A Middle Dorset Palaeoeskimo structure at Peat Garden North, northwest Newfoundland Une structure paléoesquimaude du Dorsétien moyen à Peat Garden North au nord-ouest de Terre-Neuve

Abstract

Dorset Palaeoeskimo sites on ancient gravel and cobble beaches pose exceptional challenges when researching prehistoric dwellings constructed along coastlines. At Peat Garden North, a Middle Dorset Palaeoeskimo site on Newfoundland's Great Northern Peninsula, a feature embedded in a raised dolomite beach is interpreted as a dwelling. Although all that remains of this structure is a poorly defined oval of arranged beach cobbles and gravel, the argument for this feature as a house is based on overall shape, size, internal features, artifact and faunal distribution, and similarities with other houses in the Arctic and sub-Arctic. In conclusion, we suggest that the layout of this house is a function of site seasonality.

Résumé

Les sites dorsétiens paléoesquimaux situés sur d'anciennes plages de gravier et de galets présentent des défis exceptionnels lors de l'étude d'habitations préhistoriques le long du littoral. À Peat Garden North, un site du Dorsétien moyen de la Péninsule Great Northern à Terre-Neuve, nous avons interpété un aménagement intégré dans une plage de dolomite comme étant une habitation. Bien que de cette structure il ne reste qu'un oval mal défini fait de galets et de gravier, notre interprétation se base sur sa forme générale, sa taille, ses aménagements intérieurs, la distribution des objets lithiques et des vestiges fauniques, ainsi que des similarités avec d'autres maisons de l'Arctique et du sub-Arctique. En conclusion, nous suggérons que la disposition de cette maison est une fonction de la saison d'occupation du site.

Introduction

The town of Bird Cove is located on the limestone barrens of the west coast of Newfoundland's Northern Peninsula and lies just below the southwest extent of the Strait of Belle Isle (Figure 1). The area's rich resources, coastal inlets, bays, and harbours have attracted human interest for over 4,500 years, beginning with the Maritime Archaic Indians (Reader 1998, 1999). After this initial settlement, Bird Cove and the surrounding area became home to a number of cultures including Groswater and Dorset Palaeoeskimo, several Recent Indian groups, and historically, the French, English, and Irish (Hartery 2001; Hartery and Rast 2001, 2002; Reader 1998). The intensity of human occupation is represented by 32 pre-contact and historic archaeological sites, 14 of which are located on the Dog Peninsula. The Middle Dorset site of Peat Garden North (EgBf-18), dated to 1500-1000 B.P., is characterized by lithic, faunal and shellfish remains, as well as two house features, one of which (F00-01) is the focus of this article. Data collected from the 1997-2001 field seasons are herein reported; 2002 onward can be found in Hartery (in prep.).

Site location and area description

Peat Garden North is located on the Dog Peninsula, which protrudes into the Gulf of St. Lawrence for approximately 5 kilometres and forms the eastern border of St. Margaret's Bay. During most of the span of human occupation in the area, the Dog Peninsula was divided into two islands, although today falling relative sea level has joined them to each other and the mainland. Peat Garden North is on the inner coast and landward site of the peninsula, facing the mainland and town of Bird Cove, and sheltered from the forceful northwesterly winds (Figure 1). Access to outer coastal headlands, which are suitable for catching harp seals, is only a few minutes walk away.

The site occupies a 1000 square metre area on a wide paleobeach ridge 3.4 m above sea-level. However, at the time of occupation, the sea-level was 1-1.5 m higher than it is today (Grant 1994; Hartery and Rast 2001, 2002; Reader 1998; Renouf and Bell 2000) (Figure 2). Excavated areas correspond to open clearings that were formerly used as ATV trails, berry picking patches and home lots for Bird Cove's earliest residences. Despite these uses, the site remains largely intact except in the extreme southern portion where disturbance occurs. Investigated areas in these open clearings are surrounded by thick spruce and birch forest. The site has two distinct stratigraphic layers, the first being peat that ranges from 20-100 cm deep which lies over the second, an ancient dolomite beach surface mixed with grayish-black organic soil (Figure 3). The Palaeoeskimo occupation is limited to the top of this ancient beach surface where 12,000 faunal remains, 11,873 shell fragments, 9,261 flakes, 652 artifacts and two houses have been uncovered between 1997 and 2002 (Table 1).

