BY WILL ROSCOE

The men arc strongly inclined to sodomy; but the boys that abandon themselves thus are excluded from the society of men and sent out to that of women as being effeminates. They are confused with the Hermaphrodites which they say are found in quantity in the country of the Floridians. I believe that these Hermaphrodites are none other than the effeminate boys, that in a sense truly are hermaphrodites. Be that as it may, they employ them in all the diverse handiworks of women, in servile functions, and to carry the munitions and provisions of war. They are also distinguished from the men and the women by the color of the feathers that they put on their heads and for the scorn than they bring au to themselves. --Francisco Coreal

Introduction: The Problem of Translation

This was how the Spanish traveler Francisco Coreal, who visited Florida in 1669, described the social role that anthropologists now term berdache. The presence of berdaches had been well documented by Coreal's time. In fact, Coreal was clearly aware of these reports, for he devoted as many words to clarifying their discrepancies as he did to reporting his own observations. This pattern can be traced from the earliest accounts of the Spaniards to present-day ethnographies. What has been written about berdaches reflects more the influence of existing Western discourses on gender, sexuality and the Other than what observers actually witnessed.

Typically described, in the words of Matilda Stevenson, as men who "adopt woman's dress and do woman's work," male berdaches have been documented in nearly 150 North American societies. In nearly half of these groups, a social status also has been documented for females= who undertook a man's life-style, who were sometimes referred to in the native language with the same. term applied to male berdaches and sometimes with a distinct term. Although the existence of berdaches has long been known to specialists in North American anthropology, the subject has been consigned to footnotes and marginal references. In the past twenty years, however, berdaches have become a subject of growing interest. An expanding base of empirical data concerning the social, cultural and historical dimensions of berdache status has become available.

In this essay, I take advantage of this recent work to propose a theoretical model of berdache roles as distinct gender categories (e.g., third and fourth genders). I begin with a review of the history of the anthropological study of berdache roles and a summary of recent literature on the subject. Following this, I propose a new analytical program for theorizing berdache roles and demonstrate its utility by applying it to several of the outstanding questions in berdache studies.

Until quite recently, serious investigation of berdaches has been confined to the most basic problems of description and definition. Throughout five centuries of contact, a bewildering variety of terms has been employed by Europeans and Americans to name this status, with new ones introduced in almost every generation. Such practices have created doubt not only about the nature of berdache roles but also concerning their very presence in cases in which confusing terminology makes it difficult to know whether different writers were referring to the same phenomena. The difficulty is that Euro-American cultures lack social and linguistic categories that can translate the pattern of beliefs, behaviors and customs represented by North American berdaches. Instead, writers have chosen between mutually exclusive terms that emphasize either gender variation or sexual variation - "hermaphrodite" and "sodomite," for example, or, more recently, "transsexual" (gender) and "homosexual" (sexuality). Berdache was originally an Arabic and Persian term for the younger partner in a male homosexual relationship, synonymous with "catamite" or "Ganymede." Used in North America since the seventeenth century, the term was not generally adopted until the nineteenth century, and only then by American anthropologists. The dissimulation effected by the use of all these different terms was so great that observers like Coreal arrived in the New World expecting to tine! two distinct classes of persons (i.e., Hermaphrodites and garcons effemines).

Although the principle of cultural relativity has been central to twentieth-century anthropology, its application to differences in gender and sexuality has been slow. Perhaps this is because most discourse on sexuality and gender in Euro-American societies during this period has been dominated by psychology and sexology. Perceiving the relativity of sexuality and gender patterns requires the simultaneous perception of the cultural basis of the knowledge produced by these disciplines. Not recognizing the importance of culture in constructing the desires, roles, identities and practices that constitute gender and sexuality, anthropologists and other observers have paid little attention to local beliefs, focusing instead on a much grander story, one that holds enduring fascination for the Western imagination - how culture confronts nature (and the individual confronts society) and all the possible outcomes that these livens can produce.

Above all, it took the emergence of feminist theory and its critique of biological determinism to make a serious reevaluation of the berdache role possible. This can be traced back to the work of Elsie Clews Parsons and Ruth Benedict, whose insightful, if brief, discussions of berdaches in the early twentieth century were informed by a feminist understanding of the social construction of gender roles. Between the 1920s and the 1960s, a similar perspective can be traced in references to berdaches by Ruth Landes, Ruth Underhill, Gladys Reichard, Nancy Lurie, Omer Stewart, Harry Hay and Sue-Ellen Jacobs.8 A less direct but just as significant influence has come from the field of literary criticism and the methods of discourse analysis. The degree to which poststructuralist theory has sensitized scholars to the relativity of the categories and taxonomies they use cannot be underestimated. In the field of anthropology, analyzing the "rules of discourse" that shape the texts readers rely on, whether anthropological, historical, literary or native, has become a key tool of cultural analysis.

In the 1970s, these intellectual developments combined with a social climate in which gender and sexual differences had become topics of broad public interest to produce a fluorescence in berdache studies. In 1976, Jonathan N. Katz published a collection of original texts on berdaches in his popular book, Gay American History, which Harriet Whitehead cited extensively in her 1981 essay, "The Bow and the Burden Strap:' This was followed by a major article by Charles Callender and Lee Kochems in 1983 based on a thorough review of the anthropological literature; articles on female berdaches in 1983 and 1984 by Beatrice Medicine and Evelyn Blackwood, respectively; a series of articles by me; and two major book-length studies based on fieldwork and historical research by Walter Williams and me.

As a result of these diverse contributions, a consensus on several points has begun to develop. The key features of male and female berdache roles were, in order of importance, productive specialization (crafts and domestic work for male berdaches and warfare, hunting and leadership roles in the case of female berdaches), supernatural sanction (in the form of an authorization and/or bestowal of powers from extrasocietal sources) and gender variation (in relation to normative cultural expectations for male and female genders). In the case of gender variation, crossdressing was the most common and visible marker, but it has proven a more variable and less reliable indicator of berdache status than previously assumed. As the passage from Coreal shows, in some tribes male berdaches dressed distinctly from both men and women. In other cases, berdaches did not cross-dress at all, or only partly. In the case of female berdaches, cross-dressing was even more variable. Often, female berdaches wore men's clothes only when hunting or participating in warfare.

The sexual behavior of male and female berdaches was also variable. Where data exist, they indicate that the partners of berdaches were usually nonberdache members of the same sex - that is, berdaches were homosexual, if we define that term narrowly in terms of behavior and anatomy. Some berdaches, however, appear to have been bisexual and heterosexual. This was most often the case when adult men entered berdache status primarily on the basis of visions or dreams (see discussion below). Berdaches participated in both casual encounters (reported for male berdaches) and long-term relationships (reported for both male and female berdaches). Unfortunately, little is known about what social norms and kinship rules might have governed these relationships. At least some reports suggest that berdaches observed the same incest regulations as other members of their tribes.

In sum, the most reliable indicators of berdache status were its economic and religious attributes and not gender or sexual difference alone. Further, the variation of berdaches in terms of occupational and religious pursuits surpassed rather than fell short of social norms. Again and again one finds berdaches attributed with exceptional productivity, talent and originality. The careers of such male berdaches as We'wha (Zuni), Osh-Tisch (or "Finds-Them-and-KillsThem") (Crow) (see figure 7.2) and I lastiin Klah (Navajo) (see figure 7.3), and female berdaches like Qanqon (Kutenai), Running Eagle (Blackfeet) and Woman Chief (Crow), suggest that this reputation was often deserved.

