REGIONAL COMPARISON OF THULE HARPOON HEADS 



THROUGH THICK AND THIN: 

A REGIONAL COMPARISON OF HARPOON HEADS FROM 

THULE SITES IN NUNA VUT, CANADA 

By 

Mary Jo Megginson, B.A. 

A Thesis 

Submitted to the School of Graduate Studies 

in Partial Fulfilment of the Requirements 

for the Degree 

Master of Arts 

McMaster University 

August, 2000 



McMASTER UNIVERSITY LIBRARY 



MASTER OF ARTS (2000) 
(Anthropology) 

McMaster University 
Hamilton, Ontario 

TITLE: Through Thick and Thin: A Regional Comparison of Harpoon Heads from 
Thule Sites in Nunavut, Canada 

AUTHOR: Mary 10 Megginson, B.A. (McGill University) 

SUPERVISOR: Professor P. G. Ramsden 

NUMBER OF PAGES: viii + 132 

11 



Abstract 

In 1969170 Robert McGhee first suggested the existence of regional social groups 

in Thule culture. Under the assumption that such social groupings would be reflected in 

the distribution of material culture, this study aims to investigate McGhee's hypothesis 

using one artifact class of Thule culture: harpoon heads. The study looks at harpoon 

heads from all published Thule sites from across the territory of Nunavut, in arctic 

Canada. The harpoon heads are broken down into individual attributes, and the regional 

distribution of each attribute is considered in an attempt to find patterning across space. 

Rather than confirming the existence of regional social groupings, the patterns discovered 

suggest a culture continuum moving from west to east across the Canadian arctic. This 

research is preliminary in nature, and opens a new forum for debate in Canadian arctic 

archaeology. 

111 



Acknowledgements 

Thanks to my committee: Laura Finsten, Trudy Nicks, and especially my 

supervisor Peter Ramsden for their undying patience and constructive criticism. Though I 

may never want to look at another harpoon head again, I've learned much from this 

experience. This thesis would not have been possible without the faithful support of my 

two best friends: Mom and Dan, this is for you. 

IV 



Table of Contents 

Table of Contents .................................................................................................. .iv 

List of Figures .......................................................................................................... v 

I - Introduction ......................................................................................................... l 

Background ........................................ '" ....................................................... 2 

Contributions of the Research .................................................................... .4 

II - Background ........................................................................................................ 6 

Thule Culture - An Introduction .................................................................. 6 

Origin of Thule Culture .................................................................... 7 

Archaeology and Social Boundaries ........................................................... 1 0 

Ethnicity ........................................................................................ 11 

Tyranny of the Ethnographer. ....................................................... 13 

The Meaning of Style ............. '" .................................................... 15 

The Uses of Style .......................................................................... 17 

III - Methodology .................................................................................................. 21 

Methodological Problems .......................................................................... 21 

The Study Area .......................................................................................... 25 

Western Central Arctic .................................................................. 26 

The Central Region ........................................................................ 31 

v 



The Eastern Region ........................................................................ 35 

Chronology of Sites ....................................................................... 39 

Selection of Artifacts ................................................................................ .40 

IV - Harpoon Heads: An Introduction .................................................................. .42 

Introduction ............................................................................................... 4 2 

Harpoon Head Typology ......................................................................... .43 

Function ..................................................................................................... 46 

Chronological Issues .................................................................................. 50 

Methodological Issues ............................................................................... 51 

V - Data Analysis .................................................................................................. 53 

Thule Type I Harpoon Heads ................................................................... 53 

Thule Type II Harpoon Heads .................................................................. 57 

Thule Type III Harpoon Heads ................................................................. 67 

Thule Type IV Harpoon Heads ................................................................. 76 

VI - Discussion and Conclusions ........................................................................... 82 

Distribution of Types ................................................................................ 82 

Distribution of Raw MateriaL .................................................................. 87 

Distribution of Lashing Provision Types .................................................. 88 

Distribution of Decorative Motifs ............................................................. 92 

Regional 'Tribes'? ..................................................................................... 95 

Conclusions .............................................................................................. 1 03 

Appendix ............................................................................................................. 1 06 

Bibliography ........................................................................................................ 121 

VI 



List of Figures 

Figure 3-1: Map of Major Geographical Features ofNunavut.. ............................ 27 

Figure 3-2: Map of Sites ........................................................................................ 30 

Figure 4-1: Harpoon Head Types and Attributes ................................................ .48 

Figure 4-2: Harpoon Head Types and Attributes ................................................ .49 

Figure 5-1: Distribution of Type I Harpoon Heads .............................................. 53 

Figure 5-2: Raw Material of Type I Harpoon Heads ............................................ 55 

Figure 5-3: Lashing Provision on Type I Harpoon Heads ..................................... 56 

Figure 5-4: Distribution of Type II Harpoon Heads ............................................. 57 

Figure 5-5: Distribution of Type II Fonus ............................................................ 58 

Figure 5-6: Raw Material of Type II Harpoon Heads .......................................... 61 

Figure 5-7: Lashing Provision on Type II Harpoon Heads ................................... 63 

Figure 5-8: Distribution of Type III Harpoon Heads ............................................ 68 

Figure 5-9: Distribution of Type III Fonus ........................................................... 69 

Figure 5-10: Lashing Provision on Type III Harpoon Heads ................................ 70 

Figure 5-11: Raw Material of Type III Harpoon Heads ....................................... 73 

Figure 5-12: Distribution of Lateral Waists on Type III Harpoon Heads ............. 74 

Figure 5-13: Distribution of Type IV Harpoon Heads .......................................... 77 

Figure 5-14: Distribution of Type IV Fonus ......................................................... 78 

vii 



Figure 5-15: Raw Material of Type IV Harpoon Heads ....................................... 79 

Figure 6-1: Distribution of Harpoon Head Types in the Western Region ............ 83 

Figure 6-2: Distribution of Harpoon Head Types in the Central Region .............. 83 

Figure 6-3: Distribution of Harpoon Head Types in the Eastern Region .............. 84 

Figure 6-4: Raw Material Distribution .................................................................. 88 

Figure 6-5: Lashing Provision Distribution ............................................................ 89 

List of Tables 

Table 1: Thule Type I Harpoon Heads Being Used in this Study ...................... 106 

Table 2: Thule Type II Harpoon Heads from the Western Study Region .......... 107 

Table 3: Thule Type II Harpoon Heads from the Central Study Region ......... 1 08-9 

Table 4: Thule Type II Harpoon Heads from the Eastern Study Region ..... .11 0-12 

Table 5: Thule Type III Harpoon Heads from the Western Study Region ... 113-14 

Table 6: Thule Type III Harpoon Heads from the Central Study Region .... .114-15 

Table 7: Thule Type III Harpoon Heads from the Eastern Study Region ..... 116-17 

Table 8: Thule Type IV Harpoon Heads from the Western Study Region ......... 118 

Table 9: Thule Type IV Harpoon Heads from the Central Study Region ..... 118-19 

Table 10: Thule Type IV Harpoon Heads from the Eastern Study Region: .119-20 

V111 



I 
Introduction 

Over one thousand years ago, a new marine-adapted culture arrived into what is 

now the territory of Nunavut, in arctic Canada. Thule culture, as it was named in 1927 

by archaeologist Therkel Mathiassen, would ultimately evolve over the last millennium to 

become the richly diverse Inuit culture that we know today. Archaeologists have long 

portrayed Thule culture as a uniform cultural entity, bound together by a common 

adaptation to an isolated and barren ecological niche. However, modem Inuit are proud of 

their cultural diversity, and frequently refer to local social group names, such as 'Netsilik' 

and 'Iglulik', to describe their sense of group affiliation. In an earlier work (Megginson 

1997), I argued that these vast social entities were a creation of the enthnographers of the 

late nineteenth and early twentieth centuries who lived with and studied the Inuit. The 

extremely mobile lifestyle of the protohistoric Inuit precluded the existence of such large 

social groups, and group affiliation occurred at a much smaller level, often centred around 

the maximal winter band of about one hundred people. This is not to say that these 

groupings do not have meaning today: as the hunter-gatherer lifestyle became less 

prominent in the lives of the Inuit, a more sedentary lifestyle and centralisation into town 

centres became more common. This more sedentary lifestyle brought people into contact 

with a larger group of people on a day-to-day basis, and thus the wide cultural terms 

seem to have been appropriated by the modem Inuit to serve these new needs. 

With all of the changes that have occurred from the proto historic period to the 

I 



2 
present day, one begins to wonder what the social situation might have been like in 

prehistoric times. Could it be possible that Thule culture was not the uniform entity it 

has always been thought to be? Without the luxury of written accounts that are at the 

disposal of the ethnohistorian investigating the protohistoric period, the prehistoric 

archaeologist must often rely on material culture in such an investigation. If one can make 

the assumption that social group affiliation would be reflected in material culture, then it 

seems to follow that one could make a systematic analysis of Thule material culture in an 

attempt to find evidence of discrete social groupings. The purpose of this thesis is to test 

the hypothesis that Thule culture was uniform across the Canadian arctic, and this will be 

achieved through an analysis of geographical variation in harpoon head technology. 

Background 

Many archaeologists have argued for the in situ transition from Thule culture to 

historic Inuit cultures (e.g. Mary-Rousseliere 1979; McGhee 1969170, 1972; Taylor 

1965). According to such an approach, during the centuries that followed the initial 

expansion of Thule, their way of life became modified to adapt to local conditions and in 

response to the progressively deteriorating climate (Burch 1979: 190). The result was 

increasing regional diversification in Thule culture: 

The uniform stratum of early Thule culture then developed regional 
variants, and these variants eventually developed into the various 
local populations of the historic Inupik-speaking Eskimo (McGhee 
1969170: 173). 

McGhee relates the development of these regional variants to climatic changes that have 

occurred in the Canadian arctic over the past one thousand years. While the climate 

changes themselves were probably not dramatic enough to have produced significant 



3 
cultural change, their influence on the annual and seasonal distribution of sea ice, and by 

association, the annual and seasonal availability of sea mammals, would have caused 

crucial changes in Thule subsistence strategies (Ibid.: 175). The Thule expansion occurred 

during the Neo-Atlantic Episode (900-1200 A.D.), during which mean summer 

temperatures were one to two degrees Celsius warmer that at present, and there was a 

general retreat of pack ice throughout the arctic (Ibid.). However, this was soon followed 

by a period marked by a deteriorating climate, the Pacific Episode (1200-1550 A.D.). 

