LORD FIELAKEPA MARRIED MALIA I LUTU OF CANTERBURY CITY SYDNEY AUSTRALIA

- 7

The Polynesian kingdom of Tonga in the South Pacific, as many writers have pointed out, is a highly stratified society. In this paper I discuss the interplay of rank and power in the traditional political and social system and their partial transformation, under modern political and economic conditions, into a system of socio-economic classes similar to that of Western society.

The Tongan system of stratification has been often described but little understood, even in the last 20 years when an increasing number of anthropologists and historians have added their descriptions and analyses to the remarkable contemporary accounts of Cook (1961, 1967) and Mariner (Martin 1818) and the basic ethnography of Gifford (1929). My original intention before going to Tonga was to make a conventional anthropological village study, because Gifford's Tongan Society (1929), although an excellent ethnography of its kind, did not really explain how the social and political system worked, and I found the Beagleholes' (1941) study of a village in Vava'u sketchy and unsatisfying. My plans were soon altered, however, because an unusual opportunity arose, thanks to the efforts of James Spillius, for me to work with the late Queen Sālote Tupou. I am enormously indebted to the late Queen's patient explanations of genealogies, political alliances, and personal rank which made it possible for me to develop my own understanding of the traditional system. Many of the data of the present paper are taken from these discussions, although I am responsible for the interpretation of them, and the paper is not in any sense an official view.

The ideas of the present paper were developed not only through working with the Queen but also through observing various aspects of modern Tongan society at both the domestic and the political level, for much of what the Queen was explaining about the traditional system was also happening in a somewhat different form in everyday Tongan life. 1

My own field work with other Tongans was somewhat restricted by the fact that I was working with the Queen. Whatever assurances I tried to give about confidentiality were unlikely to be believed; people regarded the Queen as the great authority on Tongan custom and were afraid they - 8 themselves were doing things wrongly. Fortunately, I have been able to use the many papers and theses that have appeared about Tonga since 1960, many of them generously supplied by the authors.

Stratification is particularly complex in Tonga because the several forms of social differentiation overlap but do not coincide and are often contradictory. I have found it helpful to think of the Tongan social system as expressing a series of interacting principles which are not mutually consistent. Each major principle or form of stratification has a context, in some cases a ceremonial context, in which it is expressed in relatively pure form—traditional political authority in kava ceremonies, rank among kin at funerals, modern Government status at Government feasts—but there are other contexts in which the principles are in conflict so that there is room for manoeuvre and for value in one system to be converted into value in another.

The Tongan kinship system was (and is) simultaneously domestic and political. 2. It contains the basic and simplest forms of stratification, a system of patrilineal authority and a system of ceremonial rank. Even within the domain of domestic kinship these two principles do not coincide. Fathers and elder brothers have authority and control over one's access to land. The basic principles of rank, on the other hand, are that sisters have higher rank than brothers and that older siblings of the same sex have higher rank than younger siblings of the same sex. Sisters have no authority over brothers even though they have the higher rank. Both fathers and sisters are respected but the basis of the respect is different. Fathers have the right to command, to give or to withhold; sisters have the right to ask to be given what their brothers produce (Biersack 1974).

The system of rank among kinsmen was elaborated into a system of rank recognised at society-wide level. In this system no two people had the same rank so that the resulting structure was, of course, radically different from a class structure, assuming one follows the usual definition of social classes as society-wide groups stratified in terms of power and/or prestige. Normally one expects members of the same elementary family to belong to the same social class.

The system of patrilineal authority, combined with the practice of territorial sovereignty, was elaborated in the traditional system into the basis of corporate group formation and political organisation in terms of title and ha'a. The enduring political-jural entities were territorial groups of subjects ruled by political leaders who usually, though not invariably, held hereditary titles. By a figurative extension of kinship terminology these territorial groups were called the leader's “kin” (kāinga); most of - 9 the people who lived on a leader's territory (fonua) were in fact his kin, but, kin or not, they were his subjects. In Tongan idiom they were his subjects and therefore his kin rather than being his kin and therefore his subjects.

A hereditary title gave a man authority, that is, the socially recognised right to rule, as distinct from the de facto capacity to do so, for which I shall use the term “power”. Traditionally the main basis of political power was the ability to build up and lead a local group. It involved getting access to land, usually through the permission of a title-holder from whom one was descended, though this was easier than it is today because there was no land shortage in most areas. It involved hard work to grow food which was used to help one's kin and thereby to put them under an obligation to oneself. By this means the aspiring leader built up his kāinga; he became what Sahlins calls a “big man”; he attracted the attention of an established chief and, in one way or another, eventually he got drawn into the system of chieftainship. Sahlins thinks of the “big men” of Melanesia as a contrast to the established chiefs of Polynesia. The Tongan case suggests that both established chiefs and “big men” were integral to the system. 3

The external relations of territorial groups were regulated by the membership of their leaders in two systems. First, the leaders belonged to a system of titles. The titles were grouped into social entities called ha'a according to their genealogical derivation and seniority; this system was supposed to be stable and unchanging. Second, the position of territorial groups was defined by networks of kinship ties created by the leaders' marriages and the marriages of their parents. Considerations of rank and power as well as formal title entered into choice of spouse and the resulting political alliances. Succession, similarly, was affected by considerations of rank and power as well as by the technically “correct” rule of succession to a brother or son of the deceased leader. The network of ties between corporate groups created by marriage was somewhat at variance with the relations established by titles and formal political authority, and thus allowed some rapprochement between the current facts of political power and the ideology of unchanging titles and stable political authority.

Taken together these two systems, authority and rank, combined to provide both for continuity and for change. Viewed in the short term, the combinations were unpredictable, labile, and complex; viewed in the long term, certain patterns were discernible. In this paper I shall expound this briefly stated thesis in greater detail, outlining the expression of authority and rank in the domestic sphere, then the elaboration of both systems in the political structure and their articulation with each other. I - 10 focus primarily on the traditional system but at the end of the paper I discuss modern trends towards the development of social classes. 4

Before proceeding, however, I should like to define the way I am using English and Tongan terms concerning power, authority, and rank, for even at this comparatively simple level the complex and apparently contradictory nature of Tongan conceptions of status makes exact translation impossible.

By “rank” I mean a quality commanding respect and deference, and inherited from one's parents; it cannot be altered either by one's own achievements or by one's failures. By “power” I mean the de facto capacity to lead a group and direct its activities. By “authority” I mean legitimate, institutionalised power. “Power” is the wider term but both “authority” and “power” concern capacity to direct and order the activities of other people.

I use the word “rank” to describe the Tongan 'eiki-tu'a (chief-commoner) distinction. Tongans have no generic term for the concept of rank. Nowadays the word toto (‘blood’) is often so used, but this usage is probably comparatively recent and taken over from English. The word “rank” is not ideal for the 'eiki-tu'a distinction but it is preferable to “status” or “prestige”, which can apply to many attributes other than birth and kinship.

There is no word to translate 'eiki adequately (let alone the 'eiki-tu'a relationship) because 'eiki is confusingly similar to but different from the English word “chief”. In English the word has a strong connotation of ruling or leading, whereas in Tongan it has an even stronger connotation of rank, that is, of being an aristocrat by blood, of having by inheritance a large quantity of sacred stuff in one's person through being descended from a king and preferably also from the sister of a king, particularly the sister of a Tu'i Tonga, the former sacred king. In Tongan the term 'eiki also has the connotation of leading or ruling, but this is less important than its connotation of rank. According to the late Queen Sālote, the term 'eiki originally meant someone of aristocratic descent; it did not mean a ruler or a title-holder, but, since the Constitution in 1875, the collective term hou'eiki has come to be used for all the title-holders who were created nōpele at the time of the Constitution, even though over half of them were not aristocratic by blood. Hence one is often told: “So-and-so is a noble (nōpele) but not a chief ('eiki)”. Since the system of rank and the system of political authority do not coincide, it is not only possible to hold a noble title without being an aristocrat; it is also possible to be an aristocrat without holding a title or being a political - 11 leader. (Such people used to be very common; they are less frequently met with today.) Such a person is called sino'i'eiki, literally “chiefly in body”. If such an aristocrat also holds a title, he is called 'eiki fakanofo, literally “appointed chief”.

The term for title is hingoa fakanofo, literally ‘appointed name’. There are two main types, hingoa 'eiki, who held estates, ruled subjects and thus had political authority; and hingoa matāpule, who are the ceremonial attendants of chiefs and did not normally have estates and rule subjects (although six matāpule were granted hereditary estates by King Tā'ufa'āhau at the time of the Constitution). Having a chiefly title (hingoa 'eiki) did not and still does not necessarily mean that one had high standing in the system of rank, but the fact that the term was used at all in this context indicates its dual meaning. It connotes authority as well as rank. Title-holders who ruled subjects but were not aristocratic were called matu'a tauhi fonua, literally “old men who look after the land”, and, since rank is a relative matter, all title-holders describe themselves as matu'a in the presence of the monarch. Conversely, within the confines of their own village, people may think of their title-holder as 'eiki and when speaking to him may use the “chiefly language”—special terms of respect which are used to people of high rank—even if he is not an aristocrat by blood, but they will not use this chiefly language to their title-holder when outsiders of higher rank are present.

There are terms in Tongan, ivilahi and ivimālohi, which mean a powerful man, though they are not necessarily specific to a political context. There is thus no exact term in Tongan to describe a man who held political power and ruled subjects regardless of whether he had a title (hingoa 'eiki) or was an aristocrat (sino'i 'eiki).

In this paper I shall follow the existing ambiguous custom of using the English word “chief” so long as it is clear from the context what is meant. When it is necessary to be more exact, I shall use the term “leader” to mean a person who held an estate and ruled subjects, whether or not he held a chiefly title and whether or not he was an aristocrat. I shall use the term “title-holder” for someone who held a chiefly title and ruled subjects, regardless of whether he was politically powerful or aristocratic, and I shall use the term “aristocrat” for sino'i 'eiki, and “aristocratic title-holder” for 'eiki fakanofo.

Tonga is an independent Polynesian kingdom in the south-west Pacific consisting of approximately 150 islands, over half of which are uninhabited and many of which are very small. The population was 56,838 in 1956, 77,439 in 1966, and 90,409 in 1976. It is estimated that - 12 during the 18th and 19th centuries the population was about 20,000 and that, except in certain areas, such as the western part of Tongatapu, the main island, there was no shortage of land. Over half the population now live on the main island. In the 18th and 19th centuries, Tonga thus did not have an economy of scarcity, an environmental factor which allowed a considerable free play in social and political manoeuvre. “People, not land, were the resources chiefs needed”, as the Queen explained.

Archaeological evidence shows that Tonga was settled from the general direction of Fiji and Melanesia about 1200 B.C. and that there has been no major immigration since that time. According to legend and oral history, the islands have always had a centralised political structure headed by a paramount chief or king, the Tu'i Tonga. 5 The details of how the political system worked when there was only one king are no longer known. There were a number of other chiefly titles besides that of the Tu'i Tonga, some of which were designated as “brother” titles to the title of the king. There were also a number of matāpule titles, these being ceremonial attendants of the king and other title-holders. The original matāpule are said to have been foreigners, Samoans or Fijians, social outsiders who were not subject to the social tapu and restrictions that affect Tongans. The matāpule in Tonga, unlike their Samoan counterparts, have never held political power and do not act as formal political councillors to their chiefs.

In spite of an ideology of support and loyalty among brothers, there are many legends of murders of a succession of Tu'i Tonga by the holders of the “brother” titles. It is said that in order to simplify the task of ruling his turbulent kingdom, one of the Tu'i Tonga decided to become a senior, sacred chief, leaving the actual work of secular government to his younger brother, whom he designated as a new, second king called the Tu'i Ha'atakalaua. It was the task of the Tu'i Ha'atakalaua to rule the kingdom on behalf of the Tu'i Tonga and to see that subjects sent certain crucial tribute that expressed political and religious submission to the Tu'i Tonga; a first fruits ceremony called the 'inasi was especially important in expressing national unity and dependence on the Tu'i Tonga as an intermediary with the gods. There were also many others sorts of occasion on which food and movable property (called koloa) in the form of mats and barkcloth were sent to superior chiefs and through them to the Tu'i Tonga.

According to Queen Sālote, the Tu'i Ha'atakalaua sent his own younger brothers and sons and other relatives to outlying parts of the kingdom to get themselves established as leaders with the local people, usually by marrying daughters of local chiefs. If successful, these younger brothers and sons were granted titles by the Tu'i Ha'atakalaua. - 13 When the time came for sending 'inasi and other tribute to the Tu'i Tonga, the younger brothers and sons of the Tu'i Ha'atakalaua would send their tribute first to the Tu'i Ha'atakalaua, who would then send it on to the Tu'i Tonga.

After several generations, the Tu'i Ha'atakalaua is said to have repeated the process adopted by the Tu'i Tonga. He appointed one of his younger sons, called Ngata, as secular king for the whole kingdom. The oldest son succeeded to the title of Tu'i Ha'atakalaua and was of higher rank because of his seniority. This “younger brother” king, the Tu'i Kanokupolu, is now the sole monarch. The first Tu'i Kanokupolu established himself in Hihifo, the western part of the main island of Tongatapu, by marrying the daughters of the local chief, 'Ahome'e. During the reign of the second Tu'i Kanokupolu, his younger brothers were given titles which were collectively called Ha'a Ngata, named after their father. Several of the sons of the third Tu'i Kanokupolu gave rise to another group of titles known as the Ha'a Havea. The Ha'a Ngata established themselves in Hihifo by marrying the daughters of local chiefs and absorbing their authority and their titles, which became defunct. The Ha'a Havea title-holders expanded into central Tongatapu by the same method. In subsequent generations, younger sons of various Tu'i Kanokupolu went to the northern islands of Ha'apai, Vava'u and Niuafo'ou and established themselves there, eventually being granted titles. The title-holders derived from the Tu'i Kanokupolu sent their ritual tribute and other forms of food and koloa (mats and barkcloth) first to the Tu'i Kanokupolu, who forwarded it to the Tu'i Ha'atakalaua, who then sent it to the Tu'i Tonga. Later, when certain chiefs became sole rulers of the northern islands, the ritual and secular tribute was sent by island group rather than by each chief separately to his own king.

This three-king system, each with its own derivative titles grouped into ha'a, is usually thought of as the classical period of Tongan history (Bott, in press). The classical period is supposed to have been peaceful and orderly, largely because Captain Cook was impressed by the people's obedience and deference and by the elegance of Tongan culture and the openness with which he was welcomed. But the legends of the Tu'i Tonga being assassinated and the early missionary accounts (Thomas n.d., also Thomas ca. 1853) of ferocious local wars in previous generations suggest that Cook's view was idealised. Conflicts and minor wars were probably endemic in the classical system. Cook saw the last efflorescence of the traditional system before the storm broke at the end of the 18th century. At that time the various branches of the Tu'i Kanokupolu's line began fighting among themselves. The title of Tu'i Ha'atakalaua became defunct. There were irregularities in the appointment and residence of - 14 the Teu'i Tonga. Th 'inasi ceremony—all-important as a ritual expression of political and religious unity—was allowed to lapse. The son of a Tu'i Kanokupolu fought a war against the Tu'i Tonga in the 1820s and then virtually imprisoned the Tu'i Tonga for 40 years. The last Tu'i Tonga died in 1865, the title became defunct, and there is now only one king, the Tu'i Kanokupolu.

All this was happening when European ships, sailors, and missionaries were arriving with trade goods, guns, and religion, though whether these environmental changes were the main cause of this particular phase of upheaval it is impossible to tell.

In spite of this turmoil, Tonga was remarkably fortunate and successful in managing its contact with European powers, partly because the European powers were not avid to gain political and economic control of the kingdom, partly because most of the missionaries wished Tonga to remain independent, and partly because Tonga was fortunate to have a political genius as the Tu'i Kanokupolu during the critical period of contact with missionaries, traders, and European governments. This was the great King George Tāufa'āhau Tupou I, great-great-grandfather of Queen Sālote. In fact and later in name, he wielded effective power over the kingdom from the 1830s until his death in 1893. He was one of the first Tongans to adopt the Wesleyan faith and he used his Christian affiliation to good advantage against his political enemies. In his later years he used a disaffected missionary, Shirley Baker, as political adviser (Rutherford 1971, 1977). By developing a quasi-European code of laws, passing a European type of constitution, encouraging education and Christianity, and developing a system of land tenure that prevented alienation of land, King Tāufa'āhau and his missionary adviser held off European occupation and laid the foundation of what Marcus (1975a) calls the “compromise culture”, a unique combination of Tongan and European custom. King George outlived all potential rivals to his title and power and was duly succeeded in 1894 by his great-grandson, Tupou II.

The reign of Tupou II was troubled by financial complications and Tonga was made a protected state by Britain in 1901 but this last vestige of European control was removed in 1970. Queen Sālote succeeded her father in 1918, her reign being marked by an imaginative integration of Tongan custom and European innovation. She was succeeded in 1965 by her son Tāufa'āhau Tupou IV who is grappling with the challenge of helping Tonga to adapt to land shortage and a developing economy of scarcity.

- 15

I do not want to describe the Tongan kinship system in detail for it has now been described by many authors. 6 All I shall attempt to do is give enough detail to expound my theme that the principles of authority and rank are rooted in the sphere of domestic kinship.

