1. These population estimates come from three different sources. The 1896 figure is from an article by P. R. Sundara Iyer published in 1906 in the Indian Review and cited by Krishnan (1959: 31). Pillai's (1930) estimate of the Nakarattar population in 1920 is based on a caste census published in the Nakarattar journal Oolian in 1921-22. The 1980 population figure is based on a 1980 caste census kindly supplied to me by the Madurai Nakarattar Sangam (the Madurai Nakarattar Association).

2. The absence of the caste from the historical record before 1600 is a rather controversial point, though one beyond the scope of the present study. Many scholars point to medieval inscriptional uses of the term nakarattar and its variants, asserting that such uses refer to a specific caste who formed the ancestors of the contemporary Nattukottai Chettiars. I find it much more likely that medieval use of the term nakarattar refers variously to merchant guilds, to municipal organizations in market towns ( nakaram ), or even simply to the residents of market towns (see Rudner 1985, 1987).

3. In addition to the literature on Nakarattar history and customs cited in the text, I have consulted Chandrasekhar (1980), Cooper (1959), Evers (1972), Manickam (1978), Masters (1957), Nadarajan (1966), Naidu (1941), Ito (1966), and Thomas (1941). Various Nakarattar writers have also been very helpful, especially writers mentioned in the discussion of traditional caste histories ( jati varalaru ) in this chapter.

4. See Baker 1984: Chapter 4, especially pp. 281-290, on Nakarattar commercial practices within Madras.

5.  Dumont (1977, 1980 [1970], 1986) characterizes Hindu ideology as a hierarchical system for evaluating ("valorizing") persons in terms of rela-

tive purity. In this characterization, he trades on an ambiguous use of hierarchy as denoting at least two concurrent relationships between members of a social group: ritual rank and segmentary inclusion. That is, Dumont asserts that, according to Hindu ideology, every person is related to every other person in a group in two ways. In the first case, every person has the property of being more or less pure than every other person, and no person is knowable independent of his place within a system of ranked relationships. In the second case, every person has the property of being a member of several social groups, each one of which is related to every other group within a single, logically inclusive taxonomic system. The first component in Dumont's definition of hierarchy—the attribution of relative purity to all members of any social group—defines a serial ordering of every person in any social group; that is, it defines a relationship between any two persons that is irreflexive, asymmetrical, transitive, and connected (R. Rudner 1966: 36). The second component in Dumont's definition of hierarchy—the attribution of segmentary membership in an inclusive taxonomy of social groups—defines a complex, connected, but nonserial ordering for every person in any social group, one that defines relationships for which it would be possible to measure the degree of similarity of persons in terms of kinship distance defined by ideas about descent (see, e.g., Evans-Pritchard 1940).

Dumont claims that societies whose members act in conformance with this kind of dual ideological ordering are "holistic"; he implies that any society whose members act in conformance with a different ideology is individualistic. Dumont seems to mean by this that holistic characterizations of persons exemplified by Hindu ideological expressions refer only to relational properties. For example, every Hindu characterization of a person will relate at least two or more persons in statements containing two place predicates such as " A is more pure than B " and '' A is a member of B , where B is a group of persons, including or included by another group of which A is also a member." By contrast, Dumont seems to claim that societies whose members conform to nonholistic, individualistic ideologies exemplified by Western democratic capitalism characterize each other only with nonrelational, single-place predicates, as in the statement " A owns 20 acres of land."

Such generalizations about civilizational ideologies are highly oversimplified and, I would argue, are shaped by and contribute to the essentialism, exoticism, and totalism that mark much of Dumont's comparative sociology (see Appadurai 1986a).

6. Marriott has greatly elaborated his ethnosociological approach to Indian culture during the sixteen years since publication of his 1976 essay, but until recently has published few of his results. Nevertheless, gleanings about the direction of his own continuing inquiry may be found in the work of his students, such as E. V. Daniel, A. Gold, M. Moreno, D. Mines,

M. Moore, and G. Raheja. See the special ethnosociology issue of Contributions to Indian Sociology (1989, reprinted as Marriott 1990). See also further commentary in Contributions to Indian Sociology (1990).

7. Dirks (1988), in a forceful indictment of the orientalist image of India, perhaps overstates the hegemony of orientalism in Western social science. Anthropologists have long been concerned with the exercise and organization of political power, whether examining the dynamics of dominant castes (Mayer 1958; Srinivas 1959) or analyzing the structure of "little kingdoms" (a phrase that Dirks himself borrows from Cohn 1962). Even Dumont, whom Dirks holds out as the champion of kingless India, does not deny the existence of kings or other holders of power so much as he argues that they legitimized their authority by subordinating it to Brahmanical ideology (1958: 45-63, 1962: 48-77, 1980 [1970]: 152-184). Such arguments can be and have been criticized for ignoring the kingly ideologies of raja dharma (Derret 1976; Spellman 1964). But they cannot accurately be said to deny India any indigenous institutions of governance or ideologies of power and authority. Nor is it the case, as Dirks asserts, that Dumont's study of the Pramalai Kallar portrays them "as examples of a ritually marginal group that exemplified the Dravidian isolation of kinship from the influence of caste hierarchy" (1988: 5-6, my emphasis). On the contrary, it is precisely Dumont's point that not even kinship is immune to the pervasive force of a Hindu hierarchicizing ideology. Dumont is clear about this in his original ethnography (1957a) and in his theoretical masterpiece (1980 [1970]). But the reader is also referred to his essay (1957b, reprinted in Dumont 1983) "Hierarchy and marriage alliance in South Indian kinship." Neither in this essay nor in Un Sous-Caste de l'Inde du Sud (1957a) does Dumont articulate his theory of the Brahmanical underpinnings of hierarchy. Indeed, in a move quite close to Dirks' own argument, he actually locates its cultural basis in ideas about kingship. But his emphasis on the pervasive intrusion of hierarchy into the domains of caste and kinship remains constant throughout the corpus of his work (see Chapter 9 of this book). The interpretation of Dumont, however, is somewhat at a tangent to my immediate concerns. My argument with Dirks in this footnote concerns his reading of anthropological history, not his interpretation of Pudukottai Kallars and the institutions of South India's ancien régime.

1. In recent years, the standard model of caste has been complicated by the findings of a structural study of a mercantile/agrarian division of castes in South India (Beck 1972) and by ethnosociological studies (Marriott 1976, 1990) that, although focused on variations in Hindu strategic interaction, produced findings that implied corresponding variation

among castes whose members follow divergent strategies. Such studies have influenced some of my own revisionary efforts in the present book. See my comparison of Kaikkolars and Nakarattars in the present chapter and also in Chapter 10.

2. Baker's (1976) treatment of late nineteenth- and early twentieth-century caste politics diverges from his (1984) treatment of precolonial castes. To my knowledge, he has not drawn any connection between caste organization in precolonial and colonial South India.

3. Complications of terminology combined with questions of fact render issues related to the nature of caste even more complex than is indicated in the text. This complexity is particularly clear in the case of the Nadar Mahajana Sangam (the Nadar Association) studied by Robert Hardgrave (1969). Hardgrave treats what in my terminology is actually a caste cluster as a single caste. On one hand, as Baker and Washbrook (1975) point out—and their view is supported by my own field inquiries —it is questionable to what extent all of the Nadar castes constituting the Nadar caste cluster ever organized themselves even in a purely modern political fashion, let alone whether they interacted with respect to occupational, affinal, and martial considerations. On the other hand, discrete Nadar subcastes do seem to have operated as corporate groups, with respect to a variety of considerations, over a wide regional territory. The whole issue requires a new investigation—one, moreover, that should be speedily undertaken, as many contemporary Nadar informants (especially the educated youth) take Hardgrave's book as an authoritative source for knowledge of their own caste!

4. I elaborate my use of the term symbolic capital in Chapter 10.

5. In his more recent work, Washbrook (1984) seems to have modified his position on precolonial caste identity. If I follow him correctly, he has come to equate "substantialization" with endogamy, and to claim that castes have always been "substantialized" on the level of the localized endogamous jati . What is new, in Washbrook's view, is the "brahmanization" of jatis : a pan-regional evaluation not simply of local jatis but of jati cluster status, based on applications of varnashrama doctrine that are novel in the South Indian context. According to Washbrook, these doctrines came to be regarded as legal precedent by colonial jurists and were enforced by colonial administrators. This "invention of tradition" kind of argument (cf. Hobsbawm and Ranger 1983) is far more radical than the earlier versions of Washbrook's position described in my text, above. While there is considerable merit to such an argument—particularly with regard to his insights about the impact of the colonial legal system—it is important to note that Washbrook has altered the use of the term "substantialization'' from its original application by Dumont (1970) and Barnett (1973). It is also important to note that such a radical view has difficulty accounting for the position of Brahmans in South Indian society and

also for ideological appeals to the varna system, both of which extend back to at least the twelfth century (for examples of such appeals, see Subrahmanya Aiyer 1954-56). Accordingly, it still seems preferable to speak of the colonially stimulated substantialization of Brahmanic doctrine.

6. Political support for an elite caste member can be expressed in a variety of ways, such as civil disobedience, migration, religious conversion, voting, and so on.

7. Srinivas uses the metaphor of the djinn of caste in several places (see, e.g., Srinivas 1962: 16).

8. It is also worth noting that Nakarattars, who share many aspects of the commercial adaptation that Washbrook claims are preadaptive for ethnicization, have proven more resistent to assimilation into ethnicized clusters of Chetti castes than have other, more localized Chetti subcastes (e.g., those belonging to the Ariya Vaisya and Aiyera Vaishya Chetti clusters).

9. Indeed, many possible objections to my interpretation of caste as symbolic capital may follow less from disagreement over empirical facts than from disagreement over which elements in the definition of a corporation are to be stressed. For my purposes, Maine's (1931: 154) notion of an enduring jural person with rights over property and Weber's (1947: 145) concept of a verband , with its reference to jural closure of membership and a definite authority structure, serve as the most useful points of departure. I would, however, qualify this synthetic view with the further, anthropologically conditioned thought that corporate authority need not be vested in a chief or panchayat , but may be allocated by the acephalous operation of other forms of political organization, such as a segmentary lineage (Fortes 1954). I employ such a maximally general definition in order to facilitate comparison both between differently organized castes and between castes and other kinds of corporate and noncorporate groups. Also relevant are Wolf's (1955) and Fried's (1957) interest in village communities as corporate groups, possessing joint ownership over land, and Mayer's (1966) and Boissevain's (1968) distinction between quasi groups, or coalitions, and corporate groups, on the basis of temporal continuity. For further discussion of definitional problems in uses of the term corporation see Appell (1983, 1984), Brown (1976, 1984), Cochrane (1971), Goodenough (1971), and Smith (1974).

