay between L-7 and L-8 at 1 1 m asl. Possible structural cobbles were associated with flakes of Mugford chert and slate. L-]8. About 1 m south of L-9, on the uppermost beach terrace at 1 2 m asl., was a rock structure 2 m in diameter that was associated with Mugford chert, Ramah chert and slate flakes. L-]9. Positioned near the eastern edge of the site between L-1 1 and L-1 2, at 1 2 m asl., this find spot contained a few Mugford chert flakes. L-20. Located between L-1 and a bare rock out- crop to the southwest, at 1 m asl., L-20 consisted of possible structural remains associated with a Ramah chert biface base and crystal quartz flakes. L-2] Southeast of L-1 was a vegetated area between two rock outcrops. A few flakes of Mugford chert and slate were observed and there is potential for structural remains beneath the vegetation. INTER-LOCUS COMPARISONS The preceding sections explored spatial variation with- in each locus. Before providing an overall assessment of the site a comparison between the excavated loci will be presented, considering variation in tool types, lithic raw materials and spatial organization. Table 91 ATTU'S POINT: A PRE-DORSET 'CENTRAL PLACE" ON WEBB BAY 309 Table 91. Attu's Point: Comparison of Tool Classes by Excavated Loci and Other Collections. L-1 N (%) L-5 N (%) L-7 N (%) Surface, L-9 N (%) Endblades 4 (8.2) 4 (4.9) 6 (25.0) Endblade preforms 4 (8.2) 1 (1.2) 1 (4.2) 1 (5.0) Endblades/points 2 (2.4) 2 (8.3) Stemmed points 1 (1.2) 1 (5.0) Lanceolate bifaces 2 (10.0) Bifaces 2 (4.1) 3 (3.7) 5 (25.0) Burins 3 (3.7) 2 (8.3) 3 (15.0) Burin spalls 1 1 (22.4) 1 7 (20.7) 6 (25.0) Microblades 1 3 (26.5) 19 (23.2) 4 (16.7) 2 (10.0) Microblade cores 2 (2.4) Other cores 2 (10.0) Scrapers 2 (4.1) 3 (3.7) 1 (5.0) Scraper/burins 1 (2.0) 1 (1.2) Utilized flakes 10 (20.4) 24 (29.3) 3 (12.5) 2 (10.0) Celts/adzes 2 (4.1) 2 (2.4) 1 (5.0) TOTAL 49 82 24 20 Table 92. Attu's Point: Comparison of Tool Raw Material Frequencies by Excavated Loci and Other Collections. L-1 N (%) L-5 N (%) L-7 N (%) Surface, L-9 N (%) Dark gray Mugford 18 (36.7) 36 (43.9) 14 (58.3) 9 (45.0) Light gray Mugford 16 (32.7) 3 (3.7) 3 (12.5) Cray-speckled chert 3 (6.1) 6 (7.3) 1 (4.2) Patinated chert 2 (2.4) 1 (5.0) Black chert 1 (2.0) 4 (4.9) 3 (15.0) Ramah chert 2 (4.1) 11 (13.4) 3 (12.5) 4 (20.0) Crystal quartz 6 (12.3) 10 (12.2) 3 (12.5) 2 (10.0) Slate 2 (4.1) 2 (2.4) 1 (5.0) Gray silicified slate 1 (2.0) 8 (9.8) TOTAL 49 82 24 20 outlines the variation in Pre-Dorset tool types between the three excavated loci and the other localities at the site. Most notable is the relatively high proportion of endblades and preforms at L-7 (the smallest sample) versus their low proportion at L-5 (the largest sample), a trend supported by the expected frequencies calcu- lated for the table. The proportion of burin spalls is consistent across the loci, although burins themselves are absent at L-1, aside from a single combination scraper-burin. L-7 is relatively lacking in implement 310 CHAPTER 1 3 Table 93. Attu's Point: Comparison of Flake Raw Material Frequencies by Excavated Loci. L-1 N (%) L-5 N (%) L-7 N (%) Dark gray Mugford 1 1 (4.8) 38 (1 1.9) 175 (40.8) Light gray Mugford 99 (43.2) 9 (2.8) Cray-speckled chert 5 (2.2) 1 1 (3.5) 73 (17.0) Patinated chert 32 (13.9) 16 (5.0) 14 (3.3) Black chert 5 (2.2) 4 (1.3) 4 (0.9) Ramah chert 2 (0.9) 44 (13.9) 23 (5.4) Biostrome chert 1 (0.3) Crystal quartz 1 8 (6.8) 4 (1.3) 7 (1.6) Slate 57 (25.6) 1 1 5 (36.3) 133 (31.0) Clastic slate 62 (19.6) Light gray silicified slate 1 3 (4.1) TOTAL 229 317 429 Table 94. Attu's Point: Comparison of Total Lithic Raw Material Weights by Excavated Loci. In Grams. L 1 g (%) L-5 g (%) L-7 g (%) Dark gray Mugford 48.6 (10.1) 89.7 (17.4) 58.0 (25.0) Light gray Mugford 59.9 (12.5) 16.5 (3.2) Gray-speckled chert 24.0 (5.0) 7.8 (1.5) 1 1.1 (4.8) Patinated chert 13.2 (2.8) 9.4 (1.8) 7.0 (3.0) Black chert 3.1 (0.6) 7.0 (1.4) 2.6 (1.1) Ramah chert 42.5 (8.9) 30.8 (6.0) 6.9 (3.0) Biostromatic chert 0.4 (0.1) Crystal quartz 28.3 (5.9) 8.3 (1.6) 6.7 (2.9) Slate 259.8 (54.2) 222.5 (43.2) 1 39.3 (60.1) Clastic slate 72.5 (14.1) Light gray silicified slate 50.1 (9.7) classes that might be considered "maintenance" tools: microblades, scrapers, and utilized flakes. Celts/adzes were also absent at L-7, despite the presence of con- siderable slate debitage. Lanceolate bifaces ("knives") are lacking from all three loci, with the possible excep- tion of a biface edge fragment from L-5, although they were surface collected elsewhere at the site. The limit- ed excavation areas may contribute to sampling error, ATTU'S POINT, A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY but there seem to be functional or depositional differ- ences here. The overall trend in the excavated and surface col- lected material is towards a high proportion of har- poon endblades and preforms, relatively few stemmed points or small bipointed bifaces that might be arrow points, relatively few burins, scrapers and bifacial knives. There is a fairly high proportion of burin spalls 31 1 relative to burins and a low proportion of microblade cores relative to microblades, both of which point to curation processes. The functional significance of this pattern for the site as a whole will be taken up below. Table 92 displays the frequency of tool raw mate- rials across the excavated and surface collected loci. Dark gray Mugford chert dominates overall. The most notable differences are that tools of light gray Mugford chert are found primarily at L I, less Ramah chert is present at L-1 and most of the gray silicified slate occurs at L-5. When expected frequencies are consid- ered it is clear that the high proportion of light gray Mugford chert at L-1 is the strongest pattern. Table 93 compares the flake raw material frequencies across the three excavated loci. The clearest patterns are the pre- dominance of dark gray Mugford chert at L-7, light gray Mugford chert at L-1 , a slightly higher proportion of gray-speckled chert at L-7, a higher proportion of patinated chert at L-1, a higher proportion of Ramah chert at L-5 and the limitation of clastic slate to L-5. These trends are confirmed when expected frequen- cies are considered. Slate occurs in high proportions across the board. If we compare the proportions of tool and flake raw materials there are several examples of inconsistencies between the two. Both L-1 and L-5 seem to be somewhat low in debitage of dark gray Mugford chert compared with their abundance of tools made of this material. Conversely, L-1 and L-7 are completely lacking in tools made of patinated chert, while each has at least a mod- est number of flakes of the material, and L-7 lacks tools of black chert but contains a few flakes of the material. L-7 also lacks tools made of slate, while it has the high- est frequency of slate flakes of the three localities. These inconsistencies point to a technological organiza- tion involving asynchronous tool production/retouch and discard; in other words, curation processes and locational staging of reduction. Variations in the quan- tities of raw materials can result from the differential access of social units to point sources, personal pref- erence, or from households being at different posi- tions on the supply/consumption cycle. Table 94 com- pares total lithic raw material weights across the loci. Generic slate is clearly dominant in all cases, followed by dark gray Mugford chert, but the other materials vary somewhat in their proportions. Organizational differences in lithic reduction between the loci might also be seen in flake size dis- tributions. Figure 226 shows the size distributions of Mugford chert flakes (all variants lumped together) at each locality.^ ^ L-1 and L-5 have similar profiles, but L-7 has a much higher percentage of material in the <10 mm category and lower percentages in the cate- gories from <20 mm and larger. While sample size may be a factor, the L-7 material suggests more intensive secondary/tertiary reduction. In contrast, the flake size profiles for slate are very similar for all three loci (Figure 227). Furthermore, slate flake sizes are signifi- cantly larger than Mugford chert flakes, the former with modes ranging from <20 mm to <25 mm, the lat- ter with modes at <10 mm and <15 mm. Thus, not only is slate the most abundant material by weight at all loci, it is also reduced with less secondary/tertiary retouch than is the case for Mugford chert. This ten- dency may result from larger sized slate raw materials, the role of grinding rather than fine retouch in creating the final working edge, or lack of concern for material conservation. The ambiguous nature of the structural remains at L-1 and L-5 render comparisons of spatial organization somewhat difficult. Nonetheless, a set of similarities and differences can be formulated. At both L-1 and L- 7 the majority of lithic materials were deposited on the seaward ("front ") side of the features. At L-7 this might be a consequence of the feature's orientation perpen- dicular to the beach front, with most activity being cen- Given the lack of screening the smaller size categories may be underrepresented. 312 CHAPTER 1 3 226/ Attn's Point: Mugford chert flake size distribution. Slate Flake Size Distribution 35 30 25 20 15 10 5 <10 <15 <20 =25 <30 <35 <40 <45 <50 >50 5 mm classes 227/ Attu's Point: slate flake size distribution. tered on a dwelling entrance. L-1 is problematic since the orientation of the apparent structure is uncertain, possibly oblique to the beach but conceivably parallel. Also, there may be more than one feature represented. One interpretation of the L-5 pattern was of an axial feature oriented parallel to the beach front with a pos- sible "external" activity area positioned along that par- allel axis. In the parallel to the beach interpretation of both L-1 and L-5, a hearth would be positioned at the ter- minal end of the feature rather than in the middle. An alterna- tive reading of L-5 as a linear series of three hearths would represent a quite different organization of space than the other two loci. At all three localities a high proportion of the tools tend to be associated with the axial fea- tures (or in the case of L-5 pos- sibly an isolated hearth), although a few implements occur on the flanks of the fea- tures. Tool frequencies are too low to identify statistically sig- nificant patterns in the spatial distribution of types, but a few observations are pertinent. At L- 7 endblades and preforms lay close to the axial structure with a refit leading to a small group on one flank of the feature. At L- 1 endblades and preforms were concentrated in a cache-like deposit, although this lay some distance from the hearth. However, if the L-1 feature was oriented parallel to the beach front, then the endblade/pre- form concentration would lie at the foot of the structure, a placement similar to L-7. At both L-1 and L-7 burin spalls were concentrated near the hearth/axial feature areas. One reason for the limited spatial patterning in tool types, which has also been observed elsewhere in the eastern Arctic (e.g., Dekin 1976; McChee 1979:52- 55; Jensen 1 996:1 57; Mikalsen 2001 :86), may