8
Marriage Alliance

Introduction

We return repeatedly to the role of trust in the conduct of commerce and to the commercial transaction as an index of trust. Besides building and signaling trust through elite endowments, Nakarattars also built and signaled trust through various kinds of nonelite transactions, especially nonelite transactions between relatives. The items exchanged in these transactions ranged from gifts, deposits, and loans to payments, goods, and services; every exchange was carefully entered in the appropriate ledgers of Nakarattar family firms. Some of these transactions appear, from an economic point of view, to be trusting to the point of irrationality—at least, until one looks beyond the short-term payoff. Yet it is clear that low-interest, long-term loans and deposits to relatives—like nonvicarious ritual gifting by elites—had substantial financial rewards.

There is a danger here. Such a formulation makes it sound as though successful Nakarattar business was prescribed by codes of conduct for kin, by recipes for relatives. Such thinking obscures the freedom with which Nakarattars could manipulate kinship ties to minimize risk in financial transactions, and it completely hides the way that Nakarattars used financial transactions to create the framework of kinship itself. But once we remind ourselves of the dangers of reification, what is important to realize is that in kinship, as in finance, trust was the essential ingredient. Its presence was required for transactions to take place; its absence was enough to sever a kin relationship.

The present chapter begins by providing a brief overview of trust between relatives, measured (quite literally) by Nakarattar terms for kin-

ship. Its nine subsequent sections then focus upon and analyze various transactions between affinal kin, or "in-laws." The preliminary section of definitions describes the overall range of Nakarattar kin relations as designated by terms for different groupings of kin. The second section identifies characteristic relations of affinity and marriage alliance that hold between members of these groups. The third section explores some of the pragmatic considerations that affected marriage alliance formation and the complex alliance structures that sometimes resulted. I give special attention to the intersection of marriage alliance and territorial segmentation in Chettinad and to additional complicating factors that arose in the wake of Nakarattar business activities in different parts of Southeast Asia. The fourth section provides an illustrative case of an alliance in operation. The fifth section describes the ceremonial prestations that symbolized Nakarattar marriage alliances, contrasting them with those that symbolized non-Nakarattar castes. The sixth through ninth sections examine the implications of Nakarattar marriage alliance for standard theories about kinship terminology, marriage rules, and affinal prestations among South Indian castes generally, and especially for the theories of Louis Dumont (1983). The conclusion looks forward to Chapter 10 where I consider why Nakarattar marriage alliance differs from the pattern predicted by standard theories of Dravidian kinship, relating the Nakarattar pattern to Marriott's (1976) theory of Hindu transactions, and viewing the Nakarattar pattern as a specific adaptation to the caste's occupational niche as merchant-bankers.

My information about Nakarattar kinship was collected primarily from directed and undirected interviews carried out during my field work in Tamil Nadu in 1981. Some data were obtained by observing contemporary weddings. But many of the practices described below are no longer performed and are only dimly remembered by some of the older Nakarattars. Accordingly, my description of Nakarattar marriage alliance is offered as a reconstruction of customs before 1930. I believe these customs were in force during the primary focal period of this book, 1870–1930, and I present my description in the past tense.

Preliminary Definitions: Terms for Kin Groups

Nakarattar terms for kin groups segmented the caste into contrasting categories that provided an index to three levels of social distance: (a) structured kinship (including relations between in-laws), (b) diffuse kinship, and (c) common membership in the Nakarattar caste, but no kinship ties. These levels are depicted in Figure 13. At its most diffuse level, Nakarattar kinship terms distinguished caste members who were kin (contakkarar ) from those who were not (contam illai ). Kin included all of one's

Figure 13.
Levels of social distance in Nakarattar kin relations.

agnatic relations, collectively referred to as pankali: that is, members of a person's clan (kovil pankali ) as well as members of one's lineage (kuttikkira pankali ).^[1] Kin also included members of a one's spouse's lineage, all of whom were referred to as campantippuram ("in-laws"). Once kinship relations were established between two sets of kuttikkira pankali through marriage, they lingered on even after the paired descent groups had terminated jural obligations contracted at the time of the marriage. Affinal relations between kin groups did not extend to members of a spouse's temple clan outside his or her lineage.

Nakarattars employed the terms tayapillai and tayati (plurals tayappillaikal and tayatikal ), in both a broad and a narrow sense. In the broad

sense the terms referred to members of descent groups with whom Nakarattars had once established a marriage alliance, even if they no longer maintained the alliance as such. The terms indicated that marriages had previously occurred between female tayatis and male pankalis . In this sense, tayati designated a more diffuse relationship of kinship than a campantippuram relationship, that is, a relationship between descent groups observing an ongoing marriage alliance. Relationships within this diffuse body of affines were recognized, and they might be activated by persons seeking favors, much as were the ties between otherwise unrelated members of the same temple-clan. There did not, however, exist any institutionalized set of rights and obligations for tayatis such as those incurred by two lineages actively allied by marriage.

Nakarattars viewed all of the husband's agnatic relatives as the basic unit of reference for the wife as well as the husband. They constituted an enormously extended family into which the wife had married and were referred to by the focal couple and its offspring simply as "our pankali ." Active campantippuram members who belonged to the wife's uterine descent groups, however, were marked by a narrow use of the term tayati , signaled in discussion by reference to specific moral obligations.

Marriage Alliance, Affinal Kindreds, and Tayatis

Exactly what was a Nakarattar marriage alliance? Minimally, it was a negotiated agreement between two sets of descent groups related to the husband and wife of a married couple. The primary groups involved were the joint family (valavu )^[2] and the encompassing lineage segment (kuttikkira pankali ) of each individual. The linked groups in a Nakarattar marriage alliance created an affinal kindred defined by consanguineal relationship to one or the other spouse of a married couple.

Membership in Nakarattar affinal kindreds was not reckoned by a criterion of genealogical distance from a specific individual,^[3] and for this reason some anthropologists may be tempted to conceive of the group formed by a marriage alliance as a kind of compound descent group, comprising the husband's clan and the wife's lineage.^[4] But conceiving the affinal kindred in this fashion, underestimates important differences between descent groups and kindreds that have been all too infrequently acknowledged.^[5] In the first place, an affinal kindred was not defined by descent from a common ancestor and, on the basis of this substantive consideration, should not be confused with any kind of descent group.^[6] Secondly, and perhaps more importantly, structural relationships between members of an affinal kindred—including relationships marked by important financial transactions—were marked by relations of reciprocity. In this, they

differed notably from the redistributive relations that mark Nakarattar descent groups. As we shall see in Chapter 9, descent groups were organized around temple cults that—like the multicaste temple cults described in Chapter 7—provided a focus for the collective pooling and redistribution of resources to their members. These cults, whether constituted of a clan, a multivillage lineage, or the multicaste inhabitants of a Chettinad village, formed corporate units that reproduced themselves from generation to generation. By contrast, affinal kindreds were marked, with one exception, by a set of prestations between two sets of allied descent groups. These affinal prestations generated, in Sahlins' (1966) words, a "between-group relationship" rather than a "within-group relationship." Moreover, among the Nakarattars, the relationship between parties of an affinal kindred was not inherently reproductive.