The Tongan kinship system does not lend itself well to traditional forms of anthropological classification but if one must fit it into a typology I think it does least violence to kinship usage to describe it as a cognatic system in which the bilateral kindred, called kāinga or nowadays often called fāmili, is the basic kinship category. But, as Garth Rogers points out, kāinga is an ideology as well as a definition of a category of kin. People are defined as kāinga only if they mutually recognise the kinship relationship, and non-related people who act towards each other as if they were kin are spoken of as kāinga (Rogers 1975: 247). Marcus (1975b) says that formerly the kāinga was the minimal segmentary lineage in Tongan social structure, but that today the lineage has “dissolved” so that the term kāinga is now applied to the bilateral kindred. He presents no evidence for this statement, however, and there is nothing in contemporary sources or oral history that supports this definition of kāinga or such a change in usage.

Normally one is not supposed to marry a kinsman but the boundary between kāinga and non-kāinga has always been elastic. In the traditional system kings and certain high-ranking aristocrats often practised matrilateral cross-cousin marriage (called kitetama), a form of political alliance which is further discussed below. Polygyny was commonly practised in the traditional system, especially by aristocrats. Divorce was easy and frequent and it was common for women of high rank to have a succession of husbands.

In the traditional system, land was controlled by titles and titles were normally inherited either by a brother or a son, though it was not uncommon for titles to go to a daughter's son if the political circumstances were considered to warrant it. Residence is a matter of choice and convenience though it is usually patrilocal. At the present time there are no descent groups with control of land or other means of production. Whether such descent groups were explicitly recognised in the traditional system is doubtful. I will return to a discussion of this issue at a later point.

The relation between father and son is the basic model of authority-conformity. The father has both authority (pule) and high rank. He commands respect (faka'apa'apa); he is 'eiki (“chiefly”, “of high rank”) whereas his son is tu'a (“common”, “low”). Rights over children, titles, land, houses, and, in the traditional system, political authority were vested in the father and his brothers. 7 Between fathers and their adult - 16 sons there is virtually an avoidance relationship. Relations between fathers and daughters are less tense and less marked by avoidance but daughters are on the low side (liongi) at their father's funeral (Rogers 1975, 1977).

In one's own generation brothers are supposed to be friendly and supportive. At the present time it is not unusual to find a group of brothers, usually actual not classificatory brothers, who share various economic resources and form what Marcus calls a “family estate”—one brother and his household contributing produce from a garden plot, another sending cash home from work overseas, etc. (Marcus 1975b). Nayacakalou (1959) and Aoyagi (1966) found similar groups or segments of them in the villages of Ha'ateiho and Nukuleka. James Spillius and I found them in the villages of Houma, 'Ahau and elsewhere, and Korn (1974) found them in the village she studied. These groups are usually called fāmili, the word having the same immediate and extended use it has in English. Korn defines kāinga as the kindred, fāmili as a village based action group, a loose association of households linked by consanguineal ties, whose members co-operate economically without being asked; membership, she stresses, is completely “optative”. She also notes that her definition is inferred from observation of behaviour and that Tongans would not define fāmili in this way if asked. Aoyagi uses the term fāmili for the set of people that Korn defines as kāinga, and the term “fāmili organisation” for the set of people defined by Korn's term fāmili. This gives some idea of the problems of analysing kinship in Tonga, and of how easy it is to suggest, almost without intending to, that Tongans themselves have a conceptual category for the group or set one infers from their behaviour.

I would agree with Rogers (1975) that if an analysis is to approach completeness it should include indigenous concepts, observation of behaviour, and analysis of the relationship between them. Analysis is especially difficult in the realm of kin group definition, because it is likely that membership in small-scale local kin groups has become much more “optative” than it used to be. The use of the English word fāmili indicates that this is a new usage. In the traditional system one belonged to the local group of one's chief and to a subdivision within this local group. One gets the impression that membership was not absolutely fixed but that one could not choose one's group affiliation as freely as one can now. It is also likely that relationships between brothers have become less openly hostile in the modern system than they used to be, especially among chiefs, because in the modern system inheritance of both titles and, in the case of commoners, land is now fixed so rigidly by the Constitution that open rivalry and competitiveness have little point.

- 17

In the traditional system there was a strong ideology that brothers should live and work together harmoniously and that younger brothers should support older brothers. In fact, sons of the same father and patrilateral parallel cousins were often in competition for political power. This rivalry was particularly strong when brothers of the same father had different mothers, which often happened in the traditional system because aristocrats were polygynous and the divorce rate was high. As Rogers notes, brothers who had the same father but different mothers were called uho tau, literally “fighting brothers” (uho meaning “umbilical cord” and tau meaning “fight”), whereas brothers who had the same mother but different fathers were called uho taha, “of one cord”, and were unlikely to compete with each other directly for titles and political power (Rogers 1977). Thus, patrilineal relatives—real and classificatory fathers and brothers—were the focus of authority, rivalry, and competition.

The basic principle of rank is that sisters have higher rank than brothers. Between siblings of the same sex older siblings had higher rank than younger siblings. Sex is the more important principle because younger sisters have higher rank than older brothers. Brothers must respect sisters (faka'apa'apa). The brother gives his sister gifts of food and generally helps her and her children. Sisters are 'eiki (“high”) whereas brothers are tu'a (“low” or “common”). There is an avoidance relationship between them. In this respect a sister's relation to her brother is similar to that of a father to his child; both sister and father are respected. Both are avoided; both are 'eiki (“chiefly”), but the two relationships are also very different because the sister has no authority, no secular power, no pule, over her brothers. In effect, the eldest brother, if the father is dead, has pule over the whole family, including his sister. But she has higher rank. In circumstances of such contradiction, it is hardly surprising that brother and sister maintain a strict avoidance relationship.

Pule is the prerogative of real and classificatory fathers and brothers; higher rank is the prerogative of both fathers and sisters. But, as Aletta Biersack (1974) shrewdly points out, the higher rank of real and classificatory fathers has a different basis from the higher rank of sisters. In the traditional system the higher rank of fathers took the form of their having potential resources to bestow, through access to land and sometimes to title, whereas the higher rank of sisters took the form of their having the right to ask to be given to, rather than the right to bestow. Sisters and their children could ask to be supported but they could not command it.

The rank of the father's sister (mehekitanga) is even greater than that - 18 of the sister. The father's sister has ritual mystical power over her brother's children. Her curse can make a niece barren. Typically she names her brother's children. She used to choose their spouses. She is an honoured guest at their funerals. In the traditional system, if there were no immediately obvious heirs to a title, her wishes were taken into consideration in choosing an heir.

These usages are extended to the children of the people concerned. In other words both men and women respect not only their father's sister but also her children (tama 'a mehekitanga). A man respects not only his sister but also his sister's children (his 'ilamutu). A woman is superior not only to her brother, but also to his children (her fakafotu). Conversely, both men and women are superior to their mother's brother (tu'asina or fa'ē tangata, “male mother”) and his children (tama 'a tu'asina).

Funerals are the ritual occasions when one sees the clearest expression of 'eiki-tu'a relationships (Rogers 1975, Kaeppler 1971). Relatives who are higher in rank than the dead person sit in the house, are fed and receive koloa. One of them is appointed as the fahu of the ceremony, meaning the one of highest rank. Relatives who rank lower than the corpse bring food, stay outside and work in the kitchen; they are called liongi. Older siblings of the same sex and their children are neither liongifahu. Grandparents and grandchildren similarly are neither liongi nor fahu. nor

In strict Tongan usage, the fahu is a specially designated position of honour at funerals, weddings and first-year birthdays. There is some disagreement among Tongans over which relative it should be, a father's sister, a father's sister's child, or a sister's child (Rogers 1975). Fahu is also loosely used as a kinship term by both Tongans and anthropologists, usually to mean “sister's child” ('ilamutu) (Aoyagi 1966, Kaeppler 1971).

A woman's children are supported and helped not only by her brothers but also by her father, for, as Radcliffe-Brown pointed out many years ago (1927) when describing Tongan kinship, “. . . there is a tendency to extend to all the members of a group a certain type of behaviour which has its origin in a relationship to one particular member of the group.”

Garth Rogers suggests that there are in the Tongan kinship system implicit matrilineages whose members hold mystical powers over their respective patrilines. He expresses it as follows (1975: 272-3):

- 19

He supports his assertion by an insightful analysis of the fahu and liongi relationships and of who is chosen to be a title-holder's kai fono at royal kava ceremonies. (The kai fono is a relative of higher kinship rank than the title-holder who takes away the offering of food that is made to the title-holder in the course of the ceremony.)

The father's side (kāinga'i tamai) has both authority (pule) and high rank ('eiki); the mother's side (kāinga'i fa'ē) has neither. The mother's side consists of a set of brothers and brothers' children to whom she and her children are superior. This principle was of great importance in the traditional political system. It meant that a leader could always expect support from the groups of his mother and his wives, also from the group of his mother's mother. It meant he had to give support to the groups into which his sister, his sister's daughter, and his father's sister married, always assuming that they had children.

The Tongan system of kinship and marriage was thus typically asymmetrical. There was no regular expectation that the group a sister married into would repay the gifts of the woman herself or of the movable property, food, and general support that constantly flowed in her direction. Rather it was up to her brother's group to find wives for themselves so that food and support would flow into their group to right the balance. Occasionally there was a direct reciprocal exchange, that is, Group A gave a woman in marriage to the leader of Group B and the leader of Group B gave a woman in marriage to the leader of Group A. But such direct reciprocity was rare. It was more usual for Group A to give a woman in marriage to Group B and to seek wives from some other group, let us say Group C.

A woman was (and is) subject to the authority of her husband. The regard in which she was held by her husband and his group, however, depended not only on her personal qualities but also on her inherited rank, her “chiefliness”. In the traditional system it also depended on the political strength of her brothers; it would have been unwise to offend a woman on whose brothers one was dependent for support.

Up to this point I have spoken of personal rank only as an attribute of familial sex and age distinctions. Personal rank is also inherited from both parents. It is inherent in one's being, fixed for life, cannot be changed either by good works or misdeeds. Although so fixed and absolute, however, its expression varies according to context and according to the rank of the other people present. A person may be fahu at one funeral and liongi at another funeral the next day; it does not of course mean that his intrinsic personal rank has changed, only that he was differently related to the two deceased persons. During our stay in Tonga people deferred to the noble Vaea because he was sino'i 'eiki, an - 20 aristocrat. They used the chiefly language when addressing him. But when the heir apparent, who then held the title of Tungī, was also present, people used the chiefly language for Tungī and ordinary words for Vaea. Once again it did not of course mean that Vaea's intrinsic rank had changed, only that his “chiefliness” was less than than of Tungī.

The rank one inherits from one's parents depends on their descent from the sacred king (the Tu'i Tonga), his sister (the Tu'i Tonga Fefine), and his sister's child (the Tamahā). I shall postpone discussion of this aspect of personal rank until discussing the extension of familial concepts of rank to the society as a whole.

In the traditional system, political authority was vested in titles. The title conferred access to land and the right to rule a local group of subjects, called the kāinga of the title. This political kāinga, as I shall call it, was the basic corporate group of the traditional society. The quality of the title-holder's authority was similar to that of the father, but the concept of territorial jurisdiction was added. A clear distinction was (and is) made between the kāinga of the title, which was a political group—nowadays it is a village—and the kāinga of the particular man who held the title, which was his personal network of recognised kin.

There is controversy in the literature over whether land and title were the corporate estate of a patrilineage or whether land and political authority accrued to the title rather than to the kin group of which the title was the head and focus (Gifford 1929, Lātūkefu 1974, Maude 1971). I hold the latter view, namely, that territory and subjects accrued to the title. Firstly, Tongans had no agreed conception or name, either generic or proper, for a patrilineage. Secondly, in all discourse and action, it is the title or leader and its territorial prerogative of rule that are emphasised, not the kin group which appointed the title-holder. There is no clearly formulated conception of a patrilineage as the group that chose the incumbent of a title in the traditional system. Of course, it is very difficult to know how title-holders and leaders were chosen because it is now more than 100 years since a major title-holder was chosen according to the rules of the traditional system, instead of by the fixed rules of succession adopted in the Constitution of 1875. From what one can gather from specialists in Tongan history, from Mariner, and from other sources, the new incumbent was usually chosen by the brothers and sons of the former title-holder, but the choosing group was not strictly defined. The father's sister of the former title-holder was often con- - 21 sulted. It is uncertain whether the brother's children of the former title-holder were involved. It is relatively certain that the political groups of the mothers of the various candidates had no direct say in the selection, even though the strength and rank of the mother's people were important factors influencing the choice. Typically, succession went to a brother or a son of the former title-holder, though succession to a daughter's son was not uncommon, particularly when considerations of rank and political power were involved. Once the kin group had made their choice, the actual appointment was made by a senior title-holder in the same ha'a or by the king from whose line the title was derived.

Whether Tonga should be described as having a system of patrilineages raises a general issue of anthropological theory. Should one speak of a society having patrilineages when they are not recognised by the people themselves, have no name, generic or proper, but are implicit in kinship terminology and behaviour, and may, in the past, actually have been the framework, however loose, of the selection of leaders? Should one follow the conceptions of the anthropologists or the Tongans? I think one must do both. It would do violence to Tongan conceptions to speak of their having patrilineages in the conventional sense. But it would also be misleading to overlook the importance of patrilineal authority and the implicit presence of patrilineages.

The Tongan focus on the title and the chief rather than the patrilineage can be viewed as a displacement. One way of regarding this focus on the title is that it marks a step in the transformation from a lineage-based to a territory-based society.

As far as the patrilineage issue is concerned, then, I agree with Garth Rogers who states (1975) that patrilineages exist implicitly but are “hypothetical”, not overtly realised; he suggests one speak of “patrilines” rather than patrilineages.

Although the typical expectation was that the ruler of a political kāinga would hold a title (hingoa fakanofo), there were several instances of politically powerful men who ruled political kāinga but were not formally appointed to a title. Such leaders were usually the younger sons of a king or another important title-holder and were in the process either of building up a new local group in support of their senior relative, or striking off on their own to build up a rival group. Eventually, if the group maintained and increased its power, its leader was likely to be granted a title. Many of the men in Mariner's account of Tonga in the early 19th century were leaders of this sort (Martin/Mariner 1818). Fīnau 'Ulukālala, though he was the hau or secular ruler of Vava'u and Ha'apai, did not hold a formally appointed title until after the time of the Constitution. Similarly, the descendants of the last Tu'i Ha'atakalaua— - 22 Mulikiha'amea, Fatukimotulalo, Tungī Halatuituia, Tuku'aho, and Tungī Mailefihi—formed an unbroken line of very important and powerful leaders at the village of Tatakamotonga on the main island of Tongatapu from the late 18th century, but did not hold a formal title until the title of “Tungī” was created for the line in the late 19th century.

A title-holder or a leader ruled all the people in his political kāinga whether they were his kin or not. If he held a title, the leader also maintained links with descendants of former holders of his title who were living elsewhere and he might call on them to help to fulfil his obligations to his king. These people were still considered to be the “children” (fānau, here meaning descendants through both men and women) of the title, even though they were living elsewhere. They thus had a double set of obligations, one to their leader under whom they were living, and another to the leader whose title they were descended from; but the obligation to the leader one was living with was much more important than the obligation to the title one was descended from. These latter ties tended to become ties of courtesy rather than jural obligation.

In theory, the leader's power over his political kāinga was absolute. He was not responsible to any higher authority for the way he ruled his people. The duty of his subjects was to support him with gifts of food both for his own use and so that he could fulfil his and his subjects' external obligations; to provide labour for such things as house building both for the leader himself and sometimes for the leader's superior title-holders; to provide warriors if need be; and to show deference, respect, and abject obedience to the leader, especially when outsiders were present. The leader's basic duty to his people was to allow them to cultivate land. He also shared with his political kāinga the food he received from other groups.

There were no courts in traditional Tongan society. The ruler governed his village without having to follow the advice of any formal advisory council. He had an 'ilokava every day, that is, a ceremonial drinking of kava attended by his matāpulekāinga, so that he knew what his people thought but he did not have to follow their opinions. The leader gave his people their orders in a meeting called a fono, at which he spoke and the people listened respectfully. There was no discussion. (ceremonial attendants), and certain minor title-holders and heads of families within his

In fact, the leader's power over his subjects was limited by the fact that in most areas there was no land shortage and subjects could slip away to live with relatives elsewhere if they found their leader unbearable. Further, since the leader's political power depended on the size and strength of his political kāinga, a wise leader did not alienate his people too much - 23 or too often.

Observations of the way people behave towards their nobles today, combined with the accounts of Cook, Mariner, and the early missionaries, suggest that the “rational” explanation—that subjects obeyed their leaders because the leaders controlled access to land—is not in itself sufficient to explain their behaviour. Even today there is an attitude of reverence towards a title which is usually combined with a very realistic evaluation of the character and abilities of the man who holds the title. The title represents the people of the village; it is the embodiment of their identity; it smacks of the sacred. In the Tongan view, however, the man who holds the title is an individual, a person like any other. If he acts in such a way that he enhances the prestige of the title in the eyes of outsiders, he enhances the prestige of his villagers too, and they will put up with an autocratic leader if he is respected by outsiders. Mariner's Fīnau 'Ulukālala was a notable example. A leader who acts the fool embarrasses and degrades his villagers as well as himself. We were told in the case of such a title-holder: “We respect the title but not the man”.