10. For an analysis of the social organization and commercial activity of "petty" Chetti castes (involved only in small-scale trading), see Fanselow (1986). For North Indian examples, see Fox (1969) and Hazelhurst (1966).

11. In commercial territory, Nakarattars resemble most closely and compete in scale with such North Indian jatis as Marwaris, who by contrast tended to confine their business ventures to India itself (Timberg 1978). They also resemble the smaller-scale Kalladaikurichi Brahman of Tirunelveli, who conducted business primarily in Tranvancore and the region comprising present-day Kerala (Krishnan 1959; Madras Provincial

Banking Enquiry Committee 1930). The chief contrast that seems to emerge between these three groups of merchant-bankers is their geographic share of the Asian market.

12. In his contrast of the organization of Kaikkolars and of agrarian castes, Mines (1984) is building on the work of Beck (1972) and Stein (1980).

13. Kaikkolars once again reveal a similarity with Nadars, who also utilize local residence-cum-business sites like natus , which in their case are called pettais , from the term for a fortified market in which itinerant traders protected their goods and cattle and in which they engaged in trade (Vaidyalingam and Sundaramurthi interviews 1980-81). In contrast to that of Kaikkolars, however, Nadars' pan-regional commercial integration is a relatively recent phenomenon. Coordination of their pettais and other local business-cum-residence concentrations is accomplished only through branches of their own recently developed caste association. There is no Nadar organization comparable to the 72- natu system.

14. For discussion of various medieval merchant guilds, see Abraham (1988) and Hall (1980). Interestingly, the localized guildlike groups were called nakarattars! It was only after the eleventh century that nakarattar came to be applied to guild groups involved in extensive long-distance trade (see Rudner 1985).

1. Parts of this chapter were previously published in Rudner (1989).

2. For the classic statement of the nationalist argument, see Dutt (1950). For more modern versions and also for historians favoring one or another version of Western modernization theory, see Morris (1969).

3. Other noteworthy studies that do address Indian commercial institutions include Habib (1960, 1969, 1982), Leonard (1979), and Das Gupta and Pearson (1987).

4. For contemporary ethnographic accounts of capitalist exploitation of peasant farmers by moneylenders, see Breman (1974), Gough (1981), J. Harriss (1982), and Mencher (1978).

5. I derived the ratio of British to Indian credit by comparing Vakil and Muranjan's (1927: 532) figures for loans and investments made by Madras Presidency and British exchange banks with those made by Indian joint stock banks. If a comparison of total deposits is made, the ratio is even higher: eight to one.

6. "Statement Exhibiting the Moral and Material Progress and Condition of India during the Year 1901-02, and the Nine Preceding Years" (Great Britain, Parliamentary Papers 1903 XLVI: 146-148, quoted in Bagchi 1972: 62).

7. Evidence of the Nattukottai Nagarathars' Association and of C. A. C. Kasinathan Chettiar (MPBEC 1930 I: 101-118, quoted in Bagchi 1972: 207).

8. In fact, Bagchi's (1972) argument stresses the collusive cooperation between British economic interests and colonial political authorities. But within this argument he accepts the colonial vision of India's financial underdevelopment. The actual availability of credit to Indian entrepreneurs would in fact underscore his primary argument about British monopolization of opportunities for industrial investment (see also Ray 1979).

9. There are wide extremes of scholarly opinion regarding the exploitative nature of postulated jajmani systems. Some authors see them as ensuring a more or less egalitarian distribution of communal goods and services (Wiser 1936; Wiser and Wiser 1969). Others see jajmani as an overtly exploitative institution, operating with particular rigor against the interests of landless laboring castes (Breman 1974; Gould 1964; Orenstein 1962). For more recent "cultural" accounts, see Commander (1983), Fuller (1989), Good (1982), and Raheja (1988).

10. By excess profits I here mean those above and beyond the amount required for subsistence and interest payments on back loans.

11. As many as 90 percent of all loans to peasant farmers in Madras were made by a small number of wealthy peasants (see Nicholson 1895: 230; MPBEC 1930: 79). Washbrook (1981: 73) estimates that this rural elite made up no more than 2-4 percent of the landowning population. See also Baker (1984), Bayly (1981), Murton (1973), Musgrave (1978), and Washbrook (1973).

12. For analyses of Indian bazaars see, for example, Fanselow (1986); Fox (1969), Hazelhurst (1966), Mines (1972), and Ostor (1984). See also Geertz (1963, 1979) for two influential essays on bazaar economy outside the Indian context.

13. In other regards, the difference in focus between moneylenders in village settings and in market towns is significant. Barbara Harriss (1981), for example, sees village moneylenders as potential entrepreneurs, or at least as stimulating studies which portray them as entrepreneurs to the extent that they shift their investments away from villages and into towns. At the same time, Harriss finds that studies of money lending in market towns—that is, studies of bazaar economy—portray moneylenders not as entrepreneurs but as "powerless agents of stagnation" (B. Harriss 1981: 5-12).

14. For studies of Saurastra, see Spodek (1976). For Gujarat, see Das Gupta (1979), Dobbin (1972), Gopal (1976), Haynes (1991), Morris

(1965), Pearson (1976). For the Malabar Coast, see Das Gupta (1967). For the southern Coromandal, see Appadurai (1974), Arasaratnam (1986), Basu (1982), Brennig (1977), Furber (1951), Raychaudhuri (1962), Suntharalingam (1974). For Golconda, see Siddiqui (1956). For all of southern India, see Subrahmanyam (1990). For India in general, see Chaudhuri (1978, 1985).

15. I exclude recent work on commercial activity in the medieval period (ca. 900-1500), including Hall's (1980) work on marketing towns in Chola, South India, 900-1200; Abraham's (1988) work on medieval merchant guilds in the same period; and Mines' (1984) and Ramaswamy's (1985) work on medieval weaving communities. This important body of Indian historiography has obvious relevance for the arguments in the present book, but an adequate evaluation of its significance for the colonial period must await another opportunity.

16. Baker's (1984) study of colonial and independent Tamil economy would also seem to call for comment, since it focuses explicitly on economic issues. But the overall orientation of his work bears more on issues in microeconomics than on the sociocultural issues that inform Bayly's (1983), Subrahmanyam's (1990), and my own work. To the extent that he does address such issues, Baker (1984: 282) explicitly mentions collective Nakarattar institutions that facilitated their emergence as a major commercial group in Tamil Nadu. He speculates that such institutions evolved in response to nineteenth-century economic conditions. But he does not describe them in detail, and he does not describe the pre-nineteenth-century institutions from which they evolved.

17. In the text leading up to his conclusion, Bayly (1978) discusses two forms of caste organization: chaudhries of endogamous ritual units—"castes" or " jatis "—which, ''though they might consist of tradesmen were not chaudhries of trades," and chaudhries , which "exercised authority over both the economic and ritual life of trades where the 'guild' and 'caste' ( jati level) were coterminous" (181). Both of these organizations were integrated within a commercial system that also included organizations structured independently of caste.

18. Arasaratnam's (1971) paper on three Tamil kingdoms represents a partial exception to historiographic blindness beyond the coastal margins. But even here, the focus on interior regions of South India emphasizes political rather than commercial organization. More recently, Subrahmanyam (1990) has attempted to remedy the omission, giving special attention precisely to inland commerce. This exceptional work represents the most comprehensive study to date and goes a long way toward remedying the kinds of omissions I have been describing. It also provides evidence of the difficulty in obtaining data about inland commerce relative to the availability of data for the coastal areas—a difficulty that no doubt contributes to the historiographic bias for coastal studies. Thus, even when

Subrahmanyam focuses on inland commerce, the inland merchants he highlights figure prominently in Indo-European commerce and are characterized by important political ties to Indian kings and European trading companies. See, for example, his treatment of a powerful family of Balija Chettis (1990: 300-314).

19. On commercial actors before 1900, see, for example, Arasaratnam (1966, 1967, 1968, 1971, 1978, 1979, 1980, 1986), Brennig (1977), and Subrahmanyam (1990).

20. Note, however, that strain between political authorities and merchant groups was equally inevitable in purely Indian contexts, so much so that it plays a role in the origin myths of virtually every South Indian merchant caste. The Nakarattar are unexceptional in this regard.

21. My interpretation that there was little change in the dominant role of elite merchants since the seventeenth century differs from that of Arasaratnam (1979) and Brennig (1977), both of whom see the eighteenth-century dubash system as distinctly different from the preceding organization of merchants headed by "chief merchants."

22. See discussion of Washbrook (1975, 1984) on caste in the preceding chapter.

23. Moreover, Subrahmanyam's (1990) focus extends to Muslim merchants as well as Hindu merchants.

24. Subrahmanyam (1990: 298-300) aligns himself on this issue with Arasaratnam (1986) and a Dutch author, H. W. van Santen, against Pearson (1976, 1987) and Chaudhuri (1985). He does, however, qualify this position geographically, leaving open the possibility that the situation in Southeastern India differed from that in Southwestern India, as described by Pearson (1976).

1. For an overview of Indian political economy during this period, see Bayly (1988). For descriptions of South India's integration with the international market, see Arasaratnam (1966, 1967, 1979, 1986), Chaudhuri (1965, 1978, 1985), Das Gupta (1967), Furber (1951), Habib and Raychauduri (1982), Raychaudhuri (1962), Subrahmanyam (1990).

2. By 1565, Vijayanagar had lost all influence in the area. In 1607, Sadaika Deva Udayan won political recognition of his de facto independence from Madurai and was appointed (or "restored") as Setupati of Maravar country (present-day Ramnad District). One hundred years later, Sadaika Deva Udayan's descendant, Kilavan Setupati, had to contend with internal factions threatening to splinter his own realm. In 1708, Rangunatha, a vassal of Kilavan Setupati, attempted to detach Pudukottai from Maravar country and merge it with the territory of the Nayak of Tanjavur. Kilavan patched up this threat by marrying Rangunatha's sister

and giving him the title of Tondaiman (chief) of Pudukottai. But this move brought only a temporary peace to the territory of Ramnad. In 1733, the Nayak of Tanjavur took advantage of a turbulent dynastic succession to invade Maravar country, and he successfully engineered its partition. Tanjore acquired for itself considerable territory and divided the rest between one of the dynastic pretenders and another palaiyakkarrar who would later be established as the Raja of Sivaganga. In 1767, the continuing incursions of the Nayak of Tanjavur brought the Nawab of Arcot, who was ruler of Madurai and nominal overlord of Maravar country, into the fray. The Nawab allied himself first with the Dutch and then with the English. Ultimately, of course, the English emerged as the rulers of the entire region, at the turn of the nineteenth century, and proceeded to reorganize Madras and India under the most powerful centralized government they had ever known. For general overview see Stein (1969); see also Rajayyan (1974); Sathyanatha Aiyar (1924).

