The Dorset-Thule Succession Revisited

In Identities and Cultural Contacts in the Arctic, edited by Martin Appelt, Joel Berglund and Hans Christian Gulløv, pp. 192-205. Danish Polar Center Publication No. 8. The Danish National Museum & Danish Polar Center, Copenhagen. The Dorset-Thule Succession Revisited By Robert W. Park, University of Waterloo, Ontario - INTRODUCTION It is clear that the Thule people replaced the Dorset in Arctic Canada and Greenland. However, I am convinced that we do not really understand what happened when the first Thule immigrants entered the Eastern Arctic - did they encounter people of the Dorset culture, or had the Dorset population died out prior to their arrival? The overwhelming majority opinion among Arctic prehistorians has long been and continues to be that these two cultures overlapped for many years in parts of the Canadian Arctic and Greenland, and that some acculturation took place between them (e.g., Appelt, et al. 1998; Bandi 1969:149; Bielawski 1979:lOl;Cache1 1997:588; Collins 1950:21; 1984:16; Crowe 1974:17;de Laguna 1947:14; Dumond 1977:145; Fitzhugh 1980:601; 1994:259; Friesen 1999; GulLav 1996; Harp 1976:138; Hood 1998:46;Jenness 1925:437;Jordan 1979:154; 1984:548; Kaplan 1985:48; Mary-Rousseliere 1979:56; Maxwell 1985:278;1984a:372; McGhee 1984b:7;1989:97-98; 1990:672; 1996:135; 1997; Meldgaard 1960a:27; 1960b:589; Morrison 1983:279; 1999:150-151; Plumet 1979:116; Plumet and Gangloff 1987:86; Rowley 1994:371; Schledermann 1976:42; 1996:lOl-102;Shields and Jones 1998; Swinton 1972:lll; Taqon 198360; Taylor 1960:86; 1963:462; Thomson 1985:49;Wenzel 1991:26). Some years ago I attempted to challenge that accepted wisdom (Park 1993).In that paper I examined and rejected many of the specific pieces of evidence that had been put forward as proof of Dorset-Thule contemporaneity and acculturation. At the end I wrote "This paper will not represent the final word on the Dorset-Thule succession, and many researchers may not accept these conclusions" (Park 1993:226).Both predictions were correct. Morrison (1999:140)described my paper as "a provocative if extreme view" while Hood (1998:46) observed "Some archaeologists may regard Park's conclusions as overly sceptical, and it does seem to be the case that overlapping radiocarbon dates from some areas are fairly secure." McGhee's (1997:210)thorough response explains why he is convinced contact occurred - a very few Thule artefacts in Dorset sites -but he envisages a very brief and ephemeral kind of interaction between the peoples of these two cultures resulting in little or no acculturation, contrary to most earlier models (e.g., Bielawski 1979; Maxwell 1985:243-244,278). For the scholars who continue to posit DorsetThule overlap and acculturation, three lines of evidence seem to be most persuasive: (1) artefacts found in contexts suggestive of direct contact; (2) evidence of acculturation, specifically the adoption by the Thule of Dorset styles; and (3) the overlapping radiocarbon dates mentioned by Hood. Given the length constraints imposed on this paper and because I dealt extensively with the others in my previous publication on the topic, here I will focus on primarily on radiocarbon dates. ARTEFACT AND DATE SAMPLE CONTEXT Dealing just with Thule, Morrison (1999:140)notes that radiocarbon dates often do not correspond to stylistic data for a variety of reasons "of which site disturbance is probably chief". Such disturbance could be due both to cryoturbation, and to human activity. At some sites it seems clear that the Thule deliberately chose to re-use Dorset structures. Excavations of several Thule winter houses at the Tungatsivvik site at the head of Frobisher Bay by Douglas Stenton and myself have produced extensive Dorset lithic assemblages (e.g., Stenton and Rigby 1995).Test-pitting in other parts of the site looking for undisturbed portions of the Dorset component - in other words, between the Thule winter houses - has not been productive. Even though not particularly constrained by topography or geology, at Tungatsivvik the Thule appear to have built their winter houses precisely on top of every pre-existing Dorset structure. While we obviously cannot and should not assume that the Thule did this everywhere, it would be incautious to assume that deliberate or inadvertent Thule re-use of Dorset sites and structures has not profoundly complicated the stratigraphy of many sites. A vivid example of this process is perhaps provided by George Lyon, who was captain of the Hecla while it wintered at Igloolik in 1822. Near the end of July he visited what he called the "winter huts" there, which would be re-occupied that fall and winter. He described them as: ... so dilapidated, that we were enabled to see the interior zuifhout entering; which zuas not to be regretted, on account of the state they were in. The ground all around was strewed with skulls and skeletons of animals ... Bones indeed were so numerous, that we literally trod on them. A large stagnant field of mud surrounded the place, adding its full share of sweets, as it was constantly ploughed up by all who walked through it to the huts: the bottom of this also felt as i f covered with bones. @yon 1824:236) This kind of process could easily introduce later materials into stratigraphically earlier deposits at any sites that weren't completely unoccupied from spring through fall. I suspect that the stratigraphy of many of the multi-component sites discussed elsewhere in this paper has been complicated in this manner. ACCULTURATION Various archaeologists have proposed that a wide variety of technology passed between the Dorset and the Thule, including snowhouse construction, cold-trap entrance passages and raised rear sleeping platforms in winter houses, breathinghole sealing, soapstone vessels, bone sled shoes, some clothing styles, some harpoon-head styles, the use of ground slate, and the extensive use of iron. But for the traits believed to have passed from Dorset to Thule, with one exception either the Thule