TONGAN AUTHORITY STRUCTURE: CONCEPTS FOR COMPARATIVE ANALYSIS
Allan Coult
Introduction
The appellation "feudalistic" has frequently been applied to the
means of allocating authority In a number of Polynesian-societies.   Most
notable among these are Hawaii, Tahiti, Tonga and the Marquesas.   The use
of the term "feudal," of course, implies a resemblance to the authority
structure of Medieval Europe.  However, since the investigators who des-
cribed Polynesian social systems often used, as their sole criteria of
feudalism, the existence of lord-vassal relationships and/or the granting
of fiefs, it follows that frequently only a limited part of the authority
structure of the above societies has been examined to determine whether
or not they compare with that of Medieval Europeo Therefore, it is reas-
onable to interpret the assertions that certain Polynesian societies were
"feudal," to mean that the implied comparison holds only for the criteria
utilized. The objection may arise that the comparison of even these few
criteria is not valid, since the lord-vassal relationships in the two
areas were not equivalent.  This Is true in term's of the total content of
the relationships; the system of ritual which symbolized the social rela-
tion- was quite different .in Polynesia and Feudal Europe. But, if the
frame of reference for comparing the two systems is in terms of the social
action only, then the above objection loses much of its validity.
If it can be assumed for the moment that the lord-vassal, fief
granting relationships in both Europe and Polynesia were socially equiva-
lent, it would follow that many other social elements in the two societies
might be equatable, inasmuch as the relationships between these elements
in any society are not random. The intent of this paper is to take the
first steps toward determining the equivalences, if any, between these
elements.  To do this, a Polynesian mfeudal" society (Tonga, during the
early 1800's) will be analyzed so as to allow future comparison with Euro-
pean feudal societies.  Obviously, in any comparison of this sort, the
units of comparison must be chosen in such a way that they are equally
applicable in either of the systems to be':compared.  This qualifi.cation
rules out the possibility of using- the categorie.s der ived for comparing
Western societies with one another, since non-Western societies frequently
lack the counterparts of what, in Western societies, appear to be discrete
institutions, such as kinship, government, technology, and economy. When
attempting cross cultural comparison, the limitations of the use of these
traditional categories become very apparent; then there is great confusion
oXrer such questions as, where does the kinship systeol end and the politi-
cal system begin; is a certain element of behavior to be classified as
religion or technolog;y; and can this or that society be sai4d :to have a
political system.  In order 'to obviate thse diff&clties the: iethodoThgy
employed in this paper is directed toward analyzing the soclal system of
Tonga in terms of the social relations between indivi4rluals and groups
without sorting these relations into preconceived categories.   It is hoped
that a similar analysis of Feudal Europe would allow comparison between
systems which are frequently held to be incomparable.
The approach used herein is as follows: first, a model will be
derived of the Tongan social system with reference to the types of social
groupings, the basis for the affiliation of the individuals in these
groupings, and the allocation of authority in these groupings. Second,.
the variations from the model will be described in order to give a more
coherent picture of the Tongan social system. Finally, an attempt will
be made to determine what mechanisms maintain the social system, in the
form which it is found to have.
Population and Economy of Tonga
The Tongan archipelago comprises over one hundred islands, which
are naturally divided into three groups, Vavau in the north, Haapai in
the center, and Tongatabu in the south. The largest island has an area
of 150 square miles, whereas the smallest of the islands are only a few
acres in area. The distance from the center of the Vavau group to the
center of Haapai is approximately 150 miles and from the center of
Haapai to the center of Tongatabu is another 100 miles.
The population of these islands at the beginning of the nineteenth
century has been estimated at 6,000 for Haapai, 6,000 for Vavau, and 8,000
for Tongatabu.  Of the total population of 20,000, approximately 6,000 can
be assumed to have been males of fighting age. (Gifford 1929:5-6)
The land was intensively cultivated, and normally enough food was
produced through both horticulture and fishing to provide the inhabitants,
throughout the year, with more food than they could consume. However,
there were occasional famines during which the lower classes fared better
than their superiors, since the former were able to hide their produce in
times of scarcity.
Sea transportation was furnished by canoes, the largest of which
could carry upwards of one hundred persons and provisions. Journeys as
far as 400 miles without touching land were common.
