UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Women's access to power : a continuous area study Légaré, Evelyn I. 1986

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-UBC_1986_A8 L43.pdf [ 12.74MB ]
Metadata
JSON: 831-1.0097034.json
JSON-LD: 831-1.0097034-ld.json
RDF/XML (Pretty): 831-1.0097034-rdf.xml
RDF/JSON: 831-1.0097034-rdf.json
Turtle: 831-1.0097034-turtle.txt
N-Triples: 831-1.0097034-rdf-ntriples.txt
Original Record: 831-1.0097034-source.json
Full Text
831-1.0097034-fulltext.txt
Citation
831-1.0097034.ris

Full Text

WOMEN'S ACCESS TO POWER: A CONTINUOUS A R E A STUDY by E V E L Y N I. LEGARE B.A.CHons.), The University of Regina, 1979 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in THE FACULTY OF GRADUATE STUDIES Department of Anthropology and Sociology We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA September 18, 1986 © Evelyn I. Legare, 1986 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at The University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the Head of my Department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of Anthropology and Sociology The University of British Columbia 2075 Wesbrook Place Vancouver, Canada V6T 1W5 Date: September 18, 1986 ABSTRACT This thesis tests empirically three hypotheses advanced in the anthropological literature on women; differences in men's and women's access to power and autonomy in various spheres are associated with 1) the degree of stratification in a society, 2) the type of descent in a society, and 3) the final postmarital residence pattern in a society. I compare the access of men and women across thirty aspects of power - a multidimensional approach -rather than comparing the power of men and women - a unidimensional approach. The hypotheses are tested using ethnographic data from thirty-nine contiguous cultural units. I use a continuous area sample rather than a cross-cultural one because a continuous area sample reduces the number of confounding variables; contiguous cultures have less variety in terms of ecology, subsistence base, language, customs, and history. The results of the analysis indicate that differences in men's and women's access to power and autonomy are associated with each of the three independent variables: degree of stratification, type of descent, and final postmarital residence. The degree of stratification is the most important of these variables. The directions of the associations between women's access to power and the degree of stratification are not always those predicted in the literature, however. For some aspects of power, greater stratification is associated with relatively less access for women across cultures; for other aspects of power, greater stratification is associated with relatively greater access for women. Descent appears to be less important than degree of stratification for ii women's access to power; there are fewer significant relationships between the dependent variables and descent. Nevertheless, matrilineal descent is linked with equal or greater access for women in two areas. The results also demonstrate that matrilineal descent tends to dampen or work against the associations between degree of stratification and women's access to power in various spheres. In cultures with other (non-matrilineal) descent, the degree of stratification remains correlated with the dependent variables while in cultures with matrilineal descent, the relationships tend to be weaker or to disappear. The findings show that the final postmarital residence pattern in a society is associated with certain dependent variables although it also is less important than the degree of stratification. Matrilocal residence is associated with women's relatively greater access. In the literature, avunculocal residence is predicted to have a negative effect on women's power. The results demonstrate that in some cases, avunculocal residence is positively associated with women's power. From these results, I conclude that women's access to power does indeed vary across cultures; women are not invariably barred from wielding power in the community or kin group. Thus, this study indicates that there are differences in women's and men's access to power and autonomy, and this variation across cultures is related to the three independent variables tested in the thesis. Some of the relationships found are in accord with those postulated in the anthropological literature; others are not. In fact, some surprising results emerge from this analysis. iii T A B L E OF CONTENTS ABSTRACT ii ACKNOWLEDGEMENTS viii I. INTRODUCTION 2 A. INTRODUCTION 2 B. THE STATUS OF WOMEN IN THE LITERATURE 5 C. THE DEPENDENT VARIABLES 11 1. The Non-Scalar and Scalar Usages of Status 11 2. Access to Power and Autonomy 25 3. The Thirty Dependent Variables 40 II. THE INDEPENDENT VARIABLES 63 A. INTRODUCTION 63 B. DEGREE OF STRATIFICATION 64 1. A Review of the Models 65 2. The Independent Variable 82 3. The Hypothesis 85 C. DESCENT AND RESIDENCE 85 1. A Review of the Models 85 2. The Independent Variables 101 3. The Hypotheses 109 III. THE DATA I l l A. Dealing with the Limitations of the Data 134 IV. THE CULTURAL UNITS 136 A. Continuous Area Studies 136 B. The Appropriate Cultural Unit 139 C. The List of Cultural Units 145 V. THE STATISTICAL METHOD 149 A. The Statistical Approach 149 B. The Statistical Technique 151 C. Interactive Effects 157 VI. PRELIMINARY ANALYSIS 160 A. The Dependent Variables 160 1. The Distributions 160 2. Factors Influencing the Distribution of Dependent Variables ... 162 3. Non-existent Role or Institution 168 B. The Independent Variables 170 VII. THE STATISTICAL ANALYSIS 175 A. THE ANALYSIS 175 1. Interactive Effects 177 2. Univariate and Multivariate Associations 180 iv a. RELATIONSHIPS WITH STRATIFICATION 183 b. LEADR 184 c. MAN.KIN.CH ATT and MAN.KIN.PROP 186 d. COMMUN.INFL 187 e. CO-OP.FRTS 191 f. W.FATE.AFT.MAR 193 g. Relationships With Stratification - Conclusions 196 h. RELATIONSHIPS WITH DESCENT 198 i. OWN.OWN.FRTS 198 j. MAR.ARRANG 200 k. Relationships With Descent - Conclusions 201 I. RELATIONSHIPS WITH RESIDENCE 202 m. OWN.OWN.FRTS 203 n. MAR.ARRANG 205 o. W.FATE.AFT.MAR 206 p. Relationships With Residence - Conclusions 207 q. SUBSIST.LEADR 209 B. CONCLUSION 211 VIII. CONCLUSIONS 220 IX. APPENDIX I: THE EFFECT OF CULTURE AREA 230 X. BIBLIOGRAPHY 235 LIST OF TABLES TABLE I: FREQUENCIES OF GENDER ACCESS TO POWER AND AUTONOMY 161 TABLE II: FREQUENCIES FOR STRATIFICATION, DESCENT, AND RESIDENCE 171 TABLE III: CORRELATIONS AMONG THE INDEPENDENT VARIABLES ..172 TABLE IV: RANGE OF VALUES OF KENDALL'S TAU FOR THE RELATIONSHIPS BETWEEN THE DEPENDENT AND INDEPENDENT VARIABLES 176 TABLE V: KENDALL'S TAU VALUES FOR THE RELATIONSHIPS BETWEEN THE DEPENDENT AND INDEPENDENT VARIABLES AND SIGNIFICANCE LEVELS OF THESE RELATIONSHIPS 181 TABLE VI: SUMMARY OF THE STRENGTH OF EACH RELATIONSHIP AND ITS STATISTICAL SIGNIFICANCE 182 vi LIST OF FIGURES FIGURE 1: NATIVE CULTURAL UNITS INCLUDED IN THE CONTINUOUS AREA STUDY 1 vii ACKNOWLEDGEMENTS I would like to acknowledge the continued support, advice, and guidance of my committee members: Dr. David F. Aberle, Dr. Michael Kew, and Dr. Margaret Stott. I extend my special thanks to Dr. David F. Aberle, not only for his guidance and encouragement, but also for the example he has set in terms of scholarship as well as for his wit, which made even his criticisms a source of enjoyment. I extend my gratitude to Dr. Neil Guppy, who taught a numerically illiterate writer that numbers can also be a source of knowledge, and who consistently listened to any and all statistical queries. I would like to thank Frank Flynn for producing the map of cultural groups and Sherry Gee for typing the initial draft of this thesis. I sincerely acknowledge my debt to fellow-students Jo-Anne Fiske, Mary Ann Tisdale, Ann Stevenson, Joanne Richardson, and Anne Underhill, who listened, gave critical support, and laughed when appropriate. I extend my gratitude to friends who encouraged me throughout the writing of this thesis: Sue, Tim, Tom, Cassie, Paul, Laura, Peter, Howard, Lome, Helene, and the women from W.E.S.T. - Alice, Ray, Gay, Barb, Jennifer, Kathleen, Sandra, and Sandy. I extend special thanks to Cheryl Exner, not only for her friendship, but also for her reading of the manuscript. Finally, my loving thanks to Greg Schwann who provided many services including map drawing, proof reading, software programming, sounding board, and faith that this study could be accomplished. vm F i g u r e 1 N a t i v e C u l t u r a l Unuts I n c l u d e d un t h e C o n t i n u o u s R r e a S t u d y L J a ZD e-E— CE 170 N 160 W 150 W 140 W 130 W 1 2 0 W 1 1 0 W 100 W 9 0 W LONGITUDE I. INTRODUCTION A. INTRODUCTION An enormous literature has developed in Anthropology within the past twelve to fifteen years which focuses on women and sexual equality and inequality. This literature includes ethnographic studies as well as theoretical and ethnological studies. Many of the ethnographic studies attempt to present a more complete picture than past ethnographies which tend to neglect women and misconstrue their roles. The theoretical and ethnological studies are concerned with topics such as comparing women's status in a small number of cultures and the processes which give rise to and perpetuate sexual equality/inequality.1 While this literature on the anthropology of women continues to expand, even a cursory examination indicates little consensus or synthesis. Rogers (1978:124) states that "a deluge of disparate studies has appeared, but little effort has been made to relate them to one another, to isolate recurrent issues and concerns for closer examination, or to make explicit the various assumptions informing them". Whyte (1978:3) notes that the debates over the exact status of women, the origins of women's status, and variation in status over time has led to "very little consensus". There are a number of reasons for the lack of agreement. One is disagreement regarding the definition and measurement of women's status. Another is a basic disagreement over women's status cross-culturally; does it 1 The subject matter of this thesis does not include the literature on women's status in modern societies (e.g., the effects of industrialization on women's status, Marxist feminist literature on the status of women in the present, or the effects of modern economic development). 2 3 vary at all or is it invariably low? A third reason is the practice of proposing hypotheses on the basis of evidence from two or three cultures without further attempts to test these hypotheses using large data sets. Only after a hypothesis has been repeatedly tested using a number of societies, can one accept or reject the hypothesis. As Jorgensen explains, The many theories that have been offered to account for various facets of Indian culture in western North America . . . have practically always been advanced to explain something about one or two tribes, and have begged for explicit tests. Formal comparisons among many tribes and many variables, to determine whether these theories could withstand scrutiny over a wide spectrum of data pertaining to a large group of societies, seemed necessary. It may be the case that many of the theories that have been advanced over the years seem plausible because their authors have marshalled only a small amount of data, selecting some and rejecting others, on one tribe so as to illustrate preconceived impressions (1980:6-8). Yet anthropologists continue to advance (and reject) hypotheses on the basis of data from a few societies.1 The result is a proliferation of competing hypotheses, each hotly supported by its author. These untested hypotheses rest mainly on plausibility, and, as Whyte notes, "in this day and age one man's plausibility is often another man's (or woman's) incredulity" (1978:6). Two areas of concern must be addressed if there is to be some consensus in the anthropological literature on women. First, the concept of women's status must be clearly defined and measured. Thus far, there is little agreement on how to define or measure the position of women, either within a culture or across cultures. Nevertheless, if the definitions used are 1 The studies by Clignet 1970, Friedl 1975, Martin and Voorhies 1975, Nielsen 1978, Sacks 1979, Sanday 1973 and 1974, and Schlegel 1972 are attempts to do larger comparative studies; Whyte 1978 is one of the few large scale statistical studies of women's status done to date. (Murdock 1965a is another.) 4 clearly enunciated, there will at least be some basis for comparison and analysis among the studies. Second, there must be systematic, large-scale comparative studies of gender relations and women's status which test the hypotheses advanced in the anthropological literature. In this way, researchers can begin to choose among the many competing hypotheses - an essential step in the process of theory building. This thesis attempts two things. First, I define a measure which allows an examination of the differential access of men and women to power and autonomy. Second, I formally test three hypotheses which recur in the literature. The research question is whether there is an association between gender differences in access to power and autonomy and one or all of the following: a culture's degree of stratification, type of descent system, or final postmarital residence pattern.1 I use a statistical model to examine .the relationship between thirty dependent variables which measure the access of women and men to power and autonomy in a culture and three independent variables: degree of social stratification, type of descent system, and final postmarital residence. Data from thirty-nine cultural groups located in a continuous geographical area are used to determine the presence, strength, and direction of relationships between the dependent and independent variables. In the remainder of this chapter, I consider the dependent variables. I review some of the concepts used to define and measure women's status in the anthropological literature and the problems associated with these concepts. Subsequently, I define and describe my dependent variables. In addition, I 1 In determining access to power and autonomy, I exclude those men and women who are not considered members of the community or local group, nor do I include slaves (i.e., outsiders). 5 discuss some of the problems encountered in obtaining data on these dependent variables. In Chapter 2, I discuss and define the independent variables. I set forth the hypotheses to be tested. Chapter 3 is a critique of the ethnographic data used in this thesis. In addition, I discuss some of the problems a researcher encounters in using these data. Chapter 4 delimits the geographic area covered in this thesis. In Chapter 5, I describe the statistics used in the study. Results are presented in Chapters 6 and 7 with the distributions of the dependent and independent variables as the topic of Chapter 6 and the results of the statistical analysis the topic of Chapter 7. Chapter 8 summarizes the results and implications of this study. B. THE STATUS OF WOMEN IN THE LITERATURE The manner in which women's status is defined contributes to the lack of consensus in the anthropological literature on women (see Rogers 1978, Whyte 1978). For example, both Cohen and Leacock (1978:258) agree that women in band societies had "a great deal of autonomy and even power". Nevertheless, Cohen asserts that only men had authority (defined as the rec-ognized right to allocate scarce resources) in these societies. Thus, he concludes that there is "male dominance" in band level societies. Leacock, on the other hand, contends these same societies had sexual equality because women were autonomous - defined as having decision-making power over one's own life and activities to the same extent as men (1978:247). The difference in their conclusions results from each author's definition and measurement of women's status. 6 The manner in which an author defines status can have a profound ef-fect on findings, as in the case above. However, the focus on the causes of inequality often makes defining women's status a secondary consideration for the researcher. In some cases women's status is loosely defined. Authors speak simply of "women's status" or "the position of women" although implicitly this variable is often equated with power (Rogers 1978:140). Even when the dependent variable is defined, its use may become less precise as the author continues. As Quinn (1977:182) points out, though writers may begin by carefully defining what they take to mean 'women's status', in terms of female political participation, economic control, personal autonomy, interpersonal equality,, legal adulthood, ideological position, or other specific indices, for the purpose of testing a particular hypothesis, they often lapse into much looser usage in nearing their conclusions, and speak broadly of 'women's low status', 'female subjugation', 'female oppression', or 'male dominance'.1 Such imprecision can lead to confusion. In other cases, researchers have chosen questionable indicators of women's status.2 For example, Young and Bacdayan (1965:230) treat the presence of menstrual taboos in a society as indicators of low status, arguing that menstrual taboos are institutionalized ways in which males in primitive society discriminate against females. Bock (1967:215), Briggs (1974:294, Leacock (1978:247), Quinn (1977:215-225), and Whyte (1978:8), on the other hand, argue that the presence of menstrual taboos need not indicate any such thing. Another commonly used indicator of women's low status is segregation of the sexes (see Whyte 1978:32). Rogers 1 See Cohen in the example above. 2 The indicators chosen by a researcher are affected, of course, by the researcher's biases and assumptions about women and their roles and status in society (see Leacock 1978, Sacks 1979). A discussion of these biases and assumptions is presented at a later point in Chapter 1 and in Chapter 3. 7 asserts, however, that such segregation need not indicate low status since the barring of one sex from the domain of the other is not necessarily a nega-tive thing, nor can it be assumed that one group is more excluded than the other if neither has access to the other's domain (1978:145). Similarly, polygyny is sometimes treated as an indicator of low status for women. Authors such as Martin and Voorhies (1975:241-46) and Whyte (1978:45) argue, on the other hand, that the presence of polygyny may be a source of power for women, as it lightens their burdens while increasing the responsibilities of the husband. Indeed, Honigmann (1949:86) reports that this view is held among the Kaska. Another indicator of low status for women sometimes used is the negative evaluation of women by men in a society. Other authors contend, however, that the presence of ideologies of male superiority held by men may actually reflect the high status of women because this male hostility is based on fear of women's power (e.g., Briggs 1974:294, Dahlberg 1981:27, Quinn 1979:218-19, and Tiffany 1979:19-22). There is obviously room for argument about the meaning of such indicators of women's status. Another common practice is to use a narrow range of indicators (or one indicator) to measure status. Since there is disagreement about how to define status and what indicates high or low status, a researcher may be neglecting other important aspects of women's status when he or she uses only one or two measures. For example, Cohen (1978:258) and Lamphere (1977:626) both claim that formal public authority is the only valid measure of status. Netting (1979:166), on the other hand, points out the perils of assessing status on the basis of formal public behavior. He uses the example of 8 Kofyar women, who have no formal political office and cannot participate in many religious rituals. Nevertheless, Kofyar women control the houses and crops as well as their own labour force participation. Can one conclude that these women have "low status"? The disagreement of Leacock and Cohen described earlier illustrates how a narrow range of indicators can lead to conflicting results and, hence, to disagreement among researchers (see Rogers 1978:145, Sacks 1979:72, and Whyte 1978 for further discussion of this point). Using a narrow range of indicators may be inappropriate for another reason. In the past, "women's status" has been treated as a unitary construct. Measures of status are treated as "so many related and covarying symptoms" (Quinn 1977:182) which are equivalent indicators of status. Evidence suggests, however, that it might be more accurate to treat women's status as a composite of many variables which are often causally independent of each other (Quinn 1977:183, Whyte 1978:116). That is, in a given society, women may have low status in some domains, approach equality in others, and in some domains surpass the status of men (e.g., see Netting's example above). Lowie made this point in the early part of this century, 'First of all, it should be noted that the treatment of woman is one thing, her legal status another, while the character and extent of her labors belong again to a distinct category. Whatever correlations exist between any two of these aspects are empirical; conceptually they are diverse, and only confusion can result from ignoring the fact' (1920:187 in Whyte 1978:95). If women's status is a multidimensional phenomenon, then to compare women's status cross-culturally on the basis of a narrow range of indicators may be conceptually unsound. 9 Whyte examines this question in great detail, asking, "do various aspects of the status of women vary quite independently in our sample, so that knowing, for example, that women hold political offices in a particular culture will not allow us to predict anything about their economic rights or their role in religious affairs?" (1978:96). His findings indicate that it is inappropriate to ask in which culture the status of women is higher; We can compare the domestic authority of women relative to men in the two cultures, or their subsistence contribution or property rights, but the total picture is a series of pluses and minuses that are different in each culture and that cannot be summed up easily to form an overall measure of women's status (1978:116). Whyte concludes that because of this, it does not make sense to compare women's status cross-culturalfy; only individual aspects of status can be compared (1978:117). Whyte's findings have another interesting implication; as he points out, "we cannot say which aspect is the better indicator of status of women since the existence of such a general phenomenon is doubtful" (1978:117). If women's status is a multidimensional rather than a unidimensional phenomenon, as these authors argue, then the status of women may vary within the same culture, depending upon the indicators of status that are used (e.g., see Sacks 1979:72, Schlegel 1977a: 17, Tiffany 1979:3-4). This adds to the lack of consensus in the literature. Whyte maintains "we can at least gain some understanding of why debates about 'the status of women' have generated so much heat and so little light in recent years" (Whyte 1978:117), given the suspicion that women's status is not a unidimensional phenomenon. 10 I believe that if some agreement and synthesis of the literature are to be achieved, researchers must, explicitly and carefully define their dependent variable(s). With precise definitions, studies can be compared and contrasted and new insights gained. A broad range of indicators ought to be used if women's status is to be accurately determined and understood. This is true whether one asserts that women's status is a unitary construct or a multidimensional phenomenon. By using a narrow range of indicators, the researcher runs the risk of choosing inappropriate or limited indicators. Where statistical paradigms are used, there is a further requirement. The dependent variable or variables must be clearly defined in such a way that the necessary data, gathered from a large number of cultures, can be readily identified and coded; that is, the dependent variable(s) must be empirically measurable. The dependent variables chosen for this thesis are empirically measurable. Like many others (e.g., Clignet 1970, Friedl 1975, Giele 1977, Gough 1975, Lamphere 1977, Lips 1977, O'Barr 1976, Rogers 1978, Sanday 1973, 1974, 1978, Schlegel 1972, 1977a), I am particularly interested in one dimension of women's status - power, and how power is distributed between men and women in a society. My interest stems from a prevalent assumption in the literature that women never hold legitimate authority, political power (e.g., Friedl 1975, Nielsen 1978:13, Sanday 1973, 1974, Schlegel 1972:6), or the right to allocate scarce resources (Cohen 1978:25s).1 In other words, many authors do not 1 Schneider (1961:7) defines "the male sex role as having authority over the statuses occupied by women" in the two spheres in which he is interested - the descent group and the domestic sphere. 11 concede that women's status varies. Rather, they argue, women always have low status. Because of such claims, I am interested in learning how power is distributed between men and women. If it is true that women never have authority, never hold public power, and never allocate resources, there will be no variation in my dependent variables. Women will have less access than men across all the categories of my dependent variables. C. T H E DEPENDENT VARIABLES 1. The Non-Scalar and Scalar Usages of Status In order to define the dependent variables used in this thesis, I return to the basic notion of status and how it has been defined in the literature. According to an article by Zeldifcch in the International Encyclopedia of the Social Sciences (1968:250-257), status has two usages, a non-scalar and a scalar usage. The non-scalar usage refers to the rights and duties of a posi-tion in a social system. It is also called "status in the Linton sense", after the social anthropologist who first defined it. The scalar usage, on the other hand, refers to evaluations of superiority/inferiority or a hierarchical ordering of people or positions on some basis such as wealth, power, prestige, or respect (1968:250). It is this scalar usage which is commonly used by social scien-tists in measuring and evaluating social and sexual inequality (i.e., women's status). Linton (1971:291) defines status as a collection of rights and duties. Statuses, therefore, are polar positions in patterns of reciprocal behavior. While a status is a collection of rights and duties, a role "represents the 12 dynamic aspect of status, the putting into effect of its rights and duties" (Goodenough 1971:310). "Status" defines who a person is, for example, a doctor, a child, an American, while "role" defines what such a person is ex-pected to do (Zelditch 1968:251). Goodenough (1971:310) points out that a role, together with its status, is often confusingly called a "status" or "posi-tion". Goodenough labels a role and its associated status (rights and duties) a social identity. In the remainder of this thesis, the terms status, role, and social identity will follow Goodenough's usage. Two points must be made regarding the non-scalar usage of status. First, Goodenough suggests that the concept of "relationship" is essential to an understanding of status (1971:310). He argues that a role can be defined only in pairs - that is, in relationship to an alter. Since statuses are polar positions in patterns of reciprocal behavior, ego's rights are alter's duties to ego, while ego's duties are alter's rights in the relationship. It can be seen that the rights and duties involved in a particular relationship (i.e., the pat-tern of reciprocal behavior) depend not only on ego's role but on alter's role. For example, if ego is a doctor, the rights and duties involved in the relationship will differ, depending on whether alter is a patient, a nurse, an-other doctor, or the hospital administrator. A knowledge of both roles is nec-essary if one is to deduce ego's rights and duties. Therefore, when statuses are being considered, the failure to discuss relationships is like trying to clap with only one hand.1 Thus, in this thesis, when I focus on one role of a pair, the other role is included implicitly. Second, there are certain "broad role dispositions" which may be relevant 1 I would like to thank Dr. Aberle for clarifying this point. 13 to the roles a person may fill and, hence, to the relationships between alter and ego. The term "broad role dispositions" is used by Eisenstadt (1956:22-29). He notes that both age and sex are of importance for defining one's roles, activities, and expectations vis-a-vis others (1956:25). This is because age and sex define one's qualifications and one's obligations at any given stage of life. Age and sex have some influence on the roles which are allocated to various people. Eisenstadt cautions that age or sex is not a prescription or expectation of a detailed role but, rather, of a general, basic role disposition into which specific roles may be built and to which they may be ascribed (1956:22). Broad role dispositions such as sex and age, therefore, may affect the roles which are available to people, the expectations held about these roles and therefore the statuses associated with these roles. One word of caution must be interjected at this point. When status is under consideration, gender may not be the only factor which affects the roles a person may hold. That is, sex1 is not necessarily the only broad role disposition which influences access to specific roles, yet it is often assumed to be the delimiting factor in social relationships and in access to roles. For example, in describing Tlingit culture, Oberg (1973:85-86) explains that al-though women wove the sacred clan emblems into baskets and blankets, they were not supposed to understand the meaning of the sacred crests. He accepts this as evidence that Tlingit women are excluded from the ceremonial sphere (and concludes women have an inferior status in Tlingit society). In other words, he assumes gender is the factor which causes the "exclusion 1 "Sex" refers to the biological distinctions between male and female, while "gender" refers to the culturally learned behavior which is appropriate to each sex (Oakley 1972). from the ceremonial sphere". Nevertheless, de Laguna (1983:76) and Olsen (1967:38, 44) were told by Tlingit informants that only clan members are supposed to know and understand the meaning of clan property. Because of the marriage pattern among the Tlingit, a woman will marry a non-clan member. Hence, she will be using her husband's crest symbols to manufacture ceremonial articles and objects for the use of her husband and his kin. Her lack of knowledge might be explained, then, not as the exclusion of women from the ceremonial life of the Tlingit, but as a factor of a woman's kin group membership.1 In her work on sexual stratification, Schlegel (1977a:20) addresses this point. She notes that there are factors besides one's sex, which affect the status of men and women (e.g., age and social status). She observes, for example, that in some societies gender differences are muted as people age and that the location of a woman in the social hierarchy of a ranked or stratified society may affect her access to positions of authority or to power. Another author, Smock, (1977:395) cautions that although some sex-typing occurs in all societies, the importance of sex in comparison to age, social class, or inheritance as a basic ordering category will vary from culture to culture. In determining status, then, the researcher must examine the data thoroughly. Having reviewed the non-scalar usage of status, I will consider this defi-nition of status as it pertains to women. According to the non-scalar usage 1 It is curious that this type of reasoning is not extended to men. For example, the carvers of a totem pole were not of the kin group which commissioned the pole. Hence, it can be surmised that they were not supposed to understand the meaning of the crests they were carving; yet nowhere does Oberg mention that men are excluded from the ceremonial sphere! 15 of status, women's roles are those roles performed by (and expected of) those people who are defined as women in the society. It should be noted that these roles include not only the roles restricted to women, but also those roles that are mainly filled by women and those roles that are equally open to men and women. There is a tendency in Western European societies, however, to define women's roles only in terms of those roles which are restricted to women. This has created a number of problems in the ethnographic literature. Because women in Western European societies are perceived primarily as child bearers and child rearers (i.e., those roles restricted to women),1 dependent on males economically and socially, this perception is projected by anthropologists onto other cultures so that women everywhere are defined in the same manner (e.g., Ortner 1974, Rosaldo 1974, Schneider 1961:7-8). For this reason, most discussions of women's roles in the ethnographic literature are limited to topics such as puberty, marriage, birth, and child rearing. In attempting to synthesize these data, and make cross-cultural generalizations about women, one is led to the conclusion that women everywhere do little but marry men and run (or get run by) households. It is unclear, however, how much information on nondomestic aspects of female roles has gone undiscovered or unreported because anthropologists have failed to ask the proper questions" (Rogers 1978:145). Rogers provides an example. While working in Ashanti, Rattray (1955) was surprised when he finally learned that women were important in the formal state hierarchy of precolonial Ashanti. When he asked why he had not been told, the response was "The white man never asked us this; you have 1 See Rogers 1978:134-37; Sacks 1979, Sutton et al. 1975:596. 16 dealings with and recognize only men, we supposed the Europeans considered women of no account, and we know you do not recognize them as we have always done" (1978:145). Sacks (1979:72) analyzes this ethnocentric perception in detail. She notes that anthropologists have an ethnocentric view of "woman's position" as wife/mother, arguing that it is dangerous to take this one aspect of "woman's place" and confuse it with a "total notion of status" (i.e., the sum total of all the roles a woman can hold in a culture). Anthropologists should consider not only those roles which are restricted to women, but also those which are filled mainly by women and those which are available equally to men and women. I will now consider the scalar usage of status, and some of the prob-lems with using this scalar definition of women's status. According to Zelditch (1968:250), the scalar usage of status refers to the evaluation of a role. "Status" is a summed value or rank placed on a role in a society. The evaluation is made by other people, and the criterion for the evaluation may be power, deference, income, occupation, honour, or some other criterion. It is this usage which is commonly meant when anthropologists speak of women's low status or high status. Sociologists and anthropologists interested in sexual inequality often utilize three criteria to evaluate roles: material rewards (privilege), prestige (deference), and power (see Jeffries and Ransford 1980, Lenski 1966; Nielsen 1978; Runciman 1968, 1969; Schlegel 1977a). If the power, prestige and/or material rewards associated with the roles held by women are less than those rewards associated with the roles held by man, women are said to have lower status. 17 In other words, there is sexual (or social) inequality present when power, prestige, and material rewards are unequally distributed. As Schlegel (1977a:3) explains, when there are sets of statuses and roles that are characteristically accompanied by greater rewards, prestige, or power than others, and when movement between one set into another is restricted, inequality exists . . . When such systematic social inequality differentiates men and women, the condition is one of sexual inequality. Determining status is more difficult than it might seem. For anthropologists, there are two major problems. First, there is the problem of how to measure status; this problem is especially crucial where empirical studies are concerned. Second, there is the problem of evaluating status. This evaluation is made by the researcher, who has his or her own cultural frame of reference. A number of the anthropologists interested in the anthropology of women argue that the researcher's cultural frame of reference colours his or her perceptions and the manner in which he or she interprets the data. In addition, if the researcher uses informants, the data are likely to come from male informants (e.g., see Chapter 3). Certainly, this is true of the ethnographic data recorded in the past. The implications of these facts will be considered in turn. First, there is the task of measuring "status". How is this to be accomplished? For instance, if one wanted to compare the status of men and women on the basis of all their roles, one would have to take all the roles available to each sex and provide a summary measure for each of the rights associated with them (e.g., one summary measure for power, one for prestige, and one for material rewards). Then the summary measures for power, prestige, and material rewards would be aggregated into one summary measure or overall status measure for each gender. The problem, of course, would be empirically to translate power, prestige, and material rewards into equivalent concepts so that one could arrive at an overall measure of women's status and men's status (see Smelser and Lipset 1966:6-7 for further discussion of this problem). For this reason, most researchers interested in the status of men and women will choose only one of the criteria associated with men's and women's roles (i.e., power, prestige, or material wealth). Women's status and men's status are then compared on the basis of this one criterion. Unfortunately, even this less complex procedure is not easily accomplished. There is growing evidence that the rights associated with a role - wealth, prestige, and power. - need not be related; that. is, they may vary independently for each role (see Nielsen 1978:13, O'Barr 1976, Runciman 1968, and Sanday 1974). If this is so, the evaluation of women's status will differ, depending on which criterion is chosen to measure status (Runciman 1968 and Shils 1968 discuss some of the empirical problems involved in accomplishing this task). Another common solution is to choose a narrow range of roles, or even one role, and to use these to evaluate the status of women and men. In order to use this method, the researcher must assume: 1) the rewards associated with all roles are the same so that all roles are equivalent indicators of status, or 2) the rewards associated with the chosen role(s) are more appropriate measures of status than are those associated with other roles. (An example of this is Cohen's claim that although women in band 19 level societies have power and autonomy, they cannot allocate scarce resources, and, therefore, men are dominant.) Some of the problems encountered with this approach were mentioned at the beginning of this chapter. At this point, the problems are discussed in greater depth. First, on what basis is the ethnographer choosing the indicator? Second, do the author's assumptions hold cross-culturally? Finally, as discussed earlier, if women's status is not a unitary phenomenon, how accurate are results based on one or two indicators? These questions bring us to a second problem in using the scalar defini-tion of status. The anthropological researcher is choosing indicators of status, and evaluating status, on the basis of assumptions which may not be appro-priate to the culture(s) he or she is considering. These assumptions are the result of the researcher's Western values and biases, and they will have an important consequence on the data collected and on findings1 (see Rohrlich-Leavitt 1975:621). A number of these assumptions and their implications for evaluating women's status are considered here; others will be dealt with in Chapter 3. I have already mentioned that ethnographers often assume the only roles filled by women are those roles restricted to women (i.e., child bearing, child rearing, domestic activities). This assumption has resulted in a dearth of in-formation about many of the roles and activities of women in other cultures, making it difficult to compare the status of men and women. For example, 1 Driver (1974a:86), speaking of the culture element lists collected in the middle of this century, admits, "Within the limits of the qariation [sic] among ethnographers and informants in this study, the personal equation of the ethnographer seems to be as important a determinant of reliability as that of the native informant." 20 Mitchell (1976:96), in her thesis on the Coast Salish, states she had to infer the pre-contact status of Coast Salish women because of a lack of informa-tion. Researchers may assume the behavior they observe in another culture has the same meaning as it does in their own (i.e., certain behavior and roles have the same meaning and evaluation cross-culturally). Danger exists in making such an assumption. For instance, among the Ingalik, women eat after their husbands. At first glance, such behavior may seem to indicate women's servitude to their husbands. The meaning of this behavior is quite different, however. Osgood (1958:196) explains that Ingalik women eat after their husbands because women are "chief of the food"; as such, they have a responsibility to oversee the eating of food by others before they eat them-selves. Thus, a responsibility associated with an important role can be incor-rectly evaluated as a sign of inferior status. A second example comes from Raunet (1984:87) who mentions that Victoria Longarm was the most important chief in her village. According to Nishga protocol, such an important chief could not negotiate with Government commissioners. Consequently, she was not present during land negotiations but sent a lesser chief to speak for her - her husband in this case. An ethnographer, not conversant with Nishga norms, could easily misinterpret such behavior as proof of women's exclusion from public affairs. A final example illustrates the manner in which women's status has been incorrectly evaluated in the ethnographic literature when the ethnographer assumed that behavior has the same meaning in other cultures as it does in his or her own. Hearne (1968:129-30) assumes that Northern Athapaskan 21 speaking women have an inferior position because they are treated in ways that would be "humiliating to a European woman" (i.e., they work hard, and they do not eat at the same time as their husbands). Referring to the fact that Kutchin women eat after their husbands, Osgood argues that this behav-ior should not be treated as evidence of male superiority and female inferiority. He points out that the woman has been working around the camp, eating whenever she is hungry, while the husband has been out roaming the country for food, usually without eating. She "considers it natural to prepare food for her husband's return and might as well be described as taking joy in merely watching a hungry person eat as in suffering a feeling of inferiority because she does not keep pace by mouthfuls like people at a dowager's dinner" (Osgood 1958:113). Thus, if women's status is evaluated on the basis of the assumption that roles and behavior have the same meaning cross-culturally, some of these evaluations may be invalid (see Leacock 1978, 1981, Leacock and Nash 1981, Sacks 1978). A third assumption which has major consequences for the manner in which women's status is evaluated is androcentrism. "Androcentrism refers to a tendency to explain social, cultural, and physical phenomena in terms of men's activities" (Nielsen 1978:99).1 In our own society, the male sex, male roles, and male activities are the "standard" or "norm" used for describing 1 For example, we are told that, in terms of human evolution, hunting, always carried out by males, was essential; "the demands of the hunt shaped the char-acter that makes us human" (Dahlberg 1981:1). Research suggests, however, that this assumption is incorrect; both men and women in foraging societies hunt (it is assumed by both groups that the state of affairs in the past can be deter-mined by examining modern hunting and gathering societies). Also, gathering, not hunting, provides the majority of calories in hunting and gathering societies (Dahlberg 1981:2, Martin and. Voorhies 1975:175-177, 180-182, Rohrlich-Leavitt 1975:621). Thus, one could argue that the demands of gathering shaped the character that makes us human. our culture. Authors interested in the anthropological literature on women attest that this androcentric bias has been projected onto the cultures researched by anthropologists (see Leacock 1975:601, Ridington 1983:68-69, Rogers 1978:143-44, 146-47, Rohrlich-Leavitt et. al. 1979:117-19, Schlegel 1977a:2, Schrijvers 1979:98-101, and Tiffany 1979:15-18). As a consequence, women are viewed as peripheral. These assumptions have a number of results. Women are defined as passive objects who are manipulated by men, rather than as social actors. For example, Sapir and Swadesh (1955:325) translate Nootka marriage as "woman purchase", yet on page 278 of the same work, they record a marriage speech which asserts that men marry women so they will have a wife to guide them and give them advice. A second example of androcentrism is provided by Oberg (1973:32). In describing the lines of inheritance for Tlingit males, he claims "after property, privileges, honors, and wives passed from one uncle to another, they finally came to the oldest nephew." As a last example, Usher (1974:54) reports that the Tsimshian Indians came to trade and prostitute their women. Other results of androcentrism include the ignoring of women and the devaluing of women and anything associated with women (i.e., women's roles, women's values, women's activities) because they differ from the male norm. Often, when women are mentioned, they are denigrated. The ethnographic data contains many examples of this. Mayne (1862:259) describes Tsimshian female prophets in these terms: "It is not uncommon to see these old witches [i.e., shamans and seers] communicating their dreams to the tribe". Osgood (1959:122), a meticulous ethnographer, comments that the bear song, a 23 lucky animal song of the Ingalik, is not highly regarded - except by women! The assumption that men are necessarily the standard by which women should be judged in non-Western European cultures has been challenged by authors such as Briggs (1974), Dahlberg (1981), Leacock (1978), Mathiasson (1974), and Schlegel (1977a). In addition to the problem of evaluation introduced by the researcher's assumptions, the evaluation of status is also affected by the informant. The majority of the ethnographic data collected until very recently - data which are used to evaluate women's status - were collected primarily from male informants (see Schrijvers 1979:105). This choice of male informants is due partly to the androcentric assumption that males, their roles, and their activities are the norm for the society. It is also due, in part, to the sex of the ethnographer; men may be blocked in their attempts to discuss women's topics or to obtain access to female informants. VanStone (1963:34), for example, explains that because he was male, information on birth, infancy, and the role of women among the Snowdrift Chipewyan was difficult to obtain. The result of using predominantly male informants is that evaluations of women and their place in the society are often made by male ethnographers (with their own assumptions and biases about women) based on information provided by male informants.1 This reliance on male informants may reinforce the devaluing of women and their roles. At the least, the measurement and evaluation of men's and women's status on the basis of information obtained 1 Evidence suggests that although women ethnographers are becoming more common, they too have the same assumptions and views as their male counterparts. Thus their research may be plagued by the same biases and assumptions (see Bujra 1975:552 and Rogers 1978:126-27). 24 exclusively from men is methodologically suspect. A number of authors have pointed out that men and women may differ in their evaluations of them-selves and the opposite sex (Berndt 1981:186, Rogers 1978:129-131, 141-144, Sacks 1979:90-91). Rogers (1978:129-31, 141-44) argues that male and female cultural perceptions may differ; thus, it cannot be assumed that male perceptions are culturally representative. She maintains that, since gender roles are rarely identical, men and women have different concerns and perhaps different values and goals. It should not be automatically assumed they do not. Certainly we recognize that perceptions differ from person to person.1 Why should it be assumed, therefore, that male evaluations of female status will be the same as female evaluations of their own status and vice versa (Berndt 1981:186, Rogers 1978:144, Sacks 1979:90-91)? Drucker's ethnographic material provides an example of differences in male and female perceptions (1951:209). He gives an account of a woman shaman who is called in to cure a patient after repeated failures by male shamans. She openly jeers at the male shamans because, although they brag that male shamans are more powerful, she succeeded where they could not (see also Guedon on the Eskimo, personal communication, 1984; McDonnell on the Kasini, 1975:273; and Sacks, 1979:90-91 for other examples). The use of information obtained exclusively from male informants for the purposes of evaluating women's status creates certain problems which should 1 McDonnell (1975:291) provides an example of how the same behavior can be interpreted in vastly different ways by the actors who are involved. In this case, the actors are two men. The author spoke to two individuals, a Kasini father-in-law and his son-in-law, about bride-service. The father-in-law evaluated his son-in-law's role as the lower status role because the son-in-law owes service to the father-in-law. The son-in-law claimed that the father-in-law had the lower status role because the father-in-law was old and dependent upon him. be acknowledged by the researcher. Rogers (1978:144) suggests that public, male valuations of women should be taken into consideration, but "such valuations are rather lopsided grounds on which to base theories about women, if women in the same societies do not accept them, and have their own counter valuations." 2. Access to Power and Autonomy To this point, I have illustrated some of the difficulties associated with measuring and, particularly, with evaluating women's status in the anthropological literature. With this background information, I now discuss the dependent variables used in this thesis. As mentioned earlier, I am interested in one particular dimension of status - power and the manner in which power is distributed between men and women cross-culturally. I begin by defining "power". Next, I discuss the construction of my dependent variables. Finally, I list and describe the dependent variables. There are many definitions of power in the anthropological literature (e.g., Fried 1967:13, Levy 1952:468, Nielsen 1978:11-10, and Sacks 1979:72). The definition of power used in this thesis is Weber's definition. Power is "the probability that an actor will be able to realize his [her] own objectives even against opposition from others with whom he [she] is in a social relationship" (in Giddens 1971:156). This ability to realize one's objectives may be based on any number of factors - personal influence, economic posi-tion, prestige, skill, knowledge, persuasion, force, or even covert manipulation. I am interested in power which is overt rather than covert (sometimes de-fined as "informal power"). In addition, I am interested in the power which 26 is associated with a role or activity and with power as a right; I am not interested in the power associated with an individual's personality. Power has been measured in a number of ways in the anthropological literature on women. Friedl (1975:6-7), for example, is interested in overt power. She uses three indicators to measure it: control of production and the domestic and extra domestic distribution of strategic resources; the right to participate in political, ritual, and religious activities, and leadership in these activities; and autonomy in decisions about sex, marriage, residence, divorce, and the lives of children. Schlegel (1977b:17-20) is interested in the roles of women and men in the institutions of a society.1 Specifically, she asks how and to what degree the two sexes participate in the institutions of the society, and how power and authority within these institutions are allocated to men and women. According to Sanday (1973, 1974, 1981), who uses the same measure of status in all her work, status is defined as "the number of economic and political rights which accrue to women" (1973:1682), i.e., power. She measures status using four indicators along a Guttman scale; low status exists when no indicators are present, high status when all four indicators are present. The four indicators are: female control over produce, demand for female produce, female participation in political activities, and the presence of female solidarity groups (1973:1659; 1974:192-93). To date, Whyte's (1978) is the most comprehensive study of how and why women's status varies. Status he defines as "the relative power, influence, rights, and importance of women" (1978:50). Whyte uses fifty-two 1 Schlegel (1977b: 18) defines institutions as "those structures of roles designed to fulfill certain functions" such as the court of law and the family. 27 indicators from the religious, political, economic, and domestic spheres. He examines the patterns of association among these dependent variables; the re-sult is nine scales which represent distinct aspects of the status of women relative to men. Of these nine scales, at least four are directly concerned with power or autonomy: property control scale, power of women in kinship contexts, domestic authority, and control over own sexuality. As can be seen, in the anthropological literature on women, indicators of power include the roles people fill which have power as one of the status rights associated with them as well as the rights and responsibilities associated with certain roles. Thus, access to power can be observed from behavior which indicates that men or women have the ability to make decisions which affect others in various spheres, the ability to influence decisions which are being made, the control, manipulation, and redistribution of forms of economic wealth (since this gives people a means of influencing others), and autonomy - the ability to make decisions which affect one's own life.1 As Runciman explains (1968:49, 52), power is observed in the behavior between people (i.e., relationships). He cautions, however, that although power is manifested in actual performance, this does not mean that power is only observed in situations where people win (i.e., where they have their way). A person may not succeed every time, but if that person has the ability to re-alize his or her own objectives, that person has power. The dependent variables I have chosen measure various aspects of power in a number of spheres. These include ten dependent variables which measure autonomy. 1 Leacock defines autonomy as having decision-making power over one's own life and activities (1978:247). 28 There are two aspects of my dependent variables which need clarification. First, I am interested in whether there are differences in the power and autonomy available to each gender (and whether these differences are affected by three independent variables); such differences are not assumed to denote high or low status, however (i.e., the non-scalar usage of status).1 By considering differences in the access of men and women to power, without weighting or evaluating these differences, I am not forced to weight indicators of power cross-culturally by arguing that some indicators are better or worse indices of power. I do not suggest that such evaluations cannot be made, but I do not choose to make such evaluations without: 1) greater knowledge of how women's status is evaluated by men and women within their own culture, 2) greater knowledge of how certain roles are evaluated within each culture, and 3) more information on the roles, rights, and responsibilities of women in the cultures examined in this thesis. Moreover, there remains the whole question of whether there exists a unidimensional phenomenon, commonly called "women's status". Second, my dependent variables consider access to power and autonomy. I use the concept of access in order to exclude two types of information when I code my dependent variables. These are the frequency of participation in certain types of activities or the frequency with which they fill certain roles or claim rights, and the amount of power held. For example, if both men and women can be shamans in culture X, they are coded as having 1 A number of sociologists make a strong distinction between differentiation and evaluation. Differentiation does not itself imply distinctions of rank or value among the differentiated elements. A rank order is an additional act of evalua-tion, imposed on top of differentiation (e.g. Dahrendorf 1969:27 and Eisenstadt 1968:74-75). This distinction should be kept in mind in any discussion of status. equal access to this role and, therefore, equal access to the power associated with being a shaman in that society. Claims that male shamans are more powerful or more common will not be considered.1 Similarly, if both men and women can host feasts, statements that the feasts held by men are bigger or more important are not taken into account; men and women are coded as having equal access. There are a number of reasons for proceeding in this way. First, this approach allows me to minimize the effect of potential biases and resulting value-laden interpretations of ethnographers. Claims about the level of participation in activities or the amount of power held by women are often suspect because they seem to be subjective interpretations on the writer's part, rather than a description of the reality of women's lives. This is not altogether unexpected if one recalls that those observing or researching other cultures often have certain androcentric and ethnocentric biases, which affect the manner in which they perceive and interpret the information they obtain about women. I shall present a few examples from the data used in this thesis. The first is an early account of an incident in a Northern Athapaskan speaking band (MacNeish 1956:142, 149). MacNeish describes a meeting where the women soundly berated a male "chief for his behavior. She concludes that the chief had lost so much power that even women felt free to criticize him. MacNeish does not consider that the reason for such behav-ior might be tied to the lack of authority of "chiefs", a term which a 1 A statement that male shamans are more common because women can only be shamans if all the men have died would be considered, however. In other words, statements that women fill roles or have certain rights less frequently than men are not treated as evidence of less access unless there is a reason provided to explain why women have access to this aspect of power less fre-quently than men, and the reason must be based on gender distinctions. 