3. "There were four varieties of salt. The first was natural salt manufactured in the Salt Pans and the second was Swamp Salt or Spontaneous Salt which was manufactured during periods of high tides. Sea water was allowed to flow over the level earth, and return flow of water was prevented by quickly throwing up small banks, and the enclosed water was then allowed to evaporate. The third was earth salt produced by lixivation, while the fourth was edible salt which was a by-product, in the manufacture of salt-petre. Natural salt, salt-petre and salt fish were manufactured throughout the east coast. Earth salt was made in the districts of Bellary, Cuddappa and Kurnool. Earth salt was made in a few villages on the sea coast of Narsapur, Bhimavaram, Amalapuram and Vizag. Manufacture of salt was associated with two other industries (1) Salt Petre (2) Salt Fish" (Krishna Rao 1964: 63).

Krishna Rao cites the Report of Godavari Collector 1844 as the source of this information. I assume similar practices were prevalent during the seventeenth century in the area south of the Northern Circars, but that the salt traded by Nakarattars derived from the Palk Bay salt beds, which extend out to present-day Point Calimere.

4. Madras Board of Revenue (1783); see also Appadorai (1936), Krishna Rao (1964: 65-66), and Saruda Raju (1941).

5. The Palani manuscripts (see Chapter 7) describe Nakarattars as transporting salt on their heads or on shoulder poles ( kavatis ) carried by themselves or by servants, but not at such a scale that they required cattle.

6. Although I have not investigated the matter, it seems clear that the East India Company did not succeed in eliminating indigenous middleman renters. Nor did the Company, despite claims to the contrary (Strachey and Strachey 1985 [1882]: 218), eliminate leveraged profits for middlemen. The Nakarattar provide a counterexample. Although they may have been driven or lured away from occupations as salt traders, present-day

Nakarattars remember their forefathers as participating in the nineteenth-century salt trade, precisely in the substantial role of salt renters (see further discussion in text).

7. Ramathan Chettiar (1953: 24-25). I was able to confirm some of the details of Ramanathan's report after returning from the field. Inquiries conducted through an intermediary revealed the existence of the tank in the village, and some memory of now-defunct relations between village inhabitants and Nakarattars. I am indebted to Linda May and Muthu Chidambaram for visiting Narasingampatti on my behalf. Narasingampatti lies inside Chettinad and thus does not appear in Ramanathan's list of trading towns.

8. The monopoly on the fisheries was ended by British Governor Horton, who abolished the grant of monopoly in favor of a license fee for the right of fishing. Nakarattar control of the pearl fisheries is discussed in Samaraweera (1972: 8-9). His discussion is based on Brohier (1964).

9. Chettiar Chamber of Commerce (1963: 140), cited in Mahadevan (1976: 93). Another caste historian, Seshadri Sharma, places Nakarattars in Ceylon as early as 1730 (1970: 100).

10. Copper (1877), cited in Weersoria (1973: 16). The full extent of the Nakarattars' economic power cannot be estimated. But the impact of their domination may be gauged by articles appearing in 1832 issues of the Colombo Journal , which singled them out for criticism from Lubbai (Tamil Muslim), Jaffna (Ceylon) Tamil, and Sinhalese traders as speculators, hoarders, and profiteers. See Colombo Journal , February 16, 19, and 27, 1832 (cited in Samaraweera 1972: 10). See also Samaraweera's (1972) article, which discusses their dominance and control of trade in all of the commodities mentioned above. Finally, for richly textured eyewitness accounts (not without an element of humor), see Capper (1877), Cave (1900), and Skeen (1906), all quoted at length in Weersooria (1973: 12-19).

11. MacKenzie (1954: 90). Ceylonese merchants had two even less happy alternatives. They could have sent their sterling bills to India, exchanged them for rupees, and then imported the rupees back to Ceylon. But this process took almost a month. Alternatively, Ceylonese merchants could have waited for their sterling bills to be discounted in London, bought silver, shipped it to India, and then had it minted into rupees for shipment to Ceylon. But this would have taken more than a year.

12. During my field work, one family specifically described to me their ancestor's acquisition of title as a major salt-bed renter in Tuticorin at the beginning of the nineteenth century. Another family claimed to have cornered a monopoly in the Madras-Ceylon salt trade during the latter part of the century.

13.  See, for example, the Report of the Select Committee of the House of Commons of Great Britain (1832: Appendix No. 109), cited in Natara-

jan's (1941) description of trade between Madras and Bengal between 1796 and 1829.

14. In many cases, specific documentation linking such banking activities to specific Nakarattar businessmen is not available, but the location of many of these transactions in the Madurai/Ramnad region of the Madras Presidency, as well as clear references that do exist in documents, such as a letter from the Madurai Collector (see later in this chapter), suggest Nakarattar involvement. Such documents make suggestive references to elite European and Indian transmission of funds and the purchase of material resources through use of hundis paid to unnamed native soukars . See, for example, Grant (1784) and volumes of the Madras Country Correspondence (1752-1757). For further discussion, see Ludden (1978: 148-155), Saruda Raju (1941: 133-142), and Sundaram (1944-45).

15. Madras Board of Revenue (September 23, 1793: 5518-5519), quoted in Ludden (1978: 151). See also Madras Board of Revenue (August 8, 1793: 4887-4889) for a longer account, along with an invective against soukars in general.

16. Simultaneously with the increase in land taxation, prices for agrarian commodities from 1800 to 1850 underwent something of a depression. According to Saruda Raju (1941: 224-243), this was due, in part, to a reversal in British-Indian relationships vis-à-vis the textile trade: the British assumed the role of exporter and Madras the role of importer. Saruda Raju suggests that the resulting trade imbalance produced deficiencies in the Madras money supply which raised the price of credit and lowered the cost of commodities.

17. Board's Consultation, Madras Board of Revenue (April 14, 1828: back no. 47). On the Settlement of the Ramnad Zemindary for Fasli 1235 and 1236, p. 4 (Tamil Nadu Archives). Quoted in Price (1979: 194).

18. Price (1979: 194). It is not so clear whether Nakarattars and other soukars were significantly involved with smaller-scale landholders and agriculturalists, particularly if we assume that late nineteenth- and early twentieth-century patterns of money lending reflect an exaggeration of earlier practices, rather than a radical change. In the later colonial period, professional moneylenders such as Nakarattars made loans directly only to a small number of landowners, focusing the bulk of their money-lending activities on urban centers of the cotton and rice trade. The majority of agriculturalists-especially in nonirrigated farming zones—met their credit needs by borrowing from these wealthy peasants or even somewhat less wealthy peasants (normally, the headmen in their own villages), who made up yet another link in the circuit of credit (Nicholson 1895: 232; Saruda Raju 1941: 143; Statement Exhibiting ... 1903 XLIV: 354; Washbrook 1973). For a qualification to this view, see Robert (1983).

19.  On Zamindars ' increasing reliance on large-scale loans, see Price (1979), Rajayan (1964-65), and Sundaram (1944-45). Price also illumi-

nates the massive, escalating, and politically necessary ceremonial costs faced by zamindars .

20. See description of contemporary discussion of the relationship between the Nayak of Tanjavur and native soukars at the end of the eighteenth century (Rajayan 1964-65: 151).

21. Madura District Records 1830, Tamil Nadu Archives. Two observations are worth noting about this letter. First, the Rs. 9,000 payment in hundis is said to be Rs. 90,000, which I believe to be a clerical error. Second, a handwritten marginal note identifies the unnamed Nakarattar seeking repayment as Sathappan Chetty, who may have been the ancestor by the same name of Raja Sir Annamalai and his son Raja Sir Muthia Chettiar in the twentieth century.

22. Pillai (1930: 1178). As a consequence of the length of time required to settle legal disputes, rural credit was supplied primarily by residents with lendable capital, who had sanctions available to them other than the courts. Not surprisingly, Nakarattars were important sources of credit for non-Nakarattar landed and landless laboring inhabitants of Chettinad, who were, in effect, their dependents.

23. Conceivably, Marwaris were better positioned than Nakarattars to take advantage of the Bombay market for cotton: see Baker (1984: 267-274).

24. In other words, during the second half of the nineteenth century, Nakarattars and other Indian financiers were excluded from the lucrative European credit market that had previously had no choice but to deal with them. The formation of the Presidency Bank (which merged with the Imperial Bank of India in 1927) was followed by the arrival of the Chartered Bank of India, Australia and China in 1853, The National Bank of India in 1877 (previously represented by the British firm of Binny's and Co.), and the Mercantile Bank of India in 1892 (Mahadevan 1976: 31-37). The arrival of the European banks in India is the subject of two studies (Natarajan 1934, Sadasivan 1939) that provide the basis for Mahadevan's discussion.

25. Calculated from ''Statements Showing the Prices of Food Grains" in Government of Madras Land Revenue Reports for Fasli 1290 (1880-81) and for Fasli 1335 (1925-26) cited in Washbrook (1973: 157).

26. Price (1979) presents other histories of Nakarattar exploitation of zamindar mismanagement that begin as early as the 1820s. See, for example, her account of the Zamin of Sivaganga.

27. The events are described in Price (1979). But it is difficult for me to identify exactly who these other families are from the data she presents. The case is clear for Al. Arunachalam, whom Price correctly identifies as the father of Al. Ar. Ramasami, who became Zamindar of Devakottai.

28.  Details of the three lawsuits described by Price (1979: 198) may be found in the Madras High Court's Documents in Regular Appeal No. 80

(1876), which contain the judgments for Original Suit No. 1 of 1872 (Chithambaram Chetty v. Raja Vijia Regunada Muthuramalinga Saithupathy, Zemindar of Ramnad), p. 12; Original Suit No. 3 of 1872 (Narainen Chetty and Vyravan Chetty v. Vijia Regunada Mootooramalinga Saithoopathy, Zemindar of Ramnad), p. 17; and Original Suit No. 5 of 1874 (The Collector of Madura and Agent to the Court of Wards on behalf of Baskarasami Setupati, Zemindar of Ramnad v. Ramasami Chetti son of Arunachellam Chetti), p. 3. Also cited in Price 1979: 198.