simply be the narrow range of types represented at the Attu's L-1 L-5 L-7 ATTU'S POINT A PRE-DORSET "CENTRAL PLACE' ON WEBB BAY 313 Point localities. The paucity of burins, bifaces and scrapers is probably related to functional or social aspects of site use (see below). Another variable is preferential deposition of tools in the vicinity of axial features after use or rehafting, rather than in the places they were used— an axial feature "attraction effect" Oensen 1 996:1 56-1 58). Jensen's (1996) analysis of the spatial sorting of raw material types at Paleoeskimo sites in east Greenland observed that: "...each raw material must therefore have had its own individual depositional his- tory" (Jensen 1996:156; see also Mikalsen 2001). There are hints of similar patterning at Attu's Point. At both L-1 and L-7 Mugford chert flakes are generally associated with the axial features while slate flakes are concentrated in separate clusters on the flanks, sug- gesting the reduction activities for these materials were deliberately separated. This does not seem to be the case at L-5, but there it is possible to discern spa- tial distinctions in the deposition of "normal" versus "clastic" slate debitage. At L-1 there is a clear pattern of crystal quartz tool and flake deposition centered on the hearth/rock feature, but this does not seem to be the case at either L-5 or L-7. The frequencies of other raw materials are either too low or too variable from locus to locus to draw any conclusions. Nonetheless, it does seem that some of the more useful behavioral inferences from Paleoeskimo dwellings will be derived from raw material distributions, combined with detailed refitting studies. Chronology The 21 loci at the site occur as three separate linear dis- tributions on two main beach terraces, suggesting that Attu's Point was repeatedly occupied over some length of time. All loci lie between 10-12 m asl., a typical ele- vation for Pre-Dorset sites in the Nain region. The two excavated localities at 10 m asl. on the lowermost beach terrace, L-1 and L-5, were radiocarbon dated 3750±60 B.P. and 3790±70 B.P., respectively. These assays provide a fairly good maximum date for the low- est beach level. As noted previously, attempts to acquire reliable radiocarbon dates from the higher beach levels were unsuccessful. Although the Pre- Dorset (and Maritime Archaic) material on the upper- most beach level near 12 m asl. could date ca. 4000- 3900 B.P., it is also possible that some of it could date ca. 3700 B.P. or later. Although the linear distribution of the Pre-Dorset occupation on three different levels might indicate three periods with several contempora- neous social units, there is no reason to believe that set- tlement of the gently sloping beach was so closely bound to the shoreline that the different levels were not used simultaneously. One might even argue that the undisturbed condition of the axial feature at L-7, on the middle beach level at 1 1 m asl., could indicate that it was the last occupation at the site, post-dating the local- ities on the 10 m level. Consequently, without further radiocarbon dating, shoreline relations provide little help in developing an internal chronology for the site. In the absence of sufficient radiocarbon dates we are left with tool typology. Unfortunately, it has proved difficult to seriate Labrador Pre-Dorset sites beyond Early (4000-3600 B.P.), Middle (3500-3000 B.P.) and Late (3000-2800 B.P.) phases (Cox 1978); indeed, it has been particularly difficult to identify sites pertain- ing to the Middle phase. Cox (n.d.:8) concludes that, typologically speaking, Pre-Dorset was fairly static until 3000 B.P., when a series of changes led to the emergence of Croswater. He suggests that "subtle" dif- ferences may signal Middle phase material: concave vs straight based triangular endblades, thicker and less well-flaked triangular endblades, lack of edge serration on scrapers and burins, less edge serration on bifaces, predominantly straight vs tapered stems on bifaces and the presence of very small burins. One might add to this an increase in facial grinding on burins. It is difficult to identify any substantial evidence for such "later" attributes in the Attu's Point collection. Typologically, the two loci radiocarbon dated ca. 3800- 3700 B.P. exhibit typical Early Pre-Dorset features such as straight-based triangular endblades with slight edge 314 CHAPTER 1 3 serration. The only unusual item is a tiny burin on a microblade at L-5 (Figure 210;i). L-7, however, has a few implements with a possible later stamp. The two burins from this locus were both unusual types made on small flakes, one a very thin bifacially retouched specimen with partial grinding on both faces, the other with striations indicative of proto- or incomplete grinding (Figure 220:h, i). Additionally, a small triangu- lar endblade was slightly ground on both faces and another triangular endblade had slight grinding on one basal corner (Figure 220: a, e). These features are the only support for the suggestion that the undisturbed nature of the axial structure at this locus may indicate a late phase in the site occupation. In conclusion, with- out further excavation and dating of other loci at the site there are no grounds for supposing that the occu- pation of Attu's Point extends beyond the 4000-3600 B.P. period. Structural Comparisons The well-preserved axial structure at L-7 bears some resemblance to the partially excavated structure at L- 9. L-7 is also broadly similar in form to Pre-Dorset features excavated at Dog Bight L-5, east of Nain (Fitzhugh 1 976a: 1 30-1 33; Cox 1978:101). The latter site contained three structures, two of which consist- ed of well defined axial features measuring 3 m by 1 m in size, with central box-hearths constructed of thick rocks and partial rings of tent anchor rocks. One of these structures had a number of small ther- moliths concentrated within the bounds of the axial feature, although there were few within the box- hearth itself (Fitzhugh 1 976a:l 33). A third structure, possibly disturbed, had an axial feature 2 by 1 m in size. The main difference between these structures and L-7 is the total lack of perimeter anchor rocks at L-7. Looking further afield, L-7 bears a striking resemblance to an axial structure at Tuapagssuit (64V1-I,10) near Nuuk, west Greenland (Gullov 1 983:51; Gullov and Kapel 1988:47). Otherwise, L-7 clearly lies within the broad tradition of Paleoeskimo axial features with box-hearths and thermoliths that is well documented from Independence I and Saqqaq contexts (e.g., Gronnow and Jensen 2003; Gronnow and Meldgaard 1991; Knuth 1967; Kramer 1996; McChee 1979; Mikalsen 2001; Mobjerg 1998, 1999; Olsen 1998; Schlederman 1990). The features at L-1 and L-5 are more difficult to evaluate given the likelihood of disturbance, but in both cases a hearth is positioned at the terminal end of a rock alignment rather than in the middle. Whether this signals a functional difference from the more stan- dard axial model at L-7 or individual choice in hearth placement cannot be determined from the present data. It is also conceivable that the spatial pattern at L- 5 should not be interpreted within the expectations of the axial model, but instead as a linear series of hearths such as at Port Manvers Run-1 (Chapter 12). Not only might a linear hearth arrangement signal a functional difference from axial features, it could also be associated with different socio-spatial practices. Functional Variability The term functional variability denotes activity- related variability attributable to resource extrac- tion practices, the latter being subject to seasonal differences. As such, functional variability at Attu's Point is considered along two dimensions: dwelling structures and lithic assemblages. Arctic archaeologists have increasingly recog- nized formal variability in Early Paleoeskimo struc- tures, some of which is probably seasonal in nature. Yet there are no solid criteria for identify- ing structures of different seasonality. The most common assumption is that cold weather struc- tures were of more substantial construction and should exhibit considerable attention to heating facilities, resulting in emphasis on the hearth/axial feature arrangement and possibly larger accumula- tions of charcoal and fire-cracked rock (e.g.. Cox 1978:98; Knuth 1967:45-51; Maxwell 1985:96-98; Odgaard 2003; Olsen 1998:111-116; Renouf ATTU'S POINT: A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY 315 1993:191-197). It is also expected that there may be larger accumulations of tools than at warm- weather sites, and that these accumulations will occur primarily inside the structures, whereas warm-weather dwellings may be associated with external activity areas. On the other hand, Maxwell (1985:98) suggest- ed that sites lacking structural remains may be traces of snow house camps. Ramsden and IVIurray (1 995) provide tentative support for this from faunal associations that suggest "substantial" structures represent warm weather tent structures, while fea- tures consisting only of small vegetation patches with rocks indicate snow-walled winter dwellings. They also note, however, that structural variability can be expected in "transitional" seasons with vari- able weather and variable demands on construction techniques (Ramsden and Murray 1995:115-116). Although the safest conclusion to draw is that corre- lating structures and seasonality is highly problem- atic in the absence of faunal remains, it is by no means the case that the presence of faunal material will be decisive for interpretation. Cronnow and Jensen (2003: 337) note significant contradictions between faunal data and structural and lithic evi- dence from Independence I sites in northernmost Greenland. None of the excavated localities at Attu's Point exhibit structural remains that might be considered substantial constructions. There is little besides a hearth and the remains of a small axial feature or rock alignment. None of the features display a clear perimeter of anchor rocks. L-7 is unusual in its degree of preservation; its classic box-hearth illus- trates the use of thermoliths, but there was minimal accumulation of fire-cracked rock and very few tools. At L-5 there were few tools or flakes associated with the possible axial feature, but larger quantities of lithics were found a short distance away in a context that could be interpreted either as a disturbed struc- ture or as an external dump/activity area. The lack of perimeter rocks might be explained either by warm season use of wooden pegs to hold down tent skins or cold season use of snow. The lack of lithic accumulation has dubious relevance to the seasonality issue, since it is linked to longevity of occu- pation, group size and staging behavior in technologi- cal organization. The location of the Attu's Point fea- tures on beaches highly exposed to westerly winds is advantageous for mitigating insect torment during warmer seasons, but disadvantageous for cold sea- sons. In sum, the structural information is, by itself, of little value for the interpretation of site seasonality. As far as lithic assemblage