The Thule people adjusted their patterns of adaptation to meet this changing climate, but 

the kind and nature of these changes varied from region to region: 

After 1200 A.D., Thule culture probably did not constitute the 
highly uniform cultural horizon suggested by early excavations. 
With further work, we may be able to distinguish and define 
regional 'tribes' on the Thule level, differing from one another to 
about the same extent as the various tribes of historic Inupik­
speaking Eskimo (McGhee 1969170: 180). 

McGhee suspects that further investigation will reveal 'tribal' separation in the following 

regions of the Canadian arctic: Amundsen Gulf, Foxe Basin, Baffin Bay, Labrador, and 

Hudson Bay. He defines the first of these groups on the basis of their extensive use of 

pottery and copper, and their preference for Thule Type II harpoon heads with lashing 

slots. l Baffin Bay Thule, on the other hand, are defined by their reliance on the hunting 

of Greenland whales, their preference for flat harpoon heads with drilled lashing holes, 

and the frequency of dolls with top-knot hairstyles (McGhee 1969170: 180). 

McGhee's main concern in proposing this model is to support his theory of in 

situ culture development in the Canadian arctic: 

Some of these 'tribal' differences on the Thule level may continue 
through time and serve as distinguishing characteristics between 

1 This refers to Therkel Mathiassen's harpoon head typology, defined in 1927. The definitions of each 
type will be discussed in a later chapter. 



historic Eskimo tribes, implying geographical stability of these 
tribes relative to one another for the past 400 to 800 years 
(McGhee 1969170: 180). 

4 

This model is most frequently applied in terms of the ethnographically described cultures 

of the historic period: Copper, Netsilik, Caribou, Iglulik, and Baffinland Inuit. Such an 

approach, however, is based on the assumption that the Inuit ofthe historic period can be 

divided into discrete social groupings. This assumption is based on the early 

ethnographic research done by such eminent anthropologists as Knud Rasmussen, 

Therkel Mathiassen, Kaj Birket-Smith, Franz Boas, and Diamond Jenness during the 

latter part of the nineteenth century and the beginning of the twentieth century. 

Contributions 0/ the Research 

Robert McGhee, who first suggested the possible existence of regional variation in 

Thule culture (1969170), was probably influenced by his strong belief in in situ 

prehistoric-historic development in the Arctic, and also the general acceptance of discrete 

social groupings as having existed in the ethnographic period. As yet, no archaeologist 

has undertaken a systematic investigation of McGhee's original suggestion. This will be 

the first research that will look for Thule regional variation without assuming unilinear 

development from Thule 'tribes' to historic period 'cultures', and as such, it will aim to 

avoid the biases inherent in past archaeological research in the Canadian Arctic. This 

research suggests a new perspective for hunter-gatherer archaeology as a whole that does 

not assume the existence of discrete social groups. Sites are compared on a local level 

rather than through more general regional survey, and in this way, one can avoid the 

danger of grouping sites together into 'cultures' that have no social or archaeological 

meaning. Archaeologists have never been able to agree upon the extent to which we can 



5 
reliably link material culture to specific social groups. This research will work on the 

assumption that this link is possible, and thus this work can contribute to the debate in 

the form of a case study. Broader issues concerning ethnicity and ethnic boundaries can 

also be addressed by considering Barth's (1969) discussion of these matters while looking 

for regional variation in the archaeological record. 

If my research shows that a culture continuum is evident in the Thule 

archaeological record, then the theory of in situ development will be supported in terms of 

the theory of historic period culture continuum. If, however, McGhee was right in 

suggesting that regional 'tribes' can be identified in the archaeological record, new 

questions must be asked as to why and how there was a change towards cultural 

uniformity through time. A consideration of subsistence strategies would probably be of 

importance in answering these questions - cooperative hunting strategies, such as those 

required by whale hunting, would assist in the creation of discrete social groups, as large, 

sedentary winter villages were the norm. In historic times, however, hunting groups were 

much smaller and spread out, thus reducing the possibility of the recognition of affiliation 

with a larger social entity. This could be argued to have led to increased uniformity in the 

stylistic attributes of artifacts throughout the Arctic, thus creating a culture continuum. 

These changes in subsistence strategies are most often explained in terms of climatic 

change, using a perspective borrowed from cultural ecology. 



Thule Culture - An Introduction 

n 
Background 

Thule culture was first defined III 1927 by Therkel Mathiassen. Having 

undertaken an excavation of Naujan, perhaps the most famous Thule site investigated to 

date, located at Repulse Bay, just north of Southampton Island - he concluded that the 

remains he had found differed sufficiently from the more modern populations of the 

region to justify calling it a separate culture. Furthermore, the complete absence of any 

artifacts of European manufacture and the deteriorated state of the house ruins led him to 

conclude that this culture was about a thousand years old (Mathiassen 1927a: 86-89). He 

saw striking similarities between this site and one that had been investigated near Thule, 

Greenland - enough similarity to conclude that they represented the same culture. 

As the first find of this culture was thus found at Thule by 
members of the Second Thule Expedition and the second find, at 
Naujan, was made by the Fifth Thule Expedition, I consider it 
warrantable to attach the name of Thule to the old Central Eskimo 
culture as represented by the Naujanfind (Mathiassen 1927a: 89). 

Thule culture has variously been defined based on chronology, artifact typology, 

dwelling types and economy. In its most general form, it can be defined as including 

" ... all the midden-building, polished-slate-making, lamp-burning, kayak- and umiak-

paddling Eskimos ... who extended from Kodiak Island to Greenland (Dumond 1977: 118). 

Any definition of Thule culture is, by necessity, concerned with determining the criteria 

by which this culture can be differentiated from what we know of the ethnographically 

described modern Inuit. This differentiation cannot be determined through exact means, 

6 



7 
and the line between the two groups is often very blurry. Part of the reason for this is 

that the transition appears to have occurred slowly, and at different times across the 

arctic. 

Origin a/Thule Culture 

The first inhabitants of the Canadian arctic were those associated with the Arctic 

Small Tool Tradition (ASTt) of the Paleoeskimo phase of arctic prehistory. These 

hunters originated in the Bering Sea region and migrated east to Greenland sometime 

between 4500 and 4000 years ago (Irving 1962; Giddings 1964). The Pre-Dorset of the 

ASTt evolved into what is known as Dorset culture in Canada by 500 B.C. (Taylor 

1968), and in Alaska it developed into the Norton Tradition by 800 B.C. The Norton 

Tradition includes the cultures known as Choris, Norton and Ipiutak, and lasted 2000 

years in Alaska (Dumond 1977). 

The Paleoeskimo stage is differentiated from the Neoeskimo stage of arctic 

prehistory by the appearance of harpoon float gear, which for the first time allowed for 

open water sea mammal hunting (Morrison 1983: 4). The oldest cultures classified as 

Neoeskimo are known as Okvik and Old Bering Sea, and they probably developed from 

the Norton Tradition in Alaska (Dumond 1977). In northwestern Alaska, these cultures 

developed into Birnirk culture by A.D. 500, and on St. Lawrence Island and the Siberian 

coast, they developed into Punuk culture (Morrison 1983: 5). 

The strong resource base provided by the increasing reliance on the hunting of sea 

mammals from skin boats allowed for population growth (Arnold 1986: 4). This 

emphasis on open water hunting increased, and Birnirk eventually evolved into the 

whaling culture that is now known as Thule culture (Ibid.). This Birnirk-Thule transition 



8 
was complete by A.D. 1000 (Y orga 1979: 287), and within two centuries they had spread 

over most of the inhabitable regions of the arctic, as far east as Greenland. The rapid rate 

at which this migration occurred is supported by linguistic evidence, as Inuktitut, the 

language spoken by the modem Inuit, varies little from the Bering Sea region to Greenland 

(Morrison 1983: 6). 

Many reasons have been proposed for this quick migration, but the reason most 

commonly evoked is climatic change. At about this time, the arctic climate was 

undergoing a warming period; which led to a reduction of pack ice in the Beaufort Sea, 

which in tum led to an increase in the numbers and distribution of certain species of 

marine mammals, including the large bowhead whales. According to many theories of 

expansion, this climatic warming rendered traditional Alaskan methods of hunting whales 

in ice-leads ineffective, which in tum led to the adoption of new open water hunting 

techniques linked with the use oflarge skin boats called umiaks. 

It is often argued that the Thule followed the whales into their new feeding 

grounds in the Canadian arctic archipelago (McGhee 1969170; McCartney 1977). Such a 

theory assumes the Thule economy was strongly based on the hunting of bowhead 

whales. This supposed focus on whaling was first articulated by Therkel Mathiassen: 

Whaling has apparently been one of the principal occupations; this is 
proved both by the construction of the houses, in which whalebone play 
such a great part, and by the material used for the implements, whalebone 
and baleen apparently being the most important; in particular, however, 
the composition of the refuse heap, the large masses of baleen and 
whalebones which appear especially in the lower strata, indicate that 
whaling has been one of the most important means of livelihood of the 
population (Mathiassen 1927b: 85). 

This postulated whale-based economy is often used as the main criterion by 

which one can distinguish Thule culture from the culture of the modem Inuit, and it has 



9 
many proponents (e.g. Mary-Rousseliere 1979; McCartney 1980; Savelle 1987; Savelle & 

McCartney 1994). It may be inadequate for many reasons. The strongest argument for 

the importance of whales in the Thule economy are the whalebone houses found 

throughout the Canadian arctic and the whaling equipment found in association with these 

sites. However, the large harpoons and floats interpreted as being used exclusively for the 

hunting of whales could have just as easily been intended for use with other large sea 

mammals, such as walrus. Also, one should be cautious to differentiate between whale 

bone utilisation and whale hunting (Freeman 1979: 279). These bones could have been 

scavenged from beached whale sites as a substitute for driftwood in the construction of 

houses. 

In more recent years, the reliance on cultural ecology as an explanation for the 

rapid migration has been challenged as being too simplistic (e.g. Morrison 1999), and 

alternative explanations have been offered. One such explanation suggests the migration 

was spurred by the desire for iron (McGhee 1984). The Cape York meteorite in 

northwest Greenland provided a large deposit of meteoric iron, and this theory assumes 

that the Thule in Alaska heard rumours of these deposits from the local Dorset 

populations. One obvious problem with this theory is that it presupposes a degree of 

cultural interaction between Dorset and Thule for which there is no proof (Morrison 

1999: 140). Other explanations for the rapid migration rely on demographic motivations, 

such as population pressure, over-hunting, or warfare (e.g. McCullough 1989: 300-303; 

Morrison 1983: 271-272). 