A second point of agreement is that berdaches were accepted and Integrated members of their communities, as their economic and religious reputations indeed suggest. In many cases, berdaches enjoyed special respect and honors. In a few cases they were feared because of the supernatural power they were believed to possess. If berdaches were scorned, hated or ridiculed by their tribespeople, however, it was likely for individual reasons and not a function of their status as berdaches. In yet other cases, Indian joking relationships have been mistakenly interpreted as evidence of nonacceptance. In fact, in many tribes, individuals were subjected to teasing precisely because they enjoyed high status or prestige's Finally, many reports attributing American natives with hostility toward berdaches have been shown to reflect the Euro-American author's values and not native judgments. Indeed, what is missing at this point is an analysis of a confirmed case of a tribe lacking such a role or genuinely hostile to it.

A third area of consensus involves the abandonment of deterministic hypotheses concerning the "cause" of berdache behavior. Viewing berdaches as wholly determined products of social forces has a long history. The very language used to describe them has foreclosed the possibility very their agency. This is typically done through the repetitive use of passive, transitive and reflexive grammatical constructions. The passage from Coreal illustrates all three: "the boys that abandon themselves thus are excluded from the society of men .... They employ them in all the diverse handi works of women" (emphasis mine). Such practices predetermine and overdetermine berdaches as the objects of action, never the< subjects. The anthropological version of these tropes takes the form of etiological theories that account for berdaches in terms of external forces alone - for example, the suggestion that the berdache role was a social status imposed on men too weak or cowardly to measure up to stringent tribal standards of masculinity. This suggestion has been convincingly disproved by evidence of males uninterested or unsuccessful in warfare who, nonethe less, do not become berdaches and by the actual participation of berdaches in warfare. Indeed, a good part of the prestige of ber daches was due to the belief that they enjoyed the same kind of supernatural sanction as successful hunters and warriors. Consequently, most recent work on berdaches acknowledges the role of individual motivations, desires and talents in determining who became a berdache. Berdaches are finally being recognized as historical subjects - individuals who actually desired to be berdaches because of the rewards that life-style offered.

A fourth area of emerging consensus addresses the problem of translation referred to above. Whereas berdaches have been traditionally conceptualized as crossing or exchanging genders, as the terms transvestite or transsexual imply (or exchanging object choice, as homosexual suggests), several investigators (including myself) have begun to argue that berdaches in fact occupied a third gender role, or, in the case of tribes with both male and female berdaches and distinct terms for each, third and fourth genders. A multiple-gender paradigm was first proposed by M. Kay Martin and Barbara Voorhies, whose 1975 book, Female of the Spedes, included a chapter titled "Supernumerary Sexes:' They noted that "physical sex differences need not necessarily be perceived as bipolar. It seems possible that human reproductive bisexuality establishes a minimal number of socially recognized physical sexes, but these need not be limited to two." In her 1983 commentary on Callender and Kochems, Jacobs referred to berdache status as a third gender, a characterization she considers more inductive than the Western paradigm of gender-crossing. The first definitive argument for a multiple-gender paradigm was put forward by Blackwood. who proposed the "rigorous identification and labeling of the berdache role as a separate gender." "The berdache gender...," she concluded, "is not a deviant role, nor a mixture of two genders, nor less a jumping from one gender to its opposite. Nor is it an alternative role behavior for nontraditional individuals who are still considered men or women. Rather, it comprises a separate gender within a multiple gender system."

Both positive and negative evidence supports the argument that berdache status constituted a culturally acknowledged gender category. On the one hand, it can easily be shown that a dualgender model fails to account for many of the behaviors and attributes reported for berdaches - for example, berdaches who did not cross-dress or attempt to mimic the behavior of the "opposite" sex or those who engaged in a combination of female, male and berdache-specific pursuits. On the other hand, the consistent use of distinct terms to refer to berdaches, a practice that prevented their conceptual assimilation to an "opposite" sex, is positive evidence that berdache status was viewed as a separate category. Such native terms have various translations, from the obvious "man-woman" (e.g., Shoshoni tanowaip) to "old womanold man" (e.g., Tewa kwido) to terms that bear no relation to the words for "man" or "woman" or simply cannot be etymologized (e.g., Zuni Ihamana).

In many tribes, the distinction of berdaches from men and women was reinforced by sartorial practices and the use of symbols, such as the distinct color of feathers worn by Floridian berdaches reported by Coreal. In other cases, as I have shown in The Zuni Man-Woman, the religious functions of berdaches and the life-cycle rites they underwent were specific to their status while paralleling the kind of functions and rites pertinent to men and women. Similarly, among such tribes as the Zunis, Navajos, Crows and others, myths accounting for the origin of berdachestatus placed that event in the same context in which male and female gender categories were defined (stating, in so many words, "when the spirit people made men and women, they also made berdaches"). Although the points made so far apply equally to male and female berdaches, it is clear that female roles were not simply mirror opposites of male berdache roles. Unfortunately, the study of female berdaches lags behind that of male berdaches, and several features of this status await clarification. Medicine concluded that "warrior women," like male berdaches, occupied "socially sanctioned role alternatives:" These were "normative statuses which permitted individuals to strive for self-actualization, excellence, and social recognition in areas outside their customary sex role assignments." Some researchers, however, have concluded that female berdache roles were less viable and female berdaches less tolerated than were their male counterparts, and others have argued that the term berdache should not be applied to women at all. Callender and Kochems found documentation of female berdaches in only thirty tribes. Whitehead concluded that "when women did the equivalent of what men did to become berdaches, nothing happened." On the other hand, Blackwood has argued that the female berdache role was socially and ontologically on par with male berdache status in the sense of being a distinct alternative identity. At Zuni, I found that the female berdache role was less visibly marked than the male role (i.e., there are no reports of cross-dressing by women) and may have been more variable from individual to individual, but linguistic and religious practices still countenanced a distinct status for women who combined male and female pursuits, as evidenced by the use of the same term, Ihamana, to refer to both male and female berdaches.

Where Do Berdaches Come From?: The Theoretical Challenge

Derived from the Latin genus - meaning "kind, sort, class"--"gender" has come to be used by researchers in several fields to distinguish socially constructed roles and cultural representations from biological sex. Indeed, throughout Western history, popular belief and official discourse alike have acknowledged the role of social learning in sex-specific behavior, but biological sex has always been considered both the point of origin and natural limit of sex roles. What we call gender, in this view, should conform to sex, a belief that is rationalized alternately on moral and naturalistic grounds. The study of non-Western cultures, however, reveals not only variability in the sociocultural features of sex roles but also, as I argue below, wide variation in beliefs concerning the body and what constitutes sex.

If gender can be multiple, and potentially autonomous from sex, it becomes crucial to clarify exactly what it denotes. (In fact, definitions of gender are rare in the literature of "gender studies.") For the purposes of cross-cultural analysis, therefore, I define gender as a multidimensional category of personhood encompassing a distinct pattern of social and cultural differences. Gender categories often draw on perceptions of anatomical and physiological differences between bodies, but these perceptions are always mediated by cultural. categories and meanings. Nor can we assume the relative importance of these perceptions in the overall definition of personhood in a given social context, or that these differences will be interpreted as dichotomous and fixed, or that they will be viewed as behavioral or social determinants (as opposed to, for example, a belief that behavior might determine anatomy). Gender categories are not only "models of difference (to borrow Clifford Geertz's terminology) but also "models for" difference. They convey gender-specific social expectations for behavior and temperament, sexuality, kinship and interpersonal roles, occupation, religious roles and other social patterns. Gender categories are "total social phenomena," in Marcel Mauss's terms; a wide range of institutions and beliefs find simultaneous expression through them, a characteristic that distinguishes gender from other social statuses. In terms of this definition, the presence of multiple genders does not require belief in the existence of three or more physical sexes but, minimally, a view of physical differences as unfixed, or insufficient on their own to establish gender, or simply less important than individual and social factors, such as occupational preference, behavior and temperament, religious experiences and so forth.

