Encyclopédie internationale
des histoires de l’anthropologie

Marriage alliance in Sumatra viewed in the light of the Panji theme

Bert van den Hoek
Référence complète

Hoek, Bert van den, 1976. “Marriage alliance in Sumatra viewed in the light of the Panji theme” (traduit par Sjoerd Zanen et Jos Platenkamp), in Bérose - Encyclopédie internationale des histoires de l’anthropologie, Paris.
URL Bérose : article3189.html

Télécharger en pdf

Publié dans le cadre du thème de recherche « Histoire de l’anthropologie néerlandophone », dirigé par Thomas Beaufils (Université de Lille, IRHiS UMR CNRS 8529).

The anthropological controversy

Since the publication of Les structures élémentaires de la parenté (C. Lévi-Strauss 1949), there has been debate over what should be given theoretical priority when studying kinship systems [1],

  1. the kin group, with the smallest unit being the family, and its continuation through marriage ; the focus of this approach is the preservation, or management, of power and property within a self-continuing group ; or
  2. the system of relationships, within which the exchange of women and other valuable goods takes place, with brother, sister, father and son as the smallest unit.
    Briefly, in case a., priority is given to the principles of descent and inheritance, in case b. to the principles of ’alliance’ and exchange. The descent–exchange opposition broadly coincides with that between English (Radcliffe Brown, Evans Pritchard) and French anthropology (Lévi Strauss, Dumont).
    Louis Dumont has recently (1971) laid out the discussion in a handy little book, highlighting the contrast all the more sharply. One might also wonder whether the whole matter is not grossly exaggerated, and is in fact a chicken-and-egg issue.
    In this paper, I propose a third way, which, however, is by no means a compromise. In perusing Indonesian material, I have become convinced that the anthropological point of contention is recognised as a problem on the ground, i.e. in the Indonesian field of study itself, and that it is strongly reflected in the social structure of different Indonesian societies. That is to say, the opposition descent–alliance does not arise solely, or primarily, from disagreements concerning anthropological theory, but is more deeply rooted, and, at least as far as Indonesia is concerned, can be traced back to the field of study itself.
    My main focus hereafter will be on Sumatra. The amount of factual material I can cite in the scope of this paper is small, but can easily be supplemented by anyone with knowledge of the field of study. My thesis will be that the problem posed above is the ultimate key to the existence of a patrilineal society, the Toba Batak (the classic work on this is by J.C. Vergouwen), right next to a matrilineal one, the Minangkabau (the classic work on this is by P.E. de Josselin de Jong) and that it can explain mixtures of the two [2] such as exist in South Sumatra (the subject of a recent dissertation by D. S. Moyer 1975).
    In the light of the descent–alliance opposition, it will become clear that these different structures, in addition to the ubiquitous tendency towards so-called double-unilineality, are a logical continuation of each other.

