Troubling kinship: Sacred marriage and gender configuration in South India



Patterns of kin-making among devadasis in northern Karnataka pose problems for anthropological charting of kinship and for state projects of family normalization. Given to the goddess Yellamma by her family in a rite of marriage, a devadasi becomes a person who is both a woman and a son. Such a person cannot be mapped within a structuralist calculus of kin in which every position is always already gendered. I elaborate kin-making as a technology for producing gender and value in persons who can inhabit, but may confound, alignments between sex, gender, and kin position that have been smuggled into the anthropological project as kinship.

I don't have any male children, with this in mind I left my daughter, she will come to leave water before I die. I was tied to take care of my mother's property, and to continue her light (deepa), in the same way, in order to take care of me and mine I tied my daughter.


Shantamma, a dedicated woman (devadasi or jogati) from northern Karnataka, South India,1 described her decision to dedicate her daughter to the devi (goddess) Yellamma in genealogical terms.2 With this decision, she continued a lineage of material and affective relations through her daughter. Within the Dravidian context, as this region in South India is often described, patrilineal and patrilocal kin arrangements are conventional, and the conduct of death rites and inheritance of property falls almost exclusively, by custom and in practice, to sons.3 In the absence of a male child, Shantamma made her female-sexed child into a son. When parents give a daughter, or, sometimes, a son, to Yellamma, they reconfigure the gendered forms of labor, value, and personhood understood to inhere in that child. This reconfiguration is recognized in the Central Deccan as the effect of marriage to Yellamma and is visually marked by the distinctive red and white beads worn by tens of thousands of devadasis living and working throughout Maharastra, Karnataka, and Andra Pradesh today.

Devadasis become persons whose gendered modes of embodiment and relatedness exceed the logic of the kinship diagram and trouble the form of the family underwritten by the postcolonial state. As wives of the Devi, they can never be widowed and are ever-auspicious persons whose presence ensures and confers fertility.4 Within Hindu logics of personhood, gender, and value, such a person can only be a woman. Indeed, boys who are given or give themselves to the Devi become sacred women, jogappas.5 Devadasis are not only auspicious women, however, but they are also sons—persons endowed with the obligations of sons whose lineage position and entitlements within the kin network are reckoned as a son's are. Sometimes they are fathers.

In my early fieldwork exercises, I attempted kinship charting as part of a comparative survey of devadasi and nondevadasi households. I was equipped with the rudimentary tools of formalist mappings of kin: circles, triangles, horizontal lines for alliance and vertical lines for lineage. My efforts in the territory of what Bronislaw Malinowski called “the bastard algebra of kinship” (1930:19) came up against a certain kind of productive failure. The kind of persons devadasis become by virtue of their marriage to the devi Yellamma—women who are also sons—cannot be situated within a formalist mapping of kin relations in which every position is always already gendered. Jogatis are socially recognized women and, therefore, “.” They are also sons whose children's relationship to the wider network of kin in terms of inheritance and marriageability is reckoned through their genetrix's patrilateral kin position, thus, “△.” The system cannot admit a person who is simultaneously “” and “△.” How are we to think about this impossible grammar of gender and kinship as itself the result of a kin-making practice, the gift in marriage of a girl to the goddess? How might the existence of this practice in the world disrupt heterosexual, biogenetic, and secular terrains as the natural and inevitable grounds of kinship?

In this article, I explore the effects of kinship with the devi Yellamma. In particular, I focus on jogatis, or devadasis, the female women–sons who are subjected to state projects of reform in a way that jogappas, the male women are not. My argument here unfolds in a series of moves. First, I offer an ethnographic account of sacred marriage. I then briefly outline the history of devadasi dedication and its reform. This ethnographic and historical background allows me to draw attention to the paradoxical status of dedication as a practice that produces forms of personhood at once unproblematically quotidian and highly contested. This paradox prompts a question: Why has the practice of dedication endured? Working through the logic of the gift, I argue that this contemporary practice of sacred marriage serves as a means for families to increase the value of daughters. Having situated the practice of dedication within the rubric of transactional personhood, I then turn to an elaboration of the gendered effects of being given and the kinship trouble it produces. Sacred marriage produces both forms of difficulty and possibility for dedicated women and their families—in relation to the Devi and in relation to the state. Sacred marriage and its effects also produce forms of difficulty and possibility, or trouble, for anthropological mapping of kinship as a system of always already fixed gendered positions. In my final move, I turn to the implications of this alliance between girls and a goddess for anthropological considerations of kin-making.

Tying beads

If you want to get into this family you must tie beads for the Devi, if someone comes to me and asks, I will tie beads for them. It is a marriage.


When the beads are tied they throw rice and turmeric on us as in a marriage, the priest ties the mangalasutra (wedding necklace) and beads. A kailasha (pot) is put beside us as a representative of the groom, as the Devi, then rice and turmeric is thrown in her name and she becomes our father, mother, husband, everything.


According to a government survey, some 30 thousand devadasis are living in Karnataka.6 Trained through apprenticeship as priests (pujaris), dedicated women enact household, temple, and festival rites in their communities. They “keep” Yellamma in small temples and serve as intermediaries between her and her devotees. They accompany the Devi wherever she goes, playing her instruments and singing devotional songs (bhajans) in the households of dominant-caste (Lingayat) landholding farmers and at public gatherings.7 At Yellamma's main temple, only Lingayat men conduct puja (worship, rites), but in towns and villages across the Central Deccan Plateau, Dalit (outcaste) devadasis act as pujaris.8 To cultivate the favor of this devi, and thereby all forms of prosperity, fertility, and well-being, dominant-caste devotees must call on the women who keep her. Devotees transact with devadasis. They give the fruits of the harvest, new cloth, and money to devadasis and receive the mediated presence of the Devi from devadasis. Dedicated women who conducted the rites of Yellamma as their main occupation reported to me earning enough to sustain themselves and, often, the extended networks of kin they headed.

This is not an insignificant source of livelihood given that most devadasis come from landless or small-landholding Dalit families eking out subsistence in an area prone to drought and largely dependent on dryland agriculture. Once dedicated, they do not marry in the conventional way: Many say, “Yellamma is my husband.” Typically, only those families with “Yellamma in them” dedicate daughters. That is, this is a kin-making practice that extends over generations. Unlike women who take men for husbands (gandulavalu), devadasis inherit land, pass the family name on to their children, and function as heads of households. They fulfill the obligations and claim some of the privileges typically assigned to sons, such as fixing marriages, making payments for jobs for their sisters’ children, and roaming as they please. In the process, they remain primarily attached to their natal families, either dwelling within the natal household throughout their lives or returning to the place they call “home” for festivals and family occasions and, at the end of life, to retire. Once past puberty, many of these young women migrate to towns and cities to work in brothels. Of those who remain in their natal villages, most are given to, or take, patrons, usually properly married men of a higher caste with whom they have long-term if not lifelong exclusive sexual relations. Still others remain celibate throughout their lives.9 In all of these cases, as women who lack husbands, they have no claim to sexual respectability. Yellamma has claimed them, and the disposition of their sexual capacity proceeds through their relationship to her. That relationship is the primary ground of their sense of who they are in the world; they are muttu kattudavaru, those who have tied beads.