The terrestrial and marine fauna of Bird Cove are similar to that which is found in most parts of western Newfoundland. Faunal remains are abundant at several sites in the Bird Cove area and are reliable indicators of the environment and available resources during the area's precontact occupation. Several sites in the area were occupied by different cultures during roughly the same period. Peat Garden, Peat Garden North, North Cove, and the Bird Cove site are crucial for understanding the resource inventory from 2200-1000 B.P. Murray's (2000) preliminary analysis of the Groswater Palaeoeskimo component of Peat Garden include species such as seal, fish, birds, caribou, bear, beaver and lynx. Stewart's (1999) analysis of the faunal remains from the dual component Recent Indian/Dorset Palaeoeskimo North Cove site is equally diverse and includes sculpins, flatfish, Canada goose, ducks, terns, black bear, ringed seal, caribou, deer and other unidentified birds and mammals. This list is augmented by the recent analysis of Peat Garden North as well (Hartery and Rast 2001; Hartery in prep.; Murray 1998). Faunal remains at this site include shellfish, birds, beaver, caribou, seals and fish. The variability in available resources also comprises variability in clues, which enables us to detect the season of occupation sites such as Peat Garden North. Though the faunal analysis of this site is in its preliminary stages, we suggest that Peat Garden North was occupied during the spring-summer season based on the interrelationship of faunal remains, artifacts and architectural elements of house feature F00-01.

Figure 1

Location of study area and Bird Cove sites mentioned in the text. Rectangle corresponds to area of sea-level reconstruction in Figure 2

Figure: 2

Peat Garden North, locations of 1 m x 1 m units excavated in 1997, 2000, and 2001. Solid black line indicates sea-level position at time of Dorset Palaeoeskimo occupation.

Figure 3

Stratigraphic profile of Peat Garden North. Level 1 is a sterile peat layer. On top of and in between the rocks of level 2, Dorset Palaeoeskimo cultural material is found. Unit shown is exactly 1 metre.

Table 1

Peat Garden North artifact count from 1997, 2000, 2001

Methodological approach

Reader's trenches at Peat Garden North in 1997 uncovered a midden in the northeast section of the site. Subsequent excavations occurred in three areas in 2000, 2001, and 2002 (Figure 2). In general, the southernmost excavated area contained few bones and artifacts and appears to have been outside of the feature to be discussed. In the northwest block, we unearthed a dense midden deposit of bone and artifacts but found no trace of any feature that may be structural. In 2000-2001, we excavated upslope from Reader's test trenches, in the northeast block, with hopes of locating additional midden remains, hearths or any other features. We were successful in uncovering more of the expansive midden and located Feature 00-01 which is interpreted as a dwelling (Figures 4 and 5). However, no hearth features were detected. In 2002, Hartery returned to Peat Garden North and located a second feature (F02-01) similar and adjacent to the first (F00-01).

The excavation of Peat Garden has proceeded slowly and in a number of stages. Shovels are used to remove sterile peat until it forms only a thin layer above the dolomite beach, trowels are used to remove this thin layer, and tea-spoons are used to excavate the grey-black organic soil of the second (cultural) level. Spoons are considered optimal since it is the only thing that can be dragged slowly over the irregular contours of the undulating beach rocks without damaging artifacts and inflicting fresh scrapes on the faunal remains. All soil was dry sieved through a 5 and 3 mm screen in the field to prevent the loss of small bird and fish bone. A provenience is assigned to all faunal remains and artifacts to the nearest centimetre for future statistical analysis. The rocks associated with Feature 00-01 have been left in situ but the artifacts and bone that have trickled down through the cracks in the cobble beach were collected if possible. Each rock has been hand drawn and mapped as a means to decipher the exact perimeter of this feature. Careful excavation and documentation such as this has ensured that small details are not missed such as individual strands of caribou hair, a tiny dehydrated berry (Rubus), and loose teeth of juvenile mammals.

