Nair-structure

The Internal Structure of the Nayar Caste
Author(s): C. J. Fuller
Source:, Vol. 31, No. 4 (Winter, 1975), pp. 283-312
Published by: University of New Mexico
Stable URL: http://www.jstor.org/stable/3629883

The matrilineal Nayars of Kerala in southwest India constitute a populous caste, which contains within it a large number of subdivisions. In contrasto the ideal model, on which most standard descriptions appear to be based, these subdivisions fall into several analytically distinguishable categories. There is a core of a few large, stable subdivisions and afringe of many small, unstable subdivisions. The latter are created and eliminated by several processes, one of which, treated in detail, is connected to the famous, hypergamous marriage system of the Nayars. The statuses of tali-tiers and sambandham partners act as diacritical markers of a subdivision’s status. Modern changes in the subdivision system are briefly considered A distinction between stable and unstable statuses in the caste system is introduced, and doubt is raised about Dumont’s and Pocock’s theories of caste.

SOME YEARS AGO, Pocock (1957) drew attention to a number of the sociological problems posed by a populous and widely distributed caste like the Patidar of Gujarat. In this respect, the Nayars of Kerala, whose estimated population in 1968 was around 2.9 million, or 14.5% of Kerala’s total population (Govt. of Kerala 1971: App. xviii), closely resemble the Patidars. The Patidars, however, differ from the Nayars in having but few named subdivisions (Pocock 1972:56). The Nayars' marriage system has made them one of the most famous of all communities in anthropological circles.

This paper attempts to analyze the internal structure of the Nayar caste. It does not set out to analyze the Nayars’ marriage system per se, although it does try to show how the famed cross-caste hypergamy of the high-ranking Nayars-involving Brahmans, Kshatriyas, and Samantans-was linked to the subdivision system within the Nayar caste.

1 I would like to thank Jonathan Parry for his useful criticisms of an earlier draft of this paper. 283

284

INTRODUCTION

Before Indian Independence, the region which now forms the mainly Malayalam-speaking state of Kerala, in southwest India, was divided into three parts: Malabar District, under direct British rule, in the north; Travancore, a “Native State” ruled by its own maharaja, in the south; and Cochin, another “Native State,” a tiny kingdom sandwiched between the two much larger units, in the center. North Malabar, the region to the north of Calicut and the Kora River referred to by Gough (1961:305) as “North Kerala,” is not discussed in any detail in this paper. The Nayars in that area had a kinship and marriage system significantly different from that of their caste-fellows to the south, and the internal structure of the caste in the north, so far as we can judge from the scanty data on it, also differed markedly from that found further south.

South Malabar and Cochin, which Gough (1961) called “Central Kerala,” is the region about which we know most. Here, the Nayars lived in large matrilineal joint families known as taravads. Marriage, as it is commonly understood, was not found among them. Before puberty, girls underwent a ritual known as the talli-rite and, when they became sexually mature, took one or more “lovers” in a relationship known as sambandham. By these lovers, they bore children who were members of their own taravad. The talli-rite and the sambandham union are discussed later in more detail. Central Kerala is the region where the famed “polyandry” was most prevalent, and it is there, too, that the hypergamous system reached its fullest efflorescence. The extreme north of Travancore (approximately equivalent to the northern part of the modern Ernakulam District) should, I think, be included in Central Kerala on cultural grounds, for the principal features of the Central Kerala Nayars’ social system seem to have been present in northern Travancore too. The area lying south of this northerly part, and north of a line drawn east-west across the center of Travancore, more or less coincides with the region known locally as “Central Travancore.” Here, the main features of the Nayars’ kinship and marriage system seem to have been basically similar to those found in Central Kerala, but partly because of the relative paucity of Brahmans-hypergamy was probably neither so prevalent nor so spectacular. Compared with data on Central Kerala, information on Central Travancore is sparse and uneven in quality, although my own field experience in this region has enabled me to make some sense out of some of it.

The data on southern Travancore are worst of all, and this is particularly regrettable because it is almost certain that the kinship and marriage system of the Nayars in this area differed substantially;

285 although Nayars belonged to matrilineal joint families, they had more or less monogamous marriages with wives residing in their husbands' taravads (Aiya 1906:358). But despite these major variations, the data on subdivisions among the Travancore Nayars are nowhere (to my knowledge) broken down by geographical area, but are presented as if they applied to the entire state. This is unlikely to be true. Since I am so unsure about the southern Travancore situation, I make no claim that any of the remarks herein apply to southern Travancore, other than extremely tentatively.

Virtually all of what is discussed in this paper is now part of history. To summarize, during the nineteenth century the tali-rite became progressively less and less important, and the sambandham union was transformed into a more or less monogamous and permanent union of a kind very similar to marriage elsewhere in India. Taravads, too, broke up, and smaller family units became the norm. The process of change occurred earlier and faster in Central Travancore than it did in Central Kerala; in the former region, the tali-rite had virtually died out by 1900, and by around 1925, practically all Nayars entered into monogamous unions which were intended to be permanent. In the same period, taravads were in an advanced state of disintegration, and by the 1940’s, the nuclear family had become the norm. By contrast, in Central Kerala, the tali-rite persisted until the 1930’s, and, even in the 1950’s and 1960’s, a significant proportion of Nayars was still involved in sambandham unions in which husbands did not live with their wives, but only visited them. In this region, too, many taravads did not finally break up until the 1950’s. However, the direction of change is clear enough, and it seems unlikely that it would halt at that point. In both regions, Nambudiri Brahmans mostly stopped taking Nayar women as lovers in the 1920’s. As we shall see, these changes in the marriage system must almost certainly have led to major changes in the subdivision system. Consequently, the analysis of this system which I present must be taken as applicable only to the past.

NAYAR SUBDIVISIONS: PREFATORY REMARKS

One possible source of confusion, present in certain older sources (e.g., Logan 1887:133-134), needs to be eliminated immediately. Each male Nayar bears a caste title as his last name; examples include Nair (the usual anglicized spelling in Kerala), Kurup, Menon, Pillai, and so forth. In the past, it is said, these titles were conferred as marks of honor by kings and chiefs, although some of them refer to specialized occupations. The title could be granted to any Nayar, irrespective of his subdivision, and thus give no indication of the latter. Further, there was

286 nothing to prevent Nayars awarding themselves these titles, as they certainly used to and perhaps still do (C. A. Menon 1911:202). Several Travancore census reports provided some information about the Nayar subdivisions (vibhagam). The most detailed was that of 1901, in which more than 130 subdivisions were discovered by the enumerators; these were compressed into 44 distinct subdivisions in the final report (Census 1901b:321; cf. Census 1875:197-198; Census 1881:52-55; Census 1891f:761; Census 1921b:106). The censuses of Cochin and Madras (of which Malabar was one district) generally paid less attention to subdivisions. An exception was the 1891 Census, which recorded 55 subdivisions in Cochin (Census 1891a:111) and 138 in Madras (Census 1891c:222). Some of the latter were subdivisions of Nayars residing outside Malabar, for only 128 were enumerated in Malabar itself (Census 1891e:380). The number of subdivisions is, as Mencher (1966:159) remarks, strikingly high; for instance, in Malabar, with a Nayar population of around 380,000 in 1891, the average membership of each subdivision would have been around only 3,000.2 But an average of this sort is somewhat irrelevant, for, significantly (as we shall see), subdivision populations were highly variable.

THE IDEAL MODEL AND STANDARD DESCRIPTIONS

In a Malayalam text entitled the Jatinirnayam, which, as Iyer (1912:15) explained, is a “work of some authority, which gives an account of Malayali castes,” eighteen Nayar subdivisions are listed (with their traditional occupations) in order of rank: (1) Kiriyam; (2) Illam; (3) Svarupam; (4) Padamangalam; (5) Tamil Padam; (6) Itasseri (herdsmen); (7) Maran (drummers); (8) Chempukotti (coppersmiths); (9) Otattu (tilemakers); (10) Pallicchan (palanquin-bearers); (11) Matavan or Puliyath (servants to Brahmans and Ambalavasis); (12) Kalamkotti or Anduran (potters); (13) Vattakkatan or Chakkala (oilmongers); (14) Asthikkuracchi or Chitikan (funeral priests); (15) Chetti (traders); (16) Chaliyan (weavers); (17) Veluttetan (washermen); and (18) Vilakkittalavan (barbers) (K. P. P. Menon 1933:192-195).3 The list bears a striking resemblance to the standard descriptions given by Aiya (1906:348-349) for Travancore, by Iyer (1912:14-18) for Cochin, and by Innes (1908:116-120) for South Malabar. Hardly different is a modern

[^2] There is a distinct possibility that the censuses themselves promoted the proliferation of subdivisions, as Nayars tried to claim higher status to the census enumerators; such attempts were widespread in India (Srinivas 1966:94-100). It is improbable, though, that Nayars would have invented so many subdivisions ex nihilo, and thus reasonable to assume that the subdivision proliferation was already prominent before the censuses began.

