* This description of current customs draws upon field work conducted in the Coimbatore District of Madras State during 1965 and 1966. Historical statements, unless otherwise specified, refer to the area of South India covered by the present states of Mysore, Andhra, and Madras. Literary sources rarely specify an area more precisely than this. This paper has used a scheme of letter by letter transliteration from written Tamil. With the following exceptions, common English letters are used as direct equivalents for Tamil ones:

a) The capital letters L, N, and T represent retroflex sounds in the middle of a word.

b) n and ṉ distinguish two Tamil letters, both pronounced like the English n.

c) ṅ is equivalent to the English ng.

d) ā, ē, ī, ō, and ū represent long as opposed to short vowels.

A guide to accurate pronunciation necessitates a long list of rules, and would require too much space to provide here. Only two difficult words recur frequently in the text. The first is the name of a region, Koṅku, which should be pronounced like “Kongu,” and the second is the name of a caste, KavuNTar, which should be pronounced a little like “Gounda(r).”

¹ Arokiaswami, M., The Kongu Country (Madras: Madras University Press, 1956) p. 272.Google Scholar

² The roots of the division were probably economic, but evidence suggests that the initial separation of social groups on these grounds became blurred with time. Not surprisingly, the ritual rivalry of the two blocs appears to have gradually increased to compensate for this sense of confusion. By the time the British arrived in South India petty bickering between the divisions was intense, while the cause of the opposition left these foreign observers bewildered. See Stein, B., “The Integration of the Agrarian System of South India” in Frykenberg, R., ed., Social Structure and Land Control (Madison: University of Wisconsin Press, 1968)Google Scholar for a general discussion of the conflict and its development.

³ Love, Henry D., Vestiges of Old Madras (London, John Murray, 1931), vol. II, p. 25 and ff. and pp. 142–43Google Scholar, Bertrand, P. J., La Mission Du Maduré (Paris: Librarie de Poussielgue-Rusard, 1947) pp. 78 and 81Google Scholar; Frykenberg, R., Guntur District 1788–1848 (Oxford: Clarendon Press, 1956) p. 114Google Scholar; and Srinivasachari, C. S., “Right and Left Hand Caste Disputes in Madras in the Early part of the 18th Century,” Indian Historical Records Commission: Proceedings of Meetings, vol. XII. (Calcutta: Government of India, 1930) pp. 68–76.Google Scholar

⁴ The term peasant refers to anyone directing agricultural activities on a day-to-day basis, whose instructions are normally decisive in the final division of the produce. Peasant in this sense refers to a powerful, landed community, not to field labourers.

⁵ Stein, Burton, “Brahman and Peasant in Early South Indian History” in The Adyar Library Bulletin, vols. XXXI–XXXII, 1967–1968Google Scholar, “Dr. V. Raghavan Felicitation Volume.”

⁶ Stein, Burton, “The State and the Agrarian Order,”Google Scholar lecture delivered at the University of Chicago, Mar. 31, 1969, as yet unpublished.

⁷ Koṅku comprises most of the District of Coimbatore, plus the southwestern part of Salem, and parts of Tiruchirappali and Madurai. It consists of one large upland plain, surrounded by mountains and watered by the upper reaches of the river Kaveri.

⁸ Locally the Kammālaṉ are referred to as Ācāri and the Cakkiliyaṉ as Mātāri. Indeed only one of my five informants (a second Cakkiliyan) was content to label his own bloc as “left.” This suggests that the labels do cary overtones of rank and that “right” is the preferred and implicitly superior category. The probable reasons which lie behind these conflicting claims are discussed further in the next section. One may further note that the tendency to reverse the labels of two opposed divisions, where the reversal favors one's own group, is not unique to my data. See, for example, Paul Radin's description of the phratries of the Winnebago tribe, Two of Radin's informants (one from the “upper” and one from the “lower” phratry) gave him inverse descriptions of the position of clan lodges, Each placed the leading clan of his phratry in the southernmost (most prestigious) location. See Maybury-Lewis, D., “The Analysis of Dual Organizations: A Methodological Critique,” in Bijdragen, vol. 116 (special issue entitled Anthropologica), 1960, pp. 18–19.Google Scholar

⁹ The kirāmam, a territory some five to ten square miles in size and containing several discrete settlements, is the traditional economic and ritual unit in rural areas. It corresponds in significance (if not always in exact dimensions) with the panchayat territories of post-independence reorganization. It is the unit into which British land records were subdivided and within which a local government appointee was responsible for revenue collection. Current usage of the term “village” is ambiguous, since some authors use it to refer to a unit like the kirāmam, and others to refer to a discrete settlement (ūr). Much argument over the significance of the Indian village as a unit of study has been generated by this confusion.