A title has thus two aspects, a practical secular aspect to do with ruling and organising a group of people, and a sacred element to do with embodying their identity and representing them to the outside world. The attitude towards the title is similar to that towards “father” in domestic kinship, in that the title carries secular authority (pule) but it is also sacred and makes one feel respect (faka'apa'apa). These two elements, authority and rank, were sometimes deliberately split, as when the Tu'i Tonga is said to have appointed the Tu'i Ha'atakalaua as the secular ruler, reserving higher rank for himself, or again, when the Tu'i Ha'atakalaua created the Tu'i Kanokupolu as the second secular king, and yet again, when 'Ahio, the oldest son of the first Tu'i Kanokupolu, decided that he had no wish to become Tu'i Kanokupolu himself but would be content to have a title of higher rank, letting his younger brother Atamata'ila assume the title Tu'i Kanokupolu, with its burden of ruling.

A title is not only the embodiment of its people; it also connects the present with the past. A title once created is thought to exist unchanged for ever. All the many holders of the title Vaea, for example, are in a sense felt to be the same person. Hence we heard such statements as: “Yes, 1852. That was the year when I fought King Tā'ufa'āhau”, when the actual person who had fought King Tā'ufa'āhau was the speaker's great-great-great-grandfather. Further, kinship terms are used in relation to titles in a way that perpetuates the original link between the individual and the title. For example, if one's father held a particular title, say the title of Vaea, then one will call all future holders of the Vaea title “father” and act towards them in certain situations as “father”, even if - 24 the actual holder of the title is a brother or even a nephew. If one is descended from a title one is its “child” or “grandchild” regardless of one's actual relationship with the man who holds the title. This usage made the early missionaries confused about the kinship relationships among title-holders.

These attitudes, that there is a sacred element in a title and that the title exists unchanged for ever, of course affects its relation with other titles, for the Tongan view is that the relation between titles was established by the original kinship between the first holders of the titles and that this relation continues unchanged regardless of personal changes in the kinship relations of the men who hold the titles. Hence, for example, it is not at all unusual to be told: “Our titles are brothers but we are not related”. In cases of this sort the two men concerned behave quite differently towards each other in different contexts. In situations involving duty to the monarch, or relations between the titles as such, they will act according to the older brother/younger brother relationship, but behaviour in other forms of social situation will be based on the personal kinship ties.

It is very difficult at this remove to discover the actual composition of political kāinga and the layout and form of land holding in the traditional system. We do know that political kāinga varied greatly in size from a handful of families in the case of minor titles to several hundred at the court of the sacred king, the Tu'i Tonga. A sizable political kāinga would consist of a title-holder or a leader, his wives and children, some of his wives' relatives, various descendants of former title-holders, a number of matāpule, and minor title-holders with their families, and some outsiders related by kinship either directly to the leader or to someone else in his political kāinga.

There are two common sorts of minor title-holder within a political kāinga, called tehina and foha, “brother” titles and “son” titles. Their duties, especially those of the tehina, were to act on behalf of the leader when he was absent and to sit at the head of his kava circle and generally to rule on his behalf. Today their functions are largely ceremonial. In the case of the foha titles, the original holder of the title was always a real son of the holder of the main title. In the case of the tehina title the original holder might have been a real younger brother of the title-holder or might have been some other relative, often a relative of one of the title-holder's wives. As generation succeeded generation, the men who held these minor titles ceased to be related to the main title-holder, or were related in a different manner from the original relationship.

The pattern of landholding within the political kāinga is difficult to discover and is a matter in which our field work was very incomplete (see - 25 Maude 1971). It is generally agreed that in theory all land was held by the sacred king, the Tu'i Tonga, but that he allowed various title-holders to use and inherit certain estates traditionally assigned to them. Although in most areas there was no land shortage, all land was held by someone and one could not use it without permission. It is said that at one time settlements were dispersed, coalescence into fortified villages having occurred only during the late 18th century, when persistent warfare among the rivals for the Tu'i Kanokupolu title led to fortification of settlements. Her Majesty Queen Sālote thought that this was probably an over-simplification. She said that she had always been told that political kāinga lived all together on the leader's 'api (hereditary plot of land) with a reed fence surrounding the houses and gardens. As the kāinga grew, the fence was enlarged. Captain Cook, however, notes that the great aristocrats of Mu'a, the sacred capital, had no garden plots inside their enclosures, presumably because they were entirely supported and fed by their subjects.

It seems likely that if a political kāinga grew and prospered, the people would begin to occupy and use parts of their leader's land that had hitherto been unused. In many cases it appears that tehina and foha were granted hereditary plots ('api) of their own, for nowadays tehina and foha often say that certain pieces of land in the noble's estate (tofi'a), used to be held by them, although these holdings were not recognised in the Constitution. Sometimes a loyal commoner would also be granted his own 'api. Judging from the elation of the missionary Vason and the sailor Mariner when they were granted 'api, having one's own 'api was an important step towards independence and higher status, even though one was obliged to give some of one's produce to the title-holder and generally to support him in all matters (Vason 1810, Martin/Mariner 1818).

Sometimes a small group of kin headed by a tehina or foha, or an untitled commoner, all worked together on one piece of land. As the group became larger it was customary for each family within the group to cultivate one portion of the land of the head ('ulumotu'a) of their kin group. Each family then gave some food to their own 'ulumotu'a and the title-holder. (This was told to me by Soakai, a knowledgeable and well-educated minor title-holder of Ha'apai, who made extensive inquiries for me on this matter.) The system was flexible and adaptable because in most areas there was no land shortage and there was, of course, nothing comparable to the present law which limits 'api (except those of nobles) to eight and one-quarter or sometimes four and one-eighth acres.

In other words, the very scanty information available suggests there were subdivisions within the political kāinga, each with its own head and - 26 each sometimes having a hereditary plot of land. I was unable to discover any generic or proper name for these sub-groups headed by an 'ulumotu'a. Lātūkefu (1974), however, has said that they are at present called fa'ahinga and Alaric Maude, speaking of the traditional system, takes it further:

In other words, Lātūkefu says there is an explicit concept of a unilineal descent group called fa'ahinga, and Alaric Maude says that in old Tonga the fa'ahinga had a corporate estate. Although I suspect our data are substantially in agreement, I think Lātūkefu and Maude push the data a little too far in the direction of explicit recognition of patrilineal corporate groups. I did not find any real agreement among my informants that fa'ahinga meant lineage, and Garth Rogers had the same experience. Most of the people I talked to thought that fa'ahinga meant “kind” or “species”. Two or three people told me, as they told Rogers, that fa'ahinga was a classification of pigs, not people. Queen Sālote said she thought one might use the term when asking how some distant set of people were connected with a particular title or ha'a; note, however, the connection with titles and ha'a, which are in the domain of authority and patrilineal descent.

When we were trying to discover in present-day villages whether people recognised or had any explicit concept of groups of kin larger than the small groups of co-operating brothers described above, we asked the town officers of 53 villages on the main island whether such groups existed and what they were called. There was no immediate agreement even that such groups existed, and it was clear we were asking a question that did not fit with Tongan concepts of village structure. Most town officers, however, having been asked, made some attempt to answer. In some cases they gave the names of minor title-holders as the heads of such groups, sometimes they named the heads of the small family groups of co-operating brothers, sometimes the names of prominent men without titles, most usually a mixture of both. There is no agreement on whether there was a generic name for these groups or what - 27 it was. Sometimes the groups were called fāmili, sometimes kāinga, and sometimes matakali (the term matakali is of Fijian origin, and the villagers whose town officers used that term have Fijian connections). No one used the term fa'ahinga spontaneously, but when I inquired about it explicitly, one informant, who was especially interested in history, used the term as Lātūkefu does, as a minimal patrilineage headed by a matāpule or title-holder, and he emphasised its connection with where one was buried. I would agree with Garth Rogers that

The word matakali was also used by occasional informants to mean a minimal patrilineage headed by a minor title-holder, but this, too, was a concept rather than a functioning group.

A ha'a is a set of titles deriving from one or other of the three ancient kings of the traditional system, the Tu'i Tonga, the Tu'i Ha'atakalaua, and the Tu'i Kanokupolu. In the traditional system ha'a membership was of great political importance because it regulated the flow of tribute and provided a framework of stability and continuity with the past. Today it is of ceremonial but not political or economic importance.

With the exception of occasional radical restructuring, titles and ha'a changed much more slowly than political power and the rank of individuals. The kava ceremony was and still is the ceremonial context in which this conservatism of title and ha'a is given explicit expression. Titles retain their place in kava ceremonies long after they have ceased to have any political importance and long after anyone, even the title-holder himself, has ceased to know how the title began (Bott 1972).

In the traditional political system, tribute went from each title-holder's group to the head of his ha'a and then to the king from whom the ha'a was derived. If the ha'a was derived from the Tu'i Kanokupolu, there were certain types of situation in which the food was then sent from the Tu'i Kanokupolu to the Tu'i Ha'atakalaua, then from the Tu'i Ha'atakalaua to the Tu'i Tonga. This was the case in the great 'inasi ceremony, held to celebrate the harvest of the first yams of the season, a ceremony which was both religious and political and involved an expression of unity and solidarity from all the political kāinga of the kingdom. The quantity of yams was not great, but large quantities of other food were brought along in order to feed the people assembled at the capital.

When the Tu'i Tonga died, when he returned from a voyage, and when - 28 a newly appointed Tu'i Tonga first received the homage of his subjects, there was a ceremony called pongipongi involving a royal kava ceremony (taumafa kava) and the presentation of food and kava on a large scale, each title-holder contributing his share through the head of his ha'a to the king from which his title was derived and so on to the Tu'i Tonga. Similarly, pongipongi were held by each of the secular kings. When a lesser title-holder himself succeeded to his title, he also had a pongipongi hingoa, which involved presenting himself and his people to his king in a kava ceremony at which large quantities of food were presented and distributed. In addition to these special occasions, gifts of uncooked food were frequently presented to the king from whom one's title was derived and labour and warriors were provided when required.

Ha'a membership thus defined a set of formal obligations to other title-holders and to the three kings. The organisation of these presentations depended somewhat on where the title was located. In the classical period the three kings lived on the main island, Tongatapu, with the Tu'i Tonga and the Tu'i Ha'atakalaua living at Mu'a and the Tu'i Kanokupolu living in the western end of the main island called Hihifo. The northern islands of Ha'apai, Vava'u, and Niuafo'ou usually sent their 'inasi and pongipongi as island groups, sending it first to the dominant secular title-holder of their own island; this secular leader then sent the tribute to the king from which his own title was derived. The island of Niuatoputapu, however, became virtually independent, for, not only was it even more remote than the others, it had also a very powerful ruler of higher ceremonial rank than the Tu'i Tonga, so that he was not subject to the usual demand for tribute.

The definition of the ha'a by anthropologists has always been controversial for there is the same ambiguity about it as there is over the relation of the title and its patrilineal kin group—did the territorial estate and its people accrue to the title or to the kin group? Is the ha'a a descent group that “owns” titles, or is it a set of titles that mobilises kin groups and territorial groups? Gifford adopted the first view and defined the ha'a as a patrilineal lineage focused on titles. Most other authors have questioned his assertion, especially pointing out that ordinary villagers do not know what the ha'a is or which one they belong to, which would be a very odd state of affairs if the ha'a were a lineage as usually defined. Gifford's original definition was as follows (1929: 29-30):

Firth (1936: 587; 1957b) questions the patrilineality of the ha'a, as does Coult (1959: 60-1); the Beagleholes (1941) question whether lineages exist at all; more recently Aoyagi (1966) notes that the villagers of Nukuleka on Tongatapu had no idea what the ha'a was; Marcus (1975a) reports that the people of 'Uiha knew nothing of the ha'a. Adrienne Kaeppler (1971) says that Aoyagi did not find out about the ha'a because she asked the wrong people, since ha'a concern only chiefs and the societal structure. Kaeppler says that ha'a “. . . are societal divisions that have political functions and can be conceptually separated from the kinship system . . .”; she also describes them as “. . . groups associated with a title originated by collateral segmentation . . .” “Associated with a title are the title holder, his descendants, and descendants of former holders of that title.” At this point she does not specify the type of descent, though later she says: “. . . demonstrating that societal inheritance is not simply by patrilineal descent, or even close blood relationship, and that ha'a are not lineages by the usual definition.” Still later she adds: “. . . an individual ideally tracing his ha'a through his father's line . . .” (Kaeppler 1971: 175, 180, 183, 188).

Shulamit Decktor Korn says (1974: 6-7):

Korn's definition agrees substantially with what I was told, but is somewhat misleading because it implies that any individual can choose whether to belong to the ha'a of his father or his mother. In fact, ha'a affiliation applies to titles more than to persons. It is true that in the past, for various reasons of political advantage, titles sometimes got drawn - 30 into a ha'a by a kinship connection through a woman, though patrilineal descent was the customary rule. But it is not consistent with the tenor of the contemporary sources or Queen Sālote's explanations to think of individual Tongans choosing which ha'a to align themselves with. They were the subjects of the title-holder they lived under; any change of affiliation came about through moving away to live under another title-holder.

It is perhaps no accident that ordinary villagers nowadays are so willing to be ignorant about the ha'a, so quick to say, “I don't know. It is a thing for the chiefs and the nobles.” The ha'a is one of those institutions that used to lead to particularly heavy fatongia (obligations) for the ordinary commoner, for it was the ordinary commoners who had to do the work that allowed their title-holders to fulfil their ha'a obligations. In forgetting or not knowing about the ha'a, a becoming modesty is happily combined with practical convenience.

At various times and in various contexts I was given three answers to the question: “What is the ha'a?”: Firstly, it includes titles only. This definition was much the most common, and was given by nobles and experts in Tongan custom. They made it clear that it was a group of titles derived from the former holders of the three kingly titles.

Secondly, it includes the titles and the close fānau of the titles, fānau being descendants of former title-holders through both men and women. This definition was given only by title-holders of a Ha'a Ngata Motu'a and Ha'a Ngata Tupu, both of which are derived from the Tu'i Kano-kupolu title. All titles have fānau, but in these two ha'a the close fānau of the titles are entitled to perform certain functions in the kava ceremony of the Tu'i Kanokupolu. The fānau of the various titles in that ha'a do not act together as a group; rather they form a pool of individuals whom the Ha'a Ngata title-holders can call upon to perform certain ritual tasks.

Thirdly, it includes all the descendants of the original holders of the titles of the ha'a. When I asked descendants through men or through women, my informants answered, after some hesitation, that it should be through men. Such a group would be a large-scale lineage, or, in view of the hesitation about the rule of descent, a large-scale non-unilineal descent group. This definition exists as a conceptual category in the minds of informants who have reflected on Tongan custom and history, but in practical operations it is not a social reality. There are no occasions when this large group of descendants assembles and acts together as a group. Many persons who should belong to such a group by descent are not aware of their membership. The descendants are scattered all over the island and their primary ties of loyalty are to the title-holders under - 31 whom they live and to their personal relatives, not to the title-holders they are descended from 13 or more generations ago. Only kings and aristocrats trace their genealogies so far.

Because Tongans so consistently stress the title rather than a descent group, and because the concept of the political kāinga is so firmly territorial, I think the most appropriate solution is to define the ha'a as a group of titles, with the political kāinga and the fānau as the groups mobilised by each title in order to fulfil its obligations to its king and senior title-holders, obligations which are now ceremonial but were formerly political as well.

A grouping of titles into ha'a is based on the kinship ties that are supposed to have existed between the first holders of the title. As I have already described, the relation between titles is supposed to be unchanged even though the personal kinship relation between the men who hold the titles may be very different from the kinship relation between the original title-holders. The various ha'a are ranked in relation to one another according to their genealogical seniority. This “ha'a ranking” determined the flow of tribute and the ordering of ceremonies organised according to the ha'a, but did not play so notable a part in everyday social interaction and political manoeuvring, in which the kinship connections and personal rank of title-holders as individuals were the important factors.

Gifford (1929) gives detailed stories of the various titles and ha'a. I have also described elsewhere the genealogical derivation of the three sets of ha'a, one derived from the Tu'i Tonga line, one from the Tu'i Ha'atakalaua line, and one from the Tu'i Kanokupolu line (Bott 1960, in press). I have also discussed elsewhere (Bott 1972), the role of titles and ha'a in the kavaha'a, together with useful and clarifying diagrams of their genealogical derivations. ceremony. Kaeppler (1971) also presents a detailed account of the

The alignment of titles and ha'a underwent a massive change when the Constitution was being developed. When the last sacred king died in 1865 he granted his kava privileges to the Tu'i Pelehake and his right to rule as the sacred king to the Tu'i Kanokupolu. The kava ceremony of the Tu'i Kanokupolu was rearranged in order to accommodate the Kauhala'uta and Ha'a Takalaua titles, the Kauhala'uta being those derived from the Tu'i Tonga title and the Ha'a Takalaua titles being those derived from the Tu'i Ha'atakalaua. Further, at the time of the granting of the Constitution, only a few of the many titles of Tonga were made into nobles and given hereditary estates (tofi'a), these titles being those whose incumbents were most powerful at the time. The other titles continued to be appointed and they retained their kava rites, but their origin and - 32 derivation of the ha'a to which they belong are slowly being forgotten.