The Indonesian problem as stated in the Panji theme

It may seem a digression or side-path from the stated problem, but for the line of thought I want to follow, it is of immediate importance to pay attention to the classical Javanese Panji theme, which Rassers (1922) has related in a magisterial way to a pre-existing social structure in Java, and to its still existing manifestations, in particular the wayang purwa. [3]
Here, I merely want to draw attention to some aspects that do not tally, neither in the Panji-story itself, nor in Rasser’s analysis of it. The fundamental opposition in both the story and its analysis is unmistakable, namely, the (complementary) opposition between sun and moon relating as male : : female (Raden Inu and Candra Birana). However, the moon appears under the different aspects of waxing, waning and full, and these aspects, according to Rassers, are related to the evil and the good marital pretender for a woman who represents the full moon. From a comparison of the Panji story with a North Celebes myth, Rassers concludes that the unacceptability of a man representing the waxing moon as a husband of a woman representing the full moon, in both cases (Java and Celebes) points to the prototype of a connubial arrangement. In his final conclusion, he follows Durkheim and Mauss (in Année Sociologique 1901–2), and assumes that Javanese society was originally divided into (two) exogamous moieties.
Earlier, however, three aspects of the moon were mentioned, and two male pretenders who, under the sign of the waning and waxing moon respectively, seek a woman (the full moon). From this point of view, the myth points to a relationship between three groups and thus to the circulating connubium commonly spread in Indonesia, for which three clans are the minimum requirement.
However, it cannot be denied that the dualism of sun and moon, male and female is also present in these (and other) Indonesian myths. The question then becomes how dichotomy and tripartition are related, and whether – and if so how – they both relate to an existing or formerly existing social structure.
It seems to me that, all in all, the decisive answer to this question has already been given by Lévi-Strauss in his article entitled ‘do dual organisations exist ?’. After comparing Indonesian with South American social structures, Lévi-Strauss states with regard to Indonesia that ’... ternary structure refers to classes, binary structure to relationships between these classes’ (in : Structural Anthropology (1963) 1967 p. 158). This distinction between classes and relationships seems to me to be of an importance not to be underestimated, even if the question of binary vs tripartite divisions does not do justice to all ethnographic materials of the Indonesian field of study. As for Sumatran societies, the Toba-Batak conform the most to the characterisation given by Lévi-Strauss. The asymmetrical connubium there is basically based on three classes ; it is also, if we follow Tobing’s selection, introduced as such in myths (1956, pp. 36–56, esp. p. 55). The relationship between the classes, which always relate as bride-giving and bride-taking, does rely on the dualism of masculine and feminine, superior and inferior. This is particularly expressed in the gifts exchanged by bride-giving and bride-taking groups, which are referred to as ulos (“cloth” ; female) and piso (“sword” ; male). Together, they thus express a totality that exists purely in the relationship between these two [categories], and need not be related to tribal division. Or, stated abstractly, the totality does not rest on a substance, but on a relationship. The image of the tree of life, which is widespread in the archipelago both in myth and ritual expresses this totality : ’... the tree of life is represented in its simplest form by a spear with a square of cloth, i.e. the banner or Panji’ (Ras, ib. p. 449).
Taking symbols out of their context and then relating them to each other is not a legitimate practice. On the other hand, it can hardly be a coincidence that it is precisely the cloth and the sword among the Toba-Batak that are the signs chosen to point to the relationship between the superior bride-giving and inferior bride-taking group.
That Rassers failed to see the distinction between trichotomy (or more) and dichotomy, between classification and relation, is, in my view, largely due to a time-specific tendency towards perfection. It was in the spirit of the times – following Durkheim and Mauss’s seminal essay on primitive classification – to subsume all aspects of a society under one encompassing whole, that is, to strive for perfect classification. Had Rassers realised that precisely this same striving existed in the culture he studied, his analysis would have looked different. Only Lévi-Strauss, decades later, pointed out the imperfection of classifications, the insoluble problems posed in myths, and the spiritual struggle to overcome contradictions.
Returning now to Panji and the banner as the simplest representation of the Indonesian tree of life : whence does this representation derive its perfection ? Or, to put it in another way, how can this representation acquire perfection ? It is tempting to think that the answer is rather simple, namely, through the union of man and woman, husband and wife. And perhaps there is indeed no better answer, yet from an Indonesian perspective it is nevertheless imperfect. For where does the man receive his wife from, and to whom is he indebted for this ?
The Panji novel begins not with husband and wife, but with Raden Inu and Candra Birana being brother and sister. They are not indebted to each other nor to anyone else. They form the smallest kin group in which both sexes are represented. On their own, they cannot enter into a [marital] relationship ; nevertheless, they fall in love with each other. From the start, the problem is unsurmountable, but Raden Inu and Candra Birana try everything to solve it. From children of gods in heaven they become children of kings on earth, going through the most terrible ordeals before meeting one another (interpreted by Rassers as a totemic initiation rite). The miraculous thing is that this meeting does indeed come about, that they marry, and unite as king and queen with Java under their rule.
So is this a solution after all ? I am in no position to judge this but I may offer the following opinion. There is a happy ending in the solace provided to the unsurmountable problem posed at the beginning. However, the question remains to what extent this final union of Raden Inu and Candra Birana differs from the incest they would have committed in the beginning as children of the gods. About one thing I am certain : their relationship does not, as Rassers argues, lead to a connubium and to an exogamous division of Javanese society. The impossibility of this is easily demonstrated, and it is incomprehensible that a great scholar like Rassers should have entangled himself in mythical pretexts.
Raden Inu and Candra Birana incarnate as son and daughter of kings who are each other’s brothers. So the marriage between them – viewed from both egos – is a union between father-brother-daughter and father-brother-son. Viewed symmetrically or asymmetrically, this relationship cannot result in a connubium. In fact, as a marriage between a classificatory brother and sister, by Indonesian standards the union is downright incestuous.
From these considerations follows the final question : if the myth had really intended to provide a solution, why was Candra Birana not incarnated as MoBrDo of Raden Inu ? The answer is that with the introduction of the avunculate, the totality of the pair would be destroyed. So the myth cannot ultimately provide a solution.
The choice, made after appalling fuss and bother, is for the incestuous relationship to be put on stage at the beginning, which finally, with all its reprehension, brings about a cosmic and political totality.
At the end of the story (in the Hikayat Cekel Wanengpati – the version used by Rassers) a turn is nevertheless made to a connubium, but not within Java itself : it is a relationship in the distance, with the foreign royal house of Keling. This last twist makes the weakness, not of the theme or its elaboration, but of the so-called solution, all the more apparent.
If my sketchy view of the Panji theme is correct, it must have been clear to any Javanese listener in those days that the relationship between Raden Inu and Candra Birana remained incestuous, and no connubium could follow from it. The connubium with the external royal house of Keling is therefore not an appendage of myth, but a subterfuge necessary for the future of society.
How now can the Panji theme shed light on Sumatran social structures ? This can best be elucidated by first considering separately the two components which the above consideration has distinguished :