Tongan Social Groups
The household group:  The smallest social grouping in Tonga, considered
relevant to this discussion, was the small household group; usually a
number of persons who lived under the same roof. The nucleus of this
group was- a male ego and his wife, the remainder of the household con-
sisting of their own and adopted children. Inasmuch as this group might
be characterized as one based on kinship, then kinship must be taken to
mean a set of reciprocal obligations, a good many of them rather diffuse,
and a feeling of solidarity with certain emotional overtones. Kinship
could not mean only a principle of affiliation, since the individuals
were associated variously by a) socially recognized biological ties
(ascriptive affiliation), b) marriage (a more or less voluntary associa-
tion established by certain rules defined by the society), c) adoption
57
(a voluntary association on the part of one party and involuntary on the
part of the other).  Therefore, a kinsman must be defined as a person
toward whom one acts and feels in a certain way, not a person with whom
one is associated vis a vis certain rules of descent.  If this definition
is accepted, then it follows that in Tonga, as well as in the rest of the
world, kinsmen are aligned in groups which may be seen to have certain
genealogical and marriage ties, but many other ties, not always obvious,
are also operative. Whereas this may be considered a; simple minded analy-
sis, it serves to point up the need to keep distinct,: the behavioral
aspects of affiliation and the criteria on which the affiliation is based.
The need for this sort of distinction, though perhaps not very convincing
in relation to the household group, will become more apparent as the
larger groupings are consideredo
The household group was associated with a plot of land to which
the male head of the household held the usufruct rights. Succession to
these rights was generally by primogeniture in the male line.  In the
event that a man had no sons, the land could be left to a daughter, a
brother, or a sister's son. The heir always had to have the approval of
the chief (see below) who owned the land.
Theoretically the- head of the household held absolute powers over
all members of this group, and also over their personal property. Obvi-
ously, so-called absolute power is always mitigated by many considerations,
including the personalities of all persons concerned and pressures from
outside groups. Thus, it is always necessary to distinguish between what
various persons can in theory do, and what they actually do. This was not
done, at the level of the household group, by any of the persons who wrote
on Tonga; therefore only a limited part of the Tongan social system is
available for analysis at this level.  Nevertheless, a few valid assump-
tions can be made on the material available. There is, for example, a
high degree of certainty that the head of the household was held responsi-
ble, by persons outside of the household group, for the actions of his
subordinates,
A man was always superordinate in terms of authority to his wife;
this applied even if his wife's rank was superior to his. A child showed
great respect toward his father and all of his father's brothers. This
respect relationship extended to the father's sister and, in this instance,
was especially severe. Extremely permissive relations existed between a
sister's son and mother's brother. The same type of familiarity applied
also to father's father, m:ther's father and mother's sister.(l).
The Subdistrict:  A number of household groups, dispersed over the country-
side, were under the authority of a titled chief who held residdual rights
to all the land in his subdistrict.  There is no evidence of any intermedi-
ate authority between the household and subdistrict levels. The ties
which existed between the various persons in a subdistrict, besides those
of territorial contiguity, have to be inferred from a very hazy analysis
given by Gifford.  He appears to have identified the subdistrict as a
Rlineagelt and describes it as follows:
The lineages are patrilineal.  Each consists of a nucleus of related
chiefs about whom are grouped inferior relatives, the lowest and
most remote of the commoners. Some commoners are not aware of their
lineage as such, but most are, and claim relationship to some chief,
usually the one under whom they live.  Thus the commoners living
under the chief Ata are reckoned as of his lineage. . . The landed
chiefs of today are all grouped in lineages. Modern representatives
of former powerful chiefs against whom the political situation has
turned, and,who have sunk to the status of commoners, are as a rule
unconnected with the modern lineages.
Everything points to the necessity of a line of powerful chiefs
for a nucleus about whiich the lineage groups itself. Without such
chiefs it appears to wilt and die and its membership gradually aligns
itself with other rising lineages. This process of realignment
naturally contravenes the rule of patrilineal descent, which theoreti-
cally, and largely in practice, determines lineage membership. Adop-
tion into lineages is practiced.
There was no rule against marriage within the haa [lineage].
Close blood relationship was the only bar. Patrilocal residence is
said to have been the rule with both chiefs and commoners, a factor
which would tend to keep the patrilineal haa localized. (Gifford
1929:30-31)
Although Gifford at f irst states that a "lineage" is a patrilineal
descent group, it may be seen that membership in a "lineage" was deter-
mined by a) pat?tilineal descent, b) non-unilineal descent, c) a claim of
non-unilineal descent, d) residence in the "lineage" territory. Any one
of these factors or any of the possible combinations of them appears to
have validated membership in the "lineage."
