Archaeological research in northern Labrador over the past three decades has uncovered numerous Palaeoeskimo habitation structures with axial or mid-passage features. Such structures span both the Early Palaeoeskimo (Pre-Dorset and Groswater phase) and Late Palaeoeskimo (Dorset) periods. In Early Palaeoeskimo axial tent structures, the mid-passage apparently divides gender-based activity areas, but evidence for this is lacking in Dorset. This paper summarizes evidence for axial structures from much of the span of Palaeoeskimo in the Okak region of north-central Labrador.
Les recherches archéologiques faites au Labrador septentrional durant les trente dernières années ont mis au jour de nombreuses structures d'habitations paléoesquimaudes avec des aménagements axiaux. Ces structures se retrouvent à la fois durant le Paléoesquimau ancien (Prédorsétien et phase groswatérienne) et le Paléoesquimau récent (Dorsétien). Dans les structures de tentes du Paléoesquimau ancien, l'aménagement axial divise apparemment les activités selon le sexe des individus, alors que durant le Dorsétien, on ne retrouve pas ce type d'évidence. Cet article résume les données concernant les structures axiales paléoesquimaudes dans la région d'Okak au nord du Labrador.
Three decades of research in central and northern Labrador by the author, the Smithsonian, Tuck and others has revealed a rich Palaeoeskimo sequence stretching from Early Pre-Dorset through Late Dorset. During virtually the entire three millennia long sequence a common thread is the presence of axial or mid-passage features within habitation structures. This continuity in turn raises questions about cultural and population continuity within the Labrador Palaeoeskimo sequence.
Most of the data for this paper are from the Okak region of north-central Labrador (Figure 1). Okak lies about 100 kilometres north of today's northernmost coastal settlement, Nain, and just to the south of the northern limit of trees on the coast. Okak's bay and island complex harboured rich resources, sufficient to support Labrador's highest Labrador Inuit population concentration during much of the 18th and 19th centuries.
Archaeological research at Okak began with the author's dissertation research in 1974 and 1975 (Cox 1977, 1978). Primarily oriented towards gathering data on prehistoric Eskimo and Indian settlement-subsistence systems, the project combined site survey with excavation of several sites, including three Dorset semi-subterranean houses. In 1987 I returned to Okak to excavate a remarkable Early Palaeoeskimo site on Nuasornak Island. Nuasornak-2 contains a series of raised beach terraces, each with multiple Early Palaeoeskimo tent structures ranging in age from Early Pre-Dorset at about 4000 B.P. to Late Groswater phase at about 2000 B.P.
The core of this paper is the Okak data, but equally important to the story is information from other regions, including William Fitzhugh's data from the Nain region (Fizhugh 1976), and data gathered by the Torngat Archaeological Project in 1977 and 1978 on the north Labrador coast (Fitzhugh 1980).
Nuasornak and the Early Palaeoeskimo sequence
Site Nuasornak-2 (HiCl-1) is located on the western side of Nuasornak Island in the Okak region. The site lies on a series of five sand and gravel raised beach terraces rising from six to twelve metres above sea level. On these terraces are a minimum of 39 Early Palaeoeskimo structures, mostly mid-passage tent rings, covering an area of approximately 200 m x 120 m (Figure 2). Most of the structures are partially buried, so that while most are detectable on the basis of surface indications, structural details are often not apparent without excavation. During the 1987 field season we excavated four mid-passage tent ring structures and a pavement feature of uncertain function. Structures from four of the five terraces were excavated, with only the lowest (5.9 m) terrace not tested.
Surface inspection of the upper (10 m) terrace revealed a concentration of artifacts near the northern end of the terrace, apparently not in association with a tent ring. The artifacts were typologically early, and we decided to excavate the area of the concentration, calling it Find Spot-2 (FS-2). We excavated 4.75 m2, uncovering a linear 2 m x 1 m concentration of cobbles running NE-SW with a charcoal concentration at the NE end (Figure 3). We suspected that the rock concentration might be a grave or cache, but there was no evidence of a pit below the rocks, and the function of the feature remains enigmatic. It is possible that this is a tent mid-passage, although outer ring rocks were not evident. Artifacts from the feature looked to be early in the Pre-Dorset sequence, a suggestion confirmed by a radiocarbon date on the charcoal from the concentration of 3960 ± 100 B.P. (Beta-25199) (see Table 1).