30 number of authors consider inapropriate for describing such cultures (e.g., Leacock 1978, Ridington 1983, Sacks 1979). A second example is provided by Drucker (1951:183). He explains that although both Nootka men and women could be shamans, "most of his informants" believed that the male shamans were more powerful. Drucker's statement is somewhat suspicious for two reasons. First, twelve of his fifteen informants are male. Second, there are alternate data which contradict this statement; Sapir's and Swadesh's informants state that shamans of both sexes can be powerful (1978:438). A third example is provided by Jorgensen (1980:293) in his cross-cultural study of Western American Indians. He states that where shamans are predominantly men, they will treat the most serious illnesses, influence the weather, and forecast events. Almost invariably, he explains, women who are shamans are assigned less important tasks, which he labels "household shaman pursuits". These include administering to girls at puberty, aiding women with menstrual difficulties, aiding women with difficult pregnancies, and curing minor illnesses. One is left to wonder why these female pursuits are less important and how he explains the existence of female seers and prophets who also treat serious illnesses, forecast events, etc. Another example of doubtful information is provided by Barnett (1955:77-80). He claims that the pursuits of Coast Salish women are so unimportant that they do not need guardian spirits to help them in their pursuits' as men do. Examination of the works by Jenness (1955), Hill-Tout (1902:361), Curtis (1970a:95-96), and Smith (1949) indicates that Coast Salish women did indeed have (and need) guardian spirits. An interesting example of the devaluation of women's roles is provided by Slobodin (1962:48-69). He defines Kutchin leaders as men who host and who settle disputes. He explains that Mrs. Nash, a Kutchin woman in the community he studied, hosts and settles disputes in her community. Slobodin concludes that Mrs. Nash is, "in effect", a leader. McKennan (1965:52) makes an interesting evaluation of gender relations among the Chandalar Kutchin. He notes that the "allocation of the more strenuous and exciting roles to men tends to make the female role appear more inferior than it actually is." Thus, the very evidence which is used by other authors to "prove" women are drudges and slaves (i.e., hard work) leads McKennan to report that men's roles appear superior to women's roles. He goes on to say that "Like many hunting and gathering societies, the Chandlar are male-centred since, in the final analysis, it is the man who provides the food". He concludes that the society is male-dominated. He does not present his basis for stating men are more important to the diet (women also hunt and gather), nor does he explain his claim that because men are the basic food suppliers, this makes them dominant. A last example is from a discussion of Haida dances and the degree of women's participation. Blackman (1981:70), quoting Curtis (1970b: 142), claims that women did not have access to the most important dances. She also notes that Swanton's (1905:160-181) material supports this claim. A survey of these two sources indicates that although Curtis does make this statement, he has no basis for doing so, since he does not rank the dances (1970b:142-145). Similarly, Swanton does not rank the dances (1905a:156). Swanton differentiates the spirits associated with the dances into "good spirits", danced by women and men, and "rough spirits" whose dances are performed by men (1905a: 161). Nowhere does he state that this division constitutes a hierarchy. Thus, the claims of Blackman are not supported by the data. 1 These examples are by no means unique. The ethnographic literature is plagued by such statements which are, in some cases, suspect and, in other cases, contradicted by other information. By utilizing the access measure, I can minimize some of the potential bias introduced by androcentrism and ethnocentrism and make use of the information. There is a cost, however, as there are costs to any of the alternatives. The term "access" may mute some of the distinctions which could be made regarding male and female participation in roles and activities. Specifically, any differences in the access of men and women had to be clear before an indicator was coded lesser or greater access. Because of this, access may be coded as equal more often than would occur if one were considering informa-tion such as amounts of power, degree of participation, or frequency of participation. Some may view this as a bias in my work. I acknowledge this criticism, but I maintain that the biases of ethnographers, aided by the use of predominantly male informants, have so affected the information on women and their roles and activities in these societies, that it is better to err on the side of conservatism and speak about differential access, which can be clearly discerned, than to engage in the analysis of differences that may not exist. In addition to using access in order to minimize some of the subjectivity 1 Jorgensen (1980:501) codes sodality membership as open to both men and women among the Haida. in the ethnographic literature on women, I decided to use the access measure for two other reasons. I use the access measure to obtain a consistent basic level of information in the coding of the dependent variables. Some authors provide information on frequencies of participation or holding roles by gender, while others do not.1 There are three ways to handle these differences in the data. First, one could omit the more detailed information on frequencies of participation, degree of participation, or amounts of power where these are provided, in order to have a consistent basic level of information for all the dependent variables. This is the approach used with the access measure. Alternatively, one could make assumptions regarding the frequencies and degree of participation or amounts of power for those cases where such information is not supplied. This alternative would also give a consistent level of informa-tion. I rejected this approach because it may involve a higher degree of evaluation bias (intentionally or unintentionally). A third alternative would be to record the available data as presented. This wouid mean, however, that both across cultures and across dependent variables, I would sometimes be recording access and sometimes be recording more detailed information. Essentially, this would amount to using a different measure for each culture and each dependent variable. Yet, in the analysis, this information would be treated as equivalent. While it might be possible 1 Even when such information is provided, it can be difficult to interpret. For example, consider the statement, "there are fewer women shamans than men." Does this mean that fifty-five percent of the shamans are men and forty-five percent are women, or does it mean that eighty percent of the men are shamans and twenty percent of the women? (Indeed, what proportion of the population is female?) The answer to this question willl affect the construction of a dependent variable and, in turn, the results one obtains from statistical analysis. to adjust the analysis to take advantage of this situation, I have chosen to adopt the more simple approach described in alternative one - particularly since this approach helps to minimize bias, as described above. My last reason for choosing the access measure is also motivated by the state of the ethnographic literature. For many of the cultures used in this thesis, the accounts of women's roles, activities, and behavior are often in direct conflict with other accounts of the same culture.1 As an example, consider the data on access to supernatural power, S U P N A T . P U B (the first dependent variable). This dependent variable considers access to supernatural power which is used for the benefit of others. In the case of one cultural unit, three contradictory sets of information were reported. One ethnographer indicates that women shamans are rare. Three others indicate that there are fewer women shamans than men but that they are equally powerful; another ethnographer states that there are two classes of shamans and women are always of the less powerful second class. This level of conflict among authors occurs for many of the cultures. For instance, of the thirty-eight observations for S U P N A T . P U B , nineteen (fifty percent) have conflicting informa-tion regarding the extent of women's participation and power. The dependent variable S U P N A T . P U B is not anomalous in terms of the conflicting accounts presented by various authors. Similar conflicts can be found for many of my dependent variables.2 In fact, the only cultures for which I had complete agreement regarding the extent of women's access to supernatural power and 1 I suspect that this conflict is due partly to the subjective interpretations made by some authors and to the fact that some authors include information on frequencies and amounts of power, while others do not. 2 For SUPNAT.PRIV, which considers access to supernatural power used for one's own benefit, forty-one percent of the observations have conflicting accounts. 35 the amount of supernatural power held by women are those cultures where I have only one ethnographic source! In some cases, I can choose among the conflicting accounts on the basis of criteria which are discussed in Chapter 3. In many cases, however, there is little basis for choosing one account over another. In the same situation, Jorgensen (1980:188-89) indicates that contradictions in the data were coded as ambiguous and such cases were thrown out. As the above example illustrates, if I had followed this procedure, there would have been few cases left for any of the dependent variables. The use of the access measure was one method of dealing with contradictory data that does not entail throwing away a great deal of information. The access measure goes a long way in alleviating these conflicts among authors, since ethnographers are more likely to be in agreement about women's access to roles or their participation in activities than they are about the frequencies of women's participation or about the amounts of power held by each gender. For all of these reasons, I use the concept of access to power. As mentioned above, there is a cost, however. The access measure may mute some of the distinctions which could be made regarding male and female participation in roles and activities; thus, access may be coded as equal more often than would occur if one were considering information such as amounts of power, degree of participation, or frequencies of holding certain roles or claiming rights. It should be noted, this means the use of the access measure reduces the chances of finding significant associations (there is less variation). The thirty dependent variables chosen for this thesis consider access to 36 power and autonomy in various spheres. Specifically, they measure the access of men and women to roles, activities, and rights associated with power. They do not focus on men's roles and women's roles. This switch in focus was made for two reasons. First, it changes the focus from women only as child bearers, child rearers, and housewives. As I mentioned earlier, there is a tendency for Western Europeans to assume that this is all women do and to ignore those roles and activities which are open to both men and women. This may lead to the omission of important information about the power which is available to men and women. Second, by focusing on the power available to men and women, rather than on men's roles or women's roles, I am able to compare men and women. Such comparisons are important. Often, when women's status or power is discussed, the limitations imposed on women are reported, while those imposed on men are forgotten. This can lead to an imbalanced picture of the power and autonomy available to each gender. As Dahlberg (1981:17) points out, evidence that women cannot choose their own marriage partners will be noted and treated as evidence of less power or status, while the same limitations on men are passed over in silence (e.g., McKennan 1959:120, 134, 146 on the Upper Tanana). By comparing the access of men and women rather than concentrating on women's access alone, the limitations imposed on both men and women are more likely to be made visible. At this point, I turn to my coding procedure. Each of the thirty de-pendent variables was coded in two stages. The first stage was to code separately the access of men and the access of women to the dependent var-iables. This was done for thirty-nine cultures. For this stage, five categories 37 were used: access, conditions on access, no access, non-existent role or institution in this culture, and insufficient information. The final step was to compare for each culture, the access of men and women to each role, activity, or right. Five categories were used for this final coding: less access, equal access, greater access, non-existent role or institution, and insufficient in-formation. Both of these stages will be briefly discussed. In order to determine whether men and women have access to power or autonomy (i.e., the first stage), two points were considered. First, who in this culture has access to this type of power? For example, the defined role of community or kin group leader may be filled by a number of statuses (as defined by Linton): shaman, any skilled male, high status people, or grandmothers. From these possibilities, I then determined whether men have access to the role and whether women have access to the role. Although more awkward than simply asking "do women have access to this role?" and "do men have access to this role?", I chose this method in order to gain some insight into whether factors other than sex are affecting access to a role. As previously explained, access to a role can be affected, not only by one's sex, but by factors such as age, rank in the society, or kinship. In the initial stage of coding access, such factors were taken into account under the category "conditions on access". That is, in those cases where access to a role is not open to all the members of one sex, this distinction is noted. For example, all the women in a culture may have access to the role of shaman while only older men have access to the role. In this case, age is a condition for men's access. Five of these conditions were commonly found in the literature: 38 social rank, age, primogeniture, kin group membership, and, for women, no close male kin. After the initial coding of male and female access, each indicator was finally coded by comparing male/female access. For each indicator, if men had access which women did not have, or if there were conditions for women's access and none for men, the final coding was less access.1 If both had the same access (both yes, both no, or both had conditions for access), the final coding was equal access. If women had access to roles which men did not have or if there were conditions for men's access, the final coding was greater access. If the role did not exist in the culture, it was coded non-existent and if there was insufficient information regarding access for one or both sexes, the final coding was insufficient information. I did not assume that a lack of information on the access of women or men implies no access to that dependent variable for two reasons. First, precedent exists for treating the data in this manner. In his study of Western Indians, Jorgensen (1980:188-89) uses this approach; he reports that when there was no mention of a practise or a trait, the case was coded "ambiguous". Second, and more important, my familiarity with the ethnographic data led me to conclude that a lack of information is not necessarily an indication that the activity or behavior does not occur. My research for this thesis clearly indicates that women are regularly ignored by ethnographers (see Chapter 3 for a detailed discussion). Thus, gaps in the data are as likely to result from androcentrism or the ethnographer's inability to obtain information on women, as they are to be a case of no access. 1 Women are the point of reference. Three examples are provided to illustrate this point. McKennan describes Upper Tanana remarriage restrictions in three separate instances (1959:120, 134, 146). On all three occasions, he describes restrictions for widows but does not mention restrictions on widowers. Near the end of his ethnography, however, in a section entitled "Taboos" (1959:168), McKennan mentions, in passing, that both widows and widowers are constrained by the same restric-tions! A second example is provided by Drucker's description of the Nootka Wolf Ritual (1951:215-218). In Drucker's account, the presence and participation of women is barely noted. Eyewitness accounts of a Wolf Ritual presented in Sapir and Swadesh (1978:90-93, 97-99, 122) provide extensive evidence of the importance of women's participation on these occasions. The third example is provided by Boas (1916:528). He states that in Haida tales about acquiring crests and supernatural help, supernatural beings give crests and help to sons or daughter's sons. From this information, one could assume that women do not acquire crests or supernatural help. Nevertheless, on pages 160, 192, 414, 415, and 418 of Boas' ethnography, there are ex-amples of women obtaining a variety of crests, skills, and other supernatural power from supernatural beings. ' Thus, a lack of data does not imply that women had no access.1 1 It should be pointed out that, while the omissions of one ethnographer were sometimes corrected by other ethnographers (as in the example of Drucker and Sapir and Swadesh above), some cultures have few data sources and, therefore, these gaps could not be filled. 40 3. The Thirty Dependent Variables The thirty dependent variables measure access to power and autonomy for a wide range of activities. That is, I did not select a narrow range of indicators of power as many authors have done in the past (e.g., Cohen 1978).1 Because I am interested in how various aspects of power are distrib-uted between men and women, and because I am not convinced that women's status should be treated as a unidimensional phenomenon, these thirty indicators are not pooled in any manner. Rather, each one is treated as a separate aspect of power or autonomy. The final list consists of thirty dependent variables which represent six areas of cultural life: supernatural power, ceremonial roles, social leadership and community decision-making, economic roles, domestic life, and aspects of autonom}' in personal life. It should be noted that these areas cannot be sharply delineated in some cases (e.g., the household or local group is often both the domestic and the political unit). 2 The dependent variables are listed and then discussed in turn. Some examples of the roles or activities which are associated with the aspect of power are presented for the dependent variables. SUPNAT.PUB: Access to supernatural power or supernaturally acquired gifts 1 Only a few authors have attempted to examine a number of aspects of women's power or status (e.g., Friedl 1975; Sacks 1978, and Schlegel 1972). Whyte (1978) provides the most comprehensive list of indicators to date - and he uses them on a large sample of cultures. 2 Fried (1968:20-21) defines political organization as those portions of social organization that specifically relate to the individuals or groups that manage the affairs of public policy or seek to control the appointment or action of those individuals or groups. Both Fried (1967:128-29) and Levy (1952:485, 500) state that predominantly politically oriented institutions need not exist in a society. The political organization ma}' be embedded in other institutional activities. which can be used for the benefit of the community or kin group (e.g., shaman, prophet, seer) SUPNAT.PRIV: Access to supernatural power or supernaturally acquired gifts which are used for one's personal benefit (i.e., gambler, canoe-maker, basket maker) L E A D R : Clearly defined leadership roles in the kin group or community (Whyte 1978:57) (e.g., house head, shaman) C O M M U N . I N F L : Influence in community or kin group affairs (e.g., council member, any community member, elder) C E R E M . H O S T : Host or sponsor of kin group, community, or intercommunity gatherings such as feasts and potlatches (This role is associated with economic power as well, since the ceremonial host is a redistributor) D O M E S T . D E C : Access to decision-making in those domestic matters involving spouses (husband, wife) SUBSIST.LEADR: Organizer/supervisor of subsistence activities involving more than two people T R A D E / R E C . D E C : Access to organizing and decision-making in the area of trade, barter, and reciprocity (e.g., house head, high status person) INH.INDIV.PROP: Inheritor of individually owned, corporeal economic property (e.g., child, aunt) INH.CEREM.PROP: Inheritor of individually owned, ceremonial property INDIV.PROP: Individual ownership of private property 42 M A N . K I N . C H A T T : Steward/trustee or manager of chattel property owned collectively by the kin group (e.g., house head, eldest child, any family member) MAN.KIN.PROP: Steward/trustee or manager of important, fixed, economic property owned collectively by the kin group M A N . K I N . I N T A N G : Steward/trustee or manager of intangible property owned collectively by the kin group OWN.OWN.FRTS: Controller of the disposal and use of one's own fruits of labour within the unit consisting of spouses (e.g., husband, wife) N O N D O M E S T . F R T S & W E A L T H : The right to expect or ask for material goods from someone who is not a spouse, whether the goods are donated, given, or lent for a period of time CO-OP.FRTS: Controller of the disposal and use of fruits of labour, primarily food, produced co-operatively by a unit consisting of spouses M A R . A R R A N G : Access to arranging the first marriages of younger kin group members (e.g., house head, mother, grandfather, sister) ED.CHILD: Educator and disciplinarian of children belonging to one's kin group (e.g., household member, parent) SIB.LEADR: Role of leadership among siblings The remaining indicators measure access to autonomy in areas of private life. P R E M A R . S E X : Access to premarital sexual freedom as an accepted cultural 43 norm O W N . M A R . A R R A N G : Access to autonomy in choosing one's first spouse E X T R A M A R . S E X : Access to extramarital sexual freedom DIVORCE: Freedom to initiate divorce REMAR: Access to autonomy in the decision to remarry D O M E S T . A U T O N : Access to autonomy in decision-making about one's everyday activities W.OTHR.ROLES: Access of women to miscellaneous roles associated with power or power mentioned in the ethnographic literature M.OTHR.ROLES: Access of men to miscellaneous roles associated with power or power mentioned in the ethnographic literature W . F A T E . A F T . M A R : Access of women to autonomy in decision-making in the area of spouse exchange and wrestling for wives M . F A T E . A F T . M A R : Access of men to autonomy in decision-making in the area of spouse exchange and wrestling for wives Each indicator will now be clarified and discussed. Both the first and second dependent variables consider access to supernatural power. I treat them separately in order to determine whether access differs for men and women, depending on whether the supernatural power is used for others or for oneself.1 The roles considered in 1 Actually, if one obtains power to help oneself (i.e., success in fishing, canoe-making, or weaving), others do benefit. It is in a more indirect manner, however. 44 S U P N A T . P U B include not only what ethnographers have labelled "shamans", but also healers, diagnosers, seers, etc. For SUPNAT.PRIV, the roles or activities could be and were performed without the aid of supernatural power, but if one obtained supernatural power to do them, or if one were very skillful, one would be recognized by others as being supernaturally endowed. L E A D R considers clearly defined roles of leadership in the kin group or local group. According to Fried (1967:1, 6, 82-83, 134), leadership is the setting of a course of action followed by others and the frequency of initiating behavioral streams as opposed to carrjdng on a behavioral stream initiated by someone else. He states that in egalitarian societies, leadership has the following characteristics: 1) it is not associated with threats or the use of sanctions; 2) it is displayed in transient fashion, moying from one competent person to another; 3) shifts in the locus of leadership are associated less with persons than with situations; 4) the limited presence of power is associated with small groups like families and vanishes as the scope of the group widens. In ranked societies, leadership is regular and repetitive, extending into various aspects of social life. Nonetheless, as Fried points out, the leader still has no power to force compliance. The coding of access to leadership was more difficult than the above de-scription would suggest. There were a number of reasons for this. The first is a tendency for ethnographers to define leaders in all societies in terms of Western European concepts of political leadership. These concepts include con-trol, male leadership, domination, and subordination. Because of these conceptions of leadership, it may be imputed where it does not exist or 45 where it does not function in a manner familiar to those from a Western European background. As Miller concludes from his finding that many of the Indian societies of North America lacked vertical authority (i.e., "legitimatized imperative control"), the only concept of authority found in Western European cultures, "This raises certain questions as to the utility of the concept 'authority' for purposes of cross-cultural analysis" (1955:287-88). Certainly, early descriptions of Mackenzie Subarctic groups indicate that the writers were surprised and chagrined at the lack of respect and obedience shown to "chiefs" by their "followers" (e.g., Hearne 1968:284; Petitot 1876:34). In fact, a great deal of the ethnographic literature treats leadership in egalitarian societies as though it were permanent in nature and as though it involves relationships of dominance and subordination. This is not a problem restricted to ethnographic material written in the past. In his work on the Chipewyan, Smith (1982:33-37) presents a formal typology of three types of Chipewyan leaders. In contrast, Mackenzie (1971:cxxv), in his 1801 account of his voyages across northern North America, reports that the Chipewyan had no government. Similarly, in his ethnographic account of the Chipewyan, Birket-Smith (1930:66) claims they had no chiefs.1 The problem of discerning whether leadership does exist and who has access to it is not limited to the egalitarian hunting and gathering societies 1 Of course, this difference may be the result, not onty of ethnocentric bias, but also of changes in the Chipewyan institution of leadership over time. The whole topic of culture change is a recurrent and confounding problem in dealing with the ethnographic material. Yerbury, for example, believes the Chipewyan described by early fur traders and explorers had already undergone cultural changes because of the fur trade; they were no longer representative of some pure, untouched aboriginal culture. Nevertheless, he argues, Subarctic specialists perpetuate the former error because they rely primarily on materials gathered in the field to reconstruct what are claimed to be the baseline conditions of the aboriginal culture (1976:257). 46 in this thesis. Recent evidence suggests that leadership on the Northwest Coast also may have been less formal than is indicated in the literature. In a recent work, Drucker suggests that chiefs were socio-ceremonial and not political leaders (1983:87). Morice (1892:118-119) recognizes that the Tlingit did not possess European style leaders, while Suttles' Salish informants denied that such a role existed among the Coast Salish speaking peoples (1951:303-04). Thus, even on the Northwest Coast, leadership was not a fixed position filled only by one person. Leadership was dependent upon the situation, with many possible "leaders" (i.e., people who initiate streams of be-havior; see Duff in Oberg 1973; Suttles 1951:274). In this thesis, the problem is not so much that some ethnographers conceptualize Northern Athapaskan and Northwest Coast leadership as a hierar-chy composed of one leader above a group of people, but that in conceptualizing leadership in this manner, they ignore or do not observe other types of leaders in the community or local band, ,and that they do not expect to observe women in any position of leadership. Bonnie C. Freeman (1976:251) suggests that women in public affairs are treated as aberrant. She argues, "It is an open question whether women so completely lack access to the higher echelons of political decision making or if the methods employed by students of community power operate systematically to remove women." Ridington (1983:76) acknowledges this problem in a recent article on the Dunne-za. In an important paper . . . Eleanor Leacock argued that personal autonomy is an important adaptive device in egalitarian hunting and gathering societies. 'I suggest that personal autonomy was concomitant with the direct dependence of each individual on the group as a whole. Decision-making in this context calls for 47 concepts other than ours of leader and led, dominant and deferent, no matter how loosely these are seen to apply' (Leacock 1978 249). Nachi's story gives substance to Leacock's assertion. When the two old men who would have been identified as the band's leaders by an observer whose own culture assumes the existence of social hierarchy and male dominance, failed to make contact with the trails of animals in their dreams, a girl and a young woman took the obvious and rational course of action. They went hunting. The following ethnographic information suggests that the ethnographer may have (unintentionally) removed evidence of leadership roles held by women. The first example is found in the literature on Mackenzie Subarctic cultures. A number of scholars who have done research in this area indicate that shamans are often leaders (e.g., McClellan and Denniston 1981:384). From the literature, it is obvious that women are shamans. Only Smith (1981:34) and Allen (1887:143) affirm, however, that as "adepts" or shamans, both men and women are leaders. The second example comes from the data on Northwest Coast cultures. Ethnographers indicate that "chiefs" (i.e., high status people) are usually leaders. The only people referred to as leaders are men. Nevertheless, the data indicate that women are chiefs as well. Drucker, for example, reports that hawih means "chief and not "rich man" (1951:243). He makes no mention of women. Curtis, on the other hand, reports that hawih is used for male and female chiefs or high status people (1970a:203). Similarly, Barnett defines siem as male chief (1955:243-251), while Suttles indicates that siem means sir or madam and probabfy refers to a wealthy man (my emphasis) or potlatch leader. He notes that leadership in other matters was apt to be in the hands of others. Since some ethnographers do report the presence of women who are chiefs (i.e., hawih and siem), and since hawih 48 and siem were leaders, one must ask whether women who were hawih and siem were also considered leaders within their own groups. What of the "other leaders" mentioned by Suttles? Did women have access to these leadership roles? Morice (1891:112, 118-119) makes the interesting point that since all Upper Carrier notables (i.e., high status people) have a great deal of influence, all notables, including women, may be considered leaders. Could this state of affairs have existed in those other societies characterized by ranking? There is some incomplete but tantalizing evidence that women in some of these cultures were considered leaders, even if ethnographers did not define them as such. A Sapir and Swadesh account (1939:195) of the advice given to young "princesses" notes that they were told to feed their people.1 There is evidence that even when the data do indicate that women hold leadership roles, these roles are devalued in the reports (e.g., Slobodin 1962:59-60). Can it be that ethnographers are ignoring evidence of women's access to leadership? One further problem was encountered in coding access to roles of leadership. This is the distinction between political and domestic affairs. The two spheres may be interrelated in groups organized on the basis of kinship; Finley (1973:19) and Smith (1983:33) suggest the two cannot always be 1 Part of the problem with determining whether women have access to leadership roles on the Northwest Coast is a tendency for ethnographers to equate "chief and "leader", as seen above. Murdoch (1965a:281) argues that there are two types of rank among the Haida - one "political" and one social. If this distinc-tion exists among the other groups on the Northwest Coast, what does it mean in terms of women's access to leadership roles? Drucker (1955:235), for example, makes the interesting statement that high ranked women could outrank chiefs. What power do such women wield? What position do they hold in the power structure of the culture? differentiated. "Is, for example, marriage a domestic and never a political matter?" (Smith 1983:33). In describing the councils held by Nootka chiefs, Boas (1890a:585) reports that they discuss important tribal affairs such as war, peace, and the marriages of their sons and daughters. Similarly, Seguin (1984:xiii) notes that among the Tsimshian, the chief and all holders of ranked names discuss topics such as marriages, alliances, resource use, successions, and defense. Oberg (1973:34) asserts that Tlingit marriages are a political decision. It is difficult to establish whether women, who can make certain decisions which might be considered political as well as domestic in nature, have access to clearly defined leadership roles and political power, or whether they wield domestic power (which is generally devalued in the anthropological literature). In this thesis, I make a distinction between domestic and social leadership. In terms of domestic matters, I am interested in how power be-tween a husband and wife is divided. Domestic power, therefore, refers to matters within the unit consisting of spouses and their children. Leadership roles in the kin group or local group are concerned with those matters which affect units larger than this basic unit of spouses and children (i.e., more than one of these units). C O M M U N . I N F L considers power in terms of influence in community affairs. This dependent variable is an attempt to discern whether women have political influence in cultures in which ethnographers ignore indications of women's access to leadership or where leadership is defined in such a way that the role of women in the public sphere goes unrecognized. Even if women are not defined as leaders by the ethnographer, I am interested in 50 whether women play some part in the public decision-making process. C E R E M . H O S T considers access to the hosting or sponsoring of intercommunity, intracommunity, or kin group ceremonies. This variable indi-cates power both in the ceremonial and economic spheres, since one who sponsors such ceremonial events is a redistributor, a role which gives great economic power (i.e., the peak position in Fried's pyramid of statuses in ranking societies; see Chapter 2). 1 There is one major difficulty in coding access to this role. Ethnographers rarely comment on women's roles outside of the role of wife. The rights and duties of women as members of their own kin group are often ignored. It is difficult to ascertain whether a woman can host a ceremony as a kin group member, whether she only aids a husband who hosts, or whether she and her husband are viewed equally as hosts. Ethnographic descriptions speak of men as hosts; the women who are mentioned are host's wives (e.g., Barnett 1939:264, 269; Drucker 1951). The terms used to define sponsors are often vague and the terms are commonly treated as interchangeable. One reads that the "father," "parents," "family," and "relatives" host a feast - often within the same paragraph (e.g., Barnett 1955:131, 192, 196, 220; Drucker 1951:371; Emmons 1911:103, 106; McDonnel l 1975:173; M c K e n n a n 1965:64, 131; Oberg 1973:40). If a number of people may individually host, or a group co-operatively hosts, a potlatch or feast (see Barnett 1978:82-83; Drucker 1978:144-45; Suttles 1969:299), are 1 Jorgensen (1980:139) defines redistribution as a practice by which a person with authority or some group (village, sodality, kingroup) collects chattels and food from several people and then redistributes these in the form of gifts, feasts, or both to people other than those who produced and collected it, but perhaps including those who produced and collected it. women involved? Clues to women's access were gained both directly (e.g., Barnett 1955:257) and, more often, indirectly - for example, from statements that the host and wife must obey the ritual taboos imposed on a potlatch giver (e.g., Guedon 1981:581), from statements that a woman gains prestige from a potlatch, or from statements that young girls and boys are ceremonially important (e.g., Sapir and Swadesh 1978:381) and are taught correct ceremonial knowledge and behavior (e.g., Mcllwraith 1948, Vol.I:708). D O M E S T . D E C measures access to the process of decision-making in the unit consisting of spouses. For this dependent variable (and for T R A D E / R E C . D E C , M A R . A R R A N G , and ED.CHILD), I am interested in the whole process of decision-making, and not only the defined "family head" or final decision-maker. Access to the decision-making process is the important factor. By regarding the whole process of decision-making, I hope to overcome the tendency to keep women invisible.1 As Berndt argues in her analysis of marriage-arranging among Australian Aborigines, too often the roles of women are excluded unless the whole process of decision-making is consid-ered. She .reports that Aboriginal women were involved in both the preliminary stages and in the final stages of decison making; "It is, and was, never just a matter of who was involved in the final decision" (1981:186). Interestingly, this focus on the whole process of decision-making does not favour women. In this thesis, many cases existed where considering the 1 This emphasis on the whole process of decision-making is simply part of the focus on access to power rather than descriptions of amounts of power or frequencies. whole process of decision-making was advantageous to men. For D O M E S T . D E C , T R A D E / R E C . D E C , M A R . A R R A N G , and ED.CHILD, a number of cultures have women as the "ultimate authority" or decision-maker. Nonetheless, if men are said to have some influence on the decisions that are made, both women and men have the ability to realize their own objectives in the situation; thus both are coded as having access to power in these areas. An examination of the whole decision-making process has an additional advantage. Since I am interested in the whole process, I do not accept unsubstantiated generalizations about power and authority. Statements that the man is "lord and master" or the woman is "ruler of the lodge" (Wheeler 1914:54) are not sufficient grounds for assuming the other partner in the domestic situation has no power. Such statements do not provide sufficient information to ascertain whether there is differentiation in the power or influence held by husbands and wives. More specific information must be sought (e.g., Klein's statement that either a Tlingit husband or wife can rep-resent the household in public 1975:212-213). Where such detailed information was unavailable, the indicator was coded "insufficient information". The reason for this caution is, again, the desire to avoid accepting subjective evaluations as facts (for example, see Osgood's analysis of the commonly quoted descrip-tion of Mackenzie Subarctic women as "drudges" and "slaves", 1958:113). SUBSIST.LEADR codes access to the role of organizer or supervisor of subsistence activities involving more than two people. If subsistence activities are segregated by gender, one might assume both men and women would organize/supervise their own subsistence activities. If they are not segregated, 53 who is in charge? Does leadership depend on the activity and the area of expertise needed or does it depend on gender? Unfortunately, there are large gaps in the ethnographic material regarding these aspects of women's activities which make it difficult to decide if women have the power to organize and supervise such activities; Koppert (1930a:56-80), for example, uses twenty-four pages to discuss the subsistence activities of Nootka men; women's subsistence activities are covered in four pages. 1 The next ten variables, T R A D E / R E C . D E C to C O - O P . F R T S , measure access to economic power. Economic power is displayed in the ability to dispose of and use resources, produce or fruits of labour, and goods and property. The ability to do so gives one leverage in dealing with others and in achieving one's objectives. T R A D E / R E C . D E C measures access to decision-making in the areas of trade, barter, and reciprocity. Jorgensen (1980:139-40) defines reciprocity as a donor giving something to a recipient who will reciprocate at some point in the future. In barter, two people bargain over the value use of two items being exchanged. In trade, standard values are used to establish the worth of an item being traded. Although these definitions seem straightforward, distinguishing among them can be difficult. Nevertheless, I a m interested in determining to what extent men and women engage in non-domestic exchange activities since their ability to do so gives them wealth which can be used to influence other people. Generally speaking, data were adequate for the 1 This dearth of information on the organization of women's subsistence activities is not limited to the geographic area studied in this thesis. In his cross-cultural study of women's status, Whyte was forced to discard a dependent variable which measured subsistence leadership, due to lack of information (1978:90). Northwest Coast cultures but sparse for the other culture areas. Observations made by traders and explorers were of value in filling some of these gaps. INH.INDIV.PROP, INH.CEREM.PROP, and INDIV.PROP are concerned with property inheritance and individual ownership of property. INH.INDIV.PROP codes access to the inheritance of individually owned economic property (i.e., property that is owned individually and not corporately). INH.CEREM.PROP codes access to the inheritance of individually owned ceremonial property, while INDIV.PROP codes access to the individual ownership of property.1 Originally, I included all three variables because I was interested in discovering whether economic power varies, depending on the type of property, and whether there is a difference in access to ownership and inheritance rights. For these three dependent variables, there was one prob-lem - a difficulty in distinguishing between personal property and corporately owned kin group property. This may be the reason that these three varia-bles provided no significant results in the analysis. Olson acknowledges the difficulty of differentiating individual, household, and clan property among the Tlingit (1967:37c). This confusion makes it extremely difficult to determine who can have control over different types of property and the limits of the controls. For example, usufruct rights to kin group owned property, at least in some cases, seem to have included the right of the individual to pass on the usufruct rights without input from the kin group. In other words, some corporately owned property was disposed of by individuals. Suttles (1951:404) points out that a name, which is owned by the kin group and is the sign 1 Property ownership involves the right to legitimately determine some significant aspects of the employment to which the property or resource is put, the alloca-tion of benefits from its use, or the right to dispose of the property or some part of it (Swanson 1966:48). 55 of membership in a kin group, was the property of the person who bore it and he (or she) could give it to a descendant and take another. Alternately, although property was said to be individually owned in some cases, others had rights to it. Goldman (1953:164-166), for example, speaks of both individually owned and clan owned fish traps, yet the fruits of the individually owned traps could be and were claimed by the joint family, as were the fruits of clan owned traps. M A N . K I N . C H A T , M A N . K I N . P R O P , and M A N . K I N . I N T A N G code access to the role of manager or steward of corporately owned or held kin group property of various types. 1 M A N . K I N . C H A T codes access to the role of steward or manager of corporately held kin group chattel property . 2 M A N . K I N . P R O P codes access to the role of steward or manager of non-movable corporeal property such as fishing sites and houses. 3 M A N . K I N . I N T A N G considers access to the role of steward or manager of kin group owned intangible property such as songs, dances, titles, rituals, and crests. 4 A careful survey of the literature indicates that at least some of the intangible property (e.g., titles, songs, dances, and rituals) was treated as personal property by group members (Garfield 1939:276; Murdock 1965:282; 1 Jorgensen (1980:137-38) states that a steward was the representative of the kin group; all the members of that group were co-owners of the property, however. Thus , stewardship did not give one the right to dispose of the property although the manager had greater control over its use. The steward was generally the highest ranking member of the kin group. 2 Jorgensen (1980:134) defines chattel property as movable, corporeal property. 3 Corporeal property is tangible, a physically material body of some sort (Jorgensen 1980:135). Driver (1969:269) includes permanent houses in this category; movable houses he classifies as chattels. 4 Intangible property does not have a physical presence. It "constitutes a right that has no physical existence" (Jorgensen 1980:135). 56 Oberg 1973:30, 62; Olson 1967:38). OWN.OWN.FRTS codes access to control over (i.e., the ability to decide about the disposal and use of property and resources) the fruits of one's own labour within the domestic sphere (i.e., the unit consisting of spouses).1 N O N . D O M E S T . F R T S & W E A L T H measures access to economic power in terms of having the right to expect or to call upon the fruits of production and material wealth of non-spouses (usually household members, kin, and affines). The produce includes produce obtained as a right to first fruits in return for the use of kin group owned property. Produce and wealth which were requested might have to be repaid. As is the case with OWN.OWN.FRTS and CO-OP.FRTS, these resources could be used for the purposes of redistribution. CO-OP.FRTS measures access to economic power in the area of domestic production. Which gender has the ability to decide about the disposal and use of fruits of labour, primarily food, produced jointly by spouses? The labour may have been carried on simultaneously or sequentially. This produce might be used for various purposes, including buying services or redistribution. M A R . A R R A N G , E D . CHILD, and SIB. L E A D R deal with other aspects of the domestic and the larger social sphere. M A R . A R R A N G measures access to the process of arranging the first marriages of younger kin group members. As was the case with D O M E S T . D E C and T R A D E / R E C . D E C , I am interested in the whole process of decision-making and not simply in the final decision. ED.CHILD measures access to control over and the education of children 1 Since reciprocity is a norm in the cultures in my sample, giving away some of one's produce to others because of expectations of reciprocity is not considered to be evidence of a lack of economic power in this area. 57 in the household. General comments that mothers had ultimate control over children were not treated as evidence of greater access unless men had no involvment. In other words, where women were said to have final authority but the data indicated that men have some input into the process, access is coded equal. S I B . L E A D R codes access to the leadership role, if any, among a group of siblings. The remainder of the dependent variables measure access to autonomy in various areas of one's life. P R E M A R . S E X examines access to premarital sexual freedom as a cultural norm. One recurrent problem was encountered in coding this measure; the lack of information regarding the behavior ex-pected of men. This omission may indicate male autonomy, or it m a y indi-cate the presence of an assumption that only the sexuality of women is. con-trolled. In order to circumvent this omission, I used statements about the consequences of such behavior if a couple were caught. I did not rank the consequences in any manner; I simply treated the presence of consequences as evidence that premarital sex was not acceptable. O W N . M A R . A R R A N G considers autonomy over the arrangement of one's own first marriage. The data regarding this dependent variable were often unclear. For instance, general statements that the boy asks the girl or that he initiates proceedings may or may not be evidence that he has autonomy. Is he simply carrying out orders (see Briggs 1974:292)? The presence of the sororate and/or levirate were also confusing since there is a tendency to treat both institutions as evidence that women are limited in their behavior; the freedom of men in the presence of these institutions is rarely questioned, 58 however. In addition, the role of the girl in the process of marriage arranging often goes unmentioned. E X T R A M A R . S E X measures sexual autonomy after marriage, as' a norm in the society.1 As with P R E M A R . S E X , information on the consequences of such behavior was used to fill gaps in the data. Unfortunately, the consequences noted are generally those imposed by a wronged husband and only rarely by a wronged wife. DIVORCE is concerned with the ability to initiate divorce. R E M A R considers autonomy in the area of remarriage. Specifically, I am interested in the presence of the sororate, levirate, mourning taboos, the bondage of the bereaved, or the inability to remarry without permission. Two problems were encountered in coding. The first is the manner in which the sororate and levirate are interpreted; that is, who is being constrained? The second problem is the vagueness of the information and omissions in the in-formation (one third of the cases were coded insufficient information). Many authors assume that only women are constrained by the levirate and sororate. The data state that a widow must marry her dead husband's brother; is he not bound by the levirate? Similarly, the data will also say that a sister must marry her dead sister's husband; is he not bound by the sororate (e.g., Jenness 1939:52; Emmons 1911:100; Olson 1967:21)? Given this type of information, it is difficult to discover whether the levirate or sororate is compulsory for the woman and optional for the man or whether 1 One must be careful when investigating "norms". As Rogers (1978:131) indi-cates, norms and values may not be viewed in the same light by men and women. I met with this problem in my research for this thesis. Ross (1872:310) states both Dogrib and Slave men and women believe they are free to have extramarital affairs, but the men do not accept that women are. the author's bias has coloured the data. Olson (1967:21), for example, states that a Tlingit widow is asked whom she will remarry - but she is expected to name the nephew. Kle in (1980:100) records a missionary's account (1888) of such an event which places a different light on the widow's role in this process. Describing her as "not quite as old as Methuselah," the missionary explains that the chiefs widow sent a messenger to call the three sons of one man, the eldest being eighteen, "that she may look at them and see if she wants either, or perhaps, all of them to take the place [as husband] of the dear departed". Likewise, one of Cruikshank's (1979:21) Yukon Subarctic female informants describes the levirate from a different perspective: "they gave my young grandfather to m y grandmother to look after her". In addi-tion to shedding light on the supposed limitations on women imposed by the levirate (and sororate), these accounts introduce a new question; is the levirate a duty of the widow as many claim, or is it her right? Dawson (1891:13), Garfield (1939:234), and M a c L a c h l a n (1981:464) acknowledge that, with the levirate, the young man has a duty towards the widow and that she has a claim on him. Who has autonomy in this situation? D a t a regarding the expected behavior of men were difficult to obtain in a number of cases (e.g., Jenness 1937:63, M c K e n n a n 1965:57-60, Morice 1897:46-47, Teit 1975b:591-92). D O M E S T . A U T O N codes access to autonomy in the daily activities of husbands and wives. This measure is less "formal" than those previously described. Explicit data were rarely forthcoming; instead, I looked for state-ments that husbands and wives have their own business affairs, that a spouse will leave on his or her own business, and that spouses are 60 independent. The last four dependent variables differ from the preceding variables for they do not compare men and women since men are rarely mentioned in these contexts. If I were to compare men and women, these dependent vari-ables would have been rejected because of lack of da ta . 1 A present/absent format is used to code the autonomy for each gender. W . O T H R . R O L E S and M . O T H R . R O L E S focus on miscellaneous power available to men and women. W . F A T E . A F T . M A R and M . F A T E . A F T . M A R ask whether women and men, respectively, make decisions about being lent as a sexual partner (i.e., spouse exchange) and, in the case of women, about wrestling for women; 2 that is, are they treated as passive objects in the transactions carried on by the other gender? F r o m the data it was difficult to discern who made decisions in these situations. Hearne (1911:104-05) describes unwilling women being dragged off by the victor of a wrestling match. Nevertheless, he also observes It has ever been the custom among these people for the men to wrestle for any woman to whom they are attached, . . . and, of course, the strongest party always carries off the prize. A weak man, unless he be a good hunter 3 and well beloved, is seldom permitted to keep a wife that a stronger man thinks worth his notice (emphasis added). Further evidence that women may have had some part to play is found in Morice (1910:984) who reports that wrestling matches were common, especially 1 Lack of data forced the exclusion of M . O T H R . R O L E S and M . F A T E . A F T . M A R from the analysis. 2 There were no cases in m y sample of contests for husbands. 3 Morice (1910:983) expands on this point, explaining that weaker men are usually old men. Because Dene women m a r r y young, and they marry older men, by the time they reach maturity, their husbands are old and declining. The women prefer and want younger and better providers. 