Since the work of Ruth Benedict and Margaret Mead, anthropological studies of sex roles have focused on the relationship between sex and gender - a relationship that has been described as both motivated and arbitrary. A multiple-gender paradigm, however, leads us to analyze the relationship between the body and sex as well. Although morphological differences in infants may motivate a marking process, in a multiple-gender paradigm the markers of sex are viewed as no less arbitrary than the sociocultural elaborations of sex in the form of gender identities and roles. North American data, for example, make it clear that not all cultures recognize the same anatomical markers and not all recognize anatomical markers as "natural" and, therefore, counterposed to a distinct domain of the "cultural."

In traditional Zuni belief, for example, a series of interventions were considered necessary to ensure that a child has a "sex" at all. This began before birth, when the parents made offerings at various shrines to influence the sex of the developing fetus. In fact, the infant's sex was still not fixed at the time of birth. If a woman took a nap during labor, for example, the Zunis believed the sex of her child might change. After birth, interventions intended to influence physical sex continued. The midwife massaged and manipulated the infant's face, nose, eyes and genitals. If the infant was male, she poured cold water over its penis to prevent overdevelopment. If the child was female, the midwife split a new gourd in half and rubbed it over the vulva to enlarge it. In this context, knowing the kind of genitals an individual possesses is less important than knowing how bodies are culturally constructed and what particular features and processes (physiological and/or social) are believed to endow them with sex.

Previous theoretical work on berdache roles has taken the correspondence between sex and gender for granted, and this has skewed the resulting interpretation of berdaches in both subtle and fundamental ways. This is perhaps best illustrated by Whitehead's 1981 essay, "The Bow and the Burden Strap:" As this essay remains the most sustained attempt to generate theory to account for berdache roles to date, it might be helpful to analyze its arguments and discursive strategies in some detail. Following Henry Angelino and Charles Shedd, Whitehead defines berdaches as "gender-crossers," "person(s) of one anatomic sex assuming part or most of the attire, occupation, and social - including marital - status, of the opposite sex." She links the presence of such roles to the widespread Native American belief in the power of dreams and visions. This "vision complex" had an implicit economic dimension in that random contact with the supernatural was believed to bestow skills and luck in activities that created wealth and prestige. Such beliefs served to rationalize differences in individual success within otherwise egalitarian communities. At the same time, the activities that generated prestige were also coded for gender, so that gender, like skills and good fortune, was viewed as somewhat random. It became a "status," in the sense of a rank or standing, something achieved rather than ascribed. The fact that Native American women had relatively equal access to this prestige system, through food and crafts production, childbearing and household administration, left open the possibility that some men might decide to pursue these activities as well. In fact, Whitehead credits berdaches with opportunistic motives: to gain the "material prosperity and cultural respect that accrued to the assiduous practitioner of female crafts:' The berdache quits "the battlefront ... of male prestige... to take up a respectable sort of lateral position - that of ultrasuccessful female." His reputation for excellence, therefore, reflects an ideology of masculine supremacy: of course, he's better, he's a man!

The images of the military and the marketplace in these passages - evoked by terms like "battlefront," "lateral position" and "material prosperity" - signal an intertextual moment, a point at which Whitehead's text links up with other, master discourses, in particular that representing the encounter between feminism and the Western intellectual tradition. In this context, berdaches become figures in a feminist just-so story in which male exploitation and female resistance are portrayed as primordial/universal events. As Whitehead writes, "The culturally dominant American Indian male was confronted with a substantial female elite," which he found profoundly "disturbing:' "It was into this unsettling breach that the berdache institution was hurled. In their social aspect, women were complimented by the berdache's imitation. In their anatomic aspect, they were subtly insulted by his vaunted superiority. Through him, ordinary men might reckon that they still held the advantage that was anatomically given and inalterable.

The real problem with Whitehead's argument, however, is not its discursive moves but its reliance on two analytical dichotomies that from a multiple-gender paradigm must be questioned. The first, as suggested above, is the sex/gender binary, which opposes biological-sex to sociocultural gender. Second is the opposition of economy and economic relations to ideology (the base/superstructure opposition of Marxist theory). Both dichotomies have an essentializing influence that undermines the attempt to understand berdache roles as cultural constructions.

As Whitehead argues, "A social gender dichotomy is present in all known societies in the sense that everywhere anatomic sexual differences observable at birth are used to start tracking. the newborn into one or the other of two social role complexes. This minimal pegging of social roles and relationships to observable anatomic sex differences is what creates what we call a 'gender' dichotomy in the first place." Callender and Kochems echo this when they state that gender "is less directly tied to this anatomical basis, although ultimately limited by it." Unpacking these formulations reveals two propositions: social gender is based on the "natural facts" of sex, and, since there are only two sexes, there are only two genders. It follows that, if an individual is not one, then she must be the other. The only variation possible is an exchange of one gender for its "opposite" or some form of gender-mixing; but there are no possible variations that cannot be defined by reference to male or female. It also follows that in such a system there can be only one sexual orientation, namely, heterosexual.

The assumptions of a dual-gender system have been criticized in recent years on both empirical and theoretical grounds. It may, indeed, be arguable that all societies have at least two genders and, as suggested above, that these two genders are linked to perceptions of physiological differences. What constitutes anatomical sex, however - which organs (or fluids or physiological processes) are considered the signs of maleness and femaleness - has been shown by scholars in several fields to be as much a social construction as what has been termed gender.

Deconstructing the sex/gender binary reveals a hierarchical relationship between the two terms. That is, anatomy has primacy over gender, and gender is not an ontologically distinct category but merely a reiteration of sex. This is apparent in Whitehead's comments on female berdaches. "For someone whose anatomic starting point was female," she argues, "the infusion of an official opposite sex component into her identity was by no means so easily effected," because, "throughout the continent, the anatomicphysiological component of gender was more significant in the case of the female than in the case of the male; and was thus less easily counter-balanced by the occupational component." But this raises the question: If gender differences are to be viewed as anchored to an "anatomic-physiological component," then on what grounds can we argue that gender roles are not, in fact, "natural" (i.e., mirroring and/or determined by biology)? And if we accept the contention that having a female body makes it more difficult to become a berdache, then have we not conceded that the difference that defines women also makes them inferior?

In sum, if berdaches are to be understood as simply exchanging one gender for another, then they can indeed be interpreted as upholding a heterosexist gender system. If they are to be understood as entering a distinct gender status, however, neither male nor female, then something more complex is occurring. A multiple-gender paradigm makes it possible to see berdache status not as a compromise between nature and culture or a niche to accommodate "natural" variation but as an integral and predictable element of certain sociocultural systems, not a contradiction in Native American beliefs but a status fully consistent with them.

The second analytical dichotomy underlying Whitehead's thesis is the opposition of economic relations to cultural belief systems. This is yet another binary in which the relationship between the terms is hierarchical - that is, beliefs and cultural forms are the product of economic relations. The limitations of a purely economic model become clearer if we test it against specific cases from North America.

Drawing on the hypothesis linking berdaches to prestige systems, for example, we might predict that as a given North American society increased its capacity to produce surpluses and support economic specialization (e.g., by adopting or intensifying horticultural production), both women and berdaches would take advantage of the new opportunities for acquiring prestige, and their status would increase accordingly. Expanding production would also support population increases and concentration and, therefore, the number of berdaches in a given community would increase as well. As this occurred, we might predict that berdaches would begin to function collectively, along the lines of a priesthood.

Evidence from North America does indeed suggest that berdaches enjoyed economic and religious prestige in sedentary, horticultural communities; and in some communities, it does appear that a berdache priesthood had developed. In Florida, for example, Spanish and French colonizers encountered large settlements of sedentary farmers, organized into castes and ruled by chieftains. As Coreal suggests, berdaches appear to have functioned collectively in carrying out their role of tending to the sick and injured and burying the dead. The famous illustration published by De Bry shows no less than six "hermaphrodites" two pairs carrying patients on stretchers and two individuals carrying full-grown men on their backs. The accompanying text speaks of them acting as a group. Other evidence comes from the earth-lodge villages of the northern Plains. Among the Hidatsa, according to Alfred Bowers, there were fifteen to twenty berdaches in a village. They were an "organized group," a "special class of religious leaders," who collectively fulfilled various ceremonial functions. Among the Crows, close relatives of the Hidatsa, who had abandoned settled, village life for nomadic buffalo hunting in the eighteenth century, berdaches pitched their tipis together and were recognized as a social group. Among the Cheyenne, who had also lived in horticultural villages before the contact period, berdaches functioned as a group in organizing the scalp dance.

