Part 9

Still another objection is implied in Dr. Sapir’s own statement of the case. It is not necessary for the natives to look at the levirate from the point of view hitherto assumed. Instead of defining the paternal uncle in terms of his potential fatherhood, they may have a word distinct from that for father to designate the step-father and the paternal uncle. Dr. Sapir cites the Upper Chinook by way of illustration. In other words, the action of the levirate is equivocal. It may affect nomenclature so as to produce the semblance of the Dakota principle, but it may also produce quite different results. It may also fail to affect terminology at all, as apparently is the case in Semitic languages with their descriptive nomenclature.

In this connection a qualification must be made that applies equally to the exogamy hypothesis. Though the ultimate cause of a terminological feature be the levirate, the immediate cause in a given instance may well be an historico-geographical one. If the Chinook nomenclature is differently affected by the levirate from that of the Yahi, the proximate reason may be simply the fact that the Chinook did not come into contact with the same peoples as the Yahi and thus had no chance to borrow their nomenclature. In other words, admitting an influence of the levirate, it is not necessary to assume that it has repeatedly produced the same terminological effects independently.

I know of at least one instance in which the hypothesis advanced by Dr. Sapir seems definitely excluded, leaving exogamy in the field as the efficient cause. The Hopi system conforms to the essentials of the Dakota type, but neither the levirate nor the marriage with two sisters is in vogue. It cannot be argued that the Dakota features were borrowed from some other Southwestern tribe possessing these usages, first, because the Dakota features are far more highly developed among the Hopi than among other Pueblo Indians; secondly, because it is very doubtful whether the practices in question occur among other Pueblo tribes.[48]

In justice to Dr. Sapir it must be pointed out that he does not advance his hypothesis as a general interpretation of the phenomena. As he suggests, it is most serviceable where the exogamous factor does not occur, or, as I should add, where diffusion of features from a system affected by exogamy seems improbable. I have examined his hypothesis as if it were designed to account for all the relevant phenomena simply in order to bring out clearly its inferiority from this point of view to the theory of exogamy.

There are two series of cases which strongly corroborate the theory of the effect of the exogamous organization on the kinship nomenclature. They constitute a distinct variant of the Dakota principle, the deviation being in the designation of cross-cousins. While these are still differentiated from parallel cousins, they are not placed together in a single category but are classed, one group of cousins with the first ascending and the complementary group with the first descending generation. In short, the generation factor which is fundamental in the Hawaiian scheme and only modified by dichotomy in the usual type of bifurcate merging schemes is here overridden by some other factor. Now what is the nature of this new determinant? Let us look at the facts.

The Hidatsa class the father’s sister’s son with the father and the father’s sister’s daughter and all her female descendants through females to infinity with the father’s sister; correlatively, the mother’s brother’s son, in the absence of special words for nephew or niece, is classed with the son, even by women. That the Crow scheme is almost identical, is readily intelligible from the historical relations of the two tribes, who speak very similar languages of the Siouan stock. But the essentials of the classification reappear among the geographically, linguistically, and culturally remote Hopi, with suggestions of similar features among the Tlingit and even in Melanesia. We are again confronted with a puzzling problem of distribution.

An analysis of the Hidatsa data clarifies the situation. According to the statements of the natives themselves, the term ‘father’ is applied to any father’s clansman irrespective of age and would accordingly include the father’s sister’s son. This suggests that the clue to the entire situation may lie in the clan feature. As a matter of fact, we find the daughter of the father’s sister’s _son_ is not classed with the daughter of the father’s sister’s _daughter_. The only difference that can be connected with this distinction is that in clan membership: the former relative, owing to the exogamous clan system, can never, and the latter relative always must, belong to the father’s sister’s clan. Hence the former, being a father’s sister’s son’s, _i.e._, a ‘father’s’, daughter, becomes in Hidatsa speech a sister, while the latter is designated by a word translated paternal aunt’ but really embracing likewise all the lower generations of females in the paternal clan. That we are dealing with the clan factor, is corroborated by the fact that in Hidatsa terminology the mother’s brother, instead of being designated by a specific word, is classed with the elder brother, a term also applied to the mother’s mother’s brother. The last-mentioned kinsman may be similarly addressed in Hopi.