variability is con- cerned, given the lack of fully published Pre-Dorset col- lections from other parts of Labrador it is difficult to provide a concrete assessment of how the Attu's Point assemblage fits into the overall range of functional vari- ation in Pre-Dorset material. The few existing accounts either provide no information on tool frequencies or lump together material from what are probably differ- ent loci, such that intra-site variation is masked (Cox 1977, n.d.; Fitzhugh 1976; Tuck 1976). Inter-site com- parisons are of limited value without control over intra- site variation. That said, Attu's Point seems to exhibit a lower proportion of burins and perhaps scrapers than elsewhere. Lanceolate bifaces, probably hafted as knives (Cronnow 1994:206-207), are absent from the excavated loci (except for a possible fragment at L-5), although two were surface collected. There is a paucity of small stemmed or bipointed bifaces, which else- where have been identified as arrow points (Crennow 1994:224; Gr0nnow and Meldgaard 1 991:135), pre- sumably for hunting terrestrial mammals. On the other hand, Attu's Point exhibits a fairly high proportion of triangular endblades, which were used as harpoon tips. Consequently, the excavated assemblages give the impression of being weighted towards "procurement" rather than "maintenance" activities and possibly towards the exploitation of seals rather than land mammals. This contrasts with the situation further north at Okak, where inner bay 316 CHAPTER 1 3 sites have relatively few triangular endblades, which Cox (1 978:1 02) interprets as indicating summer occu- pation with less emphasis on sea mammal hunting. However, endblade equipped harpoons could also be used for hunting caribou, which seems likely at Pre- Dorset inland sites from the Barrenlands of the Northwest Territories (B. Gordon 1 996:1 55-1 59). Finally, the presence of three slate adzes and two celt/adze preforms indicates that procurement and processing of wood products, and/or butchering of frozen meat, may have been an important activity at this inner bay settlement. The emphasis in the foregoing was on tool assem- blages reflecting different activities. A related dimen- sion is the articulation of activities with social group composition. Some of the observed variation in assemblages could be related to differences in the gender composi- tion of the social units responsible for each occupation. For example, the high proportion of endblades and lack of maintenance tools (e.g., L-7) might indicate a predominantly male hunting party. On the other hand, there are no compelling grounds to assume strongly gender-specific tools, and the absence of such tools would not necessarily imply the absence of a gender. The quantity of Mugford cherts used at Attu's Point is fairly typical of Labrador Pre-Dorset. Black chert, on the other hand, seems unusually low in fre- quency for the Nain region. The high frequency of slate debitage also seems unusual for the Nain area, but a geographically comparable site from deep within the inner bay zone of the Okak region, Sipukat Bay-1 (HjCn-1), also has a high frequency of celts/adzes (n=6) and slate is the dominant raw material (Cox 1 977:230-237).^ The use of large quantities of slate for celts/adzes in the inner bay zones of Nain and Okak presumably is related to wood procurement and processing in forested areas. The quantity of Ramah chert found at Attu's Point appears to be somewhat greater than that noted at other Pre-Dorset sites in the Port Manvers Run area, but it is well below that report- ed for sites closer to the Ramah chert sources, such as Tuck's (1975:1 35) localities at Sagiek Bay (5.4-37.2% of the tools). It is possible that Pre-Dorset people at Attu's Point procured some of their Ramah chert from their Maritime Archaic neighbours in the Nain region, or that they scavenged chert from abandoned Maritime Archaic sites such as Attu's Bight (see Chapter 12). These superficial observations on functional varia- tion in Pre-Dorset lithic assemblages underline the importance of more detailed studies of Pre-Dorset technological organization. More information is need- ed concerning how Pre-Dorset lithic assemblages were formed and how variability relates to seasonal and sit- uational factors as well as raw material availability. CONCLUSIONS Attu's Point was used repeatedly for short seasonal occupations, presumably over a few hundred years. These occupations were directed towards a relatively limited range of functions and in some cases might have involved social units of restricted composition. But despite the ephemeral nature of individual loci, the repeated use of the site suggests it played a significant role in the regional settlement pattern, serving as a seasonal "central place." Few other Pre-Dorset sites in the Nain area exhibit such repetitive use. The functional signature of the Attu's Point lithic assemblage points to two main activities: sea mammal hunting and wood procurement and processing. Contemporary fauna availability patterns suggest it is most likely Attu's Point was occupied either in the early spring or the fall. At both these times a variety of seal species (but especially harps) may be plentiful in Webb Bay and could have been harpooned close to the site as they moved through the narrow passage between Igloo Sipukat Bay-1 was surface collected for tools, so no quantitative information on debitage frequencies is available. ATTU'S POINT: A PRE-DORSET "CENTRAL PLACE" ON WEBB BAY 317 Island and the mainland (Figure 1 53). Today the pas- sage sometimes may not freeze over until February; if this was the case in the past, open water sealing might have been possible during the early winter. Caribou have wintering areas in the mountains to the north and west of Webb Bay (Brice-Bennett 1 977:1 58- 1 59), but in some years herds may winter on the coast, including in the valleys of neighboring South Aulatsivik Island. Their early spring and early winter movements over the ice between the mainland and South Aulatsivik Island^ ^ might be ideal periods to intercept major herd concentrations. However, the low frequency of stemmed points and lanceolate bifaces could indicate this was not a major activity at Attu's Point. Our own experience in early-middle July was that relatively little game is available at that time, or at least it is difficult to find. A shift to coastal settle- ments would be more advantageous during this period when seals, fish and sea birds are concen- trated in the outer island fringe. The situation changes in late summer and early fall when char and salmon re-enter the inner bays, geese and ducks are plentiful and black bears come down to the shores and stream mouths. Later in the fall harp seals move through Port Manvers Run. Consequently, late summer and fall might provide a mix of resources to draw Pre-Dorset people from the outer coast. Additionally, given the prominence of slate tools and debitage at Attu's Point, it is likely that fall occupations were used to stock up on needed wood supplies and to prepare wooden implements for the winter, while spring occupations could have replenished supplies exhausted during the winter. How this interpretation of Attu's Point fits into the broader view of Pre-Dorset settlement patterns in the Nain area and Labrador as a whole is taken up in Chapter 1 4. A nefarious crew at Immilikulik Island. Left to right: Eric Loring, Dosia Laeyendecker, Susan Rowley, Ben Fitzhugh, Joshua Fitzhugh. (Photo: W. Fitzhugh 1985) ^5 Such a situation was observed in late April 1994, with westward moving animals crossing the ice over the southern entrance to Port Manvers Run near Double Island Cove. 318 CHAPTER 1 3 The St.ucturat.on of Maritime Archaic/ Pre- Dorset Social boundaries in L_ a brad or One of the recurring themes in Labrador prehistory and history has been the shifting social boundaries between Inuit/Pre-lnuit and Innu/Pre-lnnu peoples (e.g., Fitzhugh 1972, 1977, 1987; Taylor 1979).With that in mind, it is now time to draw together several threads that were developed in the previous two chap- ters and address the question of social boundary rela- tions between the Maritime Archaic and Pre-Dorset peoples. Evidence accumulated since the early 1 970s demonstrates that these two cultures overlapped in time between 4000-3500 B.P., corresponding to 2500- 1800 BC (calibrated). Despite this lengthy period of coexistence, the archaeological assemblages of the two cultures exhibit virtually no unambiguous indica- tions of interaction. Up to this point, the Maritime Archaic/Pre-Dorset boundary problem has been discussed largely as a culture-historical problem, using the traditional con- cepts of contact situations with a cultural-ecology overlay. This has been useful, certainly, but the impor- tance of this unique situation for understanding the developmental trajectories of each culture necessi- tates linking the specific historical context to broader theoretical issues. This chapter considers frameworks for doing so, although some new data will also be pre- sented. These frameworks are drawn from different paradigmatic sources, one emphasizing ecological causality, the other focusing on culturally constructed landscapes. Crudely put, they correspond to the theo- retical boxes termed, respectively, processual and post-processual archaeology. The boundaries con- structed around these viewpoints originate from major ontological differences, but their architecture is also derived partly from the polemics of territorial defense within theoretical identity politics. Although the pragmatic approach is to regard the opposed arguments as representing alternative entry points to a common problem and work towards some form of integration, it is difficult to reconcile conceptual struc- tures emanating from divergent ontological first prin- ciples. Consequently, one function of this chapter is to reflect upon the dilemma of working with the con- ceptual tensions found at the boundaries between dif- ferent archaeologies. When all is said and done, it is hoped that the chapter can show how a specific prob- lem in Subarctic culture- history can, in a modest way, illuminate central issues in archaeological knowledge construction. CULTURE-HISTORICAL BACKGROUND The potential for interaction between the Maritime Archaic and Pre-Dorset in Labrador was first identified during James Tuck's 1969-1971 research at Sagiek Bay (Tuck 1975, 1976). Given virtually contemporane- ous radiocarbon dates for the cultures from superim- posed strata at Rose Island Site Q, Tuck (1975:195- A shorter preliminary version of this chapter was published in Hood (2000). THE STRUCTURATION OF MARITIME ARCHAIC/PRE-DORSET SOCIAL BOUNDARIES IN LABRADOR 319 196) speculated about possible cultural transfers: tog- gling harpoons from the Maritime Archaic to the Pre- Dorset and the bow and arrow from the Pre-Dorset to the Maritime Archaic, as suggested by the initial appearance of small projectile points in the Late Maritime Archaic. More recent research indicates these proposals are problematic. Toggling harpoons have been found in early Saqqaq contexts such as Qeqertassusuk, Greenland (Crannow 1994) and small "flake points" have been found in Maritime Archaic sites dated as early as ca. 6000-5500 B.P. (e.g., Nukasusutok-5), although these need not indicate the presence of bows and arrows. During the 1970s and 1980s, William Fitzhugh and his colleagues' fieldwork on the north coast of Labrador expanded the data base for both cultures and radiocarbon dates confirmed the overlap of the Late Maritime Archaic Rattlers Bight Phase and Early Pre- Dorset between 3900-3500 B.P. (Cox 1978, 1988; Fitzhugh 1975, 1976, 1978, 1980, 1984; Thomson 1982, 1983, 1985). Although the two cultures some- times occupied the same or nearby sites, little indica- tion of contact was seen in their tool assemblages. Fitzhugh (1978:91, 1984:22) referred to a possible Maritime Archaic copy of a Pre-Dorset burin, Cox (1 977:235) pointed to a Maritime Archaic gouge re- worked by the Pre-Dorset and the presence of a Maritime Archaic sandstone plummet at a Pre-Dorset site (Cox 1988:3), and Chapter 13 in the present vol- ume reported a Late Maritime Archaic stemmed point and possible biface on the upper terrace of the Attu's Point Pre-Dorset site. Each culture used a distinct set of lithic raw materials. The Maritime Archaic focussed almost exclusively on translucent Ramah chert for their flaked stone tools; this had to be procured from the Ramah Bay region at the northernmost periphery of the Maritime Archaic world (Cramly 1978; Lazenby 1980). The Pre-Dorset used a varied set of colored cherts, most of which were obtained in the Cape Mugford area, north of Okak (Cramly 1978; Lazenby 1980), although varying amounts of Ramah chert also appear in their assemblages. The organization of their lithic technologies was therefore anchored at different points in the landscape. Much of the 1980s' fieldwork focussed on inter- preting Maritime Archaic social organization and com- munity structure. Late Maritime Archaic people used longhouse structures— presumably interconnected tent dwellings— ranging in size from 15 to 100 m long. The individual segments of the longhouses were interpreted as the floor spaces of individual families, which suggested co-residential group sizes ranging from 50 to 100 people. The northern Labrador site of Nulliak Cove, situated between Hebron and Sagiek Bay, contained up to 27 longhouses and was interpreted as a short-term seasonal staging camp for accessing the Ramah chert sources, 50 km to the north (Fitzhugh 1980, 1984, 1985a,b). Maritime Archaic longhouse community organization, combined with mortuary cer- emonialism and exchange systems, suggested some degree of social elaboration or "complexity" (Fitzhugh 1981, 1984; Hood 1993, 1995). In contrast, Pre-Dorset usually maintained small group sizes of perhaps one to three co-residential fam- ilies housed in independent tent dwellings. There are, however, a number of Pre-Dorset sites with multiple occupation locales indicating either seasonal aggrega- tion or regular re-occupation. In contrast to the Maritime Archaic, Pre-Dorset exhibits no evidence for social "complexity," at least if one considers the most frequently used indicator variables. The spatial distribution of Maritime Archaic and Pre-Dorset sites from Nain to the Torngat Mountains suggested that the two cultures partitioned their set- tlement space. Fitzhugh (1984:21-23) proposed an "enclave model" that consisted of a Maritime Archaic "core area" along the central coast south of Nain and a Pre-Dorset core area in the northern Torngat moun- tains and the Nain-Okak regions. The Pre-Dorset core areas encompassed both the sources of their preferred cherts near Cape Mugford as well as the Ramah chert sources preferred by the Maritime Archaic (Figure 320 CHAPTER 1 4 Ramah chert Sagiek Hebron Mugford chert 100 km I I Pre-Dorset core area \^//\ Maritime Archaic core area Exx^ Overlap Hamilton Inlet 228/ Distribution of Late Maritime Arcliaic and Pre-Dorset settlement in Labrador: the macro-territorial model (after Fitzhugh 1984). 228). The Hebron-Sagiek area was an overlap zone since it contained Maritime Archaic seasonal staging camps for accessing the Ramah chert sources. Fitzhugh (1984:21-23) identified a local Maritime Archaic enclave at Nulliak Cove, situated between Hebron and Sagiek, and a local Pre-Dorset enclave at Harp Isthmus on the south side of the entrance to Hebron Fjord. In later years, the find of a substantial Late Maritime Archaic Rattlers Bight Phase site at Attu's Bight in the Main region (see Chapter 12) led Fitzhugh (1986:57) to suggest the enclave model might have to be modified. One implication of the enclave model is that the Maritime Archaic need to maintain seasonal bases in northern Labrador in order to acquire Ramah chert, the stress of maintaining a long distance chert delivery system, and territorial behavior in the overlap zone, may all have contributed to the cultural elaboration seen during the Rattlers Bight Phase (Fitzhugh 1984:23-24). Ultimately, how- ever, these features may have overloaded Maritime Archaic organizational capacities and contributed to the apparent disappearance of the culture as a recognizable entity in the archaeological record after 3500 B.P. (Hood 1993:179). Although the situation is usu- ally seen from the perspective of the Maritime Archaic (given fascination with the puzzling disappearance of "complexi- ty"), Pre-Dorset was also affect- ed by the inter-cultural dynam- ic. The visibility of Pre-Dorset sites also diminishes markedly after 3500 B.P., although to some extent this might be attributable to few radiocarbon dates and limited typological change that masks the presence of sites post-dating 3500 B.P. (Cox 1988). With this basic culture-historical framework in mind we can consider the issues involved in theorizing Maritime Archaic/Pre-Dorset social boundary relations using Fitzhugh's (1984) preliminary model as a point of departure. On the one hand, the model postulates cultural macro-territoriality: the Maritime Archaic dom- inating the central coast, Pre-Dorset the Torngats and the Nain-Okak regions. On the other hand, the model also postulates micro-territoriality occurring within a zone of seasonal overlap (Hebron-Ramah) in which the THE STRUCTURATION OF MARITIME ARCHAIC/PRE-DORSET SOCIAL BOUNDARIES IN LABRADOR 321 Maritime Arclnaic maintained "enclaves, "staging camps for Ramah chert acquisition expeditions. While useful as an initial discussion, this image is somewhat static, implying a degree of boundary equilibrium over sever- al centuries. Additionally, territoriality is a complex concept that has often been formulated in ecological terms, but w/hich has intertwined ecological and social components that need to be drawn out. The following pages will attempt gradually to re-work this prelimi- nary image into something potentially more dynamic. One issue that needs to be addressed is whether this "contact" relationship might be formulated in terms of ethnicity. Over 30 years later, it is still the case that Barth's (1969) conceptualization of ethnicity remains fundamental. Summarized briefly, it postu- lates that ethnicity is a self-ascribed category and a form of social organization used to structure interac- tion. Ethnicity arises through engagement rather than isolation, such that "...boundaries persist despite a flow of personnel across them" (Barth 1969:9). Inter- ethnic relations are ordered by rules that define poten- tial roles and statuses and boundaries are signaled by a selected range of features, behaviors or values that agents regard as significant. Fundamental as this defi- nition may be, the conceptualization is inconvenient for prehistoric archaeologists in that self-ascription and the persistence of boundaries despite social mobility render recognition of ethnic boundary processes highly problematic in the absence of supple- mental textual information. Variations in material cul- ture can be related to the formation of social identity groups at different scales (e.g., clans, gender, region- al bands, dialect groups, etc.) and it is difficult to untangle the various threads in archaeological situa- tions. For example, contrasts between the northern and southern branch Maritime Archaic, as exemplified by the intrusive Black Island Complex (4500-41 00 B.P.) at Hamilton Inlet (Fitzhugh 1975), suggest the exis- tence of different social identity groups in southern and central Labrador prior to the Pre-Dorset coloniza- tion, but to what extent can those differences be viewed as "ethnic"? Most archaeologists would agree that the material culture and behavioral differences between the Pre-Dorset and the Maritime Archaic are so marked that they were distinct social identity groups, but it is debatable whether understanding of their interaction can be enhanced by invoking ethnici- ty theory. In any event, one could say that pronounced ethnic differences may not have existed in northern Labrador prior to the Pre-Dorset colonization ca. 4000 B.P.. Ethnicity only became an emergent relation in a culture contact situation structured by marked differ- ences in language and behavior, and potentially com- petitive land-use relations. Whether or not ethnicity concepts are involved, the interaction situation must be understood in terms of two main dimensions. First, there must be ecologi- cal parameters focusing on the energetics of boundary maintenance. Population interaction models from evo- lutionary ecology may direct us towards alternative modes of resource partitioning with different out- comes for inter-cultural relations. Second, there must be social parameters involving the different organiza- tional and cultural practices used by the two societies to structure boundary maintenance strategies. Overall, how did the cultural construction and use of the land- scape articulate with these two dimensions of human action? Here we have a set of seemingly incompatible or incongruent themes— one ecological and material- ist, the other social and constructionist— that network themselves around a common problem. How do we go about constructing interpretive narratives or theory in this or similar situations? The next section lays out some of the issues that draw our local problem into the heart of recent debates in archaeological theory. WORKING AT THEORETICAL BOUNDARIES: "LANDSCAPE" AS A PLATFORM FOR DIALOGUE? After over 1 5 years of often heated debate and mutual dismissal tactics between what we have come to call processual and post-processual archaeologies, a 322 CHAPTER 14 series of statements have appeared that call for some degree of convergence, reconciliation, or modus Viven- di. Schiffer (2000) for example, advocates "building bridges" between different conceptual structures. Following Galison (1996), Wylie (2000) refers to inter- field "trading zones." Hodder (2001) and Preucel and Hodder (1996) argue for a multi-perspectival plural- ism, while Schiffer (2000) sees this as resulting in potentially dangerous fragmentation and eclecticism. Despite the proliferation of various archaeological sects since the 1980s and the groping towards some fusion of interpretive horizons, much of the disci- pline's discourse still seems to migrate towards poles defined