HOMO HIERARCHIES
serves to express all sorts of things, so here the universal form of the relation is given by the properly religious relation, as the etymological connotation of jajmani incidentally reminded us. One employs a Brahman, a gcneologist, a barber, one employs similarly a carpenter, one employs an untouchable unfree labourer: each case is always, so to speak, on the same model. In other words, the 'religious' is here the universal mode of expression, and this is perfectly coherent if one knows that the overall orientation is religious, that the language of religion is the language of hierarchy, and that the hierarchy is necessarily, as we have seen, a matter of pure and impure.
In the last analysis, the division of labour shows not a more or less gratuitous juxtaposition of religious and non-religious or 'economic* tasks, but both the religious basis and the religious expression of interdependence. Better, it deduces interdependence from religion.
Let us hasten to add that this conclusion does not exhaust the problems and topics of study, even within the jajmani system. It simply helps to locate the true problems and get rid of a pseudo-problem.
Moreover, the jajmani system is not everything, and it is well to recall this in concluding this chapter. Even in the village money plays a role; this has long been the case but is increasingly so nowadays. In addition, there are other specializations, both large-scale (villages of weavers) and urban. I have neglected all this because it seemed to me that, in the present state of knowledge, the ordinary type of village, and the jajmani, contained the main lesson in connection with this aspect of the caste system.434
The iMtvtetik o/0\
1 /Sto(f
CHAPTER FIVE
THE REGULATION OF MARRIAGE: SEPARATION AND HIERARCHY
108
It is generally agreed that caste is characterized by the obligation to marry within the group, by endogamy. Thus, at first sight, following most of the literature, the regulation of marriage is an expression of the principle of separation: castes separate themselves from one another by prohibiting marriage outside the group, just as they forbid contact and commensality between persons belonging to different groups. At a certain level of segmentation, the caste prescribes endogamy and thus ensures its own reproduction. One belongs to the caste, or rather the subcastc (etc.) of one's parents. In fact this view, no doubt satisfactory from a statistical point of view but theoretically too simple and too narrow, would give rise to numerous exceptions. Here as in other matters, we shall see the necessity for the principle of hierarchy: it in a sense encompasses the principle of separation. Wc shall first recall the importance of marriage in this society and the various rules which relate to it at the level of the caste and also in connection with kinship, then we shall summarize the common view, which stresses endogamy, and finally we shall attempt to reach a more general view. A few details will be added on the subdivisions of caste in relation to intermarriage, and the treatment of marriage from the point of view of the varnas in ancient Sanskrit literature will be recalled.
51   Importance of marriage
Marriage dominates the Hindu's social life, and plays a large part in his religion. The social importance of marriage can be seen from
109
HOMO HTERARCHICUS
a number of features. It is the most prestigious family ceremony, and at the various social levels constitutes the main occasion on which the greatest number of members of the caste and other persons gather together. It is also the most expensive, and the marriage of a daughter in particular is known to be the main cause of debt among Indian peasants, so imperative arc the dictates of prestige, even for the poor.51* For many castes, the celebration of a marriage is the occasion on which a Brahman priest is indispensable, as are the services of the barber, washerman and other castes (services which must be liberally remunerated). It is the occasion for a lengthy series of ceremonies and complicated prestations.51" Finally, it is naturally a matter which is strictly codified for each caste, certain customs having the value of positive hierarchical criteria, like infant marriage, prohibition of remarriage of widows and even the absence of divorce. By its nature, marriage constitutes to a large measure the link between the domain of caste and that of kinship, and the kinship rules relating to it arc as elaborate as the others. It is not surprising to see that adultery is very severely punished when the occasion demands, especially when it takes place between different castes; nor that premarital sexual relations are forbidden - probably universally - for the castes, unlike the tribes.51*
What I translate as 'infant marriage* designates the fact of celebrating marriage long before cohabitation can begin, in fact at a very young age, especially for the girl. The custom is, or rather was, in force among Brahmans and the high castes generally, and was a sign of high status, I say 'was' because it is one of the features which have been fought by the reformers as being both repugnant to the modern mind and devoid of clear religious foundation. However, it is an ancient custom; in Dharma literature, it was essentially a question of getting the girl married before puberty, and for not obeying this commandment a father brought upon himself a supernatural sanction. The age of the girl at marriage became lower in the course of time. In modern eyes the most shocking result was the existence of young 'widows' who had never lived with their husbands.5"1
Among the Brahmans marriage tends to be unique (monogamous) and indissoluble. I say 'it tends' because the duty to have a son makes an infertile union a legitimate ground for exceptions, and the man takes a second wife in such a case. As for indissolu-
THE REGULATION OF MARRIAGE
bility, it is expressed by the fact that divorce does not exist (at most there can be separation) and by the prohibition against remarriage of widows. It is not surprising to see the inferior marriage partner bearing the whole brunt of its indissolubility, and furthermore, the widow leads, or used to lead quite recently, a life of penitence. Among royal castes, things have a different aspect: there i3 hierarchized polygamy, but the remarriage of widows is likewise forbidden, and it is essentially these castes which must be connected with the practice of salt ('virtuous' spouse) whereby the main wife (at least) sacrificed herself on her husband's funeral pyre (a custom also sometimes practised by the Brahmans, and forbidden early on by the English).51*
It has sometimes been written that the castes in general forbade the remarriage of widows. Even as a rough approximation, tin's is very inaccurate: most castes, the overwhelming majority of the population, permit it, and a considerable part of the population even recognize divorce and marry girls after puberty, not before. But a distinction must be made: the true marriage, a woman's first marriage - primary marriage - is universally unique (but not indissoluble). The difference is between castes who forbid and castes who allow the woman, if her first marriage is ended by widowhood or divorce, to contract a kind of inferior marriage, which we shall call a secondary marriage.3,1 In direct opposition to the absence of the woman's secondary marriage, the custom of levirate - or it would be better to say quasi-levirate - is widespread; it allows the widow to marry (secondary marriage) the younger brother of her husband (the elder brother being generally excluded).