[^3] Subdivision names have been standardized, generally in accordance with the spelling in Hermann Gundert’s Malayalam-English dictionary.

287 account, Gough’s (1961:308-312). These various versions are presented in Figure 1. A few remarks about the information given in Figure 1 may be made here. Except for the Jatinirnayam, each account distinguishes the high-ranking subdivisions from those below, all of which are said to have had traditional occupations. Only Nayars from the higher subdivisions are said to have been soldiers. Kiriyam ranks highest throughout Kerala, although (except in the far north) there are few Kiriyam Nayars in Travancore. Aiya (1906) and Iyer (1912) both rank Illam second to Kiriyam, whereas Innes (1908) states that Illam and the two Charnavar subdivisions dispute for precedence in South Malabar; Gough (1961), although she places Illam fifth, also indicates that variations in ranking between districts did exist. The two Charnavar subdivisions, mainly found in Malabar, are jointly equivalent to Svarupam, which is absent in that region. Iyer, Innes, and Gough-but not Aiya for Travancore-also separate from the other Nayars the lowest three or four subdivisions, which are said to have formed a distinct group whose touch polluted those above them. Although there are undeniable discrepancies between these various descriptions, I hope the reader will agree that the overall structure of the Nayar subdivision system, as presented in Figure 1, is fairly clear and consistent. This structure has an interesting feature, for it embodies, so to speak, a caste system within a caste system. Except for high-ranking priests, the Nayar subdivisions mirror all the main caste categories: high-status aristocrats, military and landed; artisans and servants; and untouchables. But this point needs no further discussion because, as I shall now argue, this structure is ideal rather than real. The problem which obviously has to be considered now is how this structure, on which the Jatinirnayam and our authors seem to agree, is to be squared with the picture emerging from the census reports, which recorded so many more subdivisions than are listed in Figure 1. The problem deepens when one discovers that some of the subdivisions listed in Figure 1 are not in the census reports at all, while others are given with only tiny populations. For example, Chempukotti is missing from the 138 Nayar subdivisions discovered in the Madras Presidency (Census 1891e:69-70), from the 55 in Cochin (Census 1891b:i-vii) -unless Chempukotti is the same as Chembottil, a distinct possibility, but it is a subdivision with a population of less than 100-and from the 44 principal subdivisions in Travancore (Census 1901b:321). Chaliyan, to take a second example, is also absent from the Cochin and Travancore lists (loc. cit.); it was found in Madras-with a population of 169 (Census 1891c:223). One final example: Otattu had a membership of less than 100 in both Travancore in 1901 and Cochin in 1891, and

288 0 Q) CO 7O cn 0 0 u

1 M O CO 0 0 0 CO Cd cq cO Co 0 0 CO E0 ?cd Cd r.L COld .0. cl cqd C = -4 CO 0 63 cq cO . 00 +a V Cd .M 0 C? 0 ? 0 C-C C;. CO1; O C.0 0 . C’sC -?4 C,3 cl +a cqd CO - U Cd r4 +J c, 0.0 CO xCu0 CO 0. ' cd + 10 0 NO C -1- 4- J0 - C’s C W r. v o-d U- > CO o C 0 q COO M cqd c cqd m1 ~ 14 06 C0; 6 0 1 O ciC 0 c ?06 - ; 0 C.d 4 q 0C P 7 CO 00 3 CO 0U ~COc , r4 c3 cl .4 ‘sbD0 ’’ C Cd ~ 0 U.l*- 42 C~ C cis CO -CU COu mC 4 Ee?q - 0 c- d w l :: 4 -j u -J.1 4J 4- Mc cl qd ,3 cz C’ C’ c re~l~ E~r~C-C E0 CO >0- -1 - CO CO 0 a (6 CO CO ? ( ? 0 cqd ~e~ > r 00 ~ C~ 0~m-C C-C C z +j s . . d CO U c~d cd cqd 4-j P-4~ ed –4 . - > m c CO m0 CO .0 d 0MUG COO . 00 ..H 00 0.. > >0 000> 0004 O r -.4 1-4 P-4 " -f O rcqd Cd 0 Cdo d~ M ; ;~ r;~d r w cqd ‘4 cqd E cd cd(dm cqs -4 cq > d C’ , C’, (d C; C’s r cl c Cc 4 U 6 rz 06 C; C; - cli C? 06 ( ~ P-4 -1 P I P4 r- r-. C14 00 0 o 0 0 0 ? .4- 00 -0 O-1 CO ob 0 C() Q,-’ L4 00 00 0 -~ 00 -~ 0 w o 0 CO 0 0 00 0 CO .0

289 less than 500 in Madras in 1891 (loc. cit.). There is, then, a double discrepancy between the accounts summarized in Table 1 and the census reports. Not only do many of the subdivisions listed in the latter find no place in the standard descriptions; some of those listed in Table 1 are absent from the census reports.

TABLE 1 Populations of Nayar Subdivisions Travancore Cochin Madras (1901) (1891) (1891) Total Nayars 536,186 101,691 393,767 Higher subdivisions Kiriyam 25,164 23,017 115,125 Illam 325,125 23,279 42,429 Svarupam 103,123 – – Purattu Charna – 109,396 Akattu Charna – 9,096 32,446 Total higher subdivisions 453,412 55,392 299,396 Percentage of total 85% 55% 76% Lower subdivisions Itasseri 24,332 Chakkala 15,173 Pallicchan – 18,568 16,668 Vattakkatan – 8,452 30,980 Asthikkuracchi – 4,553 13,689 Total lower subdivisions 39,505 31,573 61,337 Percentage of total 7% 31% 16% Total all listed subdivisions 492,917 86,965 360,733 Percentage of total 92% 86% 92% Sources: Census 1901b:321; Census 1891a:111; Census 1891d:69-70. The explanation for this paradox, I suggest, is that all the standard descriptions stem ultimately from the Jatinirnayam, which embodied an ideal model of the subdivision system. It resembled, in a localized manner, the varna scheme and tended to be reproduced as an authentic account of the system. That this suggestion is plausible is supported by the role played by the varna scheme in descriptions of the Indian caste system; as Srinivas (1962:63) has noted, “Varna has been the model to which the observed facts have been fitted, and this is true not only of educated Indians, but also of sociologists to some extent.” Gough, like

290 the earlier authorities, is, I think, mistaken in presuming that the system as she describes it ever accorded with the historical reality.

AN ALTERNATIVE HYPOTHESIS

From the census figures presented in Table 1, it is clear that the majority of Nayars belonged to the five higher-ranking subdivisions (Kiriyam, Illam, Svarupam, Purattu Charna, and Akattu Charna) and, of those that did not, most were included in the Itasseri and Chakkala subdivisions (Travancore) or the Pallicchan, Vattakkatan, and Asthikkuracchi subdivisions (Cochin and Malabar). These few subdivisions accounted for around 90% of all Nayars in each of the three provinces. Most of the other subdivisions enumerated in the census reports had very small populations. Many had less than a hundred members, several had less than ten, and not a few had but one solitary representative! It follows fairly obviously from this that the Nayar subdivisions-at least as they are recorded in the censuses-cannot all be groupings of a single sociological type. An analysis which treats the Illam subdivision (with a population of over 325,000) in Travancore as identical to the Itattara subdivision (population 1) in the same state is clearly mistaken. Although Gough has not proposed any notion as absurd as this, she has stated in various places (e.g., 1959:28) that each of the aristocratic Nayar lineages “tended to set itself up as a separate sub-caste.” For this, she has been taken to task by Dumont (1964:98): Are the Nayar . . . a caste, and are their subdivisions really castes or sub-castes? . . . we are confronted with an enormous conglomerate of groups distinguished by their profession, social status. . . . Clearly these conglomerates are not castes, although they may appear as such in certain situations in relation to real outsider castes (Brahmans, Kshatriyas). . . . it is clear that we have to do with populations, not with castes. . . . Among the Nayar an exogamous lineage or cluster of lineages of a certain status cannot be called a sub-caste. . . . All these are actually status groups which may be absolutely (or practically) endogamous at the one end, and exogamous at the other. Dumont, then, argues that the Nayars constitute a “population,” not a caste, and that their subdivisions should be regarded as “status groups,” not subcastes. A thorough discussion of Dumont’s remarks would necessitate a lengthy digression into the formulation of definitions, which is better avoided. Briefly, however, for me, the merit of Dumont’s criticism of Gough is that it deters us from labeling as “subcastes” groups which lack certain characteristics normally regarded as typical of castes, and thus forces us to consider more carefully the nature of Nayar subdivisions. On the other hand, his criticism appears