by "processual" or "post-processual," which have now become firmly entrenched categories of research history as well as axes of social organization within archaeology. Tendencies towards polarization have been aided and abetted by clinging to resistant first principles that are played out as a series of explicit or implicit concep- tual dualisms which, when assembled, constitute two competing ontologies of archaeology. One ontology privileges the view of humans as biological beings and is founded on adaptation as its central metaphor, from which a series of more specific causal theories is derived. The other ontology privileges the view of humans as thoroughly cultural beings, suspended in webs of significance they themselves have construct- ed. Language is the dominant metaphor, from which subsidiary interpretive theories of meaning are drawn. Thus, the Nature/Culture opposition is the eminence gris lurking behind much of the debate. Attempts to move beyond the sterile processual/ post-processual debates require working at overcom- ing the troublesome Nature/Culture dualism. A recent review article by Anschuetz et al. (2001), who write from an American "liberal processual" perspective, suggests that a "landscape paradigm" can bridge the great divide between processual and post-processual views. The authors are careful to define this approach as a "methodological paradigm": a set of tools that can be used in solving common puzzles (as perMasterman 1970:70), thereby leading to accommodation or inte- gration between different theoretical frameworks (Anschuetz et al. 2001:160-164, 191). Besides the residual odor of "unified science" views (cf., Wylie 2000), which are highly problematic, this "integrative" methodological approach attempts to insulate practi- cal action from ontological issues and deliberately skirts the difficulties of bridging radically different conceptual structures. Research in cognitive history of science and on conceptual change in general suggests these issues cannot be ducked so easily (e.g., Chi 1992; Giere 1994; Kuhn 2000; Latour 1993; Nersessian and Andersen 1997; Thagard 1992). Writing from a different perspective, Tilley's (1994) landscape phenomenology has also argued for overcoming the Nature/Culture opposition. It is diffi- cult to quarrel with Tilley's statement that: "...the rela- tionship between people and it [landscape] is a con- stant dialectic and process of structuration: the land- scape is both medium for and outcome of action and previous histories of action. Landscapes are experi- enced in practice, in life activities" (Tilley 1 994:23). But Tilley's landscape phenomenology is positioned in strategic opposition to adaptationism, so he takes a thoroughly constructionist view of landscape struc- turation: "Landscape is a signifying system through which the social is reproduced and transformed, explored and structured — process organized. Landscape, above all, represents a means of conceptu- al ordering that stresses relations" (Tilley 1994:34). With Tilley it is the meaningful and symbolic that is privileged, with practice and life activities exemplified by Australian hunter-gatherers moving through a land- scape of sacred places encoded by Dreamtime mythol- ogy, but completely without reference to practices and environmental knowledge that facilitate subsistence. To adapt a phrase from Tim Ingold, Tilley treats Nature primarily as "...raw material for imaginative acts of world-making" (Ingold 1996:1 50). The landscape is so good to think we don't need to worry about eating. THE STRUCTURATION OF MARITIME ARCHAIC/PRE DORSET SOCIAL BOUNDARIES IN LABRADOR 323 Ingold's (1996:150) critique of constructionist thinking argues that constructionism reinforces rather than challenges the ecological determinist models because its logic creates separate domains in which such models can be used. With a slice of the intellectual knife, two conceptual planes are created: (1 ) Nature, a space inhabited by biological organisms (animals) that forage to extract resources, and (2) Society, consisting of persons (thinking subjects) liv- ing in a culturally constructed landscape and engaged in the production and distribution of resources (Ingold 1996:127, 149). Ingold rejects the cultural constructionist argument that social rela- tions serve as a model for ecological relations because it requires a disembedding of practical engagement with the environment from the social relations that constitute humans as social actors, a separation of Mind from World, Nature from Society, Organism from Person and Biological relations from Social relations (Ingold 1996:126-128). Ingold sug- gests hunter-gatherers do not make such conceptual separations; instead, they have similar kinds of rela- tions with both the human and non-human con- stituents of the environment. In hunter-gatherer worlds, animals can be persons and humans have social relations both with other humans and with ani- mals, such that harvesting animal resources is not just a technical relation (Ingold 1996:127-131). Consequently, landscapes are not simply "socially constructed" by imposing a conceptual schema over Nature. Movement through the landscape involves engagement and learning, discovery rather than imposition of a predefined mental grid; practice in the environment constitutes persons and persons constitute their environments (Ingold 1 996: 1 38- 1 46). There is no radical break between ecological and social relations and "environments are constituted in life, not Just in thought, and it is only because we live in an environment that we can think at all" (Ingold 1996:1 50-1 51). While Ingold's argument is attractive, it is far from clear how such ontological reflection can be translated into research practice. The analysis of concrete archae- ological/anthropological problems may be difficult without some sort of fundamental Nature/Culture con- ceptual categorizations (Ellen 1996:29-30). Analytical arguments, as opposed to abstract ontological reflec- tions, require a choice of entry points, each of which commit the entrant to navigate certain directions along conceptual networks rather than others. The alterna- tive is eclectic incoherence or a discussion conducted at such a high level of abstraction that it would lose the very connection with human practice that Ingold views as the key issue. One of the hallmarks of "paradigm" shifts is the emergence of new concepts, changes in the meaning of old concepts and the shifting of old concepts into new systems of relations, sometimes by analogical extension, which involves major changes in lexical systems and results in incommensurability problems (e.g., Chi 1992; Kuhn 2000; Nersessian and Andersen 1997; Thagard 1992). For Ingold's (2000) perspective to become a viable alternative research program it is necessary to build up a new conceptual system that embeds his terms— such as "enskillment," "affordances" and "taskscapes"— within a semantic network of greater depth and precision that includes archaeologically relevant categories. Gamble (1999) has made a valiant effort in that direction, but the con- ceptual affordances for steering concrete research problems remam very limited. All things considered, during the current period of stalemate between competing ontologies of archae- ology there is something to be said for Anschuetz et al.'s (2001) advocacy of a landscape perspective. Although their view of landscape as an integrative methodological paradigm should be rejected, there is no reason why "landscape" cannot serve as a platform for dialogue, a point of departure for debating the problems and commitments that separate contempo- rary archaeologies. 324 CHAPTER 14 PARTIAL READINGS: NAVIGATING THE BOUNDARIES ...tension in theories, as in politics, is a creative force. ..we may see theories. ..as articulated within an intertextual field wherein they derive much of their mean- ing from what they oppose and thus, in a sense, confirm the importance of what they deny. ..Let us call this dissonance, facetiously, 'hypertension'... (Edwards 1994:261-262). If there are incommensurable frame- works or conceptual schemes, as Collingwood, Kuhn and others affirm, there could be no neutral overarching standards in virtue of which a rational adjudication between them could be made. Yet the mutual non-translatability of such schemes does not prohibit one from pursuing each in turn— much as a bilingual uses one language and then another— all the while constrained by such practical matters as the seemingly obvious requirement that one must speak one language at a time or by the different aims inherent in each frame- work (Harre and Krausz 1 996:2 1 7). Arguments that today's competing archaeologies are irrevocably incommensurate are exaggerated and serve mostly as a boundary maintenance strategy in theoretical identity politics. Despite significant concep- tual differences, competing "paradigms' must overlap to some extent, otherwise discussion between their respective proponents would be impossible (Bernstein 1983:85-93). Paradigms may be compared in various ways by moving back and forth between them in order to draw out elements of overlap and contrast (Bernstein 1 983:86, 90); Wylie (1 989) uses C.S. Pierce's "cables and tacking" metaphor to characterize these interpretive strategies at the boundaries of different frameworks. Working at these points of contact and contrast may help expose the processes through which intellectual boundaries are constructed. The opposi- tional tone between processual and post-processual archaeologies was understandable and perhaps inevitable, but such "hypertensional" rhetoric may serve as a cloaking device, disguising the process whereby each side "...derive[s] much of their meaning from what they oppose and thus, in a sense, confirm the importance of what they deny" (Edwards 1994:261). Also cloaked is how these conceptual boxes create artificial closures around what are best described (metaphorically) as idea networks. Such polemical closures are ironic for post-processualists who otherwise write of intertextuality: they ignore the consequences of their own constructionism. Intertextuality notwithstanding, navigating con- ceptual structures cannot be a random walk through a network. Choices of entry point are required and these choices have consequences for the direction of the entrants' navigation route. One cognitive-philosophical approach sees knowledge as structured in type hierar- chies, with different type hierarchies implying different ontologies about how the world is organized (e.g., dif- ferent type hierarchies for the adaptationist vs cultural constructionist ontologies). Since concepts at lower portions of a hierarchy inherit properties from super- types positioned higher in the hierarchy, activating concepts at any level of the hierarchy implicitly down- loads other properties of the ontological system (Aronson et al. 1995:27-54). This is not to say that such structures are governed by rigid necessity; if they are to function in practice and develop further they must be marked by an "open texture" that permits some flexibility in ordering (Harre 2002:264). But it does suggest that