How do families who give their daughters to Yellamma account for this decision? In this discussion, I take my hermeneutic from the everyday language of dedicated women and their families. My sense of this language is drawn from 24 months of ethnographic research conducted in northern Karnataka and southern Maharastra between 2001 and 2012. Much of this research was conducted in a small village I call “Nandipur,” located in the Belgaum district of northern Karnataka. This research consisted of extensive participant-observation, over 100 interviews with dedicated women, and a comparative set of life-history interviews with brothel-based dedicated women, village-based dedicated women, and village-based women with husbands (gandulavalu). Dedicated women and members of their families describe the relationship between Yellamma and the women who are given to her as children in two major idioms: gift and trouble. Daughters are “given” (kodu), “left” (bittu, implied: in the care of), or “tied” (kattu) to Yellamma in fulfillment of a religious vow (harake) taken as a means to resolve the trouble Yellamma can send.

Pratima, a village-based devadasi in her forties, gave the following account:

Still no one has understood (Yellamma's) power, that is how much power is there. Still no one has understood her play (aata). As for me, listen, I will tell you. When I was small, I got blisters on my head, all over my body, filled with blood and puss, I could not be held, could not nurse properly … Only my mother would touch me, no one would touch me, some knowledgeable person said, “It's from god (devardinda), ask god [perform a rite of divination] then see what happens and then [if it's from god] apply Yellamma's turmeric saying her name, leave her as a jogamma (devadasi).” Then my parents took her turmeric and applied it to my head, after three days the sores were all dried up. When it dried up, they left me as a devadasi.

Her mother added, “We threw turmeric on her in the Devi's name and she became well, then we took her to Yellamma's temple, tied beads, and brought her back home.”

When devadasis account for how they came to be tied to Yellamma, they detail terrible skin ailments, fever, persistent and inexplicable illness exacerbated by medical treatments, deaths in the family, loss of livestock, failure of crops, serious quarreling within the family, and stubborn and terrible poverty. This is the trouble (kadaata) of Yellamma, her play (aata), and it is often articulated as a form of divine possession (Bradford 1983). She is said to have come “onto the body,” to be “in the family” or “with us.” Like other South Indian goddesses, Yellamma is hot in her aroused state and must be propitiated (cooled) with offerings of devotion and gifts of buttermilk, turmeric, parched grain, new cloth, silver, and cattle (Kinsley 1988:205). The highest-stakes version of this type of gift is a daughter whose dedication secures the favor of the Devi within the family. “We threw turmeric on her in the Devi's name and she became well, then we took her to Yellamma's temple, tied beads, and brought her back home.” The daughter is given, but she is not given away; she is brought home.

The daughter comes home, and Yellamma comes with her, in her. Recall Kamala's description of tying beads: “If you want to get into this family you must tie muttu (beads) for the Devi, if someone comes to me and asks, I will tie beads on the disciples. It is a marriage.” “Who do disciples marry?” I asked. “The Devi. Jamadagni [Yellamma's spouse] is not the father of the disciple, he is only the husband of Yellamma,” she said. In Kamala's description, those dedicated to Yellamma constitute a family. Entrance to this family is negotiated through an alliance with the goddess. This marriage does not follow the heterosexual norms governing conventional marriages: The disciple (whether male or female) marries the goddess. Moreover, the normative pattern of kin relations does not obtain: Jamadagni may be the husband of Yellamma, but he is not the father of the disciples. Yellamma is both their husband and their mother.

The trouble Yellamma brings is resolved through sacred marriage, an alliance that transforms the valence of her presence from affliction to fecundity. But this alliance also disrupts normatively modern patterns of kinship and gendered embodiment, producing its own pattern of kin-making. Thus, sacred marriage resolves one kind of trouble and produces another.10

Sacred marriage and its reform

Epigraphical evidence places the practice of dedicating girls to deities and temples across South India from about the ninth century on (Kersenboom-Story 1987; Orr 2000:5). A specialized class of temple servants, these women were choreographers, dancers, musicians, and ritual performers whose sexual capacity was mediated by and harnessed to their position in the temple as wives of the deity. While the particular ritual duties and performative arts attached to dedicated women varied by region, the structural features that they shared included kinship with a deity and a relationship to a temple, its economy, and its patronage. Most scholars agree that these women were sexually active, as an aspect of their ritual efficacy as well as a feature of their place in the system of temple patronage (Apffel-Marglin 1985; Kersenboom-Story 1987; Parasher and Naik 1986).

The uncontained sexuality of dedicated women has attracted the attention of generations of reformers. Along with the reform of other suspect cultural practices such as sati (widow immolation) and child marriage, the reform of devadasis in the colonial period took shape in the context of debates between officials of the British Empire and Indian elites who sought to demonstrate their worthiness to self-rule. These debates typically focused on women as the bearers of culture and “tradition” and the embodiment of the future of the nation (Mani 1998; Sangari and Vaid 1989; Sinha 1998). Within such debates, the extended sexuality of devadasis was set against law and respectable society. As Kalpana Kannabiran describes, “[They] … did not come within the very narrowly redefined boundaries of the monogamous family: an adult female has to be either a wife or a prostitute” (1995:66).

Unlike the devadasis who were subjected to reform in the colonial era, jogatis have never been associated with a highly elaborated dance form or a very wealthy temple complex. However, like their more elite counterparts, jogatis have long been subjected to projects of reform that produced their nonconjugal sexuality as illicit and their presence in the temple as contaminating. Jogatis are predominately rural-dwelling Dalit women, and their reform has emerged largely within anticaste movements in Maharastra and Karnataka.

Colonial-era acts banned dedications, but they stopped short of criminalizing any of the ritual or performative arts belonging to devadasis or of reformulating their kinship with the deity, as contemporary projects have. The 1982 Karnataka Devadasis (Prohibition of Dedication) Bill nullifies the dedication of any woman (it does not mention men) to any deity and confers legal status on any existing conventional marriages entered into by dedicated women and legitimacy on any children born from such newly legalized unions. It prescribes penalties of up to five years' imprisonment and fines of up to 5,000 rupees for dedication. It provides funds for economic rehabilitation schemes such as loans for the purchase of sewing machines or dairy cows and payments of 5,000 rupees to men who married devadasis (Jordon 2003:153). This law rewrites marriage to the deity as a crime against the state, and marriage to men as rewardable by the state, which has set itself up in loco paternus, arranging marriages and paying dowries. It codifies conventional marriage as the only legitimate form for reproductive sexuality. Patterns of relatedness between Dalit families and the goddess and the forms of gendered embodiment and value that they establish have been criminalized in an act that privileges heterosexual, monogamous, reproductive, and patrilineal patterns of kin-making in relation to the state. These patterns, in turn, regulate and reproduce dominant norms of caste endogamy, female sexual purity, and female economic dependence.