Powerful corroborative evidence is supplied by a second series of facts. Among the Omaha, where descent is reckoned in the paternal line, the father’s sister’s daughter is no longer classed with the father’s sister but with the sister’s daughter. These, it may be noted incidentally, would belong to the same division if the moieties of the Omaha were at one time exogamous, for which there is some evidence. But the essential point is that here the mother’s brother’s son and all his male descendants through males are indiscriminately classed with the maternal uncle. It is clear that they are all members of the same gens, and corresponding to our Hidatsa experiment we find that as soon as we pass outside the gens the terminology changes: my mother’s brother’s _daughter’s_ son is not my maternal uncle but my brother since his mother, the uncle’s daughter, is called ‘mother’, belonging as she must to my mother’s gens.[49]

The Omaha phenomena are absolutely paralleled not only among other Southern Siouans but also among a number of Algonquians, _viz._, the Miami, Sauk and Fox, Kickapoo, Menomini and Shawnee. The area covered is an absolutely continuous one, and it is impossible not to explain such a distribution by diffusion. This conclusion is accentuated by the fact that the Ojibwa, though an Algonquian people with a gentile system, do not share the Omaha variant of the Dakota scheme but conform to the more usual type found among their neighbors, the Dakota. The mere presence of a gentile organization, though doubtless a favorable basis for the development or adoption of the Omaha scheme, is not the only determining condition; the presence of terminological features in a particular tribe is also a function of its geographical position or historical connections. This does not interfere with the ultimate interpretation of such features but it shows the necessity of taking into account the geographico-historical situation. At present I cannot suggest what may have been the differential condition that produced the Hidatsa variant among some tribes with a clan system but not among the Iroquois; or the Omaha variant among certain Algonquian tribes but not the Ojibwa.

The exogamy hypothesis, with special reference, to the phenomena just mentioned, has recently been discussed by Professor Kroeber.[50] He accepts the empirical correlation between exogamy and the merging of lineal and collateral kin with bifurcation of the parental lines, but interprets it as due rather to the differentiation of male and female lines of descent than to exogamy itself, which latter he regards as ‘perhaps a common but not necessary development, and an overlying development of the former’. “The basic condition,” argues Dr. Kroeber, “would be that in which a woman would be felt to be a very different thing from a man in relationship--less perhaps as an existing individual than as a factor in the relations of other people. Once this point of view prevailed, cross-cousins would necessarily be felt to be something very different from parallel cousins, and cross-uncles and aunts from parallel ones; and the distinction would find expression in nomenclature.” Accentuation of the male and female lines of descent with greater weighting of the one would possibly lead to clan groups.

As a theory of the origin of exogamous groups I have no particular objection to offer to the foregoing. For reasons to be stated below (p. 163) I heartily concur in the assumption that the family, in America at all events, preceded the clan or gens. If I understand him correctly, Dr. Kroeber’s remarks merely paraphrase the fact of this sequence. But I do not see that acceptance of his view on this point involves a rejection of the influence of the clan when that has once developed. Of course it is not directly exogamy that is expressed but the alignment in groups which exogamy brings about. On Dr. Kroeber’s assumption it is unintelligible why father’s sister’s son and mother’s brother’s son should so frequently be classed together since the one is clearly related through the father, the other through the mother. We can hardly credit the native mind with a tendency to algebraic equalization of a plus and minus quantity by which the product of a male and a female relationship shall be standardized by a common designation. Generally speaking, Dr. Kroeber’s factors explain only bifurcation but not merging. The fact that even remote father’s cousins are grouped with the father is what the clan or gentile hypothesis explains over and above the dichotomy of relatives. That such merging occurs among tribes with definite exogamous groups, and generally not in loosely organized ones, can hardly be an accident. Dr. Kroeber’s case is, however, weakest as regards the Hidatsa and Omaha variants of the Dakota scheme. If ‘unilaterality of descent’ rather than clan or gentile affiliation is the determinant here, then why is the Hidatsa variant uniformly found among matrilineal tribes and the Omaha variant uniformly with a gentile system? In other words, why does not the Omaha call his father’s sister’s son ‘father’ and his father’s sister’s daughter ‘aunt’? The cross-cousins in question are as clearly related to me through the father among the Omaha as among the Hidatsa, but in the former case they are not, and in the latter they necessarily are, my father’s clansfolk. Similarly, the mother’s brother’s son and his male offspring are as emphatically related to me through my mother among the Hidatsa as anywhere, but they are not aligned in the same social group with one another and they are not classed together in terminology. For the sake of clearness I will, at the risk of repetition, formulate what I consider the probable course of events. Among certain loosely organized tribes the bifurcation of immediate kin evolved, as we find it among a number of our Far Western tribes. This tendency was amplified and became superseded by a definite clan or gentile scheme. As this scheme developed, possibly as a part of its growth, kinship terminology became not only forked but more inclusive as well. Finally, the fully established organization was able, in certain instances to exert the extreme retro-active influence on nomenclature revealed in the Hidatsa and Omaha variants.