Figure 4

Plan drawing of F00-01 and associated artifacts at the end of the 2001 field season. Dark grey area shows where rocks and gravel were removed to form the inside of the dwelling. Inset shows distribution of artifacts from 2000 and 2001 excavations.

Feature 00-01

Feature 00-01 was first exposed in the last two weeks of the 2000 field season. The exact nature of the feature was not immediately obvious, but it was apparent that the beach surface in the northeast block had been manipulated. In the western portion of the feature the peat layer rested directly on bedrock instead of the jumbled varying sized dolomite beach cobbles, which usually occur in about a 20 cm thickness above the bedrock. The house builders had removed the cobbles down to the bedrock in some areas and re-arranged the cobbles in others. By the end of this season, only half of the feature was excavated but enough information was gained to indicate that this feature might be the footprint of a dwelling. At the end of the 2000 field season we had exposed: a) a wide semi-circular arc of stones which seem to form a wall perimeter, b) the clearing of the larger beach rocks down to either parent bedrock or very small beach cobbles, leaving a slightly raised internal axial feature, and c) the presence of a soapstone vessel for either heat/light/cooking on top of this axial feature which suggested domestic activity. Further excavation over two seasons revealed the feature in full as well as a second feature that is exactly the same size (F02-01). However, intrasite analysis is avoided here since the second feature will not be completely excavated or analysed until the end of 2003.

Once Feature 00-01 was fully excavated, new evidence further strengthened the argument that this structure was a house and we were able to determine the house boundaries although they were somewhat vague. The arrangement of the dolomite beach cobbles, the distribution of artifacts and faunal material all contribute to understanding the house feature. Figure 4 depicts the slight depressions and clearings of rocks as well as the locations of the northeast block artifacts found in 2000 and 2001.

Feature 00-01 is roughly ovate and spans approximately 5 m east-west and 4 metres north-south. The outer perimeter is loosely defined by large cobbles stacked directly on the beach surface. No entrance is clearly demarcated, but given the position of both the beach and the midden to the north, south and east of the dwelling, it is presumed to be located somewhere on the dwelling's east side. The house's northern perimeter is not as clearly defined on the palaeobeach surface as the south.

A poorly defined axial feature composed of large beach cobbles lies in the middle of the house. The axial feature is not linear, but rather squarish. The most linear element of the axial feature is an arrangement of a half dozen long narrow rocks lying parallel to one another, like railway ties (Figure 5). To the north and northeast, and south of this central elevated area, are slightly depressed areas where large beach cobbles were removed to expose the underlying bedrock, or tiny cobbles that fit together forming "cobblestone-like" flooring, and gravel, leaving behind a narrow shelf of original beach surface in the axial feature. The elevated rocks and the cleared space have about a 10-15 cm difference in elevation (Figure 5). A number of rocks that are stacked on the ring's perimeter are tilted forward towards the centre of the dwelling. Consequently, the areas of cleared rock appear narrower than perhaps they would have been during the initial phases of construction. Although F00-01 is not semi-subterranean, other similar early Paleoeskimo houses constructed on boulder fields are found in southeastern and central eastern Hudson Bay at sites such as Kuujjuarapik (Gendron and Pinard 2000: 133).

The distribution of artifacts and faunal remains across the beach provides further evidence for the location and size of the house feature. Distinct activity areas are present within the house; in the northeast excavation area there are concentrations of tip-flute spalls, endblades, microblades and cores, whereas in the west and southwest microblades and pieces of a soapstone vessel predominate. In 2000 and 2001, we opened 34 1 m x 1 m units in the northeast block and recovered 368 identifiable artifacts (excluding flakes and faunal material). Of these 34 units, 15 lie within the area of the house and 19 are external. The artifact density inside the area of the house was 15.66 artifacts per m², while outside of the house it averaged 7.00 per m². The external artifact density is slightly skewed by a very artifact rich area northeast of the house, which may relate to a second house feature and will be discussed in future reports. The artefacts in the 12 m² units south of the house averaged only 3.17 per m², but had considerably more bone. The artifact density inside the house feature is between 2 and 5 times greater than the density of artefacts outside of the house. The density of the faunal distribution is the opposite of the artifacts, with the majority of elements found outside of the house. The artifact rich areas inside the house have accompanying faunal remains but in smaller quantities than that occurring outside the house.