Marriage as a gift relation

The practice of dedication persists, the law notwithstanding. Why? To be tied to Yellamma is to be tethered to the natal family but loosened from the expectations of conjugality and its imperatives of reproduction and gendered hierarchy. Giving daughters to a goddess is consistent with the general logic of kin-making and reveals something of its investments. I am making a point here about the materiality of kin-making, about the nature of “the ties that bind.”

Anthropologists have worked through the force of the gift in social life in three ways relevant to this issue. First, as reciprocity—from Marcel Mauss (1990), the proposition that there is no such thing as a disinterested gift and that gift economies operate though three principles: the obligation to give, the obligation of return, and the obligation to receive. Second, as gender—from Gayle Rubin (1975) reading Mauss and Claude Lévi-Strauss (1969) against the grain, the idea that the exchange of that “most precious of goods—women” is constitutive of a sexual economy that vests men with rights in women that women do not have in themselves. Third, as value—from Marilyn Strathern (1988) and ethnographers working across South Asia, the theorization of exchange relations as productive of forms of embodied distinction and worth in persons, such as caste and gender (Lamb 2000; Marriott 1976; Raheja 1988).

The gift carries a burden; it confers a debt that must be repaid. The obligation to participate in exchange creates forms of social integration through reciprocity and forms of stratification through distinctions between givers and receivers. Thus, the asymmetry of gender is explained in part by Rubin in “The Traffic in Women” (1975) through the difference between exchanger and exchanged, a distinction that is maintained through the constraint of female sexuality: Women cannot give themselves away but must be given. Gender, in short, is the effect of transactions (Strathern 1988). Daughters are given, whether to husbands or the Devi, to form alliances. The gift of a daughter to the Devi creates an alliance with her and sets in motion a circulation of debt, gendered obligations, and care different from that circulation produced through an alliance with a man. When daughters are given to husbands, they and the wealth they may create through productive, reproductive, or sexual labor goes out of the family; when they are given to the Devi, this wealth remains within the natal family and is increased.11 Recall Pratima's mother's formulation: “We threw turmeric on her in the Devi's name and she became well, then we took her to Yellamma's temple, tied beads, and brought her back home.”

In both cases, women's value as wealth is put to work for the family. Properly married women, however, are not seen as exploited by their families in the way that devadasis are, at least, not by reformers. For reformers, proper marriage is the solution to the problem of sacred marriage. I am here arguing against the distinction this formulation poses between “good” (nonexploitative) marriage and “bad” (exploitative) dedication. This point is in keeping with feminist critiques of marriage as an exploitative property relation and common site of violence against women as well as accounts of extramarital sexual personhood, such as that of the devadasi or the courtesan, as a site of possibility (Basu 1999; Kannabiran and Kannabiran 2003; Nair 1994; Oldenburg 1990). In the anthropological sense, both dedication and conventional alliance are marriages organizing gender and value in persons and inaugurating flows of exchange between givers and receivers of wives (McKinnon 1991; Rubin 1975). Depending on how we understand the gift relation, both are exploitative or both are reciprocal.

Within the Dravidian patterns of kin-making found across much of South India, bilateral cross-cousin marriage is ideal. As scholars working in the region have noted, this is a system of alliance that produces equality between lineages over time through relations of reciprocity. In Thomas R. Trautmann's description, “Dravidian marriage always has the character of an exchange of sisters” (1981:24). In such a system the distance between wife givers and wife takers as an interval of time is measured in generations. This interval may well mask the exploitation of the system, if exploitation is what we call the burden of bearing gift value. But perhaps this mask too is gendered. Properly married women often recognize this burden; hence, the Tamil expression “Kinship burns!” Critiques of kin relations are rarely spoken by respectable women, and it was not until the very end of her research that Karin Kapadia (1996) was taught this expression by women who found the debt of marriage to outweigh its gains.12 Kapadia found this negative appraisal of marriage especially prevalent among women in upwardly mobile, low-caste families, in which secular modern ideals of conjugality and Brahmanical norms of chastity were converging. Women increasingly found themselves subject to seclusion in the home, where men's interests progressively held sway.13 In the Dalit households in Nandipur, I found these transformations in gender, caste, and class to be well under way.

In the context of the patrilocal and patrilineal kin arrangements dominant in the Central Deccan Plateau, women move out of their natal homes into their husbands’ households. The position of the youngest daughter-in-law in her new household is not enviable; typically, the physically most demanding forms of domestic labor fall to her. Her mobility outside the household is subject to male prerogative. If she works in the fields, weeding sugar cane or picking vegetables, her earnings come to the household. Her children belong to their father's lineage, and if the marriage fails, she will be expected to leave them behind. From the point of view of the natal family, in the short term, little is gained by giving a daughter to a son-in-law. Some relief from the burden of caring for her is shifted to his family, which has gained rights in her reproductive and productive labor and an interest in her respectability. In the longer term, over generations, through the principle of reciprocity and the pattern of mother's brother and cross-cousin marriage, the gift comes back to the natal family as a daughter-in-law. But, for my purposes here, and in the context of subsistence living, the short term is where the difference lies. The devadasi is given, but she is not given away; she is kept within the family.

Shantamma, a brothel-based devadasi who migrated as a young woman from rural northern Karnataka to Bombay, described this process to me in an interview. At the time of the interview, in 2003, she was living in a city in southern Maharashtra, where she was an active member of a collective of sex-worker peer educators working against the spread of HIV and for the rights and dignity of their community. Shantamma's mother was also a dedicated woman.

Shantamma: When my mother died, my uncles thought, she is the only child, if I were married someone would take my mother's property [meaning it would go out of the lineage], if I were left, if I were tied, I could continue my mother's land. Her property would come to me. They tied beads. I was tied when I grew up [matured].

Lucinda: What did you think when you were tied?

S: In the beginning I didn't understand, I didn't think anything. When I came from working in the fields, sitting in front of the house, I used to watch those women who had gone to Bombay to work. They used to come visit their relatives, I saw them and thought, if I go to Bombay, I can also earn like them, I can also wear good clothes like them. Maybe my life would become good, because at that time our condition was very bad. We didn't have enough food to eat or clothes. I decided to come here and work …

I used to send money to my uncles. Then my uncles said, it's enough, you have done lots for us, now you come to Marad, take a house for yourself and stay happily on whatever you earn …

L: How much did you give to your uncles?

S: Monthly I used to give 1,000, 2,000 sometimes I brought clothes for everyone.

L: Do you have any debt?

S: One lakh [100,000 rupees], we purchased land to cultivate, it lost money, the interest grew, for irrigation pipes in our field, we have some debt.

L: This is your debt, your uncles’ debt?

S: It's their debt, it's our debt, I should help pay it back because they are depending on my life, I have taken the responsibility of running the family, I must help them in their hard times.

Later in the interview, she explained her decision to dedicate her own daughter. “I don't have any male children, thinking like that I left my daughter, she will come to leave water before I die. I was tied to take care of my mother's property, and to continue her light (deepa), in the same way, in order to take care of me and mine I tied my daughter.”