The basic familial principles of rank have been described above: sisters rank higher than brothers; older siblings of the same sex rank higher than younger siblings of the same sex. Other things being equal (which in Tonga they seldom are), the descendants of sisters rank higher than the descendants of brothers and the descendants of older siblings higher than the descendants of younger siblings of the same sex. The principles of rank are also developed and extended so that they apply not only in the domestic sphere but also throughout the society in such a fashion that a person's rank is an ascribed part of his identity. It is inherited from both parents. It does not change from one situation to another, although the expression of deference to the person concerned will depend on who is present and on the social occasion.

The sacred king, the Tu'i Tonga, held the most senior and therefore the highest-ranking title in Tonga. He was the sacred father of the whole nation. The authoritarian, coercive, cruel aspects of authority were split off from his title and lodged in the role of hau, the secular ruler, first the Tu'i Ha'atakalaua and later the Tu'i Kanokupolu, leaving the Tu'i Tonga as the sacred, benign role, the quintessence and symbolic embodiment of the nation. In theory the Tu'i Tonga should have been the highest-ranking person in Tonga as well as holding the highest-ranking title. But of course his sisters had higher rank than he. A method was evolved for recognising their special position and arranging marriages for them which created offspring of higher rank than the Tu'i Tonga without threatening the special position of his title. It is said that the older sister of Tu'i Tonga Fatafehi (roughly 14 generations ago, in the early 17th century), was given the special title of Tu'i Tonga Fefine (literally female Tu'i Tonga). According to legend, this Tu'i Tonga Fefine married a Fijian called Tapu'osi in very romantic circumstances. They had a son who started a new title and a new ha'a, the Fale Fisi (the house of Fiji). In the usual Tongan fashion where titles are concerned, the titles of the Fale Fisi were regarded as foreign for ever because the original ancestor was Fijian. Henceforth the Tu'ilakepa and the Tu'iha'ateiho, the two main titles of the Fale Fisi, usually married the Tu'i Tonga Fefine, the theory being that they were foreigners by title and therefore their children, being only half sacred because they were also half foreign, did not threaten the special title and prerogatives of the Tu'i Tonga title, even though they had higher personal rank than the Tu'i Tonga. These children of the Tu'i Tonga Fefine and the Tu'ilakepa or the Tu'iha'ateiho were loosely referred to as Tamahā (sacred child).

- 33

Anderson, the surgeon on Cook's third voyage, made a particularly astute observation in this connection, for he noted that Tu'i Tonga Paulaho deferred to Lātūnipulu (tama 'a mehekitanga) and his sisters on ordinary social occasions, but that at the 'inasi'inasi ceremony Lātūnipulu was acting according to his title (Tu'ilakepa), whereas on ordinary social occasions he was acting as Paulaho's tama 'a mehekitanga. ceremony Lātūnipulu “. . . assisted only in the same manner as the other principle men” (Beaglehole 1967, III, 2: 954). In the

Kinship principles were used to link the Tu'i Tonga and the hau (the secular ruler) in such a way that the Tu'i Tonga's higher rank was constantly reinforced. Thus, personal rank and the rank of titles were made to support each other. The hau, at first the Tu'i Ha'atakalaua and later the Tu'i Kanokupolu, sent his daughter to be the great chief wife, the moheofo, of the Tu'i Tonga. She was always the mother of the heir.

TT	Tu'i Tonga	d	daughter of
TK	Tu'i Kanokupolu		son of
TH	Tu'i Ha'atakalaua	———	half or classificatory siblings
THT	Tu'iha'ateiho	♀	woman
THNg	Tu'iha'angana	(1)	indicates order of spouses
TP	Tu'ipelehake	OR	which spouse is parent of a child.
FU	Fīnau 'Ulukālala	Additional titles in italics

- 34

Hence the group of the Tu'i Ha'atakalaua and later of the Tu'i Kanokupolu was called the Ha'a Moheofo. Unlike the enduring relations between titles, these relations of rank established by marriage concerned the persons who held the titles and lasted only a generation or so. The marriages had constantly to be repeated to maintain the high personal rank of the Tu'i Tonga compared with that of the Tu'i Kanokupolu. It was, of course, a marriage of the Tu'i Tonga to his matrilateral cross-cousin, called kitetama. Similarly, the marriage of the Tu'ilakepa or the Tu'iha'ateiho to the Tu'i Tonga Fefine became a kitetamamoheofo. marriage, especially if repeated generation after generation, which was the case, though not so consistently, with the Tu'i Tonga and his

I have already described above the way the title embodies and symbolises the village, the political kāinga. Similarly, the Tu'i Tonga embodied and symbolised the whole of Tonga. His personal kinship relationships in a sense involved the whole nation and his rank in relation to that of his mother's brother, the Tu'i Kanokupolu, pertained to the whole nation not just to his own personal kin network. It is this conception that serves as a model for chiefliness embedded in a person's flesh—sino'i 'eiki—which is recognised throughout the whole society. The essential “formula” for being sino'i 'eiki was based on this type of marriage, for, according to Queen Sālote, a sino'i 'eiki in traditional Tonga was the offspring of an 'eiki fakanofo of Kauhala'uta (a title-holding aristocrat of the ha'a deriving from the Tu'i Tonga) and the daughter of a king, meaning the hau. A sino'i 'eikihau and a woman who was herself the daughter of a marriage between an 'eiki fakanofo of Kauhala'uta (such as the Tu'ilakepa or the Tu'iha'ateiho) and the sister of an 'eiki fakanofo of Kauhala'uta (such as the Tu'i Tonga Fefine). That is, a person would also be sino'i 'eiki if his father was the Tu'i Kanokupolu and his mother was the Tamahā, that is, a child of the Tu'i Tonga Fefine and the Tu'ilakepa or the Tu'iha'ateiho. According to Queen Sālote, there were only three properly designated and appointed Tamahā: Tu'imala, Lātūfuipeka, and 'Amelia Fakahiku'ouiha. But all the children of marriages between the Tu'i Tonga Fefine and the Fale Fisi chiefs were very high rank and were loosely referred to as “tamahā”. could also be the offspring of the

According to Queen Sālote, special terms of reference were used for these very great sino'i 'eiki. The Tu'i Tonga, the Tu'i Tonga Fefine, and the Tamahā were called “'eiki” if they were men or “ta'ahine” if they were women. Aristocrats of slightly lower rank were called “tama” for men and “ta'ahine” (again) for women. Similarly certain special words are used only when addressing the king and other special words are used only when addressing other high-ranking aristocrats.

- 35

Queen Sālote thought that this efflorescence of titles of rank and special terms of reference and address for people of high rank had come about at the same time as the emergence of the Tu'i Kanokupolu title. She thought it was a compensatory development because of the Kanokupolu emphasis on secular power. She attributed it to the influence of the Samoan mother of Ngata, the first Tu'i Kanokupolu, for Samoans emphasise titles, and Ngata's mother brought a large contingent of Samoan followers with her to Tonga.

- 36

When discussing rank within the family, I noted that rank depended not only on sex and age within the family, but also on inheritance from both parents. Such inherited “chiefliness” (sino'i 'eiki) depends on descent from kings and the sisters of kings, but particularly on descent from the Tu'i Tonga Fefine and the Tamahā. Nearly all the great aristocrats of Tonga in 1958-60 were descended from Lātūfuipeka, the last Tamahā who had children (Figure 2). Most of the great aristocrats of 1958-60 were descended from Lātūfuipeka by several lines and from various Tu'i Tonga Fefine as well.

- 37

Now that the titles of Tu'i Tonga, Tu'i Tonga Fefine, and Tamahā have become defunct, the Fale Fisi title-holders and their titles have decreased in rank. In 1960 it was often said that there are no more really great sino'i 'eiki as there had been before, only the two sons of Queen Sālote qualifying for the special term “'eiki”. But there were all shades of aristocrat, all descended with varying degrees of dilution from the former Tu'i Tonga, Tu'i Tonga Fefine, and Tamahā. It was generally surmised that in the future the connection with and descent from the present royal family, would come to have the same effect as connection - 38 with the Tu'i Tonga and descent from the Tu'i Tonga, the Tu'i Tonga Fefine, and the Tamahā.

In Tonga being a great aristocrat (sino'i 'eiki) is considered an end in itself. It means that one is given universal recognition and deference, accorded the special language of respect, given gifts of food, and sought after in marriage. “To be known” is the synonym for high rank; to be “not known” is to be of low rank. In the traditional system, titles involved ruling and responsibility. Being an aristocrat meant pure privilege, so much so that many of the great aristocrats of the 18th and 19th centuries, even though they ruled political kāinga and had titles, did not bother to use their titles or even to get themselves appointed, and were known by their personal names instead. Nor were they particularly interested in personal power. The title was and still is spoken of as a “garland” (kakala), meaning that it can be taken away whereas the “blood” (toto) is one's own for ever.

Situations often arise in which there is conflict between a person's rank at the level of domestic kinship and his rank in society as a whole. For example, in 1959 a woman called 'Elenoa died. According to principles of domestic kinship, the heir apparent, then holding the title of Tungī, should have been liongi at the funeral because 'Elenoa was his classificatory father's sister (mehekitanga). But her father was “no one”, that is, his rank was very low, whereas Tungī was not only heir to the throne but was a very great sino'i 'eiki as well.

'Elenoa's child, who was organising the funeral, solved the conflict by invoking another much more remote kinship link which made 'Elenoa “low” and Tungī “high”. According to this link, 'Elenoa's father, Kumā, was descended from a brother of Ikatonga, the woman from Fua'amotu who married Mulikiha'amea, son of Tu'i Ha'atakalaua Maealiuaki and ancestor of the Tungī line. Since Kumā was descended from the brother and Tungī from the sister, Tungī's kinship rank could be regarded as higher than 'Elenoa's.

Similarly, it often happens that a sister marries a person of lower rank whereas her brother marries someone who is sino'i 'eiki. The children are then in a dilemma. According to the kinship principle that the children of a sister have higher rank than the children of a brother, the sister's children should be higher in rank. But according to the principle of inheriting rank from one's parents, the children of the brother should have higher rank than the children of the sister. The outcome in cases of this sort, of which there are many, depends on the social context and on the characters of the people involved. Tongans are very skilled in manipulating and resolving such conflicts.

- 39

Indeed, it was because situations of this sort were constantly arising that I found it more appropriate to analyse Tongan social structure as a set of interacting principles than looking for well-defined structures. In any given type of social situation I eventually learnt to know which principles would be involved, but I could never predict the empirical outcome.

Finally, it is important to note how different this system of rank is from a system of social class. In Tonga, no two people have the same rank, even within the immediate family. There can easily be conflict between rank according to kinship criteria and rank in the society at large, even though rank in the society at large is derived from an extension of principles of kinship rank to the kings, their sisters, and their sisters' children. In the traditional system there was a notable lack of fit between the system of rank, of de facto political power, and of titles. Even nowadays the system of rank is somewhat in conflict with the emerging social classes.

- 40

Because the traditional system of rank and authority worked so differently from what is ordinarily understood by sociological concepts of social and economic class, I think it best to avoid the term “class” except when discussing the specifically class-based features of the modern system. In this respect I differ from Kaeppler (1971) and Marcus (1975a, 1975b).

The enduring corporate groups of traditional Tongan society, then, were territorial groups of subjects ruled by political leaders who usually, though not invariably, held hereditary titles (hingoa fakanofo “appointed name”). The leader's subjects were called his kin, his kāinga; many of them actually were his kin but those who were not acted as if they were. Residence was what made one a subject but kinship was the idiom used to describe the leader-subject relationship. The leader was called 'eiki. Like the role of father, the role of 'eiki had elements both of secular authority and of the mystical power of rank; the title embodied the fonua, the land and its people.

The external relations of these corporate local groups were regulated by the membership of their leaders in two systems; the system of titles and ha'a, described above and the network of kinship ties created by their leaders' marriages. The system of titles and ha'a was a sort of charter that in theory was supposed to fix for ever the genealogical position of the titles and their relative rank and obligations. But the vagaries of personal ability to command men and resources, and the kinship ties created by leaders' marriages, led to marked fluctuations of political power and personal rank, so that the actual situation of political power and personal rank was infinitely more complicated than the formal charter of titles and ha'a implied. Very occasionally the charter of titles and ha'a had to be brought up to date to take account of the actual changes that had taken place through the growth of new titles and the decline and demise of old ones. The stable genealogical charter of titles and ha'a, combined with the flexibility in marriage rules, allowed for both continuity and change. Viewed in the short term, the interplay of titles, personal power and personal rank was exceedingly complex and unpredictable. Over a time, however, certain patterns are discernible, and it is to a discussion of these that I turn now.

When a leader of a local group married, it was a political act, for, provided there were children, it set up relationships between the groups concerned that lasted for a generation, usually for two generations. What was personal kinship for the leader was politics for his subjects. It involved the whole local group in rights and obligations. The usual - 41 expectation, according to the principle that brothers honour and support their sisters and their sisters' children, was that a woman's original local group would give food and support to the group into which she married and to which she gave children. Sometimes her group gave a piece of land, especially if one of their woman's children was likely to succeed to the title and position of their woman's husband. The strength in land and manpower of the potential wife's people was therefore an important factor in choosing wives; a girl's rank was also important since it affected the rank of her children. Similarly, the mother's people and the rank of the mother were important in choosing which of the many possible heirs would succeed to a title or a leadership position. All through Tongan history, legend, and contemporary accounts the theme of the chiefs relying on their mothers' people is endlessly repeated. Because chiefs practised polygyny, because high-ranking women practised serial monogamy, and because divorce was easy, the web of kinship and local group alliance was exceedingly complex.

(a) The marriages contracted by the men of a patriline

If one examines Tongan genealogies to compare the history of marriages and alliances of successful and unsuccessful lines of chiefs, it becomes apparent that there is a typical sequence from marriages to secure political supporters to marriages to women of high rank who would raise the rank of the heirs to the title and eventually would raise the rank of the title itself. The more successful lines of title-holders were those who kept a balance between the two types of marriage. Unsuccessful lines were those whose chiefs married for rank too soon or too often, or those who married too often for local support. Successful lines also made judicial use of the alliance potential of their women.

In an earlier paper (Bott, in press), I have outlined in detail Queen Sālote's account of the way the ancient kings of Tonga used to send their younger brothers and sons out to distant villages and islands to get hold of these territories on behalf of their king. At first the brothers and sons were sent by the Tu'i Tonga, then by the Tu'i Ha'atakalaua and finally by the Tu'i Kanokupolu. I do not want to repeat the Queen's detailed account of this process, but I will outline certain aspects of it so as to show the use of marriage in the process of developing power and title.

Typically the younger brother or son sent out by a central king had low rank at the central court because of his age but high rank in the outlying village or island—“half commoner at court, half king in the bush”—as the Queen put it. As a person of high rank, he was likely to be fed and supported when he went to the outlying community. It was also an - 42 advantage if he had some sort of kinship connection which would make him superior to the people he was going to attempt to take over; it was particularly useful, for example, for his mother or his mother's mother to have come from the island concerned. In many cases he went to the outlying island with a supporting group of warriors as well as his high rank and his kinship connections. Once he had managed to land and to get a foothold in the local community, he then married the daughters of an important local title-holder in order to secure the support of the local title-holder and his people. The usual rule of patrilineal succession of leaders was waived so that the son of the immigrant aristocrat and the chief's daughter could succeed to the leadership position of the old chief. Thus, the local title was either taken over by the immigrant aristrocrat, or forgotten, or drawn into the ha'a of the immigrant, who was sooner or later granted an official title by the king from whom he was descended.

Of the many examples of this process I will briefly describe four.

1. Ngata, the first Tu'i Kanokupolu

Ngata, the first Tu'i Kanokupolu, was sent by his father, a Tu'i Ha'atakalaua, to see if he could establish himself in Hihifo, the western part of the main island of Tongatapu. He married two daughters of a local chief, 'Ahome'e. The sons of Ngata became the founding ancestors of the several titles that form the Ha'a Ngata, with the exception of the one who became Tu'i Kanokupolu himself, Atamata'ila. The title of 'Ahome'e became known as “grandfather” to the Ha'a Ngata and was eventually drawn into the Ha'a Ngata.

2. The marriages of Tu'i Kanokupolu Mataeletu'apiko and the establishment of the Ha'a Havea

The third Tu'i Kanokupolu, Mataeletu'apiko, married several women, among them, Papa and 'Umukisia, whose fathers were title-holders in the central part of Tongatapu. These women had several sons to Mataeletu'apiko and these sons became the founding ancestors of the titles of the Ha'a Havea. These sons in turn married women who were also daughters of title-holders in central Tongatapu, and the sons took over the positions and lands of their respective fathers-in-law and thus established themselves as the rulers of central Tongatapu. The titles of the fathers-in-law became defunct.

3. Mumui, ancestor of the present Tu'i Kanokupolu line

Mumui was one of the younger sons of the sixth Tu'i Kanokupolu, Ma'afu-'o-tu'itonga. He went to live with his wife's people in - 43 Kolomotu'a on the main island of Tongatapu, taking over the leadership position of her father. He married many other women, all of low rank, who brought their male relatives to live with him, and thus gradually built up a large and strong political kāinga. Mumui became Tu'i Kanokupolu. Three of his sons also became Tu'i Kanokupolu: Tuku'aho, Tupoumālohi, and 'Aleamotu'a.