1 ittle will be said here about kinship. The place of marriage in the explicit kinship system is very different in the south (the lands of Dravidian language) and the north (of Indo-Aryan language, with some simplification: Maratha country, of Indo-Aryan language, is transitional). In the former case, marriage is central, and the affinal relationships based on it are developed to the point of perfectly counterbalancing blood relationships. In the north, on the contrary, at first sight marriage seems to have a role almost as secondary as in our own culture. However, in the one case as in the other, there arc exogamous groups, and whilst marriage between close relatives is forbidden in the north, this prohibition itself is extended and elaborated among the Brahmans, in theory if not always in practice, in a manner which looks like a way of going one
no
III
HOMO HIERARCHICUS
better. In fact it has been shown in a detailed study that affinal relationships, or what we would call in the south marriage alliance relationships (alliance being an extended and permanent affinity), play an important part, although their direct expression is absent and, so to speak, repressed by 'Aryan' orthodoxy. In spite of the presence of another feature, hypergamy, which will be mentioned again later, and in spite of the prohibition against the marriage of close relatives, the ceremonial roles and prestations which are produced by marriage alliance in the south are certainly also present in the north. The Dravidian aspect is there made incomplete and non-conscious, but it underlies what actually takes place. Therefore it is inaccurate to claim that caste and kinship are two utterly distinct domains: they are united through the importance of marriage, an importance which is obvious in the case of caste and which, in the case of kinship, is from the structural point of view sometimes explicit and sometimes implicit.51 h
52   Endogamy: the usual view and its limitations
For Western common sense, the 'caste' is above all a 'closed' group: permanent, exclusive and self-sufficient. A man of caste X marries a woman of caste X and the children belong to caste X. This fact is expressed in various ways, by saying that the group reproduces itself by itself from generation to generation, is 'endo-recruiting', etc. On analysis, it can be seen that two distinct features are here combined: one marries within the group (endogamy), or rather one is forbidden to marry outside; and descent ((he transmission of group membership) depends on both parents. The first feature contrasts the caste with the tribe and with the majority of societies which tolerate marriage outside the group, even though it generally takes place within. The second feature contrasts the caste with the clan in which descent is (uni-)linear, whether in the paternal or maternal line. Clans, like castes, depend by definition on their regulation of marriage: clan - generally at any rate - is accompanied by exogamy, the obligation to marry outside; caste on the contrary, entails, in relation to its mode of descent, the obligation to marry within.52* Let us note in passing that, compared with a tribe which breaks down into clans, caste society represents a higher order of complexity, since each caste generally has its exogamous clans or their equivalent.