Shantamma's rationale for dedication is similar to the explanation supplied by her mother's brothers: the continuation of social and material survival. This is not simply a matter of property relations and earned income, as she also emphasizes the continuation of forms of care secured through the dedication of her daughter. Indeed, the two are inextricable, as they often are in kin ties. As a devadasi, her daughter can perform death rites for her mother: “She will come to leave water before I die,” thus ensuring Shantamma's personhood through time as an ancestor. Of her own mother, she says, “I was tied to take care of her property, and to continue her light, I was tied, like this to take care of me and mine I tied my daughter.” The burden of debt and solace of affective ties intertwine. Forms of care and movable and immovable wealth circulate between dedicated women and their natal families. To put it more precisely, these material and affective goods circulate through devadasis as conduits and producers of wealth whose value within the natal family is enhanced through an alliance with the goddess.

Consider the status of the dedicated woman. She is given; she is the gift. She is by no means, however, a simple object of exchange. She is the gift and the countergift; she comes back to the family transformed by her marriage to the goddess, whom she now embodies, whose auspiciousness she bestows, whose oracle she becomes. Through an alliance with the goddess, she is empowered to make claims on the resources of dominant-caste devotees, whose offerings of grain, cloth, and money, in turn, provide the means of livelihood for her family. The dedicated woman is the form that relations between Yellamma and her devotees take. I mean form here in Strathern's sense. A person is the form that relationships take, a composite of relations, rather than a proprietary individual, as Strathern puts it on the basis of Melanesian material.14 Both material and symbolic forms of value are produced through this sacred alliance, and this value is not only the property of the one given but it is also hers to bestow. Devadasis are both producers and conduits of value in transactional economies of material and symbolic wealth. They are both gifts and givers. Or, rendered in the idiom of gender, they are both women and men. This complicates the gendered grammar of kinship.

Kinship trouble

What happens when kinship goes awry? To recapitulate, in “The Traffic in Women,” Rubin argues that gender is the effect of the exchange of women between men, or marriage. In a published conversation with Rubin about this essay, Judith Butler makes the point in these terms: “You then speculated that it might be possible to get beyond gender … if one could do something like overthrow kinship” (1997:72). In her own writing, Butler explores this question of an “overthrown” kinship. Through the figure of Antigone, she suggests that destabilizations in normative kinship can produce disruptions in gender: Oedipus's sons become like daughters, his daughters like sons. In Butler's words, “And so we have arrived at something like kinship trouble at the heart of Sophocles” (2000:62).

As I have noted, trouble is something devotees of Yellamma are very familiar with. The kinship trouble in Antigone's family seems to be the effect of a transgression of the founding law of incest. In the families of Yellamma's devotees, it is an effect of the presence of the Devi and a sign of her power. In a conversation in Nandipur about how her mother-in-law came to put aside her sari and began to wear pants and a shirt, a devotee said, “The Devi gave her lots of trouble. First she had four children and then the Devi said I will not give you a family life (janma), I need you. Her children died. She left home and began living alone and the Devi made her wear white clothes and bells (geje). Yellamma is so powerful she gave her a different manifestation (rupa).”

The loss of “family life” and a change in bodily manifestation is a sign of the Devi's trouble, her desire, and her power. Here, kinship trouble produces unconventional forms of gendered personhood, as in Butler's account. But it does more than that. It also opens up forms of personhood that articulate themselves in relation to the gods and thus disturb secular modern conceptions of the social as bereft of the gods and of the human as ontologically singular. As families with Yellamma in them know, becoming kin with the Devi, getting tied to her, is the best way to resolve this trouble, to convert it into well-being. This well-being, however, does not map onto conventional narratives of familial belonging, gendered conduct, and sexual life.

Indeed, Yellamma has her own complicated family history. The universalizing modern family ideal in which men provide, women are protected, and children are innocent goes unrealized in her life, as it does in the lives of those women who wander in her name.15 According to the history her devotees tell, Yellamma, or Renuka, as she is also called, was married to a great sage, Jamadagni. She was so pure she was able to perform miracles. Forming a pot from loose sand on the bank of the river and coiling a cobra for a pot rest, each day she brought water to Jamadagni for his morning puja. One day, as she was walking and balancing this pot on her head, she saw ghandarva (celestial beings) playing in the river. She lost her concentration, the pot dissolved, the snake slipped away. When she returned home, Jamadagni immediately recognized that she had been tempted by desire. He became very angry and ordered their three sons to cut off her head. The two eldest refused and were cursed with impotency by their father for this failure of paternal loyalty. Yellamma failed to maintain the feminine ideals of ritual purity and sexual chastity; she was distracted by pleasure. This is the story of Renuka–Yellamma, wife of Jamadagni, as published in pamphlets sold by the main temple trust and illustrated on posters for sale in every market town in the region.

As a fact of social life and means of organizing sexuality, kinship is trouble for Yellamma. It burns. At the same time, Yellamma's loss of sexual propriety and exile from the conjugal family is the ground of her power. She is reborn as an autochthonous devi who wears a mustache and distributes fertility across the region through the women she has taken as wives. As I show above, sacred marriage has its productive possibilities even as, or perhaps because, it produces transformations in gender. Here I consider three transformations: devadasis as sons, as mamas (mother's brothers), and as fathers.

As is often noted in the literature on Karnataka devadasis, one of the most common local explanations for their dedication is the absence of sons in a family (Joint Women's Programme [JWP] 1983; Mahale 1986).16 This production of what they refer to as “honorary” sons through sacred marriage is seen by some scholars as simply a strategy of place holding that seeks to maintain the continuity of the patriline through hoped-for male children of the dedicated woman (Epp 1997). My conversations with devadasis and their families suggest, however, that dedication has more to do with the work gender is made to do under conditions of hardship than simply with the continuation of the patriline. Within dominant constructions of gender, labor, and kinship, sons are produced as economic actors who earn for their families, whereas daughters and the reproductive and productive wealth they produce belongs to their husbands’ families. Moreover, even in families with male children designated as sons, in which the patriline, as such, is secure, daughters are given to the Devi. As was the case in Yamuna's family, many become heads of households over male brothers by virtue of their ability to better earn and sustain the kin network, that is, to more successfully meet the obligations of sons. This suggests not only that daughters may be produced as sons but also that the presence of jogatis in a kin network may disrupt the articulation between male-sexed offspring and the designation of persons as sons. Patterns of gender, kinship, and obligation here are not simply transposed from absent men onto women, as the category “honorary son” suggests. Instead, these patterns are reconfigured in such a way as to maximize the resources available to the natal family.