  1. Dichotomy versus tripartition, and
  2. The nature of dualism.


As explained, the tripartite division, which can also be a multiple division, relates to tribal division or the objective classification of society. The dichotomy that is manifested in its most widespread form as the male : : female ratio does not relate directly to the tribal division itself but to the relations between the parts of the tribe. Both, however, have to do with marriage. The tribal division rests on exogamous kin groups ; the dualism male-female reflects the relationship between bride- giving and bride-taking groups, which together form a cosmic totality but not a classificatory totality. Indeed, starting from a bride-giving and a bride-taking kin group, a third group is needed, which gives women to the bride-giving, and takes from the bride-taking group.
If within any culture the pursuit of perfection may be presupposed, the Indonesian problem in this case would be to match the cosmic totality with the classificatory totality. Rassers does not pose the problem, but does give its solution, though not exactly. Onvlee (1949), in a famous article based on field experience, also hints at the solution. Both point to phratry-exogamy, but neither indicates how the system works. How such a system works is depicted in the following model, which I borrow from P.E, de Josselin de Jong (1952, p. 42), who has shown that this model is indeed valid for Minangkabau.

Fig. 1.
Model done by Van den Hoek

It is necessary to slightly modify or simplify the model here, because de Josselin de Jong himself directly relates it to double-unilineality, a topic not yet discussed in this article.
A, B, C and D are four exogamous clans (with three the solution is possible !), which are in a circulating connubium (MBD marriage). If the women go from A to B, from B to C, from C to D, and from D to A, then it logically follows that A and C together, and B and D together, constitute two exogamous phratries. This logical implication in itself has little value if the phratries were not indeed manifest in the social structure. In Minangkabau, this is the case ; the phratries exist as two parties that, especially in marriages, ritually oppose each other (de Josselin de Jong, pp. 71–82). It is worth noting here that it is not the phratries (though exogamous) but the clans that exchange women. It is not clear (or unknown) whether the phratries relate as male : : female. However, there is evidence that dualism of this nature is recognised in Minangkabau, particularly in the contrast between the Koto Piliang and the Bodi Tjaniago adat [legal codes]. De Josselin de Jong concludes that (p. 82) : ’... it is clear that, just as the clash of the phratries at a wedding cemented the unity of the nagari, so the ritual fight of the nagari was a manifestation of the coming-together, always antagonistically and yet always complimentarily of the two larèh (i.e. the two groups adhering to the two forms of adat) which together make up the totality of the Alam Minangkabau’.
While the Minangkabau social structure classificatorily rests on a quadrilateral division (within which one must imagine the circulating connubium), the totality, as in Java and with the Toba-Batak, is expressed by dualism. It may be taken as probable that this cosmic duality of the Alam Minangkabau rests on the relationship between male and female principles (see also de Josselin de Jong in Contributions 1975, pp. 277–305, esp. p. 292).
The first thesis of de Josselin de Jong’s dissertation reads, ’Underlying Minangkabau social organisation is phratry-dualism and an asymmetrical connubial system’, and it is, of course, this combination that is at stake : cosmic duality incorporated into, or contained by, tribal division. Formally stated, this way the distinction between classification and relation, which Lévi-Strauss signalled in his article on dual organisations, disappears. Other aspects of social structure aside, it seems that here, among the Minangkabau, a point has been reached where at least this issue – dualism vs circulating connubium – has been put in order. To show that this is not the case, we return to the Panji theme.


One can observe that the cosmic dualism of masculine and feminine is widespread in the Indonesian archipelago, operating as an organising principle in all kinds and perhaps all societies of this field of study, and be satisfied with this observation. One can also ask the question of what nature this dualism is, penetrating to the roots of the Indonesian tree of life – and this is the question that dominates the Panji novel and which I repeat here. Male and female, sun and moon, Raden Inu and Candra Birana were introduced in the Panji myth as brother and sister. Is that then the essence of cosmic dualism ; does it exist in the bond between brother and sister ? It has been found that the answer to this must be in the negative : the situation is highly unsatisfactory because brother and sister are not allowed to enter into a relationship, cannot conceive offspring and hence cannot establish a society. To meet the latter requirement, Raden Inu would have to marry off his sister, choosing another woman as a wife for himself. But this would logically bring at least one other party into play. In the Indonesian context, however, this would add at least two other parties. The bride-giving, male group cannot simultaneously be the bride-giving, female group, because the dualism of male and female is asymmetrical. Raden Inu would thus be unable to accept a wife from the party to which he gives his sister and would thus necessarily be indebted to a third party, thus losing perfection. Or rather, the couple, viewed from both sides, no longer forms a totality in that case.
At this point, I do not believe that a major leap of thought is needed to recognise in the Panji theme the problem posed at the beginning of this article from an anthropological perspective, i.e. where the theoretical priority should be placed : with one’s own kin group, or with the union with others outside one’s own group ? The smallest kin group is the family, but as Raden Inu and Candra Birana say goodbye to heaven, and thus to their parents, in the Panji novel, they form the smallest kin group as brother and sister.
Rassers believes that the Panji novel lays the foundation for a Javanese exogamous tribal division ; this is refutable. That the original incest is maintained is defensible on good grounds, but not decisive : why else would both lovers descend from heaven to carry out their intention on earth ? Incest is both avoided and carried out, and the ultimate conclusion, though highly unsatisfactory, must be that the problem remains unresolved. If this conclusion is valid with regard to the Indonesian material, it is also valid for anthropological theory, because no general theory can ever be formulated that does not do justice to all the available material, i.e. anthropological theory cannot go beyond the observation of the opposition of descent vs alliance. However, that observation is extremely valuable where the Indonesian field of study is concerned.
The only possible reconciliation between descent and alliance in the Indonesian context would be a marriage between brother and sister, but this is unacceptable. The corollary of this – assuming the theme occurs throughout the archipelago – is that everywhere there must be a tension in the cosmic dualism of masculine and feminine. The complementarity of this can never be complete, but must always carry signs of antagonism within it – because brother and sister cannot have a connubial relationship.
With all this, it remains possible that the problem, as stated here, is completely absent in some parts of Indonesia. In Sumatra, however, it is so intense that other problems and other structural aspects are dwarfed by it.
Finally, to clarify this consideration, I turn briefly to the phratry dualism among the Minangkabau whose theoretical importance is emphasised by de Josselin de Jong. The relationship between the phratries shows all the signs of the ambivalence of a brother-sister relationship, The phratries are both complementary and antagonistic ; although they are exogamous (p. 82), they cannot enter into a mutual marriage (p. 71). The beauty of the Minangkabau system, however, and this can only now be fully appreciated, is that the circulating connubium of the four suku prevents the two phratries from being directly in a connubial relationship with each other. Incest, in other words, is avoided, although the solution is as apparent as in the Panji story, and the tension remains.