291 to lead to an over-radical reinterpretation of the data in that he implies that neither the Nayars as a whole, nor any subdivision within that whole, forms a unit which can meaningfully be termed a caste. Given that the Nayars live in a caste society, they must evidently fit into the caste system at some level or another. The key to the problem, I submit, is to recognize that there are a number of analytically distinct categories within the system of Nayar subdivisions. The first of these categories is made up of the handful of populous subdivisions, such as Illam or Pallicchan. At the risk of oversimplification, the subdivisions in this category can be visualized as forming the stable “core” of the system. Around the core is a “fringe,” comprised of a large number of relatively very much smaller subdivisions. The fringe is more volatile than the core, and the smaller subdivisions are (or were) being created and eliminated by several social processes, of which three may be analytically distinguished. The first of these is the process by which lower-ranking groups-castes or sections of them-are absorbed into the Nayar community as part of a process of upward mobility. The second is the process in which exogamous lineages are joined together to form localized subdivisions, which may sometimes remain distinct subdivisions, or may sometimes coalesce with larger subdivisions, thereby eliminating the previous smaller subdivisions. In the third type of process, relatively very small sections of the community-exogamous lineages or sublineages-split off to create subdivisions in the context of a “status game” played at the highestranking levels of the community. The latter, intimately associated with the hypergamous marriage system, is the one which has attracted most attention, and I shall look at it in more detail later. An important question immediately looms: What precisely distinguishes the larger subdivisions from the others? No simple answer to this can be given. Indeed, a completely satisfactory response is, I think, impossible. A crucial aspect, however, is the stability to which I have referred. The populous subdivisions are represented throughout large regions of Kerala and have almost certainly existed for a very long time (Mencher 1966:160). Further, their statuses are relatively stable and are not a function of the fortunes of particular individuals, lineages, or other small groups. By contrast, the smaller subdivisions, especially those formed of a few lineages or even smaller kinship groupings, lack such stable statuses. It is pertinent here that one of the crucial characteristics of a caste is that its status is not subject to short-term alterations in response to the changing fortunes of individual members or small sections of the caste. Mencher (1966), arguing along similar lines, concludes that the larger subdivisions can meaningfully be regarded as subcastes. Her conclusion is evidently problematic, though

292 for my own part I am inclined to accept it. (Several other points about the criteria distinguishing the larger subdivisions from the rest have also been made by Mencher, but they are not strictly relevant here, and I refer the reader to her paper.) The distinction between groups with stable status and those with unstable status will be further developed in my conclusion.

SUBDIVISION CREATION AND ELIMINATION

The first process I isolated is that in which upwardly mobile, low-status groups are absorbed into the Nayar community. Among other features, the taking of the prestigious title “Nayar” may itself serve this end (Dumont 1964:98). The process is perhaps most apparent in the cases of the Veluttetan and Vilakkittalavan, and these two subdivisions also well illustrate the ambiguity attached to upward mobility. Nowadays, they are commonly referred to as “Veluttetathu Nayar” and “Vilakkittala Nayar”-at least in Central Travancore (where I worked) and in official publications. We have seen, too, that the Jatinirnayam included them as Nayars; on the other hand, in early census reports they were often enumerated as separate, non-Nayar castes. It is still the case, despite their names, that many Nayars belonging to higher-ranking subdivisions do not acknowledge the Veluttetathu and Vilakkittala Nayars as “real” Nayars, and they never intermarry with them. It is important to note that these ambiguities cannot be eliminated by redefining categories; they are intrinsic to the situation. Altering caste names with a view to raising status, and thereby increasing the uncertainty of status attribution, is a fundamental feature of the caste system. A precise parallel comes from the Patidars of Gujarat (Pocock 1972:54). In fact, it has to be conceded that hard evidence that these two subdivisions, like some other low-status ones such as Chakkala or Vattakkatan (oilmongers), have “risen to the status of a low-ranking Nayar [subdivision] over a long period of time” (Mencher 1966:160) is, as Mencher herself implies, scanty. Nevertheless, from both indirect evidence and examples from elsewhere in India, it does seem almost certain that such a process has occurred among the Nayars. For some Tamil groups, the evidence of their transition to Nayar subdivisions is quite plain (e.g., Iyer 1912:18). It may be remarked here that hypergamy, in that it can lead to a shortage of marriageable women for men on the lowest rungs in the caste, promotes the absorption of lower-status groups into the larger caste through marriage, and thus further expands the populous caste. But the relevance of this formal property is questionable here, for the evidence suggests that hypergamous sambandham relationships were mainly confined to the upper levels of the caste, and were rare among

293 the lower-ranking Nayars. Statistically, the shortage of women at the lower levels was probably insignificant. The second process is the creation of localized subdivisions from a number of exogamous lineages. Precisely how common such subdivisions are or were is impossible to say, for it does not emerge from census reports or gazetteers, which also fail to describe how the subdivision system functioned in any one locality. Unni’s (1958:63-67) is the best account of the process as it operated in one particular area. The situation Unni describes is complex and hard to summarize. Very briefly, however, in the two South Malabar villages which he studied, Nayar taravads were linked by hereditary duties of a ceremonial or ritual nature to Nambudiri families dominating the villages. By virtue of these duties, the taravads gained a “reflected” prestige dependent on the status of the Nambudiri family. A group of taravads with the same prestige, usually those linked to one family in one village, would tend to become endogamous, which means, in this context, that they would, for the most part, exchange sambandham partners only with each other. Over time, particularly if the group of taravads took a distinctive name, it would effectively become a subdivision. Nayar taravads serving Nayar chiefs, rather than Nambudiris, also formed similar groups, whose status depended on that of their respective chiefs. Clearly, subdivisions formed in this manner were highly localized, although they could expand by contracting alliances with other taravads of equal status. In such a case, the creation of a larger subdivision would simultaneously mean the extinction of two or more smaller ones. Unni (1958) himself remarks that his account does not do justice to the complexity of a situation in which there was a great deal of diversity. Nonetheless, despite these gaps in the evidence, there are, I think, sufficient data available for the analytical isolation of this second process to be both justified and useful.4

SUBDIVISIONS AMONG HIGH-STATUS NAYARS

The third process I isolated was that in which small subdivisions were formed as part of the status game played by high-status Nayars. Most of the rest of this paper will be concerned with its description and analysis. 4 The creation (ideally at least) of an endogamous caste subdivision from a number of exogamous lineages is not unique to the Nayars (cf. Dumont 1970:123). In parts of North Malabar (which is not treated in this paper, as it appears to be quite distinct from the region to the south in many respects), very large subdivisions, such as Purattu and Akattu Charna in the Chirakkal and Kottayam kingdoms, were said to be formed of groups of kulams, or exogamous clans. Innes (1908:118-120) provides the fullest details on the North Malabar system. His text, however, includes some mystifying details; for example, he implies that all kulams within a subdivision have equal status, but then proceeds to describe the different ritual duties, titles, etc. of these kulams, some of which even have different death pollution periods. Because I have been unable to resolve these problems, I remain quite uncertain about the North Malabar subdivision system.