already possessed them before their departure from Alaska or the traits would have been entirely plausible and logical independent developments for the Thule to have made upon their arrival in Arctic Canada. The traits proposed as having passed from the Thule to the Dorset, such as cold-trap entrance passages, were already in use by at least some Dorset prior to the earliest possible Thule arrival (Park 1993:216-219). The only compelling evidence for Dorset-Thule acculturation, and therefore face-to-face contact, lies in the similarities between some Dorset harpoon-head styles and certain Eastern Thule harpoon-head styles that have no antecedents in Alaska and the Bering Strait region. These similarities were first noted and ascribed to Dorset-Thule contact by Collins (1937:315-316),and his conclusion has been reiterated numerous times since then. However, some of these same researchers correctly note that the Dorset-influenced Thule harpoon heads are not found at stylistically early Thule sites in the Canadian Arctic and Greenland. Instead, they are characteristic of substantially later Thule occupations in these regions (Jordan 1984:542; Taylor 1960:83).This seems incongruous if one assumes that the Thule adoption of these styles would have taken place in the context of Thule interaction with people of the Dorset culture and not several centuries after the Thule arrival in the Eastern Arctic, by which time the Dorset are supposed to have become completely acculturated or marginalized geographically. Curiously, the most parsimonious explanation appears to lie in both the contemporary archaeological record and in the archaeological record encountered by the Thule hundreds of years ago. Schiffer (1987:104,118-119)has used the term "salvage" to refer to "The process of reclaiming artifacts, including structures, from occupations by earlier peoples at a site." I conclude that the Thule were salvagers par excellence! In my earlier paper (Park 1993:221-225) I presented evidence that the Thule commonly salvaged certain kinds of Dorset artefacts, especially harpoon heads. This provided an explanation for why the Thule did not adopt Dorset styles until some considerable time after their arrival in the Canadian Arctic: the means by which they came into contact with Dorset harpoon styles was through salvaging actual Dorset harpoon heads from Dorset sites. Parenthetically, I should note that my explanation remains consistent with the recent models of McGhee (1997) and Friesen (1999)postulating overlap but minimal acculturation. Also, this idea - that the Thule acquired these Dorset harpoon head styles through salvaging Dorset artefacts - is in no way inconsistent with the kinds of ideas put forth by Gullav (1996; 1997) in which the harpoon head is seen as functioning as a kind of cultural symbol for its manufacturers. Thus, the similarity in harpoon head styles provides no proof of contact between these two cultures. RADIOCARBON DATING Amongst the scholars who continue to posit Dorset-Thule overlap and acculturation, the proof most often cited involves overlapping radiocarbon dates (e.g., Appelt and Gullm 1999x69-70; Fitzhugh 1994:259; Hood 1998:46).If we accept that the two most plausible models for the DorsetThule transition in the Eastern Arctic are (a) temporal overlap but with little or no acculturation (McGhee 1997); or (b) no temporal overlap (Park 1993), then radiocarbon dating becomes especially crucial in determining which model is more plausible. Some dates presently available do indeed suggest an overlap. However, McGhee (1999a; 1999b) has expressed doubt about the pre-12thcentury Thule dates in the Canadian Arctic so, for the purposes of this brief discussion, I will concentrate on Dorset dates late enough to plausibly overlap with Thule. As I have argued (Park 1993:206-208), most can be questioned on a number of grounds. Some of the sites or structures that have produced late dates have also produced much earlier dates it is therefore not clear if the late dates derive from (1) a very prolonged Dorset occupation of the site; or (2) a very late Dorset reoccupation of a previously abandoned site; or (3) an archaeologically unrecognized Thule occupation at an abandoned Dorset site; or (4) an aberrant date. Several of the remaining very late Dorset dates are the only dates obtained for that site or structure, for reasons of expense or because the goals of the researcher didn't require multiple dates. Both of these situations (i.e., where multiple dates from a site or structure differ considerably, or where a site's age is estimated from just one "C date) are examples where the 14Cdating technique is unable by its very nature to produce robust results. As noted by radiocarbon dating expert R.E. Taylor (1987:105): Unfortunately, it is often difficult to evaluate directly the various factors that could influence the accuracy of a single ^C value. For this reason, little reliance should be placed on an individual 14C "date" to provide an estimate of age for a given object, structure, feature, or stratigraphic unit. A critical judgement of the ability of ^C data to infer actual age can be best made with a suite of 14C determinations on multiple samples drawn from the same context or with multiple 14C determinations obtained on different fractions of the same sample (Waterbolk 1971 :19; 1983:18). Concordance of values on different sample types or fractions of the same sample from well-defined stratigraphic contexts provides one of the strongest arguments for the accuracy of age assignments based on 14C values. In the context of Arctic archaeology, Schledermann (1996:4-5)has reached the same conclusion. He writes: ... A side from these problems, radiocarbon dates remain a valuable tool, as long as the results are used with caution and i n conjunction with other data. A refinement of the radiocarbon dating technique now makes it possible to date much smaller samples. By using this "accelerator" technology, we can, in theory, date a variety of samples from the same