Descent vs alliance in Sumatra

In necessarily brief fashion, some Sumatran societies will now be considered in the light of this Panji theme. The strongest and most explicit expression of the descent vs alliance theme is found among the Toba-Batak. However, for reasons that will become clear later, I will start from the Minangkabau structure treated by de Josselin de Jong (1952) from the point of view of descent.
Not because this approach would be wrong, but in order to trace an assumed dialectic in the Sumatran structures themselves, I will adopt the position of alliance here. The contrast becomes immediately clear if we turn to the theme of ’double descent’ advanced by de Josselin de Jong (pp. 82–91).
Leaving aside whether descent along the paternal line also exists, no one will deny that the most important kin groups in Minangkabau – which determine the inheritance of power and property – are the matriclans. But, building on the Panji theme, it must immediately be added that these matriclans are imperfect in one respect : in order to continue themselves, they must enter into relationships. The alternative to alliance would be incest (i.e. claiming women for marriage among the members of one’s own kin group). Such an option of the myth [4] does not exist in reality, and the choice inevitably lies with alliance, i.e. each bride-giving kin group must be linked to a bride-taking one. If there are basically four exogamous suku, as is the case among the Minangkabau, the idea of alliance idea, assuming a male ego, entails the following :
A male ego marries his mother’s brother’s daughter (MBD), and is indebted to his mother’s brother (MB), or rather, to MB’s kin group, for accepting this woman. The peculiarity of the Minangkabau structure, however, is that ego, or ego’s kin group, does not actually receive this woman. She and her children remain part of the clan into which she was born. Nor does ego himself become part of his wife’s kin group (i.e. there can be no exchange of men, as is sometimes suggested !). This means that in the next generation, ego’s wife’s children are not blood relatives, but affines to him ; the girls among them are potential wives for his sister’s sons.
Special attention should be drawn here to the fact that a marriage is always decided one generation earlier than when it takes place, so that the father – referred to as ego in the foregoing – is directly concerned with the marriage of his daughters, as an affine, and in this case as a bride-taking party. However, this position of ego is not absolute (as his position in his matriclan is), but fits into a system of relationships. For ego also maintains a relationship as part of a bride-giving party, namely his sister’s husband, or rather, with this man’s mother and mother’s brother. Again : marriage is decided one generation earlier than when it takes place, and it is the elder affine who accepts (or gives) the gift. The most significant affine of ego – now considered the bride-giver – is located four terms away from ego itself : it is the brother of the mother of his sister’s husband. The main relative of this ZHMB (a) in turn, is again the brother of the mother of his sister’s husband (b) – note that we are now two clans and two generations removed from ego. The same step can be made again two times (c and d), but in the latter case it takes us back to the matriclan of ego we started from – four generations in the past. These four steps, a, b, c and d form a chain of relationships, and it is important to realise that these are relationships of affinity. They ran through the husband of the sister of ego, and extend over four suku and four generations – at the fifth generation, the cycle starts all over again. And that, among the Minangkabau, is exactly the period that must elapse before a household (parui) is allowed to split up (the so-called five-generation rule). If there were three or five suku, the chain would span three and five generations, respectively.
Now is the time to consider the diagram of the total exchange and descent system given by de Josselin de Jong on p. 38, and depicted below in modified form – namely from the perspective of alliance :