Tool types are diverse and indicate a wide variety of activities occurred at Peat Garden North (Table 1 and Figure 6). The high frequency of microblades, scrapers, ground slate tools and knives indicates a range of cutting, scraping, and processing activities occurred on site. Various performs, cores/core fragments and burin-like tools indicate that both bone and stone tools were being manufactured at Peat Garden North. The presence of foreshafts, harpoon head fragments and endblades suggests seal hunting occurred while symbolic artefacts such as a miniature ivory harpoon head and a polar effigy represent the spiritual side of daily life. The variegated range of tasks represented by the artifact inventory is good evidence for a habitation site.

We attempted to use the identification of postholes as a guideline to determining the limit of house walls or areas of internal support. This did not prove to be a useful method of detecting house boundaries since the beach cobbles distributed throughout all of the excavated areas have cracks, crevices and holes between them. Any attempt by the Dorset Palaeoeskimo to use these holes, or even to create new postholes, remains impossible to detect since they will blend in with those already naturally occurring. We have found no indications of construction material so far. If this was a spring-summer site as we suggest, then perhaps substantial building materials, such as whalebone support beams were not necessary (Renouf 1993), and they used lighter materials for the milder conditions of a warm season. This house was not semi-subterranean; fighting the cold did not seem to be an important consideration.

Figure 5

Peat Garden North; view east toward south half of F00-01. The area to the left of the strings is the poorly defined axial feature with the "railway tie" rocks. The space between the strings is an area where the Dorset Palaeoeskimo excavated through the cobbles down to bedrock. The area to the right of the strings has stacked beach cobbles and marks the southern perimeter of the house feature.

Figure 6

A range of artifacts from Peat Garden North; (L-R) endblades, a knife and an ivory harpoon head

Chronological data

The submission of radiocarbon samples has been hindered by the small amount of charcoal present at the site. Reader (1998) was able to locate a hearth (F97-02) in a midden (F97-03), 5 metres south of house feature F00-01, which was dated to 1570 ± 60 B.P. (Beta-113160). Since that time, we have found only a few tiny pieces of charcoal throughout the site's eastern excavation area. One of these small samples, located along the north side of the axial feature of F00-01, was dated to 959 ± 45 B.P. (BGS 2254). A second radiocarbon assay was run on charcoal outside the house's northern boundary. Unfortunately it was highly fragmented and only thin flecks remained interspersed in the black organic soil. The result was 677 ± 45 B.P. (BGS 2321) and is outside Middle Dorset range in Newfoundland. Land mammal bone from inside the northwest portion of F00-01 was sent to Isotrace for AMS dating and the results were 1030 ± 290 B.P. (TO-9555) and 1490 ± 50 B.P. (TO-10468). This standard deviation for the first date is quite high; a result of low collagen yield in the bone, but the second has a smaller margin of error, and fits well with the other dates presented here. F02-01 (a second house feature) has been dated to 1380 ± 50 B.P. (GX-29745) based on the dating of a tiny piece of wood charcoal. The site was occupied towards the end of the Middle Dorset period in Newfoundland, possibly 1500-1000 B.P.

The difficulty of finding charcoal may indicate that few fires were active at this site and may reflect the warmth already experienced at the site due to season of occupation. The small rectangular soapstone vessel (12 x 8 x 5 cm), with a rim of black soot on the upper half of the inside wall, and found inside the F00-01, seems to be the only evidence, in addition to the one hearth feature, that so far represents human production of heat and light at the site.

Season of occupation

Although the innumerable dolomite beach cobbles propose a great challenge in determining the boundaries and size of the Feature 00-01, they fortunately provide great preservation of organic remains and hence clues for interpreting site function and seasonality.

All artifacts and faunal remains are found lying directly on the dolomite beach surface and in between these rocks. However, some rocks have been removed in the northwest midden area to reveal that bones, shells, and artifacts have percolated down through open cracks to a depth of 5-10 cm. The chemical formula of the dolomite is CaMg (CO3)2 (Gilluly et al. 1968: 599). The presence of calcium, magnesium and carbonates provides the conditions that promote bone preservation at this site. In general, the faunal remains, like the artifacts, are in direct contact with the dolomite rock by either resting on top of it or being wedged in crevices between cobbles.