4. Tuituiohu, ancestor of the Fīnau 'Ulukālala title and the Ha'a Ngata Tupu

Tuituiohu was a younger son of the fourth Tu'i Kanokupolu, Mataeleha'amea, from the western part of the main island of Tongatapu. Tuituiohu went to Vava'u and established himself there by getting the support of his mother's people and the support of the people of his maternal grandmother, who was a woman of Vava'u. He then married local women whose fathers and brothers gave him support.

Having got started in the manner just described, the next task for a new line of chiefs was to build up its strength. The proper thing according to Tongan ideology was for the younger brothers of the new chief to support him which they usually did at the beginning of the development of the new line when his success had not yet been securely established. For example, the second Tu'i Kanokupolu was strongly supported by his brothers, who were the founding ancestors of the Ha'a Ngata Motu'a (Figure 4). This particular set of brothers is said to have had the same mother, which diminished their rivalry. By the time of the third and fourth Tu'i Kanokupolu, when the Ha'a Havea had been formed and the Kanokupolu chiefs were spreading out into central Tongatapu, competitiveness between brothers became marked. Vuna, who eventually became the fifth Tu'i Kanokupolu, is said to have been very resentful because his younger brother, Mataeleha'amea, was chosen as the fourth Tu'i Kanokupolu in preference to Vuna himself. Vuna went off to Vava'u ostensibly to support his younger brother but actually to found a new ha'a. Mataeleha'amea's other brothers, who were the founding ancestors of the Ha'a Havea, are also said to have been resentful and competitive and not as supportive of the Tu'i Kanokupolu as the Ha'a Ngata were. These attitudes are said not only to have existed at the beginning but to have been continued so that the Ha'a Ngata always think of themselves as very strong supporters of the Tu'i Kanokupolu and are somewhat critical of the Ha'a Havea on the grounds that their support is traditionally less wholehearted.

After building up its strength by marriages to the daughters of local powerful chiefs, a new line of chiefs would begin to marry great aristocrats as well. Thus, the third Tu'i Kanokupolu, Mataeletu'apiko,

- 44

married a Tamahā, Tu'imala, and another woman of high rank. The chief wife of the fourth Tu'i Kanokupolu, Mataeleha'amea, was the daughter of a Tu'i Ha'atakalaua. All his other wives were her fokonofo (secondary wives closely related to the chief wife). Mataeleha'amea's children began to spread out. His son Tupouto'a established himself in - 45 Ha'apai; his sons Kafoa and Tuituiohu went to Vava'u, where each started once again the process of marrying local women and then building up their power. All these chiefs at first supported the Tu'i Kanokupolu but later, after they had become more securely established, they competed with each other over who should be the ruler (hau) of Vava'u and Ha'apai, the Vuna line or the Fīnau 'Ulukālala line, which was descended from Tuituiohu.

The sixth Tu'i Kanokupolu, Ma'afu-'o-tu'itonga, had four main sons; Ngalumoetutulu, Tupoulahi, Maealiuaki and Mumui (Figures 4 and 5). Two of these lines failed in the leadership competition and two succeeded and I want now to examine why.

The mother of Ngalumoetutulu was a local woman of Ha'apai and Ngalumoetutulu's descendants formed a large and well-supported kāinga in Ha'apai. But the men of this line married local women who were not aristocrats and their descendants became “not well-known”. Some of - 46 the women of this line were very much sought after in marriage because of their strong local support and their descent from a king. Several present-day aristocrats are descended from 'Ulukilupetea, a daughter of Ngalumoetutulu. She was called “the woman with the ivory stomach” because she married so many great chiefs and had so many children of high rank. Tuita is descended from her, and so is Fīnau 'Ulukālala and the Tu'i Kanokupolu. Also descended from a woman of this line (actually from 'Anaukihesina, a sister of 'Ulukilupetea) are the Kaho family, a set of forceful and talented brothers who attempted to dominate the political scene for several years during the reign of Queen Sālote. Thus, Ngalumoetutulu's daughters produced well-known descendants, some of high rank, some of great power, but his descendants through men are “not well-known”. According to Queen Sālote, they were kept busy in the early days of the line by supporting their sisters and their sisters' eminent children. They themselves married only to secure local support.

Tupoulahi became Tu'i Kanokupolu. His descendants lived at court in Mu'a, and married aristocrats. Their women were much sought after in marriage because of their high rank but the men of the line lacked strong local support and this line too eventually became “not well-known”. This line declined for the opposite reason from Ngalumoetutulu's line—too much rank and not enough local support.

Maealiuaki also lived at Mu'a though his wives were not so aristocratic as Tupoulahi's. He became both Tu'i Kanokupolu and Tu'a Ha'atakalaua, his claim to be Tu'i Ha'atakalaua being based on the fact that his father's mother, Kaloafūtonga, was a daughter of a Tu'i Ha'atakalaua. His line went briefly into decline after the death of his son Mulikiha'amea in the wars at the end of the 18th century, but re-emerged as the major title and leader of the Ha'a Takalaua. Mulikiha'amea was in effect, the founding ancestor of the Tungī title, a line of powerful chiefs of extremely high personal rank (Figure 6, the Tungī line).

Once again the early leaders of this new line, Mulikiha'amea and Fatukimotulalo, married local women to secure support and their descendants married women of high rank. Eventually Tungī Mailefihi married Tu'i Kanokupolu Sālote Tupou III and was the father of the present monarch, Tu'i Kanokupolu Tāufa'āhau Tupou IV.

- 47

Mumui was one of Tu'i Kanokupolu Ma'afu-'o-tu'itonga's younger sons, much younger and lower in rank than Tupoulahi and Maealiuaki, who were full brothers, and Mumui is the ancestor of the present ruling line. Mumui himself became Tu'i Kanokupolu and, according to Thomas (n.d., also Thomas ca. 1853), at one time he was also Tu'i Ha'atakalaua, although there was no tradition of this in Tonga in 1960. According to Queen Sālote, Mumui never expected to become Tu'i Kanokupolu. He lived at Nuku'alofa in the fonua of his first wife, and brought many of his other wives (16 in all) and their relatives to live with him so that he built up a large and strong kāinga. He did not marry aristocrats. Three of his sons became Tu'i Kanokupolu: Tuku'aho, - 48 Tupoumālohi, and 'Aleamotu'a. Tuku'aho married both women of low rank and strong kāinga, and a great aristocrat, the Tamahā, though she did not have children. Tupoumālohi married an aristocrat and so did his son Setaleki Mumui. 'Aleamotua'a married local women of strong kāinga.

Tu'i Kanokupolu Tuku'aho was a very strong and cruel king and was murdered by Fīnau 'Ulukālala-'i-Feletoa and his brother (uho tau) Tupouniua in 1799. Both Fīnau 'Ulukālala-'i-Feletoa and Tupouniua were descendants of Tuituiohu, who was a son of Tu'i Kanokupolu Mataeleha'amea. There followed a period of strife and wars among the various contenders for the Tu'i Kanokupolu title and this period of warfare coincided with the first mission to Tonga and the arrival of ships and runaway sailors. Eventually Tuku'aho's son Tupouto'a succeeded to the title of Tu'i Kanokupolu. Then 'Aleamotu'a was appointed and finally Tāufa'āhau, son of Tupouto'a. Tu'i Kanokupolu Tupouto'a, like his father Tuku'aho, married both women of high rank and women who gave him local support.

Thus both Maealuiaki's line and Mumui's line were successful and both showed the pattern of marrying at first to secure local support and then later to keep a balance between marriages to women of high rank and marriages to secure support.

If one compares other lines derived from the Tu'i Kanokukpolu, such as the Vuna linc and the Fīnau 'Ulukālala line, one finds the same pattern. Vuna's line failed for lack of local support. Fīnau 'Ulukālala's line succeeded by marrying first for local support and then by combining marriages for local support with marriages to great aristocrats (Bott in press).

There were thus two “commodities” men married for: local support and high rank. A successful title-holder had to achieve a delicate balance between the two. If a man married an aristocratic woman, his children had much higher rank than himself but lower rank than their mother. The children could not expect the usual degree of political or military support from the male relatives of their mother because, in the case of such a marriage, the male relatives of the mother normally had higher rank than their sisters' children. Hence such a marriage increased the rank of the title-holder's children and gave them good connections at court, but it weakened their local political strength. A title-holder could only marry for rank if he were sure of support from his own local subjects. Such support was achieved partly by the breadth of his kinship connections with his mother and his mother's mother, and partly by his personal ability to lead, cajole and coerce. If a title-holder was politically insecure he had to relinquish the possibility of having a high-ranking heir - 49 in favour of marrying local women whose villagers would support him and his sons. Polygyny permitted a man to make both types of marriage. The crucial decision was the choice of the new incumbent: should he be of high rank or should he be a man with strong local support? Rank, age, personal ability and the strength of the mother's political kāinga were the critical factors. The choice was not prescribed. It was up to the close relatives of the deceased title-holder to decide which of his brothers or sons would best further their interests. There were several cases before the Constitution of succession of a title to a daughter's son rather than to a son or brother in order to secure an heir of high rank of particular ability or power. There are even more cases in which the ha'a position of a title was changed, the title being drawn into a different ha'a because of marriage. I have discussed several such cases previously, especially those of the Vaha'i Lango'ia; Makahokovalu as heir to the title of Tu'ilakepa; the Fakafanua title; Mohulamufu'amotu; and the title Mā'atu (Bott in press).

Unsuccessful lines were those that pursued marriage for rank or marriage for local support to the exclusion of the other principle. If there were too many marriages and choices of heir to improve rank, the title ceased to have local support and the title-holder often went to live in the Tu'i Tonga's court in the sacred capital of Mu'a, as in the case of Tu'i Kanokupolu Tupoulahi and his descendants. Since the title-holder had little to offer at court in the form of political power and since at court he was only one man of high rank among many, the line ceased to attract offers in the exchange of rank for political power. The title was often forgotten once the title-holder had severed his connection with his local subjects by going to live at court. If, on the other hand, a line of title-holders married and chose their heirs only to secure local political support, as in the case of Ngalumoetutulu's patrilineal descendants in Ha'apai, their tie with the court was weakened, the rank of the leaders gradually declined and the women of the line eventually ceased to be an attractive marriage proposition for the daughters of either aristocratic or local chiefs. Eventually the exact origin of the ancestor and his title might be forgotten. The holders of such titles were often lax in fulfilling their obligations to their king. Eventually a new aristocrat would arrive from the court of the king and the process of exchanging rank for political support would begin afresh.

The same two commodities, power and rank, entered into choice of marriage partners for the women of a patriline. One example is provided by the marriages of Fusipala, daughter of the fourth Tu'i Kanokupolu, - 50 Mataeleha'amea (Figure 7). Fusipala was at first happily married, so the story goes, to a man of high rank, Tongatangakitaulupekifolaha, son of Tu'ihoua, son of Tu'i Tonga 'Uluakimata, by whom she had a son called Fuatakifolaha. At that time a man called Fisilaumāli was living in eastern Tongatapu where he held a large piece of land. He was a typical “big man”, very hard-working and generous so that he had attracted a large and strong kāinga. Tu'i Kanokupolu Mataeleha'amea did not have any supporters in the eastern part of Tongatapu, which was where the Tu'i Tonga and his various derivative title-holders lived. So Fusipala's brothers kidnapped their sister, forced her to leave her husband and baby and took her off to marry Fisilaumāli. She had a son with Fisilaumāli whose name was Lekaumoana. He was the founding ancestor of the Tu'ipelehake title.

In her first marriage Fusipala was of lower rank than her husband because he was a son of the Tu'i Tonga and she was only the daughter of the hau, so her father and brothers had to give food and support to her first husband according to the usual rule that a woman's brothers support her and her child. But the rank of her second husband, Fisilaumāli, was much lower than Fusipala's rank, because he was only a hardworking capable leader without a title, whereas she was the daughter of a king. The usual role of support by the brothers was reversed. Fisilaumāli - 51 supported the Tu'i Kanokupolu and became his ally. Note, in passing, that the fact that Fusipala had higher rank than her brothers because of being their sister did not stop them from coercing her into doing what they considered to be politically advantageous.

Later marriages of Fisilaumāli's descendants raised the personal rank of the title-holder and then, several generations later, a descendant confirmed the connection of the title with the Tu'i Kanokupolu by a crucial marriage to Sālote Pilolevu, daughter of Tu'i Kanokupolu Tāufa'āhau. The title is now held by the royal family, the present incumbent being a younger brother of the present king.

Fusipala's two marriages, one to an aristocrat and the other to a powerful local leader, are typical of the use the Kanokupolu chiefs made of their women for making alliances. In the first type of marriage the woman married a chief of high rank; the Tu'i Tonga, or the Tu'iha'ateiho, or another very high-ranking aristocrat. In this case the Tu'i Kanokupolu was put in the position of having to use his immense resources of food and men to support the 'eiki fakanofo, the Tu'i Tonga or the Tu'iha'ateiho. But it was a statement of his very special position and power, a statement which said that his resources were so great that he could afford to give lavish support to his sister and her aristocratic husband. In effect, it was a form of conspicuous display. But, in the fashion I shall describe below, what the Tu'i Kanokupolu lost in power and food, he gained indirectly in rank.

In the second type of marriage the Kanokupolu woman ranked so much higher than her husband that the usual rule of brothers supporting their sister and their sister's children was waived and the leader whom the Tu'i Kanokupolu's daughter or sister had married lent his support to the Tu'i Kanokupolu.

Of 37 marriages by daughters of the Tu'i Kanokupolu that I counted in the genealogies (not an exhaustive count), 14 were of the first type and 14 of the second, leaving only nine marriages in which the Tu'i Kanokupolu's group had to support another group without gaining much political advantage from it.

I have already described the process by which local titles used to get swallowed up or forgotten when a relatively high-ranking leader arrived from the central court and married the daughters of the local chief. A similar process used to take place between senior and junior lines in which the genealogically senior and higher ranking line of title-holders got superseded by the junior line whose leaders were more powerful, though in this case the junior line was of lower rank. As usual it was - 52 accomplished by marriage. There are several examples:

Marriage of Kaloafūtonga, daughter of Tu'i Ha'atakalaua Vaea, to Tu'i Kanokupolu Mataeleha'amea (the fourth Tu'i Kanokupolu) (Figure 8).

The Tu'i Ha'atakalaua title was genealogically senior to Tu'i Kanokupolu title and was therefore of higher rank. Originally the title was more powerful as well. In this marriage, therefore, the daughter of a higher ranking king married a lower ranking king. It is possible that this marriage of Kaloafūtonga to Tu'i Kanokupolu Mataeleha'amea was one of those in which the rank of the woman was considered to be so high that the usual role of wife-giving group as supporters to the wife-taking group would be reversed. There is no way of knowing how this particular marriage was defined at the time, but certainly the balance of power between the Tu'i Ha'atakalaua and the Tu'i Kanokupolu was very different after it. After this marriage the succession of Tu'i Ha'atakalaua became irregular, the Tu'i Kanokupolu took the place of Tu'i Ha'atakalaua as the secular ruler of the kingdom, and the Tu'i Kanokupolu became the giver of the chief wife, the Moheofo, to the Tu'i - 53 Tonga. Mataeleha'amea was the first Tu'i Kanokupolu who successfully sent a daughter to the Tu'i Tonga as moheofo and henceforth the Tu'i Kanokupolu succeeded the Tu'i Ha'atakalaua as “Ha'a Moheofo”. The Tu'i Kanokupolu also superseded the Tu'i Ha'atakalaua as wife receiver from the Fale Fisi.

Marriage of 'Ulukilupetea and Sisifā to Tu'i Kanokupolu Tuku'aho

'Ulukilupetea and Sisifā were daughters of Ngalumoetutulu, eldest son of the fifth Tu'i Kanokupolu, Ma'afu-'o-tu'itonga. As eldest son Ngalumoetutulu was senior and higher in rank than Tupoulahi, Maealiuaki and Mumui, but his mother was an ordinary woman of Ha'apai. Ngalumoetutulu also married an ordinary woman of Ha'apai, Siu'ulua, daughter of Malupō of 'Uiha. 'Ulukilupetea and Sisifā were her children. They both married Tu'i Kanokupolu Tuku'aho, this being one of the rare instances of marriage between parallel cousins. (Ngalumoetutulu and Mumui, Tu'i Kanokupolu Tuku'aho's father, were half-brothers, uho tau.) This marriage put Po'oi, the brother of 'Ulukilupetea and Sisifā, in the inferior wife-giving, support-giving position which contradicted his genealogical status as the elder and senior line. According to Queen Sālote, Po'oi was kept busy looking after his sisters because they - 54 made such illustrious marriages and he made such unimportant marriages himself that he and his line soon became “not well-known”.

Marriage of Tupou'ahome'e, daughter of Tu'i Kanokupolu Tupoulahisi'i and Tamahā Lātūfuipeka, to Tu'i Kanokupolu Tupouto'a (Figure 10).