61 if there were no children "to deter a passer-by from attempting to win her over to his side" (emphasis added). This confusion about the woman's consent extends beyond wife wrestling to situations of wife exchange. Both Honigmann (1949:164) and Osgood (1958:205), for example, state that women were active in these exchanges and encouraged them. There is a second complication. In a number of the accounts of wife wrestling, the situation seems to arise when different local bands meet (e.g., on trading trips or at a fort). Although the traders and early explorers may have considered these various groups to be all the same people if they spoke the same language, there is no evidence to indicate that local bands would so consider non-band members. If men wrestle for wives only with non-band members, one must ask whether the treatment accorded non-members should be considered in the same light as the treatment of fellow band members (when one considers women's status, for example, one considers the status of women within the community - not those who are not members of the community, i.e., slaves). To claim that women have no autonomy because non-band members victimize them and their spouses (after all, husbands had no choice in the matter, either), may not be appropriate. This completes the discussion of the dependent variables. Although not exhaustive, this list of dependent variables is comprehensive, covering a wide range of aspects of power. Any differences in access which are observed on these dependent variables are statistically tested against three independent vari-ables: degree of stratification, type of descent, and final postmarital residence pattern. In the next chapter, I discuss the independent variables used in this thesis. In addition, I review a number of predictions which have been made by researchers about the manner in which these independent variables affect women's status and power vis-a-vis men. II. T H E INDEPENDENT VARIABLES A. INTRODUCTION A major topic in the anthropological literature on women is the factors that give rise to and perpetuate sexual inequality. Some anthropologists argue that "female subordination" and "male domination" are universal facts (e.g., Cohen 1978, Ortner 1974, Rosaldo 1974, Schneider 1961). M a n y of these researchers expend their efforts in attempting to find universal explanations to account for the universal subordination of women. The majority of these explanations are biological. Women's roles (which are defined by culture) are viewed as biologically determined (see Friedl 1975 and Sacks 1979 for further discussion of this topic). Other anthropologists disagree with this approach, both on empirical and methodological grounds (e.g., Barrett 1980, Leacock 1975, Leacock and N a s h 1981, Poewe 1980, Rohrlich-Leavitt 1975, Sacks 1979a, Schlegel 1979a, Tiffany 1979, Whyte 1978). These authors maintain that women's status varies, and they attempt to explain how and why sexual inequality occurs and is perpetuated. Whyte, for example, declares, "The whole notion of assuming universal male dominance and then looking for universal explanations for that dubious assumption seem to us an unproductive enterprise" (1978:168). H e argues that if a pattern is truly universal, then explaining its origins or reasons for existence is methodologically difficult because one cannot test the validity of different explanations of a phenomenon unless that phenomenon varies in the societies under study. To explain the origins of such a universal pattern, one is reduced to speculating about events before the dawn of recorded history or referring to other universals (such as the aggression-related hormones 63 64 in males used in Goldberg, 1973). However, as long as a pattern is universal these explanations rest mainly on plausibility, rather than on rigorous testing (1978:6). Whyte concludes that it would be more useful to examine the relative share of women in political power "and how this varies cross-culturally than to simply stop with the assertion that men tend to monopolize political positions in every society" because such variation is worthy of study (1978:4-6). E v e n those who assert that women's status varies, however, often base their explanations for the variation on plausibility rather than vigorous testing (Friedl 1975, M a r t i n and Voorhies 1975, Nielsen 1978, Sanday 1973, 1974, Schlegel 1972, Whyte 1978 are exceptions). In this thesis, I test for a relationship between gender differences in access to power and three independent variables, degree of stratification, type of descent, and final postmarital residence pattern. These independent variables are the focus of ongoing debate in the anthropological literature on women's status. In the remainder of this chapter, I consider each independent variable in turn. In the first part of each section, I review some of the predictions made by various authors regarding the effect of the independent variable. Since the independent variable can be defined in different ways, I describe my definition of the variable in the second part of each section. I conclude each section by presenting the hypotheses to be tested. B. DEGREE OF STRATIFICATION 65 1. A Review of the Models A number of the authors interested in the conditions which facilitate sexual equality and inequality have suggested the amount of sexual inequality in a culture is related to the amount of social stratification in the culture. These authors include Leacock, Gough, Friedl , Schlegel, Mart in and Voorhies, and Whyte. Before discussing the models of these six authors, I define the term "social stratification." According to Sahlins, social stratification has two aspects: degree of stratification and form of stratification. '"Degree of stratification' refers to the complexity of the status system, i.e., the number of different kinds of ranks and the extent to which they confer unequal privilege in economic, social and religious life" while '"form of stratification' refers to the social organization of the status system, that is, the sociological principles that determine rank differences" (1958:x). These principles or qualifications for rank are any principles beyond sex, age, and personal attributes, since, as Sahlins explains (1958:1-2), Theoretically, an egalitarian society would be one in which every individual is of equal status, a society in which no one outranks anyone. B u t even the most primitive societies could not be described as egalitarian in this sense. There are differences in status carrying differential privilege in every human organization. Although differences in status are regular features of human organization, the qualifications for status are not everywhere the same. In certain societies . . . the only qualifications for higher status are those which every society uses to some extent, namely, age, sex, 1 and personal characteristics. Sahlins (1958:1-2) defines an egalitarian society as a society in which 1 Sahlins assumes, then, that in all societies sex will be a basis for differences in status. the only principles of rank allocation are the universal principles: age, sex, and personal characteristics. These societies are labelled egalitarian because, "given these qualifications, every individual has an equal chance to succeed to whatever statuses may open." A stratified society is one in which statuses are filled by a mechanism beyond these universals - for example, by inheritance (1958:2). Fried (1968:464) has created a typology based on the degree of stratification in a society. The categories are: egalitarian, rank, and stratified. Like Sahlins, Fr ied assumes that all societies differentiate among their mem-bers and assign them greater or less prestige on the basis of age, sex, and what he calls subsistence skills. Nevertheless, he maintains that such societies may be regarded as egalitarian since they are "undifferentiated enough". According to Fried's typology, an egalitarian society is one in which there are as many positions of prestige in any given ,age-sex grade as there are persons capable of filling them. If within a certain kin group or territory there are four big men, strong, alert, keen hunters, then there will be four 'strong men'; if there are six, or three, or one, so it is (1968:465-66). Fr ied observes that the unit of production in egalitarian societies is the household; no specialization exists. Each family group does essentially the same tasks, and although some people may be recognized for particular skills, there is no favoured economic role, no regular division of labour, and no political power residing in any of the recognized roles. Reciprocal exchange characterizes the egalitarian economy. Exchange m a y or may not be balanced. If one person is more successful at subsistence, he or she may, over time, give away more than he or she receives. The prestige gained does not give 67 this person a privileged economic or political role, however, because all recog-nize that one's luck can turn at any time (1968:465). Fr ied (1968:466-68) defines a rank society as one in which access to valued statuses is limited on the basis of criteria other than age, sex, or personal attributes. the rank society is characterized by having fewer positions of valued status than individuals capable of handling them. Furthermore, most rank societies have a fixed number of such posi-tions, neither expanding them nor diminishing them with fluctuations in the population, save as totally new segmented units originate with fission or disappear . . . The simplest technique of limiting status . . . is to make succession to status dependent upon birth order (1968:466). Fried conceptualizes the rank society as a triangle of statuses with leading statuses at the top. The leading status has economic significance as central collector of allotments and -redistributor of these supplies. Although there ex-ists no exploitative economic power nor genuine political power, Fr ied notes that status differences are marked by "sumptuary specialization and ceremonial function" (1968:468). There is one essential characteristic which differentiates Fried's rank society from the next category, stratified society, the rank society operates on the principle of differential status for members with similar abilities, but these statuses are devoid of privileged economic or political power, the former point being the essential one . . . Meanwhile, the stratified society is distinguished by the differential relationships between the members of the society and its subsistence means - some of the members of the society have unimpeded access to its strategic resources [subsistence resources] while others have various impediments in their access to the same fundamental resources (1968:469-70). The categories of stratification used in my thesis are based on Fried's typology. For the purposes of the thesis, the important distinction is the one between egalitarian societies (where access to statuses is unlimited on any basis other than sex, age, and personal attributes), and rank societies (where access is limited on the basis of other criteria). Fried's third category, stratified society, is not represented in this study. H a v i n g defined social stratification, I shall now return to the discussion of how different authors have conceptualized a relationship between sexual equality/inequality in a society and the degree of social stratification in that society. Some authors who assert social stratification is related to sexual inequality focus specifically on relations of production and the control of resources 1 (e.g. Leacock in Etienne and Leacock 1980, Sacks 1979). In terms of Fried's model, these authors are focusing on those valued statuses having economic significance at the top of the triangle (1968:466-68). This focus on the economic aspects of social stratification and, indeed, the whole concept of a relationship between social stratification and sexual inequality, seems to have originated with the nineteenth century evolutionists' theories regarding the development of class societies and the role of class societies in the rise of women's inequality. Engels (1972 [1884]) argued that the development of private property effected a change from matriarchies (societies dominated by women), based on female descent, to patriarchies 1 Etienne and Leacock (1980:8) define relations of production as "the relations set up among people as they produce, distribute, exchange, and consume the goods upon which they live". Godelier (1977:24-25) defines relations of production as "man's relationship to man, producers and non-producers, in the appropriation and control of means of production and the products of labour." Where access to these relations of production are restricted, on the basis of qualities other than age, sex, and personal attributes, Godelier claims that some degree of social stratification exists. (societies dominated by men), based on male descent, and then to class societies (1972:117-20, 129). Female subordination to men was tied to male control of wealth in the form of domesticated animals. T h a t is, communal matrilineal kinship gave way to patrilineal kinship because, as men came to control private wealth in the form of domesticated an imals , 1 they desired to leave this property to their children rather than to members of their matrilineal kin groups (1972:119-20). Thus , men overthrew matrilineality; they formed the patriarchal family based on patrilineal descent, and then they formed the monogamous family. Families became the basic economic unit and the institution that appropriates and perpetuates private property. M e n and women no longer produced for the use of their own kin group but for the family. This led to inequality in the ownership of property and later to exploitative class societies (Sacks 1974:207-08). The development of private property and the differential ownership of wealth led not only to the destruction of egalitarian social organization, but also to sexual inequality. Engels claimed (1972:129, 137) that women within the newly developed monogamous families became the domestic servants of men. Women now produced only for domestic consumption within the production unit. B y producing for domestic consumption, women were producing for men, who were, therefore, free to use the wealth they controlled for exchange. With these developments women were transformed from free and equal producers and members of society to subordinate and dependent wives (1972:137). O n the other hand, men, because they controlled private wealth, now produced primari ly for their own benefit and only secondarily for their 1 Engels did not explain why men come to control private wealth. 70 family unit (Schlegel 1977a: 10-11). Leacock (Etienne and Leacock 1980:5-16), in her work on sexual inequality, uses Engels' basic ideas of advances in technology leading to new unequal relations of production. However, she contends that changes in the relations of production which accompany technological changes, and not technological changes per se, create inequality for women. 1 It is the breakdown of communal ownership of property and the isolation of the individual family as the primary economic unit that reduces women to the role of domestic servants in the home. Leacock (Etienne and Leacock 1980:6) asserts that while women's rights and duties vary from society to society, "Al l students of the subject agree . . . that women's degree of personal autonomy in band societies contrasted sharply with the oppression that characterizes their position in hierarchically organized societies." 2 A s for the relationship between degree of social stratification and sexual inequality, Leacock remarks, there is a rough correlation between the position of women vis-a-vis men and the degree of socioeconomic inequality in the society as a whole. This does not mean that female inequality follows from the development of class differences generally, instead, it is our under-standing that the origins of both socioeconomic and sexual hierarchy are inextricably bound together. (1980:8) In order to explain the rise of social and sexual inequality, Leacock 1 Leacock is concerned with the question of whether women control the conditions of their work and the disposal of the fruits of their labour. 2 Leacock uses the term "autonomy" rather than "equality" because she believes that equality is often assumed to mean "the same as men". Thus , in societies where women and men have different roles and spheres of activities, women are assumed to have lower status. She holds that men and women m a y have different roles and spheres of activities, yet women may have sexual equality with men in the society (Leacock 1978:248, 270; Etienne and Leacock 1980:4). 71 defines four types of relations of production which have evolved in sequence. E a c h is characterized by a different degree of social stratification: egalitarian, ranking or transitional, preindustrial hierarchical, and industrial capitalist. In egalitarian societies, all adults participate more or less equally in production, distribution or exchange, and consumption. The term "participation" does not mean that all people perform the same tasks. Rather, as Leacock (Etienne and Leacock 1980:8-9) explains, people contribute work according to their abilities and interests. Thus , in egalitarian societies, there are as many statuses in any age-sex category as there are people who wish to fill them. People cannot be. alienated from their right of access to basic resources. The only division of labour is based on sex. There is no division between a private, domestic, female sphere and a public, political, male sphere as exists in more stratified societies. In egalitarian societies, authority is dispersed and decisions are made by those who will carry them out. Social arts, such as myth-making, rituals, and gossip, are used equally by men and women to influence others (Etienne and Leacock 1980:9-12). In contrast, ranking societies are characterized by inequalities in production, distribution, exchange, and consumption. The inequalities are less marked, however, than they are in stratified societies. In ranking societies, high ranking people contribute to the organization of production and do few of the onerous tasks. H i g h ranking people are important in the distribution and exchange of goods, and they consume more (although not much more in comparison to stratified societies). These high ranking people tend to consume luxuries rather than necessities, however. Low ranking people have access to land in general though it may be somewhat restricted. Leacock also notes, 72 in ranking societies, the division of labour goes beyond sex; there are manufacturers of desirable items or collectors of raw materials for trade. Unlike specialists in stratified societies, however, these people continue to produce some of their own food and necessities (Etienne and Leacock 1980:12-15). Since preindustrial hierarchical societies are not represented in my sample, I shall only briefly mention Leacock's (Etienne and Leacock 1980:15-16) salient distinction between these societies and ranking societies. Like Fried (1968:469-70), Leacock asserts the major distinction between ranking and preindustrial stratified societies is that in preindustrial stratified societies, some people have no access to basic sources of subsistence. Thus a minimally consuming class of producers supports a maximally consuming class. A s for the causes of inequality, Leacock alleges that sexual inequality is based on trade (Etienne and Leacock 1980:13). The growth of trade was important for the development of ranking; early trade with the Europeans, and the exchange of local goods for European goods undercut the interdependence of co-operatively organized bands and villages. While egalitarian societies also engaged in trade, at some point trade became important enough to lead to specialization and, therefore, to the reorganization of productive relations. This development made possible unequal access to important resources. That is, with the development of some specialization in trade, adults of each sex were producing and procuring significant quantities of different things (instead of the same things). This created for the first time the possibility of ties of economic dependence between people. In ranking societies, a sector of the economy concerned only with production for trade 73 began to be differentiated from the lineage or household sector which was involved with production for use and sharing. This development laid the basis for inequalities among people and between men and women. In terms of sexual inequality, Leacock (Etienne and Leacock 1980:13-14) maintains that where women produced for the new market and participated directly in trade, they still held a recognized public status and autonomy in family life. Hunting and warfare often gave men the advantage, however, because these allowed men to be responsible for trade and external political relations. This led to the strengthening of patrilineal over matrilineal ties and the long range undermining of women's position. In this analysis, I test Leacock's proposition that in those societies which are egalitarian, women have access to public roles of power and to autonomy in private life. More specifically, I test for a relationship between egalitarian societies and women's access to these aspects of power. Another author, Kathleen Gough (1975), uses Engels as a point of departure in her work which postulates a causal relationship between social stratification and sexual inequality. Although she alleges that women are "always in some sense" the second sex, Gough (1975:69-70) argues, In general in hunting societies, however, women are less subordinated in certain crucial respects 1 than they are in most, if not all, of the archaic states, or even in some capitalist nations . A s Engels saw, the power of men to exploit women systematically springs from the existence of surplus wealth, and These crucial respects include: control over their own sexuality, control over their own labour and produce, control over their own children, autonomy in their own movements, not being used as objects in male transactions, and having access to areas of the society's knowledge and cultural attainments, a lack of the male possessiveness that leads to institutions such as savage punishment or death for female adultery, the jealous guarding of female chastity and virginity, denial of divorce to women, and a ban on remarriage of widows (1975:69-70). 74 more directly from the state, social stratification, and the control of property by men. With the rise of the state, a male monopoly of weapons and men's freedom from child care allow them to perform specialized economic and political roles. Access to these roles gives ruling class men control over some other men and over women in general (1975:70). Gough's argument about the effect of social stratification on sexual inequality is not totally applicable to this study. First , Gough is speaking of the differences between hunting and gathering societies with no degree of social stratification and the state which is characterized by a degree of stratification greater than obtained in the societies I investigate. Thus my sample is not characterized by the level of stratification indicated by Gough. Nevertheless, it will be interesting to observe whether degree of stratification remains correlated with sexual inequality where the degree of stratification is less than the one indicated by Gough. Unlike the other authors reviewed to this point, Friedl (1975) reasons that increased social stratification is beneficial for women, at least in horticultural societies with patrilineal descent (1975:98). Friedl does not deal with the question of stratification in terms of hunting and gathering societies; she assumes that these societies are not characterized by social inequality. 1 In these societies, although men and women have access to subsistence resources, to roles of leadership, and to autonomy, men have greater control over the primary sources of power - extradomestic distribution and the exchange of 1 She observes that the hunting and gathering societies of the Northwest Coast have some degree of stratification but she does not discuss the issue further (1975:15). 75 valued goods and services (1975:98). They have this greater access to these sources of power because of the hunt; men in hunting and gathering societies have a monopoly over individual and small group hunts. M e n , therefore, have the opportunity for large extradomestic exchanges of food, a source of power not available to women. Friedl notes, however, that if women contribute to subsistence and are involved in extradomestic distribution and the exchange of valued goods and services, then there are fewer differences in power and autonomy between men and women (1975:13s).1 Male dominance is greatest in hunting and gathering societies where hunting is the sole source of food; equality is greatest where women and men work together in subsistence activities (1975:31-32). 2 Friedl's views regarding the positive effect of greater stratification come into play in her study of those societies based on horticulture. Friedl maintains that in those horticultural societies where all men have an equal There is some confusion about the type of power men and women have. Originally, Friedl uses extradomestic distribution to measure power (i.e., as a de-pendent variable). A t the same time, however, extradomestic distribution is Friedl's independent variable (1975:135-36). 2 This aspect of Friedl's model is suspect for a number of reasons. First , Whyte's empirical results indicate that although women have less power in the areas of kin group leadership and less informal influence in hunting and gathering societies, women have greater domestic authority, more ritual solidarity, more control over property, less ritualized fear from men, and perhaps, more val-ue placed on their lives. Moreover, these associations are strongest where there is hunting of large and medium-sized animals (generally assumed to be the domain of men), rather than small game (assumed to be the domain of men and women, 1975:127-29). Second, m y data indicate that even in those societies where hunting is said to be almost exclusively done by men, women still control the food in a number of instances (e.g., Sharp 1981 on the Chipewyan). T h i r d , m y data indicate that, even in those societies where women's access to certain types of power is less than men's, men and women work together in some subsistence activities and help each other with certain other subsistence activities (e.g., Barnett 1955:89, Drucker 1950:247, 262 on the Nootka, Drucker 1951:38, 39, Garfield 1939:270 on the Tsimshian, Murdock 1934a:251 on the Haida , and Boas 1916:400, 404). This contradicts Friedl's hypothesis that where men and women share subsistence activities, there will be more opportunity for equality. opportunity to engage in redistribution and where redistribution is reciprocal (i.e., one obtains as much in return as one gives), women have less power and authority (1975:62-64). This is because men have greater opportunity to organize and control extradomestic exchange and, hence, political and religious affairs. In these societies men control the allocation of land (men are the ones who clear the land, since uncleared lands are located on the borders of their property where warfare is most likely to break out). This ability to al-locate land gives men a greater opportunity to accumulate goods for redistribution; all men are free to compete for the status of "big man" and the economic and political leadership associated with it. According to Friedl , this state of affairs exists regardless of the type of descent system. In nonegalitarian horticultural societies (i.e., those societies where the distributor has an institutionally higher rank and receives more than he or she gives), women have a greater opportunity to control and organize extradomestic exchange and political and religious affairs. In these societies, only those who inherit the appropriate political and ritual positions of power can organize extradomestic activities (1975:49). Friedl asserts that in these societies women, through inheritance, have access to political power, trade, movable wealth, and the allocation of land. Thus , women can engage in extradomestic exchange. This gives them autonomy and the ability to have control over others; "among horticulturalists with nonegalitarian systems of economic distribution and political control, women have access to inherited posi-tions of authority and power far more often than they have the right to enter into open competition for them" (1975:98). Although the hunting and gathering societies in this thesis are not 77 comparable to the horticultural societies of Friedl's model, it will be of interest to learn whether the ability of women to inherit rank in ranked hunting and gathering societies is, in any way, associated with extradomestic economic power and political power not available to women in egalitarian societies. Like Leacock and Gough, Schlegel (1977a:5) states that there is a nega-tive relationship between the degree of social stratification in a society and sexual inequality. Nevertheless, she is uncertain about the nature of this relationship. In recent years there has been considerable discussion about the relationship between social stratification, that is, a social class sys-tem, and sexual inequality. While it is true that most classless foraging societies appear to be more sexually egalitarian than m a n y other more complex societies, and socially stratified societies as we know them tend to be sexually inegalitarian, the data do not sup-port a simple or direct correlation. Classless societies at the 'middle range' of complexity - horticultural, herding, or fishing societies - can go in either direction . . . A n d in looking at European stratified societies over the past 100 years, we see real gains . . . in avenues of participation and achievement open to women (1977a:4-5). 1 Schlegel (1977a:26) posits that the correlation between egalitarian societies and sexual equality m a y be due to the greater flexibility in the division of labour among foragers than among more complex societies. In foraging societies women hunt as a secondary subsistence activity while men gather as a secondary activity. In these societies, even with a domestic mode of production that has a division of labour based on sex, women are freer from the constraints of child bearing and rearing than are women in societies 1 L ike Friedl , Schlegel maintains that a woman's rank may positively affect her status and her access to power (1977a:20). 78 where economic and social success for a woman depends on the labour and loyalty of her children and her affines (1977a:26). In these latter societies, a woman's priority is procreation, rather than production. She thus has less opportunity to participate in non-domestic spheres of activity. This view is of interest for my study because data for m y ranked societies indicate that this flexibility in the division of labour existed among the ranked societies of the Northwest Coast; women fished although men were the primary fishers, and men aided women in the gathering of marine vertebrates, etc. (e.g., Barnett 1955:89, Drucker 1951:247, Osgood 1958:259). Thus , according to SchlegePs model, the egalitarian societies of m y study should be characterized by sexual equality. The ranked societies m a y "go in either direction." In his cross-cultural study of the status of women in preindustrial societies, Whyte (1978:38-41), considers the relationship between women's status and the degree of social stratification within the more general context of cultural complexity and differentiation. Whyte's concepts of complexity and differentiation refer to the existence of more specialization in roles and organizations as well as more varied and efficient technology in modern industrial societies than in preindustrial ones (1978:38). Thus , some societies are more complex than others. Whyte provides an example. In present day hunting and gathering societies, no political organizations exist apart from kin groups such as lineages and clans; more complex societies have non-kin based political organizations which are specialized. He stresses, however, that 'simple' and 'complex' are used only in this technical sense; their use "does not imply that people in simpler societies are simpleminded, or that these simple societies do not have other kinds of complexities of their own" (1978:39). Whyte uses the concept of social complexity to test whether 'civilization' and social evolution affect women's status, a major topic of concern in the literature on women's status. He argues that one can deal with questions of social evolution in his model through the thirteen indicators he uses to measure societal complexity. H e notes that he does not have a true evolutionary sequence in his sample, since data on each culture refer to one point in time. Nevertheless, some of the indicators of societal complexity "emerged early in the scale of human evolution (pottery-making, nomadic bands)" while others "appeared relatively late (the plow, settled towns and cities)". Therefore, his indicators of social complexity serve as his "proxy" in discussing social evolution (Whyte 1978:17, 172). The degree of social stratification present in a society is one of the thirteen indicators of societal complexity used by Whyte. Stratification is measured along a four point scale: largely absent; differences in access to resources, but no distinct classes; dual stratification - elite and commoners; and complex stratification - three or more classes. Whyte postulates that in more complex cultures, women will have lower status than women in less complex cultures. His results (1978:153-56) support the hypothesis. Indeed, this relationship was one of the few that were supported by his data. Specifically, the results show that in simple societies: 1 1) the division of authority between husbands and wives is relatively equal, 2) women engage in 1 These are nomadic or semi-nomadic hunting and gathering societies with small , autonomous political units where: 1) any existing leadership is exercised by the family head or lineage chief, 2) any social differences are based on age, sex, or personal skills, 3) hunting and gathering provides the subsistence base, and 4) there is little inheritable property. their own activities, 3) women are not more sexually and maritally restricted, 1 4) women have little informal influence, 2 5) women are not especially disadvantaged in their control of property , 3 and 6) unexpectedly, there is perhaps less of a tendency for women to share in kin group leadership. In the sample's complex cultures, 4 Whyte found: 1) women tend to have less authority in the home, 2) women tend to lack joint rituals and solidarity with other women, 3) women have more unequal restrictions on their sexual lives, 4) "perhaps" women have less control over property, and 5) women tend to exercise influence informallj ' , 5 often through their access to men who hold positions of formal authority. According to Whyte , "the pattern of associations in table 12 [his indicators of complexity] is sufficiently general that it is hard to say what specific aspect of complexity [i.e., politics, technology, stratification, or some other area] is most important" in affecting the lot of women. H e considers they are all interwoven and, in some general fashion, all affect women's status (Whyte 1978:139-40). Nevertheless, Whyte does assert that perhaps stratification and the nature of family farming - particularly land ownership -are features which place a premium on controlling women's sexual and 1 H e observes this may be due to the absence of strict sexual restrictions for anyone (1978:155). 2 Whyte reports this may be due to the absence of positions and roles from which women are barred (1978:155). 3 There exists little property for anyone to monopolize (1978:155). 4 These are cultures with intensive agriculture, advanced tools and techniques, large and stable settlements, large political units with a hierarchy of specialized offices, and stratification involving a complex set of classes with differential con-trol over private property and other resources. 5 The presence or absence of informal influence was gauged as a response to the question, "Does the ethnographer(s) say that women have more informal influence than the formal norms of the society would make it appear?" (Whyte 1978:88) 81 marital lives, and hence, on property inheritance. In addition, the presence of these indicators causes the development of conjugal family estates and family farms. This sets the family apart from wider sodalities, including those with other women. Finally, the presence of many specialized roles outside of the family, which are dominated by m e n , 1 complex political hierarchies, and an ideology which legitimates male control of these structures, m a y also play a part in limiting women's domestic authority (1978:172). In terms of societal complexity, Whyte's results (1978:139) indicate, the most important evolutionary transition for women is between societies of intermediate and high complexity . . . rather than be-tween societies of low and intermediate complexity. Thus , for ex-ample, the shift from subsistence based on hunting and gathering to one based on shifting or extensive agriculture (horticulture) seems to be less important than the further shift to intensive, plow agriculture. The ranked cultures considered in this thesis lie between egalitarian and complex societies in terms of Whyte's levels of societal complexity. If m y re-sults support Whyte's findings, they should indicate greater sexual inequality than Whyte observed in his egalitarian societies, but less inequality than he observed for his complex societies. If I obtain results similar to Whyte's, however, then Whyte's model is inexact, since m y results will have indicated that even a low level of societal complexity is important for women's access to power and autonomy. 1 Whyte admits that he cannot explain why men control these and falls back on the statement that it may be due to women's child care role (1978:163); that is, he turns to biology to provide an explanation. 82 2. The Independent Variable Having reviewed a number of models which predict a relationship be-tween the degree of social stratification and women's access to power and autonomj', I shall now describe the manner in which degree of social stratification is defined in this thesis. On the basis of my sample, degree of stratification was divided into three categories: 1) ranked, 2) intermediate, and 3) egalitarian. Originally, I proposed two categories - ranked and non-ranked, or egalitarian. However, six of the cultural groups in my sample seemed to be in transition between these two types. Although these cultural groups could have been arbitrarily assigned to one category or the other, an intermediate category was created on the grounds that, by using three categories, I would retain more detailed data; by using two, I would lose some information. In order to code a cultural group as ranked, intermediate, or egalitarian, a number of indicators of degree of social stratification were employed. Indicators were chosen which measure differential access to power, privilege, and material rewards among groups of people, but not between the sexes.1 Eight indicators were chosen. The presence of all eight in a cultural group clearly denotes a ranked society, the absence of all eight, an egalitarian society. Where the trait is said to be recent, or where it is in the process" of becoming a norm in the culture, the trait is coded intermediate. The indicators are: 1) the presence of some tendency to relieve persons of high rank of 1 For example, I had originally included one indicator, "Are some people excluded from decision-making in community affairs?" Since women as a group may be excluded from decision-making in community affairs, this indicator was rejected. 83 some or all participation in some phases of work (Fried 1967:130-31), 2) the presence of hereditary s lavery 1 (Aberle 1961:694, 2 Driver 1969:332), 3) the presence of inherited ranks or statuses (Driver 1969:334) (Technically, one should speak of degrees rather than actual presence or absence, since, in some of the groups studied, ethnographers disagree about whether one really inherits a rank or whether one's birth simply entitles one to strive for high status. Nevertheless, even in the second case, those born without this potential would have little chance to gain high status. Since there is some differentiation occurring on the basis of inheritance, the trait is considered present in these cases.), 4) the presence of certain goods consumed only by some segments of the population, and seen as evidence (by the ethnographer or informants) of high rank (Sahlins 1958:8), 5) restrictions on intermarriage between certain segments of the population (Gough 1961:482, Sahlins 1958:9), 6) the presence of differences in the kind and/or elaboration of life crises rites among some segments of the population (Sahlins 1958:9), 7) the presence of enhanced marriage payments among certain segments of the population (Gough 1961:477), 1 Lane (1953:201) argues that although "slaves" may be stigmatized for a time, if they are later considered fully participating and free individuals with no stigma attached to them, they are not true slaves. M y criteria for discerning the pres-ence of slavery is, therefore: 1) are the children of such persons considered slaves? and 2) is there a stigma attached to the non-slave if a non-slave marries a slave? 2 His data are based on Murdock's World Ethnographic Sample. 8) the presence of ownership/control of fixed economic resources such that some segments of the community are refused access to it (Sahlins 1958:7) (This means that members of the community who are not mem-bers of the kin group controlling the resources do not have automatic access to the use of those resources. They will have access to their own kin group's resources, however).1 Other indicators could have been included in this list, for instance: the presence of craft specialization (Driver 1969:331, Fried 1967:62), the prevalence of redistribution rather than reciprocity as the primary form of exchange (Fried 1967:117), or the size of the largest continuously functional political unit (Fried 1967:118, 174). In the end, the eight indicators listed were chosen on the basis of their clarity and their applicability to the data which were available. If the majority of the indicators were present in a cultural group, the group was coded as ranked. If the majority of the indicators were absent, it was coded egalitarian. If the majority were in transition, it was coded intermediate. In those cases where two categories had equal numbers of indicators, (i.e., one case did exist where two indicators were coded present, three were coded intermediate, and three were absent), the case was coded as the more stratified of the two tied categories (i.e., in this case intermediate rather than egalitarian). In those situations where no data were available on an indicator, two choices were possible. In the cases where data on one or two indicators 1 If some members of the society have no access to any resources, the society would be stratified and not ranked, according to Fried's classification (Fried 1968:469). 85 were missing, but the remainder of the data clearly indicated the degree of stratification of the society, the missing indicators were unnecessary and the culture could be coded without the missing data. In those cases where there was a tie or where the number of present and absent were similar, the missing information became crucial. Driver (1975:9), facing a similar problem, decided to code a trait as present if: 1) one or more neighbouring units have the trait and 2) if he knew of no reason why it should be absent (if most or all of the surrounding groups did not have the trait, this is a good reason for its absence). These two criteria were used to decide whether a trait should be coded present. 3. The Hypothesis The following hypothesis is tested in this thesis: there is a relationship between the degree of stratification in a society and differences in the access of men and women to power and autonomy. C. DESCENT AND RESIDENCE 1. A Review of the Models The other independent variables to be tested are type of descent and final postmarital residence pattern. These two will be discussed together since most of the authors who predict a relationship between sexual inequality and these independent variables have considered descent and residence simultaneous-iy. M y interest in the effect of matrilineality on women's access to power 86 was stimulated by the ongoing and largely untested debates in the anthropological literature about the existence of a relationship between matrilineality/matrilocality and women's status. In this section, I review some of the models which deal with the topic of sexual inequality or women's status and type of descent and/or residence pattern. In the next section I describe the variables I use in this thesis. Interest in the relationship between matrilineal descent and sexual inequality seems to have originated in the nineteenth century and has been a matter of contention ever since. Nineteenth century evolutionists such as Morgan (1974)[1877] and Engels (1972)[1884] believed that matr iarchy 1 was a universal stage of cultural evolution, which was displaced by patriarchy, based on patrilineal descent. Morgan (1974:5) claimed that early cultures were organized socially on the basis of kinship ties. Families had not existed in the earliest societies; rather, marriages occurred communally among groups of men and women (1974:393). Later , through the progressive restriction of possible marriage partners, the gens had developed, allowing social organization on the basis of kinship ties. Because one's mother, but not one's father, could be traced with biological certainty, descent was traced through women (1974:67). A t this stage of social evolution, women had equality with men. Morgan contended that societies based on political organization and monogamous families could not develop until kinship ties gave way to territorial boundaries as the organizing principle in society. This could occur only after the development of private property and urban centres (1974:263, 477). These, in turn, could 1 Matr iarchy is rule by women based on the indisputable legitimacy of children which comes from reckoning descent through the mother (Leacock 1972:34-35). only arise when agriculture and the domestication of animals became sufficiently productive (Morgan 1974:263-64, 272). According to Morgan, prior to the shift from the gens to political organization, matrilineal descent shifted to patrilineal descent (1974:63-64). Thus, political organization, patrilineal descent, and later, monogamous marriage, arose because of the development of private property and urban centres, which came with agriculture. Morgan asserted that patrilineal descent arose as a result of the devel-opment of private property in the form of land ownership and cattle. With the development of private property, the concept of inheritance took on meaning. A man would desire to transmit his property to his own children, rather than to members of his own, and, therefore of his mother's, gens (1974:363, 366). Thus , patrilineal descent and, later, monogamous marriage were the results (Leacock 1974:IIxv). The position of women was adversely affected by these changes from gentile to political organizaton, from matrilineal descent to patrilineal descent, and by the emergence of the monogamous family (1974:477-78, 485). Engels was also interested in the question of matrilineal descent and women's equality. He maintained that female subordination was tied to male control of wealth, initially in the form of domesticated animals (1972:117-120). Like Morgan , Engels claimed that the development of personal property led to the overthrow of matrilineal descent, which had been the basis of women's equality with men (1972:113). Matril ineal descent was overthrown because men wished to have their own children inherit their property (1972:119-20). Thus , matrilineality gave way to patrilineal descent and to monogamous families. Wives became domestic workers for their husbands, dependent upon 88 husbands and their property, rather than having their own rights to communally owned matrilineal kin group property as they once had. Differential ownership of wealth led to economic inequality in the family and between the sexes in general (Engels 1972:120, 129, 137-39; Leacock 1972:29). Although evidence of matriarchy, as defined by these nineteenth century scholars, has never been found, as Whyte (1978:7, 34) points out, the pres-ence of matrilineal descent continues to be treated as an indicator of high status for women by some authors (e.g., Nielsen 1978:16), while others deny that this is so (e.g., LeVin e 1966:186-87). When evolutionary theory came under attack after the turn of this century, anthropologists such as Bardis (1963, in Whyte 1978:7), Hobhouse (1924, in Whyte 1978:7), and Lowie (1920, in Whyte 1978:7) argued that descent rules and postmarital residence had little effect on the status of women, and that in matrilineal societies, men still held all the authority; 1 "'where it [matrilineal descent] most flourishes, it is perfectly possible for the position of women to be as low as the greatest misogynist could desire . . . The woman is not necessarily any better off because she is ruled by a brother in place of a husband'" (Hobhouse in Whyte 1978:7). Anthropologists continue to have assumptions about the effect or lack of effect that matrilineal descent has on women's position in society. Oberg (1973:38), for example, states that the Tlingit world was a thoroughly masculine one, even with matrilineal descent and inheritance (1973:38). 1 This assumption is still held by a number of authors (e.g., Schneider 1961:7; it is interesting that in the same work, Gough provides evidence that women hold authority in some societies, 1961:519-20). • M c K e n n a n (1965:52), on the other hand, observes that because of matrilocal residence, the roles of Chandalar Kutchin women are not as inferior as they seem. The models of six scholars are reviewed at this point in order to determine the manner in which matrilineal descent and matrilocal residence are predicted to affect women's access to power. These six scholars are Friedl (1975:72), Gough (1975:55), M a r t i n and Voorhies (1975:7, 184), Nielsen (1978:37-39), Schlegel (1972:141 and 1977a:29), and Whyte (1978:132-34). Friedl (1975) argues that descent and residence affect women only in horticultural societies. Specifically, because they affect inheritance, succession, and rules of residence, rules of descent affect women's domestic authority. Descent and residence have no effect in hunting and gathering societies, however (1975:14-15). In these societies there is no accumulation of commodities. Consequently, inheritance is of no major importance. In addition, the lack of accumulated property precludes the development of differential wealth among either men or women and, therefore, the development of differential economic or political power and sexual inequality. Since a number of the hunting and gathering societies used in this thesis are indeed characterized by differential wealth and differential economic and political power, an examination of Friedl's model as it pertains to horticultural societies may be of interest. Friedl asserts that descent and residence do affect women's status in horticultural societies. Friedl reports that inheritance and succession (i.e. the descent system), especially as these are combined with residence rules for the married couple, have "weighty consequences" for the domestic relations between the sexes and for relationships with extended kin in horticultural societies (1975:66). Patterns of strains and tensions, devotion, and loyalty between spouses, siblings, and parents and children will differ on the basis of the descent sys tem. 1 She observes that the patterns of relationships affected by inheritance and succession go beyond the kin group itself. Expectations of loyalty and antagonisms will also influence attitudes about the sexes/genders in the society. In other words, the relations between men and women in general are also affected by residence, inheritance, and succession (1975:66-67). In her comparison of the effects of patrilineal and matrilineal descent, Friedl (1975:67-69) argues that although in-marrying women are valued for their procreative role in a horticultural, patrilineal society with patrilocal residence, this does not mean they have a position of power. The bride is a stranger among people who are tied by blood and marriage. She is de-pendent on a husband whose pr imary ties are to his kin group and to his mother. Faced with these strains, the bride has little domestic power. The greatest opportunity for overt dominance by husbands over wives exists in these societies. In patrilineal societies where women cannot be traders (trade provides one means whereby women can accumulate economic resources) and where public office is not inherited, women have little access to power unless: she has kin of her own nearby to support her or she has economic resources of her own through the control of the fruits of her own labour. A woman also may have informal power through her sons; this latter source of power is available only later in her married life, however. A s mentioned earlier in this chapter, this picture of patrilineal societies 1 Due to constraints in space, Friedl considers only unilineal descent systems. Moreover, she does not consider avunculocal residence. 91 with patrilocal residence is modified in more stratified societies (1975:62-64, 69). 1 In non-egalitarian societies, where economic and political power are often inherited with one's rank, women may have access to political power (1975:49). In addition, a woman may have economic and domestic power because her position as head wife of a chief or king can involve significant control of major resources and the supervision of the household (1975:69). Nevertheless, Friedl concludes that even in these more stratified societies, the ordinary wife of a common man has a position similar to the one described for women in more egalitarian patrilineal societies. The situation of women in matrilineal societies is quite different (1975:70-74). Women have a greater chance for domestic equality with men than do women in patrilineal societies with virilocal residence (1975:72). M e n still allocate land, unite for war and defense, and have the main ritual and political roles, but relations between the spouses are less tense - partly because they matter less. A woman is not perceived as her husband's property; she continues to produce for her own kin group, and he produces for his (1975:72). A s a result, women in matrilineal societies tend to have domestic equality with men. To summarize, Friedl maintains that descent and residence have no effect on women's access to domestic power in hunting and gathering societies, but these factors do in horticultural societies. If there exists some stratification in a patrilineal society, some women may have domestic, economic, and political power; otherwise, women in patrilineal societies with patrilocal residence are 1 Friedl indicates that social stratification is a characteristic of horticultural societies, but not of hunting and gathering societies, with the exception of the Northwest Coast area (1975:15). 92 subordinate to men. O n the other hand, women have equality in the area of domestic authority in societies with matrilineal descent and matrilocal residence. Two aspects of Friedl's model are investigated in this thesis. First , she predicts that patrilineal descent and patrilocal residence have a negative effect on women's power in horticultural societies. Second, she suggests that some degree of stratification combined with patrilineal descent gives women greater political power and domestic authority, perhaps more than women have in societies with matrilineal descent; that is, degree of stratification is predicted to destroy the negative relationship usually observed between women's power and patrilineal descent and patrilocal residence. It will be interesting to observe whether Friedl's model for horticultural societies, whether egalitarian or ranked, can be extended to the egalitarian and ranked hunting and gathering societies of this thesis. The next author reviewed is Gough (1975). In "The Origin of the Family", Kathleen Gough (1975:54) argues that women do have greater autonomy in matrilineal societies; where property, rank, office, and group membership are inherited through the female line, it is true that women tend to have greater independence than in patrilineal societies. This is especially so in matrilineal tribal societies where the state has not yet devel-oped, and especially in those tribal societies where residence is matrilocal. According to Gough, women have greater respect, dignity, and freedom in all pre-agricultural societies (i.e. hunting and gathering and horticultural societies), 1 regardless of descent and residence. Even in these societies, 1 Gough suggests the development of state level societies is due to the 93 though, "matrilocal societies, with matrilineal inheritance, offer greater freedom to women than do patrilocal and patrilineal societies of the same level of productivity and political development" (1975:75). The direction of the relationship is very clear in this model. Although the agricultural revolution (and the subsequent rise of the state) is the most important factor in sexual equality and inequality, 1 matrilineal descent with matrilocal residence are associated with greater independence for women. A n important point should be noted about the models presented by Friedl (1975) and Gough (1975). Both consider the effects of two independent variables at the same time. They predict that descent and residence work in conjunction with the development of greater societal complexity (however that is defined - as greater social stratification or the rise of the state). I shall return to this topic in Chapter 5. Like Friedl , the next authors assert the effect of descent and residence are much greater in horticultural societies than in hunting and gathering societies. O n the basis of their analysis of fifty-one foraging societies, Mart in and Voorhies (1975) argue that there is "no simple correlation" between descent and the relative social status of men and women among foragers (1975:187). There are differences, however, in the way that men and women are regarded (1975:187-88). Nevertheless, this statement is not empirically supported. The only relationship which is tested in their sample of hunting and gathering societies is the relationship between premarital sexual autonomy continued . . . development of agriculture. Changes accompanying these developments include the growth of individual and family property in herds, the growth of durable craft objects and trade objects, and the growth of stable, irrigated farm sites or other forms of heritable wealth (1975:75). 1 Whyte claims his results support this hypothesis (1978:139). 94 and descent and residence. Their results indicate that patrilineal or patrilocal foragers exert a greater degree of control over the premarital sexual lives of young women (1975:188); the authors do not test whether the sexuality of young men is also controlled, however. Mart in and Voorhies argue that among more complex societies (i.e., horticultural societies), descent and residence have a more widespread effect. The key to power is the localization of related individuals since this is a crucial determinant of the assignment of food production and distribution rights in the kin groups. The right of women to determine how the products of their labour will be allocated to others determines women's access to power. That is, the residence pattern strongly influences women's economic power and, therefore, "their ultimate status in the community" (1975:228-229, 231). Common matrilocal residence (and matrilineal descent) 1 give kinswomen this power to control resources and wealth since, in these societies, women form the locus of social groups (1975:228). Hence, women define social and political relationships among women and men. Although positions of public authority are consistently assigned to elder males in matrilineal societies, in each descent group actual power is often concentrated in the hands of senior women. These women wield "considerable influence" in decision-making within the community (1975:225). Women in societies with matrilocal residence have much greater latitude in extra-domestic activities and power than do women in many other societies. In contrast, in those societies where women leave their own kin at 1 Mart in and Voorhies (1975:184), among others, contend that residential patterns, standardized over a number of generations, give rise to the patterns of descent. marriage and live the rest of their lives with strangers, males are highly dominant. A s M a r t i n and Voorhies explain, it follows that any "mode of postmarital residence that disperses related women jeopardizes their collective relationship to land and production (1975:228)." This is the case with patrilocal residence and patrilineal descent. Patrilineal descent and patrilocal residence are postulated to have a negative effect on women's access to wealth, social prominence, and power in horticultural societies (1975:235-41). Although women remain the primary cultivators in these societies, just as they are in matrilineal cultures, the nature of the productive and distributive groups is different. Women in these societies relate to land and other resources as a consequence of marriage rather than birth (1975:235). Consequently, no local aggregates of women exist, and the focus of their collective labour is destroyed. In patrilineal societies and patrilocal societies, women are producers for, and domestic appendages of their male spouses. H a v i n g left the security of her kin group to live among strangers, a woman's only road to advancement is through her sons. In patrilineal societies, women have fewer opportunities to achieve high status or to obtain power in the community (1975:238-241). Nevertheless, if a woman can estab-lish economic independence outside the household, access to positions of influence are possible for her, even in patrilineal societies (1975:238). M a r t i n and Voorhies also consider the question of avunculocal residence. Although it is associated with matrilineal descent, this pattern of residence "almost inevitably results in the transfer of some or all production and distri-bution rights to males who reside together (1975:22s)."1 A s a result, 1 The authors do not explain why this transfer is almost inevitable. 