Fig. 2.
Model done by Van den Hoek

As mentioned above, ego’s main affine (as a bride-giver) is the brother of the mother of his sister’s husband ; this figure represents the party who ’received’ his sister, just as ego’s own mother’s brother represents the ’giving’ party (the inverted commas are to indicate that, among the Minangkabau, this ’gift giving’ does not actually take place).
From a patrilineal perspective, the most important affine is his most important blood relative : it is his own father (a) ! The most important relative of (a) is again his own father (b), and so on, up to (d). Does patrilineality not then follow logically from affinity ? So are we dealing with a double-unilineal system ?
The latter position, as is well known, is taken by de Josselin de Jong. He assumes descent along both paternal and maternal lines, and in this way can also explain the five-generation rule. In any kinship system based on exogamous matriclans and MBD- marriage, matrilines are logically implied. If there are four exogamous matriclans, each individual in the fifth generation belongs to the same patriline as his matrilineal relative of five generations before. Since this is difficult to make clear in words, one has to add de Josselin de Jong’s diagram (p. 38). However, the explanatory value of this model depends on whether each individual indeed belongs to both a patri- and a matrilineal descent group. In the absence of knowledge about Minangkabau society, it is difficult for me to judge whether the clues cited by de Josselin de Jong (pp. 82–87) are sufficient. His reasoning assumes that this empirical condition is met and runs as follows (de Josselin de Jong, p. 87) : ’If we turn to Diagram VIII, Chapter IV, which summarises the working of a double-unilateral system with four clans taking part in an asymmetrical connubium, we see that successive generations of members of one matrilineal clan belong to each patrilineal clan in turn. Each matri-clan circulates, as it were, through the ranks of the patri-clans until they have each had their turn. The same combination of matri- and patri-clan recurs after as many generations as there are unilateral clans participating, in this case after four. The fifth generation reproduces the type of the first. This, in our opinion, is the explanation, and indeed the only satisfactory explanation, of the situation that obtains in Minangkabau. The woman in the fifth generation equals the ancestress of five generations back, and only she may therefore assume the rôle of becoming an ancestress of a new independent unit, a parui. Furthermore, before the fifth generation, the cycle was not yet complete ; a breaking-off the connubial process during that period would damage the integration of the entire community in its two aspects, patrilineal and matrilineal’.
The double-unilineal descent system is presented here as the only satisfactory explanation of the five-generation rule, so that the conclusion is actually that this system must logically exist, even if it were not otherwise provable. [5] This does not facilitate a critical dissection of the overall argument ; I make an attempt here :

