EVIDENCE FOR COMPLEX ALLIANCE IN TIKOPIAI David B. Eyde University of California Berke ey -r,. I -= 'rdv , I I~ Irrduoi.o lon Recent years have seern a great deal- of interest in patterns of cog- nitive organization in cu1tral materlal--as opposed to the sort of emotion1 patterning that was a focus of attention in t-he flifties0 The newer approach seems to have been developed by a runber of people independently, so that ttpattern, t? gestal1t, "st ructue"g pramnar,'1 and "sworld view" are all phrases meaning ;rvery simila r thi%.s In the course of orkirng w_1tf1 Raynrn Flia hs mraterial on Tikopia, a pattern of this sOrt ha,s merged wioh seems to me to have some import- ance, both because It Is more Comp-ex than any simple notion of binary opposition, and becatuse ir various permmutations it appears to turn up elsewhere in Oceania--in the T arobriads, and in the Asmat of southwest New Guinea, at least. iJ ihe Fetrn Thi's pattern, whJich ia;cves the notrion o.f two haltves, four corners, and a central position, turnS up in marny areas o? the culture of Tikopia. Let me start with t,he hsA: , pThn0 Frth (l.93667-8l) says: The QU hcY se ti tn-e i ic: 0 0 0 0 ;dThe general floor space is d;` e 2rt Te A-d rS ate-ier ar phas :S jCt I ls wi'thout clearly defined boumnda:i;2 Th c.t i ; .l a:oa known technically as roto a pi, or :ct 1or= the d2> f 9 h: eo ;I. Is oMmTon ground to all the m,em~rs wif #heY c;^sehsL..) ,bject>+ es .) Jc. In from outside are temporarily set down there, meals are spread out, and it is a kind of neutrai area for the performance of a-ll sorts of actirvity. On the one hand lies mata p on the other tuaum.u The former signifies literally the t"face" or "eye" of the houses it is the ceremonial side, where men alone may sift, and on which food is laid out in a ritual manner0 In many houses a large portiorn of it., that towards the eaves, remains permanently unoccupied, excep.t uring a funeralo It is treated with respect, in that pe_ple dnot turn -e? 1backs towards it, and when they lie down to sleep the`y orient th-e ir heads in that direction-- or at least refralin from poinrting t. ei feei B t.-hither0 o 0 The oppo- site side, tuaumu, Signifles literally "the back of the oven," though the expression is really figurative0 Here are the household fires, close to the Wall, and in their vicirnity sit the women and children, facing towards the nata paito. If .he terms are not understood in too strict a sense9 tuaumu may be called t'e prcfane side of the house, as against iata p , the sacred slde, with rote a paito, the centre, as neutral ground between themo o o * o * o o o o a o u 0 0 o o o 0 o * o e 0 o Q e * o e * o * o o The use of the vari-ous doorways depends upon thelr position in relation to m_ata aito. Tlis is always left, blank of entrances0o 0 0 Tuaumu contains the side entrances used normally by women and children, 4O and basual'Ly by neighbours living on that side or by men golng to the took-house. For more ceremonious purposes men go in by a doorway at the end of the house, near which the most obvious path runs. Entering by this it is usual to find one has mata paito on one's rlght hand, tuaumu on one's left, Another feature of the house demands attention. As part of its structure four stout posts stand some distance in from the corners0 . 0 0 It is a mark of some consideration to be allowed to use a house-post as a back-rest; women do not share in this privilege, held only by the senior male members of the household0 0 o In an old house or that of a man of rank a certain precedence is followedo One of the posts on mata paito is the private back-rest of the head of the house; the others may be left vacant as posts for occasional vilsitors of importance, or for religious reasons may never be utilized at allo The posts on tuajmu will serve the resident brothers or married sons of the head of the house6 0 o 0 It ls clear that both rank and religious beliefs are refletted in the use made of the ordinary struttural features of the house, its floor space and covering9 posts and doorwayso So much for the layout of the house. Now look at what Firth (1955' 313-314) has to say about the 'layout of the tniverse0 After distussing the ten levels of heaven, Firth sayso RathLer Ill-fused with this schematic arrangement o Is another, rather less systematized, but more elaborateo There is a directional allocation0 The Tikopia of course traditionally had no compass points, but they divided the circle of their horizon up into something very like it, in the form of wind