working at the boundaries of ontolo- gies and attempting to develop new "conceptual blends" (Fauconnier and Turner 2002) cannot escape the dilemma of conceptual inheritance that is an auto- matic consequence of choosing a starting position. THE STRUCTURATION OF MARITIME ARCHAIC/PRE-DORSET SOCIAL BOUNDARIES IN LABRADOR 325 Considering the multitude of problems attendant with attempts to build new theory at the boundaries between current archaeological ontologies, the follow- ing discussion of theoretical models is presented in two parts: ecological space and culturally constructed social and ideological landscapes. Selected aspects of these models are used to frame the subsequent cul- ture-historical narratives. Although this mode of pres- entation remains trapped in the Nature/Culture dual- ism, each model and narrative is considered a partial reading, using one language at a time, launched from entry points that activate different conceptual domains relevant to human practices in the landscape. ECOLOGICAL SPACE From the perspective of the humans as adaptive beings ontology, the landscape is an ecological space with varying distributions of resources from which foragers must make rational selections in order to reproduce or meet social obligations. Interaction between popula- tions of organisms inhabiting the same ecological community and engaged in foraging for similar resources (niche overlap) leads ecologists towards competition theory. Two main types of competition have been identified: exploitation and interference competition. In exploitation competition, foraging by one organism reduces the quantity of a resource need- ed by another organism. In interference competition, one organism reduces the ability of another to make use of a common resource, through mechanisms such as territoriality (Carothers and Jaksic 1984; Keddy 1989). Although the first type is the most traditional view of competition, the second has become increas- ingly significant to researchers and seems more rele- vant for discussing human practice in the environment, so interference competition will be the focus here. Another approach to niche overlap— mutualism— has been less theorized in ecology (perhaps for ideological reasons,^ ^ Keddy 1989), but may have considerable relevance for human practice (Spielmann 1986). Competition theory has been marked by the extensive use of mathematical models. Most of these are con- cerned with the relationship between resource avail- ability and population dynamics in situations of niche overlap. The following discussion will not be con- cerned with this quantitative dimension and popula- tion dynamics will play little role in the discussion. Much of the focus will be on spatial dynamics, since this has the most visible archaeological consequences. The central element in competitive or mutual relations concerns how organisms partition resources and habitats. Partitioning can be accom- plished by scheduling the use of foraging sites at dif- ferent points in time, by dispersing foraging sites in space, or by shifts in foraging behavior towards spe- cialization on different resources. Preventing tempo- ral overlap at foraging sites can minimize the poten- tial for direct conflict in interference competition, such as fighting over the best fishing or sealing loca- tions, stealing food etc. Temporal partitioning of the same foraging sites may be accomplished by the potential competitors selecting resource types that are available at different times, or by exploiting con- tinually renewing resources at different times. An effective temporal partitioning of activity may not require or result in the partitioning of other variables, such as habitat space or particular food resources (Carothers and Jaksic 1984:405-406). Some highly desirable resources, however, may be available only for short time frames (e.g., seasonally), decreasing the likelihood that temporal partitioning would be an effective means of limiting competition. Partitioning of space is perhaps the most frequent means of preventing direct competition. In patchy ' ' "Competition" is a category in need of deconstruction. Its conceptual privileging relative to other relational possibilities- such as mutualism— may reflect its metaphorical role in the intertextual relations between biological and sociopolitical dis- course (Keddy 1989:160-162). Keene (1983) discusses the sociopolitical baggage of evolutionary ecology in archaeology. 326 CHAPTER 14 environments where resources are dispersed, dispersal of foraging sites in space minimizes competition. Of course, patches may differ markedly in their quality of resources such that competition may arise from forag- ing decisions to prioritize high quality patches. The likelihood of competition increases if such resources are also temporally limited in availability. For example, seasonally migrating harp seals have, in a general sense, a patchy distribution. But they may have higher accessibility to human foragers in certain predictable places, such as narrow inlets and passages that restrict and funnel herd movements. Such places may become the focus of competition. One of the central (and traditional) issues in spa- tial partitioning is territoriality. As humans use such a wide range of interaction mechanisms it is very sim- plistic to categorize behavior as either territorial or non-territorial (Dyson-Hudson and Smith 1978). One might say there is a continuum of territorial behaviour from highly territorial competitive exclusion, to coop- erative interaction as non-territorial behaviour. Dyson- Hudson and Smith (1978) regard territoriality as a response to interaction situations in which resources are dense and predictable. The articulation of these two properties promote territoriality by setting condi- tions of economic defendability. Kelly's (1 995) discus- sion of territoriality among foragers builds largely upon their observations. Although he gives lengthy consideration to the social mechanisms involved a group's extending permission to others to use their territories (Kelly 1995:181-189), his arguments turn on the centrality of resources and demography, seen in terms of cost-benefit analysis: "...for any given case we should be able to translate the interplay between envi- ronmental and population variables into their signifi- cance in terms of intragroup variance and intergroup correlation, and from these predict the specific form of land tenure" (Kelly 1994:201). Although Kelly is sensi- tive to the social mechanisms involved in regulating access to the landscape, his view of "land tenure" is premised on a reductionist attempt to predict social relations from ecological relations (Ingold [1986:130- 1 33] provides an alternative conceptualization of these terms, see below). It should be noted, however, that most of Kelly's ethnographic examples pertain to boundary processes between social units within single ethnic/linguistic units, rather than across such units. To this point the discussion has focused on the properties of available resources, but variations in the behavioral properties of the competitors must also be considered. Competitors may exhibit varying techno- logical capacities for resource extraction or differences in organizational characteristics, such as labor mobi- lization or social differentiation. These abilities may change over time, resulting in either shifts of compet- itive advantages/disadvantages, or the development of coexistence. In the ecological literature, the relative degree of specialization/generalization in resource exploitation (niche breadth) is a frequently used dimension of variation.^ ^ In some situations both competitors may specialize, thereby minimizing niche overlap and potential competition. If two populations specialize on different resources we might expect the development of relatively weak territoriality, or per- haps even reciprocal resource exchange in a mutualis- tic relation between the two populations (cf., Spielmann 1986). In other situations one competitor may shift to more generalist strategies, exploiting the resource patches not used by the specialist. Depending on the degree of niche overlap, this could be an unstable coexistence marked by some degree of territoriality. The specialists might be expected to invest in defending their resources since they have less alternative resource flexibility than generalists and '° How such a distinction could actually be measured empirically among human foragers is a serious problem, particularly in archaeological cases for which we lack zooarchaeological data, and draw subsistence inferences primarily from site loca- tion, such as in the Pre-Dorset/Maritime Archaic context. THE STRUCTURATION OF MARITIME ARCHAIC/PRE-DORSET SOCIAL BOUNDARIES IN LABRADOR 327 thus much to lose. The generalists would be unlikely to incur defence costs for gaining only a small increase in resource variety unless the niche overlap with the spe- cialist included resources that were seasonally critical for the generalist. In all these cases, as long as the interaction effects are fairly equal for both parties, the competition could be termed symmetric (Keddy 1989:1 5-16). But an inability to shift niche breadth or foraging characteristics may militate against coexistence, lead- ing to exclusion of one of the competitors. Marked organizational differences between populations raise the possibility of competitive dominance, or asymmet- ric competition (Keddy 1 989: 1 6). Asymmetric relations might involve avoidance strategies on the part of sub- ordinate populations, such as: a) refuging behavior, with movement to poorer quality habitats or favored retreat sites, perhaps linked with an increase in spe- cialization, b) a "floating-nomadic" existence, moving through the interstices between the favoured locations of the dominant population, perhaps involving a shift to more generalist strategies, or c) shifts in activity times, such as the subordinate exploiting a dominant's favoured resource during a non-optimal period (Carothers and Jaksic 1984:405; Schoener 1974b; Smith 1978; the latter two cited in Carothers and Jaksic 1984:405). Dominance patterns may also be related to the temporal dimension of niche relations. Yodzis (1 978:25) discusses spatial pre-emption resulting from the history of colonization sequences. It is generally assumed that already established organisms have a competitive advantage over new colonizers, resulting in what Yodzis (1 978:26) calls founder-controlled com- munities. In such communities, colonization by new organisms may involve a "reshuffling" of species among patches, a rearrangement structured by founder-dominance, which results in re-equilibration. In some cases, however, repeated colonization events may produce what Yodzis (1978:49) terms "quasi- cycles": a series of colonizations without equilibrium. This discussion is important in that it draws attention to the problem of historical contingency in competitive relations and highlights the static nature of the stan- dard equilibrium models. Another dimension that may lead away from sim- ple equilibrium models is temporal variation in the environment, which may be