  1. That double-unilineal descent is a logical implication of the diagram given on p. 38 is unquestionable.
  2. Whether that implication explains the five-generation rule can only be determined on factual grounds ; in other words, it depends on whether both kin groups are indeed recognised (n.b. the fact that the father plays some role with regard to his children is not a sufficient indication of patrilineality [6]).
  3. That double-unilineal descent would be the only satisfactory explanation is immediately debatable. The alternative explanation, based not on descent but on alliance, is that the cycle of reciprocity is completed after four generations.
    However, when the alternative explanation is also exposed to critical dissection, it immediately leads to the next and apparently unsolvable question. Namely, if, as with the Minangkabau, neither women nor men are exchanged, what form of reciprocity exists in the kinship system ?
    I am not sure if I can answer this question in a satisfactory way, but at least it makes sense at this point to include the patrilineal structure of the neighbouring culture, the Toba Batak, in the discussion because there, women are indeed exchanged.
    The bias present in this comparison, which for that matter dominates this entire article, is that there is ultimately a tension between descent and alliance, and that this tension is reflected in both the Panji story and the Sumatran structures.
    If it had not already become clear from the Panji story, the kinship structures of the Toba-Batak show that the notions of descent and alliance, that is, relations of consaguinity and of affinity, are not simply a Western projection on Indonesian material.
    Without pursuing general theoretical arguments, Vergouwen (1933) has convincingly argued this for the Toba-Batak. The Toba-Batak material I aim to cite in the context of my discussion is not voluminous, but the reasoning that follows is more difficult than any of the previous ones. In short, I am aware that I am demanding the utmost from the reader ; the only justification can be that this line of reasoning was once followed, however one imagines it, in Java, among the Batak and among the Minangkabau.
    The additional difficulty in the following consideration entails the distinction between the cosmological and the empirical level in Batak and Minangkabu culture, that is, between ordre conçu and ordre vécu. There is a rigorous separation in the ordre vécu of Toba-Batak between consanguineal relatives and affinal relatives (dengansabutuha and pardongan sabutuhhaon) in such a way that contact between siblings and parallel cousins is to be avoided to the utmost. In contrast, the alliance between bride-giving and bride-taking groups (hula-hula and boru) is emphasised to the extreme. Phratry exogamy does not exist among the Batak [7] and the cosmic dualism of masculine and feminine expresses itself directly in the relationship between the bride-taking and bride-giving group. If this dualism were indeed based on the relationship of brother and sister, surely one might expect that the contact between bride-taking and bride-giving group would precisely be avoided as much as possible, or put in terms of antagonism (cf. the phratries among the Minangkabau).
    With the Toba-Batak, there seems to be a contradiction between the fact that ’incest’ is avoided to the extreme, but at the same time an alliance is emphasised that on a cosmological level – it is suspected – means incest. I am only hinting at the solution in this case, and in this I am plainly speculative.
    How is the alliance between the bride-giving and bride-taking groups expressed ? In its most concrete form in the gift of women and other female goods – ulos – including land, from hula-hula to boru. And conversely in the gift of male goods (but not men) – piso – including money, from boru to hula-hula. This relationship is irreversible, and father-sister-daughter marriage is forbidden, for ’How is it possible that water can flow back to its origin’ (Loeb, in American Anthropologist, volume 35, 1933, p. 27).
    With the Toba Batak, it is not the relationship between father and son – as with the Minangkabau – but rather that between father and daughter (and hence that between brother and sister) is a relationship of affinity. In the system of three exogamous patrilineal groups with MBD marriage, a three-generation matriline is logically implied (cf. above), but there are no signs, in my view, that such a descent group is indeed recognised. There is evidence that affinal relations exceed those of consaguineal relations in importance. However, at a cosmological level – and this is the difficult jump in a train of thought that runs through this consideration – consanguinity is valued higher, that is to say, that the bond between the bride-giving and bride-taking group is valued as one between brother and sister [as if it were a] ’Panji’ relationship. In my view, it is this appreciation that is reflected in the sacred nature of the relationship between hula-hula and boru : as brother and sister, they express original perfection. The exchange that takes place in the circulating connubium, and through which each clan is both hula-hula of its boru, and boru of its hula-hula, cannot be reconciled with this. As in the Panji novel, the contradiction between descent and alliance [at the cosmological level] is insurmountable. But at the empirical level, one cannot afford to let an insurmountable contradiction exist. To put it crudely : life must go on and choices must be made. That choice, in the case of the Toba-Batak, lies with an extreme – that is, sacred – alliance, and an extreme, sacred respect of the bride-taking vis-à-vis the bride-giving group. Unlike the Minangkabau, the idea of the bride-as-gift is indeed realised among the Toba-Batak : the father renounces his daughter, the brother divorces his sister. In return, the bride-receiving group boru is expected to show sacred respect towards the bride-giving group hula-hula, which is indeed worthy of such respect. Apart from the most valuable gifts of ulos, a divine blessing (sahala) emanates from the hula-hula, and this blessing becomes paramount if the given wife remains childless (Vergouwen, pp. 63–65). Thus, the need for exchange is not minimised by the Toba-Batak – which would also be a possibility – but rather the alliance of bride-giving to bride-taking group is emphasised to the extreme, so that their relationship seems to constitute a perfection – which, however, disappears if the system of marriage relations is considered as a whole.
    It has been argued above that among the Toba-Batak there is an extreme emphasis on marriage alliance, and that this alliance can be traced back on a cosmological level to the bond between brother and sister. I have also argued that this line of thought implies a contradiction because brother and sister are not allowed to enter into a relationship with each other. Many criticisms can be made of this assumption, it seems to me, especially that the relationship ordre conçuordre vécu is little elaborated. However, there is one question that seems to me to be of paramount importance : does affinity among the Toba-Batak, as described by Vergouwen (pp. 50–78), indeed constitute the extreme of conceivable alliance relations between exogamous groups, or can this alliance be taken even further ? In my view, the question itself is the key to matrilineality among the Minangkabau. For there is indeed an even more far-reaching form of alliance possible than with the Toba-Batak. This extreme is achieved if the husband returns the children to the party from which he received the bride. Put in a somewhat old-fashioned way, I believe this constitutes the explanation to the origin and essence of the Minangkabau system. Although the man in Minangkabau does not receive his bride, it is also not true that no gift takes place : there is indeed a circulating connubium. However, not one, but two gifts are made, the first of which consists of the bride herself, the second of this woman’s children ; and these gifts go in opposite directions. One might say (but this formulation only serves to clarify) that there is a form of direct reciprocity alongside the circulating connubium, in which the woman is given, and the children are given back. However, this formula is incorrect because the gifts are unequal (NB : it is the children, and not the woman, who continue the lineage group).
    However, I do think it is correct to say that among the Minangkabau there is an extreme of alliance which in practice leads to the inversion of descent (assuming patrilineality) viz. alliance, i.e. matrilineality. This thesis would be overreaching in the absence of corroborative material from other parts of Sumatra, i.e. from Batak other than the Toba (Loeb, 1933), and especially from South Sumatra (Moyer, 1975). It is clear from the opposition of jujur vs ambil anak, which will be discussed further in section 4 below, that, among the most valuable goods exchanged in Sumatra, are not only the women, but also – and partly independently – the children.
    Along this line, the different social structures of Sumatra can be understood, and it becomes clear that the matrilineal structure of the Minangkabau is not a contradiction but a logical continuation of the patrilineal structure among the Toba-Batak, i.e. a continuation of the theme of alliance vs descent.
    In the Minangkabau, the extreme of alliance is achieved – except in the princely house, from which one might suggest that dynastic considerations led to emphasising descent (because royalty was patrilineal).
    Only one effort is now required of the reader, and that is to move to the cosmological level of Minangkabau structure – the ordre conçu. At various levels, the Minangkabau have strived for perfection in dealing with the cosmic theme masculine : : feminine :
  • in the dualism of “monarchy vs people” ;
  • in adat dualism ;
  • in phratry-dualism ;
  • in the dualism of the bride-giving vs bride-taking group.
    In the ordre vécu, as stated above, an extreme of alliance (i.e. the return of the children by the father to the bride giving group leads to a reversal of alliance and descent, i.e. to matrilineality. In the ordre conçu, this maximum of alliance simultaneously means the minimum, or rather its negation. By returning the children to the woman’s group from which they came (in practice : leaving them there), it is suggested that the union never existed, that bride-giving and bride-taking groups never entered into a relationship. I do not want to claim that this cosmic consideration took priority in the formation of the system, but only give my opinion, that if anywhere a solution has approached the Panji theme, this honour belongs to the Minangkabau.