points--the directions from which the major winds comeo Each of these major. wind points, of which there are four, is the home of a major deity of the Tikopia, who is a prime clan and lineage god. He controls the wind and weather from that particular directiono * o . Each of these quarters, as we may call them, is In spirit terms made up of a number of Rangi. Names are given to these heavens, according to their various characteristics. They are arranged in layerso 0 0 0 In the topmost division lives the deity who is known as the Post or Stay of the Heavens, the controller, who presides over his whole set of heavens. Below him live the "small gods," as they are sometimes called, and the souls of ordimary men. So we see that the universe, like the house, has four corners, each marked by a posto Is there also a division of the heavens into halves? It appears that this is the caseo Consider the following quotation from Work of the Gods (1940:2,-3), bearing in mind that each of the four corners of the universe is identified with a windo Discussing the alternation of the trade wind with the monsoon period, Firth says, Th'is seasonal change, which is accompanied by changes in economic pur- suits such as fishing, is the min bsis for the seasonal ritual9 as is shown by the fact that the ritual is called by the names of "the Work of the Trade Wind" and "the Work of the Monsoop" respectively. About April the wind, after veering through several points, settles down to blow steadily with moderate force fmm the south-east or east south-east and continues thus with hardly a9t iotermissioa for about six months. 9 About Octobgr, the tsra Liwd diet asw a}d 41 replaced by a variable period of normally light winds alternating with flat calms. The days are often very hot and cloudless. Occasionally, the wind, which may have come from any direction from south-west to north, suddenly sets in strongly from the north-west and may rise to gale force for a few days. Rarely, perhaps once a decade, it becomes a hurricane which destroys houses and crops. Such a hurricane occurs only in the raki--the monsoon, and never in the t-ona--the trade wind season. I think it is clear that Just as in the house the roto a paito is conceived of as in the center, between four corner posts, and at the meet- ing point of the two halves--mata paito and tuaumu--so the island of Tikopia is conceived of as in the center, between the four corners of the universe, and at the meeting point of the two halves: raki and tonga. This is what I try to show in figures I and II. It my be that the island of Tikopia is itself conceived of on the same pattern. At least the island is divided into two halves: Faea and Ravenga, with the sacred center of the island at Uta, which is special in a number of respects, although it is reckoned a part of Ravenga. The people of Faea and Ravenga have a ritually competitive relationship, but cooperate jointly in the sacred ceremonies at the Marae in Uta. I can find no refer- ence to an idea of four corners to the island. Now lets look at the marae. Firth (1940:207) says: The dancing place in Uta is known as Marae. The term is applied descriptively to other open spaces used for religious rites, but is reserved as a proper name for this one alone, which is the chief of them all. Marae is one of the most sacred spots in Tikopia, both from its general association with religious rites, and also, more particu- larly, because in it stand a number of rough stone slabs, each of which is representative of one of the chief gods of the island. According to native ideas the slab serves as his "seat" or more strict- ly, back-rest, during any kava ceremonies performed there. Please note here, that aside from holding up the roof, the most important function of the house post is as a back-rest. So that these stone slabs are in some sense functionally equivalent to posts. The layout of Marae is fairly complex when all details are taken into account, but the following characteristics stand out. The marae is divided into halves, each with a large low mound on which the kava bowls of the chiefs are prepared. In one corner are two stones representing the back-rests of the two brothers who are ancestors to the Kafika clan. These two brothers are known as Pu Ma. In the next corner, going counter- clockwise is the stone of the Atua i Taumako, deity of the Taumako clan; in the next, the stone of Atua i Niunano, a deity of the Tafua clan, and in the fourth corner, the temple of Rarofiroki, for the deity of the Fangarere clan, who for