294 Like most other writers of his time, Stuart (Census 1891c), the author of the 1891 Madras Census Report, viewed castes as rigidly defined units. Nonetheless, he made a significant observation when he remarked that some of the subdivision names “are only names of families [taravads], and not of separate sub-castes. Some of the names again are purely fanciful” (Census 1891c:222). Undoubtedly, the fanciful names to which Stuart referred were ones like Karattakkanonarkkacharna or Sekkarivarma Rajavamsam-to be found in the list of subdivision names given in the report (Census 1891e:69-70). That many of these subdivisions were only single taravads ivirtually certain, for, as I have mentioned, many subdivisions had minuscule populations. Some were probably only sections of a taravad, and there were, in addition, the single-member subdivisions-Nayars satisfying their vanity, I suppose, through the medium of the census. Almost certainly, this process of status raising occurred predominantly among the highest-ranking and most powerful or wealthy Nayars. Thus, we can begin by looking at the Kshatriyas and Samantans, the two castes to which the kings and chiefs claimed to belong; however, most unbiased observers (Dumont [1961:27] is an exception) have concluded that the Kshatriya and Samantan subdivisions should be treated merely as supereminent Nayar subdivisions. (Such a view was not, of course, shared by the authors of publications sponsored by the royal governments of Travancore and Cochin.) Like the Nayars, the Kshatriyas and Samantans are matrilineal, and they had the same marriage system. The Kshatriyas were split into two principal subdivisions: Tambans (or Tamburans) and Tirumulpads. Included among the former are the Cochin royal family and the Cranganore chiefly family (Iyer 1912:151). In Travancore, the categorization differed slightly from that of Central Kerala; there, Tamburans were divided into three categories of which only the highest-ranking-a grouping made up of ten chiefly families and known as Koil Tamburans-were Kshatriyas. The second grouping, known as Rajas, comprised nine chiefly families (including the Travancore royal family) who were Samantans, like the third grouping of ordinary Tamburans (Mateer 1883:115, 118; Pillai 1940:850-853). In Central Kerala, there were seven principal Samantan subdivisions: Eradi, Nedungadi, Vallodi, Unnitiri, Adiyodi, Tirumulpad, and Nambiyar. (Some of these names, of course, are also employed as subdivision names or titles by Kshatriyas and Nayars). Eradi is the subdivision to which the Zamorin (king) of Calicut belonged; the raja of Walluvanad was a member of Vallodi. The rajas of Chirakkal (and Travancore?) belonged to Unnitiri (Census 1891c:229-231).¹

295 The raja of Travancore, however used to perform an extraordinary ceremony known as hiranyagarbhan, “golden womb.”[^6] The essential feature of this ceremony was the casting of a hollow golden vessel through which the raja passed. On emerging from the vessel, the raja’s caste status rose from Samantan to Kshatriya. Unfortunately for the royal family, Kshatriya status so acquired was not hereditary, and thus the ceremony had to be performed for each new raja. But the misfortune of the royal house was a blessing for the Travancore Brahmans, for when the ceremony was over, the golden vessel was cut into pieces which the raja then distributed among the Brahmans. It has been suggested (Drury 1890:188), although so far as I know without any very sound evidence, that the Brahmans’ lust for gold sometimes tempted them to do away with ailing rajas, in the expectation that another hiranyagarbhan would soon have to be performed. (For a description of the ceremony, see Mateer 1871:169-175.)

Each Kshatriya and Samantan subdivision was minute.7 The Eradi subdivision, to take one example, had a population of only 356 in 1891 (Census 1891c:320), and this is probably entirely accounted for by the Zamorin’s lineage. However, the significance of the Kshatriya and Samantan subdivisions lay not in their numbers, but in the model they provided for other Nayars involved in the status game-an example of “imitation” (Pocock 1957:24) in the caste system. A Nayar taravad, especially if it were wealthy or powerful, could attempt to transmute itself into a new, separate subdivision. In such an endeavor, various stratagems could be employed: for instance, the severance of all connections with any demeaning occupation, the Sanskritization of various customs, or the taking of a new name. But perhaps most crucial of all was alteration of the taravad’s marital connections-by finding men of higher status to perform the tali-tying ceremony for the girls in the taravad, and by beginning to accept only men of higher status as sambandham partners for the women (cf. Mencher 1966:159). It is in the construction of a hypergamous marriage system that the Kshatriyas and Samantans supplied such an excellent prototype.

NAYAR MARRIAGE

Before proceeding further, a brief outline of the Nayars’ unique works on Travancore have rarely made reference to Cochin and Malabar, or vice versa, and this on occasion engenders confusion. 6 The precise connection between this ceremony and the golden egg (hiranyagarbhan), from which the primeval creator, Prajapati, emerged in Hindu mythology, is unclear. 7 The Kshatriya and Samantan castes both had very small populations. In Travancore in 1931, the Kshatriya population was 3,673 (0.07% of the total population), and the Samantan population was 97. In Cochin in 1931, the populations were, respectively, 2,128 (0.18%) and 571 (0.05%). In Malabar in 1931, the Kshatriya population was recorded as less than 0.1% of the total population; for 1921, the Samantan population was given as 4,663 (0.15%). Sources: Census 1931d:153-163; Census 1931a:lxxii-lxxiv; Census 1931b:306-310; Census 1921a: 110-123.

296 marriage system is desirable. Much of the following will be familiar to many readers, and some of them will be conversant too with the theoretical debate about the implications of the system. Since this is not an appropriate place to join that debate, however, I shall try to avoid taking issue with other writers on the subject. In South Malabar, Cochin, and the northern part of Travancore, each Nayar girl underwent the talikettukalyanam (tali-rite) before puberty. The rite incorporated a number of distinct rituals, but its only absolutely indispensable element was the actual tying of the tall itself (Menon 1911:192-193). (The tali is a leaf-shaped emblem, usually made of gold or silver, which is worn on a string around the neck. For most South Indian women, it indicates their status as married women and is equivalent to our wedding ring.) Theoretically, although probably not in practice, were a girl to reach menarche before the rite had been performed, she would have been ejected from her family and outcasted. Normally, each taravad held a tali-rite only once every ten years or so; on this occasion, all the immature girls of the family, babies included, would have their talis tied at the same time. In Kshatriya families, however, tali-rites were held more often, and talis were only tied on girls age ten to thirteen or so (Iyer 1912:152). When a Nayar girl attained sexual maturity, she began to receive lovers, thus inaugurating a relationship known as sambandham. Sambandham partners visited a woman at night in her room in the taravad house, but they rarely stayed at her taravad during the day or ate there. It was usual, although not universal, for a man to inaugurate the sambandham by giving a piece of cloth to the woman, as well as presenting her with other personal gifts at the three principal Malayali festivals. Apart from exogamic restrictions (which barred a woman from taking as a partner a member of her own matrilineal descent group or a close patrilateral relative), there was only one restriction on the choice of a woman’s sambandham partners: that they belonged to a caste or subdivision of equal or superior status to her own. However, the head of the taravad, the karanavan, had to approve the partners taken by his female dependents. The man who tied the girl’s tali could, if he were of a suitable category and so inclined, become one of her sambandham partners, but he was not preferred over other men. Both men and women could have several partners at once, and either party was free to break the relationship, for any reason or for none, whenever they wished. When a woman became pregnant, one or more of her partners was expected to claim paternity; if none did, then it was assumed that the woman had had sexual relations with a man of lower status. The punishment for this crime, for herself and her baby, was outcasteing followed by execution or sale into slavery.

297 Tying the tali around a girl’s neck was a ritual recognition of her right to receive men of appropriate status as sexual partners in the sambandham relationship, and to bear by such men children who, if one or more of these men claimed paternity, were legitimate-that is, in this context, possessed of full rights in their mother’s taravad and caste. It is, I think, clear, as has been generally recognized, that this is the fundamental meaning of the tali-rite. The sambandham relationship gives rise to only two important theoretical questions here: Why was there a status-based restriction on a woman’s choice of partner, and why was there the insistence that a man of appropriate status claim paternity of her children? The answer is surely that a caste is a bilateral grouping and a child’s place in the caste society cannot be determined by only one parent. Further, the Indian system of status attribution, under most circumstances, proscribes sexual relations between a woman and a man of status lower than herself, and generally denies to any children born of such a union membership of either parent’s caste. For these reasons, some recognition of paternity and an assurance that the genitor is of the right status is necessary-even if it is only the minimal one of a man asserting paternity.8 The analysis of Indian marriage has engendered a great deal of controversy. Cumulatively, various writers have successfully shown that Nayar marriage can be understood within its Indian context-that is, in terms of purity and pollution, the natUre of caste, legitimacy and affinity, etc. I have my disagreements with most of these writers individually. However, to cut a swathe through the controversy, I think it may be most sensible to regard Indian marriage, for a woman at least, as divisible into two basic types, which may be labeled simply “first marriage” and “second marriage” (cf. Iyer 1912:27-28; Dube 1953). (I use these labels because I do not mean quite the same as Dumont when he refers to “primary” and “secondary” marriages.) In my opinion, the first marriage is primarily a rite of passage for the girl, marking her progress, in a social sense, from girlhood to womanhood. It does not necessarily coincide with the physiological passage; among the Nayars, for example, a girl’s first menstruation was the occasion for a rite (tirandukuli) which was quite separate from the tali-rite (Iyer 1912:29-30). In the second marriage, a woman is linked as a sexual partner to one or more men, by whom she is expected to bear 8 There are many sources on Nayar martiage. Descriptions may be found in Iyer (1912:22-38), Thurston (1909:313-332), and Gough (1955, 1959). For the theoretical discussion, see-in addition to the works of Gough (1952; 1955; 1959; 1961; 1965)-Dumont (1961; 1970:119-120), Leach (1961:Ch. 4), Yalman (1963; 1967:365-374), Mencher (1965), and Carter (1974). I have discussed Nayar marriage, as well as modern changes in the Nayar kinship and marriage system, at greater length in a book which I hope to have published in the not too distant future.