Tupoulahisi'i was descended from Tu'i Kanokupolu Tupoulahi and was therefore senior in rank to Tupouto'a, because Tupoulahi was the elder brother and Mumui was the youngest brother. Tupouto'a, however, was a chief of great power and strength partly because of the strength of his father Tuku'aho as a leader and partly because of the support of his mother's people in Ha'apai. (Tupouto'a's mother was 'Ulukilupetea; Figure 6.)

Tupou'ahome'e was of extraordinarily high rank not only because her father was a Tu'i Kanokupolu of the genealogical senior line, but even - 55 more because her mother was Lātūfuipeka, the Tamahā, the person of highest rank in her day in all Tonga. The daughter of Tu'i Kanokupolu Tupouto'a and Tupou'ahome'e was called Halaevalu Mata'aho. She was of very high rank and married Tu'iha'ateiho Afi'afolaha and then Tu'i Tonga Laufilitonga. The marriage of Tupou'ahome'e to Tu'i Kanokupolu Tupouto'a put Tupoulahisi'i in the inferior support-giving position and hastened the decline of his line which was already suffering from lack of a firm local base.

Marriage of Ulakai, son of Tu'i Kanokupolu Tuku'aho to Tupoukolotolu, daughter of Tu'i Kanokupolu Tupoulahisi'i

Ulakai was a half-brother of Tu'i Kanokupolu Tupouto'a, having the same father but a different mother. Tupoukolotolu was a half-sister of Tupou'ahome'e (same father, different mother). The same argument applies as in the case of Tupouto'a and Tupou'ahome'e.

Marriage between Tu'ifuamatāpule, a son of Tu'i Kanokupolu Mumui, and 'Atuhakautapu, daughter of Tu'i Kanokupolu Tu'ihalafatai

Again a man from a junior powerful line put the senior line of higher rank in the wife-giving, support-giving inferior position.

Thus, marriage was used to “upstage” senior lines of higher rank but lesser power and to hasten their decline. If a line was to withstand this sort of depletion of its rank and resources, its men needed to make marriages with women of powerful and/or high-ranking groups which would restore the balance. But the original Tu'i Ha'atakalaua line and the senior Kanokupolu lines were already on the decline.

The exchange of power for rank in the circulating connubium

Unlike the relationships between titles, relationships between groups established by marriage did not outlast the lifetimes of the children of the marriage. If the relationship between two groups that had been established by a marriage was to be continued, another marriage of the same sort had to be contracted. Marriage of a man to his mother's brother's daughter (called kitetama) was such a marriage, for it continued the flow of food and political support in the same direction as that established by the mother's marriage.

This is a different interpretation of matrilateral cross-cousin marriage from that of Kaeppler, who describes it as occasionally practised (1971: 192-3),

Kaeppler's explanation fits the marriage she describes between Tu'ipelehake Fatafehi Toutaitokotaha (son of Tu'ipelehake Filiaipulotu and Sālote Pilolevu, daughter of King Tāufa'āhau Tupou I) and Fusipala Tauki'onetuku (daughter of Tevita 'Unga, son of Tu'i Kanokupolu Tāufa'āhau Tupou I) but it does not fit the many other instances of matrilateral cross-cousin marriage in the genealogies.

In the case of the moheofo, the great wife of the Tu'i Tonga, who was the eldest daughter of the secular king, this marriage was prescribed and institutionalised; it was repeated generation after generation in such a fashion that the ritual supremacy of the Tu'i Tonga was constantly reaffirmed even though the secular king was more powerful than the Tu'i Tonga. Kinship in other words was used to reinforce genealogical seniority (Figure 1).

There was a similar prescribed marriage to the mother's brother's daughter in the case of the Tu'i Tonga Fefine, the sister of the Tu'i Tonga, who married the Tu'ilakepa or the Tu'iha'ateiho, the major chiefs of the “outsider” ha'a, the Fale Fisi (see above). Thus, the Tu'i Kanokupolu were wife-givers to the Tu'i Tonga, and the Tu'i Tonga were wife-givers to the Fale Fisi. Only the eldest daughter of the Tu'i Kanokupolu had to marry the Tu'i Tonga, however, and only the Tu'i Tonga Fefine had to marry the Tu'ilakepa or the Tu'iha'ateiho. Other daughters of the Tu'i Kanokupolu and the Tu'i Tonga made many other types of marriage and even the moheofo could marry other men once they had produced heirs for their respective chiefs.

As noted above, marriage in Tonga was usually asymmetrical, especially mother's brother's daughter's marriage. There was no explicit expectation of an immediate reciprocal return when a group gave a woman in marriage, no bride price and only occasional wife exchange. Righting the balance had to be done by the chief of the group which was being depleted by the marriage of its sisters and daughters. In such a system there was pressure to “marry in a circle”, that is, for Group A to give wives to Group B who gave wives to C who in turn gave wives to A, to form a “circulating connubium” (Leach 1961). In Tonga it took the form of the eldest daughter of the Tu'i Kanokupolu marrying the Tu'i Tonga, the elder sister of the Tu'i Tonga marrying the Tu'ilakepa or the Tu'iha'ateiho and the Tamahā marrying the Tu'i Kanokupolu (Biersack 1974).

All three appointed Tamahā (Tu'imala, Lātūfuipeka and 'Amelia Fakahiku'ouiha), married Tu'i Kanokupolu, as did three other Fale Fisi - 57 women of high rank (Lātūtama, daughter of Tu'iha'ateiho Fakatakatu'u, Lupemeitakui, daughter of Tu'iha'ateiho Tungīmana'īa, and Lupepau'u, daughter of Makamālohi). In these marriages, the usual rule that the woman's group must give support and food to the group she marries into, was waived. The gift these very high-ranking women brought to the Tu'i Kanokupolu's group was their very high rank which raised the rank of the children they had with the Tu'i Kanokupolu.

Unlike the marriages of the moheofo and the Tu'i Tonga Fefine, however, the marriage of Fale Fisi women to the Tu'i Kanokupolu was not prescribed and institutionalised. Further, the circulating connubium was not part of Tongan ideology. In this respect it differed from societies that prescribe asymmetrical cross-cousin marriage, such as the Kachin, who present the circulating connubium as their own model of how their social system works, although the empirical facts do not entirely fit the model (Leach 1961). The Tongan conceptual model was the charter of titles and ha'a. In my view, the circulating connubium in Tonga was an empirical result of the two prescribed marriages of the moheofo and the Tu'i Tonga Fefine combined with the empirical pressure to complete the circle. In this circle political power was eventually exchanged for rank. Along with his daughter, the Tu'i Kanokupolu gave to the Tu'i Tonga power in the form of political support, protection and food. Along with the Tu'i Tonga Fefine, the Tu'i Tonga gave high rank and a certain amount of political protection to the Fale Fisi chiefs. In the person of their daughters, the Fale Fisi chiefs gave high rank to the Tu'i Kanokupolu. Political power, in other words, was eventually converted into high rank.

The circulating connubium was not a fixed one-directional system of marriage. Some Fale Fisi women were married to the Tu'i Tonga, and some Kanokupolu women were married to the Tu'iha'ateiho. (Thus Fusipala Tauki'onetuku, daughter of Tu'i Kanokupolu Tuku'aho, married Tu'iha'ateiho Fā'otusia; Halaevalu Mata'aho, daughter of Tu'i Kanokupolu Tupouto'a, married Tu'iha'ateiho Afi'afolaha.) In these - 58 marriages it is possible that the Tu'iha'ateiho was being treated as if he were the sacred king, a sort of substitute for the declining Tu'i Tonga. In fact, Halaevalu Mata'aho should have been moheofo to Tu'i Tonga Laufilitonga, whom she married after having had a child with Afi'afolaha. She was never considered a real moheofo because she was not a virgin when she went to the Tu'i Tonga. Hence her half-brother, Tu'i Kanokupolu Tāufa'āhau, had a suitable pretext for claiming that her son, Kalaniuvalu, would not become Tu'i Tonga.

In summary, relations between local groups were governed by the membership of their leaders in two systems, a system of titles and ha'a and a network of kinship ties created by marriage. Titles and ha'a formed a relatively stable and permanent charter of genealogical seniority which regulated the flow of tribute and was supposed to define the relative authority of titles. Marriage and the system of rank allowed for shifting alliances and rapid fluctuations of power and rank. Three repeated empirical patterns have been discussed: successful lines were those whose incumbents married first for local political support and then for a balance between marriage for local political support and marriage for rank; failed lines were those who married too often for local support or too often for high rank; marriage was used to “upstage” genealogical senior lines; matrilateral cross-cousin marriage was a common feature leading, in the case of the Tu'i Kanokupolu, the Tu'i Tonga, and the Tu'iha'ateiho, to a circulating connubium in which locally based political power was converted into high rank.

I will now consider some of the changes that have taken place in this system in the modern period.

Political, economic, and religious changes in the 19th and 20th centuries have modified the traditional system of stratification so that, instead of independent but inter-penetrating systems of rank, locally based political power and formal political authority in the form of titles and ha'a, a system of social classes of the Western type has gradually developed. There is now a ruling elite of aristocratic nobles, a set of nobles of much lower rank who play little part in government, a variegated middle class and a peasantry. Even now the working of this class structure is to some extent countered by the principles of rank, by the comparative abundance of the subsistence economy and by Tongan propensity for operating somewhat contradictory principles of social differentiation by expressing them in separate social contexts (Bott 1972). But it seems inevitable that the increasing land shortage, coupled with overseas employment, dependence on money and desire for Western con- - 59 sumer goods, will enhance the development of a Western type of class structure and will further weaken the traditional system of stratification.

I will not attempt to describe in detail the events in the 19th and 20th centuries that have led to the development of present-day Tongan social structure. These developments have been outlined by Noel Rutherford and others (Rutherford 1971, 1977; Fusitu'a and Rutherford 1977). I want to focus on those aspects of political and social change that have especially affected the system of stratification.

In 1839, 1850, 1862, 1875 and 1881, King George Tāufa'āhau Tupou I, Tu'i Kanokupolu, issued various proclamations intended to establish the rule of law, weaken the power of the chiefs, and emancipate the people. With the help of Shirley Baker, he was attempting to control the chiefs, many of whom were opposing him, and to make Tonga into a civilised nation so that it would be recognised by European powers. He wanted to keep Tonga independent, an aim in which he was largely successful. Tonga became a British protected state in 1901, but this foreign control was removed in 1970. Tupou's measures, however, had certain unexpected consequences.

In 1875 and 1881 he made 30 important chiefs into “nobles” and assigned a hereditary estate (tofi'a) to each. Six matāpule were also given estates though they were not made nobles. All these estate-holders were supposed to grant hereditary leases of eight and one-quarter acres to each adult male over 16 years of age in return for an annual rent and the Government was to pay the estate-holder. Thus, the nobles were made dependent on the king, instead of being supported by their own political kāinga. All these noble titles and the matāpule titles who were granted estates were to be inherited according to fixed British rules of succession instead of by the flexible rules of the traditional Tongan system. Thus, the system of titles was frozen, made inflexible. These measures led Tupou I into a troubled period of conflict in the 1880s with certain chiefs and title-holders who had not been made nobles and who realised that they would soon lose all their traditional political power and social standing. They had no tofi'a, their villages were considered Government land and they had no role in the new legislative assembly.

Tupou's aim was to weaken the chiefs and free the common people from bondage, but it worked out rather the other way round. The nobles became independent of their villagers but retained a certain measure of control over them. It did not matter whether a noble led his local kāinga - 60 well, or badly, or even not at all; he was assured of his title, income and an important role in the newly established legislative assembly. The nobles were thus no longer dependent either on their own kāinga or on the political kāinga of their mother or their mother's mother. But most nobles retained a considerable measure of economic control over their former subjects, because the traditional rules of land usage were continued. No allotments were actually registered until 1915 (Maude 1971) and even at the time of our study in 1958-60, much land was being used that had not been formally registered. In spite of this reversal of Tupou's aims, however, there has been a very great change in the arbitrary exercise of power by chiefs. By the time of our visit, low-ranking chiefs were treated with scant respect. Nobles of high rank were treated with great respect by their villagers and were given gifts of food and even money, but even these nobles were asking, not commanding.

At first Tupou I paid little attention to the legislative assembly he had established in the Constitution of 1875, but gradually, especially in the reign of Queen Sālote, the rule of law was established. By European standards the legislative assembly is heavily weighted with nobles. It consists of cabinet, seven nobles' representatives, and seven people's representatives elected by general franchise. Cabinet and the people's representatives are usually opposed to each other, so that the nobles' representatives often hold the balance of power. The cabinet is mainly composed of nobles. In 1958 six of its eight members were aristocratic nobles, one was a commoner and one a European. There is also a small civil service which in 1958-60 consisted mainly of Europeans, so that the ministers were permanent but the civil servants rotated. This arrangement also contributed to the power of the ministers.

In the modern system there is no autonomous village government. Town and District officers explain the law and see that it is enforced. Town and District officers were appointed by the Government in 1958-60, though later these offices were made elective. The nobles, in other words, have now no formal political function in their villages.

The local political power of the nobles, which was the basis of their power in the traditional system, has thus been taken over by the Central Government. The Central Government is of course dominated by nobles, especially by the aristocratic nobles, but their local roots have been weakened.

In effect, the noble title-holders now belong to two systems, a modern system of government and a traditional system of titles and ha'a. Traditional activities are now largely ceremonial, being concerned with the installation and funeral ceremony of the sovereign and of title-holders, both of which involve kava ceremonies. These ceremonies and the system - 61 of titles and ha'a according to which they are organised, have an important function in symbolising and expressing Tongan culture and identity. They also involve considerable quantities of food and therefore of economic and social effort, but the ceremonies are no longer directly concerned with power and authority.

Tonga is still basically a subsistence economy of small peasant farmers, though production of cash crops began in the 1860s, as soon, according to Rutherford (1971), as people felt secure enough in their land tenure for it to be worth their while to produce coconuts for export. The population has grown rapidly and there is now not land enough for every adult male to have a registered allotment. Paradoxically, registration of individual allotments became customary only when the land shortage became acute. Economists usually point out that the system of land tenure, although it has prevented the alienation of land to Europeans, is so rigid that it makes efficient farming difficult (Maude 1971). With the exception of tourism and some small-scale processing of coconut, commercial ventures have not been successful in Tonga (Baker 1977). The country is still very dependent on the export of coconuts and bananas. Because of the limited opportunity for work in Tonga and the increasing shortage of land, large numbers of Tongans now work abroad in New Zealand and Australia and send money home.

Money and consumer goods are becoming increasingly important as land gets scarcer and money flows in from the overseas workers. Even at the time of our visit there was considerable conflict, among the more affluent families, over how money should be used: should it be confined to one's own elementary family, especially for the education of one's children, or should it be used to support a large household of dependents in the traditional fashion of a man who was building up a kāinga?

Tongans have been keen to be educated ever since the first mission schools were established in the 19th century. At the present time schools are run both by the churches and by the Government. All children attend school and can achieve the Tongan Lower and Higher Leaving Certificate. One Government school prepares its pupils for the New Zealand School Certificate. Increasing numbers of students are sent to school and for higher education to New Zealand and Australia. The Government and the churches provide scholarships for many students in the expectation that they will return to Tonga. Others are sent by parents. There is a special scheme for the education of heirs to noble titles, the aim being to - 62 improve their abilities as leaders and their chances of securing important Government positions.

In addition to its other functions, education now provides a new way of achieving status and entry to government, though there are far more educated Tongans than there are government positions available for them. In a sense, education has taken the place of building up a political kāinga as the first step in acquiring political power. Marked educational achievement may entitle one to an important Government position and, in a few cases, highly successful commoners of this sort have married aristocratic women.

Wesleyan missionaries arrived in 1826 and rapidly converted many important leaders, among them the future king George Tāufa'āhau Tupou I. His political opponents remained pagan or later on became Catholics. In 1885 the Wesleyan Church split over the issue of sending funds to Australia. The breach was partially healed by Queen Sālote in the 1920s, but there is still a small independent Methodist Church as well as the established church. There are also Catholics, Seventh Day Adventists, a small Church of England and, since the 1939-45 war, Mormons.

The churches provide a focus for many village and national activities. Since the mid-19th century they have also provided a new form of achieved status, for many gifted and highly educated individuals seek positions in the church rather than in the Government. As Marcus points out (1975a), there is a tendency towards intermarriage among church families and other educated families, but there is no systematic tendency towards marriage with the ruling elite. Marcus also suggests that the church has become the focus of conspicuous competitive gift-giving with no expectation of reciprocal return of the sort that formerly was focused on chiefs. The church, in other words, has taken over something of the mystique that the chiefs used to have.

The basic principles of rank have changed very little; sisters still rank higher than brothers, older siblings rank higher than younger siblings. But the church's success in eliminating polygyny and discouraging serial monogamy, combined with Tupou I's changes in the system of titles, has changed the role of rank in political manoeuvring.

Conflict between siblings of the same father but different mothers used to be endemic. Nowadays most sets of siblings, especially in the case of noble title-holders, are likely to have the same mother as well as the same father and, since the rules of succession and inheritance are fixed - 63 and rigid, there is less material basis for rivalry than there was in the traditional system. Many nobles have illegitimate children, but these children are barred by law from succeeding to titles and, although they are recognised and looked after by the family of the noble, their social position is not very different from that of ordinary commoners.