29. The Court of Wards—a "proper" British institution—would approve and pay for only one marriage for the young prince.

30. Information on the acreage of Ramasami's domain was provided in an interview with Al. Ar. Somanathan Chettiar by Raman Mahadevan in 1973 (Mahadevan 1976: 45).

31. S. Rm. M Chidambaram Chettiar was the elder brother of Raja Sir Annamalai Chettiar and grandson of S. Rm. M. Rm. Muthia in the story of the Settupatti of Ramnad. He was also father of Sir M. Ct. Muthia Chettiar. All members of the S. Rm. family hail from the village of Kanadukathan.

32. Sundara Iyer (1906, cited by Krishnan 1959: 31). Useful population figures for castes are notoriously difficult to obtain. For example, the 1901 census (Francis 1906: 149) provides a virtually meaningless number of 289,457 for the population of Chettis in India. There is no way of knowing what groups were counted as Chettis, let alone of determining the population of a subgroup such as the Nakarattars. The only other figure available for this period comes from Sundara Iyer's (1906) article in the Indian Review (cited by Krishnan 1959). I do not know the basis for his estimate, but he is reported to have had close personal knowledge of the Nakarattar community (Krishnan 1959: 31), and his figures may have been based on a caste census undertaken by one or another Nakarattar caste association. Census figures would most likely have been obtained by consulting with the karyakkarars (trustees) of Nakarattar clan temples, who maintained lists of all Nakarattar marriages.

33. It is impossible to be more precise about the number of large-scale landholding Nakarattars in the absence of other data. The range of variation in my estimate reflects uncertainty regarding the number of people constituting an average large-landholding family. For discussion of the organization of Nakarattar families and family firms, see Chapter 6.

34. Correspondence with Lakshman Chettiar, March 1982.

35.  The nature of these principal business location "circles" (Pillai 1930) is somewhat mysterious. They fall within Tamil-speaking districts, but they do not exhaust the Tamil parts of Madras in which Nakarattars carried out their business. Interestingly, they seem confined within the geographic boundaries established by the medieval Pandyan kingdom, leaving out, for example, Nakarattar business locations in Madras or Tan-

javur (i.e., within the medieval Chola kingdom). Moreover, the criteria for distinguishing between the different circles of Chettinad is obscure to me, as are the implications of making such distinctions for segmenting the Nakarattar caste.

36. The British banks in Ceylon included the Bank of Ceylon (established in 1841), the Oriental Banking Corporation (1848), the Chartered Mercantile Bank (1854), and the Bank of Madras (1867): see Kannangara (1960), Mahadevan (1976: 100-101).

37. Capper (1877), quoted in Weersooria (1973: 16), notes that Nakarattar bankers were loaning substantial sums to indigenous coffee growers. But Ceylonese holdings were quite small relative to those of the British.

38. The coffee blight actually began to make inroads in production some years earlier. Coffee reached its maximum production figure in 1870. In that year, 1,054,000 pounds were grown on only 185,000 acres. This proportion contrasts with 437,000 pounds grown on 322,000 acres in 1881. Only the increasing price for coffee (from Rs. 54 to Rs. 100 per hundred pounds) and the uncertainties involved in switching to a new crop kept the acreage under coffee cultivation expanding. The sources for these figures are the Ceylon Blue Book Statistics (Ceylon Colonial Office 1870-81) and Ferguson's Handbook and Directory (1870-81). For detailed statistical analysis of the growth of Ceylon's plantation economy, see Rajaratnam (1961).

39. Vysiamitran (a Nakarattar caste journal published out of Colombo), December 25, 1916, cited in Mahadevan (1976: 112).

40. Ceylon Banking Commission (1934 II: 253, 316, 354-355). See Chapters 5, 6, and 10 in this book for discussion of the structural role of adathis in the Nakarattar banking system.

41. Evidence of J. Tyagarajah (head shroff for the Hongkong and Shanghai Banking Corporation), Ceylon Banking Commission (1934 II: 353-354). In Ceylon, the British banks engaged shroffs (normally from one of the non-Nakarattar Chetti castes of Colombo) to recommend and guarantee loans made to local bank clients. Their business depended on their knowledge of potential borrowers (Wright 1907: 317, cited in Weersooria 1973: 25). The term shroff has broad currency all over India. It is, in fact, a cognate form of chetti and is used in North India in much the same way that chetti is used in South India.

42. I am not confident that this is the correct tolil vilacam ("business name") formed from the initials of ancestors in the proprietor's patriline (see Chapter 6). For example, it might have been Ar. Ar. S. M.

43.  Although a detailed analysis of Nakarattars and Ceylon's legal system is beyond the scope of the present investigation, I note that comparison of Ceylonese legal interpretations of Nakarattar practice (described by Weersooria 1973) with Nakarattars' own interpretations of these practices (described in Chapter 6) indicates significant and, for Nakarattars, serious

discrepancies. Among the relevant legislative actions were the Money-lenders Ordinance Number 2 and the Business Names Registration Ordinance, both of 1918; the establishment of an income tax in 1932 and an estate tax in 1938; and the enactment of the Pawnbrokers Ordinance of 1942. Weersooria (1973) gives excellent, brief descriptions of these acts and of their implications for the Nakarattar. In addition, he reviews the considerable body of legal decisions concerning Nakarattar banking practice in Ceylon. See especially his discussion of decisions regarding the usage of Nakarattar firm names ( tolil vilacams ) and the legal interpretation of such usage with respect to the financial liabilities of proprietors and agents. See also his discussion of legal precedent regarding the application of income taxes and estate taxes to Nakarattars, only part of whose business was in Ceylon (Weersooria 1973: 76-126).

44. On the Burmese rice market, see also Adas (1974), Andrus (1948), Cooper (1959), Mahadevan (1976, 1978a, 1978b), Siegelman (1962). My summary of this body of work (in the text) provides an adequate background for the primary focus of the present investigation: namely, internal Nakarattar social organization. Readers interested in more detailed accounts of relationships between Nakarattars and their clients are referred to the publications just cited.

45. See Chapter 5 for analysis of the interest rate structure of the Burmese credit market.

46. Nor have I described what I interpret as a third phase of Nakarattar commercial evolution, manifested in elite Nakarattar investment in industry in Madras, Burma, and Malaya. See Mahadevan (1976) for an account of Nakarattar industrialization in Indochina and Madras.

1. Descriptions of Chettiar interest rates were published in the Madras and Burma Provincial Banking Enquiry Committee Reports (MPBEC 1930; BPBEC 1930) and the Ceylon Banking Enquiry Committee Report (CBEC 1934) and reproduced in various publications thereafter. I discuss difficulties involved in interpreting these descriptions in the Appendix A.

2. Deposit accounts were separate from loan accounts, which were secured against various kinds of promissory notes or mortgages. A loan, moreover, was paid out in one lump sum and could not be recalled by issuing a hundi . In some cases (typically for petty shopkeepers in towns and cities), a particular kind of loan called a kandu kisti loan might be advanced. These loans (normally for small amounts, with interest deducted in advance and payments of the principal scheduled over a short period) yielded a high return to the moneylender (25 percent per annum or more: see MPBEC 1930 I: 227). However, kandu kisti accounts, like other loan accounts, were not subject to the drawing of hundis .

3. Many merchants extended their business relationship with Nakarattar bankers beyond purely financial transactions. For example, most Burmese and many small-scale Nakarattar merchants had no storage "go-downs" (warehouses) of their own. In such cases, merchants frequently stored and insured their paddy in a go-down conveniently owned by the bankers. Ultimately they might even sell the paddy to the bankers.

4. In practice, I am told, Nakarattar bankers were normally liberal in converting thavanai hundis into dharsan hundis if their clients were faced with unexpected cash flow problems. At the same time, the Burmese banking analyst Tun Wai observes that thavanai deposits, technically cashable by hundis after as short a term as two months, often remained with the bank of deposit "for decades" (1962: 45).

5. Pay order hundis are not mentioned in any published description of Nakarattar banking practices, but a retired banker informed me that they used this English phrase in reference to hundis drawn as dowry payments.

6. For discussion of the institutional sanctions operating to minimize conflict and resolve disputes, see Chapter 6.

7. I found no evidence for any use of "secret" codes, as described by Thurston (1909 V: 270-271), based on the alphabetic characters used to spell the names of Tamil deities and temples.

8. Without a general ledger, Nakarattar account books do not seem to provide a picture of the overall profitability of an agency. This seems peculiar. And it is possible that the books which I examined did not represent a complete set. In this regard, I note that Somalay (S. M. Lakshmanan Chettiar), a Nakarattar writer and a retired banker, makes ambiguous references to a ledger called an ainthugai in a rough manuscript for an unpublished book. In one instance, he uses the term to describe the duplicate ledger sheets sent by an agent to his proprietor and copied by the proprietor's son. This reference is similar to Krishnan's (1959: 34) mention of periodically mailed balance sheets by the same name. In another instance, Somalay presents an explicit definition, translating ainthugai as "balance sheet" and stipulating that "this contained five items—capital, borrowings, investments, other outstandings, and profits." I note that this statement contains no evidence for the use of a double-entry system. Nor does Krishnan's description of indigenous accounting methods alter my belief that double-entry systems were not used.

9. I am deeply indebted to Rachel Winslow, CPA, who helped me to prepare the general ledger and balance sheet from figures translated from the agency's peredu .

10. Weersooria (1973: 109) provides a similar list of the types of Nakarattar account books.

11. Interview with Lakshmanan Chettiar, March 25, 1981. Although the mempanam subcategories identified were gleaned from analyzing the Burmese agency's account books described in the text, only minor and

obvious modifications are required to adjust for differences in the funding sources available to Nakarattars in Malaya, Ceylon, or Singapore—or Madras, for that matter.

12. Written evidence about thanadumaral or thandu morai deposits (alternative transliterations for tanatu murai panam ) submitted to the Madras Provincial Banking Enquiry Committee suggest that the term may also have been used as an umbrella term for all deposits from relatives, including accimar panam deposits from in-laws. (see evidence given by M. R. Ry. C. A. C. Kasinathan Chettiar, MPBEC 1930 II: 1116) and by the Nattukottai Nakarattars Association (MPBEC 1930 II: 1103-1109).

1. Weber, of course, nowhere considers such a paradox and proposes instead a contrast between what he terms rational individualism and traditional egocentricism as an apparently exhaustive dichotomy. Thus, in at least one place, he contrasts wealth accumulated unselfishly by a Protestant's pursuit of his calling with wealth accumulated by a pre-Protestant merchant for the sake of his family. The former he attributes to an individualistic effort to avoid worldly contamination of actions undertaken solely for the glory of God. The latter he attributes to an egocentric extension of the family founder's personality (1958: 276). Whatever may have been the case for the pre-Protestant Europeans that Weber had in mind, for the Nakarattars, just the opposite seems to have been the case. That is, familial goals were internalized and projected onto individuals, rather than the other way around.