The status of devadasis as sons is also evident in their right to inherit land. As well as having implications for the gendered entailments of property inheritance, the recognition of devadasis as sons within a system of kin is linguistically marked. The children of devadasis do not call their genetrix's male-sexed sibling “mama” (mother's brother), they call him “cakka,” the term in Kannada for father's brother. These terms mark a distinction between matrilateral and patrilateral kin essential in a society in which cross-cousin marriage is ideal and parallel-cousin marriage is taboo. When the children of a devadasi call their mother's brothers “cakka,” they are indexing her position as father to them and as brother to her male siblings. This is a recognition that governs decision making about the marriage of the children of devadasis. Whereas the ideal marriage partner in this region for a daughter is the mother's younger brother,17 this person would be taboo for the daughter of a devadasi.

This was first brought to my attention by Meena Sesho, who has worked with devadasi and nondevadasi sex workers for many years building an internationally renowned model program for HIV prevention and rights-based community organizing in Sangli, Maharastra. In listening to a conversation among devadasis about the marriage of one of their daughters, Meena gathered that they were trying to sort out whether the prospective groom was kin to the bride: “What does it matter if you are in the same gotra (patrilineal clan), mother's brother's marriage is there.” In response, Devaki, the mother of the bride, exclaimed, “But as a devadasi I am as a son, he would be cakka to my daughter, not mama!” All the other devadasis jumped in in spirited agreement. Over the course of my research, I came to understand that this effect of kinship with the Devi was taken for granted and explicit among devadasi families and those familiar with them.

Meena's account of Devaki's marriage deliberations focused on the kinship effect of dedication and its gendered implications: “I had not realized the extent to which they are a man in the family. If I could have the status my brother has in my family I would choose it in a minute. It is not just that they are like the man, they are the man.” The distinction Meena made raises two interesting questions: What is the difference between being a man and being like a man? Why do we make this distinction? What is at stake and for whom in how the gender of embodiment is parsed from the gender of kin position?

If devadasis are “like” sons, if they are “honorary” sons, this is not the same as being sons. The former is an analogical or indexical position, the latter an ontological status. The use of the term honorary in reference to persons recognized both as female and as sons seems to insist on the limit of culture in the face of nature. Sex (maleness, femaleness) is situated as natural—understood as essential, perpetual, and real—while an “honorary” gendered kinship position (son, daughter) is cultural—ascribed, adopted, fictive. In other words, the logic would seem to be that devadasis cannot actually be sons but only like sons, because they are not persons recognized as male or masculine, and a person who is a son is necessarily a male or masculine person.

If, however, we untether kin position from the logics of sexual dimorphism and binary gender assignment and follow, instead, the logic of dedicated women and their families, other forms of gendered personhood and anthropological recognition come into view. Devadasis are described, recognized, and related to as persons who are both women and sons. They are not either women or men, either daughters or sons. They are both women and sons. I learned more about how to think about this both–and form of gendered personhood on a follow-up research trip in the winter of 2004.18 I met with a group of brothel-based devadasis, including many with whom I had conducted life-history interviews in 2003, and I asked them more about the kinship effects of their dedication. “As devadasis many of you have fixed marriages, bought land, and made payments for education or jobs for your sister's children.” “Yes, yes,” they chorused around, seemingly unmoved by this banal line of questioning. “This is the work of mamas (mother's brothers), these are the things that mamas do for their sister's children.” “Yes, yes.” “Because you are a devadasi, your brother is a cakka (father's brother) to your children.” “Yes, yes.” “Then are you mamas to your sister's children?” To my surprise, the circle erupted with retorts, “No, no, no. No one would say we are men (gandisu).” “No, we are not like that, you have not understood.”

We went back and forth in a lively conversation, in which I attempted to clarify that I was not suggesting that anyone saw them as male persons or men. Meena added that she did not see them as men (as her earlier comments to me might suggest) but as really having the status of men in the family. I clarified that I was trying to learn if they occupied the structural position and fufilled the material and affective responsibilities mamas have toward their sister's children: “Do you do the work of mamas?” This was not the sticking point; they quickly agreed that they were expected to, wanted to, and did fulfill these obligations. What seemed to upset them was the possibility that I had misrecognized them as persons whose bodily comportment signifies gandisu, man. Finally, one of the most vocal contributors to the exchange said in a suddenly quiet voice, “The children of two sisters cannot marry, but because I am a devadasi, my [married] sister's children can marry my children, in this way we are mamas.”

This was another surprise, one that clarified two things. The first is that devadasis’ position as sons within their natal families does not transform their embodiment of sex/gender or their recognition as women. They transact within their kin network as sons and incorporate the affective obligations of mamas to their sister's children, but this incorporation does not disturb their social recognition as women, persons whose bodily comportment successfully signifies “woman.” The second is that the kinship effects of dedication are not limited to the dedicated person. They ripple across the whole extended kinship network in terms of patterns of care and obligation as well as inheritance rights and marriage rules. Their status as sons is incorporated within and across the kin network and, through time, across generations. They are not men, but they are sons. As a matter of bodily comportment or perceived essence, they are not men. As a matter of structural position, affective orientation, and transactional obligation, they are sons. This seemingly paradoxical social fact of gendered personhood puts what is often taken to be the ontological foundation of the category “son” into crisis. I return to this point below. For now, I pose it in the form of two questions: What role does bodily masculinity or male sex assignment play in constituting who is or can become a son? How might devadasis trouble this?

The structural and transactional masculinity of devadasis within the grammar of kinship is not limited to their recognition as sons. Sometimes they are also seen as fathers, as the following exchange with a brothel-based devadasi indicates.

L: What was it like in your family when you were young?

Margawwa: My father and mother were working hard and filling their stomach. They left me to Yellamma, from this I started filling my stomach. I have two children. Our situation was very bad. I have brothers, but they are [only] taking care of themselves. My father died, my mother is still there. I have two children, one is in school. I am working and earning my food, taking care of them.

L: Before coming here, where were you?

M: I was in the village, working hard and feeding us.

L: How did you come here?

M: I came here to earn in the name of Yellamma …

L: How did you decide to come to this town?

M: I came here to fill the stomach on my own.

L: Your children's father?

M: Father? I am the father.

In her terms, as the one who provides for her children and sees to their education, she is their father. She fulfills the obligations of a father and claims for herself that kinship position. In the course of my conversations with them, devadasis rarely made this claim so directly. However, the conviction it captured was not unusual among devadasis, almost all of whom have children for whom they are the sole providers. As I have described, many of them also support extended family kin. Their claim on the position of father was also sometimes corroborated by children. As Ishwara, the son of the most prominent devadasi in the village I lived in, said to me over tea one day on the subject of the man who contributed biological material but not material sustenance to his making, “He is not my father, I have no father, my mother is my father, she is everything to me.” When this young man married, he brought his bride home to his mother's house, in keeping with the dictates of patrilocality and the place of his genetrix as father. This pattern was observed by all the sons of devadasis still living in Nandipur.

Devadasi dedication produces female-bodied forms of personhood not available to domesticated women; devadasis are economically productive lineage bearers (as sons or fathers are) and conduits of auspiciousness (as only married women can be). Further, they are not only conduits of but also transactors in auspiciousness, reproducing not only children for the next generation but also fertility and fecundity within the households of devotees and the agricultural body of the village. The value of men in a patrilineal society—as economic actors, property holders, and lineage bearers—and the value of women within Hindu conceptions of life and the conditions for its renewal—as auspicious persons—are consolidated in the person of the devadasi. This consolidation is achieved through a kin-making practice, the marriage of a girl to a goddess.