linked to variable compe- tition over time (Wiens 1 977). In general, there should be limited competition when resources are abundant and maximal competition at resource lows. In environ- ments marked by strong spatial variations in resource availability, periodic resource minimums will lead to greater competition for the limited high quality resource patches or extraction locales. Although for modelling purposes temporal variation in resource availability has often been depicted using shifting equilibrium models, cycling between high and low abundance, more recent non-linear systems ecology draws in the abrupt and chaotic aspects of environ- mental change. This puts a premium on the historical- ly contingent in Nature/Culture relations. For the purposes of further discussion related to the specifics of the Maritime Archaic/Pre-Dorset social boundary issue, three aspects of spatial partitioning are extracted for particular attention. First, "territorial- ity," involving either: a) symmetric relations or mutual- ism, or b) asymmetric relations marked by the relega- tion of one group to less favored patches or refuging behaviour. Second, "floating" strategies, involving inter-patch mobility through cultural interstices to avoid contact. Third, "cycling," involving repeated col- onization events on varying time scales within regions, either as a partial avoidance strategy or simply failure to colonize permanently. CONSTRUCTED SOCIAL AND IDEOLOGICAL LANDSCAPES From the perspective of the humans as intrinsically cul- tural beings ontology, space is not a neutral resource foraging arena, but a culturally constructed landscape imbued with social and ideological meanings. Instead 328 CHAPTER 14 of locations or sites that are the settings or physical frames for activities— simply utilized spaces— land- scapes are actively constructed and experienced by minds and bodies. Natural spaces become cultural landscapes by being enveloped in conceptual grids and a history of practices as well as through the con- struction of built environments that physically and conceptually act back upon the humans who experi- ence these places. The central point is that in the course of social practice, agents "draw on their set- tings" as resources for action and, in turn, that places constitute agents as social actors (Giddens 1979:206; Tilley 1994: 16,19-20, 23). The "existential spaces" in which agents act are not static, but are continually pro- duced and reproduced through practices involving movement through the landscape (Tilley 1 994: 1 6-17). Thus, the biographies and identities of individuals and groups are constructed in relation to places and in the course of movement along the paths that tie places into meaningful networks, and these relations between place, people and movement are central to the con- struction of cultural memory (Tilley 1994:27-28). The mutual structuration of practices and networks of places gives a spatial dimension to cultural knowledge and power relations (Hood 1988). Some of the best examples of this approach applied to the study of hunter-gatherers come from Australia (for other examples see Bender 1 993; Hirsch and O'Hanlon 1995; P.Jordan 2003; Tilley 1994; Ucko and Layton 1999). Morphy's (1995) discussion of the Yolngu makes clear the relationship between land- scape, the ancestral past and social reproduction. All parts of the Yolngu landscape are imbued with the actions of ancestral beings such that the landscape consists of ancestral tracks and mythological coordi- nates anchored to specific places. The ancestral past becomes part of subjective experience as the individ- ual moves through and experiences the totemically coded landscape. Kinship relations (moieties and clans) are mapped onto the landscape and thus inte- grated with the ancestral grid, and rituals relate groups and places in relation to ancestral tracks. Thus, a set of long-term structures encoded in the landscape constitute a framework for active social structuration processes: The ancestral past is continuously re-cre- ated by the sedimenting of past and present experiences and political outcomes on pre- existing loci. ..In reality places are continually being reformed into new sets. New divisions of the landscape are made as clans die out and new ones emerge. ..Thus the articulation of social groups with the landscape is always changing, but the mythic screen that covers the landscape makes the relationship appear unchanging. ..The acting out of individual lives ultimately produces the cumulative changes which force the ancestral screen to adjust in order to mirror present circum- stances. ..The ancestral past, though changed and reproduced through present human action, is absorbed as a precedent for future action (Morphy 1995:204-205). The aforementioned perspectives provide impor- tant analogies for thinking about hunter-gatherer action in the landscape. Yet, as discussed previously, they present a rather one-sided cultural constructionist view that generally ignores resource extraction prac- tices or treats them as epiphenomenal to meaning. The consequence is an extremely idealist view of human relationships with the landscape in which Nature is viewed primarily as a source of raw materials for cul- tural acts of imagination (Ingold 1996:150). But the landscape needs to be both good to think and a source of raw materials for cultural acts of survival. If we take more seriously the mix of constraints and opportunities afforded by the natural world and different types of mobility systems, then we might identify some general patterns or at least some gener- al consequences for cultural construction of the land- scape. For example, there is surely a difference in land- scape construction practices between populations that THE STRUCTURATION OF MARITIME ARCHAIC/PRE DORSET SOCIAL BOUNDARIES IN LABRADOR 329 have been established in an area for many generations and those which are initially colonizing an area. As in the Yolgnu example, established populations operate within a pre-existing network of places and paths that are implicated deeply in the reproduction of social relations and cultural memory. The landscape is a long-term structure that is both the product and prece- dent of action. Colonizers enter new landscapes for which there is no pre-existing cognitive map, no pre- cise knowledge of resource distributions, no coded network of known places or paths to relate to. They must act without structural precedents, except insofar as they can extend analogies from previous experi- ences elsewhere. Consequently, besides engaging in flexible resource-searching and knowledge-gathering practices, colonists must actively construct a land- scape by building up an archive of experiences with places and paths. These differing implications for social action in landscapes have obvious relevance for the Maritime Archaic/Pre-Dorset situation. Another argument relevant to a critique of the constructionist privileging of Culture is Ingold's (1 986: 1 30ff.) discussion of the concepts of tenure and territoriality, which stand in a dialectical relation that illustrates what is involved in working with the Nature/Culture dualism. Ingold (1986:131, 136-137) considers tenure to be the schema and practices through which the landscape is appropriated, claims are made on resources and individuals are constituted as agents. Tenure binds groups into historical relation- ships with landscapes, which are reproduced as long- term structures. Hunter-gatherer tenure is a relation- ship with places and paths through the landscape— a dispersed network— not a partitioning of space into two-dimensionally bounded patches, which is the usual thinking with "territoriality." For Ingold (1986: 1 30-1 31, 136, 141-145, 156-157), territoriality \s x.h& schema and practices used in appropriating resources and for communicating about the locations of persons and resources. Although it does involve divisions of the landscape, it is not about maintaining exclusive access to resources, but is a set of practices aimed at minimizing interference in resource extraction or the disturbance of sacred sites. Unlike the structural conti- nuity of tenure over time, however, territoriality is sit- uationally contingent since different modes of territo- riality are implemented in response to changes in envi- ronmental (and social) contexts. To summarize: "...tenure engages nature in a system of social rela- tions, territoriality engages society in a system of nat- ural relations" (Ingold 1986:136). Such a brief summary does not do complete jus- tice to Ingold's ideas, but in conjunction with the pre- vious discussion it outlines a framework that can be used to stimulate thinking regarding questions of landscape and social boundaries in Labrador. Details of the archaeological context for Maritime Archaic/Pre- Dorset relations are presented in the next section, which is followed by an interpretative integration of the archaeological context with the threads cast out in the theoretical sections. BACK TO THE ARCHAEOLOGICAL CONTEXT We know that the Maritime Archaic and Pre-Dorset shared the north-central Labrador coast for several hundred years and preliminary interpretations have pointed to spatial partitioning as one of the mecha- nisms by which that coexistence was sustained. The preceding theoretical discussion established frame- works for understanding that coexistence from differ- ent entry points to human use of the landscape. Now, the empirical aspects of Maritime Archaic and Pre- Dorset settlement need to be considered in greater depth. The foraging characteristics of the two cul- tures are difficult to ascertain, since we completely lack archaeozoological material from Pre-Dorset sites and that from the Maritime Archaic consists mostly of calcined bone fragments. Subsistence-settlement inferences are therefore drawn primarily from site locations, structural information and lithic assem- blage variation. 330 CHAPTER 1 4 229/ Distribution of Late Maritime Arclnaic and Pre-Dorset Settlement in the Nain Region. Original map © 2004 produced under licence from Her Majesty the Queen in Right of Canada, with permission of Natural Resources Canada. Broad similarities in site location have led to both cultures being characterized as having "interior-mar- itime" settlement systems, which involved spring to fall land-use ranging across the inner and outer coastal islands and inner bays for sea mammal hunting, fish- ing and caribou hunting, and winter occupation of the inner bays or near-interior based on caribou hunting and fishing (Cox 1978:102; Fitzhugh 1978:83-84). As of yet, however, none of the inner bay sites known for either culture provide convincing evidence for winter occupation. The discussions of the Late Maritime Archaic site Attu's Bight in Chapter 12 and the Pre- Dorset site Attu's Point in Chapter 1 3, both inner bay localities, highlighted the difficulties of drawing sea- sonal inferences. Nonetheless, the settlement data are sufficient to make the case that there was potential for resource competition between the two cultures. On the other hand, the settlement typology is static and insen- sitive to possible organizational changes over time. As will be discussed further below, Early/Middle Maritime Archaic settlement patterns may have differed from those of the Late Maritime Archaic. For Pre-Dorset, the paucity of radiocarbon dates makes it impossible to track temporal shifts in the configuration of their set- THE STRUCTURATION OF MARITIME ARCHAIC/PRE-DORSET SOCIAL BOUNDARIES IN LABRADOR 331 tiement patterns, although regional variations can be considered. The following pages outline the settlement pattern data on a region-by-region basis. The Nain Region Figure 229 depicts the distribution of Late Maritime Archaic and Pre-Dorset sites in the Nain region.