Main Page


Theorizing the Third

or How I Became a Queen in the Empire of Gender



Related Links

"The concept of multiple genders is a useful one, theoretically and politically. The potential of assimilation or co-optation of such a direct challenge to heterosexist binarism seems to me much less than that of an identity based on a difference that exists only in the fleeting moments of the bedroom...."

This paper was originally presented at the “Lesbian and Gay History: Defining a Field” conference at the Center for Lesbian and Gay Studies, City University of New York, October 7, 1995.

I became interested in the possibilities of multiple genders as a result of my research on Native American two-spirits or berdaches. The first problem I encountered when I began this research was that much of the evidence I found didn’t fit the standard anthropological definition of berdaches, which explained these diverse tribal roles as instances of “a person of one anatomic sex assuming part or most of the attire, occupation, and social—including marital—status, of the opposite sex” (Whitehead, 85). I was learning about two-spirits who did not cross-dress, or who dressed in styles distinct from both women and men, or who cross-dressed but made no attempt to “pass” or disguise their original sex. In many cases they engaged in behaviors and activities of both their anatomical sex and those of the so-called opposite sex, and many of their behaviors and activities were unique to their role, especially their ritual and spiritual lives. Finally, there were myths that accounted for the origins of two-spirit roles much as male and female genders were explained. In short, many elements of these roles were inconsistent with the idea that these were persons of one sex trying to be the opposite sex.

When The Zuni Man-Woman was published in 1991 it was the first time a Native American berdache role had been comprehensively examined using a multiple gender paradigm. I was not the first, however, to recognize the role as a distinct gender status. In the 1970s, feminist anthropologists Martin and Voorhies described such roles as “supernumerary sexes,” and in 1983 Jacobs used the term “third gender.” In 1986, Williams referred to the role as the “berdache gender,” among other things. It was a paper by Evelyn Blackwood in 1987, however, later published in the Journal of Homosexuality, that convinced me to adopt a multiple gender paradigm. Blackwood argued that such a paradigm was closest to the actual native conceptualization of two-spirits, and it offered the best fit with the data. Given this, she called on researchers to abandon Western terminology and models as misleading and inaccurate, and to adopt “a more rigorous identification and labeling of the berdache role as a separate gender” (179).

Following Blackwood's advice, I began asking to what extent a third gender paradigm helped to explain my Zuni data. Rather than having to dismiss or treat as exceptions evidence that did not conform to a gender-crossing model, I found that a third gender paradigm made it possible to see all the elements of the berdache role as part of a coherent pattern. Male and female two-spirits at Zuni and elsewhere did not behave like either men or women, nor were they labeled or thought of as such, but rather as lhamana, winkte, bote, tubas', and so forth. In The Zuni Man-Woman I made the argument for a third gender model in terms of a specific case, which allowed me the luxury of postponing some of the larger issues raised by this concept. On what basis, for example, are we justified in calling a social status or identity a third gender? Answering this question requires that we define gender first, something that is not often done in the literature. Here I realized, with help from feminist and deconstructionist theory, that part of the problem is the sex/gender binary itself—an analytical tool that has, until now, served us quite well. But as I analyzed the deployment of this binary in a variety of texts it became apparent to me that as long as we continue to anchor gender in physical sex, gender becomes merely another version of sex. It collapses back into the transcendental signified of a precultural state of nature. The apparent symmetry of the binary, as Derrida would show in a much more complicated way, masks a hierarchical relationship between the terms.



[Enki] opened his mouth to speak, saying to the Lady of Birth, the Mother-womb
"O Lady of Birth, Creatress of the Fates
... let there be a third category among the people
(Let there be) among the people bearing women and barren women ...
They shall indeed be tabooed, and thus cut off from child-bearing.

Akkadian myth of Atrahasis, ca. 1700 b.c.e., 3.7.1ff.

The [clay] she [Ninmah] made into a woman who cannot give birth
Enki, upon seeing the woman who cannot give birth
Decreed her fate, destined her to be stationed in the "woman house."