some reason does not have a stone back-rest, but uses the temple as a seat. (I must emphasize that there are other, important, and confusing deities represented on the marae that I am not considering.) During the rites of Marae each of the four clans on Tikopia "thad its basic station; this was in the rear of its own chief, his seating-mat 42 being near the seat of his principal god. . . . Thus the chief s and their clansmen were roughly at the four corners of a rectangle, each group facing towards the center of Marae (Firth 1940:211)." So once again we see a pattern of a rectangle split in half with posts (i.e. backrests) at each of the four corners. Note, too, that just as the posts in the house are ranked with regard to one another, so are the clans and, by implication, the corners of the maraeo There is no clear evi- dence that there is an important center position on the marae,.however, and I will comment upon this later. Given that each of the four clans is associated with a particular corner of the heavens, and each is associated with a particular corner of the marae, it is reasonable to wonder if this division into two pairs and a center does not have some significance for social organization. I think there is a good deal of evidence for this in the interaction of the clans in the Work of the Gods, which I am currently trying to analyze in structural terms, but I don't want to go into the matter in this paper beyond noting two points: 1. A grouping of the clans into interacting pairs' does appear in a number of cases: 2. In a number of cases the relationship between clans is phrased as one of affinality or, what is the same thing, as the relationship between mother's brother and sister's son (tuatina-tama tapu). Much clearer evidence for organization of ceremonial in terms of two pairs and a center is to be found in the life cycle ceremonies. As is shown in figures V to VIII, each of these ceremonies involves interaction of five groups of people in an affinal chain of wife-givers and wife-takers. More- over, with the single exception of the rituals at birth, it is clear that ego's lineage, C, stands alone while lineages in categories A and B, wife- taker lineages, are grouped together as cooks and are associated with the oven, tuaumu, side of the house, and at the same time lineages in categories D and E are linked together and are associated with the mata paito side of the house. For evidence of this last point, let me quote from We the Tikopia (1936:445). This is from the section on initiation. "After being fed the boys are carried on the backs of their tuatina. . . ' When they arrive at the house of the parents they are set down and the tuatina squat outside the house on the mata rito side.t I think it is clear that the same pattern of two pairs and a center, which occurs in the house and the universe, and, less clearly in the dis- tricts of the island and the layout of the marae, is also expressed in the organization of the life cycle ceremonies. III. Complex Alliance Now, al1 this reeks of an alliance system, and there is plenty of other evidence that something like an alliance system is going on, at least as a preferential system. Some of this other evidence is: 1. The constant affinal relationships between clans expressed in the Work of the Gods. 143 2. The fact that a mother's brother's son normally succeeds to the position of tuatina when his father dies (Firth 1936:215) and, even while h3T?~aher is still living, my refer to his father's sister's son as tama tapu (Firth 1936:269). This turns the termin- ology almost into an Omaha-system, and I am among those who believe that Crow and Omaha terminologies are indicative of alliance Sys- tems. 3. The fact that wife-givers are associated with mata paito while wife- takers are associated with the oven clearly indicates a difference of rank between wife-givers and wife-takers. Can it be that the relative ranking of clans in Tikopia (and of lineages within the village) is based on a conception of rank as established by connu- bium? I think there is clear evidence that this is the case. One of the problems for a society which is establishing rank by connu- bium is that connubium is circular, and there must be some point at which either (a) women enter and leave the system, or (b) a wife- taker outranks his wife-giver despite their relationship. There must be a "circle breaker" if connubium is to result in absolute rather than relative rank. In fact, listen to the first portion of this myth about the ancestral gods of the