2. Mahadevan (1976) provides the best documentation of the shift of Nakarattar investment into capitalized industry.

3. The functional equivalents of these parent bankers performed similar roles in the seventeenth century, although they did so under a different title: ettis .

4. I make use of three primary sources of information in my analysis. These consist of notes for an unpublished book on the Nakarattars written by a retired Nakarattar banker and caste historian, Somalay (a nom de plume derived from S. M. Lakshmanan Chettiar), interviews with Somalay and other Nakarattars conducted as part of my field research in Tamil Nadu in 1981, and a sample of a Nakarattar firm's account books from its agency house in Burma for the period from 1912 to 1915. These various primary oral and written sources are used to flesh out, verify, and refine more general accounts of the system available in already published work by Krishnan (1959), Mahadevan (1976), Naidu (1941), Pillai (1930), Siegelman (1962), Tun Wai (1962), and Weersooria (1973). The triangulation of several different sources permits one to be reasonably confident of the reconstruction of the Nakarattar banking system at its chronological

and spatial peak, before the community shifted massively from mercantile and financial to industrial ventures.

5. This practice has not been followed consistently since the 1930s. In the postwar period, many men have simply indicated their father's name and their village name when establishing their own firm; for example, P. Ct. Allagappa signifies ''Palattur village, Chidambaram Allagappa." Occasionally one also finds the occupational title "Chettiar," but no indication of natal village, as with Ct. C. Allagappa for Chidambaram Chettiar Allagappa. But this last practice has lost popularity since the 1960s (this illustration as well as those provided in the text are taken from Lakshmanan Chettiar 1953).

6. In this case, "Tevar" designates the employee's caste affiliation.

7. " Pulli parkka = to conjecture how much a sum may be; to estimate" (Fabricius 1972: 727).

8. As an example of their use of pullis as units, I obtained figures from the Madurai Nakarattar Sangam for the total number of pullis presently registered in all nine Nakarattar clan temples (about 20,000) and hence, even in the absence of government census figures indexed by caste, feel relatively confident in accepting their estimate for the contemporary Nakarattar population at about 100,000. This coincides with Chandrasekhar's (1980) estimate, which was probably derived in the same fashion. It differs from recent estimates by Moreno (1981) and Timberg (1978).

9. A considerable body of case law has been generated specifically in reference to estate and income taxes levied against Nakarattar individuals outside of India. In these cases, there was no question of refusing to grant jural status to the Nakarattar valavu . These and other Hindu joint families were legally recognized as juristic individuals similar to corporations and subject to the application of Hindu Mitakshara law (cf. Kane 1946 III; Tambiah 1973a). Litigation arose only in determining whether some body of assessed property belonged to an individual or to the joint family of which he or she was a member. See Weersooria (1973: 110-116) for a review of the issues and the relevant cases in Ceylon.

10. The large family houses shared by a valavu may be responsible for the popular name of the Nakarattars: the Nattukottai Chettiar ("Country Fort Chettiar"). Alternatively, it is sometimes speculated that the name represents an abbreviation of Nattarasankottai, a town within Chettinad thought to be one of the Nakarattars' earliest residences.

11. Nakarattars also employ the common Tamil term kutumpam , marked by an emphatic article: ore kutumpam "one (same) family."

12. A passing comment by Evers (1972: 637) apparently supports my interpretation. There, he mentions that Nakarattar "spheres of accumulation," in contrast to "spheres of consumption," were administered by the heads of Nakarattar joint families. But he does not elaborate on the issue.

13. For further discussion of Nakarattar marriage alliances, see Chapter 8.

14. See Chapter 5, note 2 on kandu kisti loans.

15. I am not able to confirm at the present time that the twelve largest landowners represent all the adathis in Burma. Nor am I able to speculate about the presence or absence of adathi status among other Nakarattars outside this group. Further investigation could be carried out by querying surviving Nakarattar bankers from the period or by locating copies of the adathi lists maintained by the Imperial Bank of India.

16. A list of the Nakarattar firms represented on the Burma Indian Chamber of Commerce in 1925-26 gives the following firms: A. K. A. Ct. V., A. A. Krm. M. Ct., A. K. Rm. M. K., P. K. N., Rm. P., S. A. A., S. A. Rm., S. K. R. S. K. R., S. M. A. Ra., S. Rm. M. A., S. Rm. M. Ct. Sir., S. Rm. M. Rm., and T. S. N. From the Burma Indian Chamber of Commerce (1929: v), cited in Mahadevan (1976: 187).

17. See also the evidence of a Ceylon bank shroff —a semi-independent loan guarantee officer—presented to the Ceylon Banking Enquiry Committee in 1934 (quoted in Chapter 6).

18. Nakaravitutis should not be confused with the headquarters for Hindu sectarian orders, also called matams or atinams .

19. For a cultural account of the differing ritual roles of Nakarattars and Kongu Goundar Vellalars in Palani, see Marriott and Moreno (1990).

20. The "information transfer" function of these ritual events continued even when the rituals were augmented in this respect by the publication of caste journals sponsored by local caste associations. Such journals included Dhanavanikan (a monthly journal published from Kottaiyur in Chettinad and Rangoon, ed. A. K. Chettiar), Dhana Vysia Ooliyan (a weekly journal published from Karaikudi in Chettinad, ed. S. Mooragappa Chettiar), Kumari Malar (a monthly journal published from Madras, ed. A. K. Chettiar), Ooliyan (a weekly journal published from Karaikudi in Chettinad, ed. Rai Chokalingam), Vysiamitran (published from Devakottai in Chettinad, ed. S. T. Ramanathan Chettiar), and Nakarattar Malar (published from Madurai).

21. The elder mediators or arbitrators of a panchayat were conventionally five in number ( panch means "five"), although reality often deviated from this ideal. The difference between a panchayat that resolved a dispute through mediation and one that resolved a dispute through arbitration is formal. In the latter case, both parties agreed in advance to abide by the decision by signing a legal document called a muchalika , which could be used as evidence in a civil suit if either party found it necessary to go to court. Muchalikas (from a Persian term) were broadly used in similar contexts throughout colonial India. However, it is not clear when this practice was adopted by the Nakarattars. According to Thurston (1909 V: 263),

agreements were not made in writing as recently as 1909. The presumption, then, is that muchalikas were a recent response to the colonial legal system.

22. As discussed in note 18, Saivite matams or atinams should not be confused with Nakaravitutis , which were sometimes called Nakarattar matams. Matam panchayats were apparently receding into the background by the end of the nineteenth century (Thurston 1909). None of my informants remember them in operation, although the palm-leaf manuscripts from Palani temple (Chapter 7) offer a glimpse of conditions in which they would have played an important role during the seventeenth century. Kovil (temple) panchayats (see Chapter 9), however, continued until at least the 1940s, although much of their role was taken over by caste associations and by the court system of colonial India during the course of the twentieth century.

23. Thurston (1909: 263) mentions a panchayat held for a dispute between two families from the same clan that arose over adoption.

24. I am aware of a 1940s example of a temple clan panchayat meeting in which a wealthy member was denied permission to marry his daughter to any family within the community until he had paid his temple dues. I am also aware of several outcaste marriages (albeit among extremely wealthy, Westernized, and modern Nakarattars) that brought no sanction from any community institution.

25. Outsiders sometimes mistake a line of the S. Rm. family for a royal Nakarattar lineage because it was represented by Raja Sir Annamalai Chettiar and his son, Raja Sir Muthia Chettiar. But Nakarattars have no royal lineages. The title "Raja Sir" was conferred by the British on Raja Sir Annamalai in the twentieth century and gives him superior, zamindar -like land rights over a village also created at that time. It was Annamalai's idea to name his village by the same name as the entire territory comprising the Nakarattar homeland of Chettinad—no doubt precisely in order to play on uninformed British sentiments about royalty.

26. Among works that continue to view individualism as essential to capitalism I include not only historical studies such as Shoji Ito's (1966) essay about the Nakarattar, discussed at the beginning of this chapter, but also general theoretical works, from the classic sociological theories of Marx, Durkheim, Weber, Simmel, and Mauss, through the modernization theorists of the 1960s, to recent studies in the culture of capitalism such as those by Dumont (1977, 1986) and Macfarlane (1987). For recent critical evaluations of these tendencies see Carrithers, Collins, and Lukes (1986).

27. Caste studies by Conlon (1977), Dirks (1987), Leonard (1978), and Mines (1984) do consider some of the corporate functions of caste.

28. This would be consistent with application of Weber's Jewish "double ethic" (1958: 271) to Hinduism.

1. The Six Deeds of Gift contain two lists of ritual practices as well as six "deeds." All the manuscripts are presently stored in Palani Temple, but transcripts have been compiled and published by the Tamil professor and Nakarattar authority V. Sp. Manickam (1963). I have not seen the original manuscripts (which are likely to have been recopied in any case). But Manickam vouches for their authenticity on the basis of internal linguistic evidence as well as personal knowledge about the conditions of their storage and preservation. The texts are written in an archaic mixture of classical Tamil and the colloquial Tamil of the seventeenth and eighteenth centuries, presenting difficult translational problems to the nonexpert. My analysis would not have been possible except for the translation provided for me by Professor K. Paramasivam (1981a), parts of which I reproduce with his kind permission.

2. I have analyzed the origins of the Nakarattar Palani pilgrimage in greater detail elsewhere (Rudner 1987).

3. Paramasivam (1981a: 3). The figure "96-4-0" makes reference to units of currency smaller than a rupee: either chakrums and pannams or pannams and kasus .

4. "The deity is not given food, which it has to reciprocate with a 'counterprestation' [referring to Babb 1975: 57] of prasada . Instead, the deity is served a meal, which the worshiper later consumes. The food prasada is not a return gift, but the same food transmuted—like all other substances that become prasada —by its contact with the deity in its image form. There are no food transactions and prestations in puja , because they are not what the food offering together with the receipt of prasada amount to" (Fuller 1992: 78). Nonanthropologists may be unfamiliar with the standard but technical term prestation introduced by Mauss (1954) to denote a gift-giving action that incurs a recursive obligation for the recipient to return a prestation.

5. Appadurai has explored the political dimensions of ostensibly purely domestic relationships between husband and wife in his (1981c) essay "Gastro-politics in South Asia."