Troubling kinship

Anthropological attention to anomalous patterns of gender and kinship is not new. For instance, in 1951, E. E. Evans-Pritchard wrote about a rare practice among the Nuer, in which a man marries a woman to his deceased childless sister. A woman so married “then counts as a man” (1951:112) and becomes a female pater for the homestead. Subsequent literature on female husbands, or woman–woman marriage, in African societies identified maintenance of the patriline, continuation of property, and wealth accumulation and consolidation as reasons for the practice (Greene 1998). Denise O'Brien (1977) points out that within the societies where this practice obtains, female husbands are not considered deviant or abnormal, as they may seem to outsiders. Described as a creative or adaptive practice, woman–woman marriage increases the status and power of the female husband and the wealth of her natal kin (Greene 1998). The female husband shares many of the features of the dedicated devadasi, and much of my own analysis has unfolded along the lines laid by the scholars I reference here. However, my thinking departs from anthropological accounts of female husbands in two ways. I elaborate them as a means of clarifying my analytical project here within the body of anthropological work on anomalous configurations of gendered personhood, for instance, Albanian sworn virgins (Young 2000), Native American berdache (Williams 1986), and Brazilian travestis (Kulick 1998).

Regina Smith Oboler's definition demonstrates one of the tendencies in this literature I have sought to avoid: “A female husband is a woman who pays bridewealth for, and thus marries (but does not have sexual intercourse with) another woman” (1980:69). For Oboler, a female husband is a female person, a woman filling the role of a husband. The nature of her sexed and gendered personhood, in accounts such as this, is self-evident. This kind of description reasserts the foundational status of the sexed body or gendered person: A person is a man or a woman; however, he or she might occupy a differently gendered kin position in the sense of a social enactment and economic function. “Woman-marriage” Beth Greene writes, “in particular, gave some women advantages that normally accrued to men and formed an avenue for social mobility for the women who served as husbands” (1998:396–397). To serve as is not the same as to be.

The second tendency in this literature performs the same underlying foundationalist theory of gender in relation to kin position. Scholars writing on female husbands debate whether, and the extent to which, the female husband is or becomes a man. Evidence of gender ascription or achievement is made to speak to the degree to which masculine comportment and social conduct are performed and recognized. There are two possible sex/gender positions here: female/woman and male/man. Oboler's rich and detailed consideration of this question concludes, “Female husband is thus a category which, in some sense, occupies an intermediate position between male and female” (1980:83). To situate the female husband as caught in between feminine women and masculine men is to locate the problematic in the person rather than in the categories of gender and kinship and the relations we pose between them. Within such logic, female husbands can be men to the extent that they successfully fulfill the economic, social, and political duties of a pater or husband and accomplish recognition as a person of masculine bodily comportment. They might be in between, almost men, but they cannot be both husbands and women at the same time. The gendered character of the kin position remains ontologically stable: A husband is self-evidently a man. A female husband is a woman who may be understood to have become a man to the extent that her gendered comportment and enactment are masculine and align with the always already gendered character of “husband.” If this comportment or enactment is incomplete or mixed, she is in between, or third.

The extensive anthropological literature on “third gender” is also relevant here, especially for the ways this scholarship has documented the multiple possible articulations between the sexed status of bodies and gendered character of personhood and has challenged the binary conception of sex/gender (Cohen 1995, 2005; Herdt 1994; Reddy 2005). Literatures on social men and third gender have facilitated my thinking here about the gendered character of devadasi personhood up to a point. Devadasis are unproblematically women and sons at the same time. They are not “neither men nor women” (Nanda 1990) or “intermediate” and in between; they are both–and. To the extent that anthropological considerations of anomalous patterns of gender and kinship have remained within the logic of the binary or the spectrum, they have not gone far enough. As models of gender, both “either–or” and “man–in between–woman,” stabilize and naturalize “men” and “women” as discrete, unitary, and complementary forms of personhood.

I make two points here about the implications of this cultural practice of dedication—in which persons are made and unmade through kin relations with a goddess—for how anthropologists have thought about relatedness, gender, and the body. My first point is very much in keeping with existing anthropological work on female husbands and male daughters. It is also indebted to feminist analyses of marriage as a form of exchange relations that produces gendered distinction and hierarchy (Hutchinson 1992; McKinnon 1991; Rubin 1975; Udvardy 1995; Weiner 1992). In Strathern's (1988) terms, transactions do not occur between self-evident men and women; transactions produce persons as men and women. The social masculinity of devadasis suggests that gender cannot only be understood as the property of a person, as in accounts that naturalize gender through recourse to the sex of the body (e.g., Busby 2000). Kin positions are maintained by transacting with kin in specific and gendered ways: A father is a father because he meets the obligations of fathers. A man who does not fulfill these is insulted: “He is not my father”; a devadasi who does fulfill them claims the position: “I am the father.” This suggests that gender can function as the property of transactions, rather than simply as the property of particularly sexed bodies or a person's successful performative iteration of normative femininity or masculinity. Thus, a devadasi is a son, although her body is sexed female and her embodiment is unambiguously feminine. She is a son because she is given the structural position and social obligations of a son in the natal family and she inhabits them. Gender, I am arguing, is not only performative but also transactional. It is also necessarily tied to strategies of and for relatedness. The transactional aspect of gender emerges when we are attentive to the operations of the gift and if we take kin-making to be a technology of the human, a tool people use to produce certain kinds of persons endowed with particular obligations to and claims on others, social functions, and forms of value.

My second point about the kinship effects of dedication is this: The kinship chart is performing what it purports to merely describe.19 It presumes correspondences between the sex of the body, the gender of a person and that person's kin position—a male-sexed person △ is a man △, is a father △, brother △, son △—that do not everywhere occur. The disarticulations between the sexed body, the embodiment of gender, and the assignment of kin position that occur among Yellamma's kin demonstrate that these correspondences are not inevitable. The devadasi in a sari is not a daughter in the house but the son. The jogappa in a sari is a daughter and a son. Thus, kin-making with the goddess denaturalizes kinship as a system of fixed gendered positions.

David Schneider (1984) dethroned this genealogical conception of kinship as a universally applicable grid of human relatedness in A Critique of the Study of Kinship. As he pointed out, this is a folk model based on Western understandings of the “natural facts of life.” Predominant human practices of kin-making tend to take on the status of the “natural facts of life” within their own societies (Collier and Yanagisako 1987). They can come to inform what counts as relatedness elsewhere, among other persons. Since Schneider's critique, few anthropologists working on kinship and relatedness have taken sexual difference and heterosexual reproduction to be the “natural facts of life” so neatly mapped in the form of the kinship chart.20 At the same time, foundationalist understandings of sex and gender in relation to kin position continue to be smuggled into the anthropological project through the diagram as a basic tool of our trade.