David Morrison has effectively demonstrated that the Thule did not follow 

bowhead whales as has been suggested in earlier theories, pointing out that evidence from 

the field of biology (Dyke et al. 1996 in Morrison 1999: 146) has shown that the Pacific 

2 For an interesting perspective on the Dorset-Thule transition, see Park (1993). 



10 
and Atlantic whale populations have not met since the Holocene. Morrison, however, 

does not deny that whaling played an important role in Thule culture, and identifies at 

least two migrational episodes, which he refers to as the 'Natchuk Phase' and the 'Ruin 

Island Phase'. Although the migration of earlier groups may have been motivated by 

whale hunting, Morrison asserts that the later migrations may have differed in their 

motivations and the routes taken (Morrison 1999: 151). 

Regardless of how and why Thule people migrated across the Canadian Arctic, it 

is generally agreed that this event began about one thousand years ago. Thule chronology 

is often broken down into three periods: Early Thule, Classic Thule, and Modified Thule. 

Although the dates for each of these periods vary by region, the general agreement is that 

Early Thule falls between 900 and 1100 A.D., Classic Thule is the period between 1100 

and 1200 A.D., and Modified Thule is the period between 1200 and 1600 A.D. 

(McCartney 1977: 219). 

Archaeology and Social Boundaries 

An analysis of variability in material culture in an attempt to delineate regional 

variation requires an important assumption: that the degree of stylistic similarity between 

assemblages is directly related to the degree of social contact, and therefore social 

relatedness, between the groups in question (McGhee 1984: 87). This assumption has 

led to some controversy in the history of archaeology (e.g. Binford & Binford 1966; 

Bordes & Bordes 1970; MeHars 1970), and so a brief discussion of the issues surrounding 

the linking of archaeology and social boundaries is necessary at this time. 

McGhee's original hypothesis regarding regional variation in Thule culture 

(1969170) used the term 'tribe' in defining these units. The term 'tribe' can be defined as 



11 

" ... a social group speaking a distinctive language or dialect and possessing a distinctive 

culture that marks it off from other tribes. It is not necessarily organized politically" 

(Hoebel 1958: 661). This term is now outdated and its link with evolutionary ideas of 

culture renders it ethnocentric. However, the entity which the definition describes is vital 

to a study of this sort. 

Others have provided definitions for a different level of social grouping, called 

'bands'. A 'band' is defined as " ... an explicit community comprising a small population 

in constant face-to-face interaction, sharing life in a locality, although usually wandering 

about some restricted range, camping together. Being exogamous, each band must 

maintain at least some relations of affinality with one or more adjacent counterparts, but 

these gossamer ties do not constitute any significant alteration of the independent nature 

of the group" (Sahlins 1961 and Service 1962 in Fried 1968: 11). By this definition, 

'tribes' are a higher level of social organisation, consisting of a number of 'bands' 

connected by complex social institutions and sharing the common goal of defending their 

territory in wars with other tribes. By many definitions, a 'tribe' must be a discrete unit 

with well-defined boundaries (e.g. Sahlins 1961: 343, n.3), a requirement that cannot be 

met in the real world as the ethnographic record has shown us that the boundaries 

between tribes are often no clearer than those between bands (Fried 1968: 13). The term 

'social group' will be adopted in an attempt to capture the meaning originally intended by 

the above definitions without carrying any of the negative connotations with which the 

terms are now linked. 

Ethnicity 

Ethnicity, most simply defined, " ... signifies a self-perception of common social 



12 
identity expressed in a people's shared traditions, ideology, and common history as a 

people" (Fitzhugh 1987: 141). Although the idea of a 'social group' within a wider 

culture is not equivalent to the idea of 'ethnicity', some of the major concepts described 

by the term apply here. Raoul Naroll (1964: 284) lists six criteria for defining an ethnic 

unit. The first criterion considers the distribution of traits being studied. These traits 

should be relatively uniform within the area occupied by the ethnic unit, and should differ 

from those traits found outside of this area. Secondly, an ethnic unit should be expected 

to be found in an area of territorial contiguity. The third and fourth criteria listed by 

Naroll concern language and political organisation, both of which would be expected to be 

uniform throughout the territory in question. The final two criteria involve the specific 

ecological adaptation expressed by the group, and the local community structure found 

there. It is apparent that not all of these criteria are accessible archaeologically, and the 

model is of questionable value in certain areas of the world (Greaves 1982: 7-8). While 

ecological adaptation may prove to be of vital importance to a study of Canadian arctic 

archaeology, it has been shown that " ... where ecologically homogeneous areas were 

larger than the annual cycles of the local groups could encompass, the social groupings 

were multiple within the area, and the societies were marked by cultural and linguistic 

diacritica, even when following ecologically identical subsistence patterns" (Graburn 

1979: 186). This is the case for Thule culture, where almost all groups relied on the 

exploitation of sea mammals to the virtual exclusion of other species. 

Barth's definition of an ethnic group (1969: 10-11) is much more general, and 

therefore more universally applicable. For him, such a group is biologically self-

perpetuating, shares fundamental cultural values, composes a field of communication and 

interaction, and finally, this unit must identify itself and be identified by others as 



13 
constituting a distinguishable category. While this definition is appealing for its 

generality, only the second criterion is accessible archaeologically, as one would expect 

that shared cultural values would lead to a unity in cultural forms, expressed at the level 

of the artifact. 

In an ideal world for an archaeologist, social groups would always be marked by 

sharp boundaries, and a self-awareness of membership in a particular social group would 

lead to uniformity of material culture within the boundary, and a marked discontinuity 

with material culture outside of this boundary. Unfortunately for the archaeologist, 

human behaviour is far too complex to allow for this ideal situation. Boundaries are often 

blurred, with intermarriage, trading and migration common throughout the territory in 

question. One must also ask whether one can use 'social group' as a meaningful 

archaeological unit if it didn't have meaning for the people themselves. A perfect example 

of this occurs in the historic era of the central Canadian arctic, where ethnographers of the 

late nineteenth and early twentieth centuries (e.g. Birket-Smith 1929a, 1929b; Boas 1888; 

Jenness 1922; Rasmussen 1929, 1931, 1932) delineated Inuit culture into five discrete 

'culture areas' that bore little resemblance to the complex spheres of social interaction 

which marked Inuit culture at that time3
• 

Tyranny of the Ethnographer 

Using a perspective borrowed from H. Martin Wobst, it can be argued that the 

informants on which ethnographers usually rely tend to dichotomise the continuum of 

space into a bounded unit with predictable behaviours on the inside and unpredictable 

behaviour on the outside (Wobst 1978: 305). This spatial construct is then translated by 

the ethnographer into a bounded cultural unit containing a finite set of individuals with 

3 See Megginson (1997). 



14 
shared behavioural patterns. As a consequence, ethnographic work often divides a spatial 

process of hunter-gatherers into populations surrounded by boundaries, whether or not 

these boundaries have behavioural significance. As spatial variability is reduced, pattern 

and homogeneity are artificially produced or exaggerated, and cultures or societies are 

created (Ibid.: 306). Wobst has termed this process the 'tyranny of the ethnographer', 

and it can be shown to have been active in the central Canadian arctic, where the largest 

meaningful social units, at least prior to 1960, were the winter settlements of about one 

hundred people (Maxwell 1979: 85). An examination of Thule regional adaptations in 

terms of these artificially created social groupings, therefore, would perpetuate this 

'tyranny' and bias the results. 

Wobst is not unique in his identification of a problem with the creation of 

ethnographic 'cultures': 

Boundaries of social groups are often obscure in living societies and 
probably were equally unclear in prehistoric ones, not to speak of their 
uncertainty in archaeological remains (Fitzhugh 1987: 141). 

Fitzhugh thus addresses an issue not specifically discussed by Wobst - the fact that social 

groups fluctuate in the living world, yet classic ethnographic studies and archaeology 

based on the idea of culture history describe the relationship between material culture and 

ethnicity in static terms (Fitzhugh 1987: 141). Fitzhugh cautions the archaeologist that 

'tribal' ethnic units are not always equivalent to archaeological culture units, and that the 

only way to reliably extend historic social divisions into a prehistoric sequence is through 

a direct historical approach (Ibid.: 142). This approach, though, must work on the 

assumption that these ethnographically derived 'cultures' are real, discrete entities, a 

concept which this author is not ready to accept. Regardless of whether or not 

Fitzhugh's suggestions are followed in their entirety, his cautions regarding the limits of 



15 

archaeological data must be accepted when he asserts that the data with which an 

archaeologist must contend " ... represent specific events which archaeologists are too 

frequently tempted to interpret as long-standing patterns" (Fitzhugh 1987: 149). 

The Meaning of Style 

Returning to the work of Wobst (1977), he has further suggested that stylistic 

behaviour should be considered in terms of the life history of an artifact and its jUnction as 

a means of information exchange. All human behaviour involves the potential for 

information exchange, and therefore the context of transmission is as diversified as human 

behaviour. The artifactual mode of transmission is one such context. The types of 

information that are best transmitted through stylistic content are messages of emotional 

state, identification of ownership and authorship, religious and political affiliation, and 

ethnic affiliation (Wobst 1977: 324). The utility of style for conveying a message 

decreases as the emitter and potential receivers become more closely acquainted with one 

another - it is at this point that the message becomes redundant. The amount of stylistic 

behaviour, then, should positively correlate with the size of the social networks in which 

the individuals participate (Ibid.). Such a correlation should allow us to make hypotheses 

concerning changes in the extent and nature of social networks as reflected in stylistic 

change. 

Variation in artifacts can be caused by a variety of interconnected factors. One 

such factor is the personality of the individual artisan: 

If different people are given the same task, under similar physical 
conditions and using similar materials, results of their activities are 
distinguishable (Clegg 1977: 60). 

A second influencing factor is the medium used to produce that artifact, as certain raw 



16 
materials can be characterised by technological constraints. Another factor is the function 

for which an artifact will be used once completed, as certain traits may only be 

advantageous for particular functions. A final influencing factor concerns the culture 

within which the artifact is being produced. This type of variation can be identified when 

" ... the artifacts in question are different though made by the same means from the same 

material, and for the same function" (Ibid.: 60). 

As this study proposes to identify regional differences in harpoon head attributes, 

the fourth factor listed above will be of primary importance. Clegg gives the following 

advice for isolating culture as a cause of artifact variation: 

If we want to know about cultural differences, we should look at artifacts 
made in the same medium, for the same function, by people of the same 
personality, but from different cultures, thus examining the effects of one 
variable at a time (Clegg 1977: 60). 