Affinal kindreds could be initiated by a marriage between two joint families or two larger lineage segments without any previous participation in a marriage alliance. They lasted for at least two generations, during which time their members bore rights and responsibilities to each other. But, counter to most models of marriage alliance in Dravidian kinship, Nakarattar marriage alliances were not automatically renewed. At most, affinal kindreds might, under appropriate conditions, be renewed indefinitely by successive two-generation alliances.

Some researchers studying Dravidian marriage have found it useful to distinguish between, on the one hand, an alliance relationship , formed by the first marriage between families of extended lineages, and, on the other hand, an alliance , where the relationship is perpetuated (Burkhart 1978). In the former case, the alliance seems oriented primarily toward the respective joint families of the married couple. In the latter case, the couple's total affinal kindred is implicated, with ramifications that may extend to a core group of repeatedly intermarried families in a marriage circle (Burkhart 1978). On the assumption that Dravidian kinship terminology necessarily entails perpetual alliance, scholars such as Dumont (1957b) reserve the term alliance for the recursive, self-generating form. In the following discussion, I differ from this usage and employ alliance and alliance relationship interchangeably, for both perpetual and "one-shot" alliances. I qualify my usage with appropriate adjectives as needed to indicate whether the alliance is regenerative or terminated after a finite period.

Members of Nakarattar affinal kindreds cooperated in a variety of ways. They were tied together, in part, by a recurring sequence of ceremonies and rituals that provided opportunities to exchange information and lobby one another for assistance in various enterprises, of both a business and a nonbusiness nature. These same cyclic and singular ceremonies

constituted occasions on which resources were accumulated and deposited to form a trust fund in the bride's name, over which she maintained nominal control. In effect, the deposits—referred to as accimar panam (married woman's money)—constituted a bequest, insofar as they were normally intended for the benefit of descendants of the newly married couple, and especially for any daughters that might be born to the couple. Since a share of the bride's natal family's estate was set aside for her and her descendants (especially her daughters) through the mechanism of the dowry, it is accurate to view the formation of these trusts as reflecting a secondary line of inheritance through the female line. If the bride died before bearing any children, all monies deposited as accimar panam trusts traditionally reverted to the bride's natal family.

A considerable length of time could pass before the bride's children were old enough to benefit from or control their mother's accimar panam in any direct way. Normally, it was not spent until they incurred educational or marriage expenses. In some cases, the fund or some portion of it even remained indefinitely as a kind of communal property under control of the husband's joint family. In the meantime, the accumulated resources provided financial reserves on which all the different joint families of the alliance kindred could make various claims (see Chapter 5).

This is not to say that affinally allied joint families and other descent groups made equal contributions or had equal claims on the accumulated resources. On the contrary, the status of a married person's respective descent groups varied according to the rights, duties, and privileges each spouse possessed with respect to his or her hearthhold (pulli ), joint family (valavu ), or lineage (kuttikkira pankali ). These claims varied widely according to the individual circumstances of the family group involved. In principle, as already noted, the resources in question belonged to the bride. The major part was given by her father (amman, mamakkarar ) or brother (attan, maittunan ) in the form of cir tanam ("dowry") and in related gifts.^[7] However, since she was largely subject to control by her husband, and he was normally subject to control by the senior male in his joint family (see Chapter 6), her rights were largely nominal. On the other hand, the lack of direct control over decisions allocating the use of funds in the trust did not eliminate all privilege of access for the wife's natal joint family or other joint family among the couple's tayatis . In some cases, the long process of negotiation leading up to the marriage could result in an agreement that some portion of the fund be deposited in the business enterprises of the tayatis .

In general, Nakarattar families felt some aversion to investing an affinal trust in businesses owned by either family in a marriage alliance. One

of my informants went so far as to say that such funds were never invested in family business since "it was risky to put all the eggs in one basket." Accordingly, affinal gifts were frequently deposited in the business of third parties, either in interest-bearing accounts or in the form of special, interest-bearing hundis . Although these hundis were technically discountable on sight, they were used as long-term certificates of deposit and were rarely cashed without considerable advance warning (see Chapter 5). In some situations, an independent banking business might be established in the bride's name. This was especially the case in situations where a son-in-law had no financial assets of his own and could not expect to inherit any. In all of these cases, however, the effect was to open up a line of credit nominally to the bride, but in effect to the various members of the kindred according to their specific claims in the alliance.

Some of the factors contributing to the formation of a marriage alliance have already been examined from the point of view of a family's choice of grooms for its daughters (Chapter 6). A candidate was normally selected after he had already completed his first three-year term as the agent for a Nakarattar banking firm. Accordingly, his future potential as a businessman was subject to knowledgeable evaluation in the marriage-arranging season that followed his return to Chettinad. In fact, although I have no statistical data, it is my impression that a young man was often groomed for the part of son-in-law by a prospective affine, who first hired the young man as an agent for his business firm. If the young man proved himself abroad and relations remained cordial between agent and principal, the family could then propose a long-term alliance through marriage with a daughter.

Selection of the bride for a family's son proceeded along a different route. For example, more emphasis was placed on a bride's physical beauty than on a groom's. Another difference was that a bride's potential capital contributions to an alliance were not evaluated on the basis of expected inheritance or demonstrated business acumen. In fact, until 1947, Hindu women had no rights of inheritance, and there was no way to test in advance the potential contribution a wife's council might eventually make to her family's decision making. Nevertheless, a bride generally did make a substantial contribution to her husband's family business. The bride's counterpart to a groom's inheritance was her dowry along with other affinal prestations that resulted from a marriage alliance between two families.

The size of a dowry was determined by delicate negotiations between families. If the groom's family was seen as too greedy or the bride's as too stingy, a wedding could be called off, although blame would be placed else-

where—for example, on the sudden discovery of a previously unnoticed misalignment of horoscopes or a previously unknown "parallel cousin" common to both families.^[8] Other affinal prestations were not subject to explicit agreement, but a general understanding and set of expectations would be generated and generally followed.

Vattakais and Territorial Endogamy

Nakarattar marriage alliances reflected and influenced principles of territorial organization. The major territorial factor affecting them was a system of prohibitions that circumscribed bounded, endogamous territories called vattakais . In these vattakais , Nakarattars established panchayats (councils of influential elders) for making certain kinds of collective decisions and for resolving certain kinds of conflict. That is, the Nakarattar caste encompassed what amounted to incipient subcastes or microcastes governed by the external sanctions of judicial bodies. They did not, however, form marriage circles by adherence to normative obligations of reciprocal marriage exchange. Accordingly, their social organization did not reflect any strictly "elementary structure of kinship" (Lévi-Strauss 1969).

The pre-1930s system of Nakarattar vattakai -bound microcastes survives today only in fossilized form, as a statistical pattern among contemporary alliances and in the memory of a handful of Nakarattar historians who have taken an interest in their own past. Thus, the evidence for reconstructing colonial structures of Nakarattar marriage is meager and difficult to come by. Such as it is, I am encouraged by its fit with other components of Nakarattar caste organization during the colonial period. However, the organization and functional implications of vattakai endogamy did not become apparent to me until after I had left Tamil Nadu, and I did not pursue a systematic inquiry as part of my field investigation.