¹⁰ This point is discussed at length in Dumont, Louis, Homo Hierarchicus (Paris: Gallimard, 1966).Google Scholar

¹¹ See McKim Marriott's extended treatment of this situation in “Interactional and Attibutional Theories of Caste Ranking,” Man in India, vol. 39, no. 2, 1959, pp. 92–107Google Scholar and “Caste Ranking and Food Transactions: A Matrix Analysis” in Singer, Milton and Cohn, Bernard eds., Structure and Change in Indian Society, (Chicago, Aldine, 1968) pp. 133–172.Google Scholar

¹² This choice is beautifully symbolized by the act of plowing. According to orthodox Hindu docsignificance trine, plowing land is a polluting activity since overturning the soil threatens life within it. (See, for example, Crooke, W., The Tribes and Castes of the N.W. Provinces and Oudh, vol. I, Government of India, Calcutta, 1896, p. cxlixGoogle Scholar). On the other hand, plowing is an essential agricultural activity, directly linked to high productivity. A Brahman will never plow himself but always hire a man of another caste to do this task for him. A caste which has opted to contest its rank in these worldly terms, by contrast, will plow and do so with pride. Steve Barnett of the University of Chicago has recently collected data of great interest on this point. He worked with a caste who stand just below the Brahmans in ritual status. This caste attempts to compete both in orthodox, ritual terms and in this worldly terms as landowners. Their decision about the plow balances the two views as delicately as would seem possible, yet ultimately opts for the orthodox position. They will touch the plough during an annual ceremony, yet will not use it to actually turn over earth in cultivation. The KavuNTars in Coimbatore, by contrast, boast of their ploughmanship to the point where a recent book about the caste published by one of its leaders has a vivid color picture of a man wielding a plow on the front cover.

¹³ These are the Tamil terms. There are equivalents which can be given the same English transalation in Kanada and Telugu, the other two principal languages spoken in areas where the right-left division was once prominent.

¹⁴ Other examples of this are a local saying to the effect that “if the breath be taken in through the right nostril at conception the child will be male; if through the left a female,” and the popular belief that bodily features such as moles and muscular ticks are auspicious when they occur on the right side of a man's body, or on the left side of a woman's. Furthermore, a woman is sometimes asked to use her left foot or hand to contrast with the male right in rituals where both sexes participate.

¹⁵ Indeed, it is difficult to find any conceptual antithesis which is truly symmetric except in a formal sense. Nearly all such oppositions have some evaluative connotation implying a superiorinferior relationship. It is not surprising, therefore, to find this problem of relative value or rank intruding at the level of social groupings wherever such a pair of terms has been used to describe them. This point has been made previously by Maybury-Lewis, op. cit., p. 42.

¹⁶ Coomaraswamy, A. K., Spiritual Authority and Temporal Power in the Indian Theory of Government (New Haven: American Oriental Society, 1942) p. 1–3.Google Scholar

¹⁷ See Dubois, Abbé, Hindu Manners, Customs and Ceremonies (Oxford: Clarendon Press, 1966) p. 25Google Scholar. Burton Stein, in a personal communication, mentions that he has found similar examples of attempts at reversal of the two labels in early temple inscriptions which he has studied. Furthermore, in the 18th and 19th centuries die Brahmans brought several law suits against the artisans in an attempt to dampen their exalted claims to Brahman style knowledge and status. See Oppert, Gustav, On the Original Inhabitants of Bharatavarsa or India (London: Westminster, 1894) p. 58–63.Google Scholar

¹⁸ Subcaste is a very difficult concept to define. Here the term will refer to a group within a general occupational or social category which identifies itself by an additional modifying adjective (usually one with ritual significance) and which attempts to impose sanctions on marriages across its boundaries. Subcastes are primarily kin groupings and only secondarily occupational units.

¹⁹ Up to this point differences between the two divisions have been described by reference to castes as the basic social unit. This was done in order not to confuse the reader with a caste-subcaste distinction prematurely. Strictly speaking, however, the observations contained in this paper refer to subcastes. It is here that detailed markers of ritual and social custom take on their importance and it is at this nodal point where identification with one division or the other was ultimately affirmed. According to available historical sources, when switching between blocs took place members of a subcaste moved across the divide together.

²⁰ A particular diet, either vegetarian or nonvegetarian, and including specific restrictions on certain animals, is socially sanctioned by a subcaste as a whole. This is asserted as the norm when ranking the group vis-à-vis outsiders, though individuals within the subcaste may at times deviate from the rule in practice.