Moreover, as Elizabeth Povinelli (2002) has pointed out in the context of her work on aboriginal Australians and land rights, the genealogical imagination continues to function as a grid of intelligibility, especially as concerns state recognition or citizenship. These terms of state recognition are not without their relationship to anthropological accounts of kinship. Camille Robcis (2004) has made this case in relation to contemporary debates in France over civil union (Pacte Civil de Solidarité, or PACS) and same-sex parenting arrangements, in which Lévi-Strauss's conception of the symbolic has figured significantly. Robcis quotes a French politician: “I respect different ways of living one's sexuality, there is no question of homophobia, but the Republican marriage answers to our anthropological principles. Homosexuality is not a social model that allows society to continue; it is contrary to the existence of sexual difference which founds marriage” (2004:111). In her essay, “Is Kinship Always Already Heterosexual?” Butler (2002) situates the possibilities of queer kinship as a question of the limits of recognition outside the terms of what Schneider called our “folk model” of kinship, namely, sexual difference and heterosexual reproduction. In short, the challenge anomalous patterns of kin-making pose continues to offer a useful corrective to modernist conceptions of the nature of the family and the question of relatedness.

My argument here is with the logic of the chart and the categories it stabilizes. I am interested in where this anthropological tool fails. As feminist and queer appraisals have shown, a focus on sexual dimorphism and human sexual reproduction as the sine qua non of kin-making has produced exclusions in the field of kinship studies (Hayden 1995; Weston 1997). Modes of relatedness that do not conform to the presumptively universal principles of descent and alliance have been foreclosed. Schneider's critique notwithstanding, these foreclosures have circumscribed kinship as a knowledge project among anthropologists who have failed to describe relatedness where they did not see reproductive heterosexuality replicated over time. Such failures discursively reproduce the social death that awaits those exiled from kinship. John Borneman (1996) has made this point in his consideration of marriage as an object within anthropological discourse. Writing about A. R. Radcliffe-Brown's commentary on “unnatural” offenses (“incest, bestiality, in some societies homosexuality, and witchcraft”), Borneman notes that sex outside marriage has been placed outside the “social,” as a matter of morality, perhaps, but not a constitutive element of human sociality in the way that sex–marriage is.

As I have described, my efforts to situate devadasis and their kin within the rubric of the kinship chart met with failure. Devadasis are women “,” readily socially recognized as female or feminine persons. They are also persons whose genealogical and transactional personhood is reckoned as a son's is, so they are “△.” This kind of personhood cannot be situated within the diagram. It falls outside the social, to the extent that the diagram represents the elemental relations understood to constitute the social: kinship. At the same time, devadasis’ inadmissible personhood is itself the consequence of one of these elemental relations: marriage. That they are both “” and “△” at the same time is precisely an effect of their marriage to the Devi. And yet scholarly treatments have tended to render this marriage as a merely symbolic act—a “marriage” (Orchard 2007; Orr 2000; Soneji 2012). This designation suggests that alliance between a girl and a goddess does not have the same force as a “real” marriage in organizing social and material relations and forms of value in and among persons, including gender. There is no need to register it as a form of kinship. The appraisal of sacred marriage as fictive sits uneasily next to the fact that the state found it necessary in 1984 to legally annul marriages between devadasis and Yellamma to make devadasis available for conventional marriage. If it is not “real,” why does it need to be abolished?

The gift of a child to this popular South Indian devi produces a distinct mode of intelligible human life. It changes one kind of person, with a particular kin position and reproductive and familial trajectory, into another kind of person. It reconfigures relations between those given and their natal kin as well as relations among devadasis, the Devi, and her devotees. It is in this sense that I refer to kin-making as a technology, a means of transforming persons that reiterates techniques of the body (Mauss 1973) such as gender and caste as well as forms of knowledge about the nature of persons and their place in the world (Strathern 2005). “Kinship,” Donna Haraway has written, “is a technology for producing the material and semiotic effect of natural relationship, of shared kind” (1997:53). Building on the insight of Strathern 1992 and Yanagisako and Delaney 1995 that, for Euro-Americans, kinship mixes nature and culture, Sarah Franklin and Susan McKinnon write, “[It is] a cultural technology not only for naturalizing relationships, but also … for producing what will count as the difference between nature and culture” (2001:16). Focusing less on the naturalizing effects of kinship and more on its function as a means of self-fabrication, Christopher Roebuck offers a somewhat different formulation, “As a human technology, kinship both produces particular kinds of humans and is a tool by which humans come to craft individual and collective lives” (2013:91–92). Like these scholars, I frame kin-making as a technology to point to the social and semiotic character of kinship as a set of arrangements produced through human activity and productive of persons able to accomplish recognition as persons. Further, I turn to the language of technology in an effort to capture the systematic as well as the inventive aspects of kin-making.

Technologies are products of human activity and innovation and, as such, are flexible and creative. They are also limited by available knowledge and techniques. Technologies can make things into other things, they are transformative. But they cannot make any one thing into any other thing; the possibilities of transformation are delimited by available forms of knowledge and materials. For instance, medical technologies routinely make a part of a pig into a part of a human but only as the result of specific anatomical knowledge about the human body and the porcine body as partible and analogous as well as of the requisite surgical techniques and antirejection pharmaceuticals and access to them.

As a technology, kin-making is innovative. Possibilities of human being and doing are made and unmade through everyday practices such as marriage, adoption, and domiciliation. Innovative kin-making practices open new pathways to forms of social and political recognition and inclusion. For instance, marriage is sometimes pursued as a pathway to legal citizenship, access to health insurance, or other means to and forms of human thriving. As a way of securing the positive regard of the state, however, marriage is not a tool everyone is able to grasp. Kin-making practices are thus also constrained: They occur within regulating and normalizing fields of power that shape legitimated forms of relations and recognizable kinds of persons. To the extent that they confirm modern ideals of the family and biogenetic or genealogical terms of kinship, anthropological categories and tools can and do continue to play among these disciplinary discursive regimes.

To frame kin-making as a technology is to draw on formalist understandings of kinship as a rule-bound system of positions as well as post-Schneiderian conceptions of relatedness as an open field of affiliation characterized by relations of care (Borneman 1997), choice (Weston 1997), or endurance (Butler 2002). It is also to refuse either model as adequate on its own, or, put the other way around, to insist on both constraint and creativity as features of kin-making. To think of kin-making as a technology of human generation and transformation is to get beyond the distinction between “real” and “fictive” kin and out of a field conscripted by blood and alliance. It might seem that we long ago left this field, but to the extent that the genealogical grid continues to delimit our categories of gendered personhood and relatedness, we have not. Within a field populated by binary, unitary, and discrete categories of gender and personhood, the relationship between Yellamma and her wives can only be a “ritual” marriage, its gendered effects merely “honorary.” But when we take the realness of their marriage and the wives’ (both–and) status as women who are also sons for granted—as ontological facts—in the way that devadasis, their neighbors, and families do, we bring our categories into productive crisis and open up the field of intelligible kin and kind.