The apparently muddled composition of the Tongan "lineages,"R on
analysis, can be seen to conform to a few simple principles. Obviously
the "lineages" were not groups of individuals related through patrilineal
descent lines.  However the nucleus of chiefs about which the "lineage"
was formed were members of the same patri-descent group since inheritance
of chiefly titles was by primogeniture in the patrilineal lines, and it
was from the name of the chiefs' descent groups that the "lineages" t9ok
their names. Therefore, all persons living on land belonging to chiefs
who were in the same patri-descent line, were members of the same
Itlineage.91 This partially accounts for the fact that the same "lineage"
name occurred among widely separated groupso
Although chiefs of different subdistricts had the same "lineage"
name as chiefs of a number of other subdistricts, there was no necessary
connection between them for purposes of administration, unless one of
them happened also to be the district chief (see below). In this latter
event, the inferior units would also be considertd as "sublineoges" in
relation to the superior unit, although it was only-the chiefs who were
related in tht patrilineal line.
59
It is evident from the above that the members of a 'subistri ct,
besides the line of chiefs which governed that unit, had no co-sis tent
genealogical ties with one another (either actual or stipulated)'. They
were variously related by non-unilineal descent and affinal. ties. Many
of them could probably trace descent non-unilineally to a common ancestor
with their chief. However, a few did not even claim that they were rela-
ted to the chief in this way. Since the concern of Polynesi.an groups
with genealogies is usually related to the validation of titles and
property holding, this latter fact indicates that-such alid04on was of
no particular importance to these persons, who had only usufuct rights
to the land. The household groups held the land at the discre.tion of the
chiefs, and this fact overruled all genealogical considerations.
.Briefly sumnmarized, the structure of a subdistrict apears to have
been as follows:  the only principle of affiliation applicable for all
members of the group was common residence on land owned by a chief. Suc-
cession to the chieftainship, .and thus inheritance .of all the land, was
through the patrilineal line of the chief. Inheritance of the usufruct
rights to the land was also through the patrilineal line. However, the
chief could revoke the rights of any person to use his land. The menmbers
of the household groups were affiliated, variously, by non-unilineal ti es
and marriage. Many of the household groups claimed a common.ancestor
with the chief, however some made no such claim.
The presence of patrilineages among the chiefs of Tonga and the
supposed membership of the chiefs' subjects in these patrilineages suggests
that the Tongan social system, in relation to descent. groups, landholding
and authority structure was a development from a former unilineal descent
system wherein all persons were members of an actual patrilineage which
owned a certain territory, on which only members of the patrilineal des-
cent line and their wives and children lived.  The reasoning behind this
is as follows:
Goodenough has indicated that unilineal landholding groups are not
compatible with a scarcity of land:
In any community where cultivatable land is not overabundantt in rela-
tion to population, and all rights to land depend on mmbership. in a
.strictly unilinear kin group, a serious problem must soon arise.
Unilinear groups inevitably fluc.tuate .considerably ..in size.  The
matrilineal lineages on Truk, for example, readily doublee or halve
the ir membership in the space of :one-. or two generations.' As a result,
one lineage may have twice as much land as its members need while
another has not enough to go around.  Unless 'devices are developed to
redistribute land rights to.persons outside the'owning group, intra-
community conflict is inevitable.  (Goodenough  955:l9).
When land was scarce among the matrilineage:s on,Truk., it was redis-
tributed by separating use' 'r'ights from membership in the`- oih:ng group.
Thereby a lineage could grant usufruct rights in Its land to. the sons of
a lineage member who was living patrilocally.  The sons would form the
nucleus of a new matrilineage, and on the death of their fath,er would
receive ownership rights in the land.
60
If the original Tongans were divided into patrilineages, as
described above, which were faced with a land shortage, and if they
accomplished a redistribution of the land in the same way as the Trukese,
by gra'nting usufruct rights to non-lineage members, but not giving even-
tual ownership rights, then the Tongan system described herein would be
a logical consequence. The patrilineages holding rights to the land
could eventually emerge as an uptler class through the disfranchisement of
the groups who could never have more than use rights to the land. This
situation need not have arisen on Truk where groups which were given use
rights were eventually granted ownership.'