My first clue to the historical operation of the system came, typically, while discussing another topic with an informant. I had been trying to ascertain whether any Nakarattar employed the Tamil term nakaram or the English phrase "marriage circle" in reference to exogamous Nakarattar clans (Chapter 9)—a usage reported by various non-Nakarattar writers on the topic.^[9] (Chandrasekhar 1980; Price 1979). My informant at the time—a knowledgeable man in his seventies with a long history as an industrialist, a regional politician, and a man active in the internal politics of the Nakarattars—agreed with other informants that the usage was unheard of. But he wondered whether the writers in question were referring to a territorial division of Chettinad into "West Circle," "East Circle,"

and "South Circle." On further questioning he referred to these circles as vattakais ,^[10] and offered the information that these vattakais were traditionally endogamous, although vattakai endogamy was now breaking down. According to this man, the cultural principle underlying endogamy was simply a preference for geographical proximity: "No family wanted to send its daughter too far away."

Although I am confident that this rationalization accurately reports Nakarattars' sentiment about their daughters, it cannot account for vattakai endogamy. In a relatively uniform topographic terrain such as Chettinad, such a preference would result in an evenly distributed network of marriage alliances, not the tripartite system of bounded endogamous segments that existed. I continued to enquire about Nakarattar vattakais with all of my informants. But during the course of my field work I found only three or four other Nakarattars who had heard the term vattakai , and none of them were able to describe the whereabouts of vattakai boundaries or identify the sets of Chettinad villages that were members of the different vattakais .

I received no further clues until I returned to America and began to go closely over some of the Tamil documents I had collected during my stay in Tamil Nadu. At this time, I noticed a significant reference in the book Chettinadum Tamilum (Chettinad and Tamil) written by my principal informant, Lakshmanan Chettiar, known as Somalay (1953: 12): "In Chettinad, Chettiars still hold their community meetings on the pattern of Chola village assemblies. The sub-divisions, vattakai and karai are still mentioned." An appendix to Somalay's book contains a table of seven territorial divisions including three vattakais (for a version of this table, reordered according to the Latin alphabet, see the key to Map 4; see also Maps 3 and 4).

A tabulation of 91 sample marriages that occurred in 1980–1981, taken from the marriage register of Neman clan temple, indicated that 64 of the marriages (70%) were endogamous within these clusters even today. Moreover, 10 of the 27 mixed vattakai marriages (37%) occurred within Melappattur and Mela Vattakais, and another 12 within Pathinattur Vattakai, Kila Vattakai, and Kilappattur (44%). My speculation is that the seven-way territorial division represents a post-seventeenth-century fission of an earlier three-way division, and that these two large clusters of interterritory marriage actually maintain an underlying vattakai endogamy. But even if this theory is incorrect, there remains a considerable nonrandom clustering of territorially bounded marriage zones (see Table 10).

I wrote Lakshmanan Chettiar asking about the function and organization of vattakais . He returned the following reply:

A cluster of villages, in Chettiar terminology, is called vattakai . There are three major vattakais and three minor ones. Till fifty years ago, there used to be held vattakai conferences of Nagarathars [Nakarattars]. No such meet has since met. Representatives from the village temples (of the Nakarathars) in each vattakai still meet on rare occasion to decide austerity measures in marriages, etc.

Till say 1939, there were not many roads in Chettinad. Transport was few and far between. For marriage and vevu ceremonies, death rituals, etc., the Pangalis [pankalis: members of local lineage segments] used to march on foot. Therefore, it was the custom to contract matrimonial alliances within a ten-mile radius. In those days, it was possible to have first-hand knowledge of the families only within such short distance. If a boy (groom) could not get a suitable match within the vattakai or if a girl could not be given away within this territorial jurisdiction, it was considered as a stigma to that family.

Adoption was not confined to the vattakai .

Since 1939, things have changed vastly. Even now 2/3 of all Chettiar marriages take place within the vattakai . But marriage within the vattakai is no longer the rule. All villages are connected by road, by public transport and by telephone. To check up anyone's credentials or verify family background is very easy. During World War II, families which stayed on in Malaysia and Singapore contracted marriages among themselves without reference to vattakai and this has caught on. In [one of the Chettinad villages] for instance, people have contracted marriages with almost every village in Chettinad. [A textile industrialist residing in the village] set the ball rolling by

effecting marriage alliances with all textile mill-owner families in Chettinad. In Devakottai, it was the tradition to finalize marriages with families within the town. There are 2,000 families and so it was not a problem to find a suitable match. But, in Devakottai too, people are slowly and in increasing numbers effecting marriages elsewhere. If two families are in business in Salem or say in Coimbatore and are sufficiently acquainted with each other over the years, the vattakai ceases to be a bar. I can speak of at least a thousand instances where the vattakai system has been given the go-by. (Lakshmanan Chettiar, personal communication, March 1981)

In this letter, Lakshmanan Chettiar subscribes to the notion that preferences for affinal proximity explain the formation of vattakai . The explanation is no more satisfactory coming from him than from my previous informant. But his further amplification, along with the comments in his book, suggest some of the reasons for the preference and some of the reasons why it should now be breaking down. Apparently, vattakais at one point constituted regional assemblies built upon ancient political assemblies operating in the Chettinad region of Pandyanadu.^[11] As such, they must have provided a crosscutting organizational framework that complemented those of lineage, clan, and business station described in Chapters 6 and 9. As we shall see more clearly in Chapter 9, the village temple provided a framework for political coordination within a single village. The clan temples provided an institution capable of sanctioning the behavior of dispersed clan members from multiple villages. The vattakai panchayat apparently provided a coordinating institution for Nakarattars belonging both to different villages and to different clans. Thus, it must have operated in Chettinad like vitutis or caste associations operated in Nakarattar business stations (see Chapter 6). Significantly, the most prominent feature in my informants' descriptions of vattakai deterioration concerned the substitution of principles of business association for residential proximity in determining suitability for marriage alliance. This substitution most likely took place because, both in the case of Nakarattars sharing common business stations and in the case of Nakarattars sharing a common industry, business-related sodalities provided alternatives to vattakais as institutions for distributing information and resolving disputes among segments of dispersed clans and localized lineages. The vattakai system declined with the creation of the new colonial political framework—especially with the creation of the system of electoral district boards in 1936. By the end of this period, vattakais retained (significantly in this context) only some coordinative functions in setting limits on the cost of a dowry.

Affinal Gifts as Interest-Earning Trusts: An Illustration

In 1980–81, most of my informants complained about the high cost of a dowry which, they felt, had grown out of control during the last fifty years (see Table 11). It is difficult to judge the accuracy of this nostalgic vision of times gone by. Memories of low-cost dowry do not fit well with the prominent role given to dowry deposits (accimar panam ) in turn-of-the-century banking practice (Chapter 5). Unfortunately, in the absence of pre-1930s documentation for dowry expenses, it is necessary to consider a modern example. Given my contemporary informants' nostalgia, one must be cautious in taking the magnitude of money involved as a basis for historical projection. I am more confident about continuity in the structure of affinal prestations and in the social and financial ties they created.

I illustrate the kinds of financial claims that affinal allies may place on each other—especially the claims that a married couple (pulli ) may place on members of the wife's natal family (tayatis , in the narrow sense)—by presenting a brief account of a relatively extended case. The story begins with a marriage that occurred in the late 1930s. It continues with the marriages of this initial couple's children. As an analysis of a single case, it should not be regarded as typical. It will be apparent that I collected more information about the marriages of my informant's daughters than about his own marriage forty years before. Nevertheless, similarities between the two generations of marriages suggest that no significant change had taken place.