Of the 12,000 bones recovered, more than half are identifiable to Class or lower taxonomic level; the remainder is simply classified as unidentifiable. The assemblage has not been fully quantified, however, based on a preliminary analysis of 2,418 identified faunal elements from the eastern and southern portions of the site, Murray (1998: 3) suggests a warm season occupation based on goose remains containing medullary bone (see also Hodgetts et al. 2003). She further states that seal accounts for roughly 43% of the identified assemblage, a number which differs quite markedly from other sites in Newfoundland such as Phillip's Garden, Port aux Choix, where roughly 98% of the faunal remains have been identified as seal (Murray 1998; Renouf 1993; Renouf and Murray 1999). Just over 50% of the Peat Garden North faunal assemblage is avifauna (Murray 1998). This is an unprecedented number of avifauna and is the first of its kind when compared to other Middle Dorset sites in Newfoundland and Labrador, such as Port aux Choix, where roughly 98% of the faunal remains have been identified as seal (Renouf 1993). However, bird hunting is common at a few other Palaeoeskimo Arctic sites such as Nipisat (Saqqaq) and those in Hatherton Bay in Greenland (Late Dorset) (see Bendix 2000; Gotfredsen 1996). Canada goose (Branta canadensis) is the dominating species and a number of elements, such as femurs and tibiotarsi, have medullary bone present, indicating these geese were harvested during the egg laying period (Hartery in prep.). Calcium deposits in Canada goose bones point to a narrower time band for their capture of mid-May to mid-June. Hartery (in prep.) is working on the 12,000 animal bones which are associated with the house features and the eastern and western portions of the site. While still in a preliminary stage, early analysis corresponds with the initial analysis of the test trenches of 1997 (Hartery in prep.; Murray 1998; Reader 1998).

Both juvenile and adult seals have been found at Peat Garden North. The list of seals includes harp (Phoca groenlandica), harbour (Phoca vitulina), hooded (Cystophocacristata) and ringed (Phoca hispida) seals. Seals are usually the predominant mammal in Palaeoeskimo sites throughout the Canadian Arctic as well as Newfoundland and Labrador, often adding up to more than 90% of all faunal material for any given site (e.g., Mary-Rousselière 1976; Renouf 1994; Renouf and Murray 1999; Schledermann 1990). The number of caribou in this assemblage is less than bird and seal, but greater than fish, and accounts for roughly 9% of all identifiable bones (Murray 1998).

The presence of 11,873 shells came as a surprise at Peat Garden North since this is a first for Dorset Palaeoeskimo research on the Island of Newfoundland. Though the majority are fragments, many complete specimens give insight as to what the remaining highly fragmented specimens could be. The list of species identified to date includes soft shelled clams (Mya arenaria), mussels (Mytilus edulis), periwinkle shells (Thiaslapillus) and scallops (Placopecten magellanicus) (Hartery and Rast 2002). Clams, mussels, periwinkle shells, and scallops, are available throughout Newfoundland year round in small quantities or in special places (Tuck and Pastore 1985) but are most abundant from May to October especially since pack-ice usually precludes access to these types of molluscs in winter. The shellfish could have been used for either primary consumption or as bait to catch other large fish. Regardless of function, they were most easily acquired during a warm season when they were most abundant and available.

The above synopsis of the faunal material provides only broad parameters for interpreting this large site and subsequent reports about faunal material will not occur until all bones are analysed. However, a late spring-summer occupation of Peat Garden North is likely because all of the above named resources are available during that narrow time period, especially waterfowl and molluscs. While caribou may be lingering in the area after fall, they would be low in numbers and their meat and fur would be less desirable. The wide range of animals hunted and the intensity of the deposit suggest that Peat Garden North may have been occupied for an extended period of time, possibly an entire season or several times over a number of years or a combination of both of these scenarios.