96 avunculocal residence shares many features with patrilocal residence and patrilineal descent (1975:228). The relationship between descent/residence and sexual inequality is primary in this model. The authors postulate that, "Whereas the position of females is quite variable in patrilineal societies, it is almost universally high in matrilineal ones" (1975:224-25). In terms of testing for a relationship between descent and/or residence and women's access to power in hunting and gathering societies, the model presented by M a r t i n and Voorhies adds a new dimension - the effect of avunculocal residence on women's access. Another author, Nielsen (1978), presents a model similar to that advanced by M a r t i n and Voorhies. Nielsen (1978:24-29) views the effect of descent and residence as the pr imary factor in women's status. 1 L ike Friedl and M a r t i n and Voorhies, Nielsen directs her attention to horticultural societies. 2 Nielsen argues that descent has a greater effect in horticultural societies than in hunting and gathering societies because the sedentary lifestyle of horticulturalists allows for resource accumulation. Thus , inheritance rules and those from whom one inherits take on great importance. Nielsen focuses solefy on women's status in her study, having made the assumption that men's status is not affected" by the kinship structure of a 1 Status is measured in terms of access to the rewards associated with roles, namely, power, prestige, material rewards, and ps3'chological gratification (Nielsen 1978:10-19). 2 She uses type of subsistence (and, therefore, presumably the level of productivity) as an index of cultural development, rather than relations of production (as Leacock does), or societal complexity (as Whyte does). Nielsen characterizes hunting and gathering societies as having little surplus food or wealth, and therefore, as having little opportunity to develop status differences. society (1978:27); 1 "Perhaps it is more accurate to say that women's status is affected by kinship structure, while men's status remains constantly high. In all societies where public authority exists, it is in the hands of the men ." 2 The subsistence base of the society determines the division of labour which, in turn, determines residence. The residence pattern is essential in determining which sex will be the household head and, hence, the owner, collector, and distributor of food and other resources. In addition, the residence pattern affects the type of descent system. Type of descent and control over resources directly influence "sex status" (1978:37-39). Nielsen associates the various patterns of women's status found in horticultural societies with three kinship patterns: 1) matrilineal descent with matrilocal residence, 2) matrilineal descent with avunculocal residence, and 3) patrilineal descent with patrilocal residence. E a c h of these three .patterns of descent and residence has a different consequence for women's status. According to Nielsen, matrilineal societies are collective in nature. Generally, women form the collective work groups and collectively distribute food and other resources. For this reason, no one individual can accumulate wealth. The matrilineage as a whole controls food resources and this gives female household heads a "definitive source of power". Nielsen claims that women's high status is confirmed by their importance in the kinship structure and in their high degree of personal and sexual freedom (1978:29). Matril ineal descent with avunculocal residence, on the other hand, like 1 For example, she claims that in matrilineal societies brothers have legal responsibility over the household because they have authority over sisters and their sister's children. In addition, she claims that even in societies with matrilineal descent, men and not women hold society-wide positions of authority (1978:27). 2 Nielsen seems to equate high status primarily with roles of public authority. patrilineal descent with patrilocal residence, is associated with surplus production (in contrast, societies with matrilineal descent and matrilocal residence are associated with subsistence production). Nielsen asserts that the change in residence pattern is significant for women because it usually entails a change in the control over the distribution of land and products. In these societies, the household head is a member of the husband's lineage, rather than the wife's lineage. Because descent is still matrilineal, however, a young wife can still seek protection and help from her own family. This provides a young woman with resources not available to women in non-matrilineal societies (1978:29). The situation is different in patrilineal societies with patrilocal residence. Women still do the majority of the productive work, although there may be increased male participation. A wife, however, is a stranger in her husband's home and is under the authority of her mother-in-law. Food collection is no longer a collective activity done by a group of related women. A wife's children, the fruits of her labour, and all resources belong to her husband's kin group for whom she works, and she is separated from her own kin group (1978:32). In these societies, a woman has little opportunity to gain status. In essence, then, the relationship postulated between matrilineal descent, matrilocal residence, and women's access to roles of power is a positive one. Avunculocal residence, it is postulated, will have a negative effect, as do patrilineal descent and patrilocal residence. Like Friedl , Schlegel (1977a) maintains that in many preliterate societies, kinship is one of the main channels (beyond the household) through which bonds are created among people, "loyalty is activated", and "social action is mobilized (1977a:29)". Control over corporately held property, the principles of inheritance of resources and positions, the right to form alliances through marriage - all are features of kinship and all can support or modify the power and authority held by women and men (1977a:29). Schlegel (1977a:28-29) argues that residence rules determine which sex forms the nucleus of the household. Ties of loyalty and obligation will tend to be strongest among members of this core group, and, other things being equal, "power or authority would reside with the sex that brings in husbands or wives, as the case may be." She cautions, however, that other things are never equal, so "there is no simple equation between matrilocality and female domestic power, on the one hand, and patrilocality or avunculocality and male power, on the other (1977a:29)." In an earlier work on matrilineal societies, which compares the character-istics associated with husband domination and brother domination in domestic affairs, Schlegel postulates that the type of descent system characterizing a society will affect the way people organize their activities and the way people define their attitudes about the sexes. In addition, the type of descent sys-tem affects the way people behave toward members of their own and the opposite sex (1972:140-41). She concludes, E v e n where female autonomy is low, m y general impression is that it is higher in matrilineal than in patrilineal societies within any given culture cluster . . . It is probable that the importance of the woman as the linking factor in the descent group gives to womanhood a dignity that may be lacking in societies which do not have this belief. In psychological terms, this is saying that in matrilineal societies there is a cognitive set toward the importance of women that has an effect in mitigating male dominance (1972:140-41). 100 Thus , like Friedl , Gough, Mart in and Voorhies, and Nielsen, Schlegel claims there exists a correlation between descent and women's access to power and autonomy as well as between residence and women's access to power and autonomy. Like these authors, she reasons that matrilineal descent and matrilocal residence are positively associated with women's power and status, while patrilineal descent, patrilocal residence, and avunculocal residence are neg-atively associated with women's power and status (1972:135, 141). The last author to be reviewed is Whyte (1978:33). Us ing a sample of ninety-three cultures, he tests two hypotheses relevant to this discussion: 1) in cultures with matrilineal descent, women will have higher status than women in other cultures, and 2) in cultures with matrilocal postmarital residence rules, women will have higher status than in other cultures. Whyte suggests that matrilocal residence affects women's status because women continue to live with female kin; avunculocal residence is therefore of no added advantage to women (1978:34-35). Whyte's results (1978:132-33, 171) indicate a persistent relationship be-tween descent and/or residence and some of his dependent variables; women in matrilineal and matrilocal societies have more property control than women in other cultures. (Property control is an index consisting of five variables: inheritance rights, dwelling ownership, control of fruits of male labour, control of fruits of joint labour, and control of fruits of female labour 1978:98). Somewhat weaker relationships are supported for matrilineal and matrilocal residence and more domestic authority for women, more ritualized female solidarity, more equality in sexual restrictions, and more value placed on women's lives. Thus , Whyte obtains some empirical support for his hypotheses 101 that matrilineal descent and matrilocal residence are positively associated with higher status for women. 2. The Independent Variables Having completed the review of the manner in which matrilineal descent and matrilocal residence are expected to affect women's status and power, I will describe the independent variables, type of descent and postmarital residence pattern, as they are defined in this thesis. For the purposes of the analysis, descent was divided into two categories, matrilineal and other. Clearly, descent may be divided into finer categories than this. However, this simple dichotomy allows me to test all of the propositions concerning the effect of matrilineal descent on women's status that have been reviewed in this thesis. If an individual is assigned to a unit of consanguineally related kin on the basis of a descent line traced through females, the society is coded matrilineal (Schneider and Gough 1961:2). If the individual in a society is assigned to a unit of consanguineally related kin on the basis of a descent line traced through males (patrilineal, Schneider and Gough 1961:2), or, if a person is associated with a group of close kin irrespective of genealogical ties (bilateral, Murdock 1965c: 15), the society is coded other. Coding descent was a straightforward task. The coding of residence was not as straightforward. In a number of cases, there were multiple contradictions in the data (e.g., the Tanaina), the result of factors such as the manner in which authors classify residence and changes in the residence pattern over time. These factors will be discussed in more detail before I 102 present the residence categories used in this thesis. Contradictions in the classification of a cultural unit's residence pattern may be the result of the manner in which the authors define and classify residence. Unfortunately, most ethnographers classify the residence pattern without explaining their classification system. It is difficult, therefore, to ascertain the factors that led the author to classify as he or she has. One must ask if the author has incorporated information on the type of descent system into his or her classification. Residence may be described onry with reference to one member of the bridal couple. Thus , patrilocal residence is often defined as residence with the groom's family; matrilocal residence is de-fined as residence with the bride's family (e.g., Nielsen 1978:167, 169). O n the other hand, what if residence is with the groom's mother's people? Some may still consider this to be patrilocal residence while others m a y not. Those in the latter category might argue that, because residence rights result from membership in a kin group, the descent system and the kin member sponsoring the couple should be considered when classifying residence (e.g., Aberle 1981:27, Goodenough 1956). One example of the contradictions inherent in these two approaches is found in Fischer (1958). Fischer (1958:508-09) presents the example of a society (Truk) where descent is matrilineal, but residence is with the husband's kin group. He notes Goodenough (1956) had defined this residence pattern as avunculocal; Fischer argues that residence is actually patrilocal. According to Fischer (1958:509), Goodenough's reasons for labelling residence as avunculocal in this case were: 1) the culture was matrilineal and 2) in matrilineal societies, males own the property and sponsor the residence of 103 newly married kin. The residence link is, therefore, uncle-nephew. Fischer, observing that uncles do not even reside with their nephews in this particular society, maintains that descent should be ignored in classifying the residence pattern. Thus , he classifies the residence pattern as patrilocal. (In fact, if descent is matrilineal and the groom takes his wife to his kin group, the person sponsoring the husband and his wife is the husband's mother; that is, the link is mother-son). This type of confusion may be the cause of some of the contradictions found in the data. Another potentially confusing problem in classifying residence is the level of aggregation at which the ethnographer categorizes residence. Aberle (1981:29, 33-34) explains that among the Navajo, a daughter and her husband may live in a separate hogan on the corporate kin group lands belonging to her mother. If only the smallest household is considered, one could classify this residence pattern as neolocal. O n the other hand, if the larger co-operating economic unit is considered, it would be classified as matrilocal. A further source of contradiction is whether to classify according to the stated norms of residence or the frequency of residence. Is the ethnographer classifying residence on the basis of number of occurrences in the population or on the basis of the stated preferred norm? There is also the factor of time. In the case of changes over time, I originally determined to classify a culture's residence pattern according to the earlier pattern. However, if the majority of the data on a cultural unit had been collected after the change in residence pattern had occurred, resulting changes in the social organization would be in contradiction with the earlier 104 residence pattern. Therefore, I considered the residence pattern that was reported by the author who provided the majority of the data on a cultural unit. In addition, I considered the author's familiarity with the cultural unit and, where these were provided, the author's reasons for classifying residence as he or she did. The question of changes over time was particularly troublesome for three cultural units, the Beaver, the Slave, and the Tahltan. Goddard describes the Beaver as matrilocal (1916:221). A t the same time, however, Goddard indi-cates that, in response to a question about matrilocal residence, an older male informant explained, "Now I go everywhere my son-in-law goes". This de-scription does not suggest matrilocal residence. To complicate matters, Ridington reports bilocal residence among the Beaver (1981:352-53). One must also keep in mind that the Beaver, also had initial matrilocal residence. Could Goddard have mistaken initial matrilocal residence for final matrilocal residence (this possibility would explain the statement by the older informant that he, not the son-in-law, was the dependent resident)? Is Ridington's report of bilocal residence the correct one? Is the contradiction the result of change over time or of variations between different groups of Beaver Indians? In any case, how should residence be coded in this instance? I coded the Beaver as having bilocal residence. In the second case, Honigmann notes that matrilocal residence is more frequent for the Fort Nelson Slave (1946:69, 84). Helm, on the other hand, asserts that the L y n x Point Slave had a bilocal residence pattern (1961:67). Is the confusion due to changes over time, differences between local groups, or the result of Honigmann's reporting the more "frequent" pattern while 105 H e l m reports the available options? H e l m brings up the question of residence among the Slave in a later publication (1965a:382). Although a number of scholars since Honigmann have coded the Slave as matrilocal, she could find no evidence for doing so. She concludes that they all base their classification on Honigmann's material. H e l m argues that Honigmann is incorrect; the Slave are bilocal, although they did have initial matrilocal residence. The Slave were bilocal in the past, and bilocal residence would have been more adaptive then, as it is now. After deliberation, I coded the Slave bilocal. The third case is that of the Tahltan. Emmons in 1911 wrote that matrilocal residence and avunculocal residence were the most frequent forms of residence (1911:27-28, 98). Teit argues that it did not matter whether the couple went to live with the woman's family or the man's, although virilocal residence was more frequent (1956:145-46). Albright records a bilocal residence pattern for the Tahltan (1982). M a c L a c h l a n asserts territorial rights, not residence patterns, determined where a couple would live; they might live with a man's matrilineal kin or a woman's matrilineal kin (1981:458-68). I coded the Tahl tan as bilocal since there is evidence that there was variation in the residence pattern. 1 In retrospect, perhaps these groups should have been coded matrilocal; on the other hand, the evidence suggests that the couple did have choices about where they would live. It seemed that the category bilocal 1 These terms for classifying residence patterns are from a system of classifica-tion proposed by Aberle (1981). There are two parts to each term,the first reflecting the person of orientation (the person claiming residence rights - i.e., the bride or groom), the second half reflecting the person sponsoring the person of orientation. The first half of the term is either viri (groom) or uxori (bride) and the second half is either patrilocal or matrilocal. Thus , in the example provided by Fischer above, residence is classed as virimatrilocal. This categorization of residence reflects the fact that the groom's father is in a dependent position in the residence group, and it is the mother who provides the groom with a legitimate claim to live there. 106 residence could best capture this dimension of choice. Where contradictions exist, the classification of residence type was finally made on the basis of time depth and the authors' familiarity with the cultural group. Some weight was given to a consensus among the majority of the authors. One further problem was encountered in defining m y variable, postmarital residence pattern - how to classify the category avunculocal residence. Mart in and Voorhies (1975), Nielsen (1978), Schlegel (1972), and Whyte (1978) all claim that avunculocal residence is similar to patrilocal residence in its effect on sexual equality/inequality. Mart in and Voorhies (1975:228), Nielsen (1978:29), and Schlegel (1972:135, 141) argue that this is because with both of these types of residence, women no longer have economic control. Whyte states this is due to physical distance from female kin after marriage (1978:34-35). Given these two different models, one encounters a problem with the pattern of avunculocal residence found in some of the cultures used in this thesis. These groups combine avunculocal residence and preferred cross-cousin marriage (e.g., Haida , Tlingit, and in some cases, the Tsimshian). This re-sults in a system where a young man moves in with mother's brother and marries a woman who is mother's brother's daughter. Thus , a woman continues to reside with her consanguineal kin after marriage (e.g., for the Haida , Blackman 1982:30, Harrison in Blackman 1982:26, Murdock 1934a:239, 1934b:359). Kle in (1975:80-83), who maintains that this pattern characterizes both the Tlingit and the Haida , discusses this pattern in detail for the Tlingit. Two aspects of this type of marriage/residence pattern are important 107 for women. First , although Tlingit women lived in the same house, genera-tion after generation, the house itself was owned by the lineage of their husbands. Second, men of the lineage owning the house all tended to m a r r y women who were lineage sisters to each other. 1 The result of the ongoing intermarriage between two lineages is that a man married a woman who was both father's sister's daughter and mother's brother's daughter, because father's sister often married one's mother's brother. Thus , a woman grew up in her father's clan house and often remained there after marrj'ing a father's sister's son who had moved in at puberty (Klein 1975:81). Kle in (1975:82-83) describes the benefits which accrued to women with this pattern of residence. In this situation, two sets of links existed, nephew-uncle and parent-daughter. In addition, although she did not live in her own lineage house, all the women in the house were a woman's lineage sisters. Their support and loyalty could be counted on; a woman's parents and lineage sisters were there to lend support and help. Moreover, a woman's affines were members of her father's lineage; they had watched her grow up in their midst. No mother-in-law was present to dominate her. Rather, the woman in authority over the new bride was often husband's mother's brother's wife, her own clanswoman (see Oberg 1973:35-36) who had been a person in authority during her childhood as well. Final ly , because of this pattern of residence, a woman owed allegiance to only two lineages, her own and her father's (and husband's) lineage, with whom she had grown up, rather than to three lineages - her own, her father's and her husband's. 2 1 The Tlingit believed that all the women of a household should be of the same lineage (de L a g u n a 1960:162, Olson 1967:22). . 2 Citing Durlach (1928) as evidence, Rosman and Rubel (1971:39-44) argue that the Tlingit marriage pattern (They call it "wife exchange". According to Oberg 108 Returning to the problem of how to categorize avunculocal residence in m y thesis, the avunculocal residence pattern described above has a number of unique features. First , there are two residence links, mother's brother -sister's son (avunculocal) and father - daughter (uxoripatrilocal). Second, a woman is a member of her mother's kin group and although she maj' reside with members of her mother's kin group, they do not own the house or subsistence resources. They do in some cases control distribution, however. If Whyte's (1978) reasoning is accepted, avunculocal residence, in this case, should be positively associated with women's status, since a woman remains in close proximity to her own female kin. If the logic of Mart in and Voorhies (1975), Nielsen (1978), and Schlegel (1972) are accepted, avunculocal residence should have a negative effect on women's status because women do not control the house and other important economic resources. I chose to treat avunculocal residence as a separate category rather than make an arbitrary decision about the validity of one model over the continued .. . (1973:38), the Tlingit marriage pattern is one of "nephew exchange") involved three lineages: wife givers, wife takers, and father's lineage (ibid:44). These authors disagree with the evidence of de L a g u n a (1952) and Oberg (1973) which supports Klein's contention about the two lineage pattern. These authors claim that de L a g u n a is mistaken about the two lineage marriage system; they even quote de L a g u n a in order to prove that de L a g u n a is mistaken. In fact, the quotation supports de Laguna's claim; " A young m a n who is succeeding to the title of his maternal uncle would m a r r y his uncle's widow, and she would be ideally his paternal aunt" (1952:7, m y emphasis). Given a two lineage rather than a three lineage marriage pattern, a man's maternal uncle's widow would be his paternal aunt, for as Klein points out, "father's sister often married one's mother's brother" (Klein 1975:81, see above). Moreover, information provided by Olson (1967), a fourth author, also supports the claim of a two lineage system. Olson notes that father-in-law and son-in-law were frequently of the same lineage (1967:14). This agrees with Oberg's (1973:36) chart of a two lineage marriage system, but not Rosman and Rubel's chart (1971:41) of a three lineage system. Olson (1967:14) also states that a man ideally calls his son-in-law "nephew" (also illustrated in Oberg's chart, 1973:36). Final ly , Olson (1967:22) states outright that the majority of marriages occur between two lineages. 109 other. It was recognized that avunculocal residence could be combined with one of the other categories at a future time, if desired. Thus , final postmarital residence 1 has three categories; these categories are based on Aberle's classification system (1981): 1) uxorimatrilocal, referred to as matrilocal, 2) viravunculocal, referred to as avunculocal, and 3) other residence, which incorporates all other patterns (e.g., virimatrilocal, viripatrilocal, bilocal). 3. The Hypotheses The following hypotheses are tested: 1) there is a relationship between the type of descent in a society and differences in the access of men and women to power and autonomy, 2) there is a relationship between the final postmarital residence pattern in a society and differences in the access of men and women to power and autonomy. Before proceeding to a discussion of the data (Chapter 3), two points remain to be discussed. First , descent and degree of stratification vary inde-pendently of each other. Residence, on the other hand, does not vary inde-pendently of descent or degree of stratification. T h a t is, residence is partly determined by descent since other descent and avunculocal residence do not occur together. Moreover, residence is partly dependent on degree of stratification; in my sample, matrilocal residence occurs only with egalitarian and intermediate societies, and avunculocal residence occurs only with ranked societies. If all the independent variables varied freely, eighteen (3x3x2) possi-ble types of cultural groups could occur. However, six combinations do not occur in my sample. These are: 1) an egalitarian or intermediate matrilineal 1 Initial residence is not considered an indicator of residence. 110 society with avunculocal residence, 2) an egalitarian or intermediate other society with avunculocal residence, 3) a ranked matrilineal society with matrilocal residence, 4) a ranked other society with avunculocal residence, 5) a ranked other society with matrilocal residence, and 6) an intermediate other society with avunculocal residence. The remaining twelve combinations do occur in the sample. The second point concerns the cultures investigated in this thesis. Many of the models discussed in this chapter postulate a relationship between the dependent and independent variables (or postulate that a relationship is stronger) in societies with a horticultural or agricultural subsistence base, rather than a hunting and gathering subsistence base. I test the relationships on hunting and gathering societies. If the hypotheses are supported, then the relationships may also exist outside of horticultural cultures (i.e., in cultures . based on hunting and gathering), and the models can be extended to include hunting and gathering societies. The purpose of this chapter has been to describe the three independent variables tested in this thesis. These variables have been a source of debate in the literature on the anthropology of women for a long time. Some authors link one or more of these variables with variations in women's power and status; others contend there is no association. This thesis seeks statistically to support or reject the existence of the postulated relationships on data from thirty-nine hunting and gathering societies within a limited geographic region. In the next chapter, I discuss the quality of the data sources used in the analysis. III. T H E DATA This chapter focuses on the ethnographic data used in m y analysis. Specifically, the type of materials used is outlined. In addition, I evaluate the sources' reliability in terms of the problems posed by the data for those interested in the position or status of women in other cultures. Historians routinely engage in "source criticism"; anthropologists rarely do . 1 The reason often provided is that "Historians study written documents" while "anthropologists study living peoples" (Carmack 1972:229). In reality, anthropologists do use written documents in their research - comparative statistical studies are an example. E v e n the data provided by "living peoples" should not be accepted uncritically, however. 2 The historian Bloch has argued that "much of a firsthand observer's information is as indirect as that of a historian working with documents; that is, much of what a fieldworker 'sees' comes to him through the eyes of others" (Hudson 1973:130). Some anthropologists have recognized the need for source criticism. Authors such as Sturtevant and Evans-Pritchard "have generally castigated anthropologists for uncritical use of documentary sources" (Carmack 1972:232). Evans-Pri tchard warns that "Ethnographic monographs are documents too . . . and we have trusted them far too uncritically" (Carmack 1972:233). Carmack observes that Whyte (1978:14), for example, indicates he will simply assume accuracy rather than question the reliability of the data. Driver (1974a; first published in 1938) and Jorgensen (1980:301-02) are exceptions. Indeed, the scarcity of studies on the accuracy of data is illustrated by the statement of Narol l , Michick, and Narol l (1974:141) that no study done since Driver 1938 is nearly as comprehen-sive, thorough, or useful. 2 A recent article published in the June, 1986 issue of American Anthropologist (Romney, Weller, and Batchelder 1986) provides a mathematical model for analysing informant consensus on questionnaire data that estimates 1) the cultural competence or knowledge of each informant and 2) the correct answer to each question asked of informants. There is no consideration of the ethnographer's impact on the data, although they do acknowledge that this is important. 112 the reader must learn to evaluate sources and that in this area we have a great deal to learn from historians. Carmack insists that for the reader to properly evaluate the sources from which the information comes, the ethnographer should provide information on the sociological positions of informants, their relations to power, and personality factors which might uniquely influence the informant's information. The same information is needed about the ethnographer recording the data (Carmack 1972:233). 1 Some of these details regarding sources are considered in this chapter. Although it may be uncomfortable for anthropologists to admit that the ethnographic data may not be as reliable on the issue of women's status (or on other topics) as we have assumed, it is only by assessing our sources and recognizing the limitations of the data, not by ignoring them, that the researcher can find methods of dealing with these limitations. B y being aware of, and making allowances for, the limitations of the data, the researcher can utilize the data more effectively and have confidence in the re-sults he or she obtains. The source criticism presented in this chapter is a general one; I do 1 In their attempts to understand and describe women's status, those scholars interested in the anthropology of women have made an important contribution to the discipline in this area of source criticism, not only by their exploration of ethnographer and informant bias and what these have meant for descriptions of women's status, but also in their attempts to write new ethnographies without these biases (e.g., Albright 1982, Berndt 1981, Briggs 1974, Bujra 1975, Dahlberg 1981, Etienne and Leacock 1980, Fiske 1982, Freeman 1975, Kle in 1975, Kle in 1980, de Laguna 1983, Leacock 1978, 1979, 1981, Leacock and N a s h 1981, Matthiasson 1974, Quinn 1977, Rapp 1979, Ridington 1983, Rogers 1978, Rohrlich-Leavitt 1975, Sacks 1976, 1979, Schlegel 1977a, Schrijvers 1979, Smock 1977, Sutton, Makiesky, Dwyer and Kle in 1975, Tiffany 1979). Whereas some ethnographic data may be reliable (see Driver 1974a), evidence presented by scholars such as those listed here, suggests that this may not be the case for the ethnographic data on women, their roles and their status in general. This v topic will be taken up momentarily. 113 not assess the sources for each work individually. Rather, I discuss the manner in which the views of ethnographers and informants, coupled with the reality of fieldwork techniques, have affected the data on women that I use. Final ly , I indicate some of the methods I use to deal with the limitations in the data. The data were obtained primarily from major ethnographies. I rarely used secondary sources or ethnographic material on special topics. In addition, where possible, at least one early account by a non-ethnographer was included for each cultural group in the sample. The majority of these early accounts were by explorers, fur traders, and missionaries. In the preliminary stages of thesis research, I expected that two or three sources would be sufficient for each cultural group; ultimately, the number of works needed to obtain a reasonable amount of corroborated information was far greater. The dearth of comprehensive information on women and their activities, a desire to avoid selectively choosing data which would support m y expecatations, 1 as well as a desire to verify information and to depend on more than one main ethnographer for each cultural group, made this expansion of sources neces-sary. The major problem in obtaining and coding information on women and their access to power is the invisibility of women in the ethnographic literature. E v e n where information about women is provided, women are viewed as marginal , the information is minimal , and the activities of women are generally devalued. Some of the reasons for this state of affairs in the anthropological and ethnographic literature were discussed in Chapter 1. These 1 Jorgensen (1980:301) observes that a reseacher can introduce bias by selectively choosing a source that supports his or her own hypotheses. 114 reasons include the manner in which status is defined and evaluated, the manner in which field work is carried out, the choice of informants, and the use of ethnocentric and androcentric models by ethnographers raised in western cultures. Another factor is discussed at this point - the failure of ethnographers, and anthropologists in general, to recognize the effects of western contact and the resulting culture change on the position of women in other societies. A s Tiffany indicates, these factors are all interrelated (Tiffany 1979:7-8; also see Leacock 1978 and Perry 1979). A number of the scholars interested in why women's status varies have argued that women are often seen as universally subordinate to men because the effects of culture change, brought about by contact with European cultures, are ignored by anthropologists, who often fail to recognize these changes. These scholars argue that factors such as colonial policies, missionization, trade, the introduction of a cash economy, and industrialization have had consequences for women's status (e.g., Etienne and Leacock 1980, Fiske 1982, Kle in 1979, Leacock 1978, 1981, O ' B a r r 1976, and Tiffany 1979). 1 Leacock (1975:604-05, 1978:250) argues that anthropologists view cultures as static, closed systems without a past. Anthropologists generally collect recent data, assuming it accurately describes the past, thereby "freezing those people in some timeless 'traditional culture' that does not change or de-velop, but only becomes lost" (1978:250). This conceptualization of native Indian cultures is evident in the ethnographic material used in this study. A 1 Adaptations and diffusion, of course, are not phenomena limited to contact be-tween Native Indian cultures and Western European cultures. It is tempting to forget that Native Indian cultures were interacting and undergoing change prior to contact with Europeans (i.e., the issue of Galton's Problem, which I discuss in Chapter 4). 115 number of ethnographers claim they are describing the "aboriginal" or traditional culture, even though they are reconstructing nineteenth century traditions from data collected in the twentieth century; the information is recalled by older informants who are remembering the past (e.g., Barnett 1955, Barbeau 1929, Honigmann 1946, M c K e n n a n 1965, Osgood 1936a, Seguin 1984). 1 A number of authors assert the cultures have changed little from the time of contact so the present-day culture can be treated as an approximation of the aboriginal culture (e.g. He lm 1965, M c K e n n a n 1965, Sharp 1979) - this claim cannot be justified without a great deal of research. The blame should not be placed totally at the feet of anthropologists. The field technique is also at fault. In reference to Barbeau's work among the Ts imshian from 1914 to 1929, Halpin (1973) points out that Barbeau was working with sixty and seventy year olds who were remembering the 1860's to the 1880's. A s she observes, this type of memory ethnography means "people were attempting to describe what they perceived as the essen-tially timeless cultural patterns of, or persisting from, an immediately preceding traditional age" (Halpin 1973:21). The purpose of this critique is not to suggest these memory ethnographies are not of value; the information collected in this manner is essential. We should not, however, expect more from the data; the limitations of memory ethnography must be taken into account when this type of material is being collected and used. I dealt with the problem of historical cultural change in a number of ways. First , as far as possible, I used early sources rather than later ones. I hoped in this way to obtain information on the access of women and men 1 Both Barnett and Osgood acknowledge there are problems with such reconstructions. 116 to power prior to changes brought about by contact with European cultures. This would be difficult since most of the ethnographies have been written after the mid-nineteenth century. Nevertheless, I hoped that by concentrating on the earliest ethnographic information, I could keep the major effects of culture change to a minimum. In addition to early ethnographies, I used early non-professional accounts (i.e., accounts by traders, explorers, and missionaries, rather than trained anthropologists). Unfortunately, it was not always possible to use early ethnographies, particularly for a number of the Northern Athapaskan speaking cultures. Information on these groups often consists of spotty references in the early documents written by traders, explorers, and missionaries, and ethnographic research done after 1941 (Rogers 1981:19-21). In those cases where more recent accounts were the only possibility, I planned simply to ignore more recent borrowings but soon recognized that changes are rarely limited to a single trait. Generally, a number of modifications arise. A change in the economic system for example, m a y affect the social and ceremonial life of the culture. It is difficult to distinguish where changes begin and end. To avoid arbitrary decisions, I decided to accept the description of each cultural group as it is presented in the data. Whenever older patterns are distinguished, however, this information is utilized. In addition, by using a continuous area sample, I hoped to control for the situation to some extent, since most of the cultural units would have undergone similar historical experiences. One way of dealing with this problem would have been to choose one time period per group and to limit m y sources to this particular time period. Unfortunately, not enough material was available, in some cases, to allow 117 excluding all the ethnographic material not of the appropriate time period. In addition to changes in women's roles brought about by contact with European cultures, the information on women is coloured by the ethnographer who collects and records the information about women and their place in the society; the ethnographer's background and assumptions about the position of women affects the data on women. In Chapter One, I discussed the point that ethnographers tend to hold certain ethnocentric and androcentric assumptions about men and women, affecting the type of data they collect and the manner in which they interpret their data. Thus , one finds the fol-lowing type of statement in the literature: "It is not uncommon to see these old witches communicating their dreams to the tribe; men and women standing by with open mouths, and impressed wonder-stricken faces" (Boas 1916:562). This statement is made even though Boas recognizes that these women were influential. Similarly, Niblack (1890:340) declares, "The war canoes were generally steered by some old crone whose courage and influence had been found oftentimes to be of the greatest incitement to the warriors". While it is obvious these women had power, it is also obvious the ethnographers view them in a negative and demeaning light. A n interesting example of how eas-ily information about women can be altered to meet the reality of the ethnographer who is less than meticulous is presented by Suttles (1951:188). In his study of the Coast Salish of the Haro and Rosario Straits, Suttles confesses he wanted to discount information that Coast Salish women engaged in reef net fishing. In the end he did not discount it for two reasons, first, because of the universal agreement that they did, and second, because he ac-tually saw women using nets that may be ancestral to reef nets. 118 Although the preconceptions and stereotypes of the ethnographer are not acknowledged or, in many cases, even recognized, one can argue (see Chapter 1) and can see from the examples above that ethnographers do have definite ideas about women and their position in other cultures. 1 These views, added to the almost exclusive focus on men in all cultures and the resulting marginality of women (see Chapter 1), surely affects the ethnographer's final portrait of the community or cultural group. F o r example, one ethnographer asserts, the Tlingit world is a thoroughly masculine one. Women own noth-ing but their clothes and ornaments and are only permitted to pass on their deceased husband's individual property. Women are, on the other hand, well treated, and socially they stand very high . Women are the means through which position, rank, and property flow from one generation to another, but thej' have no legal power to possess or use these privileges (Oberg 1973:38). In a recent study of the aboriginal social organization of the Tlingit, which includes an analysis of the aboriginal status of women, de L a g u n a paints a far different picture. "In economic, political, and ceremonial matters, Tlingit women have more authority than is usually recognized, perhaps even the final say" (de L a g u n a 1983:81). Given these contradictory views of the position of Tlingit women, the information provided by one early explorer among these people is of particular interest. Vancouver reported in 1793, In all the commercial transactions the women took a very principal part, and proved themselves by no means unequal to the task. These views are sometimes made explicit. In a discussion of gender in linguistics, Hi l l -Tout has determined, from an examination of the language that the Salish people, like "Our own A r y a n ancestors", distinguish the masculine as the "nobler gender"; "Anything that is large, strong, fine, or excellent, is . . . masculine; anything that is small, weak, mean or contemptible, is . . . feminine" (1902:375). 119 Nor did it appear, that either in these or in any other respect they were inferior to the men; on the contrary, it should rather seem that they are looked up to as the superior sex, for they appeared in general to keep the men in awe, and under their subjection. (Vancouver in de L a g u n a 1983:81). A review of the goals and interests of these three writers m a y help to ex-plain, to some extent, the difference in their interpretations. Oberg's work, based on informant information, focuses primarily on the economy and clan property which male house heads control; de Laguna is interested in describing the aboriginal Tlingit sociopolitical organization at the time of first European contact; Vancouver provides an eye-witness account. Thus , these authors' interests, goals, biases, and assumptions have shaped their definition of women and women's roles in Tlingit society. For the researcher interested in women's status or power, attempting to distinguish between the ethnographer's interpretations of facts and the reality of the situation can be frustrating. Given the almost exclusive focus on men and their activities, one of the most baffling problems is the endeavor to learn, from the data provided, whether women are even present at day-to-day events and, if so, what roles they play. The invisibility of women is com-plete except for the minimal attention given to what are discerned as "women's roles" - child birth, child rearing, domestic activities, and subsistence activities. The problem is aggravated by the ethnographer's tendency to use pronouns and adjectives which refer to men, when references to both sexes are appropriate (e.g., Chapman 1914:29, 54, 86, 105, 109, 145; Olson 1967:48; Swanton 1905a:129 versus Murdock 1965b:277, Swanton 1912:34). Olson (1967:48), for example, defines the Tlingit term anKau as "rich man"; de 120 Laguna , on the other hand, reports that it refers to a rich man or woman (1983:74). E v e n more confusing is the use of the pronouns "he" and "his" when women are involved. In three instances, Barbeau (1929) refers to informants as "he". In all three cases, the informant is a woman. Since two of these informants are chiefs, Barbeau's choice of pronouns is a serious oversight; women have been made invisible even though they occupy important positions (1929:28, 70-71, 83). A distinction is often made between the "chief and a "chiefs wife" (Barnett 1939:264, 269; Drucker 1951), even when terms in the native lan-guage do not make such a distinction. F o r example, the Haida term K t T S L a-i, "chief, used in Swanton (1912) is applied to both men and women (Swanton 1912:15, 23, 40, 42 - Swanton's initial translations of the songs on these pages make this clear). Nevertheless, in his final translation of each song, the term is applied to men. Where it applies to women, it is variously translated as "chiefs daughter (song 16 on page 15) or "chief woman" 1 (song 17 on page 15). A study of the literature led me to suspect women in many cultural groups in m y sample were more directly involved in community life than the ethnographic accounts illustrate. Striking support for this suspicion is provided by a comparison of Drucker's (1951:386-439) account of the Shaman's Dance and the precise translation of an eyewitness account in Sapir and Swadesh (1955:89-120, 122). Another example is found in Curtis (1970b:36) - a public speech made by the wife of a successful Nootka whaler. Her speech makes it clear that whaler's wives are ritually essential to the hunt's success, and 1 The term for "chief woman" is qloldjat. 121 Nootka people (or, at least, Nootka women) make no distinction between ritual activity and the actual harpooning of the whale in terms of their importance to the enterprise. Women seem an integral part of whale hunting. Given such examples, can it be assumed that when women are not mentioned, they have no role to play? One common example in the ethnographic literature is the statement that prerogatives are passed on through the mother - often by a grandfather, through his daughter's dowry, to his son-in-law or to his future grandsons. Women do not seem to play an active part in this process (e.g., Drucker 1951:268, Suttles 1955:28-29). Yet , Suttles himself, in the paragraph following his statement that "a man gave his inherited names to his sons' sons as a matter of course, but he also gave them to his daughters as part of their dowries to be used by their sons to show their mother's origin", explains, Thus Simon's father's father's father . . . married a Kwant len woman who brought with her three Kwantlen names which she gave to their three sons . . . his own mother, through Kwantlen, bore a Samish name. These two descriptions, provided by the same author, present two different views of Stalo women and their roles in passing on family names, as well as their ability to receive names. Indeed, if one were to accept the descriptions of women's lives in the ethnographic literature, one could safely state that: women are born, go through puberty, are married off, and give birth (although male informants cannot provide many details). A t this point, most descriptions become vague, except for references to plant gathering, the snaring of small game, 1 and 1 de L a g u n a (1983:83), for example, points out that there do not exist any tales about women as adulterers and widows. Information on women's involvement in community affairs and ceremonial activities are minimal and often confusing. Indeed, from discussions of death in the literature, the reader would be justified in concluding that women rarely die. E v e n when women are mentioned, the information is often based on assumptions that women are inferior and subordinate to men or that women are passive objects. Rarely is an attempt made to discover whether such assumptions are appropriate. Whyte (1978:45) discusses this tendency in the ethnographic literature to treat any norm as evidence of women's subordination. In describing how women's contribution to subsistence is assessed, he observes, "In other words the exist-ing literature generally takes the position that women are damned if they do [most of the work] and damned if they don't." 1 E v e n the choice of words can convey very different images. One learns that tattooing among the Chandalar Kutchin "was confined to the women . . . Face painting was the privilege of the men" (McKennan 1965:46-47) and that "Hunting is the privilege of man . . . the gathering of the berries and edible roots falls entirely to the lot of the woman" (Morice 1910:419; emphasis added). So strong is the tendency to view women as subordinate to men and limited in their social roles that evidence to the contrary is ignored. " continued . . . adequate accounts of Tlingit methods of preserving fish. Such an omission is difficult to comprehend, given the importance of fish to the Tlingit diet and way of life. 1 Osgood is one of the few ethnographers to recognize that women's activities are sometimes evaluated incorrectly (1936a:131-32). Although he believes Kutchin women's status is "probably on the whole inferior", he notes early observers did not understand what they were observing. He argues that the lot of women as "beasts of burden" is exaggerated. Historical sources "show emphatic disagreement, as to whether the woman is socially inferior or not". 123 Mcl lwra i th (19481:171-172) presents an amusing example. After declaring that women are rarely established in a chiefs seat, usually only when a chief has no male descendants, he gives an example of an important male chief who "had thus obtained a seat for his daughter, as well as ones for his two sons" (emphasis added). Another example is provided by Goldman (1953:81-82), who states that, theoretically, a bereaved Alkatcho Carrier spouse, regardless of sex, was under the rule of the siblings of the deceased and must perform appropriate mourning rituals. He reports that men were not really bound by these restrictions, however, noting he received no accounts of widowers having their faces burned, etc. Goldman seems to have forgotten the existence of two such accounts which occur on pages 72 and 345 of his study. A n important factor in the quality and reliability of the data is the informant selected by the ethnographer. One criterion for assessing data is how representative the informants are of the cultural unit they describe. In other words, how knowledgeable are they? Do they have an interest in describing the culture in a certain light? F o r example, Hearne's description of his journey in 1770-71 with a Chipewyan band, under the leadership of Mattonobee, has been the basis of many descriptions of aboriginal Chipewyan life. Leacock (1978) cautions, however, that the economic ties of this band of Chipewyan with European fur traders make them non-representative of the aboriginal Chipewyan culture as a whole. Certain statements by Hearne (who greatly admired Mattonobee) imply he was not an "average" Chipewyan male. A n examination of his background supports this suspicion; Mattonobee was the son of a Chipewyan man and a slave owned by Southern (Cree) Indians. He was not raised among Chipewyan; he was born, and lived at least part 124 of his life, at Prince of Wales Fort (1968:328-34). Another author, Frankl in (1969a), is also a pr imary source on early Chipewyan culture. His data come mainly from his observations of Akaitcho, the Copper Indian chief who served as his guide (1969a:289). A statement by Frank l in that no other Chipewyan group had a leader of authority equal to Akaitcho raises the suspicion that the fur trade m a y have already had an impact among this group of Chipewyan. Should Akaitcho be considered a "typical" Chipewyan male (that is, one who lives by the norms of the aboriginal Chipewyan culture)? Regardless of the answer to this question, many anthropologists continue to use the accounts of Frankl in (and Hearne) as descriptions of typical Chipew3 ran culture at the time of contact. The informant's position, knowledge, and interests should be considered by the researcher using the ethnographic material. For example, Barnett (1955:7) reports his information on the Pentlatch came from a person who had not grown up among these people. His first Pentlatch informant refused to co-operate; thus, with some personal misgivings, Barnett obtained his informa-tion on the Pentlatch from his first informant's wife who had grown up among the Kwakiut l . Jenness (1937) obtained data on the L o n g Grass Sekani from a Gitksan woman married to a Long Grass Sekani, while de Laguna (1960:13) indicates that half her information on the Tlingit was obtained from an "atypical", well-educated male. Oberg (1973), who did his fieldwork on the Tlingit in 1932, relied heavily on a K l u k w a n "halfbreed" who had lived there as a child. This man was a marine engineer and served as Oberg's interpreter and informant. Moreover, he collected data for Oberg on his own initiative. Garfield's (1939) Tsimshian informants include a long time male 125 resident of White communities and two Christian men, one of whom was married to a clan sister (an unheard of situation among less acculturated Tsimshian). Honigmann (1954) admits his principal Upper L i a r d (Kaska) informant is more representative of the Teslin Lake , Upper T a k u , and even Tlingit cultures, than he is of Upper L i a r d culture. Similarly, Barbeau (1929:46) reports his male interpreter-informant "had been educated by missionaries and seemed at times not to possess sufficient knowledge of some of the old customs", while the information provided by another informant "may not be wholly relied upon, owing to his having lived away from his tribe most of his life" (1929:100). While data collected from such informants is not necessarily inadequate, the possibility must be considered. The informa-tion can be valuable but might be more appropriately treated as the perceptions of someone living in two different cultures, rather than information from someone knowledgeable about the culture being studied. Util izing a num-ber of informants is one way to check the quality of the data. In many cases, however, time and monetary constraints make the number of informants interviewed quite small , with the data heavily reliant on one or two (e.g., Barnett 1939, Drucker 1950, Mason 1946, Ray 1942, Sapir 1915). The assessment of these factors becomes almost impossible, of course, when there is no information provided about the number or the background of informants (e.g., Boas 1890a, 1890b, 1892, 1895a, 1895b, Emmons 1911, F a r r a n d 1900, Hil l -Tout 1902, Koppert 1930, Slobodin 1962, Teit 1914, 1975a). One must also consider the informant's interests, biases, and goals, which determine the type of information he or she is willing or able to provide. For example, Dav id Thompson, travelling in the Subarctic from 1784 to 1812, 126 complains, I have always found it very difficult to learn their [Cree] real opinion on what may be termed religious subjects. Ask ing them questions on this head is to no purpose; they will give the answer best adapted to avoid other questions and please the enquirer. (Hopwood 1971:10) In his studj' of the Alkatcho Carrier , Goldman (1953:14-15) reports one of his informants, a Bella Coola man (!), considers the potlatch system to be of great significance in Alkatcho Carrier life. This claim is apparently based on the informant's own heavy involvment in the potlatch system and his interest in rank; as Goldman observes, rank and the potlatch system were of less interest to the other informants. Jorgensen indicates a number of factors which may introduce informant bias into the data. He points out that, unconsciously, the ethnographer may phrase questions in such a way as to elicit the responses he (or she) expects to hear. A n informant m a y have his or her own interests and goals to achieve. H e or she may not want to offend the ethnographer. The informant m a y be in financial need and not wish to jeopardize future fees, and therefore, provides information even when he or she knows nothing about the topic. Final ly , the informant may wish to mislead the "probing, snooping, ethnographer" (1980:301-02). Thus , the informant's biases and interests can profoundly affect the final picture of the culture presented. In addition to the ethnographer and the informant, the manner in which field work is conducted can affect the quality and reliability of the data. For example, if the ethnographer has little knowledge of the language, he or she is dependent on an interpreter. Indeed, the ethnographer may restrict the 127 choice of possible informants because of the language problem. I suspect this factor m a y be one reason ethnographers sometimes chose younger, western educated informants rather than older informants who may have had a greater knowledge of the more traditional culture. Time constraint is also a factor. How well did the writer know the culture he or she is describing and how much time was devoted to data collection. In some cases, the amount of time spent was rather limited (e.g., Birket Smith 1930 - one month, Jenness 1937 - four weeks, M c K e n n a n 1965 - nine weeks, Mason 1946 - less than two months, Sapir 1915 - two days, Teit 1975a - two weeks). In those cases where ethnographic material was produced by non-professionals who had spent years among a group of people, the accounts may be of comparable value to those produced by professionals who were not able to spend any length of time collecting the materials . 