In addition to FS-2, we excavated one definite tent ring (S-26) on the 10 m terrace. The structure was roughly 3 metres in diameter, with an east-west axial feature defined by a line of columnar rocks, roughly square in cross section (Figure 4). Few flat slab rocks were found, and there was no evidence of an upright slab hearth. The southern lobe of the tent ring was largely missing, represented by only a few rocks, and most of the cultural material was concentrated in the northern lobe. This suggests the possibility that this was a "D" shaped structure rather than a bilobed tent ring, although I am more inclined to believe that it was bilobed, and that part of it had been scavenged in construction of other, nearby tent rings. The structure is not directly dated, but several characteristics of the artifact collection suggest a date slightly later than that of FS-2, perhaps within the range 3800-3500 B.P.
We excavated one structure completely on the 7.9 m terrace. Structure 4 is an oval or subrectangular tent ring measuring approximately 5 m x 3 m (Figure 5). Details of the structure were difficult to distinguish because of a large number of naturally-occurring rocks which had eroded from the front of the 10 m terrace. The axial structure runs NW-SE and consists of a double row of rocks, mostly rounded cobbles rather than slabs or columnar rocks. There was no evidence of a formal hearth, but there were several charcoal concentrations within the structure. There may be partial overlap with a second structure on the western side of S-4.
Interestingly, the collection from S-4 does not differ greatly from those of higher structures. It includes a series of small triangular points, a small tapered stem point of quartz crystal, and two large stemmed bifaces. Fine edge serration is present on several of the triangular points and one of the large stemmed bifaces. Other artifacts include unground burins, an endscraper, several pointed or side scrapers, and a flaked adze.
Charcoal samples from two different concentrations within the structure were submitted for dating. One sample returned a date of 4250 ± 90 B.P. (Beta-25200) while the other dated to 3210 ± 80 B.P. (Beta-25198). The former date is clearly too early, both for the terrace elevation and for the typological characteristics of the collection from the structure. It may be the result either of old driftwood within the sample, or incorporation of charcoal from the eroding upper terrace. The later date is reasonable for the terrace elevation and for dates associated with structures above and below S-4. If accurate, it strengthens our impression that Labrador Pre-Dorset culture remained fairly static, with few changes in stone tool technology or habitation structures, from near its beginnings 4,000 or so years ago down to sometime after 3,200 radiocarbon years ago.
Examination of the distribution of artifacts within Structure 4 suggests a differential distribution for some types (Figure 6). Within the structure all of the triangular points and scrapers are found on the right (NE) side of the structure, while microblades are more numerous on the left (SW) side. This distribution is reminiscent of McGhee's Independence I evidence from Port Refuge (McGhee 1979: 52-55), and like his suggests gender-related activity areas on either side of the tent. If so, it is interesting that burins and burin spalls are present in roughly equal numbers on both sides. Also note the sparsity of artifacts along the NW side of the structure — the reason for this is unclear, as is the location of the tent entrance.
Structure 3, at 7.1 m above sea level, is a roughly round mid-passage tent ring about 5 m in diameter (Figure 7). The N-S axial structure included cobbles, flat slabs and upright slabs. In about the center of the mid-passage there was a hearth marked by a blocky upright rock on the north side and a roughly square structure of small upright and flat slabs extending to the south. Charcoal in and around the hearth was collected for dating, returning a date of 2900 ± 90 B.P. (Beta-25197). On top of the hearth structure there was a sterile layer of pebbles and sand, apparently deliberately placed. Two upright slab rocks within the mid-passage to the south of the hearth, one of them slightly notched on top, were not associated with charcoal and may have been a lamp stand.