The [clay] she made into one who has no male organ, who has no female organ
Enki, upon seeing him who has no male organ, who has no female organ
To stand before the king, decreed as his fate

Sumerian, Creation of Man, ca. 2000 B.C.E.


India and Southeast Asia

Kliba. Impotent, emasculated, a eunuch;...the neuter gender.

Oxford Sanskrit Dictionary, s.v.

Tritiya prakrita. The third nature, a eunuch.

Oxford Sanskrit Dictionary, s.v. (appears in Mahabharata ca. 200 B.C.E.-200 C.E., 4.59)

On certain rites requiring a kliba: "And as to why it is of a long-haired man,—such a long-haired man is neither woman nor man; for being a male, he is not a woman, and being long-haired (a eunuch), he is not a man."

Satapatha Brahmana ca. 800-500 B.C.E.,

Napumsaka (lit. "not-man"): third type (neither male nor female).

Natyasastra ca. 200 B.C.E.-200 C.E., 24.68-69

"We are neither men nor women."

Contemporary hijra, in Nanda, p. 15

Nang Itthang Gaiya Sangkasi (f.) is born from the earth and Pu Sangaiya Sangkasi (m.) is born from fire. From the four elements they conjure three sexes: female, male, and hermaphrodite (i.e., napumsaka).

Pathamamulamuli, Yuan manuscript ca. early 16th C.E.



Gallus. Someone who has been orgified by the Mother of the Gods; he who has changed his nature from male to neither man nor women.

Etymologicum Magnum, s.v.

Neither is he changed into a woman, nor does he remain a man.

Augustine, De civitate Dei, 7.24

You [pagans] too have your third race; not indeed third in the way of religious rite, but a third race in sex (tertium sexus).

Tertullian, Ad nationes, 1.20.4

Both sexes are displeasing to her holiness, so he [the gallus] keeps a middle gender (medium genus) between the others."

Prudentius, Peristephanon, 10.1071-3

He would say that eunuchs were a third sex of humans [tertium genus hominum].

S.H.A., Alexander Severus, 23.7



berdache < berdache (French) < bardascia (Italian) < bardaje/bardaja/bardaj (“sodomita paciente,” Spanish; , s.v.) < bardaj (“slave,” Arabic) < bardah (“prisoner,” Persian) < vartak (Middle Persian) < vádhri (“eunuch,” fr. vadh, “to strike,” Sanskrit; used in Atharva Veda in reference to third gender figures) < *varta- (“seized, prisoner,” Old Iranian) < *wele- (“to strike, wound,” Indo-European)

American Heritage Dictionary, 3d ed.; Alonso, Enciclopedia del idioma; Oxford Sanskrit Dictionary.

But how to define gender without depending on sex? It's easier said than done! In my contribution to Third Sex/Third Gender I argued that identifying a social status as a third gender required showing consistent labeling and other linguistic practices in a society that distinguished a class of individuals from both men and women, and attributed them with a constellation of traits comparable to those traits used to define other genders. The presence of mythologies and narratives describing the origins of such roles, as well as rites of passage for individuals entering them, are also evidence of socially recognized multiple genders.

Subsequently, I expanded my research to the rich field of ancient history. I began by studying the priests of the Greco-Roman gods Cybele and Attis known as galli, who were sometimes referred to as members of a tertium sexus. Along the way, I discovered that parallel roles existed in the Near East and Mesopotamia, as well, going all the way back to what historians like to call the "dawn of civilization." Seeking to identify the full extent of the distribution of such roles, I turned next to south Asia, where I again found gender-defined religious roles, like the contemporary hijra of northern India, with remarkable similarities to the galli. Hijra and their counterparts throughout south Asia can be traced back to Sanskrit sources from the early first millennium BCE. Indeed, as I argued in an article in History of Religions (“Priests of the Goddess: Gender Transgression in Ancient Religion”), multiple genders linked to specialization in religious-administrative functions are a feature of the entire area from the Mediterranean to south and southeast Asia, a region of agrarian city-state societies that have been socioeconomically interconnected from prehistoric through Islamic times.