This statement is problematic for many reasons, one of which is Clegg's failure to clarify 

what is meant by the term 'culture'. A very loose definition states that a culture is 

marked by " ... people living in one place at one time" (Clegg 1977: 60), but this is too 

vague to be of any use to a study looking for variation on a local and regional level. Also, 

while the medium with which the artifact is produced is easily controlled for in 

archaeological research, functional differences are not always as easy to determine from 

artifact analysis alone4
• As for the personality of the manufacturer, it is unclear how this 

could ever be accurately accessed through archaeological inquiry. Clegg's suggestion is 

that idiosyncrasies caused by differences in personalities can be simply treated as 'noise', 

and left to cancel themselves out (Ibid.), but this is of limited value in a culture in which 

each site only contains a few specimens of each artifact type. Clearly, Clegg'S model 

cannot be adopted without significant revision, though the general concepts with which it 

4 See discussion concerning the debate over harpoon head functions in Chapter IV below. 



17 

is concerned will be used throughout this study. 

The Uses of Style 

A stylistic analysis of artifacts must first pose the question, what is 'style'? 

Wobst (1977: 317) has argued that while style is integral to most archaeological research, 

it lacks meaning. By this he is arguing that style is most often treated as a negative 

category - artifact variability that cannot be attributed to other factors, such as function, 

is labeled 'stylistic', and not subjected to any further testing. Style is most frequently 

treated in isolation from other factors, with little awareness of how it articulates. with 
! 

other cultural variables or the adaptive advantages it may bestow on the artifact (Ibid.). 

Style is described as being an afunctional aspect of artifact variation, and as such, it is 

inaccessible archaeologica1ly. 

Style should, Wobst argues, be seen as an important conveyer of communicative 

information. The maintenance of a particular style can be explained as a product of 

homeostasis in communication processes, and the uniformity of this style over geographic 

space can be interpreted as implying high levels of communication between people living 

within this area (Ibid.: 318). Alternatively, a break in the continuity of style could 

indicate a break in communication density, and therefore a separate socio-cultural unit. 

Wobst argues that archaeologists can measure the degree of communicative equilibrium 

directly through analysis of the temporo-spatial distribution of stylistic form (Ibid.: 319). 

Style can be defined as formal variation in material culture that relates to the role 

of artifacts in processes of information exchange (Ibid.: 321). Information exchange refers 

to all communicative events in which a message is emitted or received. Although a 

message will not be emitted unless there is a potential receiver, the emitter and receiver 



18 

can be separated both spatially and temporally (Ibid.). Artifact style will tend to convey 

simple messages, such as emotional state, ownership, authorship, or religious and political 

affiliation (Ibid.: 323). The utility of style for conveying a message decreases as the 

emitter and potential receivers become more closely acquainted with one another. 

Stylistic messages are of maximum value if the potential receivers are unlikely to receive 

the message through any other means, yet are still close enough to receive it (Ibid.). 

Wobst argues that the amount of stylistic behaviour should positively correlate 

with the size of the social networks in which the individuals participate, such that band 

societies are not expected to show much stylistic behaviour (Wobst 1977: 325). 

Although it is apparent that Thule culture will not show the same amount of variability as 

would be found between state societies, it has been argued that style can be expressed at 

many different levels of ethnic resolution, from great culture-historical complexes to clans 

in a region, to kin groups within settlements (Sackett 1990: 33). To define regionally 

distinct social groups in Thule culture using an analysis of artifact style, then, does not 

seem an impossible task. 

Wobst's approach remains an important source of information that will allow one 

to predict which artifacts are most likely to be used as conveyers of communicative 

information, and thus will show the most stylistic variability. The less an artifact is 

visible to members of a group, the less likely it is to carry a stylistic message (Wobst 

1977: 328). By this, one would predict that those artifacts which never leave the 

household, such as kitchen tools, will be the least likely to show stylistic variation. In 

contrast, those items which are potentially visible to all members of a social group, such 

as outer layers of clothing or house styles, are more likely to show a specific expression 

of style (Ibid.: 329). However, this stylistic signal can only be expected to differ from 



19 
that of surrounding groups if its function is to explicitly broadcast social group affiliation 

and is active in the process of boundary maintenance (Ibid.). 

A similar opinion, based on the idea of a culture-historical approach, is expressed 

by Glynn Isaac: 

Experience up to now suggests that the peculiarities of the most 
highly 'designed' components of stone tool industries provide the 
best markers of idiosyncratic 'phases' and 'provinces' within the 
culture transmission system - and of continuity and interchange 
between phases and provinces (Isaac 1977: 8). 

Unfortunately, Isaac does not provide a definition for 'designed components', nor does he 

explicitly state how one would determine which artifacts would be most sensitive to 

'stylistic' and 'fashion' differentiation (Ibid.). Isaac suggests a study of style be 

restricted to stylistic attributes, and a study of function should be restricted to functional 

attributes (Ibid.), but Wobst has stated that these categories are not mutually exclusive 

(1977: 317). 

While the social structure of Thule culture is merely inferred from ethnographic 

analogy, enough is known to allow us to test Wobst's theory. One would predict that 

household objects, such as those most commonly identified with women, would be the 

least likely to show regular stylistic variation, while house styles and men's tools, both 

visible over a wider area, would be more likely to show this variation. Harpoon heads are 

generally classified as a men's tool, and so according to Wobst's theory, they would be 

expected to be good indicators of social affiliation. 

Another argument in favour of the use of artifact style in determining social 

boundaries is put forward by Sackett (1990), using the concept of isochrestic variation. 

This term, a neologism from Greek translating as 'equivalent in use', is based on the idea 

that there is a spectrum of equivalent alternatives for attaining any given end when making 



20 
or using material items (Sackett 1990: 33). Artisans are aware of a few options at their 

disposal, and their choices are dictated by the technological tradition to which they 

belong. These choices tend to be consistent within the group, but are subject to change 

over time due to exposure to alternative options from outside influences. The implication 

of this is that " ... each social group or unit of ethnicity tends to possess its own 

distinctive style, and the overall degree of stylistic similarity represented by two groups' 

material cultures taken as wholes can be regarded as a direct expression of their ethnic 

relatedness" (Ibid.). In other words, stylistic distance is equivalent to social distance. 

According to Sackett, isochrestic variation can manifest itself anywhere. The 

most obvious source of stylistic data is decoration on artifacts, and it has infinite 

possibilities because it is not constrained by functional considerations. Decoration is 

considered an adjunct form because it is added to an artifact once it is complete. 

Instrumental form, however, is also a good source of style data - choices must be made 

regarding such things as raw material and production techniques (Sackett 1990: 33). 

While Wobst (1977) sees style as a multidimensional entity, varying too randomly 

to be properly analysed, Sackett (1990) sees style as varying in regular and patterned 

ways. The whole idea of isochrestic variation is that style is governed by a finite number 

of choices on the part of the person doing the manufacturing. This is an appealing idea 

since it would facilitate the interpretation of style as a marker of social identity. If style 

is simply idiosyncratic as Wobst would have us believe, this study would be an exercise 

in futility. Perhaps any findings resulting from this study will contribute to this 

seemingly endless debate. 



ill 
Methodology 

Methodological Problems 

Thule culture offers a virtual cornucopia of artifact types over which an 

archaeologist may pore. Therkel Mathiassen's 1927 treatise alone defines eighty-seven 

basic functional categories of Thule artifacts (Mathiassen 1927b: 4-6), each divided into 

numerous subcategories, and this number is constantly being expanded (e.g. Morrison 

1983; McCartney 1977). The functional categories were defined in reference to the 

toolkit used by the historic Inuit and observed in ethnographic study, and while some 

proposed analogues may be questioned from time to time 5, almost all arctic archaeological 

monographs to this day are modeled on Mathiassen's work. However, rarely does an 

archaeologist have an entire toolkit available for analysis. A number of factors in arctic 

archaeology make artifact analysis difficult, and each of these factors will be discussed in 

turn. 

Thule sites are ubiquitous in the Canadian arctic, large tracts of coastline are often 

covered by a series of small sites, sometimes so close together that it is difficult to tell 

where one site ends and the other begins (e.g. McCartney 1977). Partly due to this 

abundance of sites waiting to be examined, and partly due to the shortened field season 

and often impermeable permafrost in the arctic, Thule sites are rarely, if ever, excavated in 

their entirety. Arctic archaeologists often frame their research on a regional survey scale, 

doing test pits and mapping surface features at a number of sites, with only a few houses 
5 e.g. see (Morrison 1983) for an interesting reanalysis of what Mathiassen (and almost all archaeologists to 
follow him) have called 'snow knives'. 

21 



22 
being excavated out of the multitude of houses located. Nor are these houses chosen in a 

random way: archaeologists will choose which features to excavate based on their specific 

research design, such that a researcher studying issues of initial Thule migration would 

specifically target those houses which, by outward appearance, seem the oldest: 

The large winter houses, well preserved and lavish with whale 
bones, were not touched.... They would probably yield much data, 
but such structures require, for their detailed excavation, more time 
than the small field party could afford them. More importantly, 
the more worn-down houses give at least some hint of being older 
and early Thule information was a prime goal of the work. 
Possibly, those less prominent houses were rather temporarily 
occupied or occupied during transitional seasons. They might also 
be less conspicuous only because much of their building material 
had been removed to make the prominent houses whose age might 
be only slightly less old (Taylor & McGhee 1981: 11). 

Another problem in Thule sites is the probability that many houses were 

reoccupied periodically, in some cases perhaps spanning hundreds of years. Stratigraphy 

is often disturbed by periodic cleanings by house occupants, and so artifacts showing 

chronological differences thought to be separated by centuries may be found side-by-side. 

It is almost impossible to control for these chronological differences as most arctic 

chronologies are based on the seriation of artifact styles and thus are quite relative and 

subjective. Radiocarbon dating is unreliable in an arctic environment due to the 'marine 

reservoir effect' (Arundale 1981; McGhee & Tuck 1976), which causes dates obtained 

from sea mammal bones to be much earlier than the actual age of these bones due to the 

recycling of fossil carbon into the marine food chain. Although correction curves have 

been developed (e.g. Stuiver & Pearson 1986), this curve does not take into account 

regional variation in marine reservoir effect, and even if this were possible, the migratory 

nature of sea mammals would render it inaccurate (Tuck & McGhee 1983: 9). Due to 



23 
these seemingly unsolvable problems, many archaeologists today have followed Tuck and 

McGhee's recommendation (1983) that no attempt should be made to derive radiocarbon 

dates from sea mammal bone. Dates obtained on some types of wood in the arctic have 

also been called into question (e.g. Schledermann & McCullough 1980: 840), largely due 

to their likelihood to become saturated with sea mammal oil, although some naturally 

occurring factors have also been considered (Maxwell 1985: 253). 