1 O f course, in the case of non-professionals, one must be aware of their motives for being there, as this may affect the reliability and quality of the informa-tion they provide. Final ly , for the researcher interested in women's roles and status, the sex of the informants becomes an important consideration. In the ethnographic sources I studied, most of the informants were men. The result is a dearth of information on women, and a view of women and their roles which may be one-sided. 2 A s explained in Chapter 1, men are the principal informants, 1 This is not necessarily so, of course. Cultural blinkers will prevent an observer from understanding what he or she is observing, regardless of the amount of time spent in the community. 2 Tiffany and A d a m s (1975:2) make this point with a quotation from Ardener (1972:138): The fact is that no one could come back from an ethnographic study 128 both because of the ethnographer's androcentric bias and, in other instances, because the ethnographer cannot have access to female informants. Regardless of the reasons, the results are the same. Barnett's (1955) Coast Salish informants, for example, include twenty men and one woman (she was a second choice because the first male informant selected was uncooperative); Drucker (1950) had fourteen primary informants, all men, as well as four women who provided specialized information. Honigmann (1954) had six male informants; Mcllwraith's Bella Coola informants were primarily men. McKennan's (1965) informants were twenty men and two women. Mason (1946), Oberg (1973), Swanton (1905), and VanStone(1963) interviewed only men. Osgood (1936a) had four male informants in 1936 and nine male informants in 1937 (1966). R a y (1942) spoke with fifteen men and two women. Sapir (1915) had three male informants; Sapir and Swadesh (1939) had eight male informants. Smith (1973) interviewed three men and one woman. A few ethnographers do include a higher proportion of women informants (e.g. Sue 1980 - eleven men and eleven women, Olson 1967 - five men and two women, Garfield 1939 - four men and two women), but these are relatively rare. In general, the use of female informants is an extremely recent devel-opment. A s mentioned above, this state of affairs has two consequences which di-rectly affected m y research. First , there exists little information on women and their roles. Second, there is a male-centered focus throughout the continued .. . of 'the X ' , having talked only to women about men, without professional comment and some self-doubt. The reverse can and does happen constantly (emphasis in the original). 129 ethnographies which, I suspect, may have resulted in a biased view of women and their roles in the culture. A s Whyte (1978:22) explains, Since most fieldworkers have been males and their informants more often men than women, we may get a picture of a culture as the men think it is or is supposed to be, which m a y distort the actu-al role and status of women. The first consequence is evident. If the ethnographer does think of asking questions about women in that society, men cannot be as reliable primary informants regarding women's affairs as women would be; they generally have less knowledge about women's activities, and they may have no interest in those aspects of the culture which do not affect them directly. The male informants may provide a male perspective on women's roles, but this perspective should not be considered sufficient. The perspective of women must also be included. The women's perspective is largely missing from the data used in this analysis. It is not surprising, then, that women and the activities that are perceived as "women's activities" are largely ignored or devalued. F o r exam-ple, Drucker (1950:273) had to exclude the topic of pregnancy from his material on Northwest Coast culture traits because "consistent and well-balanced accounts could not be obtained from male informants". Smith (1982) encountered a similar problem in discussing childbirth with his male Chipewyan informant. In his publication on Northwest Coast culture traits, Drucker (1950:261) confesses that his section on textiles is "without a doubt the poorest of the list, owing to the difficulty of getting male informants to differentiate between even the simplest weaving techniques" and, since it is one of the "few fields of female endeavor in which men took no part . . . 130 men have no interest in the subject." This may well be the reason that Koppert's description of Nootka men's subsistence activities occupy twenty-four pages while those of Nootka women are detailed in four (1930:56-8).1 Similarly, Barnett (1955:77-124) uses forty-one pages to describe men's industries; he covers women's industries in six. While some authors do recognize that women may be better informants on some topics (e.g., Drucker 1950:261), few have recognized that women's views about their culture in general may add a new dimension to the ethnographic description. McDonnell (1975:273) is one exception I uncovered. This ethnographer admits not only that men and women have different perceptions of their world, but that both must be included to get a complete picture of the culture. The major limitations in the data on women's access to power is the dearth of information on women, and the evaluations of women and their roles which are based on the assumptions and potential biases of the ethnographer and on information provided by largely male informants. The ethnographic data must be carefully examined, therefore, to distinguish value laden interpretations of data from a more, objective description. Comparing a number of sources for each cultural group is one method for clarifying, refuting, or corroborating information. In the ethnographic literature I examined, the amount of contradictory material on women's roles and position within societies is an indication of how much the ethnographic material m a y be affected by the perceptions of the ethnographer and informants. 2 To 1 Unfortunately, he provides no background on his informants. 2 Sometimes it is the ethnographer himself or herself who contradicts his or her own data. In other cases, other sources do so. 131 distinguish areas of concern and to clarify, corroborate, and refute information, I chose to utilize a larger number of sources for each cultural unit. Gaps in the literature were sometimes filled from data regarding actual events which occurred in the past or are observed by the ethnographer (i.e., data on actual behavior in addition to data on cultural norms), and sources such as songs and speeches. With these types of data, glimpses of the day-to-day interactions of men and women can be obtained. Leacock (1983:19) and Whyte (1978:23) both recognize that observed behav-ior m a y provide information unavailable when only behavioral norms are con-sidered. E a r l y accounts, for example, often describe the interaction between Indian people and traders, explorers, or missionaries. Even reconstructed ethnographies contain accounts of events experienced by the informants as children and by their older kin. Although such accounts are recounted through the eyes of the informant and/or the recorder, and although such accounts cannot provide systematic information, they can provide clues about the day-to-day roles and activities of men and women in the community. A s Leacock (1983:19) asserts, such observed incidents can round out, qualify, or even discount statements regarding behavioral norms. Thus, information on ac-tual behavior can be a source of additional information about the status of women in a society. 1 A word of caution must be included regarding actual behavior. First , 1 In order to round out and clarify the data on women's access to roles found in the ethnographic accounts, I recorded information on both actual behavior and cultural norms. The data were recorded separately and I hoped to utilize these data in separate statistical analyses at a later date. Unfortunately, there was not enough information to make an analysis of the actual behavior feasible. Moreover, I could think of no appropriate way in which to integrate these two types of data to create one data set. Thus , the data set on actual behavior is not utilized in this analysis. 132 there exists the possibility that the behavior observed is not representative of behavior in the community. Authors such as Fried (1967:84-85), McDonnell (1975:236, 242, 245, 263) and Osgood (1966:22) all acknowledge that there m a y be a great difference between cultural norms and the actual behavior of people in a society. 1 Second, actual behavior, in the last analysis, is defined through the eyes of the observer; his or her frame of reference provides the meaning for the behavior. Thus , Fr ied (1967:86) cautions that observations of other cultures made by people from "complex cultures" are suspect, as the behavior may have a different meaning in both cultures (e.g., see Osgood's analysis of Kutchin women who are seen as drudges and slaves, 1936a: 131-32). A l l the above points must be kept in mind when making use of data on actual behavior. Songs and speeches, accurately translated, provide clues about, and clarification on cultural norms. One can learn, for example, that resource ownership, as conceptualized by H a i d a people, involves both men and women. Haida songs recorded by Swanton (1912: 36, 38, 50, 51, 51, 52) speak of "grandfather's house" and "uncle's house", but "grandmother's land". A Tlingit marriage speech, made to in-marrying spouses by the new in-laws illustrates that both men and women are economically important; the in-marrying spouse is assured that he or she will never be poor if he or she marries "my son" or "my daughter" (Olson 1967:20). A glimpse of the status opportunities available to women is provided by a Tlingit mourning song recorded by Swanton (1909:411). A woman of the L l e n e ' d i sings, "if I were helped by 1 A description of both types of behavior would, of course, be the ideal situation, but as Osgood (1966:22) points out, when an ethnographer reconstructs a culture, one cannot compare idealized norms with actual behavior. 133 Taxgwas [a supernatural being?], I should rebuild m y uncle's house". A Haida potlatch song recorded by Swanton may help to clarify a debate in the Ha ida ethnographic literature. Murdock (1965b:12) claims the most important potlatch, the house-building potlatch, is sponsored by the wife. Rosman and Rubel (1971:57-58) argue that the husband is the host, inviting his own people as his guests, rather than inviting his opposites, as occurs in all other potlatches among the Haida , Tlingit, and Ts imsh ian . 1 In a text of Haida songs, Swanton (1912:13) includes a potlatch song which says, "For you are a woman, you are a woman . . . Skidegate Inlet the woods (i.e. timber) you are going to command you are a woman". Swanton explains its meaning; "the girl will command when housepoles are to be raised." Final ly , one more Haida potlatch song is quoted here because it gives a tantalizing glimpse of women's status vis-a-vis men: "Women are better than men, Women are better than men, Women have more property, Chiefs of m y family, where are you?" (1912:50). 1 Rosman and Rubel's (1971:58) disagreement with Murdock is based on their claim that, in matrilineal societies, all important power positions are in the hands of men. They point out that neither a woman's brother nor her mother's brother is mentioned by Murdock, "and it is impossible to imagine that the W'algal potlatch, the most important type among the Haida , and a central institution in the rank system, should be consistently given by female hosts and donors." This type of reasoning is based on ethnocentric and androcentric models, rather than evidence. Blackman contends that both a husband and wife host, although she points out that Murdock contradicts his own argument when he states that a woman could not hold this potlatch for a child (Blackman 1982:36-37). A n exami-nation of the statement indicates that Blackman has incorrectly interpreted Murdock. Murdock states that a woman with an illegitimate child cannot potlatch to raise the status of that child (1936:19). Given the emphasis on the impor-tance of the family in obtaining and maintaining status, this inability to hold a house raising potlatch for an illegitimate child is not surprising. (Drucker notes that among the Ha ida and the Tlingit, there is no place for an illegitimate child with only one-half the necessary kin to support it, 1950:278.) 134 A. DEALING WITH T H E LIMITATIONS OF T H E DATA A number of factors have been discussed which may have affected the lack of information on women in m y data and evaluations of women's roles which may not be accurate. I have used a number of means to deal with these limitations in the data. Some of these methods have already been discussed. They include utilizing earlier, rather than more recent, sources and using a number of sources on each cultural unit in order to clarify and round out the information. Information on actual behavior was reviewed, in addition to the data on cultural norms. Also, the absence of data regarding participation by either gender was coded insufficient information. I did not assume lack of data means lack of access. Although some might consider this procedure overly cautious, it prevents the marginal position of women (due to the ethnographer's viewpoint and the predominant use of male informants) from being interpreted in a negative light, without some basis in fact. Another means for dealing with the data is to continually compare the access of men and women; this was done to check the tendency of some ethnographers to define women as limited or restricted in their autonomy, while ignoring the same restrictions applied to men (e.g. marriage choices, the sororate and levirate, etc). The access measure is another means used to deal with the limitations of the data. Moreover, it was useful for dealing with conflicting accounts in the ethnographic data. The emphasis on using a number of sources to clarify, fill in, and refute more subjective statements resulted in great numbers of contradictory claims about women's access to power and the extent of that power. A s I explained in Chapter 1. the use of the access 135 measure was of help in making the coding of data more manageable, given the amount of contradictory and confusing information. Moreover, six criteria were also used to help resolve contradictions: 1) has one author spent more time in the community than the other? 2) What is the training of each au-thor and what is his/her focus of interest? 3) Does one author speak the language of the community, or alternately, do the informants speak English? 4) Was the information in question part of a detailed description or was it simply mentioned in passing? 5) Does either author lack internal consistency in his or her discussion of the topic? 6) Is one source earlier in time (this includes consideration of the ages of informants)? Where no decision could be reached, even using the above criteria, the indicator was coded i n s u f f i c i e n t i n f o r m a t i o n . In this chapter I have considered the qualit}' and reliability of the data used in the analysis, particularly as it pertains to the treatment of women in the ethnographic literature. I have discussed some of the problems encoun-tered while coding this ethnographic material as well as some of the methods used to extract information and resolve difficulties with coding the data. In the next chapter, I shall discuss the cultural units from which the ethnographic data were drawn. IV. T H E C U L T U R A L UNITS This chapter reviews the cultural units selected for this thesis and discusses some concerns involved in doing a continuous area study. The chapter is divided into three sections. Section A deals with the advantages of a continuous area study. This includes a review of Galton's Problem. Section B deals with the selection of appropriate cultural units for this study. Finally, in section C I list the cultural units selected. A. CONTINUOUS AREA STUDIES When this analysis was first proposed, I planned to use a small num-ber of cultural units (four to six) to examine the hypotheses which have been advanced about the relationships between the dependent and independent varia-bles. Almost immediately, however, the idea of doing a comparative statistical anatysis was substituted as the research design. A t this point, there were two options - to use a large world-wide sample of cultures (i.e., a hologeistic study) or to use a geographically controlled study of continuously distributed cultures, commonly called a continuous area study. A hologeistic study has one major advantage; it allows the researcher to examine a wider range of variation in culture and social structure than studies based on one society or a few societies (Whyte 1978:16). Despite this, the second alternative was chosen. 1 A continuous area study has several advantages. First , since a continuous area sample consists of cultural units which are continuously distributed, the cultural units share a similar ecology, 1 This is the only comparative statistical study of women's power or status to use the second alternative, a continuous area study, known to this author. A l l other empirical studies of women's status examined to this point are based on cross-cultural samples. 136 137 subsistence base, language, history, and customs. In other words, a continuous area study has less variety in terms of ecology, subsistence base, language, and history, than a hologeistic study, hence, the number of possible confounding variables are reduced. Continuous area studies provide some con-trol over historical developments, culture contact, the environment, and the subsistence base (see Driver and Massey 1957:422, Pelto and Pelto 1978:269). This was the main reason for using a continuous area study in this analysis (e.g., see the discussion of historical developments and their effect on women's access to power in Chapter 3). In addition, a continuous area study allows the researcher to examine variation in the same manner as a world-wide sample, yet, because it is restricted in area, it allows for greater familiarity with the cultures and the ethnographic literature of the area. A s Jorgensen observes in Western Indians, Comparative analysis applied to many societies within a continuous geographic area has the strength of allowing the researcher to com-mand ethnographic information on a large number of societies with greater depth and breadth than is possible for a worldwide sample of the same number of societies, while it forces ethnologists to rec-ognize similarities and differences among tribes that are often overlooked by scholars who analyze one society in depth (1980:8). Final ly , a continuous area study is one strategy which allows the researcher to make some assessment of the effect of diffusion (i.e., Galton's Problem; Driver 1970:635-36, Jorgensen 1974:195-96, Pelto and Pelto 1978:266-67) and retention in accounting for relationships among variables. This is done by controlling for variables such as language family, culture area, and proximity. Galton's Problem is concerned with the effect of diffusion in the sample 138 of cultures used in a statistical study. Statistical theory is based on the proposition that a sample can be used to test whether a relationship observed "out there" in the world exists or whether it is due to chance. The statistics used to test the relationship are based on a number of assumptions. One is of concern for Galton's Problem; each unit of observation in the sample is independent of all the other units. Galton's Problem asks to what degree the customs of tribes are independent of each other. In other words, how can one be sure that the relationship observed between two variables is not due to diffusion from culture A to culture B (if they did diffuse together, culture A and culture B cannot be treated as two independent units of observation). It is important to know whether this is affecting one's sample; if so, the researcher must control for the effect of diffusion. Anthropologists since the time of Galton have attempted to deal with Galton's Problem in a number of ways. One method used by Murdock (and others) is to select sample units believed to be unrelated in terms of common origins and borrowings. Murdock suggests that only one or two cultural units should be selected from each culture area or geographic region. If more than one cultural unit is selected from a subarea (defined by Murdock in his World Ethnographic Sample 1961 or Ethnographic Atlas 1967), they are chosen from different linguistic groups or from non-contiguous localities (Pelto and Pelto 1978:267). The focus of this strategy is to ensure the independence of the sample units (Jorgensen 1974:196). Some scholars disagree with this technique. Driver, for example, contends this approach does not provide any real controls for interdependence since diffusion can occur across verj' large areas. Selecting units from 139 different culture areas or language families is no guarantee of independence (1970:636). Indeed, Driver suggests that in attempting to get rid of interdependence among cultures by choosing samples of units from different culture areas and language families, Murdock and others have actually obscured the question of the independence of the units because they have totally obscured the interdependencies among the units (Driver 1970:636, Driver 1974, 37, Driver and Chaney 1974:119, Jorgensen 1974:196). Jorgensen explains the problem succinctly, "As I pointed out a few years ago (Jorgensen 1966), drawing samples from a judgemental sample that 'controls' for time and space by disregarding both is still a judgemental sample of societies with unknown interdependencies" (1974:200). According to Driver , the solution is not to obscure history and "propinquity". Rather, he has used continuous area studies in an effort to control for the influence of borrowing and inheritance while testing for relationships among variables (Driver 1970, Jorgensen 1974:196). Stated more accurately, there is still no straightforward solution to Galton's Problem. Continuous area studies are of value for this purpose, however. Although they do not strictly control for diffusion, they do allow the researcher to evaluate the effects of diffusion. A preliminary evaluation is attempted in this thesis. The results of the evaluation are presented in Appendix 1. B. THE APPROPRIATE CULTURAL UNIT Having reviewed the advantages of using a continuous area sample, I turn to a discussion of the cultural units which were selected for the study. Choosing the cultural units for this thesis forced a decision regarding what 140 unit of measurement to use. Cultural groups can be divided and subdivided in a number of ways, depending on the criteria used to make the divisions and on the boundaries selected for each unit. F o r example, in The Indian History of British Columbia (1964), Duff analyzes the basis for making ethnic divisions, pointing out that a number of criteria can be employed, including physical type, culture, or language. H e notes that anthropologists often use language because it allows more definite boundaries to be drawn. Duff presents a classification of British Columbia Indian groups which comprises three levels of division: major ethnic divisions, dialect and regional groups, and tribal or band units. According to Duff, a major ethnic division "comprises the speakers of one language or a number of related languages, occupying a continuous area and sharing a basically similar culture" (1964:12). Regarding these major ethnic divisions, Duf f acknowledges: 1) that the Indians did not formally recognize the major divisions which are employed by anthropologists (see also Drucker 1955:108-109), and 2) that the divisions are "somewhat illogical and inconsistent" since the groups are not linguistically equivalent units (1964:12). For example, Nootka and Kwakiut l , treated as totally separate linguistic divisions, are more closely related than some languages included under the single division "Coast Salish". The Bella Coola provide a second example; they speak a Coast Salish language, but they are considered a separate division because of their geographical location and distinct culture (Duff 1964:12). D u f f s second level of classification, the dialect and regional groups, are classified according to linguistic, geographical, and cultural criteria. These units are "groups which share single languages or major dialects and form 141 convenient and meaningful regional and cultural units" (Duff 1964:15). The au-thor again stresses that: 1) these subdivisions were not recognized by the people themselves, "although their members usually recognized that they all shared the same language, culture, and territory" (Duff 1964:15), and 2) the subdivisions are somewhat arbitrary and inconsistent and the groups are not equivalent in all respects. D u f f s third level of classification is not based on the same criteria. Duff asserts that, at this level, the linguistic criterion fails and "Indian principles of social and political organization" must be used to distinguish and classify the units created and recognized by the Indian people themselves (Duff 1964:16). These smaller units are called "tribes" or bands. They, in turn, are constructed from the small localized groups of people organized around a core of kinfolk who owned the resource areas and lived together throughout the year - either in small, migratory hunting bands or at more sedentary village sites or defined localities (Duff 1964:16-17). Other scholars have considered this question of classification. In his study of the distribution of Northern Athapaskan speaking Indians, Osgood (1936b:3) explains: The greatest difficulty in preparing a map [a distribution map] lies in presenting a consistently equivalent nomenclature. Theoretically, one might proceed from the standpoint of culture, of language, or even of physical type. One discovers, however, that the literature contains hundreds of native terms, or translated equivalents, concerning which it is frequently dubious whether a village, a band, a tribe or a larger grouping is indicated, or whether the distinction is cultural or linguistic . . . The seemingly simple problem of setting down "tribal" names results in complications over the ques-tion of what groups among the Athapaskans should be regarded as tribes (emphasis added). 142 In her introduction to Volume Six of the Handbook of North America Indians, H e l m (1981a: 1-2) admits that the identification and demarcation of Subarctic "tribes" for Volume Six was "in most cases a matter of judgement." Ultimately, a number of difficult and somewhat arbitrary decisions were made in the classification of people. 1 Which consistently equivalent cultural units would form the appropriate cultural units for my sample? Clearly, manj' difficulties are inherent in defining the cultural unit of analysis. The unit of analysis can vary, depend-ing on the criteria used to classify groups and on the level of classification desired (i.e., from major ethnic units to local groups). How can one ensure a representative sample which is not biased when the "cultural groups" are not equivalent - geographically, demographically, or linguistically (i.e., some groups will speak different languages while others speak dialects of the same language). Delimiting cultural units is difficult. F o r example, Jenness (1937:8) asserts that the Beaver and Sekani are so similar that "even today it is impossible to draw a sharp line between Sekani and Beaver Indians". Indeed, he suggests that omissions in the ethnographic material on the Beaver can be filled with data on the Sekani. A second example is provided by Morice. In listing those groups which border the Carrier , he observes, 1 Osgood divides the Northern Athapaskans into twenty-five "major groups" and only two "cultural groups", the Pacific Drainage culture and Arctic Drainage culture. His linguistic classification consists of eight divisions and a number of unclassified groups (1936b:19-22). Volume Six of the Handbook also delimits twenty-five Athapaskan speaking groups (four of these are not included in Osgood's classification, while Osgood includes four which the Handbook of North American Indians does not include in its classification), but it delimits four cultural-areal divisions (based on an environmental - ecological perspective) (Helm 1981a: 1). The linguistic classification contains twenty-three languages (Krauss and Golla 1981:67). 143 But, as to all practical purposes, the Chi Koh'tin [Chilcotin]-and, [sic] indeed, the Western Nah'ane [Kaska?] as well-have [sic] the same general characteristics and, in the main, possess similar social institutions as the Carrier , it may truly be said that the latter's neighbours are: In the south, the Salish . . . the Kawichan in the south-west (1892:11). The last example is provided by M c K e n n a n (1981:563) in his description of the Upper Tanana: It should be emphasized that the grouping of Indian bands treated here under the name T a n a n a is to a certain extent an arbitrary one. The Athapaskans on the T a n a n a and Yukon Rivers, from the Tutchone to the Ingalik, do not fall easily into a number of discrete cultural or linguistic blocks; rather they constitute a continuum of local bands whose respective microcultures and dialects differ only slightly from those of their immediate neighbours. Over a span of several bands the linguistic differences are compounded. 1 A s Aberle indicates in his cross-cultural study of matrilineal societies, "we still do not always have a rule for deciding when we have one culture and when we have two or more" (1961:724). Two criteria are important in defining the appropriate cultural units for this study. First , in order to conduct a statistical study, an adequate sample size is essential. Therefore, the level of classification used to define the cultures must allow for an adequate sample size. The second criterion is ensuring that the cultural units are sufficiently distinct from each other. A t one level, if the units of observation selected for this study were D u f f s major ethnic divisions, for example, the sample would consist of a very few, large groupings. A t the other extreme, if one chose individual villages as the units of observation, it would be difficult to distinguish any variation from He reports that three different languages are spoken among the T a n a n a (McKennan 1981:563). one village to another in terms of their observable traits. In addition, the use of these small, local villages would entail dependency on only one, and occasionally two, ethnographic sources. Consequently, I use the cultural groups or units found in Jorgensen (1980), Osgood (1936b), and the Handbook of North American Indians (1981). These units are basically equivalent to D u f f s second level of classification -dialect and regional groups. M y subsample of Northwest Coast units is based on Jorgensen's (1980) sample in Western Indians. The rest of m y sample consists of the more or less "conventionally accepted" groups as found in Osgood (1936b) and Volume Six of the Handbook of North American Indians. Moreover, where variation exists in the degree of stratification, type of descent, or postmarital residence patterns within a cultural unit, the subgroups displaying this variation are treated as separate cultural units. The cultural units are selected on the basis of three criteria. First , the cultural units must lie within a continuous geographical area from the Northwest Coast eastward. The Chipewyan form the eastern limit of the continuous geographical area. Second, the amount of information available on a cultural unit is an important consideration. Sparsely documented cultural groups are excluded from the sample. T h i r d , time constraints have affected the composition of the sample. For example, the sample does not extend below the forty-ninth parallel, nor does it include non-Athapaskan speakers in the Eastern Subarctic; these arbitrary cut-off points were chosen in an attempt to limit the geographic area and, therefore, the time spent on pr imary data collection. Even with these restrictions in place, the data collection took ten months. C. T H E LIST OF C U L T U R A L UNITS The ethnographic data used in this thesis were obtained for thirty-nine geographically contiguous cultural groups, located in three adjoining culture areas: the Northwest Coast, the Subarctic, and the Plateau. The thirty-nine cultural units are presented alphabetically within their language family classifications: * E Y A K A T H A P A S K A N L A N G U A G E FAMILY 1 A h t n a 2 Bear Lake 3 Beaver 4 Carrier (Alkatcho) 5 Carrier (Bulkley River) 6 Chilcotin 7 Chipewyan 8 Dogrib 9 H a n 10 Hare (Fort Simpson and Fort Good Hope) 11 Ingalik 12 Kas in i 13 K a s k a (Upper L i a r d and Dease River) 14 Kutchin (Chandalar) 15 Kutchin (Peel River and Crow River) 16 Mountain 17 Sekani (Long Grass) 18 Sekani 19 Slave 20 Tahltan 21 Tanana (Upper) 22 Tanaina HAIDA ISOLATES 23 Haida (Massett) 24 Haida (Skidegate) TLINGIT ISOLATES 25 Tlingit (Northern) 26 Tlingit (Southern) SALISHAN FAMILY Coast Salish 27 Bella Coola 28 Cowichan 29 Klahuse 30 Lower Stalo 31 Pentlatch 32 Sanetch (Western) 33 Squamish 34 Upper Stalo Interior Salish 35 Shuswap (Eastern and Southern) 36 Shuswap (Northern and Western) 147 * TSIMSHIAN LANGUAGE ISOLATE 37 Coast Tsimshian 38 Gitksan * WAKASHAN FAMILY 39 Nootka (Central) A s mentioned earlier, some cultural units within this continuous area are not included because of lack of information - (e.g., the Inland Tlingit, the Tutchone, the Kolchan, and the Tsetsaut) . 1 Only one Northwest Coast ethnic division is excluded, the division consisting of Kwakiut l speaking groups. They are excluded because of the massive, confusing amount of material available; this could be the basis for a separate study. Two non-Athapaskan speaking groups from the Plateau culture area are included: the Northern and Western Shuswap and the Eastern and Southern Shuswap. I include these non-Athapaskan speaking Plateau groups because it would be of interest to learn whether the results for these groups conform to the results obtained for the other egalitarian cultures, which are all Athapaskan speaking groups. Unfortunately, time constraints prevented the inclusion of more than these two non-Athapaskan speaking groups. I have completed the discussion of the continuous area sample. One cautionary note is introduced at this point. A continuous area study represents a universe (i.e., a whole population of cultural units within a geographic area) and not a "sample" of a universe. This has two 1 I did not learn of sources on certain groups until later (e.g., Legros 1981 on the Tutchone and Gillespie's work on the Yellowknife, 1975). B y that time, I had completed m y data collection. implications. First, the empirical generalizations obtained from this statistical analysis apply only to the geographical area in question (Jorgensen 1974:198). They should be extended to the whole world with great caution. Second, because the cultural units are not a random sample, certain statistical tests cannot be used. Some scholars suggest that statistics based on "proportional reduction in error" are preferable for this type of study. I take up the question of appropriate statistics in the next chapter. V. T H E STATISTICAL METHOD This chapter describes the statistical technique and the statistics 1 used to test for the existence of relationships between the dependent and independent variables. The first section delimits the advantages and the limitations of using a statistical approach. The next section describes and discusses the statistical technique and the statistics. The last section is a discussion of in-teractive effects. A. T H E STATISTICAL APPROACH A major reason for using a statistical model to study cultures is the ability to test empirically for relationships between variables on a large num-ber of cases. Relationships are often reported in the literature based on evidence from one or two cultures. Consequently, there can be little reliance on the results. Only after a hypothesis has been tested and validated using a variety of data can one begin to have some confidence in a postulated relationship. Us ing a statistical approach to analyze relationships among cultures is not without its limitations, however. A n awareness of these limitations is 1 The term "statistics" has a number of meanings. According to Derek Rowntree's "primer for non-mathematicians", the word statistics has four meanings: F ir s t of all, it can indicate, very broadly, a whole subject or discipline . . . Secondty, and more specificalty, the term may refer to the methods used to collect or process or interpret quantitative data. Thirdly , the term may be applied to collections of data gathered by those methods. A n d fourthly, it m a y refer to certain specially calculated figures (e.g., an average) that somehow characterize such a collection of data (1981:17). 149 150 important to an understanding of the implications of the findings. One limitation is that variation within a cultural group, by design, is not incorporated into a statistical analysis. E a c h cultural unit in this analysis is assigned to one category of the dependent variable - greater access, equal access, or less access. Hence, the complexity and variation within each culture are disregarded. A s Whyte ackowledges in his study of women's status, "Al l of this means that people interested in differences in, and subtle features of, the status of women within societies, say between high and low classes or between rural and urban residents, will find little direct evidence" (1978:19). Similarly, it is difficult to study variations over t ime . 1 The infor-mation on cultural units which is coded often refers to only one point in time; information on other time periods is not incorporated into the coding. A further, limitation is the inability to make use of general statements found in the literature. Unless the information provided is fairly specific, coding is problematic. Niblack (1890:239), for example, reports the existence of "political and industrial equality of the sexes" among northern Northwest Coast groups. Although this information relates directly to the topic of women's access to power, it is unproductive because he does not define "political" and "industrial" equality. One might argue, of course, that the val-ue of this type of information is limited in any kind of scientific inquiry. Thus , this situation might be considered an advantage rather than a limitation since it provides a check against using vague, unqualified statements which are, unfortunately, used all too frequently as evidence in studies of women's status and power. 1 Cross-section time series analysis is possible, but it entails a more complex statistical formulation than has been used by anthropologists to date. 151 B. T H E STATISTICAL TECHNIQUE This section outlines the statistical technique used to analyze the relationships between the dependent and independent variables. In all statistical studies, variables are used to measure theoretical constructs in terms of readily observable data. To choose an appropriate statistical technique, one must first consider the level of measurement of the variables. In this study, all the variables are measured (coded) in terms of categories. M y dependent variables, for example, have three categories or values: men having access that women do not have (i.e., less access for women), equal access, or women having access that men do not have (i.e., greater access for women). E a c h of these categories can be ordered along a continuum. Therefore, the dependent variables are ordinal level variables. A n ordinal variable is differentiated into categories which can be ordered by degree or ranked along a continuum, for example, from high to low or small to large (e.g., less access, equal access, greater access; Blalock 1979:16-18). The three independent variables are also ordinal level variables. Degree of stratification has the categories egalitarian, intermediate, ranked. E a c h category can be ranked or ordered according to its degree of stratification. The variable, type of descent, is actually a nominal level variable. A nominal level measure "consists of 'mutually exclusive and non-ordered categories' . . . they constitute different 'values' or alternative possibilities of some, more general, concept or class" (Pelto and Pelto 1978:143). T h a t is, due to the nature of the categories, no assumptions can be made about ordering them. Type of descent can be treated as an ordinal variable, however, because it takes two values, matrilineal or other. According to Nie 152 et al., the creators of the Statistical Package for the Social Sciences or SPSS Programme, a dichotomous, nominal level variable (i.e., a variable with two categories) can be treated as an ordinal level variable. T h a t is, a rank ordering may be imposed on the variable by treating one category as the "high" and the other as the "low" category. It is immaterial which category is treated as the high and which the low; if the categories were switched, however, the measures of association would change sign (the sign indicates the direction of the relationship) (1975:5-6, 227). The third independent variable, final postmarital residence pattern, is also regarded as an ordinal level variable in this analysis. M y basis for rank ordering residence is that, as one goes along a continuum from matrilocal residence to patrilocal residence, women's status is said to decrease. Avunculocal residence can theoretically be placed part way between the two on the continuum, since it is associated with the same descent system as matrilocal residence and is somewhat more positively associated with women's access than patrilocal residence, but less than matrilocal residence (see M a r t i n and Voorhies 1975, Nielsen 1978). Thus , the categories of residence are ordered along a line or continuum from matrilocal to avunculocal to bilocal and patrilocal (other residence). Contingency table analysis (the technique is also known as Crosstabulation), an appropriate technique to use with nominal and ordinal level variables, was chosen for this study. A contingency table provides the joint frequency distribution of cases along two or more dimensions. That is, it displays the distribution of observed cases in terms of two or more varia-bles. 153 A variety of statistics can be derived from a contingency table. These statistics are measures of association and tests of significance. A s with the choice of a statistical technique, the measures of association and tests of significance chosen depend on the levels of measurement of the variables. The statistics selected for use in this analysis are from the set of statistics computed by the Statistical Package for the Social Sciences (Nie et al. 1975) or S P S S Programme, a preexisting statistical package created for social scien-tists. A measure of association indicates how strongly two or more variables (i.e., the dependent and independent variables) are related. In other words, a measure of association indicates the extent to which categories of one variable and those of another variable occur together. Two variables m a y appear strongly related to each other, yet the relationship itself m a y be due to chance. F o r this reason, tests of significance are often used jointly with measures of association. Tests of significance are used to determine the probability that a relationship between two variables is simply the result of chance. If this probability is at or below a pre-selected value or "level of significance", the relationship is said to be statistically significant; that is, the relationship is probably not due to chance. The level of significance conventionally accepted is five percent; that is, in only five cases out of one hundred, the relationship observed between the variables is due to chance (Nie et al. 1975:222). There is one complication with using tests of significance, and measures of association based on them. I indicated at the end of the last chapter that, because this analysis is a continuous area study rather than a random 154 sample (i.e., a sample in which each element of the universe has an equal chance of being selected for the sample, Erickson and Nosanchuk 1977:128-29), there is a problem with using inferential statistics (i.e., statistics created for use with random samples). Nevertheless, the use of such statistics can be justified. Pelto and Pelto, in their review of methods of anthropological research, observe that although a "statistical purist" would insist that inferential statistics cannot be used on a nonrandom sample, in actual fact "some statistical analysis is better than none at all". One can still present useful information about the probabilities of certain relationships occurring by chance (Pelto and Pelto 1978:265-66). Pelto and Pelto conclude their discussion of nonrandom samples by suggesting that such investigations might be regarded as exploratory and the results seen, not as definitive, but as supporting evidence in association with other studies using other samples. With these points kept in mind, using statistics based on the assumption of a random sample can provide valuable information which furthers our under-standing of some factors which are related to women's access to power vis-a-vis men. Certainly, there is a long tradition in anthropology of using inferential statistics such as the chi square test and other statistics based on chi square to test the relationships among variables in continuous area studies (see Driver 1966, Driver and Coffin 1975, Driver and Massey 1957, Driver and Schuessler 1957 in Jorgensen 1974:176-195, Jorgensen 1969, Jorgensen 1974:195). Evidence suggests that certain statistics are appropriate for use with a continuous area study. These P R E measures are based on the concept of "proportional reduction in error". They measure the proportion by which 155 "uncertainty" in the dependent variable is reduced by having some knowledge about the independent variable (Nie et al. 1975:225-226). The formulas of P R E measures are not based on an assumption that a sample is random. Kendall's T a u B and C , the measures of association used in this analysis, are P R E measures (Loether and M c T a v i s h 1980:234). Thus , Kendall's T a u B and C are appropriate for use in a continuous area study. Moreover, because a continuous area study is really a universe rather than a non-random sample, as such, the interpretation of the results are straight forward; that is, I. do not have to worry about the implications of generalizing from a non-random sample to a universe. Kendall's T a u B and C assume that the variables are ordinal. These T a u values are calculated by using information on the ordering of the categories of variables. Because we have information on the ranking of the categories, these statistics use the rank of pairs of cases on the two ordinal variables. E v e r y possible pair of cases in the contingency table is considered in order to determine whether or not knowledge about the rank ordering of pairs of cases on one variable is useful in predicting their rank order on the other variable. E a c h pair of cases is considered by checking whether its relative ordering on the first variable is the same as its relative ordering on the second variable (concordant), or whether the ordering is reversed (discordant). If the cases are at the same position on one or both variables, the pair is tied (Nie et al. 1975:227-228). Kendall's T a u B and C consider ties, an important factor in this analysis, since there are a number of ties. If the knowledge of ranking of pairs on one variable is of no use in predicting rank order on the other variable, then the measure of association 156 will have a value of zero. This situation would be equivalent to predicting the rank order of cases randomly by tossing a coin (Loether and M c T a v i s h 1980:227-28). Kendall's T a u B is used for tables where the number of rows and columns are equal. T a u C is used where the number of rows and col-umns are unequal. Kendall's T a u B and C give information on the direction as well as the strength of the relationship. A positive association exists when one can pre-dict the same rank order on the second variable as the pair had on the first variable. A negative association exists when one can predict the opposite rank order of cases on the second variable as opposed to the first. A s mentioned above, a problem exists with using tests of significance with a continuous area study because they are based on nonrandom samples. O n the other hand, as mentioned earlier, inferential statistics are of value if their limitations are kept in mind and, as Aberle indicates in his cross-cultural statistical analysis of matrilineal descent in Matrilineal Kinship, it seems advisable to have some device to decide whether it is reasonable to assume that the data show a trend, or to assume that the results could have arisen by chance (1961:674-75). Some indication of statistical. significance is of value in analyzing the measure of association. Therefore, although it is not treated as definitive, a test of significance is used in this analysis. The more commonly used test of significance, the Chi Square Test, was rejected because it has a number of unique problems (Blalock 1960:220-21, 225-26, Pelto and Pelto 1978:152). Fortunately, the S P S S package provides a significance value for Kendall's T a u B and C . This significance value gives the probability that the value of Kendall's T a u B or C is the result of 157 chance. Thus , if the significance value is at or below the level of .05, then the strength of the association we observe between the dependent and independent variables did not arise by chance (Blalock 1960:324). There are situations, however, when the significance value associated with T a u is not provided. In these cases, a second test of significance is substituted. This test of significance is called an Exact Probability Test. It is the exact probability of obtaining the result observed in the Contingency Table plus the probabilities of getting even more unusual outcomes in the same direction as those observed in the table (Blalock 1960:221-222). For two by two tables, the S P S S Programme automatically calculates the exact probability. It is called the Fisher Exact Probability. A n exact probability is not calculated by the S P S S Programme for tables with more than two rows or columns, nor for tables with more than twenty-one cases. Thus , a custom written programme, not available in any of the standard statistical routines available at the University of British Columbia, was written which calculates an exact probability for contingency tables of all dimensions. When an Exact Probability value is provided in the thesis, it is this statistic which is being reported. C. INTERACTIVE EFFECTS The final topic of concern in this chapter is that of interactive effects. Independent variables can affect a dependent variable in two ways, in an additive manner or an interactive manner. A n additive effect is the specific effect of each independent variable on the dependent variable. There may, however, be another or an additional effect caused by the interaction of two 158 or more independent variables . 1 When the effects of the independent variables on a dependent variable are additive, the relationship between the dependent variable and any given independent variable is the same across all the categories of the third independent variable which is being held constant. When an interactive effect exists, the effect of the given independent variable varies from one category of the remaining independent variable to another because the remaining independent variable is affecting the relationship between the dependent variable and the given independent variable (Nie et al. 1975:372, 403). Interaction between two independent variables is manifested in a number of ways. First , an interaction may produce a spurious correlation between an independent and a dependent variable. This occurs when two independent var-iables are closely related. Consider the case of a dependent variable X , which is closely related to an independent variable Y , but which is not related to the independent variable Z . In addition, Y and Z are closely related. In this case, a crosstabulation of the dependent variable X by the independent variable Z may show a significant relationship where none exists because the variable Z , which is actually related to Y , is picking up the correlation between X and Y . If one repeats the crosstabulation between X and Z but controls for Y , the relationship between X and Z would disappear. Second, interaction between two variables can also obscure the relationship between independent and dependent variables. This would occur if the 1 This interactive effect is often ignored in anthropological studies. Jorgensen (1974:199) notes that with the exception of three studies, the practitioners of cross-cultural studies have seldom tried to analyze more than one pair of varia-bles at a time. A n examination of interactive effects, done here, is one means for analyzing more than two variables at a time. 159 dependent variable X is positively related to the independent variable Y , the dependent variable X negatively related to the independent variable Z , and Y and Z are positively related. A crosstabulation of X by Y alone would actu-ally be an amalgam of the positive effect of Y on X and the negative effect of Z on X . Thus , the two variables obscure each other's effect. Again , by controlling for one independent variable, this relationship would be made clear. B y controlling for one independent variable and comparing the uncontrolled tables with the controlled tables, one can learn more about the relationships among the three variables. Nie et al. (1975:413) caution, however, that where a model contains a large number of nominal or ordinal level independent variables, the interpretation of interaction effects may be difficult. I hoped to control simultaneously for two variables at a time while analyzing the pure effect of the third variable on the dependent variable. Unfortunately, because of m y small sample size, this was not possible. Thus , I have analyzed the relationship between one independent and the dependent variable, holding only one variable constant at a time. The statistics described above are used to analyze the data on women's access to power and autonomy. In the next two chapters, the results of the analysis are reported and discussed. VI. PRELIMINARY ANALYSIS This chapter describes and discusses the frequency distributions of my de-pendent and independent variables. That is, I describe the variation in the categories of variables. The frequency distributions of my dependent variables are given first, followed by possible reasons for the distributions. Finally, the frequency distributions of my independent variables are presented. A. THE DEPENDENT VARIABLES 1. The Distributions In Chapter 1 I stated there are five categories for each dependent vari-able: (1) women have less access than men to power in various spheres and to autonomy, (2) women and men have equal access to power in various spheres and to autonomy, (3) women have greater access than men to power in various spheres and to autonomy, (4) insufficient information on men's and women's access to enable coding, and (5) role is non-existent in the culture. The first three categories are the primary categories in this study; the last two may be regarded as missing values. In the final analysis, the fifth category was combined with the second one. This matter is taken up later in the chapter. Table I presents the frequency distributions of each dependent variable (i.e., the distribution of cultures across the categories of my dependent varia-bles). Seven of the thirty dependent variables in Table I have cases in just one primary category, equal access for men and women.1 The seven 1 Contingency tables were not constructed for these dependent variables. 160 161 Table I: Frequencies of Gender Access to Power and Autonomy Women Have Less Access E q u a l Access Greater Missing Access (Unknown) Dependent Variable # % # % # % # % S U P N A T . P U B _ 38 97 1 3 S U P N A T . P R I V - 31 79 - 8 21 L E A D R 16 41 14 36 - 9 23 C O M M U N . I N F L 3 8 18 47 - 18 46 C E R E M . H O S T - 25 64 - 14 36 D O M E S T . D E C 1 3 18 46 4 10 16 41 S U B S I S T . L E A D R 2 5 14 36 - 23 59 T R A D E / R E C . D E C 3 8 11 28 - 25 64 I N H . I N D I V . P R O P - 30 77 2 5 7 18 I N H . C E R E M . P R O P - 23 59 16 41 I N D I V . P R O P - 30 77 - 9 23 M A N . K I N . C H A T T 10 26 22 57 - t 18 M A N . K I N . P R O P 12 31 21 54 - 6 15 M A N . K I N . I N T A N G 1 3 33 84 - 5 13 O W N . O W N . F R T S - 27 69 5 13 7 18 N O N - D O M E S T . F R T S + W E A L T H 2 5 22 57 . 