While the C-14 Method of dating has perhaps been the most important tool ever developed for prehistorians, the results of this technique have not been an unmixed blessing. This is especially true in the far north where wood charcoal, the standard medium for C-14 determinations in most other areas, is only infrequently recovered. Other substances—bone, ivory, blubber, and skin from sea mammals; sod, grass, and peat; bone and antler from terrestrial mammals; and occasionally driftwood or locally available twigs either burned or unaltered—have been submitted for radiocarbon analysis in the hope of unraveling the complex relationships between man and nature in the Arctic through the development of accurate absolute chronologies of events which characterized Eskimo prehistory during the last four or five thousand years.

Most prehistorians engaged in Arctic research have long been aware of gross inconsistencies in the results obtained from C-14 laboratories, and several specialists from these laboratories have offered suggestions for correcting our anomalous results. To date, however, no one seems to have taken the trouble to collect radiocarbon determinations from the Arctic and evaluate them with respect to the known but only partially explained discrepancies between dates obtained on wood charcoal and those obtained on other substances. The following attempt to do so for Arctic Canada and northern Greenland is an indication of what might be accomplished by a less indiscriminate use of radiocarbon dates.

Most of our knowledge regarding the Paleoeskimos of Arctic Canada is derived from the “core area” of Paleoeskimo occupation, a rough circle of some 1,000 km diameter including the coasts of Fury and Hecla Strait, Hudson Bay, and Hudson Strait. This is the area in which Dorset culture was first recognized, the source of most of our larger collections, and the area where continuity of development throughout the Paleoeskimo sequence has been demonstrated (Meldgaard 1962; Taylor 1968a; Maxwell 1973). The number and size of archaeological collections from this area suggest that it supported a larger Paleoeskimo population than did other regions of Arctic Canada, while the temporal distribution of components and continuities of style suggests that the region was occupied continuously throughout the Paleoeskimo period.

In the fringe areas surrounding this central core, continuous occupation has not yet been demonstrated through any major segment of the Paleoeskimo sequence. Work in these fringe areas has rapidly progressed during the past decade, and it now seems certain that most of the temporal gaps and cultural discontinuities are not the result of poor archaeological sampling but reflect a situation of sporadic occupation occurring at different times in different regions. One of the striking features of the Paleoeskimo population was its propensity for expanding and retracting its geographical range, and this is the phenomenon which this paper will attempt to document. The primary aim of the paper is to sort out who lived where and when; a secondary aim is to suggest how they may have got there and what happened to them.

Paleoeskimo artifacts from the Pond Inlet region have been known since Mathiassen carried out archaeological excavations in 1923, although for some time they were not recognized as such by their discoverer. A short description of the main sites known at the present time and of the most characteristic artifacts will be given here. Some conclusions will then be drawn concerning the local aspects of the Pre-Dorset and Dorset cultures, and their implications.

By a strange coincidence, the oldest dated remains of the whole region come from a point only 6 m distant from the Thule house excavated by Mathiassen at Mittimatalik (in which he found a few Dorset and perhaps Pre-Dorset artifacts) (Fig. 1). The site is situated on the west bank of a stream, in the middle of the present village of Pond Inlet, and is now covered in part by new houses and roads. In 1969, a pre-Dorset layer was found at 8 m above sea level, under modern, Thule, and Dorset remains. Underlaid by gravel and sand, it does not seem to have been too much disturbed by solifluxion, but the stratigraphy is confused by numerous crevices cutting through the sod and peat. Besides a few flint artifacts, mostly burins and burin spalls, a complete barbed harpoon head and a similar one with the point cut off were found. Seal bones from the same layer have provided a date of 4385±155 radiocarbon years (S-589), or 2435 B.C. (adjusted date: 2035 B.C.). Local conditions have made it impossible to carry out investigation of the site during the last four years.