THE REGULATION OF MARRIAGE
It is roughly true that the caste system defines endogamous groups, but it is, of course, a little too simple to be true. We must not forget, first and foremost, that whilst it looks in general self-sufficient for its reproduction, yet the caste is strictly dependent upon other castes from the hierarchical point of view and in virtue of the division of labour. It would be surprising if nothing in the caste's internal organization, in particular marriage, reflected these external features. Further, we have already seen that in general a subdivision of the same kind as the caste, a segment rather than the caste itself, is the unit of endogamy in the sense of the group outside which marriage is forbidden,326 Fvcn so, there is often a tendency to believe that the system is more rigid than it really is: it is commonly fancied that any breach of the rule of endogamy would automatically lead to excommunication. Now it does happen in certain cases that a man X marries a woman Y and that the children are legitimate, and, even more often that a man X has children by a woman Y without either him or them being expelled from group X. Status is really the essential principle here: certain irregularities are penalized simply by a loss of status, and it is at bottom the need U> mainlain the group's status which governs endogamy. In short, castes are self-reproducing because this is a condition for the application of the hierarchical principle by which they are arranged in order. Now the hierarchical principle which ranks castes and their segments docs not stop at the bounds of the unit of endogamy, it permeates it, in a more or less effective way, and endogamous marriage does not necessarily unite spouses of equal status, Finally, when we speak of endogamy or of a unit of endogamy we place oursekes at the level of the rule or law, of what ought to happen; but what happens in actual fact? In practice, one often marries not throughout the whole range of the unit of endogamy but only into a part of it, often a territorial part.52c
If one confined the regulation of Hindu marriage to endogamy pure and simple, one would have to admit a large number of exceptions. But it will be seen that most of the exceptions disappear if two general principles are introduced: (1) endogamy is a corollary of hierarchy, rather than a primary principle; (2) the first marriage must be distinguished from subsequent freer marriages and, a fortiori, from illegitimate unions.SM
1 iz
«3
HOMO HIERARCHIES
THE REGULATION OF MARRIAGE
Si   Hierarchy of marriages and conjugal unions
Let us recall first of all that neither premarital sexual relations nor adultery are tolerated. Next, the only true and complete marriage whereby one moves from the category of an unmarried person to that of a married person is the first. But the ceremony which effects this transition is especially important for the woman, and one must distinguish the case of a male from that of a female. In the ease of a woman \vc shall call the first marriage the primary marriage. Once this marriage has heen contracted, cither it is indissoluble: even by the death of the spouse (superior castes) or else the woman may, after her husband's death or even after divorce, contract another union, legitimate, but infinitely less prestigious, involving much less ritual and expense, which we shall call secondary marriage. Secondary marriage, being of tower status, is freer, sometimes much freer, than primary marriage. In the case of a man his first marriage becomes the principal marriage only if it bears him children, preferably sons. But a man has the option, either in the case of the barrenness of the first marriage, or freely in other castes (royal, etc.), of taking other wives, either with full rite (necessary for the wife if she has not been married before) or with secondary rite (if the wife has already been married). Thus for a man there are supplementary or subsidiary marriages, with a corresponding hierarchy of wives.
Here there are good reasons to say there are two contrasting patterns: the Brabmanic pattern consisting of monogamy (except in cases of barrenness) and a' royal' polygynous pattern (subsidiary wives may be of lower status than the principal one and the sons ranked accordingly). Polygyny is often sororal in the middle castes (naturally it is limited as elsewhere by economic reasons). It must be insisted that in all these cases it is a question of marriages, that is to say, of legitimate conjugal unions uniting not only individuals but, through them, families or small lineages. In addition, the married man who can afford it may have one or several concubines; such a relationship depends only on the parties concerned; the offspring are not legitimate, the sons do not inherit (or only a lesser share). In short, it is seen that in addition to our distinction of legitimacy (between marriages and other unions) another distinction, which is in the main a status or hierarchical distinction, must be applied, between primary and principal
marriage on the one hand and other unions in general on the other.53*
A spectacular example of the usefulness of this distinction was given by Chambard in Malwa (Central India), He found, among I a caste of middle rank, that there were 'woman fairs' in which the women were, at first sight, bought more or less freely from their possessors. This may seem astonishing in India (given that the famous 'marriage by purchase' has never existed anywhere any more than the no less famous hmarriage by capture'). We must reflect that it is only a matter of secondary marriages: the women concerned have first been married (primarily) with all the usual care and solemnity, and it is the husbands who can afterwards 'sell' them, or rather abandon them to someone else in exchange for a money payment. Extreme as it is, the case represents an example of the freedom of secondary marriage and does not affect what has been said about primary marriage, except to this extent: that it can be terminated in an unexpected way."6
There is an important difference between the north and the south of the country. In the south, there is a clearly marked status difference between the two sorts of marriage and their issue whenever these are encountered in the same family. In the north on the contrary the difference is scarcely more than one of ritual and of the prestige of the spouses, and it is not passed on to their descendants.530