clans of Tikopia (Firth 1961:30): Now was born the deity Semoana, the eldest of the brethren. He was born, but born badly, while his brothers behind were still living in the belly of their mother. They urged him to hurry and descend. They called to hiim, "Lad! there! What's that you doing there?" Their mother called out "Wait, your brother is still going there; he has not come free." Thereupon Tafito glanced upwards, glanced upwards from the bottom of the womb of his mother, saw that her head showed the light, rushed up, and appeared on top of her. He was the youngest, but when he rushed up he became the oldest indeed! His brothers had not yet appeared. So he was termed the original. Tafito means the original. He is the ancestor god of the lowest ranking clan, Fangarere. It appears to me very likely that we have here a charter for the fact that even though Fangarere is wife-giver to Kafika, Fangarere is nevertheless outranked by Kafika. However, we now run into a difficulty. Most alliance theory assumes marriage with mother's brother's daughter as in figure IX. The Tikopia can not be practicing such a marriage system for two reasons: 1. Marriage with any first cousin is looked upon with disfavor (Firth 1936:221). 2. Marriage with mother's brother's daughter requires a model with only three lines, as shown in figure IX. We need a model which requires five lines. It is, therefore, interesting to note that, after marriage with mother's brother's daughter, the next simplest alliance system we can imag- ine is a five line alliance system of the type shown in figure X. Here, marriage is with mother's mother's brother's daughter's daughter, who is the same person as father's mother's brother's son's daughter. Such a model of course fits perfectly the pattern that we have found running through Tikopian culture, for on the five line model we do not have simply a chain of affinality, but a situation in which A and B are both wife-takers from C, while D and E are both wife-givers to C. We have precisely the pattern of two pairs and a center that we are looking for. There is some final confirmation that we are on the right track, when we return to the mythology. In the first place, there is one rather aberrant myth of the origins of mankind (Firth 1961:23) that goes, in part, as follows: They were singing to produce men. They gazed at first without results, but as they sang a man sat there. They sang that the sticks of wood should grow into men. Then as they counted ten, there sat there five men and filve women. That was the growing of people. In the second place, in discussing the versions of the myth of the emergence, Firth says (1961:33): Again, the versions differ in numbers of the Brethren mentioned. Pa Raropuka cites the rames of the four senior offspring, and also refers to the subsequent formation from placenta and umbilicus of two others [who, however, float away on the ocean]. Pa Tauraro mentions only Semoana and Taf ito at first, but later refers to the deities of Tafua and Taumako. Neither account mentions Oatuatafu, one of the generally recognized members of the Brethren, invoked in various kava rites. Both of these items indicate that five lines are necessary for the functioning of Tikopia social organization. Firth's last discussion gives a new idea of what is going on. Can it be that what is usually expressed as four lines as in the four brethren, is in fact five lines, with the fifth line submerged or hidden? If so, this might explain why there are only four clans on Tikopia, rather than the expectable five. Let me go back for a minute to the layout of Marae. You will remem- ber that there was no central representation of a deity with an associated clan. But note that in fact there are five deities represented in the corners of the marae: the individual gods who are ancestors of the other clans and the two brothers Pu Ma, who are jointly ancestors of the Kafika clan, the highest ranking clan. Let me sunmarize the myth which deals with these two brothers: they first dwelt in Tikopia as gods. Then they fought with the Brethren, the ancestors of the other clans. Finally, they left in anger and went to the land of the gods, but then they came back as men in disguise. Here is how Firth (1961:56) quotes the relevant section of the myth: Thereupon they went to the land of the gods, but appeared in the ocean, appeared with their canoe into which they jumped and came here again. On they came over the ocean. Then the two brothers, Pu Ma, spoke to- gethero One said to the other, When we go, do not let us go as brothers; they will know it is we two who have returned. 