^ ^ One concentration of Pre-Dorset sites is found in the outer island area. Most of these outer coast sites are small, consisting of lithic scatters with and without traces of dwelling structures. On Natsutuk Island, in the outer island fringe, the September Harbour-1 site is a rela- tively large locality with repeated occupations, although few clear structures (Fitzhugh 1976a:136; Fitzhugh, personal communication). On Dog Island, Dog-Bight L-5, dated 3810±75 B.P., contains three closely spaced axial structures, but very few lithics (Cox 1978:99, 101; Fitzhugh 1976:130-133). Finally, one of the few late Pre-Dorset sites in the Nain region is Nukasusutok-2 (Chapter 4), which consists of two substantially built axial structures, one of which is dated 3315±85 B.P. (3055±85 with ^^C/'^^C correc- tion; Cox 1978:99, 101; Fitzhugh 2002). The other concentration of Pre-Dorset sites is found along Port Manvers Run, the long passage west of South Aulatsivik Island. The Run can be used as a protected "inside" travel route to avoid rougher outer coast waters. In spring and fall, migrating harp seal herds may be funnelled through the Run and during the winter strong tidal currents maintain a polynya at the "Second Rattle" that can be used for open water sealing. Caribou of the Kingurutik and Kiglapait Mountain herds may be accessed from the Run and small herds may move across the Run onto South Aulatsivik Island during the winter. Three large Pre- Dorset sites with multiple localities indicative of fre- quent re-occupation are found on or near the Run. At the seaward end of the Run, Thalia Point contains sev- eral localities, one dated 3660±140 B.P., which are marked by a high frequency of burins and burin spalls (Fitzhugh 1 976b:l 06-1 07). About one third of the way southward along the Run is Approach Point-2, located adjacent to the polynya. The site consists of at least a dozen localities, some with ambiguous structural remains, two of which have fairly substantial accumu- lations of tools. The location is favorable for spring and fall sealing as well as winter open water sealing in the polynya. Oddly enough, harpoon endblades are rare in the surface material; a considerable amount of slate debitage is present (Smithsonian field notes, 1980). The third site is Attu's Point on Webb Bay, slightly west of the entrance to Port Manvers Run. As outlined in Chapter 1 3, Attu's Point contains over 20 small localities, including several dwelling structures, two of which are dated 3750±60 B.P. and 3790±70 B.P.. The lithic assemblage is marked by a fairly high frequency of endblades, abundant slate materials and a paucity of scrapers. Its location and assemblage characteristics suggest a focus on spring and fall seal hunting at a nearby narrow channel. Most of the other sites along the Run are small, with single or two or three localities. Two of these. Port Manvers Run-1 and Double Island Cove-1 (the latter with dates of 3640±70 B.P. and 3320±70 B.P.), were described in Chapter 12. To summarize, there are two frequently re-used "central places" on the outer coast, at September Harbour and Thalia Point. A third central place is found at an inner bay location, Attu's Point, while a fourth, ' " The site distribution information is derived from the files of the Cultural Heritage Division, Government of Newfoundland and Labrador. Several caveats must be noted regarding these data. (1) The distribution data are most reliable for the Nain and Sagiek Bay regions, since these areas have been subjected to the most intensive and long-term survey programs. (2) Generally speaking, the inner bay areas have not been surveyed as extensively as the outer coastal regions. (3) There are problems in separating Late Maritime Archaic sites from Early and Middle period sites, since chronological information has not been entered systematically in the site files. Site selection was based on preliminary reports, personal experience, archae- ological oral history, and elevation above sea level. 332 CHAPTER 1 4 Aunospheht diu from Reimer el al i^fX^JjinxCal v3 HI Rroiik R.imvey l2fX)5), ^.iibr S >d ! 2 |iriih iisp|i-hrtiii| Attu's Bight 408()±1()()BP Attu's Bight 4080±90BP Dog Bight L5 3810±75BP Attu's Point 379()±7()BP Attu's Point 3750±60BP Quest Cove- 1 376()±9()BP ThaHaPoint-2 366()±14()B^ Double Is. Cove 364()±70BP Double Is. Cove 332()±70BP Nukasusutol<-2 3055±85BP 1 1 1 1 1 1 i 4000CalBC 3000CalBC 2000CalBC Calibrated date lOOOCalBC 230/ Calibrated radiocarbon date probability distributions for Late Maritime Archaic and Pre-Dorset sites in the Nain region. From OxCal 3. 1 0. Approach Point, lies in an intermediate position on Port IVIanvers Run. As a tentative proposal for a "local" settlement pattern within the Nain region, one might treat the Thalia Point, Port Manvers Run, Webb Bay area as a north-south linear settlement system. The abun- dance of Pre-Dorset sites along this stretch suggests the Run was an important "highway" between the outer coast and the inner bay zone. Winter settlement might be suggested at the northern seaward end near Thalia Point, which is close to the sina, and at Approach Point, near the Second Rattle polynya. In spring, Pre- Dorset people may have shifted to the southern anchor of the system at Attu's Point, exploiting seals during and after breakup, hunting caribou moving back to the mainland from South Aulatsivik Island and replenish- ing wood supplies. Summer might see a shift back to the outer coast, then another return to Webb Bay in the fall for wildfowl, seals, fish, caribou and wood sup- plies. The concentration of Pre-Dorset sites in the outer islands might be linked to a similar outer coast/inner bay movement, perhaps through the area south of South Aulatsivik Island and west towards Nain Bay, but multi-season occupa- tion of the outer island region is also a possibility. Although Early and Middle Maritime Archaic sites are abun- dant in the Nain region, Late period Rattler's Bight Phase sites are extremely rare. The only substantial locality is the Attu's Bight site on Webb Bay near the entrance to Port Manvers Run, only 1.8 km east of the Pre-Dorset central place at Attu's Point (Chapter 12). Although the Maritime Archaic site is extensive, the low fre- quency of tools and paucity of features suggest it was used as a short term transit camp for movements north and south along Port Manvers Run. A small Rattlers Bight locality is located on the outer coast at Ford Harbour (Paul Island), where there is also a trace of Pre-Dorset activity. The other late Maritime Archaic site relevant to the discussion is located south of the Nain region, slightly inland up the Kogaluk River, which empties into Voisey's Bay. This locality, Pukutakan Emish- shipu-2, has not been investigated in detail, but appears to be a camp on a portage route into the inte- rior (Labreche et al 1 997:1 35). This site suggests that in addition to the documented Maritime Archaic move- ments south-north along the coast related to Ramah chert procurement there was also an east-west coast- interior axis. Recent research in the interior at Lake Kamistastin indicates that Maritime Archaic activity in the interior may have begun as early as 7000 B.P., with possible traces of activity ca. 5000 B.P. (Loring 1999); this parallels finds from Lake Mushuau Nipi (Indian House Lake, Samson 1978) Table 95 summarizes the radiocarbon dates for Pre-Dorset and Maritime Archaic sites in the Nain region and Figure 230 presents their calibrated BC THE STRUCTURATION OF MARITIME ARCHAIC/PRE-DORSET SOCIAL BOUNDARIES IN LABRADOR 333 Table 95. Radiocarbon Dates for Pre-Dorset and Late Maritime Archaic Sites in the Main Region. SITE (CULTURE) DATE (B.P.) LAB NO. 16 Calibrated BC OxCal 3.10 REFERENCE Attu's Bight L-3/6 (MA) 4080±100 Beta-571 26 2860-2490 Chapter 1 2 Attu's Bight L-3/6 (MA) 4080±90 Beta-71477 2860-2490 Chapter 1 2 Attu's Point L-1 (PD) 3750±60'' Beta-7761 1 2280-2030 Chapter 1 3 Attu's Point L-5 (PD) 3790±70' Beta-77612 2350-2050 Chapter 1 3 Structure 1 (PD) Double Island Cove L-1 3640±70 Beta-571 25 2140-1910 Chapter 1 2 Structure 2 (PD) Double Island Cove L-1 3320±70' Beta-71476 1690-1 520 Chapter 1 2 Dog Bight L-5 (PD) 3810±75 SI-2521 2440-21 30 Fitzhugh 1 976a: 1 30-1 33; Cox 1978:99 Quest Cove-1 (PD) 3760+90 SI-4826 2340-2030 Clark and Fitzhugh 1990:301 Thalia Point-2-Al 9 (PD) 3660±140 CSC-1264 2280-1 780 Fitzhugh 1976b:107; Cox 1978:99 Nukasusutok-2 (PD) 3055±85(F) SI-2988 1430-1210 Cox 1978:99 * '-^C corrected (F) burned fat sample, 331 5±85 prior to '^C correction probability distributions. At one sigma there is no over- lap between the Maritime Archaic dates from Attu's Bight and the Pre-Dorset sites. A closer look at the archaeological contexts of the Attu's Bight and Attu's Point dates at least opens up the possibility of chrono- logical overlap, although it is important not to fall prey to wishful thinking. At L3/6 of the Attu's Bight Maritime Archaic site there was a faint trace of an occupation layer 20 cm above, and therefore possibly later than, the dated 4000 B.P. component, although deflation and redeposition processes in the sand matrix have to be taken into account. At the Pre-Dorset site Attu's Point, the two 3700 B.P. dates are both from structures on the lowest 10 m beach level, while additional features are found up to and including the uppermost 12 m beach, suggesting that some of these features could be earli- er than the radiocarbon dated loci. Additionally, the uppermost 12 m beach of the Pre-Dorset site had sur- face finds of a Late Maritime Archaic stemmed point and a possible Maritime Archaic biface. Although these implements might hint at some form of cultural con- tacts, they could also be traces of an earlier Maritime Archaic occupation or items Pre-Dorset people scav- enged from a nearby Maritime Archaic site. In any event, the stratigraphic factors suggest that chronolog- ical overlap between the two sites cannot be ruled out. If so, we would have two "central places" within a 1.8 km distance of each other. Overall, however, the evi- dence for contemporaneity is thin. Although the discussion of Pre-Dorset settlement patterns assumed a year-round presence in the Main region, there is no reason to suggest that the sparse evidence for Late Maritime Archaic occupation should be interpreted similarly. It is quite conceivable that the Maritime Archaic sites Attu's Bight and Ford Harbour may represent only seasonal forays into the Main area 334 CHAPTER 14 23 1/ Distribution of Pre-Dorset sites in the OI