6. My suggestion that anthropologists overemphasize the importance of status purity or impurity in Hindu ritual specialists deserves elaboration. For discussion of some of the relevant issues see Claus (1978), Das (1976), and Parry (1982).

7. For discussion of Dumont's Brahman-centered model of caste hierarchy, see papers in the Journal of Asian Studies (1976).

8. A distinctive "Cambridge School" represented by Baker (1976; Baker and Washbrook 1975), Seal (1973), and Washbrook (1976) views the trend as a consequence of the cost of empire, "the perennial dilemma of the Raj: If the administrative cost of intervening was not to overtake the

returns and the security of the state to be put at risk, Indian collaboration would have to be much extended. . . . Sys">

9. In general, the postulated pre-twentieth-century stage of centralized colonial rule is not clearly distinguished from precolonial regimes with respect to the question of local influence on regional politics. In fact, in both cases the issue is rather vague and undefined. Most discussions take a ''top-down" orientation and focus on the revenue powers of the state. Some others pay attention to the intrusion of the state into villages by its appointment of village officers. Still others pay attention to the intrusion of the state into local temples. But little or no consideration is given to small-scale political regimes that extended beyond individual villages or temples. For South Indian examples of such regimes, see Beck (1976), Dirks (1988), and Price (1979).

10. As with my comments on recent interpretations of the nineteenth-century evolution of government in India, I only summarize here a well-developed historiography chronicling and analyzing the evolution of Anglo-Indian law and attitudes towards religious gifting and temples. The standard reference is Mudaliar (1974, 1976). Major monographic treatments may be found by Appadurai (1981), Fuller (1984), and Presler (1987). My view of the subject has been particularly influenced by Appadurai (1981) and Breckenridge (1976, 1977, 1983).

11. See Mudaliar (1974) on the HRCE Acts of 1817, 1863, 1925, and see Breckenridge (1977, 1983) for particularly perceptive interpretations of the social context and consequences of these acts.

12. The Nakarattar elite comprised perhaps 5 to 10 percent of the 10,000-member caste in 1900 (see Rudner 1985).

13. Other branches of the S. Rm. family and families supported by S. Rm. were also prominent participants in Madras politics. Annamalai's brother, Ramaswamy, had started a secondary school in Chidambaram sometime during the end of the nineteenth century or the beginning of the twentieth century and was praised as the first Chettiar to "have boldly deviated from the medieval ways of debasing charity and diverted it along more modern and useful channels" ( Madras Mail , June 26, 1920, cited in Arooran 1980: 124). Annamalai built on his brother's philanthropy by founding Sri Minakshi College in Chidambaram in 1920. In

1928, Annamalai's political ally, Chief Minister P. Subbaroyan, formed a committee to decide whether there was a need to establish a Tamil university. The committee recommended a federal affiliating type of University with six principal centers at Madurai, Tirunelveli, Tiruchirapalli, Coimbatore, Kumbakonam and Chidambaram. At the same time Annamalai offered an endowment of Rs. twenty lakhs to form a unitary, residential university just at Chidambaram. Another Chettiar offered Rs. 35 lakhs to establish a unitary university at Madurai ( Madras Mail , May 5, 1928, cited in Arooran 1980: 62). Present-day informants say the other Chettiar was Karimuthu Thiagarajan, on the advice of Annamalai's son-in-law, Venkatalchalam (C. V. Ct.) Chettiar. But the government accepted Annamalai's offer, and the Annamalai University bill, with its stipulation of a matching 20-lakh grant and an annual 1.5-lakh grant was passed by the Legislative Council, the governor, and the viceroy by the end of 1928 (for further details, see Arooran 1980).

14. I am told that Annamalai and his son Muthia exercised complete control of the Indian Bank from the time of its founding until its nationalization in 1969, and that Muthia continued to exercise considerable behind-the-scenes influence until his death in 1984.

15. Speech in Rangoon by Raja Sir Muthia Chettiar, sometime in the 1930s (S. M. Lakshmanan Chettiar, personal communication, March 1981).

16. See Baker (1976: 153-154) for a rather scandalous account of Annamalai's 1923 election.

17. From his election as mayor in 1933 until his death in 1984, Muthia held continuous formal political office. Upon completing his term as Mayor of Madras he became the Chettiar representative to the Legislative Assembly, in which capacity he served until 1952. In between he held various ministerial posts in Justice party governments (see Appendix C). From 1952 until 1957, he was elected (with Congress party support) as the independent Tirupattur representative of the Assembly. From 1957 to 1962, he served another term as the Congress party representative from Karaikudi. Following this and up until his recent death he was a nominated member of the Legislative Council (the upper house of the Madras bicameral legislature).

18. Erskine to Willingdon, March 11, 1935, Erskine Papers, vol. 5, India Office Library; cited in Baker 1976: 243.

19. Not that Muthia had a monopoly on such actions: other zamindar Justice leaders, including the Raja of Bobbili, the Raja of Venkatagiri, and the leader of the Legislative Council, Sir Mahomed Usman, were all said to have made or to have attempted to make similar deals with the government (cf. Baker 1976).

20. The collapse of the roles of temple trustee and kingly protector into a single category was due less to any victory of merchants over kings than

to the colonial transformation of kings into litigants on par with any other elite member of society (cf. Breckenridge 1977).

21. My information about the process of Annamalai's involvement in the educational movement comes from discussions with informants during my field work in Tamil Nadu. Thus, the events were important enough to be remembered sixty years later. I was subsequently able to confirm my informant's memories with documents cited in Arooran (1980).

22. Complaints from dissident Chettiar political factions provide the best documentation in support of this hearsay about Annamalai's and Muthia's use of power for their own purposes. See, for example, a variety of Government Orders from the Madras Department of Local Self Government in the 1920s; reports in the Hindu (a prominent newspaper published in Madras), April 8-9, 1921; and editorials in Ooliyan (a Chettiar journal published by a non-S. Rm. A. M. faction between 1935 and 1937). Most of these allege S. Rm. A. M. manipulation of the ministry and local boards to influence Chettiar-controlled temples and markets (see also Baker 1976: 269).

23. For an account of Annamalai University's history, see Arooran (1980).

24. See Kochanek (1964: 151-155) on S. Rm. A. M. influence on the Chamber of Commerce.

25. This definition of mercantile elite is similar to the use of the term dominant caste as refined by Beck (1976a). (The original formulation is from Mayer 1958 and Srinivas 1959). Like agrarian power, mercantile dominance is a function of rule and not necessarily of ownership of resources (see Neale 1969).

1. By lineage , I mean here the maximal lineage segment active in Nakarattar kinship organization, not the patrilineal line of descent extending without limit from a focal individual. Nakarattar clans and lineages are described in relation to Nakarattar control of Chettinad in Chapter 9.

2. The Nakarattar joint family ( valavu ) is best considered as the minimal lineage segment of Nakarattar kinship organization. Even an inmarrying woman undergoes a ceremony at marriage in which she is placed under protection of the lineage segment's tutelary deity ( kula teyvam ), becoming, in effect, a member of her husband's lineage.

3. See Freeman (1961), for example, for a definition of kindred based on a criterion of genealogical distance.

4. The differential linkage to the husband's and wife's descent groups is signaled by the couple's obligation to pay a religious tithe ( pulli vari , see Chapter 9) to the husband's clan temple, but not to the temple of the wife's father's clan.

5. For an unusual differentiation between descent group and kindred , see Carter (1973).

6. A general definitional discussion of the term kindred is not relevant in the present context. Suffice it to say that in dealing with this definitional problem, as in dealing with many others, I follow Goodenough (1970). See also Dumont's (1957b: 21-22) critique of Emeneau's (1941) theory of Dravidian double-unilineal descent.

7. The Tamil phrase cir tanam is easily confused with the Sanskrit stri dhanam (see, for example, Dumont 1957b: 31-32). The latter, also interpreted as a term for dowry, is generally translated as "woman's property." But the Tamil phrase here employs the word cir for "glory," "fame," or "beauty,'' not the Sanskrit word stri . Accordingly, I translate cir tanam as "property which brings fame or glory.'' Although Sanskrit terms are often incorporated in Tamil ritual vocabularies, the Nakarattar (and also Dumont's Kallar) clearly adopt a Tamil frame of reference for describing affinal gifts between two families allied by marriage.

8. Nakarattars practiced Dravidian cross-cousin marriage. Accordingly, the presence of common parallel cousins between two families rendered the families parallel to each other and union between them incestuous. The definitions of Dravidian "cross" and "parallel" relationships (also termed "affinal" and "kin" relationships, cf. Dumont 1957a, 1957b), and the distinction between them, have preoccupied anthropologists since Lewis Henry Morgan (1870). There is still no generally accepted interpretation. For more recent efforts to arrive at a solution see Carter (1973), David (1973), Dumont (1953, 1957a, 1957b, 1961, 1975), Kay (1965, 1967), Scheffler (1971, 1972a, 1972b, 1984), Trautmann (1981), Tyler (1966), and Yalman (1971). For a discussion of aspects of the history of these efforts, see Rudner (1990).

9. In contemporary South Indian society, nakaram refers to a city. Historically, nakaram refers to a medieval guild or municipal council. Kovil refers to a temple or palace. The compound term, nakarakkovil , is a contraction of nakaram kovil ("temple of the nakaram "), but refers to both the Nakarattar clan (literally, "temple-clan") and the Nakarattar clan temple. The derivation is not clear. See Rudner (1985).

10. Vattakai is an archaic form of vattam (literally, "circle"). Both terms have been used to denote a territorial division since the Chola period (900-1200).

11. Similarly, karais may once have constituted medieval regional assemblies. The terminological distinction between vattakais and karais may indicate a jurisdictional difference between Colamandalam and Pandyamadalam.

12. This shift to a later age for marriage is common throughout India.

13. The basis for judging a figure to be "odd" (hence to be "rounded off") is not clear to me. My informant gave examples of Rs. 1,420 and Rs. 1,900.

14. In describing the ritual construction of Nakarattar marriage alliance, I adopt the standard anthropological convention for abbreviating genealogical relationships, e.g., F for father, M for mother, etc. For complete list of abbreviations, see Appendix D.

15. For a description of Nakarattar nuptial ceremonies, see Thurston (1909 V: 265-268).

16. Patrilateral prestations were not institutionalized and not very much in evidence during the course of my field study. My data for them fall into three categories.

The first type of data consisted of personal observations of patrilateral gifts to first children in a small number of families. I did not collect a Tamil term for these gifts. The few examples in my data may represent examples of murais or vevus . They may exemplify a separate cultural category in their own right. Or they may simply reflect the custom of one or two families and not the Nakarattar caste as a whole.