298 children. The second marriage centers on the continuation of her or her husband’s line through the birth of legitimate children, while the first marriage is principally concerned with the status of the woman herself. In most Indian communities, the two types of marriage are not actually separated, but are both embodied in one “ordinary” marriage, which is thus a rite of passage for the woman as well as a rite concerned with the progeniture of children. Hence, we can regard the tali-rite as a first marriage, and the sambandham union as a second marriage. Although I have completely ignored the important question of affinity, this short summary of the ethnography and analysis of Nayar marriage will, I hope, suffice for our present purposes.

MARRIAGE AND THE STATUS OF SUBDIVISIONS

The elaboration of the tall-rite was closely correlated with the wealth and prestige of the taravad for whose girls the rite was being celebrated. But it was the status of the actual tali-tier-and the manner in which he performed his role in certain instances-which were the most significant aspects of the rite in relation to the status of the girl’s group. Conversely, for whom a man tied a tall could affect the status of his group. The choice of sambandham partners was also significant in the status game, very much more so for a woman than a man. As I shall show later, the statuses of tali-tiers and sambandham partners can be seen as diacritical markers of subdivision status. The entire process can also be understood as one of “inclusion” and “exclusion” (Pocock 1957:28); for instance, a girl’s group attempts to include itself with the higher-ranking tali-tier’s or sambandham partner’s group, while excluding itself from those refused as tali-tiers or sambandham partners, and therefore accorded lower status. It is important to note that among the higher-ranking Nayars (and Kshatriyas and Samantans) in contradistinction to the “commoner” Nayars, no two subdivisions admitted to equal status. Thus the relations set up by the tall-rite and the sambandham union were always hypergamous.9 As Gough (1959:28) phrases it, each high-status family acknowledged ritual superiors and inferiors but acknowledged no peers. Complete information on the statuses of tali-tiers and sambandham partners is not available for all the high-ranking subdivisions, on which the following account will concentrate; that which we do have is not

299 always consistent. Nonetheless, the general principles underlying the system can be ascertained.

The families of highest status were those in which only Nambudiri Brahmans (the highest-ranking caste in Kerala) acted as tali-tiers and only Nambudiris were accepted as sambandham partners by the women. In Central Kerala, the sole family definitely practicing Nambudiri exclusivism was the Cochin royal house, belonging to the Kshatriya caste. A Nambudiri tied the tali for only one princess at a time, and he was paid for his services (Iyer 1912:152; Census 1901a:145-146). Neither Iyer nor the author of the census report makes it absolutely clear whether other Kshatriya families, in particular the Cranganore chiefly family, also employed Nambudiris to tie their girls’ talis. Gough (1955:49, 53; 1959:29; 1961:377) implies that Nambudiris tied girls’ talis in many aristocratic families but, apart from the Cochin royal family, fails to specify which. On the other hand, Mencher and Goldberg (1967:93-94), who have carried out fieldwork among the Nambudiris, state flatly that in Central Kerala, except for the Cochin princesses, Nambudiris never tied the tali for any girl of another caste. For a number of reasons, I am inclined to believe that Mencher and Goldberg are correct-principally because their assertion makes the most logical sense in the context of what happened in other aristocratic families. Of note is the fact that other Nambudiris considered the tali-tier for even the Cochin princesses to have been polluted and degraded (Mencher and Goldberg 1967). There was thus a kind of status exchange operating here: the Nambudiri tier losing status as the Cochin royal family gained it. I have been unable to find any concrete evidence that such a status exchange was present in all tall-rites, so that a higher-ranking tier always lost status when a lower-ranking group employed him. If an exchange of this sort was general, however-and it seems logical that it should have been-it has notable implications for the analysis of Nayar marriage within its wider Indian context, particularly in relation to the question of affinity (see, especially, Dumont 1961). Germane to this would be the contrast with the sambandham relationship in which, it appears, a status exchange either did not operate or was only of minimal significance. In other words, Nambudiri men, for example, lost nothing or very little by way of status if they became sambandham partners of Nayar women. In Kshatriya families of Central Kerala, apart from the Cochin royal family, the tali-tier was an Aryapattar (a category of Tamil Brahmans), and the women took as sambandham partners either Nambudiris, Tamil or South Kanara Brahmans, or Kshatriyas (Census 1901a:145-146; Iyer 1912:152). (“Foreign” Brahmans rank below Nambudiris in Kerala.) However, Iyer also notes, rather confusingly, that in

300 “Kshatriya houses of noble ancestry,” only Nambudiris acted as sambandham partners. A question arises because the identity of Kshatriya families not of “noble ancestry” is obscure; the caste was small, and many families-I would have suspected almost all-must have been fairly closely related, through junior lines, to the royal families. (Stuart’s statement [Census 1891c:231] that in Kshatriya families, the tali-tier was either a Nambudiri or a Kshatriya seems to me, in the light of other evidence, implausible; his “Nambudiri” probably ought to refer to other categories of Brahmans.)

According to Pillai (1940:851), among the ten families of Koil Tamburan Kshatriyas in Travancore, the girls’ talis were tied by Aryapattars or, more frequently, by Nambudiris. However, these Nambudiris were almost certainly relatively low-ranking, as none of the highest-status Nambudiri families resided in Travancore (Mencher and Goldberg 1967:90). Koil Tamburan women took only Nambudiris as sambandham partners.

I shall now turn to the Samantans, continuing for the present in Travancore. The talis of girls belonging to the nine Raja families were tied by men from the Koil Tamburan families; in other Tamburan families, the tali-tiers were Aryapattars (Pillai 1940:852-853). Pillai provides no data on their sambandham partners, but according to Mateer (1871:118, 171-172), Travancore princesses took only Koil Tamburan men as partners.

For the Central Kerala Samantans, too, the data are incomplete, but the overall picture is reasonably clear. In the Zamorin’s family, that is, the Eradi subdivision, talis were tied by Kshatriyas from the Cranganore chief’s family, which the Zamorin recognized as more ancient and thus of higher rank. The talis of girls belonging to the Walluvanad raja’s family, equivalent to the Vallodi subdivision, were tied by men of the higher-ranking Beypore chief’s family (Iyer 1912:147; Gough 1955:49). For other Samantan girls, the talis were usually tied by Kshatriyas, although in poorer families they may have been tied by Samantans of the Tirumulpad subdivision. In the latter case, the Tirumulpad sometimes tied the talis of a number of girls at once (Census 1891c:231; Iyer 1912:147), which is less prestigious than having one tier for each girl. In the Zamorin’s and other chiefly families, all (idem.) or nearly all (Gough 1959:29) the women’s sambandham partners were Nambudiris, the possible exception being Kshatriya men from ancient royal families.

Finally to the Nayars themselves: here, there is a great deal of diversity. According to Iyer (1912:24-25), the tali-tier among Cochin aristocratic families was usually a Tirumulpad, probably a Samantan rather than a Kshatriya. Sometimes there was one Tirumulpad for each

301 girl; sometimes one tied the talis for all the girls at the ceremony. In northern Travancore, girls in aristocratic Nayar families had their talis tied either by Tirumulpads or Aryapattars (Census 1901b:329), although Aiya (1906:353) believes that men from certain Ambalavasi (templeservant) castes, which were ranked between Samantans and Nayars, could take this role. Again, there was sometimes one tier for each girl and sometimes not. Data on the tali-rite among aristocratic Nayars are sparse, and none of the older accounts explains exactly who these aristocratic Nayars were. According to Iyer (1912:23), Cochin aristocratic families had to ask the raja’s permission to hold a tall-rite, and make a donation on the occasion. However, it would appear most likely that the aristocratic Nayars were those belonging to small, high-ranking subdivisions who were also usually district chiefs or village headmen (Gough 1959:28). Their hypergamous marriage connections, which distinguish them from commoner Nayars, are, of course, both a means to, and an end of, their higher status.