The artifact assemblage from within the structure consisted mostly of microblades, with few diagnostic forms. However, 6 m south (inland) of the structure we found a concentration of material on the surface including a ground burin or burin-like tool and two notched bifaces. Our initial impression that this concentration was associated with Structure 3 was subsequently confirmed by several artifact fits between material from the structure and the surface locus. Within the structure there is too little diversity in the artifact assemblage to be certain, but similar to the pattern in S-4 most of the microblades were found on the western side of the tent ring while two of three bifaces occurred on the eastern side.
The Structure 3 assemblage belongs to the Early Groswater phase, and the associated 2900 B.P. date appears appropriate. A number of the artifacts from the structure are made of mottled red, brown and gray cherts almost certainly from Newfoundland sources, and this is the first appearance of Newfoundland chert in the Nuasornak sequence. Indeed, this is the earliest radiocarbon date associated with definite Newfoundland chert that we have for northern Labrador, and the occupation of Structure 3 may have occurred not long after Palaeoeskimo peoples first reached Newfoundland. Similarly, an oval soapstone lamp preform from within the structure is the earliest dated soapstone vessel in Labrador.
The lowest of the excavated structures, Structure 1 at 6.7 m, produced a date of 2150 ± 70 B.P. (Beta-25196) on wood charcoal from a hearth. The date is consistent with the structures elevation, and suggests a late Groswater phase occupation. However, the artifact assemblage is somewhat puzzling. It consists mostly of microblades (88 of 107 artifacts). Two burin tips and a base recovered from the structure are more extensively polished than that of S-3, suggesting a later date. However, the structure also produced a chipped and ground graver, a type previously known only from Pre-Dorset and Early Groswater assemblages (S-26 produced one). Even more unusual is the complete absence of Newfoundland chert in the assemblage. These cherts are ubiquitous in virtually all Groswater phase collections, even those from northernmost Labrador. Instead, the S-1 collection is composed entirely of Labrador materials, including Ramah chert, chert from the Cape Mugford region just north of Okak, and quartz crystal. Thus, we are not entirely confident of the date and cultural affiliation of this structure.
Structure 1 itself appears to have been a large (6 m diameter) round or oval tent structure with a mid-passage running SE-NW (Figure 8). Both the tent ring and the axial structure appeared to have been disturbed. For the most part hold-down rocks were present only in the northwestern quadrant of the tent ring, although given the fact that this is the direction of the prevailing fall wind it is possible that the original structure had relatively few rocks along the rest of the tent perimeter. The mid-passage structure, consisting of both rounded cobbles and flat slabs, had clearly been disturbed, but a slab box hearth containing abundant charcoal near the southeastern (inland) end of the passage did survive.
The Nuasornak sequence documents the presence of mid-passage tent rings through the entire two millennia span of early Palaeoeskimo in Labrador. Throughout the sequence, settlement at Nuasornak appears to have occurred as encampments of one to three tents, based on observed grouping of structures at the site. The earliest Palaeoeskimo occupations of Nuasornak, and of northern Labrador in general, date to about 4,000 radiocarbon years ago and exhibit Independence I-like traits such as edge serration on bifaces, unifaces and burins; small triangular, bipointed and contracting stem endblades; unpolished burins; and habitation within tents with axial structures. During the subsequent millennium, Labrador Pre-Dorset culture was very static, with few apparent changes.
After a long period of stability, at or shortly before 3000 B.P., there is a sudden and rapid series of changes in Labrador Pre-Dorset culture that culminates in the Groswater phase. Many of these changes are similar to those occurring simultaneously in other portions of the eastern Arctic. Notched harpoon endblades and knives replace or supplement earlier stemmed and triangular forms. Microblades greatly increase in frequency. Facial and working edge polish appears on burins, which are often very small. Soapstone lamps make their first appearance. At the same time these and other changes were happening, Palaeoeskimos reached the island of Newfoundland and established exchange or mobility systems that resulted in the movement of Newfoundland cherts in quantity north along the Labrador coast.