hearth. This would not only give u s a better idea of the variance betzveen materials, but provide a better statistical basis for calculating the approximate period the hearth was i n use.. . . Dogmatic dependence on the result of one or two radiocarbon dates rarely advances the understanding ofprehistone events. When archaeologists have been able to apply this approach in Arctic research (i.e., dating multiple samples from the same structure), the results have been extremely useful in resolving specific chronological questions and in revealing the varying dating characteristics of different sample materials (e.g., Arundale 1981; McCullough 1989; McGhee and Tuck 1976; Morrison 1989; Tuck and McGhee 1983).Given the caveats expressed by experts in 14Cdating, I believe that we should concentrate our present interpretive efforts on sites for which multiple ^C dates are available. three hundred years? Surprisingly to many archaeologists, I suspect, we would certainly expect to see some overlapping dates - those Dorset dates producing results later than the actual age of the last Dorset sites, and those Thule dates producing results earlier than the actual age of the earliest Thule sites -but there would be a bimodal distribution in the dates overall. Conversely, what pattern of radiocarbon dates would we expect to see if the Dorset and Thule occupations had overlapped for one or more centuries? I contend that we should expect to see no major gap in the Dorset dates prior to the arrival of the Thule, and no overall bimodal distribution in the dates in the dates. Given these expectations, what do the dates from well-dated sites suggest? RADIOCARBON DATING EXPECTATIONS But before examining the actual dates, I think it is also worthwhile to think for a moment about what we might expect to find. Based on the principles underlying the radiocarbon technique and its well understood limitations, we should expect a large series of dates run on samples of identical age to produce a histogram in the form of a normal curve, centering on the actual age but with some dates considerably older than and some considerably younger (Figure 115a).A nice illustration of this can be seen from 17 dates from materials derived from Martin Frobisher's visits to Kodlunarn Island in Frobisher Bay (Figure 115b). Historical accounts allow us to know precisely when that site was created, but when looking at the ^C dates it is only the central tendency of the dates that is consistent with the known age of the site's occupation. Keeping these examples in mind, what pattern of radiocarbon dates might we expect to see if the Dorset and Thule occupations of the Arctic had been separated by two or Hypothetical series of 14C dates from a sample of age 0 x 0 A. THE DORSET AND THULE 14CDATES For my 1993 study I was able to assemble 383 dates run on terrestrial materials from Dorset and Thule sites in the Canadian Arctic and Greenland. Four implausibly early dates were eliminated from consideration, leaving 379 dates. Since that time many more dates have been run and, at least as important, Morlan (1999)has undertaken the monumental task of assembling all Canadian Martin Frobisher radiocarbon dates from Kodlunam Island Histogram Actual dates B. 150 Years more recent than 100 actual age 50 Actual Age 50 Years older than 100 actual age 150 200 -n dates I- n dates - A single calibrated radiocarbon date with muitiplc intercepts 1 Fig. 115. Hypothetical histogram (A) illustrating what we might expect from a large series of radiocarbon dates run on samples of identical age, compared with (B) an actual series of 17 radiocarbon dates from the 16"' century Frobisher expedition site on Kodlunarn Island, Frobisher Bay. Fig. 116. Regions of the Arctic providing the radiocarbon dates discussed in this paper. radiocarbon dates and making them available online. From that source and from published dates from Greenland, I was able to assemble 669 dates on terrestrial materials, excluding from my analysis only those same four implausibly early dates. As in my 1993 study, I found it useful to subdivide the dates geographically into subsets. For the sake of brevity I will focus on only two subsets, from regions where overlap and acculturation has been postulated (Figure 116).One of these regions encompasses Newfoundland, Labrador, Ungava, and southern Baffin Island, while the other includes Central and High Arctic Canada east of llOOW,and Greenland. This subdivision is useful because the kinds of materials dated differed substantially between these two regions. In the former, organic preservation is generally poor and only 3% of the dates come from bone or antler; almost 80% of the dates were obtained from wood or plant material, overwhelmingly charcoal. In the latter region, 56% of the dates were run on bone or antler, and only 35% derive from charcoal or other wood. All dates were normalised and then calibrated with CALIB 4.1.2 using the INTCAL98.14C database (Stuiver and Reimer 1993; Stuiver, et al. 1998). In this paper I will draw upon Maxwell (1985:217,239-241) and make a terminological distinction between "Late Dorset" which I will define as characterised by changes in harpoon head styles and some types of lithics and commencing perhaps A.D. 500, and "Terminal Dorset" which is mostly defined by very late 14Cdates plausibly overlapping the earliest Thule dates in the Canadian Arctic. Starting first with dates from sites in the region encompassing Newfoundland, Labrador, Ungava, and Southern Baffin Island, Figure 117 presents a series of histograms including all the 14Cdates obtained from Dorset and Thule sites in this region. These graphs show clearly the bimodal distribution of dates the dates between AD 1and 1600 that are ascribed to Dorset. The latter peak, ascribed to Terminal Dorset, coincides very closely with the relatively small number of dates ascribed by excavators to Thule. From this, archaeologists such as Fitzhugh (1994:241,259)have concluded that the Dorset and Thule co-existed in this region for