At the same time, other North American horticultural societies appear to have lacked berdache roles altogether - in particular, the sedentary communities of Algonquian- and Iroquoian-speaking tribes of the Northeast and Atlantic Coast regions. Berdaches are well attested among Algonquian speakers west of the Appalachians. Marquette, for example, wrote of Illinois berdaches that "nothing can be decided without their advice." Certainly the prestige and economic status of women was high among both Northern Iroquois and Coastal Algonquians. They produced durable trade goods and played key roles in food production and tribal politics. Several historical. cases of women chiefs and warriors among New England Indians are known from the colonial period. In fact, it may be that a careful review of seventeenth and eighteenth-century sources will reveal the existence of berdaches in these eastern tribes. Nonetheless, a hypothesis based on economic factors alone would lead us to expect a prominence on the part of berdaches that would have been more generally noted. At the same time, when we examine the social organization of these groups we find little in terms of mode of production or prestige systems to distinguish them from comparable tribes with berdache roles.

In other cases, a prestige-based model fails to account for the presence of berdache roles where they do occur. According to Whitehead, North American societies lacked "full prestige differentiation," and to accommodate a fuller range of social specialization and individual variation, they simply elaborated gender into multiple categories. The berdache role was a "cultural compromise formation founded on an incipient, though never fully realized, collapse of the gender-stratification system." Presumably, societies with more elaborate prestige systems would not produce such a role.

Pacific Northwest Coast tribes, however, provide several examples of North American societies with elaborate systems of prestige differentiation - including social ranking into classes of nobility, commoners and slaves; ranking of lineages and tribes; inherited titles, names, property and leadership roles; elaborate displays of wealth and symbols; distinctions between prestigious and lowly occupations; and so forth. Contrary to Whitehead's hypothesis, male and female berdache roles have been documented in nearly half of the thirty-eve societies in the coastal region between northern California and southern Alaska.

The most complete picture comes from reports of Franz Boas and T.F. McIlwraith on the Bella Coola, which reveal well-developed male and female berdache roles with the same key features found elsewhere in North America. Entry into berdache status was based on preference for the work of the other gender, which was credited to the influence of Skheents, the supernatural patron of berdaches. In mythology, Skheents is the first picker of berries an important seasonal food item - and he guards a bevy of young maidens. He was also portrayed in masked dances. According to Mcllwraith, "His face is that of a woman, his voice that of a woman, but he has masculine characteristics as well."

Berdache status appears to have been well integrated into the Bella Coola system of social ranking. According to McIlwraith, the profession of first berry picker, the woman who had the right to decide when berry harvesting could begin, depended on the ownership of an appropriate ancestral name. Occasionally, a male berdache was given one of these names. Again, we see that prestige, even in a system of ranking and inherited privileges, is still gender based. At the same time, gender is not seen as dichotomous and fixed and, as a result, a third category becomes viable.

As this review has shown, efforts to develop theoretical models berdache roles can no longer rely on the analytical dichotomies of sex/gender and economy/ideology. New analytical tools are needed to explain not only the occurrence of berdache roles but their similarities and differences across diverse societies as well.

How to Become o Herdache: Toward a Unified Analysis

In an earlier essay, I argued for the employment of multidimensional models in analyzing social and cultural differences in sexuality and gender.58 l suggested that definitions based on single traits such as "gender identity" or "sexual object choice" be replaced with a multidimensional inventory of all the differences to be found in a given cultural context associated with a status such as "berdache" or "homosexual," whether in terms of social role, gender variation, economic specialization, religious roles, sexuality or subjectivity. A counterpart to multidimensional description and cultural translation is now needed at the level of cultural analysis. Jane Collier and Sylvia Yanagisako offer such an approach in their program for a "unified analysis" of gender and kinship. They point to a growing recognition among theorists that the social phenomena of gender and kinship are manifestations of the same sociocultural processes. What is required at this point, they argue, is a methodology capable of analyzing these larger processes without relying on such analytical dichotomies as sex/ gender, nature/culture or domestic/public. They propose a threepart analytical program for this purpose.

The first phase entails the cultural analysis of meaning. In this phase the objective is to explicate the cultural meanings people realize through their practice of social relationships. At this stage of analysis, the investigator needs to ask, "What are the socially meaningful categories people employ in specific contexts, and what symbols and meanings underlie them?" In terms of gender diversity, we would want to know what kind of beliefs are associated with and are necessary for the formulation of berdache gender categories. Were gender and sexuality viewed as natural, constructed, inborn or acquired traits? Was the berdache, therefore, an anomaly, a monster or a prodigy? How were berdaches conceptualized in terms of the categories and meanings associated with kinship, economics, politics, religion and other social systems?

The second phase of a unified analysis involves the construction of systemic models of inequality. This is accomplished by analyzing the structures that people create through their actions and tracing the "complex relationships between aspects of what using conventional analytical categories - we might call gender, kinship, economy, polity and religion:' Such ideal-typical models of how power and social difference are organized in various societies are particularly valuable for comparative purposes. In the case of berdache roles we would want to know: What was the position of the berdache as a producer and a consumer within the larger system for creating and circulating power? What avenues of economic specialization were available to the role? What was the position of berdaches in relation to the organization of other genders and the division of labor between them (with special attention to the economic standing of women)?

Being synchronic in nature, systemic models have a built-in bias toward the persistence and continuity of social orders. For this reason Collier and Yanagisako include historical analysis as the third element of their program, pointing out that the ideas and practices that seem to reinforce and reproduce each other from a systemic perspective can be seen to undermine and destabilize each other from a historical perspective. Historical analysis also leads to the consideration of individual factors in social developments -- the motivations, desires and self-generated meanings of the individ uals who participate in "events" and occupy social "roles" -- and to an analysis of the construction of subject positions.

In what follows, I hope to illustrate the utility of a unified analysis in generating a theoretical model of berdache roles that can help answer some of the key questions in berdache studies as illustrated by ethnographic data from the Pueblo, Navajo and Mohave Indians.

The Pueblos

As we have seen, the most widely shared features of berdache roles were in the areas of economic and religious specialization. Nonetheless, some key questions remain unanswered. What was the relative importance of economic and spiritual dimensions in determining whether an individual became a berdache? Was it possible to become a berdache based on spiritual qualifications alone? Is a "vision complex" the only kind of belief system in which alternative gender roles can arise? More generally, what is the relationship between division of labor and gender ideology?

Perhaps the clearest evidence of the degree to which beliefs functioned independently of economic factors were those cases of vision-induced entry into berdache status by adult men who had not previously manifested berdache traits. This occurred in visioncomplex cultures, especially among Plains tribes, where following supernatural instructions was considered mandatory. Indeed, even successful warriors, if visited with dreams or visions considered specific to berdaches, assumed berdache identity. In other words, even though the economic dimension of berdache roles was their most common feature, native beliefs concerning supernatural experience were sufficient to sanction entry into the status on their own. Future research should focus on the possibility that multiple gender statuses might exist in societies without any economic correlate or where such a correlate had lapsed.