It is no longer common for women of high rank to have several husbands and lovers of high rank leaving a trail of high-ranking children behind them. The last woman who did this on the grand scale was Tupoumoheofo, who died in the influenza epidemic of 1918. Tupoumoheofo's mother was Lavinia Veiongo, a very great aristocrat, daughter of the last Tu'i Tonga Laufilitonga, and Halaevalu Mata'aho, who was a half-sister of Tupou I (Figure 9). Tupoumoheofo's father was 'Isileli Tupou, an illegitimate son of Tupou I by a low-ranking mother (Figure 12). This love affair was one of the few instances of a liaison between a man and his father's sister's daughter. Nearly all cousin marriages and liaisons among chiefs occurred between the man and his matrilateral cross-cousin.

Tupoumoheofo had children with three very important chiefs. First, she had a child, a daughter called Vaohoi, with Siale'ataongo, who was the son of Ma'afu “Fiji”, who was the son of Tu'i Kanokupolu 'Aleamotu'a. Vaohoi married the Tongan chief Veikune and had three children, one of whom is Heu'ifanga, the mother of the present Queen, Halaevalu Mata'aho. Tupoumoheofo also had a second child, called Sālote Fisi, with Tui Cakau of Fiji. She had yet another child, Vilai, whose father was Tāufa'āhau Tupou II, the father of Queen Sālote. According to at least one of the genealogical books ('Etueni Tupou, p.41) Tupoumoheofo had three other children with three other men.

As polygyny by men and serial monogamy by high-ranking women died out, the enormous complexity of alliances between political kāinga has been reduced and the discrepancies between ha'a and rank have become less marked. There are now very few men of high rank who are not members of the ruling elite. In the old system, even when Queen Sālote was a girl, there were several men of high rank who had no formal title or position in government but had enormous social importance because of their rank. Such a man was Uili Kalaniuvalu, or rather he would have been such a man had he not died young. The Queen remembered that her father and Losaline Fatafehi (a daughter of Tu'i Tonga Laufilitonga) used to have long discussions about who had the highest rank in all Tonga, Uili Kalaniuvalu or Tungī, Queen Sālote's future husband. Uili Kalaniuvalu had no title, but he was descended from Tamahā Lātūfuipeka, Tu'i Tonga Fefine Fatafehi Lapaha, Tu'i Tonga Tu'ipulotu-'i-Langitu'oteau, Tu'i Kanokupolu Tupoulahi, Tu'i

- 64

Kanokupolu and Tu'i Ha'atakalaua Maealiuaki, Tu'i Kanokupolu Mumui, Fīnau 'Ulukālala, etc. (Losaline argued convincingly, however, that Tungī's rank was higher than that of Uili Kalaniuvalu.)

Vilai, Queen Sālote's half-brother, was the last of these great aristocrats without title. As described above, he was an illegitimate son of the Queen's father Tāufa'āhau Tupou II, and the somewhat promiscuous Tupoumoheofo described above. Queen Sālote said that the people of the village of Pelehake wanted Vilai to become their Tu'ipelehake, but he could not succeed because he was illegitimate. He was brought into the system by marriage to Tupouseini, eldest daughter of a Ha'a Havea title-holder called Vaea, an important title held by a capable man who was not aristocratic. Vaea had no sons, so that when he died, - 65 Vilai's son with Tupouseini succeeded to the title, according to the new rules of the Constitution. Vilai was also brought into the government of the system by being A.D.C. to the Queen for many years.

In the new system aristocratic women tend, as in the traditional system, to marry politically powerful men. The power is now concentrated in the Central Government which is dominated by educated nobles so that aristocratic women tend to marry educated nobles. Their aristocratic daughters, similarly, tend to marry other educated nobles, and the aristocratic nobles' sons tend to marry daughters of other aristocratic nobles. Hence a certain degree of class endogamy has developed.

In the old system, some of the sons of aristocratic parents, especially the younger sons, would have gone to outlying islands to marry the daughters of local powerful chiefs, but in the new system there is no inducement for men to marry down in rank but up in power in this way, because power as well as rank is increasingly concentrated in the elite nobles who dominate the Central Government.

Education as well as noble origin can help to secure a Government position. In a sense education is the modern substitute for acquiring political power by building up a strong local kāinga. Marriage is now often used to combine education and high rank. A well-educated commoner is quite likely to marry an aristocratic girl, which eventually absorbs him and his descendants into the ruling elite. But nowadays, unlike the situation in the traditional system, there is no pressure in the other direction. A high-ranking man does not normally marry a highly educated girl of low rank, for there is no way he can use such a girl's education to further his political aspirations. Hence the contrary tendencies in the traditional system, which tended to maintain the heterogeneous bases of stratification, are tending to be replaced by a system in which social and economic forces are much more likely to work in one direction.

Table 1 shows the modern tendencies for women of high and middling rank to marry either nobles or well-educated and successful commoners.

Thus, of 44 women of high or middling rank who were alive in 1958, 22 were married to nobles or noble heirs, 13 were married to successful commoners, one had an illegitimate child with an aristocratic man, and eight had “pleased themselves”, that is, they had not used their rank in an exchange of rank for noble position or education.

Table 2 shows the marriages of noble men and matāpule ma'u tofi'a (matāpule holding estates) according to rank. Thus, high-ranking nobles tend to marry aristocratic women; low-ranking nobles tend to marry low-ranking women; six low-ranking nobles were married to women of high or middle rank and this is particularly likely to happen if the noble concerned holds a Government position.

- 66

Marriages of women of high and middling rank (1958-1960)

Marriage to or liaison with:	Women of high rank	Women of middling rank	Totals
Nobles or noble heirs of high or middling rank	6	7	13
Nobles of low rank	7	2	9
Aristocratic man without title		1	1
Educated and successful commoners	8	5	13
Undistinguished commoners	3	2	5
Europeans	2	1	3

Rank of noble title or tofi'a holder in relation to rank of wife (1958-60)

	Rank of wife
Rank of husband	High or middling	Low
High or middling	10	3
Low	6	16

Hence power and rank tend increasingly to coincide. A ruling elite of educated nobles of high rank has developed in place of the former system in which rank and power were virtually independent variables which were constantly being converted into each other.

A somewhat similar coalescence has taken place between sacred and secular titles, in that the split between the sacred title of the Tu'i Tonga and the secular title of the Tu'i Kanokupolu no longer exists. The present kingship combines both aspects. In part, this has happened because the rigid rules of succession of the Constitution allow one man to hold two or more titles, which did not occur in the traditional system. When the last Tu'i Tonga died he bequeathed his rights as sacred king to Tupou I, who later abolished the title of Tu'i Tonga. Still later, the title Kalaniuvalu became recognised as the main title descended from the Tu'i Tonga, but it is no longer considered sacred or higher in rank than the - 67 title of Tu'i Kanokupolu. Tupou II inherited the title of Tu'i Kanokupolu through his mother and the title Tu'ipelehake from his father. The title of Tu'i Ha'atakalaua became defunct at the end of the 18th century but the title of Tungī became the representative of the old Tu'i Ha'atakalaua title, as Tungī was a direct patrilineal descendant of the last Tu'i Ha'atakalaua (Figure 6). The marriage of Tungī Mailefihi and Tu'i Kanokupolu Sālote Pilolevu brought the two titles together. Thus, the royal family now holds the titles Tu'i Kanokupolu, Tu'ipelehake, and Tungī. It also acquired the title Tupouto'a because there were no heirs to the title and the royal family were next of kin. The same thing happened to the titles of Mā'atu, Fīnau 'Ulukālala, and Ata. Thus, seven major titles can now be held or appointed by the royal family, though the titles of Tungī, Fīnau 'Ulukālala and Ata have not been appointed at the present time (1980).

The system of sacred titles of higher rank and secular titles of lower rank is thus being replaced by one in which there is a coincidence of sacred and secular, of rank and power. The circulating connubium and matrilateral cross-cousin marriage are being replaced by a system in which men marry women of equal or higher rank, and there is little inducement, other than romantic love or rebelliousness, for aristocratic men to marry down.

At the top of the modern class structure of 1958-60 was a group of 11 nobles of high rank. They tended to live in the capital, Nuku'alofa, and held positions in the Government. Three more nobles were of slightly lower rank; six others were of low personal rank but were married to women of high rank. The remaining 16 nobles and matāpule with estates were of low personal rank and were married to women of low personal rank. They tended to live in their villages and were unlikely to hold Government posts. Table 3 shows the residence of these various types of nobles and matāpule with estates, and Table 4 shows their holding of positions in the Government.

The landholding title-holders were thus separated into two polarised groups, a ruling elite of high-ranking nobles and a low-ranking set of nobles living in their villages. Marcus (1975a) uses the term “gentry” nobles for the last group. In Tonga there have always been low-ranking title-holders. What is new is the absence of marriages by men of the high-ranking group to women of the low-ranking group, for, as described above, such marriages were an intrinsic part of the traditional system but are no longer politically useful.

- 68

Rank and residence of nobles and matāpule with estates, 1958-60

		Living mainly in:
	In capital	Own village near capital	Own village but frequent visits to capital	Own village and few visits to capital
Nobles of high rank	8	1	1	1
Nobles of middling rank	2	1
Nobles of low rank married to women of high or middling rank	1	1	1	3
Nobles of low rank married to women of low rank			3	13

Rank and government posts held by nobles and matāpule with estates, 1958-60

	Government post held	No position in Government
Nobles of high rank	10	1
Nobles of middling rank	1	2
Nobles of low rank married to women of high or middling rank	2	4
Nobles of low rank married to women of low rank	0	16

The ruling elite is thus composed of aristocratic title-holding families, periodically leavened by the addition of highly educated commoners who hold important positions in the Government and who marry aristocratic women. The ruling elite try to secure the best education for their children, who are expected to succeed them in occupying Government positions.

- 69

The gentry nobles usually live like ordinary villagers except that they sometimes demand and are shown a certain respect by their villagers because of the respect traditionally given to the title. They may have a certain power over those villagers who want tax allotments. Some non-aristocratic nobles have large personal estates which, if exploited efficiently for personal use, allow them to become individually wealthier than other villagers and permit more expensive overseas education of their children, which is the path towards higher occupational and Government status, if the children choose to follow that path. The title has a seat in the royal kava ceremony, and the villagers will contribute comparatively willingly to any national ceremony that brings recognition to their noble and hence to their village. But a noble who is not of high rank (sino'i 'eiki) is not shown any particular respect either inside or outside his village if he expects deference to his person rather than to his title.

Production for export, overseas work, and the changes in the system of titles and landholding have led to the development of a new middle class based partly on income but, even more, on education and occupation with a heavy bias towards the church, teaching, medicine, law and Government posts involving technical skill. In Tonga, money may be converted into having a large household of personal dependants which gives a man considerable prestige comparable to building up a kāinga in the traditional system, though it cannot so easily be used to convert power into rank as it could in the traditional system. Money may also be used to buy imported consumer goods and status symbols. But most important, it can be used to buy education abroad which is a prerequisite for achieving secure middle class status or for joining the governing elite.

Many of these middle class families have a long tradition of educational achievement and service to the community in some form or other, often through the church. Most of the families concerned hold a tax allotment as well as a salaried occupation, so that they can become wealthier than the ordinary villager. They tend to adopt European standards of housing and education for their children and often find themselves in conflict between their traditional obligation to relatives and educational aspirations for their children. The members of this developing middle class do not appear to think of themselves as a distinctive group. The ties of kinship and infinite gradations of personal rank link them both with the governing elite and the ordinary villagers. Many of them are “well-connected”, meaning that they had an ancestor of high rank somewhere in their genealogy and that they are fairly closely connected with people of high rank at the present time. A few have married - 70 aristocratic women. On certain issues they find themselves at one with the governing elite, whose educational sophistication is similar to their own. Well-educated men, especially those in Government positions, sometimes marry women of high rank so that the bonds of kinship blur and soften the division between the governing elite and the middle class.

Below the established middle class and the “moneyed” individuals who have large households and/or are educating their children for social mobility, there are the ordinary people who are content with the modest subsistence on a tax allotment supplemented by cash crops of copra and bananas and nowadays also by cash sent home by a relative working overseas. This group has always formed the great bulk of the population in both the traditional system and in the modern period. But they used to be the subjects and/or kin of local leaders and aristocrats and these bonds are now being weakened as the locus of power, authority, rank and occupational achievement becomes increasingly centred on the capital.

Further, land shortage is such that it is now impossible for every male over 16 to have a tax allotment. In 1958 there was concern about the tendency for landless young men to congregate in the capital looking for jobs, amusement and trouble. In the 1960s they began to work overseas, sending money home and thus easing the financial burden of their families. But it is as yet far from clear whether and how they will be reabsorbed by Tongan society when they return.

Transition from the traditional to the modern system of stratification has been greatly eased by the fact that there has always been discordance between the various modes of stratification in the traditional system and Tongans have developed great skill in allowing the co-existence of contradictory principles of stratification. One of the main devices for ensuring such co-existence is that each major mode of stratification had a particular social context in which it was expressed in relatively pure form. Thus, the system of titles and ha'a was expressed in the structuring of the kava ceremony into which kinship rank and personal power were not supposed to enter. Access to land and the giving of tribute were also organised according to title and ha'a. Kinship rank is the organising principle of funeral and marriage ceremonies in which title and political power are not supposed to matter.

New dimensions of stratification such as position in the Government have been added to the system on the same model. Being a noble gives - 71 one a marked initial advantage in the pursuit of a Government position, but the noble must also meet the educational and intellectual requirements if he is to secure an important Government post. Highly educated commoners are also able to achieve important Government positions. Less highly qualified nobles can become nobles' representatives and commoners can become people's representatives.

Although the immediate context of the Government is the day-to-day work, there are also ceremonial expressions of it. The most striking of these that we observed were the feasts provided for visiting Governors from Fiji or for Colonial Officers. On these occasions all cabinet ministers, regardless of personal rank or title, sat at the head table with the Queen. Depending on the occasion, the heads of the leading churches might also be seated at the head table. The British Consul was always included as well. The other tables were composed of a mixture of nobles, civil servants, town and district officers and Europeans. These seating arrangements were totally different from those of a royal kava ceremony, in which only title-holders, including those who were not nobles and the ceremonial attendants of title-holders (matāpule), could sit in the kava circle. It occasionally happens, today as formerly, that a man may have high position in one system and low position in another. At Government feasts an educated commoner of low rank who is a cabinet minister will be shown a deference and respect becoming to his Government position, but if he should venture into a formal kava ceremony, this same man without question or argument will be placed at the bottom of the circle.

In everyday life and at ceremonies, the various principles of stratification are often in conflict. A man may be wealthy, for example, but of low personal rank; or a very great aristocrat, the holder of an important title, may be technically on the low side (liongi) at the funeral of a woman of much lower rank than himself. (See above.) A man may be aristocratic and therefore deferred to without being wealthy, having a title, being well-educated, or having a Government position, though such men are now much rarer than they used to be.

Thus, in conditions of increasing economic stringency and the weakening of the local bases of power, a European type of class structure has developed although it is still countered by the system of rank to some extent and, even now, by the comparative abundance of the subsistence economy and by the system of overlapping, independent principles of social differentiation.

Traditionally Tonga had a complex system of stratification in which - 72 titles, power and personal rank were constantly being manipulated, exchanged and converted into one another. It was a system in which no two people had the same rank—hence radically different from a system of social or economic class.

Principles of authority and rank were (and are) simultaneously domestic and political. The father and the father's side of the family have secular authority and high rank. Both aspects were extended to the role of chief ('eiki), which had both a secular aspect of ruling and a sacred, tapu, aspect of high rank and mystical identity with the land, its people, and with the past. In certain cases these aspects were split; certain chiefs such as the Tu'i Tonga were designated as sacred and of high rank, whereas their immediately genealogically junior titles, the Tu'i Ha'a-takalaua and later the Tu'i Kanokupolu, were designated as secular, “working” chiefs, greater in power but lower in rank.

Within the family there is an inherent contradiction in that fathers and elder brothers have authority but sisters have higher rank than brothers, a contradiction that was frequently duplicated in the political realm in the lack of fit between power and rank.

Principles of kinship rank were also extended nationally in the sense that the kinship connections of the paramount chiefs were the political connections of the nation. The Tu'i Tonga was the sacred “father” of all Tonga. The Tu'i Ha'atakalaua and later the Tu'i Kanokupolu held the coercive, authoritarian aspects of the father role for the whole nation. The Tu'i Tonga's great wife, the Moheofo, was in a sense the “mother” of all the nation. The Tu'i Tonga's sister, the Tu'i Tonga Fefine, was mehekitangatama 'a mehekitanga (father's sister's child) to the whole nation. In the family the offspring of sisters have higher rank than the offspring of brothers. In the nation descent from the Tu'i Tonga Fefine and the Tamahā were the essential bases of high personal rank recognised at society-wide level. (father's sister) to the whole nation. The Tamahā, sacred child, of the Tu'i Tonga Fefine, was

The father's side of the family is 'eiki (high). The mother's side is tu'a (low). A man is supported by his mother's brother and by his mother's brother's children; he in turn must support and help his father's sister, his own sister, his sister's child. In the traditional political system these kinship usages were of great political importance, because what was kinship for chiefs was politics for their subjects. A chief and his group could expect support in food and fighting men if need be from the local groups of his mother, his mother's mother and his wives; from others as well but these were the main ones. He had in turn to give support to the groups into which his sister and his sister's daughter married. The system was thus asymmetrical. A chief's group did not expect an immediate recipro- - 73 cal return for the gift of their woman and her childbearing capacity, or for the gifts of food, fighting men, and sometimes even land that were given to the sister and her child.