Among the commoner Nayars of Central Kerala, the tali-tier was normally an enangan, that is, a member of a lineage linked to that of the girl by certain ritual duties, of which tali-tying was the most important. The enangan lineages of commoner Nayars were of equal status and belonged to the same subdivision; in the tali-rite, there was one tier for each girl. In Travancore, enangans could also tie the tali, but in most of the country except the extreme north, the tier more often belonged to one of the macchampikkar families. There were three or four such families in each village, and they were ancient Nayar families appointed as tali-tiers by royal writ. Like tali-tiers from equal-status enangan lineages, those from macchampikkar families only tied the tali on one girl at one time (Aiya 1906:353). Unfortunately, I have been unable to find any more data on the macchamppikkar families, making the obviously relevant comparison with the enangan institution not feasible. In very poor families, girls’ talis were tied at a small ceremony held at the side of a proper tall-rite, or in front of a temple, a Brahman’s house, or a chief’s residence. The tier was not infrequently the girl’s mother or aunt (Government of Madras 1891:19; Census 1901a:159; Census 1901b:329; Innes 1908:175; Iyer 1912:27). Although the situation is not absolutely clear, it is, I think, best to regard these poor families as having been forced to deviate from the norm by economic privation, rather than as crucial exceptions to generalizations made with reference to the tali-rites of the majority of Nayars. Throughout South Malabar, Cochin, and the northern part of Travancore, Nayar women-both aristocratic and commoner-accepted men of any caste or Nayar subdivision of status equal or superior to their own as sambandham partners. According to one source (Census

302 1901a:162), Nambudiris only took Nayar women as partners if the latter were from aristocratic families. If this source is correct, then it would imply (compare my earlier remarks) that there was a danger to a Nambudiri’s status in consorting with lower-status women; however, the statement in the 1901 census report is widely contradicted elsewhere. In spite of the permissibility of hypergamy, the “great majority” of commoner women must, as Gough (1959:26) points out, have taken commoner men as sambandham partners for demographic reasons. In the remainder of Travancore, women very rarely took partners from subdivisions other than their own; indeed, Aiya (1906:357) implies that intra-subdivision unions were much more strongly preferred than they were further north, although the exact difference between Travancore and Central Kerala in this respect remains obscure. Despite its unavoidable lacunae, my summary has, I hope, given some indication of the way in which the statuses of those who tied the talis for a group’s girls, and of those who were sambandham partners to members of the group, acted as diacritical markers of that group’s status. One family, and thus one subdivision, demonstrated its superiority over another by, say, having Kshatriya rather than Samantan tali-tiers, or by accepting only Nambudiris, rather than men of any Brahman community, as sambandham partners for its womenfolk. And, conversely, although this appears rather less distinctly in the data, one subdivision may have demonstrated its superiority over another by refusing to supply talli-tiers to an allegedly lower-status group. Now, clearly, such a demonstration of superiority is necessarily implied by refusal. But what is unclear is strategy-whether or not subdivisions would actually try to rise by refusing to act as talli-tiers to alleged inferiors. Evidence from elsewhere in India suggests that they might have done so. It should also be recalled that according to certain classical texts, upward mobility is sometimes possible through repeated isogamous marriages (Tambiah 1973:202). Forming isogamous, rather than hypergamous, links at the tali-rite and through sambandham unions might have been a route to higher status (as, perhaps, in the subdivisions described by Unni [1958]). Again, evidence on whether such a strategy was in fact adopted is not, to my knowledge, present. I must briefly refer to the question of how many tiers were present at the tall-rite. I have already indicated the prestige connotations of this, but it has a wider theoretical implication as well. If there was only one tier for several girls, this would imply equal status for all the girls and their immediate matrilineal descendants. If each girl had her own tier, however, this was not necessarily so, and thus the possibility of different descent lines within the taravad having various statuses was opened up. We have here the “parcelization” of status-the creation of minute

303 units, each with different status, which in turn increases the permutations of ranking within the caste, and the opportunities for mobility within it.

In concluding this section, I would like to re-emphasize the dynamism of this process whereby subdivisions, equivalent to taravads or sections of them, were created and eliminated in the competition for status. If a commoner taravad, probably after it had become wealthy or otherwise powerful, wished to raise its status, it could try to form hypergamous relations in place of its previous isogamous ones. It might have begun by permitting only higher-status sambandham partners for its womenfolk. As we have seen, the choice of a man’s sambandham partners did not really affect his taravad’s status. Severing enangan ties was probably more difficult, although severance would be an essential part of the status-raising process, for, as we have seen, the tall-tier was always of higher rank in aristocratic families. An aristocratic taravad whose relations were already hypergamous could attempt to rise still further by finding men of yet higher status to act as tall-tiers and sambandham partners. An illustration of this comes from the Zamorin’s family, which, in the eighteenth century, began to accept only Nambudiri partners for its women, whereas it had previously accepted certain Kshatriya men as well (Gough 1955:47). But as taravads could rise, so too could they fall if those who had acted as tall-tiers and sambandham partners refused to continue, thus forcing the taravad to accept men of status lower than their predecessors to perform these roles. Mobility was no one-way process.

MODERN CHANGE IN THE SUBDIVISION SYSTEM

Much of what has been discussed here has, of course, now disappeared. A principal feature of the subdivision system has vanished as a result of the decline in significance, and then the disappearance, of the tali-rite, together with the transformation of sambandham into monogamous marriage. Concrete data, as opposed to general impressions, are sparse on the subject of Nayar subdivisions in modern times, and few generalizations applicable to all of Kerala can be hazarded. However, I do have some data from my fieldwork in Central Travancore.10 This region, though, is almost certainly atypical because the Nayar Service Society (N.S.S.), the caste association of the Nayars, was founded in this area in 1914, and still has most of its support there. Arguing that only as a united community could Nayars form a powerful political force in the state, the N.S.S. has always campaigned 10 Fieldwork in Central Travancore was carried out between August 1971 and October 1972. It was financed by the United Kingdom Social Science Research Council and the Smuts Memorial Fund, University of Cambridge, to whom I express my gratitude.

304 for the abolition of subdivisions among the Nayars. It has not only encouraged intermarriage between subdivisions, but has asked Nayars to refuse to reveal their subdivision membership to census enumerators. Partly because of this campaign, the 1931 Travancore Census Report (Census 1931c:364, 377) stated that the N.S.S. campaign to fuse the subdivisions had been a success, and that there were now only two subdivisions remaining. However, the author of the 1941 census report was more cautious, noting that the “correctness” of the remarks made ten years earlier “could not be vouched for” (Census 1941:131). His caution was wise, for although there were commentators such as the author of the Travancore State Manual who claimed that “even intermarriages are now quite common and meet with the full approval of society” (Pillai 1940:858), the validity of this claim is highly dubious. Even today, in Central Travancore, intermarriages between subdivisions are very rare. In the area where I worked, many Nayars see the persistence of subdivisions within their community as a social and political problem, and they tend to deny their existence when outsiders ask about them. Only my frankest informants, once they realized that I had deduced that subdivisions still existed, would discuss them with me. In Central Travancore today, four Nayar subdivisions are found: Illam, Svarupam, Itasseri, and Chakkala. There are also the Veluttetathu and Vilakkittala Nayars, but these two are not regarded by other Nayars as “real” Nayars at all, and they are ranked by most people-Nayars as well as non-Nayars-considerably below “real” Nayars, being accorded a status approximately equal to that of the artisan castes. Turning to the other subdivisions, the reader will recall (see Table 1) that these four, together with Kiriyam, were by far the largest subdivisions in Travancore in 1901, and that together they accounted for over 90% of all Nayars in the state. Kiriyam, as I have mentioned, was mainly confined to the area of Travancore bordering on Cochin. There are two obvious hypotheses as to why Illam, Svarupam, Itasseri, and Chakkala are, apparently, the only subdivisions currently existing in Central Travancore. The first, and simpler, is that these four are the only ones which have ever been represented in the region. The second hypothesis is that other, smaller subdivisions have been absorbed into the four larger ones. Although I know of no data which would permit us to clinch either of these hypotheses, the first seems less likely solely on the grounds that I can find no apparent reasons why there should not have been many small subdivisions in this part of Travancore. Admittedly, this is an extremely weak argument, but without any information bearing on the distribution of subdivisions within Travancore, it cannot be reinforced.