²¹ See Hare, P. V., History of Dharmasastra (Poona: Bhandarkar Research Institute, 1941) vol. II, pt. I, pp. 274–96 and pp. 583–636.Google Scholar

²² The term KavuNTar is related to the Sanskrit grama kuta, a title meaning “village or regional chief of Shudra origin.” See Emeneau, M. B. and Burrow, T., Dravidian Borrowings from Indo-Aryan (Berkeley: University of California Press, 1962) p. 21.Google Scholar

²³ In such an undertaking one would have to take account of local, regional, and all-India sources, and of the considerable disagreement within as well as between such materials. Nonetheless, such work must form the critical base for the initially seminal, but currently vague and overworked concept of “Sanskritization.” For work to date on Sanskritization, see Srinivas, M. N., Social Change in Modern India (Berkeley: University of California Press, 1966).Google Scholar

²⁴ The rank of a caste was determined by analyzing the results of twenty-five systematic interviews with members of thirteen castes. In the interviews one male and one female from each caste (with one exception) were asked to evaluate all the others in terms of restrictions they applied to those communities concerning house entry, seating, acceptance of food, and bodily contact. The fewer the restrictions the higher the score assigned to any particular caste evaluated. For details of the ranking procedure see the author's doctoral dissertation, “Social and Conceptual Order in Koṅku: A Region of South India,” Oxford University, 1968, pp. 460–464Google Scholar. The method used closely resembles that described by Pauline M. Kolenda in her article “A Multiple Scaling Technique for Caste Ranking,” Man in India, vol. 39, no. 2, pp. 128–47.Google Scholar

²⁵ The service castes are not adequately represented in my field material and their full complement, therefore, does not appear in the chart. Traditionally these service groups, particularly barbers, washermen, and drummers, were divided. Some subcastes served the right bloc of castes and the other subcastes served the left. Presumably if sample were larger, subcastes identifying with each division would follow the same pattern and fall on the same half of the chart as do those they traditionally served.

²⁶ See the appendix for a discussion of how the regression lines were drawn and for a justification of the contrast in relative slope.

²⁷ Riker, Theodore, A Theory of Political Coalitions (New Haven: Yale University Press, 1962).Google Scholar

²⁸ Many sources indicate that this was the case. See for example, Hutton, , op. cit., p. 9Google Scholar, Dubois, , op. cit., p. 25Google Scholar and Reddi, N. Subba, “Community Conflict among the Depressed Castes of Andhra,” Man in India, vol. 30, no. 4, pp. 1–12.Google Scholar

²⁹ The current stress on “Black is Beautiful” may be an example of this.

³⁰ Various details which are cited in the literature about Tanjore support this conclusion. See, for example, Gough, E. K., “Brahman Kinship in a Tamil Village,” American Anthropologist, vol. 58:5, p. 827, and p. 49.Google Scholar

³¹ Being a landlord requires more worldly interaction with men of other castes than is seemly for a priest. It also limits the amount of time one can spend in ritual-related activities.

³² Burton Stein, personal communication.

³³ Joan Mencher indicates, in a recent article, that the Nayars of Central Kerala resemble the Nāmbudiris very closely indeed. See her “Nāmbudiri Brahmans: An Analysis of a Traditional Elite in Kerala,” Journal of Asian and African Studies, vol. I:3, PP. 183–196Google Scholar. An explanation of the absence of a right—left division in Kerala lies, perhaps, in the unusual position of the Nāmbudiri Brahmans in this area. Unlike their counterparts in other areas of the South, the Nāmbudiris of Kerala have a tradition of “marriage” with the Nayar community whose social position falls just beneath their own. This intimate linkage of the two groups has reinforced Brahman preeminence in the region. Furthermore, in much of Kerala the Nāmbudiris are the major landowners and in areas where they do not control land directly their customary behavior continues to serve as an uncontested model for the region. For further detail see Miller, Eric, “Caste and Territory in Malabar,” American Anthropologist, No. 56, 1954Google Scholar; Mayer, Adrian, Land and Society in Malabar (Oxford: Oxford University Press, 1952)Google Scholar; and Gough, E. Kathleen, “The Nayar” in Schneider, David and Gough, E. K. eds. Matrilineal Kinship (Berkeley: University of California Press, 1961)Google Scholar. The possible connection between Nayar hypergamy, Brahman emulation, and the absence of a right-left division was initially suggested to me by David J. Elkins, Department of Political Science, University of British Columbia.

³⁴ Artisan groups all over India refer to themselves as “Visvakarma Brahmans,” a term that refers both to the god Visvakarman or “world-creator” of the Rg. Veda and to the demi-god Visvakarman or “craftsman-creator” of the Ramayana. Use of the term today plays on this double meaning, indicating that group members see their status as craftsmen in this world as a counterpart to that of divine creator in another. Because of their insistence on familial purity and their refusal to accept food from other high-ranking communities, they tend to be difficult to place in a unidimensional hierarchy. In some places they are also associated with extremist and puritanical sectarian movements. Sec such studies as Mayer, Adrian, Caste and Kinship in Central India (London: Routledge and Kegan Paul, 1960)Google Scholar; Pocock, David, “The Movement of Castes,” Man 1955, no. 79, pp. 71–2Google Scholar; and Channa, Dev Raj, “Sanskritization, Westernization and India's Northwest,” Economic Weekly, vol. XIII, no. 9, 03 4, 1961, pp. 409–414.Google Scholar