Forms of institutional and material support for this work were generously provided by the University of California, Berkeley; Cornell University; Harvard University; the University of Kentucky; and the American Institute for Indian Studies. I am indebted to several colleagues for their critical engagement, especially Srimati Basu, Cori Hayden, Sarah Pinto, and Rachel Prentice. Years of conversation about queer lives and kinship matters with Lawrence Cohen and Chris Roebuck have made this writing possible. The assistance of Jyoti Hiremath and Ambuja Kowlgi facilitated the research. I am grateful most of all to the dedicated women who patiently taught me about their lives.


  1. 1

    Devadasi (servant or slave of the deity) is a pan-Indian Sanskrit term used to refer to women dedicated to deities in a variety of regionally distinct practices across the subcontinent, some of which date back to the ninth century C.E. In the region under description—the Central Deccan Plateau—dedicated women most often refer to themselves as jogatis. In governmental and nongovernmental reports and journalistic representations, however, they are referred to as devadasis, a term they have come increasingly to apply to themselves. In this article, I use both the regionally specific term, jogati, and the pan-Indian designation, devadasi.

  2. 2

    Yellamma is the most popular deity in the region; during the pilgrimage season, half a million devotees from all castes throng to her main temple by the side of the Malaprabha River, making it the most significant pilgrimage site in northern Karnataka. Like other South Indian amma (mother) devis, Yellamma is ambivalent, a potential source of cure as well as affliction. Sometimes invoked as Shiva's consort, sometimes as an autochthonous mother goddess, Yellamma is mythologized and ritually propitiated in ways that place her within, and in between, two of the three major strands of Hinduism: Shaivism and Shakta.

  3. 3

    As Srimati Basu (1999) has argued, although granted the right to inherit paternal property under the Hindu Code Bill in 1955–56, Hindu women across India have mostly forgone this right in favor of maintaining gendered codes of sisterly loyalty and self-sacrifice.

  4. 4

    Devadasis are auspicious women (Apffel-Marglin 1985; Kersenboom-Story 1987). Frédérique Apffel-Marglin writes, “Auspiciousness [is] a state which unlike purity does not speak of status or moral uprightness but of well-being and health or more generally of all that creates, promotes, and maintains life” (1985:19).

  5. 5

    In a compelling structuralist analysis of this phenomenon, Nicholas J. Bradford writes, “Indeed, it is considered to be part of Yellamma's power and character to change a person's sex: ‘ganda hoogi henna maadataala, henna hoogi ganda maadataala, come as a man, she will make a woman, come as a woman, she will make a man’” (1983:310).

  6. 6

    A 2007 Government of Karnataka survey found 30,000 devadasis in 14 districts. See Kumar 2009. In the 1990s, Karnataka reported 22,941 identified devadasis to the National Commission on Women, Andhra Pradesh reported 16,624, and the Government of Maharastra, which has repeatedly accused Karnataka of being the “source” of the devadasis who work in the brothels of Bombay and Pune, reported only those receiving government pensions and other forms of support. See JWP 2001–02.

  7. 7

    The concept of a dominant caste was developed by M. N. Srinivas (1955) to convey the social fact that although Brahmans rank at the top of the Varna system, they are not always or everywhere the most economically and politically powerful or “dominant” caste.

  8. 8

    Dalit (Marathi, lit. smashed or broken) is the self-designation of politicized members of communities formerly labeled “untouchable.”

  9. 9

    Estimates of the percentage of dedicated women working in brothels vary widely, from 80 percent, a figure cited within public health contexts (Gilada 1993; Shreedhar 1995) to 40 percent (Shankar 1994:107). In the village where I worked, no dedicated woman in living memory had left the village to take up brothel-based sex work. All had, or had had, patrons who were otherwise properly married within caste. In other areas, virtually all dedicated women take up “the line” for at least some period of their lives (Orchard 2007). Unlike most village-based devadasis, brothel-based devadasis in towns or cities are not monogamous, finding cash for sex transactions more lucrative than patronage relations.

  10. 10

    This other trouble is, in part, what Judith Butler, writing in another context, has called “gender trouble”—the subversive confusion, and promiscuous proliferation of genders—and therefore identity, beyond the binary. In her words, “There is no gender identity behind the expressions of gender; … identity is performatively constituted by the very ‘expressions’ that are said to be its results” (1990:25).

  11. 11

    For an extended consideration of the question of value in relation to sexual economy and sexual citizenship, see Ramberg 2011.

  12. 12

    “Kinship burns!” is Kapadia's (1996:44) translation of the Tamil expression “Sondam sudum!”

  13. 13

    See also Anupama Rao's (2009:232–236) useful discussion of the relationship between sexual economy and caste, itself the result of the regulation of sexuality through practices of monogamy and endogamy. As Rao points out, another sexual economy, one in which upper-caste men claim extramarital sexual access to Dalit women, works to reproduce caste hierarchies. Attention to this sexual economy provides a useful backdrop against which to consider the investments of Dalit men in securing the respectability of “their” women.

  14. 14

    This is not to make a claim about sex acts or sexual ritual. The value of dedicated women, as I am describing it, exists as the effect of their position in this sexual economy, whether or not they are exchanging sex for means of livelihood with patrons or clients. Whatever compensation they receive for sex is, of course, part of the material value they convey to their natal family. I never found any indication that sexual relations between jogatis and their patrons, clients, or lovers took the form of a ritual, blessing, or encounter with the Devi. This suggests that the sexual act has been thoroughly secularized, although, as I am arguing, the disposition of sexuality has not, as yet. That is, however, clearly the intention of the antidedication law and its annulment of sacred marriage. See Gurumurthy 1996:132–133 for a historical consideration of sex acts as fertility-renewal rites in the space of the temple at Saundatti.

  15. 15

    Itinerant life, renunciation of domestic family life or householder status, and dependency on the offerings of householders for livelihood are all features of a strain of South Asian asceticism. See, especially, Ramaswamy 1997 and Khandelwal et al. 2006.

  16. 16

    For a study of the cultural production in Albania of female-sexed offspring as masculine persons who occupy the social and structural position of sons, see Young 2000.

  17. 17

    Variations on this pattern are common. For instance, the ideal partner may be mother's brother's son, mother's male cousin's son, and so forth. From the perspective of a male ego, elder sister's daughter is the ideal marriage partner.

  18. 18

    See Ramanujan 1989 for a delightful consideration of both–and thinking as an aspect of Indian epistemology.

  19. 19

    I am grateful to Chris Roebuck for this phrasing.

  20. 20

    Jane Collier and Sylvia Yanagisako make a similar point: “The standard units of our genealogies, after all, are circles and triangles about which we assume a number of things. Above all, we take for granted that they represent two naturally different categories of people and that the natural difference between them is the basis of human reproduction and therefore, kinship” (1987:32). For them, the presumptions of the chart demonstrate the mutually constituted character of two fields of study: gender and kinship.