These were not, however, tribal societies with a gender-based division of labor and egalitarian social structure such as those we find in native North America. The element of specialization in crafts production so prominent in the case of berdaches is missing in the examples of galli and hijra. However, an examination of the belief systems of these societies reveals how multiple genders were constructed. In Greek, Roman, and Indian medical texts, sex is a function of biology, but the biology involved is inherently unstable—maleness and femaleness depend on relative levels of heat and cold, dryness and moisture, solid (e.g., bones) and soft (e.g. flesh), which individuals need to carefully preserve to maintain their sexual identity. In such a belief system, gender is redundant. There is no need for recourse to social influence or disease or life experience to account for variations in maleness and femaleness. The hydraulic model of the body easily accounts for a wide range of individual differences.

Conversely, in a society like Zuni, where infants are considered “raw” and ungendered until cultural intervention makes them “cooked,” gendered adults, biology is reduced to a minimum. “Sex” counts for little compared to the requisite ritual and social experiences that render raw infants into cooked people. In contemporary North American and northern European societies, on the other hand, both sex and gender are important categories. Gender, a learned social role, helps explain why individuals can vary from what is otherwise considered the biological destiny of their sex. Gender polices sex and creates a new domain of morality. That is, gender can vary from physical sex—but it shouldn’t and it’s our job as moral citizens to make sure it doesn’t.

In other words, the concept of “sex/gender system” introduced by Gayle Rubin cannot be applied universally. Not every society has a sex and gender system. Some emphasize one element to the exclusion or minimization of the other. My research has underscored for me the importance of examining the specific beliefs about sex (i.e., what makes a body male or female or other) and gender (i.e., what makes a male body a social man and so forth) before determining whether a given society has multiple genders.

Of course, the third gender concept has its limits, and they have been clearly pointed out by my colleague Stephen O. Murray in his provocatively titled commentary in Current Anthropology, “Subordinating Native American Cosmologies to the Empire of Gender.” (It was this title that led Steve to comment that my advocacy of third gender entitles me to the rank of queen in the empire of gender.) Murray writes, “I am uncomfortable with projections of contemporary Northern American and European conceptions of distinctions between sex, gender, and sexuality on peoples in other times and places” (60). He makes the point that not every social status is a gender role. Determining if a third gender is present, he argues, requires native contrast sets that consistently distinguish a third category in relation to other genders, and clear statements from natives to the effect that there are three genders. The use of pronouns other than those for women and men would also be strong evidence of a distinct gender (60).

I have argued that Murray's criteria are, in one sense, too strict, and, in another, too easy. Terms like medium genus and tertium sexus applied to galli, and trhytîyâ prakrhyti in Sanskrit texts, all of which literally mean “third gender,” and statements like those of contemporary hijra who define themselves as “neither men nor women,” seem to me to meet his criteria. But I would not pin determinations about the presence of third genders strictly on linguistic evidence. As I have suggested, we need to determine generally how men and women are defined and constructed in a culture, and then compare these sets of traits to the evidence of alternative statuses to really determine if these statuses are parallel constructions to other genders. Although the distinction between etic, or outside concepts and models applied to cultural data, and emic, or native, concepts and models is important, I think we unnecessarily limit our analysis when we insist that only native concepts, categories, models, and terminology are valid for discussing a given culture. While I’ve cited instances that meet Murray’s linguistic criteria, I also believe that there are cases where use of the term “third gender” is justified even though the people we are concerned with might not use that precise terminology, because the concept of “third gender” still gets us closer to their understanding of gender difference—closer, for example, than homosexual, transvestite, or transsexual.