The relative flexibility of the system can be seen in the case of illegitimate children, insufficiently known as it is. We may think that excommunication applied only to cases presenting flagrant difference of status, and that the treatment of illegitimate offspring was largely a function of environment and circumstances. The universal principle is that the illegitimate child has a status markedly inferior to that of the legitimate children. Circumstances ptay a part in determining the intensity and expression of the difference of status: will the child be relegated to an inferior position within the caste - a position which may or may not be transmitted to his descendants - or will all relations with the child, and perhaps with his father, be broken off, or will he be attached to his mother's caste, if it is inferior (the most common case)? In the absence of fixed principles, dominance or simply wealth, even locality, and finally the caste's situation, whether or not it is isolated from castes of neighbouring status, must have played a large part. This is how
114
"5
HOMO HTERARCHICUS
one can reasonably picture the situation, though there is some risk of hasty generalization, due to our slender information. At any rate, we can be sure that there was no shortage of bastards, especially in the houses or retinues of princes. This is already something with which to temper our prejudices about the rigidity of the system,33d
54   Isogamy and hypergamy
We have not finished with the multifarious manifestations of status in marriage. We must again distinguish two patterns. In the first, there must be equality of status between the spouses in the primary and principal marriage (it being understood that the woman can be of slightly inferior status in the man's subsidiary marriage -which again is already the case in a sense if the woman has already been married, that is, if the marriage is secondary for her). In this pattern the (principal) marriage will be called isogamous. In the present state of our knowledge, this pattern seems to be general for southern India where, let it be recalled, one often marries what wc would call a close relative, in particular a maternal uncle's daughter. This pattern obviously requires that one be sure of the status of the family into which one marries, or into which one marries one's daughter. And this requires, and above all formerly required, that the unit of endogamy be confined to a limited area, where everyone can know one another directly or indirectly, and that there be sanctions against irregular practices and bastardy, so that the group's status is not endangered. In short, the isogamous pattern must have been accompanied by a high degree of fissiparity of groups within the caste.J4a
The word hypergamy was introduced at the end of the last century to designate a different pattern which is sometimes encountered in the north of India, although it is not universal there.5'"5 In this pattern, a slight status difference, a slight inferiority of the wife's family in relation to the husband's, is considered normal and does not in the least affect the offspring's status. Of course, this relates to the principal marriage, and in no way excludes endogamy. It is because in such a marriage the daughter marries into a superior family (she 'marries up') that it iias, rightly or wrongly, been called ' hyper'gamous. Three remarks can be made: (i) the term 'hypergamous' is not used to designate all
116
THE REGULATION OF MARRIAGE
marriages or unions in which there is a status difference in the direction indicated, but more precisely the fact that such a difference is, within certain limits, neutralized normatively in the first marriage. The term even implies if not an obligation at least a strong recommendation for the girl's parents to find her a superior partner. (2) As the woman is in general considered inferior to the man, the pattern would seem natural to the people concerned. (3) More importantly, the pattern harmonizes best with the Brahmanic-classical and universal ideology of a girl's marriage being a 'gift of a maiden' {kartyu dan). The gift in general is an extremely meritorious action: one acquires merit by the gift of goods to the Brahmans, meaning that one thereby exchanges raw materials of no value for spiritual goods. Now the 'gift of a maiden* is a special form of gift, and it is meritorious on condition that no payment is received for the girl; here the girl is, on the whole, assimilated to a material good, and the giving of her is in fact accompanied by material gifts and by as lavish receptions as possible; in the hypergamous pattern the superior status of the bridegroom's family makes it more demanding about the prestations it receives with the girl, as if it would only accept marriage into an inferior family on condition of receiving hard cash; but this precisely squares with the pattern of the gift: one gives a daughter and goods to a superior in exchange, not in this case for spiritual merits, but for something similar, namely the prestige or consideration which results from inter mar riage with him.5*3
In this case, in contrast with the isogamous pattern, the endo-gamous unit tolerates notable differences of status within itself and even some degree of uncertainty in this respect, since one may take a wife either within an equal group or within a group to a certain extent inferior to one's own. This fact probably accounts for the existence of extremely large endogamous units, for example in the Gangetic plain, which would be inconceivable in the south.
One can formally distinguish between obligatory hypergamy and optional hypergamy, but our knowledge just enables us to see that perhaps the distinction does not in fact have much interest. Imagine a caste which is not segmented but which is composed of a number of clans, each covering a very large area. This is how the Rajput caste (or should one rather say the subcaste of Rajputs proper?) appears in the literature.■* If these clans are strictly hicrarchized in relation to each other, as is. theoretically the case,
117
HOMO HTERARCHICU6
THE REGULATION OF MARRIAGE
then one could not marry an equal since one must marry outside the clan, and given that one cannot marry a woman of superior status,^r hypergamy will be obligatory. In another pattern, of which there are many examples, it is not clans which are ranked but groups having the nature but not the name of subcastes, which, curiously enough, are distinguished by numbers. In this case one may marry within one of these groups (isogamously) as well as outside, and thus hypergamy is only optional. We shall see an example of this.
In the theoretical model of obligatory hypergamy, and to a lesser extent in the case of optional hypergamy, it was observed long ago that men would be supernumerary in the inferior part of the endogamous group, since the women will have married for preference into a superior group, and on the contrary the women would be supernumerary in the superior part, unless recourse were had to large-scale polygyny. The first difficulty was overcome by unions of inferior status with women of other castes, the second by the infanticide of daughters, which the British government fought with success. This situation is characteristic of the Rajputs: infanticide of daughters at the top of the ladder, breakdown of endogamy at the bottom, polygyny among the powerful.