'When we go we two shall be mother's brother and sister's son." This is, of course, a charter for clan endogamy. In fact, what is called one clan has been split into two halves, of which one is wife-taker to the other. Thus it is that the need for a fifth line is filled by a charter for incest in the highest ranking group. IV. Surmmary To sum this all up, the following pattern runs through Tikopian culture . 1. A rectangle is divided into four quarters and a center. 2. The four corners are ranked. 3. The four quarters are grouped into two pairs, one of which is higher ranking, to the right, male, and "away from the oven." 4. Congruent with this pattern is a five line compLex alliance system, with ranking of lines. 5. The five line complex alliance system is identified with a system of four ranked clans. The discrepancy between five lines and four clans is resolved by a mythological charter for incest in the highest ranking clan. NOTE 1This paper was presented at the Eighth Annual Meeting of the Kroeber Anthropological Society, held at the Claremont Hotel, Berkeley, California on April 25, 1964. I am using the term "complex alliance" to refer to a prescriptive assymetric marriage system which is more compli- cated than that involved in simple prescriptive marriage with mother's brother's daughter. BIELIOGRAPHY Firth, Raymond W. 1936 We, the Tikopia. London, George Allen and Unwin, Ltd. 1939 Primitive Polynesian economy. London, George Routledge and Sons. 1940 The work of the gods in Tikopia. London, London School of Economics Monographs in Social Science nos. 1 and 2. 1955 The fate of the soul: an interpretation of some primitive con- cepts. Cambridge, at the University Press. Reprinted as "The fate of the soul"f in Charles Leslie (ed.), Anthropology of folk religion. New York, Vintage Books, 1960. Page references are to the latter. 1956 Ceremonies for children and social frequency in Tikopia. Oceania 27:12-55. 1961 History and traditions of Tikopia. Wellington, The Polynesian Society. 46 Figure I (after Firth 1936) TIKOPIA HOUSE PLAN Figure II (after Firth 1955) RAN OF THE TIKOPIA UNIVERSE Figure III (after Firth 1936) DISTRICTS AND SUBDISTRICTS OF TIKOPIJA * MATA PAITO A ROTO - - - - A _ _ _ _ _ _ _ PAITO * TUAUMU 0 a RAKI a TIKOPIA - - - - - - 0 TONGA 0 - - | - Kva Mound 47 ava Mound . 0 Fangarere Tafua Fi gure IV (after THE MARAE AT UTA Firth 1940) A B C Wife- Wife- BabyI Takers Takers Patri from B from C linea FARE MASIKITANGA, (FaFaSiDa)1 (FaSi) "COOKS" (FaFaFa~r- (FFSH)"Ego's |DaDaHu-) (FaFaSiHu) | G|oup' Soko Fai SokoFai Tuatina Matua Figure V (after Firth 1956) CLASSES oF PARTICIPANTS IN BIRTH RITUAL s ge D Wife- Giver s to C E Wife- Givers to D Notes: 1. This kin type is not mentioned in the text, but is shown on the chart, figure 1, page 27. If correct, it runs counter to the hypothesis being developed here. 2. Father's mother's group is said (p. 17) to participate in payments as member of this group. If correct, it runs tounter to the hypothesis being developed here. 3. This kin type is not mentioned in the text, but is shown on the chart, figure 1, page 27. Kaf ika Taumako 48 A B COOKS 'Soko Fai "Soko Fai Tat ina" Matua" Husbands of Husbands of Tama Tapu Daughters of the Fami ly C AI MATUA| |"Fathers"| D TUATINA (MoBr) I Figure VI (after Firth 1936) CLASSES OF PARTICIPANTS IN INITIATION RITUAL A B COOKS "Soko Fai "Soko Fa'i TuatinalV Matual" Husbands of Husbands of Tama Tapu Daughters C Figure VII CLASSES OF (aftber Firth 1936) PARTICIPANTS IN MARRIAGE RITUAL (1) COOKS (2) Deceased's Sister's- Husbands Deceased' s Daughters' Husbands Tama Tapu Mourning Group Figure VIII (after Firth 1939) CLASSES OF PARTICIPANTS IN FUNERAL RITUAL Notes: see p. 49. E D E Recipients of Valuables Bride I sBride'Is Fathers Mother's Brothers I f I Notes to Figure VIII: 1 There seems no reason to doubt that any of the deceased's father's sister's husbands who survived would also fall into the cook class. 2. Husbands of tama tapu (Fai Soko Tuatina) are not mentioned in the dis- cussion of funeral ritual, but are assumed, since they appear always to be included in the category of "cook." Definition of symbols used in Figures IX and X: G Generation a (Patri) line of descent < - a Siblings (sisters) given as spouses > a Spouse (bride) price G1 A B C G3 | __ |-,-? _ I ~ ~ ~ ~~~~~~~~~~~~~~~~~~~~~~~~~ I I.-. Figre IX THREE LINE ALL IANCE SYSTEM A G1 G2 G3< _ --> *; .G3 l B Figure X FIVE LINE ALLIANCE SYSTEM D E