Jujur vs semendo marriages in recent research (Moyer, 1975).

This section contains no new thought, but has been added to seek confirmation of the foregoing, and further because of its connection to recent research (Moyer, 1975). Here I largely abandon the Panji theme and mythic level, and limit myself to the results of the preceding analysis for the ordre vécu of Sumatra. It has emerged, in my view, that the matrilineal system of the Minangkabau – in which the position of the wife and her husband’s children have always been cause for wonder – becomes comprehensible if we assume a patrilineal system – like the Toba-Batak – and thereby assume the maximum of alliance. That maximum is reached when the husband’s children are ’returned’ to the bride-giving party, thereby establishing (still from a patrilineal point of view) a form of direct reciprocity between the bride-giving and bride-giving groups – consisting of the exchange of both wife and children (but in opposite directions). This exchange leads directly to, or is nothing but, matrilineality. That is, a brother does not divorce his sister, but neither does he enter into an incestuous relationship with her.
And, to return for a moment to the ordre conçu : that a man gives his children back to his wife’s party means on a cosmological level that the relationship between the bride-taking and bride-giving parties (still considered from a patrilineal point of view) is in fact nullified, and that bride-giving and bride-taking groups, being the expression of the relationship brother : : sister, have not entered into connubium. Although this cosmological consideration rests on the denial of connubial relations between patrilineages, it leads positively to the establishment of a matrilineal system. Whether such reasoning was followed in Minangkabau thinking itself must remain a question. My hints that the problem, or the opposition brother –sister vs man–woman, descent vs alliance, is recognised in the Indonesian field of study, I initially derived from the Javanese Panji novel.
The gap in my consideration at the level of the ordre vécu is whether the two gifts – wife and children – are really distinguished. That this idea is indeed alive and well in Sumatra is most clearly shown by the marriage law of South Sumatra. In the Sungai Lemau laws (see Moyer, pp. 106–158), jujur and semendo (ambil anak) marriage are opposed. The opposition relates primarily to patri- vs matrilocality, but also, as Moyer shows through fasal (paragraph) 24, to descent (pp. 129–135).
For a patrilineal marriage, a higher bride price has to be paid than for a matrilineal one (there is also an intermediate form, samendo baliek djoerai, see p. 126), clearly showing that a gift of a woman is distinguished from the gift of her children. That the children can belong to both the man’s and the woman’s party implies a real opening in South Sumatra to double-unilineality – if the line were to continue beyond one generation, i.e. if jujur and semendo are considered two forms of connubium. Unfortunately, as far as South Sumatra is concerned, the material is not present from which to derive the total kinship system – as has been done by Vergouwen for the Toba-Batak, and by de Josselin de Jong for the Minangkabau. The following empirical questions remain : a. are lineages or clans recognised in South Sumatra (and how) ? b. If so, is there an asymmetrical connubium between them, and c. is the distinction between the various marriage forms perhaps related to the different ranks in society ?
General double-unilineality is, in my view, the most unlikely form that the kinship system could take, because in that case blood relatives would be confused with kin. Or we would have to assume that the opposition of alliance vs descent, and consequently the ’incest’ problem (as in the Panji novel : consisting in the desirability vs reprehensibility of marriage and sexual relationship with direct or classificatory blood relatives) is absent in South Sumatra. That the latter is not the case can be convincingly demonstrated through the Sungai Lemau laws. The way I can demonstrate this implies a certain criticism of Moyer’s methodology, without, however, affecting his own conclusions. My argument rests entirely on the position of the last fasal (paragraph) 25, in the overall system. Moyer’s method is to work from the parts to the whole : he starts with fasal 1 and 2, then with fasal 3 and 4 – and so on – contrasting them with each other. From a structuralist point of view – according to which view the parts derive their meaning mainly or entirely from their position in the overall system of relations – this method is open to serious criticism. The second criticism is that Moyer is preoccupied with numbers, without valuing the surplus meaning of those numbers ; or rather without demonstrating that these numbers have more meaning than their function in the classification of laws and the monetary amounts contained therein. One could put the question this way : is the numerical system related to a more comprehensive system of thought ? I do not know the answer to this question by and large, but for fasal 25, it must be in the affirmative. Fasal 25 revolves around sumbang (incest), as defined above. From Moyer himself, I know that sumbang can also mean ’the wrong place’, i.e. at a ritual event. 25 is a wrong number and has a negative value. What remains are the 24 other laws. Without denying that a decimal system is also used in the division of these laws, it is obvious that the total is governed by a system based on 12 numbers, in which the main denominator (divisor and multiplier) is 2 (the most obvious proof of this 12-2 division is the wang penurun in fasal 7).
The first 12 laws deal with the construction (grades and taxes) and continuity (marriage) of society. It is only at the 13th that things go wrong : then we come to murder, robbery, theft, illicit sex, and insufficient reciprocity in marriage (24) ; in addition to the sanctions to be taken against this breakdown of society. As Moyer notes, Acts 6 to 12 and 24 all revolve around marriage (interrupted by laws on other subjects), while 24 indicates an important aspect of marriage. (Although they make a droll appearance, the various deadly occurrences in fasal 24 through which the marriage covenant, and also the offspring from it, can be affected, are of paramount importance – which is also recognised by Moyer. A Western reader would do well to separate intention from verbal expression here, as in myths. Starting from the duodecimal system, it must now be concluded that the marriage alliance (rightly seen by Moyer in its aspects of both residency and descent), i.e. the continuity of society (more specifically of one’s own group within it) is central to this system of laws. The finale comes in fasal 25, dealing with incest. While the first dozen laws deal with the positive aspects of society (construction and continuity), the second dozen with the negative aspects (degradation and discontinuity ; and the measures against it) the last law deals with the complete negation of society and the – very severe – measures against it. Acts 6, 12 and 24 indicate, both in content and by their numerical relationship, the importance of marital alliance ; 25 relates to the inversion of alliance, i.e. incest (sumbang). I do not think an example can be found anywhere in which alliance and incest (i.e. the union between ’descendants’, blood relatives) are so sharply contrasted as they are here. Ultimately, however, it concerns the same theme that is subtly and romantically displayed in the Panji novel.