The problems that are introduced at the data collection stage of archaeological 

study in the arctic are perpetuated at the analytical stage. Stylistic differences in artifacts 

are generally attributed to chronological differences. Therkel Mathiassen developed a 

general chronology of artifact attributes based on a general progression from simple to 

complex, and inferences made regarding the age of particular sites based on elevation 

above sea level (1927a: 2). This latter technique requires the assumption that sites were 

always placed near the shoreline, and that subsequent isostatic rebound has raised these 

sites to higher elevations. While isostatic rebound is certainly an important factor in the 

placement of some sites in the arctic, this has been shown not to hold true for the entire 

arctic. Allen P. McCartney (1977) has convincingly shown that sites on the coast of 

northwestern Hudson Bay were consistently placed on high elevations with panoramic 

views, regardless of age. Despite this, Mathiassen's chronology is now often accepted 

without question, with archaeologists deeming a site 'Early' or 'Late' based on 

Mathiassen's criteria (e.g. Savelle 1987; Schledermann 1975, Taylor & McGhee 1981). 

His site chronology is accepted without question, despite of the fact that these same 

archaeologists would never use Mathiassen's criteria for determining the age of artifact 

styles (Park 1994: 32). 

Seriation as a dating technique requires two assumptions: that styles changed over 



24 

time in a directional manner, and that time is the only factor causing this variation (Park 

1994: 31). In an interesting study, Robert W. Park (1994) compared the perceived 

relative age (based on Mathiassen's estimations) of individual harpoon head attributes 

from seventeen sites across the arctic with radiocarbon dates collected from those same 

sites. It was found that many attributes considered to be 'Early Thule' were found just 

as frequently in sites deemed to be 'Late Thule'. This calls into question the entire 

sequence of Thule occupation that has been proposed for the arctic, and leads one to 

consider the possibility that some of the variation in artifact attributes can be explained 

through regional rather than temporal variation: 

The diversity of Thule adaptations has long been recognised (e.g. 
Taylor 1966; 1968) but when we encounter variability in the 
archaeological record we usually attribute it to particular types of 
change over time without examining the possibility of synchronic 
differences (Park 1994: 44). 

From the above, it is apparent that arctic archaeologists are plagued with small and 

selective sample sizes and with a largely undefined chronological framework. What is a 

researcher to do? There is no simple answer to such a complex conundrum. One cannot 

completely eliminate the bias inherent in a regional study of Thule artifacts. Statistical 

analysis seems inappropriate in a field where random sampling is rarely if ever practiced. 

Small sample sizes are another factor that render statistical analysis difficult, as a given 

site assemblage may contain over a hundred functional or stylistic classes, each 

represented by only one or two specimens (Morrison 1983: 7). Despite these 

difficulties, some minor quantitative investigations will be attempted in this thesis. 

It is apparent that it would be an impossible task to attempt to control for 

chronology in a study of Thule archaeology. For the purposes of this study, sites 

classified as belonging to all three periods of Thule Culture (Early, Classic and Modified) 



25 

will be considered in the analysis. If stylistic differences are identified between sites, 

these sites will then be cross-referenced with their generally accepted chronological 

classification to determine whether the perceived stylistic difference is due to 

chronological or regional factors. 

The Study Area 

Thule culture covered a vast area of the arctic, from Alaska to Greenland, and 

south along both sides of Hudson Bay. The large size of the Thule culture area makes it 

very difficult to select a specific area for analysis. The fact that this study aims to 

discover regional differentiation requires that a large area be examined, but it would be far 

beyond the scope of this work to make an attempt to investigate the entire arctic. Ideally, 

the study area would comprise regions with slightly different local environments, and 

containing the traditional lands of more than one traditional Inuit social group. Because 

only one artifact type has been selected for study, this area will in fact be larger than the 

absolute minimum required by the study, in order to maximise the potential of the 

analysis to find regional trends. 

The central Canadian Arctic is the area of focus for this study. This region 

extends from 65°W to 115°W, and from 63°N to 80oN. Additionally, the study area will 

include land no further south than 63°N, and no further north than 800 N. This area 

comprises the land claimed by the following Inuit groups: Baffinland Inuit, Iglulik, 

Sadlermiut, Netsilik, Copper Inuit and the Caribou Inuit6 
, and spans an area that contains 

many different ecological zones. The study area will be described in terms of three 

divisions: Western, Central, and East. 

6 There is much debate over how long the Caribou Inuit have inhabited this region (e.g. Clark 1977; Burch 
1978). 



26 
Western Central Arctic 

The western central arctic comprises Victoria Island, Banks Island, and the 

mainland surrounding Coronation Gulf. This area is less known archaeologically than the 

eastern arctic, but a few important archaeological studies have taken place here. Diamond 

Jenness visited this area during the Canadian Arctic Expedition of 1913 -1918, and 

described the Copper Inuit who reside there in a volume published in 1922. William E. 

Taylor, Jr. visited the area in 1963, and published a full report of his findings in 1972. 

Finally, Robert McGhee has done much work in the western central arctic, with works 

being published in 1970 and 1972 based on his excavations in the Bloody Falls region and 

western Victoria Island. 

Three sites on the western Coronation Gulf coast are considered in this analysis. 

Clachan (NaPi-2) is located on a small bay south of Cape Hearne on the western coast of 

Coronation Gulf. It contains three houses located on a high point of land, but one of 

these houses demonstrates evidence of previous disturbance. Because these three houses 

were found to cluster rather tightly, it was possible for the researchers to excavate the 

entire site in sixty-two contiguous two metre squares (Morrison 1983: 47-49). The 

Beulah site (NcPf-2) is located on the western coast of the peninsula. This site is much 

larger, containing twelve houses in a line parallel to the sea, and three to five metres above 

sea level. Only one house was chosen for excavation at this site as time was limited 

(Ibid.: 68-72). The latter two sites are both located near the junction of Dolphin and 

Union Strait and Coronation Gulf, the strong current from which does not allow for the 

production of thick ice. These rare conditions allow for an abundance of sea mammals 

such as bearded seal and ringed seal. Caribou also appear to have been abundant in the 

vicinity (Ibid.: 69). A total of one thousand eight hundred fourteen artifacts were found 



1-~1tlmd 
2-C~Gulf 
l-1Gng William. Itlmd 
4.-Booihia Pminruh. 
:5- SOJYU!IJet Itlmd 
6- Com.waJIU Itlmd 
'-Bdwnt Itlmd 
~-Devon Itlmd 

!)i- Chestafi.d.d Inlet 
1(1- S01dha:mpton Itlmd 
ll-R.tpuI.t;e Bay 
12-:r.IEhoiIle Pminruh. 
ll-Bylot Itlmd 
14.-Pcobidu!l" Bay 

km 
iii , i 
o 200 400 

Figure 3-1: Map of Nunavut, showing some of the geograpbkal features 
mentioned in the text 



28 
such as bearded seal and ringed seal. Caribou also appear to have been abundant in the 

vicinity (Ibid.: 69). A total of one thousand eight hundred fourteen artifacts were found 

at these three sites, the majority of which originate from the most completely excavated 

site, Clachan. 

A large scale regional site survey between Cape Parry and Cambridge Bay was 

undertaken by William E. Taylor, Jr. in 1963, resulting in the mapping of numerous 

archaeological sites dating from the Pre-Dorset occupation to that of the historic Copper 

Inuit. Only those sites which produced Thule artifacts are considered in this study. One 

of these sites, Lady Franklin Point, was thought temporarily to be the source of the 

Semmler collection of Thule artifacts, discussed in an earlier publication by Taylor (1963: 

458). This site is located on western Victoria Island, and contains twenty-one Thule 

houses. Excavation was limited to nine test cuts in probable midden areas, and the 

resulting artifacts showed that this site was of much more recent age than the Semmler 

collection, the source of which remains to be found (Taylor 1972: 35). 

More Thule sites were found further east on Victoria Island, near Cambridge Bay. 

The Pembroke site (NgNc-2) was found to contain a total of twelve tent rings and Thule 

houses on a small knoll which overlooks a creek bank. One house was excavated, and 

fifteen artifacts were recovered (Ibid.: 44). The Clare site (NgNc-3) is also found in the 

vicinity of Cambridge Bay, and it contains seven oval tent rings arranged in a row. A test 

cut made within one of the tent rings, yielded a few Thule artifacts (Ibid.: 46, 49). 

Finally, the Bell site (NiNg-2) is located slightly to the west of the former two sites, in 

the Ekalluk River area of Victoria Island. This site contains sixteen Thule houses, only 

one of which was investigated. Both Dorset and Thule artifacts were found in the two 

test cuts made in this house (Ibid.: 54). 



29 
During the summers of 1966 and 1968, Robert McGhee investigated several sites 

in the western Victoria Island region, as well as in the vicinity of Kugluktuk7 on the 

mainland. Two of these sites concern the Thule period of occupation and so are of 

importance to this study. Memorana (OdPq-1) is located on the Amundsen Gulf coast of 

western Victoria Island, nine miles southeast of Holman, in an area that enables very 

productive seal hunting. Four small Thule houses were found here, and all four were 

excavated, as well as 90% of the midden deposits. The shallow nature of the middens 

suggested this site was occupied for a relatively short period of time 8 (McGhee 1972: 21, 

23). 

Another Thule site discussed by McGhee is the Bloody Falls site (MkPk-3), 

located on the western bank of the Coppermine River, adjacent to Bloody Falls, the 

location of the famous massacre of Inuit by Chipewyan Indians recounted by Samuel 

Hearne in 1771 (Hearne 1972: 98-104). Caribou and musk-oxen are rare sightings in the 

region, but the falls have been an important fishing spot for various hunting groups for 

over three thousand years (McGhee 1972: 39). Nearby is an important source of native 

copper, which was traded throughout Thule territory. Two of the five large Thule houses 

are found at this site were completely excavated. The artifacts recovered reflect the poor 

preservation of organic material found at this site (Ibid.: 39-40). 

7 This settlement was formerly known as Coppermine. 
8 Based on dry bone weights of the fauna recovered, McGhee estimates that five individuals, each 
consuming about two pounds of meat per day could have survived four months at this site. Taking into 
account the amount of food likely consumed elsewhere, he suggests the faunal findings are consistent 
with a single winter occupation of this site by a group of about twenty people (McGhee 1972: 23). 



WESTERNRECION • EASTERNRECION. 
:M:airJl.aM. W. 1-BeuhlL., N.nruk N\V:llJ.WGJl. 21-Silum.iut S BdIin 31-Niut.mg ( Q) 

\id(IJia. I. 

2-Ckchm 
3-Blood'j Fills 
4-Mem.0J:iIrIA 
5-:w.r, Pmlklin R. 
6-Bell 
1-Pem."Jmu.:e. Clm 

CEN'I'RtU. RECION • 

Md\'iIle 1'. 