305 The second hypothesis is, in my estimation, more likely. Gough (1961:308) and Mencher (1966:161) both refer to the current tendency toward amalgamation of Nayar subdivisions, although they do not produce much concrete evidence to support their statements. In Central Travancore, there is the additional factor of the N.S.S.’s campaigns, which may have led to the absorption of smaller subdivisions into the larger ones, even though they have not succeeded in eliminating the latter as well. But I must repeat: much of the above argument is speculative, and further, the distinct possibility exists that the impression gained by Gough, Mencher, and myself-an impression shared by many Nayars-that subdivisions have tended to amalgamate is false.

CONCLUSION

In writing this paper, I have had a number of objectives in mind. The first was to provide a more accurate empirical description of the Nayar subdivision system than presently exists in the literature. Using early census reports in particular, I hope to have shown that the system does not consist of a relatively small number of clearly defined subdivisions, but rather of a relatively large number of ill-defined groupings (cf. Beck [1972:258], who argues similarly for subcastes throughout India). These fall into four analytically distinguishable categories. At one extreme, there is a small number of populous subdivisions, incorporating the vast majority of Nayars, which can be regarded as subcastes, whereas at the other extreme, there is a multitude of tiny subdivisions-mainly high-status-which consist of exogamous groups or sections of such groups. To describe these latter as subcastes would be an abuse of the term for, perhaps most importantly, they lack stable statuses. I should draw the reader’s attention to a factor which would, I suspect, be of considerable importance in a comprehensive account, but which I have been unable to treat in any detail; this factor is the relation between subdivisions and territory. We have already seen that the kinds of subdivisions described by Unni (1958) were confined to particular localities; further, both Mencher (1966:160) and Miller (1954:416-417) refer to the geographically limited distribution of subdivisions. In particular, there was a contrast between the higher subdivisions, which were represented over large areas (although members of different subdivisions were rarely all found in one village), and the lower subdivisions, which tended not to be so widespread. In more modern times, there has been a tendency toward greater mixing of subdivisions within a particular locality. Comparative evidence from elsewhere in India-for example, the Rajputs of Malwa (Mayer

306 1960:152-161) or the castes in Konku, Tamilnadu (Beck 1972:2, 66-78 & passim), as well as Miller’s paper (1954)-would suggest that the relation between subdivisions and territory is of considerable significance. But, as I have mentioned, there is little useful information on the geographical distribution of subdivisions, making it impossible to discuss the issue in any detail. The second objective has been an attempt to demonstrate how Nayar marriage, especially in its hypergamous dimension, is connected with the subdivision system. We have seen that hypergamy is a crucial component of the status game at the top. The existence of this connection has long been recognized, by Gough (1959:28-29), for example, but I think I have been able to spell out the details more fully than previous writers, for they have worked with an inaccurate picture of the subdivision system. There has also been an understandable tendency for discussions of Nayar marriage to focus on the conundrums posed for theories of kinship, marriage, and the family. Without in any way decrying the value of these discussions, I would suggest that they have meant that the status implications involved in the choice of talitier and sambandham partner have been somewhat neglected. Further, the attempt to fit the Nayars into Indian society on a wider level has meant that the nature of the Nayar community itself has not received the attention it deserves. The latter remark brings me to the third objective, which has been to ask again the question: What is the Nayar caste? Pocock, on the basis of his work among the Patidars of Gujarat, was forced to pose an identical question (1972:52). Some of the parallels between the Nayars and the Patidars have already been mentioned, but further ones no doubt exist. For instance, Pocock stresses the heterogeneity within the Patidar community-there is no single set of beliefs or customs uniting all Patidars, nor is there economic homogeneity within the caste-and his remarks apply equally to the Nayars. In other words, the Nayars, like the Patidars and indeed many other castes, cannot be seen as some kind of solidary block incorporated into a stable system with a number of similarly solidary castes. There is nothing original in this conclusion; although many earlier writers did tend to see castes as blocks of this kind, more modern research has demonstrated clearly enough that such a picture is wrong. However, there has been no simple or satisfactory replacement for the old view. Dumont and Pocock represent two approaches to the problem, and the issue between them revolves around the question of whether or not there is only one kind of status principle operating in the caste system. Briefly, Dumont contends that this is the case, whereas Pocock proposes a distinction between the kind of status ordering

307 relations between castes and the kind of status appertaining within the caste. For Dumont, the old view of castes as solidary blocks was permeated with the “substantialist fallacy.” Castes are not homogeneous status groups hierarchized from outside, an ordered series of pigeon-holes. This is a substantialist fallacy. The hierarchical principle does not stop at the outward boundary of each particular caste-group, it permeates it, and the caste boundary is only one more marked cleavage than others (Dumont 1964:83). Later, he stressed the segmentary structure of the caste system. Rather than try to identify or define a “caste” or “subcaste,” one should focus on the total system, defined according to a uniform principle of hierarchy (1970:41-42, 61-64). But this directive poses its own problems. It is, of course, a defining characteristic of a segmentary system that each segment is structurally homologous with each other segment and with the whole, and that the principle of segmentation is uniform at all levels of the structure. This does not mean that each segment has identical functions. But, in fact, it is not clear that structural homology is always present in castes and their subdivisions. Dumont discusses the case of yet another huge and heterogeneous caste, the Sarjupari Brahmans of Uttar Pradesh, among whom localized patrilineal groups bear titles which are indicative of a certain “status.” But, according to Dumont (1970:123), “The ‘status’ in question has no effect outside the relation of intermarriage; one would therefore rather speak of a certain consideration or prestige. . . .” Within the Sarjupari Brahman, and also the Rajput and other populous castes, Dumont (1970:123) emphasizes that “status is ascribed not only to endogamous but also exogamous groups.” Whether this fact would imply merely that segments in this system are functionally differentiated, as one would expect, or whether it implies that they are not structurally homologous, and that some modification of the segmentary model of the caste system is therefore required, is unclear. At this point, Dumont seems indecisive; although uncomfortable and tempted toward historical speculation, he avoids any direct confrontation with the issue. Pocock, on the other hand, is explicit. For him, the caste boundary does represent a real divide, and status within the caste is not identical to status between castes: It seems analytically necessary to speak of standing of a person, a family, or a group of families within the caste as opposed to the status which they may have in their relationship with other castes (Pocock 1972:65). If one were to use the term “status” to describe the relation between, say, two Patidars, it would imply that this relation was homologous to

308 that between castes, which, asserts Pocock, is absurd. But to my mind anyway, Pocock fails to show why it is so absurd. Nonetheless, Pocock is making an analytically significant and probably ethnographically valid point. I, however, would like to redefine his proposition somewhat; I would suggest that the essential distinction which needs to be made is not between status within the caste and status between castes, but that between relatively stable and relatively unstable statuses. For short, let us refer simply to stable and unstable status. In addition, I would insist that no hard and fast dividing line between stable and unstable status can be drawn. Rather, they represent poles of a continuum, and, further, the conversion of unstable status to stable status is a fundamental element in the system. What I am saying, in other words, is that the distinction between status and standing is better expressed as one between stable and unstable status, and that it should not be conceived of as a simple binary opposition. No doubt my argument needs further explication. It is plain, I think, that the status attached to a caste is normally more or less constant over longish periods of time, and that it is not subject to short-term fluctuations in the fortunes of members of the caste. The status attached to a kinship group, a family, or, most extremely, an individual, is, on the other hand, much less permanent, and, unlike caste status (which is-ideologically anyway-defined by supposedly immutable ritual criteria), it is subject to relatively short-term alterations in response to changing economic and political fortunes. This dichotomous distinction gives rise to a series of points. First, I want to make it clear that I am defining neither kind of status in terms of economic or political variables. Normally, either kind will be ideally conceived of as being independent of economics and politics. Of course, they are not-but what is relevant here is not that fact in itself, but the “lag”: how long it takes for status to alter in accordance with changes occurring outside its own domain. At the same time, however, and this is my second point, the stable/unstable distinction will often also refer to different sorts of status criteria; most especially, stable status will tend to depend on ascribed ritual criteria (purity and pollution), while unstable status will tend to be accorded on the basis of achievements-successful marriages, occupation of eminent offices, military valor, etc. The distinction thus implies a nonuniform set of principles of status attribution. The third point, though, is that despite this dichotomy, there is a continual striving toward the conversion of unstable, achieved status into stable, ascribed status, and this is one important aspect both of, say, an individual taravad’s attempts to contract higher-status marital connections, and of endeavors by castes or sections of them to follow the classic path of Sanskritization.