One of the major assumptions involved in the above thesis is that
the patrilineal group would not grant ownership rights, whereas it is
known that certain matrilineal groups in the same situation did so. How-
ever this assumption seems justified in that certain structural elements,
which in mrtrilineal groups would facilitate the relinquishment of owner-
ship rights, are not present in patrilineal descent groups. Thus in
matrilineal Tnik, redistribution of ownership rights involved granting
lineage land to the sons of the lineage members.  The parallel situation
in Tonga would have been the granting of lineage land to the sisters'
sons of lineage members. Since there is good psychoanalytic as well as
empirical data that the bond between a man and his son in a mtrilineal
society is stronger than the bond be tween a man and hi s si ster 's son in a
patrilineal society, it follows that the transfer of ownership rights to
one's sister's son is less -likely to occur than the transfer of these
rights to one's own son, in the respective matrilineal and patrilineal
situations.
Logically the next step in this analysis of the Tongan social
system would be to demonstrate how the subdistricts link into the larger
groupings of Tongan societyo  However, since this demonstration is rEade
easier by an undlerstanding of the Tongan class structure, a discussion of
titles, ranks, and professional occupations follows:
Titles and Rank: The literature on class structure in Tonga tends to be
as confused as the data on the descent and residence gropsu    This is due,
in part, to the failure of the writers to distinguish consistently between
title holding and the system of ranking.  The holder of a title always had
either administrative powers over the persons in a given territory or he
was retained in an advisory capacity by a person having administrati ve
powers.  Title holding, however, did not necessarily imply cerembnial prece-
dence over all persons holding lower titleso  In contrast, superior rank
always implied ceremonial precedence but not necessarily administrative or
advisory status- Usually, the holder of a title had to have a certain
rank (2).
There were four distinct ranks in Tonga, hierarchically graded in
the following order; robles, matabules, mutL and tua a  The nobility included
the chiefs and the ChbIdre. of the cMiefs by their oob-e v4ves   The child
of a chief by a non-noble woman was a non-noble.  The child of a noble
woman by a   rnon-nobie 'n was always a noble.  Thus nobility descerded
through the female line.  This fact has been used as evidencet to sbstan-
tiate the presence of a forimer descent system in Tonga, wherein title as
161i
well as rank were inherited matrilineally. However, the inheritance of
rank through the mother and title through the father can be explained
without postulating former matrilineality. The Tongan chiefs were poly-
gynous, and if all of their children had been nobles, the nobility would
soon have numerically exceeded the lower classes. If, as was the case,
only the offspring of noble women were nobles, then the number of the
nobility would be kept within reasonable limits. Furthemore, the much
closer identification of a child with its mother than with its father,
which is usual in polygynous families, would also contribute toward the
ranking of a child commensurate with its mother's rank.  These processes
are in no way incompatible with patrilineal inheritance of titles.
The matabule class included the advisors to the chiefs, their
wives and (it is assumed) a number of their offspring (the exact nature
of the inheritance of matabule and mua rank cannot be determined from the
literature). The mua were the warrior class and the descendants of war-
riors. Finally, the lowest class of people were the tua, who worked the
land, and were the servants of the chiefs.
The descendants of the chiefs and their advisors (Matabules), who
were not in line for succession to their titles, reverted to increasingly
lower rank, and in one or two generations would normally be tuas. Thus,
birth was only one of the criteria by which rank was determirnd. Valida-
tion of rank was also dependent on such factors as title holding, occupa-
tion, prestige and relationships with titled and high ranking persons.
The son of a chief by a matabule wife might be ranked as a matabule, mua
or tua, depending on his occupation. Similarly the younger sons of chiefs
by their noble wives might hold lower than noble rank if they were not
favored by their father.
The normal rule in the Tongan class system was downward mobility.
Gifford states that "regression is a simpler process than progression and
for every person who rises in rank there are many who correspondingly
sink." (Gifford 1929:112) The primary asset necessary for upward mobility
was, usually, fighting prowess.  The sons of tuas could rise to the rank
of mua by being successful in war and being favored by a chief. A chief
mIght also raise a tua or mua to matabule rank if the person showed partic-
ular abilities as an advisor. Only a few persons ever rose to the rank of
noble, and this was usually accomplished through warfare, wherein a chief
would appoint one of his matabules or muas as governor of a conquered
territory, and this position would subsequently validate the matabule's or
mua's claim to nobility.
The highest title in Tonga was that of chief   A chief theoretical-
ly held absolute power in the territory over which he had jurisdiction.