This evidence leads us to ask whether a vision complex is the only kind of belief system in which berdache roles could occur. The Pueblo Indians, for example, in contrast to many Plains tribes, lack any manifestation of a vision complex, as Benedict pointed out in her classic study, Patterns of Culture. The cooperative values of Pueblo communities, arising from the collective nature of agricultural production and communal living, deemphasized all forms of individualism. Direct contact with the supernatural was not sought and not welcomed. As Benedict notes, "If a Zuni Indian has by chance a visual or auditory hallucination it is regarded as a sign of death. It is an experience to avoid, not one to seek by fasting." The very notion of individual contact with and use of supernatural power was suspect - this was the activity of witches. Instead, all dealings with the supernatural were invested in priesthoods and religious societies. Although Pueblo berdaches were often religious specialists and their supernatural counterparts were portrayed in ceremonies, if they were considered "holy" it was not because of their berdache status as such but because of their religious training, which required the mastery of complex oral literature and ceremonial procedures.

Despite the absence of a vision complex, berdaches have been documented in a majority of Pueblo communities. The economic basis of their status was similar to that of Plains berdaches. Pueblo women produced and distributed both food and durable goods, and these products were coded as being female. Specialization in these areas by males entailed no loss in social standing. Nonetheless, Pueblo and Plains belief systems were distinct in how they legitimated multiple genders. Among the Zunis, the berdache role was sanctioned not by individual contact with the supernatural but through tribal myths that relate the creation of berdache status as an autonomous cultural category, much as gender distinctions, kinship categories and other social statuses are accounted for.65 Unfortunately, not enough data have been collected on Pueblo beliefs concerning the origins of berdache inclination in individual cases. According to Jacobs, the Northern Tewa believe that the exposure of an infant's genitals to the light of the moon "cause" it to be a berdache. The moon, among Plains tribes, is often associated with berdache status, but, significantly, Tewa belief involves not a dream or vision of the moon but merely accidental exposure.

The examples of Plains and Pueblo societies provide evidence that berdache status can exist in conjunction with diverse subsistence patterns and belief systems. At the same time, despite these differences, neither Plains nor Pueblo economies produced significant inequalities of wealth, and both afforded a basis for berdache status in terms of economic specialization. Similarly, while Plains and Pueblo societies rationalized berdache roles in different ways, both groups shared a basically constructivist view of gender in that neither viewed gender as determined by sex or, for that matter, made a distinction between sex and gender.

We might go one step further and question the dichotomy between economic and spiritual domains altogether. As Collier and Yanagisako point out, social structures are realized through practices; they are part of one process. The practices related to the production and circulation of spiritual power create structures just as much as the production and exchange of material goods. The religious specialist transforms and redistributes "raw" spiritual energies to meet a variety of human needs. Spiritual power is constantly being transformed into material goods (and back again) through mechanisms such as feasting, fees for healing services, ceremonial giveaways and so forth. It can be accumulated and invested; it can be used to control other resources and bodies. In many cases it is possible to attach quantifiable values to spiritual power, depending on the importance and role attributed to it in the larger belief system.

Viewing "spiritual power" and "material goods" as differential products of a unified system, we can see that in both Plains and Pueblo societies "power" could be accrued through food production, crafts production and provision of religious services. The "mode of production" for spiritual power, however, differed. Among Plains societies, spiritual power was acquired through individual means, whereas among the Pueblos it was a strictly collective production. In sum, the Plains spiritual economy lent itself to the kind of development whereby berdaches could augment gender difference with spiritual power - both acquired in the same way. The Pueblo organization of spiritual power did not lend itself so readily to the linkage between berdache status and religious ability. But if the opportunities for accumulating spiritual power (as an individual) were fewer among the Pueblos, the economic opportunities for crafts specialization and food production were certainly just as great.

The Navajo Nadleehe

If one were to select the North American tribe in which berdaches enjoyed the highest status on the basis of prestige system, one would be much more likely to select the Pueblos than their neighbors, the Navajos. Although one of the largest and most successful tribes today, the Navajos have been characterized in the anthropological literature as nomadic foragers and hunters whose culture was only slightly modified by the adoption of farming from the Pueblos and sheepherding from the Spaniards. Such a social system would not seem to be one capable of supporting the economic and religious specialization associated with berdache status. Yet to judge from reports from the past one hundred years, Navajo nadleehe may have enjoyed the greatest economic opportunities, prestige and respect of any berdaches in North America.

As Willard Hill reported, Navajo families welcomed a nudleehe in their midst. Children with berdache tendencies were given special care and encouragement, and "as they grew older and assumed the character of nadle, this solicitude and respect increased, not only on the part of their families but from the community as a whole:' Berdaches were often given control of family property and acted as the head of the household, supervising both agricultural and domestic work. They wove, made pottery, tanned leather, made baskets and did other female crafts, and they often became religious specialists, a male role. Certain ceremonials required their participation. They also served as gobetweens in arranging affairs between women and men. In terms of sartorial practices, they might dress like men, women or neither. Unfortunately, little is known about Navajo female berdaches, except that the same term was applied to them as to male berdaches, and, according to Hill, they were equal in number.

The comments of Navajo elders recorded by Hill in the early 1930s reveal the high status nadleehe enjoyed: "If there were no nadle, the country would change. They are responsible for all the wealth in the country. If there were no more lest, the horses, sheep, and Navaho would all go. They are leaders just like President Roosevelt." " A nadle around the hogan will bring good luck and riches." "You must respect a nadle. They are, somehow, sacred and holy."

Gary Witherspoon's work on Navajo thought and language provides some suggestive insights into the Navajo conceptualization of third-gender status. All beings, including natural phenomena, were believed to have both an inner and outer form. Sometimes these forms were of different genders. Speech, for example, which is female, is the outer form of Thought, which is male. Nor are inner and outer forms necessarily fixed; an inner form may have many different outer forms. Thus, we find that the term nadleehe literally translates as "one who changes continuously," suggesting that berdaches were seen not as crossing genders or entering a distinct category but as fluctuating between outer and inner dimensions of male and female forms - a third process rather than a third category. This is consistent with Witherspoon's conclusion that the Navajos envision "a cosmos composed of processes and events, as opposed to a cosmos composed of facts and things."

A synchronic model of the Navajo socioeconomic system, however, would not account for the social and cultural elaboration of the berdache role in that tribe. This is a case in which historical analysis, the third element of Collier and Yanagisako's program, is especially valuable.

For this purpose, three phases in the Navajo history need to be distinguished. When the ancestors of the Navajos first arrived in the Four Corners area in late prehistoric times, they were huntergatherers, living in small nomadic bands. There is no reason to believe that a berdache role did not already exist among them, as it did among the historic Great Basin tribes, whose subsistence patterns were similar to those of the protohistoric Navajos. Even so, opportunities for economic specialization for either women or berdaches would have been limited. The encounter of the Navajos with the ancestors of the Pueblo Indians was a fateful one, however, for the Navajos acquired not only the means of growing their own food but new crafts and religious practices as well. Significantly, Navajo women seem to have been both the vehicles and beneficiaries of many of these developments. Farming and weaving, male pursuits among the Pueblos, became a joint activity of women and men in the first case and a strictly female art in the second. It was during this phase of Navajo history, before their first contact with Europeans, that a semisedentary lifestyle replaced earlier nomadic patterns.

The acquisition of the horse and sheep by the end of the seventeenth century marked the beginning of a second phase in the history of the Navajo economy. Sheep became the property of the matrilocal household and in many cases the outright property of women. This, together with their significant contributions to food production, meant that women's labor alone could provide families a basic subsistence. At the, same time, women also produced textiles, pottery, basketry and other valuable exchange goods. This freed men to engage in the lucrative if sometimes risky practice of raiding Pueblo and Hispanic villages, an extension of their traditional pattern of traveling in bands to collect resources and hunt. The potential now existed for surpluses and the accumulation of real wealth sufficient to support talented family members in various specializations, especially religious practice. In this period, the role of the medicine man appears to have grown accordingly in importance. But a key point is that the changes in subsistence patterns fostered by these historical events all tended to increase the social and economic autonomy of Navajo women, or, rather, of the group of women who constituted the matrilocal household and managed the family's affairs in the absence of men. At this point, the Navajos meet the conditions for the elaboration of alternative gender statuses predicted by a prestige-based hypothesis.