My informant described both families in the 1930s as upper middle-class. The groom's family business was located primarily in Burma. Its income was estimated at about Rs. 40,000; its assets at about Rs. 500,000–600,000. The business interests of the bride's family were located in Ceylon and Malaysia. The income of her family was about Rs. 50,000–60,000, with assets of about Rs. 300,000–400,000. Their natal villages were located within six kilometers of each other. There had been no previous mar-

riage alliance between their respective families. My informant summarized the arrangements for the marriage ceremony in the following words.

Wedding expenses (kalyanaccelavu ) were shared by both families. The bride's family contributed about Rs. 5,000; the groom's, Rs. 4,000. Normally, the groom's family would only have spent about Rs. 2,000. But my father spent an extra Rs. 1,500 for special musical performances.

My wife's dowry (cir tanam ) was Rs. 5,000, a high figure at that time.

We received wedding gifts (cir varicai ) worth about Rs. 25,000.

Some gifts (mamiyar caman —i.e., from the bride's mother's brother) were also given to the groom's mother including saris, jewelry, utensils, etc. Their value was less than Rs. 2,000. No money gifts were given to the groom's mother before 1941. (Interview with Lakshmanan Chettiar, May 1981)

The marriage produced four daughters, born between 1940 and 1960. They were married at ages that ranged from sixteen to twenty-three (the later they were born, the older they were married).^[12] Table 12 itemizes the various expenses incurred on the occasion of their weddings, along with my informant's explanatory comments.

Based on this information, it is possible to make a few observations about the financial investment that marriages represented to this Nakarattar family. It is also possible to indicate something of the kinds of financial cooperation that occurred within the affinal kindred established at the time of the initial marriage alliance. To begin with, there is a discrepancy in the kind of information collected for the father's and the children's weddings. In the former case, we have information both about the expenses of marrying one's daughter, and also about the income and assets of the intermarrying families (notice, it is my informant's wife's natal family who bore the brunt of the expenses). But we do not know what other kinds of expenses these families faced—especially what expenses other than my informant's wedding were borne by his wife's natal family. Nevertheless, it is possible to observe that the total cost of the wedding to her natal family of Rs. 37,000 represented about 10 percent of its assets and about 60 percent of one year's income.

In the case of the children's generation, we do not know the assets and income of the allied families. But we do know that my informant's family acted as in-laws under similar obligations for the weddings of at least his four daughters. We are also able to estimate the relationship that each wedding bore to the others as a proportion of the expenses incurred by all the weddings. I note, further, that my informant had only one son, his

youngest child. This proportion of daughters to sons and especially the absolute number of daughters represents a piece of financial misfortune for Nakarattars and for Tamil families generally, as indicated by the common blessing "May you have a hundred sons," and also by the Tamil proverb "Five daughters will ruin the wealthiest man." Accordingly, this potential source of bias should be kept in mind if generalizations are drawn from my illustrative case.

Finally, my informant indicated that his family had lost most of their money when the majority of Nakarattars left Burma in 1941, never to receive compensation for their substantial landholdings and other Burmese assets. This is confirmed by information he provided me about the amount of dowry appropriate for Nakarattars falling into different economic classes. I present his estimates in Table 11. They demonstrate clearly that he views his family's fortunes as having declined from upper middle-class to lower middle-class and perhaps even to "poor" status. They also give some indication of trends in dowry payments during the last fifty years.

All of these considerations indicate that my informant could be regarded as something of a hardship case. Nevertheless, my impression is that the means he employed for financing the enormous expenditures required by the marriages of his daughters are typical of Nakarattars generally and cast considerable light on economic transactions between Nakarattar kin groups within affinal kindreds. The sources of marriage funds are presented below.

The important lesson here is that 45 percent of the considerable expenses incurred by my informant in marrying off his daughters were borne in turn by his wife's uterine family. The process by which these funds were accumulated reflects even more deeply the ongoing operation of a Nakarattar marriage alliance. My informant's second daughter was

named after her maternal grandmother. In recognition of their special relationship, this grandmother gave her granddaughter Rs. 1,001 one week after she was born. The amount was kept in deposit as accimar panam in a firm of tanners in Dindigul. Every year, the principal plus interest was redeposited. Occasionally, the grandmother took the opportunity to round off the amount when the deposit reached an "odd" figure.^[13] Periodic festivals such as Deepavali also marked occasions for new contributions to the fund. Assuming that these gifts made up the bulk of the funds donated by the family comprising my informant's in-laws for the marriage of his second daughter, and accepting the figure of 30 percent as accurately reflecting their contribution to the cost of her wedding, the trust established in her name must finally have reached perhaps Rs. 16,500. It is a sum that reflects a substantial credit line directly linking the recipient firm of tanners and my informant's joint family and indirectly available to other joint families among my informant's affinal kindred.

The Ritual Construction of Nakarattar Marriage Alliances

The ritual construction of a Nakarattar marriage alliance provided a detailed picture of many normative relationships between members of affinal kindreds.^[14] In the following analysis, I focus on Nakarattar ceremonial transactions commencing with prenuptial rituals and continuing long after the wedding celebration, for the life of a marriage alliance.^[15] Such transactions were carefully recorded by Nakarattars in special notebooks that, as we shall see, formed a thoroughgoing bookkeeping system for tracking relations of trust and cooperation between kin. My point of departure is Dumont's (1957b) analysis of a similar chain of the prestations and counterprestations that symbolize marriage alliances among non-Nakarattar castes—specifically, in a Vellalar ("farmer") caste of Tirunelveli and some Kallar and Maravar ("warrior") castes of Chettinad and neighboring regions.

If I correctly understand Dumont, these affinal prestations fall into essentially two categories. Dumont describes the first category of prestations as external prestations —called cir or curul —which are presented back and forth between families allied by the marriage of their son and daughter.

[W]ith the ceremony proper [the couple] begins the series of visits to and stays with F [the wife's family], the couple being every time accompanied by a number of baskets [cir ] containing foodstuffs and other articles for consumption, from M [the husband's family] to F, increased back from F to M. Prestations from F dominate more and

more as time goes on until finally—it may be two or three years after the marriage ceremony—the young couple establish a separate household near M, and receive the necessary pots and pans from F without any return gift. This is the "cir of going apart." (1957b: 30)

Dumont describes the second category of prestations as "internal prestations" — called moi:

During the marriage ceremony, in both houses, money is collected among the bridegroom's relatives on the one hand and the bride's relatives on the other. This is called moi; its effect is to make the relatives contribute to the expenses of the family. (1957b: 30)

Dumont notes that moi takes various forms among the different groups he studied: among the Mudukkulattur Maravar, moi is absent at weddings but present at funerals and female puberty rites; among the Nangudi Vellalar, moi is replaced by collections where the two groups of pankali pool their gifts in a common fund rather than contributing to two separate funds; similarly, among the Paganeri Kallar, all who attend the feast contribute moi , which consists of a small contribution of rice and also a collection of money, called revei , which is accompanied by a gift of thanks in return.