Throughout the geographic and temporal range of dorset there is little yet documented to indicate major cultural changes. The Pre-Dorset part of the Paleoeskimo continuum is better known by its lithic artifacts, and to date too few sites have produced non-lithic ones to afford a clear picture of technological development. On the other hand, there is enough available information on the two periods to indicate that some non-lithic artifacts reflect a regularity in the shifting styles of discrete attributes over those thirty-five hundred years involved. Such minor style shifts do not readily appear functional in the sense of producing a more efficiently adaptive weapon, and some seem to entail more manufacturing energy coupled to greater risk of breakage without apparent gain in efficiency. Regardless of the reasons for these changes in style, they appear to occur simultaneously over relatively broad areas and appear to indicate an intensity of information exchange and cultural conformity within such areas. This is most notable within what has been called the core area (see Introduction): a region including North Baffin and Bylot Islands, the Fury and Hecla Straits, Southampton, Coats, and Mansel Islands, both shores of Hudson Strait, and southern Baffin Island at least as far north as Cumberland Sound. As McGhee (this volume) has suggested, there are periods in which this stylistic homogeneity extended into the High Arctic, and different periods in which the Central Arctic as far west as Victoria Island (but not the High Arctic) participated in many of these discrete style shifts.

Perhaps the most neglected aspect of archaeological research in the Arctic has been the within-site dimension of archaeological data. Few archaeologists have “attempted to see their material as remains left by social groups” (Anderson 1968:397) and fewer still have attempted to infer patterns of social organization and the within-site organization of activities. One measure of the degree to which Arctic archaeologists have failed to contribute to contemporary archaeological method and theory is the fact that the 37-page bibliography of Contemporary Archaeology: A Guide to Theory and Contributions recently edited by Mark Leone (1972) contains not one reference to Arctic archaeology. In part, this is because much of the contents of this collection of papers considers data from the southwestern United States and Mesoamerica, but it also demonstrates that Arctic archaeologists have contributed little to discussions of archaeological method and theory.

In a recent chapter on the development of Arctic archaeology (Dekin 1973), I expressed optimism regarding the potential theoretical contributions of archaeology in the Arctic, but this will not occur without a renewed emphasis on precise excavation, analytic sophistication, and a revitalized sense of “problem.”

This paper limits itself to a description of four complexes revealed by our excavations at Saglek Bay, northern Labrador, during the years 1969, 1970, and 1971. A “Pre-Dorset” complex and three phases of a considerably longer Dorset Eskimo occupation are examined and some comparison within the Saglek sequence and a few comparisons with nearby geographic areas are offered. No attempt is made to generalize about Paleoeskimo cultures in the entire Canadian Arctic, especially as regards causative relationships, since such attempts seem premature given our present lack of control over basic cultural and environmental data.

Over two dozen Paleoeskimo components were excavated or surface-collected during our three years at Saglek Bay. Analysis is now nearly complete and four representative components from these sites have been selected to illustrate four more or less distinct cultures or culture phases in this little known region of eastern Canada.

As sites were discovered at Saglek Bay they were given consecutive letter designations. The three sites considered herein are Upernavik Island Site K, a Pre-Dorset station, Rose Island Site E, which contained early and middle Dorset components, and Rose Island Site W, a late Dorset station. Each site location, the artifacts found therein, and evidence for chronology are summarized briefly below.

Paleoeskimo archaeology has a peculiar history in that it is best known by its more distinctive regional and chronological variants. Thus, we have had rather full reports from such edge-areas as Greenland (Knuth 1952; Larsen and Meldgaard 1958), Southampton Island (Collins 1956a, 1956b, 1957), and Newfoundland (Wintemberg 1939, 1940; Harp 1964). More recent fieldwork at Igloolik (Meldgaard 1960b, 1962), and full publication from northwestern Quebec (Taylor 1968a) and Lake Harbor (Maxwell 1973) have begun to close the information gap for the geographic core area of the Central Arctic. However, the continuing research by Taylor, McGhee, and Müller-Beck in the peripheral regions of Coronation Gulf, Banks, and Victoria Islands, by Nash and Harp in southern Hudson Bay, by McGhee on Devon Island, and by Tuck, Plumet, Fitzhugh and Linnemae in Labrador-Ungava and Newfoundland has made it even more necessary to tie these regions to developments in the demographically more stable central regions. A large number of questions, such as relationships with a fluctuating environment, population shifts, and regional interrelationships, hinge on more complete information for the Dorset core area.