Groswater phase people continued living in mid-passage tents, although there were some structural changes. Somewhat surprisingly, none of the early and middle Pre-Dorset tent structures found to date in Labrador feature flat slab rocks in hearth construction, and such slabs are rare in axial structures. It is only in the terminal Pre-Dorset period around 3000 B.P., or perhaps a century or two earlier, that we begin to regularly see flat horizontal and vertical slabs used for axial structures, particularly in hearths. Thereafter, they are a common feature of Groswater phase habitation structures, as well as later Dorset structures.
At Nuasornak, tent rings become larger during the Groswater phase. However, given that smaller terminal Pre-Dorset and Groswater phase tent rings are known from elsewhere in Labrador, this may reflect a seasonal rather than cultural shift. Lacking faunal remains we do not have direct evidence for seasonality at Nuasornak. We do know that Nuasornak Island was the site of one of three Labrador Inuit fall-early winter sod house villages in the Okak region at the time of the establishment of the Okak Moravian mission station in 1776 (Taylor 1974), and that it was probably a less desirable location in the late winter and spring, when resources were concentrated at the ice edge, considerably seaward of Nuasornak and with a large island, Iglusuaktalialuk, in between. Excavation of the earlier and smaller structures on the 10 and 7.9 m terraces revealed that most of the cultural material occurred within the structure boundaries, perhaps indicating cold weather occupations paralleling the Labrador Inuit fall-early winter occupation. In contrast, the lower Groswater phase structures were larger, contained a narrower range of tools (primarily microblades), and at least in the case of Structure 3, had outdoor activity areas associated. This may reflect a shift to warmer weather settlement (summer-early fall?) later in the sequence.
Dorset — the pattern continues
The Groswater phase persisted until a century or two into the Christian era. However, during the middle of its span a new culture, Early Dorset, appeared in northern Labrador, spreading at least as far south as Nain. Labrador Early Dorset culture is classic Dorset, with virtually all of the defining Early Dorset characteristics including triangular tip-fluted harpoon endblades, multiple notched lance endblades, also tip-fluted, extensively polished burin-like tools made of chert and nephrite, soapstone lamps and cooking pots, and semi-subterranean houses. In this and later Dorset phases there is a shift in settlement pattern towards a more seaward, maritime orientation, with settlement nearer the ice edge during cold months and less settlement in general back in the inner reaches of the bays. Most of the Labrador Early Dorset radiocarbon dates fall within the range 2500-2400 B.P.
Our sample of Early Dorset sites and houses is relatively small compared to that of later phases, perhaps reflecting a relatively short time span for the phase. Two sets of Early Dorset semi-subterranean houses have been tested, one a single house at Okak (site Iluvektalik-1 [HhCk-1]) and the other a group of three houses on the north coast, in Komaktorvik fiord (Komaktorvik-1 [IhCw-1]). In neither case were the houses fully excavated, but it was apparent that all of the houses were around 6 m in their longest dimension, all lacked entrance passages, and there was no evidence of a formal rock mid-passage structure. The Iluvektalik house produced a few faunal remains, including walrus which were historically in the Okak region mainly during late winter (February-March).
At site Nukasusutok12 (HcCh-14) in the Nain region, Bryan Hood excavated several axial structures whose associated lithic assemblages exhibit some Early Dorset characteristics (Hood 1986). Unfortunately, none of the structures are reliably radiocarbon dated, and Middle Dorset traits are also present in the collections, making it difficult to assess the true age and affiliation of the structures.
There appears to have been a gap in occupation between Early Dorset, with latest dates around 2400 B.P., and the succeeding Middle Dorset phase which begins a few decades before 2000 B.P. While this gap may be an artifact of sampling, as the number of radiocarbon dated Early and Middle Dorset sites continues to rise it seems increasingly likely that there was a hiatus in occupation of at least portions of the north coast between Early and Middle Dorset.
Middle Dorset begins slightly before 2,000 radiocarbon years ago and lasts until between 1200 and 1300 B.P. Technological changes between Early and Middle Dorset are relatively minor, but there are significant changes in housing. These changes are perhaps best illustrated by another semi-subterranean house in the Okak region, at site Iglu-4West (HhCj-5). The site lies on the east side of Iglusuaktalialuk Island and contains a single Middle Dorset semi-subterranean house (Figures 9 and 10).