The title of Matabule was held by the senior advisors and attendants to the
superior chiefs, and the title of Mua was held by the advisors to inferior
chiefs and the junior advisors of superior chiefs.
All of the titles were inherited patrilineally. Usually the oldest
son inherited his father's title, however, if he were too young to fulfill
the duties his father's younger brother might hold the title until the
former had come of age.
62
Professional occupations: The professional occupations in Tonga varied
as to whether they were hereditary or not. The more skilled professions
were -inherited, not in the sense that a man was obliged to follow the
occupation of his father, but in that persons usually chose to practice
the skills taught them by their fathers. A few of the less valued skills
were practiced by anyone who had the inclination to do so. The two low-
est occupations, farming and cooking, for the chiefs, were hereditary by
legal requirement, since the chiefs demanded that the children of their
cooks and peasants perform the same services as their parents. The fol-
lowing list of Tongan occupations, indicating the rank of the persons
who practiced them, is taken from Mariner (1818:vol II:91).
[ canoe builders         ]
cutters of whale-teeth 3    Followed by both
ornaments                   matabules and muas
[ superintendents of     ]
Hereditary     [ funeral rites          ]
by choice
[ stone masons           ]
net makers
[ fishermen              ]
large house builders    >   Followed by both
[                        ]    nmuas and tuas
-  tattooers             ]
Hereditary    <[ club carvers
or not         [c
barbers
cook]
Hereditary     [ cooks                   >   Followed only by
by demand        peasants                    tuas
Individuals gained prestige and material benefits, including
presents from the chiefs, for practicing the higher skillso The more
skilled a man was in his profession, the more esteemed he was in the com-
munity. As indicated above, it is probably - fair assumption that the
rank of a person derived from his occupation and his skill in it, as well
as from his genealogical connections.
The District and Higher levels of Authority: To continue the interrupted
discussion of the authority structure, it will be recalled that the sub-
district was the next higher level of authority over the household groups.
A number of subdistricts were united in a district under the authority of
a single chief. While the subdistrict chiefs owned the land in relation
to their peasants, these chiefs held only provisional rig.hts to the land
in relation to the district chiefs. These latter, in turn, held only
provisional rights to the land in relation to their superiors and so on
up to the king. The subdistrict chief who was autonomous in his own ter-
ritory, owed tribute in the form of food and other goods, and military
63
aid to the district chief, If the tribute was not paid, then the district
chief had the legal right to deprive the inferior chief of his position.
Whether the district chief would in fact have done this would have depended
on his strength in relation to that of the subdistrict chief.  A district
chief could create new subdistricts by granting portions of his own land
to his noble relatives or to other nobles, whom he judged worthy, in
return for the usual tribute and military aido
The level of authority above that of the district was the island
group,-corresponding to the three natural divisions of Tonga (Vavau,
Haapai, and Tongatabu). The chiefs of the island groups stood in the
same relation to the chiefs of the district as these latter did to the
chiefs of the subdistrict.  Finally, a king, standing in similar relation-
ship to the chiefs of the island groups, was the top level of authority in
the system. Since the king lived in one of the island groups, he was also
the chief of that group0 Whether or not he was also the chief of a dis-
trict is not ascertainable from the literature.
From the foregoing it may be seen that a chief at any level of
integration was theoretically autonomous within his own territory and owed
tribute and military aid to his superior chief. He was also obliged to
appear at the court of his superior chief when requested to do soo A
superior chief in all instances held residual rights to all land of his
subordinates, and the latter nmintained their positions by carrying out
their obligations in good faith.