15 38 C O - O P . F R T S 1 3 7 18 17 44 14 36 M A R . A R R A N G - 33 85 5 13 1 3 E D . C H I L D - 31 79 2 5 6 15 S I B . L E A D R - 11 29 - 28 72 P R E M A R . S E X - 28 72 1 3 10 26 O W N . M A R . D E C 4 10 31 79 - 4 10 E X M A R . S E X 2 5 28 72 - 9 23 D I V O R C E 1 3 33 85 - 5 13 R E M A R - 30 77 1 3 8 21 D O M E S T . A U T O N - 32 82 7 18 Present Absent Missing (Unknown) Dependent Variable # % # % # % W . O T H R . R O L E S 24 62 1 3 14 36 M . O T H R . R O L E S 9 23 1 3 29 74 W . F A T E . A F T . M A R 16 41 3 8 20 51 M . F A T E . A F T . M A R - 1 3 38 97 162 dependent variables with no variation are: S U P N A T . P U B , SUPNAT.PRIV, C E R E M . H O S T , INH.CEREM.PROP, INDIV.PROP, SIB.LEADR, and D O M E S T . A U T O N . The frequency distributions of these variables are presented in Table I. In addition to these variables which indicate equal access between men and women, there are eleven variables in which variation in a second primary category is limited to less than ten percent of the total variation (excluding the missing cases). These variables, INH.INDIV.PROP, M A N . K I N . I N T A N G , N O N - D O M E S T . F R T S & W E A L T H , E D . C H I L D , P R E M A R . S E X , E X M A R . S E X , DIVORCE, R E M A R , W.OTHR.ROLES, M . O T H R . R O L E S , and M . F A T E . A F T R . M A R are excluded from the statistical analysis. The remaining variables in Table I show some variation in. their distri-bution of observations among the primary categories. Notably, for most of the variables in this subset of the dependent variables, the variation is limited to two of the three primary categories. The variables, analyzed statistically in this study, are L E A D R , C O M M U N . I N F L , D O M E S T . D E C , SUBSIST.LEADR, T R A D E / R E C . D E C , M A N . K I N . C H A T T , MAN.KIN.PROP, OWN.OWN.FRTS, CO-OP.FRTS, M A R . A R R A N G , O W N . M A R . D E C , and W . F A T E . A F T . M A R . The results of the analysis are presented in Chapter 7. 2. Factors Influencing the Distribution of Dependent Variables One must consider why a number of the dependent variables indicate equal access to various aspects of power for men and women. A number of factors may have influenced this outcome. 163 To begin, the definition of m y dependent variable in terms of access may have lowered the degree of variation in m y measures of women's power and autonomy. Specifically, by using the access measure, I do not consider the ethnographic information on frequencies of participation and filling roles or the amount of power said to be held by men and women. A s explained in Chapter 1, I use the concept of access for three reasons. First , I use it in order to minimize the biases and subjective interpretations of ethnographers (see Chapter 1 and Chapter 3). Second, I use access to obtain a consistent basic level of data in coding the dependent variables. Some authors provide information such as the frequencies of participation or compare the amounts of power available to men and women, while others do not. A s I explained in Chapter 1, the differences in the . amount of detail and information provided by ethnographers results in difficulties in obtaining comparable measures of power across the different de-pendent variables and across observations. The use of the access measure is one way to handle this problem (see Chapter 1 for a more detailed discussion). The. third reason for using the access measure is also motivated by the state of the ethnographic literature. For many cultures in m y sample, the ethnographic accounts of women's participation in various spheres of their society are in conflict. I provided the example of S U P N A T . P U B (dealing with supernatural power which is used for others' benefit) in Chapter 1, noting that this dependent variable is not unusual in terms of the contradiction among the ethnographic accounts for one culture. The access measure goes a long way in alleviating these conflicts, since authors are more likely to agree 164 that women perform certain roles or participate1 in activities than they are to agree about frequencies with which they perform certain roles or the amounts of power they hold, etc. The access measure makes the operation of coding manageable and maintains a basic consistent level of data. However, as mentioned, this is at the cost of perhaps muting some of the distinctions which could be made about differences in access. Coding procedures are not the only cause of the lack of variation in certain dependent variables. There is a second possibility. Women may actu-ally have more access to certain aspects of power and autonomy than is as-sumed in the anthropological literature - the assumption being based on the anthropologists' cultures rather than on information about the cultures under discussion.2 Some examples are presented from this study. INH.CEREM.PROP considers the right to inherit individually owned ceremonial and intangible property. Of the thirty-nine observations, sixteen are coded "missing informa-tion". Of the remainder, eight (twenty-one percent) are coded equal access because men and women equally have the right to inherit such property. There were no contradictions in the accounts, nor were there any claims that women inherit less frequently or that the property inherited by women is less valuable. The remaining cases (thirty-eight percent) are coded equal access because neither men nor women inherit such property; individually owned ceremonial property does not exist. Hence, there is no opportunity for one gender to have control of such property while the other gender does not. 1 Even this level of agreement is not always obtained. 2 Whyte's results (1978:79) support this possibility; he found there are "many areas of substantial equality between the sexes". 165 Similarly, an examination of the dependent variables which are concerned with access to premarital sexual freedom and extramarital sexual freedom provides some interesting insights. Contrary to assumptions in the literature, the sexual freedom of men and women is equally limited. O f the twenty-nine observations on premarital sexual freedom, twenty-one cases (Fifty-four percent), clearly indicate equal access or lack of access. O f the remaining eight cases which have conflicting accounts, onty three indicate the existence of a double standard in favour of men. For all three, other authors contradict this claim, stating that neither men nor women have access to premarital sexual freedom. For the majority of the observations indicating that neither men nor women had this type of freedom, a man was more severety punished than a woman for transgressions. O f the thirty observations on extramarital sexual freedom, four cases cited women as more likely to be punished; in all four cases, the data are contradicted by other data. In twelve cases, the man's punishment is more severe. The level of punishment was not considered in determining access - partly because this type of information was not consistently provided and partly because of the difficulty in assessing the degree of severity of a punishment. The basic point is that such behavior is unacceptable and will have consequences for both individuals. 1 The final example of equal access presented here is the variable measuring access to the management or control of kin group owned intangible 1 Both Whyte and Gough provide some support for the observed equality. Whyte's findings provide empirical evidence that, cross-culturally, men and women have equality in the areas of divorce and freedom to make one's own marriage arrangements (1978:74, 78). Gough has argued that in societies without great societal complexity (i.e., hunting and gathering societies and societies based on horticulture), women have control over their own sexuality, divorce, remarriage, and autonomy of movement (1975:69-70). 166 property, M A N . K I N . I N T A N G . O f the thirty-four observations on this dependent variable, only one observation is coded less access. The remainder are coded equal access. The equal access of men and women in this area is rather unexpected, given the strong relationships between stratification and the other two dependent variables measuring the management of other types of kin group owned property, i.e., chattels and immovable economic property. The distinctive nature of the intangible property maj' help to explain the difference. The titles, crests, and other intangible property belonging to a kin group are essential proof of kin group membership for all kin group members. They are also signs of rank (e.g., Sapir 1915:6, 28). Thus , all kin group members obtain some intangible property in the form of names. Higher ranked people will have even more intangible property. While ceremonial ob-jects which are movable and economic property which is not, are also forms of kin group property, these generally remain with the core of the kin group which is located in one place. Distance obviously makes the control of such property difficult for a woman residing away from the natal home after marriage. Intangible property, on the other hand, need not be tied to the lo-cation of the kin group. Thus , logically, those who remain in the family home will be put in charge of the family property which cannot be moved or which is used on ceremonial occasions in the family home. Those who do not permanently reside in the kin group home, however, can logically have rights over intangible property. Hence, men and women equally have access to intangible kin group property and pass it on. Indeed, Olson records that pubescent Tlingit girls were told, "you alone can give the names of the noted people of our family line [to 3rour children]" (Olson 1967:49). Only Tlingit 167 women can pass on kin group membership and status within the kin group. 1 Although some might argue women have "less status" because they have access to control over intangible kin group property but not kin group owned economic property (chattels and immovable property), this claim cannot be supported by the data on intangible property. M a n y authors have noted the overwhelming importance of intangible property such as crests, names, and privileges (e.g., Oberg 1973:51-52, Suttles 1951:500). Halpin (1984:60), for instance, points out that among the Tsimshian, names are wealth. Names controlled property and privilege and allowed people to participate in potlatch wealth transfers. Thus , the intangible property held by high status women and men is of great value in those cultures where it exists. W h y should m y results indicate women have equal access to certain aspects of power when it is commonly assumed they do not have such power or autonomy? These results may be unique to m y sample. The cultures used in this thesis consist solely of hunting and gathering societies. Although some of these societies are regarded as having greater societal complexity than is true for m a n y hunting and gathering societies in general, the level of complexity is not great in comparison to agricultural, industrial, and even some horticultural societies. Both Gough (1975) and Whyte (1978:139) argue that a lesser degree of societal complexity has a smaller effect on women's status and power than a greater degree of complexity would. Gough indicates 1 It may be significant that in his study of the Coast Salish, Barnett (1955:250) acknowledges that men and women own privileges equally. He adds, however, that important men have "the most important" privileges. This statement is not explained in any way. One is left wondering how degree of importance is deter-mined and how important women fare in this matter. 168 that in hunting and gathering and in horticultural societies, women are well-off (although even here, they do not have total equality. Thus , even in these societies, matrilineal descent and matrilocal residence will have a positive influence on women's personal autonomy and status). She maintains that with the growth of class societies, women's subordination increases greatly (1975:75). Whyte's results are quite specific. They indicate the shift from a subsistence pattern based on hunting and gathering to one based on horticulture is less important for women than the further shift to plow agriculture. A s argued in Chapter 2, since the level of societal complexity in this study is between the levels described by the scholars whose models I reviewed (i.e., they compare egalitarian hunting and gathering societies and horticultural societies), if the models are correct, my results should indicate fewer differences in women and men's access to power. It may be that the degree of stratification found in my sample (which is an indicator of the level of societal complexity) has some effect on certain areas of women's lives (particu-larly economically and politically), but not enough to influence other areas of their lives, as could be expected for societies with even greater societal com-plexity. 1 Further research of a cross-cultural nature would be necessary in order to test this hypothesis. 3. Non-existent Role or Institution Before turning to the distributions of the independent variables, I consider whether the category 'nonexistent role or institution' should be treated as a missing value. This thesis will not treat it as a category of missing data, 1 It may be that autonomy in women's private lives is maintained longer than power in the political and economic spheres. 169 but rather as one of the pr imary categories. Thus , the categories equal access and nonexistent role or institution have been combined into the sin-gle category, equal access. The decision to recategorize the cases of nonexistent role or institution was arrived at in two stages. The impetus for the change came with the recognition that the treatment of cases as missing when neither sex has access to a non-existent role does not make optimal use of information. This inadequacy became apparent when contingency tables were constructed for some of the dependent variables - in particular, I N H . P R I V . P R O P , M A N . K I N . C H A T T , M A N . K I N . P R O P , N O N - D O M E S T . F R T S & W E A L T H , and C O - O P . F R T S . These variables deal with various aspects of men's and women's access to economic power. Egalitarian societies do not have such roles or rights. Some ranked societies not only have them but, more importantly, allow women less access to them. This important . association could not be analyzed, however, when all of the cases in egalitarian societies were coded as missing information. The decision to recategorize nonexistent as equal access was made in an attempt to correct this problem. A precedent exists for treating cases of nonexistent roles or institutions as cases of equal access. Whyte (1978:155) does so, arguing that in cases where neither men nor women have access to a role or right, neither gender has the oppportunity to manipulate status not available to the other gender. Where neither men nor women have access to power, access is equal. A l l cases originally coded nonexistent role or institution have been recoded equal access. The main effect of recoding is to increase the overall sample size and reduce the number of missing cases. 170 B. THE INDEPENDENT VARIABLES This section considers the independent variables: degree of stratification, type of descent, and final postmarital residence. For these variables, I shall discuss the lack of missing data, the distribution of the cases across their categories, and relationships among the three independent variables. Table II provides information on the distribution of cases among the categories of each variable. Although it was difficult to categorize the cultures in terms of the independent variables, the difficulty was rarely due to insufficient information. In fact, there are no missing cases for the independent variables. This is due, primari ly , to the interest of ethnographers in the three independent variables used in this study. While the ethnographic accounts do sometimes contradict each other, few omit discussion of these topics. For the variable degree of stratification, the largest number of cultural units fall in the category, ranked. Almost one-half of the cultural units (eighteen cases/forty-six percent) are coded ranked; thirty-six percent of the cultural units (fourteen cases) are coded egalitarian; and eighteen percent of the units (seven cases) are coded intermediate. The distribution of the cases between the two categories of the variable type of descent is almost equal; forty-four percent of the cultural units (seventeen cases) are coded matrilineal and fifty-six percent (twenty-two cases) are coded other descent. The final postmarital residence pattern is more unevenly represented. Fifteen percent of the cases (six cultural units) are matrilocal residence; twenty-one percent (eight cultural units) are avunculocal residence. Twenty-five cultural units are coded other residence (sixty-four percent). Table II: Frequencies for Stratification, Descent, and Residence Categories I N T E R M E D # 7 18 R A N K E D # % 18 46 O T H E R # % 22 56 A V U N C # % 8 21 O T H E R # % 25 64 Independent Variable Degree of Stratification E G A L # % 14 36 Type of Descent M A T R I # % 17 44 Fina l Postmarital Residence M A T R I # % 6 15 172 Table III: Correlations Among the Independent Variables Descent X Degree of Strat MATRI OTHER TOTAL EGAL 4 10 14 INTERMED 4 3 7 RANKED 9 9 18 TOTAL: 17 22 39 Exact Probability: .11498 Kendall's Tau C : -.19198 Sig: .1311 Residence X Degree of Strat MATRI AVUNC OTHER TOTAL EGAL 5 - 9 14 INTERMED 1 - 6 7 RANKED 8 10 18 TOTAL: 6 8 25 39 Exact Probability: .00030 Kendall's Tau B : .02068 Sig: .4445 Descent X Residence MATRI OTHER TOTAL MATRI 5 1 6 AVUNC 8 8 OTHER 4 21 25 TOTAL: 17 22 39 Exact Probability: .00000 Kendall's Tau C : -.68639 Sig: .00000 173 Table III presents the results of cross-tabulating the frequencies of the independent variables, two by two. The values of Kendall's T a u B and C and the test of significance are also included. Crosstabulating descent by the degree of stratification brings to light an interesting interrelationship. Matri l ineal and other descent are found equally among ranked and intermediate societies. Egalitarian societies, however, tend to have other descent. I do not know whether this concentration is typical of all egalitarian societies (as compared with ranked and intermediate societies), or whether it is a phenomenon limited to the geographical area under consideration in this thesis. The relationship between type of descent and degree of stratification is insignificant. T a u C indicates a weak, insignificant relationship. These two independent variables do not appear to be closely associated. When the postmarital residence pattern and degree of stratification are crosstabulated, other residence is found across the three categories of degree of stratification. Avunculocal residence is limited to ranked societies, however, while matrilocal residence is limited to non-ranked societies. Thus , there seems to be some correlation between the final postmarital residence pattern and the degree of stratification. In terms of the statistical analysis, this association between residence and the degree of stratification implies that it is difficult to distinguish the individual effects of each variable. The relationship is difficult to analyze statistically. The Kendall's T a u statistic indicates a weak relationship, and the significance level associated with it indicates statistical insignificance. A t the same time, other statistics generated by the relationship (e.g., the Exact Probability) indicate the relationship is highly statistically sig-nificant. O n closer examination, it seems the relationship between residence 174 and stratification is significant, but is non-linear, hence the low T a u value . 1 It is, therefore, difficult to ascertain the nature of the relationship between residence and degree of stratification. A crosstabulation of descent by residence demonstrates these two varia-bles are correlated. Matril ineal descent is observed across all postmarital residence categories (there is a tendency to find slightty more of them among societies with avunculocal residence). O n the other hand, virtually all other descent is associated with other residence. This supports many claims that type of descent and postmarital residence pattern are interrelated (e.g., see Gough 1961:551-554, M a r t i n and Voorhies 1975:223). The relationship between descent and residence is extremely strong according to the T a u value. In addition, it is highly statistically significant. Thus, the effect of each is difficult to untangle from the other. This description of the distributions of the variables provides the background information for the next chapter which presents the results of the statistical analysis. 1 The effect of avunculocal residence, documented in Chapter 7, provides further evidence that the relationship is a non-linear one. VII. T H E STATISTICAL ANALYSIS The statistical analysis provides a number of insights into women's access to power and autonomy vis-a-vis men's and how access is related to the independent variables: degree of stratification, type of descent, and final postmarital residence. This chapter discusses these findings. In Section A, the relationships observed between the dependent and independent variables are described and discussed. In Section B, the findings are summarized. A. THE ANALYSIS The relationships between the dependent and independent variables in this study are analyzed on the basis of one measure of association, Kendall's Tau. I chose Kendall's Tau because: 1) it indicates the direction of the relationship, 2) unlike certain other measures, it is not affected by the distributions of the marginal totals,1 3) Tau can be used when there are large numbers of ties (an important consideration for my data), and, finally 4) there is a significance level for this measure of association. A level of significance of .05 is used as the cutoff level for determining statistical significance. This means that the association would occur due to chance in less than five cases out of one hundred. In some cases, the Exact Probability is used as the significance test. This occurs when the significance level associated with Kendall's Tau is not available. The significance level provided by the Exact Probability is not tied to the Tau values; because of this, the Exact Probability values are consistently less significant than the other significance test values. 1 Tau only reaches unity when all the cells are empty except those on the di-agonal (Blalock 1960:323). 175 176 Table IV: Range of Values of Kendall's T a u for the Relationships Between the Dependent and Independent Variables Weakest Strongest Average Relationship Relationship Relationship Stratification .018 .749 .435 (Tau B) .218 (Tau C) Descent .000 .520 .208 (Tau B) .142 (Tau C) Residence .000 .299 .240 (Tau B) .125 (Tau C) In Table I V , I have compiled the range of values of Kendall's T a u gen-erated by the associations in m y data. Included in Table IV are: the smallest (weakest) recorded relationship, the largest (strongest) recorded relationship, and the average value of the relationships across the thirty de-pendent variables for each of m y independent variables (since I a m concerned with the magnitude or strength of the measures of association, the absolute values of the statistics are compared; the directions of the relationships are ignored). The T a u statistics range from 0.0 to .75, with the average values ranging from .125 to .218 for T a u C , and .208 to .435 for T a u B . This means the associations observed in this study are relatively weak. A glance at the table indicates, however, that the values for the relationships with the independent variable stratification tend to be larger than those for the other independent variables. T h a t is, the relationships between stratification and the dependent variables tend to be stronger than those between descent and the dependent variables or residence and the dependent variables. Before discussing the relationships between the dependent and independent variables, I shall define some of the terms and concepts used. First , "stratification", "descent", and "residence" are employed throughout as short forms for the independent variables degree of stratification, type of descent system, and final postmarital residence pattern. "Tau B " and "Tau C " refer to Kendall's T a u B and C . The term "a negative relationship" indicates that as the independent variable increases, the dependent variable decreases. F o r example, in this thesis, a negative relationship between stratification and women's access to power means, as stratification increases, women's access decreases (from equal access to less access or from greater access to equal access). In terms of descent, as descent goes from matrilineal to other, women's access decreases. Final ly , in terms of residence, as residence goes from matrilocal to avunculocal to other, women's access also decreases. Similarly, "a positive relationship" means that as the independent variable increases, women's access vis-a-vis men's also increases. 1. Interactive Effects A s explained in Chapter 5, an interaction or interactive effect is an ef-fect produced by the interaction of two or more independent variables on a dependent variable. When the effects of the independent variables are additive, "the relationship between the dependent variable and any given independent variable is the same across all values of the remaining independent variables" (i.e., the remaining independent variables have no effect on the relationship, Nie et al. 1975:372). When an interactive effect exists, the effect of the given independent variable varies from one category of the 178 remaining independent variable to another because the remaining independent variable is affecting the relationship between the dependent variable and the given independent variable (Nie et al. 1975:372, 403). A s an example, one might assume a positive relationship between level of education and income, but if a third independent variable "sex" is held constant, one might observe that the relationship between level of education and income is weaker (and/or less significant) for women than for men. One's sex affects the relationship between education and income. To test for interactions, partial associations are computed - that is, I control the effect of the remaining independent variables so the effect of the controlled independent variables are removed. Unfortunately, because my sample is small, only one independent variable can be held constant at a time. Thus, the effect of the independent variables can only be removed one at a time. One note of caution must be added regarding interactive effects; it is sometimes difficult to interpret interactive effects in nominal and ordinal level data. Although a number of interesting partial associations are observed for the relationships examined in this thesis, only consistent patterns are described and analyzed; discussing each partial association in detail would be a monumental and thoroughly confusing task, due to the amount of material and the detail involved. To briefly summarize these patterns, the most obvious and recurring in-teractive effect involves the effect of descent on relationships between stratification and eight of the dependent variables. In each case, in cultures with matrilineal descent, the impact of stratification on the dependent variables 179 is weaker, while in cultures with other descent, stratification is strongly associated with the dependent variables. Matril ineal descent appears to counteract or dampen the effect of stratification on women's access. In Chapter 2, I described the ongoing arguments in the anthropological literature about the effect (or lack of effect) of matrilineal descent on women's power and status. This interaction provides some support for the hypothesis that matrilineal descent is associated with greater access to power for women, at least indirectly, since it prevents women from being adversely affected by the degree of stratification characterizing their culture. Other interactive effects are not as obvious. Controlling for stratification reveals that stratification has a general tendency to weaken the effect of descent on the dependent variables. The interactive effects between stratification and residence are difficult to discern. A s I explained in Chapters 2 and 6, these two independent variables are linked in their construction to some extent; matrilocal residence is found only in the egalitarian and intermediate cultures of this thesis, and avunculocal residence, in the ranked cultures. Hence, these two variables are correlated from the outset. In general, when it is controlled for, each of these independent variables weakens the relationship between the other independent variable and the dependent vari-able. Specifically, when stratification is held constant, the relationships between residence and the dependent variables become weaker. Similarly, when residence is held constant, the relationships between stratification and the de-pendent variables become weaker. The results indicate some tendency for residence to interact with the relationships between descent and the dependent variables. In cultures with 180 matrilocal residence, the relationships between descent and the dependent varia-bles tend to be weak, while in those cultures with other residence, the relationships are strong and significant. However, this interactive effect is not as strong as the interactive effect of descent on the relationships between stratification and the dependent variables. Descent does not have a consistent effect on the relationships between residence and the dependent variables. 2. Univariate and Multivariate Associations In this section, I report and discuss the associations between the depend-ent and independent variables. The relationships observed between the depend-ent and independent variables are presented in Tables V and V I . Table V presents the values of the measure of association, T a u B or C , and the significance level for this association. Table V I provides a summary of the relationships. First , the relationships support the hypotheses that women's access to various types of power is associated with degree of stratification, type of descent, and final postmarital residence pattern, although which independent variable is the important one depends on the dependent variable. Women's access to autonomy in various areas of their lives, on the other hand, is rarely associated with the independent variables ( W . F A T E . A F T . M A R is the only exception). Second, the results demonstrate that degree of stratification is the strongest and most significant factor in women's access to power. When stratification is associated with a dependent variable, the relationship is generally stronger than the associations observed between descent and the 181 Table V: Kendall's Tau Values for the Relationships Between the Dependent and Independent Variables and Significance Levels of These Relationships Dependent Stratification Descent Residence Variable Tau B Tau C Signif. Tau B Tau C Signif. Tau B Tau C Signif. LEADR -.436 .011 .009 .481 .062 .370 COMMUN.INFL .354 .016 -.240 .141 -.045 .393 DOMEST.DEC. .242 .113 -.181 .146 -.235 .119 SUBSIST.LEADR -.250 .081 -.378 .072 -.250 .081 TRADE/REC.DEC. .163 .242 .389 .080 .163 .249 INH.INDIV.PROP .141 .062 -.357 .024 -.125 .070 MAN.KIN.CHATT -.703 .000 -.009 .481 -.008 .481 MAN.KIN.PROP -.749 .000 .164 .177 .154 .173 MAN.KIN.INTANG -.055 .183 .208 .116 .055 .173 OWN.OWN.FRTS .148 .138 -.520 .002 -.266 .018 NON-DOMEST.FRTS + WEALTH -.208 .048 .000 .500 .000 .500 CO-OP.FRTS .628 .001 -.102 .296 -.244 .103 MARR.ARRANG -.111 .174 -.276 .047 -.277 .006 ED. CHILD -.107 .118 -.023 .448 -.055 .254 PREMAR.SEX .057 .208 -.196 .150 .057 .208 OWN.MAR.DEC .144 .108 .073 .335 -.157 .069 EXMAR.SEX .018 .429 -.234 .104 -.098 .143 DIVORCE .087 .080 .196 .130 .104 .035 REMAR .054 .206 -.201 .135 -.054 .206 W.OTHR.ROLES -.083 .163 -.196 .168 -.064 .214 M.OTHR.ROLES -.160 ? -.333 -.200 ? W.FATE.AFT.MAR .332 .032 -.167 .239 -.299 .048 182 Table V I : S u m m a r y of the Strength of E a c h Relationship and Its Statistical Significance Dependent Variable Stratification Descent ? 7 + + ? ? + Residence L E A D R C O M M U N . I N F L D O M E S T . D E C . S U B S I S T . L E A D R T R A D E / R E C . D E C . I N H . I N D I V . P R O P M A N . K I N . C H A T T M A N . K I N . P R O P M A N . K I N . I N T A N G O W N . O W N . F R T S N O N - D O M E S T F R T S + W E A L T H C O - O P . F R T S M A R R . A R R A N G E D . C H I L D P R E M A R . S E X O W N . M A R . D E C E X M A R . S E X D I V O R C E R E M A R W . O T H R . R O L E S M . O T H R . R O L E S W . F A T E . A F T . M A R . . . * + + * + ? + + + * ? 7 ? ? ? ? ? ? ? + + * ? ? ? ? 7* ? + + + + + + A statistically significant relationship A strong positive relationship A moderately strong positive relationship A modest tendency for a positive relationship A weak or doubtful relationship A modest tendency for a negative relationship A moderately strong negative relationship A strong negative relationship dependent variables or residence and the dependent variables. Moreover, degree of stratification is the independent variable most often significantly related to women's access. Descent and residence are of secondary importance; they tend 183 to be only moderately or modestly related to the dependent variables with which they are associated. T h i r d , residence has an interesting pattern of association with women's access to power. It is significantly associated with dependent variables only when stratification or descent is also associated with that same dependent var-iable (see Table V) . Stratification and descent, however, are never both sig-nificantly associated with the same dependent variables. The discussion of the associations will take the following form. I begin with a discussion of those dependent variables associated with stratification. I then consider those variables associated with descent, and then those associated with residence. Next, I briefly consider those dependent variables which have weak or insignificant associations with the dependent variables. The chapter concludes with a discussion of the findings. For each of the significant relationships, I describe the direction and strength of the association and report any interactions or spurious correlations. The results are analyzed and interpreted throughout. a. RELATIONSHIPS WITH STRATIFICATION . Stratification has its strongest relationships with the dependent variables dealing with public leadership, community influence, economic power and autonomy in one area: L E A D R , C O M M U N . I N F L , M A N . K I N . C H A T T , M A N . K I N . P R O P , C O - O P . F R T S , and W . F A T E . A F T . M A R . 1 1 Degree of stratification is significantly associated with N O N - D O M E S T . F R T S . & W E A L T H , a dependent variable which deals with the right to expect produce and wealth from people who are not spouses. This significant relationship is based on a small number of cases, however; for this reason, it is omitted from further analysis. 184 b. LEADR The dependent variable, L E A D R , deals with access to defined roles of leadership in the kin group or community. The thirty observations on this dependent variable are distributed between the two categories, less access and equal access. Clearly, women hold formal roles of leadership, contrary to some claims made in the anthropological literature. The contingency tables for L E A D R indicate a strong and significant nega-tive association with stratification; as the degree of stratification increases, women's access to roles of leadership decreases vis-a-vis men's (Tau C = -.436, statistically significant at a level of .011). Descent and residence are only weakly associated with L E A D R . Thus , the effect of stratification is dominant. The association between stratification and L E A D R is modified by an interaction with descent. In cultures with matrilineal descent, the effect of increased stratification on women's access to leadership, described above, is non-existent (Tau C = 0.0). In cultures with other descent, however, the relationship between increased stratification and less access for women is stronger than in the uncontrolled relationship (Tau C is now -.817, statistically significant at .001). 1 It seems that matrilineal descent counteracts the negative effect of increased stratification on women's access to leadership, while in cultures with other descent (i.e., non-matrilineal descent), the negative effect of stratification is unimpeded. These findings support the claims of authors who argue that greater stratification is associated with decreased status and power 1 In the controlled tables, the measures of association are accurate, but the smaller sample size will tend to decrease the significance levels (there is less in-formation available for ascertaining a predictable pattern from chance). Thus , obtaining greater significance with a smaller sample, as we observe here, is difficult. 185 for women and indirectly, the claims of those who argue that matrilineal descent is associated with women's increased status and power. The results for this dependent variable, however, do not support two of the contentions made by authors reviewed in Chapter 2. The first is the claim by Friedl (1975:70-74) and Nielsen (1978:27) that, even in matrilineal societies, men have the main political functions and public authority. The dis-tribution for L E A D R indicates this is not so. Women do have equal access to leadership in some cultures, particularly in egalitarian ones (regardless of the descent system). 1 The second claim not supported by the results is Friedl's (1975:62-64, 69) statement that in patrilineal, patrilocal, non-stratified societies (she is actually referring to horticultural societies), women have no power, while in patrilineal, stratified societies, women have political power and control. M y results reveal the opposite; women in the other (patrilineal and bilateral), non-stratified cultures in this thesis are more likely to have equal access to leadership than are women in patrilineal, stratified cultures. Either Friedl's hypothesis is incorrect or stratification in hunting and gathering societies does not work in the same manner as predicted for horticultural societies. It is surprising that women and men have equal access to leadership roles in egalitarian cultures with other descent. One would assume that where women move into a community after marriage, they would be less likely to be accepted as leaders in the community. This does not seem to be the case, however. It m a y be that their access to leadership roles is due to their husbands (i.e., women who are leaders are wives of leaders); unfortunately, this type of infor-mation was not provided in the ethnographic material I examined. 186 c. MAN.KIN.CHATT and MAN.KIN.PROP These dependent variables consider access to economic power as manager or trustee of kin group owned chattels and immovable economic property. For M A N . K I N . C H A T T , thirty-two observations are distributed between the two categories, less and equal access. Women can and do control important economic property, although the distribution indicates men are more likely to do so. The relationship between stratification and M A N . K I N . C H A T T is nega-tive, very strong (Tau C = -.703), and statistically significant (significance = .0000). Thus , as stratification increases, women's access to the control of kin group owned chattels decreases vis-a-vis men's. The relationships with descent and residence are almost nonexistent. Thirty-three observations on M A N . K I N . P R O P are distributed across the categories less access and equal access. The relationship between M A N . K I N . P R O P and the independent variables is almost identical to the relationship with M A N . K I N . C H A T T . The relationship between stratification and M A N . K I N . P R O P is negative, very strong (Tau C = -.749), and statistically signif-icant (significance =.0000). A s stratification increases, women's access to the control of kin group owned immovable economic property decreases vis-a-vis men's. The relationships between M A N . K I N . P R O P and descent and residence are weak and statistically insignificant. The results clearly support the hypotheses that degree of stratification has a negative effect on women's access to the control of kin group owned chattels and immovable property. A s degree of stratification increases, women's access to this type of economic power decreases. The equality observed in non-ranked societies is due primarily to the lack of such economic 187 power in these societies, so that neither men nor women have economic power not available to the other gender. M y results agree with Whyte's findings that there is a tendency for women in more complex societies to have somewhat less control over property (1987:137, 140). 1 The interaction previously observed between descent and the relationship between stratification and women's access to clearly defined roles of leadership is perhaps weakly observed with these dependent variables. In societies with matrilineal descent, the relationship between M A N . K I N . C H A T T and stratification is somewhat weaker but it remains significant (Tau C = -.663, significant at a level of .01). In societies with other descent the relationship is slightly stronger than the uncontrolled one (Tau C = -.731, significant at a level of .0005). Similarly, for M A N . K I N . P R O P , T a u C = -.710, statistically significant at .007 in matrilineal societies, and T a u C = -.720, significant at a level of .0006 in other societies. These interactions disclose that stratification is by far the most important influence on women's access to control of kin group owned chattels and immovable property. d. COMMUN.INFL The relationships discussed to this point indicate that women in ranked societies have less access to leadership roles in the community and to kin group owned economic property, as predicted by many of the authors 1 There is a difference, however, between Whyte's dependent variable and mine. Whyte's "property" scale includes economic control of all property. I have a num-ber of dependent variables which consider different types of property and ownership of this property (e.g., kin group owned chattels and immovable property, kin group owned intangible property, privately owned economic property, privately owned ceremonial property); only some of these are significantly related to stratification. 188 discussed in Chapter 2. Three of the relationships between dependent varia-bles and stratification do not take this predicted direction. The first is the relationship between C O M M U N . I N F L and stratification. I use the dependent variable, C O M M U N . I N F L , to measure whether women have influence and power in the community or kin group which is not associated with a formal leadership role (i.e., women's opinions and wishes affect the behavior and decisions made by others, and they have a voice in the outcome of events and situations under debate). Twenty-one observations on this dependent variable are distributed between the categories less and equal access. The relationships between C O M M U N . I N F L and descent and C O M M U N . I N F L and residence are negative, weak, and insignificant. C O M M U N . I N F L has a moderately strong (Tau C = .354), positive relationship with stratification - the opposite of the relationship between L E A D R and stratification. T h a t is, as stratification increases, women are more likely to have equal access with men to influence in community or kin group decisions. The observed relationship is statistically significant (significance =.016) . 1 There are at least four possible explanations for the unexpected direction of this relationship. First , a third factor m a y be involved - the fur trade. Some authors contend the fur trade adversely affected the position of women in the Mackenzie Subarctic. Since the majority of egalitarian cultures in m y sample are located in this area, it would explain -why the access of women in the egalitarian cultures of m y sample is lower than the access of women in ranked and intermediate cultures. Perry (1979), for example, has suggested that the "low status" of Northern Athapaskan women resulted from change 1 Once descent is held constant, the relationship loses significance, but these interactions are based on a small number of cases. 189 brought by the fur trade in this area. Goods had to be transported long distances and women were seen as a form of cheap transportation. H e argues that, in the ethnographic literature on Northern Athapaskan women, the contradictory evidence about the status of women actually represents two different time periods - before development of the fur trade, when women had high status, and after development of the fur trade, when the literature attests that these women had low status. Similarly, Leacock (1978) maintains that reports of low status for women in Northern Athapaskan societies do not represent aboriginal conditions. She alleges that these reports of low status are due to changes brought about by the fur trade. While this explanation has much to recommend it, one problem remains. If the fur trade is the cause of the unexpected direction of the relationship between stratification and C O M M U N . I N F L , why did the fur trade not adversely affect women's access to leadership in egalitarian societies (i.e., L E A D R ) ? A second possible explanation is that the unexpected direction of the relationship may be the outcome of errors in the data. These errors could arise in a number of ways. Errors could arise, for example, from the ethnographer's interpretations of the information, from the informant's errors, or from errors in m y coding of the data for this study. This explanation does not seem likely, however, as one must then explain why the error is present only for egalitarian societies and not for intermediate or ranked ones and why the error is present only for this dependent variable and riot for L E A D R . A third possibility is suggested by Whyte (1978:127-28) who argues that the greater informal influence of women in complex societies (as defined by Whyte) is a compensation or reaction to increasing restrictions placed on women's formal authority (1978:140). A s formal roles of leadership are increasingly restricted to men in the intermediate and ranked societies of m y sample (see L E A D R ) , women gain equality with men in their ability to influence the kin group or community. This explanation is intriguing. If it is correct, the hypothesis is supported, both cross-culturally (Whyte 1978) and within the more limited geographic area examined in this thesis. Fourth , one's rank may be more important than one's sex in determining who has influence in the kin group or community for the ranked and intermediate societies in this thesis. Thus , as the degree of stratification increases, so does women's influence. That is, where women can inherit high status, both high status men and women have the ability to influence decisions affecting the kin group or community. Although this explanation, suggested by both Friedl (1975:62-64) and Schlegel (1977a:20), was not supported by results observed for L E A D R , it seems to be supported in this case. O f course, one must then ask why inherited rank should be associated with women's influence in community and kin group affairs, but not with their ability to exercise clearly defined leadership roles. C a n it be that high status women do not have less access to formal roles of leadership but, rather, that ethnographers define formal roles of leadership in such a way that women's participation is made invisible (see Chapter 1)? This would ex-plain why Friedl's argument is not supported for L E A D R and is supported for C O M M U N . I N F L . 191 e. CO-OP.FRTS This variable considers control of economic resources in the domestic sphere. It asks which spouse has control over the use, allocation, and distri-bution or redistribution of fruits of labour (particular^ food) they jointly produce. The twenty-Five observations are distributed across the three primary categories, less access, equal access, and greater access. The relationships be-tween C O - O P . F R T S and descent and C O - O P . F R T S and residence are weak and insignificant. The relationship with stratiFication is strong (Tau B = .628, statistically signiFicant at a level of .0005) and positive. A s stratiFication increases, women's access to control over this type of economic wealth increases. A positive relationship between C O - O P . F R T S and stratiFication is unexpected; numerous authors have argued that increased stratiFication will have a negative impact on women's power and status. Possible reasons for this unexpected result will be considered momentarily. Holding the independent variables constant, one by one, provides additional information about this type of power. The pattern of interaction al-ready observed between stratiFication and descent is repeated. Matril ineal descent tends to dampen the relationship between stratification and C O - O P . F R T S , although the relationship remains signiFicant (Tau C = .496, statistically signiFicant at .049). In societies with other descent, the relationship between increased stratiFication and women's greater access to C O - O P . F R T S is stronger than before (Tau C = .714, statistically significant at a level of .0041). Similarly, residence has this same interactive effect on the relationship between C O - O P . F R T S and stratification (Tau B = .333, the Exact 192 Probability value is 1.0001 in matrilocal societies; Tau B = .738, statistically significant at a level of .0008 in societies with other residence). Stratification thus has a positive effect, especially in those societies with other descent or other residence. Matrilocal residence and, to a lesser degree, matrilineal descent, work to prevent women from having relatively lower access than men (i.e., only equal access, rather than greater access) in societies with intermediate or egalitarian degrees of stratification. Why increased stratification should be associated with increased access for women is an interesting question. The results for MAN.KIN.CHATT, MAN.KIN.PROP, and CO-OP.FRTS suggest that in ranked societies there is a distinction as to the type of property controlled by each gender. Men tend to control and have rights over kin group owned economic and ceremonial property in this study; women tend to have control over the wealth and goods produced jointly by themselves and their spouses.2 In egalitarian societies (and, perhaps, intermediate societies) these distinctions do not occur. In general, men and women have equal access to control of kin group owned property in egalitarian and intermediate societies (see MAN.KIN.CHATT, MAN.KIN.PROP, and CO-OP.FRTS). The positive association between stratification and women's access to the economic resources produced jointly by As pointed out in Chapter 5, the Exact Probability is a test of significance; it is the exact probability of obtaining the result observed in the contingency table, plus the probabilities of getting even more unusual outcomes in the same direc-tion as those observed in the table (Blalock 1960:221-222). Because it is not di-rectly associated with the Kendall's Tau value, however, the Exact Probability is consistently larger (i.e., less significant) than the significance value for Tau. 2 One cannot assume the wealth controlled by women is less important than the wealth controlled by men; a number of authors have stated that this domestic wealth is important (e.g., Klein 1975:100, Oberg 1973:66, Olson 1967:13, Portlock 1786:290). Indeed, some have admitted this wealth is essential to the establishment and maintenance of kin group status (e.g., Klein 1975:100-101, Suttles 1951:495). 193 spouses in ranked societies seems to be the result of differential property con-trol by women and men in these societies. In other words, although women in societies with increased stratification seem to lose control over property and resources which are owned communally by the kin group, they still maintain control over domestic wealth, wealth which is essential for the creation and maintenance of social status. f. W.FATE. AFT.MAR This dependent variable is concerned with women's autonomy over their own bodies and fate after marriage in the areas of spouse exchange and wrestling for wives, an activity described primarily for the Northern Athapaskan speaking peoples of my sample. The nineteen observations are distributed between the two categories, present and absent. When women have this type of autonomy, the dependent variable is coded present for that cultural unit. W h e n women do not have this autonomy, the variable is coded absent. The relationship between stratification and W . F A T E . A F T . M A R is unexpectedly positive and moderately strong (Tau C = .332, significant at .032); as stratification increases, women's access to this type of autonomy increases. The relationship with residence is negative and modest (Tau C = -.299, statistically significant at .048); this relationship will be discussed later in the chapter. The relationship with descent is weak and insignificant. Holding the independent variables constant, one by one, results in an interaction with descent. A m o n g societies with matrilineal descent, the relationship with stratification is weaker than the uncontrolled one (Tau 194 C = .160; E . P . = .400), while among societies with other descent, the relationship is much stronger than the uncontrolled one (Tau C = .494, E . P . = .1667). 1 Thus, matrilineal descent seems to work against a decrease in women's autonomy in egalitarian societies; indeed, descent may be partly responsible for the relationship observed between W . F A T E . A F T . M A R and stratification (a significance level for the T a u value is needed to determine this). A s was the case with C O M M U N . I N F L and C O - O P . F R T S , the direction of the relationship between stratification and the dependent variable is unexpected - greater stratification is expected to be associated with decreased autonomy for women. W h y should stratification be associated with greater autonomy for women in this area? There are at least three possible explanations. First , it may be due to a third factor, the fur trade. In the ethnographic literature, authors, writing about the same culture, sometimes disagree about the existence of the institutions of "wife exchange" and "wife wrestling" (e.g., Osgood 1936:143, and Honigmann 1946:86 versus Jenness 1977:391, Mason 1946:32 versus Wentzel 1889:86, Jenness 1977:393 and H e l m 1972:74 versus Richardson 1852:256 and Kei th 1890:114-115). A s noted earlier, Perry has argued these contradictions in the data represent change over time, brought about by the nature of the fur trade in the Subarctic. Thus , the fur trade m a y be the reason that women in the egalitarian societies of m y sample tend to be at a disadvantage compared with men. A second possible explanation lies with errors in the data. For example, authors disagree about the extent of the woman's involvement in the decisions being made about wife exchange or wrestling for wives (e.g., Honigmann 1 Exact Probability values are consistently less significant than the significance values obtained for T a u B or C . 195 1954:71, Nielson 1978:42, and Osgood 1932:78). It is difficult to determine the extent, if any, of women's input into these decisions. In addition, I suspect that some of the ethnographic descriptions of wrestling for wives are borrowed from Hearne's description of such an event among the Chipewyan , 1 and his description has simply been projected onto another cultural unit without further research as to its appropriateness for that partic-ular cultural unit (e.g., Birket-Smith 1930:67-68, Mason 1946:32, Richardson 1852:256). Moreover, some of the descriptions of such institutions are not adequately referenced by the authors (e.g., in Jenness 1956) and are, therefore, uncertain. I suspect that selective error cannot be ruled out; further research is necessary. in order to determine this. Third , the increased autonomy may be the result of a combination of increased stratification and its effect on unilineal descent systems. In the ranked societies of the Northwest Coast, women and men inherit status at birth (or they inherit the possibility of attaining high status). Marriages are arranged between men and women of equal status and represent ties between families, not just the individuals involved. E v e n after marriage, a man's or woman's ties with the kin group remain strong. Marr ied people retain their claims on family economic resources (see Jones 1914:25, and de L a g u n a 1983:79) and continue to be involved in family affairs (much to the chagrin of certain missionaries - see Duncan C2155 November 21, 1862, pages 10290 and 10977 and M a y 31, 1866, page 10895). Indeed, Olson (1967:70) states that clan loyalty outweighed nuclear family and household loyalty in a feud. In return, the family took a personal interest in the success or failure of a 1 In Chapter 3, I expressed doubts about how representative Mattanobee, Hearne's guide, is of aboriginal Chipewyan culture. 196 family member's marriage (e.g., Olson 1967:23). The family had a responsibility to protect and revenge married family members, even from their spouses (e.g., Curtis 1970b:121, Drucker 1950:279, Swanton 1905b:408), and could demand payment if that family member were harmed by a spouse (Murdock 1934b:372, Olson 1967:22-23). This strong family interest in married family members (both economic and personal) is associated with greater stratification in the geographic area of my thesis and may be partly responsible for the freedom of women from spouse exchange and wrestling for wives - at least in those societies characterized by other descent. The autonomy of women in matrilineal societies does not seem to be affected by the degree of stratification in the society (see the interactive effect of descent described above). g. Relationships With Stratification - Conclusions The findings to this point indicate that the degree of stratification in a cultural unit does have an effect on women's access to power, particularly in those societies with other (non-matrilineal) descent, and, in some cases, other residence. The results indicate that women in ranked societies tend to have less access than men to formal roles of leadership and to the management of kin group owned chattel property and immovable economic property. Thus , the development of social stratification in a society has a detrimental effect on women's ability to manipulate power in certain spheres, particularly the political and non-domestic economic spheres. O n the other hand, women in these ranked societies tend to have equal access with men to influencing decisions which affect the community or kin group. Although women have 197 less access to leadership roles (as defined by ethnographers), they have greater community influence. Perhaps their ability to inherit high r a n k 1 provides high status women with the same access as high status men to an active role in influencing important community or kin group affairs. Increased stratification seems to have other effects which increase the power of women in certain spheres. Women in ranked societies tend to have greater access than men to control over the fruits of labour produced jointly by spouses. 