The house is rectangular, measuring approximately 5.5 m x 3.5 m (interior dimensions). There was a 3 m long entrance passage extending from the center of the southwestern wall, pointed away from the shore. A stone mid-passage structure ran through the center of the house. It was bounded on one side by a row of upright flat slabs; there was no corresponding row on the other side. At the rear of the house there were several large flat paving rocks within the mid-passage, and in the middle of the mid-passage, in front of the paved area, there was a hearth bounded by two flat upright rocks to the north and south. Between the two upright rocks there was a flat hearth rock covered with a layer of burnt fat. In front of the hearth there were several more flat pavement rocks, and then a break in the mid-passage until one reaches the inner mouth of the entrance passage, where there was an assemblage of flat and round rocks which may mark a second hearth.
The house appears to have burned down, possibly not long after the beginning of occupation, to judge from the paucity of associated cultural material. A layer of charcoal extended throughout the interior of the house, in places reaching a maximum thickness of 6 cm. In some areas two distinct layers of charcoal could be discerned, separated by a layer of fire-reddened sand — possibly the floor and collapsed roof of the dwelling. Where individual burnt logs could be distinguished they were generally 5-10 cm in diameter and oriented along the long and short axes of the house.
The walls of the house were clearly evident during excavation. The houses living floor was composed of leached gray (E horizon) sand, whereas the walls were composed of mineralized brown sand. Additionally, the charcoal layer ended abruptly at the wall. At least as they survived, the walls were quite low, rising about 30 cm above the house floor. The areas on either side of the mid-passage were for the most part clear of rocks, and were presumably sleeping as well as activity areas. No evidence of raised platforms or benches was found.
The small collection from the house interior included typical Middle Dorset material such as tip-fluted triangular endblades, polished nephrite burin-like tools and a rectangular soapstone cooking pot. Two radiocarbon dates on charcoal from the house interior were run, producing dates of 1685 ± 70 B.P. (SI-2157) and 1860 ± 90 B.P. (SI-2508).
While the Iglu-4W house has the most formal mid-passage structure known from any excavated Middle Dorset houses in Labrador, it is not alone in its combination of an entrance passage and axial structure. Fitzhugh's (1976) Koliktalik-1 Middle Dorset semi-subterranean house from the Nain area had an entrance passage. While it lacked a formal paved mid-passage, it did have a central linear concentration of pits and cultural material, perhaps more reminiscent of Newfoundland houses (Renouf 1993). Additionally, a number of Middle Dorset surficial tent rings with mid-passages are known, including examples from the Nain area (Fitzhugh 1976) and from Avayalik Island near the northern tip of Labrador (Jordan 1986: 138).
As in just about all other areas of the eastern Arctic, the appearance of Late Dorset in Labrador around 1,200 years ago brought a series of dramatic changes in culture, rivalling the degree of change between Pre-Dorset and Dorset. This was a period of uniformity in the eastern Arctic, with part of that uniformity lying in the widespread presence of rectangular or subrectangular semi-subterranean houses lacking entrance passages but featuring paved mid-passage structures. A number of such structures are known from northern Labrador, including excavated structures on Avayalik Island, Peabody Point in Seven Islands Bay, Shuldham Island in Saglek, and Okak. Again perhaps the most elaborate example comes from Okak, from site Okak-3 (HjCl-3) on the north side of the main Okak Island (Figures 11 to 13).
The north and west walls of the Okak-3 house were marked by straight, steeply sloping walls of earth and stone. The eastern (shoreward) and southern walls were less clearly delineated, but the house appears to have been roughly rectangular, 7-8 m long (north-south) and about 6 m wide. The southern wall may have been more rounded than the others. The house floor was about 50 cm below the present ground surface.