The second type of data were reports of patrilateral prestations of gold coins to a goldsmith prior to matrilateral prestations of sireduththal ( cir tanam ) (Thurston 1909 V: 265). According to one informant, the coins were traditionally beaten into a thin sheet and then melted into the gold used for the kalutturu , a massive ceremonial necklace presented to a Nakarattar bride in addition to her tali . My informants, however, said that the gold was not called paricam or "bride price" (cf. Dumont 1957b: 83). Moreover, the practice of presenting a kalutturu is becoming prohibitively expensive. In fact, most contemporary marriages rent a kalutturu for ceremonial purposes. It may be the case that the groom's family covers the rental cost. But I have no further information on the topic.

The third type of data collected were reports of patrilateral prestations occurring at the uppu eduththal ("salt carrying") ceremony during a wedding (Thurston 1909 V: 266). I did not witness any such prestation in connection with the salt ceremony during weddings in which the salt ceremony was included. Among my informants who knew anything about the salt ceremony, nothing was known about any patrilateral prestations.

17. In fact, the implications of the distinction between systems of classification and systems of social groups have never been a subject of common understanding. See, for example, the differences in opinion between Leach (1961 [1951]), Lèvi-Strauss (1969), and Needham (1962) with regard to the minimal number of local groups necessary for maintaining various elementary kinship structures.

18. For a more elaborate discussion of some of the difficulties in maintaining the distinctions between perfect and virtual affines, see Rudner (1990).

19.  Burkhart (1978) and Tambiah (1973a) both point to uses of the term taymaman in cases where WF is identical with MB. But the only lessons they draw from this usage are that "the tie with the mother's agnatic

group is stressed" (Burkhart 1978: 172) and, even more weakly, that "distinctions such as these represent a mapping of notions of genealogical distance on the kin categorical distinctions (Tambiah 1973a: 124-125).

20. A systematic analysis of Nakarattar rites of marriage would be digressive in the present context. I note briefly that not only do WF and MB receive titles that are specially marked versions of purely consan-guineal kinship terms, but also the bride is ritually transferred to the protection of her husband's family deity ( kula teyvam ).

21. Questions can be raised about the applicability of Dumont's (1957b) "personnel replacement" argument even to his non-Nakarattar castes. None of the replacements he describes for cir -givers ever extend beyond the second generation. Thus, he fails to demonstrate that the successive replaceability of cir -givers within a two-generation span generates a replaceability for descent groups that succeed each other in the formation of affinal kindred. As we have seen, alliances are not automatically renewed among the Nakarattar. The question is open in the case of non-Nakarattar groups.

22. Amman is used for both the MB and the FZH. If two families contract a marriage alliance, both the HF and the WF are designated by maman or mamanar , indicating their "in-law," murai -obligated status (regardless of their identity as a real MB or FZD). For both an unmarried MBS and an unmarried FZS, attan is used if older than ego, and maittunan if younger than ego. Again, the inferior term, maittunan , is used for the WB even if older than ego, thereby marking his murai obligations, but not a lateral preference for marriage. Attiyandi is used for the MBD and the FZD (regardless of age in relation to ego). For a complete list of Nakarattar kin terms, see Appendix D.

1. For traditional caste histories, see Chinnaiya Chettiar (1941-42), Chockalinga Ayyah (1919), Lakshmanan Chettiar (1953), Ramanathan Chettiar (1953), Sharma (1970), and Subramaniyan Ayyar (1895). In the absence of independent documentary evidence, these histories are best understood as ideological tracts or Malinowskian charters for behavior. For review of available documentation, see Rudner (1985). For further discussion of the Nakarattar varalaru , see Chapter 10.

2. According to the varalaru , the original seven families first arrived in Chettinad in the town of Nattarasankottai along with two other groups of non-Nakarattar Chettis. The three Chetti groups then moved to the town of Ilayathakudi. The non-Nakarattars left Chettinad entirely. Six of the seven families established themselves in their own village, separate from Ilayathakudi, and the remaining Ilayathakudi family split into three

groups, of which one remained in Ilayathakudi and the other two settled in Iraniyur and Pillaiyarpatti, respectively.

3. Ramanathan Chettiar (1953) also provides more detailed information on the individual religious gifts.

4. On religious endowment and control of religious trusts, see Appadurai (1981), Breckenridge (1976), and Mudaliar (1974). For description of a specific Nakarattar endowment in the context of a detailed analysis of nineteenth-century religious endowment generally, see Breckenridge's (1976) account of the Nagappa Chettiar kattalai in Meenakshi Temple in Madurai.

5. For village temples, trusteeship was frequently vested in the dominant family of the dominant lineage of the village. For the nine Nakarattar clan temples ( nakarakkovils ), karyakkarars were recruited on a hereditary basis, rotating in a regular fashion between selected small groups of prominent families in the clan from throughout Chettinad. In both cases, the ancestors of these families had demonstrated their generosity ( vallanmai ) and trustworthiness ( nanayam ) by contributing to the founding or renovation of the temple. For major public temples, the Hindu Religious and Charitable Endowments Board appointed individuals to nonhereditary positions on the boards of trustees. In general, however, one qualification for appointment was an act of endowment by oneself or one's ancestors.

6. Ownership of a temple depended, among other things, on the legal definition of its congregation, another complex legal problem (see Appadurai 1981). Depending on the temple under litigation, Nakarattar plaintiffs or defendants might be joint families, local lineage segments, clans, or the caste as a whole. For an account of litigation between two Nakarattar factions for control of a temple in Colombo, as well as accounts of other Nakarattar litigation in colonial Ceylon, see Weersooria (1973).

7. According to the administrative report to the Board of Trustees of Ilayathakudi Temple (Ilayathakudi Devasthanam 1939), the temple karyakkarars (trustees) had originally presented their argument to the Madras High Court in 1912 as part of a suit to confirm their position as hereditary trustees (Madras High Court, Appeal No. 72 of 1923, "Scheme for administration of Ilayathakudi Temple"). The 1939 report indicates that the proposed scheme was accepted in 1926 (i.e., following the Religious Endowments Act of 1925; see Mudaliar 1974). Under provisions of the accepted scheme, the temple's board of trustees was to consist of five members: two from the four traditional karyakkarar families, and three to be appointed by the court.

8.  The Ilayathakudi Temple Report (Ilayathakudi Devasthanam 1939) may even omit reference to inam holdings outside the temple's immediate vicinity. My impression is that temple devastanam lands included zamindari or inam lands in Tanjavur; however, this impression requires confir-

mation. If I am correct about these omissions as well as omissions of non- inam investments, the selectivity of information contained in the report is somewhat puzzling and raises questions about the purpose of the report. I note, however, that, unlike the temple's account books, the administrative report was prepared in both Tamil and English and was still freely distributed to temple visitors forty years after its publication. In addition, before the report presents its description of selective temple expenditures, it describes the history of Nakarattar involvement with the temple and the court decisions that recognized the temple as privately owned by the Nakarattar community. For these reasons, it seems likely that the report was designed primarily for an external audience, to demonstrate how responsibly Nakarattars were meeting their obligations as owners.

9. This is not to say that Nakarattars understood or subscribed to Western, biological theories of genetic heritability. Rather, they followed their own version of Hindu beliefs about membership in different kinds of descent groups. This Nakarattar understanding of descent was similar to relevant tenets of Hindu textual traditions exemplified in the Dharmasutras, the early Dharmasastras, and Kautilya's Arthasastra. The texts maintain that every person is born into a particular caste ( jati ), clan ( kula ), and family ( parivara ). Membership in these groups is determined in part by genetic transmission of the male "seed" from the Vedic ancestor of an individual's lineage ( gotra ) and in part by the pure or impure combination of this seed with the female "field" in which it quickens and grows. Specific combinations of these parental contributions produce offspring with a characteristic blend of elemental substances ( gunas ) and concomitant moral codes for behavior ( dharma ). Shared parentage produces shared substances, shared moral codes, and comembership in common descent groups. The consequences of mixed marriages (i.e., marriages between members of descent groups possessing different blends of gunas ) for generating a multitude of descent groups are the topic of considerable speculation. For Indological and ethnographic interpretations of the relevant texts, see Davis (1976). Inden and Nicholas (1977). Kane (1941 v. I, II), Marriott (1976), Marriott and Inden (1974). Rocher (1981), and Tambiah (1973a). For a detailed cultural account of a folk Tamil version of these beliefs, see Daniel (1984).

10. The Nakarattar conjugal family, the pulli , employed principles of marriage as well as descent in recruiting its members. That is, husbands and wives became members of a pulli at the time of their marriage, in contrast to legitimate children, who were born or adopted into the pulli . I distinguish the Nakarattar joint family, the valavu , from various forms of joint households for which it sometimes formed a basis. Valavus are accurately regarded as minimal lineage segments subject to the application of Mitakshara laws regarding inheritance. For more detailed discussion of the relationship between Hindu concepts of joint family and sociological considerations in the definition of joint households, see Tambiah (1973a: 75).

11. The extent to which ideas about inherited moral codes were fostered or even created in some novel fashion by interaction with the colonial government is difficult to gauge. Recent studies (Carroll 1978; Cohn 1960, 1983) seem almost to suggest they were a product of the colonial period. Against such a possibility, it must be remembered that the Nakarattar jati and its descent groups were in place by the seventeenth century (see Chapter 7).

12. The processes that generated growth in the number of Nakarattar residential villages have apparently been reversed in recent times as the number of villages that Nakarattars acknowledge as under their control has diminished. Nakarattars attribute this reversal to a Nakarattar migration to larger towns that depleted the population of the smaller villages.

13. The seasonal festivals of village temples included tirttan kutittal or marantu kutattal (drinking sanctified water in the seventh month of pregnancy), putumai (child's first birthday), suppiti (boy's coming-of-age ceremony), tiruvatirai (girl's coming-of-age ceremony), kalyanam (marriage), shastiaptapurti kalyanam (man's sixtieth birthday and marriage renewal ceremony). For a description of these ceremonies, see Thurston (1909).

14. See Harper (1959) for a discussion of similar beliefs about ghosts ( peys ), spirits ( avis ), family deities ( kula teyvams ), and village deities ( grama teyvams ).

15. See Dumont (1959) and Whitehead (1921) for, respectively, structural and descriptive accounts of the standard non-Brahmanic village pantheon of South India.