Seasonality and its relationship to the structure

If, as the faunal analysis and features suggest, this site was occupied in the spring- summer season, then this is an extremely important discovery. Most of the houses in Port aux Choix have been interpreted as winter to spring houses (Harp 1976; Renouf 1993; Renouf and Murray 1999). These interpretations are based not only on the substantial constructions of these houses (e.g., House 2 and Feature1) but also on the intensity with which harp seal and other seals were hunted at Port aux Choix. Midden deposits associated with House 4, another winter dwelling, had 28,000 associated bones; 98% of which is seal (Harp 1976: 128).

Harp (1976) believed House 5 at Phillip's Garden in Port aux Choix may have been a summer dwelling because of its insubstantial construction. The size of House 5 was 3.3 m x 5.9 m, which is similar in size to the one at Peat Garden North. Both houses were also characterized by shallow depressions and were oval clearings of detectably fewer rocks than those outside the vague perimeter of the house. Harp (1976: 130) further noted that House 5 did not posses an internal hearth or any other distinguishing features. More houses with these characteristics may be present at Port aux Choix since not all houses have been excavated or fully analysed (Harp 1976).

Since Harp's initial excavations at Port aux Choix, Renouf and Murray (1999: 125) have compared two winter dwellings, House 2 and Feature 1 to discern that these two features are alike in interior size, architectural elements and layout, season of occupation and range of artifact classes. They further argue that although Phillip's Garden may have been occupied at any one season during its initial occupation, as time passed and "the abundance, predictability and accessibility of late winter harp seals became apparent, Phillip's Garden evolved into a permanent hunting site where families gathered every March" (ibid.). Over 50 houses are present at Phillip's Garden, 24 have been analysed and three have been published in full. Harp (1976) tested approximately 20, and assigned one of these to a summer dwelling type while the rest were identified as winter houses, with House 2 as the model for the winter house type.

Renouf (1986, 1987, 1993) excavated three houses that she describes as cold weather dwellings. Despite the fact of having seasonality in common, these houses still have some structural differences, while maintaining basic characteristics of internal hearth features and ground depressions. At Phillip's Garden there is a warm weather feature (No. 42) which is considered a windbreak structure (Renouf 1991). At Point Riche, occupation occurs on a smaller scale than at Phillip's Garden. Renouf (1986) suggests these may be summer dwellings based on location, dwelling type (i.e. external hearth/axial feature and loose boundaries), and general observation of resources available in the area at that time of year. However, unlike Phillip's Garden and Bird Cove, statements about seasonality and housing cannot be supported by faunal evidence.

Studies of Dorset Palaeoeskimo houses in Newfoundland outside of Port aux Choix are rare. LeBlanc (1999: 2 and this volume) has excavated two houses on Dildo Island. House 1 is a circular pattern of large rocks which form a tent ring. Because of "site location and abundance of marine resources" around the island, it is hypothesized that this may be a spring-summer dwelling (LeBlanc 1997: 68). House 2 is a semi-subterranean house and is believed to be a cold season dwelling (ibid.). However, LeBlanc thinks that site seasonality is still speculative since there is no accompanying faunal remains.

At Point Riche in Port au Choix, Eastaugh (2002 and this volume) excavated a Dorset Palaeoeskimo house feature similar to feature F00-01 at Peat Garden North. The structure was a shallow semi-subterranean dwelling dug into the beach, with a slightly built up perimeter. The dwelling (Feature 30) was not clearly delineated with rocks or postholes, but was initially outlined through magnetometer survey and defined through careful attention to stratigraphic clues in the excavation. The house foundation was built in silty and peaty soil, which provided much greater resolution in defining the house than the coarse cobble beach at Peat Garden North. The diameter of the circular house was 6.5 m, with a 1.6 m wide raised soil wall (Eastaugh 2002: 57). Similar to the house at Peat Garden North, the artefact density inside the house outnumbered the artifact density outside the house by a ratio of three to one (Eastaugh 2002: 111). Feature 30 has an internal axial feature similar to F00-01 and Eastaugh contrasts this with an external axial feature associated with a second house at Point Riche. Eastaugh argues that the external focus of the nearby house may indicate that it was a warm season dwelling where activity was focused out of doors, while Feature 30 may have been a cold season habitation (ibid.). This may have been the case at Point Riche, but at Peat Garden North the faunal evidence suggests that warm season dwellings could have internal axial features.