centuries. Here, however, I'd like to reiterate my 1993 suggestion that, given the high degree of similarity in age ranges between the Terminal Dorset and Thule dates from this region, all the "C dates later than approximately AD 800-900 may in fact derive from the Thule occupation of this region. The archaeological rationale for that hypothesis relies on several factors specific to this region: intensive reuse of Dorset sites by the Thule; poor preservation of organic artifacts; extremely limited use of stone by the Thule; and poor stratigraphic separation between Middle Dorset, Late Dorset, and Thule occupations. Therefore, at a site occupied sequentially in Dorset, Thule, and historic times one might expect to find numerous Dorset flaked and ground-stone artefacts, an extremely small number of Thule lithics or none at all, and historic artefacts. However, datable charcoal from any of these components could be expected to survive. The evidence for this suggestion is outlined at length in Park (1993:208-213). Turning to an even more detailed analysis, and consistent with the principles outlined above concerning the interpretation of radiocarbon dates, I will focus just on thoroughly dated sites. The graphs in Figure 4 show all the 14Cdates obtained from all the sites for which three or more 14Cdates have been obtained from terrestrial materials. In other words, the only dates left out of this series All Dorset and Thule Dates from Newfoundland, Larbrador, Ungava, and Southern Baffin Island Fig. 11 7. Histograms illustrating all the radiocarbon datesjrom Newfoundland, Labrador, Ungava, and Southern Baffin Island. From left to right, the histograms illustrate just the Dorset dates, just the Thule dates, the Thule dates combined with the Dorset dates, and a hypothetical reconstruction of the cultural events represented by these dates. of graphs come from sites producing just two or one 14Cdate on terrestrial materials. The dates are expressed as 1-sigma ranges showing multiple intercepts. In the graphs I have not altered the cultural affiliation or context of the dates as assigned by the excavators, but I have included some dates that excavators later rejected as aberrant. Figure 118a shows all the 14Cdates obtained from such sites in Newfoundland, Labrador, Ungava, and Southern Baffin Island, illustrating even more clearly the bimodal distribution evident in Figure 117. Long and apparently continuous Dorset occupations in all these regions appear to terminate by around AD 800. In Newfoundland there are no later dates ascribed to Dorset. In Labrador, Ungava, and Southern Baffin Island there is a much later group of dates ascribed to Terminal Dorset. The best evidence in this entire region for continuity between the earlier and later series of dates is found at the Dia.4 and adjacent sites in Ungava, which I have previously discussed at some length, outlining the reasons why I believe that there is an unrecognised Thule occupation to which I ascribe this impressive series of dates (Park 1993:211-212).Significantly, I believe, this latter sequence of dates at Dia.4 is essentially identical to the series of Thule dates from sites such as Silumiut, Learmonth, and Eskimobyen (Figure 118b).At practically all the other sites producing such Terminal Dorset dates, they appear to coincide almost exactly with the Thule dates from the same or nearby sites. I will address this further below but for the moment I will simply assert that that these dates are all derived from the Thule occupation of this region, and have been mistakenly assigned to "Terminal Dorset" due to the very low archaeological visibility of the Thule archaeological record in this region (Park 1993:209-211). Turning to the radiocarbon dates from central and High Arctic sites, the situation is more complex for a variety of reasons. First, we have far fewer Dorset dates than Thule ones and those dates derive from a much wider range of materials, with presumably more diverse intrinsic dating characteristics. We have a good under- 1 Newfoundland 1 8 , assigned by excavator to: 500 Labrador , 8 0 I Unaava Southern Baffin Islan # 1 Sites from the Central and High Canadian Arctic, and Greenland I I l l -1 1 1 I I B. I l l I l l I I [1 Nunauvik Staffe Island 1 Qeaertaaraa 1 1 Tasiarulik I Bronze pot Fig. 118. Graphs showing all the radiocarbon dates on terrestrial materzalsfrom all sites producing three or more dates. In each graph the period between A D 800 and 1150 is indicated by the dotted lines. 0Dates from sites in Newfoundland, Labrador, U n p v a , and southern Baffin Island; 0Dates from sites in the Central and High Canadian Arctic, and Greenland; and (C)Dates from four sztes, showing the dates from individual structures at those sites. standing of one factor - marine reservoir effect that would result in some dates being consistently too old, although the exact amount of distortion cannot reliably be estimated (McGhee and Tuck 1976; Tuck and McGhee 1983).We attempt to avoid that problem by not dating marine mammal materials, but it is likely that in some well-preserved semisubterranean houses other materials such as wood could have become saturated with sea mammal oil and thus date earlier than their actual age. Another factor that would result in dates that are older than they should be is the use of old driftwood, and indeed some of the dated samples from the Learmonth and Cape Carry Thule sites have been identified as probably driftwood (McCartney, in Rutherford, et al. 1981:117). Finally, as I suspected some time ago (Park 1994:31) and as McGhee (1999b)has now pointed out in far greater detail, there may be features of the permafrost environment affecting non-marine dates as well - essentially a variable terrestrial reservoir effect (Taylor, et al. 1999) although its effects are probably considerably less than that of the marine reservoir effect. Figure 118b shows all the 14Cdates obtained from all the sites in the Central and High Arctic region outlined in Figure 116, for which three or more ^C dates have