However, Murray’s point is well-taken. There are many nuances in beliefs concerning sex, gender, and sexuality, and these categories are not universal. Sexual and gender differences can be conceptualized as states of non-differentiation or as multifarious states of differentiation; as androgyny or as the product of fuzzy categories. A third gender is only one possibility, but I do think it is useful for describing those instances of roles that cannot be explained in terms of binary male or female genders, in belief systems that do not fix gender in sex or define sex as stable and finite. I also believe that cross-cultural typologies are feasible and useful; that some historically-defined culture regions are characterized by the presence of multiple genders and some by their lack, and that these patterns have correlates in the social formations and histories of the societies in those regions.

But if calling both North American berdaches and ancient galli members of a third gender leads us to think that these roles have the same socioeconomic basis, or that they are thought of and valued in the same ways, then the limits of the concept of third gender have been reached. This is not a shortcoming that is intrinsic to the term itself, however, but a matter of how we go about using it. Its use needs to be justified in each case by consideration of a variety of evidence.

I think the greatest value of the concept of third gender is heuristic and conceptual. On the one hand, it helps us see the cultural and historical coherency of roles that until recently have been treated in isolation. At the same time, third gender helps us to break the vicious cycle of projection, in which Western heterosexist binarism is constantly replicated onto the world’s cultures (and most animal species for that matter) by the language we use. Third gender suspends the assumption that hetero-binarism is universal, that behind every gender lurks a sex to pull difference back to one of two versions.

What does all this have to do with sexual minorities today? In fact, I think we have a real stake in the discovery of multiple genders. As long as the language for talking about difference is confined to the possibilities of two mutually exclusive, fixed positions, I am convinced that lesbians, gay men, and others are bound to come off looking bad—as defective, counterfeit, or imitation males and females. The only alternative in such a system is to imagine androgyny, the mixture of two. But isn’t the mixture only more of the same? Androgyny is intrinsic to a heterosexist imaginary, not a point of opposition. Third gender, on the other hand, helps us to perceive all that is left over when the world has been divided into male and female, all the feelings, perceptions, and talents that are neither male nor female. While I would not go so far as to claim that we are a third gender, I would say that like a third gender we occupy an ontologically coherent, historically constructed, multidimensional subject position called—well—call it anything, as long as we abandon the binary language of heterosexism.

I find that the idea of third gender seems to capture how many people today feel about their sexuality and gender—not only gays, lesbians, bisexuals, and transgender people, but some heterosexuals as well. Many of us feel that our personalities are founded on a strong element of gender difference in relation to hetero-normativity, in addition to an emotional-sexual difference, but defining this difference as either female or male misses the mark. We are not males being feminine or females being masculine, but queers being queer—or whatever we call it. What we do and how we do it, what characterizes us, is not what either men or women do—it’s what gays, lesbians, and queer folk do.

Third gender helps us get past another confining definition, which says that our difference is singularly sexual. Scholars have shown how this “sexed being” called the homosexual was constructed and they’ve shown who did the constructing—elite European males and institutions of social control. Given these origins, I’ve never understood why we continue to employ this definition as if it were empirically-based. Did the progenitors of this term ever go out and observe “homosexuals”? Did they base the theories and terms they invented on a representative sample? No, of course, not. In fact, as George Chauncey has shown, medical-psychiatric labels such as “homosexual” had little impact on how American fruits, pansies, queers, fairies, dykes, and others defined themselves until after World War II. So much for the power of the mighty label, capable of leaping over all social boundaries, able to construct genders, sexualities, and identities in a single discursive move!

In 1987, I argued for the use of a multidimensional model in studying sexual and gender diversity, which would recognize that roles like “berdache” or “the modern homosexual” often have social, economic, religious, sexual, gender, and subjective dimensions. Here I learned of Rodney Needham’s work on polythetic vs. monothetic classification is useful. Polythetic categories are based on multiple traits and can be compared in more complex fashions than monothetic or single-trait categories, which limit one to either/or comparisons. The relationship of modern gay people to berdaches, for example, is not a matter of yes or no, but of two sets of traits. Comparing them, we see some traits are shared, others not. A third gender model does much of the same work, helping us to think about these roles without resorting to heterosexist language, in a way that encompasses sexuality but is not arbitrarily limited to it. The common connotations of “gender” are always-already multidimensional; those of sexuality already narrow and foreclosed.