55   Some examples
Let us now give some illustrations of the foregoing, starting with the distinction between primary and secondary marriages. This distinction accounts for an arrangement which would otherwise constitute an exception from the point of view of endogamy. In various groups, in order to secure for women great freedom of (secondary) marriage or of sexual unions in general, primary marriage is, or rather was, reduced to a mere ritual formality. Sometimes women are married in this way to a god, an object, a fruit, or a man who immediately disappears from their lives: in the south this was the case for the Devadasi or ritual prostitutes, in certain districts for the Basavi, girls devoted to giving their fathers a son, in Malabar (Kerala) for the Nayar girls, in order to maintain matrilineal filiation, and, as a remarkable fact, at the other end of the country, for the Newar girls of Nepal (for a purpose which is not quite clear). After such a marriage the Devadasi was allowed to prostitute herself, the Basavi and the
Nayar girls to have unions in which they play the social role normally devolving to the man (transmitting unilineal descent), and the Newar girls probably to have unions with men of inferior status.
The Nayar case is very remarkable and deserves brief mention.55* A girl of the highest status among the Nayars was first of all married with pomp to a Nambudiri Brahman, for whom she and her children would subsequently have to go into mourning. Whether or not there was cohabitation and a divorce ritual, this relationship ceased almost at once, and the girl then contracted secondary marriages with different men, of status at least equal to her own, cither Nambudiri Brahmans or Nayars. The difficulty here is that the primary spouse is not an object but actually a man, and a man not only of very high caste, but also of very high varna: a Brahman, where the Xayars arc considered Shudras. Here then a major misalliance features as the principal marriage! On closer inspection it is nothing of the kind: from the hierarchical point of view it would be a misalliance only for the Brahman, since for the girl the relation is, on the contrary, one of the most honorific (this is its raison d'Stre: to secure her a husband of the highest possible status). Now precisely what is a marriage for the girl and the Nayars is not one for the Nambudiri priests. Among them only the eldest son marries, a Nambudiri girl, of course, whilst the younger sons arc treated in the way illegitimate sons are treated elsewhere: they do not inherit, and must be content with concubinage with Nayar women, the offspring of such unions being Nayars. As for the Nayar girl's primary marriage, for the Nambudiri who figures in it as the 'spouse' it is only a ritual in which he plays the part which devolves on the father in Nambudiri marriages (tying the tali round the woman's neck), a rite de passage in which there is no husband. This is a crucial case. The notion that marriage is universally governed by the principle of endogamy is invalidated by this exception. Let us, on the contrary, posit that endogamy is the result, nowadays general, of the law whereby a caste refuses to take spouses of clearly inferior status, and conversely cannot get spouses of clearly superior status, although it would have no objection. Then the fact that the ceremony does not constitute a marriage for the Nambudiri is enough to make this comedy possible. One can see from this example why it is necessary to regard as primary the principle of hierarchy and not the principle
"9
HOMO HIERARCHICUS
of separation, or a so-called 'repulsion1 in the manner of BougIe\SSb Let us observe further that, contrary to what the use of the same word by certain authors might suggest, the symbiosis of the Nam-budiri and the Nayar has nothing to do with hypcrgamy: there is indeed a difference of status in the usual direction, but this difference is not neutralized; on the contrary it remains relevant in the highest degree, for it is the search for a social father (pater) and physiological fathers {genitores) of the highest status which is, together with the maintenance of matrilineal filiation in a patrilineal environment, at the root of the institution.
A-O
Marriage ponitiiMliei 1-1
fur mtn A for women O
FJgura ?. Hyperpamv among the Rartii Brahmaru in Hangar.
One can take as a relatively simple example of hypergamy the classic case of the Rarhi Brahmans of Bengal. They arc divided into twosubcastes, Kulin (literally 'of [good] lineage') and Srotriya. The Kulins, being superior, were so sought after as husbands for Srotriya girls that they used to marry up to several dozen of them, each time in exchange for substantial presents. Naturally they could not maintain so many wives, so the wives remained in their own families and simply received visits from their husbands (a situation similar to that of the Nayar girls who receive their secondary husbands, Nambudiri or otherwise, in their own homes). Moreover, the Kulin asked for a payment each time. The children were often brought up in their mother's family. The institution did not recover from the High Court judgment which decreed that
THE REGULATION OF MARRIAGE
husbands were obliged to maintain their wives. In detail: the Srotriya were segmented into three and the lowest subdivision was endogamous whilst the two higher ones gave their daughters in marriage to their superiors (to the exclusion of the inverse relation), namely, the first only to the Kulin - and its own members - the second to the Kulin, to members of the first division, and to its own members.55'1 Figure 2 shows that with equal numbers, and not taking into account the Kulins* polygyny, unmarried girls would accumulate among the Kulins, and unmarried men in the second division of the Srotriya. In this example, only the lowest subdivision is endogamous whereas the three others, one of which is first order in relation to the caste, the other two of second order, have optional hypergamous relationships. Such situations have led people to conclude that the caste was endogamous in a stricter sense than its segment, but this does not say enough: it is impossible to reduce the arrangement simply to a principle of endogamy: there is a tendency to secure husbands of high rank, and, when this is impossible, the group closes in on itself. We have insisted on the exchange of goods (a girl, as wife, and money in exchange for prestige).