NBdirLI. 

I<:in( Willi.un.I. 8-Mal.eruilib: 
9-PeJ.1y Ba'!. 
Km"Igemf:igs~ 

Bont.ialS(JJ"J1IIDS lO-Spen.ce Bll'j 
II-N.ldlI.Jlm 
12-Levesque Habour 
13-FOIt Ross 
14 -C:reS'iWll Bll'j 
15 -I.e mmord:h., Q.1o Ilb: 

22-Kmn.;mrib: 32-~ (Ja..l) 
23-E.Jb: 33-B-l 
24-Nmjm 34- C:rj"ml. II 
25-Lyon Jnlet... Vms:d:tm 
26-Pingerqilib:. Igloolib: 
21-N.mgu:trib: 
28-Button Po:ird. 
29- Qil.a1ub:m 
30-M:itim.M.ilik 

16-1(1.l(2.~.LIlb:e 
11· Broomm Po:ird. 
18·Deblicqu'f 
19·Port Refuge 
20·Pordm Pow. RbJr·1 

Figure 3-2: Map of Nunavut, showing sites being used in this study. 



31 
The Central Region 

The central region of the central Canadian arctic is perhaps the best known 

archaeologically. This area contains Devon Island, Somerset Island, Boothia Peninsula, 

and King William Island. It is here that we begin to enter the territory explored by 

Therkel Mathiassen, though the bulk of his sites will be reserved for the discussion of the 

eastern central arctic. Malerualik is located on the southern coast of King William Island 

at Simpson Strait, between McClintock Bay and Douglass Bay. This is a very large site, 

containing sixty-eight house ruins and over one hundred tent rings and other features. 

Thirteen of the houses were excavated by Knud Rasmussen in the summer of 1923, and 

two hundred ten artifacts were recovered. This site is located near an important caribou 

migration route (in fact, the name means 'the place where one follows the caribou'), and 

also boasts good sealing and salmon fishing (Mathias sen 1927a: 305-307). 

Many sites are located in the Boothia Peninsula-Somerset Island region. Two 

thousand artifacts were collected by L.A. Learmonth, a post manager at Fort Ross, 

between 1939 and 1949. These artifacts have been described in a report by James W. 

VanStone, published in 1962. While the trained archaeologist may consider such a 

collection of artifacts questionable because of unskilled collection techniques and inexact 

provenience recording, VanStone justifies his report as follows: 

Ordinarily, the value of a report of this kind might be considered 
dubious, and this would certainly be true in an area that is well 
known archaeologically. This can by no means be said about the 
central arctic .... (VanStone 1962: 2). 

In 1962, VanStone's comments regarding the paucity of archaeological information in the 

central region were not exaggerated; other than Mathiassen's 1927 report, very little work 

had been done here. Although this situation has happily changed in the decades that have 



32 
passed since VanStone's study, his report will still be considered in this study. 

The Levesque Harbour site, which contains four whalebone houses, is located on 

the northeastern tip of Boothia Peninsula, just south of the entrance to Bellow Strait 

(VanStone 1962: 8). Three sites at Creswell Bay will be considered as one unit, and all are 

located on the southeastern side of the bay near Cape Garry. Each site contained twenty 

to thirty whalebone houses (Ibid.: 15). Nudlukta is a site located on a small stream that 

connects Nudlukta Lake and Nudlukta Inlet, sixteen miles north of Levesque Harbour. 

An unspecified number of Thule houses were found here, of which five were excavated 

(Ibid.: 18). The Fort Ross site is located near an old Hudson Bay Company post on 

Brentford Bay, on a peninsula that juts from southeastern Somerset Island close to the 

entrance of Bellot Strait. Five Thule houses were located at this site, but the artifacts 

were recovered mainly through surface collection (Ibid.: 23). The Spence Bay site is 

located at Netsilik Lake, and the artifacts here were also recovered through surface 

collection (Ibid.: 28). VanStone also describes artifacts found at historic grave sites, but 

these will not be included in this study as they fall outside of the Thule period. 

VanStone's archaeological work in the Somerset Island/Boothia Peninsula region 

was continued by William E. Taylor, Jr. and Robert McGhee in 1979, who investigated 

three sites along eight kilometres of shoreline at Creswell Bay, Somerset Island. The 

Learmonth site (PeJr-I) is located on the north shore of the bay (93°48'W, 72°47'N), in 

an area with abundant sea mammals, migratory birds and caribou. This site contains 

seventy-eight dwelling structures in total, twenty-eight of which have been defined as 

Thule houses. Eight trenches were dug in possible midden areas, and resulted in the 

collection of four hundred fourteen artifacts identified as being of Thule origin (Taylor & 



33 
McGhee 1979: 5, 8-11, 22).9 A nearby site was named, quite suitably, the Near site 

(PeJr-2). Forty house ruins were found here, none more than ten metres above sea level. 

Little time was spent investigating this site, with only eight Thule artifacts being collected 

from the surface, and no excavation taking place (Ibid.: 49). Five kilometres from the 

Near site is the Quoak site (PeJq-1), at which were found fifty-five house ruins, twenty-

nine of which were classified as being of Thule origin. These houses lie between three and 

six metres above sea level, and they are arranged into three rows. Excavation at Quoak 

consisted of surface collection and six test cuts, all of which were limited to a depth of 

twelve centimetres due to permafrost. One hundred three artifacts were recovered (Ibid.: 

51-52). 

While many other archaeological investigations have taken place in the central 

region of the Canadian arctic, most of these have centred on high arctic locations. The 

earliest of these were conducted by Henry B. Collins, who investigated four Thule 

villages located near the Resolute weather station on Cornwallis Island. Three of these 

sites are known as Mi, M2 and M3, and will be discussed together as the sample size 

from each site is very small, and the sites are all located very close together1o
• M 1 

contains twelve Thule houses, of which three were excavated in 1949. Two of nine 

houses have been excavated at M2, and at M3, located on the southern side of Cape 

Martyr, only one of five houses was excavated (Collins 1952: 48-49). A final site in this 

region investigated by Collins is known as the Lake site. This site is located between two 

lakes, a half mile inland, and a quarter mile from the weather station. Nine house ruins 

were discovered here, of which three were chosen for excavation (Ibid.: 49-50). The total 

9 Taylor and McGhee also found two hundred ninety-six artifacts which they termed 'culturally non-specific' 
(1979: 22) which are not included in this study in an attempt to maintain a small degree of chronological 
control. 
10 M1 and M2 are separated by only 280 yards (Collins 1951: 50). 



34 
number of Thule artifacts recovered from the four sites was one thousand one hundred 

specimens (Ibid.: 51). 

On Bathurst Island, there are two sites which have undergone major excavation. 

The first of these sites, Deblicquy (QiLe-l), is located on the central east coast of 

Bathurst Island (75°29'N, 97°29'W), one hundred kilometres from the Collins sites 

discussed above. This site, excavated by William E. Taylor Jr. and George R. Carruthers 

in July of 1961, contains twenty-four house ruins, located twenty-two metres above sea 

level and two hundred fifty metres inland. Three houses were excavated at this site, 

yielding a total of three hundred forty-seven artifacts (Taylor & McGhee 1981: 1-11, 25). 

The second major site on Bathurst Island, Brooman Point (QiLd-l), was excavated during 

the summers of 1976, 1979 and 1980 by Robert McGhee. This Thule village appears to 

have been built on a former Late Dorset settlement, and contains twenty houses 

designated as Thule. Ten houses and one midden were excavated, resulting in the 

recovery of six hundred fifteen identifiable artifacts (McGhee 1984: 2, 8, 41). 

Three sites located on the Grinnell Peninsula, Devon Island complete this 

discussion of the central portion of the central arctic. These sites, excavated by Robert 

McGhee in the summers of 1972, 1976 and 1977 have been reported in great detail in a 

Masters thesis written by Robert Park in 1983. The first site is known as PordenPoint 

Brook Village (RbJr-l), and is located on the southeastern corner of the Grinnell 

Peninsula, on the western tip ofPorden Point (76°15'N, 93°40'W). This site straddles a 

brook, and contains nine Thule houses in two rows on either side of the brook. Two 

houses were excavated here, resulting in two hundred thirty-two artifacts (Park 1983: 1, 

11). Porden Point Pond Village (RbJr-4) is located one hundred sixty metres southwest 

ofthe former site. Three Thule houses were located here, and one was excavated. Eighty-



35 
eight artifacts were collected (Ibid.: 77-83). The final Thule village on the Grinnell 

Peninsula is Port Refuge (RbJu-1), located thirty kilometres west of Porden Point on the 

southern coast of the peninsula (76°17'N, 94°45'W) (Ibid.: 1). Five Thule houses were 

found here, all of which showed some evidence of previous disturbance. Two houses and 

part of a midden were excavated, resulting in one hundred ninety-seven artifacts (Ibid.: 

117). A cache (RbJr-7) found in this region was also excavated. It is located two 

kilometres west of the tip of Porden Point, at an elevation of twenty-two metres. Sixty-

six artifacts, mostly hunting equipment, were collected from this cache (Ibid.: 101). 

The Eastern Region 

A discussion of the eastern region of the study area must begin with the most 

famous Thule site, Naujan. This site is located on the northern coast of Repulse Bay, 

which separates Melville Peninsula from the mainland. During the summer of 1921, 

Naujan was excavated by Therkel Mathiassen, and was the basis for his definition of 

Thule culture. A total of twenty Thule houses were observed at this site, and twelve of 

these were excavated. By modem archaeological standards, this excavation was far from 

being thorough since for most of the two months the site was being excavated, 

Mathiassen was working alone (Mathiassen 1927a: 4). It would be a logistical 

impossibility for a solo excavator to fully excavate twelve houses and a large patch of 

midden in two months, yet this site remains the generally accepted Thule type site, and 

researchers working across the arctic continue to draw comparisons between their findings 

and those of Mathiassen. A total of three thousand specimens were discovered here by 

Mathiassen, of which 800 were deemed unfinished or indeterminable (Ibid.: 5). 

A nearby site published in the same seminal work by Mathiassen (1927a) is 



36 
known as Kuk. This site is located in Duke of York Bay, on an estuary of the Thomsen 

River on northern Southampton Island. This bay is rich in a variety of sea mammals, and 

the river is abundant in salmon. Three groups of Thule houses were discovered here, 

totaling twenty-one ruins. Eleven of these houses and seventy-six square metres of 

midden were excavated by Mathiassen between August 20th and September 6th of 1921, 

and the resulting find consisted of seven hundred eighty artifacts (Ibid.: 223). 