2 A s previously explained, this greater access to the control of domestic fruits of production may be due to a differentiation in the type of property controlled by men and women, a differentiation arising with the de-velopment of greater stratification. Although high status men are inclined to have greater control of kin group owned resources and wealth, women control the domestically produced fruits of labour and wealth goods. Thus, both men and women have access to the economic property essential to validate and maintain one's own and one's family's social status. 3 Final ly , the results reveal that women in ranked societies generally have more autonomy in the area of spouse exchange and wrestling for wives than women in intermediate or egalitarian societies. This increased access may be 1 The results for the dependent variables which measure women's access to the role of ceremonial host and women's access to the control of kin group owned intangibles, indicate that women did have high rank. Furthermore, the ethnographic data indicate women had high rank (e.g., Boas 1916:496 and Duff 1959:38-40). For example, Garfield reports that among the Tsimshian, it was equally important to ensure the status of children of both sexes (1939:196). Kle in notes that a Tlingit woman's rank was as important as a man's (1975:102). 2 This is not true for women in societies with matrilocal residence, which is associated with egalitarian societies in this thesis. 3 Women retained influence in their own families. Sapir and Swadesh, for instance, comment that young Nootka boys were warned to be "manly" (i.e., hard working and industrious in an attempt to gain and maintain prestige), or their brothers and sisters would not aid them (1939:189). 198 an outcome of the fur trade and its effect on women in egalitarian societies. It may also be a product of increased stratification and the resulting greater hold families retain over their own members, and hence, their continued interest in the fate of family members after marriage. h. RELATIONSHIPS WITH DESCENT Two dependent variables display a strong and significant association with descent: O W N . O W N . F R T S , and M A R . A R R A N G . 1 These dependent variables con-sider women's access to power in the economic and domestic spheres. The associations are not as strong as those between stratification and the depend-ent variables. i. OWN.OWN.FRTS Descent is significantly associated with a dependent variable that measures the ability of wives and husbands to control the use and disposal of the fruits of their own labour within the domestic unit consisting of spouses. Sanday (1973) argues that the ability to control these resources is an important measure of status. Thirty-two observations on this dependent variable are located in the categories equal access and greater access. Stratification has a weak and statistically insignificant relationship with O W N . O W N . F R T S . A relationship is observed between O W N . O W N . F R T S and residence; this relationship is discussed later. Descent has a strong, negative association with this dependent variable (Tau B = -.520, statistically significant 1 A third dependent variable, dealing with the right to inherit individually owned economic property, also is significantly related to descent. Because it is based on a small number of cases, however, it is not included in the analysis. 199 at a level of .0019). 1 In societies with matrilineal descent, wives are more likely to have access to control over the fruits of their own labour while men are not. Indeed, wives are more likely to control the fruits of labour of their husbands. This is an unexpected result if one accepts the common assumption that women do not have economic power. In societies with other descent, wives and husbands generally control the fruits of their own labour. The lack of an association between stratification and O W N . O W N . F R T S is surprising, since hypotheses advanced in the literature suggest that women's greater access or control of their own fruits of labour can be expected to decrease in ranked societies. 2 Controlling for the presence of interactive effects reveals that in ranked societies with matrilineal descent, women have greater access than men, while in ranked societies with other descent, women and men have equal access. When the effect of stratification is held constant, the relationship between descent and O W N . O W N . F R T S is stronger than ever (Tau B = -.588, significant at .009). Thus , descent and not stratification is the important factor here. Residence modifies the relationship between descent and O W N . O W N . F R T S . This relationship is strong. In societies with other descent and other residence, women are more likely to have equal access with men while in societies with matrilineal descent and other residence, women have greater access (Tau B = -.548, statistically significant at .007). In societies 1 Whyte's (1978:133) results are similar to mine, although his dependent varia-bles take a different form. He found that women in matrilineal societies and or/matrilocal societies are more likely to control the fruits of men's labour and of women's labour, in addition to their control of other types of property. 2 Sacks (1979), for example, argues that control of family property is an important determinant of women's domestic authority in societies without clearly demarcated social classes (Whyte 1978:147). 200 characterized by matrilocal residence, regardless of the type of descent, the relationship between descent and O W N . O W N . F R T S is no longer significant (Tau B = -.333; no significance level is available, although the Exact Probability value is .750). Thus , matrilocal residence dampens the effect of descent. j. MAR.ARRANG The second variable to have a significant association with descent is M A R . A R R A N G . This dependent variable determines who arranges the first marriages of younger kin. Although the ability to make such arrangements is treated as part of the domestic sphere, these arrangements can have economic and political ramifications as well. For example, McDonnell (1975:251, 288, 339) notes that among the Kas in i , control over marriage arrangements gives the arranger access to the hunting and gathering areas of the group his or her child is marry ing into, an important advantage to the marriage arranger. Seguin (1984:xiii) states that among the Tsimshian, a chief and all holders of ranked names discuss important topics such as alliances, succession, defense, resource use, and marriages. Thus , the ability to arrange first marriages can give men and women access to additional economic and/or public power. The thirty-eight observations on this dependent variable are distributed across the two categories equal and greater access. The relationship with stratification is negative, weak, and statistically insignificant. The relationship with descent, on the other hand, is negative, of moderate strength (Tau B = -.276) and statistically significant at a level of .047. T h a t is, women in societies with matrilineal descent are more likely to have greater access to this type of domestic power than men, while women in those societies 201 characterized by other descent are more likely to have only equal access with men. Residence is also associated with this dependent variable. It will be discussed later in this chapter. Interactions modify the relationship between descent and M A R . A R R A N G . First , when stratification is held constant, the relationship between descent and M A R . A R R A N G is insignificant in all three categories of stratification. In other words, stratification is at least partly responsible for the relationship observed betweeen descent and M A R . A R R A N G . Residence also modifies the relationship between descent and M A R . A R R A N G . In societies with matrilocal residence, the negative relationship between descent and M A R . A R R A N G disappears altogether. Matrilocal residence dampens the negative effect of other descent. In societies with other residence, the relationship is stronger and more significant (Tau B = -.466, sig-nificant at .013). That is, in societies characterized by other residence, other descent is associated with equal access for women and men while matrilineal descent is associated with greater access. k. Relationships With Descent - Conclusions F r o m the relationships observed between descent and the two dependent variables, a number of insights have been gained. First , matrilineal descent is associated with greater access for women to the control of the fruits of their own labour (as well as the fruits of labour produced by their spouses) and to the arranging of the marriages of younger kin members. Other descent is associated with a decrease in women's access to these various aspects of power. 202 The presence of the relationships between descent and the dependent var-iables indicate that Friedl (1975:14-15) and Nielsen (1978:24-29) are incorrect in their claims that descent (as well as residence) has little or no effect on women's power in hunting and gathering societies (see Chapter 2). 1 Certainly, descent affects women's ability to control the fruits of their own labour in both the egalitarian and ranked hunting and gathering societies examined in this thesis and the ability to arrange the first marriages of younger k i n . 2 I. RELATIONSHIPS WITH RESIDENCE Like stratification and descent, residence is associated with women's access to certain aspects of power. A s with descent, however, these associations are not as strong or as numerous as the associations observed between stratification and the dependent variables. Residence is associated with three dependent variables; one is concerned with economic power, one with domestic power, and one with autonomy. The variables are O W N . O W N . F R T S , M A R . A R R A N G , and W . F A T E . A F T . M A R . In fairness, it must be recalled that these authors are comparing hunting and gathering societies with horticultural societies. They may be treating the subsistence base as an index of societal complexity and the cultures of the Northwest Coast do not fit the usual pattern of societal complexity in hunting and gathering societies. 2 However, the degree of stratification has some effect on the relationship be-tween descent and women's access to arranging the first marriages of younger kin; the relationship with descent is no longer significant when stratification is controlled for. 203 m. OWN.OWN.FRTS Like descent, residence is significantly associated with O W N . O W N . F R T S . The relationship with residence is negative, of modest strength (Tau C = -.266), and statistically significant at a level of .018; this association is due almost entirely to the positive effect of avunculocal residence on women's control over the fruits of their own labour. Women in societies with avunculocal residence are more likely to have greater access than men while women in societies with other residence are more likely to have equal access with men. Women in societies characterized by matrilocal residence are between these two posi-tions. It seems that matrilineal descent, matrilocal residence, and especially, avunculocal residence, are positively associated with women's greater access to these economic resources. The results agree with Whyte's findings that matrilineal descent and matrilocal residence give women more control over property than do patrilineal descent or patrilocal residence. T h a t is, the hypothesis is supported cross-culturally, as well as within a limited geographical area. The association between avunculocal residence and O W N . O W N . F R T S is unexpected. Authors such as Mart in and Voorhies (1975:228-29) and Nielsen (1978:24-29), have argued that avunculocal residence will have the same nega-tive effect on women's power and status as does patrilocal residence (i.e., the majority of my "other residence" category) in more stratified societies. A s ex-plained in Chapter 2, these authors contend that in societies with avunculocal residence or patrilocal residence, women become workers for their spouses and are not part of the core group that produces and distributes together. Thus , there is little o p p o r t u n e for women to gain power or prestige. M y results 204 indicate the opposite; women in societies with avunculocal residence are more likely to have greater control over their own fruits of labour, providing them with an opportunity to redistribute and gain power and prestige. Women in societies with matrilocal residence are more likely to have equal access with men. Avunculocal residence may have this effect because it is linked with both matrilineal descent and the ability of women to inherit rank in societies with some degree of stratification. Where descent is matrilineal and rank is hereditary for men and women in societies with some degree of stratification, and, as a consequence, where there are more things for men and women to inherit and control, certain aspects of women's power may be enhanced. 1 The partial associations support the result that avunculocal residence is correlated with greater access for women because it is tied to greater stratification. When the effect of stratification is held constant, the effect of residence is no longer significantly associated with O W N . O W N . F R T S (for the category ranked, T a u B = -.381, no longer significant at .064; for egalitarian, T a u B = -.509; no significance level is provided, but E . P . = .300). Thus, degree of stratification seems to be a strong factor in the relationship observed be-tween O W N . O W N . F R T S and avunculocal residence. The statistics for this dependent variable sustain the hypotheses that matrilineal descent, and, to some extent, matrilocal residence are associated 1 Friedl's contention that where women have access to inherited positions of social status, they will have greater opportunity for power (1975:62-64) is partially upheld by this result. Friedl argues that the ability of women to inherit rank is most strongly associated with their increased status in those societies with patrilineal descent. M y results indicate that women have greatest access in those societies characterized by matrilineal descent and the inheritance of rank, however. Thus , matrilineal descent and inherited rank are positively related to women's access in this instance. 205 with greater economic power for women. 1 This is especially true in ranked societies. The results do not support the hypothesis that avunculocal residence is correlated with less access for women in the same manner as patrilocal residence (i.e., other residence). n. MAR.ARRANG Residence, like descent, is significantly associated with the dependent vari-able M A R . A R R A N G . The thirty-eight observations are distributed between the categories equal access and greater access. The relationship with residence is negative, of moderate strength (Tau C = -.277), and statistically significant (at a level of .006). In cultures with matrilocal residence, women are more likely to have greater access than men, while in cultures with other residence, they are more likely to have equal access with men. For this dependent variable, avunculocal residence has the same negative effect as other residence. Two interactive effects modify this relationship. First , unlike the relationship between descent and M A R . A R R A N G , the relationship between residence and M A R . A R R A N G is not made insignificant by holding stratification constant; the relationship remains strong for all three categories of stratification. F o r two of the three categories, intermediate and egalitarian, the strength of the relationship between residence and M A R . A R R A N G remains significant. Thus , other residence remains correlated with less access for women and matrilocal residence, with more access. In ranked societies, however, residence no longer has a significant effect on women's access. The 1 A s mentioned earlier, some of Whyte's (1978:133) conclusions are similar to mine. H e finds women in matrilineal societies and/or those societies with matrilocal residence are more likely to have control over property; this property includes the fruits of men's labour and women's labour. results demonstrate, contrary to the claims of Friedl and Nielsen that in hunting and gathering societies (i.e., egalitarian societies) descent and residence have little effect on women's power and status, that residence is associated with women's access. The second interactive effect is revealed by controlling for descent. When descent is held constant, residence is only weakly related to MAR.ARRANG in cultures with matrilineal descent (Tau C = -.138, statistically insignificant at a level of .271). In cultures with other descent, however, residence is a perfect predictor of women's access; Tau B = -1.00, statistically significant at a level of .000. Thus, matrilineal descent dampens the negative effect of avunculocal residence and other residence on women's access to power vis-a-vis men's. Again, these results uphold the claims of authors who argue that matrilineal descent and/or matrilocal residence are positively associated with women's power and status. o. W.FATE. AFT.MAR The last significant relationship with residence occurs with the dependent variable W.FATE.AFT.MAR. Thus, both stratification and residence are signifi-cantly related to this dependent variable. The relationship between residence and W.FATE.AFT.MAR is negative and moderately strong (Tau C = -.299, statistically significant at .048). Women in cultures with matrilocal residence and avunculocal residence have autonomy not shared by women in cultures with other residence. This result sustains the hypothesis that matrilocal residence is linked to women's access to greater power. In addition, as with OWN.OWN.FRTS, avunculocal residence is positively associated with equal 207 access for women. The direction of this association with avunculocal residence is not startling when one recalls the relationship between greater stratification (avunculocal residence is found only in ranked cultures in m y sample; see Chapter 6) and women's increased access to autonomy for this dependent vari-able. Descent modifies this relationship between residence and W . F A T E . A F T . M A R . In matrilineal cultures, the negative effect of other residence on women's autonomy is stronger than in the uncontrolled table (Tau C = -.320; there is no significance level provided). In societies with other descent, the relationship is weaker (Tau B=- .189 , there is no significance level provided). Thus , in matrilineal cultures, other residence is negatively associated with women's access. In cultures with other descent, other residence is no longer significantly associated with less access for women. This result is likely due to the confounding effect of stratification - other residence is found across all the categories of stratification (unlike matrilocal residence and avunculocal residence). Hence, when the effect of stratification is not controlled for, the pattern of associations between stratification, residence, and W . F A T E . A F T . M A R are not kept distinct. The result is the diminishing of the link between other residence and W . F A T E . A F T . M A R . p. Relationships With Residence - Conclusions The statistical anafysis of the relationships between residence and the de-pendent variables has provided several insights. The relationships between residence and the dependent variables indicate that other residence has a neg-ative effect on women's access to 1) control over the fruits of their own 208 labour, 2) arranging the first marriages of younger kin in intermediate and egalitarian cultures or in cultures with other descent, and 3) autonomy over their fate after marriage, particularly in those cultures with matrilineal descent. Matrilocal residence has a positive effect. A s with descent, then, residence does affect women's access to power and autonomy, even in hunting and gathering cultures, contrar}' to the expectations of authors such as Friedl (1975) and Nielsen (1978). The results also provide some insight into the manner in which avunculocal residence is related to women's access to power and autonomy vis-a-vis men's; avunculocal residence is not consistently tied to decreased access for women, as some authors claim, Rather, the effect of avunculocal residence varies. In one case, both avunculocal residence and other residence were negatively correlated with women's access. In another case, avunculocal residence was positively associated with women's access - much the same as matrilocal residence. In the third case, avunculocal residence was positively associated with women's access, matrilocal residence with somewhat less access, and other residence with even less access. Obviously, the relationships are not simple ones. I shall return to this topic at the end of the chapter. Thus far, all the multivariate relationships revealed by holding a second independent variable constant, are in addition to a primary relationship be-tween the dependent and independent variables. The statistical analysis reveals that for one of the dependent variables, S U B S I S T . L E A D R , controlling for each independent variable in turn produces significant relationships not observed when the independent variables are not held constant. 1 T h a t is, holding one 1 One other dependent varible displa3's this same pattern, i.e., no significant primary relationships occur, but when certain independent variables are held 209 independent variable constant uncovers a significant relationship between the de-pendent variable and a second independent variable. q. SUBSIST.LEADR Neither stratification, descent, nor residence is significantly associated with S U B S I S T . L E A D R , which considers access to leadership roles in subsistence activities (the associations between this dependent variable and the three independent variables are close to significance, however). Nevertheless, the search for interactive effects (i.e., partial associations) reveals an interesting pattern. The results indicate: 1) in cultures with other descent or other residence, women's access decreases as the degree of stratification increases (Tau C = -.750, no significance level is provided for this association, but the Exact Probability value is .036); 2) in ranked cultures with oiher descent, women, in comparison to men, have less access to this type of leadership than women in ranked cultures with matrilineal descent (Tau B = -1.00, E . P . = .036); 3) women in ranked cultures with avunculocal residence are more likely to have equal access with men than are women in ranked cultures with other residence (Tau B = -1.00, E . P . = .036). Thus , there appears to. be a set of conditions which must be present before women will have less access than men to leadership roles in subsistence activities, specifically, a ranked culture with other descent or other residence. These interactions between the dependent and independent variables lend credence to the hypotheses that greater stratification, other descent and/or other residence have a negative continued . . . constant, a significant relationship is observed. Because the relationship with E D . C H I L D (authority over the education and discipline of children) is based on a small number of cases, however, it is excluded from the analysis. 210 effect on women's access to leadership in subsistence activities. Significant relationships were not found between the independent variables and a number of dependent variables. These include the variables which deal with domestic decision-making ( D O M E S T . D E C ) , access to decision-making in the area of trade, barter, and reciprocity ( T R A D E / R E C . D E C ) , and extramarital sexual freedom ( E X M A R . S E X ) . In Chapter 6 I discussed some of the factors which may have led to the equal access between men and women observed for a number of dependent variables. Two of these factors m a y help explain why some of the relationships do not show strong or significant associations. First , the definition of m y dependent variable in terms of access may have lowered the degree of variation in m y measures of women's power and autonomy. This point has been discussed in various places in this study. The discussion will not be repeated here, except to note that the access measure may mute some of the distinctions which could be made about differences in access. Second, the lack of significant or strong relationships between some of the dependent variables and the three independent variables may be unique to my sample. A s explained in Chapter 6, the cultures included in this analysis consist solely of hunting and gathering societies. Although some of these societies do have greater societal complexity than many hunting and gathering societies, the degree of societal complexity may not be as important for women's access to power as would the greater societal complexity described by Gough (1975) and Whyte (1978). 1 A s explained in Chapter 6, only a 1 Gough maintains that with the growth of class societies, women's subordination will increase greatly (1975:75), whereas in hunting and gathering societies and societies based on horticulture, women are well-off. Whyte suggests that the shift from a subsistence pattern based on hunting and gathering to one based on 211 cross-cultural study or research on contiguous cultures characterized by greater societal complexity than obtains on the Northwest Coast can lend support to this hypothesis. B. CONCLUSION This study provides abundant new information concerning the manner in which power is distributed between men and women and how stratification, descent, and residence affect these distributions. The results support certain hypotheses commonly found in the literature concerning the manner in which stratification, descent, and residence do or do not affect women's power and status. The results do not support other hypotheses. In both these respects, this study represents an important contribution to the anthropological literature on women, which seldom makes any direct attempt at rigorous hypothesis testing. The major results will be reiterated here. Firs t , the findings indicate stratification, and to a lesser degree, descent and residence, have an effect on women's access vis-a-vis men's to power and autonomy in various aspects of their lives. Stratification is the most pervasive factor affecting women's access. In some cases, increased stratification is associated with decreased access for women; in other cases, with increased access for women. The positive associations between increased stratification and greater access for women are not predicted by most authors who contend that as the degree of stratification increases, women's power and status will decrease. 1 continued .. . horticulture is less important for women than the further shift to plow agriculture (1978:139). 1 The unexpected direction of these positive relationships lends support to the statement by Schlegel (1977a:5), discussed in Chapter 2, that the relationship 212 The results show that increased stratification is associated with less access for women to clearly defined roles of social leadership, particularly in those societies with other (non-matrilineal) descent. Increased stratification is associated with less access for women to the management of kin group owned chattels and non-movable economic property. A t the same time, increased stratification is associated with women's equal access to community influence. It has been postulated by Whyte (1978) that this positive association may be tied to women's lack of access to roles of formal leadership; as women's access to formal roles of leadership decreases, their access to influence within the community (i.e., more informal leadership) increases. To some extent, this relationship also lends supports to Friedl's argument that in societies with inherited positions, women have political power and domestic control not availa-ble to women in societies where rank is achieved rather than inherited (1975:62-64). Increased stratification is associated with women's greater control over the fruits of labour produced jointly by themselves and their husbands. I have postulated that the development of some stratification in a society results in differentiation in the economic property controlled by each gender. F r o m this analysis, it is clear high ranking men come to control the economic resources and certain economic and ceremonial goods owned collectively by the kin group. The control of such property gives these men great power and the right to call on the goods and services of others. Women in these societies continued . . . between degree of stratification and women's status is not a simple linear relationship. Schlegel notes that in egalitarian hunting and gathering societies, there is sexual equality but that in those societies in the middle ranges of social stratification (i.e., fishing), the relationship can "go either way". Thus , the unexpected direction of some of the relationships between stratification and women's access to various aspects of power imply Schlegel is correct. 213 retain control over domestic goods, which are also an important source of power since they allow women to influence others and to obtain prestige through redistributive activities. Further research is necessary to ascertain whether an even greater degree of stratification than obtains in this study continues to take increasingly more political power out of the hands of women and decreases women's control over kin group owned economic resources, economic property, and ceremonial property. Likewise, further research is needed to ascertain whether women in these societies retain control over the domestic fruits of labour produced by themselves and their spouses or whether even greater stratification is associated with women's increasing alienation from the control of important economic resources and other property. In this latter case, for women, the result would be a loss of valuable channels for wielding power and influencing the behavior and decisions of others. Final ly , increased stratification is associated with women's greater control over their own fate after marriage, in terms of spouse exchange and wrestling for wives. This autonomy for women in ranked societies may be brought about by increased family interest in absent members who move away after marriage and in their children, who represent valuable sources of political and economic power. Although absent, married women retain kin group membership and their status (and fate) is of importance for the whole kin group. 1 O n the other hand, this relationship may be due to the form the fur trade took in the egalitarian hunting and gathering societies of the 1 This explanation also lends support to Friedl's conjecture that where status and power are inherited rather than attained, women have an important advantage not available to women in other societies (1975:62-64). 214 Mackenzie Subarctic, as Leacock (1978) and Perry (1979) postulate. Besides these important associations between stratification and women's access to various aspects of power and autonomy, the degree of stratification also has indirect or partial effects on women's access. Stratification has an indirect effect on women's access to arranging the marriages of younger kin, since, when the effect of stratification is controlled for, the significant association between descent and M A R . A R R A N G becomes insignificant in all three categories of stratification. T h a t is, although stratification does not display a direct association with M A R . A R R A N G , it is at least partly, if not wholly responsible for the association observed between M A R . A R R A N G and descent. Stratification also has an indirect effect on women's access to leadership in subsistence activities; in ranked societies, other descent and other residence are associated with less access for women to this role, while in egalitarian and intermediate societies, other descent and other residence are not associated with less access for women. In this instance, the interaction does not sustain Friedl's claim that in those societies where women inherit their rank, they will have access to power not available to women in societies where rank or status must be achieved. Descent and residence are also associated with differential access to power and autonomy for men and women. The results indicate that other descent is associated with less access for women to two aspects of power in the domestic sphere. In societies with other descent, men and women have equal access to control over the fruits of their own labour, whereas, in societies with matrilineal descent, women have greater control over these fruits of labour. Thus , women in societies characterized by matrilineal descent have 215 domestic power not available to women in societies characterized by other descent. In addition, in societies with other descent, men and women have equal access to the arranging of first marriages of younger kin, while matrilineal descent is associated with greater access for women. Consequently, matrilineal descent is associated with greater access to certain aspects of domestic power for women, as Friedl (1975), Gough (1975), Mart in and Voorhies (1975), Schlegel (1972, 1977a), and Whyte (1978) have argued. O n the other hand, descent has less of an effect on women's access to power than is claimed in the anthropological literature (e.g., Engels 1972, Morgan 1974). O n the basis of his empirical results, Whyte (1978:133-134, 171) draws this same conclusion. His results reveal that matrilineal descent and matrilocal residence are associated with women's control over property, while other descent and other residence are not. H e also finds that matrilineal descent and matrilocal residence are weakly associated with somewhat more domestic authority, more equal sexual restrictions, and more ritualized female solidarity. With the exception of control over property, however, the benefits associated with matrilineal descent and matrilocal residence "are not very powerful" (1978:132-34). He concludes, Matri l ineal descent and matrilocal residence are associated with modest benefits for women in certain areas (particularly in property rights), and the existence of private property is associated with some forms of male bias . . . However, it should be noted that neither of these factors has the sweeping consequences for women that some of the earlier literature supposed. M y results add one important qualification to this finding. Matril ineal descent does have an indirect effect on some of the associations between stratification and certain dependent variables and on the associations between 216 residence and certain dependent variables. In societies with matrilineal descent, degree of stratification is not associated with less access to power for women or, in some cases, the effect of stratification is weaker and statistically less significant than in societies with other descent. In other words, matrilineal descent tends to dampen the effect of increased stratification. The results demonstrate that residence has some effect on women's access to power. Other residence is associated with women's less access to certain aspects of power than obtains in societies with matrilocal residence. These aspects of power include: 1) arranging the first marriages of younger kin, 2) control over the fruits of their own labour, and 3) autonomy in choices about spouse exchange and wrestling for wives. Thus , residence, like descent, has a moderate effect on a few dependent variables. These relationships support the hypotheses predicting that other residence is associated with less access to power for women, and matrilocal residence with greater access. Like descent, however, residence does not have the "sweeping consequences for women that some of the earlier literature supposed" (Whyte 1978:171). The manner in which avunculocal residence is related to women's access deserves attention. A number of authors have argued, for various reasons, that avunculocal residence will have a negative effect on women's power and status. Mart in and Voorhies (1975:228-29) and Nielsen (1978:27-32), for ex-ample, hold that avunculocal residence arises as a response to a shift to male control of property in societies with matrilineal descent, and, therefore, avunculocal residence shares a number of the negative effects for women's status associated with patrilocal residence. Whyte agrees that avunculocal 217 residence has a negative effect on women's power for a different reason; with avunculocal residence, women no longer reside with their female kin (1978:34-35). My results do not consistently support the claims of these authors that avunculocal residence, much like other residence, will be associated with less access for women. Where avunculocal residence is concerned, it may be, as Schlegel maintains (1977a:28-29), there is no simple equation between matrilocal residence and women's power versus avunculocal residence or patrilocal residence and men's power. This study shows that avunculocal residence may have either a positive or negative effect on women's access to power and autonomy. Avunculocal residence seems to behave differently from other residence (i.e., contrarj' to the hypothesis) in those cases where greater stratification is also positively associated with women's access. This result suggests that as long as stratification is based on inherited rank, and both men and women inherit rank, and where there is matrilineal descent (as is the case for the matrilineal, avunculocal societies of the Northwest Coast), women have access to certain aspects of power vis-a-vis men. Where rank is not inherited or, perhaps, where women do not inherit rank equally with men, women have less access to these aspects of power. I would like to add a caveat to this discussion of the statistical results. While the results have supported many hypotheses appearing in the literature and have not supported others, this does not prove that certain hypotheses are true while others are false. Statistics can only render some hypotheses more plausible than others. They demonstrate that certain hypotheses (and theories) are more or less useful in prediction and explanation; 218 Theories in all branches of science are supported to varying degrees by accumulated evidence . . . numerical analyses are intended to show some plausibility - that is, some empirically demonstrable relationships - that fit with the broader theoretical constructs. The statistics present us with patterns - relationships that make some of the other, competing theoretical propositions less plausible and less useful in predicting outcomes and events. (Pelto and Pelto 1978:280). To conclude, in this thesis, I have substituted a multidimensional perspective on access to power in place of the more common unidimensional perspective (for another example, see Whyte). I have shown how each dimension or aspect of power is associated with the three independent varia-bles, degree of stratification, type of descent, and final postmarital residence pattern. What has emerged clearly from my analysis is the pre-eminent im-portance of degree of stratification, tempered by descent, for women's access to power and autonomy. What also has emerged is the fact that the different dimensions of women's power can be simultaneously influenced by more than one independent variable . 1 It would be possible to extend this analysis in several ways. First , while I have used a large range of indicators of power and autonomy, the analysis could be repeated using other dependent variables. In addition, the analysis could be extended to other independent variables. Final ly , it would be well to test these relationships using other continuous area samples. It 1 Analyses which take into account such simultaneity are rare in both the theoretical and empirical anthropological literature. There are exceptions, however. In the theoretical literature, Friedl (1975:62-64) and Gough (1961, 1975) consider the manner in which stratification and descent jointly combine to affect women's status. Schlegel (1977a:5) asserts that although stratification is associated with women's status, it is not a simple linear relationship. In the empirical literature, Jorgensen (1966), Driver and Schuessler (1967), and Chaney and Ruiz-Revilla (1969) have attempted to analyze more than one pair of variables at a time (see Jorgensen 1974b: 199 on this point). might also be useful to construct a larger sample of cultures in order address the question of independent variable interactions more fully. VIII. CONCLUSIONS I began this study by pointing out that there is a lack of consensus in the anthropological literature on women. I indicated that if some consensus is to be achieved, two of the issues which must be addressed are: 1) the defi-nition and measurement of women's status or position, so that comparisons can be made among studies, and 2) the lack of systematic rigorous testing of the hypotheses advanced to explain various aspects of women's status and sexual inequality. In this thesis, I examine the associations between the differential access of men and women to various aspects of power and three independent varia-bles, degree of stratification, type of descent, and final postmarital residence pattern. The findings have provided new insights regarding the manner in which access to power is distributed between men and women across a number of cultural units, and the manner in which access is affected by the three independent variables. To begin, contrary to the expectations of some scholars who argue that women are "universally subordinate" to men (this subordination is generally in reference to two specific spheres, political leadership and economic power), m y results indicate that there is variation in women's access to power vis-a-vis men's across cultures. 1 The findings demonstrate that women's access to power varies from culture to culture. The results do not support claims such as: "In all societies where public authority exists, it is in the hands of men" 1 The results indicate, for instance, that women can have access to leadership roles and to economic power. In addition, they indicate the conditions under which women's access to these aspects of power will decrease. 220 221 (Nielsen 1978:27), or that, even in matrilineal societies, men hold the public authority (Martin and Voorhies (1975:225), or the assertion that women in band societies did not have the right to allocate scarce resources (Cohen 1978:258). Women can and do have public authority and the right to allo-cate important economic property. This means that in order to gain some understanding of the processes giving rise to, and perpetuating, sexual inequality, scholars must look to factors which vary across cultures (e.g., degree of stratification, type of descent, women's control over food or their ability to trade, etc.), rather than universal explanations (e.g., biology), as some scholars have attempted to do. The results of this analysis indicate that women's access to power (i.e., their access to roles and rights) vis-a-vis men's varies, not only across cultures, but within the same culture. This variation in women's access to specific roles, rights, and autonomy within the same culture implies that women's status is not a unidimensional phenomenon (see Chapter l ) . 1 Moreover, the finding that the dependent variables are not related to the same independent variables (some dependent variables are related to stratification, others to descent, and others to residence) may also be an indication that women's access is not a unitary phenomenon. 2 This finding has important implications for those interested in the anthropological literature on women. First , it means that scholars may obtain a very skewed portrait of the overall position of women in a society if they utilize only one or two 1 In his study of women's status, Whyte e m p i r i c a l ^ tests this hypothesis (1978:95-120). His results indicate that women's status cannot be measured along a single continuum. 2 Presumably, if all the dependent variables were equivalent indicators, each should be related to the independent variables in the same manner. 222 measures of women's power or status and assume that these measures capture "the status" of women. The effects of assuming unidimensionality are even worse from a cross-cultural perspective. In this case, not only is the scholar assuming that a narrow range of indicators adequately captures women's power or status, but in addition, he or she is assuming that the same indicators are equally relevant indicators of women's status in all cultures. Given these implications, it is no wonder that the anthropological literature on women abounds with contradictory claims. In addition to the insights provided regarding the manner in which men's and women's access to power varies within and across cultures, m y re-sults provide information regarding the manner in which differences in their access to power is associated with the three independent variables tested in this stud}r. To summarize briefly, according to the results, the access of men and women is affected by the three independent variables, degree of stratification, type of descent, and final postmarital residence. Thus , the statistical anyalysis provides support for hypotheses which maintain that these independent variables are associated with women's status or women's power. Conversely, the results do not support hypotheses which maintain that these independent variables have no effect. The findings also provide new information about the manner in which the three independent variables affect women's access to power vis-a-vis men's. First , in the cultures examined in this study, the degree of stratification is the most pervasive factor associated with women's access to power. Contrary to expectations, however, the direction of the relationships between women's access to power and degree of stratification is not consistent. In some cases, 223 increased stratification is linked with less access for women, in other cases, with equal or greater access for women. Whether these disadvantages and advantages are particular to the ranked societies of this studj' or whether any increase in the degree of stratification is associated with women's decreased or increased access to these aspects of power, can only be deter-mined by testing other cultures characterized by even greater stratification. According to the results, descent is associated with women's access to power in certain spheres. Matril ineal descent and matrilocal residence have a positive effect on women's access to power in the cultures in this study. Thus , those hypotheses which maintain that the type of descent and final postmarital residence have little or no effect in hunting and gathering societies, are not supported by m y study. 1 In terms of their effect on women's access to power vis-a-vis men's, descent and residence are of secondary importance in comparison to the degree of stratification, however. 2 The last finding regarding the relationships among the variables I shall review concerns the associations with avunculocal residence. Avunculocal residence is not consistently associated with women's decreased access to power, as some scholars have maintained. In some instances, avunculocal residence is not correlated with less access for women. Rather, it is associated with women's greater access to power vis-a-vis men's. There are at least two possible explanations for this unexpected finding. The first is 1 Aga in , it should be reiterated that the scholars (i.e., Friedl 1975 and Nielsen 1978) seem to use the hunting and gathering subsistence base as an indicator of low societal complexity, since they argue that there is no accumulation of surplus or wealth to make inheritance and, hence, descent, of any importance in these societies. 2 This parallels Whyte's finding that "this pattern of associations with aspects of complexity accounts for the large majority of the findings." 224 the unique character of avunculocal residence found in certain societies on the Northwest Coast. This is the preferred cross cousin marriage pattern and intermarriage between two lineages which characterize certain matrilineal ranked societies (e.g., the Massett Haida , the Skidegate Haida , the Northern Tlingit and the Southern Tlingit). This pattern was described in Chapter 2. If Whyte's explanation of why matrilocal residence has a positive effect on women's status is correct (i.e., where women live in close proximity to their female kin, they have greater status, 1978:34-35), then one could expect to find that avunculocal residence in this study will be associated with greater access to certain aspects of power for women in these ranked matrilineal societies. Given this explanation, however, why avunculocal residence is not consistently associated with equal or greater access for women for all the de-pendent variables, remains a mystery. A second possible explanation lies with Friedl's assertion that where women inherit rank equally with men (in patrilineal societies), women have greater economic and political control. Friedl's association between greater stratification, patrilineal descent, and women's power is only partly supported by the results of this analysis; matrilineal descent and not patrilineal descent seems to be associated with greater stratification and women's greater access to power. The concept that women's access to power is related to women and men having equal access to the inheritance of rank is an intriguing one. Certainly, it seems that in those societies where both men and women can inherit positions of high rank - the same societies which tend to have greater ownership and control of resources, wealth, and produce - and which are characterized by matrilineal descent and avunculocal residence (i.e., the ranked, 225 matrilineal, avunculocal societies of the Northwest Coast), women have greater access to power in certain spheres. It remains to be seen whether this relationship between the inheritance of rank, matrilineal descent, avunculocal residence, and women's access to certain types of power breaks down at some point, perhaps when women no longer inherit rank equally with men. Further research would be necessary to determine how women's access to power is affected by such changes. In addition to the information gained from the associations between the dependent and independent variables, the examination of partial associations (controlling for the effect of one variable on the dependent variable while holding a second independent variable constant) provided a number of key discoveries. First , it is clear that few of the relationships are simple linear relationships. This finding supports Schlegel's arguments that the data do not support a simple or direct correlation between sexual equality or inequality and degree of stratification (1977a:5), or a simple equation between matrilocality and female domestic power and patrilocality or avunculocality and male power (1977a:28-29). Certainly, the variety of interactions observed among the variables support these claims. Nevertheless, although there are a large number of partial associations (interactions) observed, there are few consistent patterns. The most obvious and recurring pattern of partial associations is observed for matrilineal descent, which has an indirect effect on the relationships between women's access to power vis-a-vis men's and the degree of stratification. Matril ineal descent has a dampening effect on the negative association between increased stratification and a decrease in women's access to certain aspects of power in comparison to men. This is a notable 226 discovery and lends empirical support to Gough's proposition that, even in hunting and gathering societies and horticultural societies where women tend to have greater independence and power, matrilineal descent (and matrilocal residence) are associated with greater independence for women (Gough 1975:74-75). The only other consistent result involves matrilocal residence. The results show that matrilocal residence has a dampening effect on the relationships between the dependent variables and descent. Thus , other descent does not decrease women's access in those societies characterized by matrilocal residence. Although this partial association is consistent, it does not recur as often as the one between matrilineal descent and the relationships between the dependent variables and degree of stratification. A s observed in Chapter 7, empirical studies rarely examine the relationships between more than two variables at a time. Thus , the results for the partial associations among the variables provide new information about the manner in which the three independent variables and dependent variables interact. A s well as making some hypotheses more plausible than others, the findings of this study have other implications for the study of women in the anthropological literature. The results show that there exists sufficient variation in women's and men's access to power within hunting and gathering societies to produce the kinds of associations which are ascribed to horticultural societies. T h a t is, although a number of the scholars reviewed in Chapter 2 assert that degree of stratification, descent, and residence have an effect on women's status in horticultural societies but little or no effect in hunting and gathering societies, my results show that their models can be 227 extended to the hunting and gathering societies included in this analysis. Final ly , the relationships observed in this continuous area study replicate some of the major results of Whyte's (1978) cross-cultural study. Most of the empirical studies of women's status or power have utilized cross-cultural samples, generally based on Murdock's (1961) World Ethnographic Sample. These include the studies by M a r t i n and Voorhies (1975), Sanday (1973, 1974, 1981), Schlegel (1972), and Whyte (1978). O f these, Whyte's study is the most directly comparable to this study. I did not set out to replicate the findings of Whyte's study - the variables differ, as do the statistical techniques used. Y e t the results regarding certain relationships with degree of stratification, type of descent, and final postmarital residence are similar. Also noteworthy is the similarity of the finding that degree of stratification (in this study) or societal complexity, which includes degree of stratification (Whyte's study), is the most important of the independent variables affecting women's access to power (or status, in Whyte's study). Pelto and Pelto (1978:267) observe that "The logic of cross-cultural research requires that any valid relationship between variables should hold up under statistical analysis within culture areas as well as worldwide." They conclude that if the correlations between cultural elements or items are found to hold for certain areas but not for others, "The purported associations m a y be accidental or historical rather than arising from 'true' causal factors." In so far as the variables in Whyte's study and this study are comparable (and a number of them are), the similarity of some of our results lends support to both studies. That is, the validity of the relationships, which are observed both cross-culturally and within the continuous area used in this study, is enhanced. 228 Although a number of insights have been gained about the differential access of men and women to power, and the manner in which the degree of stratification, type of descent, and final postmarital residence pattern affect access in this study, there is room for further research in this area. In Chapter 7 I suggested the possibility of repeating the analysis using other de-pendent variables or extending the analysis to other independent variables. I pointed out it would be useful to test the relationships using other continuous geographical areas. It might be of particular interest to use a continuous area that includes both hunting and gathering and horticultural societies. The question of the reliability of the ethnographic data on women and their roles in other societies also would be a profitable area for research. Two research ideas, both utilizing statistical techniques, come to mind. The first would be to replicate the present study with one major alteration - to incorporate all of the variation found in the sources for each cultural unit. If this were possible, it would not be necessary to choose only one source from among all the sources describing one culture as the "correct" one. Such a study would, of course, involve a greater degree of statistical finesse than the present study. A second possibility is a study to test the reliability of the ethnographic data on women. One culture would be selected, a culture which has been described by a large number of authors. A n extended list of culture traits could serve as the dependent variables. E a c h source would be used in turn to code the information on the culture traits. Statistical analysis could then be used to determine whether there in only one culture being described by all the sources, or whether there are many cultures (i.e., each author 229 describes a culture which is statistically different from the descriptions provided by the other authors). In addition, such an analysis might provide informa-tion about which aspects of culture are more likely and less likely to be reliable. B}r way of conclusion, I would like to reiterate the value of the statistical technique to anthropology. There is a wealth of ethnographic data available for the researcher who is interested in utilizing them. Careful con-sideration of the data, their strengths and their limitations, can allow a researcher to obtain "new views from old data". This study illustrates the value of using a statistical model to lend sup-port to certain hypotheses commonly found in the literature, and to gain new information about the manner in which these variables affect women's access to power vis-a-vis men's. Large scale comparative studies can provide the support needed to allow us to generalize about and understand the manner in which phenomena interact, as well as to begin to understand the processes which give rise to or perpetuate these phenomena. By allowing us to find meaningful patterns in the data, empirical studies can allow us to test, to support, and to reject, various models which attempt to explain cultural phenomena such as differences in the access of men and women to power. IX. APPENDIX I: T H E E F F E C T OF CULTURE A R E A In Chapter 5 I discussed Galton's Problem and its implications for empirical studies dealing with relationships among cultural phenomena. To reiterate, Galton's Problem is concerned with the effect of cultural diffusion be-tween cultures used in an empirical study. Galton's Problem asks, are the cultures independent of each other or have they been affected by diffusion because of a common history or borrowing? If the cultures under investigation are interdependent, then one must take this interdependence into account when examining covariation between cultural variables across cultures. A s explained in Chapter 5, a continuous area study is one approach for dealing with Galton's Problem. Driver, the major advocate of this approach (1970, Driver and Chaney 1974), asserts that this type of study can be used to control for the influence of diffusion, while testing for relationships among variables, because the researcher can evaluate, rather than ignore the factor of diffusion. In a study originally published in 1966, Driver explicitly tests for the influence of diffusion on kin avoidance in a continuous area of North American Indian cultures. H e incorporates "geographical-historical" factors in addition to "psycho-functional" factors such as descent, residence, and kinship terminology (i.e., the hypothesized causes of kin avoidance) to explain the presence or absence of kin avoidances. The geographical-historical factors are measured by two variables, culture area groupings and language family mem-bership. Driver's results indicate that these latter variables are more powerful in predicting the presence or absence of kin avoidances than are his psycho-functional variables (1974c:40). 