We have mentioned the existence of a pattern in which the caste is segmented into units indicated by numbers. There are two cases. In the first, the higher group is indicated by a higher number, for example, the Agarwal (Vaishyas of Gujarat and the Punjab) have the 'twenty' (superior) and the 'ten' (inferior). The two divisions are endogamous. The numbers are to be understood as '20/20' (full status) and ' 10/20' (reduced status). As the second division corresponds to a degradation of full status, it is not surprising to find the same exogamous groups represented within the two divisions. The two modes of division, into 20 and 10 and into clans, are independent and their function is different (Blunt, p. 49).
The other pattern, the converse of the preceding one, is more widespread. Its meaning may be understood in Maharashtra, and it is very frequent in Uttar Pradesh. The clans of the Maratha caste are ranked into five levels of decreasing status, corresponding to the ascending numbers: five, seven, etc., ninety-six: the numerical increase is related to the fact that each level includes the previous ones: in theory there are ninety-six clans in all and the first five are included in the second seven (thus the second level in fact consists of two clans). Karve pictures the whole in the form of
121
HOMO HIERARCHICUS
THE REGULATION OF MARRIAGE
concentric circles, in which the smaller arc superior in status to the larger. These groups, being unnamed and probably relatively unstable, are by nature subcastes and not exogamous groups: one can in effect marry within the five clans, and women necessarily do so, for 'the five' accept wives from the inferior groups, but refuse to reciprocate."4
The same arrangement of numbers is found again in Uttar Pradesh, although in this case, so far as we know, the less numerous arc not included in the more numerous. Further, whilst the descriptions, already ancient, generally indicate hypergamy, they are often complex and confused.55' I shall take as an example the Sarjupari or Sarvariya Brahmans, which I have studied. This group will be called a caste or a subcaste depending on whether one considers the Brahmans as a varna or as being also a caste. They are localized in the eastern part of the State, their name indicating a locality, 'beyond (to the east of) the River Sarju (the Rapti)', though they are no longer confined to it nowadays. One hears repeatedly, and it has been printed, that they comprise 'the three and a half houses, the thirteen houses, and the one hundred and twenty-five thousand'. It is difficult to obtain a serious enumeration of the 'three and a half and the 'thirteen'. The word 'house' seems to refer not to a clan, but to a localized patrilineal group, let us say a local lineage group and its descendants even if they have emigrated to another place, (The group, like others, has in addition another division, this being into two sections.) The inquiry confirms that status is in fact ascribed 'to such and such people of such and such a place', for instance to the 'Tivari of Rampur' (Tivari being one of the titles in use in the caste), and this is independent of the overall pattern quoted. This point is essential in many respects. In the first place, the caste is immense, having perhaps two million members. The superior section has very few members, and the inferior section, although comprising the overwhelming majority of the caste, still constitutes a (theoretical) unit of endogamy. The grouping within which one in fact marries is naturally much more restricted. Further, sustained attention is given to status only at the higher levels. A dozen titles are in use, which are evidently quite inadequate to distinguish someone's status. They are the usual titles in these regions: Tivari, Dube, etc. Besides, are wc entitled here to speak of 'status'? The main concern is to prevent the reversal of a given relation of intermarriage, and the
fact of giving a wife makes the group which gives inferior to the one which receives. The 'status' in question has no effect outside the relation of intermarriage; one would therefore rather speak of a certain sort of consideration or prestige, especially as this quality is extremely segmented, being ascribed in each ease to a small local or regional group. In these conditions it is tempting to consider the overall pattern as a rationalization effected at a certain moment, which has subsequently worn thin and been more or less forgotten, and which gave a 'total1 pattern for a prestige ladder which by its very nature tended to disintegrate. It is tempting to generalize, which would account for the diversity and confusion of the descriptions. For example, we find in a legendary account a king of Bengal, Ballal Sen, intervening to codify the order of precedence in certain castes. More importantly, here, as in the ideal, more or less Rajput pattern, mentioned above, status or its equivalent is attributed not to caste segments, but to exogamous groups. The difference with the Rajputs is that they claim to hierarchize entire clans whereas here only localized lineages are in question.s5« This fact must be stressed: status is on the one hand attached to the caste and its segments (these latter being sometimes designated as quantitative aggregates of exogamous groups); on the other hand it is apportioned amongst local exogamous groups. In other words, within these enormous castes, in which hypergamy enables different statuses to be distinguished without the unity of the caste being impaired, status is ascribed not only to endogenous hut also exogamous groups. This can explain the ease with which the native, preoccupied above all with status, can pass from one to the other, calling both of them, when occasion arises, by the term 'jdli', 'birth' which designates, rather, the caste and its segments.55 b
56 Conclusion
To sum up, from the point of view of Hindu India as a whole, endogamy is, in modern times, only an average and general result at one level or another of the principle of hierarchy. In effect:
(1) The separation or closure of one group with respect to those above results fundamentally from the closure of the other groups with respect to those below (Nayar).