The region of northwestern Hudson Bay contains many more sites, few of which 

have been published in any detail. The work of Allen P. McCartney in this region is an 

exception to this trend. In 1962, McCartney led a large regional site survey of the 

western coast of Hudson Bay between the western coast of Roes Welcome Sound and 

Chesterfield Inlet. Many sites were located, but only a few were excavated (McCartney 

1977: 36). Silumiut (KkJg-1) is located on an island north of Cape Silumiut (63°41 'N, 

90° 05'W), and contains twenty-eight Thule house ruins. Seven houses were excavated 

here, and six middens were tested. The resulting collection (not including the thousands 

of quartzite flakes recovered) was about six hundred artifacts (Ibid.: 6-7). The Kamarvik 

site (LeHv-1) is located further north along the coast, on a peninsula that juts into 

Hudson Bay just south of Wager Bay (64°45'N, 87°19'W). This site contains sixteen 

houses arranged into three distinct clusters. Two houses were excavated here, and one 

house was tested, resulting in seventy-eight artifacts (Ibid.: 152-170). A site known as 

Igluligardjuk was also excavated, but as no diagnostic artifacts were found here (Ibid.: 

327), it will not be included in this study. 

Between the region referred to as northwestern Hudson Bay and northern Baffin 

Island, is a strip of land called the Melville Peninsula. Mathiassen (1927a) refers to 

artifacts from a number of sites in this region, and they will be considered in this analysis. 



37 

Two of these sites are located just to the east of Naujan, and they are Vansittart Island 

and Lyon Inlet (Mathiassen 1927a: 124). Only one harpoon head is known from the 

former site, six from the latter. Further to the north are Pingerqalik and Igloolik, from 

which one and two harpoon heads were found, respectively. The artifacts from these 

sites were not excavated by Mathiassen himself, so little information is given regarding 

these sites, many of which are simply listed as 'scattered finds' (Ibid.). Some of these 

harpoon heads appear to be from grave sites, and this difference will be considered in the 

analysis. 

Two more sites that were excavated by Therkel Mathiassen during the Fifth Thule 

Expedition were Mitimatalik and Qilalukan, both located on northern Baffin Island, near 

the modem settlement of Pond Inlet. These two major sites, as well as several minor sites 

in the vicinity were excavated during the summer of 1923, aided in part by fellow 

researcher Peter Freuchen. Mitimatalik is smaller than most of the sites examined by 

Mathiassen, with only two Thule houses, and two 'autumn houses'. One house was 

excavated, as well as four square metres of midden (Mathiassen 1927a: 133-134). 

Qilalukan (the name means 'narwhals') is located three kilometres from Mitimatalik, and 

has a total of nineteen Thule houses. Five houses were excavated by Mathiassen and 

Freuchen, as well as thirty-two square metres of midden (Ibid.: 136-139). The total 

number of artifacts recovered from the Pond Inlet region was two thousand eight hundred 

specimens: one thousand nine hundred of there were from Qilalukan, two hundred fifty 

from Mitimatalik, two hundred fifty from graves in surrounding Qilalukan, and four 

hundred from Button Point (Ibid.: 132). Also on northern Baffin Island is a site excavated 

by Guy Mary-Rousseliere, and known as Nunguvik. This is one of the largest Thule 

sites, containing fifty Thule houses as well as many Dorset ruins. Three houses were 



38 
excavated, but only a general discussion of the findings has been published (Mary-

Rousseliere 1979). 

Southern Baffin Island is the most easterly region being considered in this study, 

and four sites will be discussed. The first of these is known as Crystal II, excavated 

between July 17th and August 21st, 1948 by archaeologist Henry B. Collins. This site is 

located at the head of Frobisher Bay, and contains four houses, all 140-200 feet from the 

riverbank and twelve feet above sea level. Two of these houses had already been 

excavated by soldiers stationed at the nearby Air Base, but the remaining two were 

excavated by Collins, as well as the middens adjacent to the former two houses. A total 

of six hundred artifacts were recovered from this site (Collins 1950: 18-19), but 

descriptions are limited to those considered 'diagnostic' by Collins. 

During the summers of 1971-1973, Peter Schledermann conducted an intensive 

site survey in the area of Cumberland Sound, and mapped twenty-three Thule sites. Of 

these, three were chosen for excavation (Schledermann 1975: 15,34). The first of these is 

known as Niutang (MbDc-l) or simply site 'Q '. This site is located in a small valley on 

the eastern side of Kingnait Fjord, and was first recorded by Franz Boas in 1888. Sixteen 

houses were found here, nine of which were selected for excavation based on availability 

of undisturbed areas and location within the clusters thought to represent different 

components. A total of one hundred five artifacts were found (Ibid.: 53-55). The second 

site is known as Anarnitung or A-I (MbDj-l), and it is located on an island near Bon 

Accord. Fifteen houses were found here, but most had already been disturbed before 

excavation began. Excavation centred on a midden area (Ibid.: 66-68). Finally, site B-1 

(LlDj-l) is located to the south of Anamitung, and also contains fifteen houses. Three 

houses were tested, and four houses and a midden area were excavated, resulting in one 



39 
thousand two hundred fifty artifacts (Ibid.: 68-71). 

Chronology o/Sites 

Although radiocarbon dating is unreliable in the arctic, for reasons already 

discussed, each of the above sites has been tentatively placed within the three phase 

system of Thule chronology, based on the few carbon dates on substances other than sea 

mammal bone, and the commonly accepted (though problematic) harpoon head seriation. 

Of these sites, Ml on Cornwallis Island, Brooman Point on Bathurst Island, Nunguvik on 

northern Baffin Island, Crystal II on southern Baffin Island, Naujan in Repulse Bay, and 

Malerualik on King William Island have all been placed in the period referred to as 'Early 

Thule'. The sites that fall within the 'Classic Thule' period are the later occupations at 

Naujan, Qilalukan and Mitimatalik on northern Baffin Island, Kuk on Southampton 

Island, Levesque Harbour on Boothia Peninsula, and Learmonth on Somerset Island. 

Labelled as 'Modified Thule' are the later occupations at Kuk and Cumberland Sound 

(Park 1983: 8). 

It must be kept in mind that Thule chronology is far from being definite, and its 

problematic nature dictates that all Thule sites be investigated regardless of their 

placement within this chronology. Many sites were occupied continuously for the entire 

Thule sequence, and regional variation may be obscured by an imperfect chronology 

which has very little basis in absolute dating. Attributes that have been deemed 'early 

traits' may be shown to be a factor of space rather than time, and so they will all be 

considered in this analysis. 



40 

Selection of Artifacts 

An analysis of all artifact categories in Thule material culture would be beyond the 

scope of a Masters thesis, and so much care has been taken in the artifact selection 

process. Ideally, we will rule out those artifacts which result from minimal modification 

and whose form is strictly dictated by function. Some examples of such artifacts are awls, 

probes, and wedges. Also not included are those artifacts which show remarkable 

uniformity in appearance throughout the Arctic, and finally, those artifacts which vary 

both within and between sites to such a large extent that the differences must be regarded 

as idiosyncratic and void of cultural meaning in the larger sense. 

Upon consideration of the entire range of Thule artifacts, harpoon heads were the 

only category selected for analysis, based on their wide distribution across the study area, 

and their perceived potential to show regional variation, based on the criteria discussed 

above. For the purposes of this study, each harpoon head will be broken down into a 

finiate number of attributes or variables. The values for each variable will be recorded for 

each artifact, and these data will then be analysed at the site level to look for site-specific 

trends, and the local level to look for regional patterns, and finally, at the level of the 

entire study area. This final level of analysis will compare various regional trends to look 

for regional variation. 

Mathiassen's seminal study was not limited to a mere description of the 

thousands of artifacts recovered through the Fifth Thule Expedition, it also included the 

grouping of these artifacts into stylistic 'types' within each functional category 

(Mathias sen 1927b). These types are still commonly used in site reports to this day, 

with the unfortunate result that many archaeologists delve no further in their descriptions 



41 

of the artifacts!! (e.g. Collins 1951; Taylor 1972). In an attempt to avoid the bias 

inherent in Mathiassen's typology, this study will go beyond the categories provided by 

the commonly accepted typology, and artifacts will be analysed at the level of individual 

attributes. The assertion is that Mathiassen' s 'types' are too general to reveal the 

subtleties of regional variation. 

The time constraints imposed on the present study permit that only published 

sources be considered for analysis. An aideal analysis would include consideration of the 

many site reports that have never been published in any complete form. Many of the 

sources cited above were intended merely as preliminary reports of the findings from 

particular sites (e.g. Collins 1951), or were meant only to provide an overview of the 

general findings (e.g. Mary-Rousseliere 1979), and as such, individual artifacts are often 

not reported in any detail. This study will focus on those artifacts for which there are 

adequate and detailed descriptions, preferably accompanied by photographs of the 

artifacts under discussion. In the few rare instances where no descriptions of specific 

artifacts exist (e.g. Collins 1950, 1951, 1952; Taylor 1972) , the analysis will be based on 

observations I have made from the pictures alone. Any individual attributes not clearly 

indicated will be given the value 'unknown' in the analysis, rather than assuming these 

attributes based on the classification of the artifact into one of Mathiassen' s (192 7b) 

'types' . 

11 This refers particularly to Mathiassen's harpoon head typology, which divides Thule harpoon heads into 
five general 'types' (1927b: 12-13). 



IV 
Harpoon Heads: An Introduction 

Introduction 
Harpoon heads are perhaps the most studied class of artifact in arctic 

archaeology. The quest for the perfect harpoon head typology has had a long history, 

beginning with Mathiassen's 1927 scheme, and continuing to this day. Harpoon heads 

are ubiquitous in Thule sites, vary in regular ways, and the presence or absence of 

particular attributes has long been used to support theories of cultural affiliation or 

temporal phases: 

Harpoon heads appear in a multitude of forms and variations and, 
as they furthermore are among the most frequently occurring 
objects in older Eskimo finds, they are better than any other 
element of Eskimo culture for showing cultural connection and 
chronology in the Eskimo culture (Mathiassen 1927b: 11). 

For these reasons, the analysis of the distribution of harpoon head attributes across the 

arctic will playa vital role in this study. 

In historic times, harpoon heads were one of the essential tools to Inuit survival. 