309 I should further point out that my definitional distinction, in itself, incorporates neither the intra-caste/inter-caste distinction, nor the endogamous/exogamous group distinction. This point is particularly relevant in the Nayar context for, as I tried to show, the handful of large subdivisions had stable statuses; in other words, they had statuses comparable in their stability to that of the caste as a whole. In this respect, these large subdivisions differed from the many small, highranking Nayar (and Kshatriya and Samantan) subdivisions, whose statuses were unstable and sensitive to changing economic and political fortunes in the short-term. Within the caste, therefore, subdivisions differed from each other in the kinds of status attaching to them. This conclusion implies a rejection of Dumont’s uniform segmentary model and, at the same time, a rejection of Pocock’s analysis, although the latter did provide the catalyst for the argument I have developed. To return to the question of what the Nayar caste is (or was): it is a large, named social group (or, perhaps preferably, category) with a stable status, vis-A-vis other castes in Kerala. It is not, however, a solidary group, and, the efforts of the N.S.S. notwithstanding, it is never likely to become one. It is internally divided into a number of different types of group. First, the few large groupings, to which the overwhelming majority of Nayars belong, also have stable status. (It would, of course, be foolish to repeat a mistake and conclude that the large subdivisions represent some sort of solidary blocks. Obviously they do not.) Second, there are (or rather were) many very small groups with unstable status; these were lineage-based groupings engaged in a status game at the highest levels of the caste. Third, there were subdivisions consisting of either upwardly mobile groups or conglomerates of lineage-based groupings, which had statuses less stable than the large subcastes, but more stable than the small, high-ranking subdivisions. Over time, this third type of subdivision would normally (though not inevitably) gain increasingly stable status. This is hardly an exhaustive definition of the Nayar caste, but it may perhaps be accounted a step forward, and a contribution toward the understanding of those vast, amorphous castes, distributed throughout India, which have arguably received less attention than their numbers warrant.

BIBLIOGRAPHY

AIYA, V. NAGAM 1906 The Travancore State Manual, vol. 2. Trivandrum: Government Press. BECK, BRENDA E. F. 1972 Peasant Society in Konku: A Study of Right and Left Subcastes in South India. Vancouver: University of British Columbia Press.

310 CARTER, A. T. 1974 A Comparative Analysis of Systems of Kinship and Marriage in South Asia. Proceedings of the Royal Anthropological Institute 1973:29-54. CENSUS 1875 Census of Travancore. Trivandrum (1876). 1881 Census of Travancore. Trivandrum (1884). 1891a Census of Cochin, pt. I. Cochin (1893). 1891b Census of Cochin, pt. II. Cochin (1893). 1891c Census of India, Madras, vol. XIII. Madras (1893). 1891d Census of India, Madras, vol. XIV. Madras (1893). 1891e Census of India, Madras, vol. XV. Madras (1893). 1891f Census of Travancore, vol. I. Trivandrum (1894). 1901a Census of India, vol. XX, Cochin, pt. I. Ernakulam (1903). 1901 b Census of India, vol. XXVI, Travancore, pt. I. Trivandrum (1903). 1921a Census of India, vol. XIII, Madras, pt. II. Madras (1922). 1921b Census of India, vol. XXV, Travancore, pt. I. Trivandrum (1922). 1931a Census of India, vol XXI, Cochin, pt. II. Ernakulam (1933). 1931 b Census of India, vol. XIV, Madras, pt. II. Madras (1932). 1931c Census of India, vol. XXVIII, Travancore, pt. I. Trivandrum (1932). 1931d Census of India, vol. XXVIII, Travancore, pt. II. Trivandrum (1932). 1941 Census of India, vol. XXV, Travancore, pt. I. Trivandrum (1942). DRURY, HEBER 1890 Reminiscences of Life and Sport in Southern India. London: W. H. Allen. DUBE, S. C. 1953 Token Pre-puberty Marriage in Middle India. Man 53:18-19. DUMONT, LouIS 1961 Les Mariages Nayar Comme Faits Indiens. L’Homme 1:11-36. 1964 Marriage in India: the Present State of The Question: Postscript to Part I; Part II. Contributions to Indian Sociology 7:77-98. 1970 Homo Hierarchicus. London: Weidenfeld & Nicolson. GOUGH, E. KATHLEEN 1952 Changing Kinship Usages in the Setting of Political and Economic Change Among the Nayars of Malabar. Journal of the Royal Anthropological Institute 82:71-87. 1955 Female Initiation Rites on the Malabar Coast. Journal of the Royal Anthropological Institute 85:45-80. 1959 The Nayars and the Definition of Marriage. Journal of the Royal Anthropological Institute 89:23-34. 1961 Nayar: Central Kerala. Pp. 298-384 in Matrilineal Kinship (ed. by D. M. Schneider and E. K. Gough). Berkeley: University of California Press. 1965 A Note on Nayar Marriage. Man 65:8-11. GOVERNMENT OF KERALA 1971 Report of the Backward Classes Reservation Commission, Kerala, 1970, vol. 2. Trivandrum: Government Press.

311 GOVERNMENT OF MADRAS 1891 Report of the Malabar Marriage Commission. Madras: Government Press. INNES, C. A. 1908 Madras District Gazetteers: Malabar and Anjengo (ed. by F. B. Evans), vol. 1. Madras: Government Press. IYER, L. K. ANANTHA KRISHNA 1912 The Cochin Tribes and Castes, vol. 2. Madras: Higginbotham and Co. LEACH, E. R. 1961 Rethinking Anthropology. London: Athlone Press. LOGAN, WILLIAM 1887 Malabar, vol. 1. Madras: Government Press. MATEER, SAMUEL 1871 The Land of Charity: A Descriptive Account of Travancore and Its People. London: John Snow. 1883 Native Life in Travancore. London: W. H. Allen. MAYER, ADRIAN C. 1960 Caste and Kinship in Central India. London: Routledge and Kegan Paul, Ltd. MENCHER, JOAN P. 1965 The Nayars of South Malabar. Pp. 163-191 in Comparative Family Systems (ed. by M. F. Nimkoff). Boston: Houghton Mifflin. 1966 Kerala and Madras: A Comparative Study of Ecology and Social Structure. Ethnology 5:135-171. MENCHER, JOAN P., AND HELEN GOLDBERG 1967 Kinship and Marriage Regulations Amongst the Namboodiri Brahmans of Kerala. Man (n.s.) 2:87-106. MENON, C. ACHYUTA 1911 The Cochin State Manual. Ernakulam: Government Press. MENON, K. P. PADMANABHA 1933 History of Kerala, vol. 3. Ernakulam: Government Press. MILLER, ERIC J. 1954 Caste and Territory in Malabar. American Anthropologist 56:410- 420. PILLAI, T. K. VELU 1940 The Travancore State Manual, vol. 1. Trivandrum: Government Press. POCOCK, D. F. 1957 Inclusion and Exclusion: A Process in the Caste System of Gujerat. Southwestern Journal of Anthropology 13:19-31. 1972 Kanbi and Patidar: A Study of the Patidar Community of Gujarat. Oxford: Clarendon Press. SRINIVAS, M. N. 1962 Caste in Modern India. Bombay: Asia. 1966 Social Change in Modern India. Berkeley: University of California Press.

312 TAMBIAH, S. J. 1973 From Varna to Caste Through Mixed Unions. Pp. 191-229 in The Character of Kinship (ed. by Jack Goody). Cambridge: Cambridge University Press. THURSTON, EDGAR 1909 Castes and Tribes of Southern India, vol. 5. Madras: Government Press. UNNI, K. RAMAN 1958 Polyandry in Malabar. Sociological Bulletin 7:62-79, 123-33. YALMAN, NUR 1963 On the Purity of Women in the Castes of Ceylon and Malabar. Journal of the Royal Anthropological Institute 93:25-58. 1967 Under the Bo Tree. Berkeley: University of California Press. DEPARTMENT OF SOCIAL ANTHROPOLOGY UNIVERSITY OF MANCHESTER MANCHESTER M13 9PL, ENGLAND