Third gender really brings us full circle. David Halperin’s “one hundred years of homosexuality” may in fact be only two or three decades, the years since Stonewall in which the sexual definition represented by the term “homosexual” has held sway. But our history is not as impoverished as some social constructionists make it sound. There have been two modern homosexualities, not one—the type defined by object-choice alone without special reference to gender, about which one learns a great deal from Foucault, Weeks, and others, but also the type exemplified by the Molly and the Uranian, roles in which gender difference and same-sex desire are interpenetrated. Both rightly can be characterized as "modern" in that both were constructed in the modern era and both are understood as psychological phenomena characterizing types of persons (per Davidson 309). The individuals who founded the first lesbian/gay movement, the men and women who populated the urban subcultural enclaves of bohemians and sexual deviants that eventually became the space we now call our community, thought of themselves, defined themselves, and labeled themselves as Mollies, Uranians, intermediate types, and fairies—not homosexuals. At the same time, drawing an arbitrary line between these two modern homosexualities overlooks their genealogical relationship and their extensive co-existence; the way in which the homosexual was produced both from within as well as against the Uranian, and the debt contemporary lesbian and gay identities owe to both.

I would like to conclude by anticipating one other critique of the third model, one that in fact was made some time ago. In Homosexual Desire, originally published in 1972, Guy Hocquengham, refers to the third sex theory of Hirschfeld, and says “When it is not totally fascist, the third sex theory is dangerous” (121). This certainly gave me pause for reflection. Of course, Hocquengham considers any universalization of homosexuality to be complicit with the project of regulating and containing it, of enlisting it for the larger cause of Oedipus. This is also an early version of today’s critique of identity politics. And certainly to the extent that third gender conveys the notion of a role with definite contents, it is susceptible to this critique.

But anti-identitarianism itself is due for some revision—as several theorists have recently suggested (see, for examole, Bersani, Morton, Hennessy). The risks of not having an identity and not specifying its contents are just as great, if not greater, than those of having one. Invisibility and assimilation are not options—or at least, not options for a movement whose goal is liberation. In the same way that terms alone cannot be blamed for the ways they are used, it is not identity itself that is the pitfall but the way in which it is deployed. Identities can be constructed without recourse to crude essentialism, they can be compared as well as contrasted, and they can be powerful rallying points for progressive as well as conservative causes. If the risk of identity is that a group places itself within certain discursive regimes, what is the alternative? To talk about homosexuality in the absence of bodies that actually touch? But if we formulate the question in existential terms, knowing that construction is a part of the human condition so that there is no question of some other way of being, then the issue becomes how we live as ethical beings in these constructions; not identity vs. something else, but which identity, and what sort of contents will it have? The distinction between saying lesbians or gays are like third and fourth genders and saying that they are distinct genders, is subtle but important. The first statement is a hypothesis, a metaphor that invites elaboration; the second takes the form of a pronouncement that narrows discourse to assent or dissent.

Here, again, I think third gender is useful. The danger of assimilation or co-optation of such a direct challenge to heterosexist binarism seems to me much less than the potential for assimilation of an identity based on a difference that exists only in the fleeting moments of the bedroom. With or without the heuristic of third gender, I am convince that the discussion of our differences must be re-opened without foreclosing any lines of exploration for fear they might not conform to pre-existing ideological commitments or personal comfort levels. Much greater is the danger of saying nothing, of allowing phantasmagoric images of homosexuals to continue to stalk the landscape, of remaining culturally and historically anomalous, unable to articulate a relationship the greater strivings of humanity except by the self-effacing gesture of denying our difference in order to rejoin its ranks at the lowest common denominator. In my work, I’ve tried to take another approach—uncovering and exploring our queer differences as distinct and integral elements of the story of how humanity has come to reach the state of affairs that is the present.

©1995/2010 by Will Roscoe — do not reproduce or distribute without permission.