The retinue of superior chiefs (district or higher) consisted of
inferior chief, untitled nobles and titled and untitled matabules. The
retinue of the inferior chief consisted of a body of fightingm nn, most
of whom had the rank of mua and some of whom were tuaso The process by
which this retinue came to be attached to a chief is described by Mariner
(1818:vol II:285ff) as follows:
We will suppose that the present king or any other great chief hs a
son six or seven years of age, his playmates are the sons of the
Inferior chiefs, matabooles, and muas of his father's establishnent,
who freely associate with him, accompany him upon excursions, an
imitate, in-many respects, the habits of their parents:  he does not,
however, designedly play the chief, and conduct hinmelf with arrogance
towards them; they know his superior rank without being reminded of
it; 0 . . in some of his country excursions, he perhaps meets with
two or three of the sons of tuas, who by their strength and agility
in wrestling, or bravery in boxing, or some other ostensible quality,
recommend themselves to his notice, and therefore also become his
companions0  Thus they grow up in years together, and as the young
chief approaches towards manhood, he does not exact9 but he receives,
with more or less affability, the respect and attention which his
inferior associates readily pay him, and who now may be termed his
cow-tangata, i0e0, associates, supporters :and defenders of his cause0
He has not yet however, any matabooles [titled matabulesl in his
train, for all these are in the immediate service of the old chief,
and the son of a mataboole cannot be a mataboole till his father is
64
dead, and then he would not perform the functions of a mataboole,
unless he were grown to man's estate, and even then he would not
be in the service of the younger chief, but of the elder. By and by
the old chief dies, and the young one succeeds to his authority, and
all the matabooles of his father become his matabooles, and the
inferior chiefs and mooas also enter his service in addition to
those he had before. . 0 0
The members of a superior chief?s retinue who actually dwelt
within or near his fencing were as followss  all of his titled matabules,
since their presence was constantly required in their advisory capacity;
a few of the infer ior chiefs ; and approximately half of the muas both
titled and untitledo The chiefs were responsible for the feeding ard
quartering of all of these persons0 It is interesting to note, in this
connection, that during famines more of these dependents of the chiefs
died than persons of the lower classes since the latter were always able
to reserve food for themselves despite the great taxes0  (Mariner:01818:,
400)
The primary loyalties of any Tongan were theoretically to the
titled chief immediately superior to him0 The ideal Tongan followed the
orders of his chief without question, even if this involved fighting an
even higher chief. That this ideal pattern seemed to have been followed
in practice is borne out by the details of the wars in Tonga, as they are
described by Mariner. Frequently, close relatives, such as father and
son or brothers, fought one another since they lived under different
chiefs. However, a person holding a high position might be excused by
his superior chief from fighting against a relati'veo
Public meetings in Tonga were held at the subdistrict, district
or higher levelo Unlike similar meetings in other areas of Polynesia
there was no discussion among the persons holding various titleso Only
the chief or his Matabule talked and communicated his order to the peopleo
Of course, many of the directions given at these meetings were decided on
at conferences between the chiefs and their advisorso
Tle outline of the Tongan social system so far given may be viewed
as a model of the authority structure0  It is not a statisti cal model and
it is doubtful whether it is an ideal model, either from the point of view
of how the people thought the system actually operated, or how they thought
that it should operate. It is certainly not a model of how the system did
in fact operateo It might best be thought of as a frameework which is
arbitrarily derived so as to give coherence to the authority structure,
and to act as a base from which variation can be measured. Variations
from this base are not to be thought of as aberrations or as not conform-
ing to the model. The variations may statistically be more frequent than
similar elements presented by the model; this, however, should not throw
doubt on the validity of the model, since it is used only as a frame of
reference to further study of the workings of the system0  What follows
then, is a description of some of the more important variations from the
model.
Whereas the model does not provide for any intermediate levels of
authority between the household groups and subdistrict, Tongan matabule's
and mua's frequently had their own plantations. They held these lands
under subdistrict, district or higher chiefs. There is no information to
indicate the degree of authority that these persons held over the peasants
working their land. It is likely, however, that the peasants on a mata-
bule's or mua's plantation owed allegiance primarily to their superior
chief and that the powers of the matabule's and mua's over their planta-
tion workers were much less arbitrary than the power of the chiefs over
their peoples,
In the model system it was indicated that each person had only
one immediately superior chief under whom he held land,  It was, however,
also possible for an inferior chief to hold land under a number of super-
ior chiefs. Thus a chief's holdings might be spread out over a number of
districts and islands. Exactly to whom a chief owed allegiance in such a
case is not stipulated in the references, however, it seems probable that
his obligations were primarily to only one of his superiors and only
secondarily to the other grantors of land, (this is similar to the land
holding system of late feudal Europe).
The model system pictures a graduated series of territorial units
with subdistricts integrated into districts, districts into island groups
and the island groups united under one ruler. Frequently, the various
sub-groups asserted their independence from the larger groups. The chief
of the Vavau islands did not pay tribute regularly to the king; this
rebellion was facilitated by his separation from the king who at that
time lived in the southern group of islands. A much more nmrked example
of- the independent actions of various chiefs occurred at the death of
the king Toogoo Ahoo, who left neither son nor brother to succeed him.