The impressive mid-passage structure consisted of two parallel rows of narrow columnar boundary rocks, about a m apart and running about 5 m north-south (parallel to shore), with flat paving rocks in between and in strips along the outside of the boundary rocks. The total mid-passage structure was 2-2.5 m wide. There were two hearths along the mid-line, and a third at the eastern margin of the mid-passage structure. Each hearth was marked by an upright slab, notched on its upper end, next to a horizontal slab on which there was a deposit of charcoal or burnt fat. The southern mid-line hearth had two upright rocks, one of them columnar and unnotched, to the south of the horizontal slab. Extending along each side of the mid-passage structure there was an area of sand and pea-gravel, probably brought up from the beach, that was relatively clear of rocks and about a m wide — perhaps sleeping areas.
The cultural deposit within the house consisted of a 10-12 cm thick dark brown humus layer made up of decomposed peat and sand. No peat lenses indicative of superimposed floors were found, but the presence of more than 800 stone artifacts within the house suggests multi-year habitation. Charcoal from the hearths returned two radiocarbon dates: 1005 ± 95 B.P. (SI-2154) and 895 ± 85 B.P. (SI-2506).
Conclusions and questions
The entrance of Thule culture into Labrador in the 13th century A.D. brought an end to over three millennia of Palaeoeskimo occupation of the north Labrador region, and to a similar span of habitation within mid-passage structures. With the possible exception of the Early Dorset period, all phases of Labrador Palaeoeskimo saw habitation structures with axial features, ranging from relatively simple surficial tents to substantial semi-subterranean houses. We know that most of the axial features contain hearths, and that at least during the early Palaeoeskimo period, including Pre-Dorset and probably Groswater phase, there is some evidence to suggest that the mid-passage divided gender-based activity areas.
Beyond that we know very little, and are left with a series of questions. Most fundamentally, what was the function of the axial feature? Was it to divide social space, as suggested by the early Palaeoeskimo activity patterning? Why then don't we see (to my knowledge) similar patterning in Dorset houses? Do the more informal and amorphous central linear patterns of houses such as those of Koliktalik and Port au Choix serve the same function as the more formal rock pavement features seen in the Okak region houses, and why the difference? Does the presence of axial features throughout the northern Labrador Palaeoeskimo sequence suggest population continuity, either within Labrador or within a broader eastern region of which Labrador was a part? Hopefully continuing research of the sort reflected in the papers of the St-Pierre Symposium will shed light on these questions.
- Cox, Steven L., 1977 Prehistoric Settlement and Culture Change at Okak, Labrador, Ph.D. dissertation, Cambridge, Harvard University, Department of Anthropology.
- Cox, Steven L., 1978 Palaeo-Eskimo Occupations of the North Labrador Coast, Arctic Anthropology, 15(2): 96-118.
- Fitzhugh, William W., 1976 Preliminary Culture History of Nain, Labrador; Smithsonian Fieldwork, 1975, Journal of Field Archaeology, 3: 123-142.
- Fitzhugh, William W., 1980 Preliminary Report on the Torngat Archaeological Project, Arctic, 33(3): 585-606.
- Hood, Bryan, 1986 Nukasusutok-12: Early/Middle Dorset Axial Structures From the Nain Region, Labrador, in Palaeo-Eskimo Cultures in Newfoundland, Labrador and Ungava, St. John's, Memorial University of Newfoundland, Reports in Archaeology, 1: 49-64.
- Jordan, Richard H., 1986 Palaeo-Eskimos in Atlantic Canada: A Regional Comparison of Newfoundland and Labrador Middle Dorset, in Palaeo-Eskimo Cultures in Newfoundland, Labrador and Ungava, St. John's, Memorial University of Newfoundland, Reports in Archaeology, 1: 135-150.
- McGhee, Robert J., 1979 The Palaeoeskimo Occupations at Port Refuge, High Arctic Canada, Ottawa, National Museum of Man, Archaeological Survey of Canada, Mercury Series, 92.
- Renouf, M.A.P., 1993 Palaeoeskimo Seal Hunters at Port au Choix, Northwestern Newfoundland, Newfoundland Studies, 9(2): 185-212.
- Taylor, J. Garth, 1974 Labrador Eskimo Settlements of the Early Contact Period, Ottawa, National Museum of Man, Publications in Ethnology, 9.