230 231 Since Driver's results illustrate the importance that geographical and historical factors can have in a continuous area study, I attempt to explicitly test for this association with the access to power of men and women. I utilize culture area as the geographical-historical variable 1 and degree of stratification as the psycho-functional variable (since stratification is the most important of these hypothesized factors). The culture area variable has two categories, Northwest Coast and Interior. The culture areas are those defined by Driver (1969). 2 The analysis consists of three steps. First , I examine the relationships between the de-pendent variables and culture area. The strength of the associations (measured by Kendall's T a u B or C) and the statistical significance of the strenth of the associations are compared with those values obtained for the relationships between the dependent variables and degree of stratification. Next, I examine the relationships between the dependent variables and culture area, while controlling for the effect of stratification. If the culture area var-iable is more powerful than the degree of stratification for predicting the access of women compared to men, then these relationships should remain strong. Final ly , I examine the relationships between the dependent variables and degree of stratification when the effect of culture area is controlled for. If culture area is the more powerful predictor of women's access, the relationships between the dependent variables and degree of stratification should 1 The Athapaskan languages are not spoken on the Northwest Coast and Salishan languages and language isolates are not spoken in the Subarctic culture area. Therefore, the variable language family membership is not tested. 2 According to Driver's classification, there are only two cultural units from m y study located in the Plateau culture area (the Northern and Western Shuswap and the Eastern and Southern Shuswap). A maximum of two cases for one category is impractical, thus, these two cultural units were combined with the Subarctic into one category, Interior. 232 become weaker. Unfortunately, the situation in this study is not ideal for such a test. First , only two culture areas are represented. Second, one of these, the Northwest Coast, is characterized as having ranked societies. Thus , the psycho-functional variable being tested (degree of stratification) is already encompassed within the definition of the Northwest Coast culture area. Since the degree of stratification is known to be associated with the dependent vari-ables in this study, one might expect that the results of this methodological test will reflect these interrelationships. The findings bear this out. The association of the Northwest Coast culture area with ranked societies does not allow enough variation within the Northwest Coast culture area to test the effect of stratification when the ef-fect of culture area is held constant. There is only one category of stratification observed in the Northwest Coast culture area - ranked. In the Interior category, however, one observes ranked, intermediate, and egalitarian societies. Similarly, the interrelationship between the Northwest Coast culture area and ranked societies does not allow enough variation within the categories intermediate and egalitarian, when the effect of degree of stratification is held constant. Egalitarian and intermediate societies are observed only in the Interior category. Ranked societies, however, are observed in both culture areas. This means that statistics are generated for only some of the relationships; any results are, therefore, difficult to interpret. E v e n with the inadequate variation, the results of the analysis do pro-vide some interesting findings. First , it seems culture area is a more powerful predictor than degree of stratification for four of the dependent 233 variables. These include LEADR - access to clearly defined leadership roles; TRADE/REC.DEC - access to organizing and decision-making in the areas of trade, barter, and reciprocity; OWN.MAR.ARRANG - ability to influence the arranging of one's own first marriage; and W.FATE.AFT.MAR - autonomy in the areas of wife wrestling and spouse exchange. This last relationship bears out my supposition, presented in Chapter 7, that the relationship between W.FATE.AFT.MAR and degree of stratification is a phenomenon of either the Northwest Coast or Subarctic culture areas. Contrary to the expectation that culture area is the most powerful predictor of women's access in a continuous area study, degree of stratification seems to be the more powerful of the two for at least three relationships. These include: COMMUN.INFL - access to influence in the community or kin group; MAN.KIN.PROP - access to kin group owned nonmovable economic property; and CO-OP.FRTS - access to the domestic fruits of labour produced co-operatively by spouses. In all cases, the relationships are the same as those reported in Chapter 7. It is impossible to tell which of the two independent variables, degree of stratification or culture area, is the more powerful predictor of women's access to power vis-a-vis men's in the remaining relationships. From the minimal evidence available, it seems that stratification does have some effect on certain aspects of women's access to power, over and above any effect produced by culture area. In conclusion, I would add that including geographical-historical variables with the psycho-functional variables being tested in a continuous area study is a sound methodological approach. In this analysis, however, the results are 234 confounded by the small number of culture areas represented and particularly, by the association between one of these culture areas and the degree of stratification characterizing the societies in the culture area. This methodological study must be done with more culture areas if adequate results are to be obtained. X. BIBLIOGRAPHY Abercrombie, W . R . 1900 Alaska 1899 - Copper River Exploring Expedition. Washington: Government Printing Office. Aberle, Dav id F . 1961 Matril ineal Descent in Cross-Cultural Perspective. In Matri l ineal Kinship. David M . Schneider and Kathleen Gough, eds. Pp. 665-727. Berkeley: University of California Press. 1972 Outline of Materials for Research on Southwestern Ecology and Economy. Vancouver, University of Brit ish Columbia, Teaching Materials . 1980 Navajo Exogamic Rules and Preferred Marriages . In The Versatil ity of Kinship. L inda S. Cordell and Stephen Beckerman, eds. Pp. 105-143. New York: Academic Press. 1981 A Century of Navajo Kinship Change. Canadian Journal of Anthropology 2:21-36. Albright, Sylv ia L . 1982 Women's Roles in Tahl tan Society. Paper presented at the A n n u a l Meeting of the Canadian Ethnology Society, Vancouver, Brit ish Columbia, M a y 9, 1982. Al lard , Rev. E . 1929 Notes on the K a s k a and Upper L i a r d Indians. Primitive M a n 2:24-26. Al lardt , E r i k 1968 Theories about Social Stratification. In Social Stratification. J . A . Jackson, ed. Pp. 14-24. Cambridge: Cambridge University Press. 235 236 Al len, Henry T . 1887 Report of an Expedition to the Copper, Tanana , and Koyukuk Rivers, in the Territor}' of Alaska , in the Y e a r 1885. Washington: Government Printing Office. Alsop, Richard 1967[1815?] A Narrat ive of the Adventures and Suffering of John R. Jewitt: Only Survivor of the Crew of the Ship Boston, Dur ing a Captivity of Nearly Three Years A m o n g the Savages of Nootka Sound. Reprint. Fairfield, Washington: Y e Galleon Press. Asch , Michael J . 1975 The Impact of Changing F u r Trade Practices on the Economy of the Slavey Indians. In Proceedings of the Second Congress of the Canadian Ethnology Society 2:646-657. National M u s e u m of M a n , Mercurj ' Series, Canadian Ethnolog}' Service 28. 1981 Slavey. In Handbook of North American Indians. Vol .6 The Subarctic. June He lm, ed. Pp. 338-349. Wil l iam Sturtevant, Gen. ed. Washington: Smithsonian Institution. A z m a n , Yae l 1981 Sex, Power, and Authority. The Brit ish Journal of Sociology 32:547-559. Barbeau, Mar ius 1929 Totem Poles of the Gitksan, Upper Skeena River, Brit ish Columbia. National Museum of Canada, Bulletin 61, Anthropological Series 12. Barnett, Homer G . 1938 The Coast Salish of Canada. American Anthropologist 40:118-141. 1939 Culture Element Distributions: Gul f of Georgia Salish. University of California Anthropological Records 1(5):221-295. 1955 The Coast Salish of British Columbia. University of Oregon Monographs, Studies in Anthropology 4. 237 1968 The Nature and Function of the Potlatch. Department of Anthropology. Eugene: University of Oregon. 1978[1938] The Nature of the Potlatch. In Indians of the North Pacific Coast: Studies in Selected Topics. T o m McFeat , ed. Pp. 80-91. Reprint. Toronto: The Macmil lan Co. of Canada Limited. Barrett, Michele 1980 Women's Oppression Today: Problems in Marxis t Feminist Analysis . London: Verso Editions and N L B . Bell , James M . 1903 The Fireside Stories of the Chippewyans. The Journal of Amer ican Folk-Lore 16:73-84. Bender, Donald R. 1967 A Refinement of the Concept of Household: Families, Co-Residence, and Domestic Functions. American Anthropologist 69:493-504. Benedict, Ruth 1936 Mar i ta l Property Rights in Bilateral Society. American Anthropologist 38:368-373. Berndt, Catherine H . 1981 Interpretations and Facts in Aboriginal Austral ia . In Woman the Gatherer. Frances Dahlberg, ed. Pp. 153-204. New Haven: Yale University Press. Bernick, K a t h r y n 1975 The Socio-Economic Role of Carrier Women. Kumtucks Review (Spring) 1975:77-86. 238 Birket-Smith, Kaj 1930 Contributions to Chipewyan Ethnology. Report of the Fifth Thule Expedition 1921-1924, Vol.6(3). Copenhagen: Gyldendalske Boghandel, Nordisk F o r - L a g . Bishop, Charles A . 1983 Limit ing Access to Limited Goods: The Origins of Stratification in Interior British Columbia. In The Development of Political Organization in Native North Amer ica . 1979 Proceedings of the American Ethnological Society. Elisabeth Tooker, ed. Pp. 148-161. Washington, D . C . : American Ethnological Society. Black, Samuel 1955 A Journal of a Voyage F r o m Rocky Mountain Portage in Peace River to the Sources of Finlay's Branch and Northwestward in Summer 1824. The Publications of the Hudson's B a y Record Society: Black's Rocky Mountain Journal 1824. E . E . Rich, ed. London: The Hudson's B a y Record Society. Blackman, Margaret B . 1973 Totems to Tombstones: Culture Change as Viewed Through the Ha ida Mortuary Complex, 1877-1971. Ethnology 12:47-56. 1982 During M y Time: Florence Edenshaw Davidson, A Ha ida Woman. Seattle and Vancouver: University of Washington Press and Douglas & Mclntyre , L t d . Blalock, Hubert M . 1960 Social Statistics. New York: M c G r a w - H i l l Book Co. 1979 Social Statistics. Revised 2nd E d . New York: M c G r a w - H i l l Book Co. Boas, F r a n z 1890a The Nootka. Sixth Report on the North-Western Tribes of Canada. The Brit ish Association for the Advancement of Science 60:582-604, 668-679. 1890b The Shuswap. Sixth Report on the North-Western Tribes of Canada. The Brit ish Association for the Advancement of Science 60:632-647. 239 1892 The Bilqula. Third Report on the Indians of British Columbia. The British Association for the Advancement of Science 61:408. 1895 The Nass River Indians, The Nisk • A'. Fifth Report on the Indians of British Columbia. The British Association for the Advancement of Science 65:569-592. 1897 The Religious Ceremonials of Other Tribes of the North Pacific Coast. In The Social Organization and the Secret Societies of the Kwakiutl Indians. Report of the U.S. National Museum 2:632-664. 1912 Tsimshian Texts (New Series). Publications of the American Ethnological Society 3:65-284. 1916 Tsimshian Mythology. Thirty-First Annual Report of the Bureau of American Ethnology. Pp. 29-1037. Bock, Philip K. 1967 Love Magic, Menstrual Taboos, and the Facts of Geography. American Anthropologist 69:213-217. Bompas, William C. 1888 Diocese of Mackenzie River. London: Society for Promoting Christian Knowledge. Bone, Robert M., Earl N. Shannon, and Stewart Raby. 1973 The Chipewyan of the Stony Rapids Region. Mawdsley Memoir I. Saskatoon, Saskatchewan: Institute for Northern Studies, University of Saskatchewan. Brathwaite, Jean, and W.J. Folan 1972 The Taking of the Ship Boston: An Ethnohistorical Study of Nootkan-European Conflict. Syesis 5:259-266. Briggs, Jean L. 1974 Eskimo Women: Makers of Men. In Many Sisters: Women in Cross-Cultural Perspective. Carolyn J. Matthiasson, ed. Pp. 261-304. New York: The Free Press. 240 Brooks, Alfred Hulse 1953 Blazing Alaska's Trai ls . B . L . Fryxe l l , ed. University of A laska and the Arctic Institute of North America . Bujra , Janet 1975 Women and Fieldwork. In Women Cross-Cultrually: Change and Challenge. Ruby Rohrlich-Leavitt, ed. Pp. 551-557. The Hague: Mouton Publishers. Burch , Ernest S., J r . 1979 The Ethnography of Northern North America: A Guide to Recent Research. Arctic Anthropology 16(1):62-146. Caamano, Jacinto 1938 The Journal of Jacinto Caamano. Henry R. Wagner and W . A . Newcombe, eds. Captain Harold Grenfell, translator. The Brit ish Columbia Historical Quarterly 2:189-222, 265-301. Campbell , K e n 1984 Hartley B a y , B . C . : A History. In The Tsimshian: Images of the Past, Views for the Present. Margaret Seguin, ed. Pp. 3-26. Vancouver, B . C . : University of Brit ish Columbia Press. Camsell , Charles 1954 Son of the North. Toronto: The Ryerson Press. Carmack, Robert M . 1972 Ethnohistory: A Review of its Development, Definitions, Methods, and Aims . A n n u a l Review of Anthropology 1:227-246. Chaney, Richard P. 1974 O n the Concepts of "Culture A r e a " and "Language Culture". In Comparative Studies by Harold E . Driver and Essays in His Honor. Joseph G . Jorgensen, ed. Pp. 237-245. New Haven: H R A F File. 241 Chapman, John W . 1914 Ten'a Texts and Tales F r o m Anvik , A laska . Publications of the American Ethnological Society 6. 1968[1906] Notes on the Tinneh Tribe of Anvik , A laska . Congres International des Americanistes. X V Volume 2:7-38. Reprint. Nendeln, Liechtenstein: K r a u s Reprint. Clignet, Remy 1970 M a n y Wives, M a n j ' Powers: Authority and Power in Polygynous Families. Evanston: Northwestern University Press. Cohen, Ronald 1978 Comments. In Women's Status in Egalitarian Society: Implications . For Social Evolution. B y Eleanor Leacock. Current Anthropology 19(2):257-259. Cohen, Ronald, and James W . VanStone 1963 Dependency and Self-Sufficiency in Chipewyan Stories. Canada. Department of Northern Affairs and National Resources 2:29-55. Collins, June M . 1949 John Fornsby: The Personal Document of a Coast Salish Indian. In Indians of the U r b a n Northwest. M a r i a n W . Smith, ed. Pp. 287-341. New York: Columbia University Press. 1950 Growth of Class Distinctions and Political Authority A m o n g the Skagit Indians Dur ing the Contact Period. Amer ican Anthropologist •52:331-342. Colson, Elizabeth 1953 The M a k a h Indians: A Study of an Indian Tribe in Modern American Society. Manchester: Manchester University Press. 1980 The Resilience of Matrilineality: Gwembe and Plateau Tonga Adaptations. In The Versati l ity of Kinship. L inda S. Cordell and Stephen Beckerman, eds. Pp. 359-374. New York: Academic Press. 242 Cowichan Chiefs 1885 Speeches of Cowichan Chiefs in Regard to the Potlatch Act . Department of Indian Affairs . (RG10, Volume 3628, File 6244-1). Victoria, Provincial Archives of British Columbia, Reel B-283. Cruikshank, Julie 1979 Athapaskan Women: Lives and Legends. National M u s e u m of M a n , M e r c m y Series, Canadian Ethnology Service 57. Curtis , Edward S. 1970a[1913] The North American Indian. Vol .9 Salishan Tribes of the Coast. Reprint. New York: Johnson Reprint Corporation. 1970b[1916] The North American Indian. Vol .11 The Nootka, The Haida . Reprint. New York: Johnson Reprint Corporation. Dahlberg, Frances 1981 Introduction. In Woman the Gatherer. Frances Dahlberg, ed. Pp. 1-33. New Haven: Yale University Press. Dahrendorf, R. 1969[1968] O n the Origin of Inequality A m o n g M e n . In Social Inequality: Selected Readings. Andre Beteille, ed. Pp. 16-44. Reprint. Baltimore, Mary land: Penguin Books. Davis , Kingsley, and Wilbert E . Moore 1945 Some Principles of Stratification. American Sociological Review 10:242-249. Davis , Philip W . , and Ross Saunders 1980 Bella Coola Texts. British Columbia Provincial Museum, Heritage Record No. 10. 243 Davydov, G.I . 1976 A Selection F r o m G.I. Davydov: A n Account of Two Voyages to America . Richard A . Pierce, ed. Colin Bearne, translator. Arctic Anthropology 13(2): 1-30. Dawson, George M . 1880 Appendix A . O n the Haida Indians of the Queen Charlotte Islands. In Report on the Queen Charlotte Islands 1878. Geological Survey of Canada. Pp. 103B-175B. Montreal: Dawson Brothers. 1888 Notes on the Indian Tribes of the Y u k o n District and Adjacent Northern Portion of British Columbia. A n n u a l Report of the Geological and Natura l History Survey of Canada 3:191B-213B. 1891 Notes on the Shuswap People of Brit ish Columbia. 2nd Section. Transactions of the Royal Society of Canada 9:3-44. Denniston, Glenda 1981 Sekani. In Handbook of North Amer ican Indians. Vol .6 The Subarctic. June He lm, ed. Pp. 433-441. Wil l iam Sturtevant, Gen. ed. Washington: Smithsonian Institution. Densmore, Frances 1939 Nootka and Quileute Music . Bureau of American Ethnology, Bulletin 124. Driver , Harold E . 1956 A n Integration of Functional, Evolutionary, and Historical Theory by Means of Correlations. Indiana Univers i ty Publications in Anthropology and Linguistics, Memoir 12. Pp. 1-35. 1969 Indians of North America . 2nd ed. Chicago: University of Chicago Press. 1970 Statistical Studies of Continuous Geographical Distributions. In A Handbook of Method in Cultural Anthropology. Raoul Narol l and Ronald Cohen, eds. Pp. 620-639. Garden City , N . Y . : The Natura l History Press. 244 1974a[1938] Culture Element Distributions: The Reliability of Culture Element Data . In Comparative Studies by Harold E . Driver and Essays in His Honor. Joseph G . Jorgensen, ed. Pp. 81-96. Reprint. New Haven: H R A F Press. 1974b[1939] Culture Element Distributions: Northwest California. In Comparative Studies by Harold E . Driver and Essays in his Honor. Joseph G . Jorgensen, ed. Pp. 96-106. Reprint of the Introduction. New Haven: H R A F File. 1974c[1966] Geographical-Historical Versus Psycho-Functional Explanations of K i n Avoidances. In Comparative Studies by Harold E . Driver and Essays in His Honor. Joseph G . Jorgensen, ed. Pp. 27-60. Reprint. New Haven: H R A F Press. Driver, Harold E . , and Richard P. Chaney 1974[1970] Cross-Cultural Sampling and Galton's Problem. In Comparative Studies by Harold E . Driver and Essays in His Honor. Joseph G . Jorgensen, ed. Pp. 109-121. Reprint. New Haven: H R A F Press. Driver , Harold E . , and James L . Coffin 1975 Classification and Development of North Amer ica Indian Cultures: A Statistical Analys is of the Driver-Massey Sample. Transactions of the American Philosophical Society n.s. 65. Drucker, Philip 1950 Culture Element Distributions: Northwest Coast. University of California Anthropological Records 9(3): 157-294. 1951 The Northern and Central Nootkan Tribes. Bureau of American Ethnology, Bulletin 144. 1963 Indians of the Northwest Coast. 2nd E d . New York: American M u s e u m of Natura l Historj^. 245 1983 Ecology and Political Organization on the Northwest Coast of America . In The Development of Political Organization in Native North America . 1979 Proceedings of the Amer ican Ethnological Society. Elisabeth Tooker, ed. Pp. 86-96. Washington, D . C . : The American Ethnological Society. Duff, Wilson 1952 The Upper Stalo Indians of the Fraser Valley, Brit ish Columbia. Anthropology in Brit ish Columbia Memoir 1. 1964 The Indian History of British Columbia. Vol.1 The Impact of the White M a n . Anthropology in British Columbia, Memoir 5. Duff, Wilson (ed.) 1959 Histories, Territories, and Laws of the Kitwancool. Anthropology in Brit ish Columbia, Memoir 4. Duncan, Wil l iam [Papers of Wil l iam Duncan]. Reels C2155, C2158, and C2159. L i b r a r y Microfilm Collections, Univers i ty of British Columbia. Vancouver. Durlach, Theresa M a y e r 1928 The Relationship Systems of the Tlingit, Haida and Tsimshian. Publications of the American Ethnological Society 2. Dyen , Isidore, and David F . Aberle 1974 Lexical Reconstruction: The Case of the Proto-Athapaskan Kinship System. London: Cambridge Univers i ty Press. Eisenstadt, S . N . 1956 F r o m Generation to Generation: Age Groups and Social Structure. New York: The Free Press of Glencoe, Collier-Macmillan Limited. 1968 Prestige, Participation and Strata Formation. In Social Stratification. J . A . Jackson, .ed. Pp. 62-103. Cambridge: Cambridge University Press. 246 Emmons , G . T . 1911 The Tahl tan Indians. University of Pennsylvania Museum, Anthropological Publications 4(1):3-121. Endicott, Wendell 1928 Adventures in A laska and Along the Tra i l . New York: Frederick A . Stokes Company. Engels, Frederick 1972[1884] The Origin of the Fami ly , Private Property and the State. Reprint. Eleanor Burke Leacock, ed. New York: International Publishers. Erickson, B . H . , and T . A . Nosanchuk 1977 Understanding Data . Toronto: M c G r a w - H i l l Ryerson Limited. Etienne, Mona , and Eleanor Leacock (eds.) 1980 Women and Colonization: Anthropological Perspectives. New York: Praeger Publishers. F a r r a n d , Livingston 1899 The Chilcotin. The British Association for the Advacement of Science 68:645-648. 1900 Traditions of the Chilcotin Indians. Memoirs of the American M u s e u m of Natura l History: The Jesup North Pacific Expedition 2:pt . l . Field, Poole 1957 [1913] The Poole Field Letters. June H e l m MacNeish , ed. Anthropologica 4:47-60. 247 Finley, M . I . 1973 The Ancient Economy. Berkeley and Los Angeles: University of California Press. Fischer, J . L . 1958 The Classification of Residence in Censuses. American Anthropologist 60:508-517. Fiske, Jo-Anne 1982 Protection for Young Ladies: Wil l iam Duncan's Mission to the Tsimshian. Unpublished manuscript. Franchere, Gabriel 1969 Journal of a Voyage on the North West Coast of North America during the Years 1811, 1812, 1813, and 1814. W . K a y e L a m b , ed. Wessie Tipping L a m b , trans. Toronto: The Champlain Society. Frankl in , John 1969a[1823] Narrat ive of a Journey to the Shores of the Polar Sea, In the Years 1819, 20, 21, and 22. Reprint. New York: Greenwood Press. 1969b[1828] Narrative of a Second Expedition to the Shores of the Polar Sea, In the Years 1825, 1826, and 1827. Including an Account of the Progress of a Detachment to the Eastward. Reprint. New York: Greenwood Press. Freeman, Bonnie Cook 1976 Power, Patriarchy, and 'Political Primitives'. In Beyond Intellectual Sexism: A N e w Woman, A New Reality. Joan I. Roberts, ed. Pp. 241-264. New York: David M c K a y Company, Inc. Freedman, Maurice 1978 M a i n Trends in Social and Cultural Anthropology. New York , N . Y . : Holmes and Meier Publishers, Inc. 248 Fried, Morton H . 1967 The Evolution of Political Society: A n Essay in Political Anthropology. New York: Random House, Inc. 1968 O n the Evolution of Social Stratification and the State. In Readings in Anthropology. Vol.2 Cultural Anthropology. 2nd E d . Morton H . Fr ied , ed. Pp. 462-478. New York: Thomas Y . Crowell Company. Friedl , Ernestine 1975 Women and Men: A n Anthropologist's View. New York: Holt, Rinehart and Winston. Galbreath, J . C . 1888 Notes on the T a h l - T a n Indians. In Notes on the Indian Tribes of the Yukon District and Adjacent Northern Portion of Brit ish Columbia. George M . Dawson, ed. A n n u a l Report of the Geological and Natura l History Survey of Canada 3:195B-199B. Garfield, Viola 1939 Tsimshian Clan and Society. University of Washington Publications in Anthropology 7(3): 167-340. 1947 Historical Aspects of Tlingit Clans in Angoon, Alaska . Amer ican Anthropologist 49:438-452. 1966[1951] The Tsimshian and Their Neighbours. In The Tsimshian Indians and Their Art s . Pp. 1-70. Reprint. Seattle: University of Washington Press. Giddens, Anthony 1971 Capital ism and Modern Social Theory: A n Analysis of the Writings of M a r x , Durkheim, and M a x Weber. Cambridge: Cambridge University Press. 249 Giele, Janet Zollinger 1977 Introduction: Comparative Perspectives on Women. In Women: Roles and Status in Eight Countries. Janet Z. Giele and Audrey Chapman Smock, eds. Pp. 3-31. New York: John Wiley & Sons. Gillespie, Beryl C . 1975 A n Ethnohistory of the Yellowknives: A Northern Athapaskan Tribe. In Contributions to Canadian Ethnology, 1975. Dav id B . Carlisle, ed. Pp. 191-245. National M u s e u m of M a n , Mercury Series, Canadian Ethnology Service 31. 198.1a Mountain Indians. In Handbook of North American Indians. Vol .6 The Subarctic. June H e l m , ed. Pp. 326-337. Wil l iam Sturtevant, Gen. ed. Washington: Smithsonian Institution. 1981b Yellowknife. In Handbook of North American Indians. Vol .6 The Subarctic. June Helm, ed. Pp. 285-290. Wil l iam Sturtevant, Gen. ed. Washington: Smithsonian Institution. Goddard, Pliny E . 1916 The Beaver Indians. Anthropological Papers of the American M u s e u m of Natura l History 10(4):203-293. Godelier, Maurice 1977 Perspectives in Marxis t Anthropology. Robert Bra in , trans. Cambridge: Cambridge University Press. Goldman, Irving 1941 The Alkatcho Carrier: Historical Background of Crest Perogatives. American Anthropologist 43:396-418. 1953 The Alkatcho Carrier of Brit ish Columbia. Unpublished Manuscript, Provincial Archives, Victoria. 1963[1940] The Alkatcho Carrier of British Columbia. In Acculturation in Seven American Indian Tribes. Ralph Linton, ed. Pp. 333-386. Reprint. Gloucester, Mass . : Peter Smith. 250 Goodenough, W a r d H . 1956 Residence Rules. Southwestern Journal of Anthropology 12:22-37. 1971[1965] Rethinking 'Status' and 'Role': Toward a General Model of the Cultural Organization of Social Relationships. In Readings in Kinship and Social Structure. Nelson Graburn , ed. Pp. 309-321. Reprint. New York: Harper and Row. Goody, Jack 1976 Production and Reproduction: A Comparative Study of the Domestic Domain. Cambridge: Cambridge University Press. Gough, Kathleen 1961 Matril ineal Kinship. David M . Schneider and Kathleen Gough, eds. Berkeley: University of California Press. 1975[1971] The Origin of the Fami ly . In Toward an Anthropology of Women. R a y n a R. Reiter, ed. Pp. 51-76. Reprint. New York: Monthly Review Press. Government of Canada 1900 Lis t of Indian Chiefs and Councillors, Brit ish Columbia. Government Sessional Papers (No. 14). 63 Victoria. Ottawa: Department of Indian Affairs. Grossman, Daniel 1965 The Nature of Descent Groups of Some Tribes in the Interior of Northwestern North America . Anthropologica n.s. 7:249-262. Guedon, Marie-Francoise 1974 People of Tetlin, W h y A r e You Singing? National M u s e u m of M a n , Mercury Series, Ethnology Division 9. 1981 Upper T a n a n a River Potlatch. In Handbook of North American Indians. Vol.6 The Subarctic. June H e l m , ed. Pp. 577-581. Wil l iam Sturtevant, Gen. ed. Washington: Smithsonian Institution. 251 1984a A n Introduction to Tsimshian Worldview and Its Practitioners. In The Tsimshian: Images of the Past, Views for the Present. Margaret Seguin, ed. Pp. 137-159. Vancouver, B . C . : University of British Columbia Press. 1984b Tsimshian Shamanic Images. In The Tsimshian: Images of the Past, Views for the Present. Margaret Seguin, ed. Pp. 174-211. Vancouver, B . C . : University of Brit ish Columbia Press. Gunther, E r n a 1942 Reminiscences of a Whaler's Wife. The Pacific Northwest Quarterly 33:65-69. Hale, Horatio 1890 Remarks on the Ethnology of Brit ish Columbia: Introductory to the Second General Report of D r . F r a n z Boas on the Indians of T h a t Province. The British Association for the Advancement of Science 60:553-562. Hardisty , Wil l iam L . 1872 The Loucheux Indians. In Notes on the Tinneh or Chipewyan Indians of British and Russian America . Annua l Report of the Smithsonian Institution for 1866. Pp. 311-320. Halpin , Marjorie M . 1973 The Ts imshian Crest System: A Study Based on M u s e u m Specimens and the Marius Barbeau and Wil l iam Beynon Field Notes. P h . D . Dissertation, Department of Anthropology, University of British Columbia, Vancouver. 1984a Feast Names of Hartley Bay . In The Tsimshian: Images of the Past, Views for the Present. Margaret Seguin, ed. Pp. 57-64. Vancouver, B . C . : University of Brit ish Columbia Press. 1984b Seeing in Stone: Tsimshian Masking and the T w i n Stone Masks . In The Tsimshian: Images of the Past, Views for the Present. Margaret Seguin, ed. Pp. 281-307. Vancouver, B . C . : University of British Columbia Press. Harmon, Daniel W . 1957 Sixteen Years in the Indian Country: The Journal of Daniel Will iams H a r m o n , 1800-1816. W . K a y e L a m b , ed. Toronto: The Macmil lan Company of Canada Limited. Harrison, Charles 1892a F a m i l y Life of the Haidas, Queen Charlotte Islands. The Journal of the Anthropological Institute of Great Britain and Ireland 21:470-476. 1892b Religion and Fami ly A m o n g the Haidas (Queen Charlotte Islands). The Journal of the Anthropological Institute of Great Britain and Ireland 21:14-29. Hearne, Samuel 1968[1775] A Journey F r o m Prince of Wales Fort in Hudson's Bay to the Northern Ocean in the Years 1769, 1770, 1771, and 1772. Reprint. New York: Greenwood Press. Helm, June 1961 The L y n x Point People: The Dynamics of a Northern Athapaskan Band. National M u s e u m of Canada , Bulletin 176, Anthropological Series 53. 1965a Bilaterality in the Socio-territorial Organization of the Arctic Drainage Dene. Ethnology 4:361-385. 1965b Patterns of Allocation A m o n g the Arctic Drainage Dene. In Essays in Economic Anthropology. Proceedings of the 1965 A n n u a l Spring Meeting of the American Ethnological Society. June H e l m , ed. Pp. 33-45. Seattle: University of Washington Press. 1968 The Nature of Dogrib Socio-territorial Groups. In M a n the Hunter. Richard B . Lee and Irven DeVore, eds. Pp . 118-125. New York: Aldine Publishing Co. 1972 The Dogrib Indians. In Hunters and Gatherers Today. M . G . Bicchieri, ed. Pp. 51-89. New York: Holt, Rinehart, and Winston, Inc. 1981a Dogrib. In Handbook of North American Indians. Vol .6 The Subarctic. June H e l m , ed. Pp. 291-309. Wil l iam Sturtevant, Gen. ed. Washington: Smithsonian Institution. 253 1981b Introduction. In Handbook of North American Indians. Vol .6 The Subarctic. June Helm, ed. Pp. 1-4. Wil l iam Sturtevant, Gen. ed. Washington: Smithsonian Institution. He lm, June, and Eleanor B . Leacock 1971 The Hunt ing Tribes of Subarctic America . In North American Indians in Historical Perspective. Eleanor B . Leacock and Nancy O. Lur ie , eds. Pp. 343-374. New York: Random House Hill-Tout, Charles 1902 Ethnological Studies of the Mainland Halkome'Lem, A Division of the Salish of British Columbia. The British Association for the Advancement of Science 72:355-449. 1978[1904] Ethnological Report on the Stseelis [Chehalis] and Skaulits [Scowlitz] Tribes of the Halkomelem Division of the Salish of British Columbia. In The Salish People: The Local Contributions of Hil l -Tout. Vol .3 The Mainland Halkomelem. Ralph M a u d , ed. Pp. 95-158. Vancouver: Talonbooks. Honigmann, John J . 1946 Ethnography and Acculturation of the Fort Nelson Slave. Yale University Publications in Anthropology 33. 1949 Culture and Ethos of K a s k a Society. Yale University Publications in Anthropology 40. 1954 The K a s k a Indians: A n Ethnographic Reconstruction. Yale University Publications in Anthropology 51. 1981 K a s k a . In Handbook of North American Indians. Vol .6 The Subarctic. June He lm, ed. Pp . 442-450. Wil l iam Sturtevant, Gen. ed. Washington: Smithsonian Institution. Hopwood, Victor G . (ed.) 1971 David Thompson: Travels in Western North Amer ica , 1784-1812. Toronto: MacMi l lan . 254 Hosley, Edward H . 1981 Kolchan. In Handbook of North Amer ican Indians. Vol .6 The Subarctic. June Helm, ed. Pp. 618-622. Will iam Sturtevant, Gen. ed. Washington: Smithsonian Institution. Hudson, Charles 1973 The Historical Approach in Anthropology. In Handbook of Social and Cultural Anthropology. John J . Honigmann, ed. Pp. 111-141. Chicago: Rand M c N a l l y College Publishing Co. Jacobsen, Johan Adr ian 1977 Alaskan Voyage 1881-1883: A n Expedition to the Northwest Coast of America . Translated by E r n a Gunther from the G e r m a n text of Adr ian Woldt. Chicago and London: The University of Chicago Press. Jeffries, Vincent, and H . Edward Ransford 1980 Social Stratification: A Multiple Hierarchy Approach. Boston: A l l y n and Bacon, Inc. Jenness, Diamond 1937 The Sekani Indians of British Columbia. National M u s e u m of Canada, Bulletin 84, Anthropological Series 20. 1943 The Carrier Indians of the Bulkley River: Their Social and Religious Life. Anthropological Papers 25, Bureau of American Ethnology, Bulletin 133:468-586. 1955 The Fa i th of a Coast Salish Indian. Anthropology in Brit ish Columbia, Memoir 3. 1977 The Indians of Canada. Seventh Edition. Toronto: University of Toronto Press. N . D . The Saanich Indians of Vancouver Island. Unpublished, uncompleted manuscript. Victoria, B . C . , Provincial Archives. 255 Jenness, Diamond (ed.) 1956 The Chippewyan Indians: A n Account B y an E a r l y Explorer. Anthropologica 3:15-33. Jewitt, John R. 1976[1807] A Journal Kept at Nootka Sound. Reprint. New York: Garland Publishing Inc. Jones, Livingston F . 1914 A Study of the Thlingets of Alaska . N e w York: Fleming H . Revell Company. Jones, Strachan 1872 The Kutchin Tribes. In Notes on the Tinneh or Chipewyan Indians of Brit ish and Russian America . A n n u a l Report of the Smithsonian Institution for 1866. Pp. 320-327. Jorgensen, Joseph G . 1969 Salish Language and Culture: A Statistical Analys is of Internal Relationships, History, and Evolution. Bloomington, Indiana: Indiana University Press. 1974a O n Continuous A r e a and Worldwide Studies in Formal Comparative Ethnology. In Comparative Studies by Harold E . Driver and Essays in His Honor. Joseph G . Jorgensen, ed. Pp. 195-203. New Haven: H R A F Press. 1974b O n Methodology: Reliability, Statistics, Sampling. In Comparative Studies by Harold E . Driver and Essays in His Honor. Joseph G . Jorgensen, ed. Pp. 81. New Haven: H R A F Press. 1980 Western Indians: Comparative Environments, Languages, and Cultures of 172 Western American Indian Tribes. San Francisco: W . H . Freeman and Company. Josephy, A l v i n M . 1968 The Indian Heritage of America . New York: Bantam Books. 256 Kalifornsky, Peter 1977 Kahtnuht'ana Qenaga The Kena i People's Language. J i m K a r i , ed. Fairbanks, Alaska: A laska Native Language Center, University of Alaska. Kei th , George 1890 Letters to M r . Roderic McKenzie 1807-1817. In Les Bourgeois de la Compagnie D u Nord-Ouest, Vol.2:61-132. L . R . Masson, ed. 2 Vols. Quebec: A . Cote et C . Kew, J . E . M . 1970 Coast Salish Ceremonial Life: Status and Identity in a Modern Village. P h . D . Dissertation, Department of Anthropology, University of Washington, Seattle. Kle in , L a u r a F . 1975 Tlingit Women and Town Politics. P h . D . Dissertation, Department of Anthropology, New York University, New York . 1980 Contending with Colonization: Tlingit M e n and Women in Change. In Women and Colonization: Anthropological Perspectives. M o n a Etienne and Eleanor Leacock, eds. Pp. 88-108. New York: Praeger Publishers. K n a p p , Frances, and Rheta L . Childe 1896 The Thlinkets of Southeastern Alaska . Chicago: Stone and Kimbal l . Kobrinsky, Vernon H . 1977 The Tsimshianization of the Carrier Indians. In Problems in the Prehistory of the North American Subarctic. J . W . Helmer, S. V a n Dyke, and F . J . Kense, eds. Pp. 201-210. Calgary, Alberta: University of Calgary Archaeological Association. Koppert, Vincent A . 1930a Contributions to Clayoquot Ethnology. The Catholic University of America , Anthropological Series 1. 257 1930b The Nootka Fami ly . Primitive M a n 3(3,4):49-55. Krause , Aure l 1956 [1885] The Tlingit Indians: Results of a Tr ip to the Northwest Coast of Amer ica and the Bering Straits. E r n a Gunther, trans. Vancouver: Douglas and Mclntyre . Krech , Shepard, III 1976 The Eastern Kutchin and the F u r Trade, 1800-1860. Ethnohistory 23:213-235. 1983 The Influence of Disease and the F u r Trade on Arctic Drainage Lowlands Dene, 1800-1850. Journal of Anthropological Research 39(1):123-146. de Laguna , Frederica 1953 Some Problems in the Relationship between Tlingit Archaeology and Ethnology. In Memoirs of the Society for American Archaeology. No.9 A s i a and North Amer ica Trans-Pacific Contacts. M a r i a n W . Smith, ed. Pp. 53-57. Salt Lake City , Utah: The Society for Amer ican Archaeology. 1960 The Story of a Tlingit Community: A Problem in the Relationship between Archaeological, Ethnological, and Historical Methods. Bureau of Amer ican Ethnology, Bulletin 172. 1972 Under Mount Saint Elias: the History and Culture of the Yakutat Tlingit. 3 Vols. Smithsonian Contributions to Anthropology, Vol .7 . 1975 Matri l ineal K i n Groups in Northwestern North America. In Proceedings: Northern Athapaskan Conference, 1971. Vol .1 . Pp. 19-145. A . M c F a d y e n Clark, ed. National M u s u m of M a n , Mercury Series, Canadian Ethnology Service 27. 2 Vols. 1983 Aboriginal Tlingit Sociopolitical Organization. The Development of Political Organization in Native North America . In 1979 Proceedings of the American Ethnological Society. Elizabeth Tooker, ed. Pp. 71-85. Washington, D . C . : American Ethnological Society. 258 de Laguna , Frederica, and Catharine McCle l lan 1981 Ahtna . In Handbook of North Amer ican Indians. Vol .6 The Subarctic. June Helm, ed. Pp. 641-663. Wil l iam Sturtevant, Gen. ed. Washington: Smithsonian Institution. Labovitz, Sanford 1970 The Assignment of Numbers to Rank Order Categories. American Sociological Review 35(3):515-524. Laforet, Andrea 1984 Tsimshian Basketry. In The Tsimshian: Images of the Past, Views for the Present. Margaret Seguin, ed. Pp. 215-280. Vancouver, B . C . : University of British Columbia Press. Lamphere , Louise 1977 Review Essay: Anthropology. Signs 2(2):612-627. Lane , Robert B . 1953 Cultural Relations of the Chilcotin Indians of West Central British Columbia. P h . D . Dissertation, Department of Anthropology, University of Washington, Seattle. 1981 Chilcotin. In Handbook of North American Indians. Vol .6 The Subarctic. June He lm, ed. Pp. 402-412. Wil l iam Sturtevant, Gen. ed. Washington: Smithsonian Institution. Leacock, Eleanor Burke 1972 Introduction. In The Origin of the Fami ly , Private Property and the State. B y Frederick Engels. Pp . 7-67. New York: International Publishers. 1974 Introduction. In Ancient Society or Researches in the Lines of H u m a n Progress F r o m Savagery Through Barbarism to Civilization. B y Lewis Henry Morgan. Pp. Ii-Ixx, Ili-IIxx, Illi-IIIviii, IVi-IVii i . Gloucester, Mass . : Peter Smith. 1975 Class, Commodity, and the. Status of Women. In Women Cross-Culturally: Change and Challenge. Ruby Rohrlich-Leavitt, ed. Pp. 601-616. The Hague: Mouton Publishers. 259 1978 Women's Status in Egal i tarian Society: Implications for Cultural Evolution. Current Anthropology 19:247-275. 1979 Women, Development, and Anthropological Facts and Fictions. In The Politics of Anthropology: F r o m Colonialism and Sexism Toward a View F r o m Below. Gerrit Huizer and Bruce Mannhe im, eds. Pp. 131-142. The Hague: Mouton Publishers. 1981 Myths of Male Dominance: Collected Works on Women Cross-Culturally. New York: Monthly Review Press. 1983 Ethnohistorical Investigation of Egal i tarian Politics in Eastern North America . The Development of Political Organization in Native North America . In 1979 Proceedings of the American Ethnological Society. Elisabeth Tooker, ed. Pp. 17-31. Washington, D . C . : American Ethnological Society. Leacock, Eleanor, and June N a s h 1981[1977] Ideologies of Sex: Archetypes and Stereotypes. In Myths of Male Dominance: Collected Articles on Women Cross-Culturally. Eleanor Burke Leacock, ed. Pp. 242-263. Reprint. New York: Monthly Review Press. Legros, Dominique 1981 Structure Socio-Culturelle et Rapports de Domination Chez Les Indiens Tutchone Septentrionaux D u Yukon au Dix-Neuvieme Siecle. 2 Vols. Vancouver: University of British Columbia. 1982 Reflexions sur l'origine des inegalites sociales a partir du cas des Athapaskan Tutchone. Culture 2(3):65-84. Leland, Donald 1983 Was Nuu-chah-nulth-aht (Nootka) Society Based on Slave Labour? In The Development of Political Organization in Native North America . 1979 Proceedings of the American Ethnological Society. Elisabeth Tooker, ed. Pp. 108-119. Washington, D . C . : American Ethnological Society. 260 Lenski , Gerhard E . 1966 Power and Privilege: A Theory of Social Stratification. New York: M c G r a w - H i l l Book Co. Levine, Robert A . 1966 Sex Roles and Economic Change in Afr ica . Ethnology 5:186-192. Levy , Marion J . 1952 The Structure of Society. Princeton: Princeton University Press. Linton, Ralph 1971[1935] Status and Role. In Readings in Kinship and Social Structure. Nelson Graburn , ed. Pp. 291-297. Reprint. New York: Harper and Row. Lips , Hi lary M . 1981 Women, M e n , & the Psychology of Power. Englewood Cliffs, New Jersey: Prentice-Hall, Inc. Loether, H e r m a n J . , and Donald G . M c T a v i s h 1980 Descriptive and Inferential Statistics: A n Introduction. 2nd E d . Boston: A l l y n and Bacon, Inc. McClel lan, Catharine 1954 The Interrelations of Social Structure with Northern Tlingit Ceremonialism. Southwestern Journal of Anthropology 10:75-96. 1956 Shamanistic Syncretism in Southern Yukon. Transactions of the New York Academy of Sciences, Series 2 19:130-137. 1961 Avoidance Between Siblings of the same Sex in Northwestern North America . Southwestern Journal of Anthropology 17:103-123. 261 1975 Feuding and Warfare among Northwestern Athapaskans. In Proceedings: Northern Athapaskan .Conference, 1971. Vol .1 . A . M c F a d y e n Clark, ed. Pp. 181-258. National M u s u m of M a n , Mercury Series, Canadian Ethnology Service 27. 2 Vols. 1981 Inland Tlingit. In Handbook of North American Indians. Vol .6 The Subarctic. June H e l m , ed. Pp. 469-480. Wil l iam Sturtevant, Gen. ed. Washington: Smithsonian Institution. McCle l lan , Catharine, and Glenda Denniston 1981 Environment and Culture in the Cordillera. In Handbook of North American Indians. Vol .6 The Subarctic. June He lm, ed. Pp. 372-386. Wil l iam Sturtevant, Gen. ed. Washington: Smithsonian Institution. M c C u r d y , David W . , and James P. Spradley (eds.) 1979 Issues in Cultural Anthropology: Selected Readings. Boston: Little, Brown and Company. McDonald , James A . 1984 Images of the Nineteenth-Century Economy of the Tsimshian. In The Tsimshian: Images of the Past, Views for the Present. M . Seguin, ed. Pp. 40-54. Vancouver, B . C . : University of Brit ish Columbia Press. McDonnel l , Roger F . 1975 Kas in i Society: Some Aspects of the Social Organization of an Athapaskan Culture Between 1900-1950. P h . D . Dissertation, Department of Anthropology, University of British Columbia, Vancouver. McFeat , T o m (ed.) 1978 Indians of the North Pacific Coast: Studies in Selected Topics. Toronto: The M a c M i l l a n Company of Canada Limited. 262 Mcl lwra i th , T . F . 1948 The Bella Coola Indians, 2 Vols. Toronto: University of Toronto Press. M c K e n n a n , Robert A . 1959 The Upper Tanana Indians. Yale University Publications in Anthropology 55. 1965 The Chandalar Kutchin. Arctic Institute of North America Technical Paper 17. 1969 Athapaskan Groups of Central A l a s k a at the Time of White Contact. Ethnohistory 16:335-343. 1981 Tanana . In Handbook of North American Indians. Vol .6 The Subarctic. June Helm, ed. Pp. 562-576. Wil l iam Sturtevant, Gen. ed. Washington: Smithsonian Institution. Mackenzie, Alexander 1966 Exploring the Northwest Territory: Sir Alexander Mackenzie's Journal of a Voyage by Bark Canoe from Lake Athabasca to the Pacific Ocean in the Summer of 1789. T . H . McDonald , ed. Norman: University of Oklahoma Press. 1971[1801] Voyages F r o m Montreal on the River St. Laurence Through the Continent of North Amer ica to the Frozen and Pacific Oceans In the Years 1789 and 1793, Wi th a Prel iminary Account of the Rise, Progress, and Present State of the F u r Trade in That Country. Reprint. Edmonton: M . G . Hurt ig L t d . MacLach lan , Bruce B . 1981 Tahltan. In Handbook of North American Indians. Vol .6 The Subarctic. June Helm, ed. Pp. 458-468. Wil l iam Sturtevant, Gen. ed. Washington: Smithsonian Institution. MacNeish , June H e l m 1956 Leadership Among the Northeastern Athabascans. Anthropologica 2:131-163. 263 M a r t i n , M . K a y and B a r b a r a Voorhies 1975 Female of the Species. New York: Columbia University Press. Mason , J . Alden 1946 Notes on the Indians of the Great Slave Lake A r e a . Ya le University Publications in Anthropology 34. Mason , Michael H . 1934U924] The Arctic Forests. Reprint. London: Hodder and Stoughton Limited. Matthiasson, Carolyn J . 1974 Conclusion. In M a n y Sisters: Women in Cross-Cultural Perspective. Carolyn J . Matthiasson, ed. Pp. 421-437. New York: The Free Press. Mayne , R . C . 1862 Four Years in British Columbia and Vancouver Island: A n Account of Their Forests, Rivers, Coasts, Gold Fields, and Resources for Colonization. London: John M u r r a y . Michea, Jean 1959 Les Chitra-Gottineke: Groupe Athapascan des Montagnes Rocheuses. Journal de la Societe des Americanistes 48:197-235. Mil ler , J a y 1984 Feasting With the Southern Tsimshian. In The Tsimshian: Images of the Past, Views for the Present. M . Seguin, ed. Pp. 27-39. Vancouver, B . C . : University of British Columbia Press. Mil ler , Walter B . 1955 Two Concepts of Authority. American Anthropologist 57:271-289. 264 Mitchell , Marjorie R. 1976 Women, Poverty, and Housing: Some Consequences of Hinterland Status for a Coast Salish Indian Reserve in Metropolitan Canada. P h . D . Dissertation, Department of Anthropology, University of British Columbia, Vancouver. Morgan , Lewis H e n r y 1974[1872] Ancient Society or Researches in the Lines of H u m a n Progress F r o m Savagery Through Barbar ism to Civilization. Reprint. Eleanor Burke Leacock, ed. Gloucester, Mass . : Peter Smith. Morice, Adrien G . 1892 A r e the Carrier Sociology and Mythology Indigenous or Exotic? Proceedings and Transactions of the Royal Society of Canada s e r . l , sect.2, Vol . 10:109-126. 1895 Notes Archaeological, Industrial, and Sociological, on the Western Denes. Transactions of the Canadian Institute 4:1-222. 1897 A u Pays de l'Ours Noir: Chez les Sauvages de la Colombie Britannique. Paris: Delhomme et Briguet, editeurs. 1903 The Nah-ane and Their Language. Transactions of the Canadian Institute 7:517-534. 1906 The Great Dene Race. Anthropos l(2):229-277, 483-508; l(2):695-730. 1907 The Great Dene Race. Anthropos 2:1-34, 181-196. 1909 The Great Dene Race. Anthropos 4:582-606. 1910 The Great Dene Race. Anthropos 5:113-142, 419-443, 643-653, 969-990. Mozino, Jose Mariano 1970 Noticias de Nutka: A n Account of Nootka Sound in 1792. Iris Higbie Wilson, ed. and trans. Toronto: McClel land and Stewart L t d . 265 Mueller, John H . , K a r l F . Schuessler, and Herbert L . Costner 1977 Statistical Reasoning In Sociology. 3rd E d . Boston: Houghton Mifflin Company. Murdock, George Peter 1934a The Haidas of British Columbia. In Our Primitive Contemporaries. Pp. 221-263. New York: The M a c M i l l a n Company. 1934b Kinship and Social Behavior A m o n g the Haida . Amer ican Anthropologist 36:355-385. 1965a[1937] Comparative Data on the Division of Labor by Sex. In Culture and Society. George Peter Murdock, ed. Pp. 308-310. Pittsburgh: University of Pittsburgh Press. 1965b[1936] Rank and Potlatch A m o n g the Haida . In Culture and Society. George Peter Murdock, ed. Pp. 262-289. Reprint. Pittsburgh, P . A . : University of Pittsburgh Press. 1965c Social Structure. Free Press Edition. New York: The Free Press. 1981 Atlas of World Cultures. Pittsburgh, P . A . : University of Pittsburgh Press. Murdock, George Peter, and Caterina Provost 1980[1973] Factors in the Division of Labour by Sex: A Cross-Cultural Analysis . In Cross-Cultural Samples and Codes. Herbert B a r r y III and Alice Schlegel, eds. Pp. 289-311. Reprint. Pittsburgh, P . A . : University of Pittsburgh Press. Namboodiri, N . Kr i shnan , Lewis F . Carter, and Hubert M . Blalock, J r . 1975 Applied Multivariate Analysis and Experimental Designs. New York: M c G r a w - H i l l Book Company. 266 Narol l , Raoul 1970 Galton's Problem. In A Handbook of Method in Cultural Anthropology. Raoul Naroll and Ronald Cohen, eds. Pp. 974-989. Garden City, N . Y . : The Natural History Press. Naroll , Raoul, G a r y L . Michik, and F r a d a Narol l 1974 Hologeistic Theory Testing. In Comparative Studies by Harold E . Driver and Essays in His Honor. Joseph G . Jorgensen, ed. Pp. 121-148. New Haven: H R A F File. Nelson, Edward W . 1978 [1880]. E . W . Nelson's Notes on the Indians of the Yukon and Innoko Rivers, Alaska . James W . VanStone, ed. Fieldiana Anthropology 70. Netting, Robert M c C . 1979 Women's Weapons: The Politics of Domesticity A m o n g the Kofyar . In Issues in Cultural Anthropology: Selected Readings. David W . M c C u r d y and James P. Spradley, eds. Pp. 166-175. Boston: Little, Brown and Company. Niblack, Albert P. 1890 The Coast Indians of Southern A l a s k a and Northern British Columbia. Report of the National Museum. Pp. 225-386. Nielsen, Joyce M c C a r l 1978 Sex in Society: Perspectives on Stratification. Belmont, Calif.: Wadsworth Publishing Company, Inc. Oakley, A n n 1972 Sex, Gender & Society. New York: Harper and Row, Publishers. 267 O ' B a r r , Jean 1976 Third World Women: Factors in their Changing Status. Duke University Center for International Studies, Occasional Paper 2. Oberg, Kalervo 1973 The Social Economy of the Tlingit Indians. Seattle: University of Washington Press. Olson, R . L . 1967 Social Structure and Social Life of the Tlingit in Alaska . University of California Publications, Anthropological Records 26. 1968[1936] Some Trading Customs of the Chilkat Tlingit. In Essays in Anthropology: Presented to A . L . Krober in Celebration of His Sixtieth Birthday. Pp. 211-214. Reprint. Freeport, New York: Books For Libraries Press, Inc. Ortner, Sherry B . 1974 Is Female to Male as Nature is to Culture? In W o m a n , Culture, and Society. Michelle Z . Rosaldo and Louise Lamphere , eds. Pp. 67-87. Stanford: Stanford University Press. Osgood, Cornelius 1932 The Ethnography of the Great Bear Lake Indians. National Museum of Canada, Bulletin 70, Annua l Report for 1931. Pp . 31-98. 1933 Tanaina Culture. American Anthropologist 35:695-717. 1936a Contributions to the Ethnography of the Kutchin. Yale University Publications in Anthropology 14. 1936b The Distribution of the Northern Athapaskan Indians. Yale University Publications in Anthropology 7. 1940 Ingalik Material Culture. Yale University Publications in Anthropology 22. 1958 Ingalik Social Culture. Yale University Publications in Anthropology 53. 268 1959 Ingalik Mental Culture. Yale University Publications in Anthropology 56. 1966U937] The Ethnography of the Tanaina . Reprint. New Haven: H R A F Press. 1971 The H a n Indians: A Compilation of Ethnographic and Historical Data O n the Alaska-Yukon Boundary A r e a . Yale University Publications in Anthropology 74. Parsons, Elsie Clews 1921-22 A Narrat ive of the Ten'a of Anvik , Alaska . Anthropos 16-17:51-71. Parsons, Talcott 1954 A n Analytical Approach to the Theory of Social Stratification. In Essays in Sociological Theory by Talcott Parsons. Pp. 69-88. Revised Edition. New York: The Free Press. London: Coll ier-Macmillan L t d . Pelto, Pertti J . , and Gretel H . Pelto 1978 Anthropological Research: The Structure of Inquiry. 2nd E d . Cambridge: Cambridge University Press. Perry, Richard J . 1977 Variations on the Female Referent in Athapaskan Cultures. Journal of Anthropological Research 33:99-118. 1979 The F u r Trade and the Status of Women in the Western Subarctic. Ethnohistory 26:363-375. Petitot, E . 1876 Monographie des Dene-Dindjie. Paris: Ernest Leroux, Editeur. 1891 Autour du Grand L a c des Esclaves. Deuxieme Edition. Paris: Nouvelle Librairie Parisienne. Albert Savine, Editeur. 269 Petroff, Ivan 1884 Report on the Population, Industries, and Resources of Alaska . Tenth Census, 1880. Washington: U . S . Government Printing Office. Pirenne, Henr i 1959[1931] What are Historians T r y i n g to Do? In The Philosophy of History in Our Time. Hans Meyerhoff, ed. Pp. 87-100. Reprint. Garden City , N . Y . : Doubleday & Company, Inc. Poewe, K a r l a O. 1980 Matril ineal Ideology: The Economic Activities of Women in Luapula , Zambia. In The Versatility of Kinship. L inda S. Cordell and Stephen Beckerman, eds. Pp. 333-357. New York: Academic Press. Powell, Addison M . 1910 Trai l ing and Camping in Alaska . New York: Wessels & Bissell. Quinn, Naomi 1977 Anthropological Studies on Women's Status. A n n u a l Review of Anthropology 6:181-225. Rapp, R a y n a 1979 Review Essay: Anthropology. Signs: Journal of Women in Culture and Society 4(3):497-513. Raunet, Daniel 1984 Without Surrender Without Consent: A History of the Nishga L a n d Claims. Vancouver: Douglas and Mclntyre . Ray , Verne F . 1939 Cultural Relations in the Plateau of North Western America . Publications of the Frederick Webb Hodge Anniversary Publication Fund 3. 270 1942 Culture Element Distributions: The Plateau. University of California Anthropological Records 8(2):99-262. Reed, T . B . , and Elsie Clews Parsons 1922 Cries-For-Salmon, A Ten'a Woman. In American Indian Life: B y Several of its Students. Elsie Clews Parsons, ed. Pp. 337-361. New York: B . W . Huebsch, Inc. Richardson, Sir John 1852 Arctic Searching Expedition: A Journal of a Boat-Voyage Through Rupert's L a n d and the Arctic Sea. New York: Harper & Brothers, Publishers. Ridington, Robin 1969 K i n Categories Versus K i n Groups: A Two-Section System Without Sections. Ethnology 8:460-467. 1971 Beaver Dreaming and Singing. Anthropologica 13:115-128. 1981 Beaver. In Handbook of North American Indians. Vol .6 The Subarctic. June He lm, ed. Pp. 350-360. Wil l iam Sturtevant, Gen. ed. Washington: Smithsonian Institution. 1983 Stories of the Vision Quest A m o n g Dunne-Za Women. Atlantis 9(l):68-78. Rogers, E d w a r d S. 1981 History of Ethnological Research in the Subarctic Shield and Mackenzie Borderlands. In Handbook of North American Indians. Vol.6 The Subarctic. June He lm, ed. Pp. 19-29. Wil l iam Sturtevant, Gen. ed. Washington: Smithsonian Institution. Rogers, Susan Carol 1978 Woman's Place: A Critical Review of Anthropological Theory. Comparative Studies in Society and History 20:123-162. 271 Rohrlich-Leavitt, Ruby 1975 Conclusions. In Women Cross-Culturally: Change and Challenge. Ruby Rohrlich-Leavitt, ed. Pp. 619-641. The Hague: Mouton Publishers. Rohrlich-Leavitt, Ruby, Barbara Sykes, and Elizabeth Weatherford 1979 Aboriginal Woman: Male and Female Anthropological Perspectives. In The Politics of Anthropology: F r o m Colonialism and Sexism Toward a View F r o m Below. Gerrit Huizer and Bruce Mannheim, eds. Pp. 117-129. The Hague: Mouton Publishers. Romeny, A . Kimbal l , Susan C. Weller, and Wil l iam H . Batchelder 1986 Culture as Consensus: A Theory of Culture and Informant Accuracy. American Anthropologist 88(2):313-338. Rosaldo, Michelle Z . 1974 Woman, Culture, and Society: A Theoretical Overview. In Woman, Culture, and Society. Michelle Z . Rosaldo and Louise Lamphere , eds. Pp. 17-42. Stanford: Stanford University Press. Rosenfeld, E v a 1951 Social Stratification in a 'Classless' Society. American Sociological Review 16:766-774. Rosman, A b r a h a m , and Paula G . Rubel 1971 Feasting With Mine Enemies: Rank and Exchange A m o n g Northwest Coast Societies. New York and London: Columbia University Press. Ross, Bernard R. 1872 The Eastern Tinneh. In Notes on the Tinneh or Chipewyan Indians of Brit ish and Russian America . A n n u a l Report of the Smithsonian Institution for 1866. Pp. 304-311. 272 Rowntree, Derek 1981 Statistics Without Tears: A Primer for Non-mathematicians. Harmondsworth, Middlesex, England: Penguin Books. Rubel, Paula G . , and A b r a h a m Rosman 1983 The Evolution of Exchange Structure and Ranking: Some Northwest Coast and Athapaskan Examples. Journal of Anthropological Research 39:1-25. Rubin, Gayle 1975 The Traffic in Women. In Toward an Anthropology of Women. R a y n a R. Reiter, ed. Pp. 157-210. N e w York: Monthly Review Press. Runciman, W . G . 1968 Class, Status and Power? In Social Stratification. J . A . Jackson, ed. Pp. 25-61. Cambridge: Cambridge University Press. • 1969U966] The Three Dimensions of Social Inequality. In Social Inequality: Selected Readings. Andre Beteille, ed. Pp. 45-63. Reprint. Baltimore, Mary land: Penguin Books. Rushforth, Scott 1984 Bear Lake Athapaskan Kinship and T a s k Group Formati