In his extremely valuable paper on Miwok organization[51] Mr. Gifford also suggests a rival explanation in place of exogamy. The Miwok of California are organized in approximately exogamous moieties, and their nomenclature bears some resemblance to that of the Omaha. More particularly is the mother’s brother’s son (and his male descendants through males?) classed with the mother’s brother. According to Mr. Gifford, this is due to the custom of a man marrying, either polygamously or after his wife’s decease, the daughter of his wife’s brother. This form of marriage is actually practised among the Miwok in addition to the more generally diffused marriage with the mother’s brother’s daughter. Obviously, the facts of terminology are consistent both with this usage and with the moiety principle. Mr. Gifford objects that among the Miwok “there are no clan or moiety brothers and sisters, all relationship being based on blood and marriage ties.” This, however, is not the essential point. It does not matter whether the unrelated members are _called_ brother or sister provided they are aligned together in the same social group; the very existence of such social groups implies a differential attitude towards fellow-members as compared with the rest of the tribe. That mere affiliation along moiety lines does not solve all the mysteries of Miwok terminology, is quite true since a sharp distinction is drawn between the mother’s brother’s daughter and the father’s sister’s daughter. Since both these relatives are eligible mates from the point of view of exogamy while as a matter of fact marriage with the paternal aunt’s daughter is prohibited, Mr. Gifford’s objection seems to be sustained. That is to say, here the social organization explains the classing together of certain relatives but not the exclusion of certain other relatives, while the specific marriage regulations of the tribe do account for this phenomenon. But on the other hand, the marriage rules fail where the moiety hypothesis succeeds. Why are the mother’s younger sister, who cannot be married, and the father’s brother’s wife classed with the marriageable cross-cousin and the wife’s brother’s daughter unless it is because they are all members of the same moiety?

So far as the merging of a maternal uncle’s male descendants through males with the uncle himself is concerned, I do not see how any marriage rule would directly explain the extension of the term _ad infinitum_ while moiety alignment at once renders it intelligible. An advantage which the exogamous principle enjoys over every special marriage rule is the universality of its sway over the population. An individual’s wife may not have a brother and her brother may not have a daughter for the husband to marry, but where exogamous groups exist every tribesman is by birth a member of a particular group.

To the subject of specific marriage rules I shall have to revert below. My position as to the Miwok nomenclature is that special regulations undoubtedly account for some of its features while the dual organization successfully explains others and more particularly the Omaha variant of the Dakota principle.

We may sum up our discussion of the Dakota principle with the statement that its distribution, coupled as it is with exogamous groups, supports the theory of an organic connection between the two phenomena. On the question which I have hitherto shelved, _viz._, whether it is exogamy in any form or more particularly the dual organization that gave rise to the features under discussion, I am at present unable to reach a definite decision. Though the distribution of the moiety is far more restricted than that of exogamous groups generally, there is no doubt that not a few elements of the Dakota principle are most readily derived from a dual organization. It remains for the future to determine what is the relative part taken by the multiple kin group and the moiety organization in fashioning kinship nomenclature.

Before leaving the Dakota principle, it seems desirable to allude to two important theoretical problems with which it seems connected--its relations to the Hawaiian principle and its bearing on the antiquity of the clan organization. The Dakota scheme in its more usual form may be logically regarded as merely a complication of the simpler Hawaiian one. As Morgan pointed out, the two coincide in practically half of all the relationships. Inspired no doubt by the general trend of evolutionary thought in his day, Morgan converted the logical connection into an historical sequence and assumed the priority of the simpler system. He indicated how, if grafted on the Hawaiian scheme, the clan or gentile organization would transform it into the Dakota type. It does not seem to have occurred to him that the evolution might have taken place in the reverse direction. Development, as shown precisely by linguistic phenomena, such as the history of the English language--and kinship terms, no matter what else they may be, are elements of human speech--is not always from the simple to the complex. Morgan’s belief was influenced by the view that humanity started their social existence at an extremely low level, for which opinion he found support in the social conditions he inferred from the Hawaiian schedules. These, he argued, suggest brother-sister marriage since such marriages would explain the use of the same term for mother’s brother and father. Such unions certainly _would_ produce the observed terminology but Morgan failed to consider that an alternative explanation was at hand. His fundamental error lay in attaching to the primary kinship terms of the Hawaiians and other peoples the notion of actual cohabitation. From this starting-point he consistently argued that all men addressed as father had actual access to the speaker’s mother. As Cunow has well shown,[52] there is not a tittle of evidence that this represents the native point of view, from which the term ‘father’ merely indicates tribal status with reference to the speaker. When we have once recognized this fact, there is nothing so intrinsically primitive in the Hawaiian scheme of ranging kin as to demonstrate hoary antiquity.