been obtained from terrestrial materials. Some Thule dates substantially predating AD 1000 are clearly aberrant, probably for the reasons outlined above. Although we have fewer Dorset dates than is the region to the south, at several sites - Qeqertaaraq, Nunguvik, Saatut, and perhaps Tasiarulik - we see the same pattern of an early series of dates ending around A.D. 800 separated by several centuries from a small later set of dates ascribed to Terminal Dorset but also essentially synchronous with Thule dates at the same sites or in the same region. Some sites in this region do, however, provide rather better evidence for a Dorset occupation between AD 800 and the later Terminal Dorset sites. Intriguingly, however, these sites are all in the High Arctic. I will return to this issue below. Figure 118c shows the dates obtained from four obviously relevant sites. Unlike the preceding graphs, in which all dates from a site were combined, this figure subdivides the dates by structure. At the extensively dated Nunguvik site, House 73 produced an impressive and consistent series of Dorset dates, plus one falling into the Terminal Dorset category. However the dated sample (S- 1615) is described as coming from the "bottom of earth-filled fissure into House 73,55 cm depth" (Morlan 1999).This stratigraphic context, combined with the fact that the date coincides precisely with Thule dates from other parts of the Nunguvik site, suggests that it and a similar date from House 71 derive from the Thule occupation of the site. An artifact recovered from House 73 and ascribed to a Norse origin (Sutherland 1999)may also derive from the Thule occupation here although the radiocarbon date is inconsistent with a Norse origin. At the Qeqertaaraq site in North-western Greenland, House 4 has the most extensive series of ^C dates and it also produced a Norse bronze pot fragment which provides a terminus post quem date of approximately AD 1300 (Appelt, et al. 1998:145).Although all the dates from House 4 are ascribed by the excavators to the Dorset culture (1999a:69;Appelt and Gullm 1999b:19-20), the presence of Thule artefacts in this house and its evidently complex history of occupation (Appelt and Gullnv 1999b:12, Fig. 7) clearly raises the possibility that some or all of the later series of dates derives from a Thule reoccupation of the house. These dates are consistent with a date on Thule structure 294 at the site, and with the second and much later date on Structure 161. The Staffe Island 1 site is situated in northern Labrador. Fitzhugh (1994) uses this site to document the presence of Terminal Dorset there at approximately the time the Thule arrived. He assigns only one 14Cdate at the site to what I am calling Terminal Dorset (Fitzhugh 1994:Table3) but at least one of the dates that he identifies as Thule (SI-3891, from House 9) was originally identified as Dorset (Kaplan 1983:219).I take this to indicate that there is rather less certainty about the exact cultural affiliation of the dated samples than is suggested by Fitzhugh. Fascinatingly, Fitzhugh (1994:258-259)also notes that there is no artifactual evidence of Dorset-Thule contact at this site, despite arguing that Dorset-Thule contact must have taken place in Labrador, and despite the fact that the radiocarbon dates he reports cannot distinguish between his postulated Thule and Terminal Dorset occupations of the site. As outlined above, I conclude that his Terminal Dorset date actually derives from the Thule occupation of the site. Unfortunately, only the briefest of descriptions of the Late Dorset Tasiarulik site on Little Cornwallis Island have been published (Helmer 1996; LeMoine and Darwent 1998)but the available dates from the site (Morlan 1999) show the same gap following AD 800. However, any real assessment of the relevance of this site to the question of the Dorset-Thule succession probably must await publication of the finds from the site. DISCUSSION One possible explanation for the patterning that appears to be so clear in the Late and Terminal Dorset dates is that some unknown phenomenon was affecting the abundance of "C in the Arctic environment and causing the calibrated dates from sites occupied during the centuries from approximately AD 800-1100 to be either older than or younger than their actual ages. In the absence of any known mechanism that would have this effect, I will provisionally reject that hypothesis and turn to explanations that are predicated on these dates accurately reflecting the ages of the sites. Researchers have long accepted that the people of the Dorset culture disappeared, probably due to local or regional extinction, from some regions of the Arctic prior to the arrival of the Thule. One of the most dramatic of these events took place in Newfoundland in the 7thor 8thcenturies AD, as can be seen quite clearly in Figure 118a. This extinction, following an extended Dorset occupation of Newfoundland that has been described as "remarkably successful," has been explained quite plausibly as being due to periodic failures of critical resources (Tuck and Pastore 1985:71,77).Tuck and Fitzhugh (1986:166)and Fitzhugh (1994:243244) are slightly more hesitant to postulate a comparably complete depopulation of Labrador at the same time, but recognize the hiatus in radiocarbon dates and accept the possibility that such an event took place here as well. Extended episodes of local abandonment have also been proposed for southern Baffin Island, including one essentially simultaneous with that of Newfoundland and Labrador (Odess 1998:431), and for parts of the High Arctic (Schledermann 1996:101103). Dealing with the Arctic more generally, Helmer (1996:305)notes that one of the "more noteworthy features of this cultural episode [is] the fact that a large part of the Eastern Arctic appears to have been re-colonised during the Late Dorset period following an extended interval of local abandonment." The same observation has been made less succinctly by several authors. What I will call the "Dorset refugium" - the