³⁵ The Heir, the Viceroy, the General and the President stood on the right, while the Master of Ceremonies and the Minister of Commerce stood on the left. Hocart, , op. cit., p. 208.Google Scholar

³⁶ Hocart, , op. cit., pp. 208–9Google Scholar; and Yalman, Nur, “Dual Organization in Central Ceylon,” Journal of Asian Studies, vol. XXIV, no. 3, (1965) pp. 197–223.Google Scholar

³⁷ Dubois, , op. cit., p. 25Google Scholar. Also Nelson, J. H., The Madura Country (Madras: Government Press, 1868) pt. II, p. 4.Google Scholar

³⁸ Mukherjee, N., The Ryotwari System in Madras (Calcutta: Mukhopadhyay, 1962) pp. 332–3 and 346Google Scholar. See also Kumar, Dharma, Land and Caste in South India (Cambridge: Cambridge University Press, 1965) p. 34.Google Scholar

³⁹ Ibid., Kumar, Dharma, p. 9.Google Scholar

⁴⁰ This appears to be a very old and persistent characteristic of Indian society. There are references to a similar separation as early as the second half of the first millennium B.C. See Romila Thapar, “The Elite and Social Mobility in Early India,” paper read at the University of Chicago, Apr. 1968, pp. 9–11, as yet unpublished.

⁴¹ Polanyi, Karl, The Great Transformation (Boston: Beacon Press, 1944) p. 178.Google Scholar

⁴² Kumar, Dharma, op. cit., p. 192Google Scholar and Frykenburg, R., op. cit., p. 2.Google Scholar

⁴³ Metcalf, Thomas R., The Aftermath of Revolt (Princeton: Princeton University Press, 1964) p. 178.Google Scholar

⁴⁴ See Rudolph, Lloyd, The Meaning of Party: From the Politics of Status to the Politics of Opinion in Eighteenth Century England and America, unpublished Ph.D. dissertation, Harvard, Department of Government, 1956, pp. 83–86Google Scholar; Bulmer-Thomas, , The Growth of the British Party System (London: John Baker, 1965) vol. I, p. 9Google Scholar; and Mitchell, Austin, The Whigs in Opposition, 1815–1830 (Oxford: Clarendon Press, 1967) pp. 10–17.Google Scholar

⁴⁵ This description is actually for the end of the 16th century, but it is said that the rituals which were associated with Parliamentary sessions persisted nearly unchanged in later periods. See Neale, J. E., The Elizabethian House of Commons (London: Jonadian Cape, 1949) p. 421Google Scholar. It is interesting that when the House of Lords met alone the representatives of the Church stood on the right of the sovereign and the temporal peers on the left. With only the landed present, the Church was given the place of honor, just as the Brahmans take ritual precedence over kings in India.

⁴⁶ Bulmer-Thomas, , op. cit., p. 7Google Scholar, and Love, , op.cit., vol. II, p. 143Google Scholar. It is interesting that the landed group, in both cases, preempted the color red, possibly because of its ready association with physical prowess and military valor.

⁴⁷ Mitchell would date the beginnings of change from the period of the great Reform Bill of 1832, op. cit., p. 4. See also Polanyi, K., op. cit., p. 180.Google Scholar

⁴⁸ Some researchers have suggested that all societies depend to a greater or lesser extent on a set of core themes, and that these themes are generally expressed in terms of a few elementary dichotomous categories. If this is the case, then the question becomes why these themes, so common at an ethical and linguistic level, find expression in the actual organization of some societies, but not in others. More specifically, examples of dichotomous social organization (as opposed to dichotomous categories of thought) come largely from “primitive” or at least pre-industrial societies. This paper suggests, implicitly, that such a finding can be connected with economic facts. Societies which support themselves by means of a few relatively discrete and straight forward economic activities often have the kinds of basic social cleavage (due to specialization in easily contrasted activities) which lend themselves to symbolic elaboration. As the economic base of a group becomes more complex, however, it would appear that these dualistic aspects of social organization “fit” less well with reality. At the same time, such practical complexity does not lend itself as easily to symbolic manipulation. It is hypothesized, therefore, that a pattern of dual organization (where previously existent) will tend to disappear under conditions like those described in this paper. For detail on the general assertion concerning the dualistic nature of major cultural themes see Durkheim, Emile and Mauss, Marcel, Primitive Classification, Needham, Rodney, trans., London, 1963Google Scholar, and Hertz, Robert, Death and the Right Hand, Needham, Rodney, trans., London, 1960.Google Scholar