These variations in the composition of moi -giving groups cast doubt on the generality of Dumont's contrast of cir and moi as a difference between external and internal gifts—that is, between gifts given between the bride's and groom's pankalis and gifts given within each pankali . But Dumont never elaborates on the significance of this distinction, even among those castes where it does seem to hold. I will offer some speculative thoughts on the issue after examining the kinds of prestations found among the Nakarattar.

Cir

Nakarattar affinal prestations took a variety of forms, some of which overlap with those described by Dumont. To begin with, Nakarattars applied the term cir in compound terms referring to two kinds of gifts: cir tanam (dowry) and cir varicai (gifts to the couple on the occasion of the wedding). As is partially the case in the groups studied by Dumont, both of these varieties of cir can be viewed as forms of prestation by the bride's family to the groom's family, or to the new pulli formed on the occasion of the marriage and assimilated into the groom's family. Nevertheless, there are some important aspects of Nakarattar varieties of cir that deserve special notice. Most importantly, unlike the prestations of cir described by Dumont, neither cir tanam nor cir varicai was reciprocated by the groom's

family. Nor was there any continuing cycle of cir exchange through the years, although the dowry was sometimes paid in installments. This is not to say that there were no continuing prestations between the families. But these were referred to by different terms, described below.

Cir tanam consisted of a large sum of money which technically remained the property of the bride. In practice, it was deposited in the business belonging to the groom's valavu or was used to open up a long-term, interest-paying deposit account (accimar panam ) in another Nakarattar family's business. In either case, the groom's valavu exercised considerable control over its use. However, if the bride died without giving birth to any children, the sum was returned to her family.

A similar arrangement was made concerning cir varicai . In this context, the bride's and the groom's families both maintained lists of the bridal gifts, called kalyanattukku caman vaitta vivaram . If the bride died without any issue, the gifts were returned to the bride's family. Prestations of cir varicai (varicai = "row on row") were quite similar to prestations of cir among the groups described by Dumont. They constituted a variety of items, including those for use by the bride, items for use by future children (including items reserved for the cir varicai of any daughters that might be born), and items for use by the groom (which would not be returned to the bride's family in the event of the childless death of the bride). The difference was that among the Nakarattar, cir varicai was given all at once on the occasion of the wedding itself.

Moi

Nakarattar applied the term moi to gifts given by any person attending a wedding. As with the Nangudi Vellalar, but unlike among the Kallar, Nakarattar moi were not confined to internal prestations, within the bride's and groom's respective pankalis . It was the case that these groups were ritually marked in the presentation of moi . But the marking was a complicated affair involving a variety of groups both inside and outside the couple's affinal kindred. Nakarattar moi did not represent any complementary gifting to the married couple, let alone a reciprocal exchange between the two separate sets of pankalis comprising affinal kindred. Nor was moi directed primarily toward providing the young couple with a substantial sum of money for use in establishing their pulli . Instead, moi generally took the form of symbolic tokens: a single rupee or the gift of a small amount of copper. These gifts were recorded in an account book (moippana etu ).

The order in which gifts of moi were entered reflected an internal segmentation of Nakarattar affinal kindred into five categories. The first

entry was that of the deceased father of the groom's oldest living patrilineal relative in direct line of descent (e.g., the groom's deceased FFF if his FF was still living). In this way the groom's valavu —the joint family or minimal lineage segment, which was midway between his kuttikkira pankali and his pulli —was given special recognition. The gift, however, was normally only one rupee and thus was symbolic rather than substantial.

After this entry, the names of all members of the groom's kuttikkira pankali were listed in genealogical order, whether or not they attended. Each attending pankali gave a small amount of copper—another symbolic gift—and a mark was placed by his name. By marking off the names of the attending pankalis from a comprehensive list in the moippana etu , Nakarattars maintained a record of the current state of solidarity and cooperation among different segments of their lineage.

The bride's uterine pankalis were listed after the groom's attending pankalis , as were relatives of the bride through her mother's line. In short, the couple's tayatis comprised the next group listed in the moippana etu . Nakarattars recognized lineal distinctions between the couple's tayatis by the order in which each was listed. The wife's father (amman or mamakkarar ), who was first on the list of tayatis , gave a substantial amount (I did not collect information about standards for judging its substantiality). Other tayatis of the couple gave symbolic tokens.

The fourth group in the moippana etu were termed catunkukkarar (ceremony doers). They consisted of the groom's sisters (nattanars ), who were entitled to certain rights, duties, and ceremonies during the wedding ceremony and for whom the groom and his sons acted (or would eventually act) as tayatis , observing cir -giving and murai -giving obligations for the nattanars' pulli .

Finally, there was a residual category of moi -givers that consisted of anybody else who attended the wedding. This group could even include friends or employees who did not belong to the Nakarattar caste.

Murai

Murai constituted customary and continuing external gifts from the wife's family to the husband's family. They partially corresponded to Dumont's description of cir among non-Nakarattar groups, and complemented Nakarattar cir tanam and cir varicai . However, they continued beyond the immediate occasion of the wedding. Like the two Nakarattar cir prestations, murai constituted asymmetrical prestations from the wife's family to the husband's. The primary kin groups involved were her murai -giving uterine valavu and her murai -receiving affinal pulli (i.e., the wife's father's pulli and the pulli formed by her marriage to her hus-

band). The groups of kuttikkira pankalis and tayatis to which these smaller kin groups belonged were indirectly allied through the alliance of their children. Their token moi prestations and their optional presence on the occasion of murai prestations ritually distinguished their status from the directly allied valavus and pullis .

Occasions for giving murai included seasonal festivals such as those occurring on the Hindu holidays of Pongal or Deepavali. Nakarattars also gave murai on the occasion of the wife's father's sixtieth birthday (shastiaptapurti santi kalyanam ), by which time he should have determined the eventual division of his properties (if he had not already divided them) and should have satisfied all dowry obligations for any other daughters. Finally, Nakarattars presented murai to the husband's family on the occasion of any other marriage occurring in their families.

The obligation to give murai on these occasions was taken seriously. Even if there was a quarrel, say, between a woman and her brother, such that he refused to attend her husband's sixtieth birthday celebration, he was still obligated to send murai through an intermediary and to cover the cost of a new gold marriage pendant (tali ) and sari for his sister, a veshti for her husband, and garlands and other auspicious articles for the ceremony.

The limits of murai obligations were especially significant for indicating the duration of a marriage alliance. Murai obligations between a woman's uterine valavu and her affinal pulli held force for the life of the married couple and of their unmarried children. Initially, murais were given by a married woman's father (her husband's amman or mamakkarar ). But even after her father's death, murai continued to be given by her brothers (her husband's attan or maittunan ). If she died before her husband and children, her father or her brothers continued to present murai on appropriate occasions. Finally, even if both husband and wife died, surviving members of the wife's uterine pulli continued to present murai to her surviving children until they married and formed a new pulli .