source of the re-peopling of these abandoned regions - is seen as the socalled "Core Area." In its simplest form, the Core Area hypothesis (Fitzhugh 1997; Maxwell 1976a) has been based on the continuous occupation of a region including "at least both shores of Hudson Strait, the islands at its western mouth, the vicinity of Fury and Hecla Strait, northern Baffin Island, and at least southern Bylot Island" (Maxwell 1976b:5).Frustratingly, we do not have the extensive series of "C dates from important parts of this region that we would need to evaluate the hypothesis of Late Dorset continuity into the 12thcentury or later, especially from the important sites in the vicinity of Igloolik. Rather, according to Maxwell (198592)in this region the evidence supporting continuity comes from the absence of obvious typological gaps or from the absence of unoccupied beach ridges in some locales. But the Dia.4, Nunguvik and Saatut sites are thoroughly dated Dorset sites within Maxwell's Core Area boundaries, and none provides evidence of continuity from Late Dorset (ca. A.D. 500-800) through to so-called Terminal Dorset (AD 1000-1500). If I am incorrect and all the very late Dorset dates at sites in Labrador do indeed represent a Dorset repopulation of that region, we need to determine where that repopulating came from. Implicitly, the accepted answer has been from Ungava or southern Baffin Island, but there clearly is a hiatus in the Dorset occupation there as well following AD 800. We see the same pattern at the northern Baffin Island sites. Surprisingly, the few sites that are plausible candidates for a continued Dorset occupation in the centuries immediately following AD 800 based on presently-available dates are all in the High Arctic, north of Lancaster Sound. Perhaps it is there that we should be seeking the Dorset refugium, if indeed there was one. We also need to explore why the Terminal Dorset peoples reoccupied so many sites that their ancestors had abandoned centuries earlier, just at the time that the Thule were moving in to those same sites (e.g., Staffe Island 1, Nunguvik) at least to the degree that radiocarbon dating allows us to identify contemporaneity. This appears quite inconsistent with models invoking Terminal Dorset avoidance of the Thule (Friesen 1999; McGhee 1997). If, however, I am correct in my inference that the radiocarbon dates ascribed to "Terminal Dorset" actually all derive from Thule occupations at those sites, then we clearly need to ask why this happened. Most archaeologists are understandably reluctant to explore the possibility that the regional extinction in Newfoundland and Labrador might have been paralleled by, or perhaps even precipitated, the extinction of the entire Dorset population. However, one of the observations made commonly by many archaeologists concerns the unity/interconnectedness of the Late Dorset world as revealed by stylistic and technological homogeneity over vast distances (e.g., Helmer 1996:306; LeMoine and Darwent 1998:74; Maxwell 1985227).It is also likely that the Late Dorset population within any given region was not large (e.g., Odess 1998:420; Schledermann 1996:101; c.f. Murray 1999). Such a small scattered population would have required some minimum population level for biological reasons, and for the kind of social network required by foragers in such an environment (Odess 1998:421).The widespread homogeneity suggests that local Dorset groups were in regular contact with, and probably relied upon, distant groups. It is thus possible to imagine a regional extinction, such as has been accepted for Newfoundland (Tuck and Pastore 1985) or for parts of the High Arctic, putting neighboring regional populations in a very precarious situation indeed through the loss of this social network. As noted by Schledermann (1996:101),"A small population of any kind can be endangered in a number of ways, particularly in a region easily affected by minor environmental and ecological conditions. One disastrous hunting season or the accidental death of the best hunters could have led to starvation and demise of a small group of people." I submit that we should explore to what extent that kind of process, compounding over time could have resulted in the disappearance of all the Dorset prior to the arrival of the Thule. ACKNOWLEDGEMENTS Robert McGhee, Richard Morlan and Patricia Sutherland generously shared data and ideas with me, although they bear no responsibility for my conclusions. REFERENCES Appelt, Martin and Hans Christian Gullov Conclusion. In Late Dorset i n High Arctic Greenland: Final Report on the Gateway t o Greenland Project, edited by M. Appelt and H. C. Gullov, pp. 69-70. The Danish National Museum & Danish Polar Center, Copenhagen. Qeqertaaraq. In Late Dorset i n High Arctic Greenland: Final Report on the Gateway t o Greenland Project, edited by M. Appelt and H. C. Gullov, pp. 7-20. The Danish National Museum &Danish Polar Center, Copenhagen Appelt, Martin, Hans Christian Gullov and Hans Kapel 1998 The Gateway t o Greenland. Report on the Field Season 1996. In Man, Culture a n d Environment i n Ancient Greenland, edited by J. Arneborg and H. C. Gullov, pp. 136-196. The Danish National Museum & Danish Polar Center, Copenhagen. Arundale, Wendy H. 1981 Radiocarbon Dating in Eastern Arctic Archaeology: A Flexible Approach. American Antiquity 46(2):244271. Bandi, Hans-Georg 1969 Eskimo Prehistory. University of Alaska Press, College. Bielawski, Ellen 1979 Contactual Transformation the Dorset-Thule Succession. In Thule Eskimo Culture. A n Anthropological Retrospective, edited by A. P. McCartney, pp. 100109. Mercury Series Paper, No. 88. Canadian Museum of Civilization, Archaeological Survey of Canada, Ottawa. Cachel, Susan 1997 B. Grennow, pp. 201-214.Publication No. 1. Danish Dietary Shifts and the European Upper Palaeolithic Transition Current Anthropology 38(4):579-603. Polar Center, Copenhagen. 