Dumont, L. 1971. Introduction à deux théories d’anthropologie sociale. Groupes de filiation et alliance de marriage – Paris/The Hague.

Durkheim, E. Mauss M. 1963 (1903). Primitive Classification. English translation and introduction by R. Needham – London.

De Josselin de Jong, P.E.(1952). Minangkabau and Negri Sembilan. Socio-political Structure in Indonesia – Leiden.

De Josselin de Jong, P.E. 1975. The Dynastic Myth of Negri Sembilan (Malaya), Bijdragen tot de Taal-, Land- en Volkenkunde, vol 131, 1ste and 2e installment.

Lévi-Strauss, C. 1949. Les structures élémentaires de la parenté – Paris.

Lévi-Strauss, C. 1967 (1963). Do Dual Organisations Exist ? (Les organisations dualistes existent-elles ?), Structural Anthropology (1974) – New York.

Loeb, E.M. 1933. Patrilineal and matrilineal organisation in Sumatra : The Batak and the Minangkabau, American Anthropologist 35.

Moyer, D.S. 1975. The logic of the laws. A structural analysis of Malay language Legal Codes from Benkulu – The Hague.

Onvlee, L. 1949. “In response to the dam at Mangili” ; in Contributions to the Language, Land and Ethnology, vol. 105, 4e issue, pp. 445–459.

Ras, J. J. 1973. ‘The Panji-Romance and W. H. Ras’s analysis of its theme’, in : Contributions to Language, Land and Ethnology, 129 (4).

Rassers, W.H. 1922. The Panji Novel – diss. Antwerp/Leiden.

Tobing, Ph.O.L. 1956. The structure of the Toba Batak belief in the High God – diss. Utrecht.

Vergouwen, J.C. 1933. The legal life of Toba Bataks – The Hague.

[1Manuscript translated from Dutch into English by Sjoerd Zanen & Jos Platenkamp.

[2Only tentatively so designated ; that these cases involve something other than double-unilineality will be discussed in the following.

[3For a summary and discussion, including an addition to Rasser’s analysis, see J.J. Ras, in : Contributions 1973, Vol. 129, 4e issue

[4Whether a corresponding myth like the Javanese Panji story exists or has existed among Minangkabau is unknown to me.

[5Yet the last sentence of the given quote, if I understand it correctly, hints at alliance.

[6In anticipation of my conclusions, I would like to advance one logical ground on which it can be expected that the search for patrilineal descent should be barren. The fact that ego’s principal affine would at the same time be his blood relative gives the most hideous confusion imaginable, because it directly implies incest.

[7That there has indeed been an effort to reconcile a classificatory tripartite division with a relational (cosmic) dichotomy is shown by Tobing’s (1952) dissertation, which, incidentally, again did not pose the problem.