When a woman's sons married, her father and sons performed a modest ceremonial role at the wedding. But the major cir -giving and murai -giving responsibilities were assumed by the new bride's father's pulli . In principle, it was possible for a woman's son to marry her brother's daughter. In this case, the original marriage alliance between her uterine family and her husband's family was renewed, and the flow of prestations continued unaltered. But there was no requirement to renew an alliance in this way, and it happened infrequently. My informants felt that perhaps 90 to 95 percent of the time, marriages took place with families who were not already in-laws (campantippuram ). When a woman's daughters married,

her husband and sons assumed murai obligations for their in-laws. But a major portion of their obligations was absorbed by the mother's own cir tanam and murai , deposited for the purpose in an accimar panam trust. She and her husband and sons made use of the bulk of this money to cover the cost of her daughters' cir and murai .

A marriage alliance between descent groups—at least in the sense in which it was symbolized by murai prestations—thus had a kind of developmental cycle of its own, which lasted about two generations. Longer-term alliances between Nakarattar descent groups were possible, in principle, and were effected by arranging for marriage between first cross-cousins (i.e., FZD=MBS or MBD=FZS). But the incidence of such renewals was quite scarce. I consider the implications of the infrequent exercise of this option to renew marriage alliances between valavus (i.e., marriage of first cross-cousins) in subsequent sections of this chapter.

Vevu

Vevu constituted prestations of special ceremonial murai given by all pankalis on important life-cycle ceremonies for family members including the initial wedding, the birth of a first child (wealthy families gave vevu for the first boy and the first girl), the first menses of the first daughter, and the sixtieth birthday of the father. The difference between vevu and murai was that occasions for giving vevu required the physical presence of pankalis and tayatis (not just a representative or messenger of the wife's father or brother). In addition, vevu was given on exceptional ceremonial occasions. In contrast, murai were given on a periodic and continuous basis and could be sent by an intermediary.

Finally, Nakarattar tayatis were also responsible for a special form of funerary vevu called pirantu itattukkoti "bringing unbleached (ceremonial) cloth from the house of birth (of the wife)." When a man or woman died, the wife's father or brothers, accompanied by their pankalis , brought appropriate funeral garb for the deceased, along with sandalwood and other items required for the funeral pyre. They also supplied baskets of rice and vegetables to feed the mourners.

Summary of Affinal Prestations

The preceding observations highlight significant differences between Nakarattar and non-Nakarattar expressions of marriage alliance, especially with regard to the Pramalai Kallar described by Dumont. In the case of the Kallar, external prestations (all termed cir ) were initially symmetrical between husband's and wife's pankalis . Over the course of time, they became increasingly asymmetric until, in the end, the bride's family bore

an enormously disproportionate share of the burden (1957b: 32). Yet Kallar marriage alliances made a symbolic show of reciprocity through the initial exchange of cir . Thus, even though subsequent prestations tilted in favor of the husband's family, members of a Kallar marriage alliance could still point out that cir had been exchanged on both sides.

Among the Nakarattar, cir tanam, cir varicai, murai, vevu , and the funerary prestations of pirantu ituttukkoti together corresponded to the single category of Kallar external prestations. Two differences stand out. Firstly, Nakarattar external prestations exhibited greater qualitative variety, marked by an elaborated terminological system. Secondly, this variety was associated with a pronounced asymmetry in the direction of prestations. The uniform matrilateral asymmetry of Nakarattar prestations and the nonuniformity and variety of their classification combined to produce a mirror image of the Kallar pattern. Nakarattar alliances were not balanced by even initial reciprocal prestations of cir . All of their different external prestations flowed asymmetrically from the wife's family to the groom's.^[16]

Nakarattar and Kallar castes also differed with respect to what Dumont called "internal prestations," or at least with respect to prestations termed moi . In the case of the Kallar, moi defined two separate groups. Prestations were made within ("internal" to) lineage segments, each of which made substantial contributions to the member family whose son or daughter was married. In the case of the Nakarattar, individuals from different descent groups of the entire affinal kindred all gave token moi directly to the newly married couple. The different subgroupings of donors were distinguished by the order and classification of names on the moippanu etu . But all of them coalesced as a single group, allied by the common focus of its members on the married couple. The symbolic differentiation of status between categories of moi -givers reflected degrees of social distance between the couple's kin (see Figure 13) and corresponded to differences in access to credit generated by a woman's accimar panam deposits. But this single graded structure was markedly different from the bifurcated structure generated by moi -giving practices among the Kallar segments, each of which made substantial contributions to the member family whose son or daughter was married.

Positive Marriage Rules and Virtual Affinity

According to Dumont (passim ), all intermarrying descent groups in South Indian castes classify one another as either marriageable (affines ) or as nonmarriageable (kin). In addition, affines are themselves classified into two subcategories: (a) perfect affines , whose affinity is expressed by an

existing bond of marriage between their respective descent groups, and (2) virtual affines , whose affinity is expressed only by their terminological classification (1957b: 25, 27). Perfect affinity exists only when there is an ongoing alliance between two descent groups. Virtual affinity exists when there is no such alliance. Virtual affinity may even exist when there has never been an alliance. In such cases, Dravidians employ (and when queried, agree that they employ) a recursive algorithm: "Kin of my kin are kin; affines of my kin are affines; affines of my affines are kin." Dumont's distinction corresponds to the Nakarattar classification of, on one hand, campantippuram (in-laws) and, on the other hand, both tayatis (in the broad sense, meaning members of previously allied descent groups) and also all marriageable non-kin (see Figure 13).

The distinction between perfect and virtual affines is parallel to Leach's (1961) distinction between local descent groups and descent lines . Both distinctions follow from careful analyses of the social organization of specific kinship systems and contain a lesson worth reviewing in the context of any subsequent analysis: namely, that it is important to distinguish between a system of classification and a set of social groups. In Leach's words,

The notion of a local line is to be distinguished from the parallel concept of a descent line (line of descent) which has frequently been used by Radcliffe-Brown and his pupils. Descent lines have nothing whatever to do with local grouping, they are merely a diagrammatic device for displaying the categories of the kinship system in relation to a central individual called Ego. The number of basic descent lines in such a diagram depends merely upon how many different kinds of relative are recognized in the grandfather's generation. It has nothing to do with the number of local descent groups existing in the society. (1961: 57).

After forty years, anthropologists continue to have difficulty maintaining these distinctions.^[17] And it has proven especially difficult to maintain them in the context of analyses of affinal relationships.^[18] The biggest difficulty arises due to a misconception about the operation of prescriptive marriage rules, such as the putative Dravidian marriage rule "Thy children must marry affines." According to Dumont, for example,

The regulation causes marriage to be transmitted much as membership in the descent group is transmitted. With it, marriage acquires a diachronic dimension, it becomes an institution enduring from generation to generation, which I therefore call "marriage alliance" or simply "alliance."

... [In South India], sons of affines are ipso facto affines, at least in a virtual or rather a general sense, before or without becoming so individually.

I submit that, in societies where there are (positive) marriage regulations: (1) marriage should be considered as part of a marriage alliance institution running through generations; (2) the concept of affinity should be extended so as to include not only immediate, individual relationships (affines in the ordinary sense) but also the people who inherit such a relationship from their parents, those who share it as siblings of the individual affines, etc.; (3) there is likely to be an affinal content in terms which are generally considered to connote consanguinity or "genealogical" relationships (such as mother's brother, etc.). (1957b: 25)

In this passage, Dumont talks first of the transmission of marriage, not affinity. He then asserts that relationships of affinity apply to ("should be extended to") not only perfect affines, who have actually established a marriage alliance, but also virtual affines, who are not married, who may never marry, and who are, in fact, only marriageable. In so doing, Dumont sacrifices the distinction between perfect and virtual affines in order to argue the central theme of his essay, that affinity is an important principle of Dravidian kinship. Yet, granting that affinity is an independent principle, in what does it consist? Apparently, all that is truly common to perfect and virtual affines is the marriageability of their children. They share none of the other relationships or obligations that hold between families allied together by marriage as affinal kindred.