The basic corporate group of the traditional political system was the political kāinga, consisting of a chief ('eiki) and his subjects, who were called his kin or kāinga. Many of them actually were his kin, but residence was the crucial criterion. If one lived in the chief's fonua (land, but it includes the people on the land), one belonged to his kāinga even if one could not trace any known kinship tie. I have suggested above that this use of kinship idiom to express a political and residential definition represents a process of transition from kin-based to territory-based political organisation.

There has been considerable controversy about whether Tonga had or did not have a system of patrilineal lineages. My own view is that it does least violence to the facts to define the kinship system as cognatic, though with implicit, unnamed, unconceptualised patrilineages. The basic corporate group was the political kāinga headed by a chiefly title-holder. In Tongan idiom, land and political authority reside in the title, rather than in the kin group of which the title is the head and focus.

Similarly, the ha'a is best regarded as a set of genealogically related titles with the political kāinga and the fānau (descendants of former holders of a given title through both men and women) of the respective titles as the groups mobilised by each title to fulfil its obligations to its king and senior title-holders. The ha'a was formerly of great political significance because it was the organising principle of political tribute. Its only importance now is ceremonial.

The external relations of territorial groups (political kāinga) were regulated by membership of their leaders in two systems: the ha'a and a network of kinship ties created by marriage.

The relations defined by title and ha'a were supposed to be perpetual. Obligations between titles were defined as the obligations that existed between the original founding ancestors of the titles. The Ha'a Ngata titles were originally “brothers”, for example, and in certain social contexts the present-day title-holders act like brothers regardless of the actual kinship relationship between the men who hold the titles. In Tongan idiom all holders of a title are identified. There is a mystical union between, for example, the Tu'iha'ateiho of today and the Tu'iha'ateiho of the 18th century. Similarly, if one's father held the title of, say, Fīnau 'Ulukālala, one would address all subsequent Fīnau 'Ulukālala as “father” regardless of personal kinship.

Even though titles and ha'a are no longer the basis of political definition and national organisation, the kava ceremony is conducted accord- - 74 ing to title and ha'a, not according to personal kinship and rank, so that this ceremony perpetuates the traditional forms. It has additional importance as a symbolic expression of Tongan identity.

Thus, titles and ha'a acted as an historical charter in much the same way as a model of lineages acts as an historical charter in a society organised according to unilineal descent. The assumption of unchanging perpetual titles and ha'a is, of course, a fiction. New titles emerged periodically and, similarly, others were forgotten and became defunct.

The ties between politico-local groups that were created by the marriages of their leaders as individuals were less long-lasting than the ha'a relationships. They lasted only a generation or two, but they were infinitely more complex than the ties defined by title and ha'a. Choice of wife was affected by her rank and by the political strength of her local group as well as by various personal factors. In theory, a woman's group was supposed to give to her husband's group not only the woman and her childbearing capacity but also food, general support and sometimes even land “to feed the child”. As usual in Tonga, however, the fulfilment of these obligations depended on rank. If the woman was of very high rank, the giving of her alone was enough. Sometimes the usual obligation was even reversed, so that the wife-receiving group became the supporters of the wife-giving group.

Succession was also much affected by considerations of rank and power. Because of polygyny there were usually many potential heirs, the choice being made at a meeting of patrilineal kin according to the various heirs' ability, rank, age and the strength of the political groups of their respective mothers. Normally a brother or son succeeded, but the title could go to a daughter's son if it was considered expedient for reasons of rank or power.

Looked at historically, certain patterns emerge in the manipulations of marriage, rank and power, that went on in the process of establishing new chiefs and titles and obliterating old ones. When an outlying local group became lax in fulfilling its ha'a- 75 of local low-ranking but powerful chiefs and marriages to women of high rank at court. Unsuccessful lines were those who married too often for local support, which eventually attracted another immigration by a king's son, or too often and too soon for rank, which drained the resources of the new title and its group and weakened their local support. obligations to the central court, it was customary for the secular ruler to send out a younger brother or younger son to “get hold of” the offending territory. He did not always succeed, but when he did, he accomplished it by using his existing kinship connections with the community (if any), his high rank as a king's son, which was respected in the “bush”, and military strength. He then married daughters of the local chief and one or other of his sons by the local chief's daughters would succeed to the old chief's position and title. The next step was to build up the strength of the new leader, which was done at first by marriage to local women whose groups would support the new line. Successful lines achieved a balance between marriage to daughters

Judicious use also had to be made of the marriage potential of the new line's women. At first it was politic to marry one's women to local chiefs of low rank but considerable resources, so that one could waive or even reverse the customary obligation to support one's sisters and their children. Later, once one's line was securely established, it became politic to marry one's sisters to chiefs of great power and/or high rank, so as to show that one's own resources were so great and rank so high that one could afford to give lavish support to one's sister and her children.

Once a line was successfully launched and developed, marriage was used to “upstage” genealogically senior lines of title-holders who in theory had higher rank because of their genealogical seniority, but who, in fact, were declining in power. By marrying their daughters, the powerful junior line put the senior chiefs in the inferior wife-giving, support-giving position and hastened their decline.

The relations established by marriage lasted only a generation or two so that if the respective groups wished to continue the type of relationship established by previous marriage, they had to contract other marriages of the same sort. Thus, for example, it was customary for the daughters of the secular king to marry the Tu'i Tonga, which put the more powerful secular king in the inferior wife-giving position and maintained the higher rank of the Tu'i Tonga. Similarly, the Tu'i Tonga Fefine always married a chief of the Fale Fisi, an arrangement which allowed her children to have higher personal rank than the Tu'i Tonga without threatening his title and his political position, since the Fale Fisi were perpetual “outsiders” because of their foreign origin. There was a certain pressure to “complete the circle”, that is, for the Fale Fisi women to marry the Tu'i Kanokupolu, thus creating a circulating connubium in which the great power and resources of the Tu'i Kanokupolu were eventually converted into high rank.

The network of ties between corporate groups created by kinship and marriage thus differed markedly from the relations defined by title and ha'a. The difference allowed rapprochement between the current facts of political power and the ideology of perpetual, unchanging titles. Taken together the two systems combined to provide for growth as well as continuity.

Political, economic, educational and religious developments in the - 76 19th and 20th centuries have changed the Tongan system of stratification so that it now resembles a system of social classes.

The Constitution of 1875 and related acts transferred much of the chiefs' power to the Central Government in the form of king, ministers and legislature and fundamentally altered the basis of political power. Some 36 of the many political leaders remained “nobles”, were granted estates, and were instructed to give permanent plots of land to their subjects. The chiefs who were not made nobles gradually lost their powers and their titles have tended to become “not well-known”. Nobles are no longer dependent on their local kāinga for political support, though the kāinga remain dependent on their chief for land, at least in practice, although this had not been the intention of the drafters of the Constitution. The Constitution also made rules of succession rigid and prevented succession by illegitimate children so that there was no more manoeuvring of choice of heir according to the exigencies of the immediate political situation. Polygyny was abolished and serial monogamy discouraged, so that the political complexities of rank and kinship have been considerably reduced.

The population has increased four- or fivefold since the 19th century and there is now a land shortage. Cash crops have been exported since the 1860s but cannot support the population entirely and schemes for diversification of the economy have had limited success. Overseas working has increased markedly together with increased use of money and demand for Western consumer goods.

Education has to some extent replaced the building up of a local kāinga as the first step in gaining political power. Some of the highly educated and successful commoners marry aristocratic girls and their offspring gradually get absorbed into the ruling elite. In contrast to the traditional system, however, there is no inducement under modern conditions, for aristocratic men to marry down in rank but up in power, for there is no way in which an aristocratic man can make use of his wife's education for political purposes. In the traditional system, in other words, there were situations in which it was advisable for an aristocratic man to ally his high rank with a woman's political power, or rather, with the political power of her local group. Equally, there were other situations when it was advisable for a politically powerful man to ally his power with a woman's high rank. In the modern system only the second choice is available. All forces, in other words, are tending to work in one direction. Hence the complex system of stratification of the traditional system, with its independent but interpenetrating and overlapping bases of stratification is changing into a class system in which rank, power in the form of access to the Government, economic resources and education - 77 are tending to coincide.

In 1958-60 the social classes consisted of the nobles, a heterogeneous middle class and peasants. The nobles were increasingly polarised into two subgroups, a ruling elite of aristocratic nobles who held the main Government positions, were highly educated and lived in or near the capital, and a group of “gentry nobles” who were of low personal rank, did not hold Government positions, were usually not highly educated, and lived in their villages. Their life-style was not very different from that of their villagers, though they were usually accorded a modicum of respect because of holding the title.

The middle class consisted of clergy, teachers, medical practitioners, lawyers, shopkeepers and Government employees of varying degrees of skill and education. Placing value on education was the common feature, but in other respects the middle class was heterogeneous and did not think of themselves as a power group.

The peasants have always been the largest group both traditionally and in the modern period. Socially they are less oppressed by their chiefs than they were in the traditional system, but economically they are now much more vulnerable.

Transformation to a class system is far from complete. It is encouraged by the economic and political changes described above. It is opposed by the system of personal rank, by the Tongan propensity for expressing different principles of social differentiation in separate contexts, and by the relative abundance of the subsistence economy, even now.

The field work on which this paper is based was carried out from 1958 to 1960, when I accompanied James Spillius, who was employed in Tonga as anthropological consultant to the World Health Organisation. I was employed by the Tonga Traditions Committee of the Government of Tonga. I am grateful to the Social Sciences Research Council for subsequent financial support which has made possible further documentary research and the writing of this paper. I am especially grateful to His Majesty King Tāufa'āhau Tupou IV for permission to publish this paper and to Garth Rogers and Elizabeth Wood Ellem for much helpful criticism.

- 78

• AOYAGI, Machiko, 1966. “Kinship Organisation and Behaviour in a Contemporary Tongan Village.” Journal of the Polynesian Society, 75: 141-76.

• BAKER, John R., 1977. “Contemporary Tonga—An Economic Survey,” in N. Rutherford (ed.), Friendly Islands: A History of Tonga. Melbourne, Oxford University Press, pp.228-46.
• BEAGLEHOLE, Ernest, and Pearl, 1941. Pangai: Village in Tonga. Wellington, Polynesian Society Memoir 18.
• BIERSACK, Aletta, 1974. Matrilaterality in Patrilineal Systems: The Tongan Case. Unpublished Curl Bequest Prize Essay.

- 79

• BOTT, Elizabeth, 1960. Tongan Society at the Time of Captain Cook. Unpublished manuscript in Palace Records Office, Nuku'alofa, Tonga. (Substantially the same as Bott in press.)
• —— “Psychoanalysis and Ceremony”, in J. S. La Fontaine (ed.), The Interpretation of Ritual London, Tavistock Publications, 205-37; 277-82.
• —— 1972 see copy with the assistance of Tavi, in press. Tongan Society at the time of Captain Cook's visits: Discussions with Her Majesty Queen Sālote Tupou. Polynesian Society Memoir 44.

• COOK, Capt. James, 1961 and 1967. The Journals of Captain James Cook on his Voyages of Discovery. Edited by J. C. Beaglehole. Cambridge, Cambridge University Press for the Hakluyt Society. Vol. II, The Voyage of the Resolution and Adventure, 1772-1775; Vol. III, The Voyage of the Resolution and Discovery, 1776-1780. Part 1 (Cook's Journal) and Part 2 (Journals of Anderson, Samwell, and several officers).
• COULT, Allan, 1959. “Tongan Authority Structure: Concepts for Comparative Analysis.” Kroeber Anthropological Society Paper, No. 20, p.56-70.

• FIRTH, Raymond, 1957a. We, The Tikopia: A Sociological Study of Kinship in Primitive Polynesia. London, Allen and Unwin.
• —— 1957b. “A Note on Descent Groups in Polynesia.” Man, 57: 4-8.
• FUSITU'A, 'Eseta, and Noel RUTHERFORD, 1977. “George Tupou II and the British Protectorate,” in N. Rutherford (ed.), Friendly Islands: A History of Tonga. Melbourne, Oxford University Press, pp.173-89.

• GIFFORD, Edward W., 1929. Tongan Society. Honolulu, Bernice P. Bishop Museum Bulletin, 61.
• GROUBE, L. M., 1971. “Tonga, Lapita Pottery, and Polynesian Origins.” Journal of the Polynesian Society, 80: 278-316.

• KAEPPLER, Adrienne L., 1971. “Rank in Tonga.” Ethnology, 10: 174-93.
• —— 1978a. “Me'a Faka'eiki: Tongan Funerals in a Changing Society,” in Niel Gunson (ed.), Essays in Honour of Harry Maude. Canberra, Australian National University Press, pp.174-202.
• —— 1978b. “Exchange Patterns in Goods and Spouses: Fiji, Tonga and Samoa.” Mankind, 11: 246-52.
• KORN, Shulamit R. Decktor, 1974. “Tongan Kin Groups: The Noble and the Common View.” Journal of the Polynesian Society, 83: 5-13.

• LĀTŪKEFU, Sione, 1974. Church and State in Tonga. Canberra, Australian National University Press.
• LEACH, Edmund, 1961. “The Structural Implications of Matrilateral Cross-Cousin Marriage,” in Edmund Leach, Rethinking Anthropology. London, Athlone Press, pp.54-104.

• MARCUS, George Emanuel, 1975a. The Ancien Regime in the Modern Kingdom of Tonga: Conflict and Change among the Nobility of a Polynesian Constitutional Monarchy. Unpublished PhD thesis, Harvard University, Cambridge, Mass.
• —— 1975b. “Alternative Social Structures and the Limits of Hierarchy in the Modern Kingdom of Tonga.” Bijdragen Tot de Taal-Land-En Volkenkunde. 131: 34-66.

- 80

• —— 1977. “Contemporary Tonga—The Background of Social and Cultural Change,” in N. Rutherford (ed.), Friendly Islands: A History of Tonga. Melbourne, Oxford University Press, pp.210-27.
• MARTIN, John and MARINER, William, 1818. An Account of the Natives of the Tonga Islands, compiled and arranged from the extensive communications of Mr William Mariner, several years resident in those islands. 2 vols. London, John Murray, Second Edition.
• MAUDE, Alaric, 1971. “Tonga: Equality Overtaking Privilege,” in R. Crocombe (ed.), Land Tenure in the Pacific. Melbourne, Oxford University Press, pp. 106-28.
• —— 1973. “Land Shortage and Population Pressure in Tonga,” in H. Brookfield (ed.), The Pacific in Transition. London, Edward Arnold, pp.163-85.

• NYACAKALOU, R., 1959. “Land Tenure and Social Organization in Fiji, Tonga and Samoa.” Journal of the Polynesian Society, 68: 93-114.

• POULSEN, Jens, 1977. “Archaeology and Prehistory,” in N. Rutherford (ed.), Friendly Islands: A History of Tonga. Melbourne, Oxford University Press, pp.4-26.

• RADCLIFFE-BROWN, A. R., 1924. “The Mother's Brother in South Africa.” South African Journal of Science, 21: 542-55. Reprinted in Structure and Function in Primitive Society. London, Cohen and West, 1952.
• RIVERS, William H. R., 1914. The History of Melanesian Society. 2 vols. Cambridge University Press.
• ROGERS, Garth, 1975. Kai and Kava in Niuatoputapu: Social Relations, Ideologies and Contexts in a Rural Tongan Community. Unpublished PhD thesis, University of Auckland.
• —— 1977. “The Father's Sister is Black: A Consideration of Female Rank and Powers in Tonga.” Journal of the Polynesian Society, 86: 157-82.
• RUTHERFORD, Noel, 1971. Shirley Baker and the King of Tonga. Melbourne, Oxford University Press.
• —— 1977. “George Tupou I and Shirley Baker,” in N. Rutherford (ed.), Friendly Islands: A History of Tonga. Melbourne, Oxford University Press, pp. 154-72.

• SAHLINS, Marshall D., 1958. Social Stratification in Polynesia. Seattle, University of Washington Press.
• —— 1968. Tribesmen. Foundations of Modern Anthropology Series. Englewood Cliffs, Prentice-Hall.
• —— 1972. Stone Age Economics. London, Tavistock Publications.

• THOMAS, John, (n.d.). Tongatabu or the Friendly Island. Unpublished manuscript formerly in the archives of the Methodist Missionary Society, London, now (1980), transferred to the Library of the School of Oriental and African Studies, London University.
• —— ca. 1853. History of Tonga. Unpublished manuscript formerly in the archives of the Methodist Missionary Society, now (1980), in the Library of the School of Oriental and African Studies, London University.

• VASON, George, 1810. An Authentic Narrative of Four Years' Residence at

- 81

• Tongataboo. London, Longman, Hurst, Rees and Orme.
• VE'EHALA and Fanua, Tupou Posesi, 1977. “Oral Tradition and Prehistory”, in N. Rutherford (ed.), Friendly Islands: A History of Tonga. Melbourne, Oxford University Press, pp.27-39.

- 82 Page is blank