(2) Apart from illegitimate unions, whose issue usually suffer
122
HOMO HIERARCHICUS
degradation of status rather than excommunication, marriages are strictly hierarchized, the primary marriage (for a woman) being strictly regulated but capable of being made fictitious, whilst the secondary marriage, where it exists, may be very free. The institutions tend to uphold the group's status but do not prevent as an addition the proliferation of inferior statuses.
(3) Hierarchy, in the form of hypcrgamy, penetrates to the very core of the institutions of marriage and kinship. Not only does it 'temper' the endogamy of the caste segment and transfer strict endogamy to a higher level (caste), but in certain cases (Rajput) it even produces a breakdown of endogamy at the group's lower limit.
It must be repeated that at the present time the specialists are not agreed on this interpretation. Yet the advance that such a view represents on those that are still currently accepted should be readily appreciated. For example, we now see the impossibility of claiming that caste and kinship arc two absolutely watertight compartments. And above all, while unifying the domain under consideration, we obtain an infinitely truer picture of things than if we supposed that the men living in castes either automatically respected the most drastic rulcsor were excommunicated ; moreover, this would contradict the spirit of the system, which is more concerned to classify human possibilities in an hierarchical order than to exclude or punish those who do not conform to its ideal. 1 am convinced that, seen in this light, many apparent oddities become comprehensible. Let us consider a complicated example of sub-caste ranking and relations in Uttar Pradesh, where such arrangements are not uncommon. It concerns the Dhanuk caste in the district of Cawnporc, according to the Census of 1911. Blunt and Ilutton give the case as a curiosity.*6* There are five subcastes whose relations vary from one part of the district to another. A reasonable hypothesis, given the environment, is to suppose a hypergainous pattern, We can then construct a diagram in which the horizontal order from right to left corresponds to a hypothetic-ally decreasing status, and the vertical order, from top to bottom, to an increasing separation between subcastes. The numbers correspond to the order of enumeration (1: Laungbarsa, etc.).
It is seen that the order of the subcastes (horizontally) is variable for the lower numbers: 3 occupies a lower rank on the first line than on the following ones, 1 and 2 change their respective positions between the first two lines and the last two. These are the only
THE REGULATION OF MARRIAGE
arbitrary variations. This is not a serious matter, and these changes are explicable if one considers that, as the caste is small, the classification of subcastes must take into account differences in the behaviour of local groups.
To conclude, let us note that the most common modern change in this matter consists in marriage between different subcastes of
(hypo!helical) order of decreasing status
(a)
'- 2	NE		........y	I
3	NW		-.:.■..!.. .... .--2 00	
4	S	□ □ nca.....LI		
5	E	□ □ □ □ □		
la}
Number of segments [subcastes) not allowing intermarriage
.........O : Girls may marry into another segment.
Figure 3. Segmentation in the Dhanuk caste, Kanpur District (U.P.).
the same caste (n. 112c). This is the better understood if one admits that endogamy is not only not attached ne varietur to a fixed level of segmentation, but also is more an implication of hierarchy than an independent principle.
57   The classical theory: marriage and varna
In general, the authors see in the classical theory either something very different from modern reality, not to say something arbitrary, or else a picture of a state of affairs infinitely more fluid, more flexible, than exists today. Now whilst the differences leap to the
124
HOMO HIERARCHIES
eye - in the first place it is a matter of varnas rather than jatis -yet it can be shown that the general view of the modern situation we have just formulated brings these ancient texts much closer to us. This is not to say they are devoid of difficulties. To keep to what is essential, and to try to grasp the spirit of these texts, we shall leave on one side the long enumerations of the people called 'mixed', said to be the offspring of marriages or unions between different varnas, and designated by names which seem ethnic or professional, but which may also be the names of genuine castes. On occasion I shall suggest a possibly rash interpretation of certain points in the light of modern knowledge.57*
At the outset one can lay down the principle that the marriage which is preferred, to say the least, is the marriage within the varna. As for marriages or more generally unions between persons of different varnas, the texts treat them from two main viewpoints: first from the point of view of marriage (for example Mann, III, 12 ff.), then from the point of view of the categories of people who are deemed to issue from unions of this sort, those called the 'mixed' (Manu, X, 6-39). Less important viewpoints must be added, like the relative classification of sons of mothers of different varnas with respect to inheritance (Manu, IX, 151 ff.).
In the theory of the 'mixed' people, the products of intcr-varna unions, two categories are distinguished according to whether the father is of superior or inferior status to the mother: in the first case, the union is anuloma literally, 'following the hair', or as we should say 'with the grain', which means in conformity with the natural order (the woman being from a general point of view inferior to the man). In the opposite direction the union is' brushed the wrong way' or 'against the grain', pratiioma. In both cases the offspring of such a union have inferior status (see below), but the natural order gives rise to offspring superior to those of unions which go against nature.5711 It is not specified here whether this concerns marriage or just any sexual union. When they are concerned elsewhere with marriage, the authors ignore this distinction, but in practice they exclude pratiioma unions in connection with marriage. Only anuloma unions can count as marriages, although such marriages are not always approved. The point is confirmed by considerations relating to inheritance, in which only a man's sons by equal or inferior women are counted, since it is reasonable to identify right to inheritance with legitimacy.