A third stage in Navajo history was ushered in by their defeat at the hands of the American military in 1863 and the onset of the reservation period. The changes that this entailed, however, tended to limit men's economic activities more than women's. Whereas Navajo men could no longer engage in raiding, most of the traditional economic pursuits of women remained open to them, including gardening, sheepherding and weaving. Gradually, Navajo men turned to large-scale sheepherding and the art of jewelry making. Our first documentation on Navajo berdaches dates from this period, and it reveals the extraordinarily high status they enjoyed.

The career of the famous Navajo nadleehe Hastiin Klah (1869-1939) provides a glimpse of how historical and structural factors interact in the formation of a status like that of the berdache (see figure 7.3).6 Born at the onset of the reservation period, Klah was nonetheless raised by his family as a traditional berdache. He became an accomplished weaver and medicine man - prestigious female and male activities, respectively. By combining these skills, he was able to create an entirely new artifact - large weavings depicting ceremonial designs. Before this time (c. 1920), Navajo weaving was strictly secular and purchased primarily for use as floor coverings. Traders paid weavers for their work by the pound. Klah's tapestries, on the other hand, were purchased by wealthy art collectors and museums, whose interest eventually extended to traditional weaving styles, so that what was once a "craft" became a "fine art."

At the same time, Klah's skills as a medicine man were in high demand, as the Navajos turned to traditional religion in response to the stresses of reservation life. With income from his weaving, his medicine practice and the herds of sheep he owned with his female relatives, Klah had the ability to accumulate significant wealth in Navajo terms - except that he chose to give most of it away and devote his attention to spiritual pursuits. In the 1920s and 1930s he collaborated with several scholars and researchers to record his ceremonial knowledge. He also traveled extensively in the white world, extending the traditional function of Navajo berdaches as cultural and social mediators into the realm of intercultural relations.

Klah was equally innovative when it came to Navajo religion. Gladys Reichard credited him with the creation of a systematic synthesis of Navajo philosophy out of what had been a diffuse set of beliefs and practices. He also appears to have elaborated the role of the supernatural berdache known as Begochidiin. Accounts of this figure are complex. Karl Luckert traces Begochidiin back to ancient preagricultural hunting patterns? He is the son of the sun and often linked with the moon. He is the patron god of hunters, but these associations are overlaid with features of a culturebearer. He is credited with providing the first seeds for agriculture and inventing pottery. In myths told by Klah, Begochidiin is especially prominent. Indeed, in Klah's rendering, he becomes a transcendental figure who bridges not only gender and economic differences (hunting and farming) but age distinctions and racial differences as well (Klah described him as fair-haired with blue eyes).

In sum, Klah flourished during the early reservation period because rapid changes in Navajo subsistence patterns provided unique social, economic and intellectual opportunities for individuals with the skills traditionally associated with his role. Being ambitious and talented, he took advantage of these opportunities and was undoubtedly not the only berdache to have done so. In the process, he contributed to the further elaboration of berdache status and an extension of its vitality into the twentieth century."

A synchronic analysis of Navajo social structure alone would not account for the high status of berdaches like Klah. The case of the Navajo nadleehe reveals how historical factors can be crucial in accounting for the development of multiple genders. In other cases, historical study may provide answers concerning the absence of berdaches in tribes in which we might expect to find them. In sum, a unified analysis that includes a historical dimension provides the best approach to accounting for the formation of gender roles and identities.

The Mohave Alyha and Hwame

As long as berdaches are culturally recognised as berdaches (i.e., consistently referred to with a distinct term) and individuals and communities do not engage in a social fiction concerning their anatomy (by suppressing or "forgetting" the individual's actual anatomy or pretending that it had somehow been changed), crossdressing alone is not necessarily indicative of a gender-crossing pattern. The sartorial practices of both male and female berdaches have been shown to be far more variable than previously assumed. Clothing and ornament in most North American societies constituted a semiotic system for signaling not merely gender but social standing, kinship status, religious status, personal accomplishments, age and so forth. Cross-dressing itself often occurred in ritual and mythological contexts with little or no reference to berdache status. However, if male berdaches not only wore women's clothing but imitated women's reproductive processes and female berdaches did the reverse, then a sex/gender belief system would appear to be operative, with "berdaches" behaving according to the logic of dual and dichotomous sexes. Such an example appears to be provided by the Mohaves of the Colorado River area.

This case is worth examining because it seems to provide the strongest evidence against a multiple-gender paradigm of berdache roles and because the Mohave example is frequently cited in the secondary literature as illustrating North American berdaches in general. This is due largely to the vivid ethnographic account of Mohave male berdaches, or alyha, and female berdaches, or hwame, by the psychoanalytic anthropologist George Devereux in the 1930s.80 Although the earliest Spanish explorers noted the presence of berdaches in this area, we are largely dependent on Devereux's report for our knowledge of them.

According to Devereux, Mohave berdaches consistently, indeed rigorously, behaved according to the precepts of a cross-gender model - as individuals of one anatomical sex striving to become the "opposite sex:" Alyha insisted on being referred to by female names and with female gender references. They only practiced receptive anal and oral intercourse, and although they appeared to have achieved orgasm, they discouraged personal contact and even reference to their male genitals, using the Mohave word for clitoris to refer to their penises, the term for labia majors to describe their testes and the word for vagina to refer to their anuses. A special ceremony served to confirm male berdache status, during which clothes of the "opposite" sex were made and presented to them. They subsequently received female facial tatoos. Both alyha and hwame might enjoy casual or long-term relationships with nonberdache men and women. If in a partnership, according to Devereux, alyha. and hwame were consistently referred to as a "wife" or "husband," respectively.

What has earned the alyha, in particular, a permanent place in the ethnographic literature is Devereux's account of their elaborate mock pregnancies. These were carried out in excruciating detail, including the simulation of pregnancy through self-induced constipation followed by the "birth" of a stillborn fecal fetus. The whole production culminated with the burial of the "fetus" and the observance of the appropriate mourning rites, in which the alyharequired her husband to participate. Alyha were also reported to simulate menstruation by scratching themselves until they bled.

Devereux's data on female berdaches are less consistent. Hwame. are not said to have employed male physiological terminology to refer to their genitals, and one informant told Devereux that hwame did not necessarily take male names. Some women became hwame after having children. They ignored their own menses but followed the taboos required of husbands when their wives menstruated or were pregnant. They did not necessarily cross-dress. Sex between a hwame. and a woman was performed in a variety of positions, without distinct active or passive role-playing. Like male berdaches, hwame were often shamans.

Although Devereux's report provides convincing evidence of cross-gender beliefs and practices on the part of Mohave male berdaches, it poses some difficulties when used to make generalizations. The extreme gynemimetic behavior that Devereux attributes to the alyha is unique in North America. There are one or two reports of berdaches taking measures to hide their male genitalia when appearing naked before women, but nothing remotely similar to the fake pregnancies of the alyha has been reported elsewhere, not even among linguistically related neighboring tribes. The challenge, therefore, is to account for this apparent discrepancy, and here, again, a unified analysis provides an effective approach, leading us to consider, in turn, cultural meanings, socioeconomic structures and historical factors.

The first test of whether Mohave berdaches represent a case of culturally patterned gender-crossing would be to determine whether the Mohaves believed gender to be dichotomous and fixed and whether, therefore, a third position was conceptually impossible. In fact, a close examination of evidence reported by Devereux and A.L. Kroeber reveals that the social labeling and conceptual patterns of the Mohave are more compatible with a multiple-gender paradigm than a cross-gender model. The use of distinct terms for male and female berdaches, for example, is not consistent with the maintenance of a social fiction of gendercrossing and transformation. In fact, although Devereux states that Mohaves consistently used "he" and "she" in referring to hwame and alyha, respectively, he later presents extensive quoted material from an informant who repeatedly uses "he" in referring to alylia. Perhaps the Mohaves' cross-gender references to berdaches were less literal than Devereux understood them. They may have been meant in the same sense conveyed when the pronouns "he" or "she" are placed in quotation marks in written texts.