1997 From Middle Ages to Colonial Times: Archaeological a n d ethnohistorical studies o f the Thule culture i n Collins, Henry B. 1937 1950 South West Greenland 1300-1800AD. Meddelelser Archaeology o f St. Lawrence Island, Alaska. Smith- om Grenland - Man & Society Vol. 23. Kommissionen sonian Miscellaneous Collections Vol 96. for Videnskabelige Undersegelser I Grenland, Excavations at Frobisher Bay, Baffin Island, N.W.T. Copenhagen. Annual Report o f the National Museum o f Canada, 1984 Bulletin No. 1 18:18-43. Harp, Elmer Jr. History o f Research Before 1945.In Arctic, edited by 1976 Dorset Settlement Patterns in Newfoundland and D. Damas, pp. 8-16 Handbook o f North American Southeastern Hudson Bay. Memoirs o f the Society Indians, vol. 5, W C. Sturtevant, general editor. for American Archaeology No. 31 .I 19-138. Smithsonian Institution, Washington, D.C. Helmer, James W. Crowe, Keith J. 1996 1974 A History o f the Original Peoples o f Northern A Tale of Three Villages: Assessing the Archaeological Potential o f Three Late Dorset Settlements on Canada. Arctic Institute of North America, Montreal. Little Cornwallis Island, N.W.T. In The Paleo-Eskimo Cultures o f Greenland - New Perspectives i n Green- de Laguna, Frederica 1947 The Prehistory o f Northern North America as Seen landic Archaeology, edited by B Grennow, pp. 295308. Publication No. 1. Danish Polar Center, Copen- from the Yukon. Memoirs o f the Society For Amer- hagen ican Archaeology No. 3. Hood, Bryan C. Dumond, Don E. 1977 1998 Theory on Ice: The Discourse o f Eastern Canadian The Eskimos a n d Aleuts. 1st ed. Thames and Hudson, Arctic Paleo-Eskimo Archaeology. Acta Borealia London. 15(2):3-58. Fitzhugh, William W, Jenness, Diamond 1980 1925 Preliminary Report on the Torngat Archaeological Project. Arctic 33(3):585-606. 1994 graphical Review 15:428-437. Staffe Island 1 and the Northern Labrador DorsetThule Succession. In Threads o f Arctic Prehistory: Jordan, Richard H. Papers i n Honour o f William E. Taylor, Jr., edited by 1979 D. A. Morrison and J.-L. Pilon, pp. 239-268.Mercury 1997 A New Eskimo Culture in Hudson Bay. The Geo- Inugsuk Revisited: A n Alternative View of NeoEskimo Chronology and Culture Change in Green- Series Paper, No. 149.Canadian Museum o f Civiliza- land. In Thule Eskimo Culture: A n Anthropological tion, Archaeological Survey o f Canada, Hull. Retrospective, edited by A. P. McCartney, pp. 149- Biogeographical Archaeology in the Eastern North 170.Mercury Series Paper, No 88.Canadian Museum American Arctic. Human Ecology 25(3)'385-418. o f Civilization, Archaeological Survey of Canada, Ottawa Friesen, T. Max 1999 1984 Neo-Eskimo Prehistory of Greenland. In Arctic, Social Differences Between Palaeo- and Neo- edited by D. Damas, pp. 540-548.Handbook o f North Eskimos: Implications for Dorset-Thule Interaction. American Indians, vol. 5,W. C. Sturtevant, general Paper presented a t the Identities a n d Cultural Con- editor. Smithsonian Institution, Washington, D.C. tacts i n the Arctic Conference, November 30December 2, Copenhagen. Kaplan, Susan A. 1983 Economic and Social Change in Labrador Neo-Eskimo Gullev, Hans Christian Culture. Unpublished Ph.D. dissertation, Department 1996 of Anthropology, Bryn Mawr College, Bryn Mawr, In Search of the Dorset Culture in the Thule Culture. In The Paleo-Eskimo Cultures o f Greenland - New Perspectives i n Greenlandic Archaeology, edited by Pennsylvania. 1985 European Goods and Socio-Economic Change in Early Labrador Inuit Society In Cultures i n Contact: the Impact o f European Contacts o n Native Amer- dian Arctic Islands, edited by C. R. Har~ngton,pp. ican Cultural Institutions A.D. 1000-1800, edited by 666-676, Vol. 2. Canadian Museum of Nature, W. W. Fitzhugh, pp. 45-69. Smithsonian Institution Ottawa. Press, Washington, D.C. 1996 1997 LeMoine, Genevieve M. and Christyann M. Darwent 1998 Ancient People o f the Arctic. UBC Press, Vancouver. Meetings Between Dorset Culture Palaeo-Eskimos and Thule Culture Inuit- Evidence from Brooman The Walrus and the Carpenter: Late Dorset Ivory Point. In Fifty Years o f Arctic Research: Anthropolog- Working i n the High Arctic. Journal o f Archaeolog- ical Studies from Greenland t o Siberia, edited by R. ical Science 25(1)'73-83 Gilberg and H. C. Gullov, pp. 209-213. Ethnographical Series, Vol. 18 Department of Ethnography, National Museum o f Denmark, Copenhagen. Lyon, George Francis 1824 The Private Journal o f Captain G.F. Lyon. John 1999a Discussant's Comments. Paper presented at the Canadian Archaeological Association 32nd Annual Con- Murray, London. ference, April 28-May 2, Whitehorse. Mary-Rousseliere, Guy 1979 1999b The Nature and Timing of the Thule Migration. The Thule Culture on North Baffin Island: Early Thule Paper presented at the Identities and Cultural Con- Characteristics and the Survival o f the Thule Tradi- tacts in the Arctic Conference, November 30- tion. In Thule Eskimo Culture- A n Anthropological December 2, Copenhagen. Retrospective, edited by A. P. McCartney, pp. 54-75. Mercury Series Paper, No. 88. Canadian Museum of McGhee, Robert and James A. Tuck Civilization, Archaeological Survey o f Canada, 1976 Un-Dating the Canadian Arctic. In Eastern Arctic Prehistory: Paleoeskimo Problems, edited by M S Ottawa Maxwell, p p 6-14. Memoirs of the Society for American Archaeology, No. 31, Washington, D.C. Maxwell, Moreau S. (editor) 1976a Eastern Arctic Prehistory: Paleoeskimo Problems. Memoirs o f the Society for American Archaeology, Meldgaard, Jergen No. 31. Society for American Archaeology, Wash- 1960a Eskimo Sculpture. Methuen, London. ington, D.C. 1960b Prehistoric Culture Sequences in the Eastern Arctic as Elucidated By Stratified Sites at Igloolik. In M e n a n d 1976b Introduction. In Eastern Arctic Prehistory: Pale- Cultures - Selected Papers o f the Fifth International oeskimo Problems, edited by M. S. Maxwell, pp. 1-5. 