If this is the case, it raises a serious problem for Dumont's thesis, since marriageability—even terminologically marked marriageability—is not a consequence of any positive marriage rule. To see this, consider the case of Dumont's Pramalai Kallar, who, unlike the Nakarattar, do seem to subscribe (at least partially) to a rule prescribing marriage of the eldest son with the mother's brother's daughter (MBD), and not just with any woman also designated by the same term (Dumont 1957b: 14). When the rule is followed, marriage alliances are established between distinct Kallar descent groups that persist from generation to generation. Are similar alliance relationships established between the descent groups of virtual affines? No. The defining feature of virtual affinity is precisely its lack of a real alliance. Virtual affinity establishes no relationship between specific groups, but only the potential for a relationship.

Positive marriage rules only make sense in the case of perfect affines—that is, between bounded descent groups which maintain a perpetual relationship of alliance between them. What would it mean to say that there

is a positive marriage rule requiring marriage with a marriageable category whose "virtually married" members number hundreds or even thousands of different descent groups? Virtually nothing. No enduring relationship need be generated between any two groups of virtual affines from marriage to marriage because there is no marriage.

There is only one common feature besides marriageability that is shared by Dravidian kin whom Dumont classifies as perfect or virtual affines. This additional feature is nothing less than the misinterpreted formal aspect of marriageability: that is, its terminological marking by identical kinship terms. But it is a non sequitur to argue that applicability of a common term, signifying marriageability, eliminates any distinction between the two otherwise separate categories.

Marriage alliance between perfect affines may follow from adherence to a positive marriage rule. But it may also result from a variety of pragmatic considerations. The Pramalai Kallars illustrate the former case. Nakarattars illustrate the latter: they do not maintain any positive marriage rule for allied perfect affines. In neither caste is there either rule or alliance between virtual affines.

I suspect that the temptation to conflate perfect and virtual affines actually arises from a conflation of semantic and jural rules. The former classify kin categories in a semantic domain defined by descent and marriage. The latter stipulate rules for behavior between members of actively allied kin groups. The two sets of rules are quite distinct and independent. Thus, Nakarattars share the terminological marking of affines (both perfect and virtual) with Kallars. But they do not stipulate rules for perpetuating marriage alliance, even among perfect affines.

Ironically, it should be easier to recognize the distinction between semantic and jural rules among the Kallar than among the Nakarattar. Kallars maintain two sets of jural rules that apply differentially to the terminologically unmarked categories of perfect and virtual affines. The composite category of Kallar affines is subject to rules stipulating jural marriageability. But only (apparently unmarked) perfect affines are subject to normative rights and duties—termed uravin murai —that perpetuate marriage alliance between pairs of descent groups.

In-Laws, Murai , and Terminological Marking

In fact, among many South Indian castes, including the Nakarattar, perfect affines were distinguished from virtual affines in discussing ceremonial occasions where the affine was a male in-law with obligations to give gifts of cir or murai to the married couple (see above). In these cases, the normal kin term designating the kinsperson's relation to the groom as a virtual

affine was modified or replaced by an alternative term that indicated his change of status from a virtual affine to a perfect affine. The Nakarattar, for example, exhibited the following usages: amman versus mamakkarar (for MB vs. WF) and attan versus maittunan (for MBS vs. ZH). I address aspects of the precise import of this status marking immediately below. In the present context, however, all that matters is that a terminological marker for perfect affines with cir -giving or murai -giving obligations designated a discrete social identity subject to positive jural rules.^[19]

Among the Nakarattar, the marked status of a cir - or murai -giving, perfect affine did not carry any additional obligation to renew the marriage alliance as stipulated by a positive marriage rule. That is, although perfect and virtual affines were distinguished by the presence of murai obligations among the former, they were similar in the absence of any positive marriage rule. This property of the Nakarattar variant of Dravidian kinship raises two further questions in regard to Dumont's theories. In the first place, did the restriction of terminological markers in these contexts just to perfect affines with murai -giving obligations indicate that virtual affines who did not give murai were therefore not affines in any sense? The question is implied by Dumont himself, whose argument that affinity and marriage alliance are synonymous is designed to combat any doubts, either about the "affinity" of virtual affines or about the "independence" of affinity from consanguinity. But this argument is both dubious and unwarranted. Not only were relationships of marriageability between kin independent of consanguineal and alliance ties, but the only times when there was no incidental overlapping of consanguineal and affinal categories were precisely those cases where no real marriage had occurred—that is, precisely in cases of virtual affinity between consanguineously unrelated members of the same endogamous caste. Conversely, marriage ceremonies performed precisely the functions of conferring consanguineal status on otherwise nonconsanguineal affines and conferring perfect affinal status on otherwise merely virtual affines, regardless of their prior status as kin.

The Maternal Uncle, the Wife's Father, and Murai

A second question arises in consideration of an obscure component of the elaborate ceremonies performed at a Nakarattar wedding.^[20] According to one of my informants, the wife's father (amman, mamakkarar ) enacted a distinctive role in these ceremonies that seems to have expressed the right of his family (the couple's tayatis ) to renew the alliance by marrying future children in their family to any subsequent children born in the groom's pulli . That is, according to this informant, murai obligations were

morally linked to the murai -giver's right to enforce an alliance-renewing marriage rule.

During a wedding, the mamakkarar stands in attendance throughout the marriage and is within early reach. He wears a ceremonial uniform with a gold-laced shawl in reddish color around his waist or belly. From the moment the bride groom leaves his home for the marriage celebrations to the bride's place and until the bridal couple's ceremonial setting foot at their new home, the mamakkarar leads the bride or the bridegroom at every step. He has to lead the way, to guide the bride or groom, to help the groom to mount up and dismount from the horse, to direct him to the prayers of the temple and to assist the Brahman priest in the rituals. If the mamakkarar family has a child of the opposite sex and suitable in age, to match the sister's child, marriage is contracted amongst them. If for any reason such a marriage is not consummated, the parties may agree to disagree. It is only as proof of this "no objection certificate" that the presence of the mamakkarar is insisted on throughout. In other words, a mamakkarar has the right of first refusal.

Here, apparently, is concrete evidence of a positive marriage rule operating in association with a rule about murai prestations. Both rules apply to two specific kin categories, WF and HF. The marriage rule, in particular, apparently generates the WF's right to renew a marriage alliance between two families in a way that is similar to marriage rules called uravin murai among Kallars, Nadars, and other South Indian castes. In the face of such a cogent and forceful opinion, it is important to note that this was the only expression even resembling uravin murai rules that I discovered among the Nakarattar. I heard no reports, for example, of sanction-free rights to "kidnap" a bride by tayati groups, as was reported about some Kallar groups. Nor did Nakarattars refer to their MBD as urimai pen (rightful woman) as was the case among representatives of Nadar and Kallar groups from whom I gathered information. It is the case that my informant successfully established a claim on the son of his in-laws' family as a groom for his second daughter (see comments in Table 12). But I have no indication that this alliance might be renewed in any subsequent generation. Moreover, as already noted, the incidence of multigenerational alliances was extremely low among all Nakarattars, and it may be that my informant (who could be an extremely persuasive negotiator) had managed to create a nonstandard "right" in the process of arranging the marriage (see Linda May, 1985). This lone exception aside, it had not occurred to any of my other Nakarattar informants that their caste maintained uravin murai . On the contrary, even those familiar with its operation denied that it operated among the Nakarattar.