16. Mines (1984) describes processes of deification and cult formation among Kaikkolars that are quite similar to those exhibited by Nakarattars. In addition, he describes what might be considered a two-tier system of temples: one set concerned with matters of kinship, the other concerned with matters beyond the realm of kinship (e.g., trade and the state). Like Kaikkolars, Nakarattars also participated in supra-kinship temples. But these did not correspond to either the village or the clan temples described in the text. Nor, for that matter, did they correspond to any intermediate form of quasi-kinship temple. Rather, they formed a third tier of temple organization. Notice that it is not clear from Mines' account whether Kaikkolar temple-based cults of descent were segmented into two tiers.

17. Most devastanam inam lands had been given to the Nakarattar clan temples in past centuries. The gifts are often recorded in stone inscriptions carved on the temple walls. They are also recorded in the Inam Register of the colonial government. For a large but incomplete listing of Nakarattar money endowments during the late colonial period, see Ramanathan Chettiar (1953).

18. In general, care should be taken not to overemphasize the solidarity of kuttikkira pankalis . One of the fiercest political fights for control of the

North Ramnad District Board in 1937 occurred between the son-in-law (an affinal relation) and a member of a collateral pankali of the powerful Nakarattar leader Raja Sir Annamalai Chettiar. In this instance, the son-in-law, C. V. Ct. (Venkatachalam) Chettiar ran on the ticket of the Justice party as an ally of Annamalai. But the Raja's pankali , Nakappa Chettiar, ran on the Congress party ticket. One should not conclude from this alliance, however, that the converse relationship is therefore all-important. Both are open-ended relationships with options to renew or terminate. In this case, the S. Rm C. V. Ct. alliance was not permanent. A few years later, C. V. Ct. attempted to thwart Annamalai's plans to initiate Annamalai University by advising Kalimuthu Thiagarajan on tactics for gaining governmental support of a Thiagarajan University in Madurai. Ultimately, Annamalai won. Thiagarajan University was never founded (Lakshman Chettiar interview, June 22, 1981; see also Arooran 1980). For more about this story, see Chapters 7 and 10.

1. Specifically, the new group of "descendants" claims hereditary rights for one of its members to act as sami ati (literally, "god dancer"), a medium possessed by Murugan in the Tai Pucam pilgrimage festival. Some aspects of the post—seventeenth-century history of Nakarattar involvement in the Palani festival are described in Moreno (1981).

2. Following Lévi-Strauss, many anthropologists confuse semantic distinctions with jural norms, arguing that some kind of positive marriage obligation is created by a Dravidian kin classification. I have dealt with the disciplinary history of confusion surrounding this issue, elsewhere (Rudner 1990; see also Needham 1986). I mention it in this context to underscore the radical departure from most standard models of Dravidian kinship represented by Nakarattar resistance to self-perpetuating alliances.

3. In order to avoid any risk of being misunderstood, I wish to note that I am not saying that Nakarattars lacked corporate descent groups, but only that Nakarattars did not generally establish and perpetuate these groups through marriage alliance.

4. See Baker (1976) and Washbrook (1973, 1975, 1976) for pertinent descriptions of elite South Indian colonial politics. Baker's book, in particular, contains relevant accounts of the Ramnad elections of 1937 and of the role of Raja Sir Muthia Chettiar in Madras. For additional details, see Chapter 7, above.

5. For more recent analyses of the organization of caste groups see Beck (1972) and Dirks (1979, 1983, 1988).

6. This generalization about clan and lineage control of Chettinad villages does not apply to clusters of devastanam villages owned by both clan and village temples. But such clusterings do not affect my claim. Devas -

7. Readers unfamiliar with the variety of standard schemes available for transliterating Tamil should note that Beck (1972) renders the caste name Goundar as "KavuNTar."

8. It is noteworthy that other analysts dispute Dumont's (1957b) claim that hierarchy informs every social relationship in Hindu society. Anthony Carter (1973), for example, argues that intracaste relations differ from intercaste relations precisely in virtue of their absence of hierarchy. I suspect that both Dumont and Carter have overgeneralized from their data: the military castes with which Dumont was most familiar were internally ranked (see also Dirks 1987), and the castes with which Carter was most familiar resembled the Goundar of Coimbatore and mercantile castes such as the Nakarattar and Kaikkolar (see Chapter 2, above). For a general discussion of the history and present status of cross-cultural studies of such organizational differences outside of India, see Service (1985: Chapter 9).

9. The Pudukottai Kallar differed from Dumont's Pramalai Kallar primarily in the degree to which they emphasized kingship as a concept for defining and ordering ritual and political relationships. For this reason, they are especially useful for delineating a contrast with a group such as the Nakarattar.

10. Other forms of service were also exchanged for land control and honors, especially by non-Kallar subjects. Typically, all such transactions were mediated by a redistributional ritual in the context of cult worship, within the system of temples described above. For example, a king might grant an inam to a retainer who would act more or less as a trustee. Produce from the land would be donated to the deity of a temple, consecrated by the appropriate act of worship, and redistributed to cult members in such a way that it symbolized and even created the status approved by the king. In this way, the king had considerable powers to influence the hierarchy of Kallar social groups.

11. Pudukottai Kallars refer to these regions also as natus , but to keep them distinct from the encompassing natus associated with caste groups I will refer to them as sub- natus .

12. Cervaikkarars are not invariably selected from the royal natu because considerations of service and honor override considerations of

descent at this level of organization.

13. Dirks (1983: 30) describes the system in the following way: "There are lineage deities, village deities (sometimes the deity of the highest lineage but often separate), and subcaste deities.... Temples both served to establish a unity—to represent a whole—and to gradate and rank the parts of that whole. Membership in a village, a lineage, and a subcaste was ultimately talked about in terms of whether one had kaniyacci (a right to worship and receive temple honors) in the relevant temples; while all kaniyaccikkarars were equal in that they all held equal right ( urimai ) or share ( panku ) to participate in the affairs of the temple, the nature of participation was ranked. The ampalams (headmen of subcastes) were honored first, followed in order either by a ranked list of villages (represented by their respective ampalams ) or of lineages (likewise represented)."

14. The episode referred to in the text (the Pandyan king's abduction of a Nakarattar maiden) takes place after various catastrophic adventures in the Chola port city of Kaveripumpattinam, prior to the Nakarattar migration to Chettinad. The most easily available English version of the Nakarattar varalaru remains Thurston (1909), which, however, leaves out many details of the varalaru published by the Nakarattars themselves in Tamil (e.g., Sharma 1970) and familiar to present-day Nakarattars as part of their oral tradition. In particular, although Thurston's account presents an incident of a royal Chola kidnapping of a Nakarattar girl and its consequences, it does not offer the story of the Pandyan kidnapping described in the text.

15. Narayana Rao (1986) provides one major exception to the general lack of systematic attention to caste histories. Although the Telugu castes —the subjects of his analysis—do not employ the left-hand/right-hand metaphor to distinguish themselves, and although Rao identifies structural features in their histories that are different from those which emerge in Beck's analysis, Rao nevertheless finds a mercantile/agrarian division of castes similar to Beck's finding in Kongu. Kenneth David (1977) provides the only other systematic analysis of caste histories of which I am aware. But his discussion of jati varalaru in a Sri Lankan context inexplicably fails to address the Nattukottai Chettiar, even though they figure in other parts of his essay.

16. For further discussion of the "economic" model of variation between right- and left-hand castes, see, in addition to Beck (1972), Mines (1984). For a political interpretation of the variable occurrence of right-hand and left-hand factions in South Indian social structure, see Appadurai (1974). Stein (1980) provides a view of the matter that seems to share elements of both positions.

17. I am not suggesting that Beck (1972) has mischaracterized Kongunadu, but that Kongunadu should not be taken as paradigmatic for all South India. There is no need to belabor the ethnographer's perennial

18. An early version of this section was originally published in Rudner (1989).

19. "The media of Hindu transactions are substance-codes that may be scaled from the relatively 'gross' ( sthula ) to the relatively 'subtle' ( suksma ). 'Gross'—that is, lower, less refined, more tangible, and less widely transformable substance-codes—are contrasted with higher, less tangible substance-codes that are "subtler," more capable of transformation, and therefore imbued with greater power and value. For example, knowledge may be considered subtler than money, and money subtler than grain or land, but grain or land not so gross as cooked food or garbage, which have less power of generation. Such a scale may be understood as resembling the distinctions among communications codes capable of generating more and fewer messages; but Indian thought understands subtler substance-codes as emerging through processes of maturation or (what is considered to be the same thing) cooking. Thus subtler essences may sometimes be ripened, extracted, or distilled out of grosser ones (as fruit comes from Plants, nectar from flowers, butter from milk); and grosser substance-codes may be generated or precipitated out of subtler ones (as plants come from seed, feces from food)" (Marriott 1976: 110).

20. See Chapter 5 for a full description of the various types of Nakarattar deposits.

21. Some readers will be unable to free themselves from suspicions that talk about the social functions of Nakarattar institutions necessarily implies analogy with physical or biological systems in equilibrium. Yet alternative metaphors for nonequilibrium systems are readily available. One obvious physical analogy that suggests itself is the chaos of fluid turbulence—a concept that has developed increasing fascination for both anthropologists and "harder" scientists during the last decade. Anthropologists, for diverse reasons, have become increasingly resistant to considering any contribution from the nonsocial sciences. But affinities remain: affinities that, in some cases, make explicit appeals to metaphors of fluid dynamics (e.g., Daniel 1984; Marriott 1990). All that is lacking in such anthropological interpretations is a historical perspective that looks beyond the forces at work in a single moment to the dynamics of those forces over a time series of sociocultural transformations.

Here lies the real attraction of new theories about fluid turbulence as analogies or even models for nonequilibrium, complex behavior, generally. We may never be able to describe all the individual interactions that constitute a dynamic social system. But, like very complex nonsocial systems, the behavior these interactions exhibit is hardly random. Nor are the transformations they undergo. Both social and nonsocial systems manifest collective modes of behavior, describable in terms of restrictions on possible transformations. A turbulent fluid, for example, has a potentially infinite number of degrees of freedom. Yet only a very small number of possibilities are ever realized, and these possibilities can be related to the external environment of the fluid, regardless of its initial conditions and complexity. Moreover, under certain conditions, there can exist large-scale, dissipative structures in fluid turbulence that have predictable dynamics in the short term, that persist in the long term, and that alter in predictable ways under changed environmental conditions. We do not know whether such systems can serve as models for social systems. But the possibility is intriguing and suggests interesting questions about the patterning of social life, even if these are not answerable at present. For an excellent, nontechnical introduction to the multidisciplinary work in this area see James Glieck (1987). For a recent effort to apply theories of complex systems to modeling social systems, see Lansing (1993).