Several distant relatives claimed the sovereignty, and the island of
Tongatabu was soon divided into at least thirteen warring factions, each
of which had erected a fortress and waged war with all the rest.
In the model, the chiefs have arbitrary powers over all persons
within their territory. This situation was, of course, mitigated by the
usual internal checks on arbitrary power, such as discontent of subjects,
fear of assassination and the counsel of advisors. In addition to these
checks the matabules of the superior chiefs kept watch over the activi-
ties of the inferior chiefs, especially the younger ones, and if they
were found to be abusing their people, their superiors could exert pres-
sure to rectify such circuimstanceso
Another variation from the model system was in the riethod by
which a successor to a chiefly title was chosen, The mode of inheritance
was described as patrilineal by primogeniture, except in the instance that
the oldest son was too young to assume power, wherein the title would be
held by a father's brother until the son came of age.  It was usual, how_*
ever, for a number of other considerations to enter into the determination
of succession to a title.  At the death of a chief any number of his patri-
lineal relatives might claim the titleo In this event, the strength and
number of the followers of these persons, together with their reputations
66
and the preferences of a superior chief, would all enter into the deter-
mination of the successoro
The followers of a chief might decide that the "legal" heir to the
title was incompetent, and decide among themselves that another person,
usually a close relative of the former chief, should rule. Wars, over-
ruled any considerations of inheritance.  A chief could dispose of newly
won land among his followers, who would depose the former cwners of the
land and move themselves into the latters' positions.  Territory won in
war might also be left in the hands of the former owners, who would take
an oath of allegiance to their new chief and py tribute to him.
From the foregoing it may be concluded that the principles on
which succession to a title were based were much more complicated than
the simple preference for c.ertain genealogical kin outlined in the model.
A number of interacting preferences giving form to social actions deter-
mined who would govern a given territory".  It is, in fact, a fair assump-
tion that any facet of the.social structure in Tonga, and elsewhere, is
similarly determined.- The tendency of many anthropologists to focus on
the genealogical determinants of social structure frequently obscures
other important determinantso  One examnple of this type. of obfuscation
is the use -of the catch-all category "fictional kinship" which serves to
equate relationships which are based on a whole hos t of different prefer-
enceso
Social systems may be analyzed in terms of their structural elem-
ents only.  F.or example,- it can be determined that a group of people are
aligned spatially, by patrilineal affil-lation, without determining what
preferences for social action result in this type of affiliation.  How-
ever, inasmuch as this type of study allows. of predictability, the prefer-
ences for social action have to be correctly assumed.  Thus, Goodenough's
prediction of the breakdown of matrilineal.. inheritance u'nder pressure of
land shor-tage assumes a pref"erence to avoid Intracommunity. conflict and a
preference not to -redistribute land in such a way as to keep the matri-
lineages intact.  These two assumptions. and many more were nade in this
paper in predicting the fate. of patrilineages under.-similar conditions of
land shortage. Obviously a much greater predictability would.result if
the presence or absence of the assumed variables would be empirically
determined, One reason that this is usually not done is that many of the
preferences for social action are taken to be different kinds of phenomena
than so-called "structural eleentnts.'"  Yet, they both appear to be the
same thing viewed from different points6 For example, patrilineal affilia-
tion is considered. to 'he a structural element, inasch as It represents a
way of aligning persons which can be expressed sociometrically, A prefer-
ence for warfare is not.usually considered to be a structural .element but
is variously categorized ag a Wvaluew or.an element of 'tethOs,W This lat-
ter, however, seen from. another point of view, Is also a structural elem-
ent, since the alignm.ent of individuals for warfare, which may result
from this.. preference., can also be. expressed sociof'tricallye -To reverse
this illustration, actual patrilineal affiliation may result from a
preference for patrilineal affiliation or it may be the'.result of a number
of other preferences for social, action. Similarly a preference for war-
fare does not necessarily result in a social alignment based on this
67
preference, since other preferences may be more influential. Thus it may
be seen that what are usually considered to be discrete categories, i.e.;-
elements of social structure and elements of ethos or values can and
should be treated as two sides of the same coin.