The hunting of seals and other sea mammals seems to have played a vital role in the 

economies of both the Thule and historic Inuit people, and so it is not surprising that the 

harpoon heads used for this purpose are frequent archaeological finds. A harpoon head is 

composed of four main parts: a wooden shaft, a socket piece, a foreshaft, and a harpoon 

head. The head is attached to the shaft by way of a line running through a hole on the 

head to a tension piece on the shaft itself. The head is designed to separate from the shaft 

once it is embedded in the animal being hunted. Thule harpoon heads were made of a 

42 



43 

variety of organic materials, including bone, ivory and antler. All harpoon heads have a 

socket, by which the head is attached to the fore shaft of the harpoon. This socket can be 

either open or closed, and the open socketed forms normally have some sort of provision 

for lashing the harpoon head to the foreshaft, such as drilled lashing holes, lashing slots, or 

a sunken lashing bed. Some specimens have sharp points and/or lateral barbs, and were 

intended to be used as is, others have a slot in one end for the insertion of a separate 

blade. 

Harpoon Head Typology 

Therkel Mathiassen was the first archaeologist to create a typology of harpoon 

heads (1927b: 11-27), and this typology is still used in almost every Thule study to this 

day. Mathiassen established a 'type' system, based on his observation that harpoon 

heads seemed to vary in regular ways, with certain attributes consistently co-occurring. 

At its most basic level, his typology breaks down all harpoon heads into five 'types' 

based on the following attributes: position and alignment of the line hole; whether the 

shaft socket is open or closed; whether there is an inserted blade; and if so, whether the 

slot is aligned parallel or perpendicular to the line hole; and the number and position of 

barbs, and the number and shape of spurs. 

Thule Type I harpoon heads are defined as being very thin, with the line hole 

positioned straight through from one face to the other, one aslant dorsal spur, an open 

shaft socket, and no barbs or separately inserted blade (Mathiassen 1927a: 24). Thule 

Type II harpoon heads are described as being similar to Type I, but with two powerful, 

opposite barbs (Ibid.). Thule Type III harpoon heads are again very similar to Type I, 

but differ in having a slot for blade insertion positioned parallel to the plane of the line 



44 
hole (Ibid.: 25). Thule Type IV harpoon heads are described as being thin, with a line 

hole going direct from one face to the other, with a closed socket and a blade slot 

positioned perpendicular to the plane of the line hole (Ibid.). Finally, Thule Type V 

harpoon heads have a more rounded cross-section than the above types, have two dorsal 

spurs, no barbs, and a blade slit parallel to the plane of the line hole (Ibid.: 26). The Type 

V specimens also generally have a curved line hole, with both openings on one face of the 

specImen. 

For the purposes of his analytical volume (1927b), Mathiassen offered a more 

detailed typology, perhaps after recognising that the five-type system was not adequate 

to describe much of the variation he was observing. The following breakdown of harpoon 

head attributes can be used to describe almost any Thule harpoon head: 

A. Thin; the line hole runs directly from side to side. 
1. Open shaft socket, which is closed by a lashing; one 
slanting spur. 

a. Neither barb nor inserted blade. 
b. With barb; no inserted blade. 

1. Two opposite barbs. 
2. One barb. 
3. Several barbs. 

c. Without barbs; inserted blade. 
1. Blade parallel with line hole. 
2. Blade at right-angles to line hole. 

d. Both barbs and inserted blade. 
1. One barb, blade parallel to line hole. 
2. One barb, blade at right-angles to line hole. 
3. Two barbs, blade parallel to line hole. 
4. Two barbs, blade at right-angles to line 

hole. 
II. Closed shaft socket. One or two spurs. 

a. Neither barbs nor inserted blade. 
b. With barbs; no inserted blade. 
c. Without barbs; inserted blade. 

1. Blade parallel to line hole. 
2. Blade at right-angles to line hole. 



d. Both barbs and inserted blade. 
B. Flat; the line hole having a curved path, with both openings on 
the upper side. 

I. Open shaft socket. Mostly two spurs. 
a. Neither barbs nor inserted blades. 
b. With barbs; no inserted blade. 
c. Without barbs; inserted blade (parallel to line 

hole). 
d. Both barbs and inserted blade. 

II. Closed shaft socket. Mostly two dorsal spurs. 
a. Neither barbs nor inserted blade. 
b. With barbs; no inserted blade. 
c. Without barbs, with inserted blade. 

1. Almost round. 
2. Very flat. 

d. Both barbs and inserted blade. 
(Mathiassen 1927b: 12-13) 

45 

It is clear that some definitions are in order here, thoughtfully provided by 

Mathiassen himself. By 'thin' harpoon head, Mathiassen was referring to those 

specimens for which the greatest width was found perpendicular to the plane of the line 

hole. Conversely, by 'flat' he meant harpoon heads in which the greatest width was 

parallel to the plane of the line hole. The term 'spur' specifies a barb that is found at the 

proximal end of the artifact, as opposed to 'barbs' which are always found at the side of 

the artifact (Mathiassen 1927b: 12). Many other typologies of harpoon heads have been 

proposed (e.g. Collins 1937; Ford 1959; Holtved 1944; Wissler 1916), but most of these 

are based on Mathiassen' s categories, with accompanying elaboration on specific 

distinctive attributes (Schledermann 1975: 105). 

Thule Type V harpoon heads belong to the class of 'flat' harpoon heads, and are 

defined by Mathiassen as " ... [r ]ound or slightly flattened, so that the back is the natural 

resting surface; 2 dorsal spurs, no barbs; blade slit parallel with the line hole" (1927a: 26). 

This type, which Mathiassen observes to come from later site components (Ibid.: 27), is 



46 
comparatively rare in Thule sites. In fact, absolutely no specimens are reported from 

either the western or central regions of our study area. The strong presence of this type 

in the eastern region may be biased by the fact that the majority of the sites in this region 

were published by Mathiassen, who may have included many specimens of historic origin 

within his Thule types. Indeed, sixteen of the Type V specimens reported come from 

sites excavated by Mathiassen (1927a: PI. 2(3), 40(2,4,6,11), 67(2), 69(6), 72(5), Fig. 18). 

The remaining eleven specimens all come from Schledermann's Cumberland Sound sites 

(1975: PI. 3a-f,h,i; PI. 4a-h). 

Individual specimens belonging to the flat variety of harpoon head will not be 

discussed in this study because they seem to vary idiosyncratically, and because their 

inclusion would make it too difficult to control for chronology. This study will focus on 

Mathiassen's Types I through IV, but will not be limited by these categories in the 

descriptions of individual artifacts. If someone could effectively demonstrate that all of 

these specimens are prehistoric, or at least protohistoric, their restricted distribution 

could be very meaningful. Is it possible that what has been known as 'Thule Type V', 

based on Mathiassen's findings at Naujan could actually be a product of a regional social 

group centred in the eastern region of the central Canadian arctic? This question must be 

left for future research. 

Function 

Mathiassen's basic categories of harpoon head types remain the basis for most 

descriptions of Thule assemblages, but the question of whether these types reflect 

functional categories, chronological changes or regional styles has never been 

satiisfactorily answered. Mathiassen also considered the question of function based on 



47 

his ethnographic experiences with contemporary Inuit, surmising that the blunt point and 

large barbs of Type II would make it suitable for an animal whose skin was easily 

penetrated, but also subject to easy tearing, such as seals or white whales. The small size 

and sharp inserted blade of Type III would be suitable for use on an animal with strong 

hides, such as walrus (Mathiassen 1927a: 26). In his work with the Iglulik Inuit, 

Mathiassen observed that type AIIc2 harpoon heads were used for walrus, types BIIb 

and BIIc for narwhal and white whale, and type Alia for salmon fishing (Mathiassen 

1927b: 14). 

More recently, Robert Park discussed the weak arguments that have been made 

for the function of specific harpoon head styles, stating that archaeologists often ascribe 

function based on the size of the artifact because of the lack of related ethnographic 

information, citing as an example Allen P. McCartney's 1977 report, in which it is argued 

that Thule Types III and IV were used for seal, walrus, and beluga. Park notes that if 

these two types are specialised for the same function, as McCartney argues, then we 

would not expect them to co-occur as frequently as they do in Thule sites (Park 1983: 

171). Park points out the need to clarify the issue of function before these artifacts are 

used in chronological interpretations, to determine whether particular styles reflect 

functional differences, individual preferences, or chronological phases (Ibid.: 176). As 

individual preferences may be due to social group affiliation, this distinction is also 

important to this study, and so the idea of function and chronology as factors also causing 

variation will be considered, mostly in terms of the authors' own interpretations of the 

collections they are describing. 



1\ 
/ \ 
o 1.inehole 

o 0 ~slots>l."''€> 
opmsoc}(et 

Thule Type I 
(Ala) 

Thule Type ill 
(AIel) 

l.i:ne ho I.e 

AO 
10 

bM.e slot pmllil 
to pl.me of 1.ine 
hole 

1.ine hole 

.......... opm SOc}(et 

s:puI'-

Thule TWe IT (AThI) 

endbM.e--

P"PotIdkulm: ~ n 
bWhslot ~ I I 

I I 
I I 

1.ine hole __ ...:,..1, 0 l/lnmJ. wrist. 
I '. 
i \ 

opmsoc}(et 

Thule Type ill 
(Ak2) 

Figure 4-1: Hatpoon Head Types and Attributes. 



~ 
holes 

Clathan Open 
Socket Type 
(AIel) 

linehole ---+--0 

clo!=ed !=ocb:et 

Thule Type IV 
(Alli2) spur 

I 

\ l 

o 
-~=---- clo!=ed!=ocb:et 

Clathan Closed Socket 
Type (Alli2) 

Figure 4-2: llaJpoon Head 
Types and Attributes 



50 
Chronological Issues 

A number of harpoon head attributes have been used to derive a chronology of 

Thule sites. A classic example of this can be found in Peter Schlederrnann's 1975 report 

of his archaeological work around Cumberland Sound. Among the early Thule harpoon 

head attributes listed are lashing slots or sunken lashing grooves, sharply angled base 

spurs, incised Y-line decorations, and lateral ridges at the base (McCartney 1977: 226-

227; Morrison 1983: 84-87; Schlederrnann 1975: 241). Perhaps the most frequently cited 

chronological marker is the type of perforation used to facilitate lashing to the harpoon 

foreshaft - lashing slots are consistently seen as an earlier trait than drilled lashing holes 

(e.g. Collins 1937; Jenness 1928; Maxwell 1985; McCullough 1989; Morrison 1983). 

Robert Park tested this assumption in 1994 by seeing if sites dominated by harpoon 

heads with lashing slots were indeed occupied at an earlier period. He found that harpoon 

heads with drilled lashing holes were more often from sites with earlier radiocarbon dates 

than sites dominated by lashing slots (Park 1994: 

40). Park also tested the correlation between radiocarbon dates and some of the other 

commonly accepted early traits, to find that the