THE REGULATION OF MARRIAGE
On the topic of the regulation of marriage, the authors seem ambiguous and contradictory. On the one hand it is stated or implied that marriage should he between people of the same varna; on the other it is admitted (elsewhere) that not all children arc born in this way and, even here, there are provisions for certain inter-varna unions as marriages. This is not a matter of a change occurring at the time of the writing of a particular text, for all the texts present this difficulty. Thus Gautama, who in connection with marriage prescribes endogamy of the varna, mentions elsewhere the existence of 'mixed' people. Likewise Yajnavalkya prescribes marriage within the varna or (for a man) in the varna immediately below only, but in his treatment of the 'mixed' people he mentions every possible anuloma combination. Manu has several successive verses which are characteristic (II, 12-15). The first recommends a twice-born to marry first of all a girl of his own varna and then, if driven by desire, girls from inferior varnas. The second indicates that a woman may marry either in her own varna or into a superior varna. The possibility of pratiioma marriage is thus removed. The last verse excludes a Sudrd wife for a Brahman or a Kshatriya, giving as a reason that this would lead to the offspring being degraded to the rank of Sudrd. (Note that a little later (17) Manu condemns not only marriage but even sexual relations with a Sudrd woman.) As usual all this is found again in other texts. By reference to contemporary observations or even directly with the help of other passages (Rcnou), one is tempted to understand that the principal wife, mother of children of full status, must be of the same status, but that one can add inferior, secondary wives, whose children will also be inferior. This corresponds pretty well to the hierarchy of sons, in which the principal son -the only one who is both legitimate and natural - born of an isogamous marriage is ranked above the secondary sons who are either legitimate or natural. The Sudrd wife of a 'twice-born' is often referred to as a pleasure opposed todharma(Vas.,XVIII, 18), and it is not just a question of concubinage, for Vasistha knows of such a marriage (without mantra or Vedic formula) but docs not approve of it (I, 25 6). There are direct references even to the ceremony of this type of marriage (Manu, III, 44): the Sudrd woman shall touch only the edge of the husband's garment, etc. Now we have seen above that Manu prohibits this marriage for the first two varnas. We thus come to perceive a distinction between
126
127
HOMO HIEKARCHICUS
THE REGULATION OK MARRIAGE
what is possible and what is commendable: the ideal is clear, practice adds inferior customs to it, and with respect to these the authors take attitudes which vary both from one to another and within the same text: thus Manu, apart from a word on the abhorred pratilomas, indicates both what is done and what it would be best not to do. (Cf. Ill, 155, where he excludes from the iraddha the Brahman married to a Sudra.) One must thus recognize in these texts both the (religious) recommendations of dharma, and the properly juridical prescriptions, the two being linked and as it were dominated by the overriding and constant concern to rank marriages and unions, wives and children, in relation to each other. It can be clearly seen that marriages are ranked in three categories: (1) intra-varna marriage, indispensable to the twice-born for the maintenance of the status of the lineage (except for certain devices designed to regain it once lost) and the only one permissible for Sudrd men; (2) for the 'twice-born', marriage to a woman of inferior twice-born category; (3) for the same, marriage to a 3udrd woman which, although it exists, is not recommended. It can be understood that this last embarrasses the authors: in particular, is a twice-born really bound by a marriage without the recitation of mantra, or is this only a means of attaching the £udrd woman, who is really from this point of view little short of a concubine? This seems to be borne out by the ambiguity of the position of a Sudrd woman's son from the point of view of legitimacy or inheritance. Sometimes Manu prescribes (IX, 152-3) that if a Brahman has nnc son each from wives from eacli of the four varnas the shares in the inheritance shall be respectively, in tenths, in descending order of the varnas, 4, 3, 2 and 1; that is, he transcribes the status hierarchy into proportions and implicitly assumes that the son of a Sudra woman is legitimate. Sometimes, on the contrary, he disqualifies him, only allowing him what his father was able to give him in his lifetime. In the hierarchy of secondary sons, some authors put the Sudra woman's son in the last, the thirteenth, place. In a more problematic way, one may wonder whether certain complications in these texts do not come from another source. Did there not exist, then as now, two dilferent patterms: isogamous monogamy among the Brahmans, and graduated polygyny among the Kshatriyas? If this were the case, the hierarchical schema would require both the acceptance of Kshatriya customs, though as subordinate, and that the Brah-
128
man should not be refused a prerogative so abundantly enjoyed by the Kshatriya. This might be what has led to many difficulties. But we arc not even at all sure of a close relation between these prescriptions and the practice of the period (what period ?). Did not the Shastras translate caste practices into the language of the varnas ? Here we could only try to reconstruct a part of the conceptual scheme presented by these texts.
139