1985 Memoirs of the Society for American Archaeology, Congress o f Anthropological a n d Ethnological Sci- No. 31, Washington, D.C. ences, edited by A. F. C. Wallace, pp 588-595. Univer- Prehistory o f the Eastern Arctic. Academic Press, sity of Pennsylvania Press, Philadelphia. Orlando. Morlan, Richard E. McCullough, Karen M. 1989 1999 Canadian Archaeological Radiocarbon Database. The Ruin Islanders: Early Thule Culture Pioneers i n WWW website, Canadian Museum of Civilization. the Eastern High Arctic. Mercury Series Paper No. <http://www.canadianarchaeology.com/radio- 141. Canadian Museum of Civilization, Archaeolog carbon/card/card.htm>. ical Survey of Canada, Ottawa. Morrison, David A. McGhee, Robert 1983 N.W.T. Damas, pp. 369-376. Handbook of North American Civilization, Archaeological Survey of Canada, Indians, vol. 5, W. C. Sturtevant, general editor. Ottawa. Smithsonian Institution, Washington, D.C. 1989 S4(1):1-7. Radiocarbon Dating Thule Culture. ArcticAnthrop010gy 26(2):48-77. 1984b The Timing of the Thule Migration. Polarforschung 1989 Thule Culture i n Western Coronation Gulf, Mercury Series Paper No. 116. Canadian Museum of 1984a Thule Prehistory o f Canada. In Arctic, edited by D. 1999 The Earliest Thule Migration. Canadian Journal o f Archaeology 22(2).139-156. Ancient Canada. Canadian Museum o f Civilization, Hull. 1990 The Peopling of the Arctic Islands. In Canada's Murray, Maribeth S. Missing Dimension: Science a n d History i n the Cana- 1999 The Importance o f Being Last: Real and Perceived Differences Between Dorset and Thule Site Structure. Past. Komatik Series No. 5. Arctic Institute o f North Paper presented at the Canadian Archaeological America, Calgary. Association 32nd Annual Conference, April 28-May 2. Whitehorse. Shields, Edward D. and Gregory Jones 1998 Dorset and Thule Divergence from East Central Asian Odess, Daniel Roots. American Journal o f Physical Anthropology 1998 1061207-218. The Archaeology of Interaction: Views from Artifact Style and Material Exchange in Dorset Society. American Antiquity 63(3):417-435. Stenton, Douglas R. and Bruce G. Rigby 1995 Park, Robert W. 1993 Community-Based Heritage Education, Training and Research: Preliminary Report on the Tungatsivvik The Dorset-Thule Succession in Arctic North America: Archaeological Project. Arctic 48(1):47-56. Assessing Claims for Culture Contact. American 1994 Antiquity 58(2)'203-234. Stuiver, Minze and Paula J. Reimer Approaches t o Dating the Thule Culture in the 1993 Eastern Arctic Canadian Journal o f Archaeology carbon Calibration Program. Radiocarbon 35(215- 18:29-48 230). Plumet, Patrick 1979 Extended 14C Database and Revised CALIB Radio- Stuiver, Minze, Paula J. Reimer, E. Bard, J.W. Beck, G.S. Burr, Thuleens et Dorsetiens dans I'Ungava (Nouveau- K.A. Hughen, B. Kromer, F.G. McCormac, J. v. d. Plicht Quebec). In Thule Eskimo Culture: A n Anthropolog- and M . Spurk ical Retrospective, edited by A. P. McCartney, pp. 1998 110-121. Mercury Series Paper, No. 88. Canadian INTCAL98 Radiocarbon Age Calibration 24,000 - 0 cal BP. Radiocarbon 40:1041-1083. Museum o f Civilization, Archaeological Survey of Canada, Ottawa. Sutherland, Patricia 1999 Strands o f Culture Contact: Dorset-Norse Interactions Plumet, Patrick and Pierre Gangloff i n the High Arctic. Paper presented at the Identities 1987 Contribution a I'etude d u peuplement prehistorique and Cultural Contacts i n the Arctic Conference, des cotes d u Quebec arctique et de son cadre November 30-December 2, Copenhagen. paleogeographique. Etudes Inuit Studies 11(1):67-89 Swinton, George Rowley, Susan 1994 1972 The Sadlermiut: Mysterious or Misunderstood? In Sculpture o f the Eskimo. McClelland and Stewart, Toronto. Threads o f Arctic Prehistory: Papers i n Honour o f William E. Taylor, JK, edited by D. A. Morrison and J.- Taqon, Paul S.C. L. Pilon, pp. 361-384. Mercury Series Paper, No. 149. 1983 Canadian Museum o f Civilization, Archaeological An Analysis o f Dorset Art in Relation t o Prehistoric Culture Stress. ~ t u d e s l l n u i t l ~ t u d i e 7(1):41-65. s Survey o f Canada, Hull. Taylor, R.E., C. Vance Haynes, Jr., Donna L. Kirner and John R. Rutherford, A.A., Juergen Wittenberg and Roscoe Wilmeth 1981 University o f Saskatchewan Radiocarbon Dates IX. Southon 1999 Radiocarbon Analyses of Modern Organics at Monte Verde Chile: No Evidence for a Local Reservoir Effect. Radiocarbon 23(1):94-135. American Antiquity 64(3)'455-460. Schiffer, Michael B. 1987 Formation Processes o f the Archaeological Record. Taylor, Royal Ervin University of New Mexico Press, Albuquerque. 1987 Radiocarbon Dating - A n Archaeological Perspective. Academic Press, Orlando. Schledermann, Peter 1976 The Effect o f Climatic/Ecological Changes on the Taylor, William E., Jr. Style of Thule Culture Winter Dwellings Arctic a n d 1960 Alpine Research 8(1):37-47 1996 Voices i n Stone: A Personal Journey into the Arctic A Description o f Sadlermiut Houses Excavated a t Native Point, Southampton Island, N.W.T. Annual Report o f the National Museum o f Canada, Bulletin No. 162-53-100 1963 Tuck, James A. and Ralph T. Pastore 1985 A Nice Place t o Visit, But.. Prehistoric Human Extinc- Hypotheses on the Origin of Canadian Thule Culture. tions on the Island o f Newfoundland. Canadian American Antiquity 28(4):456-464. Journal o f Archaeology 9(1):69-80. Thomson, J. Callum Waterbolk, Harm Tjalling 1985 1971 Working w i t h Radiocarbon Dates. Proceedings o f the 1983 Thirty Years of Radiocarbon Dating: The Retrospec- Dorset Shamanism - Excavations in Northern Labrador. Expedition 27(1):37-49 Prehistoric Society 37(15-33) Tuck, James A. and William W. Fitzhugh tive View o f a Groningen Archaeologist. In 14Cand 1986 Palaeo-Eskimo Traditions of Newfoundland and Archaeology, edited by W. G. Mook and H. T. Water- Labrador: A Re-Appraisal. In Palaeo-Eskimo Cultures bolk, pp. 17-27. Council of Europe, Strasbourg i n Newfoundland, Labrador a n d Ungava, pp. 161167. Reports in Archaeology, No. 1. Memorial Univer- Wenzel, George sity of Newfoundland, St. John's. 1991 Animal Rights, Human Rights: Ecology, Economy a n d Ideology i n the Canadian Arctic University o f Tuck, James A. and Robert McGhee 1983 Sea Mammal Dates: Science Or Science Fiction? Quarterly Review o f Archaeology 4(2):9-10. Toronto Press, Toronto.