Similarly, as the Mohave "emphatically stated" to Devereux, the purpose of the initiation ceremony for male berdaches was not to effect a transformation in their "personal habits" but merely to acknowledge them. This ceremony follows the common pattern of rites of passage as outlined by Victor Turner, with the phases of separation, liminality, and incorporation (the boy spends a night covered up while shamans sing over him; the next day he is led in a dance to the river where he publicly strips and enters the water: after four days his face is painted and he reenters the community as an alyha.). The passage, however, is not from male to female, but from boy to alyha, a transition of both gender status and age.

The full-length version of the origin myth of the alyha, originally told to Korea in 1902 and summarized in his Handbook of California Indians (published in full in 1972), provides valuable evidence on this points.85 The account begins in the house of the god Mastamho, where four women are seated in a directional circuit around him. Mastamho proceeds to assign each a particular identity or function, creating, in effect, female gender subcategories: one is to be a shaman (and not marry), one is to be a "loose" woman, the third is a mother about to give birth and the fourth woman is a midwife. With instructions from Mastamho, the infant is delivered. As the narrator relates:

Now the baby lay there, looking around. "Sit back from him there," Mastamho said, "That boy knows much more than you all: he will be a leader:' The baby was looking this way and that, its eyes winking. Then it said: "1 want a name. What will you call met" Mastamho said: "He is a boy, but 1 think we will give him not a boy's name, nor a man's but a girl's. I call him Hatshinyo-hai-kwatsh'iae."
Mastamho picked up the baby, held it in his hands, "I will tell you all about him. I want you to learn what I will teach about this child." Then he sang, swaying his hands from side to side with the child on them, and the four women danced to his motions.
When he laid the child down, the boy thought: "I am a boy; but shall I wear a breech-clout or not? Shall 1 wear girl's clothes or boy's?"

The myth foregrounds the capriciousness of gender assignments.87 In the next passage, Mastamho initiates a series of tests that reveal the boy's preference for girls' toys. Acknowledging these preferences, Mastamho instructs the boy in the skills of women's games and dancing, placing him in charge of both. He provides the boy with a dress and gives him a new (female) name.

Throughout the account, the boy is referred to as "he," and while he is instructed in certain manners and skills of women, -he is also described as having traits unique to his status (e.g., leadership, skill in gambling and ability to cure venereal disease). There is no suggestion of anatomical change or the imitation of female reproductive functions. In fact, neither this account nor the briefer ones cited by Devereux provide mythological precedence for gynemimetic behavior. Mohave mythology presents alyha. status as a distinct and autonomous category of personhood, on par with other gender-based statuses such as the shaman, the chief, the midwife, the sexually active woman and so forth.

The account of the al alyha initiation ceremony collected by Kroeber is of particular interest because of the description of the< face painting applied to the alyha at the conclusion of the cere mony. Mohave men and women painted their faces with distinct male and female styles. The face painting described by Kroeber's< informant consists of "a vertical stripe down from each eye and another down the nose to the mouth." This design is illustrated in Kroeber's Handbook and in an article by Edith S. Taylor and William J. Wallace as a male style. Similarly, although male berdaches assumed female personal names, they did not assume the clan name borne by all the women within a lineage.

In fact, various comments recorded by Devereux suggest that, in the minds of most Mohave, alyha. and hwame. retained qualities of their "original" gender and combined them with those of the "opposite" gender, and for this reason they were always thought of as distinct from both men and women. "You can tease an hwame," one Mohave told Devereux, "because she is just a woman, but if you tease an alyha, who has the strength of a man, he will run after you and beat you up." Alyha were not courted like ordinary girls; nor were they viewed as equivalent to female wives. "He must be awfully hard up to marry a womanly man," was a typical comment concerning a man who chose to marry an alyha. According to Devereux, many Mohave wondered whether the husbands of alyha. "really thought they were having intercourse with a woman" - in other words, they knew that the alyha was "really" a man; surely the husband of the alyha. knew as well.

In explaining the development of berdache orientation in individual cases, Mohaves credited a combination of predestination, occupational preferences, social influences and, above all, dreams. As a contemporary Mohave explains, "Dreaming was the very core of Mohave life. It was the source of each individual's special skill, of his prowess as a warrior, and of his success in all his undertakings. A Great Dream might foretell the birth of a male child to his father. Later the dream would return to the child as he grew to manhood and listened to others tell of their dreams. Then he dreamed his own and became a man."91 In other words, dreaming was key to acquiring gender identity, whether male, female or berdache. Although all dreams were believed to have originally occurred in the mother's womb, the dreams of alyha and hwame. were not apparent until, as children, they began to express preferences for particular work activities. Devereux was told, "When there is a desire in a child's heart to become a transvestite (sic) that child will act different. It will let people become aware of that desire. They may insist on giving the child the toys and garments of its true sex, but the child will throw them away." Although berdache identity is presented as predetermined and involuntary, this passage clearly points to a psychic and not a physiological point of origin.

In sum, Mohave beliefs combined the two modes of rationalizing berdache status described earlier: supernatural sanction similar to the Plains pattern end mythological precedence, as in the case of the Zunis and Navajos. Both rationales are more consistent with a multiple-gender paradigm than with the gender-crossing model.

If Mohave work activities were not gender coded, the markers of berdache status might be expected to shift to other areas, such as anatomy, but there is no evidence for this. We find the same correlates of berdache status, in terms of social, economic and religious specialization, in Mohave culture as in other North American groups. Overall, the division of agricultural and domestic work appears to have been fairly informal (as it was among the Pueblos and Navajos). Wives and husbands worked together in their fields, and, according to Devereux, alyha were not the only males to perform housework: tribal heroes who take care of ailing families and middle-aged men who marry very young girls are also described as performing domestic tasks. However, prestige was still gender based. For men, it was achieved through military exploits, leadership roles and religious specialization; for women, through food and crafts production and female forms of shamanism.

In sum, neither the belief system of the Mohave nor their socioeconomic patterns provide an explanation for the crossgender behavior attributed to Mohave berdaches. The expectations and beliefs are those found in a multiple-gender paradigm. Although some berdaches may have insisted on a fiction of gender-crossing, the community as a whole did not go so far. This leaves the consideration of historical factors - or, more broadly stated, the possibility of nonstructural and nonpatterned sources for this behavior.

We might begin with a reexamination of Devereux's report. Gilbert Herdt has pointed out the extent to which Devereux's account of Mohave berdaches conformed to the Freudian theory of homosexuality as a phenomenon of sexual inversion.95 This tended to lump together individuals now distinguished by such terms as "gay," "lesbian," "transvestite" and "transsexual:' In fact, it would be tempting to define Mohave berdaches as an instance of transsexualism, but this would be misleading. The goal of modern transsexuals has been to appear so convincing as members of their chosen sex that others never suspect they had ever been anything but that sex .% As we have seen, the practice of holding public ceremonies to confirm alyha status made any such fiction impossible among the Mohave. Further, the alyha described by Devereux appear to have aspired specifically to the acquisition of female reproductive functions. Western male-to-female transsexuals, on the other hand, tend to be preoccupied with the inappropriateness of their male genitalia in relation to their gender identification. They aspire more to female morphology than reproductive functions, with the acquisition of breasts taking priority over a vagina. If alyha. were identically motivated, we might expect not elaborate fake pregnancies and simulated menstruation but attempts to castrate themselves and/or enlarge their breasts (both of which are possible with methods available in a preindustrial society).