The Dorset dwelling at Cape Ray is described by Fogt (1998: 88) as having "very little in common with the dwellings of Port aux Choix, and continues to defy the notion of a standardized Dorset dwelling." That author further explains that these differences occur even though the two sites were likely spring seal hunting sites. The house at Cape Ray possesses two axial features, several interior hearths, a more rectangular shape, and even some evidence for an entrance passage (Fogt 1998: 87). As different as this dwelling is from those at Phillip's Garden, at this point it also seems very different from the house at Peat Garden North. Several other Dorset Palaeoeskimo dwellings have been identified in the Exploits River and Trinity Bay area by Devereux (1966), Evans (1981) and Robbins (1985), respectively, but they are not as well documented as those mentioned above.

In Labrador, a number of Middle Dorset Palaeoeskimo structures, described as tent rings, are found at Avayalik Island-2 (Jordan 1980) and in the Nain area (Fitzhugh 1976). The Avayalik-2 structure has a sub-rectangular outline but lacks any central definition except for a loosely defined central rock feature (Jordan 1986: 138). The dimensions of this feature, 5.85 m x 3.15 m, are comparable in size to House 5 in Port aux Choix and F00-01 in Peat Garden North. Nain tent rings were defined by central cobble or slab pavement features, but had a rectangular to sub-rectangular outline. Jordan (1986) makes a comparison between winter dwellings in Newfoundland and Labrador but avoids doing so for tent rings "given the poorly known variation in Newfoundland temporary dwellings." When fully analyzed, hopefully Peat Garden North can enhance the Newfoundland data concerning these types of dwellings.

Discussion

In summary, the house feature and faunal evidence from Peat Garden North suggests that it functioned as a residential site, which was likely occupied in late spring to summer, and maybe even throughout the summer if we consider that the Dorset Palaeoeskimo, when compared to earlier Palaeoeskimo groups, may have had long term commitment to place and lower residential mobility (Renouf 1993). Furthermore, its location may have been selected to take advantage of the local distribution of birds, seal, and caribou. The role of fish in the Palaeoeskimo diet is not fully understood but Hodgetts et al. (2003) suggest, based on evidence from Phillip's Garden, Port aux Choix, that as the site is occupied through time, an increased reliance on fish and bird is apparent. Harp seal hunting probably ceased with the arrival of summer and the disappearance of sea-ice around the Dog Peninsula. Harbour seal populations would have remained, but in considerably lower numbers than the northbound harp or ringed seals.

The faunal collection from Peat Garden North suggests a broad-based Dorset Palaeoeskimo economy, which makes use of all possible types of animal resources that were available within the region. The higher than expected frequency of waterfowl suggests a possible seasonal variant from a subsistence strategy based solely on sealing practices which is normally suggested for Dorset Palaeoskimo in Newfoundland. Sometime in the late spring, the Dorset Palaeoeskimo moved into Peat Garden North (or other west coast summer camps), began hunting harp seals, and also captured other seals close to shore. When seal hunting subsided, caribou and bird hunting continued, until changes in resource availability prompted these people to move to another location.

If Dorset Palaeoeskimo low residential mobility, coupled with the evidence of this large midden, indicates the length of stay, it seems the occupation of Peat Garden North could have spanned an entire season, perhaps during multiple years, or even during different seasons since re-occupation of this site is possible. Futhermore, the dolomite rocks have not been removed from excavated units for fear of compromising undetected house features. When the rocks are cleared, it is expected that additional zooarchaeological remains will be found.

An in-depth study of spring to summer Dorset Palaeoeskimo houses and subsistence has not been carried out in Newfoundland. The reason for this is simple; strong evidence for this season has so far been lacking. In Bird Cove, there is a collective sample of a dwelling, faunal material and artifacts, all of which makes such an analysis finally possible. This is rare for any culture but very important for the Dorset in Newfoundland as we have a biased view of their hunting strategies due to the recovery of winter-spring season sites and dwellings.