Some factors which contributed to the relativer
autono   of the Tongan territorial units
It appears to be a valid assumption that the various chiefs of
Tonga placed a high value on autononsy for, inasmuch as they could, they
directed their efforts so as to be completely free of the control of other
chiefs. That the majority of chiefs were forced to surrender some of their
autonony by subjecting themselves to other chiefs, indicates that they also
valued such things as territorial expansion, loyalty, security, and surren-
der in hopeless situations. The degree to which they were autonomous, thus
depended on the degree to which these other values could be upheld in an
autonomous situation. This in turn depended on certain empirical factors,
such as the relative strength of the chiefs, and the distribution of power
at a given time. On Tonga, a chief had to have a vast numerical superior-
ity in fighting men in order to defeat a rival on the latter's territory.
This was due to the strong defenses, similar to early medi-eval fortifica-
tions, to which the chiefs and their followers retreated when attacked. A
chief having a superior force of ten to one had attempted for eleven years
to breach one of the fortresses, located among the thirteen warring fac-
tions of Tongatabu, and had repeatedly failed. Finally, with the introduc-
tion of carronades, he was able to take this fortress, but did not follow
up his victory.  Even the carronades, however, were ineffective against
fortresses erected on high levels. The strong defenses of the Tongans
were thus a major factor in preserving the autonomry of numerically weak
factions.
Another factor which promoted the independence of the Tongan chiefs
was the long distances between island groups, which made it difficult for
a superior chief to maintain control over his distant subjects, and encour-
aged the latter's recurrent declarations of independence.
Since the Tongans did not perceive any external threats in the
form of invasions from other areas, there was no pressure for centraliza-
tion of authority in order to defend against foreign attack.
Conclusions
Although the primary intent of this paper was not to draw compari-
sons between earl Feudal Europe and Tonga, in a preliminary analysis, the
authority structuZres of these two areas give the impression of being, to
a large extent equivalent, and predicated on the same values. This is not
to say that the elements of ritual and belief were equivalent.  However,
as far as the system of social action is concerned, these elements are
unimportant.  The same values may be derived from vastly different systems
of belief and, it is the values, not the beliefs, which determine what
social actions will occur. This is nicely illustrated by Mariner:
68
The Tongan people do not indeed believe in any future state of
rewards and punishments, but they believe in that first of all
religious tenets, that there is a power and intelligence superior
to all that is human, which is able to control their actions, and
which discovers all their most secret thoughts; and though they
consider this power and intelligence to be inherent in a number of
individual beings, the principle of belief is precisely the same;
it is perhaps equally strong, and as practically useful as if they
considered it all concentrated in their chief god. They firmly
belie-ve that the gods approve of virtue, and are displeased with
vice; that every man has his titular deity, who will protect him
as long as he conducts himself as he ought to, but if he does rot,
will leave him to the approaches of misfortune, disease and
death. (Mariner 1818,140ff)
A Tongan may uphold the social norms because he believes the gods
will punish him in this life if he transgresses them; a European may hold
the same value because he fears hell, but the value, however derived,
serves the same social puose in either society. This particular illus-
tration is used since one of the most frequent objections to comparing
so-called primitive Pblynesia with feudal Europe is that Europe had a
Catholic Church while Polynesia did not. Within therframiework of this
analysis the belief system is not germame. The church as a social system
is another problem which, however, is easily met. The church, or more
precisely, the persons holding ecclesiastical positions in early feudal
society were, for purposes of the social system, no different than non-
clerics. Thus a bishop might hold a fief, and render service to his lord
just as any other man,
Although it would take another paper to attempt to demonstrate
adequately and convincingly the equivalences noted above, between early
Feudal Europe and Tonga, it is hoped that at least the validity of
attempting such comparison has been indicated.
Notes
(1) These relations applied, of course, to all persons who were addressed
by the terms indicated, whether they were actual genealogical rela-
tives or not,
(2) Since some of the Tongan terms for titles are identical to those for
rank, the terms will be capitalized when the reference is to title
and not capitalized when the reference is to rank.
69
REFERENCES
Gifford, E. W.
1929     Tongan society. Bishop Museum Bulletin No. 61.
Goodenough, W. H.
1955     A problem in Malayo-Polynesian social organization.  American
Anthropologist 57:No, 1.
Mariner, W.
1818     An account of the natives of the Tongan Islands. London.
Other Sources Consulted
Cook, J.
1788     A voyage to the Pacific. Vol. 1. London.
Thompson, B.
1894     Diversions of a Prime Minister.  London.
1902     Savage island. London.
Williamson, R. W.
1924     The social and political systems of Central Polynesia.
London.
Wilson, W.
1805     First missionary voyage to the South Sea Islands. London.
West, T.
1865     Ten years in South Central Polynesia. London.,
70