Although there is no reason to doubt the accuracy of his reporting, Devereux himself made no direct observations of berdaches, relying instead on the memories of informants recalling events of the late nineteenth century and often relying on secondhand information. Although referring to the existence of other informants, Devereux cites only three specific sources for his data on berdaches - Nahwera, reputedly the last Mohave who knew the alyha initiation songs (but not a berdache himself); an eightyyear-old woman who had heard about (but not seen) the alyha ceremony when in her youth; and Hivsu Tupoma, a male shaman who had known Kuwal, a Mohave who, in the late nineteenth century, had had more than one alyha wife. It appears that most of Devereux's information concerning berdache pregnancies was provided by Hivsu Tupoma, based on stories he heard from Kuwal. Of course, it must be kept in mind that in a small community the behavior of two or three berdaches might constitute "tradition" for a given generation.

It seems possible, therefore, that the behavior Devereux described had its origins in individual factors more than in cultural expectations and may have been specific to the individuals known to (or heard of by) Devereux's informants, in particular Kuwal's wives and the dynamics of Kuwal's relationships with them. This leads us to ask which motives besides the desire to cross genders might underlie the imitation of pregnancy by Mohave berdaches. A comment Devereux recorded provides one clue: "Some men who had enough of it [marriage to an alyha] tried to get rid of them politely, alleging barrenness of the alyha. But no alyha would admit such a thing. They would begin to fake pregnancy."97 In other words, faking pregnancy may have been the somewhat desperate strategem of an individual threatened with the loss of a lover, not unlike cases of hysterical pregnancy in females. Getting the husband to participate in mourning rites for the stillborn "infant" (and burial rites were serious business among the Mohave) amounted to his capitulation to the fantasy and, therefore, a victory for the alyha.

In fact, the culturally patterned dimensions of the fake pregnancies attributed to Mohave berdaches can be discerned, but they are related less to beliefs concerning gender than to expectations concerning the behavior of shamans and a general anal preoccupation in Mohave culture. As Devereux reports elsewhere, any unusual manipulation of the genitals by children, such as hiding the penis between the legs or holding it backward to urinate like a mare, was considered typical behavior of future shamans. Similarly, the fecal fetus of the alyha is consistent with a pattern of anality evidenced in Mohave mythology, naming practices and individual fantasies. The multilayered significance of the alyha's offspring becomes apparent when read in light of the mythological account of how the daughter of the culture-hero Matavilye is impregnated after swallowing her father's feces, an act that also causes his death. All this suggests that the mimetic pregnancy of the alyha was a symbolic, not a literal, enactment - a ritualistic manipulation of "dirt" (in Mary Douglas's sense of pollution) to mediate life and death - and therefore on an entirely different order than the behavior of modern transsexuals. There was no real possibility of "passing" as a woman in traditional Mohave society (after all, Mohave women were traditionally bare-breasted most of the year), but through magic one might at least approach the mystery of the creation of life.

Yet another line of investigation is suggested by a brief passage in Devereux's original article. Reporting on the contemporary (i.e., 1930s) status of homosexuality among the Mohave, Devereux describes three men "accused of" "active and passive homosexuality" - none of whom cross-dressed or had undergone the alyha initiation. All three men lived together, and two were half-brothers. According to Devereux, "They are usually referred to as each other's wives and are said to indulge in rectal intercourse."101 What is striking here is the indefiniteness of the role attributions. Devereux invokes the distinction between active and passive homosexuality, but he fails to indicate who is active and who is passive. In the rest of his article, these distinctions are crucial. One of these men would have to be a husband and another a "wife," and the "wife" would be expected to vehemently insist on the distinction.

Certainly traditional practices were lapsing by the 1930s, which would account for these nonberdache forms of homosexuality, but another possibility worth considering is that not only casual but committed .homosexual relationships such as these were a viable option in traditional Mohave culture - alongside the option of being a berdache or a partner of a berdache. Because such a possibility does not fit Devereux's preconceived theory of "homosexuality" as sexual inversion, he does not explore it (although he does report that casual homosexual relations were "frequent" in traditional times). This would indeed distinguish the Mohave from other tribes, where the best evidence at present suggests that committed, sexual partnerships and cohabitation between members of the same gender were rare. If this were an option in Mohave culture, the only motivation for entering the berdache role would be to express a strong sense of gender difference.

As this analysis suggests, Mohave beliefs and practices are more complex than a gender-crossing model would predict. A review of Devereux's original report reveals how Western assumptions concerning gender and sexuality can powerfully shape ethnographic observations and even lead the ethnographer to overlook the presence of social patterns - in this case, the possibility of nonberdache homosexuality. The Mohave case also underscores the importance of allowing for the divergence of individual meanings and motivations from normative belief's. In small-scale societies, idiosyncratic behavior can too easily be mistaken for a "traditional" cultural practice.

Conclusion

Berdache status was not a niche for occasional (and presumably "natural") variation in sexuality and gender, nor was it an accidental by-product of unresolved social contradictions. In the native view, berdaches occupied a distinct and autonomous social status on par with the status of men and women. Like male and female genders, the berdache gender entailed a pattern of differences encompassing behavior, temperament, social and economic roles and religious specialization - all the dimensions of a gender category, as I defined that term earlier, with the exception of the attribution of physical differences (the Navajos may be one exception; see n.74). But physical differences were constructed in various ways in Native American perception, and they were not accorded the same weight that they are in Western belief. Social learning and personal experiences (including ritual and supernat ural experiences) were considered just as important in defining individual social identity as anatomy. Viewing female and male berdache roles as third and fourth genders, therefore, offers the best translation of native categories and the best fit with the range of behaviors and social traits reported for berdaches. Conversely, characterizations of berdaches as crossing genders or mixing genders, as men or women who "assume the role of the `opposite' sex," are reductionist and inaccurate.

Given the presence of multiple genders, what are their social and cultural correlates) The three cases discussed here suggest that most of the variations in the berdache role among different North American societies were related to cultural systems of meaning and historical factors more than differences in prestige systems. Despite a wide variety of subsistence patterns, North American modes of production and division of labor did not in most cases produce significant or fixed differences in wealth and status. At the same time, they also afforded an economic dimension to berdache status in terms of productive specialization. Even so, economic potential alone does not predict the presence of multiple genders. Whereas sedentary horticultural communities may have provided more opportunities for specialization, it was among the Navajos, for historical reasons, that the berdache achieved highest status. Similarly, in terms of belief systems, although a vision complex can serve to rationalize alternative gender statuses and foster entry into the status by individuals who do not manifest the typical traits of berdaches, berdache roles can flourish within cultural systems lacking a vision complex altogether. Finally, as in the Mohave case, even with economic opportunities and beliefs similar to other North American groups, individual motivations, both psychological and situational, could powerfully shape the construction of what otherwise appear to be "traditional" features of social roles and the meanings surrounding them.

There are no definitive variables for predicting the presence of multiple genders, but I believe we can specify a set of minimal conditions for the possibility of such statuses. First is a division of labor and prestige system organized in term's of gender categories, so that the potential exists for female specialization in production and distribution of food or exchange goods. Second is a belief system in which gender is not viewed as determined by anatomical sex or in which anatomical sex is believed to be unstable, fluid and nondichotomous, and, therefore, an autonomous third category is viable. Third are the occurrence of historical events and individuals motivated to take advantage of them in creating and shaping gender identities. If these conditions are present, then multiple gender roles can develop - and it becomes possible to become a berdache. Conversely, I would hypothesize that, for a given society in which multiple genders were present, it would take not only the elimination of the economic dimension of such statuses but a lapse in the belief systems rationalizing them and the introduction of a dual-sex ideology to effect a full collapse of such roles.

The next step in berdache studies will be the recognition that gender diversity is not an isolated feature of North American societies but a worldwide phenomenon, represented in most culture areas as well as in certain historical periods of Western societies. Gender diversity will become one more part of the story of human culture and history that is anthropology's job to tell.