This is significant. Positive rules enjoining asymmetrical, matrilateral cross-cousin marriage seem to have played a role in the social organization of nonmercantile castes such as the Kallars. The implication of my informant's claims is that his exercise of uravin murai rights represented an expression of a positive marriage rule among Nakarattars, although one that compelled symmetrical, bilateral cross-cousin marriage. If my informant was accurate about present-day Nakarattars, then it would not be unreasonable to assume that their ancestors also subscribed to such a rule. But in this case my informant's opinion is suspect. Whatever may be the case among non-Nakarattar castes, the Nakarattar expression of the rule—if expressed at all—is so feeble as to have no discernible effect, either manifest or latent.

The issue of possible linkage between murai obligations and marriage rules lies at the core of Dumont's argument about the self-renewing or perpetual character of Dravidian marriage alliance. According to Dumont, obligations to give external gifts such as cir or murai between allied descent groups symbolize the operation of a positive marriage rule. He provides no explicit rationale for this assertion. But he does offer several examples in which a mother's brother is replaced in his role as gift-giver by his sons or other agnates (1957b: 89–90). The implication is that uravin murai rights to renew the alliance are associated with obligations to give gifts; since the latter, at least, seem to be inherited by subsequent generations, so must be the former.

As with Dumont's non-Nakarattar groups, Nakarattars also engaged in a continuing series of external prestations from tayatis representatives to their daughter's (or sister's) pulli . Moreover, as with Dumont's non-Nakarattar groups, Nakarattar murai -givers replaced one another as representatives of the obligated tayatis . In the Nakarattar case, however, gifts and personnel replacements did not extend beyond the second generation. They had a distinct termination point, marked by the wedding of the final child born to the pulli that initiated the alliance. Either party had an option to renew the alliance. But they were not obligated to do so. Indeed, perhaps 95 percent of the time, the alliance was relinquished, and a different group established tayati obligations. In other words, although Nakarattar marriage alliance had what Dumont calls a diachronic quality, it was not therefore perpetual and self-renewing.^[21]

A Note on Ethnographic Reports of Nakarattar Marriage Rules

If (contrary to my informant) Nakarattars did not subscribe to a positive marriage rule enjoining symmetrical cross-cousin marriage, did they

subscribe to any other positive marriage rule? According to Thurston (1909 V: 265), Nakarattars did follow a positive rule for asymmetrical, patrilateral cross-cousin marriage. In the ideal exemplification, a man married his father's sister's daughter (FZD), a perfect affine in the Dravidian system. That is, an initial alliance in one generation stimulated a second alliance in the succeeding generation. Compliance with this rule would have precluded both bilateral cross-cousin marriage and matrilateral cross-cousin marriage.

Thurston's report seems to be confirmed by similar observations that contrast matrilateral cross-cousin marriage among landholding castes and patrilateral cross-cousin marriages among mercantile castes in the Kongu region of Coimbatore, located northwest of Chettinad. According to Brenda Beck (1972), who provided this report, the contrast is associated with the presence of status hierarchy among descent groups within landholding castes and the absence of descent groups (and hence, the absence of hierarchical relations between descent groups) within mercantile castes. The issue is somewhat complicated by an additional, associated contrast between the kinship term systems of these groups. But, in the end, Beck interprets the various differences with reference to Lévi-Straussian theories about the elementary structures of kinship.

The [landholding] sub-castes make no distinction between the terms for mother's brother and father's sister's husband, nor between terms for matrilateral and patrilateral cousins. The [mercantile] sub-castes, by contrast, use slightly inferior terms for matrilateral uncles and cousins, while they reserve slightly superior terms for their patrilateral equivalents. This tendency to distinguish between wife-givers and wife-takers is associated with their special emphasis on dowry and their determination that the receiver of the bride be superior. This does not contradict their patrilateral cross-cousin marriage preference when it exists, however; for the superior-inferior distinction is easily reversed in succeeding generations. This is so because among the mercantile sub-castes, clan lines are weak. The hierarchy of individual givers and takers rather than that of whole descent groups is considered of prime importance. (Beck 1972: 13)

Having read Thurston and Beck before arriving in India, I was looking forward to analyzing the functional consequences of Nakarattar patrilateral marriage rules, a kin term system that marked wife-givers and wife-receivers, and an absence of clanlike descent groups. Surely, I thought, Beck is correct. This is a mercantile transformation of Dravidian kinship, which must have some practical consequences for Nakarattar commercial organization.

As we shall see in the following chapter, there is evidence in Nakarattar descent and territorial organization of an absence of hierarchy. But I am now quite convinced that, whatever situation obtains in Kongu, the Nakarattar differed from Beck's description of mercantile castes. Nakarattars maintained a nested organization of descent groups including clans (although these were territorially dispersed, and this dispersal, I suspect, is what Beck was really talking about). I am also convinced that Thurston was misinformed about the presence of a marriage rule prescribing patrilateral cross-cousin marriage. There was some kind of status marking going on in their kinship term system. But it was a marking that designated murai obligations, not a lateral preference for marriage.^[22] Finally, there is no evidence that Nakarattars were constrained by any marriage rule, not even a rule applicable just to perfect affines and requiring marriage only to a "real" FZD. Although active marriage alliances between Nakarattar descent groups (campatippuram relations) were occasionally renewed by the marriage of first cross-cousins, my impression is that there was a slight preference for MBD marriage over FZD marriage. Moreover, when gently queried, when asked overtly leading questions, when chivied, harassed, and pursued beyond the bounds of civility (for I was determined to prove the existence of a FZD marriage rule), my Nakarattar informants insisted that the whole idea was silly.

Final Comment

In this chapter, we have examined the operation of Nakarattar affinity and marriage alliance. I noted that the inclusive or nested relationships of Nakarattar descent groups (ranging from hearthhold to joint family to local lineage segment to clan) generated a correspondingly multiplex set of affinal ties. But these affinal ties exemplified characteristic structural properties unlike those exemplified by non-Nakarattar castes. In particular, they contributed functionally to the organization of Nakarattar commerce by providing mechanisms for accumulating liquid resources, for undertaking cooperative ventures with limited liability, and for participating in numerous occasions for the exchange of commercial information. In the following chapter, we shall see that Nakarattar descent organization, like Nakarattar affinal organization, exemplified a characteristic mercantile adaptation. In particular, we shall see that the Nakarattars lacked the trappings of hierarchy associated with kinship organization in nonmercantile castes: namely, relations of blood purity, territorial precedence, or royal honors. We shall also see that the Nakarattar descent organization differed from that of non-Nakarattar castes with respect to a pattern of crosscutting agnatic relationships that tied together Nakarattar residential villages throughout Chettinad.