All empirical considerations, indeed, point in the opposite direction. For one thing, all the peoples whose systems are characterized by the Hawaiian feature rank relatively high in the scale of civilization. No one would dream of placing the Maori culture below that of, say, the Fijians. Secondly, we have the most powerful circumstantial evidence from distinct quarters of the globe to prove that Hawaiian features develop secondarily within the Dakota scheme. Thus, among some Iroquois tribes, the tendency has developed to call the father’s as well as the mother’s sister ‘mother’. The Crow differ from all other Siouan tribes, even from their closest relatives, the Hidatsa, in similarly extending the word for mother in direct address. Among the Torres Straits Islanders a corresponding change of usage was recorded by Dr. Rivers,[53] and similar developments seem to have occurred among the Gilyak.[54] Relevant data from West Africa have already been cited in another connection.

All this does not prove that as a general proposition Morgan’s sequence must simply be inverted. For this there is no evidence in North America, where complete Hawaiian schemes, or even approximations thereto, are lacking. But the data at our disposal do indicate that in so far as a tendency toward Hawaiian elements appears it is often due to secondary development.

To turn next to the problem of the exogamous kin group. Some theoretical writers have assumed the priority of the clan or gens to the ‘loose’, _i.e._, clanless or non-gentile, organization in which the family and local group usually form the only important social units. To support such a view appeals have sometimes been made to kinship nomenclatures. So far as North America is concerned, this argument is certainly without foundation. It was Dr. Swanton, I think, who first showed that in North America the exogamous system is found precisely among the more highly cultured tribes while generally speaking it is lacking among the more primitive peoples. Now as I have shown above, exogamy in North America largely goes hand in hand with the Dakota principle. It is therefore rather remarkable that the more primitive clanless North American tribes of the Plateau and neighboring regions also lack the Dakota principle. The suggestion sometimes offered that a clan or gentile system has once existed and simply eluded the field worker’s scrutiny on account of the degeneration of aboriginal life under modern conditions thus breaks down. We cannot argue positively that where the Dakota principle reigns exogamy must necessarily have occurred, because the correlation, while high, is not perfect and because the principle may have been borrowed without the social organization. But an exogamous organization is so frequently associated with the Dakota principle and there is so little reason for a change of kinship terminology provided the native language is preserved that the total lack of Dakota features over a wide area may be regarded as exceedingly strong evidence against the former or at least ancient existence of exogamous groups.

_Supposed Features of ‘Classificatory’ Systems._ Under the misnomer ‘classificatory systems’ some writers have included consideration of the principle of differentiating elder and younger brothers and sisters. The distribution of this distinction is simply staggering when one attempts to trace it more or less systematically. Of North American systems, I can offhand recall only two, the Pawnee and Kiowa, in which it does not appear. We find it in association with the Hungarian and Chukchee terminologies, both of which lack the Dakota principle, and it occurs with the Hawaiian no less than the vast majority of bifurcate systems. So far as I know, the only one who has offered any explanation of the phenomenon is Dr. Rivers, who once connected it with a difference in the time of tribal initiation.[55] But since there are many peoples, _e. g._, in North America, who do not practise any form of tribal initiation, the hypothesis hardly seems tenable and we must rest content to accept the facts of distribution.

Another feature that is often erroneously treated in association with the Dakota principle is that of reciprocity, which has already been referred to as the usage of designating a pair of relatives, more particularly two belonging to different generations, by a single term. Thus, the Shoshone call the mother’s father and the daughter’s son (man speaking) by one term. Such usage would be manifestly opposed to the Hawaiian principle with which it does not seem to be associated. It is found in connection with the Dakota scheme in Melanesia and particularly in Australia, but is markedly absent from the merging systems of North America. Since here it is highly developed where the Dakota principle does _not_ occur, it cannot be regarded as an essential element of ‘classificatory systems’. The question remains how we are to account for the facts of distribution. Australian data forcibly suggest that, there at least, the reciprocal feature is a reflection of social organization. Grandparents and grandchildren, by the curious rule of descent that regulates affiliation with the matrimonial classes of the area, are necessarily in the same class, _i.e._, a father’s father and a son’s son or a mother’s father and a daughter’s son (man speaking) are fellow-members of a class. The fit seems too close to admit of an accidental association. But when we turn to the North American region of reciprocal features the interpretation no longer holds since no vestige is found there of any institution that might align the relatives under discussion in a common group. The inference is that there has been convergent development, and perhaps the most plausible explanation of the North American terms is that they are designations not so much of the relatives as of the relationship itself.[56]