One area that until recently remained completely unknown was the Labrador coast (Taylor 1964a). This 800-mile stretch of coastline carries a strip of tundra environment south from Hudson Strait to Newfoundland, and its rich marine resources, including a large seasonal harp seal population, provided a suitable habitat for early Eskimo culture, as well as the only feasible migration route from the northern regions into Subarctic Newfoundland.

The history of dorset eskimo archaeological research now spans a half-century and appears to be fully mature. At the outset, progress in this research was sporadic, but in the last two decades momentum has increased and we have achieved many significant clarifications of the general Dorset problem. These include a firm definition of Dorset people as true Eskimos, and further insights with respect to their temporal and geographic spread throughout the central and eastern Arctic, their localized adaptations to changing environments, and, as a persistent theme, their noteworthy cultural conservatism through time. This paper deals primarily with the middle and late stages of Dorset culture as seen in marginal settlements in western Newfoundland and southeastern Hudson Bay.

In reviewing past investigations of Arctic prehistory, I have been struck by our substantive preoccupation with material culture, mainly with artifacts. That is quite natural because our powers of interpretation and reconstruction are circumscribed in well-known ways by the paucity of our field data, particularly so as we are dealing with the cultures of Arctic hunter-gatherers. Therefore, we should be forgiven a natural tendency to submerge in the warm realities of artifacts and the comforting procedures of mensuration and taxonomy. These are legitimate scientific concerns, and they may lead us toward useful statistical formulations, definitions of cultural parameters and complexes, and so on. However, to the extent that we think solely in such statistical and materialistic terms, the fundamental human nature of our quest may be diminished, if not lost altogether.

Despite continued attention from archaeologists the origin of Dorset culture in the Eastern Arctic remains a persistent problem in northern prehistory. The transition from late Pre-Dorset to early Dorset culture is marked by a relatively rapid and consistent typological shift throughout a large geographic area at between about 1000 and 800 B.C. Although hampered by a distinct paucity of information, available chronological and distributional data do not supply a ready explanation for such consistent changes over such a wide-flung territory. However, the lack of a time slope and the presence of regional variants do not suggest that Dorset origins will be found in a single locale; rather, the transition appears to have occurred throughout a band of interacting populations within the core area of the eastern Central Arctic. Subsequent changes may be seen as a result of stylistic evolution, diffusion, and migration into more peripheral regions during the maximum extension of the Dorset sphere.

There can be little doubt that the majority of the area through which the Arctic Small Tool horizon spread was unoccupied by human populations. This is particularly true of the Eastern Arctic, with the exception of the southern fringes where they may have come into contact with Archaic peoples near the ecotone between the tundra and the taiga.

In the Western Arctic, it is apparent that there had been earlier human populations in many areas, but these later migrants appear to have occupied a previously unoccupied ecological niche. Geographically, they spread along the Arctic coasts until they had reached the maximum extent of seasonally frozen coasts, usually with adjacent tundra. While it may be that the rising sea level had flooded earlier evidence of such a coastal occupation arid that this habitat had not been unoccupied as it now appears, the present data support the inference that Arctic Small Tool populations were the first to accomplish a successful adaptation to these particular conditions in the American Arctic. For the present, I wish to focus on this coastal aspect of Arctic Small Tool peoples, ignoring (or not modeling) the expanded distribution of sites in interior Alaska which also occurred at this time.